**Preface to "An Evolutionary and Environmental Perspective of the Interaction of Nanomaterials with the Immune System-The Outcomes of the EU Project PANDORA"**

This book collects scientific contributions aiming to describe the mechanism common to all living species by which innate immunity interacts with nanomaterials. The overall goal is to harness such interactions for improving environmental and human safety and exploiting them for modulating immunity in vaccination strategies.

Most of the studies gathered here are the results of a successful collaboration effort across Europe, funded by the Horizon 2020 project PANDORA. The PANDORA scientists wish to dedicate this book to Valeria Matranga, a rigorous scientist and a good friend. Although Valeria left us at the very beginning of the PANDORA project, we have always kept in mind her approach to being a scientist.

> **Diana Boraschi** *Editor*

## *Editorial* **An Evolutionary and Environmental Perspective of the Interaction of Nanomaterials with the Immune System**

**Diana Boraschi 1,2,3**


Assessing the modes of interaction between engineered nanomaterials and the immune system is a topic of particular interest for research in several fields, from a toxicological and safety perspective to potential nano-based immunomodulatory strategies for medical use. This Special Issue gathers results and new information—mostly collected within the EU Horizon 2020 project PANDORA (probing the safety of nano-objects by defining the immune responses of environmental organisms), which specifically focuses on nano-immune interaction across living organisms, from plants to human beings. The underlying concept is that several of the immune defensive mechanisms used for tackling exogenous agents (including nanomaterials) are conserved across evolution with little modification. Thus, we looked for common mechanisms of recognition and reaction, based on the high evolutionary conservation of innate immunity, the ancient and highly efficient defensive system shared by all living organisms [1]. We wanted to find answers to the following questions:


First question: **are nanomaterials seen by the immune system as a threat?** Yes, in some cases, in that the innate immune system, in particular phagocytes, can "see" the nanomaterials and begin action to eliminate the nanomaterials and maintain the organism's functionality. Most interestingly, we should also consider the inverse interaction, i.e., how nanomaterials "see" the immune system and are modified by their interaction with it [2]. Notably, upon interaction, we can even observe beneficial biological effects, as in the case of decreased stress responses and growth promotion in the model plant *Arabidopsis thaliana* [3]. In many other cases, it is possible to observe an immune reaction, with morphological and functional changes in innate immune cells; however, these changes are not long-lasting and do not hamper the organism's integrity. This means that a successful immune reaction has taken place, which has recognised the nanomaterials as a possible threat and has acted to eliminate them and re-establish normal tissue/organism functions [4–7]. A very important point that should be considered is that exposure to nanomaterials may affect immunity indirectly by interacting with immune-modulating entities. In particular, when nanomaterials are ingested, the interaction of nanomaterials with the resident microbiota must be considered, as microbiota are well known to shape intestinal and systemic immunity [8].

Second question: **can we use our knowledge of nano-immune interactions for designing "smart" nano-based vaccines**? The use of nanoparticles is a very promising

**Citation:** Boraschi, D. An Evolutionary and Environmental Perspective of the Interaction of Nanomaterials with the Immune System. *Nanomaterials* **2022**, *12*, 957. https://doi.org/10.3390/ nano12060957

Received: 13 February 2022 Accepted: 7 March 2022 Published: 14 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

approach to vaccination because the particles may double their scope by acting as a carrier for the vaccine antigens, being able to shuttle them preferentially to antigen-presenting cells while being active as an adjuvant, i.e., they are able to induce an innate/inflammatory response that is necessary for the optimal induction of a specific long-lasting immunity [9]. Two aspects have been considered here: the induction of a specific anti-infective protective immunity in poultry [10] and the possibility of modulating innate memory in human innate cells towards a more protective secondary response, thereby generating not only adaptive memory (resulting in enhanced secondary specific response) but also innate memory (i.e., a long-lasting amplification of the specific response) [11]. In both cases, nanoparticles derived from bacterial cells were used, and were able to act both as antigen carriers and adjuvant particles.

Third question: **can we design common assays that enable us to assess the crossspecies effects of nanomaterials on immunity** (i.e., valid for both environmental species and human beings)? To this end, we have compared the most representative methods for evaluating immune reactivity across species in order to identify common conserved innate immune responses activated by interaction with nanomaterials. Excluding plants, whose extreme specialisation also impacts the type of immune defensive tools and responses, we have compiled a list of common assays, both in vivo and in vitro, that can be used to evaluate a response to nanomaterials (in terms, for instance, of safety) across animal species [12–14]. Notably, this implies that we can use some invertebrate models in vivo for predicting the effects of nanomaterials on human innate immunity [14].

To conclude, by examining immune response to nanomaterials across living organisms we have observed that immunity is, in general, able to cope with the nano-challenge and prevent detrimental effects to the organism. Both environmental species (marine and terrestrial invertebrates) and human beings display an array of common defensive mechanisms that are engaged in the interaction with nanomaterials, which allows us to identify model assays (in vivo and in vitro) which are predictive of nano-effects across species, making them useful for both environmental and human nano-safety testing. By evaluating nano-effects on the immune system, we can design nano-based vaccination strategies that exploit the immunomodulatory capacity of nanomaterials to achieve optimal long-term protective immunity.

**Funding:** This research was funded by the EU Commission H2020 project PANDORA (GA 671881) and by the Presidential International Fellowship Programme (PIFI) of CAS (2020VBA0028).

**Acknowledgments:** The author is grateful to all the PANDORA partners for their enthusiastic and active collaboration. Special thanks to Giuliana Donini for her relentless support of this project.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


## *Review* **The Interactions between Nanoparticles and the Innate Immune System from a Nanotechnologist Perspective**

**Lena M. Ernst 1, Eudald Casals 2, Paola Italiani 3, Diana Boraschi 3,4,5 and Victor Puntes 1,6,7,\***


**Abstract:** The immune system contributes to maintaining the body's functional integrity through its two main functions: recognizing and destroying foreign external agents (invading microorganisms) and identifying and eliminating senescent cells and damaged or abnormal endogenous entities (such as cellular debris or misfolded/degraded proteins). Accordingly, the immune system can detect molecular and cellular structures with a spatial resolution of a few nm, which allows for detecting molecular patterns expressed in a great variety of pathogens, including viral and bacterial proteins and bacterial nucleic acid sequences. Such patterns are also expressed in abnormal cells. In this context, it is expected that nanostructured materials in the size range of proteins, protein aggregates, and viruses with different molecular coatings can engage in a sophisticated interaction with the immune system. Nanoparticles can be recognized or passed undetected by the immune system. Once detected, they can be tolerated or induce defensive (inflammatory) or anti-inflammatory responses. This paper describes the different modes of interaction between nanoparticles, especially inorganic nanoparticles, and the immune system, especially the innate immune system. This perspective should help to propose a set of selection rules for nanosafety-by-design and medical nanoparticle design.

**Keywords:** nanoparticles; immune system; innate immunity; inflammation; tolerance

#### **1. Introduction**

The immune system of higher vertebrates encompasses a collection of different specialized cells and specialized soluble molecules distributed throughout the body, being present in all organs and tissues, circulating in blood and lymph (to reach every corner of the body in case of need), and concentrated in some lymphoid organs (lymph nodes, spleen, bone marrow, where hematopoiesis takes place in adult life). These cells have been classified into two functional branches, namely innate and the adaptive immunity, which have different roles, complementing each other very efficiently in complex organisms such as mammals (simpler organisms such as invertebrates only display a perfectly efficient innate immunity). The innate immune system's role is to scan the body to remove apoptotic bodies, cell debris, and protein aggregates; recognize and eliminate pathogens or abnormal cells; and keep commensals outside tissues. Additionally, it promotes the repair of damaged tissue and is involved in the control of embryogenesis and delivery. We can say that the innate immune system is the actual immune system, active throughout evolution with conserved and very efficient defensive mechanisms. The other system, adaptive immunity, developed much

**Citation:** Ernst, L.M.; Casals, E.; Italiani, P.; Boraschi, D.; Puntes, V. The Interactions between Nanoparticles and the Innate Immune System from a Nanotechnologist Perspective. *Nanomaterials* **2021**, *11*, 2991. https:// doi.org/10.3390/nano11112991

Academic Editors: Horacio Cabral and Thierry Rabilloud

Received: 16 September 2021 Accepted: 3 November 2021 Published: 6 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

later as a complement of innate immunity, providing slower but more specific protective responses, good for long-living and mobile organisms that do not stably reside in the same environment [1]. The adaptive immune responses are tools for the innate immune system with subordinated or programmed functions—tools because they develop without making any decision [2]. It is the innate immune system that detects, categorizes, and triggers the immune response and, in the case of additional needs, calls for adaptive immunity to come in when the innate activation has reached a certain threshold level indicative of excessive danger and the need for more specific defensive tools.

These complex defensive actions that body tissues perform in response to harmful stimuli, such as pathogens or damaged cells, are described as inflammation. Inflammation requires an excess biological workout and therefore it is closely related to metabolism. Immunometabolism has become increasingly popular since the publication of Mathis and Shoelson's perspective in 2011 [3]. This is crucial in the context of interactions with nanoparticles (NPs), since they have been observed to have the capacity to increase or decrease reactive oxygen species (ROS), which directly correlates with the onset or remission of inflammation [4]. ROS are free radical molecules resulting from natural metabolism, which, when excessive and unregulated, may contribute to cell damage and aggravate human pathologies such as cancer [5], neurodegeneration, and stroke, among others [6].

Following the great oxidation event some 2.3 billion years ago, oxidation has been the leading force of metabolism. A delicate equilibrium between heat generation (enthalpy) and biological organization (entropy) was established, which allowed natural systems to decrease their free energy in a particular controlled fashion [7]. Deregulation of a living system, for instance, in the case of a disease, increases enthalpy generation at the expense of entropy. The system over-burns, which in biological terms is described as inflammation (literally *setting in flames*). Inflammation is correlated with a particular metabolic pathway, anaerobic glycolysis, providing higher energy power output, in which cells defend themselves from aggression. Furthermore, aerobic glycolysis, with a broken Krebs cycle, provides important metabolic intermediates and ROS [8]. Inflammation provokes the unbalance between endogenous production of free radicals and antioxidant defenses, resulting in oxidative stress [9]. While this metabolic defense mechanism is an ability of all eukaryotic cells, it is reasonable to imagine that, through evolution, some cells adapted the unbalanced energy equation to becoming professional defensive cells forming a whole discontinued system distributed across the body and responsible for the maintenance, defense, and repair of our biological tissues. In normal conditions, these cells have a patrolling role based on scanning and surveying tissues to eliminate senescent or damaged cells and become aggressive when encountering some possible dangers, capable of initiating, developing, and controlling inflammation.

The innate cell response is different, depending on the type of stimulus or combination of stimuli, the stimulus intensity (quantitative and temporal), the location of the innate cells (the tissue and its specialization), and the microenvironmental conditions. All these cues trigger a defined activation profile in innate cells, which is different based on the combination of microenvironmental conditions that have triggered it. Engineered NPs may share several characteristics of microbial agents, such as size and ordered molecular surface patterns, presenting "eat-me" or "eat-me-not" surface signals that favor or prevent macrophages from engulfing them. Thus, they are expected to develop complex and intense interactions with immunity. Bachmann et al. [10] showed that the immune system readily recognized antigen repetitive organization on the surface of viral particles.

In contrast, poor antigen organization does not induce an immune response. The same holds for complement (in particular C1q, an ancient version of immunoglobulins) [11], which recognizes ordered antigenic structures as those present on microorganisms but do not react to disordered patterns as those present in mammalian cell surfaces. The same has been observed with NP coatings [12]. These interactions mainly concern innate immunity, as responsible for detecting and categorizing foreign matter inside the body.

In order to navigate the described interactions between NPs and the immune system, it is recommendable to remember the different type of immune cells and their different functions (Figure 1). A major role in the innate immune system is played by macrophages, which in mammals develop some specialized functions depending on the tissue where they reside and are named accordingly: Langerhans cells in the skin, Kupffer cells in the liver, osteoclasts in the bone, microglia in the brain. Other innate immune cells are the innate lymphoid cells (ILCs), such as natural killer cells. ILCs contribute to patrolling tissues (abundant in the barrier tissues such as mucosal surfaces), identifying and killing/eliminating abnormal cells and microorganisms, and contributing to tissue development and homeostasis. Contrary to macrophages, they cannot phagocytose, but they are endowed with cytotoxic tools that literally kill the target. Similarly, mast cells are highly efficient defensive cells, abundant in all barrier tissues, endowed with an array of pre-formed proteolytic enzymes and other bioactive substances, which they release upon challenge and can detoxify snake and bee venoms, release factors that initiate/enhance a tissue-localized protective inflammatory reaction against parasites, and contribute to tissue repair and remodeling. Other important innate cells are neutrophils (short-lived, very abundant in the blood, highly phagocytic and inflammatory, strong producers of reactive oxygen species (ROS) in response to microbes), basophils (functionally similar to mast cells but residing in the blood), and eosinophils (with partially overlapping functions with mast cells and basophils, involved in response to multicellular parasites). Moreover, cells of adaptive immunity include T and B lymphocytes, which develop membrane receptors or antibodies able to recognize different pathogenic molecules/antigens specifically. In between, there are dendritic cells, which share with macrophages the capacity of taking up, processing, and presenting pathogen-derived antigens to adaptive immune cells, thereby enabling T and B cells to develop their antigen-specific membrane receptors and antibodies.

**Figure 1.** Representative classification of the most common mammalian immune cells.

The majority of the works on NP and immune system interactions [13] have focused on circulating blood monocytes (macrophage precursors) and tissue-resident macrophages [14]. Blood monocytes come from the bone marrow, while tissue macrophages can be a mixture of self-replicating cells that have populated the tissue during embryogenesis (developed from precursors in the yolk sac or fetal liver) and cells developed from blood monocytes that enter the tissue for replenishing the resident macrophage pool [15]. Macrophages present higher phagocytic activity than monocytes and can be easily identified based on size (monocytes are smaller than macrophages) and some biochemical markers (e.g., esterases) and surface molecules (e.g., CD14, CD16, CD68, CD11b, MAC-1) that are differentially expressed between the two cell types. The functional profile of monocytes and macrophages is exceptionally plastic, as the role of these cells is that of rapidly reacting to different microenvironmental signals by adopting an appropriate activation profile, which will contribute to danger elimination and, eventually, instructing the subsequent adaptive immune responses.

#### **2. The NP-Immune System Interactions**

It is essential to realize that inorganic matter is commonly nanostructured and that NPs and nanostructures have naturally occurred on the planet's surface since its origin before life emerged. This suggests that the immune system of living organisms has developed in an environment rich in such structures and substances, and therefore it should know how to deal with them. However, since the advent of nanotechnology, we are building more artificial nanostructures and artificial combinations of nanostructures and molecules, which may result in stronger interactions with the immune system, beneficial or detrimental. They depend on the nature of the employed material and its evolution in the environment before encountering the immune cell. The observed interactions between NPs and the immune system can be classified as in Figure 2. It is essential to understand that the interaction of NPs and the immune system can be multifactorial; size, shape, and surface state (including composition, structure, charge. and hydrophilicity) are primary factors, together with the presence of bioactive moieties in the sample, or the promotion of chemical reactions resulting in immune activation. These factors are closely related to the NP evolution in different media, which may result into aggregation, dissolution, or associate with bystander (bio)molecules. All these factors should be taken into account to develop safe NPs and functional NPs.

**Figure 2.** NP-immune system interactions. NPs can be undetected or detected by cells of the immune system, depending on different parameters such as their size, surface charge, and hydrophobicity/hydrophilicity of the surface coating. If detected, NPs can be either tolerated (either ignored or eliminated in a silent fashion, i.e., without inducing an inflammatory reaction) or generate an inflammatory response allowing for resolution of inflammation and tissue regeneration or an anti-inflammatory. With a proper NP design, these responses can be harnessed for developing different immunomodulating activities for medical exploitation (e.g., self-adjuvanted vaccines based on virus-like particles (VLPs), or outer membrane vesicles (OMVs), etc.).

#### *2.1. When NPs Are Not Detected by the Immune System*

The first scenario is when the NPs can escape from immune system detection. Indeed, the progress in the construction of synthetic nanostructures as delivery vehicles, contrast agents, or medical devices has allowed for the development of NPs able to escape the immune system and reach their target without inducing an undesirable inflammatory reaction. In order to escape from pattern recognition receptors inserted in the cell membrane of immune cells, or opsonization by complement molecules or antibodies (which mark foreign/dead cells for recycling, thereby enhancing phagocytosis), the use of polymers to camouflage the NP surface has been thoroughly developed. In pharmacology, polyethylene glycol (PEG) and polyvinylpyrrolidone (PVP) have been the most used polymers to stealthily NPs from the immune system [13,16,17], historically developed together with the liposomal formulation of antineoplastic drugs, such as in the case of Doxil® and, more recently, mRNA vaccine platforms [16]. Studies showed that these polymer functionalized NPs may appear invisible (stealth) to the immune system [17,18] by mimicking non-dangerous biological structures. As a consequence, the PEG coating increased the halflife of organic and inorganic NPs in the bloodstream from minutes to hours [19,20], similar to previous observations with PEGylated proteins [21]. Similarly, oligosaccharide- and peptide-derived NP coatings seem to afford escape from the immune system and allow for longer circulation times. Nevertheless, these coatings can reduce/delay opsonization and phagocytosis, but they do not completely prevent it. Thus, for example, the development of anti-PEG antibodies has been reported in several patients, which led to faster clearance of subsequent doses of PEG-coated formulations [22–24]. To circumvent these problems, researchers have explored the possibility of coating NPs with natural substances that can more finely deceive the immune system, for instance, by coating NPs with albumin or serum mixtures. As far as proteins are not denatured, and the resulting object is not too big, NPs seem to pass undetected [25].

The different coatings employed to escape immune system detection are listed in Table 1.


**Table 1.** Coatings employed to avoid immune detection.

Another way of camouflaging NPs from immune elimination is using coatings with proteins that are downregulatory immune signals. This is the case of the CD47 protein, a marker of "self" and "eat-me-not" that is expressed on all cell membranes [34]. In the work of Rodriguez et al. [32], the attachment of "self" peptides computationally designed from human CD47 protein onto polystyrene NPs achieved a delayed macrophage-mediated clearance of NPs in mice. In addition, this increased the circulation time of the NPs and enhanced the drug delivery to lung adenocarcinoma xenografts. Likewise, Hu et al. [33] coated PLGA NPs with a red blood cell (RBC)-membrane shell. These RBC-based polymeric NPs also showed a longer circulation half-life and sustained in vivo drug release compared with that achieved by using PEG-coated NPs. The coating with specific "self" molecules can also be used for the opposite reason, i.e., to induce an immune system activation by coating NPs with endogenous danger-associated molecular patterns. As an example, Aldossari et al. [35] coated AgNPs with high-density lipoprotein (HDL), which is recognized by scavenger receptors (SR-B1) expressed by macrophages. Once administered to mice, HDL-coated AgNPs provoked the recruitment of inflammatory cells, whereas SR-B1-deficient mice showed reduced cell recruitment. This strategy allowed the antimicrobial activity of AgNPs to be enhanced by targeting delivery. This indicates how important the NP coating is to escape the immune system, where a large body of knowledge has been developed to allow NPs to serve as drug delivery vehicles.

#### *2.2. When NPs Are Detected by the Immune System and Tolerated*

When the immune system detects NPs, they can be tolerated or induce an activation. Being tolerated means that NPs are silently removed, without inducing inflammation, as if they were protein aggregates, apoptotic cells, or cellular debris. This can be controlled mainly by modulating NP size, surface charge, and hydrophobicity/hydrophilicity of their surface [13,16,17,32,33], where small sizes, hydrophilicity, and negative surface charges often result in tolerable NPs [36]. In general, one can say that NPs below 4–6 nm can pass undetected and undergo rapid renal clearance after i.v. administration [37]. As the NP diameter increases, NPs become the target of the different immune cells. NPs of viruslike size (a few tens to a few hundreds of nm) can be endocytosed without triggering inflammation [12]. Larger objects, of micrometric size, like bacteria, are phagocytosed, while for sizes larger than 10–20 microns, objects are encapsulated [38].

This has critical consequences for the biodistribution of NPs inside the body. NPs are transported and accumulated in different organs depending on the administration route, their physicochemical properties, and their detection by the immune system. Accordingly, after intravenous (i.v.) injection, common NPs are often filtered in the liver by hepatocytes [17,18] or eventually Kupffer cells (the liver macrophages) depending on if they are detected or pass undetected by the immune system [31]. The first studies of biodistribution of colloidal particles (a few hundred nm) were reported in the 1970s in the *Journal of the Reticuloendothelial Society* (now *Journal of Leukocyte Biology*). Singer et al. [39] and Adlersberg et al. [40], by treating mice with i.v. and i.p. colloidal Au, found that after one hour, 90% of the administered dose was accumulated in the liver and 10% in the rest of the body (mainly kidneys). Subsequent histological studies with similar colloidal gold particles i.p. administered found them in the liver and lymph nodes primarily localized inside macrophages [41]. These results were later confirmed by numerous studies of the pharmacokinetic and biodistribution of different NPs. Sadauskas et al. [42], using AuNPs of different sizes (below 40 nm), showed that Kupffer cells were central in accumulating NPs once they entered the body. Similar results were also obtained with metal oxides, quantum dots, carbon nanostructures, etc. [43]. Yokel et al. [44] administered citrate capped nanoceria (5, 15, 30, and 55 nm) at 50 and 100 mg/kg bw i.v. into Sprague-Dawley rats and measured Ce content over time (1 h, 20 h, and 30 days). Remarkably, in all these works and many others, no inflammation or systemic injury was observed, except at larger doses (>100 mg/kg bw). Accordingly, we have observed by mass spectroscopy that after i.v. administration of albumin-conjugated nanoceria (CeO2) at low doses (0.1 mg/kg bw), twice a week during two weeks, in control and fibrotic Wistar rats, that most of the Ce is in the liver (84% of the administered dose after one hour and 75% eight weeks after administration) [26].

This non-inflammatory capture of NPs can be exploited for harnessing these phagocytic immune cells to transport NPs towards the target area, be it a wound, an infection, or a tumor. For such delivery, circulating monocytes have been proposed as a sort of Trojan Horse or Cellular Shuttle, since they naturally migrate from the blood to the sites of damage and disease. Hence, they can be loaded with NPs to be transported through the body [45,46]. Thus, Choi et al. [47] explored the use of monocytes containing AuNPs for transport into tumor regions for subsequent photothermal therapy. This study showed the phagocytosis of AuNPs by both monocytes and macrophages and their recruitment into the tumor. Oude-Engberink et al. [48] showed the accumulation of monocytes laden with iron oxide NPs (30 nm) in the affected cerebral sites in a rat model of experimental neuroinflammation. More recently, Moore et al. [49], using a microfluidic in vitro model, showed increased activity of monocytes/macrophages to transport NP across a confluent endothelial cell layer, advancing in the design of cellular shuttles loaded with NPs. This tolerated elimination of NPs may limit the dispersibility of NPs inside the body. However, the immune system is by itself an important therapeutic target where nanocarriers can efficiently transport drugs assisting immunotherapy.

#### *2.3. When NPs Are Detected by the Immune System and Not Tolerated*

Many reports show that NPs may induce harmful immune responses and toxicity. NPs can induce an inflammatory immune activation because of aggregation or dissolution or because they accidentally carry immune-activating moieties (such as endotoxin, detergents, allergens, or cationic molecules). These biological effects are rather independent of the composition, size, or shape of the individual NP, described as extrinsic factors of NP toxicity [4]. Similarly, the organization of molecular epitopes in a non-conventional form (upon adhesion of the NP surface) may generate new antigens or allergens. The activation of the immune system induced by NPs can be classified as follows:

#### 2.3.1. NP-Induced Oxidative Stress

The more universal inflammatory reaction to NPs corresponds to the most non-specific and rapid defense mechanism of macrophages, the overproduction of reactive oxygen species (ROS), which results in oxidative stress, responsible of lipid oxidation and DNA damage and, eventually, structural alterations, DNA mutations, and cell death. ROS refers to biogenic free radical molecules resulting from natural metabolism characterized by being highly oxidant. These free radicals are involved in different critical physiological processes, such as gene expression, signal transduction, growth regulation, and, significantly, inflammation, where high ROS concentrations are needed to sustain the energetic demands of a proinflammatory immune response [50].

Accordingly, independently of composition, large aggregates of TiO2 [51], Al2O3 [52], and Fe2O3 [53] NPs showed a similar capacity to increase oxidative stress. Moreover, the corrosion process of metallic NPs itself produces a high concentration of free radicals, which may trigger an inflammatory immune response [54–56]. These processes are often neglected in NPs made up of bulk non-biodegradable materials. However, biodegradation of Ag, Fe3O4, and CdSe/ZnS NPs due to enzymatic or hydrolytic activities in lysosomes [57,58] have been described. Even the physiological disintegration of AuNPs through oxidative etching by cysteine and chlorine has been described [59–61]. Similarly, carbon nanotubes (CNTs) have been observed to dissolve in vivo through enzymatic catalysis [62]. Subsequently, an increased number of reports has established relationships between observed inflammatory effects after NP exposure and NP disintegration [63–66]. Related to that, it is worth mentioning the works of Burello et al. [67] and Zhang et al. [55]. They developed theoretical and experimental models to predict the oxidative stress potential of oxide NPs by looking at their bandgap energy and their ability to perturb the intracellular redox state. Note that NP dissolution may become a source of toxic cations. For instance, in the early 2000s, the studies of Derfus et al. [68] and Kirchner et al. [69] showed that the released Cd ions were responsible for the intracellular oxidation and toxicity caused by CdSe NPs. Similar effects were found later when comparing the toxicity of Ni NP and ions as a function of time [70].

#### 2.3.2. When Phagocytosis Is Not Sufficient

When the immune system detects a foreign object, phagocytosis is the first mechanism for elimination that comes into play. However, when the object is too big (usually larger than 10–20 μm), rather than engulfing it, the immune cells start spreading on it to form a layer of cells that secludes the object from the rest of the tissue and initiate a chemical defense against the material that, if not non-biodegradable, is permanently kept secluded into a fibrous capsule or granuloma.

Historically, chronic inflammation has been observed in the case of penetration of non-biodegradable (persistent) large size (micrometric) particles in the lungs, as the wellknown cases of particle-induced granulomatosis such as silicosis and asbestosis [71]. This is because when a phagocytic cell fails to digest these particles, phagolysosomal rupture, the release of lysosomal enzymes and particles, and subsequent activation of the inflammasome and other cytoplasmic sensing mechanisms may happen, thereby triggering inflammation. This may lead, as the material persists, to chronic inflammation, permanent oxidative stress, tissue damage, and alterations that favor tumorigenesis in the long term. Brandwood et al. [72] found that murine macrophages phagocytosed inert carbon fiber-reinforced carbon particles up to 20 microns in diameter, but larger particles were not engulfed and became surrounded by aggregations of macrophages. This reaction may have pathological aspects; fibromas and granulomas are non-functional neo-tissues, similar to scars, that may hamper the organ functions and be active, i.e., growing, for a long time. In some instances, the reaction can be overtly pathological, as in the case of long fibrous materials. Accordingly, when Poland et al. [73] instilled high doses of multiwalled carbon nanotubes between the membranes lining the lungs and abdominal organs in mice, they found that long straight nanotubes caused inflammation and lesions in membrane cells similar to those leading to cancer, just like asbestos fibers [74]. Similarly, Ag nanospheres did not elicit any immune response or toxicity, while Ag nanowires can elicit a high inflammatory response, directly correlated to nanowire length, in murine macrophages [75]. The same effects were observed by Ji et al. [76] in THP-1 cells when comparing nanoceria nanorods and high aspect ratio nanoceria nanowires at high doses and aggregation states. This suggests that the needle-like shape of NPs is prone to provoke inflammation. It has been observed that macrophages engulfing needle-shaped crystals and fibers end up getting pierced by the needle-like structures and, consequently, start inflammation [77]. Parental NPs usually are never grown to these sizes, but uncontrolled aggregation can transform objects of tens of nm to tens of μm.

#### 2.3.3. NPs, Intendedly or Accidentally, Can Display Antigens, Allergens, or Toxins

The unintended or accidental absorption of biomolecules onto the NP surface may be a cause of concern. NPs may associate with specific bio-molecules, toxic by-standers, or pollutants, and present them to the immune system in an ordered pattern, thereby mimicking microorganisms and triggering the innate immune reaction of the host.

It is important to note that NPs have a strong tendency to adsorb many different molecules (hetero-aggregation) at their surface due to their high surface energy. Consequently, they are usually surrounded by a molecular coating, either provided intentionally (NP functionalization) and/or spontaneously by molecules present in the environment, forming the NP biocorona. These coatings also take part in the NP morphology and functions. The consequences of this are diverse; NPs can be good molecular aggregators and substrates for molecules to be presented to the immune system.

Among essential immunoactive biomolecules present in the environment, bacterial endotoxin is one of the most common and abundant. Endotoxins (also known as lipopolysaccharides (LPSs)) are large molecules present in the outer membrane of Gram-negative bacteria, able to elicit strong innate/inflammatory immune responses. Endotoxin is a ubiquitous environmental contaminant and can be present in all chemicals and glassware used in laboratories, even after sterilization (depyrogenation is needed to get rid of it). The presence of endotoxin, if not recognized, can be responsible for many of the in vitro and in vivo effects attributed to NPs [78]. Our study [79] showed that the endotoxin present on AuNPs turned those NPs from inactive to highly inflammatory and able to induce secretion of IL-1β in human primary monocytes. This could be an underlying factor in inflammatory responses and toxic effects associated with other metallic NPs and carbon nanomaterials [80,81]. Hence, special attention is needed to avoid endotoxin contamination when preparing NPs, which includes working in endotoxin-free conditions and glassware depyrogenation [82].

In other cases, the toxic ingredient may come from the formulation or derived chemicals employed during NPs preparation. If the synthesis process does not involve proper purification steps, the use of such NP samples may entail deleterious responses due to excess surfactants or unreacted precursors. This is the case of PEI molecules, a common NP stabilizer to enhance NP endocytosis, but with safety concerns due to the attachment to the negatively charged cell membranes that modify permeability and compromise viability [83]. Indeed, it is well-established that positively charged macromolecules can cause higher

toxicity and immune activation than their neutral or negatively charged counterparts, as in the case of monolayer-coated silicon nanoparticles [84]. Similar is the case of amphipathic molecules such as cetyltrimethyl ammonium bromide (CTAB), employed in preparing Au nanorods [85], which act as a detergent to lyse cell membranes. Another example is in the work of Dowding et al. [86]. These authors prepared different nanoceria NPs using the identical precursor (cerium nitrate hexahydrate) through a similar wet chemical process but using other bases: NH4OH, which yields negatively charged nanoceria, or hexamethylenetetramine (HMTA), which yields positively charged nanoceria at neutral pH. Results showed that HMTA-nanoceria NPs were readily taken into endothelial cells and reduced cell viability at a 10-fold lower concentration than the other NPs, which showed no toxicity.

Another type of immunoactive molecule that NPs can adsorb is allergens. This is unlikely to happen in the case of NPs since the concentration of allergens in the environment is very low, and the NP surface would be passivated before encountering them. However, it must be taken into account. Note that it has been reported that car combustion emission microparticles, when coated by pollen grains, enhance allergenic responses [87]. Radauer et al. [88] observed the formation of a stable allergen coating around NPs when exposed to different types of allergens (Der p1 and Bet v1), enhancing allergic responses against them. A recent review about the potential of NPs to trigger allergies via adsorption of allergens can be found in reference [89]. Here, it is essential to remark that allergy, understood as an anomalous immune response towards substances that are generally tolerated, has never been observed for engineered nanomaterials per se.

Regarding immune effects induced by biomolecules adsorbed on the NP surface, another possible source of inflammation comes from the potential modification of the structure of proteins upon adsorption at the NP surface [90]. Lynch et al. [91] pointed out how partial protein misfolding at the NP surface may result in the exposure of protein fractions usually buried in the core of the native structure. These cryptic epitopes may be recognized by immune cells and trigger inappropriate defensive reactions. Accordingly, in the work of Falagan [92], such modifications of the adsorbed proteins structure have been indicated as responsible for the long-term toxicity observed after a single low-dose exposure of AuNPs.

#### 2.3.4. NPs Presenting Vaccine Antigens and Working as Vaccine Adjuvants

Regarding the intentional use of NPs for vaccination, conjugation of antigens to NPs can help both attain a more efficient presentation of poorly immunogenic soluble antigens and provide an adjuvant effect targeted to innate immunity (the NP as a foreign agent) [12]. An interesting example is the development of AuNP-based virus-like particles (VLPs), where the NP replaces the virus core, which scaffolds the proteic capsid structure [93]. Typically, capsid proteins need the highly negatively charged dense core of DNA/RNA to self-assemble properly. This core can be replaced by dense and highly negatively charged AuNPs. Nikura et al. [93] demonstrated that the size and shape of AuNP-VLPs allowed for shaping of the in vitro and in vivo immune response in terms of the production of antibodies against West Nile virus. This implies that by modulating the NP size and shape, and consequently the arrangement of viral proteins on the NP surface, it could be possible to obtain highly effective and efficient vaccines. NPs can also be employed as vaccine adjuvants by exploiting their capacity to target and modulate the activity of innate immune cells. For a long time, vaccines were prepared by precipitation of antigens within some matrix, initially bread crumbs (in the XIX century), and currently alum powder, where the antigens are absorbed, forming aggregates that vary in size from 1 to 20 μm, acting as an antigen depot [94]. In this way, slow release of antigens is achieved, prolonging antigen presence, improving its processing and presentation. Other NP aggregates have been used as adjuvants. Skarastina et al. [95] used silica NPs (10–20 nm) as adjuvants for the hepatitis B vaccine in a mouse model. The monodisperse silica NPs formed heterogeneous aggregates larger than 1 μm after formulation, resulting in the same IgG2a/IgG1 ratios as in the case of immunization with alum as an adjuvant. Other nanostructure used as

vaccine adjuvants are nano-sized emulsions (sometimes called lipidic NPs). This is the case of the oil-in-water MF59 emulsion, which is used as an adjuvant, mainly for influenza vaccines (Flaud®, Novartis) and has been licensed in more than 20 countries. The MF59 adjuvant allows for significant cross-reactivity against viral strains and reduces antigen concentration to 50–200-fold lower doses [96].

The induction of inflammatory responses with non-pathogenic triggers has also been proposed as a preventive approach against exposure to unknown pathogens. Behind this concept, preventive activation of the innate immunity, there is the capacity of innate immune cells to develop a different response to a challenge as a consequence of previous contact with a different threat, a capacity known as innate memory [97,98]. All innate immune cells are able to develop a long lasting memory, despite their short lifespan in circulation. The reason lies mainly on the fact that the precursors in the bone marrow can be primed by a trigger and generate circulating immune cells with a different capacity to react against threats, as currently shown in monocytes/macrophages. Thus, after having previously experienced an inflammatory activation, the innate immune system becomes more efficient in preventing the rooting of newly incoming pathogens. For instance, this has been observed in the case of the administration of bacille calmetteguerin (BCG), the vaccine for tuberculosis, which increases resistance to other diseases [99]. The generation of innate memory thus represents an alternative, or better a complement, to the highly specific adaptive memory induced by vaccines. The strategy of innate memory induction leads to outcomes (enhanced protection) that have advantages (wider range of protection) and disadvantages (more unpredictable and less controllable side effects). Despite the controversy regarding safety, NPs can be used as adjuvants for the non-specific amplification of immune responses, and, even more, they can be excellent tools for generating or modulating innate memory [100]. In this regard, administration of AuNPs alone was observed to have little/no impact on the subsequent capacity of human monocytes to mount an innate/inflammatory response to a microbial challenge (LPS) [101]. However, the co-administration of AuNPs, or Fe3O4NPs, with memoryinducing microbial agents (e.g., LPS, BCG, muramyl dipeptide (MDP), Helicobacter pylori) led to a modulation of the innate memory response induced by the microbial agents depending on the priming stimulus and the NP type, shape, and size [102]. The implication is that vaccination with antigens and NPs could bring about a protective specific immunity based on adaptive immune memory and a non-specific innate memory induced by the antigen-NP combination.

The proinflammatory activation effects are listed in Table 2. In all these aspects, the uncontrolled proinflammatory activation of the immune system is, in principle, a common source of NP toxicity. In contrast, controlled activation can be employed for vaccination and other modes of defense against pathogens.



#### *2.4. When NPs Act as Enzymes and in This Way Can Modulate Immune Reactions*

Rare-earth oxide NPs have been found to be biocompatible antioxidants able to buffer excess ROS in physiological conditions, showing powerful anti-inflammatory effects. Mineral antioxidants may offer superior activity to currently available substances due to their enhanced bioavailability and stability, longer tissue residence time, and resistance to biological degradation [106]. These features can be exploited in many diseases based on excessive immune/inflammatory activation, such as autoimmune diseases, chronic inflammation, organ rejection, asthma and other allergic diseases, neurodegenerative pathologies (Alzheimer's disease, Parkinson's disease), and aging [106]. They have been described as engineered inorganic materials with enzyme-like activities, especially cerium oxide NPs, nanoceria [106,107]. Nanoceria has been reported to display superoxide dismutase (SOD)-like activity (conversion of superoxide anion into hydrogen peroxide and finally oxygen) [108], catalase-like activity (conversion of hydrogen peroxide into oxygen and water) [109,110], peroxidase-like activity (conversion of hydrogen peroxide into hydroxyl radicals) [111], as well as NO scavenging ability [103]. Consequently, nanoceria has been shown to safely down-regulate oxidative stress by scavenging the excess of ROS in diseases such as retinal degeneration [112,113], neurological disorders (including Alzheimer's disease, Parkinson's disease, and ALS) [114–116], ischemia [117], cardiopathies [118], diabetes [119], gastrointestinal inflammation [120], liver diseases [26–28], and cancer [121,122], as well as in regenerative medicine [123] and tissue engineering [124], with better performance than other antioxidant substances in both efficacy and efficiency. Interestingly, nanoceria become active at high ROS concentrations. Otherwise, at homeostatic ROS levels, the NPs become inactive. This is because several free radicals have to be simultaneously absorbed onto the NP surface in order to be recombined into non-radical adducts, a condition that only happens for high ROS concentrations. In other words, the ROS scavenging capacities of nanoceria are ROS concentration dependent. With time, these NPs dissolve into innocuous ions, which are excreted via the urinary route [19]. The solid NPs have been observed to be excreted through the hepatobiliary route [26,125].

This aspect is significantly different from the previous ones, where activation of the immune system results in inflammatory responses. In this case, the enzyme-like catalytic activity of rare earth NPs results in anti-inflammatory activity. The different observed responses are listed in Table 3.


**Table 3.** Immune responses to NP exposure.


**Table 3.** *Cont.*

#### **3. NP Evolution and Transformations in the Exposure Media**

The interaction between NPs and the immune system strongly depends on the conditions in which such interaction occurs (route of exposure, co-exposure with other agents) and on the characteristics of both NPs and the host immune system. Here we have presented the variety of immune responses to NPs and how these responses can help us design immuno-active and immune-benign NPs, which could either avoid immune recognition and activation in order to persist in the body long enough for completing their theranostic tasks or directly interact with immune cells for triggering inflammatory or anti-inflammatory responses as desired for therapeutic purposes. Indeed, the scientific community is still struggling with the apparent contradiction of similar materials being toxic and non-toxic (even beneficial) at the same time. This paradox can be attributed to undescribed effects of NP modifications during their dispersion in the working media, such as aggregation and corrosion. The main modifications NPs may suffer during their dispersion in different media are shown in Figure 3. Basically, NPs can be coated with molecules (e.g., hydrophilic polymers) to both pass undetected by the immune system and avoid aggregation (1). They can also be coated with soluble antigenic molecules to induce a response against them (2). In the opposite direction, when dispersed in physiological media, NPs can aggregate (3) and adsorb other molecules present in the medium (e.g., protein corona) (4) or both (5). In addition, depending on the core composition, NPs can be used as ROS scavengers, thereby down-regulating inflammatory responses (6), or they can dissolve and act as an ion reservoir that may increase the level of oxidative stress and generate an inflammatory response (7).

NPs have different ways to minimize their high surface energy, basically aggregation and corrosion. These are common phenomena in nature, widely studied by geochemistry, where a NP is an intermediate state between a micrometric particle and the dissolved ions. Thus, NPs may aggregate or associate with coating molecules in different media. They may also disintegrate through corrosion (defined as the chemical degradation of a solid material) and dissolution.

**Figure 3.** Intended or spontaneous NP modifications and their impact on immune responses.

Aggregation deserves particular attention. NPs are colloidally stable by repulsive electrostatic or steric forces or a combination of both. Aggregability depends on intrinsic NP parameters such as morphology, surface coating, and charge, and extrinsic parameters such as electrolyte concentration, pH, presence of organic matter, etc. Aggregation is common in physiological environments where NPs aggregate to submicrometric or micrometric sizes when not properly stabilized. To avoid it, one has to provide repulsive forces to the NP surface, either by electrostatic (high surface charge) or steric (entropic) means, usually provided by large soluble molecules associated with the NP surface; otherwise, they will aggregate and their unique physicochemical properties that arise at the nanoscale (quantum confinement, superparamagnetism, extreme catalytic activity, etc.) progressively lost. Note that aggregation entails modifications in terms of specific surface area, concentration, mobility, and dosing. Protein adsorption, the formation of a protein corona, is a particular aggregation phenomenon between NPs and proteins present in the dispersion media. It is a dynamic process in which, initially, proteins adsorb and desorb at the surface, followed by a set of re-organizational arrangements, which make this absorption more stable and finally irreversible [126,127]. This depends on NP size, surface state, type of protein, and protein-NP incubation media and conditions, where sophisticated functional patterns can be obtained [128]. The most straightforward strategy to cope with this issue is to passivate the NP surface in a controlled manner, e.g., by albuminization, PEGylation, or addition of PVP. These strategies usually decrease aggregation, even in high salt media, and the adsorption of microenvironmental biomolecules on the NP surface.

In addition to aggregation, chemical transformations, corrosion, and dissolution can also be a cause of immune activation via the alteration for the cellular redoxome, and the delivery of toxic cations. In this regard, the nanochemist or nanoengineer needs to control the redox potential (and the oxidative/reductive environment) where the NP will be stored, employed, and disposed of. In this regard, using NPs at their higher valence state is recommendable when possible [14] (passivating the surface with a continuous layer of oxide is sometimes an alternative). The chemical transformation and dissolution of NPs, which can cause immune activation or toxicity, is fundamental to determine NP fate and reduce its presence and persistence in the environment.

#### **4. Concluding Remarks**

After considering the different interactions NPs may have with the immune system, one can draw indications on how NPs have to be designed to control these interactions and

consequent responses. Indeed, while many NP functions can be attributed to their core structure, the surface coating defines much of their bioactivity. By controlling the nano-bio interface, NPs can be designed to be safe and innocuous or active and have therapeutic benefits. From the NP point of view, and for the nanochemist and the nanoengineer, the NP immunological properties can be summarized as depicted in Figure 4. The composition of the NP core determines its chemical potential and catalytic activity, while the surface coating largely determines its bioactivity. NPs can aggregate, either with other NPs or with macromolecules present in the physiological media (e.g., biocorona), or they can dissolve, being redox-active and acting as an ion reservoir, consequently increasing the levels of oxidative stress.

**Figure 4.** The NP properties that can impact the immune system.

In order to address the NP immune interactions, one has first to deal with the instability of the NP surface, and it should be passivated before introduction into biological systems. Otherwise, they may spontaneously aggregate, resulting in objects of increased size and decreased dose. Polymeric coatings have traditionally been the most commonly employed materials for such purposes. However, this surface engineering is sometimes costly and involves multi-step synthesis approaches, sometimes in the organic phase. One simple and effective solution could be to promote NP solubility in physiological media by prealbuminization during the preparation process [118,119,128], a similar approach employed by Abraxane®, one of the first approved nanomedicines [30]. In addition to providing colloidal stability and avoiding opsonization, NP coatings can be designed to directly interact with the immune system, such as CD47 [32] for avoiding complement activation, LPS for inducing innate/inflammatory activation [79], or viral proteins for vaccination [95]. Finally, NPs that belong to the family of natural antioxidants, such as nanoceria that catalytically scavenge free radicals (ROS in the context of inflammation), provide powerful immunomodulatory effects.

Thus, by mainly playing with surface characteristics, it is possible to adjust the NP physicochemical characteristics (aggregation, surface display of given biomolecules, chemical stability) and consequently their modes of interaction with the immune system.

**Author Contributions:** V.P., L.M.E., E.C., P.I. and D.B. wrote the manuscript. L.M.E. also prepared the figures and managed references together with E.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the EU Commission H2020 project PANDORA (GA 671881; to D.B., P.I. and V.P.). Additional funds were provided by the EU Commission H2020 project ENDO-NANO (GA 812661; to P.I. and D.B.), the Italian MIUR InterOmics Flagship projects MEMORAT and MAME (to D.B. and P.I.), the Italian MIUR/PRIN-20173ZECCM (to P.I.), the CAS President's International Fellowship Programme (PIFI; award 2020VBA0028; to D.B.), Spanish Ministerio de Ciencia, Innovación y Universidades (MCIU) (RTI2018-099965-B-I00, AEI/FEDER, UE), and Generalitat de Catalunya (2017-SGR-1431) (V.P.).

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish.

#### **References**


## *Article* **Growth-Promoting Gold Nanoparticles Decrease Stress Responses in Arabidopsis Seedlings**

**Eleonora Ferrari 1, Francesco Barbero 2,3, Marti Busquets-Fité 4, Mirita Franz-Wachtel 5, Heinz-R. Köhler 6, Victor Puntes 2,7,8 and Birgit Kemmerling 1,\***


**Abstract:** The global economic success of man-made nanoscale materials has led to a higher production rate and diversification of emission sources in the environment. For these reasons, novel nanosafety approaches to assess the environmental impact of engineered nanomaterials are required. While studying the potential toxicity of metal nanoparticles (NPs), we realized that gold nanoparticles (AuNPs) have a growth-promoting rather than a stress-inducing effect. In this study we established stable short- and long-term exposition systems for testing plant responses to NPs. Exposure of plants to moderate concentrations of AuNPs resulted in enhanced growth of the plants with longer primary roots, more and longer lateral roots and increased rosette diameter, and reduced oxidative stress responses elicited by the immune-stimulatory PAMP flg22. Our data did not reveal any detrimental effects of AuNPs on plants but clearly showed positive effects on growth, presumably by their protective influence on oxidative stress responses. Differential transcriptomics and proteomics analyses revealed that oxidative stress responses are downregulated whereas growth-promoting genes/proteins are upregulated. These omics datasets after AuNP exposure can now be exploited to study the underlying molecular mechanisms of AuNP-induced growth-promotion.

**Keywords:** engineered nanomaterial (ENM); nanoparticle (NP); gold nanoparticle (AuNP); plant; *Arabidopsis thaliana*; plant growth; stress response; transcriptomics; proteomics

#### **1. Introduction**

Engineered nanomaterials (ENMs) are distributed into the environment in drastically increasing amounts, yet knowledge on the resulting effects of ENMs on the environment is limited [1,2]. Although naturally occurring nanomaterials have always existed, in the last decade the emission rate of anthropogenic nanoparticles (NPs), intentionally or unintentionally released, has been continuously rising [3,4]. For these reasons, novel nanosafety approaches to assess the environmental impact of ENMs are required [5].

Gold ENMs are used worldwide in various fields, including medicine, biology, chemistry, physics, electronics, and cosmetics [6–8]. The unique optical and electrochemical properties of AuNPs [9], as well as their accessibility for various surface functionalizations [10], have been exploited in many applications ranging from diagnostics and cancer therapy [11] to industrial catalysis [12] and water purification [13]. Furthermore, the latest developments in nanotechnology have opened up new opportunities in the food safety industry [14,15] and agronomy [16]. Several studies have shown that biosynthesized AuNPs

**Citation:** Ferrari, E.; Barbero, F.; Busquets-Fité, M.; Franz-Wachtel, M.; Köhler, H.-R.; Puntes, V.; Kemmerling, B. Growth-Promoting Gold Nanoparticles Decrease Stress Responses in Arabidopsis Seedlings. *Nanomaterials* **2021**, *11*, 3161. https:// doi.org/10.3390/nano11123161

Academic Editor: Anne Kahru

Received: 19 October 2021 Accepted: 18 November 2021 Published: 23 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

have larvicidal and nematicidal effects in crop cultivation without adversely affecting their growth and development [17,18], implicating the application of more agro-ecological practices in the future.

The increasing production, use, and disposal of ENMs has translated into a higher and uncontrolled release of such materials in the environment [19]. As a result of the unique size-dependent physicochemical properties of NMs, such as higher reactivity than their respective bulk materials, concerns about potential adverse effects have arisen [20,21]. In order to assess the risk of ENMs on biosystems, a comprehensive characterization of the material is required, as their physicochemical properties and behavior after release into the environment depend on their chemical composition, size, shape, and surface [22,23]. Stable colloidal ENMs can be obtained by steric or electrostatic surface functionalization with coating materials [24,25]. Despite the stabilizing function of such surface modifications, it must be taken into account that the chemistry of coatings can change under natural conditions, affecting the properties of ENMs and their biocompatibility [26,27].

To study the risk of ENMs, ecological effects should be studied under natural conditions, but model systems also need to be optimized for stable, non-toxic assay conditions and adequate characterization of ENMs. As ENMs can change their properties in different environments, characterization is necessary for the starting material as well as under experimental conditions.

Positive, negative, and no effects of ENMs on human health, animals, and plants have been described. In vitro toxicity studies of AuNPs did not find detectable changes in the concentration of inflammatory markers in either humans or animals [28–31]. On the other hand, AuNPs were associated with dose-dependent imbalances of oxidative stress levels in vitro [32,33], with higher doses of particles responsible for initial oxidative cell damage [34,35]. Many studies have reported that the effects of AuNPs on living organisms are strictly related not only to their concentration, but especially to their size and the physicochemical properties of the coatings [36,37]. Environmental AuNP concentration predicted by screening models are 0.14 μg L−<sup>1</sup> in natural waters and 5.99 μg kg−<sup>1</sup> in soils [38]. Under lab conditions, for these concentrations no measurable physiological effect on plants have been reported [39]. To evaluate whether AuNPs have any effect on plants, higher concentrations are tested to assess the maximum potential risk that could be caused by AuNP accumulation. The effects of AuNPs on plants were reviewed by Siddiqi and Husen [40]. Overall, despite a few contradictory studies, AuNPs have been found to have detrimental effects at high concentrations (≥100 mg/L) [41], with particles with a diameter below 5 nm showing increasing toxicity [42–44]. By contrast, lower concentrations of AuNPs can enhance seed germination and chlorophyll content, and improve growth and productivity in several crops and model plants under laboratory conditions [45–47]. Enhanced and reduced toxic effects and stress responses, such as reactive oxygen species (ROS) production and activation of immune responses, have both been reported in plants after AuNP treatment [40,47–50]. Furthermore, growth-promoting effects are common for nanofertilizers such as ZnNPs, for example, but were also observed for inert AuNPs [18,47]. AuNPs are widely inert and the special physicochemical properties of NPs can be studied without physiological side effects of the bulk material [51]. The mechanism underlying AuNP-induced growth-promotion is not yet understood. Understanding how NPs influence plant growth could be useful for improving crop yield in the future [52].

AuNP uptake is controversially discussed in the community, with studies showing uptake and even transport within the plant and others ruling it out [53–56]. The pore size of cell walls has been determined to be approximately 3–6 nm, a size that precludes the uptake of larger molecules, but some properties of NPs, such as surface charge, may induce morphological changes in the cell wall, thereby affecting pore size and uptake rate [57,58]. Though many plants have been tested for their ability to take up various kinds of ENMs, our knowledge is still limited and many aspects remain elusive. The mechanisms underlying the physiological effects caused by AuNPs in or outside the plant cells will be interesting to elucidate in the future.

Transcriptomics and proteomics studies after exposure to multi-walled carbon nanotubes (MWCNT), titanium dioxide (TiO2), cerium dioxide (CeO2), and silver (Ag) NPs have been reported in Arabidopsis plants [59–62]. Furthermore, Simon et al. [63] performed a transcriptome sequencing study on the eukaryotic green alga *Chlamydomonas reinhardtii* after treatment with Ag, TiO2, zinc oxide (ZnO) NPs, and quantum dots (QDs). Conversely, the transcriptome and proteome changes in plants exposed to AuNPs have not yet been adequately studied.

Here, we show that well-characterized AuNPs stabilized with sodium citrate and tannic acid (SCTA) are stable, sterilizable, and functional in promoting the growth of Arabidopsis seedlings and do not show any negative effects at moderate concentration levels. Stress responses are downregulated after AuNP exposure at the ROS burst level and also on the transcriptome and proteome level. We studied transcriptome and proteome changes after AuNP-SCTA treatment, and those omics data revealed candidate genes/proteins that could explain the growth-promoting effect of AuNPs on a molecular level.

#### **2. Materials and Methods**

#### *2.1. AuNP Synthesis*

Aqueous dispersions of citrate-stabilized AuNPs were synthesized following two kinetically controlled seeded growth approaches as reported by Bastús et al. [64] and Piella et al. [65]. Tetrachloroauric (III) acid trihydrate (99.9% purity), sodium citrate tribasic dihydrate (99%), and tannic acid were purchased from Sigma-Aldrich (Madrid, Spain). Briefly, 150 mL of sodium citrate (SC) aqueous solution (2.2 mM) were brought to a boil in a three-neck flask under reflux; subsequently, 1 mL of 25 mM chloroauric acid (HAuCl4) was injected into the citrate solution. After few minutes the solution became reddish, indicative of AuNP formation (~10 nm, seeds). Afterwards, different sequential steps of growth, consisting of sample dilution plus further addition of gold precursor, led to the desired AuNP size. In the second method, the main difference was the addition of traces of tannic acid (TA) (200 μM) as a co-reducer and an increase in the starting pH to produce highly monodispersed and stable AuNPs. Several batches of the AuNPs were synthesized with and without the addition of TA. All batches presented very similar features and produced similar results in the assays described. The AuNPs were synthesized by the Catalan Institute of Nanoscience and Nanotechnology and purchased from Applied Nanoparticles SL.

The AuNP dispersions were concentrated by Amicon® Ultra Centrifugal Filters (100 kDa) (Merck, Darmstadt, Germany).

#### *2.2. AuNP Characterization*

#### 2.2.1. Size Determination by Electron Microscopy

The diameter of the synthesized AuNPs was measured by analysis of images obtained by scanning electron microscopy (SEM) with FEI Magellan XHR SEM (FEI, Hillsboro, OR, USA) in transmission mode (STEM) operated at 20 kV. Samples were prepared by drop-casting 3 μL of the NP dispersion onto a carbon-coated copper TEM grid and left to dry under mild vacuum. To prevent aggregation of the NPs during the drying procedure, they were previously conjugated with 55 kDa polyvinylpyrrolidone (PVP) (Sigma-Aldrich, Madrid, Spain). More than 500 particles from different regions of the grid were measured.

#### 2.2.2. UV-Vis Spectroscopy

The UV-Vis absorption spectra of AuNPs are due to the collective oscillation of their metallic surface electrons, called localized surface plasmon resonance (LSPR). The LSPR profile and maximum position strictly depend on the material, shape, and size of the NPs, as well as the refractive index of the solvent and the vicinity of the NP surfaces. The LSPR profile is highly sensitive to NP aggregation. At inter-particle distances that are less than their diameter, the NP near-field electromagnetic coupling applies, leading to significant UV-Vis spectra changes that translate to a LSPR red-shift and/or to the occurrence of a

second peak at a higher wavelength. For large aggregates, an increase in the baseline can be observed [66–68].

UV-Vis spectra were acquired with a Shimadzu UV-2400 spectrophotometer (SSI, Kyoto, Japan). One mL of sample was placed in a plastic cuvette and analyses were performed at time zero or over time in the 300–800 nm range at room temperature. In the case of solidified media, samples were poured into the cuvette prior to jellification. MilliQ water or <sup>1</sup> <sup>2</sup> Murashige and Skoog (MS) agar (Duchefa, Haarlem, The Netherlands) were taken as reference for the different samples.

#### 2.2.3. Size and Zeta Potential Measurements

Laser doppler velocimetry and dynamic light scattering were used to determine the Z potential and the hydrodynamic diameter of the AuNPs, respectively, employing a Malvern Zetasizer Nano ZS instrument (light source wavelength at 638.2 nm; detector at a fixed scattering angle of 173◦) (Malvern Panalytical Ltd., Malvern, UK). Measurements were performed at 25 ◦C. Diameters were reported as Z-average and polydispersity index (PDI) calculated by cumulative analysis.

#### 2.2.4. Au Quantification

Samples were digested with aqua regia (1:3 HNO3 (70%):HCl (37%)) for 24 h and then diluted with MilliQ water to be further analyzed by induced coupled plasma-mass spectroscopy (ICP-MS) using an ICP-MS NexION 300 from Perkin Elmer (Shelton, CT, USA).

#### *2.3. AuNP Sterilization*

AuNP suspensions were sterilized by filter sterilization with cellulose mixed ester (CME) and polyethersulfone (PES) filters (Carl Roth, Karlsruhe, Germany), both with a pore size of 0.2 μm, according to the manufacturer's protocol. UV-Vis spectra of AuNPs before and after filtration were acquired as previously mentioned.

#### *2.4. Plant Growth Conditions and Plant NP Exposition*

#### 2.4.1. Seed Sterilization

*Arabidopsis thaliana* ecotype Columbia 0 (Col-0) seeds were surface sterilized by chlorine gas treatment in a desiccator containing 50 mL of 12% sodium hypochlorite and 2 mL of 37% HCl for 4 h. The chemicals were purchased from Merck (Darmstadt, Germany). The seeds were dried in a laminar air flow sterile bench for 30 min. The seeds were grown on agar-solidified medium or in hydroponic cultures.

#### 2.4.2. Plant Growth Conditions

All plants were grown in long day conditions with 16 h light, 8 h dark, 22 ◦C, 110 μmol m−<sup>2</sup> s−<sup>1</sup> light, and 60% relative humidity.

#### 2.4.3. NP Exposure in Agar-Solidified Medium

Sterilized seeds were sown on agar plates containing <sup>1</sup> <sup>2</sup> MS plant medium and stratified at 4 ◦C for 2 days in the dark. Afterwards, plates were incubated for 7 days under controlled long day conditions. Plates were placed vertically to allow root growth along the agar surface. The reducing and stabilizing agents sodium citrate and tannic acid (SCTA) or AuNP-SCTA were mixed with the medium at the indicated concentrations before jellification. Physicochemical characterization of AuNP-SCTA dispersed in <sup>1</sup> <sup>2</sup> MS agar showed high colloidal stability up to 3 weeks, allowing for 1 week exposure experiments.

#### 2.4.4. NP Exposure in Hydroponic Culture

Sterilized seeds were sown on a thin layer of <sup>1</sup> <sup>2</sup> MS agar medium and stratified at 4 ◦C for 2 days in the dark. Subsequently, the seeds were germinated and grown for 2 weeks under controlled long day conditions. The seedling roots grew through the agar into liquid <sup>1</sup> <sup>2</sup> MS plant medium. After 2 weeks, SCTA or AuNP-SCTA were mixed in the indicated concentration with the <sup>1</sup> <sup>2</sup> MS medium and were incubated for 6 h. Since UV-Vis spectroscopic analyses of AuNP-SCTA dispersed in <sup>1</sup> <sup>2</sup> MS revealed no changes in the shape of the spectrum at 6 h, whereas a typical aggregation profile was shown after 9 h, this interval was chosen for short-term experiments.

#### *2.5. Physiological Effects*

*Arabidopsis thaliana* seedlings were grown in agar-solidified medium, as described previously, with SCTA or AuNP-SCTA to a final concentration of 10 mg/L. Photographs of 7 day-old seedlings were taken. Growth parameters, i.e., rosette diameter, primary root length, and lateral root length, were measured using the software ImageJ. The lateral root number was determined by counting the number of lateral roots per seedling with 20 seedlings being analyzed for each individual parameter (n = 20).

#### *2.6. Statistical Analysis*

Statistical significance between groups was evaluated using one-way ANOVA combined with Tukey's honest significant difference (HSD) test. FOX assay data were tested with a two-way nested ANOVA followed by Dunnett's post-hoc test; data were normally distributed (Shapiro–Wilk test) and showed homogeneity of variances (Levene's test). Significant differences are indicated with different letters (*p* < 0.01). Statistical evaluations were performed using JMP (version 15.0.0, Heidelberg, Germany) software.

#### *2.7. Detection of Immune-Related Responses*

#### 2.7.1. Oxidative Burst

Production of reactive oxygen species (ROS) was measured in a luminol-based assay using a microplate luminometer (CentroPRO LB 962; Berthold Technologies, Bad Wildbad, Germany) as described by Albert et al. [69]. The elicitor flg22 (final concentration 100 nM) was used in the assay as positive control. The horseradish peroxidase, in the presence of ROS, catalyzed the oxidation of luminol to 3-aminophthalate with emission of light at 428 nm. The monitored oxidative burst was measured as emitted light and recorded as relative light units (RLU). The ROS burst was monitored for 30 min for 3 plants per treatment and three leaf pieces per plant (n = 9).

#### 2.7.2. FOX Assay

The level of lipid hydroperoxides (LOOHs) was assessed with the modified colorimetric ferrous oxidation xylenol orange (FOX) assay as described by Hermes-Lima et al. [70] and adjusted by Schmieg et al. [71]. Leaves of 5 week-old *A. thaliana* plants were cut into square pieces (about 2 mm2) and left to equilibrate overnight in milliQ water. Then the leaf pieces were elicited for 30 min with flg22 (100 nM) or the tested compounds and immediately stored at −80 ◦C. Three plants and three leaf pieces per plant (n = 9) were used for each sample. Samples were homogenized in ice-cold HPLC-grade methanol in a 1:15 ratio, and 30 μL of the supernatant was used in the reaction mixture. Cumene hydroperoxide equivalents (CHPequiv./mg wet weight) were calculated using the following equation:

$$\begin{array}{l} \mathsf{CHPPE} = \frac{\begin{array}{c} \mathsf{ABSS70} \\ \mathsf{ABSS70}+\mathsf{CHP} \end{array}}{} \times Volume\,\mathsf{CHP} \times \frac{\begin{array}{c} \mathsf{Total\ Volume} \\ \text{Sample\ Volume} \end{array}}{} \times \begin{array}{c} \mathsf{Distance} \\ \text{Alternative\end{array}} \times \begin{array}{c} \mathsf{Dilution\ Factor} \\ \text{A} \end{array} \end{array}$$

#### *2.8. Transciptomics*

#### 2.8.1. RNA Extraction

RNA was extracted from 100 mg of Arabidopsis seedling roots treated for 6 h and 7 d with 10 mg/L Au-SCTA or SCTA (SC 2.2 mM; TA 200 μM) in triplicate using the RNeasy Plant Mini Kit (QIAGEN, Hilden, Germany) followed by on-column DNA digestion with the RNase-Free DNase Set (QIAGEN, Hilden, Germany). The total RNA concentration, RNA Integrity Number (RIN) value, and rRNA ratio (28S/18S) were evaluated using Agilent2100 Bionalyzer (RNA 6000 Nano Kit; Agilent, Waldbronn, Germany).

#### 2.8.2. Transcriptome Sequencing Analysis

The BGI Group (Shenzhen, China) performed the total transcriptome sequencing (RNA-Seq) analysis. Samples were sequenced on the Illumina HiSeq platform. The internal software SOAPnuke v1.5.2 was used to filter low-quality reads, reads with adaptors, or those containing more than 5% of unknown bases (N). Genome mapping of clean reads was performed using HISAT v2.0.4 (Hierarchical Indexing for Spliced Alignment of Transcripts) software [72]. The assembler of RNA-Seq alignments into potential transcripts StringTie v1.0.4 was used to reconstruct transcripts [73]. Cuffcompare, a tool of Cufflinks [74], was used to identify novel transcripts by comparing reconstructed transcripts with genome reference annotation information. The coding ability of those new transcripts was predicted using CPC v0.9-r2 [75]. After novel transcript detection, novel coding transcripts were merged with reference transcripts to get a complete reference and clean reads were mapped to it using Bowtie2 v2.2.5 [76]. For each sample, the gene expression level was calculated with RSEM, a software package for estimating gene and isoform expression levels from RNA-Seq data [77]. Differentially expressed genes (DEGs) were detected with the nonparametric approach NOIseq method (parameters: fold change ≥ 2.00 and probability ≥ 0.8) as described by Tarazona et al. [78].

#### *2.9. Mass Spectrometry Analysis*

#### 2.9.1. Total Protein Extraction

For total protein extraction from seedlings, 100 mg of material were ground in liquid nitrogen and mixed in a ratio of 1:3 with ice-cold extraction buffer (10% glycerol, 150 mM Tris/HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, 10 mM DTT, 0.2% Nonidet P-40, 2% PVPP, 1 tablet of proteinase inhibitor cocktail (Roche, Mannheim, Germany) per 10 mL solution). Protein extraction was performed on a rotor at 4 ◦C for 1 h and the extract was purified by a centrifuging at 4 ◦C, 5000× *g* for 20 min. The supernatant was then transferred through a one-layer Miracloth (Merck, Darmstadt, Germany) in a fresh pre-chilled 1.5 mL tube on ice.

#### 2.9.2. NanoLC-MS/MS Analysis

The Proteome Center Tübingen performed the nanoscale liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) on total protein extracts as described.

Proteins were purified in a 12% NUPAGE Novex Bis-Tris Gel (Invitrogen, Karlsruhe, Germany) for 10 min at 200 V and stained with Colloidal Blue Staining Kit (Invitrogen, Karlsruhe, Germany). In-gel digestion of proteins was performed as previously described [79]. Extracted peptides were first desalted and then labeled using C18 StageTips [80] as described elsewhere [81]. Samples were labeled with dimethyl "light" ((CH3)2) and dimethyl "intermediate" ((CH1D2)2). Complete incorporation levels of the dimethyl labels were achieved in all cases.

Eluted peptides were mixed in a 1:1 ratio according to the measured protein amounts. The analysis of the peptide mixture was performed on an Easy-nLC 1200 system coupled to an LTQ Orbitrap Elite or a QExactive HF mass spectrometer (all Thermo Fisher Scientific) as described elsewhere [82] with slight modifications: Peptides were injected onto the column in HPLC solvent A (0.1% formic acid) at a flow rate of 500 nL/min and subsequently eluted with a 227 min (Orbitrap Elite) or 127 min (QExactive HF) gradient of 10–33–50–90% HPLC solvent B (80% ACN in 0.1% formic acid). During peptide elution the flow rate was kept constant at 200 nL/min.

In each scan cycle, the 15 (Orbitrap Elite) or 12 (Q Exactive HF) most intense precursor ions were sequentially fragmented using collision-induced dissociation (CID) and higher energy collisional dissociation (HCD) fragmentation, respectively. In all measurements, sequenced precursor masses were excluded from further selection for 60 (Orbitrap Elite)

or 30 s (Q Exactive HF). The target values for MS/MS fragmentation were 5000 and <sup>10</sup><sup>5</sup> charges, and for the MS scan 10<sup>6</sup> and 3 × <sup>10</sup><sup>6</sup> charges.

#### 2.9.3. MS Data Processing

The MS data were processed with MaxQuant software suite v1.5.2.8 and v1.6.3.4 (Cox and Mann 2008), respectively. A database search was performed using the Andromeda search engine [83], which is a module of the MaxQuant. MS/MS spectra were searched against an *Arabidopsis thaliana* database obtained from Uniprot, and a database consisting of 285 commonly observed contaminants. In the database search, full tryptic specificity was required and up to two missed cleavages were allowed. Protein N-terminal acetylation and oxidation of methionine were set as variable modifications. Initial precursor mass tolerance was set to 4.5 ppm and to 0.5 Da at the MS/MS level (CID fragmentation), or 20 ppm (HCD fragmentation). Peptide, protein, and modification site identifications were filtered using a target-decoy approach at a false discovery rate (FDR) set to 0.01 [84]. For protein group quantitation a minimum of two quantified peptides were required.

Perseus software (v1.6.1.3), a module from the MaxQuant suite [85], was used for calculation of the significance B (psigB) for each protein ratio with respect to the distance of the median of the distribution of all protein ratios as well as the intensities. All proteins with a fold change ≥2.00- and psigB <0.01 in a pairwise comparison were considered to be differentially expressed.

#### **3. Results**

#### *3.1. Physicochemical Characterization of AuNPs Dispersed in Plant Growth Media*

Two different types of gold nanoparticles (AuNPs) with an average diameter of about 12 nm were synthesized with the two seeded-growth methods reported by Bastús et al. [64] and Piella et al. [65], with the only difference being the addition of tannic acid (TA), which can interact with the NP surface, inferring higher colloidal stability. The physicochemical characterization of one of the used batches of AuNPs prepared in the presence of TA (AuNP-SCTA) is shown in Figure 1a,b, whereas the characterization of AuNP-SC is reported in Supplemental Figure S1a,b.

It is important to characterize the evolution of the AuNPs once dispersed in the working media to correctly correlate the pristine and the evolving NP features with the observed biological effects [86]. Thus, over time, physicochemical characterization of AuNP-SC and AuNP-SCTA in the used working media, i.e., <sup>1</sup> <sup>2</sup> MS and agar-solidified <sup>1</sup> <sup>2</sup> MS media, was performed. Once dispersed in <sup>1</sup> <sup>2</sup> MS, AuNP-SC underwent fast aggregation, pointed out by an immediate emergence of a second localized surface plasmon resonance (LSPR) peak at around 650 nm in the UV-Vis spectra (Supplemental Figure S1c) [87]. This aggregation was probably due to the increase in the ionic strength by mono- and divalent inorganic ions in the media ( <sup>1</sup> <sup>2</sup> MS has a salinity of 23 mM). The ions in the media can screen the negative charges provided by the SC present on the surface of the AuNPs, responsible for the electrostatic repulsion between particles [88,89].

By contrast, the UV-Vis spectra of AuNP-SCTA dispersed in <sup>1</sup> <sup>2</sup> MS showed no changes until 6 h of exposure. After 9 h, changes in the spectrum shape were observed, showing the start of a typical aggregation profile that led to complete aggregation after 15 h, detectable by a drastic change in the UV-Vis spectrum (Figure 1c) [87]. This result indicates that AuNP-SCTA had a good colloidal stability up to 6 h of exposure to the hydroponic medium, whereas after 9 h the NPs started to slowly aggregate. Therefore, in this study all experiments carried out in hydroponic cultures were short time exposures, in a time range of 6 h. The presence of TA was the only difference between the two types of AuNPs. Thus, we hypothesize that this organic molecule functions as a NP stabilizer, increasing the particle stability against salt-driven aggregation. TA confers an effective higher surface charge or partial steric stabilization, preventing NP aggregation. Regarding the observed aggregation of the AuNP-SCTA in <sup>1</sup> <sup>2</sup> MS after 9 h, it could be speculated that organic molecules (e.g., sucrose), present in the medium in excess compared to the NP

stabilizers, could progressively replace the NP stabilizers on the NP surface, conferring a negative effect on stabilization and supporting aggregation. However, further studies will be necessary to precisely understand the role and nature of TA in the stabilization of AuNP-SCTA.

**Figure 1.** Physicochemical characterization of AuNP-SCTA dispersed in H2O, <sup>1</sup> <sup>2</sup> MS, and <sup>1</sup> <sup>2</sup> MS agar. (**a**) Bright field—scanning transmission electron microscopy (STEM) of AuNP-SCTA. (**b**) NP average diameter measured by STEM; hydrodynamic diameters (in H2O) measured by dynamic light scattering, reported as Z average and poly dispersity index (PDI); Z potential of the AuNP-SCTA dispersed in H2O (pH 6.5, conductivity 0.85 mS/cm). (**c**) UV-Vis spectra of AuNP-SCTA dispersed in H2O (red) and over time (from time 0 to 15 h) in <sup>1</sup> <sup>2</sup> MS (black). (**d**) UV-Vis spectra of AuNP-SCTA dispersed in <sup>1</sup> <sup>2</sup> MS agar at time 0 (red dashed) and after 3 weeks of exposure (blue); representative photograph of Arabidopsis seedlings germinated and grown for 1 week on an agar plate containing 1 <sup>2</sup> MS and AuNPs, showing a typical reddish color of non-aggregated AuNPs-SCTA. Absorbance A in arbitrary units (a.u.). All experiments were repeated twice with similar results.

UV-Vis spectroscopy of AuNP-SCTA exposed to <sup>1</sup> <sup>2</sup> MS agar showed high stability at least up to 3 weeks (Figure 1d), allowing for long-term exposure experiments. Conversely, AuNP-SC dispersed in <sup>1</sup> <sup>2</sup> MS agar showed an initial aggregation that, unlike in <sup>1</sup> <sup>2</sup> MS, did not evolve over time (Supplemental Figure S1d), probably due to the interaction with agar molecules and the fast viscosity increase due to medium jellification as well as to the reduced number of particles (aggregation is directly proportional to concentration). Note that below 1010 NP mL−<sup>1</sup> the collision probability decreases to almost zero, so even if their surface is not passivated, NPs do not aggregate.

In light of these observations, the AuNP-SCTA were chosen for the following physiological and molecular studies, permitting all the experiments to be conducted with stable AuNPs, thereby allowing for the correct NP size to be correlated with the observed effects on Arabidopsis.

Several AuNP-SCTA batches with very similar physicochemical features were produced and tested. The results of the physiological studies after AuNP-SCTA treatment were fully reproducible between the different batches, showing that the synthesis protocol is very robust and produced reliable results.

#### *3.2. AuNP-SCTA Sterilization*

In order to grow seedlings under sterile conditions and to discriminate the AuNP effects from the possible physiological and molecular changes induced in plants by microbial contaminants such as, e.g., (pathogenic) bacteria and fungi, the sterility of the colloidal solution is a fundamental requirement. To sterilize AuNP-SCTA solutions, physical filtration methods were chosen. Two different filter materials, i.e., cellulose mixed ester (CME) and polyethersulfone (PES), both with a pore size of 0.2 μm, were tested. To analyze possible changes in the particle concentration or their aggregation state, UV-Vis spectra before and after filtration were acquired (Figure 2). Both filtering procedures were effective in removing all contaminating microorganisms and allowed plant cultivation under sterile conditions. The CME filter, a standard hydrophilic membrane commonly used for a broad range of applications, was shown to significantly affect the amplitude of the spectra, revealing a drastically reduced AuNP-SCTA concentration. By contrast, the PES filter, a hydrophilic and low protein-binding membrane, did not change the spectra and therefore did not affect the concentration of AuNP-SCTA. No changes in the overall shape of the UV-Vis spectrum were observed, indicating that no alterations of the physicochemical properties of the AuNPs occurred. Therefore, PES membranes were used in all subsequent experiments for NP sterilization.

**Figure 2.** UV-Vis spectra of AuNP-SCTA before and after sterilization with different filter materials. UV-Vis spectra of AuNP-SCTA before (red dashed) and after sterilization with PES (blue) and CME (black) filters were acquired with a Shimadzu UV-2400 spectrophotometer. Absorbance A in arbitrary units (a.u.). The experiment was repeated twice with similar results.

#### *3.3. Physiological Effects of AuNPs*

Although gold (Au) can be present in the environment from natural sources, in the last decade the increased use and disposal of AuNPs has affected the level of this chemical element in soil and water [1]. Although many studies on the accumulation and physiological effects of Au in various plant species have been conducted [90], a comprehensive investigation on AuNP fate and action after their release into plant growth media and their effects on plants at the physiological, transcriptomic, and proteomic level is missing. In the present study, *Arabidopsis thaliana* was used as a model plant to investigate the effects of AuNPs on growth and development. As shown in Supplemental Figure S2b, AuNP-SCTA in a range from 0 to 20 mg/L affected Arabidopsis root growth in a dose-dependent manner with a maximal effect at 10 mg/L, whereas the NP stabilizer SCTA (SC 2.2 mM; TA 200 μM) did not affect the primary root length at any of the tested concentrations (Supplemental Figure S2a). For this reason, 10 mg/L AuNP-SCTA was chosen as the final concentration in all subsequent experiments.

Physiological analyses were performed in order to evaluate Arabidopsis responses to abiotic stress caused by AuNP-SCTA exposure. Seedlings, grown under controlled long day conditions, were harvested after 7 days, and representative parameters were recorded, i.e., primary root length, rosette diameter, number of lateral roots, and lateral root length. For each parameter, another set of plants was grown in the presence of SCTA (SC 2.2 mM; TA 200 μM) as a control.

Although SCTA did not affect plant growth and development, AuNP-SCTA had a positive influence on all parameters tested. Seedlings germinated and grown on AuNP-SCTA-containing medium developed a longer primary root, with an enhancement of 1.2 folds compared to control seedlings (Figure 3a). Furthermore, the lateral root number and length were positively affected upon AuNP-SCTA treatment, displaying, compared to the controls, an increase of 1.7- and 1.5-fold, respectively (Figure 3e,f). Shoot development was also influenced by AuNP-SCTA exposure in the same way as the root system (Figure 3c). The size of the rosette diameters was enhanced by 1.3-fold in comparison to the control seedlings. These data show that AuNP-SCTA is not acutely toxic to plants, but rather have a positive effect on plant growth.

**Figure 3.** AuNP-SCTA enhances growth of Arabidopsis seedlings. Wild-type Arabidopsis seedlings were grown for 7 d on agar-solidified <sup>1</sup> <sup>2</sup> MS medium containing 10 mg/L of AuNP-SCTA or SCTA (SC 2.2 mM; TA 200 μM) (control) or on un-supplemented medium (untreated). Growth parameters were scored: (**a**) primary root length and (**b**) representative picture of (**a**). (**c**) Rosette diameter and (**d**) representative photograph of (**c**). (**e**) Lateral root number. (**f**) Lateral root length. Results shown are means ± SE with n = 20. Different labels a and b indicate statistically different groups according to multiple comparisons following one-way ANOVA analysis at a probability level of *p* < 0.01. All experiments were repeated twice with similar results.

#### *3.4. Immune Responses upon AuNP Treatment*

NPs have been reported to affect the innate immune system in animals [5]. To assess whether they have an influence on plant immune responses, different innate immune defensive reactions in plants were evaluated. The production of reactive oxygen species (ROS) in the apoplast and lipid peroxidation are typical cellular events triggered by the plant surveillance system that detects highly conserved microbe- or pathogen-associated molecular patterns (M/PAMPs) via cell surface-located pattern-recognition receptors (PRRs) in a process called pattern-triggered immunity (PTI) [91,92].

The cellular response of *Arabidopsis thaliana* to abiotic stress resulting from NP exposure was initially measured as reactive oxygen species (ROS) production or oxidative burst, using a luminol-based chemiluminescence assay. As shown in Figure 4a, no ROS production was detected after exposure to milliQ water (untreated control), AuNP-SCTA (100 mg/L), or coating solution SCTA (SC 2.2 mM; TA 200 μM) (control). This also confirms that the NP suspensions were free of endotoxins such as LPS that would induce ROS in plants [93]. As positive control, the PAMP flg22 was added at a final concentration of 100 nM. The same concentration of the elicitor was used as treatment also in combination with AuNP-SCTA (10 or 100 mg/L) or SCTA as control. Although the coating solution did not affect the level of ROS production caused by flg22 treatment, AuNP-SCTA influenced the level of recorded ROS. In particular, in the presence of 10 mg/L AuNP-SCTA the PAMP (flg22) signal decreased. Furthermore, a 10× higher NP concentration (100 mg/L) was tested, and a further decrease in the level of ROS was detected.

**Figure 4.** AuNP-SCTA decreases ROS production and lipid peroxidation levels. (**a**) ROS production measured with a luminol-based assay in leaf squares of Arabidopsis Col-0. ROS production is represented as relative light units (RLU) after elicitation with milliQ water (untreated control), flg22 (100 nM) (positive control), AuNPs-SCTA (100 mg/L), SCTA (control), flg22 + SCTA, or flg22 + AuNPs-SCTA 10 or 100 mg/L. Results are mean ± SE (n = 9). The experiment was repeated two times with similar results. (**b**) Lipid peroxides level, expressed as CHP equiv./mg ww, was measured in Arabidopsis leaves with the FOX assay. The results after treatment with milliQ water (untreated), flg22 (100 nM) (positive control), or flg22 + AuNPs-SCTA (10 or 100 mg/L) are presented as mean ± SE of three independent experiments. Based on two-way nested ANOVA followed by Dunnett's post-hoc test, data were normally distributed (Shapiro–Wilk test) and showed homogeneity of variances (Levene's test). Different letters indicate statistically significant differences at *p* < 0.01. Different labels a–c indicate statistically different groups according to multiple comparisons following two-way nested ANOVA analysis at a probability level of *p* < 0.01.

In order to discriminate between a real decrease in the ROS production and a mere technical interference with the light detection, a lipid peroxidation assay was performed. Cellular and organelle membranes, due to their high polyunsaturated fatty acid (PUFA) content, are particularly susceptible to ROS-induced peroxidation [94]. The applied colorimetric ferrous oxidation xylenol orange (FOX) assay was modified to quantify lipid

hydroperoxides (LOOHs) in plant extracts. Upon treatment with 10 and 100 mg/L AuNP-SCTA plus flg22, the level of lipid peroxidation decreased significantly in comparison to flg22 alone (Figure 4b). Treatment with 100 mg/L AuNP-SCTA resulted in a more pronounced decrease in the lipid hydroperoxide level compared to 10 mg/L AuNP-SCTA. The FOX assay confirmed the oxidative burst assay results, clearly pointing out that co-exposure to PAMPs and AuNP-SCTA reduced the PAMP-induced ROS burst and subsequent lipid peroxidation. The underlying mechanism of this effect is still elusive, but the data suggest that AuNP-SCTA might be able to detoxify ROS and shift the balance between growth and immunity trade-off to the growth side.

#### *3.5. Transcriptomics Analysis of AuNP-SCTA-Exposed Arabidopsis Seedlings*

To untie the molecular nature of the plant–AuNP interaction, whole transcriptome analyses were performed on Arabidopsis seedling roots after short (6 h) and long (7 d) exposure to 10 mg/L AuNP-SCTA in hydroponic culture and agar-solidified medium (6 h and 7 d, respectively). As controls, the seedlings were treated with SCTA (2.2 mM SC; 200 μM TA). Samples were sequenced with an Illumina HiSeq platform. The average genome mapping rate was 94.66% and the average gene mapping rate was 92.04%. Raw data for both experimental conditions and all three replicates are shown in Supplemental Table S1.

As shown in Figure 5, a total of 651 differentially expressed genes (DEG) were identified after short-term treatment and 6 DEGs after long-term exposure. Whereas 121 genes were upregulated after 6 h of AuNP treatment, 530 genes were downregulated. After 7 d, 3 genes were upregulated and 3 genes were downregulated. DEGs with expression information are listed in Supplemental Table S2. Gene ontology (GO) (molecular biological function, cellular component, and biological process) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway classification are reported in Supplemental Figures S3 and S4. In both conditions, genes involved in the response to external stimuli and cellular and metabolic processes are overrepresented within the DEGs. In particular, after short-term exposure the majority of genes involved in disease resistance, defense response, response to oxidative stress, and metal response were downregulated. This indicates that immune and oxidative stress responses were negatively affected during AuNP exposure.

DUF642 L-GalL-responsive gene 2 (DGR2, At5g25460), a gene involved in growth and development of Arabidopisis plants, was up-regulated. DGR2 has a key role in Arabidopsis root elongation and shoot development [95,96]. Downregulation of immune response genes and upregulation of growth factors indicate a shift in the trade-off between immune and growth effects and may explain the growth-promoting effects of AuNPs. The Nicotianamine synthase 2 gene, (NAS2, At5g56080), the only shared DEG between the two conditions, encodes for a protein involved in the synthesis of nicotianamin. Mutants in NAS2 show altered metal contents, indicating a role in metal uptake or response [97] (Supplemental Table S2). After 7 d of NP exposure, only 6 DEGs were detected, compared to the 651 genes identified after 6 h, clearly pointing out that transcriptome changes are relevant only at early time points after AuNP treatment.

**Figure 5.** DEGs after AuNP-SCTA treatment. MA plot representing DEGs (upregulated genes: red dots; downregulated genes: blue dots) and non-DEGs (grey dots) in Arabidopsis seedling roots after (**a**) short- and (**b**) long-term AuNP-SCTA treatment (6 h and 7 d, respectively), detected by RNA-seq data analysis. The X-axis represents value M (log2 transformed fold change of a gene expression value) and the Y-axis represents value A (log2 transformed mean expression level). (**c**) Venn diagram displaying the total number of up- (red) and downregulated (blue) differentially expressed genes in both treatments and the name of the single overlapping gene.

#### *3.6. Proteomic Analysis of the Effect of AuNP-SCTA in Arabidopsis*

To further understand the mechanisms underlying the effects of AuNPs on *Arabidopsis thaliana* seedlings, proteomic analyses were performed on seedlings using mass spectrometry. Global changes in protein expression were investigated in Arabidopsis seedlings in the same experimental setup as used for the transcriptome analyses. Protein extracts were analyzed via nano-liquid chromatography double mass spectrometry (NanoLC-MS/MSspectrometry).

As shown in Figure 6, from a total of 2727 detected proteins after 6 h exposure and 2503 after 7 d exposure, 119 and 59 differentially expressed proteins (DEPs), respectively, were identified. All identified up- and downregulated proteins, along with their expression profiles, are listed in Supplemental Table S3. Furthermore, we sorted the DEPs into gene ontology (GO) categories (molecular biological function, cellular component, and biological process) and KEGG pathways, as shown in Supplemental Figures S5 and S6. DEPs significantly overrepresented after both treatments were involved in metabolic processes, protein synthesis, and response to stimuli (Supplemental Table S3).

**Figure 6.** DEPs after AuNP-SCTA treatment. Volcano plots representing DEPs (upregulated proteins: red dots; downregulated proteins: blue dots) and non-DEPs (grey dots) in Arabidopsis seedlings after (**a**) short- and (**b**) long-term AuNP-SCTA treatment (6 h and 7 d, respectively) and detected by nano-liquid chromatography with tandem mass spectrometry (NanoLC-MS/MS) of total protein extracts. The X-axis represents the log2 transformed fold changes; the Y-axis represents the log10 transformed intensity (in log10), with H representing the treatment and L the control. (**c**) Venn diagram displaying the total number of up- (red) and downregulated (blue) differentially expressed proteins in both treatments and the protein names of the 4 overlapping proteins.

Oxidative stress-related proteins were mainly downregulated, as shown on the transcriptome level. The overlap analysis of the different timepoints revealed the protein DGR1 (DUF642 L-GalL-responsive gene 1, At1g80240), which was investigated for its role during the development of *Arabidopsis thaliana* [95]. After 7 d of treatment DGR1 and DGR2 were both upregulated, whereas after 6 h of treatment only DGR1 was initially downregulated. In the transcriptome analyses, the gene encoding for DGR2 was also detected to be upregulated. GSTF6 (Glutathione S-transferase F6, At1g02930), another DEG shared between treatments, encoded for a downregulated glutathione transferase involved in defense mechanisms (Supplemental Table S4). The finding of DGR2 and GSTF6 in both DEGs and DEPs indicates that these were reproducibly and robustly regulated genes/proteins upon AuNP exposure. As DGR1 and DGR2 have been previously described to be involved in growth and development, the differential regulation of these genes/proteins may explain why AuNPs have a positive effect on Arabidopsis growth. Our well-controlled transcriptome and proteome dataset provides a source for future analysis of the molecular mechanism underlying AuNP-induced growth-promotion.

#### **4. Discussion**

The widespread use of NP-containing products has led to the direct exposure of the terrestrial environment to these nanosized materials, raising concerns regarding their safety and biocompatibility with both living organisms and the environment [1,16,98–100]. Therefore, the risks and hazard assessment of NP exposure for plants, soil organisms, and consequently, humans, as a result of contamination of the food chain need to be addressed [1,101,102]. To date, despite recent developments in plant nanotoxicology, an unequivocal understanding of the effects of NPs on terrestrial plants is lacking, with fundamental information gaps about their mechanisms of action [103–105].

Plant-based nanosafety research focuses on a number of key aspects, i.e., the physicochemical properties of NPs such as material, size, and surface chemistry; the interaction of NPs with the surrounding environment; and the plant type and route of exposure [104–108]. Possible alterations in the properties and colloidal stability of NPs once released into an environment other than that of synthesis make studies under natural conditions difficult to interpret; thus, nanosafety assessments under reproducible and controlled conditions help to interpret investigations in ecotoxicological assays [24,86,109].

AuNPs, due to their unique intrinsic optical, biological, and catalytic properties [110–112] and their biocompatibility with mammalian systems, have been exploited in numerous medical and technological applications and used as model particles under laboratory conditions in nanotechnological research [113–116]. The effects on plants at the physiological and molecular level remain controversial [40,54,57,117]. In this light, this study aimed at studying the behavior of engineered AuNPs as a starting material and after dispersion in plant growth media, along with their physiological and molecular effects on the model plant *Arabidopsis thaliana*.

The high salt concentration of plant growth media may facilitate the aggregation of NPs and alterations in their bio-identity [118–121]; thus, surface-stabilizing agents are used to stabilize colloidal suspensions through electrostatic, steric, or electrosteric repulsion [122]. NP surface-stabilizing agents can play a key role in plant and animal toxicity tests [123,124]. In particular, the ionic charge conferred by particle coatings may influence the physical interaction between NPs and cell membranes, with positively charged NPs being more effective than negatively charged ones [125]. Barrena et al. [124] showed that, in germination tests of cucumber and lettuce seeds, toxic effects can be attributed to NP solvents rather than to NPs themselves. In this light, we showed that exposure of Arabidopsis seedlings to 10 mg/L of the negatively charged surface stabilizer SCTA did not influence Arabidopsis root growth.

Effects of electrostatically or sterically stabilized AuNPs have been studied in liquid and agar-solidified plant media, though their behavior and possible state of aggregation have not been further described in all cases [45,126–128]. By contrast, in their physiological and toxicological studies on Arabidopsis seedlings, Siegel et al. [41] dispersed SC-capped AuNPs in 1/16-diluted low-salinity MS, detecting a slight aggregation of AuNPs and consequently suboptimal conditions of plant growth assays. In this study, we tested the overtime stability of SC-stabilized and SCTA-stabilized AuNPs. The presence of traces of TA, the only difference between the two types of synthetized NPs, increased the stability of the particles by conferring a higher surface charge or partial steric stabilization and providing the necessary stability against salt-driven aggregation. A comprehensive physicochemical characterization of newly synthesized NPs, prior to and during their use in Arabidopsis treatments, was carried out, showing that the NPs are stable, dispersed, and usable for reproducible plant exposure experiments.

Since contaminants in plant growth media allow for the growth of microorganisms, NPs need to be sterile before their use in plant assays. As shown by previous studies, autoclaving and radiation sterilization might result in the aggregation of the NPs, loss of the coating, and contamination with potential microbial toxins [129–133]. Sterile filtration has been shown not to directly affect the physical properties of NPs, but filter materials should be tested to exclude possible interactions with particle surfaces resulting in NP retention or coating removal [132]. We tested CME and PES filters and revealed that PES filters are suitable for sterilization of AuNPs, whereas CME filtering resulted in a significant reduction in the number of NPs in the filtered samples. Therefore, PES filters are considered suitable for AuNP sterilization to allow for sterile plant cultivation in the presence of AuNP-SCTA.

A number of studies have addressed AuNP responses in plants, reporting both positive and negative effects [40]. In this study, growth-promoting effects of AuNP-SCTA at a moderate concentration (10 mg/L) were revealed. As AuNP concentrations in the environment are very low, studies with lower concentrations might reflect more natural conditions [134]. Previous studies have found that AuNPs at high concentrations (≥100 mg/L) cause detrimental effects on plants, whereas for lower concentrations of AuNPs larger than 5 nm growth-promoting effects have been shown, supporting our findings that AuNPs have positive effects on plant growth [45–47,57]. Furthermore, Siegel et al. [41] tested three different sizes of AuNPs (10, 14, and 18 nm) at increasing concentrations (1, 10, and 100 mg/L) and showed that at the highest concentration the smaller particles reduced the length of the *Arabidopsis thaliana* root more than the larger ones. It has been hypothesized that a high concentration of NPs negatively affects plant growth by particle adsorption onto the cell wall of the root system, decreasing pore size and inhibiting water transport [124,135,136]. On the other hand, some contradictory studies have been reported. Feichtmeier et al. [135] reported a decrease in the biomass of *Hordeum vulgare* after AuNP treatment at a final concentration of between 3 and 10 mg/L. Some of these discrepancies can be explained by differences in specific experimental settings and different behavior of NPs under test conditions, which make a clear assessment of AuNP responses more difficult. Therefore, a careful evaluation of each study is necessary to draw a complete picture of the effects of AuNPs on plants.

Although the mechanisms of action and effects of AuNPs on plants are not yet fully understood [40,54,57,117], for the Au bulk counterpart the results are clearer. As plants have revealed their potential in the green synthesis of AuNPs, many studies have been produced on the physiological responses of plants to Au salts [137], used as starting material in order to obtain NPs. Gold is required by plants in traces, but its absorption in higher amounts can cause drastic changes in plant growth [40]. A previous study demonstrated that Arabidopsis seedlings treated with 10 mg/L of potassium tetrachloroaurate(III) (KaAuCl4) showed the formation of AuNPs in the roots and shoots and enhanced vegetative growth [138], whereas higher amounts of KAuCl4 or gold(III) chloride (AuCl3) (100 mg/L) negatively affected the root length and shoot development [139].

Immune responses are reported to be activated upon NP exposure in many plant and animal models, including reactive oxygen species (ROS) production and lipid peroxidation [48,140–142]. Here, we found that AuNP-SCTA alone did not induce these classical plant defense responses. In addition, ROS production induced by the 22-amino acid peptide derived from bacterial flagellin (flg22), sensed in plants as a pathogen-associated molecular pattern (PAMP), was significantly reduced in the presence of increasing amounts of AuNP-SCTA. To exclude a biophysical quenching effect of the light emitted by luminol, we performed lipid peroxidation assays. The same results were obtained by measuring lipid peroxidation, an indicator of oxidative stress in animals and plants [143], showing that the PAMP-triggered ROS burst was indeed reduced. Kumar et al. [47] showed that AuNPs at 80 mg/L significantly improved the free radical scavenging activity of Arabidopsis seedlings by increasing the activity of enzymes involved in the defense system against ROS, whereas plants treated with AuNPs in the range of 100 to 400 mg/L showed reduced growth, which was considered to be a consequence of increased free radical stress [40,144]. After 6 h of treatment with 10 mg/L of AuNP-SCTA, we found 10 peroxidases to be downregulated on the transcript level and 6 at the protein level, which were likely involved in oxidative stress reactions. These data indicate a correlation between ROS production and AuNP effects and show that AuNPs can reduce stress responses triggered by immune stimulatory peptides. Whether this effect is based on a direct effect on the peptide, e.g., through adsorption to the NP surface or changes in the peptide accessibility (and consequent alteration of its mode of perception) [145], or on a protective effect of AuNPs on PAMP recognition or downstream signaling, will be interesting to study in the future.

We performed transcriptomics and proteomics analyses to study the alterations caused by short (6 h) and long (7 d) AuNP exposure at the molecular level. In particular, after shortterm AuNP treatment, genes involved in disease resistance, defense response, oxidative stress, and auxin- and metal-response were downregulated. In the proteomic analysis after short treatment we detected two distinct categories of upregulated proteins, i.e., proteins involved in responses to oxidative stress and abiotic stimuli, whereas after long NP exposure all upregulated proteins were annotated as involved in development processes. To our knowledge, this is the only study analyzing the transcriptomic and proteomic changes after AuNP treatments in Arabidopsis, whereas such analyses have been performed on the roots of Arabidopsis seedlings upon gold (KAuCl4) exposure, which leads to AuNP formation by the plant [138]. As for the physiological effects mentioned above, at the molecular level the changes induced by gold exposure also showed some effects similar to those induced by AuNP exposure. A comparative analysis between our transcriptomics data and Tiwari et al.'s [138] shows that there is a significant overlap of up- and downregulated genes (three common upregulated and 22 common downregulated genes, Figure S7). In particular, between the upregulated DEGs two metal response genes (MT1C, At1g07610 and ALMT1, At1g08430) and DGR2 (At5g25460) were found. In both studies, disease and defense response and oxidative stress genes were downregulated. By contrast, Tiwari et al. [138] found that developmental, auxin-responsive, and metal-responsive genes were upregulated after Au treatment, whereas in our study the same categories of genes were downregulated after AuNP treatment. These differences were likely caused by different effects caused by Au ion uptake compared to exposure to nanoparticles. As metal AuNPs are very inert, the significant overlap between Au salt and AuNP is potentially caused by NP effects in both experiments, as Au ions are taken up and converted into AuNPs inside the plant, where they may cause similar effects as external NP exposure.

An overlap analysis of our proteomic and transcriptomic studies revealed two particularly interesting candidates: DUF642 L-GalL-responsive genes 1 and 2 (DGR1, At1g80240 and DGR2, At5g25460), which were also found to be upregulated by Au salt exposure [138]. DGR1 and DGR2 encode for two proteins belonging to the DUF642 protein family, whose members are part of the cell wall proteome [96] and have shown in Arabidopsis a complementary expression pattern in young and developed roots, suggesting a similar but non-redundant function [95]. As Gao et al. reported in their study [95], DGRs are involved in the development processes of Arabidopsis, and in particular in root elongation. DGR2 seems to have a predominant role, as *dgr2* single mutants show a short, undeveloped root phenotype [95]. These results suggest the potential involvement of these proteins in the root growth-promoting effects induced by AuNPs and can be used as a starting point for further studies aimed at dissecting the pathways underlying the beneficial effects of AuNP-SCTA on Arabidopsis development.

#### **5. Conclusions**

NPs are released into the environment in increasing amounts and their high reactivity may cause problems that are not associated with the respective bulk material. Therefore, an ecotoxicological assessment is necessary to evaluate their risk in nature, but to understand the molecular mechanisms underlying the effects of NPs on the environment, controlled model systems are necessary. Here, we describe the establishment of a stable and reproducible system to study plant responses to AuNPs after short- and long-term exposure. Both initial and overtime characterization of NPs, especially after dispersal in new environments, is essential. The effects resulting from NP-plant interaction need stable, sterile, and reproducible colloidal solutions, ensured by the use of non-toxic NP surface stabilizing agents. In this study, we demonstrated that these AuNP-SCTAs positively influence the growth of Arabidopsis seedlings, while also conferring partial protection against oxidative stress caused by triggering immune-responses. Transcriptomics and proteomics studies show downregulation of (oxidative) stress and immune responses and upregulation of growth-promoting genes and support the scenario that the trade-off between growth and immune/stress responses are shifted to the growth side after AuNP exposure (Figure 7).

The identified DEGs and DEPs provide a useful data source for future analysis of the molecular mechanism underlying AuNP-induced growth stimulation.

**Figure 7.** Model of AuNP effects on Arabidopsis seedlings. AuNPs stabilized with SCTA have growthpromoting effects on Arabidopsis seedlings and can reduce oxidative stress genes/proteins and ROS burst after triggering with the pathogen-associated molecular pattern (PAMP) flg22, indicating that the NPs can shift the trade-off between growth and defense responses to the growth side.

**Supplementary Materials:** The following information is available online at https://www.mdpi. com/article/10.3390/nano11123161/s1, Figure S1: Physicochemical characterization of AuNP-SC dispersed in H2O, <sup>1</sup> <sup>2</sup> MS, and <sup>1</sup> <sup>2</sup> MS agar; Figure S2: Growth of Arabidopsis seedlings in the absence and presence of AuNPs; Figure S3: GO classification of DEGs after AuNP-SCTA treatment; Figure S4: Pathway classification of DEGs after AuNP-SCTA treatment; Figure S5: GO classification of DEPs after AuNP-SCTA treatment; Figure S6: Pathway classification of DEPs after AuNP-SCTA treatment; Figure S7: Venn diagram of DEGs after AuNP-SCTA exposure (this study) and DEG after exposure to KAuCl4 resulting in in planta AuNP formation [140]. Table of the overlap between our DEGs and those published by Tiwari et al. (2016); Table S1: Clean reads quality metrics and summary of genome mapping; Table S2: DEGs list after short (6 h) and long (7 d) AuNP-SCTA treatment; Table S3: DEP list after short (6 h) and long (7 d) AuNP-SCTA treatment; Table S4: Transcriptomic and proteomic studies: overview and overlap.

**Author Contributions:** Conceptualization, E.F. and B.K.; methodology, E.F., F.B., M.B.-F., H.-R.K. and B.K.; validation, E.F. and B.K.; formal analysis, E.F.; investigation, E.F. and F.B.; resources, M.F.-W.; data curation, E.F., F.B. and M.F.-W.; writing—original draft preparation, E.F. and B.K.; writing review and editing, all authors; visualization, E.F. and B.K.; supervision, V.P. and B.K.; project administration, B.K.; funding acquisition, B.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received funding from the European Union's Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement PANDORA, no. 671881.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Acknowledgments:** We thank Helene Eckstein for her support with the FOX assays and Silke Wahl and Irina Droste-Borel for their technical support with NanoLC-MS/MS analysis.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **In Vitro Interactions of TiO2 Nanoparticles with Earthworm Coelomocytes: Immunotoxicity Assessment**

**Natividad Isabel Navarro Pacheco 1,2, Radka Roubalova 1, Jaroslav Semerad 1,3, Alena Grasserova 1,3, Oldrich Benada 1, Olga Kofronova 1, Tomas Cajthaml 1,3, Jiri Dvorak 1, Martin Bilej <sup>1</sup> and Petra Prochazkova 1,\***


**Abstract:** Titanium dioxide nanoparticles (TiO2 NPs) are manufactured worldwide. Once they arrive in the soil environment, they can endanger living organisms. Hence, monitoring and assessing the effects of these nanoparticles is required. We focus on the *Eisenia andrei* earthworm immune cells exposed to sublethal concentrations of TiO2 NPs (1, 10, and 100 μg/mL) for 2, 6, and 24 h. TiO2 NPs at all concentrations did not affect cell viability. Further, TiO2 NPs did not cause changes in reactive oxygen species (ROS) production, malondialdehyde (MDA) production, and phagocytic activity. Similarly, they did not elicit DNA damage. Overall, we did not detect any toxic effects of TiO2 NPs at the cellular level. At the gene expression level, slight changes were detected. Metallothionein, fetidin/lysenin, lumbricin and MEK kinase I were upregulated in coelomocytes after exposure to 10 μg/mL TiO2 NPs for 6 h. Antioxidant enzyme expression was similar in exposed and control cells. TiO2 NPs were detected on coelomocyte membranes. However, our results do not show any strong effects of these nanoparticles on coelomocytes at both the cellular and molecular levels.

**Keywords:** earthworm; coelomocyte; TiO2 nanoparticles; reactive oxygen species; innate immunity; lipid peroxidation; alkaline comet assay; phagocytosis; apoptosis; gene expression

#### **1. Introduction**

Titanium dioxide nanoparticles (TiO2 NPs) are commonly used in different industries because of their physico-chemical properties. TiO2 NPs have photocatalytic properties, protect against UV radiation, are used as semiconductors, etc. These nanoparticles are used, e.g., in cosmetics, food industry, paints, ceramics, devices development, and the agriculture industry [1–3]. In the last decade, TiO2 NPs have been used in wastewater treatment plants for their ability to degrade some organic pollutants [1]. Thus, TiO2 NPs reach the soil system from different sources including sludge, nanofertilizers, and nanopesticides. These nanoparticles then interact with the soil biota. It is therefore very important to assess the potential risk of TiO2 NPs to soil organisms.

Earthworms are dominant soil invertebrate animals. They possess a strong immune system because of their permanent contact with soil bacteria, viruses, and fungi. Defense mechanisms are used in earthworm protection against soil pollutants including nanoparticles. Earthworms *Eisenia andrei* and *E. fetida* are used as model organisms to monitor ecotoxicity according to OECD guidelines [4–6]. TiO2 NPs do not affect earthworm viability and growth [2,3]. In some cases, reproductive inhibition was observed [7]. Further, these nanoparticles can induce, e.g., oxidative stress, DNA damage, apoptosis,

**Citation:** Navarro Pacheco, N.I.; Roubalova, R.; Semerad, J.; Grasserova, A.; Benada, O.; Kofronova, O.; Cajthaml, T.; Dvorak, J.; Bilej, M.; Prochazkova, P. In Vitro Interactions of TiO2 Nanoparticles with Earthworm Coelomocytes: Immunotoxicity Assessment. *Nanomaterials* **2021**, *11*, 250. https://doi.org/10.3390/ nano11010250

Received: 22 December 2020 Accepted: 14 January 2021 Published: 19 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

and affect gene expression [3]. Earthworm cellular defense mechanisms are based on coelomocytes present in the coelomic fluid. Coelomocytes can be divided into free chloragogen cells called eleocytes, with a mainly nutritive function, and amoebocytes, which are the immune effector cells [8]. Amoebocytes can be further divided into granular (GA) and hyaline (HA) amoebocytes.

Various nanoparticles were described to impair earthworm defense mechanisms. Hayashi et al. showed that Ag NPs altered the expression of some genes involved in coelomocyte oxidative stress and immune reactions [9]. Further, Ag nanowires detected on coelomocyte membranes increased intracellular esterase activity [10]. ZnO NPs were internalized by coelomocytes, with consequent DNA damage [11]. However, similar mechanisms were not described for TiO2 NPs in earthworms. TiO2 NPs cause significant mitochondrial dysfunction by increasing mitochondrial ROS levels and decreasing ATP generation in macrophages. Moreover, TiO2 NPs exposure activated inflammatory responses and attenuated macrophage phagocytic function [12]. TiO2 NPs interacted with sea urchin immune cells and increased the antioxidant metabolic pathway in vitro [13]. In earthworms, only increased apoptosis was observed following TiO2 nanocomposites exposure [7,14–16].

A compromised immune system may result in a decreased reproductive rate and increased mortality of earthworms. Thus, nanoparticle toxicity risk assessment is extremely important, as the adverse health effects remain poorly characterized for many nanomaterials. We aimed to assess the potentially dangerous impact of TiO2 NPs exposure on earthworms' cellular function, including the immune responses to harmful stimuli.

*E. andrei* coelomocytes were exposed to 1, 10, and 100 μg/mL of TiO2 NPs for 2, 6, and 24 h in vitro. After exposure, viability, oxidative stress (reactive oxygen species and malondialdehyde production), immune functions (phagocytosis), and genotoxicity (DNA damage) were assessed. Further, electron microscopy (transmission and scanning) enabled TiO2 NPs localization on the cell surface. Gene expression changes were also followed to better understand the underlying cellular mechanisms.

#### **2. Materials and Methods**

#### *2.1. Animal Handling, Sample Collection, and Culture Medium Preparation*

Clitelate, adult *Eisenia andrei* earthworms were obtained from the laboratory compost breeding. Earthworms were first kept on moist filter paper for 48 h to depurate their guts. Coelomocytes were harvested by applying 2 mL of extrusion buffer (5.37 mM EDTA (Sigma-Aldrich, Steinheim, Germany); 50.4 mM guaiacol glyceryl ether (GGE; Sigma-Aldrich, Steinheim, Germany) in Lumbricus Balanced Salt Solution (LBSS; [17]) per earthworm for 2 min. The cells were then centrifuged and washed twice in LBSS (200× *<sup>g</sup>*, 4 ◦C, 10 min). Subsequently, cells were counted and diluted to 10<sup>6</sup> cells/well for scanning electron microscopy (SEM) and lipid peroxidation assessment. 1 × <sup>10</sup><sup>5</sup> cells/well, <sup>2</sup> × <sup>10</sup><sup>5</sup> cells/well, and 3 × <sup>10</sup><sup>5</sup> cells/well were used for the ROS production analysis, apoptosis detection, and phagocytosis assay, respectively.

RPMI 1640 culture medium supplemented with 5% heat-inactivated fetal bovine serum (FBS; Life technologies, Carlsbad, USA), 1 M HEPES (4-(2-hydroxyethyl)-1-piperazineethane sulfonic acid; pH 7.0–7.6, Sigma-Aldrich; Gillingham, UK), 100 mM sodium pyruvate (Sigma-Aldrich, Steinheim, Germany), 100 mg/mL gentamycin (Corning, Manassas, VA, USA), and antibiotic–antimycotic solution (Sigma-Aldrich, Steinheim, Germany) was diluted with autoclaved MilliQ-water to 60% (*v*/*v*) to obtain R-RPMI 1640 medium [18]. Subsequently, TiO2 NPs were dispersed in R-RPMI 1640 medium and incubated with cells in darkness at 20 ◦C for 2, 6, and 24 h in triplicate.

#### *2.2. TiO2 NPs Characterization*

Aeroxide TiO2 P25 nanoparticles (irregular and semi-spherical shape; mexoporous NPs, anatase, and rutile 4:1; primary size between 10 and 65 nm) were purchased from Evonik Degussa (Essen, Germany). TiO2 NPs were previously characterized in several

aqueous solutions, as described by Brunelli et al. [19]. Nanoparticle physico-chemical properties were determined by ZetaSizer Ultra (Panalytical Malvern; Malvern, UK), transmission electron microscope (TEM), and TECAN 200 Pro plate reader. Powder TiO2 NPs were weighed and dispersed in distilled water. Then, diluted TiO2 NPs were vortexed thoroughly for 5 min prior to further dilution [20]. TiO2 NPs were diluted either in R-RPMI 1640 medium or distilled water to a concentration of 1, 10, and 100 μg/mL, and incubated for 2, 6, and 24 h. Experiments were carried out in triplicate. Culture medium and distilled water without NPs were used as negative controls.

#### *2.3. Electron Microscopy Analyses*

#### 2.3.1. Cell Preparation

Coelomocytes were exposed to 1, 10, and 100 μg/mL TiO2 NPs for 2, 6, and 24 h. Cell viability was measured by propidium iodide (PI; 1 μg/mL) staining using flow cytometer. Then, samples were collected and fixation solution (5% glutaraldehyde in PBS) was added in a 1:1 ratio (*v*:*v*). Fixed cells were shaken gently for 15 min and kept overnight at 4 ◦C.

#### 2.3.2. Scanning Electron Microscopy (SEM)

For SEM, fixed cells were washed with LBSS buffer three times at room temperature for 20 min, and centrifuged at 150× *g* for 10 min. Then, they were allowed to adhere onto poly-L-lysine coated round 13 mm Thermanox Plastic Coverslips (Nunc, Thermo Fisher Scientific; Roskilde, Denmark) overnight at 4 ◦C. The coverslips with attached cells were washed with ddH2O and fixed with 1% OsO4 for one hour at room temperature. The coverslips were then washed three times for 20 min, dehydrated through an alcohol series (25, 50, 75, 90, 96, and 100%), and were critical-point dried from liquid CO2 in a K850 Critical Point Dryer (Quorum Technologies Ltd., Ringmer, UK). The dried coverslips were sputter-coated using a high-resolution Turbo-Pumped Sputter Coater Q150T (Quorum Technologies Ltd., Ringmer, UK) with 3 nm of platinum. Alternatively, for EDS microanalysis, the samples were coated with 10 nm of silver or 5 nm of carbon. The final samples were examined in a FEI Nova NanoSEM scanning electron microscope (FEI, Brno, Czech Republic) at 5 kV using CBS and TLD detectors. An electron beam deceleration [21] mode of the Nova NanoSEM scanning electron microscope performed at a StageBias of 883.845 V and accelerating voltage of 5 kV was used for high-resolution imaging. The EDS microanalysis was performed at 15 kV using an Ametek® EDAX Octane Plus SDD detector and TEAM™ EDS Analysis Systems (AMETEK B. V.; Tilburg, The Netherlands).

#### 2.3.3. Transmission Electron Microscopy (TEM)

For TEM, a TiO2 NPs suspension (500 μg/mL; 5 μL) was applied onto glow-dischargeactivated [22] carbon-coated 400-mesh copper grids (G400, SPI Supplies, Structure Probe, Inc., West Chester, PA, USA). Nanoparticles were sedimented for 1 min and the remaining solution was then blotted with filter paper and the grids were air-dried. A Philips CM100 electron microscope (Philips EO, Eindhoven, The Netherlands; Thermo Fisher Scientific) equipped with a Veleta slow-scan CCD camera (EMSIS GmbH, Muenster, Germany) was used to examine the grids. TEM images were processed in the proprietary iTEM software (EMSIS GmbH, Muenster, Germany).

#### *2.4. Flow Cytometry Assays*

Coelomocytes were incubated with TiO2 NPs (1, 10, and 100 μg/mL) for 2, 6, and 24 h. Cells were then treated as described below and analyzed with a laser scanning flow cytometer. Through flow cytometry, coelomocytes were subdivided into eleocytes, granular (GA), and hyaline amoebocytes (HA). The coelomocytes subset detection was based on the cell size (FSC) and the cell inner complexity/granularity (SSC). Cell viability was assessed for every assay. All flow cytometry assays were performed by three independent experiments with three replicates per each treatment and time interval. The minimum collected events

were 1000 per population. Event counts per each gate were calculated by Flowjo (9.9.4 version, BD Biosciences, San Jose, CA, USA). In each flow cytometry assay, coelomocytes were exposed to H2O2 as a positive control (Sigma-Aldrich, Steinheim, Germany; 10 mM H2O2 for 30 min incubation for apoptosis and phagocytosis, and 1 mM H2O2 for ROS production assesment). Controls with and without PI (1 mg/L; Sigma-Aldrich, Steinheim, Germany) were included in each experiment. Further, control analysis of 1, 10, and 100 μg/mL TiO2 NPs incubated with or without cells for 2, 6, and 24 h were performed (Figure S1).

For ROS production determination, 20.6 μM 2 ,7 -dichlorofluorescin diacetate (DCF-DA; Sigma-Aldrich, Steinheim, Germany) was added to the washed cell suspension (LBSS, 200× *g*, 4 ◦C, 10 min) for 15 min in darkness. Subsequently, the cell suspension was washed twice with LBSS (200× *g*, 4 ◦C, 10 min) and stained with PI.

To detect the apoptotic process a cell suspension was washed twice with Annexin V buffer (200× *g*, 4 ◦C, 10 min; 0.01 M HEPES (pH 7.4), 0.14 M NaCl, and 2.5 mM CaCl2 solution), and subsequently stained with 5 μL of Alexa Fluor 647-Annexin V (15 min in darkness; Thermo Fisher Scientific, Eugene, OR, USA). PI was then added to the cell suspension and measured by flow cytometry. The apoptosis % represented the apoptotic cell number out of each subpopulation. The necrosis % represented the necrotic cell number out of each subpopulation.

The phagocytosis assay was performed using latex beads (Fluoresbrite® Plain YG; 1 μm microspheres diameter; Polysciencies Inc., Warrington, PA, USA) added to the incubation plates in a 1:100 ratio (cells:beads) and kept in darkness at 17 ◦C for 18 h. Then, cell suspensions were washed twice with LBSS (200× *g*, 4 ◦C, 10 min), stained with PI, and analyzed by flow cytometry. The % phagocytic activity was determined by the % of alive cells, which were able to engulf at least one bead out of each subpopulation. Each experiment included samples with NPs dispersed in the medium in order to detect effects exerted by NPs alone.

#### *2.5. Malondialdehyde (MDA) Production and Alkaline Comet Assay*

Coelomocytes were incubated with TiO2 NPs (10 and 100 μg/mL) or CuSO4 (100 μg/mL; positive control) for 2, 6, and 24 h. Afterward, cell suspensions were collected and MDA production was measured. MDA production was detected by high-performance liquid chromatography with fluorescence detection (HPLC/FLD) using derivatized MDA-TBA2 [23]. MDA analysis was performed in three independent experiments with 3 replicates for each treatment and time interval.

For the alkaline comet assay, 1.5 × <sup>10</sup><sup>4</sup> cells exposed to 1, 10, and 100 <sup>μ</sup>g/mL TiO2 NPs for 2, 6, and 24 h were mixed with 2% 2-hydroxyethyl agarose (Sigma-Aldrich, Steinheim, Germany) at 37 ◦C. Glass slides containing agarose with cells were kept at 4 ◦C for 10 min. Subsequently, samples were incubated for 2 h in lysis buffer (2.5 M NaCl, 10 mM Tris-HCl, 100 mM EDTA, 1% Triton X-100, pH 10). Then, slides were immersed three times in unwinding buffer (0.03 M NaOH, 2 mM EDTA, pH 12.7) for 20 min. Gel electrophoresis was carried out at 24 V, 300 mA for 25 min. Subsequently, slides were rinsed with neutralizing buffer (0.4 M Tris, pH 7.5) and stained with PI (3 μg/mL) for 20 min. The excess dye was removed with distilled water (5 min). Then, samples were stored in humidified chambers until the analysis by LUCIA Comet Assay software. One hundred cells per replicate of each treatment and time interval were analyzed, and the mean of DNA content in 100 comet tails (%) was calculated as a parameter of DNA damage. Positive control (100 mM H2O2; 30 min incubation; Sigma-Aldrich, Steinheim, Germany) was included with the assay. The comet assay was repeated in three independent experiments with three replicates for each treatment and time interval.

#### *2.6. mRNA Levels Quantification*

Cells were incubated with TiO2 NPs (1 and 10 μg/mL) for 2, 6, and 24 h. Cellular RNA was isolated using the RNAqueous®-Micro Kit (Invitrogen, Vilnius, Lithuania). RNA (500 ng) was reverse-transcribed with the Oligo(dT)12–18 primer and Superscript

IV Reverse Transcriptase (Life Technologies). Non-RT controls were included to show the elimination of gDNA contamination.

Quantitative PCR (CFX96 Touch™ Real-Time PCR detection System, Bio-Rad) was performed to detect changes in mRNA levels encoding proteins participating in metal detoxification (metallothionein, phytochelatin), oxidative stress (manganese superoxide dismutase, Mn-SOD; copper-zinc-superoxide dismutase CuZn-SOD; catalase), immunity (endothelial monocyte-activating polypeptide II, EMAPII; fetidin/lysenin, and lumbricin), and signal transduction (MEK kinase I, MEKK I; and protein kinase C I, PKC I). Sequences of primers used in qPCR assays are referred in Table S1. The PCR reactions were performed in a 25 μL volume containing 4 μL of cDNA (dilution 1:10, except for 1:5 dilution for SODs). The cycling parameters were similar to Roubalova et al., with slight changes [24]: 4 min at 94 ◦C, 35 cycles of 10 s at 94 ◦C, 25 s at 60 ◦C (at 58 ◦C for MEKK I, PKC I, and catalase), 35 s at 72 ◦C, and a final extension for 7 min at 72 ◦C. Gene expression changes were calculated according to the 2−ΔΔCT (Livak) method. Two reference genes (RPL13, RPL17) were selected as internal controls for gene expression normalization. Non-template control was included in each experiment. The fold change in the mRNA level was related to the change of the corresponding controls. The results were expressed as the mean ± SEM of the values. mRNA levels quantification was performed by three independent experiments with duplicates per each treatment and time interval.

#### *2.7. Statistical Analyses*

Statistical analyses were performed using GraphPad Prism (8.3.1 version, San Diego, CA, USA). Flow cytometry assays, lipid peroxidation, alkaline comet assay, and gene expression were analyzed by two-way ANOVA with Bonferroni post-test.

#### **3. Results**

#### *3.1. TiO2 NPs Characterization*

TiO2 NPs were dispersed and stabilized in distilled water and in R-RPMI 1640 culture medium evenly. However, differences in NPs characteristics were observed between both mediums along the exposure time (2, 6, and 24 h) (Table 1). In the UV/Vis spectra, NPs exerted a similar wavelength range: 300–370 nm for distilled water; 320–380 nm for R-RPMI 1640 medium. Although NPs absorbed similar UV/Vis wavelengths, differences were observed in the hydrodynamic size distribution. TiO2 NPs dispersed in R-RPMI 1640 medium were not stabilized and tended to aggregate. The aggregation was detected between 6 and 24 h of incubation. After 6 h, the hydrodynamic size distribution was 35.5 ± 3.94 nm, and it increased to 597 ± 447 nm after 24 h. At 2–6 h, the hydrodynamic size of TiO2 NPs was stable (31.34 ± 1.55 to 35.5 ± 3.94 nm, respectively). In distilled water, the hydrodynamic size distribution was stable between 2–24 h 581 ± 23.30 and 480 ± 64.3 nm, respectively. Regarding zeta potential, TiO2 NPs dispersed in both distilled water and R-RPMI 1640 did not change significantly over time (Table 1).

**Table 1.** Characterization of 100 μg/mL TiO2 nanoparticles (NPs) suspension in milliQ water and R-RPMI 1640 medium.


(a) ultraviolet-visible = UV/Vis spectra absorbance (nm), (b) Z-Avg = Hydrodynamic size determined by multi-angle dynamic light scattering (MADLS), and (c) ζ = zeta potential values are expressed as mean of 3 measurements ± SD.

#### *3.2. Electron Microscopy*

The TiO2 NPs size given by the manufacturer was 10–65 nm. However, we were not able to verify this information because of a great aggregation of TiO2 NPs in concentrations detectable by TEM. According to our measurements, the nanoparticles ranged between 20 and 100 nm. The TiO2 NPs were rode/spherical (Figure 1).

**Figure 1.** Transmission electron microscopy of 500 μg/mL TiO2 NPs clustered in distilled water. The scale bar represents 200 nm.

In coelomocytes exposed to 100 μg/mL TiO2 NPs, nanoparticles were observed on cell membranes by scanning electron microscopy (Figure 2). Moreover, EDS microanalysis confirmed Ti presence in nanoparticle clusters on the cell surface (Figure 3). At 10 μg/mL TiO2 NPs exposure, nanoparticles were also detected on the coelomocyte surface but with lower frequency. EDS microanalysis of non-treated cells is shown in Figure S2.

**Figure 2.** Scanning electron microscopy of coelomocytes. (**A**) cells exposed to 100 μg/mL TiO2 NPs for 2 h; (**B**) control cells cultured in the medium. Images recorded with B + C segments of a CBS detector at 3 kV. White arrow indicates a TiO2 NPs cluster on sample support. Clusters of the same morphology can be seen on the cell surface (white double arrow). The scale bar represents 5 μm.

**Figure 3.** EDS microanalysis of coelomocytes incubated with 100 μg/mL TiO2 NPs. (**A**) An image showing the area of interest taken with EDX TEAM software at 15 kV using a SED detector. The spectra collection places are marked with EDS labels. Increased charging effects caused by the non-conductive nature of Thermanox coverslips used for sample preparation deteriorated image quality. (**B**) EDS microanalysis confirmed Ti in NPs clusters found on the cell surface (e.g., EDS Spot 2 label) and also in the cluster labeled EDS spot 1. Blue arrow indicates TiO2 NPs cluster (EDS Spot 2 label), green arrow points to the cell surface without NPs clusters (EDS Spot 4 label). Corresponding EDS spectra in matching colors are shown in B. The scale bar represents 5 μm.

#### *3.3. Flow Cytometry Assays*

Coelomocyte subpopulations were differentiated by flow cytometry (Figure S3). Thus, the viability of HA and GA were analyzed. The eleocyte subpopulation was excluded from the results because of the interaction between their autofluorescence and the fluorescences used in the assays.

HA and GA viability (the percentage of alive cells in each subpopulation) was similar in non-treated cells and TiO2 NPs-exposed cells. No differences in viability were observed between amoebocyte subpopulations.

We did not observe any significant changes in ROS production in HA or in GA after exposure to any of the TiO2 NPs concentrations (Figure 4). HA population exerted two times lesser fluorescence intensity in comparison to the GA population. This suggests that HA population is less potent to produce ROS than GA population (Figure 4). Illustrative histograms of ROS production between control samples and positive control (1 mM H2O2), indicating a clear shift in sample fluorescence, are shown in Figure S4.

**Figure 4.** ROS production by hyaline (HA) and granular (GA). ROS production was measured in HA and GA after incubation with 1, 10, and 100 μg/mL TiO2 NPs for 2, 6, and 24 h using a cell-permeant tracer 2 ,7 -dichlorofluorescein diacetate (DCF-DA). Coelomocytes were also exposed to 1 mM H2O2 (positive control) for 30 min. The results are shown as the mean of fluorescence intensity (DCF-DA) ± SEM of three independent experiments with 3 replicates in each. \*\*\* *p* < 0.001, and \* *p* < 0.05 according to two-way ANOVA and Bonferroni post-test.

Similarly, we did not detect any significant differences in the apoptosis level between TiO2 NPs exposed and control cells (both in HA and GA; Figures 5 and 6). In both populations, the early apoptosis percent is similar over time, while late apoptosis slightly decreased after 24 h (Figures 5 and 6). Necrosis increased along the exposure time in GA (Figure 6). However, statistically significant differences were not detected between treated and control cells. Representative distributions of the apoptotic/necrotic cell stages in GA and HA cell subpopulations are shown in Figures S5 and S6, respectively.

**Figure 5.** Early and late apoptosis, viability and necrosis of hyaline amoebocytes (HA). Early and late apoptosis, viability and necrosis of HA of non-treated cells, cells exposed to 1, 10, and 100 μg/mL TiO2 NPs after 2, 6, and 24 h. 10 mM H2O2 was used as positive control for 30 min exposure. The results are shown as mean (%) ± SEM of three independent experiments with 3 replicates in each. \*\*\* *p* < 0.001, and \*\* *p* < 0.01 according to two-way ANOVA and Bonferroni post-test.

**Figure 6.** Early and late apoptosis, viability and necrosis of granular amoebocytes (GA). Early and Late apoptosis, viability and necrosis of GA of non-treated cells, cells exposed to 1, 10, and 100 μg/mL TiO2 NPs after 2, 6, and 24 h. 10 mM H2O2 was used as positive control for 30 min exposure. The results are shown as mean (%) ± SEM of three independent experiments with 3 replicates in each. \*\* *p* < 0.01, and \* *p* < 0.05 according to two-way ANOVA and Bonferroni post-test.

The viable amoebocyte phagocytic activity was measured in both amoebocyte subsets (HA and GA). Representative phagocytic activity density plots of GA and HA cell subpopulations are shown in Figures S7 and S8, respectively. The phagocytic activity was similar in both amoebocyte subpopulations (GA and HA; Figure 7). A decrease in the phagocytic activity of HA control cells and TiO2 NPs-exposed cells occurred after 24 h (Figure 7). This slight decrease may indicate the greater sensitivity of HA to external conditions. However, phagocytic activity was not significantly affected by NPs treatment or by the incubation time. Phagocytic activity of untreated cells with and without Fluoresbrite® YG Plain 1μm microspheres was also compared (Figure S9).

**Figure 7.** Phagocytic activity of HA and GA. Phagocytic activity was measured after incubation with TiO2 NPs (1, 10, and 100 μg/mL) for 2, 6, and 24 h. Coelomocytes were also exposed to 10 mM H2O2 (positive control) for 30 min. Results are represented as the mean ± SEM of three independent experiments with 3 replicates in each. \*\*\* *p* < 0.001, \*\* *p* < 0.01, and \* *p* < 0.05 according to two-way ANOVA and Bonferroni post-test.

#### *3.4. MDA and Alkaline Comet Assay*

Malondialdehyde (MDA) is a lipid peroxidation subproduct, and it is therefore used as an oxidative stress biomarker in cells. We did not detect any significant increase in MDA production in cells exposed to TiO2 NPs (10 and 100 μg/mL) at the tested timepoints (2, 6, and 24 h; Figure 8).

**Figure 8.** Relative malondialdehyde (MDA) production in coelomocytes exposed to 10, 100 μg/mL TiO2 NPs and positive control (100 μg/mL CuSO4) for 2, 6, and 24 h. Values are expressed as mean (%) ± SEM of three independent experiments each with three replicates. \*\*\* *p* < 0.001 according to two-way ANOVA and Bonferroni post-test.

The DNA damage in coelomocytes exposed to 1, 10, and 100 μg/mL TiO2 NPs for 2, 6, and 24 h was assessed by the alkaline comet assay. DNA damage was evaluated by the mean tail intensity (% DNA in tail) of 100 comets in each incubation. The observed DNA damage was not greater than 40% during exposure with TiO2 NPs and the non-treated cells (Figure 9).

**Figure 9.** DNA damage in coelomocytes after their exposure to 1, 10, and 100 μg/mL TiO2 NPs for 2, 6, and 24 h. Coelomocytes were also exposed to 100 mM H2O2 (positive control) for 30 min. Values are expressed as the mean of DNA content in tail (%) ± SEM of three experiment with three replicates. \*\*\* *p* < 0.001 and \* *p* < 0.05 according to two-way ANOVA and Bonferroni post-test.

#### *3.5. mRNA Levels of Detoxification, Immune, Antioxidant, and Signal Transduction Molecules*

The change in mRNA levels of appropriate molecules after coelomocyte exposure to TiO2 NPs was assessed (Table 2). Metallothioneins involved in metal detoxification were significantly upregulated in coelomocytes exposed to 1 μg/mL TiO2 NPs for 2, 6 and 24 h, and in coelomocytes exposed to 10 μg/mL TiO2 NPs for 6 h. Further, significant Mn-SOD downregulation was detected in coelomocytes incubated with 10 μg/mL TiO2 NPs for 6 h. Then, fetidin/lysenin and lumbricin were upregulated upon coelomocyte exposure to 10 μg/mL TiO2 NPs for 6 h. MEKK I upregulation after 1 μg/mL TiO2 NPs exposure for 24 h, and PKC I downregulation after 10 μg/mL TiO2 NPs exposure for 6 and 24 h were detected. Surprisingly, the mRNA levels of catalase and CuZn-SOD (antioxidant enzymes) were not significantly altered.



Values were normalized to two reference molecules (RPL13 and RPL17). Fold changes (±SEM) in mRNA levels in TiO2 NPs exposed coelomocytes are relative to the mRNA levels in control cells. Two-way ANOVA and Bonferroni post-test were performed to evaluate data significance (\* *p* < 0.05, \*\* *p* < 0.01). mRNA levels quantification was performed by three independent experiments with duplicates per each treatment and time interval. Mn-SOD: manganese superoxide dismutase; CuZN-SOD: copper-zinc-superoxide dismutase; EMAP II: endothelial monocyte-activating polypeptide-II; MEKK I: MEK kinase I; PKC I: protein kinase C I.

#### **4. Discussion**

The physico-chemical properties of TiO2 NPs were analyzed in R-RPMI 1640 medium to understand their behavior in cell cultures. The analyses in distilled water were performed to observe possible changes in nanoparticles behavior in the stock over time. UV/Vis spectra, hydrodynamic size, zeta potential, and TEM were used for the NPs characterization. TiO2 NPs dispersed in distilled water showed an aggregation behavior at the greatest concentration (100 μg/mL), and the size remained approximately the same between 2 and 24 h. The zeta potential was also stable at all TiO2 NPs concentrations (Table 1). However, different NPs behavior was observed when dispersed in the R-RPMI 1640 culture medium. Between 2 and 6 h of incubation, changes were not observed in the size distribution, while zeta potential indicated instability (Table 1). Between 6 and 24 h, we observed a great increase in particle size distribution in comparison with previous intervals. These changes indicate that NPs were dispersed in R-RPMI 1640 medium, and they started to precipitate only after 6 h of incubation. Magdolenova and colleagues assessed the

relationship between the cytotoxic effects and the dispersion of TiO2 NPs [25]. They showed that tested cell culture medium types did not influence TiO2 NPs dispersion. However, they observed that different dispersion protocols and the use of serum in stock solution affected nanoparticles aggregation and size distribution. Accordingly, Ji et al. showed the improvement in TiO2 NPs dispersion upon addition of bovine serum albumin (BSA), although the dispersion also depended on cell culture media phosphate concentration [26]. TiO2 NPs tended to aggregate in R-RPMI 1640 medium (Table 1), which may be related to the low FBS concentration or the effect of phosphate ions in the cell culture medium.

By TEM, the aggregation of 500 μg/mL TiO2 NPs was also observed in distilled water (Figure 1). Therefore, it was not possible to determine the nanoparticles' size. UV/Vis spectra were similar in exposed and control samples in both distilled water and R-RPMI 1640, as well as during the experiment, indicating that NPs properties did not change. Previously, the addition of HEPES and FBS into RPMI-1640 medium led to NPs re-dispersion [14].

Scanning electron microscopy showed the NPs cluster in contact with the cell membranes (Figure 2). EDS spectra showed TiO2 NPs that are present on cells at the 100 μg/mL concentration (Figure 3), but not at the lesser concentration (10 μg/mL). This may be because of the EDS microanalysis detection limit. TiO2 NPs are internalized by *E. fetida* coelomocytes. Bigorgne et al. determined their presence in the cell cytoplasm, but not in the nucleus or mitochondria [14]. However, we were unable to detect TiO2 NPs inside coelomocytes. This could be because TiO2 NPs aggregates are large. Earthworm coelomocytes are probably unable to engulf large NP clusters via phagocytosis and/or endocytosis, the most probable routes of TiO2 NPs entry into coelomocytes [1,14]. Phagocytic cells are potentially the most affected because they engulf NPs. Coelomocyte viability was not affected by exposure to 1, 10, and 100 μg/mL TiO2 NPs for 2, 6, and 24 h. Similar results were observed in *E. fetida* coelomocytes exposed to TiO2 NP [14]. Nanoparticles often trigger reactive oxygen species (ROS) production in cells, resulting in biomolecule oxidative damage [27,28]. We did not detect any statistically significant differences in ROS production in TiO2 NPs-exposed cells in comparison with control cells (Figure 4). Cells exposed to other nanoparticles, such as Ag NPs, nZVI NPs or ZnO NPs release significantly greater ROS amounts. Contrary to TiO2 NPs, ROS production could be elicited by the metal ions released from these nanoparticles [11,29,30].

We evaluated the apoptotic process in cells treated with TiO2 NPs, and did not detect any significant differences between exposed and control coelomocytes (Figures 5 and 6). Late apoptosis was similar in GA and HA, with the greatest difference observed after 24 h of incubation. HA population exerted relatively greater early apoptosis than GA. Excess ROS production led to decreased cell viability and apoptosis [31,32]. Homa et al. suggested that coelomocytes are susceptible to bacterial or fungal products that may induce programmed cell death [31]. TiO2 NPs did not increase ROS production, and simultaneously, apoptosis was not increased as compared to control cells (Figures 4–6). We suggest that TiO2 NPs do not affect ROS production, and thus do not trigger the apoptotic pathway in amoebocyte subpopulations (HA and GA).

Amoebocytes are earthworm immune effector cells with the ability to phagocytose. At the phagocytic activity level, control cells and cells exposed to TiO2 NPs (1, 10, and 100 μg/mL) did not show any statistically significant changes (Figure 7). The results are in accordance with Bigorgne et al., who reported that there were no phagocytic activity changes in coelomocytes exposed to 1, 5, 10, and 25 μg/mL of TiO2 NPs, although TEM images demonstrated that TiO2 NPs were engulfed by the coelomocytes [14]. Thus, we can confirm that phagocytic activity is not compromised due to TiO2 NPs exposure.

ROS production initiates harmful radical chain reactions on cellular macromolecules, including DNA mutation, protein denaturation, and lipid peroxidation. At the lipid peroxidation level, MDA production was similar in both exposed and control cells. MDA is a subproduct derived from the reaction of free radical species with fatty acids [33]. We did not observe elevated lipid peroxidation (Figure 8). Ayala et al. explained that MDA is more

stable and has a greater lifespan than ROS, and therefore it is more toxic [33]. Therefore, it could be a better biomarker for cellular oxidative stress detection. Excess ROS leads to MDA production [34]. Two oxidative stress markers, ROS and MDA, were produced at similar levels in control cells and TiO2 NPs-exposed cells (Figures 4 and 8). The same results were also observed after THP1 human cells and sea urchin cells were exposed to TiO2 NPs [20,35]. UVA light could also enhance ROS production and increase toxicity several fold [36]. However, in this instance, the cells were mimicking the environmental conditions in the soil ecosystem, where UVA light was not present.

Significant differences between exposed and control cells were not detected regarding DNA damage. The alkaline comet assay results showed that there is no significant DNA damage in coelomocytes exposed to 1, 10, and 100 μg/mL of TiO2 NPs for 2, 6, and 24 h (Figure 9). A relationship between ROS, MDA, and DNA damage has been suggested. As mentioned previously, ROS may induce MDA production, which, in turn, affects nucleosides and results in DNA damage [33,34]. This mechanism has been described in coelomocytes exposed to pollutants, antibiotics, or pathogens [34]. Reeves et al. showed that GFSk-S1 cells (primary cell line from goldfish skin) exposed to different doses of TiO2 NPs (1, 10, and 100 μg/mL) could result in slight DNA damage, whereas co-exposure with UVA caused a significant increase in toxicity [36]. In vitro analysis described in this study did not reveal substantial changes in cellular physiologic activities, but the long-term exposure experiments can reveal different findings [37]. Zhu et al. described transcriptomic and metabolomic changes in earthworms as a global response to TiO2 NPs exposure that cannot be observed by conventional toxicity endpoints [38].

Treatment of coelomocytes with TiO2 NPs induced slight changes in the mRNA levels of distinct molecules. Metallothioneins are proteins protecting against metal-induced oxidative stress [9]. Metallothionein was upregulated in coelomocytes exposed to 10 μg/mL TiO2 NPs for 6 h, respectively (Table 2). This is in agreement with Bigorgne et al., who immunostimulated coelomocytes with lipopolysaccharides (LPS) (500 ng/mL) for 5 h prior to TiO2 NPs addition. After 12 h of incubation with 10 and 25 μg/mL TiO2 NPs, metallothioneins were upregulated [14]. We determined that even 1 μg/mL TiO2 NPs concentration upregulated metallothionein expression during the whole experiment (Table 2). Interestingly, the highest upregulation was detected in coelomocytes incubated with 1 μg/mL TiO2 NPs already after 2 h. Further, the induction of metallothionein expression in cells exposed to 10 μg/mL TiO2 NPs started at 6 h, and afterward decreased after 24 h (Table 2). Bigorgne et al. similarly showed increased metallothioneins expression after 12 h of incubation, with a subsequent decrease after 24 h [14].

Further, the antioxidant enzymes were not affected, except for Mn-SOD, which was downregulated after 6 h of coelomocyte exposure to 10 μg/mL TiO2 NPs (Table 2). Mn-SOD is a mitochondrial protein that protects cells against oxidative stress [39]. It seems that macrophages (RAW 264.7 cell line) and coelomocytes can engulf TiO2 NPs. These nanoparticles affect mitochondria even if they are not located inside the mitochondria [12,14]. Moreover, TiO2 NPs decreased ATP production in the macrophage RAW 264.7 cell line [12]. Thus, engulfed TiO2 NPs could target mitochondria and cause mitochondrial malfunction [12]. Mn-SOD downregulation and loss in mitochondrial oxidative phosphorylation function was also reported in primary rat hepatocytes [40].

Elevated levels of the antimicrobial proteins fetidin/lysenin and lumbricin were detected in cells exposed to 10 μg/mL TiO2 NPs for 6 h (Table 2). Similarly, Bigorgne et al. observed that fetidin was upregulated in cells exposed to 10 μg/mL TiO2 NPs after 12 h [14]. As previously described in related earthworm species *E. fetida*, lysenin regulation is changed rapidly by environmental stressors and it is suggested as an early biomarker of stress [41]. However, we cannot exclude that the increase in antimicrobial protein mRNA levels could be caused by used TiO2 NPs that were not LPS-free.

Referring to the signal transduction molecules, PKC I was strongly downregulated after coelomocyte exposure to 10 μg/mL TiO2 NPs for 6 and 24 h (Table 2). PKC I is important in cellular homeostasis and is involved in the cell proliferation signaling cas-

cade [42,43]. This downregulation could suggest a coelomocyte homeostasis destabilization upon TiO2 NPs exposure. Another signal transduction molecule, MEKK, was upregulated in coelomocytes exposed to 1 μg/mL TiO2 NPs for 24 h and in coelomocytes exposed to 10 μg/mL TiO2 NPs for 6 h (Table 2). This molecule is involved in the MAPK cascade participating in many cellular processes, besides others in stress signaling [9,43]. Generally, coelomocyte exposure to TiO2 NPs results in slight changes in the mRNA levels of various molecules, however, these changes seem not to be significant enough to affect the observed cellular functions.

#### **5. Conclusions**

Coelomocytes exposed to TiO2 NPs (1, 10, and 100 μg/mL) did not show any impaired cellular responses as compared to control cells. The oxidative stress pathway and phagocytic activity were not affected as well. Nanoparticles do not cause greater DNA damage in treated cells than in non-treated cells. We also detected some gene expression alterations involved in metal detoxification, oxidative stress, defense reactions, and signal transduction. However, these changes do not seem to affect the observed cellular functions. In summary, we did not determine any detrimental effects of TiO2 NPs on *E. andrei* coelomocytes.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2079-4 991/11/1/250/s1, Figure S1. Illustrative figure of NPs distribution incubated without and with the cells. Figure S2. EDS spectra from standard non-treated cells. Figure S3. Illustrative figure of coelomocytes subpopulations detected by flow cytometry. Figure S4. Illustrative histogram of ROS production between control samples and positive control. Figure S5. Illustrative figure of apoptosis of GA after 24 h of exposure to 100 μg/mL TiO2 NPs. Figure S6. Illustrative figure of apoptosis of HA after 24 h of exposure to 100 μg/mL TiO2 NPs. Figure S7. Illustrative figure of phagocytic activity of GA after 2 h. Figure S8. Illustrative figure of phagocytic activity of HA after 2 h. Figure S9. Detection of Fluoresbrite® YG Plain 1μm microsphere. Table S1: Primer sequences used for qPCR.

**Author Contributions:** Conceptualization, N.I.N.P. and P.P.; methodology and data curation, N.I.N.P., J.S., J.D., O.B., A.G. and O.K.; writing—original draft preparation, N.I.N.P., J.S., R.R.; writing—review and editing, N.I.N.P., J.S., O.B., R.R. and P.P.; supervision, R.R. and P.P.; project administration, P.P., M.B. and T.C.; funding acquisition, P.P. and T.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by Grant QK1910095 of the Ministry of Agriculture of the Czech Republic and by the Center for Geosphere Dynamics (UNCE/SCI/006). This project also received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 67188.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article or supplementary material.

**Acknowledgments:** The authors gratefully acknowledge the access to the electron microscopy facility, supported by the project LO1509 of the Ministry of Education, Youth and Sports of the Czech Republic. The authors also acknowledge the Cytometry and Microscopy Facility at the Institute of Microbiology of the ASCR, v.v.i. for the support of the staff and the use of their equipment, and we acknowledge the support of CMS-Biocev Biophysical techniques (LM2015043 funded by MEYS CR).

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **Stressor-Dependant Changes in Immune Parameters in the Terrestrial Isopod Crustacean,** *Porcellio scaber***: A Focus on Nanomaterials**

**Craig Mayall 1, Andraz Dolar 1, Anita Jemec Kokalj 1, Sara Novak 1, Jaka Razinger 2, Francesco Barbero 3, Victor Puntes <sup>3</sup> and Damjana Drobne 1,\***


**Abstract:** We compared the changes of selected immune parameters of *Porcellio scaber* to different stressors. The animals were either fed for two weeks with Au nanoparticles (NPs), CeO2 NPs, or Au ions or body-injected with Au NPs, CeO2 NPs, or lipopolysaccharide endotoxin. Contrary to expectations, the feeding experiment showed that both NPs caused a significant increase in the total haemocyte count (THC). In contrast, the ion-positive control resulted in a significantly decreased THC. Additionally, changes in phenoloxidase (PO)-like activity, haemocyte viability, and nitric oxide (NO) levels seemed to depend on the stressor. Injection experiments also showed stressor-dependant changes in measured parameters, such as CeO2 NPs and lipopolysaccharide endotoxin (LPS), caused more significant responses than Au NPs. These results show that feeding and injection of NPs caused an immune response and that the response differed significantly, depending on the exposure route. We did not expect the response to ingested NPs, due to the low exposure concentrations (100 μg/g dry weight food) and a firm gut epithelia, along with a lack of phagocytosis in the digestive system, which would theoretically prevent NPs from crossing the biological barrier. It remains a challenge for future research to reveal what the physiological and ecological significance is for the organism to sense and respond, via the immune system, to ingested foreign material.

**Keywords:** gold nanoparticles; cerium nanoparticles; woodlice; immune response; haemocyte

#### **1. Introduction**

The role of the immune system is to distinguish between the self and non-self and to choose the most effective response to exogenous or endogenous threats, in order to maintain the integrity and homeostasis of the organism [1–3]. The immune recognition and response mechanisms of crustaceans depend entirely on the innate immune system.

The most prominent and well-studied cellular effector reactions of the innate immune system are phagocytosis, nodulation, encapsulation, cell-mediated cytotoxicity, and clotting. Cellular responses act in conjunction with humoral factors [4].

In crustaceans, in response to natural infections and challenges that mimic natural infections, such as lipopolysaccharide endotoxin (LPS), the response mechanisms and the recognition system that governs them are well -studied [5–10]. However, much less is known about the immune effector reactions in response to other challenges, such as pollutants [11]. Recently, the research on adverse or beneficial interactions between engineered nanomaterials and model organisms has provided a new model system to study if and how organisms sense and respond to novel engineered material of sizes comparable to those of viruses or bacteria [3].

**Citation:** Mayall, C.; Dolar, A.; Jemec Kokalj, A.; Novak, S.; Razinger, J.; Barbero, F.; Puntes, V.; Drobne, D. Stressor-Dependant Changes in Immune Parameters in the Terrestrial Isopod Crustacean, *Porcellio scaber*: A Focus on Nanomaterials. *Nanomaterials* **2021**, *11*, 934. https:// doi.org/10.3390/nano11040934

Academic Editor: Diana Boraschi

Received: 6 February 2021 Accepted: 27 March 2021 Published: 6 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Terrestrial isopods are a valuable model organism to study defences against challenges to homeostasis, as isopoda are the only order of crustacean able to occupy freshwater, marine, and terrestrial environments, which indicates that they have been exposed to a large variety of stressors during their evolution [12].

Isopods, as well as other arthropods, primarily rely on the cuticle as a sensing and defence system and the innate immune system to cope with different challenges to homeostasis. The cuticle is the relatively thin but tough and flexible layer of noncellular material that is capable of inducing many biochemical processes and able to respond to different environmental cues [13].

Haemocytes play a central role in the internal defence of the animal, although their role and function are not solely limited to purely immunogenic responses. In oysters, haemocytes were seen to participate in calcium carbonate shell crystal production, transportation, and shell regeneration [14,15]. Moreover, haemocytes in a range of organisms have been found to be involved in muscle fibre degeneration and regeneration; adult neurogenesis; and the digestion, storage, and distribution of nutrients [16–19]. The total haemocyte count (THC) is traditionally taken as a measure of an organism's change in immunocompetence and is used in assessing the stress or health status in crustaceans [20–23]. It has been shown that the THC can vary not only in response to infection but also due to environmental stresses and endocrine activity during the moulting cycle [24,25].

An integral component of the innate immune system is the extracellular phenoloxidase (PO) cascade, which is part of humoral response that catalyses the formation of cytotoxic intermediates quinones, essential precursors for melanisation and sclerotisation [26–29]. Melanin accumulates at wound sites and around invading microorganisms [30]. It is not only involved in innate immune responses through melanotic encapsulation, it is also critical for cuticle tanning, which has been extensively studied in insects [19]. The PO cascade also has a role is ecdysis and exoskeleton formation [28].

The primary source of PO activity comes from the haemocytes [8,31]. However, prophenoloxidase (proPO) transcripts have been detected in nonhaemocyte cells too, including the hepatopancreas, stomach epithelium, anterior midgut caecum, glia cells in the nervous cord, and neurosecretory cells in ganglions [27,32]. Besides the plasma PO enzyme, literature also reports the existence of haemocyanin-derived PO (Hd–PO) activity [33,34]. Therefore, in isopods, PO-like activity in the haemolymph is usually shown, because, at this point, it is impossible to distinguish whether the observed activity is purely due to the plasma PO enzyme or haemocyanin-derived [5,33].

Activity of PO is taken as one of most frequently used parameters of assessing immune system activity. Phenoloxidase (PO) activity is increased as a result of clotting process's, phagocytosis, encapsulation of foreign material, antimicrobial action, and cell agglutination [35]. The moult cycle, for example, is also a source of variation of phenoloxidase activity and correlates with the presence of large granular cells, although this appears to be species-specific [36,37]. Correlative measures of phenoloxidase (PO) activity and resistance to infection against vibriosis have been also investigated during the moult cycle [38]. In addition, it is documented in insects that, when soft cuticles are damaged, the pro-enzymes can become activated, [39] and the active enzymes can take part in both wound healing and in defending against invading microorganisms. The PO cascade not only results in the formation of cytotoxic pigment precursors, including quinones, quinone methides, and semiquinones, but also reactive oxygen species (ROS) and reactive nitric species (RNS), such as nitric oxide (NO) [40].

The concentration of NO is another well-studied and often used measure of immune system activity. The first evidence for NO to have a role as an humoral molecule in an invertebrate was provided by Radomski et al. (1991) [41]. Both RNS and ROS are produced by activated phagocytic cells. These molecules are involved in the breaking down of phagocytosed infectious agents such as viruses and bacteria, as well as dead and dying cells [42–44]. Otherwise, at lower concentrations, NO is involved in diverse housekeeping functions, such as neurotransmission, mucus secretion, and in controlling certain activities of the circulatory system, while at higher concentrations, NO is cytotoxic [45].

In the work presented here, we have used different types of immune challenges, i.e., lipopolysaccharide (LPS) and two types of nanoparticles and metal salt. LPS is a constituent of the cell membrane of Gram-negative bacteria. It has been widely used in investigating the immune function of a spectrum of model organisms [46–49]. For nanoparticles (NPs), two particles with well-known and frequently studied biological interactions were selected. Metal salt was used as a positive control to compare the effect of different forms of a metal (ions and nanoparticles).

Gold nanoparticles (Au NPs) have been used extensively in different experimental studies as their size, shape, and surface chemistry can be readily manipulated [50]. Due to their unique properties, they have a wide potential for applications in industry and biomedicine [50,51]. Numerous previous studies have found Au NPs to be relatively inert and nonharmful to organisms [52,53]. While some authors have suggested Au NPs have anti-inflammatory and antiangiogenic effects [54–56], others have shown Au NPs have toxic potential and can modulate immune processes in vitro and in vivo in some studies [50,54,57].

CeO2 NPs also represent nanoparticles with great industrial potential, due to their catalytic and redox properties [58,59]. Compared to Au NPs, they are more often considered as biologically potent and can induce oxidative stress; inflammation; DNA damage; and affect reproductive capability, growth, and the survival of exposed animals [59,60]. In the study of Kos et al. (2017), honeybees were exposed to the very same CeO2 NPs as in the present study. Significant alterations in Acetylcholinesterase and glutathione-S-transferase activities were evidenced at concentration 2 mg/L. However, in another experiment testing the toxicity of Au NPs and CeO2 NPs on earthworms, where the same exposure duration (as this paper) was used, CeO2 NPs showed a low toxic potential [59]. CeO2 NPs also possess dual properties, contrary to adverse effects some studies report on their antioxidant and anti-inflammatory properties [61,62].

Both CeO2 NPs and Au NPs have low dissolution rate [54,58]; therefore, the particles, and not the dissolved ions, are primarily responsible for the biological effect.

We provide a comparison of the responses of the innate immune system of terrestrial isopod *P. scaber* to different types of challenges to homeostasis (Au and Ce nanoparticles, salt, and LPS). We have analysed and compared some well-studied immune parameters (THC, haemocyte viability, PO-like activity, and NO concentration) after dietary exposure and in the case of body injection [63].

The aim of our work was to investigate if and how the most well-known and studied parameters of the immune system respond to nanomaterials in the food and to compare this response to that when nanomaterials are injected directly into the haemolymph of a terrestrial isopod, *Porcellio scaber*. Our focus was also directed towards comparing the response to nanoparticles vs. ions in the feeding experiment and towards comparing the response to nanoparticles vs. LPS in body injection experiments. We hypothesise that the response to NPs directly injected into the haemolymph or consumed with food is substantially different. We discuss the differences in response to the selected types of noninfectious immune system challenges in a model terrestrial invertebrate species.

#### **2. Materials and Methods**

#### *2.1. Nanoparticle Synthesis and Characterisation*

Au NPs were synthesised by the Catalan Institute of Nanoscience and Nanotechnology (ICN2). Au NPs (26.4 ± 3 nm) were synthesised according to the seeded-growth method developed and detailed in Bastús et al. (2011) [64]. Au NPs were purified and surface coated in PVP following the procedure outlined in Alijagic (2020) to produce PVP–Au NPs, referred to in this paper as Au NPs [65]. Dynamic light scattering (DLS) analysis showed the particles had an average hydrodynamic diameter of 67.2 nm, and laser doppler anemometry showed a Z potential of−3.9 ± 0.8 mV (pH 5.5, conductivity 0.9 mS/cm) (Malvern Zetasizer Nano ZS, Malvern Panalytical Ltd, Worcester, UK). Diameters were reported as distribution by intensity calculated by non-negative least squares (NNLS). A typical UV–Vis spectrum profile with a surface plasmon resonance band peaking at 525 nm showed the particles were monodispersed (Agilent Cary 60 spectrophotometer, Santa Clara, CA, USA). Tetrachloroauric (III) acid trihydrate (99.9% purity), sodium citrate tribasic dihydrate (≥99%), and polyvinylpyrrolidone (55 KDa) were purchased from Sigma– Aldrich. Particle characterisation can be found in Figure A1.

Stabiliser-free, uncoated spherical CeO2 NPs as an aqueous dispersion in dH2O (batch number PROM-CeO2-20 nm-2306/5a) were supplied by NanoMILE PROM (Promethean Particles, Nottingham, UK, http://www.prometheanparticles.co.uk/ 2 April 2021) within the framework of the EU FP7 NanoMILE project. The CeO2 NPs were synthesised using supercritical fluid synthesis, followed by a washing step postsynthesis to remove unreacted species. The mean particle diameter (TEM) was 4.7 ± 1.4 nm (JEOL JEM2100F, Tokyo, Japan), the Z-average size was 172.1 ± 1.705 nm, the polydispersity index (PDI) was 0.272 ± 0.009, and the zeta potential was 50.3 ± 0.719 mV (Malvern Zetasizer 5000, Malvern Panalytical Ltd, Worcester, UK). Further particle characterisation can be found in Kos et al. (2017) [66].

#### *2.2. Experimental Animals*

*Porcellio scaber* were collected from a compost heap in noncontaminated, pollution-free garden in Kamnik, Slovenia. Prior to the experiment, animals were cultured for a several months under constant temperature (20 ± 2 ◦C) and illumination (light:dark 16:8 h) regime in a climate-controlled chamber at the University of Ljubljana. Animals were kept in a glass terrarium filled up with a humus soil (moistened at 40% of the water holding capacity; WHC) and fed with dry leaves of common hazel (*Corylus avellana*), common alder (*Alnus glutinosa*), and carrots, as described by Jemec Kokalj et al. (2018) [67]. Adult animals of both sexes with body mass greater than 25 mg were selected for the experiment, while individuals with evident signs of moulting, gravid females, and those with symptoms of bacterial or viral infection were excluded [5]. We have followed the the ARRIVE (Animal Research: Reporting of *In Vivo* Experiments) guidelines (http://www.nc3rs.org.uk/page. asp?id=1357) for reporting experiments with using living animals. Experiments with terrestrial isopods do not need to be approved by the ethics committee.

#### *2.3. Feeding Experiment*

The feeding experiment was carried out for 14 days in petri dishes with a single animal at constant temperature (20 ± 2 ◦C), illumination (light:dark 16:8 h) regime, and moisture in a climate-controlled chamber, a semichronic exposure period. The animals were fed dried leaves covered with either cerium oxide nanoparticles (CeO2 NPs), gold nanoparticles (Au NPs), or ionic gold (AuCl3) as salt control or pure dried leaves, which served as controls. Nanoparticles and metal ions were spread on a leaf at a concentration of 100 μg/g of dry weight (d.w.) leaf. This concentration was chosen, as it is lower than the previously observed lowest effect concentration yet higher than the estimated soil compartment concentration, in order to provide an example of the model response [68–71]. Animals' weights were taken before and after the experiment; additionally, dry leaf weights after nanoparticle application at the start and end of the experiment were taken. During the experiment, animal faeces were removed from the petri dishes every 48 h in order to measure defecation rate and to reduce coprophagy. At the end of the 14-day experiment, fresh haemolymph was collected from the animals, according to Dolar et al. (2020), to measure the selected immune parameters [5]. Details on the number of animals used can be found in Table A1.

#### *2.4. Injection Experiment*

For injection experiments, animals were injected with 0.5 μL of the lipopolysaccharide (LPS), CeO2 NPs, Au NPs or Milli-Q water as a trauma control, using a 25 μL Hamilton microsyringe (700 series, 33 gauge, blunt tip) and a repeating dispenser (PB 600-1, Hamilton, Bonaduz, Switzerland). Cerium and Au NPs were suspended in Milli-Q water at a concentration of 0.3 μg/μL. A nonlethal dose of LPS (from *E. coli* O111:B4, Sigma-Aldrich, St. Louis, MO, USA) was used as a positive control for reference immune response (50 μg/μL). Injected animals were left for 48 h in a petri dish with a dry leaf for food at a constant temperature (20 ± 2 ◦C) and illumination (light: dark 16:8 h) regime and moisture in a climate-controlled chamber. Afterwards, fresh haemolymph was collected from the animals to measure the selected immune parameters. Details on the number of animals used can be found in Table A2.

#### *2.5. Haemolymph Collection*

A sterile syringe needle was used to puncture the intersegmental membrane on the dorsal side of the animal between the 5th and 6th segment; then a drop of haemolymph was withdrawn using glass microcapillary pipette. Where not enough haemolymph could be collected from a single animal, a pooled sample was prepared for nitric oxide and phenoloxidase-like measurements.

#### *2.6. Immune Parameters*

To determine haemocyte viability and total haemocyte count (THC), 3–5 μL of freshly collected haemolymph was diluted in 100 mM PPB (pH 7) up to a volume 28 μL, and then 2 μL of nigrosin dye was added (final concentration of 80 μg/mL), which stains dead cells, while viable haemocytes remain unstained. After that, ten microliters of haemocyte suspension was loaded on to each chamber of the Neubauer haemocytometer to assess haemocyte viability and THC under a phase contrast microscope (Axio Vert.A1, Zeiss; magnification: 40×, Oberkochen, Germany). In order to fulfil the requirement for a statistically significant count, at least 100 cells per square were counted (according to the instructions of the haemocytometer). Each haemolymph sample measurement was performed in duplicate.

Nitric oxide (NO) levels in the haemolymph were measured using Griess reagent after a modified protocol Faraldo et al. (2005) [72]. Twenty microliters of a pooled haemolymph sample, collected from an average of 4 animals, was placed into 28 μL of 1% sulphanilamide dissolved in 5% phosphoric acid (H3PO4) and kept on ice. After that, 20 μL of this solution was diluted at a ratio of 1:1 (v:v) with 1% naphthylethylenediamine dihydrochloride (NEED, Sigma). The absorbance of NO was measured in a 384-well assay plate at 543 nm after a 5-min incubation period using a Cytation 3 imaging reader (Biotek, Winooski, VT, USA). Each haemolymph sample measurement was done in duplicate. NO concentrations (μM) were calculated from a standard curve for NaNO2 (2.5–200 μM).

Phenoloxidase (PO)-like activity was assessed photometrically in the haemolymph obtained from 1–3 animals, after a modified protocol described by Jaenicke et al. (2009) [34]. Five microliters of haemolymph was placed into 200 μL of Dulbecco's phosphate buffered saline buffer (DPBS, pH 7.1–7.5), containing 4 mM dopamine hydrochloride (Sigma-Aldrich, St. Louis, MO, USA) and 2 mM sodium dodecyl sulphate (SDS, Sigma), necessary for in vitro PO activation. Forty microliters of this reaction mixture was pipetted into 384-well plate (Greiner Bio-One, Kremsmünster, Austria). The formation of a reddish–brown pigment, which is nonenzymatically synthesised from the dopamine catalytic product dopaquinon, was measured using a Cytation 3 imaging reader (Biotek, Winooski, VT, USA) at 475 nm and 25 ◦C for at least three hours. Haemolymph PO-like activity was calculated as the change in absorbance from the linear part of the absorbance slope per minute per haemolymph volume, normalised to control and expressed as a percentage [73]. PO-like activity assessment was done in duplicate for each haemolymph sample.

#### *2.7. Statistical Analysis*

Statistical analysis of the data was performed using OriginPro v2020 software program OriginLab, Northampton, MA, USA). The data was tested using Kruskal–Wallis

test, followed by pairwise comparison of treatments with control group, using Mann– Whitney U-test. Values of *p* < 0.001 (\*\*\*), *p* < 0.01 (\*\*) and *p* < 0.05 (\*) were considered as significantly different.

#### **3. Results**

#### *3.1. Feeding Activity of Isopods and Mortality*

The feeding activity of isopods, calculated as the mass of leaves ingested per each animal's mass, in two weeks, was not significantly altered in animals that were fed CeO2 NP- or Au NP-spiked leaves. In addition, Au ions (AuCl3) as a salt control for Au NPs did not induce any changes in feeding activity. No signs of animal mortality were evidenced (data not shown) (Figure 1). These data show that the tested concentration (100 μg/g d.w. leaf) did not induce significant adverse effects and can be considered not toxic.

**Figure 1.** Feeding activity of isopods in comparison to control (%) after 14 days of feeding on CeO2 nanoparticle (NP)-, Au NP-, and Au ions-spiked leaves. Columns represent the mean (±SE). n = number of biological repeats in each column.

#### *3.2. NO Levels and PO-Like Activity*

Nitric oxide (NO) levels were significantly increased after the animals were fed or injected with CeO2 NPs, while Au NPs caused increases in NO levels only when injected. The observed increase was more significant in the case of CeO2 NPs, in comparison to Au NPs (Figure 2a).

PO-like activity was significantly increased in the case of both CeO2 NPs and Au NPs after feeding and injection. The increase was again more significant in response to CeO2 NPs (Figure 2b). The injection procedure, itself, had no effect on NO levels or PO-like activity, as evidenced by no significant changes in the trauma controls. Injection with LPS, a positive control, significantly increased the NO levels and caused a decrease in PO-like activity.

#### *3.3. Haemocyte Count*

The total haemocyte count (THC) was significantly increased in the case of CeO2 NPs and Au NPs, in comparison to control after feeding, but decreased in the case of Au ions. However, after injections, neither CeO2 NPs nor Au NPs affected the THC. The viability of the haemocytes was decreased after both feeding and injection CeO2 NP exposures, but Au treatments had no effect on viability after either exposure route. Injection, itself (evidenced

as trauma control), had no effect on the THC or viability of haemocytes; however, both THC and viability were affected after LPS injection, serving as a positive control (Figure 3).

**Figure 2.** (**a**) Nitric oxide concentrations and (**b**) phenoloxidase (PO)-like activity in *P. scaber* exposed to CeO2 NP- and Au NP-spiked leaves (Feeding exp.) and injected with CeO2 NPs and Au NPs dispersion (Injection exp.). Au ions correspond to AuCl3, trauma control are animals injected with dH2O and lipopolysaccharide endotoxin (LPS); animals injected with 50 μg/μL LPS. Columns represent the mean (±SE). n represents the number of analysed samples. For nitric oxide (NO) measurements, on average, 4–5 animals were grouped into one sample, and for the PO-like measurements, 1–3 animals were grouped into 1 sample. (\*) *p* < 0.05, (\*\*) *p* < 0.01, (\*\*\*) *p* < 0.001 in comparison to control (Kruskal–Wallis test followed by Mann–Whitney U-test).

**Figure 3.** Total haemocyte count (THC) and haemocyte viability in *P. scaber* exposed to CeO2 NP- and Au NP-spiked leaves (Feeding exp.) and injected with CeO2 NPs and Au NPs dispersion (Injection exp.). Au ions correspond to AuCl3, trauma control are animals injected with dH2O and LPS; animals injected with 50 μg/μL. Columns represents mean (±SE). n represents the number of biological repeats. (\*\*) *p* ≤ 0.01, (\*\*\*) *p* ≤ 0.001 in comparison to control (Kruskal–Wallis test followed by Mann–Whitney U-test). When \* is drawn inside the column, it refers to THC values.

#### **4. Discussion**

We report a comparison of the responses of well-studied immune parameters (THC, haemocyte viability, PO-like enzyme activity, and NO concentration) of the terrestrial isopod *P. scaber* to different types of stressors (Au and CeO2 nanoparticles, salt, and LPS) after feeding exposure and after direct injection into the animal's body. While injection is not a realistic exposure route for NPs, it can help to study consequences of direct biological interactions between NPs and isopod haemocytes inside the body (in vivo).

#### *4.1. Feeding on Au NP-Dosed Diet*

In animals fed for 14 days on a Au NP-dosed diet, we detected a significant increase in the THC, no effect on haemocyte viability, a slight increase in PO-like activity, and no change in the NO concentration. Literature explains such a response as immunostimultion by oral administration [74] or an indication of increased immune capacity [75]. Setyawan et al. (2018) also provided evidence that diet composition (crude fucoidan from three tropical brown algae i.e., Sargassum, Padina, and Turbinaria) was able to provoke increased PO, THC, and relative superoxide dismutase (SOD) activity in white shrimp [74]. In addition, in lobster, haemocyte levels, as well as the haemolymph protein levels, were found to be influenced by diet more than by temperature [76]. Pascual et al. (2004) reported that the amounts of dietary protein were positively corelated to the TCH response in *L. vannamei* juveniles [77]. In their study, shrimp fed an optimal level of protein had an elevated concentration of haemocytes, indicating that optimal dietary protein levels promoted blood cell synthesis [77]. In contrast, shrimp fed suboptimal protein had more proPO per cell, showing that shrimps could be enhancing the content of the haemocytes (i.e., proPO) as a response of cell deficit induced by imbalanced dietary protein [77]. There are also reports in different insect species that the THC increased when they were withheld food [78,79]. Furthermore, Matozzo et al. (2011) observed a positive relationship between the THC and PO activity in starved crabs. They hypothesised that the number of circulating haemocytes increased in starved crabs to allow them to mobilise energy reserves during starvation [80]. In addition, Sequeira et al. (1996) suggested that increases in the THC in Crustacea may be either due to a more active mobilisation of haemocytes from tissues into the haemolymph or a faster division of circulating haemocytes [81]. In line with our findings are reports of Muralisankar et al. (2014), who found an elevated THC and haemocyte production after ZnNP exposure to the freshwater prawn, *Macrobrachium rosenbergii*, within a certain exposure dose range [82]. We can conclude that increases of the THC or in PO-like activity are the result of a complex response of the organism responding to ingested material that is different from its usual diet but one that is not necessarily an indication of adverse effect. What this complex response is and why it has physiological or ecological consequences needs to be discovered.

There are only a few studies on the mechanism of how components in the diet challenge immune processes. In crustaceans, for example, the hepatopancreas, which comes into direct contact with ingested material (not protected by cuticle or peritrophic membrane), produces many proteins that contribute to humoral immunity, including the two most abundant proteins in the haemolymph, clotting protein and haemocyanin [83,84]. Both of these proteins play roles in the immune response [83,84], especially haemocyanin, which is believed to carry out phenoloxidase-like functions in *P. scaber* [34].

We suggest that ingested Au NPs could come into contact with hepatopancreatic cells, and the NPs' environmental corona could be recognised by hepatopancreatic cells [85]. Since a feeding rate reduction was not detected in this study, food quality, and not food deprivation, must be the explanation for the increased THC or PO activity. As ingested Au NPs did not cause an increase in the NO concentration, we conclude that there is a lack of evidence of phagocytic activity of damaged material (damaged-self) [86]. In addition, with no changes in haemocyte viability, we speculate that the changes in immune parameters via hepatopancreas–humoral immunity communication/signalling are primarily not damagerelated (damaged-self-related).

#### *4.2. Feeding on CeO2 NP-Dosed Diet*

In animals fed for 14 days on the CeO2 NP-dosed diet, increased PO-like activity and reduced viability of haemocytes was accompanied by a significant increase of NO activity, suggesting phagocytic activity of damaged or non-self material. However, NO is also produced in the PO cascade, which results in the formation of cytotoxic pigment precursors (quinones, quinone methides, and semiquinones) as well as reactive oxygen intermediates (ROI) and reactive nitrogen intermediates (RNI) [40].

In CeO2 NP-fed animals, we explain the increased THC, reduced haemocyte viability, increased PO-like activity, and increased NO concentration as an immune response, indicating either increased melanisation via increased PO-like activity and/or increased production of toxic radicals (via NO), which could occur either during haemocyte phagocytic activity or melanisation. This response to CeO2 NPs could be provoked due to cuticle (external surfaces) damage and subsequent melanisation of the cuticle either on the body surface or in the digestive system [13]. It is assumed that hepatopancreatic cells, as a possible contact site between cells and NPs, have probably not been damaged. If damage had occurred here, it would likely have been reflected in a reduced feeding rate, which was not seen in this study.

We speculate that the organism responded to microdamages or changes in the cuticle structure, which surrounds the gut epithelium, when exposed to CeO2 NP-dosed food. As reported by Parle et al. (2017), for insects, an injury that penetrates the epidermis results in a scab which is never resorbed [87].

In the case of the CeO2 NP-fed animals in our study, the increased THC and PO-like activity was accompanied by a significant reduction in the viability of haemocytes and a significant elevation in NO levels, suggesting phagocytosis [88], which may indicate a (micro)damage-related response. The response pattern of *P. scaber* to CeO2 NPs is significantly different from that in Au NP-fed animals, indicating a different type of interaction between the two particles and *P. scaber*. The Au NPs appeared more biologically inert than CeO2 NPs.

Even if CeO2 NPs demonstrated some biological reactivity, according to Giese et al. (2018), environmentally relevant concentrations of CeO2 NPs are those 100-fold lower that the one recorded to provoke toxicity—concentrations ranging from 100 to 3000 mg/kg or mg/L [89].

#### *4.3. Feeding on Au Ions-Dosed Diet*

As expected, the Au ions provoked a different type of response, as measured via the selected immune parameters, than the Au NP consumption. We detected a significant decline in the THC but no change in feeding, mortality, haemocyte viability, PO-like activity, or NO concentration. It is interesting that feeding on Au NPs provoked a completely opposite response compared to Au ions. Au NP particles in the diet caused significant increases in the THC and of PO-like activity, while Au ions in the diet resulted in a THC decline.

Similarly, a decrease in the number of circulating haemocytes (TCH) was demonstrated in mercury-exposed prawns [90] and in crabs after bacterial infection [91] or after temperature stress [92]. A decrease in the number of circulating haemocytes is explained as a consequence of haemocyte immobilisation in different tissues, as demonstrated in mercury-exposed prawns [90] and in crabs after bacterial infection [92,93]. Consequently, this leads to a stress-induced decrease in immunocompetence [94] and renders organisms more susceptible to disease. Le Moullac et al. (2000) reported that environmental stress from pollutants induced immunosuppression in crustaceans, in terms of haemocyte number, phagocytosis, and a change in PO activity [21]. Further, Quin et al. (2012) reported that the total haemocyte count in Cd-exposed groups was decreased significantly, when compared with the control groups of freshwater crab, *Sinopotamon henanense* [22]. In addition, Singaram et al. (2013) reported that 14 days of exposure to Hg at environmentally

relevant concentrations suppressed THC, superoxide generation, phagocytosis, and PO activity, among other immune parameters [95].

Matozzo et al. (2011) hypothesised that increased PO activity in haemolymph from temperature-stressed crabs was a physiological response of animals to compensate for the lower THC, in order to increase immunosurveillance in both haemolymph and peripheral tissues [92], corroborating findings by Hauton et al. (1997) [96]. In our study, the PO-like activity was not changed to potentially compensate for the decreased THC. However, we confirm negative consequences of Au ion exposure on the THC, which is in line with findings after mercury exposure.

#### *4.4. Injection of Stimuli into Dlood*

The second approach in our study to challenge the immune parameters of isopods was by the injection of different stimuli (LPS or NPs) directly into the haemocoel and to leave the animals for 48 h to develop a response. A lipopolysaccharide (LPS) is an endotoxin, which is an integral component of the outer membrane of Gram-negative bacteria and, thus, can be used to mimic a bacterial infection [97].

In our study, injected LPS resulted in an increased THC, decreased haemocyte viability, the generation of NO, and reduced PO-like activity. Other authors have also reported alterations in the THC after LPS injection. In insects, adult male *Acheta domesticus* crickets, an LPS injection caused a significant decrease in the number of circulating haemocytes 2 h post injection, which was followed by an increase in haemocyte numbers 1 day after injection and then maintained levels for at least 7 days [73]. Similarly, Bogolio et al. (2000) reported that, after an initial decrease in haemocyte counts in juvenile prawns, a sublethal infection with *V. alginolyticus* induced higher haemocyte counts than controls at 8 days postinfection [98]. In an experiment by Lorenzon et al. (1999), multiple shellfish were injected with an LD50 dose of LPS *E. coli* [99]. In most of the species, the LPS caused an initial decrease in haemocytes (40% at 3–5 hpi), then an increase in haemocytes to control animal levels after 48 hpi (*Palaemon elegans*, *Crangon crangon*, and *Squilla mantis*). A challenge with LPS injection is also reported by Chiaramonte et al. (2019) on the sea urchin *Paracentrotus lividus* [100]. Here, LPS treatment significantly increased the number of coelomocytes after the first hour of treatment. At 6 and 24 h post-LPS treatment, values obtained did not significantly differ from the untreated controls. LPS injection in crayfish [48] caused an increase in haemocytes 2 h post-LPS injection, as well as a significant decrease in haemocyte viability. Furthermore, Xu et al. (2015), who exposed crab haemocytes to LPS, reported that LPS causes a decrease in haemocyte viability, due to the increase apoptosis rate [101]. They found that LPS damaged DNA and caused morphological changes in the haemocytes, resulting in cell shrinkage, nucleus membrane and chromatin fracturing, and the formation of apoptotic bodies [101].

Similar to our work is a study conducted with juvenile *Litopenaeus stylirostris*, which evaluated the impact of the injection of (1) formalin-killed *Vibrio penueicida* cells (vaccine) and (2) a sublethal infection with a moderately pathogenic strain of *Vibrio alginolyticus* on total haemocyte counts. After an initial decrease in haemocyte counts, both the vaccination and the sublethal infection induced higher haemocyte counts than controls. In order to investigate the possibility of increasing the number of circulating haemocytes in pondreared juvenile shrimp, the vaccine was added to the diet. No effect on the haemocyte counts was observed over a 20-day feeding period [98].

Contrary to our expectations, only LPS injection caused changes in the TCH. However, the level of NO was increased in all three injection treatments: LPS, CeO2 NPs, and Au NPs. Literature reports that LPS stimulates non-self-induced nitric oxide generation involving nitric oxide synthase in haemocytes and the production of nitric oxide during phagocytosis as a cellular immune reaction to fight the pathogen [6]. Yeh et al. (2006) also postulate that the observed increase in nitric oxide synthase (NOS) and NO generation increases bacterial adhesion to haemocytes, and as such, LPS increases both haemocyte adhesion and the phagocytic activity of the haemocytes [102]. Gopalakrishnan et al. (2011) found that, in the crab *Scylla paramamosain,* LPS stimulation caused the levels of NO to increase from 3 h to 96 h post-injection [6]. In crayfish injected with LPS, after 1 h, a two-fold increase in NO levels was found, compared to animals injected with only saline [102]. Similarly, NP injection may also provoke phagocytic activity of the haemocytes, but only the CeO2 NPs affected their viability.

We also evidenced alterations of PO-like activity in all cases of injected stimuli. LPS injection caused a reduction of PO-like activity, while both injected NPs increased the PO-like activity. Our results are in line with those reported by Charles & Killian (2015), who found that, in crickets, LPS caused an increased THC and decreased PO activity at 7 days [73]. However, these results were with LPS from *S. marcensces* and not *E. coli* [73]. Contrary to our study, Gopalakrishnan et al. (2011) found that, in the crab *Scylla paramamosain,* LPS injection stimulation caused an increase in PO activity from 3 h to 48 h, while at the same time causing a decrease in the THC [6]. Salawu et al. (2016) suggested that the role of activated PO in the crab *Uca tangeri* was to limit the survival of Gram-negative bacteria upon entry into the haemolymph, and the PO is immediately necessary upon the LPS being identified, with there being an increase in activated PO in the haemolymph 10 min after exposure to LPS [103]. In our study, LPS caused increases in the THC but no increase in PO-like activity; in fact, PO-like activity was decreased again, showing a complex response to a stimulus, as evidenced also in the ingestion exposure.

To sum up, our results show that haemocyte viability and NO response is more severe in case of LPS and least in case of Au NPs (LPS > CeO2 NPs > Au NPs). Excessive NO production is explained by increased phagocytosis activity of cells [88]. Injection of CeO2 NPs and Au NPs resulted in the significant increase in PO-like activity and NO concentration. CeO2 NPs also caused a significant decrease in haemocyte viability, a response also seen after LPS injection, suggesting CeO2 NPs were causing more damage than the Au NPs to the haemocytes, with the former causing a higher instance of haemocyte death. The same, CeO2 NPs being more biologically potent than Au NPs, was observed in feeding exposure.

#### **5. Conclusions**

There are significant differences in the immune response after feeding or injection exposure to different stressors. In both types of exposure, feeding or injection of both NPs, increased the PO-like activity. However, injected LPS caused a reduction of PO-like activity. Increased PO-like activity is interpreted to lead to increased melanisation, which is needed for wound healing [29] but may have other functions, as well. Furthermore, NO was increased in all cases of injection, but only in the case of CeO2 NPs in the feeding experiment and was most significantly increased after LPS injection. NO is associated with the phagocytosis of non-self and damaged-self [86,104]. Therefore, we conclude that CeO2 NPs, and not Au NPs, were causing (micro)damage/alterations to the isopods' tissues and cells, resulting in the increase in phagocytosis, as evidenced by NO levels increasing, to remove the cell debris.

Next, the THC was observed to increase after feeding of both NPs and LPS injection, but feeding on a diet with Au ions caused the THC to decrease. The number of circulating haemocytes can change dramatically over the course of an infection. In other crustaceans exposed to LPS, haemocyte numbers have been found to increase [81,105] in the hours and days after exposure. However, many studies have also reported significant decreases in the haemocyte count and PO activity after exposure of different crustacean species to environmental pollutants [106]. A low number of circulating haemocytes in crustaceans is strongly correlated with a greater sensitivity to pathogens [107] and, hence, a low THC indicates a higher susceptibility to infectious disease [21]. On the other hand, recruitment of haemocytes, as well as increased phagocytic, phenoloxidase, and antioxidant enzymatic activities, among others, are described as a result of immunostimulantion [63,108]. It remains a challenge for the future research what are the physiological and ecological significances for the organism to sense foreign material in the food.

**Author Contributions:** Conceptualisation, C.M., D.D., and A.J.K.; methodology, C.M., A.J.K., A.D., and S.N.; investigation, resources, J.R., F.B., and V.P.; writing, review, and editing, C.M., D.D., A.J.K., A.D., and S.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement PANDORA No 671881 (Probing safety of nano-objects by defining immune responses of environmental organisms).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the study design; in the collection, analyses, and interpretation of the data; in the writing of the manuscript; or in the decision to publish results.

#### **Appendix A**

Figure A1: Physiochemical characterisation of PVP-coated Au NPs and scanning electron microscopy (SEM) of dried leaves covered with PVP–Au NPs. Table A1: Details the number of animals used in the feeding experiment. Table A2: Details the number of animals used in the injection experiment.

**Table A1.** Details the number of animals used in the feeding experiment, the mortality rate, and the stages of the investigation. \*, pooled from 4–5 animals; \*\*, pooled from 1–3 animals.


**Figure A1.** Polyvinylpyrrolidone coated Au NP physicochemical characterisation. (**A**) Bright field scanning transmission electron microscopy (STEM); dynamic light scattering measurement of the Au NPs (Top left); UV–Vis spectra (Top right). (**B**,**C**) Scanning electron microscopy (SEM) of dried leaves covered with PVP Au NPs.


**Table A2.** Details regarding the number of animals used in the injection experiment and the stages of the investigation. \*, pooled from 4–5 animals; \*\*, pooled from 1–3 animals.

#### **References**


## *Article* **Functional and Morphological Changes Induced in** *Mytilus* **Hemocytes by Selected Nanoparticles**

**Manon Auguste 1,\*, Craig Mayall 2, Francesco Barbero 3, Matej Hoˇcevar 4, Stefano Alberti 5, Giacomo Grassi 6, Victor F. Puntes 3, Damjana Drobne <sup>2</sup> and Laura Canesi <sup>1</sup>**


**Abstract:** Nanoparticles (NPs) show various properties depending on their composition, size, and surface coating, which shape their interactions with biological systems. In particular, NPs have been shown to interact with immune cells, that represent a sensitive surveillance system of external and internal stimuli. In this light, in vitro models represent useful tools for investigating nano-biointeractions in immune cells of different organisms, including invertebrates. In this work, the effects of selected types of NPs with different core composition, size and functionalization (custom-made PVP-AuNP and commercial nanopolystyrenes PS-NH2 and PS-COOH) were investigated in the hemocytes of the marine bivalve *Mytilus galloprovincialis*. The role of exposure medium was evaluated using either artificial seawater (ASW) or hemolymph serum (HS). Hemocyte morphology was investigated by scanning electron microscopy (SEM) and different functional parameters (lysosomal membrane stability, phagocytosis, and lysozyme release) were evaluated. The results show distinct morphological and functional changes induced in mussel hemocytes depending on the NP type and exposure medium. Mussel hemocytes may represent a powerful alternative in vitro model for a rapid pre-screening strategy for NPs, whose utilization will contribute to the understanding of the possible impact of environmental exposure to NPs in marine invertebrates.

**Keywords:** hemocytes; *Mytilus*; in vitro; scanning electron microscopy; immune response

### **1. Introduction**

Due to the rapid expansion of production and use of nanoparticles (NPs) and their consequent release in different ecosystems, there is increasing concern on the utilization of alternative, affordable biological animal models for investigating nanosafety in environmental species. In vitro testing that focuses on specific cellular functions (i.e., immune responses) may provide an ideal starting point for developing a rapid prescreening strategy for NPs [1]. This would apply not only to vertebrate models, but also to invertebrates, which account for over 95% of animal species.

In the aquatic environment, which represents the final sink for all anthropogenic contaminants, including NPs, suspension-feeding invertebrates, due to their filtration ability for nutrition and respiration needs, have been identified as an unique target group for NP ecotoxicity [2–5]. In particular, in bivalve molluscs, abundant cell types involved

**Citation:** Auguste, M.; Mayall, C.; Barbero, F.; Hoˇcevar, M.; Alberti, S.; Grassi, G.; Puntes, V.F.; Drobne, D.; Canesi, L. Functional and Morphological Changes Induced in *Mytilus* Hemocytes by Selected Nanoparticles. *Nanomaterials* **2021**, *11*, 470. https://doi.org/10.3390/ nano11020470

Academic Editor: Diana Boraschi Received: 18 December 2020 Accepted: 9 February 2021 Published: 12 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

in digestion or immune responses (digestive cells and hemocytes, respectively), have the capacity to internalize particles from the nano- to the micro-size via endocytic/phagocytic pathways [4–6].

The bivalve immune system relies solely on innate immunity, which mainly involves circulating hemocytes acting in collaboration with other soluble factors present in hemolymph serum [7,8]. This system responds very fast upon its encounter with foreign particles, making it a suitable tool for studying the effects of NPs using in vitro methods [9–13]. In particular, in the hemocytes of the marine mussel *Mytilus galloprovincialis*, the utilization of a battery of immune-related biomarkers has proven useful in the evaluation of the effects of a number of NPs, that can modulate responses with consequent immunotoxic effects or stimulation of immune parameters, leading to inflammation, depending on the NP type and on the conditions of exposure [12,14]. These studies underlined how mussel hemocytes represent a sensitive target for different NPs and their potential application as a starting point to more accurately design further studies that are relevant at the whole animal level.

The utilization of primary cells or cell lines is already established for vertebrates but these existing protocols need to be adapted and/or improved for bivalve in vitro assays [1,15,16]. Hemocytes freshly isolated from adult mussels (*M. galloprovincialis*) can be maintained in different media (artificial seawater, salt-enriched culture medium or hemolymph serum) for some hours, without impacting cell survival and biochemical/functional features [9,12,17,18]. All these media, characterized by a high ionic strength, will however affect NP behavior in terms of agglomeration and surface charge [19] as in the natural marine environment. Moreover, the behavior of NPs in mussel biological fluids (i.e., hemolymph serum) has been shown to be distinct for different types of NPs and important in determining interactions with hemocytes. In mussels, some types of NPs have been shown to associate with serum soluble components, organized into a "hard protein corona", providing a specific biological identity for immune recognition and subsequent cellular responses [20,21]. All these factors are important to consider in the evaluation of the possible biological impact of NPs on the cells of marine invertebrates, to properly utilize these cellular in vitro models as a suitable alternative strategy for testing the immunosafety of NPs in environmental organisms.

In the present study, we investigated the in vitro responses of mussel hemocytes to different types of NPs in different exposure media with the aim of:


Selected NPs were chosen on the basis of two criteria: first, the need to utilize a reference NP that is generally considered biologically inert as in mammalian model systems (i.e., AuNPs) [22,23]. To this aim, custom-made polyvinylpyrrolidone-PVP coated gold NPs (PVP-AuNPs), were designed to be stable in seawater medium and provide baseline information for further NPs testing. Second, the utilization of commercially available surface-modified NPs that could be utilized as model NPs of environmental interests (e.g., nanoplastics). Commercial surface modified polystyrene nanoplastics (amino- and carboxy-modified nanopolystyrene, PS-NH2 and PS-COOH) were utilized to provide an estimate of the role of different surface characteristics in hemocyte nano-bio-interactions.

The short-term in vitro effects of these selected NPs at different concentrations were compared in freshly isolated hemocytes from *M. galloprovincialis* exposed in either artificial seawater (ASW) or filtered hemolymph serum (HS). Upon exposure, the morphological changes induced in different experimental conditions were investigated by scanning electron microscopy (SEM). Complementary experiments were performed to measure different functional immune markers, from functional integrity of lysosomes (lysosomal membrane

stability) and phagocytic ability to extracellular defense mechanisms (e.g., lysozyme release and the production of ROS-reactive oxygen species).

#### **2. Materials and Methods**

#### *2.1. Nanoparticle Synthesis and Characterization*

Custom-made PVP-AuNPs were synthesized by the Catalan Institute of Nanoscience and Nanotechnology (ICN2) (Barcelona, Spain), according to the previously reported seeded-growth methods. Detailed synthetic procedure can be found in Bastús et al. (2011) [24]. Tetrachloroauric (III) acid trihydrate (99.9% purity), sodium citrate tribasic dihydrate (≥99%), and polyvinylpyrrolidone (55 KDa) were purchased from Sigma-Aldrich (Madrid, Spain). Artificial seawater (ASW) was prepared at 35 ppt salinity, pH 7.9–8.1 [25,26]. In brief, AuNPs were synthesized using a sodium citrate aqueous solution, and different synthesis, postproduction, and characterization steps are reported in more detail in Figure S1. The AuNP seeds (~10 nm) were grown to obtain the desired size (~30 nm) and measured by scanning electron microscopy in transmission mode (STEM) (FEI Magellan 400 XHR, Hillsboro, OR, USA) (Figure S1A,B). AuNPs were further coated with PVP, followed by several washing steps. UV-vis analyses using a Carry 4000 spectrophotometer (Agilent, Santa Clara, CA, USA) were performed to verify the AuNP aggregation state and sample concentration (Figure S1C). The final hydrodynamic diameter (Figure S1D) and ζ-potential were determined by dynamic light scattering (DLS) and laser Doppler velocimetry using a Zetasizer Nano ZS instrument (Malvern Panalytical, Malvern, UK).

Commercial surface-modified nanopolystyrenes were from Bangs Laboratories, Inc. (Fishers, IN, USA). Amino-modified nanopolystyrene (PS-NH2), nominal size 50 nm, were previously characterized in different media (milliQ-MQ water, ASW, and HS) as described in Canesi et al. (2015, 2016) [11,12]. Carboxylated nanopolystyrene (PS-COOH), nominal size 60 nm, kindly provided by Ilaria Corsi's lab (Univ. Siena, It), were from the same batch previously characterized in MQ water by DLS analysis [27]. In addition, in the present study, particle behavior was evaluated in suspensions of both ASW and *Mytilus* hemolymph serum (HS). The results on particle characterization are summarized in Table 1.

**Table 1.** Physicochemical characterization of NPs tested in vitro, PVP-AuNP, nanopolystyrenes (PS-NH2 and PS-COOH) behavior in exposure medium.


PDI = polydispersity index; ζ = zeta potential; MQ = MilliQ water; ASW= artificial seawater; HS = *Mytilus* hemolymph serum; <sup>1</sup> from [11]; <sup>2</sup> from [12]; <sup>3</sup> from [27].

Possible contamination of NPs by lipopolysaccharides (LPS) was not of first concern, as mussels have shown to be resistant to LPS effects, due to their filtering ability and vicinity with human activities. In particular, mussel hemocytes in vitro show no sensitivity to LPS up to 10 μg/mL [28]. These concentrations are much higher than those of NP-associated LPS [29,30].

#### *2.2. Mussels, Hemolymph Collection, Preparation of Hemolymph Serum and Hemocyte Monolayers and Exposure Conditions*

Mussels (*Mytilus galloprovincialis* Lam.) were purchased from an aquaculture farm (Chioggia, IT), transferred to the laboratory, and acclimatized in static tanks containing aerated ASW (1 L/animal), 35 ppt, at 16 ± 1 ◦C for 24 h. For each sample, hemolymph was extracted from the adductor muscle of 4–5 animals, using a syringe via a noninvasive method, filtered with gauze, and pooled in Falcon tubes. To obtain hemolymph serum-HS (i.e., hemolymph free of cells), whole hemolymph was centrifuged at 900× *g* for 10 min, and the supernatant was passed through a 0.22 μm filter. All procedures were performed as previously described [11,12].

Hemocytes were incubated at 16 ◦C with different concentrations of NPs in ASW or HS, for 30 min to 1 h (depending on the functional parameter measured) as previously described [9–11,13]. Untreated (control in ASW or HS) hemocyte samples were run in parallel. The in vitro short-term exposure of mussel hemocytes has long been successfully applied to screen the effects of different types of NPs on immune function (see [9–11,13]). The underlying reason is that in these cells, induction of functional responses, as well as of stress responses and apoptotic processes, are particularly rapid, occurring within 1 h of exposure, in line with the physiological role of bivalve hemocytes as the first line of defense against non-self-material (see also [4]).

#### *2.3. Hemocyte Morphology Using Scanning Electron Microscopy (SEM)*

Hemocyte monolayers prepared on membrane filters (hydrophilic mixed cellulose esters membrane, Millipore©) of 3 μm pore size were incubated for 30 min with NP suspensions in ASW or HS. Control cells were run in parallel in ASW or HS medium. Cells were fixed for 2 h in 2.5% glutaraldehyde, 0.4% paraformaldehyde in modified PBS (100 mM phosphate-buffered saline adjusted to 450 mM osmolarity by addition of NaCl). Fixed cells were rinsed with the same buffer, postfixed in 1% osmium tetroxide in buffer for 1 h, and then stained with TOTO (thiocarbohydrazide/osmiumtetroxide/thiocarbohydrazide/ osmiumtetroxide) conductive stain (adapted from [31,32]). Samples were then dehydrated by graded ethanol solutions and dried with hexamethyldisilizane (HMDS).

The dried samples were mounted on aluminum holders and sputter-coated with gold/palladium using precision etching coating system (Gatan 682, Pleasanton, CA, USA). Hemocytes were then examined with a field emission scanning electron microscope (SEM) (SEM, JEOL JSM-6500F, Tokyo, Japan) equipped with an energy dispersive X-ray spectroscopy (EDX) system. Samples exposed to AuNPs were sputter-coated with carbon in order to detect Au by EDX.

#### *2.4. Hemocyte Functional Assays*

Hemocyte functional parameters (lysosomal membrane stability, phagocytosis, lysosomal enzyme release, and extracellular ROS production) were evaluated essentially as previously described [9,11,13].

Lysosomal membrane stability (LMS) was evaluated by the NRR (neutral-red retention time) assay. Hemocyte monolayers (in triplicates) on glass slides were incubated with NP suspensions in filtered ASW or HS for 30 min and then incubated with a neutral-red (NR) solution (final concentration 40 μg/mL). Controls (unexposed samples) were run in parallel. The slides were examined under an optical microscope every 15 min until 50% of the cells showed sign of lysosomal leaking and the results are reported as percentage of control cells.

The percentage of phagocytic cell was evaluated by the uptake of neutral-red-stained zymosan on hemocyte monolayers. After 30 min of NP exposure, the neutral-red-stained zymosan in 0.05 M TrisHCl buffer (TBS) was added to the monolayers and left to incubate for 60 min. Monolayers were then washed three times with ASW, fixed with Baker's formol calcium, and mounted in Kaiser's glycerol gelatine medium for microscopical examination with an optical microscope. For each slide, the percentage of phagocytic hemocytes was calculated from a minimum of 200 cells in triplicates.

Extracellular ROS production was measured by the reduction of cytochrome c. The aliquots of hemolymph (in triplicates) were incubated with 500 μL of cytochrome c solution (75 mM ferricytochrome c in TBS buffer) in presence or not of NPs. The samples were read at 550 nm at different times (0, 30, and 60 min), and the results are expressed as changes in OD per mg of protein.

Lysozyme activity in the extracellular medium was measured in aliquots of serum. Whole hemolymph samples were incubated with or without NPs for 30 to 60 min. Lysozyme activity was determined spectrophotometrically at 450 nm utilizing *Micrococcus lysodeikticus* and was expressed as percentage of controls.

Total protein content was determined according to the Bradford method using bovine serum albumin (BSA) as a standard.

#### *2.5. Statistics*

Data are the mean ± SD of four independent experiments (n = 4), with each assay performed in triplicate. Data of functional parameters and hemocyte diameters were analyzed by one-way ANOVA followed by Tukey's test at 95% confidence intervals (*p* ≤ 0.05). All statistical calculations were performed using the GraphPad Prism version 7.03 for Windows, GraphPad Software, San Diego, CA, USA.

#### **3. Results**

#### *3.1. NP Characterization*

Data on NP characteristics and behavior in exposure media are reported in Table 1.

Custom-made PVP-coated gold nanoparticles (PVP-AuNPs) in ASW were characterized by STEM to measure the size of the gold core and by DLS to obtain the hydrodynamic diameter. AuNPs had a gold core of 28.7 ± 3.2 nm (Figure S1B) and a PVP (55 KDa) coating leading to a final hydrodynamic diameter of 58.3 ± 0.16 nm (Figure S1D). PVP-AuNPs ζ-potential in ASW was −4.1 ± 0.6 mV.

For both nanopolystyrenes, in MQ the Z-average were in the same range as reported by the manufacturer, 57 ± 2 and 64.5 ± 0.6 nm for PS-NH2 and PS-COOH, respectively. The ζ-potential was positive (+42.8 ± 1 mV) for PS-NH2 and negative (−59.7 ± 2.6 mV) for PS-COOH [11,27]. In both ASW and HS, PS-NH2 showed a similar behavior, with formation of small agglomerates (~200 nm) and a lower ζ-potential (+14 mV) [11,12]. A distinct behavior was observed for PS-COOH, that in ASW formed large agglomerates (1822 ± 373.6 nm) and a less negative charge value (−26.9 mV). In HS, the size of the agglomerates was greatly reduced (189.1 ± 48.6 nm), and the ζ-potential showed even fewer negative values (−11.6 ± 1.5 mV).

For all NPs, PDI values were smaller in MQ that in ASW and HS, indicating a more homogenous dispersion in the absence of salts.

#### *3.2. Effects of NPs on Hemocyte Morphology: SEM*

The effects of different NPs on hemocyte morphology were investigated by SEM. For all NP types, the hemocytes were exposed for 30 min at a concentration of 10 μg/mL in ASW or HS suspension. Control hemocytes (in absence of NPs) were run in parallel in both media.

Control hemocytes in ASW showed a characteristic extremely flat shape, with cells fully spread onto the filter support, with diameters ranging from 20 to 40 μm and showing a rather smooth surface, several short cell-surface extensions and absence of filopodia (Figure 1A,B). Due to the extreme flatness of the cells, fractures were often observed between the periphery and the center of the cells, which was thicker and contained granular structures presumably in the perinuclear region.

**Figure 1.** Scanning electron microscopy (SEM) images of control hemocytes from *M. galloprovincialis* in different media. (**A**,**B**) in artificial seawater (ASW); (**C**,**D**) in hemolymph serum (HS). Hemocytes were attached to a paper filter support with pore size of 3 μm.

A different shape was observed in control hemocytes maintained in HS medium. They were not fully spread on the support and therefore thicker with smaller diameters (~20–28 μm). Their surface was irregular with several membrane ruffles and filopodia of variable shapes (Figure 1C,D).

After the addition of PVP-AuNP in ASW, cell morphology was similar to that of controls, even though some cells showed the presence of vesicles of micro-/submicrometric size (≤1 μm diameter) on the edge of the plasma membrane (Figure 2). When the presence of Au was investigated by EDX, no gold was detected either inside the cells or in the surrounding environment (Figure S2). Similar observations were made in HS (not shown).

In contrast, PS-NH2 induced changes in hemocyte shape in both media (Figure 3). In ASW, some cells were still spread on the substrate, but they were not fully attached to the support, showing variable shapes (Figure 3A). Other hemocytes were smaller and thicker, showing the typical morphology of activated cells (Figure 3B). Such changes in shape were more evident in HS, where a large number of filopodia was observed (Figure 3D,E). Moreover, in both conditions, the edge of the cells often showed a complex lace-like pattern, due to a network of short filopodia (Figure 3C,F).

Exposure to PS-COOH in ASW did not affect gross morphology of cells with respect to control hemocytes (Figure 4A). Large particle agglomerates were observed around some cells (arrowheads), appearing very white in contrast with the rest of the biological material (Figure 4B and Figure S3 for details of PS-COOH). Moreover, short membrane protrusions could be observed along the border of some cells, as well small extracellular vesicles (Figure 4C). The presence of small PS-COOH agglomerates on the cell surface was observed near vesicles with apparent concave shapes (Figure 4D).

In hemocytes exposed to PS-COOH in HS, cell morphology was apparently similar to those of control cells in HS; however, the formation of thin filopodia of about 3–8 μm in length was observed, apparently enabling communication between adjacent cells (Figure 5A,B). Among these, some longer (>15 μm) filopodia were observed (Figure 5C, arrow), sometimes apparently making contact with apoptotic cells/cell bodies. When observed in more detail (Figure 5D), these filaments appeared thicker (300–400 nm) and

containing vesicles (asterisk). Moreover, other shorter filopodia appeared to be formed by a chain of vesicles connected by very thin necks, as in a necklace-like structure (arrowheads). In HS, no particle agglomerates were observed.

**Figure 2.** SEM images of hemocytes of *M. galloprovincialis* incubated for 30 min with PVP-AuNPs (10 μg/mL) in ASW suspension. In the framed cell, the presence of PVP-AuNPs was investigated using Energy-dispersive X-ray (EDX) analysis, and the details are presented in Figure S2.

**Figure 3.** SEM images of hemocytes of *M. galloprovincialis* incubated for 30 min with PS-NH2 in ASW suspension (**A**,**B**) and in HS suspension (**D**,**E**). Details of membrane structures observed in ASW (**C**) and HS (**F**).

**Figure 4.** SEM images of *M. galloprovincialis* hemocytes incubated for 30 min with PS-COOH suspensions (10 μg/mL) in ASW (**A**). Arrowheads in (**B**) indicate large agglomerates of PS-COOH; (**C**) short extensions and vesicles along the plasma membrane (dotted frame); (**D**) Enlargement of the white frame in (**C**) indicating the presence of PS-COOH small agglomerates on the cell surface (arrowhead) as well as of concave vesicles (arrow).

**Figure 5.** SEM images of *M. galloprovincialis* hemocytes incubated for 30 min with PS-COOH suspensions (10 μg/mL) in HS (**A**,**B**,**D**) Detail of cytoplasmic extensions observed in (**C**), indicating thicker filaments containing vesicles (asterisk), and filopodia with necklace-like structure (arrowheads).

#### *3.3. Hemocyte Functional Parameters*

#### 3.3.1. PVP-AuNP

PVP-AuNP exposure in ASW did not affect hemocyte LMS and phagocytosis (Figure 6A,B) at any concentration tested up to 100 μg/mL (Figure 6A,B). Moreover, PVP-AuNPs did not affect extracellular ROS production up to 50 μg/mL (Figure 6C). Similarly, no effects were observed using HS as suspension medium (data not shown).

**Figure 6.** Effects of PVP-AuNPs in ASW on hemocyte: lysosomal membrane stability (LMS) (**A**), phagocytosis (**B**) and extracellular ROS production (**C**).

#### 3.3.2. Amino-Modified Nanopolystyrene—PS-NH2

The effects of PS-NH2 on *Mytilus* hemocytes in different exposure media were previously reported in [11,12]. However, to mirror the effects observed for the changes in cell morphology, in this work further experiments were performed at the same concentration, 10 μg/mL in both media, and the results are reported in Figure 7.

**Figure 7.** Effects of PS-NH2 in different suspension media in ASW (white bars) or hemolymph serum (HS) (grey bars) on mussel *M. galloprovincialis* hemocytes. (**A**) LMS; (**B**) phagocytosis; and lysozyme activity at time 0, after 30 and 60 min in ASW (**C**) and HS (**D**). Data, representing the mean ± SD of four experiments in triplicate, were analyzed by ANOVA followed by Tukey's post hoc test, (*p* < 0.05): \* = all treatments vs. controls; # = ASW vs. serum.

A dose-dependent decrease in LMS was observed from 10 μg/mL in ASW, down to −50% at 50 μg/mL; a significantly stronger effect was observed in HS, at both concentrations (Figure 7A). The phagocytic activity showed a similar decrease (about −40% with respect to controls) at 10 and 50 μg/mL and in both media (Figure 7B). PS-NH2 also stimulated increase in lysozyme activity: the effect was transient in both media with highest values recorded directly immediately after NP addition and after 30 min exposure (Figure 7C,D).

#### 3.3.3. Carboxy-Modified Nanopolystyrene—PS-COOH

The effects of PS-COOH are reported in Figure 8. In ASW suspensions, LMS was decreased from −10 to −45% with respect to controls, at concentrations ranging from 10 to 100 μg/mL, whereas no effects were observed in HS (Figure 8A). Phagocytic activity was unaffected in all experimental conditions (Figure 8B). Addition of PS-COOH immediately triggered lysozyme release by hemocytes in both media although only at 50 μg/mL. Interestingly, in ASW this rise in lysozyme activity was persistent up to 60 min exposure, whereas in HS a rapid decrease over time was observed (Figure 8C,D).

**Figure 8.** Effects of PS-COOH in different suspension media in ASW (white bars) or hemolymph serum-HS (grey bars) on mussel *M. galloprovincialis* hemocytes. LMS (**A**), phagocytosis (**B**), and lysozyme activity at time 0, after 30 and 60 min in ASW (**C**) and HS (**D**). Data representing the mean ± SD of four experiments in triplicate, were analyzed by ANOVA followed by Tukey's post hoc test, (*p* < 0.05); \* = all treatments vs. controls; # = ASW vs. serum.

Because PS-COOH in HS showed distinct agglomeration and effects on hemocyte morphology, lysosomal membrane stability and time course of lysozyme release, its possible interactions with soluble serum components (i.e., the formation of a protein corona) were investigated. PS-COOH was incubated with hemocytes HS and isolation of the corona proteins was performed by centrifugation and 1D-gel electrophoresis as previously described for PS-NH2 and other NPs ([12,20] and Figure S4 for details). When samples were analyzed by gel electrophoresis (Figure S5), no detectable protein bands specific of the

corona sample obtained with PS-COOH were observed, indicating the absence of proteins stably bound to the NPs (hard protein corona).

#### **4. Discussion**

In the present work, the in vitro interactions and effects of NPs in different media were investigated in the hemocytes of the marine mussel *M. galloprovincialis*. Custom-made PVP-AuNPs stable in ASW were first used as a reference material, and the results were compared with those obtained with two types of nanopolystyrene bearing distinct surface properties (PS-NH2 and PS-COOH). The results highlight distinct morphological and functional responses of mussel hemocytes depending on the core composition and chemicophysical properties of the NPs used and on exposure medium. To our knowledge, these are the first data on the effects of different NPs on detailed cell morphology evaluated by SEM coupled to evaluation of functional parameters in the immune cells of marine invertebrates.

Gold NPs represent a valuable tool for biomedical applications, i.e., diagnosis or targeted drug delivery, due to the peculiar properties of this material [33,34]. AuNPs appear generally inert toward biological systems, since they have been reported to poorly affect natural cell functions [22]. In particular, in mammalian systems, AuNPs have been shown to be relevant to meet the need for immunosafe particles for clinical purposes [23]. In this light, AuNPs can be utilized as a model to compare the interactions of NPs with innate immunity across different species. However, for a proper comparison, the main characteristics and behavior of NPs should be maintained in different experimental settings. Various NPs have been shown to strongly agglomerate in ASW, due to the high salt concentration which reduces the electrostatic repulsion of the particles [35]. To circumvent this problem, custom-made AuNPs, coated with PVP to maintain their dispersion in ASW, were successfully synthesized to allow for exposure of marine invertebrate cells in vitro.

Previous in vitro studies on *M. galloprovincialis* hemocytes exposed to citrate-coated AuNPs of different sizes (5, 15, 40 nm) for 24 h reported a slight decrease in cell viability at concentrations >50 μg/mL, with stronger effects caused by smaller-size AuNPs; however, the citrate coating used appeared to play a major role in the toxicity observed [16]. In gill explants of the clam *Ruditapes philippinarum*, the utilization of an optimized STEM-in-SEM methodology allowed for detection of AuNPs (24 nm) inside the gill cells, attached to the outer membrane of the mitochondria and to the nuclear envelope [36]. In the present work, no internalization of AuNPs was observed by *Mytilus* hemocytes; however, differences in phagocytic activity may be due to the cell type, the incubation time, or the AuNP characteristics. Moreover, the results here obtained show that short-term exposure to PVP-AuNPs did not affect the morphology or main functional parameters of mussel hemocytes. When the presence of Au was investigated by EDX, no gold was detected either inside the cells or in the surrounding environment. However, in some AuNP-exposed cells, small vesicles on the edge of the plasma membrane were observed, resembling ectosomes. Ectosomes are vesicles generated by the direct outward budding of the plasma membrane, which produces microvesicles, microparticles, and large vesicles in the size range of ~50 nm to 1 mm in diameter [37]. Human polymorphonuclear neutrophils (PMNs) upon activation release ectosomes which are characterized by the expression of phosphatidylserine and show anti-inflammatory/immunosuppressive activities toward macrophages [38]. In mussel hemocytes, the production of ectosomes and exchange of signaling molecules might represent an additional component of intercellular communication among different cell subpopulations, contributing to down-regulation of responses to AuNP exposure and immunoregulation as in human PMNs. However, this possibility requires further investigation.

Overall, these data indicate that AuNPs did not strongly interact with hemocytes and/or were lost during the fixing procedure. The data obtained in the present work confirm the suitability of the newly synthetized AuNP to be utilized as a negative control for testing the effects of NP exposure on *Mytilus* hemocytes, as well as any other marine invertebrate cell types. These data reinforce the concept of the importance of the choice of NPs and suitability of the protocols for handling marine invertebrate cells for in vitro testing of NP safety. Many parameters can influence the biological activity of NPs, including core composition, surface charge, and agglomeration state. To evaluate the effects of NP functionalization on *Mytilus* hemocytes, polystyrene NPs (PSNPs) of similar size (50–60 nm) but carrying different surface modifications (–NH2 and –COOH) were compared in vitro. PS-NH2 showed little agglomeration (~200 nm) in ASW, and slightly smaller (~178 nm) in HS medium, while retaining a similar positive surface charge (+14.2 mV) in both media [12]. PS-NH2 induced significant changes in functional parameters in both ASW and HS, with stronger effects observed in HS as previously described [11,12]. These results are supported by the present data here obtained by SEM on cell morphology in HS with respect to ASW.

The difference observed between the two media is in line with the reported formation of a stable biomolecular corona recorded for PS-NH2 in HS [12], whose unique protein component was identified as the extrapallial protein precursor (EPp) (also called MgC1q6). EPp is the most abundant serum protein encountered in *Mytilus* HS, and it is known to play a key role in the specific recognition of both selected bacterial strains and NP types [39,40]. The presence of a protein corona on PS-NH2 would likely increase the stability of NPs and contribute to reducing agglomeration, as well as to regulate nano-bio-interactions with hemocytes and the consequent outcome of the cellular response.

Distinct observations were made with PS-COOH, in terms of both particle behavior in exposure media and morphological and functional responses of hemocytes. In ASW, PS-COOH showed the formation of large agglomerates (>1500 nm) and retained a negative surface charge (−26 mV). Such a high agglomeration in ASW is likely to occur due to the interactions between the negative surface charge (-COO−) retained at pH ≥ 5 and high concentrations of divalent cations (e.g., Ca2+ and Mg2+) naturally present in seawater [41,42]. The results of SEM analysis indicate the absence of gross morphological changes in hemocytes exposed to PS-COOH at 10 μg/mL in ASW; accordingly, in these conditions, only a slight decrease in LMS was observed, whereas lysozyme release and phagocytosis were unaffected. This confirms that large NP agglomerates show little interactions with hemocytes.

When suspended in HS, PS-COOH formed smaller agglomerates (~180 nm) that were not detectable around the cells by SEM; in addition, under these experimental conditions, gross cell morphology and functional parameters were not significantly affected. However, SEM allowed for the detection of some subtle morphological changes, like formation of multiple filopodial extensions of variable length and shape that were not observed in all the other experimental conditions tested. The roles of filopodia can be various, from sensing or communication with the surrounding environment [43], to mobility, particle engulfment, and the production and release of molecules [44]. In mussel hemocytes exposed to PS-COOH in HS, we observed thicker filopodia containing vesicles, and/or shorter and thinner filopodia apparently formed by a chain of spherical vesicles connected by ultrathin necks. Some of them seemed to connect with the neighboring cell or with apoptotic bodies/vesicles. Drab et al. (2019) recently reported several mechanisms that could explain the presence and formation of these necklace-like structures, which can represent an intermediary stage in the formation of structures known as tunneling nanotubes (TNTs). TNTs are a new emerging cell-to-cell communication tool, that allows for the selective transfer of cellular components, signaling molecules, and pathogens between cells [45]. The presence and roles of TNTs have been mainly described in mammalian cells, including immune cells, where they play several roles in responding to stress, including the transport of nanomaterials [46,47]. Interestingly, we have recently reported the presence of similar structures exchanging lysosomal vesicles and mitochondria in live *M. galloprovincialis* hemocytes [48]. The results here obtained suggest that, under certain experimental conditions, NPs may induce TNT formation in invertebrate immune cells, thus probably allowing for an intercellular communication system that may participate in the protection against NP toxicity. Further work will be needed to confirm and define the presence of these peculiar

structures in *Mytilus* hemocytes and appreciate their role in the effects that different NPs have on their immune function.

The results obtained with PS-COOH are the first data on the effects of negatively charged nanoplastics in the cells of marine bivalves. Data obtained in the presence of HS suggest that the observed effects may be due to the formation of a NP-protein corona, in analogy with previous data obtained with other NPs, including PS-NH2 and nanooxides [20]. However, we could not identify any protein stably bound to PS-COOH (hard corona). These data do not exclude the possible interactions with other hemolymph proteins loosely bound to PS-COOH (soft corona) that may participate in modulating the responses of hemocytes and that cannot be detected with the method utilized. Grassi et al. (2019) studied the corona formation and composition in sea urchin coelomic fluid for PS-NH2 and PS-COOH and reported a similar hard-corona proteomic pattern for both PS-NP types [27]. The discrepancies with this work not only may be due to the profound differences in protein composition between biological fluids of sea urchins and mussels but also to the methodological approach using different protein separation (2D-electrophoresis). Finally, in mussel hemocytes, in contrast with data obtained with PS-NH2, where the presence of a stable protein corona elicited stronger effects/damages, the presence of proteins more loosely interacting with PS-COOH (soft corona) would participate in protective mechanisms, contributing to the biological response, as indicated by the results obtained measuring functional parameters.

Overall, the results further underline the importance of the surface charge, rather than the core material, in determining NP behaviour in exposure medium and consequent specificity of the nano-bio-interactions with mussel hemocytes leading to biological responses in vitro.

The application of SEM revealed for the first time the detailed morphology of hemocytes from *M. galloprovincialis* kept in different media and allowed for the evaluation of the morphological changes induced by different NPs in different experimental conditions. The results obtained on cell morphology are in line with those of determination of functional parameters, because the experimental conditions that elicited more evident morphological changes were always accompanied by the activation of stronger functional responses. The morphological and functional approach thus provides a better understanding of the results obtained when testing different types of NPs in vitro. This basal information will contribute to developing more standardized protocols for the in vitro screening of nanosafety in marine invertebrate cell models. This knowledge will also help better define protocols for in vivo exposure to NPs at environmentally realistic exposure concentrations.

In conclusion, the experimental system used in our study represents a suitable approach for fast evaluation of the biological activity of engineered nanomaterials in mussel immune cells. The most important characteristic of a prescreening tests is its capacity to discriminate among particles with different characteristics and what responses are biologically relevant. Our results confirm the specificity of the responses of mussel hemocytes (morphology and functional parameters) to different NPs.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2079-499 1/11/2/470/s1, Figure S1: Characterization of synthesis and postproduction steps of synthesized PVP-AuNPs; Figure S2: Energy-dispersive X-ray (EDX) analysis of a SEM sample of a hemocyte exposed to PVP-AuNPs in ASW; Figure S3: SEM images showing in more details PS-COOH suspensions in ASW; Figure S4: Schematic overview of the protocol utilized to identify the mussel PS-COOH corona (C) from *Mytilus galloprovincialis* hemolymph serum (HS); Figure S5: Separation of PS-COOH protein complexes from HS of *M. galloprovincialis* proteins by SDS-PAGE and staining with Coomassie Brilliant Blue.

**Author Contributions:** Conceptualization, M.A. and L.C.; methodology, C.M. and D.D.; formal analysis, M.H. and S.A.; investigation, resources, F.B., G.G., and V.P.; writing—original draft preparation, M.A.; writing—review and editing, M.A. and L.C.; supervision, D.D. and L.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement PANDORA N◦ 671881 (Probing safety of nano-objects by defining immune responses of environmental organisms). Partial support was given by the Italian Antarctic Project NANOPANTA Nano-Polymers in the Antarctic marine environment and biota PNRA16\_00075 B.

**Institutional Review Board Statement:** The Mediterranean mussel, *M. galloprovincialis*, is not considered an endangered or protected species in any international species catalog, including the CITES list (www.cites.org), and not included in the list of species regulated by EC Directive 2010/63/EU. Therefore, no specific authorization is required to work on mussel samples.

**Informed Consent Statement:** Not applicable.

**Acknowledgments:** We would like to thank I. Corsi (Univ. Siena, IT) for providing the PS-COOH used in the present work.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


*Article*

## **Safety Evaluation of TiO2 Nanoparticle-Based Sunscreen UV Filters on the Development and the Immunological State of the Sea Urchin** *Paracentrotus lividus*

### **Riccardo Catalano 1, Jérôme Labille 1,**†**, Daniela Gaglio 2,3, Andi Alijagic 4, Elisabetta Napodano 3, Danielle Slomberg 1, Andrea Campos <sup>5</sup> and Annalisa Pinsino 4,\*,**†


Received: 9 September 2020; Accepted: 20 October 2020; Published: 23 October 2020

**Abstract:** Sunscreens are emulsions of water and oil that contain filters capable of protecting against the detrimental effects of ultraviolet radiation (UV). The widespread use of cosmetic products based on nanoparticulate UV filters has increased concerns regarding their safety and compatibility with both the environment and human health. In the present work, we evaluated the effects of titanium dioxide nanoparticle (TiO2 NP)-based UV filters with three different surface coatings on the development and immunity of the sea urchin, *Paracentrotus lividus*. A wide range of NP concentrations was analyzed, corresponding to different levels of dilution starting from the original cosmetic dispersion. Variations in surface coating, concentration, particle shape, and pre-dispersant medium (i.e., water or oil) influenced the embryonic development without producing a relevant developmental impairment. The most common embryonic abnormalities were related to the skeletal growth and the presence of a few cells, which were presumably involved in the particle uptake. Adult *P. lividus* immune cells exposed to silica-coated TiO2 NP-based filters showed a broad metabolic plasticity based on the biosynthesis of metabolites that mediate inflammation, phagocytosis, and antioxidant response. The results presented here highlight the biosafety of the TiO2 NP-based UV filters toward sea urchin, and the importance of developing safer-by-design sunscreens.

**Keywords:** cosmetic formulation; cosmetic lifecycle; hydrophobic compound; nano-TiO2; nano safety; marine invertebrate; metabolomics

#### **1. Introduction**

Ultraviolet (UV) radiation is a main risk factor for skin disorders such as erythema, photoaging, and keratinocyte cancer. As a consequence, effective photoprotection is of utmost importance to humans. Several approaches for skin protection have recently been developed, including organic (e.g., benzophenone, octocrylene) and inorganic UV filters (e.g., zinc oxide, titanium dioxide), topically applied antioxidants, DNA repair enzymes, and oral photoprotective strategies based on nutritional supplements [1]. Sunscreens are emulsions of water and oil that contain UV filters capable of protecting human skin from the detrimental effects of UV radiation [2].

The widespread use of these products has increased concerns regarding their safety and compatibility with both the environment and human health [3]. Thus, sunscreen product-related environmental health risk assessment and management are important issues that need to be carefully considered. Some biological models have shown harmful effects in the presence of organic filters (e.g., benzophenones, camphor) used in sunscreen formulations [4–6]. Moreover, most organic UV filters are able to pass through the skin barrier after one single topical application, raising new concerns regarding consumer safety [7]. Thus, inorganic UV filters are becoming even more promising candidates for photoprotection [8]. They consist of either zinc oxide (ZnO) or titanium dioxide (TiO2) mineral particles. However, to date, knowledge on the potential health risks associated with sunscreens containing such inorganic UV filters remains limited. ZnO-based UV filters are often less preferred because of their high solubility in water. The dissolved Zn species are more bioavailable than ZnO particles, which increases the potential risk, notably in seawater ecosystems [9–11]. On the other hand, TiO2-based UV filters are generally considered to be more safe because they are weakly soluble in aqueous media [4,11]. The nanoparticulate form of ZnO and TiO2 (size < 100 nm) is generally used in sunscreens because their small size leads to a more transparent film on the skin after application, which enhances user acceptance. Moreover, these metal oxides are also more efficient UV blockers when finely dispersed in the sunscreen product [12,13], which also contributes to the considerable use of nanoparticles (NPs).

However, using NPs also implies specific safety concerns due to their well-known higher surface reactivity and bioavailability [14]. Nanoparticle risk assessment must, therefore, be developed in the context of the sunscreen lifecycle. Although nano-TiO2 is one of the most studied nanomaterials in nanosafety research, few of the studied NPs were relevant to the nanoparticulate TiO2 UV filters used in sunscreens. In a review of more than 200 scientific articles, Minetto et al. [15] revealed that almost all the safety investigations on TiO2 NPs were performed with bare anatase or with the highly photocatalytic P25 (mixture of anatase:rutile, 4:1), whereas the TiO2 NPs used as UV filters in sunscreen consist of pure rutile generally coated with a surface passivation layer [16,17]. The aging, transformation, and environmental fate of TiO2 NP-based UV filters have scarcely been studied to date [18]. While the important role of the NP coating in controlling fate in the suspension and surface reactivity has already been pointed out, questions as to how its nature and lifetime may influence both exposure and hazard in aquatic systems remain. These are key questions regarding the risk assessment of inorganic UV filters, which cannot be evaluated with the widely studied bare TiO2 NPs.

Moreover, the dispersing medium carrying the NPs in the original cosmetic formulation may also play an important role on the environmental fate. Either an oil or water phase can be used at the product fabrication stage to disperse the hydrophobic or hydrophilic TiO2 NP-based UV filters, respectively. While the hydrophilic filters pre-dispersed in the water phase can readily be dispersed, diluted, and transported in the aqueous environment, the hydrophobic filters pre-dispersed in the cosmetic oil may have a different environmental fate [19,20]. Once the sunscreen is washed off the skin, the hydrophobic compounds tend to remain in the oil phase of the cosmetic emulsion and float at the water surface. We can reasonably assume that before NP aging takes place, the local NP concentration in the oil droplet floating at the surface should correspond to that in the original cosmetic oil formulation [19]. This is a NP concentration leveling at 5–10 wt%, much higher than any predicted environmental concentration at the ng or μg/L level [21,22]. Nevertheless, organisms living in coastal zones at low water depths could be exposed to these higher NP concentrations. In such an exposure scenario, the oil droplet may act as a vector, transporting the concentrated NP UV filters between the seawater and living organisms, and significantly impacting the biological effects. Currently, there is no reported environmental threshold for TiO2 NP-based UV filters inducing a biological effect. It is, thus, mandatory to address the impacts they may have both in the environment and in the organisms.

In the present work, we explored the biosafety of three TiO2 NP-based UV filters on the sea urchin (*Paracentrotus lividus*) model both in the embryonic and adult life-cycle stages. Sea urchins, found in almost all the marine environments [23], are an effective sentinel of environmental stress, notably in coastal areas that are potentially more impacted by recreational bathing activity. The sea urchin embryonic development and the immunological state of adult sea urchin immune cells were investigated in culture conditions. A wide range of TiO2 NP concentrations was explored, corresponding to different levels of dilution or aging, starting from the original cosmetic dispersion. Considering the contrasted exposure scenarios expected, with an oil phase carrying hydrophobic UV filters and a water phase carrying hydrophilic UV filters, we varied the nature of the TiO2 NP coating (hydrophobic: Polydimethylsiloxane and stearic acid; hydrophilic: Silica) and of the cosmetic dispersing medium (water, oil) that were injected into the culture medium. The TiO2 NP-based UV filters appear safe, as they did not significantly compromise the growth of the sea urchin embryos at the pluteus stage (from 0.001 to 1 mg/L). The hydrophilic silica-coated TiO2 NPs did not affect immune cell viability or produce toxicity at concentrations representative of the cosmetic lifecycle, and the cells showed a broad metabolic plasticity based on the biosynthesis of metabolites promoting an increase in antioxidant activity and phagocytosis.

#### **2. Materials and Methods**

#### *2.1. Commercial TiO2 NP-Based UV Filters and Sunscreen Oil Phase*

Four different types of TiO2 NPs were used in this work. Three of them were commercial nanoparticulate UV filters commonly used in sunscreen formulations, namely, T-S (Titanium Dioxide and Alumina and Stearic Acid), T-Lite (Titanium Dioxide and Aluminum Hydroxide and Dimethicone/Methicone Copolymer), and T-AVO (Titanium Dioxide and Silica). Both T-S and T-Lite UV filters have a primary mineral layer of aluminium oxide and an outermost organic and hydrophobic layer composed of stearic acid or polydimethylsiloxane, respectively. In contrast, the T-AVO UV filter has one single mineral, hydrophilic silica (SiO2) coating. In addition, P25 TiO2, was used as an uncoated nano-TiO2 reference.

These NPs were directly purchased from the suppliers as dry powders. The respective trade names together with the chemical compositions and primary particle size provided by the manufacturers (when available) are reported in Table 1.


**Table 1.** TiO2 NP powder commercial names, supplier name, chemical composition, and primary particle size declared by the suppliers.

\* ND = not declared.

Two different dispersing media were used to mimic the release of these NPs and their respective life cycles in the environment. The hydrophobic UV filters (T-S and T-Lite) were pre-dispersed in a typical cosmetic oil, while the hydrophilic NPs (T-AVO and P25) were pre-dispersed in pure water. Further dilution was then completed to reach the desired concentrations, as detailed below.

The cosmetic oil was prepared by mixing together two emollient oils and an emulsifying agent provided by the respective suppliers (see Table 2), in a 2:2:1 ratio. The mixture was gently homogenized by magnetic stirring for 10 min. The emulsifying agent contained two surfactant molecules (Octyldodecyl xyloside and PEG30 dipolyhydroxystearate) known to play a role in the

exposure route of the hydrophobic NPs via the creation of oil droplets containing the NPs that are stable in pure water [13].

**Table 2.** Components constituting the typical sunscreen oil phase used in this work: Product names, supplier names, function of each product, and related chemical composition.


#### *2.2. Sea Urchin Paracentrotus Lividus Embryo Exposure during Development*

Six males and six females were induced to spawn by injecting 1–2 mL of 0.1 M KCl into the sea urchin body cavity, through the peristomal membrane surrounding the mouth. Eggs were collected by placing spawning females on 100 mL beakers with 0.45 μm filtered artificial seawater (ASW). Egg quality and sperm motility were inspected by observing the gametes under an optical microscope (OLYMPUS CKX31, Olympus, Tokyo, Japan); 10 μL seminal fluid was added to the egg suspension (sperm/egg ratio 50:1) and fertilization success was verified under the microscope (formation of the fertilization membrane with a fertilization rate >95%). After fertilization, the embryonic culture (500 embryos per mL) was transferred into 50-mL disposable, sterile tubes (10 mL in each tube), and embryos were immediately exposed for 48h to increasing NP concentrations (between 0.001 and 1 mg/L). The hydrophilic P25 TiO2 and T-AVO NPs were pre-dispersed and diluted in water in order to reach the desired exposure concentrations. The hydrophobic T-S and T-Lite NPs were pre-dispersed in the oil phase, as typically done during sunscreen formulation. The oil dispersion was then emulsified by dilution in pure water. An oil/water emulsion with a nominal concentration of 5 mL/L was obtained. This emulsion was then added to the embryo culture medium at nominal concentrations of 0.005, 0.05, and 0.5 mg/L. In preparing the different dispersion concentrations, a constant volume of oil was injected into the culture medium, and only the NP concentration was varied. A pure oil/water emulsion was also prepared as an oil control reference. Tests were accepted if the percentage of control embryos at 48 h of development was ≥80%, as recommended by standard procedure [24].

The degree of toxicity per treatment was calculated using the standard criteria of evaluation based on the calculation of the percentage of normal versus abnormal embryos for a sample of 100 embryos (in triplicate) by optical microscopy (48 h of development endpoint). Embryonic development was also kept under observation to evaluate the occurrence and timing of several morphological events related to the endoderm, ectoderm, and mesoderm (germ layers) development and differentiation, as reported by Pinsino et al. [25].

#### *2.3. Adult Sea Urchin Immune Cell Exposure*

Adult sea urchins (*Paracentrotus lividus*) were collected along the northwest coast of Sicily and were acclimated and maintained under controlled conditions of temperature (16 ± 2 ◦C), pH (8.1 ± 0.1), salinity (38–39%), and density (1.028–1.030 g/cm3) in oxygenated ASW (Aqua Ocean Reef Plus Marine Salt, Aquarium Line, Bari, Italy).

Approximately 0.5 mL of coelomic fluid containing freely circulating immune cells were collected from each sea urchin using a 1-mL sterile syringe already containing 0.5 mL of anticoagulant solution, namely Coelomocyte Culture Medium (CCM), composed of 1 M NaCl, 10 mM MgCl2, 40 mM Hepes-4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid- and 2 mM EGTA - ethylene glycol tetra-acetic acid- at pH 7.2. After collection, the immune cells were counted in a Fast-Read chamber (Biosigma, Madison, Wisconsin, USA), and morphological analysis of cells was performed using the optical microscope.

After counting, immune cells were plated at a density of 1 <sup>×</sup> 105 cells/well in a 96-well white, opaque-walled plate (Thermo Fisher Scientific, Waltham, Massachusetts, USA) and exposed to a final volume of 100 μL T-AVO NPs that had been pre-dispersed in stock solutions of pure water (1 and 100 mg/L nominal concentration), sterilized under UV light, and vortexed for 5 min in order to homogenize the dispersions. The NPs were added to each cell culture medium to achieve five different concentrations (0.1, 1, 10, 100, 500 mg/L final concentration). Culturing was performed in the dark at 16 ± 2 ◦C. Cell viability and cytotoxicity were measured using the RealTime-Glo MT Cell Viability Assay (Promega, Madison, Wisconsin, USA) and the non-lytic CellTox™ Green Cytotoxicity Assay (Promega, USA), respectively, as previously described [26]. Luminescence and fluorescence were detected using a GloMax Discover high-performance Microplate Reader (Promega). All assays involved at least five biological replicates (specimens).

#### *2.4. Metabolite Renewal Analysis by Mass Spectrometry in Untargeted Liquid Chromatography*

Liquid chromatography-mass spectrometry (LC-MS analysis) was performed to analyze the metabolite profile of sea urchin cells exposed to hydrophilic NPs, according to previously established protocols [27]. Metabolites were isolated in 0.5 mL ice-cold 1% acetic acid water–acetonitrile solution (70:30 *v*/*v*). Supernatants were recovered in glass inserts for solvent evaporation, dried at 30 ◦C for approximately 2.5 h (Concentrator plus/Vacufuge® plus, Eppendorf, Juelich, Germany,), re-suspended in 150 μL of H2O LC-MS grade, and injected into the UHPLC (Ultra high performance liquid chromatography)–MS system for reversed-phase liquid chromatography (RPLC). Data analysis and isotopic natural abundance correction were performed with MassHunter ProFinder and Mass Pro le Professional software (Agilent, Santa Clara, California, USA).

Samples were analyzed using an UHPLC system (Agilent 1290 Infinity UHPLC system) coupled with a quadrupole time-of-flight hybrid mass spectrometer (Agilent 6550 iFunnel Q-TOF) and equipped with an electrospray Dual JetStream source. The LC reversed phase was separated using an InfintyLab Poroshell 120 PFP-pentafluorophenyl-column (2.1 × 100 mm, 2.7 μm; Agilent Technologies) at a volume of 15 μL, a flow rate of 0.2 mL/min, and a temperature of 35 ◦C. Both the mobile phase A (100% water) and B (100% acetonitrile) contained 0.1% formic acid. The elution gradient was (1) 0 min, 100% A, (2) 2 min, 100% A, (3) 4 min, 99% A, (4) 10 min, 98% A, (5) 11 min, 70% A, (6) 15 min, 70% A, and (7) 16 min, 100% A followed by 5 min of post-run. Mass spectra were recorded in centroid mode in a mass range from *m*/*z* 60 to 1050 *m*/*z*. The mass spectrometer operated using a capillary voltage of 3.7 kV. Source temperature was set to 285 ◦C with a 14 L/min drying gas and a nebulizer pressure of 45 psig (pounds per square inch gauge). Fragmentor, skimmer, and octopole voltages were set to 175, 65, and 750 V, respectively.

Active reference mass correction was performed through a second nebulizer using the reference solution (*m*/*z* 112.9855 and 1033.9881) dissolved in the mobile phase 2-propanol-acetonitrile-water (70:20:10 *v*/*v*).

#### *2.5. Characterization of the NPs by High-Resolution Scanning Electron Microscopy*

The primary particle length of the NPs was measured using high-resolution scanning electron microscopy (HR-SEM). Three to four milligrams of the four pristine TiO2 NP powders (T-S, T-Lite, T-AVO, P25 TiO2) were dispersed on carbon adhesive tape and analyzed using a Zeiss Gemini500-Field emission SEM. To obtain surface-sensitive imaging at nanoscale resolution, images were recorded at low voltage (1–5 kV) with an in-lens secondary electron detector. Image analysis was then completed in order to distinguish the smallest particulate units constituting the powder grains. The primary particle lengths were obtained from the largest dimension of these smallest units. Fifty particles per sample were measured in order to calculate an average primary particle length.

#### *2.6. Characterization of the NP Dispersions by Dynamic Light Scattering*

The size distribution of the studied NPs was measured in their respective original dispersing medium (water or oil) using dynamic light scattering. These dispersions corresponded to the NPs as prepared before injection into the exposure media for the in vivo assay. P25 TiO2 and T-AVO NPs were each dispersed in pure water at a nominal concentration of 125 mg/L by agitating the appropriate mass of the pristine powders in 5 mL.

The oily dispersions of T-S and T-Lite NPs were prepared by dispersing the UV filters in the commercial cosmetic oil by mechanical agitation at 1000 rpm rotation speed for 10 min, at a nominal NP concentration of 25 g/L using a *Heidolph Hei-Torque* 400 stirrer equipped with a pitcher blade impeller. The hydrophobic UV filters were analyzed in oil at a solid concentration 200 times as high as that for hydrophilic UV filters because in the in vivo assay the oil dispersion was further emulsified by dilution in water but the NP concentration remained locally high in the oil droplets. Moreover, since such complex dispersion/emulsion systems cannot be measured by dynamic light scattering (DLS), only the original oil dispersion was measured here in order to provide insights on the NP aggregation state in the oil droplet.

The size measurements were performed in triplicate at 25 ◦C with 11 runs per measurement, normal resolution analysis, and 0.01 cumulant fit error tolerance, using a *Zetasizer Nano* (Malvern Instruments, Malvern, UK).

The characterization of the particles and the degree of dispersibility in each pre-dispersant medium are reported in Supplementary Materials.

#### *2.7. Statistical Analysis in Biological Assays*

Statistical analyses were performed by GraphPad Prism Software 6.01 (USA). Statistical differences among selected groups were estimated by one-way ANOVA (followed by the multiple comparison tests). The p-value lower than 0.05 was deemed statistically significant. Data were expressed as mean ± standard deviation (SD).

#### **3. Results and Discussion**

#### *3.1. Influence of TiO2 NP-Based UV Filters on the Growth of Sea Urchin Embryos*

The TiO2 NPs used as commercial UV filters in sunscreen consist of the rutile lattice form, which is known to be less photoreactive and less toxic than other forms (i.e., anatase, brookite) [16,17]. They are generally coated with a surface passivation layer aimed at suppressing the rutile photocatalytic activity (e.g., Al2O3, SiO2) and a secondary layer added to enhance particle dispersion in the cosmetic formulation (e.g., hydrophobic polydimethylsiloxane, hydrophilic Na-polyacrylate). Recent field studies of recreational bathing waters have evidenced a distinct fate for the hydrophobic compounds released from sunscreen [19,20]). Notably, measured TiO2 concentrations were higher (up to 1000x) in the water top surface layer (0.1–0.9 mg/L) than in the water column (0.02–0.05 mg/L).

Here, we investigated the effects of increasing concentrations of three TiO2 NP-based UV filters on the sea urchin *P. lividus* in the embryonic life-cycle stage. Embryos were classified as normal only when they satisfied all the following morphological criteria: (1) Acceptable schedule in reaching the developmental endpoint; (2) dorso/ventral and left/right embryonic axis symmetry; (3) correct differentiation of oral/aboral endoderm and ectoderm; and (4) correct mesenchyme differentiation, distribution pattern, and shape. At the gastrula stage (24 h of development), embryos exposed from fertilization maintained a regular time schedule and proper sites of spicule elongation at all concentrations tested (not shown). In agreement, at the pluteus stage (48 h) TiO2 NP-based UV filter-exposed embryos displayed a low number of abnormalities (Figure 1). A slight increase in the incidence of potential morphological abnormalities was observed only in embryos exposed to 1 mg/L of the T-AVO UV filter (hydrophilic) and 0.05 mg/L of the T-Lite UV filter (hydrophobic). Specifically, about 20% of the embryos exhibited problems on arm and skeleton rod development and/or

randomly distributed atypical big cells (Figure 2). The most common skeletal malformations observed were: (1) Crossed or separated tips at the hood apex arms (Figure 2, red arrows); (2) asymmetrical arm lengths; and (3) decrease or increase in arm growth and supporting skeletal rods (Figure 2, black arrows, oral and post-oral skeletal rods). Similar skeleton-defective embryos were previously observed in *P. lividus* embryos exposed to other types of NPs, suggesting that skeleton abnormalities could be considered a sensitive target to NP exposure [28,29]. Besides, the formation of big cells or masses (Figure 2, blue arrows) was previously observed in *P. lividus* embryos exposed to PS-NH2 NPs and in *Strongylocentrotus droebachiensis* embryos exposed to Ag NPs [25,30]. These cells could be mesodermal cells involved in NP internalization (immune defense). Others have reported that NP internalization can initiate the immune response responsible for the formation of big mesodermal cells [25], such as those seen in Figure 2. However, this hypothesis remains essentially speculative because NP internalization was not verified here. Notably, other plausible explanations cannot yet be excluded, such as the possibility that the formation of big cells or masses may be caused by other compounds and not by NP internalization. However, similar effects were not observed in embryos exposed to the pure sunscreen oil phase only (Figure 1b), indicating that the oil phase compounds could not be incriminated in the formation of big cells or masses. Further studies are needed to clarify the underlying mechanism.

**Figure 1.** The hydrophilic and hydrophobic TiO2 NP-based UV filters at different concentrations and dispersant phases influence the sea urchin embryonic development without producing a relevant developmental impairment. Histograms represent the results expressed as mean percentage (%) of normal embryos ± standard deviation after 48 h of exposure to the (**a**) hydrophilic (T-AVO and P25) and (**b**) hydrophobic (T-S and T-Lite) TiO2 NP-based UV filters. \* *p* < 0.05.

**Figure 2.** Optical images of representative sea urchin *Paracentrotus lividus* exposed to TiO2 NP-based UV filters. Black arrows indicate decrease or increase in arm growth and supporting skeletal rods. Red arrows indicate crossed or separated tips at the hood apex arms. Blue arrows indicate the atypical big cells found in abnormally developed embryos.

In addition, we would like to point out that the morphological abnormalities observed in the sea urchin embryos exposed to NPs may not even be of ecological significance. Indeed, there was no evidence that these abnormalities reduced larvae survival ability, thus endangering a whole of one population.

Li et al. proposed a model of interaction between NPs and lipid layers of the plasma membrane based on the hydrophobic/hydrophilic nature of particles [31]. Based on the model, they suggested that hydrophobic NPs, which are thermodynamically stable around the core of a bilayer hydrophobic membrane, lead to lipid molecule deformation and distribution, whereas hydrophilic NPs that adsorb on the membrane surface (ready to be phagocytized) enter the hydrophobic core of the membrane, maintaining it intact. The primary particle shape also influences and modulates the particle behavior in a medium and its subsequent biological effect on a living system. Brown et al. argued that rod-shaped NPs could interact more strongly with biological systems than round-shaped NPs because the van der Waals interaction forces in lengthwise-oriented NPs increase proportionally to their length, typically reaching values several orders of magnitude above that of spheres [32]. Based on this theory, the elongated shape of T-AVO NPs compared to the P25 TiO2 NPs (Figure S1, Supplementary Materials) would facilitate the interaction of particles with the membranes.

It is unclear how the dispersion states of the different UV filters measured here in the aqueous or oil cosmetic medium (Table S1, Supplementary Materials) are altered after dispersion and dilution in the aqueous culture medium. Aggregation and sedimentation will surely occur in artificial seawater for the hydrophilic UV filters pre-dispersed in water because salt-induced aggregation is a well-known effect in such a system [33]. On the contrary, further aggregation is not expected for the hydrophobic UV filters, as they should remain in the oil phase during the exposure. It is important to mention that hydrophobic filters trapped in the surface microlayer may not be bioavailable to the organisms living in suspension because the oil phase does not naturally mix with water. However, the surfactant molecules used in the cosmetic formulation may favor the mixing of the aqueous and oil phases and might play a determining role here (Table 2). For example, it is known that the octyldodecyl xyloside has high affinity for T-Lite NP surface [13]. Thus, we can reasonably assume that once washed of the consumer's skin, these molecules can modify the NP fate in the aquatic environment and their subsequent bioavailability for the marine organisms.

Overall, our results confirmed that the tested hydrophilic and hydrophobic TiO2 NP-based sunscreen filters do not elicit significant harmful effects on sea urchin embryonic development. Of note, we were not fully satisfied with the efficacy of our evaluation assay with the hydrophobic UV filters pre-dispersed in the oil. Increasing the NP concentration in the oil implies an increase in viscosity. This makes the approach difficult to apply at high concentrations, as well as in tests under static conditions (e.g., primary immune cell culture). For this reason, only the hydrophilic TiO2 NPs were used in subsequent immune cell-based assays.

### *3.2. Sea Urchin Adult Immune Cells: Health State and Metabolic Typing under Hydrophilic TiO2 NPs*

An ideal sunscreen active agent should remain localized close to the skin surface without penetrating into the deeper layers because by entering in the systemic circulation it could stimulate immune reactions. The sea urchin *P. lividus* can function as a proxy for humans for in vitro immunological studies [34]. Thus, to focus on the sea urchin immunological tolerance to the silica-coated TiO2 NP-based UV filters we assessed the viability and cytotoxicity of the exposed immune cells to the hydrophilic T-AVO NPs for 48 h and we characterized their metabolic profile after 72 h of exposure. Cell viability and cytotoxicity were monitored and measured in real time for cells exposed to increasing concentrations of T-AVO NPs (0.1, 1, 10, 100, 500 mg/L). Only the measurements at 48 h are shown (Figure 3). The highest concentrations (100, 500 mg/L) were used as positive controls to demonstrate the appropriateness of the procedures.

**Figure 3.** Impact of T-AVO UV filters on the sea urchin immune cell viability and toxicity. Real-time viability over two days of continuous monitoring, of which one measurement point (48) is shown (0.1, 1, 10, 100, 500 mg/L final concentration). The highest doses (100 and 500 mg/L) provoked decreases in cell viability and increase in cell toxicity. Levels are expressed in arbitrary units as fold increase or decrease compared to controls assumed as 1 (dot line). Data are reported as the mean ± SD; stars (\*) indicate significant differences among groups (\* *p* < 0.05; \*\*\* *p* < 0.001).

Our results confirmed those obtained from the sea urchin embryonic development assay, where no significant toxic effects were found at concentrations from 0.1 to 10 mg/L. At the two highest concentrations of exposure (100 and 500 mg/L), a significant decrease in viability and a high cell toxicity were measured, as expected. At these concentrations, faster NP aggregation likely led to particle destabilization and sedimentation, which, in turn, affected cellular viability due to the aggregate–cell physical contact. For samples exposed to 0.1–10 mg/L T-AVO, the RealTime-Glo MT cell viability assay indicated an increase (not statistically significant) in cell viability/metabolic activity compared to unexposed controls. Notably, the chemistry assay is based on the reducing potential of the cell, which is a known metabolic marker of cell viability. This result may be due to a hysteresis response usually observed under drug administration, in which the effect of a drug declines despite its continued presence (drug tolerance) [35]. Studies elucidating metabolic profiles of immune cells exposed to T-AVO were, therefore, carried out to confirm this increased metabolic activity, as reported below.

Immune functions are bio-energetically expensive, requiring accurate management of metabolites coordinated by intracellular and extracellular signals, which direct the uptake, storage, and utilization of substrates (e.g., glucose, amino acids, fatty acids). In turn, metabolites renew and control immune responses [36]. In order to obtain integrated data on the sea urchin immune metabolic state during exposure to hydrophilic TiO2 NP-based UV filters and their related tolerance state, we characterized the metabolic profile of cells exposed to T-AVO and compared it with the profiles of cells exposed to P25 and unexposed cells (72 h in culture) (Figure 4). To this purpose we used an initial higher dose of particles (2 mg/L, loading dose) to achieve a lower maintenance dose (1 mg/L).

**Figure 4.** Sea urchin immune cell metabolic profile under hydrophilic nano-TiO2-based UV filters. Untargeted metabolic profiling of T-AVO and P25 exposed for one day at 2 mg/L and for two days at 1 mg/L and unexposed immune cells (Ctr) for 72 h. Hierarchical clustering heatmaps display significantly (*p* ≤ 0.05) different intracellular metabolites by LC-MS. Metabolic typing was performed on *P. lividus* primary immune cell cultures obtained from five individual donors.

Metabolite profiling identified level changes of only eight metabolites among groups (Group 1: T-AVO 1-day exposure at 2 mg L−<sup>1</sup> followed by 2-days exposure at 1 mg L<sup>−</sup>1; Group 2: Control, 3 days; Group 3: P25 1-day exposure at 2 mg/L followed by 2-days exposure at 1 mg/L), including proteinogenic amino acids (serine, glutamine), amino acid derivatives (L-pyroglutamic acid), organic acids (L-lactic acid, glutaric acid, pyruvic acid), sulfate metabolites (sulfate), and acetamides (acetamidovalerate). Sea urchin T-AVO-responsive metabolites predominantly involved in mediating inflammatory signals and phagocytosis were found. Specifically, glutamine and acetamidovalerate were significantly increased compared to unexposed controls, while L-pyroglutamic acid and L-lactic acid were significantly decreased. It is well known that to re-establish normal cellular and molecular function, immune cells have higher glutamine needs during inflammatory states [37]. Notably, acetamides are known to relieve inflammatory events; in fact, they are used as chemotherapeutic agents for inflammation-associated cancers [38].

Lactic acid is produced in high amounts by innate immune cells during inflammatory activation (anaerobic glycolysis product). A reduction in its amount may be translated into a negative feedback signal to silence or attenuate inflammatory responses [39].

Increased levels of pyroglutamic acid (also called 5-oxoproline) are able to promote lipid and protein oxidation and to enhance hydrogen peroxide content, thus promoting oxidative stress [40]. Consequently, the significantly decreased pyroglutamic acid levels observed under T-AVO exposure may be considered a signal of an increased antioxidant metabolic activity. Although slight, L-serine and pyruvic acid levels were also increased compared to those in unexposed cells. Serine racemase enzyme catalyzes the α, β elimination of water from L-serine to produce pyruvate and ammonia [41]. Notably, L-serine metabolism is required for fueling one-carbon metabolism and nucleotide biosynthesis, and it is known to lower the inflammatory responses in mice during infection [42].

The P25 TiO2 bare NPs were used here as a negative control for the silica-functionalized T-AVO. In our recent studies we demonstrated that under TiO2 NP exposure (1 mg/L TiO2 NPs, 24 h endpoint) the innate sea urchin immune system is able to control inflammatory signaling, excite antioxidant metabolic activity, and acquire immunological tolerance [27,43]. Notably, sea urchin immune system metabolic typing under P25 exposure for 72 h (one day of exposure at 2 mg/L followed by two days of exposure at 1 mg/L) highlighted a different scenario compared to the respective T-AVO and control groups. L-glutamine, acetamidovalerate, and pyriglutamic acid levels all presented a trend similar to controls, while lactic acid levels were similar to T-AVO-exposed cells and pyruvic acid levels were considerably increased. Interestingly, L-serine, sulfate, and glutaric acid levels presented a trend completely different to both T-AVO and control groups. Specifically, L-serine and sulfate were reduced compared to controls, while glutaric acid levels increased, highlighting an ongoing inflammatory state. These findings show the superior immune compatibility of T-AVO compared to P25 when particles stay in contact with the sea urchin immune cells for 72 h (one day of exposure at 2 mg/L, and two days of exposure at the half dose). Based on the notion that the silica coating of commercial TiO2 NP-based UV filters undergoes a fast degradation once released into aquatic media (88–98% silica removal within 96 h) [44], we speculated that the T-AVO in contact with cells after 72 h of exposure consisted of mainly pure rutile. Worth noting is the fact that not only the shape but also the crystal composition of TiO2 particles (e.g., T-AVO: Pure rutile; P25: Anatase:Rutile) makes their use more or less safe, as documented in the literature [45].

Overall, our results demonstrate that the commercial TiO2 NP-based UV filters tested in this work do not show any significant harmful impact toward the development or immunity of the sea urchin (*P. lividus*) at NP concentrations expected near the seashore during summer recreational activities. A summary view of the experimental setup and the results is schematized in Figure 5.

These results are in accordance with the majority of risk assessment studies on different marine organisms using pure rutile or anatase NPs that are summarized in the most relevant reviews [3,46]. The different coatings on the commercial UV filters, together with particle shape and the original dispersant phase, slightly modulate the effects toward embryo development but without causing a relevant developmental impairment. In this context, the particle shape of hydrophilic NPs pre-dispersed in water, predominantly influences the interaction with sea urchin embryos compared with other physical features such as primary particle size or aggregation state. Particularly, the elongated, rod-shaped T-AVO NPs slightly impacted the development compared to the more spherical P25 NPs. The effects of hydrophobic UV filters are, instead, in line with their respective dispersion capacity in the former sunscreen oily dispersant medium, which is directly related to their different particle external coatings. The more finely dispersed T-Lite NPs showed a few visible effects compared to the T-S NPs, probably because of specific interactions with the octyldodecyl xyloside surfactant present in the sunscreen oily medium, which would likely ease its transportation inside the embryonic culture medium.

Viability and toxicity tests on immune cells showed no toxicity of T-AVO NPs at concentrations consistent with that of the cosmetic life cycle. Furthermore, metabolic profile characterization performed on P25 and T-AVO NPs showed that the T-AVO NPs have a superior immune compatibility to that of the P25 after 72 h of interaction with sea urchin immune cells, ultimately supporting the safety of this type of commercial TiO2-based UV filter on the immunological state of the sea urchin.

Notably, the application of metabolomics in the environmental field is relatively new, but the technique is actively attracting the attention of the scientific community because results obtained successfully describe a "picture" of the biochemistry of an organism, cell, or tissue at any one time [47].

**Figure 5.** Schematic illustration of the results obtained from the evaluation of the TiO2 NP-based sunscreen UV filters on the development and the immunological state of the sea urchin *Paracentrotus lividus*. Models: Sea urchin embryos and adult *P. lividus* immune cells in vitro. Endpoints: Sea urchin embryos at the pluteus stage (48 h), immune cell viability/toxicity (48 h), and metabolomics (72 h). Commercial nanoparticulate UV filters with three surface coatings: T-S, T-Lite, and T-AVO tested on embryonic development and T-AVO on immune cells. P25 was used as an uncoated nano-TiO2 reference. The results presented here highlight the biosafety of TiO2 NP-based UV filters on sea urchin, and the importance of developing safer-by-design sunscreens and evaluating the associated risk. Created by Andi Alijagic using BioRender.com.

#### **4. Conclusions**

Our findings underline the importance of developing sustainable sunscreen formulations based on nanoparticulate UV filters, rather than the bare NP counterpart, to minimize the environmental risk posed by sunscreen products. Further studies with a similar approach need to be performed in the future, on different biological models and with different experimental conditions, in order to fully confirm the safety of these nano-products. Primary cell cultures accurately represent the biological microenvironment in which cells reside in tissues, as cell–cell signaling remains preserved [48]. However, the next step should be to study the immunological state of the sea urchins and other organisms that live in the beaches during recreational activity (field study).

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/10/11/2102/s1, Figure S1: High-resolution scanning electron microscopy (HR-SEM) analysis of the pristine TiO2-based NPs. Table S1: Comparison of primary particle size and hydrodynamic aggregate size of the TiO2NPs.

**Author Contributions:** R.C.: investigation, writing-original draft preparation; J.L.: conceptualisation, data interpretation, writing—review and editing, project administration; D.G., A.A., E.N., and A.C.: methodology and data curation; D.S.: writing—review and editing; A.P.: conceptualisation, supervision, data interpretation, writing—review and editing, project administration. All authors have read and agreed to the published version of the manuscript.

**Funding:** This project was funded from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement [No 671881], the Marie Skłodowska-Curie grant agreement [No 713750], and from the Excellence Initiative of Aix-Marseille University—A\*MIDEX, a French "Investissements d'Avenir" program, through its associated Labex SERENADE project. This work is also a contribution to the OSU-Institut Pythéas.

**Acknowledgments:** The authors gratefully acknowledge M. Biondo for his technical support and sea urchin husbandry.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **The E**ff**ects of In Vivo Exposure to Copper Oxide Nanoparticles on the Gut Microbiome, Host Immunity, and Susceptibility to a Bacterial Infection in Earthworms**

**Elmer Swart 1,\*, Jiri Dvorak 2, Szabolcs Hernádi 3, Tim Goodall 1, Peter Kille 3, David Spurgeon 1, Claus Svendsen 1,\* and Petra Prochazkova <sup>2</sup>**


Received: 9 June 2020; Accepted: 6 July 2020; Published: 9 July 2020

**Abstract:** Nanomaterials (NMs) can interact with the innate immunity of organisms. It remains, however, unclear whether these interactions can compromise the immune functioning of the host when faced with a disease threat. Co-exposure with pathogens is thus a powerful approach to assess the immuno-safety of NMs. In this paper, we studied the impacts of in vivo exposure to a biocidal NM on the gut microbiome, host immune responses, and susceptibility of the host to a bacterial challenge in an earthworm. *Eisenia fetida* were exposed to CuO-nanoparticles in soil for 28 days, after which the earthworms were challenged with the soil bacterium *Bacillus subtilis*. Immune responses were monitored by measuring mRNA levels of known earthworm immune genes. Effects of treatments on the gut microbiome were also assessed to link microbiome changes to immune responses. Treatments caused a shift in the earthworm gut microbiome. Despite these effects, no impacts of treatment on the expression of earthworm immune markers were recorded. The methodological approach applied in this paper provides a useful framework for improved assessment of immuno-safety of NMs. In addition, we highlight the need to investigate time as a factor in earthworm immune responses to NM exposure.

**Keywords:** innate immunity; infection; microbiome; survival; nanomaterials; nanoparticles; copper; earthworms; *Eisenia fetida*

#### **1. Introduction**

Nanomaterials (NMs) are increasingly used in various applications including surface coatings, biocide pesticides, and electronics [1,2]. The potential risks of NM to human health and the environment have long been identified [3–5]. Over the last decade, research has provided vast amounts of toxicity data that have reduced many of the initial uncertainties around NM risk. There are, however, still some aspects that need further investigation. One of these remaining issues relates to the immuno-safety of NMs [6–8]. Owing to their particulate nature, NMs have an increased potential to interact with the innate immune system of organisms [9–12] and to induce both pro- and anti-inflammatory responses [13]. Most of the current research to investigate such effects has used in vitro models to

characterize NM–immune interactions. Although these studies have provided crucial information on how immune systems may interact with NM, it remains unclear how responses in vitro will translate to in vivo effects. Further, immuno-modulation by NM does not necessarily indicate that the immune system is being compromised. In fact, immune reactions are part of a healthy response by the host towards foreign objects. In order to assess whether NMs actually compromise host immunity, co-exposure with infectious pathogens is necessary [7,8].

A major application of NMs is as antimicrobial agents in pesticides and coatings. The effects of biocidal NM on soil microbial communities have been relatively well studied [14–16]. There are, however, uncertainties concerning the impact of NMs on microbes associated with plants and animals (commonly referred to as 'microbiome') [17]. Common roles of the microbiome in host health include the provision of essential nutrients and aiding in digestion, while the role of the microbiome in host immunity is now increasingly being recognized [18–20]. Microbes associated with mucosal surfaces can contribute to immunity by providing resistance against invading pathogens [21–23] and by stimulating the release of antimicrobial peptides by the host [24]. Disruption of the healthy microbiome by, for example, chemical exposure, can lead to reduced immune functioning and reduced survival of bacterial infections [25–27]. Previous studies have shown that, when NMs alter the microbiome of animals, the expression of host immune genes can also change [28–31]. Host immunity and the microbiome are thus a complex and integrally linked system, and it is thus important to include microbiome analysis in the immuno-safety assessment of NM. Invertebrate animals such as earthworms, bivalves, and sea urchins provide suitable models to study in vivo effects of NM on immune functioning under more realistic environmental conditions [31–33].

Earthworms provide crucial ecosystem services in soils through mixing, organic matter degradation of plant material, and nutrient cycling, and thereby contribute to enhanced crop production [34]. Living in soil, earthworms inhabit an environment with high microbial activity. To provide protection from pathogens, cellular immunity in earthworms is provided by immune cells called coelomocytes, which circulate the coelomic cavity. A crucial element of the innate immune system is the recognition of microorganism associated molecular patterns (MAMPs) by host pathogen recognition receptors (PRRs). A well described PRR in earthworms is coelomic cytolytic factor (CCF). This PRR, upon binding to specific MAMPs, induces the prophenoloxidase pathway, which ultimately leads to the production of antimicrobial factors [35–37]. Earthworm pathogens are also controlled by various humoral factors. One of these is lysozyme, an enzyme that can hydrolyse components of the cell wall of Gram-positive bacteria [38]. In the earthworms *Eisenia fetida* and *Eisenia andrei,* immunity is also supported by the humoral factors lysenin [39] and fetidin [40,41], the modes of antibacterial action of which are not fully understood [40,42]. Recent work shows that changes in gene expression of these immune factors can be used as a marker of immune-modulation in earthworms [38,43,44]. In vitro studies have shown that exposure to NM can also alter the expression of earthworm immune genes [45,46]. Effects of in vivo exposure to NM earthworm immune system regulation have not been studied extensively [33] and it remains uncertain whether NM exposure can compromise earthworm immunity by affecting the host susceptibility to infections.

The earthworm gut microbiome has been relatively well described. Gut communities have been identified as being composed of both transient bacteria associated with ingested soil and food [47] and resident bacteria more closely associated to intestinal surfaces [48–50]. Loss of some core earthworm symbionts can lead to reduced host fitness and juvenile development [51,52]. Environmental pollutants can alter the microbiome of earthworms [53–56] and lead to the loss of core symbionts important to host health [57]. An increasing body of literature now indicates that exposure to NMs can also disrupt the microbiome of soil invertebrates [58–60]. In the earthworm *Enchytraeus crypticus*, for example, exposure to CuO-NP can significantly reduce the abundance of core intestinal *Plantomycetes* bacteria [58]. The interplay between microbiome, NMs, and host immunity in earthworms, however, has not been studied. Disruption of host–microbiome interactions can be expected for chemicals that are designed to target microbes (biocides). In agriculture, copper-based NM formulations are being developed

for biocidal applications [61]. In widespread application, there is thus the potential for such NMs to negatively affect earthworms through alterations to their microbiome structure.

Because of the integrate link between the microbiome, host immunity, and health status of animals, studies on the immuno-safety assessment of NM require a holistic approach. This paper aims to study whether exposure to copper oxide nanoparticles (NPs) has an effect on the gut microbiome structure, host immunity, and susceptibility to a bacterial infection in earthworms. For this purpose, earthworms were exposed in soil to concentrations of copper forms known to alter the earthworm gut microbiome for a duration of 28 days. The earthworms were subsequently removed from soils and challenged with the bacterium *Bacillus subtilis* for a further four days. The effects of the bacterial challenge were assessed by looking at survival and tissue damage, and by measuring mRNA levels of known immune markers. An analysis of the gut microbiome was concurrently conducted through a metabarcoding approach to link the effects on microbiomes to immune responses. The effects of NP were compared to those of metal salts, to test whether any effects were attributed to particles or ions. We hypothesized that (i) earthworms that have their microbiome changed through exposure to CuO-NP and copper salts are more susceptible to a bacterial infection; and (ii) exposure to CuO-NP, copper salts, and the proceeding bacterial challenge will have an effect on the gene expression of tested immune markers, in line with previous studies [44,62]. The holistic methodological approach applied in this paper provides a useful framework for improved assessment of immuno-safety of NMs.

#### **2. Materials and Methods**

#### *2.1. Test Organism, Test Chemicals, and Soil Spiking*

*Eisenia fetida* were reared at 20 ◦C in a medium consisting of loamy top soil, composted bark, and garden compost in 1:1:1 ratio by volume basis. Earthworms were fed with field collected horse manure from horses grazing on unpolluted pastures and free from recent medical treatment. All earthworms used in the experiment had a stripe patterned outer body characteristic of *E. fetida* with fully developed clitella and were within a weight range between 300 and 600 mg.

Molecular grade CuCl2·2H2O was supplied by Sigma-Aldrich (Poole, UK). CuO-NPs were manufactured by Promethean Particle Ltd. (Nottingham, UK) and were dispersed in water. Nanoparticles were cuboid in shape, with a stated mean dimension of 20 by 50 nm. Size distributions of NPs were determined with nanoparticle tracking analysis using a Nanosight (Malvern Instruments, Salisbury, UK). Derived mean and modal dimensions were 183 nm (±SE 5.2). Zeta potential of CuO-NPs (33 mV ± SD 0.3 mV) was determined using phase analysis light scattering using a Malvern Zetasize Nano ZS.

All exposures in soil were conducted in LUFA 2.2 natural soil (LUFA-Spreyer, Germany), a sandy loam soil widely used in ecotoxicological testing. Test soils were spiked with a nominal concentration of CuO-NP (160 mg·kg−<sup>1</sup> d.w. soil) or CuCl2 (160 mg·kg−<sup>1</sup> d.w. soil) or a negative control (0 mg·kg−<sup>1</sup> d.w. soil). Test concentrations were based on a previous study that showed changes in the microbiome structure and loss of core symbionts at these copper concentrations [56]. CuO-NP treated soils were spiked one day before the initiation of the exposure. CuCl2 treated soils were spiked five days before the start of the exposure to allow the metal speciation in the soil to reach a quasi-equilibrium [63]. A control consisting of the liquid carrier of the CuO-NP dispersion was not included, as previous work has established that the liquid carrier of this NP dispersion is not toxic to earthworms and does not alter the earthworm microbiome structure [56].

Mixing of the chemicals with the soil was done for each treatment following Waalewijn-Kool et al. [64]. Briefly, the total amounts of CuO-NP and CuCl2 required for all replicates were dissolved in 60 mL of de-ionised water. These stock solutions were then each mixed with 250 g of d.w. LUFA 2.2 soil using a spatula. The mixture was then thoroughly mixed with the remaining soil and subsequently wetted with de-ionised water to reach 55% of the water holding capacity (WHC) before final mixing. The soil mixtures were divided into replicates each consisting out of a 60 g w.w. aliquot in a 100 mL clear plastic round tub.

#### *2.2. Overview of the Experimental Design*

Earthworm were initially exposed to a pre-treatment of either CuO-NP, CuCl2, or a negative control for 28 days, after which time they were removed from soil and challenged with either *Bacillus subtilis* or a negative control for four days (Figure 1). *B. subtilis* was chosen as a model pathogen for the bacterial challenge on the basis of a pilot experiment, which showed that the coelomic fluid of *E. fetida* had inhibiting effects on the growth of this bacterium, possibly indicating an effective immune response by cellular or humoral components (Figure S1). After this bacterial challenge, earthworms were returned to their original soil for recovery for another 28 days, during which time they were periodically sampled for analysis. Details of each of the three experimental steps are described below.

**Figure 1.** Schematic overview of the study design and collected samples. Dashed lines indicates sampling points. Tick marks indicate the samples collected at the respective sampling point. 'DNAgut' indicates sampling of DNA from gut tissue for microbiome analysis; 'RNAgut' and 'RNAcc' indicate sampling of RNA for gene expression analysis from gut tissue and coelomic fluid, respectively; and 'Histlgut' indicates sampling of gut tissue for histological analysis. NP, nanoparticle; PBS, phosphate buffered saline.

#### *2.3. Pre-Treatment Exposure*

Prior to the start of the pre-treatment exposure, earthworms were acclimatized to LUFA 2.2 soil for two weeks under the same conditions as the main exposure assay. Before exposure initiation, earthworms were rinsed and weighed. To start the test, one adult *E. fetida* was added to each test replicate. The pre-treatment exposure was conducted at 20 ◦C for 28 days. Once a week, each replicate received 0.5 g of spiked horse manure on a d.w. basis as food. At the end of the exposure, earthworms from all replicates were removed from soil, rinsed, and weighed. Collected earthworms were subsequently depurated on wetted filter paper to allow egestion of their gut content for two days before bacterial challenge.

#### *2.4. Bacterial Challenge*

Following depuration, each earthworm was challenged in a petri dish with either *Bacillus subtilis* (*B. subtilis* subsp. *subtilis*, CCM2217; Czech Collection of Microorganisms, Brno, Czech Republic) or a negative control in a medium consisting of re-wetted paper pellets. The pellets were re-wetted using either 10 mL of phosphate buffered saline (PBS) containing 5 <sup>×</sup> <sup>10</sup><sup>8</sup> *B. subtilis* cells per mL or 10 mL PBS only. *B. subtilis* cultures used for the challenge were in exponential growth phase at the time of the start of the challenge. Cell concentration was determined by measuring OD600. The bacterial challenge was initiated by placing a depurated earthworm into the prepared Petri dish. The bacterial challenge was conducted at 20 ◦C in dark conditions for four days. Earthworm survival was monitored at day one, two, and four.

#### *2.5. Recovery Period*

After the bacterial challenge, earthworms were rinsed, weighed, and subsequently returned to the original soil exposure replicates to assess responses to exposure after the bacterial challenge. The recovery exposure was conducted at 20 ◦C and lasted 28 days. Earthworms were fed with 0.5 g of d.w. spiked horse manure every week.

#### *2.6. Sample Points*

Six sampling points were used: 'pre-treatment day 0', 'pre-treatment day 28', 'bacterial challenge day 2', 'bacterial challenge day 4', 'recovery period day 1', and 'recovery period day 28' (Figure 1). At every sampling point, except 'bacterial challenge day 4', gut tissue and coelomic fluid from five earthworms were collected (see below). At every sampling point, one additional earthworm was collected for histological analysis (see below). All earthworms collected from soil exposure replicates (i.e., 'pre-treatment day 0', 'pre-treatment day 28', 'recovery period day 1', and 'recovery period day 28') were depurated for two days prior to sampling. Earthworms collected at day two of the bacterial challenge were rinsed in de-ionised water, but not depurated and immediately dissected. At the end of the pre-treatment exposure, 10 g w.w. soil was collected for metal analysis.

#### *2.7. Sampling of Gut Tissue and Coelomic Fluid*

Each depurated earthworm was placed in a petri dish containing 500 μL PBS and coelomic fluid was extruded by electrification for 5 s using a 4.5 V battery. The mixture of coelomic fluid and PBS was collected and mixed with 500 μL of 2× RNA/DNA Shield (Zymo Research, Irvine, CA, USA) and placed in a lysis tube. The extruded earthworm was subsequently euthanized in pure ethanol and the midgut (spanning 20 segments posterior to the clitellum) was dissected using sterile equipment. Small incisions were made along the length of the midgut and rinsed in 1 mL PBS for 1 min using a vortex to facilitate removal of any residual soil, and subsequently placed in a lysis tube. All lysis tubes were bead beaten using an MP FastPrep-24TM set at 4.5 m/s for one minute. Lysis tubes were placed at 6 ◦C overnight and subsequently stored at −20 ◦C until DNA or RNA extraction.

#### *2.8. Histological Analysis*

Earthworms collected for histological analysis were fixed and processed according to Dvorak et al. [44]. Briefly, a whole body sample spanning a 10 segment region posterior to the clitellum was fixed in 4% paraformaldehyde overnight, dehydrated, and embedded in paraffin. For each sample, three 2 μm sections were cut using a microtome, dried overnight, deparaffinised using xylene, rehydrated, and stained using hematoxylin/eosin following Kiernan [65]. Sections were visually inspected for tissue integrity using a light microscope. Damage to the gut epithelium and chloragogen tissue was scored using an ordinal scoring method with four categories ((1) no effects, (2) mild effects, (3) moderate effects, and (4) severe effects), following Gibson-Corley et al. [66].

#### *2.9. Soil Metal Measurements*

Soil copper concentrations were measured in 130 mg of d.w. soil, which was mixed with a 4:1 mixture of nitric acid and hydrochloric acid on a volume basis and digested for seven hours at 150 ◦C. Copper concentration was determined using atomic absorbance spectrometry at the Vrije Universiteit

Amsterdam (The Netherlands). Copper recovery from exposure soils was on average 60.4% and 71.3% for CuO-NP and CuCl2 spiked soils, respectively. The lack of full recovery may be linked to the loss of copper during preparation of stocks solutions and owing to heterogeneity in the distribution of copper forms in soils.

#### *2.10. DNA and RNA Extraction and cDNA Synthesis Procedure*

DNA was extracted from gut tissue and soil using a Quick-DNA Fecal/Soil Microbe Miniprep Kit (Zymo Research) according to the protocol supplied by the manufacturer. RNA was extracted from gut tissue and coelomic fluid using Quick-RNA™ Miniprep Kit (Zymo Research) following the protocol supplied by the manufacturer and included a DNA removal step using DNase. Visual inspection through agarose gel electrophoresis under denaturing conditions verified that the RNA in all samples was not degraded. Extracted RNA was subjected to a further clean-up using a Clean and ConcentratorTM-5 kit (Zymo Research). RNA quantity of the cleaned samples was determined using Qubit™ RNA HS Assay Kit (ThermoFisher Scientific, Waltham, MA, USA). Per sample, 250 ng of RNA was reverse transcribed to cDNA using Reverse Transcription System A3500 (Promega, Madison, WI, USA) following the standard protocol supplied by the manufacturer.

#### *2.11. Earthworm Genotyping*

Gene expression analysis relies on accurate binding of primers to target genomic regions. Genetic variation in binding sites between different individuals is likely to reduce the efficacy to elucidate patterns of gene expression. Within species genetic diversity in earthworms is high [67–69]. On the basis of cytochrome c oxidase I (COI) sequence similarities, two distinct genetic *E. fetida* clades have so far been recognised [70]. To screen whether earthworms used in this experiment were part of a single genetic clade, earthworms sampled at day two of the bacterial challenge were genotyped by amplification and sequencing of the mitochondrial COI DNA. PCR reactions were set up using forward primer COI\_1490 (5 -GGTCAACAAATCATAAAGATATTGG-3 ) and reverse primer HCO\_2189 (5 -TAAACTTCAGGGTGACCAAAAAATCA-3 ) [71] using One*Taq*® Hot Start polymerase and reaction buffer (New England Biolabs) using the following programme: initial denaturation at 94 ◦C for 2 min followed by 35 cycles of (1) denaturing at 94 ◦C for 30 s, (2) annealing at 47 ◦C for 30 s, and (3) extension at 68 ◦C for 1 min, followed by a final extension step at 68 ◦C for 10 min. Amplification of a single fragment was verified through gel electrophoresis. PCR products were cleaned using a QIAquick PCR purification kit (QIAGEN) and DNA quantity was assessed using a Qubit dsDNA HS Assay Kit (ThemoFisher Scientific). Then, 7.5 ng of PCR product was sequenced using Sanger sequencing using 3.2 pg of the forward primer at the University of Birmingham (UK). Sanger sequences were submitted to the National Center for Biotechnology Information (NCBI) BLASTn for taxonomical assignment. Pairwise alignment of sequences was performed using MUSCLE alignment in Geneious 9.1.8. Ambiguous bases and erroneous inserts were manually resolved and low quality ends of sequences were trimmed. The remaining 632 bp alignment was used as input for genetic analysis in MEGA software v7. Gamma-distributed Hasegawa, Kishino, and Yano model was calculated to best fit the data and used to calculate a maximum-likelihood phylogenetic tree using 500 bootstraps. Pairwise between groups genetic distance was calculated in MEGA.

Phylogenetic analysis on the COI gene revealed the existence of three separate genetic clusters. From each cluster, five samples were selected and subjected to further genetic analysis through random amplification of polymorphic DNA (RAPD). The RAPD reactions were conducted using the primer 5 -CAGGCCCTTC-3 [72] and One*Taq*® polymerase and reaction buffer (New England Biolabs) following the thermal cycling programme: initial denaturation at 94 ◦C for 2 min followed by 35 cycles of (1) denaturing at 94 ◦C for 1 min, (2) annealing at 37 ◦C for 1 min, and (3) extension at 68 ◦C for 2 min, followed by a final extension step at 68 ◦C for 10 min. Genomic DNA extracted from the earthworm *Lumbricus rubellus* was used as outgroup. PCR product was run on a 1.5% agarose gel for three hours at 120 V using 1 kb HyperLadder (Bioline, London, UK) as reference. Band patterns

were manually scored in a blind manner. Rooted neighbourhood-joining tree was calculated in R 3.5.0 (www.r-project.org) using the package "ape" [73].

#### *2.12. Gut and Soil 16S Sequencing Metagenomics Bioinformatics*

The prokaryotic community in genomic DNA extracted from soil and gut tissue was determined by PCR amplification and sequencing following the method outlined by Kozich et al. [74]. Briefly, a ~555 bp fragment spanning the V3–V4 region of the 16S-rRNA gene was amplified using the forward primer 5 -CCTACGGGAGGCAGCAG-3 and reverse primer 5 -GGACTACHVGGGTWTCTAAT-3 , each modified with the addition of a sequencing primer, an indexing region, and an Illumina flow-cell adaptor such that each sample was uniquely barcoded. PCR amplification was done using Q5® High-Fidelity DNA Polymerase and reaction buffer (New England Biolabs, Ipswich, MA, USA) using the following programme: initial denaturing at 95 ◦C for 2 min, followed by 30 cycles of (1) denaturing at 95 ◦C for 30 s, (2) annealing at 55 ◦C for 15 s, and (3) extension at 72 ◦C for 40 s, followed by a final extension step at 72 ◦C for 10 min. Gel electrophoresis was used to verify amplification of a single product. PCR product was normalized using SequalPrep™ Normalization Plate Kit (ThemoFisher Scientific) and samples from each normalization plate were pooled. The pooled samples purified using QIAquick Gel Extraction Kit (QIAGEN, Venlo, The Netherlands). Gel extracted libraries were quantified using Qubit dsDNA HS Assay Kit (ThemoFisher Scientifc) and equimolary pooled and diluted to 7 pM. The pooled library was sequenced with 10% PhiX on a MiSeq using MiSeq Reagent Kit v3—600 cycles (Illumina, Inc., San Diego, CA, USA). The Illumina demultiplexed sequences were processes using the DADA2 bioinformatics pipeline [75] to generate an amplicon sequence table from the forward reads. DADA2 settings were maxEE(2), maxN(0), and truncQ(2). Sequences were trimmed to 290 bases. Sequences were dereplicated and the DADA2 core sequence variant inference algorithm was applied. Chimeric sequences were removed using removeBimeraDenovo default settings. Amplicon sequence variants (ASVs) were subjected to taxonomic assignment using assignTaxonomy at default settings and the Silva database [76]. ASVs assigned to mitochondria, chloroplasts, Archaea, Eukaryotes, and ASVs with unknown kingdom or phylum were removed from the dataset. Nucleotide sequence data have been submitted to NCBI and are available under submission number SUB7500125 as part of BioProject number PRJNA610159.

#### *2.13. Quantitative PCR*

Quantitative PCR was used to determine differential levels of mRNA of several earthworm immune genes in gut tissue and coelomic fluid (Table 1). Both tissues were screened for coelomic cytolytic factor (CCF), lysozyme, and lysenin/fetidin. Primer pairs were mapped against reference transcriptomes of both *E. fetida* and *E. andrei* to estimate the binding potential to all known allelic variants. Primer pairs targeting CCF were designed to both *E. fetida* and the *E. andrei* versions of the CCF gene using Primer 3. Amplification efficiency of primer pairs was verified through serial dilution and was between 90% and 110% for all pairs. Amplification of the target fragment was verified by Sanger sequencing of the PCR products. qPCR reactions were conducted using GoTaq® qPCR Master Mix (Promega) in a 20 μL reaction volume using 6.25 ng of cDNA as input. qPCR was performed using a Roche LightCycler® 480II with PCR conditions: initial denaturation at 95 ◦C for 3 min, followed by 40 cycles of (1) denaturation at 95 ◦C for 10 s and (2) annealing and extension at 60 ◦C for 30 s. Melt curve analysis was conducted to verify single PCR product. Changes in gene expression were calculated using the 2−ΔΔCt method [77]. EF1α was used as a reference gene for the normalization of the target immune genes. Log2 fold change was expressed in relation to the negative control (earthworms exposed to control soils in the pre-exposure and PBS in the bacterial challenge).

#### *2.14. Statistical Analysis*

All data analysis was done in R (www.r-project.org). Non-metric dimensional scaling (NMDS), distance-based redundancy analysis (db-rda) (using Bray–Curtis distance matrix), permutational

analysis of variance (Permanova), and calculation of diversity indices were conducted using the R package 'vegan' [78] using datasets rarefied to 4959 reads per sample with removal of samples below this threshold. Differences between treatments in diversity indices and gene expression values were tested using two-way analysis of variance (2w-ANOVA) and Tukey's post hoc test. Differential abundance analysis of bacterial taxa was done using Kruskal–Wallis Rank Sum Test and Mann–Whitney test using datasets rarefied to 2448 reads. For the differential abundance analysis, rarefication to this lower read number was done to prevent losing replicates and, therefore, statistical power.


**Table 1.** Details of primers targeting housekeeping gene (EF1α) and target immune system genes.

<sup>a</sup> from Dvorak et al. [43].

#### **3. Results**

#### *3.1. Earthworm Population Genotypes*

Sequencing of the COI gene from earthworms sampled during the bacterial challenge indicated three separate genetic clusters (Figure 2A). Local alignment using NCBI BLASTn indicated that one of those clusters was most similar to *E. andrei* COI, while the COI of the two other clusters aligned best with *E. fetida* COI. The two *E. fetida* COI sub-clusters did not group together in the phylogenetic tree. However, the overall genetic distance between the two *E. fetida* COI clusters was smaller (0.160) than that between the *E. andrei* COI cluster and *E. fetida* COI cluster 1 (0.186) (Figure 2B). RAPD analysis based on 26 polymorphic markers indicated the existence of two genetic clusters (Figure 2C). One of these clusters consisted of individuals carrying an *E. andrei* COI gene copy. The other cluster comprised individuals carrying an *E. fetida* COI copy and one COI assigned *E. andrei* individual.

**Figure 2.** Phylogenetic trees showing relationship between cytochrome c oxidase I (COI) cluster and genetic distances between COI clusters. (**A**) Maximum-likelihood phylogenetic tree of showing phylogenetic relation between the three COI clusters. Samples are collapsed to COI cluster level, see Figure S2 for full tree. Bootstrap values are derived using 500 bootstraps. (**B**) Table with genetic distances between the three COI clusters. (**C**) Rooted neighbourhood-joining tree based on random amplification of polymorphic DNA (RAPD) profiles with *L. rubellus* (*Lr*) as outgroup. Different number indicate sample number. Colours indicate COI grouping.

#### *3.2. The E*ff*ects of Pre-Treatment Exposure and Bacterial Challenge on the Gut Microbiome*

The earthworm gut community at the start of the pre-treatment was composed of a consortium of bacteria comparable to that found in previous studies [56,79]. The gut community was dominated by *Verminephrobacter*(*Proteobacteria*), '*Candidatus* Lumbricincola' (*Mollicutes*), a member of the *Spirochaetaceae* family (*Spirochaetes*), and multiple ASVs belonging to the genus *Aeromonas* (*Proteobacteria*) (Figure S3). Transfer of earthworms from culture soil ('pre-treatment day 0') to LUFA control soils ('pre-treatment day 28') did not significantly alter the total community structure in the gut (Permanova: *F*(1,8) = 1.059, *p* = 0.356) (Figure S3) nor Shannon diversity (*F*(1,8) = 0.882, *p* = 0.375) or species richness (*F*(1,8) = 0.074, *p* = 0.375) (Table S1). Average Shannon diversity and richness across all replicates were 3.0 (±SD 1.1) and 239 (±SD 146), respectively.

At the end of the pre-treatment exposure (i.e., 'pre-treatment day 28'), there were no significant differences in overall community structure between the treatment groups (Permanova: *F*(2,12) = 1.252, *p* = 0.256) (Figure 3A). Bacterial diversity was also not affected by pre-treatment exposure (Shannon: (*F*(2,12) = 0.176, *p* = 0.84; richness: *F*(2,12) = 1.695, *p* = 0.225) and was on average 2.5 (±0.9) (Shannon) and 189 (±114) (richness) (Table S1). Among the most abundant gut bacteria (ASVs with a relative abundance >1% in any of the samples), three ASVs were significantly negatively affected by exposure to both copper forms (Kruskal–Wallis: *p* < 0.05). These ASVs included '*Candidatus* Lumbricinola' (ASV 4876) and two ASVs belonging to *Luteolibacter* (ASV 4960 and 4963) (Figure 3C–E). These taxa together comprised an average of 8% of the total community in controls, but were at or below the limit of detection in the copper-treated earthworms.

**Figure 3.** Bacterial community composition and structure of earthworm gut samples at the end of the pre-treatment exposure (i.e., 'pre-treatment day 28'). (**A**) Plot of non-metric dimensional scaling (NMDS) showing ordination of samples at amplicon sequence variant (ASV) level. (**B**) Relative abundance of dominant ASV at class level per sample. All ASVs with a relative abundance <1% of the total community are grouped under "Other". Mean relative abundance (±se) per treatment as percentage of total community of (**C**) '*Candidatus* Lumbricincola' (ASV 4876), (**D**) *Luteolibacter pohnpeiensis* (ASV 4960), and (**E**) *Luetolibacter* (ASV 4963).

In earthworms that were sampled at 'bacterial challenge day 2', treatment (i.e., pre-treatment exposure followed and bacterial challenge combined) had a nearly statistically significant effect on the bacterial community composition in earthworms, with treatment explaining 24% of the total variance (Permanova: *F*(5,23) = 1.423, *p* = 0.075) (Figure S4). Pre-treatment exposure and bacterial challenge treatment alone explained 10% (Permanova: *F*(2,26) = 1.453, *p* = 0.108) and 4% of the total variance (Permanova: *F*(2,26) = 1.222, *p* = 0.265), respectively (Table 2). No significant effect of treatment on diversity indices, which averaged 1.95 (±0.6) (Shannon) and 57 (±22) (richness) across all samples, was found (Table S1). Among the most dominant gut bacteria (ASVs with relative abundance >1%), '*Candidatus* Lumbricincola' (ASV 4876) relative abundance was significantly negatively affected by copper treatment (Figure S4D) (Kruskal–Wallis: *X*2(5) = 12.1, *p* < 0.05), while *Aeromonas* (ASV 10149) was positively affected (Figure S4D) (Kruskal–Wallis: *X*2(5) = 14.0, *p* < 0.05).

**Table 2.** Outcomes of models testing the relationship between bacterial community composition in the gut of earthworms sampled at 'bacterial challenge day 2' and different combinations of explanatory variables. Db-rda: distance based redundancy analysis using Bray–Curtis distance matrix and applying square root transformation and Wisconsin double standardization. Permanova: Permutational multivariate analysis of variance using Bray–Curtis distance matrix and 999 permutations. 'Explained' and 'Unexplained' represent in db-rda models the proportion of inertia either explained or unexplained by explanatory variables. In brackets (in the 'Explained' column), the fraction of the inertia that is conditioned is shown (i.e., removed). In Permanova models, 'Explained' and 'Unexplained' refer to the model R2 and residual R2 values.


\* indicates *p* < 005, \*\* indicates *p* < 0.01.

#### *3.3. Relation between Earthworm Genotype and Bacterial Community Structure*

Permanova and db-rda indicated that genotype was a better predictor for the bacterial community composition than either pre-treatment exposure or bacterial challenge treatment (Table 2) with samples clustering primarily by COI genotype (Figure 4). After removal of the variation associated to COI genotype using partial db-rda models, the effect of pre-treatment and bacterial challenge treatment on community composition was still not significant (Table 2).

**Figure 4.** (**A**) NMDS plot showing ordination of 'bacterial challenge day 2' samples. (**B**) First two-axis of distance based redundancy analysis (db-rda) including 'pre-treatment exposure', 'bacterial challenge treatment', and 'genotype' as explanatory variables. Percentage following axis labels in (**B**) indicate percentage of total inertia explained by the respective axis. In both figures, different colours indicate different COI genotypes, while different shapes and filling indicate different treatments. Ellipses indicate 90% confidence interval (CI) of the respective COI cluster. In the legend, the text before the vertical bar indicates 'pre-treatment exposure' and the text following the vertical bar indicates 'bacterial challenge treatment'. 'Ea' (*E. andrei*), 'Ef1' (*E. fetida* 1), and 'Ef2' (*E. fetida* 2) indicate COI genotype.

#### *3.4. Impact of Treatments on Bacillus Subtilis Abundance in Gut Tissue, Earthworm Survival, Tissue Integrity, and Immune Responses*

All control earthworms exposed to the PBS control for four days survived. The survival of earthworms exposed to *B. subtilis* was on average 79% at day four, with no statistically significant effects of pre-treatment on the survival rate observed (*X*2(2) = 0.875, *p* = 0.646) (Figure 5A). Abundance of *Bacillus* in gut tissue from earthworms challenged with *B. subtilis* was significantly higher than in control animals, indicating successful inoculation (Figure 5B). No differences in the abundance of *Bacillus* in gut tissue were observed during the recovery period. Histological analysis indicated a possible effect of pre-treatment exposure with copper (in both forms) on the integrity of the gut epithelium and longitudinal muscle tissue. The average integrity scores in copper treatment were between 0.8 and 1.3 points higher than controls (Table S2). The effects of copper treatment were manifested as the thinning of the gut epithelium tissue lining as well as the thinning of muscle fibres (Figure S5). Gene expression levels were assessed through qPCR analysis, targeting several known earthworm immune genes using EF1α as reference gene (Figure S6). No significant effect of treatments was found on immune gene expression in both tissue types (2w-ANOVA: *p* > 0.05) (Figure 5C–H).

**Figure 5.** (**A**) Survival at day four of the bacterial challenge. (**B**) Boxplots of rarefied read abundance of *Bacillus* per sample point. Annotation at top of the graph indicates bacterial treatment: na (not applicable), − (PBS control), and + (*B. subtilis* treatment). Triple asterisks indicate statistical significance between bacterial challenge control and bacterial challenge treatment with *p* < 0.001. Relative high abundance of *Bacillus* at 'pre-treatment day 0' is driven by the two samples in that sample group (Figure S2). Boxplots of fold change in gene expression of earthworm immune responses in (**C**–**E**) gut tissue and (**F**–**H**) coelomic fluid at 'bacterial challenge day 2'. All gene expression values are log2 fold change values derived through 2−ΔΔCt method. Expression values represent fold changes of treatments compared to control animals (control pre-treatment + control bacterial challenge) and are normalized to the expression of the housekeeping gene EF1α. There were no significant differences in gene expression between groups for all tested genes (α = 0.05). Different colours in all panels indicate different bacterial challenge treatments. CCF, coelomic cytolytic factor.

#### *3.5. Relation between Earthworm Genotype and Gene Expression*

No significant relation between gene expression and COI genotype was found, with the exception of lysozyme expression in coelomic fluid samples. For this gene, expression in the *E. andrei* COI genotype was marginally, but significantly higher than the *E. fetida* group 2 (2w-ANOVA: *F*(2,26) = 4.646, *p* < 0.05; Tukey's post hoc test: *p* < 0.05), with the difference in mean fold change of 1.4 indicating a small magnitude effect (data not shown).

#### **4. Discussion**

Innate immunity provides a first line of defence against invading pathogens. NMs are known to interact with the immune system of organisms and can induce both pro- and anti-inflammatory responses [8,13]. NMs are developed and applied as antimicrobial agents in personal care products and in an agricultural setting. In many animals, microbial symbionts play an important role in host defence. Therefore, when animals are exposed to biocidal NM, disruption of their microbiome can be expected, which, accordingly, may lead to effects on host immunity. It remains unclear, however, whether NM exposure can compromise host immunity through effects on the microbiome when hosts are infected by pathogens. This paper aimed to study the impact of biocidal CuO-NPs and its ionic counterpart on the gut microbial community, host immune responses, and infection susceptibility in an earthworm.

Previous in vitro studies have shown that NPs can be taken up by earthworm immune cells (coelomocytes) [45,80,81], and can alter the expression of earthworm immune markers [45,46], leading to cellular toxicity [82,83]. In this study, in vivo exposure to metal biocides (in both metal salt and NP form) caused changes to microbiome structure, with several bacterial symbionts being negatively affected by exposure to copper. Following these exposures and microbiome changes, survival rates were unaffected when earthworms were challenged with a high dose of the soil bacterium *B. subtilis*. Histological analysis indicated possible tissue damage owing to copper exposure, although this analysis is based on observations for a limited number of samples, and thus further assessment of this response is needed. Overall, we found no evidence for altered infection susceptibility or altered immune gene regulation at biocide concentrations where the gut microbiome is already affected.

The lack of an immune response after NP exposure contrasts with the results of previous studies [45,46]. Hayashi and colleagues, for example, found that in vitro exposure in earthworm immune cells to Ag-NP significantly alters the temporal expression of immune genes [45]. Studies in other invertebrates, such as mussels and sea urchins, have also shown that both in vivo and in vitro exposure to NMs can modulate the immune system of these animals [30,31,84]. NMs that are released into the environment are likely to undergo transformations [85–87]. Uptake by organisms may further modify the shape, size, and form of NMs [88,89]. Therefore, the NMs that earthworm immune systems are exposed to in vivo may be different to the pristine forms that have often been used in in vitro studies. The discrepancy between the impact of NM on earthworm immune reactivity in vivo and those in vitro may thus be linked to the transformations of NMs in soil media and the resulting change in the immuno-reactivity of NMs.

Previous research has shown that the microbiome of animals can be altered by NM exposure. In rodents, for example, exposure to Ag-NP can negatively affect the abundance of *Firmicutes* and *Lactobacillus* and induce histological damage to intestinal tissue [28,29]. In soil invertebrates such as springtails and earthworms, metal NP exposure was also shown to alter intestinal microbiomes [58–60]. Exposure to Ag-NP in the springtail *Folsomia candida*, for example, negatively affects the abundance of *Firmicutes* and *Actinobacteria* in the gut of these soil invertebrates [59]. Although implications on host functioning were not further studied, other research shows that microbiome dysbiosis induced by environmental pollution can be associated to changes in the isotopic composition of springtails, suggesting an impact on nutrient turnover [90]. In this study, we found that the earthworm symbiont '*Candidatus* Lumbricincola' is negatively affected by an exposure to copper forms. '*Candidatus* Lumbricincola' is a bacterium exclusively associated to earthworms and has a possible role in the degradation of polysaccharides [47,91]. However, the implications of the near loss of this symbiont for the health and functioning of earthworms require further investigation.

Contrary to our hypothesis, a two-day exposure to a high bacterial level did not change the expression of known earthworm immune markers. Successful inoculation with the bacterium was confirmed by the mortality data and the abundance of *Bacillus* in gut tissue. The lack of an immune response at concentrations of bacteria at which 21% of the exposed individuals die is thus unexpected. Earthworms can show large variation in the expression of immune genes over time [45,92]. Timing of the expression of components of the immune system can be gene-specific [44], but can also depend on the specific pathogen to which the earthworm is exposed [92]. For example, in *E. andrei*, lysozyme is expressed within several hours of exposure to *Escherichia coli*, but only after 16 h following exposure to *B. subtilis* [38]. Similarly, in *E. andrei,* lysenin/fetidin has been reported to be upregulated after six hours in response to a *Staphylococcus aureus* exposure, but downregulated when exposed to *E. coli* [93]. The methodology adopted in this study was based on studies by Dvorak and colleagues, who showed that exposure to high levels of *B. subtilis* can induce changes in immune regulation of earthworms [43,44]. Discrepancy between measured immune responses in *E. fetida*, as reported in this study, and those measured in previous studies in a related species under similar condition show that earthworm immune responses are also species-dependent. Investigations into the molecular structure of the earthworm immune gene CCF in eight different species have shown that some earthworm species have a wider recognition capacity than others [37]. Even within closely related *Eisenia* spp., there are differences in the reactivity of immune genes and immunity related enzymatic activity [43]. These differences in immune reactivity between related earthworms may reflect differences in microbial environments, which may require niche-specific immune responses and lead to differing basal immune reactivities. Time as a factor in earthworm immune responses is thus not fully understood [33] and requires further species-specific investigation. Sufficient sampling over a time-course is needed to fully elucidate patterns of immune expression in earthworms under various environmental stressors and, in particular, to identify the specific points of highest upregulation for key genes.

Earthworm coelomocytes are composed of three subpopulations, each with a unique function [94,95] and molecular immune-expression profile [93–96]. This cellular complexity means that it is possible for different cell subpopulations to have different sensitivities to pollutant exposure [82,97]. Exposure to metals and xenobiotics, but also immunostimulants like LPS [98], has been demonstrated to change the ratio between the different coelomocyte cell subpopulations [82,97,99] and to alter the expression of earthworm immune markers [62]. In this study, coelomic fluid was extruded and sampled without separation of different cell subpopulations. Accordingly, the measured immune responses to *B. subtilis* exposure are an average of the expression levels of these genes across these different subpopulations. This, in combination with high variation between individuals in expression of some of the tested genes as previously reported [100], may limit the ability to elucidate differences in the patterns of gene expression within any individual cell subpopulation [82].

In this study, we found that 7 out of the 30 genotyped earthworms carried an *E. andrei* COI copy. Moreover, for the remaining *E. fetida* individuals, two COI clades were recorded. COI genotype, however, did not affect measured immune responses. The finding of clade structure for earthworms from the *E. fetida*/*E. andrei* complex is in agreement with previous studies [70,101–103]. *E. fetida* are phenotypically characterized by their stripped pigmentation pattern, whereas *E. andrei* are classically more uniformly red coloured. These two species were formally described as two different subspecies (e.g., *E. fetida fetida* and *E. fetida andrei*) [104], but, on the basis of crossbreeding experiments and differences in biochemical markers, were classified as separate species [105]. More recent research has shown that, in laboratory conditions, *E. fetida* and *E. andrei* can hybridize and produce fertile hybrid offspring [101,106]. Field studies also confirm that gene flow between these two species does occur [102]. Earthworms in this study were characterized by typical *E. fetida* pigmentation. Previous studies, however, report that pigmentation is not always a good predictor for COI genotype [101,103]. Here, RAPD profiling suggests that COI genotype does not always predict genomic variability, as indicated by the presence of an individual carrying an *E. andrei* COI copy within a clade consisting of *E. fetida* COI carrying individuals. The COI genotype was shown to be a better predictor for the bacterial community composition than any treatment. Host genetics is one of the components shaping the human gut microbiome [107,108], but similar relationships have also been observed in other animals

such as mice [109] and invertebrates. In the water flea *Daphnia manga*, for example, host genotype shapes not only the structure, but also the functionality of the gut microbiome, in particular its ability to respond to toxic cyanobacteria [110]. The gut microbiome of *Eisenia* spp. is dominated by a consortium of bacteria that are vertically transmitted from parental animal to offspring [79,111]. The relation between COI genotype and gut microbiome structure may thus be linked to the concurrent maternal transmission of both mitochondria and bacterial symbionts.

#### **5. Conclusions**

We show that the microbiome of earthworms can change when exposed to a copper (in both NP and salt form). However, these biocide-mediated changes of the microbiome do not lead to altered susceptibility to a bacterial infection. Despite mortality when challenged with a bacterium, no effects of treatment on the measured earthworm immune markers were observed. The absence of an effect on immune function needs to be further validated by studies of gene expression using a greater time resolution of immune responses in earthworms and further identification of markers of immunity through, for example, full transcriptomic analysis. The methodological approach applied in this paper may guide future studies to improve the assessment of immuno-safety of NMs.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/10/7/1337/s1, Figure S1: Growth over time of (**A**) *Bacillus subtilis*, (**B**) *Bacillus thuringiensis*, and (**C**) *Citrobacter rodentium* in LB medium (i.e., 'Control'), filter sterilized culture medium consiting out of 800 μL LB medium and 200 μL 10x diluted coelomic fluid ('Coelomic fluid') or culture medium containing both diluted coelomic fluid and antibiotics ('Antibiotics'); Figure S2: Maximum-likelihood phylogenetic tree of showing phylogenetic relation between all samples; Figure S3: Bacterial community composition and structure of earthworm gut samples from control replicates at start and end of pre-treatment; Figure S4: Bacterial community composition and structure of earthworm gut samples during bacterial challenge (i.e., 'bacterial challenge day 2'); Figure S5: Examples of cross sections of *E. fetida* stained with hematoxylin/eosin representing a range of tissue integrity scores; Figure S6: Boxplots of Cp values of elongation factor 1 alpha (ef1α) per treatment in coelomic fluid and gut tissue; Table S1: Mean Shannon diversity index and bacterial richness (±SD) per sampling point and treatment group; Table S2: Integrity scores of gut epithelium and chloragogen tissue per sample point and treatment. Values are means of the scores of three sections from a single replicate.

**Author Contributions:** Conceptualization, E.S., D.S., P.P., P.K., and C.S.; Methodology, E.S., D.S., P.P., P.K., and C.S.; Formal analysis, E.S. and J.D.; Investigation, E.S., J.D., S.H., and P.P.; Data curation, E.S.; Data interpretation: E.S., J.D., P.K., D.S., C.S., and P.P.; Resources: P.P., T.G., S.H., and P.K.; Writing—Original draft preparation, E.S.; Writing—Review and editing, J.D., D.S., T.G., and P.P.; Visualization, E.S.; Supervision, D.S., P.K., C.S., and P.P.; Project administration, E.S., D.S., C.S., and P.P.; Funding acquisition, D.S., C.S., P.K., and P.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 671881. The contributions by David Spurgeon and Claus Svendsen were supported by the UK Natural Environment Research Council (NERC) UK Centre for Ecology & Hydrology Institute Funding award.

**Acknowledgments:** The authors would like to thank Denny Rigby (UK CEH), Natividad Isabel Navarro Pacheco (IMIC), and Frantisek Skanta (IMIC) for their help.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
