**About the Editors**

#### **Patrick C. Baer**

Patrick C. Baer is a cell biologist and Associate Professor of Experimental Medicine at the Hospital of the Goethe University in Frankfurt/M. He completed his studies in biochemistry at the Technical University of Darmstadt and received his doctorate at the Goethe University of Frankfurt/M. P.C.B. has currently published over 100 research articles, including three book chapters and three patents. P.C.B. has been working with cell culture models of proximal and distal tubular epithelial cells of the kidney for more than 25 years. The research areas of P.C.B. also focus on the isolation, culture and differentiation of mesenchymal stromal/stem cells (MSCs) and the transplantation of MSCs or their derivatives (conditioned medium, extracellular vesicles) to improve organ regeneration.

#### **Ralf Schubert**

Ralf Schubert Ph.D., is an Associate Professor with tenure at the Department of Pediatrics at the University Hospital Frankfurt, Goethe University and is the Head of the Pneumological/ Immunological Laboratory. He graduated from the Technische Universitat Darmstadt with a Ph.D. ¨ degree in Immunology and recieved his postdoctral training at the Medical School at the University of California, San Diego. He has been working in the field of immunology for more than 20 years and the research activity of his group is focused on the investigation of cellular and molecular mechanisms of inflammatory lung diseases such as bronchial asthma. There work is especially concerned with rare diseases such as Ataxia telangiectasia, and bronchiolitis obliterans. He has authored more than 180 scientific publications in international journals on this topic and has received numerous national and international grants. Current projects focus on the regulatory role of microRNAs in the resolution of inflammation in in vivo and in vitro models.

### *Editorial* **In Vitro Models of Tissue and Organ Regeneration**

**Patrick C. Baer \* and Ralf Schubert \***

Division of Allergology, Pneumology and Cystic Fibrosis, Department for Children and Adolescents, University Hospital, Goethe-University, 60596 Frankfurt/M., Germany

**\*** Correspondence: p.baer@em.uni-frankfurt.de or pcbaer@arcor.de (P.C.B.); ralf.schubert@kgu.de (R.S.);

Tel.: +49-69-6301-83611 (R.S.); Fax: +49-69-6301-83349 (R.S.)

The recovery of cells after tissue and organ injury is a complex process. To understand the underlying molecular biological mechanisms, more detailed insights into the cellular processes of repair and regeneration are urgently needed. Based on this knowledge, this Special Issue focuses on current in vitro systems exploring repair and regeneration mechanisms. Experimental research approaches to investigate the mechanisms involved and laboratory methods to establish and optimise models for tissue and organ repair and regeneration, as well as theoretical modelling and computational models, but also review papers are included here. Eleven articles are published in the Special Issue, which deals with various tissue and organ regeneration questions or the modelling or summary of the research models used in this process.

Shyam et al. comprehensively summarise various methods involved in developing 3D cell culture systems, emphasising the differences between 2D and 3D systems and methods involved in recapitulating the organ-specific 3D microenvironment [1]. They also discuss the latest developments in 3D tissue model fabrication techniques, microfluidicsbased organ-on-a-chip, and imaging as a characterisation technique for 3D tissue models. Lieto et al. summarise current research to accurately evaluate ocular toxicity and drug effectiveness [2]. The recent achievements in tissue engineering of in vitro 2D, 2.5D, 3D, organoid and organ-on-chip ocular models and in vivo and ex vivo ocular models were discussed in terms of their advantages and limitations. Another review by Young et al. looks in detail at different strategies for fatty liver treatment in non-alcoholic fatty liver disease [3]. They discuss various defatting strategies, including in vitro use of pharmacologic agents, machine perfusion of extracted livers, and genomic approaches targeting specific proteins. Another work by Hentabli et al. deals with modelling a neural network for bioactivity prediction [4]. This paper describes a novel technique based on a deep learning convolutional neural network for predicting chemical compounds' bioactivity. The authors explain the importance of this work by stating that determining and modelling the possible behaviour and effects of molecules requires the study of the basic structural features and physicochemical properties that determine their behaviour in chemical, physical, biological and environmental processes.

Two original works use in vitro models of mesenchymal stromal/stem cells (MSCs) to investigate regenerative purposes. Barbon and coworkers use an in vitro conditioning regimen of MSCs towards the endothelial lineage to stimulate coagulation factor VIII production [5]. The background of this work was the development of a cell therapy for the treatment of Haemophilia A and, therefore, for future pre-clinical investigation using preconditioned MSCs. Leppik and coworkers demonstrate a new perspective on bone tissue engineering [6]. The work shows that MSCs survive cryopreservation on scaffolds and, after thawing, could be released as ready-to-use products for permanent implantation during surgery.

In addition, two other in vitro studies use epithelial cell systems to investigate their differentiation or their involvement in inflammatory processes. Primary alveolar epithelial cells' main limitation is the difficulty of maintaining the type II phenotype in culture.

**Citation:** Baer, P.C.; Schubert, R. In Vitro Models of Tissue and Organ Regeneration. *Int. J. Mol. Sci.* **2023**, *24*, 14592. https://doi.org/10.3390/ ijms241914592

Received: 7 September 2023 Revised: 13 September 2023 Accepted: 22 September 2023 Published: 26 September 2023

**Copyright:** © 2023 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/).

Marhuenda and coworkers show that culturing primary alveolar epithelial cells on lung extracellular matrix-derived hydrogels facilitated the prolonged culturing of these cells and enhanced the preservation of the type II phenotype [7]. Baer and coworkers characterised the mRNA expression of renal proximal tubular epithelial cells and the cargo in extracellular vesicles in an inflammatory microenvironment [8]. This study demonstrates the altered miRNA expression of epithelial cells and their released vesicles during induced inflammation, with only three miRNAs overlapping between cells and vesicles. The background to this study is that understanding the precise molecular and cellular mechanisms that lead to inflammation is the most important way to identify targets for the prevention or treatment of inflammation.

Steyn-Ross and coworkers describe the ex vivo quantification of tissue oxygen consumption by measuring oxygen partial pressure as a function of probe depth using thin slices of cortical brain tissue [9]. The authors confirm that a previously published diffusionconsumption model provides an excellent description of the oxygen-tension distribution in a thin slice of active tissue.

Finally, this Special Issue contains two publications addressing issues using in vivo models. Azam and coworkers investigated the prevention of neuroinflammation in vitro and in vivo using an herbal extract and purified dioscin [10]. The in vitro study demonstrates protection against lipopolysaccharide-activated inflammatory responses in microglial cells. The following in vivo study shows that dioscin upregulates brain-derived neurotrophic factor and cAMP-response element binding protein phosphorylation in the cerebral cortex and hippocampus regions of the mouse brain. The authors conclude that dioscin protects against neurotoxicity. Esteves-Monteiro and coworkers evaluated changes in ileum and colon histomorphometry and Angiotensin II reactivity in a rat model of diabetes mellitus. They showed the structural remodelling of the gut wall with a decreased contractile response to Angiotensin II [11]. They summarise that these findings may help to explain diabetic dysmotility.

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

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

## *Review* **Fabrication and Characterization Techniques of In Vitro 3D Tissue Models**

**Rohin Shyam 1,2, L. Vinod Kumar Reddy <sup>3</sup> and Arunkumar Palaniappan 2,\***


**Abstract:** The culturing of cells in the laboratory under controlled conditions has always been crucial for the advancement of scientific research. Cell-based assays have played an important role in providing simple, fast, accurate, and cost-effective methods in drug discovery, disease modeling, and tissue engineering while mitigating reliance on cost-intensive and ethically challenging animal studies. The techniques involved in culturing cells are critical as results are based on cellular response to drugs, cellular cues, external stimuli, and human physiology. In order to establish in vitro cultures, cells are either isolated from normal or diseased tissue and allowed to grow in two or three dimensions. Two-dimensional (2D) cell culture methods involve the proliferation of cells on flat rigid surfaces resulting in a monolayer culture, while in three-dimensional (3D) cell cultures, the additional dimension provides a more accurate representation of the tissue milieu. In this review, we discuss the various methods involved in the development of 3D cell culture systems emphasizing the differences between 2D and 3D systems and methods involved in the recapitulation of the organ-specific 3D microenvironment. In addition, we discuss the latest developments in 3D tissue model fabrication techniques, microfluidics-based organ-on-a-chip, and imaging as a characterization technique for 3D tissue models.

**Keywords:** in vitro models; 2D and 3D cell cultures; 3D tissue models; 3D bioprinting; confocal microscopy

#### **1. Introduction**

In vitro two-dimensional (2D) cell culture methods are a widely used tool for understanding biological functions such as cellular interaction, mechanisms of disease initiation and progression, production of proteins, cellular biology, and, more recently, the development of engineered tissue mimics. In a 2D environment, cells are grown as a monolayer over a flat plastic surface, where they adhere and spread. However, the simplicity of this model makes the depiction and simulation of complex tissue structures challenging. Two-dimensional monolayer cultures have been used for decades to study the cellular responses to biochemical and biophysical cues. These systems do not always mimic human physiological conditions despite providing significant advancements in the understanding of cellular behavior [1], thereby resulting in non-predictive results.

In recent years, the paradigm has shifted towards three-dimensional (3D) cell cultures. Increasing research-based evidence suggests that 3D tissue models are a better option for mimicking complex tissue or organ architecture (cell–cell and cell–matrix interactions) and physiology [2]. These models are gaining importance from basic research to advanced application-based research such as drug testing/screening and other translational purposes. In human tissue, cells are encapsulated within extracellular matrix (ECM) proteins in a 3D environment [3]. The ECM function under defined biophysical and biochemical signals, which regulate cellular functions such as proliferation, adhesion, migration, differentiation,

**Citation:** Shyam, R.; Reddy, L.V.K.; Palaniappan, A. Fabrication and Characterization Techniques of In Vitro 3D Tissue Models. *Int. J. Mol. Sci.* **2023**, *24*, 1912. https://doi.org/ 10.3390/ijms24031912

Academic Editors: Patrick C. Baer and Ralf Schubert

Received: 27 November 2022 Revised: 30 December 2022 Accepted: 1 January 2023 Published: 18 January 2023

**Copyright:** © 2023 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 morphogenesis and maintain homeostasis [4]. Hence, different 3D models are evolving with the combination of cells and proteins to recapitulate native organs and the cellular microenvironment. This aids in understanding the various organs and tissue functions under a controlled laboratory setting and offer the possibility to generate organ-specific and personalized drug testing platforms [5].

Recent advances in microfabrication techniques and tissue engineering technology have influenced the development of complex culture systems and biomimetic microfluidic platforms to capture the structural and functional complexity of the native physiological environment. Tissue engineering is a subfield of regenerative medicine that aims to repair, replace, or regenerate tissues or organs. This is achieved through the translation of fundamental principles of physics, chemistry, and biology combined with the principles of materials engineering and cell transplantation. The goal of this approach is to mimic native tissues that can function as medical devices with therapeutic benefits to regenerate damaged tissues, function as a platform to study drug cytotoxicity at a cellular and molecular level, and model disease under laboratory conditions [6,7]. With these advancements, 3D models with ECM-mimicking proteins could recapitulate the microarchitecture and functional cellular environment of the native organ. In recent years, organ-on-the-chip technology has been gaining prominence due to its ability to simulate organ-level physiology by recreating the multicellular connections and interfaces, vascular perfusion, mechanical cues, and chemical gradient under highly controlled environments.

The bioengineering and designing of complex biomimetic tissue for model systems involve considering several design characteristics and parameters. A 3D tissue model system can be generated through the fabrication of spheroids and organoids; however, while being able to provide a 3D microenvironment, a critical challenge with these systems is the lack of vasculature, which is essential in providing oxygen and nutrients while removing metabolic waste from cells. Alternatively, a scaffold that mimics the ECM is generated via techniques such as 3D bioprinting, electrospinning, and solvent casting/particulate leaching (SCPL) to create porous structures that house the cells, growth factors, vasculature, and transcription factors. The choice of biomaterial to generate the ECM is critical. There are a variety of natural and synthetic biomaterials available, with each having its own benefits and limitations. There has been an increased interest in the combination of biomaterials to generate hybrid biomaterials, which enhance the structural and biological properties of biomaterials. Another consideration is the choice of cells, which is dependent on the tissue being modeled. Stem cells are state-of-the-art in tissue engineering due to their differentiation potential into any cell lineage. Figure 1 summarizes the essential considerations in the realm of 3D tissue models. In the following sections, we discuss the types of 3D tissue models, types of biomaterials and their key characteristics, techniques used in the mimicking of tissue architectures and generation of porous scaffold structures, the types of cells used in 3D models, and their advantages and disadvantages, concluding with the imaging modalities of tissue architecture.

It is often challenging to directly study and observe the complex mechanisms of human development and disease due to a lack of experimental accessibility to biological processes. As a result, the use of model systems that recapitulate these functions ex vivo has been of primary interest to researchers. Two-dimensional culture systems have been established as standard protocols to observe cellular behavior; however, these systems do not completely recapitulate the cellular microenvironment. For instance, the epithelia of the small intestine is an active and rapidly renewing tissue that can undergo tissue widening and form compact folds, invaginations, evaginations, and wavy morphologies [8]. Similarly, in the cardiovascular system, myocardial fibrillar proteins form a 3D complex structure that changes orientation during systole and diastole, resulting in the cardiac tissue undergoing cyclic stress, torsion, and compression [9,10]. These microenvironments represent a major challenge to replicate in vitro. The replication of tissue-specific conditions within a 3D model can offer the ability to study these complex mechanisms and enable a deeper understanding of the role in human development and disease progression and

as a platform for drug testing. The key differences between 2D and 3D cell cultures are tabulated in Table 1. There are several strategies to fabricate and characterize 3D tissue models. In this review, we have explored the various types of 3D model strategies involved in the fabrication of complex models. Importantly, we have also explained unique imaging techniques involved in the characterization of 3D tissue models.

**Figure 1.** Schematic representation of considerations in the fabrication of 3D tissue models.


**Table 1.** Key differences between 2D and 3D cell cultures for modeling in vivo conditions.


#### **Table 1.** *Cont.*

#### **2. Types of 3D Tissue Models**

*2.1. Anchorage Independent (Non-Scaffold Based) 3D Tissue Models*

2.1.1. Spheroids

Spheroids are perfectly spherical cellular aggregates in suspension generated from primary cell types and cell lines. The term was coined in 1970 by Sutherland et al. when the group dissociated Chinese Hamster V79 lung cells which formed spherical aggregates [21]. There are various techniques involved in the fabrication of spheroid, including the hanging drop technique, microwell hanging drop technique, liquid overlay technique, microwell array from micropatterned agarose wells, rotating wall vessel, and magnetic levitation (Figure 2A(i–vi)) [22,23]. Microfluidic technology and 3D bioprinting have also been utilized in the generation of spheroids [24–26]. The most common applications of spheroids are in creating tumor models, stem cell research, tissue engineering, and transplantation therapy. The key advantages of using this method as a 3D tissue model are that it facilitates cell–cell and cell–matrix interactions providing a physiochemical environment similar to in vivo while maintaining intrinsic phenotypic properties and improving the viability and proliferation of cells [23].

**Figure 2.** Common fabrication techniques used for the creation of spheroids, organoids, and cell sheet (**A**) (i) Hanging drop method (ii) Spontaneous spheroid formation (iii) suspension culture (iv) ECM method (v) Magnetic levitation method, (vi) Microfluidic device method. Altered and reproduced with permission from [27] under Open Access CC BY 4.0. MDPI (**B**) Schematic representation of cell sheet engineering.

However, there are several drawbacks to this method. Due to the lack of vasculature within the aggregates, the supply of nutrients to the core of spheroids is limited, and this limitation becomes pronounced with larger spheroid aggregates as it forms a diffusion gradient [23]. Additionally, despite the various techniques utilized in spheroid formation, each has its own unique challenges. For example, the hanging drop method is a simple method to implement and provides uniform spheroid shapes with greater control over spheroid shapes. However, it is tedious to handle and time-consuming, and inefficient due to low throughput. In other spheroid formation techniques, long-term survivability and tedious media exchange are the key challenges [28]. Despite exhibiting a 3D structure, inherently, spheroids lack the complex architecture of tissues in vivo and, therefore, cannot completely recapitulate the physiological environment.

#### 2.1.2. Organoids

Organoids are 3D self-aggregating assemblies containing multiple cell types arranged spatially, such as cells in a tissue, recapitulating cellular and molecular stages in early organ development [29,30]. They have been used as tissue models to explore mechanisms of organ development. Organoids are increasingly being used in medical research, specifically in preclinical studies and in 3D tissue models, to study cellular interactions and drugtoxicology, pharmacology, and microbiology [29]. The 3D architectural and functional similarities to the tissue of origin make organoids an excellent model for studying complex cell–cell interactions and tissue development. The fabrication of organoid models is similar to the processes involved in the generation of spheroids (Figure 2A(i–vi)) [27,31]. However, the key difference is that in organoid formation, pluripotent stem cells and embryonic stem cells are given specific signaling cues that act as instructions to form 3D organoids of a variety of tissues [31]. Organoids have been employed in the generation of optical cups, liver, brain, lung, and heart [32–36]. They have also been used to model disease conditions to study disease development and progression. For example, in a recent study by Richards et al., cardiac organoids with oxygen-diffusion gradients were fabricated to model the human heart after myocardial infarction while recapitulating the hallmarks of myocardial infarction [36]. Yang et al. developed a mice 3D testicular organoid using testicular cells from BALB/c mice to investigate Zika-virus-induced mammalian testicular damage [37]. The key challenge of using organoids is the lack of vasculature. Optimization of the conditions for incorporating more than one type of cells to mimic in vivo structure is required [38]. Additionally, the effect of ECM composition and cell–matrix interaction requires further investigation to develop robust model systems. While there have been significant advances to overcome this challenge, research into multi-organ communication requires further investigation.

#### 2.1.3. Cell Sheet Engineering

Cell sheet engineering is a form of tissue engineering methodology that does not require a scaffold. In this method, cells are grown in vitro by placing a single-type cell on a stimuli-sensitive polymer (Figure 2B). In a culture environment suitable for cell growth, cells are grown till a three-dimensional cell sheet is generated. By inducing a stimulus such as heat, the polymer becomes hydrophilic, enabling the detachment of the cell sheet from the polymer base [39]. Cell sheet engineering has applications among various organs such as the heart, cornea, bladder, liver, and bone. The key advantage of using cell sheet engineering is the ability to co-culture cells and generate a vasculature network. For example, Sakaguchi et al. observed that endothelial cells within cell sheets spontaneously form blood vessel networks as in vivo capillaries [40]. Wu et al. investigated the therapeutic benefits of cell sheets derived from umbilical cord mesenchymal stem cells on rat models with induced ischemic heart failure [41]. The authors subjected H9C2 cardiomyocytes under hypoxia conditions and starvation to observe cell apoptosis as a 2D model, and an ischemic model was made by subjecting rats with Left Anterior descending artery (LAD) ligation to induce ischemic conditions [41]. The study observed that the cell sheets improved cell retention in the myocardium affected by ischemic heart failure, improved cardiac function, attenuated cardiac fibrosis, and induced neovascularization [41]. While recent research indicates that cell sheet engineering may pose a viable therapeutic solution, a major drawback of this method is the generation of hypoxic conditions within thicker cell sheets. Additionally, the lack of well-developed vascular networks within the cell sheet at the time of the generation of sheets poses further translational limitations [42].

#### *2.2. Anchorage Dependent (Scaffold Based) 3D Tissue Models*

3D tissue models offer the versatility of generating mini-organs that mimic in vivo physiology of a specific tissue. However, these models do not completely recapitulate the characteristics of the tissue. Spheroids and organoids have major drawbacks, such as poor mechanical strength and closed 3D geometry. This results in decreased oxygen and nutrients delivery to the center and hampers the use of conventional assays and instrumentation for screening studies such as nutrient and oxygen transport, absorption kinetics of drugs, and cell–cell interactions [43,44]. The paradigm of tissue engineering involves the conglomeration of living cells within bioartificial support to generate a 3D living structure with mechanical, structural, and functional properties equivalent to human tissue [45]. While the generation of artificial constructs is primarily for regenerative purposes, artificial tissues are being developed to replace reliance on animal models, which are dissimilar to human physiology and do not provide accurate predictions for human tissue responses. The conventional methods, from the perspective of tissue engineering for regenerative purposes, rely on the generation of support structures that act as a temporary scaffold to aid tissue regeneration while gradually degrading and being replaced by autologous tissues [46]. However, from the perspective of modeling, tissue replication should be designed to recapitulate the specific conditions being mimicked. This process is extremely complex due to several factors involved in the mimicking of tissue. Specifically, each tissue exhibits varying features such as porosity, ECM composition, cell phenotypes, and signaling pathways [47]. Ergo, the fundamental elements to consider in the designing of artificial tissue are the material for scaffolds, the cell source, the chemical stimuli, and the method for generating the correct tissue architecture. Additionally, it is pertinent that the choice of material is significantly dependent on the tissue being mimicked as the material will form the ECM, and therefore, the scaffold must meet the specific mechanical, chemical, physical, and biological requirements to achieve cell diffusion, proliferation, viability, and functionality [46]. The key modalities used in the generation of scaffolds for 3D tissue models are Solvent Casting Particulate Leaching (SCPL), Electrospinning, and 3D Bioprinting. Figure 3 provides a schematic representation of the methods, and Table 2 highlights the advantages and disadvantages each method has to offer.

**Figure 3.** Schematic representation of the methods used in the generation of 3D tissue architecture.


**Table 2.** The advantages and disadvantages of fabrication methods used in creating 3D architectures.

#### 2.2.1. Solvent Casting Particulate Leaching (SCPL)

SCPL is a popular technique used in the fabrication of highly porous polymer scaffolds for hard tissues such as bone and teeth. In this method, a salt that is insoluble in the polymer is admixed in a polymer solution followed by an evaporation process to remove the solvent, resulting in a salt-polymer composite. The composite matrix is then submerged in water to leach out the salt resulting in a highly porous structure (Figure 3a) [48]. Through this method, 50–90% porosity is achieved [49]. A key advantage of this method is the relative ease and low cost associated with the fabrication of highly porous and tunable pore size that enables the migration of cells within the scaffolds [50]. Similar processes that are employed in the generation of highly porous structures are freeze-drying [51,52], thermal-induced phase separation (TIPS) [53], and gas foaming [54]. The advantages and disadvantages of this method are covered in Table 2.

#### 2.2.2. Electrospinning

The term is derived from electrostatic spinning and is a method that utilizes a highvoltage electric field to draw charged threads of ultrafine nanometric scale fibers from polymer melts or solutions [55,56]. The technique is complicated and involves a process where a charged droplet of polymer in a liquid phase under high voltage results in an electrostatic repulsion counteracting surface tension and elongation of the droplet to a critical point of liquid stream eruption termed a Taylor cone [56]. As shown in Figure 3b, a standard electrospinning system consists of a syringe pump, a metallic needle, a high-voltage DC supply, and a grounded collector. In the process of electrospinning, solvents evaporate, and the resulting fibers are solidified to form nonwoven fibrous membranes. Typically, cells suspended in cell culture media are seeded on electrospun mats in tissue culture well plates to cultivate cells within the scaffold [57]. Recently, there have been advances in incorporating cells within the polymer solution as a cell-laden bioink to generate cell-laden fibrous structures [39]. This technique was first introduced by Townsend-Nicholson et al., who used a coaxial system to encapsulate cells in a bio-suspension within an outer core of PDMS [40]. The key material and process parameters that need to be considered in the generation of either electrospinning or cell-electrospinning are viscosity, applied electric field, feed rate, and the distance between the nozzle and collector plate, along with environmental factors such as room temperature, relative humidity [39]. Table 2 summarizes the advantages and disadvantages of using such a system.

#### 2.2.3. Bioprinting

3D bioprinting is the layer-by-layer deposition of cell-laden biomaterials in 3D space based on a predetermined geometry. Complex geometries and shapes are designed through computer-aided design (CAD) software or geometries extracted from medical images. The main modalities of 3D bioprinting are based on the delivery system of the cell-laden biomaterials termed bio-inks and include extrusion-based (extrusion can be achieved via pneumatic, piston, or screw), inkjet (thermal or piezoelectric), and laser-assisted [58] (Figure 4). In a typical extrusion-based 3D bioprinting system, bioink is extruded via a needle, and based on the pattern generated in a CAD file, a 3D structure in a bottom-up approach is generated (Figure 4A). Three-dimensional bioprinting is a rapidly evolving technology employed to print a variety of tissue structures of various organs, and the frontier of 3D bioprinting is the printing of a complete artificial whole organ, which was most recently achieved by Mirdamadi et al. [59] using a novel technique termed Freeform Reversible Embedding of Suspended Hydrogels (FRESH). In the study, the authors modified an extrusion-based bioprinter and embedded alginate in a support bath comprised of gelatin microparticles suspended in a calcium chloride solution [59]. The core principle is that the gelatin microparticles act as a support bath with multiple crosslinking strategies to gel the different types of hydrogels while providing support for embedded hydrogels that would normally collapse in conventional additive manufacturing processes as they are being printed (Figure 4F) [60]. Senior et al. modified the FRESH bioprinting approach to generate stable hydrogels with low viscosity, termed Suspended Layer Additive Manufacturing (SLAM) [61]. In their study, bioinks with low viscosity in the liquid phase prior to gelation were extruded in an agarose gel that exhibited shear thinning property as the material was extruded and regained its structure upon removal of the shear force entrapping the suspended hydrogel [61]. A crosslinker was then allowed to diffuse through the agarose fluid gel, which resulted in the hydrogel forming stable structures and could be easily removed from the fluid gel [61] (Figure 4G). While microgel support baths have been used to demonstrate full organ printing [59], a shift from this paradigm is the bioprinting of a sacrificial bioink within a slurry-support bath comprised of cellular spheroids in a technique termed sacrificial writing into functional tissue (SWIFT) (Figure 5). Skylar-Scott et al. reported the use of this technique to generate a living matrix primarily composed of tightly compacted tissue-specific organ building blocks from iPSC-derived embryoid bodies, multicellular spheroids, or organoids [62]. Within the living matrix, a sacrificial ink is patterned and embedded via 3D printing, which, when removed, yields perusable branching channels and conduits, thereby resembling vascularized networks (Figure 5) [62,63]. While Support bath systems with extrusion-based bioprinting could be an effective platform in the fabrication of microtissues, however, controlling the position within the 3D space is a challenge [63]. Novel methods to circumvent these challenges are emerging within the scientific community and have been covered elsewhere [63].

**Figure 4.** Modalities of 3D Bioprinting (**A**) Extrusion based printing (**B**) Inkjet Printing (**C**) Laser Induced Printing (**D**) Kanzen Spheroid and needle array (**E**) Stereolithography (**F**) FRESH 3D printing method (**G**) SLAM 3D Printing. (**D**) reproduced under the terms and conditions of the Creative Commons CC BY 4.0 License [64] Copyright 2017, The Authors. Published by Springer-Nature Publishing. G reproduced under the terms and conditions of the Creative Commons CC BY 4.0 License [61] Copyright 2019, The Authors. Published by Advanced functional materials.

**Figure 5.** Schematic representation of the SWIFT process. Copyright © 2023 The Authors [62], some rights reserved, exclusive licensee American Association for the Advancement of Science. Distributed under a Creative Commons Attribution Non-Commercial License 4.0.

3D bioprinting offers versatility in controlling essential parameters such as bioink composition, printing speed, needle gauge, extrusion pressure, and scaffold geometry. However, despite the plethora of biomaterials available for this technique (Table 3), each biomaterial has unique properties that must be optimized to generate suitable constructs. A major benefit of this method is the ability to generate a complex vasculature network via bioinks laden with endothelial cells, as shown in recent research by Noor et al. [65]. Despite the large library of biomaterials that can be used, not all materials have gelling properties required to hold the shape fidelity of the final printed structure and need to be modified to enhance mechanical strength along with chemical, physical, and biological properties. In certain circumstances, bioinks are stabilized through post-processing crosslinking mechanisms via photon activation through UV light in the presence of a photoinitiator or via ionic crosslinking in the presence of divalent cations. While 3D bioprinting offers vast opportunities, it is severely limited by the availability of printers capable of printing whole organs [66]. Additionally, further research into improving the print resolution of the printed construct and encapsulation of cell densities from a clinical translation outlook remains a challenge [67].

#### 2.2.4. Organ-on-a-Chip

The process of developing novel drugs and medical interventions requires the use of in vitro modeling, followed by animal studies, to test the safety and efficacy of newly developed drugs before testing on humans. However, animal models do not provide accurate predictions for human responses. Clinical trials are time-consuming and not cost-effective in the long run. Most novel drugs fail in clinical trials, and therefore, there is a need to develop a system or model that mimics human physiology, remains cost-efficient, and has the capability to provide accurate data. In contrast to biological approaches to generate 3D tissue models, organ-on-a-chip (OOC) systems are used to recapitulate tissue and organ structure by leveraging microfluidic physics along with microfabrication engineering techniques and biomaterials to create micro-physiological systems that model tissue structure and disease conditions. Research into the development of microfluidic channels to study signal pathways, drug responses, and tissue functions is ongoing [68]. For example, Zhao et al. employed OOC to create a platform to generate chamber-specific cardiac tissue and disease modeling to measure contractile force in ventricles and atriums and their response in the presence of drugs [69]. Similarly, Parsa et al. developed a platform to study mechanisms of cardiac hypertrophy with low cell volume [70].

There is a wide range of organ systems that have been modeled on an OOC platform, including the heart [71], kidney [72], brain [73], lung [74], intestine [75], liver [76], and eyes [77]. Additionally, OOC has been employed to study tissue-specific diseases. Costa et al. reported the use of microfluidics to mimic arterial thrombosis in vitro [78]. The study was designed to replicate a three-dimensional architecture of coronary arteries under healthy and stenotic conditions by modeling healthy and stenotic arteries to create a microfluidic chip with inlets and outlets to allow perfusion through the system [78]. This enabled the authors to study the effect of shear rates within arteries and enable a better understanding of arterial thrombosis [78]. Microfluidic technology has also been leveraged as a tool to generate spheroids and organoids [26], study drug pharmacokinetics, and the generation of micro bioreactors where 3D bio-printed tissue constructs can receive oxygen and nutrients under laminar flow conditions. While most microfluidic systems use a design-based approach and leverage fluid behavior on a microscale, the lack of ECM or an in-vivo-like microenvironment is a drawback in OOC. OOC technology is based on the use of soft lithography to generate molds of microchannels with the use of polydimethylsiloxane (PDMS) as a substrate material. The high resolution offered by stereolithography and the ability to miniaturize the microenvironment enables researchers to study complex diseases and their behavior in a heterogeneous environment. In designing and production of OOC, the selection of cells and biomaterials must be given extensive consideration. In order to improve the relevance of OOC, it is crucial to include vascular networks that can provide efficient nutrient and oxygen diffusion across the tissue or microfluidic channels. There has been a focus on the incorporation of scaffolds or hydrogels into microfluidic systems to overcome this drawback. The presence of an ECM-like matrix to house cells provides both biophysical and chemical cues that aid in the development of a more in-vivo-like microenvironment. Figure 6 provides a schematic representation of the microfluidic system integrated with hydrogels to generate 3D in vitro models to study disease. For example, Shang et al. used 3D bioprinting to generate biomimetic hollow blood capillaries [79]. The authors created microchannels using 3D printing and injected a composite of GelMA and Alginate incorporated with human umbilical cord endothelial cells (HUVECS), the hydrogel being crosslinked with either barium or calcium chloride Figure 7i and studied the proliferation of cells in the hollow chamber [79]. Hong et al. used 3D bioprinting to fabricate cancer spheroids for evaluating the drug resistance of cancer cells [80]. In their study to evaluate the efficacy of drug resistance of cancer cells, the authors printed 3D miniwells using poly (lactic acid) in a grid structure. The authors then embedded drug-resistant MCF-7 breast cancer cells in a gelatin–alginate hydrogel bioink and 3D bioprinted into the mini-wells to encourage single spheroid formation [80]. The use of hydrogels encapsulated within microfluidic devices to provide a more comprehensive in-vivo-like environment could change the research field towards lesser reliance on animal models.

**Figure 6.** Microfluidic system integrated with hydrogels and cells to provide in-vivo-like 3D microenvironment with biochemical and biophysical cues that result in enhanced differentiation of stem cells or reprogrammed cells, generate functionally mature tissue specific cells and enable a structurally organized microenvironment.

**Figure 7.** Fabrication of microfluidic chip to generate an in vitro model to simulate hollow biomimetic capillary using 3D bioprinting and hydrogel. (**i**) fabrication of templates and microfluidic device (**ii**) simulating hollow blood capillary. Comparison of templated printed using 3D printing, with (**iii**) fused deposition model (FDM) and (**iv**) using hydrogels and extrusion-based bioprinting. (**v**) fabrication and characterization using fluorescence microscopy of diverse hollow structures (**vi**) barrier function of hollow hydrogel microfiber with cells. Image reprinted with permission from [79] Copyright © 2023 by the authors ACS Biomaterials Science and Engineering.

Despite the significant advantages this technology has to offer, there are several challenges that need to be addressed. For example, PDMS is the most common material used as a substrate to build a microfluidic device. However, it is known that PDMS absorbs small molecules such as drugs and may have an impact on drug bioactivity in OOC devices designed to study cell behavior and drug efficiency [81]. While there are other materials that can be used in the generation of microfluidic devices, PDMS is one of the most predominantly used materials that is used in the generation of microfluidic devices, and these materials have been thoroughly reviewed elsewhere [82]. The lack of multi-organ interaction and communication is a drawback of this technology. However, researchers have reported the generation of multi-organ/human-on-a-chip. Abaci et al. reported a conceptual study on the design parameters and considerations in developing such a model [83]. The benefits offered by OOC technology outweigh the drawbacks, which have resulted in the continued development of this technology. With the incorporation of novel biomaterials and nanotechnology, OOC platforms are expected to evolve with technological advancements in the future.

#### **3. Biomaterials for 3D Tissue Modelling**

Advances in research have led to the development of improved 3D tissue models for in vitro studies. Cells in nature reside in a molecular matrix composed of protein, glycosaminoglycan, and glycoconjugate, termed the extracellular matrix (ECM). The ECM provides physical scaffolding, biochemical cues, and mechanical stability to cells and is necessary for morphogenesis and homeostasis [84]. The engineering of ECM that mimics native tissue matrix begins with the identification of a biomaterial that is critical in the formation of a scaffold. The choice of biomaterial is dependent on the tissue being modeled. Biomaterials are based on three categories (a) Polymers, (b) metallic, and (c) ceramics. Factors that influence the choice of materials are the type of tissue being mimicked, structural integrity, adequate mechanical environment, bioactivity, biocompatibility, and biodegradability [84]. The biomaterial should provide structural support for cellular attachment, growth, proliferation, and migration while consisting of adequate mechanical properties and an environment native tissue matrix provide to cells. Materials should be bioactive and biocompatible to provide bioactive cues and growth factors while reducing the risk of immunological response in the presence of an artificial scaffold. Additionally, the scaffold or matrix should act as a support structure facilitating correct localization and retention at the site of tissue damage [85]. While biodegradability is key for the formation of the vascular network and allows for patients' own ECM to replace the scaffold and degrade over time without any cytotoxic effects [86], this factor is organ-specific. For example, in regenerative medicine for hard tissues such as bone or teeth, materials are engineered from metallic or ceramic biomaterials to reduce the rate of biodegradability. Table 3 provides a summary of the various biomaterials and their pros and cons.

**Table 3.** List of Biomaterials, both natural and synthetic employed in tissue engineering and their advantages and disadvantages.



#### *Characterization and Optimization of Biomaterials*

There is a plethora of biomaterials available in the generation of ECM, such as structures, and the choice of biomaterial is highly dependent on the tissue of interest. On the formation of stable tissue-like constructs through any of the biofabrication techniques, the constructs should be subjected to various characterization techniques to ensure that they meet the parameters as close as possible to native tissue. Table 4 highlights the fundamental properties and the quantitative methods utilized in the characterization of these properties of biomaterials.

**Table 4.** List of quantitative methods utilized in the characterization of the fundamental properties of biomaterials.


#### **4. Cell Sources**

The incorporation of cells is essential in the generation of functional 3D tissue models. In general, cells can either be seeded on an existing carrier matrix or can be encapsulated within a biomaterial [108]. The factor governing cell incorporation is dependent on the tissue architecture fabrication method. While there are various methods to create scaffolds for 3D tissue models, the choice of the cell is highly dependent on the tissue being modeled. Primary cells closely mimic in vivo physiological state of the tissue or organ of interest; however, not all organs or tissues have primary cells in sufficient quantities or have limited proliferative potential. Figure 8 provides a schematic representation of the various cell sources [108–112].

**Figure 8.** Schematic representation of the Cell Sources. (**A**) skeletal myoblasts, (**B**) adipose-derived stem cells, (**C**) cardiac 'progenitor' cells and cardio sphere-derived cells (**D**) bone marrow-derived stem cells (**E**) Embryonic stem cells derived from the blastocyst (**F**) induced pluripotent stem cells derived from skin. Modified and adapted with permission from [113] under creative commons license CC BY 4.0. Copyright 2015 The Authors. Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd. and Foundation for Cellular and Molecular Medicine.

Recent findings on the differentiation of stem cells towards any tissue-specific lineages have led to significant advancements in tissue modeling and have been reviewed exhaustively elsewhere [108,111,112]. The conglomeration of novel biomaterials fabrication strategies, advances in stem cell biology, and 3D bioprinting has evolved as a next-generation technology for in vitro tissue model development. Table 5 provides a summary of the advantages and disadvantages of the various cells used in 3D bioprinting.


**Table 5.** Advantages and Disadvantages of the various types of stem cells found within the human body.

#### **5. Imaging Modalities of 3D Tissue Models**

There are a variety of methods used in the generation of 3D tissue models which have been discussed. While characterization methods are employed to ensure that the physical, mechanical, chemical, and biological parameters are met, these methods are often destructive and do not provide an insight into what is happening in the tissue once it is constructed. Therefore, additional methods are required to characterize and ensure that the final tissue model works as intended. In 2D cell culture and model systems, imaging, molecular, and immunohistochemistry techniques are commonplace. However, in a 3D system, advanced techniques are essential. Imaging techniques allow observation of the live-cell morphology and other organelles within the cells from 3D tissue models. Scanning electron microscopy analysis helps to find the cell morphology, migration, attachment, and cell–cell and cell–matrix interaction. Recent techniques have shown the real-time analysis of biological parameters in 3D cell/tissue models. Ruslan et al. used polymer-conjugated nanoparticles to identify O2 in cells present in the 3D tissue models [116]. Muller et al. used nanoparticle-based fluroionophore to study live analysis of K+ flux in 3D tissue models and animals [117]. Cell density can be analyzed with the presence of nucleated cells with H&E-stained histologic section photographs by using the ImageJ tool [118]. Table 6 provides a summary of the common advanced imaging techniques used for the analysis and examination of 3D tissue models, and Figure 9 provides the use of Optical Coherence Tomography (OCT) to characterize hydrogels.

**Table 6.** Common advanced imaging techniques used to analyze 3D tissue models and live cells within scaffolds.



**Table 6.** *Cont.*

**Figure 9.** (**a**) Schematic representation of FRAP method, (**b**) Example of a FRAP experiment (**c**) Anatomy of a typical FRAP curve. Modified and reprinted with permission from [123] under creative commons CC BY 4.0. Copyright © 2023 by the authors; licensee MDPI, Basel, Switzerland.

#### *5.1. Fluorescence Recovery after Photobleaching (FRAP) Using Confocal Microscopy*

Confocal fluorescence microscopy is an advanced method of fluorescence microscopy where high-resolution images can be obtained by the introduction of a spatial pinhole before the light source and the detector [128]. The aperture of the pinhole can be controlled to limit diffraction and thereby eliminate out-of-focus light from the sample. FRAP is a method used to study the movement of molecules that have been doped with a fluorescent dye (Figure 9) [123]. In FRAP, mobile fluorescent molecules are bleached by a high-intensity laser source. The bleached molecules are exchanged with fluorescent molecules from the surrounding area resulting in a recovery of fluorescent intensity. This information is plotted on a recovery curve and can be used to study the behavior of the molecules (Figure 9) [129]. The key advantage of using FRAP with confocal microscopy is that a small region in high resolution can be observed. For example, it is possible to study oxygen diffusion in a scaffold. Lee et al. used this method to examine the microscale diffusion of oxygen in scaffolds generated via electrospinning [130]. By introducing simulated cell concentrations, the study reports the ability to predict the efficiency of the scaffold. However, this technique requires further standardization protocols to be established as a viable method to characterize 3D tissue models.

#### *5.2. Optical Coherence Tomography (OCT)*

OCT is a type of imaging modality that performs high-resolution, cross-sectional imaging of microstructures in biological materials by measuring optical backscatter from different microstructural features within materials and tissues [126]. OCT can be used to observe the spatial and temporal changes of these features in real-time and in three dimensions, allowing the screening, identification, and optimization of parameters that govern the usability of tissue [131]. A key feature of OCT is capturing details in high resolution between 15–20 μm depths, thereby allowing the ability to observe scaffold architecture in intricate details [131]. The characteristics of scaffold architecture include parameters such as porosity, pore size, and degree of pore interconnectivity, which influence cellular activity, including cell adhesion, distribution, and proliferation [119,131]. A nondestructive method, OCT imaging, can be used to quantify changes in porosity as the scaffold degrades and cellular growth profile. For example, Zheng et al. used OCT to demonstrate the importance of OCT in the reconstruction of scaffold architecture and cell adhesion by capturing high-resolution images of two scaffolds with different seeding densities of human embryonic kidney cells [131]. Their study concluded that OCT is a viable method that can be used to optimize the parameters of scaffolds. More recently, Wang et al. used OCT to capture high-resolution images of the inner microstructures of cell-laden 3D-printed scaffolds. The study incorporated C3A cells in the gelatin–alginate hydrogel with varying pore sizes and utilized OCT to quantify morphological features, including pore size, pore shape factor, volume porosity, and the interconnectivity of the pores, as shown in Figure 10 [132]. Ultimately, this imaging modality has the capability to improve the understanding of the intricate structures, thereby leading to improved scaffold architecture designs, efficiently mimicking in vivo architecture and improving the efficacy of 3D tissue models.

**Figure 10.** Cell-laden 3D bioprinted structures with varying pore size characterized using OCT (**A1**–**A6**) Macrographs (**B1**–**B6**) Micrographs, (**C1**–**C6**) C OCT Cross-sectional images to a depth of 3 mm. (**D1**–**D6**) en-face OCT images. (**E1**–**E6**) rendering in 3D and (**F1**–**F6**,**G1**–**G6**) 3D reconstruction of hydrogel exhibiting variation in pore size. Image reprinted with permission from [132] under the creative commons license CC BY 4.0. Copyright © 2023 by the authors; Scientific Reports [132].

#### **6. Other Imaging Modalities**

While FRAP and OCT are imaging modalities that can be utilized to perform characterization on 3D tissue models, confocal microscopy imaging provides other methods to characterize and analyze 3D tissue models. Such modalities include Fluorescence Loss in Photobleaching (FLIP), Fluorescence localization after photobleaching (FLAP), Fluorescence Resonance Energy Transfer (FRET), Fluorescence Lifetime Imaging Microscopy (FLIM), Phosphorescence Lifetime Imaging Microscopy (PLIM), and Micro-Computerized Tomography (MCT), and a summary of their characteristics and application can be found

in Table 6. Figure 11 provides a workflow of the modalities. Ishikawa-Ankerhold et al. have provided an exhaustive review of the same [123].

**Figure 11.** (**a**) workflow of FRET (**b**) Workflow of FLIP, (**c**) workflow of FLAP, (**d**) workflow of Photoactivation. Images reprinted with permission from [123] under creative commons CC BY 4.0. Copyright © 2023 by the authors; licensee MDPI, Basel, Switzerland.

#### **7. Conclusions and Future Perspective**

This review highlights the vast potential of 3D in vitro models for the generation of tissue mimics, disease modeling, and assessment of innovative drugs toward personalized medicine over 2D models. While 2D modeling is a traditional and established method, it lacks the capability to replicate human physiology and diseased conditions. In the context of tissue engineering, the various methods used in the generation of artificial constructs, along with their advantages and disadvantages, are discussed. The potential role of these methods in regenerative medicine is also highlighted. Biomaterials play an important role in the generation of such constructs and models. The choice of biomaterials that have the capability to closely replicate human physiology and promote cellular functions within artificial constructs is critical when considering modeling. With the advent of various stem-cell types, specifically iPSCs (induced pluripotent stem cells), research in disease

modeling and personalized medicine has taken an innovative direction. The key advantage of employing the strategy of using 3D in vitro model systems is a reduced dependence on animal models, which are dissimilar to human physiology.

Most reviews discuss the state-of-the-art in tissue engineering research and regenerative medicine; however, methods used in the assessment of artificially generated constructs are a key area that is often neglected. An important aspect of 3D models and tissue engineering is to ensure that the artificial construct has the capability to replicate physiological conditions as closely as possible. Methods such as FTIR, mechanical testing, and biological activity assays to determine cell proliferation and survivability, to name a few, enable researchers to establish artificial tissues as efficient models and maintains standardization from a regulatory perspective. This review provides an exhaustive analysis of the various characterization methods used to evaluate artificially constructed 3D models along with various imaging modalities. Imaging has the capability to provide researchers with a tool to observe the functioning of cells at a microscopic level. It provides a platform where researchers can develop a deeper understanding of the attributes involved in the development and progression of the disease through direct observation. Methods such as optical coherence tomography are used in observing the structure of scaffolds in 3D, while FRAP and FRET can be employed to observe cellular functions.

A key challenge with 3D in vitro modeling is that while it has the capability to closely mimic human physiological conditions, it is an incomplete model, hence the reliance on animal models. Towards the future (Figure 12), it is imperative to focus on research towards the development of models that completely considers and mimics various factors and functions within the human body. The choice of biomaterials to have the right cellular microenvironment, appropriate mechanical properties as that of the relevant tissues of interest, the right orientation of cell/s, vascular networks, immune cells, the spatio-temporal release of necessary factors needed for the differentiation or growth of cells, and other factors unique to the tissues of interests, such as conduction properties in case of cardiac and neural tissues. This will allow researchers to work with improved 3D models, develop an improved understanding of diseases, and provide targeted solutions which are easy to manufacture, economically viable, and safe to administer. Furthermore, with the recent implementation of FDA Modernization Act 2.0, we envision that more emphasis will be given to complex and more sophisticated human physiology-relevant 3D in vitro tissue models for drug testing applications in the near future.

**Figure 12.** The future of in vitro models: a perspective.

**Funding:** Arunkumar Palaniappan would like to kindly acknowledge the financial support from the Science and Engineering Research Board (SERB), Department of Science and Technology, Government of India through its start-up research grant scheme (SRG/2020/001115).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** All the data and materials that support the results or analyses presented in the paper will be made available upon request.

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

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Review* **Looking into the Eyes—In Vitro Models for Ocular Research**

**Krystyna Lieto 1,†, Rafał Skopek 2,†, Aneta Lewicka 3, Marta Stelmasiak 4, Emilia Klimaszewska 5, Arthur Zelent 2, Łukasz Szyma ´nski 2,\* and Sławomir Lewicki 1,2,4,\***


**Abstract:** Animal research undoubtedly provides scientists with virtually unlimited data but inflicts pain and suffering on animals. Currently, legislators and scientists alike are promoting alternative in vitro approaches allowing for an accurate evaluation of processes occurring in the body without animal sacrifice. Historically, one of the most infamous animal tests is the Draize test, mainly performed on rabbits. Even though this test was considered the gold standard for around 50 years, the Draize test fails to mimic human response mainly due to human and rabbit eye physiological differences. Therefore, many alternative assays were developed to evaluate ocular toxicity and drug effectiveness accurately. Here we review recent achievements in tissue engineering of in vitro 2D, 2.5D, 3D, organoid and organ-on-chip ocular models, as well as in vivo and ex vivo models in terms of their advantages and limitations.

**Keywords:** in vitro eye models; 3D eye models; tissue engineering; ocular toxicity; eye irritation; corneal equivalents

#### **1. Introduction**

The number of factors that can damage human tissues increases every year. For example, smog, substances contained in cosmetics, unnatural food additives, and UV radiation have a harmful effect on our skin and eyes. Moreover, every year industry delivers thousands of new chemical substances which are necessary for new medicines, chemicals, or food additives. Therefore, biocompatibility assessments of each new compound, especially one that involves animal testing, is impossible. Moreover, the number of studies performed on animals has to be limited in Europe by the law (i.e., Directive 2010/63/EU).

Optic neuropathies, such as glaucoma, anterior ischaemic optic neuropathy (AION), traumatic optic neuropathies, optic neuritis, etc., need new treatment options, which in turn require the development of disease models [1]. However, the pathophysiological mechanisms of these diseases are not fully understood; therefore, developing an animal model is a tough challenge. Moreover, due to the physiological differences, animal models differ significantly from human diseases [2]. For example, rodents' eyes do not have maculae or foveae, and 85–90% of their optic nerve axons decussate to the other side of the brain [1]. On the other hand, monkeys' anatomy of the retina and optic nerve is almost identical to that of human eyes. Still, monkey breeding is complicated, very expensive, and time-consuming; therefore, the number of tests performed on individual animals is limited. As a result, monkeys are often used in the stage just before clinical trials on humans [1].

**Citation:** Lieto, K.; Skopek, R.; Lewicka, A.; Stelmasiak, M.; Klimaszewska, E.; Zelent, A.; Szyma ´nski, Ł.; Lewicki, S. Looking into the Eyes—In Vitro Models for Ocular Research. *Int. J. Mol. Sci.* **2022**, *23*, 9158. https://doi.org/10.3390/ ijms23169158

Academic Editors: Patrick C. Baer and Ralf Schubert

Received: 13 June 2022 Accepted: 11 August 2022 Published: 15 August 2022

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

**Copyright:** © 2022 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/).

Historically, one of the most popular experiments performed on animal eyes is the Draize test. Developed in 1944 by American toxicologists John H. Draize and Jacob M. Spines, it was widely used to study cosmetics and other chemicals. However, the test arouses many controversies due to the lack of reliable and objective results. In fact, the test was never correctly validated. Briefly, the test is based on applying the test substance directly to the eye, but the exposition time is not well defined. After observing the eye reaction for some time, the substance is washed from the eye, and the animal is observed for another two weeks. The result is subjectively assessed by the operator [3]. Moreover, the test is considered incorrect mainly due to anatomical and biochemical differences between the human and the animal (mostly the rabbit) eye. Therefore, currently, the Draize test is not performed. Instead, chemicals are usually tested using the EpiOcular eye irritation test, in vitro cytotoxicity assay, and irritation tests on the rabbit's skin. Each of these tests has advantages and disadvantages, but none of them allows for multifactorial compound-eye interaction evaluation. The EpiOcular eye irritation test is an in vitro alternative which allows for the assessment of acute eye irritation in response to the topical administration of chemicals onto the EpiOcular cornea epithelial model. The test makes cytotoxic effect measurement possible and provides a tool for eye-hazardous chemical identification.

For these reasons, it is crucial to develop new in vitro tissue models to study all substances for ocular treatment and to understand the development and molecular causes of eye diseases [4]. Here, we summarized all cellular and tissue-specific animal models used in in vitro eye studies.

#### **2. Eye Structure**

The eye is a highly complex biological machine (Table 1). The human eye has the shape of a sphere about 24 mm in diameter. It is filled with a vitreous body that allows the shape of the ball to be maintained. It is located in the eye socket, which reduces the risk of mechanical damage. The eye is divided into two parts: external and internal. The eye's outside layers are tough, elastic structures, with a white sclera and transparent cornea which provide eye shape [5]. The second inner layer is the vascular membrane, including the iris, ciliary body, and choroid. The third layer is the retina. It consists of light receptors, cells, nerve fibers, and blood vessels originating from the central retinal artery [6]. The primary function of the eye is to convert light pulses into electrical signals, which are transmitted to the brain and converted into images. Light is refracted by the cornea and the lens, which results in a sharp, inverted, reduced image formed on the retina. The amount of light reaching the receptors is regulated by the iris, which changes pupil diameter [6].

The retina in vertebrates is characterized by light-sensitive structures and covers 60% of the back of the eyeball. It is located above the choroid. It consists of 10 cell layers and contains photoreceptors called rod cells and cone cells [7]. These cells contain a visual pigment which is located in the cell membrane. Rod cells allow the recognition of shapes and motion at low light intensity [8]. Cone cells are responsible for seeing color and detail in a more intense light than rod cells. Individual cone cells differ in sensitivity at wavelength, which allows us to distinguish colors. In humans, cone cells are concentrated in the macula—a small pocket of the retina center. The optic nerve's axons that leave the retina at the ONH form a blind spot. There are no photoreceptors in that spot [9].

The eyeball is surrounded by the connective tissue—the sclera. It is a hard layer protecting the inner structures of the eye. In addition, it stiffens the eyeball like a bag of collagen and elastic fibers [10]. The sclera is thinner, more permeable to light, and creates a transparent cornea on the front. It refracts the light rays so that they fall on the lens. It is susceptible to pain and able to partially regenerate. It is avascular and consists of six layers (epithelial, Bowmann, stroma, Dua's, Descemet, and endothelial).

The conjunctiva lines the inner surface of the eyelid. It contains a lot of mucous cells, which ensures the constant humidity of the eyeball—it produces mucus and tears. It covers the eyeball up to the edge of the cornea. It has a very high regeneration ability. The conjunctiva is sensitive to any irritation, such as smoke, dust, or chemical substances. These factors can lead to conjunctivitis. During this type of inflammation, blood vessels are firmly filled with blood, causing redness and swelling of the eye [11]. Glands are dispersed in the conjunctiva, and the lids secrete mucus, water, and lipids, forming a tear film whose primary function is to moisturize and cleanse the eye from undesired foreign bodies if needed. There are many glands dispersed in the conjunctiva and the lids that secrete mucus, water, and lipids, thus forming the tear film. However, the main lacrimal gland (responsible for emotional tears) is outside the eye structure.


**Table 1.** Layers of the eyeball and their functions.

Even though the eye is a very specialized organ, there is significant progress in the development of tissue engineering, and newer and more suitable in vitro models are emerging. The search for such models is caused, among others, by increasing awareness of the welfare of animals used in experiments, including toxicological effects [12].

#### **3. In Vitro Ocular Models**

Over the years, many alternative assays were developed to accurately evaluate ocular toxicity and drug effectiveness. Here we present recent achievements in tissue engineering of various ocular models in terms of their advantages and limitations (see also Figures 1 and 2).

#### *3.1. 2D Eye Models*

Currently, 2D models are the most popular ones in ocular research. Two-dimensional cell line culture is an inexpensive, well-established model providing results that are easy to compare with the vast literature. However, the unquestionable drawback of these culture systems is the lack of predictivity in research connected with the fact that cells growing on a flat surface are not an equal representation of the cell environment in the organism.

#### 3.1.1. Pigment Epithelium Cell Lines

One of the most common 2D models is immortalized retinal pigment epithelium (RPE) cell lines. Primary cultures of retinal cells are challenging to handle. Obtaining a homogenous cell line that is not contaminated with other eye cells is challenging. Furthermore, isolated cells often quickly change their properties. For example, cells can lose keratin-containing intermediate filaments [13]. Cell transformation using the SV40 virus managed to obtain a line that retains the characteristics of retinal cells [14]. These cells are characterized by appropriate polarization and monocellular epithelial cell formation.

In 1995, RPE cells were first isolated by Davis et al. from a patient. However, the RPE cell line was only used in toxicity tests because the cells lost the characteristics of normal metabolism, adequate cytoskeleton polarization, and enzyme activity [14]. In the literature, primary models of RPE cell culture obtained from mice (i.e., Mouse Retinal Pigment Epithelial Cells-Hpv16 E6/E7, Immortalized) [15], rats (RPE primary cells isolated from PVG rats susceptible to experimental uveitis development; RPE isolated from Long Evans rats) [16,17], chickens (primary RPE cells isolated from domestic chickens embryos at stages 29–31 of development) [18], bovines (primary RPE cells) [19], and frogs (Xenopus laevis isolated primary RPE cells) [20] have also been described.

The human cell line ARPE-19 has structural and functional properties characteristic of RPE cells in vivo (in rats, RPE-J) [11]. This line is essential because the number of tissue donors is limited [21]. Studies on ARPE19 showed several features confirming the usefulness of this line for retinal pigment epithelial examination, such as expression of characteristic RPE cell markers, CRALBP, and RPE65, secretion of IL-6 and IL-8, as well as morphological polarization in monolayers, and ability form tight-junctions [22,23].

RPE-340 are primary cells isolated from humans which have epithelial morphology, but after several passages, their ability to replicate is limited [24]. Human RPE cells are good models for pharmacodynamic and physiological evaluation of a drug's effect on the choroid-RPE-photoreceptor, but after 40–60 population doublings, they go into a senescence state [24]. To develop a cell line with an extended lifespan, RPE-340 was transfected with a plasmid expressing the human telomerase reverse transcriptase subunit (hTERT), creating a new cell line—hTERT-RPE-1 (human retinal pigment epithelial RPE-1) [25]. This way, the lifespan of hTERT-RPE-1 is extended without any alterations in the population, doubling time and RPE-340 characteristic features [26]. hTERT-RPE-1 is reported to be an excellent model for epigenetic regulation studies [27–29]. Unfortunately, this line still has its limitations. The handling lasts 20 passages longer than RPE-340, but after this period, the cells change their morphology and function under the phenomenon called deadaptation [30,31]. Due to the low availability of primary human RPE cultures, validating and comparing this cell line with immortalized cell lines is challenging. The perfect line of human RPEs has yet to be developed.

**Figure 1.** Schematic diagram of of eye models types. The figure was created using SMART (Servier Medical ART) modified graphics, licensed under a Creative Commons Attribution 3.0. Generic License.

**Figure 2.** Schematic diagram of eye organoids. The figure was created using SMART (Servier Medical ART) modified graphics, licensed under a Creative Commons Attribution 3.0. Generic License.

The R28 immortalized retinal precursor cell line originating from postnatal day 6 rat retinal culture has been frequently used in in vitro and in vivo studies [32]. R28 provides an important system for understanding retinal cell behavior aspects such as differentiation, cytotoxicity, light stimulation, and neuroprotection. Although R28 originated from single clones, they remained highly heterogenous, suggesting the precursor character of these cells [32]. This cell model has been used in various toxicity experiments in vitro [33–37]. In addition, R28 exerts a high potential for studying the neuroprotective properties of chemical compounds [38–40]. Latanoprost was one of the drugs validated on R28 under the angle of cytoprotective properties [33].

RGC-5 was previously described as a rat-derived, transformed retinal ganglion cell line and is widely used in glaucoma research [41]. After more than 220 published papers worldwide involving the use of the RGC-5 cell line, it was reported that these cells are in fact 661W, a mouse SV-40 T antigen transformed photoreceptor cell [41,42]. The 661W cell line was present in the laboratory of origin of RGC-5; therefore, the most probable scenario was the cross-contamination of the newly developed cell line with 661W. This incident has shown how crucial the proper culture protocols and DNA profiling of newly-developed cell lines are. 661W is a model of cone photoreceptor cells. This cell line was widely used as a model for research on macular degeneration, but studying retinal ciliopathies such as retinitis pigmentosa is believed to be possible [43]. 661W shows properties of both retinal ganglion and photoreceptor cells, providing a functional photoreceptor model [43,44]. Moreover, 661W are believed to be an alternative model to the hTERT-RPE-1 cell line previously used for small molecule screening to identify new treatments for retinal ciliopathies [45]. 661W shows potential in studying ciliopathy disease genes not expressed or expressed at a low level in hTERT-RPE-1 cells [43].

#### 3.1.2. Cornea Cells

The primary cornea cultures on which the individual in vitro models were developed come mainly from rabbits. Rabbit corneal epithelial cells (RbCEpC) help assess drug safety, pharmaceutical effects, corneal development, pathology, glaucoma, viral infections, keratitis, ocular hypertension, and even special contact lenses that provide sustained, extended-release of ophthalmic drugs [31]. Human corneal epithelial cells (HCEpC) have been used as models for studying corneal damage and reconstruction, re-epithelialization of the eye following surgery, and the effects of degradative enzymes. Corneal research mainly focuses on developing a model of drug permeation through this structure [46]. The models of corneal culture used in cellular research concern simple monolayers, the multilamellar epithelium, and very complex three-dimensional (3D) tissues resembling the functional cornea. HCEpC was used to create a single cellular layer later used for transplantation [47].

Several commercially available in vitro cornea models are destined to be cultured in 2D models (monolayers). One such model is HCE-T, in which cells are grown on the collagen membrane and are located at the air-liquid interface with the serum-free medium. The cells have the features of the primary cell line and form a stratified epithelium whose morphology can be modulated with calcium. Moreover, the cells expressed specific corneal epithelial cell markers such as epidermal growth factor (EGF), EGF receptor, basic fibroblast growth factor (basic FGF), transforming growth factor-beta 1 (TGF-beta 1), and interleukin-1 alpha (IL-1 alpha) [48].

#### 3.1.3. Corneal Endothelial Cells

The role of corneal endothelial cells (CECs) is to control corneal transparency. Unfortunately, the cells exhibit limited proliferative capability; therefore, their dysfunction may be one of the causes of blindness. One of the gold standards in treatment of corneal endothelial dysfunction is the donor isolated corneal transplant [49]. The first culture of human corneal endothelial cells from donors was established by Pistov et al. in 1988 [50], and from that time, plenty of protocols and newly designed biomaterials for the propagation of these cells were developed [51]. However, due to the low proliferation rate of primary CECs culture, protocols for immortalized cells were established [52,53]. Currently, several immortalized human CECs are available in the market and are used mainly to understand corneal endothelial cell dysfunctions. Two clonal cell lines derived from the immortalization of human corneal endothelial cells (obtained from the donor) were described by Valtink et al. [54]: B4G12 and H9C1 cells. B4G12 cells are polygonal, strongly adherent cells, which form a strict monolayer and H9C1 cells are less adherent and formed floating spheres. Both cell lines exhibited the characteristic expression of corneal endothelial cell markers; however, on different levels. Therefore, the authors concluded that the B4G12 cell line is a good model of differentiated CECs, and H9C1 is a good model for developing or transitional CECs. Alternatives for donor corneal endothelial cells or cornea endothelial cell lines may be pluripotent stem cell-derived corneal endothelial cells. The cells generated from a cryopreserved human embryonic stem cell (hESC) are stable, express corneal endothelial cell markers, and have an improved proliferation rate compared to primary CECs [55,56].

#### 3.1.4. Conjunctival Cells

The most popular eye conjunctival test model is the rabbit conjunctiva. For the first time in 1996, Saha et al. isolated the primary culture of the rabbit's conjunctiva. The model created by him represents a tight epithelial barrier [57]. This model is constantly being improved. The most significant difficulty was adapting the cells of this model to contact with air, just like in the natural eye. The use of additional filters (for example, the Transwell filter) allowed cell growth at the air-liquid interface. The layers of the conjunctival epithelial cells showed transepithelial resistance and a difference in potential [58]. The conjunctival epithelial cells are polygonal with many microvilli [59]. The primary culture of conjunctival cells was also obtained from bovines [59] and rats. The immortalized rat conjunctival (CJ4.1A) cell line was created by transfection of SV 40 [60]. CJ4.1A expresses the SV40 T antigen, conjugal cytokeratin 4, and cytokeratin specific for goblet cells 7, but not the cytokeratin 12. The line's lifespan is very long—line cells can be cultured for over 60 passages, and the population doubling times were 22 ± 7h[60].

The development of methods for obtaining the primary culture of conjunctival cells has contributed to the development of transplantation techniques for heterotopic or allogeneic grafts. After severe damage to the conjunctiva, it is possible to restore its function by taking a piece of epithelium from a healthy eye, multiplying it in a cell culture, and implanting it in the affected conjunctiva [61]. Besides the primary cell culture, several established conjunctiva cell culture lines also exist. An example would be two human immortalized conjunctival cell lines: HCjE [62] and IOBA-NHC [63]. These cells have a typical epithelial morphology of the human epithelium, and after exposure of the cells to inflammatory mediators (IFNγ and/or TNFα), they increase the expression of the intercellular adhesion molecule (ICAM)-1 and MHC class II cell surface receptor (HLA-DR) [63].

The above-mentioned 2D models have their advantages but also their limitations. Firstly, these models are exceptionally delicate, and their manipulation must be meticulous. For example, the layer is easily damaged and dried. In addition, the models do not take into account cell to cell communication and the influence of immunological factors, which probably have a tremendous impact on the regeneration of this structure. Finally, based on these models' results, it is impossible to recapitulate all the processes occurring in the cornea in the human eye [64]. Therefore, the researchers decided to develop more complex, multicellular eye models.

#### *3.2. 3D Models*

Three-dimensional models better replicate the organism-environment compared to two-dimensional cultures. Cells grow in every dimension and closely replicate tissue in vitro, which complements 2D cell culture [65]. Although 3D models provide us with more information than 2D models, they are more challenging to handle. Multilayer models respond to more and more questions about corneal damage and disfigurement, but they are still far from the complex equipment that the eye is. For example, they lack the lacrimal apparatus responsible for cleansing and supporting regeneration [66]. Few 3D cornea models have been developed to this point.

EpiOcular™, developed in 2010, was obtained from cultured human epithelial cells. The cells showed a morphology and expression of biomarkers similar to the intact human cornea and maintained its thickness and permeability [67]. The EpiOcular model was used to assess the eye irritation potential of surfactant and surfactant-based formulations. Based on the protocol, the compound is considered to be an irritant when more than 50% of the cell die as compared to the negative control [68,69]. This test is validated and under review by the European Center for the Validation of Alternative Methods (ECVAM).

Clonetics (cHCEC) was developed in 2011 and was obtained from human corneal epithelial cells. Research using the model provides information on the assessment of corneal penetration by various chemical compounds (e.g., ophthalmic drugs) [68]. cHCEC was examined by RT-PCR for the expression profile of drug-metabolizing enzymes (e.g., CYP P450s and UGT1A1) and transporters in cHCE in comparison to the human cornea [70].

The SkinEthic (HCE) 3D cornea model comprises immortalized human mucosa cells; cells are grown at the air-liquid interface using a polycarbonate membrane. Under appropriate conditions, the cells differentiate and form the three-dimensional (3D) stratified epithelium and have non-keratin structures [69]. The advantage of the model is the possibility of administering dissolved substances in organic and inorganic solvents at any concentration (a very concentrated solution or minimal drug application concentrations can be applied) and the preservation of conditions similar to the eye mucosa of the human eye. Immortalized cells are grown in a dedicated medium and form a histologically multilayered construct with a thickness of 60 μm. The HCE secretes the same mucins found in the human cornea in vivo and expresses CD44 and keratin. This model is used to study phototoxicity, irritation, corrosivity, and the transport of substances [71,72].

The LabCyte CORNEA-MODEL is produced from normal human cornea epithelial cells [73]. It was developed by differentiating and stratifying cornea epithelial cells and is meant to be used to identify irritant chemicals in eye irritation tests. The corneal epithelial cells are cultivated on an inert filter substrate for 13 days with a medium containing 5% FBS. Proliferating cells build up in a multilayer structure consisting of a fully differentiated epithelium with features of the average human corneal epithelial tissue [74].

The limited source of corneal tissue to form a 3D-model, the short-lived life cycle of the corneal cells themselves, and the time-consuming culture contribute to problems with the industrialization of culture. The use of many commercially available 3D models was limited by the rapid differentiation of cells leading to problems with maintaining cell culture [75]. Many attempts have been made to increase the in vitro culture cycle of corneal epithelial cells concerning telomerase reverse transcription gene transfection, viral transfection, and the induction of spontaneous mutations. Nevertheless, the abnormal phenotype of these cells, which can lead to the potential risk of tumorigenesis, is not desired in the construction of new cornea models. Therefore, Li et al. enriched cornea cells with limbal stem cells providing additional expansion and development stimulation [75]. The addition of limbal stem cells promoted development and cell expansion. Moreover, it enabled the large-scale production of a new 3D model. Use of the corneal stromal layer of the animal to stimulate a specific microenvironment for limbal stem cells resulted in their differentiation into cornea epithelial cells.

Zuguo et al. proposed a new in vitro xeropthalmia model by dissecting the conjunctival epithelium and subconjunctival matrix, culturing it on a collagen I coated dish submerged in a culture medium. The in vitro dry eye model is obtained after 4–20 days. The invention can be used to research dry eye squamous metaplasia, ocular surface epithelial barrier damage, epithelial mucin change, to test new drugs, or to find new methods for dry eye treatment [76].

The model created by Minami et al. consists of bovine epithelial, stromal, and endothelial cells in a collagen gel matrix. The epithelium consists of five to six layers, and the epithelial cells produce keratin, which is a fundamental multilayer model for the cornea [77]. In addition, some corneal models use cell lines from different animals. In these models, individual layers come from mice, rabbits, bovines, and pigs [65,78,79].

#### *3.3. 2.5D Models*

2.5D models seem to be an alternative approach compared to 2D and 3D cultures. In 2D, cells are grown on a flat surface, while 3D models are based on cells embedded in an extracellular matrix (ECM) and/or scaffolds that provide a proper three-dimensional environment. In 2.5D cultures, cells are grown in an extracellular matrix (ECM) layer which often is not flat but unregular with projections and grooves, thus, providing an intermediate between 2D and 3D conditions [80].

#### *3.4. Ex Vivo Models*

One alternative to the Draize test is harvesting organs for examination from animals used for meat (ex vivo model). Eyeballs are isolated from bovines (BCOP), rabbits (IRE), pigs (PCOP), and chickens (ICE). This test was accepted internationally in 2009 and is used to research if significant tissue damage can occur [81]. Tests on the models mentioned above are based mainly on histological and light transmittance through cornea analysis. The pigs' cornea provides the highest degree of similarity to the human cornea, especially in tests involving substances dissolved in water [82]. Unfortunately, all the models mentioned above have serious drawbacks, primarily resulting from anatomical differences. In addition, these models can only be used to study individual eye structures. Therefore, they do not allow for general-purpose research. It is vital to create cell microenvironments that support tissue differentiation and changes, tissue-tissue communication, and spatiotemporal chemical and mechanical gradients of the microenvironment of living organs [83].

Yu F. proposed an ex vivo mammalian cornea culture system used for chemical tests of consumer products [84]. This system closely resembles in vivo testing by maintaining the corneal structure, architecture, and epithelial cell interaction. The cornea or the whole eye is excised and placed on an agar or collagen scaffold. It is then submerged in a culture medium until the medium covers the limbus. The upper part of the cornea is not submerged in medium. The tested reagent is administered directly to the cornea. The inventor claims that the system may be used to replace the use of Draize's test in many situations. This system allows drug testing without using live animals. The corneas or eyeballs may be, for example, easily acquired when dissecting rabbits for meat or fur industry purposes.

#### *3.5. Spheroids, Organoids, and Organ-on-Chips*

New techniques and technologies in cell culture allow the development of more proper and scientific-useful models for ocular research. Here we described three types of it: spheroids, organoids, and organ-on-chips, a summary of which is presented in Table 2.

#### 3.5.1. Spheroids

Spheroids are self-assembly aggregated cells that spontaneously organize themselves into spherical-shaped structures. This phenomenon occurs naturally during embryogenesis, morphogenesis, or organogenesis. In in vitro culture, single cells may constitute multicellular spheroids after applying appropriate cell culture techniques (i.e., pellet culture, the hanging drop method, culture in the extracellular matrix, or others) [85]. Spheroids may have a different biological response to various factors due to the presence of a concentration gradient of nutrients, oxygen, or metabolites between cells from the outside and the inside part of the spheroid. Spheroids are mainly used in cancer research [86,87]. However, the technique of 3D multicellular culture with spheroids is also used in cellular research. Lu et al., using air-lifting 3D spheroid formation techniques, developed an in vitro model for research on the ocular surface and tear film systems. The model was composed of rabbit conjunctival epithelium and lacrimal gland cell spheroids [88]. The model allowed for the creation of the aqueous and mucin layers of the tear film, which may facilitate research on dry eye. A Japanese-German research group generated multicellular spheroids from human-donor RPE cells cultured in a methylcellulose matrix [89,90]. The model mimics the in vitro drusen model, which might help understand the pathogenesis of drusen-related diseases such as AMD. Sherwin's group from New Zeeland developed methods for isolation and propagation of spheroid human peripheral cornea using a clear cornea component of the rim isolated from a donor [91,92]. They found that generated spheroids implanted into frozen-stored corneoscleral tissue worked as limbal stem cell centers and proliferated to reproduce limbal cells. Spheroids are also used in ocular cancer research. There are several spheroid models of retinoblastoma (cells isolated from human intraocular tumors) which are used to develop new cancer treatments [93] or to understand retinoblastoma pathophysiology [94,95].

#### 3.5.2. Organoids

Organoids are stem cell derived 3D structures with organ-level functions. They are composed of self-organizing organ-specific cells derived from embryonic stem cells, induced pluripotent stem cells, or organ-restricted adult stem cells [96,97].

One of the most well-known ocular organoid models is a model described by Eiraku et al. [98]. The authors used mouse embryonic stem cells and show that ESCs in differentiation medium are self-organizing into optic-cups in 3D culture. Susaimanickam et al. developed an organoid model based on human embryonic stem cells (ESCs) or human induced pluripotent stem cells (iPSCs) cultured in a retinal differentiation medium supplemented with noggin [99]. The addition of noggin is crucial because of the protein's (a BMP inhibitor) involvement in the retinal differentiation of pluripotent stem cells during embryonic and organoid development. After two weeks, the culture gave rise to retinal and corneal primordia, and after six to eight weeks, primordia developed into minicorneas with specific

morphological and marker similarities to the human cornea. This model may be used in basic research and regenerative applications. In addition, the use of organoid models with different ranges of time culture could provide us with data regarding drug toxicity in different stages of eye development. A congruous model or cornea organoids was developed by Foster et al. [100]. In this model of the cornea, three distinct cell types with the expression of key epithelial, stromal and endothelial cell markers were obtained. Mellough et al. in 2012 showed that ESC and iPSC cultured in ventral neural induction media (VNIM) supplemented with noggin, Dickkopf-1, Insulin-like growth factor 1, Lefty A, Human Sonic Hedgehog, and 3, 30, 5-triiodo-L-thyronine may develop retinal photoreceptor cells [101]. Later, they showed that VNIM can differentiate both EPS and iPSC cells, but the presence of IGF-1 is essential for the development of 3D ocular-like structures containing retinal pigmented epithelium, neural retina, primitive lens, and corneal-like structures [102]. In the latest work, Mellough et al. found that different embryoid bodies' (EBs) generation protocols affect the method and maintenance conditions that determine the later differentiation and maturation of retinal organoids [103]. The generation of more advanced in vitro multiocular organoids from human iPSCs cells was proposed by Isla-Magrané et al. [104]. In this protocol, organoids are differentiated in three different media, which leads to obtaining multicellular organoids after 150 days. Firstly, 75% confluent hiPSCs were cultured on Matrigel in an induction medium (DMEM/F12, 5% fetal bovine serum, nonessential amino acids, GlutaMax, N2, B27, β-glycerol phosphate, nicotinamide, Noggin, DKK1, bFGF) for 30 days. Next, all-trans retinoic acid (ATRA) was added for the next 60 days. Finally, cells maintained in a medium with ATRA for the next 60 days develop multiocular and corneal organoids, and cells cultured without ATRA and with triiodothyronine develop retinal organoids, RPE organoids, and multiocular organoids.

Recently, the National Centre for the Replacement Refinement & Reduction of Animals in Research (NC3Rs) and The National Eye Institute established a relationship that will result in the construction of organoids for drug screening, disease modeling, and regenerative medicine [105]. Therefore, a retinal 3D model is constantly being developed to fulfill those criteria [105–107]. The retinal 3D model contains bioprinted Müller cells, microglia, neurons, and RPE cells [108].

#### 3.5.3. Organ-on-Chips

Organ-on-chips (OoC) are structures created by combining microfluidic technology, biomaterials, and cell culture methods [97]. Many organ-on-chips were used to research the permeability of the epithelium. Puleo et al. created a microfluidic device consisting of a bilayer structure of a corneal epithelial layer, a layer of stromal cells, and collagen vitrigel substrate [109]. Bennet et al. invented a cornea organ chip including epithelial layers, Bowman's membrane, basement membrane, and a device simulating tear flow dynamics. The measurement of epithelium permeability underflow showed results similar to in vivo measurements [110]. Cornea and retina chips are powerful and promising in vitro tools to study drug effects and therapeutic approaches, yet the chips are still minimal and straightforward 60 [97].

Recently, Seo and Huh proposed a cornea-on-chip "human blinking eye model" [111]. The system mimicked spontaneous eye blinking in humans with keratinocytes cultured to mimic the epithelial cells and form a corneal structure. Blinking imitation was performed by integrating a tear chamber in a 3D-printed eyelid [112].

DynaMiTES' Dynamic Micro Tissue Engineering System was developed from cornea immortalized cells. The system allowed for the measuring of transepithelial electrical resistance in real-time by implementing two electrodes into the system, providing a noninvasive way to monitor cell conditions [113].

Although organoids and organ-on-chips carry indisputable benefits, their potential in drug testing has yet to be closely examined. The main issues concerning drug assays relate to permeation and accessibility of the ocular surface of the tested models [97].

#### *3.6. In Silico Analysis*

In silico analysis is often used to meet the 3Rs regulations (replacement, reduction, and refinement) [114]. Many in silico models have been proposed up to this point in time. One of them is a quantitative structure-property relationship (QSPR) model proposed by Vincze et al. to study corneal permeability. The model is based on corneal-PAMPA (Parallel artificial membrane permeability assay) experimental data and different in silico drug transport parameters (Caco-2 and jejunal permeability) [115]. The test provided good predictions and is suitable for efficiently shortening the examined drugs list, provided we have comparable experimental data at our disposal. However, although promising, in silico studies currently do not provide us with enough data to regard drugs as safe. Therefore, experimental testing should be carried out to confirm the result of the studies.

**Table 2.** Spheroids, organoids, and microphysiological models for in vitro ocular research.



#### **Table 2.** *Cont.*

#### **4. Conclusions**

Tissue engineering is one of the most rapidly developing scientific disciplines. It allows an easy and more reliable study of the effects of various factors and substances (including drugs). The development of this field will contribute to the invention of more advanced methods of combating diseases, repairing damaged tissues as a result of trauma, and to the ability to change and improve the function of given structures. At the same time, it will limit the number of animals used for experiments, which are now often indispensable research models. Currently, scientists are trying to fine-tune in vitro models and combine as many elements as possible to create a fully functional organ. One of the paths leading to this goal is the development of bioreactors. Bioreactors extend the time of in vitro culturing through specific, periodic exchanges of the culture medium. Physical factors are strictly controlled, e.g., temperature, pH, oxygen, and carbon dioxide. Additionally, they enable the precise delivery of nutrients and the removal of unnecessary metabolites from the nutrient solution [116,117].

All of the research models mentioned above have their limitations and advantages. Different Draize test alternative models provide more extensive flexibility in our research. Currently, 2D cultures are the most common research models. The reason is that 2D cultures are relatively inexpensive, more modulable, and easy to maintain [118]. Because of reproducible results obtained in controlled conditions [119], big-scale screening assays should be performed on these models. The main weakness of 2D models is their low ability to recreate the complexity of different cell classes and matrices interaction [118]. On the other hand, 3D multilayer models seem to be sufficient for small-scale drug toxicity and irritation assays. These models more closely resemble the eye microenvironment and consider cell-to-cell interactions, providing more relevant results. Moreover, 2D and 3D models seem to be limited in immunological disorders, such as allergy or sensitivity, because of their low of complexity. This problem could be addressed with organoids, which generate remarkable research outcomes, but only after long and arduous steps of standardization and testing [118].

Both in vitro and ex vivo models share one major limitation: the lack of vascularization [120]. The immune cells and vascularization should be introduced to these models

to address this problem more appropriately. Organ-on-chip technology may be applied to facilitate the manipulation of more complex research models [120,121]. For example, including blood vessels in the model is possible by applying a forced flow supplied by on-chip technology. All that remains is to hope that the current development of in vitro models in ocular research allows for the complete elimination of the need to conduct tests on living organisms in the near future.

**Author Contributions:** Conceptualization: S.L. and Ł.S.; investigation: K.L., R.S. and A.L.; writing original draft: R.S., K.L., M.S., E.K. and A.L.; writing—review and editing: A.Z., Ł.S. and S.L.; visualization: R.S. and Ł.S.; supervision: S.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study was supported by Polish National Centre for Research and Development project no: DOB-1-6/1/PS/2014.

**Institutional Review Board Statement:** The study did not require ethical approval.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data reviewed in this study are available on request from the corresponding author. The data are not publicly available due to founding agreement limitations.

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

#### **References**


## *Review* **A Review of Defatting Strategies for Non-Alcoholic Fatty Liver Disease**

**Erin Nicole Young, Murat Dogan, Christine Watkins, Amandeep Bajwa, James D. Eason, Canan Kuscu and Cem Kuscu \***

> Transplant Research Institute, James D. Eason Transplant Institute, Department of Surgery, College of Medicine, The University of Tennessee Health Science Center, Memphis, TN 38163, USA

**\*** Correspondence: ckuscu1@uthsc.edu

**Abstract:** Non-alcoholic fatty liver disease is a huge cause of chronic liver failure around the world. This condition has become more prevalent as rates of metabolic syndrome, type 2 diabetes, and obesity have also escalated. The unfortunate outcome for many people is liver cirrhosis that warrants transplantation or being unable to receive a transplant since many livers are discarded due to high levels of steatosis. Over the past several years, however, a great deal of work has gone into understanding the pathophysiology of this disease as well as possible treatment options. This review summarizes various defatting strategies including in vitro use of pharmacologic agents, machine perfusion of extracted livers, and genomic approaches targeting specific proteins. The goal of the field is to reduce the number of necessary transplants and expand the pool of organs available for use.

**Keywords:** liver; steatosis; defatting; molecular biology
