*Article* **SPR-Based Kinetic Analysis of the Early Stages of Infection in Cells Infected with Human Coronavirus and Treated with Hydroxychloroquine**

**Petia Genova-Kalou 1, Georgi Dyankov 2,\*, Radoslav Marinov 1, Vihar Mankov 2, Evdokiya Belina 2, Hristo Kisov 2, Velichka Strijkova-Kenderova <sup>2</sup> and Todor Kantardjiev <sup>1</sup>**


**Abstract:** Cell-based assays are a valuable tool for examination of virus–host cell interactions and drug discovery processes, allowing for a more physiological setting compared to biochemical assays. Despite the fact that cell-based SPR assays are label-free and thus provide all the associated benefits, they have never been used to study viral growth kinetics and to predict drug antiviral response in cells. In this study, we prove the concept that the cell-based SPR assay can be applied in the kinetic analysis of the early stages of viral infection of cells and the antiviral drug activity in the infected cells. For this purpose, cells immobilized on the SPR slides were infected with human coronavirus HCov-229E and treated with hydroxychloroquine. The SPR response was measured at different time intervals within the early stages of infection. Methyl Thiazolyl Tetrazolium (MTT) assay was used to provide the reference data. We found that the results of the SPR and MTT assays were consistent, and SPR is a reliable tool in investigating virus–host cell interaction and the mechanism of action of viral inhibitors. SPR assay was more sensitive and accurate in the first hours of infection within the first replication cycle, whereas the MTT assay was not so effective. After the second replication cycle, noise was generated by the destruction of the cell layer and by the remnants of dead cells, and masks useful SPR signals.

**Keywords:** SPR; cell-based assay; viral growth kinetics; human coronavirus; hydroxychloroquine

#### **1. Introduction**

The cell is the minimum functional unit of living organisms. Knowledge of the basic cellular components and the way cells work is fundamental to life sciences, including molecular biology, cell biology, cell physiology, etc. With the traditionally used cell-based assays, it is a common practice to use radioactivity, chemiluminescence, or fluorescence to produce a measurable signal. Label-free cell-based assays have sparked interest due to their ability to measure cell responses without additional reporter compounds. Among the different label-free techniques, optical methods have been widely adopted for cell-based assays. The most effective one—surface plasmon resonance (SPR)—has been applied in the study of a variety of cellular processes.

In its conventional approach, SPR detects the binding of molecules in the detection volume on a sensor chip in real-time without any labeling. The signals are generated by a change in the biomolecule layers and are linearly related to their thickness. This is true in the first approximation since the layers are uniform and much smaller than the penetration depth of the plasmon wave. The situation is different in cell-based SPR assays where cells are immobilized on the sensing surface and serve as sensing elements. Nevertheless, SPR

**Citation:** Genova-Kalou, P.; Dyankov, G.; Marinov, R.; Mankov, V.; Belina, E.; Kisov, H.; Strijkova-Kenderova, V.; Kantardjiev, T. SPR-Based Kinetic Analysis of the Early Stages of Infection in Cells Infected with Human Coronavirus and Treated with Hydroxychloroquine. *Biosensors* **2021**, *11*, 251. https://doi.org/10.3390/ bios11080251

Received: 20 June 2021 Accepted: 21 July 2021 Published: 26 July 2021

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

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

sensing has been extended into a powerful method for sensing large biological objects such as cells [1].

#### *1.1. Penetration Depth and Detection Depth*

An essential concept in SPR sensing is the nature of the evanescent field of the plasmon wave. This is especially relevant in functional cell-based assays because the penetration depth of 150−500 nm from the metal surface is only a fraction of the height (vertical dimension) of commonly used cells, which is in the range of several micrometers. Thus, it was suggested that the SPR signal is provoked by biological events in the area near the plasma membrane, whereas events inside cells, especially in their upper area, cannot be detected [2–4].

Two approaches have been used to achieve a greater penetration depth of the plasmon wave: excitation of long-range surface plasmon in a specific SPR biochip [5] and plasmon excitation in the UV range [6,7]. The latter approach seems to be more effective—the penetration depth could reach several microns. Even though these modifications are clearly advantageous, it may turn out that the more distant cell regions can be detected without applying them.

Although in the majority of cases the penetration depth is within the range of several hundred nanometers, SPR sensing is not necessarily applied to this limited range. SPR has been successfully used in detecting cell responses to external triggers such as drugs [8–11] and external stimuli [12].

It has been demonstrated [1,13,14] that a refractive index (RI) near the plasma membrane might reflect the accumulation and rearrangement of proteins activated by intracellular signal transduction provoked by exogenous stimuli. The SPR signals generated by the cellular response originate from complex biological events that have a local impact on RI. Additional experiments are required to find out what biological matter elicits the SPR signal.

#### *1.2. Application in Drug Research*

SPR technology has been widely applied in studying drugs. These studies have been generally limited to bimolecular binding assays outside living cells where purified biomolecules have been immobilized and a binding reaction with the target drug molecules has been detected [15–17].

Instead of immobilizing purified biomolecules in binding assays, a whole-cell adhesion to a sensing surface would provide on-site signals from drug–living cell interactions. Therefore, the pharmacokinetic parameters obtained by the cell-based assays would be more accurate and reliable than those obtained by the biochemical binding assays. A number of research groups have reported cell-based SPR assays used in evaluating the efficiency of a variety of drugs. A comprehensive review of cell-based SPR assay can be found elsewhere [18].

#### *1.3. Cell-Based SPR Assay in Virus Research*

The SPR technique has been widely used in studying viruses. Its well-known advantages are as follows: (i) the fact that it is label-free, thus eliminating functionalization of multiple antibodies, which occurs with ELISA; (ii) dynamic measurement of binding– unbinding kinetics; and (iii) high sensitivity, providing reliable virus detection. Reasonably, more research has been focused on viral diagnosis. The recognition elements used have included antibodies, antigens, DNA, and aptamers. Viral kinetic analysis of cells infected with SARS [19], SARS-CoV [20], and SARS-CoV-2 [21] has been performed. Surprisingly, the cell-based SPR assay has never been used in virus research so far.

#### *1.4. Aim of the Present Study*

Coronaviruses are disease-causing agents that infect many species of mammals and birds. Some, such as HKU1, OC43, 229E, and NL63, circulate seasonally and cause res-

piratory diseases in children and adults, which are not life-threatening. At the end of 2019, the identification of SARS-CoV-2 as the causal agent of atypical pulmonary diseases was the latest example of these emerging coronaviruses. It is essential to investigate the way in which the virus replication cycle occurs. MTT and immunofluorescence have been widely applied in investigating virus kinetics. However, the necessity of studying virus kinetics in the first hours of infection requires the application of new methods. Although the cell-based SPR assay has been successfully used to study intracellular processes, it has never been applied in examining the kinetics of the ultrastructure of virus-infected cells. The aim of this study was to prove that the cell-based SPR assay can be applied in the kinetic analysis of the early stages of viral infection of cells and the antiviral drug activity in the infected cells.

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

There are seven human coronaviruses (HCoV), highly diverse and causing respiratory diseases with mild to severe outcomes [22]. Currently, no specific antiviral drugs to treat HCoV infection are available, although hydroxychloroquine (HCQ) has been suggested as appropriate [23]. The way in which HCQ exerts its effect on time-dependent HCoV growth is well known and helps us analyze the SPR signal and compare it with other methods.

*Human cell line culture:* HCoV, strain 229E (HCoV-229E) was isolated using Vero E6 (African green monkey kidney) cell line, obtained from the National Center of Infectious and Parasitic Diseases (NCIPD), Bulgaria. The cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM, Sigma-Aldrich, Sent Luis, MO, USA), supplemented with 10% fetal bovine serum (FBS, GibcoTM by Life Technology, Darmstadt, Germany) and antibiotics at 37 ◦C, 5% CO2 atmosphere.

*Virus propagation:* The HCoV-229E (from the NCIPD viral collection) was propagated in Vero E6 cells that reached 70% confluence in DMEM media, supplemented with 2% FBS at 37 ◦C, 5% CO2 atmosphere.

*Virus titration:* Confluent Vero cells (3 × 104 cells/well) were cultured in 96-well plates (100 μL/well). Serial 10-fold dilutions of the HCoV-229E stock (10−<sup>1</sup> to 10<sup>−</sup>8) were prepared in DMEM supplemented with 2% FBS, at 37 ◦C and 5% CO2 atmosphere for 4 days. Cell viability and yields of virus progeny were measured post-infection (*p.i.*) every 24 h for a 96 h period of total incubation time. The infected cells were monitored microscopically daily for cytopathic effects (CPE) in the infected cells caused by HCoV-229. The titer of the purified HCoV-229E was 104.5 (high-titer) 50% tissue culture infection doses (TCID(50)/mL.

*MTT assay:* The MTT assay (3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (Methyl Thiazolyl Tetrazolium; MTT) is used to measure cellular metabolic activity as an indicator of cell viability, proliferation, and cytotoxicity. MTT is reduced by mitochondrial dehydrogenases to the water-insoluble pink compound formazan, depending on the viability of cells. Vero cells were seeded in 96-well microtiter plates (3 × 104 cells/mL) and infected with HCoV-229E, multiplicity of infection (MOI) 0.1, and treated with different HCQ non-toxic concentrations at different hours. Measuring the optical density (OD) by the MTT assay has been used as a sensitive method to quantify the density of the HCQ-treated infected cells [24]. The OD values have been measured at a wavelength of 540 nm using ELISA reader (Sunrise Basic Tecan, Männedorf, Switzerland), whereby the concentration of viable cells is found [25]. The same approach was used in our study as well.

*Cell-based SPR assay:* The SPR slides were derived from a recordable compact disc (CD-R). A gold layer with thickness 80–100 nm was deposited onto the polycarbonate substrate by vacuum thermal evaporation.

Before seeding the cells, the slides were immersed in isopropyl alcohol and cleaned ultrasonically for 10 min. Then, they were rinsed thoroughly with high purity water, dried, and illuminated by UV light for 24 h.

The Vero E6 cells were cultured in DMEM (Dulbecco's Modified Eagle Medium), supplemented with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics at a density 3 × <sup>10</sup><sup>4</sup> cells/mL and incubated for 24 h at 37 ◦C and 5% CO2 conditions to allow

cell adhesion to the SPR slide surface. When the cells achieved appropriate density (about 70% confluence on the SPR slide surface), a monolayer was washed twice with phosphate buffer solution (PBS), pH = 7.3, the supernatant was carefully removed, and the cell culture medium was supplemented with 2% FBS and HCoV-229E with multiplicity of infection (MOI) 0.1. After virus adsorption, the infected cells were treated with a non-cytotoxic concentration (1 mg/mL) of the antiviral drug HCQ. The infected and treated cultures were incubated at 33 ◦C in a humidified 5% CO2 atmosphere. Cell morphology was observed every 6 h for microscopically detectable morphological alterations, such as loss of confluency, cell rounding and shrinking, and cytoplasm granulation and vacuolization. The viability of the infected and treated cells from each well of the 96-well culture plate was determined every 6 h by an MTT-assay, and the SPR spectral shift of the cell-based SPR assay was also measured.

*SPR measurement:* The gilded diffraction grating is part of a continuous CD-R spiral groove with a period of 1.55 μm. A Θ-2θ goniometer with a 0.01 deg accuracy was used for the SPR excitation and registration. Spectral interrogation was used for the SPR registration. A collimated beam of p-polarized white light under angle of incidence in the range 35–42 degrees excited resonances between 710 nm and 610 nm in a bare grating. A spectrometer registered the spectrum in the zero-order reflection. The optical setup is depicted in Figure 1.

**Figure 1.** Optical setup for the SPR measurements.

Figure 2 shows the experimentally observed resonances in the bare SPR slides and the cell-based SPR assays: a slide with cells obtained at 12 h after seeding and a slide with infected cells at 12 h *p.i.* The changes in cellular morphology, which in turn led to a variation of the effective refractive index at the interface between the cell membrane and the metal layer, caused a well observable spectral shift, marked in Figure 2. Reference resonances were established for bare grating—the black curve in Figure 2. The spectral shift of the slides with non-infected cells was evaluated against reference resonances—marked as "A" in Figure 2. The spectral shift established in this way is referred to as "cell control". The SPR shift for infected cells was evaluated against the cell control—marked as "B" in Figure 2. The spectral shift established in this way is referred to as "virus control". The SPR spectral shift of the treated cells was evaluated against the virus control, after which it was compensated for by dividing the difference between cell control and virus control. The signal established was referred to as "SPR compensated signal". The SPR responses were measured at different time intervals between 4 and 48 h.

*AFM examination:* An Atomic Force Microscope (AFM) (Asylum Research MFP-3D (Oxford Instruments) was supplied with silicon nitride probes: frequency of 30 kHz, spring constant of 0.27 N/m, and radius <15 nm. The experiments were carried out at ambient conditions using the AFM contact mode. For the purposes of scanning, the cell-based SPR assays were fixed with 2.5% glutaraldehyde.

In the next section, we describe what type of biological events occur at different moments and determine the registered SPR shift.

**Figure 2.** Experimentally observed resonances of bare grating and cell-based SPR assays.

#### **3. Results**

*3.1. Cells Growth Kinetics*

Non-human primate kidney cell line Vero E6 was seeded at concentration <sup>3</sup> × 104 cells/mL on SPR slides and on glass plates. The cells on the glass plates were counted by MTT. Every 6 h, the SPR spectral shift of the cell-based SPR assay was measured. The results are presented as a dotted line in Figure 3. A reference measurement with an MTT assay was provided. The data obtained—the mean values of three indep endent experiments—are presented in Figure 3: an SPR assay measurement (dotted line) and an MTT measurement (solid line). Obviously, the SPR signal follows the temporal change of cell viability established by the MTT assay. We also observed that the Vero E6 cells grew rapidly, producing a confluent monolayer. Even after a prolonged period, the Vero E6 cells showed the typical morphological characteristics of spindle-shaped fibroblasts—flat, without prominence in their shape and surface, with intact cytoplasm and oval nucleus, 15–20 μm long, and about 5 μm high—all of which was microscopically proven by the AFM examination (Figure 4). The AFM study showed that the cells had adhered tightly to the grating surface, which ensured an effective penetration of the plasmon wave into the cells. The population doubling per day (Pd/D) and cell density were found to increase with the prolongation of cell cultivation expressed as a steeper SPR curve in the range 25–38 h (Figure 3). The growth curves constructed from both the SPR and MTT assays showed the typical pattern of a growth curve. The SPR curve clearly indicated a lag phase (until 20 h), an exponential phase (25–38 h), a plateau (around 40 h), and reaching a death phase (40–48 h). This suggested that the cell density used was appropriate for running the experiments with the present virus.

**Figure 3.** Growth kinetics of Vero cells: dotted line—SPR results (cell control); solid line—MTT results.

**Figure 4.** Vero cell at 24 h *p.i*.; AFM scan of the diffraction grating.

#### *3.2. Viral Growth Kinetics*

To estimate the viral growth curve characteristics, the infectivity titer of HCOV-229E was determined every 6 h *p.i.* by MTT (solid curve in Figure 5). The MTT assay measurement revealed the main phases of the viral growth kinetics. The initial increase in viability by 10 h is due to an increase in the number of uninfected cells as a result of the cell growth process.

**Figure 5.** Growth kinetics of HCoV-229E-infected cells: dotted line—SPR results (virus control); solid line—MTT results.

The increase in the SPR spectral shift (dotted curve in Figure 3) lasted until 18 h, probably due not only to the increased cell density on the grating surface but also to the attachment of the viruses to the cell membrane. Then, the SPR assay measurement accurately indicates the infection efficiency.

We would like to point out that the measurement time interval of MTT assay was almost twice as long as the SPR time interval. This is due to the lower time resolution of MTT. The highest time resolution of SPR assay can explain observed local maximums of the cell control around 12 h (Figure 3) and of virus control around 18 h (Figure 4).

MTT showed a substantial decrease in cell viability in the interval 20–24 h (Figure 5), which corresponded to dramatic ultrastructural changes—marked granulation of cell cytoplasm, particularly around the nucleus with the fragmentation of the latter, and proliferation of pseudopodia at the cell periphery. This was confirmed by the AFM examination (Figure 6).

**Figure 6.** Infected Vero cell at 42 h *p.i*.; AFM scan of the diffraction grating.

The SPR signal also decreased in this interval and reached its minimum around 30 h. This is close to the minimum of cell viability (MTT curve—Figure 5), well pronounced in the interval 30–40 h, confirming the cell monolayer destruction, cell deterioration, and detachment from the surface as a result of the increased virus production.

At the end of the first replication cycle (24–30 h), the virus was expressed in the intercellular space. This was clearly observed by the AFM study as shown in Figure 7, which represents a magnified part of an area (shown in Figure 4) located near the cell membrane. Viruses (marked with arrows) have just been expressed from the host cell at 24 h p.i and are still close to the cell membrane. After this moment, the second replication cycle starts: virus attachment to the cell membranes of new cells. As a result of this, compaction of the cell membrane occurred, and the refractive index increased. This coincided with the increase in the SPR signal after 30 h.

**Figure 7.** Viruses expressed in the intercellular space at 24 h *p.i*.

The change in the MTT signal was more inert—it increased after 36 h due to the competing processes of cell growth and viral replication.

#### *3.3. Kinetics of Antiviral Activity of HCQ*

Hydroxychloroquine (HCQ), used to treat malaria and some autoimmune disorders, might be of certain use in the clinical management of infections caused by HCoV, especially SARS coronavirus (SARS-CoV-1) and SARS-CoV-2, by potently inhibiting the infection, a fact that has been found in cell culture studies [26]. Here, we report an in vitro kinetic analysis of the antiviral activity of HCQ against the HCoV-229E strain. We would like to point out here that such a study has not been carried out so far.

The cytotoxicity of HCQ in Vero E6 cells was measured in advance for the purposes of determining the concentrations that would not cause injury or death to the treated cells (data not shown). This experiment was conducted three times and the results obtained are shown in Figure 8. To gain an initial insight into the stages of the viral replication cycle, at which HCQ is likely to exert its antiviral activity, time-of-drug-addition assays, such as SPR and MTT, were elaborated.

**Figure 8.** Kinetics of HCQ activity: dotted line—SPR results; solid line—MTT results.

The experiment involved Vero E6 cells (3 × <sup>10</sup><sup>4</sup> cells/mL) infected with HCoV-229E (MOI = 0.1) and treated with an HCQ standard (Sigma-Aldrich) at a concentration of 1 mg/mL (maximum non-toxic concentration). The replication cycle of HCoV-229E has demonstrated rapid viral propagation inside host cells, reaching maximum levels at 24 h *p.i.* [27]. This was confirmed by the AFM scans performed in our study—Figure 7, as well.

The inhibition of post-translational glycosylation with subsequent reduction in SARS-CoV-2 binding to and fusion with the angiotensin-converting enzyme 2 (ACE2) receptor of the host cell is an important antiviral effect of HCQ used in the treatment of SARS-CoV-2 infections [28]. Cleavage of SARS-CoV-2 spike (S) proteins by HCQ in the autophagosomes has also been reported [29]. Thus, the highest antiviral activity has to be expected at the stage of virus attachment to cells.

Indeed, our MTT assay confirmed the same mechanism of action in HCoV-229E: HCQ inhibited its replication in Vero E6 cells until 18 h *p.i.* A maximum antiviral activity was observed around 18 h—Figure 8. This coincided with the local maximum of virus control established by SPR (Figure 5), which confirmed that HCQ effectively inhibited virus replication at the stage of its attachment to the membrane and penetration into the cells.

The SPR study showed that maximum antiviral activity was reached around 12 h. This is also the stage of the virus replication cycle corresponding to the attachment to the membrane and penetration into the cells. The temporal shift against the MMT results could be explained by the method of compensation of SPR signals generated by the processes of cell growth and virus growth, as expanded in Section 2. The compensation procedure accumulates the error of surface plasmon resonance measurements (about 1.5 nm) and influences the peak position. However, the SPR signal is well pronounced at the expense of decreased accuracy.

#### **4. Conclusions**

A cell-based SPR assay was used to study cell growth, virus growth kinetics, and hydroxychloroquine antiviral kinetics. The MTT method was used as a reference since it is widely adopted for assessing cell metabolic activity. The kinetics revealed by the cell-based SPR assay was consistent with the findings of the MTT assay. Although the principles of the SPR and MTT methods are very different, the results obtained were very similar. All that showed that cell-based SPR is a reliable tool in investigating virus–host cell kinetics and antiviral drug activity. As expected, we found that the SPR assay provides better time resolution than MTT.

To the best of our knowledge, the present study is the first one focusing on the inhibiting effect of HCQ on the HCoV-229E virus. Both the SPR and MTT assay revealed that the antiviral efficiency is highest at the first stages of infection.

However, a few points have to be considered for the correct SPR measurement. First, the cells have to be seeded on the SPR slides with almost uniform density, which would ensure a reliable SPR signal across the slide. Additionally, the cell density has to prevent light scattering so that the reflection from the grating can be detected reliably. In addition, the method for compensating the signals generated by the processes of cell and virus growth has to be carefully considered.

There is a significant limitation to the cell-based SPR assay in investigating virus–host kinetics—it cannot be applied for a period lasting more than two virus replication cycles. After this period, the SPR signal is masked by destruction of the cell monolayer, detachment from the grating surface, and presence of remnants of the destroyed cells on the surface. As a result, the SPR signal is not unambiguously defined by the virus–host interaction.

In the present research, we showed that the cell-based SPR assay is applicable for in vitro studies. However, an extension of the SPR assay for in vivo application is a matter of engineering solutions.

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

**Funding:** This research was funded by Bulgarian National Science Fund, grand number PN33/1, and by the Bulgarian Ministry of Education and Science, grant number D01-392.

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article.

**Acknowledgments:** The authors thank Veneta Koceva, Institute of Optical Materials and Technology, for SPR slides elaboration.

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

#### **References**


## *Communication* **Surface Plasmon Resonance for Protease Detection by Integration of Homogeneous Reaction**

**Ning Xia 1, Gang Liu 1,2 and Xinyao Yi 3,\***


**Abstract:** The heterogeneous assays of proteases usually require the immobilization of peptide substrates on the solid surface for enzymatic hydrolysis reactions. However, immobilization of peptides on the solid surface may cause a steric hindrance to prevent the interaction between the substrate and the active center of protease, thus limiting the enzymatic cleavage of the peptide. In this work, we reported a heterogeneous surface plasmon resonance (SPR) method for protease detection by integration of homogeneous reaction. The sensitivity was enhanced by the signal amplification of streptavidin (SA)-conjugated immunoglobulin G (SA-IgG). Caspase-3 (Cas-3) was determined as the model. A peptide labeled with two biotin tags at the N- and C-terminals (bio-GDEVDGK-bio) was used as the substrate. In the absence of Cas-3, the substrate peptide was captured by neutravidin (NA)-covered SPR chip to facilitate the attachment of SA-IgG by the avidinbiotin interaction. However, once the peptide substrate was digested by Cas-3 in the aqueous phase, the products of bio-GDEVD and GK-bio would compete with the substrate to bond NA on the chip surface, thus limiting the attachment of SA-IgG. The method integrated the advantages of both heterogeneous and homogeneous assays and has been used to determine Cas-3 inhibitor and evaluate cell apoptosis with satisfactory results.

**Keywords:** surface plasmon resonance; protease; caspase; avidin-biotin interaction

#### **1. Introduction**

Proteases play an important role in a wide variety of biological processes, including protein digestion, wound healing, apoptosis, fertilization, growth differentiation, and immune system activation [1]. In the human body, at least 1.7% of human genes are encoded by proteases. The activities of proteases are closely related to many diseases, such as cancer, cardiovascular disease, Alzheimer's disease, human immunodeficiency virus (HIV), thrombosis, and diabetes [2]. Thus, extensive efforts have been made to screen protease inhibitors as potential drugs. This provides a powerful motivation for the development of sensitive, selective, and robust methods to detect protease and discover potential inhibitors.

Until now, many homogeneous and heterogeneous biosensors have been reported for the detection of proteases and screening of their inhibitors [2,3]. In homogeneous analysis, the substrate and protease sample are present in the aqueous phase. For instance, in the fluorescence resonance energy transfer (FRET) assay, the commonly used method for protease activity detection, the peptides labeled with two different fluorophores at two ends are digested by protease in the aqueous phase [4]. The activity of protease can be measured by monitoring the change of fluorescence signal after the cleavage of the peptide. On the contrary, the peptide substrate is anchored on a solid surface in the heterogeneous assay, and the enzymatic reaction happens at the solid-liquid interface [5,6]. Both the homogeneous and heterogeneous methods have their own advantages and disadvantages. Usually,

**Citation:** Xia, N.; Liu, G.; Yi, X. Surface Plasmon Resonance for Protease Detection by Integration of Homogeneous Reaction. *Biosensors* **2021**, *11*, 362. https://doi.org/ 10.3390/bios11100362

Received: 7 September 2021 Accepted: 27 September 2021 Published: 29 September 2021

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

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

homogeneous biosensors have the advantages of easy operation, rapid response, excellent sensitivity, and high throughput, but they show poor anti-interference ability and require large sample volumes and complex sample handling procedures. Conversely, heterogeneous assays exhibit the advantages of less sample consumption, ultra-high sensitivity and selectivity, and low instrument investment. Overall, the heterogeneous biosensors provide tremendous advantages over conventional homogeneous assays since numerous peptide substrates are immobilized at a discrete location on the solid interface [2]. However, immobilization of peptides on the solid surface will cause a steric hindrance to prevent the interaction between the substrate and the active center of protease [7], thus limiting the enzymatic cleavage of the peptide. Although the steric hindrance can be reduced by the use of nanomaterials-modified interface and the well-design of peptide substrate [8–10], the surface chemistry and coverage of peptide on the solid surface demands laborious optimization. Therefore, it is of importance to integrate the advantages of both heterogeneous and homogeneous assays for the design of general protease biosensors.

Surface plasmon resonance (SPR) is a simple, label-free technology to monitor the protein-protein interactions by measuring the refractive index change at the sensor surface [11–14]. The technology can be used to monitor the cleavage of protein or peptide fixed on the chip surface, providing a label-free detection method for protease analysis due to the advantages of fast response, real-time detection, high signal-to-noise, and good compatibility with the microfluidic system. For example, Steinrücke and co-workers suggested that cleavage of the helical protein with 78 amino acids by protease caused a detectable SPR signal [15]. However, the cleavage of low molecular weight peptides leads to a small, undetectable change in the refractive index [16–18]. Thus, it usually requires a signal amplification strategy to detect protease by labeling the peptide substrate with nanomaterials or specific groups [19–24]. Biotin is usually used to label peptide substrate for the design of heterogeneous biosensors. It can interact with avidin or its analogs of neutravidin (NA) and streptavidin (SA) with a binding coefficient as high as ~10<sup>15</sup> M−1. Such an interaction allows for the immobilization and recognition of peptide substrate at the solid-liquid interface [25–28]. By integrating the advantages of homogeneous assays, herein, we proposed a novel SPR method for protease detection by the signal amplification of SA-conjugated immunoglobulin G (SA-IgG). Caspases, a family of cysteine proteases, play an important role in apoptosis. To demonstrate the feasibility of the method, caspase-3 (Cas-3) that can specifically recognize and cleave the C-terminal of the peptide with the DEVD sequence was determined as the model. A peptide labeled with two biotin tags at the N- and C-terminals (bio-GDEVDGK-bio) was used as the substrate (Scheme 1). In the absence of Cas-3, the peptide substrate can be captured by the NA-covered chip through the avidin-biotin interaction (Channel 1). The biotin group at the other end of the peptide allows for the capture of SA-IgG, thus resulting in a strong SPR signal. When the peptide substrate was digested by Cas-3 in the aqueous phase, the biotinylated products (bio-GDEVD and GK-bio) would compete with the substrate to bond NA on the chip surface (Channel 2). This prevents the attachment of bio-GDEVDGK-bio and the follow-up capture of SA-IgG by the avidin-biotin interaction. However, when the activity of Cas-3 was suppressed by inhibitor, more bio-GDEVDGK-bio substrates would be anchored on the chip surface, which facilitating the capture of SA-IgG. The method was used to evaluate the inhibition efficiency of the inhibitor and monitor the activity of Cas-3 in cell lysates.

**Scheme 1.** Schematic representation of SPR method for the detection of Cas-3 using NA-covered gold chip. The signal was amplified by SA-IgG conjugates.

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

#### *2.1. Chemicals and Materials*

NA protein was purchased from Thermo Fisher Scientific (Shanghai, China). Cas-3 was obtained from New England BioLabs (Ipswich, MA, USA). Thrombin, beta-secretase, prostate-specific antigen (PSA), and bovine serum protein (BSA) were acquired from Sigma-Aldrich (Shanghai, China). SA-IgG and glutathione (GSH) were purchased from Sangon Biotech (Shanghai, China). Peptides were provided by China Peptide Co., Ltd. (Shanghai, China). Other reagents were ordered from Aladdin Reagent Co., Ltd. (Shanghai, China). All aqueous solutions were prepared daily with ultrapure water collected from a Milli-Q purification system.

#### *2.2. Preparation of SPR Chips*

The gold chips were annealed in a hydrogen flame to eliminate the surface contaminant. Then, the cleaned gold chips were incubated with 1 μM NA protein in carbonate buffer (pH 10) for 2 h. NA was capped on the gold surface through the hydrophobic and Au-S interactions [29]. The unbound NA proteins were removed by rinsing the chip with the carbonate buffer. The unreacted gold surface were blocked by incubation of the chip with 10 μM BSA and 100 μM GSH. Finally, the NA-covered chips were thoroughly washed with the carbonate buffer.

#### *2.3. Procedure for Cas-3 Detection*

The peptide bio-GDEVDGK-bio was digested by Cas-3 in the HEPES buffer with the optimized reaction conditions. The reaction mixture was delivered to the flow cell by a syringe pump. When the baseline is stable, SA-IgG in phosphate buffer (10 mM, pH 7.4) was injected into the SPR channel by the pump. The signal was collected by measuring the change of SPR dip shift on a BI-SPR 3000 system (Biosensing Instrument Inc., Tempe, AZ, USA).

#### *2.4. Inhibitor Detection and Cell Lysate Analysis*

For the detection of Cas-3 inhibitor, DEVD-FMK at different concentrations was mixed with a given concentration of Cas-3 for 10 min. Then, the resultant solution was incubated with the peptide substrate. For real sample assays, HeLa cells were cultured and the cell lysates were extracted with our reported procedures [30,31]. Then, the peptide substrate was incubated with the diluted cell lysates to react for 2 h. Finally, the reaction mixture was delivered to the flow cell, followed by injection of SA-IgG to the channel after the baseline stable was attained.

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

#### *3.1. Feasibility of the Strategy*

Based on the avidin-biotin interactions, NA or SA-modified magnetic beads, chromatography columns and solid surfaces have been widely used for the immobilization and separation of biotinylated biomolecules. In this work, an NA-covered gold chip was used to capture the biotinylated peptide. Figure 1 depicts the SPR responses when injecting SA-IgG, SA, and IgG to the sensor channel. A negligible change in the dip shift was observed after injecting SA-IgG conjugate to the NA-covered chip (curve a), demonstrating that SA-IgG showed no interaction with the sensor chip. Interestingly, the SPR dip shift reached 207 mD when injecting the conjugate to the chip treated by bio-GDEVDGK-bio (curve b). No significant change was observed when injecting IgG onto the bio-GDEVDGK-bio treated channel (curve c) and a smaller SPR dip shift (59 mD) was attained when injecting SA onto the channel (curve d). Thus, the change in curve b should be attributed to the avidinbiotin interaction and the signal was greatly amplified by IgG due to its large molecular weight (~150000 Da). We also found that the signal was intensified with the increase in bio-GDEVDGK-bioconcentration from 0.01 to 20 nM and began to level off beyond 5 nM. To attain higher sensitivity and a wider linear range, 5 nM bio-GDEVDGK-bio was used as the substrate for the assays of Cas-3.

**Figure 1.** SPR sensorgrams when injecting 0.1 mg/mL SA-IgG to the fluidic channel wherein the NA-covered chip had been treated by blank buffer (curve a) and 20 nM bio-GDEVDGK-bio (curve b). Curves c and d were acquired when injecting 0.1 mg/mL IgG and 0.05 mg/mL SA onto the NA-covered chip treated by bio-GDEVDGK-bio.

#### *3.2. Detection of Cas-3 and Its Inhibitor*

The analytical performances were first investigated by determining different concentrations of Cas-3. In Figure 2A, the SPR signal decreased gradually with the increase in Cas-3 concentration in the range of 0~2000 pg/mL. The plateau beyond 1000 pg/mL is indicative of the completion of the enzymatic hydrolysis (Figure 2B). The signal did not decrease to the background value, indicating that not all the substrate peptides were cleaved by Cas-3 even at a higher concentration with a very long reaction time. The detection limit was estimated to be 0.5 pg/mL by measuring the sensor response to a dilution series and determining the smallest target concentration at which the signal was clearly distinguishable from the response to the blank solution. The value is lower than

that attained by the homogeneous methods, such as fluorescence (128 pg/mL) [32], colorimetric assay (5 ng/mL) [33], differential pulse voltammetry (27.4 ng/mL) [34], and mass spectrometry (3.02 ng/mL) [35]. The value is comparable to or even lower than that achieved by heterogeneous methods based on the signal amplification of enzymes and nanomaterials (Table 1). The high sensitivity of the method should be attributed to the "immobilization-free" hydrolysis reaction and the large molecular weight of SA-IgG.

**Figure 2.** (**A**) SPR sensorgrams when injecting 0.02 mg/mL SA-IgG onto the NA-covered chips treated by the mixture of 5 nM bio-GDEVDGK-bio and a given concentration of Cas-3 (from top to bottom: 0, 0.5, 5, 50, 200, 500, 1000, and 2000 pg/mL). (**B**) Dependence of SPR signal on the concentration of Cas-3. The inset shows the effect of inhibitor concentration on the SPR signal wherein the concentration of Cas-3 was 500 pg/mL.



Abbreviations: DPV, differential pulse voltammetry; LSV, linear sweep voltammetry; SWV, square wave voltammetry; EIS, electrochemical impedance spectroscopy; ECL, electrochemiluminescence; GO, graphene oxide; rGO, reduced graphene oxide; AuNPs-MCM, gold nanoparticle-coated silica-based mesoporous materials; MB, magnetic bead; HRP, horseradish peroxidase; AgNPs, silver nanoparticles; FNP, self-assembled biotinphenylalanine nanoparticle.

As a proof-of-concept experimental for evaluation of Cas-3 activity, the inhibitor DEVD-FMK at different concentrations was incubated with 500 pg/mL Cas-3. The inset in Figure 2B shows the dependence of SPR dip shift on the concentration of inhibitor. The increase in inhibitor concentration induced the enhancement in the SPR signal, indicating that DEVD-FMK is an effective Cas-3 inhibitor. The half-maximal inhibitory concentration (IC50) was estimated to be 98 nM, which is in agreement with that measured by other methods [30,36]. Thus, the method has bright prospects for the screening of protease inhibitors.

#### *3.3. Selectivity*

To evaluate the specificity of the method, the method was first challenged by determining other proteases (e.g., thrombin, beta-secretase, and PSA) to replace Cas-3. As shown in Figure 3, the tested proteases did not induce a significant decrease in the SPR dip shift, suggesting that the method shows high selectivity toward Cas-3 (*cf.* curves 1~4). However, trace biotin or other molecules in real samples may interact with biotin, thereby limiting the practical application of the technique. For this consideration, the interferences from avidin and biotinylated biomolecule such as bio-GLRRASLG were examined. As envisaged, both avidin and bio-GLRRASLG caused a significant decrease in the SPR signal (curves 5~6). The result is understandable as avidin can bind to the peptide substrate (bio-GDEVDGK-bio) and the biotinylated peptide can compete with the substrate to bind NA on the chip surface, thus preventing the attachment of bio-GDEVDGK-bio on the chip. To resolve this problem, a certain amount of biotin was added to the sample in advance to eliminate the interference of avidin. The free biotin or biotinylated peptide was then removed by the commercial SA-modified magnetic beads. As a result, the interferences from avidin and biotinylated peptide have been well eliminated (curves 7~8).

**Figure 3.** Selectivity of the method: bar 1, thrombin; bar 2, beta-secretase; bar 3, PSA, bar 4, Cas-3; bar 5, avidin; bar 6, bio-GLRRASLG; bar 7, the mixture of avidin and biotin pretreated by SAmodified magnetic beads; bar 8, bio-GLRRASLG pretreated by SA-modified magnetic beads. The concentrations of thrombin, beta-secretase, PSA, Cas-3, avidin, biotin, and bio-GLRRASLG were 5 ng/mL, 5 ng/mL, 5 ng/mL, 500 pg/mL, 200 ng/mL, 12.5 nM, and 5 nM, respectively.

#### *3.4. Evaluation of Cell Apoptosis*

Apoptosis is a highly regulated physiological process, which is of great significance in the life cycle of organisms. However, the imbalance of apoptosis may directly lead to the occurrence of many diseases. Therefore, the death caused by apoptosis has attracted extensive attention from experts in pathology, pharmacology, and toxicology. Among various types of caspases, Cas-3 is the central molecule to mediate the apoptotic pathway inside and outside cells. Therefore, Cas-3 has been regarded as the biomarker and therapeutic target for the diagnosis and treatment of apoptosis-related diseases. To verify the feasibility of this method for monitoring cell apoptosis, HeLa cells were used as the models. As shown in Figure 4A, when the peptide substrates were incubated with the cell lysates extracted from normal HeLa cells, the SPR signals were high and no significant changes were observed with the increase in cell number. However, when the cells were treated by STS (a common apoptosis inducer), the SPR signals decreased gradually with the increase in cell number. This indicated that the apoptosis was triggered by STS and

the activity of Cas-3 was activated during apoptosis. STS-induced apoptosis was also confirmed by characterizing the cell morphology with a microscope (Figure 4B). The result is consistent with that obtained by other methods, indicating that the method can be used for the evaluation of apoptosis by monitoring the Cas-3 activity.

**Figure 4.** (**A**) Dependence of SPR signal on the concentration of normal and STS-treated HeLa cells. (**B**) Confocal images of normal and STS-treated HeLa cells.

#### **4. Conclusions**

In summary, we reported a heterogeneous SPR method for protease detection by integration of homogeneous enzymatic hydrolysis reaction. The signal was amplified by SA-IgG because of its large molecular weight. The method was used to determine Cas-3 activity and evaluate cell apoptosis with satisfactory results. The method exhibited high sensitivity and obviated the use of enzymes or nanomaterial for signal amplification. The "immobilization-free" strategy for the enzymatic reaction should be valuable for the design of novel heterogeneous biosensors to eliminate the effect of steric hindrance.

**Author Contributions:** Conceptualization, G.L. and N.X.; methodology, N.X.; investigation, G.L.; writing—original draft preparation, N.X.; writing—review and editing, X.Y.; project administration, N.X.; funding acquisition, X.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Program for Innovative Research Team of Science and Technology in the University of Henan Province (21IRTSTHN005), and the National Natural Science Foundation of China (21705166).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

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

#### **References**


## *Article* **Effect of Graphene vs. Reduced Graphene Oxide in Gold Nanoparticles for Optical Biosensors—A Comparative Study**

**Ana P. G. Carvalho 1,\*, Elisabete C. B. A. Alegria 1,2, Alessandro Fantoni 3,4, Ana M. Ferraria 5,6, Ana M. Botelho do Rego 5,6 and Ana P. C. Ribeiro <sup>2</sup>**


**Abstract:** Aiming to develop a nanoparticle-based optical biosensor using gold nanoparticles (AuNPs) synthesized using green methods and supported by carbon-based nanomaterials, we studied the role of carbon derivatives in promoting AuNPs localized surface plasmon resonance (LSPR), as well as their morphology, dispersion, and stability. Carbon derivatives are expected to work as immobilization platforms for AuNPs, improving their analytical performance. Gold nanoparticles (AuNPs) were prepared using an eco-friendly approach in a single step by reduction of HAuCl4·3H2O using phytochemicals (from tea) which act as both reducing and capping agents. UV–Vis spectroscopy, transmission electron microscopy (TEM), zeta potential (ζ-potential), and X-ray photoelectron spectroscopy (XPS) were used to characterize the AuNPs and nanocomposites. The addition of reduced graphene oxide (rGO) resulted in greater dispersion of AuNPs on the rGO surface compared with carbon-based nanomaterials used as a support. Differences in morphology due to the nature of the carbon support were observed and are discussed here. AuNPs/rGO seem to be the most promising candidates for the development of LSPR biosensors among the three composites we studied (AuNPs/G, AuNPs/GO, and AuNPs/rGO). Simulations based on the Mie scattering theory have been used to outline the effect of the phytochemicals on LSPR, showing that when the presence of the residuals is limited to the formation of a thin capping layer, the quality of the plasmonic resonance is not affected. A further discussion of the application framework is presented.

**Keywords:** biosensors; AuNPs; metal–graphene hybrid; simulations; Mie theory

#### **1. Introduction**

Plasmonic biosensors are widely explored as promising sensing tools due to their low cost, simplicity, and short response time. These devices can be useful in a variety of situations, such as medical emergencies in more isolated populations or/and with poor access to health services [1]. For these applications, plasmonic structures can be integrated in point-of-care systems with optical, electrical, or thermal signals delivered in response to certain stimuli, mediated by biomolecules immobilized on biosensor surfaces biological recognition elements (BREs)—thereby allowing the selective detection of analytes of interest [2].

**Citation:** Carvalho, A.P.G.; Alegria, E.C.B.A.; Fantoni, A.; Ferraria, A.M.; do Rego, A.M.B.; Ribeiro, A.P.C. Effect of Graphene vs. Reduced Graphene Oxide in Gold Nanoparticles for Optical Biosensors—A Comparative Study. *Biosensors* **2022**, *12*, 163. https:// doi.org/10.3390/bios12030163

Received: 22 January 2022 Accepted: 2 March 2022 Published: 4 March 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/).

Gold nanoparticles (AuNPs) have been widely used in the development of optical biosensors [3]. Their utilization in optical signal analysis is based on their characteristic surface plasmon resonance (SPR) effects. No additional material is required for the generation of surface plasmons after interaction with light [4], making their use advantageous in label-free sensing devices.

When AuNPs are used as optical transducers, the region of confinement in which the evanescent wave is propagated is smaller than the wavelength of the incident light. In this case, the phenomenon is called localized surface plasmon resonance (LSPR), where the collective oscillation of the electrons originates a dipolar or multipolar moment in the nanoparticle. LSPR is excited at a specific light wavelength, which is determined by the morphology, size, and composition of the nanoparticle [5]. Any modification on the surface of AuNPs affects the behavior of the LSPR, allowing the exploration of this phenomenon in the development of biosensors through the functionalization of its surface with biomolecules.

When the analyte of interest is recognized by the BRE, the surface refractive index of the AuNPs changes, promoting a shift in the LSPR wavelength [4].

The interest in AuNPs for the development of biomedical diagnostic devices is related to their sensitivity and the possibility of controlling and optimizing the limit of detection (LOD) [5].

Moreover, these interesting properties may act synergistically with graphene when this carbon nanomaterial is added to form a nanocomposite [6]. For example, Banerjee [7] highlighted the efficiency of nanocomposites formed with metallic nanoparticles and graphene in biomedical applications as compared to conventional materials. This greater efficiency is due to their small size and high surface-to-volume ratio, which improves their responses to external stimuli. Carbon allotropes, such as graphene (G), graphene oxide (GO), and reduced graphene oxide (rGO), have similar optical, electronic, and electrochemical properties, and have therefore been investigated as potential materials for biosensor design, drug delivery, bioimaging, and tissue engineering applications. The unique properties of graphene derivatives are also related to their 2D dimensionality, high electrical conductivity and thermal properties, malleability, and functionalization potential. The properties of graphene, which is composed of a single 2D sheet of carbon atoms forming vertices of hexagons with covalent bonds and *sp2* hybridization [7], can be manipulated by chemical modification either through oxidation (from G to GO) or reduction (from GO to rGO) reactions, increasing its functionalization capabilities and broadening its range of applications.

The oxidation of graphene, yielding graphene oxide (GO), allows the addition of oxygen functional groups to its surface, reducing electrical conductivity and malleability, and increasing solubility in polar solvents, as well as reducing the aggregation of carbonbased nanomaterials in aqueous solutions.

Graphene oxide shows a greater functionalization capacity due to the presence of a larger number of oxygen groups, but for the same reason there is also restriction of the mobility of electrons on its surface. Reduction of the number of oxygen groups, yielding reduced graphene oxide, allows the restructuring of electrical and thermal conductivity towards pristine graphene, maintaining functionalization capacity and the distance between graphene sheets [7]. The maintenance of this distance is mandatory to prevent agglomeration between the sheets and avoid the graphite state, since this material has different properties to graphite-derivative nanomaterials [8]. The AuNPs act synergistically with these nanoscale carbon-based nanomaterials, not only in optical properties but also in the dispersion of both nanoparticles and graphene sheets. Studies report that nanoparticles can function as nano-spacers of these materials [9], enhancing AuNPs' stability after the addition of rGO.

In this work, the influence of a variety of carbon-based materials on gold nanoparticles' (AuNPs) localized surface plasmon resonance (LSPR) has been studied and, considering environmental concerns, an ecofriendly approach in the synthesis of gold nanoparticles

was chosen, using tea extract as a reducing/capping agent. As a matter of fact, the phytochemicals present in tea, such as polyphenols and tannins, have the interesting capacity of reducing metallic salts, such as chloroauric acid (HAuCl4) [10–12].

Although the use of composite materials based on metal nanoparticles and graphene allotropes to explore plasmonic properties in sensing applications has already been reported in the literature [13,14], a green method of synthesis for nanoparticle fabrication represents a novel approach in the preparation of hybrid materials. Since a residual layer of polyphenols remains on the surface of the nanoparticles, this approach results in the introduction of an additional factor in the set of parameters necessary to obtain the desired LSPR tuning which deserves to be investigated. In this paper, the frequency and intensity of the LSPRs for three different types of graphene are analyzed, compared, and discussed, combined with a green protocol for the synthesis of gold nanoparticles as well as different sequences in the fabrication of the composites.

#### *Application Framework*

Sensors based on the local surface plasmon resonance of metal nanoparticles are characterized by simple structures and good sensitivities. Despite the good sensing properties of LSPR structures, full commercialization has been prevented by the production costs associated with the bio-functionalization and the high-precision systems necessary to extract the optoelectronic output. There is a great interest in new strategies for bringing the excellent detection properties of LSPR sensors into play in low-cost devices made with low-cost materials [15]. The combination of carbon-based nanomaterials (CNMs) with MNPs has been demonstrated to enhance LSPR response [16] and facilitate functionalization with specific and selective antibodies [17]. In addition, the introduction of CNMs in the plasmonic layer allows a tuning of the LSPR central frequency. With the double dependence of the LSPR on MNP size and the presence of CNMs, it is possible to create a set of plasmonic layers whose LSPR wavelengths are distributed in a spectral range of a few tenths of a nanometer. This consideration paves the way to an LSPR sensor with an arrayed structure, where each element maximizes its specific LSPR at its own wavelength. Illumination with a broad light source will produce a different response in each one of the elements and the biomarkers' immobilization in the surrounding medium will cause a transition to a different state. In such a configuration, the output can be extracted by the application of an image analysis approach based on a color-recognition algorithm [18]. The experimental characterization presented in this work represents a first step toward the development of an arrayed LSPR sensor whose elements are composed by metal nanoparticles with different dimensions and supported by different CNMs, joined to a reading scheme provided by a CCD imager, supported by an image processing algorithm. The use of low-cost materials together with a simplified interrogation scheme aims to overcome the elevated costs related to highprecision mechanical systems and wavelength selective light sources. The result will be a proof of concept for a low-cost LSPR sensor with potential for large-scale biosensing applications in environmental monitoring or the medical stratification of diseases.

The main advantages of these nanomaterials are their biocompatibility, high chemical stability, and high surface areas. The reduced dimensions and weight, high resistance, simplicity of use, and low cost makes graphene and its derivatives excellent candidates for the development of biosensors.

An optical sensor has, by definition, the ability to convert an external stimulus into an optical output signal.

The π–π binding capacity of graphene with biomolecules allows the use of this carbonderived nanomaterial as a substrate, and the SPR technique can be used to detect the interaction with the analyte of interest, in which the angle of an incident polarized light is adjusted as a result of the change in the refractive index caused by the interaction between them. rGO has advantages over graphene, since its oxygen functional groups improve the interaction of this nanomaterial with biomolecules [19].

Optical sensors, such as those based on fluorescence, surface-enhanced Raman scattering (SERS), optical fiber biological sensors, among other kinds of optical sensors, are being developed due to the amazing properties of graphene and its derivatives. Various sensing applications, such as single-cell detection, cancer diagnosis, and protein and DNA sensing, have been reported in recent years [19].

Due to its unique optical and electrical properties, graphene and its derivatives are widely used in photonic and optoelectronic devices as they have displayed several ideal properties, including broadband light absorption, the ability to quench fluorescence, excellent biocompatibility, and strong polarization-dependent effects, making them very popular platforms for optical sensors. Graphene and its derivatives-based optical sensors have numerous advantages, such as high sensitivity, low-cost, fast response time, and small dimensions. The use of metallic nanoparticles, namely, gold nanoparticles, may improve the response of biosensors, amplifying the signals obtained and increasing the sensitivity of these devices [19].

Several applications of graphene and its derivatives-based optical sensors are summarized in the following Table 1:


**Table 1.** Several applications of graphene and its derivatives-based optical sensors.

It is undeniable that there are still many challenges in this area. The need to synthesize high-quality graphene, to achieve a low-cost, environmentally friendly method for synthesizing graphene and its derivatives are issues still to be addressed. With this study, the use of a simple green method to produce light-responsive material is our aim and contribution.

Addressing the above challenges, we hope to show the potential of green methods as well as the importance of graphene and its derivatives in the development of optical sensing technologies, which will ultimately increase the quality of future life.

#### **2. Results and Discussion**

#### *2.1. Sequence 1—Addition of Carbon-Based Derivatives to AuNPs (SQ1)*

After the addition of chloroauric acid (HAuCl4·3H2O) to a tea extract (5% *w*/*w*), a change of color from yellow to red was observed, which is consistent with the formation of AuNPs. The AuNPs' characteristic LSPR was observed only after one week (t1w), the absence of LSPR at the initial time (t0) possibly being associated with the period of reduction required to originate AuNPs or due to the fact that in the case of hybrid nanostructures, carbon materials may be responsible for radiation absorption, inhibiting the detection of AuNPs, as described by Biris et al. [37]. Figure 1 shows the UV–Vis spectrum of the aqueous solution containing AuNPs with the corresponding LSPR occurring at 556 nm. The presence of a single band suggests a spherical morphology for the synthesized AuNPs [37]. The solution remained stable for several weeks without any signal of aggregation.

**Figure 1.** LSPR of AuNP samples synthesized with 5% tea extract (*Thea sinensis*) at t1w (after 1 week) and t2w (after 2 weeks).

The stability of the synthesized nanoparticles was evaluated. We found that between the first (t1w) and second week (t2w) after synthesis no significant differences in LSPR response was observed, confirming that polyphenols guarantee a certain stability to the synthesized AuNPs. As proven previously by the authors of [10], polyphenols prevailed as reducing agents of the metallic salt and capping agents of the produced AuNPs [10], maintaining their capacity for at least two weeks after the synthesis. This is due to the ionic force of polyphenols that promote good dispersion by efficiently counteracting the mutual attraction caused by Van der Waals forces and which are responsible for the aggregation of nanoparticles [38].

Figure 2 reports the LSPR of AuNPs synthesized with 5% tea extract after the addition of G, GO, and rGO. For the naked AuNPs (5%\_AuNPs), the LSPR at t1w occurred at 556 nm, with a shift to higher frequencies for the mixtures containing graphene (G) (LSPR = 541 nm) and reduced graphene oxide (rGO) (LSPR = 542 nm). The composite containing graphene oxide (AuNPs/GO) did not show any LSPR shift after the addition of carbon-based nanomaterials. For the AuNPs/G hybrid nanostructure, band amplitude decreased by ~10% when compared to the sample without any of the carbon-based nanomaterial, while for AuNPs to AuNPs/rGO, it increased by ~0.29, which corresponds to 46% of the naked/unsupported AuNPs' absorbance. This result may be ascribed to a dispersion of the AuNPs [38] in the presence of the polyphenols [37].

**Figure 2.** LSPR of AuNPs synthesized with 5% tea (*w*/*w*) extract and after the addition of G, GO, and rGO, one week after their synthesis (t1w).

LSPR properties depend not only on the AuNPs' shape, size, and dispersion but also on the surrounding dielectric constant. TEM characterization showed the formation of spherical AuNPs (Figure 3). Although the addition of rGO caused a LSPR shift to lower wavelengths, the AuNPs supported on rGO exhibited a larger diameter (61 nm ± 3 nm) when compared to AuNPs only encapsulated by polyphenols (30 nm ± 2 nm). The shift to lower wavelengths after the carbon-based material addition may be related to the increased dispersion of AuNPs, with no change in the diameter of the nanoparticles, and to charge transfer interactions between graphene and AuNPs [39]. By acting as nano-spacers [6], when graphene is used as a support, the nanoparticles benefit from increased dispersion, while at the same time they avoid graphene sheet agglomeration. Additionally, the LSPR deviation can also be caused by the surface energy of neighboring nanoparticles [40].

**Figure 3.** TEM images. (**a**) AuNPs in 5% tea extract. (**b**) rGO-supported AuNPs in 5% tea extract.

The probable formation of π–π bonds between AuNPs and the functional oxygen groups of rGO [6] may be responsible for the stable dispersion confirmed by the measured zeta potential value −20.17 mV when compared to −15.59 mV obtained for the sample containing AuNPs dispersed in 5% tea extract (Table 2). Since LSPR did not occur at t0, we assumed that the nucleation of metallic salts occurred near the rGO functional oxygen groups [6] which may have promoted a higher stability due to the dispersion of the AuNPs as well as the increase in the diameter of the nanoparticles [41].


**Table 2.** Zeta potential values of AuNPs synthesized before and after rGO addition.

The reason for using a pre-prepared solution of AuNPs stabilized with polyphenols then mixed with rGO is to help anchor the polyphenol-protected AuNPs to the rGO surface. This process could be assisted by: (i) *π*–*π* interactions between AuNPs and functional oxygen groups of rGO; (ii) *π*–*π* interactions between benzene rings of the polyphenols and the surface of rGO; and (iii) electrostatic interactions between the OH on the surface of rGO and the polyphenols capping the AuNPs. After mixing, the shape of the AuNPs appeared to be the same when observed by TEM (Figure 3), but the presence of the rGO seemed to have anchored them, as expected. Schematically (Figure 4), the introduction of rGO leads to the settling of the AuNPs [42].

**Figure 4.** Schematic illustration of AuNPs anchored on the surface of reduced graphene oxide (rGO).

In the case where GO is added to the AuNPs, no significant deviation in the LSPR occurs relative to the value observed prior to the addition of this carbon-based material (λ = 556 nm). TEM images (Figure 3) show a greater dispersion of AuNPs on the surface of the carbon-based nanomaterial which is probably responsible for the increase in the LSPR band amplitude observed by UV–Vis [38,43,44]. As with the addition of rGO, nucleation may have occurred next to the functional oxygen groups [41]. To confirm this hypothesis, Figure 5 shows dispersed nanoparticle clusters on the GO surface. The wide number of oxygen functional groups on the GO may have promoted AuNP agglomeration and significant dispersion of diameters. These results are in accordance with those reported by Parnianchi et al. [45] related to the difficulty of controlling both the morphology and the homogeneous distribution of AuNPs when GO is present. The frequency versus size distributions for the three TEM images are presented in ESI (Figure S1).

**Figure 5.** TEM image. GO supported on AuNPs synthesized with 5% tea extract.

After the addition of both GO and rGO, sharper resonance bands as well as an increase in the resonance intensity or amplitude were observed (Figure 2). In fact, both enhancements of plasmonic resonance intensity and shift in the position are indicators of enhanced sensitivity of the LSPR sensor.

The partial reduction of oxygen groups in GO driven by tea polyphenols [46] may contribute to reducing the availability of these phytochemicals both as reduction and capping agents. The dispersion of diameters verified after GO addition and the fact that this nanomaterial is insulating (due to the functional groups of oxygen, the electronic mobility is reduced) compromises the use of this nanocomposite in the development of biosensors.

We also measured the electrostatic potential at the electrical double layer surrounding a nanoparticle in solution, commonly referred to as the zeta potential. Nanoparticles with a zeta potential between −10 and +10 mV are considered approximately neutral, while nanoparticles with zeta potentials of greater than +30 mV or less than −30 mV are considered strongly cationic and strongly anionic, respectively [47]. Since most cellular membranes are negatively charged, zeta potential can affect the nanoparticle tendency to permeate membranes, with cationic particles generally displaying more toxicity associated with cell wall disruption. It can be observed that the initial 5%\_AuNPs are weakly anionic. With the addition of rGO, they became strongly anionic due, most likely, to the increase in acidity by the introduction of rGO.

According to the classical physical theory for the electromagnetics of metals, the plasma frequency (ω*p*) of the free electron gas depends linearly on the density of the electrons (Ne). Such a consideration leads, in the Drude model, to a linear dependence of the metal dielectric function on the value of Ne. Within the dipole approximation, i.e., when the nanoparticle size is much smaller than the light wavelength, and under the Fröhlich condition, it can be shown that the LSPR frequency (ω*LSPR*) can be directly related to ω*<sup>p</sup>* by the following equation:

$$
\omega\_{LSPR} = \frac{\omega\_p}{\sqrt{1 + 2\varepsilon\_m}}.
$$

where *ε<sup>m</sup>* is the dielectric constant of the surrounding medium [48]. Therefore, a change in the electron density is expected, leading to a blue or red shift of the localized plasmon resonance. This effect has been reported in the literature, supported by electrochemical experiments [49] and exploited for sensing applications [38]. Optical gas sensors based on gold nanoparticles and carbon nanomaterials have also been demonstrated to rely on the reactions for both reducing and oxidizing gases and the correspondent injection or subtraction of electrons to and from graphene oxide, implying a shift in the observed LSPR [50]. The charge transfer interaction between AuNPs and graphene has been also demonstrated to be suitable for active modulation of surface plasmon resonance. These considerations are in agreement with our experimental findings, namely, the blue shift observed in the LSPR of the composites (Figure 2) and the negative charge transfer between the graphene allotropes and the nanoparticles observed in the zeta potential measurements (Table 2).

The degree of oxidation of graphene before and after the addition of AuNPs, the type of oxygen functional groups, the oxidation state of gold at AuNP surfaces within the AuNPs/G, AuNPs/GO, or AuNPs/rGO composites, as well as relevant relative atomic amounts were assessed by XPS. The characterization of AuNPs prior to any graphene addition has already been reported [10]. Here, the three different types of graphene and the corresponding gold composites are studied.

XPS confirms that samples are composed mainly of carbon and oxygen, and, where expected, gold. GO-based samples also contain some sulphur, and some samples have a residual or low relative amount of silicon. Table 3 shows the corrected binding energies (BE) of the peaks fitted in the different XPS regions: C 1s, O 1s, Au 4f, S 2p, and Si 2p.


**Table 3.** Corrected BE ± 0.1 eV and corresponding assignments.

( 1) see text.

C 1s regions of AuNPs/rGO and AuNPs/G are dominated by the peak assigned to carbon atoms in C-C and C-H sp2 bonds, centred at 284.4 ± 0.1 eV, and by a long tail at the high BE side, detected roughly between 287 eV and 297 eV (Figure 6a–c), corresponding to π–π\* excitations, typical of extended delocalized systems, such as graphene. In addition, at BE > 297 eV plasmon loss features are detected. Included in the C 1s envelope, peaks attributed to carbon atoms bonded to oxygen in different functional groups are identified in Table 3. The latter are particularly intense in GO and AuNPs/GO, where the loss of electron delocalization due to the oxidation of graphene is evident: the sp2 carbon peak has a lower relative intensity compared with AuNPs/rGO and AuNPs/G, and the energy loss features are hardly detected (Figure 6d). In GO, a peak at 284.4 eV was fitted, but it is almost cancelled with the fitting, leaving only the peak centred at 285 eV (Figure 6d).

Au 4f regions are doublet peaks (Figure 7) with a spin–orbit energy separation of 3.7 eV. Au 4f7/2 is centred at 84.1 ± 0.1 eV in AuNPs/rGO and AuNPs/G, which is attributed to Au0. In AuNPs/GO, Au 4f7/2 is centred at 84.6 ± 0.1 eV, which has been identified as Au+ in HAuCl [53]. However, in this case, no chlorine was detected to corroborate this assignment. Still, positive BE shifts for Au 4f photoelectrons have been reported for very small gold nanoparticles (diameter ≤20 nm) [54]. In addition, a contact potential effect between the metal nanoparticles and the organic substrate may be present, leading to an underestimation of the charge shift. It is noteworthy that the full widths at half maximum, for all Au 4f fitted peaks, are very similar to each other (1.1 ± 0.1 eV), which also sustains the hypothesis of having reduced gold nanoparticles in all the samples. Finally, sulphur is only present in GO-based samples. S 2p is a doublet with a spin–orbit split of 1.1 ± 0.1 eV, and the main component, S 2p3/2, centred at 168.8 ± 0.2 eV, is assigned to sulphate groups. The silicon detected in some of the samples may come from the polysiloxane-based tape used to mount the powder for XPS analysis. Table 4 shows the XPS quantitative analysis.

**Figure 6.** C 1s regions of (**a**) AuNPs/G and G; (**b**) AuNPs/rGO and rGO; (**c**) G with fitting (similar to rGO and AuNPs/rGO and AuNPs/G); and (**d**) GO and AuNPs/GO.

**Figure 7.** Au 4f XPS regions.


**Table 4.** XPS atomic concentrations (%) and relevant atomic ratios.

It is interesting to note that the larger relative amount of Au in AuNPs/rGO compared with AuNPs/G, computed from XPS data, is compatible with the UV–Vis absorbance spectra shown in Figure 2: AuNP LSPR absorbance is much larger for AuNPs/rGO than for AuNPs/G. Moreover, the relative amount of Au, detected by XPS, is much lower in AuNPs/GO than in AuNPs/rGO. Actually, as attested by Table 3 and Figure 6, graphene oxide has a very different chemical composition from G or rGO, with many more oxygen functional groups than G or rGO. These oxidized carbonaceous groups establish stronger intermolecular interactions with polyphenols surrounding the AuNPs, allowing for the formation of sandwich-like GO/AuNPs/GO, significantly attenuating the Au 4f photoelectron signal detected by XPS. It is also clear from the quantification results that samples with gold nanoparticles are slightly more oxidized than the parent samples with no gold. Actually, since AuNPs are capped with phenolic functional groups, a larger relative amount of oxygen is expected in the samples with AuNPs. In addition, a further reduction of gold may occur when in contact with G, GO, or rGO with the simultaneous oxidation of graphene. In Table 4, the ratios computed were obtained after subtracting the contribution

of sulphate and silicone, these being, exclusively, the ratios in the graphene-based samples discarding the contaminations. Other effects of the introduction of Au nanoparticles can be found in the C 1s spectral differences presented in ESI (Figure S2).

#### *2.2. Sequence 2—Addition of Carbon-Based Derivatives Prior to AuNP Formation (SQ2)*

The addition of carbon-based nanomaterials prior to AuNP formation seems to compromise phytochemicals' reducing and capping capacities. This may be related to the adsorption of phytochemicals by carbon-based nanomaterials [6] which compromise their availability to reduce the metallic salt. Some of these samples revealed a slight LSPR at t0 (Figures S4–S6, Supplementary Materials). We believe that the addition of these derivatives initially increases the contact between the metallic salt and polyphenols, promoting the synthesis of AuNPs. However, at t1w and t2w the LSPR did not occur. We believe that the adsorption of polyphenols by carbon-based nanomaterials through electrostatic bonds and Van der Waals interactions [6] compromises the efficiency of these species as capping agents. In this case, we did not verify stability between t1w and t2w in any of the samples. No further testing was performed.

#### *2.3. AuNP Stability Study*

The stability of the synthetized nanoparticles was also evaluated by the sample's characterization, 1 week (t1w) and 2 weeks (t2w) after synthesis. Figure 8 reports the absorbance spectra for the different composites taken at time intervals of one week. We found that the resonance of the surface plasma had a red shift one week after synthesis (t1w vs. t0). These data show that the synthesis of the nanoparticles did not cease after 20 min of vigorous stirring of chloroauric acid with the tea extract. This shift could mean that the diameters of the AuNPs increased between t0 and t1w. As mentioned above, the increased band amplitude after a week suggests that there was an increase in the concentration of nanoparticles, indicating that the formation of AuNPs was still happening [44].

**Figure 8.** AuNP samples stability (t1w–t2w). Carbon-based nanomaterials added after AuNP formation.

UV–Vis analysis of the samples after 2 weeks (t2w) showed that LSPR did not shift in comparison to that observed after 1 week (t1w) (Figure 8). The polyphenols acted as capping agents [10] of the AuNPs contrary to the Van der Waals forces that promote the agglomeration of AuNPs. This result leads us to conclude that polyphenols can be good candidates for green synthesis of AuNPs, not only because they were efficient as reducing agents but also because they gave stability to synthesized nanoparticles. The synthesis with tea extract allows the use of phytochemicals for the production and stabilization of AuNPs, simplifying the process.

Between t1w and t2w, we can observe a slight increase in the absorbance value without a change in the behavior of the plasmonic response. Tea contains a multitude of different chemicals. Some of these, e.g., tannins, are fairly dark to begin with, but, if they are allowed to react with the oxygen [55] in the air, they oxidize, producing other compounds that are even darker in color. We can observe this phenomenon by UV spectroscopy because, in general, bigger molecules absorb more light, and as the oxidized tannins tend to aggregate over time [56], creating bigger molecules, this leads to a change in tea spectra. GO and rGO oxygen functional groups contribute to a more efficient oxidation of samples. The difference between t1w and t2w is more obvious in the 5%\_AuNPs/GO in comparison with the 5%\_AuNPs/rGO sample, since GO has more oxygen groups on its surface.

#### *2.4. Simulation (Mie Theory)*

The variation in the transmission spectra caused by the plasmonic resonance of the nanoparticles can be calculated by recourse to the Mie Theory [57], and the intensity of the LSPR effect can be correlated with the material properties of the surrounding medium [58]. Regarding gold spherical nanoparticles, as a bottom line of the Mie analysis, it can be stated, as a rule, that increasing AuNP size results in an LSPR shift towards the red part of the spectrum. Such a behavior can also be observed in experimental measurements [59]. Moreover, as the ratio between AuNP radius and light wavelength increases, multipolar behavior is to be expected and a widening of the peak waist observed. Increasing the values of the refractive index of the surrounding medium will also lead to a red shift of the LSPR peak, accompanied by a significant enhancement of its maximum value. For the analysis of the specific case of the AuNPs, produced by combining HAuCl4 with the phytochemicals present in tea extract as a reducing agent, we have considered the gold nanosphere capped by a thin uniform layer of tea polyphenols, immersed in pure water. This approach agrees with the morphology observed in Figure 3a. The most abundant polyphenol encountered in tea is epigallocatechin gallate (EGCG) [60], whose reported refractive index is 1.857 [61], which was the value used in the simulations.

Figure 9 reports the light extinction profile calculated for a AuNP with increasing radius and capped by a thin layer of EGCG (1–30 nm). From the analysis of this Figure, the LSPR wavelength (560 nm for a 10 nm radius of the AuNP) is only slightly affected by the thickness of the EGCG layer, but if it is too thick, LSPR intensity is reduced. Anyway, when compared with the LSPR produced by AuNPs in pure water, without EGCG capping, the LSPR peak always suffers a red shift. This red shift is observable for any thickness of the capping layer.

**Figure 9.** Simulated LSPR intensity for AuNPs with increasing dimensions (radius between 10 and 50 nm). Gold nanospheres are immersed in pure water and have a capping layer of EGCG with a thickness between 1 and 30 nm.

Figure 10 shows the results for the simulation where the light extinction is calculated for a AuNP with a radius of 30 nm and a capping layer thickness between 1 and 100 nm.

The effect of increasing the thickness of the cover layer results in the LSPR shifting towards the red region. Simultaneously, the intensity of the LSPR shows a marked reduction as the capping thickness increases.

We can conclude from this simulation analysis that the residual polyphenols which remain after AuNP fabrication have a negligible effect on the quality of the plasmonic response. An excessive accumulation of the residual polyphenols on the AuNP surface would reduce the LSPR intensity, but the simulation shows that for the range of the capping thickness observed in the TEM images, such a level of EGCD accumulation is not reached. The red shift foreseen by the simulation has no impact on the operation mechanism of a sensor device built with these materials. At the same time, the presence of the capping layer can be expected to physically separate the nanoparticles, preventing their aggregation. Thus, AuNPs synthesized by this green method combine the advantage of a simplified fabrication method that avoids aggregation with a reliable plasmonic resonance that can be successfully exploited in a sensing device.

The wavelength and intensity of LSPR are strongly dependent on the refractive index of the surrounding medium. An alteration of this parameter provokes a shift of the LSPR peak which can be used as the output value of a sensing system. To evaluate the sensing efficiency of the AuNPs, the spectral shift (Δλ) of the resonance wavelength, as a function of the variation in the refractive index (Δn) of the surrounding medium, can be translated into a sensitivity parameter (S):

#### S = Δλ/Δn

Figure 11 reports the position of the LSPR central wavelength as a function of the refractive index of the surrounding medium and the corresponding sensitivity. The presence of the EGCG capping layer also acts as a separation layer between the AuNPs and the surrounding medium, reducing the sensitivity of the AuNPs. The simulation results show that when the thickness of the EGCG layer remains below a few tenths of a nanometer, even if reduced, the sensitivity value is maintained within the limits described in the literature [62,63]. The variation of the refractive index also produces a modification in the LSPR peak intensity. The combined analysis of peak wavelength and intensity can be used to improve the sensor signal-to-noise ratio (SNR), allowing a higher tolerance to the sensitivity reduction introduced by the EGCG capping layer.

**Figure 11.** (**a**) Variation of the central wavelength for the LSPR resonance as a function of the medium refractive index for different thickness of the EGCG capping layer. (**b**) Sensitivity of the NPs' LSPR as a function of the EGCG capping layer thickness.

#### **3. Materials and Methods**

Chloroauric acid (HAuCl4·3H2O) was purchased from Sigma-Aldrich (Sigma-Aldrich, Munich, Germany); the black tea leaves (*Thea sinensis)* for the tea extract preparation used as a reducing agent for chloroauric acid was purchased from Pingo Doce (Pingo Doce, Portugal, tea brand—batch 1832); graphene [64] and derivatives were synthetized using the modified Hummers method [64].

Black tea extracts with 5% concentration were prepared as reported previously [10], the main difference being that the resides used in the addition of graphene and its derivatives were added in a different order to see if this would produce different UV spectra (different transitions). This led to the following sequences (Figure 12):

**Figure 12.** Preparation of samples by SQ1-AuNPs/rGO (**a**) or SQ2-rGO/AuNPs (**b**).

Sequence 1 (SQ1): 1 mL of chloroauric acid (0.1 M) was added to each tea extract *Thea sinensis* concentration (5%) with a 6 mL total volume. This mixture was stirred for 20 min at 400 rpm at room temperature. After that, 1 mg of graphene or derivative was added to every 2 mL of the samples previously obtained (Figure 12a).

Sequence 2 (SQ2): The same weight of graphene or derivatives was added before the addition of chloroauric acid (Figure 12b). The remaining synthesis process of nanoparticles was maintained as described in SQ1.

Both nanoparticles and nanocomposites were characterized by UV–Vis spectroscopy (UV-2501PC Schimadzu, Waltham, MA, USA) and transmission electron microscopy (Hitachi 8100, Tokyo, Japan) with a ThermoNoran EDS light and zeta potential (Litesizer 500— Anton-Paar, Gratz, Austria). The characterization by UV–Vis was undertaken at 3 distinct moments, t0, t1w and t2w, to determine the stability of the AuNPs and the nanocomposites over time, t0 being the starting moment, immediately after stirring the samples, t1w one week after, and t2w two weeks after. All samples were preserved in cold conditions and protected from light. TEM images were obtained between t0 and t1w, as was the zeta potential of the samples.

Modified and unmodified graphene were analyzed by X-ray photoelectron spectroscopy with a XSAM800 spectrometer from KRATOS. Non-monochromatic radiation from a Mg Kα source was used (hν = 1253.6 eV). Powdered samples were fixed on the XPS holder with a double face tape and analyzed at UHV, at TOA = 45◦. The BE was corrected considering the charge shift observed for the sp<sup>2</sup> C-C and C-H peak set at 284.4 eV [52]. Other operational conditions and data treatment details were as published elsewhere [65].

#### **4. Conclusions**

The black tea extract (*Thea sinensis*) showed reducing capacity of chloroauric acid, allowing the synthesis of spherical gold nanoparticles. The production of AuNPs through this green synthetic approach proved to be sustainable, not only due to its low cost but also because of the reducing capacity of the tea and its coating agent function, conferring stability to the synthesized AuNPs. We found that 2 weeks after stirring of the reagents, polyphenols prevailed as coating agents, sustaining the stability of the AuNPs. Simulation results based on the Mie theory for the LSPR effects support the conclusion that, even with a thin capping layer of residuals, no significant reduction of plasmonic resonance should be expected.

The addition of carbon-based nanomaterials before stirring of the tea extract with chloroauric acid proved not to be efficient with respect to the stability of synthesized AuNPs, and, in some cases, LSPR did not take place at any of the moments of characterization. The reverse order of material addition, i.e., the addition of carbon-based nanomaterials after the stirring of the tea extract with metallic salt, proved to be more efficient, both in terms of the synthesis and stability of the AuNPs.

Of the three materials studied, rGO proved to be the most efficient carbon-based nanomaterial used as a support for AuNPs to be applied in biosensors. The stability revealed by the zeta potential, the greater dispersion of AuNPs, and the conductivity of this nanomaterial reported by several authors support this statement. Khalil et al. [6] corroborate our conclusion by stating that rGO has advantages as a support of AuNPs, since this conjugation enhances stability, inhibiting agglomeration through a closer contact between rGO and AuNPs, as also suggested by XPS results. The possibility of producing a metal–graphene hybrid nanostructure, composed of AuNPs and graphene allotropes, opens an interesting avenue for the exploration of biosensing applications, as these composites can be tuned to a specific wavelength of resonance, while at the same time they are known to simplify functionalization with bioelements (antibodies or antigens) for selective detection of specific biomarkers or disease carriers. Once joined to a low-cost optoelectronic setup for output extraction, a LSPR sensing element fabricated with these graphene–metal hybrid nanostructures (AuNPs-G-GO-rGO) could be of great use in a situation where a large-scale, low-cost, and timely disease screening action is needed, as, for example, in a future pandemic crisis or in Third World countries, where access to laboratory facilities is problematic.

The topic of this paper has not been widely explored in the literature and it will be necessary to carry out more exhaustive research in order to reach conclusions about certain questions raised in our study. Nevertheless, we believe that this work can be a good starting point for this investigation. Regarding future work, it would be interesting to monitor the synthesis of AuNPs between t0 and t1w so that we can determine when this reaction ceases. In addition, trying other brands of black tea could supply additional valuable information, since different conditions, both soil and climatic, may be associated with different antioxidant properties.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/bios12030163/s1: Figure S1: Frequency vs. size distribution: (a) for Figure 3a; (b) for Figure 3b and (c) for Figure 5; Figure S2: C 1s spectral differences between (a) AuNPs/rGO and rGO; (b) AuNPs/G and G and (c) AuNPs/GO and GO; Figure S3: TEM image for (a)—G; (b)—rGO and (c)—GO; Figure S4: LSPR of AuNPs and G composite (SQ1 and SQ2) at t0, t1w and t2w; Figure S5: LSPR of AuNPs and GO composite (SQ1 and SQ2) at t0, t1w and t2w; Figure S6: LSPR of AuNPs and rGO composite (SQ1 and SQ2) at t0, t1w and t2w.

**Author Contributions:** Conceptualization, E.C.B.A.A. and A.P.C.R.; investigation, A.P.G.C., A.F., A.M.F. and A.P.C.R.; writing—original draft preparation, A.P.G.C., E.C.B.A.A., A.F., A.M.F. and A.P.C.R.; writing—review and editing, A.P.G.C., E.C.B.A.A., A.F., A.M.F., A.M.B.d.R. and A.P.C.R.; supervision, E.C.B.A.A. and A.P.C.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by EU funds through the FEDER European Regional Development Fund (project LISBOA-02-0145-FEDER-031311) and by Portuguese national funds by FCT— Fundação para a Ciência e a Tecnologia with projects PTDC/NAN-OPT/31311/2017, PTDC/QUI-QIN/29778/2017, project UIDB/00100/2020 of the Centro de Química Estrutural, LA/P/0056/2020 of Institute of Molecular Sciences, projects UIDB/04565/2020 and UIDP/04565/2020 of iBB and LA/P/0140/2020 of i4HB, project UID/EEA/00066/2020 from the Center of Technology and Systems, and from the Instituto Politécnico de Lisboa with IPL/2018/STREAM\_ISEL and IPL/2020/AGE-SPReS\_ISEL projects. APCR and AMF thank the Instituto Superior Técnico for the scientific employment contracts IST-ID/119/2018 and IST-ID/131/2018, respectively.

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

**Informed Consent Statement:** Not applicable.

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

#### **References**

