*2.7. Data Analysis*

Origin version 7.5 software (Microcal Software Inc., Northampton, MA, USA) was used for curve-fitting and data analysis. Data were obtained from a minimum of 3 independent experiments.

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

#### *3.1. Development of Acoustic Biosensor for the Detection of Trypsin Activity at Surfaces*

For the detection of trypsin activity at surfaces, it is crucial to optimize the methods of preparation of the protein layers that serve as a substrate for the protease of interest. The preparation of the protein layers on the surface of the transducers is a common application of acoustic biosensors. For instance, the preparation of casein layers is attractive for future applications in the pharmaceutical and food industries [32].

In this study, we have monitored the activity of trypsin at various concentrations (from 0.1 to 20 nM) in the hydrolysis of a β-casein layer immobilized onto a gold surface by a carboxylate terminated self-assembled monolayer (SAM) of MUA using a TSM technique. MUA strongly binds to gold through thiol groups in a high level of molecular dimension order, forming a stable SAM [33]. The formation of the SAM itself enables the coupling of activated carboxylic groups with free amino groups in the β-casein, which is an effective method for immobilizing proteins on a gold surface [34–36].

Figure 3 illustrates typical kinetic changes of the frequency, Δf, and motional resistance, Δ R m, obtained during the TSM experiment. The TSM crystal covered by the MUA layer activated by EDC/NHS established in a flow cell has been first washed by PBS. As soon as the stable baseline was established, the β-casein dissolved in PBS in a concentration of 0.5 mg/mL has been added. The sharp drop of the resonant frequency and an increase of the motional resistance were observed, indicating the adsorption of the β-casein to the quartz crystal/liquid interface. The washing of the surface by PBS resulted in only a slight increase of the frequency, which is evidence of removal of weakly adsorbed β-casein molecules from the surface. Thus, the frequency did not recover to the original value obtained when the crystals were exposed to the buffer. This suggests that there were two modes of casein binding to the MUA surface, a tightly bound layer and a weakly bound layer, and that only loosely bound casein layers were removed during the PBS washing [37]. Since the increase in resonant frequency after PBS washing was so small, we can speculate that β-casein adsorbed on the MUA formed a stable immobilized layer, which makes this result attractive for its potential applications in biosensors for the detection of protease activity.

The resulting frequency shift after the adsorption of the β-casein to the surfaces of the crystal was around −199.43 Hz. Furthermore, the buffer was changed to a 20 nM trypsin solution. The frequency increased asymptotically to reach a stable value, indicating that the proteolysis process occurred. Washing of the surface by PBS did not result in significant changes of frequency and motional resistance, which is evidence that the cleaved peptide residues were removed from the surface in a flow mode during the application of trypsin. The kinetics of the changes of the resonant frequency and motional resistance were recorded for different trypsin concentrations, each one with a new quartz crystal and a newly adsorbed β-casein layer.

**Figure 3.** Typical kinetics of the changes of resonant frequency, Δf, and motional resistance, ΔRm, of the thickness-shear mode (TSM) transducer for various modifications. The additions of β-casein, trypsin, and washing of the surface by phosphate-buffered saline (PBS) are shown by arrows.

Earlier works indicated that the Sauerbrey Equation (1) can be applied to obtain a rough estimate for the surface concentration of the adsorbed β-casein layer [23,38,39], which is valid only for the specific case of a crystal being loaded with rigid, well-adhered layers in air with a minor contribution to the surface viscosity [19,40,41]. As we mentioned in Section 2.6, the contribution of viscosity into the frequency changes can be estimated from the ratio |Δf/ΔRm|. At the highest concentration of trypsin (20 nM) studied and at the steady-state conditions (Figure 1), |Δf/ΔRm| = 199.43 Hz/7.4 Ω = 26.95 Hz/Ω. This value is much higher than the threshold value (10.37 Hz/Ω). This means that the changes of frequency are related mainly to the changes of the mass.

Therefore, with an awareness of the limitations stated above, Equation (1) can be used to estimate the amount of proteins on the surface (ΓQCM, ng/cm2) [38]. The average value of the frequency shift after the adsorption of the β-casein to hydrophilic surfaces was −165.26 ± 47.7 Hz. Using this value, as well as A = 0.2 cm<sup>2</sup> for the area of the electrode of an AT-cut quartz crystal (f0 = 8 MHz fundamental resonant frequency), a surface concentration of 228.1 ± 65.8 ng/cm<sup>2</sup> was obtained for β-casein.

This is in good agreemen<sup>t</sup> with earlier experimental works based on ellipsometry that reported 200–300 ng/cm<sup>2</sup> for a full-coverage monolayer of β-casein [42–44]. Furthermore, QCM studies by Tatarko et al. estimated a mass density of 350 ng/cm<sup>2</sup> for the immobilized β-casein monolayer [23]. These results support the interpretation that the surface concentration of β-casein obtained by TSM measurements corresponds to monolayer formation.

Based on the kinetic curves obtained for the concentration range of trypsin 0.1–20 nM, we prepared a plot of the frequency and motional resistance changes as a function of trypsin concentration (Figure 4). It can be seen that the frequency changes increase with increasing the trypsin concentration and started to saturate at CTRY > 10 nM. In contrast with frequency, Rm decreased with increasing the concentration of the protease, which is evidence of dominant mass changes.

**Figure 4.** Plots of changes of frequency, Δf, and motional resistance, ΔRm, vs. trypsin concentration (CTRY). Statistically, a value for the standard deviation was obtained from three independent experiments at each trypsin concentration.

For practical purposes, for the detection of trypsin in food or in other biological samples such as blood or blood plasma, it is convenient to analyze the effect of trypsin on the cleavage of β-casein by changes of resonant frequency of the quartz crystal under steady-state conditions. In order to minimize the effect of variation of the properties of β-casein layers at the monolayer of 11-mercaptoundecanoic acid (MUA) on the resonant frequency, we plotted the normalized frequency changes: ΔfN = (ΔfTRY/Δfcasein) × 100% vs. concentration of trypsin, CTRY (ΔfTRY is the frequency change corresponded to the cleavage of β-casein layer after incubation with a certain concentration of trypsin and Δfcasein is the frequency changes corresponded to the adsorption of β-casein at the MUA layer before trypsin addition). This dependence shown on Figure 5 can be fitted by the Langmuir isotherm (see Section 2.6 and Equation (2)).

The fitting of calibration plots yielded (ΔfN)max = 70.36 ± 4.60 and KM =0.92 ± 0.44 nM. The limit of detection (LOD) has been determined from the linear part of the dependence presented in Figure 5 using the 3.3(SD)/S rule (SD is standard deviation at the lowest concentration of trypsin, S is the slop of the linear dependence) as LOD = 0.48 ± 0.08 nM. Thus, in the presence of 20 nM trypsin, almost 70% of the casein layer is removed due to protease cleavage. This value is close to the maximum cleavage obtained by fitting the Langmuir isotherm. It can be assumed that due to the restricted access of the trypsin to the casein layer at the surface of the TSM transducer, around 30% of the casein remained at the surface after protease cleavage.

β-casein interacts with the immobilized MUA layer preferably with N-terminus. This part of the protein contains most of the charge [45]. It also contains numerous free amino groups that amino-reactive MUA can bind. β-casein is composed of 209 amino acids starting with arginine at the N-end (Arg1) [46]. The immediate binding of Arg1 to MUA is possible. Following the addition of trypsin, the cleavage of available peptide bonds toward the C-terminus of lysine and partially arginine occurs. These cleavage sites for trypsin are mostly identical to that of the plasmin [47]. The only unique cleavage site for trypsin is located between Arg202-Gly203, near the C-terminus [48]. The most common hydrolysis takes place at Lys28-Lys29, Lys105-His106, and Lys107-Glu108 with the

subsequent cleavage at Lys97-Ala98, Lys99-Glu100, and Lys113-Tyr114 [49]. The cleavage of these peptide bonds should cause release of the β-casein fragments (or so called γ-casein fragments) that corresponds to up to 88% of the β-casein molecular weight. This ratio can be affected by the β-casein assembly on the MUA layer and thus the availability of such bonds to the trypsin. Considering that approximately 70% of casein fragments are released from the sensing, we can speculate that the closest site for its cleavage by trypsin at the MUA layer is probably after Lys48. According to the ExPASy Peptide Cutter tool [50], the cleavage of β-casein by trypsin at Lys48 is highly probable.

**Figure 5.** Plot of the normalized changes of the resonant frequency ΔfN vs. trypsin concentrations, CTRY. Standard deviation values are obtained from three independent experiments. The red line is the fit according to the Langmuir isotherm (Equation (2)) with accuracy R<sup>2</sup> = 0.99.

In the paper by Chen et al. [51], the detection of trypsin activity based on the electrochemical method has been reported. They applied a gold working electrode modified with a short peptide substrate conjugated with graphene oxide (GO) and the thionine redox label. The incubation of the sensor with trypsin for2hresulted in cleavage of the peptide substrate, removal of the redox probe, and a decrease of the current amplitude. Although the authors reported a lower detection limit, down to 0.05 nM, and a high selectivity to trypsin, this method has some drawbacks. First, the biosensor was based on the peptide substrate labeled by the graphene oxide (GO)–thionine conjugate, which is not available commercially. The peptide–GO–thionine conjugates are more expensive in comparison with the β-casein used in our work. Therefore, this limits the practical application of such an electrochemical sensor. Moreover, unlike the label-free approach presented in our work, the method by Chen et al. cannot monitor the trypsin activity in real time, because this activity was detected only after 2 h of incubation of the trypsin with the peptide-modified electrode. It should be also mentioned that commercially available enzyme-linked immunosorbent assay (ELISA) kits for trypsin possess also high selectivity and sensitivity similar to the work of Chen et al. (down to 0.012 nM) [52]. However, those kits require expensive antibodies, and detection is carried out in several steps. Furthermore, ELISA does

not allow monitoring of the kinetics of the trypsin activity. The acoustic sensor developed by us is sufficiently sensitive (LOD of 0.48 ± 0.08 nM) to detect such dangerous diseases as cystic fibrosis, acute pancreatitis, or the acute phase of chronic pancreatitis that are characterized by raised concentration of trypsin in blood in the range of 2.1–71.4 nM [7,8]. In contrast with ELISA, the TSM biosensor is label-free, straightforward, and facile regarding the evaluation of the response. In addition, the TSM method can be used in samples that are not transparent.

#### *3.2. Sprectophtometric Assay of Protease Activity*

In colorimetric sensor applications, AuNPs are most widely used due to their high stability, facile synthesis, excellent biocompatibility, and strong surface plasmon resonance effect. This effect can be utilized to produce visual color changes in a process termed the colorimetric method [53,54]. Here, we report the results of a simple colorimetric assay based on the optical properties of functionalized AuNPs (Figure 2). The purpose of this study was a comparison of the sensitivity of surface-based (TSM) and volume-sensitive methods of trypsin activity detection. We used a slightly modified version of the method reported by Chuang et al. [26]. However, instead of gelatin, β-casein has been used as a substrate for trypsin digestion. Briefly, for the protease assay, AuNPs were first modified by β-casein and subsequently with MCH. The molecules of MCH are chemisorbed to the AuNPs through a thiol group (-SH) substitution and the hydroxyl group (-OH) exposed on the AuNPs surface enhances the attraction force between AuNPs. Additionally, MCH molecules on the AuNPs act as blockers, while covering the surface area of the AuNPs that are not conjugated with casein and blocking adsorption of the protease on the surface of the AuNPs [26]. The addition of MCH to the AuNPs–β-casein solution led to a color change from wine-red to violet. When trypsin digests the casein at AuNPs/MCH–casein, NPs aggregated due to the removal of the protective layer of casein and the color change from violet to blue occurred within minutes; then, the solution became colorless.

The absorption spectra of AuNPs in the absence of β-casein (black curve), presence of β-casein (red curve), presence of β-casein and MCH (blue curve), and AuNPs with chemisorbed MCH (magenta curve) are shown in Figure 6. The absorption peak of pure AuNPs is centered at 520 nm as expected. This indicates that the gold colloids are not aggregated but well dispersed as individual particles [55]. After the modification of AuNPs with β-casein, the position of the maximum absorption of AuNPs shifted from 520 to 525 nm, which indicates the formation of bioconjugates [55]. The shift is identical with those reported in [26] for AuNPs modified by gelatin. The red shift in the position of the plasmon absorption band is produced by a perturbation in the dielectric constant around the nanoparticles due to the chemisorption of β-casein molecules [56]. No significant broadening of the spectrum was observed after the β-casein adsorption process, which indicates that the separation distance between AuNPs is higher than their radii, and that AuNPs do not experience aggregation into larger nanoparticles upon the adsorption of β-casein [55]. Further modification with MCH resulted in a significant red shift around 60 nm accompanied by the broadening of the spectrum. This broadening is indicative of an aggregation of nanoparticles This is due to the replacement of the β-casein protective layer with MCH, which in turn makes the nanoparticles closer to each other [57,58]. The modification of AuNPs with MCH resulted in a significant red shift, indicating strong aggregation of the nanoparticles.

Additionally, a less expressed maximum at 280 nm is observed after the modification of AuNPs by β-casein. This is due to the absorption of β-casein's amino acids at this wavelength. The amplitude of this peak decreases after the chemisorption of MCH, which is probably due to the removal of weakly adsorbed casein molecules from the surface of AuNPs. Furthermore, we carried out a quantitative analysis of trypsin activity via the UV-vis spectroscopy method. For this purpose, trypsin was added to the AuNPs solution. We recorded the changes of absorbance spectra of the AuNPs suspension during the trypsin cleavage at 0 min, 0.01 min, 15 min, 30 min, 45 min, and 60 min. Figure 7 illustrates the

changes in spectra over time in a 10 nM concentration of trypsin. A substantial red shift (up to 640 nm) of the spectra and a decrease in absorbance with time was observed at this concentration of trypsin, due to trypsin-induced aggregation caused by the cleavage of the AuNPs' protective shell as well as the MCH induced increase of attractive force between the AuNPs. Moreover, the absorbance spectra showed a decrease in the absorption spectra at 280 nm when the AuNPs/MCH–casein was digested by trypsin. It can also be seen that the absorbance decreased with time. The absorbance also started to decrease after maximum shifting. Our results are in good agreemen<sup>t</sup> with those previously reported by Chuang et al., whose work served as our inspiration to design a colorimetric assay based on an AuNPs/MCH-protein platform. Chuang et al. demonstrated that protein modified AuNPs aggregation after treatment with protease can be successfully monitored via the red shift of absorption spectra [26].

**Figure 6.** UV-vis absorption spectra of gold nanoparticles (AuNPs): bare (black), modified by βcasein (red), and subsequently modified by 6-mercapto-1-hexanol (MCH) (blue) as well as AuNPs modified by MCH (magenta).

**Figure 7.** UV-vis absorption spectra of β-casein and MCH-conjugated AuNPs treated with 10 nM trypsin at different time points. Note that at time 0 and 0.01 min, the spectra are almost identical.

Trypsin at concentrations ranging from 0.1 to 10 nM was used in the study to estimate the detection limit of the optical AuNPs assay. In order to construct the calibration curve, we have plotted the changes in relative values of absorbance measured at around 640 nm after 15 min of trypsin exposure against the trypsin concentration (Figure 8a). As in the case of the analysis of trypsin activity via the TSM method, we were able to use an inverse Michaelis–Menten (MM) model expressed by Equation (3) to analyze the obtained calibration curve for trypsin activity in volume.

**Figure 8.** Calibration plots of colorimetric assay. (**a**) Changes in relative values of absorbance after β-casein and MCH functionalized gold nanoparticles (AuNPs/MCH-β-casein) exposure to trypsin (A0—exposure time 0 min, A15—exposure time 15 min.) vs. concentration of trypsin (CTRY). Symbols are experimental data, and the red line is the best fit of Equation (3). (**b**) Linear part of the calibration curve for calculation of the limit of detection (LOD). Values are means ± SD (*n* = 3). Red line is the linear regression fit.

The fitting of calibration plots with the MM model yielded *vmax* = 14.98 ± 0.81% and KM = 0.56 ± 0.10 nM. As was already mentioned, in the inverse MM model, the roles of the enzyme and substrate are swapped, and the concentration of the enzyme is changed while the substrate is presented in excess.

To calculate the LOD, we used only part of the calibration curve from 0 to 5 nM, where the dependence was almost linear. The obtained LOD was 0.42 ± 0.03 nM, according to the rule 3.3 (SD)/S, where SD is the standard deviation of the sample with the lowest concentration and S is slope calculated from the fit of the linear part of the calibration curve [59]. The results are shown in Figure 8b.

It is interesting to compare the properties of the AuNPs assay and the TSM method used to detect trypsin activity (Table 1). On one hand, both methods successfully detected protease activity at the sub-nM level, within a similar time range in a real-time mode. However, it should be noted that a major drawback of the AuNPs assay is that the method is of limited application in a turbid medium. On the other hand, unlike the TSM method, detection using the AuNPs assay can be carried out in only one step, as the signal detection simply involves the direct measurement of the absorbance values at A640. It is also interesting to compare the reverse Michaelis–Menten constants for both methods. As can be seen in Table 1, a lower KM value has been obtained for the AuNPs-based colorimetric assay. This can be attributed to trypsin's better access to the β-casein substrate. Certainly, the β-casein layer is formed at MUA monolayers by covalent binding of the casein hydrophilic amino groups. Thus, the cleavage sites are closer to the quartz crystal surface with limited access to the trypsin. In addition, due to covalent binding of casein molecules at the self-assembled MUA, the casein layer is compactly packed, which creates additional restriction of access of trypsin to the cleavage sites. A similar conclusion was also obtained for chymotrypsin detection [57]. In contrast, at AuNPs, the casein is physically adsorbed at

the gold surface. This means that casein molecules are randomly oriented, which make the access of trypsin to the casein cleavage sites more advantageous.

**Table 1.** Comparison of TSM biosensor and AuNPs platform (colorimetric biosensor) used for detection of trypsin.


We should also mention that in contrast with the colorimetric method, the acoustic TSM technique is sensitive to air bubbles presented in the sample and to the pressure changes caused by handling of the flow cell. Air bubbles are more prone to growth at the hydrophilic interface, which likely altered to a hydrophobic case upon the adsorption of the casein layer. Therefore, special care can be taken in avoiding this effect, for example by degassing the sample before starting the experiments.

Finally, we briefly discuss the most often used techniques employed for trypsin detection. The advantages and disadvantages of these techniques as well as their LOD are summarized in Table 2. Nowadays, researchers' efforts are focused on the development of simple and rapid biosensors for the sensitive determination of trypsin because traditional methods such as enzyme-linked immunosorbent assay (ELISA), gelatin-based film technique, and high-performance liquid-chromatography (HPLC) are time-consuming and require specialized instruments and trained personnel [60]. Moreover, those methods do not allow for the monitoring of the kinetics of protease activity. Recently, many efforts have been reported regarding trypsin determination. Several biosensors based on fluorescent, electrochemical, and colorimetric methods have been developed to detect trypsin [15,26,60,61]. Fluorescence-based homogeneous assays are the most popular ones for trypsin activity monitoring due to their simple processes, high sensitivity, and convenient operation. These methods usually need peptide-based molecular probes containing fluorochrome and quencher pairs to monitor specific proteases by fluorescence resonance energy transfer. Nevertheless, these labeled fluorogenic substrates are expensive and are difficult to synthesize [62]. Colorimetry is another method reported for the detection and screening of trypsin [63]. It is the simplest, less expensive, and most widely used method. It can be directly observed with the naked eye or accurately quantified via UV-vis spectrophotometer. Unfortunately, the colorimetric method is limited to only optically transparent liquids. Electrochemical methods are rather sensitive. However, they require the conjugation of a specific peptide substrate by redox probes, longer incubation time with protease, and cannot be used for measuring the kinetics of enzyme reaction [51]. Acoustic methods are among the most effective and promising approaches for the detection of trypsin activity. Their advantage lies in high sensitivity, which reaches levels comparable to state-of-the-art techniques such as ELISA. Since most of the biochemical samples are acoustically transparent, measurements can be performed in a wider range of solutions without the need for a chemical probe, as well as in opaque and high-concentration samples that are difficult to measure with optical methods. Most recently, we successfully demonstrated the feasibility of a volume-sensitive acoustic method for the detection of proteolytic activity of trypsin [64]. However, it is worth noting a possible limitation of the proposed acoustic methods: namely, the influence of air bubbles and temperature stability. This limitation must be addressed in future research.


**Table 2.** Comparison of the most used analytical methods for trypsin determination.
