Peptidic Sulfhydryl for Interfacing Nanocrystals and Subsequent Sensing of SARS-CoV-2 Protease

Abstract

There is a need for surveillance of COVID-19 to identify individuals infected with SARS-CoV-2 coronavirus. Although specific, nucleic acid testing has limitations in terms of point-of-care testing. One potential alternative is the nonstructural protease (nsp5, also known as M^pro/3CL^pro) implicated in SARS-CoV-2 viral replication but not incorporated into virions. Here, we report a divalent substrate with a novel design, (Cys)₂–(AA)_x–(Asp)₃, to interface gold colloids in the specific presence of M^pro leading to a rapid and colorimetric readout. Citrate- and tris(2-carboxyethyl)phosphine (TCEP)-AuNPs were identified as the best reporter out of the 17 ligated nanoparticles. Furthermore, we empirically determined the effects of varying cysteine valence and biological media on the sensor specificity and sensitivity. The divalent peptide was specific to M^pro, that is, there was no response when tested with other proteins or enzymes. Furthermore, the M^pro detection limits in Tris buffer and exhaled breath matrices are 12.2 and 18.9 nM, respectively, which are comparable to other reported methods (i.e., at low nanomolar concentrations) yet with a rapid and visual readout. These results from our work would provide informative rationales to design a practical and noninvasive alternative for COVID-19 diagnostic testing—the presence of viral proteases in biofluids is validated.

Graphical Abstract

INTRODUCTION

The colloidal stability of nanostructures is typically determined by their interfacial functionalities, which can shift the balance of ligand–metal and ligand–solvent interactions in either direction.^1,2 Nanocrystals self-assemble when the coordinating ligands favor interactions with other metal cores or surface molecules. Such ligand-induced coupling has evolved into a popular strategy of signal transduction in stimuli-responsive sensors, photonic devices, and biomedical imaging.^3-6 In particular, near-field plasmonic coupling promoted by peptides has been harnessed to measure proteases.^7-10 Several peptide coatings have been exploited to provide nanocrystals with enhanced colloidal stability while preserving their photophysical features.^11-15 Importantly, these peptide functionalities on the surface are protease-responsive and therefore address the analytes via optical signals.¹⁶ Examples of plasmonic coupling through proteolysis have been reported for metalloproteinase, caspase-3, furin, etc.^17-19

The key principle for nanoplasmonic sensing is to modulate the substrate’s physicochemical features via enzymatic stimulus (e.g., interfacial affinity,²⁰ amphiphilicity,²¹ charge valence,^22,23 and reactivity²⁴). This can subsequently induce a colloidal phase transformation. In widely cited examples, cysteine (C) and histidine (H) residues enable surface functionalization via convenient metal–thiol/histidine interactions (i.e., Au–S = 126–167 kJ/mol).^25-29 Sequences comprising proline, alanine, and serine in order (i.e., PASylation) result in highly water-soluble polypeptides with properties similar to poly(ethylene glycol) (PEG).³⁰ Alternative glutamic acid and lysine [i.e., (EK)₃] create a zwitterionic hydration layer that prevents the formation of protein corona.³¹ Various reactive groups can also be introduced on the peptide backbone for bioconjugations.^32,33 As a result, the use of peptide ligands for colloidal assembly appears favorable due to the automatic synthesis, interfacial binding with satisfactory affinity, and enzymatic responsiveness.¹⁴ Indeed, peptides featuring a head and tail thiol [i.e., Cys–(AA)_n–Cys] have been extensively applied to disassemble colloidal aggregates via the proteolysis of responsive modules.^20,34 Although effective, this conventional design suffers from the lack of hydrophilic domains for maintaining a balance between the water solubility and responsiveness.

Inspired by the concepts from the modular ligand field,^1,7,35 here, we designed and tested peptides with a novel formula of (Cys)_n–(AA)_x–(Asp)₃ to assemble metal nanostructures and improve their performance in aqueous media. We then validated this system for nanoplasmonic sensing of SARS-CoV-2 main protease (M^pro, or nsp5/3CL^pro). These nonstructural proteins (nsp), in particular M^pro, are crucial to viral replication and thus may be useful for diagnosis and antiviral therapeutics (e.g., Pfizer Nirmatrelvir) development.³⁶ The ligand-like peptide used here has three domains: (i) cysteine anchors, (ii) an M^pro recognition site, and (iii) a solubilizing block comprising aspartate acids. There was a sharp transition between the colloidal stabilization promoted by zero- and multivalent peptides. Gold nanoparticles (AuNPs) assembled via thiol–Au bridges after enzymatic cleavage because of changes in the balance of interfacial affinity and hydrophilicity on the peptides. We then explored the effects of varying a few parameters (e.g., number of cysteines and testing media) on the resulting colloidal aggregates. We compared the properties such as optical spectra, hydrodynamic size, and colloidal stability. The divalent substrate (i.e., two cysteines) yields the most intense color changes, and the detection limit of M^pro is 10–20 nM in biological biofluids (e.g., exhaled breath and saliva) with good specificity. These detection limits are slightly higher than those of the reported fluorogenic probes (i.e., 5–10 nM³⁷) but are much less time-consuming (<10 min) and with a visual readout. Considering the broad interest in rapid diagnosis and mass surveillance of COVID-19, these findings may have important implications for the development of portable biological sensors for SARS-CoV-2 proteases.

RESULTS AND DISCUSSION

Prior work in colorimetric assays for proteases have described the combined use of AuNPs and thiol-flanking peptides, i.e., Cys–(AA)_x–Cys, where –(AA)_x– is the responsive module (Figure 1a).^20,34 This configuration often requires stringent water solubility of the responsive modules with limited performance when the catalytic domain of the enzyme contains hydrophobic amino acids (AAs). Our target here is the coronavirus main protease (M^pro), but M^pro has well-defined substrate specificities with cleavage at most sites containing a hydrophobic Leu at the P₂ position, Gln at the P₁ position, small aliphatic residues (i.e., Ser, Ala, and Gly) at the P₁' position, and bulky aromatic Phe at the P₃' position (Figure 1e).^37-40 As such, the oligopeptide substrates of M^pro are usually quite hydrophobic. This motivated us to reframe the conventional thiol-flanking design and expand the peptide portfolio of existing colorimetric assays for the SARS-CoV-2 main protease.

Figure 1.

Modular peptide design for colorimetric sensing of SARS-CoV-2 M^pro. (a) Conventional flanking design modifies the peptide head and tail with a cysteine (top). The novel peptide contains three modules, i.e., the surface anchor, the cleavage site, and the hydrophilic segment (bottom). (b) Ligand-like divalent peptide stabilizes nanoparticles (top), while the proteolytic fragments favor interparticle bridging (bottom). The green cartoon denotes the protease; Au₁ and Au₂ represent two different AuNPs. Transmission electron microscopy (TEM) images of the monodispersed 13 nm AuNPs mixed with the ligand-like divalent peptide (c) and the proteolytic fragments (d). The aggregation is induced via dithiol bridging. (e) Synthetic peptides with increasing number of cysteines, including CYS0, CYS1, CYS2, CYS4, and cys2. Here, the number after CYS implies the number of Cys residues and the lowercase cys is the scramble control. The AAs on the recognition site and functional motifs are color coded. SARS-CoV-2 main protease (M^pro) cleaves the peptide at Gln↓Ser (or Q↓S at the P₁ and P₁’ site). HPLC (f) and ESI-MS (g) data confirm the cleavage of CYS2 by M^pro. The peak with // is DMSO solvent with one CYS2 fragment, and the peak with * is the target of interest.

Modular Peptide Design

We designed and tested the ligand-like peptide and combined them with AuNPs for colorimetric sensing of M^pro. We hypothesized that the inclusion of multiple cysteine and hydrophilic aspartic acid residues in the peptidic backbone resemble the established multifunctional ligands for interfacing AuNPs.⁷ The products of peptide breakdown by protease would induce clustering of AuNPs via thiol bridging, thus changing the dispersion color (Figure 1b-d). Here, peptides in (Cys)_n–(AA)_x–(Asp)₃ were synthesized to encompass three functional domains (Figure 1a,e): (i) several proximate Cys near the C-terminus for binding metal surfaces, (ii) a centered enzyme recognition sequence, and (iii) an N-terminal solubilizing segment made of three aspartic acids (Asp) for promoting hydrophilicity and colloidal stability. Importantly, the incorporated thiol bridge [in part (i)] mimics a simple dithiothreitol (DTT) molecule that would favor interparticle interactions due to the high affinity of the thiol to metal surfaces.^41,42 The repeating aspartic acid residues improve the peptide performance in aqueous media regardless of the hydrophobicity of the recognition sequence. Overall, this modular design obviates the need for major revisions when targeting other proteases—the specificity of the sensor can be tuned simply by changing the cleavage site.

A highly water-soluble peptide was synthesized following the above criteria: DDDTSAVLQ↓SGF-(Cys)_n (↓ marks the cleavage site; n is the cysteine valence, Figure 1e). To determine the effect of cysteine valence n on the colloidal assembly, CYS0, CYS1, CYS2, and CYS4 peptides were used corresponding to zero-, mono-, di-, and tetravalent ligands, respectively.^43,44 Cleavage at the C-terminal Gln (or Q) is rarely seen for human proteases with the exception of kallikrein-3 that is expressed only in the prostate.⁴⁵ For example, the specific cleavage of the CYS2 parent peptide by M^pro has been confirmed by HPLC and ESI-MS, which resulted in the formation of a dithiol fragment SGFACGAGC (Figure 1f,g). A scramble sequence (cys2) that cannot be cleaved by M^pro was also synthesized and tested for colloidal stability.

Screening Ligands for Colorimetric Assays

In addition to the peptide design, protecting ligands on the metal surfaces are of paramount importance for nanoplasmonic sensing. We tested the commonly used ligands (n = 17, mostly negative charge), which range from weakly coordinating monomers (e.g., citrate, phosphine molecules) to the strong binding polymers with multiple anchors (e.g., derivatives of lipoic acid, La);^{1,41,42,46,47} 13 nm citrate-AuNPs were successfully derivatized with these ligands and their characteristic properties (i.e., size, surface potential, and stability) were compared side-by-side via dynamic light scattering (DLS), gel electrophoresis, and dithiothreitol (DTT) destabilization (Figures S2 and S3 and Table S2). The presence of DTT redox molecules with high affinity to gold surfaces would competitively displace any weakly bound ligands and thus impart interparticle interactions and vice versa.^48,49

Figure 2 shows that citrate, tannic acid (TA), TCEP, polyacrylic acid (PAA), polystyrene sulfonate (PSS), and polyvinylpyrrolidone (PVP)-capped AuNPs produced intense color changes in the presence of DTT. The comparisons of the aggregation patterns, manifesting in versatile colors, also confirm the unique derivatization of the pristine (or citrate-capped) AuNPs with different surface molecules. For example, the agglomerated TA-AuNPs appear as green due to the mixed color of clustered particles (blue) and tannic acid ligands (faint yellow).⁵⁰ The AuNPs had a fast color shift at low amounts of DTT (e.g., 5–20 μM); a reverse course was observed at high concentrations.⁵¹ The behavior of phosphine-gold varied broadly when DTT was added: TCEP-AuNPs aggregated, but particles coated with phenyl phosphines remained stable [i.e., diphenylphosphinobenzene sulfonate (DPPS) and bis(p-sulfonatophenyl)phenylphosphine (BSPP); see structures in Figure S2]. The stability of DPPS and BSPP-coated gold is attributed to the large steric hindrance from the bulky aromatic rings that prevent interparticle attractions.^1,52 In contrast, most thiol ligand-protected AuNPs were insensitive to DTT aggregation including mercaptopropionic acid (MPA), mercaptopropane sulfonate (MPS), glutathione (GSH), and thiolate PEG oligomers.^42,53 Particles functionalized with polymers (i.e., PAA and PSS) have narrow concentration windows (i.e., 5–10 μM) for the DTT-induced visual readout. Thus, subsequent work tested citrate-, TA-, or TCEP-AuNPs with the multivalent peptides.

Figure 2.

Dithiothreitol (DTT) test. AuNPs (13 nm, ~3.4 nM) coated with different ligands (n = 17) and incubated with increasing concentrations of DTT (i.e., from 0 to 100 μM, with 10 mM NaCl) 10 min after addition.^42,51 Citrate-, TA-, TCEP-, PAA-, and PSS-capped AuNPs show color changes via a dithiol bridging mechanism. The red color represents dispersed nanoparticles while violet-blue indicates colloidal aggregates. Characterizations of these particles are given in Figures S2 and S3 and Table S2.

Colloidal Stability Enhanced by the Peptide

We performed optical, size, and stability measurements to evaluate whether the proposed peptide acts as a stabilizing ligand and promotes the colloidal stability in peptide-AuNPs. Figure 3a shows that the optical absorption from pristine nanocrystals and after peptidic ligation are essentially identical (Δλ_SPR ≤ +6 nm), implying the photophysical integrity of the gold nanocolloids. However, a shoulder at ~600 nm increases for the absorption spectrum of CYS0-AuNPs suggesting aggregates. This is in line with the hydrodynamic diameter (D_H) measurement (Figure 3b). For example, the Au dispersion with the CYS0 peptide clearly indicates particle agglomeration (D_H at ~64.3 nm) versus the pristine AuNPs with a D_H ~ 18.2 nm. This is because the CYS0 peptide cannot coordinate with the nanoparticle surface to stabilize it. Thus, there is an increase in aggregation during particle processing such as centrifugation. Importantly, only a slight increase in the D_H value is measured after coating AuNPs with other thiolate peptides such as CYS1 (monovalent, D_H ~ 24.3 nm), CYS2 (divalent, D_H ~ 24.4 nm), and CYS4 (tetravalent, D_H ~ 28.2 nm). These minor increases in D_H after ligand substitution confirm the more complex peptide architecture relative to citrate anions on surfaces.

Figure 3.

Characterization of peptide-capped AuNPs. (a) Absorption spectra collected from AuNP dispersion, as-grown (red) and after ligation with the synthetic peptides. (b) DLS profiles of citrate-AuNPs (D_H = 18.2 nm, red) and other peptide-AuNPs. The D_H for AuNPs ligated with CYS0, CYS1, CYS2, and CYS4 peptides is 64.3, 24.3, 24.4, and 28.2 nm, respectively. (c) Profiles for the aggregation factor vs time extracted from various dispersions of peptide-AuNPs (1 nM) in the presence of 20 mM DTT and 400 mM NaCl. PEGylated AuNPs were used as a reference. (d) Time-dependent progression of the SPR peak at 520 nm, extracted from a series of mixtures of peptide-AuNPs (1 nM) with 0.5 mM NaCN. (e–g) Absorption spectra collected from CYS1 (monovalent)-, CYS2 (divalent)-, and CYS4 (tetravalent)-AuNPs, following the reaction with 500 equiv of maleimide-TAMRA dye (structure in panel g). Inset schematics show that AuNPs with mono-/dithiol peptides cannot conjugate with the dye (7–25 dye/NP). In comparison, conjugates prepared starting with tetracysteine peptide-AuNPs yield an increasing conjugate of ~229, implying uncoordinated thiol groups on the Au surfaces for maleimide activation. The orange circle and purple and blue blocks in the ligand cartoon represent the peptide domain of the anchor, the cleavage site, and the solubilizing group. A close look on panel g is given in (h), where the spectrum of the deconvoluted dye resembles that of the dye alone (bottom, dash line).

We further tested the ability of the peptide coatings to stabilize the Au cores against competition from DTT and chemical digestion by sodium cyanide (NaCN).^42,54 DTT (20 mM) probes the ligand affinity to the metal core, and NaCN (0.5 mM) evaluates the ligand packing density. Here, a standard reference using PEGylated AuNPs was also tested. The DTT destabilization manifests in a progressive loss of the plasmon features with an increase of the aggregation factor (or Abs₆₀₀/Abs₅₂₀). Figure 3c shows that the pristine and CYS0-ligated AuNPs have a rapid increase in the aggregation factor. Conversely, the ratiometric absorbance for the mono-/dithiol peptide-AuNPs remains near the baseline over time. This is also observed using the PEGylated NPs, which are attributed to the presence of strong coordinating moieties. Despite this, the tetravalent peptide (CYS4)-coated AuNPs have a modest colloidal stability after addition of DTT: Presumably there are dangling reactive sulfhydryl groups on the surfaces. Figure 3d shows an essentially unchanged optical density over time suggesting that thiolate peptides (including those of mono-/di-/tetravalent) can protect the Au core from CN⁻ complexation and dissolution, i.e., little or no loss of SPR features. The relatively unstable pristine and CYS0-AuNPs showed a fast decay in the plasmonic band. Thus, we conclude that the mono- or divalent peptides stabilize gold colloids in solution across a wide range of conditions.

The binding modes of peptide sulfhydryl to Au surfaces were investigated. Steric effects may prevent coordination of all thiol groups onto the gold surface.⁵⁵ To evaluate the number of free thiols, we used the highly efficient maleimide activation of peptidic sulfhydryl to attach the dye [tetramethylrhodamine (TAMRA)] to the nanocrystals. The absorbance spectra of AuNP-dye conjugates were deconvoluted to extract the number of bound dyes per NP using their molar absorption coefficients. The monothiol peptide (CYS1) lacks extra sulfhydryl for TAMRA coupling other than metal coordination, thus resulting in a negligible valence (i.e., ~7; Figure 3e). The divalent CYS2-capped AuNPs after maleimide activation also showed essentially no dye attachment (i.e., ~25, Figure 3f). That is, most cysteine moieties on the mono- or divalent peptides were anchored on the metal surfaces. In comparison, the tetracysteine CYS4-coated AuNPs showed accessible thiols for dye coupling, with a high valence estimated to be ~229. Figure 3g,h shows an increasing contribution from the conjugated dyes to the composite absorption profiles commensurate with high cysteine valence on the peptide. This explains the weak colloidal stability provided by tetracysteine CYS4 peptides (Figure 3c): The dangling and reactive sulfhydryls could undermine the overall colloidal stability by reactively attracting soluble specimens (e.g., other AuNPs or thiolates) close to the Au surfaces.^41,51 This also agrees with the experimental observations showing that exchange of AuNPs with excess tetravalent peptides (i.e., 130,000 in molar ratio) yielded irreversible aggregation. In addition, the cysteine valence on the peptide has important implications for colorimetric assays.

Aggregation Induced by the Proteolytic Product

In contrast to the stability induced by the parent/intact peptides, proteolytic fragments rapidly aggregated colloidal gold because the balance between the water solubility and interfacial affinity is largely flipped.^4,12,56 Here, we limited our study to simple citrate-AuNPs. In Figure 4a, the absorption profiles of citrate-AuNPs with the intact divalent peptide (c_CYS2 = 0–90 μM) remain essentially the same. In comparison, the products of CYS2 breakdown by M^pro led to a pronounced decrease in the SPR band commensurate with a noticeable shift to 600 nm. Figure 4b shows the time progression of the absorbance ratio (Abs₆₀₀/Abs₅₂₀) collected by incubating citrate-AuNPs (2.8 nM, 120 μL) with increasing concentrations of CYS2 fragments (i.e., 0–90 μM). No convergence of the ratiometric absorbance values over time was observed at various concentrations, indicating that a continuous application of an oligo-sulfhydryl stimulus is required for maximizing this aggregation. We defined a 10 min readout time for rapid protease detection, but a continuous measurement system could also be designed with improved detection limits. Under these conditions, the minimum amount of CYS2 fragments required to trigger optical changes exceeds ~30 μM (Figure 4c). The responsiveness of citrate-AuNPs to the proteolytic dithiol products in the range of micromolar concentrations agrees with the initial DTT test, as shown in Figure 2. A similar trend was found using the TCEP-AuNPs (Figure S5). Nevertheless, the dithiol fragments of CYS2 breakdown could not successfully aggregate TA-capped AuNPs under the testing conditions.

Figure 4.

M^pro-induced particle aggregation using the modular peptide. (a) Concentration-dependent optical absorption of citrate-AuNPs (2.8 nM, 120 μL), when incubated with intact CYS2 peptides (left) and the products of M^pro breakdown (c = 0–90 μM, right). Arrows designate sizable peak changes at 520 and 600 nm. (b) Time progression of ratiometric absorbance (Abs₆₀₀/Abs₅₂₀), where citrate-AuNPs (2.8 nM, 120 μL) are incubated with increasing concentrations of the CYS2 parent or its proteolytic fragments (0–90 μM). Tris buffer (5 mM, pH 8.0) was used and the time interval is 1 min. Error bar = standard deviation (n = 3). (c) Ratio of Abs₆₀₀/Abs₅₂₀ at 10 min collected from citrate-AuNPs (2.8 nM, 120 μL) incubated with various amounts of the CYS2 parent (red) and fragments (blue). The best working window is >30 μM. See also color evolution of citrate-AuNPs (2.8 nM, 120 μL) in the presence of the CYS2 parent (d) and fragments (e). These are cropped images where purple represents more aggregation (see color bar). (f) Fractional RGB (=V_B/V_R+G+B) extracted from panels (d, e) at 10 min. DLS profiles (g) and zeta potential (h) of citrate-AuNPs (2.8 nM, 120 μL) incubated with increasing concentrations of the CYS2 parent (blue) and its fragments (red). Error bars represent triplicate measurements for one sample. Ratiometric absorbance at 10 min collected from citrate-AuNPs (2.8 nM, 120 μL) incubated with various amounts of the CYS0 peptide (i), CYS1 peptide (j), CYS4 peptide (k), and their corresponding proteolytic fragments. The inset illustrates the decreased absorbance without the SPR shift. (l) View of MANTA⁵⁷ size measurements shows that citrate-AuNPs scatter more blue light (small-sized, left) whereas the CYS2 fragment-induced colloidal aggregates scatter more red light (large-sized, right).

The red shift of optical absorption (i.e., 520 → 600 nm) leads to a color change. Red color was observed upon mixing particles with the intact CYS2 peptide, but the addition of CYS2 fragments (>30 μM) to the gold dispersion caused a visual color response to purple (Figure 4d,e). We next used RGB analysis (V_B/V_R+G+B) via a smart phone to obtain a quantitative response in Figure 4f that resembles to the one collected by a laboratory spectrophotometer (Figure 4c). The color changes were further collaborated with DLS (Figure 4g), zeta potential (Figure 4h), TEM, (Figures 1c,d and S4), and multispectral advanced nanoparticle tracking analysis (MANTA, Figure 4l) measurements. These data confirmed that the AuNPs were monodisperse with intact peptides, while agglomerated significantly in the presence of CYS2 fragments. Despite the neutrality of the dithiol fragment of CYS2, the surface potential of the cross-linked AuNPs increased from −30.2 to −6.5 mV. Presumably the thiolate specimens displace the native citrate anions. Additionally, this covalent linking was irreversible in the presence of other stabilizing surfactants such as sodium dodecyl sulfate or PEG likely due to the strong affinity of the thiol─Au bond.²⁹ As a control, the zero-valence peptide (CYS0) could not cluster particles regardless of whether it was intact or cleaved (Figure 4i).

In the case of the monothiol peptide (CYS1), the application of colorimetric assays was partially successful presumably because of the reduced hydrophilicity/negative charge of the corresponding fragment (Figure 4j). The tetravalent peptide (CYS4) incubated with the citrate-AuNPs led to macroscopic sediments but without color changes. This manifested in a decreased optical density without spectral shape change (see Figure 4k and the inset, Δλ_SPR ~ 9 nm). Such discriminative responses of AuNP aggregation with distinct multiple-thiolate ligands were also reported by others, indicating the fundamental differences in the peptide structures and the resulting interparticle spacing.^58-61 Therefore, we focus on using the divalent peptide for subsequent colorimetric assays (i.e., CYS2 at 100 μM).

Subsequent Protease Detection

The activity of M^pro is essential for the viral proliferation processes and releases mature proteins for functional virus.⁹ The release of M^pro to the respiratory droplets such as saliva (i.e., upper airway from the oral cavity) and exhaled breath (i.e., lower airway from the lung), in principle, should occur in SARS-CoV-2-infected individuals.^62-65 Here, we examined the limit of detection (LoD) of M^pro in four matrices: Tris buffer (TB, 5 mM, pH 8.0), exhaled breath condensate (EBC), saliva (50% dilution), and human plasma (50% dilution).^66,67 The overall DTT level in assays was maintained below 2 μM for the sake of stabilizing both protease and citrate-AuNPs. Citrate-AuNPs were stable in the EBC or saliva samples collected from 10 single donors (n = 10, COVID-negative by PCR). In comparison, these AuNPs aggregated in the corresponding EBC or saliva matrices when spiked with M^pro in the presence of the CYS2 peptide (Figure S4f,g). Though there might be a background signal when the AuNPs are placed in saliva media, the colors between M^pro-blank versus M^pro-spiked saliva were still visually different.

Next, the protease assay was performed by first incubating the CYS2 substrate (100 μM) with various amounts of M^pro spiked in different media (i.e., 0–200 nM, recombinant M^pro without His-tag⁶⁸). The pristine AuNPs (3.2 nM, 100 μL) were then added as a readout at 10 min (Figure 5a). More colloidal aggregation and rapid color change appeared at high concentrations of M^pro and vice versa. Here, the protease LoD was determined to be 12.2 nM in Tris buffer, 18.9 nM in the EBC matrix, and 28.5 nM in diluted saliva (Figure 5b). No color change occurred in human plasma (50% dilution) presumably due to the formation of a stabilizing protein corona on AuNPs.⁶⁹ The clinically relevant level of M^pro in respiratory fluids from COVID-infected patients is still unclear,⁶² but the M^pro LoDs of our system are similar to other reported fluorogenic probes (i.e., at low nanomolar concentrations) but much less time-consuming (~10 min) and with a visual readout.^37,66,70,71

Figure 5.

Determination of sensitivity and specificity. (a) Scheme of stepwise M^pro assay using a divalent peptide, including peptide/protease incubation in different media and subsequent use of AuNPs as the color/absorbance readout. (b) Ratiometric absorbance as a function of M^pro concentration. The CYS2 substrate (100 μM) and the 10 min readout time are applied. The linear region used to calculate LoDs can be found in Figure S5.⁷² Error bar = standard deviation (n = 2). (c) Sensor responsiveness by other mammalian proteins (50 nM), including bovine serum albumin (BSA), hemoglobin, trypsin, thrombin, α-amylase (50 U/mL), and neuraminidase (5 U/mL). Samples with and without M^pro were the positive and negative control. (d) Time progression of the ratiometric signal in inhibitor (i.e., GC376 chemical) assays. The molar ratio of [GC376]/[M^pro] varied from 0:1 to 5:1. (e) Typical inhibition titration curve with a guideline collected by using GC376 chemical.⁷³ The inset shows the structure of GC376. Error bar = standard deviation (n = 2).

A scramble peptide, cys2, was used as the control experiment and showed no aggregation even at the highest concentration of M^pro (200 nM, Figure S4h). Notably, enzymatic cleavage and colloidal aggregation could not occur simultaneously; this is because the fast stabilization of colloidal Au by thiolates takes place before any dithiol (by enzymatic catalysis) links with the particles (Figure S6f). This is also in agreement with the notion that DTT cannot destabilize AuNPs protected by thiolate ligands (Figure 2).

We further cross-tested whether other related analytes can off-site activate our sensing system such as BSA, hemoglobin, trypsin (i.e., cleaves at the C-terminus of Arg or Lys), thrombin (i.e., cleaves the Arg─Gly bond in fibrinogen), salivary α-amylase (i.e., digests α-1,4-glucosidic bonds in starch), and viral neuraminidase (related to influenza viruses).^74,75 Figure 5c shows that only the positive control (with 50 nM M^pro) had distinguishable optical signal changes due to the CYS2 fragment. Neither particle aggregation nor changes of the optical signal were measured in the presence of mammalian proteins (e.g., BSA and hemoglobin) or other enzymes (e.g., α-amylase, thrombin, trypsin, and neuraminidase). This indicates the remarkable selectivity and specificity of our sensor to the SARS-CoV-2 virus, which could be used to detect a wide range of clinical-relevant samples. In addition, a covalent inhibitor (GC376) was used to explore the effect of varying inhibitor molarity (i.e., 0–250 nM) on M^pro (50 nM) and its substrate (CYS2 peptide, 100 μM).^73,76 Figure 5d shows that the ratiometric absorbance was greatly reduced with increasing inhibitor concentration due to the suppressed activity of M^pro. Examination of the optical signals at 15 min yields a typical inhibitor titration curve: The extrapolated line suggests that the amount of active M^pro was 38 nM (or 76% of the stock solution) (Figure 5e).⁷⁷

This approach is simple and offers colorimetric detection at the point of care. Tests like these can be easily affixed to routine face coverings turning them into easy-to-use diagnostic kits (see Figure S7). For example, we have previously demonstrated the design and integration of a simple strip architecture with several mask types for amylase detection by visualizing the colors.⁷⁵ In related work, the SARS-CoV-2 virus and viral RNA were directly detected from the exhaled breath (or coughs) and serum in COVID-19-positive individuals.^62,65,78 Recently, the IgG antibodies specific for M^pro were also reported to be detected in saliva and serum.⁶² Our ongoing work focuses on the discovery and quantification of the M^pro level in the exhaled breath and saliva from the tracked COVID-19-positive patients. As suggested in Figure 5b, the presence of >20 nM M^pro marker in the sample leads to obvious color changes and could complement the existing detection portfolio.

CONCLUSIONS

In this study, we designed and tested a peptide with a new formula, (Asp)₃–(AA)_x–(Cys)_n, to modulate the dispersity of gold colloids as a readout for rapid M^pro detection. Four peptides with distinct cysteine valence were compared in terms of their value in colorimetric sensing. We found that the fragments of divalent CYS2 breakdown by M^pro (i.e., dithiol) strongly aggregate the AuNPs via thiol-to-Au bridging, thus yielding strong color changes. Based on the DTT test, we selected citrate- and TCEP-capped AuNPs as the colorimetric agents. We then determined the sensor LoDs to be 12.2 nM in Tris buffer, 18.9 nM in exhaled breath, and 28.5 nM in saliva (50% dilution). Remarkably, we cross-tested the responsiveness of this sensing system to many related mammalian proteins and enzymes and found no nonspecific and off-site activations; the scramble control was also negative. Additionally, this peptide and Au-based colorimetric assay does not require bioconjugation techniques or sophisticated instrumentation. These advantages may allow us to incorporate the reagents into a portable diagnostic strip and integrate it onto face coverings for designing a practical and noninvasive alternative to PCR-based COVID-19 diagnostic testing.

EXPERIMENTAL SECTION

Peptide Synthesis

Peptides were chain-assembled by Fmoc-SPPS (solid-phase peptide synthesis) on Rink Amide resin (0.67 mmol/g, 200 mg) using an automated Eclipse peptide synthesizer (AAPPTec, Louisville, KY). Amino acid couplings were performed with Fmoc-amino acid (5 equiv), 0.4 M HBTU in DMF (5 equiv), and DIPEA (7.5 equiv). The number of coupling cycles followed the sequence analyzer built in the peptide synthesizer. Finished peptides on the resin were transferred into a syringe filter, washed with five rounds of DCM (~5 mL), and dried under vacuum. Then, peptides were cleaved from the resin using a cleavage cocktail (3 mL) that contained TFA (82.5%), EDDET (2.5%), phenol (5%), thioanisole (5%), and H₂O (5%). Resins were treated with the cleavage cocktail for 2 h. After cleavage, the resin was filtered and the filtrate containing the crude peptides were precipitated/washed with three rounds of cold ether (15 mL, −20 °C), suspended in 50% ACN/H₂O (5 mL), and lyophilized. Peptides were purified by the reversed phase HPLC, characterized by the ESI-MS, aliquoted (ε₂₀₅ ~ 74.6 mL·mg⁻¹ cm⁻¹), and stored under dry conditions at −20 °C for use.

Colloidal Gold Synthesis

The citrate-stabilized AuNPs (~13 nm diameter, TEM) were prepared using the Turkevich method by rapidly injecting an aqueous solution of sodium citrate tribasic dihydrate (150 mg, 5 mL) into an aqueous solution of HAuCl₄·3H₂O (45 mg, 300 mL) under boiling conditions and vigorous stirring. The reaction mixture was left boiling while stirring for another 20 min and then cooled down to room temperature. The deep-red dispersion was then purified by applying one round of centrifugation at 18,000 g for 30 min and the pink supernatant was discarded. The resulting pellet of citrate-AuNPs was redispersed in DI water by sonication and stored at ambient conditions (ε₅₂₀ = 4.0 × 10⁸ M⁻¹ cm⁻¹).

Surface Functionalization

The ligand solution (in PB buffer 5 mM, PH 7.5) was added to a vial containing citrate-AuNPs (10 mL, 3.4 nM) equipped with a stir bar. This yields a molar ratio of ligand (or anchor)-to-AuNP = 130,000. The mixture was stirred at 1600 rpm for 18 h at room temperature. The resulting gold colloids were precipitated once by centrifugation at 18,000 g for 40 min. The supernatant was discarded, and the pellet was redispersed in 10 mL of DI water and stored at 4 °C for further use. Remark: note that modified ligand exchange procedures were applied for a few ligations, e.g., the addition of MPS/MPA/LA-PIMA-PEG ligands to Au dispersion was completed under sonication conditions; the ligation time for TCEP, MPS, and MPA was shortened to 5 h, 3 h, and 30 min, respectively; and the removal of excess TCEP/MPS/MPA ligands was done using a membrane filtration device (M_w cutoff =100 kDa) at 5000 rpm for 5–10 min. The methods of gel electrophoresis, peptide coating, dye conjugation, and stability test are given in the Supporting Information, Sections 2 and 3.

LoD Measurement

The M^pro of a desirable amount were spiked into one of the below media Tris buffer (5 mM, pH 8.0), EBC, pooled human saliva (50% dilution), and human plasma (50% dilution) to reach a final c_enzyme at 0, 1, 2, 5, 10, 15, 20, 30, 50, and 100 nM (i.e., with respect to 140 μL volume). Then, the CYS2 peptide (20 μL, 616 μM) was added to the above mixtures in microtubes and the total assay volume was brought to 40 μL by adding the corresponding media. Next, the above mixtures were tapped, centrifuged, and incubated at 37 °C for 1 h. After this time, the assay was transferred into a 96-well plate and incubated with citrate/TCEP-AuNPs (100 μL, 3.4 nM). The absorbance at 600 and 520 nm was read out in a microplate reader every 1 min for 1 h. The ratiometric signal (λ_600/520, or aggregation factor) at 10 min was extracted for LoD analyses. The LoD was calculated using a statistical method previously reported in the literature:⁷²

$$\begin{gathered} {\text{LoD}}_{\text{int}}={\text{mean}}_{\text{blank}}+1.645\times ({\text{SD}}_{\text{blank}}) \\ +1.645\times ({\text{SD}}_{\text{low concentration sample}}) \end{gathered}$$