Enzyme Sensing Using 2-Mercaptopyridine-Carbonitrile Reporters and Surface-Enhanced Raman Scattering.

The high sensitivity and functional group selectivity of surface-enhanced Raman scattering (SERS) make it an attractive method for enzyme sensing, but there is currently a severe lack of enzyme substrates that release SERS reporter molecules with favorable detection properties. We find that 2-mercaptopyridine-3-carbonitrile ( o-MPN) and 2-mercaptopyridine-5-carbonitrile ( p-MPN) are highly effective as SERS reporter molecules that can be captured by silver or gold nanoparticles to give intense SERS spectra, each with a distinctive nitrile peak at 2230 cm-1. p-MPN is a more sensitive reporter and can be detected at low nanomolar concentrations. An assay validation study synthesized two novel substrate molecules, Glc-o-MPN and Glc-p-MPN, and showed that they can be cleaved efficiently by β-glucosidase (K m = 228 and 162 μM, respectively), an enzyme with broad industrial and biomedical utility. Moreover, SERS detection of the released reporters ( o-MPN or p-MPN) enabled sensing of β-glucosidase activity and β-glucosidase inhibition. Comparative experiments using a crude almond flour extract showed that the presence of β-glucosidase activity could be confirmed by SERS detection in a much shorter time period (>10 time shorter) than by UV-vis absorption detection. It is likely that a wide range of enzyme assays and diagnostic tests can be developed using 2-mercaptopyridine-carbonitriles as SERS reporter molecules.

Introduction

Enzyme assays are ubiquitous in biomedical research, clinical diagnostics, drug discovery, and environmental monitoring.[1−3] The range of assay designs and experimental conditions continues to expand, creating an ongoing need to develop new ways of detecting enzymes and evaluating enzyme inhibition. Classic enzyme assays use optical techniques to monitor changes in color,[4] electronic absorption,[5] fluorescence,[6] or chemiluminescence.[7] While these methods are broadly useful, they have application-specific drawbacks such as insensitivity, background interference, high cost, or long preparation time.[8,9] These technical concerns have been motivating research efforts to develop enzyme assays with alternative detection strategies.[10,11] Advances in nanotechnology have led to new classes of nanoparticles with plasmonic properties that have great promise for exploitation within next-generation enzyme assays.[12,13] Surface-enhanced Raman scattering (SERS) is attracting a lot of attention because of its high sensitivity and functional group selectivity.[14−16] In principle, SERS is capable of enhancing the Raman scattering of the reporter molecules adsorbed on rough metallic surfaces by a factor of ∼106–1010.[17,18] The metallic surface is often in the form of silver nanoparticles (AgNPs) or gold nanoparticles (AuNPs), which are easily prepared as monodispersed suspensions or aggregated clusters.[19]

Although the idea of incorporating SERS detection into bioanalytical methods, including enzyme assays, has been discussed for two or more decades,[17,20−22] there has been a limited applied impact. A major roadblock slowing the development of effective SERS enzyme assays is the challenge of designing suitable enzyme substrate molecules with a favorable combination of properties, including (a) rapid and selective cleavage by a target enzyme and (b) release of a reporter molecule that can be captured by AgNPs and made to elicit a large SERS response.[23] To date, only a few enzyme substrates have been reported based on a small number of releasable SERS reporter molecules such as azo, coumarin, or naphthol dyes.[17,24,25] We decided to expand the structural scope in a new direction and develop a novel set of versatile SERS reporter molecules. We were drawn to the possibility of enzyme substrates that release derivatives of 2-mercaptopyridine, which are known to have very high affinity for AgNPs and produce distinctive SERS spectral patterns.[26−30] However, there was very little literature precedence indicating if, and how, an enzyme substrate could be designed to release a suitable 2-mercaptopyridine derivative for SERS detection.

Here, we report two modified versions of 2-mercaptopyridine as highly effective SERS reporters that are captured by AgNPs with very high affinity. The compounds include a nitrile group whose vibrational stretching band is readily apparent at 2230 cm–1 within the open window that has very little interference from other vibrational bands (Scheme a).[23] Moreover, we describe a protype SERS-active enzyme assay, which compares enzymatic cleavage of two new substrates, (2-mercaptopyridine-3-carbonitrile β-d-glucoside) (Glc-) and 2-mercaptopyridine-5-carbonitrile β-d-glucoside (Glc-), by the enzyme β-glucosidase (β-Glcase), to give the SERS reporters and , respectively (Scheme b). We chose to focus on β-Glcase sensing for several related reasons. It is an industrially important enzyme that is distributed widely throughout the biosphere,[31] and it is employed extensively in the industry for production of food and biofuels.[32−34] It is also a biomarker for several different diseases that require new classes of β-Glcase sensors.[35,36] From a technical perspective, β-Glcase is a good choice for early-stage, validation studies because it commercially available and its substrate selectivity profile is well-defined with reliable benchmark kinetic data (Table S1). We find that Glc- and Glc- are both excellent substrates for the β-Glcase enzyme, although cleavage of Glc- produces a more intense SERS spectrum. The results lead us to infer that or can be broadly applied as reporter molecules for many bioanalytical applications using SERS detection.

Scheme 1

(a) Capture of 2-Mercaptopyridine-carbonitrile Reporter Molecules by Aggregated AgNPs (or AuNPs) Enables Detection Using SERS and (b) Structure and Enzymatic Cleavage of the Substrates Glc- or Glc- by the β-Glucosidase Enzyme Gives the SERS-Active Reporter Molecules or , Respectively

Experimental Section

Synthesis and Characterization of SERS Substrates

Substrate synthesis and compound characterization are described in the Supporting Information. In short, Glc- was synthesized in 98% yield through a modified Koenig–Knorr glycosidation between 2,3,4,6-tetra-o-acetyl-α-d-glucopyranose bromide and 2-mercaptopyridine-3-carbonitrile () using 2 molar equivalents of cesium carbonate in acetonitrile. The acetylated intermediate was purified by column chromatography (silica gel) and then deacetylated using aqueous triethylamine to produce Glc- in 89% yield. The substrate Glc- (2-mercaptopyridine-5-carbonitrile β-d-glucoside) was synthesized in one step by conducting a nucleophilic aromatic substitution reaction. Equimolar equivalents of 1-thio-β-d-glucose sodium salt and 2-chloro-5-cyanopyridine were stirred overnight in methanol to produce Glc- in 38% yield after purification by column chromatography. Chemical stability studies revealed that stock supplies of Glc- and Glc- are highly stable when stored as dry powders (Figure S2).

Enzyme Studies Using UV–Vis Absorption

Stock solutions of 1 mM Glc- and Glc- were prepared in distilled water and used within a few hours. A 1 mg/mL stock solution of commercial β-glucosidase (purchased from Sigma who derived it from almonds) was prepared in 1X PBS buffer, pH 5.33, and stored at 2–8 °C. An aliquot of Glc- (50 μM) and Glc- was added in a cuvette containing 1× PBS buffer, pH 5.33, followed by the addition of 200 μg/mL β-glucosidase. The final 1 mL reaction mixture was mixed by inversion five times before monitoring the change in the absorption spectra over an hour. Enzyme inhibition studies were also performed using castanospermine (CAT) (10 μg/mL), a plant alkaloid inhibitor of β-glucosidase (purchased from Sigma). The order of addition was Glc- or Glc- (50 μM), CAT (10 μg/mL), and β-glucosidase (200 μg/mL). The Michaelis–Menten kinetic parameters were determined in 1× PBS buffer, pH 5.33, at room temperature by varying the concentration of Glc- (10–500 μM) or Glc- (5–200 μM) in the reaction solutions. β-Glucosidase (100 μg/mL) was added to each reaction, and the change in absorption at 300 nm for the appearance of and 312 nm for the appearance of was measured at 5 min intervals over an hour. A Lineweaver–Burk plot was created, and the Michaelis–Menten constant (Km) and maximum rate (Vmax) were calculated.

Synthesis and Characterization of Silver (Ag) Nanoparticles

AgNPs were prepared using the Lee and Meisel method.[37] Briefly, 91 mg of silver nitrate in 500 mL of water was brought to boiling under continuous stirring. Then, 10 mL of 1% (w/v) sodium citrate was added dropwise and boiled with stirring for 30 min. Upon cooling to room temperature, the colloidal dispersion was diluted to 1 L. Transmission electron microscopy (TEM) analysis of the AgNPs revealed an average particle diameter of roughly 50 nm, and UV–vis absorption showed λmax = 405 nm (Figure S6).

Synthesis and Characterization of Gold (Au) Nanoparticles

Quasispherical AuNPs were synthesized according to a modified Frens method.[38] Briefly, 180 mL of 10–2% (by wt) tetrachloroaurate trihydrate salt solution was brought to boiling. Then, 1.2 mL of 1% (by wt) trisodium citrate was added rapidly under vigorous stirring. Boiling was continued for 30 min before the deep red suspension with hues of yellow was allowed to cool to room temperature. SEM analysis of the AuNPs revealed an average particle diameter of 60 nm, and UV–vis absorption showed λ = 540 nm (Figure S12).

Enzyme Studies Using SERS

For all SERS studies involving AgNPs, an aliquot of Glc- or Glc- (10 μM) was added to a 1 mL vial containing preaggregated AgNPs (monodisperse NPs aggregated by addition of 1 M NaBr) and 1× PBS buffer, pH 5.33, followed by addition of β-Glcase (200 μg/mL). For the enzyme inhibition studies, CAT (5 μg/mL) was added prior to the addition of β-Glcase. SERS spectra were measured with a custom-built Raman setup using a 633 nm HeNe laser (Thor Labs). The laser was focused onto the sample using an inverted microscope objective (Nikon, 20×, NA = 0.5) with 120 μW power, measured at the sample. The backscattered radiation was passed through a Rayleigh rejection filter (Semrock) and then dispersed with a spectrometer (Princeton Instruments Acton SP2300, grating = 600 g/mm). Light was detected using a back-illuminated, deep depletion CCD camera (PIXIS, Spec-10, Princeton Instruments) and recorded using Winspec32 software (Princeton Instruments) with a typical acquisition time of 60 s. All experiments were conducted in triplicate.

The studies that compared UV–vis and SERS detection (Figure ) used the following conditions. The UV–vis assay tracked the changes in the spectral profile for a single 1 mL solution of Glc- (50 μM) + almond flour extract (protein concentration of 200 μg/mL) in 1× PBS buffer, pH 5.33. The SERS assay employed a 2 mL solution of Glc- (10 μM) + almond flour extract (protein concentration of 200 μg/mL) in buffer and removed a 1 mL aliquot at two incubation time points (1 and 24 h). Each 1 mL aliquot was added to a vial containing preaggregated AuNPs (preaggregated by addition of 1 M NaBr solution to monodisperse AuNPs), and the SERS spectrum was acquired.

Figure 5

(a) Representative UV–vis spectra of 50 μM Glc- before and after 1 h or after 24 h of incubation with crude almond flour extract (protein concentration of 200 μg/mL). (b) Representative SERS spectra of 10 μM Glc- before and after 1 h or after 24 h of incubation with a crude almond flour extract (protein concentration of 200 μg/mL). In all cases, the assay solution was 1× PBS, pH 5.33.

Almond Flour Extract

Food-grade almond flour was purchased from a local baker and washed three times with ethyl acetate and two times with acetone to remove lipids and water. The powder was then dried immediately in a vacuum desiccator and stored at 4 °C. Afterward, 1 g of defatted flour was added to 25 mL of PBS (50 mM, pH 7.0). The supernatant (crude almond flour extract) was collected after centrifugation (room temperature, 4.4 rpm, 15 min). A Bradford assay was performed using BSA as a standard to determine the protein concentration in the crude extract (Figure S14), and a sample of the crude extract solution with a protein concentration of 200 μg/mL was subsequently tested, by UV–vis or SERS assay, for the capacity to cleave Glc- (β-Glcase activity).[39]

Results and Discussion

The results of UV–vis absorption assays (Figure ) show that β-Glcase catalyzes the cleavage of Glc- or Glc-, releasing or , respectively, which has red-shifted absorption bands. In the case of Glc-, the cleavage reaction was confirmed by thin-layer chromatography analysis of the assay solution (Figure S3). Enzyme inhibition studies were performed to confirm that the enzyme is responsible for the substrate cleavage. As shown by the UV–vis spectra in Figure S5, addition of the known β-Glcase inhibitor CAT to a solution containing Glc- or Glc- followed by addition of the enzyme greatly slowed the appearance of absorption bands corresponding to the cleavage products, or .

Figure 1

Representative UV–vis absorption spectra and Lineweaver Burk plots for solutions containing 50 μM Glc- or Glc- + 100 μg/mL β-glucosidase (β-Glcase) over a one-hour period in 1× PBS Buffer pH 5.33. (a,b) Glc- and (c,d) Glc-.

Enzyme efficiency was quantified by conducting a series of kinetic assays that determined the Michaelis–Menten kinetics for both substrates (Figure S4). The measured values of Km and Vmax for Glc- and Glc- are similar to the values reported in the literature for other β-Glcase substrates (Table S1). The Km for Glc- (228 μM) was slightly higher than the Km for Glc- (162 μM), but Glc- exhibited a slightly higher turnover number and catalytic efficiency (kcat = 0.33 s–1 and kcat/Km = 1445 M–1 s–1) than Glc- (kcat = 0.21 s–1 and kcat/Km = 1300 M–1 s–1). Combined, these results indicate that Glc- has a slightly higher affinity for the active site of the β-Glcase enzyme; however, the enzyme active site can more efficiently convert Glc- to than Glc- to .

The efficient cleavage of Glc- and Glc- by β-Glcase is a remarkable finding since aryl thioglycosides are known to resist the action of glucosidase enzymes. Indeed, thioglycosides are often prepared and evaluated as nonreactive glucosidase inhibitors.[40] However, substrate cleavage has been reported before when the glucoside leaving group is a thiol-substituted heterocycle with an ortho nitrogen atom.[41] This reactivity trend suggests that the pyridyl nitrogen in Glc- and Glc- is protonated in the β-Glcase active site, which activates C–S bond cleavage as illustrated in Scheme a.

Scheme 2

(a) Proposed Active Site Catalysis of Glc- and Glc- Cleavage by the β-Glcase Enzyme and (b) Tautomers of and

An absorption titration experiment determined the pKa of to be 6.46 (Figure S1); thus, it favors the acid form at the working pH of 5.33 used throughout this study. It is worth noting that the acidic forms of 2-mercaptopyridine compounds exist in a tautomeric equilibrium as illustrated in Scheme b.[42−44] The “thione” tautomer, shown on the left of the equilibrium arrows, is the more prevalent species in aqueous solution; however, both tautomers are very likely to form the same Ag-bonded structure when or is captured by the surface of the AgNP as shown in Scheme a.[28,29]

After proving that Glc- and Glc- are efficient substrates for the β-Glcase enzyme, the study moved to experiments using SERS analysis. The AgNPs were prepared by standard methods, and TEM images indicated an average particle diameter of 50 nm (Figure S6). Shown in Figure are SERS spectra for separate samples of and captured by AgNPs. The normalized spectra in Figure a show that both reporters produced a peak at 2230 cm–1, which represents stretching of the CN bond. However, at other spectral locations, there are noticeable differences in the peak wavenumber, and especially notable are the very intense peaks at 1064 cm–1 for and at 1096 cm–1 for . This suggests that and have high potential for incorporation into multiplex detection methods that quantify the amount of each reporter in the same sample, and a demonstration of this concept is provided in Figure S11.[45,46] A comparison of the relative intensity spectra in Figure b shows that produces a substantially more intense spectrum than , suggesting that it is likely to be a more useful reporter for a high sensitivity detection assay using a single substrate. Indeed, serial dilution experiments showed that low nanomolar concentrations of can be captured by the AgNPs and easily detected at 1096 cm–1 (Figure ).

Figure 2

SERS spectra of separate samples containing equal amounts of or (10 μM) in the presence of preaggregated AgNPs. (a) Normalized intensity and (b) relative intensity.

Figure 3

SERS spectra of at different concentrations in the presence of AgNPs. The spectra were collected using a 633 nm laser, 1200 g/mm grating, and 120 μW power for an acquisition time of 30 s.

The greater SERS sensitivity of was apparent in enzyme experiments that used AgNPs to capture the or that was released when Glc- or Glc- was cleaved by β-Glcase. The standard conditions for these enzyme experiments added an aliquot of Glc- or Glc- (10 μM) to a vial containing preaggregated AgNPs in PBS buffer, pH 5.33, followed by addition of β-Glcase (200 μg/mL) in the presence or absence of the CAT inhibitor (10 μg/mL). Control experiments proved that the SERS spectra for the enzyme-cleaved substrates matched the spectra for separate samples of AgNPs with added amounts of authentic or (Figure S7), thus confirming that the SERS spectra were reporting capture of the released SERS reporter molecules by the AgNPs. In the case of Glc-, the intensity of the SERS spectrum before enzyme addition was very weak, and there was a large increase in signal intensity once β-Glcase had cleaved all the substrate and the AgNPs had captured all the released (Figure S8a). As expected, much less was produced over the standard 1 h incubation period when the assay was repeated in the presence of the inhibitor CAT (Figure S8b). The bar graph in Figure a shows a 20-fold increase in the SERS peak area upon complete cleavage of the substrate and substantial reduction of the peak area when β-Glcase was strongly inhibited by CAT. Similar SERS changes were also observed with Glc- (Figure S9), but enzymatic cleavage of the substrate to release as the reporter molecule only produced a sixfold increase in the CN peak area at 2230 cm– 1 (Figure b).

Figure 4

(a) (Left) Representative SERS spectra of 10 μM Glc- after incubation for 1 h in the presence or absence of 200 μg/mL β-Glcase and 5 μg/mL CAT (1:8 molar ratio) in 1× PBS, pH 5.33, and (Right) bar graph showing the average peak area at 2230 cm–1. (b) (Left) Representative SERS spectra of 10 μM Glc- after incubation for 1 h in the presence or absence of 200 μg/mL β-Glcase and 5 μg/mL CAT in 1× PBS, pH 5.33, and (Right) bar graph (N = 3) showing the average peak area at 2230 cm–1. The peak area has units of counts mW–1 s–1 cm–1.

The selectivity of this SERS-based enzyme assay was tested by conducting a set of experiments that mixed separate samples of Glc- with high amounts of other analytes that might have affinity for the AgNPs and thus possibly produce an artifact (i.e., bovine serum albumin, cysteine, glutathione, glycine, or lipase).[47] As shown in Figure S10, only β-Glcase produced a large SERS signal corresponding to released . Additional experiments proved that the SERS-based assay worked equally well with preaggregated AuNPs (Figure S13).

The corollary of the high sensitivity gained by SERS detection is the possibility of diagnostic tests that can detect enzyme activities in a much shorter time period compared to detection by UV–vis absorption. We tested this idea by performing an experiment that compared the relative capabilities of the SERS and UV–vis detection methods to report positive evidence for cleavage of Glc- by β-Glcase. The comparative experiment employed a sample of crude almond flour extract and mimicked a real-world circumstance, namely quality control screening of a food source for evidence that it contains a desired threshold level of flavor-enhancing β-Glcase activity.[32−34] Since the β-Glcase activity in a typical crude almond flour extract is quite low, a practically useful diagnostic test needs to detect the appearance of with high sensitivity.

Two separate samples from the same stock solution of crude almond extract were tested for capacity to cleave Glc- and produce (β-Glcase activity). One sample was assessed by UV–vis, and the other was assessed by SERS. In both cases, the spectrum for a solution of Glc- was acquired before extract addition and again at 1 or 24 h after the addition of the crude almond extract (protein concentration of 200 μg/mL). Inspection of the UV–vis spectra in Figure a shows that appearance of the band at 312 nm (or 360 nm) was not very clear after 1 h but was quite apparent after 24 h, with both spectra exhibiting significant background absorption at <300 nm due to other proteins present in the crude extract. In contrast, the SERS spectra in Figure b show that appearance of the peak 2230 cm–1 was very apparent after 1 h with no evidence of any background SERS signal due to other proteins. The results clearly show that the presence of β-glucosidase activity in the crude almond extract is confirmed by SERS detection in a much shorter time period (>10 times shorter) than by UV–vis absorption detection. Obviously, this substantial time saving would greatly facilitate any industrial food production process that must perform a large number of quality assurance measurements to confirm the threshold levels of β-glucosidase activity.

Close inspection of the SERS data in Figures and 5 suggests that dependence of the SERS signal intensity on the concentration is not perfectly linear, which we attribute to the use of preaggregated nanoparticles for the SERS measurements. Thus, future work to optimize the assays will need to develop a suitable immobilized metal surface for SERS detection, in particular a surface with a more uniform and reproducible morphology that enables improved quantitative analysis. This is an ongoing community-wide research challenge that must be solved before SERS-based assays will be routinely used for quantitative enzyme detection.[25,48,49]

Conclusions

The two 2-mercaptopydrine-carbonitrile compounds, or , have great potential as reporter molecules for efficient capture by silver or AuNPs and detection by SERS. The specific focus of this study was on enzyme sensing, and the results show that Glc- and Glc- are efficient substrates for the β-Glcase enzyme, which enable SERS detection of β-Glcase activity and β-Glcase inhibition. The substrate Glc- is superior because the released is detected with higher sensitivity. Comparative sensitivity experiments using an almond flour extract showed that the presence of β-glucosidase activity in the crude extract could be confirmed by SERS detection in a much shorter time period (>10 times shorter) than by UV–vis absorption detection. With further development, it is possible that β-Glcase assays using Glc- will be broadly useful in environmental science[31] and disease diagnosis.[35,36] SERS detection will be especially helpful with heterogeneous samples such as saliva[50] or bacterial cell culture,[33] where optical assays fail due to strong scattering of light and background interference.

Beyond the specific application of β-Glcase sensing, it is very likely that many other enzyme assays can be developed using released 2-mercaptopydrine-carbonitriles as SERS reporter molecules. In addition to single-use assays, it should be possible to spatially pattern the AgNP capture agent in an array format that enables high-throughput screening of enzyme inhibitors for drug discovery.[8,48] Alternatively, enzyme–antibody conjugates can be developed for enzyme-catalyzed signal amplification with SERS detection in ELISA-based diagnostics.[51] The narrow peaks and high signal dispersion of SERS spectra favor multiplex diagnostics and imaging, which raises the possibility of additional 2-mercaptopydrine-carbonitrile compounds, beyond and , as reporter molecules with distinct spectral SERS signals that can act as detection barcodes.[52]