Paired Spectroscopic and Crystallographic Studies of Proteases

Abstract

The active sites of subtilisin and trypsin have been studied by paired IR spectroscopic and X-ray crystallographic studies. The active site serines of the proteases were reacted with 4-cyanobenzenesulfonyl fluoride (CBSF), an inhibitor that contains a nitrile vibrational reporter. The nitrile stretch vibration of the water-soluble inhibitor model, potassium 4-cyanobenzenesulfonate (KCBSO), and the inhibitor were calibrated by IR solvent studies in H₂O/DMSO and the frequency-temperature line-slope (FTLS) method in H₂O and THF. The inhibitor complexes were examined by FTLS and the slopes of the best fit lines for subtilisin-CBS and trypsin-CBS in aqueous buffer were both measured to be −3.5×10⁻² cm⁻¹/°C. These slopes were intermediate in value between that of KCBSO in aqueous buffer and CBSF in THF, which suggests that the active-site nitriles in both proteases are mostly solvated. The X-ray crystal structures of the subtilisin-CBS and trypsin-CBS complexes were solved at 1.27 and 1.32 Å, respectively. The inhibitor was modelled in two conformations in subtilisin-CBS and in one conformation in the trypsin-CBS. The crystallographic data support the FTLS data that the active-site nitrile groups are mostly solvated and participate in hydrogen bonds with water molecules. The combination of IR spectroscopy utilizing vibrational reporters paired with X-ray crystallography provides a powerful approach to studying protein structure.

Graphical Abstract

Temperature-dependent infrared spectroscopy and X-ray crystallography were used in conjunction with a nitrile-modified inhibitor to probe the active site of subtilisin and trypsin.

INTRODUCTION

Proteases are ubiquitous enzymes that efficiently cleave amide bonds in proteins and peptides and are involved in such diverse biological functions as angiogenesis, apoptosis, blood coagulation, cell differentiation, and immunity.^[1,2] From a mechanistic perspective, they are divided into six classes, metalloproteases, serine, cysteine, aspartic, threonine, and glutamic proteases.^[3] There are over 500 human proteases and homologs and serine proteases comprise almost one-third of these enzymes.^[4] The serine proteases are found in diverse species from mammals to bacteria, and yet many share the serine-histidine-aspartate catalytic triad, making these enzymes a prototypical example of independent convergent evolution.^[1] Given their prevalence and importance, proteases have been extensively studied with over 350,000 scientific articles published on them.^[3] However, proteases have not been evaluated by nitrile vibrational reporters^[5] and the frequency-temperature line-slope (FTLS) method^[6–8] to access their active site solvation state. Herein we describe studies of two serine proteases using this method paired with X-ray crystallographic analysis.

Site-specific vibrational reporters - such as nitriles - have been demonstrated to be effective reporters of local biomolecule environments.^[5,9,10] Nitriles are especially effective vibrational reporters due, in part, to the position of the nitrile symmetric stretch vibration in an open region of the IR spectrum, the sensitivity of this vibration to local environment, the relatively strong oscillator strength, the relatively localized transition, and the relatively small size (only two atoms).^[5,11,12] This vibrational reporter has been incorporated into peptides and proteins synthetically, genetically, and post-translationally.

The non-canonical amino acid (ncAA) 4-cyano-L-phenylalanine (pCNF) is a vibrational reporter which has been incorporated into a number of peptides including the MLCK peptide,^[13] the villin headpiece subdomain,^[14] the human islet amyloid polypeptide,^[15] and the amyloid peptide (Aβ16–22),^[16] by standard solid-phase peptide synthesis. Genetic incorporation through the amber codon suppression method has also been used to incorporate pCNF into several protein systems including myoglobin,^[17] the N-terminal domain of the L9 protein (NTL9),^[18] cytochrome c,^[19] superfolder green fluorescent protein (sfGFP),^[8,20,21] the N-terminal Src homology 3 domain of the murine adaptor protein Crk-II (nSH3),^[22] plastocyanin protein (Pc),^[23] and the heme nitric oxide and/or oxygen (H-NOX) protein from Caldanaerobacter subterraneus (Cs H-NOX).^[10] Finally, nitrile vibrational reporters have also been incorporated post-translationally though the conversion of a cysteine to a thiocyanate (SCN) or by using enzyme inhibitors containing nitriles. For example, Boxer first demonstrated that cysteine residues can be converted into thiocyanates using Ellman’s reagent for IR spectroscopic studies of several proteins including ribonuclease S-protein, human aldose reductase (hALR2), and the bacterial photosynthetic reaction center.^[24] This method has also been used to study other peptides and proteins.^[25–28] Boxer and coworkers also studied the active site environment of hALR2 using two different inhibitors containing nitrile groups. The electrostatic fields of wild type and mutant hALR2 were evaluated using the vibrational Stark effect.^[29,30] Hochstrasser and coworkers studied the allosteric hydrophobic binding site of HIV reverse transcriptase with TMC278, a non-nucleoside inhibitor that contains two nitrile groups with two-dimensional IR spectroscopy to measure the dynamics of the interaction of each nitrile group with the enzyme.^[31]

Here we have studied the active site solvation environments of the proteases subtilisin (type VIII from Bacillus licheniformis) and bovine trypsin from Bos taurus pancreas using 4-cyanobenzenesulfonyl fluoride (1, CBSF, Figure 1A), a suicide inhibitor containing a nitrile vibrational reporter (Figure 1) paired with infrared spectroscopy and X-ray crystallography. Subtilisin is a 274-residue serine protease that is 30% helical and 19% β-sheet and trypsin is a 223-residue serine protease that is 10% helical and 34% β-sheet. Both proteases were reacted with CBSF at the active site serine residue. The sensitivity of the nitrile symmetric stretch of water-soluble analogue potassium 4-cyanobenzenesulfonate (2, KCBSO, Figure 1) and CBSF to solvent and temperature were investigated. The solvents were selected to mimic various protein environments and the temperature-dependent measurements provide a calibration for the application of the frequency-temperature line-slope (FTLS) method to more accurately assess the local solvation state.^[6–8,10] The FTLS method is based upon the differential temperature sensitivity of the nitrile symmetric stretch when the reporter is involved in hydrogen bonds. The protease-inhibitor complexes were studied by the FTLS method to assess the solvation state of the active sites and these results were correlated with structural data determined from the X-ray crystal structures of the protease-inhibitor complexes.

Figure 1.

Structure of the inhibitor, 4-cyanobenzenesulfonyl fluoride (1) and potassium 4-cyanobenzenesulfonate (2).

RESULTS AND DISCUSSION

Synthetic Chemistry

The synthesis of 2 (KCBSO) was accomplished in good yields by the Sandmeyer reaction of 4-benzenediazonium sulfonate with CuCN generated in situ from CuSO₄, sodium ascorbate, and KCN (Scheme 1). This water soluble salt was needed as a model of the temperature-dependence of inhibitor-enzyme complex in an aqueous environment. To confirm the band assignment, the ¹³C-isotopomer was also prepared by the same reaction but using ¹³C-labeled Cu¹³CN. Inhibitor 1 (CBSF) is soluble in THF and thus served as the model for temperature-dependence in a non-hydrogen bonding environment.

Scheme 1.

Synthesis of K CBSO and K ¹³CBSO using copper (I) cyanide.

Infrared Spectroscopy

Nitrile Stretch Band Assignment of KCBSO.

The room-temperature IR spectrum of KCBSO dissolved in water shows a single, symmetrical absorbance band centered at 2240.3 cm^–1 resulting from the nitrile symmetric stretching vibration (Figure 2). This assignment was based upon the position of the nitrile IR absorbance band of other nitrile-containing molecules, such as 4-cyano-L-phenylalanine (pCNF),^[13,20] and confirmed through isotopic editing of the nitrile group. Specifically, a ¹³C-labeled KCBSO (K¹³CBSO), was synthesized and the resulting room-temperature spectrum in water is also shown in Figure 2. The IR spectrum of K¹³CBSO shows a single, symmetric absorbance band centered at 2188.1 cm^–1. Thus the absorbance band centered at 2240.3 cm^–1 for KCBSO has red-shifted 52.2 cm^–1 upon ¹³C-labeling of the nitrile group. The direction and magnitude of the isotopic shift is consistent with the DFT gas-phase predicted shift of 53.1 cm^–1 at the B3LYP/6–311++G(3df,3pd) level for CBSO and is consistent with the experimentally measured isotopic shift of pCNF^[20] upon ¹³C labeling of the nitrile group (53.2 cm^–1).

Figure 2.

FTIR spectra of KCBSO and K¹³CBSO dissolved in water at a concentration of 250 mM in the region 2140 – 2280 cm^–1 recorded at 25 °C. The spectra were intensity normalized and baseline corrected.

Solvent Sensitivity of Nitrile Stretch of KCBSO.

The sensitivity of the nitrile stretching frequency of KCBSO was explored in several solvents selected to mimic various local protein environments. Specifically a series of dimethyl sulfoxide (DMSO) / water mixtures were utilized to represent either solvated or de-solvated local environments similar to previous work on other nitrile containing molecules such as pCNF^[20,21] and 2’-azido-5-cyano-2’-deoxyuridine (N₃CNdU).^[32] The IR spectra of KCBSO in water, DMSO, and three DMSO / water mixtures each show a single band for the nitrile stretch vibration (Figure 3). The nitrile stretch vibration shifts monotonically from 2227.1 cm⁻¹ (pure DMSO) to 2240.3 cm⁻¹ (pure water) as the DMSO percentage decreases (see Figure S1 in the Supporting Information for details of the line shape analysis). This 13.2 cm⁻¹ blue shift is the result of hydrogen bonding between solvent water molecules and the nitrile group of KCBSO, which are absent in the DMSO solution. The direction of this shift is similar to previous literature studies of the effect of hydrogen bonding on the nitrile stretching frequency of pCNF and N₃CNdU, although the magnitude of the shift is slightly larger than the observed shift in pCNF (10.9 cm⁻¹),^[20] and N₃CNdU (12.5 cm⁻¹)^[32] suggesting a greater sensitivity of the nitrile stretching frequency of KCBSO. The blue shift in the nitrile stretch frequency going from DMSO to water was also observed for ¹³C-labeled KCBSO as expected (see Figure S2 in the Supporting Information). The dependence of the nitrile stretch was also measured in methanol for CBSF (see Figure S3 in Supporting Information). CBSF instead of KCBSO was utilized for methanol based upon solubility limitations. As expected based upon literature precedent,^[33] the nitrile IR absorbance band of CBSF in methanol consisted of two components where the high frequency component corresponded to the nitrile hydrogen bonding with methanol and the low frequency component was not participating in hydrogen bonding.

Figure 3.

FTIR spectra of KCBSO in DMSO and H₂O mixtures (v/v) at a concentration of 250 mM in the region 2200 – 2270 cm^–1 recorded at 25 °C. The spectra were intensity normalized and baseline corrected.

Probing Protease Active Site Local Environments.

The CBSF inhibitor was then reacted with subtilisin or trypsin to probe the active site environment. Upon reaction with the active-site serine the protein-inhibitor complexes subtilisin-CBS and trypsin-CBS are formed along with one equivalent of hydrogen fluoride as a byproduct. Reaction conditions were selected to maximize inhibitor binding to the active site serine and minimize undesired side-reactions of the inhibitor with other residues in the protein. Mass spectral analysis was utilized to optimize this procedure to achieve the optimal balance between these two goals (see Supporting Information Figure S15). The best conditions utilized 1.0 equivalent of CBSF with subtilisin or trypsin in an aqueous buffer (10 mM Hepes, 20 mM KCl, pH 7.5) at room temperature overnight. X-ray crystallography confirmed the preferential binding of the inhibitor to the active site of both proteases (see below). Unfortunately, nonspecific binding of the inhibitor (i.e., binding to other sites) was also observed with ~40% of the sample containing more than one equivalent of CBS bound to the enzyme. This complicates the analysis of the IR data and is one reason why X-ray crystallographic analysis was coupled with the IR analysis.

The room-temperature IR spectrum of the nitrile stretching frequency of subtilisin-CBS and trypsin-CBS each show a single, symmetrical absorbance band at 2241.4 cm^–1 and 2244.3 cm^–1, respectively (Figure 4). The similarity of these frequencies with the room-temperature nitrile stretching frequency of KCBSO in water suggests that the nitrile group is involved in hydrogen bonding interactions with either the solvent or neighboring residues. However, surprisingly both of these frequencies are blue shifted from the nitrile stretching frequency of KCBSO in water. Specifically, the nitrile stretch vibration in subtilisin-CBS and trypsin-CBS are blue-shifted 1.1 or 4.0 cm⁻¹, respectively. This blue shift suggests either (1) an altered hydrogen bonding geometry or strength of the nitrile with solvent/residues in the active site compared to the nitrile group of KCBSO dissolved in water and/or (2) a different electrostatic environment of the active-site nitrile group compared to the nitrile group of KCBSO in water. X-ray crystallographic evidence suggests that the hydrogen bonding partner of the nitrile group of the bound inhibitor is likely water molecules (see below). The FWHM of the nitrile IR absorbance band is also ~2 cm⁻¹ larger in subtilisin-CBS and trypsin-CBS than KCBSO in water on average (see Tables S1, S3, and S4 in Supporting Information), which is likely due to specific binding of CBS to the active site and nonspecific binding in ~40% of the sample.

Figure 4.

FTIR spectra of CBS bound to subtilisin (A, open circles) or trypsin (B, open squares) dissolved in an aqueous buffer (10 mM Hepes, 20 mM KCl, pH 7.5) in the region 2210 – 2270 cm^–1. The protein concentration was 2.5 mM. The spectra were recorded at 25 °C, intensity normalized, baseline corrected, and fit to a linear combination of a Gaussian and Lorentzian function (solid curves).

The solvation environment of the active-site nitrile group was further assessed through temperature dependent IR spectroscopy. Specifically, the frequency-temperature line slope (FTLS) method was utilized where the sensitivity of the nitrile stretching frequency to temperature is correlated to the local environment. In order to interpret the temperature dependence of the active-site nitrile stretching frequency, the temperature dependence of the nitrile stretching frequency of KCBSO in aqueous buffer and CBSF in tetrahydrofuran (THF) were measured (Figure 5). These serve as models of a fully solvated, high dielectric and de-solvated, low dielectric local protein environments, respectively. The nitrile (KCBSO or CBSF) selected for each solvent based upon solubility limitations. THF was selected here instead of DMSO since the lower dielectric of THF is a more apt model of a buried environment in a protein, while DMSO was selected in Figure 3 due to the solubility of KCBSO in water, DMSO, and mixtures thereof.

Figure 5.

Temperature dependent shifts in the nitrile stretching frequency of CBSF dissolved in THF (open circles) or KCBSO dissolved in an aqueous buffer consisting of 10 mM Hepes and 20 mM KCl at a pH of 7.5 (open squares). The temperature-dependent frequency shifts were fit to a straight line.

Figure 5 shows the temperature dependence of the nitrile stretching frequency of KCBSO (open squares) dissolved in an aqueous buffer (10 mM Hepes, 20 mM KCl, pH of 7.5) and CBSF (open circles) dissolved in THF fit to a straight line. The decreased temperature range employed for the THF measurements compared to water is due to the lower boiling point of THF relative to water. The frequency shifts were referenced to the nitrile stretching frequency recorded at the lowest temperature (20.5 °C) for each solvent. The corresponding temperature dependent IR spectra and the results of the line shape analysis to determine the nitrile stretching frequencies are shown in the Supporting Information (Figures S4 and S5, Tables S1 and S2). The slope of the best fit line for the frequency shift of KCBSO in aqueous buffer was −4.5±0.1×10⁻² cm⁻¹/°C, while the slope of the frequency shift of CBSF in THF was −4.4±2.4×10⁻³ cm⁻¹/°C. The relatively large (an order of magnitude greater) temperature dependence of the nitrile stretching frequency of KCBSO in aqueous buffer is due to the sensitivity of the nitrile stretch vibration to the geometry of hydrogen-bonding interactions between the nitrile and water molecules.^[34,35] The nitrile stretching frequency of CBSF in THF is nearly independent of temperature due to the lack of specific interactions between the nitrile and the solvent. The direction and magnitude of these temperature dependent frequency shifts are in agreement with previous studies using pCNF.^[7,8]

Figure 6 shows the temperature dependence nitrile stretching frequency shifts of subtilisin-CBS (open circles) dissolved in an aqueous buffer consisting of 10 mM Hepes and 20 mM KCl at a pH of 7.5 fit to a straight line. These data are the result of the average of two temperature dependent measurements. The temperature range was selected to ensure reversibility of the measurements. Representative temperature dependent IR spectra and the corresponding line shape analysis results are shown in the Supporting Information (Figure S6, Table S3). The slope of the best fit line for the nitrile stretching frequency shift of subtilisin-CBS in aqueous buffer was measured to be −3.5±0.5×10⁻² cm⁻¹/°C. This slope is in between the slope of KCBSO in aqueous buffer and CBSF in THF, indicating that the nitrile group of subtilisin-CBS is involved in hydrogen bonds of moderate strength, likely with solvent molecules (see X-ray crystallographic discussion below). Specifically, the magnitude of the subtilisin-CBS line slope is a factor of 1.3 less than the line slope of KCBSO in aqueous buffer and factor of 8.0 greater than the line slope of CBSF in THF. Thus, the IR temperature dependent measurements of subtilisin-CBS suggests that the active-site nitrile is mostly solvated. This FTLS result is a more robust and accurate means of assessing the local solvation environment of the nitrile group compared to using only the room-temperature nitrile stretching frequency of subtilisin-CBS.^[6–8,10]

Figure 6.

Temperature dependent shifts in the nitrile stretching frequency for CBS bound to subtilisin (open circles) dissolved in an aqueous buffer (10 mM Hepes, 20 mM KCl, pH 7.5). The protein concentration was 2.5 mM. The temperature-dependent frequency shifts were fit to a straight line with a slope of −3.5±0.5×10⁻² cm⁻¹/°C.

The corresponding temperature dependent IR spectra, line shape analysis, and temperature dependent nitrile stretching frequency shifts for trypsin-CBS are shown in Supporting Information (Figures S7 and S8, Table S4). The FTLS results showed the same line slope as subtilisin-CBS, suggesting the nitrile group is mostly solvated in the trypsin-CBS complex. However, these trypsin temperature dependent measurements suffered from a lack of reversibility in most trials. The FTLS analysis for trypsin-CBS reported in the supplemental and referenced above corresponds to a set of measurements that demonstrated reversibility. The lack of reversibility for trypsin attributed to the autolytic nature of this protease^[36] and mass spectral results show that some free protease is still present when one equivalents of the inhibitor is added to the protease solution. Excess inhibitor was not used for the IR experiments to minimize reaction of CBSF with serine residues other than the active site serine (see below). The autolytic propensity of this residual free trypsin explains the irreversibility in some of the temperature dependent experiments given the time and temperature required for these measurements.

Protein X-ray Crystallography

Formation of the subtilisin-CBS complex was confirmed through X-ray crystallography (Figure 7). Three conformations of the active site residue 220 were modelled: two different conformations of the bound inhibitor (CBS_-A and CBS_-B) and an unreacted serine (SER_-C) (Figure 7B). Even with four equivalents of CBSF added prior to crystallization, SER_-C was modelled into the active site to account for unreacted subtilisin to resolve –|F_o-F_c| difference density around the sulfonates when only CBS_-A and CBS_-B were modeled. The presence of unreacted subtilisin under these reaction conditions was confirmed by mass spectrometry. SER_-C220 was a likely hydrogen bonding partner along with the carbonyl oxygen of SER124 in coordinating a water residue (HOH_C) that would be absent upon inhibitor binding because of steric limitations. Both CBS_-A and CBS_-B were modelled at sites along the surface of subtilisin. Electron density for the CBS_-A conformation indicated the nitrile group was partially solvated and directed towards GLY126, GLY127, and a calcium cation displaying partial occupancy. A potential hydrogen bonding partner along the backbone for CBS_-A was the amide hydrogen of a peptide bond between GLY126 and GLY127, however the orientation of the backbone points the hydrogen away from the cyano-group, reducing the potential for interactions. Density 3.8 Å from the nitrile in CBS_-A was modeled as a Ca²⁺ ion at 58% occupancy as the density was spherical, not fully accounted for by a water molecule, and there was an excess amount of Ca²⁺ in the crystallization condition. CBS_-B was modelled with the nitrile directed towards a solvent exposed environment and within hydrogen bonding distance (2.5 Å) of an ordered water molecule (HOH_B) (Figure 7B and C). It is possible that this water molecule reflects bulk water interactions with the inhibitor within the IR spectroscopic data. It is important to note that the IR experiments were performed with 1.0 equivalents of CBSF, unlike the 4.0 equivalents used in the crystallography work, to maximum inhibitor binding to the active site while minimizing undesired side reactions of CBSF with other serine residues so that the observed nitrile stretch frequency can be correlated with the active site solvation environment.

Figure 7.

Crystal structure of subtilisin-CBS complex. A) subtilisin ribbon structure shown in pink with CBS inhibitor shown in sticks modeled in two conformations at the active site serine 220. B) Zoom in on active site in same orientation as A with modelled CBS inhibitors shown in sticks along with 2F_O - F_C electron density at 1σ shown in blue mesh, water shown as red sphere, and partially occupied calcium shown as green sphere C) Approximately 90° rotation from A/B orientation with protein surface shown in pink and CBS inhibitor shown in sticks.

In the subtilisin-CBS structure the histidine (HIS63) of the catalytic triad occupies two conformations: one pointing towards the active site serine and one pointing away (Figure S10). Similar changes induced by inhibitor binding were observed in previous published crystal structures with HIS63 modelled in two conformations.^[37,38] Moreover, structural alignments of subtilisin-CBS with subtilisin complexes with phenylmethylsufonyl (PMS) or vinyl-PMS, and with unreacted subtilisin all revealed that the overall tertiary structures were similar (Figure 8A).^[37–39] Alignment of the subtilisin-CBS active site with subtilisin-PMS and subtilisin-vinyl-PMS active sites were illustrative in that CBS_-A adopted a similar conformation to vinyl-PMS and CBS_-B adopted a similar conformation to PMS (Figure 8B). While the two highest occupancy conformations of CBS are modeled, the active site serine:CBS complex maintains a great deal of flexibility and the less favorable and lower occupancy conformations are not explicitly illustrated by the two conformations modeled here. This flexibility at the inhibitor-bound active site is supported by the difference density of the final structure (Figure S9B). An ethylene glycol molecule is modeled adjacent to CBS_B (~4 Å away), however this molecule was introduced in the cryoprotection solvent and was not present in the buffer for IR spectroscopy. Therefore, it did not interrupt the electronics nor solvation dynamics of the active site during the IR experiments.

Figure 8.

Comparison of various subtilisin structures. A) Ribbon structure alignment of subtilisin-CBS in pink, wild-type subtilisin from Bacillus Lentus in yellow (PDB ID: 1NDQ, RMSD 0.515Å), subtilisin-PMS in cyan (PDB ID: 3VYV, RMSD 0.306Å), and subtilisin-vinylPMS in grey (PDB ID: 5AQE, RMSD 0.476Å). B) Active site alignment show in sticks for subtilisin-CBS in pink, wild-type subtilisin in yellow, subtilisin-PMS in cyan, and subtilisin-vinylPMS in grey.

Formation of the trypsin-CBS complex was also confirmed through X-ray crystallography (Figure 9). As with subtilisin, four equivalents of CBSF were reacted with trypsin prior to crystallization although the IR experiments were performed with 1.0 equivalents of CBSF to minimize undesired side reactions. The trypsin active site was modelled with a single conformation of the inhibitor, CBS_-A, and the native active site serine, SER_-B (Figure 9B). Mass spectrometry indicated that under these reaction conditions some unreactive trypsin remained so SER_-B was modelled to account for this. The CBS_-A molecule is oriented towards the protein surface and accessible to solvents (Figure 9A and C). This orientation is consistent with the IR spectroscopic data that suggested the nitrile group was in a solvated environment (Figure 5 and S8). Structural alignment of the trypsin-CBS structure to similar phenyl-based inhibitors such as benzamide, PMSF, and benzene-boronic acid revealed that the overall tertiary structure of the protein was not significantly perturbed relative to these structures (Figure 10A).^[40–42] Alignment of the inhibitors at the active site was similar between the trypsin-CBS and trypsin-PMS, while the trypsin-benzamide structure shows the feasibility for other inhibitor conformations in the active site (Figure 10B). Sulfonated serines were modelled in addition to serine at four solvent accessible serines in the structure (sites 55, 121, 131, and 169) to account for additional 2F_o-F_c electron density and +|F_o-F_c| difference density (Figure S14). The highly solvated nature of these three sites resulted in a conformational flexibility for the CBS molecules modeled at these sites and did not allow for the cyanobenzyl component of the inhibitor to be modelled. The sulfonated serine residues were each of relatively low occupancy (37, 53, 44, and 44% sulfonated serine occupancy for sites 55, 121, 131, and 169, respectively) and are potential sites where reaction with the CBSF inhibitor had occurred, consistent with mass spectrometry results indicate more than one equivalent of CBS is bound to the protein when four molar equivalents of CBSF were used. Overall, the crystal structure of trypsin-CBS suggests that the inhibitor is present in primarily one conformation at the active site and is consistent with the solvent accessibility observed in the IR experiments.

Figure 9.

Crystal structure of trypsin-CBS complex. A) Trypsin ribbon structure shown in green with CBS inhibitor shown in sticks modelled at the active site serine 200. B) Zoom in on active site in same orientation as A with modelled CBS inhibitor shown in sticks along unreacted serine with 2F_O - F_C electron density at 1σ shown in blue mesh. C) Approximately 90° rotation from A/B orientation with protein surface shown in green and CBS inhibitor shown in sticks.

Figure 10.

Comparison of various trypsin structures. A) Ribbon structure alignment of trypsin-CBS in green, trypsin-benzamide in yellow (PDB ID: 4I8H, RMSD 0.141Å), trypsin-PMS in cyan (PDB ID: 1PQA, RMSD 0.810Å), and trypsin-benzeneboronic acid in magenta (PDB ID: 2A32, RMSD 0.319Å). B) Active site alignment show in sticks for trypsin-CBS in green, trypsin-benzamide in yellow, trypsin-PMS in cyan, and trypsin-benzeneboronic acid in magenta.

CONCLUSIONS

The active sites of two proteases have been studied by paired IR spectroscopic and X-ray crystallographic studies. The active site serines of subtilisin and trypsin were reacted with CBSF, a protease inhibitor that contains a nitrile vibrational reporter. The nitrile stretch vibration of the inhibitor and its water soluble salt (KCBSO) were calibrated by IR solvent studies in H₂O/DMSO and the FTLS method in H₂O/THF. The nitrile stretch vibration of the subtilisin-CBS and trypsin-CBS complexes were blue-shifted 1.1 or 4.0 cm⁻¹, respectively, compared to the aqueous KCBSO frequency. This suggested a fully solvated environment in both proteases. The inhibitor complexes were examined by FTLS and the slopes of the best fit lines for subtilisin-CBS and trypsin-CBS in aqueous buffer were both measured to be −3.5×10⁻² cm⁻¹/°C. These slopes were intermediate in value between that of KCBSO in aqueous buffer and CBSF in THF. Thus, the more robust and accurate FTLS results suggest that the active-site nitriles in both proteases are mostly solvated. X-ray crystal structure of the subtilisin-CBS and trypsin-CBS complexes were solved at 1.27 and 1.32 Å, respectively. The inhibitor was modelled at the active site in two conformations in the subtilisin complex and in one conformation in the trypsin complex. The crystallographic data support the FTLS data that the active-site nitrile groups are mostly solvated and participate in hydrogen bonds with water molecules. The crystal structures are similar to previous structures for subtilisin- and trypsin-inhibitor complexes with low RMSDs (Figures 8 and 10).^[39–42] The similarity in the structures indicates that CBS does not perturb the active sites of these proteases, thus the vibrational reporter provides insight on the native protease active sites. The combination of IR spectroscopy utilizing vibrational reporters paired with X-ray crystallography provides a powerful approach to studying protein structure. When the vibrational reporter is attached to an inhibitor this approach becomes a general method to study enzyme active sites. However, the utility of CBSF as an effective reporter of local solvation environments is hampered by its undesired side reactions with non-active site serine residues, although the active site serine reactivity remained highest. This is the result of the electron withdrawing nature of the nitrile group which creates a more reactive species compared to protease inhibitors such as PMSF. Thus, current work is underway using azide or selenocyanate vibrational reporters which should have significantly lower reactivity with non-active site serine residues while still possessing an effective vibrational reporter of local solvation environment. This general methodology pairing IR spectroscopy and X-ray crystallography with a inhibitor modified with a vibrational reporter is also applicable for the study of other proteases including cysteine proteases such as papain.