Mechanical cues often influence the factors affecting the transition states of catalytic reactions and alter the activation pathway. However, tracking the real-time dynamics of such activation pathways is limited. Using single-molecule trapping of reaction intermediates, we developed a method that enabled us to perform one reaction at one site and simultaneously study the real-time dynamics of the catalytic pathway. Using this, we showed single-molecule calligraphy at nanometer resolution and deciphered the mechanism of the sortase A enzymatic reaction that, counter-intuitively, accelerates bacterial adhesion under shear tension. Our method captured a force-induced dissociation of the enzyme–substrate bond that accelerates the forward reaction 100×, proposing a new mechano-activated catalytic pathway. In corroboration, our molecular dynamics simulations in the presence of force identified a force-induced conformational switch in the enzyme that accelerates proton transfer between CYS184 (acceptor) and HIS120 (donor) catalytic dyads by reducing the inter-residue distances. Overall, the present study opens up the possibility of studying the influence of factors affecting transition states in real time and paves the way for the rational design of enzymes with enhanced efficiency.
Force as a determinant of specific enzyme conformations leading to specific biochemical reactions is well established, from both theoretical perspectives and single-molecule force measurements [1–5]. However, exploring the role of force in driving specific enzymatic pathways has been challenging [6,7]. Using a combination of single-molecule force-ramp measurements with atomic force microscopy (AFM), and single-molecule fluorescence measurements, we developed a method of trapping the enzymatic reaction intermediate (TERI) to track the force-activated reaction pathways in real time. We implemented this method on the sortase A enzymatic reaction that not only mediates the contact adhesion of Gram-positive bacteria to hosts against shear flow [8–10] but also, counter-intuitively, accelerates the rate of adhesion with increasing shear tension .
The sortase family of enzymes catalyze bacterial cell adhesion via a multistep process [10,12,13] in all Gram-positive bacteria (including Listeria monocytogenes, Streptococcus pyogenes, Streptococcus pneumoniae) . It first recognizes the membrane-bound cell-adhesion molecules (e.g. staphylococcus protein A, fibronectin binding proteins FnBPA and FnBPB, clumping factors ClfA and ClfB, etc.) from their conserved amino acid sequence (LPXTG) and forms a thioacyl-linked protein–sortase complex. This complex is formed by the nucleophilic attack of thiolate (S−) of cysteine184 (CYS184) of sortase A on the peptide-carbonyl linking threonine (T) and glycine (G) in the LPXTG polypeptide.
Subsequently, the nucleophilic amine from polyglycine that is attached to cell wall precursor molecules, lipid II, reacts with the thioester–acyl complex and forms a tetrahedral intermediate (intermediate-II, Supplementary Figure S1). Cleavage of the enzyme–substrate bond of intermediate-II finally anchors the cell adhesion molecules to the bacterial cell wall and establishes the foundation for host–pathogen interactions. This overall reaction is commonly known as sortagging, which is a continuous reaction process where the product regains the substrate binding and cleaving motif, i.e. LPXTG, and thus serves as a substrate for sortase (Supplementary Figure S2). This reversible nature of sortagging sets the dynamics between the anchorage and obliteration of cell adhesion molecules to/from peptidoglycans forming the cell wall and subsequently, modulates bacterial adhesion to hosts. While the sortagging mechanism is well established, the mechanism by which the sortagging is catalyzed by shear flow, contributing towards faster cell adhesion is poorly understood. Since the shear flow on the bacteria generates a tensile force on the complexes linking bacteria to host surfaces , we studied the reaction kinetics of sortagging under tensile force at single-molecule resolution using TERI. We used sortase A from Staphylococcus aureus as a representative of the sortase enzyme family of proteins where three conserved amino acid residues, HIS120, CYS184, and ARG197 (numbering as per sortase A), serve as triads in the catalytic activity of the entire protein family .
Cloning, expression, and purification of wild-type and LPETG-tagged sortase A
Sortase A (Δ59) in the pET28a plasmid (Addgene plasmid # 51138) with an N-terminal 6x-His tag was used as the wild-type (WT). For LPETG-tagged sortase A, the peptide was genetically incorporated at the C-terminus of the WT construct using PCR and cloned into NdeI and BamHI sites of the pET28a vector. Both proteins were expressed in Escherichia coli BL21 (DE3) (Stratagene) cultured in LB (Luria-Bertani) media and induced at the OD600 of 0.6 with 500 µM IPTG at 16°C for 16 h. These pellets were freeze-thawed three times and suspended in resuspension buffer (50 mM HEPES buffer with 100 mM NaCl, 50 mM KCl, 2 mM CaCl2). The cells were lysed by sonication followed by centrifugation to remove the cell debris. Both proteins were obtained in the soluble fractions that were immediately loaded on to a Ni-NTA (Qiagen) column and eluted in the same buffer containing 200 mM imidazole (Hi-media). Finally, the proteins were purified using size-exclusion chromatography with a Superdex 75 16/30 column (Wipro-GE Healthcare) in 25 mM HEPES, 25 mM KCl, 100 mM NaCl, and 2 mM CaCl2 (pH 7.5) (Supplementary Figure S14).
The glass coverslips and Si3N4 cantilevers (Olympus, OMCL-TR400PSA-1) were cleaned with plasma for 1 min and piranha (H2SO4 : H2O2 in a ratio of 3 : 1) (Merck) for 3 h prior to any modifications. Subsequently, the surfaces were silanized using 2% APTES (Sigma-Aldrich) in acetone and cured at 110°C for 1 h. The amine-exposed surfaces were reacted with a mixture of 10% NHS-PEG-maleimide in NHS-PEGm (LaysanBio) in a basic buffer (100 mM NaHCO3, 600 mM K2SO4, pH 8.0) for 4 h. The PEG-modified surfaces were then incubated in 100 µM polyglycine (GGGC) for 7 h at 25°C to anchor the peptide to surfaces via the cysteine-maleimide reaction. This polyglycine serves as a nucleophile for sortagging chemistry. To covalently attach the LPETG-tagged sortase A enzyme onto the surfaces, the surfaces were incubated in a mixture of sortase A (WT) and LPETG-tagged sortase A in a 4 : 5 volume ratio at 200 nM concentration of each.
Single-molecule fluorescence using total internal reflection fluorescence microscopy (TIRFM)
LPETG-tagged sortase A was conjugated with the Cy7 NHS ester non-specifically to locate the reaction sites on the surface. To track the thioacyl complex formation, the LPETGSS peptide; leucine (L) labeled with Cy5-NHS, was added to the surface at a concentration of 10 nM and the extent of the reaction was monitored by exciting the molecules with a 640 nm laser at 200 ms exposure. Next, we washed off the unreacted LPETGSS peptide from the solution and added the GGGC peptide, labeled with Cy3 maleimide at cysteine (C), onto the surface. The formation of intermediate-II was imaged by exciting the surface-attached dyes simultaneously using 532 nm and 640 nm lasers at 10% intensity and in total internal reflection mode. All the image processing, image subtraction, centroid fitting, and overlapping of images from different channels were performed using MATLAB programs written in-house.
Objective-based total internal reflection fluorescence microscopy (TIRFM) (Olympus) equipped with 3 lasers (488 nm, 532 nm, and 640 nm diode laser system, 100 mW power), excitations and an EMCCD camera (Q-Imaging Roller Thunder) with QUADVIEW options were used for fluorescence experiments. The filters used are for a Quad-band LF405/488/532/635-A-000 Bright Line Full Multi-Band Laser Filter set.
Single-molecule force-ramp spectroscopy using AFM
Force-ramp measurements were performed at 5625 spots with a 400 nm gap between two consecutive points, thus covering an area of 30 µm × 30 µm. We performed such measurements at five different pulling velocities (100, 1000, 2000, 5000, and 10 000 nm s−1) and a constant contact time of 0.5 s. The contact time is the time the cantilever is in contact with the surface or ligands. To standardize the contact time, we first performed experiments at contact times varying from 0.05 s to 1 s and observed maximum events at 0.5 s.
The spring constant of the chemically modified cantilever was measured using the thermal fluctuation method before each experiment. Each approach and retract distance was set to 200 nm with data acquisition at 6 kHz.
Molecular dynamics (MD) simulation methods
The co-ordinates for sortase A were obtained from Protein Data Bank entry 1T2P. The hydrogen atoms were added using the AutoPSF plugin from VMD . The protein was placed in a TIP3P  water box using the VMD plugin SOLVATE. This was followed by the addition of ions using the plugin AUTOIONIZE to obtain a total concentration of 250 mM after neutralization of the system. All the molecular dynamics simulations were performed using the NAMD program version 2.12 [19,20] and CHARMM36 force field . The system was first minimized for 5000 steps using a conjugate gradient algorithm with the backbone atoms of the proteins constrained, followed by 5000 steps of minimization with all the atoms free to move. The system was then slowly heated up in steps of 20 K every 2 picoseconds (ps) with the backbone atoms constrained using harmonic restraints. The system was then equilibrated in an NPT ensemble for 5 ns. The effect of force on the conformation of sortase was studied using steered molecular dynamics (SMD) simulations. The C-terminus was restrained with a spring to a point moving with a constant velocity of 2.5 Å/ns away from the catalytic residue, i.e. CYS184. The value of the spring constant was chosen to be 97.3 pN/Å. All the simulations were performed using periodic boundary conditions, with the temperature kept constant at 300 K using Langevin temperature control  and pressure was maintained at 1 atm with a Nose-Hoover piston . The long-range electrostatic interactions were calculated using the Particle Mesh Ewald (PME) method. A time step of 1 femtosecond (fs) was used for annealing, and it was increased to 2 fs with the bonds to the hydrogen atoms constrained used the SHAKE algorithm  for the equilibration and SMD phase. The simulation trajectories were visualized using VMD and the distances were calculated using the Carma program .
Results and discussion
TERI at the single-molecule level
We trapped the sortase A–substrate intermediate-II with an AFM cantilever and measured the real-time dissociation of the enzyme–substrate bond from the deflection of the cantilever (Movie S1). Each dissociation leads to the covalent attachment of ligands to the glass coverslip substrate. We employed this feature and performed ‘single-molecule calligraphy’ (Figure 1A) using AFM as the proof-of-principle experiment of intermediate trapping and subsequent dissociation. We covalently attached the C- terminal of sortase A to the AFM cantilever prior to the experiment , and used that as a pen for the calligraphy. The sortase recognition peptide (sortag), LPETGSS, labeled with Cy5-NHS at the leucine (L) site (Cy5-LPETGSS), was used as red fluorescent ink. The glass coverslip modified with polyglycine (GGGC) was used as paper (Figure 1A). The amine (–NH2) group of polyglycine was set away from the surface, available to execute the nucleophilic attack necessary for intermediate-II formation (Supplementary Figure S1). We used total internal reflection fluorescence microscopy (TIRFM) to monitor the extent of reaction from the fluorescence signal.
Single-molecule calligraphy using TERI.
For the experiment, we added sortag (Cy5-LPETGSS) in the reaction buffer of pH 7.5 which readily forms the thioester–acyl complex with the sortase A attached to the cantilever (Figure 1A (1–2)). This complex is stable for hours in the buffer (Supplementary Figures S3 and S4). Next, we brought the cantilever with the thioester complex down into contact with the polyglycine-coated surface and waited for 0.5 s (contact time) for the nucleophilic attack from the amine (–NH2) group to occur (Figure 1A (3)). This contact facilitates the formation of intermediate-II. The contact time was optimized to 0.5 s from the measurement of the frequency of the trapping reaction intermediate in single-molecule force spectroscopy (SMFS) for different contact times, and subsequently from the fluorescence intensity of the surface (Supplementary Figure S5). Upon pulling the cantilever away in the next step, wherever the intermediate-II is formed, we observed the typical stretching of polyethylene glycol (PEG) molecules in force spectroscopy (Figure 1A (4), Supplementary Figure S6). PEG here is used as a spacer to differentiate the non-specific surface interactions [27,28]. PEG stretching manifested the trapping of reaction intermediate-II, and the detachment of the cantilever from the surface (peak-maxima of PEG-stretch) due to the dissociation of the enzyme–substrate bond, leading to the attachment of dye-labeled peptides onto the surface. Further, the stretching of PEG was confirmed from the Worm-like Chain (WLC) model fitting (Supplementary Figure S6) [27,29].
We programmed the tethering of four molecules in 4 reaction spots, 50 nm away from each other and repeated the pattern at five sites after unidirectionally translocating 4 µm for each step (scheme in Figure 1B). This experimental scheme should result in five bright fluorescent spots beyond the diffraction limit where each spot should contain four fluorophores within diffraction. We monitored the extent of TERI from the single-molecule fluorescence measurement of the grafted dyes using TIRFM, as circled in Figure 1C (center panel). Direct correlations between the successful trapping of intermediates to the grafting of molecules are shown in Figure 1C (side panels). The number of PEG-stretching events (inset of Figure 1C side panels) at each spot corresponded to the successful trapping of intermediate-II leading to tethering, and the number of steps in the intensity-time photobleaching profile of fluorophores corresponded to the number of tethered molecules (Figure 1C side panels). In spot one, all four attempts were successful as reflected by the four PEG-stretching events and four-step photobleaching profile. However, in consecutive spots only two, one, three, and three molecules were attached, respectively. The position of each tethered molecule on spot one was determined beyond the diffraction limit, using two-dimensional Gaussian fitting of the intensity image (Supplementary Figure S7) [30,31]. This experiment overall validates the concept and the successful execution of TERI which could now be used as a potential tool to measure the kinetics of sortase A enzymatic reactions and the influence of tensile force on its reaction pathway.
TERI to decipher the counter-intuitive adhesion of S. aureus against tensile force
To decipher the counter-intuitive adhesion mechanism of S. aureus against a tensile force we performed TERI as before, except the density of molecules on both surfaces was reduced to 10%, so that separation between each molecule was 600 ± 10 nm, ensuring capture of proper single-molecule events (Materials and Methods) . We observed 7.6% of PEG-stretching events on average (Supplementary Figure S8), and in accordance with Poisson predictions, 96.8% of the PEG-stretching featured single unbinding events fitting to the WLC model (Supplementary Figures S6 and S9). While repeating the experiments in the absence of any LPETGSS peptide in the buffer, or without polyglycine on the surface, we obtained 0.4% PEG-stretching events (Supplementary Figure S8). This accounts for the non-specificity of the surfaces. In addition to non-specific PEG-stretching, we also anticipated PEG-stretching events for unsuccessful reactions due to the incomplete interaction between intermediate I and the nucleophile on the surface. To avoid such false attempts of the reactions, we repeated the measurements five times consecutively on each spot and analyzed those spots, which featured PEG-stretching only once at either the first or second attempt out of five (Supplementary Figure S10).
Figure 2(A–E) shows the force distributions of the dissociation of the thioester complex of intermediate-II, at five different pulling rates (nm s−1) of the cantilever. At low pulling rates 100 nm s−1 and 1000 nm s−1, we observed predominantly unimodal force distributions. At higher loading rates, a second distribution appeared at the higher force regime (Figure 2C–E). The dissociation force (pN) of a single bond, the S–C+ bond of intermediate-II in this case, and the survival of the bond (lifetime, s) are related to the rate at which the bond is loaded to the pulling force, defined as loading rate (pN s−1) [32,33]. We estimated loading rate from the most probable unbinding forces, obtained from the peak maxima of each distribution . In the case of the higher loading rate we have two peak maxima and thus two loading rates at each velocity (Figure 2A–E). Using the analytical expression relating lifetime, force, and loading rate, as given by Dudko-Hummer-Szabo (DHS) , we converted the mean force of each bin in the histogram in Figure 2A–E to survival time of the bond at that force and depict in Figure 2F. The limitation of the DHS model is that it is only applicable to unimodal force distribution. We segregated the bimodal distribution to individual contributions using Gaussian fit as shown in Figure 2C–E and implemented DHS. Force-lifetime data obtained from different loading rates showed excellent overlap amongst each other. This collapse of the data from widely distributed loading rates is one of the major criteria to check the validity of the model. We observed non-linear behavior in the force-lifetime plot, as shown in Figure 2F. The non-linearity in the model is expected with increasing loading rate and accounted for the spatial shift in the peak of the potential barrier towards the bound state. In other words, the model considers the decrease in the distance between the bound state to the transition state , thus stiffening the loaded bond with increasing loading rate, along with the tilting of the barrier towards the load. However, the DHS model doesn't predict the appearance of a new force distribution at the higher loading rate. The new force distribution can be justified if the contribution is from a new conformation of the intermediate-II complex, which is a force-induced transient conformation.
Role of mechanical tension in the dissociation of tetrahedral intermediate-II in sortagging.
A similar observation was previously reported in the case of the von Willebrand factor (VWF) binding to platelets during hemostasis under the influence of hydrodynamic drag [35,36]. A switch from one bound state to another bound state of the VWF–platelet complex was observed with loading rates in SMFS. At a lower loading rate, only one bound state existed, and dissociation from that bound state was featured in a unimodal force distribution. Above a critical loading rate, the second pathway of bond dissociation appeared and became increasingly dominating with loading rates. The differences in the potential barrier crossing of the bound state were reflected as bimodal distributions in the unbinding force histogram at higher loading rates. We also observed similar bimodal force distributions appearing at higher loading rates, confirming the force-induced conformational switch, and the same has been reflected as two slopes in the force-lifetime plot. The plot can be divided into two Bell's regimes and can be fitted to the Bell equation (dissociation rate, ). At a lower force regime (<300 pN), the dissociation occurred from state I with an off-rate and of 1.78 s−1 and 0.6 Å, respectively. The dissociation from the switched conformation is represented by the slope in the high force regime, with off-rate and of 100 s−1 and 0.2 Å, respectively.
The faster off-rate and smaller of the intermediate-II obtained at the higher force regime indicate that the force-induced conformational switch (state-II) accelerates the dissociation kinetics. At lower loading rates, where intermediate-II dissociates at forces below Fcrit, it does not experience the force-induced conformational switch and thus, follows slower dissociation kinetics from state I. However, with increasing loading rates, the bonds are loaded to a force above the Fcrit, faster than the survival time of intermediate-II at state I and thus, the switch occurs. We estimated the Fcrit as 370 pN from the intersection of the decays (Figure 2F). This switch to state II possibly contributes to the flow-induced faster adhesion.
SMD simulations depict the conformational switch by mechanical tension
For the structural elucidation of the phenomenon observed in the AFM experiments, we performed multiple constant velocity SMD simulations. To maintain the pulling geometry similar to AFM, we kept the S-group of CYS184 anchored, and pulled from the C-terminal of sortase A at a velocity of 2.5 Å/ns (and 7.5 Å/ns) and calculated distances between the residues contributing to catalysis, i.e. CYS184, HIS120, and ARG197 (Materials and Methods). We noticed a flip of the HIS120 residue towards CYS184 above a critical force that gradually decreased the distance between the S− of CYS184 and the ND1 atom of HIS120 to 5.0 Å from 8.2 Å (from crystal structure) (Figure 3A,B, left of the vertical blue line, Movie S2). The same was reflected from the bimodal distributions of HIS120-CYS184 distances obtained from SMD (Figure 3C). The peak at 8.0 Å corresponded to the unaffected crystal structure dominated at the low force regime, whereas the peak at 4.9 Å corresponded to the switched conformation observed at the high force regime. No change was measured in the distance between CYS184 and the third member in the catalytic triad, ARG197 (Supplementary Figure S11). This conformation switch favors the proton transfer from the ND1 atom of HIS120 to the thiolate of CYS184, which eventually makes the substrate-enzyme dissociation faster. This flip is highly resilient, and upon switching off the force, HIS120 reverted to the original position as shown in the crystal structure (Figure 3B, right of the vertical blue line). Reassuringly, thermal fluctuations alone could not induce the flip as observed from multiple MD simulations of 10 ns each (Figure 3B and Supplementary Figure S12). Based on the experimental fit and SMD, we proposed a dissociation-pathway model via an intermediate as a function of mechanical tension (Figure 3D) [37,38]. At a force regime lower than Fcrit, the rate-limiting step is the escape from the thermodynamically stable conformation as in the crystal structure (primary bound state) to the dissociated state (black dotted line in Figure 3D). In the presence of force, the barrier heights decrease and at a force higher than Fcrit, the state corresponding to the conformational switch becomes the deepest bound state (red line in Figure 3D). The escape from this state to the dissociation state is now rate determining (red dots in Figure 3D), which is featured as the second peak in the force distribution.
Effect of force on the conformation of sortase A.
Our results show how sortase A switches its conformation to adapt to a tensile force and accelerate the sortagging of cell adhesion molecules to the cell wall. Acceleration in the kinetics overall helps the limited number of enzymes at host–pathogen contact to anchor a number of cell adhesion molecules to the cell wall within the given contact time and thus, strengthen adhesion (Supplementary Figure S13). The existing theory of force-induced strong adhesion is catch bonding between cell adhesion molecules as observed for multicellular organisms  and Gram-negative bacteria . Catch bonds extend bond lifetime with increasing force . We cannot rule out the possibility of an additional catch-bonding mechanism between cell adhesion molecules from the bacterial cell wall to the host to explain the counter-intuitive cell adhesion against force . However, single-molecule force spectroscopy with FnBPs that dimerize for intercellular adhesion in S. aureus has shown slip bonds , which reduce the bond lifetime with increasing force .
The molecular details presented here have raised an important question, i.e. if the reaction rate is higher with HIS120 and CYS184 closer to each other then why did the enzyme not evolve with the force-induced twisted conformation in the natural state. One possible explanation comes from the NMR structure of sortase A bound to the LPAT* substrate , and the MD simulations performed earlier. It appears that the force-induced conformation observed in our SMD simulations would promote the ‘THR-out’ conformation of the peptide, i.e. THR pointing away from the active site by forming H-bonds with HIS120 . This conformation is catalytically non-productive as it hinders the formation of the thioester from the cleavage of the THR–GLY peptide bond and consecutively, the nucleophilic attack from a polyglycine amine. Therefore, in our force-induced conformation, if it existed intrinsically, the covalent attachment of the peptide to CYS184 would have slowed down the reaction drastically.
Due to ‘conformational switching’, sortase A is a viable target for allosteric drugs . The discovery of molecules that can mimic the effect of force and make sortase A adopt the conformation seen on applying force would push the peptide into the ‘THR-out’ non-productive conformation, effectively inhibiting the sortase-catalyzed reaction. Also, if the orthosteric inhibitors are biased towards one of the available conformers of sortase, either with HIS120 and CYS184 close together or far apart, then allosterically acting compounds that enrich the needed conformation can be used along with the orthosteric inhibitors to enhance their efficacy.
Two movies have been uploaded. Movie 1 describes the procedure followed for single-molecule intermediate trapping, and Movie 2 highlights the change in the distance between CYS184 and HIS120 during the course of SMD simulations.
S.R. has supervised the project. J.P.H. and N.A. performed the expression and purification. J.P.H., N.A., S.S. and S.R. recorded and analyzed AFM. J.P.H. and S.R. recorded and analyzed TIRF. A.S. and S.R. performed and analyzed MD and SMD. J.P.H., A.C., and S.R. performed the data fitting and proposed model. J.P.H., N.A., and A.S. made the figures. S.R., A.S., and A.C. wrote the manuscript. J.P.H., A.S., N.A., A.C., and S.R. edited the manuscript. N.A. and A.S. are co-second authors.
S.R. acknowledges the financial support by the Indian Institute of Science Education and Research Mohali (IISERM) and the Centre of Excellence (COE) in Frontier Areas of Science and Technology (FAST) program of the Ministry of Human Resource Development, Government of India. J.P.H sincerely thanks IISERM for financial support. N.A. is grateful to the Council for Scientific and Industrial Research, India (CSIR), for funding. A.S. is thankful to the Centre of Excellence (COE) in Frontier Areas of Science and Technology (FAST) program of the Ministry of Human Resource Development, Government of India for financial support. A.C. acknowledges the financial support by the Indian Institute of Science Education and Research Mohali (IISERM).
We thank Professor N. Sathyamurthy, Indian Institute of Science Education and Research Mohali, India for editing this manuscript, Professor Sri Rama Koti Ainavarapu, Tata Institute of Fundamental Research, Mumbai and Professor S. Ramakrishnan, Indian Institute of Science, Bangalore, for providing critical comments that improved the scientific content of the manuscript. Authors also thank Satavisa Jana (MS project student, IISERM) for helping N.A. with the cloning.
The Authors declare that there are no competing interests associated with the manuscript.
These authors contributed equally to this work.