A hydrophobic patch surrounding Trp154 in human neuroserpin controls the helix F dynamics with implications in inhibition and aggregation

Neuroserpin (NS) mediated inhibition of tissue-type plasminogen activator (tPA) is important for brain development, synapse formation and memory. Aberrations in helix F and β-sheet A movement during inhibition can directly lead to epilepsy or dementia. Conserved W154 residue in a hydrophobic patch between helix F and β-sheet A is ideally placed to control their movement during inhibition. Molecular Dynamics (MD) simulation on wild type (WT) NS and its two variants (W154A and W154P) demonstrated partial deformation in helix F and conformational differences in strands 1A and 2A only in W154P. A fluorescence and Circular Dichroism (CD) analysis with purified W154 variants revealed a significant red-shift and an increase in α-helical content in W154P as compared to W154A and WT NS. Kinetics of tPA inhibition showed a decline in association rates (ka) for W154A as compared to WT NS with indication of complex formation. Appearance of cleaved without complex formation in W154P indicates that the variant acts as substrate due to conformational misfolding around helix F. Both the variants however showed increased rate of aggregation as compared to WT NS. The hydrophobic patch identified in this study may have importance in helix F dynamics of NS.

Scientific RepoRts | 7:42987 | DOI: 10.1038/srep42987 Earlier, a Molecular Dynamics (MD) simulation study has reported the distinct movement of helix F away from β -sheet A during inhibition mechanism, and helix flexibility was found to be of importance during the final stages of NS mediated inhibitory mechanism 23 . Besides, Gettins PG 2002 24 also predicted the movement of helix F during insertion of strand 4A, followed by locking the stable acyl-protease-inhibitor complex after complete insertion. Analysis of AAT variants by Cabrita et al. 25 clearly showed melting and restructuring of helix F at both N and C-terminal end at the later stages of loop insertion during inhibition mechanism. Further, protein engineering based experiments have elucidated that certain mutations in the helix F accelerates polymerisation mechanism of AAT 26 . Intriguingly these predictions were contrasted by detection of an inactive late stage intermediate in the crystal structure of L55P from AAC. The structure revealed that the loop connecting helix F to strand 3A of β -sheet A partially fills the space between the sheet A 27 , leading to loss of its activity. In view of these facts, it is likely that NS inhibition of tPA would also require the movement of the helix F, however the residues contributing to this movement and their exact role in the mechanism that links inhibition and polymerisation still remains unclear.
A conserved W154 residue in helix F is part of a large hydrophobic patch with a network of hydrophobic interactions 28 with residues of helix F and β -sheet A (Fig. 1a). The patch contain residues Y150, I151, W154 and V155 from helix F, A135 of strand 1A, A109 and F113 from strand 2A and I181 from strand 3A of the β -sheet A. Residues of the hydrophobic patch are conserved in the orthologs of NS and the serpin family ( Supplementary Fig. S1). Moreover, a structural superposition of WT (PDB ID: 3FGQ) 29 and cleaved loop inserted NS (PDB ID: 3F02) 30 conformation indicated that the loop insertion causes more shift at the bottom of helix F, than at the top, along with the bottom of strands 1A, 2A and 3A of β -sheet A indicating a coordinated movement between the two regions (Fig. 1b). The structural superposition also showed conformational shifts in the loops connecting helix F to the strand 3A at its C-terminal end and strand 1A at the N-terminal end of the cleaved NS. Therefore, in the present study we performed site-directed mutagenesis on NS by mutating W154 to alanine (W154A) and proline (W154P) to investigate the effect of loss of interactions between helix F and β -sheet A on conformation, activity and dynamics of NS with implication on its inhibition and aggregation. The W154A and W154P variants were designed with the premise that strands 1A, 2A and 3A of β -sheet A interacting with W154 are involved in shifting of these strands may increase the polymerisation propensity if interactions with helix F are lost. On the other hand, any change that restricts the helix F movement will impede the insertion of RCL as strand 4A during inhibition. The results of the present study demonstrated that hydrophobic interactions between W154 of helix F and β -sheet A are important and needs to be thoroughly tested for helix F movement during loop insertion and in preventing aggregate formation during inhibition.

Results
Molecular Dynamics Simulation Studies. The initial coordinates were taken from PDB ID: 3FGQ 29 for MD studies and were first assigned the missing atom by using Swiss-Model 31 . The rationale for using it was primarily its high resolution (2.09 Å) as compared to PDBID: 3F5N (3.15 Å) 30 . The study was initiated by performing MD simulations for WT, W154A and W154P NS variants ( Supplementary Fig. S2). The Root Mean Square Fluctuation (RMSF) analysis showed a marked increase in fluctuation in strand 1A and decreased mobility in strand 2A in both the variants. Of note the differential fluctuation was found to be lowest in the strand 3A in both the variants as compared to the WT NS. We observed that W154P variant demonstrated greater mobility in helix F, loop and strand 1A region; however, the mobility was reduced in strand 2A than the W154A variant (Fig. 2). In contrast, for W154A, we observed significantly reduced mobility in the helix F region; however helix F at N-terminal of the W154P variant showed comparatively higher mobility. Interface between strands 3C and 4C also showed significantly higher mobility in W154P than W154A. In addition, the superposition of the simulated structures of W154P and W154A highlighted that helix F at the N-terminal end of W154P was deformed,  29 (pink) and cleaved NS PDB ID: 3F02 30 (blue). The images were generated using Chimera 47 and the distance was calculated through the program module.
but not in the case of W154A variant (Fig. 3a). The loop connecting strand 1A with helix F was also observed to be shifted, indicating the significant movement of Y150 away from the hydrophobic patch. A structural superposition of the modeled structure of W154P and cleaved NS also demonstrated the partial loosening of helix F at the C-terminal end as compared to cleaved NS structure and also moving away of the loop connecting helix F with strand 3A of the β -sheet A (Fig. 3b). Both the loop connecting helix F to strand 3A and 1A were found to be shifted in case of W154P variant (but not in W154A and WT NS) thereby, resulting in an increased conformational flexibility (Figs 2 and 3). Therefore, it is more likely that hydrophobic patch around W154 might be contributing to helix deformation and formation of an intermediate type of structure where the strands of β -sheet A were still in the native like conformation (Fig. 3b). Based on these in silico analysis, we hypothesized that the presence of this hydrophobic patch may contribute to the helix F dynamics in inhibition mechanism of NS.

Expression and purification.
To test our hypothesis based upon in silico analysis, site-directed mutagenesis was performed to obtain the W154A and W154P variants using pET28b vector containing WT NS gene encoded with N-terminal poly-His-tag and the amplified variants (Fig. 4a,b) were assessed by DNA sequencing of the complete gene ( Supplementary Fig. S3). The variants and the WT NS were transformed into E. coli BL21/DE3 cells for expression and purification. The WT NS and 154 variants were purified by two-step chromatography using Ni-NTA Sepharose column and HiLoad 16/60 Superdex 200 column. The variant migrated corresponding to its molecular weight with the same mobility as WT NS (Fig. 4c,d and e). Lanes 5 and 6 showing Ni-NTA elution at 300 mM and 600 mM imidazole were further concentrated, buffer exchanged and loaded onto a Gel-filteration (GF) column for isolating active fraction. Active NS peak eluting at 79 ml was concentrated and buffer exchanged for further analysis. Concentration was determined using molar extinction coefficient of the protein at 280 nm. Only 0.5-1.0 mg of monomer W154P and W154A protein from 4 liter of culture was obtained as compared to a yield of 8.0-9.0 mg of WT NS. These observations point towards a decrease in the expression levels of variants and/or slight conversion to oligomeric forms. Eluted fractions run on 12% SDS-PAGE were confirmed to be homogenous. Conformation of WT and mutant proteins. To assess the conformational differences between WT NS and the 154 variants, their fluorometric and CD analysis were conducted and the results are summarized in Fig. 5. Fluorescence emission spectra of W154A and W154P showed a reduction in the emission intensity as compared to WT NS (Fig. 5a). A polymeric NS showed a clear red-shift with characteristic decrease in the emission intensity 32 , matching with spectra of the chemically denatured WT NS 33 . Besides, W154P also revealed a red-shift although the emission intensities of both the variants were comparable. Further, the W154P also showed an increase in the surface hydrophobicity in the presence of 4,4′ -Bis-1-anilino naphthalene 8-sulfonate (bis-ANS) as compared to W154A and WT NS indicating that the surface environment of W154P and W154A were possibly different (Fig. 5b). Far-UV CD spectra of the proteins showed a marginal increase in α -helical content of the W154P variant as compared to WT NS and W154A variant (Fig. 5c), indicating a more structured protein, agreeing with the previous study 34 .

Inhibitory activity of WT and mutant proteins.
To obtain insight into the functional role of mutant, inhibition assay was performed to assess inhibitory activity WT NS and the W154 variants. Different concentrations of inhibitor (NS and the W154 variants) were incubated with a constant amount of tPA along with its chromogenic substrate. Progress of the reaction (tPA inhibition) was followed by hydrolysis of substrate with time. Association rate constant (k a ) was calculated using equation (2) (materials and methods) and is presented in Fig. 6. The experiments were performed under pseudo-first order rate conditions 8,35 . W154P revealed a low association rate constant (0.09 × 10 3 M −1 s −1 ) as compared to W154A (0.54 × 10 4 M −1 s −1 ) and WT NS (3.19 × 10 4 M −1 s −1 ), agreeing with the rates shown previously using single chain t-PA (sct-PA) 2,36 . The rate of inhibition of W154P variant was significantly reduced, a 6-fold decline in the k a of W154A NS variant was also observed.
To investigate the nature of stable-complex formation between tPA and NS, both were incubated for varying time and assessed on 12% SDS-PAGE (Fig. 7). We observed covalent complex formation in WT NS (Fig. 7a) and W154A (Fig. 7b) but not in case of W154P (Fig. 7c), moreover the cleaved NS was detected indicating a purely substrate behavior with no covalent complex formation (Fig. 7d,e).

Propensity of polymerisation.
Since, NS has also been shown to be less stable that can polymerise upon mild thermal stress 8,34,37 . Therefore, WT NS and 154 variants were incubated at 45 °C and monitored for the polymer formation on non-denaturating PAGE after taking the aliquots of reaction at different time intervals (Fig. 8). Although, slight conversion to large and small molecular weight polymer were observed, however the major portion of native was not able to form polymer even after 240 min of incubation (Fig. 8a). In contrast, almost all of W154A were converted to short and long chain oligomers just after 15 min of incubation (Fig. 8b). W154P showed conversion to high molecular weight aggregate within 45 min. The W154P reaction solution became turbid after 15 min of incubation at 45 °C, which might be due to the precipitation of the unfolded amorphous material due to partial unfolding of the protein, consequently resulting in unfolded aggregates with absence of oligomeric polymers (Fig. 8c). This might be attributed to the loss of interaction between helix F and strands of β -sheet A.

Discussion
The increased propensity of NS to form polymer was proposed to be due to the labile nature of helix F and its role during inhibition mechanism 25 . The dynamics of helix F after initial docking of tPA to NS through Michaelis complex and during the transition of RCL as strand 4A and the terminal kinetically trapped acyl-intermediate steps still remains poorly understood 38 . A Fluorescence Resonance Energy Transfer (FRET) based analysis had shown two distinct stages of RCL insertion in AAT, describing late stage changes at the interface of helix F and β -sheet A and predicted that energy stored in helix F contributes to the insertion process 39 . We have identified a previously unknown conserved hydrophobic patch at the interface between helix F and β -sheet A centered around W154 which can influence helix F conformational transition during inhibition mechanism. Residues of the hydrophobic patch originating from strand 1A (A135 and N137), strand 2A (A109 and F113) and strand 3A (I181) interacts with the W154 of helix F (Fig. 1a). W154 also interacts with N-terminal (Y150 and I151) and C-terminal (with V155) helix F residues, which may contribute towards helix dynamics. Moreover, the N-terminal and C-terminal of helix F connects the strand 1A and strand 3A respectively (of β -sheet A) through loops. Helix deformation can control the dynamics of the loop to influence the shift in the strands. The strand can shift only if the loops are relaxed possibly due to deformation of the helix F during inhibition mechanism. Previous work demonstrating the increase in polymerisation of AAT due to C-terminal end deformation of helix F revealed that increase in flexibility of the loop is important for inhibition 26 .
We made W154A and W154P variants by site-directed mutagenesis, purified them to homogeneity, and evaluated them for structure-function using WT NS as control. The decrease in fluorescence intensity was observed in both the variants with W154P showing a red-shift indicating partial unfolding and a more hydrophilic environment (Fig. 5a). Further, the rise in surface hydrophobicity and an increased α -helical content was observed in W154P (Fig. 5b,c) but not in W154A and WT NS. Althogether these observations indicate that overall α -helical content of the W154P was higher in contrast to the simulation studies that pointed towards a deformation of the helix F. These results support that both the variants have different conformations. The variants were also denatured in the presence of varying concentration of guanidinium chloride (GdmCl) to assess if helix F has a role in folding mechanism, monitored by intrinsic fluorescence of Trp residues. A two-step unfolding transition was seen in W154A and WT NS at 0.  the denaturation profile of W154P but not in W154A and WT NS ( Supplementary Fig. S4), resulting in a faster two-step transition. Curiously, W154 contributes in the folding mechanism of NS only when helix F is conformationally deformed (W154P). These results are in accordance with other helix F variants in serpins that have been shown to cause destabilization and that helix F is part of a folding intermediate 25 .
We have compared the MD trajectories and RMSF plots of WT NS and the W154 variants to show that there are marked changes in helix F region, strands 1A and 2A (Fig. 2). The W154A variant exhibited reduced mobility in the helix F region and strand 2A and increased mobility in strand 1A and RCL. Furthermore, a structural superposition of the modeled structures suggested that W154P helix F was deformed at N-terminal end resulting in more flexibility in the loop connecting helix F to strand 1A (Fig. 3a). Consequently fluctuations in helix F of W154P and strand 1A are more as compared to W154A. The in silico observations implied that perturbation in the hydrophobic patch surrounding W154 affects strand 1A and strand 2A of β -sheet A whereas strand 3A remains unperturbed. The results agree with a previous study that investigated the role of strand 3A by engineering a disulfide bond with helix F, and showed that the bond didn't affect inhibition mechanism but it influenced the rate of polymerisation 21 .
The analysis of the W154P variant demonstrates absence of NS-tPA complex formation and appearance of cleaved NS (Fig. 7c). Previous studies have shown that NS-tPA complex are unstable and can dissociate in minutes unlike other serpin-protease complexes that can have half life of upto weeks 30,36,41,42 . A similar type of increase in substrate pathway has also been demonstrated by binding a ligand specific for helix F 43 in PAI-1. W154A forms a complex with tPA and showed inhibition (Figs 6 and 7b). We observed that most of the hydrophobic interactions with W154 are with indole ring of the tryptophan (Fig. 1a) and hypothesized that, W154A variant will reduce the number of contacts with residues originating from strands 1A, 2A and 3A of the β -sheet A, but will maintain the integrity of helix F as the interaction with V155 and Y150 will be maintained (Figs 1a, 2 and 3a). Increased polymerisation propensity was observed for W154A variant based on the faster disappearance of band of NS and appearance of polymers as compared to very slow polymer transition observed in the WT NS (Fig. 8). W154P native band disappeared at 45 min with only large molecular weight aggregates, indicating that this variant is in a partially unfolded conformation and a W154P type variant might affect the integrity of helix F (Figs 2  and 3). Indeed, the complete deletion in helix F of PAI-1 showed complete loss in inhibitory activity and purely substrate behavior 44 .
In the present study we report that a conserved hydrophobic patch surrounding W154 may play a crucial role in the control of helix F dynamics. We hypothesize that partial deformation of helix F and relaxation in the loop connecting strand 1A will shift helix F in a synchronized manner with strands 1A and 2A. This coordinated movement is important for not allowing a gap to form between the strands of β -sheet A before insertion of the strand 4A, any variant which alters this organization will increase the polymerisation propensity. This hypothesis needs to be tested more vigorously using variants of conserved hydrophobic residues identified in this study.

Methods
Molecular Dynamics (MD) Simulations. Initial coordinates of WT NS (PDB ID: 3FGQ) 29 were obtained from Brookhaven Protein Data Bank RCSB-PDB (www.rcsb.org). The missing loop coordinates were generated using Swiss-Model web server by keeping 3FGQ as a template 31 . MD simulations were carried out on WT, W154A and W154P models of NS under similar environmental conditions using AMBER_14 running on Tesla K20c CUDA workstation. For an explicit solvent environment, both the proteins were immersed into a virtual box having distance between protein to box wall equal to 0.5 Å. Each box is then solvated with TIP3P water molecules 45 (18, 000 and 19,000 TIP3P box water molecules respectively) before subjecting the proteins to energy minimization. This is followed by addition of counter ions (17 Na + ) for charge neutralization of the minimized system, followed by re-minimization of the ionized solvated systems. Equilibration of the minimized systems was performed at a low temperature of 100 K and gradually heating up to 300 K by maintaining constant volume temperature with no pressure coupling. Long-range electrostatic interactions were evaluated by the Particle Mesh Ewald (PME) method, where Van der Waals and Coulomb interactions were truncated using a switch function within 1 nm cutoff. SHAKE algorithm was applied to constraint all the covalent bonds with H atoms, using non-bonded cutoff set to 12 Å. Thermal bath of 300 K was coupled to 1.0 Atm pressure in each of the subjected simulations and the coupling parameters for both pressure and temperature were set as 1 ps. Using Integration step of 2 fs, a final 40 ns production run simulation was carried out for all the minimized and equilibrated systems. The resulting trajectories were analyzed using AmberTools 14 to determine the dynamics of W154 dependent interactions and associated mutational consequences.
Sub cloning and site-directed mutagenesis. The plasmid pQE81L containing human NS gene was kindly provided by Prof. David A. Lomas, University of Cambridge, Cambridge, UK. It was sub cloned with restriction sites NheI and XhoI into the pET28b (Novagen) expression vector encoding N-terminal poly-His-tag and human NS protein was confirmed by MALDI-TOF analysis. The W154P mutant was constructed by  Table showing k a of WT NS, W154A and W154P, with ± standard deviation of at least three independent experiments. *The data for k a was found to be statistically significant with P < 0.001 for both W154A and W154P. One way ANOVA was used for determining statistical significance and P value was determined using t-test.
Scientific RepoRts | 7:42987 | DOI: 10.1038/srep42987 polymerase chain reaction (PCR) based site-directed mutagenesis with mutagenic primers forward: 5′ AACTA CATCAATAAGCCGGTGGAGAATAAC3′, reverse: 5′ GTTATTCTCCACCGGCTTATTGATGTAGTT3′; and mutagenic primers for W154A mutant were: forward: 5′ AACTACATCAATAAGGCGGTGGAGAATAAC3′, reverse: 5′ GTTATTCTCCACCGCCTTATTGATGTAGTT3′ using plasmid containing NS gene as a template. PCR product was transformed in E. coli DH5α (Novagen) cells following Dpn1 treatment. Both the mutations were confirmed by DNA sequencing of complete gene (Supplementary Fig. S3).  150 mM NaCl, 10 mM imidazole, 5 mM β -mercaptoethanol (β ME) and 5% glycerol] followed by washing with 30 ml of 20 mM Tris-HCl [pH 8.0] were carried out. Finally, protein was eluted using 300 mM and 600 mM imidazole in buffer A at 4 °C. Eluted fractions were run on 12% SDS-PAGE to check the presence of NS. The elution fractions containing protein were pooled, concentrated using Amicon ultracentrifugal filters (Millipore) and loaded on to HiLoad 16/60 Superdex 200 column (GE Healthcare) at flow rate of 1 ml/min for further purification. The column was pre-equilibrated with running buffer; 30 mM Tris-HCl [pH 8.0], 150 mM NaCl, and 5% glycerol before loading the protein. Monomeric purified proteins were used for carrying out the experiments.
Fluorescence measurement. Structural differences between WT NS and 154 variants were monitored by fluorescence measurement study using JASCO FP-6300 spectrofluorometer at 25 °C. The tryptophan fluorescence was detected upon excitation at 295 nm and emission spectra were recorded from 300-450 nm. The excitation and emission slit widths were set at 5 nm. All experiments were performed in 20 mM sodium phosphate buffer [pH 7.4] with 2 μ M of protein samples. Each spectrum represents an average of three individual scans and was corrected for the contribution of the blank solution.
4,4′-Bis-1-anilino naphthalene 8-sulfonate (bis-ANS) fluorescence spectra. Bis-ANS is a widely used extrinsic fluorescence probe to check the surface hydrophobicity of proteins 46 . Post mutation changes in the overall surface hydrophobicity of NS were assessed by fluorescence spectra of bis-ANS (Sigma-Aldrich, St. Louis MO, USA). For this experiment protein samples (2 μ M) were incubated with bis-ANS in a molar ratio of 1:5 at 25 °C for 1 hr in dark. Fluorescence measurements were performed using a JASCO FP-6300 spectrofluorometer. Bis-ANS emission spectra were detected using an excitation wavelength of 390 nm, and spectra were recorded from 400-650 nm 46 . Each spectrum represents the mean of three individual experiments, and were baseline corrected by subtracting with appropriate buffer and protein blanks. Circular Dichroism studies. Circular Dichroism (CD) spectra were recorded on Chirascan TM Plus CD Spectrometer from Applied Photophysics with peltier temperature control. Far-UV CD spectra were recorded at protein concentration of 0.2 mg/ml in the range of 260-200 nm wavelength at rate of 1 nm step. Far-UV CD spectrum was performed at least thrice where each reading represents the average of 10 scans taken at 25 °C. All CD spectra were baseline-corrected by substracting the buffer spectrum and smoothed. CD experiments were performed in 20 mM sodium phosphate buffer [pH 7.4].

Inhibition assay.
To determine the kinetic parameter of NS, inhibition assays were performed under pseudo-first order rate reaction, using the progressive curve method as described previously 8,35 . Briefly, inhibition assay was initiated with the addition of 100 nM tPA (> 85% sctPA, Molecular Innovations, USA) to the mixture of 1 mM chromogenic substrate (T2943 [CH 3 SO 2 -D-HHT-G-R-pNA.AcOH] from Sigma-Aldrich, St. Louis MO, USA) and varying concentration of WT NS (100 nM, 200 nM, 400 nM, 600 nM and 1000 nM) in kinetic buffer 50 mM HEPES [pH 7.4], 150 mM NaCl, and 0.1% Tween-20 at 25 °C. Progress of reaction was recorded with release of the product with absorbance at 405 nm as a function of time. Inhibition assay of W154A was performed at the same concentration as that for WT NS, but W154P was performed at a higher concentration (0.6 μ M, 1 μ M, 2 μ M and 3 μ M) as negligible inhibition was observed at lower concentration. Values of pseudo-first order rate constant (k obs ),