Apr 5, 2017 - Internal Water Dynamics Control the Transglycosylation/Hydrolysis. Balance in the Agarase (AgaD) of Zobellia galactanivorans. Benoit Dav...
Nov 1, 2016 - P680, and a quinone acceptor, QA. The radical pair is ..... averaged, baseline-corrected, and normalized spectra were fit and plotted using IGOR (Wavemetrics, Lake Oswego, OR) software. To summarize the information above, TRFT-IR spectr
Nov 1, 2016 - algorithm in Bruker OPUS software (see figure legends). The ...... (26) Berthold, D. A.; Babcock, G. T.; Yocum, C. F. A Highly. Resolved ...
Nov 3, 2006 - Three main quality aspects for analytical laboratories are internal method validation, internal quality control (IQC), and sample result uncertainty ...
Nov 1, 2016 - This background spectrum was subtracted from the TRFT-IR data in order to provide a flat baseline (Figure S1). The difference .... (24) With our current method of baseline subtraction and subsequent curve fitting (Figure 3A), the center
Nov 1, 2016 - (21-23) Previously, our group has detected an infrared signal attributed to protonation of this internal water network during the S1-to-S2 transition. .... or H218O resuspension buffer already contained the appropriate amount of calcium
Nov 1, 2016 - (12, 13) S4 converts to S0 in the dark, and oxygen release accompanies this transition. Calcium and chloride ions are required for optimal ...
Nov 6, 2013 - (4-6, 8) While ligand migration in Mb has been extensively studied, only a few studies have examined the involvement of internal water molecules in the ...... MRD measurements on gel and solution samples, we have arrived at the followin
Nov 1, 2016 - *Address: Department of Chemistry and Biochemistry, Georgia ... an open access article published under an ACS AuthorChoice License, which ...
Nov 6, 2013 - R. Bryn Fenwick , David Oyen , H. Jane Dyson , and Peter E. Wright .... Gillian C. Lynch , John S. Perkyns , Bao Linh Nguyen , B. Montgomery ...
Nov 6, 2013 - Kuo-Jung Chang , Yun-Hsuan Kuo , and Yun-Wei Chiang ... Peter Pohl , Fabio Sterpone , David van der Spoel , Yao Xu , and Angel E Garcia.
Subscriber access provided by BALL STATE UNIV
Article
Internal water dynamics control the transglycosylation/hydrolysis balance in the agarase (AgaD) of Zobellia galactanivorans. Benoit David, Romain Irague, Diane Jouanneau, Franck Daligault, Mirjam Czjzek, Yves-Henri Sanejouand, and Charles Tellier ACS Catal., Just Accepted Manuscript • Publication Date (Web): 05 Apr 2017 Downloaded from http://pubs.acs.org on April 5, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Internal water dynamics control the transglycosylation/hydrolysis balance in the agarase (AgaD) of Zobellia galactanivorans. Benoit David,§,‡ Romain Irague,§,‡ Diane Jouanneau,♯Franck Daligault,§ Mirjam Czjzek,♯YvesHenri Sanejouand,§ Charles Tellier*,§ §
♯
UFIP, CNRS, Université de Nantes, 44322 Nantes, France
Integrative Biology of Marine Models, CNRS, UPMC Univ Paris 06, Sorbonne Université, 29680, Roscoff, France
ABSTRACT: In retaining glycoside hydrolases (GHs), transglycosylase activity is often low due to the natural hydrolytic activity that is favored in water. Improving the relative transglycosylase activity of these enzymes is of particular interest to obtain enzymes suitable for the synthesis of oligosaccharides. We explored the effect of engineering the water dynamics within the endo-βagarase AgaD on the transglycosylation/hydrolysis (T/H) balance. By mutating three aminoacids (D341, Q342 and S351), which could control water access to a putative water channel ending close to the active site, we obtained AgaD variants with an inverted T/H balance. For the best mutant, D341L/Q342H/S351F, the hydrolysis activity was reduced 50-fold compared to the wild type, while the transglycosylase activity was maintained and even slightly improved. This
variant produced large amount of oligo-agaroses by a disproportionation reaction with decaagarose as the substrate. Molecular dynamics simulations showed that these enzymatic modifications were correlated with higher water dynamics, revealed by a marked reduction in the water survival time and a decrease in the purge time of water in a channel ending close to the active site. These results suggest that modifying the water dynamics in GHs could be a rational basis for engineering of transglycosylase activity.
KEYWORDS:
glycoside
hydrolase,
transglycosylation,
water
dynamics,
agarase,
oligosaccharide synthesis
1. INTRODUCTION Although water is of major importance in the mechanism of hydrolases since it participates as a substrate in the reaction, the role of water dynamics in protein catalysis has rarely been studied,1 2,3 except for enzymes having a specific substrate access tunnel to the active site such as dehalogenases.4,5 Assuming that water exchange with the bulk water solvent is rapid within the active site,6 water is usually considered not limiting since it is also the solvent. A puzzling observation in some glycoside hydrolase families (GH1, GH16, GH33, GH13) is that enzymes with either transglycosylase or hydrolase activity are naturally found while structural and even sequence similarities are very high. For example, the sialidase of Trypanosoma rangeli and the trans-sialidase of Trypanosoma cruzi, have dramatically different hydrolytic activities while sharing 70 % sequence identity.7,8 Other homologous pairs of transglycosylase and hydrolase enzymes have also been characterized within family 16 (endoxyloglucanases and XET)9 and
family 13 (amylases and cyclodextringlucano-transferases)10. Numerous mutational analyses have been carried out in these enzymes in order to understand the origin of their catalytic differences, but no definitive conclusions were drawn from these studies.11,12,13 Understanding the mechanisms that control the balance between hydrolysis and transglycosylation is fundamental to elucidating the evolution pathway of these enzymes. It is also of great interest for biotechnological applications, since rational engineering of transglycosylases from glycoside hydrolases would provide a panel of useful enzymes for the synthesis of oligosaccharides and glycoconjugates, without the need for nucleotide activated sugars as substrates.14 One method to turn hydrolytic enzymes into glycosynthases was elegantly demonstrated by Mackenzie et al,15 but this reaction relies on synthetically modified sugar donors, which are not always straightforward to obtain. The second approach, used in this work, is based on the observation that enzymes that proceed by retention of the configuration of the anomeric carbon can naturally occur as transglycosylases. Here, the ultimate goal is to invert the equilibrium between hydrolysis and transglycosylation reaction by rational engineering of hydrolytic enzymes. The fact that the accessibility of water to the enzyme active site can be similar in transglycosylases and hydrolases led us to hypothesize that, in order to perform hydrolysis, water may have to access the active site by specific pathways such as channels or cavities.16In fact, protein structures reveal the presence of numerous internal water molecules, which are firmly packed and usually considered an integral part of the protein structure.17,18,19 However, proton exchange and magnetic relaxation experiments have shown that some of these internal water molecules can exchange with the bulk water on timescales ranging from nanoseconds to hundreds of microseconds.20,21Recently, we demonstrated by deuterium mass spectrometry exchange experiments and molecular dynamics simulations that highly conserved internal water
molecules in enzymes of GH1 family can rapidly exchange with external water defining preferential water channels within the protein.22Furthermore, some of these channels end close to the active site. Herein, our main objective was to investigate the possible relationship between water dynamics within a glycoside hydrolase and the balance between hydrolysis and transglycosylation pathways in the enzyme mechanism. The β-agarase AgaD from Zobellia galactanivorans23 was chosen since it belongs to GH family 16, in which both hydrolases (xyloglucan endo-hydrolases24, endo-1,3-β-glucanases25,26, and endo-1,3-β-galactanases27) and strict transglycosylases (such as xyloglucan endo-transglycosylases28 or chitin β-1,6 glucanosyltransferases29) are found, suggesting a common evolutionary relationship between both types of activity. For example, based on structural data, a loop deletion variant of the GH16 xyloglucanase from Nasturtium was designed to convert the xyloglucanase into xyloglucan endo-transglycosylases.9 A slight improvement in its minor transglycosylase activity was indeed obtained, but the hydrolysis reaction remained very high. Among characterized agarases, AgaD is mainly hydrolytic, producing tetrasaccharides as the dominant reaction products, and no naturally transglycosylating enzyme has been described in this GH16 subfamily so far. In addition, no attempts to produce an engineered agarosynthase or agaro-transglycosylase have been reported, despite the fact that generalizing these principles to enzymes active on marine polysaccharides would be of biotechnological interest. In this study, we have identified a putative water channel in AgaD that connects the bulk water to the catalytic acid/base in the active site. We hypothesized that the amino acids that line the water chain could control the flux of water and affect the hydrolytic activity of this hydrolase. By mutating these residues, we observed a dramatic change in the transglycosylation/hydrolysis
balance, which correlates well with a strong modification of the water flux as revealed by molecular dynamics simulation.
2. MATERIALS AND METHODS 2.1 DNA manipulations DNA-modifying enzymes were purchased from ThermoFischer Scientific and used according to the manufacturer’s instructions. DNA plasmids were isolated using the QIAprep® Spin Miniprep Kit (Qiagen). DNA was sequenced by GATC Biotech SARL (Mulhouse, France). The mutants were constructed from the pFO4-AgaDcat plasmid23 used as the starting template DNA. Point mutations were introduced by PCR using the Quickchange site-directed mutagenesis kit (Agilent Technologies) following the manufacturer’s instructions. Primer pairs employed for mutagenesis are listed in Table S1. All constructions were verified as carrying the desired mutations by DNA sequencing. 2.2 Protein expression and purification Large-scale expression of AgaD enzyme and mutants was performed using E. coli BL21 (DE3) strain (Invitrogen). Recombinant cells were grown at 20°C for 3 days in a flask containing 1 L of ZYP5052 medium 26 supplemented with 100 µg/mL ampicillin. Cells were than pelleted by centrifugation (5000 rpm, 20 min, 4°C) and resuspended using 30 mL of extraction buffer (TrisHCl 50 mM, NaCl 300 mM, MgCl2 1 mM, CaCl2 1 mM, imidazole 50 mM, pH 7.5). The suspensions were sonicated on ice using cycles of 10 s ‘on’, 10 s ‘off’ during 4 min with the power output set at 30 % of the maximal power. The sonication step was repeated one more time after a rest of 10 min on ice for each sample. The cell extracts were then centrifuged (10 000 rpm, 30 min, 4°C) and the supernatants collected. The 6xHis-tagged proteins were purified by
gravity flow using affinity chromatography on Ni-NTA superflow resin (Qiagen). After binding to the resin, the recombinant proteins were washed with 2 x 15 mL of extraction buffer followed by 2 x 15 mL of washing buffer (Tris-HCl 50 mM, NaCl 300 mM, MgCl2 1 mM, CaCl2 1 mM, imidazole 70 mM, pH 7.5). Finally the enzymes were eluted with 1 mL of elution buffer (TrisHCl 50 mM, NaCl 300 mM, MgCl2 1 mM, CaCl2 1 mM, imidazole 200 mM, pH 7.5) and stored at 4°C. The purity of eluted enzymes were verified by SDS-PAGE analysis and the concentrations were determined by capillary electrophoresis (Agilent 2100 Bioanalyser) and Bradford assay using serum bovine albumin as a standard. 2.3 Preparation of the oligo-agarose substrates Commercial agarose (25 g) was dissolved in boiling deionized water (500 mL). The mixture was cooled down to 37°C, and then 250 µL of thawed purified AgaD enzyme was added to initiate agarose hydrolysis under stirring for 345 min. The reaction was stopped by soaking the mixture into ice before centrifugation (4785 g) to eliminate the white pellets (undigested agarose). Oligo-agaroses contained into the supernatant were concentrated by rotary-evaporation (40°C) to a final volume of approximately 60 mL and then ultra-filtered by centrifugation (2000 g, 30 min) using 10 kDa Amicon cells (Millipore). The filtrate was concentrated to a final volume of approximately 11 mL by rotary-evaporation at 40°C and then filtered through a 0.45 µm syringe filter (Millipore) prior to purification. The solution of oligo-agaroses (3.5 mL) was purified by preparative size exclusion chromatography. The system was composed of three columns of Superdex 30 (26/60) in series, preceded by a Superdex 30 (26/10) guard column. Detection was carried out by a refractive index detector. Elution was performed using a 50 mM ammonium carbonate elution buffer, pH 9, at a flow rate of 1 mL/min, at 0.5 MPa. Five-mL fractions of oligo-agaroses were collected between 550 min and 915 min of elution before being
analysed by thin layer chromatography: 1 µL of samples was loaded onto silica gel plates (TLC silica gel 60 F254, Merck) and products were separated in 5/5/3 (v/v/v) butanol/ethanol/H2O eluent. After separation, the TLC plate was dried and immersed in 0.1% (w/v) orcinol, 10% (v/v) sulfuric acid in ethanol. Then products were revealed by warming the gel plate at 165°C for 0.5 to 1 min. Fractions containing pure oligo-agarose with identical degree of polymerization were pooled and freeze-dried. 2.4 Transglycosylase activity assays For the screening of the transglycosylase activity, 12 µL of purified enzymatic preparations was mixed with 3 µL of reaction buffer (MOPS 500 mM, NaCl 3 M, pH 7.5), 3 µL of 10 mg/mL agaro-decaose and 12 µL of water. Mixtures were incubated at 25 °C with stirring. One microliter of each reaction medium was analyzed after 2, 4, 6, 8 and 24 h of incubation using thin layer chromatography. The kinetics of enzymatic transglycosylation experiments were carried out using purified WT (0.21 µg) or enzyme variants, Q342H (0.44 µg), D341L/S351A (1.9 µg), D341L/S351F (4.2 µg), D341L/Q342H/S351A (2.4 µg) and D341L/Q342H/S351F (0.9 µg), mixed with 10 µL of reaction buffer (MOPS 500 mM, NaCl 3 M, pH 7.5), 10 µL of 100 mg/mL agaro-decaose and adjusted to 100 µL with water. Mixtures were incubated at 25 °C with stirring. Each hour, an aliquot (4 µL) of the reaction medium was taken and mixed with 4µL of a maltotriose solution (5 mg/mL) as an internal standard. These aliquots were submitted to derivation by adding 30 µL of 0.15 M ANTS in 15% acetic acid (v/v) and 30 µL of 1 M NaBH3CN in DMSO. The mixtures were incubated at 55°C for 7 h prior to capillary electrophoresis analysis. Separation was performed using a Beckman P/ACE System 5000 with an uncoated fused silica capillary 54 cm long with an internal capillary diameter of 120 µm. Samples were loaded into the
capillary under hydrodynamic injection mode at 40 mbar pressure for a 6 s period. The running buffer was prepared from a 50 mM phosphoric acid solution adjusted to pH 2.5 with triethylamine. Electrophoresis was performed at −25 kV at 30°C. The electropherograms were recorded at 214 nm (20 nm slit) using a UV–Vis diode array. Between each run, the capillary was rinsed using the following sequence: 3 min deionized water, 2 min NaOH 0.1 M, 2 min deionized water, 5 min running buffer. 2.5 HPAEC-PAD analysis To detect the oligosaccharides produced by the enzymatic reactions, the reaction mixtures (filtered on 0.22 µm) were analyzed with a Dionex ICS5000 system equipped with a Carbopac PA100 (4x250 mm) column associated with a PA100 guard column (4x50 mm). The system was equilibrated in 150 mM NaOH and samples (20µL) were eluted at a flow rate of 0.5 mL.min−1 with 150 mM NaOH and a linear gradient from 0 to 0.5 M sodium acetate in 30 min. Oligosaccharides were detected with an electrochemical detector (gold electrode), using the recommended PAD waveform: 0.00 s (+0.1 V), 0.20 s (+0.1 V; start data acquisition), 0.40 s (+0.1 V; stop data acquisition), 0.41 s (−2.0 V), 0.42 s (−2.0 V), 0.43 s (+0.6 V), 0.44 s (−0.1 V), 0.50 s (−0.1 V). 2.6 Computational analysis All calculations were carried out using the GROMACS software, version 4.6.5.30 The crystal structure of AgaD was retrieved from the PDB (ID: 4ASM)23. All ligands coming from the crystallization medium (PEG, imidazole, calcium ion) were removed from the initial structure. The mutations were introduced using the fixbb module of the Rosetta 3.5 software.31 In order to accommodate the mutation, side-chain repacking was systematically performed. The protonation state of all protein titratable residues was assigned using the pdb2gmx module of the
GROMACS software. In accordance with the catalytic mechanism of β-glycosidases, the carboxylate moiety of the catalytic acid/base residue (E179) was kept protonated. A 8.88 nm3 cubic simulation cell was generated around the protein and filled with the standard TIP3P water model.32 In order to satisfy the minimum image convention, the simulation cell was built such that the solute was separated from the edge of the box by a distance of 1.2 nm in all three dimensions. Sodium and chloride counter ions were introduced into the system so as to neutralize the net charge of the protein and to reach a final NaCl concentration of 0.115 M (0.150 M without taking into account the volume of the protein). The energy of the system was then minimized in three steps using the steepest descent algorithm with a tolerance threshold of 500 kJ/mol/nm. The first energy minimization of 500 steps was conducted under harmonic restraints on protein heavy atoms (1000 kJ/mol/nm2) in order to relax the geometry of water molecules and protein hydrogen atoms. Protein side chains were then minimized along with the solvent in 1000 steps while keeping the backbone restrained. All restraints were finally released in order to minimize the geometry of the whole system until convergence of the algorithm. Constant volume-temperature (NVT) and constant pressure-temperature (NPT) simulations were finally then conducted under harmonic restraints on protein heavy atoms (1000 kJ/mol/nm2) to equilibrate the solvent around the protein. Temperature and pressure were controlled using the modified Berendsen thermostat33 and the Parrinello-Rahman barostat34 with a damping coefficient of 0.1 and 1 ps, respectively. All simulations were performed with the GROMACS software using the AMBER99SB-ildn force field35 in an NPT ensemble at 300 K and 1 bar. The leapfrog integrator was used to integrate the equations of motion with a time step of 2 fs. Hydrogen bonds were constrained using the LINCS algorithm36. Periodic boundary conditions were applied. Long-range electrostatics were calculated with the particle mesh Ewald (PME)
method.37 A cutoff value of 1.2 nm was used to limit short-range Lennard-Jones interactions calculations. The simulation snapshots were saved every 100 ps. Protein internal water channels and pockets were identified using in-house algorithms (available at https://gitlab.univ-nantes.fr/Ben/DynaWatProt/) and the protocol was conducted in several steps. The first step consisted in isolating clusters of buried water molecules for each frame of the trajectory. Then they were defined as groups of water molecules whose oxygen atoms were less than 3.2 Å from one another and did not establish any hydrogen bond with the bulk solvent. Note that this cutoff value is within the optimal hydrogen bond distance range separating two water oxygens in accordance with spectroscopy and molecular dynamics simulation studies. The second step of the algorithm consisted in searching for all pairs of consecutive frames in order to link water clusters. Two buried water clusters isolated in two consecutive frames (cluster A at time t and cluster B at time t+100 ps, respectively) were considered linked if they were found within 3.2 Å from one another and if at least one water molecule identified in the cluster A was also present in the cluster B. Since a water channel is defined by a set of linked water clusters, all pairs of linked water clusters were then extracted along the whole trajectory in order to reconstruct water channels. To ensure reliable water clustering, all frames were previously aligned (on the protein backbone) to the initial one (at t=0) using the GROMACS module trjconv. Due to the inherent dynamics of water molecules, not all water channels exist as continuous water chains since a given water channel may be emptied and rebuilt several times during the simulation. In this case, a water channel can be defined as a cluster of its constitutive transient water channels formed within discrete and successive time periods over the course of the simulation. The last step of the algorithm thus consisted in aggregating all transient water channels having similar trajectories
into a single water channel cluster. Transient water channels with similar trajectories were clustered based on their distance from a given group of crystallographic water molecules from the crystal structure of AgaD (aligned to the initial frame). As an example, the water channel studied in the present work (channel 1) was isolated by clustering all transient water channel(s) with at least one water oxygen atom in the vicinity (within 1.6 Å) of one of the three water oxygen atoms 2329, 2330, and 2554 from the crystal structure of AgaD. These three crystallographic oxygen atoms were used as probes to isolate the water channel of interest since they form the core of the crystallographic water chain identified in the AgaD crystal structure.
2.7 Water dynamics analyses The channel purge time is defined as the time needed for a given cluster of water molecules present in the channel of interest at a specific time to leave the channel. By computing the distribution of channel purge times, the fluctuations of water dynamics in a given water channel can be quantified. The survival time of water molecules in a given water channel corresponds to their average residence time in this channel. Knowing the fraction of water molecules present in the channel at a given time t of the simulation, a survival probability can be determined by measuring the fraction of water molecules remaining in the channel at a later time t+τ. By calculating and averaging the survival probabilities at all times of the trajectory, a survival probability curve can be traced and used to estimate the water survival time in the channel of interest. The decay rate of the curve is given by the equation: ்
where T is the simulation length in ps, τ is the time interval between two frames in the trajectory, N(t) is the number of water molecules counted at time t, and N(t, t+τ) the number of remaining water molecules in the channel at time t+τ among those found at time t. P(τ) is the survival probability. All water dynamics analyses were conducted using in-house Python scripts. 2.8 Protein dynamics analyses The protein backbone Root Mean Square Deviation (RMSD) is an indicator of conformational changes in the protein backbone over the course of the simulation. Prior to its calculation, the backbone Cα atoms of the protein conformation taken at each time t of the simulation are superimposed on those of the starting conformation (at t=0). The RMSD was calculated using the g_rms module of GROMACS. The Root Mean Square Fluctuation (RMSF) is an indicator of the average flexibility of each protein residue over the course of the simulation. This parameter can be used to identify the most flexible regions of the protein over the whole simulation. Prior to its calculation, the backbone Cα atoms of the protein conformation taken at each time t of the simulation are superimposed on those of the starting conformation. The RMSF was calculated using the g_rmsf module of GROMACS. The conformational flexibility of catalytic residues side chain was estimated by calculating the distribution of their dihedral angles χ1 and χ3 using in an in-house Python script. Hydrogen bonds between water oxygens and protein residues atoms were identified and counted using the h_bond module of GROMACS. Using the default parameters of the program, a hydrogen bond was identified if the distance between the donor and the acceptor atoms was less than or equal to 3.5 Å and if the angle formed between the acceptor and the hydrogen atoms was less than or equal to 30°.
3. RESULTS 3.1 Buried water in AgaD There are 28 internal water molecules in the crystal structure of AgaD (PDB 4ASM). Among them, we identified two water chains that could potentially deliver water to the active site (Figure 1A). The first one, hereafter called channel 1, ends with a water molecule that is 7.3 Å from the acid base residue (E179) and is separated from the active site pocket by a gatekeeper formed by two residues, Q342 and Y181. The water chain is composed of 7 water molecules, which align along a channel about 13 Å long heading to the solvent accessible surface (Figure 1B). This potential channel is also the most pertinent one, according to CAVER Analyst.38,39 A second chain of 7 water molecules was detected at the opposite side of channel 1, relative to the axis of the active site. This channel 2 is wide open to the protein surface and presents a bottleneck in the direction of the active site, which makes it unlikely that water molecules can move freely in the direction of the catalytic residues. An interesting feature of the two buried water molecules of the channel 1 that are the closest to the acid base residue is that they exhibit a geometry (Figure S1) with respect to the acid/base that is close to the geometry observed in family 1 glycoside hydrolases (GH1 CAZY). It was shown that they belong to a water channel able to provide water molecules to the active site, right over the anomeric carbon, where the nucleophilic attack takes place during the deglycosylation step of the catalytic process.22 This observation is all the more intriguing since the protein folds are totally different, namely (α/β)8 for GH1 family and β-jelly-roll for GH16, suggesting the possibility of convergent evolution driven by the necessity of the mechanism.
3.2 Kinetic effect of mutations along the water chain The identified water channel 1 is lined with two pairs of residues, Q342/Y181 and D341/S351, which face each other forming two putative bottlenecks at each end of the water channel. We anticipated that mutations in these residues could modify the water dynamics and possibly affect the hydrolysis in AgaD, if this channel is indeed involved in supplying water molecules in the hydrolysis mechanism. The activity of AgaD variants was tested with neoagarodecaose (LA-G)5 as a substrate, which is degraded by AgaD to give the hydrolysis products, neoagarotetraose (LA-G)2 and neoagarohexaose (LA-G)3 (Scheme 1). As this enzyme uses a double displacement/retaining mechanism, AgaD also has the potential to form transglycosylation products, which were clearly identified as (LA-G)7 by TLC and capillary electrophoresis, together with the hydrolysis products at the initial stage of the reaction. The transglycosylation products are formed from neoagarodecaose, which acts as both donor and acceptor, and are subject to secondary hydrolysis during long-term incubation. Neoagarodecaose was shown to be the minimum donor substrate that allows the formation of transglycosylation products. In fact, only hydrolysis was observed with octa-agarose (Figure S2). Quantification of both hydrolysis and transglycosylation under initial rate conditions was determined by capillary electrophoresis after ANTS labelling. The initial rates of appearance of (LA-G)2, (LA-G)3, (LA-G)7 and (LA-G)8 products were determined (Figure S4), from which transglycosylation/hydrolysis (T/H) balance was calculated (Table 1). The sum of the initial rates of (LA-G)7, (LA-G)8 and (LA-G)9 release provided the transglycosylation rate of the reaction (VT), while the rate of (LA-G)2 or (LA-G)3 release came from both the hydrolysis and transglycosylation reactions. The hydrolysis rate (VH) was calculated from the following
equation: VH = [V(LA-G)2+V(LA-G)3-V(LA-G)7)-V(LA-G)8-V(LA-G)9]/2. As expected, the WT enzyme was mainly hydrolytic with a small transglycosylation activity on deca-agarose, which was not detected previously.23 The transglycosylation/hydrolysis balance was strongly in favor of the hydrolysis with a transglycosylation/hydrolysis rate ratio of about 0.24.
First, we targeted two amino acid residues at the end of the putative channel 1, close to the catalytic acid/base residue. As no efficient screening method was available for transagarase activity, only single amino-acid exchange was done by site directed mutagenesis. The conservative mutation Y181F resulted in a mutant that was almost inactive (Figure S2). This negative effect could be due to the removal of a hydrogen bond between the catalytic acid/base residue E179 and the hydroxyl group of tyrosine 181, which may stabilize the orientation of E179. Indeed, analysis of the wild-type AgaD molecular dynamics trajectory over 500 ns showed that hydrogen bonding of Y181 to E179 almost exclusively requires the insertion of one or two water molecule(s) between their respective side-chains. The residue Q342, which faces Y181 at the end of the water chain, was mutated to histidine. This mutation was suggested by the sequence alignment of 32 agarases from GH16 at least 20% identical to AgaD (Figure S3), which showed that only glutamine (91%) and histidine (9%) are present at this position. The GH16 enzymes having a histidine at the homologous position 342 are putative agarases, which have not been characterized yet. However, this alignment suggested that the mutation Q342H might be compatible with an agarase or a trans-agarase activity. Interestingly, this mutation resulted in a dramatic improvement in the transglycosylation/hydrolysis rate ratio (Table 1), with transglycosylation becoming the major reaction (90%) (Figure 2).
The second bottleneck in water channel 1 is located between amino acids D341 and S351, which have their side chains lying face to face. Sequence alignment revealed that these amino acids are not well conserved in the agarase family (Figure S3), although at least one of the positions is generally occupied by a polar residue. Since both residues are involved in hydrogen bonds with water molecules observed in channel 1 in the crystal structure (Figure S1), we decided to replace this pair of amino acids located at the entrance of the putative channel by an apolar pair in order to modify the hydrogen bonding network of the internal water molecules. Among all the possible pairs of amino-acid side chains, we selected two amino acid combinations with a strong structural difference at one position, which were structurally compatible according to the Rosetta algorithm (Table S2).31 Double mutants D341L/S351A and D341L/S351F were constructed, expressed and assayed for the activity. Both double mutants exhibited improved transglycosylase activity (Table 1), despite the fact that these mutations are located at least 13 Å from the acid/base residue E179. Moreover, when these mutations were incorporated in the context of the Q342H mutation, additive effects were observed and the resulting
triple
mutants
showed
lower
hydrolytic
activity
and
a
higher
transglycosylation/hydrolysis ratio, with T/H > 30 (Figure 2). For long-term accumulation, these triple mutants formed multiple oligosaccharide products ranging from (LA-G)2 to (LA-G)14 due to efficient disproportionation reactions, which became significant when transglycosylase activity was high together with a low hydrolytic activity (Figure 3). It is interesting to note that all these mutations had a larger effect on the hydrolysis activity than on the transglycosylation rate, since a 50-fold reduction in the hydrolysis rate was observed compared to the wild type, while the transglycosylation rate was only improved by a factor of 2.7. The mutant Q342H had
the highest absolute transglycosylation rate but, due to the rather high hydrolytic rate, this mutant formed fewer disproportionation products (Figure 3B).
3.3 Molecular dynamics of AgaD and variants Internal water dynamics Previous observations made on family GH1 enzymes suggested that water molecules could transit within water channels from the protein surface to the active site close to the acid/base residue.22 In AgaD, the bottleneck formed by both the Q342/Y181 and D341/S351 residue pairs could prevent water from reaching the acid/base E179 or at least control its access. To obtain information on the mobility of water inside putative channels, 500-ns long molecular dynamics simulations were performed on the free AgaD and mutant enzymes. As no structure was available for the glycosyl-enzyme intermediate, and because force-field parameters were not available for anhydro-galactose residues, molecular dynamics simulations were carried out on free enzymes. Starting from the AgaD enzyme (PDB 4ASM), the mutants Q342H, D341L/S351A, D341L/Q342H/S351A and D341L/Q342H/S351F were constructed and minimized before calculating molecular dynamics simulations. The RMSD profiles (Figure S5) revealed that the overall structure was stable throughout all simulations for the WT and the variants. Higher RMSD are observed for all mutants, suggesting that their dynamical behavior could prove significantly different in localized parts of the structure. The RMSF of each residue along the sequence (Figure S6) indicated also some differences between the WT and mutant enzymes, mainly on surface loops in the sequence regions 55-60 and 205-225, but in each case far from the active sites (~ 15-17 Å from the catalytic acid/base residue). The sequence region
320-345 also showed an increased flexibility in the best triple-mutant, but the most important structural change was minor since it corresponded to an interconversion between two helical conformations (Figure S7). Then, the α-helix carrying the mutated residues does not undergo any structural change, and we can conclude that the mutations did not significantly alter the secondary structure of the enzyme. However, these fluctuations indicated that the targeted mutations had a small but significant effect on the overall dynamics of the proteins. Particularly, the distribution of the internal water molecules inside the protein exhibited clear differences between WT AgaD and its mutants, noteworthy near the catalytic residues (Figure S8). Then, we had a closer look at the water dynamics inside the identified channel 1, in which the two pairs of amino acids (Q342/Y181 and D341/S351) were mutated. First, the bottleneck role of these two pairs of residues was confirmed during the long range simulation. The opening of the D341/S351 pair occurs rarely, since these residues are often linked by a hydrogen bond. The second bottleneck, formed by the residues Q342 and Y181, may control water access from the channel 1 to the catalytic site. Indeed, hydrogen bonding between Q342 and Y181 in the long range simulation is less frequent, suggesting that the corresponding bottleneck is more open. This has been confirmed by distance measurements between both side-chains, which revealed more fluctuations than the bottleneck formed by residues D341/S351 (Figure S9). In the structure of AgaD, three internal water molecules were identified within the space delimited by these two amino-acid pairs. Analysis of the water dynamics inside this particular region revealed interesting differences between the WT AgaD and the variants. First, besides the fact that the space explored by the water molecules during the simulation showed some difference (Figure 4), all the mutants with an improved transglycosylase activity exhibited a water purge time in this
channel that was significantly lower for the transglycosylase mutants than for the WT enzyme (Figure 4), showing that the internal water dynamics nearby the active site speed up in the mutants. Correspondingly, the average survival probability of water molecules in channel 1 decayed on a much shorter timescale for the mutants than for the WT enzyme (Figure 5). Taken together, these results reveal that hydration sites between the targeted mutations have a much shorter lifetime in the mutants with high transglycosylase activity, suggesting that mutations at both ends of the channel 1 lead to a leakier channel. Conformational dynamics The simulation analysis provided strong evidence for modifications of internal water dynamics in mutants with enhanced transglycosylase activity. However, such variations could also be the result of a secondary effect through alterations of the conformational dynamics of residue side chains, located close to these mutations. In particular, as the Q342 residue is within hydrogen bond distance of the acid/base catalytic residue (E179), we anticipated that mutation of this residue could strongly affect the side-chain conformation of E179. The conformational mobility of this residue was followed during the dynamic simulation by measuring the variation in the side chain dihedral angles χ1 and χ3 (Figure 6). Mutation Q342H slightly changed the conformational space explored by the acid/base residue compared with the behavior in the WT enzyme. By contrast, the double mutation D331L-S341A did not significantly change the conformational mobility of the glutamate E179, which is consistent with the long distance (13 Å) between these mutations and E179. More intriguing were the strong conformational modifications observed in the triple mutants (D341L/Q342H/S351A and D341L/Q342H/S351F), in which long-distance mutations D341L/S351A and D341L/S351F greatly affected the
conformation of the acid base residue in the context of the Q342H mutation. Similar behavior was observed for the conformational mobility of the Y181 position, which forms a hydrogen bond with E179 (Figure S10).
4. DISCUSSION Previous observations made on family GH1, GH13 and GH117 enzymes suggested that water molecules could transit through a water channel from the protein surface to the active site close to the acid/base residue.22,40,41 However, no clear relationship could be established between the water dynamics and the functional activity of these enzymes. In AgaD, which belongs to the GH16 family, we identified a water chain close to the catalytic residues and this putative water channel was used to define mutations targeting amino acids that line this water chain. Four amino-acid positions were identified Q342, Y181, D341 and S351 (Figure 1B), which have their side chain pointing toward the water cavity in such a way that they could act as gatekeepers.16 Mutation at these positions resulted in mutant agarases that had lost most of their hydrolytic activity, while keeping and enhancing their transglycosylase activity. For the best variant (D341L/Q342H/S351F), the hydrolytic activity was reduced 50-fold compared to the wild type enzyme, while the transglycosylase activity showed a 2.7-fold enhancement. As a result, this variant exhibited a strong disproportionation activity on oligo-agaroses, which compares well with the activity of the natural GH16 endotransglycosylases, XET, on oligoxyloglucan.42 This result is, to our knowledge, the first example of a rational conversion of a natural endoglycosidase into an efficient endo-transglycosylase working on natural substrates. These mutants are also of interest since they provide access to oligo-agaroses by in vitro enzymatic synthesis. This result is all the more remarkable since only one mutation is close to the active site and the
others are far from the active site: 13 Å from the catalytic residues. Furthermore, the improvement in the transglycosylase activity was observed on natural and non-activated substrates unlike for most engineered transglycosylases in which this effect is observed only with activated substrates such as PNP-sugars or sucrose.14,43 The mechanisms that control the transglycosylation/hydrolysis activity in retaining glycoside hydrolases is still not known and many hypotheses have been proposed in recent years based on structural data,7 mutational analysis44 or comparisons between natural glycosidase/ transglycosidase pairs.45 Although water is one of the major actors involved in this balance, few studies have focused on the dynamic behavior of the water inside hydrolases.1 In this work, we demonstrate a strong correlation between the increase in the transglycosylation/hydrolysis ratio of AgaD and the water dynamics in a specific channel located close to the acid/base catalytic residue. Mutations improving water leakage from this channel strongly reduce the hydrolytic activity of the mutant, whatever the position of these mutations at both ends of the channel. Interestingly, reduction in the hydrolytic activity is greater for mutated gating residues (D341 and S351) located far from the active site. By contrast, transglycosylation activity is barely affected by these mutations compared to that of WT enzyme, except for the Q342H mutant, which exhibits improved transglycosylation activity compared to the WT. The mutation Q342H is clearly important for the control of the transglycosylation/hydrolysis balance. Firstly, the histidine side chain is more rigid than that of glutamine, resulting in less favorable hydrogen bond between H342 and Y181. This hypothesis was validated by the increased conformational mobility of Y181 (Figure S9) when this mutation was present. As a result, the H342/Y181 gatekeeper pair was more widely open, which may explain the shorter lifetime of water in the channel. The small increase in the transglycosylase activity is more difficult to explain since the
AgaD structure does not contain the substrate. The most homologous agarase to AgaD is AgaB (60% similarity), which has been crystalized with an octaagarose (pdb_code 4ATF).20 In this structure, the homologous position to Q342, Q310, is in hydrogen bonding distance to the galactosyl O-4 in the subsite -1, which emphasizes the possible role of this residue in substrate binding and transition state stabilization. Mutation to histidine may provide an additional hydrogen bond with the galactosyl O-6 in the subsite +2, which might promote better binding of the acceptor and slightly favor the transglycosylase activity. The way in which water dynamics can affect the transglycosylation/hydrolysis activity is not easy to infer, since no structural information is available on the glycosyl-enzyme intermediate for the WT enzyme and variants. Moreover, reliable molecular dynamics simulations are difficult to obtain in the absence of available parameters for the glycosyl part of the complex, which contains anhydro-galactose structures. However, because several of the mutations considered in the present study are far from the active site, we hypothesized that the dynamical behavior of the free enzyme may reflect the behavior of the glycosyl-enzyme well enough, allowing us to further interpret the experimental results. As a matter of fact, our simulations on free enzymes reveal dramatic kinetic changes in the water dynamics of the mutants compared to WT enzyme. The water residence time in the water chain located close to the acid/base residue drops from ≈350 ns for AgaD to 50-70 ns in the transglycosylase variants, which is correlated to a marked decrease in the purge time of the channel in these variants. An increase in the water flux in the acid/base vicinity seems counter-intuitive to explain the reduced hydrolysis activity in mutants. However, a decrease in hydrolytic activity could also results from a mispositioning of the water molecule that has to be activated by the catalytic E179 in order to hydrolyze the glycosyl-enzyme. This latter mechanism could also be dependent on the orientation of the acid chain of the acid/base
residue (E179). However, in the variant (D341L/S351A), the conformational mobility of E179 is similar to the WT (Figure 6), while this variant exhibits a marked reduction in hydrolytic activity. This observation strongly suggests that the acid/base conformational mobility might not be the main factor governing the efficiency of the hydrolytic step. In turn, the correct positioning of the hydrolytic water molecule may require a particularly rigid environment to anchor a network of hydrogen bonds able to maintain the hydrolytic water molecule in a favorable geometry over the glycosyl linkage. Thus, during the transglycosylation reaction, the correct positioning of the sugar is obtained through interactions with the protein, and speeding up water dynamics around the active site may prove sufficient to reduce the efficiency of the hydrolytic reaction.
5. CONCLUSION By mutating amino-acid residues along a putative water channel in agarase AgaD, we dramatically modified the dynamics of water access to the active site. As a result, the AgaD variants exhibit a strong inversion in the transglycosylation/hydrolysis balance, which leads to mutants that have lost most of their hydrolytic activity, while keeping their synthetic ability. The transglycosylase activity of these mutants is remarkable since it is based on natural and nonactivated oligosaccharide donors. More generally, these results emphasize the importance of water dynamics in hydrolase catalysis and illustrate a new approach to the engineering of glycoside hydrolases by manipulating water access to the active site. Moreover, they suggest that by studying differences in water dynamics in natural pairs of glycoside hydrolases and transglycosylases might provide a way to understand how these enzyme activities have evolved from a common ancestor.
AUTHOR INFORMATION Corresponding Author *E-mail for C.T.: [email protected] Notes ‡These authors contributed equally The authors declare no competing financial interest.
ASSOCIATED CONTENT Additional experimental details and computational results including Table 1-2, figures S1-S10 (PDF)
ACKNOWLEDGMENTS This work was supported by the French National Research Agency with regard to the investment expenditure program IDEALG (http://www.idealg.ueb.eu/, grant agreement no. ANR-10-BTBR-04). RI and DJ were funded by a postdoctoral fellowship supported by the program IDEALG. The PhD fellowship of BD was funded by the University of Nantes. We thank Johann Hendrickx for his help in molecular modeling. ABBREVIATIONS
REFERENCES (1) Grossman, M.; Born, B.; Heyden, M.; Tworowski, D.; Fields, G. B.; Sagi, I.; Havenith, M. Nat. Struct. Mol. Biol. 2011, 18, 1102–1108. (2) Syrén, P.-O.; Henche, S.; Eichler, A.; Nestl, B. M.; Hauer, B. Curr. Opin. Struct. Biol. 2016, 41, 73–82. (3)
Benson, S. P.; Pleiss, J. J. Chem. Theory Comput. 2014, 10, 5206–5214.
(4) Sykora, J.; Brezovsky, J.; Koudelakova, T.; Lahoda, M.; Fortova, A.; Chernovets, T.; Chaloupkova, R.; Stepankova, V.; Prokop, Z.; Smatanova, I. K.; Hof, M.; Damborsky, J. Nat. Chem. Biol. 2014, 10, 428–430. (5) Pavlova, M.; Klvana, M.; Prokop, Z.; Chaloupkova, R.; Banas, P.; Otyepka, M.; Wade, R. C.; Tsuda, M.; Nagata, Y.; Damborsky, J. Nat. Chem. Biol. 2009, 5, 727–733. (6)
García, A. E.; Hummer, G. Proteins 2000, 38, 261–272.
(7) Buschiazzo, A.; Tavares, G. A.; Campetella, O.; Spinelli, S.; Cremona, M. L.; París, G.; Amaya, M. F.; Frasch, A. C.; Alzari, P. M. EMBO J. 2000, 19, 16–24. (8) Buschiazzo, A.; Amaya, M. F.; Cremona, M. L.; Frasch, A. C.; Alzari, P. M. Mol. Cell 2002, 10, 757–768. (9) Baumann, M. J.; Eklöf, J. M.; Michel, G.; Kallas, A. M.; Teeri, T. T.; Czjzek, M.; Brumer, H. Plant Cell 2007, 19, 1947–1963. (10)
Skov, L. K. J. Biol. Chem. 2001, 276 (27), 25273–25278.
(11) Kelly, R. M.; Leemhuis, H.; Rozeboom, H. J.; van Oosterwijk, N.; Dijkstra, B. W.; Dijkhuizen, L. Biochem. J. 2008, 413, 517–525. (12) Cambon, E.; Barbe, S.; Pizzut-Serin, S.; Remaud-Simeon, M.; André, I. Biotechnol. Bioeng. 2014, 111, 1719–1728. (13)
Bissaro, B.; Monsan, P.; Fauré, R.; O’Donohue, M. J. Biochem. J. 2015, 467, 17–35.
Carugo, O. Curr. Protein Pept. Sci. 2015, 16, 259–265.
(19)
Park, S.; Saven, J. G. Proteins 2005, 60, 450–463.
(20)
Persson, F.; Halle, B. J. Am. Chem. Soc. 2013, 135, 8735–8748.
(21)
Kaieda, S.; Halle, B. J. Phys. Chem. B 2013, 117, 14676–14687.
(22) Teze, D.; Hendrickx, J.; Dion, M.; Tellier, C.; Woods, V. L.; Tran, V.; Sanejouand, Y.-H. Biochemistry (Mosc.) 2013, 52, 5900–5910. (23) Hehemann, J.-H.; Correc, G.; Thomas, F.; Bernard, T.; Barbeyron, T.; Jam, M.; Helbert, W.; Michel, G.; Czjzek, M. J. Biol. Chem. 2012, 287, 30571–30584. (24) Martinez-Fleites, C.; Guerreiro, C. I. P. D.; Baumann, M. J.; Taylor, E. J.; Prates, J. A. M.; Ferreira, L. M. A.; Fontes, C. M. G. A.; Brumer, H.; Davies, G. J. J. Biol. Chem. 2006, 281, 24922–24933. (25) Keitel, T.; Simon, O.; Borriss, R.; Heinemann, U. Proc. Natl. Acad. Sci. U. S. A. 1993, 90, 5287–5291. (26) Ilari, A.; Fiorillo, A.; Angelaccio, S.; Florio, R.; Chiaraluce, R.; van der Oost, J.; Consalvi, V. FEBS J. 2009, 276, 1048–1058. (27) Kotake, T.; Hirata, N.; Degi, Y.; Ishiguro, M.; Kitazawa, K.; Takata, R.; Ichinose, H.; Kaneko, S.; Igarashi, K.; Samejima, M.; Tsumuraya, Y. J. Biol. Chem. 2011, 286, 27848–27854. (28) Piens, K.; Fauré, R.; Sundqvist, G.; Baumann, M. J.; Saura-Valls, M.; Teeri, T. T.; Cottaz, S.; Planas, A.; Driguez, H.; Brumer, H. J. Biol. Chem. 2008, 283, 21864–21872. (29) Blanco, N.; Sanz, A. B.; Rodríguez-Peña, J. M.; Nombela, C.; Farkaš, V.; HurtadoGuerrero, R.; Arroyo, J. FEBS J. 2015, 282, 715–731. (30) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435–447. (31) Leaver-Fay, A.; Tyka, M.; Lewis, S. M.; Lange, O. F.; Thompson, J.; Jacak, R.; Kaufman, K.; Renfrew, P. D.; Smith, C. A.; Sheffler, W.; Davis, I. W.; Cooper, S.; Treuille, A.; Mandell, D. J.; Richter, F.; Ban, Y.-E. A.; Fleishman, S. J.; Corn, J. E.; Kim, D. E.; Lyskov, S.; Berrondo, M.; Mentzer, S.; Popović, Z.; Havranek, J. J.; Karanicolas, J.; Das, R.; Meiler, J.; Kortemme, T.; Gray, J. J.; Kuhlman, B.; Baker, D.; Bradley, P. Methods Enzymol. 2011, 487,
545–574. (32) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926–935. (33) Berendsen, H. J. C.; Postma, J. P. M.; Gunsteren, W. F. van; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684–3690. (34)
Parrinello, M.; Rahman, A. J. Appl. Phys. 1981, 52, 7182–7190.
(35) Lindorff-Larsen, K.; Piana, S.; Palmo, K.; Maragakis, P.; Klepeis, J. L.; Dror, R. O.; Shaw, D. E. Proteins Struct. Funct. Bioinforma. 2010, 78, 1950–1958. (36) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J. Comput. Chem. 1997, 18, 1463–1472. (37)
Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089–10092.
Table 1. Steady state kinetics of AgaD variants at 25°C with neoagarodecaose (10 mg/ml) as the substrate. Initial velocities are expressed in µmole/min/mg.
Figure 1. A) Internal water molecules (in blue spheres and stars) within the crystal structure of AgaD (PDB 4ASM) with putative water channels 1 (red mesh) and 2 (yellow mesh) as predicted by Caver 3.0. B) Water chain corresponding to channel 1 and the amino acids lining the water channel 1 (in magenta and orange). The catalytic residues are in green and crystallographic water molecules closest to the catalytic acid base residue (E179) are in purple. The figure was prepared using the PyMOL Molecular Graphics System, v1.82 (Schrödinger)
1A 51F 51F 51A S35 /S3 /S3 / /S3 L L H H 1 1 2 2 34 34 D34 D34 L/ Q L/Q 1 1 4 4 D3 D3
Figure 2. Balance between hydrolysis and transglycosylation in steady-state conditions for WT AgaD and the variants. Percentages of hydrolysis and transglycosylation are plotted in gray bars and black bars, respectively.
Figure 3. HPAEC/PAD analysis of the products formed with WT AgaD, and mutants. A) Time course evolution of the products formed by the D241L/Q342H/S351F mutant. B) Profile of the product obtained with WT AgaD and mutants after long-term incubation (24h at 25°C) with deca-agarose (10 mg/ml) as a single substrate. Small peaks along the main (LA-G)n signals correspond to different methylation states of oligo-agaroses purified from natural agarose.
Figure 4. Distribution of the purge times of water molecules within the channel. Water purge time is related to the time needed for a cluster of water molecules present in the channel at a specific time to escape. The tip of each bar indicates the median purge time value whereas the whiskers range from the first to the third quartile of the purge time distribution. The blue bar refers to the WT AgaD whereas bars in green, red, yellow and purple correspond to the Q342H, D341L/S351A, D341L/S351A/Q342H, D341L/S351F/Q342H mutants, respectively. The structures above the bars represent all the positions explored by water within the channel during the molecular dynamics simulation.
Figure 5. Survival function for water molecules inside the channel. For each time t of the simulation, the survival probability P(t) is related to the average number of water molecules remaining in the channel at time t. X-axis intercepts indicate the average residence times of water molecules in the channel. The color code is the same as in Figure 4.
Figure 6. Scatter plot of E179 side-chain (χ1,χ3) dihedral angles over a 500-ns molecular dynamics simulation. A) WT; B) Q342H; C) D341L/S351A; D) D341L/Q342H/S351A; E) D341L/Q342H/S351F
Scheme 1. Kinetic scheme of AgaD for the minimal donor substrate neoagarodecaose (LA-G)5. LA-G corresponds to the repetitive neoagarobiose disaccharide unit (3,6-anhydro-L-galactose-α1,3-D-galactose) of agarose.