Steric Control of the Rate-Limiting Step of UDP-Galactopyranose Mutase

Abstract

Galactose is an abundant monosaccharide found exclusively in mammals as galactopyranose (Galp), the six-membered ring form of this sugar. In contrast, galactose appears in many pathogenic microorganisms as the five-membered ring form, galactofuranose (Galf). Galf biosynthesis begins with the conversion of UDP-Galp to UDP-Galf catalyzed by the flavoenzyme UDP-galactopyranose mutase (UGM). Because UGM is essential for the survival and proliferation of several pathogens, there is interest in understanding the catalytic mechanism to aid inhibitor development. Herein, we have used kinetic measurements and molecular dynamics simulations to explore the features of UGM that control the rate-limiting step (RLS). We show that UGM from the pathogenic fungus Aspergillus fumigatus also catalyzes the isomerization of UDP-arabinopyranose (UDP-Arap), which differs from UDP-Galp by lacking a -CH₂-OH substituent at the C5 position of the hexose ring. Unexpectedly, the RLS changed from a chemical step for the natural substrate to product release with UDP-Arap. This result implicated residues that contact the -CH₂-OH of UDP-Galp in controlling the mechanistic path. The mutation of one of these residues, Trp315, to Ala changed the RLS of the natural substrate to product release, similar to the wild-type enzyme with UDP-Arap. Molecular dynamics simulations suggest that steric complementarity in the Michaelis complex is responsible for this distinct behavior. These results provide new insight into the UGM mechanism and, more generally, how steric factors in the enzyme active site control the free energy barriers along the reaction path.

Graphical abstract

Galactofuranose (Galf) is an important structural component of the cell wall of fungi and mycobacteria and is a cell surface virulence factor in several parasitic human pathogens.^1–3 Galf biosynthesis starts with the isomerization of UDP-galactopyranose (UDP-Galp) to UDP-galactofuranose (UDP-Galf), which is catalyzed by UDP-galactopyranose mutase (UGM) (Scheme 1A).^4,5 UGM is a flavin-dependent enzyme that requires the reduced form of FAD for activity, despite the reaction being redox neutral.

Scheme 1. Reactions Catalyzed by UDP-Galactopyranose Mutase (UGM), (a) the Reaction of UGM with the Natural Substrate UDP-Galactopyranose (UDP-Galp) and (B) the Reaction of UGM with the Non-Natural Substrate UDP-Arabinopyranose (UDP-Arap)^a.

^aThe difference between the two substrates is highlighted in orange.

Extensive mechanistic and structural studies have provided a clear description of galactose isomerization catalyzed by UGM (Scheme 2).^6–10 The initial stage involves the reaction of the oxidized flavin with NAD(P)H yielding the reduced form of the enzyme, which is required for activity.¹¹ The reduced flavin, acting as a nucleophile, attacks the anomeric carbon atom of Galp, yielding a Flavin–galactose adduct and breaking the glycosidic bond to UDP. After the covalent adduct is formed, the sugar ring is opened, leading to formation of a FAD–iminium intermediate. Next, the ring closure step forms Galf. Ring closure is considered to be the rate-limiting step (RLS) of the mechanism.⁷ Lastly, the catalytic cycle is completed when the flavin–Galf adduct breaks and the glycosidic bond is reestablished, yielding the final UDP-Galf product.^6–10

Scheme 2. Detailed Mechanism for the Reaction Catalyzed by Wild-Type AfUGM with the Natural Substrate UDP-Galp^a.

^aRed color denotes the H atom initially bonded to N5_FADH and subsequently transferred as a proton to the substrate. Steps a → b and b → c involve the FAD reduction by NADPH. Once FAD is reduced, the substrate (UDP-Galp) enters the active site (c → d). A covalent adduct is formed between the substrate and the FADH cofactor (d → e), temporarily breaking the glycosidic bond, followed by an internal proton transfer between N5_FADH and O6_FADH (e → f). Step f → g involves linearization of the sugar and formation of the iminium ion species. In the next step (g → h), the sugar cyclizes into its five-membered ring form. This step is isotope sensitive and is considered to be the chemical rate-limiting step (RLS) of the mechanism.⁷ Once the furanose form of the sugar is reached, another internal proton transfer takes place (h → i), and the glycosidic bond to UDP is formed (i → j). Finally, the product of the reaction exits the active site of the enzyme.

Crystal structures have provided support for several of the steps in the mechanism. The structural basis for enzyme cofactor reduction is known from the structures of oxidized Aspergillus fumigatus UGM (AfUGM) complexed with NAD-(P)H.¹² The structure of the E–S Michaelis complex is known for several UGMs, including AfUGM,^13,14 another eukaryotic UGM from Trypanosoma cruzi (TcUGM),¹⁵ and bacterial UGMs.^16–24 Furthermore, the flavin–galactose adduct has been trapped in crystallo, providing direct structural evidence of flavin functioning as a nucleophile.²⁵ These and other structures have also revealed large conformational changes associated with reduction of the cofactor and substrate binding.²⁶

Herein, we explore the features of the active site that control the RLS of UGM. Using kinetic solvent isotope effects (KSIEs), kinetic solvent viscosity experiments (KSVEs), and rapid reaction kinetic analyses, we show that the RLS of AfUGM changes from ring contraction to product release when using the non-natural substrate UDP-arabinofuranose (UDP-Araf).

This finding suggested that residues near the C6 hydroxyl of UDP-Galp/f are important for fast product release. Previous crystal structures of AfUGM show that Trp315 is uniquely positioned to play this role. Mutation of Trp315 to Ala in the AfUGM variant W315A switched the RLS with UDP-Galf to product release, mimicking the kinetic signature of the wild-type enzyme with UDP-Araf. To investigate the process at the atomic level, we also performed classical molecular dynamics (MD) simulations to characterize the Michaelis complex and quantum-classical MD simulations to study the chemical steps of the mechanism. The computational analysis shows that furanose ring formation is the RLS with UDP-Galp, because of restricted rotation of the hydroxyl groups in the compact space imposed in part by Trp315. In contrast, with UDP-Arap, the lack of the -CH₂-OH group decreases the size of the physical constraints in the active site, lowering the activation energy for recyclation. Altogether, these results provide new insight into the UGM mechanism and show how steric factors in the enzyme active site control the free energy barriers along the reaction path.

EXPERIMENTAL DETAILS

Materials.

UDP and UDP-Galp were acquired from Sigma, and UDP-Araf was acquired from the Complex Carbohydrate Research Center (University of Georgia, Athens, GA). PfuUltra hotstart high-fidelity DNA polymerase was obtained from Agilent Technologies (Santa Clara, CA). DpnI was purchased from Fisher Scientific (Hampton, NH). Escherichia coli TOP-10 chemically competent cells were obtained from Invitrogen (Carlsbad, CA). E. coli BL21 (DE3) chemically competent cells were obtained from Promega (Madison, WI). The plasmid miniprep kit was from Qiagen (Valencia, CA). Primers were from IDT Integrated DNA Technology (Coralville, IA). All other buffers and chemicals were purchased from Fisher Scientific.

Site-Directed Mutagenesis.

Primers of 25 bp were designed to insert a point mutation to create the A. fumigatus W315A mutant (W315A). Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit following instructions provided by the manufacturer. The UGM gene cloned into pVP55A was used as the DNA template.¹³ The mutation was confirmed by DNA sequencing.

Expression and Purification.

Wild-type UGM and mutant UGM from A. fumigatus were expressed in autoinduction medium as a fusion to His₈ tags and purified following protocols previously reported.^7,11,13,15 In general, 0.6 mg of purified protein was obtained per gram of bacterial cell paste.

Kinetic Characterization.

The enzymatic activity for recombinant UGM was determined by monitoring the formation of UDP-Galp or UDP-Arap from UDP-Galf or UDP-Araf on a high-performance liquid chromatography (HPLC) instrument. The assay was performed as described previously by Oppenheimer et al.⁷ The assays were performed in 25 mM HEPES, 125 mM NaCl, and 20 mM dithionite (to reduce the flavin and make the solution temporarily anaerobic), at pH 7. The enzyme concentration was determined on the basis of the flavin bound to UGM using an extinction coefficient at 450 nm (ε₄₅₀) of 10.6 mM⁻¹ cm⁻¹. The reaction was initiated by addition of 30 nM AfUGM or 1 μM W315A. The reaction mixture for AfUGM was incubated at 37 °C for 1 min, and the mutant was incubated for 5 min. The reaction was stopped by heat denaturation and the denatured protein pelleted by centrifugation. The supernatant was injected onto a Dionex CarboPac PA100 column. UDP-Galp and UDP-Galf were eluted isocratically with 75 mM KH₂PO₄ (pH 4.5) at a rate of 0.80 mL/min and monitored at 262 nm. Under these conditions, UDP-Galp eluted at 27.4 min, UDP-Galf at 34.2 min, UDP-Arap at 30.3 min, and UDP-Araf at 41.1 min. The conversion was quantified by the percentage of the substrate and product peaks. Kinetic parameters were determined by fitting the initial velocity data to the Michaelis–Menten equation.

Kinetic Solvent Viscosity Effects.

To determine whether product release limits the rate of the reaction, viscosity effects were measured for AfUGM and W315A Reactions were performed as described above with 1000 μM UDP-Galf or UDP-Araf in the presence of 0, 5, 10, 20, and 30% glycerol. The relative viscosity (η_REL) of the reaction was calculated using a reference table for different percentages of glycerol in solution.²⁷

Kinetic Solvent Isotope Effects.

To determine the solvent kinetic isotope effect for AfUGM and W315A, initial velocities were measured at various concentrations of UDP-Galf or UDP-Araf in either H₂O or 90% D₂O at pD 7.0 as previously reported.⁷ The enzymatic activity was determined as described above.

Monitoring the Reaction of Reduced Flavin with UDP-Araf.

The assay was performed in the stopped-flow spectrophotometer under anaerobic conditions. Removal of oxygen from the stopped-flow system was achieved through addition of a solution of 100 mM glucose with 0.1 mg/mL glucose oxidase from Aspergillus niger in 0.1 M sodium acetate (pH 5.0) overnight. Sodium phosphate at 50 mM (pH 7.0) was degassed by five cycles of vacuum and argon flushing, each for 30 min. The enzyme solution was made anaerobic by degassing with six cycles of vacuum for 15 min and flushing with anaerobic argon between cycles. UDP and UDP-Araf were solubilized in degassed buffer, and concentrations were verified by a spectrophotometer at 262 nm. To reduce the degassed AfUGM, 20 mM sodium dithionite was added to the enzyme and the excess dithionite was removed using a desalting column. Reduced UGM at a final concentration of 20 μM was mixed with a final concentration of 0.25 mM UDP-Araf or 0.25 mM UDP. Spectra were collected on a logarithmic time base from 1 ms to 30 s using a photodiode array spectrophotometer.

Initial MD Settings and Equilibration Protocol.

The crystal structure of reduced AfUGM complexed with UDP-Galp (PDB entry 3UTH) was used to generate the initial coordinates of the AfUGM/UDP-Galp simulation. Galp was replaced with Arap to create the initial coordinates for the AfUGM/UDP-Arap simulation. This assumes that the introduced modification of the substrate does not significantly alter the structure of the enzyme–substrate system and that misleading conformations will be corrected during the simulation. Standard protonation states were assigned to all the residues except for His62, which was protonated according to experimental results.¹⁵ The structures were solvated in a 12 Å periodic truncated octahedral cell of TIP3P water molecules, and counterions were used for charge neutralization. The Amber99SB force field was used to compute the potential energy of the protein. The FADH cofactor and substrates UDP-Galp and UDP-Arap were parametrized as described previously.²⁸ Systems were minimized at a constant volume and then heated to 303 K during 500 ps using the weak-coupling algorithm with a τ_tp value of 3.5 ps. This was followed by a 500 ps period of equilibration under constant-temperature and -pressure conditions at 303 K and 1 bar using 3.5 ps for both τ_tp and τ_p. The electrostatic cutoff was set to 10 Å. We monitored the temperature and density of the system until they reached plateaus. The obtained structure was used as the starting point for classical and quantum-classical molecular-mechanics (QMMM) simulations.

Classical MD Simulations.

From the equilibrated structures, we performed 10 equivalent classical MD simulations for each Michaelis complex: AfUGM/UDP-Galp and AfUGM/UDP-Arap. In each case, the trajectories were started considering distinct initial atomic velocities, which were randomly chosen from a Maxwellian distribution at 303 K. The production phases lasted 100 ns, and snapshots of the trajectories were collected every 0.5 ns. Data corresponding to the distributions presented were computed considering the whole group of trajectories, while temporal evolution of the distances corresponds to individual trajectories that are representative of the behavior of the whole group of simulations.

QMMM MD Simulations.

QMMM MD simulations were performed considering the QM subsystem constituted by the flavin cofactor, the substrate (either UDP-Galp or UDP-Arap), Gly62, His63, and the side chains of Arg91, Arg182, Arg327, and Arg447. The self-consistent charge density functional tight binding (scc-DFTB) method was implemented to describe the potential energy of this subsystem. This method has proven itself to be suitable for describing the energetics of chemical and biochemical reactions.^29–31 It has also been shown to provide the best semiempirical description for six-member carbohydrate ring deformation.³² Starting from the corresponding Michaelis complex of each system (AfUGM/UDP-Galp or AfUGM/UDP-Arap), we employed the umbrella sampling technique to investigate each chemical step of the catalytic mechanism. In Figure S1, we present a detailed representation of how each reaction coordinate was defined for each step of the mechanism. For the sake of completeness, we will briefly describe the procedure here, placing a specific emphasis on the ring closure step because it is considered the rate-limiting one. The first chemical step of the catalytic mechanism involves the formation of the flavin–Arap/Galp adduct (step d → e in Scheme 2 and step 1 in Figure S1). Then, an internal proton transfer between N5_FADH and O4_FADH occurs (step e → f in Scheme 2 and step 2 in Figure S1) prior to the ring opening process (step f → g in Scheme 2 and step 3 in Figure S1). At this stage, the sugar reaches a linear form. The next step involves cyclization of the sugar into a five-member ring form. This process has been implicated in the rate determinant step for the Galp–Galf conversion by both theoretical and experimental studies^7,28,33 and is sensitive to the solvent isotopic kinetic effect. In this step of the catalytic mechanism, O4_GAL/ARA transfers a hydrogen atom to O4_FADH and forms a new bond with the anomeric sugar carbon (C1_GAL/ARA), producing closure of the ring into its five-membered form (step g → h in Scheme 2 and step 4 in Figure S1). To sample the configurations from this process, we defined the reaction coordinate (ε₄) using a linear combination of three distances related to the atoms involved in the reaction. In particular, ε₄ = d₈ – d₉ – d₁₀, where d₈, d₉, and d₁₀ represent the distances between 04_GAL/ARA and H, H and 04_FADH, and 04_GAL/ARA and C1_GAL/ARA, respectively. Reaction coordinate ε₄ was sampled from −5.00 to −0.48 Å considering 0.08 Å wide windows. Harmonic restraints of 225.0 kcal mol⁻¹ Å⁻² were applied to force the system to wander around the selected values of the reaction coordinate. Within each window, an equilibration phase of 50 ps was followed by a production phase of 0.2 ns. The actual values of the reaction coordinate were recorded every 2 fs. The last 30000 values of each window were used to compute the unbiased probability by means of the weighted histogram analysis method (WHAM).³⁴ The DHAM methodology was also implemented as a way to check the consistency of the results.³⁵ When following each reaction coordinate, the last structure of a given window was used as the starting point for the next window.

To further check the convergence of the free energy computations, we performed other tests. First, we compared the free energy profiles obtained using the first half of the data (i.e., the first 15000 values selected at each specific value of the reaction coordinate) with those obtained using the second half. The reaction coordinate was sampled both forward and backward. Finally, we repeated the reaction path three times starting from a different initial configuration. The average values for the barrier of this reaction and the standard deviation were computed considering six distinct but equivalent paths, formed by each forward and backward path from each of the three independent calculations.

The same protocol, except for the triple repetition of the independent simulations, was implemented for every step of the catalytic mechanism. The distances involved in the definition of each reaction coordinate are depicted in Figure S1. These are the same as the ones employed by us to study the Galp–Galf conversion in the catalytic mechanism of TcUGM.²⁸ When each reaction intermediate was reached, simulations of 0.5 ns without any restraint were performed to check their stability.

RESULTS

AfUGM Catalyzes the Interconversion of UDP-Arap/f.

UDP-Araf has been shown to be a substrate for bacterial UGM; however, to the best of our knowledge, it has not been evaluated with any eukaryotic UGM.³⁶ As a baseline for this analysis, the kinetic parameters of wild-type AfUGM with the natural substrate were first determined. Activity was measured by monitoring the amount of UDP-Galp formed at various concentrations of UDP-Galf The protein was reduced with excess dithionite and the product separated and analyzed using the HPLC assay described in Experimental Details. The initial rates exhibited Michaelis–Menten behavior (Figure 1A). Fitting of the data yielded a k_cat of 100 s⁻¹ and a K_M of 450 μM (Table 1).

Figure 1.

Activities of AfUGM and W315A with UDP-Galf or UDP-Araf as the substrate. (A) Wild-type AfUGM with UDP-Galf as the substrate. (B) Wild-type AfUGM with UDP-Araf as the substrate. (C) W315A single mutant with UDP-Galf as the substrate. Empty circles correspond to assays performed in H₂O and filled circles those in D₂O.

Table 1.

Kinetic Parameters for AfUGM and W315A Using UDP-Galf or UDP-Araf as the Substrate ^a

	AfUGM		W315A
	UDP-Galf	UDP-Araf	UDP-Galf	UDP-Araf
kcat (s−1)	100 ± 15	0.14 ± 0.005	0.27 ± 0.02	0.23 ± 0.05
KM (μM)	450 ± 150	94 ± 11	193 ± 40	700 ± 200
kcat/KM (M−1 s−1)	230000 ± 10000	1540 ± 150	1400 ± 200	331 ± 46
D2Okcat b	2.30 ± 0.25	1.16 ± 0.04	1.05 ± 0.14	not determined
KSVE slopec	0.25 ± 0.01	1.17 ± 0.10	0.80 ± 0.10	0.90 ± 0.06

^aAll the reactions were performed in 25 mM HEPES, 125 mM NaCl, and 20 mM dithionite (pH 7.0).

^bKinetic solvent isotope effects were calculated by dividing the value determined in H₂O by the k_cat value measured in 90% D₂O at pD 7.0.

^cKinetic solvent viscosity effect obtained with glycerol as the viscogen at 0.5, 10, 20, and 30%.

The activity of AfUGM with UDP-Araf was determined as described for UDP-Galf (Figure 1B). The k_cat value with UDP-Araf was ~700-fold lower than for UDP-Galf, while the K_M value was ~5-fold lower, which resulted in a 150-fold decrease in k_cat/K_M. Thus, AfUGM catalyzes UDP-Arap/f conversion, albeit with a catalytic efficiency much lower than that of the natural substrate.

AfUGM Forms an Iminium Intermediate with UDP-Araf.

The flavin–sugar iminium ion is a hallmark of the UGM reaction mechanism (Scheme 2, species g), so we asked whether such a species is formed with UDP-Araf. Changes in the absorbance spectrum of reduced AfUGM were monitored in the stopped-flow spectrophotometer under anaerobic conditions. The enzyme was initially mixed with buffer alone, and no absorbance changes were observed (data not shown).

Similarly, the spectra remained unchanged when UDP was added (Figure S2A). In contrast, when UDP-Araf was introduced, a decrease in absorbance at ~450 nm and an increase at ~380 nm were detected (Figure S2B). These absorbance changes are consistent with those previously reported by us and others for iminium ion formation in UDP-Galp/f isomerization catalyzed by UGMs.^7,8 These results suggest that the AfUGM-catalyzed isomerization of UDP-Arap to UDP-Araf also proceeds via an iminium ion intermediate.

The AfUGM W315A Variant Is Kinetically Compromised.

We sought a structural explanation for the low activity of AfUGM with UDP-Araf. The only difference between UDP-Arap and UDP-Galp is the presence of a -CH₂-OH group at position C5 of the latter substrate (Scheme 1). The crystal structure of AfUGM complexed with UDP-Galp shows that this group packs tightly against Trp315 (Scheme 2d, inset). This residue is highly conserved in eukaryotic UGMs. Sequence alignment shows that it is conserved in 114 of 124 sequences analyzed (Figure S3).

W315A was purified following the procedure developed for AfUGM. The protein contained tightly bound FAD and upon reduction was able to catalyze UDP-Galp/f isomerization; however, the activity was much lower than that of wild-type AfUGM. The k_cat and k_cat/K_M values ofW315Awere 160–370-fold lower than those of the wild type, while the K_M value changed by ~2-fold (Table 1 and Figure 1C). In summary, the kinetic parameters of W315A with the natural substrate are similar to those of wild-type AfUGM with UDP-Araf. The activity of W315A with UDP-Araf was also determined. Compared with those of AfUGM, the k_cat and K_M values were 1.6- and ~7-fold higher, respectively. These changes resulted in a modest ~5-fold decrease in k_cat/K_M (Table 1).

Kinetic Solvent Isotope Effects.

To investigate if a chemical step in the reaction is rate-limiting, we measured the KSIE. In KSIE experiments, H₂O is replaced by D₂O, allowing the exchange of ionizable protons. The rate measured under these conditions should decrease if proton transfer takes place in the RLS of the reaction.³⁷ The results are expressed as ${}^{{\text{D}}_2\text{O}}{k}_{cat}$, the ratio of k_cat measured in H₂O to that measured in D₂O.

The KSIE was measured for wild-type AfUGM and W315A. Using wild-type AfUGM with UDP-Galf, a ${}^{{\text{D}}_2\text{O}}{k}_{cat}$ close to 2 was obtained, consistent with proton exchange being part of the RLS. In contrast, a ${}^{{\text{D}}_2\text{O}}{k}_{cat}$ close to 1 was obtained with the wild-type enzyme and UDP-Araf, indicating that a proton transfer is not part of the RLS. Similarly, a ${}^{{\text{D}}_2\text{O}}{k}_{cat}$ of ~1 was obtained using W315A and UDP-Galf (Figure 1 and Table 1). Thus, the KSIE of W315A with the natural substrate is similar to that of wild-type AfUGM with UDP-Arf.

Kinetic Solvent Viscosity Effects.

KSVE experiments were performed to determine if product release contributes to the RLS. The ratio of k_cat in water [(k_cat)_o] to k_cat in a glycerol-containing solution [(k_cat)_η] was plotted as a function of the relative viscosity of the solution. The slope of this plot provides an indication of the degree to which diffusion out of the active site limits k_cat. Using wild-type AfUGM and UDP-Galf, a KSVE slope of 0.25 was obtained, indicating that product release does not contribute substantially to the RLS (Figure S4 and Table 1). This result is consistent with the one obtained for the same reaction but catalyzed by TcUGM.⁷ In contrast, a strong dependence of the k_cat value on solution viscosity was observed when UDP-Araf was used as the substrate with the wild-type enzyme. In this case, the slope increased to a value close to 1 (Table 1). Similarly, with W315A, the slope with either substrate was 0.8–0.9 (Figure S4 and Table 1). These results suggest that product release contributes to the RLS of both the wild-type enzyme with UDP-Araf and W315A with either substrate. Thus, similar to the KSIE measurements, the KSVE of W315A with the natural substrate is similar to that of wild-type AfUGM with UDP-Araf.

Molecular Dynamics Simulations of the Michaelis Complexes.

Altogether, the kinetic data suggested that perturbing the enzyme either by using a non-natural substrate or by mutating Trp315 to Ala changed the RLS from a chemical step to product release. Classical MD simulations were performed on the AfUGM/UDP-Galp and AfUGM/UDP-Arap Michaelis complexes to understand how perturbation of the natural substrate impacts the active site behavior. The focus of this analysis was to investigate differential aspects between the systems, which could provide insight into the observed differential rate of product release.

A first clear difference that was noticed by visually analyzing the trajectories was the distinct position of the sugar moiety in AfUGM/UDP-Arap with reference to the AfUGM/UDP-Galp Michaelis complex. In AfUGM/UDP-Galp, the O6 hydroxyl group of Galp occupies a region between the benzo group of FADH and the Trp315 wall (Figure 2A). In the AfUGM/UDP-Arap Michaelis complex, the absence of the -CH₂-OH group at C5 in Arap allows a rotation of the sugar ring, which, in turn, moves it farther from the benzo ring of the FADH cofactor, generating an empty space in the active site (Figure 2B).

Figure 2.

Binding modes of substrates in the active site of AfUGM. The structures shown are the most representative active site conformations from the MD simulations: (A) AfUGM with UDP-Galp and (B) AfUGM with UDP-Arap. The substrate, Asn207, Trp315 (or Ala315), and the isoalloxazine ring of the FADH cofactor are highlighted. Distinct characteristic distances are also indicated.

Noticing the void in the active site in the AfUGM/UDP-Arap simulation, we analyzed the diffusion of water into the active site. Thus, we counted the number of water molecules within 6 Å of the N5_FADH atom of the cofactor in all the snapshots of the MD trajectories. In Figure 3, we present the distributions of the number of water molecules in the active site for the two cases. On average, the AfUGM/UDP-Galp Michaelis complex has six water molecules around the substrate, while the AfUGM/UDP-Arap complex has 9–11 water molecules in the active site. This result is consistent with the lack of a -CH₂-OH group at C5 in Arap.

Figure 3.

Distribution of the number of water molecules within 6 Å of N5_FADH computed from the MD simulations.

Umbrella Sampling Calculations of the Chemical Rate Determinant Step.

Quantum-classical MD simulations were performed to compare the free energy profiles and the conformational changes in the ring closure step of the Galp–Galf and Arap–Araf isomerizations catalyzed by AfUGM. Experimental and theoretical studies are consistent in indicating that after substrate binding a covalent bond is formed between the sugar moiety of the substrate and the FADH cofactor (d → e in Scheme 2), followed by the sugar ring opening process (e → f → g in Scheme 2).^7,9,18,25,28 After that, the linear form of the sugar closes into its furanose form (g → h in Scheme 2). Both computational studies and kinetic analyses of the catalytic mechanism of TcUGM have identified this cyclization process as the chemical RLS of the mechanism.^7,28,33 In the current analysis, this cyclization process was also determined to be the one that presents the highest free energy barrier for the chemical steps in each case (data not shown). Therefore, we focused the analysis on this step (g → h in Scheme 2).

The free energy profiles for the ring closure step are presented in Figure 4. The calculated free energy barrier related to the Galp–Galf conversion catalyzed by AfUGM was 23 ± 1 kcal/mol. Also, a significant increase in the free energy exists at the beginning of the process (from a value of approximately −5.0 to approximately −3.0 of the reaction coordinate), which is not observed in the simulation of the Arap–Araf conversion (Figure 4). During this period of the reaction, there are concerted changes in the orientation of the sugar hydroxyl groups before any bond is broken or formed. While 04_GAL approaches C1_GAL, torsional changes around the C–C bonds caused the O5_GAL and O6_GAL hydroxyl groups to move away from C1_GAL. These conformational rearrangements appear to be restrained by the presence of Trp315. Movie S1 depicts the conformational changes mentioned above.

Figure 4.

Free energy profiles for the ring contraction step in the catalytic mechanism of AfUGM. The highest barrier (23 ± 1 kcal/mol) is observed for the wild-type form of the enzyme catalyzing Galp–Galf conversion (blue line). For the Arap–Araf isomerization catalyzed by AfUGM (pink line), the barrier is significantly smaller. Movies available as Supporting Information depict these simulations.

With regard to the cyclization step for the Arap–Araf conversion catalyzed by AfUGM, the calculated barrier was 16 ± 1 kcal/mol, which is ~30% lower than that obtained with the native substrate (Figure 4). Unlike the Galp–Galf conversion, the initial reorientation of the sugar group has almost no energy cost (Figure 4, pink line). This can be explained considering that, as the O6_GAL group is not present in arabinose, its restrained movement is avoided and its absence facilitates the conformational rearrangements of the O4_GAL and O5_GAL hydroxyl groups that are needed to form the five-membered ring. Movie S2 describes these movements in this case.

DISCUSSION

Enzymatic reaction mechanisms can be divided into three parts: Michaelis complex formation, chemical reaction, and product release. Any of these stages can contribute to the RLS of the catalytic cycle.^29,38,39 For the Galp–Galf tautomerization catalyzed by UGM, it has been shown that ring cyclization controls the overall rate of the catalytic cycle.⁷ In this work, we studied how the RLS switches from a chemical reaction to product release either by changing the substrate or by replacing an active site residue in the mutase reaction catalyzed by AfUGM. In particular, we first measured the activity of AfUGM catalyzing either the Galp–Galf or Arap–Araf conversion and that of the W315A single mutant catalyzing the Galp–Galf isomerization. Then, to better understand the difference in the activities, kinetic solvent viscosity and kinetic solvent isotopic experiments with classical and quantum-classical molecular dynamics simulations were performed.

For the Galp–Galf conversion catalyzed by AfUGM, a KSIE value of ~2 suggested that a slow proton transfer step is the main RLS, while the KSVE results indicated that product release is only partially rate limiting. We assigned the rate-limiting solvent sensitive step to the ring contraction step (g → h in Scheme 2). This is supported by the umbrella sampling free energy calculations, which showed that the estimated barrier for this sugar contraction step is 23 ± 1 kcal/mol. This barrier was the highest obtained for the cases studied and is consistent with the values found for the same step in the Galp–Galf isomerization catalyzed by TcUGM.²⁸ Analysis of the conformational changes that occur during this cyclization process indicated that, at the beginning of this reaction and prior to any bond breaking/formation event, the torsional rearrangements around the C–C bonds of the sugar that orient O4_GAL, O5_GAL and O6_GAL hydroxyl groups are restrained by the presence of the Trp315 wall. In addition, when the atomic details of the Michaelis complex were investigated by classical MD simulations, it was observed that only a small number of water molecules are present in the active site.

The Galp–Galf conversion catalyzed in the absence of the Trp wall in W315A occurred with a k_cat value that was several hundred-fold lower than for AfUGM. KSIE results for W315A indicate that there are no protons in flight in the RLS, in contrast to what was observed with AfUGM. Furthermore, the KSVE results suggest that product release is significantly more rate-limiting than the same reaction catalyzed by the wild-type form. A possible explanation is that the extra space afforded by the absence of the Trp315 wall lowers the free energy barrier of the cyclization step of the catalytic mechanism, allowing product release to become rate-limiting.

For Arap–Araf isomerization catalyzed by AfUGM, it was found that the k_cat was 0.14 ± 0.005 s⁻¹, which is significantly lower than that for the Galp–Galf conversion. However, it is similar to the k_cat value for Galp–Galf isomerization catalyzed by W315A. In this case, KSIE results also indicate that the ring contraction step is not rate-limiting and KSVE results confirm that product release is the rate determinant step. Thus, the experimental results revealed that for AfUGM, both the substitution of the substrate Galp for Arap and the replacement of Trp315 with Ala cause the same effect, a decrease in the turnover number and a switch in the rate-limiting step from ring contraction to product release. The computed free energy barrier for the sugar contraction process for Arap–Araf conversion was ~30% lower than that of the Galp–Galf conversion catalyzed by wild-type AfUGM. This lower energy value for the Arap–Araf conversion is supported by the lack of a KSIE. For this case, it could also be observed that the reorientation of hydroxyl groups of the sugar that are needed during the initial period of the cyclization process takes place with no energy cost. We attribute this to the absence of the -CH₂-OH substituent; thus, the rearrangement of the other sugar hydroxyl groups is not hindered. Classical MD simulations of this system showed that there are almost twice as many water molecules in the active site. This fact could have a detrimental effect on the rate of product release. Together, the data presented here provide further insight into the roles of the Trp wall in the reaction of UGM and of steric control in enzyme catalysis.

Supplementary Material

Movie S1.

Conformational rearrangements of the sugar moiety at the beginning of the ring closure step for the AfUGM/UDP-Galp system. In this case the rotations of the O5_GALP and O6_GALP hydroxyl groups take place restricted by the presence of the Trp315 wall.

Movie S2.

Conformational rearrangements of the sugar moiety at the beginning of the ring closure step for the AfUGM/UDP-Arap system. As this sugar lacks a carbonyl group at position 5, its smaller size allows the movement of the O5_ARAP hydroxyl group without being restrained by the presence of the Trp315 wall.

ABBREVIATIONS

Galp

galactopyranose

Galf

galactofuranose

Arap

arabinopyranose

Araf

arabinofuranose

UGM

UDP-galactopyranose mutase

RLS

rate-limiting step

AfUGM

A. fumigatus UDP-galactopyranose mutase

TcUGM

T. cruzi UDP-galactopyranose mutase

KSVE

kinetic solvent viscosity effect

KSIE

kinetic solvent isotope effect

PDB

Protein Data Bank

MD

molecular dynamics

US

umbrella sampling