Enzymatic Production of β-Glucose 1,6-Bisphosphate Through Manipulation of Catalytic Magnesium Coordination

Production of β-Glucose 1,6-Bisphosphate Through Manipulation Manipulation of enzyme behaviour represents a sustainable technology that can be harnessed to enhance the production of valuable metabolites and chemical precursors. b-glucose 1,6-bisphosphate (bG16BP) is a native reaction intermediate in the catalytic cycle of b-phosphoglucomutase (bPGM) that has been proposed as a treatment for human congenital disorder of glycosylation involving phosphomannomutase 2 (PMM2). Studies of both bPGM and PMM2 could benefit from a green and high-yielding method for bG16BP production. Three strategies have been reported previously for the synthesis of bG16BP; however, each of these methods either delivers low yields or uses chemicals and procedures with significant environmental impacts. Herein, through combined use of NMR spectroscopy, kinetic assays and site-directed mutagenesis, we report the efficient enzymatic synthesis of anomer-specific bG16BP using a variant of bPGM. Further purification, employing a simple environmentally considerate precipitation procedure requiring only a standard biochemical toolset, results in a product with high purity and yield. Moreover, this synthesis strategy illustrates how manipulation of the catalytic magnesium coordination of an enzyme can be utilised to generate large quantities of a valuable metabolite. Abstract Manipulation of enzyme behaviour represents a sustainable technology that can be harnessed to enhance the production of valuable metabolites and chemical precursors.  -glucose 1,6-bisphosphate (  G16BP) is a native reaction intermediate in the catalytic cycle of  -phosphoglucomutase (  PGM) that has been proposed as a treatment for human congenital disorder of glycosylation involving phosphomannomutase 2 (PMM2). Studies of both  PGM and PMM2 could benefit from a green and high-yielding method for  G16BP production. Three strategies have been reported previously for the synthesis of  G16BP; however, each of these methods either delivers low yields or uses chemicals and procedures with significant environmental impacts. Herein, we report the efficient enzymatic synthesis of anomer-specific  G16BP using a variant of  PGM. Further purification, employing a simple environmentally considerate precipitation procedure requiring only a standard biochemical toolset, results in a product with high purity and yield. Moreover, this synthesis strategy illustrates how manipulation of the catalytic magnesium coordination of an enzyme can be utilised to generate large quantities of a valuable metabolite. containing different concentrations of MgCl 2 (0, 0.1, 0.3, 0.6, 1.0, 1.5, 2.5, 5, 10, 20, 50 and 100 mM), 1 mM NAD + , 5 U/mL G6PDH, 1 mM  G1P and either 1 nM  PGM WT with 100  M  G16BP, or 10  M  PGM D170N with 1250  M  G16BP. Initial rates of the reactions were obtained from the linear steady-state region of the kinetic profiles and were fit to Equation 1 using an in-house non-linear least-squares fitting program, which uses bootstrap error estimation.

Studies of both bPGM and PMM2 could benefit from a green and high-yielding method for bG16BP production. Three strategies have been reported previously for the synthesis of bG16BP; however, each of these methods either delivers low yields or uses chemicals and procedures with significant environmental impacts. Herein, through combined use of NMR spectroscopy, kinetic assays and site-directed mutagenesis, we report the efficient enzymatic synthesis of anomer-specific bG16BP using a variant of bPGM. Further purification, employing a simple environmentally considerate precipitation procedure requiring only a standard biochemical toolset, results in a product with high purity and yield. Moreover, this synthesis strategy illustrates how manipulation of the catalytic magnesium coordination of an enzyme can be utilised to generate large quantities of a valuable metabolite.

Main
Enzyme engineering represents an emerging technology with the potential to deliver solutions to many sustainable development problems [1,2]. Biofuel production, plastic degradation and the clean generation of industrial reagents and precursors are three examples of areas where enzymes already make a significant contribution [36]. With a comprehensive understanding of enzyme mechanism still lacking, research into the fundamentals of enzyme catalysis is of great interest. Phosphoryl transfer enzymes are at the forefront of research models for investigating the origins of enzyme catalysis because they exhibit some of the largest enzymatic rate enhancements known [78]. In addition, phosphate esters are often covalently incorporated into pharmaceutical products to improve bioavailability [9.
-phosphoglucomutase (PGM; EC 5.4.2.6) has emerged as an archetypal enzyme in the study of phosphoryl transfer, and substantial progress has been made in understanding its mechanism of catalysis [11. This magnesium-dependent enzyme from Lactococcus lactis (subspecies lactis IL1403) catalyses the isomerisation between -glucose 1-phosphate (G1P) and glucose 6-phosphate (G6P) via a -glucose 1,6-phosphate (G16BP) intermediate, which is released to solution before rebinding in the alternate orientation ( Figure 1) [11,17]. The G1P substrate of PGM is commercially unavailable, but appropriate quantities for research have been produced enzymatically from maltose using a simple method involving maltose phosphorylase [18]. To initiate the catalytic cycle, PGM requires priming with a phosphorylating agent to generate the active phospho-enzyme (PGM P , phosphorylated on residue D8) and G16BP performs this role primarily in vivo. Since G16BP is also commercially unavailable, alternative phosphorylating agents such as acetyl phosphate (AcP), fructose 1,6bisphosphate (F16BP) and -glucose 1,6-bisphosphate (G16BP) have been used to generate PGM P in vitro, but these compounds are less effective and produce complicated kinetic behaviour [12,19].
G16BP has also been identified as a potential pharmacological chaperone for the management of a human congenital disorder of glycosylation involving phosphomannomutase 2 [20]. Acting as a weakly binding competitive inhibitor, G16BP is able to rescue the compromised activity of pathological variants of phosphomannomutase 2 by stabilising the protein fold. Therefore, further investigations of phosphomannomutase 2 and of PGM are reliant on the availability of substantial quantities of G16BP. Three strategies have been reported previously for the synthesis of G16BP; however, each of these methods either delivers low yields or uses chemicals and procedures with significant environmental impacts. Firstly, the chemical synthesis of G16BP from -glucose involves an eight step protocol [12], requiring considerable time and technical expertise, together with the use of harmful and environmentally hazardous reagents. Low yields are obtained, since the -anomer must be selected carefully on the basis of solubility from a racemic mixture of glucosaccharide products. Secondly, an enzymatic production method utilises a non-native reaction of phosphofructokinase to generate G16BP from G1P using adenosine triphosphate as the phosphoryl donor [17,20]. Purification of the product, though, cannot be achieved simply using precipitation procedures, since contaminating adenosine diphosphate co-precipitates with G16BP [21] and therefore ion-exchange HPLC purification is required. The use of HPLC columns is inherently damaging to the environment owing to the use of triethylammonium bicarbonate as a volatile buffer mobile phase, which during its production results in enormous quantities of carbon dioxide being released into the atmosphere as a greenhouse gas [22]. Thirdly, an extraction method involves the removal of G16BP from a variant of PGM that copurifies with a stoichiometric quantity of the molecule [18]. This method suffers from low yields, since it relies on very high recombinant PGM production levels, and requires a week-long protein growth and purification procedure for each new batch of G16BP. The limited availability of G16BP therefore represents a significant barrier to the structural, kinetic and therapeutic investigations of phosphomutase enzymes. Herein, we describe a room-temperature, enzymatic method using the D170N variant of PGM (PGM D170N ) for the production of 100% anomer-specific G16BP, which requires only micromolar quantities of enzyme and a simple environmentally considerate purification procedure that can be performed easily by a non-chemist over the course of two days. More generally, this enzymatic synthesis strategy illustrates how manipulation of catalytic magnesium coordination can be utilised to generate large quantities of a valuable metabolite.
PGM has two phosphoryl transfer steps in its catalytic cycle: Step 1 comprises phosphoryl transfer from PGM P to the G1P substrate forming the G16BP reaction intermediate, whereas Step 2 involves phosphoryl transfer from G16BP to PGM forming the G6P product and regeneration of PGM P ( Figure 1). When wild-type PGM (PGM WT ) is incubated in the presence of Mg 2+ ions, with 20 mM AcP as the phosphorylating agent and 10 mM G1P as a substrate, G16BP generated in the catalytic cycle does not accumulate to detectable levels when monitored using 31 P NMR experiments [18].
Instead, G16BP rebinds the enzyme with micromolar affinity in the alternate orientation, for the Step 2 reaction. Thus, the tight binding and high reactivity of G16BP maintains a low steady state concentration, which precludes the harvesting of this species in useful quantities. The crystal structures of substrate-free PGM WT (PDB: 6YDL; [23]) and of the PGM WT P analogue complex (PGM WT :BeF 3 complex, PDB: 2WFA; [15]) indicate that the catalytic magnesium ion (Mg cat ) is coordinated through three enzyme atoms in the former and four phospho-enzyme atoms in the latter ( Figure 2). Therefore, the differential coordination and affinity of Mg cat provides an appropriate target with which to manipulate PGM to shift the balance in the rates of Step 1 and Step 2 so that G16BP will accumulate to a greater extent. Two potential strategies emerged where the rate of Step 2 could be retarded with respect to the rate of Step 1, which involved either performing the reactions of the catalytic cycle under Mg 2+ -free conditions or perturbing Mg cat coordination through point mutation to alter its binding properties. In either scenario, it was hypothesised that PGM with a compromised Mg cat site could be phosphorylated efficiently by reactive phosphorylating agents such as AcP, thereby generating PGM P and subsequent reaction with G1P to produce G16BP in Step 1 ( Figure 1). In contrast, phosphorylation of PGM by G16BP in Step 2 is less likely under these circumstances, which would lead to an accumulation of the reaction intermediate that could be harvested.
To explore whether AcP is able to phosphorylate Mg cat -free PGM WT , 31 P NMR experiments were acquired to measure the change in AcP concentration over time in the presence and absence of 300 M PGM WT . The addition of PGM WT resulted in a 25% increase in the rate of AcP hydrolysis ( Figure   3A), implying thatPGM WT P is generated in the absence of Mg cat . Consequently, the Step 1 reaction between Mg cat -free PGM WT (300 M) and 10 mM G1P in the presence of 20 mM AcP together with the Step 2 production of G6P was monitored using 31 P NMR time-course experiments. However, there was no detectable accumulation of G16BP ( Figure 4AC) and the appearance of G6P product proceeded with a rate constant of 6.7 × 10 -3 s -1 , which is 4 orders of magnitude slower than the rate constant observed in the presence of 5 mM MgCl 2 [18]. Hence, the observed enzymatic activity appears to arise simply due to the presence of very low levels of residual Mg 2+ ions associated with the reagents.
Taken together, these results indicate that both Mg cat -bound PGM WT P and Mg cat -free PGM WT P can be generated by AcP, but both the Step 1 and Step 2 phosphoryl transfer reactions are seriously impaired by the absence of Mg cat .
Given the low rate of G16BP production in the absence of Mg 2+ ions in the reaction buffer, a more subtle modification of the enzyme Mg cat site was engineered. In PGM WT , Mg cat is coordinated octahedrally by a carboxylate oxygen atom of residue D8, a carboxylate oxygen atom of residue D170 and the carbonyl oxygen atom of residue D10, together with three water molecules. In PGM WT P , one of the water molecules (water 3) is displaced by a phosphate oxygen atom of the D8 aspartyl phosphate moiety, creating bidentate coordination of Mg cat in a six-membered ring of atoms ( Figure 2). Point mutations involving residue D8 have been reported to result in the complete loss of measurable catalytic activity [19]. Therefore, perturbation of Mg cat was achieved through the generation of the D170N variant (PGM D170N ), where the carboxamide group of residue N170 retains an oxygen atom with which to coordinate Mg cat , but the sidechain is not charged. Accordingly, the reaction of PGM D170N with 10 mM G1P and 20 mM AcP in the presence of 100 mM MgCl 2 was monitored using 31 P NMR time-course experiments and in contrast to PGM WT , the G16BP intermediate was observed to accumulate to a level comparable with the initial G1P concentration ( Figure 4DE). The G6P product was only generated to a measurable extent once the AcP concentration had reduced significantly ( Figure 4F).
Hence, perturbation of Mg cat in PGM D170N (in the presence of excess AcP) results in a retardation in the rate of phosphorylation of PGM D170N by G16BP (Step 2) with respect to the rate of phosphorylation of G1P by PGM D170N P (Step 1), therefore allowing G16BP to accumulate.
Further NMR time-course experiments were conducted to assess whether the G16BP accumulation could be enhanced through changes in the concentration of Mg 2+ ions. At 5 mM MgCl 2 the accumulated G16BP was converted to G6P more rapidly than at 100 mM MgCl 2 ( Figure 4GH). The combined G6P peak intensities (G6P and G6P peaks) at maximum G16BP accumulation, although a relatively crude metric, corresponded to 174% of the G16BP peak intensity at 5 mM MgCl 2 and only 15% at 100 mM MgCl 2 . The observation that PGM D170N is active at similar MgCl 2 concentrations to PGM WT is supported by analysis of the initial rates of reaction for the conversion of G1P to G6P by PGM D170N at increasing concentrations of MgCl 2 ( Figure 3B). This experiment resulted in an apparent K m for Mg 2+ of 690 ± 110 M, which is only 4 times higher than that determined for PGM WT (apparent K m = 180 ± 40 M). Hence, for the enzyme form involved in the rate-limiting step, the Mg cat binding affinity has not been disrupted substantially by the removal of the negative charge from the coordinating sidechain of residue D170, although the observed catalytic rate of PGM D170N (k obs = 0.0086 s -1 ) is reduced 30,000fold with respect to PGM WT (k obs = 285 s -1 ) under the same conditions. The accumulation of G16BP at higher concentrations of MgCl 2 therefore suggests that the affinity of PGM D170N for Mg cat is not the predominant factor influencing this observation. To ascertain whether the effect is instead caused by Cl  ions, experiments were conducted in which the reaction of PGM D170N with G1P and AcP was monitored in the presence of 200 mM sodium chloride and 5 mM magnesium acetate. The conversion of G16BP to G6P progressed at a similar rate to the experiment performed in the absence of NaCl ( Figure 4I), indicating that the accumulation of G16BP is a MgCl 2 -dependent phenomenon.
One plausible explanation for these observations is that in substrate-free PGM D170N , the loss of the negative charge from the sidechain of residue D170 could be mitigated by the binding of a Cl  ion in the active site of the enzyme, with the displacement of a water molecule (water 3 in Figure 2B) to confer charge balance in the presence of Mg cat . In this scenario, the Cl  ion in turn can bind a separate Mg 2+ ion (with attendant counter ions) occupying the position adjacent to the nucleophilic oxygen of residue D8 and analogous to the position of the phosphorus atom of PGM P (Figure 2A). In this arrangement a six membered ring is formed consisting of both carboxylate oxygen atoms of residue D8, Mg cat , a Cl  ion and a Mg 2+ ion (a Mg cat ClMg moiety), mimicking the to the BeF 3  and AlF 4  moieties, which are representative of the ground state and transition state of PGM P hydrolysis, respectively [13,15].
This hypothesis is supported by the observation of an analogous enzyme-bound fluoride species in 19 F NMR experiments conducted with PGM WT (Figure 4JK), in which the chemical shift difference between free F  and MgF + is similar to that between MgF + and the putative enzyme-bound Mg cat FMg moiety, the chemical shift of which is comparable to that of the bridging F  atom in the PGM WT :MgF 3 :G6P complex [24]. A weakly bound Mg cat ClMg moiety would impede Recombinant PGM D170N is overexpressed in high yields from Escherichia coli BL21(DE3) cells (>100 mg/L) using routine culture techniques and is readily purified using a two-step protocol involving ionexchange chromatography followed by a size-exclusion chromatography step (see materials and methods). PGM D170N can be stored at 20 °C for long periods and responds well to multiple freezethaw cycles, meaning that once purified, a batch of enzyme can be used for numerous G16BP preparations. 31 P NMR time-course experiments were used to monitor the reaction between PGM D170N and G1P to determine the optimal point at which to harvest G16BP. The resulting enzyme-free solution contained G16BP alongside contaminants that included significant amounts of G1P, G6P and inorganic phosphate (P i ), in a ratio of 1 : 0.07 : 0.2 : 3.9, respectively. As substrates of PGM, these phosphorylated impurities are undesirable, therefore the solution was subjected to a barium salt precipitation protocol to obtain G16BP with high purity. Barium salts of phosphate species are relatively insoluble [25], and the difference in relative solubility of the G16BP barium salt compared with those of G1P and G6P was exploited to enable further purification [2628].
The solution was passed through a 20 x 10 mm column packed with IR120 (H + ) ion-exchange resin, which had been washed with 15 mL of milliQ water. This step acidified the solution, which was then neutralised using 0.2 M barium hydroxide solution, resulting in significant precipitation. The solution was kept on ice during neutralisation to increase the solubility of the mono-phosphorylated glucosaccharide barium salts [29]. Fractions obtained along the course of the barium salt formation were analysed using 31 P NMR experiments, which indicated that the G16BP barium salt was contained mainly in the precipitate, and that the G1P and G6P barium salts remained in solution. The precipitate was pelleted using centrifugation at 4 ºC (4,500 rpm, Thermo Scientific Heraeus Labofuge 400 R) and the supernatant was discarded. Conversion of the G16BP barium salt to the more soluble sodium salt involved resolubilising the pellet in a large volume (~1 L) of cold milliQ water and passing the resulting solution through a 20 x 10 mm column packed with IR120 (Na + ) ion-exchange resin. The flow-through was then frozen at 80 ºC and lyophilised to leave a fine powder as the final G16BP product.
To confirm the identity and assess the purity of the final G16BP product, a sample of the fine powder was resolubilised in 100% 2 H 2 O containing 1 mM sodium trimethylsilyl propionate (TSP) as a reference, and analysed using 1 H, 13 C and 31 P NMR experiments ( Figure 5). The identity of the resulting compound was established to be G16BP by comparison of 1 H and 13 C chemical shifts with previously reported values [12]. Glucose and maltose contaminants were identified in the sample using 1 H chemical shifts and scalar coupling constants (BMRB: bmse000015, BMRB: bmse000017). Based on integral values of the anomeric proton signals and of the phosphorus signals in quantitative 1 H and 31 P NMR spectra, the G16BP concentration was determined to be 67 mM, which represented 98% of the total phosphorylated glucosaccharide components and 72% of the total glucosaccharide components present in the final sample. G1P, G6P and glucose comprised <1%, 1% and 3%, respectively, of the total glucosaccharide content. Maltose was present at a greater concentration in the sample (24%), but as a bystander in the reactions of PGM, and not known to bind to phosphomannomutase 2, this contamination is unlikely to be problematic for users. P i was also present at a concentration 2.9 times higher than that of G16BP. The glucose, maltose and P i components, which were carried through into the final G16BP product are contaminants derived from the enzymatic synthesis of G1P and would otherwise not be present if a purer source of G1P were used. Residual HEPES buffer and acetate were also present as minor contaminants. The final yield for the PGM D170N -catalysed conversion of G1P to G16BP was 33.6% and the yield for the overall conversion of maltose to G16BP was 7.7%. Since the equilibrium for the enzymatic conversion of maltose to G1P lies in favour of maltose, conducting the reactions for the maltose phosphorylase synthesis of G1P and the PGM D170N synthesis of G16BP in a one-pot system is likely to lead to higher G16BP yields. The removal of G1P by PGM D170N would drive the maltose reaction to produce more G1P, which in turn would result in a greater overall yield of G16BP. This approach has been demonstrated previously for the protocol involving maltose phosphorylase and phosphofructokinase [17]. The removal of G1P by PGM D170N drives the maltose reaction to produce more G1P, which in turn results in a greater overall yield of G16BP. To demonstrate the biochemical effectiveness of the final G16BP product at activating PGM WT , a kinetic experiment was conducted using a glucose 6-phosphate dehydrogenase coupled assay (see Materials and Methods). PGM WT was mixed with the G1P substrate and activated using either 1 M G16BP or 8 mM AcP as the phosphorylating agent. The kinetic profile obtained was linear for the G16BPcontaining reaction, but exhibited a lag phase when AcP was used ( Figure 3C). As G16BP is the only phosphorylating agent known to induce linear initial kinetics in PGM [23], this experiment provided a clear demonstration of the activity of the final G16BP product.
The successful manipulation of PGM behaviour to facilitate G16BP production is a demonstration of how detailed structural and mechanistic knowledge of an enzyme can lead to novel engineering strategies. Specifically, the modification of the metal binding site of the enzyme dramatically increases the steady state concentration of its reactive metabolite. This highlights the transformative potential that enzymes have within chemical industries and vindicates the intensive study of these useful biomolecules.

Conflicts of interest
There are no conflicts to declare.

Reagents
Unless stated otherwise, reagents were purchased from Sigma-Aldrich, Fischer Scientific, Alfa Aesar and VWR. Isotopically enriched 15 NH 4 Cl was purchased from CortecNet.
Gene sequence for PGM D170N .

NMR Spectroscopy
All NMR spectra were acquired at 298 K, unless otherwise stated. resonating at 0.0 ppm, sealed inside a glass capillary and inserted into the sample NMR tube or were referenced indirectly to TSP using the gyromagnetic ratios of the 1 H and 31 P nuclei. 19      2) with 5 mM magnesium acetate and 200 mM NaCl, 36 min after the addition of AcP. The peak at 1.92.0 ppm in panels AI corresponds to inorganic phosphate (P i ), which is present in the stocks of both G1P and AcP. 31 P chemical shifts were referenced to external 1 M HPO 3 = 0.0 ppm, which was sealed inside a glass capillary and inserted into the sample NMR tubes. The samples in GI were recorded using experiments with proton-phosphorus decoupling in order to simplify the identification of the relevant species. (JK) 19 F NMR experiments acquired at 5 ºC involving PGM WT . (J) Control experiment with 5 mM MgCl 2 and 10 mM NH 4 F in 50 mM K + HEPES buffer (pH 7.2). (K) 1 mM PGM WT together with 5 mM MgCl 2 and 10 mM NH 4 F in 50 mM K + HEPES buffer (pH 7.2). The peak at 119 ppm corresponds to free fluoride (F  ) in solution, the peak at 156 ppm corresponds to MgF + and the peak at 173 ppm in (K) corresponds to an enzyme-bound Mg cat FMg moiety that has been assigned on the basis of chemical shift differences between the fluoride-containing species in solution. .76 ppm (triplet) and 1-phosphate, 2.55 ppm (doublet)) and the signal corresponding to inorganic phosphate (2.70 ppm (singlet), truncated for clarity). (D) Natural abundance 1 H 13 C-HSQC spectra comparing the final G16BP product (orange) with chemically synthesised G16BP (blue, [12]). Peaks are labelled with carbon ring atom assignments.