Role of connecting loop I in catalysis and allosteric regulation of human glucokinase

Abstract

Glucokinase (GCK, hexokinase IV) is a monomeric enzyme with a single glucose binding site that displays steady-state kinetic cooperativity, a functional characteristic that affords allosteric regulation of GCK activity. Structural evidence suggests that connecting loop I, comprised of residues 47–71, facilitates cooperativity by dictating the rate and scope of motions between the large and small domains of GCK. Here we investigate the impact of varying the length and amino acid sequence of connecting loop I upon GCK cooperativity. We find that sequential, single amino acid deletions from the C-terminus of connecting loop I cause systematic decreases in cooperativity. Deleting up to two loop residues leaves the k_cat value unchanged; however, removing three or more residues reduces k_cat by 1000-fold. In contrast, the glucose K_0.5 and K_D values are unaffected by shortening the connecting loop by up to six residues. Substituting alanine or glycine for proline-66, which adopts a cis conformation in some GCK crystal structures, does not alter cooperativity, indicating that cis/trans isomerization of this loop residue does not govern slow conformational reorganizations linked to hysteresis. Replacing connecting loop I with the corresponding loop sequence from the catalytic domain of the noncooperative isozyme human hexokinase I (HK-I) eliminates cooperativity without impacting the k_cat and glucose K_0.5 values. Our results indicate that catalytic turnover requires a minimal length of connecting loop I, whereas the loop has little impact upon the binding affinity of GCK for glucose. We propose a model in which the primary structure of connecting loop I affects cooperativity by influencing conformational dynamics, without altering the equilibrium distribution of GCK conformations.

Introduction

Mammalian tissues contain four enzyme isoforms that catalyze the ATP-dependent phosphorylation of glucose.¹ Hexokinases I-III are 100 kDa polypeptides that display ∼70% sequence identity with one another.² The N- and C-terminal domains of hexokinases I-III are structurally superimposable, and likely resulted from a gene duplication and fusion event.^3–6 Hexokinase IV, also known as glucokinase (GCK), is a monomeric, 52 kDa enzyme that displays ∼50% sequence identity with the N- and C-terminal domains of hexokinases I-III.^2,7 GCK is mainly expressed in the liver and pancreatic β-cells, where it performs the rate-limiting step of glycolysis.⁸ Pancreatic GCK activity must be tightly regulated in vivo, as it controls insulin secretion into the bloodstream in response to a glucose challenge. GCK is unique among the four hexokinase isozymes by its sigmoidal steady-state response to increasing glucose concentrations, which is characterized by a Hill coefficient of 1.7 and a relatively high glucose midpoint value (K_0.5 = 7 mM).⁹ These kinetic features facilitate glucose-mediated allosteric regulation of GCK and allow the enzyme to fulfill its physiological role as the body's glucose sensor.¹⁰

Genetic lesions in the gck gene lead to several disease states associated with impaired glucose homeostasis, including maturity onset diabetes of the young type II (MODY-II), permanent neonatal diabetes mellitus (PNDM), and persistent hyperinsulinemia of infancy (PHHI).^11–13 The identification of these disease states has led to increased interest in the mechanism of GCK cooperativity, and has prompted searches for therapeutic agents that target this important metabolic enzyme. Such efforts have resulted in the identification of several new classes of pharmaceutical agents that activate GCK, one of which has progressed through phase II clinical trials for the treatment of diabetes.^14,15 Activators of GCK stimulate enzyme activity by reducing the enzyme's cooperativity and by increasing the enzyme's responsiveness to glucose. Similar traits are observed in activated, PHHI-associated GCK variants. Understanding the structural feature(s) responsible for the enzyme's cooperativity and modest glucose affinity is important for understanding disease states such as PHHI, and for elucidating the mechanism of action of synthetic allosteric activators.

Current models suggest that GCK cooperativity results from slow, glucose-mediated conformational reorganizations in the enzyme's partially disordered small domain.^16,17 To produce steady-state cooperativity these structural changes must occur on a time scale slower than k_cat.^18–20 In support of this model, NMR and X-ray crystallographic investigations have revealed dynamic conformational heterogeneity in unliganded GCK.^17,21,22 Crystal structures demonstrate that GCK undergoes a more dramatic induced-fit hinge motion than has been observed in other hexokinase isozymes. In the absence of glucose, the enzyme adopts a “super-open” conformation in which the large and small domains are separated by an angle of 99° [Fig. 1(A)].²¹ The closed, glucose-bound crystal structure of GCK superimposes upon the glucose-bound structures of other hexokinase isozymes. This observation has led to the speculation that GCK's ability to adopt a much wider opening angle than other hexokinases is responsible for slow conformational reorganizations upon glucose binding.

Figure 1.

Connecting loop I links the large (light blue) and small (gray) domains of GCK. (A) Crystal structure of unliganded GCK (PDB 1V4T²¹) depicting the position of connecting loop I (red), and the 99° opening angle between the large and small domains. (B) Crystal structure of GCK (PDB 1V4S²¹) in the presence of glucose (green) depicting the position of connecting loop I (red), which forms part of the allosteric activator binding site (yellow).

Connecting loop I, which is comprised of 25 residues spanning Leu47 to Val71, connects the large and small domains of GCK. On the basis of the topology of the GCK scaffold, connecting loop I appears to play an important role in dictating the range of motions sampled during the catalytic cycle (Fig. 1). A variety of activating variants associated with PHHI have been identified in connecting loop I, including S64P, S64F, T65I, and S69P.^12,23,24 Several inactivating, MODY-II associated variants are also localized here.^11,25 Residues within connecting loop I interact with the α5- and α13-helices, two secondary structural elements whose sequence has been shown to dictate cooperativity.^26,27 Connecting loop I also forms part of the binding site for small-molecule allosteric activators of GCK. Finally, a proline residue (Pro66) located in connecting loop I has been observed to adopt a cis conformation in one crystal structure of human GCK.²⁸ Studies in other systems have shown that cis/trans isomerization of proline residues occurs on a sufficiently slow time scale to modulate protein conformational dynamics.^29–31 In this study, we investigate the possibility that cis/trans isomerization of Pro66 contributes to the kinetic cooperativity of human GCK. We also explore the contribution of the length and amino acid sequence of connecting loop I to the enzymatic activity and allosteric regulation of GCK.

Results

Kinetic and thermodynamic consequences of truncating connecting loop I

To investigate how the length of connecting loop I impacts the activity and cooperativity of human GCK, we constructed the following deletion variants: Δ71, Δ70–71, Δ69–71, Δ68–71, Δ67–71, and Δ66–71. Following expression and purification of recombinant enzymes from a glucokinase-deficient bacterial host, to avoid contamination with endogenous glucokinases, steady-state kinetic analyses were conducted on each variant. As shown in Table I, variants lacking one or two residues (Δ71, Δ70–71) display k_cat values, glucose K_0.5 values and ATP K_m values similar to wild-type GCK values. The equilibrium glucose binding affinity is also unaffected by removal of these two residues. The Hill coefficients displayed by the Δ71 and Δ70–71 variants are slightly lower than the wild-type value of 1.7, indicating that cooperativity is slightly reduced in each enzyme.

Table I.

Kinetic and Thermodynamic Parameters of Connecting Loop I Deletion Variants

Enzyme	kcat (s−1)	K0.5, glucose (mM)	kcat/K0.5, glucose (M−1 s−1)	Hill coefficient	Km, ATP (mM)	KD, glucose (mM)a
Wild-type	40.1 ± 0.1	6.3 ± 0.2	(6.3 ± 0.2) × 103	1.8 ± 0.1	0.50 ± 0.10	2.9 ± 0.2
Δ71	33.0 ± 0.1	3.1 ± 0.2	(1.1 ± 0.1) × 104	1.5 ± 0.2	0.29 ± 0.01	2.3 ± 0.5
Δ70–71	29.0 ± 0.2	10.6 ± 0.6	(2.7 ± 0.3) × 103	1.3 ± 0.1	1.2 ± 0.1	6.5 ± 1.8
Δ68–71	(2.6 ± 0.2) × 10−2	2.5 ± 0.2	11 ± 1	1.2 ± 0.1	0.20 ± 0.02	3.0 ± 0.9
Δ67–71	(4.2 ± 0.4) × 10−2	3.2 ± 0.3	13 ± 1	1.1 ± 0.1	0.30 ± 0.01	7.2 ± 1.9
Δ66–71	(3.7 ± 1.3) × 10−2	3.6 ± 0.9	10 ± 4	1.2 ± 0.2	0.20 ± 0.01	2.3 ± 0.5

^aMeasured by fluorescence.

The removal of three or more residues from the C-terminus of connecting loop I resulted in substantial reductions in protein function. Recombinant production of the Δ69–71 variant yielded little soluble protein, resulting in insufficient quantities for kinetic analysis. The Δ68–71, Δ67–71, and Δ66–71 variants produced reasonable yields of soluble, monomeric protein; however, each of these enzymes exhibit k_cat values that were 1000-fold lower than the wild-type value. Removal of four or more residues from connecting loop I also reduced the Hill coefficient to values that are statistically indistinguishable from 1, indicating a complete loss of cooperativity. Surprisingly, the glucose K_0.5 values, the glucose K_D values and the ATP K_m values of the Δ68–71, Δ67–71, and Δ66–71 truncated variants differ from the wild-type values by less than 3-fold.

Installing a stabilizing interaction in connecting loop I

A comparison of the crystal structures of human GCK and human hexokinase I (HK-I) reveals a putative stabilizing interaction between residues Arg518, Asp521, Thr523, and Glu524 within the HK-I connecting loop I, which is absent in GCK [Fig. 2(B)]. In an attempt to recapitulate this internal loop interaction in GCK, we created the S64R–E67D–S69T triple variant. Kinetic characterization of this variant indicates a modest 1.4-fold increase in the k_cat value, and a 2-fold decrease in the glucose K_0.5 value (Table II). The triple variant displays a similar cooperative kinetic response as wild-type GCK, with a Hill coefficient of 1.7.

Figure 2.

Structural and sequence differences between GCK (blue, PDB 1V4S²¹) and the C-terminal domain of HK-I (green, PDB 1HKC³²). (A) The connecting loop from HK-I (black) forms more extensive interactions with the small domain than does the connecting loop from GCK (red). (B) A putative intraloop stabilizing interaction present in HK-I connecting loop I (left) is absent in GCK (right). (C) Sequence comparison of connecting loop I between HK-I and GCK. The three residues that were mutated in an attempt to recapitulate the intraloop interaction shown in B are colored red. The proline residue (Pro66) that adopts a cis conformation is indicated with an asterisk.

Table II.

Kinetic and Thermodynamic Parameters of Connecting Loop I Substitution Variants

Enzyme	kcat (s−1)	K0.5, glucose (mM)	kcat/K0.5, glucose (M−1 s−1)	Hill coefficient	Km, ATP (mM)	KD, glucose (mM)b
Wild-type	40.1 ± 0.1	6.3 ± 0.2	(6.3 ± 0.2) × 103	1.8 ± 0.1	0.50 ± 0.10	2.9 ± 0.2
S64R–E67D–S69T triple variant	56.0 ± 0.2	3.3 ± 0.6	(1.7 ± 0.6) × 104	1.7 ± 0.1	0.24 ± 0.05	3.8 ± 0.5
HK-I identity loop variant	27.5 ± 0.1	4.7 ± 0.9	(5.8 ± 1.2) × 103	1.0 ± 0.2	0.16 ± 0.05	9.5 ± 1.8
HK-I identity loop variant + GKA-22	32.4 ± 0.3	1.6 ± 0.2	(2.0 ± 0.1) × 104	1.0 ± 0.1	n.d.b	n.d.b
P66G	45.0 ± 2.0	6.8 ± 0.4	(6.6 ± 0.7) × 103	1.8 ± 0.1	0.40 ± 0.05	7.3 ± 0.1
P66A	45.0 ± 1.0	9.8 ± 0.3	(4.6 ± 0.2) × 103	1.6 ± 0.1	0.23 ± 0.03	14.0 ± 0.7

^aMeasured by fluorescence.

^bn.d., not determined.

Transplanting connecting loop I from human HK-I yields an active, noncooperative GCK

To explore the extent to which the primary structure of connecting loop I impacts the kinetic features of human GCK, we constructed a variant in which the entire connecting loop was replaced by the corresponding loop sequence from the noncooperative HK-I isozyme [Fig. 2(C)]. To do so, 11 residues in GCK were altered by site-directed mutagenesis. The resulting variant, hereafter referred to as the HK-I identity variant, displays a k_cat value that is 1.5-fold lower than wild-type GCK and an ATP K_m value that is 3-fold lower (Table II). The glucose K_0.5 value of the HK-I identity variant is decreased by 1.3-fold, whereas the glucose K_D value is increased by 3.3-fold. Notably, the kinetic cooperativity of the HK-I identity loop variant is completely abolished, as evidenced by a Hill coefficient of unity.

We also investigated whether transplanting connecting loop I from HK-I abolished the ability of synthetic activators to alter the steady-state kinetics of human GCK. Assays conducted in the presence of 20 µM GKA-22, which is near the solubility limit of this compound, reveal a decrease in the apparent glucose K_0.5 value from 4.7 to 1.6 mM. The k_cat value of the identity variant is also slightly increased (1.2-fold) upon addition of the activator. By comparison, the glucose K_0.5 value of wild-type GCK assayed in the presence of 20 µM GKA-22 is reduced from 6.3 to 1.0 mM, while the k_cat value is unchanged.³³

GCK kinetic cooperativity does not require Pro-66 cis/trans isomerization

Proline 66, which resides in connecting loop I, has been observed to adopt both cis and trans conformations in a crystal structure of GCK determined in the presence of a synthetic allosteric activator (Fig. 3).²⁸ To investigate whether cis/trans isomerization of proline 66 could contribute to the slow conformational changes associated with kinetic cooperativity, we constructed the P66A and P66G variants. Kinetic characterization of these enzymes indicates that the positive cooperativity observed in wild-type GCK is maintained in both enzymes (Table II). The Hill coefficients, k_cat values and glucose K_0.5 values of the P66A and P66G variants are largely unaltered from those of wild-type GCK.

Figure 3.

Crystal structure of GCK (PDB 3F9M²⁸) demonstrating that Pro66 adopts a cis conformation in the presence of a synthetic allosteric activator.

Discussion

Of the four hexokinase isozymes found in humans, GCK has unique structural and functional characteristics that enable it to function as the body's primary glucose sensor.¹⁰ GCK displays a cooperative kinetic response to glucose that is characterized by Hill coefficient of 1.7 and a K_0.5 value of 7 mM.⁹ In contrast, the other human hexokinase isozymes are noncooperative and are ∼100-fold more sensitive to glucose.² Both the kinetic cooperativity and glucose responsiveness of GCK are physiologically important features. A loss of GCK cooperativity is associated with persistent hyperinsulinemic hypoglycemia, and increasing the enzyme's glucose sensitivity with small-molecule allosteric activators is an established diabetes therapeutic strategy.^15,34–36

Models of kinetic cooperativity in monomeric enzymes indicate that sigmoidal steady-state velocities can be observed if the conformational changes associated with substrate binding occur more slowly than does substrate turnover.^18–20 For GCK, substrate turnover occurs with a rate constant of ∼50 s⁻¹ at room temperature. By comparison, typical protein conformational changes are thought to occur in the millisecond to microsecond time regime. Notable exceptions to this expectation are provided by conformational changes involving the cis/trans isomerization of proline residues, which can occur with rate constants as slow as 0.002 s⁻¹.²⁹ That proline cis/trans isomerization might contribute to GCK cooperativity is suggested by the X-ray crystallographic data of Petit et al.,²⁸ which demonstrate that Pro66 in connecting loop I adopts a cis conformation (Fig. 3). This structure, like an earlier structure from Kamata et al.,²¹ was determined in the presence of a small-molecule allosteric activator. In the Kamata structure, the region surrounding Pro66 is also highly strained. Both the threonine residue that precedes Pro66 and the glutamate residue that follows Pro66 occupy disallowed regions of the Ramachandran plot. In the Petit structure, the Pro66 cis conformation is putatively stabilized via a hydrogen bond between the nitrogen atom of the activator and the carbonyl oxygen of Pro66. Here, we demonstrate that replacing Pro66 with alanine or glycine leaves the kinetic parameters and cooperativity of human GCK largely unchanged. Our results suggest that isomerization of this residue does not, by itself, play a significant role in dictating the overall rate of conformational change(s) linked to kinetic cooperativity. The present results do not eliminate the possibility that Pro66 may adopt a cis conformation at some point during the GCK catalytic cycle.

The X-ray crystallographic structures of unliganded and glucose-bound GCK suggest that connecting loop I may play an important role in dictating the conformational transitions upon which cooperativity relies. Connecting loop I separates the large and small domains of GCK, and the opening angle between these two segments changes substantially throughout the catalytic cycle (Fig. 1).²¹ Thus, it is reasonable to postulate that the length of connecting loop I could impact cooperativity. To test this hypothesis, we systematically shortened connecting loop I by sequentially deleting amino acids from the loop's carboxyl terminus. The removal of one or two residues resulted in only minor changes in the k_cat value and Hill coefficient of GCK. However, further truncation of the loop dramatically reduced both turnover number and cooperativity. Cooperativity and turnover number are linked in GCK; cooperativity is only observed when the glucose-mediated conformational reorganizations occur on the same (or a slower) time scale as k_cat. We postulate that the reduction in Hill coefficient observed in the deletion variants is due to the reduced k_cat values, rather than alterations in the nature of the conformational changes themselves. The decreasing turnover rates observed in the deletion variants provide more time for the conformational transitions linked to cooperativity to reach equilibrium. This situation is expected to reduce the appearance of hysteresis in steady-state kinetic profiles.¹⁹ This hypothesis is supported by the observation that the glucose equilibrium binding affinities for all deletion variants are unchanged from the wild-type value. The loop is far removed from the glucose binding site, so no direct impact on glucose affinity is expected by altering the loop. If the equilibrium between the open and closed GCK conformations was altered by loop truncations, one might expect a change in glucose K_D values. The absence of such an effect suggests that the conformational equilibrium between unliganded and glucose bound states is not impacted by changing the length of connecting loop I by up to six residues.

A structural comparison of GCK's connecting loop I with the corresponding loop from the noncooperative human HK-I isozyme reveals some striking differences (Fig. 2). In the unliganded state, the connecting loop of GCK is extended, allowing the large and small domains to be separated by 99°. In this “super-open” conformation, the connecting loop does not form interactions with other regions of the small domain. In contrast, the connecting loop of unliganded HK-I is closely packed against the small domain, forming interactions that restrict the mobility of the α13 helix.³⁷ This compact loop structure is stabilized by an intricate series of five intraloop hydrogen bonds between four loop residues, Arg518, Asp521, Thr523, and Glu524 [Fig. 2(B)]. Together, these features appear to restrict the degree of opening between the large and small domain of HK-I. Of the four residues that participate in this intraloop interaction, GCK possesses only one residue in the same location (Glu524). When the full interaction network is installed into GCK, via the S64R–E67D–S69T triple variant, the cooperativity is unchanged. These results demonstrate that introducing a putative stabilizing loop interaction analogous to that found in HK-I is insufficient to alter GCK cooperativity.

To further modify the properties of the GCK connecting loop I toward that of hexokinase I, we transplanted the entire HK-I connecting loop sequence onto the GCK scaffold. The resulting HK-I identity loop variant required the substitution of 11 amino acids in the GCK connecting loop. The glucose K_m value of the identity loop variant resembles the glucose K_0.5 values of other activated, single site PHHI variants. The glucose K_D value for the HK-I identity loop variant is identical to the wild-type value, suggesting that the overall conformational equilibrium between open and closed conformations is unchanged by substitutions in connecting loop I. Notably, this variant lost all traces of steady-state cooperativity and displays Michaelis–Menten kinetics. The k_cat value of wild-type GCK was only slightly reduced by loop transplantation, suggesting that the loss of cooperativity in this variant is not caused by a decrease in turnover rate that allows the conformational changes to equilibrate during the catalytic cycle. Instead, we postulate that the HK-I identity loop variant displays reduced cooperativity because it possesses altered intrinsic dynamics.

Our results allow us to develop a model for the functional role of connecting loop I in GCK catalysis and allosteric regulation. Our data demonstrate that GCK requires a minimal loop length to achieve high k_cat values. A variant lacking three or more residues is unable to achieve a proper arrangement of catalytic groups in the active site for efficient turnover. In contrast, shortening the length of connecting loop I by as many as six residues does not appear to alter the equilibrium properties of GCK, as evidenced by similar glucose K_D values for the various deletion variants investigated herein. The appearance of kinetic cooperativity in GCK depends upon the amino acid composition of connecting loop I, indicating that the primary sequence dictates the functional properties of the loop. Two characteristics that are likely to be affected by the loop sequence are the flexibility and the extent of interactions formed between the loop and other components of the GCK scaffold. We postulate that connecting loop I in HK-I is intrinsically less flexible than the corresponding sequence from GCK. The decreased flexibility in this loop sequence explains why the identity variant constructed herein lacks cooperativity, while maintaining unchanged steady-state k_cat and K_0.5 values. This model is consistent with the observation that replacing Ser64 or Ser69 with proline, which is conformationally restricted, reduces cooperativity.³⁸ Similarly, small-molecule allosteric activators, which serve to organize and constrain connecting loop I, reduce cooperativity. Ongoing structural and NMR-based dynamic investigations of GCK promise to shed light on the link between connecting loop I flexibility and cooperativity.

Methods

Site-directed mutagenesis

Variants were created via the Quikchange site-directed mutagenesis strategy (Stratagene) using oligonucleotides purchased from Integrated DNA technologies. DNA sequencing using overlapping primers in both the forward and reverse directions confirmed successful mutagenesis.

Protein expression and purification

GCK and variant thereof were produced as recombinant N-terminal hexa-histidine tagged polypeptides in glucokinase-deficient E. coli K-12 strain BM5340(DE3) to avoid contamination with endogenous glucokinases.³⁹ Bacterial cultures were grown at 37°C in Luria-Bertani medium containing ampicillin (150 μg/mL), kanamycin (40 μg/mL), and chloramphenicol (25 μg/mL) and gene expression was induced at 20°C by IPTG (1 mM) when the OD_{600 nm} reached 0.9. After 20 h of growth, bacteria cells were harvested by centrifugation at 8000g for 10 min and the pellet was resuspended in buffer A containing HEPES (50 mM, pH 7.6), KCl (50 mM), imidazole (25 mM), dithiothreitol (10 mM), and glycerol (30% w/v). Cells were lysed using a French Press and subjected to centrifugation at 25,000g at 4°C for 30 min. The supernatant was immediately loaded onto a 5 mL HisTrap Fast Flow affinity Column (GE Healthcare) pre-equilibrated in buffer A. The column was washed with 10 column volumes of buffer A followed by 5 columns of buffer A containing 55 mM imidazole. GCK was eluted with buffer A containing 250 mM imidazole, and the enzyme was dialyzed at 4°C against 1 L of potassium phosphate buffer (25 mM, pH 8.0), containing KCl (25 mM), and dithiothreitol (10 mM). Following dialysis, the protein was injected onto a Superdex 200 16/60 gel filtration column pre-equilibrated with potassium phosphate buffer (25 mM, pH 8.0), containing KCl (25 mM), and dithiothreitol (10 mM). The gel filtration column was run at a flow rate of 0.12 mL/min, and the fractions that contained the highest absorbance readings at 280 nm were pooled and retained for further analysis. The molecular mass of the purified P66A and P66G variants was confirmed via MALDI mass spectrometry on a Bruker autoflex III spectrometer housed in the Biochemical Analysis and Synthesis Service Laboratory at Florida State University. The purity of recombinant enzymes was judged to be greater than 95% based on the results of SDS–PAGE analysis.

Kinetic assays

The GCK activity of individual enzymes was measured spectrophotometrically at 340 nm by coupling the production of glucose 6-phosphate to the reduction of NADP⁺ via glucose 6-phosphate dehydrogenase (G6PDH). Assays were conducted at 25°C in reaction mixtures containing HEPES (250 mM, pH 7.6), KCl (50 mM), NADP⁺ (0.5 mM), dithiothreitol (10 mM), G6PDH (10 units), ATP (0.1–50 mM), GCK (20 nM), and glucose (0.05–100 mM). MgCl₂ was provided at a concentration exceeding the ATP concentration by 1 mM. Initial rate data from the first 10% of reaction progress curves were fitted to the Hill equation or the Michaelis–Menten equation, depending upon the substrate under investigation and the extent of cooperativity detected in individual enzymes. Assays involving variable glucose concentrations were conducted at a saturating concentration of ATP, and assays involving variable ATP concentrations were conducted at a saturating concentration of glucose. Assays were initiated by the addition of ATP and were conducted in duplicate for each substrate concentration. In assays involving GKA-22, the activator was added to a final concentration of 20 µM from a DMSO stock solution, such that the concentration of DMSO did not exceed 0.5%. The kinetic constants reported herein are the average of data obtained from at least two independent preparations of each enzyme.

Equilibrium binding assays

Glucose dissociation constants were determined by monitoring the change in intrinsic tryptophan fluorescence at 338 nm upon addition of varying glucose concentrations (0.010–100 mM), following excitation at 280 nm using a 10 nm slit width. Enzyme (1–30 μM) and glucose were mixed in 0.4 mL, 1 cm path length cuvettes in a buffer containing sodium phosphate (5 mM at pH 7.0), KCl (25 mM), and DTT (10 mM). Reaction mixtures were allowed to equilibrate for several minutes at 25°C prior to measuring fluorescence emission intensity. Equilibrium binding experiments were performed on a Cary Eclipse Fluorescence Spectrometer housed in the Protein Biochemistry Laboratory in the Institute for Molecular Biophysics. Data were collected in duplicate, averaged, and fitted to the following equation: