GK (glucokinase) is an enzyme central to glucose metabolism that displays positive co-operativity to substrate glucose. Small-molecule GKAs (GK activators) modulate GK catalytic activity and glucose affinity and are currently being pursued as a treatment for Type 2 diabetes. GK progress curves monitoring product formation are linear up to 1 mM glucose, but biphasic at 5 mM, with the transition from the lower initial velocity to the higher steady-state velocity being described by the rate constant kact. In the presence of a liver-specific GKA (compound A), progress curves at 1 mM glucose are similar to those at 5 mM, reflecting activation of GK by compound A. We show that GKRP (GK regulatory protein) is a slow tight-binding inhibitor of GK. Analysis of progress curves indicate that this inhibition is time dependent, with apparent initial and final Ki values being 113 and 12.8 nM respectively. When GK is pre-incubated with glucose and compound A, the inhibition observed by GKRP is time dependent, but independent of GKRP concentration, reflecting the GKA-controlled transition between closed and open GK conformations. These data are supported by cell-based imaging data from primary rat hepatocytes. This work characterizes the modulation of GK by a novel GKA that may enable the design of new and improved GKAs.
GK (glucokinase; hexokinase IV) plays a central role in maintaining glucose homoeostasis and is the major glucose-phosphorylating enzyme expressed in hepatocytes and pancreatic β-cells . GK is unique among hexokinases in that it displays a sigmoidal substrate dose–response curve, demonstrates low affinity and positive co-operativity for substrate glucose, and is not susceptible to product (glucose 6-phosphate) inhibition [2,3]. These properties are critical to the role GK plays as the glucose sensor. Given its pivotal role in regulating glucose homoeostasis, there has been significant interest in GK as a target for treating T2DM (Type 2 diabetes mellitus). A number of small-molecule GKAs (GK activators) have been discovered that allosterically enhance GK activity . In vitro, these GKAs increase GK activity by enhancing the apparent affinity for glucose and/or increasing the turnover rate (kcat) of the enzyme . In vivo studies revealed that GKAs were able to decrease glucose levels in normal and diabetic animal models [4,6–8]. However, although early-generation dual-acting (liver and pancreas) GKAs were efficacious, many were also reported to have significant hypoglycaemia risks that have been attributed, in part, to increasing insulin secretion at inappropriately low glucose levels. As an alternative, hepatoselective activators may avoid this hypoglycaemia liability, albeit while potentially offering less efficacy . Hepatic GK activity, unlike pancreatic GK activity, is regulated by an endogenous 68 kDa regulatory protein inhibitor called the GKRP (GK regulatory protein) . Under conditions of limited nutrients, GKRP associates with and sequesters GK in the nucleus. Under nutrient-rich conditions, the GK/GKRP complex dissociates and GK translocates to the cytoplasm [11,12]. There is evidence from rodent models of T2DM that this translocation is impaired in the disease state  and human genome-wide association studies support a linkage between GKRP polymorphisms and serum-fasting plasma glucose levels [13,14]. Physiologically, the association between GK and GKRP is enhanced by F-6-P (fructose 6-phosphate) and diminished by F-1-P (fructose 1-phosphate) , with evidence suggesting that both sugar phosphates bind to the same site on the regulatory protein . In addition, GKRP associates with GK in a manner that is competitive with glucose , and there is evidence that GKRP is a tight-binding inhibitor of GK .
A model based on crystallographic evidence has been proposed to account for the kinetic co-operative behaviour of GK. The model assumes two distinct slowly interconverting conformations induced by a major reorientation following ligand binding. The first conformation is an inactive super-open form with low affinity for glucose, whereas the second is an active closed form with a higher affinity for glucose . When glucose binds to the higher-affinity lower-abundance conformer, equilibration increases the proportion of the higher affinity form [19,20]. Pre-steady-state kinetic analysis supports this model . Further refinements demonstrate a landscape of conformations available between the extremes of the forms super-open and closed the forms as well as the existence of a GK–GKA complex in the absence of glucose . Therefore both kinetic and structural studies support the hypothesis that multiple interconverting conformations of GK exist with discrete rate-limiting domain reorganizations. This model allows one to kinetically explore GK modulation by glucose, small-molecule activators and GKRP.
In the present study, we show with steady-state kinetic methods that there are at least two kinetically distinct forms of GK that interconvert through a slow conformational change and that this interconversion is affected by glucose concentration and a liver-specific GKA (J. A. Pfefferkorn, A. Guzman-Perez, J. Litchfield, R. Aiello, J. L. Treadway, J. Pettersen, M. L. Minich, K. J. Filipski, C. S. Jones, M. Tu, G. Aspnes, H. Risley, S. W. Wright, J. Bian, B. Stevens, P. Bourassa, T. D'Aquila, L. Baker, N. Barucci, A. S. Robertson, F. Bourbonais, D. R. Derksen, M. MacDougall, C. Over, J. Chen, A. L. Lapworth, J. A. Landro, K. Atkinson, N. Haddish-Berhane, B. Tan, L. Yao, R. E. Kosa, M. V. Varma, B. Feng, D. B. Duignan, A. El-Kattan, S. Murdande, S. Liu, M. Ammirati, J. Knafels, P. DaSilva-Jardine, L. Sweet, L. Spiros and T. P. Rolph, unpublished work). We demonstrate that GKRP is a slow tight-binding inhibitor of GK, and explore the modulation of this mechanism with a GKA. Corroborative modulation of GK by glucose and a GKA is demonstrated using a cellular image-based assay. A detailed understanding of the regulation of GK in the liver by GKRP in the presence of a GKA is of particular importance for understanding the effects of small-molecule GK activation in the liver and for the design of new GKAs.
G6PDH (glucose-6-phosphate dehydrogenase), insulin, dexamethasone M2 anti-FLAG resin and hexokinase I from Saccharomyces cerevisiae were purchased from Sigma. Collagen-coated plates were from BD Biosciences. Hoechst 33342, FBS and all DMEM (Dulbecco's modified Eagle's medium) were purchased from Invitrogen. Alexa Fluor® 594-labelled goat anti-rabbit antibody was purchased from Santa Cruz Biotechnology. All other chemicals and reagents were purchased from various commercial sources at the highest level of purity available. All data were analysed with the indicated equations using GraphPad Prism. Statistical analysis was carried out using SAS 9.3 software.
Protein expression and purification
Recombinant human β-cell N-terminally hexahistidine-tagged GK was purified from Escherichia coli as described previously . The purified protein was stored in 25 mM Hepes, 150 mM NaCl, 5 mM DTT (dithiothreitol) and 5% (v/v) glycerol pH 7.5 at −80°C. Recombinant human FLAG-tagged GKRP was cloned as described in  and expressed in and purified from Sf9 insect cells using baculovirus-infected cell technology . Cells were harvested 72 h post-infection and protein purified using affinity column chromatography (M2 anti-FLAG resin) followed by size-exclusion column chromatography (Superdex 200). Purified GKRP was stored in 25 mM Tris/HCl, pH 7.5, 150 mM NaCl, 10% (v/v) glycerol and 2 mM TCEP [tris-(2-carboxyethyl)phosphine] at −80°C.
GK activity was measured by monitoring the rate of glucose 6-phosphate formation using the G6PDH/NADP-coupled assay  with minor modifications. All enzyme assays were carried out in 50 mM Hepes, pH 7.1, 10 mM MgCl2, 25 mM KCl, 2.5 mM ATP, 1.1 mM NADP, 0.1% BSA, 2 mM DTT and 1 unit/ml G6PDH at 25°C in a final volume of 40 μl. GKA stock solutions were prepared in 100% DMSO and the final concentration of DMSO in the assay never exceeded 1%. All reactions were carried out in a 384-well microtitre plate (Corning 3702) and the data were collected continuously at 340 nm on a Spectramax 384 Plus.
To study the kinetics of GK activation, various concentrations of glucose or glucose plus compound A were added to GK in a glucose-free assay buffer. Data acquisition was initiated within 15 s and the reaction was monitored continuously every 10 s for time periods up to 15 min. To study the kinetics of GK inhibition by GKRP, various concentrations of GKRP or GKRP plus F-6-P were added to GK in assay buffer containing 20 mM glucose or 20 mM glucose and a GKA. Data acquisition was initiated within 15 s and the reaction was monitored continuously every 15 s for time periods up to 30 min. All reactions were run in triplicate or more and results are means±S.D.
Analysis of the steady-state kinetics of GK inhibition by GKRP
where ET is the total concentration of active enzyme, IT is the total concentration of inhibitor, Kiapp is the apparent dissociation constant of inhibitor, v is the initial velocity of the reaction in the presence of inhibitor at concentration [I] and v0 is the initial velocity of the reaction in the absence of inhibitor. For the case of substrate and inhibitor competing for the same enzyme active site, Kiapp is related to the true dissociation constant Ki by the relationship shown in eqn (2):
where Km is the Michaelis constant and [S] is the competing concentration of the substrate.
Analysis of GK activation and inhibition kinetics
Progress curve analysis allowed for the determination of the first-order rate constant (k) for the transition of GK from the initial rate (v0) to the steady-state rate (vs). The progress curves for both activation and inhibition of GK were analysed by eqn (3) :
where product p represents the absorbance at 340 nm at any time t, v0 and vs are the initial and steady-state velocities of the reaction in the presence of the inhibitor respectively, and k is the rate of the transition from v0 to vs.
The derived rate constants (k) for the inhibition of GK with varying amounts of GKRP, with and without F-6-P, were further analysed using eqns (4)–(6) to determine Ki′. For the case of tight-binding inhibition, the concentration of the EI complex is not negligible in comparison with the concentration of I, so the assumption that free and total inhibitor concentrations are equivalent is not valid. Eqn (4) allows for calculation of [EI] for a tight-binding inhibitor:
where [EI] is the concentration of enzyme–inhibitor complex, IT is the total concentration of inhibitor and ET is the total concentration of enzyme. The relationship between the rate of transition from initial to final velocity described above (k) and the kinetic constants for the forward rate of formation of the final, tight EI complex (kf) and the reverse rate back to the initial EI complex (kr), is given by eqn (5):
where IF is the concentration of free inhibitor calculated from the relationship IF=IT−EI and Kiapp is as described above. Global fitting of data to eqns (4) and (5) yields the best-fit values for kf and kr. The relationship between the initial velocity Ki' and the steady-state velocity Ki is given by eqn (6):
GK translocation assay
The translocation of GK from the nucleus to cytoplasm in primary hepatocytes was performed as described in  with minor modifications. Cryopreserved hepatocytes from Wistar rats were plated into collagen-coated 96-well plates (60000 cells/well) in DMEM containing FBS, 100 nM insulin, 10 nM dexamathasone and 2 mM sodium pyruvate. Following 4 h of incubation in a humidified incubator at 37°C and 10% CO2, fresh medium containing Matrigel was added to the cells and they were incubated overnight. The medium was changed to DMEM (no glucose) with 10% FBS, 100 nM insulin, 10 nM dexamethasone, 2 mM L-glutamine and the cells were incubated for 12–16 h. Medium was aspirated and glucose at concentrations of 2.5 or 8.9 mM was added to DMEM containing 0.07% BSA, 1 nM insulin and 10 nM dexamethasone. Compound A was added to the cells at concentrations ranging from 0.3 nM to 100 μM. Following a 1-h incubation at 37°C and 10% CO2, cells were fixed with 8% (v/v) paraformaldehyde for 15 min followed by 4% paraformaldehyde for 30 min at 4°C. Cells were washed with PBS and permeablized using 0.1% Triton X-100 in PBS for 20 min at room temperature (23°C). Following washing with PBS, cells were treated with 50 μl of blocking solution (3% BSA and 1% goat serum in PBS) and incubated at room temperature for 1 h. An anti-GK antibody (diluted 1:250 in blocking solution) was added to the cells and incubated at 4°C overnight. Cells were washed and incubated with an AF594-conjugated goat anti-rabbit antibody (diluted 1:500) with 2 μM Hoechst in PBS at room temperature for 1 h.
Image analysis for GK translocation
Array Scan (Cellomics, Thermo Fisher Scientific) imaging was performed using an λex of 364 nm for the Hoechst 33342 dye and an λem of 594 nm for the AF594/GK signal. Images were analysed using a compartmental analysis algorithm to determine the GK distribution between the nucleus and cytoplasm. Compound response (y) was defined as change in pixel intensity in the red GK signal channel between the nucleus and cytoplasm relative to the values at different glucose concentrations (2.8 or 8.9 mM) in the extracellular medium. GraphPad Prism was then used to fit dose–response curves to eqn (7):
where x is the concentration of the compound, y is the compound response and Hill indicates the Hill coefficient.
Observation of the assay transient
The previous work has demonstrated that at least two distinct forms of GK exist and interconvert via a substrate-induced conformational change as observed using standard kinetic methods . We wanted to confirm this observation and determine the effect of a GKA on the phenomenon. By monitoring GK progress curves after GK was transferred from a glucose-free assay buffer to a buffer containing glucose, we were able to observe a lag in product formation where the initial velocity v0 was lower than the steady-state velocity vs (Figure 2). The transition time from the initial to the steady-state velocity was dependent on the concentration of glucose, and the transition rate constant kact could be calculated from eqn (3). For glucose concentrations up to 1 mM, the progress curves were linear for 5 min with no observed difference between the initial and steady-state velocities (Figure 2A, left-hand y-axis). For glucose concentrations above 1 mM, there was a measurable difference between the initial and steady-state velocities, and this was reflected in the value of kact (Figure 2A, right-hand y-axis and Table 1). At 5 mM glucose, the value of kact was 0.025±0.001 s−1 and remained constant up to 50 mM glucose (results not shown).
Observation of the GK assay transient as a function of glucose and compound
|v0 (nmol·mg−1·s−1)||vS (nmol·mg−1·s−1)||kact (s−1)|
|Glucose (mM)||DMSO||Compound A||DMSO||Compound A||DMSO||Compound A|
|v0 (nmol·mg−1·s−1)||vS (nmol·mg−1·s−1)||kact (s−1)|
|Glucose (mM)||DMSO||Compound A||DMSO||Compound A||DMSO||Compound A|
The transition of GK from initial to steady-state velocity was also measured in the presence of a novel GKA, compound A (J. A. Pfefferkorn, A. Guzman-Perez, J. Litchfield, R. Aiello, J. L. Treadway, J. Pettersen, M. L. Minich, K. J. Filipski, C. S. Jones, M. Tu, G. Aspnes, H. Risley, S. W. Wright, J. Bian, B. Stevens, P. Bourassa, T. D'Aquila, L. Baker, N. Barucci, A. S. Robertson, F. Bourbonais, D. R. Derksen, M. MacDougall, C. Over, J. Chen, A. L. Lapworth, J. A. Landro, K. Atkinson, N. Haddish-Berhane, B. Tan, L. Yao, R. E. Kosa, M. V. Varma, B. Feng, D. B. Duignan, A. El-Kattan, S. Murdande, S. Liu, M. Ammirati, J. Knafels, P. DaSilva-Jardine, L. Sweet, L. Spiros and T. P. Rolph, unpublished work). Compound A is a potent and selective activator optimized for recognition by liver specific OATPs (organic anion transporting polypeptides). This activator has enhanced hepatic uptake, low hepatic oxidative metabolism and low passive permeability to minimize distribution into peripheral tissues that lack OATP transporters such as pancreas. Biochemically, compound A demonstrated in vitro activation of human recombinant GK with an EC50 of 90 nM. Moreover, compound A shifted the S0.5 for glucose from 13 to 5.9 mM, and increased the Vmax 1.5-fold at concentration of 2.3-fold EC50 (i.e. 210 nM). In the present study, when tested at 210 nM, there was a noticeable difference between the initial and steady-state velocities for all glucose concentrations tested. The value of kact measured at 1.0 mM glucose was 0.01±0.002 s−1, whereas at 5 mM glucose, the value of kact was 0.006±0.0007 s−1, on average 3.6 times larger than in the absence of activator. The progress curves for 1 and 5 mM glucose are shown (Figure 2A, left- and right-hand y-axis respectively). Values for v0, vs and kact for 1, 3, 5 and 10 mM glucose are shown in Table 1. Two experiments support that the observation of the transient is not due to an artefact of the assay system. When GK is pre-incubated with compound A and glucose (5 mM), there is no observable difference between the initial and steady-state velocities, which is distinct from the observation when there is no pre-incubation step (Figure 2B). Secondly, no transient is observed under any conditions when hexokinase family members I, II or III, enzymes that do not display co-operative behaviour with substrate, are substituted for GK. It should be noted that the value of kact, although influenced by the on-rate of compound A for GK, was not dominated by this event (results not shown).
Effect of compound A on time dependence of GK inhibition by GKRP
GKRP is an endogenous inhibitor of GK in hepatocytes. Previous work has demonstrated that GKRP inhibits GK in a manner that is competitive with glucose  that the association between GK and GKRP is enhanced by F-6-P  and that GKRP inhibits GK in a tight-binding manner . Given the tight-binding nature of the inhibition of GK by GKRP, we investigated the time dependence of the interaction. Preliminary experiments indicated that GK inhibition by GKRP was time dependent and slow enough to be measured under standard steady-state assay conditions. Experiments monitoring GK inhibition by GKRP as a function of time in the absence and presence of 210 nM compound A were performed after a 20 min pre-incubation of GK with glucose, DMSO vehicle or compound A, and F-6-P. The assay was initiated by the addition of standard assay reagents plus various concentrations of GKRP, and had a final concentration of 100 μM F-6-P. The data clearly show that the inhibition was dependent on time and GKRP concentration for experiments performed without and with compound A (Figures 3A and 3B respectively). In the absence of compound A, non-linear regression analysis of progress curves using eqn (3) showed that the apparent rate constant describing the time dependence for the transition from initial velocity v0 to steady-state velocity vs, kobs, increased with GKRP concentration. When kobs was plotted as a function of GKRP concentration, the data followed a biphasic hyperbolic function (Figure 3C, closed squares), indicating that a fast equilibrium preceded a final slow-dissociating complex thus supporting a two-step tight inhibition mechanism. When these data were analysed with eqns (5) and (6), the values of kr and kf were 0.0014±0.0002 and 0.011±0.003/s respectively, and the steady-state Ki value reported was 12.8 nM, whereas the initial Ki value (Ki′ in eqn 6) was 113 nM. The steady-state value of Ki was similar to the value of 3.1±0.2 nM determined by a pre-incubation of GK (40 nM) with variable GKRP concentrations for 20 min in the presence of 500 μM F-6-P, and analysis of the data using eqns (1) and (2). When compound A was included in the pre-incubation mixture, time-dependent inhibition was still observed, but the value of kobs did not appear to vary with GKRP concentration (Figure 3C, closed circles), suggesting the approach to vs was governed by a conformational change reflecting the transition of GK from a compound A-induced closed form to the open form. Statistical analysis of these data was performed using both a linear model and ANOVA. When GKRP concentration was treated as a continuous variable, a linear model between the kobs value and the independent variable GKRP suggested that kobs was not likely to be linearly associated with GKRP concentration (the P value of the significance testing for the slope was 0.1031). When GKRP concentration was treated as categorical variable, ANOVA of kobs with GKRP concentration as fixed effect suggested that kobs was not likely constant across GKRP concentrations (the P value of the significance testing for the fixed effect of levels of GKRP was <0.0001). Although there exist statistically significant differences between some concentrations of GKRP in terms of the kobs value, those differences do not appear to be a function of GKRP concentration.
Progress curves for the inhibition of GK by GKRP
Cell-based GK translocation assay
A cell-based assay was used to translate our biochemical observations of the effect of compound A on the modulation of GK activity by GKRP to a more physiological setting. The translocation of GK from the nucleus to the cytoplasm was measured in primary rat hepatocytes and monitored by high-content image analysis . In primary rat hepatocytes cultured under low glucose concentrations, GKRP retained GK in the nucleus in an inhibited state. In the presence of increased glucose concentration, the interaction between GK and GKRP was de-stabilized; GK was subsequently released and translocated to the cytoplasm, where the active conformer phosphorylated glucose. In addition to glucose concentration, the effect of compound A was tested in this assay. As shown in Figure 4(A), increasing glucose concentration caused a reduction in the nucleus to cytoplasm GK intensity, reflecting the translocation of GK from the nucleus to cytosol. Compound A further increased the cytoplasmic shift of GK in a dose–dependent manner. As shown in Figure 4(B), the EC50 value decreased with an increasing glucose concentration, from an EC50 of 1.46 μM at 2.5 mM glucose to 0.26 μM at 8.9 mM glucose.
Effect of compound A on GK translocation from the nucleus to the cytoplasm in cryopreserved hepatocytes from Wistar rats
The unique properties of GK, a monomeric protein displaying positive co-operativity with respect to glucose, permits responsiveness to glucose in the relevant physiological range of 2.5–5.0 mM glucose observed in the normal glycaemic state and over a broader range (7.0–15 mM) under conditions of hyperglycaemia characteristic of T2DM. Although a number of models have been proposed to account for experimental observations, the ‘mnemonical’ model , which has been further refined as the pre-existing equilibrium model , appears to be supported by a wealth of experimental information . In the simplest scenario, GK exists in two forms with different glucose-binding affinities to give the same kinetically indistinguishable enzyme–glucose complex. As glucose concentration increases, the co-operativity of glucose binding is reflected in an equilibrium shift from a higher abundance lower glucose affinity form E to a higher affinity form E*. The lower-affinity form was observed in our experiments (Figure 2) at concentrations of glucose up to 1 mM, as evidenced by the linear GK progress curve. As glucose concentration increased beyond 1 mM, the transition from the low-affinity form to the high-affinity form was reflected in the steady-state transition from v0 to vs. The transition kact, which we observed by our steady-state method, reflected a combination of glucose binding and kinetic constants describing conformational changes from E to E*. The addition of compound A alters the GK affinity for glucose and the kinetic rate constants describing the conformational changes leading to GK activation. We showed that compound A increases the steady-state velocity of GK at the lowest concentration of glucose tested (1 mM), by driving the equilibrium towards a higher activity state. At 1 mM glucose, the v0 to vs transition was measurable and the value of kact was found to be 0.01 s−1. We also observed an increase in transition time of approximately 4-fold at the higher glucose concentrations tested. Several potential models may explain the increased transition time: compound A may simply interfere with the transition from E to E* along the glucose-induced conformation path; compound A may direct the transition from E to E* along an alternate conformational pathway; or compound A may induce an alternative energy state of GK that is distinct from E*. Given the observable transition from v0 to vs at 1 mM glucose, the increased glucose affinity, the increased Vmax, and longer transition time from v0 to vs for glucose concentrations at or above 3 mM, we propose that compound A induces an alternative GK conformation E** that is distinct from E*. It is noteworthy that compound A activates GK at glucose concentrations that are considerably lower than what are considered euglycaemic. However, given that compound A is largely confined to the liver, the possibility of hypoglycaemia resulting from pancreatic insulin secretion is minimized.
The interaction between GK and GKRP may shed light on the how compound A influences GK conformation and enzyme activity. Compound A did not alter the affinity between GK and GKRP; however, compound A increased GK activity by 3.5-fold relative to the control in the presence of GKRP, but interestingly in the absence of GKRP, compound A only increased the activity of GK by 1.5-fold over glucose alone (results not shown). This increased relative activation of GK by compound A in the presence compared with the absence of GKRP suggests that in the presence of GKRP, a second mechanism may be operative. Glucose is competitive with GKRP for binding to GK and we report that compound A increased the affinity of GK for glucose. It follows that in the presence of compound A, less GK would be bound to GKRP, creating a larger active GK population. This phenomenon was confirmed in rat hepatocytes when GK compartmentalization was monitored using cell imaging technology, where compound A was shown to potentiate the action of glucose in disrupting the interaction between GK and GKRP and promoted the translocation of GK from nucleus to cytosol. We also observed that in the presence of compound A, the time-dependent inhibition of GK by GKRP was no longer dependent on GKRP concentration. This observation suggests that the rate-limiting step of the inhibition may be governed by an E** to E* transition. Alternatively, compound A may simply drive the equilibrium from E towards the closed more active GK conformer E*. Although plausible, this explanation does not account for the changes in glucose affinity, increases in GK activity in response to compound A, and does not account for the effect compound A has on the value of kact. A third possibility is that by binding to the allosteric site on GK, compound A may slow the transition from E* back to E. This mechanism is consistent with the increase in transition time from v0 to vs, but does not account for the increased affinity for glucose nor does it explain the increase in enzyme activity. Therefore a transition from E** to E* is the likely rate determining step in the time-dependent inhibition of GK by GKRP in the presence of compound A.
In summary, our observations about the effects of GKAs on GK kinetics in the presence and absence of GKRP may, in part, enable the further characterization of GKAs and the relationship between activation and pharmacological effects.
All authors contributed to the writing of the paper. In addition, specific contributions are as follows: Frank Bourbonais and Cong Huang performed the biochemical experiments and made intellectual contributions to the design and interpretation of those experiments. Jing Chen performed the cell-based translocation experiments and made intellectual contributions to the design and interpretation of those experiments. Yanwei Zhang performed a statistical analysis and interpretation of the data appearing in Figure 3(C). Jeffrey Pfefferkorn is the project leader for the Pfizer Glucokinase programme and made intellectual contributions to various aspects of the work. James Landro made intellectual contributions to all aspects of the work appearing in the paper.
We thank Paul DaSilva-Jardine, Ann Aulabaugh, Dave Beebe, Jessica Ward and Rich Derksen for a critical reading of the paper.