Enzymatic characterization of the pancreatic islet-specific glucose-6-phosphatase-related protein (IGRP).

Glucose is the main physiological stimulus for insulin biosynthesis and secretion by pancreatic beta-cells. Glucose-6-phosphatase (G-6-Pase) catalyzes the dephosphorylation of glucose-6-phosphate to glucose, an opposite process to glucose utilization. G-6-Pase activity in pancreatic islets could therefore be an important factor in the control of glucose metabolism and, consequently, of glucose-dependent insulin secretion. While G-6-Pase activity has been shown to be present in pancreatic islets, the gene responsible for this activity has not been conclusively identified. A homolog of liver glucose-6-phosphatase (LG-6-Pase) specifically expressed in islets was described earlier; however, the authors could not demonstrate enzymatic activity for this protein. Here we present evidence that the previously identified islet-specific glucose-6-phosphatase-related protein (IGRP) is indeed the major islet glucose-6-phosphatase. IGRP overexpressed in insect cells possesses enzymatic activity comparable to the previously described G-6-Pase activity in islets. The K(m) and V(max) values determined using glucose-6-phosphate as the substrate were 0.45 mm and 32 nmol/mg/min by malachite green assay, and 0.29 mm and 77 nmol/mg/min by glucose oxidase/peroxidase coupling assay, respectively. High-throughput screening of a small molecule library led to the identification of an active compound that specifically inhibits IGRP enzymatic activity. Interestingly, this inhibitor did not affect LG-6-Pase activity, while conversely LG-6-Pase inhibitors did not affect IGRP activity. These data demonstrate that IGRP is likely the authentic islet-specific glucose-6-phosphatase catalytic subunit, and selective inhibitors to this molecule can be obtained. IGRP inhibitors may be an attractive new approach for the treatment of insulin secretion defects in type 2 diabetes.

In pancreatic ␤-cells, glucose is the primary physiological stimulus for insulin synthesis and secretion. Phosphorylation of glucose to glucose-6-phosphate is essential for glucose-dependent insulin secretion, triggering an increase of the ATP: ADP ratio, inhibition of ATP-sensitive K ϩ channels, depolarization of ␤-cell membranes, activation of voltage-gated Ca 2ϩ channels, increase of intracellular Ca 2ϩ concentrations, and release of insulin from ␤-cells (1,2). Glucokinase functions as the glucose sensor in the ␤-cell, controlling the rate of entry of glucose into the glycolytic pathway and its subsequent metabolism. Heterozygous mutations in glucokinase lead to maturity-onset diabetes of the young (MODY2) (3,4). Glucose-6-phosphatase (G-6-Pase, 1 EC 3.1.3.9) catalyzes the conversion of glucose-6-phosphate to glucose. Thus, G-6-Pase activity could potentially antagonize the signaling cascade that leads to insulin secretion. Indeed, stable overexpression of the liver G-6-Pase catalytic subunit was shown to attenuate glucose sensitivity of insulin secretion from a mouse pancreatic ␤-cell line, MIN6 (5). In addition, islets from ob/ob mice, a genetic obese/ diabetic model, showed dramatically increased rates of glucose/ Glc-6-P cycling and glucose-6-phosphatase activity when compared with islets from wild-type control mice (6). Similar results have also been obtained in other rodent hyperglycemia models (7,8). These experiments suggest that pancreatic G-6-Pase activity may play an important role in the control of glucose metabolism. G-6-Pase activity has been detected in pancreatic islets (9,10). However, there are conflicting reports on the detection of low levels of G-6-Pase activity in islets (9 -12). In the liver, glucose-6-phosphatase activity is encoded by LG-6-Pase (NCB accession NP_000142). This gene, however, is unlikely to be responsible for G-6-Pase activities in pancreatic islets, since it has only been detected in the liver, small intestine, and kidney (13)(14)(15). Recently, a mouse gene encoding for a protein with 50% amino acid sequence identity to the liver G-6-Pase was cloned from mouse islets and named islet-specific glucose-6phosphatase catalytic subunit-related protein (IGRP). In mice, IGRP is expressed exclusively in pancreatic islets. Surprisingly, no phosphatase activity was detected from the recombinant protein expressed in several mammalian transient transfection systems (12,16). Cloning of the human ortholog was subsequently reported, and no enzymatic activity was demonstrated for the recombinant protein (17). The presence of a specific G-6-Pase catalytic activity in islets therefore remains unresolved.
We examined the enzymatic activity of human IGRP, and found robust enzymatic activity in mammalian and insect cell membranes overexpressing IGRP. IGRP mRNA levels are upregulated in cultured mouse islets under elevated glucose con-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  centrations. Taken together, these data strongly suggest that IGRP is likely the gene responsible for G-6-Pase activity in the islet. Our high-throughput screening of a small molecule compound library (ϳ800,000 compounds) identified several inhibitors that are specific for IGRP activity.

EXPERIMENTAL PROCEDURES
Animal Usage-Male ob/ob, db/db, A y /a and tub/tub in C57BL/6J strain, along with littermate controls, were purchased from the Jackson Laboratory (Bar Harbor, ME). Mouse pancreatic islets were isolated as described previously (18). For the glucose treatment experiment, islets isolated from 6 -8-week-old male B6Af1 mice (The Jackson Laboratory) were dispersed in 20 -30 ml of warm phosphate-buffered saline in a 50-ml tube by vigorous shaking. Dispersed islet cells were resuspended in RPMI 1640/10% fetal bovine serum, and pipetted onto Matrigel (BD Biosciences)-coated plates. After overnight culture, the plates were washed with warm phosphate-buffered saline and replaced with media containing various concentrations of glucose.
Northern blot, Western blot, Quantitative PCR, and in Situ Hybridization-For Northern blot analysis, the mouse pancreas total RNA was prepared from male ob/ob, db/db, A y /a, tub/tub and their respective littermate control mice at 15 weeks of age (RNAgents, Promega). Twenty micrograms of total RNA were loaded to each lane. The blot was prepared using a standard protocol with 1.25% MOPS-agarose gel (Em-biTec, San Diego, CA) and Biotrans membrane (ICN, Costa Mesa, CA) (22). The probe was generated from the mouse IGRP coding region by PCR. The DNA fragment was radiolabeled with [ 32 P]dCTP using Prime-It RmT Random Primer Labeling Kit (Stratagene, La Jolla, CA) and the probe was purified by ProbeQuant TM G-50 Micro Columns (Amersham Biosciences, Piscataway, NJ). Hybridization was done using ExpressHyb solutions (Clontech, Palo Alto, CA) according to the manufacturer's protocol.
For quantitative PCR analysis (Taqman), RNA was extracted from isolated islets, and then treated with deoxyribonuclease I (Invitrogen). First strand cDNA was synthesized using random hexamers (Superscript First-Strand Synthesis System for RT-PCR, Invitrogen). The Taqman primers and probe sets, designed using Primer Express 1.5 software (Applied Biosystems, Foster City, CA), are as follows: forward primer TGAGTTGATGGCGAAACCC, reverse primer ACGCCTTTTG-CTGGACTCG, and probe AGCCAAAGAGGACCCCGAGGTTTCTC. The samples were run on ABI PRISM 7700 Sequence Detector and data was analyzed by Sequence Detector v1.7 (Applied Biosystems). Expression levels of IGRP were first normalized to that of 18 S RNA in each sample. Arbitrary units were then assigned to different samples to indicate their relative abundance.
Western blotting was done according to the protocol of ECL Western blotting analysis system (Amersham Biosciences). His-probe (H-15) from Santa Cruz Biotechnology was used as primary antibody (1:1000 dilution); anti-rabbit IgG, horseradish peroxidase-linked whole antibody (Amersham Biosciences) was used as secondary antibody (1:5000 dilution).
For in situ hybridization, pancreas from db/db and littermate lean mice at the age of 10 weeks were dissected, frozen immediately in liquid nitrogen and stored at Ϫ80°C. 10-micron sections were cut on a Microm HM505E microtome and mounted on Superfrost plus slides (VWR Scientific, Westchester, PA). Slides were air dried for 15 min and stored at Ϫ80°C. Complementary RNA probes were generated from the mouse IGRP coding region by PCR. The hybridization was performed as described previously (23). The controls for the in situ hybridization included the use of corresponding sense probes to show no signal above background in all cases.
Baculovirus Expression and Enzyme Preparation of IGRP-The IGRP cDNA was cloned from a human islet cDNA library prepared. The LG-6-Pase gene was also cloned by RT-PCR using human liver cDNA as the template. Both cDNAs were cloned into pcDNA3.1 for protein expression and enzyme assays. All constructs were confirmed by sequencing. For baculovirus expression, the exact coding sequences encoding the human IGRP and the human LG-6-Pase were cloned into pFastBac (Invitrogen) EcoRI-XhoI sites as either N-terminal-His 6 or C-terminal-His 6 fusions (Fig. 2). Recombinant baculovirus was generated using the Bac-to-Bac TM Baculovirus expression system (Invitrogen). The recombinant pFastBac vectors were transformed into DH10Bac and plated onto Luria Agar plates containing 50 g/ml kanamycin, 7 g/ml gentamicin, 10 g/ml tetracycline, 100 g/ml Bluo-gal, and 40 g/ml isopropyl-1-thio-␤-D-galactopyranoside. White colonies were picked for inoculation and isolation of recombinant Bacmid. Recombinant Bacmid was verified by PCR with the pUC/M13 primers. To generate Baculovirus, the recombinant Bacmid was used for transfecting SF-9 cells using CellFECTINE. At 5 days post-transfection, recombinant baculovirus was harvested, and the cells collected. Cell extract was prepared in lysis buffer containing 50 mM sodium phosphate, pH 8.0, 300 mM NaCl and 1% Tween 20 and total cell extract and soluble fractions were analyzed by Western using polyHis antibodies (Pierce). For protein production, cells were infected with a multiplicity of infections of 0.1 pfu/cell and cell density at harvest of 2.9 ϫ 10 6 cells/ml. The cell pellets were resuspended at about 1 g/7.5 ml of rinse buffer, hand mixed, and microfluidized four times. The disrupted IGRP-expressing cells were centrifuged 15,000 rpm for 30 min at 4°C. The pellets were washed with an equal volume of rinse buffer twice to remove inorganic phosphate. The final protein concentration was determined by the Pierce BCA Protein Plus assay with bovine serum albumin as standard. Attempts were made to solubilize catalytically active IGRP and LG-6-Pase using a wide variety of detergents. These attempts failed due to the instability of the enzyme after its solubilization from membranes (24,25).
Glucose-6-Phosphatase Assay using Malachite Green Detection Method-The IGRP enzyme assay was run using a concentration of 0.38 mg/ml of protein preparation for each 20-l assay. Protein amounts were kept constant and monitored using the Pierce BCA Protein Plus assay. 4 l of ϳ1.9 mg/ml IGRP and 16 l of assay buffer containing 0.62 mM Glc-6-P, 62 mM MES K ϩ pH 6.5 were added to each well to give final concentrations of 0.5 mM G-6-P, 50 mM MES pH 6.5, and 0.38 mg/ml of IGRP enzyme preparation. The reaction was incubated for 40 min and quenched using 80 l of the detection reagent with 0.067% Tween 20. Then, the reaction was incubated at ambient temperature for 35 min for color development. Absorbance was measured at 650 nm using the Molecular Devices SpectraMax (Sunnyvale, CA). Dependence of enzyme activity on divalent and monovalent metal ion concentration, pH, detergents, buffer effect, Me 2 SO tolerance, and incubation time were all determined to forward the final assay conditions. Controls were run in the absence of enzyme and/or substrate. LG-6-Pase was characterized under conditions identical to those for the IGRP assay. LG-6-Pase is more active than IGRP under these assay conditions and thus 10-fold less LG-6-Pase was used in the assays. The final concentration per well of LG-6-Pase was 0.022 mg/ml. For determination of the Michaelis constant of Glc-6-P, the enzyme was incubated at 25°C with 50 mM MES, increasing concentrations of Glc-6-P from 50 M to 5 mM for IGRP and Glc-6-P from 50 M to 5 mM for LG-6-Pase, respectively (Fig. 3).
Glucose-6-Phosphatase Assay using Glucose Oxidase/Peroxidase-coupling Enzymes-Phosphohydrolase activity of IGRP was assayed by monitoring release of glucose from glucose-6-phosphate using modified glucose oxidase/peroxidase coupling system (26). Glucose oxidase catalyzes oxidation of ␤-D-glucose to ␦-D-gluconolactone and hydrogen peroxide. The hydrogen peroxide is then used by peroxidase to convert Amplex Red to resorufin, a molecule with high fluorescence. Fluorescence was monitored using excitation at 530 nm and fluorescence detection at 590 nm. For a glucose standard curve, the appropriate amount of the 50 mM glucose standard solution was diluted into reaction buffer to yield glucose concentrations of 0 to 60 M. The reaction mixture contained 50 M Amplex Red, 5 units/ml glucose oxidase, 0.5 units/ml horseradish peroxidase and 50 mM MES (pH 6.5) in a final volume of 50 l. Fluorescence was monitored in the BMG FluoStar Galaxy (Durham, NC) using excitation at 530 nm and fluorescence detection at 590 nm. For comparison to the kinetic constants for inorganic phosphate formation determined by malachite green assay as described above, the steady-state kinetic constants for glucose formation by IGRP were determined using the same buffer and substrate concentrations as in the malachite green assay plus the 50 M Amplex Red, 5 units/ml glucose oxidase and 0.5 units/ml horseradish peroxidase (Table I and Fig. 3).
Determination of Effects of pH and Metal Concentration on IGRP Activity-All reactions were performed as mentioned above with the addition of one of the following reagents: NaCl, FeCl 2 , CoCl 2 , ZnCl 2 , MnCl 2 , or MgCl 2 at a final concentration of 10 M or 1 mM. The buffers used in the pH study were MES, Bis-Tris propane, Chaps, NaOAc, and Pipes at 50 mM, respectively. Membrane proteins were treated with increasing concentrations of the buffer from 10 to 200 mM to determine the effect of buffer concentration on phosphohydrolase activity of IGRP using glucose-6-phosphate as the substrate.
Stability of Enzyme-The IGRP membrane preparation tolerates one freeze-thaw cycle. A freeze-thaw cycle means flash freezing in a mixture of isopropyl alcohol and dry ice and storing in a Ϫ80°C freezer and thawing on ice before use. After the second cycle of freeze-thaw, ϳ30% of initial activity was lost. In order to prevent loss in enzyme activity and maintain consistent assay condition, aliquots of enzyme were stored at Ϫ80°C and unused portions were discarded after the first freeze thaw cycle.
Determination of Kinetic Constants for IGRP and LG-6-Pase-The optimal assay conditions were determined to be 50 mM MES K ϩ pH 6.5, 0.5 mM Glc-6-P, 0.38 mg/ml IGRP membrane prep concentration in a 20-l total volume in a 384-well black plate with a clear bottom. The reactions were performed as described above and substrate concentrations ranged from 50 M to 5 mM for IGRP and from 50 M to 15 mM for LG-6-Pase. The kinetic constants were obtained by fitting the data to equation 1 using Prism (Graph Pad Software Inc., San Diego, CA), where v is the initial velocity, V max is the maximal velocity, S is the substrate concentration, and K m is the Michaelis constant, shown in Equation 1.
The kinetic data are summarized in Table I. IC 50 Determination of Inhibitors-Known human LG-6-Pase inhibitors, 1 and 2 were tested against the activity of IGRP and LG-6-Pase in protein membranes. From our library of ϳ800,000 compounds, compound 3 has been found to inhibit IGRP activity in a dose-dependent fashion. Peroxyvanadium compounds were also tested as inhibitors of IGRP as well as LG-6-Pase. The results are summarized in Table II. In general, the reaction was run at room temperature for forty minutes in a total volume of 20 l containing 50 mM MES K ϩ pH 6.5, 0.5 mM Glc-6-P, 0.38 mg/ml of IGRP, 5% Me 2 SO and various concentrations of inhibitors from 1.0 nM to 500 M.

RESULTS
The IGRP mRNA Expression in Obese/Hyperglycemic Mice-Many studies have shown that the G-6-Pase activity is detectable, albeit at very low levels, in pancreatic islets (6, 7, 16, 17).
To further explore its biological function, we studied the IGRP mRNA expression level in the pancreas of several obese/hyperglycemia mouse models because up-regulation of G-6-Pase enzymatic activity has been previously reported in diabetic rodent models (13,14). We confirmed that IGRP expression was restricted to pancreatic islets (Fig. 1A). Increased levels of IGRP mRNA were observed by Northern analysis in pancreas isolated from ob/ob, db/db, Ay/a, and tub/tub mice versus their respective control mice. However, it appears that the up-regulation was a consequence of ␤-cell hypertrophy and/or hyperplasia because the IGRP mRNA level was decreased as measured by quantitative PCR in ob/ob (2.8-fold), db/db (1.8-fold), Ay/a (3.0-fold), or unchanged in tub/tub islets (Fig. 1B).
Because IGRP may be an important regulator of glucose utilization, we hypothesized that its levels would change following glucose concentration changes. To examine this, we measured IGRP mRNA levels in cultured mouse islets with various glucose concentrations. As shown in Fig. 1C by quantitative PCR, IGRP mRNA was up-regulated 2.3 and 3.5-fold when normalized to 18 S RNA levels in islets cultured in hyperglycemic concentrations of glucose (11 and 20 mM) for 48 h, respectively.
Glucose-6-Phosphatase Activity by IGRP Protein Expressed in COS7 Cells-Enzyme assays were first conducted at 30°C under the previously described conditions (12) using microsome preparations from the human untagged full-length IGRP cDNA (in pcDNA3.1) transfected COS7 cells. The human untagged full-length liver G-6-Pase cloned into the same vector was used as positive control and the empty vector as negative control. We were able to detect a robust G-6-Pase activity from our IGRP microsome preps consistently. The experiment was repeated many times using independently transfected cells, three of which are shown in Fig. 2A. The enzyme activity was approximately one-twentieth of that measured with liver enzyme preparations under the same conditions while the background activity of COS7 cells with the empty vector was approximately one-tenth of that measured with IGRP microsome preparations. We also obtained similar results with a mouse IGRP clone (data not shown).
Glucose-6-Phosphatase Assay using Malachite Green or Coupling Enzyme System-Membranes containing overexpressed IGRP were characterized for their ability to hydrolyze Glc-6-P using two different methods: detection of free phosphate using malachite green and detection of free glucose using a glucose oxidase/peroxidase coupling system (see "Experimental Procedures"). Initially, the time course of phosphate production was determined using 0.5 mM Glc-6-P to identify the linear ranges and activities at 37°C and ambient temperature. To minimize background signal, fresh Glc-6-P stock solution was prepared daily to reduce the free phosphate present from self-hydrolysis.
Our initial experiments showed a linear increase in the amount of phosphate produced up to 60 min at 37°C and up to 300 min at room temperature, followed by a plateau due to signal saturation. Similar experiments were performed with glucose detection as the read-out, and very similar results were obtained (Table I). Based on these experiments, we chose 20 min at 37°C and 40 min at ambient temperature as our standard assay conditions, and all the data presented below were generated using these conditions.

Baculovirus Expression and Enzyme Preparation of IGRP and LG-6-Pase-The
His-tag was fused to either the N terminus or C terminus of human IGRP. Sf9 cells were transfected with both constructs, and membranes prepared from these cells were evaluated for protein expression (Fig. 2B) and for enzyme activity using a modified malachite green assay (Fig. 2C). Since the C-His-tagged G-6-Pase construct was the most active and  2. IGRP protein expression and activity. A, glucose-6-phosphatase activity by the expressed IGRP protein. Assay was done at 30°C, pH 6.5. Microsomal preparations were from COS-7 cells transfected with human untagged full-length IGRP construct (3 independent transfections), human untagged full-length liver G-6-Pase construct, and pcDNA3.1 empty vector (Null). Shown in B is Western blot of C-His-tagged human IGRP and C-His-tagged human liver G-6-Pase. Mock-transfected cells were used as controls. The IGRP proteins on the immunoblot were visualized by an anti-His monoclonal antibody and 5 min of exposure. Shown in C are activities of mock, irrelevant membrane protein control, N-His-tagged IGRP and C-His-tagged IGRP in Sf9 cells. Shown in D is a time course for release of inorganic phosphate from Glc-6-P by 10-fold above background activity (Fig. 2C), this construct was used for all subsequent experiments. The activity of islet glucose-6-phosphatase was Ͼ10 fold higher than the mock transfection and linear throughout the course of the assay (Fig. 2, C  and D).
Effects of pH, Buffers, Metal Ions, and Me 2 SO on Catalytic Activity-In this study, the effects of pH (4.5 to 9.5), metal ions and buffers on the IGRP activity were characterized. Both IGRP and LG-6-Pase showed the highest activity at pH 5.5-6.0 (Fig. 4). Buffers that contain sulfonic or phosphonic acid moiety were tested as potential inhibitors of the enzyme due to their isosteric characteristics with the substrate and product of the catalyzed reaction. Buffers containing these moieties had little or no effect on the IGRP activity or LG-6-Pase activity using 3,3-dimethylglutaric acid as control buffer (data not shown). For example, the concentration of MES that contains sulfonic acid moiety was varied from 10 mM to 200 mM and had no effect on phosphohydrolase activity of the enzyme. The addition of mono-or divalent metal ions (Na ϩ , Fe 2ϩ , Co 2ϩ , Zn 2ϩ , Mn 2ϩ , or Mg 2ϩ ) to the reaction had no effect on the IGRP or LG-6-Pase enzyme activity. Me 2 SO had very little effect on the IGRP or LG-6-Pase activity and the enzyme could tolerate up to nine percent without a significant loss in activity. A final concentration of 1% Me 2 SO was used for inhibitor characterization.
Determination of Kinetic Constants for IGRP and LG-6-Pase-Kinetic constants for IGRP as well as LG-6-Pase were determined using both inorganic phosphate detection and glucose detection as read-outs at room temperature (Fig. 3). The kinetic constants were measured with concentrations of Glc-6-P ranging from 71 M to 10 mM. Our data clearly show that the recombinant human IGRP has G-6-Pase activity. Importantly, similar kinetic constants were obtained using the two different detection methods (Table I). In addition, the affinity constants determined are comparable to the constants obtained for IGRP as well as LG-6-Pase in its native context in a variety of mammalian species (10).
Inhibition of Glucose-6-Phosphatase-Inhibition of IGRP by previously described LG-6-Pase inhibitors as well as nonspecific inhibitors of phosphatases such as peroxyvanadates was determined. Racemic mixtures of compounds 1 and 2 that are known LG-6-Pase inhibitors were synthesized according to the published procedure (19) and tested as inhibitors of IGRP. As previously reported (27), compounds 1 and 2 inhibited human LG-6-Pase with IC 50 values of 0.84 and 0.31 M, respectively (Table II). Interestingly, compounds 1 and 2 showed no inhibitory action against human IGRP at concentrations up to 100 M (Table II). We screened our library for small molecule in-hibitors of IGRP and found several active compounds, one of which is shown as compound 3. Compound 3 inhibits IGRP activity in a dose-dependent fashion, and with an IC 50 value of 42 M by malachite green detection method and 64 M by glucose oxidase and peroxidase coupling assay. However, compound 3 showed no inhibitory action against human LG-6-Pase at concentrations up to 200 M.
It has been reported that the tungstate and peroxyvanadium compounds are potent inhibitors of G-6-Pase (27,28). It was found that tungstate is a competitive inhibitor against Glc-6-P with a K i value in the range of 10 -25 M with intact rat liver microsomes and in the range of 1-7 M with detergent-treated rat liver microsomes. Peroxyvanadium compounds such as bpV(phen) and bpV(pic) are competitive inhibitors against the substrate, Glc-6-P, with K i values of 0.96 and 0.42 M with intact rat liver microsomes and 0.50 and 0.21 M with detergent-treated rat liver microsomes. We determined the IC 50 values for bpV(HOpic) (compound 4), and bpV(bipy) (compound 5) for human IGRP, respectively, to be 0.60 and 0.60 M (Table II). DISCUSSION Glc-6-Pase catalyzes the hydrolysis of Glc-6-P to glucose and inorganic phosphate. In the liver, this reaction is a critical step in glucose production via gluconeogenesis and glycogenolysis. In the islets, G-6-Pase functionally antagonizes the glucokinase activity, which is critical to glucose utilization and glucose-stimulated insulin secretion (1,2). Our present enzymatic activity data indicate that the IGRP is likely to be responsible for the G-6-Pase activity in islets.
Regulation of IGRP Activity-IGRP activity appears to be regulated in response to glycemic conditions. Increased IGRP mRNA is observed in the hyperglycemic partial pancreatectomy model (29). Interestingly, we show that IGRP mRNA from cultured mouse islets is up-regulated in the presence of elevated glucose levels. Glucose cycling that interconverts between glucose and Glc-6-P has been reported in mouse islets (30), and the cycling activity is reported to increase in islets of the hyperglycemic ob/ob mice, likely due to increased IGRP activity (31,32). We have found that IGRP mRNA levels are relatively modest in ob/ob mice. The precise physiological roles and molecular mechanisms for the regulation of IGRP activity and transcription are not fully understood to date. Taken together with the knowledge that reduced glucokinase activity leads to maturity-onset diabetes of the young (MODY2) (3,4), and our finding that IG-6-Pase transcript is regulated by glucose levels indicates that IGRP may play critical roles in reg-C-His-tagged IGRP (q) and mock control (OE). The enzyme assay was run using a concentration of 0.38 mg/ml of protein preparation for each 20-l assay. 4 l of ϳ1.9 mg/ml either IGRP or mock control and 16 l of assay buffer containing 0.62 mM Glc-6-P, 62 mM MES K ϩ pH 6.5 were added to each well to give final concentrations of 0.5 mM Glc-6-P, 50 mM MES pH 6.5, and 0.38 mg/ml of enzyme preparation. Further details are under "Experimental Procedures." Amount of inorganic phosphate in panel D is net value that background free phosphate was subtracted from the total phosphate. Shown in E is a time course for release of glucose from Glc-6-P by C-His-tagged IGRP and mock control. The reaction was detected by glucose oxidase/peroxidase and Amplex Red. The reaction mixture contained 0.5 mM Glc-6-P, 50 M Amplex Red, 5 units/ml glucose oxidase, 0.5 units/ml horseradish peroxidase and 50 mM MES (pH 6.5) in a final volume of 50 l. Fluorescence was monitored with a fluorescence-based microplate reader using excitation at 530 nm and fluorescence detection at 590 nm. Solid line, IGRP prep; dashed line, mock control; dotted line, background fluorescence determined for a no-glucose oxidase control reaction with mock preparation. ulating glucose utilization in islets, and consequently, pathogenesis in type 2 diabetes. Cloning, Expression and Characterization of the Human Islet-specific Glucose-6-Phosphatase-As expected from its predicted transmembrane topology (33), the IGRP was only detected in the insoluble membrane fraction (Fig. 2B). Consistent with our prediction that higher protein expression would improve the assay, we were able to detect much higher enzymatic activity using the baculoviral C-His-tagged IGRP preps than with N-His-tagged preps (Fig. 2, C and D). The LG-6-Pase was also prepared by expressing the C-terminal His-tagged LG-6-Pase in Sf9 cells. The K m for IGRP is 5-fold lower than that of the LG-6-Pase, and the V max value is 75-fold lower. The use of membrane preparations made it impractical to quantitatively compare the enzymatic efficiency of these two enzymes. However, one plausible explanation as to why K m of LG-6-Pase is higher than that of IGRP is that the intraluminal glucose-6phosphate concentration in liver might be higher than that in islets. The liver T1-translocase that is the rate-limiting enzyme for phosphohydrolysis of Glc-6-P may have different properties from its islet isozyme. In fact, Khan et. al. (34,35) found that both 3-mercaptopicolinic acid and S-3483, the derivative of chlorogenic acid, specifically inhibited rate-limiting hepatic translocase activity of the liver enzyme but not the islet enzyme. IGRP activity is little affected by mono-and divalent metal ions, Me 2 SO, or buffers containing sulfonic or phosphonic acid moiety. Both IGRP and LG-6-Pase have pH optimum at pH 6.0 (Fig. 4).
Glucose-6-Phosphatase Inhibitors-Selective inhibition of IGRP versus LG-6-Pase by small molecules was studied in this work. A series of 4,5,6,7-tetrahydrothieno[3,2-c]-and [2,3c]pyridines was recently reported to be potent inhibitors of the  Phosphohydrolase activity was measured in malachite green system and coupling enzyme system with varying concentrations of Glc-6-P as described under "Experimental Procedures." The inset panels represent the Lineweaver-Burk plots.
LG-6-Pase (19). These compounds (compound 1 and 2) were highly potent with baculoviral LG-6-Pase membrane prep, with IC 50 values as low as 0.31 M. However, compounds 1 and 2 do not inhibit IGRP at a concentration of 100 M (Table II). Although IGRP possesses a 50% sequence homology to LG-6-Pase, this result indicates that highly selective, active sitedirected IGRP inhibitors can be found. In fact, compound 3 obtained from our small molecule screen was shown to inhibit IGRP activity with an IC 50 value of 42 M, and lacked inhibition of LG-6-Pase at Ͼ200 M. Collectively, these results demonstrate that although existing LG-6-Pase inhibitors do not inhibit IGRP, identification of selective IGRP inhibitors is possible. Although the inhibitors are selective against the IGRP versus liver G-6-Pase, it has yet to be shown that they have the same inhibition profile against Glc-6-P hydrolytic activity in crude islet homogenates as that against the recombinantly expressed IGRP.
On the other hand, the nonspecific inhibitors such as tungstate and peroxyvanadium compounds represent a group that inhibits both IGRP and LG-6-Pase. Our results show that the IC 50 values of these compounds were not significantly different between IGRP and LG-6-Pase. The competitive mode of inhibition of these compounds versus Glc-6-P on G-6-Pase is likely due to an isosteric structure shared between tungstate, vanadate, and phosphate. Vanadate and tungstate tend to adopt pentavalent geometries that mimic the transition state for the G-6-Pase catalyzed reaction. The lack of significant difference between IC 50 values determined with IGRP and LG-6-Pase may reflect that the vanadates likely act at the catalytic active site of IGRP.
Comparison with Other Members of Phosphatase Superfamily-It has been shown that many membrane phosphatases including lipid phosphatases as well as a soluble vanadate-dependent chloroperoxidase share conserved consensus sequence, K(X) 6 RP, PSGH, SR(X) 5 H(X) 3 Q/D (36 -38). On the basis of sequence analyses and LG-6-Pase model, it has been proposed that the amino acids responsible for G-6-Pase catalysis are Arg, His, and His (the boldface residues in the signature motif) (39,40). It is not surprising for us to find the enzymatic activity of IGRP because human IGRP, like other members of the phosphatase superfamily, contains this signature motif. The inability to detect high IGRP activity in previous studies may have been the result of different expression systems/conditions (9,10,12). Since human LG6-Pase and IGRP possess 50% amino acid sequence identity, it is possible to speculate on the location of these amino acid residues in human IGRP relative to the membrane. The hydropathy profiles of IGRP predict that the enzyme consists of nine transmembrane helices, which would position Arg-79, His-115, and His-174 on the same side of endoplasmic reticulum membrane. Based on LG-6-Pase model, it is speculated that Arg-79 forms hydrogen bonds to phosphate and stabilizes the transition state, and His-174 serves as the catalytic acid to protonate the 6-hydroxyl group of the glucose and His-115 is the nucleophile to form a phosphohistidine enzyme intermediate in the G-6-Pase reaction. The role of these residues in the IGRP reaction mechanism could be characterized by mutational studies.
In conclusion, our enzymatic activity data support that IGRP is the G-6-pase that catalyzes the hydrolysis of Glc-6-P in pancreatic islets. Future characterization of this gene product using the small molecule inhibitors we identified in this work as well as the knockout mouse model will provide further insight into its biological functions in vivo.