Regulation of the Interaction of Inosine Monophosphate Dehydrogenase with Mycophenolic Acid by GTP*

Inosine monophosphate dehydrogenase (IMPDH), a rate-limiting enzyme in the de novo synthesis of guanine nucleotides, is a major therapeutic target. A prototypic uncompetitive inhibitor of IMPDH, mycophenolic acid (MPA), is the active form of mycophenolate mofeteil (CellCept®), a widely used immunosuppressive drug. We have found that MPA interacts with intracellular IMPDH in vivo to alter its mobility on SDS-polyacrylamide gels. MPA also induces a striking conformational change in IMPDH protein in intact cells, resulting in the formation of annular aggregates of protein with concomitant inhibition of IMPDH activity. These aggregates are not associated with any known intracellular organelles and are reversible by incubating cells with guanosine, which repletes intracellular GTP, or with GTPγS. GTP also restores IMPDH activity. Treatment of highly purified IMPDH with MPA also results in the formation of large aggregates of protein, a process that is both prevented and reversed by the addition of GTP. Finally, GTP binds to IMPDH at physiologic concentrations, induces the formation of linear arrays of tetrameric protein, and prevents the aggregation of protein induced by MPA. We conclude that intracellular GTP acts as an antagonist to MPA by directly binding to IMPDH and reversing the conformational changes in the protein.

Inosine monophosphate dehydrogenase (IMPDH) 3 catalyzes the first step in the de novo synthetic pathway for the formation of guanine nucleotides by converting IMP to xanthosine 5Ј-monophosphate (XMP) with the concomitant reduction of NAD ϩ . As the rate-limiting step in this pathway, the enzyme has been identified as an important regulator of cell proliferation. Inhibitors of IMPDH are in clinical use as immunosuppressive agents and have potential utility in the treatment of neoplastic and viral diseases (1)(2)(3)(4)(5). Two isoforms of IMPDH have been identified, the type I form that is expressed at lower levels in all cell types and the type II form that is more highly expressed in proliferating and transformed cell types (6,7). These enzymes have 84% amino acid identity, and both are catalytically active as tetramers of 55-kDa subunits with similar, although not identical, kinetic profiles.
Studies on the regulation of IMPDH activity have demonstrated strong transcriptional up-regulation of the type II mRNA in response to mitogenic stimuli in peripheral blood lymphocytes and in response to enzyme inhibition and guanine nucleotide depletion in a number of cell types (8,9). Until recently, very little attention has been paid to the role of protein structure in regulating its activity. The finding that mutations within the IMPDH type I coding region are associated with a familial form of retinitis pigmentosa (10) has initiated renewed interest in the structure of IMPDH, first crystallized in 1996 (11). These mutations occur within a region of the protein termed the cystathione ␤-synthase domain that has recently been shown to bind oligonucleotides up to 100-base pair in length (12). In addition, this domain binds ATP, resulting in allosteric activation of the enzyme (13). It has thus been postulated that IMPDH may act as an energy sensor in cells and have a more complex role in cellular physiology than previously imagined.
In investigating the potential of IMPDH inhibitors as antineoplastic drugs, we have discovered that these inhibitors induce macroaggregates of IMPDH protein that are reversible by the addition of GTP and to a lesser extent by ATP. More spectacularly, most aggregates take the form of ring structures in the cytoplasm of intact normal and neoplastic cells that regress with the repletion of intracellular guanine nucleotides. The unique nature of this protein self-association may account in part for the effects of IMPDH inhibitors and suggest that intracellular levels of GTP may regulate the interaction of IMPDH with small molecule inhibitors.

EXPERIMENTAL PROCEDURES
Cell Culture and Treatment Conditions-The human breast cancer cell line MCF7 was cultured in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum and 100 g/ml penicillin and streptomycin. The human acute lymphoblastic leukemia cell line CEM and hypoxanthine-guanine phosphoribosyl transferase (HGPRT)-deficient hybridoma K6H6/B5 cells (ATCC, Manassas, VA) were cultured in RPMI 1640 medium with the same supplements. HGPRT-deficient CEM cells were maintained in 5 g/ml 6-thioguanine in RPMI 1640 medium with supplements.
Mycophenolic acid (Sigma) was solubilized in ethanol and used at a final concentration of 2 M. Equal volumes of ethanol were added to all untreated controls. Guanosine was added to a final concentration of 100 M. GTP and non-hydrolyzable GTP␥S (Sigma) were used at 1 mM.
Leukemic Cell Isolation and Western Blots-Peripheral blood samples were obtained from a patient with chronic myelogenous leukemia treated with CellCept, an ester pro-drug of MPA, and a normal volunteer. Mononuclear cells were separated using Histopaque-1077 (Sigma). The cells were lysed by three freeze-thaw cycles in an extraction buffer containing 200 mM Tris-HCl, pH 7.5, 200 mM NaCl, 10% glycerol, 2 mM dithiothreitol, 5 mM MgCl 2 , 1 mM EGTA, 2 g/ml leupeptin and pepstatin, and 0.1 mM phenylmethylsulfonyl fluoride. Total lysates (10 g) were run on a 15% SDS-polyacrylamide gel. The Western blot was probed using a polyclonal antibody kindly provided by Vertex (Cambridge, MA).
IMPDH Activity Assay-The enzymatic activity of IMPDH was measured by the increase in absorbance at 290 nm (formation of XMP) in crude cell extracts or at 340 nm (formation of NADH) for purified IMPDH at 37°C. The assay buffer was prewarmed to 37°C and contained 100 mM Tris/HCl, pH 8.0, 100 mM KCl, 50 g/ml BSA, 1 mM dithiothreitol, 2 mM EDTA, 5 mM MgCl 2 , 200 M IMP, 200 M NAD ϩ . The activities from 40 l of whole cell extracts or 5 g of purified IMPDH were recorded over 10 min after adding 1 ml of assay buffer by a temperature-controlled UV-visible recording spectrophotometer UV160U (Shimadzu Co., Kyoto, Japan). The amount of protein was determined by Bradford protein assay, and the activity is expressed as nmol/min/mg protein.
Immunohistochemical and Electron Micrograph Analysis of IMPDH-MCF7 cells were cultured on coverslips in 12-well plates in the absence or presence of 2 M MPA and the indicated additives. Cells were fixed for 15 min in 1 ml of 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) followed by treatment with 0.2% Triton X-100 for 5 min. After 30 min of blocking, the cells were incubated with 100 l of rabbit anti-IMPDH antibody diluted 1:300 in 50% blocking solution (2% fetal bovine serum, 2% BSA, 2% normal goat and rabbit serums in phosphate-buffered saline) for 1 h at room temperature. The cells were then incubated with fluorescein isothiocyanate-labeled goat anti-rabbit conjugates diluted 1:300 in 50% blocking solution for 1 h, washed three times with phosphate-buffered saline, and mounted with Vectorshield (Vector, Peterborough, UK) for fluorescent and confocal microscopy.
For immunonanogold staining and electron microscopy, cells were fixed with 1% glutaraldehyde and blocked with 1% BSA in phosphatebuffered saline for 30 min. The IMPDH antibody in 1% BSA was incubated with cells for 1 h, and the cells were washed and incubated with 1:200 NANOGOLD reagent (Nanoprobe Inc., Yaphank, NY) for 1 h at room temperature. Cells were post-fixed with 1% glutaraldehyde. The nanogold particles were then silver enhanced using a HQ Silver Enhancement kit (Nanoprobe Inc.) according to the manufacturer's instruction. The ultrathin sections were stained with lead-citrate solution and were examined with a Philips CM12 transmission electron microscope.
Gel Filtration of Proteins with FPLC-MCF7 cells cultured with 2 M MPA for 6 h were harvested and briefly sonicated on ice in phosphatebuffered saline containing 100 mM KCl, 1 mM dithiothreitol, 0.1% Tween 20, and protease inhibitor mixture. The whole cell extracts were clarified by centrifugation at 14,000 rpm for 20 min and applied to a Sephacryl S-500 column connected to an FPLC system (Amersham Biosciences). Fractions were collected at a flow rate of 12 ml/h. Equal volumes of proteins from each fraction were resolved by SDS-PAGE and immunoblotted using anti-IMPDH-specific antibody. The relative protein levels of IMPDH were quantitated by a densitometer. Protein standards (Sigma) were applied to the same column to generate a standard curve, from which the molecular masses for IMPDH and its aggregates were calculated.
Recombinant Human IMPDH Purification and Treatment in Vitro-To facilitate purification, the type II IMPDH cDNA was cloned into the pET-24C vector and transformed into Escherichia coli BL21 (DE3). After induction with isopropyl thiogalactoside, the cell pellets were resuspended in 1ϫ binding buffer (Invitrogen) containing 5 mM dithiothreitol, 1 mg/ml lysozyme, and EDTA-free protease inhibitor mixture. Following sonication on ice and centrifugation at 25,000 ϫ g for 20 min at 4°C, the His-tagged IMPDH II proteins were purified on a nickelnitrilotriacetic acid column (Invitrogen). Fractions were collected using a 50 -200 mM imidazole gradient elution and were dialyzed with buffer containing 50 mM HEPES, pH 7.9, and 100 mM KCl. Glycerol was added to a final concentration of 20%, and the samples were stored at Ϫ20°C.
For in vitro treatment, the purified IMPDH was further diluted to a final concentration of 0.1 mg/ml, and 2 M MPA was added for 15 min at room temperature. IMPDH proteins were spread on polylysinecoated glass slides, stained using the immunohistochemical staining procedure, and visualized by fluorescent microscopy at a magnification of ϫ1000. Purified IMPDH was also prepared for electron microscopy at a final concentration of 50 g/ml in HEPES buffer. After treatment with nucleotides, MPA, or the combination, each sample was mounted on glow charged thin carbon foils and stained with 2% uranyl acetate (14). Micrographs were taken on a Philips CM12 electron microscope at 80 kV and a magnification of ϫ45,000.
GTP Binding Assay-Binding of GTP to purified IMPDH was determined with a rapid filtration technique (13,15). 1.0 M [␣-32 P]GTP was mixed with increasing concentrations of cold GTP and incubated with 0.25 M (55 g/ml) IMPDH for 10 min at room temperature in 20 l of reaction buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM KCl, 2.5 M BSA. Nonspecific control values were obtained under the same conditions substituting BSA for IMPDH. 10 l of the reaction mixture were loaded on a Millipore MF filter membrane (Millipore Corp.) rapidly under vacuum and washed with 1 ml of ice-cold reaction buffer. Radioactivity on the filters was measured by scintillation counting, and data were plotted using SigmaPlot (SYSTAT).

RESULTS
Effect of MPA on IMPDH Protein-To study the direct effect of MPA on IMPDH protein, we performed Western blot analysis of IMPDH following MPA treatment. In MCF7 cells, a distinct decrease in the mobility of IMPDH occurred within 5 min of MPA addition (Fig. 1A,  lane 3). The addition of guanosine, which restores guanine nucleotide pools through the HGPRT salvage pathway, completely prevented this band shift (Fig. 1A, lane 2). Approximately 50% of the protein migrated more slowly, and the amount of protein shifted did not change with increased time of incubation with drug (Fig. 1A, lanes 4 and 5). A similar shift has previously been reported for the yeast protein (16). Importantly, this shift was also observed for human IMPDH obtained from the mononuclear cells of a patient treated with immunosuppressive doses of CellCept, an ester pro-drug of MPA (Fig. 1B). The addition of guanosine with MPA completely prevented the shift, whereas guanosine treatment for the last 10 min slightly reduced the shift (Fig.  1C, lanes 3 and 4). The shifted band disappeared after withdrawal of MPA (Fig. 1C, lane 5), reflecting the rapid metabolism of the bound MPA by glucuronidation (17,18). IMPDH activity in MCF7 cells treated with 2 M MPA for 24 h was decreased by ϳ50% (Fig. 1D); this inhibition was completely reversed by the addition of guanosine for 60 min. The changes in activity of IMPDH following MPA treatment correlate with the amount of protein shifted on Western blots.
Aggregation of IMPDH Protein by MPA Treatment in Intact Cells-We then examined the effect of MPA treatment on intracellular IMPDH distribution by immunofluorescent staining. As shown in Fig.  2A, IMPDH is located diffusely in the cytosolic compartment in untreated cells. Within 30 min of MPA treatment, IMPDH protein started to form elongated structures. Within 24 h, we observed a striking aggregation of the protein into perinuclear linear arrays more than 10 microns in length and subsequently into clearly defined annular configurations. This aggregation was seen in all cell lines examined, including a number of leukemic and solid tumor cell lines, as well as in normal peripheral blood lymphocytes. Using laser-powered confocal microscopy, we examined the shape of the aggregates in greater detail. As shown in Fig. 2B, there was a marked predominance of ring formations. Finally, we examined the structures using high resolution electron microscopy with immunonanogold staining of IMPDH in MCF7 cells (Fig. 2C). After silver enhancement, the nanogold staining again demonstrated IMPDH organized into large linear or circular arrays in the cytoplasm. Of particular note is the fact that IMPDH is not associated with any organelles or other defined intracellular structures by this high resolution technique. In other studies, we did not find co-localization of mitochondria, Golgi apparatus, endoplasmic reticulum, lipid bodies, or cytokeratins with IMPDH by light microscopy (data not shown). In addition, extensive co-immunoprecipitation experiments and tandem affinity purification using tagged IMPDH did not demonstrate any interacting protein other than endogenous IMPDH.
Reversal of IMPDH Aggregates by GTP-We then tested whether guanosine reverses IMPDH aggregation by repleting guanine nucleotide pools. MPA-induced IMPDH aggregates are reversed within 1 h by adding 100 M guanosine to MCF7 cells cultured for 24 h in the presence of 2 M MPA (Fig. 3C). In data not shown, the addition of equivalent concentrations of guanine, GMP, GDP, or GTP were equally effective in reversing aggregation, whereas the addition of NAD ϩ , PRPP, xanthine, hypoxanthine, adenosine, AMP, ADP, or ATP had no effect in either MCF7 or CEM cells. We then asked whether the intracellular HGPRT salvage pathway that converts guanine to GMP is necessary for this effect. Guanosine is metabolized to guanine by the enzyme purine nucleoside phosphorylase that is present in fetal bovine serum. The use of HGPRT-deficient cells enabled us to determine whether either guanosine or guanine had a direct effect on IMPDH aggregation or whether they were acting through conversion into guanine nucleotides. As shown in Fig. 3, A and B, guanosine was unable to reverse the IMPDH aggregates in HGPRT-deficient CEM and K6H6/B5 cells. However, the addition of 1 mM GTP resulted in complete dispersal. This result is surprising in that there is no well defined mechanism for the transport of GTP directly from the extracellular to the intracellular compartment. Therefore, these results were confirmed by adding GTP␥S, a non-hydrolyzable GTP analog, to MPA-treated MCF7 cells. The aggregates dispersed, with the subsequent appearance of IMPDH in nucleoli and diffusely within the cytoplasm within 60 min (Fig. 3C). Subsequent experiments using [ 35 S]GTP␥S confirmed that the nucleotide is able to get across the MCF7 plasma membrane in a time-and dose-dependent manner (data not shown). These results confirm that guanosine reverses the MPA-induced IMPDH aggregates through salvage to guanine nucleotides and that GTP is the direct effector. The structural changes in IMPDH correlate with changes of activity. The ϳ50% reduction in activity induced by MPA was completely reversed by the addition of 1 mM GTP␥S to cells for 1 h (Fig. 3D). These data indicate that  increasing the intracellular GTP level overcomes the inhibition of IMPDH induced by MPA.
IMPDH Aggregation Induced by MPA-To further define IMPDH aggregation, the distribution of IMPDH in control or MPA-treated cell extracts was examined using a Sephacryl S-500 size exclusion column (Fig. 4A). Treatment of MCF7 cells with MPA for 6 h resulted in an increase in the size of the IMPDH complex, as detected by Western blot of IMPDH monomers, from a peak of 400 kDa to a peak of 600 kDa with sizes ranging upwards of 2,000 kDa and extending into the void volume. These data support the view that MPA promotes self aggregation of IMPDH protein and that the higher molecular mass multimers are in equilibrium (11,19). Further support for this concept comes from studies on purified recombinant IMPDH type II. His-tagged IMPDH II expressed in E. coli was purified to homogeneity. The protein was treated with MPA, spread on polylysine-coated glass slides, and visualized using standard immunofluorescence microscopy (Fig. 4B). Treatment with MPA results in the formation of large aggregates with a tendency to annular formation, although without the same level of definition found in intact cells. These data demonstrate that MPA causes self aggregation of IMPDH in the absence of other proteins or cellular elements, most likely through conformational change (20).
Effect of GTP on IMPDH in Vitro-To examine the molecular sequelae of both MPA-and GTP-induced changes on the purified pro-tein, we performed uranyl acetate negative staining of protein followed by electron microscopic examination (Fig. 5). This technique allows visualization of IMPDH homotetramers that are 14.6 Ϯ 0.9 nm (n ϭ 10) in diameter. Protein treated only with ethanol exists predominantly as single or ditetrameric units (panel A). MPA treatment for 15 min resulted in the formation of large aggregates of the tetramer ranging in size from 0.1 to Ͼ1 m in diameter (panel D). These aggregates are stable in solution over time at room temperature. The addition of 1 mM GTP alone to IMPDH protein in solution resulted in the formation of linear arrays consisting of IMPDH tetramers (panel B). GTP reversed the IMPDH aggregates formed as a consequence of MPA exposure (panel E), again resulting in linear arrays. ATP, an allosteric activator of IMPDH (13), also increased the formation of linear IMPDH chains (panel C) and reduced the size of MPA-induced aggregates (panel F), although neither effect was as pronounced as that caused by GTP. Both GTP and ATP at 1-2 mM concentrations increased the activity of purified IMPDH by ϳ20% (data not shown). Neither CTP nor UTP had any effect on IMPDH alone or in the presence of MPA (not shown). MPA treatment did not induce protein aggregation after preincubation with GTP or ATP, but did with CTP or UTP (data not shown). These results are summarized in Fig. 6.
IMPDH Binding of GTP-Although it has recently been demonstrated that the cystathione ␤-synthase domains of IMPDH can bind ATP (13), GTP binding has not been demonstrated. Fig. 7 shows that GTP binds to purified IMPDH protein, albeit with low affinity. Data were fitted to a four-parameter logistic curve using SigmaPlot. The apparent K d is 72.9 Ϯ 8.1 M, which is within the physiologically relevant range. This binding also displays positive cooperativity, with a Hill slope of 1.7.

DISCUSSION
IMPDH has long been considered to be an important pharmacologic target, and a number of inhibitors have been developed and used in preclinical studies (21,22). Of these, three are in widespread clinical use. Ribavirin, a broad spectrum antiviral agent used in the treatment of Hepatitis C in combination with interferon-␣, is a relatively weak IMPDH inhibitor (23). Mizoribine (Bredinin) and Mycophenolate mofeteil (CellCept) are used as immunosuppressive agents in the setting of organ transplantation, graft-versus-host disease, and vasculitis in Japan and the United States, respectively (24). Given the clinical importance of IMPDH as a drug target, the enzyme has been crystallized from a number of species and the structural basis of its inhibition has been examined in detail as the basis for new drug design (11,25,26).
Human IMPDH is a tetramer of 55-kDa subunit consisting of ␣/␤ barrels, with the active site located at the monomer-monomer interface. The catalytic site is an ␣/␤ barrel of ϳ400 residues. The substrate (IMP) and cofactor (NAD ϩ ) bind in a continuous cleft on the C-terminal face of each barrel (25). The oxidation of IMP to XMP proceeds through a covalent mechanism involving an active site cysteine residue. The uncompetitive inhibitor MPA binds directly at the NAD ϩ site and traps the XMP as a covalent intermediate in an open configuration of the enzyme. Mizoribine, in contrast, must first be phosphorylated to its monophosphate form and then binds at the substrate binding site, inducing a closed conformation of the enzyme that does not trap XMP (26).
We have shown that MPA induces a shift in IMPDH mobility in cell lines treated with the drug and that this shift is both prevented and reversed by the addition of guanosine to cell cultures, as well as by removal of the MPA from the medium. Because a similar shift occurs when the purified protein is incubated with drug in the presence of IMP and NAD ϩ but is not found in cell lysates following mizoribine treatment (data not shown), we suggest that it results from the enzyme-XMP   intermediate bound to MPA. Recently, a similar mobility shift was reported for the yeast family of IMD enzymes (16) and conclusively shown to result from trapping of the covalently bound XMP intermediate in the presence of MPA. This shift was also reversible with the removal of MPA. These data support the conclusion that the shift results from a conformational change of the IMPDH protein, because the magnitude of the shift considerably exceeds the molecular weight increment of the enzyme-XMP⅐MPA complex. We have now demonstrated the same shift of IMPDH in a mononuclear cell lysate from a patient receiving standard therapeutic doses of Mycophenolate mofeteil (Fig. 1B). This alteration in gel mobility could provide a surrogate marker for IMPDH inhibition in clinical studies of drugs that bind to the NAD ϩ binding site.
We have here demonstrated a unique consequence of IMPDH inhibition, the formation of large intracellular aggregates of protein that result in distinct annular configurations in the cytoplasm. These aggregates are formed in response to MPA, mizoribine, and VX-944, a new IMPDH inhibitor, in all cell types tested. They do not appear to involve any defined intracellular organelles either by immunocytochemical costaining techniques or by electron microscopy. Despite extensive investigation, we have not been able to identify any other interacting proteins in this complex. Finally, treatment of highly purified IMPDH protein with MPA ex vivo results in similar aggregation by both light and negative staining electron microscopy. Thus, this striking configuration of IMPDH appears to be a direct function of an alteration in protein conformation. The ability of mizoribine to induce the ring-like aggregation of IMPDH in intact cells, but not to induce the gel mobility shift, implies that the trapping of XMP is not required for this effect.
The ability of guanosine to reverse the effects of IMPDH inhibition has been well documented in numerous studies on the biologic sequelae of IMPDH inhibition (5,27,28) and has been used to prove that guanine nucleotide depletion is specifically responsible for the biologic effects of IMPDH inhibitors. We have now shown, however, that guanine nucleotides also reverse the conformational change induced in IMPDH by MPA treatment, both in vivo and in vitro. We have also shown that GTP itself can bind to IMPDH protein at physiologic concentrations. Although it is clear that GMP is a competitive inhibitor of IMPDH (29), the direct binding of GTP to IMPDH has not previously been demonstrated to our knowledge. We therefore hypothesize based on our microscopy studies that the IMPDH tetramer forms linear arrays by binding either ATP or GTP, putatively through the cystathione ␤-synthase domains. MPA, by causing conformational change in the protein and by reducing intracellular GTP levels, induces the formation of high molecular weight aggregates. The restoration of intracellular GTP levels reverses MPA-induced aggregation and restores enzymatic activity. The in vitro studies confirm that GTP not only reverses but also prevents IMPDH aggregation.
Although the specific physicochemical basis for these conformational changes has not been established, these studies indicate that the regulation of IMPDH by GTP levels may be important in the multitude of biologic effects of IMPDH inhibitors. Inhibition of IMPDH has been found to inhibit T and B cell activation (30), induce the differentiation of leukemic cell lines (31), cause apoptosis in hematopoietic and solid tumor cell lines (27,32), impair the maturation and function of dendritic cells (33), and interfere with glycosylation, cell adhesion, and nitric oxide synthesis (34,35). The reversal of these important biologic effects by increasing intracellular GTP levels may be in part attributable to effects on IMPDH enzymatic activity and the restoration of de novo guanine nucleotide biosynthesis.