Isolation and characterization of mycophenolic acid-resistant mutants of inosine-5'-monophosphate dehydrogenase.

Mycophenolic acid (MPA) is a potent and specific inhibitor of mammalian inosine-monophosphate dehydrogenases (IMPDH); most microbial IMPDHs are not sensitive to MPA. MPA-resistant mutants of human IMPDH type II were isolated in order to identify the structural features that determine the species selectivity of MPA. Three mutant IMPDHs were identified with decreased affinity for MPA. The mutation of Gln277 → Arg causes a 9-fold increase in the Ki of MPA, a 5-6-fold increase in the Km values for IMP and NAD, and a 3-fold decrease in kcat relative to wild type. The mutation of Ala462 → Thr causes a 3-fold increase in the Ki for MPA, a 2.5-fold increase in the Km for NAD, and a 1.5-fold increase in kcat. The combination of these two mutations does not increase the Ki for MPA, but does increase the Km for NAD 3-fold relative to Q277R and restores kcat to wild type levels. Q277R/A462T is the first human IMPDH mutant with increased Ki for MPA and wild type activity. The third mutant IMPDH contains two mutations, Phe465 → Ser and Asp470 → Gly. Ki for MPA is increased 3-fold in this mutant enzyme, and Km for IMP is also increased 3-fold, while the Km for NAD and kcat are unchanged. Thus increases in the Ki for MPA do not correlate with changes in Km for either IMP or NAD, nor to changes in kcat. All four of these mutations are in regions of the IMPDH that differ in mammalian and microbial enzymes, and thus can be structural determinants of MPA selectivity.

Inosine-monophosphate dehydrogenase (IMPDH) 1 catalyzes the oxidation of IMP to XMP with the concomitant conversion of NAD to NADH (Fig. 1). This reaction is the rate-limiting step in guanine nucleotide biosynthesis, and rapidly growing cells have increased levels of IMPDH (1). Inhibitors of IMPDH have antiproliferative activity and are used clinically for cancer, viral, and immunosuppressive chemotherapy (2)(3)(4). Moreover, differences in the properties of microbial and mammalian IM-PDHs suggest that species-selective IMPDH inhibitors can be designed, which will be useful for anti-infective chemotherapy (5)(6)(7). Two human IMPDH isozymes exist; type I is constitutively expressed, while type II is expressed in rapidly proliferating cells (8 -11). The IMPDH reaction involves attack of Cys 331 (human type II numbering) at the 2-position of IMP, followed by expulsion of the hydride to NAD (Fig. 1). The resulting covalent E-XMP* intermediate is subsequently hydrolyzed to XMP (12,13).
Mycophenolic acid (MPA) is a potent and specific inhibitor of mammalian IMPDHs, and a MPA derivative, mycophenolate mofetil, is a promising immunosuppressive drug (14,15). MPA affinity varies greatly among microbial and mammalian IM-PDH, for example K i ϭ 22 nM for human IMPDH, 500 nM for the Bacillus subtilis, 20 M for the Escherichia coli and 14 M for the Tritrichomonas foetus enzymes (16 -18). 2 MPA traps the E-XMP intermediate in mammalian IMPDHs (Fig. 1), and the crystal structure of the E-XMP⅐MPA complex of IMPDH from Chinese hamster has recently been solved (12,19,20). MPA stacks against E-XMP in the likely nicotinamide binding site, as predicted by multiple inhibitor studies (17). Of the residues that contact MPA, only Arg 322 and Gln 441 differ in mammalian and microbial IMPDHs; Arg 322 is replaced by Lys, and Gln 441 is replaced by Glu (21)(22)(23). While the mutation of Gln 441 3 Ala decreases MPA sensitivity by 20-fold, activity is also decreased 20-fold (20). Mutations at Arg 322 have not been reported. Moreover, both B. subtilis and E. coli IMPDHs contain Lys 322 and Glu 441 , although the K i values of MPA for these enzymes vary by 40-fold. Thus residues 322 and 441 cannot be the only structural determinants of MPA selectivity.
Random mutagenesis followed by selection for the ability to grow in the presence of MPA can identify mutations in IMPDH that confer MPA resistance. Selection for MPA resistance has previously been reported in both mammalian and parasite systems. In most cases MPA resistance resulted from increased expression of IMPDH, usually via gene amplification (24 -26). Mutant IMPDHs with altered sensitivity to MPA have been reported in murine lymphoma and leukemia cells, although identity of these mutations and their effect on enzyme activity were not characterized (27,28). A MPA-resistant neuroblastoma cell line has been isolated in which a mutant IMPDH is overexpressed by gene amplification (29). While this mutant IMPDH is less sensitive to MPA, it is also much less active than wild type. This MPA-resistant IMPDH contains two mutations: Thr 333 3 Ile and Ser 351 3 Tyr. Recent mutagenesis experi-ments suggest that resistance results from the alteration of Thr 333 (20). This residue is strictly conserved in all IMPDHs sequenced to date, and therefore cannot be a determinant of species selectivity. In addition to these examples, MPA resistance can also result from alterations in purine salvage pathways or in enzymes that utilize guanine nucleotides and from inactivation of MPA via glucuronidation (30 -32).
We have isolated and characterized MPA-resistant mutants of human type II IMPDH in order to explore the structural basis of the species selectivity of MPA. We have identified three mutant enzymes that have a decreased sensitivity to MPA. These enzymes are 3-8-fold less sensitive to MPA. Four substitutions are identified in regions of the protein that are conserved in mammalian IMPDHs, but different in microbial enzymes. Thus these residues may be structural determinants of MPA sensitivity. Interestingly, these substitutions are in residues that do not contact MPA.

MATERIALS AND METHODS
IMP, NAD, and mycophenolic acid were purchased from Sigma. Plasmid pHIMP containing human IMPDH type II cDNA was the generous gift of Dr. Frank Collart (8).
Construction of an Expression System for Human IMPDH Type II-The 1.5-kb NcoI/SalI fragment containing the IMPDH coding sequences from a partial digest of pHIMP was ligated to the NcoI/SalI fragment of the pBluescript II KS ϩ -based vector pST (33) (pBluescript II KS ϩ from Stratagene). This construct (pSI) was used as a template for Kunkel mutagenesis (34). Four silent mutations were inserted into the IMPDH coding sequences, creating four unique restriction sites (pSI5). The This construct was sequenced in its entirety to confirm that only the desired mutations were introduced. One difference from the published cDNA sequence was noted (8); we found that the sequence of bases 608 -611 is GCA GGC, not CGC AGC as reported previously. As a consequence, residues 190 and 191 are Ala and Gly, respectively, rather than Arg and Ser. This sequence was also found in pHIMP, indicating that this discrepancy did not arise during the construction and mutagenesis of pSI. Our results agree with the cDNA sequence reported by Natsumeda in this region of human IMPDH type II (9) and with sequences of the IMPDH type II gene (35,36). The 1.5-kb NdeI (filled in by Klenow extension)/HindIII fragment of pSI5 containing the IMPDH coding sequences was ligated to the NcoI (treated with S1 nuclease to create blunt ends)/HindIII fragment of pKK233-2 to create pHIA5 (37). This construct can complement the guaB deficiency of H712 cells. The junction of the NcoI-NdeI sites contains the sequence (ribosome binding site is underlined) AGGAAA-CAGAC-A(1)TG, which indicates that a T normally found at Ϫ1 was deleted during subcloning. Similar expression systems for human type II IMPDH have been reported previously (18,38,39).
Isolation of MPA-resistant Clones-A library of randomly mutagenized pHIA5 was created by transforming pHIA5 into the mutD5 E. coli strain NR9072 (40). The plasmid was reisolated from a culture grown on LB broth containing 100 g/ml ampicillin. This mutagenized plasmid was used to transform E. coli strain H712, which carries a partial deletion in the guaB gene (41). 3 MPA-resistant colonies were selected by growth on minimal media containing 200 M mycophenolic acid, 9.6 g/ml tryptophan, 48 g/ml histidine, 5.0 g/ml tyrosine, 0.1 g/ml thymine, 2% glucose, and 25 g/ml ampicillin. Plasmid was isolated from these colonies and used to transform H712 cells. The new transformants were tested for the ability to grow on minimal medium in the presence of MPA as described above.
DNA Sequencing-DNA sequencing was performed using either 35 S-dATP with a Sequenase kit (U. S. Biochemical Corp.) or a PRISM Dyedeoxy Terminator Cycle Sequencing kit (Applied Biosystems, Inc.) and an Applied Biosystems 373A DNA sequencer at the Brandeis Sequencing Facility.
IMPDH Purification-Wild type enzyme was purified in two steps using affinity chromatography. H712 cells carrying pHIA5 were grown overnight in LB broth containing 1 mM IPTG and 100 g/ml ampicillin. The cells were harvested by centrifugation and resuspended in 50 mM Tris, pH 7.5, 1 mM dithiothreitol, 10% glycerol (Buffer A). The cells were disrupted by sonication and cell debris removed by centrifugation at 12,000 ϫ g for 20 min. The crude lysate was applied to a Cibacron Blue Sepharose column pre-equilibrated in Buffer A. IMPDH was eluted in a linear gradient of 0 -1 M KCl in Buffer A. The IMPDH containing fractions were pooled, diluted 4-fold in Buffer A, and applied to an IMP affinity column (42). IMPDH was eluted in Buffer A containing 0.5 mM IMP. The preparation is homogeneous as judged by SDS-polyacrylamide gel electrophoresis analysis. IMPDH was concentrated using Centriprep filters (Amicon). IMP was removed by dialysis against buffer A as needed. Protein was quantified using the Bio-Rad assay with IgG as a standard. This assay overestimates the IMPDH concentration by a factor of 2.6 (7), and measurements were adjusted accordingly. The concentration of IMPDH determined by this method agreed with that determined in the MPA inhibition experiments.
Purification of Mutant IMPDHs-F456S/D470G, D255G, M482I and G340E were purified as described above. A462T did not bind to the IMP affinity column. Therefore crude lysate of A462T was chromatographed on a Bio-Gel A5m column. A462T containing fractions were applied to a Cibacron Blue Sepharose column and eluted as above. This preparation was homogeneous as determined by SDS-polyacrylamide gel electrophoresis. L30F/Q277R, L30F, and Q277R are unstable in the absence of IMP. Therefore fractions from the Cibacron Blue Sepharose column were collected in tubes containing IMP such that the final IMP concentration was 0.5 mM. Activity was lost upon further purification of these enzymes. No NADH oxidase or phosphatase activity was detected in these preparations, which suggests that the partially purified enzymes are suitable for kinetic studies. It should be noted that the specific activity of wild type IMPDH does not increase after the Blue Sepharose column, which indicates that this single purification step yields Ͼ 80% homogeneous enzyme preparations. The concentrations of L30F/ Q277R, L30F, and Q277R were determined in the MPA inhibition experiment. Q277R/A462T is also unstable in the absence of IMP. The Cibacron Blue Sepharose fractions of Q277R/A462T were further purified on a POROS CM cation exchange column using a BioCAD Sprint perfusion chromatography system (PerSeptive Biosystems). The column was equilibrated in 7 mM Hepes, 7 mM Mes, 7 mM acetate buffer, pH 6.0 (Buffer B). Enzyme was eluted in a linear gradient of 0 -1 M NaCl in Buffer B into test tubes containing IMP.
Enzyme Assays-The standard assay solution contains 50 mM Tris, pH 8.0, 100 mM KCl, 1 mM dithiothreitol, 3 mM EDTA, 125 M IMP, and 100 M NAD. The production of NADH is monitored by the change in absorbance at 340 nm on a Hitachi U-2000 spectrophotometer. For K m determinations, IMP and NAD are varied as appropriate. Enzymes were dialyzed to remove IMP prior to K m determinations. MPA affinity was determined by varying NAD and MPA concentration in the presence of 1 mM IMP. All assays were performed at 25°C.
Data Analysis-Michaelis-Menten parameters were determined by fitting initial rate data to a sequential mechanism (Equation 1) using v is the initial velocity, V m is the maximal velocity, A is the concentration of IMP, B is the concentration of NAD, K ia is the dissociation constant for IMP from the binary EA complex, K b is the K m for NAD, and K a is the K m for IMP. MPA affinity was determined by fitting initial rate data to Equation 2, which describes a tight binding uncompetitive inhibitor (43,44).
I is the concentration of MPA, v 0 is the initial velocity in the absence of I, E is the concentration of IMPDH active sites, and K i is dissociation constant for MPA.

RESULTS
Expression of Human IMPDH Type II in E. coli-The cDNA (1.5 kb) of human IMPDH type II was modified by silent mutations to insert restriction sites at the following bases: Ϫ1 (NdeI), 473 (XhoI), 750 (KpnI), and 1232 (SacII). The modified cDNA was cloned into the E. coli expression vector pKK233-2 as described under "Materials and Methods" (37). The NdeI site was lost in the cloning process. This construct, pHIA5, expresses human IMPDH type II under the control of the trc promoter. pHIA5 was transformed into E. coli strain H712, which contains a partial deletion in guaB, the gene encoding IMPDH (41). H712 cells carrying pHIA5 can grow on minimal medium in the absence of guanine, which indicates that IM-PDH is expressed in the absence of IPTG. These cells cannot grow on minimal medium in the presence of 200 M MPA (data not shown).
Isolation of MPA-resistant Clones-A library of randomly mutagenized IMPDH was generated using a mutD5 strain of E. coli. This library was transformed into E. coli strain H712. MPA-resistant clones were selected by growth on minimal medium containing 200 M MPA. Thirty clones were selected from 10 6 colonies screened in two separate experiments. Plasmids were isolated from these clones and used to transform H712 cells. All of the retransformed clones could grow on minimal medium in presence of 200 M MPA, which indicates that MPA resistance is contained on the plasmid. Twenty-three of these clones were chosen for further characterization.
Screen of IMPDH Activity and MPA Resistance-MPA resistance can result either from a decrease in the affinity of IMPDH for MPA or from an increase in IMPDH levels. Therefore, crude extracts were prepared from the MPA-resistant clones in order to screen for clones that might contain altered IMPDHs. Crude extracts were prepared from cultures grown in LB broth in the presence of IPTG. These growth conditions were chosen in order to maximize the amount of IMPDH in the crude extracts. IMPDH activity was assayed in the presence of 125 M IMP and 200 M NAD. These substrate concentrations are saturating for wild type IMPDH. The relative specific activities of these preparations are shown in Table I. Six of the crude extracts had relative specific activities within 0.5-1.5-fold of wild type, 14 had specific activities greater than 1.5-fold, and three had specific activities less than 0.5-fold of wild type. The IC 50 for MPA inhibition was also measured under these conditions (Table I). Six clones were identified with relative IC 50 values of 4 or more: MA5, MA9, MA14, MA17, MA18, and MPA11. All of these mutants have specific activities less than 1.6-fold of wild type, which suggests that MPA resistance may result from alteration of IMPDH rather than an increase in IMPDH concentrations.
Sequencing of MPA-resistant IMPDHs-Plasmids isolated from the MPA-resistant colonies were sequenced in order to identify mutations in IMPDH. Of the six clones identified above, four express mutant IMPDHs. MA5 and MA18 express an IMPDH where Ala 462 is changed to Thr (henceforth this mutant IMPDH is denoted A462T). MA17 expresses an IMPDH with two mutations; Leu 30 is changed to Phe and Gln 277 is changed to Arg (L30F/Q277R). MPA11 also expresses an IM-PDH with two mutations; Phe 456 is changed to Ser and Asp 470 is changed to Gly (F456S/D270G). A mutation is found at Ϫ6 in the IMPDH coding sequence of MA14. This mutation could confer MPA resistance by increasing the expression of IMPDH. No mutation is present in the IMPDH coding sequences in MA14, which suggests that a mutation in another region of the plasmid must be responsible for MPA resistance.
The remaining 17 clones were also sequenced. Overall, 17 mutations were identified in the IMPDH coding and adjacent sequences of the 23 plasmids (Tables I and II). Only two of the mutations were transversions, which is consistent with previous observations of mutD5-mediated mutagenesis in rich media (40). Five clones contained substitutions at Ϫ6, which could Characterization of Mutant IMPDHs-The mutant proteins were purified as described under "Materials and Methods." A462T did not bind to an IMP affinity column, which indicates a change in IMP binding. L30F/Q277R is unstable in the absence of IMP. The remaining mutant IMPDHs have purification characteristics similar to wild type.
The K i values for MPA inhibition of wild type and mutant IMPDHs are summarized in Table III. MPA is a tight binding inhibitor of IMPDH under the conditions in these assays. In all cases, the data best fit an uncompetitive inhibition mechanism versus NAD. However, while a competitive mechanism can be definitively eliminated, the fit to a noncompetitive mechanism is only marginally worse than the uncompetitive fit. The parameters for human IMPDH type II are similar to those reported elsewhere (18,39). The K i for MPA inhibition is increased relative to wild type in the three mutant enzymes originally identified in the crude extract screen. The K i of MPA is increased 3-fold in A462T. This mutation was isolated five times; although these isolations do not necessarily represent independent mutational events, this observation suggests that MPA resistance can result from this modest decrease in MPA affinity. A 3-fold increase in K i is also observed for F456S/ D470S. The K i of MPA is increased 8-fold in L30F/Q277R. No increase in the K i for MPA is observed in the other mutant IMPDHs, in agreement with the crude extract experiments.
Alterations are also observed in the steady state kinetic parameters of the MPA-resistant mutants (Table III). In A462T, K m for IMP is increased by 2.5-fold relative to wild type, and k cat is increased 1.5-fold, while no significant change is observed in the K m for NAD. In L30F/Q277R, K m for NAD is increased by 4.5-fold relative to wild type, and k cat is decreased 5.5-fold. Unfortunately, the K m for IMP could not be measured because this enzyme is unstable in the absence of IMP. In F456S/D470S, neither the K m for IMP nor the K m for NAD are significantly changed, although k cat is decreased 2.4-fold. The K m values and k cat values of D255G, M482I, and G340E are indistinguishable from wild type.
Characterization of Single Mutants L30F and Q277R-L30F/Q277R is the only mutant with a dramatic increase in the K i of MPA. Therefore the single mutants L30F and Q277R were constructed, purified, and characterized in order to determine if both mutations are necessary for MPA resistance. The K i for MPA inhibition of L30F is similar to wild type, while the K i for MPA inhibition of Q277R is 6-fold greater than wild type. Thus the Gln 277 3 Arg mutation is sufficient to confer MPA resistance. The K m for NAD of L30F is also similar to wild type, although k cat is decreased by 5-fold. The K m values of both IMP and NAD are increased 5-6-fold in Q277R, while k cat is decreased by 3-fold.
Characterization of Q277R/A462T-A double mutant was constructed in order to determine if the two mutations that confer MPA resistance, Gln 277 3 Arg and Ala 462 3 Thr, function independently. The K i for MPA inhibition of Q277R/A462T is similar to Q277R; thus the effects of the mutations are not additive and the mutations are not independent. Interestingly, the K m for IMP of Q277R/A462T is similar to Q277R; however, the K m for NAD is increased 3-fold and k cat is increased 4-fold. Thus the Ala 462 3 Thr mutation restores the defect in k cat caused by the Gln 277 3 Arg mutation. Q277R/A462T is the first mutant IMPDH with wild type levels of activity and decreased sensitivity to MPA.

Isolation of MPA-resistant Mutants of Human IMPDH-We
have used a random mutagenesis and selection approach to identify mutations in human IMPDH type II that can confer MPA resistance. Human IMPDH type II was expressed in an E. coli strain containing a deletion in guaB. This strain cannot grow in the presence of MPA and was used to select MPAresistant clones. Unlike previous approaches, this method can screen large numbers of clones and other mutations in the purine biosynthetic pathways can be readily eliminated. Twelve MPA-resistant clones were identified, which contained mutations in IMPDH. Three mutant IMPDHs were isolated with decreased K i values for MPA: L30F/Q277R, A462T, and F456S/D470G. The greatest increase in the K i for MPA (8-fold) is observed in L30F/Q277R. This increase can be attributed to the substitution of Arg for Gln 277 . A 3-fold increase in the K i for MPA is observed in A462T. A 3-fold increase in MPA K i is also observed in F456S/D470G. Since this increase is modest, no attempt has been made to determine if both mutations are   (12,20). Therefore MPA resistance can result from a change in the MPA/nicotinamide site and might be manifest in the K m for NAD. Since MPA binding involves a stacking interaction with E-XMP*, changes in the IMP site can also increase the K i for MPA. In addition, the K i for MPA will be increased if E-XMP* no longer accumulates during the IM-PDH reaction. For example, a mutation that changed the rate of a conformational change could change the accumulation of E-XMP*. These mechanisms need not be mutually exclusive; indeed, while the K m values of both NAD and IMP are higher in IMPDH from T. foetus and E. coli than in the human enzyme, the k cat values are also higher in the microbial enzymes, which suggests that both mechanisms are important. A462T, Q277R, and Q277R/A462T display increases in both the K m values for IMP and the K m values for NAD, while the substrate K m values of F456S/D270G are similar to wild type. The k cat for A462T is 1.5-fold higher than wild type, and Q277R/A462T is similar to wild type, while the other mutations decrease k cat . These results suggest that the mutations may both change the MPA binding site and the accumulation of E-XMP*.
Structural Context of Mutations That Confer MPA Resistance-Unfortunately, the coordinates of the structure of the E-XMP⅐MPA complex are not yet available (20), so it is difficult to evaluate the effects of these mutations on the structure of IMPDH. In addition, while the residues that contact MPA and the E-XMP intermediate appear to be clearly delineated, the residues that contact NAD have yet to be identified. These residues are also expected to influence MPA sensitivity. IM-PDH is an ␣/␤ barrel; the active site is located in loops on the C-terminal ends of the ␤ strands. The active site Cys 331 is found in the loop between ␤6 and ␣6 (residues 325-342), and additional active site residues are in the loops between ␤4 and ␣4 (residues 275-280) and ␤8 and ␣8 (residues 400 -450). The large loop after ␤8 forms a flap over the active site; the flap residues interact with both E-XMP and MPA.
Gln 277 is in the loop between ␤4 and ␣4. The adjacent residues, Asp 274 , Ser 275 , and Ser 276 , contact MPA in the structure. Thus, while Gln 277 does not contact MPA directly, substitutions at position 277 can easily affect the residues that do contact MPA. While residues 274 -276 are conserved among IMPDHs, 277 is His and Asp in the IMPDHs from E. coli and T. foetus, respectively (21, 23). Thus residue 277 may be a determinant of species selectivity. In addition, residues 279 -281 are involved in intersubunit contacts in the IMPDH tet-ramer. Therefore, substitutions at 277 may also affect the function of the adjacent subunit. Ala 462 is located in the middle of ␣8 and would appear to be removed from the active site. Phe 456 is also located in ␣8, at the beginning of the helix. Interestingly, residue 462 is the n ϩ 6 residue from 456, and thus would be on the same side of the helix. Asp 470 is the first residue in the loop at the end of ␣8, and would therefore be on the opposite end of the barrel from the active site, although on the same side of the helix as residues 456 and 462. This cluster of mutations suggests an important role for ␣8 in the IMPDH reaction. It is possible that ␣8 may influence the flap, and thus modulate the accumulation of E-XMP* and MPA affinity. Helix ␣8 is highly conserved in mammalian IMPDH, but varies widely in IMPDHs from other sources (Fig. 2). Therefore, ␣8 may be a structural determinant of MPA selectivity.