The functional basis of mycophenolic acid resistance in Candida albicans IMP dehydrogenase.

Candida albicans is an important fungal pathogen of immunocompromised patients. In cell culture, C. albicans is sensitive to mycophenolic acid (MPA) and mizoribine, both natural product inhibitors of IMP dehydrogenase (IMPDH). These drugs have opposing interactions with the enzyme. MPA prevents formation of the closed enzyme conformation by binding to the same site as a mobile flap. In contrast, mizoribine monophosphate, the active metabolite of mizoribine, induces the closed conformation. Here, we report the characterization of IMPDH from wild-type and MPA-resistant strains of C. albicans. The wild-type enzyme displays significant differences from human IMPDHs, suggesting that selective inhibitors that could be novel antifungal agents may be developed. IMPDH from the MPA-resistant strain contains a single substitution (A251T) that is far from the MPA-binding site. The A251T variant was 4-fold less sensitive to MPA as expected. This substitution did not affect the k(cat) value, but did decrease the K(m) values for both substrates, so the mutant enzyme is more catalytically efficient as measured by the value of k(cat)/K(m). These simple criteria suggest that the A251T variant would be the evolutionarily superior enzyme. However, the A251T substitution caused the enzyme to be 40-fold more sensitive to mizoribine monophosphate. This result suggests that A251T stabilizes the closed conformation, and this hypothesis is supported by further inhibitor analysis. Likewise, the MPA-resistant strain was more sensitive to mizoribine in cell culture. These observations illustrate the evolutionary challenge posed by the gauntlet of chemical warfare at the microbial level.

tosine, azoles, and echinocandins are currently used to treat candidiasis, although toxicity, limited availability, and the emergence of resistance create an urgent need for new drugs (3,4). C. albicans is sensitive to mycophenolic acid (MPA) 1 and mizoribine when cultured in vitro (5,6). Both of these compounds inhibit IMP dehydrogenase (IMPDH), so this enzyme presents a novel target for antifungal chemotherapy. In addition, the IMPDH gene IMH3 is a useful genetic marker for C. albicans because multiple copies confer MPA resistance (7). MPA R , an IMH3 allele encoding an MPA-resistant IMPDH, confers drug resistance with a single gene copy (8 -10). A second MPA-resistant allele of C. albicans IMPDH that carries mutations I47V, S102A, and G482D has been reported, although the functional effects of these mutations on enzyme activity have not been characterized (11,12).
IMPDH catalyzes the rate-limiting step in guanine nucleotide metabolism: the oxidation of IMP to XMP with concomitant reduction of NAD ϩ (Fig. 1). IMPDH is a target for the immunosuppressive drugs mycophenolate mofetil (CellCept®) and mizoribine; the active metabolites of these drugs are MPA and mizoribine monophosphate, respectively ( Fig. 1) (13,14). The mechanism of the IMPDH reaction is complex, with important consequences for inhibitor selectivity (Fig. 1). IMPDH undergoes a large conformational change in mid-catalytic cycle that converts the enzyme from a dehydrogenase to a hydrolase (15). In the first half of the reaction, substrates bind, and hydride transfer occurs to form the covalent E⅐XMP* intermediate. NADH departs, and the mobile flap folds into the NADH site. This conformational change positions the conserved Arg-Tyr dyad to activate water, and E-XMP* is hydrolyzed to XMP (16). MPA traps E-XMP* by competing with the flap for the NADH site (18,19). In contrast, mizoribine monophosphate (MZP) binds to the IMP site and induces the flap to close (15). Thus, the equilibrium between open and closed conformations controls drug sensitivity: the open conformation favors MPA binding, whereas the closed conformation favors MZP binding.
To explore the potential of C. albicans IMPDH as a chemotherapeutic target and to understand the mechanism of MPA resistance, we have characterized the wild-type enzyme as well as the enzyme encoded by MPA R . The MPA R enzyme contains a single substitution (A251T) that is outside of the MPA-binding site. Here, we show that this mutation stabilizes the closed conformation, causing MPA resistance, but increasing MZP sensitivity. * This work was supported in part by National Institutes of Health Grants GM54403 (to L. H.) and R01 AI33317 (to N. A.) and by Bundesministerium fü r Bildung und Forschung Grant O1 K18906-0. 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 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY864854.

EXPERIMENTAL PROCEDURES
Materials-IMP, ADP, NADH, MPA, mizoribine, and Tris were purchased from Sigma. NAD ϩ was purchased from Roche Applied Science. Dithiothreitol was purchased from Research Organics, Inc. Glycerin, EDTA, and KCl were purchased from Fisher. Tiazofurin was obtained from NCI. MZP was a gift from Drs. S. Shuto and A. Matsuda (Hokkaido University, Sapporo, Japan).
Strains and Growth Conditions-The C. albicans Ura Ϫ auxotrophic strain CAI4 was kindly provided by Dr. William A. Fonzi (Department of Microbiology and Immunology, Georgetown University Medicine Center, Washington, D. C.). C. albicans cells were propagated in yeast extract/peptone/dextrose at 30°C. The C. albicans strains used in this study are listed in Table I. Escherichia coli strain DH5␣ was used for cloning and plasmid isolation. E. coli strain H712, which lacks endogenous IMPDH, was obtained from the E. coli Genetic Stock Center of Yale University (New Haven, CT). Bacteria were grown at 37°C in LB medium or M9CA medium with suitable supplements.
Isolation and Sequencing of a Mutant IMH3 Gene-MPA-resistant mutants of C. albicans strain CAI4 were obtained by incubating 10 10 electrocompetent cells on uridine-containing synthetic dextrose medium consisting of yeast nitrogen base, uracil-deficient complete supplement mixture (Bio 101, Inc., Vista, CA), 100 g/ml uridine, 2% glucose, and 2% agar with a sub-lethal concentration of MPA (3.5-5 g/ml) for 3 weeks at 30°C. Several colonies appeared that were streaked onto uridine-containing synthetic dextrose medium plus a higher concentration of MPA (10 g/ml). Two strains (EC3 and EC5) that were able to grow on elevated MPA concentrations were isolated. Assuming that a mutant IMH3 allele confers MPA resistance to strains EC3 and EC5, we amplified the entire coding regions of the IMH3 genes plus the flanking regions of both strains with the primer pair IMH3RXF (5Ј-GATTAGAATTCTCTAGATGTTTATGATAC-3Ј, with the introduced restriction sites in boldface) and IDHBXR (5Ј-GTATTGGATC-CTCTAGAACTCAGTATATC-3Ј). Two IMH3 libraries of PCR products derived from the respective strains were generated by cutting the amplicon with EcoRI and BamHI, followed by cloning into pBluescript (Stratagene). The inserts were released using EcoRI and BamHI and used in two separate transformations of C. albicans strain CAI4. Both transformations yielded MPA-resistant C. albicans clones, thereby confirming our hypothesis that a mutant IMH3 allele confers MPA resistance. We re-amplified the IMH3 genes from two MPA-resistant C. albicans clones (EC352 and EC551) using the IMH3RXF/IDHBXR primer pair and cloned the amplicons into pBluescript, yielding pEC352 and pEC551, respectively. We sequenced both plasmid inserts derived from clones EC352 and EC551 and found that they contain the same point mutations. The sequence of the MPA-resistant IMH3 allele MPA R has been deposited in the GenBank TM /EBI Data Bank with the accession number AY864854.
Generation of IMH3 Mutants-A heterozygous imh3/IMH3 mutant was generated using the URA3 flipper method (20) in C. albicans strain CAI4. For this purpose, plasmid pI4F was constructed by amplification of a 2.7-kb DNA fragment containing the IMH3 gene plus the flanking regions using the IMH3RXF/IDHBXR primer pair, followed by insertion of the EcoRI/BamHI-cut amplicon into pBluescript. Subsequently, the SalI and NotI restriction sites of pI4F were removed. The resulting plasmid was used as template for divergent PCR with primers KO-IMHS (5Ј-CAGGGACGACTGGTCGACNRN 2 RN 3 RNCCCGGGAGAA-ATGGGAATGGGTAG-3Ј) and NIMHKO (5Ј-TAATTGCGGCCGCTG-CAGNWN 3 WN 2 SNGCATGCCCAAGGTGTCTCTGGTTC-3Ј), which introduced new SalI and NotI sites (underlined), respectively. (Degenerate regions of the primers were used to tag strains for a different study.) After a NotI/SalI restriction digest, the amplicon was ligated to the NotI/XhoI URA3 flipper cassette of pSFU1 (20), resulting in pSFI-MH3. The gel-purified ApaI/SacI fragment of pSFIMH3 containing the URA3 flipper with the 5Ј-and 3Ј-flanking regions of IMH3 was used for targeted IMH3 gene disruption in strain CAI4. C. albicans strain CAI4 was transformed by electroporation as described previously (7). Transformants were selected on uracil-deficient synthetic dextrose medium (yeast nitrogen base, complete supplement mixture without uracil, 2% glucose, and 2% agar) and subsequently tested for correct insertion of the disruption cassette by Southern analysis of genomic DNA (data not shown).
For this study, we used the heterozygous Ura ϩ imh3/IMH3 mutant Cl1M7 to generate mutants with a single copy of MPA R substituted for IMH3. This was achieved by introducing the MPA R mutation into strain  Cl1M7 via transformation with a PCR fragment amplified from pEC352 using primers IMHSF (5Ј-GCTCCGTCGACTAGATGTTTATGATAC-3Ј) and IDHCLAR (5Ј-GCATGGCATCGATGGAAC-3Ј). MPA-resistant Ura ϩ transformants were selected on uracil-deficient synthetic dextrose medium plus 10 g/ml MPA, and the MPA R mutation was verified by restriction digest (G-to-A transition in MPA R eliminates an Fnu4HI site) and direct sequencing of PCR products. Two transformants (MPAR1 and MPAR2) were chosen for further testing. Susceptibility Testing-The MPA and mizoribine resistance of C. albicans strains was measured quantitatively in a microtiter plate assay as described previously (7). Growth of the strains in synthetic medium in the presence of serial dilutions of the inhibiting substances was determined by A 560 measurements after 48 h using a microplate reader. Experiments were performed in triplicate.
Recombinant Expression of C. albicans IMPDH and MPA R -Both the IMH3 gene (7) and MPA R contain an intron at amino acid 150, which would prevent bacterial expression. Therefore, we isolated poly(A) ϩ RNA from strains CAI4 and EC3 using an Oligotex mRNA mini kit (Qiagen Inc.). Reverse transcription was carried out with Superscript II (Invitrogen) and primer DTNOT (5Ј-AATTCGCGGCCGCT 17 -3Ј). The cDNAs were amplified using the primer pair CA5CDS (5Ј-CGGCG-GAATTCATGGTGTTTGAAACTTCAAAAG-3Ј) and CACDS3 (5Ј-TG-GAACTGCAGTTAGTTGTGTAATCTCTTTTC-3Ј), which span the start and stop codons of the IMPDH genes, respectively. The PCR products were cut with EcoRI and PstI using the restriction sites introduced by the PCR primers and cloned into pUC18, resulting in pUCCDS4 (IMH3) and pUCCDS9 (MPA R ). These inserts were subsequently amplified using primers 5TGCDS (5Ј-P i -TGGTGTTTGAAACT-TCAAAAG-3Ј) and CDS3HIND (5Ј-TGGTCAAGCTTAGTTGTGTA-ATCTCTTTTCATA-3Ј) to introduce a HindIII restriction site for cloning into pTACTAC (21). pTACTAC was digested with NdeI, and the resulting ends were filled in with the Klenow fragment of DNA polymerase I and further digested with HindIII. The blunt end ligation at the NdeI site regenerated the ATG start codon of IMH3 and MPA R , and the 3Ј-ligation regenerated a TAA stop codon. The ligation products were transformed into E. coli strain H712, a guanine auxotroph harboring a guaB Ϫ mutation. Cells were plated on modified M9CA minimal medium (10 g/liter M9 minimal salts (Invitrogen), 17 mM NaCl, 0.1 mM CaCl 2 , 1 mM MgSO 4 , 1 g/liter casamino acids, 0.79 g/liter complete supplement mix, 1 g/liter thiamine, 2 g/liter glucose, 1 mM isopropyl 1-thio-␤-Dgalactopyranoside, and 50 mg/liter ampicillin) to isolate transformants correctly expressing the IMH3 or MPA R plasmid and complementing the guaB Ϫ mutation. For further studies, we used plasmids pTAC1_5 and pTAC4_13, which express wild-type IMH3 and MPA R IMPDHs, respectively. The sequences of both plasmid inserts were determined and proved free of errors.
Protein Purification-E. coli H712 cells harboring either the pTAC1_5 or pTAC4_13 vector were cultured in LB medium containing 100 g/ml ampicillin and 1 mM isopropyl 1-thio-␤-D-galactopyranoside to express wild-type and A251T IMPDHs, respectively. The cells were harvested by centrifugation, and the pellet was resuspended in buffer A (50 mM Tris-HCl (pH 7.4), 1 mM dithiothreitol, and 10% glycerol). All of the following manipulations were performed at 4°C. The cells were disrupted by sonication. Debris was removed by centrifugation, and the supernatant was applied to a Cibacron blue-agarose column (Sigma) previously equilibrated with buffer A. IMPDH was eluted in a linear gradient of 0 -1 M KCl in buffer A. Fractions containing IMPDH activity were pooled, concentrated by ultrafiltration, and dialyzed overnight against buffer A containing 0.5 mM IMP. IMPDH was further purified on an HQ anion exchange column using a BioCAD Sprint perfusion chromatography system (Applied Biosystems, Framingham, MA). Enzyme was eluted in a linear gradient of 0 -1 M NaCl in 50 mM Tris-HCl (pH 8.0) and 1 mM dithiothreitol. The fractions containing IMPDH were pooled, dialyzed overnight against buffer A containing 0.5 mM IMP at 4°C, and then applied to an HS cation exchange column. IMPDH was eluted in a linear gradient of 0 -1 M NaCl in 50 mM Tris-HCl (pH 8.0) and 1 mM dithiothreitol. The protein concentration was measured using a NanoOrange® protein quantitation kit (Molecular Probes, Inc.) with bovine serum albumin as a standard. This protocol produced ϳ12 mg of purified protein from 1 liter of cell culture.
Enzyme Kinetics-Standard IMPDH assays contained 100 mM KCl, 3 mM EDTA, 1 mM dithiothreitol, and 50 mM Tris (pH 8.0) (assay buffer). The production of NADH was monitored spectrophotometrically at 340 nm (⑀ ϭ 6.22 mM Ϫ1 cm Ϫ1 ) using a Hitachi U-2000 spectrophotometer. The concentrations of IMP and NAD ϩ were varied for K m determinations. Initial velocity data were fit to the Michaelis-Menten equation (Equation 1) and the uncompetitive substrate inhibition equation where v is the initial velocity, k cat is the turnover number, K m is the Michaelis constant of IMP or NAD ϩ , and K ii is the substrate inhibition constant.
where v is the initial velocity; V max is the maximum velocity; S is substrate; K m is the Michaelis constant for the substrate; K is and K ii are the slope and intercept inhibition constants, respectively; I is inhibitor; and E is enzyme. The best fits were determined by the relative fit error. MZP is a slow tight-binding inhibitor under standard assay conditions. The K i value was determined from the steady-state inhibition data, and the k on and k off values were determined from progress curves, both analyzed using Dynafit (22). The analysis presumed that MZP is a competitive inhibitor with respect to IMP, as has been observed for IMPDH from other organisms.
Multiple inhibition experiments with tiazofurin and ADP were performed with constant concentrations of IMP (600 M for wild-type IMPDH and 200 M for A251T IMPDH) and NAD ϩ (250 M for wildtype IMPDH and 4000 M for A251T IMPDH) at 25°C in assay buffer. Data were fit to an equation describing multiple inhibition (Equation where I and J are inhibitors; v 0 is the initial velocity in the absence of inhibitors; ␣ is the interaction constant; and K i and K j are the inhibition constants for I and J, respectively.

Isolation of MPA-resistant Strains of C. albicans-Strain
CAI4 is sensitive to MPA, with an MIC 50 of 0.25 g/ml. Electrocompetent C. albicans strain CAI4 was cultured in the presence of sub-lethal concentrations of MPA (3-5 g/ml) as described under "Experimental Procedures." After prolonged incubation, several colonies appeared; cells were streaked onto agar plates with an increased concentration of MPA (10 g/ml); and two strains were isolated (EC3 and EC5). The MIC 50 values for both strains EC3 and EC5 as determined by microdilution assays were ϳ1.7 g/ml MPA (Supplemental Fig. S1).
Identification of the MPA R Mutation-Assuming that a mutant IMH3 allele confers MPA resistance, we amplified the entire coding regions of the IMH3 genes plus the flanking regions of both strains EC3 and EC5. Transformation of strain CAI4 with these amplification products yielded MPA-resistant clones, confirming that drug resistance results from mutation(s) in IMH3. Isolation and sequencing of the IMH3 genes from these strains revealed several differences from the published nucleic acid sequence of the IMH3 gene from strain SS   (7). With one exception, all sequence differences within the coding region were silent. The G-to-A transition at nucleic acid 751 of the IMH3 coding sequence creates an Ala-to-Thr substitution at position 251 of the protein sequence. The mutant IMH3 gene also lacks the 200-bp solo phi element in the 3Јuntranslated region, a long terminal repeat of a putative transposable element of C. albicans. phi was recently described by Goodwin and Poulter (23). We found that phi is present in one IMH3 allele of strains SC5314 and CAI4 (data not shown).
Because neither SC5314 nor CAI4 shows MPA resistance similar to that shown by EC3 or EC5, we concluded that the phi element is not involved in MPA resistance. A single copy of the IMH3 gene carrying the A251T mutation could confer MPA resistance to C. albicans. This allele was subsequently developed as a dominant selection marker gene in Candida (8,10,24). However, the mechanistic basis of the drug resistance of MPA R was unclear because Ala 251 , although well conserved among IMPDHs from different sources, is not located in the MPA-binding site. Characterization of Wild-type and A251T IMPDHs-Both the wild-type and MPA R alleles of C. albicans IMPDH were expressed in E. coli by isopropyl 1-thio-␤-D-galactopyranoside induction using vector pTACTAC. Both constructs complement the guanine auxotrophy of E. coli strain H712, indicating that active IMPDHs were produced, and both enzymes were purified to Ͼ95% homogeneity as described under "Experimental Procedures" (Supplemental Fig. S2).
The wild-type enzyme displayed strong NAD ϩ substrate inhibition, which is generally attributed to formation of a nonproductive E-XMP*⅐NAD complex. Initial velocity data collected at saturating IMP concentrations suggested that the K m value for NAD ϩ is greater than the K ii value, which complicates determination of steady-state parameters. This observation was confirmed when both substrates were varied (Figs. 2  and 3). Therefore, the k cat and K m values should not be considered well determined. Nevertheless, it is apparent that the kinetic parameters for C. albicans IMPDH are significantly different from those for human IMPDH, which is promising for the development of specific inhibitors. The parameters for human IMPDH type II are shown in Table II for reference; the parameters for human IMPDH type I are very similar to those for human IMPDH type II (25,26).
In contrast to the wild-type enzyme, A251T IMPDH did not exhibit NAD ϩ substrate inhibition (Figs. 2 and 3). Although the k cat value appears to be similar to that for the wild-type enzyme, the K m values for both substrates are significantly lower. Surprisingly, high concentrations of IMP inhibited the A251T IMPDH reaction. To our knowledge, this is the first report of IMP substrate inhibition in an IMPDH. This inhibition is competitive versus NAD. Inhibition of C. albicans IMPDHs-We determined the sensitivity of both C. albicans enzymes to well known IMPDH inhibitors. MPA is an uncompetitive inhibitor of IMPDHs from other sources (27); it binds selectively to the E⅐XMP* intermediate in the nicotinamide end of the dinucleotide site (after NADH departs) (Fig. 1) (18, 19). As expected, MPA is also an uncompetitive inhibitor of wild-type C. albicans IMPDH (Fig. 4 and Table III). The K ii value is very similar to that of human IMPDH type II, which suggests that the drug-binding sites are also very similar in the fungal and human enzymes. MPA is also an uncompetitive inhibitor of A251T IMPDH. The K ii value is 4-fold higher than that for the wild-type enzyme, comparable with the increase in MIC 50 observed in the growth of the MPA R strains. These results confirm that drug resistance results from the substitution of Ala 251 with Thr.
We further probed the structure of the dinucleotide site by investigating the inhibition of the C. albicans enzymes by tiazofurin and ADP (see Fig. 1 for tiazofurin structure). Like MPA, tiazofurin binds to the nicotinamide end of the dinucleotide site of IMPDHs from various sources. However, unlike MPA, tiazofurin binds to free enzyme and E⅐IMP as well as E-XMP*; therefore, tiazofurin is usually a noncompetitive inhibitor versus NAD. (The nomenclature of Cleland is used here (28); inhibitors of this type are also referred to as mixed.) Tiazofurin is also a noncompetitive inhibitor of C. albicans IMPDH, although the K is value is much larger than the K ii value. This observation suggests that tiazofurin binds mainly to E-XMP* under these assay conditions. The K ii value for wild-type C. albicans IMPDH is similar to that for the human enzyme, further suggesting that the nicotinamide subsites, which are also the MPA-binding sites, are similar in the two E-XMP* complexes. As observed with MPA, the A251T mutation increased the K ii value for tiazofurin.
In contrast to MPA and tiazofurin, ADP binds to the adenosine end of the dinucleotide site. As observed with other IMP-DHs, ADP is a competitive inhibitor versus NAD ϩ of C. albi-  cans IMPDH (27). The K is value is 4-fold greater than that for the human enzyme. This difference probably reflects differences in the structure of the adenosine end of the dinucleotide site. The A251T mutation had a modest effect on ADP inhibition, decreasing the K is value by a factor of 2.
We also investigated the MZP inhibition of the C. albicans enzymes. Like other IMPDHs, MZP is a slow tight-binding competitive inhibitor versus IMP of wild-type C. albicans IMPDH (Fig. 5 and Table III). MZP is a more potent inhibitor of the fungal enzyme, with the K is value decreased by a factor of 4 relative to that for the human IMPDH (Table III) (26). Surprisingly, the A251T mutation increased the potency of MZP, with the K is value decreased by a factor of 40 relative to that for the wild-type enzyme (Fig. 5 and Table III). In Tritrichomonas foetus IMPDH, MZP induces the closed conformation, where the flap has folded into the dinucleotide site (15). Therefore, one possible explanation for the increase in MZP affinity is that the equilibrium between the open and closed conformations has shifted to favor the closed conformation. Because the flap competes with MPA, such a shift in the conformational equilibrium could also explain MPA resistance.
We performed a multiple inhibitor experiment to measure the equilibrium between the open and closed conformations (16). As noted above, tiazofurin binds to the nicotinamide subsite, whereas ADP binds to the adenosine subsite. However, if the closed conformation is favored, the binding of the first inhibitor will convert the enzyme into the open conformation, so the binding of the second inhibitor will be enhanced. This synergy will be manifest in the value of the interaction constant ␣, which will be Ͻ1. Little synergy was observed when tiazofurin and ADP inhibited wild-type C. albicans IMPDH ( Fig. 6 and Table III). This observation suggests that the wildtype enzyme is predominantly in the open conformation. The A251T mutation decreased the ␣ value by a factor of 2 ( Fig. 6 and Table III), which indicates that the equilibrium between the open and closed conformations has shifted toward the closed conformation.
Mizoribine Susceptibility of C. albicans-Surprisingly, A251T IMPDH was more susceptible to inhibition by MZP as described above. Therefore, we tested whether C. albicans cells harboring the MPA R allele are also more susceptible to mizoribine using microdilution assays. In agreement with earlier studies (6), mizoribine is a very potent inhibitor of C. albicans growth (Ͼ50% inhibition of homozygous IMH3/IMH3 strains such as SC5314 between 1 and 2 g/ml) (data not shown). For comparison, we constructed heterozygous imh3/IMH3 and imh3/MPA R strains and compared their susceptibility to mizoribine. The imh3/IMH3 strain was somewhat more sensitive to MPA than the IMH3/IMH3 strain CAI4, as expected given that less of the IMPDH target will be present in this strain. A single copy of MPA R clearly conferred resistance to MPA as noted previously, but rendered the cells more sensitive to mizoribine (Fig. 7). This growth effect was not as dramatic as the effect on enzyme inhibition (a factor of 3 versus 40, respectively). This diminished effect was expected because drug efficacy will be modulated by substrate and target concentrations as well as the limitations of drug uptake and activation.
Mizoribine sensitivity was decreased by the presence of gua- nine and guanosine (Supplemental Fig. S3), which indicates that IMPDH is the target of mizoribine in vivo. Xanthine did not protect the cells from mizoribine, as expected because Candida does not appear to have the ability to salvage xanthine to bypass a drug block at IMPDH. 2 Moreover, both MPA and mizoribine inhibited Candida growth on blood agar plates, and MPA R also conferred MPA resistance and mizoribine sensitivity under these conditions (Supplemental Fig. S4). These observations further validate IMPDH as an antifungal target. Surprisingly, uridine and inosine (but not cytidine) protected the cells against mizoribine. Uridine and inosine protection likely results from competition for a common transporter (29,30). Although mizoribine transport has not been investigated, similar compounds such as ribavirin utilize adenosine/inosine/ uridine transporters (31). DISCUSSION Resistance to MPA can arise from a change in the topology of the MPA-binding site (which is also the nicotinamide subsite of the NAD ϩ -binding site) or from a shift in the conformational equilibrium to favor the closed conformation (Fig. 1). These two mechanisms are not mutually exclusive and can be difficult to distinguish. Nevertheless, we believe the A251T mutation acts primarily by shifting the equilibrium to the closed conformation. First, the mutation had similar effects on both MPA and tiazofurin inhibition, increasing the K ii values by factors of 3-4. These inhibitors have very different structures (Fig. 1), so it seems unlikely that a structural change would have similar effects on their potency. In contrast, changes in the open/closed conformational equilibrium should have similar effects on all inhibitors that bind to the open conformation. Second, the affinity of MZP increased ϳ40-fold; because MZP induces the closed conformation, this increase in affinity can also be explained by a shift in the equilibrium to favor the closed conformation. Finally, the interaction between tiazofurin and ADP was more synergistic in the A251T enzyme as measured by the 2-fold decrease in the interaction constant ␣. This increase in synergy also indicates that the mutation favors the closed conformation (16).
Most importantly, these data are remarkably self-consistent. The (16). Both of these assumptions have been verified in experiments with IMPDH from Tritrichomonas foetus (16). Although not well determined, the ␣ values are consistent with the above estimates (␣ ϭ E open ϭ ϳ0.7 and ϳ0.3 for the wild-type and A251T enzymes, respectively).
Unfortunately, at present, the structure of C. albicans IM-PDH has not been solved, and the sequence of the flap is too different from that of the T. foetus enzyme to allow reliable modeling of the closed conformation. Therefore, we can only speculate about how the A251T mutation stabilizes the closed conformation. Ala 251 corresponds to Gly 237 in T. foetus IMPDH (see Supplemental Fig. S5 for alignment) (Fig. 8). Gly 237 is on the same segment as Arg 241 , which interacts with NAD in the open conformation and with the flap in the closed conformation. With the exception of the side chain of Arg 241 , this segment has the same structure in both the open and closed conformations. It is possible that a substitution at Gly 237 could propagate to Arg 241 , changing the equilibrium between the open and closed conformations. In addition, Gly 237 is within 5.0 Å of Asp 261 , which also interacts with NAD ϩ in the open conformation and with the flap in the closed conformation. Position 237 can easily accommodate an Ala residue, but Thr creates a potential steric conflict with the ␤-carbon of Asp 261 (Fig.  8). Perhaps the repositioning of Asp 261 in response to the Thr substitution alters the equilibrium between the open and closed conformations. Interestingly, the IMPDH encoded by IMD2 in Saccharomyces cerevisiae also harbors a Ser residue at the equivalent position (Ser 253 , S. cerevisiae numbering). The IMD2 gene is the only member of the IMPDH gene family in S. cerevisiae that confers resistance to MPA (note that IMD1 is a pseudogene) (32,33). Like the wild-type IMH3 gene in C. albicans, the IMPDHs encoded by S. cerevisiae IMD3 and IMD4 contain Ala at this position. Whether Ser 253 in S. cerevisiae Imd2p is indeed responsible for the MPA resistance of this enzyme remains to be elucidated.
Evolution is generally presumed to favor the more catalytically efficient enzyme, yet the wild-type C. albicans enzyme is less efficient than the A251T enzyme as measured by the value of k cat /K m . Perhaps IMPDHs evolve in response to inhibitor 2 S. Bentink and G. A. Köhler, unpublished data. pressure rather than to optimize catalysis (34). Unfortunately, we do not know how C. albicans originated or when it adapted to inhabit warm-blooded animals, so we can only speculate on the selective pressures driving its evolution. Both MPA and mizoribine are natural products, and it is likely that the C. albicans progenitor encountered both compounds. MPA is produced by several Penicillium species and has been identified in diverse environments such as sewage and cheese; it has even been suggested that immunosuppression can result from the MPA derived from these sources (35)(36)(37)(38). Mizoribine is produced by the common soil fungus Eupenicillium brefeldianum (6). The extreme MPA sensitivity of mammalian enzymes suggests that C. albicans is protected from MPA in the gastrointestinal tract, although C. albicans has occasionally been isolated from soil, grass, water, and dairy products, where Penicillium and E. brefeldianum reside. Nevertheless, the opposing effects of the A251T mutation on MPA and mizoribine sensitivity illustrate the evolutionary challenge posed by the gauntlet of microbial chemical warfare.
More antifungal drugs are urgently needed, and our results suggest that IMPDH may be a new antifungal target. C. albicans is highly sensitive to IMPDH inhibitors, which suggests that guanine nucleotides are primarily supplied by de novo biosynthetic pathways. Although MPA resistance can be overcome in vitro by guanine supplementation, guanine levels in blood are too low to support growth, and C. albicans lacks the ability to salvage xanthine. 3 Although MZP is a more potent inhibitor of C. albicans IMPDH compared with the human enzyme, it is not sufficiently selective to be useful in chemotherapy. Nevertheless, MZP (as well as ADP) demonstrates that selective inhibition is possible. Furthermore, alignment of the human and C. albicans sequences indicates that the adenosine subsites are very different. It may be possible to exploit the interactions of the adenosine subsite to develop fungusspecific IMPDH inhibitors.