Kinetically Controlled Drug Resistance

Background: Mycophenolic acid sensitivity varies by 103-fold among IMP dehydrogenases (IMPDHs) from different sources even though the drug-binding site is completely conserved. Results: Resistant IMPDHs have a different kinetic mechanism compared with sensitive enzymes. Conclusion: Resistance results from a failure to accumulate the drug-sensitive intermediate. Significance: This is a novel mechanism of drug resistance. The filamentous fungus Penicillium brevicompactum produces the immunosuppressive drug mycophenolic acid (MPA), which is a potent inhibitor of eukaryotic IMP dehydrogenases (IMPDHs). IMPDH catalyzes the conversion of IMP to XMP via a covalent enzyme intermediate, E-XMP*; MPA inhibits by trapping E-XMP*. P. brevicompactum (Pb) contains two MPA-resistant IMPDHs, PbIMPDH-A and PbIMPDH-B, which are 17- and 103-fold more resistant to MPA than typically observed. Surprisingly, the active sites of these resistant enzymes are essentially identical to those of MPA-sensitive enzymes, so the mechanistic basis of resistance is not apparent. Here, we show that, unlike MPA-sensitive IMPDHs, formation of E-XMP* is rate-limiting for both PbIMPDH-A and PbIMPDH-B. Therefore, MPA resistance derives from the failure to accumulate the drug-sensitive intermediate.


Plasmids-Constructs for the recombinant expression of N-terminally His-tagged AnImdA, PbIMPDH-A, and
PbIMPDH-B in Escherichia coli were constructed as described previously (5). Site-directed mutagenesis of PbIMPDH-B R429A was performed using a method based on QuikChange mutagenesis (Stratagene, La Jolla, CA). (This is residue 440 in PbIMPDH-B numbering.) Briefly, a T7 forward primer and reverse primer (GTC GTT CTC AGA GAA GTA GGC GGA AGC GCC AGC) were used to amplify by PCR a portion of the PbIMPDH-B gene carrying the R429A mutation. This PCR product was used as the forward and reverse mutant primers to complete the remainder of the QuikChange procedure according to the manufacturer's directions. Plasmids carrying the desired R429A mutation were verified by sequencing (GENEWIZ, South Plainfield, NJ).
Expression and Purification of His-tagged Proteins-Enzymes were expressed and purified as described previously (5). Briefly, plasmids were expressed in a ⌬guaB derivative of E. coli BL21(DE3) (12); purified using a HisTrap affinity column (GE Healthcare) on an Ä KTApurifier TM system (GE Healthcare); and dialyzed into buffer containing 50 mM Tris (pH 8.0), 100 mM KCl, 1 mM DTT, and 10% (v/v) glycerol. All enzymes were purified to Ͼ90% purity as determined by SDS-PAGE.
Enzyme Concentration Determination-Enzyme concentration was determined by the Bio-Rad assay according to the manufacturer's instructions using IgG as a standard. The Bio-Rad assay overestimates IMPDH concentration by a factor of 2.6 compared with the concentration determined by active site titration using the irreversible inactivator 5-ethynyl-1-␤-D-ribofuranosylimidazole-4-carboximide 5Ј-monophosphate (13), so a correction factor of 2.6 was used.
Enzyme Kinetics-Standard IMPDH assay buffer consisted of 50 mM Tris (pH 8.0), 100 mM KCl, 1 mM DTT, and various concentrations of IMP and NAD ϩ or APAD ϩ . Enzyme activity was measured by monitoring the absorbance increase at 340 nm corresponding to the production of NADH or at 363 nm corresponding to the production of reduced 3-acetylpyridine adenine dinucleotide (APADH) on a Cary Bio-100 UV-visible spectrophotometer at 25°C. Initial rates were calculated using ⑀ 340 ϭ 6.2 mM Ϫ1 cm Ϫ1 and ⑀ 363 ϭ 9.1 mM Ϫ1 cm Ϫ1 . Initial velocity data were fit to either the Michaelis-Menten equation (Equation 1) or an uncompetitive substrate inhibition equation MPA inhibition assays were performed at saturating IMP and half-saturating NAD ϩ concentrations to avoid complications arising from NAD ϩ substrate inhibition. The concentrations of IMP and NAD ϩ used for each enzyme are listed in the figure and table legends. MPA inhibition data were fit to either an uncompetitive inhibitor equation (Equation 3) or a noncompetitive/mixed inhibitor equation (Equation 4) using Tight-binding inhibitor treatment was used for AnImdA. Data were fit to an uncompetitive tight-binding inhibitor equation

Multiple-inhibitor Experiment-Experiments
where I and J are the inhibitors; v 0 is the uninhibited initial velocity; K i and K j are the inhibition constants for I and J, respectively; and ␣ is the interaction constant between the two inhibitors.
Primary Deuterium Isotope Effects-[2-2 H]IMP was used as the substrate in standard IMPDH assay buffer. Activity was measured by holding IMP at a constant saturating concentration and varying the concentration of NAD ϩ or APAD ϩ .
Solvent Deuterium Isotope Effects-Assay buffer was prepared in either H 2 O or D 2 O. pH meter readings were corrected to pD by adding 0.4 units. Activity was assayed by holding IMP at a constant saturating concentration and varying the concentration of NAD ϩ or APAD ϩ .
Measurement of E-XMP*-Reaction mixtures contained 2 or 3 M enzyme, 0.25 or 1 mM [8-14 C]IMP (for AnImdA and PbIMPDH-A/B, respectively), and 5 mM NAD ϩ in standard IMPDH assay buffer at 25°C. Enzyme, [8-14 C]IMP, and NAD ϩ were mixed to initiate the reaction, and the reaction was quenched during steady state by precipitation with TCA to a final concentration of 10% (w/v). Enzyme samples were collected using HA-nitrocellulose filters (Bio-Rad) and washed with 10% (w/v) TCA. Radioactivity was measured using a scintillation counter. Control reactions in which enzyme was omitted were included for each experiment. We examined the effects of varying substrate concentration on MPA action to determine the mechanism of inhibition for all three fungal enzymes. MPA concentrations approximated enzyme concentration in experiments involving AnImdA, necessitating tight-binding inhibitor treatment (14). The data best fit uncompetitive inhibition for both IMP and NAD ϩ ( Fig.  2  MPA is also an uncompetitive inhibitor of PbIMPDH-A and PbIMPDH-B versus NAD ϩ . However, MPA inhibition fits best to a noncompetitive/mixed pattern of inhibition versus IMP. This change in mechanism indicates that MPA no longer binds exclusively to E-XMP*. Perhaps the accumulation of E-XMP* open is so low that the association of MPA with other enzyme forms becomes significant.

Mechanism of Inhibition
MPA-resistant IMPDHs Have Less E-XMP* total -Whereas most IMPDHs accumulate E-XMP* in the steady state, the above observations suggested that less E-XMP* may accumulate in PbIMPDH-A and PbIMPDH-B. To test this hypothesis, the reactions were performed with radiolabeled IMP. The reactions were acid-quenched, and the protein-associated radioactivity was assessed by filter binding. This experiment measures all E-XMP* complexes (E-XMP*-NADH, E-XMP* open , and  (15,16). We performed a multiple-inhibitor experiment to probe the open/closed equilibrium (6). Tiazofurin binds to the nicotinamide subsite of the NAD ϩbinding pocket, and ADP binds to the adenosine subsite. If the closed conformation is favored, one inhibitor will pull the equilibrium toward the open conformation, resulting in enhanced binding of the second inhibitor. This will be reflected in the interaction constant ␣, which will be Ͻ1 if such synergism is observed. In contrast, if E-XMP* open predominates, the two inhibitors will be independent, and ␣ will be ϳ1.
The value of ␣ is ϳ1 for both AnImdA and PbIMPDH-B (Table 1 and

Hydride Transfer Is Rate-limiting in Reactions of
PbIMPDH-A and PbIMPDH-B-To further assess the accumulation of E-XMP*, we used isotope effects to identify the ratelimiting steps for each enzyme reaction. If hydrolysis is the ratelimiting step as typically seen, a solvent deuterium isotope effect will be observed. However, if hydride transfer is the slow step, there will be a primary deuterium isotope effect when [2-2 H]IMP is used as the substrate.
No primary deuterium isotope effect is observed in the reaction of AnImdA, indicating that hydride transfer is fast, as observed for other IMPDHs (Table 2) (6). A small solvent deuterium isotope effect is observed on k cat , indicating that hydrolysis is partially rate-limiting step, also as expected. Therefore, E-XMP* open is the predominant enzyme form during the reaction of AnImdA, accounting for MPA sensitivity.
In contrast, PbIMPDH-A shows a primary deuterium isotope effect on k cat in addition to a solvent deuterium isotope effect, indicating that both steps are partially rate-limiting. This observation explains the relatively low amount of E-XMP* total that accumulates during the reaction of PbIMPDH-A.
PbIMPDH-B also shows a primary deuterium isotope effect on k cat but no solvent deuterium isotope effect, indicating that hydride transfer is completely rate-limiting in this reaction. E-IMP-NAD ϩ is therefore the predominant enzyme form for PbIMPDH-B. This kinetic mechanism minimizes the accumulation of E-XMP* in the steady state, rendering the enzyme resistant to MPA.
Increasing E-XMP* Makes PbIMPDH-A and PbIMPDH-B More MPA-sensitive-If the MPA resistance of PbIMPDH-A and PbIMPDH-B results from the low accumulation of   The amount of E-XMP* can also be increased with the NAD ϩ analog APAD ϩ . This dinucleotide has a greater reduction potential than NAD ϩ (EЈ 0 ϭ Ϫ0.258 for APAD ϩ Ϫ0.320 for NAD ϩ ) (17) and a lower affinity for IMPDH (6,18). These properties will shift the reaction toward E-XMP* open , Importantly, because the reduced cofactor must dissociate before MPA can bind, the structure of E-XMP* formed in the reaction with APAD ϩ will be identical to the E-XMP* formed in the reaction with NAD ϩ . As predicted, all of the enzymes became more MPA-sensitive by factors of 2-4 (Table 3). When APAD ϩ and D 2 O were combined, the effects were approximately additive for all three enzymes (Table 3).
Mutation R429A Increases MPA Sensitivity of PbIMPDH-B-Arg-429 is believed to act as a general base to activate water in the hydrolysis of E-XMP* (19). The mutation of the analogous Arg-419 to Ala decreases the hydrolysis step by a factor of 500 but has no effect on hydride transfer in Tritrichomonas foetus IMPDH (16). The R429A mutation caused a 1000-fold reduction in the value of k cat for PbIMPDH-B (Table 2). Hydrolysis is now completely rate-limiting as evidenced by a solvent isotope effect of 4.4. A burst of NADH production is observed in the pre-steady state, further confirming that hydrolysis is rate-limiting (supplemental Fig. S2). This mutant is 10-fold more sensitive to MPA than PbIMPDH-B.

DISCUSSION
The MPA resistance properties of PbIMPDH-A and PbIMPDH-B derive from changes in the kinetic mechanism. Unlike all other IMPDHs characterized to date (6,21), hydride transfer is at least partially rate-limiting for both enzymes. This simple kinetic switch minimizes the accumulation of the drugsensitive species E-XMP* open , allowing these enzymes to evade MPA inhibition.
The resistance of PbIMPDH-A can be almost completely reversed by increasing the accumulation of E-XMP*. The remaining 3-fold resistance can be attributed to the equilibrium between the open and closed conformations of E-XMP*, which favors the closed conformation in PbIMPDH-A. This shift in equilibrium may be due to the presence of Ser-249. MPA-sensitive IMPDHs contain an Ala at this position (Fig. 1B). Ser at this position accounts for the MPA resistance of IMD2 in Saccharomyces cerevisiae (20). The substitution of Thr causes a 3-fold increase in MPA resistance in Candida albicans IMPDH (15), shifting the equilibrium toward the closed conformation.
The resistance of PbIMPDH-B is only partially reversed by increasing E-XMP*. Like MPA-sensitive IMPDHs, PbIMPDH-B contains Ala-249. We have recently found that changes in residues 498 -527 account for ϳ7-fold of the 60-fold difference in MPA resistance between PbIMPDH-A and PbIMPDH-B; these residues form part of an allosteric site that binds K ϩ (8). However, many of the structural determinants of the MPA resistance of PbIMPDH-B remain a mystery.