Mutations that perturb cyclophilin A ligand binding pocket confer cyclosporin A resistance in Saccharomyces cerevisiae.

In complex with the peptidyl-prolyl isomerase cyclophilin A, the immunosuppressive antifungal drug cyclosporin A (CsA) inhibits a Ca/calmodulin-dependent protein phosphatase, calcineurin, which regulates signal transduction. We isolated and characterized cyclophilin A mutations that confer CsA resistance in a Saccharomyces cerevisiae strain whose growth is CsA-sensitive. Three mutations (G70S, H90Y, and G102A) alter single amino acids conserved between yeast and human cyclophilin A, which structural analyses implicate in CsA binding to human cyclophilin A. By Western analysis, all three mutant proteins are expressed in yeast. In vitro, two purified mutant cyclophilins (G70S, G102A) retain prolyl isomerase activity and have moderately reduced affinity for CsA and calcineurin but, when bound to CsA, do bind and inhibit calcineurin phosphatase activity. In contrast, the purified H90Y mutant cyclophilin is dramatically decreased in prolyl isomerase activity, CsA affinity, and calcineurin binding and inhibition. These studies identify conserved cyclophilin A residues that participate in CsA binding and catalysis.

In complex with the peptidyl-prolyl isomerase cyclophilin A, the immunosuppressive antifungal drug cyclosporin A (CsA) inhibits a Ca 2؉ /calmodulin-dependent protein phosphatase, calcineurin, which regulates signal transduction. We isolated and characterized cyclophilin A mutations that confer CsA resistance in a Saccharomyces cerevisiae strain whose growth is CsAsensitive. Three mutations (G70S, H90Y, and G102A) alter single amino acids conserved between yeast and human cyclophilin A, which structural analyses implicate in CsA binding to human cyclophilin A. By Western analysis, all three mutant proteins are expressed in yeast. In vitro, two purified mutant cyclophilins (G70S, G102A) retain prolyl isomerase activity and have moderately reduced affinity for CsA and calcineurin but, when bound to CsA, do bind and inhibit calcineurin phosphatase activity. In contrast, the purified H90Y mutant cyclophilin is dramatically decreased in prolyl isomerase activity, CsA affinity, and calcineurin binding and inhibition. These studies identify conserved cyclophilin A residues that participate in CsA binding and catalysis.
The natural product cyclosporin A (CsA) 1 is a potent antifungal and immunosuppressive compound via its ability to inhibit signal transduction (for review, see Refs. [1][2][3]. CsA is a hydrophobic cyclic peptide that binds to a family of intracellular proteins, the cyclophilins, which are abundant, ubiquitous, highly conserved, and found in multiple forms in different intracellular compartments. In addition, cyclophilins are enzymes that catalyze protein folding by cis-trans isomerization of peptidyl-prolyl bonds. Cyclophilin A, an 18-kDa cytoplasmic protein, mediates CsA actions in yeast (4 -6). The immunosuppressive effects of CsA are also presumed to be mediated by cyclophilin A (7). The target of the cyclophilin A⅐CsA complex is calcineurin (8,9), a calcium/calmodulin-dependent serine-threonine-specific protein phosphatase highly conserved from yeast to man (10 -12). Calcineurin is also the target of the macrolide FK506 bound to a different prolyl isomerase, FKBP12 (13). Calcineurin regulates nuclear import of the nuclear factor of activated T-cells transcription factor during T-cell activation (14 -16), participates in recovery of yeast cells from pheromoneinduced cell cycle arrest (4,10,12), and is essential in CsAsensitive yeast strains (5,17).
A wealth of structural information is available for cyclophilins, both free and in complex with CsA or substrate. The x-ray crystal structures have been solved for cyclophilin A alone (18,19) and in complex with CsA (20,21), a tetrapeptide model substrate or a dipeptide (Ala-Pro) (22)(23)(24), and a CsA analog (25). Similarly, NMR structures are available for the cyclophilin A-bound conformation of CsA (26), the human cyclophilin A⅐CsA complex (22,27), and for wild-type and an F112W mutant of Escherichia coli periplasmic cyclophilin (28,29). Finally, the x-ray structures of cyclophilin B-CsA and cyclophilin C⅐CsA have been solved (30,31). These studies reveal that CsA and substrate bind a common hydrophobic pocket and provide a foundation for mutagenic and biochemical studies to analyze cyclophilin functions in vivo.
Previous studies employing genetic and biochemical analyses established the roles of specific cyclophilin residues in CsA binding, active site function, calcineurin inhibition, and in vivo function. E. coli periplasmic cyclophilin is distantly related to human cyclophilin A, sharing only 34% sequence identity and having prolyl isomerase activity relatively resistant to CsA (32,33). In E. coli cyclophilin, a tryptophan residue invariant in other cyclophilins is replaced by phenylalanine. The CsA sensitivity of an F112W mutant E. coli protein was increased 75-fold, whereas substitution of Trp 121 by phenylalanine in human cyclophilin A decreased CsA affinity 17-fold (34). In the x-ray and NMR structures of cyclophilin A, Trp 121 lies in the middle of the Thr 116 -Gly 130 loop in the CsA binding pocket, near CsA residue 11 and hydrogen-bonded with CsA residue 9 (21,22). A potential role for cysteine residues in catalysis by cyclophilin A was excluded by isolating active mutants lacking each of four different cysteines (32). Based on the x-ray crystal structure of human cyclophilin A, active site residues (Arg 55 , Phe 60 , Phe 113 , and His 126 ) were identified by site-directed mutagenesis (35). Interestingly, cyclophilin mutants lacking peptidyl-prolyl isomerase activity retained the ability to bind CsA and inhibit calcineurin (35). Residues surrounding the cyclophilin A ligand pocket (Arg 69 , Lys 125 , and Arg 148 ) that participate in interactions between cyclophilin A⅐CsA and calcineurin were identified by site-directed mutagenesis (36). Finally, mutations in the Drosophila NinaA cyclophilin homolog (37,38) were found amongst mutations that disrupt targeting of rhodopsin and distort the fly eye (39). This analysis revealed NinaA regions critical for biological function, including residues within and surrounding the ligand binding pocket and hydrophobic membrane anchor (39 -42).
Here we report the isolation of cyclophilin A mutations that confer CsA-resistance in a CsA-sensitive Saccharomyces cerevisiae yeast strain. These mutations identify three residues, Gly 70 , His 90 , and Gly 102 , which participate in CsA binding to yeast cyclophilin A and are identical in human cyclophilin A (Gly 72 , His 92 , and Gly 104 ). In the structure of human cyclophilin A, these residues all lie within the CsA binding pocket, and, based on molecular modeling, substitutions at these residues should perturb the cyclophilin A ligand binding pocket. In addition, modeling suggests that the H92Y mutation would most profoundly perturb the ligand binding pocket, and, in vitro, the H90Y cyclophilin A mutant protein has the most severe defect in CsA binding and prolyl isomerase activity. The isolation of spontaneous cyclophilin A mutations that confer CsA resistance in yeast has provided further insight into cyclophilin A structure and corroborates and extends our view based on structural analyses.

MATERIALS AND METHODS
Strains, Media, and Mutant Isolation-Yeast media was prepared as described previously (43)(44)(45). CsA was obtained from either Sandoz (crystalline) or from the bone marrow transplant unit at Duke University (dissolved at 50 mg/ml in 30% ethanol, 70% cremophor EL).
Yeast strains were isogenic derivatives of the CsA-FK506 sensitive S. cerevisiae strain IL993/5c (5,46)  Spontaneous independent CsA-resistant mutants of strain TB24 were isolated on YPD medium with 100 g/ml CsA. Mutants resistant to CsA but not to FK506 were identified by streaking to YPD medium with 100 g/ml CsA or 1 g/ml FK506. The resulting mutants (MATa ura3 ILV ϩ ) were crossed to strain IL993/5c (MAT␣ URA ϩ ilv5), and prototrophic diploids selected on YNB minimal medium and scored for CsA resistance to determine if mutations were dominant or recessive. Each mutant strain was transformed with the control CEN URA3 plasmid pRS316 (47) and a derivative, pTB4, bearing the yeast CPR1 gene encoding cyclophilin A (5). Transformants selected on synthetic medium lacking uracil were scored on medium containing 100 g/ml CsA to determine if mutations were complemented by the cyclophilin A gene. The o mitochondrial mutation and mutations conferring CsA-FK506 sensitivity prevent sporulation of diploid IL993/5c strains.
Plasmid Constructions, PCR, and Sequence Analysis-Genomic DNA was prepared as described previously (48). Wild-type and mutant alleles of the yeast gene encoding cyclophilin A, CPR1, were PCR amplified with primers JH123, 5Ј-CGCGGGATCCGATGTCCCAAGTCTATTT-TGATG, and JH124, 5Ј-GGCCATGAATTCCAAGCCTGGCAACAT-ACTCCG, which contain a BamHI and an EcoRI site, respectively. PCR reaction conditions were as follows: 5 min at 94°C, 30 cycles of 30 s at 94°C, 30 s at 55°C, and 30 s at 72°C, followed by 5 min at 72°C. Resulting PCR products (ϳ700 bp) were gel purified, cleaved with BamHI and EcoRI, and cloned between BamHI and EcoRI sites of the his6-pTrcHisB expression vector (Invitrogen). The resulting plasmids were introduced into E. coli strain Top 10.
The sequences of the wild-type and CsA mutant alleles of the cyclophilin A CPR1 gene were determined using as templates PCR products amplified from genomic DNA (New England Biolabs, cycle sequencing kit) and denatured plasmid DNA of the cloned genes. Primers were as follows: JH45, 5Ј-GAATTCGGATCCCCGCTAATACTACCATGTCC, and JH46, 5Ј-AGATCTGGATCCTATTGTTCCAGGCAGAGCGG, and internal, JH111, 5Ј-CGGCGGTAAGTCTATCTACGG, and JH112, 5Ј-CTGTCGTGGTGCTTCTTGAAG. In some cases, flanking primers JH123 and JH124 described above were also used. Mutations were identified by comparison with the wild-type sequence, run in parallel, and comparison with the known gene sequence (49).
Cyclophilin A: Overexpression, Purification, and Generation of Antisera-Wild-type yeast cyclophilin A was tagged with a stretch of six histidines at the amino terminus and overexpressed and purified by Ni 2ϩ -nitrilotriacetic acid-agarose affinity chromatography as described previously (50). The His 6 -cyclophilin A protein was further purified by excision and electroelution from KCl-stained SDS-PAGE gels. The resulting protein was emulsified in an equal volume of Freund's complete adjuvant and used to immunize a rabbit that was subsequently boosted with four additional injections in Freund's incomplete adjuvant at 3-week intervals to yield a high affinity rabbit polyclonal antisera against yeast His 6 -cyclophilin A. This antisera specifically detects an ϳ18-kDa protein present in extracts from wild-type yeast strains and absent in extracts lacking cyclophilin A prepared from a cpr1 mutant strain (see "Results" and Fig. 1).
CsA Binding and Peptidyl-prolyl Isomerase Assays-CsA binding to the purified wild-type and CsA-resistant His 6 -cyclophilin A proteins was determined by LH-20 assay using [methyl-butenyl-methyl-threonine-␤-3 H]CsA (Amersham Corp., specific activity ϭ 11.1 Ci/mmol) as described previously (45,51). Cyclophilin A and CsA were preincubated prior to LH-20 assay for 15-60 min to ensure equilibrium binding. cis-trans peptidyl-prolyl isomerase assays were performed by the chymotrypsin-coupled cleavage assay, using the synthetic substrate Ala-Ala-Pro-Phe-p-nitroanilide (Sigma), as described previously (45,52). Release of p-nitroanilide was monitored at 395 nm and 10°C with a Beckman DU640 spectrophotometer coupled with a Peltier temperature controller, an automated sample mover, and a microcomputer and disc reader. Initial reaction rates were determined by assaying prolyl isomerase activity of 20 ng of enzyme in the absence and presence of increasing drug concentrations, from 0 to 1000 nm CsA, added from a 1 mM stock in methanol and preincubated with wild-type or mutant cyclophilin A for 30 -60 min prior to the addition of substrate and chymotrypsin. The concentration of drug necessary for 50% enzyme inhibition or K i was extrapolated from plots of initial velocity rates plotted versus the log of CsA concentration. The results presented were representative of at least three different determinations in which the wild-type and mutant enzymes were assayed in parallel.
Cyclophilin A Affinity Chromatography-For cyclophilin A and cyclophilin A⅐CsA affinity chromatography, the purified wild-type and mutant His 6 -yeast cyclophilin A proteins were coupled to Affi-Gel 10 beads as described previously (50). Preparation of yeast protein extracts from a cyclophilin A-deficient strain (MH250 -2C), cyclophilin affinity chromatography, and Western blot with affinity purified antisera directed against the yeast calcineurin A catalytic subunit CMP1 were as described previously (50).
Calcineurin Phosphatase Assay-Calcineurin phosphatase activity was assayed as described previously (45,53) with a synthetic peptide from the RII subunit of cAMP-dependent protein kinase (DLD-VPIPGRFDRRVSVAAE) phosphorylated with cAMP-dependent protein kinase. Reactions contained 2 M phosphopeptide, 40 nM bovine calcineurin (Sigma), 80 nM bovine calmodulin (Sigma), and, where indicated, wild-type or mutant His 6 -cyclophilin A and CsA in a final reaction of 50 l. Protein concentrations were determined by comparison of Coomassie Blue-stained SDS-PAGE gel bands to standards of known concentration. Release of 32 PO 4 2Ϫ by calcineurin was quantitated by ion-exchange chromatography and scintillation counting.

RESULTS AND DISCUSSION
Isolation of CsA-resistant Yeast Mutants-Previous studies identified an unusual strain of S. cerevisiae, IL993/5c, whose growth is inhibited by CsA (46) and FK506 and in which mutations in cyclophilin A or FKBP12 confer resistance to CsA or to FK506, respectively (5). Our findings established that this strain is CsA-FK506-sensitive because calcineurin is essential for viability (5). These studies provide the foundation for a genetic dissection of the interactions between cyclophilin A and CsA, and of the cyclophilin A⅐CsA complex with calcineurin.
Here we isolated and characterized 200 spontaneous independent CsA-resistant mutants of the CsA-FK506-sensitive strain TB24. 193 mutants were resistant to both CsA and to FK506 and presumably result from mutations that render calcineurin resistant to both drugs or no longer essential. The remaining seven mutants were CsA-resistant but were as FK506-sensitive as the isogenic wild-type parental strain (Table I). When each of these mutant strains was mated to an isogenic CsA-sensitive strain of opposite mating type (strain IL993/5C), CsA-sensitivity was restored in all of the resulting diploid strains, indicating that the mutations are recessive (Table I).
Cyclophilin A Mutations Confer CsA Resistance-As it had been previously established that cyclophilin A mutations confer recessive CsA resistance in yeast (5,46), we tested if any of the seven CsA-resistant mutations were allelic with the cyclophilin A-encoding gene CPR1. For this purpose, each mutant strain was transformed with a centromeric low copy number plasmid expressing the cyclophilin A gene CPR1 (pTB4) or, as a control, lacking the CPR1 gene (pRS316) (5,49). Introduction of the cloned CPR1 gene complemented a cpr1::LEU2 disruption strain (TB40) and restored CsA sensitivity. In each of the CsA-resistant mutant strains, introduction of the cloned CPR1 gene restored CsA sensitivity, whereas introduction of the control plasmid did not. Thus, the mutations that confer CsA resistance are all alleles of the cyclophilin A-encoding gene CPR1 (Table I).
To determine the molecular nature of the cyclophilin A mutations that confer CsA resistance, the CPR1 gene was retrieved from wild-type and mutant genomic DNA by PCR with oligonucleotides flanking the CPR1 gene coding region. DNA sequence analysis of the resulting PCR products revealed that four of the mutations (7, 46, 141, and 47) are single-nucleotide changes that result in amino acid substitutions, G70S, G102A, H90Y, and A10P (Table I). In one additional mutation (183), a single nucleotide substitution replaces His 90 with arginine, but a single-base deletion in the adjacent codon renders the gene out of frame (H90R⌬). In two cases (48 and 50), no changes were identified in the coding portion of the cyclophilin A gene, even though these are clearly cyclophilin A mutations based on failure to complement a cyclophilin A mutation and complementation by the cloned cyclophilin A gene. Because only the coding region of the gene was amplified and sequenced, these two mutants may harbor promoter mutations. In accordance with this interpretation, cells bearing these mutations do not express any cyclophilin A protein (Table I and described below).
As a first step to determine how these cyclophilin A mutations confer CsA resistance, extracts from wild-type and mutant strains were subjected to Western analysis with antisera specific for yeast cyclophilin A. As shown in Fig. 1, this antisera detects wild-type cyclophilin A, an 18-kDa protein, expressed in the wild-type CsA-sensitive strain (Fig. 1, lane 1), and also in a mutant resistant to both CsA and to FK506 (Fig. 1, lane 2) and a CsA-resistant calcineurin mutant (Fig. 1, lane 3), which both express wild-type cyclophilin A. This protein is not present in extract from a cyclophilin A deletion strain (⌬cpr1, Fig. 1, lane  4). In extracts from the mutant strains, three express wild-type levels of cyclophilin A (G70S, H90Y, and A10P) (Fig. 1, lanes 5,  6, and 7), one expresses about one-fourth the wild-type level (H90Y) (Fig. 1, lane 10), and the remaining three do not express cyclophilin A (Fig. 1, lanes 8, 9, and 11). Thus, cyclophilin A mutations that confer CsA resistance can result from a complete lack of protein expression or expression of normal levels of a mutant protein.
Conserved Residues Are Altered in CsA-resistant Cyclophilin A Mutants-Yeast and human cyclophilin A share 65% identity and many conservative amino acid substitutions. In Fig. 2, the sequences of human and yeast cyclophilin A were aligned, and the single-amino acid substitutions that confer CsA resistance (A10P, G70S, H90Y, and G102A) are indicated. Ala 10 is conservatively replaced by valine in the human cyclophilin A, whereas Gly 70 , His 90 , and Gly 102 are all conserved between yeast and man. Thus, yeast cyclophilin A mutations that confer CsA resistance occur in highly conserved or invariant amino acids.

CsA-resistant Cyclophilin A Mutants Have Reduced
CsA Affinity-To assist with purification, wild-type cyclophilin A and three cyclophilin A mutant proteins that bear substitutions at residues conserved in human cyclophilin A and which are stably expressed in yeast (G70S, H90Y, and G102A) were fused to a stretch of six histidines at their amino termini, over-expressed in bacteria, and purified by Ni 2ϩ affinity chromatography to near homogeneity (see "Materials and Methods").
To assess their ability to bind CsA, the purified cyclophilin A proteins were subjected to LH-20 ligand binding assays (51). As shown in Fig. 3A, wild-type yeast cyclophilin A and the G70S and G102A mutant cyclophilins bound CsA, whereas the affinity of the H90Y mutant protein for CsA was dramatically decreased. When LH-20 assays were performed with 5-fold lower amounts of cyclophilin A proteins, CsA binding to the G102A mutant cyclophilin A was reduced in comparison with

FIG. 2. CsA resistance is conferred by mutations in highly conserved cyclophilin A residues.
Human and yeast cyclophilin A sequences are aligned. Amino acid number designations are from yeast cyclophilin A, which is two amino acids larger than the human, and every 10th amino acid is underlined. In the alignment and consensus, identical residues are capitalized and indicated by double dots. Cyclophilin A mutations that confer CsA resistance in yeast are indicated below the consensus. wild-type cyclophilin A (Fig. 3B). Thus, the apparent affinity for CsA by the LH-20 assay was wild-type, G70S Ͼ G102A Ͼ Ͼ H90Y.
Peptidyl-Prolyl Isomerase Activity of Cyclophilin A Mutant Enzymes-Because both CsA and substrates associate with the same binding pocket on cyclophilin, as a second measure of CsA affinity we assayed cis-trans peptidyl-prolyl isomerase activity of the purified wild-type and G70S, H90Y, and G102A mutant cyclophilin A proteins. As shown in Fig. 4, the G70S and G102A cyclophilin A mutant proteins had readily detectable prolyl isomerase activity that was comparable with that of wild-type cyclophilin A. This suggests that the G70S and G102A mutations do not dramatically alter the cyclophilin A active site. In contrast, the H90Y mutant protein did not exhibit prolyl isomerase activity (Fig. 4), indicating that this mutation may perturb the cyclophilin A active site such that neither CsA (Fig.  3) nor substrate binds with high affinity.
We next assessed the ability of CsA to inhibit prolyl isomerase activity of the wild-type, G70S, and G102A cyclophilin A enzymes. Prolyl isomerase assays were conducted following preincubation of 20 ng of the different enzymes with CsA at a range of concentrations from 0 to 1000 nM, and the initial reaction velocities were calculated and used to determine the CsA concentration resulting in half-maximal reaction velocity (see "Materials and Methods"). By this analysis, the K i for inhibition of wild-type cyclophilin A by CsA was approximately 10 nM, whereas for the G70S and G102A, the K i was increased approximately 10-fold, to 100 nM. Thus, the G70S and G102A mutations result in a moderate reduction in cyclophilin A affinity for CsA. Because the CsA minimum inhibitory concentration in strain TB24 is approximately 25 g/ml and these mutations confer resistance to 100 g/ml CsA or 4-fold higher than the MIC, a small reduction in CsA affinity of cyclophilin A should suffice to render the cell drug-resistant. This appears to be the case for the G70S and G102A mutations.
Binding and Inhibition of Calcineurin by Mutant Cyclophilin A⅐CsA Complexes-To determine if these mutations alter the ability of cyclophilin A to interact with calcineurin, purified wild-type cyclophilin A and the G70S, G102A, and H90Y mutant proteins were coupled to Affi-Gel 10 to produce four different affinity chromatography matrices (see "Materials and Methods"). Equal amounts of cell extract from a cyclophilin A-deficient strain (MH250 -2C, cpr1::LEU2 mutant) were incubated with each affinity matrix and washed, and bound material was eluted, fractionated by SDS-PAGE, and transferred to nitrocellulose. The yeast calcineurin A catalytic subunit CMP1 was then detected by Western analysis with affinity-purified antisera raised against a ␤-galactosidase-CMP1 fusion protein (11,50). Calcineurin associated with the wildtype cyclophilin A⅐CsA complex and with the G70S⅐CsA and G102A⅐CsA mutant complexes; however, the binding to the mutant complexes was approximately 3-fold reduced compared with the wild-type complex (Fig. 5, lanes 2, 4, and 6), suggesting that these mutations have a slight affect on the affinity of the cyclophilin A⅐CsA complex for calcineurin. In contrast, substantially less calcineurin associated with the H90Y cyclophilin A mutant⅐CsA affinity matrix (Fig. 5, lane 8), which is consistent with the pronounced CsA binding defect of the H90Y mutant protein.
Finally, we assessed the ability of the wild-type and mutant cyclophilins to inhibit calcineurin protein phosphatase activity in vitro. For this purpose, we used the assay described by Hubbard and Klee (53). Bovine calcineurin was reconstituted with calmodulin and incubated with the DLDVPIP-GRFDRRVSVAAE phosphopeptide and purified wild-type yeast cyclophilin A, or the G70S, G102A, or H90Y cyclophilin A mutant enzymes, either in the absence or the presence of CsA. As shown in Fig. 6, wild-type yeast cyclophilin A⅐CsA and the G70S⅐CsA and G102A⅐CsA complexes inhibited bovine calcineurin activity in vitro. However, in comparison with wildtype cyclophilin, calcineurin inhibition by the G70S and G102A mutant proteins required higher levels of cyclophilin, which is consistent with their decreased affinity for CsA. In addition, these mutations might also perturb nearby residues, such as Arg 67 (Arg 69 in human cyclophilin A), involved in calcineurin binding. In contrast, the H90Y cyclophilin A mutant protein only weakly inhibited calcineurin activity, which is consistent with its profound CsA binding defect. These observations underscore that, whereas the G70S and G102A mutations subtly alter cyclophilin A properties, the H90Y mutation more dramatically alters the cyclophilin A ligand binding pocket.
Molecular Modeling Based on Human Cyclophilin A-The structures of human cyclophilin A, alone and in complex with either CsA or a peptide substrate, have been determined by x-ray crystallography and NMR (18, 20 -24, 27). Residues that comprise the ligand binding pocket of human cyclophilin A (His 54 , Arg 55 , Gln 63 , Asn 71 , Thr 73 , Asn 102 , and Trp 121 ) are con-served in yeast cyclophilin A (His 52 , Arg 53 , Gln 61 , Asn 69 , Thr 71 , Asn 100 , and Trp 119 ). We therefore employed the human cyclophilin A structures as a guide to understand the mutations that render yeast cyclophilin A CsA-resistant. In Fig. 7, these mutations are indicated at the corresponding positions in the x-ray crystal structure of human cyclophilin A (22). The CsA and substrate binding pocket is indicated by the filled C ␣ -C ␣ bonds.
All four of the mutations alter single amino acids that lie within the CsA binding site on cyclophilin A. Three mutations alter Gly 70 , His 90 , and Gly 102 in yeast cyclophilin A, which correspond to Gly 72 , His 92 , and Gly 104 in human cyclophilin A. By molecular modeling, 2 a G72S substitution is predicted to perturb the conformation of the 70s loop of cyclophilin A in which Asn 71 and Thr 73 bind water molecules that pack against CsA. A G104A substitution would probably alter the conformation of adjacent residues Ala 101 , Asn 102 , Ala 103 , which form part of the hydrophobic pocket (Ala 101 , Ala 103 ) or hydrogen bond the amino nitrogen of CsA residue Abu2 (Asn 102 ). The G72S and G104A substitutions could also alter the conformation of residues required for cyclophilin A⅐CsA binding to calcineurin, given our finding that the yeast G70S and G102A cyclophilin A⅐CsA complexes bound to calcineurin 3-fold less well than wild-type (Figs. 5 and 6). Finally, a H92Y substitution would change the conformation of His 126 and possibly also Trp 121 , which, respectively, hydrogen bond with the carbonyl oxygens of CsA residues MeVal 11 and MeLeu 9 . At present, it is not clear, based on molecular modeling, how an alteration of Val 12 (corresponding to yeast A10P) would alter the CsA-binding pocket. CONCLUSION Our studies identified yeast cyclophilin A mutant proteins with single-amino acid substitutions at residues conserved between yeast and human cyclophilin A. These mutant proteins were cloned, overexpressed, purified, and characterized in vitro. These analyses reveal that the G70S and G102A cyclophilin A mutant proteins have moderately reduced CsA binding affinity, whereas the H90Y mutant has a more profoundly perturbed ligand binding pocket and fails to bind CsA or catalyze prolyl isomerization. Based on the structure of human cyclophilin A, molecular modeling predicts that these mutations would perturb the ligand binding pocket. None of these residues were identified in earlier studies employing site-directed mutagenesis (32, 34 -36), and thus this random collection of mutations has revealed additional residues that partic-2 Jörg Kallen, personal communication.
FIG. 6. Inhibition of calcineurin phosphatase activity by wildtype and mutant yeast cyclophilin A-CsA complexes. Calcineurin phosphatase activity assays were conducted as described under "Materials and Methods" in the absence (Ϫ) or the presence of 20, 50, and 100 nM purified wild-type (WT) or G70S, G102A, and H90Y mutant cyclophilin A. Reactions contained no CsA (Ϫ) or 200 nM CsA (ϩ). Results are expressed as percentage of the control reaction in the absence of cyclophilin A or CsA and are representative of at least two independent determinations.
FIG. 7. Cyclosporin A resistance by alteration of the cyclophilin A drug-binding site. The ␣-carbon tracing of human cyclophilin A, as determined by x-ray cystallography (20), is depicted in stereo. ␣-Carbons of several residues are indicated by single-letter amino acid abbreviation and residue number. Regions of the protein comprising the CsA and substrate binding pocket are indicated as filled bonds. Residues Val 12 , Gly 72 , His 92 , and Gly 104 correspond to residues in yeast cyclophilin A (Ala 10 , Gly 70 , His 90 , and Gly 102 ), which, when mutated, confer CsA resistance. By molecular modeling of human cyclophilin A, G72S, H92Y, and G104A substitutions are expected to disrupt the ligand binding site. The ␣-carbon tracing of human cyclophilin A was modified with permission from Pflugl and co-workers (20). ipate in cyclophilin A-ligand interactions. One mutant of the Drosophila NinaA cyclophilin homolog, G96E, has a substitution at the residue corresponding to Gly 70 in yeast cyclophilin A (39). Thus, two different studies converged to implicate this residue in cyclophilin interactions with CsA and a presumed in vivo substrate. Further studies in yeast should allow a more precise dissection of cyclophilin A interactions with substrates, ligands, and calcineurin.