Molecular Analysis of the Interaction of Calcineurin with Drug-Immunophilin Complexes*

The calcidcalmodulin-regulated phosphatase cal- cineurin (CN) is the site of action of the immunosuppressive drugs cyclosporin A (CsA) and FK506. CN has re- cently been established as a key signaling enzyme in the T cell signal transduction cascade and an important regulator of transcription factors such as NF-AT and OAP/Oct-1, which are involved in the expression of a number of important T cell early genes. CsA and FK506 act by forming complexes with their respective intracellular receptors cyclophilin and FKBP (immunophilins), which can then bind to CN, inhibiting its enzymatic activity and thereby preventing early gene expression. CN is comprised of two subunits: a 59-kDa catalytic subunit (CNA), which contains a calmodulin binding domain and autoinhibitory region, and a 19-kDa intrinsic calcium binding regulatory subunit (CNB). In this study, we have utilized a series of deletion mutants of the CNA subunit to investigate the subunit and molecular requirements that govern the interaction of CN with drug-immunophi-lin complexes. The calmodulin binding and autoinhibi- tory domains of the CNA subunit were found to be dispensable for the binding of CN to drug-immunophilin complexes. In contrast, we found that the regulatory CNB subunit appears to play an obligatory role in this interaction and have defined an amino acid sequence of the CNA subunit which forms the binding site for CNB. Although necessary, the CNB subunit per se is not

ceptors, cyclophilin and FKBP , respectively, collectively known as the immunophilins (11). The binding of each drug to its cognate receptor imposes a gain of function on the corresponding immunophilin, resulting in the formation of an inhibitory complex that directly interferes with a key Ca2+-sensitive step in the T cell signal transduction cascade (3, 8,9,[11][12][13], thereby preventing the activation of specific transcription factors (such as NF-AT and OAFVOct-1) involved in lymphokine gene expression (14).
Recent studies have afforded new insights into the mechanism of action of these two important immunosuppressant drugs. First, Schreiber and his colleagues used drug-immunophilin affinity chromatography to identify the major cellular target of the drug-immunophilin complex as the Ca2+/calmodulin-regulated serinehhreonine phosphatase, CN (PP2B) (15,16). Both FKBP.FK506 and cyclophilinCsA complexes were found t o completely inhibit the enzymatic activity of CN toward a well characterized model phosphopeptide substrate in uitro.
Second, in vivo studies demonstrated that overexpression of CNA and CNB subunits was able to render T cells markedly resistant to the inhibitory effects of CsA and FK506 and to augment significantly the activity of calcium-dependent promoters, such as NF-AT, OAP/Oct-1, and interleukin 2 (17)(18)(19). Moreover, a calcium-independent, constitutively active mutant of CN was able to synergize with phorbol ester and activate these latter promoters, replacing the normal requirement for an increase in intracellular calcium (18,19). Taken together, these studies established CN as the i n uiuo target of the immunosuppressive drugs CsA and FK506 and as a key component of the T cell signal transduction cascade and an important downstream effector of the Ca2+ signal and regulator of T cell transcription factor activity.
CN is comprised of two subunits, a calmodulin-binding 59-kDa catalytic subunit (CNA) and a Ca2+-binding 19-kDa regulatory subunit (CNB) (20)(21)(22). In the present study, we have utilized a series of deletion mutants of the CNA subunit to investigate the molecular and subunit requirements that govern the interaction of drug-immunophilin complexes with CN. We have identified an obligate requirement for the CNB subunit in this interaction and have defined the sequences in the CNA subunit which represent the binding site for the CNB subunit. The CNB subunit, although necessary, is not sufficient to mediate binding to drug-immunophilin complexes. Thus, we have additionally identified sequences of the CNA subunit, within the conserved protein phosphatase catalytic domain, which are also essential for the CN/drug-immunophilin interaction.
EXPERIMENTAL PROCEDURES Materials-The murine CNA,, (23) and CNB (24) cDNAs were gifts of Dr. R. L. Kincaid, NIAAA. The plasmid encoding the GST-FKBP-12 (15) was a gift of Dr. S. Schreiber, Department of Chemistry, Harvard University. The plasmid encoding GST-cyclophilin A (25) was obtained from Dr. R. Bram. Yeast strain Y153 and plasmids PAS1 and PACT (26) 26431 were generous gifts of Dr. S. J. Elledge, Baylor College of Medicine. The plasmid containing the Saccharomyces cereuisiae CNB subunit (27) was provided by Dr. M. Cyert, Stanford University. GAL4-E47 and ACT-lyl were gifts from Drs. L. Naumovski and M. Cleary, Stanford, CA. FK506 was a gift from Fujisawa USA Inc., CsA was obtained from Sandoz Pharmaceuticals Corp. The VA1 mAb (28) was kindly provided by Dr. J. Wang, University of Calgary, Canada. All restriction enzymes and DNAmodifying enzymes were purchased from New England Biolabs.
Plasmid Construction-For carboxyl-terminal CN deletion constructs, the influenza virus hemagglutinin epitope (EFYPYDVPDYA) (29) was fused in-frame to the amino terminus of murine CN with the polymerase chain reaction (PCR). All CN carboxyl-terminal deletion constructs, except CA355, were generated by PCR using specific oligonucleotide primers and standard methodologies. wically, reactions were carried out in 10 m M Tris, pH 8.4,50 m M KCl, 1.5 m~ MgCl,, 100 pg/ml gelatin, 200 p dNTPs, a 0.25 p concentration of each primer, and 2.5 units of Taq polymerase (15-20 cycles; 1-min denaturation at 94 "C, 1.5-min anneal at 55 "C, 2-min extension at 72 "C). Amplified DNA fragments were digested with the appropriate restriction endonucleases and introduced into pBJ5-stop using standard molecular cloning techniques. pBJ5-stop is a derivative of pCDL-SRa296 (30) and contains translational termination codons in all three reading frames immediately downstream of the EcoRI site in the polylinker. CA355 was generated by utilizing the PZfmI site in murine CNA,,, which was rendered blunt with T4 DNA polymerase and fused to the EcoRI site of pBJ5-stop that had been blunted with the Klenow fragment of DNA polymerase I. For N1-CA394, NA32, NA99, and NA347-CA394, DNA fragments were amplified by PCR with specific oligonucleotide primers and inserted into pBEXl (25) at the appropriate restriction sites. pBEXl was derived from pBJ5-stop by the insertion of a synthetic oligonucleotide that allows open reading frames to be translationally epitope (EFYF'YDVPDYA). NA347 and NA364 were generated by PCR fused at their carboxyl termini with the influenza virus hemagglutinin using specific oligonucleotide 5' primers and a 3' primer that adds the influenza virus hemagglutinin epitope as a carboxyl-terminal translational fusion to the carboxyl terminus of CN. Amplified fragments were digested with the appropriate restriction endonucleases and introduced into pBJ5-stop. All constructs generated by PCR were sequenced to eliminate potential mutations incorporated during the PCR. Wherever possible the extent of DNA fragments derived by PCR was limited by utilizing existing restriction sites for plasmid reconstructions.
GAL-FKBP was generated by making a translational fusion between the human FKBP-12 open reading frame and the GAL4 DNA binding domain (1-147) contained within plasmid pASl (26). ACT-CNA394 and ACT-CNA332 were generated by in-frame fusion of CA394 and CA332 to the GAL4 transcriptional activation domain contained in PACT (26).
DNA Dansfections and Preparation of Cell Extracts-COS-7 cells grown in RPMI 1640 medium supplemented with 10% (v/v) fetal calf serum (growth medium) were transfected with the indicated plasmids by electroporation using a Bio-Rad Gene Pulser. Briefly, lo7 cells harvested by trypsinization and resuspended in 400 p1 of growth medium were combined with 5 pg of the indicated plasmid DNA in a 0.4-cm gap electroporation cuvette (Bio-Rad). After a 5-min incubation at room temperature an electric field of 230 V, 960 microfarads was applied, cells were allowed to recover for 5 min, and were then replated in prewarmed growth medium at 37 "C. 48 h after transfection, cells were harvested by trypsinization, washed twice with phosphate-buffered saline, resuspended in 200 pl of lysis buffer (25 m~ Hepes, pH 7.2, 100 m~ NaCl, 2 m M MgCl,, 2 m M CaCl,, 0.5% %ton X-100, 1 m M phenylmethylsulfonyl fluoride, 1 pg/ml leupeptin, 1 pg/ml antipain, 1 pg/ml aprotinin, 1 p & l d pepstatin) and incubated on ice for 15 min. Nuclei were removed by centrifugation at 1,000 x g for 5 min 4 "C, and the resulting supernatant was clarified by centrifugation at 13,000 x g for 5 min 4 "C and stored at -70 "C.
Drug-Immunophilin Binding and Immunoprecipitation-Glutathione S-transferase immunophilin fusion proteins were prepared as described previously (25). Preformed immobilized GST-FKBP-12 fusion protein.FK506 or GST-cyclophilin A fusion protein.CsA complexes (10 pg) were mixed with aliquots (50 pl) of extracts derived from transiently transfected COS-7 cells in a total volume of 200 pl of lysis buffer. Where indicated 25 m M EGTA was added. After incubation for 2 h at 4 OC, GST-immunophilin-bound complexes were washed five times with lysis buffer and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by immunoblotting with the 12CA5 mAb (BABCO). Immunoprecipitates were prepared by incubating aliquots (50 pl) of transiently transfected COS-7 cell extracts with preformed 12CA5 mAb.proteinA-Sepharose complexes in a total volume of 200 pl of lysis buffer for 2 h at 4 "C. After extensive washing with ice-cold lysis buffer, immunoprecipitates were analyzed by SDS-PAGE and immunoblotting with the VA1 mAb (28). For analysis of total cell extracts, 8 pl of the transfected cell extract was analyzed by SDS-PAGE and immunoblotting with the 12CA5 mAb. Immunoblotting was performed as described in detail (31), using primary antibodies at a dilution of l/l,OOO ascites fluid. Immunoreactive bands were visualized using horseradish peroxidase-coupled rabbit anti-mouse immunoglobulin (1/1,000; Zymed) and the ECL system (Amersham Corp.) according to the directions provided by the manufacturer.
Yeast Strains, Manipulations, and lho-hybrid Screening"Y153 (26), MATa leu2-3,112, ura3-52, trpl-901, his3-A200, ade2-101, ga14A, gal8OA, URA3::GAL-lacZ, LYS2::GAL-HISS. To generate Y153ACNB, a CNB disruption vector (cnbA:ADEZ) was created by cloning a 4.7-kilobase genomic BamHI fragment containing the ADE-2 gene into BgZIIdigested BLINKS, encoding the S. cereuisiae CNB gene (27). The entire 5.2-kilobase disruption fragment was liberated with BamHIIPstI, and gene disruption was carried out by standard yeast genetics. Correct targeting of the CNB locus was confirmed by Southern blot analysis and pheromone halo assay (27). Yeast transformations were performed by the lithium acetate procedure (32). Yeast two-hybrid screening was performed essentially as described previously (26), except that 50 m~ 3-aminotriazole (Sigma) was included in the plating media. Where indicated, FK506 was added to SC plates at a final concentration of 200 ng/ml. Transformed colonies were selected by growth at 30 "C on SC plates lacking the appropriate nutrients; interaction in the two-hybrid system was observed by growth in the absence of histidine and confirmed by p-galactosidase assay (26).

RESULTS
To investigate the molecular requirements for the interaction of CN with drug-immunophilin complexes, we initially generated a series of progressive carboxyl-terminal deletions of the CNA subunit (CNA,, isoform (23)) (see Fig. 7) and tested each of them for its ability to interact with either FKBP.FK506 or cyclophilin A-CsAin an in vitro binding assay. The recombinant CN molecules used for these studies were generated by transiently transfecting each of the CNA deletion mutants, together with a n expression vector encoding the CNB subunit, into COS-7 cells. To facilitate detection of the ectopically expressed CN molecules, each of the CNA carboxyl-terminal deletion mutants was epitope tagged with a peptide epitope from influenza virus hemagglutunin (HA), which is recognized with high affinity by the 12CA5 mAb (29). SDS-PAGE analysis of transiently transfected COS-7 cell extracts followed by immunoblotting with the 12CA5 mAb revealed that each of the carboxylterminally truncated CNA deletion series was efficiently expressed with the expected relative molecular mass ( (Fig. 1B). As expected, the full-length wild type CNA subunit (wtCNA 521 amino acids) bound to the FKBP.FK506 complex, as did CNAmutants truncated to amino acids 394 (CA394) and 376 (CA376); however, truncation of the CNA subunit to amino acid 355 (CA355) abolished the ability of CN to bind to FKBP.FK506. Similarly, wtCNA, CA394 and CA376, but not further carboxyl-terminal truncations of CNA, bound to cyclophilin A G A complexes (Fig. 1C). Thus, it appears that sequences of the CNA subunit between amino acids 355 and 376 are required for the interaction of CN with drugimmunophilin complexes.
The interaction of wtCN with drug-immunophilin complexes is known to be highly dependent upon calcium and is stimulated by calmodulin (15). Thus, the interaction of wtCNA with drug-immunophilin complexes is prevented when 25 m~ EGTA is included in the binding buffer (Fig. 2). In contrast, we found that CA394, which lacks the carboxyl-terminal 127 amino acids including both the calmodulin binding site and the autoinhibitory domain (see Fig. 71, interacted with drug-immunophilin complexes in the presence of EGTA in an apparently calciumindependent fashion (Fig. 2). Hence the CN carboxyl-terminal domain containing the calmodulin binding site is required for the calcium-dependent interaction of wtCN with drug-immunophilin complexes.
The region of the CNA subunit between amino acids 355 and 376 is located carboxyl-terminal to the phosphatase homology domain and is contained within a region of CNA which is highly conserved between species (21,23,(33)(34)(35)(36). This high degree of conservation indicates that this region might represent the binding site for the regulatory CNB subunit (21). To test this hypothesis, each of the carboxyl-terminally truncated CN mutants was immunoprecipitated with the 12CA5 mAb, and the immunoprecipitates were analyzed by immunoblotting with the VA1 mAb (281, which is specific for the CNB subunit. As shown in Fig. 3, the CNB subunit was coimmunoprecipitated with wtCNA, CA394, and CA376, but not further carboxylterminal truncations of the CNA subunit. In addition, the CNB subunit was also found to be associated with wtCNA, CA394, and CA376 eluted from drug-immunophilin complexes (data Thus, it appears that sequences of the CNA subunit between amino acids 355 and 376 are required for the interaction of CNAwith CNB. Interestingly, as shown above, this same region also appears to be required for the interaction of CN with drug-immunophilin complexes. Given the concordance between the amino acid sequences of the CNA subunit required for the interaction with both drugimmunophilin complexes and the CNB subunit, we next attempted to determine whether CNB itself was required for the interaction between drug-immunophilin complexes and the CNA subunit. To this end we took advantage of the yeast twohybrid system (26,37). This system allows the analysis of protein-protein interactions, by virtue of the ability of GAL4 chimaeric fusion proteins to reconstitute GAL4-dependent transcription. Accordingly, the 1-147 amino acid residue DNA binding domain of GAL4 was fused in-frame to human FKBP-12 sequences (GAL-FKBP), whereas the GALA activation domain was fused with the (23394 fragment (ACT-CNA394) of CNA. We monitored the interaction between GAL-FKBP and ACT-CNA394 by utilizing the Y153 indicator yeast strain (26). This strain of yeast, which is a histidine auxotroph and harbors a disrupted endogenous GAL4 gene, contains the His-3 gene under the control of a GAL4-dependent promoter. Growth of this strain on media lacking histidine therefore re- quires an interaction between the GAL-FKBP and ACT-CNA394 fusion proteins to reconstitute GAL4-dependent His-3 expression. Thus, Y153 were transformed with plasmids encoding GAL-FKBP and either ACT-CNA394 or ACT-CNA332. As a control, Y153 was also transformed with GAL4-E47 and ACTlyl. E47 and lyl are proteins that are known to interact strongly with each other independently of FK506. "ransformed colonies were streaked on selective plates in the presence or absence of FK506. As expected, since FK506 is required to promote the interaction between FKBP and CN, Y153 containing plasmids encoding both GAL-FKBP and either ACT-CNA394 or ACT-CNA332 did not grow on plates lacking FK506 (Fig. 4A). In contrast, however, significant growth of yeast containing both GAL-FKBP and ACT-CNA394 was observed when FK506 (200 ng/ml) was included in the plating media (Fig. 4B). As expected, since CA332 did not interact with drug-immunophilin complexes in vitro (see Fig. 1, B and C), Y153 transformed with GAL-FKBP and ACT-CNA332 did not promote growth even in the presence of FK506 (Fig. 4B). Thus, in the yeast two-hybrid system, as in vitro, FKBP interacts with CN in a strictly FK506-dependent fashion and exhibits the same sequence requirements. Having established the integrity of the interaction between FKBP and CN in the two-hybrid system, we next investigated whether CNB was necessary for this interaction. Accordingly, we generated a CNB null in Y153 by targeted disruption of the endogenous yeast CNB locus and then transformed this mutant strain (Y153ACNB) with plasmids encoding the same chimaeric fusion proteins as those used above. In contrast to the wild type Y153 strain, we were unable to detect an FK506-dependent interaction between GAL-FKBP and ACT-CNA394 in the mutant strain (Y153ACNB) containing the disrupted CNB gene, whereas the control interaction between GAL4-E47 and ACT-IyI was readily observed (Fig. 4C). Thus, it appears that the CNB regulatory subunit is necessary for the interaction of CN with the FKBP.FK506 complex.
As a result of the clearly important role of the CNB regulatory subunit in the interaction between FKBP.FK506 and CN, we wished to define the CNA subunit sequences required for runs anomalously on SDS-PAGE with an apparent molecular mass greater than that predicted. binding CNB. In preliminary experiments we had established that the CNB binding site was located in the carboxyl-terminal half of the CNA subunit distal to the phosphatase homology domain. To define further the sequences required for CNB binding we generated further amino-terminal truncations of the CNA subunit, epitope-tagged at the carboxyl terminus with the HA epitope, and tested them for their ability to coprecipitate the CNB subunit from transiently transfected COS-7 extracts. The expression of these amino-terminal truncation mutants is shown in Fig. 5A. For unknown reasons the NA364 fragment runs aberrantly on SDS-PAGE, with a slower mobility than the NA347 mutant. Extracts from COS-7 cells cotransfected with the amino-terminal CNA mutants and a plasmid encoding the CNB subunit were immunoprecipitated with the 12CA5 mAb and the immunoprecipitates analyzed by immunoblotting with the CNB-specific mAb, VA1. As shown in Fig. 5B, CNB was coimmunoprecipitated with NA347 but not NA364. Furthermore, a fragment of CNA encompassing amino acid sequences 347-394 was sufficient to bind CNB (NA347-CA394; Fig. 5). Taken together with the results from Fig. 3, since CA376 and not further carboxyl-terminal truncations of CNA were able to bind CNB, it appears that the CNA subunit sequences between 347 and 376 form the minimal binding site for the regulatory CNB subunit.
Having established the essential role of the CNB subunit in the interaction between FKBP.FK506 and CN, we wished to establish whether the CNB subunit itself was sufficient to mediate binding to drug-immunophilin complexes. Accordingly, we tested the ability of the amino-terminal truncations of CNA from Fig. 5 to bind to either GST-FKBP-12.FK506 or GSTcyclophilin ACsA complexes. As shown in Fig. 6, B and C, we were unable to detect any interaction between either NA347 or NA347-CA394 and drug-immunophilin complexes even though both mutants interact with CNB. Thus, although necessary, CNB is not sufficient for binding to FKBP.FK506 and cyclophilinCsA, indicating that sequences from the CNA subunit also play a major role in mediating the interaction of CN with drug-immunophilin complexes. To define those amino acid sequences of the CNA subunit required for binding, we tested the ability of a series of progressive amino-terminal truncations of CNA to bind to both FKBPeFK506 (Fig. 6B) and cyclophilin ACsA (Fig. 6C). Thus, extracts from COS-7 cells transiently cotransfected with each of the HA epitope-tagged amino-terminal CNA truncations together with the CNB subunit, were incubated with either GST-FKBP-12.FK506 or GSTcyclophilin ACsA complexes, and specifically bound material was analyzed by SDS-PAGE followed by immunoblotting with the 12CA5 mAb. N1-CA394 was retained by both the GST-FKBP-12.FK506 and GST-cyclophilin ACsA complexes, as was NA32, which lacks the first 32 amino acids of CNA; however, further truncation of CNA to amino acid residue 99 (NA99) resulted in a loss of binding to both FKBP-12.FK506 and cyclophilin ACsA. Thus, in addition to CNB, sequences of the CNA subunit between amino acid residues 32 and 99 are also required for the interaction of drug-immunophilin complexes with CN. These results are summarized in Fig. 7.

DISCUSSION
The inhibitory effects of the FKBP.FK506 and cyclophilin A-CsA complexes on the enzymatic activity of CN underlie the molecular basis of action of these potent immunosuppressive drugs (15,(17)(18)(19). In the current study, we have investigated the subunit requirements and the molecular sequence determinants that govern the interaction of CN with drug-immunophilin complexes (Fig. 7). Analysis of progressive carboxylterminal deletion mutants of CNA revealed that the autoinhibitory and calmodulin binding domains were completely dispensable for the interaction with drug-immunophilin complexes. Thus, CN mutants CA394 and CA376, which lack both the autoinhibitory and calmodulin binding domains, interacted strongly with both FKBP.FK506 and cyclophilin ACsA complexes (Fig. 1, B and C). Further truncation of the CNA subunit to amino acid 355, however, completely abolished the ability of CN to bind to drug-immunophilin complexes.
The interaction of wtCN with drug-immunophilin complexes is strongly calcium-dependent and appears to be stimulated in the presence of calmodulin (15). In contrast, the interaction of drug-immunophilin complexes with CA394 occurs in the presence of 25 mM EGTA and consequently appears to be calciumindependent (Fig. 2). These results are therefore consistent with the notion that the CN carboxyl terminus normally precludes the interaction of drug-immunophilin complexes with CN and that in the presence of elevated calcium a conformational change occurs in the CN molecule which makes the drugimmunophilin binding site accessible. The role of either calmodulin or the calcium-binding regulatory CNB subunit in this putative conformational change remains to be fully explored.
The region of CNA between amino acids 328 and 375, which lies immediately distal to the protein phosphatase homology region, is highly conserved in CNA subunits from a wide variety of species (21,23,(33)(34)(35)(36). This has led to the speculation that this region is involved in the binding of the regulatory CNB subunit (21,351). Consistent with this notion is the observation that CNB binds to CA376, but not further carboxyl-terminal truncations of the CNA subunit (Fig. 3). Thus, CNA sequences located between amino acid 376 and and the carboxyl terminus are completely dispensable for CNB binding. Moreover, a polypeptide corresponding to the sequence between amino acids 347 and 394 of the CNA subunit is sufficient to mediate binding of CNB (Fig. 5). Taken together, these results indicate that a 29-amino acid region between amino acids 347 and 376 most likely forms the minimal CNB binding site, but do not rule out the possibility that other CNA amino acid sequences may influence the affinity of the CNB interaction with the CNA subunit.
The observation that carboxyl-terminal truncations of the CNA subunit which failed to bind the regulatory CNB subunit also failed to bind to drug-immunophilin complexes, prompted us to investigate the potential role of CNB in the interaction of CNA with the drug-immunophilin complex. In this regard, utilizing the yeast two-hybrid system, we were able to show that the regulatory CNB subunit was in fact necessary for the interaction of the CNA subunit with FKBP.FK506 (Fig. 4). This result is in accord with both the 1:l stoichiometry of the CNM CNB subunits isolated after drug-immunophilin chromatography (15) as well as recently published enzymatic (391, biochemical (40) and genetic evidence (41). Although necessary, however, it appears that the CNB subunit per se is apparently not sufficient to mediate an interaction with drug-immunophilin complexes, since the CN amino-terminal mutants NA99, NA347, and NA347-NA394, which all bind to CNB (Fig. 5 and data not shown), fail to interact with FKBP.FK506 and cyclophilin ACsA complexes (see Fig. 6, B and C). This notion is further supported by the observations that the isolated CNB subunit is not retained on a drug-immunophilin affinity column (38) and is incapable of being directly biochemically cross-linked to drug-immunophilin complexes (40,44). It should be noted, however, that chemical cross-linking can potentially identify weak or transient interactions between molecules, whereas the affinity chromatography assay used in the present study is a more demanding test of intermolecular interactions. Whether our observed requirement for amino acids 32-99 indicates that this region is directly involved in the interaction with drug-immunophilin complexes or alternatively, is simply required for the structural integrity of the CN catalytic domain, thereby conferring an appropriate structural conformation that permits a high affinity interaction with drug-immunophilins, is not known. Why is the CNB subunit required for the interaction of CN with drug-immunophilin complexes? Certainly, the requirement for the CNB subunit can in part explain why FKBP.FK506 and cyclophilinCsA specifically interact with CN and not with the related serinelthreonine phosphatases, PP1 and PP2A (15,38). In this respect, although CN, PP1, and PPSA all share a common homologous protein phosphatase catalytic core (23,(33)(34)(35)(36), only CN interacts with the CNB regulatory subunit (20)(21)(22). One possible explanation for the CNB requirement is that the primary interaction between CN and the drugimmunophilin complex may indeed be with the CNB subunit. However, it is clear from the analysis of the CNA amino-terminal truncation mutants that the CNB subunit per se is not sufficient to mediate a n interaction with drug-immunophilin complexes. Thus, in this model the association of the CNA subunit with CNB would appear to be a prerequisite to confer an appropriate conformation on CNB to permit the interaction with drug-immunophilin complexes. However, since wtCN, NA32, NA99, NA347, and NA347-NA394 all bind CNB (Fig. 3,  5, and data not shown), yet only wtCN and NA32 bind drugimmunophilin complexes (Fig. 6), the conformation of CNB when bound to wtCN and NA32 would have to be fundamentally different than when bound to the other CN mutant proteins. A second possibility is that the CNB subunit is simply required to confer an appropriate conformation onto the cata-lytic CNA subunit, which then allows CNA to interact directly with the drug-immunophilin complex. A third more parsimonious possibility, however, is that the CNA and CNB subunits both make direct contact with the drug-immunophilin complex. Several recent studies have used cross-linking approaches in an attempt to identify the subunit of CN which directly contacts the drug-immunophilin complex (40,42,44). In two of the studies, drug-immunophilin complexes were selectively crosslinked to the CNB subunit (40,44) but only in the presence of the CNA subunit; cross-linking to isolated CNB subunits was not observed. In the other case, the CNA, but not the CNB subunit, was selectively cross-linked (42). The apparent disparity between these studies may be explained by the different chemical nature of the cross-linking reagents used, which may bias for or against potentially accessible amino acid residues available for cross-linking. Despite this lack of consensus, these cross-linking data indicate that drug-immunophilin complexes are most likely in very close approximation to both the CNA and the CNB subunits, although the exact contact sites remain to be identified.
Interestingly, none of the CN mutants described in the current study was able to discriminate between FKBP.FK506 and cyclophilin-CsA, consistent with previous findings that FKBP.FK506 and cyclophilinCsA appear to compete for the same binding site on CN (15,40). Structure-function analysis of the FKBP-12.FK506 interaction with CN has revealed the importance of several charged residues in FKBP-12 which appear to be involved in making direct contact with CN. These residues have defined a 100 A2 surface patch that corresponds to the putative CN binding site (43). It will be interesting to identify the corresponding residues on CN which are involved in mediating the interaction with drug-immunophilin complexes. In this regard, the CN two-hybrid system described herein, combined with a random mutagenesis approach, should prove useful for rapidly identifying contact residues in CN. Ultimately, however, either x-ray crystallographic or NMR structural analysis will be required to resolve the precise three-dimensional relationship of CN with drug-immunophilin complexes.