Cyclophilin A Binds to Linear Peptide Motifs Containing a Consensus That Is Present in Many Human Proteins*

Cyclophilin A (CypA) is a peptidyl-prolyl cis/trans-isomerase that is involved in multiple signaling events of eukaryotic cells. It might either act as a catalyst for prolyl bond isomerization, or it can form stoichiometric complexes with target proteins. We have investigated the linear sequence recognition code for CypA by phage display and found the consensus motif FGPXLp to be selected after five rounds of panning. The peptide FGPDLPAGD showed inhibition of the isomerase reaction and NMR chemical shift mapping experiments highlight the CypA interaction epitope. Ligand docking suggests that the peptide was able to bind to CypA in the cis- and trans-conformation. Protein Data Bank searches reveal that many human proteins contain the consensus motif, and several of these protein motifs are shown to interact with CypA in vitro. These sequences represent putative target sites for binding of CypA to intracellular proteins.

Cyclophilin A (CypA) 1 is a ubiquitously expressed protein that has been found in a variety of functional contexts. On the one hand, CypA serves as immunophilin and binding partner for the immunosuppressive drug cyclosporin A (1), and on the other hand CypA was found to catalyze the cis/trans-isomerization of proline imide bonds in peptides (2). Peptidyl-prolyl cis/trans-activity of CypA was shown to be relevant for protein folding in vitro (3), but it has been difficult to prove the relevance of catalysis in vivo. Recent experiments suggest that the tyrosine kinase Itk is regulated by the catalytic activity of CypA (4) and for the HIV Vpr protein cis/trans-interconversion of a critical proline residue might be catalyzed by CypA (5). The role of CypA as stoichiometric binding partner has also gained recent interest in the light of HIV infectivity and T cell signaling. For certain HIV-1 strains, the ability of the capsid protein (CA) to bind CypA correlates well with the infectivity of these strains (6). Structural analysis shows that the GP-dipeptide of the CA 86 -93 loop is deeply inserted into the CypA binding site, and mutation of glycine to alanine reduces ground state binding, presumably fostering the cis/trans-interconversion rate (7). Therefore, binding of the X-P-dipeptide bond may result in stable complex formation or transient interaction and catalysis, depending on the sequence or conformational context of the critical proline. The sequence context for catalysis by CypA has been investigated in detail (8) and revealed no stringent requirements for the nature of the amino acid flanking the central proline. For immunophilins as binding modules it was suggested that peptide conformation is central to the formation of complexes (9); however, a global analysis of the sequences binding to CypA is missing. Here we apply phage display from a randomized 9-mer peptide library to map the recognition code for linear peptide sequences that interact with CypA. We identify the consensus sequence as FGPXLp and confirm the importance of the individual amino acids for the peptide FGPDL-PAGD. The latter peptide is an active site inhibitor, and NMR spectroscopy shows that the binding epitope overlaps with the interaction site of cyclosporin A and other CypA ligands. The recognition signature is present in a number of human proteins, and peptides derived from these proteins bind to CypA when spotted onto a nitrocellulose membrane. The identification of novel CypA binding sites sets the stage for the detection of yet unknown CypA interaction partners. As a first example, we show that a phage display-derived peptide motif is bound by CypA in the context of the entire cytoplasmic domain of the T cell adhesion molecule CD2.

EXPERIMENTAL PROCEDURES
Preparation of Recombinant Proteins-To construct GST-CypA fusion protein, the gene encoding human cyclophilin A was subcloned into pGEX-4T-1 vector (Amersham Biosciences) using BamHI and EcoRI restriction sites. Escherichia coli BL21 (DE3) cells harboring the recombinant plasmid were grown in 2 ϫ YT medium (16 g of tryptone, 10 g of yeast extract, 5 g of NaCl per 1 liter of medium) at 37°C till A 600 of 0.5, gene expression was induced with 1 mM isopropyl ␤-D-thiogalactopyranoside, and the culture was harvested after 4 h of growth. GST-CypA was purified from the soluble fraction using affinity chromatography on a prepacked HiTrap glutathione-Sepharose column (Amersham Biosciences) and subsequent gel filtration (Superdex® 75, Amersham Biosciences) in phosphatebuffered saline (pH 7.9) supplemented with 1 mM ␤-mercaptoethanol. 15 N-Labeled CypA was prepared using the expression plasmid pTFT74 (10) with the His 6 -tag engineered at the N terminus of the protein. E. coli BL21 (DE3) culture was grown at 37°C in 1 ϫ A minimal medium (60.3 mM K 2 HPO 4 , 33.1 M KH 2 PO 4 , 1.7 mM sodium citrate, or 15 N-labeled NH 4 Cl, 0.8% w/v glucose, 1 mM MgSO 4 ) until A 600 of 0.2 and induced by 50 M of isopropyl ␤-D-thiogalactopyranoside overnight at 18°C. Protein was isolated using a HiTrap nickelagarose column (Amersham Biosciences) and a gel filtration step as described above. The CD2 cytoplasmic domain was expressed as fusion protein with the Trp-leader sequence in the vector pTMHa (courtesy of P. Kim, Whitehead Institute). The protein was then solubilized from inclusion bodies, cleaved with cyanogen bromide, and purified by high performance liquid chromatography on a Vydac 218TP reversed-phase column (Grace GmbH & Co.).
Peptidyl-Prolyl cis/trans-Isomerase Assay-PPIase activity was determined in a coupled assay with chymotrypsin using the synthetic tetrapeptide N-Suc-Ala-Ala-Pro-Phe-p-nitroanilide (Merck) as substrate at a final concentration of 100 M (11). Measurements were carried out at 10°C, and increase in absorbance was followed at 390 nm for 2-10 min. Concentrations of chymotrypsin and PPIase in the reaction mixture in routine assays were 10 M and 5-50 nM, respectively. Kinetic progression curves were analyzed by a non-linear least squares fit to a single exponential function assuming a first order rate reaction at [S] 0 Ͻ ϽK m . The rate constant of non-enzymatic spontaneous isomerization was estimated in the absence of the PPIase and subtracted from the values for the enzymatic reaction. To test inhibition of PPIase, rate constants for the first order reaction were estimated in the presence of 10 -400 M of the peptide selected via the phage display procedure. The IC 50 was derived using a single exponential decay function.
Phage Display Screening-A randomized nonapeptide library (X9) fused to the gpVIII protein of the phagemid vector pC89 (12) was used for the phage display procedure. Screening of the library was performed as follows: 100 g of GST-CypA fusion protein bound to glutathione-Sepharose 4B gel (Amersham Biosciences) was incubated overnight with 10 10 infectious particles in phosphate-buffered saline supplemented with 5 mg/ml BSA, at 4°C in a total volume of 400 l. After washing 10 times with phosphate-buffered saline, 0.1% Tween 20, the bound phages were eluted with 100 mM glycine/HCl (pH 2.2). E. coli XL-1Blue cells were infected by the eluted phage particles and the phages amplified using the helper phage VCSM13 (Stratagene) according to the standard protocol (13).
Binding Studies of Membrane-attached Peptides-Single substitutions of the peptide EGEFGPDLPAGD were generated by semiautomated spot synthesis on Whatman 50 cellulose membranes as described by Kramer and Schneider-Mergener (14). The membranes were incubated with GST-CypA (10 g/ml) overnight. After washing, bound GSTfused protein was labeled with rabbit polyclonal anti-GST antibody (Z-5, Santa Cruz Biotechnology) and horseradish peroxidase-coupled anti-rabbit IgG antibodies (Rockland). An enhanced chemiluminescence substrate (SuperSignal West Pico, Pierce) was used for detection on a LumiImager TM (Roche Applied Science).
NMR Analysis of CypA-Ligand Interaction-All NMR experiments were performed at 298 K using a Bruker DRX600 instrument equipped with a standard triple resonance probe. Data processing and analysis were carried out with the XWINNMR (Bruker) and SPARKY software packages (15). In the NMR titration experiment increasing amounts of the synthetic peptide (0.1-1.5 mM) of the sequence Ac-FGPDLPAGD-NH 2 or of the entire cytoplasmic domain of CD2 (amino acids 241-351) were added to a 0.2 mM sample of 15 N-labeled CypA in 50 mM sodium phosphate buffer (pH 7.0), 1 mM ␤-mercaptoethanol. The sum of the chemical shift changes for 15 N and 1 H atoms in the heteronuclear single quantum correlation spectra of free and ligand-bound CypA was determined as [(⌬ 1 H_cs) 2 ϩ (⌬ 15 N_cs) 2 ] 1/2 , where ⌬ 1 H_cs is in units of 0.1 ppm, and ⌬ 15 N_cs is in units of 0.5 ppm.
Modeling of the CypA-Peptide Complex-Since the cyclophilin-bound CA N loop has a similar sequence (XGPX) as the peptide investigated here, we used the x-ray structure of the complex CypA/HIV-1 CA N , Protein Data Bank entry 1M9C (7), as a template for modeling. The FlexX package (16) was used for docking of the peptides FGPDLPAGD and FGPDLP to CypA. All atoms of CypA and Pro 90 (chain D) were fixed during the docking, and the remaining residues of the peptide were subjected to exhaustive conformational analysis. Two complex conformations displaying the highest binding energy score were analyzed in more detail. After docking, the two candidate complexes were solvated in cubic boxes using TIP3P water molecules (17) with an initial minimum distance of at least 8 Å between the boundaries of the box and the nearest solute atom. The systems were first optimized by 500 steps energy minimization each using the GROMACS3.14 package (18) and the OPLSAA force field (19). Subsequently, a 1-ns molecular dynamics simulation was carried out for each candidate with restrained positions for all the template atoms in CypA and Pro 90 (chain D). Finally, two complete 10-ns molecular dynamics simulations with no positional restraint were performed to fully optimize the structure of the two candidates. The LINCS procedure (20) was applied to constrain all bond lengths. The time step of the simulation was set to 2 fs. A 9 Å cutoff was used for the short range non-bonded interactions and the lists of nonbonded pairs were updated every 10 steps. The Particle Mesh Ewald method (21) with a grid size of 1.2 Å was used to calculate long range electrostatic interactions. In the simulations, temperature and pressure were maintained by weak coupling to an external bath (22). All simulations were performed at 300 K temperature. Cluster analysis was applied after the simulations. Main-chain atoms and C ␤ atoms were selected for calculating the root mean square deviation matrix. The root mean square deviation cutoff was set to 1.0 Å.

RESULTS
Phage Display-Phage display was performed with a gene-VIII-derived vector that presents multiple copies of peptides from a randomized X9-library on the surface of phage M13 (12). The resulting sequences obtained after five rounds of panning are shown in Fig. 1A. The GP-dipeptide motif was found in all sequenced clones in full agreement with the structure-based prediction of previous investigations (7,23). The residue directly C-terminal to the proline at position i is assumed to point toward the solvent, while the amino acid at position iϩ2 shows a preference for the hydrophobic side chains of leucine or tyrosine. Proline is preferred at position iϩ3, while residues further C-terminal did not show enrichment of any given amino acid. The identified consensus sequence GPXLp highlights the importance of the GP motif and hydrophobic interactions as major components of CypA-peptide interactions. The peptide FGPDL-PAGD was further investigated for its potential to inhibit catalysis of cis/trans-prolyl isomerization of the model substrate N-Suc-Ala-Ala-Pro-Phe-p-nitroanilide by CypA. As can be seen from Fig. 1B, the IC 50 for inhibition is 27.8 Ϯ 2.4 M and suggests that the peptide interacts with the active site of CypA.
Peptide Substitution Analysis-To refine the phage display results we performed a peptide substitution analysis of the most frequently selected peptide GPDLPAGD (Fig. 1C). The four N-terminal amino acids, EGEF, belonging to the phage capsid protein gVIII were also considered for the analysis, since they are present in all the selected ligands directly N-terminal to the consensus motif and may therefore contain residues that are essential for binding. In addition to the absolute requirement of proline at position i, glycine at position iϪ1 must be present in this sequence to allow interaction of the peptide with CypA. Surprisingly, the phage protein derived phenylalanine at position iϪ2 is exclusively needed for high affinity binding. While position iϩ1 tolerates most amino acid substitutions, position iϩ2 shows preference for leucine, although substitution by other hydrophobic amino acids, as well as by arginine still results in detectable spot intensities. At position iϩ3, proline is one of the amino acids resulting in favorable binding, but arginine and cysteine mutant peptides show comparable affinities. For the other positions, no clear preference for certain amino acids can be observed, and we conclude that these residues do not contribute to complex formation with CypA. The combined results from phage display and peptide substitution analysis converge on FGPXLp as the consensus motif for efficient CypA binding.
NMR Analysis of CypA Binding to the FGPDLPAGD Peptide-To map the binding interface for the phage display derived peptide, we monitored 15 N-1 H chemical shift changes of CypA amide groups upon ligand addition. Fig. 2A shows three superimposed heteronuclear single quantum correlation spectra of samples containing CypA alone (red) and CypA in the presence of 0.1 mM (green) and 1.5 mM (blue) concentrations of the peptide. Chemical shift changes are significant for resonances that define the epitope accommodating the peptide. Additionally, since the off-rate of the ligand binding is fast in regard to the NMR time scale, the dissociation constant of the interaction could be determined using titration data (the curve in the inset of Fig. 2A). We obtained a K D of 50 Ϯ 13 M, assuming a simple A ϩ B 7 AB binding model. This value is similar to the IC 50 for the competitive inhibition of the model peptide substrate. The amino acids of CypA that are affected by peptide binding are shown as colored surface in Fig. 2B. They define a contiguous area that superimposes well with the interaction sites of known CypA-ligand structures, as can be seen for the residues His 87 -Pro 93 of the CA N ligand. In Fig. 2B the most prominent chemical shifts are observed for amino acids situated in several ligand recognition areas. First, the N-terminal "iϪ2 recognition area" comprising residues Gln 63 , Thr 73 , Lys 82 , Ala 101 , Thr 107 , Asn 108 , and Gly 109 ; second, the hydrophobic pocket accommodating the central proline that consists of Met 61 , Phe 113 , and Leu 122 ; and third, the "iϩ2 recognition area" involving Ile 57 , Trp 121 , and Asn 149 , and surrounding loops (Fig. 2). In addition, Arg 55 , a critical residue for catalysis (24), also displays large chemical shift changes (Fig.  2). The hydrophobic pocket is occupied by a proline residue in several CypA-ligand structures (summarized in Refs. 7 and 9), and the importance of the GP motif for binding (Fig. 1C) suggests a similar interaction site for the phage display-derived peptide. Anchoring of the central GP motif at the position observed in the crystal structure of the CypA/HIV-1 CA N complex was used to model the interaction between CypA and the phage displayderived ligand.
Model of CypA Bound to the Phage Display-derived Peptide-The docked and optimized model of CypA bound to the FGP-DLP motif of the phage display-derived peptide as well as the superposition of the cis-and trans-variants of the peptide are shown in Fig. 3, A and C. To analyze possible interactions of the ligand with CypA a cutoff was set to 3.4 and 5.0 Å for hydrogen bonds and van der Waals contacts, correspondingly. The GP motif adopts a trans-conformation and fits well to the hydrophobic pocket defined by residues Ile 57 , Phe 60 , Met 61 , Ala 101 , Ala 103 , and Leu 122 , involving also Gln 63 , Asn 102 , Phe 113 , and His 126 (Fig. 3A). The glycine at position iϪ1 is essential, since any other residues would result in a clash between the ligand side-chain and CypA. The phenylalanine at position iϪ2 fits well into the indentation formed by Lys 82 , Ala 101 , Asn 102 , Ala 103 , Thr 107 , Asn 108 , Gly 109 , and Gln 111 . These results are in agreement with the peptide substitution analysis, which shows the FGP segment to be exclusively required for high affinity binding. At position iϩ1, the side chain of aspartic acid points toward the solvent, which is in accordance with the mutational analysis (Fig. 1C), where most amino acid substitutions are tolerated at this position. The leucine residue at position iϩ2 is involved in van der Waals interactions with Ile 57 , Asn 71 , and Arg 148 of CypA; however, membrane spot analysis suggests that other hydrophobic amino acids at this position can support similar interactions (Fig. 1C). Superposition of the modeled peptide with the x-ray structure shows a good fit of the first five residues (FGPDL) to the CA fragment despite the difference in the amino acid sequence (Fig. 3B). The proline at position iϩ3 in our model points toward the solvent, while it is oriented toward CypA in the experimental structure (Fig. 3B). Fig. 3C shows the superposition of the two docking models with transand cis-conformations of the GP motif. It can be seen that the CypA binding site could accommodate both variants, since the major conformational change affects G iϪ1 , whereas the hydrophobic interactions can be maintained. Finally NMR studies of the peptide in complex with CypA will allow to experimentally validate this model.
CypA Binding to Protein Data Bank-derived Sequences-To identify possible interaction partners of CypA a profile search with the phage display-derived FGPXLp signature was performed for the human genome. 104 peptide sequences from intracellular human proteins that match the consensus motif most closely were spotted onto a cellulose membrane and evaluated for binding to GST-CypA (Fig. 4). As a positive control, phage display-derived ligands were used (Fig. 1A) and peptides were grouped into three distinct categories according to their binding efficiency (Table I). Several of the peptides bind with FIG. 1. Phage display-derived ligands of CypA. A, ligands of CypA selected from a randomized 9-mer peptide phage displayed library. The first four amino acids in all sequences are part of the phage gene VIII protein, while the nine subsequent positions were randomized. B, kinetics of inhibition of peptidyl-prolyl isomerase activity by the nonapeptide FGPDLPAGD. A proteasecoupled enzyme assay was used to test the release of the colored reaction product (11). C, substitution analysis of the peptide FGPDLPAGD. All possible single site substitution variants of the peptide were synthesized on a nitrocellulose membrane. The single letter code above each column indicates the amino acid that replaces the corresponding wild-type residue; the row defines the position of the substitution within the peptide. Spots in the first column display wild type peptide in all cases. The membrane was incubated with GST-CypA, and bound protein was detected with an anti-GST primary antibody and a horseradish peroxidase coupled secondary antibody. The relative spot intensities correlate qualitatively with the binding affinities (39).
affinities that are comparable with the phage display peptide (approximated K D values: 10 -100 M) and are therefore approaching the affinities of known binding partners of CypA (e.g. CA). Peptides having the highest homology with the consensus "FGPXLp" display the strongest binding. For example, a sequence from signal-induced proliferation-associated protein-1 matches exactly with the consensus motif and is a promising candidate for a physiological interaction with CypA. None of the sequences is known to interact with CypA yet, but it is interesting to note that several zinc finger proteins of the Sna family contain binding motifs for CypA and might indicate a link for CypA to transcription. Surprisingly, a sequence from the ryanodine receptor, a protein that is bound with high affinity by the FKBP12 protein, shows binding to CypA. However, it has to be kept in mind that the linear sequences displayed on the membrane might not be freely accessible in the context of the full-length proteins. Or else, as part of a structured domain, these sequences might be solvent-exposed but adopt a restricted conformation that does not allow the CypA-compatible structure to be formed. To test the potential of CypA to bind to sequences derived from a known structural context, we investigated the binding to such peptides presented on a cellulose membrane. Since CypA is suggested to show a slight preference for cis-proline bonds (25,26), we used a selected set of 240 sequences of cis-proline-containing peptide motifs of Protein Data Bank-deposited proteins (www.rcsb.org/ pdb/) for binding analysis. Seven of these peptides bind with reasonable affinity to CypA showing that peptides adopting a cis-conformation of the X-P bond in naturally folded proteins could be targets of cyclophilin. On the other hand, we did not observe binding of sequences from established CypA binders as for example CA or Itk when displayed as linear peptide motifs (Fig. 4, spots 1-4 and 11). This might imply, in these cases, that the binding conformation of the peptides is only adopted in the context of the folded protein and is unlikely to be populated to a significant degree in the unconstrained peptides as displayed on the membrane. Our results therefore show that the requirements for binding of constrained peptides within folded domains and linear sequences derived by phage display are different and that CypA is able to interact with extended peptide structures that are putative interaction sites within human proteins. CypA Binding to an Expressed Protein Domain Containing the Consensus Sequence GPXLP-One of the data base-derived sequences that resulted in binding to CypA belongs to the reductase domain of the neuronal nitric-oxide synthase isoform , which contains the motif KGPPLP. Sequence comparison showed the T cell adhesion molecule CD2 to contain this sequence (designated as proline-rich region 4 or PR4) as part of its cytoplasmic domain. Since CD2 is suggested to be involved in Itk signaling (27), and since Itk directly binds to CypA (4), we further investigated a possible direct link between CypA and CD2. The entire cytoplasmic domain of human CD2 was added to 15 N-labeled CypA, and 15 N-1 H NMR correlation spectra were recorded. Fig. 5 shows the superposition of a spectra of 15 N-CypA alone (red) and of CypA in the presence of equimolar amounts or a 6-fold excess of CD2 cytoplasmic domain (green and blue, correspondingly). Resonances of amide groups of the CypA active site residues display small, but significant, chemical shift changes upon CD2 ligand addition and indicate a weak but specific interaction between the two proteins. This result shows that a peptide sequence similar to the phage display consensus motif can bind to CypA in the context of an intact protein domain. In respect to CD2 signaling, our findings suggest a CypA-mediated link between Itk and CD2.

DISCUSSION
Identification of the FGPXLp consensus motif suggests a common binding mode for linear peptides that bind to CypA with considerable affinity. Since we show active site binding of the phage display-derived peptide FGPDLPAGD by inhibition experiments and NMR spectroscopy, it is likely that most of the sequences of the phage display ligands (Fig. 1A) and of the motifs derived from human proteins (Fig. 4) are recognized by CypA at the site surrounding the proline binding pocket. The preference of CypA for the GP dipeptide has been observed previously (7) and is supported by our peptide substitution analysis (Fig. 1C). In addition, in our model of the CypA bound peptide the GP motif is stably anchored in the binding cleft, while additional van der Waals interactions are provided by phenylalanine at position iϪ2 and leucine at position iϩ2 of the   Fig. 4 A, peptides derived from sequences of human proteins on the basis of similarity to the phage display consensus; B, peptides adopting cis-conformation of the proline residue in naturally folded proteins (Protein Data Bank entry codes are indicated). The numbers in the right column correspond to the numbers on the membrane. 16-mer sequences are shortened for clarity and aligned to demonstrate dependence of affinity on the resemblance of the consensus. Binding consensus is shown above the peptide sequences in each section of the ligand. The second proline at position iϩ3 is introducing a "kink" in the backbone conformation of the ligand, thereby pointing away from the CypA surface. However, in the experimental structure of the CypA-CA N complex this proline is a part of a PPII-helical conformation (Fig. 3B). Our substitution analysis for proline iϩ3 demonstrates that most residues in this position with high PPII-helix propensities (e.g. Pro, Gln, Asp, Ala, Arg) (28), show detectable binding, while those with low PPII-helix preferences (His, Thr, Ile, Val) prohibit stable interactions (Fig. 1C). Therefore, the proline at position iϩ3 of the phage display-derived ligand may also be a part of PPIIhelix and residues supporting formation of the helix are preferred at this position.
Taken together our data give a rational for the observed binding of a linear peptide ligand to CypA. Since the binding affinity of the phage dipsplay peptide (K D ϭ 50 M) is similar to the affinity of natively folded proteins, for example immature HIV-1 capsid protein p24 (K i ϭ 8.2 M), or its domains (K D ϭ 10 -42 M) (29 -31), we propose that linear peptide binding can be of physiological relevance. Such binding motifs, when present in unstructured parts of cytoplasmic proteins, may well represent interaction partners of CypA in vivo. In addition, a possible chaperone activity of CypA may depend on its ability to interact with unconstrained regions of human proteins. As an example, we have investigated the binding of CypA to the T cell adhesion molecule CD2, which contains five proline-rich sequence stretches within its cytoplasmic domain. Proline-rich sequence 4 of CD2 comprises the relaxed consensus motif GPXLp, that is present in one of the peptides derived from the data base search (Fig. 4, spot 37). Clearly, there is a weak, but specific, interaction between the two proteins (Fig. 5). The residues that show the most significant chemical shifts upon addition of CD2 are Arg 55 , Gln 63 , and Asn 102 , which are involved in catalysis by CypA, as well as Ile 57 , Ala 101 , Phe 113 , Trp 121 , and Leu 122 , which are known to form the ligand binding site (Fig. 5, also compare with Fig. 2). Whether such a weak interaction is of physiological relevance remains to be investigated, but it is noteworthy that both CD2 and CypA are implicated in Itk signaling in T cells and might therefore be part of a larger assembly involved in the regulation of cytokine production (4,27).
With regard to the well characterized role of CypA in cyclosporin A-induced immunosuppression its cellular function becomes only slowly unraveled. The function of CypA as an immunophilin suggested a role in immune cell signaling, and recent experiments with CypA Ϫ/Ϫ mice show reduced interleukin-4 production and enhanced inflammatory responses, thereby highlighting the role of CypA in T cell signaling (32). Interestingly, we found a peptide from the N-terminal SH3 domain of Vav protein (Vav-SH3 N ) to interact with CypA (Fig.  4, spot 236). While the native Vav-SH3 N did not bind to CypA, we observed an acceleration of its in vitro refolding in the presence of the PPIase (data not shown). This indicates that CypA can interact with Vav-SH3 N domain in the unfolded or partially structured state, as observed for other proteins (3). Since Vav and CypA were both shown to interact with Itk in T cells (4,33), it will be interesting to investigate a possible interplay between the three proteins.
Furthermore, CypA is highly up-regulated in neuronal tissues (34) and has also been implicated in the response of cells to oxidative stress (35). Several proteins with sequences that bind with significant affinity to CypA are expressed in brain, as for example the neuron navigator (36). Another set of proteins identified from our list as potential candidates for CypA binding is involved in cell cycle progression (Table I). Since CypA has been implicated in proliferation of neuronal (34) and endo-thelial cells (37), it will be interesting to validate a direct interaction in more physiological settings. The finding that cytochrome P450 might be directly recognized by CypA is a possible link between oxidative stress and CypA function (35). Interestingly, the ryanodine receptor 3 from brain comprises a putative CypA binding site. Ryanodine receptors are Ca 2ϩ channels that allow the influx of Ca 2ϩ into the cytoplasm from the endoplasmatic reticulum, and their operation is known to depend on the peptidyl-prolyl cis/trans-isomerase FKBP12 (38). An overlapping function of CypA in this context calls for further investigations.