Escherichia coli cyclophilin B binds a highly distorted form of trans -prolyl peptide isomer Michiko Konno 1 , Yumi Sano 1 , Kayoko Okudaira 1 , Yoko Kawaguchi 1 , Yoko Yamagishi-Ohmori 1 , Shinya Fushinobu 2 and Hiroshi Matsuzawa 2, * 1 Department of Chemistry, Ochanomizu University, Otsuka, Bunkyo-ku, Tokyo, Japan; 2 Department of Biotechnology, The University of Tokyo, Yayoi, Bunkyo-ku, Tokyo, Japan Cyclophilins facilitate the peptidyl-prolyl isomerization of a trans-isomer to a cis-isomer in the refolding process of unfolded p roteins to recover the natural folding state with cis-proline conformation. To date, only short peptides with a cis-form proline have been observed in complexes of human and Escherichia coli proteins of cyclophilin A, which is present in cytoplasm. The crystal structures analyzed in this study show two complexes in which peptides having a trans- form proline, i.e. succinyl-Ala-trans-Pro-Ala-p-nitroanilide and acetyl-Ala-Ala-trans-Pro-Ala-amidomethylcoumarin, are bound on a K163T mutant of Escherichia coli cyclo- philin B, the preprotein of which has a signal sequence. Comparison with cis-form peptides bound to cyclophilin A reveals that in any case the proline ring i s inserted into the hydrophobic pocket and a hydrogen bond between CO of Pro and N g2 of Arg is formed to fix the peptide. On the other hand, in the cis-isomer, the formation of two hydrogen bonds of NH and CO of Ala preceding Pro with the protein fixes the peptide, whereas in the trans-isomer formation of a hydrogen bond between CO preceding Ala-Pro and His47 N e2 via a mediating water molecule allows the large distor- tion in the orientation of Ala of Ala-Pro. A lthough loss of double bond character of the amide bond of Ala-Pro is essential to the isomerization pathway occurring by rotating around its bond, these peptides have forms impossible to undergo proton transfer from the guanidyl group o f Arg to the prolyl N atom, which induces loss of double bond character. Keywords: cyclophilin; isozyme; peptidyl-prolyl cis-trans isomerase; peptide binding; refolding. Cyclophilins (CyPs) exist abundantly an d ubiquitously in a broad range of organisms from Escherichia coli to humans [1–3]. Two cyclophilins, E. coli CyPA and E. coli CyPB, have been identified [4,5], and at least five cyclophilins in mammals (human CyPA–D, CyP40) [3] have been identi- fied. CyPB homologs have a membrane-binding signal sequence in the amino-terminal, w hereas CyPA homologs are present in the cytoplasm [1–5]. It has been reported that E. coli CyPB exhibits almost equal activity of peptidyl- prolyl isomerization from cis- to trans-form of short peptides to E. coli CyPA [5]. Because cis-proline conforma- tion in the polypeptide spontaneously converts to trans- conformation, acceleration of the isomerization of the cis-isomer to the trans-isomer is not thought to be the function of the CyPA and CyPB proteins in these cells. On the other hand, CyPs facilitate the step of the isomerization in which a trans-isomerisconvertedtoacis-isomer in the process of the refolding of unfolded proteins. Although the natural roles of CyPs are still poorly understood, they may be related to the fixing of distorted trans-isomers at the intermediate step of converting the trans-tothecis-isomer of proteins. It has been reported that in the N-terminal domain of the HIV-1 capsid protein, a loop containing Gly- trans-Pro binds to the human CyPA; this occurs at a position where the Gly residue assumes //w angles in the regions disallowed for residues with side chains [6]. Because the distortion of the loop of the c apsid protein is concen- trated in the torsional angles of the Gly residue, but as most proteins with a cis-proline, the refolding process of which is accelerated by CyPs, have no flexible Gly residue at the position immediately p receding Pro [7–11], this binding conformation of the capsid p rotein is not sufficient to serve as a model of t he intermediate formed during isomerization of the trans-isomer to t he cis -isomer. To date, the b inding structures of peptides in th e cis-proline form have been reported for human CyPA [12–15] and E. coli CyPA [16], but the peptides f orming the disto rted trans-conformation have not yet been identified. Based on NMR measurements, Eisenmesser et al.[17] proposed a possible mechanism by which the C-terminal peptide segment containing a Pro residue of a tetrapeptide rotates around the Phe-Pro peptide bond. Their results indicated that the cis-form in the initial state is inserted into a hydrophobic pocket of human CyPA, whereas the trans-form in the final state is released from the pocket. The Correspondence to M. Konno, Department of Chemistry, Ochanom- izu University, 2-1-1 Otsuka, Bunkyo-ku, Tokyo 112–8610, Japan. Fax: +81 359785717, Tel.: +81 359785718, E-mail: konno@cc.ocha.ac.jp Abbreviations: CyP, cyclophilin; Suc, succinyl; pNA, p-nitroanilide; Ac, N-acetyl; AMC, amidomethylcoumarin; PPIase, peptidyl-prolyl cis-trans isomerase. *Present address: Department of Bioscience and Biotechnology, Aomori University, 2-3-1 Kohbata, Aomori, 030-0943, Japan. (Received 23 March 2004, revised 1 August 2004, accepted 4 August 2004) Eur. J. Biochem. 271, 3794–3803 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04321.x trans-form, the proline of which is bound to the pocket, is only the loop of the capsid protein. It has been reported that E. coli CyPA accelerates the conversion of trans- to cis-form on peptide bonds of not only S er54-Pro55 but also Tyr38- Pro39 in the final refolding process of the unfolded RNase T1, w hereas h uman CyPA accelerates only the conversion on the Ser54-Pro55 peptide bond [8]. The fact that E. coli CyP is bound to structurally diverse substrates, and the notion that CyPA and CyPB should affect d ifferent target proteins, led us to consider whether E. coli CyPB might bind to peptides of the disto rted trans-isomer. E. coli CyPB [4,5] consists of 190 amino acids and has a signal sequence of 24 residues at the N-terminal end. CyPB molecules are processed and forms of CyPB without the signal peptide are present in the periplasm of E. coli cells (here, residues number 25–190 are defined as 1–166). Analysis of a complex in which a protein is bound to a peptide containing a Pro residue of trans-form in the initial state of the isomerization reaction from a trans-tocis- isomer will be needed. As the isomerization of t he trans-to cis-isomer of the short peptides proceeds very slowly or does not occur at all in CyPB molecules, we used short peptides in order to obtain the complexes with the distorted trans- isomer. We could obtain no crystals of E. coli CyPB having suitable size for X-ray analysis. Thus, we designed a K163T mutant protein of E. coli CyPB that can crystallize under conditions similar to those for the crystallization of E. coli CyPA. Finally, we succeeded in crystallizing K163T mutant proteins complexed with a tripeptide, succinyl-Ala-Pro- Ala-p-nitroanilide (Suc-Ala-Pro-Ala-pNA), and w ith a tetrapeptide, acetyl-Ala-Ala-Pro-Ala-amidomethylcouma- rin (Ac-Ala-Ala-Pro-Ala-AMC), and w e found highly distorted trans-peptides. We carried out comparisons of these complexes with a complex of E. coli CyPA and the tripeptide containing cis-proline , and with a complex of human CyPA bound to the loop containing the proline in the trans-form of t he capsid protein. In addition, we discussed the importance of the conversion from the sp 2 -to sp 3 -hybrid configuration on the prolyl N atom in the isomerization of the proline from t he trans-tocis-isomer of unfolded proteins in the refoldin g process, as well as in the isomerization of the short peptides containing the proline from the cis-totrans-isomer. Materials and methods Preparation of mutant E. coli CyPB The plasmid pATtrpEPPIa, which has a ClaIandaBamHI site added at the 5¢ and 3¢ ends, respectively, of the gene encoding amino acid residues 1–190 of E. coli CyPB, was kindly supplied by N. Takahashi, Tokyo University of Agriculture and Technology, Tokyo, Japan [5]. PCR was carried out using a 32-mer primer of 5¢-AAAAAGAAT TCATCGATATGTTCAAATCGACC-3¢ with EcoRI added at the 5¢ end of ClaI, a 23-mer primer of 3¢-CGAGA CGGCATTCCTAGGTTTTT-5¢, and pATtrpEPPIa as a template. The P CR fragment was cleaved with Eco RI and BamHI, and was subcloned into pUC118, producing an expression vector, pUCPPIb. Mutation was introduced into pUCPPIb u sing a QuickChange TM Site-Directed M utagen- esis Kit (Strategene, La Jolla, CA). The codon for Lys187 (AAA) was replaced with ACA (as the purified proteins lose signal sequences of 24 amino acids at the N-terminal, the mutant protein is designated K163T in this paper). The sequences of the mutated DNA were checked using an ABI 373 DNA sequencer (Applied Biosystems). The ClaI- BamHI fragment of each mutant plasmid was replaced with the ClaI-BamHI fragment of pATtrpEPPIa. This expression plasmid p ATtrpEPPIa (KT) was transformed in E. coli HB101 cells. The cells were cultured at 37 °Cin M9CA medium [0.05% (w/v) NaCl, 0.1% (w/v) NH 4 Cl, 0.2% (w/v) casamino acid, 0.2% (w/v) glucose, 2.0 m M MgSO 4 ,0.1m M CaCl 2 ,0.6%(w/v)Na 2 HPO 4 ,0.3%(w/v) KH 2 PO 4 , pH 7.4] and the proteins w ere purified as described previously [5]. The PPIase activity was measured using the synthetic peptide Suc-Ala-Ala-Pro-Phe-AMC (Peptide Institute, Inc., Osaka, Japan). The synthetic peptide [a 40 lL solution containing 1.67 m M of peptide, 17% (v/v) dimethylsulfoxide and 35 m M HEPES buffer, pH 7.8] was preincubated with proteins (1.1–4.8 n M )in2mLof35m M HEPES buffer containing 5 m M 2-mercaptoethanol (pH 7.8), and the assay was started by mechanical mixing with 40 lLof 0.58 m M chymotrypsin in a spectrophotometer cell. The fluorescence of AMC from the cleaved trans-peptide [18] was measured at 460 nm using a FP-770 fluorescence spectrophotometer (JASCO Corp., Tokyo, Japan) at 10 °C. Crystallization, data collection, structure determination and refinement Whole crystals of K163T mutant proteins were grown by vapor diffusion in hanging drops at 20 °C. A 9 lLdrop containing 0.57 m M (12 mgÆmL )1 )proteinand6m M trip- eptide (Suc-Ala-Pro-Ala-pNA (Peptide Institute, Inc.), or tetrapeptide (Ac-Ala-Al a-Pro-Ala-AMC (Bachem, Switzer- land), 38–40% (w/v) saturated ammonium, sulfate, 9% (v/v) methanol, 0.04% (w/v) NaN 3 ,and50m M Tris/HCl buffer (pH 8.0) was equilibrated against a reservoir solution containing 46% saturated ammonium sulfate, 0.04% (w/v) NaN 3 , and 100 m M Tris/HCl buffer (pH 8.0). When the peptide dissolved in methanol was added to the drop containing the protein under the above co nditions, the solution became turbid, but it became transparent again when left overnight. The crystals of the free K163T mutant protein were grown from a drop c ontaining 9.5 mgÆmL )1 protein, 38% (w/v) saturated ammonium sulfate, 0.04% (w/v) NaN 3 ,and80m M Tris/HCl buffer (pH 8 .0) equilibrated against a reservoir solution of 46% (w/v) saturated ammonium sulfate, 0.04% (w/v) NaN 3 ,and 100 m M Tris/HCl buffer (pH 8.0). Intensity data for crystals were collected using a Weis- senberg camera for macromolecules [19] installed on the beam line BL6A of the Photon Factory (PF) at Tsukuba, Japan. Data sets were reduced using DENZO and SCALEPACK [20]. The structure for the complex of E. coli CyPB K163T mutant bound to tripeptide was solved b y m olecular replacement with XPLOR [21] using E. coli CyPA [16] as a search model. The model was built using O software [22], and refinement was assisted by molecular dynamics using XPLOR . The structure of the complex with t he tetrapeptide was started from the final structure of the complex with the tripeptide. The crystal structure of the free E. coli CyPB Ó FEBS 2004 E. coli cyclophilin B/distorted trans-isomer (Eur. J. Biochem. 271) 3795 K163T was solved by m olecular replacement using a model of the CyPB K163T protein of the complexes, and the final refinements w ere made using CNS [23]. The model buildings on the basis of electron density maps revealed that in all three kinds of crystals CyPB proteins do not have 24 residues of s ignal sequence at the N -terminal and were cut off in the stage of their cultivation or purification. After refinements of the model containing two proteins and water molecules for two complexes, model building of the peptide was tried. Both of these complexes have an asymmetric unit containing two CyPB molecules and one peptide. The model of distorted trans-form gave good fitting into the electron densities in two complexes. The final results are summarized in Table 1. Results Structure of the K163T mutant of E. coli CyPB We constructed a mutant that satisfies the following two conditions. Firstly, the mutant residue must not disturb the folding of the overall skeleton and secondly it must not affect the binding of peptides containing a proline. A mutant, K163T, in w hich lysine was r eplaced by threonine at residue 163, was selected under these criteria, as Lys163 is located in b-strand and has no intramolecular interaction with any residue s. The k cat /K m value of the K163T mutant (1.9 · 10 7 M )1 Æs )1 )inthecis-totrans- isomerization reaction for Suc-Ala-Ala-Pro-Phe-AMC was almost the same as that of the wild-type protein (1.8 · 10 7 M )1 Æs )1 ). Crystals of complexes of mutant K163T CyPB molecules, and tripeptide (Suc-Ala-trans- Pro-Ala-pNA) or tetrapeptide (Ac-Ala-Ala-trans-Pro-Ala- AMC), g rew f rom a solution containing ammonium sulphate under conditions almost identical to those under which crystals of a complex of E. coli CyPA and tripeptide (Suc-Ala-cis-Pro-Ala-pNA) grew. The c rystals of free K163T CyPB also grew from a protein drop containing no methanol, which was used for dissolving peptides. The structural alignments of E. coli CyPB, E. col i CyPA [16] and human CyPA [6,14] are shown in Fig. 1. The CyPB molecules, as well as the E. coli and human CyPA molecules, have a b-barrel structure, consisting of upper and lower b-sheets of four antiparallel b-strands enclosed by two a-h elices at the top and bottom (Fig. 2). In both complexes, CyPB molecule A, i.e. the molecule without a peptide, is packed along one 3 2 axis, whereas CyPB molecule B, which is bound to a peptide, is packed along the other 3 2 axis. CyPB molecules A and B exist in a ratio of 1 : 1 in the crystal. Molecules A and B, related by a local quasi-twofold rotation axis, make contacts through the side chains of Ile159, Ser161, Thr163 (mutant residue) and Leu165 on the b8 strand, and through Leu8 and Thr10 side chains on the b1 strand. In the crystal of free CyPB, a crystallographic twofold rotation axis is observed, and the symmetry of the space group increases from P3 2 , which is found in the Table 1. Data collection and refinement statistics f or E. coli CyPB mutant K163T. Values in parentheses are for the highest resolution shell. R merge ¼ S h S i |I h,i ) <I> h |/S h S i I h,i ,where<I> h is the mea n in tensity. R factor ¼ S||Fo|-|Fc||/S|Fo|. A subset of the data ( 10%) was excluded from therefinementandusedtocalculateR free . Peptide Suc-Ala-Pro-Ala-pNA Ac-Ala-Ala-Pro-Ala-AMC none Space group P3 2 P3 2 P3 2 21 a ¼ ,b¼ (A ˚ ) 79.28 78.73 82.55 c ¼ (A ˚ ) 56.63 56.16 52.35 Z66 6 Data collection Number of used crystals 2 2 2 Maximum resolution (A ˚ ) 1.7 (1.76–1.70) 1.8 (1.86–1.80) 1.8 (1.86–1.80) Measured reflections 187361 263488 133992 Unique reflections 40355 35961 19219 Completeness (%) 92.0 (73.3) 99.4 (97.5) 99.0 (96.5) I/rI 16.7 9.3 14.8 R merge (%) 9.0 (23.8) 9.0 (37.1) 6.9 (40.2) Refinement Resolution (A ˚ ) 6–1.7 (1.78–1.70) 6–1.8 (1.88–1.80) 8–1.8 (1.82–1.80) Used reflections 39354 34891 18959 Protein atoms 2497 2481 1273 Peptide atoms 35 39 none Solvent atoms 185 204 81 R factor (%) 18.4 (27.0) 18.8 (26.2) 20.1 (28.7) R free (%) 22.1 (29.0) 22.9 (28.4) 22.5 (31.1) Root-mean-square deviations from ideal values Bond lengths (A ˚ ) 0.006 0.008 0.006 Bond angles (°) 0.874 1.38 0.842 Average B factors (A ˚ 2 ) Main chain 17.35 13.46 18.60 Side chain 20.24 15.53 21.11 Solvent 37.24 34.95 30.38 Peptide 39.82 34.26 3796 M. Konno et al.(Eur. J. Biochem. 271) Ó FEBS 2004 complexes, to P3 2 21. In the region from 144 to 151 residues of the molecule in the crystal of free CyPB (Fig. 3; shown in green), the turn structure is broken by the strong interaction with the adjacent molecule. On the other h and, the CyPB molecules A and B (Fig. 3; red and blue, respectively) of the complexes possess the T5 turn of Type II¢ and E. coli CyPA molecule has also the similar Type II¢ turn [16]. Short peptides containing a trans -form proline bound to E. coli CyPB A left-to-right-running cleft is shown in the upper b-sheet of the b3, b4, b6andb5 strands of the CyPB proteins (Fig. 2). In the center of the cleft (Fig. 4), there is a hydrophobic pocket formed by the side chains of Phe53, Met54, Phe112, Leu113 and Tyr122, and by the side chain of Phe104 at the bottom. Mainly hydrophilic residue s occupy the l eft half of the cleft, whereas mainly hydrophobic residues occupy the right half. The tripeptide Suc-Ala1-Pro2-Ala3-pNA and the tetrapeptide Ac-Ala1-Ala2-Pro3-Ala4-AMC have the distorted tra ns-form, allowing them to bury deeply into the cleft of CyPB m olecules B. In t he tripeptide, the //w torsional angles of the main chains are )170/100°, )56/118° and )86/138° for Ala1, Pro2 and Ala3, respectively. In the tetrapeptide the se angles are )177/91°, )52/ 122° and )79/ 132° for Ala2, Pro3 and Ala4, respectively. The Ala-Pro-Ala segments of the tripeptide and tetrapeptide occupy the exactly same positions (Fig. 4), when 166 Ca atoms of CyPB molecules B in the two complexes were superimposed. The / and w angles of the Ala residue of Ala-Pro are rarely observed for residues of linear short peptides containing no glycine. The Pro2 and Pro3 residues of the tripeptide and tetrapeptide have an envelope form in which their Cc Fig. 1. Structural alignments of E. coli CyPB, E. coli CyPA [16], and human CyPA [6,14]. Pale blue boxes: b-st rands; pink: a-helices; orange: 3 10 helices; and blue: turns. Amino acid residues are shown usin g a one-lette r code and dots indicate deletions. The 14 conserved residues of the region, in which the peptides areplaced,arewritteninboldtype.Asterisks are shown in 10-residue intervals of E. coli CyPB. In E. coli CyPB K163T mutant, a Lys at residue 163 is replaced by a Thr. Fig. 2. The ribbon model of the b-barrel structure of E. coli CyPB consisting of the upper and the lower b-sheets enclosed by two helices. The colors of ribbon are shown corresponding to those of Fig. 1. The loop colored in green is the region expected to affect the selection of the substrate. T hr163 is located outside of b8 strand. The Suc-Ala-trans- Pro-Ala-pNA is also shown by ball-and-sticks model. Figures 2,3,4,5,6 and 7 were prepared using the programs MOLSCRIPT [35] and RASTER 3 D [36]. Fig. 3. Superimposed traces. The superim- posed traces of Ca of CyPB molecule A (red) without a peptide and CyPB molecule B (blue) bound by a tripeptide in the complexes, and CyPB molecule (green) in the crystal of free CyPB. The molecules shown in Figs 2 and 3 are in the same orientation. Ó FEBS 2004 E. coli cyclophilin B/distorted trans-isomer (Eur. J. Biochem. 271) 3797 atoms lie both above the plane of the coplanar Cb-Ca-N-Cd as shown in Fig. 5 so that their tra ns-proline rings do not collide with p electron over the ring of Phe104 at the bottom of the pocket. This envelope form is also observed in cis-Pro2 of the tripeptide bound to E. coli CyPA and in trans-Pro90 of the loop of the capsid protein bou nd to human CyPA but does not belong to any of t hree conformations shown in a side-chain rotamer library [24]. The Cc atom of most Pro residues lies down the plane of the coplanar Cb-Ca-N-Cd as observed in Pro51 (w ¼ 141 °), Pro71 (w ¼ 149°), Pro153 (w ¼ 133°) and Pro156 (w ¼ 137°) in the CyPB molecule. There is often another envelope form observed, in which the Cb atom lies down the plane of the coplanar Ca-N-Cd-Cc as observed in Pro69 (w ¼ 176°), Pro72 (w ¼ 161°) and Pro148 (w ¼ )6°). In the tripeptide (tetrapeptide) (Fig. 4), CO and NH of the amide bond of Ala-Pro form no hydrogen bond with the CyPB molecule, and CO of the amide bond of suc-Ala1 (Ala1-Ala2) forms a hydrogen bond with a distance of 2.77 A ˚ (2.59 A ˚ ) t o a water molecule, which forms a hydrogen bond with His47 N e2 with a distance of 2.87 A ˚ (2.78 A ˚ ). The carbonyl o xygen atom of Pro2 of the tripeptide forms a hydrogen bond to the N g2 atom o f the guanidyl group of Arg48 wit h t he d istance o f 2 .62 A ˚ ,and 2.72 A ˚ for t hat o f Pro3 of the tetrapeptide. B ecause the w angles of Pro2 (w ¼ 118 °)ofthetripeptideandPro3(w ¼ 122°) of the tetrapeptide deviate from w angles of 130–150° observed for trans-Pro residues, it has became possible to form a hydrogen bond to the side chain of Arg48. The aromatic ring of p-nitroanilide in the C-terminal end of the tripeptide overlaps parallel to the ring of Phe112 and the same is observed for that of coumarin of the tetrapeptide. Thus the binding of peptides to the cleft of CyPB is mainly due to the hydrophobic interaction. In CyPB molecules B bound to the tripeptide and tetrapeptide (Fig. 4) , t he Arg48 N e atom forms a hydrogen bond to the Gln56 O e1 atom with the distance of 2.85 A ˚ for the tripep tide, and 2.94 A ˚ for t he tetrapeptide. The Gln56 N e2 atom forms a hydrogen bond to the Gln102 O e1 atom with the distance of 2.88 A ˚ for the tripeptide, and 2.74 A ˚ for the tetrapeptide. On the other hand, in the CyPB molecules A without peptide and the CyPB molecule in the crystal of free CyPB, the conformations of the side chains of Arg48 differ from those of t he CyPB molecules B bound to a peptide, and no hydrogen bond between Arg48 N e and Gln56 O e1 is formed, whereas these Gln56 N e2 atoms form a hydrogen bond to Gln102 O e1 . Comparison between peptides containing trans -Pro bound to CyPB and the loop containing Gly- trans -Pro of the capsid protein bound to human CyPA The c rystal structure of the human CyPA complex [6] revealed that the Ala88-Gly89-trans-Pro90-Ile91-Ala92- Pro93-Gly94 region of t he most mobile loop in the N-terminal domain of the HIV-1 capsid protein is bound to the cleft on the u pper b-sheet. When the cores of E. coli CyPB and human CyPA were superimposed, the position of trans-Pro90 of the capsid protein displaces by half of a ring from that of Pro3 of the tetrapeptide as shown in Fig. 6 . The main chain of the glycine residue immediately preceding Pro90 makes the torsional angles of / ¼ 149° and w ¼ 158°, which are in t he regions in which o ther amino acids with side chains are disallowed due to steric hindrance. The plane of Ca, C and O of Gly89 has the rotation angle x of  20° from the plane of N, Ca and Cd of Pro90 around the Gly-Pro amide bond. The //w torsional angles for Pro90 are )78/141° and the ring of Pro90 is inserted into the hydrophobic pocket. Pro90 C O forms hydrogen bonds to Arg55 N g1 and N g2 atoms with distances of 2.67 A ˚ and 2.91 A ˚ ,andtheArg55N e atom does not form a hydrogen Fig. 4. A stereo view of Suc-Ala-trans-Pro-Ala-pNA (green) and Ac-Ala-Ala- trans-Pro-Ala-AMC (y ellow) b ound to superimposed E. coli CyPB molecules. The hydrogen bonds are shown in broken lines. The CyPB molecules shown in Figs 4, 6 and 7 were rotated by 45° around the horizontal axis from those shown in Figs 2 and 3. Fig. 5. The model of the e nvelope form of the proline ring. CO-Ala2- Pro3-Ala4 portion of the tetrapeptide is shown by ball-and-sticks model. 3798 M. Konno et al.(Eur. J. Biochem. 271) Ó FEBS 2004 bond. On the other hand, in E. coli CyPB molecules B the side chain of Arg48 is fixed by formation of a hydrogen bond between the Arg48 N e atom and the Gln56 O e1 atom and Pro CO of the tripeptide and tetrapeptide forms a hydrogen bond only to Arg48 N g2 . The difference of the hydrogen bond formation to the side chain of the Arg residue of E. coli CyPB and human CyPA indicates that these hydrogen bonds are f ormed after the Pro r esidue of peptides is fixed in the hydrophobic pocket. A hydrogen bond is formed between NH of Ala88 and CO of Gly72 of human CyPA with a distance of 2.84 A ˚ . T he fixing by hydrogen bonds of Ala88 and Pro90 to CyPA generates the deviation of x angle of  20° from the plane around the Gly-Pro amide bond and the distortion of the torsional angle of Gly89. This fixing and hydrophobic interaction agree with the finding that the capsid protein binds rigidly to human CyPA; even i n the presence of a high concentration of salt or detergent, the binding between them was detected [25]. This is the most mobile loop of the capsid protein, and it will bind to the human CyPA molecule in conjunction with a conformational change in the loop. The conforma- tion of acetyl-Ala1 of the tetrapeptide shows a big deviation from that of His87-Ala88 of the capsid protein. Comparison between E. coli CyPB with a peptide of the trans -proline form and E. coli CyPA with a peptide of the cis -proline form Crystals were obtained in which Suc-Ala-trans-Pro-Ala- pNA is bound on E. coli CyPB (Fig. 7 A), whereas c rystals were reported [16] in which Suc-Ala-cis-Pro-Ala-pNA is bound on E. coli CyPA (Fig. 7 B). In the cis-form of the tripeptide binding in the cleft of E. coli CyPA, torsional angles //w for Ala1, Pro2 and Ala3 are )119/146°, )69/143° and )67/146°, respectively. The cis-proline ring of Pro2 is also inserted into the hydrophobic pocket, and CO of Pro2 forms a hydrogen bond with N g2 of the guanidyl g roup of Arg43, whereas the succinyl group in the N-terminus protrudes from the cleft, and NH and CO of Ala1 form hydrogen bonds with CO and NH of Arg87. The p-nitro- anilide ring in the C-terminal end lies over the side chain of the Ile45 residue, parallel with the ring of Phe48. Because all E. coli CyPA molecules are bound to the peptide with cis-proline in solution, crystals obtained consist only of complexes. On the other hand, in crystals of E. coli CyPB analyzed in this study, the CyPB molecule A without peptide and the CyPB molecule B with a peptide exist in the ratio of 1 : 1. The findings that the peptides of the distorted trans-proline form occupy half of the E. coli CyPB mole- cules can be attributed to the fact that the binding affinity of peptides of the distorted trans-proline form for CyPB is smaller than that of peptides of the cis-proline form for CyPA. Because the concentration of complexes with trans- peptide may be smaller than that o f free p roteins, crystals consisting of only complexes were not obtained but crystals consisting of the complex and t he free protein, the ratio of which is 1 : 1 due to crystal contact, were obtained. The fact that crystals of these complexes of E. coli CyPB and CyPA grew under the same conditions also indicates that the binding affinity of the trans-form of the tripeptide for CyPB molecule is higher than that for CyPA, whereas the binding affinity of the cis-form i s higher for CyPA to the contrary. We were unable to identify any difference of conformation in 13 of the 14 conserved residues (CyPB/CyPA; His47/42, Arg48/43, Ile50/45, Phe53/48, Met54 /49, Gln56/51, Ala91/ 86, Arg92/87, Thr100/95, Gln102/97, Phe104/99, Phe112/ 107, Leu113/108 and Tyr122/120) of the region in which the tripeptide is placed (Fig. 7), whereas the nonsubstantial difference betwe en the CH 3 -group of Met54 in CyPB and that of Met49 in CyPA r eflects only the presence of steric hindrance in the case of the complex o f CyPB and the distorted trans-form. On the other hand, we found that the different forms of b inding peptides are generated due to the difference in areas close t o t he peptide-binding region, i.e. the loop continuing from b5 and the loop connecting the b4andb5 strands (Figs 2 and 7). Such regions are expected to determine the orientation of the substrate at the P2 site, i.e. the second residue of the N-terminal side from proline in the isomerization reaction of refolding proteins as they proceed from the trans-form t o the cis-form. The P2 site is responsible for the difference in the rate of the isomerization reaction accelerated by CyPA and CyPB molecules. The T4 turns in t he loop between the b5andb6 strands consist o f Asp95, Lys96, Asp97 and Ser98 in CyPB and Ala90, Pro91, Fig. 6. Comparison between a loop (His87-Ala88-Gly89-trans-Pro90-Ile91-Ala92-Pro93-Gly94) of the HIV-1 capsid protein bound to human CyPA [6] (PDB code 1AK4) (yellow) and a distorted tetrapeptide Ac-Ala-Ala-trans-Pro-Ala-AMC bound to E. co li CyPB (green). The cores of the proteins were superimposed. Residues of E. coli CyPB and human CyP A are shown in gree n and yellow with residue numbers (E. coli CyPB/human CyPA), respectively. Ó FEBS 2004 E. coli cyclophilin B/distorted trans-isomer (Eur. J. Biochem. 271) 3799 His92 and Ser93 in CyPA. The difference in these side chains will generate the difference in conformations of the side chains of Thr93 in CyPB and Thr88 in CyPA. The side chains of Thr93 and Ala94 in CyPB and those of Thr88 and Gln89 in CyPA will come in contact w ith t he su bstrate at the P2 s ite. The part o f the loop connecting the b4andb5 strands, i.e. L ys68, P ro69, A sn70 and Pro71 in CyPB and Ala63, Thr64, Lys65 and Glu66 in CyPA occupies the bottom left side of the cleft and is close to the peptide- binding region. The difference of hydrophobicity of these areas between the CyPB and CyPA molecules is expected to affect the selection of the substrate. Discussion Short peptides with a trans -proline are distorted to come to complex with CyPB CyP proteins accelerate the isomerization reaction f rom cis- to trans-form of Suc-Ala-Xaa-cis-Pro-Phe-pNA (Xaa stands for any amino acid residue) [5,17,18,26–29] and shorter peptides such as Ala-Pro, Ala-Pro-Phe, Ala-Ala- Pro-Ala and Ala-Ala-Pro-Ala-Ala i nhibit cis to trans isomerase activity of calf thymus CyPA [27]. It was suggested that in the mechanism of the observed isomeri- zation reaction from cis-totrans-form on CyP proteins, the N atom of Pro receives a proton from the guanidyl group of Arg55 on the b3 strand for human CyPA [17]. The proton transfer to the prolyl N atom results in weakening the double bond character of the amide bond of Xaa-Pro. The double bond character of this C-N amide bond is due to the hyperconjugation between the p orbital on the sp 2 - hybrid type N atom of a proline and the p orbital system CO of Xaa preceding Pro. The N ato m of a proline with a proton transferred thereto has an sp 3 -hybrid rather than sp 2 -hybrid orbital; this change in electronic configuratio n is essential to the transient state of cis to trans isomerization pathway of short peptides catalyzed by CyP proteins. As the mutation of Arg t o Ala in human CyPA loses the above- mentioned pathway via the proton transfer from the guanidyl group of Arg to the prolyl N atom, the mutant wasreportedtoretainlessthan1%ofwildtypecatalytic activity [30]. The local energy d iagram of the rotating C-N bond of Xaa-Pro for two distinct pathways contributing to cis to tra ns isomerization o f short peptides, which explains the enzymatic activity difference between the wild-type and the mutant, is illustrated in Fig. 8A. If the isomerization from a cis-totrans-isomer occurs by rotating around the Xaa-Pro amide bond while preserving the planarity of three bonds around the N atom of the Pro residue, the energy barrier DE 1 for the rotation corresponds to loss of stability of energy due to the hyperconjugation of the Xaa-Pro amide bond having sp 2 -hybrid configuration on the N a tom. On the other hand, if the peptide rotates after the N atom of Pro in the C-N amide bond of Xaa-Pro accepts a proton to be convertedintoansp 3 -hybrid c onfiguration, the barrier for Fig. 7. The stereo views of clefts of (A) E. coli CyPB complexed w ith Suc-Ala-trans-Pro-Ala-pNA and ( B) E. coli CyPAcomplexedwithaSuc-Ala-cis- Pro-Ala-pNA [ 16] (PDB code 1LOP). These were viewed from the same direc tion using coordination of CyPA superimposed on CyPB. The hydrogen bonds are shown in broken lines. The conserved residues of the region, in which the tripeptide are placed, are shown in light gray, and the residues, which is expected to affect the selection of th e substrate , are shown in green. 3800 M. Konno et al.(Eur. J. Biochem. 271) Ó FEBS 2004 the rotation due to only steric hindrance is low (DE 2 ), as sp 3 - hybrid configuration on the N atom has no interaction with CO of Xaa. For the reverse conversion from a trans-isomer having the sp 2 -hybrid configuration on the N atom of Pro to a cis-isomer having also the sp 2 -hybrid configuration, in addition to DE 1 , a difference in the energy between these states, DE 3 , is also required, and the activation e nergy for short peptides is too high for the reverse conversion to occur. However, in the observed complex of E. coli CyPA with Suc-Ala-cis-Pro-Ala-pNA (Fig. 7B), CO of cis-Pro2 forms a hydrogen bond with the distance of 2.75 A ˚ to the N g2 atom of the guanidyl group of the Arg43 on the b3 strand, but the N atom of Pro2 is 4.02 A ˚ away from the N g2 atom of Arg43. T he N g1 atom of Arg43 i s not within 5.5 A ˚ from the N atom of Pro2. In addition, NH and CO of Ala1 of the cis-form have hydrogen bonds with the CO and NH of Arg87, respectively. The amide bond of Ala-cis-Pro is planar and t he N atom of Pro has the sp 2 -hybrid configuration. In E. coli CyPB molecules B bound to Suc-Ala-trans-Pro- Ala-pNA and Ac-Ala-Ala-trans-Pro-Ala-AMC (Figs 4 and 7A), the N g2 atom of Arg48 forms a hydrogen bond with the distance of 2.62 A ˚ and 2.72 A ˚ to CO of trans-Pro of the tripeptide and tetrapeptide, respectively. In these complexes, the distance of the N atom of Pro and the N g2 atom of Arg48 is 3.08 A ˚ and 3.02 A ˚ , and the angle between the N g2 - N vector and normal vector to the amide bond plane of Ala- Pro is 16° and 17°, respectively. The N g1 atom of Arg48 i s not within 5.0 A ˚ from the N atom of Pro. For both peptides, the amide bond of Ala-trans-Pro is planar and the NatomofProhasansp 2 -hybrid configuration. In conclusion, the proton transfer from the guanidyl group of Arg to the prolyl N atom no longer occurs in the complex of E. coli CyPA, as hydrogen bonding to CO of Ala1 gives rise to the increased double bond character of the Ala1-Pro2 amide bond and the guanidyl group of Arg43 is far from the N atom of Pro2. In the case of the observed complex of CyPB, because the hydrogen bond between the guanidyl group of Arg48 and CO of Pro plays an essential role in the binding of the peptide, it is impossible for the guanidyl group of Arg48 to be involved in proton transfer to the prolyl N atom. These facts show that in both cases the enzyme cannot carry o ut the isomerization of these peptides observed in crystals of complexes. Similarly, a crystal structure [15] of the tetrapeptide Ac-Ala-Ala-cis-Pro-Phe- pNA staying in unreacted form on human CyPA also demonstrated that Arg55 forms a hydrogen bond with the C O of cis-Pro of the tetrapeptide, and that this peptide retains still planarity of the Ala-cis-Pro amide bond. In the case of CyPB in complexes with Suc-Ala-trans-Pro- Ala-pNA and Ac-Ala-Ala-trans-Pro-Ala-AMC, the large distortion in the orientation of Ala1 and Ala2 of Ala-Pro with //w torsional angles of )170°/100° and )177°/91° allows the fixing of CO of Suc and Ala1 to His47 N e2 via a mediating water molecule. T he fixing t o Arg48 and His47 makes a major contribution to binding of peptides in the trans-form. In contrast, as for CyPA in complex w ith Suc- Ala-cis-Pro-Ala-pNA, two hydrogen bonds of NH and CO of Ala1 are formed, and the orientation of Ala1 has a little distortion. When CO of Xaa preceding Pro has no formation of a hydrogen bond, the double bond character of this C-N amide bond is due only to the hyperconjugation. Proton transfer to the prolyl N atom results in weakening the double bo nd c haracter o f the amide bond of Xaa-Pro. As result, the N atom of a proline with a proton transferred thereto has an sp 3 -hybrid rather than an sp 2 -hybrid orbital, and thus the planar configuration of the N atom of proline is converted to an ammonium type configuration. As the contribution of the hyperconjugation into the stability by sp 2 -hybrid orbital on the N atom of Pro is lost owing to the small rotation around the amide bon d of Xaa-Pro, the process for the conversion from sp 2 -tosp 3 -hybrid confi- guration is enhanced. The a bove mentioned mechanism for short peptides, where t he conversion from sp 2 -tosp 3 -hybrid configuration has a critical contribution to in vitro c is to trans peptidyl- prolyl isomerization activity of CyP proteins, may be extended to understand the in vitro trans to cis peptidyl- Fig. 8. The local energy diagram of the rotating C-N bond of Xaa-Pro. (A) Two distinct pathways contributing to cis to trans isomerization of short peptides. (B) Two distinct pathways contributing to trans to cis isomerization of unfolded proteins, which possess the cis-proline conformation in the natural folding state. Ó FEBS 2004 E. coli cyclophilin B/distorted trans-isomer (Eur. J. Biochem. 271) 3801 prolyl isomerization in volved in t he reported e nhancement by CyP proteins in the refolding process of unfolded proteins (e.g. RNase A, carbonic anhydrase I I, and RNase T 1 proteins [7–11]), which possess a cis-proline conformation in the natural folding state. In the in vitro refolding experiment without CyP proteins, the in vitro trans to cis peptidyl-prolyl isomerization was observed to progress slowly, which may be attributed to such a d irect path requiring larger activation energy DE 5 where the trans- isomer with the sp 2 -hybrid configuration o n the N atom of Pro is directly isomerized to the cis-isomer with the sp 2 - hybrid configuration (Fig. 8B). On the other hand, in the in vitro system with CyP proteins, the in vitro trans to cis peptidyl-prolyl isomerization was enhanced remarkably, which may be reasonably understood to be due to such an indirect path requiring a smaller activation energy DE 4 .The conversion from the sp 2 -tothesp 3 -hybrid configuration on the N atom of Pro of the trans-isomer is enzymatically conducted by means of CyP proteins, being followed by the quick isomerization from the trans-isomer with the sp 3 - hybrid configuration to the cis-isomer with the sp 3 -hybrid configuration. At the stage of the enzymatic conversion from the sp 2 -tothesp 3 -hybrid configuration on the N atom of Pro of the trans-isomer, the guanidyl group of Arg should be used only for the proton t ransfer to t he N atom o f Pro rather than the hydrogen-bonding to the CO o f Pro so that the guanidyl group of Arg may be positioned much closer to the N atom of Pro than that observed in CyPB in complex with Suc-Ala-trans-Pro-Ala-pNA or Ac-Ala-Ala-trans-Pro- Ala-AMC analyzed here. Thus, in place of hydrogen bonding of the guanidyl group of Arg to CO of Pro, another amino acid residue located somewhat behind the Pro concerned is predicted to be involved in the binding of unfolded proteins to CyP proteins. In such a predicted binding form, the proline containing the peptide portion of the unfolded proteins embedded in the cleft of CyP proteins may have not only the distortion of Xaa of Xaa-Pro but also additional distortion in the C-terminal region next to the Pro. These distortions may be effectively used as driving force for quick isomerization from the trans-isomer with the sp 3 -hybrid configuration to the cis-isomer with the sp 3 - hybrid configuration in the enhanced refolding process of unfolded proteins observed in the the in vitro system with CyP proteins. It has been reported that strains of the yeast Saccharo- myces cerevisiae in which all eight identified CyP family genes w ere d isrupted survived [31–33]. Therefore, CyPs appear to be irrelevant to the in vivo folding process for native proteins that possess a cis-proline c onformation. As explained above, CyPA and CyPB may have an advanta- geous potential for binding distorted peptide portions of partially unfolded proteins in its cleft. If such a distorted peptide portion of a partially unfolded protein resulting from extrinsic c auses (for e xample, heat shock) is bound in the cleft of CyPA or CyPB protein, further progress of the protein denaturation induced by the extrinsic causes would be successfully blocked. In such a case, when the e xtrinsic cause is r emoved from the partially unfolded proteins held in the CyPA or CyPB protein, successful refolding of this is achieved to make a quick recovery from damage due to the extrinsic causes. Such a possible f unction of CyPs to block the extensive denaturing course of proteins promoted by extrinsic c auses m ay provide a more probable explanation for previous reports [34] that yeast strains lacking CyPA and CyPB are sensitive to heat shock, and that both of these proteins facilitate the survival of cells exposed to high temperatures. Protein Data Bank access codes Coordinates of the structures have been deposited in the Protein Data Bank (accession codes 1V9T and 1VAI for the two kinds of complexes of the E. coli CyPB K163T mutant bound by Suc-Ala-trans-Pro-Ala-pNA and Ac-Ala-Ala- trans-Pro-Ala-AMC, a nd accession code 1J2A for the E. coli CyPB K163T mutant). Acknowledgements We thank Dr Mamoru Suzuki and Dr Noriyoshi Sakabe at the High Energy Accelerator Research Organization, KEK, for their help in the data collection. This work was su pported in part by a grant from the N ational Project on Pro tein Structural and Func tional Analyses to M. K. References 1. Harding, M.W., Handschumacher, R.E. & Speicher, D.W. (1986) Isolation and amino acid sequence of cyclophilin. J. Biol. Chem. 261, 8547–8555. 2. Galat, A. (1993) Peptidylproline cis-trans-isomerases: immuno- philins. Eur. J. Biochem. 216, 689–707. 3. Hunter, T. (1998) Prolyl isomerases and nuclear function. Cell 92, 141–143. 4. Liu, J . & Walsh, C.T. ( 1990) Peptidyl-prolyl cis-trans -isomerase from Escherichia coli: a periplasmic homolog of cyclophilin that is not inh ibited by cyclosporin A. Proc. Natl Acad. Sci. USA 87, 4028–4032. 5. Hayano, T., Takahashi, N., Kato, S., Maki, N. & Suzuki, M. (1991) Two distinct forms of peptidylprolyl-cis-tarns-isomerase are expressed separately in periplasmic and cytoplasmic compart- ments of Escherichia coli cells. Biochem. 30, 3041–3048. 6. Gamble,T.R.,Vajdos,F.F.,Yoo,S.,Worthylake,D.K.,Hous- eweart, M., Sundquist, W.I. & Hill, C.P. (1996) Crystal structure of human cyclophilin A bound to the amino-terminal domain of HIV-1 capsid. Cell 87, 1285–1294. 7. Lang, K., Schmid, F.X. & Fischer, G. (1987) Catalysis of protein folding by prolyl isomerase. Na ture 329, 268–270. 8. Scho ¨ nbrunner, E.R., Mayer, S., Tropschug, M., F ischer, G., Takahashi, N. & Schmid, F.X. (1991) Catalysis of protein folding by cyclophilins from different species. J. Biol. Chem. 266, 3630–3635. 9. Freskga ˚ rd, P O., Bergenhem, N., Jonsson, B H., Svensson, M. & Carlsson, U. (1992) Isomerase and chaperone activity of prolyl isomerase in the folding of carbonic anhydrase. Science 258, 466–468. 10. Fransson, C., Freskga ˚ rd, P O., Herbertsson, H., Johansson, A ˚ ., Jonasson, P., Ma ˚ rtensson, L G., Svensson, M., Jonsson, B H. & Carlsson, U. (1992) Cis-trans isomerization is rate- determining in the reactivation of denatured human carbonic anhydrase II as evidenced by proline isomerase. FE BS Lett. 296, 90–94. 11. Schmid, F.X. (1993) Prolyl isomerase: Enzymatic catalysis of slow protein-folding reactions. Annu. Rev. Biophys. Biomol. Struct. 22, 123–143. 12. Kallen, J., Spitzfaden, C., Zurini, M.G.M., Wider, G., Widmer, H., Wu ¨ thrich,K.&Walkinshaw,M.D. (1991) S tructure of 3802 M. Konno et al.(Eur. J. Biochem. 271) Ó FEBS 2004 human cyclophilin and its binding site for cyclosporin A determined by X-ray crystallography and NMR spectroscopy. Nature 353, 276–279. 13. Kallen, J. & Walkinshaw, M.D. (1992) The X-ray structure of a tetrapeptide bound to the active site of human cyclophilin A. FEBS Lett. 300, 286–290. 14. Ke, H., Mayrose, D. & Cao, W. (1993) Crystal structure of cyclophilin A complexed with substrate Ala-Pro suggests a solvent-assisted mechanism o f cis-trans isomerization. Proc. Natl Acad. Sci. USA 90, 3324–3328. 15. Zhao, Y. & Ke, H. (1996) Crystal structure implies that cyclophilin predominantly catalyzes the trans to cis isomerization. Biochem. 35, 7356–7361. 16. Konno, M., Ito, M., Hayano, T. & Takahashi, N. (1996) The substrate-binding site in Escherichia coli cyclophilin A preferably recognizes a cis-proline isomer or highly distorted form of the trans isomer. J. Mol. Biol. 256, 897–908. 17.Eisenmesser,E.Z.,Bosco,D.A.,Akke,M.&Kern,D.(2002) Enzyme dynamics during catalysis. Science 295, 1520–1523. 18. Fischer, G., Bang, H. & Mech, C. (1984) Determination of enzymatic catalysis for the cis-trans-isomerization of peptide binding in proline-containing peptides. Biomed. Biochim. Acta 43, 1101–1111. 19. Sakabe, N. (1991) X-ray diffraction data collection system for modern protein crystallography with a Weissenberg camera and an imaging plate using synchrontron radiation. Nucl. Instrum. Methods Phys. Res. A 303, 448–463. 20. Otwinowski, V. & Minor, W. (1997) Processing X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276,307– 326. 21. Bru ¨ nger, A.T. (1993) X-PLOR, Version 3.1: a System for X-ray crystallography and NMR. Yale University Press, New H aven, CT. 22. Jones, T.A. (1978) A graphics model building and refinement system for macromolecules. J. Appl. Crystallogr. 11, 268–272. 23. Bru ¨ nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J S., Kuszewski, J., Nilges, M., Pannu, N.S., Read, R.J., Rice, L.M., Simonson, T. & Warren, G.L. (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D 54, 905–921. 24. Ponder, J.W. & Richards, F.M. (1987) Tertiary templates for proteins. J. Mol. Biol. 193, 775–791. 25. Luban, J., B ossolt, K.L., Franke, E.K., Kalpana, G.V. & Goff, S.P. (1993) Human imm unodeficiency v irus type 1 g ag prote in binds to cyclophilins A and B. Cell 73, 1067–1078. 26. Liu, J., A lbers, M.W., Chen, C., Schreiber, S.L. & Walsh , C.T. (1990) Cloning, expression, and purification of human cyclophilin in Escherichia coli and assessment of the catalytic role of cysteines by site-directed mutagenesis. Proc. Natl Acad. Sci. USA 87, 2304– 2308. 27. Harrison, R.K. & Stein, R.L. (1990) Mechanistic studies of pep- tidyl prolyl ci s-trans isomerase: Evidence for catalysis by distor- tion. Biochem. 29, 1684–1689. 28. Kofron, J.L., Kuzmie ` ,P.,Kishore,V.,Colo ´ n-Bonilla, E. & Rich, D.H. (1991) Determination of kinetic constants for peptidyl prolyl cis-trans isomerases by an improved spectrophotometric assay. Biochem. 30, 6127–6134. 29. Kakalis, L.T. & Armitage, I.M. (1994) Solution conformation of a cyclophilin-bound proline isomerase sub strate. Biochem. 33, 1495– 1501. 30. Zydowsky, L.D., Etzkorn, F.A., Chang, H.Y., Ferguson, S.B., Stolz, L.A., Ho, S.I. & Walsh, C.T. (1992) Active site mutants o f human c yclophilin A s eparate peptidyl-prolyl isomerase activity from cyclosporin A binding and calcineurin inhibition. Protein Sci. 1, 1092–1099. 31. Haendler, B., K eller, R., Hiestand, P.C., Koc her, H.P., Wegmann, G. & Movva, N.R. (1989) Yeast cyclophilin: isola- tion and characterization for the protein, cDNA and gene. Gene 83, 39–46. 32. Tropschung, M., Barthelmess, I.B. & Neupert, W. (1989) Sensivity to cyclophilin A is mediated by cyclophilin in Neurospora crassa and Saccharomyces cerevisiae. Nature 342, 953–955. 33. Dolinski, K., Muir, S., Cardenas, M. & Heitman, J. (1997) All cyclophilins and FK506 binding proteins are, individually and collectively, dispensable for viability i n Saccharomyces cerevisiae. Proc.NatlAcad.Sci.USA94, 13093–13098. 34. Sykes, K., Gething, M J. & Sambrook, J. (1993) Proline iso- merases fun ctio n du rin g hea t s hock. Proc. Natl Acad. Sci. USA 90, 5853–5857. 35. Kraulis, P.J. (1991) M OLSCRIPT: a program t o produce both detailed a nd schematic plots of protein structures. J. Appl. Crys- tallogr. 24, 946–950. 36. Mettitt, E.A. & Murphy, E.P. (1994) Raster 3d, Version 2.0. A program for p hotorealistic m olecular graphics. Acta Crystallogr. D 50, 869–873. Ó FEBS 2004 E. coli cyclophilin B/distorted trans-isomer (Eur. J. Biochem. 271) 3803 . succinyl-Ala-Pro- Ala-p-nitroanilide (Suc-Ala-Pro-Ala-pNA), and w ith a tetrapeptide, acetyl-Ala-Ala-Pro-Ala-amidomethylcouma- rin (Ac-Ala-Ala-Pro-Ala-AMC), and w e found highly distorted trans-peptides molecules B bound to Suc-Ala-trans-Pro- Ala-pNA and Ac-Ala-Ala-trans-Pro-Ala-AMC (Figs 4 and 7A) , the N g2 atom of Arg48 forms a hydrogen bond with the distance of 2.62 A ˚ and 2.72 A ˚ to CO of trans-Pro. O of cis-Pro of the tetrapeptide, and that this peptide retains still planarity of the Ala-cis-Pro amide bond. In the case of CyPB in complexes with Suc-Ala-trans-Pro- Ala-pNA and Ac-Ala-Ala-trans-Pro-Ala-AMC,