Ssp1, a Site-specific Parvulin Homolog from Neurospora crassa Active in Protein Folding*

Peptidyl-prolyl cis-trans-isomerases (PPIases) are enzymes capable of isomerizing a Xaa-Pro peptide bond. Three families of PPIases are known: cyclophilins, FKBPs, and parvulins. The physiological functions of the PPIases are only poorly understood. Eucaryotic members of the parvulin family have recently been shown to be essential for regulation of mitosis. Here we describe the purification and characterization of Ssp1, an abundant parvulin homolog from Neurospora crassa, which is unique among the known eucaryotic parvulins in containing a polyglutamine stretch between the N-terminal WW domain and the C-terminal PPIase domain. Ssp1 is asite-specific PPIase with respect to the amino acid N-terminal to the proline residue. Peptides with glutamate, phosphoserine, or phosphothreonine in the −1-position proved to be the best substrates. Ssp1 is not only able to isomerize small peptides but is also active in protein folding, as shown with mouse dihydrofolate reductase. Using the substrate specificity of Ssp1, we could identify Glu81-Pro82 as a PPIase-sensitive site in folding of dihydrofolate reductase. These results demonstrate that Ssp1 is a potent mediator of protein folding and that parvulins can serve as tools to elucidate rate-limiting steps in protein folding reactions.

accelerate slow refolding steps in certain proteins in vitro (see Ref. 2). Mitochondrial CyPs from Neurospora crassa (6,7) and yeast (8) have been shown to be part of the protein folding machinery of the organelle involving molecular chaperones (6,7).
The first member of the parvulin family was discovered in Escherichia coli (9 -11). Eucaryotic parvulins were found in yeast (Ess1/Ptf1; Refs. 12 and 13), Drosophila (dodo; Ref. 14), and humans (Pin1; Ref. 15). In contrast to the small E. coli parvulin (10 kDa), which consists only of a PPIase domain, the other members of the eucaryotic parvulin family have an additional WW domain at their N terminus. The WW domain is a structural motif containing two invariant tryptophan residues thought to be involved in protein-protein interactions by binding to short proline-rich segments of target proteins (16,17).
Deletions of ESS1 in yeast are lethal (12). This phenotype can be rescued by expressing either Pin1 or dodo (14,15). Furthermore, overexpression of Pin1 in HeLa cells inhibits G 2 /M transition, whereas depletion of Pin1 induces mitotic arrest (15). These data point to an important function of the eucaryotic parvulins in mitotic control. Ranganathan et al. (18) suggested that substrate recognition of Pin1 is phosphorylation-dependent. Yaffe et al. (19) and Shen et al. (20) indeed showed that Pin1 binds mitotic phosphoproteins containing phosphothreonine-proline or phosphoserine-proline bonds.
Here we describe the purification, cloning, and biochemical analysis of a homolog of Pin1 from the fungus N. crassa, called Ssp1. Ssp1 is unique in containing a polyglutamine stretch between the WW and the PPIase domains. Like Pin1, Ssp1 is a site-specific PPIase, highly preferring acidic residues (glutamate, phosphoserine, or phosphothreonine) amino-terminal to a proline. We show for the first time that an eucaryotic parvulin (Ssp1) is not only active in peptide PPIase assays but also in protein folding. As a substrate, we chose dihydrofolate reductase (DHFR), which is widely used as a model protein in the analysis of intracellular protein folding and trafficking (see e.g. Refs. [21][22][23][24][25][26][27]. Yet, all steps of the complex folding pathway of DHFR are not fully resolved. Using Ssp1 with its site-specific prolyl isomerase activity, we could identify the isomerization of Glu 81 -Pro 82 in mouse DHFR as being rate-limiting in the folding reaction.

EXPERIMENTAL PROCEDURES
Preparation of N. crassa Lysate-Cultures of N. crassa were grown as described (28). Hyphae from 10-liter cultures were frozen at Ϫ80°C for 1 h and then homogenized in a Waring blender for 10 min in the cold room with 2 volumes of lysate buffer (50 mM HEPES/NaOH, pH 7.8; 5 mM dithiothreitol; 1 mM EDTA; Boehringer complete protease inhibitors). Cell debris were pelleted by centrifugation (30 min, 15,000 ϫ g), and the cloudy supernatant was spun again at 100,000 ϫ g for 90 min. The clear cell lysate was withdrawn, frozen in liquid nitrogen, and stored at Ϫ80°C until use. A typical protein yield was 4 -5 mg of protein/ml. * This work was supported by Deutsche Forschungsgemeinschaft Grant SFB 388/B3 and the Fonds der Chemischen Industrie. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AJ006023.
Throughout the procedure, the presence of Ssp1 was monitored by Western blot analysis using a cross-reacting polyclonal antiserum raised against yeast Ptf1 (13).
Amino Acid Analysis and Cloning of Ssp1-Purified Ssp1 was digested in a gel after electrophoresis using protease Lys-C, and resulting peptides were separated according to Ref. 29. Two peptides (see Fig. 2A) were used to design degenerated oligonucleotides for PCRs using the codon usage of N. crassa (30): SSP-A (5Ј-TGG GAG CCI CCI (A/T)(C/G)I GGI AC-3Ј) and SSP-B (5Ј-TT(T/C) TGC ATG TCI CCI C(T/G)I CCG AAG TA-3Ј). PCR reactions using a N. crassa cDNA library made by using a Marathon cDNA kit (CLONTECH) resulted in a Ssp1-specific band, as judged by sequencing. Further PCR reactions using Ssp1specific primers SSP-C (5Ј-AGT CCG ACT GCT CCT CTG-3Ј) and SSP-E (5Ј-TAA TCT CGG ACT CGC GCC A-3Ј) and cDNA adaptor primers AP1 and AP2 (Marathon cDNA kit, CLONTECH) finally yielded a full-length cDNA for Ssp1. For expression of Ssp1, oligonucleotides SSP forward (5Ј-GAA TTC ATG TCC AAC ACC ATC GAG ACC-3Ј) and SSP reverse (5Ј-GGG CTA CTC GAG CCG TTC AAT CAA ATG-3Ј) were used.
All PCRs were performed using Pwo polymerase (Boehringer Mannheim). Three independent full-length clones were sequenced on both strands using a Sequenase kit (Amersham Pharmacia Biotech).
Recombinant Expression and Purification-Ssp1 was expressed as a glutathione S-transferase fusion protein using plasmid pGEX5-1 (Amersham Pharmacia Biotech). Glutathione S-transferase-Ssp1 was purified using glutathione-Sepharose 4B (Amersham Pharmacia Biotech) according to the manufacturer's protocol. The purified fusion protein was cut on the beads with factor Xa (Boehringer) overnight at 25°C. Final purification was done by gel filtration using a Superdex 75 16/60 column (Amersham Pharmacia Biotech). Fractions were concentrated by ultrafiltration using a 10-kDa exclusion membrane (Amicon).
PPIase Assays-PPIase assays were carried out according to Ref. 31. Helper proteases were chymotrypsin (Sigma) at a final concentration of 800 g/ml for peptides ending with -Phe-pNA and trypsin (Boehringer) at a final concentration of 300 g/ml for peptides ending with -Arg-pNA, respectively.
Peptide stock solutions were made in Me 2 SO at a concentration of 8 mM. Assay buffer, peptide substrate, and PPIases were combined to a final volume of 990 l and cooled for 10 min. The assay was started by adding 10 l of the respective protease. A 390 was monitored, and the data were fitted to a first order reaction kinetics using the Origin software. k cat /K m was calculated as Refolding of Mouse DHFR-A cDNA coding for mouse DHFR (32) was cloned into pGEM4 (Promega) as described (6,7). After transcription/translation in a TNT lysate (Promega), [ 35 S]methionine-labeled DHFR was precipitated and denatured by urea as described previously (6,7). Refolding was initiated by adding 10 l of denatured DHFR to 390 l of refolding solution (62.5 mM MOPS/KOH, pH 7.2, 75 mM KCl) with or without added PPIase. At various time points, 40 l were removed and added to 5 l of proteinase K solution (2 mg/ml). The mixture was incubated for 10 min on ice and digestion of unfolded DHFR was stopped with 8 mM phenylmethylsulfonyl fluoride. The samples were precipitated with trichloroacetic acid, analyzed by SDSpolyacrylamide gel electrophoresis and autoradiography, and subjected to quantitation by two-dimensional densitometry.
Construction of DHFR Mutants-DHFR mutants were constructed by using a PCR-based site-directed mutagenesis protocol (Quickchange; Stratagene). To generate the DHFR E81A mutant we used the oligomers 5Ј-AGT AGA GAA CTC AAA GCA CCA CCA CGA GGA GCT-3Ј and 5Ј-AGC TCC TCG TGG TGG TGC TTT GAG TTC TCT  Miscellaneous-CyP20 was purified from N. crassa as described (7).

RESULTS
Purification, Cloning, and Expression of Ssp1-In initial experiments, we detected an abundant cytosolic protein in N. crassa extracts that cross-reacted with a polyclonal antiserum against yeast Ptf1 (13). We set out to purified this protein (Ssp1) from N. crassa hyphae using DEAE-Sepharose chromatography, Affi-Gel Blue chromatography, chromatofocusing, and a final gel filtration step as described under "Experimental Procedures." A typical yield of 4 -5 g of Ssp1 starting with 1 g of total protein was obtained. Pure Ssp1 is a protein with an apparent molecular mass on SDS gels of 23 kDa (Fig. 1, lane 5). We prepared antibodies against Ssp1 in rabbits and by using analytical Western blotting, we measured a Ssp1 content of 0.05-0.1% (using total extracts of N. crassa hyphae).
In order to get sequence information for cloning the cDNA, purified Ssp1 was excised from a Coomassie-stained gel and digested with protease LysC, and the resulting peptides were purified according to Ref. 29. Two internal peptides were sequenced. Peptides 1 and 2 yielded the sequences KTS(Q)WE-PPSGT(D)V(V)K and KXGDLGYFGRGDMQK, respectively (see underline in Fig. 2A; amino acids in parentheses represent the most likely amino acid from high performance liquid chromatography).
Based on the sequences of the peptides and by using the codon usage of N. crassa (30), two degenerated oligonucleotides (SSP-A and SSP-B) were synthesized and used in PCRs with N. crassa cDNA libraries (33)(34)(35). Sequencing of a resulting 300-bp PCR product showed that the deduced amino acid had high homology to the three eucaryotic parvulins known to date (data not shown).
Further PCRs were used to identify and clone a cDNA that contained the entire open reading frame coding for Ssp1.
The deduced amino acid sequence of Ssp1 is shown in Fig.  2A. Ssp1 is a hydrophilic protein with a molecular mass of 20,674 Da (182 amino acids). The calculated isoelectric point is 7.4.
Ssp1 exhibits a high degree of sequence identity to the three other eucaryotic parvulins known to date (Fig. 3). Sequence identities are 50% (Pin1), 45% (Ess1/Ptf1), and 44% (dodo), respectively. In contrast to the other three members of the eucaryotic parvulin family, Ssp1 is unique in containing a polyglutamine stretch consisting of 11 glutamine residues and a proline residue between the conserved WW and PPIase domains, respectively (Fig. 2B). Polyglutamine regions are characteristic regulatory modules being present in various proteins involved in replication, transcription, and cellular regulation (see e.g. Refs. 36 -40).
In order to get high amounts of Ssp1 for biochemical analysis, we expressed a glutathione S-transferase fusion protein in E. coli. Purification of the recombinant protein, cleavage using factor Xa, and purification of recombinant Ssp1 resulted in a pure protein with an apparent molecular mass of 23 kDa comigrating with purified Ssp1 on SDS gels (not shown).
PPIase Activity and Substrate Specificity of N. crassa Ssp1-In order to examine the PPIase activity and the substrate specificity of Ssp1, we carried out in vitro PPIase assays (31) using 10 different peptide substrates varying in the Ϫ1position with regard to the proline (see Table I).
As shown in Table I, Ssp1 is a site-specific PPIase with regard to the Ϫ1-position. Using peptides with His, Leu, Ala, Phe, Gln, and Glu in the Ϫ1-position, respectively, Ssp1 catalyzes only the cis-trans-isomerization of a peptide containing Glu in front of a proline. Exchanging the Glu N-terminal to the proline with basic or neutral amino acids greatly reduces the PPIase activity.
Interestingly, the highest k cat /K m value was found with the peptide substrate Ac-Ala-Ala-Ser(P)-Pro-Arg-pNA, where the Ser in the Ϫ1-position is phosphorylated (Table I). In contrast, the nonphosphorylated peptide is a bad substrate for the enzyme. The same is true for a peptide containing Thr in the Ϫ1-position. However a peptide containing a phosphorylated Thr is a good substrate for Ssp1 (Table I).
We conclude that Ssp1 is a site-specific PPIase, highly preferring acidic residues in the Ϫ1-position relative to the proline and perhaps also in the Ϫ2-position (see below).
Protein Folding Activity of Ssp1-Is Ssp1 able to accelerate the refolding of proteins? Scholz et al. (41) and Dolinski et al. (50) have demonstrated that there may be a difference when measuring PPIase activity with either peptide assays or pro-tein folding activity, respectively. Protein folding activity has been shown to date for members of the cyclophilin and the FKBP families, respectively (6 -8, 35, 43). The small E. coli parvulin, which does not contain a WW domain, is also active in protein folding (42) in vitro; for eucaryotic parvulins, no protein folding activity has been demonstrated to date.
We used mouse DHFR as a model protein for folding. This enzyme is widely used as a model protein in studies of cellular protein maturation and trafficking (e.g. Refs. [21][22][23][24][25][26][27]. The folding pathway of DHFR is complex and not fully understood (44 -46). Prolyl isomerization 2 has been demonstrated to be important for the E. coli enzyme (47).
As shown in Fig. 4, Ssp1 is indeed able to accelerate the refolding of mouse DHFR. It exhibits about the same catalytic efficiency in the refolding assay as an abundant cyclophilin from N. crassa, namely CyP20 ( Fig. 4; see also Refs. 6 and 7). This is the first demonstration that a eucaryotic parvulin (Ssp1) is able to accelerate protein folding in vitro.
Use of the Site Specificity of Ssp1 to Identify a PPIase-sensitive Glu-Pro Site in Mouse DHFR-Taking into account that Ssp1 is very specific with respect to the amino acid in the Ϫ1-position relative to proline (see above), we concluded that Glu-Pro (and perhaps Asp-Pro) bonds should be critical PPIasesensitive sites during folding of mouse DHFR. Mouse DHFR has 13 Xaa-Pro sites (see Fig. 6), which all might be possible candidates for the action of a PPIase. But there is only one acidic (glutamate) residue amino-terminal to a proline in the DHFR sequence (Glu 81 -Pro 82 ; see Fig. 5). Fig. 5 shows a homology alignment of the amino acid sequences of mouse DHFR (32) and E. coli DHFR (48). As can be seen from this figure, Glu 81 -Pro 82 of mouse DHFR aligns with Gln 65 -Pro 66 of E. coli DHFR. Strikingly, cis-trans-isomerization of the Gln 65 -Pro 66 bond in E. coli DHFR was suggested to be important for fast folding of the protein (47). This observation and the site specificity of Ssp1 in the peptide assay (see above) prompted us to test whether the homologous Glu 81 -Pro 82 bond in mouse DHFR is also a critical PPIase-sensitive site for the folding process.
We therefore generated mouse DHFR mutants, namely DHFR E81A and DHFR E81L. The mutant proteins folded efficiently after synthesis in reticulocyte lysates; folding was tested as resistance against proteinase K (6, 7).
We then tested the influence of Ssp1 on the folding of DHFR E81A and DHFR E81L. Despite the fact that Ssp1 is able to accelerate folding of the wild type protein (see above), it is not able to accelerate folding of the respective mutant proteins (Fig. 6, A and B).
Cyclophilin 20 from N. crassa (CyP20) was used as a control to see if the mutations had made the resulting protein insensitive for the action of a PPIase. Cyclophilins are known to have a low substrate specificity regarding the Ϫ1-position (49) . Fig.  6, A and B, shows that CyP20 was able to accelerate folding of the mutant proteins with almost the same refolding efficiency as the wild type protein.
In addition, we constructed mutant DHFR P82A. CyP20 and also Ssp1, surprisingly, are efficiently able to accelerate refolding of this mutant (see Fig. 6C). Uncatalyzed refolding of DHFR P82A is in the same range as wild type DHFR (compare Figs. 4B and 6B).
How can the removal of a proline still lead to effective refolding activity of both PPIases? We suggest that the proline 2 The term "prolyl isomerization" is used throughout this paper for the cis-trans-isomerization of the peptide bond preceding proline in an amino acid sequence. Similarly, the term "prolyl bond" is synonymous with the peptide bond preceding proline. Ϫ1-position denotes an amino acid amino-terminal to a proline in an amino acid sequence. following immediately after Pro 82 (Pro 83 ; see Fig. 5) might also play a role in the folding pathway of mouse DHFR and in catalysis by CyP20 and Ssp1.
In summary, we conclude that (a) the substrate specificity of Ssp1 as determined by standard PPIase assays is not only true for small peptides but also for a protein substrate; (b) Ssp1 can be used as a site-specific PPIase to unravel critical Glu-Pro (and perhaps Asp-Pro) bonds that might be enzymatic targets of cis-trans-isomerization during protein folding; (c) Glu 81 -Pro 82 has an important role in the folding process of mouse DHFR; and (d) Pro 83 in mouse DHFR might also play a role during the folding process; a Glu in a Ϫ2-position relative to a Pro seems to create a Ssp1-sensitive site; however, further experiments are necessary to clarify the roles of Pro 83 and of Glu Ϫ2 .

DISCUSSION
This paper describes the first biochemical isolation of a eucaryotic member of the parvulin family. In addition, we have characterized this novel PPIase from N. crassa. In comparison with the other eucaryotic members of the parvulin family known to date, Ssp1 is unique in two respects. (a) It is an abundant enzyme (0.05-0.1% of total cellular protein), in contrast to Pin1, whose total cellular concentration in HeLa cells FIG. 3. Alignment of the amino acid sequence of Ssp1 with these of other eucaryotic parvulins. Pin1, human Pin1; dodo, D. melanogaster dodo; Ptf1, yeast Ess1/Ptf1. Identical residues are boxed. Note the poly-Q stretch, which is unique to Ssp1.

TABLE I k cat /K m values of peptide substrates
PPIase activity and substrate specificity of Ssp1 are shown. PPIase activity was measured using the peptide assay (31). Ssp1, peptide, and reaction buffer were preincubated in a final volume of 990 l at 10°C for 10 min. The reaction was started by adding 10 l of the protease solution, and the rise in A 390 was monitored for 5 min. The data were fitted to a first order kinetic and k cat /K m was calculated using k cat /K m ϭ (k obs Ϫ k 0 )/[Ssp1]. The amino acid N-terminal to the proline is indicated.

Substrate
Ϫ1-position Suc-AXPF-pNA Glu 3.48 ϫ 10 6 Gln Ͻ1000 Phe 1670 Ala 8560 Leu Ͻ1000 His Ͻ1000 Ac-AAXPR-pNA Ser(P) 6.5 ϫ 10 6 Ser 8360 Thr(P) 2.64 ϫ 10 6 Thr Ͻ1000 FIG. 4. Protein refolding activity of Ssp1. A, wild type mouse DHFR, denatured in 8 M urea, was subjected to refolding with Ssp1 and N. crassa CyP20 in refolding buffer, respectively, or without any PPIase (control; refolding buffer alone; see also Ref. 6). Aliquots were removed at the indicated time points, and the amount of refolding was assayed by resistance against digestion with proteinase K. An autoradiogram of a dried gel is shown. The 100% lane shows the input of DHFR. B, autoradiograms shown above were quantitated by two-dimensional densitometry; indicated are percentages of refolded protein compared with the total amount of DHFR in the refolding assay. was estimated to be 0.5 M (20). Ssp1 concentration therefore is in the same range as for other abundant PPIases like CyP20 (33) and FKBP13 (35) in N. crassa. In addition, Ssp1 seems not only to be located in the nucleus (like human Pin1; Ref. 15), since we can detect it in and purify it from cytosolic fractions of N. crassa. 3 (b) In contrast to the known eucaryotic members of the parvulin family from Saccharomyces cerevisiae (12,13), Drosophila melanogaster (14), or human cells (15), N. crassa Ssp1 contains a polyglutamine stretch in the interconnecting region between the conserved WW and the PPIase domains, respectively. Polyglutamine regions are characteristic regulatory modules present in various proteins involved in replication, transcription, and cellular regulation (see e.g. Refs. 36 -40). The functional role of the polyglutamine stretch in Ssp1 remains to be clarified.
Taken both points together, we speculate that there might exist more than one parvulin-like protein in higher eucaryotic cells than yeast and that we might have isolated the first member of this new type. In this respect, it is interesting to note that the only parvulin member in yeast (Ess1/Ptf1) is essential (12) but dodo in D. melanogaster is not (14). Whether Pin1 is essential in human cells is a matter of debate (15,51,52), so it might be the case that in higher eucaryotic cells parvulin isoforms exist, as has long been known for CyPs and FKBPs.
In contrast to CyPs and FKBPs, Ssp1 shows a very high substrate specificity in standard PPIase assays using short peptides (31). Our results reveal that peptides containing acidic residues like glutamate, phosphoserine, or phosphothreonine N-terminal to a proline are highly preferred substrates for the PPIase activity of Ssp1. Therefore, the substrate specificity of Ssp1 and the k cat /K m values are comparable with that of the human homolog Pin1 (18,19). Due to the substrate specificity and the binding to mitotic phosphoproteins, Ranganathan et al. (18) and Yaffe et al. (19) postulated that Pin1 might control mitotic steps by accelerating structural changes in proteins after these have been phosphorylated at distinct Xaa-Pro sites by mitotic kinases.
This theory implies that Pin1 is not only able to act as a PPIase on peptide substrates but also exhibits the same substrate specificity regarding proteins. Recent publications show, however, that results obtained in the peptide PPIase assays might not hold true for protein substrates (41,50). Both papers compare the PPIase activity of active site mutant PPIases, as deduced from the peptide PPIase assay, with their ability to refold denatured protein substrates. Mutant cyclophilins and FKBPs, inactive in the peptide assay, however, were active in a protein folding assay. Therefore, one has to be cautious in applying results from the protease-coupled peptide assay to the protein folding activity of a PPIase. With this in mind, we tested whether Ssp1 is also able as a PPIase to accelerate folding of a protein.
We used mouse DHFR as a model substrate for the refolding 3 O. Kops and M. Tropschug, unpublished results. experiments. This protein has been used for a long time in a variety of intracellular folding and transport studies (e.g. Refs. [21][22][23][24][25][26][27]. However, the folding pathway of DHFR is complex and not known in complete detail (44 -46). There are, for example, 13 Xaa-Pro sites in the molecule, which all might be candidates for the action of PPIases.
Using unfolded DHFR as substrate, we found that Ssp1 is indeed able to accelerate folding in vitro. Ssp1 exhibits about the same protein folding activity on DHFR as does CyP20 of N. crassa, which has been shown to be part of the mitochondrial protein folding machinery including molecular chaperones (6,7). Given the high substrate specificity of Ssp1 for acidic residues N-terminal to a proline, we speculated that Glu 81 -Pro 82 is a crucial site for peptidyl-prolyl cis-trans-isomerization. The generation of two mutant proteins DHFR E81A and DHFR E81L strengthens our speculation. Ssp1 is inactive in refolding the mutant proteins, whereas refolding is efficiently catalyzed by CyP20, reflecting the low substrate specificity of CyPs.
Surprisingly, refolding of a mutant where Pro 82 is exchanged for Ala (DHFR P82A) can also be accelerated by both CyP20 and Ssp1 (Fig. 6C). We speculate that in this mutant Ala 82 -Pro 83 becomes the PPIase-sensitive site; Glu 81 seems to make this region accessible for Ssp1. Further experiments are necessary to unravel this novel finding, that a Glu in the Ϫ2position relative to a proline is able to generate a Ssp1-sensitive site.
Exchanging both Pro 82 and Pro 83 for Ala resulted in an unstable protein that did not fold correctly (as tested by protease resistance) and therefore was not accessible for refolding assays (data not shown). Obviously, at least one proline is necessary for correct folding in this region of DHFR.
We conclude that the cis-trans-isomerization of the Glu 81 -Pro 82 bond in mouse DHFR represents a rate-limiting step in the folding pathway. This notion is supported by former studies using E. coli DHFR. Furthermore, Ssp1, a member of the eucaryotic parvulin family, is indeed able to accelerate rate-limiting steps in protein folding due to its PPIase activity. Also, the substrate specificity of Ssp1, as determined by the peptide assay, holds true for protein substrates, like DHFR, with the interesting finding that acidic residues N-terminal (and perhaps also in the Ϫ2-position relative to a proline) are highly preferred, if not mandatory, for the action of this specific PPIase. In view of these results, we suggest that Ssp1 and its homologs from other organisms might act as site-specific PPIases in vivo. However, whether (a) Glu-(Xaa)-Pro (and perhaps Asp-(Xaa)-Pro) or (b) Ser(P)-(Xaa)-Pro and Thr(P)-(Xaa)-Pro, respectively, are the real substrates for Ssp1 and are used with the same efficiency in vivo remains to be clarified.
Further experiments will be necessary to elucidate the exact biochemical mechanism of action of Ssp1 during DHFR folding and to study the role of the two adjacent prolines Pro 82 -Pro 83 . Also, the role of acidic residues (Glu and Asp) in a Ϫ2-position relative to a proline will be clarified. In any case, the use of the site-specific PPIases of the parvulin family can aid in a better understanding of the refolding pathways of certain proteins that depend on prolyl cis-trans-isomerization.