Sequence-Specific and Phosphorylation-Dependent Proline Isomerization: A Potential Mitotic Regulatory Mechanism

See allHide authors and affiliations

Science  12 Dec 1997:
Vol. 278, Issue 5345, pp. 1957-1960
DOI: 10.1126/science.278.5345.1957


Pin1 is an essential and conserved mitotic peptidyl-prolyl isomerase (PPIase) that is distinct from members of two other families of conventional PPIases, cyclophilins and FKBPs (FK-506 binding proteins). In response to their phosphorylation during mitosis, Pin1 binds and regulates members of a highly conserved set of proteins that overlaps with antigens recognized by the mitosis-specific monoclonal antibody MPM-2. Pin1 is here shown to be a phosphorylation-dependent PPIase that specifically recognizes the phosphoserine-proline or phosphothreonine-proline bonds present in mitotic phosphoproteins. Both Pin1 and MPM-2 selected similar phosphorylated serine-proline–containing peptides, providing the basis for the specific interaction between Pin1 and MPM-2 antigens. Pin1 preferentially isomerized proline residues preceded by phosphorylated serine or threonine with up to 1300-fold selectivity compared with unphosphorylated peptides. Pin1 may thus regulate mitotic progression by catalyzing sequence-specific and phosphorylation-dependent proline isomerization.

Proline-directed protein kinases, such as cyclin-dependent protein kinases, play important roles in regulating eukaryotic cell division (1). At the G2-to-M phase transition of the cell cycle, a rapid increase in Cdc2 kinase activity results in phosphorylation of many proteins on serine or threonine residues that are followed by proline (S/T–P sites) (2). Phosphorylation of these proteins is thought to trigger many of the structural modifications that occur during mitosis (1-3). However, it is not clear how this abrupt change in phosphorylation state leads to an organized and programmed set of mitotic events.

Pin1, originally isolated as a protein that interacts with and inhibits the essential mitotic kinase NIMA, is highly conserved and possesses both a WW domain and PPIase activity (4). PPIases catalyze the relatively slow peptidyl-prolyl isomerization of proteins, allowing relaxation of local energetically unfavorable conformational states (5). The cyclophilins and FKBPs, two major families of PPIases, are targets of clinically relevant immunosuppressive drugs. However, the immunosuppressive activity of these drugs is unrelated to inhibition of PPIase activity, and neither cyclophilin nor FKBP genes are essential (5). Thus, evidence for the biological importance of this enzymatic activity is limited.

In contrast, Pin1, a member of a family of PPIases that differ from cyclophilins and FKBPs, is essential for cell growth and requires a catalytically competent PPIase domain (4, 6). Pin1 both negatively regulates entry into mitosis and is required for normal progression through mitosis in human cells and yeast (4). We have recently shown that Pin1 specifically interacts with and regulates the activity of a subset of mitotic proteins in a phosphorylation-dependent manner (7). However, little is known about how these specific interactions are achieved.

To investigate the mechanism by which Pin1 interacts with essential mitotic proteins, we used a glutathione S-transferase (GST)–Pin1 fusion protein to screen oriented degenerate peptide libraries (8). The crystal structure of Pin1 containing an Ala-Pro dipeptide substrate revealed a sulfate ion located 5 Å from the Cβ carbon of Ala, suggesting that phosphorylated Ser (pS) might be preferred at this site (9). To test this idea, we prepared a pS-containing degenerate peptide library of general sequence H2N-MAXXXpSXXXAKK, where X is any amino acid except Cys (8). The GST-Pin1 protein preferentially bound a subset of peptides with Pro immediately COOH-terminal to pS (Fig.1A).

Figure 1

Pin1 binding specificity and its structural basis. (A) Preference of Pin1 for peptides containing Pro immediately downstream of pS. GST-Pin1 was used to screen a peptide library of general sequence H2N-MAXXXpSXXXAKK, where X is any amino acid except Cys, and the bound peptides were sequenced. Values indicate the relative abundance of each of the 19 amino acids at the sequencing cycle corresponding to the position immediately downstream of pS compared with their abundance in the starting library mixture. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. (B) Molecular modeling of the optimal WFYpSPR peptide bound to Pin1. (Top) The molecular surface shaded by surface electrostatic potential shows the aromatic NH2-terminus of the peptide in a hydrophobic channel, the pS phosphate in a basic pocket of R68 and R69, and the aliphatic Pro side chain within the hydrophobic cluster of L122, M130, and F134. The ω angle of the pS-P peptide bond is 90°. The surface potential ranges from −58.073 (dark red) to −29.037 (light red) to 0 (white) to 62.582 (light purple) to 125.165 (dark purple). (Bottom) Pin1 secondary structure showing selected residues. R68 and R69 coordinate the phosphate group (orange), C113 and H59 are implicated in the isomerization mechanism (yellow), and W73 and K117 participate in stacking the peptide Trp residue (green). (C) Isomerase activity of Pin1 and its mutants, expressed and purified as GST fusion proteins, compared with that of parvulin punified fromEscherichia coli. Activity was assayed as described in Table 1.

To investigate whether peptides containing pS-P were preferred substrates for the isomerase activity of Pin1, we synthesized oligopeptide substrates and assayed for cis-trans isomerization by Pin1 and by members of the cyclophilin (Cyp18) and FKBP (FKBP12) families of PPIases (10). Neither Cyp18 nor FKBP12 effectively catalyzed isomerization of peptides with pS/pT–P moieties (Table1). In contrast, both Y-P and pY-P bonds were good substrates for both enzymes. Thus, phosphorylation of S/T–P, but not Y-P, renders the prolyl-peptidyl bond resistant to the catalytic action of conventional PPIases, suggesting the need for a different enzyme to catalyze this reaction.

Table 1

Failure of cyclophilin (Cyp18) and FKBP12 to catalyze effectively the isomerization of the pS/pT–P peptide bond. The assay was performed with purified proteins in triplicate by monitoring the absorbance of released 4-nitroaniline (pNA) with the use of chymotrypsin as an isomer-specific protease.

View this table:

In contrast to that of cyclophilins and FKBPs, the isomerase activity of Pin1 was highly specific for peptides with pS/pT–P bonds. Whereas Pin1 exhibited little isomerase activity with substrates containing S/T–P bonds, phosphorylation of these peptides on Ser or Thr resulted in up to a 300-fold increase ink cat/K m (Table2). Pin1 showed low isomerization activity with peptides containing an A-P peptide bond; incorporation of Glu or Asp immediately preceding Pro in order to mimic pS increased isomerization activity. Peptides containing Y or pY preceding Pro were poor substrates for Pin1. This substrate specificity distinguishes Pin1 from the conventional PPIases in the cyclophilin and FKBP families.

Table 2

Sequence-specific and phosphorylation-dependent PPIase activity of Pin1. Assays were performed as described in Table 1, with the exception that trypsin replaced chymotrypsin as an isomer-specific protease when peptides with Pro-Arg-pNA sequences were used as substrates.

View this table:

To define further the sequence specificity of Pin1, we used a degenerate peptide library containing a fixed pS-P sequence flanked by three degenerate positions on each side. Pin1 selected Arg or aromatic residues at the −1 and +1 positions of the pS-P motif (Table3). Aromatic amino acids were also selected at the −3 position, Phe or Ile at the −2 position, and Leu or Ile at the +2 position. On this basis, several additional peptides were synthesized as Pin1 substrates (10). Peptides with Arg introduced at the P+1 position proved more effective substrates, with kcat/ Km values up to 1300 times those of their nonphosphorylated counterparts (Table2). Placing aromatic residues NH2-terminal to the pS-P position further improved these peptides as substrates (Table 2). The best substrate identified thus far (WFYpSPR- p NA) is the optimal sequence selected from the peptide library (Tables 2 and 3). The apparent Michaelis constant ( Km) of Pin1 for this peptide was 10 μM (10).

Table 3

Binding specificity of Pin1 and MPM-2. GST-Pin1 and MPM-2 were incubated with the pS-P–oriented degenerate peptide library H2N-MAXXXpSPXXAKK, where X is any amino acid except Cys. After extensive washing, peptides bound to GST-Pin1 or MPM-2 were eluted and sequenced. Amino acids selected at each degenerate position are shown.

View this table:

Pin1 binds a large subset of mitotic phosphoproteins that are recognized by the monoclonal antibody MPM-2 (7) (Fig.2A). We therefore investigated the sequence specificity for MPM-2 recognition. When immobilized MPM-2 antibody was used to probe a peptide library containing only pS as the orienting residue (8), peptides with Pro at the pS+1 position were preferentially selected (11). With the pS-P degenerate peptide library, MPM-2 preferentially selected peptides with aromatic or aliphatic amino acids at the −3, −1, and +1 positions relative to pS-P (Table 3). This MPM-2 binding motif is similar to the sequence motif selected by Pin1 (Table 3) and explains the observation that Pin1 specifically interacts with MPM-2 antigens (7).

Figure 2

Interaction between Pin1 and MPM-2 antigens or selected proteins predicted to bind Pin1 from a database search. (A) Interaction between Pin1 and MPM-2 antigens. After arrest at G1-S (I) or mitosis (M), HeLa cells were harvested and lysed. The resulting soluble proteins were incubated with agarose beads containing GST-Pin1, and, after extensive washing, proteins associated with the beads were subjected to immunoblot analysis with MPM-2. Control GST beads did not precipitate any specific MPM-2 antigens, as described (7). The positions of molecular size standards (in kilodaltons) and GST-Pin1 (arrow) are indicated. (B and C) Interaction between Pin1 and S6 kinase (S6K) or Rab4. Beads containing GST-Pin1 or GST were incubated with extracts of HeLa cells in G1-S or M phase and then washed, after which proteins associated with the beads were subjected to immunoblot analysis with antibodies to the COOH-terminal peptide of S6 kinases [which recognize both nuclear (p70) and cytoplasmic (p85) forms of the enzyme] (B) or with antibodies to Rab4 (C).

The structural basis for Pin1 substrate specificity was investigated by molecular model-building and site-directed mutagenesis. On the basis of the Pin1-substrate crystal structure determined by Ranganathan et al. (9), the optimal peptide (WFYpSPR) was modeled onto the Pin1 structure, with the assumption that the phosphate group of pS occupies the position of sulfate in the previous structure (12) (Fig. 1B). In this model, R68 and R69 of Pin1 coordinate the pS phosphate, a hydrophobic groove accepts the preceding aromatic tripeptide, and the side chain of C113 and H59 coordinate the isomerizing pS-P peptide bond with an ω angle of 90°, stabilizing the transition state between the cis and trans configurations. To test these predictions, we generated site-specific Pin1 mutants and assayed their PPIase activity (13).

Substitution of both R68 and R69 by Ala reduced the kcat/ Km for the phosphorylated substrate by a factor of ∼500 relative to that of the wild-type enzyme (Fig. 1C). The catalytic activity of this mutant was identical to that of wild-type Pin1 with the unphosphorylated peptide substrate. Thus, this cluster of basic residues appears to participate in coordinating the phosphate of pS or pT. In parvulin (6), the prototype of the Pin1 family of PPIases, R68 and R69 of Pin1 are replaced by Glu; this enzyme failed to catalyze the isomerization of pS-P peptidyl bonds, although it was effective in catalyzing Pro isomerization in the unphosphorylated peptide (Fig. 1C). Replacement of the catalytic H59 residue of Pin1 with Ala markedly reduced PPIase activity toward both phosphorylated and unphosphorylated peptides (Fig. 1C); however, the preference for phosphorylated over unphosphorylated substrates was unchanged (14). Thus, H59 appears to play an important role in catalyzing Pro isomerization or in binding the substrate Pro residue.

On the basis of amino acid preferences deduced for the six positions surrounding the pS-P motif for optimal Pin1 binding, we undertook a weighted screening of the SWISS-PROT sequence database (15). This scan revealed within the top 5% of high scores several potentially important mitotic phosphoprotein targets for Pin1. Many of these proteins contribute to regulation of the cell cycle, cytoskeletal or spindle structure, DNA replication, transcription, or RNA processing (see supplementary data).

Several of the proteins, such as Rab4, Cdc25, and NIMA, undergo mitosis-specific phosphorylation (3, 16). Cdc25 and NIMA have been previously shown to bind Pin1 in a phosphorylation-dependent manner (7). Other proteins identified in this search, however, had not been previously suspected of interacting with Pin1; a few were therefore further investigated as example cases (7, 17). Rab4 and ribosomal S6 kinases were shown to interact with Pin1 specifically in extracts of mitotic, but not interphase, cells (Fig. 2, B and C). NIMA contains at least five predicted binding sites for Pin1, which may explain why Pin1 was originally isolated as a NIMA-interacting protein (4). Thus, Pin1 binds a wide functional range of mitotic phosphoproteins.

Differences in isomerase activity between Pin1 and other PPIases result from different organization of the X-P binding pocket. In all PPIases, a hydrophobic pocket sequesters the aliphatic Pro side chain (5); hence, the residues responsible for determining substrate preference must reside at the entrance to the Pro binding pocket. In Pin1 and its homologs (4, 18), a cluster of basic residues coordinates the pS phosphate and determines substrate specificity. The absence of a basic pocket in the cyclophilins, FKBPs, and other members of the parvulin family of PPIases may explain their failure to catalyze isomerization of the pS/pT–P bond.

The specificity of Pin1 is consistent with the previous identification of Pin1 binding proteins (7) and also predicts other potential Pin1 substrates, some of which we have confirmed as targets of Pin1 in vitro. Furthermore, Pin1 and MPM-2 bind similar sequences and proteins and have similar mitosis inhibitory phenotypes (7), indicating that the wide conservation of MPM-2 epitopes across various species (16, 19) can be explained by recognition of this epitope by a highly conserved mitotic regulator, Pin1.

Our results suggest a two-step mechanism for mitotic regulation. Phosphorylation at specific S/T–P sites by mitotic kinases creates a binding site for Pin1, which, in turn, induces conformational changes by catalyzing prolyl isomerization (Fig.3). Such local conformational changes could alter the activity of a phosphoprotein such as NIMA (4) or Cdc25 (7), its ability to interact with other proteins, or its degradation. In this manner, Pin1 would provide an additional posttranslational level of control, allowing the general increase in protein phosphorylation to result in the organized and programmed set of structural modifications that occur during mitosis. Given that inhibition of Pin1 induces mitotic arrest and apoptosis (4), this Pin1-mediated mechanism is a potential therapeutic target for cancer.

Figure 3

A model for the Pin1-dependent regulation of mitosis-specific phosphoproteins that are phosphorylated by Cdc2 and other mitotic (M) kinases.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: klu{at}


View Abstract

Navigate This Article