Proteomic Screen Finds pSer/pThr-Binding Domain Localizing Plk1 to Mitotic Substrates

See allHide authors and affiliations

Science  21 Feb 2003:
Vol. 299, Issue 5610, pp. 1228-1231
DOI: 10.1126/science.1079079


We have developed a proteomic approach for identifying phosphopeptide binding domains that modulate kinase-dependent signaling pathways. An immobilized library of partially degenerate phosphopeptides biased toward a particular protein kinase phosphorylation motif is used to isolate phospho-binding domains that bind to proteins phosphorylated by that kinase. Applying this approach to cyclin-dependent kinases (Cdks), we identified the polo-box domain (PBD) of the mitotic kinase polo-like kinase 1 (Plk1) as a specific phosphoserine (pSer) or phosphothreonine (pThr) binding domain and determined its optimal binding motif. This motif is present in known Plk1 substrates such as Cdc25, and an optimal phosphopeptide containing the motif disrupted PBD-substrate binding and localization of the PBD to centrosomes. This finding reveals how Plk1 can localize to specific sites within cells in response to Cdk phosphorylation at those sites and provides a structural mechanism for targeting the Plk1 kinase domain to its substrates.

Activation of signaling cascades in eukaryotic cells involves the directed assembly of protein-protein complexes at specific locations within the cell. This process is controlled by protein phosphorylation on serine, threonine, or tyrosine residues that directly or indirectly regulate protein-protein interactions, often through the actions of phosphopeptide binding domains. Src-homology 2 (SH2) and protein tyrosine binding (PTB) domains bind to specific phosphotyrosine-containing sequence motifs (1), whereas 14-3-3, Forkhead-associated (FHA), and WW domains bind directly to short pSer- or pThr-containing sequences to control cell cycle progression and coordinate the response to DNA damage (2).

To design a general proteomic screen capable of identifying novel phosphoserine- or phosphothreonine binding modules, we took advantage of the observation that protein kinases and phosphopeptide binding domains tend to recognize overlapping sequence motifs. For example, the basophilic protein kinase Akt phosphorylates substrates at sites that contain the core motif RXRSX(S/T) (3) [R, Arg; (S/T), Ser or Thr; X, any amino acid except Cys], and 14-3-3 proteins bind to a subset of these phosphorylated sites with the optimal motif RSX(pS/pT)XP [(pS/pT), phosphoserine or phosphothreonine; P, Pro] (4).

To identify pSer or pThr binding domains, we biased a library of partially degenerate (varied in some but not all residues) phosphopeptides toward the phosphorylation motif of a kinase and then used an immobilized form of this library as bait in an interaction screen against a library of proteins produced by in vitro expression cloning (5, 6) (Fig. 1). This library versus library screening approach is the reverse of traditional peptide library screens in which a single purified domain is assayed against a degenerate peptide library to reveal its optimal binding motif (7). Using a collection of peptides biased toward the motif of a protein kinase superfamily, we cast a larger net than would be possible if only a single peptide were used as bait.

Figure 1

A novel phospho-motif–based library versus library screen to identify pSer or pThr binding domains. pTP and TP libraries correspond to the sequences biotin-ZGZGGAXXBXpTPXXXXAKKK and biotin-ZGZGGAXXBXTPXXXXAKKK, respectively, where pT is phosphothreonine; X indicates all amino acids except Cys; Z denotes aminohexanoic acid; and B indicates the amino acids Pro, Leu, Ile, Val, Phe, Met, and Trp. Libraries were screened against pools of [35S]methionine-labeled proteins translated in vitro from HeLa cell cDNA plasmid pools. Pin1 and a fragment of the mitotic kinase Plk-1 (asterisks) were isolated as clones that preferentially associated with the pTP peptide library. In each panel, the first lane shows 10% of the input. Second and third lanes show binding of proteins within this pool to the phosphorylated and nonphosphorylated libraries, respectively. Identification of Pin1 and Plk1 occurred through progressive subdivision of their respective pools to single clones (right). Molecular sizes are indicated on the left (in kD).

To specifically identify pSer or pThr binding domains involved in cell cycle regulation, we constructed a pThr-Pro library biased to resemble the motif generated by the action of cyclin-dependent kinases and mitogen- activated protein kinases and recognized by the mitotic phosphoprotein-specific monoclonal antibody MPM-2 (see legend to Fig. 1) (8–10). To control for phosphorylation-independent binding, we constructed an identical peptide library with Thr substituted for pThr. The pThr-Pro–oriented peptide library and its nonphosphorylated Thr-Pro counterpart were immobilized on streptavidin beads and screened against 680 pools of in vitro translated 35S-labeled proteins (6). Most proteins produced by in vitro translation did not bind to either library or bound more strongly to beads linked to nonphosphorylated peptides. However, we identified seven distinct pools with proteins that bound preferentially to the pThr-Pro library. Two clones were recovered, one of which (109-B7) was found to encode the prolyl isomerase Pin1, a protein known to bind and isomerize pThr-Pro motifs (10), validating the feasibility of our screening approach.

A second clone, 407-C6, encoded the COOH-terminal 80% of the mitotic kinase Plk1 (polo-like kinase-1; amino acids 95 to 603) (6, 11, 12). This clone was missing critical components of the Plk1 kinase domain, including the glycine-rich loop (amino acids 60 to 66) and the invariant lysine (K82), which implies that phosphopeptide binding was independent of Plk1 kinase activity. Preferential binding to pThr-Pro peptides by full-length transcripts of this clone was less pronounced than that of Pin1. However, partial translation products or proteolytic breakdown fragments showed strong discrimination for the phosphorylated peptide library (Fig. 1, arrowheads), suggesting that Plk1 likely contains a discrete phosphopeptide binding domain.

A hallmark feature of the polo kinase family is the presence of a highly conserved COOH-terminal region that includes two blocks of strong similarity, termed polo boxes (13). Therefore, we generated a series of deletion constructs based on an alignment of the COOH-terminal regions of polo kinase family members (fig. S1) and analyzed them for phosphopeptide-specific binding (Fig. 2). A construct that began immediately after the kinase domain and extended to the last residue of the protein (residues 326 to 603) demonstrated strong and specific binding to the pThr-Pro peptide library. This construct was superior to the parent clone 407-C6 in discriminating for phosphopeptides. Neither polo-box 1 (PB1) nor polo-box 2 (PB2) alone, nor a construct containing both polo boxes but lacking the linker between the kinase domain and PB1, bound phosphopeptides. Thus, the linker together with both polo boxes appears to function as a single phosphopeptide binding module that we call the polo-box domain (PBD). This region regulates the subcellular localization of Plk1 during mitosis (14, 15), and both polo boxes are necessary for this function (13).

Figure 2

Deletion mapping of the phospho-binding domain of Plk1. COOH-terminal truncations of Plk1 were translated in vitro and assayed for selective binding to the phosphorylated peptide library in Fig. 1. Shaded regions in the COOH-terminus of Plk1 correspond to its polo boxes PB1 and PB2, as defined by Pfam (23). Clone 407-C6 is the fragment of Plk1 isolated from the screen in Fig. 1.

A critical aspect of our screening technique is that any domain isolated through binding to bead-immobilized peptide libraries will yield an optimal consensus binding motif when analyzed by traditional peptide library screening (6, 7). We determined the optimal binding motif for the PBD by using multiple pThr- and pSer-containing peptide libraries. Extremely strong selection for Ser in the (pThr or pSer) −1 position was revealed by screening peptide libraries with a degenerate (pThr or pSer)-1 position (Fig. 3, A and B; table S1). This Ser selection is among the largest we have observed for any domain whose specificity has been determined by peptide library screening. Because Plk1 PBD was isolated in a screen for domains that bind pThr-Pro motifs, we investigated the relative importance of Pro in the (pThr or pSer) +1 position by assaying the PBD with libraries containing a degenerate +1 position. Little selection for Pro was observed when Ser was not fixed at (pThr or pSer) −1 (table S1), whereas inclusion of Ser at the –1 position unmasked a moderate Pro selection (Fig. 3, C and D). These results suggested that Pro in the (pThr or pSer) +1 position, although helpful, was not absolutely required for binding, as confirmed by PBD interaction with bead-immobilized libraries containing either pThr or pThr-Pro (Fig. 3E).

Figure 3

The Plk1 PBD binding motif. The PBD (residues 326-603) of Plk1 was immobilized as a glutathioneS-transferase (GST) fusion on glutathione beads and incubated with pThr- or pSer-oriented degenerate peptide libraries consisting of the sequences MAXXXXpTPXXXX- AKKK (A), MAXXXpSPXXXAKKK (B), MAXXXXSpTXXXXAKK (C), or MAXXXXSpSXXXXAKK (D). After extensive washing, bound peptides were eluted and sequenced. Bar graphs show the ratio of each amino acid (24) at a given cycle of sequencing relative to its abundance in the starting peptide library mixture. (E) Absolute Pro requirement in the pThr +1 position for peptide binding to Pin1 but not for binding to the Plk1 PBD. Full-length Pin1 and the PBD (residues 326 to 603) of Plk1 were translated in vitro in the presence of [35S]methionine and assayed for binding to four immobilized peptide libraries that differed by phosphorylation status and/or the presence of Pro in the pThr +1 position. pTP and TP sequences are given in Fig. 1; pT represents biotin-ZGZGGAXXXXpTXXXXXAKKK and T represents biotin-ZGZGGAXXXXTXXXXXAKKK.

Isothermal titration calorimetry (ITC) (6) showed that the optimal phosphopeptide ligand (PoloBoxtide-optimal) containing the core sequence Met-Gln-Ser-phosphoThr-Pro-Leu bound tightly to Plk1 PBD with a dissociation constant of 280 nM (fig. S2). Furthermore, it formed a 1:1 protein/peptide complex, indicating that separate phosphopeptides were not interacting simultaneously with the two polo boxes within the PBD. Substitution of Thr for pThr within the optimal PBD motif (PoloBoxtide 8T) eliminated peptide binding. Replacing pThr with pSer maintained binding, albeit with one-seventh the affinity, whereas replacing pThr with pTyr completely abrogated binding. Strong selection for Ser in the (pThr or pSer) −1 position was confirmed by complete loss of peptide binding when Ser was substituted by Val. Replacing the (pThr or pSer) +1 Pro with Asn, another β-turn–forming residue, increased the dissociation constant by a factor of five, verifying moderate selection for Pro in this position. On the basis of oriented peptide library screening data and these ITC results, we therefore propose that a core consensus motif recognized by Plk1 PBD is S-(pT/pS)-(P/X).

The monoclonal antibody MPM-2 (mitotic phosphoprotein monoclonal-2) (8) recognizes a conserved (pSer or pThr)–Pro epitope (9, 10) present on about 50 mitotic phosphoproteins. The pThr-Pro library used in the initial screen was partially biased to resemble the MPM-2 epitope. A number of important mitotic regulators recognized by MPM-2, including Cdc25, Wee1, Myt1, topoisomerase II α, and INCENP, contain one or more exact matches of the S-(pS/pT)-P PBD binding motif. We therefore investigated whether Plk1 PBD bound to MPM-2–reactive proteins that were generated by arresting HeLa cells in G2 or M phase of the cell cycle with nocodazole. Many MPM-2 reactive phosphoproteins were specifically bound by Plk1 PBD (Fig. 4A), which suggests that Pro-directed mitotic kinases generate PBD binding sites. Furthermore, Plk1 PBD bound to a different and somewhat smaller subset of MPM-2 epitope-containing proteins than those that bound to Pin1 (10). Incubation of Plk1 PBD with its optimal phosphopeptide ligand inhibited binding to MPM-2 epitopes (fig. S3, opt). In contrast, the nonphosphorylated analog (8T) or a peptide with Val substituted for Ser in the pT-1 position (7V) had no effect, demonstrating that binding was occurring through the phosphopeptide binding pocket of the PBD.

Figure 4

Association of Plk1 PBD with mitotic phosphoproteins in HeLa cells through its phosphopeptide binding pocket. (A) Lysates from HeLa cells, arrested in G1-S with aphidicolin (lanes A) or in G2-M with nocodazole (lanes N), were incubated with GST, GST-Pin1, and the GST-Plk1 PBD (residues 326 to 603); bound proteins were detected by blotting with MPM-2. (B) Interaction of the Plk1 PBD phosphopeptide pocket with mitotically phosphorylated endogenous Cdc25C from HeLa cells. Interaction of GST-Plk1 PBD with Cdc25C was disrupted by incubation with its optimal phosphopeptide ligand (opt) but not with an unphosphorylated equivalent peptide (8T) or a phosphopeptide whose Ser at pThr –1 was mutated to Val (7V). (C) Binding of the Plk1 PBD to Cdc25C in vivo. HeLa cells were transfected with a His-Xpress–tagged Plk1 PBD construct (residues 326 to 603) or a non-phospho–binding Plk1 PBD construct (residues 326 to 506) (Fig. 2) and arrested in G2-M with nocodazole. Lysates were pulled down with Ni2+ beads and blotted for Cdc25C. (D) Interaction of Plk1 PBD with Thr130 of human Cdc25C. Lysates were prepared from HeLa cells transfected with wild-type, T130A, or S129V HA-Cdc25C (human), arrested in G2-M with nocodazole, and normalized for equal loading of the mitotically upshifted form of Cdc25C. Interaction of GST-Plk1 PBD (residues 326 to 603) with mitotically phosphorylated Cdc25C was detected by separation on 11.4% SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and HA antibody blotting. (E) Lysates, prepared as in (D), were analyzed by 9% SDS-PAGE to enhance separation of the hyperphosphorylated (P) form of Cdc25C from partially phosphorylated and unphosphorylated (U) forms.

During mitosis the cell cycle–regulated protein phosphatase Cdc25C undergoes a shift in gel mobility due to extensive phosphorylation in its NH2-terminal region (16, 17). The Plk1 PBD phosphopeptide binding pocket interacted only with this hyperphosphorylated form of human Cdc25C in vitro and in vivo (Fig. 4, B and C). One phosphorylation site in the NH2 terminus of Cdc25C, T130, contains a conserved Plk1 PBD consensus motif. HeLa cells were therefore transfected with hemagglutinin (HA)-tagged wild-type Cdc25C or with T130A or S129V point mutants of Cdc25C expected to disrupt the PBD binding motif. In mitotically arrested lysates, Plk1 PBD bound strongly to wild-type Cdc25C but barely to either point mutant, which indicates direct interaction between Plk1 PBD and a mitotically phosphorylated PBD consensus motif in Cdc25C (Fig. 4D). Furthermore, both point mutants displayed incomplete phosphorylation when analyzed on lower percentage gels, indicating that Cdc25C mutations that impair PBD binding result in incomplete mitotic activation in vivo (Fig. 4E).

Plk1 localizes to centrosomes and kinetochores in prophase and to the spindle midzone during late stages of mitosis through its COOH-terminus (14, 15). Centrosomal localization requires both the PB1 and PB2 regions but not kinase activity (13). We therefore arrested U2OS cells with nocodazole, permeabilized them with streptolysin-O, and incubated them with GST-Plk1 PBD in the absence or presence of peptide competitors (6). Plk1 PBD localized to the centrosomes of nocodazole-arrested cells (Fig. 5), as verified by costaining with an antibody to γ-tubulin. This localization was disrupted by incubation of Plk1 PBD with its optimal phosphopeptide ligand but was unaffected by the nonphosphorylated analog (Fig. 5; fig. S4).

Figure 5

Centrosomal localization of Plk1 PBD blocked by its optimal phosphopeptide ligand. (Left) U2OS cells, arrested in G2-M with nocodazole, were incubated with 4 μM GST-Plk1 PBD in cell permeabilization buffer containing Streptolysin-O (1 unit/ml) in the presence of no peptide (upper), optimal phosphopeptide (250 μM) (optimal) (middle), or the corresponding unphosphorylated analog(250 μM) (8T) (lower). Cells were washed, fixed, extracted with Triton X-100, immunostained for GST and γ-tubulin, and counterstained with 4′,6-diamidino-2-phenylindole (DAPI) to visualize the nucleus. Overlap of the GST (Alexa Fluor 488) and γ-tubulin (Texas Red) signals is shown in the merged figure in the far right column and quantitated in fig. S4.

Our identification of Plk1 PBD as a pSer or pThr binding domain that mediates Plk1 localization to substrates and centrosomes (6) (i) provides a molecular mechanism for the essential function of the Plk1 COOH-terminus (13–15), (ii) adds another member to a growing collection of pSer/pThr-binding modules, and (iii) demonstrates the general utility of our phospho-motif–based affinity screen for identifying binding modules interacting with substrates of any kinase whose phosphorylation motif is known. Parallels between the PBD in Plk1 and SH2 domains in Src family kinases are apparent. Like the Src SH2 domain, the PBD binds to the kinase domain, inhibiting its phosphotransferase activity in the basal state (18). Thus, both the PBD and SH2 domains target their respective kinases toward phosphorylated substrates upon activation but inhibit their catalytic activity in the basal state (19, 20). A requirement for priming phosphorylation of Plk1 substrates is similar to that observed for the kinase GSK3 (21). In the case of mitotic regulation by Plk1, small amounts of Cdc2-cyclin B activity during prophase are insufficient to fully activate Cdc25C but could provide priming phosphorylation of Cdc25C for interaction with the PBD. Activation of Plk1 later in mitosis might then lead to an initial wave of Cdc25C phosphorylation and activation (17), generating more Cdc2-cyclin B activity, priming additional Cdc25C molecules for phosphorylation by Plk1, and resulting in a positive feedback loop (12) (fig. S5). This model may explain the observation that maximal intracellular targeting and activation of Cdc25C by constitutively active Plk1 requires coexpression of cyclin B1 (22). In this regard, the PBD may be an attractive target for the design of anticancer chemotherapeutics because its compact tripeptide binding motif may be particularly amenable to the design of small molecule mimetics.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Table S1

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


Stay Connected to Science

Navigate This Article