Structure–function relationship of the Polo-like kinase in Trypanosoma brucei Free

Polo-like kinases (Plks) represent a family of evolutionarily conserved serine/threonine protein kinases that are localized to various subcellular structures and play multiple essential roles in the eukaryotic cell cycle, including mitotic entry, bipolar spindle formation, chromosome segregation, mitotic exit and cytokinesis (Barr et al., 2004; Archambault and Glover, 2009). Plks share a characteristic structure, with a canonical kinase domain (KD) in the N-terminus and a regulatory Polo-box domain (PBD) consisting of one or two Polo boxes in the C-terminus (Barr et al., 2004).

Plk activity is regulated temporally and spatially during the cell cycle. It is controlled by the localization of Plks to specific subcellular structures and by phosphorylation-mediated activation (Barr et al., 2004), both of which require the PBD. Localization of Plks is mediated by the PBD through its association with different substrate proteins at certain locations (Lee et al., 1998; Song et al., 2000; Jang et al., 2002a; Reynolds and Ohkura, 2003). The PBD binds to a phosphoserine/threonine-containing motif, which stimulates the kinase activity of Plk (Elia et al., 2003a; Elia et al., 2003b). Before mitotic entry Plk activity is inhibited, partly owing to the intramolecular binding of the PBD to the KD, which prevents substrate binding and kinase activation (Jang et al., 2002a). However, upon mitotic entry, Plks are activated by Aurora A kinase-mediated phosphorylation of a well-conserved threonine residue (Thr210 in human PLK1 and Thr201 in Xenopus Plx1) in the activation loop (T-loop) of the KD (Jang et al., 2002b; Macurek et al., 2008; Seki et al., 2008a). This activation of Plk1 requires prior association of Bora, an activator of Aurora A (Hutterer et al., 2006) and a substrate of PLK1 (Seki et al., 2008b), with the PBD, which allows access of Aurora A to PLK1 Thr210 (Seki et al., 2008a). In addition to Thr210, phosphorylation of a conserved serine residue outside of the T-loop (Ser137 in PLK1 and Ser128 in Plx1) also confers substantial kinase activation (Lee and Erikson, 1997; Smits et al., 2000; Jang et al., 2002b), but the corresponding protein kinase remains unidentified. Moreover, a number of other phosphorylation sites, including another conserved threonine residue in the T-loop (Thr214 in PLK1), have been mapped in PLK1 and Plx1 (Kelm et al., 2002; Wind et al., 2002; Dulla et al., 2010), but the significance of this phosphorylation remains to be determined. Although Plk activity is likely to be regulated in all eukaryotic organisms, the regulatory mechanisms might differ substantially between systems.

The crystal structure of the PLK1 PBD–phosphopeptide complex has been determined (Cheng et al., 2003; Elia et al., 2003b). Despite a low sequence identity between the two PLK1 Polo boxes, they exhibit similar folds with each Polo box comprising a six-stranded β-sheet and an α-helix. The two β-sheets run in anti-parallel directions to form a β-sandwich domain, and the phosphopeptide binds along a positively charged cleft between the two Polo boxes (Cheng et al., 2003; Elia et al., 2003b). Four residues in the PBD – Trp414, Leu490, His538 and Lys540 – make direct contact with the phosphopeptide, and mutation of His538/Lys540 or Trp414 abolishes phosphopeptide binding and centrosomal localization of PLK1 (Elia et al., 2003b; Garcia-Alvarez et al., 2007). The single Polo box in the PBD of the murine Plk, Sak (Plk4), also consists of a six-stranded β-sheet and an α-helix, but it forms an intermolecular dimer that resembles the structure of the two Polo boxes of the PLK1 PBD (Leung et al., 2002). Strikingly, despite the single Polo box being sufficient for localizing Sak to centrosomes and the cleavage furrow, the residues intimately involved in phosphopeptide binding by the PLK1 PBD, such as Trp414, His538 and Lys540, are not conserved in the Polo box of Sak (Leung et al., 2002). It remains unknown whether the Sak Polo-box dimer is capable of binding to phosphopeptides and whether localization of Sak requires substrate binding to its Polo box.

The subcellular localization and function of the Plk homolog in Trypanosoma brucei (TbPLK), which is an early branched unicellular eukaryote, appear to differ from that of any other Plks studied so far. TbPLK possesses a well-conserved KD in the N-terminus and a putative PBD consisting of two tandem Polo boxes in the C-terminus (Graham et al., 1998). However, unlike other Plk homologs, TbPLK is excluded from the nucleus and is instead enriched at the anterior tip of the new flagellum attachment zone (FAZ) (Kumar and Wang, 2006; de Graffenried et al., 2008; Umeyama and Wang, 2008). The anterior tip of the FAZ is believed to constitute the initiation site of cytokinesis in trypanosomes (Vaughan and Gull, 2003; Briggs et al., 2004; Li et al., 2008b; Li et al., 2009). Earlier in the cell cycle, before the new FAZ is formed, TbPLK is concentrated in basal bodies and in a bilobed structure known to be involved in Golgi duplication (de Graffenried et al., 2008). Consistent with its localization, TbPLK is required for cytokinesis and Golgi duplication without any essential involvement in mitosis (Kumar and Wang, 2006; Hammarton et al., 2007; de Graffenried et al., 2008). TbPLK is the first Plk homolog known to play no essential role in mitosis.

Despite the essential function of TbPLK in trypanosomes, little is known about the regulation of TbPLK and the role of the PBD in regulating TbPLK activity and localization. In this report, we carried out a structure–function analysis of TbPLK by coupling RNAi-mediated gene silencing against the 3′-UTR of TbPLK with ectopic expression of various TbPLK mutants harboring individual point mutations in the KD and the PBD. We also expressed several truncation mutants of TbPLK to investigate the requirement of the PBD for TbPLK localization. Moreover, we examined the potential intramolecular interaction between the KD and the PBD of TbPLK as well as the potential formation of an intermolecular TbPLK homodimer mediated by the association between the PBDs and/or KDs. Finally, we tested whether the PBD of TbPLK is involved in binding to TbPLK substrates. Our results suggest that the PBD is necessary, but not sufficient, for TbPLK localization, and that it is not involved in direct substrate binding. The PBD of TbPLK appears to be involved in regulating TbPLK activity through intramolecular association with the KD to inhibit its activity.

To investigate the structure–function relationship of TbPLK, we employed a method that couples RNAi-mediated gene silencing with ectopic expression of TbPLK mutants in the same cell line (supplementary material Fig. S1A,B). To avoid interfering with the ectopically expressed TbPLK mutants, we first tested whether RNAi against the 3′-UTR of TbPLK is sufficient to knock down the expression of TbPLK. A 549 bp DNA fragment immediately downstream of the TbPLK coding region was cloned into a modified RNAi vector, pZJM-PAC, and the resulting construct was introduced into the 29-13 cell line (Wirtz et al., 1999). Northern blot showed that ~95% of TbPLK mRNA was depleted from the cells after RNAi induction for 2 days, and this led to a significant growth arrest (supplementary material Fig. S2A). Flow cytometry analysis showed an accumulation of cells with 4C DNA content after RNAi induction for 2 days and then the emergence of cells with 8C DNA content after RNAi for 3 days (supplementary material Fig. S2B). Moreover, upon RNAi induction, cells with two nuclei and one kinetoplast (2N1K) and cells with multiple nuclei were significantly increased (supplementary material Fig. S2C), which is similar to the effects of RNAi against the coding sequence of TbPLK (Kumar and Wang, 2006; Hammarton et al., 2007). Together, these observations suggest that RNAi against the 3′-UTR of TbPLK is sufficient to dysregulate TbPLK expression.

To determine whether TbPLK deficiency caused by RNAi against the TbPLK 3′-UTR can be rescued by ectopic expression of wild-type TbPLK, we transfected the pZJM-TbPLK-UTR-PAC cell line with pLew100-TbPLK-3HA-Phleo, and induced RNAi and ectopic expression of TbPLK by tetracycline. In vitro kinase assays showed that the ectopically expressed TbPLK-3HA was active (Fig. 1A), and immunofluorescence assays showed that TbPLK-3HA was correctly localized (Fig. 1B). Further, western blot showed that TbPLK-3HA was expressed and RT-PCR confirmed that the endogenous TbPLK mRNA was depleted (Fig. 1C, right panel), consistent with the northern blot (supplementary material Fig. S2A, inset). We found that ectopic expression of wild-type TbPLK alleviated the growth defect of the TbPLK-UTR RNAi cell line (Fig. 1C, green line versus purple line), suggesting that ectopic expression of TbPLK was able to complement the RNAi mutant. It also suggests that epitope tagging at the C-terminus of TbPLK does not interfere with TbPLK function and localization. This agrees with a previous report that tagging of a triple HA epitope at either the N-terminus or the C-terminus of TbPLK does not affect TbPLK localization (Umeyama and Wang, 2008). However, in contrast to wild-type TbPLK, ectopic expression of TbPLK-K70R, a kinase-dead mutant (Fig. 1A), was unable to rescue the growth defect of the RNAi mutant (Fig. 1D, green line) despite being correctly localized (Fig. 1B). This result further confirmed that RNAi against the 3′-UTR of TbPLK was specific.

We also overexpressed 3HA-tagged TbPLK and K70R mutant in the wild-type 29-13 cell line (supplementary material Fig. S1C) and, as reported previously (Hammarton et al., 2007), overexpression of wild-type TbPLK led to growth inhibition (Fig. 1C, red line). It appears that TbPLK was only overexpressed moderately (supplementary material Fig. S1C), but this slight increase in TbPLK level is toxic to the cells, suggesting that the TbPLK level is under tight control in trypanosome cells. Intriguingly, unlike other kinase-dead mutants that generate dominant-negative effects when overexpressed (Jang et al., 2002b; Li and Wang, 2006), overexpression of TbPLK-K70R in the wild-type 29-13 cell line did not affect cell growth (Fig. 1D, red line) despite being correctly targeted (data not shown), suggesting the lack of any dominant-negative effect for TbPLK-K70R.

Since the 29-13 cells expressing wild-type TbPLK and the TbPLK-UTR RNAi cells expressing TbPLK-K70R have a growth defect, we compared their phenotypes. Upon tetracycline induction, cells with two nuclei and one kinetoplast (2N1K) appeared and then decreased, which was followed by a gradual increase in cells with multiple nuclei and one kinetoplast (XN1K) in both cell lines (Fig. 1E). Similarly, the number of 2N2K cells also increased and then decreased, and this was followed by a gradual increase in XN2K cells. There was a slight enrichment of cells with multiple nuclei and multiple kinetoplasts (XNXK), but zoid (0N1K) cells were not increased (Fig. 1E). The increase of 2N1K cells in both cell lines suggests the inhibition of kinetoplast segregation, which was likely to be due to the defect in basal body duplication/segregation (Fig. 1F; data not shown). Moreover, we found that TbPLK-UTR RNAi/TbPLK-K70R overexpression (OE) cells accumulated multiple Golgi bodies (Fig. 1G), which resembles the defect caused by TbPLK RNAi (de Graffenried et al., 2008).

Human PLK1 is activated by aurora A-mediated phosphorylation on Thr210 in the T-loop (Seki et al., 2008a). Additionally, Thr214 in PLK1 is also phosphorylated in vivo, but it is not known whether this is required for PLK1 activation (Dulla et al., 2010). TbPLK also possesses the two conserved threonine residues Thr198 and Thr202 in its T-loop (Fig. 2A), but it is unclear whether they are phosphorylated and whether this would contribute to TbPLK activation. We mutated Thr198 or Thr202 to alanine and aspartic acid and expressed each of the four mutant proteins in the TbPLK-UTR RNAi cell line. In vitro kinase assays showed that mutation of either residue to alanine disrupted their ability to phosphorylate TbCentrin2, an in vitro substrate of TbPLK (de Graffenried et al., 2008), but mutation to aspartic acid restored kinase activity that was several-fold higher than that of wild-type TbPLK (Fig. 2B), indicating that phosphorylation of both threonine residues is required for TbPLK activation.

Immunofluorescence assays showed that all four mutant proteins were localized correctly to the anterior tip of the new FAZ (Fig. 2C), suggesting that phosphorylation of TbPLK on Thr198 and Thr202 is not necessary for TbPLK localization. When the two inactive mutants were each expressed in TbPLK-UTR RNAi cells, cell growth was still arrested. However, when the two phosphomimic mutants were each expressed in the TbPLK-UTR RNAi cell line, cell growth was restored (Fig. 2D, green lines). Although the two phosphomimic mutants are more active than wild-type TbPLK, the protein level of the two mutants is significantly lower than that of the wild-type protein, such that the overall kinase activity of the two mutants is similar to that of wild-type TbPLK (Fig. 2B). This might explain why overexpression of the more active kinase in TbPLK RNAi cells can still rescue the deficiency in endogenous TbPLK. It remains unknown how the protein level of the two phosphomimic mutants is regulated. Moreover, both the TbPLK-UTR RNAi/TbPLK-T198A OE and TbPLK-UTR RNAi/TbPLK-T202A OE cell lines contained multiple Golgi bodies (Fig. 2E), whereas the RNAi cell lines co-expressing the phosphomimic mutants had normal numbers of Golgi (data not shown). These results suggest that the deficiency in endogenous TbPLK was complemented by the phosphomimic mutants but not by the phosphodeficient mutants.

Additionally, like the kinase-dead mutant TbPLK-K70R, overexpression of the two inactive mutants TbPLK-T198A and TbPLK-T202A in the wild-type 29-13 cell line did not inhibit cell growth, whereas overexpression of the two phosphomimic mutants TbPLK-T198D and TbPLK-T202D led to a significant growth defect (Fig. 2D) and inhibited kinetoplast segregation and cytokinesis (supplementary material Fig. S3), similar to TbPLK overexpression in the 29-13 cell line (Fig. 1C). These results confirm that TbPLK activation requires phosphorylation on both Thr198 and Thr210.

A characteristic feature of Plks is the C-terminal regulatory PBD, which is implicated in binding to Plk substrates and targeting Plks to various subcellular structures during the cell cycle (Archambault and Glover, 2009). The single Polo box in human SAK forms an intermolecular homodimer, whereas the two Polo boxes of human PLK1 exhibit similar folds and run in anti-parallel directions, with the substrate bound between the two Polo boxes (Leung et al., 2002; Cheng et al., 2003; Elia et al., 2003b). Like human PLK1, the PBD of TbPLK appears to possess similar folds, with each Polo box comprising a six-stranded β-sheet and an α-helix, except for the insertion between β5 and β6 of Polo box 1, an additional α-helix (α1′) in Polo box 1, and the insertion between the two Polo boxes (supplementary material Fig. S4A). The PBD of TbPLK exhibits a high degree of symmetry, with the potential substrate capable of binding in between the two Polo boxes, as in its human counterpart (supplementary material Fig. S4B,C).

In the PBD of human PLK1, the four residues Trp414, Leu490, His538 and Lys540 are known to make direct contact with the phosphopeptide (Cheng et al., 2003; Elia et al., 2003b), and mutation of Trp414, His538 or Lys540 significantly impairs substrate binding and centrosomal localization (Lee et al., 1998; Elia et al., 2003b; Garcia-Alvarez et al., 2007). TbPLK appears to lack all four of these residues in its PBD (supplementary material Fig. S4A, gray boxes), but a careful scan of the PBD of TbPLK identified His710 and Lys712, which are adjacent to, but are not aligned well with, the conserved His538 and Lys540 in PLK1 and other Plk homologs (supplementary material Fig. S4A, arrows between β9 and β10). This raised the question of whether His710 and Lys712 are important for TbPLK localization and function. Moreover, because the Trp414 equivalent in the TbPLK PBD is replaced by phenylalanine (Phe561; supplementary material Fig. S4A, gray box in β1), we examined whether Phe561 is essential for TbPLK localization and function. Furthermore, we investigated whether the adjacent tryptophan residue (Trp557; supplementary material Fig. S4A, arrow in β1) has taken over the role of Phe561 for substrate binding.

We mutated Trp557 to phenylalanine and Phe561, His710 and Lys712 to alanine and expressed the proteins in TbPLK-UTR RNAi cells. In vitro kinase assays showed that mutation of any of the four residues did not affect kinase activity (Fig. 3A). Immunostaining with anti-HA antibody showed that each of the four mutants was correctly localized to the anterior tip of the new FAZ (Fig. 3B), suggesting that mutation of these residues does not abrogate TbPLK localization. Importantly, ectopic expression of the four TbPLK mutants in TbPLK-UTR RNAi cells restored cell growth at a rate similar to the uninduced controls (Fig. 3C). Moreover, overexpression of any of the four TbPLK mutants in the wild-type 29-13 cell line resulted in a severe growth defect (Fig. 3C) and inhibited kinetoplast segregation and cytokinesis (supplementary material Fig. S5), similar to overexpression of wild-type TbPLK (Fig. 1C). These results suggest that mutation of the four potentially important residues Trp557, Phe561, His710 and Lys712 does not impair the activity, localization or function of TbPLK. It also implies that the trypanosome PBD might recognize TbPLK substrates that are different from those in yeasts and mammals or may bind to the substrate through different residues in the PBD.

The unusual features of the PBD of TbPLK raised the question of whether it could be correctly targeted to various subcellular locations such as centrosomes, spindle poles, the spindle midzone and cleavage furrow in human cells, like its human counterpart. EYFP-tagged TbPLK was ectopically expressed in HeLa cells. Interestingly, TbPLK-EYFP was predominantly nuclear in interphase cells (Fig. 4A), whereas in mitotic cells TbPLK-EYFP was localized throughout the cytoplasm, suggesting that, in contrast to PLK1, TbPLK is unable to be targeted to spindle poles, the central spindle or the midbody in HeLa cells.

Conversely, to examine whether human PLK1 is correctly targeted in trypanosomes, 3HA-tagged PLK1 was ectopically expressed in T. brucei. We found that PLK1-3HA was distributed throughout the cytoplasm in all cell types (Fig. 4B), indicating that, unlike TbPLK, human PLK1 is incapable of localizing to the anterior tip of the new FAZ. Collectively, these experiments suggest that the PBD of PLK1 and the PBD of TbPLK are neither functionally equivalent nor interchangeable.

Despite lacking the four key residues (supplementary material Fig. S4A), the PBD might still be involved in targeting TbPLK to the anterior tip of the new FAZ. We deleted the entire PBD from TbPLK and expressed the truncation mutant in T. brucei (Fig. 5A,B). Surprisingly, overexpression of the PBD deletion mutant resulted in a severe growth defect (Fig. 5C), and the mutant protein was predominantly localized to the nucleus (Fig. 5D), in striking contrast to the overexpressed wild-type TbPLK, which was localized to the anterior tip of the new FAZ as well as the anterior tip of the cell (Fig. 5D). Localization to the anterior tip of the cell was only observed with overexpressed TbPLK and not with the endogenous TbPLK (Kumar and Wang, 2006; Umeyama and Wang, 2008). Localization of the PBD deletion mutant to the nucleus appeared to impair mitosis by causing unequal nuclear division to produce a small nucleus (data not shown). However, cytokinesis was apparently unaffected and cells without a nucleus (zoids) or with a small nucleus were detected (data not shown). It is not clear whether this mitotic defect is due to a dominant-negative effect of the PBD deletion mutant by interfering with a small fraction of TbPLK in the nucleus. If this is the case, then TbPLK might also play a mitotic role.

A scan of TbPLK sequence identified a putative nuclear localization signal (NLS) sequence downstream of the KD (Fig. 5A). To investigate whether this NLS is responsible for nuclear localization of the PBD deletion mutant, we further deleted the NLS sequence. Like the PBD deletion mutant, overexpression of the NLS-PBD deletion mutant also led to significant growth inhibition (Fig. 5C). However, this mutant protein was distributed in the cytoplasm (Fig. 5D), indicating that the nuclear localization of the PBD deletion mutant can be attributed to this NLS. To test whether mutation of this NLS affects TbPLK function, we expressed a TbPLK mutant with all of the arginine and lysine residues in the NLS mutated to alanine in TbPLK-UTR RNAi cells. We found that the NLS mutant was still able to complement the RNAi mutant (data not shown), suggesting that the small fraction of TbPLK in the nucleus, if any, is unlikely to be essential for mitosis and cell proliferation. Together, these results suggest that the PBD is necessary for TbPLK localization.

To further examine whether both Polo boxes are required for TbPLK localization, we deleted the second Polo box (PB2) (Fig. 5A). Overexpression of the PB2 deletion mutant also caused growth inhibition, but the growth defect was less severe than that caused by overexpression of the PBD deletion mutant or the NLS-PBD deletion mutant (Fig. 5C), presumably owing to the lower expression level of the PB2 deletion mutant in the cell (Fig. 5B). Surprisingly, the PB2 deletion mutant was also distributed to the cytoplasm (Fig. 5D), suggesting that the presence of Polo box 1 (PB1) downstream of the NLS might prevent its exposure to the nuclear targeting machinery and consequently abolish its nuclear localization. This also suggests that correct targeting of TbPLK requires the entire PBD. Finally, to test whether the PBD itself is sufficient for localization, we expressed the PBD alone in trypanosomes (Fig. 5A,B). Unlike the other three truncation mutants, overexpression of the PBD did not inhibit cell growth (Fig. 5C) and the PBD was spread throughout the cytoplasm (Fig. 5D). This contrasts with the localization of the PBD of yeast and metazoan Plks, which is virtually identical to that of the endogenous Plks (Lee et al., 1998; Seong et al., 2002; Hanisch et al., 2006).

Collectively, our results suggest that the PBD in TbPLK is essential, but not sufficient, for TbPLK targeting. Localization of TbPLK appears to require both the KD and the PBD, consistent with a previous report (Archambault et al., 2008).

The PBD in Plks is known to be involved in substrate binding, and four well-conserved residues (Trp414, Leu490, His538 and Lys540 in human PLK1) are implicated in direct association with the substrate (Cheng et al., 2003; Elia et al., 2003b). Strikingly, TbPLK lacks all of the four residues in its PBD (supplementary material Fig. S4A, gray boxes), and ectopic expression of TbPLK in HeLa cells failed to target TbPLK to centrosomes, spindle poles, spindle midzone and the cleavage furrow (Fig. 4A), thus arguing that the PBD of TbPLK binds to its substrate through a different mechanism or might not be involved in substrate binding.

To investigate whether the PBD of TbPLK associates with its substrates, we tested the interaction of the PBD as well as the KD of TbPLK with TbCentrin2, a known in vitro substrate of TbPLK (de Graffenried et al., 2008). In vitro GST pull-down showed that TbCentrin2 was able to bring down wild-type TbPLK, TbPLK-K70R and the KD, but not the PBD (Fig. 6A). We then tested the interaction between TbPLK and p110, another substrate of TbPLK (our unpublished data), by GST pull-down, and found that p110 was also capable of precipitating the KD, but not the PBD, from the cell lysate (Fig. 6B). Finally, to test whether phosphorylated TbCentrin2 is able to bind to the PBD, GST pull-down was performed to precipitate TbCentrin2 expressed in trypanosome cells. We found that the KD, but not the PBD, was able to pull down phosphorylated TbCentin2 (Fig. 6C). When the cell lysate was treated with Lambda protein phosphatase (λPPase), the KD was still able to bring down the dephosphorylated TbCentrin2 (Fig. 6C). It should be noted that GST-PBD exhibited weak interaction with both phosphorylated and dephosphorylated TbCentrin2, but as the GST control also showed non-specific binding to TbCentrin2, the weak association between GST-PBD and TbCentrin2 could be attributed to non-specific binding (Fig. 6C).

Together, these results confirm that the KD of TbPLK mediates the interaction with its substrates TbCentrin2 and p110. It also suggests that priming phosphorylation of TbCentrin2 is not required for binding to the KD.

It has been well documented that an intramolecular interaction between the KD and the PBD of human PLK1 inhibits PLK1 activity and that T-loop phosphorylation prevents PBD binding and therefore relieves this inhibition (Jang et al., 2002a; Lowery et al., 2005). The PBD of TbPLK appears to be different from that of other Plks (see above), which raised the interesting question of whether it can bind to the KD and inhibit TbPLK activity. Through yeast two-hybrid assays, we found that the PBD and the KD of TbPLK do interact, but both domains are also capable of interacting with themselves in yeasts (Fig. 7A). Additionally, the kinase-dead mutant TbPLK-K70R, but not the wild-type TbPLK, also interacts with itself in yeasts (Fig. 7A).

To confirm these interactions, GST pull-down was carried out. Unlike in yeast cells, in which TbPLK exhibits weak interactions with the KD and the PBD (Fig. 7A), TbPLK was able to bring down wild-type and kinase-dead TbPLKs, the KD and the PBD from trypanosome lysate (Fig. 7B). Similarly, the KD was also capable of pulling down all four proteins, whereas the PBD only precipitated the KD and the PBD from the lysate (Fig. 7B). Collectively, these results suggest that TbPLK is capable of forming an intermolecular dimer through interactions between the KDs, the PBDs, or between the KD and the PBD. However, the interaction between the KD and the PBD detected by both yeast two-hybrid and GST pull-down suggests a potential intramolecular association between the two domains.

Intramolecular binding of the PBD to the KD might inhibit the kinase activity of TbPLK. To test whether the kinase activity of TbPLK is inhibited by PBD binding, an in vitro kinase assay was performed. Whereas the activity of wild-type TbPLK was slightly decreased in the presence of purified recombinant PBD, the activity of the KD of TbPLK was significantly inhibited by the PBD (Fig. 7C). By contrast, the activity of the phosphomimic mutant TbPLK-T198D was unaffected by the PBD (Fig. 7C). These results suggest that the PBD plays an essential role in inhibiting the activity of TbPLK and that this inhibition is relieved when Thr198 in the activation loop is phosphorylated.

In this paper we investigated the structure–function relationship of TbPLK by ectopically expressing various TbPLK mutants in trypanosome cells that are deficient in endogenous TbPLK. This allowed us to identify a few residues that are essential for TbPLK function (Figs 1, 2) and to examine their requirement for TbPLK localization (Figs 1, 2 and 3). Although there is concern about potential off-target effects as well as the possible effect on the expression of the immediate downstream gene when performing RNAi against the 3′-UTR of TbPLK, we found that the deficiency caused by RNAi of the TbPLK 3′-UTR can be readily complemented by ectopic expression of wild-type TbPLK, but not the kinase-dead TbPLK (Fig. 1), indicating that RNAi against the 3′-UTR of TbPLK is specific. Previous reports have also confirmed the specificity of RNAi through targeting the 3′-UTR in trypanosomes (Rusconi et al., 2005; Ralston et al., 2011).

Like its homologs in yeasts and metazoa, TbPLK is also activated by phosphorylation of a universally conserved threonine residue in the activation loop of the KD. This occurs on Thr198 of TbPLK (Fig. 2), which is equivalent to Thr210 in human PLK1 and Thr201 in Xenopus Plx1 (Qian et al., 1999; Jang et al., 2002b). Abolishing this phosphorylation abrogated TbPLK activity and failed to complement TbPLK deficiency, whereas mimicking constitutive phosphorylation restored TbPLK activity and rescued TbPLK deficiency (Fig. 2A,B,D). Aurora A kinase is responsible for phosphorylating Thr210 in human PLK1 (Macurek et al., 2008; Seki et al., 2008a). However, trypanosomes do not express an Aurora A-like kinase (Tu et al., 2006), and the only functional Aurora-like kinase, TbAUK1, acts as a member of the chromosomal passenger complex that is localized to chromatin, kinetochores, central spindle and the anterior tip of the new FAZ (Li et al., 2008b; Li et al., 2008a; Li et al., 2009). Throughout the cell cycle, TbAUK1 and TbPLK do not colocalize at the anterior tip of the new FAZ (Li et al., 2010) and do not interact with each other (our unpublished data), suggesting that TbAUK1 is unlikely to be responsible for Thr198 phosphorylation. In human cells, aurora A and PLK1 localize to centrosomes and promote mitotic entry (Macurek et al., 2008; Seki et al., 2008a). Trypanosomes, however, do not have centrioles, but instead possess the basal body as the microtubule-organizing center despite the fact that the basal body does not play any role in spindle formation in trypanosomes (Ogbadoyi et al., 2000). Interestingly, earlier in the cell cycle, TbPLK is concentrated in the basal body (de Graffenried et al., 2008) and appears to control basal body duplication/segregation (Hammarton et al., 2007). This suggests that TbPLK could be activated by a novel protein kinase in the basal body. We are currently trying to identify this kinase.

Phosphorylation of another threonine residue in the activation loop of TbPLK, Thr202, appears to be also required for TbPLK activation (Fig. 2). As with Thr198, the corresponding protein kinase remains to be identified. It is known that phosphorylation of a conserved serine residue (Ser137 in human PLK1 and Ser128 in Xenopus Plx1) also confers kinase activation (Lee and Erikson, 1997; Smits et al., 2000; Jang et al., 2002b), but in TbPLK the equivalent residue of Ser137/Ser128 is replaced by Thr125. Whether Thr125 is phosphorylated and whether it contributes to TbPLK activation require further investigation. In addition to Thr210 and Ser137, in PLK1 Ser326 and Ser335 are also phosphorylated (Wind et al., 2002; Santamaria et al., 2011); however, the corresponding sites in TbPLK are occupied by Gln318 and Ala327, respectively. Conversely, TbPLK is phosphorylated on a cluster of serine and threonine sites (Ser462, Thr465, Thr466, Thr468 and Thr469) between the KD and the PBD (Nett et al., 2009), but at the corresponding positions in human PLK1 all these residues are replaced by non-serine/threonine residues. Evidently, the in vivo phosphorylation pattern of TbPLK is drastically different from that of human PLK1, suggesting the presence of distinct mechanisms for regulating TbPLK activity and function in trypanosomes.

Although the PBD appears to be necessary for TbPLK localization, the PBD alone is not sufficient to target TbPLK to the anterior tip of the new FAZ (Fig. 5). This contrasts significantly with the PBD of yeast and metazoan Plk homologs, as this is essential and sufficient for localizing Plk to various subcellular locations (Lee et al., 1998; Song et al., 2000; Jang et al., 2002a; Reynolds and Ohkura, 2003; Hanisch et al., 2006). It has been well established that the PBD-dependent localization can be attributed to the substrates of Plks that bind to the PBD. However, we found that the KD of TbPLK, but not the PBD, binds to its substrates TbCentrin2 and p110 (Fig. 6A,B). Moreover, both phosphorylated and dephosphorylated TbCentrin2 interact with the KD but not the PBD (Fig. 6C), suggesting that the interaction of TbPLK with TbCentrin2 does not require priming phosphorylation of TbCentrin2. Our findings argue that the PBD is not necessarily required for substrate binding, at least for TbCentrin2 and p110 as the only known substrates. This conclusion is further supported by the absence in the PBD of TbPLK of the equivalent residues of Trp414, Leu490, His538 and Lys540 in the human PLK1 PBD (supplementary material Fig. S4), all of which make direct contact with phosphopeptides and are each required for PLK1 localization and function (Cheng et al., 2003; Elia et al., 2003b; Garcia-Alvarez et al., 2007). Moreover, we found that mutations in the adjacent residues, such as Trp557, His710 and Lys712, in the PBD of TbPLK did not impair the localization of TbPLK (Fig. 3), thus ruling out any possibility that these residues compensate in the absence of equivalents to Trp414, His538 and Lys540 in TbPLK. However, it should be noted that Drosophila Polo kinase has been shown to interact with Map205 without priming phosphorylation of this substrate, and that the PBD, although required, is not sufficient for this interaction (Archambault et al., 2008). Additionally, the yeast Plk homolog Cdc5 was found to bind to Dbf4 via a PBD surface distinct from that used to bind phosphoproteins (Chen and Weinreich, 2010). These observations suggest that Plks in other systems interact with their substrates by more than one mechanism, some of which might be conserved with TbPLK.

Despite lacking direct involvement in substrate binding, the PBD apparently regulates the activity of TbPLK by binding to its KD and inhibiting its activity (Fig. 7), indicating a conserved role of the PBD in regulating Plk activity. The inhibition through PBD binding to the KD is relieved when TbPLK is activated by phosphorylation of Thr198 in the activation loop (Fig. 7C). This may release the PBD from the KD so that the partner proteins of TbPLK can bind to the PBD. Lowery et al. proposed a ‘distributive phosphorylation’ model in which Plk is targeted through PBD-mediated binding to a phosphorylated scaffold or docking protein at certain subcellular locations, and then the KD of Plk binds to and phosphorylates its substrates at these locations (Lowery et al., 2005). If this mechanism is present in trypanosomes, the PBD of TbPLK could be involved in binding to the scaffold or docking proteins. It should be noted that although TbPLK displays a subcellular localization pattern that differs from its yeast and human counterparts, it is also targeted to various subcellular structures, such as the basal bodies, the bilobed structure adjacent to the Golgi, and the anterior tip of the new FAZ (Kumar and Wang, 2006; de Graffenried et al., 2008; Umeyama and Wang, 2008). Further efforts to identify the scaffold or docking proteins located at these subcellular structures and to investigate their potential roles in TbPLK targeting could help elucidate the regulatory pathway that mediates TbPLK localization and function in trypanosomes.

The procyclic form of T. brucei strain 29-13 (Wirtz et al., 1999) was cultured at 27°C in SDM-79 medium supplemented with 10% fetal bovine serum (Atlanta Biologicals), 15 μg/ml G418 and 50 μg/ml hygromycin B. Cells were routinely diluted when the density reached 5×10⁶/ml.

A 549 bp fragment from the 3′-UTR of TbPLK was amplified from genomic DNA and cloned into the pZJM-PAC vector. The resulting construct, pZJM-TbPLK-UTR-PAC, was electroporated into the 29-13 cell line as described (Li et al., 2008b). Transfectants were selected under 1.0 μg/ml puromycin and cloned by limiting dilution. To overexpress TbPLK, TbPLK was cloned into the pLew100-3HA-Phleo vector. Mutation of Lys70, Thr198, Thr202, Trp557, Phe561, His710 and Lys712 was performed by site-directed mutagenesis (Agilent Technologies). The resulting constructs were electroporated into the 29-13 and pZJM-TbPLK-UTR-PAC cell lines. Transfectants were selected under 2.5 μg/ml phleomycin and cloned. To induce RNAi and overexpression, 1.0 μg/ml tetracycline was added to the culture media, and cell growth was monitored daily.

Total RNA was purified from T. brucei cells using TRIzol reagent (Invitrogen). Northern blot was performed as described (Li et al., 2003). The same blot was stripped and reprobed for α-tubulin as a loading control. Semi-quantitative RT-PCR was performed as described (Li and Wang, 2003). Briefly, total RNA was treated with DNase I, and then 200 ng DNA-free RNA was used to produce first-strand cDNAs with Superscript II reverse transcriptase (Invitrogen). TbPLK and actin were co-amplified in the same PCR reaction.

Full-length TbCentrin2 was cloned into the pGEX-4T-3 vector (GE Healthcare). Recombinant GST-TbCentrin2 was expressed in E. coli BL21 cells and purified through a column of glutathione Sepharose 4B beads (GE Healthcare). Purified GST-TbCentrin2 was dialyzed against 50 mM Tris-Cl, pH 7.6, 50 mM NaCl. Wild-type and various mutant TbPLKs were immunoprecipitated from trypanosome cells by anti-HA antibody. Purified GST-TbCentrin2 was incubated with immunoprecipitated kinases in kinase buffer (10 mM HEPES, pH 7.5, 50 mM NaCl, 10 mM MgCl₂, 1 mM DTT) containing 1 μCi [γ-³²P]ATP for 60 minutes at room temperature. Reactions were stopped by adding 1×SDS-PAGE sampling buffer and boiled for 5 minutes. Proteins were separated by SDS-PAGE and the gel exposed to X-ray film. Equal loading of TbCentrin2 was verified by Coomassie Blue staining of the gel after exposure, and immunoprecipitated wild-type and mutant TbPLKs were detected by western blot with anti-HA antibody. To inhibit TbPLK activity, 10 μM GW843286X (Tocris Bioscience), an inhibitor of human PLK1 (Lansing et al., 2007) and trypanosome TbPLK (Li et al., 2010), was added to the kinase reaction. To test the effect of the PBD on the kinase activity of TbPLK, 1 μg GST-PBD was added to the kinase reaction.

Sequences encoding TbPLK and TbPLK-K70R were cloned into the pET41 vector (Novagen). KD, PBD, TbCentrin2 and the C-terminal fragment of p110 (p110-C) were cloned into pGEX-4T-3. The p110 protein sequence is available from the corresponding author on request. Recombinant proteins were expressed in E. coli BL21 and purified through a column of glutathione Sepharose 4B beads. The beads were incubated with T. brucei lysate expressing 3HA-tagged proteins, which was prepared by incubating the cells with immunoprecipitation (IP) buffer (25 mM Tris-Cl, pH 7.6, 100 mM NaCl, 1 mM DTT, 1% Nonidet P40, protease inhibitor cocktail) on ice for 30 minutes and cleared by centrifugation (Li et al., 2008a). The beads were then washed five times with IP buffer, and proteins were eluted by boiling the beads in SDS-PAGE sampling buffer. Eluted proteins were fractionated on SDS-PAGE and immunoblotted with anti-HA monoclonal antibody as described (Li and Wang, 2008). GST was used as the negative control.

For western blot, procyclic cells expressing 3HA-tagged proteins were spun down, boiled in SDS-PAGE sampling buffer and cleared by centrifugation. The lysate was fractionated by SDS-PAGE, transferred onto a PVDF membrane and immunoblotted with anti-HA monoclonal antibody (Sigma-Aldrich) or anti-TbPLK polyclonal antibody (de Graffenried et al., 2008). The same blot was stained with Coomassie Blue or reblotted with anti–α-tubulin antibody to serve as the loading control.

Cells were washed with PBS and fixed in 4% paraformaldehyde. The fixed cells were adhered to coverslips pretreated with poly-l-lysine, and then incubated in blocking buffer (1% BSA and 0.1% Triton X-100 in PBS) for 1 hour at room temperature. Cells were then incubated with FITC-conjugated anti-HA antibody or anti-GRASP antibody (He et al., 2004) or YL1/2 antibody (Kilmartin et al., 1982) diluted in PBS containing 1% BSA. After three washes with wash buffer (0.1% Triton X-100 in PBS), the slides were mounted in Vectashield mounting medium (Vector Labs) containing DAPI and examined with a fluorescence microscope.

HeLa cells were cultured in DMEM medium (Invitrogen) supplemented with 10% fetal bovine serum at 37°C in 5% CO₂. To overexpress TbPLK-EYFP in HeLa cells, TbPLK-EYFP was cloned into the pcDNA3.1 vector. Transfection of HeLa cells was performed using the FuGENE 6 transfection kit (Roche). Plasmid DNA (4 μg) was mixed with 200 μl OPTI-MEM medium and then mixed with another 200 μl OPTI-MEM medium containing 4 μl FuGENE 6 transfection reagent. The mixture was incubated at room temperature for 15 minutes before adding to HeLa cells cultured on poly-l-lysine-coated coverslips in 2 ml OPTI-MEM medium. After 5–6 hours, the transfection mixture was removed, and cells were cultured in DMEM medium containing 400 μg/ml G418. After 24 hours, cells were fixed in 4% paraformaldehyde, mounted in Vectashield mounting medium containing DAPI, and visualized under a fluorescence microscope.

Sequences encoding full-length TbPLK and its kinase-dead mutant TbPLK-K70R, as well as the KD and the PBD of TbPLK, were each cloned into the pGADT7 vector (Clontech) for expression of Gal4 activation domain fusion proteins (prey) or into the pGBKT7 vector (Clontech) for expression of Gal4-binding domain fusion proteins (bait). Prey plasmids were transformed into yeast strain AH109 (mating type a), whereas bait plasmids were transformed into strain Y187 (mating type α). The strains carrying different combinations of bait and prey plasmids were obtained by mating the haploids and then plating on SD-Leu-Trp plates. Each combination strain was spotted in three 10-fold serial dilutions onto SD-Leu-Trp and SD-Leu-Trp-His plates; yeast growth on the latter indicates interaction between the bait and prey proteins.

We thank Dr George A. M. Cross of Rockefeller University for providing the 29-13 cell line and pLew100 vector; Dr Paul Englund of Johns Hopkins Medical School for providing pZJM vector; and Dr Jianping Jin of the University of Texas Medical School at Houston for providing the HeLa cell line, pcDNA3.1 vector and human PLK1 cDNA clone. Anti-TbPLK and anti-GRASP antibodies are generous gifts from Drs Graham Warren and Christopher de Graffenried of Max F. Perutz Laboratories in Austria. We are very grateful to the two anonymous reviewers for their suggestions and comments.