Mitotic effects of a constitutively active mutant of the Xenopus polo-like kinase Plx1

During mitosis the Xenopus polo-like kinase 1 (Plx1) plays key roles in the activation of Cdc25C, in spindle assembly, and in cyclin B degradation. Previous work has shown that the activation of Plx1 requires phosphorylation on serine and threonine residues. In the present work, we demonstrate that replacement of Ser-128 or Thr-201 with a negatively charged aspartic acid residue (S128D or T201D) elevates Plx1 activity severalfold and that replacement of both Ser-128 and Thr-201 with Asp residues (S128D/T201D) increases Plx1 activity approximately 40-fold. Microinjection of mRNA encoding S128D/T201D Plx1 into Xenopus oocytes induced directly the activation of both Cdc25C and cyclin B-Cdc2. In egg extracts T201D Plx1 delayed the timing of deactivation of Cdc25C during exit from M phase and accelerated Cdc25C activation during entry into M phase. This supports the concept that Plx1 is a "trigger" kinase for the activation of Cdc25C during the G(2)/M transition. In addition, during anaphase T201D Plx1 reduced preferentially the degradation of cyclin B2 and delayed the reduction in Cdc2 histone H1 kinase activity. In early embryos S128D/T201D Plx1 resulted in arrest of cleavage and formation of multiple interphase nuclei. Consistent with these results, Plx1 was found to be localized on centrosomes at prophase, on spindles at metaphase, and at the midbody during cytokinesis. These results demonstrate that in Xenopus laevis activation of Plx1 is sufficient for the activation of Cdc25C at the initiation of mitosis and that inactivation of Plx1 is required for complete degradation of cyclin B2 after anaphase and completion of cytokinesis.     

Identification of phosphorylation sites in the polo-like kinases Plx1 and Plk1 by a novel strategy based on element and electrospray high resolution mass spectrometry

A novel strategy for the determination of protein phosphorylation sites is described and applied to the polo-like kinases Plx1 (Xenopus laevis) and Plk1 (Homo sapiens). The strategy comprises the sequential application of the following techniques: proteolytic digestion, capillary liquid chromatography (LC)-inductively coupled plasma mass spectrometry with phosphorus detection, capillary LC-electrospray mass spectrometry and electrospray tandem mass spectrometry. In this approach, phosphopeptides are generated, their elution time in capillary LC is determined, candidate phosphopeptides at the corresponding elution times are identified, and positive identification and sequencing of phosphopeptides is performed in the last step of the analysis. Using this technique, Ser25/26, Ser326, and Ser340 were identified as phosphorylation sites in recombinant Plx1, and Ser340 was identified as the major phosphorylation site in a kinase-dead mutant of Plx1 expressed in okadaic acid-treated Sf9 insect cells. A site corresponding to Ser326 in Plx1 was also shown to be phosphorylated in the human polo-like kinase Plk1 (Ser335). Element mass spectrometry with phosphorus detection provides a quantitative phosphorylation profile of all phosphorylation sites accessible by LC.     

A feedback loop in the polo-like kinase activation pathway

The Xenopus polo-like kinase Plx1 plays important roles during entry into and exit from mitosis (M phase). Previous studies revealed that Plx1 is activated by phosphorylation on serine and threonine residues, and purification of an activating enzyme from mitotic Xenopus egg extracts led to cloning and characterization of Xenopus polo-like kinase kinase (xPlkk1), which can phosphorylate and activate Plx1 in vitro. In the present study, a positive feedback loop between Plx1 and xPlkk1 was shown to result in each kinase phosphorylating and activating the other. Sequencing of radiolabeled xPlkk1 after phosphorylation by Plx1 in vitro identified three phosphorylation sites each spaced three amino acids apart, two of which have the consensus acidic-X-pSer-hydrophobic described for other polo-like kinase substrates. In addition, endogenous xPlkk1 in oocytes was phosphorylated on these sites in M phase but not in interphase. A mutant xPlkk1 in which these three amino acids were changed to alanine (xPlkk1(SA3)) was unable to be phosphorylated or activated in vitro by Plxl. Depletion of Plx1 from oocyte extracts prior to stimulation of the G(2)/M transition blocked the activation of xPlkk1, but depletion of xPlkk1 before stimulation did not block Plx1 activation. These results indicate that xPlkk1 may function downstream as a target of Plx1 rather than as an upstream activating kinase during the G(2)/M transition.     

Phosphorylation of threonine 210 and the role of serine 137 in the regulation of mammalian polo-like kinase

The mammalian polo-like kinase (Plk) plays a critical role in M-phase progression. Plk is phosphorylated and activated by an upstream kinase(s), which has not yet been identified in mammalian cells. Phosphopeptide mapping and phosphoamino acid analyses of Plk labeled in vivo and phosphorylated in vitro by Xenopus polo-like kinase kinase-1 (xPlkk1) or by lymphocyte-oriented kinase, its most closely related mammalian enzyme, indicate that Thr-210 is a major phosphorylation site in activated Plk from mitotic HeLa cells. Although the amino acid sequence surrounding Ser-137 is similar to that at Thr-210 and is conserved in Plk family members, Ser-137 is not detectably phosphorylated in mitotic mammalian cells or by xPlkk1 in vitro. Nevertheless, the substitution of either Thr-210 or Ser-137 with Asp (T210D or S137D) elevates the kinase activity of Plk. The kinase activity of the double mutant S137D/T210D is not significantly different from that of T210D or S137D, demonstrating that substitution of both residues does not have an additive effect on Plk activity. Expression of the S137D mutant construct arrested HeLa cells in early S-phase with slightly separated centrosomes, whereas cells expressing wild type and T210D were arrested or delayed in M-phase. These data indicate that the Ser-137 may have an unexpected and novel role in the function of Plk.     

Identification of β-catenin as a novel substrate of polo-like kinase 1

Polo-like kinase 1 (Plk1) is a serine/threonine kinase that plays an important role in M phase progression by regulating various downstream substrates via phosphorylation. Here, we identified β-catenin as a novel substrate of Plk1 and determined that Ser-718 is a phosphorylation site for Plk1 by using a phospho-specific antibody that cross-reacts with Plk1-dependent phosphorylation sites. Ser-718 of β-catenin was directly phosphorylated by recombinant Plk1 in vitro, with the phosphorylation signal in cells increasing with overexpression of Plk1 and decreasing when endogenous Plk1 was depleted by small interfering RNA. The phosphorylation at Ser-718 was correlated with the cell cycle-dependent expression of Plk1 which reached a maximum in M phase. We also confirmed that there is a physical interaction between β-catenin and Plk1 using coimmunoprecipitation and a GST pull-down assay. These results demonstrate that β-catenin is a physiological substrate of Plk1 in cells, which may provide a novel insight into the role of β-catenin in M phase.     

Polo-like kinase 1 and Chk2 interact and co-localize to centrosomes and the midbody

Chk2 is a protein kinase intermediary in DNA damage checkpoint pathways. DNA damage induces phosphorylation of Chk2 at multiple sites concomitant with activation. Chk2 phosphorylated at Thr-68 is found in nuclear foci at sites of DNA damage (1). We report here that Chk2 phosphorylated at Thr-68 and Thr-26 or Ser-28 is localized to centrosomes and midbodies in the absence of DNA damage. In a search for interactions between Chk2 and proteins with similar subcellular localization patterns, we found that Chk2 coimmunoprecipitates with Polo-like kinase 1, a regulator of chromosome segregation, mitotic entry, and mitotic exit. Plk1 overexpression enhances phosphorylation of Chk2 at Thr-68. Plk1 phosphorylates recombinant Chk2 in vitro. Indirect immunofluorescence (IF) microscopy revealed the co-localization of Chk2 and Plk1 to centrosomes in early mitosis and to the midbody in late mitosis. These findings suggest lateral communication between the DNA damage and mitotic checkpoints.     

Uncoupling anaphase-promoting complex/cyclosome activity from spindle assembly checkpoint control by deregulating polo-like kinase 1

Polo-like kinase 1 (Plk1) plays a role in numerous events in mitosis, but how the multiple functions of Plk1 are separated is poorly understood. We studied regulation of Plk1 through two putative phosphorylation residues, Ser-137 and Thr-210. Using phospho-specific antibodies, we found that Thr-210 phosphorylation precedes Ser-137 phosphorylation in vivo, the latter occurring specifically in late mitosis. We show that expression of two activating mutants of these residues, S137D and T210D, results in distinct mitotic phenotypes. Whereas expression of both phospho-mimicking mutants as well as of the double mutant leads to accelerated mitotic entry, further progression through mitosis is dramatically different: the T210D mutant causes a spindle assembly checkpoint-dependent delay, whereas the expression of the S137D mutant or the double mutant results in untimely activation of the anaphase-promoting complex/cyclosome (APC/C) and frequent mitotic catastrophe. Using nonphosphorylatable Plk1-S137A and Plk1-T210A mutants, we show that both sites contribute to proper mitotic progression. Based on these observations, we propose that Plk1 function is altered at different stages of mitosis through consecutive posttranslational events, e.g., at Ser-137 and Thr-210. Furthermore, our data show that uncontrolled Plk1 activation can uncouple APC/C activity from spindle assembly checkpoint control.     

The polo-like kinase Plx1 is required for M phase exit and destruction of mitotic regulators in Xenopus egg extracts.

Polo-like kinases (Plks), named after the Drosophila gene product polo, have been implicated in the regulation of multiple aspects of mitotic progression, including the activation of the Cdc25 phosphatase, bipolar spindle formation and cytokinesis. Genetic analyses performed in yeast and Drosophila suggest a function for Plks at late stages of mitosis, but biochemical data to support such a function in vertebrate organisms are lacking. Here we have taken advantage of Xenopus egg extracts for exploring the function of Plx1, a Xenopus Plk, during the cell cycle transition from M phase to interphase (I phase). We found that the addition of a catalytically inactive Plx1 mutant to M phase-arrested egg extracts blocked their Ca2+-induced release into interphase. Concomitantly, the proteolytic destruction of several targets of the anaphase-promoting complex and the inactivation of the Cdc2 protein kinase (Cdk1) were prevented. Moreover, the M to I phase transition could be abolished by immunodepletion of Plx1, but was restored upon the addition of recombinant Plx1. These results demonstrate that the exit of egg extracts from M phase arrest requires active Plx1, and they strongly suggest an important role for Plx1 in the activation of the proteolytic machinery that controls the exit from mitosis.     

Cloning and characterization of Plx2 and Plx3, two additional Polo-like kinases from Xenopus laevis

Members of the family of Polo-like kinases are implicated in the regulation of cell cycle progression in all eukaryotes. In Xenopus laevis, only one member of this family, Plx1, has previously been described. Here we report the cloning and characterization of X. laevis Plx2 and Plx3, the likely homologs of mammalian Plk2 (Snk) and Plk3 (Fnk/Prk), respectively. RNA expression studies indicate that all three Xenopus Plks are present in both oocytes and unfertilized eggs. Further analysis by in situ hybridization revealed that Plx1 RNA is ubiquitously expressed in early embryos, but shows more restricted expression at later stages. In contrast, Plx2 and Plx3 expression is highly restricted in both early and late-stage embryos. Using Plx-specific antisera, Plx1 and Plx3 polypeptides could readily be detected on immunoblots of oocyte and egg extracts. Both Plx1 and Plx3 protein levels remained virtually constant during oocyte maturation. However, whereas Plx1 is more active in M phase than in I phase (P. Descombes and E. A. Nigg (1998) EMBO J. 17, 1328-1335), Plx3 protein and activity levels remained constant upon release of meiotic metaphase II-arrested egg extracts into interphase. Finally, microinjection of in vitro-transcribed RNAs for Plx1, Plx2, and Plx3 increased the rate of progesterone-induced oocyte maturation, and concomitantly, all three kinases became activated. Conversely, overexpression of the corresponding catalytically inactive kinases delayed maturation. This suggests that, at least in oocytes, all three kinases may be regulated by similar mechanisms, and they may also share common substrates. However, the strikingly restricted pattern of expression of Plx2 and Plx3 observed in embryos strongly suggests that individual Plk family members perform at least partly distinct functions at later stages of development.     

Novel in vitro and in vivo phosphorylation sites on protein phosphatase 1 inhibitor CPI-17

CPI-17 is a protein phosphatase 1 (PP1) inhibitor that has been shown to act on the myosin light chain phosphatase. CPI-17 is phosphorylated on Thr-38 in vivo, thus enhancing its ability to inhibit PP1. Thr-38 has been shown to be the target of several protein kinases in vitro. Originally, the expression of CPI-17 was proposed to be smooth muscle specific. However, it has recently been found in platelets and we show in this report that it is endogenously phosphorylated in brain on Ser-128 in a domain unique to CPI-17. Ser-128 is within a consensus phosphorylation site for protein kinase A (PKA) and calcium calmodulin kinase II. However, these two kinases do not phosphorylate Ser-128 in vitro but phosphorylate Ser-130 and Thr-38, respectively. The kinase responsible for Ser-128 phosphorylation remains to be identified. CPI-17 has strong sequence similarity with PHI-1 (which is also a phosphatase inhibitor) and LimK-2 kinase. The novel in vivo and in vitro phosphorylation sites (serines 128 and 130) are in a region/domain unique to CPI-17, suggesting a specific interaction domain that is regulated by phosphorylation.     

Plk1 Protein Phosphorylates Phosphatase and Tensin Homolog (PTEN) and Regulates Its Mitotic Activity during the Cell Cycle

PTEN is a well known tumor suppressor through the negative regulation of the PI3K signaling pathway. Here we report that PTEN plays an important role in regulating mitotic timing, which is associated with increased PTEN phosphorylation in the C-terminal tail and its localization to chromatin. Pulldown analysis revealed that Plk1 physically interacted with PTEN. Biochemical studies showed that Plk1 phosphorylates PTEN in vitro in a concentration-dependent manner and that the phosphorylation was inhibited by Bi2635, a Plk1-specific inhibitor. Deletional and mutational analyses identified that Plk1 phosphorylated Ser-380, Thr-382, and Thr-383, but not Ser-385, a cluster of residues known to affect the PTEN stability. Interestingly, a combination of molecular and genetic analyses revealed that only Ser-380 was significantly phosphorylated in vivo and that Plk1 regulated the phosphorylation, which was associated with the accumulation of PTEN on chromatin. Moreover, expression of phospho-deficient mutant, but not wild-type PTEN, caused enhanced mitotic exit. Taken together, our studies identify Plk1 as an important regulator of PTEN during the cell cycle.     

Phosphorylation of TPX2 by Plx1 enhances activation of Aurora A

Entry into mitosis requires the activation of mitotic kinases, including Aurora A and Polo-like kinase 1 (Plk1). Increased levels of these kinases are frequently found associated with human cancers, and therefore it is imperative to understand the processes leading to their activation. We demonstrate that TPX2, but neither Ajuba nor Inhibitor-2, can activate Aurora A directly. Moreover, Plx1 can induce Aurora A T-loop phosphorylation indirectly in vivo during oocyte maturation. We identify Ser204 in TPX2 as a Plx1 phosphorylation site. Mutating Ser204 to alanine decreases activation of Aurora A, whereas a phosphomimetic Asp mutant exhibits enhanced activating ability. Finally, we show that phosphorylation of TPX2 with Plx1 increases its ability to activate Aurora A. Taken together, our data indicate that Plx1 promotes activation of Aurora A, most likely through TPX2. In light of the current literature, we propose a model in which Plx1 and Aurora A activate each other in a positive feedback loop.     

Polo-Like Kinase 1 in the Life and Death of Cancer Cells

The polo-like kinase family plays a vital role in many cell cycle related events. The family includes mammalian Plk1, Snk (Plk2), and Fnk/Prk (Plk3), Xenopus laevis Plx1, Drosophila polo, fission yeast Plo1, and budding yeast Cdc5. These enzymes, in addition to a conserved kinase domain at the N-terminus, have highly conserved sequences called polo-box(s) in the non-catalytic C-terminal domain.1 Genetic and biochemical experiments with several different organisms have documented that polo-like kinases are involved in many aspects of the cell cycle, such as activation of Cdc2, centrosome assembly and maturation, activation of the anaphase-promoting complex (APC) during the metaphase-anaphase transition, and cytokinesis.(1-3)     

Site-specific phosphorylation differentiates active from inactive forms of the human T-cell leukemia virus type 1 Tax oncoprotein

The human T-cell leukemia virus type 1 oncoprotein Tax is a phosphoprotein with a predominately nuclear subcellular localization that accomplishes multiple functions via protein-protein interactions. It has been proposed that regulation of this protein's pleiotropic functions may be accomplished through phosphorylation of specific amino acid residues. We have conducted a phosphoryl mapping of mammalian-expressed Tax protein using a combination of affinity purification, liquid chromatography tandem mass spectrometry, and site-directed substitution mutational analysis. We achieved physical coverage of 77% of the Tax sequence and identified four novel sites of phosphorylation at Thr-48, Thr-184, Thr-215, and Ser-336. Previously identified potential serine phosphorylation sites at Ser-10, Ser-77, and Ser-274 could not be confirmed by mass spectrometry. The functional significance of these novel phosphorylation events was evaluated by mutational analysis and subsequent evaluation for activity via both CREB and NF-kappaB-responsive promoters. Our results demonstrate that phosphorylation at Thr-215 is associated with loss of both Tax functions, phosphorylation at Thr-48 was specifically deficient for activation via NF-kappaB, and phosphorylation at Thr-184 and Ser-336 had no effect on these Tax functions. Semiquantitation of phosphopeptides revealed that the majority of Tax was phosphorylated at Thr-48, Thr-184, Thr-215, and Ser-336, whereas only a minor population of Tax was phosphorylated at either Ser-300 or Ser-301. These results suggest that both positive and negative phosphorylation signals result in the maintenance of a subfraction of Tax as "active" protein.     

Phosphorylation of protein phosphatase inhibitor-1 by Cdk5

Protein phosphatase inhibitor-1 is a prototypical mediator of cross-talk between protein kinases and protein phosphatases. Activation of cAMP-dependent protein kinase results in phosphorylation of inhibitor-1 at Thr-35, converting it into a potent inhibitor of protein phosphatase-1. Here we report that inhibitor-1 is phosphorylated in vitro at Ser-67 by the proline-directed kinases, Cdk1, Cdk5, and mitogen-activated protein kinase. By using phosphorylation state-specific antibodies and selective protein kinase inhibitors, Cdk5 was found to be the only kinase that phosphorylates inhibitor-1 at Ser-67 in intact striatal brain tissue. In vitro and in vivo studies indicated that phospho-Ser-67 inhibitor-1 was dephosphorylated by protein phosphatases-2A and -2B. The state of phosphorylation of inhibitor-1 at Ser-67 was dynamically regulated in striatal tissue by glutamate-dependent regulation of N-methyl-d-aspartic acid-type channels. Phosphorylation of Ser-67 did not convert inhibitor-1 into an inhibitor of protein phosphatase-1. However, inhibitor-1 phosphorylated at Ser-67 was a less efficient substrate for cAMP-dependent protein kinase. These results demonstrate regulation of a Cdk5-dependent phosphorylation site in inhibitor-1 and suggest a role for this site in modulating the amplitude of signal transduction events that involve cAMP-dependent protein kinase activation.     

Plx1 is required for chromosomal DNA replication under stressful conditions

Polo-like kinase (Plk)1 is required for mitosis progression. However, although Plk1 is expressed throughout the cell cycle, its function during S-phase is unknown. Using Xenopus laevis egg extracts, we demonstrate that Plx1, the Xenopus orthologue of Plk1, is required for DNA replication in the presence of stalled replication forks induced by aphidicolin, etoposide or reduced levels of DNA-bound Mcm complexes. Plx1 binds to chromatin and suppresses the ATM/ATR-dependent intra-S-phase checkpoint that inhibits origin firing. This allows Cdc45 loading and derepression of DNA replication initiation. Checkpoint activation increases Plx1 binding to the Mcm complex through its Polo box domain. Plx1 recruitment to chromatin is independent of checkpoint mediators Tipin and Claspin. Instead, ATR-dependent phosphorylation of serine 92 of Mcm2 is required for the recruitment of Plx1 to chromatin and for the recovery of DNA replication under stress. Depletion of Plx1 leads to accumulation of chromosomal breakage that is prevented by the addition of recombinant Plx1. These data suggest that Plx1 promotes genome stability by regulating DNA replication under stressful conditions.     

Identification of a consensus motif for Plk (Polo-like kinase) phosphorylation reveals Myt1 as a Plk1 substrate

Plk1 (Polo-like kinase 1), an evolutionarily conserved serine/threonine kinase, is crucially involved in multiple events during the M phase. Here we have identified a consensus phosphorylation sequence for Plk1, by testing the ability of systematically mutated peptides derived from human Cdc25C to serve as a substrate for Plk1. The obtained results show that a hydrophobic amino acid at position +1 carboxyl-terminal of phosphorylated Ser/Thr and an acidic amino acid at position -2 are important for optimal phosphorylation by Plk1. We have then found that Myt1, an inhibitory kinase for MPF, has a number of putative phosphorylation sites for Plk1 in its COOH-terminal portion. While wild-type Myt1 (Myt1-WT) served as a good substrate for Plk1 in vitro, a mutant Myt1 (Myt1-4A), in which the four putative phosphorylation sites are replaced by alanines, did not. In nocodazole-treated cells, Myt1-WT, but not Myt1-4A, displayed its mobility shift in gel electrophoresis, due to phosphorylation. These results suggest that Plk1 phosphorylates Myt1 during M phase. Thus, this study identifies a novel substrate for Plk1 by determining a consensus phosphorylation sequence by Plk1.     

A novel role for Cdk1/cyclin B in regulating B-raf activation at mitosis

MAPK activity is important during mitosis for spindle assembly and maintenance of the spindle checkpoint arrest. We previously identified B-Raf as a critical activator of the MAPK cascade during mitosis in Xenopus egg extracts and showed that B-Raf activation is regulated in an M-phase-dependent manner. The mechanism that mediates B-Raf activation at mitosis has not been elucidated. Interestingly, activation of 95-kDa B-Raf at mitosis does not require phosphorylation of Thr-599 and Ser-602 residues (Thr-633 and Ser-636 in Xenopus B-Raf), previously shown to be essential for B-Raf activation by Ras. Instead, we provide evidence for Cdk1/cyclin B in mediating mitotic activation of B-Raf. In particular, Cdk1/cyclin B complexes associate with B-Raf at mitosis in Xenopus egg extracts and contribute to its phosphorylation. Mutagenesis and in vitro kinase assays demonstrated that Cdk1/cyclin B directly phosphorylates B-Raf at Serine-144, which is part of a conserved Cdk1 preferential consensus site (S(144)PQK). Importantly, phosphorylation of Ser-144 is absolutely required for mitotic activation of B-Raf and subsequent activation of the MAPK cascade. However, substitution of a phospho-mimicking amino acid at Ser-144 failed to produce a constitutive active B-Raf indicating that, in addition of Ser-144 phosphorylation, other regulatory events may be needed to activate B-Raf at mitosis. Taken together, our data reveal a novel cell cycle mechanism for activating the B-Raf/MEK/MAPK cascade.     

Hierarchical phosphorylation at N-terminal transformation-sensitive sites in c-Myc protein is regulated by mitogens and in mitosis.

The N-terminal domain of the c-Myc protein has been reported to be critical for both the transactivation and biological functions of the c-Myc proteins. Through detailed phosphopeptide mapping analyses, we demonstrate that there is a cluster of four regulated and complex phosphorylation events on the N-terminal domain of Myc proteins, including Thr-58, Ser-62, and Ser-71. An apparent enhancement of Ser-62 phosphorylation occurs on v-Myc proteins having a mutation at Thr-58 which has previously been correlated with increased transforming ability. In contrast, phosphorylation of Thr-58 in cells is dependent on a prior phosphorylation of Ser-62. Hierarchical phosphorylation of c-Myc is also observed in vitro with a specific glycogen synthase kinase 3 alpha, unlike the promiscuous phosphorylation observed with other glycogen synthase kinase 3 alpha and 3 beta preparations. Although both p42 mitogen-activated protein kinase and cdc2 kinase specifically phosphorylate Ser-62 in vitro and cellular phosphorylation of Thr-58/Ser-62 is stimulated by mitogens, other in vivo experiments do not support a role for these kinases in the phosphorylation of Myc proteins. Unexpectedly, both the Thr-58 and Ser-62 phosphorylation events, but not other N-terminal phosphorylation events, can occur in the cytoplasm, suggesting that translocation of the c-Myc proteins to the nucleus is not required for phosphorylation at these sites. In addition, there appears to be an unusual block to the phosphorylation of Ser-62 during mitosis. Finally, although the enhanced transforming properties of Myc proteins correlates with the loss of phosphorylation at Thr-58 and an enhancement of Ser-62 phosphorylation, these phosphorylation events do not alter the ability of c-Myc to transactivate through the CACGTG Myc/Max binding site.