Mitosis-specific MPM-2 phosphorylation of DNA topoisomerase IIalpha is regulated directly by protein phosphatase 2A

Recent results suggest a role for topoIIalpha (topoisomerase IIalpha) in the fine-tuning of mitotic entry. Mitotic entry is accompanied by the formation of specific phosphoepitopes such as MPM-2 (mitotic protein monoclonal 2) that are believed to control mitotic processes. Surprisingly, the MPM-2 kinase of topoIIalpha was identified as protein kinase CK2, otherwise known as a constitutive interphase kinase. This suggested the existence of alternative pathways for the creation of mitotic phosphoepitopes, different from the classical pathway where the substrate is phosphorylated by a mitotic kinase. In the present paper, we report that topoIIalpha is co-localized with both CK2 and PP2A (protein phosphatase 2A) during interphase. Simultaneous incubation of purified topoIIalpha with CK2 and PP2A had minimal influence on the total phosphorylation levels of topoIIalpha, but resulted in complete disappearance of the MPM-2 phosphoepitope owing to opposite sequence preferences of CK2 and PP2A. Accordingly, short-term exposure of interphase cells to okadaic acid, a selective PP2A inhibitor, was accompanied by the specific appearance of the MPM-2 phosphoepitope on topoIIalpha. During early mitosis, PP2A was translocated from the nucleus, while CK2 remained in the nucleus until pro-metaphase thus permitting the formation of the MPM-2 phosphoepitope. These results underline the importance of protein phosphatases as an alternative way of creating cell-cycle-specific phosphoepitopes.     

Interactions between protein kinase CK2 and Pin1. Evidence for phosphorylation-dependent interactions

The peptidyl-prolyl isomerase Pin1 interacts in a phosphorylation-dependent manner with several proteins involved in cell cycle events. In this study, we demonstrate that Pin1 interacts with protein kinase CK2, an enzyme that generally exists in tetrameric complexes composed of two catalytic CK2 alpha and/or CK2 alpha' subunits together with two regulatory CK2 beta subunits. Our results indicate that Pin1 can interact with CK2 complexes that contain CK2 alpha. Furthermore, Pin1 can interact directly with the C-terminal domain of CK2 alpha that contains residues that are phosphorylated in vitro by p34(Cdc2) and in mitotic cells. Substitution of the phosphorylation sites of CK2 alpha with alanines resulted in decreased interactions between Pin1 and CK2. The other catalytic isoform of CK2, designated CK2 alpha', is not phosphorylated in mitotic cells and does not interact with Pin1, but a chimeric protein consisting of CK2 alpha' with the C terminus of CK2 alpha was phosphorylated in mitotic cells and interacts with Pin1, further implicating the phosphorylation sites in the interaction. In vitro, Pin1 inhibits the phosphorylation of Thr-1342 on human topoisomerase II alpha by CK2. Topoisomerase II alpha also interacts with Pin1 suggesting that the effect of Pin1 on the phosphorylation of Thr-1342 could result from its interactions with CK2 and/or topoisomerase II alpha. As compared with wild-type Pin1, isomerase-deficient and WW domain-deficient mutants of Pin1 are impaired in their ability to interact with CK2 and to inhibit the CK2-catalyzed phosphorylation of topoisomerase II alpha. Collectively, these results indicate that Pin1 and CK2 alpha interact and suggest a possible role for Pin1 in the regulation of topoisomerase II alpha. Furthermore, these results provide new insights into the functional role of the mitotic phosphorylation of CK2 and provide a new mechanism for selectively regulating the ability of CK2 to phosphorylate one of its mitotic targets.     

The MPM-2 antibody inhibits mitogen-activated protein kinase activity by binding to an epitope containing phosphothreonine-183.

Mitogen-activated protein (MAP) kinases are a family of serine/threonine kinases implicated in the control of cell proliferation and differentiation. We have found that activated p42mapk is a target for the phosphoepitope antibody MPM-2, a monoclonal antibody that recognizes a cell cycle-regulated phosphoepitope. We have determined that the MPM-2 antibody recognizes the regulatory region of p42mapk. Binding of the MPM-2 antibody to active p42mapk in vitro results in a decrease in p42mapk enzymatic activity. The MPM-2 phosphoepitope can be generated in vitro on bacterially expressed p42mapk by phosphorylation with either isoform of MAP kinase kinase (MKK), MKK1, or MKK2. Analysis of p42mapk proteins mutated in their regulatory sites shows that phosphorylated Thr-183 is essential for the binding of the MPM-2 antibody. MPM-2 binding to Thr-183 is affected by the amino acid present in the other regulatory site, Tyr-185. Substitution of Tyr-185 with phenylalanine results in strong binding of the MPM-2 antibody, whereas substitution with glutamic acid substantially diminishes MPM-2 antibody binding. The MPM-2 phosphoepitope antibody recognizes an amino acid domain incorporating the regulatory phosphothreonine on activated p42mapk in eggs during meiosis and in mammalian cultured cells during the G0 to G1 transition.     

At least two kinases phosphorylate the MPM-2 epitope during Xenopus oocyte maturation

MPM-2 antigens, a discrete set of phosphoproteins that contain similar phosphoepitopes (the MPM-2 epitope), are associated with various mitotically important structures. The central mitotic regulator cdc2 kinase has been proposed to induce M-phase by phosphorylating many proteins which might include the MPM-2 antigens. To clarify the relationship of cdc2 kinase and the MPM-2 antigens, we developed an in vitro assay that enabled us to specifically detect the kinases that phosphorylate the MPM-2 epitope (ME kinases) in crude cell extracts. Two different ME kinase activities were identified in unfertilized Xenopus eggs, neither of which was cdc2 kinase, but both appeared to be activated by the introduction of cdc2 kinase into oocytes or oocyte extract. The two ME kinases differed in molecular size, substrate specificity, peptide components, and MPM-2 reactivity. The larger one, ME kinase-H, phosphorylated several MPM-2 antigens, while the smaller one, ME kinase-L, phosphorylated mainly one. We purified ME kinase-L to near homogeneity by sequential chromatography and showed that it has the characteristics of the 42-kD microtubule-associated protein (MAP) kinase. Our results support the previous finding that MAP kinase is activated during Xenopus oocyte maturation and suggest that MAP kinase may contribute to oocyte maturation induction by phosphorylating one subtype of MPM-2 epitope.     

Unusual phyletic distribution of peptidases as a tool for identifying potential drug targets

Profound changes in the phosphorylation state of many proteins occur during mitosis. It is well established that many of these mitotic phosphorylations are carried out by archetypal mitotic kinases that are activated only during mitosis, shifting the equilibrium of kinases and phosphatases towards phosphorylation. However, many studies have also detailed the phosphorylation of proteins at mitosis by kinases that are constitutively active throughout the cell cycle. In most cases, it is uncertain how kinases and phosphatases that appear to be constitutively active can induce phosphorylations specifically at mitosis. In this issue of the Biochemical Journal, Escargueil and Larsen provide evidence of an interesting alternative mechanism to attain specific mitotic phosphorylation. A mitosis-specific phosphorylation site in DNA topoisomerase IIalpha, which is recognized by the MPM-2 antibody, is phosphorylated by protein kinase CK2. The authors found that phosphorylation of this site is suppressed during interphase due to competing dephosphorylation by protein phosphatase 2A. Interestingly, protein phosphatase 2A is excluded from the nucleus during early mitosis, allowing CK2 to phosphorylate topoisomerase IIalpha. It is possible that similar mechanisms are used to regulate the phosphorylation of other proteins.     

Dietary and hypothyroid hypercholesterolemia induces hepatic apolipoprotein E expression in the rat: direct role of cholesterol

Ser-473 is solely phosphorylated in vivo in the tail region of neurofilament L (NF-L). With peptides including the native phosphorylation site, it was not possible to locate responsible kinases. We therefore adopted full-length dephosphorylated NF-L as the substrate, and employed MALDI/TOF (matrix-assisted laser desorption and ionization/time of flight) mass spectrometry and a site-specific phosphorylation-dependent antibody recognizing Ser-473 phosphorylation. The antibody showed that casein kinase I (CK I) as well as casein kinase II (CK II) phosphorylated Ser-473 in vitro, while neither GSK-3beta nor calcium/calmodulin-dependent protein kinase II did so. However, the mass spectra of the tail fragments of the phosphorylated NF-L indicated that CK II was the kinase mediating Ser-473 phosphorylation in vitro as opposed to CK I, because CK I phosphorylated another site as well as Ser-473 in vitro. The antibody also demonstrated that NF-L phosphorylated at Ser-473 was abundant in the neuronal perikarya of the rat cortex, indicating that phosphorylation of Ser-473 may take place there. This result may support the suggestion that CK II is the kinase responsible for Ser-473 phosphorylation. Despite many reports showing that CK I mediates phosphorylation of neurofilaments, CK II may phosphorylate NF-L in vivo.     

cdc25 is one of the MPM-2 antigens involved in the activation of maturation-promoting factor.

MPM-2 antigens, a discrete set of phosphoproteins that contain similar phosphoepitopes recognized by the monoclonal antibody MPM-2, are phosphorylated during M-phase induction. Our previous studies suggested that certain MPM-2 antigens are involved in the appearance of maturation-promoting factor (MPF) activity. Because the central mitotic regulator cdc2 kinase has been shown to exhibit MPF activity, we explored the possibility that certain MPM-2 antigens are regulators of cdc2 kinase. We found that MPM-2 binding of its antigens would inhibit the autoamplification of cdc2 kinase in Xenopus oocytes and interfere with cyclin-activation of cdc2 kinase in Xenopus interphase egg extract. Immunodepletion of MPM-2 antigens from cyclin-induced M-phase egg extract caused the inactivation of cdc2 kinase, which was accompanied by an inhibitory phosphorylation of p34cdc2 on Thr 14 and Tyr 15, indicating that at least one MPM-2 antigen is a positive regulator of p34cdc2 dephosphorylation. We then showed that cdc25 from M-phase arrested egg extract is an MPM-2 antigen. These results suggest that phosphorylation of the epitope recognized by MPM-2 may be a crucial event in the activation of cdc25 and that the kinase(s) that phosphorylates this MPM-2 epitope may be an important regulator of cdc2 kinase activation.     

有丝分裂期MPM-2磷蛋白家族及其调控因子

由于有丝分裂磷蛋白单克隆抗体(mitotic phosphoprotein monoclonal antibogy-2,MPM-2)识别很大一组在M期特异磷酸化的蛋白质,因而引起研究者的极大兴趣。该抗体亦成为研究M期功能性磷蛋白的重要工具。多个MPM-2磷蛋白已经定位在各种有丝分裂装置上,如中心体、着丝粒、纺锤体、染色体骨架及中体等,显示MPM-2磷蛋白在调控有丝分裂进程方面所扮演的重要角色。结合作者的工作和这一领域的相关研究,概述了MPM-2磷蛋白的细胞周期分布特点及其在细胞有丝分裂期进程中的重要功能,重点介绍MPM-2磷酸化位点的研究状况以及到目前为止已鉴定的MPM-2磷蛋白家族成员的基本情况,并总结了MPM-2磷蛋白相关调控因子的研究进展。     

Identification of peptides mimicking the ligands of proteins phosphorylated by protein kinase CK2

Synthetic peptides containing a phosphorylation site for protein kinase CK2 were used to investigate their binding properties to other peptides/proteins. The aim of this work was to find an efficient procedure to search for these peptide/protein ligands. The goal was successfully achieved through screening of random peptide libraries displayed on phage. Peptides corresponding to the amino terminal region of topoisomerase I were synthesized in both phosphorylated and unphosphorylated form and used to screen the libraries. Four of the selected sequences were also tested for their reactivity with synthetic peptides corresponding to the carboxy terminal region of the largest subunit of RNA polymerase II. The positive reaction detected supports the hypothesis that the isolated sequences may represent mimics of ligands of proteins phosphorylated by protein kinase CK2.     

Threonine phosphorylation is associated with mitosis in HeLa cells

Phosphorylation and dephosphorylation of proteins play an important role in the regulation of mitosis and meiosis. In our previous studies we have described mitosis-specific monoclonal antibody MPM-2 that recognizes a family of phosphopeptides in mitotic cells but not in interphase cells. These peptides are synthesized in S phase but modified by phosphorylation during G2/mitosis transition. The epitope for the MPM-2 is a phosphorylated site. In this study, we attempted to determine which amino acids are phosphorylated during the G2-mitosis (M) transition. We raised a polyclonal antibody against one of the antigens recognized by MPM-2, i.e. a protein of 55 kDa, that is present in interphase cells but modified by phosphorylation during mitosis. This antibody recognizes the p55 protein in both interphase and mitosis while it is recognized by the monoclonal antibody MPM-2 only in mitotic cells. Phosphoamino acid analysis of protein p55 from 32P-labeled S-phase and M-phase HeLa cell extracts after immunoprecipitation with anti-p55 antibodies revealed that threonine was extensively phosphorylated in p55 during G2-M but not in S phase, whereas serine was phosphorylated during both S and M phases. Tyrosine was not phosphorylated. Identical results were obtained when antigens recognized by MPM-2 were subjected to similar analysis. As cells completed mitosis and entered G1 phase phosphothreonine was completely dephosphorylated whereas phosphoserine was not. These results suggest that phosphorylation of threonine might be specific to some of the mitosis-related events.     

MPM-12: a monoclonal antibody that predominantly stains mitotic cells and recognizes a protein kinase.

The monoclonal antibody MPM-12, raised by using partially purified extract of mitotic HeLa cells as the immunogen, preferentially stains the cytoplasm of mitotic cells by indirect immunofluorescence without exhibiting any species specificity. On immunoblots, MPM-12 recognizes three bands, of 155, 88, and 68 kDa, in mitotic HeLa cell extract but only the 68-kDa band in interphase cell extract. The 68-kDa band seems to be associated with chromatin while the other two are not. All three MPM-12 reactive peptides are phosphorylated, and the phosphorylation seems to be required for MPM-12 reactivity. The MPM-12 immunocomplexes exhibit autophosphorylating and histone H1 kinase activity.     

Phosphorylation of the human cell proliferation-associated nucleolar protein p120

The human cell proliferation-associated nucleolar protein p120 was found in a variety of human cancer specimens but not in most normal resting cells. Polyclonal antibodies raised against bacterially expressed p120 were used to immunoprecipitate the p120 protein isolated from 32P-labeled HeLa cells. The p120 protein was phosphorylated at serine, threonine and tyrosine residues. A tryptic peptide map showed it contained three labeled peptides. One of these peptides comigrated with a p120 peptide phosphorylated in vitro by casein kinase II. This peptide was phosphorylated in vitro both at Ser-181 and Thr-185. This region is juxtaposed to the epitope site recognized by the anti-p120 monoclonal antibody.     

Cell-cycle-dependent changes in labelling of specific phosphoproteins by the monoclonal antibody MPM-2 in plant cells

The MPM-2 antibody, which recognizes a mitosis-specific phosphorylated epitope, has been used to study cell-cycle-related proteins in partially synchronized cell suspension cultures and root meristem cells. Immunofluorescence revealed that the epitope recognized by MPM-2 is located in the nucleus in interphase cells. In mitotic cells, MPM-2 labels the prophase nucleus, the spindle and some cytoplasmic components. The relative amount of the epitope changes significantly during the cell cycle. Labelling is lowest in G1 and S-phase cells and increases 2–3-fold during G2. Prophase and metaphase show four to five times the labelling of G1 cells. Labelling decreases rapidly after metaphase and is at a very low level by telophase. One- (1-D) and two-dimensional (2-D) immunoblots showed that MPM-2 labels a family of phosphorylated proteins. The labelling shows significant cell cycle dependence. Subfractionation shows at least one of these proteins is a component of the detergent-insoluble cytoskeleton cell fraction. This component is resolved on 2-D immunoblots to two to three spots of slightly different isoelectric point, possibly charge isomers, at a relative molecular mass of approximately 65 kDa. The same spots are labelled by IFA, an antibody against intermediate filament proteins. Another three of the spots at lower relative molecular mass are labelled on 2-D immunoblots of the nuclear matrix fraction.     

Distribution of cytoskeletal proteins sharing a conserved phosphorylated epitope

A group of antigens related by their reactivity with monoclonal antibodies MPM-1 and MPM-2 appear as cells enter mitosis. These antibodies bind to a phosphorylated epitope on certain proteins, and therefore the antigens are presumed to be a group of phosphoproteins. A subset of these proteins has been shown previously to be components of mitotic microtubule organizing centers in PtK1 cells. We present here evidence that the mitosis-specific appearance of these phosphoproteins is a phenomenon common to all eukaryotic cells. The MPM reactive phosphoproteins were localized to mitotic spindle poles regardless of whether the spindle formed in the cytoplasm after nuclear envelope breakdown (open mitosis) or within the nucleus (closed mitosis). This reactivity was not dependent upon the presence of centrioles at the spindle poles. Proteins that contained the phosphorylated epitope were not, however, restricted to mitotic cells. Cells of neuronal derivation and flagellated cells showed specific localization of MPM antibody to the microtubule network and basal bodies respectively. On immunoblots, the MPM antibody reacted with brain MAP-1 among a number of other phosphoproteins. The identification of microtubule-associated protein (MAP)-1 correlates with the localization of the antibody to microtubules of neuroblastoma cells. These results suggest, that different phosphoprotein molecules detected by the MPM antibody may be specific for different mitotic microtubule organizing centers, basal bodies, and other specialized cytoskeletal structures; and the presence of a related phosphorylated domain on these proteins may be important for their proper function and/or interaction with microtubules.     

Pea DNA topoisomerase I is phosphorylated and stimulated by casein kinase 2 and protein kinase C

DNA topoisomerase I catalyzes the relaxation of superhelical DNA tension and is vital for DNA metabolism; therefore, it is essential for growth and development of plants. Here, we have studied the phosphorylation-dependent regulation of topoisomerase I from pea (Pisum sativum). The purified enzyme did not show autophosphorylation but was phosphorylated in an Mg(2+)-dependent manner by endogenous protein kinases present in pea nuclear extracts. This phosphorylation was abolished with calf intestinal alkaline phosphatase and lambda phosphatase. It was also phosphorylated by exogenous casein kinase 2 (CK2), protein kinase C (PKC; from animal sources), and an endogenous pea protein, which was purified using a novel phorbol myristate acetate affinity chromatography method. All of these phosphorylations were inhibited by heparin (inhibitor of CK2) and calphostin (inhibitor of PKC), suggesting that pea topoisomerase I is a bona fide substrate for these kinases. Spermine and spermidine had no effect on the CK2-mediated phosphorylation, suggesting that it is polyamine independent. Phospho-amino acid analysis showed that only serine residues were phosphorylated, which was further confirmed using antiphosphoserine antibody. The topoisomerase I activity increased after phosphorylation with exogenous CK2 and PKC. This study shows that these kinases may contribute to the physiological regulation of DNA topoisomerase I activity and overall DNA metabolism in plants.     

Unique features of HIV-1 Rev protein phosphorylation by protein kinase CK2 ('casein kinase-2')

The HIV-1 Rev transactivator is phosphorylated in vitro by protein kinase CK2 at two residues, Ser-5 and Ser-8; these sites are also phosphorylated in vivo. Here we show that the mechanism by which CK2 phosphorylates Rev is unique in several respects, notably: (i) it is fully dependent on the regulatory, beta-subunit of CK2; (ii) it relies on the integrity of an acidic stretch of CK2 beta which down-regulates the phosphorylation of other substrates; (iii) it is inhibited in a dose-dependent manner by polyamines and other polycationic effectors that normally stimulate CK2 activity. In contrast, a peptide corresponding to the amino-terminal 26 amino acids of Rev, including the phosphoacceptor site, is readily phosphorylated by the catalytic subunit of CK2 even in the absence of the beta-subunit. These data, in conjunction with the observation that two functionally inactive derivatives of Rev with mutations in its helix-loop-helix motif are refractory to phosphorylation, indicate the phosphorylation of Rev by CK2 relies on conformational features of distinct regions that are also required for the transactivator's biological activity.     

Alzheimer's Disease Neurofibrillary Tangles Contain Mitosis-Specific Phosphoepitopes

Abstract: Paired helical filaments (PHFs) are the major components of neurofibrillary lesions present in Alzheimer's disease (AD). PHFs are composed of the microtubule-associated protein (MAP) τ, which is abnormally phosphorylated in AD. Normal fetal τ is also phosphorylated and shares certain phosphoepitopes with PHF-τ. The abnormal phosphorylation of PHF-τ is considered to be involved in the formation of PHFs and subsequent degeneration of AD neurons. We have previously shown that other neuronal MAPs, such as MAP1B, contain mitosis-specific phosphoepitopes. In addition to mitotic cells, these epitopes are also expressed in fetal brain and PC12 cells during differentiation and neurite outgrowth. One hypothesis regarding the etiology of AD involves the reactivation of a fetal-like state and mitotic conditions in selected neurons. To determine if similar mitosis-associated phosphoepitopes appeared in AD, sections of hippocampal tissue were stained for immunoreactivity with antibodies recognizing both τ and mitotic phosphoepitopes. Both the MPM2 mitotic phosphoepitope antibody and the AT8 PHF-τ antibody stained neurofibrillary lesions and colocalized to pyramidal neurons in AD samples. In addition, PHFs isolated from an AD brain reacted with both antibodies. The MPM2 antibody specifically reacted with τ in the isolated PHF fraction but not normal adult τ. In addition, MPM2 failed to react with normal fetal or adult τ obtained from rat brains. The MPM2 antibody also recognized human MAP1B; however, MAP1B was not present in the PHF fraction. Our results indicate that MPM2 recognized a phosphoepitope present on PHF-τ. Because normal fetal or adult rat brain τ did not express the MPM2 epitope, it is likely that this phosphoepitope is specific for the disease state.     

Modulation of protein kinase CK2 activity by fragments of CFTR encompassing F508 may reflect functional links with cystic fibrosis pathogenesis

Deletion of F508 in the first nucleotide binding domain (NBD1) of cystic fibrosis transmembrane conductance regulator protein (CFTR) is the commonest cause of cystic fibrosis (CF). Functional interactions between CFTR and CK2, a highly pleiotropic protein kinase, have been recently described which are perturbed by the F508 deletion. Here we show that both NBD1 wild type and NBD1 DeltaF508 are phosphorylated in vitro by CK2 catalytic alpha-subunit but not by CK2 holoenzyme unless polylysine is added. MS analysis reveals that, in both NBD1 wild type and DeltaF508, the phosphorylated residues are S422 and S670, while phosphorylation of S511 could not be detected. Accordingly, peptides encompassing the 500-518 sequence of CFTR are not phosphorylated by CK2; rather they inhibit CK2alpha catalytic activity in a manner which is not competitive with respect to the specific CK2 peptide substrate. In contrast, 500-518 peptides promote the phosphorylation of NBD1 by CK2 holoenzyme overcoming inhibition by the beta-subunit. Such a stimulatory efficacy of the CFTR 500-518 peptide is dramatically enhanced by deletion of F508 and is abolished by deletion of the II507 doublet. Kinetics of NBD1 phosphorylation by CK2 holoenzyme, but not by CK2alpha, display a sigmoid shape denoting a positive cooperativity which is dramatically enhanced by the addition of the DeltaF508 CFTR peptide. SPR analysis shows that NBD1 DeltaF508 interacts more tightly than NBD1 wt with the alpha-subunit of CK2 and that CFTR peptides which are able to trigger NBD1 phosphorylation by CK2 holoenzyme also perturb the interaction between the alpha- and the beta-subunits of CK2.     

A site-directed antibody that inhibits phosphorylation of the rat-brain sodium channel by cyclic-AMP-dependent protein kinase.

Antibodies were raised against three peptides corresponding to the potential protein phosphorylation sites of rat-brain sodium channels by the cAMP-dependent protein kinase (PKA). One of the antibody against sequence (C561-575) reacted to the channel molecule. This immunoreaction occurred in a sequence-specific manner, as it was inhibited by the antigen peptide itself but not inhibited by two other peptides. Although PKA phosphorylates two synthetic peptides, C561-575 and C681-689, of the three, anti-(C561-575) antibody can only inhibit the phosphorylation of peptide (C561-575). PKA catalyzed the incorporation of 3.1-3.5 mol of phosphates into the alpha subunit of the purified sodium channel. The anti-(C561-575) antibody inhibited the channel phosphorylation by 40%. Digestion of the phosphorylated sodium channel with lysyl endoproteinase yielded four major phosphorylated fragments of 3.5, 5.0, 7.0, and 10 kDa. However, similar digestion of the channel that was phosphorylated in the presence of anti-(C561-575) antibody did not yield the phosphorylated fragment of 3.5 kDa and gave the 7.0 kDa fragment in reducing yield. Inspection of these phosphorylated fragments by the predicted sizes of the peptide fragments containing the five potential phosphorylation sites gives a conclusion that anti-(C561-575) antibody inhibits the phosphorylation on Ser-573 completely, and on either Ser-610 or Ser-623 partially, probably due to their proximity orientation in the tertiary structure.