Phosphorylation of Centrin during the Cell Cycle and Its Role in Centriole Separation Preceding Centrosome Duplication*

Once during each cell cycle, mitotic spindle poles arise by separation of newly duplicated centrosomes. We report here the involvement of phosphorylation of the centrosomal protein centrin in this process. We show that centrin is phosphorylated at serine residue 170 during the G2/M phase of the cell cycle. Indirect immunofluorescence staining of HeLa cells using a phosphocentrin-specific antibody reveals intense labeling of mitotic spindle poles during prophase and metaphase of the cell division cycle, with diminished staining of anaphase and no staining of telophase and interphase centrosomes. Cultured cells undergo a dramatic increase in centrin phosphorylation following the experimental elevation of PKA activity, suggesting that this kinase can phosphorylate centrinin vivo. Surprisingly, elevated PKA activity also resulted intense phosphocentrin antibody labeling of interphase centrosomes and in the concurrent movement of individual centrioles apart from one another. Taken together, these results suggest that centrin phosphorylation signals the separation of centrosomes at prophase and implicates centrin phosphorylation in centriole separation that normally precedes centrosome duplication.

Once during each cell cycle, mitotic spindle poles arise by separation of newly duplicated centrosomes. We report here the involvement of phosphorylation of the centrosomal protein centrin in this process. We show that centrin is phosphorylated at serine residue 170 during the G 2 /M phase of the cell cycle. Indirect immunofluorescence staining of HeLa cells using a phosphocentrin-specific antibody reveals intense labeling of mitotic spindle poles during prophase and metaphase of the cell division cycle, with diminished staining of anaphase and no staining of telophase and interphase centrosomes. Cultured cells undergo a dramatic increase in centrin phosphorylation following the experimental elevation of PKA activity, suggesting that this kinase can phosphorylate centrin in vivo. Surprisingly, elevated PKA activity also resulted intense phosphocentrin antibody labeling of interphase centrosomes and in the concurrent movement of individual centrioles apart from one another. Taken together, these results suggest that centrin phosphorylation signals the separation of centrosomes at prophase and implicates centrin phosphorylation in centriole separation that normally precedes centrosome duplication.
The centrosome is the major microtubule organizing center of the cell, and as such it determines the number and polarity of cytoplasmic microtubules, as well as the general form of the interphase microtubule array (1). Once in each cell cycle, the centrosome is duplicated to give rise to two centrosomes (spindle poles) that organize the microtubule array of the mitotic spindle. Centrosomes consist of a closely associated pair of centrioles and a pericentriolar matrix that surrounds and connects the individual centrioles to one another and to microtubules. Pericentriolar matrix has the properties of a gel as evidenced by the exclusion of small cytoplasmic particles such as ribosomes and membrane vesicles from the centrosome. At three times during the cell cycle centrioles move apart from one another. The first time occurs immediately following mitosis when the centriole pair transiently splits during or just after telophase. At this time the older centriole remains near the cell center while the younger centriole wanders extensively throughout the cytoplasm before returning to reside near its older partner (2). The second separation of the centriole pair occurs later in G 1 phase as cells pass the restriction point and the pair of centrioles become disoriented and slightly separated from one another in the first identifiable event of centrosome duplication (3). The third time centriole pairs move apart from one another occurs as cells enter prophase when newly duplicated centrosomes (each containing a pair of centrioles) separate and migrate to opposite sides of the nucleus where they function as mitotic spindle poles (4). Changes in centrosome function at the time of the G 2 /M transition are regulated by phosphorylation since centrosomes undergo an increase in protein phosphorylation at this time (5)(6)(7)(8). It is likely that specific phosphorylation events also take place at the centrosome at other times during the cell cycle (9). Numerous kinases known to regulate cell cycle progression localize at centrosomes and mitotic spindle poles, including Cdk2 and Cdk4/6, polo-like kinase 1 (PLK1), NIMA-related kinase (Nek2), pEg2, and cAMPdependent kinase (PKA). 1 Several of the centrosome-specific substrates of these kinases have been identified (for recent reviews, see Refs. 9 and 10). Protein phosphorylation has been implicated in a variety of centrosome functions including centrosome duplication, maturation and separation, microtubule nucleation, and specification of cleavage furrow formation (9 -12). Cdk2/cyclin E activity coordinates centrosome duplication with the DNA replication cycle (11)(12)(13)(14)(15).
Centrin is a 20-kDa protein of the EF-hand superfamily of calcium-binding proteins that is a component of centrioles themselves and the surrounding pericentriolar matrix (16 -18). Gene disruption experiments demonstrate that yeast centrin (Cdc31p) is essential for cell viability (19). Additionally, two distinct centrin mutations have been described that result in either failure of centrosome (spindle pole body) duplication and separation resulting in monopolar spindles and cell cycle arrest, or precocious and inappropriate centrosome separation at the time of cell division resulting in cells with the incorrect number of centrioles/basal bodies (20 -22). Three separate human centrin genes that encode and express centrin have been identified (23)(24)(25). A conserved carboxyl-terminal region of centrins from diverse species includes a consensus motif for protein phosphorylation that is typical for serine/threonine kinases (26).
Centrin phosphorylation in lower eukaryotes has been shown to correlate with extension of centrin-containing fibers associated with centrioles/basal bodies (27,28). More recently, aberrant centrin phosphorylation has been demonstrated in human breast tumors that have amplified centrosomes con-taining supernumerary centrioles and/or excess pericentriolar material (29). The objective of this study was to investigate the cell cycle-dependent phosphorylation of centrin and to characterize the specific site of phosphorylation on the protein. Using an antibody that is specific for phosphorylated centrin and standard biochemical methods, we demonstrate that in cultured vertebrate cells, centrin is phosphorylated near its carboxyl terminus at serine residue 170 early in mitosis when the newly duplicated centrosomes separate to give rise to the mitotic spindle poles. The spindle pole localization of phosphocentrin remains high until metaphase and then diminishes to basal levels by telophase. The timing of centrin phosphorylation suggests that phosphorylation of centrin may initiate the separation of duplicated centrosomes in preparation for mitotic spindle formation. Experimental elevation of protein kinase A (PKA) activity in interphase cells also results in the phosphorylation of centrin at serine residue 170 and the concomitant movement of centrioles away from one another in a manner similar to the transient separation of the pair of centrioles that normally occurs preceding centrosome duplication which begins at about the time of the G 1 /S transition. These observations suggest that centrin phosphorylation plays a role in mitotic spindle pole and centriole separation at key stages of the cell cycle.
Cell Culture-HeLa and QT6 cells were grown in 100-mm plates in modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (and 1% chicken serum (Life Technologies, Inc.) for QT6 cultures), 1 mM sodium pyruvate, and 20 mM HEPES, pH 7.3, at 37°C and 5% CO 2 . For immunofluorescence studies, HeLa cells were plated onto 12-mm diameter glass coverslips and used before cells reached confluence. Cells were arrested in S phase of the cell cycle by 2.5 mM thymidine treatment for 16 h or in G 2 /M phase by 1 M nocodazole treatment for 8 h (30). Cell cycle arrest was analyzed by flow cytometry. For in vivo phosphorylation studies, cultured cells were grown as described above except that during the final 4 h of incubation they were washed and then incubated in phosphate-free Dulbecco's modified Eagle's medium (Specialty Media) with dialyzed serum containing the appropriate drug treatments and 0.2 mCi of 32 PO 4 (Amersham Pharmacia Biotech). Because high phosphatase activity in HeLa extracts resulted in substantial loss of centrin phosphorylation, QT6 fibroblasts, which have lower phosphatase activity in cell lysates, were used in the experiments reported here involving 32 PO 4 incorporation in vivo.
Peptide Synthesis-A synthetic peptide HCT-12 (EFLRIMKKTSLY) corresponding to the carboxyl-terminal 12 amino acid residues of human centrin 2 (CETN2, see Table I) was synthesized by Merrifield solid phase on 0.5 mmol of 4-hydroxymethyphenoxy (Wang) copolystyrene resin (Advanced Chemtech, Louisville, KY) using methods described previously (31) (Table I). Each peptide was synthesized with 9-fluorenylmethoxycarbonyl-L-amino acid (Fmoc) derivatives (Advanced Chemtech, Louisville, KY) containing the following side chain protecting groups: N ␣ -boc-S-triphenylmethyl (trityl) for cysteine; O-t-butyl for threonine and tyrosine; N ⑀ -boc for lysine; O-t-butyl ester for glutamic acid; and N␥-2,2,5,7,8-pentamethylchroman-6-sulfonyl for arginine. The HCT peptides were assembled with the Fmoc-L-amino acids on a Rainin PS3 automated peptide synthesizer using PyBOP/N-methylmorpholine as the coupling reagent, and the hydroxyl amino acid serine was incorporated into the peptide as its unprotected side chain derivative Fmoc-Ser-OH. After synthesis, the peptide resins were weighed, divided into two portions, and one portion of each peptide resin was used for phosphorylation of the free serine hydroxyl group by "global" phosphite-triester phosphorylation (32). Both HCT-12 and HCT-P peptides were also synthesized with a free amino-terminal cysteine residue for specific coupling to maleimide activated carriers (keyhole limpet hemocyanin and bovine serum albumin) and chromatography resins.
Following synthesis and phosphorylation, the peptides were deprotected and cleaved from the resin by acidolysis with 93% trifluoroacetic acid containing 1% ethanedithiol, 3% anisole, and 3% ethylmethyl sulfide for 90 min at room temperature. Each deprotected peptide was filtered from the resin and precipitated into cold methyl t-butyl ether. All peptides were purified by reverse-phase high performance liquid chromatography using a Vydac C8 column (2.2 cm ϫ 25 cm, Vydac, Hesperia, CA) in 0.1% trifluoroacetic acid/water and a gradient of 10 to 80% acetonitrile in 0.1% trifluoroacetic acid/water, and collected as single homogeneous peaks. The composition of each peptide was verified by mass spectrometry on a Bio-Ion 20 mass analyzer (PerkinElmer Life Sciences, Forest City, CA).
In Vitro Phosphorylation-Recombinant human CETN2p expressed in Escherichia coli was prepared and purified as described earlier (33). Centrin and HCT-12 (Table I)  ochraceus) for 1-2 h. Phosphorylation reactions were started by addition of kinase and were terminated by freezing in liquid N2 followed by lyophilization.
Phosphopeptide Mapping and Phosphoamino Acid Analysis-Lyophilized phosphorylation reaction mixtures were resolved by 15% SDS-PAGE according to the method of Laemmli (34) or by SDS-Tricine PAGE for centrin peptides according to the method of Schagger and von Jagow (35), and transferred to Immobulin-P membrane (Millipore) according to the methods of Hulen and co-workers (36). The region of the membrane containing phosphocentrin was located by autoradiography, and subjected to tryptic phosphopeptide and phosphoamino acid analysis as described by Boyle and co-workers (37).
Antiserum Production and Affinity Purification of ␣HCT-P Antibodies-The synthetic peptide HCT-P (Table I) was conjugated to keyhole limpet hemocyanin and used to immunize female New Zealand White rabbits. Antisera from the third and fourth bleeds were used for the experiments described in this work. An IgG fraction was prepared using a protein-A-Superose TM column (Amersham Pharmacia Biotech) and the methods supplied by the manufacturer. The IgG fraction was precleared on a Sulfolink TM resin (Pierce Chemical Co.) that had been coupled with HCT (Table I). Following a 12-h incubation at room temperature, unbound Ig was washed from the resin, concentrated, and applied to an affinity column consisting of Sulfolink TM gel to which had been coupled the phosphopeptide HCT-P. The bound IgG fraction was eluted (0.05 M glycine, pH 2.5, 0.1% Triton X-100, 0.15 M NaCl, 0.01 M NaF, 0.01 M Na pyrophosphate, and 0.04% azide), neutralized using 1 M Tris buffer (pH 7.2), concentrated, and stored in PBS containing 0.04% sodium azide. This affinity purified IgG fraction is referred to as ␣HCT-P IgG.
Specificity of affinity purified ␣HCT-P IgG was characterized in four separate analyses (Fig. 3). 1) Specificity against the peptides HCT-P versus HCT was determined by a Western dot blot analysis. The synthetic peptides HCT and HCT-P (200 ng/well) were blotted onto polyvinylidene difluoride membrane using carbonate antigen binding buffer (CB; 0.09 M NaHCO 3 , 0.01 M Na 2 CO 3 , and 0.02% sodium azide, pH 9.4) and a Bio-Dot apparatus (Bio-Rad). Membranes were blocked with 400 l/well of blocking buffer (0.01 M Tris-HCl, pH 7.4, 0.15 M NaCl, 2% bovine serum albumin, 10% normal goat serum, 0.05% Tween 20, and 0.04% sodium azide) and incubated with Ig fractions (diluted according to the figure legend) in 10% fetal bovine serum and 0.04% sodium azide (100 l/well). Bound Ig was detected by incubation with alkaline phosphatase-conjugated goat antibodies against rabbit immunoglobulins (Cappel) using the alkaline phosphate substrate 5-bromo-4-chloro-3indolyl phosphate and nitro blue tetrazolium. 2) Specificity of ␣HCT-P IgG for phosphorylated versus nonphosphorylated centrin was determined by autoradiography and Western blotting of recombinant centrin following in vitro phosphorylation by the catalytic subunit of PKA. 3) Specificity of ␣HCT-P IgG for phosphorylated centrin versus other sub-strates for PKA was determined by autoradiography and Western blotting following in vitro phosphorylation by the catalytic subunit of PKA of an equal molar mixture of recombinant centrin, citrate lyase, Lphenylalanine hydroxylase, and phosphorylase kinase. 4) Specificity of ␣HCT-P IgG for centrin phosphorylated by PKA versus PKC or Cdc2 kinase was determined by autoradiography and Western blotting following in vitro phosphorylation of centrin by each of the respective kinases.
Immunoprecipitation and Western Analysis-Cell lysates were prepared for immunoprecipitation and Western blot analysis as described previously (23). For the experiments described here, extracts were immunoprecipitated using polyclonal serum 26/14-1 that recognizes centrin regardless of its phosphorylation state or the preimmune serum from the same rabbit. For Western blot analysis, centrin was resolved by SDS-PAGE, transferred to Immobulin-P (polyvinylidene difluoride) membrane (Millipore), and fixed with 0.2% glutaraldehyde followed by standard procedures according to Towbin and co-workers (59) and probed using either monoclonal anti-centrin 20H5 that recognizes centrin regardless of phosphorylation state or ␣HCT-P IgG.

Analysis of Centrin Phosphorylation Status during the Cell
Cycle-Immunoprecipitation and Western blot analysis of centrin (20 kDa) from cultured cells is shown in Fig. 1A. Cell cycle analysis and the corresponding phosphorylation status of centrin in untreated (control asynchronously cycling) cells or cells subject to cell cycle arrest by thymidine (S phase) or nocodazole treatment (G 2 /M phase) are shown in Fig. 1, B and C, respectively. Incorporation of 32 PO 4 into centrin was highest in cultures arrested with 4C DNA (G 2 /M) content by nocodazole treatment, while cultures arrested in S phase by thymidine treatment show the lowest levels of incorporation and control (cycling) cultures showed an intermediate level of 32 PO 4 incorporation. These results suggest that centrin phosphorylation is maximum during mitosis.
Phosphorylation of Centrin by PKA in Vitro-Centrins from a variety of species share a highly conserved sequence at their carboxyl termini (Table I) that includes a consensus amino acid motif (KKXS*X) for phosphorylation by PKA (38). Bacterially expressed centrin and a synthetic peptide, HCT-12, corresponding to the carboxyl-terminal 12 amino acids of CETN2 (Table I), are readily phosphorylated by PKA in vitro (Fig. 2). Tryptic digestion and two-dimensional analysis of the phosphorylated products reveals a single negatively charged major and a neutral minor 32 P-labeled tryptic peptide for both the recombinant protein and the synthetic peptide (Fig. 2, B and C,  respectively). The two 32 P-labeled peptide fragments comigrate when tryptic digests of centrin and the synthetic peptide are first mixed and subsequently run together by two-dimensional peptide analysis (Fig. 2D). Phosphoamino acid analysis demonstrates that serine is the 32 P-labeled residue for both the major and minor peptide fragments (Fig. 2, E and F). Comparison of migration of tryptic peptides from centrin with the predicted pattern based on the computer program of Hunter (37) and the fact that there is only one potential serine phosphorylation site in the synthetic HCT peptide, allowed the unequivocal identification of serine residue 170 in the carboxylterminal sequence KKTS P Y as the site for in vitro phosphorylation of centrin by PKA. The major negatively charged phosphopeptide (TS P LY) represents the carboxyl terminus (residues 169 -172) of centrin and the minor neutral peptide fragment (KTS P LY) represents an alternative partial tryptic product (residues 168 -172) resulting from the less efficient trypsin cleavage activity between adjacent lysine residues (37).
Production and Characterization of Antibodies to Phosphorylated Centrin-Because the carboxyl-terminal phosphorylation site is highly conserved among divergent species (Table I) we raised a phosphocentrin-specific antibody to study in vivo phosphorylation of centrin in cultured HeLa cells. We first prepared a 13-amino acid synthetic peptide, HCT-12, which contains the last 12 carboxyl-terminal amino acids of human centrin (residues 160 -172, see Table I) plus an amino-terminal cysteine residue for convenience in coupling via its sulfhydryl group to either immunogenic haptens or to affinity matrices. This peptide was prepared either in a non-phosphorylated form (human centrin carboxyl-terminus, HCT) or with the serine residue synthetically phosphorylated (HCT-P). Rabbit antiserum raised against the phosphorylated peptide HCT-P was found in preliminary experiments to label mitotic spindle poles. In order to obtain antibodies that were uniquely specific for phosphorylated centrin and that showed no cross-reactivity against the nonphosphorylated centrin, we first precleared the IgG fraction against HCT and then affinity purified the preparation using resin-conjugated HCT-P as described under "Experimental Procedures." Analysis using a Western dot blot assay demonstrated exquisite specificity of the affinity purified ␣HCT-P IgG fraction for the phosphorylated HCT-P peptide with little or no reactivity of this IgG fraction against the non-phosphorylated HCT peptide (Fig. 3A). This affinity purified IgG fraction was further tested for its specificity toward the whole protein using recombinant centrin that was first phosphorylated in vitro by PKA using [ 32 P]ATP (Fig. 3B). In this experiment, centrin was incubated with PKA under conditions in which only about half of the protein became phosphorylated. Subsequent analysis by SDS electrophoresis in a 15% polyacrylamide gel resolved centrin into two closely migrating bands of around 20 kDa that could be detected by Western blot analysis using a monoclonal anti-centrin ascites (20H5) that recognizes centrin regardless of its phosphorylation state (Fig. 3B). Autoradiography (Fig. 3B, 32 P-label) demonstrated that the slower migrating centrin band represented the phosphorylated protein while the more rapidly migrating band showed no 32 P label. Analysis of a companion gel transfer prepared with the identical sample, but Western blotted with the affinity purified ␣HCT-P IgG, demonstrated a remarkable specificity for the PKA-phosphorylated form of centrin and showed little if any reactivity with the non-phosphorylated protein (Fig. 3B: ␣HCT-P).
A control experiment was performed in which several known substrates for PKA were combined at equal molar ratios with centrin, followed by in vitro incubation under complete reaction conditions with PKA and [ 32 P]ATP, and subsequent analysis by SDS-PAGE, autoradiography, and Western blotting (Fig. 3C). In this experiment the affinity purified ␣HCT-P IgG specifically and exclusively recognized phosphocentrin and no other PKA-phosphorylated proteins (Fig. 3C, Western (w)), even when present at equivalent levels and having significant 32 P incorporation (Fig. 3C, autoradiograph (a)). A second control experiment was performed in which centrin was phosphorylated by one of three distinct protein kinases: PKA, PKC, or p34 cdc2 kinase (Cdc2) and subsequently analyzed by Western blotting with monoclonal 20H5 or ␣HCT-P IgG (Fig. 3D). Centrin is phosphorylated by all three kinases in vitro (Fig. 3D, 32 P) and analysis by phosphopeptide mapping demonstrated that each kinase tested phosphorylated distinct sites on centrin (not shown). Only PKA phosphorylation resulted in gel mobility retardation of centrin and only the PKA-phosphorylated form of centrin reacted with the affinity purified ␣HCT-P IgG (Fig. 3D, ␣HCT-P). Taken together, these results demonstrate that affinity purified ␣HCT-P IgG specifically recognized centrin only when it was phosphorylated on the extreme carboxylterminal amino acid serine located at residue 170. In addition, this antibody recognized no other PKA-phosphorylated proteins or centrin itself when it was exclusively phosphorylated at sites other than serine residue 170.

Phosphorylated Centrin Localizes at Mitotic Spindle Poles in
HeLa-HeLa cells grown on glass coverslips were processed for indirect immunofluorescence using affinity purified ␣HCT-P IgG. When fields of subconfluent cells were observed by indirect immunofluorescence microscopy a vast majority of the cells showed no apparent labeling above background when stained with ␣HCT-P IgG (Figs. 4C and 5, A and B). Remarkably, however, dividing cells present throughout the microscope field showed clear and intense staining of their mitotic spindle poles (Figs. 4, C-F, and 5, A and B). The specificity of immunostaining of mitotic spindle poles by ␣HCT-P IgG was evaluated following competition of the antibody preparation with the peptides HCT-P and HCT. The affinity purified ␣HCT-P IgG was used at a constant and known dilution (1.5 g/ml IgG), preincubated with known amounts of the phosphorylated peptide HCT-P or the nonphosphorylated peptide HCT, and subsequently used for indirect immunofluorescence labeling of HeLa (Fig. 4, A-F). Preincubation of ␣HCT-P IgG with HCT-P peptide at a molar ratio of 1:1 (antibody:peptide, calculated based on antigen-binding equivalents on the IgG) completely eliminated subsequent staining of prophase and metaphase spindle poles (Fig. 4A), while preincubation at a molar ratio of 1:0.1 resulted in a reduction but not complete elimination of subsequent staining (Fig. 4B), and ␣HCT-P IgG that was not preincubated with any peptide resulted in bright staining of spindle poles (Fig. 4C) as described above. However, preincubation of the ␣HCT-P IgG with the nonphosphorylated peptide HCT showed reduction in staining of mitotic spindle poles at 1:0.1 (Fig. 4F), 1:1 (Fig. 4E), or even 1:1000 (Fig. 4D) 3. Specificity of ␣HCT-P IgG against phosphocentrin. A, dot blot analysis of the affinity purified ␣HCT-P IgG (dilution series) against the non-phosphorylated (HCT) and phosphorylated (HCT-P) peptides. The control in which no primary antibody was used in the immunoprecipitation step is shown at the far right. B, Western blot of a descending dilution series (left to right) of centrin phosphorylated in vitro with PKA and corresponding 32 P autoradiograms of the same gels. Centrin is resolved into two closely migrating bands in this 15% polyacrylamide gel. The Western blot stained with monoclonal 20H5 (upper gel) shows two closely migrating centrin bands and the corresponding autoradiograph demonstrates that only the more slowly migrating band is phosphorylated. The Western blot stained with ␣HCT-P IgG (lower gel) shows that only the phosphocentrin band is recognized by this antibody. C, Western blot (w) using ␣HCT-P IgG and corresponding autoradiograph (a) of an equal molar mixture of recombinant centrin, citrate lyase, L-phenylalanine hydroxylase, and phosphorylase kinase phosphorylated in vitro using PKA and [␥-32 P]ATP. Only centrin at 20 kDa reacts with ␣HCT-P IgG. D, autoradiograph ( 32 P) and Western blots (20H5) and (␣HCT-P) of centrin phosphorylated in vitro by PKA, PKC, and Cdc2 kinase. All three kinases phosphorylated centrin, yet only the PKA product reacts with ␣HCT-P IgG.
IgG:HCT). These results demonstrate that spindle pole labeling by ␣HCT-P IgG is exquisitely specific for the phosphorylated carboxyl-terminal serine residue 170 of centrin.
Careful analysis of a large number of mitotic cells (Fig. 5, A  and B) shows that prophase and metaphase spindle poles stained most intensely with ␣HCT-P antibodies. Staining of anaphase spindle poles was present, albeit at diminished levels, and telophase spindle poles (see Fig. 4E) failed to stain at all. That centrin is present in interphase centrosomes and at the spindle poles during all stages of mitosis is demonstrated by staining with the polyclonal serum 26/14-1 (Fig. 5C). We interpret these indirect immunofluorescence studies to demonstrate that HeLa centrin becomes phosphorylated at serine residue 170 in a cell cycle-specific manner making the carboxylterminal domain recognizable by ␣HCT-P IgG. Since staining by ␣HCT-P IgG is most intense during prophase and metaphase, the phosphorylation state of centrin is likely to be highest at these times.
Activation of PKA Stimulates in Vivo Phosphorylation of Centrin and Centriole Separation in Interphase Cells-The observations presented above do not necessarily demonstrate that PKA is the kinase responsible for phosphorylation of centrin in vivo, nor do they eliminate the possibility that other kinases may also phosphorylate distinct sites on centrin either at the time of mitosis or at other times during the cell cycle. In order to address the former question, cell cultures were incubated under control conditions or under treatment conditions that elevate endogenous cAMP levels to activate PKA. Cultures were allowed to incorporate 32 PO 4 for 4 h prior to treatment for 2 min with the membrane permeable cAMP analog (100 M Bt 2 cAMP) that selectively activates PKA, with or without concomitant treatment with the cAMP phosphodiesterase inhibitor IBMX (1 mM). Cells were lysed, centrin was immunoprecipitated using 26/14, resolved by SDS-PAGE, and subsequently analyzed by Western blotting and autoradiography for centrin phosphorylation. Control cells show low but detectable levels of centrin phosphorylation (Fig. 6), while cells treated with Bt 2 cAMP alone or Bt 2 cAMP and IBMX show substantially increased centrin phosphorylation. The centrin Western blot appears as a closely spaced double band, with the more slowly migrating band corresponding to the major phosphorylated product. A third band also appears following treatment with both Bt 2 c-AMP and IBMX and represents a minor hyperphosphorylated product migrating higher in the gel. This experiment demonstrates that treatment of living cells with cAMP analogs to activate PKA results in an increase in the phosphorylation of centrin. Immunofluorescence staining using ␣HCT-P IgG of HeLa cells similarly treated to elevate PKA activity (in the presence of calyculin A, an inhibitor of protein phosphatases 1 and 2A) demonstrates staining of mitotic cells and shows discrete ␣HCT-P IgG staining of their spindle poles as seen before. Remarkably however, in individual interphase cells treated to elevate PKA activity, centrosomes acquire the phosphocentrin epitope necessary for labeling with ␣HCT-P IgG (Fig. 7, B and C). In addition, the ␣HCT-P IgG-labeled interphase centrosomes appear as two spots separated by 2-4 m, suggesting interphase centrosomes have separated into two discrete entities (Fig. 7, B  and C). Typically, one of the spots is brighter than the other. Control cells (not treated with cAMP analogs) labeled with the polyclonal anticentrin serum 26/14-1 that recognizes centrin regardless of phosphorylation state show a single staining spot of a non-separated centrosome (Fig. 7D). Cent-  6. Stimulation of centrin phosphorylation in cultured cells following activation of PKA. Western blot using monoclonal 20H5 antibody and corresponding 32 P autoradiograph of cell lysates from cultures that were untreated (control) or treated for 2 min to stimulate PKA activity (Bt 2 cAMP) and to inhibit phosphodiesterase activity (Bt 2 cAMP/IBMX). Significantly elevated levels of phosphocentrin are seen following activation of PKA. Double arrows indicate the position of the two closely migrating centrin bands, arrowhead indicates minor hyperphosphorylated centrin band. rosome separation into two discrete spots in interphase cells was seen in 44% (22/50) of the Bt 2 cAMP-treated cells while only 10% (5/50) of control cells showed separated centrosomes. These observations demonstrate that elevation of PKA activity in living cells results in centrin phosphorylation at the carboxyl-terminal serine residue 170 and concomitant separation of the centrosome into two discrete units, suggesting that PKA phosphorylation of centrin is involved in centriole separation.

DISCUSSION
The primary objective of this investigation was to study the in vivo phosphorylation of the centrosomal protein centrin during the cell cycle. Observations presented here demonstrate that centrin serves as a substrate for in vitro phosphorylation by PKA at its carboxyl-terminal serine residue 170, and that a phosphocentrin specific antibody (␣HCT-P IgG) recognized centrin only when it was phosphorylated at this site. Analysis of cultured cells for 32 PO 4 incorporation into centrin demonstrated a significant increase in the level of centrin phosphorylation in G 2 /M phase. Indirect immunofluorescence staining of HeLa cells using the phosphocentrin-specific antibody resulted in intense labeling of mitotic spindle poles during prophase and metaphase of the cell division cycle, while the staining was diminished in anaphase and was absent in telophase and interphase centrosomes. These observations suggest that centrin phosphorylation is largely restricted to the period of the cell cycle at G 2 /M when centrosome separation and mitotic spindle assembly occur. Furthermore, cultured cells showed a dramatic and rapid increase in centrin phosphorylation following experimental treatment (Bt 2 cAMP and IBMX) that elevates PKA activity, suggesting that this kinase can phosphorylate centrin in vivo. Artificial elevation of in vivo cAMP levels resulted in phosphocentrin-specific antibody staining of centrosomes in interphase HeLa cells in addition to the mitotic spindle pole label seen without drug treatment. Surprisingly, elevated cAMP also resulted in precocious separation of centrosomes into two distinct spots that we interpret to represent the displacement of individual centrioles a short distance from one another.
Taken together these results indicate human centrin is phosphorylated near its the carboxyl terminus at serine residue 170 early in mitosis by PKA or another protein kinase with a substrate specificity similar to PKA. Observations which favor the in vivo phosphorylation of centrin by PKA include the following: 1) a conserved PKA consensus phosphorylation motif in the carboxyl terminus of centrin; 2) the in vitro phosphorylation of centrin by PKA at this site; and 3) the stimulation of centrin phosphorylation in vivo by cell permeable cAMP analogs and treatments that reduce phosphodiesterase and phosphatase activity (IBMX and calyculin A, respectively). Moreover, PKA is localized at the centrosome in vertebrate cells, including HeLa (39, 40) through interaction of its regulatory subunit (RII) and the protein kinase A anchoring coiled-coil domains of AKAP450 and the centrosomal structural protein pericentrin (41)(42)(43). Tethering of PKA to protein kinase A anchoring proteins is thought to target the enzyme to the proximity of relevant substrates, thereby conveying spatial specificity to cAMP/PKA signaling.
Numerous studies demonstrate that cellular levels of cAMP in HeLa and other cultured cells are maximal in G 1 of the cell cycle and minimal in G 2 /M (44 -46). High levels of cAMP seen during G 1 phase correspond to the time of transient centriole separation following mitosis. However, the period when centrin phosphorylation levels appear highest (G 2 /M) corresponds to the time when cellular cAMP levels are at their lowest. Furthermore, cAMP has been shown to delay G 2 progression, inhibit cell proliferation, and negatively regulate mitotic Cdc2 kinase activity (47)(48)(49). These observations confound a possible role for PKA-mediated phosphorylation of centrin at the onset of mitosis. It is possible that PKA acting at the centrosome is regulated through local changes in the cAMP pool that are not reflected in the overall cellular levels of cAMP. This could be accomplished through the activation of adenylate cyclase at times when the cAMP pool is low as is seen at G 2 /M (44), and/or through competition between PKA and protein phosphatases (50,51) that are tethered at the centrosome near their substrates. Alternatively, a kinase(s) other than or in addition to PKA may phosphorylate centrin at G 2 /M.
As indicated earlier, ϳ20 kinases have been localized at the centrosome, some only transiently and others throughout the cell cycle (9). As far as kinases are concerned the centrosome is a very crowded place. The cyclin-dependent kinases Cdk2 and Cdk4/6 are of particular interest because of their role in regulation of cell cycle progression and because centrosome duplication has been demonstrated to be dependent on Cdk2/cyclin E or cyclin A activity in Xenopus cell-free extracts and somatic cells, respectively (3,12,15,52). Our studies show that centrin can indeed be phosphorylated by Cdk in vitro, however, at a site that is distinct from serine 170 recognized by the phosphocentrin-specific antibody described in this study.
The functional consequences of centrin phosphorylation are not known. Genetic studies in lower eukaryotes suggest centrin plays an essential role in centrosome duplication and/or sepa- ration (20 -22). Yeast centrin, Cdc31p, is a component of the spindle pole body half-bridge that extends between the nascent and the pre-existing spindle pole body during its duplication, and might play a role in the templated process involved in new spindle pole body formation (53). In higher eukaryotes centrin is a component of centrioles themselves and of fibrous material connecting centrioles/basal bodies to one another (28,54,55). Similarly disposed fibers in vertebrate epithelial cells lengthen and shorten during the centriole duplication cycle (56). Studies in budding yeast have implicated centrin (Cdc31p) in activation of a protein kinase (Kic1p) that is required for cell integrity and bud morphology, in addition to its role in spindle pole body duplication (57). A unifying functional theme has recently emerged (53) wherein the centrins may act as small regulatory molecules that modulate the activity of target proteins in centrosome-associated contractile fiber systems, in a manner similar to that of another member of the calmodulin superfamily, troponin C, which regulates the Ca 2ϩ -dependent contraction of skeletal muscle. In this regard, calcium binding by centrin stimulates centriole-associated fiber contraction while centrin phosphorylation and Ca 2ϩ release are associated with fiber extension (27,28,58). Thus, the observations reported in the present work suggest that centrin phosphorylation may play a role in solation or extension of the gel-like pericentriolar matrix at during G 1 phase when centriole pairs undergo disorientation initiating centrosome duplication and at G 2 /M when newly duplicated centrosomes separate from one another to form the bipolar mitotic spindle.