Identification of a Phosphorylation Site in the Hinge Region of the Human Progesterone Receptor and Additional Amino-terminal Phosphorylation Sites*

We have previously reported the identification of seven in vivo phosphorylation sites in the amino-terminal region of the human progesterone receptor (PR). From our previous in vivo studies, it was evident that several phosphopeptides remained unidentified. In particular, we wished to determine whether human PR contains a phosphorylation site in the hinge region, as do other steroid receptors including chicken PR, human androgen receptor, and mouse estrogen receptor. Previously, problematic trypsin cleavage sites hampered our ability to detect phosphorylation sites in large incomplete tryptic peptides. Using a combination of mass spectrometry and in vitro phosphorylation, we have identified six previously unidentified phosphorylation sites in human PR. Using nanoelectrospray ionization mass spectrometry, we have identified two new in vivo phosphorylation sites, Ser20 and Ser676, in baculovirus-expressed human PR. Ser676 is analogous to the hinge site identified in other steroid receptors. Additionally, precursor ion scans identified another phosphopeptide that contains Ser130-Pro131, a likely candidate for phosphorylation. In vitro phosphorylation of PR with Cdk2 has revealed five additional in vitro Cdk2 phosphorylation sites: Ser25, Ser213, Thr430, Ser554, and Ser676. At least two of these, Ser213 and Ser676, are authentic in vivo sites. We confirmed the presence of the Cdk2-phosphorylated peptide containing Ser213 in PR from in vivolabeled T47D cells, indicating that this is an in vivosite. Our combined studies indicate that most, if not all, of the Ser-Pro motifs in human PR are sites for phosphorylation. Taken together, these data indicate that the phosphorylation of PR is highly complex, with at least 14 phosphorylation sites.

Human progesterone receptor (PR), 1 a ligand-activated transcription factor and member of the steroid receptor superfamily (1), is expressed as two forms: the full-length PR-B and the shorter form, PR-A, which lacks the first 164 amino acids of PR-B (2,3). These proteins differ in their relative ability to activate target genes and in the unique repressor activity of the PR-A isoform (4 -8). Recently, the physiological roles for the different isoforms have been examined by the generation of isoform-specific knockout mice. Interestingly, knockout of the PR-A isoform in mice demonstrated a strong tissue-specific role for PR-A in uterus not shared by PR-B (9). In addition, overexpression of either PR-A or PR-B results in aberrant mammary gland development, and the phenotypes of the two overexpressing lines differ (10,11). Moreover, recent studies have suggested that differential coactivator recruitment may be responsible for the isoform-specific differences observed in transactivation assays (8).
In addition to regulation dependent on isoform, PR is a phosphoprotein (12)(13)(14)(15)(16)(17) whose activity can be regulated by phosphorylation. There is ample evidence that regulation of cell signaling pathways alters the activity and phosphorylation of PR (18 -21) as well as other steroid receptors (22)(23)(24)(25)(26)(27)(28)(29)(30). Some of these changes are due to direct alterations in receptor phosphorylation, whereas others appear to affect associated proteins (29,(31)(32)(33)(34)(35). Additionally, the role of individual sites has been examined for PR. Mutation of Ser 190 results in decreased activity of PR (21). Since different kinases target unique subsets of sites and these phosphorylations may have opposing effects on receptor activity, it is important to know which sites are coregulated by each kinase for mutagenesis studies to determine the impact on receptor function.
We have previously identified seven phosphorylation sites in human PR isolated from 32 P-labeled T47D breast cancer cells. Each of these sites is located in the amino-terminal (A/B) domain. Three of the previously identified sites, Ser 81 , Ser 102 , and Ser 162 , are unique to the B form (15). In vivo labeling of the endogenous PR in T47D cells showed that the PR is phosphorylated basally in the absence of hormone at serines 81, 162, 190, and 400. Upon hormone stimulation, the net phosphorylation of these basal sites is dramatically increased within 5 min of treatment. In addition, hormone induces the phosphorylation of three new phosphorylation sites, serines 102, 294, and 345, in a temporally delayed manner, requiring 2 h for maximal phosphorylation. These sites are referred to as the hormonedependent phosphorylation sites (16). Recombinant human PR expressed in Sf9 insect cells displays the same phosphorylation pattern as PR from hormone-treated T47D cells; however, the hormone dependence is lacking in Sf9 cells (36). Additionally, we have shown that PR is a substrate for several kinases in vitro. We reported previously that three in vivo phosphorylation sites, Ser 162 , Ser 190 , and Ser 400 , are phosphorylated in vitro by the cyclin A-Cdk2 complex (17) and that Ser 81 is phosphorylated by casein kinase II (15).
Although many of the phosphorylation sites have been identified in PR, several phosphotryptic peptide peaks detected in the HPLC analyses of in vivo labeled PR have yet to be identified. Most of the phosphorylation sites in steroid receptors contain Ser/Thr-Pro motifs, and in the case of the chicken PR all four of these motifs are phosphorylated (37,38). Human PR contains 15 of these sequences, several of which have already been identified as phosphorylation sites (15)(16)(17). We were particularly interested in determining whether the conserved Ser-Pro in the hinge region between the hormone and DNA binding domains is phosphorylated as has been described for chicken PR (37), mouse estrogen receptor (39), and human androgen receptors (26). Initial analyses by Sheridan et al. had suggested that this site is not phosphorylated in human PR (13). Arg-Pro and Lys-Pro motifs are not efficiently cleaved by trypsin (40), and PR has an unusually large number of these sequences. Analyses of candidate Ser-Pro phosphorylation sites including Ser 676 in the hinge region reveal that many are located within peptides bounded by these highly resistant cleavage sites. If left uncleaved, these phosphopeptides will be very large and difficult to recover by HPLC using a C 18 reversed-phase column. Using trypsin modified by reductive alkylation, which has decreased susceptibility to autolysis, we have increased the percentage cleavage at these problematic sites, allowing resolution of new phosphopeptides both in PR phosphorylated in vitro by Cdk2 and by mass spectrometry of phosphopeptides isolated from PR expressed in Sf9 cells.
Here we describe the identification of six candidate phosphorylation sites identified either in in vitro phosphorylation experiments using Cdk2 or by mass spectrometry of peptides derived from PR expressed in Sf9 cells. These sites include Ser 676 in the hinge region, which was identified as both an in vitro Cdk2 site and by mass spectrometry. Three additional sites, common to PR-A and PR-B, and two sites unique to the PR-B form have also been found. Finally, an additional site has been localized to a region between amino acids 107 and 159 in the portion of the receptor unique to PR-B. These studies suggest that most, if not all, of the Ser/Thr-Pro motifs in PR are phosphorylated.

EXPERIMENTAL PROCEDURES
Materials-Phosphate-free Dulbecco's modified Eagle's medium and Hanks' balanced salt solution were purchased from Life Technologies, Inc., and fetal bovine serum was from Hyclone (Logan, UT). Radioinert R5020 was obtained from Amersham Pharmacia Biotech. [ 32 P]H 3 PO 4 and [␥-32 P]ATP were obtained from ICN (Irvine, CA). For affinity purification, nickel-NTA resin was purchased from Qiagen, Inc. (Valencia, CA), and glutathione-Sepharose 4B and protein A-Sepharose were obtained from Amersham Pharmacia Biotech. Rabbit anti-mouse antibody was obtained from Zymed Laboratories Inc. (South San Francisco, CA). Phenylisothiocyanate and HPLC grade solvents were purchased from J. T. Baker Inc. Sequencing grade-modified trypsin was bought from Promega Corp. (Madison, WI). The endoproteinases Asp-N, Glu-C, Arg-C, and Lys-C were obtained from Roche Molecular Biochemicals. For manual Edman degradation, triethylamine and trifluoroacetic acid were obtained from Sigma, and the Sequelon-AA reagent kit was purchased from Millipore Corp. (Milford, MA). TLC plates from EM Science (Gibbstown, NJ), ninhydrin from Pierce, and phosphoamino acids (phosphoserine, phosphotyrosine, and phosphothreonine) from Sigma were used for the phosphoamino acid analysis.
Baculovirus-expressed His Tag PR Purification-Infection of Sf9 insect cells with baculovirus encoding His tag PR-B or PR-A and treatment with 200 nM R5020 was performed as previously described (41). Cells from a 500-ml culture were centrifuged to harvest, and pellets were stored at Ϫ80°C until use. Cells were thawed and homogenized in a glass-Teflon homogenizer in 20 ml of His tag homogenization buffer (HHB; 20 mM Tris-HCl, pH 8.0, 350 mM NaCl, 15 mM imidazole, 1 mM B-mercaptoethanol, 10% glycerol) plus 50 mM NaF and a mixture of protease inhibitors (18). The homogenate was centrifuged at 40,000 rpm for 30 min at 4°C. Ni 2ϩ -NTA resin (2 ml packed) was washed in HHB twice and resuspended to 40 ml. The Sf9 extract was then added to the resin and incubated for 1 h at 4°C in a 50-ml conical tube on a rocking platform. The resin was centrifuged and washed with HHB twice before transferring the resin suspension to a 3-ml column. The resin was washed with 30 ml of HHB followed by 30 ml of HHB containing 0.6 M NaCl. Finally, the column was washed with 10 ml of low salt HHB (NaCl concentration reduced to 50 mM in HHB). His tag PR was eluted in elution buffer (low salt HHB containing 250 mM imidazole) and collected as 1-ml fractions. Protein concentrations were determined by Bradford assay. Estimated purity from silver staining of samples run on SDS gels was greater than 95%.
Purification of HA-Cdk2⅐Glutathione S-Transferase-cyclin A Complex-This purification was performed essentially as described previously (42). In brief, Sf9 cells were infected with the baculoviruses encoding HA-Cdk2 and glutathione S-transferase-cyclin A at a 1:1 ratio for 40 h at 27°C. Cells were harvested, washed with phosphate-buffered saline (PBS), and harvested by centrifugation at 2000 rpm. The pellet was then resuspended in 2 volumes of NETN buffer (0.5% Nonidet P-40, 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 5 mM NaF, 30 mM p-nitrophenyl phosphate, 1 g/ml leupeptin, 1 g/ml antipain, 1 mM phenylmethylsulfonyl fluoride) and incubated on ice for 10 min. The extract was prepared by sonication and cleared by centrifugation. Five volumes of kinase buffer (20 mM HEPES, pH 7.5, 15 mM MgCl 2 , 1 mM dithiothreitol, 10 mM NaF, 5 g/ml leupeptin, 1 g/ml antipain, 1 mM phenylmethylsulfonyl fluoride) and 1 mM ATP were added to the extract and incubated at 30°C for 30 min. Glutathione-Sepharose was then added to the lysate for 1 h at 4°C. The beads were washed with NETN and then with 100 mM Tris-HCl, pH 8, containing 150 mM NaCl. To elute the active Cdk2-cyclin A complex, 100 mM Tris-HCl, pH 8, containing 150 mM NaCl and 40 mM glutathione was added to the beads and incubated at 4°C for 15 min. The supernatant containing the kinase complex was removed, and aliquots were frozen at Ϫ80°C.
In Vitro Phosphorylation of His Tag PR-Purified recombinant His tag PR-A or PR-B (1 g) was incubated with 5-10 l of purified Cdk2cyclin A and 50 M [␥-32 P]ATP (37,000 cpm/pmol) in Cdk2 buffer (20 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 ) in a final volume of 40 l for 30 min at 30°C. The reaction was terminated by the addition of 4ϫ SDS-Laemmli sample buffer. Samples were separated on a 6.5% SDS-PAGE gel. The phosphorylated bands corresponding to PR, visualized by autoradiography, were excised and digested with trypsin.
Preparation and Separation of Phosphopeptides-Sequencing grade trypsin modified by reductive alkylation to prevent autolysis (Promega) was reconstituted in acetic acid as described in the manufacturer's instructions. Two g of trypsin were added to each tube containing a gel slice prepared as described previously (15) in 1 ml of 50 mM ammonium bicarbonate, pH 7.8, and incubated overnight at 37°C on a rotating platform. The tryptic digest was dried in a SpeedVac Concentrator® (Savant, Farmingdale, NY), and the phosphopeptides were separated on a C 18 reversed-phase column using a Beckman System Gold HPLC. A linear 0 -45% acetonitrile gradient in 0.1% trifluoroacetic acid was run for 90 min followed by an increase to 100% acetonitrile by 110 min at a flow rate of 1 ml/min. The phosphopeptide peaks were detected using an online radioactive flow detector as described previously. 2-ml fractions were collected, counted by Cerenkov counting, split for analysis, and dried in a SpeedVac.
Manual Edman Degradation-To determine the cycle or position of the phosphoamino acid in the peptide, manual Edman degradation was performed as previously described (43) using the Sequelon-AA Reagent kit. Briefly, the tryptic phosphopeptides were covalently coupled to Sequelon disks and subjected to Edman degradation. Each "cycle" of Edman degradation begins by modifying the first NH 2 -terminal amino acid of the peptide, which destabilizes the first peptide bond. Treatment with trifluoroacetic acid cleaves this bond and releases the amino acid from the disc. Cerenkov counting of the supernatant measures the amount of 32 P released. The disc was then subjected to additional cycles of Edman degradation.
Secondary Endoproteinase Digestions-Cleavage by the endoproteinases Asp-N, Glu-C, Arg-C, and Lys-C was used to further characterize the phosphopeptides. For Asp-N digestion, the sample was resuspended in 50 mM sodium phosphate buffer, pH 8.0, and incubated with 0.2 g of Asp-N for 8 h at 37°C. Glu-C digestion was performed with 1 g of Glu-C in 25 mM ammonium bicarbonate, pH 7.8, for 8 h at 30°C. The typical Arg-C digest was performed by dissolving the dried peptides in 85 l of incubation buffer, adding 10 l of activation solution (provided by the manufacturer) and 5 l (or 0.5 g) of Arg-C, and incubating for 12 h at 37°C. For Lys-C, the sample was redissolved in 100 l of digestion buffer (25 mM Tris-HCl, pH 8.5, 1 mM EDTA) and incubated with 0.2 g of Lys-C at 37°C for 12 h.
Phosphoamino Acid Analysis-Phosphoamino acid analysis was performed as described (44). In brief, phosphopeptide samples containing 4 g of each phosphoamino acid standard (phosphoserine, phosphothreonine, phosphotyrosine) were hydrolyzed in 6 M HCl and then dried in a SpeedVac. The sample was resuspended in pH 1.9 (2.5% (v/v) formic acid, 7.8% (v/v) acetic acid) buffer and spotted on a TLC plate, followed by electrophoresis for 24 min at 1500 V in the first dimension using a HTLE-7000 electrophoresis system (CBS Scientific). After the plate was dry, the plate was rotated 90 o counterclockwise and electrophoresed in the second dimension in pH 3.5 buffer (5% (v/v) acetic acid, 0.5% (v/v) pyridine) at 1300 V for 17 min. The negatively charged phosphoamino acids migrate toward the (ϩ)-terminal during electrophoresis in both dimensions. The amino acid standards were visualized by spraying the plate with a 0.5% solution of ninhydrin in acetone. Phosphorylated amino acids were detected by autoradiography.
In Vivo Labeling of T47D Cells and Extraction-T47D cells were plated at a density of 3.5 ϫ 10 6 cells in 75-cm 2 T-flasks in 5% charcoalstripped fetal calf serum in Dulbecco's modified Eagle's medium. Prior to labeling, the medium was replaced with serum-free phosphate-free Dulbecco's modified Eagle's medium for 1.5 h at 37°C. [ 32 P]H 3 PO 4 (0.833 mCi/ml) was added to the phosphate-free medium, and the cells were incubated overnight at 37°C. 2 h prior to harvest, the cells were treated with hormone at a final concentration of 10 nM R5020. The medium was aspirated, and the cells were washed once with calcium and magnesium-free Hanks' balanced salt solution. TEN buffer (40 mM Tris-HCl, pH 8.0, 1 mM EDTA, 150 mM NaCl) was added to the flasks and allowed to incubate for 5 min. Cells were dislodged by shaking, and the suspension was pooled and transferred to 50-ml conical tubes. The cells were pelleted by centrifugation at 3000 rpm. Cell pellets were washed with Hanks' balanced salt solution and then KPFM (50 mM potassium phosphate, pH 7.4, 50 mM sodium fluoride, 1 mM EDTA, 1 mM EGTA, and 12 mM monothioglycerol), with centrifugation between washes. To make whole-cell extract, the cells were suspended in KPFM containing 0.5 M NaCl plus a mixture of protease and phosphatase inhibitors (18) and homogenized by 50 strokes in a glass-Teflon homogenizer. The homogenates were diluted 1:1 with KPFM buffer to reduce the salt concentration, transferred to Ti75 tubes, and centrifuged at 40,000 rpm for 30 min. The extract supernatant was directly added to the prepared antibody-coupled Sepharose beads.
Immunoprecipitation and Gel Separation of PR-Protein A-Sepharose was preswollen in water and washed with PBS. Per six-flask pool, 200 l of bead suspension in PBS (3:1 beads/buffer) were incubated with 25 g of rabbit anti-mouse for 4 h at 4°C on a rocking platform. The beads were then washed twice with PBS. 40 g (per six-flask pool) of the PR antibody 1294 (45) was added to the bead suspension and incubated overnight at 4°C on a rocking platform. The antibody-conjugated beads were washed once with PBS and twice with TEG buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 10% glycerol). The cell extract was then added to the prepared beads and incubated overnight at 4°C on a rocking platform. The beads were washed once with TEG, three times with TEG containing 0.3 M NaCl, and again with TEG by suspending in the indicated buffer followed by centrifugation at 3000 rpm to remove nonspecific proteins. PR was extracted from the beads by heating at 100°C in 1ϫ SDS-sample buffer and loaded on a 6.5% SDS-polyacrylamide gel. Labeled PR was detected by autoradiography.
Protease Digestion of His Tag PR for Mass Spectrometric Analysis-Purified His tag PR was dialyzed for 8 h against 50 mM ammonium bicarbonate in a Tube-O-dialyzer (Research Products International Corp., Mount Prospect, IL). PR (40 g) was digested with 1 g each of trypsin, Asp-N, and Glu-C for 4 h at 37°C. An additional 1 g of each enzyme was added after 4 h, and incubation continued for an additional 4 h. The digest was then dried in a SpeedVac.
FeNTA Isolation of Phosphopeptides-FeNTA columns were prepared and used essentially as described (46)  Mass Spectrometry-Analyses were performed on a PE Sciex (Foster City, CA) API 3000 tandem quadrupole mass spectrometer equipped with a Protana (Odense, Denmark) nanoelectrospray source. The samples were dissolved in an aqueous solution of 50% methanol and 1% formic acid, and 2-3 l were deposited in the gold/palladium-coated glass nanoelectrospray capillaries. The samples were analyzed for phosphorylated peptide identification using negative ion precursor ion scanning for m/z 79.1 with nitrogen as the collision gas and a collision energy of 100 eV. Positive ion full scanning spectra were then recorded utilizing the first quadrupole as the mass filter and a cone potential of 70 V. Product ion spectra of the appropriate positively charged precursor ions were then recorded to identify the site of phosphorylation with nitrogen as the collision gas and collision energies of 20 -40 eV.

Identification of Sites Phosphorylated by Cdk2
HPLC Analysis-Baculovirus-expressed His tag human PR (A or B) (1 g) was phosphorylated by cyclin A-Cdk2 as described under "Experimental Procedures" and isolated by SDS-PAGE, and the PR was in-gel digested using modified trypsin. The tryptic digest was loaded on a C 18 reversed-phase HPLC column that yielded the phosphopeptide patterns seen in the two lower panels in Fig. 1. Comparing the phosphopeptide profiles of PR-B and PR-A obtained with the new digestion procedure (panels III and IV) to the previous Cdk2 profile obtained with the original procedure (panel II), we observed several novel late eluting phosphopeptide peaks (Fig. 1, peaks A-D) as well as the previously identified peaks 1 (Ser 190 ), 2 (Ser 400 ), and 4/6 (Ser 162 ). These new peaks (A-D) are specific to Cdk2, since inclusion of the Cdk2 inhibitor roscovitine in the phosphorylation reaction blocks their induction (data not shown). One of these peaks, peak C, is clearly found only in PR-B (arrow). Comparing the new Cdk2 phosphopeptides, peaks A-D (shown in panel III), with the previously published profile of phosphopeptides from in vivo labeled PR-B treated with R5020 (panel I), we observe that the Cdk2 sites represent only a subset of the total sites found in vivo. Several of the major phosphopeptide peaks from the in vivo profile have not yet been identified. Interestingly, although the elution times vary slightly due to a change in the HPLC apparatus used, peaks with elution positions comparable with peaks A-D are detected in the in vivo profile, suggesting that they are previously unidentified authentic sites. As demonstrated below, even the modified cleavage procedure is insufficient to produce complete cleavage of resistant sites; thus, the magnitude of the peaks cannot be used as an indication of the extent of phosphorylation of these sites.
Phosphopeptide Analysis-To identify these new phosphopeptides, we used a combination of manual Edman degradation, secondary endoproteinase digestion, and phosphoamino acid analysis. Table I displays the data from the manual Edman degradation of the phosphopeptides in peaks A-D. Peak A and B both release in cycle 2, with peak B having an additional smaller release in cycle 8. Peak C has a weak release in cycle 15, and peak D releases in cycle 4. Weaker releases are expected in late cycles due to gradual loss of yield due to incomplete coupling and cleavage as well as the low efficiency of manual Edman degradation when the reactions are not performed under an inert atmosphere (40).
To further characterize these phosphopeptides, secondary digests were performed with the endoproteinases Asp-N and Glu-C followed by separation on a 40% acrylamide peptide gel. Glu-C digests were also split for manual Edman degradation and peptide gel analysis. Table II summarizes the manual Edman degradation data and secondary digest analysis.
Peptide gel analysis of peak A indicated the presence of a phosphopeptide that did not cleave with Asp-N but cleaved well with Glu-C. There was no change in the cycle of release after Glu-C digestion (cycle 2). Since the possible candidates did not include a Ser-Pro motif (a Ser/Thr-Pro sequence is required for phosphorylation by Cdk2), phosphoamino acid analysis was performed. Threonine was the predominant residue phospho-rylated in this peptide pool ( Fig. 2A). Of the three threonines in PR predicted to be in position 2 of a tryptic peptide, only the Thr 430 -containing peptide is situated in a Thr-Pro motif (Fig.  2B). However, the Thr 430 tryptic peptide does not fulfill the secondary digest characteristics. Additionally, it is a very small peptide and, if bound, should elute early from the HPLC column. Closer inspection revealed that incomplete tryptic digestion of the Thr 430 peptide would result in a larger phosphopeptide with the correct digestion patterns. The peptide containing Thr 430 ends with Arg followed by Pro, a sequence very resistant to cleavage by trypsin (40). To confirm that this peptide is a larger incomplete tryptic phosphopeptide containing Thr 430 , additional digests with Arg-C and Lys-C were performed. These endoproteinases cleave Arg-Pro and Lys-Pro sequences more effectively than trypsin. Digestion with Arg-C generates a very small fragment, consistent with the Thr 430 peptide (Fig.  2C). As expected, Lys-C does not cleave the peptide (Fig. 2C). The size of the incomplete tryptic phosphopeptide containing Thr 430 is consistent with its elution time on the HPLC profile. All of these criteria unambiguously identify Thr 430 as one of the Cdk2 phosphorylation sites.
Peak B contains two phosphopeptides as detected by peptide gel analysis. Phosphoamino analysis indicates that serine is the predominant modified residue (data not shown). By collecting individual fractions under the peak, the two phosphopeptides could be partially separated. In this case, the fraction containing the phosphopeptide with the cycle 2 release (B2) cleaved with both Asp-N and Glu-C. Of the three peptide candidates with a cycle 2 release that also contain both Asp and Glu residues, only two peptides in PR contain a serine in position 2 followed by a proline (Fig. 3A). Furthermore, since the first candidate peptide (containing Ser 108 ) is from the region unique to PR-B, the candidate can be eliminated because the phosphopeptide was observed in both PR-A and PR-B profiles. Additionally, the Ser 345 -containing peptide is not a likely candidate, since we have previously shown that this phosphopeptide elutes much earlier than Peak B (see Fig. 1I). Examination of the remaining candidate peptide containing Ser 213 reveals that digestion with Glu-C would greatly change its mobility, whereas the Ser 345 peptide would be resistant to cleavage due to the location of the glutamic acid. Glu-Pro motifs are inefficiently cleaved, especially at the carboxyl terminus of a peptide (40). Glu-C digestion of this peptide results in a faster migrating phosphopeptide (data not shown); thus, the phosphorylated peptide can only be Ser 213. In the case of the phosphopeptide with a cycle 8 release (B8), digestion of the entire peak B region with Glu-C prior to manual Edman degradation converts the cycle 8 to a cycle 4 release, while the cycle 2 release remained unchanged as expected (Fig. 3C). These data conclusively identify the peak B8 phosphopeptide from the candidates (Fig. 3B). Only the peptide containing Ser 554 would cleave to produce a peptide with a cycle 4 release. Therefore, the two phosphopeptides in peak B induced by Cdk2 in vitro correspond to the peptides containing Ser 213 and Ser 554 .
Peak C, observed solely in the PR-B profile, exhibits a weak release on cycle 15 by manual Edman degradation. Only two of the three candidate peptides with cycle 15 releases are located in the B-specific region of PR (Fig. 4A). Of these, only Ser 25 is contained within a Ser/Thr-Pro motif. Examination of the sequence of the tryptic peptide containing Ser 25 shows that redigestion of the tryptic peptide with Glu-C should convert the cycle 15 release to a cycle 3 release. Indeed, we observe a cycle 3 release after Glu-C digestion (Fig. 4B), whereas the other B-specific peptide candidate (which does not contain the required Ser-Pro motif) would produce a release in cycle 4. These results identify the phosphorylated site in peak C as Ser 25 .
Peak D elutes very late in the profile, suggesting a large and hydrophobic peptide. Manual Edman degradation data show that the phosphoamino acid is in position 4. Of the eight candidates with Ser in position 4, only one contains a Ser-Pro motif. Secondary digestion with either Asp-N or Glu-C causes a significant change in mobility of the peptide observed by peptide gel analysis (Fig. 5B). Of the two peptides with a serine in position 4 that contain both Asp and Glu residues (Fig. 5A), only the Ser 676 site is consistent with these results. First, Ser 676 is the only site located within a Ser-Pro motif, and second, although the peptide containing Ser 711 contains a glu- tamic acid, it is adjacent to the C-terminal arginine, which is known to be resistant to cleavage (40). Finally, the size and hydrophobicity of the Ser 676 tryptic peptide is consistent with the late elution of this peptide from the HPLC. By these criteria, we have conclusively identified Ser 676 as the Cdk2-induced peptide in peak D. Interestingly, this site is analogous to the "hinge" site identified in other steroid receptors.

Identification of Additional in Vivo Phosphorylation Sites in PR Expressed in Sf9 Cells
Most of the phosphorylation sites identified in steroid receptors are in Ser/Thr-Pro motifs. Human PR contains 15 of these motifs, of which six have been shown previously to be phosphorylated (15)(16)(17). An analysis of the remaining sites reveals that conventional tryptic digests would produce peptides ranging from about 30 to 70 amino acids in length, too long for optimal binding and elution from a C 18 column. Our previous radiolabeling studies had indicated that PR expressed in Sf9 insect cells exhibited the same phosphopeptide map as PR isolated from T47D cells (36). We therefore took advantage of the selectivity of mass spectrometers to look for additional phosphorylation sites in baculovirus-expressed PR isolated from Sf9 cells. Initial analysis of a tryptic digest of purified recombinant PR-B in precursor ion mode revealed a number of potential phosphorylated candidates, but the spectrum in positive ion mode was too complex to successfully isolate phosphopeptides for sequence analysis. Fractionation of the digest into 15 fractions by C 18 4Ϫ . The mass deconvoluted spectrum displayed a peptide of mass 5481.30, corresponding to amino acids 107-159, which contained a single phosphorylation. This peptide contains multiple serines including Ser 130 , which resides in a Ser-Pro motif (data not shown). Since none of the serines in this region of PR have been previously identified as phosphorylation sites, this must represent a novel site. Phosphopeptides with masses consistent with peptides containing previously identified sites, Ser 162 and Ser 345 , were also detected.
In subsequent analyses, ferric ion affinity resin was used to preferentially absorb the phosphopeptides from digests. Purified recombinant PR-B was digested with trypsin, Asp-N, and Glu-C, and the phosphopeptides were purified on the ferric ion affinity resin as described under "Experimental Procedures." Fig. 6A shows the full scan of the m/z 400 -1100 range for the ion affinity column eluate, and Fig. 6B shows the corresponding precursor m/z 79 scan. Collisionally activated decomposition products of the doubly charged peak at m/z 598 yielded a sequence for amino acids 11-22 produced by cleavage with trypsin and Glu-C containing a phosphoserine at amino acid 20, the only serine in the peptide (Fig. 7A and Table III). Analysis of the products of the phosphopeptide at m/z 863

Summary of secondary digests and manual Edman degradation
Peptide gel analysis of peak B indicated the presence of more than one phosphopeptide. Peak B peptides could be partially separated by collecting individual fractions. To distinguish the peptides in peak B, the number represents their initial manual Edman degradation (MED) release. ND, not determined.

FIG. 2. Identification of Thr 430 in peak A.
A, His tag PR-B phosphorylated by Cdk2 in vitro was digested with trypsin, and the peptides were separated by reversed-phase HPLC. Fractions corresponding to peak A were collected and analyzed by phosphoamino acid analysis. Threonine is the major phosphoamino acid in these fractions. P-Ser, P-Thr, and P-Tyr, phosphoserine, phosphothreonine, and phosphotyrosine, respectively. B, of all of the complete tryptic fragments of PR, only three peptides contain a Thr residue in position 2. Ser/Thr-Pro motifs are required by Cdk2 for phosphorylation. C, HPLC fractions of peak A were split for secondary endoproteinase digestions with Asp-N, Glu-C, Arg-C, and Lys-C followed by alkaline peptide gel analysis. reveals that it corresponds to 672-679 produced by a double trypsin and Asp-N cut, which contains phosphoserine at position 676 (Fig. 7B). Previously identified phosphopeptides detected in this analysis included the doubly charged peptide at m/z 492 identified as the phosphotryptic peptide containing Ser 162 , the doubly charged peptide at m/z 571 identified as the phosphotryptic peptide containing Ser 102 , and the phosphopep-tide at m/z 680 as the phosphopeptide containing Ser 190 (data not shown).

Analysis of T47D Receptor for Newly Identified Sites
To determine whether any of the newly detected sites phosphorylated by Cdk2 in vitro are in vivo phosphorylation sites, we performed in vivo labeling of endogenous PR from T47D cells in the absence and presence of hormone. PR was purified by immunoprecipitation, separated by SDS-PAGE, and processed as previously indicated for the in vitro samples. HPLC peaks with identical retention times from in vivo and in vitro samples were analyzed simultaneously by alkaline peptide gel electrophoresis. Most of the analyses were inconclusive, presumably due to the poor efficiency of cleavage and extraction from gel slices and the background from additional peptides. However, comparison of the HPLC fractions from in vivo labeled PR with the same retention time as the fractions containing Ser 213 phosphorylated by Cdk2 reveal that they contain a phosphopeptide with the same mobility on alkaline gels as the phospho-Ser 213 peptide (Fig. 8). To determine whether phosphorylation of this site is strictly hormone-dependent or whether the site is basally phosphorylated, its phosphorylation was compared in the presence and absence of hormone with that of the Ser 162 phosphopeptide, which has previously been shown to be a basal phosphorylation site and whose phosphorylation is increased upon treatment with hormone. Fig. 8 shows that the relative change in intensity of Ser 213 phosphorylation as a result of hormone treatment is similar to that of Ser 162 . These data suggest that Ser 213 is phosphorylated in vivo in T47D cells and has the characteristics of a basal phosphorylation site.

DISCUSSION
The presence of unidentified peptides in our phosphopeptide maps (17) and the failure to identify Ser 676 (the conserved Ser-Pro that is homologous to the Ser 530 hinge site in chicken PR (37)) as a phosphorylation site led us to develop new means to identify phosphorylation sites in PR. An analysis of PR sequence revealed that the Ser-Pro sequences that had not been identified as phosphorylation sites are likely to reside in large peptides produced by incomplete digestion at resistant trypsin cleavage sites. Using trypsin modified to prevent au- FIG. 3. Identification of Ser 213 and Ser 554 as the phosphorylation sites in peak B. Partial separation of the two peptides in peak B followed by manual Edman degradation and secondary endoproteinase digestions indicate that the peptide with a cycle 2 release contains both Asp and Glu residues. A, the list of candidate peptides for peak B2 is shown. Glu residues are underlined. B, comparison of the sequences of the candidate peptides is shown for the second peptide in peak B with a cycle 8 release. C, the fraction pool of peak B (containing both phosphopeptides) was digested with Glu-C followed by manual Edman degradation. The graph of the release data is presented. The sequence of the peptide containing Ser 554 is shown prior to and following Glu-C digestion to illustrate how the cycle 8 release is converted to a cycle 4 release. The cycle 2 release of the Ser 213 -containing peptide remains unchanged. Comparison of the sequences of the candidate peptides is shown. Glu residues are underlined. B, HPLC fractions corresponding to peak C were pooled and digested with Glu-C followed by manual Edman degradation. On the graph of the release data, the sequence of the peptide containing Ser 25 is shown prior to and following Glu-C digestion to illustrate how the cycle 15 release was converted to a cycle 3 release. tolysis as well as other proteases, we have identified Ser 676 as a phosphorylation site in PR by mass spectrometry and have shown that it is a substrate for Cdk2. Using these techniques, we have also identified five other candidate phosphorylation sites. We previously identified three phosphorylation sites, Ser 162 , Ser 190 , and Ser 400 , which were phosphorylated by Cdk2 in vitro (17). Using the specially modified trypsin that cleaves Lys-Pro and Arg-Pro sequences at a low but detectable efficiency, we have detected four small, late eluting peaks in the HPLC profile and have identified five additional Cdk2 sites in these peaks. Our studies here underscore the difficulty in detecting phosphorylation sites in large peptides produced by incomplete digestion. Comparison of the trypsin we used in our previous studies with the new modified trypsin showed that there was a significant difference in the phosphopeptide peaks detected by HPLC. Although the new peaks appear to be minor compared with those previously identified, Arg/Lys-Pro motifs will still be poorly cleaved relative to other trypsin cleavage sites (40). Evidence that we are only achieving partial cleavage is shown in Fig. 2. Redigestion of the purified peptide with excess Arg-C cleaves the resistant bond. Therefore, peak size cannot be considered representative of the extent of phosphorylation for peptides bounded by these difficult cleavage sites. This problem probably contributes to our failure to unambiguously identify most of these new peptides in digests of T47D receptor. The increased difficulty in releasing large peptides from gel slices coupled with a higher background as a well as phosphorylation by other kinases in vivo greatly increases the complexity of the analyses. For example, although some of the new peptides elute in positions similar to peaks 10 and 11, our previous characterization of peptides derived from PR from T47D cells suggests that there is at least one additional phosphopeptide in this region that has not been identified.
Employing the techniques of mass spectrometry to analyze peptides derived from PR expressed in Sf9 cells and in vivo labeling of T47D breast cancer cells, we have identified two new phosphorylation sites (Ser 20 and Ser 676 ) and two novel phosphopeptides (a peptide containing Ser 130 and a peptide that comigrates with the Cdk2-phosphorylated Ser 213 -containing peptide). The mass spectrometric analyses have allowed us to identify additional phosphorylation sites in PR as well as to confirm previously identified sites. However, mass spectrometric methods also have limitations, requiring purified protein in larger quantities than the analyses following peptides only by radiolabel. For our studies, we used His tag PR amplified by baculovirus infection of Sf9 cells and purified by affinity column to achieve the amount and quality needed for analysis. Although we could detect phosphopeptides in a negative ion precursor ion scan, we found that some form of fractionation of peptides was required to enrich for phosphopeptides for sequencing by tandem mass spectrometry. FeNTA chromatography allowed us to preferentially obtain the phosphopeptides. Although this approach was useful in obtaining phosphopeptides of moderate length, we did not detect the very large phosphopeptides arising from incomplete digestion with trypsin in the FeNTA column eluate.
Previous studies from our laboratory have shown that in vivo phosphorylation patterns of PR from hormone-treated T47D and baculovirus-infected Sf9 cells are indistinguishable, and no aberrant phosphorylation was observed in Sf9 cells (36). Therefore, we expect that the new phosphorylation sites identified by mass spectrometry are also present in receptor expressed in T47D cells. The poor yield of these peptides suggests that it will be difficult to detect the sites in T47D PR either by radiolabeling or by mass spectrometry of purified protein. Phosphorylation site-specific antibodies have been produced to the Ser 190 and Ser 294 sites, and we have utilized them to study hormonedependent phosphorylation of PR in T47D cells (45). Therefore, in the case of many of these new phosphorylation sites, generation of phosphorylation site-specific antibodies will be useful in confirming their phosphorylation in vivo and studying their regulation by hormone and cell signaling pathways.
Including the three sites previously identified, our laboratory has identified a total of eight sites phosphorylated by Cdk2 in vitro; five of these have been confirmed in vivo (162, 190, 213, 400, and 676). Notably, only subsets of the 15 Ser-Pro motifs in PR are substrates for Cdk2 phosphorylation, illustrating the site-specific discrimination by Cdk2. In fact, we observed that Cdk2 phosphorylates Ser 25 , but not Ser 20 , despite the fact that both phosphorylation sites are located within the same tryptic peptide (Table I and Fig. 4). Seven of the sites phosphorylated by Cdk2 in vitro, Ser 25 , Ser 162 , Ser 190 , Ser 213 , Ser 400 , Thr 430 , and Ser 554 , are situated within the phosphorylation-rich A/B domain (Fig. 9). Interestingly, both Thr 430 and Ser 554 border the ligand-independent activation domain AF-1. Perhaps most significantly, Ser 554 is located within only 13 amino acids of the DNA binding domain. The proximity of these sites to the functional domains suggests their potential ability to regulate or be regulated by activation or DNA binding.
Only one Cdk2 site, Ser 676 , was identified outside of the A/B domain, located in the hinge region between the DNA and ligand binding domains (Fig. 9). Ser 676 is analogous to the in vivo phosphorylation sites Ser 530 identified in chicken PR (37) and Ser 650 in human androgen receptor (26). Ser 676 was not detected as a phosphorylation site in a CNBr digest of radiolabeled PR (13). However, this peptide would have been rather small (about 6500 kDa) for detection by SDS-polyacrylamide gel electrophoresis followed by nitrocellulose transfer and would have contained only one site, whereas the other peptides detected contain multiple (up to five) sites. In androgen receptor and chicken PR, mutation of the serine to alanine resulted in decreased transactivation (26); however, for chicken PR, the effect was only observed at subsaturating hormone conditions (47). Interestingly, this site in human PR has been mutated, and the effect of an alanine substitution was examined. De-  8. Identification of the Ser 213 -containing peptide from in vivo labeled T47D cells. PR from in vivo labeled T47D cells (treated with R5020 or left untreated) was immunopurified and subjected to trypsin digestion. Tryptic peptides were separated by reversed-phase HPLC. Fractions corresponding to peak 6 (Ser 162 ) and the peak B region were analyzed on an alkaline peptide gel and compared with the Ser 213 phosphopeptide generated by in vitro phosphorylation with Cdk2. An arrow indicates the Ser 213 -containing peptide.
FIG. 9. Schematic diagram of the phosphorylation sites in human PR (hPR). Shown here is an updated diagram of the phosphorylation sites in PR, including the new sites identified by in vitro phosphorylation by Cdk2, mass spectrometry of baculovirus-expressed PR from Sf9 cells, or both. All of the new sites, with the exception of the newly described "hinge" site, Ser 676 , are located in the NH 2 -terminal domain. Interestingly, one of the other new sites, Ser 554 , is in very close proximity to both the AF-1 and DNA binding domain (DBD). Since Ser 130 is the only serine located within a Ser-Pro motif in the large peptide (residues 107-159) in which the actual phosphorylated residue could not be determined, the site is indicated in parentheses. creases in activity range from 20 to 50% depending upon the cell and promoter context, suggesting that this is an important site in human PR (21). We have confirmed by mass spectrometry that Ser 676 is an in vivo phosphorylation site. Due to its proximity to the LBD, it is tempting to speculate that agonistand antagonist-induced conformational changes could dramatically alter the phosphorylation of Ser 676 . The role of Cdk2 phosphorylation in receptor function has been examined for several steroid receptors. In the case of the human estrogen receptor, overexpression of cyclin A, a regulatory partner of Cdk2, increases estrogen receptor-mediated transactivation (28). Furthermore, two serine residues, Ser 104 and Ser 106 , have been shown by in vitro phosphorylation and site-directed mutagenesis studies to be involved in this pathway (30). Rat glucocorticoid receptor has also been shown to be a substrate for Cdk2. In vitro, Cdk2 complexes modify two serines, Ser 224 and Ser 232 (27). Analysis of glucocorticoid receptor transactivation in yeast reveals that hormone-induced transactivation is greatly diminished by perturbation of the Cdk pathway (27). Since Cdk2 and its cyclin partners are key players in cell cycle progression, it is interesting to note that the phosphorylation of mouse glucocorticoid receptor is cell cycle-dependent (48). All of this evidence points to a role for Cdk2-cyclin complexes in regulating steroid receptor function by direct phosphorylation. In the case of PR, the large number of Cdk2 phosphorylation sites strongly suggests regulation of PR via the Cdk pathway.
In summary, six new phosphorylation sites have been identified in PR (Fig. 9). Five of these are phosphorylated in vitro by Cdk2. One Cdk2-induced site, Ser 676 , as well as two B-specific phosphorylation sites, Ser 20 and a peptide containing the putative site Ser 130 , were identified as in vivo phosphorylation sites by mass spectrometric methods. The phosphopeptide containing the newly identified Cdk2 site Ser 213 exhibited characteristics of a basal phosphorylation site in vivo. Identification of these additional phosphorylation sites in PR argues that we have only begun to understand the complex nature of PR phosphorylation.