INTRODUCTION

Protein-tyrosine phosphorylation and dephosphorylation play an important role in regulating cellular differentiation, proliferation, and activation. The role of dephosphorylation by CD45 protein-tyrosine phosphatase (PTPase)¹ in lymphocyte signaling has been the subject of intense investigation (1-4). CD45 (T200, B220, L-CA) is a transmembrane PTPase of hematopoietic cells of 1268 total amino acids, with a cytoplasmic domain of 702 amino acids containing tandem repeated PTPase homology domains (designated D1 and D2). The membrane-proximal PTPase domain (D1) is considered to be constitutively active, and the second PTPase domain (D2) is usually considered to be inactive. CD45 has been intensely studied because of its well documented role in the antigen-specific activation of B and T cells (1-3). T cells lacking CD45 fail to respond to stimulation of the T cell antigen receptor (5, 6), and the catalytic activity of the CD45 D1 domain is required for T cell receptor activation (7). Chimeric proteins, in which the extracellular domain of CD45 was replaced, restored normal T cell receptor activation (8-10). Similarly, CD45-deficient B cells do not respond to stimulation of the IgM receptor (11). It is believed that CD45 activates the Src family PTKs by dephosphorylating the regulatory Tyr(P) near the C terminus of T cell receptor or B cell receptor-associated Src family kinases (12-15). Since CD45 is abundantly expressed on all nonerythroid hematopoietic cells, it has also been hypothesized that CD45 may be involved in the regulation of other fundamental cell processes such as cell growth and cell cycle (1, 16, 17).

Phosphorylation of the cytoplasmic domain of CD45 has been described, and this phosphorylation has been proposed to play a role in the regulation of biological function by providing docking sites (18) or by altering PTPase activity (18-20). The modulation of CD45 PTPase activity has been found to correlate with phosphorylation of the cytoplasmic domain in several studies. 1) Treatment of T cell clones with a Ca²⁺ ionophore decreased the Ser phosphorylation of CD45 and simultaneously decreased the PTPase activity of the molecule (20). 2) Phosphorylation of human CD45 (by overexpression of CD45 in Cos cells with p50^csk PTK) increased the PTPase activity of CD45 and increased the association of CD45 with a putative substrate, p56^lck (18). The p50^csk PTK phosphorylation site was identified as Tyr¹¹⁹³. 3) The sequential in vitro phosphorylation of human CD45 by Abl PTK and casein kinase II increased the activity of CD45 for certain substrates (19). 4) The increase in CD45 phosphorylation due to phorbol ester treatment of cells was associated with a decrease in PTPase activity (21). Other reports indicate that the phosphorylation of CD45 increased after stimulation by phorbol esters (22) or interleukin-2 (23) without apparent change in CD45 activity.

The elucidation of the role of CD45 phosphorylation in PTPase activity and in signaling has been hampered by the lack of precise knowledge of naturally occurring CD45 phosphorylation sites. The large number of potential phosphorylation sites in the CD45 cytoplasmic domain make identification by sequencing or mutagenesis extremely difficult. To resolve this problem, we have identified the in vivo phosphorylation sites of CD45 in 70Z/3.12 cells (a mouse pre-B cell line) by two-dimensional phosphopeptide mapping, reverse-phase HPLC, phosphoamino acid analysis, enzymatic degradation, and matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS).

EXPERIMENTAL PROCEDURES

The mouse pre-B lymphocyte cell line 70Z/3.12 was obtained from the American Type Culture Collection (ATCC) and grown at 37 °C in 90% RPMI 1640 medium (Life Technologies, Inc.) containing 10% heat inactivated fetal bovine serum (FBS) (Life Technologies), 25 mM HEPES, 50 µM 2-mercaptoethanol, 100 units/ml penicillin, 100 units/ml streptomycin (complete RPMI 1640 medium) in 5% CO₂. Cells were counted using a Coulter Cell Counter (Coulter Electronics, Hialeah, FL) and maintained in an exponential growth state (0.1-4.0 × 10⁵ cells/ml). Cultures were harvested at 8 × 10⁵ cells/ml with viability in excess of 97%.

70Z/3.12 cells (1 × 10⁷) were washed three times with phosphate-free RPMI 1640 medium (supplemented to 10% with dialyzed FBS; Life Technologies), then resuspended in 1 ml of phosphate-free RPMI 1640 medium, followed by incubation for 1 h at 37 °C. After centrifugation, cells were resuspended in 1 ml of fresh, phosphate-free RPMI 1640 medium. One mCi of [³²P]orthophosphate (DuPont NEN Research, Boston, MA) was added and cells were incubated for 4 h at 37 °C.

70Z/3.12 cells were pelleted, washed twice with ice-cold phosphate-buffered saline (8 mM Na₂HPO₄, 1 mM KH₂PO₄, 137 mM NaCl, 3 mM KCl, pH 7.4), and then lysed (2 × 10⁷ cells/ml lysis buffer) in Nonidet P-40 lysis buffer containing 20 mM Tris, pH 8.0, 137 mM NaCl, 1.0% (v/v) Nonidet P-40 (Pierce), 2 mM phenylmethylsulfonyl fluoride, 0.3 µM aprotinin, 1 µM leupeptin, 5 mM EDTA, 1 µM pepstatin A, 50 µg/ml DNase I and RNase A, 1 µM okadaic acid (Life Technologies, Inc.), and 6 mM NaF for 30 min in ice. Lysates were cleared by centrifugation at 12,000 rpm at 4 °C for 20 min and precleared with 25 µl of packed Gammabind Plus Sepharose (Pharmacia Biotech Inc.). Anti-CD45 M1/9.3.4 monoclonal antibody (2 µg, ATCC) was conjugated to 25 µl of Gammabind Plus Sepharose for 1 h at 4 °C in phosphate-buffered saline (pH 7.4), followed by incubation with 0.5 ml of cell lysate (from 10⁷ cells) for 1 h at 4 °C under constant shaking. For SDS-PAGE, immunoprecipitates were washed once with Nonidet P-40 lysis buffer, twice with phosphate-buffered saline (pH 7.4, containing 1 µM okadaic acid, 6 mM NaF), once with 0.5 M LiCl (pH 7.4, containing 1 µM okadaic acid, 6 mM NaF), and twice with 50 mM Tris buffer (pH 8.0, containing 1 µM okadaic acid, 6 mM NaF). Immunoprecipitates were suspended in SDS-PAGE sample buffer and boiled for 3 min.

CD45 immunoprecipitate samples were subjected to electrophoresis on 4-15% SDS-PAGE gels for 1.5 h at 120 V. Silver staining was performed to determine the amount of CD45 proteins. CD45 bands were quantitated by optical scanning using an Ambis scanner (Ambis, Inc., San Diego, CA). For ³²P-labeled samples, CD45 proteins were transferred to polyvinylidene difluoride (PVDF) membrane (0.2 µm) (Bio-Rad) by electroblotting in 25 mM Tris, 192 mM glycine, 20% (v/v) methanol, pH 8.3, for 4 h at 400 mA. CD45 bands were visualized by autoradiography.

³²P-Labeled CD45 bands from the PVDF membrane were excised, washed three times with methanol, blocked in 1 ml of 0.5% PVP-360 (Sigma) in 100 mM acetic acid for 30 min at 37 °C, and washed five times with water. Tryptic digestion was performed with 10 µg of trypsin (sequencing grade) (Promega, Madison, WI) for 2 h in 200 µl of 50 mM NH₄HCO₃, pH 8.3, at 37 °C, and an additional 10 µg of trypsin was added and incubated for another 2 h at 37 °C. Tryptic peptides were recovered in the supernatant and concentrated in a SpeedVac (Savant Instruments, Inc., Farmingdale, NY). The samples were separated by thin layer electrophoresis (TLE) on 20 × 20-cm Kodak cellulose thin layer chromatography (TLC) plates (Eastman Kodak Co.) in the first dimension in pH 1.9 buffer (formic acid (88% w/v)/glacial acetic acid/water, 25:78:897, v/v/v) at 0 °C at 1000 V for 30 min, dried in air, and separated in the second dimension in chromatography buffer (n-butanol/pyridine/glacial acetic acid/water, 15:10:3:12, v/v/v/v) (24, 25). The phosphopeptides were visualized by using a Betascope 603 blot analyzer (Betagen, Waltham, MA).

³²P-Labeled tryptic phosphopeptides eluted from TLC plates or collected from reverse-phase HPLC were hydrolyzed in 100 µl of 5.7 N HCl for 1 h at 100 °C. The recovered phosphoamino acids were dried in a SpeedVac. Unlabeled phosphoamino acids were added to each sample as standard markers. The samples were analyzed by one-dimensional electrophoresis in pH 2.5 buffer (66.7% pH 3.5 buffer (glacial acetic acid/pyridine/water, 50:5:945, v/v/v, containing 0.5 mM EDTA) and 33.3% pH 1.9 buffer (glacial acetic acid/88% formic acid/water, 78:25:897, v/v/v)) on 20 × 20-cm cellulose TLC plates (Kodak) for 1 h at 0 °C and 500 V. The markers were visualized by ninhydrin and the ³²P-labeled phosphoamino acids were visualized by Betascope (Betagen) analysis.

70Z/3.12 cells (5 × 10⁹ at 8 × 10⁵ cells/ml) were collected, lysed, and immunoprecipitated with anti-CD45 M1/9.3.4 monoclonal antibody to yield approximately 40 µg of CD45 protein. This large scale, unlabeled CD45 immunoprecipitate was then mixed with a smaller amount of a radioactive marker consisting of ³²P-labeled CD45 immunoprecipitate from cells grown at the same density as described above. After SDS-PAGE, Western transfer, and tryptic digestion, the digested mixture was fractionated using a microbore reverse-phase HPLC system (Michrom BioResources, Inc.) with a 0-86% gradient of acetonitrile in 0.1% aqueous trifluoroacetic acid and a flow rate of 50 µl/min on a Reliasil C18 column (5 mm, 300 A, 1.0 × 150 mm, Michrom BioResources). The solvent gradient started at 0% solvent B, then increased to 20% solvent B at 15 min, 60% solvent B at 20 min, and 95% solvent B at 21 min, where solvent A was 0.1% trifluoroacetic acid in water and solvent B was a mixture of acetonitrile/0.1% trifluoroacetic acid in water (9:1). The eluant was monitored by a UV absorbance at 214 nm, and fractions were collected every 30 s. A 1-µl sample of each HPLC fraction was spotted on a TLC plate, followed by detection of ³²P radioactivity by Betascope (Betagen) analysis.

All MALDI-MS spectra were obtained on a Voyager Elite time-of-flight mass spectrometer (PerSeptive Biosystems, Framingham, MA), used in linear mode, equipped with a nitrogen laser (337 nm, 3-ns pulse). The accelerating voltage in the ion source was 26 kV. Data were acquired with a transient recorder with 2-ns resolution. The matrix used in this work was -cyano-4-hydroxycinnamic acid, dissolved in water/acetonitrile (1:1, v/v) to give a saturated solution at room temperature. To prepare the sample for analysis, 1 µl of the peptide solution (1-10 pmol/µl 0.1% trifluoroacetic acid) was added to 1 µl of the matrix solution and applied to a stainless steel sample plate. The mixture was then allowed to air dry on the sample plate before being introduced into the mass spectrometer. Each spectrum was produced by accumulating data from 50-256 laser pulses. Time-to-mass conversion was achieved by external or internal calibration using bradykinin (MH⁺ at m/z 1061.2) and insulin (MH⁺ at m/z 5734.6). The accuracy of mass assignments was approximately ± 0.1% (± 1 Da/1000 Da). A computer program, MSU MassMap (26) was used to calculate the average masses of all possible peptide and phosphopeptide fragments from CD45, and the m/z value of the mass spectral peak for the corresponding MH⁺ ion. Dephosphorylation of 5 µl of the HPLC fractions containing phosphopeptide was achieved by incubation of the peptide with 1~2 units of calf intestine alkaline phosphatase in 50 mM NH₄HCO₃ buffer (pH 8.0) at 37 °C for 4 h.

Digestion of TLC isolated peptide a with hydroxylamine was performed in 2 M guanidine·HCl, 2 M NH₂OH·HCl, 0.2 M K₂CO₃ (Sigma), pH 9, for 4 h at 45 °C (24). After NH₂OH digestion the peptide was purified by HPLC using conditions described above to remove salts (yielding a single radioactive peak at 21 min, retention time) and was analyzed by one-dimensional TLC using the conditions described above. The NH₂OH-digested peptide was then digested with 1 µg of endoproteinase Glu-C (Boehringer Mannheim) in 50 mM NH₄HCO₃, pH 7.8, 18 h at 25 °C. The NH₂OH- and endoproteinase Glu-C-digested peptide a was then analyzed by one-dimensional TLC as described above. Radioactivity was detected using a Molecular Dynamics PhosphorImager.

Purified rat brain PKC was provided by Dr. A. Nairn (Rockefeller University, New York, NY), and the purified, bacterially expressed cytoplasmic domain of murine CD45 was provided by Dr. P. Johnson (University of British Columbia, Canada) (27). The phosphorylation of 1 µg of cytoplasmic CD45 was carried out with 1 µg of PKC at 30 °C for 30 min in 20 mM Tris·HCl (pH 7.4), 5 mM MgCl₂, 1 mM dithiothreitol, 50 µg/ml phosphatidylserine, 100 nM PMA, 2 mM CaCl₂, and 0.01 mCi of [-³²P]ATP (3000 Ci/mmol, DuPont NEN). The reaction was terminated by the addition of SDS-PAGE sample buffer. The sample was then boiled and resolved on a 4-15% SDS-PAGE gradient gel and transferred to a PVDF membrane, followed by autoradiography. The CD45 band was excised, digested with trypsin, and analyzed as described above.

RESULTS

In an effort to map the in vivo phosphorylation sites of CD45, we first prepared a two-dimensional map of tryptic phosphopeptides from 70Z/3.12 cells cultured at high density (8 × 10⁵ cells/ml). Cells were labeled with [³²P]orthophosphate for 4 h and immunoprecipitated CD45 was separated by SDS-PAGE and transferred to PVDF for autoradiography (Fig. 1A). The CD45 band was excised, digested with trypsin, and subjected to two-dimensional phosphopeptide mapping (Fig. 1B). Tryptic phosphopeptides were designated a-h (Fig. 1B). Phosphoamino acid analysis of each of the radioactive peptides eluted from the TLC plate (Fig. 2) revealed that each contained only Ser phosphorylation, except for phosphopeptide a, which contained both Ser and Thr phosphorylation in about a 1:1 ratio (Fig. 2, lane 2). No tyrosine phosphorylation of CD45 was detected. Betascope and phosphorimage analysis was used in this work, because it was quantitative and it allowed the detection of very low levels of radioactivity.

Reverse-phase HPLC was performed to isolate larger quantities of the tryptic phosphopeptides for detailed analysis of the phosphorylation sites of CD45. Approximately 40 µg of CD45 (about 200 pmol) was purified from 5 × 10⁹ 70Z/3.12 cells (8 × 10⁵ cells/ml) by immunoprecipitation with M1/9.3.4 monoclonal antibody. A radioactive tracer was prepared by immunoprecipitating CD45 from 4 × 10⁷ 70Z/3.12 cells (8 × 10⁵ cells/ml) labeled with 4 mCi of ³²P_i for 4 h, and added to the nonradioactive preparation. After tryptic digestion, the CD45 peptide mixture was subjected to fractionation by HPLC. Fractions were collected every 30 s and numbered according to HPLC retention time (Fig. 3A). The radioactive fractions were identified by application of 1 µl of each fraction to a TLC plate, followed by Betascope detection (Fig. 3B). Fractions at the following times were found to be radioactive and are hereafter designated by their retention time: 4, 18, 18.5, 20.5, 21, 21.5, 22, and 23. 

Each radioactive HPLC fraction was aligned with each phosphopeptide from two-dimensional phosphopeptide map by subjecting each HPLC fraction to one-dimensional TLE and one-dimensional TLC (Fig. 4, A and B). The unique mobility in one dimension of each spot allowed us to correlate the HPLC fraction with the two-dimensional phosphopeptide map. TLE analysis indicated that each HPLC fraction contained primarily one radioactive peptide, except for fraction 4, which contained two peptides (designated peptide b/c on the two-dimensional phosphopeptide map). TLC analysis indicated that each HPLC fraction contained primarily one radioactive peptide, except for fraction 21.5 which was separated into two phosphopeptides on TLC (designated peptide e,f). The TLE and TLC results matched the HPLC results in which both fraction 4 and 21.5 eluted as doublets. The HPLC chromatographic pattern was consistent with TLE (each consecutive fraction (except part of 4 and 18.5) generally became less charged; Fig. 4A) and TLC (each consecutive fraction became more hydrophobic; Fig. 4B). The results from one dimensional analysis allowed the correlation of each radioactive HPLC peptide fraction with each two-dimensional phosphopeptide spot (peptide designation shown at the bottom of Fig. 4). The analysis of fractions 18 and 20.5 is not shown since these were found to be the same as peptides 18.5 and 21, respectively. Phosphoamino acid analysis of each radioactive HPLC fraction indicated the presence of Ser(P), except for fraction 18.5, which also contained Thr(P) (data not shown), thus confirming that this fraction was the two-dimensional phosphopeptide spot a.

MALDI-MS analysis was performed on each radioactive HPLC fraction to determine the identity of the CD45 phosphopeptides. The observed mass values obtained for each radioactive HPLC fraction were compared with a table of predicted masses of all possible tryptic peptides of CD45 cytoplasmic domain below 3000 Da (including partially digested, unphosphorylated, and phosphorylated peptides). Each sample was then treated with alkaline phosphatase and subjected to re-analysis by MALDI-MS to observe loss of phosphate (80 Da or multiples of 80 Da shifts in the spectra). The results of MALDI-MS analysis for each radioactive HPLC fraction are described below.

The MALDI-MS spectrum of HPLC fraction 18.5 contained a single peak at m/z 2960.2 (Fig. 5A). Comparison with a table of calculated mass values of all possible tryptic peptides of CD45 cytoplasmic domain (724 possible peptides) indicated a close match to the predicted peak at 2958.9, corresponding to the monophosphorylated peptide 1239-1268 (Tables I and II). Dephosphorylation of the sample followed by MALDI-MS clearly showed a loss of only one phosphate (80 Da) to m/z 2880.0. Because phosphoamino acid analysis showed peptide a was phosphorylated equally on Ser and Thr, it must therefore consist of a mixture of phosphopeptides with the same sequence: one with a single Ser(P) and one with a single Thr(P). The Ser(P) and Thr(P) sites were localized using the following strategy. Digestion of peptide a with NH₂OH at Asn-Gly resulted in a single ³²P-phosphopeptide, which eluted at 21 min from HPLC, compared with undigested peptide a, which eluted at 18.5 min (data not shown). This HPLC-purified peptide contained both Ser(P) and Thr(P) (data not shown) and had a slightly slower mobility by TLC than peptide a (Fig. 5B, lanes 2 and 3). (HPLC purification was required to remove salts before TLC or enzymatic digestion.). Thorough digestion of the HPLC purified radioactive peptide with endoproteinase Glu-C resulted in only one ³²P-phosphopeptide with altered TLC mobility (Fig. 5B, lane 4), which contained both Ser(P) and Thr(P) (data not shown). Thus we conclude that the phosphorylation sites of peptide a were Thr¹²⁴⁶ and Ser¹²⁴⁸. This conclusion is based on the observation that only the N-terminal portion of peptide a, after digestion at Asn-Gly, contains Glu. Furthermore, extensive digestion of this peptide with endoproteinase Glu-C results in only one product with a single Ser and a single Thr.

Table I. Observed and calculated m/z values for in vivo labeled CD45 phosphopeptides HPLC fraction Peptide Peptide site No. of Pa MH+ MH+after phosphatase Observed Calculated Observed Calculated 18.5 ab GVGTPEPT1246NpS1248AEEPEHAANGSASPAPTQSSb 1239-1268 1 2960.2 2958.9 2880.0 2878.9 21 d KTNpS1204QDKIEFHNEVDGGK 1201-1218 1 2127.4 2127.2 2048.3 2047.2  (np)c 573-585 0 1548.0 1547.6 1547.9 1547.6 21.5 e TNpS1204QDKIEFHNEVDGGK 1202-1218 1 1999.6 1999.0 1921.6 1919.0 f KTNpS1204QDKIEFHNEVD 1201-1215 1 1886.2 1884.8 1807.5 1804.8  (np)c 573-585 0 1548.0 1547.6 1549.2 1547.6 22 g TNpS1204QDKIEFHNEVD 1202-1215 1 1757.4 1756.6 1677.1 1676.6 23 h NRNpS939NVVPYDFNR 936-948 1 1676.1 1675.7 1595.4 1595.7      NSNVVPYDFNR  (np)c 938-948 0 1325.6 1325.4 1324.4 1325.4 a No. of phosphates in each peptide determined from phosphatase treatment. b Peptide a consists of a monophosphorylated form containing either phosphothreonine or phosphoserine (pS) as indicated. c (np), a non-phosphorylated CD45 peptide.

Table II. Partial list of mass values of all possible CD45 tryptic peptides around m/z 2960.2 (peptide a) Peptide MH+ No. of Phosphates Peptide site Uncut sitesb Observed Calculateda 2939.5 0 993 1017 1 2947.2 0 658 684 1 2951.2 1 915 937 5 a 2960.2 2958.9 1 1239 1268 0 2966.4 1 1100 1124 4 2968.3 1 936 959 3 2970.4 0 694 717 2 a Partial list of calculated mass values out of 724 possible tryptic peptides from CD45 cytoplasmic domain below a mass of 3000 Da. b No. of undigested trypsin sites in the predicted peptide.

Fraction 4 contained two extremely hydrophilic Ser(P)-containing phosphopeptides, b and c, which were found in the non-retained fraction from the HPLC column. Mass spectral signals for this fraction were not obtained. Their signals were probably suppressed by co-eluting salt contaminants, and efforts to remove the peptides from these contaminants were not successful.

MALDI-MS analysis of fraction 21, peptide d (Fig. 6A), exhibited one phosphorylated peak and one non-phosphorylated CD45 peptide. The mass spectral peak at m/z 2127.4 shifted to m/z 2048.3 (80 Da) after treatment with alkaline phosphatase (Table I). Ser¹²⁰⁴ was identified as the phosphorylation site, since there is only one Ser in this peptide and Thr(P) was not detected by phosphoamino acid analysis. The other major peak at m/z 1548.0 did not move after phosphatase treatment and was identified as an unphosphorylated CD45 peptide (573-585) from the membrane-proximal region of the cytoplasmic domain.

MALDI-MS analysis of fraction 21.5 showed three major peaks that correspond to CD45 peptides in the fraction (Fig. 6B). One at m/z 1548.0 represented the same unphosphorylated fragment identified in the previous fraction. The other two peaks, m/z 1886.2 and 1999.6, were found to shift by 80 Da after dephosphorylation: m/z 1999.6 to m/z 1921.6 and m/z 1886.2 to m/z 1807.5 (Fig. 6B). Two phosphopeptides were also found in this fraction by one-dimensional TLC (Fig. 4B). The peak at m/z 1999.6 was compared with the calculated mass values and identified as peptide e with a calculated m/z of 1999.0 (Table I). The only possible phosphorylation site, Ser¹²⁰⁴, was the same phosphorylation site of peptide d, and the difference between phosphopeptide d and e is one N-terminal Lys.

The observed mass of peptide f, m/z 1886.2 (Fig. 6B), did not match any tryptic phosphopeptide of CD45 but instead matched a product of peptide d involving loss of the C-terminal GGK from peptide d (loss of 242.4 Da (calculated)). Instability of the Asp-X (D-X) peptide bond at low pH is well known and likely resulted in backbone cleavage of the D-G bond in our peptides (28-31), and the procedure used involved dissolving the peptides in pH 1.9 buffer before electrophoresis or HPLC. Peptide f was phosphorylated since 80 Da was lost after phosphatase treatment. Peptide f, resulting from loss of GGK, was found in the appropriate position in HPLC (more hydrophobic), TLC (more hydrophobic) and TLE (more negatively charged) compared with peptide d. The TLC and TLE positions of peptides e and f were distinguished because a portion of peptide e was found in the preceding HPLC fraction 21 (Fig. 4B), which resulted in a weak signal of m/z 1999.6 in the mass spectra of fraction 21. Despite the fact that these peptides resulted from chemical cleavage, the peptide maps were remarkably reproducible.

The MALDI-MS spectrum of fraction 22 displayed a number of signals. Only one peptide, m/z 1757.4 (Fig. 6C, Table I), shifted by 80 Da to m/z 1677.1 after phosphatase treatment. No putative tryptic phosphopeptide has a m/z value close to 1757.4, but this number matched that expected from the loss of GGK from peptide e. Thus the four peptides, d, e, f, and g, were all derived from the same sequence region and therefore represent a single phosphorylation site, Ser¹²⁰⁴.

MALDI-mass spectrometry of HPLC fraction 23, peptide h, showed two mass spectral peaks at m/z 1325.6 and m/z 1676.1 (Fig. 7, Table I). After dephosphorylation with alkaline phosphatase, the peak at m/z 1325.6 remained unchanged (m/z 1324.4) and the peak at m/z 1676.1 shifted to m/z 1595.4 (-80 Da, loss of one phosphate). This mass matched a tryptic phosphopeptide with a calculated mass of 1675.7 designated h (Table I). Since there is only one Ser residue in this peptide, the phosphorylation site is Ser⁹³⁹. This phosphopeptide includes one site (Arg⁹³⁷) that is not cleaved by trypsin when the sequence -RXpS- is present (24). The second peak in this fraction, at m/z 1325.6, matched unphosphorylated form of the same peptide with a slightly shorter N terminus. Thus the unphosphorylated and phosphorylated Ser⁹³⁹ peptides were present in the same mass spectrum.

Since CD45 phosphorylation increases after PMA treatment of cells (21, 22), we wished to determine if any of the sites that we identified could serve as substrates for PKC phosphorylation. The phosphorylation sites at Ser⁹³⁹ (h) and Ser¹²⁰⁴ (d-g) conform to the consensus sequence described for PKC (see "Discussion"). Activation of 70Z/3.12 cells with PMA resulted in increased phosphorylation on both Ser⁹³⁹ and Ser¹²⁰⁴ (data not shown), but it is difficult to conclude that this phosphorylation was a direct result of PKC activity. To resolve this question, bacterially expressed, purified murine CD45 cytoplasmic domain was phosphorylated by a purified preparation of rat brain PKC (Fig. 8). Phosphorylated CD45 was gel-purified to separate it from the lightly autophosphorylated PKC present after the in vitro kinase reaction (Fig. 8A). Tryptic phosphopeptide mapping of CD45 indicated that there were several CD45 peptides phosphorylated. However, only one peptide matched the mobility of an in vivo phosphorylated peptide, i.e. peptide h. That peptide h is an in vitro target of PKC was confirmed by mixing the eluted PKC peptide labeled h (from Fig. 8B) with in vivo labeled CD45 peptides (Fig. 8D). PKC-labeled peptide h contained only Ser(P) (data not shown), and its position was precisely coincident with in vivo labeled peptide h.

DISCUSSION

The identification of CD45 in vivo phosphorylation sites is essential to determine which of the many possible sites may be functionally relevant to the modulation of activity and to the docking of CD45 associated proteins. The current study was designed to characterize the in vivo phosphorylation sites of CD45. Four CD45 in vivo phosphorylation sites were identified using a combination of two-dimensional phosphopeptide mapping, reverse-phase HPLC, phosphoamino acid analysis, MALDI-MS, and specific enzymatic degradation. The method used is advantageous because phosphopeptides can be detected in the picomole range, because phosphorylation can be confirmed by observation of a loss of 80 Da after dephosphorylation, and because the target phosphopeptide does not need to be absolutely pure (26). The major type of phosphorylated amino acid observed in CD45 was Ser(P) with a minor amount of Thr(P) in peptide a. This is the first time that Thr(P) has been observed in CD45. Three phosphorylation sites were identified in the C-terminal tail, Ser¹²⁰⁴, Thr¹²⁴⁶, and Ser¹²⁴⁸ (summarized in Fig. 9) and one phosphorylation site, Ser⁹³⁹, was in the D2 domain (peptide h). No phosphorylation sites were observed in the D1 domain or in the connector between PTPase domains.

The phosphorylation site at Ser⁹³⁹ (peptide h) is conserved in mouse, rat, and human CD45; instead of Ser⁹³⁹, chicken and shark CD45 have a Ser in an adjacent residue. In addition, the sequence around the Ser⁹³⁹ is highly conserved among PTPases. The three-dimensional structure of the phosphatases has been found to be remarkably conserved (32), and comparison of CD45 and PTP1B predicts that Ser⁹³⁹ is located at the apex of the substrate binding loop at a position homologous to Arg⁴⁷ in PTP1B (33). This suggests that phosphorylation of Ser⁹³⁹ may be important in the regulation of binding of substrates in the D2 domain or in the binding of signaling proteins which may interact with the D2 active site. Ser⁹³⁹ also falls in the homologous interaction region resulting in dimer formation of RPTPD1, which blocks the active site (34). Potential dimerization of CD45 (D1-D1 or D1-D2) may modulate activity since the PTPase activity of a CD45-epidermal growth factor receptor chimeric molecule was functionally inactivated upon induced dimerization with epidermal growth factor (9).

Peptide h is also of interest because it contains the target for Abl kinase, which is proposed to modulate the PTPase activity of CD45-D2 (19). Phosphotyrosine was not detected in our in vivo study, even when the PTPase inhibitor (phenylarsine oxide) was included in the isolation. It is possible that CD45 tyrosine phosphorylation occurs transiently or at very low stoichiometry. Peptide h also overlaps with a fodrin-binding site (E⁹³⁰ENKKKNRNS⁹³⁹) of CD45, and fodrin binding increases the PTPase activity of CD45 (35, 36). Phosphorylation at Ser⁹³⁹ may serve to regulate the binding of CD45 to the cytoskeleton. Other studies have suggested that CD45 interacts with the cytoskeleton and has a role in the coordination of cytoskeletal remodeling (37, 38).

In the current report, peptide h (Ser⁹³⁹) was not the major phosphorylated peptide observed. However, it is unlikely that phosphorylation reached equilibration during the labeling since CD45 is expressed at high levels and the turnover of CD45 protein is very slow (39). Thus, the phosphorylation of CD45 in our study probably resulted primarily from turnover of phosphate at individual phosphorylation sites. It is difficult to reach conclusions about stoichiometry from our ³²P incorporation experiments, since, for example, a very stable phosphorylation site would only become weakly phosphorylated. In the current study the stoichiometry of phosphorylation at Ser⁹³⁹ was more directly estimated by comparison of the MALDI-MS signals for peptide h and an almost identical non-phosphorylated peptide containing Ser⁹³⁹ in the same HPLC fraction (Table I and Fig. 7). Assuming comparable trypsin cleavage, the intensity of the MALDI-MS signal of the dephosphorylated form of peptide h was of the same magnitude as the equivalent non-phosphorylated peptide in the same mass spectrum, suggesting equal abundance.

The phosphorylation site at Ser¹²⁰⁴ is conserved in mouse and rat but not in human, chicken, or shark CD45. The other sites in the C-terminal tail, Thr¹²⁴⁶ and Ser¹²⁴⁸, are not precisely conserved among these species, but this region (within 1-2 residues) contains multiple Ser and/or Thr residues suggesting that this may also be a phosphorylation site common to all species. The MALDI-MS signals of peptides b and c (HPLC fraction 4) were not observed because of the presence of salts in this fraction. It is possible that phosphopeptides in this fraction are homologous to a peptide obtained from a tryptic digest of in vivo ³²P-labeled, human Jurkat cell CD45 (40). This sequence, which is highly conserved in mouse CD45, contains casein kinase II phosphorylation sites and in vitro phosphorylation of the cytoplasmic domain of CD45 by casein kinase II revealed one major phosphopeptide which migrated very close to peptides b and c in two-dimensional mapping (data not shown).

The phosphorylation sites Ser⁹³⁹ and Ser¹²⁰⁴ conform to the consensus sequence described for PKC (41). We have confirmed that, of these potential in vivo phosphorylation sites, only Ser⁹³⁹ is a target for in vitro PKC phosphorylation. Several other peptides were also labeled in the in vitro PKC reaction, which did not appear in in vivo labeled CD45. Increased phosphorylation at Ser⁹³⁹ and Ser¹²⁰⁴ was observed after PMA treatment of 70Z/3.12 cells (data not shown), and other studies have indicated that CD45 was phosphorylated after PMA treatment (21, 22, 42). It has been reported that other PTPases such as PTP1B, RPTP, and PTP-PEST are also substrates of PKC (43-45), and in vitro PKC phosphorylation of PTP-PEST resulted in decreased PTPase activity (45).

Our results suggest that Ser⁹³⁹ may be an important phosphorylation site, which is phosphorylated in vivo by PKC. This phosphorylation may play a role in the regulation of D2 domain activity, in potential dimerization involving the D2 domain, or in the interactions of other molecules with the active site of the D2 domain. The C-terminal phosphorylation sites may also have a regulatory role in CD45 activity. Direct confirmation of the role of these phosphorylation sites in the function of CD45 is currently under investigation by site-directed mutagenesis.