INTRODUCTION

Human single-stranded DNA-binding protein (HSSB),¹ also called RPA, is a heterotrimeric complex that consists of three subunits, p70, p34, and p11 (1-4). HSSB was initially discovered as a protein essential for several stages of the in vitro replication of SV40 DNA. It stimulates SV40 T antigen (T-Ag) catalyzed unwinding of origin-containing duplex DNA (2, 5, 6) and is an essential component of the initiation complex, which includes T-Ag, HSSB, and the DNA polymerase -primase complex. Specific protein-protein interactions have been observed between these proteins (7-9). Furthermore, HSSB stimulates the elongation of primed DNA templates catalyzed by DNA polymerases , , and  (8, 10-13).

HSSB is also required for nucleotide excision repair (14). It participates in the recognition of UV-damaged DNA by forming a complex with XPA and stimulates the incision activities of XPG and XPF-ERCC1 (15-18). Formation of a protein complex between HSSB and the human homologue of Rad52 (19) and stimulation of human homologue paring protein-1 by HSSB (20) indicate that HSSB is important in DNA recombination and double-stranded DNA break repair. This is underscored by the observations that mutations within the p70 subunit of Saccharomyces cerevisiae SSB affect DNA recombination and double-stranded DNA break repair (21-23). It has also been shown that HSSB binds to the acidic domains of transcription factors VP16 and p53 (24-26) and in S. cerevisiae ScSSB binds to DNA sequences that regulate transcription (27, 28). These observations suggest that HSSB also has important functions in transcription and its regulation.

The p34 subunit of HSSB is phosphorylated in a cell cycle-dependent manner (29-31). Phosphorylated forms first appear during the G₁ to S transition and persist through the S phase. As the cell cycle progresses through late M phase, HSSB-p34 is dephosphorylated. In x-ray or UV-irradiated cells, the level of p34 phosphorylation also increases dramatically (32, 33). While the protein kinases responsible for the phosphorylation of p34 in vivo are not known, in vitro studies have shown that the HSSB-p34 subunit is phosphorylated by both the Cdc2 kinase and the DNA-dependent protein kinase (DNA-PK) (29, 30, 34, 35).

Although it has been suggested that the phosphorylation of HSSB-p34 may play an important role in cell cycle regulation of DNA synthesis and in coordinating DNA replication and repair (29-31), the biological significance of HSSB-p34 phosphorylation is still not clear. Several studies have shown that HSSB-p34 phosphorylation does not affect its ability to bind single-stranded DNA, support SV40 DNA replication (35-38), or nucleotide excision repair (38).

Alanine substitutions at Ser-23 and Ser-29, two putative Cdc2 kinase phosphorylation sites in HSSB-p34, do not affect the binding of HSSB to single-stranded DNA or its ability to support SV40 DNA replication (35, 37). HSSB, reconstituted with a mutant p34 subunit containing a deletion of 30-33 N-terminal amino acids, was not phosphorylated by either the Cdc2 kinase or DNA-PK, but efficiently supported SV40 DNA replication (35, 37). In contrast, a mutant HSSB, reconstituted with a C-terminal 30-amino acid truncated p34 subunit, was phosphorylated by both kinases, but did not support SV40 DNA replication (37). These studies showed that phosphorylation of the p34 subunit of HSSB has no detectable effect on the in vitro replication of SV40 DNA.

The phosphorylation sites in HSSB-p34 have been inferred from the presence of putative kinase consensus sequences but have not been identified. In this study, a combination of biochemical and biophysical approaches were used to define both DNA-PK and Cdc2 kinase phosphorylation sites in the p34 subunit of HSSB. We have mapped two DNA-PK phosphorylation sites in the HSSB-p34 subunit at Thr-21 and Ser-33 and have identified Ser-29 as the major Cdc2 kinase phosphorylation site.

MATERIALS AND METHODS

HSSB, DNA-PK, and Ku were purified from HeLa cell as described previously (12, 39). The Cdc2-cyclin B kinase complex was purchased from Upstate Biotechnology Inc. Synthetic peptides were made using the ABI-433 peptide synthesizer (Microchemistry Core Facility, Memorial Sloan Kettering Cancer Institute).

Standard reaction mixtures (20 µl) contained 50 mM Hepes (pH 7.5), 10 mM MgCl₂, 0.1 mM EDTA, 0.5 mM dithiothreitol, 4 µg of bovine serum albumin, 50 mM KCl, 1 µg of sonicated calf thymus DNA (average size ~300 base pairs), 16 ng of Ku protein, and 3 units of DNA-PK (1 unit of enzyme incorporated 1 nmol of ³²P into peptide PK53 (40) in the presence of 16 ng of Ku at 30 °C after 30 min). Reactions with synthetic peptides contained 2 nmol of each peptide and 500 µM [-³²P]-ATP (~2000-4000 cpm/pmol). Incubation was for 2 h at 30 °C after which reaction mixtures were subjected to 15% SDS-PAGE analysis. Gels were dried using the Bio-Rad vacuum drier at 70 °C for 1 h followed by autoradiography.

In reactions containing HSSB, unless otherwise indicated, 300 ng (2.7 pmol) of HSSB were used with 100 µM [-³²P]-ATP (~10,000-20,000 cpm/pmol). Reactions were incubated for 1 h at 30 °C followed by 12.5% SDS-PAGE analysis. Gels were dried and followed by autoradiography as described above, or proteins were transferred to a nitrocellulose membrane for Western blot analysis.

Standard reaction mixtures (10 µl) contained 40 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 1 mM dithiothreitol, 50 µM [-³²P]ATP (~20,000 cpm/pmol), and 5 ng of Cdc2-cyclin B. Substrates used were as indicated in figure legends. Incubation was for 1 h at 37 °C followed by 12.5% (for HSSB) or 15% (for peptides) SDS-PAGE analysis. Gels were dried and subjected to autoradiography.

Phosphorylated HSSB derivatives were resolved through a 12.5% SDS-PAGE and transferred to a nitrocellulose membrane (0.2 mm, Schleicher & Shuell). The protein bands of interest (³²P-phosphorylated and unphosphorylated proteins) were excised from the Ponceau S-stained nitrocellulose membrane, and peptides were generated by in situ proteolysis (41, 42). Briefly, digestion was carried out in reaction mixtures (25 µl) using either 0.2 µg of trypsin (Promega, Madison, WI) or chymotrypsin (sequencing grade; Boehringer Mannheim) in 100 mM NH₄HCO₃ (supplemented with 1% Zwittergent 3-16) at 37 °C for 2 h. The resulting digest was reduced and S-alkylated with 0.1% -mercaptoethanol (Bio-Rad) and 0.3% 4-vinylpyridine (Aldrich), respectively, and fractionated by reverse-phase HPLC. Solvents and HPLC system configuration were as described elsewhere (43), except that a 2.1-mm 214 TP54 Vydac C4 (Separations Group, Hesperia, CA) column was used with gradient elution at a flow rate of 100 µl/min. Tryptophan-containing peptides were identified by ratio analysis of UV absorbance at 297 and 277 nm, monitored in real time using an Applied Biosystems (Foster City, CA) model 1000S diode-array detector (44). Peak fractions were collected and aliquots (2 µl) were subjected to liquid scintillation counting, and all fractions were stored at 70 °C prior to further analysis.

Peak fractions that contained ³²P-labeled peptides were analyzed by a combination of automated Edman degradation and matrix-assisted laser-desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) as described previously (43-45). After storage, fractions were supplemented with trifluoroacetic acid to give a final concentration of 10% before loading onto sequencer disks and mass spectrometer targets. Mass analysis (on 2% aliquots) was carried out using a model Voyager RP MALDI-TOF MS instrument (Vestec/PerSeptive, Framingham, MA) in the linear mode, with a 337-nm output nitrogen laser, a 1.3-m flight tube, and -cyano-4-hydroxy cinnamic acid (pre-made solution obtained from Linear Sci., Reno, NV) as the matrix. A 30-kV ion acceleration voltage (grid voltage at 70%; guide wire voltage at 0.1%) and 2.0-kV multiplier voltage were used (positive ion mode). Laser irradiance and number of acquisitions were adjusted as judged from optimal deflections of specific maxima, using a TDS 520 Tektronix (Beaverton, OR) digitizing oscilloscope. m/z (mass to charge) spectra were generated from the time-of-flight files using GRAMS (Galactic Ind., Salem, NH) data analysis software. Each sample was analyzed twice, in the presence and absence of two calibrants (25 fmol of APID and P8930), as described elsewhere (45). Chemical sequencing (on 95% of the sample) was done using a model 477A instrument from Applied Biosystems. Stepwise liberated PTH-derivatives were identified using an "on-line" 120A HPLC system (Applied Biosystems) equipped with a PTH C18 (2.1 × 220 mm; 5-µm particle size) column (Applied Biosystems). Instruments and procedures were optimized for femtomole-level PTH-derivative analysis as described previously (44, 46). Average isotopic masses of predicted tryptic (chymotryptic) peptides were calculated using the ProComp version 1.2 software (obtained from Dr. P. C. Andrews, University of Michigan, Ann Arbor, MI).

Subtractive mass analysis of peptides partially truncated by Edman degradation was done as follows. Polybrene (Applied Biosystems) was applied to a glass-fiber filter (9-mm diameter; Applied Biosystems) and precycled in a flow-through microcartridge as described previously (46). The filter was then removed and carefully cut with a surgical scalpel into six equally sized (pie-shaped) pieces. Aliquots (25% each) of the sample were applied to one filter segment, which was then inserted into a "Blott-cartridge" (Applied Biosystems) for automated Edman sequencing. To reduce sample washout, trifluoroacetic acid was delivered as vapor from the X1 bottle position. After a predetermined number of cycles, the filter piece was removed from the instrument and the truncated peptide eluted by adding 5 µl of 10% trifluoroacetic acid, 30% acetonitrile and incubated for 5 min inside a capped micro-Eppendorf tube at room temperature. Samples were then centrifuged for 10 min at 13,000 rpm. The eluate (1 µl) was mixed with matrix prepared as above for MALDI-TOF MS analysis (see above).

Proteins or peptides were transferred from SDS-PAGE to polyvinylidene difluoride membrane (Schleicher & Shuell) as described elsewhere (47) with the following modifications. Peptides were transferred at 100 V for 1 h, whereas the p34 protein was transferred at 200 mA for 1 h. After transfer, the polyvinylidene difluoride membrane was washed with three to four changes of water every 3 min. The filter was air-dried and exposed to x-ray film. The film and filter were carefully aligned, and the radioactive protein or peptide bands were excised and cut into small pieces.

The pieces of polyvinylidene difluoride membrane were placed into a tube (Corning Pyrex tube) for hydrolysis. The membrane was briefly submerged in methanol followed by the immediate removal of the solvent; 20 µl of 6 N HCl were added to the tube, which was placed in a microvial; 0.2 ml of 6 N HCl solution was added to the bottom of the microvial, which was capped and placed in the PICO-TAG workstation (Waters Inc.). A vacuum was applied to the microvial until the 6 N HCl solution started to bubble (about 15 s). The vacuum was closed, and the microvial was filled with N₂ for 10 s. After repeating this procedure three times, the microvial was incubated at 110 °C for 1 h to hydrolyze the protein and peptides.

After hydrolysis was complete, the supernatant from each tube was recovered, and the small pieces of filter were rinsed three times, each with 15 µl of 0.1 N HCl and 30% methanol. The rinsing solutions were combined with the original supernatant and dried in a Savant speed vacuum concentrator. The dried hydrolysate was dissolved in 3-5 µl of double-distilled H₂O, and 0.5 µl of an unlabeled standard phosphoamino acid mixture was added (1 mg/ml each of phosphoserine, phosphothreonine and phosphotyrosine). The same amount of radioactivity (1000 cpm) from each sample was spotted onto a TLC plate (Kodak 160-mm cellulose TLC plate without fluorescent indicator, 20 × 20 cm) and allowed to air dry at room temperature. Vertical chromatography was carried out in a solution containing isobutyric acid, 0.5 M NH₄OH (5:3, v/v). After chromatography, TLC plates were dried, sprayed with ninhydrin to visualize the position of each phosphoamino acid standard, and then exposed to x-ray film.

RESULTS

We have previously shown that the p34 subunit of HSSB (HSSB-p34) is phosphorylated by a HeLa cell fraction containing both DNA-PK and Ku in a DNA-dependent manner (34). To determine the role of Ku in the DNA-dependent phosphorylation of HSSB-p34, HSSB was incubated with highly purified DNA-PK in the presence or absence of Ku and calf thymus DNA. The phosphorylation of HSSB-p34 was analyzed by measuring ³²P incorporation from [-³²P]ATP into the p34 subunit (Fig. 1, upper panel) and examining the mobility changes of the p34 subunit following SDS-PAGE by immunoblot assay using a p34 monoclonal antibody (Fig. 1, lower panel). As shown in Fig. 1, lane 6, the p34 subunit was efficiently phosphorylated in the presence of DNA-PK, Ku, and calf thymus DNA. Omission of either Ku (lane 2), calf thymus DNA (lane 3), or DNA-PK (lane 4) markedly reduced ³²P incorporation. The incorporation of phosphate(s) into the p34 subunit resulted in the appearance of p34-P1, a slower migrating form of p34 following SDS-PAGE (p34-P1, Fig. 1, lower panel). The incorporation of additional phosphate residues into p34 resulted in a hyperphosphorylated species of the subunit, leading to further retardation of its migration (p34-P2, Fig. 1, lower panel).

These results are consistent with our previous findings and further demonstrate that Ku and DNA are required for the efficient phosphorylation of HSSB-p34 by DNA-PK.

We mapped the amino acid residues of HSSB-p34 that were phosphorylated by DNA-PK using the procedure outlined in Fig. 2. For this purpose, the two species of ³²P-labeled phosphorylated p34 were resolved by SDS-PAGE (Fig. 1) and transferred to nitrocellulose membranes. The phosphorylated forms of p34 were subjected to trypsin and chymotrypsin digestion, and the digested mixture was fractionated by reverse-phase HPLC, as described under "Materials and Methods." Four ³²P-radiolabeled peptides consisting of two major peaks, T3 and T4, and two minor peaks, T1 and T2, were isolated from the tryptic digestion mixture of p34-P1 after separation by reverse-phase HPLC (Fig. 3) and structurally characterized by MALDI-TOF MS. All four peptides contained overlapping sequences, an identical C-terminal residue (Lys-37), and variable truncated N termini, most likely due to the action of a contaminating chymotryptic-like protease activity (Table Ia). Added diversity was derived from the oxidative state of the acetylated N-terminal Met (methionine versus methionine sulfoxide). As shown in Table Ia, the mass data indicated the presence of a single phosphate group, located between amino acid residues 15 and 37 of HSSB-p34. This region contains four putative phosphorylation sites (Ser-23, Ser-29, Ser-33, and Thr-21) (Table I).

Table I. Analysis of tryptic and chymotryptic phosphopeptides derived from HSSB-p34 (p34-P1) phosphorylated by DNA-PK The peptides listed here were selected on the basis of 32P incorporation and were assumed to include all phosphorylation sites in the p34-P1. Experimental mass (m/z) was obtained by MALDI-TOF MS (see "Materials and Methods"); chemical sequencing data (including initial yields in pmol) are also listed. The theoretical mass shown for each predicted peptide was calculated as the sum of the isotopic masses of the composite amino acids and all the modifying groups. [MH+] represents molecular mass [M] plus one proton [H+]. x (in chemical sequencing results), uninterpretable or no signal; IY, initial yield. The peptides listed here were selected on the basis of 32P incorporation and were assumed to include all phosphorylation sites in the p34-P1. Experimental mass (m/z) was obtained by MALDI-TOF MS (see "Materials and Methods"); chemical sequencing data (including initial yields in pmol) are also listed. The theoretical mass shown for each predicted peptide was calculated as the sum of the isotopic masses of the composite amino acids and all the modifying groups. [MH+] represents molecular mass [M] plus one proton [H+]. x (in chemical sequencing results), uninterpretable or no signal; IY, initial yield. Peptide Experimental mass Chemical sequencing (IY, pmol) Position in HSSB-P34 Modifications Calculated mass a. Tryptic digest m/z [MH+] T4 3815.2 NDa Acetyl-1-37b + 1 phosphate 3814.90 T3 3828.4 ND Acetyl-1-37[Metox]b,c + 1 phosphate 3830.90 T2 2669.3 ND 10 -37 + 1 phosphate 2670.69 T1 2188.9 ND 15 -37 + 1 phosphate 2189.21 b. Chymotryptic digest C2 1236.4 GGAGGYxQSPGGx (3.0) 15 -27 + 1 phosphate 1236.20 C1 789.3 GGAGGYxQ (1.5) 15 -22 + 1 phosphate 790.72 a ND, not determined. b Acetyl, N-terminus acetylated. c Metox., methionine sulfoxide.

Two relatively small phosphorylated chymotryptic fragments (C1 and C2) were also isolated from the p34-P1 chymotryptic digestion mixture. As shown in Table Ib, the combined sequencing and mass spectrometry data of these peptides narrowed the phosphorylated site to an 8-amino acid peptide (C1, GGAGGYTQ; see Table I), which enabled positive identification of Thr-21 as the single phosphorylated residue in this peptide. This conclusion was further supported by the following observations. (a) Thr-21 was not observed during automated chemical sequencing of either peptide (C1 and C2 in Table I), which indicated that it was modified. Tyr-20 and Ser-23, on the other hand, were detected in amounts equivalent to the other amino acids (G, A, Q, and P) present in these peptides. Furthermore, neither phosphotyrosine nor its dephosphorylated form have been detected during noncovalent chemical sequencing (54). (b) The phosphorylated peptide C1 (residues 15-22) contains the sequence that includes a single threonine (at position 21) but no serine.

Attempts to isolate ³²P-labeled peptides following fractionation of trypsin digests of p34-P2 were not successful due to the low recovery of the hyperphosphorylated p34 protein. The reason for this difficulty is unknown. An alternative approach, however, was utilized to determine additional DNA-PK phosphorylation sites in the HSSB-p34.

It has been shown recently that HSSB-p34 is rapidly degraded by trypsin to a ~28-kDa fragment (which contains the middle and C-terminal domain) and a ~4-kDa N-terminal fragment (48) which contains all the residues phosphorylated by DNA-PK (49). Our trypsin in situ digestion experiments showed that this ~4-kDa N-terminal fragment corresponded to amino acids 1-37 of the p34 subunit (Tables Ia and IIa). One threonine and 8 serine residues were located in this region (Thr-21, Ser-4, -8, -11, -12, -13, -23, -29, and -33). Among them, two serine residues, Ser-23 and Ser-33, contained a DNA-PK target sequence ((S/Q) or (Q/S)) in addition to Thr-21, which, as described above, was one of the DNA-PK phosphorylation sites. To determine whether these two serine residues were also phosphorylated by DNA-PK, two peptides were synthesized that included Ser-23 and Ser-33, respectively (see Fig. 4B for the sequences of the peptides). These peptides were then examined as substrates for DNA-PK. As shown in Fig. 4A, peptide II, containing Ser-33, was phosphorylated, whereas peptide I, which contained only Ser-23 (Thr-21 in this peptide was changed to alanine to focus on the phosphorylation of Ser-23), was not. The phosphorylation of peptide II required both DNA (Fig. 4A) and Ku (data not shown).

Table II. Analysis of tryptic and chymotryptic phosphopeptides derived from HSSB-p34 phosphorylated by Cdc2-cyclinB The peptides listed here were selected on the basis of 32P incorporation and were assumed to include all phosphorylation sites. Experimental mass (m/z) was obtained by MALDI-TOF MS (see "Materials and Methods"); chemical sequencing data (including initial yields in pmol) are also listed. Theoretical mass calculations for predicted peptides were carried out by summing isotopic masses of the composite amino acids and of all the modifying groups. [MH+] represents molecular mass [M] plus one proton [H+]. x (in chemical sequencing results), uninterpretable or no signal; IY, initial yield. The peptides listed here were selected on the basis of 32P incorporation and were assumed to include all phosphorylation sites. Experimental mass (m/z) was obtained by MALDI-TOF MS (see "Materials and Methods"); chemical sequencing data (including initial yields in pmol) are also listed. Theoretical mass calculations for predicted peptides were carried out by summing isotopic masses of the composite amino acids and of all the modifying groups. [MH+] represents molecular mass [M] plus one proton [H+]. x (in chemical sequencing results), uninterpretable or no signal; IY, initial yield. HSSB-P34 Phosphorylation by CDC2 Kinase Peptide Experimental mass Chemical sequencing (IY, pmol) Position in HSSB-P34 Modifications Calculated mass a. Tryptic digest m/z [MH+] T6 3814.2  a Acetyl-1-37b + 1 phosphate 3814.90 T5 3829.9  a Acetyl-1-37b[Metox]c + 1 phosphate 3830.90 T4 3339.2 xGFExYx (0.42) 4 -37 + 1 phosphate 3341.37 T3 2670.3 ND 10 -37 + 1 phosphate 2670.69 T2 2190.0 xxAGGYTQxPGGF (0.98) 15 -37 + 1 phosphate 2189.21 T1 1726.9 TQxPGxFxxPA (0.61) 21 -37 + 1 phosphate 1726.74 b. Chymotryptic digest C2 2098.8 TQSPGGFGxPAPSQAEKKSR (1.4) 21 -40 + 1 phosphate 2098.20 C1 1854.0 TQxPGGFx (0.49) 21 -38 + 1 phosphate 1854.91 C1-1a.a.d 1890.0 NDe 22-38 + PTCf + 1 phosphate 1888.83 2024.3 ND 22-38 + 2(PTC)f + 1 phosphate 2023.83 C1-3a.ad 1808.4 ND 24-38 + 2(PTC)f + 1 phosphate 1808.62 a No signal observed. b Acetyl, N terminus acetylated. c Metox, methionine sulfoxide. d -1 (-3) a.a., one (or 3) amino acids removed by Edman degradation. e ND, not done. f PTC, phenyl thiocarbamyl derivative of lysine side chain (-amine).

Since peptide II contained two serine residues (Ser-29 and -33), one of these serine residues was substituted by an alanine to determine the residues phosphorylated in this peptide. As summarized in Fig. 4B, the peptide containing Ser-33, which is within the DNA-PK target sequence, was phosphorylated, whereas Ser-29 was not. Furthermore, following substitution of the Gln-34 in peptide II with a glutamate, which changed the motif essential for DNA-PK recognition (a glutamine following a serine residue), the phosphorylation of peptide II by DNA-PK was abolished (Fig. 4B). These results suggest that, in addition to Thr-21, Ser-33 is phosphorylated by DNA-PK, whereas Ser-23 is not.

To further confirm that Thr-21 and Ser-33 are DNA-PK phosphorylation sites in HSSB-p34, a mutant HSSB was constructed that contained 7 alanine substitutions within the HSSB-p34 amino acids 1 to 37. In this mutant all possible residues that could be phosphorylated in the ~4-kDa N-terminal fragment, except Ser-33 and Thr-21, were changed to alanines. This mutant protein was subjected to phosphorylation by DNA-PK. Both wild-type and mutant HSSB were phosphorylated by DNA-PK in a Ku and DNA-dependent manner (Fig. 5). As shown in Fig. 5, the phosphorylation of this mutant protein gave rise to the same slow migrating species of p34 as wild-type HSSB following SDS-PAGE (lanes 3, 4, 7, 8, p34-P1 and p34-P2, respectively). This result was consistent with the notion that Thr-21 and Ser-33 are DNA-PK phosphorylation sites in HSSB-p34, whereas the other serine residues in the region spanning amino acids 1-37 are not. Compared with the wild-type HSSB, the mutant protein was phosphorylated with a somewhat reduced efficiency (~70%). The reason for this discrepancy is not clear.

Phosphorylated p34 species (p34-P1 and p34-P2, shown in Fig. 1) were isolated and subjected to acid hydrolysis. As shown in Fig. 6A, hydrolysis of p34-P2 led to the detection of both phosphoserine and phosphothreonine (lane 3), whereas hydrolysis of p34-P1 resulted in the detection of phosphothreonine and only minor levels of phosphoserine (lane 2). This observation is in keeping with the results shown above, indicating that both Thr-21 and Ser-33 are phosphorylated by DNA-PK in hyperphosphorylated HSSB-p34, p34-P2, and only Thr-21 is phosphorylated in the p34-P1 species.

Studies on the rate of HSSB-p34 phosphorylation by DNA-PK were also carried out. As shown in Fig. 6B, the phosphorylated HSSB-p34 form, p34-P1, was detected after 15 min of incubation. The level of p34-P1 increased as the phosphorylation reaction proceeded and plateaued after 90 min. In contrast, the hyperphosphorylated product, p34-P2, was not clearly visible until after 30 min of incubation, and the amount of p34-P2 continued to increase up to 2 h. In the presence of a saturating amount of DNA-PK and lower levels of substrate, all of the HSSB-p34 subunit was converted to the hyperphosphorylated form, p34-P2 (data not shown). These observations suggest that DNA-PK initially phosphorylated Thr-21, resulting in the p34-P1 species, followed by phosphorylation of Ser-33 and conversion to the p34-P2 species.

Previously we and others have shown that, in addition to DNA-PK, the Cdc2 kinase also phosphorylates HSSB-p34 in vitro (29-31, 34). This reaction yielded two forms of p34 that migrated slower than the unphosphorylated p34 subunit following SDS-PAGE, a predominant phosphorylated form p34-PI and a minor band, p34-PII (see Fig. 7, inset). Approximately one phosphate residue was incorporated into each molecule of p34-PI (34). p34-PI was isolated and subjected to trypsin digestion followed by reverse-phase HPLC fractionation. Six ³²P-radiolabeled peptides were isolated (one major peak, T6, and five minor peaks, T1 to T5) (Fig. 7). These peptides were characterized by chemical sequencing and MALDI-TOF MS (Table II). All six peptides contained overlapping sequences, an identical C-terminal residue (Lys-37), and variable truncated N termini, most likely due to the action of a contaminating chymotryptic-like protease activity (Table IIa). Added diversity was derived from the oxidative state of the acetylated, N-terminal Met (methionine versus methionine sulfoxide). The combined mass data indicated the presence of a single phosphate group situated between residues 20 and 37 of HSSB-p34, a region that contains three serines and one threonine (Ser-23, -29, -33, and Thr-21).

Following chymotrypsin digestion of p34-PI, two ³²P-labeled peptides, both mapping to the same region (21-38 and 21-40) of the HSSB-p34 subunit, were isolated following reverse-phase HPLC fractionation. Each peptide was shown to be singly phosphorylated (C1 and C2, respectively, Table IIb). After stepwise removal of the first three residues in peptide C1 (Thr-Gln-Ser) followed by mass analysis, we concluded that the ³²P-phosphate moiety was still associated with the truncated peptide and that the single phosphorylated site was either Ser-29 or Ser-33 (Table IIb; peptide C1-1a.a. and C1-3a.a.).

Though only Ser-29 within this peptide is located in a Cdc2 kinase target sequence (serine-proline), direct evidence was needed to further define which of these two serine residues was phosphorylated by the Cdc2-cyclin B kinase. Synthetic peptides were constructed to include these two serine residues (Fig. 8, top, wild type). Peptides in which either one of these two serines was changed to alanine (mutant 1 and mutant 2, see Fig. 8, top) were also synthesized. As shown in Fig. 8, bottom, the Cdc2-cyclin B kinase phosphorylated both the wild-type and mutant peptide 2, which contained an alanine substitution at Ser-33. Mutant peptide 1, which contained an alanine substitution at Ser-29, was not phosphorylated. This suggests that Ser-29 but not Ser-33 is the major Cdc2 kinase phosphorylation site in HSSB-p34.

Other residue(s) within HSSB-p34 may be targets for phosphorylation by the Cdc2 kinase, as indicated by the presence of a minor hyperphosphorylated p34 species, p34-PII (Fig. 7, inset). Mutant HSSB-p34 in which the Ser-29 was changed to an alanine residue was phosphorylated by Cdc2-cyclin B kinase, although with much reduced efficiency (36). These results indicate that HSSB-p34 probably contains multiple Cdc2 phosphorylation sites, with Ser-29 being the predominant one.

DISCUSSION

We have shown that the p34 subunit of HSSB is efficiently phosphorylated at residues Thr-21 and Ser-33 by DNA-PK in the presence of Ku and calf thymus DNA. In this reaction, Thr-21 was more rapidly phosphorylated than Ser-33. We also identified Ser-29 as the primary Cdc2 phosphorylation site in HSSB-p34 (Fig. 9).

DNA-PK preferentially phosphorylates serine or threonine residues that are followed or preceded by glutamine residues ((S/T)-Q or Q-(S/T)) (50). Such target sequences are found within the HSSB-p34 subunit, including five serine residues (Ser-23, -33, -52, -72, and -174) and one threonine residue (Thr-21). However, of the six potential target sites, only Thr-21 and Ser-33 were efficiently phosphorylated by DNA-PK. Earlier studies have shown that, in addition to the target sequences described above, poorly characterized additional sequences and/or tertiary protein structures may be required for efficient DNA-PK phosphorylation (40). These additional considerations may explain why only a few of the synthetic peptides that contain Ser-23, Ser-33, Ser-52, Ser-72, or Ser-174 were efficiently phosphorylated by DNA-PK (data not shown). Since only Thr-21 and Ser-33 in HSSB-p34 were phosphorylated efficiently, the other potential phosphorylation sites may not meet the poorly defined additional requirements. Another consideration is that the sites not phosphorylated may be inaccessible to DNA-PK due to the association of p34 with p70 and p11 subunit. It has been shown that the p34 subunit is tightly associated with the p70 and p11 subunits via interactions with the middle and C-terminal regions of p34 (51), although the last 33 amino acid residues in the C-terminal region of p34 are not essential for this interaction (37). This association may sterically block the phosphorylation sites (Ser-52, -72, and -174) located within these regions of p34. Proteolysis experiments done by Gomes et al. (48) and protease digestion results shown in this report suggest that the N-terminal region of HSSB-p34 may be more accessible for enzymatic modification.

The association of p34 with p70 and p11 enabled a more efficient phosphorylation of Thr-21 and Ser-33 in the N-terminal region. As we have previously shown, whereas the p34 subunit alone was a poor substrate for DNA-PK, its association with the p11 subunit dramatically increased phosphorylation of the p34 subunit to a level equivalent to that observed with the p70-p34-p11 trimeric complex (39). This suggests that, although only the p34 subunit contains DNA-PK phosphorylation sites, the other two associated subunits, especially the p11 subunit, are important for the efficient phosphorylation of HSSB-p34. The mechanism underlying this observation is not known. One possibility is that the interaction between the p11 subunit and the p34 subunit renders the substrate more accessible to DNA-PK.

It has been shown recently that HSSB interacts with DNA in at least two different modes. Different physical changes occur in the HSSB protein structure when HSSB binds to single-stranded DNAs of different length. This could alter the efficiency of HSSB-p34 phosphorylation by DNA-PK (52). Indeed, depending on the type of DNA used in the reaction, an additional phosphorylation product was observed. For example, when single-stranded DNAs such as X174 or poly(dT)₁₇₀ were used in the phosphorylation reaction, an extra hyperphosphorylated p34 protein band was observed migrating even slower than p34-P2 on SDS-PAGE, which suggests that additional phosphorylation sites might exist in the p34 subunit of HSSB (34). However, this species of phosphorylated HSSB-p34 accounted for less than 4% of total input substrate (34).² Furthermore, under the phosphorylation conditions described in this report, p34-P1 and p34-P2, which resulted from phosphorylation of Thr-21 and Ser-33, accounted for more than 99% of the phosphorylated products.

The consensus motif for the Cdc2 kinase is a serine or threonine residue followed by proline ((S/T)-P). There are two such serine residues in HSSB-p34 (Ser-23 and Ser-29). We identified Ser-29 as the predominant phosphorylation site. This result is consistent with the observations made by Henricksen and Wold (53), who found that HSSB containing a serine to alanine substitution at position 23 in the p34 subunit was phosphorylated as efficiently as the wild-type protein. However, a mutant HSSB with an alanine substitution at Ser-29 in the p34 subunit was poorly phosphorylated. When both Ser-23 and Ser-29 in HSSB-p34 were changed to alanine residues, this mutant protein was still phosphorylated by Cdc2 kinase in vitro. Although this result points to the presence of an additional Cdc2 kinase target sequence differing from the (S/T)-P motif in the p34 subunit, the same mutant protein, however, was not phosphorylated under DNA replication conditions (53). Thus, the phosphorylation of this mutant protein may be an in vitro artifact.

HSSB is an essential protein in DNA replication, nucleotide excision repair, transcription, DNA recombination, and double-stranded DNA break repair. The biological significance of the cell cycle-dependent phosphorylation of HSSB-p34 is still unclear. One possibility is that HSSB phosphorylation may act as a signal for ongoing DNA replication to be sensed by other cell cycle check point proteins. This would ensure that premature cell division does not occur before DNA replication is complete. Alternatively, HSSB-p34 phosphorylation may have a more direct role in changing the interaction between HSSB and other essential DNA replication proteins contributing to the regulation of replication.

In UV-irradiated cells, enhanced phosphorylation of HSSB-p34 is associated with the cessation of cell cycle progression and a decrease in replication activity (32, 33). This suggests that the phosphorylation of HSSB-p34 may also play a role in coordinating the repair of damaged DNA with normal DNA synthesis and cell cycle progression.

Currently the construction of HSSB containing mutations in both the DNA-PK and the Cdc2 kinase phosphorylation sites is underway. Biochemical characterization of this mutant protein may allow us to determine directly the role of the phosphorylation of HSSB-p34 in the cell.