The Phosphorylation Domain of the 32-kDa Subunit of Replication Protein A (RPA) Modulates RPA-DNA Interactions EVIDENCE FOR AN INTERSUBUNIT INTERACTION*

Replication protein A (RPA) is a heterotrimeric (sub-units of 70, 32, and 14 kDa) single-stranded DNA-binding protein that is required for DNA replication, recombination, and repair. The 40-residue N-terminal domain of the 32-kDa subunit of RPA (RPA32) becomes phosphorylated during S-phase and after DNA damage. Recently it has been shown that phosphorylation or the addition of negative charges to this N-terminal phosphorylation domain modulates RPA-protein interactions and in-creases cell sensitivity to DNA damage. We found that addition of multiple negative charges to the N-terminal phosphorylation domain also caused a significant decrease in the ability of a mutant form of RPA to destabilize double-stranded (ds) DNA. Kinetic studies suggested that the addition of negative charges to the N-terminal phosphorylation domain caused defects in both complex formation (nucleation) and subsequent destabilization of dsDNA by RPA. We conclude that the N-terminal phosphorylation domain modulates RPA interactions with dsDNA. Similar changes in DNA interactions were observed with a mutant form of RPA in which the N-terminal domain of the 70-kDa subunit was de-leted. This suggested a functional link between the N-terminal domains of the 70- and 32-kDa subunits of RPA. NMR experiments provided evidence for a direct interaction following the manufac- turer s specifications and annealed to equal molar concentrations of the complementary sequence (SV40 5 (cid:4) -TCTGAGCTATTCCA- GAAGTAGTGAGGAGGCTTTTTTGGAG-3 (cid:4) ) to make a 40-nt double-stranded DNA substrate for helix destabilization assays. A double- stranded 40-nt substrate containing an eight-nt bubble was made by annealing SV40 top to SV40 bottomB (5 (cid:4) -TCTGAGCTATTCCAGAGAC- GACAGGGAGGCTTTTTTGGAG-3 (cid:4) ). The underlined portion indicates a region of non-complementary bases. Annealing reactions (10 m M Tris-HCl, pH 7.5, 20 m M MgCl 2 , and 50 m M NaCl) were placed in a PCR machine at 95 ° C for 3 min ramped to 22 ° C by a rate of 0.01 ° C/s with a hold temperature of 4 ° C. Annealing was monitored by 15% polyacrylamide gel electrophoresis (1 (cid:3) TBE). In all experiments, greater than 95% of labeled DNA was in double-stranded form.

RPA is composed of three subunits with multiple structurally related, functional domains (Fig. 1). The primary ssDNA binding activity is localized to the central region of the 70-kDa subunit (RPA70) (10). This region is composed of two structural domains called DNA binding domains (DBDs) A and B (Fig. 1A) (11,12). Both domains form an oligosaccharide/oligonucleotide binding fold (OB-fold) that contains five ␤-strands arranged into a Greek-key ␤-barrel capped by an ␣-helix between the third and fourth strands (13). The N-and C-terminal domains of RPA70 are also composed of OB-folds. Structural studies by NMR have shown that the N-terminal 168 residues of RPA70 form two distinct domains (14). The first 108 residues form an OB-fold, defined as DBD F based on structural homology and DNA binding ability (15). The remaining 60 residues are a flexible linker to DNA binding domains A and B in RPA70. In addition to binding DNA, DBD F has also been shown to interact with various proteins involved in DNA metabolism (1). The C-terminal region of RPA70, DBD C, exhibits weak DNA binding and is necessary for heterotrimeric complex formation (16,17). The OB-fold of DBD C has two unique characteristics, a three helical cap and a zinc ribbon motif (18). Two additional OB-fold domains are located in the 32-and 14-kDa subunits. The 32-kDa subunit has a central OB-fold domain, DBD D, that has weak DNA binding, flanked by the N-terminal phosphorylation domain and a C-terminal winged helix domain (19). The latter has been shown to interact with proteins involved in DNA repair (19). The 14-kDa subunit (RPA14) is also composed of an OB-fold that has been defined as DBD E although there is currently no evidence that this domain interacts with DNA. RPA14 is necessary for RPA complex formation (1,2).
In addition to binding ssDNA and proteins involved in DNA metabolism RPA can destabilize double-stranded DNA (dsDNA) (20,21). Although RPA binds dsDNA with low affinity, it is able to promote helix destabilization by stabilizing single-strand regions in dsDNA (20,22). RPA destabilization of duplex DNA is functionally distinct from that of helicases, because ATP and Mg 2ϩ are not required (20). Linearized SV40 dsDNA is preferentially denatured by RPA at internal regions rich in A-T pairs and at both ends (21). Although other ssDNAbinding proteins are capable of destabilizing dsDNA, RPA is more efficient than the ssDNA-binding proteins tested: Escherichia coli SSB, herpes simplex virus ICP8 protein, bacteriophage T4 gp32, and bacteriophage T7 gp2.5 (20,21,23). Association and helix destabilization reactions of RPA with duplex DNA and duplex DNA containing a central eight-nt ss bubble indicated that the denaturation of dsDNA occurs in at least two steps (16,23). The two steps are (i) the formation of stable RPA/ssDNA with a small region of ssDNA (nucleation), followed by (ii) efficient strand separation. The destabilizing activity of RPA may play a role in the initiation of replication, as well as the denaturation of damaged DNA, in nucleotide excision repair (16).
Helix destabilization by RPA requires the core ssDNA binding region (DBDs A and B) and is stimulated by the presence of DBD F (22). DBD F contains a basic cleft in the ␤-sheet-rich OB-fold (Fig. 2) and interacts with the acidic domains of the transcriptional activators Gal-4, VP16, and p53 (24,25). NMR data indicate that residues within the basic cleft of DBD F interact with ssDNA (15). A direct competition between ssDNA and the acidic activation domains of Gal-4, p53, and VP16 has been proposed to regulate DBD F function (15).
RPA is phosphorylated in a cell cycle-dependent manner (during S and G 2 ) (26,27) and in response to DNA damage (28,29). Hyperphosphorylation occurs in response to DNA-damaging agents, such as UV or ionizing radiation (30, 31), hydroxyu-rea or camptothecin treatment (31,32), and cellular apoptosis (33). Phosphorylation occurs primarily within the N-terminal 33 residues of RPA32 (28,34). Proteolytic and NMR analyses indicated that this domain, termed the N-terminal phosphorylation domain, exists in an extended, flexible conformation (19,35). Phosphorylation or mutations that add multiple negative charges to the N-terminal phosphorylation domain of RPA32 cause altered interactions with p53, T antigen, and DNA polymerase ␣/DNA primase (37). 2 Thus the N-terminal phosphorylation modulates RPA activity.
The effect of phosphorylation on RPA interactions with DNA is not well understood. Independent studies have reported that RPA phosphorylation causes stimulation of RPA helix destabilization activity (38), dramatic decreases in DNA binding (39), and dissociation of the RPA complex (40). Although these reports have not been reproduced in subsequent studies (41,42,47), 3 it seems likely that phosphorylation affects DNA interactions. To test this hypothesis, we examined the interactions between ssDNA or dsDNA and mutant forms of RPA with modifications in the N-terminal phosphorylation domain of RPA32. We found that whereas all forms of RPA examined had equal affinity for short oligonucleotides, a mutant form containing serine/threonine to aspartic acid mutations in the Nterminal phosphorylation domain of RPA32 had a significant defect in helix destabilization activity. A similar defect was observed with a mutant form of RPA that had the N-terminal domain of RPA70 (DBD F) deleted, suggesting a functional linkage between the N-terminal domains of RPA70 and RPA32. Helix destabilization assays with different substrates and time course experiments demonstrated that the observed decreases in helix destabilization activity were caused by deficiencies in both the nucleation and strand separation steps. Finally, we present evidence of a direct interaction between a negatively charged N-terminal phosphorylation domain of RPA32 and the basic cleft of DBD F of RPA70 using heteronuclear single quan-  (29) are indicated. The activities of each of the mutants in relation to wild-type RPA determined previously (22) and in these studies are shown to the right. ssDNA binding activity was determined with (dT)30: ϩϩϩ, wild-type binding; ϩϩ, binding reduced by an order of magnitude. dsDNA unwinding activity is shown as follows: ϩϩϩ, wild-type unwinding; ϩϩ, unwinding reduced by an order of magnitude; ϩ, unwinding reduced by two or more orders of magnitude. The following RPA mutants were used in these studies tum correlation NMR (HSQC). These findings suggest that there is a conformational change in RPA upon phosphorylation involving an intersubunit interaction between DBD F in RPA70 and the N-terminal phosphorylation domain of RPA32. This conformational change appears to regulate RPA function through modulation of RPA interactions with both proteins and DNA.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP (6000 Ci/mmol) was purchased from Amersham Biosciences. The 35-residue peptides, RPA32-WT and RPA32-Asp, corresponding to the wild-type or aspartic acid-substituted Nterminal phosphorylation domain of RPA32, respectively (Fig. 1), were from the Howard Hughes Medical Institute Biopolymer/Keck Foundation Biotechnology Resource Laboratory at Yale University. Biotinylated dT30 for surface plasmon resonance was purchased through the University of Iowa DNA Core (Sigma Genosys).
DNA Templates and Manipulation-A 40-bp dsDNA fragment of the early palindromic SV40 origin was made for helix destabilizing studies by labeling an oligonucleotide (SV40 top, 5Ј-CTCCAAAAAAGCCTCCT-CACTACTTCTGGAATAGCTCAGA-3Ј) with [␥-32 P]ATP by T4 polynucleotide kinase (New England Biosystems) following the manufacturer's recommendations. The labeled DNA was separated from free ATP with a P-30 Tris chromatography column (Bio-Rad) following the manufacturer's specifications and annealed to equal molar concentrations of the complementary sequence (SV40 bottom, 5Ј-TCTGAGCTATTCCA-GAAGTAGTGAGGAGGCTTTTTTGGAG-3Ј) to make a 40-nt doublestranded DNA substrate for helix destabilization assays. A doublestranded 40-nt substrate containing an eight-nt bubble was made by annealing SV40 top to SV40 bottomB (5Ј-TCTGAGCTATTCCAGAGAC-GACAGGGAGGCTTTTTTGGAG-3Ј). The underlined portion indicates a region of non-complementary bases. Annealing reactions (10 mM Tris-HCl, pH 7.5, 20 mM MgCl 2 , and 50 mM NaCl) were placed in a PCR machine at 95°C for 3 min ramped to 22°C by a rate of 0.01°C/s with a hold temperature of 4°C. Annealing was monitored by 15% polyacrylamide gel electrophoresis (1ϫ TBE). In all experiments, greater than 95% of labeled DNA was in double-stranded form.
Helix Destabilization Assay-The proteins were dialyzed against HI-30 (30 mM KCl). The extent of dialysis was monitored by solution conductivity. Helix destabilization assays were carried out as described previously (22). Briefly, 15-l reactions with HI buffer containing 30 mM KCl, 2 fmol radiolabeled DNA, increasing amounts of RPA (0 -3160 fmol), and 50 g/ml bovine serum albumin were incubated for 20 min at 25°C. Reactions were terminated by adding SDS to a final concentration of 0.2% (to disrupt RPA⅐DNA complexes), followed by 4% glycerol and 0.01% bromphenol blue. The reaction products were separated on a 15% polyacrylamide gel (1ϫ TBE) at 200 volts for 2.5 h. The gels were dried on Whatman 3-mm chromatography paper, and the radioactive bands were visualized by autoradiography. The radioactivity in each band was quantified using a Packard Instant Imager. The amount of dsDNA (%) was plotted against the concentration of RPA and then analyzed by non-linear least squares fitting to a Langmuir binding equation using Nonlin (22,45). Although the destabilization of DNA by RPA is not a simple bimolecular binding reaction, the Langmuir binding equation was used to precisely determine the midpoint of the transition between dsDNA and ssDNA. The midpoint (given with units of M Ϫ1 ) of the fitted curves was used as a value for comparing the activities of RPA forms in this assay.
Time Course of Helix Destabilization-Saturating amounts of RPA (562 fmol for wild-type RPA, RPA⅐32Ala10, and RPA⅐32⌬33 and 3160 fmol for RPA⅐32Asp8, RPA⅐70⌬N168, and RPA70⌬C442) were combined with HI-30 to create a 120-l reaction mixture. The DNA mixture consisted of 60 l containing 50 g/ml bovine serum albumin and 120 fmol of radiolabeled double-stranded 40-nt substrate or a doublestranded 40-nt substrate with an eight-nt bubble. The reaction mixture was incubated with the DNA mixture. Samples (15 l) were removed at the indicated time points and immediately placed in microcentrifuge tubes with SDS (0.2% final concentration) to dissociate RPA⅐DNA complexes. The zero-min time point was made by adding the saturating amount of RPA to a mixture containing 2 fmol of DNA, 50 g/ml bovine serum albumin, and 0.2% SDS in HI-30. The products were separated on a 15% polyacrylamide gel (1ϫ TBE). The amount of ssDNA at each time point was quantitated by the Packard Instant Imager and plotted as a function of time.
RPA70⌬169 -616 Interactions with Peptides-The NMR data collection was performed as described previously (14). Lyophilized RPA32-WT peptide and RPA32-Asp peptide were dissolved in the same buffer as [ 15 N]RPA70⌬169 -616. Two-dimensional 1 H-15 N correlation maps were acquired on two identical [ 15 N]RPA70⌬169 -616 samples after addition stoichiometric and superstoichiometric amounts of either RPA32-WT or RPA32-Asp peptide to each RPA sample. Final spectra contained nearly a 20-fold excess of peptide relative to protein, and spectral changes were minimal in final spectra with RPA32-Asp, indicating an approach to saturation of peptide binding. Spectra were acquired at 750 MHz on a Varian Inova console, using water flip-back HSQC. We used a spectral width of 3000 Hz, 128 increments in the indirect dimension. Temperature was 25°C.
Surface Plasmon Resonance-Interaction of RPA with ssDNA was monitored using a surface plasmon resonance biosensor instrument, BIAcore 3000. The streptavidin biosensor surface was prepared by manually injecting 5Ј-biotinylated dT30 diluted to 0.5 nM in 10 mM sodium acetate, pH 4.8, and 1.0 M NaCl into the desired flow cell. Proteins were diluted in HBS-EP buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, and 0.005% polysorbate-20) from BIAcore brought to 1 mM dithiothreitol. Protein (20 l of 4 nM solution) was loaded by the kinject option with the dissociation time of 500 s and a flow rate of 10 l/min. Each experiment was repeated at least twice. Data were analyzed with the BIA Evaluation program and fit to a bimolecular Langmuir binding curve.

Role of N-terminal Phosphorylation Domain in RPA-DNA
Interactions-RPA is composed of a high affinity DNA binding core (DBDs A and B) and three additional domains that can interact with DNA (DBDs F, C, and D; see Fig. 1). Mutations that decrease ssDNA binding also show defects in doublestranded DNA binding and helix destabilization activities (see Table I and Ref. 22). There have been several reports of phosphorylation modulating DNA interactions, but no consistent models have emerged for how RPA phosphorylation affects DNA interactions (see the Introduction). Our initial studies with RPA phosphorylated with CDK2 family kinases or DNAdependent protein kinase (DNA-PK) suggested that phosphorylation had minimal effects on RPA binding to single-stranded oligonucleotides. 4 To extend these observations we compared the ssDNA binding activity of various mutant forms of RPA including N-terminal phosphorylation domain mutants (Fig.  1). We have shown that addition of negative charges to the N-terminal phosphorylation domain of RPA32 modulates RPAprotein interactions and that a mutant form, which has aspartic acid substitutions in the N-terminal phosphorylation domain of RPA32 (RPA⅐32Asp), has properties similar to those of phosphorylated RPA. 2 Binding of the mutant forms of RPA to a 30-residue oligonucleotide was determined by surface plasmon resonance. The apparent binding constants for the phosphorylation domain mutants were the same as the wild-type RPA complex (Table I) effect on the binding constant with oligo(dT)30. This is consistent with previous deletion analysis demonstrating that the core DNA binding domain of RPA70 (DBDs A and B) is necessary and sufficient for high affinity binding to oligonucleotides (10,22).
Single-stranded oligonucleotides are the most simple form of DNA to which RPA can bind. In the cell most single-stranded DNA is part of partially duplex structures. Therefore, we next examined the ability of the phosphorylation domain mutants to destabilize a short DNA duplex. In this assay increasing concentrations of RPA are incubated with radiolabeled duplex DNA, and the relative amount of dsDNA and ssDNA is determined at each protein concentration. A radiolabeled 40-nt oligonucleotide from the SV40 early palindromic region annealed to its complementary strand was used as the substrate. This region of DNA has been shown to require only T antigen and RPA for denaturation during the early steps of SV40 initiation (46). Protein⅐DNA complexes were dissociated with SDS, and the products were separated on a 15% polyacrylamide gel. The amount of ssDNA and dsDNA in each reaction was quantitated, and the midpoint of the transition was determined. Deletion of the N-terminal phosphorylation domain (RPA⅐32⌬33) or introduction of multiple alanine mutations (RPA⅐32Ala10) had no effect on helix destabilization (Fig. 3A). In contrast, introduction of multiple aspartic acidic residues in the N-terminal phosphorylation domain caused a 4-fold increase in the amount of protein needed to destabilized 50% of the DNA template (Fig. 3A). This decrease in helix destabilization activity is unusual. Other forms of RPA have been shown to have decreased helix destabilization activity. For example, forms of RPA with mutations or deletions in the core DNA binding domain (DBDs A and B) or in the conserved zinc finger domain (DBD C) have decreased oligonucleotide binding and helix destabilization activity (22). However, in all but one case, these mutants also had decreased affinity for oligonucleotides (Table  I; see also Ref. 22). The one exception to this general rule is RPA⅐70⌬N168, which has wild-type affinity for oligonucleotides but also shows an activity in helix destabilization onequarter that of wild-type RPA (see Table I and Fig. 3A). We conclude that both RPA⅐32Asp8 and RPA⅐70⌬N168 have altered interactions with double-stranded or partially duplex DNA substrates that are not caused by differences in ssDNA binding activity.
Helix-destabilizing Activities of RPA with an Eight-nt Bubble Substrate-Helix destabilization by RPA appears to be a multistep process in which there is first the formation of an RPA⅐DNA complex at a small region of ssDNA (nucleation) followed by efficient strand separation coupled to binding of additional RPA molecules. A pseudo-origin substrate with an eight-nt bubble in the center has been used to investigate the steps that are effected in these RPA mutants (46). The ratelimiting step is hypothesized to be the generation of a small region of ssDNA, because RPA binds rapidly and with high affinity to ssDNA. The decreased helix destabilization of RPA⅐32Asp8 and RPA⅐70⌬N168 could be because of deficiencies in either nucleation or subsequent strand separation steps. If the decreased helix destabilization activity of RPA⅐32Asp8 and RPA⅐70⌬N168 could be rescued by a substrate with an eight-nt bubble, it would suggest a defect in the nucleation step.
When helix destabilization was examined using an eight-nt  bubble substrate, two levels of helix destabilization activity were observed. Both RPA⅐32⌬33 and RPA⅐32Ala10 had activity similar to wild-type RPA, although RPA⅐32Ala10 consistently had a slightly lower activity (Fig. 3B). In contrast, RPA⅐32Asp8 and RPA⅐70⌬N168 had one-half the destabilization activity with the eight-nt bubble substrate compared with wild-type RPA (Fig. 3B). Thus, the presence of an eight-nt bubble partially restored helix destabilization activity observed for RPA⅐32Asp8 and RPA⅐70⌬N168 with dsDNA. This indicates that these forms of RPA have defects in both nucleation and strand separations steps of unwinding.
Kinetics of Helix Destabilization-To investigate the kinetics of the nucleation event, a time course of helix destabilization was examined. Saturating amounts of RPA were incubated under helix destabilization conditions, and 15-l samples were removed at the indicated time points. The amount of ssDNA present at each time point was quantitated and plotted as a function of time.
Wild-type RPA, RPA⅐32Ala10, and RPA⅐32⌬33 all completed destabilization of both DNA substrates within 5 min (Fig. 4, A  and B). The RPA⅐32Asp8 mutant took 15 min longer to complete destabilization of the dsDNA substrate than wild-type RPA (Fig. 4A) but was still able to reach the same final level of destabilized DNA. In contrast, RPA⅐32Asp8 required only 1 min for destabilization of the eight-nt bubble substrate. This indicates that RPA⅐32Asp8 has a kinetic defect that affects the generation or recognition of a small region of ssDNA. RPA⅐70⌬N168 shows similar activity to RPA⅐32Asp8 helix destabilization is slower than wild-type RPA, and the time required to destabilize the eight-nt bubble substrate (10 min) is much faster than with the dsDNA substrate (60 min) (data not shown). We conclude that RPA⅐32Asp8 and RPA⅐70⌬N168 mutants have defects in both the nucleation step and in the helix destabilization steps of unwinding. These studies cannot resolve fast kinetics in the Ͻ10 s time scale.
A striking result of these studies is that removing DBD F (RPA⅐70⌬N168) or the addition of negative charges to the phosphorylation domain of RPA32 (RPA⅐32Asp8) caused similar effects on RPA interactions with complex DNA substrates. Furthermore, the absence of an altered affinity to oligonucleotides and altered kinetics of helix destabilization in these mutants indicated that these mutations are acting through a mechanism distinct from those that affect ssDNA binding. These data suggest a link between DBD F and the N-terminal phosphorylation domain of RPA32. We hypothesized that an interaction between the negatively charged phosphorylation domain and the basic cleft in DBD F may exist that prevents the N-terminal region of RPA70 from contacting DNA and contributing to destabilization (Fig. 5).
Residue-specific Interactions Monitored by HSQC NMR-To test the intersubunit interaction hypothesis, peptides corresponding to the non-phosphorylated (RPA32-WT) and the negatively charged (RPA32-Asp) phosphorylation domain (Fig. 1) were individually incubated with 15 N-labeled RPA70⌬169 -616 (DBD F; the first 168 residues of RPA70). HSQC NMR was used to monitor changes in the backbone of RPA70⌬169 -616 at five different concentrations of peptide. The final difference in 1 H and 15 N chemical shift upon addition of RPA32-WT or RPA32-Asp peptide was determined and plotted against residue number (Fig. 6, A and B). Only the addition of RPA32-Asp peptide caused a significant shift in either the backbone nitro-gen or the proton spectra of RPA70⌬169 -616. A summary of the residues perturbed by the RPA32-Asp peptide is shown in Fig. 6C. The residues perturbed by the RPA32-Asp peptide correlated well with the area of the basic cleft and residues that interact with ssDNA (15). We conclude that the addition of negative charges to the phosphorylation domain (RPA⅐32Asp8) can cause it to interact with DBD F. This suggests that phosphorylation can cause a conformational change in RPA in which there are increased interactions between the N-terminal domains of RPA70 and RPA32. DISCUSSION Several lines of evidence now indicate that the phosphorylation domain of RPA32 modulates RPA activity. (i) RPA is phosphorylated in a cell cycle-dependent manner (26,27). (ii) Cell extracts exposed to UV radiation that contain phosphorylated RPA are deficient in SV40 replication. Replication activity is restored with the addition of purified non-phosphorylated RPA (30). (iii) RPA phosphorylation modulates RPA-protein interactions. The addition of negative charges either by mutation or phosphorylation of the phosphorylation domain causes decreased interactions with T antigen and DNA polymerase ␣. The negatively charged phosphorylation domain also enhances RPA-p53 interactions (37). 2 (iv) A form of RPA lacking the phosphorylation domain, RPA⅐32⌬33, was not able to support SV40 DNA replication after pretreatment with DNA-PK. Without pretreatment of DNA-PK, RPA⅐32⌬33 is able to support SV40 replication. This suggests that under some conditions the phosphorylation domain may be required to overcome inhibitory effects of DNA-PK on components of the replication machinery (34). (v) Saccharomyces cerevisiae with the serine/threonine to aspartic acid mutations in the phosphorylation domain as the only functional copy of RPA32 is hypersensitive to methyl methane sulfonate and hydroxyurea treatment indicating that DNA damage is not being repaired. 2  There is one report in the literature that cdc2-phosphorylated calf thymus RPA stimulates helix destabilization (38). These findings are not consistent with those presented here or recent studies by Oakley et al. (37). They are also not consistent with a decrease in helix destabilization activity observed with RPA phosphorylated by DNA-PK (47). The disparity remains unresolved but could be because of differences in phosphorylation sites or differences in assay conditions.
Helix Destabilization-RPA efficiently destabilizes dsDNA (20,21). The destabilizing reaction has been functionally separated into two steps. The first step, termed nucleation, is the generation of a small region of ssDNA, either by DNA breathing or RPA association, culminating with the stable binding of RPA. The second step is efficient strand separation. As the first step of helix destabilization includes the stable binding of RPA to a small region of ssDNA, RPA mutants that effect ssDNA binding also effect destabilization (Table I) (22); there is a direct correlation between ssDNA binding and helix destabilization. There are two exceptions to this generalization. RPA⅐70⌬N168 has wild-type ssDNA binding activity but decreased dsDNA binding and helix destabilization (22). In addition, the phosphorylation domain mutant, RPA⅐32Asp8, also showed wild-type ssDNA binding but decreased helix destabilization activity ( Table I). The finding that separate mutations affecting either the basic cleft of DBD F (RPA⅐70⌬N168) or the RPA32 phosphorylation domain (RPA⅐32Asp8) do not affect ssDNA binding but do have similar effects on helix destabilization suggests a link between these two domains. We hypothesized a direct interaction between the phosphorylation domain and the basic cleft. This interaction may prevent the N-terminal region of RPA70 from contacting the DNA and contributing to destabilization.
A direct interaction between the basic cleft of RPA70 and the negatively charged phosphorylation domain was observed by HSQC NMR experiments. A peptide corresponding to the neg-atively charged (RPA32-Asp) phosphorylation domain was found to interact with DBD F whereas a peptide corresponding to the native, unphosphorylated sequence did not (Fig. 6). Furthermore, the residues perturbed by the RPA32-Asp peptide were predominantly in the basic cleft and correlated well with residues that interact with ssDNA (15). These data support the hypothesis that the negatively charged phosphorylation domain of RPA32 interacts specifically with DBD F in RPA70.
Model for RPA Intersubunit Interaction-We have presented evidence for a specific intersubunit interaction between a negatively charged phosphorylation domain of RPA32 and the basic cleft in DBD F. The negative charges can be a result of phosphorylation or aspartic acid mutations as in RPA⅐32Asp8. Increased interactions between the phosphorylation domain and DBD F are likely to cause a global conformational change in RPA. These intersubunit interactions also probably keep DBD F from interacting with DNA and contributing to dsDNA destabilization (Fig. 4). This model provides a mechanism by which the phosphorylation domain can modulate RPA activity. This conformational switch is consistent with the finding that the phosphorylation domain affects RPA-protein interactions indirectly by regulating the contacts of DBD F with T antigen and DNA polymerase ␣ (37). 2 The proposed interactions are also consistent with recent evidence that DBD F rotates independently of DBD allowing domain F to switch between conformations (15).
The Basic Cleft and Phosphorylation Domain-There are six residues, five arginines and one lysine, that form the basic cleft in DBD F of human RPA (Fig. 2). The charge of the basic cleft in human DBD F has been found to be conserved in other RPA homologues, with the exception of two eukaryotes that are missing the entire DBD F (Crithidia fasciculata (trypanosome), 23.9% identity; Pyrococcus furiosus (archeae bacteria), 11.8% identity). The basic cleft is also not conserved in Cryptospordium parvum, a protozoal parasite. We conclude that the basic cleft is evolutionarily conserved and present in all metazoans and yeast that have been examined.
Several mutations within DBD F have been characterized in yeast. Many of those with UV and methyl methane sulfonate hypersensitivity phenotypes map to regions within the basic cleft (48). This sensitivity to methyl methane sulfonate and UV is similar to the phenotype observed in RPA32 mutants with negatively charged N-terminal phosphorylation domains of RPA32. 2 Decreased rates of recombination, double-strand break repair and HO-gene conversion are also associated with mutations near the basic cleft (48 -51). In addition, some DBD F mutations are temperature-sensitive (49) and have defective DNA damage checkpoints (52,53). Recently, S. cerevisiae RPA trimer containing the rfa1 mutant t11 (K45Q) has been expressed and found to have similar ssDNA binding affinity to wild-type S. cerevisiae RPA but impaired Rad51 displacement from ssDNA (54). The genetic data demonstrates that DBD F is playing a role in DNA metabolism, particularly DNA repair. The studies presented here show that mutations that alter the structure and charge distribution of the basic cleft or the addition of negative charges to the phosphorylation domain have similar effects on RPA interactions with dsDNA. These mutations may prevent the basic cleft from participating in RPAprotein or RPA-DNA interactions that are needed for an appropriate cellular response to DNA damage.
The serines and threonines in the phosphorylation domain of RPA32 are conserved to varying degrees in eukaryotes. The total number ranges from eight to 12 for most eukaryotes examined (data not shown). The two RPA homologues that have less than eight serine and threonine residues have the remaining serines and threonines in conserved sites. The major exception to the high level of conservation is Schizosaccharomyces pombe, which has only three serine/threonines at nonconserved locations in its N-terminal phosphorylation domain. This suggests that S. pombe uses a different mechanism of regulation not involving a basic cleft. A homolog of RPA32 called RPA4 was found to be expressed in quiescent cells (55). RPA4 lacks four serines (S12I, S23L, S29T, and S33K) and one threonine (T21D) found in RPA32. The absence of the serines in RPA4 may reflect the proliferation status of the cell types it is expressed in or that it is part of complexes that are regulated differently.
Implications for RPA32 Phosphorylation-The effect RPA phosphorylation has on DNA repair pathways is not known. Phosphorylated and non-phosphorylated forms of RPA are equally active in in vitro SV40 DNA replication and nucleotide excision repair assays (56). 2 However, deletion of the phosphorylation domain and part or all of DBD F decreases the ability of RPA to support nucleotide excision repair (57). The phosphorylation state of RPA has no effect on RPA-xeroderma pigmentosum complementation group A (XPA) interactions (57). 5 XPA has been shown to inhibit RPA-dependent separation of a 24mer from M13 (58). This may be an example of a coordinated effort of phosphorylated RPA and XPA to limit the opening of DNA around the lesion site. It remains possible that phosphorylation of RPA may modulate RPA function in nucleotide excision repair in vivo. In addition, the effect of RPA phosphorylation on recombination and base excision repair remains to be determined. We have shown that negative charges in the N-terminal phosphorylation domain can modulate RPA-DNA interactions and present data that these negative charges promote an intersubunit interaction with the basic cleft of the N-terminal region of RPA70. This conformational change can modulate RPA70-mediated interactions with DNA and proteins and cause modulation of RPA activity.