Functional characterization of a cancer causing mutation in human Replication Protein A

Abstract

Replication protein A (RPA) is the primary single-stranded DNA-binding protein in eukaryotes. RPA is essential for DNA replication, repair and recombination. Mutation of a conserved leucine residue to proline in the high affinity DNA binding site of RPA (residue L221 in human RPA) has been shown to have defects in DNA repair and a high rate of chromosomal rearrangements in yeast. The homologous mutation in mice was found to be homozygous lethal and cause high rates of cancer when heterozygous (Wang et al. (2005) Nat Genet 37, 750). To understand the molecular defect causing these phenotypes, we created the homologous mutation in the human RPA1 gene (L221P) and analyzed its properties in cells and in vitro. RPA1(L221P) does not support cell cycle progression when it is the only form of RPA1 in HeLa cells. This phenotype is caused by defects in DNA replication and repair. No phenotype is observed when cells contain both wild-type and L221P forms of RPA1 indicating that L221P is not dominant. Recombinant L221P polypeptide forms a stable complex with the other subunits of RPA indicating that the mutation does not destabilize the protein; however, the resulting complex has dramatically reduced ssDNA binding activity and cannot support SV40 DNA replication in vitro. These findings indicate that in mammals the L221P mutation causes a defect in ssDNA binding and a non-functional protein complex. This suggests that haploinsufficiency of RPA causes an increase in the levels of DNA damage and in the incidence of cancer.

Introduction

Efficient maintenance of the genome is required for cell survival and proliferation. The cell must faithfully replicate billions of base pairs as well as identify and repair a variety of DNA lesions. Replication protein A (RPA), the major eukaryotic single strand DNA (ssDNA) binding protein, is required for cell viability and plays essential roles in DNA replication, repair, and recombination (1–3). RPA also plays a role in coordinating DNA metabolism with the cell cycle (4–6). RPA expression is up-regulated in tumors including breast and colon cancer and has been shown to play an important role in cell proliferation during cancer growth and progression (7, 8).

The RPA complex binds to ssDNA with high affinity and low sequence specificity (9–11) and interacts with proteins involved in DNA replication, repair, recombination and checkpoint pathways (3, 6, 12, 13). RPA is a stable three-subunit complex composed of RPA1, RPA2, and RPA3 (1–3). The genes for all three subunits are essential for cell viability and chromosome stability (14, 15); however, the largest subunit, RPA1, is responsible for high affinity ssDNA binding and many protein-protein interactions (3). RPA1 is composed of four structurally related DNA binding domains (DBD) (3). The N-terminal domain, DBD-F, contains many protein interaction sites involved in DNA repair, recombination, and cell cycle regulation (16). The C-terminal domain, DBD-C, is involved in complex formation with the other RPA subunits and also plays a role in recognizing DNA damage (17, 18). The central two domains, DBD-A and DBD-B, have the highest DNA affinity in the RPA1 subunit and constitute the ssDNA-binding core of the complex (3, 19–21).

Extensive genetic analysis of the yeast homolog of RPA1 (RFA1) has shown that viable mutations in this gene exhibit defects in DNA repair, recombination, and elevated chromosome rearrangements and mutation rates (15, 22). In mammalian cells, RPA is essential for life and depletion leads to an accumulation of cells in S phase due to deficient DNA replication and activation of DNA checkpoints due to persistence of DNA damage (5, 23).

One mutation in RPA1 that has been studied in both yeast and mammals is a leucine to proline substitution at position 221 in DBD-A of the human RPA1 subunit. The mutation is found within a hydrophobic patch at the base of an extended β-ribbon region in DBD-A near the RPA-DNA interface. The L221P mutation was first described in a yeast mutagenesis screen and exhibited sensitivity to DNA-damaging agents, defects in checkpoints and homologous recombination, and gross chromosomal rearrangements (22, 24). However, no evidence for a major defect in DNA replication was observed. Homozygous expression of the L221P mutant RPA1 in mice was shown to be embryonic lethal. The heterozygous mice were viable but had shortened lifespans and elevated cancer rates (25). The tumors were found to have widespread chromosomal rearrangements similar to those found in human cancers (25).

In order to understand the molecular defect causing chromosomal instability, the equivalent L221P mutation was made and characterized in human RPA1. A knockdown/reconstitution system in HeLa cells was utilized to assess its function in human cells. Cells containing only RPA1(L221P) exhibit abnormal cell cycle progression characterized by an accumulation in S and G2/M phases. Studies with synchronized cellular populations demonstrated that RPA1(L221P) is unable to support S phase progression indicating a replication defect. The mutant protein was unable to localize to damage-induced sites of DNA repair in cells suggesting it is also defective in DNA repair. Studies in cells expressing both wild-type and mutant forms of RPA1 indicated that the L221P form is not dominant. Biochemical analysis with recombinant RPA1(L221P) showed that the mutant subunit is able to associate with the other two subunits in a stable complex but is unable to bind ssDNA or support in vitro replication. These results indicate that the L221P mutation disrupts RPA ssDNA-binding function resulting in a non-functional complex. These studies indicate that haploinsufficiency of RPA caused by the expression of a defective form of RPA results in genomic instability.

Methods

Construction of RPA1(L221P)

For the cell culture studies, a previously constructed EGFP-tagged version of RPA1 (23) was modified using quick-change site-directed mutagenesis to mutate leucine 221 to proline. Primers used were: 5’-CCAGTTCTAGGGAGAAAGGCTTCCCTTCCCC-3’ and 5’-GGGGAAGGGAAGCCTTTCTCCCTAGAACTGG-3’.

RNAi knockdown and replacement of RPA1

Methodology for knockdown of endogenous RPA1 and expression of exogenous RPA1 was as described (23). HeLa cells (obtained from ATCC) grown in Dubellco’s modified Eagle’s medium (DMEM) with 10% calf serum at 37°C with 5% CO₂ were seeded in six-well tissue culture plates at 2 × 10⁵ cells per well. 200 pmol siRNA was added 24 hours after seeding plates to knockdown endogenous RPA1. Transfections were performed with 5µL of Lipofectamine 2000 (Invitrogen). At 24 hours after transfection of siRNA, cells were transfected with 250 ng of plasmids expressing GFP fusions of wild-type or mutant RPA1. The RPA1 siRNA target sequence was 5’-GGAAUUAUGUCGUAAGUCA-3’.

Flow cytometry analysis

Cells were collected at 96 hours post-transfection of siRNA, washed with PBS and fixed overnight in 70% methanol. The cells were rehydrated in PBS for 30 minutes and washed in PBS. For cell cycle analysis, 0.1 mg/mL propidium iodide was added to each sample. For analysis of ChK2 activation, cells were incubated in 1:100 p-ChK2 primary antibody (Cell Signaling) overnight, then in 1:100 PE secondary (Invitrogen) for 2 hours. Cells were examined on a FACScan II, and the data were analyzed using FlowJo software (TreeStar).

Immunofluoresence analysis

HeLa cells were seeded on coverslips in six-well tissue culture plates and subjected to RNAi knockdown and replacement of RPA1 as described above (23, 26). At 92 hours post-transfection of siRNA, 20µM camptothecin was added to each well. The cells were incubated for 4 hours at 37°C and 5% CO₂. Coverslips were washed twice in cold CSK buffer (10mM HEPES, 300 mM sucrose, 100mM NaCl, 3mM MgCl₂). Non-chromatin bound RPA was extracted with CSK/0.5% Triton X-100 for 5 min. Coverslips were fixed with 4% formaldehyde for 20 min, then washed three times with PBS. To detect RPA2 or phosphorylated H2AX, coverslips were incubated in blocking solution (5% calf serum, PBS) for 1 hour at room temp, then primary antibody for RPA2 (71-9A) or p-H2AX (Cell Signaling) at 1:500 overnight at 4°C. Coverslips were washed three times with PBS, then incubated in anti-rabbit Texas Red secondary antibody (Cell Signaling) at 1:800 for two hours. Coverslips were washed in PBS, incubated in DNA staining solution (1 µg/µL DAPI), washed again in PBS, and mounted to slides. Slides were examined with a Leica immunofluorescence microscope and images were collected with SPOT software (Diagnostic Instruments, Inc.). Adobe Photoshop was used to process and overlay images.

Purification of recombinant RPA complex

Wild type and L221P mutant RPA complexes were expressed in BL21(DE3) and purified as described in Binz et al (26).

SV40 replication and ssDNA binding reactions

Reactions were carried out as described previously (26).

Briefly 25 µL SV40 reactions contained 30 mM HEPES (pH 7.5), 7mM MgCl2, 40mM creatine phosphate, 2.5 µg creatine kinase, 4mM ATP, 0.2 mM each of CTP, GTP, and UTP, 0.1 mM each of dATP, dGTP, and dTTP, 0.05 mM a32-P-dCTP, and 50 ng pUC•HSO DNA template, and 6 µl RPA-depleted HeLa cytosolic extract. HeLa cell extract was depleted of RPA using 35–65% ammonium sulfate fractionation (27). 1.9 µg SV40 T-antigen (Chimex) and 400 ng of purified wild type or L221P mutant RPA were added as indicated. Reactions were incubated for 2 hours at 37°C. Reactions were quenched by addition of 0.1 M sodium pyrophosphate, precipitated with 10% trichloracetic acid and DNA filtered through glass microfiber filters. Amount of synthesized radiolabled DNA was quantified by scintillation counting.

ssDNA binding reactions contained 30mM HEPES, 100mM NaCl, 5mM MgCl2, 0.5% inositol, 1mM DTT, 2 fmol labeled (dT)₃₀, BSA (50ng/µL), and 0 to 316 fmol of mutant or wild type RPA. Reactions were incubated for 20 minutes at 25°C and then separated on a 1% agarose gel in 0.1× TAE buffer (4 mM Tris acetate and 0.2 mM EDTA). Position of free and bound DNA was quantified using a Packard Instant Imager and the fraction of free ssDNA was plotted against RPA concentration. The data was analyzed by nonlinear least squares fitting to the Langmuir binding equation using KaleidaGraph (Synergy) to calculate the apparent binding constant (11).

Results

L221 is highly conserved in evolution

The leucine at position 221 in DBD-A of human RPA1 is highly conserved. This residue is absolutely conserved in animals and some unicellular eukaryotes including yeast (Table 1). In RPA1 homologues where the leucine is not present, some plants and unicellular eukaryotes, the leucine is replaced with another hydrophobic residue. This conservation suggests that this residue is important for some aspect of RPA1 structure or function. L221 is located near the base of the β2 strand of the β-sheet that forms the DNA-binding surface of DBD-A (Figure 1A; (20, 28)). Although L221 does not appear to interact directly with DNA, the leucine side chain does extend toward the DNA in the crystal structure (Figure 1A; (28)) and NMR analysis indicated that the chemical environment of L221 changes upon DNA binding (20). Conversion of leucine to a conformationally-constrained proline residue is predicted to disrupt the β sheet which is the core of the DNA binding site in DBD-A (Figure 1B).

Table 1. Alignment of L221 residue of RPA1 subunit from various species.

Selected residues of DBD-A in RPA1 from various organisms are shown. L221 and the homologous residues are highlighted.

Species - Accession number	Start	Sequence	Stop
Homo sapiens NP_002936.1	183	KVVPIASLTPYQSKWTICARVTNKSQIRTWSNSR-GEGKLFSLELVD-ESG	231
Pan troglodytes XP_511254.2	237	KVVPIASLTPYQSKWTICARVTNKSQIRTWSNSR-GEGKLFSLELVD-ESG	285
Canis lupus familiaris XP_868487.1	181	KVVPIASLTPYQSKWTICARVTNKSQIRTWSNSR-GEGKLFSIELVD-ESG	229
Bos taurus NP_001068644.1	183	KVVPIASLTPYQSKWTICARVTNKSQIRTWSNSR-GEGKLFSIELVD-ESG	231
Mus musculus NP_080929.1	192	KVVPIASLTPYQSKWTICARVTNKSQIRTWSNSR-GEGKLFSLELVD-ESG	240
Rattus norvegicus XP_213389.4	183	KVVPIASLTPYQSKWTICARVTNKSQIRTWSNSR-GEGKLFSIELVD-ESG	231
Gallus gallus NP_001006221.1	180	KVVPIASLNPYQSKWTICARVTQKGQIRTWSNSR-GEGKLFSIELVD-ESG	228
Danio rerio NP_956105.2	170	KVVPIASLNPYQSKWTIRARVTNKSAIRTWSNSR-GDGKLFSMELVD-ESG	218
Drosophila melanogaster NP_524274.1	168	---PISSLSPYQNKWVIKARVTSKSGIRTWSNAR-GEGKLFSMDLMD-ESG	213
Anopheles gambiae XP_321709.3	179	---PISSLSPYQNKWVIKARVMSKSGIRTWSNAK-GEGKLFSMDVMD-ESG	224
Schizosaccharomyces pombe NP_595092.1	178	IIYPIEGLSPYQNKWTIRARVTNKSEVKHWHNQR-GEGKLFSVNLLD-ESG	226
Saccharomyces cerevisiae NP_009404.1	183	PIFAIEQLSPYQNVWTIKARVSYKGEIKTWHNQR-GDGKLFNVNFLD-TSG	231
Kluyveromyces lactis XP_451388.1	187	PIFAIEQISPYQNNWTIKARVSFKGDLKKWQNNR-GEGHILNVNLLD-SSG	235
Eremothecium gossypii NP_985540.1	261	PIFAIEQLSPYQNMWTIKARVSFKGDIKTWHNQR-GEGKLFNVNFLD-TSG	309
Magnaporthe grisea XP_368559.1	178	NIYPIESISPYQHKWTIKARVSQKSDIRTWHKAS-GEGKLFSVNLLD-ETG	226
Neurospora crassa XP_322908.1	173	TIYPIEGLSPFSHKWTIKARVTSKSDIKTWHKAS-GEGKLFSVNFLD-ESG	221
Arabidopsis thaliana NP_567576.2	227	KIIPVNALSPYSGRWTIKARVTNKAALKQYSNPR-GEGKVFNFDLLDADGG	276
Arabidopsis thaliana NP_199353.1	298	RINPIAALNPYQGRWTIKVRVTSKADLRRFNNPR-GEGKLFSFDLLDADGG	347
Oryza sativa NP_001054445.1	306	RIIPITALNPYQPKWTIKARVTAKSDIRHWSNAR-SSGTVFSFDLLDAQGG	355
Cryptosporidium parvum(short) AAD42062.1	1	MPIREVNVNRQTISIKGRIIQKSSLQIL---K-SGLRFFHLDIIDKDND	45
Cryptosporidium parvum(long) AAW71479.1	191	PVYPIKNITSYLHRWRIVGRVVSKSDIRKFSSSKTKEGKVFSFEICDADGS	240

Figure 1. Structures and modeling of wild-type and L221P.

(A) Structures of DBD-A from wild-type human RPA1 (blue) bound to DNA (purple; 1JMC (28)) and yeast RPA1 (yellow; 1YNX (39)). Position of L221 side chain is shown extending toward DNA from left side of binding cleft. (B) β1 and β2 strands of the DNA binding site shown for (left) structures of wild-type human and yeast and (right) modeled structures of human and yeast L221P. Shown from outside looking toward DNA centered on position 221. β1 and β2 labeled and discontinuity indicated (*) in human structure. Models determined using Modeller (40) from structures shown in (A)). DSSP was used for secondary structure assignment (41); structures rendered in PyMOL.

RPA1(L221P) is unable to rescue cell cycle progression

We initially wished to directly determine whether RPA1 containing the leucine to proline mutation at position 221 could function in human cells. We have previously established protocols for using siRNA knockdown/exogenous gene reconstitution to examine RPA1 function in cells (23). HeLa cells are treated with a siRNA targeted to the 3’UTR region of RPA1 that causes endogenous RPA1 levels to decrease to less than 5% of normal levels (23). 24 hours after transfection of siRNA, plasmid expressing different forms of RPA1 lacking the targeted 3’ UTR fused to GFP were introduced to the cells and the distribution of cells in the cell cycle was determined by flow cytometry analysis after propidium iodide (PI) staining of DNA. Optimal knock-down of endogenous RPA1 and expression of exogenous RPA1 occurred between 96–120 hours after siRNA transfection (data not shown; (23)). As has been observed previously, knockdown of endogenous RPA1 causes an accumulation of cells in S- and G2/M-phase (Figure 2A, compare left two panels; (5, 23)). This phenotype is caused by defects in DNA replication and repair. Expression of the exogenous form of RPA1 was monitored by green fluorescence and only GFP-expressing cells were selected for analysis in the reconstitution experiments. GFP-tagged wild-type RPA1 was able to rescue RPA1 depleted cells to give a normal cell cycle distribution with the majority of cells in the G1 phase and a moderate G2/M peak (Figure 2A, 3^rd panel). The GFP-RPA1 expression plasmid was mutated at position 221 to produce a leucine to proline substitution. The resulting GFP-RPA1(L221P) was also expressed in RPA1 depleted cells. In contrast to wild-type RPA1, expression of GFP-RPA1(L221P) resulted in S-phase and G2/M populations similar to the RPA1-depleted phenotype (Figure 2A, rightmost panel).

Figure 2. Effect of L221P on cell cycle progression.

(A) Cells transfected with RPA1 siRNA and GFP-tagged wild-type (WT) or L221P RPA1 vector where indicated as described in Methods. At 96 hours, DNA content was analyzed with flow cytometry. Top row shows GFP expression (with % of GFP-positive cells indicated in upper left corner); bottom row shows DNA content as a histogram. (B) and (C) show analysis of the combined data from the experiment described in (A) and three similar, independent experiments. (B) Average transfection efficiency (GFP-positive cells) from cells treated with RPA1 siRNA and transfected with GFP-tagged wild-type (WT) or L221P RPA1 with standard deviation between experiments shown (error bars). (C) Average GFP fluorescence intensity was determined for mock treated cells and all GFP-positive cells (e.g. upper box in (A)) for RPA1 siRNA treated (mock), GFP-wild-type (WT) or GFP-L221P transfected cells shown with standard deviation between experiments indicated (error bars). Because gain settings varied between experiments, values from each experiment were normalized to wild-type RPA1 (WT). (D) Cells were treated and analyzed as described in (A), at 96 hours after mock or siRNA-transfection, cells were synchronized with 5µg/ml aphidicolin for 24 hours, then released media. Flow cytometry was used to analyze DNA content at 0, 8, and 24 hours after release. For both (A) and (D) DNA content shown for GFP-negative cells for Mock samples and GFP-positive cells for WT and L221P.

To determine whether the two expression plasmids were behaving similarly in HeLa cells, we compared the transfection efficiency of the two plasmids in four independent experiments. The average transfection efficiency of GFP-RPA1(L221P) plasmid was slightly lower than with the wild-type GFP-RPA1 plasmid; however, this difference was not statistically significant (Student Test P=0.05; Figure 2B). The average green fluorescence per cell was also quantiated to determine the expression levels of the two forms of RPA1. There was significant variation in the protein level in individual cells (Figure 2A, top row) but overall the average fluorescence of cells expressing GFP-RPA1 and GFP-RPA1(L221P) were not significantly different (P=0.18; Figure 2C). It has been shown previously that the level of expression of GFP-RPA1 obtained under these conditions is comparable to the endogenous levels of RPA1 (23). This demonstrates that wild-type and L221P are expressed at similar levels in these experiments. We conclude that L221P did not complement the RPA1 depletion and that this mutant form is not able to support some essential function of the RPA complex.

Increased ChK2 phosphorylation in the presence of RPA1(L221P)

To determine whether the G2/M accumulation observed in cells expressing L221P was due to initiation of a checkpoint, we assessed the activation of the checkpoint effector kinase, checkpoint protein 2 (ChK2) (29). Cells expressing wild-type or L221P forms of RPA1 were stained with an antibody to phosphorylated ChK2 (p-Chk2) and analyzed by flow cytometry. Untreated cells contained undetectable levels of p-Chk2 (Figure 3A). Depletion of RPA1 resulted in two populations of cells: one with low p-ChK2 levels similar to untreated cells and a second population with high levels of p-ChK2 (Figure 3A). These two peaks probably correspond to cells that are in G1 and cells that have an activated checkpoint after attempting to progress through S phase in the absence of RPA1 triggering checkpoint activation, respectively. Expression of exogenous wild-type RPA1, restored low levels of p-ChK2; although, the staining level of p-ChK2 in reconstituted cells was slightly elevated over untreated cells (Figure 3A). Reconstitution with RPA1(L221P) resulted in all cells having high levels of p-ChK2 (Figure 3A). The elevated phosphorylation of ChK2 in cells expressing RPA1(L221P) indicated activation of a cell cycle checkpoint and was consistent with the altered cell cycle distribution observed in these cells (Figure 2A).

Figure 3. Cellular response to DNA damage.

(A) Activation of ChK2 in the absence of induced DNA damage. At 96 hours after mock or RPA1 siRNA-transfection (si1), cells were stained for phosphorylated ChK2 and analyzed via flow cytometry. (B) Localization of wild-type and L221P mutant RPA complexes to sites of DNA damage. Cells were grown on coverslips and treated with 20 µM camptothecin for 4 hours. Non-chromatin bound RPA was extracted prior to fixing slide. Row one-DAPI staining; Row two-GFP-tagged RPA1; Row three-RPA2 antibody staining; Row four-merge of rows two and three. (C) Ability to recognize sites of DNA damage. Cells treated as in (B). Row one-DAPI staining; Row two-GFP-tagged RPA1; Row three-stained with phosphorylated H2AX antibody; Row four-merge of rows two and three. (D) Coexpression of L221P mutant and endogenous wild-type RPA1. 48 hours after wells were seeded, the cells were mock-transfected or transfected with GFP-tagged wild-type or L221P mutant RPA1 vector. At 96 hours, DNA content of cells was stained with propidium iodide and analyzed via flow cytometry. GFP-negative cells were selected for Mock samples; GFP-positive cells were selected for WT and L221P samples.

RPA1(L221P) is unable to support progression through S phase

To directly examine cell cycle progression in cells expressing only RPA1(L221P), we analyzed a synchronized culture of RPA1(L221P) expressing cells. Cells were synchronized with aphidicolin, an inhibitor of DNA polymerases, for 24 hours and released to allow resumption of cell cycle progression. The cell cycle distribution was assessed by PI-staining and flow cytometry analysis at 0, 8, and 24 hours after release from aphidicolin treatment. At 0 hours the majority of the cellular population were in G1 phase (Figure 2D, top row). Mock-treated cells were able to resume progression into S-phase indicated by the rightward progression of the peak at 8 hours (Figure 2D, leftmost column). At 24 hours, a majority of the population had reached G2/M-phase and some had completed cell division and returned to G1-phase (Figure 2D, leftmost column). In cells lacking RPA1 only a small proportion of the population had begun to move into S-phase indicated by the slight thickening of the G1 peak at 8 hours while the majority of cells remain in the G1 peak even at 24 hours after release (Figure 2D, 2^nd column). Cells reconstituted with exogenous wild type RPA1 were able to progress into S-phase as indicated by the rightward progression of the peak at the 8 and 24 hour time points (Figure 2D, 3^rd column). Cells expressing only RPA1(L221P) did not progress into S-phase. The cell cycle distribution in the RPA1(L221P) cells was similar to the RPA1 depleted cells: minimal change in the G1 peak at 8 hours and a slight thickening of the G1 peak at 24 hours after release from aphidicolin (Figure 2D, 4^th column). These results indicate that cells expressing only RPA1(L221P) do not progress through S phase. Together with the results in unsynchronized cells, these data indicate the L221P mutation in RPA1 causes a disruption in the ability of cells to progress normally through cell cycle and suggests a defect in the ability to support cellular replication.

RPA1(L221P) is unable to localize to damage-induced foci

After DNA damage, wild-type RPA localizes to sites of DNA damage resulting in a punctuated staining pattern within the nucleus. These spots, termed DNA repair foci, contain proteins involved in DNA damage recognition and repair and are sites of DNA repair (30). We used fluorescent microscopy to assess whether RPA1(L221P) could localize to DNA repair foci after DNA damage. HeLa cells were treated with the topoisomerase inhibitor camptothecin for 4 hours to induce DNA damage. The non-chromatin bound RPA was extracted before fixing the cells. No GFP-staining was observed cells not transfected with GFP-tagged RPA1 (Figure 3B, 2^nd row left 2 images). An antibody to RPA2 showed that the endogenous RPA complex was localized to repair foci in the mock-treated cells (Figure 3B, 3^rd row left image). No RPA2 staining was observed for the cells treated with RPA1 siRNA confirming that the RPA complex is not able to localize to repair foci in the absence of RPA1 (Figure 3B, 3^rd row 2^nd column). Exogenous GFP-tagged wild-type RPA1 was able to form the punctuated staining pattern in the nucleus, indicating localization to damage-induced foci (Figure 3B, 2^nd row 3^rd column). RPA2 co-localized with the wild type RPA1 confirming localization of the entire RPA complex as expected (Figure 3B, 3^rd row 3^rd column). In contrast, neither GFP-RPA1(L221P) nor RPA2 foci were observed after DNA damage in cells reconstituted with RPA1(L221P) (Figure 3B, 2^nd and 3^rd row leftmost image). GFP-RPA1 was expressed in cells at similar levels to RPA1 based on GFP fluorescence in cells treated in parallel but not extracted prior to microscopy (data not shown). These results indicate that the L221P mutation disrupts ability of the RPA complex to localize to DNA repair foci, suggestive of a defect in DNA repair.

RPA1(L221P) causes elevated phosphorylation of H2AX

The previous experiments argue that the L221P mutation causes a defect in the RPA complex’s ability to function properly in DNA replication and repair. A lack of effective DNA replication and repair should be expected to cause an accumulation of DNA damage in cells. To test this hypothesis, we assessed the level of cellular DNA damage by measuring phosphorylation of the histone variant H2AX, a marker associated with DNA damage (31). Cells were stained with an antibody to phosphorylated H2AX (p-H2AX). The fluorescent microscopy was carried out as described above except that the extraction step was omitted to enable identification of cells expressing RPA1(L221P). No p-H2AX staining was detected in the mock-treated cells (Figure 3C, 3^rd row left image). A significant level of p-H2AX staining was evident in cells lacking RPA1 indicating elevated cellular DNA damage (Figure 3C, 3^rd row 2^nd column). Only background levels of p-H2AX were detected in cells reconstituted with wild type RPA1 (compare GFP expressing cell to non-transfected cells in Figure 3B, 3^rd column). Cells reconstituted with RPA1(L221P) exhibited significant p-H2AX staining indicating elevated cellular DNA damage (Figure 3B, 4^th column). We conclude that RPA1(L221P) is unable to function in DNA repair and results in increased accumulation of DNA damage.

RPA(L221P) is not dominant over wild type RPA

To investigate if the L221P mutation has a dominant phenotype, we assessed the effect of RPA1(L221P) expression in the presence of endogenous RPA in HeLa cells. Cells were mock-treated or transfected with wild type RPA1 or mutant RPA1(L221P) without siRNA treatment. Mock-treated cells showed a normal cell cycle distribution as expected (Figure 3D, left panel). Introduction of exogenous wild-type RPA1 did not cause a detectable change in the cell cycle distribution (Figure 3D, middle panel). Introduction of RPA1(L221P) also did not cause a disruption in the cell cycle distribution (Figure 3D, left panel). This indicates that RPA1(L221P) is not dominant over endogenous wild-type RPA in a short term assay. These results suggest that the abnormal cellular progression phenotype observed in Figure 2 was caused by the absence of functional RPA.

RPA1(L221P) forms a stable RPA complex

We next analyzed the biochemical properties of RPA1(L221P). RPA1(L221P) was expressed in combination with the RPA2 and RPA3 subunits in E. coli and purified as previously published (26). The RPA1(L221P) complex had identical chromatography properties and was purified with similar yields wild-type RPA (data not shown). The purified RPA1(L221P) complex contained all three subunit of RPA; although, the RPA1(L221P) polypeptide appeared to migrate slightly slower than the wild type RPA1 in SDS-PAGE (Figure 4A, compare lanes 1 and 2). The purified RPA1(L221P) containing complex, termed RPA(L221P), was biochemically characterized to gain insight into the mechanism of the replication and repair defects observed in the cells.

Figure 4. Biochemical properties of RPA-L221P.

(A) Gel analysis of purified L221P RPA complex. 1µg of wild type RPA or RPA containing the L221P RPA1 subunit were separated on an 8–14% SDS-PAGE gel and stained with silver nitrate. Lane 1-wild type RPA; lane 2-L221P complex. Non-RPA impurities indicated by *. (B) Ability to support SV40 DNA replication in vitro. Background synthesis was assessed in the absence of RPA (-RPA) or T antigen (-Tag). Synthesis of complete reactions containing 400ng either wild-type or L221P. A representative assay is shown; all reactions done in duplicate and normalized to background. (C) Single-stranded DNA binding. Binding isotherms of a representative gel mobility shift assay for wild-type and L221P complexes.

RPA1(L221P) is unable to support replication in vitro

SV40 virus replication depends on one viral protein, large T antigen (Tag), and multiple cellular proteins including RPA (32). Replication of SV40 can be reconstituted in vitro using Tag, RPA and HeLa cell extract depleted of RPA by ammonium sulfate fractionation (26). By adding different forms of RPA and monitoring DNA synthesis it is possible to directly test activity in DNA replication. The reactions lacking RPA or RPA and Tag showed background levels of synthesis (Figure 4B). Wild type RPA stimulated DNA synthesis but addition of RPA(L221P) had no effect on the total DNA synthesis (Figure 4B). These data confirm that RPA(L221P) is not able to support replication.

RPA1(L221P) is unable to bind ssDNA

To determine the molecular basis for the loss of activity in RPA(L221P), ssDNA-binding activity was quantitated by gel mobility shift assay. Wild-type RPA or RPA(L221P) were incubated with a radiolabeled ssDNA oligonucleotide and the resulting complexes separated using gel electrophoresis. The amount of free ssDNA and ssDNA-protein complex was quantitated. Figure 4C shows representative binding isotherms. Wild-type RPA bound with high affinity and an apparent association constant of ~10¹² M⁻¹. In contrast, RPA(L221P) did not form a stable complex with ssDNA at any protein concentration tested (Figure 4C). We estimated based on the highest concentration of RPA(L221P) used that these assays could have detected a binding affinity ≥10⁹ M⁻¹. These results indicate that the L221P mutation reduces the ability of the RPA complex to bind ssDNA by at least 1000-fold. The inability of RPA(L221P) to bind ssDNA explains the defect in DNA replication and repair. Together the results demonstrate that the RPA1(L221P) is a non-functional form of RPA1 leading to a disruption in DNA replication and repair function of the RPA complex due to its inability to bind ssDNA.

Discussion

The L221P mutation appears to primarily affect the biochemical activity of RPA. The L221P mutation does not appear to affect the stability of recombinant RPA or the level of the RPA1 polypeptide expressed exogenously in the cells. However, L221P causes a dramatic decrease in the ssDNA-binding activity of human RPA and results in a non-functional complex. Our studies demonstrate that L221 is important for normal DNA binding of RPA. These results provide a mechanism for why the homologous mutation in mice is homozygous lethal (25).

L221 is in DBD-A, one of two essential, high affinity DNA binding domains of RPA that are both necessary and sufficient for binding of the RPA complex to DNA (19, 20). The L221 residue is located in the center of the β-sheet that forms the ssDNA binding site (28). Modeling of a proline residue at position 221 predicted that it would cause a significant disruption of the beta sheet (Figure 1B). The DNA binding surface of DBD-A has been subjected to mutational analysis previously (19, 23). In these studies, it was shown that mutation of individual or pairs of DNA-interacting residues had only modest effects on ssDNA-binding and resulted in functional forms of RPA (19, 23). The only previously characterized form of DBD-A mutant that was non-functional all six polar residues in DBD-A that interact with DNA were mutated to alanine (23). This form of RPA also had a dramatically reduced ssDNA binding activity (less than 0.1% of wild-type RPA). We conclude that disruption of the core beta sheet in DBD-A has a greater effect on DNA-binding activity than do mutations that affect individual protein-DNA contacts.

Residue L221 is conserved in human, mice and yeast. In addition, the structures and DNA interactions of the human and yeast DBD are very similar (Figure 1A). However in contrast to human cells and mice, yeast that are homozygous for L221P are viable (22, 24). Comparison of the β1–β2 strands in the two structures suggests a possible explanation for this phenotypic difference. In human DBD-A the β1 strand is discontinuous with 3 residues near L221 not part of the beta sheet. In yeast both β1 and β2 strands are continuous (Figure 1B). This difference means that in the human protein proline at position 221 is predicted to result in a long (~6 residue) disruption of the beta strand, while the same change in yeast results in a short (~3 residue) disruption. We postulate that the shorter break in yeast allows the yeast protein to retain sufficient ssDNA-binding activity to support life. Yeast containing L221P have an increase in sensitivity to DNA damage (22, 24) consistent with this mutation having only partial activity.

Genomic instability and high cancer rate was observed in heterozygous mice with one allele of the homologous L221P mutation, but no loss of either the wild type or the L221P allele was found when the tumors were analyzed. This was interpreted by Wang et al. to indicate that the L221P mutant had a dominant phenotype (25). However, we see no evidence of a dominant phenotype when RPA1(L221P) is expressed in RPA1 containing cells. We also find that the RPA complex containing RPA1(L221P) is non-functional in vitro. These results suggest that the genomic instability and high cancer rate observed in the heterozygous mice is caused by haploinsufficiency of RPA. It seems most likely that expression of an inactive form of RPA from one allele causes a reduction in the cellular pool of functional RPA. In the long-term, this leads to elevated levels of DNA damage, genome instability and increased cancer development. While many tumor suppressor genes require both alleles be non-functional to promote cancer (33), a number of genes including many of those directly involved in DNA repair cause an increase in the rate of cancer when only a single allele is inactivated (34). For example haploinsufficiency of BRCA1, bloom syndrome helicase (BLM) or RAD51 all lead to chromosome instability and/or elevated cancer rates (35–38). These studies provide evidence that haploinsufficency of RPA causes a similar phenotype.