Human exonuclease 1 connects nucleotide excision repair (NER) processing with checkpoint activation in response to UV irradiation

hEXO1 has been implicated in several DNA transactions (16, 17), but its possible role in processing NER intermediates, generating the signal for checkpoint activation after UV irradiation, has not yet been investigated. To analyze the cellular localization of hEXO1 in response to UV treatment, we tagged both hEXO1a and hEXO1b isoforms with mCherry in MRC5VI-transformed human fibroblasts. We found that in S-phase cells (∼15–20% of the cell population), both hEXO1 isoforms accumulate in proliferating cell nuclear antigen-positive foci, as reported previously (20). Application of UV irradiation of confined areas of individual cell nuclei using filters containing pores of a defined size has demonstrated that protein factors involved in the repair of UV lesions localize at LUDs (21). We found that after UV irradiation of cells through 5-μm Isopore filters (Millipore), both hEXO1a and hEXO1b accumulated similarly at LUDs, colocalizing with NER preincision factors (Fig. 1 A and B). Quantification of LUDs in which hEXO1a or hEXO1b colocalized with other NER factors revealed that 30–35% of XPA- and XPB-positive LUDs and 45–50% of RPA-positive LUDs also contained hEXO1 (Fig. 1C). This variation can be related to the reportedly different kinetics of XPA, XPB, and RPA recruitment at UV lesions (21, 22). Cells traversing S-phase and non–S-phase cells can be distinguished after local UV irradiation by staining for ethynyl deoxyuridine (EdU) incorporation. EdU staining is bright and homogenously distributed in the nucleus in S-phase cells, and is less bright and localized at LUDs in non–S-phase cells (Fig. S1). Interestingly, the accumulation of hEXO1at LUDs seems to be restricted to non–S-phase cells; no hEXO1 accumulation at LUDs was detected in 100 S-phase nuclei analyzed (Fig. S1).

To gain some insight into the mechanism(s) leading to hEXO1 recruitment at LUDs, we tested whether hEXO1 physically interacts with known NER factors. hEXO1 is expressed at low levels in human nonproliferating tissues (23–26), and the endogenous protein is barely detectable in total crude extracts from cell lines because of its low abundance and limited affinity to the available Abs (23). However, hEXO1 can be recovered and analyzed by Western blot analysis after immunoprecipitation.

As shown in Fig. 1D, XPA, a protein that acts at an early stage of NER (27), coimmunoprecipitates with hEXO1 in the presence or absence of UV treatment. Attempts to detect interaction of hEXO1 with other NER factors by coimmunoprecipitation have so far been unsuccessful (Fig. S2).

Because hEXO1 accumulates at damage sites together with NER factors, we tested the dependency of hEXO1 recruitment at LUDs on NER functions, by analyzing fibroblasts derived from patients with XP. MRC5VI control cells and XPA, XPF, XPF+XPFD676A (corrected with the catalytically dead endonuclease), XPG, and XPG+XPGE791A (corrected with the catalytically dead endonuclease) were transiently transfected with a construct expressing mCherry-tagged hEXO1b and then locally UV-irradiated. Fig. 2A shows that XPB localizes at LUDs in control and XP-mutated cells. Interestingly, hEXO1b is not recruited at LUDs in the absence of the 5′ incision to the lesion (in the XPA, XPF, XPF+XPFD676A, and XPG cell lines). hEXO1b recruitment at LUDs was restored in the presence of the 5′ incision, even though the 3′ cut could not be executed (XPG+XPGE791A) (12). Quantification of the percentage of XPB-positive LUDs that also contain hEXO1b revealed slightly higher recruitment of hEXO1b in XPG cells expressing the catalytically dead nuclease (XPGE791A) than in MRC5VI control cells, with either dramatically reduced or totally abrogated recruitment of hEXO1b in other XP-mutated cell lines.

In conclusion, hEXO1b recruitment at the site of damage requires the NER preincision complex and at least the 5′ incision reaction.

We found that in yeast Exo1 can enlarge NER gaps at a subset of UV-induced lesions when the gap-refilling reaction is somehow slower or impeded. The large ssDNA gaps generated by Exo1 lead to checkpoint activation and are subsequently refilled by repair DNA synthesis (13). If hEXO1 performs a similar function in human cells, then we would expect to see greater recruitment of hEXO1 at LUDs in the presence of 1-β-d-arabinofuranosylcytosine (Ara-C), a DNA synthesis inhibitor. hEXO1 reportedly undergoes degradation after inhibition of replication through hydroxyurea treatment (23), but this did not occur after Ara-C incubation (Fig. S3). Cells locally UV-irradiated and then incubated with Ara-C showed increased accumulation of hEXO1b at XPB-positive LUDs, from 30% to 60% in MRC5VI cells. Such recruitment was lost in XPA cells and was unaffected by Ara-C treatment (Fig. 3). Similar results were obtained for hEXO1a-mCherry (Fig. S4).

NER is a multistep process requiring the refilling of ssDNA gaps generated by the processing of the DNA lesions. The UDS DNA resynthesis step requires lesion recognition and processing (27). NER-deficient cells, which are unable to process UV lesions, are UDS-defective. If hEXO1 were involved in extending the NER-induced gaps, similar to what occurs in yeast (13), then we would expect to find lower UDS levels in hEXO1-depleted cells.

A rapid nonradioactive technique for measuring UDS using the nucleoside analog EdU has been reported recently (28). We optimized hEXO1 silencing by using two different siRNA sequences. Human primary fibroblasts were transfected with siRNAs against hEXO1 as a negative control and against luciferase (LUC) and XPA as positive controls. Both XPA and hEXO1 were efficiently down-regulated (Fig. 4B). Mock- or UV-irradiated cells were then incubated with EdU for 3 h, after which EdU incorporation was quantified in nonreplicating cells, which are easily distinguished from S-phase cells (Fig. S5). Fig. 4A shows that EdU incorporation increased 3.6-fold above background after UV irradiation in control LUC-silenced cells, whereas in hEXO1-depleted cells UDS induction was reduced to 1.7-fold, approaching the reduction seen in siXPA cells (1.3-fold). Considering the EdU incorporation measured after UV irradiation in LUC-silenced cells to be 100%, depletion of hEXO1 causes a 52% reduction, and depletion of XPA causes a 65% reduction. These results suggest that hEXO1 activity is indeed involved in the formation of large ssDNA gaps after UV irradiation.

In response to both DSBs and UV irradiation, RNF8 ubiquitylates histone H2A, and this modification is required for recruiting downstream factors in the damage response (4, 5). Strikingly, when hEXO1 was silenced in quiescent primary fibroblasts and cells were locally UV-irradiated, XPA-positive LUDs were clearly detectable, but H2A ubiquitylation (uH2A) at damage sites was strongly reduced (Fig. 5A). Quantification of various experiments indicates that when hEXO1 was down-regulated (Fig. 5C), uH2A was reduced to ∼40% compared with control (Fig. 5B).

Formation of ssDNA regions by DNA damage processing activates a signal transduction cascade (10, 11). Previous work demonstrated that recognition and processing of UV-induced lesions by NER factors is required for checkpoint activation in nondividing yeast and human cells (6, 7). Asynchronously growing XPA primary fibroblasts exhibit checkpoint activation after UV irradiation, as detected by Chk1 Ser317 phosphorylation; however, this effect is limited to cells traversing S phase, which do not require functional NER to trigger the checkpoint response (7). Yeast Exo1 also has been found to be required for an efficient and rapid NER-dependent G1 checkpoint response (13). In quiescent human primary fibroblasts, knockdown of hEXO1 expression by siRNA causes a defect in checkpoint activation after UV irradiation. In fact, ATR-dependent phosphorylation of both Chk1-Ser317 and p53-Ser15 was reduced to a level similar to that found in XPA fibroblasts or in cells in which XPA expression was knocked down by siRNA (Fig. 6A) up to 120 min after UV irradiation (Fig. 6B).

The following Abs were used: anti–P-Ser-317-Chk1, anti–P-Ser-15-p53, and RPA70 (Cell Signaling Technology); anti-Chk1 and anti-actin Abs (Santa Cruz Biotechnology); anti-p53 (DO1; GeneSpin); anti-hEXO1 mAb (Ab4; Neomarkers); anti-uH2A (Millipore), rabbit anti-XPA (Sigma-Aldrich); and mouse anti-XPA (12F5; from R. D. Wood, Cancer Center, Smithville, TX), anti-XPB (1B3; from J.-M. Egly, University of Strasbourg), and anti-hEXO1 (F15; from S. Ferrari, University of Zurich). Ara-C (Sigma-Aldrich) was used at a final concentration of 1 mM for 1 h at 37 °C. Secondary Abs were goat anti-mouse or goat anti-rabbit conjugated to HRP (Western blot) or to Alexa Fluor 488 or Alexa Fluor 594 (immunofluorescence).

MRC5VI, XP12RO, XP2YO, XP2YO+XPF D676A, XPCS1RO, and XPCS1RO+XPG E791A (all SV40-transformed) were cultured in DMEM containing 10% FCS and kept at 37 °C in a humidified atmosphere with 5% CO₂ (12). The 48BR primary fibroblasts were cultured in DMEM containing 15% FCS. For serum starvation, cells were incubated on plates for 72 h with medium containing 0.5% FCS.

hEXO1a and hEXO1b were PCR-amplified from hEXO1b-omnitag [kindly provided by S. Ferrari (23)], and then cloned into pmCherry-N1 at BamHI/HindIII sites.

MRC5VI and XP12RO fibroblasts were transfected using Metafectene Pro (Biontex Labs) following the manufacturer's instructions. XP2YO, XP2YO+XPF D676A, XPCS1RO, and XPCS1RO+XPG E791A were transfected using Ca₂PO₄ (10 μg DNA each 1.5 × 10⁶ cells) and analyzed at 24–48 h posttransfection.

Cells were seeded and transfected twice with 50 nM siRNA (MWG siLUC: CGU ACG CGG AAU ACU UCG ATT; sihEXO1-1: GCA CGU AAU UCA AGU GAU GTT; sihEXO1-2: CGU AAA UAG AAG AAU AAU UTT; siXPA: ACC UAG AAG AUG ACA UGU ATT) using HiperFect (Qiagen) in DMEM with or without serum depending on the experiment. Cells were harvested and analyzed at 72 h after the first transfection.

The medium was removed, and cells were washed once with PBS and then irradiated with a UV lamp (254 nm) at a dose of 20 J/m². The medium was then added back, and cells were returned to culture conditions for different times. LUD was obtained by irradiating cells through a 5-μm Isopore filter (Millipore) with a UV box (254-nm wavelength) at a rate of 0.6 J/m²/s and a final dose of 40 J/m². The same protocol was used for EdU detection at LUDs.

MRC5VI and XP12RO cells were either mock-transfected or transfected with hEXO1a-mCherry or hEXO1b-mCherry, cultured up to 80% confluence for 24 h, harvested before and after irradiation (20 J/m² for 1 h) using PBS, and processed as described previously (23). A total of 1.5 mg of protein (as determined by the Bradford method), preimmune serum, and F15 anti-EXO1 Ab were used for the immunoprecipitations, as described previously (23). For total protein extract preparation, cells were lysed in 1% SDS sample buffer [62.5 mM Tris-HCl (pH 6.8), 2% wt/vol SDS, 10% glycerol, 50 mM DTT, 0.01% wt/vol bromophenol blue], sonicated for 10 s, and heated to 95 °C for 5 min. Equal amounts of proteins were analyzed by SDS/PAGE.

Cells were seeded on a coverslip, transfected, and irradiated as described above. Cells were washed once in PBS, fixed for 20 min with 2% PFA in PBS, and permeabilized for 5 min with ice-cold PBS containing 0.5% Triton X-100. Blocking was performed in 10% BSA in PBS for 30 min, followed by replacement with primary Ab diluted in PBS with 0.1% Tween 20 (PBST 0.1) for 2 h at room temperature. For uH2A staining, permeabilization was performed with PBS-0.2% Triton before fixation, and the Ab was incubated overnight at 4 °C. Coverslips were washed three times in PBST 0.1 for 10 min, after which secondary Ab diluted in PBST 0.1 was added. Nuclei were counterstained with DAPI. Cells were rinsed in PBST 0.1 three times for 10 min and mounted using ProLong Gold (Invitrogen). Images were obtained using a Leica DMRA2 Microscope (with Leica FW4000 software) with a 100× oil immersion objective (1.30 NA). The percentages of hEXO1-positive LUDs were scored, counting RPA, XPA, or XPB accumulation at local damage only in hEXO1-transfected cells. A value of 1 was assigned each time that hEXO1 showed accumulation at local damage and 0 when it did not. Approximately 40 LUDs were scored for each independent experiment.

The procedure used was essentially as described previously (28). EdU was purchased from Invitrogen. The 48BR cells were transfected in 24-well plates and then cultured on coverslips. After UV irradiation, cells were incubated with serum-free DMEM supplemented with 10 μM EdU for 3 h. Cells were then washed with PBS, followed by a chase period of 15 min with thymidine. Fixation, permeabilization, and fluorescent labeling were performed using Click-iT (Invitrogen), according to the manufacturer's instructions. Cell images were captured with a Leica DMRA2 fluorescent microscope (40× objective, 0.75 NA) equipped with a Leica DC300F CCD camera. Captured images were processed and analyzed with Adobe Photoshop software. At least 100 non–S-phase cells were selected at random for each experiment.