Research ArticleCELL BIOLOGY

PHF8-promoted TOPBP1 demethylation drives ATR activation and preserves genome stability

See allHide authors and affiliations

Science Advances  05 May 2021:
Vol. 7, no. 19, eabf7684
DOI: 10.1126/sciadv.abf7684


The checkpoint kinase ATR [ATM (ataxia-telangiectasia mutated) and rad3-related] is a master regulator of DNA damage response. Yet, how ATR activity is regulated remains to be investigated. We report here that histone demethylase PHF8 (plant homeodomain finger protein 8) plays a key role in ATR activation and replication stress response. Mechanistically, PHF8 interacts with and demethylates TOPBP1 (DNA topoisomerase 2-binding protein 1), an essential allosteric activator of ATR, under unperturbed conditions, but replication stress results in PHF8 phosphorylation and dissociation from TOPBP1. Consequently, hypomethylated TOPBP1 facilitates RAD9 (RADiation sensitive 9) binding and chromatin loading of the TOPBP1-RAD9 complex to fully activate ATR and thus safeguard the genome and protect cells against replication stress. Our study uncovers a demethylation and phosphorylation code that controls the assembly of TOPBP1-scaffolded protein complex, and provides molecular insight into non-histone methylation switch in ATR activation.


In response to genotoxic pressure from both endogenous and exogenous sources, cells have evolved a coordinated network of DNA damage response (DDR) to maintain the integrity of their genomes (1, 2). The DDR process is a signal transduction pathway that coordinates DNA replication, DNA repair, chromosome segregation, and cell cycle checkpoint control (2, 3). Unscheduled DDR may result in deleterious mutations that may be lethal to the cell or induce aberrant cellular behavior leading to malignancies such as cancer (1, 3). The major regulators of the DDR are the evolutionarily conserved phosphoinositide 3-kinase–related protein kinases, including DNA-dependent protein kinase (DNA-PK), ataxia-telangiectasia mutated (ATM), and ATM and rad3-related (ATR). DNA-PK and ATM are primarily triggered by rare occurrences of double-strand breaks (DSBs), whereas ATR, a pivotal element of genome maintenance, is activated by single-stranded DNA (ssDNA) arising at resected DSBs, stalled replication forks, or other single-strand lesions and is distinctively essential for cells coping with intrinsic or extrinsic genomic stress (4, 5).

Mechanistically, ATR senses and detects damaged DNA through association with its obligatory partner, ATR-interacting protein (ATRIP), which is necessary to recruit ATR to replication protein A (RPA)–coated ssDNA (6, 7). The RPA-ssDNA complex also stimulates the binding of the RAD17/RFC checkpoint clamp loader complex to the junction of double-stranded DNA (dsDNA) and RPA-ssDNA (7, 8). The presence of this junction-associated complex then facilitates the loading of the RAD9-HUS1-RAD1 (9-1-1) heterotrimer sliding clamp onto the 5′-ended junctions (8, 9). The physical and functional interaction between ATR/ATRIP and the 9-1-1 complex is linked by DNA topoisomerase 2-binding protein 1 (TOPBP1), a conserved multi-BRCA1 C-terminal (BRCT)–domain scaffolding protein (10, 11). Engagement of TOPBP1 to blocked or collapsed replication forks is largely via interaction between its N-terminal BRCT 0-2 domains and the phosphorylated tail of RAD9, which is constitutively phosphorylated by protein kinase CKII (12, 13).

Once ATR is engaged on the RPA-ssDNA, it could at least phosphorylate itself (14), while the full activation of ATR requires multiple regulators including RAD17, 9-1-1 complex, and TOPBP1 (6, 14). TOPBP1 is believed to allosterically activate ATR via its ATR activation domain (AAD), located between BRCT6 and BRCT7 (10, 11). Recently, Ewing tumor-associated antigen 1 (ETAA1), containing a motif with sequence similarity to AAD, was identified as a second ATR-activator protein (1517). Unlike TOPBP1, ETAA1 is an RPA-interacting protein (1517), and it mainly activates ATR during an unperturbed S phase but not in response to replication stress (18). These findings provide a rationale for the existence of two distinct ATR activators. Moreover, TOPBP1 may be the more important activator for ATR because a point mutation in its AAD (W1147R) results in cellular senescence and early embryonic lethality in mice (19), whereas mutations in ETAA1 AAD do not markedly alter the growth of cells (17). ATR is able to phosphorylate numerous protein substrates including cell cycle checkpoint kinase 1 (CHK1) at Ser345 and RPA at Ser33 (20). In particular, phosphorylation of RPA S33 may be largely taken as, if not exclusively, an ATR-specific event (20, 21).

The activation of the ATR pathway also relies on a dynamic array of protein posttranslational modifications including phosphorylation (14), acetylation (22, 23), ubiquitination (24), and sumoylation (25). As a major type of protein modification, lysine methylation is also an integral part of cellular biology and an important regulator of chromatin-based or non–chromatin-associated signal transduction (26). Yet, whether lysine methylation or demethylation is involved in ATR activation and replication stress response is still poorly understood.

As a histone lysine demethylase, plant homeodomain finger protein 8 (PHF8) recognizes lysine-methylated histone H3 lysine 4 (H3K4) with its N-terminal PHD finger and demethylates H3K9me2/me1, H4K20me1, and H3K27me2 via its Jumonji C (JmjC) domain (27, 28). With this mechanism, the enzyme is thought to regulate key cellular processes such as ribosomal RNA transcription (29), neuronal differentiation (30), and cell migration (31). Loss of Phf8 in mice causes cognitive impairments (32), and mutations in PHF8 are associated with X-linked mental retardation and cleft lip/cleft palate (28). Currently, advances in mass spectrometry (MS) and innovative strategies have led to the identification of ~5000 lysine methylation sites and revealed that non-histone substrates far outnumber histone substrates, so methylation goes well beyond modifying histones (33). However, whether PHF8 is able to demethylate non-histone substrates remains unknown.

In this study, we report that PHF8 interacts with and demethylates TOPBP1, and PHF8-promoted TOPBP1 demethylation is critically involved in replication stress response. We uncover a demethylation and phosphorylation code that controls the assembly of TOPBP1-scaffolded protein complex in ATR activation.


Histone demethylase PHF8 is required for ATR activation

To understand whether lysine methylation plays a role in ATR activation, we performed a targeted small interfering RNA (siRNA) screen to individually deplete 41 lysine methyltransferases (KMTs) and 31 lysine demethylases (KDMs) in U2OS cells (Fig. 1A and fig. S1A). Specifically, cells transfected with siRNAs were challenged with hydroxyurea (HU) for 4 hours to inhibit deoxyribonucleotide production and induce replication fork stalling. Then, cells were immunostained with antibodies against RPA2 S33 phosphorylation (RPA2 pS33), a surrogate marker to monitor ATR activity (20, 21). In the screen, VE-821, a commonly used ATR inhibitor (34), was used as a positive control to block ATR and thus RPA2 pS33. Among the screened enzymes potentially involved in ATR activation (Fig. 1A and fig. S1A), PHF8 depletion had the most dramatic and reproducible effects on foci formation of RPA2 pS33. Thus, we focused on investigating whether and how PHF8 contributes to ATR activation.

Fig. 1 Histone demethylase PHF8 is required for ATR activation and replication stress response.

(A) RNA interference screening and high-content imaging analysis of RPA2 pS33 foci formation in HU-treated U2OS cells (as shown with the color key). (B and C) Immunostaining analysis of RPA2 pS33 and RPA2 foci formation in HU-treated (B, n > 110) or camptothecin (CPT)–treated (C, n > 140) U2OS cells followed by quantification. (D) Immunoblotting analysis of ATR kinase activity with cellular extracts from U2OS cells expressing the indicated siRNAs and PHF8 variants. (E and F) Immunoblotting analysis of ATR kinase activity with biotinylated ssDNA-dsDNA hybrid and nuclear extracts from PHF8-knockout U2OS cells (E) or mouse embryonic fibroblasts (MEFs) (F). (G) DNA fiber assay with the indicated labeling strategy. Fork speed and symmetry were determined by measuring CldU track length, and the percentage of new origins was quantified with only CldU staining fibers (n > 300). (H to J) Analysis of tract lengths of restarted replication forks and the percentage of stalled or restarted forks (red) in HU-treated U2OS cells (H, n > 300; J, n > 300) or MEFs (I, n > 360). (K and L) Survival analysis of PHF8-knockout U2OS cells (K) or U2OS cells expressing PHF8 3′ untranslated region (3′UTR) siRNA and PHF8 variants (L). Data are means ± SDs for (B), (C), (K), and (L) from biological triplicate experiments, and for (G) to (J) from biological duplicate experiments. **P < 0.01, Mann-Whitney test for the scatterplots in (B), (C), and (G) to (J); two-tailed unpaired Student’s t test for each last panel in (G) to (J); two-way analysis of variance (ANOVA) for (K) and (L). Scale bars, 10 μm.

To confirm the screen results, U2OS cells were transfected with two individually distinct siRNAs, and immunofluorescent staining followed by confocal microscopy revealed that PHF8 knockdown markedly compromised HU-induced RPA2 pS33 foci formation (Fig. 1B), whereas the intensity of RPA2 was essentially not changed regardless of the absence or presence of HU (fig. S1, B and C). Similarly, RPA2 pS33 foci were monitored in the presence of camptothecin (CPT), which poisons Topoisomerase I and leads to stalling or collapse of replication forks (35). Although a moderate reduction of RPA2 foci intensity was observed in CPT-treated PHF8-knockdown cells (fig. S1C), the relative intensity of RPA2 pS33 normalized by RPA2 was markedly decreased (Fig. 1C). Meanwhile, Western blotting analysis indicated that PHF8 knockdown or knockout substantially decreased the phosphorylation level of CHK1 S345, RPA2 S33, and RPA2 S4/8 under CPT treatment (fig. S1D). At the time of analysis, PHF8 loss did not significantly alter the cell cycle distribution (fig. S1E), nor did it affect the expression of ATR, ATRIP, TOPBP1, ETAA1, and RAD9 (fig. S1F). Moreover, PHF8 knockdown–impaired activation of ATR could be reversed by forced expression of wild-type PHF8 (PHF8/wt) but not the catalytically inactive PHF8 mutant (PHF8/H247A) (Fig. 1D); thus, PHF8, particularly its enzymatic activity, plays a role in promoting ATR activity.

Because HU exposure or CPT treatment could generate a variable fraction of collapsed replication forks and resected DSBs, thus activating ATR (35, 36), PHF8-promoted ATR activation must be ruled out as a secondary effect of enhanced DNA end resection. To this end, we analyzed the function of PHF8 in RPA2 phosphorylation by in vitro assay. Specifically, an ssDNA-dsDNA hybrid with 5′ junction (linear dsDNA with a 3′ overhang of 70-nt ssDNA) was generated to mimic a perturbed replication fork and incubated with nuclear extracts to activate ATR. Because dsDNA is already resected before it is added to nuclear extracts, this assay provides a unique opportunity to test whether PHF8 has an end processing-independent function in RPA2 phosphorylation. Notably, RPA2 S33 was phosphorylated in an ATR-dependent manner, and PHF8 knockout markedly reduced the level of RPA2 pS33 (Fig. 1E). Similar results were obtained with nuclear extracts from mouse embryonic fibroblast cells carrying wild-type or knockout PHF8 alleles (Fig. 1E). With this strategy, we further confirmed that the catalytic activity of PHF8 is indispensable for ATR activation in resection-defective cells expressing CTIP siRNA (Fig. 1F). These results suggest that PHF8 is able to promote ATR activation in the absence of DNA end resection, although end resection enhanced by PHF8 as we demonstrated here (fig. S1C) and previously reported (37) may also play a role in stimulating ATR.

PHF8 is critically involved in replication stress response

Because the ATR signaling pathway is critical for replication fork elongation, replication timing control, and recovery of stalled replication (38), we wondered whether PHF8 plays a role in DNA replication integrity. To test this, we used DNA fiber assays to analyze the impact of PHF8 loss on the status of individual replication forks. In the absence of replication stress, PHF8-deficient cells showed reduced fork elongation rates, accompanied by a decrease in the symmetry of bidirectional replication forks and an increase in new origin firing (Fig. 1G and fig. S1G). Under HU treatment, PHF8 loss led to a defective response to replication stress, as evidenced by a significant reduction in replication strand lengths, an increased proportion of stalled replication forks, and a concomitant reduction in the rate of replication fork restart (Fig. 1, H and I, and fig. S1G). Furthermore, PHF8/H247A, the enzymatic dead mutant, could not efficiently promote fork progression as PHF8/wt did in PHF8-knockdown cells (Fig. 1J). These results suggest that PHF8 is critical for replication fork stability and DNA replication integrity under both unchallenged and stressful conditions.

We then investigated whether PHF8 is involved in the cellular response to replication stress. First, we analyzed the ability of PHF8-knockout cells to recover from HU-induced replication arrest and found that at 12 hours after the release from HU challenge, γH2AX foci persisted in PHF8-deficient cells but largely disappeared from PHF8-proficient cells (fig. S1H). Thus, PHF8 is important for recovery from replication arrest. Alkaline comet assay analysis indicated that PHF8 knockout led to an accumulation of more DNA breaks even in the absence of HU or CPT challenge (fig. S1I). This phenomenon is consistent with PHF8 loss accompanied by an elevated level of γH2AX (fig. S1, H and J). In response to replication stress-associated DNA damage, ATR and its effectors ultimately induce a G2-M checkpoint that prevents entry into mitosis (9). To test whether PHF8 is required for activation of the G2-M checkpoint, we measured the fraction of mitotic cells by phospho-Histone H3S10 staining. PHF8 knockout increased cells in mitosis after exposure to CPT, which suggests that the G2-M checkpoint was defective in the absence of PHF8 (fig. S1, K and L). These results indicate that PHF8 plays an important role in responding to replication stress and genome stability.

Next, we examined the effect of PHF8 deficiency on cell survival after genotoxic insults. PHF8-knockout cells showed a marked increase in sensitivity to HU or CPT (Fig. 1K). Also, PHF8-knockout cells were more vulnerable in the presence of cisplatin (Fig. 1K), a well-known chemotherapeutic drug that kills cancer cells largely by inhibiting replication by forming cross-linked DNA adducts (39). Because ATR signaling has been reported as a synthetic lethality partner to PARP inhibitors (PARPis) (34), we wondered whether PHF8 depletion could sensitize cells treated with PARPis. PHF8 depletion failed to activate ATR and decreased cell survival (fig. S1, M and N). This finding is likely due to intact ATR signaling being intrinsically required for cells to cope with challenges from the PARPis rucaparib and talazoparib, which are believed to trap PARP1 on chromatin and thus pose an obstacle to replication forks (40). Furthermore, on survival analysis, the demethylase activity of PHF8 was essential in protecting cells against CPT- or rucaparib-triggered replication stress (Fig. 1L). These data suggest that PHF8 is critically involved in replication stress response, and this is possibly associated with its role in promoting ATR activity.

PHF8 is physically associated with TOPBP1

To further understand how PHF8 regulates ATR activation, we first examined whether PHF8 is recruited to stalled or collapsed replication forks. Discrete foci of ATRIP colocalized with RPA2 were readily detected in U2OS cells after HU or CPT treatment, whereas PHF8 was not recruited to RPA-bound ssDNA regions (fig. S2A), so PHF8 does not accumulate at sites of replication stress. To further confirm this observation, we used the Lac operon–Lac repressor (LacO-LacI) targeting system (41) to monitor the distribution of PHF8 when the replication fork is perturbed. In cells expressing LacI, LacO arrays that are stably integrated into the genome are bound by LacI, and the multiple LacO-LacI nucleating complex poses an obstacle to DNA polymerase progression, thereby resulting in site-specific replication fork blockage (fig. S2B) (42). Subsequently, excess ssDNA is generated by structure-specific endonuclease and ATR pathway is activated at the stalled then possibly collapsed fork (43). Although expression of an mCherry-LacI fusion protein resulted in a robust accumulation of ATRIP, RAD9, or TOPBP1 around the arrays in EdU (5-Ethynyl-2′-deoxyuridine)-positive cells, we did not observe evident array-specific accumulation of PHF8 (fig. S2, B and C). These results further support the idea that PHF8 is not directly associated with perturbed forks upon replication stress.

Because ATR activation requires multiple DDR factors for assembly into DNA lesions featured with RPA-coated ssDNA (4), we hypothesized that PHF8 could physically and functionally associate with one or many of these factors to orchestrate the assembly of ATR-nucleated protein complex on damaged chromatin upon replication stress. To test this hypothesis, we used immunopurification and MS with stable isotope labeling with amino acids in cell culture (SILAC) to quantitatively interrogate PHF8 interactome. Quantitative MS analysis of the FLAG-PHF8–containing protein complex revealed that PHF8 was associated with a number of proteins, including ubiquitin-specific protease USP7, a deubiquitinase for PHF8 (Fig. 2A and table S1) (37). TOPBP1, a well-known ATR activator (10), was also highly enriched in PHF8-containing protein complex (Fig. 2A).

Fig. 2 PHF8 is physically associated with TOPBP1.

(A) SILAC-based quantitative MS analysis of PHF8-containing protein complex with cellular extracts from HeLa cells that allow Dox-inducible expression of stably integrated FLAG-PHF8. (B) Coimmunoprecipitation analysis of the interaction between PHF8 and TOPBP1 with cellular extracts from HeLa, MCF-7, and U2OS cells. (C) Coimmunoprecipitation analysis of the association of PHF8 with TOPBP1 under different nucleases treatment. (D) Colocalization of endogenous PHF8 with mCherry-LacI-TOPBP1. U2OS cells stably integrated with LacO arrays (U2OS-LacO) were transfected with mCherry-LacI-TOPBP1 and control siRNA or PHF8 siRNAs followed by immunostaining and confocal microscopy analysis. The intensity of PHF8 foci was quantified and normalized to the nuclear background (n > 150). Data are means ± SDs from biological triplicate experiments. **P < 0.01, Mann-Whitney test. Scale bars, 10 μm. (E) GFP-tagged truncation mutants of TOPBP1 or FLAG-tagged deletion mutants of PHF8 were transfected into HeLa cells followed by coimmunoprecipitation analysis. (F) FLAG-tagged PHF8/wt or APS deletion mutant of PHF8 (PHF8/ΔAPS) was transfected into HeLa cells followed by coimmunoprecipitation analysis. (G) Glutathione S-transferase (GST) pull-down assays with recombinant GST-BRCT 7-8 and His-tagged PHF8 deletions purified from bacterial cells. The asterisks indicate the recombinant proteins stained by Coomassie blue. (H) Surface plasmon resonance (SPR) analysis of the binding affinity of recombinant BRCT 7-8 and the APS, adjacent N-terminal (APS-N21), or C-terminal (APS-C22) control peptide.

To confirm the association of PHF8 with TOPBP1, coimmunoprecipitation experiments were performed with HeLa cell extracts, and the results showed that PHF8 but not another histone demethylase, LSD1, was efficiently immunoprecipitated by TOPBP1, although LSD1 could be effectively coimmunoprecipitated with HDAC1 (Fig. 2B) (44). Reciprocally, TOPBP1 but not NBS1 was efficiently immunoprecipitated by PHF8, although NBS1 could be effectively coimmunoprecipitated with TOPBP1 (Fig. 2B) (45). Similar observations were obtained when coimmunoprecipitation experiments were conducted with cellular extracts from MCF-7 and U2OS cells (Fig. 2B). The interaction of PHF8 with TOPBP1 was independent of nucleic acid incorporation because treating cellular extracts with deoxyribonuclease (DNase), ribonuclease (RNase), or their combination had a marginal effect on the integrity of this binary protein complex (Fig. 2C). In addition, immunofluorescent staining followed by confocal microscopy revealed that PHF8 formed bright foci colocalized with mCherry-LacI-TOPBP1 in U2OS cells carrying LacO arrays (Fig. 2D and fig. S2D), whereas PHF8 did not form discernable foci in cells expressing mCherry-LacI (fig. S2, B to D), implying that PHF8 specifically interacts with genome-anchored TOPBP1 but not the perturbed replication fork. Collectively, these results suggest that PHF8 is physically associated with TOPBP1.

To gain molecular insight into the interaction of PHF8 and TOPBP1, domain deletion mutants of green fluorescent protein (GFP)–tagged TOPBP1 were generated and transfected into HeLa cells. Immunoprecipitation analysis indicated that BRCT 7-8 in the C terminus of TOPBP1 was responsible for the interaction of PHF8 with TOPBP1 (Fig. 2E and fig. S2E). Reciprocally, domain mapping of the molecular interface of PHF8 required for TOPBP1 binding revealed that an acidic patch sequence (APS) with 22 amino acids, spanning amino acids 842 to 863 in the C terminus of PHF8, was required for the association of PHF8 with TOPBP1 (Fig. 2, E and F). The fluorescence colocalization approach with the LacO-LacI-TOPBP1 system confirmed the importance of the APS motif in the PHF8-TOPBP1 interaction (fig. S2F). To substantiate this finding, coimmunoprecipitation experiments were performed with cellular extracts in the presence of the APS peptide or peptide adjacent to this region: The binding of PHF8 to TOPBP1 was severely disrupted by the APS peptide (fig. S2G). Sequence alignment showed that this APS fragment is highly conserved from zebrafish to humans (fig. S2H), so it may play an important role in PHF8-TOPBP1 binding across different species.

Next, glutathione S-transferase (GST) pull-down experiments with a His-tagged C-terminal region of PHF8 inclusive or exclusive of APS and GST-fused BRCT 7-8 confirmed the requirement of APS for its specific interaction with BRCT 7-8 (Fig. 2G). To further characterize the affinity of this interaction, real-time binding of biotin-tagged APS to BRCT 7-8 or GST-BRCT 7-8 was monitored by surface plasmon resonance (SPR) analysis. Shown as representative sensorgrams, this quantitative biophysical analysis with global fitting curves revealed that APS but not its adjacent control regions strongly bound to recombinant BRCT 7-8 or GST-BRCT 7-8 with micromolar- or nanomolar-scale dissociation constant (Kd), respectively (Fig. 2H and fig. S2I). These results suggest that PHF8 directly interacts with TOPBP1 via APS of PHF8 and BRCT 7-8 of TOPBP1.

Crystal structure analysis identifies key determinants of PHF8-TOPBP1 binding

To further understand the molecular mechanism of the recognition between PHF8 and TOPBP1, we solved a complex structure of BRCT 7-8 (amino acids 1264 to 1493) and APS motif (amino acids 842 to 863) of PHF8 by x-ray diffraction at 1.7 Å resolution. The Rwork and Rfree of the refined structure was 17.8 and 21.5%, respectively. One PHF8-APS peptide bound to one BRCT 7-8 molecule in one asymmetric unit (Fig. 3A and table S2). All residues were well defined in the structure except regions spanning residues 1441 to 1450; the N and C termini of BRCT 7-8 (residues 1264 to 1265 and 1493) were disordered. The density map of the PHF8-APS peptide had good quality, encompassing amino acids 842 to 856 (842GACFKDAEYIYPSLESDDDDPA863, residues underlined), leaving the tandem-aspartate region 857 to 863 invisible (fig. S3, A and B). The PHF8-APS peptide is bound in a cleft formed between the BRCT7 and BRCT8 domains of TOPBP1 and runs through the entire structure of BRCT 7-8 (Fig. 3, A and B). Three interaction interfaces between PHF8-APS and BRCT 7-8 are indicated in the red circled regions of Fig. 3A, and they bury a total surface area of 834.1 Å2, thus suggesting a rather tight binding (Fig. 3A). The overall arrangement of the two BRCT domains is similar to that of the BRCA1-associated C-terminal helicase (BACH1)–BRCT 7-8 structure [Protein Data Bank (PDB) ID: 3AL3] (46) because the two structures can be superimposed with a root mean square deviation of 1.7 Å. However, the binding mode of PHF8-APS to BRCT 7-8 differs from that of BACH1. In Fig. 3B, yellow-colored BACH1 mainly binds to the positively charged region of BRCT7, whereas magenta-colored PHF8-APS interacts with both BRCT7 and BRCT8 via a mixture of hydrophobic and polar interactions. Among the three interaction interfaces of PHF8-APS and BRCT 7-8, polar interactions mainly occurred in regions 1 and 3, with hydrophobic interactions in region 2.

Fig. 3 Key determinants of PHF8-TOPBP1 binding.

(A) The overall structure of PHF8-APS binding to BRCT 7-8. BRCT 7-8 is shown in a cartoon model with BRCT7 in cyan and BRCT8 in green. PHF8-APS peptide is shown in a stick model (carbon, magenta; nitrogen, blue; oxygen, red; sulfur, yellow). Three protein interaction interfaces are highlighted with red dashed circles. (B) Electrostatic potential and substrate binding cleft on the surface of BRCT 7-8 (positive, blue; neutral, white; negative, red). The BACH1–BRCT 7-8 structure (PDB ID: 3AL3) is aligned by the BRCT 7-8 domain. BACH1 phosphor peptide and PHF8-APS peptide are shown in stick representation with carbon in yellow and magenta, respectively. (C) Detailed interactions at the three interfaces. Dashed lines indicate hydrogen bonds and distances are labeled in black and expressed in angstroms. (D) Isothermal titration calorimetry (ITC) analysis of the binding affinity of recombinant BRCT 7-8 and APS peptide or its mutant variants. (E) ITC analysis of the binding affinity of recombinant BRCT 7-8 variants and APS peptide. (F) FLAG-tagged PHF8 variants were transfected into HeLa cells followed by coimmunoprecipitation analysis. (G) Colocalization of PHF8 variants with mCherry-LacI-TOPBP1 in U2OS-LacO cells. The intensity of PHF8 foci was quantified and normalized to the nuclear background (n > 160). Data are means ± SDs from biological triplicate experiments. **P < 0.01, Mann-Whitney test. Scale bars, 10 μm. (H) FLAG-tagged C-terminal TOPBP1 carrying AAD and BRCT 7-8 domains (TOPBP1C/wt) or its counterpart mutants were transfected into HeLa cells followed by coimmunoprecipitation analysis.

Detailed structural analyses reveal the determinants of the binding at the following three interaction interfaces: (i) Region 1 through G842 to E849 of PHF8 (Fig. 3C): In this region, polar interaction occurs between D847 of PHF8 and R1413 of BRCT8 by forming hydrogen bonds with each other’s side chain. R1413 also contributes to the domain interaction between BRCT7 and BRCT8 by forming two hydrogen bonds with the side chain of E1316. C844 of PHF8 connects with C1299 of BRCT7 through disulfide linkage, but this may be an in vitro artifact because forming a disulfide bond inside a nucleus with a reductive environment is difficult. (ii) Region 2 through Y850 to P853 of PHF8 (Fig. 3C): Y852 of PHF8 inserts into a hydrophobic pocket formed by F1411, L1414, A1470, and L1319 of BRCT 7-8. Y850 and I851 of PHF8 also form hydrophobic interactions with F1411 and L1313 of BRCT 7-8, respectively. (iii) Region 3 through S854 to E856 (Fig. 3C): Both the side-chain hydroxyl group of S854 and the main-chain amide group of L855 of PHF8 form hydrogen bonds with the side chain of D1471 of BRCT8. D1471 also contributes to the inner-domain stabilization by forming polar interactions with Y1484. Meanwhile, E856 interacts with R1369 of BRCT7 through a side-chain hydrogen bond.

To test the role of these key residues in intermolecular interaction, we quantitatively measured the binding ability of APS and BRCT 7-8 variants by isothermal titration calorimetry (ITC). When residues C844, Y850, Y852, and S854 of PHF8 or residues F1411, R1413, and D1471 of TOPBP1 were individually mutated, the affinity of APS to BRCT 7-8 was greatly reduced, while alanine substitution of D847 partially weakened the binding (Fig. 3, D and E, and fig. S3C). On coimmunoprecipitation experiments performed with various PHF8 mutants, Y850 and Y852 contributed substantially to the complex formation (Fig. 3F). This finding was further supported by the fluorescence colocalization approach with the LacO-LacI-TOPBP1 system (Fig. 3G). Reciprocally, coimmunoprecipitation analysis with TOPBP1 mutants revealed that the conserved residues R1413 and D1471 are specifically required for TOPBP1 binding to PHF8, while F1411 is responsible for TOPBP1 association with both PHF8 and BACH1 (Fig. 3H and fig. S3D). In conclusion, Y850 and Y852 of PHF8 as well as R1413 and D1471 of TOPBP1 are specifically key determinants of the PHF8-TOPBP1 interaction.

Targeting the molecular interface of PHF8 and TOPBP1 suppresses ATR activity

Next, we sought to investigate the biological significance of the PHF8-TOPBP1 interaction. First, Western blotting analysis showed that the defective activation of ATR in PHF8-deficient cells could be restored by forced expression of PHF8/wt but not PHF8/∆APS (PHF8 lacking APS motif) and PHF8/Y852A, which could not form a complex with TOPBP1 (Fig. 4A). Consistently, DNA fiber assays revealed that unlike PHF8/wt, neither PHF8/∆APS nor PHF8/Y852A could help PHF8-depleted cells overcome replication defects (Fig. 4B). Next, survival analysis demonstrated that the APS motif, particularly of Y852 residue, is critically required for cells to counteract replication stress (Fig. 4C).

Fig. 4 Targeting the molecular interface of PHF8 and TOPBP1 suppresses ATR activity.

(A) Analysis of ATR kinase activity by immunoblotting with cellular extracts from U2OS cells expressing PHF8 3′UTR siRNA and PHF8 variants. (B) Analysis of tract lengths of restarted replication forks and percentage of stalled or restarted forks in U2OS cells expressing PHF8 3′UTR siRNA and PHF8 variants (n > 300). (C) Survival analysis of U2OS cells expressing PHF8 3′UTR siRNA and PHF8 variants under CPT treatment. (D) U2OS cells stably expressing NLS-GFP, NLS-GFP–fused APS, or NLS-GFP–fused control region (APS-C22) were treated with 2 mM HU for 4 hours followed by immunostaining and confocal microscopy analysis. The percentage of RPA2 pS33 foci-positive cells was quantified (n > 150). (E and F) ATR kinase activity (E) and survival (F) analysis with U2OS cells stably expressing the indicated proteins under CPT treatment. (G) U2OS cells stably expressing the indicated proteins were collected for coimmunoprecipitation analysis. (H) ATR kinase activity and survival analysis of U2OS cells expressing TOPBP1 3′UTR siRNA and TOPBP1 variants under CPT treatment. Data are means ± SDs for (B) from biological duplicate experiments, and (C), (D), (F), and (H) from biological triplicate experiments. **P < 0.01, Mann-Whitney test for the left panel of (B); one-way ANOVA for (D) and the right panel of (B); two-way ANOVA for (C), (F), and (H). Scale bars, 10 μm.

To further assess the contribution of PHF8-TOPBP1 interaction to ATR activation, the APS motif or its proximal control region was then fused with nuclear-localized GFP (NLS-GFP) and stably integrated into the genome of U2OS cells. Cells expressing GFP-APS showed lower ATR activity (Fig. 4, D and E) and were more sensitive to replication stress (Fig. 4F). Meanwhile, coimmunoprecipitation analysis showed that GFP-tagged APS but not its proximal region competed with endogenous PHF8, thus greatly impairing the association of PHF8 with TOPBP1 (Fig. 4G). These results indicate that disruption of PHF8-TOPBP1 binding could suppress ATR activity.

On the other side, we demonstrated that the expression of TOPBP1/R1413Q or TOPBP1/D1471A mutant that could not interact with PHF8 failed to efficiently activate ATR in TOPBP1-depleted cells (Fig. 4H), indicating that the binding of TOPBP1 to PHF8 is critical for TOPBP1-promoted ATR activation. Consistently, survival analysis suggested that residues R1413 and D1471 are crucial for TOPBP1 in responding to replication stress (Fig. 4H). In summary, these observations point to an important role of the physical association of PHF8 with TOPBP1 in ATR activation.

PHF8 promotes the engagement of TOPBP1 to sites of replication stress

To understand the molecular mechanism of how PHF8 controls ATR activation, we aimed to examine whether PHF8 plays a role in TOPBP1 loading at DNA damage sites upon replication stress. Confocal microscopy analysis demonstrated that the foci formation of TOPBP1 was markedly impaired in PHF8-knockout cells upon HU (Fig. 5, A and B, and fig. S4A) or CPT treatment (fig. S4, A to C), whereas chromatin-bound TOPBP1 manifested by the intensity of the remaining TOPBP1 after preextraction appeared to be not affected by PHF8 loss in unstressed cells (fig. S4D). Similar results were obtained when the LacO-LacI targeting system was used to examine endogenous TOPBP1 engagement on perturbed replication fork (Fig. 5C and fig. S4E). However, the foci formation of ATRIP was not affected in PHF8-knockdown cells (fig. S4F). Then, analysis of chromatin factors bound to nascent replicating forks by isolation of proteins on nascent DNA (iPOND) assay revealed that PHF8 ablation significantly disrupted TOPBP1 accumulation specifically at blocked replication forks (Fig. 5D and fig. S4G). Similar to TOPBP1, the binding of DNA sliding clamp proliferating nuclear antigen A (PCNA), a processivity factor for the elongation of nascent DNA strands at replication forks that functions downstream of TOPBP1 (47), was largely unaffected in PHF8-knockout cells without HU treatment (Fig. 5D), suggesting that PHF8 is unlikely involved in TOPBP1-governed replication initiation process.

Fig. 5 PHF8 promotes the engagement of TOPBP1 to sites of replication stress.

(A) PHF8-knockout U2OS cells expressing Myc-TOPBP1 were treated with 2 mM HU for 4 hours followed by preextraction, immunostaining, and quantification (n > 120). (B) Experiments analogous to (A) for endogenous TOPBP1 foci examination (n > 120). (C) U2OS-LacO cells expressing mCherry-LacI and PHF8 siRNAs were labeled with EdU for 1 hour followed by immunostaining and quantification (n > 120). (D) Proteins associated with replication forks were examined by iPOND in PHF8-knockout U2OS cells. (E) U2OS cells expressing PHF8 3′UTR siRNA and PHF8 variants were treated with HU followed by preextraction, immunostaining, and quantification (n > 120). (F) U2OS-LacO cells expressing PHF8 3′UTR siRNA and PHF8 variants were labeled with EdU followed by immunostaining and quantification (n > 120). (G) Pull-down assays with RPA-coated ssDNA-dsDNA and nuclear extracts from HeLa cells expressing PHF8 3′UTR siRNA and PHF8 variants. (H) U2OS-LacO cells were transfected with the indicated proteins followed by immunostaining and quantification (n > 200). (I) U2OS-LacO cells expressing the indicated proteins were labeled with EdU followed by immunostaining and quantification (n > 100). (J) Pull-down assays with RPA-coated ssDNA-dsDNA and nuclear extracts from HeLa cells expressing TOPBP1 3′UTR siRNA and TOPBP1 variants. Data are means ± SDs for (A) to (C), (E), (F), (H), and (I) from biological triplicate experiments. **P < 0.01, one-way ANOVA for (A), (B), and (E); Mann-Whitney test for (C), (F), (H), and (I). Scale bars, 10 μm.

To further confirm the effect of PHF8 on the loading of TOPBP1 to damaged forks, we next used pull-down assays with biotinylated and RPA-coated 5′ junction ssDNA-dsDNA that carries a 3′ overhang of 70-nt ssDNA (fig. S4H). TOPBP1 could be efficiently precipitated when native nuclear extracts were incubated with RPA-ssDNA/dsDNA complexes, whereas PHF8 knockout notably impaired the efficiency of TOPBP1 binding to these DNA structures (fig. S4I). These results support the idea that PHF8 is critically involved in the control of TOPBP1 loading on sites of replication stress.

Consistent with the molecular action of PHF8 in ATR activation, visual inspection and quantitative analysis of immunostaining of cells expressing PHF8 siRNA and siRNA-resistant PHF8/wt, PHF8/H247A, PHF8/∆APS, or PHF8/Y852A revealed that both the enzymatic activity and the TOPBP1 binding surface of PHF8 are required for the efficient accumulation of TOPBP1 at DNA damage sites undergoing replication stress (Fig. 5, E and F). Notably, this argument was further supported by pull-down assays with RPA-ssDNA/dsDNA complexes and nuclear extracts (Fig. 5G). Next, we showed that only NLS-GFP–tagged APS but not the proximal control region could attenuate the foci formation of TOPBP1 (Fig. 5H and fig. S4J). Last, in support of TOPBP1 accumulation relying on its association with PHF8, mutation of TOPBP1 including R1413Q and D1471A led to defective foci formation (Fig. 5I and fig. S4K) and binding to RPA-ssDNA/dsDNA complexes (Fig. 5J). Therefore, PHF8 acts as an essential regulator in controlling the engagement of TOPBP1 to replication stress sites.

TOPBP1 is also important for transcriptional repression of E2F1 proapoptotic target genes and, thus, suppression of E2F1-dependent apoptosis (4850). This prompted us to test whether PHF8-TOPBP1 interaction controls the expression of TOPBP1-regulated E2F1 target genes. We showed that, unlike TOPBP1, PHF8 depletion had minor effects on the expression of these genes, whereas the expression of E2F1 target genes CCNA2, CDC25A, and CCNE1 was down-regulated, albeit not markedly (fig. S4, L and M). This is consistent with previous studies and the understanding that PHF8 is a transcriptional activator in gene regulation (2729). The mild reduction of these cell cycle regulators provides an explanation for why PHF8 loss essentially does not alter cell cycle profile (fig. S1E). Next, we found that, in PHF8-depleted cells, overexpression of TOPBP1 binding defective mutant PHF8/Y852A could efficiently restore the expression of CCNA2, CDC25A, and CCNE1 (fig. S4M). These results suggest that PHF8 and TOPBP1 regulate distinct subsets of E2F1 target genes and TOPBP1-mediated E2F1 transcriptional activity is not affected by PHF8.

PHF8 demethylates K118 mono-methylation of TOPBP1 to facilitate its accumulation

Because PHF8 is not present at replication stress–associated damaged chromatin and the demethylase activity of PHF8 is required for TOPBP1 accumulation and ATR activation, we hypothesized that TOPBP1 is a potential substrate of PHF8. To test this idea, we first confirmed that BRCT 0-2 is sufficient and necessary for TOPBP1 binding to DNA damage sites upon replication stress (fig. S5, A and B), which agrees with previous reports (23, 51). Similar to the behavior of R1413Q and D1471A mutants, TOPBP1 lacking BRCT 7-8 (TOPBP1/∆BRCT 7-8) was still able to form foci but with much less efficiency than TOPBP1/wt, whereas TOPBP1 BRCT 7-8 alone completely failed to form foci (fig. S5, A and B). These observations suggest that TOPBP1 requires BRCT 7-8 to localize at replication stress sites, but BRCT 7-8 itself does not directly contribute to TOPBP1 accumulation. In terms of our finding that PHF8 binds BRCT 7-8 to control TOPBP1 accumulation, we speculated that PHF8 possibly depends on this interaction to further target the methylation site(s) of key lysine residue(s), likely on BRCT 0-2 in TOPBP1, thus promoting TOPBP1 redistribution upon DNA damage.

To test this hypothesis, we first determined the methylation site(s) with purified TOPBP1 from PHF8-knockout cells using MS and found that lysine residue K118, a conserved site across different species, is mono-methylated (K118me1) (Fig. 6A) and K126 and K154 are di-methylated (K126me2; K154me2) (fig. S5C). To identify the relevant methylation site(s) targeted by PHF8 for TOPBP1 accumulation, we generated six TOPBP1 mutants: K118A, K126A, and K154A, which are neutral substitutes with a very small side chain and could not be methylated, and K118R, K126R, and K154R, which are analogous substitutes that could mimic a constitutively nonmethylated lysine while maintaining the potential of forming hydrogen bond by side chain. Then, the foci formation of these mutants was examined by confocal microscopy. K118 and K154 are critical for TOPBP1 accumulation, as mutation of these residues to alanine severely disrupted TOPBP1 foci formation (fig. S5D). Notably, only TOPBP1/K118R but neither TOPBP1/K154R nor TOPBP1/wt could efficiently form foci in PHF8-deficient cells (Fig. 6B and fig. S5E). As the closest natural mimic to a mono-methylated lysine is methionine, which, to a certain extent, represents a monomethyl side chain (52), we next used methionine substitution to mimic K118me1. The results showed that TOPBP1/K118M failed to form foci in high efficiency (fig. S5D). Furthermore, the above findings were validated by pull-down assays with RPA-ssDNA/dsDNA complexes and nuclear extracts from cells expressing these mutants in TOPBP1- or PHF8-depleted cells (Fig. 6C). These results revealed that PHF8-promoted TOPBP1 loading is likely via removing methyl mark from K118.

Fig. 6 PHF8 demethylates K118 mono-methylation of TOPBP1 to promote its accumulation.

(A) MS analysis of TOPBP1 lysine methylation sites with purified FLAG-TOPBP1 from PHF8-knockout HeLa cells. Fragmentation spectra and parameters of K118me1 and a schematic presentation of the evolutionarily conserved K118 residue are shown. (B) U2OS-LacO cells expressing PHF8 3′UTR siRNA, mCherry-LacI, and TOPBP1 variants were labeled with EdU followed by immunostaining and quantification (n > 105). (C) Pull-down assays with RPA-coated ssDNA-dsDNA and nuclear extracts from HeLa cells expressing TOPBP1 variants and 3′UTR siRNA against TOPBP1 or PHF8. (D) TOPBP1 K118me1 level in PHF8-knockout or Dox-inducible PHF8-expressing U2OS cells was examined by immunoprecipitation assays. (E) U2OS cells expressing PHF8 3′UTR siRNA and PHF8 variants were collected for immunoprecipitation and immunoblotting analysis. (F) In vitro demethylation assay with recombinant N-terminal PHF8 (residues 37 to 483, PHF8N) variants purified from bacteria cells and K118me1 peptide followed by dot blot analysis. (G) In vitro demethylation assay with recombinant PHF8N variants and K118me1 peptide followed by matrix-assisted laser desorption/ionization–time-of-flight (MALDI-TOF) MS analysis. (H) ATR kinase activity and cell survival analysis in U2OS cells expressing TOPBP1 3′UTR siRNA and TOPBP1 variants. (I) ATR kinase activity and cell survival analysis of U2OS cells expressing PHF8 3′UTR siRNA and TOPBP1 variants. Data are means ± SDs for (B), (H), and (I) from biological triplicate experiments. **P < 0.01, Mann-Whitney test for (B); two-way ANOVA for (H) and (I). Scale bars, 10 μm.

To assess whether PHF8 could demethylate K118me1 of TOPBP1, we generated a rabbit polyclonal antibody that could specifically recognize the K118 mono-methylated TOPBP1 (fig. S5, F to H). We found that PHF8 knockout or overexpression could notably enhance or reduce the level of K118me1, respectively (Fig. 6D). In contrast, both PHF8/H247A and PHF8/Y852A mutants failed to demethylate K118me1 (Fig. 6E), whereas PHF8/Y852A was still active in removing methyl marks from histones (Fig. 6E), which indicates that PHF8/Y852A is a separation-of-function mutant of PHF8. Moreover, we performed in vitro demethylation assays with the N-terminal recombinant PHF8 and TOPBP1 K118me1 peptide. Dot blot and MS analysis revealed that the monomethyl residue could be efficiently removed by PHF8 but not its enzymatic dead mutant (Fig. 6, F and G, and fig. S5I). These results strongly support the notion that TOPBP1 K118me1 is a substrate of PHF8.

To further define the role of PHF8-promoted TOPBP1 demethylation in replication stress response, we examined ATR activity and cell survival with cells expressing various siRNA-resistant K118 mutants and TOPBP1 siRNA. Only TOPBP1/wt and TOPBP1/K118R but not TOPBP1/K118A or TOPBP1/K118M could restore TOPBP1 depletion–associated defects of ATR activation and growth retardation (Fig. 6H). Similar experiments were performed with cells expressing these TOPBP1 variants in PHF8-deficient cells: Only cells expressing TOPBP1/K118R behaved like control cells (Fig. 6I), which further indicates that PHF8 acts epistatically on TOPBP1 and confirms the functional link between PHF8 and TOPBP1. These results suggest that PHF8-promoted TOPBP1 demethylation plays an important role for cells to activate ATR and cope with replication stress.

PHF8-promoted TOPBP1 demethylation enforces TOPBP1 binding to RAD9

We next investigated the underlying mechanism of how TOPBP1 K118me1 demethylation facilitates the chromatin loading of TOPBP1 in response to replication stress. TOPBP1 is considered to mobilize to damaged forks by interacting with the MRN complex via its BRCT 3-6 domains (53) or with the RAD9 subunit of the 9-1-1 complex via its BRCT 1-2 domains (12, 13, 23). NBS1 or RAD9 depletion significantly compromised TOPBP1 foci formation in HU-treated cells (fig. S6A), and the interaction between TOPBP1 and RAD9 but not TOPBP1 and NBS1 was severely compromised upon PHF8 knockout or knockdown (Fig. 7A and fig. S6B), which implies that PHF8-promoted TOPBP1 loading largely depends on RAD9.

Fig. 7 PHF8-promoted TOPBP1 demethylation facilitates TOPBP1-RAD9 binding.

(A) Myc-RAD9 or GFP-NBS1 was transfected into PHF8-knockout U2OS cells followed by coimmunoprecipitation analysis. (B) Myc-RAD9 and siRNA-resistant FLAG-PHF8 variants were cotransfected into PHF8-knockdown U2OS cells, and cellular extracts were collected for coimmunoprecipitation analysis. (C) siRNA-resistant FLAG-TOPBP1 variants were transfected into TOPBP1-knockdown U2OS cells followed by mock or 2 mM HU treatment, and cellular extracts were collected for coimmunoprecipitation analysis. (D) Biotinylated RAD9 pS387 peptide was incubated with cellular extracts from cells expressing FLAG-BRCT 0-2 or FLAG-TOPBP1 variants followed by affinity-based precipitation analysis. (E) U2OS-LacO cells were transfected with PHF8 3′UTR siRNA, mCherry-LacI, and PHF8 variants, and cells were labeled with EdU for 1 hour followed by immunostaining and confocal microscopy analysis. The intensity of RAD9 foci in mCherry-LacI and EdU+ cells was quantified and normalized to the nuclear background (n > 150). (F) Nuclear extracts from HeLa cells expressing PHF8 variants and PHF8 3′UTR siRNA were incubated with biotinylated and RPA-coated ssDNA-dsDNA hybrid followed by pull-down and immunoblotting analysis. (G) U2OS-LacO cells were transfected with TOPBP1 3′UTR siRNA, mCherry-LacI, and TOPBP1 variants, and cells were labeled with EdU for 1 hour followed by immunostaining and confocal microscopy analysis. The intensity of RAD9 foci relative to the nuclear background in mCherry-LacI and EdU+ cells was quantified (n > 120). Data are means ± SDs for (E) and (G) from biological triplicate experiments. **P < 0.01, Mann-Whitney test for (E) and (G). Scale bars, 10 μm.

Given that TOPBP1 accumulation requires K118me1 demethylation and TOPBP1-RAD9 binding, we examined whether TOPBP1 methylation has an effect on TOPBP1-RAD9 interaction. First, on coimmunoprecipitation experiments, overexpression of PHF8/wt in PHF8-deficient cells rewired the physical link between TOPBP1 and RAD9, whereas forced expression of PHF8/H247A, PHF8/∆APS, or PHF8/Y852A failed to do so (Fig. 7B). Next, RAD9 could only be efficiently immunoprecipitated by TOPBP1/wt and TOPBP1/K118R but not TOPBP1/K118A or TOPBP1/K118M in TOPBP1-deficient cells (Fig. 7C). The essentiality for K118 in TOPBP1-RAD9 binding is supported by the crystal structure of chicken TOPBP1 in complex with a phosphorylated RAD9 peptide (54). Meanwhile, coimmunoprecipitation analysis revealed that each of these mutants equally binds to PHF8 as does TOPBP1/wt, thus ruling out the idea that the binding affinity changes between TOPBP1 and RAD9 are secondary effects of altered TOPBP1-PHF8 interaction (Fig. 7C). We found that replication stress increased the interaction between TOPBP1/wt and RAD9 as previously reported (23), whereas the increased interaction with RAD9 was abolished by K118A and K118M mutation, and enhanced interaction was constitutively detected in cells expressing K118R mutant (Fig. 7C). Furthermore, we performed a pull-down assay with RAD9 peptide carrying phosphorylation on S387, which is essential for interaction with TOPBP1 (54), and cellular extracts from U2OS cells expressing BRCT 0-2 or full-length TOPBP1 variants. Only K118R but not K118A or K118M mutant could be effectively pulled down (Fig. 7D). These results suggest that K118 methylation plays an important role in TOPBP1-RAD9 binding.

Because TOPBP1 is also involved in loading RAD9 to chromatin at stalled replication forks, which is conserved from frogs to humans (55, 56), we wondered whether disruption of PHF8-promoted TOPBP1 loading affects the accumulation of RAD9 on damaged replication forks. PHF8 depletion also markedly impaired RAD9 foci formation, which could be reversed by PHF8/wt but not PHF8 mutants including PHF8/H247A, PHF8/∆APS, or PHF8/Y852A (Fig. 7E). The same was true on pull-down assays with RPA-ssDNA/dsDNA complexes with nuclear extracts (Fig. 7F). Consistently, forced expression of TOPBP1/wt or TOPBP1/K118R but not TOPBP1/K118A or TOPBP1/K118M could efficiently restore RAD9 foci formation in TOPBP1-deficient cells (Fig. 7G). These results suggest that TOPBP1 K118 mono-methylation serves as a brake to control TOPBP1 binding to RAD9 and thus the loading of these two molecules onto damaged chromatin.

Besides RAD9, TopBP1-interacting replication-stimulating protein (Treslin), RAD9-HUS1-RAD1 interacting nuclear orphan 1 (RHNO1), and mediator of DNA damage checkpoint 1 (MDC1) have all been shown to interact with the BRCT1 domain of TOPBP1. To be specific, Treslin acts in collaboration with TOPBP1 to facilitate the loading of cell division cycle protein 45 (CDC45) onto chromatin thus replication initiation (23, 47, 57). Recently, it has been reported that Treslin can directly participate in ATR-mediated checkpoint signaling, and this also requires the phosphorylation-steered TOPBP1-Treslin binding (58). Since PHF8 knockout had a minor effect on the association of TOPBP1 with Treslin (fig. S6C) and the loading of TOPBP1 as well as replication elongation factor PCNA, which acts downstream of TOPBP1-Treslin (47), on newly replicating chromatin, it seems unlikely that PHF8 is involved in Treslin-TOPBP1 controlled replication initiation or checkpoint activation. RHNO1 independently binds TOPBP1 and the 9-1-1 complex, and it functions together with the TOPBP1/9-1-1 complex to activate ATR (59, 60), while, unlike PHF8, loss of RHNO1 does not affect the interaction of the 9-1-1 clamp with TOPBP1 or the loading of 9-1-1 onto chromatin (59). MDC1 is initially characterized as a reader of γH2AX and an ATR activation-related G2-M checkpoint protein (61), whereas Leimbacher et al. (62) reported that the MDC1-TOPBP1 complex specifically maintains chromosomal stability during mitosis and the ATR pathway is normally activated in MDC1-knockout cells. Yet, it remains to be further assessed whether the interaction of TOPBP1 with these ligands plays a role in PHF8-promoted ATR activation and replication stress response.

ATR-catalyzed S854 phosphorylation destabilizes the PHF8-TOPBP1 complex

In response to DNA damage, the assembly/disassembly of protein complexes is highly dynamic and spatiotemporally regulated, largely directed by posttranslational modifications such as phosphorylation. We found that replication stress increased TOPBP1-RAD9 binding but decreased the association of TOPBP1 with PHF8 (Fig. 7C), so demethylated TOPBP1 may switch from the PHF8-TOPBP1 to the RAD9-TOPBP1 complex. To understand how TOPBP1 is dissociated from PHF8, we reviewed the crystal structure of BRCT 7-8 and PHF8 APS motif and focused on the phosphor-potential serine residue S854. MS revealed that S854 is phosphorylated upon HU treatment (fig. S7A). Quantitative characterization of the interaction between recombinant BRCT 7-8 and native APS peptide or APS peptide carrying S854 or S857 phosphorylation by using ITC titration demonstrated that pS854 but not pS857 markedly decreased the affinity of the peptide-protein binding (Fig. 8A). Next, we created S854D and S854E mutants to mimic S854 phosphorylation and showed that both S854D and S854E greatly disrupted PHF8 binding to TOPBP1 (Fig. 8, B and C). In contrast, S857D or S857E mutation did not affect PHF8-TOPBP1 binding (Fig. 8B). Consistently, PHF8/S854D or PHF8/S854E behaved similar to PHF8/Y852A in regulating ATR activity and the cellular response to replication stress in PHF8-deficient cells (Fig. 8D). These results suggest that pS854 may act to destabilize the PHF8-TOPBP1 complex.

Fig. 8 S854 phosphorylation impairs the association of PHF8 with TOPBP1 upon replication stress.

(A) ITC results of APS peptide with different phosphorylated serine residues titrated to BRCT 7-8. (B) Coimmunoprecipitation analysis with cellular extracts from HeLa cells expressing FLAG-PHF8 variants. (C) U2OS-LacO cells were transfected with mCherry-LacI-TOPBP1 and PHF8 phospho-mimetic variants followed by immunostaining and quantification (n > 165). (D) ATR kinase activity and cell survival analysis in U2OS cells expressing PHF8 3′UTR siRNA and PHF8 variants. (E) U2OS cells expressing FLAG-PHF8 were treated with mock, HU, or CPT followed by coimmunoprecipitation analysis. (F) In vitro kinase assay with APS peptide and recombinant ATR purified from HeLa cells with high-salt and -detergent buffer in the absence or presence of VE-821 followed by MALDI-TOF MS analysis. (G) U2OS cells expressing FLAG-PHF8 were treated with CPT in the absence or presence of VE-821 followed by coimmunoprecipitation analysis. The relative intensity of pS854 to immunoprecipitated PHF8 was normalized against that in dimethyl sulfoxide–treated cells. (H) U2OS-LacO cells were transfected with PHF8 3′UTR siRNA, mCherry-LacI, and FLAG-TOPBP1/K118R followed by VE-821 treatment, immunostaining, and quantification (n > 120). (I) Working model. The color change of ATR indicates stimulation of the ATR kinase from basal to full activity by TOPBP1. Data are means ± SDs for (C), (D), and (H) from biological triplicate experiments. **P < 0.01, Mann-Whitney test for (C) and (H); two-way ANOVA for (D); one-way ANOVA for (G). Scale bars, 10 μm.

Next, we developed an antibody that specifically recognizes pS854 (fig. S7, B and C). Consistent with results from MS, the level of S854 phosphorylation was up-regulated with HU or CPT treatment (Fig. 8E). Using the online kinase prediction tool of PHOSPHONET, we found that S854 is a potential phosphor-target of ATR kinase. In vitro kinase assay with recombinant ATR and APS peptide followed by MS confirmed that S854 is phosphorylated by ATR (Fig. 8F). Consistently, ATR inhibitor treatment impaired S854 phosphorylation in vitro (Fig. 8F and fig. S7D) and in vivo (Fig. 8G). Meanwhile, replication stress–induced TOPBP1 dissociation from PHF8 was efficiently reversed by the ATR inhibitor (Fig. 8G), so ATR-catalyzed PHF8 phosphorylation blocks its association with TOPBP1. Furthermore, we found that the addition of ATR inhibitor, as a consequence of preventing PHF8-TOPBP1 separation, greatly impaired TOPBP1 foci formation (fig. S7E). To differentiate the contributions of PHF8 demethylase activity and PHF8-TOPBP1 binding to the engagement of TOPBP1 on replication stress sites, the foci intensity of TOPBP1/K118R (hypomethylation-mimicking mutant), which could be efficiently engaged to replication stress sites in PHF8-deficient cells (Fig. 6B and fig. S5E) and acts similar to TOPBP1/wt in PHF8 binding theme (Fig. 7C), was monitored with the LacO-LacI targeting system. The results showed that ATR inhibition reduced the accumulation of TOPBP1/K118R at a perturbed replication fork, while this effect was markedly suppressed in VE-821–treated PHF8-depleted cells (Fig. 8H). By contrast, RPA deposition was not affected in each circumstance (Fig. 8H). In addition, we proved that ATR is activated at the LacO-LacI sites, where RPA2 S33 is phosphorylated in an ATR kinase activity–dependent manner, while ATR inhibition does not alter the distribution of PHF8 at this locus (fig. S7F). Collectively, these results indicate that PHF8 could prevent TOPBP1 unscheduled loading through physical sequestration, and ATR-stimulated TOPBP1 release from the PHF8-TOPBP1 complex is important for TOPBP1 recruitment to replication stress sites. This may coordinate with PHF8 demethylase activity to control TOPBP1-associated protein complex formation and damaged chromatin engagement. Since ATR is able to autophosphorylate and is basally activated in the absence of TOPBP1 (11, 14, 63), we propose that, when replication stress occurs, ATR may take advantage of its basal activity to phosphorylate damaged chromatin-adjacent PHF8 likely in a hit-and-run manner, resulting in the dissociation of demethylated TOPBP1 from PHF8, and in turn, TOPBP1 efficiently forms a complex with 9-1-1 to load on stalled/collapsed replication forks and fully activate ATR (Fig. 8I).


Here, we report that histone demethylase PHF8 is critically required for ATR activation and cellular response to ATR-activating DNA lesions. With a combined approach of structural and biochemical analysis, we unveiled that PHF8 is physically associated with TOPBP1. Mechanistically, we revealed that PHF8 removes TOPBP1 K118 mono-methylation to facilitate the association of TOPBP1 with RAD9, thus loading the TOPBP1-RAD9 complex to damaged chromatin. We found that ATR-catalyzed phosphorylation of PHF8 at S854 does not favor but rather disfavors PHF8-TOPBP1 binding upon replication stress. In this manner, TOPBP1 is uncoupled from phosphorylated PHF8, and hypomethylated TOPBP1 is deposited onto the perturbed replication forks to fully activate ATR and maintain genome stability.

As a “scaffold” or “hub” protein, TOPBP1 is believed to organize numerous protein complex formations generally through distinct paired BRCT domains with phospho-specific binding features for the ligand proteins (64). This allows for dynamic assembly of different proteins into TOPBP1-nucleated complexes, controlling cellular processes such as DDR and DSB repair (6). For instance, BRCT 1-2 interacts with RAD9 at sites of replication stress (54) and BRCT 4-5 binds 53BP1 at DSBs (51). In our study, in the absence of exogenous replication stress, TOPBP1 and PHF8 formed a constitutive protein complex with high affinity via BRCT 7-8 and an APS of PHF8. In contrast to the current understanding of the BRCT binding properties, the interaction in our study was independent of posttranslational modifications. Mutational analysis identified Y850 and Y852 as critical residues for the formation of the PHF8-TOPBP1 complex, while C844, D847, or S854 also plays a role. Notably, phosphorylated S854 did not strengthen as reported currently (65), but rather abolished the association of APS with BRCT 7-8 as shown by ITC analysis with phosphopeptide and compromised the interaction between PHF8 and TOPBP1 as shown by coimmunoprecipitation analysis when S854 was mutated to aspartate or glutamate. These evidence suggest that the APS motif is engaged in BRCT 7-8 with a unique association/dissociation mode, and this extraordinary binding feature defines PHF8 as a special regulator of TOPBP1. We believe that these findings will broaden our insights into the understanding of the ligand-binding specificity as well as diversity of BRCT domain-containing proteins.

TOPBP1 is known to take advantage of distinct BRCT combinations to interact with multiple ligand proteins (64), but the dynamic regulation of TOPBP1-scaffolded complexes has not been well studied. A single TOPBP1 molecule, for instance, could at least, in principle, simultaneously bind RAD9, Treslin, RHNO1, or MDC1 via BRCT 1-2 (54, 58, 59, 62), BLM or 53BP1 via BRCT 4-5 (51, 66), and PHF8 or BACH1 via BRCT 7-8 (55). Yet, the steric constraints imposed by the remaining ligand proteins might prevent this combination. Conversely, for a given pair of BRCTs, the differences in affinity for binding to differential ligands potentially rules out their coexistence within the same TOPBP1-scaffolded complex. This property, together with the phosphorylation status of individual ligand sites, steering by cell cycle and/or DDR system (62, 67), may enable TOPBP1 to hierarchically engage with distinct binding proteins. We uncovered a bivalent regulatory mode for PHF8 in preparation for TOPBP1 switching ligands and loading: The C-terminal acidic patch motif is responsible for the physical association of PHF8 with BRCT 7-8 and the N-terminal JmjC domain removes K118me1 from BRCT 0-2. This finding suggests that an intramolecular communication from the tail to head of TOPBP1 is physically and functionally linked by the corresponding region of PHF8. In support of the hierarchy model of TOPBP1-nucleated protein complexes, TOPBP1 was dissociated from PHF8 when PHF8 S854 underwent phosphorylation upon replication stress, and the demethylated TOPBP1, in turn, was loaded onto ssDNA/dsDNA junctions by binding the 9-1-1 complex via phosphorylated RAD9. Although more molecular details are still needed to understand how K118me1 tunes the binding of TOPBP1 to RAD9, we provide a mechanism for the dynamic interplay of TOPBP1-scaffolded complexes and reveal that S854 phosphorylation and K118 mono-methylation act in concert for the exchange and selection of TOPBP1 ligand proteins.

From the crystal structure of RAD9 peptide–bound TOPBP1 (54), it clearly shows that K126 is distant from the BRCT1 binding interface and it is not surprising that alanine substitution does not alter TOPBP1 loading, while K154 forms part of a pocket for the recognition of the RAD9 sequence. However, compared to K118 of TOPBP1, which is closer to the RAD9 peptide and forms proper polar interaction with pS387 of RAD9, K154 is not close enough to form hydrogen bonds with residues on RAD9. Instead, K154 folds back to TOPBP1 itself stabilizing V158 to maintain a proper conformation of BRCT1. The importance of K154 for TOPBP1 recruitment is supported by the observation that K154A impaired TOPBP1 foci formation, whereas arginine substitution had a minor effect. This is consistent with the understanding that K154A prevents TOPBP1 interaction with RAD9, MDC1, and the MRN complex (45, 62, 66, 68). However, the observations that the K154R variant could not efficiently mobilize to perturbed replication forks in PHF8-knockout cells and PHF8 loss did not change NBS1-TOPBP1 binding indicate that K154 dimethyl should not be a functionally relevant substrate of PHF8, or the methyl mark does not alter the property of K154 itself in the binding theme.

Neither the LacO-LacI targeting system nor iPOND analysis showed evident accumulation of PHF8 at DNA damage sites of replication stress, implying that PHF8 may act unlike the canonical regulators of ATR signaling. However, we could not exclude the possibility that PHF8 might be recruited to blocked/collapsed replication forks in a dynamic and transient manner. We showed that replication stress resulted in an elevated level of S854 phosphorylation and, thus, decreased PHF8-TOPBP1 binding, and these effects were reverted by ATR inhibition. Because ATR is activated by its physical recruitment to RPA-coated ssDNA, albeit not fully (11, 14, 63), we propose that PHF8 may hand over hypomethylated TOPBP1 to the proximal territories of damaged chromatin sites of replication stress, where basally activated ATR phosphorylates PHF8, likely in a hit-and-run manner, and thus contributes to TOPBP1 release from the PHF8-TOPBP1 complex. Although we cannot directly test this hypothesis, it is supported by the evidence that ATR inhibition–induced uncoupling of PHF8-TOPBP1 impairs damaged chromatin loading of TOPBP1, which unlikely takes place after ATR is fully activated by chromatin-engaged TOPBP1. If the phosphorylation-steered PHF8-TOPBP1 dissociation event occurs after ATR is fully activated and PHF8 is required for ATR activation, ATR should paradoxically prevent its own activation. In this process, posttranslational modifications or other interaction partners might function to limit the accessibility of countering methyltransferases to TOPBP1, thus preventing TOPBP1 methylation after it disassociates from PHF8.

Although there was an increase of new origin firing after PHF8 loss upon HU treatment, the intensity of RPA foci was not increased. This seems to be in contrast to the understanding of ATR’s activity in preventing global exhaustion of RPA and excessive DSB formation (5). Considering that PHF8 loss may compromise the resection of DSBs that are derived from collapsed forks and thus reduce RPA foci formation (37), which was indeed found in PHF8-deficient cells after CPT treatment, we envisioned that this bi-functional activity of PHF8 likely influences the foci formation of RPA in HU-treated PHF8-deficient cells in a combinatorial manner. Consistent with the increased ssDNA generated by ATR inactivation (5), iPOND analysis revealed that the amount of RPA1 captured with EdU-labeled nascent ssDNA moderately increased in PHF8-knockout cells. However, RPA deposition is, at least, not locally controlled by PHF8 as the LacO-LacI targeting system showed that PHF8 deficiency had a marginal effect on RPA loading or its downstream effector ATRIP recruitment at a site-specific replication perturbed region. Although PHF8 deficiency led to mildly reduced RPA foci formation when cells were challenged by CPT, normalization of the intensity of RPA2 pS33 against that of RPA2 demonstrated that PHF8 depletion greatly impaired RPA2 S33 phosphorylation, suggesting that PHF8 is capable of promoting ATR activation in the absence of DNA end resection. This argument is also supported by observations from in vitro assays with ssDNA/dsDNA hybrid.

Because PHF8 is a well-characterized histone demethylase, we needed to differentiate whether PHF8-promoted ATR activation is a secondary effect of PHF8-catalyzed histone demethylation. When mapping the interface required for TOPBP1 and PHF8 binding, we found that the PHF8/Y852A mutant almost completely failed to bind TOPBP1. It was not able to restore PHF8 depletion–associated defects, including the failure of TOPBP1 foci formation, ATR inactivation, and cellular hypersensitivity. However, it was still able to remove histone methyl marks as efficiently as PHF8/wt. Thus, we defined this variant as a separation-of-function mutant of PHF8 and concluded that the role of PHF8 in activating the checkpoint and protecting cells against replication stress is independent of its histone demethylase activity but could largely be attributed to its involvement in the TOPBP1-RAD9 axis. The importance of this functional link is further warranted by the observation that in PHF8-depleted cells, expression of only the TOPBP1/K118R mutant but not the K118A or K118M variant or TOPBP1/wt was able to efficiently load on damaged DNA and ensure ATR activation. Methylated TOPBP1, detectable already under unperturbed conditions, may prevent the unscheduled TOPBP1 redistribution. Whether other histone (de)methylases that seem to affect RPA pS33 level in our screening, albeit not markedly, are responsible for TOPBP1 (de)methylation or ATR activation needs to be further investigated.

It has been reported that PHF8 controls TOPBP1 stability, in which PHF8 binding to TOPBP1 may protect TOPBP1 from degradation (65). However, we did not see evident alterations of TOPBP1 expression with several approaches, including knockdown or knockout of PHF8 and interrupting PHF8-TOPBP1 binding. It is possible that CKII could phosphorylate PHF8 under normal conditions (65), while in our model, ATR should be the relevant kinase that catalyzes PHF8 phosphorylation upon replication stress. Here, we propose that PHF8 interacts with and demethylates TOPBP1 to prepare TOPBP1 in a competent state for damaged chromatin binding, and replication stress unleashes the demethylated TOPBP1 to enhance its association with the 9-1-1 complex at ssDNA/dsDNA junctions. In this manner, PHF8-promoted TOPBP1 demethylation acts as a molecular switch to finely tune TOPBP1 loading and enable ATR-nucleated DDR machinery to respond properly and promptly to replication stress. ATR has been proven as a particularly attractive target for anticancer therapy, as tumor cells with high burden of replication stress rely heavily on ATR for survival (6, 9). Since both PHF8 and TOPBP1 have been implicated in tumorigenesis (31, 37, 49), our study suggests the potential interest of targeting PHF8 demethylase activity or PHF8-TOPBP1 binding surface for cancer intervention.



Cell lysates were prepared by incubating the cells in NETN buffer [50 mM tris-HCl (pH 8.0), 150 mM NaCl, 0.2% Nonidet P-40, and 2 mM EDTA] in the presence of protease inhibitor cocktails (Roche) for 20 min at 4°C. This was followed by centrifugation at 14,000g for 15 min at 4°C. For immunoprecipitation, about 500 μg of protein was incubated with control or specific antibodies (2 μg) for 12 hours at 4°C with constant rotation; 50 μl of 50% protein G magnetic beads (Invitrogen) was then added and the incubation was continued for an additional 2 hours. Beads were then washed five times using the lysis buffer. The precipitated proteins were boiled in 2× SDS–polyacrylamide gel electrophoresis (PAGE) loading buffer and subjected to SDS-PAGE followed by immunoblotting with appropriate antibodies. To avoid nucleic acid contamination, we used EDTA-free lysis buffer containing DNase I (4 U/ml), RNase A (20 μg/ml), or Benzonase (0.5 U/ml) to prepare the cell lysates at room temperature for 30 min. The reaction was stopped by EDTA (5 mM).

Stable isotope labeling with amino acids in cell culture

Control cells were labeled with “heavy isotopic lysine and arginine” (K6R10), and the doxycycline (DOX) treatment PHF8-expressing cells were labeled with “light isotopic lysine and arginine” (K0R0) using a SILAC Protein Quantitation Kit (Pierce, Thermo Fisher Scientific) according to the manufacturer’s instructions. After labeling, cells were harvested in NETN buffer in the presence of protease inhibitor cocktails (Roche) for 30 min at 4°C. Cellular extracts from an equal amount of cells (about 1 × 108 cells) under each treatment were applied to an equilibrated FLAG column of 1-ml bed volume to allow for adsorption of the protein complex to the column resin. After binding, the column was washed with cold phosphate-buffered saline (PBS) plus 0.2% Nonidet P-40. FLAG peptide was applied to the column to elute the FLAG-PHF8–containing protein complex as described by the vendor. The eluents were subjected to NuPAGE 4 to 12% bis-tris gel (Invitrogen) until all proteins were fully loaded onto the gel, and samples from different labeling were collected from the gel. The gel pieces were mixed and digested with trypsin followed by standard liquid chromatography–tandem MS (MS/MS) analysis. The resulting MS/MS data were processed using Proteome Discoverer 1.3. MS/MS were searched against the SwissProt human database. Trypsin/P was specified as a cleavage enzyme allowing up to two missing cleavages. The mass error was set to 10 parts per million for precursor ions and 0.02 Da for fragment ions. Peptide confidence was set at high, and peptide ion score was set >20. The fold change of protein enrichment was calculated from the ratio of protein intensity (sum of corresponding unique peptide intensity) between light-labeled and heavy-labeled samples. Two biological replicate experiments were performed, and detailed results are shown in table S1.

Protein expression and purification

A cDNA fragment encoding the BRCT 7-8 domain (amino acids 1264 to 1493) of TOPBP1 was cloned into a modified pET-28a-smt vector. The his-sumo–tagged protein was expressed in the BL21 (DE3) strain of Escherichia coli by induction with 0.15 mM isopropyl-β-d-thiogalactopyranoside at 16°C overnight. Cells were harvested by centrifugation and sonicated in buffer A [25 mM tris-HCl (pH 8.0), 500 mM NaCl, 10 mM imidazole, 1 mM EDTA, and 1 mM β-mercaptoethanol (β-ME)]. Cell debris was removed by centrifugation at 40,000g for 40 min at 4°C. The supernatant was loaded onto a Ni Sepharose Excel Column (GE Healthcare) followed by washing the beads with buffer A. The fusion protein was eluted with buffer B [25 mM tris-HCl (pH 8.0), 500 mM NaCl, 200 mM imidazole, 1 mM EDTA, and 1 mM β-ME] and cleaved by sumo protease to remove the sumo-tag. The tag-free BRCT 7-8 was then loaded onto a HiTrap SP HP column (GE Healthcare) pre-equilibrated with buffer C [20 mM MES (pH 6.0), 100 mM NaCl, 1 mM EDTA, and 1 mM β-ME] and eluted with a linear gradient of 0.1 to 1.0 M NaCl. The eluted protein was concentrated by ultrafiltration and further purified by using a HiLoad Superdex 200 16/60 size-exclusion column (GE Healthcare) in a buffer containing 20 mM MES (pH 6.0), 300 mM NaCl, and 1 mM β-ME. High-purity fractions were pooled and concentrated to ~40 mg/ml and stored in PBS buffer for crystallization. Expression and purification procedures for mutant proteins were the same as those for the wild-type protein.


The purified BRCT 7-8 was incubated with the PHF8-APS peptide (amino acids 842 to 863, GACFKDAEYIYPSLESDDDDPA, SciLight Biotechnology) at 1:3 molar ratio for co-crystallization. The crystals were grown at 16°C by sitting-drop vapor diffusion method by mixing 2 μl of protein solution with 1 μl of reservoir solution containing 0.4 M sodium phosphate monobasic monohydrate and 1.6 M potassium phosphate dibasic. The crystals were observed after 3 to 4 days and flash-frozen in liquid nitrogen by using a cryo-protectant prepared by adding 10% glycerol to the reservoir solution.

Data collection and structure determination

X-ray diffraction data were collected with the beamline BL19U1 at the Shanghai Synchrotron Radiation Facility (SSRF, Shanghai) by using a Pilatus3 6M detector at wavelength 0.9793 Å. Datasets were integrated, scaled, and merged by using the HKL2000 suite. The complex structure was solved by molecular replacement with Phaser-MR using the structure of BRCT 7-8 (PDB ID: 3AL2) as the search model. The remaining model was manually built by using Coot and refined with phenix.refine in the PHENIX suite. Detailed statistics for data collection and refinement are shown in table S2. Structure figures were prepared by using PyMOL ( and Coot, and the coordinates were deposited in PDB under ID 7CMZ.

Isothermal titration calorimetry

ITC measurements were performed using a MicroCal PEAQ-ITC instrument (Malvern Instruments, Malvern, UK) at 25°C in a buffer containing 50 mM Hepes and 150 mM NaCl. For each ITC measurement, 1 × 0.5 μl followed by 19 × 2 μl of APS peptide (600 μM) were titrated into a 250-μl BRCT 7-8 protein sample. Each sample was measured at least three times. The data were analyzed by using MicroCal PEAQ-ITC Analysis Software, and the one with the best fittings is presented. The sequences of peptides used in ITC experiments are provided in table S3.

Surface plasmon resonance

The SPR experiment was performed using a BIACORE T200 instrument (GE Healthcare). Biotin-labeled APS peptides (APS GACFKDAEYIYPSLESDDDDPA-GGK-biotin) or APS variants (APS-N21 YWRTESEEEEENASLDEQDSL-GGK-biotin; APS-C22 LKSRPKKKKNSDDAPWSPKARV-GGK-biotin) were immobilized on a CM5 sensor chip captured with streptavidin via standard amine-coupling chemistry, and a blank channel was set as a negative control. Proteins used in SPR were dialyzed against HBS-EP+ buffer [10 mM Hepes (pH 7.4), 150 mM NaCl, 3.4 mM EDTA, and 0.05% v/v surfactant P20] before. All experiments were performed in HBS-EP+ buffer at 25°C. The experimental data were analyzed and fitted by a simple 1:1 interaction model by using the Biacore T200 Evaluation Software v3.2 (GE Healthcare).


Cells on glass coverslips (BD Biosciences) were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100 in PBS. Samples were blocked in 5% donkey serum in the presence of 0.1% Triton X-100 and stained with the appropriate primary and secondary antibodies coupled to Alexa Fluor 488, 594, or 647 (Invitrogen). Confocal images were captured on a Zeiss LSM 800 microscope with a ×63 oil objective. To avoid bleed-through effects in double-staining experiments, each dye was scanned independently in a multitracking mode. During inspection of nuclear wide dispersed RPA or TOPBP1 foci, cells were pretreated with 0.5% Triton X-100 for 5 min on ice to extract nonchromatin fractions and fixed with 3% paraformaldehyde and 2% sucrose for 15 min at room temperature. Cells were then permeabilized with 0.5% Triton X-100 for 5 min on ice and then incubated in blocking buffer (0.1% Triton X-100, 5% donkey serum in PBS) for 1 hour at room temperature. For S phase discrimination, U2OS-LacO cells were pulsed with 10 μM EdU at 37°C for 1 hour before fixation. Incorporated EdU was click-labeled by using keyFluor 647-azide (Keygen Technologies) according to the manufacturer’s instructions.

RPA-ssDNA/dsDNA pull-down assay

To generate 3′ overhang dsDNA, biotinylated DNA oligomers (5′-AACCTGTCGTGCCAGCTGCA-biotin-3′) were first annealed to complementary ssDNA) (5′-TGCAGCTGGCACGACAGGTTTTAATGAATCGGCCAACGCGCGGGGAGAGGCGGTTTGCGTATTGGGCGCTCTTCCGCTTCGCAGCGAGTC-3′) with molar ratio 1:4. The resulting DNA structures (2 pmol) were attached to streptavidin-MyOne T1 beads (Thermo Fisher Scientific; prewashed three times with PBS) according to the manufacturer’s instruction followed by extensive washing with binding buffer [10 mM tris-HCl (pH 7.5), 100 mM NaCl, 10% glycerol, 0.01% NP-40, and bovine serum albumin (BSA) 10 microgram/milliliter] and then coated with recombinant RPA complexes purified from E. coli. Excess amounts of RPA (approximately 2 μg) were used to ensure that the ssDNA was saturated with RPA. The beads carrying RPA-ssDNA/dsDNA complexes were incubated in nuclear extracts in the binding buffer. After incubation at room temperature for 30 min, beads were retrieved and washed three times with the binding buffer. The proteins bound to beads were denatured in 2× SDS-PAGE loading buffer, separated on SDS-PAGE, and analyzed by immunoblotting.

Isolation of proteins on nascent DNA

In brief, 2 × 108 to 3 × 108 cells were labeled with 10 μM EdU (Thermo Fisher Scientific) for 10 min to detect nascent forks. For stalled forks, cells were treated with HU for 2 hours in the continued presence of EdU. The harvested cells were then fixed with 1% formaldehyde in PBS solution for 20 min at room temperature followed by quenching of the cross-linking reaction with 1.25 M glycine. Cells were then harvested and incubated in a permeabilization buffer (0.25% Triton X-100/PBS) at room temperature for 30 min. Then, cells were washed at 4°C with 0.5% bovine serum albumin (BSA)/PBS and then PBS alone, followed by incubation in the “click” (10 mM sodium ascorbate, 2 mM CuSO4, and 10 mM biotin-azide, in PBS) or “no-click” (i.e., no biotin-azide) reaction cocktail for 2 hours at room temperature. After the reaction, cells were re-suspended in lysis buffer (1% SDS and 50 mM tris-HCl, pH 8.0) containing protease inhibitor (Roche) and cell lysates were sonicated five times by using a Bioruptor (Diagenode) at 4°C. After centrifugation, EdU-labeled DNA was immunoprecipitated from supernatants by incubation with streptavidin-MyOne T1 beads (Thermo Fisher Scientific; prewashed three times with PBS) for 4 hours at 4°C. The streptavidin-agarose beads containing the captured DNA-protein complexes were then centrifuged for 3 min at 1800g. After washing for 5 min each with 1 ml of cold lysis buffer, 1 ml of 1 M NaCl, and twice more with 1 ml of lysis buffer, the beads were supplemented 1:1 (v/v of packed beads) with 2× SDS-PAGE loading buffer and incubated at 95°C for 25 min to liberate the proteins. SDS-PAGE fractionation and immunoblotting were then performed.

DNA fiber assay

To check fork symmetry and fork speed, cells were first labeled with 5-iodo-2′-deoxyuridine (IdU) (25 μM) for 30 min, washed twice with media, and labeled with 5-chloro-2′-deoxyuridine CldU (200 μM) for 30 min. Cells were then trypsinized and re-suspended in PBS to 7 × 105 cells/ml. Then, 2 μl of cells was mixed with 10 μl of lysis buffer [200 mM tris-HCl (pH 7.4), 50 mM EDTA, and 0.5% SDS] on a clean glass slide. After 2-min incubation, the slides were tilted at 15° to horizontal, allowing the lysate to slowly flow down along the slide. The slides were then air-dried, fixed in 3:1 methanol/acetic acid, and treated with 2.5 M HCl for 80 min. The slides were then blocked in blocking buffer (5% BSA in PBS) for 30 min and incubated with anti–5-bromo-2′-deoxyuridine (BrdU) antibodies (BD Bioscience, 347580 against IdU, and AbD Serotec, MCA2060GA against CldU) in blocking buffer overnight. After washing, secondary antibodies coupled to Alexa Fluor 488 and 594 were diluted in PBS containing 5% BSA and incubated with cells at room temperature for 1 hour. The slides were then washed three times with PBS. After washing, cells were mounted with an anti-fade solution and visualized under a Zeiss LSM 800 fluorescence microscope. The length of all discrete fibers was measured by using ImageJ software. Fork symmetry was analyzed by measuring the length of CldU fiber (red) on each side. Fork speed was analyzed by measuring the length of CldU fiber based on an average fork elongation rate (1 μm roughly corresponds to 2 kb for each fiber). New origin firings were measured by counting only red fibers and compared to total fiber numbers. For examining restart efficiency of stalled replication forks, cells were labeled with IdU (25 μM) for 30 min and then washed twice with media. After washing, cells were treated with HU (2 mM) for 4 hours. Cells were then recovered in fresh medium with CldU (200 μM) for 30 min. Then, fork length and fork events under stress conditions were examined by BrdU staining followed by microscopy and statistical analysis.


Data from biological triplicate or duplicate experiments are presented as means ± SDs. Two-tailed unpaired Student’s t test was used for comparing two groups of data. Analysis of variance (ANOVA) with Bonferroni’s correction was used to compare multiple groups of data. For values not normally distributed, Mann-Whitney U test was used. P < 0.05 was considered statistically significant. All statistical analyses involved using Statistical Product and Service Solutions. Before statistical analysis, variation within each group of data and the assumptions of the tests were checked.

Study approval

All procedures involving animals were approved by the Ethics Committee of the Tianjin Medical University and followed the NIH Guide for the Care and Use of Laboratory Animals (8th ed., The National Academies Press, 2011).


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: This work was supported by grants (2019YFA0802003 to Y.Z., 81972660 and 81722036 to L.S., 2019YFA0508902 and 2018YFA0107004 to N.Y., 81502408 to L.L., and 82003004 to S.M.) from the National Natural Science Foundation of China and Chinese Ministry of Science and Technology, a grant (to L.S.) from Excellent Talent Project of Tianjin Medical University, and Tianjin Funds for Distinguished Young Scientists (17JCJQJC46100 to L.S. and 17JCJQJC45900 to N.Y.). We thank D. Ai (Tianjin Medical University) for critical reading of the manuscript and for giving valuable suggestions on experimental design and data presentation. We thank H. Yang (Shanghai Tech University) and C. Cui (Tianjin University) for providing help on the purification of BRCT 7-8 and pull-down experiments. We thank Y. Yang (Core Facilities for Molecular Biology, Capital Medical University) and X. Guan (State Key Laboratory of Medicinal Chemical Biology, Nankai University) for providing help on ITC and SPR experiments. Author contributions: S.M., C.C., S.C., Y.W., D.S., Z.Y., N.Y., and L.S. conceived this project; S.M., C.C., S.C., Y.W., D.S., S.L., W.G., L.L., J.S., J.Z., Q.W., N.S., T.G., Q.G., S.T., and X.D. conducted experiments; S.M., C.C., S.C., Y.W., D.S., J.S., S.T., and K.Z. acquired data; S.M., C.C., S.C., Y.W., D.S., S.T., K.Z., X.D., F.Y., Z.Y., N.Y., and L.S. analyzed data; C.D.C., T.Z., J.W., G.Y., J.Y., K.Z., and Y.Z. provided technical supports and reagents; S.M., C.C., S.C., Y.W., D.S., Z.Y., N.Y., and L.S. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article