CSB cooperates with SMARCAL1 to maintain telomere stability in ALT cells

The nature of repetitive G-rich sequences represents an endogenous barrier for DNA replication fork progression at telomeres. If not resolved, this replication stress can drive telomere fragility, which manifests as multiple telomere signals per chromatid end seen in metaphase cells (Martinez et al., 2009; McKerlie et al., 2012; Sfeir et al., 2009). About 10-15% of human cancers rely on a homologous recombination (HR)-based mechanism, referred to as alternative lengthening of telomeres (ALT), to maintain their telomeres (Cesare and Reddel, 2010). In ALT cells, telomere transcription is deregulated and this deregulation can lead to enhanced collisions between replication and transcription complexes (Azzalin et al., 2007; Porro et al., 2010; Wilson et al., 2016), further contributing to replication stress at telomeres. Therefore, ALT telomeres represent a good experimental system to investigate genetic interactions that cancer cells rely on to resolve replication stress at their telomeres in order to support their continued growth and proliferation.

ALT cancer cells are characterized by several distinct hallmarks, including telomere length heterogeneity, a high level of extra-chromosomal telomeric DNA in the form of double-stranded linear/circular DNA or single-stranded circular DNA such as C-circles, elevated rates of telomeric sister chromatid exchange (T-SCEs) as well as PML bodies containing telomeric chromatin, referred to as ALT-associated PML bodies (APBs) (Henson and Reddel, 2010). These hallmarks are not always present together in ALT cells, indicative of the complexity of mechanisms that give rise to these features. Replication stress at telomeres is reported to trigger the processes leading to these hallmarks (Cox et al., 2016; Poole et al., 2015; Zhang et al., 2019b). DNA double-strand breaks (DSBs) induced at telomeres also drive ALT hallmarks (Dilley et al., 2016). Recently it has been reported that break-induced replication (BIR), a noncanonical form of homology-directed repair, drives telomere synthesis and promotes ALT hallmarks (Dilley et al., 2016; Zhang et al., 2019a,b). Not surprisingly, APBs are enriched with many proteins involved in DNA replication, recombination and repair such as DNA polymerase POLD3 (Dilley et al., 2016), an essential component of BIR, BLM (Stavropoulos et al., 2002), the MRE11/RAD51/NBS1 (MRN) complex (Wu et al., 2003, 2000), BRCA1, RAD51, RAD52 and RPA (Yeager et al., 1999).

SMARCAL1 is a chromatin remodeler that catalyzes branch migration and regression of stalled replication forks to protect stalled forks from collapsing (Bansbach et al., 2009; Betous et al., 2012; Yuan et al., 2009). SMARCAL1 is reported to resolve replication stress at telomeres in ALT cells (Cox et al., 2016; Poole et al., 2015). It has been reported that depletion of SMARCAL1 leads to persistently stalled replication forks at ALT telomeres (Cox et al., 2016), which are subject to cleavage by the structure-specific endonuclease MUS81-EME1-SLX1-SLX4 (MUS-SLX) complex (Fekairi et al., 2009; Hanada et al., 2007; Wyatt et al., 2013). It has also been reported that DSBs arising from collapsed forks as a result of cleavage by MUS-SLX are channelled into homology-directed BIR repair (Costantino et al., 2014; Roumelioti et al., 2016).

Cockayne syndrome group B (CSB) protein is encoded by the ERCC6 gene, mutations of which account for the majority of cases of hereditary and devastating Cockayne syndrome (CS). CS is characterized by severe photosensitivity, physical impairment, neurological degeneration, muscle degeneration and premature aging (Laugel et al., 2009). Although CSB was first described to be involved in transcription-coupled nucleotide excision repair (Troelstra et al., 1992), many studies over the last few decades have revealed that CSB is a multifunctional protein that participates in a wide range of cellular processes. CSB is recruited to DSBs to regulate DSB repair pathway choice (Batenburg et al., 2015, 2017). At DSBs, CSB evicts histones and limits RIF1 accumulation in S/G2. In addition, CSB interacts with BRCA1 in late S/G2 to promote MRE11- and CtIP-mediated DNA end resection (Batenburg et al., 2019), thereby enforcing HR as the pathway choice of DNA DSB repair. It has been reported that CSB is associated with telomeres and that primary skin fibroblasts derived from CSB-deficient patients exhibit an increase in telomere fragility (Batenburg et al., 2012), indicative of a role of CSB in the resolution of replication stress at telomeres. However, how CSB resolves replication stress at telomeres remains poorly characterized.

In this report, we show that CSB localizes to ALT telomeres and this localization is stimulated by depletion of SMARCAL1, suggesting that stalled replication forks trigger CSB recruitment to ALT telomeres. We show that CSB promotes recruitment of the MRE11/RAD50/NBS1 complex, BRCA1, BLM, RPA32 and POLD3 to ALT telomeres in a manner that requires CSB's ATPase activity and is controlled by ATM- and CDK2-dependent phosphorylation. In addition, we demonstrate that CSB suppresses telomeric accumulation of MUS81 and SLX4, known to collapse stalled replication forks, suggesting that CSB limits MUS-SLX-mediated processing of stalled forks at ALT telomeres. Furthermore, loss of CSB coupled with depletion of SMARCAL1 leads to a synergistic increase in not only telomeric accumulation of MUS81 but also the formation of fragile telomeres. Taken together, these results suggest that CSB-mediated HR repair and SMARCAL1-mediated fork regression cooperatively prevent stalled forks from being processed into fragile telomeres. Loss of CSB combined with depletion of SMARCAL1 also results in a synergistical increase in chromatid breaks and micronuclei formation, which is associated with a synthetic reduction in cell proliferation, underscoring their concerted efforts to promote cell growth.

GM16095, also known as SV40-transformed CS1AN (Troelstra et al., 1992), was derived from a CSB-deficient CS patient and is commonly used in laboratories working with CSB. We observed that while telomerase activity was readily detected in hTERT-immortalized GM10905 (hTERT-GM10905) cells, skin fibroblasts derived from a CSB-deficient CS patient (Batenburg et al., 2012), GM16095 cells did not exhibit any detectable telomerase activity (Fig. 1A), indicative of GM16095 being an ALT cell line. To gain further evidence that GM16095 might be an ALT cell line, we examined a number of ALT-associated hallmarks in GM16095 cells, including telomere length heterogeneity, t- and C-circles as well as ALT-associated PML bodies (APBs). Fluorescence in situ hybridization (FISH) analysis revealed that the telomere signal intensity on metaphase chromosome ends of GM16095 cells varied from barely detectable to fairly robust (Fig. 1B), indicative of telomere length heterogeneity. Southern blot analysis of telomere restriction fragments also revealed telomere length heterogeneity in GM16095 cells (Fig. S1B). In addition, GM16095 exhibited high levels of t-circles and C-circles (Fig. 1C,D). Furthermore, immunofluorescence (IF) analysis revealed the presence of APBs as evidenced by colocalization of telomeric DNA and TRF2 with PML bodies (Fig. 1E) in about 14-15% of asynchronous GM16095 cells. Taken together, these results demonstrate that GM16095 is an ALT cell line.

Reintroduction of Myc-tagged wild-type CSB into GM16095 cells suppressed APB formation as evidenced by a reduction in colocalization of telomere binding protein TRF2 with PML bodies (Fig. 1F,G), although it did not affect the expression of TRF2 and PML (Fig. S1A). Reintroduction of Myc-CSB into GM16095 cells also suppressed C-circle production (Fig. 1H,I); however, it had little impact on telomere length heterogeneity (Fig. S1B). These results suggest that CSB regulates APB formation and C-circle production, hallmarks of ALT cells.

To further investigate if the effect of CSB on ALT features might be unique to the patient-derived GM16095 cell line, we also examined ALT features in human osteosarcoma U2OS, another well-established ALT cell line that is either wild type (WT) or knocked out (KO) for CSB. IF and IF-FISH analysis revealed that while telomeric DNA, TRF2 and phosphorylated (pT371)TRF1, the last of which has been reported to be a component of APBs (Wilson et al., 2016), were readily observed to colocalize with PML bodies in U2OS WT cells (Fig. 2A), loss of CSB led to an increase in the percentage of cells exhibiting their colocalization with PML bodies (Fig. 2B-D). This increase was unlikely due to any change in the expression of TRF2 or (pT371)TRF1 in U2OS CSB-KO cells (Fig. S1C). Loss of CSB also induced C-circle production (Fig. 2E,F); however, it had little impact on the rate of T-SCEs (E.F. and X.-D.Z., unpublished data). U2OS CSB-KO cells also did not exhibit any significant change in telomere length heterogeneity compared to U2OS WT cells despite continuous culturing (at least 1 month) (Fig. S1D). Reintroduction of Myc-CSB wild type into U2OS CSB-KO cells suppressed not only colocalization of PML bodies with telomeric DNA (Fig. 2G), TRF2 (Fig. S2A) and phosphorylated (pT371)TRF1 (Fig. S2B) but also C-circle production (Fig. S2C). Taken together, these results suggest that CSB regulates APBs and C-circles, two features thought to be best correlated with ALT cancers (Cesare and Reddel, 2010).

CSB contains a central ATPase domain and therefore we asked if CSB might regulate APB formation through its ATPase activity. To address this question, we generated U2OS CSB-KO cells stably expressing the vector alone, Myc-tagged wild-type CSB or Myc-tagged CSB carrying a ATPase dead mutation of W851R (Batenburg et al., 2015). IF-FISH analysis revealed that while overexpression of Myc-CSB suppressed APB formation in U2OS CSB-KO cells, overexpression of Myc-CSB-W851R failed to do so (Fig. 2G). The expression of Myc-CSB-W851R was comparable to that of Myc-CSB (Fig. S2D). These results suggest that the ATPase activity of CSB is necessary to suppress APB formation.

Previously we have reported that CSB is phosphorylated on S10 by ATM and S158 by CDK2 and that these two phosphorylation events control its ATPase activity at DSBs in vivo (Batenburg et al., 2017). Therefore, we also investigated if CSB phosphorylation on S10 and S158 might regulate APB formation. We generated U2OS CSB-KO cells stably expressing the vector alone, Myc-CSB, Myc-CSB carrying a nonphosphorylatable S10A mutation, Myc-CSB carrying a phosphomimic S10D mutation, Myc-CSB carrying a nonphosphorylatable S158A mutation or Myc-CSB carrying a phosphomimic S158D mutation. IF-FISH analysis revealed that both Myc-CSB-S10D and Myc-CSB-S158D behaved like Myc-CSB wild type and suppressed APB formation whereas neither Myc-CSB-S10A nor Myc-CSB-S158A was able to do so (Fig. 2H). The inability of both Myc-CSB-S10A and Myc-CSB-S158A to suppress APB formation was unlikely due to any defect in their expression (Fig. S2E). These results suggest that CSB phosphorylation on S10 by ATM and S158 by CDK2 controls its ability to suppress APB formation.

Telomeres in ALT cells are enriched with HR repair proteins. We have previously reported that CSB promotes HR (Batenburg et al., 2015, 2017). Therefore, we asked if CSB might be associated with ALT telomeres to regulate telomeric recruitment of HR repair proteins. IF analysis revealed that on average about 5% of asynchronous U2OS cells exhibited colocalization of CSB with telomere binding protein TRF2 (Fig. 3A), suggesting that CSB is associated with ALT telomeres. Further IF and IF-FISH analysis revealed that while MRE11, NBS1, BRCA1, BLM, RPA32 and POLD3 localized at ALT telomeres in U2OS WT cells (Fig. 3B), in agreement with previous findings (Dilley et al., 2016; Grobelny et al., 2000; Wu et al., 2000; Zhang et al., 2019a), their telomeric localization was impaired in U2OS CSB-KO cells (Fig. 3C-H; Fig. S3A). Loss of CSB had little impact on the expression of these HR repair proteins (Fig. S3B). In addition, reintroduction of Myc-CSB wild type into U2OS CSB-KO cells rescued telomeric localization of these HR repair proteins (Fig. S4A-D). Furthermore, an increased localization of NBS1 and BRCA1 to ALT telomeres was also observed in Myc-CSB-expressing GM16095 cells compared to vector-expressing GM16095 cells (Fig. S4E,F). Overexpression of Myc-CSB did not affect the expression of NBS1 and BRCA1 in CSB-deficient GM16095 cells (Fig. S4G). These results altogether suggest that CSB promotes recruitment of HR repair proteins and POLD3 to ALT telomeres.

To investigate if the ATPase activity of CSB might be required to recruit HR repair proteins to ALT telomeres, we first examined telomeric localization of BRCA1 and NBS1 in U2OS CSB-KO cells stably expressing either the vector alone, Myc-CSB or Myc-CSB-W851R. IF-FISH analysis revealed that while overexpression of Myc-CSB rescued recruitment of BRCA1 and NBS1 to ALT telomeres in U2OS CSB-KO cells, overexpression of Myc-CSB-W851R failed to do so (Fig. 4A,B). The expression of Myc-CSB-W851R was comparable to that of Myc-CSB (Fig. S2D). These results suggest that the ATPase activity of CSB is necessary to promote recruitment of HR repair proteins to ALT telomeres. In support of this notion, we also observed that neither Myc-CSB-S10A nor Myc-CSB-S158A, both of which have been reported to be defective in promoting the ATPase-dependent chromatin remodeling activity of CSB (Batenburg et al., 2017), was able to rescue recruitment of BRCA1 and NBS1 to ALT telomeres in U2OS CSB-KO cells (Fig. 4C,D). On the other hand, Myc-CSB carrying a phosphomimic mutation of either S10D or S158D behaved like Myc-CSB wild type and rescued recruitment of BRCA1 and NBS1 to ALT telomeres (Fig. 4C,D). The inability of both Myc-CSB-S10A and Myc-CSB-S158A to promote recruitment of BRCA1 and NBS1 to ALT telomeres was unlikely due to any defect in their expression (Fig. S2E). These results suggest that CSB phosphorylation on S10 by ATM and S158 by CDK2 controls its ability to promote recruitment of HR repair proteins to ALT telomeres.

We observed that U2OS CSB-KO cells exhibited a significant increase in the telomeric accumulation of DNA damage marker γH2AX with TRF2 when compared to U2OS parental cells (Fig. 5A,B). This increase was suppressed when Myc-CSB wild type was reintroduced into U2OS CSB-KO cells (Fig. 5C), suggesting that CSB suppresses the formation of dysfunctional telomeres. To further investigate the role of CSB in regulating telomere stability, we performed FISH analysis of metaphase chromosomes from both U2OS WT and CSB-KO cells. We found that loss of CSB induced fragile telomeres and telomere loss (telomere signal-free ends) (Fig. 5D,E; Fig. S5). Reintroduction of Myc-CSB wild type into U2OS CSB-KO cells suppressed the formation of fragile telomeres and telomere loss (Fig. 5F,G). Taken together, these results suggest that CSB promotes telomere stability.

It has been well established that fragile telomeres arise from replication fork stalling (Martinez et al., 2009; Sfeir et al., 2009). Thus we asked if fragile telomeres and telomere loss induced in CSB-KO cells might be mediated by replication stress. To address this question, we treated both U2OS WT and CSB-KO cells with aphidicolin, which inhibits DNA polymerase. Treatment with aphidicolin led to a further induction in fragile telomeres in U2OS CSB-KO cells compared to U2OS WT cells (Fig. 5D), however it did not affect telomere loss (Fig. 5E). Reintroduction of Myc-CSB wild type into U2OS CSB-KO cells suppressed aphidicolin-induced formation of fragile telomeres (Fig. 5F) but not telomere loss (Fig. 5G). These results suggest that CSB prevents replication stress-induced formation of fragile telomeres.

It has been reported that the structure-specific endonucleases MUS81-EME1 and SLX4-SLX1, which process stalled replication forks into DSBs (Fekairi et al., 2009; Hanada et al., 2007; Wyatt et al., 2013), are associated with ALT telomeres (Wan et al., 2013). Therefore, we asked if CSB might regulate recruitment of MUS81-EME1 and SLX4-SLX1 to ALT telomeres to process stalled replication intermediates arising from replication stress. IF and IF-FISH analysis revealed that MUS81 and SLX4 were readily found to localize to ALT telomeres in U2OS cells (Fig. 6A), in agreement with a previous finding (Wan et al., 2013). Loss of CSB increased their assocation with ALT telomeres in U2OS cells (Fig. 6B,C). Reintroduction of Myc-CSB wild type into U2OS CSB-KO cells suppressed telomeric association of MUS81 and SLX4 (Fig. 6D,E). In addition, reintroduction of Myc-CSB into CSB-deficient GM16095 cells also suppressed telomeric association of MUS81 (Fig. 6F). Taken together, these results suggest that CSB limits MUS-SLX-mediated processing of stalled forks at ALT telomeres.

SMARCAL1, a chromatin remodeler, has been implicated in stabilizing stalled replication forks by catalyzing branch migration and fork regression (Bansbach et al., 2009; Betous et al., 2012; Yuan et al., 2009). It has been reported that depletion of SMARCAL1 leads to persistently stalled replication forks at ALT telomeres (Cox et al., 2016), which are processed by MUS81-EME1 and SLX1-SLX4. We reasoned that SMARCAL1-depleted cells represented a well-defined model to investigate the relationship of CSB with stalled forks at ALT telomeres. We transfected U2OS CSB-KO cells stably expressing Myc-CSB with either siControl or siSMARCAL1. IF analysis revealed that depletion of SMARCAL1 promoted telomeric association of Myc-CSB (Fig. 7A,B) but did not affect the expression of Myc-CSB in U2OS CSB-KO cells (Fig. S6A), suggesting that stalled replication forks trigger recruitment of CSB to ALT telomeres.

SMARCAL1 has been implicated in regulating the formation of APBs albeit with conflicting reports (Cox et al., 2016; Poole et al., 2015). Therefore, we asked if CSB and SMARCAL1 might act together to regulate APBs. To address this question, we depleted SMARCAL1 in both U2OS WT and CSB-KO cells (Fig. S6B). We observed that depletion of SMARCAL1 promoted APB formation in U2OS WT cells (Fig. 7C), in agreement with the previous finding from Flynn's group (Cox et al., 2016). Depletion of SMARCAL1 did not further induce APB formation in U2OS CSB-KO cells (Fig. 7C), suggesting that CSB and SMARCAL1 epistatically regulate APB formation.

We observed that while depletion of SMARCAL1 induced telomeric accumulation of DNA damage marker γH2AX and MUS81 in U2OS WT cells (Fig. 7D,E), in agreement with a previous report that SMARCAL1 inhibits processing of stalled forks into DNA DSBs at ALT telomeres (Cox et al., 2016). Loss of CSB led to a further increase in telomeric accumulation of γH2AX and MUS81 in response to depletion of SMARCAL1 (Fig. 7D,E), suggesting that CSB and SMARCAL1 cooperatively prevent MUS81-mediated processing of stalled forks at ALT telomeres.

It has been reported that DSBs arising from collapsed forks are channelled into homology-directed BIR repair (Costantino et al., 2014; Roumelioti et al., 2016). In agreement with these reports, we observed that depletion of SMARCAL1 led to an increase in telomeric accumulation of BRCA1, BLM, NBS1 and POLD3 in U2OS WT (Fig. 7F-I), suggesting that HR repair proteins and POLD3 are recruited to telomeres to repair DSBs arising from persistent replication stress in the absence of SMARCAL1 in ALT cells. We have shown that CSB promotes telomeric recruitment of HR repair proteins and POLD3 in ALT cells (Fig. 3). Therefore, we asked if CSB might mediate siSMARCAL1-induced recruitment of HR repair proteins and POLD3 to ALT telomeres. To address this question, we depleted SMARCAL1 in U2OS CSB-KO cells in addition to U2OS WT. Loss of CSB did not affect siSMARCAL1-induced recruitment of BRCA1, BLM, NBS1 and POLD3 to telomeres in U2OS cells (Fig. 7F-I). These results suggest that CSB does not mediate siSMARCAL1-induced recruitment of HR repair proteins and POLD3 to ALT telomeres.

To further investigate cooperative roles of CSB and SMARCAL1 at ALT telomeres, we examined telomere abnormalities in both U2OS WT and CSB-KO cells depleted for SMARCAL1. FISH analysis of metaphase chromosomes revealed that depletion of SMARCAL1 promoted the formation of fragile telomeres in U2OS WT cells (Fig. 8A). Depletion of SMARCAL1 also led to a slight but insignificant increase in telomere loss in U2OS WT cells (Fig. 8B). Fragile telomeres but not telomere loss was further exacerbated in U2OS CSB-KO cells depleted for SMARCAL1 (Fig. 8A,B). These results suggest that CSB and SMARCAL1 cooperate with each other to prevent telomere fragility.

In addition to fragile telomeres, we also observed a synergistic induction of chromatid breaks in U2OS CSB-KO cells depleted for SMARCAL1 compared to either U2OS WT cells depleted for SMARCAL1 or U2OS CSB-KO cells alone (Fig. 8C). Depletion of SMARCAL1 combined with loss of CSB also resulted in a synthetic induction in micronuclei formation (Fig. 8D), which was associated with a synergistic reduction in cell proliferation (Fig. 8E). These results suggest that CSB and SMARCAL1 cooperate with each other to promote genomic integrity and cell growth.

The work presented here has revealed that CSB coordinates with SMARCAL1 to synergistically prevent telomere fragility in ALT cells. In addition, we have shown that loss of CSB coupled with depletion of SMARCAL1 leads to a synergistic increase in chromatid breaks and micronuclei formation, indicative of cooperative roles of CSB and SMARCAL1 in the maintenance of genomic integrity. Furthermore, our finding that depletion of SMARCAL1 combined with loss of CSB results in a synthetic reduction in cell proliferation underscores the importance of their synergistic cooperation in promoting cell growth.

It has been suggested that depletion of SMARCAL1 leads to persistently stalled replication forks at ALT telomeres, which are processed by structure-specific endonucleases MUS81-EME1 and SLX1-SLX4 to generate DSBs (Cox et al., 2016). Our finding that depletion of SMARCAL1 promotes recruitment of CSB to ALT telomeres indicates that stalled replication forks trigger recruitment of CSB to ALT telomeres. Several lines of evidence suggest that CSB promotes HR repair to prevent stalled forks from being processed into fragile telomeres (Fig. 8F). Firstly, loss of CSB impairs recruitment of HR repair proteins the MRN complex, BLM, BRCA1 and RPA32 to ALT telomeres, suggesting that CSB promotes HR repair at ALT telomeres. Secondly, loss of CSB promotes recruitment of MUS81 and SLX4 to ALT telomeres, indicating that CSB prevents MUS-SLX-mediated processing of stalled forks. Thirdly, loss of CSB promotes the formation of fragile telomeres, reported to arise from replication fork stalling (Martinez et al., 2009; Sfeir et al., 2009). We have shown that loss of CSB impairs telomeric recruitment of POLD3, an essential component of BIR, suggesting that CSB may promote homology-directed BIR repair at ALT telomeres. In addition, we have shown that treatment with a DNA polymerase inhibitor aphidicolin exacerbates the formation of fragile telomeres in the absence of CSB, suggesting that CSB also suppresses the formation of fragile telomeres induced by exogenous replication stress.

It has been reported that SMARCAL1 catalyzes regression of stalled replication forks to protect stalled forks from collapsing (Bansbach et al., 2009; Betous et al., 2012; Yuan et al., 2009). We have shown that depletion of SMARCAL1 stimulates telomeric accumulation of MUS81 and γH2AX. In addition, this stimulation is further exacerbated in the absence of CSB. Furthermore, depletion of SMARCAL1 coupled with loss of CSB leads to a synergistic increase in the formation of fragile telomeres. Taken together, these findings led us to propose a model in which SMARCAL1-mediated fork regression cooperates with CSB-mediated HR repair to prevent stalled forks from being processed into fragile telomeres (Fig. 8F). Our finding that depletion of SMARCAL1 promotes telomeric recruitment of HR proteins and POLD3 independently of CSB suggests that fragile telomeres can also arise from CSB-independent HR repair of stalled/collapsed forks (Fig. 8F).

It has been reported that replication stress at telomeres triggers the processes leading to hallmarks of ALT cells including APBs (Cox et al., 2016; Poole et al., 2015; Zhang et al., 2019b). We have shown that depletion of SMARCAL1 in CSB-KO cells does not exacerbate APB formation but promotes further replication stress as evidenced by a synthetic increase in telomeric association of γH2AX and MUS81. How a further increase in replication stress at ALT telomeres fails to trigger a synthetic increase in APB formation in SMARCAL1-depleted CSB-KO cells is not known. Our finding that depletion of SMARCAL1 further exacerbates telomere fragility in CSB-KO cells suggests that replication stress arising from loss of CSB and SMARCAL1 impacts the formation of APBs in a manner mechanistically distinct from the formation of fragile telomeres.

It has been suggested that histone eviction by CSB serves as a prerequisite for active recruitment of BRCA1 by CSB to DSBs (Batenburg et al., 2019, 2017). Our finding that the ATPase-dead W851R mutation abrogates the ability of CSB to promote recruitment of BRCA1 and NBS1 to ALT telomeres suggests that CSB may remodel stalled/collapsed replication forks to facilitate recruitment of HR repair proteins to ALT telomeres. Previously we have reported that CSB is phosphorylated on S10 by ATM and on S158 by CDK2 and that these two phosphorylation events control its ATP-dependent chromatin remodeling activity at DSBs (Batenburg et al., 2017). We have shown that CSB carrying either a S10A or a S158A mutation fails to promote BRCA1 and NBS1 to ALT telomeres, suggesting that the chromatin remodeling activity of CSB at stalled/collapsed replication forks is likely mediated by ATM and CDK activities. It has been reported that ATR phosphorylates SMARCAL1 to prevent replication fork collapse (Couch et al., 2013). Together, these findings suggest that ATM and ATR may regulate the concerted actions of CSB and SMARCAL1 at ALT telomeres to prevent telomere fragility.

Retroviral expression constructs for wild-type CSB and CSB mutant alleles (W851R, S10A, S10D, S158A and S158D) have been previously described (Batenburg et al., 2015, 2017).

siRNA used were purchased from Dharmacon: non-targeting siRNA pool (D-001206-14-05); SMARCAL1 (D-013058-04-0002). siRNA transfection was performed with RNAiMAX (Invitrogen) according to the manufacturer's protocol.

Antibodies used include those against: TRF2 (1:250; Zhu et al., 2000) (a generous gift from Titia de Lange, Rockefeller University); (pT371)TRF1 (1:500; McKerlie and Zhu, 2011); MRE11 (1:5000) and NBS1 (1:10,000; Zhu et al., 2000) (kind gifts from John Petrini, Memorial Sloan-Kettering Cancer Center); PML (1:100; Santa Cruz, sc-966); PML (1:500; Abcam, ab53773); BRCA1 (1:10,000; Millipore, 07-434); RPA32 (1:200; Abcam, ab2175); γ-H2AX (1:2000; Millipore, 05-636); BLM (1:2000; Bethyl Laboratory, A300-110A), CSB (1:125; Fitzgerald, 10R-1587); CSB (1:1000; Bethyl Laboratory, A301-345A); Myc (1:1000; EMD Millipore, 9E10); SMARCAL1 (1:100; Santa Cruz, sc-166209), MUS81 (1;100; Santa Cruz, sc-47692); SLX4 (1:1000; Abnova, H00084464-B01P); POLD3 (1:500; Abnova, H00010714-M01) and γ-tubulin (1:20,000; GTU88, Sigma).

Cells were grown in DMEM medium with 10% fetal bovine serum (FBS) for U2OS (ATCC), U2OS CSB-KO (Batenburg et al., 2017), GM16095 (Coriell Institute) (Batenburg et al., 2012), hTERT-GM10905 (Batenburg et al., 2012), HeLaII (McKerlie et al., 2012; Saltman et al., 1993) and phoenix cells, supplemented with non-essential amino acids, L-glutamine, 100 U/ml penicillin and 0.1 mg/ml streptomycin. Retroviral gene delivery was carried out as described (Batenburg et al., 2012; Mitchell et al., 2009) to generate stable cell lines. Aphidicolin (Sigma) was used at 0.5 µM.

IF and IF-FISH were carried out essentially as described (Ho et al., 2016; Wilson et al., 2016). For IF except for POLD3 staining, cells seeded on coverslips were fixed directly in PBS-buffered 3% paraformaldehyde. For POLD3 staining, cells were preextracted with cold CSK buffer (10 mM PIPES [pH 7.0], 100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 0.7% Triton X-100) for 10 min prior to fixation. Cell images were recorded on a Zeiss Axioplan 2 microscope with a Hammamatsu C4742-95 camera and processed using the Openlab software package. Cells were scored positive if they exhibited colocalization of one or more large telomeric foci (telomere clustering) with PML bodies or if they exhibited colocalization of at least 3 small telomere foci with PML bodies.

Metaphase chromosome spreads were prepared as described (McKerlie et al., 2012; Mitchell et al., 2009). For drug treatment, U2OS WT and CSB-KO cells were treated with aphidicolin (0.5 µM) for 48 h prior to their arrest in nocodazole (0.1 µg/ml) for 4 h. For siRNA knockdown, U2OS WT and CSB-KO cells were transfected with siRNA against SMARCAL1 for 44 h, followed by their arrest in nocodazole (0.1 µg/ml) for 4 h.

FISH analysis on metaphase chromosome spreads was carried out as described (Batenburg et al., 2012). Slides with chromosome spreads were incubated with 0.5 µg/ml FITC-conjugated-(CCCTAA)₃ PNA probe (Biosynthesis) for 2 h at RT. Subsequently, slides were washed, counter-stained with 0.2 µg/ml DAPI, and embedded in 90% glycerol/10% PBS containing 1 mg/ml p-phenylenediamine (Sigma).

Telomere length analysis was carried out as described (Mitchell et al., 2009). Plugs containing the digested 3 µg genomic DNA were loaded on a 1% agarose gel in 0.5X TBE. Gels were run for 20 h at 5.4 V/cm at a constant pulse time of 5 s using a CHEF DR-II pulsed-field apparatus (Bio-Rad). C-circle amplification assays were performed as described (Ho et al., 2016). Following heat inactivation of φ29 at 65°C for 20 min, samples were loaded on the dot blots. The detection of φ29-amplified telomeric DNA was done with a ³²P-end-labeled single-strand (CCCTAA)₃ probe. Blots were exposed to PhosphoImager screens and scanned using a Typhoon 9400 PhosphorImager (GE Healthcare).

For growth curve assays, U2OS WT and CSB-KO cells were seeded at 25,000 cells per well in a 24-well plate. The next day, cells were transfected with siControl or siSMARCAL1 with lipofectamine RNAiMAX according to the manufacturer's instruction. Forty-eight hours post transfection, cells were counted and reseeded at 25,000 cells per well. This cycle was repeated three more times until day 12.

Protein extracts and immunoblotting were performed as described (McKerlie et al., 2013; McKerlie and Zhu, 2011). The activity of telomerase in cells was determined as described (Mitchell et al., 2009) using a Trapeze telomerase detection kit (Chemicon) according to the manufacturer's protocol.

A two-tailed Student's t-test was used to derive all P-values.

We are grateful to Titia de Lange for the anti-TRF2 antibody and to John Petrini for anti-MRE11 and anti-NBS1 antibodies.