|Home | About | Journals | Submit | Contact Us | Français|
The ATP-dependent chromatin remodeler CSB is essential for transcription-coupled DNA repair, and mutations in CSB lead to Cockayne syndrome. Here we examined the recruitment of CSB to chromatin after UV irradiation and uncovered a regulatory mechanism that ensures the specific association of this remodeler with chromatin. We demonstrate that ATP hydrolysis by CSB is essential for stable CSB-chromatin association after UV irradiation, and that defects in this association underlie some forms of Cockayne syndrome. We also show that the N-terminal region of CSB negatively regulates chromatin association during normal cell growth. Interestingly, in the absence of the negative-regulatory region, ATP hydrolysis becomes dispensable for chromatin association, indicating that CSB uses energy from ATP hydrolysis to overcome the inhibitory effect imposed by its N-terminal region. Together, our results suggest that the recruitment of CSB to lesion-stalled transcription is an ATP-dependent process and involves a gross conformational change of CSB.
ATP-dependent chromatin remodelers use ATP as energy to alter DNA-histone contacts, and their activities can lead to changes in nucleosome position, composition or conformation. These enzymes belong to the SWI2/SNF2 ATPase family, and most family members are conserved from yeast to humans. ATP-dependent chromatin remodelers are often the central catalytic subunit of a larger protein complex, and different complexes have distinct yet overlapping impacts on chromatin structure and biological processes (Becker and Horz, 2002; Cairns, 2007; Fan et al., 2003). Additional studies using chimeric remodeling complexes demonstrate that different ATP-dependent chromatin remodeling activities observed in vitro are not interchangeable in vivo, supporting the hypothesis that ATP-dependent chromatin remodelers are specialized molecular engines that are equipped with unique biochemical properties, allowing them to carry out their biological functions most efficiently (Fan et al., 2005).
The Cockayne syndrome complementation group B protein (CSB, also known as ERCC6) is an ATP-dependent chromatin remodeler and has an essential function in transcription-coupled DNA repair (Citterio et al., 2000; Troelstra et al., 1992). This process preferentially removes DNA lesions that stall transcription and is a sub-pathway of nucleotide excision repair (Hanawalt and Spivak, 2008). Cells with mutations in CSB fail to restore RNA synthesis after UV irradiation but do not have any defect in global genome nucleotide excision repair. CSB is important for multiple aspects of transcription-coupled DNA repair, including lesion recognition and repair complex assembly. Fluorescence recovery after photobleaching (FRAP) analysis indicates that UV-irradiation stabilizes the interaction of CSB with the transcription-elongation machinery (van den Boom et al., 2004). Furthermore, upon UV irradiation, CSB is the first protein recruited to chromatin sites containing lesion-stalled RNA polymerase, and CSB recruitment is essential for the assembly of proteins needed for DNA repair (Fousteri et al., 2006). In addition to its prominent role in transcription-coupled DNA repair, CSB is also implicated in base excision repair of oxidative DNA lesions and in transcription regulation mediated by RNA polymerase I, II and III (Bradsher et al., 2002; Newman et al., 2006; Stevnsner et al., 2008; van Gool et al., 1997; Yu et al., 2000). The developmental consequences of CSB mutations are revealed by Cockayne syndrome (CS), a disease in which approximately 80% of the patients have mutations in the gene encoding CSB. CS is an inherited, recessive disorder, associated with sun sensitivity, neurodevelopmental abnormalities and segmental premature aging.
The mechanism by which CSB functions in transcription-coupled DNA repair is not clear. CSB is a DNA-stimulated ATPase (Selby and Sancar, 1997). In vitro studies have shown that CSB enhances strand annealing and exchange, shortens DNA contour length and remodels nucleosomes (Beerens et al., 2005; Citterio et al., 2000; Muftuoglu et al., 2006). How any of these activities are utilized in the different steps of transcription-coupled DNA repair remains to be determined.
ATP-dependent chromatin remodelers can be recruited to their sites of action in several ways, such as through interactions with specially modified histones, histone variants or sequence-specific transcription factors. In addition, non-catalytic subunits of a remodeling complex can also ensure specific targeting of remodelers (Hogan and Varga-Weisz, 2007; Horn and Peterson, 2001; Wu et al., 2007). CSB does not harbor any identifiable sequence motifs known to direct interactions with specifically modified chromatin. Although CSB is found in a large complex containing RNA polymerase II (van Gool et al., 1997), the composition of this complex remains undescribed.
To gain broader insights into the mechanisms that regulate the association of ATP-dependent chromatin remodelers with chromatin, we dissected the mechanism by which CSB interacts with chromatin in response to UV irradiation. We determined how different CSB missense mutations associated with CS impact CSB-chromatin association in response to UV irradiation. We also assessed the contributions of CSB's enzymatic activity and primary structural organization on UV-induced CSB-chromatin association. This work has uncovered an ATP-dependent auto-regulatory mechanism that ATP-dependent chromatin remodelers can use to ensure their specific chromatin association in response to environmental stimuli.
To dissect the mechanism by which CSB associates with chromatin and determine how this association may be regulated, we employed a protein fractionation scheme to examine the subcellular distribution of CSB before and after UV treatment. Cells were lysed in an isotonic buffer containing non-ionic detergent (Figure 1A), and proteins were separated into soluble and chromatin-containing fractions by centrifugation. Fractionation efficiency was monitored by probing for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (soluble fraction), and total histones, histone H3 or elongating RNA polymerase II (chromatin-containing fraction). Additionally, CSB-PGBD3, an endogenously expressed fusion protein composed of N-terminal CSB sequence and sequence derived from a PiggyBac transposase, was used as a common probe for normalization between soluble and chromatin-containing fractions (Newman et al., 2008). One third of CSB-PGBD3 co-fractionated with chromatin while the remainder was soluble, and this distribution was independent of UV irradiation (Figure 1B).
Using this protocol, we studied the dynamics of CSB's association with chromatin. MRC5 cells (normal human fibroblasts) were heavily UV-irradiated (100 J/m2 or less than 10-6 survival), lysed at different times after irradiation, and CSB partitioning was determined by protein fractionation and Western blot analysis. As shown in Figure 1B, CSB was mainly in the soluble fraction during normal growth (lane 1 vs. 7). In contrast, by one hour after UV irradiation, about 90% of total CSB relocated to the chromatin-containing fraction (lane 2 vs. 8). Four hours post UV irradiation, more than 60% of total CSB once again became soluble and by seven hours, the vast majority of CSB was soluble (Figure 1C). As expected, GAPDH was always soluble and histones were always chromatin associated (Figure 1B). A similar trend in the partitioning of CSB was observed when MRC5 cells were more lightly UV-irradiated with a dose of 10 J/m2 (~ 60% survival; data not shown). Data quantification revealed insignificant changes in total CSB levels (soluble + chromatin-containing) within the first four hours after UV irradiation, while an approximately 50% reduction in total CSB was detected by seven hours post irradiation (Figure 1B, lanes 4 and 10 vs. lanes 5 and 11). The reduction in CSB levels could result form CSA-mediated degradation (Groisman et al., 2006) or from lethality due to the heavy UV dose that was used. The partitioning of CSB to the chromatin-containing fraction after UV irradiation likely represents stable chromatin-associated CSB, as chromatin immunoprecipitation (ChIP) experiments had revealed that CSB is recruited to sites of DNA lesion-stalled transcription one hour after UV irradiation (Fousteri et al., 2006). To test directly if UV-irradiation promotes stable CSB-chromatin association, we performed ChIP with anti-histone H3 antibodies. Western blot analysis demonstrated that CSB became associated with histone H3 after UV-irradiation (Figure 1D, lane 3 vs. 5), while the total amount of CSB remained unchanged (Figure 1D, lane 1 vs. 2). These observations indicate that, under normal growth conditions, CSB is either not associated with chromatin or the association is unstable; upon UV-irradiation, a stable CSB-chromatin association is induced.
We next determined if the amount of CSB co-fractionating with chromatin was directly proportional to UV dose. As the number of DNA lesions increases with increasing UV dose, such an analysis will determine whether the stable CSB-chromatin association is mediated through DNA damage. Accordingly, MRC5 cells were irradiated with a series of UV doses. Cells were lysed and proteins fractionated one-hour post UV irradiation. As the UV dose increased, a greater amount of CSB became associated with chromatin (Figure 1E-F). The fraction of chromatin-associated CSB reached a plateau at ~25 J/m2 (which introduces ~2 DNA lesions per 10 Kb of DNA (Mathonnet et al., 2003)); a dose similar to that required to detect maximally immobilized CSB by FRAP (van den Boom et al., 2004). We also examined the partitioning of two other proteins involved in nucleotide excision DNA repair: the endonuclease ERCC4/XP-F and the ERCC3/XPB helicase. ERCC4 also displayed UV dose-dependent, chromatin association, as did ERCC3 but to a lesser degree (Figure 1E). Notably, UV-irradiation did not alter the chromatin association of two other ATP-dependent chromatin remodelers, BRG1 and SNF2h (Figure 1E and data not shown). The vast majority of BRG1 and SNF2h were consistently in the chromatin-containing fraction along with histone H3 and elongating RNA polymerase II, while GAPDH was exclusively soluble (Figure 1E). We conclude that stable CSB-chromatin association occurs after UV irradiation.
We next assessed whether the UV-induced, chromatin association of CSB is compromised by CS-associated CSB mutations. Four mutations were examined: R670W, W851R, V957G and P1042L (Figure 2A). All these mutant proteins fail to complement the UV sensitivity of a mutant CSB cell line (Mallery et al., 1998). Transgenes encoding these mutant proteins were introduced into CSB-deficient CS1AN-Sv cells to generate stable lines, and we assayed for their ability to associate with chromatin in response to UV irradiation (Figure 2B-C). At least two independent clonal cell lines were examined for each protein. Similar to endogenous CSB in MRC5 cells, CSBwt reintroduced into CS1AN-Sv cells displayed UV dose-dependent chromatin association (Figures 1E and and2B).2B). Due to CSB overexpression, the saturating fraction of chromatin-bound CSB was less than that of endogenous CSB in MRC5 cells (~40% vs. ~90% respectively, Figures 2C and and1F).1F). Interestingly, unlike CSBwt, CSB proteins harboring the R670W, W851R or V957G mutation were impaired in their ability to associate with chromatin in response to UV irradiation, while the P1042L mutation did not interfere with this response. These results revealed that one molecular defect, associated with some forms of CS, is a failure of mutant CSB proteins to stably associate with chromatin in response to UV irradiation. However, additional mechanisms also contribute to this disease as evidenced by the P1042L mutation (see discussion).
Three mutant CSB proteins we examined have mutations in the ATPase domain (R670W, W851R and V957G), while one mutant protein has a mutation (P1042L) within a putative nuclear localization signal (Figure 2A). Our fractionation data revealed that nuclear localization of CSBP1042L is, however, unaffected (Figure 2B). To compare the biological properties of CSB to its biochemical activities, we examined the effect of these mutations on DNA-stimulated ATP-hydrolysis activity. CSBwt and the mutant CSB proteins were expressed in and affinity purified from SF9 cells (Figure 2D). As previously reported, CSBwt had robust DNA-stimulated ATPase activity (Figures 2E). Interestingly, CSBR670W and CSBW851R had no DNA-stimulated ATPase activity (Figures 2E and S1), and the V957G mutation reduced ATPase activity to 22% (Figures 2F and and3E).3E). In contrast, CSBP1042L displayed a level of DNA-stimulated ATPase activity similar to CSBwt (Figures 2F and and3E).3E). We next examined the DNA binding activity of CSBV957G and CSBP1042L using ATPase assays to determine their affinity for DNA (KM, the DNA concentration necessary to achieve half-maximal rates of ATPase activity). As shown in Figure 3D, the KM for DNA of CSBV957G was 2.3-fold of CSBwt, suggesting that the CSBV957G protein has a lower DNA affinity than CSBwt. In contrast, the KM for DNA of CSBP1042L was unaffected. These results suggest that the DNA-stimulated ATPase activity of CSB is important for UV-induced, stable CSB-chromatin association.
To test further the requirement of DNA-stimulated ATPase activity for stable CSB-chromatin association, we generated four additional proteins containing point mutations in conserved residues predicted to contact DNA and analyzed their effects in vitro and in vivo (Figure 3A-B). N653I lies in domain 1 and is predicted to disrupt contacts made with the DNA backbone; R894E, R950E and K979E lie in domain 2 and are predicted to interfere with DNA binding and/or ATP hydrolysis (Durr et al., 2005). As shown in Figure 3C, R894E abolished DNA-stimulated ATPase activity. Direct, side-by-side measurements of ATPase activity and apparent DNA affinity (KM) demonstrated that N653I, R950E and K979E significantly attenuated the interaction of CSB with DNA, reducing DNA affinity at least 5 fold, and decreased DNA-stimulated ATPase activity to 75%, 24% and 46% of wild-type level, respectively (Figure 3D-E).
By transient expression, we next examined the in vivo consequences of these four mutant proteins in CS1AN-Sv cells, as well as CSBR670W and CSBK538A (a Walker A motif mutation disrupting ATP hydrolysis) (Figure 3A). As shown in Figure 3F, CSBwt, transiently expressed in CS1AN-Sv cells, localized to the soluble fraction in the absence of UV treatment (lane 1 vs. 2), and the fraction of chromatin-bound CSBwt increased one hour after UV irradiation (lanes 3-4). In contrast, the R670W mutation disrupted the ability of CSB to stably associate with chromatin after UV irradiation (lanes 11-12 vs. 3-4), consistent with our observations made in stable cell lines (Figure 2B). Interestingly, CSBK538A, which is devoid of ATPase activity (Citterio et al., 1998), failed to stably associate with chromatin after UV irradiation (lanes 7-8). Similarly, CSBN653I, CSBR894E, CSBR950E and CSBK979E also failed to stably associate with chromatin after UV irradiation (Figure 3F). These results support the hypothesis that ATP hydrolysis stimulated by DNA is essential for UV-induced, stable CSB-chromatin association.
To define the structural elements of CSB that regulate chromatin association, three regions of CSB were independently expressed in CS1AN-Sv cells: CSB-N (1-507), CSB-C (1010-1493) and CSB-M (455-1009) (Figure 4A). Interestingly, CSB-M (ATPase domain alone) did not stably associate with chromatin after UV treatment (Figure 4B, lane 8), nor did CSB-N and CSB-C (Figure 4B, lanes 4 and 12). These observations suggest that the ATPase domain functions in concert with either the N- or C-terminal region to mediate UV-induced chromatin association.
To test this hypothesis, we characterized the chromatin association of N- and C-terminally truncated CSB proteins: CSBΔN (455-1493) and CSBΔC (1-1009) (Figures 4A and 4C). Deleting the C-terminal 484 amino acids disrupted the ability of CSB to become stably associated with chromatin one hour after UV treatment (Figure 4C, lane 8), indicating that the C-terminal domain of CSB cooperates with the ATPase domain to stabilize chromatin association. This notion was confirmed by our observation that CSBΔN readily associated with chromatin one hour after UV irradiation (Figure 4C, lane 12), demonstrating that the N-terminal 454 amino acids are dispensable for stable CSB-chromatin association.
Intriguingly, in contrast to CSB, a significant amount of CSBΔN was present in the chromatin-containing fraction even without UV treatment (Figure 4C, lane 10), indicating that the N-terminal region negatively regulates CSB-chromatin association and dictates substrate specificity. To confirm that CSBΔN stably associates with chromatin independently of UV irradiation, we performed ChIP-Western analysis with anti-histone H3 antibodies using CS1AN-Sv cells stably expressing CSBΔN. As shown in Figure 4D, CSBΔN co-purified with histone H3, independent of UV, indicating that CSBΔN can bind to chromatin in the absence of DNA lesions. Together, these observations suggest that both the ATPase and C-terminal domains are required for UV-induced, stable chromatin association and that the N-terminal region acts as a negative regulatory element preventing CSB from stably associating with chromatin in the absence of lesion-stalled transcription.
To dissect how the different regions contribute to CSB's biochemical activities, we measured ATP hydrolysis rates and DNA binding affinities of CSBΔN, CSB-M and CSBΔC. As shown in Figure 4F, CSBΔN displayed an increased KM for DNA (decreased DNA affinity) but elevated ATP hydrolysis activity. This was in sharp contrast to all other mutant CSB proteins examined (Figures 2--3),3), which displayed both decreased DNA affinity and ATPase activity in vitro. The increased ATPase activity of CSBΔN suggests that the N-terminal region negatively regulates the enzymatic activity of CSB in addition to chromatin association (Figure 4C-D). CSB-M exhibited a further decrease in DNA affinity and a concomitant decrease in ATPase activity (Figures 4F and 4G). Examination of CSBΔC revealed that this protein had no significant DNA-stimulated ATPase activity. These results suggest that the N-terminal region, the ATPase domain and the C-terminal regions of CSB all directly contribute to DNA association and catalytic activity.
The results from our in vivo and in vitro assays support the hypothesis that the UV-induced, stable chromatin association of CSB depends upon its DNA-stimulated ATPase activity. Moreover, these results reveal that specific targeting of CSB in vivo is ensured by a complex mechanism involving both positive (C-terminal) and negative (N-terminal) regulatory elements (Figure 7).
To verify that the N-terminal region functions as a negative regulatory element, we asked whether CSB-N could inhibit the ATPase activity of CSBΔN in trans (Figure 5). Interestingly, while CSBΔN displayed significant DNA-stimulated ATPase activity, CSB-N could reduce this activity. The inhibitory effect was proportional to CSB-N concentration, reaching a maximal inhibition of 63% at an equal or greater molar ratio of CSB-N/CSBΔN (Figure 5A, orange). Furthermore, the inhibitory effect was specifically caused by CSB-N and not a contaminating activity, as CSB-N did not inhibit the ATPase activity of another ATP-dependent chromatin remodeler BRG1 (Figure 5B).
In principle, CSB-N could inhibit the ATPase activity of CSBΔN by competing with CSBΔN for binding to DNA or by interacting directly with CSBΔN. To distinguish between these possibilities, we asked if increasing DNA concentrations could reverse the inhibitory effect. As shown in Figure 5D and Table S1, 7 nM of CSB-N inhibited ~ 60 % of the ATPase rate of 3 nM CSBΔN in the presence of 0.123 ng/μl DNA (green with filled square). Increasing DNA concentration by 3-, 10- or 30-fold did not reverse the inhibition of ATPase activity by CSB-N (Figures 5D-E, and Table S1). Additionally, inhibition was observed over a wide range of DNA concentrations, as low as 0.04 ng/μl. These results argue against a model whereby CSB-N prevents CSBΔN from binding to DNA and support a model in which CSB-N directly inhibits the ATPase activity of CSBΔN, consistent with our side-by-side comparisons of CSB and CSBΔN activities described in Figure 4. In agreement with this model, in vitro binding assays revealed that CSB-N directly interacted with CSBΔN and CSB-C (Figure S2A). Furthermore, CSB-N failed to inhibit the ATPase activity of CSB (Figure 5C), suggesting that the N-terminal region of full-length CSB normally occupies the site of interaction. Interestingly, CSB-N decreases both the Vmax and KM of CSBΔN to values observed for CSB (Figures 5F and S2B). Importantly, the ratios of Vmax to KM of CSBΔN in the presence or absence of CSB-N remained constant (Figure 5F). This observation is consistent with a notion that the N-terminal region of CSB can act as an uncompetitive inhibitor of CSBΔN, implying that CSB-N increases the nonreactive interaction of CSBΔN with DNA (which does not stimulate ATP hydrolysis).
Our results indicate that ATP hydrolysis is necessary for the UV-induced association of CSB with chromatin. In addition, the N-terminal region negatively regulates CSB-chromatin association in absence of UV irradiation. Furthermore, the N-terminal region inhibits the ATPase activity of CSBΔN, likely by increasing non-reactive interactions of CSB and DNA. Together, these results suggest that a function of ATP hydrolysis might be to remove the inhibitory effect imposed by the N-terminal region. If this were true, then a CSBΔN mutant protein that is also defective in ATP hydrolysis should still retain its ability to associate with chromatin, due to the absence of the negative regulatory region.
To test this hypothesis, we removed the N-terminal 454 amino acids from the ATPase-defective and CS-associated mutant CSB proteins, and we determined if these doubly mutant proteins interacted with chromatin in the absence of UV irradiation. As shown in Figure 6A-B, each of these proteins stably associated with chromatin regardless of UV treatment. The stable chromatin association was confirmed by ChIP-Western analysis of CS1AN-Sv cells stably expressing ATPase-defective CSBK538A ΔN (Figure 6C). Together, these results support a model in which the N-terminal region of CSB prevents the protein from stably associating with chromatin under normal growth conditions; upon UV irradiation, in the presence of lesion-stalled transcription, the energy from DNA-stimulated ATP hydrolysis can be used to overcome this inhibition (Figure 7).
CSB is an ATP-dependent chromatin remodeler that has an essential role in transcription-coupled DNA repair. In a critical early step, CSB associates with chromatin at or near sites of DNA lesion-stalled transcription. Our study provides mechanistic insights into the regulation of this association. We demonstrate that a UV-induced, stable CSB-chromatin association depends upon the ability of CSB to hydrolyze ATP. Our study also uncovered an autoregulatory mechanism, in which the N-terminal region prevents CSB from stably associating with chromatin under normal growth conditions, and the C-terminal region of CSB promotes stable chromatin association in the presence of lesion-stalled transcription.
From our results, we depict a model for the UV-induced, stable association of CSB with chromatin (Figure 7). During normal cell growth, the association of CSB with chromatin is dynamic, and this dynamic association is likely mediated through interactions of the ATPase domain and the N-terminal region of CSB with chromatin. The N-terminal region prevents stable CSB-chromatin association, by sequestering a DNA-binding surface within the C-terminal region, and promotes non-reactive chromatin interactions. In the presence of chromatin containing lesion-stalled transcription, the association of CSB with chromatin is stabilized. We propose that stabilization is initiated by a substrate recognition domain within the N-terminal region of CSB and accomplished by extending the DNA contact surface from the ATPase domain to include residues in the C-terminal region. The N-terminal region of CSB normally occludes this C-terminal DNA binding surface, and ATP hydrolysis is necessary to relieve the inhibitory effect imposed by the N-terminal region. We suggest that ATP hydrolysis drives a conformational change that is stabilized by components of lesion-stalled transcription and exposes a DNA-binding surface in the C-terminal region that leads to stable CSB-chromatin association.
Using the V8 protease to probe for structural changes, we observed that CSB displays different sensitivities to protease digestion in the presence of DNA and ATP versus DNA and ADP (Figure S3). These changes were observed by silver staining the digestion products resolved by SDS-PAGE and by Western blot analysis, using N- or C-terminal specific anti-CSB antibodies. The hydrolysis of ATP by other Swi2/Snf2 family members has been hypothesized to induce conformational changes. In the case of Rad54, conformational change associated with ATP hydrolysis has been suggested to be important for its mechano-chemical cycle (Lewis et al., 2008). Furthermore, conformational change of the Iswi2 remodeler has proposed to be part of the reaction cycle of nucleosome movement (Fitzgerald et al., 2004). Our analysis of CSB has revealed that a conformational change induced by ATP hydrolysis leads to stable association of this remodeler with chromatin in the presence of UV-induced DNA lesions.
Also in agreement with our model, we have found that the CDK9 inhibitor 5,6-dichloro-1-beta-D-ribofuranosylbenzimidazole (DRB) abolished the UV-induced association of CSB with chromatin (Figure S4). DRB inhibits transcription elongation, in part, by inhibiting Ser2 phosphorylation on the heptad repeats of RNA polymerase II-CTD. This result highlights the importance of elongating RNA polymerase II in recruiting CSB to lesion-stalled transcription, and suggests that the phosphorylation state of RNA polymerase II may be part of a control mechanism that stabilizes an open CSB conformation, permitting stable CSB-chromatin association.
P1042L is the only naturally occurring mutation we analyzed that did not perturb the ATP hydrolysis activity of CSB, and CSBP1042L stably bound to chromatin in response to UV irradiation. There is no clear phenotypic distinction between a CSB patient that is heterozygous for R670W and P1042L with a patient that is heterozygous for R670W and an apparent null allele (Q184stop), and CSBP1042L fails to rescue the UV sensitivity of mutant CSB cells (Lehmann et al., 1993; Mallery et al., 1998); therefore, the P042L mutation must be disrupting an essential CSB function. For instance, the P1042L mutation might disrupt an interface important for recruiting additional factors to lesion-stalled transcription. CSB has been reported to have functions that are independent of ATP hydrolysis, such as in the repair of oxidative DNA lesion (Selzer et al., 2002), and such ATP-independent functions might be compromised by the P1042L mutation.
Like the ATPase domains of other remodelers, the ATPase domain of CSB interacts with DNA (Figure 4). However, the affinity of the CSB ATPase domain for DNA is substantially increased in the presence of the C-terminal region, as revealed by the lower KM for DNA of CSBΔN compared to CSB-M. Additionally, when the C-terminal region of CSB is removed, the resulting protein CSBΔC has essentially no DNA-stimulated ATPase activity. Taken together, these observations indicate that the C-terminal region of CSB positively regulates the activity of the ATPase domain. Similar to the C-terminal region of CSB, the N-terminal regions of Ino80 and Sth1 have been shown to positively regulate the ATPase activities of these remodelers (Shen et al., 2003; Szerlong et al., 2008). In these instances, enhancement is achieved through the binding of nuclear actin-related proteins to the HSA domain. In the case of CSB, the enhancement in ATPase activity by its C-terminal region is likely mediated, at least in part, through interactions with DNA, as the effects of the C-terminal region were revealed in vitro with a naked DNA substrate.
The N-terminal region of CSB appears to play a pivotal role in regulating the way in which CSB interacts with DNA and hydrolyzes ATP, and in ensuring the specific targeting of CSB inside a cell. Specifically, our model suggests three main functions for the N-terminal region of CSB: (1) maintain a dynamic CSB-chromatin interaction by sequestering the DNA-binding surface of the C-terminal region, (2) reduce the ATPase activity by increasing non-reactive DNA binding and (3) ensure specific chromatin targeting by interacting with components of lesion-stalled transcription.
What sequences within CSB-N are critical for substrate recognition or autorepression? Among the conserved residues within this portion of CSB is a stretch of 39 amino acids (366-394) that is 60% acidic. It was previously shown that deletion of all or part of this region had no effect on DNA repair (Brosh et al., 1999; Sunesen et al., 2000); therefore, this acidic-rich region is unlikely essential for the recognition of UV-induced substrates or autorepression. Our preliminary result suggests that amino acids that lie between 245 and 455 are important for regulating chromatin association (data not shown). To identify sequence features important for recognizing UV-induced substrates or autorepression, we aligned CSB proteins from different chordates and found, in addition to the acidic region, there are conserved patches of basic residues and small regions of amino acid identity (Figure S5). These conserved features are likely important for substrate recognition and/or autorepression. The basic patches could be important for interactions with negatively charged surfaces, such as DNA and the phosphorylated C-terminal heptad repeats of elongating RNA polymerase II. More in-depth mutagenesis analyses coupled with protein interaction studies and CSB complex purifications should define residues important for substrate recognition and/or autorepression.
Brosch et al. (1999) suggested that the ATPase activity, and not the conserved N-terminal acidic region, is important for DNA repair. Consistent with this hypothesis, we found that over-expressing CSBΔN partially rescued the UV sensitivity of CS1AN-Sv cells (Figure S6). Interestingly, overexpressing a CSBΔN protein defective in ATP hydrolysis failed to rescue the UV sensitivity of CS1AN-Sv cells, indicating that additional ATP-dependent CSB functions are required for steps after recruitment of CSB to lesion-stalled transcription. In agreement with this notion, CSBΔN was unique among the mutant proteins we examined as it displayed elevated DNA-stimulated ATPase activity. This observation suggests that, once stably bound to chromatin, increased rates of ATP hydrolysis might be critical for CSB to carry out subsequent functions in transcription-coupled DNA repair.
Newman et al. (2008) have recently identified a second abundant protein encoded by the CSB locus, in which the N-terminal 465 residues of CSB are alternatively spliced to a C-terminal transposase domain encoded by the PiggyBac element PGBD3 located in CSB intron 5. Surprisingly, the resulting CSB-PGBD3 fusion protein continues to be expressed in a great majority of CS patients and has been suggested to contribute to the diverse features associated with CS (Newman et al., 2008). The interaction of CSB-PGBD3 with chromatin is not enhanced by UV irradiation (Figure 1A), and the N-terminal region, alone, does not associate with chromatin in cells with or without UV treatment (Figure 4). Consequently, recognition of lesion-stalled transcription by the N-terminal region may require CSB's enzymatic activity, consistent with the predictions of our model. How and to what extent the CSB-PGBD3 fusion protein may impact CSB function remains to be elucidated. Moreover, overexpressing CSB-N in cells expressing CSB or CSBΔN did not alter the partitioning of these proteins, before or after UV treatment (data not shown), suggesting that, once stably associated with chromatin, CSB is refractory to the inhibitory effects of the N-terminal region. The strategy that CSB uses to critically regulate its interaction with chromatin in response to UV irradiation represents a distinct mechanism that ensures the specific targeting of this ATP-dependent chromatin remodeler.
MRC5 and CS1AN-Sv were maintained as described previously (Troelstra et al., 1992). Transfections were carried out with lipofectamine 2000 (Invitrogen). To generate stable CS1AN-Sv cell lines, transfected cells were selected with 250 ng/ml puromycin or 600 μg/ml G418. Single cell clones or pooled transformants were used as indicated. Western blot and immunofluorescence analyses were used to characterize stable cell lines. For assays using transient transfection, constructs were introduced into equal numbers of CS1AN-Sv cells seeded on two 60-mm plates. Cells were processed 16-48 hours post transfection.
To study the kinetics of CSB-chromatin association after UV irradiation, equal numbers of cells were seeded on multiple 60-mm plates and the following day irradiated with 100 J/m2 UV (254 nm) using a Stratalinker (Figure 1A). At different times after UV irradiation, cells were lysed on ice in Buffer B (20 mM HEPES (pH7.9), 150 mM NaCl, 0.5 mM MgCl2, 1 mM DTT, 0.5% Triton X-100 and 10% glycerol). Lysates were centrifuged at 15,000 rpm for 20 minutes at 4°C to separate soluble and chromatin-containing fractions. Chromatin-containing fractions were solublized by sonication in Laemmli buffer. Lysates were resolved in 4-12% Bis-Tris acrylamide gels (Invitrogen).
To determine the effect of UV dose on CSB-chromatin association, equal numbers of cells were seeded into multiple 60-mm plates and irradiated with different doses of UV light the following day (Figure 1A). After recovery for 1 hour at 37°C, cells were lysed in Buffer B, and processed as above.
ChIP-Western analyses were preformed with MRC5 cells or CS1AN-Sv cells stably expressing either CSBΔN or CSBΔNK538A. ChIP was performed as previously described with the exception of the reverse cross-linking, which was carried out at 65°C for 16 hours in Laemmli buffer (Fousteri et al., 2006). Anti-histone H3 antibodies (Millipore #06-755 and #06-599) and protein G agarose were used for IP.
CSB proteins were detected using a polyclonal antibody against the N or C-terminal region of CSB (provided by Dr. Weiner, U. Washington). The anti-BRG1 antibody was provided by Dr. Kingston (MGH). Antibodies against Pol II (H5), GAPDH, ERCC3 and ERCC4 were from Convance, Millipore or Bethyl Laboratory. Western blots were developed using Supersignal West Pico chemiluminescent substrate (Thermo). Densitometry was performed using Image J software (NIH).
Soluble and chromatin-containing fractions were loaded on gels at a ratio of 1:1.25. The fraction (X) of soluble CSB was calculated as follows:
X= signal intensity of soluble CSB ×1.25/(signal intensity of soluble CSB × 1.25 + signal intensity of CSB co-fractionating with chromatin); the fraction of CSB (or mutant proteins) co-fractionating with chromatin was calculated as 1- X.
The mammalian CSB expression plasmid CSB-IRES-Puro was kindly provided by Dr. Weiner. CSB deletion constructs were generated by PCR amplification and cloned into the pcDNA3 expression vector. All proteins were N-terminally tagged with the Flag epitope, if not otherwise specified. Point mutations were created using QuickChange (Stratagene). For protein expression in SF9 cells, DNA encoding the various Flag-tagged CSB proteins was cloned into pFastBac. Proteins were purified using M2 affinity chromatography.
ATP hydrolysis reactions were carried out as described previously (Narlikar et al., 2001). Enzymes were used in the range of 2 to 60 nM, and DNA (pcDNA3) was used at 0 – 200 ng/μl. For experiments involving multiple proteins (Figure 5), proteins and DNA were mixed on ice. Reactions were prewarmed for 5 minutes at 30°C and initiated by adding γ-32P-ATP/Mg. Reaction aliquots were quenched with stop solution. Inorganic phosphate and ATP were separated on PEI-cellulose TLC plates (Baker) and ratios were quantified using a PhosphoImager and analyzed with ImageGauge software (Fuji). Initial rates were determined by linear least-squares fitting of the data obtained for the first 15% of free phosphate generated, using KaleidaGraph. KM for DNA and kcat for ATPase were calculated by plotting rates of ATP hydrolysis against DNA concentration and fitting the data.
This work was supported by a grant (GM 084983) from NIH (H.-Y. Fan) and by the FCCCs core grant (CA 006927). We specially thank Geeta Narlikar for her discussions of the model. We are grateful to Ken Zaret, Alan Weiner and Maureen Murphy for invaluable comments on this manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.