Regulation of the Rhp26^ERCC6/CSB chromatin remodeler by a novel conserved leucine latch motif

Significance

Cockayne syndrome group B protein/excision repair cross-complementation group 6 (CSB/ERCC6) belongs to a subfamily of SWI2/SNF2 (SWItch/sucrose nonfermentable)-related chromatin-remodeling complexes. Defects in the CSB protein result in Cockayne syndrome, which is a rare autosomal-recessive neurologic disorder clinically characterized with progeriod features, growth failure, and photosensitivity. However, it remains poorly understood how the CSB enzymatic activity is regulated for CSB to perform its desired biological functions. Here, we reveal the molecular mechanism of autoinhibition of CSB enzymatic activities. We identify a novel conserved motif (leucine latch motif) that interacts with lobe 2 of the core ATPase domain and hinge region and locks the enzyme into an inactive conformation as a latch. This work provides important novel mechanistic insights into the autoregulation of CSB/ERCC6 proteins.

Abstract

CSB/ERCC6 (Cockayne syndrome B protein/excision repair cross-complementation group 6), a member of a subfamily of SWI2/SNF2 (SWItch/sucrose nonfermentable)-related chromatin remodelers, plays crucial roles in gene expression and the maintenance of genome integrity. Here, we report the mechanism of the autoregulation of Rhp26, which is the homolog of CSB/ERCC6 in Schizosaccharomyces pombe. We identified a novel conserved protein motif, termed the “leucine latch,” at the N terminus of Rhp26. The leucine latch motif mediates the autoinhibition of the ATPase and chromatin-remodeling activities of Rhp26 via its interaction with the core ATPase domain. Moreover, we found that the C terminus of the protein counteracts this autoinhibition and that both the N- and C-terminal regions of Rhp26 are needed for its proper function in DNA repair in vivo. The presence of the leucine latch motif in organisms ranging from yeast to humans suggests a conserved mechanism for the autoregulation of CSB/ERCC6 despite the otherwise highly divergent nature of the N- and C-terminal regions.

SWI2/SNF2 (switch/sucrose nonfermentable) and related ATP-dependent chromatin-remodeling enzymes in the SNF2-like family of proteins play essential roles in many aspects of DNA metabolism, including replication, transcription, recombination, and repair (1). These SNF2-like ATPases are broadly conserved throughout evolution and share a common core ATPase domain. Whereas the core motor ATPase domain provides the driving force for DNA translocation and chromatin-remodeling activity, emerging evidence suggests important roles of the flanking regions in mediating specific interactions with nucleosomes or other protein factors, substrate specificity, and the regulation of the function of the motor (2–7).

Cockayne syndrome group B protein (CSB/ERCC6) belongs to a subfamily of the SNF2-like family of ATPases (8) that plays a crucial role in transcription elongation and transcription-coupled repair (TCR), a specialized repair pathway that repairs DNA lesions in the transcribed genome (9–17). CSB is involved in the initiation steps of TCR in a manner that depends upon its chromatin remodeling and ATPase activities (18–21). Human CSB mutations are associated with Cockayne syndrome, a rare autosomal-recessive neurologic disorder characterized with progeriod features, growth failure, and photosensitivity (13, 21, 22).

CSB/ERCC6 proteins are conserved from yeast to humans (Fig. 1A). Particularly, the core ATPase domain is highly conserved not only within the ERCC6 subfamily (Fig. 1A) but also among other SNF2-like family members. The core ATPase domain is composed of two RecA-like domains (lobe 1 and lobe 2) and shares the hallmark signature of seven SF2 helicase motifs (motif I, Ia, II, III, IV, V, and VI). In addition to two conserved lobes, the core domain contains two α-helical domain insertions (HD1 and HD2) that are present in other SNF2-like family proteins such as Rad54 (23, 24). We therefore used the structure of Rad54 as a model for the CSB core ATPase domain. As shown in Fig. 1B (PDB ID code 1Z3I) (23), lobe 1 (blue) and lobe 2 (red) are connected by a hinge region (yellow), which is flexible and allows rotation between the two lobes during each cycle of ATP hydrolysis and DNA translocation (23, 24).

Fig. 1.

CSB/ERCC6 contains a highly conserved ATPase core domain and divergent flanking regions that modulate its ATPase activity. (A) The CSB core ATPase domain (green) is highly conserved across species from yeast to humans. (B) Structure of the core ATPase domain shared among SWI2/SNF2 family members. Lobe 1, HD1, hinge, lobe 2, and HD2 are shown in blue, magenta, yellow, red, and green, respectively (23). (C) S. Pombe Rhp26 truncation constructs. The core ATPase domain is shown in green with seven conserved helicase motifs highlighted in blue. N- and C-terminal regions are shown in cyan and orange, respectively. (D) DNA-dependent ATPase activity of Rhp26 and Rhp26 truncations. Both the protein and DNA concentrations were at 500 nM. The reactions were performed at 30 °C for 15, 30, and 60 min, respectively. The error bars are the SD based on three independent experiments.

In sharp contrast, the N- and C-terminal flanking regions are not well conserved, and the length of these flanking regions varies significantly across species. Among the six representative CSB proteins shown in Fig. 1A, human CSB has the longest flanking regions whereas the shortest flanking regions are found in the Schizosaccharomyces pombe CSB homolog, Rhp26. Moreover, two previously identified domains in human CSB flanking regions, the acidic-rich region (356–394) and the ubiquitin binding domain (1400–1428) (13, 25, 26), are absent in Rhp26. Recent studies have suggested that the flanking regions of human CSB play regulatory roles in its enzymatic activities (20, 27). However, the molecular mechanisms underlying these regulatory activities by flanking regions are poorly understood. Moreover, the highly divergent nature of these flanking regions also raises the question whether there is a conserved mechanism by these regions regulating CSB activity.

To address the mechanism of autoregulation of CSB proteins, we focused on Rhp26, one of the shortest CSB homologs, as the model protein. We identified a novel conserved motif in the N-terminal region, which we term the “leucine latch” motif, that interacts with lobe 2 of the core ATPase domain and hinge region. This specific protein–protein interaction restricts the rotation of two lobes during translocation and inhibits the ATPase and chromatin-remodeling activities. The alteration of the leucine latch motif abolishes the autoinhibition of the ATPase and chromatin-remodeling activities and results in a hyperactive Rhp26 protein. Moreover, the strict conservation of this motif across species reveals an evolutionarily conserved regulatory mechanism in the ERCC6/CSB subfamily despite the highly divergent nature of flanking regions. Finally, we found that the C-terminal domain antagonizes the inhibition by the leucine latch and further showed that proper regulation by both the N- and C-terminal regions is important for cell survival and DNA damage repair. Taken together, we propose a new working model for a bifunctional autoregulatory mechanism of Rhp26/CSB activities.

Results

The ATPase Activity of Rhp26 Is Inhibited by Its N Terminus but Stimulated by Its C Terminus.

To characterize the biochemical activity of Rhp26, we first compared the DNA-dependent ATPase activities of purified full-length recombinant Rhp26 protein (Rhp26^wt, 1–973) as well as five truncation mutant proteins, Rhp26ΔN(206–973), Rhp26ΔC(1–851), Rhp26-M(206–851), Rhp26-N(1–205), and Rhp26-C(797–973) (Fig. 1C and Fig. S1). Strikingly, the N-terminal region played a strong inhibitory role in regulating the ATPase activity of Rhp26 (comparison between pair of Rhp26^wt vs. Rhp26∆N, or Rhp26∆C vs. Rhp26-M) whereas the C-terminal region played a positive regulatory role (comparison between pair of Rhp26^wt vs. Rhp26∆C, or Rhp26∆N vs. Rhp26-M) (Fig. 1D). Rhp26∆N had the highest ATPase activity whereas Rhp26∆C showed the lowest ATPase activity among all of the Rhp26 proteins containing core ATPase domain (Fig. 1D). Intriguingly, like CSB, the N-terminal region (Rhp26-N) can also inhibit the ATPase activity of core domain (Rhp26-M) in trans (Fig. S2), suggesting that Rhp26-N is functionally conserved to the N-terminal region of CSB (CSB-N) despite the highly divergent nature of these flanking regions (20). As negative controls, Rhp26-N, Rhp26-C, and the “ATPase-dead” mutant Rhp26∆NK308R (predicated by the sequence alignment with CSB) showed no ATPase activity (Fig. 1D and Fig. S3).

Deletion of N Terminus Greatly Stimulates Rhp26 Chromatin-Remodeling Activity.

We examined the impact of flanking regions on the chromatin-remodeling activity of Rhp26 via two chromatin-remodeling assays using either purified mononucleosomes or chromatin as circular minichromosomes. Both assays revealed that only Rhp26∆N has a robust ATP-dependent chromatin-remodeling activity among all of the Rhp26 variants we tested (Fig. 2). Intriguingly, Rhp26∆N can even displace the histones to form a strong naked DNA band in an ATP-dependent manner in the mononucleosome-remodeling assay. No free DNA band was observed in the absence of ATP or in the presence of slow-hydrolysable ATPγS (lanes for protein 2 in Fig. 2A). As a negative control, no remodeling activity was observed for the ATPase-dead mutant, Rhp26∆N-K308R (lanes for protein 3 in Fig. 2A). Full-length Rhp26^wt failed to show strong chromatin-remodeling activity under the same experimental conditions, suggesting that it is in an autorepressed state by the N-terminal region. INO80 was used as a positive control, which remodels the mononucleosomes toward the centered positions of the DNA region, resulting in a slow-migrating band (lanes for protein 10 in Fig. 2A). We further tested the chromatin-remodeling activity by using the restriction enzyme accessibility assay with chromatin reconstituted onto circular plasmid DNA. ACF was used as a positive control. Consistent with the results from the mononucleosome-remodeling assay, only Rhp26∆N showed strong chromatin-remodeling activity whereas no obvious chromatin-remodeling activity was observed for full-length Rhp26, Rhp26∆N-K308R, and all other truncations (Fig. 2B). Taken together, these results indicate that deletion of the N-terminal region greatly enhances Rhp26 chromatin-remodeling activity.

Fig. 2.

Deletion of the N-terminal region greatly enhances the chromatin-remodeling activity of Rhp26. (A) Chromatin-remodeling assay with mononucleosome. Rhp26ΔN showed strong nucleosome-remodeling activity in the presence of ATP whereas no nucleosome-remodeling activity was observed for Rhp26^wt and other truncations. Both Rhp26^wt and Rhp26ΔC can bind the nucleosome to form a supercomplex in the presence of 1 mM ATP or ATPγS. (B) Chromatin-remodeling assay using a circular plasmid. Rhp26ΔN reveals strong remodeling activity in the presence of 3 mM ATP (lower bands). No chromatin-remodeling activity was observed for Rhp26^wt and other truncations. The Rhp26 mutants are depicted in Fig. 1 and Fig. S5.

In sharp contrast, Rhp26∆C has no chromatin-remodeling activity but forms supershifted complexes with nucleosomes even in the presence of excess competing nucleosome and DNA (lanes for protein 4 in Fig. 2A). The formation of supershifted complexes depends upon the presence of ATP or ATPγS. Interestingly, the bands attributed to the supershifted complexes seemed stronger in the presence of ATP than that of slow-hydrolysable ATPγS. This difference may indicate that the association of Rhp26 with nucleosome is dependent on both ATP binding and hydrolysis. This tight nucleosome binding of Rhp26∆C relies on an intact N-terminal region because truncations of the N-terminal region abolish the supershifted complex bands (lanes for protein 5–7 in Fig. 2A). Full-length Rhp26 forms much weaker supershifted complex bands, suggesting that the presence of the C-terminal region weakens the nucleosome binding (lanes for protein 1 in Fig. 2A). Because of Rhp26∆C’s high nucleosome binding activity and low ATPase activity, it represents an inactive “locked state” that is inefficient for DNA translocation and chromatin remodeling.

In summary, the in vitro enzymatic characterizations clearly revealed that the N-terminal and C-terminal regions have opposing roles in regulating Rhp26 enzymatic activities. The N-terminal region greatly inhibits ATPase and chromatin-remodeling activities whereas C-terminal region stimulates these activities.

Rhp26 Activity Is Tightly Regulated and Balanced in Vivo.

To test the role of Rhp26 in DNA damage repair in vivo, we analyzed the DNA damage sensitivity of S. pombe Rhp26 mutants. Consistent with the literature (28, 29), rhp26Δ cells are more sensitive to UV irradiation than wild-type (wt) cells (Fig. 3A). This UV irradiation sensitivity is further enhanced by the deletion of uve1 (uvde), which acts in a parallel UV-damaged DNA endonuclease-dependent excision repair (UVER) pathway (Fig. S4A). In addition to UV irradiation, rhp26Δ cells are also more sensitive to DNA alkylating agents such as methyl methanesulfonate (MMS) (Fig. 3A).

Fig. 3.

Analysis of Rhp26 mutants in S. pombe. (A) DNA damage sensitivity assay for Rhp26 mutants. Rhp26 fragments were tagged with 3xFLAG epitope at the N terminus and inserted at the endogenous locus under the native promoter. (B) Overexpression of rhp26⁺ and rhp26ΔN is toxic to S. pombe. rhp26 constructs were under control of the thiamine repressible nmt41 promoter and transformed into the indicated mutant background. Under repressed conditions, leaky expression partially rescues rhp26Δ in the case of wild-type and rhp26ΔN plasmids. Vector expressing 3xFLAG alone was used as a control. For all the above experiments, fivefold serial dilutions of log-phase cells with the indicated genotypes were plated and exposed to the indicated dose of UV irradiation or alkylating agent methyl methanesulfonate (MMS).

To test the effects of flanking regions on Rhp26 function, we generated mutants that express wt or mutant Rhp26 with an N-terminal 3xFLAG and integrated them at the endogenous locus under the native promoter for testing DNA damage sensitivity (Fig. 3A). The FLAG-rhp26⁺ strain behaved similarly to untagged wt strains, indicating that the FLAG tag does not noticeably interfere with Rhp26 function. Interestingly, FLAG-rhp26ΔN behaved similarly to wild type in regards to DNA damage sensitivity (with the exception to the highest dose of MMS), suggesting that the autoinhibition of the N-terminal region might be released upon DNA damage in vivo. The other FLAG-rhp26 variants were as sensitive to DNA-damaging agents as a full deletion of Rhp26 (rhp26Δ strain), including the ATPase-dead FLAG-rhp26-K308R mutation, suggesting that the ATPase activity of Rhp26 is required for cellular resistance to DNA damage in vivo.

To further test the effects of flanking regions on Rhp26 function at different expression levels, we cloned the 3xFLAG-tagged constructs into plasmids under the control of the attenuated thiamine repressible nmt41 promoter and transformed these plasmids into the rhp26Δ uve1Δ background (Fig. 3B). Similar to the result in Fig. 3A (at the endogenous locus under the native promoter), the very low level of expression of Rhp26 proteins in the presence of thiamine (repressed conditions) was sufficient for the rhp26⁺ and rhp26ΔN constructs to partially complement rhp26Δ uve1Δ [Fig. 3B, Left (repressed)]. In sharp contrast, massive increases in protein expression levels of rhp26⁺ and rhp26ΔN via derepression of the nmt41 promoter (through the removal of thiamine) (Fig. S4B) surprisingly resulted in strong cellular toxicity even in the absence of UV irradiation (Fig. 3B, overexpressed, untreated condition). This cellular toxicity was greatly enhanced in the rhp26ΔN mutant, which failed to form visible colonies even under unchallenged conditions (Fig. 3B, overexpressed). These dominant-negative phenotypes of rhp26⁺ and rhp26ΔN overexpression depend on their ATPase activities because the ATPase-dead K308R mutants abolish the cellular toxicity, suggesting that the toxicity was due to the unbalanced superactive ATPase activity of Rhp26 during overexpression (Fig. S4C). Over-expression of other less-active rhp26 mutants did not cause cellular toxicity (Fig. 3B). Taken together, these results indicate that the in vivo Rhp26 activity is tightly regulated by its N- and C-terminal regulatory regions and protein expression to ensure efficient repair of DNA lesions while avoiding toxic effects arising from unbounded ATPase activity.

Identification of a Novel Motif That Plays an Important Regulatory Role.

Because the N-terminal domain seemed to have an important role in regulating the enzymatic activities of Rhp26, we decided to closely examine and map the specific regions responsible for its regulatory roles by systematically comparing the enzymatic activities of a series of Rhp26ΔC variants with step-wise deletions of the N terminus. Intriguingly, we identified a short region (residues 6–16) at the N terminus that is important for the autoinhibition by the N-terminal region because deletion of this short region abolishes the N terminus’s autoinhibition of Rhp26 ATPase activity (Fig. S5). The deletion of this region also abolishes the capability of Rhp26ΔC to form supershifted complexes with nucleosomes (Fig. 2A, lanes for protein 4 vs. lanes for protein 6).

Importantly, we found that this short region ((E/D)LxxLGVxxΦx) is highly conserved from yeast to humans (Fig. 4A). We termed this novel motif as the leucine latch motif for its conserved Leu residues (L₇xxL₁₀) and functional roles. Secondary structure predictions indicate that part of this leucine latch motif (LxxL) likely forms a short helix, which is further connected with a downstream long stem helix to the rest of the N-terminal region (Fig. S6). We speculate that this short helix (LxxL) and hydrophobic Leu side chains may be important for locking Rhp26 in an “inactive locked state” as a latch. To test this idea, we mutated the corresponding leucine residues in Rhp26∆C to glycine or proline, generating Rhp26∆C-L7G/L10G and Rhp26∆C-L7P/L10P, which are expected to remove hydrophobic side chains and disrupt the short helix. As shown in Fig. S5C, we observed a significant increase in the ATPase activity of Rhp26∆C-L7G/L10G and Rhp26∆C-L7P/L10P in comparison with Rhp26∆C. Intriguingly, the ATPase activities of Rhp26∆C-L7G/L10G and Rhp26∆C-L7P/L10P are comparable with the ATPase activity of Rhp26-M, in which the entire N-terminal region is deleted (Fig. 1D and Fig. S5C).

Fig. 4.

The leucine latch motif is responsible for the autoregulation of Rhp26 activities by the N-terminal region. (A) A conserved leucine latch motif is revealed. Sequence alignment was generated using ClustalX (38) and the ESPript server (39). Conserved region, blue rectangle; strictly conserved residues, solid red. (B) Rhp26L7G/L10G and Rhp26L7P/L10P rescue the N-terminal region inhibition of Rhp26^wt ATPase activity. (C) Rhp26L7P/L10P or L7G/L10G mutations fully rescue the N-terminal region inhibition of chromatin-remodeling activity of Rhp26. The 200 ng of chromatin and 50 nM Rhp26 were used for each reaction. (D) Rhp26L7/L10 mutants have the same in vivo phenotypes as Rhp26ΔN. Fivefold serial dilutions of log-phase cells with the indicated genotypes were plated. The error bars in this figure are the SD based on three independent experiments.

We further tested the effect of the mutations of L7G/L10G and L7P/L10P on the full-length Rhp26 enzymatic activities (in the presence of the C-terminal region). Consistent with the results from Fig. S5C, Rhp26L7G/L10G and Rhp26L7P/L10P effectively rescued the inhibition of the N-terminal region on Rhp26 ATPase activity (Fig. 4B). More strikingly, Rhp26L7G/L10G and Rhp26L7P/L10P showed strong chromatin-remodeling activity and significantly rescued the inhibition of the N-terminal region on Rhp26 chromatin-remodeling activity (Fig. 4C). Finally, we revealed that Rhp26L7G/L10G and Rhp26L7P/L10P mutants are functionally equivalent to the hyperactive mutant Rhp26∆N in vivo (Fig. 4D).

Taken together, these experiments revealed that the L7G/L10G and L7P/L10P mutations recapitulate the effects of hyperactive Rhp26∆N by abolishing the regulatory functions of the N-terminal region in modulating Rhp26 enzymatic activities in vitro (ATPase and chromatin-remodeling activities) as well as in vivo (cellular toxicity). Importantly, these results demonstrated that the leucine latch motif is required for the autoregulatory roles of the N-terminal region.

Mapping the Interaction of the N-Terminal Leucine Latch Motif with the Core ATPase Domain by Cross-Linking/Mass Spectrometry.

To gain further mechanistic insights into how the leucine latch motif interacts with the core ATPase domain, we performed chemical cross-linking/mass spectrometry analysis to map their interactions using Rhp26∆C. The cross-linker BS3 [Bis(sulfosuccinimidyl) suberate] allows us to cross-link the free amine group (-NH₂) at the N terminus (next to the leucine latch motif) with any nearby Lys ε–NH₂ residues of the core domain (the spacer length is 11.4 Å) (Fig. S7 A–C). These cross-links are highly specific, as we identified only 6 out of all of the possible 65 Lys residues in Rhp26∆C (K132, K552, K560, K628, K742, and K817) that can form intramolecular cross-links with the short N-terminal peptide containing the leucine latch motif (Fig. 5A and Fig. S7D). Strikingly, we found that most of these specific cross-links are located in the highly conserved core ATPase domain, including the HD2 region (K552/K560) and lobe 2 (K628/K742) but not lobe 1 and the HD1 regions (Figs. 1B and 5A).

Fig. 5.

Novel regulatory mechanism of Rhp26 by the flanking regions. (A) Specific cross-linking between the Rhp26 N terminus and the core ATPase domain. The different motifs were defined based on the sequence alignment of Rhp26 with zebrafish Rad54 (PDB ID code 1Z3I). The N-terminal region, lobe1, HD1, hinge, HD2, lobe 2, and the C-terminal region are shown in cyan, blue, magenta, yellow, green, red, and orange, respectively. The specific intramolecular cross-links between the N terminus and corresponding residues are depicted. (B) The docking site for the Rhp26 N terminus based on the above cross-linking data is shown as a dashed cyan oval (Left) and locking model of the Rhp26 N-terminal region (leucine latch motif). Four Lys residues from HD2 and lobe 2 identified from the cross-linking study are highlighted in cyan spheres. The conserved leucine latch motif [magenta helix (cylinder), Right] along with the subsequent N-terminal region residues were docked in the core ATPase structure. (C) Working model of bidirectional autoregulation of Rhp26 by flanking regions. Rhp26 can exhibit either a locked inactive state (with weak ATPase and no chromatin remodeling) or a super active site (with high ATPase and chromatin-remodeling activities). The N-terminal region, core ATPase domain, and C-terminal region of Rhp26 are shown in cyan, green and orange, respectively. DNA is shown in magenta.

Further examination of the 3D distributions of these residues in the core ATPase domain based on a zebrafish Rad54 structure (Fig. 5B) (23) provided several important structural insights. First, these residues are located in a defined area within the same side of the core ATPase structure (cyan dashed circle in Fig. 5B, Left), suggesting a specific docking site for leucine latch motif at the N-terminal region (Fig. 5B, Right). Second, we noticed that two cross-link sites (K552 and K742) are very close to the hinge region connecting two lobes. It was proposed that during the translocation along the DNA, lobe 2 may rotate up to 180 degrees relative to lobe 1 from an open conformation to a closed conformation (24). K552 is located exactly at the region between the hinge region and lobe 2. K742 is also located close to the motif VI and highly conserved “Arg-finger residue” R734. Motif VI is proposed to mediate the interlobe communication, and residue R734 connects γ-phosphate of ATP. Thus, the interaction between the leucine latch motif and lobe 2 is likely to lock the relative movement of lobe 2 and lobe 1 during DNA translocation and chromatin remodeling. This result provides important structural insights into how the leucine latch motif at the N-terminal region inhibits the ATPase and chromatin-remodeling activity. Third, the other two cross-link sites (K560 and K628) are located at the interface between the HD2 region and lobe 2. HD2 is conserved among SWI2/SNF2 family and proposed to be involved in DNA binding. The interaction between the leucine latch motif and the HD2 region likely stabilizes the structure of lobe 2 and HD2 conformation and enhances nucleosomal DNA binding affinity.

Discussion

Mechanistic Insights into the Regulatory Role of a Novel Leucine Latch Motif in CSB Activities.

Here, we reveal a novel molecular mechanism of how Rhp26^ERCC6/CSB is regulated by flanking regions through a combination of biochemical, biophysical, and genetic approaches. We identified a new conserved leucine latch motif (residues 6–16, (E/D)LxxLGVxxΦx) that functions as a latch to lock Rhp26 in a repressed state with low ATPase and chromatin-remodeling activities. This short motif is responsible for the regulatory role of the N-terminal region through specific interaction with lobe 2 and the HD2 region. Based on sequence analysis and secondary structure prediction, we anticipated that part of this motif (LxxL) forms a short helix that is important for regulation (Fig. 5B and Fig. S6). Indeed, double mutants Rhp26L7G/L10G and Rhp26L7P/L10P that disrupt this short helix can fully abolish the inhibitory effect of the N-terminal region and function equivalently to Rhp26∆N both in vitro and in vivo. However, it should be noted that the complete motif may extend through the entire conserved region, such as the segment that corresponds to residues 6–28 of Rhp26 (Fig. 4A). Intriguingly, the leucine latch motif is highly conserved among the CSB homologs across species despite the generally low conservation in the N-terminal region, suggesting that this leucine latch motif-mediated regulation is likely to be an evolutionarily conserved mechanism.

Importantly, our study reveals novel mechanistic insights into ERCC6 subfamily enzymatic regulation through the leucine latch motif at the N-terminal region. The opening and closing of the interlobe cleft and relative rotation between lobe 1 and lobe 2 during the ATP hydrolysis cycle is proposed to cause the relative DNA-affinity change of the two lobes, resulting in the translocation along the DNA and nucleosome (chromatin remodeling) (23, 24, 30). Our chemical cross-linking data suggest that the leucine latch motif is likely to dock at a specific area at lobe 2/HD2 of Rhp26. This specific interaction locks lobe 1 and lobe 2 in one conformation state and blocks the rotation of lobe 2 relative to lobe 1 during ATP-dependent translocation. This result provides a mechanistic explanation of all of the observed biochemical characteristics of the N-locked inactive state (Rhp26∆C): low ATPase activity, no detectable chromatin-remodeling activity, and high nucleosome binding affinity. Our results also provide a mechanistic explanation for a similar N-terminal inhibitory effect observed in human CSB (20).

Bidirectional Regulation of Rhp26 by Flanking Regions Is Important for Its in Vitro and in Vivo Function.

Here, we reveal that the N- and C-terminal regions of Rhp26 have opposing roles in regulating Rhp26 activity. Rhp26∆N has the highest ATPase and chromatin-remodeling activities among all Rhp26 proteins containing the core ATPase domain (termed the superactive state). In sharp contrast, Rhp26∆C has the lowest ATPase and no detectable chromatin-remodeling activities (termed the inactive locked state). The enzymatic activities of Rhp26^wt are determined by a net sum of opposite contributions from both Rhp26-N and Rhp26-C.

To further elucidate this bidirectional regulatory mechanism in controlling Rhp26 activity, we propose a new working model (Fig. 5C) in which the N-terminal region promotes an inactive locked state (as seen with Rhp26∆C) and the C-terminal region promotes a superactive state (as seen with Rhp26∆N) in comparison with basal level activity contribution from the core ATPase domain of Rhp26 (Rhp26-M). In the inactive locked state, the N-terminal region enhances nucleosomal DNA binding by locking its leucine latch motif with the core domain. In this state, the conformational change of two lobes during cycles of ATP hydrolysis and ATP-dependent DNA translocation is hindered. In sharp contrast, in the superactive state, the leucine latch motif is released from the core domain to unlock the restraint between two lobes, allowing cycles of ATP hydrolysis and ATP-dependent DNA translocation. The C-terminal region of Rhp26 partially counteracts the N-terminal region function by either promoting the release of the locked N-terminal domain, greatly enhancing ATP binding/hydrolysis, or promoting conformational changes during ATP-dependent DNA translocation, and consequently enhances ATPase activity and chromatin-remodeling activity. We propose that Rhp26 can switch between these two states during translocation along the dsDNA. Our model may also apply to all ERCC6 subfamily and other homologous motor proteins.

Our studies further reveal that this bidirectional regulation by flanking regions is finely balanced in vivo because misregulation in either direction leads to toxic effects or sensitivity to DNA lesions. Overexpression of wild-type Rhp26 results in cellular toxicity. This toxicity is greatly enhanced by overexpression of the superactive forms (Rhp26∆N, Rhp26L7G/L10G, and Rhp26L7P/L10P), which may be due to the disruption of normal chromatin structures or the depletion of ATP pools. On the other hand, cells expressing Rhp26 mutants with decreased enzymatic activity (such as Rhp26∆C) result in DNA damage sensitivity. Taken together, our studies provide novel mechanistic insights into understanding autoregulation of Rhp26 activities and their contributions to DNA damage repair and chromatin stability maintenance in vivo.

An Emerging Theme of Flanking Regions in Regulating ATP-Dependent Chromatin-Remodeling Motors.

Although the autoregulatory mechanism of Rhp26 by the leucine latch motif reported in this study is significantly different from previously reported regulatory motifs for other ATP-dependent chromatin-remodeling motors, regulation by flanking regions seems to be an emerging theme for many chromatin remodelers. For example, ISWI, a distant SNF2-like family chromatin-remodeling factor, has two ISWI-specific autoinhibitory motifs (AutoN and NegC) that negatively regulate ATP hydrolysis and the coupling ATPase with DNA translocation (2). Chd1, another distantly related SNF2-like family member, contains a Chd1-specific inhibitory chromodomain, which gated the interaction between ATPase motor and duplex DNA (3). In some members of the SNF2-like family of ATPase, N-terminal regions could also play positive regulatory roles, such as Ino80 (ATPase subunit in the INO80 chromatin-remodeling complex) and Sth1 (ATPase component of the RSC chromatin-remodeling complex) (6, 7). The variety in the details of regulatory mechanisms by flanking regions may stem from the diverse functions and specificities of these different chromatin remodelers.

Finally, our study also reveals a striking mechanistic similarity in terms of the autoinhibition of enzymatic activities of key transcription coupled repair factors in prokaryotes and eukaryotes. Prokaryotic transcription repair-coupled factor (TRCF, or Mfd) has a similar function in prokaryotic transcription-coupled repair to that of CSB albeit via a different mechanism. Intriguingly, TRCF (Mfd) also has autoinhibitory N-terminal domains that clamp with the core ATPase domain to inhibit ATPase activity via restricting the rotation of two ATPase lobes (31) although the details of interaction for TRCF are different from that for Rhp26^ERCC6/CSB. Autoinhibition is subsequently released upon recruitment to a damage stalled transcription complex to initiate transcription-coupled repair. This parallel theme suggests that very similar regulatory strategies are used in controlling transcription-coupled repair in both prokaryotes and eukaryotes.

Conclusion

Our work provides several novel insights and conceptual advances in understanding the regulatory mechanism of ERCC6 subfamily proteins. We reveal that the molecular mechanism of the N terminus inhibitory effect is due to specific interaction between a novel leucine latch motif and the core ATPase domain, which consequently locks Rhp26 in one conformation state and blocks the rotation of lobe 2 relative to lobe 1 during ATP-dependent translocation. This evolutionarily conserved leucine latch motif is responsible for the N terminus inhibitory effect on ERCC6 enzymatic activities. Moreover, our studies also reveal that the C terminus plays an opposite regulatory role in comparison with the N terminus. Finally, we propose the concept that a fine balance between positive and negative regulations for Rhp26 is not only important for tuning Rhp26 activities, but also crucial for maintaining genome integrity and preventing cellular toxicity.

Materials and Methods

A detailed description regarding the preparation of Rhp26 and its truncations and mutants, ATPase assay, chromatin-remodeling assay, general S. pombe methods, and chemical cross-linking and mass spectrometry analysis can be found in SI Materials and Methods. Briefly, ATPase assays were carried out by incubating purified Rhp26 proteins with dsDNA in the reaction buffer containing α-[³²P]-ATP (Perkin-Elmer). The products were separated by microcrystalline cellulose TLC as described (32). In a mononucleosome-remodeling assay, Rhp26 proteins were incubated with reconstituted a mononucleosomal substrate, which contains a laterally positioned “601” nucleosome positioning sequence as described (33, 34). The reaction products were subjected to 5% polyacrylamide gels. Restriction enzyme accessibility assays were performed as described previously (35). S. pombe strains were generated and manipulated using established techniques (36). Strains and plasmids used are listed in Table S1. In the chemical cross-linking and mass spectrometry analysis, Rhp26ΔC was cross-linked with BS3 [Bis(sulfosuccinimidyl) suberate]. The reaction sample was resolved on a 4–12% Bis-Tris NuPage protein gel. The gel band corresponding to the cross-linked monomer was analyzed by trypsin digestion (37) and mass spectrometry.