The XBP-Bax1 Helicase-Nuclease Complex Unwinds and Cleaves DNA

XPB helicase is an integral part of transcription factor TFIIH, required for both transcription initiation and nucleotide excision repair (NER). Along with the XPD helicase, XPB plays a crucial but only partly understood role in defining and extending the DNA repair bubble around lesions in NER. Archaea encode clear homologues of XPB and XPD, and structural studies of these proteins have yielded key insights relevant to the eukaryal system. Here we show that archaeal XPB functions with a structure-specific nuclease, Bax1, as a helicase-nuclease machine that unwinds and cleaves model NER substrates. DNA bubbles are extended by XPB and cleaved by Bax1 at a position equivalent to that cut by the XPG nuclease in eukaryal NER. The helicase activity of archaeal XPB is dependent on the conserved Thumb domain, which may act as the helix breaker. The N-terminal damage recognition domain of XPB is shown to be crucial for XPB-Bax1 activity and may be unique to the archaea. These findings have implications for the role of XPB in both archaeal and eukaryal NER and for the evolution of the NER pathway. XPB is shown to be a very limited helicase that can act on small DNA bubbles and open a defined region of the DNA duplex. The specialized functions of the accessory domains of XPB are now more clearly delineated. This is also the first direct demonstration of a repair function for archaeal XPB and suggests strongly that the role of XPB in transcription occurred later in evolution than that in repair.

The superfamily 2 helicase XPB (Rad25 in Saccharomyces cerevisiae) is an essential component of transcription factor TFIIH. The ATPase activity of XPB is required for both nucleotide excision repair (NER) 2 and transcription initiation from RNA polymerase II promoters (1,2). The NER pathway is a highly flexible system that is required for the detection and removal of a wide variety of bulky and helix-distorting lesions, including photoproducts. Mutations in the xpb or xpd genes in humans can cause the serious genetic diseases xeroderma pigmentosum, trichothiodystrophy, and Cockayne's syndrome, due to defects in both transcription and repair (reviewed in Ref. 3). Examples of xpb mutations in humans are much more rare than those seen in the xpd gene, probably due to the crucial role of the XPB protein in basal transcription (4).
Although the ATPase activities of both proteins are required for NER (5), the respective roles of the XPB and XPD helicase components of TFIIH are still a matter of debate. XPD is the more robust helicase (6), and it has been suggested to bind 5Ј of the DNA lesion and translocate in a 5Ј to 3Ј direction toward the damage site, potentially acting as a sensor or proofreader of DNA damage for the NER pathway either by jamming directly on DNA lesions (7) or perhaps through damage sensing by its iron-sulfur cluster binding domain (8,9). However, little direct evidence exists in support of these possibilities at present, and indeed it is not yet clear whether XPD or XPB binds first at repair sites or whether they bind the same or complementary strands in the repair bubble. The helicase activity of XPB is rather weak in vitro and is stimulated by its association with the TFIIH subunit p52 (10,11). Mutations that target the helicase motifs of XPB do not disrupt the function of TFIIH in NER, leading to the suggestion that XPB should be considered as an ATP-dependent molecular switch, perhaps opening DNA structure locally (11). Recent studies confirm the essential requirement for XPB ATPase activity in the recruitment of TFIIH to DNA damage sites (12). In contrast, the ATPase activity of XPD is not required. These data suggest that XPB may bind first to repair sites, perhaps locally destabilizing the DNA duplex to allow subsequent XPD binding and extension of the repair bubble.
The archaea share many informational proteins in common with eukarya. Most archaea encode clear homologues of the eukaryal NER helicases XPB and XPD (13). XPD is an active 5Ј to 3Ј helicase with an essential iron-sulfur cluster (8). The crystal structure of archaeal XPD provided a molecular explanation for the effects of mutations causing xeroderma pigmentosum, trichothiodystrophy, and Cockayne's syndrome (9,14,15). Archaeal XPB on its own is an ssDNA-dependent ATPase in vitro, with weak helicase activity under some conditions (16,17). A crystal structure of the core of XPB from Archaeoglobus fulgidus revealed the presence of two canonical motor domains and two accessory domains named the thumb (Thm) domain and damage recognition domain (DRD) with putative roles in DNA damage detection (17). The relevance of the archaeal XPB structure to an understanding of the eukaryal protein was emphasized by the finding that the Thm domain and a conserved RED motif identified in the archaeal enzyme are essential for the function of eukaryal XPB in NER (12).
Genes encoding archaeal XPB are usually found next to a gene encoding a protein of unknown function named Bax1 (16). Bax1 and XPB were shown previously to interact physically in vitro (16), and bioinformatic analyses have suggested that Bax1 might be a DNA endonuclease (18). This prediction was confirmed recently for Bax1 from Thermoplasma acidophilum (19).
Here we report the purification and characterization of a recombinant XPB-Bax1 complex from Sulfolobus solfataricus. We demonstrate that the complex functions as a helicase-nuclease partnership, unwinding and cleaving DNA substrates that are models for the early steps in NER. We show that the Thm domain of XPB is essential for DNA unwinding and Bax1-mediated cleavage, consistent with a role in DNA duplex unwinding. The DRD has a subtle but essential role in DNA processing by the XPB-Bax1 complex, and may be unique to archaeal XPBs. We conclude that archaeal XPB-Bax1 functions in archaeal NER, with Bax1 performing the role equivalent to XPG in eukaryal cells.

EXPERIMENTAL PROCEDURES
Cloning, Mutagenesis, Expression, and Purification-The xpb2 and bax1 genes (sso0473 and sso0475), which are organized as an operon in S. solfataricus (20), were amplified as a unit from S. solfataricus genomic DNA by PCR using the oligonucleotides 5Ј-CCATGGTAGGATTAGGATAC and 5Ј-CAG-GATCCTTTAAACCTCTTTGATC and cloned into the pEHISTEV vector (21) using the BamHI/NcoI recognition sites for co-expression of recombinant XPB and Bax1 in E. coli. The N terminus of XPB carried a polyhistidine tag with a TEV protease cleavage site. The construct was sequenced fully to confirm the expected nucleotide sequences of both genes. Protein expression was carried out in C43 cells induced overnight for 16 h at 28°C by the addition of isopropyl 1-thio-␤-D-galactopyranoside (0.5 mM).
For purification, cells were lysed by sonication in lysis buffer (20 mM Tris (pH 8.0), 500 mM NaCl, 0.1% Triton X-100, and 1 mM EDTA) containing a mixture of protease inhibitors (Roche Applied Science). The lysate was centrifuged (48,000 ϫ g, 30 min, 4°C), and filtered through a 0.45-m syringe filter before being applied to a HisTrap column (HisTrap HP 5 ml, GE Healthcare), previously charged with NiCl 2 and equilibrated with loading buffer (20 mM Tris (pH 8.0), 500 mM NaCl, 30 mM imidazole, and 10% glycerol). Proteins bound to the column were eluted with a linear gradient of imidazole (0.03-0.5 M) in loading buffer. Fractions containing the XPB-Bax1 complex were identified by SDS-PAGE, pooled, and purified to homogeneity by using a HiLoad 26/60 Superdex 200 size exclusion column (GE Healthcare) equilibrated with gel filtration buffer (25 mM Tris, pH 8.5, 500 mM NaCl, 1 mM EDTA, 0.5 mM dithiothreitol, and 10% glycerol). Fractions containing the XPB-Bax1 complex were identified by SDS-PAGE, pooled, and concentrated, and the polyhistidine tag was removed by cleavage overnight at 4°C using 0.2 mg/ml TEV protease in the same buffer. Following cleavage, the protein was applied to the HisTrap column in loading buffer to separate tagged and untagged protein.
The untagged complex was collected from the flow-through, pooled, and concentrated to 5 mg/ml in storage buffer (20 mM Tris, pH 8.0, 500 mM NaCl, 10% glycerol) and stored aliquoted at Ϫ80°C until needed.
The ⌬nuclease, ⌬helicase, ⌬RED, and ⌬Thumb mutants were generated using a XL QuikChange mutagenesis kit (Stratagene). The ⌬DRD mutant was generated by introducing an NcoI site before the codon corresponding to Val 54 by mutagenesis. The plasmid was then digested by NcoI to remove the DNA corresponding to residues 1-53, which constitute the DRD, and the vector circularized by ligation and transformed into E. coli. Mutations and deletions were confirmed by DNA sequencing and mass spectrometry of the pure proteins. Oligonucleotides for mutagenesis are listed in Table 1.
DNA Substrate Preparation-Oligonucleotides for DNA substrates were gel-purified and ethanol-precipitated as described previously (22). The oligonucleotides in Table 2 were purchased from Operon Biotechnologies GmbH (Cologne, Germany) and annealed to make the substrates as follows. Bubble 3, Bubble 7, and Bubble 16 were assembled by annealing oligonucleotide B50 with the appropriate Bubble oligonucleotide. For the fluorescent DNA substrates, oligonucleotide B50 5Ј-Fl was annealed with oligonucleotide H50 for the 5Ј-labeled splayed duplex, oligonucleotide X50 for the 3Ј-splayed duplex, oligonucleotide H25 for the 5Ј-overhang, and oligonucleotide X26 -50 for the 3Ј-overhang, respectively.
XPB-Bax1 Activity Assays-For DNA cleavage assays with radioactive substrates, 300 nM XPB-Bax1 (wild type or mutant) was incubated with 200 nM DNA substrate in reaction buffer (20 mM Tris, pH 8.5, 100 mM glutamate, 0.1 mg/ml bovine serum albumin) in a total volume of 10 l for 20 min at 50°C. Divalent metal ions (10 mM) and nucleotide (ATP or ATP␥S, 1 mM) were added as indicated. Reactions were stopped by the addition of loading dye with EDTA, and products were analyzed on denaturing polyacrylamide TBE gels as described previously (23).
For experiments using fluorescent DNA, 1.2 M XPB-Bax1 was incubated with 1 M DNA at 45°C, with all other conditions as described above. Size markers (A and G) were prepared

XPB-Bax1, a Helicase-Nuclease Machine in NER
from labeled substrates using standard protocols. Gels were scanned in a Fuji FLA5000 imager and analyzed using Fuji ImageGauge software.
For XPB-Bax1 assays analyzed by native gel electrophoresis, assays were performed in reaction buffer containing 750 nM XPB-Bax1 and 500 nM fluorescent DNA in a total volume of 10 l for 5 or 20 min at 45°C. Manganese chloride (10 mM) and nucleotide (ATP or ATP␥S, 1 mM) were added as indicated. Reactions were stopped by the addition of 20 l of chilled stop solution (10 mM Tris, pH 8, 5 mM EDTA, 0.5% SDS, 1 mg/ml proteinase K, 5 M unlabeled oligonucleotide B50) and incubated for 30 min at room temperature. Samples were electrophoresed on a 10% native acrylamide TBE gel for 2 h and imaged using a Fuji FLA5000 imager as described above.
Electrophoretic Mobility Shift Assay-The radiolabeled DNA (10 nM) was incubated with an increasing amount of XPB-Bax1 in binding buffer (20 mM Hepes, pH 7.05, 2 mM dithiothreitol, 50 mM NaCl, 0.002% Triton X-100, and 0.1 mg/ml bovine serum albumin) in a total volume of 10 l at 50°C for 15 min. Samples were electrophoresed on a 10% native acrylamide TBE gel for 90 min and imaged using a Fuji FLA5000 imager as described above.

Cloning, Expression, and Purification of the XPB-Bax1
Complex-The genes encoding S. solfataricus XPB (sso0473) and Bax1 (sso0475) were amplified together by PCR and cloned into the vector pEHISTEV. This allowed expression of both genes as an operon, with the XPB protein expressed with an N-terminal polyhistidine tag that is cleavable by Tev protease (21). The proteins were purified by immobilized metal affinity chromatography and gel filtration as described under "Experimental Procedures." The N-terminal polyhistidine tag was removed from the XPB protein by cleavage with TEV protease prior to gel filtration. XPB and Bax1 copurified on each column as a complex with an apparent stoichiometry of 1:1, as shown previously (16,19). Mutated variants with inactivated XPB or Bax1 were prepared as described under "Experimental Procedures" and purified as for the wild type protein complex.
The XPB-Bax1 Complex Catalyzes the Metal-dependent Cleavage of a Model NER Substrate-The XPB helicase in eukarya is thought to bind to DNA bubbles formed during transcription initiation or NER and to extend these bubbles by acting as a helicase or ATP-dependent conformational switch. We therefore tested the ability of the XPB-Bax1 complex to cleave a DNA duplex containing a 7-nucleotide centrally placed unpaired region (Bubble 7) (Fig. 1A). When ATP was present to support XPB activity, cleavage by Bax1 was observed 4 -6 nucleotides 5Ј of the ssDNA/dsDNA junction. This activity was lost when the predicted active site residue Asp 301 of Bax1 was mutated to an alanine, confirming that the nuclease activity was specific for Bax1. DNA cleavage was also dependent on the activity of the XPB helicase, because no cleavage was observed either in the absence of ATP or when using a XPB variant with a mutation in the Walker A box (K96A). These data suggest that XPB and Bax1 function together as a helicase-nuclease partnership to unwind and cleave NER-type DNA substrates.
Reactions were carried out with a range of divalent metal ions to test the metal dependence of Bax1. Cleavage activity was detected in the presence of magnesium, manganese, and cobalt, with barely detectable activity in the presence of zinc and no activity in the presence of calcium or nickel (Fig. 1B). Overall, manganese yielded the highest nuclease activity. This spectrum of metal ion dependence is typical of the nuclease superfamily, consistent with the classification of Bax1 as a metal-dependent nuclease based on bioinformatic analysis (18). The three main cleavage sites were located 4 -6 nucleotides into the duplex on

XPB-Bax1, a Helicase-Nuclease Machine in NER
the 3Ј-side of the bubble (black arrows). This suggests that XPB-Bax1 extends the unpaired region upon binding.
We next tested XPB-Bax1 against model DNA substrates with a range of bubble sizes (Fig. 2). No activity was observed against duplex or single-stranded DNA (data not shown). In the presence of magnesium and ATP, XPB-Bax1 cleaved bubbles of 7 and 16 nucleotides. The higher activity supported by manganese allowed detection of cleavage activity against a 3-nucleotide bubble substrate. All three bubble substrates had three cleavage sites in common (indicated by black arrows). This is consistent with a specific binding site size for XPB-Bax1 directed by the unpaired DNA of the bubble. For Bubble 16 and to a lesser extent Bubble 7, cleavage was observed further into the duplex region (white arrows), consistent with extension of the bubble by the XPB helicase.
The Helicase Activity of XPB Directs DNA Cleavage by Bax1-To confirm that DNA unwinding by XPB directs Bax1 cleavage, we looked in detail at XPB-Bax1 cleavage of the Bubble 16 substrate (Fig. 3). Unlike Bubble 7, this substrate was cleaved in the absence of XPB activity (either by omitting ATP, by substituting the non-hydrolyzable analogue ATP␥S, or with the Walker A box mutant of XPB). Tellingly, under these conditions, cleavage was confined to a single site at the junction of ssDNA and dsDNA on the 3Ј-side of the bubble (black arrow). When ATP was included in the reaction, the main site of cleavage (representing 85% of the cleavage products) was observed to shift 4 -6 nucleotides 3Ј in the cleaved strand, as shown previously. These data suggest that the Bubble 16 substrate is large enough to allow binding and cleavage by XPB-Bax1 in the absence of XPB activity and suggest that Bax1 cleaves near the ssDNA/ dsDNA junction. The known 3Ј to 5Ј directionality of XPB is consistent with binding on the bottom strand of the bubble, as shown in the schematic in Fig. 3 and discussed below.
The Roles of XPB Domains in XPB-Bax1-The structure of archaeal XPB revealed two helicase motor domains (HD1 and HD2), with an N-terminal DRD and a Thm domain arising from within HD2 (Fig. 4A). A conserved RED motif was also described as important for helicase activity. By analogy with other 3Ј to 5Ј helicases, strand separation by XPB is likely to occur close to helicase domain 2 (24). The most likely motif implicated in strand separation is therefore the Thm domain, with single-stranded DNA dragged across of the tops of the two motor domains, bringing it close to the position of the RED motif. Both the Thm domain and the RED motif have recently been implicated as important for binding of TFIIH to DNA damage sites (12).
The RED motif is not completely conserved across all archaeal XPB sequences (supplemental Fig. 1), and in fact the consensus sequence in archaea is ERXDG. Accordingly, to assess the importance of this motif, we made the double mutant E204A/R205A. The Thm domain was entirely deleted (mutant ⌬Thm) by removing amino acids Asp 348 -

XPB-Bax1, a Helicase-Nuclease Machine in NER
Val 406 inclusive, joining amino acids 347-407 via a glycine residue. The N-terminal DRD (residues 1-53) was removed by the introduction of an NcoI site, including a new start codon at position 54 and subcloning to generate the ⌬DRD mutant. The mutant XPB variants were co-expressed with Bax1 as for the wild type protein. Each mutant formed a stable complex with Bax1 (Fig. 4B).
We next tested the ability of the mutant proteins to unwind and cleave substrates (Fig. 5). With Bubble 7, the ⌬Thm and ⌬DRD mutants showed little or no Bax1 cleavage activity (Ͻ10% of wild type activity), similar to the situation where the helicase activity of XPB was disrupted, suggesting that these mutations interfere with the correct functioning of XPB in some way. The ⌬RED mutant showed decreased but detectable substrate cleavage (40% of wild type activity), suggesting that this motif is involved in but not essential for the function of XPB in this context. More informative results were obtained for the Bubble 16 substrate (Fig. 5B). With this larger bubble, we showed previously that the helicase activity of XPB was not required for Bax1 cleavage near the ssDNA/dsDNA junction. The ⌬RED mutant had activity comparable with the wild type protein (75%), with ATP-dependent cleavage sites introduced in the duplex region 3Ј of the bubble, suggesting that the XPB helicase was at least partially active. In contrast, the ⌬Thm mutant did not support this invasion of the duplex region, suggesting that helicase activity was abrogated. Instead, there was strong cleavage within the ssDNA bubble (gray arrow) as well as at the ssDNA/dsDNA boundary. This suggests that the loss of the Thm domain has affected the ability of XPB to position Bax1 correctly at the DNA junction. Together, these observations are consistent with a role for the Thm domain at the DNA unwinding site, potentially acting as the "wedge" or "plowshare" that physically separates the duplex DNA. Finally, the ⌬DRD mutant displayed very little cleavage activity (Ͻ2% of wild type activity) either in the presence or absence of ATP, suggesting a fundamental role in XPB-Bax1 function.
The binding affinities of XPB-Bax1 variants for ssDNA and the Bubble 7 substrate were tested by electrophoretic gel mobility shift analysis (Fig. 5C). Previously, we demonstrated that S. solfataricus XPB bound relatively weakly to ssDNA, with an apparent dissociation constant of about 1 M (16). By contrast, the XPB-Bax1 complex bound ssDNA an order of magnitude more tightly, with an apparent dissociation constant of about 100 nM (Fig. 5C). Slightly weaker binding (apparent K D of ϳ250 nM) was observed for the ⌬DRD and ⌬Thm mutants. The Bubble 7 substrate was bound with broadly comparable affinity (apparent K D values around 200 nM) by the wild type, ⌬DRD, and ⌬Thm enzymes, whereas dsDNA was bound much more weakly by all three proteins (K D Ͼ 2 M; data not shown). The  With the Bubble 16 substrate, Bax1 cuts at the ssDNA/dsDNA junction when XPB helicase activity is inactivated due to lack of ATP or point mutation K96A (black arrows). When XPB is active, the point of cleavage moves 4 -5 nt 3Ј away from the edge of the bubble, consistent with DNA strand opening by XPB (gray arrows). No activity was observed when the Bax1 nuclease was inactivated by mutation (D301A). Lane C, control, DNA alone. enhanced binding affinity of XPB-Bax1 compared with XPB alone may be due either to ssDNA binding by Bax1 or to an alteration in the ssDNA binding properties of XPB when in complex with Bax1. These data rule out the possibility that the loss of activity observed for the ⌬DRD mutation has resulted from a gross defect in DNA binding by the XPB-Bax1 complex. Potential roles for the XPB domains are examined in more detail under "Discussion." Finally, we compared the ability of the wild type and mutant XPB-Bax1 complexes to cleave model substrates, including splayed duplexes and single-stranded DNA overhangs (Fig. 6).
Using native gel electrophoresis to analyze DNA products, we observed cleavage of splayed duplex substrates within the duplex region next to the 5Ј-arm by wild type XPB-Bax1 (Fig.  6A). This is consistent with the activity seen against bubble substrates. A model substrate with a 5Ј-ssDNA overhang was also cleaved, yielding smaller labeled products than for the equivalent splayed duplex, suggesting cleavage at the junction between the single-and double-stranded DNA rather than within the duplex region. A 3Ј-overhang substrate was not cleaved appreciably by XPB-Bax1. Importantly, no helicase activity was observed for any substrates, even when the nuclease activity was abrogated by the ⌬nuc mutation (Fig. 6A). This suggests strongly that XPB cannot destabilize a 25-bp duplex under these conditions, which would require unwinding of only about 12 bp of DNA, but rather catalyzes local unwinding near the junction, consistent with the data obtained from bubble structures.
By repeating the assays using denaturing gel electrophoresis, we were able to map the cleavage sites introduced by XPB-Bax1 precisely (Fig. 6B). First, for the splayed duplex with a labeled 5Ј-arm, in the absence of ATP the wild type protein cleaved predominantly close to the junction (white arrow). When ATP was present, the cleavage pattern changed, with new sites within the duplex region cleaved more predominantly (e.g. black arrow). The ⌬Thm mutant did not appear to position so precisely on the junction, evidenced by the stronger relative cleavage in the single-stranded arm (light gray arrow) and showed no ATP-dependent cleavage within the DNA duplex, consistent with a loss of helicase activity as shown previously for bubble substrates. On the 5Ј-overhang substrate, no ATPdependent effect was observed, consistent with the model whereby XPB binds on the 3Ј-strand and moves 3Ј to 5Ј into the duplex. Very weak cutting of the 3Ј-flap substrate in the singlestranded region (dark gray arrow) may be consistent with the activity detected by the Kisker group (19) and could be due to Bax1 binding in the opposite orientation with respect to the junction.

DISCUSSION
Bax1, the Archaeal XPG?-The mechanism of NER in archaea has been the subject of much speculation since genome sequencing revealed the presence of eukaryal-type NER genes in the archaea (13,25). Although the intervening years have seen good progress made in studying the structures and activities of individual NER proteins in archaea, little is known about how they function together to effect damage recognition and repair. We have shown that XBP and Bax1 form a stable complex (Fig. 7A) and function together to cleave model NER substrates on the 5Ј-side of DNA bubbles. In large bubble substrates, such as Bubble 16, Bax1 can function in the absence of XPB activity to cleave the DNA at the junction of ssDNA and dsDNA. For smaller bubbles, there is not enough ssDNA available to allow Bax1 to function unless the bubble size is increased by the action of XPB. XPB activity extends bubbles by at most 6 -8 nucleotides, allowing Bax1 to engage the ssDNA and cleave it. Given the known 3Ј to 5Ј polarity of XPB, our data can only be explained by XPB binding on the bottom (undamaged) strand and extending the bubble on the "downstream" side (with respect to the lesion site), allowing Bax1 to introduce the cleavage site 3Ј of the damage, as indicated in the schematic diagrams in Figs. 1 and 3. This is supported by the observation that minimal substrates with a 3Ј-strand for XPB loading are partially unwound before cleavage, whereas substrates lacking this strand are cleaved only at the junction point (Fig. 6). Fig. 6A also emphasizes the fact that XPB does not function as a canonical helicase because there was no evidence for unwinding of any of the DNA substrates tested by the XPB-Bax1 ⌬nuc variant. This is consistent with recent data suggesting that eukaryal XPB may be functioning as an ATP-dependent conformational switch rather than a canonical helicase (11).
Archaeal Bax1 nuclease could be considered as the functional equivalent of the nuclease XPG (Fig. 7B). Initiation of this process from a DNA bubble as small as 3 nucleotides suggests that the XPB-Bax1 complex could initiate repair from very small regions of non-canonical duplex DNA (e.g. at a photoproduct site

XPB-Bax1, a Helicase-Nuclease Machine in NER
where base pairing is locally disturbed). The action of XPB in initiating the repair bubble may allow the XPD helicase to bind and extend the nascent bubble, a situation analogous to that suggested for eukaryal NER (12). Patch repair in the archaea may require the action of the 3Ј-flap nuclease XPF to complete the excision step (23,26). One key question is how XPB is directed to bind to the undamaged strand, thus directing Bax1 to cleave the damaged strand. This may be determined by the protein(s) involved in the initial DNA damage detection step, analogous to the role of HR23B-XPC in eukarya. This step of NER in archaea is still not understood, although the SSB protein has been shown capable of unwinding damaged DNA in vitro (27). Reconstitution of the archaeal NER pathway in vitro is a key future goal.
The Kisker laboratory has recently reported that Bax1 and XPB from the euryarchaeote T. acidophilum form a 1:1 complex in vitro (19). Endonuclease activity ascribed to Bax1 was only observed in the absence of XPB and was specific for 3Ј-flaps with cleavage 4 -6 nt from a dsDNA/ssDNA junction. In other words, the Bax1 activity was closer to XPF than to XPG. In contrast to our data, no endonuclease activity was detected in the presence of manganese, and bubble substrates were not cleaved (19). Sequence alignments demonstrate that Bax1 in the thermoplasmatales lacks many of the key residues conserved in other species, including the key nuclease active site motif. Given these differences and the lack of any activity for the T. acidophilum XPB-Bax1 complex, the relevance of these observations to the present study are hard to ascertain.
XPB Structure and the Role of the Accessory Domains-The A. fulgidus XPB crystal structure revealed an unusual conformation where helicase domain 2 was rotated through almost 180°with respect to helicase domain 1 compared with the "canonical" position that has been observed in all other Superfamily 1 and 2 helicase structures (17). It has been postulated that this conformational flexibility is important for the biological function of the XPB protein (12,17). However, A. fulgidus XPB was crystallized in the absence of its cognate Bax1 partner, and it is therefore possible that the unusual structure observed by Fan et al. (17) was due to unusual conformational flexibility induced by the absence of the Bax1 subunit. It has been noted previously that XPB from S. solfataricus is heat-labile and prone to aggregation in the absence of its Bax1 partner (16). A definitive answer to this question will require further analysis of the conformational flexibility of XPB in the presence and absence of Bax1.
The XPB protein structure revealed two accessory domains named the Thm domain and the DRD (17). As we have already stated, by analogy with several other helicases, the position of the Thm domain above helicase domain 2 is consistent with a role in DNA duplex opening by XPB. Consistent with this, we note a loss of helicase activity and significant defects in helicase positioning at DNA junctions when the Thm domain is deleted. A similar domain (Domain 2) has been noted in the archaeal Hef helicase. Deletion of this domain resulted in the loss of helicase activity that was ascribed to an inability to target forked DNA structures in the deletion mutant (28).
The N-terminal DRD of A. fulgidus XPB is structurally related to the mismatch recognition domain of the MutS protein and has little or no intrinsic DNA binding affinity (17). Deletion of the DRD in S. solfataricus XPB does not disrupt the interaction with Bax1 or with DNA substrates but has a profound effect upon the activity of the complex. The role in damage detection suggested by Fan et al. (17) (and implicit in the name of the domain) is not relevant in our assays and thus cannot fully explain the function of the DRD. Together, these data suggest a subtle but crucial role for the DRD, perhaps in positioning XPB-Bax1 correctly on the DNA or supporting the nuclease activity of Bax1. The presence of an equivalent DRD in eukaryotic XPB was predicted based largely on the observation of sequence similarity of around 40% between the Archaeoglobus and human sequences in this region (17). However, the most highly conserved residues of the eukaryal XPB family in this region of the protein are not conserved in the archaeal XPBs (supplemental Fig.  2). Furthermore, secondary structure prediction by the JPRED server (29) yields an accurate match for the known structure of archaeal XPB but predicts a very different secondary structure for the equivalent region of eukaryotic XPBs (supplemental Fig. 2). Finally, eukaryal XPBs have an extra N-terminal domain that is important for interactions with eukarya-specific partner proteins. Therefore, on balance, we suggest that the DRD sequence is unique to archaeal XPB and plays a vital role in the functional interaction with the Bax1 protein. In eukaryal XPB, this domain, if it exists, may well have a different fold and function.
Implications for Eukaryal NER-Eukaryal XPB is involved in both NER and transcription initiation. The observation that archaeal XPB is coupled functionally with a DNA endonuclease makes it more likely that the ancestral role of XPB was in excision repair. In the eukaryal lineage, XPB evolved a close inter- Oligonucleotide B50 with a 5Ј-fluorescein (gray hexagon) was annealed with a range of partner strands to generate minimal substrates. Approximate cleavage sites are indicated by white arrows. Cleavage products are indicated on the right of the gel. The wild type enzyme (WT) was incubated with each substrate for 5 or 20 min in the presence of ATP and MnCl 2 . The ⌬nuc mutant was incubated in the same buffer for 20 min. Control lanes are as follows. m, marker DNA showing migration of overhang and single-stranded B50 oligonucleotide; c, DNA substrate control; ⌬, ⌬nuc mutant. B, the reaction products were also run on denaturing acrylamide TBE gels with (A ϩ G) markers to map cleavage sites. All reactions were carried out in reaction buffer with 10 mM MnCl 2 and ATP where indicated. m, A ϩ G sequence markers; c, DNA alone; ⌬, ⌬nuc mutant.

XPB-Bax1, a Helicase-Nuclease Machine in NER
action with the p52 subunit of TFIIH, which is required for XPB helicase activity (10) (Fig. 7C). The ancestral form of TFIIH may thus have been co-opted by the RNA polymerase II transcriptional apparatus in the eukaryal lineage. In this scenario, the archaeal Bax1 nuclease was replaced with XPG, which interacts with TFIIH during NER (30,31).
The placement of archaeal XPB on the undamaged strand, moving 3Ј to 5Ј and thus opening up the damaged strand on the 3Ј-side of the lesion for cleavage by Bax1, also has implications for eukaryal NER (Fig. 7D). The repair complexes formed by the succession of NER proteins binding around the DNA damage site in eukarya are understood incompletely. It is clear that HR23B and XPC are the first proteins to recognize the damaged DNA in global NER and that these proteins help recruit the TFIIH complex to the damage site (32). TFIIH opens up the DNA further, displaces HR23B-XPC, and helps recruit the XPA and RPA proteins to the repair site (33). XPD has been proposed to bind to the damaged strand and play a role in confirming the presence of DNA damage, perhaps by stalling at the lesion (reviewed in Ref. 34), but direct evidence in support of this hypothesis is rather limited, and archaeal XPD has been shown recently to be capable of translocating past DNA lesions (35).
There are two fundamental options for XPB binding to DNA during NER. First, XPB could bind on the 5Ј-side of the damaged strand and thus move 3Ј to 5Ј away from the lesion, defining the boundary for the 5Ј cut by XPF-ERCC1. Second, XPB could bind the undamaged strand, in which case its 3Ј to 5Ј polarity would move it in tandem with XPD on the opposite strand toward the 3Ј-side of the repair bubble. The data for eukaryal NER are inconclusive (e.g. photocross-linking of XPB to a cisplatin lesion revealed extensive contacts with sites both 5Ј and 3Ј of the DNA lesion) (33).
In conclusion, the archaeal XPB-Bax1 complex is an interesting new example of a helicase-nuclease DNA-processing machine. Our data suggest an ancestral role for XPB in extending NER substrates by strictly limited translocation along the undamaged strand, generating a span of ssDNA on the damaged strand for attack by a structure-specific nuclease. This model is consistent with recent studies showing that the activity of eukaryal XPB is essential for the recruitment of TFIIH to damage sites. Thus, the detailed characterization of the archaeal enzymes can extend our understanding of eukaryal NER, where the additional complexity of the system has presented a barrier to detailed analysis.