Mechanism of action of nucleotide excision repair machinery

Abstract

Nucleotide excision repair (NER) is a versatile DNA repair pathway essential for the removal of a broad spectrum of structurally diverse DNA lesions arising from a variety of sources, including UV irradiation and environmental toxins. Although the core factors and basic stages involved in NER have been identified, the mechanisms of the NER machinery are not well understood. This review summarizes our current understanding of the mechanisms and order of assembly in the core global genome (GG-NER) pathway.

Introduction

Nucleotide excision repair (NER) is an essential DNA repair pathway that is responsible for eliminating a wide range of lesions from genomic DNA. Defects in NER result in the genetic disorder xeroderma pigmentosum (XP), which is characterized by an extreme hypersensitivity to sunlight and a greater than 2000-fold increase in skin cancer [1, 2]. NER is a highly dynamic process that operates through the sequential assembly and action of more than 20 proteins at the DNA lesion. Biochemical and cell biological studies of XP patient cell lines deficient in the seven NER-specific XP proteins (XPA-XPG) have led to an understanding of the basic steps of NER and individual functions of each protein [3, 4]. There are two main NER pathways, which diverge by the mechanisms of initial damage recognition and their associated spectra of disease phenotypes: (i) transcription-coupled NER (TC-NER), in which lesions are detected by RNA polymerase II stalling on the transcribed strand [5] and (ii) global genome (GG)-NER, where lesions are detected throughout the genome. The specifics of TC-NER have been covered extensively elsewhere [6–8], so this minireview focuses on GG-NER. Defects in GG-NER that lead to XP are also associated with increased incidence of internal tumors and accelerated neurodegeneration and aging in some cases [9].

Human GG-NER has been reconstituted in vitro with six core factors: XPC-human Rad23 homolog B (HR23B), transcription factor IIH (TFIIH), XPA, replication protein A (RPA), XPG and XPF-excision repair cross complementation group 1 (ERCC1) [10–12]. Although the exact order of assembly of these factors at sites of damage is not yet fully resolved, the trajectory of the core GG-NER machinery can be summarized in four phases (Figure 1). It is generally understood that XPC-HR23B recognizes the presence of damage in the DNA due to preferential interaction with destabilized DNA [13–15]. For less helix-distorting lesions such as cyclopyrimidine dimers (CPDs), the damage recognition process additionally requires the DDB1-DDB2 E3 ubiquitin ligase complex [16, 17], but this complex is dispensable for the core NER reaction in vitro. Once XPC binds to the damaged duplex, it recruits TFIIH to the lesion through specific protein-protein interactions [18, 19]. The two helicase subunits of TFIIH, XPB and XPD, fulfill key roles in NER: XPB melts the DNA duplex [20, 21], thereby allowing XPD to load and track along the damaged strand in the 5’ - 3’ direction until it stalls at the lesion, which is best explained as arising from an inability to pass through the DNA binding channel of the helicase [22–27]. A characteristic feature of lesions repaired by NER is the presence of local flexibility or helical kinks in the DNA [28]. While XPC appears to sense increased DNA strand flexibility induced by a lesion and does not directly interact with the lesion, XPD directly interacts with and validates the presence of the lesion through its translocation on the DNA. Importantly, the two steps of damage recognition and validation are independent of the exact chemical structure of the lesion, providing a molecular basis for the very broad substrate specificity of NER. After creation of an unwound ‘bubble’ in the DNA and assembly of the XPA-RPA scaffold, the XPG and XPF-ERCC1 nucleases are recruited. XPF-ERCC1 incises the DNA 5’ to the lesion, replication proteins initiate gap-filling synthesis, and following 3’ incision by the XPG nuclease, filling of the gap is completed and the resulting nick sealed [29–32].

Figure 1. The trajectory of dynamic assembly of the core GG-NER machinery.

Four steps are identified: (A) damage recognition, (B) damage verification, (C) pre-incision, and (D) dual incision.

Phase 1: Recognition of DNA damage

XPC functions as the key initiator of damage recognition in GG-NER [33, 34], ultimately leading to the opening of duplex DNA around the lesion and recruitment of additional NER factors [35, 36]. XPC is found tightly associated with HR23B [37] and Centrin 2 [38]. Although XPC alone is sufficient in cell-free NER systems for the damage recognition step, both HR23B [39, 40] and Centrin 2 [41] significantly enhance the activity and stability of XPC.

XPC-HR23B binds tightly to duplex DNA that is destabilized by base modifications, adducts and other distorting lesions. The first insights into how XPC-RAD23B binds DNA were provided by a crystal structure of the yeast orthologue Rad4–Rad23 complex bound to duplex DNA containing a cyclobutane pyrimidine dimer (CPD) lesion [42]. Yeast Rad4 inserts a β-hairpin into the DNA major groove such that the damaged base pairs are flipped out of the helical stack and make no direct contacts with the protein. Instead, Rad4 interacts with the undamaged strand 3’ to the lesion. This suggests an indirect recognition mechanism in which XPC senses DNA destabilization rather than direct recognition of a specific lesion type, enabling NER to act on a wide range of bulky and non-bulky lesions. Following initial damage detection by XPC, additional factors verify the presence of the lesion prior to incision (vide infra).

While NER acts on UV-induced DNA cross-links (CPDs or 6-4 pyrimidine-pyrimidone photodimers (6-4PPs)), XPC is inefficient at detecting lesions like these that lead to only minor distortions of the DNA helix [43, 44]. In this case, the initial recognition of damage involves the UV DNA damage binding (UV-DDB) complex, a heterodimer of two subunits, DDB1 and XPE (DDB2). XPE interacts directly with damaged DNA [17, 45, 46] and in fact exhibits a higher affinity and specificity for CPDs than XPC [47]. In contrast to XPC, XPE inserts a β-hairpin into the minor groove of the duplex and directly contacts the damaged strand [48, 49]. UV-DDB is part of the larger CUL4-RBX1 E3 ubiquitin-ligase [50] that ubiquitinates several DNA binding proteins including XPC, histones and UV-DDB itself. Ubiquitination of XPE reduces its affinity for damage sites but has no effect on the affinity of XPC, thus facilitating the handoff of the lesion from UV-DDB to XPC and subsequent recruitment of TFIIH [17, 51]. UVDDB-CUL4 also ubiquitinates histones H3 and H4, which in turn improves accessibility of chromatin to XPC and other NER factors [52].

After initial lesion recognition, XPC recruits TFIIH to the lesion via interactions with the p62 and XPB subunits of TFIIH (Table 1) [18, 53]. A cryo-EM structure of an ‘initiation complex’ containing the yeast homologues of XPC-HR23B-Centrin 2 (Rad4-Rad23-Rad33), plus the seven-subunit TFIIH core and a dsDNA substrate containing a carcinogenic 2-acetylaminofluorene adduct, which confirms that TFIIH binds two distinct motifs in the yeast Rad4 N- and C-terminal regions (Figure 2A) [54]. The duplex DNA is locally unwound and positioned between Rad4 and the Ssl2 (yeast XPB) helicase subunit of TFIIH, with Rad4 on the 3′ and TFIIH on the 5′ side of the lesion. Interestingly, the dsDNA is not yet engaged by the XPD helicase in this structure, indicating that additional configurational changes in TFIIH must occur as the DNA unwinding and lesion verification proceeds.

Table 1.

Summary of interactions of the core NER factors.

Protein	Interacting Residues or Subunits	Interacting Partner
XPC	154-331, 492-940	XPA
	494-741	RAD23B
	606-742	DNA
	816-940	TFIIH (XPB and p62)
	847-866	Centrin 2
TFIIH	p62 1-108, XPB	XPC
	XPD, p44, p62, XPB	DNA
	XPD, XPB and p52	XPG
	p8, p52	XPA
XPA	29-46	RPA32C
	67-77	ERCC1
	98-129	RPA70A
	98-239	DNA
	153-176	RPA70B
	161-170	PCNA
	185-226	DDB2
	226-273	TFIIH (p8 and p52)
	unknown	XPC-RAD23B
RPA	RPA 14 : 70-121	DNA
	unknown	XPG
	RPA 32 : 40-174	DNA
	RPA 32 : 237-270	XPA
	RPA 70 : 181-442	DNA
	RPA 70 : 236-382	XPA, XPF
XPG	1-65	TFIIH (p44 and p62)
	unknown	RPA
	unknown	XPB
	unknown	XPD
	unknown	DNA
XPF	1-824	RPA
	824-905	ERCC1
	unknown	DNA
ERCC1	67-80	XPA
	96-227	DNA
	224-245, 293-297	XPF

Figure 2. Cryo-EM structures of TFIIH complexe.

(A) Yeast TFIIH and XPC-HR23B-Cen2. (B) Human TFIIH and XPA.

Phase 2: DNA unwinding and verification of the site of damage

TFIIH is a central element of NER responsible for unwinding the duplex DNA around the site of damage and verifying the presence of the lesion. It contains ten subunits with a seven-subunit core complex and a three-subunit cyclin-dependent kinase (CDK) activating kinase (CAK) module. The core complex contains the XPB and XPD ATP-dependent helicases and five non-enzymatic subunits (p62, p52, p44, p34, p8). The CAK module is comprised of CDK7, cyclin H, and MAT1 subunits. NER requires only the core complex, using the two critical XPB and XPD helicases to extend the initial opening of the duplex by XPC and ultimately generate the 24-30 nucleotide asymmetric NER bubble around the lesion [55–58].

X-ray crystal structures of XPB [59, 60] and XPD [22–24, 27] have been reported. XPB contains two RecA-like domains, a DNA-damage recognition domain (DRD) domain, and an N-terminal extension domain (NTE) crucial for activity [59]. XPB is a dsDNA translocase that binds opposite to the damaged strand and unwinds the duplex in the 3′ to 5′ direction [59]. It is also involved in the recruitment of TFIIH by XPC [61]. XPD contains two RecA-like domains (RecA1 and RecA2) with conserved helicase motifs, a 4FeS cluster domain, and an additional Arch domain [22]. One of the most remarkable structural features of XPD is the presence of a DNA binding pore created by the Arch domain, the FeS domain and the RecA1 domain [62]. XPD, a ssDNA translocase, binds 5’ to the lesion on the damaged strand and moves along the DNA in the 5’-3’ direction until it stalls at the damaged site. Thus, XPD serves as the critical damage verification element [63, 64]. While both helicases are crucial for NER, their activity is also influenced by the presence of non-enzymatic TFIIH subunits [65]; XPB activity is upregulated by the p52 subunit [66] and XPD by interactions with the p44 subunit [67]. Recent studies have also shown that p8 is an essential factor for duplex opening, perhaps through its role in recruiting XPA [68, 69].

Structures of TFIIH determined by cryo-electron microscopy (EM) have revolutionized understanding of its mechanism of action [54, 70–73]. Among these, the structure of human TFIIH in the absence of substrate revealed the overall horseshoe-shaped architecture that is dominated by the side-by-side orientation of the helicase subunits, enforced by their extensive interfaces [71]. Additional contacts to other subunits include p52 and p8 with XPB, and p62, p44, and MAT1 with XPD. This network of inter-subunit interactions likely serves to modulate XPB and XPD activity [70, 71]. For example, the p62 subunit sterically interferes with and blocks the DNA binding pore of the XPD RecA1 domain [71, 74]. MAT1 is a non-globular protein that forms extensive contacts with the XPD Arch domain and tightly regulates its function [75, 76]. The long α-helix of MAT1 also separates XPB and XPD and reduces the inter-domain mobility in XPD. A high-resolution crystal structure of the Arch domain of XPD in complex with MAT1 shows that MAT1 renders XPD inactive by shielding residues essential for its helicase activity, affecting the DNA binding of XPD and interactions of XPD with XPG [76]. These observations suggest that the p62 and MAT1 subunits need to be displaced during NER to allow more conformational flexibility for XPD to bind and translocate along DNA for lesion verification.

Another breakthrough cryo-EM structure was that of TFIIH with XPA on a model NER bubble substrate, which provided many additional insights into the mechanism of action of TFIIH during NER (Figure 2B) [72]. This structure clarified the drastic shift in the arrangement of TFIIH subunits needed for DNA unwinding [71], including displacing the CAK module that ultimately allows XPD to engage ssDNA. In addition, this structure provided insight into how XPA enhances TFIIH specificity [64, 72, 77], as well as pointed to a potentially critical role of XPA as a wedge that pries apart base pairs as the helicases ‘pull’ the duplex DNA forward, as proposed by Tsutakawa et al.[78]. This structure also suggests that XPA enhances XPB activity by bridging the XPB ATPase lobes to the p52 and p8 subunits and facilitates DNA opening through contacts with XPD and ssDNA. Even though the cryo-EM complex of TFIIH was formed in the presence of XPA and XPG, XPG was not visible in the structure. However, its location near the XPD-MAT1 interaction site was inferred by chemical crosslinking, suggesting that XPG competes with MAT1 to interact with XPD [72]. These results in turn suggest that XPG can facilitate lesion scanning by blocking the kinase module interaction site on XPD, which would significantly enhance XPD’s helicase activity, consistent with the report that XPG greatly stimulates XPD [72].

These cryo-EM structures have informed models to explain how TFIIH unwinds DNA and transitions from the initial destabilization of the duplex to the fully opened NER bubble. In a structure of yeast TFIIH and XPC in complex with a lesion-containing duplex, XPC secures both the 3’ end of the DNA and TFIIH to prevent DNA rotation (Figure 2A) [54]. With the DNA held in place by XPC, XPB is then able to exert torque on the DNA duplex as it translocates in an ATP-dependent manner. The β-hairpin unit BHD3 of XPC positions itself between the two DNA strands leading to strand separation and facilitating unwinding. In this configuration, the dsDNA is far from XPD and the DNA binding pore of XPD is blocked by a plug element, hence XPD is unable to translocate. In contrast, in the structure of the complex with XPA, the plug is displaced so XPD can translocate on the ssDNA (Figure 2B) [72]. This transition requires a significant movement of the XPD and p44 subunits of ~80 Å in the 3’ direction to enable binding of ssDNA in the XPD DNA binding channel. The ssDNA threads along the DNA channel in close proximity to the FeS cluster and residues lining the pore interact with the sugar-phosphate backbone and aromatic residues proof-read DNA bases [72]. The ATPase motors in XPD drag the ssDNA through the channel to scan for lesions [79]. When XPD encounters a bulky lesion, both XPB and XPD are stalled and XPA was found to significantly enhance this lesion-induced stalling [62, 64].

Phase 3: Completing assembly of the pre-incision complex

Following DNA unwinding and lesion verification, the NER scaffolding is stabilized and the pre-incision complex (PIC) is formed [34, 35, 81]. As noted, current evidence suggests that XPA is recruited first by TFIIH to aid in generating the NER bubble [35, 64, 82, 83], although some studies suggest that XPA and RPA are recruited together [84, 85] and one early study suggested that RPA can be recruited to the damage site without the presence of XPA [86]. In all models, XPG is recruited through interactions with TFIIH and RPA, displacing XPC to complete formation of the PIC [81, 87, 88].

XPA and RPA play crucial roles to coordinate and ensure the accurate positioning of proteins in the PIC. Despite its small size (273 residues), XPA interacts with many NER factors, including TFIIH, RPA, XPC-HR23B, DDB2, and XPF-ERCC1 [89], as well as proliferating cell nuclear antigen (PCNA) required for gap-filling synthesis [13, 90–95]. XPA has a central globular DNA binding domain (DBD) spanning residues 98 to 239 [96] with a zinc binding motif and unordered, flexible N-terminal (1-98) and C-terminal (240-273) regions that mediate protein-protein interactions with other NER factors. XPA is positioned at the ss–dsDNA junction of the NER bubble [97, 98]. Most structural models place XPA at the junction 5′ to the lesion as it is known to interact with endonuclease XPF-ERCC1 [72, 90], although we have recently shown that binding to both the 5’ and 3’ junction is sterically feasible and proposed there may be switching between junctions over the course of the dynamic NER trajectory [99].

The XPA-RPA scaffolding stabilizes the PIC and is essential for NER [85, 99, 100]. RPA is a heterotrimer of RPA70, RPA32 and RPA14 subunits. It contains four DNA binding OB-fold domains (70A-70B-70C-32D) that can bind up to 30 nucleotides, which coincides with the size of the NER bubble [31, 101–103]. RPA binds to the undamaged strand of the NER bubble, protecting it from incoming nucleases [104, 105]. XPA and RPA make contact at two distinct sites: a stronger site that involves a motif in the XPA N-terminal domain and the RPA32C winged-helix protein recruitment domain [106], and a weaker site involving the tandem high affinity ssDNA binding domains RPA70AB and the XPA DBD [99].

XPG is understood to be recruited to the PIC through interactions with TFIIH and RPA (Table 1). XPG is a member of the FEN-1 (flap endonuclease) family of nucleases. The XPG active site consists of the highly conserved N-terminal (N)- and internal (I)-nuclease region separated by a large spacer region of 600 residues, the extended length of which is unique to XPG [107, 108]. This spacer region influences the substrate preference of XPG for ss-dsDNA junctions [109–112] and mediates essential interactions with the XPB, XPD, p62, and p44 subunits of TFIIH [111, 113, 114] and RPA [115–117]. The presence of XPG, but not its nuclease activity, was shown to stabilize the NER bubble and facilitate the first incision 5’ to the lesion by XPF-ERCC1 [118].

Phase 4: Dual incision

Removal of the damaged DNA depends on the concerted action of the two structure-specific endonucleases XPF-ERCC1 and XPG. These nucleases were first identified as part of the NER pathway but are now known to act in multiple DNA transactions [108, 119–121]. In NER, XPF-ERCC1 and XPG make incisions 5’ and 3’ to the lesion on the damaged strand, respectively [122, 123]. This dual incision and removal of approximately 24-30 nucleotides in the damaged strand is accompanied by gap filling synthesis, which completes the repair process. Because these nucleases introduce dangerous nicks in the DNA, their dual action is carefully coordinated by interactions by post-translational modifications and interactions with other NER factors, which regulate the order of assembly and catalytic activity during NER.

The order of assembly at the lesion and the incisions by XPF-ERCC1 and XPG were debated for many years, with conflicting studies conducted using multiple approaches including in vitro reconstitution and in vivo cell biological studies, different substrates, and various combinations of NER proteins [124–131]. In the currently accepted model based on the consensus of biochemical and biological studies, XPG is recruited before XPF-ERCC1, even though XPF-ERCC1 incises the damaged strand first; the XPF-ERCC1 5’ incision requires the presence of XPG but not its catalytic activity [129]. Because XPG catalytic activity is also not required for initiation of gap-filling repair synthesis starting from the exposed 3’-OH generated by XPF-ERCC1, it is generally understood that repair synthesis initiates prior to the 3’ incision by XPG [127]. Strengths of this model include high-fidelity to appropriately license and regulate the dual incision while minimizing potential DNA breaks due to exposure of the undamaged strand of ssDNA.

The XPF-ERCC1 complex is likely recruited through interactions between XPA and the central domain of ERCC1 (Table 1) [90, 132, 133] to form the incision complex. XPF is a member of the Mus81 nuclease family, with an archaeal SF2 helicase-like domain in the N-terminus, a central nuclease domain containing a conserved V/IERKX₃D motif [134], and two C-terminal helix-turn-helix (HhH) domains (HhH₂) [135]. Similarly, ERCC1 contains a central domain with high structural homology to the XPF nuclease domain, and two C-terminal HhH domains [135]. XPF functions as an obligate heterodimer with ERCC1 [136–140]. Dimerization is mediated by interaction between the HhH₂ domains in each protein and is required to stabilize both proteins for efficient nuclease activity [136, 137, 141, 142]. The presence of RPA and direct interaction with XPA also enhance incision activity [115, 122, 143].

XPF-ERCC1 exhibits a preference for DNA junction substrates [115, 122, 124], with DNA binding mediated by the HhH₂, XPF nuclease, and ERCC1 central domains [144]. In the current model for DNA binding, the HhH₂ domains of XPF and ERCC1 each bind ssDNA near the 5’ junction with the ERCC1 central domain binding specifically to the 5’ arm of ssDNA [142, 145–147]. A recent cryo-EM structure of XPF-ERCC1 bound to a stem-loop DNA substrate suggests a model in which DNA binding relieves an auto-inhibited state where the XPF helical domain blocks the active site [148]. This auto-inhibited state would inhibit nuclease activity until XPF-ERCC1 is stabilized by the NER scaffold and the appropriate DNA substrate is bound. With XPG already in place, the incision complex would then be fully assembled and the XPF-ERCC1 nuclease activated.

The structural and biochemical characterization of the XPG catalytic domain [149] suggests that XPG has a similar structure and function to the FEN superfamily with a conserved active site consisting of carboxylates in a β-sheet core and basic residues near a pair of gating helices [150]. These active site gating helices play a key role in activating the XPG nuclease, by undergoing a disorder-to-order transition induced by DNA binding [151–153]. Interestingly, XPG can exist as a stable homodimer that could bind both ss-ds DNA junctions in the NER bubble, but with only one subunit positioned for its 3’ incision activity [149]. During NER, the ‘other’ subunit could be displaced by TFIIH [87]. An alternate model positions the second XPG subunit with XPF-ERCC1 on the same dsDNA arm, which could serve to regulate the timing of the 3’ incision after the 5’ incision by XPF-ERCC1.

Overview

The long-term goal of biochemical and structural NER research is to elucidate how this complex multi-protein machine recognizes and repairs DNA lesions in our genome. A substantial amount of data has been accumulated over many years, including exciting new 3D structures determined by single particle cryo-EM that provide unprecedented views of the molecular mechanisms of NER factors. While providing fertile ground for hypotheses about NER function and the search for potential NER-targeted chemotherapeutics, these studies fall short of the ultimate objective to truly understand the mechanism of action of the NER machinery. Nevertheless, the outlook for the future is bright. With the development of systems to produce NER factors and a seemingly endless variety of NER substrates, the prospect of building a series of complexes that mimic the NER trajectory is clearly at hand. Characterization of these complexes by a broad range of biochemical, structural and biophysical techniques is anticipated to bring us much closer to our long sought-after objective and provide invaluable information for future efforts directed to NER-targeted drug discovery and advancing the predictive power of NER-focused cancer precision medicine.

Perspective.

(i). The importance of the field

Fundamental knowledge of the mechanism of action of NER machinery is highly desirable as a means to design new therapeutic strategies to inhibit upregulation of NER activity in response to anti-cancer therapies that function by damaging DNA.

(ii). Summary of the current thinking

Despite many years of study and revolutionary new 3D structures that have advanced understanding of the assembly and activity of NER factors, the mechanism of action of the finely tuned NER multi-protein machinery remains an enigma.

(iii). Future directions

Recent progress in the production of the primary factors and substrates required to reconstitute NER in vitro presents tantalizing new possibilities for generating and structurally characterizing the complete trajectory of the multi-step nucleotide excision repair pathway.