Excision repair in mammalian cells.

There are two types of structural anomalies that lead to mutation, a permanent change in DNA sequence. The first class involves normal bases in abnormal sequence context (mismatch, bulge, loop). The second class, which is referred to as DNA damage or DNA lesion, involves abnormal nucleotides (modified, fragmented, cross-linked) in normal sequence context. DNA lesions, in addition to causing mutations, also constitute replication and transcription blocks. Both types of structural anomalies are rectified by a series of enzymatic reactions referred to by the general term DNA repair (1-5). The repair reactions employed for correcting mismatches and lesions are similar in principle. The incorrect or damaged base is removed either as a base (base excision) or as an (oligo)nucleotide (nucleotide excision), the single-stranded gap resulting from the excision reaction is filled in by a polymerase (repair synthesis), and the newly synthesized DNA is ligated. Hence, there are two basic assays for measuring repair (6): the "incision/excision assay" and the "repair synthesis assay."

damage in 12-13-nt-Iong oligomers and eucaryotes excise 27-29nt-long fragments. This dual incision activity is referred to as excision nuclease (excinuclease), The single-stranded gap generated by either type of excision is filled in by DNA polymerases and sealed by ligase.

Genetics of Excision Repair
The excision repair genes (uvrA, uvrB, and uvrC) of E. coli show no homology to the human excision repair genes (1). In contrast, the sequences of excision repair genes in mammalian cells and yeast are highly homologous, and the enzymology of excision repair in these two systems is very similar (1,3). Only mammalian excision repair will be covered in this review. Three human diseases are caused by a defect in excision repair (9): xeroderma pigmentosum, Cockayne's syndrome, and trichothiodystrophy.
Xeroderma pigmentosum patients suffer from photosensitivity, photodermatoses including skin cancers, and in some cases from neurological abnormalities. XP patients are defective in excision repair. Mutations in 7 genes, XPA through XPG, cause XP. In addition, there is a group of patients with classic symptoms of XP but with normal excision repair. These are called XP variants (XP-V). Cells from XP-V patients are moderately sensitive to UV light but excise UV photoproducts at a normal rate and are defective in a biochemically ill defined phenomenon called postreplication repair (10).
Cockayne's syndrome patients suffer from growth failure, mental and neurological abnormalities, cataracts, dental caries, and photosensitivity and related dermatoses. Mutations in two groups of genes appear to cause Cockayne's syndrome. The CS-A and CS-B (ERCC-6) mutants exhibit classical CS symptoms without an increased rate of skin cancer. Cells from these patients have near normal UV sensitivity. A second group of patients manifest XP symptoms in addition to CS symptoms. Patients in this group have mutations in the XPB, XPD, or XPG genes.
Trichothiodystrophy (TTD) patients have ichthyosis and brittle hair and suffer from photosensitivity, skeletal abnormalities, and mental retardation. The patients mayor may not have an increased rate of skin cancer. Mutations in three genes are associated with TTD. In the XPITTD overlapping syndrome, the mutation is in either XPB or XPD. In classical TTD (TTD-A), the mutation is presumably in one of the other subunits of TFIIH (1).
In addition to the 9 genes identified by human diseases to be involved in excision repair, many rodent excision repair mutants have been isolated and characterized in order to define the entire set of excision repair genes (11,12). The rodent mutants fall into 11 complementation groups, and the majority of these correspond to human XP and CS complementation groups as indicated. In fact, some of the human XP genes were cloned by virtue of complementing rodent mutant cell lines and hence are also referred to as excision repair cross complementing (ERCC) genes. Ofthese genes, XPE and ERCC6 through ERCCll are not required for the basal excision reaction (13).

Structure and Function of Excision Repair Proteins
Table I summarizes some of the properties of excision repair proteins. Most of these proteins are in complexes in vivo, and hence the activity associated with a solitary protein in vitro mayor may not be relevant to its function in excision repair. Human excision nuclease has been reconstituted in a defined system by mixing six highly purified polypeptides or polypeptide complexes (13).
XPA-This protein of 31 kDa has a zinc finger and is involved in damage recognition (14). It also interacts with several other components of excision repair and hence may function as a nucleation factor for excinuclease. XPA interacts through its N-terminal domain with the ERCCI-XPF heterodimer (6) to form a relatively stable complex (15,16); it also binds to TFIIH through its Cterminal domain (17). Finally, RPA (HSSB) binds to XPA and increases its specificity for damaged DNA (18). In addition to XPA, 15915 This is an Open Access article under the CC BY license. Mismatch a n d damage excisio n repair. For sim plicity a deamina te d C is t aken as an exam ple of both mism atch an d damage, Th e re pair of mism atch and lesions by base excis ion follows the same pathway, Removal of dam age by nucl eotid e excis ion occurs by du al in cision, whereas re mova l of mis matches by nucl eotide excision occurs by endonu clea se/exonuclease acti on. Th e size of rep ai r patches are 1-4 nt for base excisio n (66, 67), 27-29 nt for dam age rep air by nucl eotide excision (7), and 300 -500 nt for mis match re pa ir by th e nucl eotide (general) mism atch re pair sys tem (2), other proteins that specifically bind to damaged DNA have been identified. One of these, the DDB protein , is abs ent in some XPE patients (19 -22). However, this protein is not required for excision repair (13), and it s relation to XPE gene is uncl ear at pr esent . Another class of pro tein s that bind to certain types of dam aged DNA have the HMG (h igh mobility group) domain (23,24); however, these proteins inhibit excision repair (25).
RPA(HSSB)-This trimer of (p70M p34M p ll)1 is essentia l for DNA replication (26,27) and for repair syn thes is (28). It is also abs olute ly required for the dual incision ste p of excision repair (13). It binds to damaged DNA with moderate affinity and mak es a compl ex with XPA that binds to lesion s with hi gher affinity th an eithe r compon ent alone (18).
TFIIH-This is a multiprotein (p89, p80 , p62 , p44, p41 , p38, p34 ) complex that cont ain s XPB (p89) and XPD (p80). TFIIH was initially identified as one of th e seve n gen er al transcription factors required for bas al tran scription by RNA polym er ase II (29). Th e accidental discovery that it s p89 subunit is identical to XPB (30) and the unexpected finding of failure of XP-B and XP-D muta nt cell-free extracts to compl em ent in excision assay (6) led to the event ual realization that the entire TFIIH compl ex is a repa ir factor (31,32). XPB and XPD proteins are DNA-dependent ATPases, have the so-called helicase motifs, and they, a s well as TFIIH itself (29), can dissociate short fragm ents anneal ed to single-stranded DNA (1, 3). Thi s modest helix u nwinding acti vit y is referred to a s helicase by som e in vestigators.
XPC-The sequenc e of the gene pr edicts a protein of 125 kD a . Thi s polyp eptide co-purifies with a protein of 58 kDa , which is th e human homolog of the yeast Rad23 protein (HHR23B). Th us, the functional form of XPC is a (p125)1(p58)1 het erodimer (33). XPC het erodimer binds to TFIIH loosely (31) a nd binds very ti ghtly to single-stra nded DNA (33).
ERCCI -XPF Complex-This is a ver y stable compl ex such that a heterodimer formed with a mutant protein does not exchange subunits in vitro and, as a re su lt , cell-free ext racts from XP-F and ERCC-1 do not complement for excinuclease activity (6). XPF (120 kDa) and ER CC1 (33 kDa) make a complex with (p1l2)1(p33)1 2 C.-H. Park a nd A. Sancar , u npu blish ed observation, 3 A. Kazantsev an d A. Sa ncar , unpublished observation.

Mechanism of Excision Repair
Th e three forma l steps of excision repair a re dam age recogniti on, dual incision (excision) , an d repair synthesis and ligation.
Damage Recognition-Excision repair was first ide ntified by the fai lure of UV-sensitive E, coli and human cells to re move thymine dimers from DNA. However , this re pa ir sys te m is not specific for UV dam age as it excise s all covalent DNA lesions test ed (37)(38)(39). With regard to substrate recognition an d preference, three in terre lated qu estions mu st be ad dressed. Does the enzyme system recogn ize only DNA with damaged bases, how does th e enzy me "know" which strand sho uld be cut, and fina lly , what is the molecula r ba sis for recognition?
Fir st , da maged bases are not the sole substrate for the enzyme . Huma n excinuclease excises mism atched ba ses a nd 1-3-nt loops a s well (38). However , in contrast to the true mis ma tch re pa ir sys te m , the excinuclease apparently h as no way of discriminatin g the correct a nd incorrect strands and a s a conseq uence excises the mismatch from either strand.
Rega rd ing the problem of ide ntifying the dama ged strand from th e und amaged one , the exa mp le of mismatch repair by excinuclease shows that the enzyme may not always be able to discrimin ate the dam aged and un dam aged strands. Thi s point has not been in vestigated in detail. However , with thymine cyclobutane dim er there was no excision of the undam aged strand at a rate of 5% of th e dam aged strand (the detection limit of the assay). Thus, clearly with dam age a s opposed to mismatch, the enzyme has a mechanism of discri mi nating the rig ht and wro ng strands .:' Thi s fact lead s to the third question that was raised with regard to substrate recognition: what is the molecular ba sis of dam age recognition? The sim ple answer at present is : we don't know. Th e following facts are of relevance in sea rchi ng for an answer for this qu estion . (i) Alt hough lesions th at cause gross helical deform ity are repaire d, those that do not are also repaired, and there is no lin ear re la tions hi p between the spe cificity coefficient (kca,lk m) of the excinuclease an d the degr ee of helical deformity (39,40). (ii) Recognition involves a protein complex that has a pr eference for damage d DNA (XPA-RPA) a nd a complex (TF II H) with ATP -depend ent local unwinding activity th at is rec ruited to the dam age site a nd, based on the precedent in E. coli, u nwinds DNA and mak es th e ultimate dam aged DNA-protein pre incisio n complex. (iii) Recently, it ha s been found that three re pa ir enzymes with rather narrow su bstrate specificity , nam ely DNA ph otolyase (pyri midine dim er s), uracil glycosylase (uracil in DNA), an d exonuclease III (AP site ), flip out th e lesion from the duplex into a "hole" withi n the enzyme to bri ng the active site cofactor or residu es in close contact with the target bond s (40 -42). Wheth er the excinuclease system flips out th e dam aged nu cleotid e(s) or the entire excised fragmen t rem ains to be seen (43).
Dual Incision / Excision-The molecul ar det ails of ma mma lia n excision re pair are now known in cons ide rable detail. Sixteen polypeptides are necessary an d sufficient for excin uclease activity (13) as shown in Fig. 2.
Th e XPA-RPA complex bin ds to the dam age site; the n XPA recrui ts TFIIH, which mak es a pr ein cision complex in an ATP hydrolysis-depend ent ma nner. XPC help s stabilize the pr ecin cision complex. Th e ATP-dependen t unw inding of the DNA by TFIIH primes it for nucl ease attac k by the two XP pr otein s kn own to have nucl ease activity . XPG is recruited by TFIIH an d incises on the 3' side (13,44), and ER CC1-XPF , which is rec ruite d by XPA, in cises on the 5' (44) side of the dam age. Th e du al incis ion is absolu te ly dependent on ATP hydrolysis (8).
Th e major site s of inc ision are re latively pr ecise an d are at the stoichiometry (13) an d bind to XPA through the N-terminal half of ERCC1 (15). Th e ER CC1-XPF complex is an endon uclease specific for single -stranded DNA. 2 XP G-This pr otein has a single-stranded spec ific endonuclease activity (34)(35)(36). It also acts as a double-stranded specific exonuclease (34). It bin ds loosely to TFIIH (13 ) and to RPA (18 ) a nd is apparently rec ruited by these compone nts to the excision nucl ea se complex . Step 1, ATP-independent da mage recognition. Step 2. ATP-de pendent formation of preincision com plex . It is quite like ly that only a subset of the pr otein s sho wn is present in th e actual pr eincisi on com plex . Some may act a s molecular matchmaker s and dis soci at e after formation of the preincision comp lex.
Step 3, dual incision. Th e two inci sio ns occu r in random order.
Step 4, rep lacement of th e exci sio n pr oteins by re pai r syn thesis prot eins .
Step 5 , rep air syn thes is and ligation.
does not affect th e activity of human excinuclease (57).
Th e conn ecti on betw een th e p53 tumor suppressor protein and DNA repair ha s been th e source of much speculation and debate becau se the p53 protein is sta bilized by DNA damage and it is a transcr iptional regul ator. It has been reported that p53 prot ein binds to XPB (58) an d RPA (59), both of which are essential for basa l excision repa ir . However, p53(-{-) cells excise the two major UV photo products, pyrimidine dim er s, and 6-4 photopro ducts at th e sa me rate as wild type cells a nd are equally resistant to UV , J . C. Hu an g and A. San ca r , unpublish ed observation. 5th phosph odiester bond 3' a nd th e 24t h phosphodi ester bond 5' to th e lesion (7,8). However , th e incision sites show some variability. Th e site of 3' incision exte nds from th e 3rd through th e 8th phosphodieste r bond (7,8), and the site of 5' incision exte nds from th e 20t h th rough th e 26th phosph odiester bond (8). The combina tion of th ese incision patterns usu ally resul ts in excision offra gme nts 24-32 nt in length ; however , 27-29-nt fragments are the domin an t species. Th e sites of incision are influenced by several factor s, including the type of lesion (38) and the seque nce context (37,39). Th e same incision pattern has been observe d in vivo in Xenopus eggs (8) and in cell-free extracts from Sc hizosaccharomyces pombe" a nd is considere d to be th e universal incision pa ttern for eucaryotes (4).
Repair Synthesis-In cont rast to the excision reacti on we kn ow less about th e det a ils of t he re pair synthesis ste p. It is kn own th at repair synthes is is PCNA·d ependen t a nd hen ce mu st be carried out by PolS a nd Pole (45,46), and, since PCNA is the polymerase clamp load ed onto te mplate-pri mer by RFC repl ica tion fact or , RFC may a lso be re quire d. In a study with cell-free ext ra ct Polo antibodies specifically inhibited repair synthesis (47), and in a highly purified in vitro sys te m for repair synthesis it was found th at Pole and eve n K1 en ow fragment of'Poll perform ed rep air synthes is pointing to th e difficulty of assign ing repair polym er ase from in vitro reconstitution sys te ms (48). Most likely, both Polo a nd Pole participate in th e repair synthes is ste p of excision repair (49).

Tra nscri p tion -Rep a ir Coup li ng
Transcribed seque nces a nd in particular the temp late strand within a transcribed seque nce are repa ir ed at a higher rate than non-tran scribed seque nces (50). Cells from CS patients are defective in strand-specific repa ir (5 1). In E. coli, a tran scription-repair coupling factor encoded by th e mfd gene displaces sta lled RNA polymer ase a nd releases the stalled compl ex whil e recruiting th e dam age recogn it ion complex of excinuclease (52). However , a t present the re is no in vitro system for trans cri ption-re pa ir coupling in mammali an cells. Th e CSB gene encodes a protein of 160 kDa , which contains th e so-called helicase motifs a nd is lik ely to function in a ma nn er a na logous to th e E. coli Mfd pr otein (50). Thus, a si mple model for stra nd-specific repair based on beh avior of CS-A an d CS-B mutants a nd of the proteins is as follows.
RNA polymerase II stalled at a lesion is recogn ized by the CSA-CSB complex, which causes the polym er ase to back off th e lesion with out disrupting the ternary complex. Th e CSA-CSB compl ex al so recru its XPA a nd TFIIH to th e lesion site and thus help s in th e assembly of th e excin uclease. Th e lesion is excised a nd th eexcision ga p is filled in . Th e backed off RNA polymerase elong ates th e tru nca te d transcript (53).

R egulation of Excis ion R epair
It a ppea rs that mammalian cells do not possess an SOS respon se like that in E. coli wher e DNA damage by bu lky agents increas es th e tran scription of excision repair gen es (54). Simila rly, da magein duced post-tra nslational modification of repair prot eins (55,56) (60). Similarly, p53 protein at nearly micromolar concentrations has no effect on excision repair in a defined system." Finally, the report that the p53-induced Gadd45 protein stimulates excision repair (61) has not been confirmed." Thus, existing data are consistent with the notion that p53 does not modulate excision repair either positively or negatively.
In contrast, the recent discovery that Cdk7 and cyclin H, which make up the Cdk-activating kinase, are constituents ofTFIIH (62,63) raises interesting possibilities regarding cell cycle regulation and DNA repair. Excision repair capability of the cell does not change during the cell cycle (64). However, replication and repair may be coordinated by differential effects of p21(CipIWAF) on replicative and repair DNA synthesis (65). Future research is likely to uncover interesting interconnections between DNA repair, replication, cell cycle, and apoptosis.