Damaged DNA-binding protein DDB stimulates the excision of cyclobutane pyrimidine dimers in vitro in concert with XPA and replication protein A.

Human cells contain a protein that binds to UV-irradiated DNA with high affinity. This protein, damaged DNA-binding protein (DDB), is a heterodimer of two polypeptides, p127 and p48. Recent in vivo studies suggested that DDB is involved in global genome repair of cyclobutane pyrimidine dimers (CPDs), but the mechanism remains unclear. Here, we show that in vitro DDB directly stimulates the excision of CPDs but not (6-4)photoproducts. The excision activity of cell-free extracts from Chinese hamster AA8 cell line that lacks DDB activity was increased 3-4-fold by recombinant DDB heterodimer but not p127 subunit alone. Moreover, the addition of XPA or XPA + replication protein A (RPA), which themselves enhanced excision, also enhanced the excision in the presence of DDB. DDB was found to elevate the binding of XPA to damaged DNA and to make a complex with damaged DNA and XPA or XPA + RPA as judged by both electrophoretic mobility shift assays and DNase I protection assays. These results suggest that DDB assists in the recognition of CPDs by core NER factors, possibly through the efficient recruitment of XPA or XPA.RPA, and thus stimulates the excision reaction of CPDs.

Damaged DNA-binding protein (DDB) is one of the candidates that might play such an accessory role in human cells. DDB is composed of two subunits, p127 and p48, and has a higher affinity for various types of DNA lesions compared with the damage recognition subunits (XPA, RPA, XPC⅐HR23B, and TFIIH) of the six core repair factors (17)(18)(19). DDB activity has been shown to be missing in a subset of xeroderma pigmentosum group E (XP-E) patients. XP is an autosomal recessive disease associated with abnormal sensitivity to UV lights and a high incidence of skin cancer, which is known to be caused by defects in NER. XP-E is one of eight genetic complementation groups (XP-A through XP-G and XP-V) of XP patients, and, along with XP-V, manifests the mildest clinical features and deficiency in NER. XP-E individuals lacking DDB activity were found to carry mutations in p48 gene (20 -23). Because of its high specificity for binding to damaged DNA and a defect in some XP-E cells, DDB has been postulated to be involved in the NER process, presumably in a damage recognition step. However, DDB is clearly not essential for NER in several in vitro reconstituted systems, and its precise role in the damage recognition process has not been demonstrated so far (24 -27). However, recent in vivo studies clearly showed that XP-E cells and rodent cells lacking DDB activity are selectively defective in global genome repair (GGR) of CPDs, and transfection of the p48 gene into the rodent cells complements the deficiencies (28,29), suggesting that DDB may be involved in GGR especially for CPDs.
For this study, we have overproduced DDB in a baculovirus/ insect cell system, purified it to near homogeneity, and tested its effect upon in vitro excision reactions with DNA substrates containing either a single CPD or a single (6 -4)photoproduct. Indeed the recombinant DDB stimulated the excision rate of CPDs by cell-free extracts (CFEs) prepared from AA8 Chinese hamster ovary cell lines that lack DDB activity, and the stimulatory effect was enhanced in the presence of XPA or XPA ϩ RPA. We also observed that DDB forms a complex with XPA and RPA at the damage sites based on electrophoretic mobility shift assays and DNase I protection assays. These results provide direct biochemical evidence that DDB can be involved in the recognition process of CPDs in NER.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-To overproduce FLAG-tagged DDB subunits in insect cells, 5Ј-terminal portions of cDNAs with FLAG epitope se-quences were amplified by polymerase chain reactions using the primers for p127 (5Ј-CGCGGACCATGGACTACAAGGACGACGATGACAA-GATGTCGTACAACTACGTGG-3Ј and 5Ј-ATTGAATTCCTTTTCTCG-G-3Ј) or for p48 (5Ј-CGCGGATCCATGGACTACAAGGACGACGATG-ACAAGATGGCTCCCAAGAAACGC-3Ј and 5Ј-CATGAATTCTCCCCC-TTTGG-3Ј) and digested with BamHI and EcoRI. These fragments of 0.6 kilobases for p127 and 0.4 kilobases for p48 were individually subcloned into the pFASTBac1 expression vector (Life Technologies, Inc.), and the sequences of the inserted portions were verified. The remaining 3Ј-terminal regions of p127 or p48 cDNA were excised from the original constructs (30) with BstEII and NotI or BstXI and NotI, respectively, and inserted into the pFASTBac1 constructs containing each 5Јterminal portions to give the complete cDNAs.
Protein Expression and Purification-Recombinant baculoviruses encoding the p127 or p48 subunit were prepared according to the manufacturer's protocol. Sf21 insect cells were co-infected with both viruses to jointly express p127 and p48 or with the p127 virus alone and incubated at 27°C for 42-48 h. After collecting the cells by centrifugation, extracts were prepared by method of Manley et al. (31) and dialyzed against storage buffer (25 mM Hepes-KOH (pH 7.9), 100 mM KCl, 12 mM MgCl 2 , 1 mM EDTA, 2 mM dithiothreitol, 12.5% glycerol). The extracts were first applied to SP-Sepharose (Amersham Pharmacia Biotech) equilibrated with buffer A (25 mM Hepes-KOH (pH 7.9), 12 mM MgCl 2 , 1 mM EDTA, 2 mM dithiothreitol, 10% glycerol) containing 100 mM KCl. After extensive washing with buffer A containing 100 mM KCl, the recombinant proteins were eluted with a linear gradient of KCl (0.1-1 M) in buffer A and identified by Western blotting using anti-FLAG antibody. The eluted proteins were directly applied to a column containing anti-FLAG M2 affinity gels (Sigma) and washed extensively with buffer A containing 0.15 M KCl. The proteins were eluted with 5 column volumes of buffer A containing 0.15 M KCl and 2 mg/ml FLAG peptides, dialyzed against the storage buffer, and stored at Ϫ80°C.
Recombinant RPA and (His) 6 -XPA were expressed in Escherichia coli and purified as described (7,32). XPC⅐HR23B heterodimer was expressed in Sf21 cells and purified as described previously (33). The recombinant virus encoding XPC⅐HR23B was kindly provided by Drs. J. T. Reardon and A. Sancar (University of North Carolina). The activities of these purified repair factors were verified by an in vitro complementation assay or reconstitution assay.
Substrates-Duplex DNA substrates (56 bp for binding assays and DNase I protection assays and 136 bp for excision assays) that contained a cis-syn-cyclobutane pyrimidine dimer or a (6 -4)photoproduct in the center were constructed from 4 and 6 blocks of oligomer, respectively. Details for sequences and preparations of these substrates have been described (34,35). All substrates were internally labeled with 32 P at the 4th phosphate 5Ј to the photoproduct.
Excision Repair Assay-Cell-free extracts (CFEs) were prepared by the method of Manley et al. (31). The standard reaction mixture con-tained 3-6 fmol of 136-bp substrate, the indicated recombinant proteins, and 50 g of CFE prepared from AA8 cells in 25 l of reaction buffer (32 mM Hepes-KOH (pH 7.9), 64 mM KCl, 6.44 mM MgCl 2 , 0.16 mM dithiothreitol, 0.16 mM EDTA, 2 mM ATP, and 4% glycerol). Normally, the substrates were first incubated with the indicated recombinant proteins at 30°C for 10 min, CFE was added, and incubation was continued for 45 min. DNAs were extracted with phenol/chloroform/ isoamyl alcohol and separated on 8% denaturing polyacrylamide gels. Excision products were visualized by autoradiography, or the signals were quantified by exposing gels to Bas2000 imaging screens, and the intensities were measured with a Fuji Bas2000 Bioimaging Analyzer.
Electrophoretic Mobility Shift Assay-Two fmol of 32 P-labeled 56-bp substrates were incubated with the indicated amounts of recombinant proteins or CFE at 30°C for 20 min in 25 l of reaction buffer. The protein⅐DNA complexes were separated by electrophoresis on 5% nondenaturing polyacrylamide gels at 25 mA for 2 h and analyzed by autoradiography. To identify the compositions of protein⅐DNA complexes, specific antibodies were added after the complex formation. Monoclonal antibodies against FLAG and hexahistidine tags were obtained from Sigma and Amersham Pharmacia Biotech, respectively. Polyclonal antibodies against the p34 subunit of RPA were purchased from Oncogene Science.
DNase I Protection Assay-The internally labeled 56-bp substrates (2 fmol) were incubated with the indicated amounts of proteins in 25 l of reaction buffer at 30°C for 20 min and then digested with 700 units of DNase I (Life Technologies, Inc.) in the presence of 3.8 mM CaCl 2 . The products were purified, separated on 10% denaturing polyacrylamide gels, and analyzed by autoradiography.

RESULTS
Overproduction and Purification of DDB Heterodimer and p127 Subunit-To investigate the effect of DDB on in vitro excision repair reaction, we have overproduced DDB heterodimer (p127 and p48) or p127 subunit alone in baculovirus/ insect cell system. An 8-amino acid FLAG epitope fused at the N termini of both polypeptides (p127 and p48) facilitated the purification of the recombinant proteins. After a two-step separation with SP-Sepharose and an anti-FLAG M2 affinity gel, the near homogeneous proteins (Fig. 1A) were tested for DDB activity by electrophoretic mobility shift assay (EMSA) (Fig.  1B). The heterodimeric DDB specifically bound to a 56-bp duplex containing a (6 -4)photoproduct (lanes 7 and 8), and the mobility of the complex was identical to that with the native DDB (lanes 3 and 4) from HeLa cell-free extracts. However, the recombinant p127 protein alone did not bind to DNA either were analyzed by 10% SDS-polyacrylamide gel electrophoresis followed by staining with GelCode Blue stain reagent (Pierce). B, 32 P-labeled 56-bp duplex DNA (2 fmol) either with or without a (6 -4)photoproduct was incubated at 30°C for 20 min with 10 g of CFE prepared from HeLa (lanes 3 and 4) or AA8 (lanes 5 and 6) cell lines or 70 ng of recombinant DDB heterodimer (lanes 7 and 8) and p127 subunit (lanes 9 and 10). It should be noted that 2 g of poly(dI-dC)⅐poly(dI-dC) was added to the reaction in this particular experiment to reduce nonspecific binding. The reaction mixtures were separated on a 5% nondenaturing polyacrylamide gel and analyzed by autoradiography. C, two fmol of 32 P-labeled 56-bp duplex DNA with or without a lesion (CPD or (6 -4)photoproduct) was incubated with the indicated amount of DDB at 30°C for 20 min. The protein-DNA complexes were visualized by autoradiography after electrophoresis on a 5% nondenaturing gel.
with or without a (6 -4)photoproduct (lanes 9 and 10), consistent with previous reports (36,37). We also compared the binding ability of the DDB to a 56-mer duplex DNA containing a single CPD or a (6 -4)photoproduct (Fig. 1C). As expected, DDB showed a clear preference for the DNA duplex with a (6 -4)photoproduct (lanes 10 -12) over the undamaged DNA probe (lanes 2-4), and it also showed a moderate preference for the DNA probe containing the CPD (lanes 6 -8).
Effects of DDB on in Vitro Nucleotide Excision Repair Reaction with CHO AA8 Cell-free Extracts-To test the effect of the recombinant DDB on an in vitro excision reaction with a CPDcontaining DNA substrate, we prepared CFEs from CHO AA8 cells, which have been shown to lack p48 expression and, consequently, DDB activity (29,36). Fig. 1B (lanes 5 and 6) confirmed that there is no binding activity in AA8 CFEs. The 136-bp substrate with a single CPD was preincubated with DDB heterodimer or p127 alone (as a control) and subsequently incubated with AA8 CFE (Fig. 2A). DDB (lane 2), but not p127 alone (lane 3), stimulated the excision reaction by AA8 CFEs ϳ3-fold, in agreement with in vivo data that when CHO V79 cells were transfected with the p48 gene DDB activity reappeared along with GGR of CPDs (29). This stimulation was also dependent on the concentration of DDB (Fig. 2B). On the other hand, when the DNA substrate containing a (6 -4)photoproduct was used, the same amount of DDB (280 ng) greatly inhibited the excision reaction ( Fig. 2A, lanes 4 and 5). Since DDB binds to a (6 -4)photoproduct more efficiently than a CPD, as shown in Fig. 1C, a similar experiment with less DDB was carried out (Fig. 2C). Under this condition, DDB up to 13.2 ng did not affect the excision level of (6 -4)photoproducts by AA8 CFEs. It should be noted that the excision amount of (6 -4)photoproducts is much higher than that of CPDs (3.0% versus 0.06%). Thus DDB might help the recognition of CPDs by the core NER proteins.
Enhancement of the DDB-stimulated Excision Reaction by XPA and RPA-To investigate the functional interaction between DDB and the predicted damage recognition subunits of the NER factors, DNA substrate with a CPD was preincubated with DDB in the presence of RPA, XPA, XPA ϩ RPA, or XPC⅐HR23B before the addition of AA8 CFE. As shown in  7-10), and the effects of DDB and XPA or XPA ϩ RPA appear to be additive. Interestingly, when the DNA substrates with a (6 -4)photoproduct were used (Fig. 3B), the inhibitory effect of DDB on the excision reaction was apparently reversed by XPA (lane 4) or XPA ϩ RPA (lane 5) but not by RPA (lane 3) or XPC⅐HR23B (lane 6). These results suggest that XPA and RPA might be directly involved in the DDB-stimulated excision reaction of CPDs.
Complex Formation of DDB with XPA and RPA on a Damaged Site-To determine whether DDB forms a complex with XPA and/or RPA on a damaged DNA, we first employed EMSA with various pairwise combinations of DDB, XPA, and RPA (Fig. 4). Since we had a special interest in the possible ability of DDB to recruit XPA or RPA onto a damaged DNA, we chose conditions under which XPA or RPA alone barely bound to the DNA probes (Fig. 4, A and B, lane 2). A combination of DDB and XPA (Fig. 4A) 6) were incubated at 30°C for 10 min with 280 ng of DDB heterodimer or p127 subunit, and then 50 g of AA8 CFE was added to each mixture, and incubation was continued for 45 min. DNAs were then extracted, separated on a 8% sequencing gel, and detected by autoradiography. B, three fmol of internally labeled substrates containing a CPD were incubated at 30°C for 10 min with the indicated amounts of DDB, and then 50 g of AA8 CFE was added to each mixture. After the incubation for the indicated periods, DNAs were extracted and separated on a 8% sequencing gel. Excision products were quantitated with an image analyzer. Bars indicate S.D. from three independent experiments. C, three fmol of internally labeled substrates containing a (6 -4)photoproduct and the indicated amounts of DDB were incubated at 30°C for 10 min, AA8 CFE was added to each mixture, and incubation was continued for 45 min. The DNA was then extracted and analyzed by autoradiography after electrophoresis on a 8% sequencing gel.
containing DNA and makes a ternary complex of DDB⅐XPA⅐ DNA. On the other hand, another combination, DDB and RPA (Fig. 4B), gave a more intense band compared with RPA alone (lane 2 versus lane 4). This band appears to contain RPA⅐DNA complex and DDB⅐RPA⅐DNA complex since the band was partially supershifted by ␣-FLAG antibody (lane 6). These data indicate the potential complex formation of DDB⅐RPA⅐DNA, which shows nearly the same mobility as the RPA⅐DNA complex, consistent with a previous report (38).
Finally, we mixed all three factors with the DNA probe to test whether a higher order complex forms. As shown in Fig.  4C, only one retarded band was observed (lane 5), which was supershifted by all of the specific antibodies to these factors (lanes 6 -8), indicating that this retarded band contains DDB⅐XPA⅐RPA⅐DNA. We believe that the band in lane 5 is different from the band in lane 4 for the following reason. When we added ␣-His antibody to the reaction containing RPA and DDB, as in lane 4, we could not observe any supershifted band (data not shown). However, ␣-His antibody supershifted more than 90% of the complex as shown in lane 6.
To determine whether the three factors make a complex at a damaged site or bind independently, we conducted a DNase I protection assay (39) using DNA probes containing a (6 -4)photoproduct and internal 32 P label at the 4th phosphate bond 5Ј to the lesion. This fragment was chosen for these studies due to the inability of CPD-containing substrates to show clear signals in the DNase I protection assay (data not shown). After incubation with DDB and/or XPA ϩ RPA, the probes were extensively digested with DNase I to determine the size of the DNA region that was protected from DNase I digestion (Fig. 5). DDB alone conferred a specific and distinctive band of 9 nucleotides (lane 4) in addition to nonspecific bands of 4 -6 nucleotides, indicating the presence of a DDB⅐DNA complex around a damaged site. On the other hand,

DDB Stimulates the Excision of CPDs in
Vitro-This paper provides the direct evidence that DDB stimulates the excision reaction of CPDs in vitro, consistent with the recent report that DDB activity correlates well with the GGR activity of CPDs in intact cells (28,29). Previous attempts to observe the effect of DDB on in vitro NER reactions have either shown little effect or inhibition (24 -27). One possible reason would be that the substrates used in those studies contained non-CPD lesions: UV-irradiated plasmid DNA containing (6 -4)photoproducts as well as CPDs (24), linear duplexes containing a (6 -4)photoproduct or a cholesterol moiety (25), covalent closed circular DNA containing a single cisplatin lesion (26), or a mononucleosome containing a (6 -4)photoproduct (27). In fact, we also failed to observe a stimulatory effect of DDB using (6 -4)photoproduct-containing DNA substrates. Preferential binding of excision repair factors RPA, XPA, or XPC⅐HR23B to non-CPD lesions has been reported (33, 40 -46). The recognition of these non-CPD lesions might be properly achieved by these repair factors, at least in vitro, regardless of the presence or absence of DDB. On the contrary, cis-syn-CPD is a poor substrate for in vitro NER reactions, as shown in Fig. 2, and there have been no reports of preferential binding of the core NER factors to CPDs. The preferential binding of CPD-containing DNA substrates (Fig. 1C) may assist core repair factors to recognize CPDs.
In Vitro Effect of DDB on the Excision of a (6 -4)Photoproduct-Could DDB not be involved in the repair of (6 -4)photoproducts? In vivo repair studies (28,47,48) do not support this idea. The repair of (6 -4)photoproducts in XP-E cells lacking DDB activity is significantly slower compared with normal cells, indicating the partial involvement of DDB in the repair of (6 -4)photoproducts in vivo. In intact cells, DDB may be required for the recognition of (6 -4)photoproducts at some parts of the genome. Alternatively, a possible role of DDB in the repair of DNA in chromatin may be implicated, as suggested by others (26). Why DDB inhibits the excision of (6 -4)photoproducts at higher concentrations (Fig. 2) is not known at present. A similar inhibitory effect of DDB also has been reported in a reconstituted system under XPA-or RPA-limiting conditions (25). DDB has an ϳ10-fold higher affinity for (6 -4)photoproducts compared with CPDs (18), and at high concentrations, the second band, with slower migration, increases relative to the first band in EMSA (Fig. 1C). This form of the complex might inhibit the binding of core NER factors and the complex formation of human excision nuclease.
Functional Interaction between DDB, XPA, and RPA-Another noteworthy finding in this study is that the stimulatory effect of DDB on the excision of CPDs was further enhanced in the presence of XPA or XPA ϩ RPA (Fig. 3A). Interestingly, the inhibitory effect of DDB on the excision of (6 -4)photoproducts was also suppressed by the presence of these factors (Fig. 3B). These observations have a good correlation with the complex formation detected by EMSA (Fig. 4). XPA, which readily made a complex with DDB on a damaged DNA, could enhance the stimulatory effect of DDB, whereas RPA, which partially made a complex with DDB, could not. Analogously, the combination of XPA and RPA made a complex with DDB as well as enhanced the stimulation of DDB. Furthermore, the binding of XPA to damaged DNA was profoundly enhanced in the presence of DDB (Fig. 4A). These results strongly suggest a functional link between DDB and XPA in the efficient repair of CPDs in vitro. Taken together, we propose the following model. DDB binds to a CPD in DNA, providing the kink or bending of DNA, as shown by others (18,19), and facilitates the recruitment of XPA or XPA together with RPA to the CPD site possibly through protein-protein interaction with XPA. The complex formation at a damaged site by these three factors may lead to the subsequent excision reaction.
In this study, we failed to obtain an indication of a functional interaction between DDB and XPC⅐HR23B based on the in vitro excision assays (Fig. 3) and EMSA (data not shown). This observation is consistent with the recent report (49) that DDB and XPC⅐HR23B bind to UV-damaged DNA independently as judged by EMSA. These results were rather unexpected because XPC⅐HR23B has been already known to be involved in GGR of CPDs (50,51) and was also proposed to be a primary damage recognition factor (45). Other unidentified factors suggested by others (28) or XPA⅐RPA might be needed to link DDB and XPC⅐HR23B. The further analysis is needed to verify this model and clarify the higher order complex formation.