Psoralen interstrand cross-link repair is specifically altered by an a
http://www.100md.com
《核酸研究医学期刊》
Laboratoire de Biophysique, INSERM U565, CNRS UMR5153, Muséum National d’Histoire Naturelle, 43 rue Cuvier, 75231 Paris Cedex 05, France and 1 UMR CNRS/Institut Curie 218, Institut Curie, Section de Recherche, 26 rue d’Ulm, 75248 Paris, France
*To whom correspondence should be addressed. Tel: +33 1 40 79 36 84; Fax: +33 1 40 79 37 05; Email: praseuth@mnhn.fr
ABSTRACT
Targeting DNA-damaging agents to specific DNA sites by using sequence-specific DNA ligands has been successful in directing genomic modifications. The understanding of repair processing of such targeted damage and the influence of the adjacent complex is largely unknown. In this way, directed interstrand cross-links (ICLs) have already been generated by psoralen targeting. The mechanisms responsible for ICL removal are far from being understood in mammalian cells, with the proposed involvement of both mutagenic and recombinogenic pathways. Here, a unique ICL was introduced at a selected site by photoactivation of a psoralen moiety with the use of psoralen conjugates of triplex-forming oligonucleotides. The processing of psoralen ICL was evaluated in vitro and in cells for two types of cross-linked substrates, either containing a psoralen ICL alone or with an adjacent triple-stranded structure. We show that the presence of a neighbouring triplex structure interferes with different stages of psoralen ICL processing: (i) the ICL-induced DNA repair synthesis in HeLa cell extracts is inhibited by the triplex structure, as measured by the efficiency of ‘true’ and futile repair synthesis, stopping at the ICL site; (ii) in HeLa cells, the ICL removal via a nucleotide excision repair (NER) pathway is delayed in the presence of a neighbouring triplex; and (iii) the binding to ICL of recombinant xeroderma pigmentosum A protein, which is involved in pre-incision recruitment of NER factors is impaired by the presence of the third DNA strand. These data characterize triplex-induced modulation of ICL repair pathways at specific steps, which might have implications for the controlled induction of targeted genomic modifications and for the associated cellular responses.
INTRODUCTION
Sequence-specific DNA ligands such as triplex-forming molecules are promising reagents that can interfere with DNA metabolism. TFOs have been successfully used to modulate gene transcription (initiation and elongation), and also replication and repair processes, in some cases leading to the induction of targeted genomic modifications . TFOs can serve as vectors for common DNA-damaging molecules, such as cross-linking or cleaving agents. Such bifunctional molecules provide useful tools to induce controlled DNA modifications, and the molecular events surrounding damage repair in the absence or presence of the oligonucleotide moiety must be characterized to elucidate the corresponding cellular responses.
Interstrand cross-links (ICLs) between nucleotides in complementary DNA strands are common lesions introduced into DNA by drugs such as psoralen, nitrogen mustards or cisplatin. The sequence of events involved in the removal of such cross-links is not yet understood in mammalian cells. Studies using cell lines deficient in specific repair proteins suggest that several repair pathways are probably involved in ICL repair. From these genetic approaches, both recombination-dependent and -independent pathways have been implicated (4,5), and an essential role for the ERCC1–XPF heterodimer has been described, this complex being involved in nucleotide excision repair (NER) and in other repair processes (6,7). In vitro studies have revealed that both mammalian cell extracts and purified components of NER can promote incisions on the 5' (8–10) and 3' (8,11) sides of the psoralen ICL. A double incision 5' to the psoralen ICL leads to a 5' gap (9), and this gap can be converted to a nick adjacent to the ICL following DNA synthesis (12). The implication of such futile synthesis in the repair process remains unknown. ICLs have been shown to induce DNA synthesis, not only in the damaged plasmid, but also in another plasmid present in the reaction. This hallmark of ICL repair is not well understood but requires XPF–ERCC1 and the recombination factors, XRCC2 and XRCC3 (6). The early stages of ICL repair in mammalian cells are also of intense interest with the recent identification of new components, such as the mismatch complex hMutS? (7).
In this study, we characterize the repair of a unique psoralen ICL, in the absence or presence of an adjacent triple-stranded structure, with the use of TFO–psoralen conjugates. We and others have previously used such conjugates and showed that: (i) in cells, they can induce targeted mutagenesis and stimulate recombination, both processes being influenced by the presence of the triplex structure, and both involving NER factors such as xeroderma pigmentosum A protein (XPA) (13–16); and (ii) in cell extracts, the incision near the cross-linked psoralen was inhibited by the presence of the third strand (10,17). In this report, we further elucidate the molecular basis for the different processing of these two types of cross-linked substrates, either containing a psoralen ICL alone or with an adjacent triple-stranded structure. We present evidence that: (i) in HeLa nuclear extracts, the presence of an adjacent triplex structure decreased the level of repair synthesis and mainly suppressed the production of one of the two products of futile repair synthesis stopping at the ICL; (ii) in HeLa cells, ICL removal by an NER-dependent pathway was delayed when a triple-stranded structure was present; and (iii) binding of human recombinant XPA to psoralen ICLs was inhibited by a neighbouring triplex. These data demonstrate triplex interference at specific stages of psoralen ICL repair. We will discuss the implications of our observations for the future design of strategies aimed at targeting controlled genomic modifications.
MATERIALS AND METHODS
Oligonucleotides
Psoralen TFO conjugates, Pso-S-S-16TC, Pso-16TC and Pso-15TCG, were described previously (10,13); for sequences, see Figure 1B.
Figure 1. Experimental system. (A) DNA substrates used in this study: two plasmids and two synthetic double helices (76 and 29 bp long), abbreviated as 76D and 29D, respectively (for sequences, see Materials and Methods). They all contain the oligopyrimidine·oligopurine sequence suitable for triplex formation (called HIV-PPT) (grey box). A psoralen ICL can be formed at the 5' TA 3' site coinciding with the DraI restriction site (5' TTT|AAA 3', underlined). (B) The third strand oligonucleotides are represented. C stands for 5-methylated cytosines. The psoralen molecule was attached to the phosphodiester 16TC TFO via a hexamethylene arm, either directly (Pso-16TC) or via an additional disulfide bond (Pso-S-S-16TC). The Pso-15TCG oligonucleotide was used with either a phosphodiester (PO) or phosphoramidate (NP) backbone. (C) The two types of substrates containing a psoralen ICL at the 5' TA 3' site used in this study are schematically represented. Left: DNA substrates with a unique psoralen ICL, obtained after Pso-S-S-16TC (cleavable) treatment (Pso-P, Pso-Pluc and Pso-76D, Pso-29D). Right: DNA substrates with a unique psoralen ICL and an adjacent triplex structure obtained after Pso-16TC or Pso-15TCG (uncleavable) treatment (16 or 15-Pso-P, 16 or 15-Pso-Pluc, and 16 or 15-Pso-76D, 16 or 15-Pso-29D) (for details see Materials and Methods). The triplex site is shown (grey box). The total lengths of the various DNA substrates are indicated.
Preparation of psoralen cross-linked DNA substrates
Two types of DNA substrates containing the oligopyrimidine·oligopurine sequence suitable for triplex formation (the polypurine tract sequence of HIV-1, abbreviated as HIV-PPT: 5' AAAAGAAAAGGGGGGA 3') (18) were used: either two plasmids, pSP-F47 (3360 bp) abbreviated as P (see details in Fig. 2A), and PGK/luc plasmid (8850 bp) abbreviated as Pluc (see details in Fig. 4), or two short duplex oligonucleotides (76D and 29D). The synthetic duplexes are formed with complementary oligonucleotides, and the sequence of the purine-containing strand is as follows: 5'-CAAGGCAGCT GTAGATCTTAG (CCACTTTTTAAAAGAAAAGGGGG GACTGG)AAGGGCTAATTCACTCCCAACA-3'; the 76D duplex includes the 29D duplex as indicated (parentheses). All these DNA substrates were modified at a 5' TA 3' motif adjacent to the triplex site in two different ways using psoralen conjugates: either Pso-S-S-16TC for generation of a substrate with a unique psoralen ICL, or Pso-16TC or Pso-15TCG (uncleavable) for generation of a substrate with a unique cross-linked triple-stranded structure (Fig. 1). In all cases, the DNA targets were incubated with the appropriate TFO to form the triplex (in 10 mM MES pH 6 or 10 mM Tris pH 7, 50 mM NaCl, 10 mM MgCl2), and UV irradiated according to Giovannangeli et al. (19) (monochromatic light, 365 nm). After the covalent triplex was formed, the disulfide bridge in Pso-S-S-16TC was reduced with 1 mM dithiothreitol (DTT; 37°C, 1 h incubation) in order to eliminate the third strand portion. The cleaved and unreacted oligonucleotides were removed by gel filtration chromatography (Chromaspin, Clontech) for plasmid substrates. Two constructs were obtained: plasmids or synthetic duplexes containing a single psoralen ICL at the PPT triplex site are designed by Pso-(P or Pluc) and Pso-(29 or 76)D, respectively. 15/16-Pso-P and 15/16-Pso-D stand for covalent triple-stranded structures obtained with uncleavable Pso-15TCG/Pso-16TC, respectively. Under the experimental conditions used, >90% of the double-stranded DNA target could be modified with a unique psoralen ICL (for an illustration, see Fig. 2B, lanes 1 and 2). The unmodified DNAs (P or Pluc, 76D or 29D) were treated in the same way as the corresponding modified substrate (irradiation, DTT reduction and chromatography).
Figure 2. Analysis of products obtained by repair processing of psoralen cross-linked substrates in HeLa cell extracts. Synthesis repair assays were performed (as described in Materials and Methods) and, after purification, DNA fragments were digested with the indicated restriction enzymes and revealed by denaturing PAGE. (A) Schematic representation of the pSP-F47 plasmid (P) in the region surrounding the defined psoralen ICL site. Cleavage sites of the restriction enzymes on both DNA strands are indicated by vertical bars. The numbers between the bars indicate the distances in nucleotides between the cleavage sites. The thymines that are cross-linked in the psoralen-modified substrates, Pso-P and 15-Pso-P, are indicated. The orientation of the psoralen molecule is described with the furane (F) and the pyrone (P) sides. A schematic map of the pSP-F47 plasmid is represented with the triplex site indicated by a box and the ICL site by a cross (X). The 80 or 43 fragments are unmodified fragments used as references using BglII–EcoRV or BglII–BsrI digestion, respectively. Fragments sizes are indicated. (B) Analysis of species after BglII–EcoRV digestion. Lanes 1–3: for the three plasmids, unmodified (P) or containing a unique psoralen ICL (Pso-P and 15-Pso-P), digestion fragments were 5' labelled (see Materials and Methods). Lanes 4–6 and 8–10: neo-synthesized DNA fragments obtained upon incubation with HeLa cell extracts. 69pu and 65py fragments correspond to full-length repair products, and contain the oligopyrimidine·oligopurine PPT sequence. The 52 and 17/16 nt fragments correspond to futile repair products stopping at the ICL site (pyrimidine and purine strand, respectively). The 80 nt fragments did not contain the ICL site and PPT target sequence (see A) and were used as references for quantification of relative repair activity (see Fig. 3A and Materials and Methods). Cross-linked species are indicated on the left of the gel. The asterisk indicates the fragments of neo-synthesis, religated to the cross-linked thymine (see text for details). Lane 7 is a size marker obtained by 5' labelling of digestion products; DraI digestions of neo-synthesized 65py and 69pu fragments were indistinguishable from futile synthesis products (52 and 17 nt fragments) (data not shown). (C) Analysis of neo-synthesized species after BglII and BsrI digestion for various plasmid templates, as indicated. Full-length (38pu and 32py) and futile repair (17 and 19) fragments are indicated. The 17 nt fragment is hardly detected (see stoichiometries in Fig. 3B). The 43 nt fragments were used as reference fragments since they do not contain the target sequence (see A). Lane 1 is a size marker obtained by 5' labelling of BglII/BsrI/DraI digestion products. A schematic representation of the substrates is given below (B) and (C): the triplex site is boxed, the adducted thymines are indicated and the various fragments are depicted with their respective lengths.
Figure 4. Kinetics of cross-link repair in cells. The PGK/luc plasmid (Pluc) expressing the firefly luciferase was modified in order to obtain a single cross-linked triplex (15-Pso-Pluc) or a unique psoralen ICL (Pso-Pluc) in the 5'-untranslated region (see Materials and Methods). Luciferase activities were measured at different times after transfection (24, 48 and 72 h) in HeLa cells. In all experiments, plasmids were co-transfected with a control vector expressing Renilla luciferase. Reporter activity is presented as the normalized ratio between firefly and Renilla luciferase activities in the same lysate (relative luciferase activity). Normalization from cells transfected with an unmodified plasmid P permits quantification of the cross-link repair efficiency. The error bars correspond to the mean deviation from duplicate experiments.
In vitro repair synthesis assay
The nucleotide incorporation assay was performed in a 50 μl reaction mixture, according to Wood et al. (20). Briefly, the plasmid (0.3 μg), either unmodified (P) or modified (Pso-P, 15-Pso-P), was incubated in HeLa cell nuclear extracts (80 μg) (Promega) in the presence of 2 mM ATP, 20 μM dATP, dTTP, dGTP, 8 μM dCTP and 2 μCi of dCTP (3000 Ci/mmol) in the presence of 40 mM creatine phosphate, 2.5 μM creatine phosphokinase. The buffer was 45 mM HEPES pH 7.8, 70 mM KCl, 7 mM MgCl2, 0.9 mM DTT, 0.4 mM EDTA, 3.5% glycerol, 0.36 mg/ml bovine serum albumin. Incubation was performed for 3 h at 30°C. After purification, the plasmid was cleaved with BglII and EcoRV (or BsrI), producing the fragments represented in Figure 2A. The 65py·69pu (or 32py·38pu) fragments contain the ICL site and the oligopyrimidine·oligopurine target sequence for triplex formation. In parallel, in order to characterize the nature of the fragments obtained after restriction cleavage, the plasmid (P, Pso-P or 15-Pso-P) was digested by the appropriate set of enzymes in the absence of incubation in extracts, and the fragments were 5'-labelled with T4 polynucleotide kinase (Biolabs) and ATP by standard procedures. In all cases, the products were analysed and quantitated by 8% acrylamide/bisacrylamide 19:1 sequencing PAGE.
Quantification of repair-induced neo-synthesized fragments
Full-length neo-synthesized products. Full-length synthesized fragments were quantitated taking into account the following factors. (i) The number of cytosines in various products, since we performed a dCTP incorporation assay as described above. It must be noted that the control fragments (80 for BglII–EcoRV digestion, and 43 for BglII–BsrI digestion) are composed of two strands that are not separated by electrophoresis in our conditions. The numbers of cytosines are as follows: 17 in 69pu, 16 in 65py, 42 in (80+80), five in 38pu, eight in 32py and 19 in (43+43). (ii) The detectable incorporation present in undamaged fragments (control fragments). (iii) The initial abundance of the fragments. It is thus possible to determine the amount of synthesized product specifically due to repair processing, i.e. the relative repair activity (Fig. 3A). It was calculated as explained in the following example, for the synthesis of the purine strand (69pu) with analysis with BglII–EcoRV digestion (see Fig. 2): the 80 nt fragments (abbreviated as 80+80), undamaged and adjacent to the target fragment, were used as references. Then the relative repair activity of 69pu was obtained by dividing (a) by (b) with: (a) the relative abundance of the neo-synthesized products 69pu, estimated by the ratio of radioactivity of the fragments 69pu to (80+80) obtained in the dCTP incorporation assay, corrected by the number of cytosines; and (b) the initial abundance of the products 69pu and (80+80), before reaction in the extracts, estimated by the ratio of radioactivity of the fragments (69pu + 65py) to (80+80) obtained in kinase labelling experiments. The relative repair activity obtained for the undamaged plasmid P should be close to 1 if the efficiencies of kinase labelling are identical for every fragment. In our system, higher efficiency was obtained for overhang versus blunt end labelling, and we chose to normalize to 1 the activity obtained for the undamaged plasmid P.
Figure 3. Quantification of the abundance of the neo-synthesized fragments due to repair processing of the different cross-linked DNA substrates. (A) Relative repair activities (R) of full-length repair products: purine (light grey) and pyrimidine (dark grey) strands were reported for each template (as indicated). They were obtained after dividing the relative abundance of neo-synthesized products (obtained by nucleotide incorporation) by their relative abundance in the initial mixture (obtained by kinase labelling) (average of three experiments; see details in Materials and Methods). (B) Schematic representation of the various neo-synthesis fragments obtained in vitro for the two different psoralen cross-linked substrates used in the present work. Purine and pyrimidine full-length or futile repair fragments are represented with their associated stoichiometries, calculated on the basis of fragment signal intensities and cytosine contents (mean from four experiments). For each of the full-length fragments, the background synthesis, estimated from (A) by comparison of the undamaged substrate (P) with the damaged ones (Pso-P and 15-Pso-P), was deduced in order to obtain the level of repair-induced synthesis.
It must be noted that we did not take into account recent results describing a possible stimulation of nucleotide incorporation when an undamaged plasmid was added to the solution in appropriate conditions, especially in excess compared with the damaged plasmid (6). In our experimental conditions, the amount of unmodified plasmid represented <10% and might not induce increased incorporation. If such incorporation occurs, our evaluation of relative repair activity would be under-estimated.
Futile neo-synthesized products. The abundance of these fragments was calculated on the basis of signal intensities and cytosine contents. The number of cytosines for the different fragments are: 15 in 52py, seven in 19py and four in 17pu. For information, the number of cytosines for fragments that are not observed is one for 13py, nine for 52pu and one for 21pu.
DNA repair assay in cells
The PGK/luc plasmid, abbreviated as Pluc, contained the firefly luciferase (Photinus pyralis) gene under the control of the phosphoglycerate kinase (PGK) promoter. A 34 bp insert (5' CCACTTTTTAAAAGAAAAGGGGGGACTGGAAGGG 3') containing the polypurine tract PPT/HIV-1 was cloned in the 5'-transcribed sequence of the luciferase gene (BamHI–HindIII fragment from the pGL2-basic vector; Promega), downstream of the putative PGK start site, as previously described in more detail (21,22). The Renilla luciferase (Renilla reniformis; RL) control vector (CMV-RL; Promega) was used to monitor transfection efficiency. The RL gene was inserted downstream of the cytomegalovirus (CMV) immediate-early enhancer/promoter region.
The Pluc plasmid was modified at 90–95% by a single psoralen ICL (Pso-Pluc) or by a single cross-linked triplex (15-Pso-Pluc), as described previously. The various Pluc plasmids (0.5 μg), modified or not, were co-transfected with the control vector expressing Renilla luciferase CMV-RL plasmid (0.005 μg), in HeLa cells (in 24-well plates) using Superfect transfection reagent (Qiagen), as described (21). The mixture was prepared for duplicates or triplicates. Both luciferase activities (firefly and Renilla) were measured in the same cell extract by using the dual-luciferase assay kit (Promega); the ratio (firefly/Renilla), called the relative luciferase activity, was used to monitor the cellular processing of the various modified plasmids with time.
Analysis of recombinant XPA binding to psoralen cross-linked substrates
GST–XPA (rXPA) was prepared according to Saijo et al. (23,24). Binding activity of rXPA was characterized by electrophoretic mobility shift assay (EMSA) (4% non-denaturing acrylamide gel in 1x TBE buffer), as described previously (25–27).
The duplex probes were purified on denaturating gels and 5'-32P-labelled on both strands. The radiolabelled target was incubated with rXPA (50 mM MES pH 6, 0.1 M NaCl, 10 mM MgCl2) according to Jones et al. (28). Incubation was performed for 1 h at 30°C and species were separated by EMSA. Mixtures with defined percentages of psoralen ICL at the triplex site were obtained by treating the sample (76D in the presence of Pso-S-S-16TC) for variable irradiation times. The amount of ICL was determined by denaturing electrophoresis and is expressed as the ratio Pso-76D:76D (Fig. 5A).
Figure 5. XPA binding to various types of psoralen ICL. (A) Gel retardation analysis of rXPA binding to the 76D duplex containing a unique psoralen ICL (Pso-6D) in various proportions, as indicated by the ratio Pso-76D:76D. The radiolabelled duplex concentration (Pso-76D + 76D) is 10 nM, and rXPA protein (1 μM) was added or not, as indicated. The amount of retarded complex L is indicated below the gel. (B) Triplex interferes with XPA binding to psoralen-modified duplexes. The damaged duplexes (16-Pso-76 or 29D and Pso-76 or 29D) were used as probes; complex formation was achieved by addition of rXPA (1 μM). Amounts of L complex (a.u.) obtained with various psoralen-damaged probes are reported; the values are normalized to that obtained for Pso-76/29D substrates. Values from one representative experiment that was independently repeated twice are presented. Equivalent results were obtained with the 76D and the 29D substrates.
Results
Substrates with a single psoralen interstrand cross-link
Plasmids or short synthetic duplexes containing a specific oligopyrimidine·oligopurine sequence (named HIV-PPT) were modified with two types of psoralen derivatives: the goal was to obtain at the same defined site in DNA a single psoralen ICL, either alone (by treatment with cleavable Pso-S-S-16TC) or associated with an adjacent triplex (by treatment with uncleavable Pso-16TC or Pso-15-TCG) (Fig. 1, and details in Materials and Methods). Following treatment with TFO–psoralen conjugates, the efficiency of ICL formation for each substrate was evaluated by electrophoresis analysis after 5' labelling. As an example, ICL induction in the plasmid pSP-F47 (P) was determined by 5' labelling of DNA fragments obtained by BglII and EcoRV digestion followed by electrophoresis analysis (Fig. 2B). The mobility of cross-linked products was reduced compared with that of undamaged fragments. The 69pu and 65py fragments containing the oligopyrimidine·oligopurine target were converted to cross-linked fragments with an efficiency of 90%, whereas the 80 nt fragment lacking the target was unaffected (Fig. 2B, lanes 1–3). Equivalent levels of modifications were obtained with synthetic oligonucleotide substrates.
Repair synthesis of the specific target fragment is decreased by the presence of an adjacent triple-stranded structure
The initial stages of ICL repair require incisions to be made in the vicinity of the cross-linked site; these incisions are associated with DNA repair synthesis. To analyse the repair synthesis in vitro, damaged plasmids were incubated in the presence of HeLa cell extracts, and DNA synthesis, induced by a single psoralen ICL that was introduced at a defined site in the Pso-P or 15-Pso-P plasmids, was measured as the specific incorporation of labelled dCTP into damage-containing fragments. The competency of the extracts to perform NER in our experimental conditions was evaluated by using a plasmid carrying a unique cisplatin ICL between two guanines in a GTG sequence. This cisplatin lesion is known to be a good substrate for NER (29) and therefore serves as a positive control. Repair activity was detected since a fragment containing the cisplatin cross-link (33 nt long) became specifically 3- to 4-fold more radioactive in the damaged plasmid compared with the control undamaged plasmid (data not shown).
Incorporation of labelled nucleotides reflects a neo-synthesis activity, and the different products have been analysed on denaturing PAGE. The damage-containing fragments, 69(38)pu or 65(32)py, were generated by double digestion with BglII and EcoRV (BsrI) as shown in Figure 2A. The radioactive signals corresponding to intact purine and pyrimidine fragments were significantly higher for cross-linked substrates than what would be expected from non-specific incorporation into the small residual undamaged fragments. Noticeably a background synthesis independent of repair (30,31) was also detected in the undamaged fragments present in the control sample (Fig. 2B, lane 6, and C, lane 4). This is attributed to nicking activities present in the extracts or arising as a consequence of the irradiation step. The neo-synthesis associated with repair processes (relative repair activity) was evaluated for each strand (see Materials and Methods) and is presented in Figure 3A. Levels of repair synthesis of pyrimidine and purine strands were increased (2.5- to 5-fold) in cross-linked compared with control plasmids, and were slightly decreased (30–40%) by the presence of the triplex structure.
Futile repair synthesis depends on the nature of the cross-linked substrate
In addition to the full-length fragments described above, shorter neo-synthesized products were also observed; their sizes corresponded to the length (±1 nt) between the proximal enzymatic cleavage site and the position of the psoralen ICL located at the DraI digestion site. In Figure 2B (lanes 4 and 8) and C (lane 2), short fragments of 52(19)py and 17/16pu were clearly detected (according to the digest). Triplex did not affect the production of the short purine (17/16 nt) fragment (Fig. 2B, lanes 8 and 9), whereas the pyrimidine short fragment was completely abolished (Fig. 2B, lanes 4 and 5, and C, lanes 2 and 3).
These truncated neo-synthesis fragments are consistent with the so-called futile DNA synthesis products described by Mu et al. (12). These authors reported that a psoralen ICL induced a DNA synthesis that converted an excision gap 5' to the ICL into a nick without removing the damage. Several findings support the idea that our observed truncated neo-synthesized products are the result of specific repair synthesis. First, we observed a synthesis signal 13-fold stronger for the pyrimidine strand compared with that for the purine strand (ICL in the Pso-P plasmid) (Figs 2 and 3B). Such a 10-fold ratio was observed previously for the futile synthesis induced by the pyrone-side adducted thymine compared with the furane-side adducted thymine (12), reflecting a differential incision efficiency as reported by Bessho et al. (9). Such stoichiometry is consistent with the preferential orientation of the psoralen moiety in our system, with the furane side adducted on the purine strand and the pyrone side on the pyrimidine strand . Secondly, we observed a lack of futile synthesis specifically for the pyrimidine strand in the presence of the triplex structure (15-Pso-P plasmid). These data are consistent with our previous demonstration that the incision step during repair of a psoralen cross-link was inhibited at the triplex site in the presence of the third strand in HeLa nuclear extracts (10). In addition, the fact that no product was detected for the pyrimidine strand using 15-Pso-P plasmid as a substrate reflected a synthesis caused by damage repair and not originating from a repair-independent incision downstream of the lesion. Indeed, such an incision should have been elongated and detected since DNA polymerization could initiate downstream of a covalent triplex (data not shown based on primer extension experiments with Klenow DNA polymerase) (33,34). Thirdly, production of these truncated fragments by nucleotide incorporation until the damage position required the presence of ATP (data not shown).
Finally, for the Pso-P plasmid, a product corresponding to religation of futile synthesis repair fragments to thymine still carrying the psoralen was detected (shown by an asterisk in Fig. 2B, lane 4), as reported previously in another system (12). This product was detected when Pso-P plasmid was used as a substrate, but not for the 15-Pso-P plasmid (Fig. 2B, lane 5). This lack of religated product in the presence of the triplex is consistent with the notion that (i) the ligated product resulted from the two futile synthesis products, and therefore could not be produced in the presence of the third strand; and also that (ii) for the futile-synthesized purine strand, which is produced in the presence of the third strand, the triplex could inhibit the ligation step.
A triple-stranded structure delayed cross-link repair in cells
Since we observed that a triple-stranded structure inhibited ICL repair by interfering with the repair machinery at specific stages (incision, repair synthesis and religation) in vitro, we decided to evaluate the processing of cross-linked substrates in cells. The PPT sequence was cloned between the promoter and the coding region of the luciferase reporter gene (PGK/luc plasmid, named Pluc). This plasmid was modified with either cleavable Pso-S-S-16TC (Pso-Pluc plasmid) or uncleavable Pso-15TCG (15-Pso-Pluc plasmid), and then transfected into HeLa cells. Luciferase reactivation was used to monitor ICL removal by a recombination-independent and error-prone repair pathway, involving NER factors, as described previously (5). Luciferase activity was measured at different times after transfection (Fig. 4). At 24 h post-transfection, luciferase activity was inhibited by 90–95% for Pso-Pluc and 15-Pso-Pluc, in agreement with the yield of psoralen ICL obtained for each construct. At later times, luciferase reactivation increased and, at all times, we observed that luciferase activity from Pso-Pluc was twice that for 15-Pso-Pluc. This result is consistent with a 2-fold lower repair rate of 15-Pso-Pluc compared with Pso-Pluc in this assay.
Recombinant XPA binds to a fragment containing a unique psoralen interstrand cross-link and complex formation is influenced by an adjacent triplex structure
In cells, NER factors have been shown to be sequentially assembled and XPA is involved in a pre-incision step for the loading of different components of the NER complex (35). In vitro, XPA has been described in some cases as a damage-sensing factor (25,36,37). In order to characterize further the effect of a triplex structure on NER processing of psoralen ICLs, we examined the binding affinity of recombinant human protein rXPA for duplexes containing a unique psoralen ICL, alone or in the presence of a triple helix. For this purpose, two target duplexes of different lengths (76D and 29D, see sequences in Materials and Methods) were modified with a unique psoralen ICL (in Pso-76D or Pso-29D) or with a unique covalent triplex (in 16/15-Pso-76D or 16/15-Pso-29D; Fig. 1). These substrates were incubated in the presence of recombinant rXPA and the resulting complexes were analysed by gel retardation assay (Fig. 5). The complex was detected as a retarded band (L) on top of the gel. The amount of L complex increased proportionally with the amount of psoralen-modified duplex (Fig. 5A), indicating that the L complex represents a specific association between rXPA and the damaged DNA. We have estimated the protein concentration that bound 50% of the substrate: for Pso-76D, Kd = 0.3 x 10–6 M, compared with 10–6 M for unmodified 76D duplex. These values agree with previously published values (Kd = 0.3 x 10–6 M versus 1.7 x 10–6 M for a 258 bp fragment in the presence or absence of four sites of UV damage, respectively) (28).
The proportion of complex was quantified for various damaged substrates (Fig. 5B). Complex formation was inhibited to varying extents depending on the nature of the third strand: 54% inhibition with the phosphodiester Pso-16TC (PO) (16-Pso-29D substrate), 70% inhibition with the phosphodiester Pso-15TCG(PO) and 85% with the phosphoramidate Pso-15TCG(NP) . The level of rXPA binding thus decreased as the thermodynamic stability of the triplex increased (38). However, the covalent triplex was still better recognized by rXPA than undamaged DNA. Similar results were obtained concerning rXPA binding affinities independently of the length of the various substrates (D, Pso-D or 16/15-Pso-D) in the tested range, 76 or 29 bp long, and also independently of the substrate sequence (another sequence suitable for triplex formation was tested 5' AAAAGGAGAGGGAGA 3'; data not shown). In contrast, in previous studies performed with prokaryote repair systems, triplexes were found to inhibit damage processing at the cleavage but not at the recognition step (39).
These data on rXPA binding must be considered with respect to helical distortions induced by a DNA triplex and the sensitivity of XPA to conformational alterations in the helical structure. The nature of the triplex (length of the third strand, Hoogsteen or reverse Hoogsteen motif, position of the ICL site at the 5' or 3' side of the triplex) could strongly influence induced distortion and, consequently, the affinity of XPA for the lesion (26). Further studies are needed to evaluate the precise influence of this set of factors.
Finally, the lower rXPA binding to psoralen ICL observed in the presence of an adjacent triplex structure is consistent with the decrease in luciferase reactivation described above, reflecting a decrease in cellular removal of psoralen ICL with a triplex, since this reactivation has been shown to be XPA dependent (5).
Discussion
This report shows that a psoralen ICL was differentially recognized whether alone or in the context of a DNA triple-helical structure by elements of the repair machinery, both in vitro and in cells. To summarize our findings, the presence of the third strand influenced different stages of psoralen ICL processing: (i) repair synthesis in HeLa cell extracts is inhibited by the triplex structure, as schematically described in Figure 3B; (ii) the ICL removal involving NER factors is delayed in HeLa cells in the presence of a triplex; and (iii) the triplex structure impaired recombinant XPA binding of a neighbouring psoralen ICL and the level of this inhibition is correlated with the binding affinity of the third strand to DNA. Triplex-forming molecules have been successfully used to position DNA-damaging agents such as psoralen molecules, with the goal of inducing targeted genomic modifications. In mammalian cells, there are indirect data demonstrating the contribution of NER factors, especially XPA, in mutagenesis and recombination processes induced by a cross-linked triplex (14,16). The molecular basis of processes leading to targeted DNA modifications is far from understood. In this context, we will discuss how our data may be useful in advancing this characterization and in designing future strategies aimed at targeting controlled genomic modifications.
The exact mechanism of psoralen ICL repair is still poorly understood in mammalian systems. Initially, repair processes are based on generation of various incisions in the vicinity of the lesion. At present, the factors responsible for these incisions as well as their exact sites remain to be determined. We and others did not directly detect species corresponding to 3' cleavage (9,10). However, such 3' incisions were directly detected in different experimental systems (7,8,11), and have been attributed to NER proteins with a major role in the ERCC1–XPF complex and recently also to MMRs (MutS?). Combining the results presented in this work and those of several other recent studies leads us to a putative model for the repair of the two types of studied cross-linked substrates (summarized in Fig. 6). We propose various scenarios starting from the futile synthesis fragments stopping at the psoralen ICL, as intermediates in the ICL repair process, including modulation by an adjacent triplex structure .
Figure 6. Model for repair of psoralen ICL and modulation by an adjacent triplex structure, based on the presented findings. Three possible repair pathways are described with futile synthesis repair fragments (grey arrows) as a starting point. (A) Following filling in, a small fraction of molecules are ligated to regenerate a cross-linked substrate. (B) 3' Cleavage (indicated by a vertical arrow) occurring on the same strand as the neo-synthesized fragment produces a gap which can be replenished by translesion followed by ligation and, probably, an additional cleavage process to completely remove the damage. This pathway is mutagenic. (C) 3' Cleavage (indicated by a vertical arrow) occurring on the strand opposite to the neo-synthesized product produces a DSB that can be repaired by either error-free HR or error-prone EJ, SSA and BIR processes (see text for detailed discussion). The different stages where the triplex was shown in the present study to interfere with ICL repair are indicated by (–). It is noticeable that for the cross-linked triplex, the pathway in (A) seems to be ineffective, and the pathways in (B) and (C) can only occur for the truncated neo-synthesized purine strand since a complete blockage of futile synthesis of the pyrimidine strand is observed (see Fig. 3B). The triplex site is boxed (in grey) and the adducted thymines are indicated.
A small fraction of these truncated fragments can be religated to the adducted thymines (Fig. 6A). We showed here that religation leading to regeneration of a cross-linked substrate was inhibited in the presence of the triplex structure (Fig. 2B). Consequently, the pathway shown in Figure 6A does seem to be ineffective for psoralen ICL in the presence of a triplex.
Futile synthesis products could also be processed in order to eliminate the ICL. Mainly, two pathways could be envisaged (Fig. 6B and C).
In the pathway shown in Figure 6B, a nick occurs 3' to the lesion, on the same strand as the incision 5' to the lesion which led to futile synthesis. DNA synthesis can then fill in the resulting gap by translesion bypass. Such translesion synthesis has already been characterized; it involves a specialized distributive polymerase (probably a homologue of Saccharomyces cerevisiae Rad30), more tolerant to structural defaults (40). It must be noted that the finding of the possible involvement of 3'–5' exonuclease activity of ERCC1–XPF in the presence of RPA (12) would even make the 5' cleavage unnecessary, and lead to the same end result. In any case, the remaining damage could then be removed by an additional cleavage process. This mechanism would predict that the base opposing the cross-linked thymine would be a mutation hot-spot as a result of the bypass synthesis. We have directly measured triplex interference during this pathway. Indeed, we have quantified the cellular removal of a site-specific psoralen ICL, either alone or with an adjacent triplex structure, using reactivation of a reporter gene when the damage was located in the 5'-transcribed but untranslated region. In HeLa cells, we have observed a decrease in ICL removal in the presence of the triplex structure. This result supports our in vitro data discussed above describing the triplex-induced inhibition of the repair machinery, at different levels (incision, repair synthesis). Such inhibition of this pathway might be associated with an increased contribution of the recombination pathway (Fig. 6C) for triplex-associated ICL repair.
In the pathway shown in Figure 6C, a nick is produced 3' to the lesion, on the strand opposite to the incision 5' to the lesion that led to futile synthesis. A double-strand break (DSB) can appear as an intermediate during ICL repair, as already demonstrated, both in cells and in vitro (4,7,41). In our experimental conditions we did not reveal any DSB. Another way to generate such a DSB could be an arrest of the replication fork near the ICL site in dividing cells. In any case, recombination-dependent repair will take place by different mechanisms; they could be error-free or error-prone processes.
Our data demonstrate triplex-induced inhibition of the repair machinery in vitro and in cells, and suggest an increased contribution of the pathway shown in Figure 6C compared with that shown in Figure 6B. These findings are consistent with mutagenesis and recombination results obtained with a positioned psoralen ICL, alone or with an adjacent triplex: (i) the mutation frequency (Fig. 6B) was lower in the presence of the triplex structure (17), whereas the recombination frequency (Fig. 6C) was higher (14); and (ii) two types of mutation are induced around the site of the specific ICL for both types of ICL (in the presence or absence of the triplex structure); they are point mutations facing the adducted thymine (generated from the pathway shown in Fig. 6B) and deletions (generated from the pathway shown in Fig. 6C). The deletions were longer in the presence of the triplex (17).
The present work advances the molecular understanding of processes leading to targeted modifications of DNA in mammalian cells with the use of specific DNA ligands to direct DNA-damaging molecules. Such a study might help in the improvement of, and new developments in, genetic manipulation of mammalian cells and in the use of DNA-targeted anti-tumour agents.
ACKNOWLEDGEMENTS
We thank U. Asseline for the synthesis of psoralen-cleavable oligonucleotides, J. Moggs for the gift of platinated plasmid and monoclonal anti-XPA antibodies, and J. Mello for careful reading of the manuscript. F.G. was supported by ARC (Association de la Recherche sur le Cancer) and AFM (Association Fran?aise sur la Myopathie) fellowships.
REFERENCES
Faria,M. and Giovannangeli,C. (2001) Triplex-forming molecules: from concepts to applications. J. Gene Med., 3, 299–310.
Knauert,M.P. and Glazer,P.M. (2001) Triplex forming oligonucleotides: sequence-specific tools for gene targeting. Hum. Mol. Genet., 10, 2243–2251.
Seidman,M.M. and Glazer,P.M. (2003) The potential for gene repair via triple helix formation. J. Clin. Invest., 112, 487–494.
De Silva,I.U., McHugh,P.J., Clingen,P.H. and Hartley,J.A. (2000) Defining the roles of nucleotide excision repair and recombination in the repair of DNA interstrand cross-links in mammalian cells. Mol. Cell. Biol., 20, 7980–7990.
Wang,G., Chen,Z., Zhang,S., Wilson,G.L. and Jing,K. (2001) Detection and determination of oligonucleotide triplex formation-mediated transcription-coupled DNA repair in HeLa nuclear extracts. Nucleic Acids Res., 29, 1801–1807.
Li,L., Peterson,C.A., Lu,X., Wei,P. and Legerski,R.J. (1999) Interstrand cross-links induce DNA synthesis in damaged and undamaged plasmids in mammalian cell extracts. Mol. Cell. Biol., 19, 5619–5630.
Zhang,N., Lu,X., Zhang,X., Peterson,C.A. and Legerski,R.J. (2002) hMutSbeta is required for the recognition and uncoupling of psoralen interstrand cross-links in vitro. Mol. Cell. Biol., 22, 2388–2397.
Kumaresan,K.R., Hang,B. and Lambert,M.W. (1995) Human endonucleolytic incision of DNA 3' and 5' to a site-directed psoralen monoadduct and interstrand cross-link. J. Biol. Chem., 270, 30709–30716.
Bessho,T., Mu,D. and Sancar,A. (1997) Initiation of DNA interstrand cross-link repair in humans: the nucleotide excision repair system makes dual incisions 5' to the cross-linked base and removes a 22- to 28-nucleotide-long damage-free strand. Mol. Cell. Biol., 17, 6822–6830.
Guieysse,A.L., Praseuth,D., Giovannangeli,C., Asseline,U. and Hélène,C. (2000) Psoralen adducts induced by triplex-forming oligonucleotides are refractory to repair in HeLa cells. J. Mol. Biol., 296, 373–383.
Kuraoka,I., Kobertz,W.R., Ariza,R.R., Biggerstaff,M., Essigmann,J.M. and Wood,R.D. (2000) Repair of an interstrand DNA cross-link initiated by ERCC1–XPF repair/recombination nuclease J. Biol. Chem., 275, 26632–26636.
Mu,D., Bessho,T., Nechev,L.V., Chen,D.J., Harris,T.M., Hearst,J.E. and Sancar,A. (2000) DNA interstrand cross-links induce futile repair synthesis in mammalian cell extracts. Mol. Cell. Biol., 20, 2446–2454.
Barre,F.X., Asseline,U. and Harel-Bellan,A. (1999) Asymmetric recognition of psoralen interstrand crosslinks by the nucleotide excision repair and the error-prone repair pathways. J. Mol. Biol., 286, 1379–1387.
Faruqi,A.F., Datta,H.J., Carroll,D., Seidman,M.M. and Glazer,P.M. (2000) Triple-helix formation induces recombination in mammalian cells via a nucleotide excision repair-dependent pathway. Mol. Cell. Biol., 20, 990–1000.
Raha,M., Wang,G., Seidman,M.M. and Glazer,P.M. (1996) Mutagenesis by third-strand-directed psoralen adducts in repair-deficient human cells: high frequency and altered spectrum in a xeroderma pigmentosum variant. Proc. Natl Acad. Sci. USA, 93, 2941–2946.
Wang,G., Seidman,M.M. and Glazer,P.M. (1996) Mutagenesis in mammalian cells induced by triple helix formation and transcription-coupled repair. Science, 271, 802–805.
Wang,G. and Glazer,P.M. (1995) Altered repair of targeted psoralen photoadducts in the context of an oligonucleotide-mediated triple helix. J. Biol. Chem., 270, 22595–22601.
Giovannangeli,C., Perrouault,L., Escude,C., Nguyen,T. and Hélène,C. (1996) Specific inhibition of in vitro transcription elongation by triplex-forming oligonucleotide–intercalator conjugates targeted to HIV proviral DNA. Biochemistry, 35, 10539–10548.
Giovannangeli,C., Thuong,N.T. and Hélène,C. (1992) Oligodeoxynucleotide-directed photo-induced cross-linking of HIV proviral DNA via triple-helix formation. Nucleic Acids Res., 20, 4275–4281.
Wood,R.D., Robins,P. and Lindahl,T. (1988) Complementation of the xeroderma pigmentosum DNA repair defect in cell-free extracts. Cell, 53, 97–106.
Faria,M., Wood,C.D., Perrouault,L., Nelson,J.S., Winter,A., White,M.R., Hélène,C. and Giovannangeli,C. (2000) Targeted inhibition of transcription elongation in cells mediated by triplex-forming oligonucleotides. Proc. Natl Acad. Sci. USA, 97, 3862–3867.
Faria,M., Wood,C.D., White,M.R., Hélène,C. and Giovannangeli,C. (2001) Transcription inhibition induced by modified triple helix-forming oligonucleotides: a quantitative assay for evaluation in cells. J. Mol. Biol., 306, 15–24.
Saijo,M., Kuraoka,I., Masutani,C., Hanaoka,F. and Tanaka,K. (1996) Sequential binding of DNA repair proteins RPA and ERCC1 to XPA in vitro. Nucleic Acids Res., 24, 4719–4724.
Nocentini,S., Coin,F., Saijo,M., Tanaka,K. and Egly,J.M. (1997) DNA damage recognition by XPA protein promotes efficient recruitment of transcription factor II H. J. Biol. Chem., 272, 22991–22994.
Missura,M., Buterin,T., Hindges,R., Hubscher,U., Kasparkova,J., Brabec,V. and Naegeli,H. (2001) Double-check probing of DNA bending and unwinding by XPA–RPA: an architectural function in DNA repair. EMBO J., 20, 3554–3564.
Vasquez,K.M., Christensen,J., Li,L., Finch,R.A. and Glazer,P.M. (2002) Human XPA and RPA DNA repair proteins participate in specific recognition of triplex-induced helical distortions. Proc. Natl Acad. Sci. USA, 99, 5848–5853.
Wakasugi,M. and Sancar,A. (1998) Assembly, subunit composition and footprint of human DNA repair excision nuclease. Proc. Natl Acad. Sci. USA, 95, 6669–6674.
Jones,C.J. and Wood,R.D. (1993) Preferential binding of the xeroderma pigmentosum group A complementing protein to damaged DNA. Biochemistry, 32, 12096–12104.
Moggs,J.G., Szymkowski,D.E., Yamada,M., Karran,P. and Wood,R.D. (1997) Differential human nucleotide excision repair of paired and mispaired cisplatin–DNA adducts. Nucleic Acids Res., 25, 480–491.
Branum,M.E., Reardon,J.T. and Sancar,A. (2001) DNA repair excision nuclease attacks undamaged DNA. A potential source of spontaneous mutations. J. Biol. Chem., 276, 25421–25426.
Coudore,F., Calsou,P. and Salles,B. (1997) DNA repair activity in protein extracts from rat tissues. FEBS Lett., 414, 581–584.
Takasugi,M., Guendouz,A., Chassignol,M., Decout,J.L., Lhomme,J., Thuong,N.T. and Hélène,C. (1991) Sequence-specific photo-induced cross-linking of the two strands of double-helical DNA by a psoralen covalently linked to a triple helix-forming oligonucleotide. Proc. Natl Acad. Sci. USA, 88, 5602–5606.
Guieysse,A.L., Praseuth,D., Grigoriev,M., Harel-Bellan,A. and Hélène,C. (1996) Detection of covalent triplex within human cells. Nucleic Acids Res., 24, 4210–4216.
Giovannangeli,C., Thuong,N.T. and Hélène,C. (1993) Oligonucleotide clamps arrest DNA synthesis on a single-stranded DNA target. Proc. Natl Acad. Sci. USA, 90, 10013–10017.
Volker,M., Mone,M.J., Karmakar,P., van Hoffen,A., Schul,W., Vermeulen,W., Hoeijmakers,J.H., van Driel,R., van Zeeland,A.A. and Mullenders,L.H. (2001) Sequential assembly of the nucleotide excision repair factors in vivo. Mol. Cell, 8, 213–224.
Wakasugi,M. and Sancar,A. (1999) Order of assembly of human DNA repair excision nuclease. J. Biol. Chem., 274, 18759–18768.
Araujo,S.J. and Wood,R.D. (1999) Protein complexes in nucleotide excision repair. Mutat. Res., 435, 23–33.
Escudé,C., Giovannangeli,C., Sun,J.S., Lloyd,D.H., Gryaznov,S.M., Garestier,T. and Hélène,C. (1996) Stable triple helices formed by oligonucleotide N3'P5' phosphoramidates inhibit transcription elongation. Proc. Natl Acad. Sci. USA, 93, 4365–4369.
Duval-Valentin,G., Takasugi,M., Hélène,C. and Sage,E. (1998) Triple helix-directed psoralen crosslinks are recognized by Uvr(A)BC excinuclease. J. Mol. Biol., 278, 815–825.
Woodgate,R. (1999) A plethora of lesion-replicating DNA polymerases. Genes Dev., 13, 2191–2195.
McHugh,P.J., Sones,W.R. and Hartley,J.A. (2000) Repair of intermediate structures produced at DNA interstrand cross-links in Saccharomyces cerevisiae. Mol. Cell. Biol., 20, 3425–3433.(F. Guillonneau, A. L. Guieysse, S. Nocen)
*To whom correspondence should be addressed. Tel: +33 1 40 79 36 84; Fax: +33 1 40 79 37 05; Email: praseuth@mnhn.fr
ABSTRACT
Targeting DNA-damaging agents to specific DNA sites by using sequence-specific DNA ligands has been successful in directing genomic modifications. The understanding of repair processing of such targeted damage and the influence of the adjacent complex is largely unknown. In this way, directed interstrand cross-links (ICLs) have already been generated by psoralen targeting. The mechanisms responsible for ICL removal are far from being understood in mammalian cells, with the proposed involvement of both mutagenic and recombinogenic pathways. Here, a unique ICL was introduced at a selected site by photoactivation of a psoralen moiety with the use of psoralen conjugates of triplex-forming oligonucleotides. The processing of psoralen ICL was evaluated in vitro and in cells for two types of cross-linked substrates, either containing a psoralen ICL alone or with an adjacent triple-stranded structure. We show that the presence of a neighbouring triplex structure interferes with different stages of psoralen ICL processing: (i) the ICL-induced DNA repair synthesis in HeLa cell extracts is inhibited by the triplex structure, as measured by the efficiency of ‘true’ and futile repair synthesis, stopping at the ICL site; (ii) in HeLa cells, the ICL removal via a nucleotide excision repair (NER) pathway is delayed in the presence of a neighbouring triplex; and (iii) the binding to ICL of recombinant xeroderma pigmentosum A protein, which is involved in pre-incision recruitment of NER factors is impaired by the presence of the third DNA strand. These data characterize triplex-induced modulation of ICL repair pathways at specific steps, which might have implications for the controlled induction of targeted genomic modifications and for the associated cellular responses.
INTRODUCTION
Sequence-specific DNA ligands such as triplex-forming molecules are promising reagents that can interfere with DNA metabolism. TFOs have been successfully used to modulate gene transcription (initiation and elongation), and also replication and repair processes, in some cases leading to the induction of targeted genomic modifications . TFOs can serve as vectors for common DNA-damaging molecules, such as cross-linking or cleaving agents. Such bifunctional molecules provide useful tools to induce controlled DNA modifications, and the molecular events surrounding damage repair in the absence or presence of the oligonucleotide moiety must be characterized to elucidate the corresponding cellular responses.
Interstrand cross-links (ICLs) between nucleotides in complementary DNA strands are common lesions introduced into DNA by drugs such as psoralen, nitrogen mustards or cisplatin. The sequence of events involved in the removal of such cross-links is not yet understood in mammalian cells. Studies using cell lines deficient in specific repair proteins suggest that several repair pathways are probably involved in ICL repair. From these genetic approaches, both recombination-dependent and -independent pathways have been implicated (4,5), and an essential role for the ERCC1–XPF heterodimer has been described, this complex being involved in nucleotide excision repair (NER) and in other repair processes (6,7). In vitro studies have revealed that both mammalian cell extracts and purified components of NER can promote incisions on the 5' (8–10) and 3' (8,11) sides of the psoralen ICL. A double incision 5' to the psoralen ICL leads to a 5' gap (9), and this gap can be converted to a nick adjacent to the ICL following DNA synthesis (12). The implication of such futile synthesis in the repair process remains unknown. ICLs have been shown to induce DNA synthesis, not only in the damaged plasmid, but also in another plasmid present in the reaction. This hallmark of ICL repair is not well understood but requires XPF–ERCC1 and the recombination factors, XRCC2 and XRCC3 (6). The early stages of ICL repair in mammalian cells are also of intense interest with the recent identification of new components, such as the mismatch complex hMutS? (7).
In this study, we characterize the repair of a unique psoralen ICL, in the absence or presence of an adjacent triple-stranded structure, with the use of TFO–psoralen conjugates. We and others have previously used such conjugates and showed that: (i) in cells, they can induce targeted mutagenesis and stimulate recombination, both processes being influenced by the presence of the triplex structure, and both involving NER factors such as xeroderma pigmentosum A protein (XPA) (13–16); and (ii) in cell extracts, the incision near the cross-linked psoralen was inhibited by the presence of the third strand (10,17). In this report, we further elucidate the molecular basis for the different processing of these two types of cross-linked substrates, either containing a psoralen ICL alone or with an adjacent triple-stranded structure. We present evidence that: (i) in HeLa nuclear extracts, the presence of an adjacent triplex structure decreased the level of repair synthesis and mainly suppressed the production of one of the two products of futile repair synthesis stopping at the ICL; (ii) in HeLa cells, ICL removal by an NER-dependent pathway was delayed when a triple-stranded structure was present; and (iii) binding of human recombinant XPA to psoralen ICLs was inhibited by a neighbouring triplex. These data demonstrate triplex interference at specific stages of psoralen ICL repair. We will discuss the implications of our observations for the future design of strategies aimed at targeting controlled genomic modifications.
MATERIALS AND METHODS
Oligonucleotides
Psoralen TFO conjugates, Pso-S-S-16TC, Pso-16TC and Pso-15TCG, were described previously (10,13); for sequences, see Figure 1B.
Figure 1. Experimental system. (A) DNA substrates used in this study: two plasmids and two synthetic double helices (76 and 29 bp long), abbreviated as 76D and 29D, respectively (for sequences, see Materials and Methods). They all contain the oligopyrimidine·oligopurine sequence suitable for triplex formation (called HIV-PPT) (grey box). A psoralen ICL can be formed at the 5' TA 3' site coinciding with the DraI restriction site (5' TTT|AAA 3', underlined). (B) The third strand oligonucleotides are represented. C stands for 5-methylated cytosines. The psoralen molecule was attached to the phosphodiester 16TC TFO via a hexamethylene arm, either directly (Pso-16TC) or via an additional disulfide bond (Pso-S-S-16TC). The Pso-15TCG oligonucleotide was used with either a phosphodiester (PO) or phosphoramidate (NP) backbone. (C) The two types of substrates containing a psoralen ICL at the 5' TA 3' site used in this study are schematically represented. Left: DNA substrates with a unique psoralen ICL, obtained after Pso-S-S-16TC (cleavable) treatment (Pso-P, Pso-Pluc and Pso-76D, Pso-29D). Right: DNA substrates with a unique psoralen ICL and an adjacent triplex structure obtained after Pso-16TC or Pso-15TCG (uncleavable) treatment (16 or 15-Pso-P, 16 or 15-Pso-Pluc, and 16 or 15-Pso-76D, 16 or 15-Pso-29D) (for details see Materials and Methods). The triplex site is shown (grey box). The total lengths of the various DNA substrates are indicated.
Preparation of psoralen cross-linked DNA substrates
Two types of DNA substrates containing the oligopyrimidine·oligopurine sequence suitable for triplex formation (the polypurine tract sequence of HIV-1, abbreviated as HIV-PPT: 5' AAAAGAAAAGGGGGGA 3') (18) were used: either two plasmids, pSP-F47 (3360 bp) abbreviated as P (see details in Fig. 2A), and PGK/luc plasmid (8850 bp) abbreviated as Pluc (see details in Fig. 4), or two short duplex oligonucleotides (76D and 29D). The synthetic duplexes are formed with complementary oligonucleotides, and the sequence of the purine-containing strand is as follows: 5'-CAAGGCAGCT GTAGATCTTAG (CCACTTTTTAAAAGAAAAGGGGG GACTGG)AAGGGCTAATTCACTCCCAACA-3'; the 76D duplex includes the 29D duplex as indicated (parentheses). All these DNA substrates were modified at a 5' TA 3' motif adjacent to the triplex site in two different ways using psoralen conjugates: either Pso-S-S-16TC for generation of a substrate with a unique psoralen ICL, or Pso-16TC or Pso-15TCG (uncleavable) for generation of a substrate with a unique cross-linked triple-stranded structure (Fig. 1). In all cases, the DNA targets were incubated with the appropriate TFO to form the triplex (in 10 mM MES pH 6 or 10 mM Tris pH 7, 50 mM NaCl, 10 mM MgCl2), and UV irradiated according to Giovannangeli et al. (19) (monochromatic light, 365 nm). After the covalent triplex was formed, the disulfide bridge in Pso-S-S-16TC was reduced with 1 mM dithiothreitol (DTT; 37°C, 1 h incubation) in order to eliminate the third strand portion. The cleaved and unreacted oligonucleotides were removed by gel filtration chromatography (Chromaspin, Clontech) for plasmid substrates. Two constructs were obtained: plasmids or synthetic duplexes containing a single psoralen ICL at the PPT triplex site are designed by Pso-(P or Pluc) and Pso-(29 or 76)D, respectively. 15/16-Pso-P and 15/16-Pso-D stand for covalent triple-stranded structures obtained with uncleavable Pso-15TCG/Pso-16TC, respectively. Under the experimental conditions used, >90% of the double-stranded DNA target could be modified with a unique psoralen ICL (for an illustration, see Fig. 2B, lanes 1 and 2). The unmodified DNAs (P or Pluc, 76D or 29D) were treated in the same way as the corresponding modified substrate (irradiation, DTT reduction and chromatography).
Figure 2. Analysis of products obtained by repair processing of psoralen cross-linked substrates in HeLa cell extracts. Synthesis repair assays were performed (as described in Materials and Methods) and, after purification, DNA fragments were digested with the indicated restriction enzymes and revealed by denaturing PAGE. (A) Schematic representation of the pSP-F47 plasmid (P) in the region surrounding the defined psoralen ICL site. Cleavage sites of the restriction enzymes on both DNA strands are indicated by vertical bars. The numbers between the bars indicate the distances in nucleotides between the cleavage sites. The thymines that are cross-linked in the psoralen-modified substrates, Pso-P and 15-Pso-P, are indicated. The orientation of the psoralen molecule is described with the furane (F) and the pyrone (P) sides. A schematic map of the pSP-F47 plasmid is represented with the triplex site indicated by a box and the ICL site by a cross (X). The 80 or 43 fragments are unmodified fragments used as references using BglII–EcoRV or BglII–BsrI digestion, respectively. Fragments sizes are indicated. (B) Analysis of species after BglII–EcoRV digestion. Lanes 1–3: for the three plasmids, unmodified (P) or containing a unique psoralen ICL (Pso-P and 15-Pso-P), digestion fragments were 5' labelled (see Materials and Methods). Lanes 4–6 and 8–10: neo-synthesized DNA fragments obtained upon incubation with HeLa cell extracts. 69pu and 65py fragments correspond to full-length repair products, and contain the oligopyrimidine·oligopurine PPT sequence. The 52 and 17/16 nt fragments correspond to futile repair products stopping at the ICL site (pyrimidine and purine strand, respectively). The 80 nt fragments did not contain the ICL site and PPT target sequence (see A) and were used as references for quantification of relative repair activity (see Fig. 3A and Materials and Methods). Cross-linked species are indicated on the left of the gel. The asterisk indicates the fragments of neo-synthesis, religated to the cross-linked thymine (see text for details). Lane 7 is a size marker obtained by 5' labelling of digestion products; DraI digestions of neo-synthesized 65py and 69pu fragments were indistinguishable from futile synthesis products (52 and 17 nt fragments) (data not shown). (C) Analysis of neo-synthesized species after BglII and BsrI digestion for various plasmid templates, as indicated. Full-length (38pu and 32py) and futile repair (17 and 19) fragments are indicated. The 17 nt fragment is hardly detected (see stoichiometries in Fig. 3B). The 43 nt fragments were used as reference fragments since they do not contain the target sequence (see A). Lane 1 is a size marker obtained by 5' labelling of BglII/BsrI/DraI digestion products. A schematic representation of the substrates is given below (B) and (C): the triplex site is boxed, the adducted thymines are indicated and the various fragments are depicted with their respective lengths.
Figure 4. Kinetics of cross-link repair in cells. The PGK/luc plasmid (Pluc) expressing the firefly luciferase was modified in order to obtain a single cross-linked triplex (15-Pso-Pluc) or a unique psoralen ICL (Pso-Pluc) in the 5'-untranslated region (see Materials and Methods). Luciferase activities were measured at different times after transfection (24, 48 and 72 h) in HeLa cells. In all experiments, plasmids were co-transfected with a control vector expressing Renilla luciferase. Reporter activity is presented as the normalized ratio between firefly and Renilla luciferase activities in the same lysate (relative luciferase activity). Normalization from cells transfected with an unmodified plasmid P permits quantification of the cross-link repair efficiency. The error bars correspond to the mean deviation from duplicate experiments.
In vitro repair synthesis assay
The nucleotide incorporation assay was performed in a 50 μl reaction mixture, according to Wood et al. (20). Briefly, the plasmid (0.3 μg), either unmodified (P) or modified (Pso-P, 15-Pso-P), was incubated in HeLa cell nuclear extracts (80 μg) (Promega) in the presence of 2 mM ATP, 20 μM dATP, dTTP, dGTP, 8 μM dCTP and 2 μCi of dCTP (3000 Ci/mmol) in the presence of 40 mM creatine phosphate, 2.5 μM creatine phosphokinase. The buffer was 45 mM HEPES pH 7.8, 70 mM KCl, 7 mM MgCl2, 0.9 mM DTT, 0.4 mM EDTA, 3.5% glycerol, 0.36 mg/ml bovine serum albumin. Incubation was performed for 3 h at 30°C. After purification, the plasmid was cleaved with BglII and EcoRV (or BsrI), producing the fragments represented in Figure 2A. The 65py·69pu (or 32py·38pu) fragments contain the ICL site and the oligopyrimidine·oligopurine target sequence for triplex formation. In parallel, in order to characterize the nature of the fragments obtained after restriction cleavage, the plasmid (P, Pso-P or 15-Pso-P) was digested by the appropriate set of enzymes in the absence of incubation in extracts, and the fragments were 5'-labelled with T4 polynucleotide kinase (Biolabs) and ATP by standard procedures. In all cases, the products were analysed and quantitated by 8% acrylamide/bisacrylamide 19:1 sequencing PAGE.
Quantification of repair-induced neo-synthesized fragments
Full-length neo-synthesized products. Full-length synthesized fragments were quantitated taking into account the following factors. (i) The number of cytosines in various products, since we performed a dCTP incorporation assay as described above. It must be noted that the control fragments (80 for BglII–EcoRV digestion, and 43 for BglII–BsrI digestion) are composed of two strands that are not separated by electrophoresis in our conditions. The numbers of cytosines are as follows: 17 in 69pu, 16 in 65py, 42 in (80+80), five in 38pu, eight in 32py and 19 in (43+43). (ii) The detectable incorporation present in undamaged fragments (control fragments). (iii) The initial abundance of the fragments. It is thus possible to determine the amount of synthesized product specifically due to repair processing, i.e. the relative repair activity (Fig. 3A). It was calculated as explained in the following example, for the synthesis of the purine strand (69pu) with analysis with BglII–EcoRV digestion (see Fig. 2): the 80 nt fragments (abbreviated as 80+80), undamaged and adjacent to the target fragment, were used as references. Then the relative repair activity of 69pu was obtained by dividing (a) by (b) with: (a) the relative abundance of the neo-synthesized products 69pu, estimated by the ratio of radioactivity of the fragments 69pu to (80+80) obtained in the dCTP incorporation assay, corrected by the number of cytosines; and (b) the initial abundance of the products 69pu and (80+80), before reaction in the extracts, estimated by the ratio of radioactivity of the fragments (69pu + 65py) to (80+80) obtained in kinase labelling experiments. The relative repair activity obtained for the undamaged plasmid P should be close to 1 if the efficiencies of kinase labelling are identical for every fragment. In our system, higher efficiency was obtained for overhang versus blunt end labelling, and we chose to normalize to 1 the activity obtained for the undamaged plasmid P.
Figure 3. Quantification of the abundance of the neo-synthesized fragments due to repair processing of the different cross-linked DNA substrates. (A) Relative repair activities (R) of full-length repair products: purine (light grey) and pyrimidine (dark grey) strands were reported for each template (as indicated). They were obtained after dividing the relative abundance of neo-synthesized products (obtained by nucleotide incorporation) by their relative abundance in the initial mixture (obtained by kinase labelling) (average of three experiments; see details in Materials and Methods). (B) Schematic representation of the various neo-synthesis fragments obtained in vitro for the two different psoralen cross-linked substrates used in the present work. Purine and pyrimidine full-length or futile repair fragments are represented with their associated stoichiometries, calculated on the basis of fragment signal intensities and cytosine contents (mean from four experiments). For each of the full-length fragments, the background synthesis, estimated from (A) by comparison of the undamaged substrate (P) with the damaged ones (Pso-P and 15-Pso-P), was deduced in order to obtain the level of repair-induced synthesis.
It must be noted that we did not take into account recent results describing a possible stimulation of nucleotide incorporation when an undamaged plasmid was added to the solution in appropriate conditions, especially in excess compared with the damaged plasmid (6). In our experimental conditions, the amount of unmodified plasmid represented <10% and might not induce increased incorporation. If such incorporation occurs, our evaluation of relative repair activity would be under-estimated.
Futile neo-synthesized products. The abundance of these fragments was calculated on the basis of signal intensities and cytosine contents. The number of cytosines for the different fragments are: 15 in 52py, seven in 19py and four in 17pu. For information, the number of cytosines for fragments that are not observed is one for 13py, nine for 52pu and one for 21pu.
DNA repair assay in cells
The PGK/luc plasmid, abbreviated as Pluc, contained the firefly luciferase (Photinus pyralis) gene under the control of the phosphoglycerate kinase (PGK) promoter. A 34 bp insert (5' CCACTTTTTAAAAGAAAAGGGGGGACTGGAAGGG 3') containing the polypurine tract PPT/HIV-1 was cloned in the 5'-transcribed sequence of the luciferase gene (BamHI–HindIII fragment from the pGL2-basic vector; Promega), downstream of the putative PGK start site, as previously described in more detail (21,22). The Renilla luciferase (Renilla reniformis; RL) control vector (CMV-RL; Promega) was used to monitor transfection efficiency. The RL gene was inserted downstream of the cytomegalovirus (CMV) immediate-early enhancer/promoter region.
The Pluc plasmid was modified at 90–95% by a single psoralen ICL (Pso-Pluc) or by a single cross-linked triplex (15-Pso-Pluc), as described previously. The various Pluc plasmids (0.5 μg), modified or not, were co-transfected with the control vector expressing Renilla luciferase CMV-RL plasmid (0.005 μg), in HeLa cells (in 24-well plates) using Superfect transfection reagent (Qiagen), as described (21). The mixture was prepared for duplicates or triplicates. Both luciferase activities (firefly and Renilla) were measured in the same cell extract by using the dual-luciferase assay kit (Promega); the ratio (firefly/Renilla), called the relative luciferase activity, was used to monitor the cellular processing of the various modified plasmids with time.
Analysis of recombinant XPA binding to psoralen cross-linked substrates
GST–XPA (rXPA) was prepared according to Saijo et al. (23,24). Binding activity of rXPA was characterized by electrophoretic mobility shift assay (EMSA) (4% non-denaturing acrylamide gel in 1x TBE buffer), as described previously (25–27).
The duplex probes were purified on denaturating gels and 5'-32P-labelled on both strands. The radiolabelled target was incubated with rXPA (50 mM MES pH 6, 0.1 M NaCl, 10 mM MgCl2) according to Jones et al. (28). Incubation was performed for 1 h at 30°C and species were separated by EMSA. Mixtures with defined percentages of psoralen ICL at the triplex site were obtained by treating the sample (76D in the presence of Pso-S-S-16TC) for variable irradiation times. The amount of ICL was determined by denaturing electrophoresis and is expressed as the ratio Pso-76D:76D (Fig. 5A).
Figure 5. XPA binding to various types of psoralen ICL. (A) Gel retardation analysis of rXPA binding to the 76D duplex containing a unique psoralen ICL (Pso-6D) in various proportions, as indicated by the ratio Pso-76D:76D. The radiolabelled duplex concentration (Pso-76D + 76D) is 10 nM, and rXPA protein (1 μM) was added or not, as indicated. The amount of retarded complex L is indicated below the gel. (B) Triplex interferes with XPA binding to psoralen-modified duplexes. The damaged duplexes (16-Pso-76 or 29D and Pso-76 or 29D) were used as probes; complex formation was achieved by addition of rXPA (1 μM). Amounts of L complex (a.u.) obtained with various psoralen-damaged probes are reported; the values are normalized to that obtained for Pso-76/29D substrates. Values from one representative experiment that was independently repeated twice are presented. Equivalent results were obtained with the 76D and the 29D substrates.
Results
Substrates with a single psoralen interstrand cross-link
Plasmids or short synthetic duplexes containing a specific oligopyrimidine·oligopurine sequence (named HIV-PPT) were modified with two types of psoralen derivatives: the goal was to obtain at the same defined site in DNA a single psoralen ICL, either alone (by treatment with cleavable Pso-S-S-16TC) or associated with an adjacent triplex (by treatment with uncleavable Pso-16TC or Pso-15-TCG) (Fig. 1, and details in Materials and Methods). Following treatment with TFO–psoralen conjugates, the efficiency of ICL formation for each substrate was evaluated by electrophoresis analysis after 5' labelling. As an example, ICL induction in the plasmid pSP-F47 (P) was determined by 5' labelling of DNA fragments obtained by BglII and EcoRV digestion followed by electrophoresis analysis (Fig. 2B). The mobility of cross-linked products was reduced compared with that of undamaged fragments. The 69pu and 65py fragments containing the oligopyrimidine·oligopurine target were converted to cross-linked fragments with an efficiency of 90%, whereas the 80 nt fragment lacking the target was unaffected (Fig. 2B, lanes 1–3). Equivalent levels of modifications were obtained with synthetic oligonucleotide substrates.
Repair synthesis of the specific target fragment is decreased by the presence of an adjacent triple-stranded structure
The initial stages of ICL repair require incisions to be made in the vicinity of the cross-linked site; these incisions are associated with DNA repair synthesis. To analyse the repair synthesis in vitro, damaged plasmids were incubated in the presence of HeLa cell extracts, and DNA synthesis, induced by a single psoralen ICL that was introduced at a defined site in the Pso-P or 15-Pso-P plasmids, was measured as the specific incorporation of labelled dCTP into damage-containing fragments. The competency of the extracts to perform NER in our experimental conditions was evaluated by using a plasmid carrying a unique cisplatin ICL between two guanines in a GTG sequence. This cisplatin lesion is known to be a good substrate for NER (29) and therefore serves as a positive control. Repair activity was detected since a fragment containing the cisplatin cross-link (33 nt long) became specifically 3- to 4-fold more radioactive in the damaged plasmid compared with the control undamaged plasmid (data not shown).
Incorporation of labelled nucleotides reflects a neo-synthesis activity, and the different products have been analysed on denaturing PAGE. The damage-containing fragments, 69(38)pu or 65(32)py, were generated by double digestion with BglII and EcoRV (BsrI) as shown in Figure 2A. The radioactive signals corresponding to intact purine and pyrimidine fragments were significantly higher for cross-linked substrates than what would be expected from non-specific incorporation into the small residual undamaged fragments. Noticeably a background synthesis independent of repair (30,31) was also detected in the undamaged fragments present in the control sample (Fig. 2B, lane 6, and C, lane 4). This is attributed to nicking activities present in the extracts or arising as a consequence of the irradiation step. The neo-synthesis associated with repair processes (relative repair activity) was evaluated for each strand (see Materials and Methods) and is presented in Figure 3A. Levels of repair synthesis of pyrimidine and purine strands were increased (2.5- to 5-fold) in cross-linked compared with control plasmids, and were slightly decreased (30–40%) by the presence of the triplex structure.
Futile repair synthesis depends on the nature of the cross-linked substrate
In addition to the full-length fragments described above, shorter neo-synthesized products were also observed; their sizes corresponded to the length (±1 nt) between the proximal enzymatic cleavage site and the position of the psoralen ICL located at the DraI digestion site. In Figure 2B (lanes 4 and 8) and C (lane 2), short fragments of 52(19)py and 17/16pu were clearly detected (according to the digest). Triplex did not affect the production of the short purine (17/16 nt) fragment (Fig. 2B, lanes 8 and 9), whereas the pyrimidine short fragment was completely abolished (Fig. 2B, lanes 4 and 5, and C, lanes 2 and 3).
These truncated neo-synthesis fragments are consistent with the so-called futile DNA synthesis products described by Mu et al. (12). These authors reported that a psoralen ICL induced a DNA synthesis that converted an excision gap 5' to the ICL into a nick without removing the damage. Several findings support the idea that our observed truncated neo-synthesized products are the result of specific repair synthesis. First, we observed a synthesis signal 13-fold stronger for the pyrimidine strand compared with that for the purine strand (ICL in the Pso-P plasmid) (Figs 2 and 3B). Such a 10-fold ratio was observed previously for the futile synthesis induced by the pyrone-side adducted thymine compared with the furane-side adducted thymine (12), reflecting a differential incision efficiency as reported by Bessho et al. (9). Such stoichiometry is consistent with the preferential orientation of the psoralen moiety in our system, with the furane side adducted on the purine strand and the pyrone side on the pyrimidine strand . Secondly, we observed a lack of futile synthesis specifically for the pyrimidine strand in the presence of the triplex structure (15-Pso-P plasmid). These data are consistent with our previous demonstration that the incision step during repair of a psoralen cross-link was inhibited at the triplex site in the presence of the third strand in HeLa nuclear extracts (10). In addition, the fact that no product was detected for the pyrimidine strand using 15-Pso-P plasmid as a substrate reflected a synthesis caused by damage repair and not originating from a repair-independent incision downstream of the lesion. Indeed, such an incision should have been elongated and detected since DNA polymerization could initiate downstream of a covalent triplex (data not shown based on primer extension experiments with Klenow DNA polymerase) (33,34). Thirdly, production of these truncated fragments by nucleotide incorporation until the damage position required the presence of ATP (data not shown).
Finally, for the Pso-P plasmid, a product corresponding to religation of futile synthesis repair fragments to thymine still carrying the psoralen was detected (shown by an asterisk in Fig. 2B, lane 4), as reported previously in another system (12). This product was detected when Pso-P plasmid was used as a substrate, but not for the 15-Pso-P plasmid (Fig. 2B, lane 5). This lack of religated product in the presence of the triplex is consistent with the notion that (i) the ligated product resulted from the two futile synthesis products, and therefore could not be produced in the presence of the third strand; and also that (ii) for the futile-synthesized purine strand, which is produced in the presence of the third strand, the triplex could inhibit the ligation step.
A triple-stranded structure delayed cross-link repair in cells
Since we observed that a triple-stranded structure inhibited ICL repair by interfering with the repair machinery at specific stages (incision, repair synthesis and religation) in vitro, we decided to evaluate the processing of cross-linked substrates in cells. The PPT sequence was cloned between the promoter and the coding region of the luciferase reporter gene (PGK/luc plasmid, named Pluc). This plasmid was modified with either cleavable Pso-S-S-16TC (Pso-Pluc plasmid) or uncleavable Pso-15TCG (15-Pso-Pluc plasmid), and then transfected into HeLa cells. Luciferase reactivation was used to monitor ICL removal by a recombination-independent and error-prone repair pathway, involving NER factors, as described previously (5). Luciferase activity was measured at different times after transfection (Fig. 4). At 24 h post-transfection, luciferase activity was inhibited by 90–95% for Pso-Pluc and 15-Pso-Pluc, in agreement with the yield of psoralen ICL obtained for each construct. At later times, luciferase reactivation increased and, at all times, we observed that luciferase activity from Pso-Pluc was twice that for 15-Pso-Pluc. This result is consistent with a 2-fold lower repair rate of 15-Pso-Pluc compared with Pso-Pluc in this assay.
Recombinant XPA binds to a fragment containing a unique psoralen interstrand cross-link and complex formation is influenced by an adjacent triplex structure
In cells, NER factors have been shown to be sequentially assembled and XPA is involved in a pre-incision step for the loading of different components of the NER complex (35). In vitro, XPA has been described in some cases as a damage-sensing factor (25,36,37). In order to characterize further the effect of a triplex structure on NER processing of psoralen ICLs, we examined the binding affinity of recombinant human protein rXPA for duplexes containing a unique psoralen ICL, alone or in the presence of a triple helix. For this purpose, two target duplexes of different lengths (76D and 29D, see sequences in Materials and Methods) were modified with a unique psoralen ICL (in Pso-76D or Pso-29D) or with a unique covalent triplex (in 16/15-Pso-76D or 16/15-Pso-29D; Fig. 1). These substrates were incubated in the presence of recombinant rXPA and the resulting complexes were analysed by gel retardation assay (Fig. 5). The complex was detected as a retarded band (L) on top of the gel. The amount of L complex increased proportionally with the amount of psoralen-modified duplex (Fig. 5A), indicating that the L complex represents a specific association between rXPA and the damaged DNA. We have estimated the protein concentration that bound 50% of the substrate: for Pso-76D, Kd = 0.3 x 10–6 M, compared with 10–6 M for unmodified 76D duplex. These values agree with previously published values (Kd = 0.3 x 10–6 M versus 1.7 x 10–6 M for a 258 bp fragment in the presence or absence of four sites of UV damage, respectively) (28).
The proportion of complex was quantified for various damaged substrates (Fig. 5B). Complex formation was inhibited to varying extents depending on the nature of the third strand: 54% inhibition with the phosphodiester Pso-16TC (PO) (16-Pso-29D substrate), 70% inhibition with the phosphodiester Pso-15TCG(PO) and 85% with the phosphoramidate Pso-15TCG(NP) . The level of rXPA binding thus decreased as the thermodynamic stability of the triplex increased (38). However, the covalent triplex was still better recognized by rXPA than undamaged DNA. Similar results were obtained concerning rXPA binding affinities independently of the length of the various substrates (D, Pso-D or 16/15-Pso-D) in the tested range, 76 or 29 bp long, and also independently of the substrate sequence (another sequence suitable for triplex formation was tested 5' AAAAGGAGAGGGAGA 3'; data not shown). In contrast, in previous studies performed with prokaryote repair systems, triplexes were found to inhibit damage processing at the cleavage but not at the recognition step (39).
These data on rXPA binding must be considered with respect to helical distortions induced by a DNA triplex and the sensitivity of XPA to conformational alterations in the helical structure. The nature of the triplex (length of the third strand, Hoogsteen or reverse Hoogsteen motif, position of the ICL site at the 5' or 3' side of the triplex) could strongly influence induced distortion and, consequently, the affinity of XPA for the lesion (26). Further studies are needed to evaluate the precise influence of this set of factors.
Finally, the lower rXPA binding to psoralen ICL observed in the presence of an adjacent triplex structure is consistent with the decrease in luciferase reactivation described above, reflecting a decrease in cellular removal of psoralen ICL with a triplex, since this reactivation has been shown to be XPA dependent (5).
Discussion
This report shows that a psoralen ICL was differentially recognized whether alone or in the context of a DNA triple-helical structure by elements of the repair machinery, both in vitro and in cells. To summarize our findings, the presence of the third strand influenced different stages of psoralen ICL processing: (i) repair synthesis in HeLa cell extracts is inhibited by the triplex structure, as schematically described in Figure 3B; (ii) the ICL removal involving NER factors is delayed in HeLa cells in the presence of a triplex; and (iii) the triplex structure impaired recombinant XPA binding of a neighbouring psoralen ICL and the level of this inhibition is correlated with the binding affinity of the third strand to DNA. Triplex-forming molecules have been successfully used to position DNA-damaging agents such as psoralen molecules, with the goal of inducing targeted genomic modifications. In mammalian cells, there are indirect data demonstrating the contribution of NER factors, especially XPA, in mutagenesis and recombination processes induced by a cross-linked triplex (14,16). The molecular basis of processes leading to targeted DNA modifications is far from understood. In this context, we will discuss how our data may be useful in advancing this characterization and in designing future strategies aimed at targeting controlled genomic modifications.
The exact mechanism of psoralen ICL repair is still poorly understood in mammalian systems. Initially, repair processes are based on generation of various incisions in the vicinity of the lesion. At present, the factors responsible for these incisions as well as their exact sites remain to be determined. We and others did not directly detect species corresponding to 3' cleavage (9,10). However, such 3' incisions were directly detected in different experimental systems (7,8,11), and have been attributed to NER proteins with a major role in the ERCC1–XPF complex and recently also to MMRs (MutS?). Combining the results presented in this work and those of several other recent studies leads us to a putative model for the repair of the two types of studied cross-linked substrates (summarized in Fig. 6). We propose various scenarios starting from the futile synthesis fragments stopping at the psoralen ICL, as intermediates in the ICL repair process, including modulation by an adjacent triplex structure .
Figure 6. Model for repair of psoralen ICL and modulation by an adjacent triplex structure, based on the presented findings. Three possible repair pathways are described with futile synthesis repair fragments (grey arrows) as a starting point. (A) Following filling in, a small fraction of molecules are ligated to regenerate a cross-linked substrate. (B) 3' Cleavage (indicated by a vertical arrow) occurring on the same strand as the neo-synthesized fragment produces a gap which can be replenished by translesion followed by ligation and, probably, an additional cleavage process to completely remove the damage. This pathway is mutagenic. (C) 3' Cleavage (indicated by a vertical arrow) occurring on the strand opposite to the neo-synthesized product produces a DSB that can be repaired by either error-free HR or error-prone EJ, SSA and BIR processes (see text for detailed discussion). The different stages where the triplex was shown in the present study to interfere with ICL repair are indicated by (–). It is noticeable that for the cross-linked triplex, the pathway in (A) seems to be ineffective, and the pathways in (B) and (C) can only occur for the truncated neo-synthesized purine strand since a complete blockage of futile synthesis of the pyrimidine strand is observed (see Fig. 3B). The triplex site is boxed (in grey) and the adducted thymines are indicated.
A small fraction of these truncated fragments can be religated to the adducted thymines (Fig. 6A). We showed here that religation leading to regeneration of a cross-linked substrate was inhibited in the presence of the triplex structure (Fig. 2B). Consequently, the pathway shown in Figure 6A does seem to be ineffective for psoralen ICL in the presence of a triplex.
Futile synthesis products could also be processed in order to eliminate the ICL. Mainly, two pathways could be envisaged (Fig. 6B and C).
In the pathway shown in Figure 6B, a nick occurs 3' to the lesion, on the same strand as the incision 5' to the lesion which led to futile synthesis. DNA synthesis can then fill in the resulting gap by translesion bypass. Such translesion synthesis has already been characterized; it involves a specialized distributive polymerase (probably a homologue of Saccharomyces cerevisiae Rad30), more tolerant to structural defaults (40). It must be noted that the finding of the possible involvement of 3'–5' exonuclease activity of ERCC1–XPF in the presence of RPA (12) would even make the 5' cleavage unnecessary, and lead to the same end result. In any case, the remaining damage could then be removed by an additional cleavage process. This mechanism would predict that the base opposing the cross-linked thymine would be a mutation hot-spot as a result of the bypass synthesis. We have directly measured triplex interference during this pathway. Indeed, we have quantified the cellular removal of a site-specific psoralen ICL, either alone or with an adjacent triplex structure, using reactivation of a reporter gene when the damage was located in the 5'-transcribed but untranslated region. In HeLa cells, we have observed a decrease in ICL removal in the presence of the triplex structure. This result supports our in vitro data discussed above describing the triplex-induced inhibition of the repair machinery, at different levels (incision, repair synthesis). Such inhibition of this pathway might be associated with an increased contribution of the recombination pathway (Fig. 6C) for triplex-associated ICL repair.
In the pathway shown in Figure 6C, a nick is produced 3' to the lesion, on the strand opposite to the incision 5' to the lesion that led to futile synthesis. A double-strand break (DSB) can appear as an intermediate during ICL repair, as already demonstrated, both in cells and in vitro (4,7,41). In our experimental conditions we did not reveal any DSB. Another way to generate such a DSB could be an arrest of the replication fork near the ICL site in dividing cells. In any case, recombination-dependent repair will take place by different mechanisms; they could be error-free or error-prone processes.
Our data demonstrate triplex-induced inhibition of the repair machinery in vitro and in cells, and suggest an increased contribution of the pathway shown in Figure 6C compared with that shown in Figure 6B. These findings are consistent with mutagenesis and recombination results obtained with a positioned psoralen ICL, alone or with an adjacent triplex: (i) the mutation frequency (Fig. 6B) was lower in the presence of the triplex structure (17), whereas the recombination frequency (Fig. 6C) was higher (14); and (ii) two types of mutation are induced around the site of the specific ICL for both types of ICL (in the presence or absence of the triplex structure); they are point mutations facing the adducted thymine (generated from the pathway shown in Fig. 6B) and deletions (generated from the pathway shown in Fig. 6C). The deletions were longer in the presence of the triplex (17).
The present work advances the molecular understanding of processes leading to targeted modifications of DNA in mammalian cells with the use of specific DNA ligands to direct DNA-damaging molecules. Such a study might help in the improvement of, and new developments in, genetic manipulation of mammalian cells and in the use of DNA-targeted anti-tumour agents.
ACKNOWLEDGEMENTS
We thank U. Asseline for the synthesis of psoralen-cleavable oligonucleotides, J. Moggs for the gift of platinated plasmid and monoclonal anti-XPA antibodies, and J. Mello for careful reading of the manuscript. F.G. was supported by ARC (Association de la Recherche sur le Cancer) and AFM (Association Fran?aise sur la Myopathie) fellowships.
REFERENCES
Faria,M. and Giovannangeli,C. (2001) Triplex-forming molecules: from concepts to applications. J. Gene Med., 3, 299–310.
Knauert,M.P. and Glazer,P.M. (2001) Triplex forming oligonucleotides: sequence-specific tools for gene targeting. Hum. Mol. Genet., 10, 2243–2251.
Seidman,M.M. and Glazer,P.M. (2003) The potential for gene repair via triple helix formation. J. Clin. Invest., 112, 487–494.
De Silva,I.U., McHugh,P.J., Clingen,P.H. and Hartley,J.A. (2000) Defining the roles of nucleotide excision repair and recombination in the repair of DNA interstrand cross-links in mammalian cells. Mol. Cell. Biol., 20, 7980–7990.
Wang,G., Chen,Z., Zhang,S., Wilson,G.L. and Jing,K. (2001) Detection and determination of oligonucleotide triplex formation-mediated transcription-coupled DNA repair in HeLa nuclear extracts. Nucleic Acids Res., 29, 1801–1807.
Li,L., Peterson,C.A., Lu,X., Wei,P. and Legerski,R.J. (1999) Interstrand cross-links induce DNA synthesis in damaged and undamaged plasmids in mammalian cell extracts. Mol. Cell. Biol., 19, 5619–5630.
Zhang,N., Lu,X., Zhang,X., Peterson,C.A. and Legerski,R.J. (2002) hMutSbeta is required for the recognition and uncoupling of psoralen interstrand cross-links in vitro. Mol. Cell. Biol., 22, 2388–2397.
Kumaresan,K.R., Hang,B. and Lambert,M.W. (1995) Human endonucleolytic incision of DNA 3' and 5' to a site-directed psoralen monoadduct and interstrand cross-link. J. Biol. Chem., 270, 30709–30716.
Bessho,T., Mu,D. and Sancar,A. (1997) Initiation of DNA interstrand cross-link repair in humans: the nucleotide excision repair system makes dual incisions 5' to the cross-linked base and removes a 22- to 28-nucleotide-long damage-free strand. Mol. Cell. Biol., 17, 6822–6830.
Guieysse,A.L., Praseuth,D., Giovannangeli,C., Asseline,U. and Hélène,C. (2000) Psoralen adducts induced by triplex-forming oligonucleotides are refractory to repair in HeLa cells. J. Mol. Biol., 296, 373–383.
Kuraoka,I., Kobertz,W.R., Ariza,R.R., Biggerstaff,M., Essigmann,J.M. and Wood,R.D. (2000) Repair of an interstrand DNA cross-link initiated by ERCC1–XPF repair/recombination nuclease J. Biol. Chem., 275, 26632–26636.
Mu,D., Bessho,T., Nechev,L.V., Chen,D.J., Harris,T.M., Hearst,J.E. and Sancar,A. (2000) DNA interstrand cross-links induce futile repair synthesis in mammalian cell extracts. Mol. Cell. Biol., 20, 2446–2454.
Barre,F.X., Asseline,U. and Harel-Bellan,A. (1999) Asymmetric recognition of psoralen interstrand crosslinks by the nucleotide excision repair and the error-prone repair pathways. J. Mol. Biol., 286, 1379–1387.
Faruqi,A.F., Datta,H.J., Carroll,D., Seidman,M.M. and Glazer,P.M. (2000) Triple-helix formation induces recombination in mammalian cells via a nucleotide excision repair-dependent pathway. Mol. Cell. Biol., 20, 990–1000.
Raha,M., Wang,G., Seidman,M.M. and Glazer,P.M. (1996) Mutagenesis by third-strand-directed psoralen adducts in repair-deficient human cells: high frequency and altered spectrum in a xeroderma pigmentosum variant. Proc. Natl Acad. Sci. USA, 93, 2941–2946.
Wang,G., Seidman,M.M. and Glazer,P.M. (1996) Mutagenesis in mammalian cells induced by triple helix formation and transcription-coupled repair. Science, 271, 802–805.
Wang,G. and Glazer,P.M. (1995) Altered repair of targeted psoralen photoadducts in the context of an oligonucleotide-mediated triple helix. J. Biol. Chem., 270, 22595–22601.
Giovannangeli,C., Perrouault,L., Escude,C., Nguyen,T. and Hélène,C. (1996) Specific inhibition of in vitro transcription elongation by triplex-forming oligonucleotide–intercalator conjugates targeted to HIV proviral DNA. Biochemistry, 35, 10539–10548.
Giovannangeli,C., Thuong,N.T. and Hélène,C. (1992) Oligodeoxynucleotide-directed photo-induced cross-linking of HIV proviral DNA via triple-helix formation. Nucleic Acids Res., 20, 4275–4281.
Wood,R.D., Robins,P. and Lindahl,T. (1988) Complementation of the xeroderma pigmentosum DNA repair defect in cell-free extracts. Cell, 53, 97–106.
Faria,M., Wood,C.D., Perrouault,L., Nelson,J.S., Winter,A., White,M.R., Hélène,C. and Giovannangeli,C. (2000) Targeted inhibition of transcription elongation in cells mediated by triplex-forming oligonucleotides. Proc. Natl Acad. Sci. USA, 97, 3862–3867.
Faria,M., Wood,C.D., White,M.R., Hélène,C. and Giovannangeli,C. (2001) Transcription inhibition induced by modified triple helix-forming oligonucleotides: a quantitative assay for evaluation in cells. J. Mol. Biol., 306, 15–24.
Saijo,M., Kuraoka,I., Masutani,C., Hanaoka,F. and Tanaka,K. (1996) Sequential binding of DNA repair proteins RPA and ERCC1 to XPA in vitro. Nucleic Acids Res., 24, 4719–4724.
Nocentini,S., Coin,F., Saijo,M., Tanaka,K. and Egly,J.M. (1997) DNA damage recognition by XPA protein promotes efficient recruitment of transcription factor II H. J. Biol. Chem., 272, 22991–22994.
Missura,M., Buterin,T., Hindges,R., Hubscher,U., Kasparkova,J., Brabec,V. and Naegeli,H. (2001) Double-check probing of DNA bending and unwinding by XPA–RPA: an architectural function in DNA repair. EMBO J., 20, 3554–3564.
Vasquez,K.M., Christensen,J., Li,L., Finch,R.A. and Glazer,P.M. (2002) Human XPA and RPA DNA repair proteins participate in specific recognition of triplex-induced helical distortions. Proc. Natl Acad. Sci. USA, 99, 5848–5853.
Wakasugi,M. and Sancar,A. (1998) Assembly, subunit composition and footprint of human DNA repair excision nuclease. Proc. Natl Acad. Sci. USA, 95, 6669–6674.
Jones,C.J. and Wood,R.D. (1993) Preferential binding of the xeroderma pigmentosum group A complementing protein to damaged DNA. Biochemistry, 32, 12096–12104.
Moggs,J.G., Szymkowski,D.E., Yamada,M., Karran,P. and Wood,R.D. (1997) Differential human nucleotide excision repair of paired and mispaired cisplatin–DNA adducts. Nucleic Acids Res., 25, 480–491.
Branum,M.E., Reardon,J.T. and Sancar,A. (2001) DNA repair excision nuclease attacks undamaged DNA. A potential source of spontaneous mutations. J. Biol. Chem., 276, 25421–25426.
Coudore,F., Calsou,P. and Salles,B. (1997) DNA repair activity in protein extracts from rat tissues. FEBS Lett., 414, 581–584.
Takasugi,M., Guendouz,A., Chassignol,M., Decout,J.L., Lhomme,J., Thuong,N.T. and Hélène,C. (1991) Sequence-specific photo-induced cross-linking of the two strands of double-helical DNA by a psoralen covalently linked to a triple helix-forming oligonucleotide. Proc. Natl Acad. Sci. USA, 88, 5602–5606.
Guieysse,A.L., Praseuth,D., Grigoriev,M., Harel-Bellan,A. and Hélène,C. (1996) Detection of covalent triplex within human cells. Nucleic Acids Res., 24, 4210–4216.
Giovannangeli,C., Thuong,N.T. and Hélène,C. (1993) Oligonucleotide clamps arrest DNA synthesis on a single-stranded DNA target. Proc. Natl Acad. Sci. USA, 90, 10013–10017.
Volker,M., Mone,M.J., Karmakar,P., van Hoffen,A., Schul,W., Vermeulen,W., Hoeijmakers,J.H., van Driel,R., van Zeeland,A.A. and Mullenders,L.H. (2001) Sequential assembly of the nucleotide excision repair factors in vivo. Mol. Cell, 8, 213–224.
Wakasugi,M. and Sancar,A. (1999) Order of assembly of human DNA repair excision nuclease. J. Biol. Chem., 274, 18759–18768.
Araujo,S.J. and Wood,R.D. (1999) Protein complexes in nucleotide excision repair. Mutat. Res., 435, 23–33.
Escudé,C., Giovannangeli,C., Sun,J.S., Lloyd,D.H., Gryaznov,S.M., Garestier,T. and Hélène,C. (1996) Stable triple helices formed by oligonucleotide N3'P5' phosphoramidates inhibit transcription elongation. Proc. Natl Acad. Sci. USA, 93, 4365–4369.
Duval-Valentin,G., Takasugi,M., Hélène,C. and Sage,E. (1998) Triple helix-directed psoralen crosslinks are recognized by Uvr(A)BC excinuclease. J. Mol. Biol., 278, 815–825.
Woodgate,R. (1999) A plethora of lesion-replicating DNA polymerases. Genes Dev., 13, 2191–2195.
McHugh,P.J., Sones,W.R. and Hartley,J.A. (2000) Repair of intermediate structures produced at DNA interstrand cross-links in Saccharomyces cerevisiae. Mol. Cell. Biol., 20, 3425–3433.(F. Guillonneau, A. L. Guieysse, S. Nocen)