Hydroxymethyl)-4,5',8-trimethylpsoralen Research Articles

Tools to Study the Role of Architectural Protein HMGB1 in the Processing of Helix Distorting, Site-specific DNA Interstrand Crosslinks.

High mobility group box 1 (HMGB1) protein is a non-histone architectural protein that is involved in regulating many important functions in the genome, such as transcription, DNA replication, and DNA repair. HMGB1 binds to structurally distorted DNA with higher affinity than to canonical B-DNA. For example, we found that HMGB1 binds to DNA interstrand crosslinks (ICLs), which covalently link the two strands of the DNA, cause distortion of the helix, and if left unrepaired can cause cell death. Due to their cytotoxic potential, several ICL-inducing agents are currently used as chemotherapeutic agents in the clinic. While ICL-forming agents show preferences for certain base sequences (e.g., 5'-TA-3' is the preferred crosslinking site for psoralen), they largely induce DNA damage in an indiscriminate fashion. However, by covalently coupling the ICL-inducing agent to a triplex-forming oligonucleotide (TFO), which binds to DNA in a sequence-specific manner, targeted DNA damage can be achieved. Here, we use a TFO covalently conjugated on the 5' end to a 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) psoralen to generate a site-specific ICL on a mutation-reporter plasmid to use as a tool to study the architectural modification, processing, and repair of complex DNA lesions by HMGB1 in human cells. We describe experimental techniques to prepare TFO-directed ICLs on reporter plasmids, and to interrogate the association of HMGB1 with the TFO-directed ICLs in a cellular context using chromatin immunoprecipitation assays. In addition, we describe DNA supercoiling assays to assess specific architectural modification of the damaged DNA by measuring the amount of superhelical turns introduced on the psoralen-crosslinked plasmid by HMGB1. These techniques can be used to study the roles of other proteins involved in the processing and repair of TFO-directed ICLs or other targeted DNA damage in any cell line of interest.

Efficient processing of TFO-directed psoralen DNA interstrand crosslinks by the UvrABC nuclease

Photoreactive psoralens can form interstrand crosslinks (ICLs) in double-stranded DNA. In eubacteria, the endonuclease UvrABC plays a key role in processing psoralen ICLs. Psoralen-modified triplex-forming oligonucleotides (TFOs) can be used to direct ICLs to specific genomic sites. Previous studies of pyrimidine-rich methoxypsoralen–modified TFOs indicated that the TFO inhibits cleavage by UvrABC. Because different chemistries may alter the processing of TFO-directed ICLs, we investigated the effect of another type of triplex formed by purine-rich TFOs on the processing of 4′-(hydroxymethyl)-4,5′,8-trimethylpsoralen (HMT) ICLs by the UvrABC nuclease. Using an HMT-modified TFO to direct ICLs to a specific site, we found that UvrABC made incisions on the purine-rich strand of the duplex ∼3 bases from the 3′-side and ∼9 bases from the 5′-side of the ICL, within the TFO-binding region. In contrast to previous reports, the UvrABC nuclease cleaved the TFO-directed psoralen ICL with a greater efficiency than that of the psoralen ICL alone. Furthermore, the TFO was dissociated from its duplex binding site by UvrA and UvrB. As mutagenesis by TFO-directed ICLs requires nucleotide excision repair, the efficient processing of these lesions supports the use of triplex technology to direct DNA damage for genome modification.

The crystal structures of psoralen cross-linked DNAs: drug-dependent formation of Holliday junctions

The single-crystal structures are presented for two DNA sequences with the thymine bases covalently cross-linked across the complementary strands by 4′-hydroxymethyl-4,5′,8-trimethylpsoralen (HMT). The HMT-adduct of d(CCGCTAGCGG) forms a psoralen-induced Holliday junction, showing for the first time the effect of this important class of chemotheraputics on the structure of the recombination intermediate. In contrast, HMT-d(CCGGTACCGG) forms a sequence-dependent junction. In both structures, the DNA duplex is highly distorted at the thymine base linked to the six-member pyrone ring of the drug. The psoralen cross-link defines the intramolecular interactions of the drug-induced junction, while the sequence-dependent structure is nearly identical to the native Holliday junction of d(CCGGTACCGG) alone. The two structures contrast the effects of drug- and sequence-dependent interactions on the structure of a Holliday junction, suggesting a role for psoralen in the mechanism to initiate repair of psoralen-lesions in mammalian DNA.

DNA replication is required To elicit cellular responses to psoralen-induced DNA interstrand cross-links.

Following introduction of DNA interstrand cross-links (ICLs), mammalian cells display chromosome breakage or cell cycle delay with a 4N DNA content. To further understand the nature of the delay, previously described as a G(2)/M arrest, we developed a protocol to generate ICLs during specific intervals of the cell cycle. Synchronous populations of G(1), S, and G(2) cells were treated with photoactivated 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) and scored for normal passage into mitosis. In contrast to what was found for ionizing radiation, ICLs introduced during G(2) did not result in a G(2)/M arrest, mitotic arrest, or chromosome breakage. Rather, subsequent passage through S phase was required to trigger both chromosome breakage and arrest in the next cell cycle. Similarly, ICLs introduced during G(1) did not cause a G(1)/S arrest. We conclude that DNA replication is required to elicit the cellular responses of cell cycle arrest and genomic instability after psoralen-induced ICLs. In primary human fibroblasts, the 4N DNA content cell cycle arrest triggered by ICLs was long lasting but reversible. Kinetic analysis suggested that these cells could remove up to approximately 2,500 ICLs/genome at an average rate of 11 ICLs/genome/h.

Two-photon excitation of 4'-hydroxymethyl-4,5',8-trimethylpsoralen.

Psoralens are a class of pharmaceutical agents commonly used to treat several cutaneous disorders. When irradiated with a mode-locked titanium: sapphire (Ti:sapphire) laser tuned to 730 nm, an aqueous solution of 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) emits blue light. The emission spectrum is centered at 452 nm and is identical to that obtained by one-photon excitation with UVA excitation, and its magnitude depends quadratically on the intensity of laser excitation. These results suggest that two-photon excitation occurs to a potentially photochemically active state. To estimate the two-photon absorption cross section, it was first necessary to measure the emission quantum yield of HMT using 365 nm excitation at room temperature that resulted in a value of 0.045 +/- 0.007. The two-photon absorption cross section of HMT at 730 nm is therefore estimated to be 20 x 10(-50) cm4 s (20 Göppert-Mayer). The excited-state photophysics and photochemistry of psoralens suggest potential applications to cutaneous phototherapy in diseases such as psoriasis and dystrophic epidermolysis bullosa.

High-efficiency triple-helix-mediated photo-cross-linking at a targeted site within a selectable mammalian gene.

Targeting damage to specific sites in the genome represents an attractive approach to manipulating gene function in mammalian cells. To test the applicability of triple-helix formation as a means for achieving precisely timed site-specific damage within a mammalian gene, a triplex-forming oligodeoxyribonucleotide (TFO) that binds with high affinity to a specific site within the hamster adenine phosphoribosyltransferase (APRT) gene was modified with the photochemically reactive psoralen derivative 4'-(hydroxymethyl)-4,5',8-trimethylpsoralen (HMT). The modified TFO, psorTFO1, bound with high affinity to a target site within intron 1 of the APRT gene. Upon irradiation, photomonoadducts (i.e., covalent adducts of psorTFO1 to one strand of the target duplex) were formed with high efficiency (approximately 50%). Introduction of 5'-TpA sequences (the preferred site for psoralen-induced photo-cross-links) at or near the triplex junction leads to increased efficiency of total photoadduct formation and to efficient formation of products that had the electrophoretic mobility on denaturing PAGE expected for three-stranded photo-cross-links (i.e., products containing psorTFO1 covalently linked to both strands of the duplex). Their identities as cross-links were verified by (1) identical electrophoretic mobility of products formed with either duplex strand radiolabeled and (2) coprecipitation of the radiolabeled duplex strand with its complementary biotinylated strand following denaturation. In addition, the cross-links were completely reversible upon irradiation at 254 nm, as expected for psoralen-mediated cross-links. The yield and distribution of photoadducts depended on 5'-TpA position. The most efficient photoadduct formation (approximately 90%) and photo-cross-link formation (approximately 90% of total photoadducts) were observed for a 5'-TpA adjacent to the triplex junction, with significant, but lower, cross-linking efficiency within three base pairs of the junction. Molecular models of the psoralen-conjugated triplex with its six-carbon linker suggested a simple explanation for this distance dependence: facile intercalation near the triplex/duplex junction, with increasing strain required for intercalation at more distant sites. These results indicate that psorTFO1 allows for DNA damage with high precision and high efficiency, and the likely proportion of monoadducts and cross-links can be estimated from the target sequence.

Interproton distance bounds from 2D NOE intensities: Effect of experimental noise and peak integration errors

The effect of experimental and integration errors on the calculation of interproton distances from NOE intensities is examined. It is shown that NOE intensity errors can have a large impact on the distances determined. When multiple spin ('spin diffusion') effects are significant, the calculated distances are often underestimated, even when using a complete relaxation matrix analysis. In this case, the bias of distances to smaller values is due to the random errors in the NOE intensities. We show here that accurate upper and lower bounds of the distances can be obtained if the intensity errors are properly accounted for in the complete relaxation matrix calculations, specifically the MARDIGRAS algorithm. The basic MARDIGRAS algorithm has been previously described [Borgias, B.A. and James, T.L. (1990) J. Magn. Reson., 87, 475-487]. It has been shown to provide reasonably good interproton distance bounds, but experimental errors can compromise the quality of the resulting restraints, especially for weak cross peaks. In a new approach introduced here, termed RANDMARDI (random error MARDIGRAS), errors due to random noise and integration errors are mimicked by the addition of random numbers from within a specified range to each input intensity. Interproton distances are then calculated for the modified intensity set using MARDIGRAS. The distribution of distances that define the upper and lower distance bounds is obtained by using N randomly modified intensity sets. RANDMARDI has been used in the solution structure determination of the interstrand cross-link (XL) formed between 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) and the DNA oligomer d(5'-GCGTACGC-3')2 [Spielmann, H.P. et al. (1995) Biochemistry, 34, 12937-12953]. RANDMARDI generates accurate distances bounds from the experimental NOESY cross-peak intensities for the fixed (known) interproton distances in XL. This provides an independent internal check for the ability of RANDMARDI to accurately fit the experimental data. The XL structure determined using RANDMARDI-generated restraints is in good agreement with other biophysical data that indicate that there is no bend introduced into the DNA by the cross-link. In contrast, isolated spin-pair approximation calculations give distance restraints that, when applied in a restrained molecular dynamics protocol, produce a bent structure.

Solution Structures of Psoralen Monoadducted and Cross-Linked DNA Oligomers by NMR Spectroscopy and Restrained Molecular Dynamics

We have used two-dimensional 1H NMR spectroscopy to determine the solution structures of the 4'-(hydroxymethyl)-4,5',8-trimethylpsoralen (HMT) furanside monoadducted (MAf) and the photoisomeric HMT interstrand cross-linked (XL) DNA oligonucleotide d(5'-GCGTACGC-3')2. The determination of the structure was based on total relaxation matrix analysis of the NOESY cross-peak intensities using the program MARDIGRAS. Improved procedures to consider the experimental "noise" in NOESY spectra during these calculations have been employed. The NOE-derived distance restraints were applied in restrained molecular dynamics calculations. Twenty final structures each were generated for both the MAf and XL from both A-form and B-form dsDNA starting structures. The root-mean-square (rms) deviations of the coordinates for the 40 structures for the MAf and XL were 1.12 and 1.10 A, respectively. The rmsd of the MAf with respect to the XL is 2.20 A. The local DNA structure is distorted in both adducts, with the helix unwound by 34 degrees and 25 degrees for the MAf and XL, respectively, and an overall helical repeat of 11 base pairs, caused by intercalation of the HMT. The MAf is a photochemical intermediate on the path to interstrand XL. Considerable local structural distortion is induced by both adducts, but the DNA returns to B-form structure within three base pairs of the damage site. There is no significant bend in the helix axis of either the MAf or the XL. We have evaluated the accuracy of the two major methods of converting NOESY data into interproton distances, the isolated spin-pair approximation (ISPA) and the complete relaxation rate matrix analysis (RMA). Both methods were evaluated by comparing the resulting calculated interproton distances generated to known covalently fixed distances in the HMT. The overall structures were evaluated by checking their agreement with biophysical evidence from non-NMR techniques. Only the modified RMA method gave correct interproton distances.

Transcription-coupled repair of psoralen cross-links but not monoadducts in Chinese hamster ovary cells.

We have examined the rate and extent of removal of 4'-(hydroxymethyl)-4,5',8-trimethylpsoralen (HMT) cross-linkable monoadducts and interstrand cross-links from restriction fragments within the amplicon containing the dihydrofolate reductase (DHFR) gene in the Chinese hamster ovary (CHO) cell line B11. The rate and extent of removal of HMT cross-links was significantly greater in an actively transcribed fragment than in a nontranscribed extragenic fragment of similar size. For the 5' half of the DHFR gene, approximately 80% of the HMT cross-links were removed in 8 h, in agreement with results reported by Vos and Wauthier [Vos, J. M., & Wauthier, E. L. (1991) Mol. Cell Biol. 11, 2245-2252, 1991]. However, few cross-links were removed in that period from the nontranscribed fragments, whose 5' end is approximately 7 kb downstream from the DHFR transcription unit and which includes a putative replication initiation site. Even after 24 h, only about 50% of the cross-links had been removed from this fragment. In contrast, both the rate and the extent of removal of cross-linkable HMT monoadducts were similar in the two fragments with 50% of the cross-linkable monoadducts removed in 24 h. Moreover, monoadducts formed in the bulk of the genome were removed in 24 h. Moreover, monoadducts formed in the bulk of the genome were removed at a slightly slower rate and to a lesser extent (30% in 24 hours) than those from either of these specific sequences.(ABSTRACT TRUNCATED AT 250 WORDS)

Analysis of mutations induced by 4′-hydroxymethyl-4,5′,8-trimethylpsoralen and UVA in Escherichia coli lac Z gene and its regulatory region

The mutagenic consequences of covalent adducts induced in M13mp19 DNA by 4′-hydroxymethyl-4,5′,8-trimethylpsoralen (HMT) and UVA have been determined in a forward mutational system capable of detecting all classes of mutagenic events. The photoreaction mediated by HMT has been carried out at a very low molar ratio of HMT to DNA which favours the induction of cross-links between high affinity reaction sites. When damaged M13mp19 DNA is used to transfect competent Escherichia coli K12 JM105 cells, a five-fold increase in mutation frequency is observed at 3.5% survivors when measured as a loss of β-galactosidase α-complementation. The enhanced mutation frequency is largely due to base substitutions, frameshift events and large deletions. The single nucleotide substitutions occur both in the lac Z coding sequence and in its regulatory region. Transversion and transition have been detected with a predominent form consisting of A.T-to-G.C transversion at position + 159. Frameshift mutations have been observed at five positions while three large deletions removing either part of the coding sequence or both the coding and the regulatory regions have been detected with a higher frequency. The spectrum of base substitutions detected between the M13 lac Z − phages surviving to the treatment is totally different from those appearing spontaneously whereas several frameshift events or deletions can already be detected between the spontaneous mutations. Despite the presence of these spontaneous hot spots, the spectrum of mutations recovered after HMT photoaddition appears to be unique and a detailed analysis of the different classes of mutations indicates an important role of cross-links in the production of mutations.

Recent advances in the synthesis and structure determination of site specifically psoralen-modified DNA oligonucleotides

We have developed novel methods for the preparation of multimicromole quantities of extremely pure, uniquely photoadducted psoralen—DNA cross-links, furan-side monoadducts and pyrone-side monoadducts. Psoralen cross-linked and furan-side monoadducted DNA were produced by employing high intensity argon ion and krypton ion lasers as light sources. Pyrone-side monoadducts were prepared by base-catalyzed photoreversal of psoralen cross-links. The various psoralen-adducted DNA oligomers were efficiently purified by high performance liquid chromatography. These methods have permitted us to synthesize 4 μmol each of a self-complementary 8-mer d(GCGTACGC) 4′-(hydroxymethyl)-4,5′,8-trimethylpsoralen (HMT) furan-side monoadduct and HMT cross-link. Preliminary nuclear magnetic resonance (NMR) data on the HMT cross-linked 8-mer d(GCGTACGC) have been obtained which confirmed the presence of the diadducted psoralen at the unique 5′TpA3′ site. NMR data obtained from the 8-mer furan-side monoadduct revealed that the psoralen molecule is intercalated into the DNA double helix. Preliminary crystals of 8-mer cross-linked DNA molecule have been grown. Conditions for the growth of X-ray diffraction-quality crystals and the further analysis of these crystals are now in progress.

Molecular analysis of mutations induced by 4′-hydroxymethyl-4,5′,8-trimethylpsoralen and UVA in the mouse HPRT gene

The effects of the reaction photosensitized by 4′-hydroxymethyl-4,5′,8-trimethylpsoralen (HMT) on a mouse lymphoma cell line have been examined. Using the hypoxanthine phosphoribosyltransferase (HPRT) locus as target gene, a mutagenic effect of the photoreaction can be detected concomitantly with a loss of cell viability. Isolation of HPRT deficient clones has permitted a molecular characterization of the mutational pattern induced by the photosensitization reaction mediated by HMT. Southern blotting analysis demonstrated that the HPRT deficiency could not be correlated with gene deletions larger than 300 bp. Using polymerase chain reaction on both DNA and cDNA, amplification products have been cloned into M13mp18 and sequenced. Base transversions targeted on thymine residues have been located in exon 2, 3, 8 and 9 together with spontaneous frameshift mutations occurring in a run of guanine residues in exon 3. HPRT deficiencies owing to mutations arising in the HPRT promoter region have also been observed. Dot and Northern blot analysis revealed that the photoreaction could lead to either a reduced level of gene transcription or to a complete absence of HPRT m-RNA. Using polymerase chain reaction (PCR) amplification and agarose gel electrophoresis, deletions in the HPRT promoter have been observed and correlated to deficient enzyme expression.

Both cross-links and monoadducts induced in DNA by psoralens can lead to sister chromatid exchange formation

The relative importance of DNA-DNA cross-links and bulky monoadducts in sister chromatid exchange (SCE) formation was investigated in three human fibroblast cell lines with different repair capabilities. These cell lines included normal cells, which can repair both classes of lesions; xeroderma pigmentosum (XP) cells, which cannot repair either psoralen-induced cross-links or monoadducts; and an XP revertant that repairs only cross-links and not monoadducts. SCEs were induced by two psoralen derivatives, 4′-hydroxymethyl-4,5′,8-trimethylpsoralen (HMT) and 5-methylisopsoralen (5-MIP). After activation with long-wave ultraviolet light, HMT produces cross-links and monoadducts in DNA, whereas 5-MIP produces only monoadducts. In normal human cells both psoralens induced SCEs, but if cells were allowed to repair for 18 h before bromodeoxyuridine (BrdUrd) was added for SCE analysis, the SCE frequency was significantly reduced. XP cells showed an SCE frequency that remained high regardless of whether SCEs were analyzed immediately after psoralen exposure or 18 h later. In the XP revertant that repairs only cross-links, both psoralens induced a high yield of SCEs when BrdUrd was added immediately after psoralen treatment. When XP revertant cells were allowed 18 h to repair before addition of BrdUrd, the SCEs induced by HMT were greatly reduced, whereas those induced by 5-MIP were only slightly reduced. These observations indicate that both cross-links and monoadducts are lesions in DNA that can lead to SCE formation.

Intercalation of psoralen into DNA of plastid chromosomes decreases late during barley chloroplast development

We have used a DNA crosslinking assay to measure intercalation of the psoralen derivative HMT (4'-hydroxymethyl-4,5',8-trimethylpsoralen) into barley (Hordeum vulgare) plastid chromosomal DNA during chloroplast and etioplast development. Intercalation into DNA in intact plastids in vivo and in plastid lysates in vitro shows that chromosomal DNA in the most mature chloroplasts intercalates HMT less efficiently than DNA in younger chloroplasts. In contrast, there is no change in HMT intercalation during etioplast differentiation in the dark. Our results also show that DNA in higher plant plastid chromosomes is under superhelical tension in vivo. The lower susceptibility to HMT intercalation of DNA in the most mature chloroplasts indicates that late during chloroplast development the superhelical tension or the binding of proteins to the DNA or both change.

Analysis of spontaneous and psoralen-induced Salmonella typhimurium hisG46 revertants by oligodeoxyribonucleotide colony hybridization: use of psoralens to cross-link probes to target sequences

An improved DNA colony-hybridization method for the rapid characterization of Salmonella typhimurium hisG46 revertants is described. Oligodeoxyribonucleotides (15-mers) complementary to each of 6 possible transition or transversion mutations and an extragenic suppressor mutation, underlying the His + phenotype, were prepared. Optimal sequence discrimination was achieved by hybridizing 15-mers at the apparent dissociation temperature ( T d) for 2 h with chromosomal DNA of revertant colonies affixed to Whatman 541 filters. Subsequent exposure of filters to UVA radiation (320–400 nm) in the presence of 4′-hydroxymethyl-4,5′,8-trimethylpsoralen (HMT) resulted in cross-linking of perfectly matched probes and target DNA sequences while sequences containing a single base-pair mismatch could be discriminated with a brief denaturing wash. No false negative results were obtained with the new procedure. An analysis of 204 spontaneous and 174 PUVA-induced TA100 revertants is presented.

Light affects the structure of Chlamydomonar Chloroplast chromosomes

We have analyzed changes in the structure of chloroplast chromosomes in response to light in growing Chlamydomonas cells using a crosslinking assay based on the intercalation of HMT (4'-hydroxymethyl-4,5',8-trimethylpsoralen) into DNA. Our results show that the structure of chloroplast chromosomes in at least three widely separated regions is different in light-grown vs. dark-grown cells. Structural changes in chloroplast chromosomes occur within 3 hrs after exposure to light or darkness, respectively. The response to light is not inhibited by atrazine and can be elicited by dim blue light incapable of evolving O2, indicating that it does not require photosynthesis. Inhibition of cytoplasmic protein synthesis with cycloheximide prevents this response to light, indicating that it depends, at least in part, on proteins imported from the cytoplasm.

DNA interstrand cross-links promote chromosomal integration of a selected gene in human cells.

We have used integrative pSV2 plasmids to learn how DNA lesions affect nonhomologous recombination with human chromosomes. Enhanced stable transformation of fibrosarcoma cells with a selectable gene was observed after chemical modification of the plasmid DNA; thus, cells transfected with plasmid pSV2-gpt carrying photoadducts of the cross-linking agent 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) yielded four- to sevenfold-higher levels of Gpt+ transformants than were obtained with untreated plasmid. The enhancement due to HMT interstrand cross-links was at least as great as that due to the monoadducts. DNA hybridization analysis indicated that the enhanced transformation frequency resulted from an increased number of cells carrying integrated plasmid sequences rather than from a higher copy number per transformant. The enhancement was not seen with a plasmid missing the sequences flanking the minimal simian virus 40 gpt transcription unit. Cotransfection with untreated and HMT-treated plasmids suggested that the HMT-containing DNA interacted preferentially with some cellular factor that promoted chromosomal integration of the plasmid DNA. It is concluded that (i) interstrand cross-linking as well as intrastrand DNA adducts promote nonhomologous recombination in human chromatin and (ii) DNA sequences flanking the selectable genes are the targets for such recombinational events.

DNA interstrand cross-links promote chromosomal integration of a selected gene in human cells.

We have used integrative pSV2 plasmids to learn how DNA lesions affect nonhomologous recombination with human chromosomes. Enhanced stable transformation of fibrosarcoma cells with a selectable gene was observed after chemical modification of the plasmid DNA; thus, cells transfected with plasmid pSV2-gpt carrying photoadducts of the cross-linking agent 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) yielded four- to sevenfold-higher levels of Gpt+ transformants than were obtained with untreated plasmid. The enhancement due to HMT interstrand cross-links was at least as great as that due to the monoadducts. DNA hybridization analysis indicated that the enhanced transformation frequency resulted from an increased number of cells carrying integrated plasmid sequences rather than from a higher copy number per transformant. The enhancement was not seen with a plasmid missing the sequences flanking the minimal simian virus 40 gpt transcription unit. Cotransfection with untreated and HMT-treated plasmids suggested that the HMT-containing DNA interacted preferentially with some cellular factor that promoted chromosomal integration of the plasmid DNA. It is concluded that (i) interstrand cross-linking as well as intrastrand DNA adducts promote nonhomologous recombination in human chromatin and (ii) DNA sequences flanking the selectable genes are the targets for such recombinational events.

In vitro Repair of Psoralen-DNA Cross-links by RecA, UvrABC, and the 5′-Exonuclease of DNA Polymerase I

Psoralens produce DNA interstrand cross-links which are thought to be repaired via a sequential excision and recombination mechanism in Escherichia coli. The first round of incision by UvrABC has been characterized: it results in 11-base oligonucleotide cross-linked to an intact DNA strand (Van Houten, B., Gamper, B., Holbrook, S.R., Hearst, J.E., and Sancar, A. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 8077-8081). In the present work, DNA substrates containing 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) cross-links in defined positions are constructed and used to analyze the other steps in repair. It is shown that RecA protein mediates strand transfer past an oligonucleotide cross-linked to a single-stranded DNA circle and that the resulting heteroduplex is a substrate for the UvrABC complex: it excises a double-stranded oligonucleotide which contains the HMT cross-link. It is also found that the first round of UvrABC incision does not lead directly to strand exchange but that an intervening step is needed. That step is carried out in vitro by the 5'-exonuclease activity of DNA polymerase I (pol I) which creates a single-stranded DNA region (a gap) at an incised cross-link such that RecA can initiate strand exchange. Studies using cross-linked oligonucleotides showed that the gap produced by pol I results from the inability of the polymerase to add nucleotides to a 3'-OH end two to three nucleotides away from the furan side of an HMT cross-link. Pol I can, however, extend a 3'-OH end next to the pyrone side of the cross-link. Since UvrABC incises predominantly the furan side of psoralen cross-links in duplex DNA, this discrepancy has important consequences for repair.

Phased psoralen cross-links do not bend the DNA double helix.

Although the chemical reaction of psoralens with nucleic acids is well understood, the structure of psoralen-DNA cross-linked products is still not clear. Model building studies base on the crystal structure of the psoralen-thymine monoadduct suggest that each cross-link bends the DNA double helix by 46.5 degrees [Pearlman, D. A., Holbrook, S. R., Pirkle, D. H., & Kim, S.-H. (1985) Science (Washington, D.C.) 227, 1304-1308]. On the other hand, Sinden and Hagerman [Sinden, R. R., & Hagerman, P. J. (1984) Biochemistry 23, 6299-6303] find that, in solution, psoralen cross-linked DNA is not bent. Here we use gel electrophoresis to test the validity of the current models. We have synthesized a series of DNA fragments (21-24 base pairs in length), each containing one unique T-A site for 4'-(hydroxymethyl)-4,5',8-trimethylpsoralen (HMT) cross-linking. Because of an estimated 28 degrees unwinding of the helix by HMT [Wiesehahn, G., & Hearst, J. E. (1978) Proc. Natl. Acad. Sci. U.S.A. 75, 2703-2707], one expects that the 22-bp cross-linked fragment will be repeated nearly in phase with the average helical screw when multimerized. In that sequence ligation will maximally amplify any deformation to the double helix. We find that the ligated multimers of cross-linked DNA migrate close to the multimers of non-cross-linked DNA on polyacrylamide gels. Our observations place an upper limit of 10 degrees on DNA bending induced by psoralen cross-linking and indicate unwinding by about 1 bp, as well as stiffening of the double helix. These properties are not unexpected for classical intercalators.