Mutagenesis by 8-Methoxypsoralen and 5-Methylangelicin Photoadducts in Mouse Fibroblasts: Mutations at Cross-Linkable Sites Indu

[CANCER RESEARCH 55, 1283-1288, March 15, 1995] Mutagenesis by 8-Methoxypsoralen and 5-Methylangelicin Photoadducts in Mouse Fibroblasts: Mutations at Cross-Linkable Sites Induced by Monoadducts as well as Cross-Links1

Edward J. GÃ¼nther,Toni M. Yeasky, Francis P. Gasparro, and Peter M. Glazer2

Departments of Therapeutic Radiology Â¡E.J. G., T. M. Y.. P. M. G.] and Dermatology Â¡F.P. G.I, Yale University School of Medicine, New Haven, Connecticut 06520-8040

ABSTRACT center PUVA trial, Stem et al. (4) showed a statistically significant increase in the incidence of squamous cell carcinoma in PUVA Psoralens are used clinically in the treatment of several skin diseases, patients. A comparative analysis of the incidence of squamous cell including psoriasis, vitÃligo,and cutaneous T cell lymphoma. However, psoralen treatment has been associated with an increased risk of squa- carcinoma in the U.S. and European trials was recently reported (5). nious cell carcinoma of the skin. To elucidate molecular events that may While in earlier comparisons there appeared to be differences between play a role in the psoralen-related carcinogenesis, we examined psoralen- the European and American experience with PUVA-induced inci induced mutagenesis in a mouse fibroblast cell line carrying a recoverable, dence of squamous cell carcinoma, the ongoing periodic reanalysis of chromosomally integrated A phage shuttle vector. Using the stipi- gene as each set of data with a longer follow-up period now indicates com a mutation reporter gene, we determined the spectrum of mutations parable findings. induced by photoactivation of 8-methoxypsoralen and of 5-methylangeli- It is likely that these cancers arise from the mutagenic psoralen cin. Both psoralens generated predominately I : Vto A: I and some T:A to photoadducts (and their lack of repair or misrepair) formed in kera- G:C transversions. Most of the mutations occurred at either 5' TpA or 5' tinocytes during photochemotherapy (6). Studies of lymphocytes ob ApT sites, both of which are conducive to interstrand cross-link forma tion. However, 5-methylangelicin produces only monoadducts, whereas tained from patients receiving PUVA for either psoriasis or vitÃligo S-methoxypsoralen generated 20% cross-links and 80% monoadducts have shown a significant increase in the number of 6-thioguanine- under the conditions of our experiments, as measured by direct HPLC resistant cells (7). These studies were performed on lymphocytes analysis of the DNA from the treated cells. Although most of the mutations obtained from patients who were receiving PUVA therapy, and thus occurred at potentially cross-linkable sites, these results implicate mono the lymphocytes were only exposed indirectly to significantly atten adducts, as well as cross-links, as critical premutagenic lesions in psor- uated UVA doses. It is likely that directly irradiated human lympho alen-treated mammalian cells. These findings may help in the identifica cytes (as in photopheresis) and human keratinocytes (as in PUVA) tion of carcinogenic changes induced by psoralen, and they may aid in the would reveal higher levels of mutations. improved design of psoralen-based treatment regimens in the future. In order to design treatment regimens with reduced risk of cancer induction, an understanding of psoralen mutagenesis is needed. Al INTRODUCTION though several studies on the mutagenicity of psoralens in mammalian Psoralens constitute a group of three-ringed heterocyclic furocou- cells have been performed, there have been deficiencies in most of marins that can intercalate into DNA and undergo photo-induced these studies (8-12). The distribution of psoralen photoadducts can cycloaddition with pyrimidines to generate either MA3 or interstrand vary greatly depending on the treatment regimen used, but many cross-links. Psoralen photochemotherapies have been used for the studies have lacked characterization of the extent of 8-MOP phoload- treatment of three diseases: psoriasis and vitÃligo(PUVA therapy; duct formation or the distribution of monoadducts and cross-links Refs. 1 and 2), and cutaneous T cell lymphoma (photophoresis; Ref. under the conditions used in vivo. The mutagenic data have been 3). For psoriasis, 8-MOP is ingested (or applied topically) and then interpreted in terms of presumed rather than measured levels of 1-2 h later the affected skin area is exposed to long wavelength UV monoadducts and cross-links, and inferences have been made based radiation (UVA). Clearing can be achieved over an 8-12-week period solely on the sites of mutations. Often, high doses of psoralen and with treatments three times a week. PUVA therapy can be gradually UVA (probably overdoses) have been used and thus may have led to tapered so that only maintenance therapy at monthly intervals (or less) molecular events that may not occur in human photochemotherapies. is required. In vitÃligo,thetreatment regimen is similar. Repigmenta- In some studies, adduci formation has been measured by indirect tion of the affected areas may require 200-300 treatments (2). In methods: renaturation kinetics (13), hydroxyapatite chromatography photopheresis, after the ingestion of the drug, the patient's blood is (14), and alkaline elution analysis (15). These methods are useful leukapheresed. Plasma and leukocytes are combined and then exposed indicators of DNA photomodification but do not provide information to UVA radiation prior to reinfusion. about specific photoadducts. In previous work, we developed sensi The clinical uses of psoralens and UVA, however, are not without tive HPLC techniques to detect the number and distribution of psor potential drawbacks. Dermatologists have long been concerned about alen photoadducts in the DNA of treated cells. We have now quan the potential risk of skin cancer induction by PUVA. In the most tified 8-MOP photoadducts in cells affected by PUVA treatment, recent follow-up study of the patients from the original U.S. multi- including lymphocytes (16), keratinocytes (6), and fibroblasts (this study). We have demonstrated a uniform susceptibility to adduci Received 10/14/94; accepted 1/19/95. formation and the correlation of a certain level of adduci formation The costs of publication of this article were defrayed in part by the payment of page (1-10 adducts per megabasepair) wilh the inhibilion of cell prolifer charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ation. In a recent study we showed that conditions which others had 1 This work was supported by grants from the Charles E. Culpeper Foundation, used to produce MA alone were also producing significant numbers of Leukemia Society of America, NIH (ES05775, P. M. G.), and Yale Skin Disease Research cross-links (17), calling into queslion conclusions aboul the relative Center (F. P. G., P. M. G.). We also acknowledge support from Therakos, Inc. (F. P. G.). 2 To whom requests for reprints should be addressed, at Department of Therapeutic mutagenicily of MA and cross-links. Radiology, Yale University School of Medicine, P.O. Box 208040, New Haven, CT In Ihis sludy, we combine a HPLC analysis of Ihe dislribution 06520-8040. 3 The abbreviations used are: MA, monoadducts; PUVA, 8-methoxypsoralen and UVA of psoralen photoadducts in the DNA of the trealed cells with a irradiation; 8-MOP, 8-methoxypsoralen; 5-MeA, 5-methylangelicin. determination of the frequency and spectra of psoralen-induced 1283

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 1995 American Association for Cancer Research. PSORALEN MUTAGENESIS mutations in a A phage shuttle vector, \supF, chromosomally inte RESULTS grated in mouse fibroblasts. To aid in the correlation of photoadducts A mouse L cell-derived line, LN12, containing approximately 100 with mutations, we used two different furocoumarins, 8-MOP and 5-MeA. 5-MeA was selected because it forms only MA and thus chromosomally integrated copies of a A phage shuttle vector, AsupF (Fig. 1) was previously constructed for the purpose of studying provides a method to analyze the effects of MA alone in our system (18-20). 8-MOP, which is in widespread clinical use, can produce mutagenesis in mammalian cells (22). The supF gene, an amber both MA and cross-links (21). We also examined the effects of suppressor tyrosine tRNA gene of E. coli, was used as a mutation reporter gene. The small size of the supF gene makes it amenable to repeated psoralen/UVA doses on the induced mutation frequencies DNA sequence analysis, and it is a well-studied reporter gene with and spectra. minimal bias in the detection of the different types of point mutations (25). In this A shuttle vector-based system, the AsupF DNA is rescued from within the mouse cell DNA by incubation of the mouse DNA in MATERIALS AND METHODS A in vitro packaging extracts which can identify, cut out, and package Vectors. The construction of the shuttle vector \supF, carrying the supF A DNA from within mouse DNA into viable phage particles for amber suppressor tyrosine tRNA of Escherichia coli as a mutation reporter growth and analysis in bacteria (22, 23, 26). Mutations in the supF gene, was described previously (22). gene that occur in the mouse cells are detected using standard pro- Cells. Mouse LN12 cells carrying approximately 100 copies of the \supF caryotic genetic techniques to detect mutants among the rescued shuttle vector DNA i n their genome were generated previously by transfection phage vectors. The mutants are identified by an altered phage plaque of mouse L cells with AsupF DNA (22). The LN12 cells were isolated by phenotype (22). Phage with functional supF genes yield blue plaques selection and screening for cells incorporating multiple copies of the A vector because they can suppress the amber mutation in the host lacZ gene, DNA into their genome (22). generating ÃŸ-galactosidase enzyme which can metabolize 5-bromo- Mutagenesis Protocol. Mouse LN12 cells were grown to approximately 4-chloro-3-indolyl-ÃŸ-D-galactopyranoside and turn it blue. Phage with 75% confluence in DMEM supplemented with 10% PCS (GIBCO-BRL, supF mutations produce colorless plaques because they fail to sup Gaithersburg, MD). The media was removed, and the cell monolayer was washed three times with PBS. Psoralen (either 8-MOP or 5-MeA) was diluted press the lacZ (amber) in the host bacteria. Since this assay requires from a 500 /XMstock solution 1:100 in PBS to yield a 5 JAMsolution which was the recovery of viable phage particles, large deletions greater than added to the cells. The cells were incubated in the presence of psoralen for 15 10,000 base pairs which reduce the vector size below the 38,000 base min at 37Â°Cwhile protected from light. After the incubation period, the cells pair minimum for A phage or which remove essential phage genes are were irradiated in PBS with a broad-band UVA light source centered at 365 nm not detected, and so the assay selects only for relatively small dele (Therakos, West Chester, PA). UVB radiation was filtered out using one- tions and point mutations. quarter inch of plate glass. Typical fluencies were 4â€”5mW/cm2; 15-30-s The LN12 cells were used to study mutagenesis induced by two exposures were required for a dose of 0.1 J/cm2. After irradiation, the psoralen- psoralens (8-MOP and 5-MeA) and UVA. The LN12 cells in loga containing solution was removed, the monolayer was washed twice with PBS, rithmic growth were treated with either 8-MOP or 5-MeA (at 5 JIM) and growth medium was added. The cells were allowed to grow for 1 week and UVA (0.1 J/cm2). Following the treatment, the cells were allowed before they were harvested for shuttle vector rescue and analysis. Multiple to recover and grow for 1 week before they were harvested for DNA doses of psoralen and UVA were also administered to cells in order to isolation. This period was chosen to allow sufficient time for repair accumulate mutations. After each dose the cells were allowed to recover for 1 and/or replication to process the DNA damage into mutations. High week prior to another incubation with the psoralen and the subsequent expo molecular weight DNA was prepared from the cells and was incu sure to UVA radiation. Shuttle Vector Rescue. High molecular weight DNA was prepared from bated in A in vitro packaging extracts in order to rescue the AsupF the treated cells 1 week following the irradiation (and from untreated cells to shuttle vectors for analysis. The frequency of supF gene mutations in analyze spontaneous mutations) by lysis in the presence of 10 mM Tris (pH 8), the rescued phage vectors for each treatment regimen is listed in Table 100 mM EDTA, 0.1% SDS, and 50 ug/ml proteinase K for 3 h at 37Â°C.The 1. For 8-MOP/UVA, the mutation frequency was 15.2 X 10~5 and for 5-MeA/UVA, 16.8 X 10~5. These mutation frequencies are 5-fold DNA was further purified by phenol extraction and ethanol precipitation as greater than the spontaneous frequency of 3.2 X 10~5. Also shown is described (22). Rescue of the phage vector DNA from within the mouse cell DNA into phage particles using A in vitro packaging extracts was as described the extent of cell killing induced by these two treatments (single dose (22), except that improved A packaging extracts were made from a restriction- only). As determined by colony-forming ability, 15% of the cells deficient lysogen (NM759 [E. coli K12 recA56 A(mcrA) el4Â°b(mrr-hsd- survived the 5-MeA/UVA treatment while only 1% survived the mcr)(\imm434 cits b2 redi DamlS Sam7)/\]>. This lysogen lacks methyl- 8-MOP/UVA treatment. Repetitive exposures, designed to enhance directed restriction activity that would otherwise degrade DNA methylated in the yield of mutations that were a result of the phototreatment, the mammalian pattern and reduce the yield of rescued phage (23). produced frequencies 10- and 8-fold over background for 8-MOP and Mutation Identification and Analysis. Detection of supF mutations in the rescued phage was accomplished by growth of the phage on a lawn of E. coli carrying an amber mutation in the lacZ gene in the presence of 5-bromo-4- chloro-3-indolyl-ÃŸ-D-galactopyranoside and isopropyl-ÃŸ-o-thiogalactopyrano- pBR322 amp & ori supF mmt-neo side as described previously (22), except that PG901 [E. coli C lacZ125(Am)] was used instead of SY204 [E. coli K12 hsdR lacZ125(A.m) frp(Am)] because of the increased yield of rescued phage using the restriction-deficient E. coli C cI857 cosR strain as a host (23). DNA sequences were obtained by direct sequencing of PCR-amplified supF gene fragments, as described (24). ! i HPLC Analysis of Photoproducts. The extent of 8-MOP photoadduct EcoRI Sail Sail formation was determined by using ['H]8-MOP (Amersham Corp., Arlington Fig. 1. Diagram of the AsupF shuttle vector. The vector carries the supF gene as a Heights, IL) to treat mouse cells (17). After the isolation of the DNA, the mutation reporter gene (22), along with the neomycin-resistance gene driven by the mouse extent of photobound [3H]8-MOP was determined by liquid scintillation anal metallothionein II promoter to allow for selection for stable integration into the mouse cell ysis. In addition, the distribution of 8-MOP photoadducts was determined by chromosomes. The A cohesive ends (cos), attachment site (att), and represser gene performing reversed-phase HPLC analysis of the [3H]8-MOP-modified DNA (cI857) are also indicated. Construction of the vector and establishment of mouse LN12 cells carrying multiple chromosomal copies of the vector DNA have been described as described previously (17). previously (22). 1284

Table 1 Cytotaxicity and mulagenicity of the psoralen treatment of mouse LNI2 cells However, in the case of 5-MeAAJVA, almost one half of the muta tions at T:A base pairs were T:A to G:C transversions, a higher plaques*29/892,700 Treatment"Untreated survival(%)100 fraction than in the case of 8-MOP/UVA. A similar distribution of angelicin-induced mutations was seen in the laci gene in E. coli (20). 8-MOP/UVA (IX) 1 114/751,200 15.232.4 The last two columns (Table 2) present the comparison of the types 8-MOP/UVA (3X) n.d.c 30/93,503 5-MeA/UVA (IX) 15 82/487,500 16.8 of base substitutions that are known to be detectable in the supF gene 5-MeAAJVA (3X)Cell n.d.cMutants/total 42/179,300Mutants/1053.223.4 when contained in either the AsupF vector (22, 24, 28) or in a " The cells were exposed to the indicated psoralen at a concentration of 5 /AMand SV40-based plasmid vector (25). These data indicate that there is irradiated with 0.1 Â¡/cm2of UVA as indicated. *The values represent the frequency of supF mutations within AsupF shuttle vectors minimal bias in the types of point mutations detectable using supF as rescued from the mouse LN12 cells, treated as indicated. a reporter gene. ' n.d., not determined.

5-MeA, respectively (Table 1). Such fractionated treatments are sim Table 2 Spectra of psoralen-induced mutations in mouse LN12 cells ilar to the repeated doses of photochemotherapy administered to mutations in patients with psoriasis, vitÃligo,and cutaneous T cell lymphoma (2). cells"MutationC:G Treatment of mouse supF*AsupFvector2111108101474Plasmidvector301331221713126 The molecular nature of the mutations detected in supF genes rescued from LN12 cells treated with the different psoralen regimens is shown in Table 2. Note that for 8-MOP/UVA (PUVA) treatments, -*T:AT:A whether single (Table 2, column 1) or repetitive (Table 2, column 3), -Â»C:GC:G there is a preponderance of transversions occurring at T:A base pairs, -A:TC:G -Â»G:CT:A mostly T:A to A:T. The several C:G to T:A transitions seen with -Â»A:TT:A 8-MOP/UVA likely represent a contribution from the background of -G:CDeletion<200Deletion>200Total8-MOP1232316634495-MeA733110900338-MOPX3206112200235-MeAX34311440017Untreated17271011029Detectable spontaneous mutations (Table 2, column 5), since the induced muta tion frequency in the case of a single dose of 8-MOP/UVA was only 5-fold. Theoretically, 20% of the mutations may be due to the back ground in this case. Nonetheless, the mutations over background " The types of mutations detected in supF genes rescued within the AsupF shuttle occur mostly at Ts as expected from the well-characterized photo vector from 8-MOP/UVA-treated, 5-MeA/UVA-treated, and untreated mouse LN12 cells are listed. 8-MOP X 3 and 5-MeA X 3 indicate multiple treatments. chemistry of 8-MOP (17, 21, 27). An even higher proportion of * Columns 7 and 8 present for comparison the types of base substitution mutations in mutations at T:A base pairs was seen with multiple 8-MOP/UVA the supF gene that are known to produce detectable phenolype changes, either in the AsupF vector (Refs. 22, 24, 28, this work, and unpublished data) or in a plasmid vector treatments, which is consistent with a reduction in the contribution of (25). These data are compiled from many different mutagenesis studies using supF- background mutations as the induced frequency increases. With containing vectors. Fewer mutations are listed for the AsupF vector because less work has 5-MeA/UVA, most of the mutations were, again, at T:A base pairs. been done with this vector and fewer mutations have been accumulated.

A. 8-MOP/UVA

T A TA AA AGC TGAT A GTA A 3' TAT AT ATA CC TTATGG A CCA C T A 5' TAAACTATACTACGCGGGG CATTTTCGTAATGGACACCACCCCAAGGGCTCGCCGGTTTCCCTCGTCTGAGATTTAGACGGCAGTAGCTGAAGCTTCCAAGCTTAGGAAGGGGGTGGTGGT 40 50 90 100 110 120 130 140 150 160 170 160 ATTTGATATGATGCGCCCC 5' ++ ++ + Promoter pre-tRNA (58-98) Suppressor tRNA (99-183) I AUS2-1781- I Â¿(106 or 107) I-- -A (86-120) I

B. 5-MeA/UVA

A T G CG TT C G TAT 3' TG AT TG GT TTC TGGATCTGAT 5' TAAACTATACTACGCGGGG CATTTTCGTAATGGACACCACCCCAAGGGCTCGCCGGTTTCCCTCGTCTGAGATTTAGACGGCAGTAGCTGAAGCTTCCAAGCTTAGGAAGGGGGTGGTGGT 40 50 90 100 110 120 130 140 150 160 170 180 ATTTGATATGATGCGCCCC 5'

C. Spontaneous

A TAT T AT A 3' G CTTT GAA AA CG TTATTTAA 5' TAAACTATACTACGCGGGG CATTTTCGTAATGGACACCACCCCAAGGGCTCGCCGGTTTCCCTCGTCTGAGATTTAGACGGCAGTAÅ“TGAAGCTTCCAAGCTTAGGAAGGGGGTGGTGGT 40 50 90 100 110 120 130 140 150 160 170 180

5' A 163 31

Fig. 2. Sequence analysis of supF mutations detected in phage vectors rescued from mouse fibroblasts. The 8-MOP/UVA-induced (A), 5-MeA/UVA-induced (B), and spontaneous mutations (C) are listed above each base pair, with the listed base representing the change from the sequence in the upper strand. The + signs below the sequence in A indicate sites at which mutations are known to produce a detectable phenotype change in supF (25), demonstrating that the use of supF in this assay does not bias detection of mutations at any particular site. Deletions are indicated below each sequence. 1285

We also observed several deletion mutations after the 8-MOP/UVA Table 4 DNA binding-properties of 8-MOP and 5-MeA" treatment (7/49 mutants analyzed) but none after 5-MeA/UVA. Of these, 4 were large deletions of between 1000 and 6000 base pairs *DNA(M'1)"e365nm encompassing the entire supF gene, as determined by restriction Â¿â€¢("DNA enzyme analysis of the mutant phage vector DNA (data not shown). addition8-MOP7709180.0135-MeA15603280.0074 " Data extracted from Ref. 29. The others were small deletions of 1, 26, and 34 base pairs. * K, DNA binding constant. The distribution of the mutations produced by the single-dose ' e, extinction coefficient for light absorption by the compound at 365 nm. 8-MOP/UVA and 5-MeA/UVA exposures are presented in Fig. 2, in ' <Â¿>,initialphotoreaction efficiency for addition of psoralen to DNA. comparison to those detected in the DNA from untreated cells (spon taneous mutations). Of the 42 base substitution mutations produced by 8-MOP/UVA, 8 occurred at 5' TpA sites (19%), 4 at 5' ApT sites, and part, this is because of the diversity of conditions under which the 7 at 5' TpT sites. For 5-MeA/UVA, a higher proportion (14/33, 42%) treatments were carried out and the lack of direct measurements of the were found at 5' TpA, whereas only 1 was seen at 5' ApT and 2 at 5' photoadducts. One study on 8-MOP mutagenesis of the supF gene in TpT. In the case of the spontaneous mutations, none were seen at an extrachromosomal, SV40-based shuttle vector concluded that either 5' TpA or 5' ApT sites, and only two were found at 5' TpT cross-links were more mutagenic than MA, based solely on the sites. Most of the spontaneous mutations involved C:G base pairs. occurrence of mutations at cross-linkable sites (8). This conclusion In conjunction with the mutagenesis experiments, the photoadducts was not warranted, however, because such sites are also conducive to produced in the mouse fibroblast DNA by the 8-MOP/UVA treatment monoadduct formation. In that study, no measurement of photoad were determined. The LN12 cells were treated with 8-MOP as in the ducts was performed. In a follow-up study, Sage and Bredberg (9) mutagenesis protocol, except that [3H]8-MOP was used to allow determined the photoadduct distribution in the supF gene after in vitro photoadduct detection and analysis. Immediately following the UVA treatment of the plasmid DNA, using a polymerase termination assay. irradiation, DNA was prepared from the cells and was analyzed both They found one mutation hot spot (at base pairs 44-47) to correlate for the formation and distribution of photoadducts. Table 3 summa with cross-link formation under their conditions. They used a split- rizes the yield of 8-MOP photoadducts in murine fibroblasts treated dose regimen in which isolated vector DNA was treated with a high under a variety of conditions. Under the conditions of the mutagenesis 8-MOP concentration (93 /XM)and consecutive UVA doses of 0.055 experiment (single dose, 5 JJ.M8-MOP plus 0.1 J/cm2 UVA), the yield and 3.3 J/cm2. In comparison, we treated cells, not naked DNA, with of photoadducts was: 2.2 adducts/megabasepairs; 20% cross-links, 5 /LLM8-MOP and 0.1 J/cm2 UVA. Extracellular and intracellular 6% 3,4-MA, 75% 4',5'-MA. Because radiolabeled 5-MeA was not PUVA treatments of DNA can lead to vastly different numbers of available, similar analyses could not be performed. However, on the psoralen DNA photoadducts with different photoadduct distributions basis of its known photochemical properties (Table 4; Ref. 29), the and hence have a strong effect on observed mutation frequencies and overall level of adduci formation with 5-MeA is expected to be spectra. This comparison highlights the need to perform experiments similar, except that only MA are formed. under conditions that at least approximate the clinical setting. We also analyzed the ability of the 8-MOP/UVA-treated cells to Recently the mutagenic effects of 8-MOP/UVA at the adenine remove the photoadducts. Parallel cultures of LN12 cells were treated phosphoribosyl transferase locus in Chinese hamster ovary cells have with 5 JAM8-MOP and 400 nm light to produce mostly MA (87%). been reported (10). On the basis of a "combined analysis" (photo- The cells were harvested for photoproduct analysis either immediately modification of isolated DNA in vitro for the analysis of the photoad or 48 h later. It was found that, by 48 h, 66% of the photoadducts were duct spectrum and phototreatment of cells with 8-MOP/UVA for the removed (81% of the cross-links; 75% of 3,4-MA, and 62% of mutation spectrum), it was concluded that cross-links are likely to be 4',5'-MA). the major psoralen-induced premutagenic lesion in mammalian cells. However, the 8-MOP/UVA treatment conditions used, 1080 ng/ml (5 JUM)and 0.3 J/cm2, respectively, would lead in vivo to mostly MA and DISCUSSION a minority of cross-links at strong sites of psoralen photoaddition (our Although multiple studies have been performed, the question of the estimate â€”20-30% cross-links, see Table 3 and Ref. 17). Mutations mutagenicity of specific 8-MOP photoadducts is far from clear. In were found at T:A base pairs: 16 T:A to A:T, 6 T:A to G:C, and 4 T:A to C:G. At about the same time, however, Papadopoulo et al. (12) Table 3 Photitadduct levels unti distribution in X-MOP"-treated munse LN12 cells reported the spectrum of mutations induced by 4,5',8-trimethylpsoralen in the HPRT gene in human cells. They again observed mostly dose2Radiation (%)3920137.25303,4-MA(%)96118.11404',5'-MA(%)52757684.133100T:A to A:T transversions at 5' TpA sites. However, from the site J/cm2UVA'0.1 specificity of the mutations, they inferred, in this case, that psoralen J/cm2UVA60 MA and not cross-links are the principal mutagenic lesion in human nm60min, 400 after48min, 400 nm cells. Hence, confusion may arise when inferences are made based h forrepair60 just on the distribution of psoralen-induced mutations. thenUVArfmin, 400 nm, J/cm2)5-MeA/UVA''Adducts/mbp8.44.42.91.02.72Cross-links(2 Yang et al. (11) reported the spectrum of mutations induced in the HPRT gene in human fibroblasts exposed to a split-dose protocol (30). " Cells were treated with 5 /J.M8-MOP (1080 ng/ml) plus the indicated UVA or visible In this work, most of the mutations analyzed were observed at light, unless otherwise noted. '' For UVA treatment, fluorescent lamps emitting broad-band UVA centered at 365 nm cross-linkable sites. A significant mutation induction, however, was were used, and the cells were treated as a monolayer in tissue culture dishes. For 400-nm also detected after the first dose of UVA in the split-dose protocol, in light irradiations, the cells were suspended in a cuvette and exposed to light emanating which a higher proportion of MA would be expected, although no from a monochromator (estimated effective dose ~ 1 J/cm"). 'Cells in this sample were treated with 0.46 Â¿Â¿M8-MOP(100 ng/ml) plus 2 J/cm2 measurement of adducts was performed. These mutations were not UVA. analyzed, but it was concluded that both MA and cross-links may be ''The second irradiation was carried out after washout of the psoralen. ' The total number of adducts was estimated using the known photochemical properties mutagenic. In our study, the mutations induced by both 8-MOP and 5-MeA of 5-MeA. In the only reported studies on the yield of specific angelicin photoadducts in DNA, we showed that another angelicin formed only 4',5'-MA (18). occurred mostly at 5' TpA and 5' ApT sites. A slightly higher 1286

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 1995 American Association for Cancer Research. PSORALEN MUTAGENESIS proportion of mutations was observed at 5' TpA sites with 5-MeA as thymidine permits Ihem to be accommodated wilhin Ihe DNA helix opposed to 8-MOP, but it is not clear if this is significant. 5' ApT (41), escaping repair. Although this might lead to a recognition sites, as well as 5' TpA sites, have been shown to be cross-linkable problem during replication, the "A-rule," which states thai "when in (21, 27, 31, 32). In addition, the mutation spectra produced by 8-MOP doubt insert an adenine (A)," may prevent some errors in replication and 5-MeA were similar, with both having a predominance of trans- at sites containing MA (42). The more pronounced structural changes versions at T:A base pairs. Because of its angled structure, 5-MeA induced by cross-links may lead to more efficient recognilion and cannot form cross-links (18-20), and, under the conditions used, the subsequenl repair. However, the faithful repair or replicative bypass 8-MOP treatment produced only 20% cross-links. Hence, our data of a specific adduci (monoadducl or cross-link) may be influenced by would implicate MA as well as cross-links as significant premuta- Ihe nalure and frequency of nearby photoadducts, by ils location genie lesions. As indicated in Table 4, 8-MOP and 5-MeA have within a specific DNA sequence, or by contacls with DNA-binding similar DNA-binding characteristics. On the basis of this, it appears proteins. In Ihis regard, Parris et al. (43) have demonslrated Ihe that the similar site specificity of the mutations induced by the two influence of DNA sequence conlext on UV mutagenesis in mamma psoralens is governed by the dark binding tendencies of these two lian cells. compounds (preference for 5' TpA and 5' ApT sites) and not neces The mechanism by which transversions are produced al Ihe sile of sarily by the formation of cross-links at those sites. The fact that Ihe psoralen adduci is not yet clear. Both Sage et a!. (10) and Yang et mutations occur at cross-linkable sites does not necessarily mean that al. (11) observed that the base 5' lo Ihe T in Ihe T:A base pair mulaled cross-links were responsible for them. This is not to say that cross by psoralen may determine the ultimate mutation. They detecled links are not mutagenic, but that MA as well as cross-links do play a mutation of the T in the T:A base pair to Ihe 5' base 63% and 81% of role in the observed mutagenesis. In contrast to the inferences made in the time, respeclively. One model to explain this type of mutagenesis previous studies, our results demonstrate that while psoralen-induced involves helix destabilizalion, dislocalion, synthesis using the 5' mutations tend to occur at cross-linkable sites, such mutations may flanking base as a template, and realignment (44). However, we did arise from MA (as produced by 5-MeA) as well as from cross-links. not see such a dramatic effect, with the T mutating to the 5' base in However, cross-links are likely more toxic, since there was greater only 40% of the 25 T:A base pair mutations induced by 8-MOP and cell killing but a similar induced mutation frequency with the 8-MOP/ in only 31% of the 22 mutalions at T:A base pairs induced by 5-MeA. UVA treatment in comparison to 5-MeA/UVA. The more toxic cross Another mechanism may simply be translesion synthesis, with pref links may also give rise to the deletions that were observed only in the erential incorporation of a pyrimidine opposite the psoralen-modified 8-MOP/UVA-treated cells, perhaps via strand breaks. Errors in dou thymidine. What determines the incorporation of a T or a C (leading ble-strand break repair or in strand break-initiated recombinational to T:A to A:T or T:A to G:C Iransversions, respeclively) remains to be repair between the tandemly repeated copies of the vector DNA in the elucidated. However, we did see a higher proportion of T:A to G:C mouse cell chromosome may lead to the observed deletions. Recom transversions in the case of 5-MeA. binational repair has been demonstrated as a pathway for psoralen Most of the psoralen photoadducts generated in our experiments cross-link repair in E. coli (33-35), and such a pathway may be active (>95%) occurred at Ts. However, less common reactions can occur, in mammalian cells when multiple copies of a gene are present, as in leading to other types of DNA damage. Psoralen reactions al Cs, this case. Errors in resolving recombinational intermediates could accounling for â€”5%of overall photoadducl formation, have not been conceivably lead to deletions. characterized as fully. Although, in Ihe case of Ihe single dose The pathways involved in psoralen adduci repair have not been 8-MOP/UVA, we attribute most of the mutalions at C:G base pairs to completely worked out. In vitro repair studies (33-36) suggest that a Ihe contribution of the background spontaneous mulalions, photoad cross-link may be subject to excision repair on one strand, creating a ducts at C may lead to some of the observed mutalions after 8-MOP/ gap in that strand and leaving a lesion on the other strand consisting UVA Irealment. Furthermore, in addition to photoadduct formalion, of a psoralen cross-link tethered to the oligonucleotide patch excised psoralens can also induce slrand breaks via singlel oxygen formation from the first strand. The choice of strand may depend on the orien (45, 46). This process may also lead to oxidative degradation of dG, tation and sequence context of the psoralen cross-link and on iran- leading to 8-oxo-guanine and other mutagenic products which we scriptional activity at the locus (10, 11, 37-39). The gapped substrate cannol measure using [3H]8-MOP. Allhough lesions at dG sites have may serve as an intermediate in recombinational repair or as a tem not been previously detected by chemical methods for either 8-MOP plate for repair synthesis. The gap-filling repair synthesis would or 5-MeA, mutational analyses may be more sensilive. In addilion, in require translesion bypass of the tethered psoralen-oligonucleotide altempting to correlate psoralen-induced lesions with mutations, it still attached to the template strand. Theoretically, bypass of the should be kept in mind lhal Papadopoulo et ai. (41) demonslrated thai residual psoralen adduci within Ihe gap would be similar lo Ihe bypass byslander adducts (e.g., 3,4-MA or even thymine dimers formed via synthesis required to replicate DNA conlaining simply a psoralen sensilizalion) may affecl Ihe repair of olher adducls. monoadducl. This model would predict thai psoralen MA and cross In Ihe sludies of bolh Sage et al. (10) and Yang et al. (Il), an links would lead lo similar speclra of mulalions, wilh some possible asymmelry in Ihe mulalion spectrum was observed with respect to the variation due to the properties of Ihe polymerases involved in gap transcribed and nontranscribed slrands, wilh the mutaled Ts found filling versus replicalion (40). Since Ihe sites of psoralen adduci mostly on the nonlranscribed slrand. This is consistent with strand- formalion are in Ihe firsl place governed by the intercalating tenden specific repair due to transcription/repair coupling (38, 39). We did cies of Ihe molecule, the favored sites for monoadduct and cross-link not observe the same phenomenon, since the supF gene is a bacterial formation are similar. The similarity we observed in the speclra of gene that is noi expressed in mammalian cells. mulalions produced by 5-MeA (which produces MA exclusively) and We have carried oui a comparison of mutagenesis by 8-MOP and by 8-MOP (which produces bolh MA and cross-links) supports this 5-MeA in mouse fibroblasts, in conjunction with an analysis of the model. photoadducts produced in the DNA of the 8-MOP-treated mouse cells. The relative mutagenicily of psoralen MA and cross-links, none- Based on the similarity in the mutalion speclra induced by 8-MOP and iheless, remains a complex issue. Several factors may play a role. The 5-MeA, we conclude that both MA and cross-links are mutagenic. The location of the vast majorily of 8-MOP pholoadducls al thymidine site specificity of the mutagenesis is dependent on the dark binding properties of the psoralens, with cross-linkable sites such as 5' TpA may mean thai most MA are toleraled because Ihe slructure of MA at 1287

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 1995 American Association for Cancer Research. PSORALEN MUTAGENESIS and 5 ' ApT sites favored, but with the mutations arising from MA as 18. MiÃ³lo,G., Stefanidis, M., Santella, R. M., Dall'Acqua, F., and Gasparro, F. P. 6,4,4'- well as cross-links at these sites. Deletions were observed only in the Trimethylangelicin photoadduct formation in DNA: production and characterization of a specific monoclonal antibody. J. Photochem. Photobiol. B, 3: 101-112, 1989. 8-MOP/UVA treatment, and these may have arisen from the cross 19. MiÃ³lo, G., Dall'Acqua, F., Moustacchi, E., and Sage, E. Monofunctional angular links generated by this treatment but not by the 5-MeA treatment. furocoumarins: sequence specificity in DNA photobinding of 6,4,4'-trimethylangelicin and other angelicins. Photochem. Photobiol., 50: 75-84, 1989. The exact molecular events involved in the process by which 20. Miller, S. S., and Eisenstadt, E. Suppressible base substitution mutations induced by PUVA transforms normal cells into squamous cell carcinomas have angelicin (isopsoralen) in the Escherichia coli laci gene: implications for mechanism yet to be elucidated. In this process, several factors may play a role. of SOS mutagenesis. J. Bacterio!., 169: 2724-2729, 1987. 21. Cimino, G. D., Camper, H. B., Isaacs, S. T., and Hearst, J. E. Psoralens as photoactive These include the wavelength of activating light, the gene in which the probes of nucleic acid structure and function: organic chemistry, photochemistry, and 8-MOP-absorbing molecule is located, the DNA photoadduct that is biochemistry. Annu. Rev. Biochem., 54: 1151-1193, 1985. 22. Glazer, P. M., Sarkar, S. M., and Summers, W. C. Detection and analysis of formed, the types of mutations produced, and the relative contribu UV-induced mutations in mammalian cell DNA using a lambda phage shuttle vector. tions of mutagenesis and immune suppression (well characterized for Proc. Nati. Acad. Sci. USA, 83: 1041-1044, 1986. PUVA [47]). The evaluation of the risks associated with psoralen 23. GÃ¼nther,E. J., Murray, N. E., and Glazer, P. M. High efficiency, restriction-deficient in vitro packaging extracts for bacteriophage lambda DNA using a new E. coli photochemotherapies will be assisted by a better understanding of the lysogen. Nucleic Acids Res., 21: 3903-3094, 1993. mutagenicity of specific photoadducts. At the same time, an under 24. Havre, P. A., GÃ¼nther,E. J., Gasparro, F. P., and Glazer, P. M. Targeted mutagenesis standing of the mechanism underlying the efficacy of psoralen pho- of DNA using triple helix-forming oligonucleotides linked to psoralen. Proc. Nati. Acad. Sci. USA, 90: 7879-7883, 1993. tochemotherapy may allow improvements in the risk:benefit ratio of 25. Kraemer, K. H., and Seidman, M. M. Use of supF, an Escherichia coli tyrosine such therapy. For example, if psoralen works simply via its antipro- suppressor tRNA gene, as a mutagenic target in shuttle vector plasmids. MutÃ¢t.Res., liferative effects, then a regimen that induces the requisite damage 220: 61-72, 1989. 26. Summers, W. C., Glazer, P. M., and Malkevich, D. Lambda phage shuttle vectors for while minimizing mutagenicity would be desirable. Since the distri analysis of mutations in mammalian cells in culture and in transgenic mice. MutÃ¢t. bution of photoadducts can be influenced by different dosing sched Res., 220: 263-268, 1989. ules of 8-MOP and UVA (i.e., U.S. versus European), the data 27. Boyer, V., Moustacchi, E., and Sage, E. Sequence specificity in photoreaction of various psoralen derivatives with DNA: role in biological activity. Biochemistry, 27: presented here may help in the design of effective treatment regimens 3011-3018, 1988. with decreased risk. 28. Havre, P. A., and Glazer, P. M. Targeted mutagenesis of simian virus 40 DNA mediated by a triple helix-forming oligonucleotide. J. Virol., 67: 7324-7331, 1993. 29. Gasparro, F. P. Psoralen photochemistry. In: R. G. Landes (ed.), Extracorporeal Photochemotherapy: Clinical Aspects and the Molecular Basis of Efficacy, Chap. 2, REFERENCES pp. 13-36. Austin, TX: R. G. Landes, 1994. 1. Parrish, J. A., Fitzpatrick, T. B., Tannenbaum, L., and Pathak, M. A. Photochemo- 30. Ben-Hur, E., and Elkind, M. M. Psoralen plus near-ultraviolet inactivation of cultured therapy of psoriasis with oral methoxsalen and long-wave ultraviolet light. N. Engl. Chinese hamster cells and its relation to DNA crosslinks. MutÃ¢t.Res.,18: 315-325,1973. J. Med., 291: 1207-1211, 1974. 31. Gasparro, F. P., Havre, P. A., Olack, G. A., GÃ¼nther,E. J., and Glazer, P. M. Site 2. Honigsmann, H., Fitzpatrick, T. B., Pathak, M. A., and Wolff, K. Oral photochemo- specific targeting of psoralen adducts with a triple helix-forming oligonucleotide: therapy with psoralen and UVA (PUVA): principles and practice. In: T. B. Fitz- characterization of psoralen monoadduct and crosslink formation. Nucleic Acids Res., patrick, A. Z. Eisen, K. Wolf, I. M. Freedberg, and K. F. Austen (eds), Dermatology 22: 2845-2852, 1994. in General Medicine, pp. 1728-1754. New York: McGraw-Hill, 1987. 32. Ramaswamy, M. and Yeung, A. T. The reactivity of 4,5',8-trimethylpsoralen with 3. Edelson, R. L., Berger, C. L., Gasparro, F. P., and Members of the CTCL Study oligonucleotides containing AT sites. Biochemistry, 33: 5411-5413, 1994. Group. Treatment of cutaneous T cell lymphoma by extracorporeal photochemo- 33. Sladek, F. M., Munn, M. M., Rupp, W. D., and Howard-Flanders, P. In vilro repair therapy. N. Engl. J. Med., 316: 297-303, 1987. of psoralen-DNA cross-links by RecA, UvrABC, and the 5'-exonuclease of DNA 4. Stern, R. S., Lange, R., and Members of the Photochemotherapy Follow-up Study. polymerase I. J. Biol. Chem., 264: 6755-6765, 1989. Non-melanoma skin cancer occurring in patients treated with PUVA five to ten years 34. Sladek, F. M., Melian, A., and Howard-Flanders, P. Incision by UvrABC excinucle- after first treatment. J. Invest. Dermatol., 91: 120-124, 1988. ase is a step in the path to mutagenesis by psoralen crosslinks in Escherichia coli. 5. Studniberg, H. M., and Weller, P. PUVA, UVB, psoriasis and nonmelanoma cancer. Proc. Nati. Acad. Sci. USA, 86: 3982-3986, 1989. J. Am. Acad. Dermatol., 29: 1013-1022, 1993. 35. Cheng, S., Van Houten, B., Camper, H. B., Sanear, A., and Hearst, J. E. Use of 6. Tokura, Y., Edelson, R. L., and Gasparro, F. P. Formation and removal of 8-MOP- psoralen-modified oligonucleotides to trap three-stranded RecA-DNA complexes and DNA photoadducts in keralinocytes. Effects of calcium and retinoids. J. Invest. repair of these cross-linked complexes by ABC excinuclease. J. Biol. Chem., 263: Dermatol., 96: 942-949, 1991. 15110-15117, 1988. 7. Strauss, G. H., Alberimi, R. J., Krusinski, P. A., and Baughman, R. D. 6-Thioguanine 36. Reardon, J. T., Spielmann, P., Huang, J. C., Sastry, S., Sanear, A., and Hearst, J. E. resistant peripheral blood lymphocytes in humans following psoralen, long wave Removal of psoralen monoadducts and crosslinks by human cell free extracts. Nucleic ultraviolet light (PUVA) therapy. J. Invest. Dermatol., 73: 211-216, 1979. Acids Res., 19: 4623-4629, 1991. 8. Bredberg, A., and Nachmansson, N. Psoralen adducts in a shuttle vector plasmid 37. Jones, B. K., and Yeung, A. T. DNA base composition determines the specificity propagated in primate cells: high mutagenicity of DNA crosslinks. Carcinogenesis, 8: of UvrABC endonuclease incision of a psoralen cross-link. J. Biol. Chem., 265: 1923-1927, 1987. 3489-3496, 1990. 9. Sage, E., and Bredberg, A. Damage distribution and mutation spectrum: the case of 38. Selby, C. P., and Sanear, A. Gene- and strand-specific repair in vitro: partial 8-methoxypsoralen and UVA in mammalian cells. MutÃ¢t.Res., 263: 217-222, 1991. purification of a transcription-repair coupling factor. Proc. Nati. Acad. Sci. USA, 88: 10. Sage, E., Drobetsky, E. A., and Moustacchi, E. 8-Methoxypsoralen induced mutations 8232-8236, 1991. are highly targeted at crosslinkable sites of photoaddition on the non-transcribed 39. Hanawalt, P. C. Heterogeneity of DNA repair at the gene level. MutÃ¢t.Res., 247: strand of a mammalian chromosomal gene. EMBO J., 12: 397-402, 1993. 203-211, 1991. 11. Yang, S-C., Lin, J-G., Chiou, C-C, Chen, L-Y., and Yang, J-L. Mutation specificity 40. Shivji, M. K. K., Kenny, M. K., and Wood, R. D. Proliferating cell nuclear antigen of 8-methoxypsoralen plus two doses of UVA irradiation in the hprt gene in diploid is required for DNA excision repair. Cell, 69: 367-374, 1992. human fibroblasts. Carcinogenesis, IS: 201-207, 1994. 41. Papadopoulo, D., Averbeck, D., and Moustacchi, E. High levels of 4,5',8-trimeth 12. Papadopoulo, D., Laquerbe, A., Guillouf, C., and Moustacchi, E. Molecular spectrum ylpsoralen photoinduced furan-side monoadducts can block cross-link removal in of mutations induced at the HPRT locus by a cross-linking agent in human cell lines normal human cells. Photochem. Photobiol., 47: 321-326, 1988. with different repair capacities. MutÃ¢t.Res., 294: 167-177, 1993. 42. Sahger, D., and Strauss, B. Insertion of nucleotides opposite apurinic/apyrimidinic 13. Gasparro, F. P., Saffian, W. A., Cantor, C. R., and Edelson, R. L. Wavelength dependence sites in deoxyribonucleic acid during in vilro synthesis: uniqueness of adenine for AMT crosslinking of pBR322 DNA. Photochem. Photobiol., 40: 215-219, 1984. nucleotides. Biochemistry, 22: 4518-4526, 1983. 14. AhnstrÃ¶m,G., and Erixon, K. Measurement of strand breaks by alkaline denaturation 43. Parris, C. N., Levy, D. D., lessee, J., and Seidman, M. M. Proximal and distal effects and hydroxyapatite chromatography. In: E. C. Friedberg and P. C. Hanawalt (eds.), of sequence context on ultraviolet mutational hot spots in a shuttle vector replicated DNA Repair, Vol. 1, Part B, pp. 403-418. New York: Marcel Dekker, Inc., 1981. in xeroderma cells. J. Mol. Biol., 236: 491-502, 1994. 15. Kohn, K. W., Ewig, R. A., Erickson, L. C., and Zwelling, L. A. Measurements of 44. Kunkel, T. A. Misalignment-mediated DNA synthesis errors. Biochemistry, 29: strand breaks and cross-links by alkaline elution. In: E. C. Friedberg and P. C. 8003-8011, 1990. Hanawalt (eds.), DNA Repair. Vol. 1, Part B, pp. 379-401. New York: Marcel 45. Sage, E., LeDoan, T., Boyer, V., Heiland, D. E., Kittler, L., HÃ©lÃ¨ne,C, and Dekker, Inc., 1981. Moustachhi, E. Oxidative DNA damage photo-induced by 3-carbethoxypsoralen and 16. Gasparro, F. P., Bevilacqua, P. M., Goldminz, D., and Edelson, R. L. Repair of other furocoumarins. J. Mol. Biol., 209: 297-314, 1989. 8-MOP photoadducts in human lymphocyte. In: B. M. Sutherland and A. D. Wood- 46. Oroskar, A. A., Gasparro, F. P., and Peak, M. J. Relaxation of supercoiled DNA by head (eds.), DNA Damage and Repair in Human Tissues, pp. 137-148. New York: aminomethyltrimethylpsoralen and UV photons: action spectrum. Photochem. Pho Plenum Press, 1990. tobiol., 57: 648-654, 1993. 17. Olack, G. A., Gattolin, P., and Gasparro, F. P. Improved HPLC analysis of 8-methoxyp- 47. Kripke, M. L., Morison, W. L., and Parrish, J. A. Systemic suppression of contact .soralen monoadducts and crosslinks in polynucleotide, DNA and cellular systems: anal hypersensitivity in mice by psoralen plus UVA radiation (PUVA). J. Invest. Derma ysis of split-dose protocols. Photochem. Photobiol., 57: 941-949, 1993. tol., 87: 87-92, 1983. 1288

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 1995 American Association for Cancer Research. Mutagenesis by 8-Methoxypsoralen and 5-Methylangelicin Photoadducts in Mouse Fibroblasts: Mutations at Cross-Linkable Sites Induced by Monoadducts as well as Cross-Links

Edward J. Gunther, Toni M. Yeasky, Francis P. Gasparro, et al.

Cancer Res 1995;55:1283-1288.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/55/6/1283

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/55/6/1283. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.