J. Photo&em.

43

Photobiol. B: Biol., 9 (1991) 43-59

Photochemical and photobiological 4,8-dimethyl-5’-acetylpsoralen

properties

of

E. Sage and L. Trabalzin? Insttiut Curie, Biology, Rue D’Ulm 26, 76231,Paris Ced,ex 05 (France)

A. Capozzi,

M. T. Concoui

Department of Pharmaceutical 35131 Padova (Italy]

and G. Pastoriui

Sciences of Padua University, Via Marzolo 5,

M. Tamaro Institute of Microbiology

of lkieste University, Via Fleming 22, 34127 Trieste htaly)

F. Bordintt Department of Pharmaceutical 35131 Padova (Italy]

Sciences of Padua University, Via Marzolo 5,

(Received June 18, 1990; accepted September 13, 1990)

Ke;yworck. Acetylpsoralen, singlet oxygen, DNA cross-linking, mutagenicity, photochemotherapy. Abstract The photochemical and photobiological properties of 4,8-dimethyl-5’-acetylpsoralen (AcPso), proposed for the photochemotherapy of some skin diseases, were investigated. The photoreaction of AcPso with DNA is weaker in the presence of air than in a nitrogen atmosphere, in terms of total photobinding and DNA cross-linking; when WA irradiation is performed in air, AcPso behaves as a monofunctional reagent. The quenching effect of oxygen is related to the high capacity of AcPso to produce singlet oxygen. Furthermore, it is demonstrated that AcPso photoadducts are better producers of singlet oxygen than free AcPso in solution. Using DNA sequencing methodology, two modes of DNA photosensitization by AcPso are shown, these lead to the formation of photoadducts mainly at T residues (and at C to a lesser extent) and to photo-oxidized G residues probably via singlet oxygen. Chemical or enzymatic cleavage were used as probes in these experiments. A rapid assay for the detection of the photodynamic effect of a photosensitizer on DNA, involving oxygen, is also described. Finally, the cytotoxicity and genotoxicity of AcPso on E. coli WP2 cells appear to be related to its ability to form photoadducts, in particular cross-links, rather than to its capacity to produce singlet oxygen.

1. Introduction

Psoralens are active photosensitizing drugs widely used in fundamental biochemical research and employed in photomedicine in the cure of various +Alsoat The Institute of Biological Chemistry of Siena University, Pian dei Mantellini 44, 53100 Siena, Italy. ++Authorto whom correspondence should be addressed. loll-1344/91/$3.50

0 Elsevier Sequoia/Printedin The Netherlands

44

skin diseases characterized by hyperproliferative conditions (for reviews, see refs. l-4). Their mechanism of action has been mainly connected with their ability to interact with DNA. After intercalation in the dark into duplex DNA, the furocoumarin reacts, on UVA irradiation, with the pyrimidine bases of DNA to form both monoadducts and biadducts [Z]. Interstrand cross-links issued from biadducts are generally considered as being mainly responsible for lethal and genotoxic effects induced by furocoumarin sensitization [5-91. At present, the compound currently used in photochemotherapy is %methoxypsoralen (g-MOP), a bifunctional derivative. However, some side effects (e.g. skin phototoxicity, long-term hazard in terms of possible induction of skin cancer) are observed [ 10-131. Therefore new monofunctional furocoumarins have been chemically synthesized and studied: (i) angular angelicin derivatives [ 14, 151 which are monofunctional for geometrical reasons; (ii) psoralen derivatives with one of the two reactive sites of their molecule blocked by a suitable substituent [ 11, 16). Recently, we have studied the possibility of obtaining monofunctional furocoumarins by inserting an acetyl group into the 3 or 5’ positions to block the reactive site at the pyrone or furan ring, respectively [ 171. All these compounds show poor reactivity with DNA, and one, 4,8-dimethyl-5’acetylpsoralen (AcPso), exhibits a very marked capacity for producing singlet oxygen (one order of magnitude higher than that of &MOP). Because this is a feature regarded by some workers as being responsible for certain toxic implications in furocoumarin sensitization, such as skin erythema [ 18, 191 and mutagenic activity [ 20, 211 it was of interest to study the photochemical and photobiological properties of this compound. In particular, we studied the effect of the presence of oxygen on the photoreaction with DNA and the capacity of the compound and its DNA adducts to produce singlet oxygen. We also investigated, at the sequence level, the two modes of DNA photosensitization by AcPso, which lead to the formation of photoadducts mainly at T residues and the photo-oxidation of G residues, probably via singlet oxygen. Finally, we compared these data with the results of some genotoxicity experiments carried out in E. coli WP2 cells.

2. Materials

and methods

2.1. Chemicals

The &MOP was purchased from Chinoin SPA, Milan0 (Italy). 4,8-Dimethyl5’acetylpsoralen (AcPso) was prepared as described in ref. 17. The extinction coefficients of these compounds at 365 run, the wavelength used in the irradiation, were 950 1 M-l cm-’ for 8-MOP and 8450 1 M-’ cm-’ for AcPso. Tritium-labelled thymidine (specific activity, 2.59 TBq mM_‘) was obtained from Amersham International Ltd. (U.K.). Radioactive furocoumarins (from Amersham) were purified by thin layer chromatography (TLC) on silica gel plates (Merck, cat. no. 5717), developed with chloroform and then ethyl

45

acetate-cyclohexane. Their specific activity, after appropriate dilution with cold compounds, was in the range 0.1-0.3 GBq M- ‘. For the radiochemical determination of photobinding and cross-linking, calf thymus DNA (from Sigma Chemical Co., St. Louis, MO, U.S.A.; cat. D 1501) was used. Its hypochromicity, determined according to ref. 22, was higher than 35%. For the E. coli exonuclease III assay, supercoiled replicative form (RF) DNA (form I) from phage Ml3 mp8 lac1935, served as a substrate for the photoreaction. It was obtained from cloning the first 935 KB of the lad gene (Hint II digestion of the plasmid pMC I) and the Hind III site of the prolinker of Ml3 mp8 phage. For sequencing experiments, 5’-end-labelled DNA fragment (100 base pairs (bp)) resulted from the digestion of Ml 3 mp8 lac1225-3 DNA by restriction enzymes EcoRI and Pvu II [23]. Restriction enzymes were obtained from Boehringer (Mannheim, F.R.G.); T4 DNA polymerase and exonuclease III of E. coli were obtained from Bethesda Research Laboratory (U.S.A.). Hydroxylapatite (Bio-Gel, HTP, cat. 130-0420) was obtained from Bio-Rad Laboratories, CA (U.S.A.). 2.2. W light sources All irradiations were performed using Philips HPW 125 W lamps. The emission spectrum was in the range 320-400 nm, with a maximum (over 90% of the total) at 365 nm. The irradiation intensity, determined by a potassium ferrioxalate chemical actinometer [24] and routinely checked by a UVX radiometer (Ultraviolet Products Co., Cambridge, U.K.), was 9.34 J S-l rnp2 for the experiments with calf thymus DNA, for singlet oxygen determinations and for the photobiological studies, and 20 J s-’ me2 for the enzymatic photoadduct detection. 2.3. Singlet oxygen pr6ducticm The measurements were performed according to the method of Kraljic and Mohsni [25]. Aqueous solutions, containing phosphate buffer (0.02 M, pH 7.3), p-nitroso-dimethylaniline (RNO) (2.2 x 10m5 M), imidazole (4 x low3 M) and the furocoumarin to be studied (at a concentration of 10V5 M), were irradiated for increasing times in open quartz cuvettes. Photoadducts containing DNA were obtained by UVA irradiation of DNA and furocoumarin solutions, prepared as described in the following section; DNA was precipitated and redissolved in the initial volume of phosphate buffer. RN0 and imidazole were added and the solutions were processed for singlet oxygen production as described above. RN0 bleaching was determined by reading the optical density (OD) at 440 run. 2.4. Photoreaction.

with caEf thymus

DNA

in the presence

of air

or

nitrogen

Aqueous solutions of calf thymus DNA (20 OD units) in saline citrate (10e2 M NaCI, 2 X 10m3 M sodium citrate) containing the compound to be tested (2 x 10e5 M) were irradiated for increasing times in calibrated testtubes immersed in a thermostat at 20 “C. Gas (air or nitrogen) was bubbled

46

through the solution for 30 min in the dark before irradiation and during the entire step. The reaction mixtures were then made to 1 M NaCl and the DNA was precipitated with an equal volume of ethanol and washed twice with 50% ethanol to eliminate the unbound material. DNA was redissolved in the initial volume of water or in phosphate buffer. 2.4.1. Photobinding The radioactivity bound to DNA was determined by counting 1 ml aliquots mixed with 10 ml of Instagel (Packard Chemicals, Downers Grove, IL, U.S.A.) as scintillation fluid, using a Packard A 300 CD liquid scintillation spectrometer. The efficiency of the apparatus, checked using a sample of standard tritiated water (Packard Chemicals, Downers Grove, IL, U.S.A.), was in the range 35%-40%. 2.4.2. cross-linking The determination of cross-linking was accomplished by hydroxylapatite chromatography according to Sinden and Cole [ 261 following the experimental procedure described (see ref. 17). Briefly, DNA was denatured by mild alkali treatment, the mixture was neutralized and then chromatographed on hydroxylapatite columns. The fractions were then adjusted to pH 12.5 and the optical density at 260 nm was determined. The number of cross-links was determined from y = - ln(D/c)IM, where &I, is the number average molecular weight of the untreated singlestranded DNA (0.73 X lo6 daltons, determined by sedimenting DNA in alkaline medium, using a Beckman L8-80 ultracentrifuge equipped with a preparative UV scanner), and D and C are the fractions of single-stranded, non-crosslinked DNA determined in the irradiated (0) and non-irradiated (C) samples. 2.5. Enzymatic photoadduct detect&n 2.5.1. Photoreaction with DNA The standard reaction involved a solution of 3 X 10m6 g of supercoiled RF DNA or 5’-end-labelled DNA fragment in TE buffer, 10 mM Tris-HCl, pH 8.0, 1 mM ethylenediaminetetraacetic acid (EDTA), and 10d3 ml of an AcPso solution (5 X lo-” M) in dimethylsulphoxide (DMSO), in a final volume of 0.02 ml. After standing for 30 mm in the dark at room temperature, the samples were irradiated with UVA light at a dose of 108 or 216 kJ m-‘. The unreacted furocoumarin was extracted with chloroform-isoamyl alcohol (24:1, vh) and DNA was precipitated with ethanol. 2.5.2. Photoadduct detection Photosensitized 5’-end-labelled DNA was incubated with T4 DNA polymerase 3’-5’ exonuclease in the absence of dNTPs, and the digested products were analysed on sequencing gels alongside Maxam and Gilbert

47

sequencing reaction products. The quantitation of the photoadducts was performed by cutting out slices of the gel corresponding to bands on the autoradiogram, measuring their radioactivity by Cerenkov counting and comparing it with the total radioactivity in the lane, as described in ref. 23. 2.5.3. Detection of guunine photo-oxidation Photosensitized 5’-end-labelled DNA underwent a chemical cleavage on treatment with 1 M piperidine at 90 “C for 20 min. The cleavage products were analysed on sequencing gels and quantified as described by Sage et al. [27]. 2.5.4. Exonuclmse III assay Photosensitized RF I supercoiled DNA was incubated in the buffer recommended by the manufacturer, in the absence or presence of 2 units of E. coli exonuclease III for 30 min at 37 “C, and analysed by electrophoresis on 1% agarose gel in the presence of ethidium bromide. The presence of DNA photolesions sensitive to exonuclease III was revealed by the conversion of supercoiled (RF I) DNA to circular opened (RF II) DNA. The ratio of RF II DNA to RF I DNA was determined for each sample by densitometric scanning of the negatives of the Polaroid photographs of ethidium-stained gels. The values were corrected for the difference in the staining intensity between relaxed and supercoiled DNA [ 281. 2.6. Mutagenesis tests The strain used was E. coli wP2 uvrA, caxrying nonsense mutations in the trp E gene which is reverted by UV light and most base pair substitution mutagens [ 29). Bacteria were grown overnight in a minimal Davis-Mingioli salts glucose medium supplemented with tryptophan (20 mg I-‘). E. coli cells were washed and then suspended in phosphate-buffered saline (pH 7.0) containing the compound to be studied (2X 10e5 M) at a density of lo* cells in-‘. Nitrogen or air (sterilized by filtration) was bubbled at room temperature and then the bacteria were irradiated with UVA light as described for DNA solutions. For the mutagenesis test, 0.1 ml aliquots of the irradiated suspensions were added to 2 ml of molten 0.6% top agar and poured onto plates containing 20 ml of SEM agar (MMA fortified with 0.1 mg ml-’ of Difco nutrient broth). For the determination of the surviving fraction, the irradiated cells (0.1 ml) were diluted with phosphate buffer, added to 2 ml of molten 0.6% agar and plated on Davis-Mingioli minimal medium supplemented with tryptophan. The plates were incubated for 48 h at 37 “C in the dark and then the colonies were counted. The mutation frequency was expressed as mutants per lo6 survivors, according to Witkin [30], which were computed by dividing the number of revertants observed per plate by the number of surviving bacteria at the same treatment, and subtracting from the result the number of revertant colonies per lo6 survivors observed in the controls. In this test all manipulations were performed under red light.

48

3. Results

3.1. Eflect of owgen

on DNA photobinding

To study the effect of oxygen on the capacity of AcPso to photoreact with DNA, a calf thymus DNA solution containing 3H-AcPso was irradiated with increasing UVA doses in air or in a nitrogen atmosphere. The DNA photobinding of 3H-8-MOP was used as a reference. Figure 1 shows that the photobinding of both drugs to DNA is more efficient on irradiation in a nitrogen atmosphere than in air. Similarly, a marked increase in cross-link formation on irradiation in the presence of nitrogen is observed (Fig. 2). However, the quenching effect of oxygen on the total photobinding and cross-linking is much higher for AcPso than for &MOP, as seen in Table 1. The high quenching induced by molecular oxygen reduces the number of excited AcPso molecules and f’uran-sidemonoadducts, thus affecting the AcPso behaviour towards DNA. It should be noted that the presence of the acetyl group at the 5’ position of AcPso does not prevent the formation of furan-side monoadducts, allowing the subsequent generation of interstrand cross-links in DNA [ 171. However, AcPso is poorly reactive when UVA irradiation is carried out in air, and DNA cross-linking only occurs to a very small extent; therefore, in these conditions, AcPso may be considered as a monofunctional reagent. In contrast, in a nitrogen atmosphere, DNA crosslinking is more pronounced.

UVA WSE (KJh2

1

Fig. 1. Photoreaction between calf thymus DNA and 3H-AcPso or 3H-8-MOP on UVA irradiation in different atmospheres (DNA photobinding). DNA solutions containing the radioactive furocoumarins (2 x 10-j M) were irradiated in a nitrogen atmosphere or in air with increasing UVA doses and the radioactivity covalently bound to DNA was determined as described in Section 2. The data are the means of three experiments; standard errors (SE) are in the range +0.05-kO.25. AcPso: Cl, in air; I, in nitrogen; g-MOP: X, in air; A, in nitrogen.

49

UVADOSE (KJ/m2) Fig. 2. Photoreaction between calf thymus DNA and cold AcPso or 8-MOP on UVA irradiation in different atmospheres (DNA cross-linking). DNA samples were irradiated as described in Pig. 1 and the number of cross-links per 10’ dalton was determined as described in Section 2. The data are the means of four experiments; standard errors sre in the rangefO.Ol-f0.03. AcPso: 0, in air; W, in nitrogen; 8-MOP: X, in air, A, in nitrogen.

TABLE 1 Oxygen quenching (%) of photoreaction with DNA Compound

Photobinding

Cross-Wring

8-MOP AcPso

30*2.0 66.6k3.1

23k 1.2 97rt2.1

The data are the means (*SE) of three determinations performed at a dose of 4.2 kJ mm2 of UVA light and a furocoumarin concentration of 2X lo-’ M; experimental conditions are described in Section 2.4.

3.2. Singlet oxygen production @ AcPso monoadducts The capacity of AcPso to produce singlet oxygen has been studied [ 171. We now compare it with the production of singlet oxygen by the AcPso moiety linked to DNA, i.e. its photoadducts. AcPso and ES-MOPadducts with DNA were irradiated with UVA and the production of singlet oxygen was measured. F’igure 3 shows that the amount of singlet oxygen generated by AcPso photoadducts is much higher than that produced by the free compound. Taking into consideration the poor capacity of AcPso to photoreact with DNA, the data must be corrected for the actual concentration of the bound sensitizer present in the irradiation mixture. After this correction, and at a UVA dose of 10 kJ mp2, the AcPso adducts are 61 times more effective than the free compound in solution. These data are consistent with the quasiabsence of DNA cross-linking when irradiation is performed in air and with

._ 5ol -//.’

P

10

0

iz d

8

f

6

B i

UVA DOSE (KJ/m2) Fig. 3. Singlet oxygen production by AcPso and B-MOP, free in solution or bound to DNA. The activity of the free chemicals was determined according to ref. 25 for the photoadducts; solutions of calf thymus DNA (20 OD units) containing the compound examined (2 X 10e5 M) were irradiated with UVA (5.6 kJ me2); the photoadducts containing DNA were precipitated and redissolved in the initial volume of phosphate buffer as described in Section 2. The data are the means of three experiments; standard errors are in the range f0.2-f0.5 for B-MOP and *0.4-kO.85 for AcPso. 0, AcPso; W, AcPso photoadducts; X, B-MOP; A, B-MOP photoadducts.

greater effect of oxygen quenching on cross-linking than on DNA photobinding (see Figs. 1 and 2). In contrast, free &MOP is a poor singlet oxygen producer; its adducts appear to be effective at low UVA doses, but at higher doses the singlet oxygen production slows down; this is probably due to the extensive conversion of 4’,5’ monoadducts into cross-links, so that the chromophore is no longer available for energy transfer to molecular oxygen. the

3.3. Sequence speca@ity in the photoaddition of AcPso to DNA A 100 bp DNA fragment from phage Ml3 mp8 was irradiated in the presence of an excess of AcPso. Photoadducts were detected and quantified using T4 DNA polymerase 3’-5’ exonuclease coupled with DNA sequencing methodology [31 I. In these experiments, AcPso appears to be very poorly reactive with DNA, since high UVA doses are required to detect photoadducts by this method. The total amounts of photoadducts are 0.07 and 0.11 adducts per DNA molecule for UVA doses of 108 and 2 16 kJ mW2.For comparison, g-MOP and 4,4’,6-trimethylangelicin produce 2.5 and 11.7 photoadducts per DNA molecule for a UVA dose of 18 kJ mm2 [23]. Figure 4 (Ianes 2 and 3) shows the termination sites of the 3’-5’ exonuclease associated with T4 DNA polymerase, corresponding to the sites of photoaddition. When DNA is exposed to AcPso without irradiation, the exonuclease totally degrades DNA to nucleotides and no bands appear on the autoradiogram (Iane 1). The frequency of photoaddition at each site is determined by radioactivity

51 12

3CTCGAG

4

5

6

TC TT~ CT T TT

TT

TA

Fig. 4. Detection of the photoaddition and photo-oxidation induced by AcPso on a DNA fragment. 100 bp 5’-end-labelled DNA fragment was irradiated in the presence of 5 X 10m4 M AcPso at a UVA dose of 108 kJ mm2 (lanes 2 and 5) or 216 kJ mm2 (lanes 3 and 5) or unirradiated (lanes 1 and 4). Sites of photoaddition are detected as stop sites at the 3’-5’ exonuclease of T4 DNA polymerase (lanes l-3). Sites of photo-oxidation are detected by cleavage with hot piperidine (lanes 4-6, arrows).

counting in the gel corresponding to each band on the autoradiogram. The distribution of photoadducts in the 100 bp DNA fragment is represented in Fig. 5; as for other furocoumarins [23, 31-361. AcPso reacts better with thymine located in 5’TpA. Figure 5 leads to the following observations: TA>TG; AT>GT; TA>AT; TG>TC; TG>GT. These observations are in line with a preferential intercalation in 5’Py-Pu sequences (371. Considering the number of T residues in reactive sites, AcPso reacts better with T in TA than in ‘IT or T’M’ in GC context (TA > TI” > TPT). The occurrence of photoaddition at C residues is low and only in AC/CA sites. Experiments using another DNA fragment confnm the preference of AcPso for AT-rich sequences (data not shown). This is consistent with the general rule of furocoumarin and furochromone photobinding [34, 361. Nevertheless, T residues surrounded by G/C are more reactive to AcPso than to other psoralen derivatives; in this respect, the sequence specificity in the photoreaction of AcPso is close to that of the methylated angelicins [23] or visnagin [36]. AcPso photobinding to DNA is less sequence specific than the photoaddition of other derivatives of the psoralen series [34].

52

0.2 0 $AGTGAGC;ACTCACA~AATTGCGTTGCGCTCT

CACTGCCCGC=CCAGiCGGGAAACCTGTCG3,

Distribution of photoadducts and photo-oxidation along the 100 bp DNA fragment. Open bars represent the occurrence of photoadducts and 5lled bars represent the occurrence of photo-oxidation. The height of the bars represents the frequency of damage at each reactive site. F’ig. 5.

3.4. DNA photo-oxidation induced by AcPso 3.4.1. Detection by sequencing methodology C&W- hot piperidine cleavage As already demonstrated, certain furocoumarins induce G photo-oxidation which can be detected using the DNA sequencing methodology after cleavage by hot piperidine [27]. Figure 4 (lanes 5 and 6) and Fig. 5 (filled bars) show the cleavage of AcPso-photosensitized DNA at the G position after treatment with hot piperidine. For a UVA dose of 108 kJ mW2 (lane 5) cleavage products are barely detected above the background (lane 4). In contrast with 3carbethoxypsoralen (3-CPs), all G residues appear to be sensitive to hot alkali, but exhibit different extents of oxidation. It seems that flanking sequences influence the reactivity of G residues. These differences observed in DNA photo-oxidation induced by 3-CPs and AcPso suggest that the mechanism may differ for the two compounds. The total amount of photo-oxidized G, residues is about 0.08 per DNA molecule for a UVA dose of 2 16 kJ rnm2, which is similar to the amount of photoadducts observed in these experimental conditions. The photoreaction between AcPso and DNA was performed in the presence of Tris and 5% DMSO, two radical quenchers; therefore, in our experimental conditions, we can suppose that G photooxidation is mainly due to singlet oxygen generation (see ref. 27, and references cited therein). Since AcPso photoadducts are efficient producers of singlet oxygen compared with free AcPso, it is probable that G, oxidation occurs by AcPso photoadduct sensitization, i.e. by energy transfer from the 4’,5’ monoadducts in the excited state to molecular olrygen, rather than by the free compound. The singlet oxygen generated by the photoadducts probably diffuses, since no preferential photo-oxidation of G residues is observed close

53

to the photoaddition site (Fig. 5). We recall that both ‘Oa and radical reactions are involved in DNA photo-oxidation by 3-CPs [27]. In addition, as with 3CPs, sequences of G residues are more reactive, in particular, the two G residues on the 5’ side in the sequence of three G residues (see Fig. 5). Under the conditions which produce photoadducts, &MOP and angelicin do not induce photo-oxidized G residues sensitive to hot alkali. 3.4.2. Detection of alkali-labile sites by E. coli exonu.clease III Alkali-labile G residues photoinduced in DNA by 3-CPs have been demonstrated to be substrates for the AP-endonuclease activity of E. coli exonuclease III [ 2 7 1. (AP denotes the presence of apurinic:apyrimidinic sites.) AcPso-photosensitized RF DNA was tested for its sensitivity to incision by exonuclease III, followed by the conversion of RF I DNA to RF II DNA, analysed on agarose gel. The results given in Table 2 demonstrate that singlestrand incision by exonuclease III can be detected only after the DNA-AcPso mixture has received a UVA dose of 216 kJ m-‘, and not at lower doses. Furthermore, the photoreaction between AcPso and RF I DNA leads to singlestrand breaks, which convert a fraction of RF I DNA molecules to RF II (compare the RF II DNA to RF I DNA ratios with that of untreated DNA). Moreover, by heating AP-DNA, AcPso-DNA and psoralen-treated DNA at 37 “C in the absence of enzyme, some RF II DNA is induced. This is not observed to the same extent for 3-CPs-treated DNA. This may indicate a difference in the mechanism of production of oxidative damage. Assuming a Poisson distribution of the lesions induced in DNA molecules and correcting for the lower fluorescence of ethidium bromide in supercoiled RF I DNA [28], we can calculate the number of single-strand breaks per DNA molecule. We estimate that DNA exposed to AcPso plus 216 kJ rnb2 of UVA light contains 0.5 sites sensitive to exonuclease III per DNA molecule. For comparison, the same enzymatic treatment of DNA irradiated in the presence of psoralen (UVA dose of 144 kJ mS2) or in the presence of 3-CPs (36 kJ mV2) leads to 0.65 and 0.9 single-strand breaks per DNA molecule respectively [27, 361. It appears that the strength of the photodynamic effect on DNA for furocoumarins and related compounds is in the order: 3-CPs > psoralen> AcPso B visnagin, khellin, angelicin, &MOP. However, it should be noted that for the last series of compounds, no photoreaction of DNA involving oxygen was detected. This result is not in line with the hypothesis of a photodynamic mechanism of mutation induction by ES-MOP[20, 211. In view of the size of the target DNA (about 8150 bp), this method detects less DNA damage (photoaddition) than hot piperidine treatment followed by analysis on sequencing gel. One reason may be that all the alkali-labile lesions are not substrates for exonuclease III. The other possibility is that our conditions of digestion with exonuclease III are not optimal. In principle, the rapid test combining the use of exonuclease III digestion and agarose gel electrophoresis is efficient in the detection of the photodynamic effect of photosensitizers on DNA. A similar assay, using enzymatic extracts

M AcPso

was prepared by Incubating RF Ml3

mp8 DNA at 70 “C

0.8 1.7 2.0

0.7 0.9 0.8

0.6 0.4 0.5

0.8 1.1 1.5

144 kJ rnez

108 kJ m-’

72 kJ m-’

216 k.I m-’

Pso”

DNA

III (RF II DNA

AcPso

Treated

and UVA irradiated. AP-DNA

0.6 1.2 2.8

0.4 0.3 0.3

No Incubation Buffer Exo III 2u

“Data from TrabaIsinI et al. [36]. RF Ml3 mp8 DNA was exposed to 5 x lo-’ for 5 mIn at acidic pH 1381.

AP-DNA 0 kJ m-’

by AcPso before and after incubation at 37 “C with E. coli exonuclease

Untreated DNA 0 kJ m-’

mp8 RF DNA photoinduced

Experimental conditions

Photodegradation of Ml3 to RF I DNA ratio)

TABLE 2

:

55

from bacteria or mammalian cells instead of puri6ed enzyme (as in our assay), has been developed to classify DNA damage according to its sensitivity to different classes of endonucleases present in the extracts (391. Our assay is very useful for determining the potency of an agent in inducing oxidative DNA damage. 3.5. Photobiological activity in E. coli MT2 uvrA cells E. coli WP2 uvrA cells, which are sensitive to UV-like lesions, were irradiated with increasing UVA doses in the presence of AcPso in equilibrium with air or in a nitrogen atmosphere. The surviving fraction and the number of revertants, determined as described in Section 2, are shown in Fig. 6 (8MOP was used as a reference). When the UVA irradiation is carried out in a nitrogen atmosphere, both compounds appear to be more genotoxic. For SMOP, which does not induce detectable photo-oxidative damage, the observed increase in the cytotoxicity and genotoxicity is clearly related to the weak oxygen quenching of the photoreaction with DNA. For AcPso, the

UVA DOSE (KJ/m2,

Lethal and mutagenic activity of AcPso and S-MOP in E. coli WP2 uwA cells on UVA irradiation in air or in a nitrogen atmosphere. Bacteria were irradiated in the presence of the furocoumarin studied (2 X 10m5M) in different atmospheres and at different doses and the surviving fraction and revertant number were scored as described in Section 2. The data are the means of two experiments; standard errors are in the range &0.03-f 0.06 for survival and+O.l-f0.035 for mutagenesis. AcPso: 0, in air, W, in nitrogen; ES-MOP: X, in air, A, in nitrogen. Fig.

6.

56

amount of oxidative photodamage induced in bacterial DNA in a nitrogen atmosphere should be very low or zero, and DNA photobinding should increase. This leads to two opposite effects. For AcPso in the absence of oxygen, photo-oxidative damage should be less important than DNA photoaddition. It should be noted that the E. coli strain used in this study does not cause the excision of the furocoumarin photoadduct, whereas it is efficient in repairing singlet-oxygen-induced oxidative photolesions. If singlet oxygen production by AcPso monoadducts is mainly responsible for the photooxidative damage, we can suppose that this lesion affects the neighbourhood of the monoadduct. Therefore in a section of DNA which contains an adduct and a nearby photo-oxidized guanine, one photolesion may affect the repair of the other. In air, AcPso, which forms about ten times fewer photoadducts than 8MOP, but generates singlet oxygen, is as genotoxic as &MOP. If we examine the increase in lethality and mutagenesis obtained on suppression of oxygen quenching, it can be seen that it is much better correlated with the increase in cross-linking rather than with the total number of photoadducts formed in DNA, for both compounds (compare Figs. 1 and 2 with Fig. 6). 4. Discussion In air, AcPso shows poor photoreactivity with DNA and mainly acts as a monofunctional derivative. In a nitrogen atmosphere, DNA photobinding is increased and DNA cross-links are clearly detected. Oxygen quenching is more important for AcPso than for &MOP, and is probably related to the large capacity of AcPso to generate singlet oxygen. AcPso photoadducts are better producers of singlet oxygen than the free drug in solution. This explains the observation that, in the presence of air, AcPso fur-an-side photoadducts, which absorb UVA light, transfer energy to molecular oxygen in the solution with good efficiency and do not form interstrand DNA cross-links. AcPso can also induce photo-oxidation in DNA, probably via singlet oxygen. Enzymatic or chemical cleavage of photosensitized DNA coupled with DNA sequencing methodology have enabled us to map at the nucleotide level and to quantify the two types of photolesions induced in DNA. The sites of photoaddition are the blockage sites of the 3’-5’ exonuclease of T4 DNA polymerase. This assay has previously been used to study the sequence specificity in the DNA photobinding of psoralen derivatives [33, 341, angelicins [23] and furochromones [36]. DNA photo-oxidations were detected as hot alkali-labile sites as in ref. 27. Photo-oxidized G residues are sensitive to E. coli exonuclease III. Although AcPso photobinding is less sequence specific than other derivatives of the psoralen series, it still exhibits a preference for AT-rich sequences. The extent of DNA photo-oxidation induced by AcPso is lower than that induced by 3-CPs or psoralen, which are known to be good producers of singlet oxygen [40, 411. In contrast, SMOP, a very weak producer of singlet oxygen, does not induce DNA photooxidation [27, 401.

57

The genotoxicity of AcPso in E. coli WP2, in the presence of oxygen, is about the same as that of SMOP, whereas AcPso forms fewer DNA photoadducts. In the absence of oxygen, the genotoxicity is clearly higher than in the presence of oxygen, and is higher than that of S-MOP. This is in contrast with that observed for 3-CPs in yeast. Indeed, a slight increase in survival is obtained after treatment with 3-CPs plus UVA light in the presence of a quencher of singlet oxygen [ 411. The genotoxic activity of AcPso seems to be mainly related to its capacity to form cross-links in DNA. Indeed, monoadducts, and in particular photo-oxidative damage, appear to be less effective. It has previously been assumed that photo-oxidative lesions induced by 3-CPs are not very mutagenic [27] In conclusion, these studies confirm the data previously reported by Carlassare et al. [ 171 on the possibility of obtaining a monofunctional agent for photochemotherapy: the insertion of an acetyl group into the psoralen skeleton cannot prevent the cross-linking capacity, but reduces the ability of the compound to photobind to DNA. However, on irradiation in air, the number of cross-links is very small and AcPso behaves as a monofunctional reagent. It has previously been observed that another electron-withdrawing group, a carbomethoxy moiety, inserted at the 5’ position, yields a good DNA cross-linker [ 421. The good energy transfer from AcPso, and in particular from its fur-an-sidemonoadducts, to molecular oxygen confers another peculiar mode of sensitization to the molecule: photo-oxidation; however, in spite of this, its antiproliferative activity is not very marked, skin phototoxicity is absent [ 171 and its mutagenicity is comparable with that of 8-MOP. Prom a consideration of these results, we believe that the activity of AcPso should be studied in more detail to ascertain whether or not it has the potential for use as a photochemotherapeutic agent and to clarify its effect on other biological systems, such as erythrocytes and membranes of mammalian cells.

References 1 M. A. Pathak, J. A. Parrish and T. B. Fitzpatrick, Psoralens in photochemotherapy of skin diseases, II Farmaco, Ed. Sci., 36 (1981) 479-491. 2 E. Ben-Hur and P. S. Song, The photochemistry and photobiology of furocoumarins (psoraiens), Adv. Radiat. Biol., II (1984) 131-171. 3 G. Cimino, H. Gamper, S. Isaacs and J. Hearst, Psoraiens as photoactive probes of nucleic acid structure and function: organic chemistry, photochemistry and biochemistry, Annu. Rev. Biochem., 54 (1985) 1151-1193. 4 D. Averbeck, Recent advances in psoraien phototoxicity mechanism, Photo&em. Photobid., 50 (1989) 859-882. 5 N. Babudri, B. Pani, S. Venturini, M. Tamaro, C. Monti-Bragadin and F. Bordin, Mutation induction and killing of V79 Chinese hamster cells by 8-methoxypsoralen plus near-ultraviolet light: relative effects of monoadducts and cross-links, Mutat. Res., 91 (1981) 391394. 6 R. S. Cole, Inactivation of EscherichzYu co& F’ episomes at transfer and bacteriophage lambda by psoralen plus 360 nm light: significance of deoxyribonucleic acids cross-links, J. Bacterial., 107 (1971) 846852.

58 7 T. Seki, K. Nozu and S. Kondo, Differential causes of mutation and killiig in Escherichia coli after psoralen plus light treatment: monoadducts and cross-links, Photochetn. Photobiol., 27 (1978) 19-24. 8 M. Tamaro, C. Monti-Bragadin, P. Rodighiero, F. Baccichetti, F. Carlassare and F. Bordin, Khiing and mutations induced by mono- and bifunctional furocoumarins in wild type and excision-less Escherichiu coli strains carrying the R46 plasmid, Photobiochem. Photobiophys., IO (1986) 261-267. 9 E. L. Grant, R. C. von Borstel and M. J. Ashwood-Smith, Mutagenicity of cross-links and monoadducts of furocoumarins (psoralen and angelicin) induced by 360~nm radiation in excision-repair-defective and radiation-sensitive strains of Sacckaromyces cerevisiae, Environ. Mutagen., I (1979) 55-63. 10 L. Musajo and G. Rodighiero, The skin-photosensitizing furocoumarins, Experientia, 18 (1962) 153-162. 11 L. Dubertret, D. Averbeck, E. Bisagni, J. Moron, E. Moustacchi, C. Bihardon, D. Papadopoulo, N. Nocentini, P. Viiy, J. Blais, R. Bensasson, J. C. Ronfard-Haret, E. J. Land, F. Zajdela and R. Latarjet, Photochemotherapy using pyridopsoralens, Biochimie, 67 (1985) 417-422. 12 P. Mullen, M. A. Pathak, J. D. West, T. J. Harrist and F. Dall’Acqua, Carcinogenic effects of monofunctional and bifunctional furocoumarins, Nat. Cant. Inst. Monogr., 66 (1984) 205-210. 13 A. R. Young, I. A. Magnus, A. C. Davies and N. P. Smith, A comparison of the phototumorigenic potential of 8-MOP and 5-MOP in hairless albino mice exposed to solar simulated radiation, Br. J. Dennutol., 108 (1983) 507-518. 14 F. Baccichetti, F. Carlassare, F. Bordin, A. Guiotto, P. Rodighiero, D. Vedaldi, M. Tamaro and F. Dall’Acqua, 4,4’,6-Trimethylangelicin, a new very photoactive and non skin phototoxic monofunctional furocoumarln, Photochem. PhotobioZ., 39 (1984) 525-529. 15 A. Guiotto, P. Rodighiero, P. Manzini, G. Pastorini, F. Bordin, F. Baccichetti, F. Carlassare, D. Vedaldi, F. Dall’Acqua, M. Tamaro, G. Recchia and M. Cristofohni, 6-Methylangelicins: a new series of potential photochemotherapeutic agents for the treatment of psoriasis, J. Med. Chem., 27 (1984) 959-967. 16 L. Dubertret, D. Averbeck, R. Bensasson, E. Bisagni, F. Gaboriau, E. J. Land, S. Nocentini, M. T. de San E Melo, E. Moustacchi, P. Morliere, J. C. Ronfard-Haret, R. Santus, P. Vigny, F. Zajdelaand R. Latarjet,Photophysicalphotochemical, photobiological andphototherapeutic properties of 3-carbethoxypsoralen, in J. Calm et al. (eds.), Psorakns in Cosmetics and Dermutolog~, Pergamon, New York, 1981, pp. 245-256. 17 F. Carlassare, F. Baccichetti, A. Guiotto, P. Rodighiero, 0. Gia, A. Capozzi, G. Pastorini and F. Bordin, Synthesis and photobiological properties of acetylpsoralen derivatives, J. Photo&em. Photobiol. B: Biol., 5 (1990) 25-39. 18 P. C. Joshi and M. A. Pathak, Production of singlet oxygen and superoxide radicals by psoralens and their biological significance, Bioch.em. Biophgs. Res. Commun., I12 (1983) 638-646. 19 N. J. De Mol and G. M. J. Beuersbergen van Henegouwen, Relation between some photobiological properties of furocoumarins and their extent of singlet oxygen production, Photocha. Photobiol., 33 (1981) 815-819. 20 L. Santamaria, A. Bianchi, L. Bianchi, R. PiizaIa, G. Santagati and P. Bermond, Photomutagenicity by 8-methoxypsoralen with and without singlet oxygen involvement and its prevention by pcarotene, Med. Biol. Environ., 12 (1984) 541-546. 21 L. Santamaria, L. Bianchi, A. Bianchi, R. PizzaIa, G. Santagati and P. Bermond, Carotenoids but not vitamin A can control the oxygen-dependent step of photomutagenicity by furocoumarins, Med. Biol. Environ., 13 (1985) 121-129. 22 J. Marmur and P. Doty, Determination of the base composition of deoxyribonucleic acid from its thermal denaturation temperature, J. MOL. Biol., 5 (1962) 109-118. 23 G. Miolo, F. Dall’Acqua, E. Moustacchi and E. Sage, Monofunctional angular furocoumarins: sequence specificity in DNA photobinding of 6,4,4’-trimethylangelicin and other angelicins, Photo&em. Photobiol., 50 (1989) 75-84. 24 C. G. Hatchard and C. A. Parker, A new sensitive chemical actinometer. II. Potassium

59

25 26

27

28 29

30 31 32 33 34

35

36

37

38 39

40 41 42

ferrioxalate as a standard chemical actinometer, Proc. R. Sot. London, Ser. B, 235 (1956) 518636. I. KraIjic and S. El Mohsni, A new method for the detection of singlet oxygen in aqueous solution, Photo&em. Photobiol., 28 (1978) 577-581. R. R. Sinden and R. S. Cole, Measurement of cross-links formed by treatment with 4,5’,8trimethylpsoralen and light, in E. C. Friedberg and P. C. HanawaIt (eds.), DNA Repair: A Laboratory Manual of Research Proceduws, Vol. I, Part A, Dekker, New York and Basel, 1981, pp. 69-81. E. Sage, T. LeDoan, V. Boyer, D. E. HeIland, L. Kittler, C. Helene and E. Moustacchi, Oxidative DNA damage photo-induced by 3-carbethoxypsoralen and other furocoumarins: mechanisms of photooxidation and recognition by repair enzymes, J. Mol. BioZ., 209 (1989) 297314. R. S. Lloyd, C. Haidle and D. L. Robberson, Bleomycm-specific fragmentation of double stranded DNA, Biochemistry, 17 (1978) 1890-1896. B. A. Bridges, R. P. Mottershead, M. A. RothwelI and M. H. L. Green, Repair deficient bacteria strains for mutagenic screening: test with the fungicide captan, Cksm. Biol. Interact., 5 (1972) 77-84. E. M. Witkin, Ultraviolet mutagenesis and inducible DNA repair in Eschmichia co& Bacterial. Rev., 40 (1976) 869-907. E. Sage and E. Moustacchi, Sequence context effects on 8-methoxypsoralen photobinding to defined DNA fragments, Biochemistry, 26 (1987) 33073314. H. Gamper, J. Piette and J. E. Hearst, Efficient formation of a cross-h&able HMT monoadduct at the Kpn I recognition site, Photo&em. Photobiol., 40 (1984) 29-34. E. Sage, V. Boyer and E. Moustacchi, Sequence specificity in psoralen-DNA photobinding, Biochena. Pharmmol., 37 (1988) 1829-1830. V. Boyer, E. Moustacchi and E. Sage, Sequence speciiicity in photoreaction with various psomlen derivatives with DNA: role in biological activity, Biochemistry, 27 (1988) 3011-3018. E. A. Ostrander, R. A. Karty and L. M. Hallick, High resolution psoralen mapping reveals an altered DNA helical structure in the SV40 regulatory region, Nucleic Acids Res., 16 (1988) 213-227. F. Dall’Acqua and P. Martelli, Photosensitizing action of furocoumarins on membrane components and consequent intracellular events, J. Photo&em. Photobiol. B: Bill., 8 (1991) 235-254. H. M. SobeIl, T. D. Sakore, S. C. Jain, A. Banexjee, K. K. Bhandary, B. S. Reddy and E. D. Lozansky, Beta-linked DNA, a structure that gives rise to drug intercalation and DNA breathing, and its wider significance in determining the premeltlng and melting behaviour of DNA, Cold &ring Harbor Sgmp. Quant. BioL, 47 (1982) 293-314. T. Limiahl and B. Nyberg, Rate of depurination in native deoxyribonucleic acid, Biochemistry, II (1972) 3610-3618. B. Epe, P. Mutzel and W. Adam, DNA damage by oxygen radicals and excited state species: a comparative study using enzymatic probes in vitro, Chem. Bill. Interact., 67 (1988) 149-165. C. N. Knox, E. J. Land and T. G. Truscott, Singlet oxygen generation by furocoumarin triplet states. Linear furocoumarins (psoralens), Photo&em. Photobiol., 43 (1986) 359-363. D. Averbeck, G. Innocenti, D. Vedaldi and F. Dall’Acqua, Photobiological effects of methylpsoralens in haploid yeast: the role of oxygen, Med. Btil. Environ., 14 (1986) 17-25. S. Marcia&Magno, P. Rodighiero, 0. Gia, F. Bordin, F. Baccichetti and A. Guiotto, Carbomethoxy-derivatives of psoralen: interactions with DNA and photobiological properties, Il Farmaco, Ed. Sci., 36 (1981) 629-647.