858
Chem. Res. Toxicol. 1993,6, 858-865
Characterization of the cis-syn and cis-anti Diastereoisomers of 5-Methoxypsoralen Pyrone-Side Monocycloadducts to Thymidine Catherine Anselmino, Lucienne Voituriez, and Jean Cadet’ CEAID6partement de Recherche Fondamentale sur la Matibre CondensBe, SESAMILAN, BP 85X, F-38041 Grenoble, France Received June 23, 199P
The photoreaction of 5-methoxypsoralen (5-MOP) with thymidine as a DNA model compound was investigated under dry-state conditions. In this respect, a thin film of thymidine and 5-MOP in a ratio 1O:l was exposed to 350-nm UV light. Four [2+21 photocycloadducts were isolated in a 0.5-2.2% yield with respect to 5-MOP by HPLC and characterized as two pairs of cis-syn and cis-anti diastereoisomers, respectively, on the basis of extensive spectroscopic measurements, including UV, fast atom bombardment mass spectrometry, lH and 13C NMR, and CD. Information concerning the absolute configuration of the four photocycloadducts was inferred from detailed nuclear Overhauser enhancement experiments. This is indicative of a 3R74S,5R,6S and a 3S94R,5S,6R configuration for the cis-anti cycloadducts (1, 2) and a 3S,4R,5R,6S) and a 3R,4S95S,6Rconfiguration for the cis-syn cycloadducts (3,4). In addition, conformational features of the four photocycloadducts were obtained from consideration of various l H NMR measurements including NOE data.
Introduction Furocoumarins represent a class of natural and synthetic compounds which are widely used in photochemotherapy due to their photosensitizing properties (1, 2). These include 8-methoxypsoralen (8-MOP),l5-methoxypsoralen (&MOP),and 4,5’,8trimethylpsoralen (4,5’,8TMP), which, when associated with UV-A irradiation, are efficient in the treatment of skin diseases such as vitiligo and psoriasis. Photoexcited furocoumarins have been reported to covalently bind to lipid membranes (3) and cellular macromolecules including nucleic acid and proteins (4-8). The photomodifications induced by psoralens in DNA and related models have been extensively investigated (for reviews see refs 9-12). I t is reasonable to assume that photobiological effects of furocoumarins arise, at least partly, from a [2+23 photocycloaddition involvingthe 3,4double bond of the pyrone moiety and/or the 4‘,5’-double bond of the furan moiety of the psoralen on one hand and the 5,6-ethylenic bond of the DNA thymine residue on the other hand. The isolation and the characterization of the 5-MOP adducts to DNA bases have received far less attention as compared to those of 8-MOP and 4,5’,8-TMP (13-23). Preliminary photobinding studies showed that, within DNA, the furan-side photoadducts of several bifunctional furocoumarins are the predominant photoproducts whereas in the dry state, the pyrone-side photoadducts are formed preferentially. In addition, it should be noted that only the cis-syn diastereoisomers are generallyformed in DNA. These observationswere further supported by molecular mechanics calculations of the intercalation complex of 8-MOP within a double-stranded dodecanucleotide, which indicate that energy-minimized
* To whom correspondence should be addressed. Phone: (33) 76-8849-87; FAX: (33) 76-88-50-90. *Abstract published in Advance ACS Abstracts, October 15, 1993. 1 Abbreviations: 8-MOP, 8-methoxypsoralen; 5-MOP, 5-methoxypsoralen; 4,5’,8-TMP, 4,5’,8-trimethylpsoralen;FAB-MS, fast atom bombardment mass spectrometry; ODs,octadecylsilyl, NOE,nuclear Overhauser effect.
structure and geometrical arrangement of psoralen promote the formation of furan-side monoadducts (24). The behavior of 5-MOP appears to be strikingly different, in spite of a similar structure, as inferred from molecular mechanics calculations. It was shown that, due to the position of the methoxy group, 5-MOP should react more easily through its pyrone double bond than uia its furan double bond. This received further support from a model study involving the UV photolysis of a dilute solution containing the two aromatic dThd and 5-MOP molecules linked by either a polymethylene chain of varying length or a diester chain. The photoreaction was found to give rise to the predominant formation of a cis-anti pyroneside monoadduct (25, 26). In order to complete these observations, the photoreaction of 5-MOP with thymidine as a DNA model was investigated in the dry state. The experimental conditions used were selected to preferentially induce 5-MOP pyrone-side monoadducts to the thymine moiety. We report here the isolation by highperformance liquid chromatography of the two diastereoisomeric pairs of cis-anti dThda-MOP (1, 2) and cis-syndThd < >: 5-MOP (3,4),produceduponnear-UV irradiation of thymidine and 5-MOP as a dry film. The characterization of photocycloadducts 1-4 was carried out on the basis of extensive spectroscopic measurements including UV, FAB-MS, lH and 13C NMR, and circular dichroism analysis.
Materials and Methods Chemicals. Thymidine and 5-MOP were purchased from Sigma Chemical Co (St. Louis, MO). Both compounds which were shown to be homogeneous by HPLC analysis were used without further purification. Analyticalsolvents used for HPLC analysis and other experiments were obtained from Farmitalia Carlo Erba (Milan, Italy). [6-2HlThymidine was prepared according to literature procedure (27). Typically thymidine (5 g) was dissolved in 60 mL of MezSO-de to which 20 mL of a solution of 10% of NaOD in DzO was added. Then, the resulting solution was warmed at 135 “ C under reflux, for 48 h. In a
0893-228~/93/2706-0858$04.00/0 0 1993 American Chemical Society
Chem. Res. Toxicol., Vol. 6, No. 6, 1993 859
Photocycloaddition Products to Thymidine subsequent step, [6-2H]thymidinewhich was purified by HPLC on a Waters prepacked ODS silica gel 500 column was obtained in a 3.75-g amount. The isotopic labeling as measured by lH NMR was about 75 % . UV Irradiation. The near-UV irradiation apparatus consisted of an array of six black light lamps (X = 350 nm), Model 08 RSF 40 BLB (Philips, Eindhoven, The Netherlands),arranged horizontally. The samples were photolyzed at a distance of 15 cm from the lamps. A fan provided efficient cooling of the samples during irradiations. Isolation of the 5-MOPothymidine Photocycloadducts 1-4. Thymidine (2.4 g) and 5-MOP (0.22 g) were dissolved in 800 mL of ethanol. The solution was mixed under magnetic stirring and then evaporated in a flat-bottom flask (30 cm X 40 cm) to give a thin film. The formation of the photoadducts required 10 successive irradiations, following repeated preparation of the dry film. After each irradiation, which lasted for 1 h, homogenization of the solution was achieved in order to allow the renewal of interactions between 5-MOP and thymidine. Then, the film was scraped off and dissolved in methanol. The resulting solution was filtered and evaporated under vacuum. The residue was further dissolved in a smallervolume of methanol and then filtered and evaporated to dryness. The operation was repeated until most of the thymidine was eliminated by precipitation. Under these conditions,the overallyield of formation of each of photoproducts 1-4 was between 0.5% and 2.2% as determined with respect to 5-MOP. HPLC Separations. [6-2H]dThdwas purified on a Waters prepacked ODS silica gel 500 column (300 mm X 57 mm i.d.) using a Waters Model LC 500 preparative liquid chromatograph (Mildford, MA). The eluent consisted of a mixture of water and methanol (90/10 v/v). The 5 - M O P o d T h d photocycloadducts 1-4 were purified on a 250 mm X 10 mm i.d. prepacked octadecylsilyl silicagel ODS Ultrapack column (SociBtBFranpise de Chromatographie,Paris, France). The HPLC system consisted of a L-6000pumpmodel (Merck,Darmstadt, Germany)associated with an ISCO 1840UV variable-wavelengthabsorbance detector (Roucaire, Velizy-Villacoublay, France). The mobile phase was made up with a mixture of water and methanol (65/35v/v) as an isocratic system at a flow rate of 3 mL/min. As expected from hydrophobic considerations, thymidine was eliminated in the void volume of the column, whereas the less polar 5-MOP was strongly retained on the column (Figure 1). It was the reason why 5-MOP had to be eliminated with methanol after each injection. The 5-MOP photocycloadducts to thymidine 1-4 exhibit intermediate retention time values between those of thymidine and 5-MOP. At first, the [2+21 photocycloadducts of 5-MOP to thymidine 1-4 were eluted, followed by the 5-MOP photodimers. The UV detection at 254 nm allowed the collection of each of the four main [2+2] 5 - M O P o d T h d photocycloadducts 1-4. Spectroscopic Measurements. UV and visible spectra were recorded on a Beckman Model 5230 UV spectrophotometer (Beckman, Irvine, CA). CD spectra were recorded in water on a Jobin-Yvon spectrophotometer, Model I11 (Jobin-Yvon,Longjumeau, France). FAB-MS were obtained on a VG ZAB 2-SEQ mass spectrometer using a LSIMS source in the positive mode (Fisons VG, Manchester, U.K.). 400.13-MHz lH NMR and 100.61-MHz l3C NMR spectra were recorded in the Fourier transform mode using a Bruker 400 apparatus (Bruker, Wissembourg, France). Depending on the concentration of the samples, 200-400 pulses were accumulated in a spectral width of 4000 Hz (16K data). The acquisition time was usually 2 s. For samples where the size of the residual HDO solvent signal in Me2SO was too large, a small quantity of DzO was added, shifting the HDO signal toward the low-field region. Nuclear Overhauser effect (NOE) experiments were carried out with the decoupling field gated off during the acquisition of the data. The time delay between the end of an acquisition and the beginning of the followingpulse was 1s. Spectra were averaged over an acquisition of 4000 transients, using a spectral width of 4000 Hz. 13CNMR were obtained on a Bruker 400-MHz spectrometer operating at
5-MOf-dThd
THYMIDINE
5-MOP
PHOTOADDUCT
\
1
-
5-MOPOdlhd PHOTCCVCLOADDUCTS
I-MOP dimers
,
I
o 4
.
8
I
I
I
I
$ .
.
.
I
%. ,
,
12 1 6 20 24 28 32 36 40 44 48 min
Figure 1. HPLC separation of the products of the photoreaction of 5-MOP with thymidine on a prepacked octadecylsilyl silica gel Ultrapack column (250 mm X 10 mm i.d.) in an isocratic mode. Eluent: water/methanol(65/35). Flow rate: 3 mL/min. UV detection: 254 nm. The peak numbers refer to the four 5 - M O P o d T h d monocycloadducts 1-4. 100.61 MHz. About 2000-4000 transients were recorded in the spectral width of 27000 Hz with a time delay of 3 s. Twodimensional correlated 1H NMR spectra (2D COSY) were recorded on a Bruker 400-MHz spectrometer. The decoupling field was gated off during the acquisition of the data. The time delay between the end of an acquisition and the beginning of the following pulse was 2 s. About 10 000 pulses were accumulated in a spectral width of 4000 Hz. Two-dimensionalheteronuclear correlated lH-l3C NMR experiments (XHCORRC) were also performed on a Bruker 400 apparatus operating at 400.13 MHz for lH and 100.61MHz for W, respectively. Proton assignments were achieved by either specific homonuclear decoupling experiments or 2D COSY analysis. H-2’ and H-2” proton were further assigned on the basis of coupling constant arguments (28). Simulated lH NMR spectra were obtained from iterative computer analysis by using the LAOCOON I11 program. 5-MOP Pyrone-Side Monocycloadducts to Thymidine. (A) (3~45,5~6S)-cis-anti-5-MOPthymidine Photocycloadducts (1). The combined fastest eluting HPLC (k’= 4.25) fractions were evaporated to dryness, giving rise to 2.4 mg of a UV-absorbingcompound (yield with respect to 5-MOP = 1.1% 1. UV (A-, HzO): 217 nm; CD (Am, HzO): 230 nm, 260 nm; FABMS positive mode (relative intensity): 481.1 ([M + Nal+, 5.51, 459.3 ([M + HI+, 23), 343 ([B-MOPThy + HI+, 18.5), 217 ([5-MOP + H I + , 100); 13C NMR (100.61 MHz, Men-
860 Chem. Res. Toxicol., Vol. 6, No. 6,1993 SO-&): 22.3 (CH3, dThd), 36.0 (C-2’, dThd), 38.3 (C-4,5-MOP), 41.1 (C-3,5-MOP),49.98,53.2 ((2-6, dThd), 59.0 (OCHs, 5-MOP), 62.2 (C-5’, dThd),71.0 ((2-3’)dThd), 83.6 (C-1’)dThd), 85.2 (C-4’, dThd), 93.9 (C-8, 5-MOP), 102.1 (C-4’, 5-MOP), 105.5, 112.8, 144.9 (C-5’, 5-MOP), 149.6, 151.5, 152.3, 155.8, 163.3, 172.0. (B) (3S,4&5S,GR)-ci~-an ti-S-MOPt hymidine Photocycloadduct (2). The HPLC fractions containing the second photoproduct (k’= 5.50) were pooled. Evaporation to dryness of the resulting solution yielded 1mg of a white compound (yield with respect to 5-MOP = 0.5%). UV (Aw, H2O): 220 nm; CD (Am, HpO): 230 nm, 265 nm; FAB-MS positive mode (relative intensity): 459.4 ([M + HI+, 2.6)) 217 ([5-MOP + HI+, 6); 13C NMR (100.61 MHz, MezSOd6): 22.4 (CH3, dThd), 38.01, 41.3, 49.6,54.4,58.8,61.7,70.2,86.8,87.1,93.2,97.3,97.4,101.6,105.6, 112.5, 112.3, 144.7, 149.6, 150.5, 151.3, 155.3, 163.4, 171.7. (6)(3S,4R,S&GS)-cis-syn-5-MOPthymidine Photocycloadduct (3). Evaporation to dryness of the third main combined HPLC fractions (k’= 6.12) gave 4.8 mg of a white compound (yield with respect to 5-MOP = 2.2%). UV (Am, HzO): 219 nm; CD (Am, H20): 220nm, 240 nm; FAB-MS positive mode (relative intensity): 481.1 ([M + Na]+, 28), 459.1 ([[M + HI+, 11.5))343.1 ([5-MOPThy HI+, 9.5), 217 ([&MOP + HI+,100);13CNMR (100.61MHz, Me2SO-de): 22.3 (CH3,dThd), 36.7 (C-4, &MOP), 38.6 ((2-3, 5-MOP), 38.9 (C-2’) dThd), 52.4, 52.4 (C-6, dThd), 59.7 (OCH3, 5-MOP), 61.6 (C-5’, dThd), 70.4 (C-3’) dThd), 84.9 (C-1’, dThd), 86.7 (C-4’) dThd), 93.2, 104.4, 105.1 (C-4’) 5-MOP), 112.9, 144.85 (C-5’, 5-MOP), 149.5, 150.5, 151.0, 155.3, 164.2, 171.1. (D) (3&4S,5S,GR)-cis-syn-5-MOPthymidine Photocycloadduct (4). Evaporation to dryness of the slowest HPLC eluting fraction (k’= 6.63) yielded 3.9 mg of a white compound (yield with respect to 5-MOP = 1.8%). UV (Aw, HzO): 217 nm; CD (Am, HzO): 220nm,240nm;FAB-MS positive mode (relative intensity): 481.1 ([M + Na]+, 22), 459.1 ([M HI+, ll),343.1 ([5-MOPThy + HI+, 9))217 ([5-MOP + HI+, 100);13CNMR (100.61 MHz, MezSO-d6): 22.3 (CH3, dThd), 36.7 (C-2’, dThd), 36.7, 38.9, 51.2, 52.1,59.7, 62.0 (C-5’, dThd), 70.6 ((2-3’) dThd), 82.7,85.8,93.4,104.4,105.1,113.1,145.0,149.3,150.5,151.3,155.32, 164.4, 164.2.
Anselmino et al.
+
+
Results and Discussion Characterization of the 5-MOP Photoadducts to Thymidine. Irradiation of a 1 0 1 mixture of thymidine and 5-MOP in the dry state led to the formation of five photoadducts which were efficiently separated by HPLC on an ODS Ultrasphere column (Figure 1)and identified as the two pairs of cis-syn and cis-anti diastereoisomers of 5-MOP pyrone-side monocycloadduct (vide infra). It should be noted that the slowest HPLC eluting photoadduct was assigned as a 5-MOP-thymidine photoadduct involving the covalent binding of 3,4-dihydro-5-methoxypsoralen-3-yl radical to a-thymidinyl radical (29). UV Absorption. The UV absorption spectra (Figure 2) of the four main photoproducts 1-4 show similar features with a maximum of absorption at 220 nm. The lack of an absorption peak a t 310 nm indicates that the photoreaction did not involve the 4‘,5’-double bond of the furan moiety of 5-MOP. In addition, the presence of a shoulder around 300 nm is in agreement with the fact that the photoreaction occurred between the 3,4-double bond of the 5-MOP pyrone moiety on one hand, and the 5,g-double bond of thymidine on the other hand. Mass Spectrometry. The mass spectral data were obtained by using the FAB-MS method in the positive mode. For each of the four adducts 1-4 studied, a molecular ion peak was observed at mlz = 459. It corresponds to a 1:l thymidine-psoralen adduct. The
2c3
300
400
h (nm) Figure 2. UV absorption spectraof 5-MOPand B-MOPodThd photoadducts 2 and 3 in water.
highest mass fragment (mlz = 481) differs by 22 amu from the molecular peak. This is characteristic of a quasimolecular ion [M + Nal+. The fragmentation pattern shows that a significant cleavage of the N-glycosidicbond of the thymidine moiety is taking place (mlz = 342.9). In addition, each spectrum exhibits an intense fragment at mlz = 216 which corresponds to 5-MOP together with a smaller fragment of thymidine at mlz = 243.1. This is characteristic of the splitting of a cyclobutane structure. Nuclear Magnetic Resonance. The four monoadducts 1-4 give similar ‘H NMR spectra with only slightly different features (Table I), suggesting closely related structures for the four nucleosides. The comparison of the aromatic proton chemical shifts of 1-4 with those of 8 - M O P o d T h d cycloadducts (12,16,18)provides support from the involvement of the pyrone ring in the formation of 5 - M O P o d T h d adducts. The resonance signals of H-4’) H-5’) and H-8 protons of the psoralen moiety appear in the low-field region of the ‘H NMR spectra, in agreement with their aromatic character. However, it should be noted that there is a slight upfield shift of these signals with respect to those of 5-MOP (Table I), which is indicative of a partial loss of the aromaticity of the 5-MOP moiety of the adducts 1-4. The lack of H-3 and H-4 signals in this region with respect to the corresponding resonance signals of 5-MOP is explained in terms of saturation of the 3,4-ethylenic bond of the pyrone moiety of the B-MOPdThd adducts 1-4. This provides further evidence for the presence of a cyclobutane structure derived from a [2+2] cycloaddition involving the 3,4double bond of 5-MOP and the 5,6-ethylenic bond of the thymine residue. This was further confirmed by the photoreversal reaction of photoadducts 1-4 upon exposure to 254-nm UV light. The signals of three cyclobutane protons, which resonate between 3.5 and 4.57 ppm appear as two doublets and one doublet of doublets, respectively.
Chem. Res. Toxicol., Vol. 6, No. 6, 1993 861
Photocycloaddition Products t o Thymidine
Table I. LH Chemical Shifts of Photocycloadducts 1-4, Thymidine, and 5-MOP As Inferred from Computer Iterative Analysis (LAOCOON I11 Program) of 400.13-MHz ‘HNMR Spectra in MeZSO-da 1 2 3 4
dThd
5-MOP
CHs 1.41 1.41 1.42 1.43 1.76
H-6 4.30 4.57 4.58 4.50 7.68
H-1’ 5.60 5.55 5.89 6.10 6.50
H-2’ 2.32 1.96 2.02 2.11 2.07
TRANS-ANTI
H-2” 2.06 2.02 1.97 1.96 2.04
H-3’ 4.23 4.12 4.17 4.17 4.21
H-4‘ 3.58 3.75 3.71 3.65 3.74
H-5’ H-5” 3.49 3.57 3.54 3.51 3.57
3.43 3.50 3.51 3.49 3.52
H-3 3.44 3.42 3.61 3.74
H-4 3.91 3.95 3.89 3.91
H-5’ 7.79 7.76 7.79 7.84
H-4‘
H-8
OCH3
7.14 7.15 7.14 7.18
6.89 6.84 6.87 6.91
3.95 4.06 4.09 4.10
6.30
8.18
8.03
7.39
7.33
4.25
TRANS-SYN
CIS-ANTI
CIS-SYN
35 4R 5s 6R
3s 4R 5R 6s
Figure 3. Chemical structure of four possible diastereoisomers of 5-MOP pyrone-side [2+23 photocycloadducts to thymidine.
cis-anli photocfcloadducl 1
3R 4s. SR. 6s
I
80
I
80
7c
53
50
40
60
50
I
I
40
30
I
20
I O
00 wm
Figure 5. 1 D 400.13-MHz 1H NOE spectrum of cis-anti-5MOPodThd photocycloadduct2. (a) Irradiation of the methyl group; (b) irradiation of the methoxy group.
1 I
1
70
30
20
I O
00 ppm
Figure 4. 1D 400.13-MHz 1H NOE spectrum of cis-anti-5MOPdThdphotocycloadduct1. (a) Irradiation of the methyl group; (b) irradiation of the methoxy group.
The assignment of the protons was unambiguously achieved by the preparation of pyrone-side monocycload-
ducts from thymidine specifically deuterated at the H-6 position. In all cases, the partial collapse of the lowest field cyclobutyl proton signal allows its assignment as the H-6 proton of the thymidine moiety (data not shown). The anisotropic effect of the C-4 carbonyl on pyrone H-3 may explain the upfield shift of the signal resonance of the related proton with respect to that of H-4. However, the unambiguous attribution of these signals was inferred from the determination of the stereoconfiguration of the molecule on the basis of NOE experiments (uide infra). Further structural information on photocycloadduct 3 was provided by the assignment of the carbon signals in the 13CNMR spectrum with the exception of the quaternary carbons. This was achieved by 2D lH-13C correlated NMR analysis (data not shown). Confirmation of these assignments was provided by the DEPT sequence (data not shown) which differentiates the secondary carbons from the tertiary and primary carbons. The cyclobutane carbons give signals within the 35-55 ppm region whereas the methyl carbon of the thymidine moiety exhibits the most upfield signal. The resonance signals between 39
Anselmino et al.
862 Chem. Res. Toxicol., Vol. 6, No. 6,1993
Cls-Syn photocycloadduct 4
3R. 4s. 5s. 6R
a
i\ i
cis-syn photocydoadducf 3
35,4R, 5R. 6s
I
I / ” ’
I
80
70
60
50
40
1
I
30
20
10
00
wm
Figure 6. 1D 400.13-MHz lH NOE spectrum of cis-syn-5MOPdThdphotocycloadduct3. (a) Irradiation of the methyl group; (b) irradiation of the methoxy group.
and 90 ppm are characteristic of a furan osidic moiety of a 2’-deoxyribonucleoside structure. The C-2’ is dissimulated under the solvent signal, but the DEPT experiment allows its assignment as a methylenic carbon. Determination of the Stereoconfiguration of the Four Photoadducts 1-4. Eight pairs of diastereoisomers, including the cis-syn, cis-anti, t r a m - s y n , and trans-anti structures, have to be considered (Figure 3). The cis stereochemistry denotes that the furocoumarin and the pyrimidine moiety are located on the same side of the cyclobutane ring. For the pyrone-side adducts, the s y n regiochemistry defines adducts in which C-2 of the pyrone and N-1 of the pyrimidine are bound to adjacent corners of the cyclobutane ring, whereas in the anti regiochemistry these atoms are bound to opposite corners of the cyclobutane ring. For the s y n stereoisomers, the pyrone C-4 carbon is linked to the pyrimidine C-5 carbon, and as a result, H-4 is only coupled with H-3 (30). Therefore, the related signal is expected to be a doublet. In addition, H-3 being vicinal to the H-4 and H-6 protons, the related signal should be a doublet of doublets. When the H-6 proton of the thymidine moiety is deuterated, the H-3 signal collapsed into a doublet, further confirming the proton assignment. In the anti configuration, the pyrone C-3 is located near the methyl group. As a result, the H-3 signal is expected to be a doublet whereas H-4, which is adjacent to H-6 and H-3, should appear as a doublet of doublets. Further confirmation of the stereochemistry of the photoadducts 1-4 was provided by NOE measurements. In a cis stereochemistry, the methyl protons and the H-3, H-4 pyrone protons exhibit dipolar interactions. Therefore, irradiation of the methyl group of the thymine
I
I
I
I
83
70
60
50
40
I
I
I
30
20
10
I
00 PPm
Figure 7. 1D 400.13-MHz lH NOE spectrum of cis-syn-5MOPdThd photocycloadducta4. (a)Irradiation of the methyl group; (b) irradiation of the methoxy group.
moiety is expected to lead to a NOE enhancement of the signals of the cyclobutane protons. In the case of a trans configuration,since the thymidine residue and the psoralen moiety are located on the opposite side of the cyclobutane ring, no dipolar interactions are expected between the methyl group and the 3,4-pyrone-side protons. Upon irradiation of the methyl group of 1-4, significant NOE effects were observed on either the H-3 or the H-4 pyroneside proton, indicating that all the four adducts 1-4 present a cis configuration (Figures 4a-7a). In order to establish the syn or anti regiochemistry, complementary 1D NOE experiments were carried out, based on the irradiation of the methoxy group of the psoralen moiety. Indeed, in a cis-syn regiochemistry, the methoxy group is in the vicinity of the methyl group whereas in a cis-anti regiochemistry, it is located nearby the H-6 proton. Irradiation of the methoxy group of the two photocycloadducts 3 and 4 gave rise to a significant NOE enhancement of the methyl signal of the thymidine residue (Figures 6b and 7b). This confirms the cis-syn stereoconfiguration of 3 and 4. The NOE enhancement observed on the furan H-4’ proton of the psoralen moiety did not provide any structural information on the cycloadducts. However, it is indicative of the validity of the NOE experiment. Irradiation of the methoxy group of the photomonoadducts 1 and 2 did not lead to any detectable NOE enhancement of the pyrimidine methyl
Chem. Res. Toxicol., Vol. 6, No. 6,1993 863
Photocycloaddition Products to Thymidine 0 / cm fi
7 I\
I
/ \ I
,
,
200
e
/
,
,
,
CIS-SYN 3
/CIS-SYN 4
,
,
250
,
,
,
,
,
300
cm
,
,
,
,
h
350 (nm)
,
I
3.0
I
2.0
1.0
ppm
Figure 9. 1D400.13-MHz lH NOEspectrumof 5-MOPodThd photocycloadduct 3. (a) Irradiation of the H-3 proton; (b) irradiation of the H-6 proton.
I
200
250
300
h
350
(nm)
Figure 8. CD spectra of the two cis-syn diastereoisomers of the B-MOPdThd photocycloadducts 3 and 4 in water. group. However, we note an increase in the signal intensity of the H-6 proton of the photocycloadduct 1 which is indicative of an anti configuration (Figure 4b). Sucheffect is not observed for photoadduct 2, probably because the methoxy group is orientated toward the furan side (Figure 5b). Further confirmation of the diastereoisomeric relationship between the two pairs of photoadducts 1 and 2 and photoadducts 3 and 4, respectively, was provided by consideration of their CD features (Figure 8). Each pair of diastereoisomers exhibits quite similar but opposite n-?r* transitions centered about 230 and 260 nm for 1and 2 and centered about 220 and 240 nm for 3 and 4, respectively. In conclusion,from the information provided by UV, FAB-MS, lH NMR, and NOE difference spectra measurement, it is reasonable to assign a cis-anti configuration to photocycloadducts 1 and 2 and a cis-syn configuration to photocycloadducts 3 and 4. Determination of the Absolute Configuration. (A) cis-anti Photoadducts 1 and 2. The absolute configuration of the cyclobutane carbons of both photoadducts 1 and 2 was inferred from NOE experiments. It is worth noting that irradiation of the methoxy group of photoadduct 1gave rise to an enhancement of the signal of the anomeric proton and H-2’ and H-2” osidic protons. In the case of photoadduct 2, the NOE enhancement was only observed for the anomeric proton. These observations may be rationalized in terms of a 3R,4S,5R,6Sconfiguration
for 1 and a 3S,4R,5S,6R configuration for 2, taking into consideration that both photoadducts exhibit a preferential anti or high anti conformation (vide infra). (B) cis-syn Photoadducts 3 and 4. Similarly, NOE experiments were used to assign the absolute configuration of the cyclobutane carbons of both photoadducts 3 and 4. For the photoadduct 3, irradiation of the H-6proton of the thymine moiety led to a NOE enhancement of the anomeric, H-2’ and H-2” protons (Figure 9) whereas in the case of photoadduct 4 only a weak effect is observed for the H-2’ proton (data not shown). Such observations are indicative of a 3S,4R,5R,6S configuration for 3 and a 3R,4S,5S,6R configuration for 4. Again, it has to be considered that the base of both photocycloadducts 3 and 4 is in a preferential anti or high anti orientation about the N-glycosidic bond (vide infra). Conformational Analysis of the Photoadducts 1-4. (A) cis-anti Adducts (1 and 2). One important aspect of nucleoside conformation is the osidic ring puckering. This is usually referred to as a dynamic equilibrium between the two puckered (2-2’ endo and C-3’ endo forms (31,32). The C-2’ endo form is the conformation in which C-2’ is displaced toward C-5’ from the plane of the other atoms of the sugar ring. The coupling constant values of vicinal protons are related via the Karplus equation to the apparent ~ H value. H Therefore, J 1 f 2 ~and J3’4’ coupling constants depend on the equilibrium between C-2’ endo and C-3’ endo forms. Measurement of these coupling constants (Table 11)showed that the dynamic equilibrium between the two sets of sugar conformers is displaced for both monoadducts 1and 2 toward the C-2’ endo conformer (1: 71 % ; 2: 657’% ). These have to be compared to those of thymidine (C-2’ endo = 74%) (33, 34) (Table 111).In addition, irradiation of the methoxy group gave rise to a
864 Chem. Res. Toxicol., Vol. 6, No. 6, 1993
Anselmino et al.
Table 11. Coupling Constant Values of Photocycloadducts 1-4, Thymidine, and 6-MOP As Inferred from Computer Iterative Analyses (LAOCOON I11 Program) of 400.13-MHz 1H NMR Swctra in Med73O-de 1 2
3 4 dThd
7.7 6.4 7.3 8.7 7.8
6.7 6.2 6.11 5.8 6.0
-13.8 -13.3 -13.4 -13.6
-13.2
6.4 6.4 6.2 5.7 6.1
3.5 3.9 3.0 2.6 3.0
3.0 3.5 2.6 2.5 2.7
5-MOP
4.8 4.5 4.5 4.0
-11.9 -12.0 -11.7 -11.8
8.5 9.7 9.8 2.4
Table 111. Conformational Features of Photocycloadducts 1-4 in MelSO As Inferred from lH NMR Measurements % C-2’ endoa % EE rotamersb % et + te rotamersf 1 2 3 4 dThd
4.1 4.3 4.9 3.9
71 65 74 78 74
42 50 50 46 60
58 50 50 53 40
+
a % (3-2’ endo = (100)J1tr/(Jl!r JSV) (33,34). % gg = [13.7 (54’6’ + J4~~*)1(100/9.7) (33, 34). % gt + tg = 100 - (% gg). Abbreviations: gg,gauche-gauche; gt,gauche-trans; tg, trans-gauche.
NOE enhancement of the H-1’, H-2’, and H-2” proton signals for the photoadduct 1 whereas in the case of photoadduct 2, only a weak NOE was observed for the anomeric proton (Figures 4b and 5b). These indicate that the base moiety of both diastereoisomers adopts a preferentialanti or high anti conformation with respect to the N-glycosidic bond. However, it should be noted that the two cis-anti photoadducts 1and 2 show slight differences in the chemical shift value of H-6 and osidic protons. Such differences are likely to reflect small changes in the N-glycosidic angle torsion (3.536). Distribution of staggered conformers about the C4’-C5’ bond was estimated from considering the experimental sum C = J4at + J4y~. This allowed the determination of the population of gauche-gauche (gg)rotamers and the combined population of gauche-trans (gt) and trans-gauche (tg) rotamers (1: gg = 42%; 2: gg = 50%;dThd: gg = 60%) (33). It is interesting to note a significant decrease in the percentage of gg rotamer population with respect to that of thymidine, the effect being more important for photoadduct 1. This may be rationalized in terms of a destabilization effect arising from repulsive interactions between the 5’-OH group and the C-2 carbonyl, as already observed for pyrimidine nucleosides bearing a bulky group at C-6 (36). Such effects are not observed for either thymidine or photoadducts 2. (B) cis-syn Adducts (3 and 4). For both cis-syn dThd < ;;>5-MOP photoadducts (3 and 4) the dynamic equilibrium between the two sets of C-2’ endo and C-3’ endo forms is also displaced toward the C-2’ endo conformer (3: 63%; 4: 78%) (Table 111). The ‘H NMR features also suggest for both adducts a preferential anti or high anti conformation of the oeidic moiety around the glycosidic bond. Also, the rotamer gg is predominant for both photoproducts 3 and 4 (3: 50 % ;4: 45 % 1. However, there is a significant decrease in the population of gg rotamer, for photoadducts 3 and 4 as compared to that of thymidine. This is wplained by the spatial proximity of the 5’-OH osidic grot p with either the carbonyl group at position 2 of the thymidine moiety for 3 or the lactone function of the psoralen moiety for 4.
Conclusion The photoreaction of 5-MOP with thymidine under drystate condition leads to the formation of five main
7.6 9.1 9.2
2.3 2.3 2.4 9.8
nucleoside photoproducts including the four [2+2] photocycloadducta which were isolated and fully characterized as 5-MOP pyrone-side diastereoisomers. The stereoconfiguration of the two pairs of cis-syn and cis-anti diastereoisomers was established on the basis of extensive lH and 13CNMR analyses. Work is currently in progress in our laboratory aimed at searching for the formation of such photoadducts 1-4 within DNA by using the chromatographic and physicochemical properties now available.
Acknowledgment. This work has been supported in part by L’OREAL (Ph.D. fellowship to C.A.). The authors thank J. Ulrich (Laboratoire des Prothines, Institut de Biologie Structurale) for FAB mass spectrometry measurements. The contribution of F. Sarrazin for NMR measurements is also greatly acknowledged. References (1) Pathack, M. A., and Fitzpatrick, T. B. (1992) The evolution of photochemistry with psoralen and UVA. J. Photochem. Photobiol., E: Biol. 14, 3-9. (2) Musajo, L., and Rodighiero, G. (1962) The skin photosensitizing furocoumarins. Experientia 18, 153-200. (3) Midden, W. R. (1988) Chemical mechanisms of the bioeffects of furocoumarins: The role of reactions with proteins, lipids, and other cellular constituents. In Psoralen DNA photobiology (Casparro, F., Ed.) Vol. 2, pp 1-49, CRC Press, Boca Raton, FL. (4) Averbeck, D. (1989) Recent advances in psoralen phototoxicity mechanism. Photochem. Photobiol. 50,859-882. (5) Cadet, J., Vigny, P., and Midden, W. R. (1990) Photoreactions of furocoumarinswith biomolecules.J . Photochem. Photobiol. B: Biol. 6,197-206. (6) Schiavon, O., and Veronese, F. M. (1986) Extensive crosslinking between subunita of oligomeric proteins induced by furocoumarins plus W A irradiation. Photochem. Photobiol. 43, 243-246. (7) Averbeck, D., and Papadopoulo, D. (1986) Genetic effects of DNA and diadducta photoinduced by furocoumarins in eukariotic cells. In The role of nucleic acid adducts in carcinogenesis and mutagenesis (Singer, B., and Bartach, H., Eds.) Vol. 70, pp 299-311, IARC Scientific Publications, Lyon, France. (8) Averbeck, D., Dardalhon, M., and Magana-Schwencke, N. (1990) Repair of furocoumarin plus UVA induced damage and mutagenic consequencesin eukaryotic cells. J. Photochem. Photobiol., B: Biol. 6,221-236. (9) Song, P. S., and Tapley, K. J. (1979) Photochemistry and photobiology of psoralens. Photochem. Photobiol. 29,1177-1197. (10) Ben-Hur, E., and Song, P.4. (1984) The photochemistry and photobiology of furocoumarins. Adu. Radiat. Biol. 11, 131-167. (11) Hearst, J. (1989) Photochemistry of psoralens. Chem. Res. Toxicol. 2, 69-75. (12) Cadet, J., and Vigny, P. (1990) The photochemistry of nucleic acid. I n Bioorganic photochemistry (Morrison, H., Ed.) Vol 1, pp 1-272, Wiley-Interscience, New York. (13) Cadet, J., Voituriez, L., Ulrich, J., Joshi P. C., and Wang, S. Y. (1984) Isolation and characterisation of the mono-heterodimers of 8-MOP and thymidine involving the pyrone moiety. Photobiochem. Photobiophys. 8, 35-49. (14) Cadet, J., Voituriez, L., Gaboriau, F., and Vigny, P. (1983) Characterization of photocycloaddition products from reaction between thymidine and the monofunctional 3-CPs. Photochem. Photobiol. 37, 363-371. (15) Land, E. J., Rushton, A. P., Beddoes, R. L., Bruce, J. M., Cemik, R. J., Dawson, C., and Mills, 0. S. (1982) A [2+21 photo-adduct of 8-MOP and thymidine: X-ray crystal structure; a model for the reaction of psoralens with DNA in the phototherapy of psoriasis. J. Chem. SOC.,Chem. Commun., 22-23.
Photocycloaddition Products to Thymidine (16) Cadet, J., Voituriez, L., Nardin, R., Viari, A,, and Vigny, P. (1988) A new class of psoralen photoadduds to DNA components: isolation and characterization of 8-MOP adducta to the osidic moiety of 2‘deoxyadenosine. J. Photochem. Photobiol., B: Biol. 2, 321-339. (17) Shim, S-C., and Kim, Y . 4 . (1983) Photoreaction of &MOP with thymidine. Photochem. Photobiol. 38,265-271. (18) Shim,S.-C., andChoi, S.-J.(1990) Photoreactionof 4,5’,&TMPwith adenosine. Photochem. Photobiol. 62, 669-675. (19) Kanne, D., Straub, K., Rapoport, H., and Hearst, J. E. (1982) Psoralen-deoxyribonucleic acid photoreaction. Characterization of the monoaddition products from 8-MOP and 4,5’,8-TMP. Biochemistry 21, 861-871. (20) Caffieri, S.,Lucchini, V., Rodighiero, P., Miolo, G., and Dall’Acqua, F. (1988) 3,4 and4’,5’Photocycloadduds between4’-methylangelicin and thymine from DNA. Photochem. Photobiol. 48,573-577. (21) Cadet, J., Anselmino, C., Douki, T., and Voituriez, L. (1992) Photochemistry of nucleic acids in cells. J. Photochem. Photobiol. B: Biol. 15, 277-298. (22) Piette, J. (1992) Molecular analysis of mutations induced by 4‘hydroxymethyl-4,5’,8-trimethylpsoralenand UVA in the mouse HPRT gene. J. Photochem. Photobiol. B: Biol. 12, 37-55. (23) Yun, M. H., Choi, S. J., and Shim, C. (1992) A novel photoadduct of 4,5’,8-TMP and adenosine. Photochem. Photobiol. 55,457-460. (24) Demaret, J. P., Brunie, S., Ballini, J. P., and Vigny, P. (1989) Geometry of intercalation of psoralen in DNA approached by molecular mechanics. Photochem. Photobiol. 50, 7-21. (25) DBcout, J. L., Huart, G., and Lhomme, J. (1988) Interactions and photoreactivity of 5-alkoxypsoralens with thymine. A model approach. J. Photochem. Photobiol. B: Biol. 48,583-596. (26) DBcout, J. L., Lekchiri, Y., and Lhomme, J. (1992) The thymidinepsoralen photoaddition. A model approach. Photochem. Photobiol. 55,639-645. (27) Rabi, A,, and Fox, J. J. (1973) Nucleosides. LXXIX. Facile basecatalysed hydrogen isotope labeling at position 6 of pyrimidine nucleosides. J. Am. Chem. SOC. 95, 1628-1632.
Chem. Res. Toxicol., Vol. 6, No. 6, 1993 865 (28) Wood, D. J., Ogilvie,K. K., and Hruaka,F. E. (1975) Proton magnetic resonance studies of dinucleotide conformation in aaueow solution. 2’-Deoxythymidylyl-(3’,5’)-2’-deoxy-3’-thymidylate~d(TpTp). Can. J. Chem. 63,2781-2790. (29) Anselmino, C., and Cadet, J. (1991) A new photoreaction of 5-MOP with thymidine: Characterizationof a paoralen pyrone-side monoadduct involving the pyrimidine methyl group. 4th Congress of the European Society for Photobiology, Amsterdam, The Netherlands, D-1,100. (30) Vigny, P., Gaboriau, F., Voituriez, L., and Cadet, J. (1985) Chemical structure of psoralen-nucleic acid photoadduds. Biochimie 67,317325. (31) Altona, C., and Sundaralingam, M. (1972) Conformational analyis of the sugar ring in nucleosides and nucleotides. A new description using the concept of pseudorotation. J. Am. Chem. SOC. 94,82058212. (32) Altona, C., and Sundaralingam, M. (1973) Conformational analysis of the sugar ring in nucleosides and nucleotides. Improved method for the interpretation of proton magnetic resonance coupling constants. J. Am. Chem. SOC. 95,2333-2344. (33) Lee, C-H., and Sarma, H. (1976) Aqueous solution conformation of rigid nucleosides and nucleotides. J. Am. Chem. SOC. 98, 35413548. (34) Ramaawany,H.,andSarma,H., (1974)TheuseofNMRspectroscopy to determine nucleic acid geometry and dynamics. In Nucleic acid geometry and dynamics (Sarma, R. H., Ed.) pp 1-45, University Park Press, Baltimore. (35) Schweizer, M. P., Banta, E. B., Witkowski, J. T., and Robins R. K. (1973) Determination of pyrimidine nucleoside syn anti conformational preference in solution by 1BC NMR. J. Am. Chem. Soe. 95, 377G3778. (36) Cadet, J., Taleb, C., Remin, M., Niemczura, W. P., and Hruaka, F. E. (1980) Conformational studies of a- and ,%pyrimidine 2’deoxyribonucleosides in the syn and anti conformation. Biochim. Biophys. Acta 60, 435-445.
