.l. Photochem. PhotobioL B: BioL, 14 (1992) 65-79 65

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

Srinivas S. Sastry, H. Peter Spielmann, Tammy J. Dwyer, David E. Wemmer and John E. Hearst+ Department of Chemistry, University of California and the Division of Chemical Biodynamics, Lawrence Berkeley Laboratory, Berkeley, CA 94720 (USA)

(Received November 26, 1991; accepted January 17, 1992)

Abstract

We have developed novel methods for the preparation of multimicromole quantities of extremely pure, uniquely photoadducted psoralen-DNA cross-links, furan-side monoadducts and pyrone-side monoadducts. Psoralen cross-linked and furan-side monoadducted DNA were produced by employing high intensity argon ion and krypton ion lasers as light sources. Pyrone-side monoadducts were prepared by base-catalyzed photoreversal of psoralen cross- links. The various psoralen-adducted DNA oligomers were efficiently purified by high performance liquid chromatography. These methods have permitted us to synthesize 4 pm01 each of a self-complementary 8-mer d(GCGTACGC) 4’-(hydroxymethyl)-4,5’,8-tri- methylpsoralen (HMT) furan-side monoadduct and HMT cross-link. Preliminary nuclear magnetic resonance (NMR) data on the HMT cross-linked 8-mer d(GCGTACGC) have been obtained which confirmed the presence of the diadducted psoralen at the unique 5’TpA3’ site. NMR data obtained from the 8-mer furan-side monoadduct revealed that the psoralen molecule is intercalated into the DNA double helix. Preliminary crystals of 8-mer cross-linked DNA molecule have been grown. Conditions for the growth of X-ray diffraction-quality crystals and the further analysis of these crystals are now in progress.

Keywcti: Lasers, psoralen, photochemistry, DNA photoadducts, high performance liquid chromatography, nuclear magnetic resonance.

1. Introduction

Psoralens are linear fnrocoumarins that photochemically alkylate nucleic acids. DNA is subject to direct photochemical damage when irradiated with short-wavelength UV radiation (UVB A < 310 nm). Psoralens extend the range of wavelengths that can photochemically damage DNA into the long-wavelength W (WA A= 310-400 nm) and visible regions of the spectrum. Psoralens react primarily with thymidine and to a lesser extent with the cytosine in DNA (Fig. 1). Damage to DNA is detrimental to an organism’s ability to survive and replicate. There are a number of enzyme systems in both prokaryotes and eukaryotes that repair damaged DNA. These enzyme systems recognize and act on psoralen-DNA adducts by the excision repair mechanism [l, 21. The Escherichia coli repair system UVR ABC recognizes and repairs psoralen-damaged

‘Author to whom correspondence should be addressed.

loll-1344/92/$5.00 0 1992 - Elsevier Sequoia. All rights reserved

66

HO /

dT-HMT furan-side monoadduct, cis-syn conformation

dT-HMT pyrone-side monoadduct, cis-syn conformation

HO

dT-HMT-dT diadduct. cis-syn conformation

Fig. 1. Structure of the three main types of HMT-thymidine adduct formed during the photoreaction of HMT with DNA.

DNA by excising a DNA oligomer patch that contains the adduct. The exact size of the oligo excised by the enzyme complex is different for the three main types of psoralen adduct @ran-side monoadduct (MM), pyrone-side monoadduct (MAPS) and cross-link adduct). Psoralens have been used to study the bacterial chromatin structure [3, 41 and the in vivo and in vitro RNA secondary structure, leading most recently to a new distance geometry model for the structure of 165 rRNA [5]. A detailed literature survey on the uses of psoralens as probes for nucleic acid structure and function has been published [6]. Stable transcriptional elongation complexes arrested by psoralen adducts at specific sites on a DNA template have been generated [7-lo]. Because psoralen photochemistry is highly specific for nucleic acids, these drugs have been employed in a sterilization procedure to avoid polymerase chain reaction (PCR) carry-over [ll, 121, the sterilization of human blood fractions for transfusion [13], the inactivation of viruses [14-161 and, in photophoresis, an extracorporeal photoinactivation procedure applied to patients suffering from leukemia and other white-blood-cell proliferation diseases. There is a clear need to prepare site-specifically-psoralen-adducted DNA oligomers to understand these processes further. We report here methods that allow the synthesis of micromole quantities of the three main types of psoralen- adducted DNA oligomer.

Large-scale synthetic methods for the production of psoralen monoadducts and diadducts have relied on Hg-Xe arc lamps and black-ray light sources to photoreact

67

psoralen with DNA [17, 181. The photochemical efficiency with these light sources is poor and often results in the production of multiple photoproducts in addition to the three main types of psoralen adduct. Denaturing polyacrylamide gel electrophoresis (PAGE), followed by the elution of the DNA from gel slices, is the most common method for the separation of photoadducted DNA oligos from other components in the reaction mixture after photochemistry. This method is inefficient because there are losses of the valuable adducted DNA during elution from the gel. Contaminating photoproducts and other impurities released from polyacrylamide gels are also major drawbacks of these methods. Preparative electrophoretic methods limit the amounts of psoralen-adducted DNA that can be prepared to quantities of 200 pg at one time. These conventional methods are obviously unsuitable for production of large amounts (multimicromoles) of extremely pure psoralen-adducted DNA molecules required for studies employing X-ray crystallographic or nuclear magnetic resonance (NMR) tech- niques.

The synthetic procedures for site-specifically-modified DNA oligonucleotides which have been developed in our laboratory employ two innovations which are essential for the synthesis of large quantities of high quality photo-adducted DNA oligonucleotides. These procedures are (1) the use of lasers to deliver intense monochromatic light so that the products of photochemistry lack many of the byproducts which are commonly found with broad spectrum light sources and (2) the use of both reverse phase and ion exchange high performance liquid chromatography (HPLC) in the purification of the psoralen-modified oligonucleotide following photochemistry. Psoralen-adducted oligodeoxynucleotides using these synthetic procedures have been shown to be ex- ceptionally pure by high resolution two-dimensional (2D) NMR. Preliminary crystals have also been obtained for the 4’-(hydroxymethyl)-4,5’-&trimethylpsoralen (HMT) cross-linked oligodeoxynucleotide 8-mer d(GCGTACGC)*.

2. Materials and methods

2.1. Materials HMT was a gift from HRI Associates Inc. (Concord, CA). Stock solutions (about

0.5 M) of HMT were prepared in dimethyl sulfoxide. All the oligonucleotides were synthesized by David Koh (Department of Chemistry, University of California, Berkeley, CA) on an Applied Biosystem automated DNA synthesizer using the phosphoroamadite method. [y 3zP] adenosine triphosphate (ATP) (specific activity, greater than 6000 Ci mmol-‘) was obtained from NEN Du Pont (Wilmington, DE). T4 polynucleotide kinase was purchased from New England Biolabs (Beverly, MA). Argon ion and krypton ion lasers were purchased from SpectraPhysics (Mountain View, CA). Quartz flow cells were purchased from NSG Precision Cells Inc. (Farmingdale, NY). HPLC instrumentation was purchased from Beckman (Palo Alto, CA) and Rainin Instruments (Emeryville, CA). Autoradiography was conducted using Kodak XAR5 film. 2.2. Oligonucleotide synthesis

The DNA oligomers were synthesized on an automated Applied Biosystems DNA synthesizer with the 5’-trityl group on and deprotected using standard methods. Full length product was then purified by HPLC using the Rainin on-column detritylation method [19]. The purity of the DNA was assayed by denaturing gel electrophoresis, W spectrophotometry and one-dimensional (1D) and 2D ‘H NMR. 2.3. K&ton ion laser irradiation

Irradiations were done with a SpectraPhysics 2020 krypton laser operating in broad band mode at 406.7 nm and 413 nm at 350 mW. The oligonucleotides were

68

dissolved in 35 ml of 150 mM NaCl+ 10 mM MgCl, + 1 mM ethylenediamine tetraacetic acid (EDTA) + 15 mM azide + 1.6 X 10 -’ M HMT. The DNA duplex concentration was 4x10-’ M. This solution was introduced into a quartz cuvette of 10 cm path length and stirred during irradiation. The cwette was placed in the laser beam and between two dielectric mirrors optimized for reflectivity at 406.7 nm. The mirrors were adjusted to cause the laser beam to be reflected for a total of eight passes through the sample. During the irradiation, aliquots (5 ~1) of the DNA solution were withdrawn from the cwette for gel analysis. After the irradiation was complete, the solution was brought to 200 mM NaCl with 6.1 M NaCl, and three volumes of absolute ethanol (EtOH) were added. The solution was cooled overnight at - 20 “C and the precipitate collected by centrifugation at 25 OOOg for 2 h. The supernatant was removed and the pellet washed with 95% EtOH and dried in ‘uucuo.

2.4. Argon ion laser irradiation The oligonucleotides were taken up in 100-500 ml of annealing buffer which

consisted of 50 mM T&Cl (pH 7.5) + 10 mM MgCl,+ 100 mM NaCl+ 1 mM EDTA+ 15 mM sodium axide + 1.6 X 10m4 M HMT for the 8-mer DNA. The DNA duplex con- centration was 4 X lo-’ M. The DNA solution was heated at 95 “C for 30 min in a water bath and cooled slowly over a period of 3-6 h to room temperature. The HMT + DNA solution was pumped through a jacketed quartz flow cell (NSG Precision Cells, Inc., NY; 10 mm path length) that was positioned in the path of the laser light. A cylindrical quartz lens was used to spread the light beam such that the entire area of the flow cell received the maximum amount of laser light. The output of the laser was adjusted to 5-6 W. The reaction mixture was pumped at room temperature (23-26 “C) through the cell at a rate of 10 ml min-’ using a peristaltic pump. After one cycle of irradiation, HMT was added to 1.6 X 10B4 M HMT and the DNA+HMT solution was pumped through the apparatus for a second cycle of irradiation. The reaction volume was reduced to 50 ml by lyophilixation, and dialyzed against 500-lOOO- fold excess water. After dialysis the reaction mixture was lyophilized to dryness.

2.5. Base-catalyred reversal A DNA 1Zmer (5’-GAA-GCT-ACG-AGC-3’), a complementary DNA 8-mer (5’-

TCG-TAG-CT-3’) and HMT as the psoralen were irradiated using an argon ion laser operating at 364 nm to form cross-linked molecules in 50% yield atter HPLC purification (see above). 1 x lo-’ mol of HPLC-purified 1Zmer HMT-8 cross-link was dissolved in 200 ~1 of 0.1 N NaOH or KOH and heated to 90 “C for 30 min. The reaction mixture was cooled to room temperature and neutralized by the addition of 10 ~1 of 2 N HCl added together with 50 ~1 of 1 M Tris (pH 7.5). The neutralized reaction mixture was applied to a reverse phase Cra HPLC column and the pyrone-side adducts separated from unmodified DNA and cross-link (see below).

2.6. High p@ormance liquid chromatography purification 2.6.1. Reverse phase high performance liquid chromatography analysis and preparative purification Analytical samples of the modified DNA were applied to a 4.6 mm X 25 cm reverse

phase Crs column of 60 A pore size for method develo ment. For preparative samples a 24.5 mmX 25 cm reverse phase C,s column of 60 1 pore size was employed. The products of the reaction were eluted with a linear acetonitrile (MeCN) gradient in 100 mM triethylammonium acetate (TEAA) @H 6.5) over a period of 80 min (flow rate, 1.0 ml min-l analytical, 8.0 ml min-’ preparative). The percentage MeCN was

69

changed from 9.5% to 17.5% from time 5 min to time 85 min. The fractions of interest were collected and lyophiliied. The residue was resuspended in a minimum volume of tris-ethylenediaminetetraacetic acid (TE) and adjusted to 200 mM NaCl+ 10 mM MgCl, and then three volumes of absolute EtOH were added. The solution was cooled overnight at -20 “C and the precipitate collected by centrifugation at 16 OOOg for 35 min. The supematant was removed and the pellet washed with 95% EtOH and dried in vacua. Quantification of the main products was done by UV spectrophotometry.

2.6.2. Anion exchange high performance liquid chromatography analysk and preparative purification The DNA was dissolved in 1.0-1.5 ml 30% MeCN and 20 mM sodium acetate

(pH 5.5) and applied to a Nucleogen 60-7 diethylaminoethyl (DEAE) anion exchange HPLC column (Macherly-Nagel; column dimensions, 4 mmx 125 mm). The products of the reaction were eluted with a linear 0.0 M to 1.0 M KCl gradient in a buffer of 20 mM NaOAc (pH 5.5)+30% acetonitrile over a period of 200 min (flow rate, 1.0 ml min-‘). The fractions that were enriched in cross-linked DNA were then dialyzed extensively against water as described above. The DNA was then lyophilized to dryness.

2.7. Gel analysis of high performance liquid chromatography fractions The oligonucleotide fractions isolated by HPLC were resuspended in 0.5 ml TE.

1 ~1 of each of these fractions was 5 ’ 32P end labeled using T4 polynucleotide kinase and [y 32P] ATP [20]. An aliquot of each of these reactions was loaded onto a 20 cmx 40 cmxO.4 mm 24% 19~1 acrylamide:bis 7 M urea gel and electrophoresed until bromophenol blue was at the bottom of the gel. The composition of the fractions was visualized by autoradiography.

2.8. Nuclear magnetic resonance spectroscopy NMR samples were prepared by dissolving approximately 2 mg of the HMT-

d(GCGTACGC)2 cross-link in 0.5 ml of DzO. NMR spectra were acquired on a General Electric GN-500 spectrometer. All spectra were recorded at 25 “C. 2D NOESY (nuclear Qverhauser effect spectroscopy) spectra were obtained using the pulse sequence delay-90”-t1-900-r~-900-& in the phase-sensitive mode using TPPI (time proportion phase incrementation). The spectra were collected into 1024 data points in each block using a spectral width of 5000 I-Ix and a mixing time of 5CX?50 ms. Typically, 512 tl experiments were collected and zero filled to 1 K. For each tl value, 64 scans were signal averaged, with a recycle time of 2 s. Prior to Fourier transformation, the data were apodixed using a squared sine bell with a 90” phase shift. Double-quantum- filtered COSY (correlation spectroscopy) experiments were obtained using the pulse sequence delay-90°-t1-900-9000-t~ in the phase-sensitive mode using TPPI. Spectral parameters consisted of 2048 data points in t2 using a 5000 Hz spectral width, and 512 blocks collected in tl using 64 scans per block. The data were apodized in both dimensions using a skewed sine bell with a 60” phase shift.

3. Results

3.1. Laser synthesis of j&ran-side monoadduct Synthesis of large quantities of furan-side monoadducted DNA oligomer was

achieved by irradiating the HMT+ DNA solution with ‘350 mW, 406.7 nm light from a krypton ion laser (see flow chart in Fig. 2). The absorption coefficient of psoralen

Phase

HPLC

Reverse Phase

Phase

Fig. 2. HMT furan-side monoadducted oligonucleotide synthetic scheme.

at wavelengths longer than 400 nm is only between 1 and 20 (1 mol-’ cm-‘). In order to get efficient photochemistry at these wavelengths, a high intensity monochromatic light source is necessary. The laser radiation specifically excites the psoralen to form MAfs with DNA. The resulting monoadducts do not absorb light at the irradiation wavelength, allowing them to accumulate in the reaction without the attendant formation of a cross-link. A laser is desirable for this procedure because of its high intensity light, low beam divergence and, most importantly, the monochromaticity of the beam. Because psoralen has such a low extinction coefficient at this wavelength, a long-path- length reaction cell is necessary to maximize the efficient use of the photons generated. A laser is ideal for long-path-length irradiations because the light that is emitted is highly collimated, so that there is little loss of light intensity due to beam divergence. To maximize the efficient use of photons, a multipass irradiation apparatus was used. This reaction produced a mixture of furan-side monoadducted molecules, and unmodified DNA, necessitating the use of a multistep HPLC purification procedure to arrive at material of sufficient purity to do NMR experiments.

3.2. High perfomzance liquid chromatography purification of furan-side monoadduct The monoadducted oligonucleotides have a significantly different retention time

from the parent oligonucleotides when eluted from a Cl8 reverse phase HPLC column (Fig. 3). Two successive rounds of reverse phase HPLC were used to purify the MAf from unmodified 8-mer as shown in Fig. 3. The separation is primarily due to the increased hydrophobic@ of the psoralen-containing oligonucleotide. The monoadducted oligonucleotide that is isolated in this fashion is single stranded. In order to obtain a homogeneous NMR sample of double-stranded heteroduplex, the Hm-modified DNA must be hybridized with its unmodified complement. The single-stranded 8-mer MAf was then titrated with the unmodified 8-mer to give duplex DNA. The titration was monitored by observing the appearance of two resonances assigned to the adenine C&H protons in the lH NMR at 600 MHz. These two peaks resonate with chemical shifts different from that of the adenine C&H proton in the unmodified duplex DNA. The MAf heteroduplex is approximately 1.7 kcal mol -’ more stable than the unmodified DNA duplex, and 3.5 kcal mol-’ more stable than the modified DNA homoduplex

Pll*

3.3. Laser synthesis of cws-link Figure 4 shows a flow chart scheme for the synthesis of micromole quantities of

8-mer self-complementary oligonucleotide cross-links. We used an argon ion laser

71

2

1

a

,

8 mer

8 MF

10 20 30 40

15 0.5

12.5 0.25

10 0

8 MAf

/i / / / / / _) / / ./ I I I

0 10 20 30

ae E

P

12.5

10

,

Minutes After Injection

Fig. 3. Crs reverse phase HPLC protiles and chromatogram of the purification of 8-mer MAf. See Section 2 for details. Because of the large amounts of DNA loaded on the Crs column, the absorbance was monitored at 300 nm instead of at 260 nm, so that the peaks would stay within the scale (less than 2) of sensitivity of our spcctrophotometer. The elution was also monitored at 330 nm where only the h4Af absorbs to give positive identification of the fractions containing the adduct.

Phase

HPLC

Reverse Phase

I 1 HPLC - A 1 PureMAp I-

Reverse I MAp I \ Phase

Fig. 4. HMT cross-linked oligonucleotide synthetic scheme: XL, cross-link.

operating at 364 mn to react psoralen (HIHT) with the DNA oligomer d(GCGTACGC) to form cross-linked 8-mer. The HMT + DNA solution was pumped through a jacketed quartz flow cell that was positioned in the path of the laser light. Both HhIT and the MAf absorb light efficiently at 364 nm wavelength. The MAf is an intermediate

72

on the path to the interstrand diadduct. The MAf does not accumulate in the reaction mixture because the quantum yield for its conversion to cross-link is four times larger than the quantum yield for formation of monoadduct from HMT [22]. The high intensity radiation from the laser allows for short reaction times and high yields of cross-linked molecules. This reaction produced a mixture of cross-linked molecules, pyrone-side monoadducted molecules and unmodified DNA. Because of the high level of photoaddition during this reaction, a small quantity of cross-links are formed at sites in the molecule other than at the central 5’-TpA-3’ site. This complex mixture of products necessitates the use of a multistep HPLC purification procedure to arrive at material of sufficient purity to perform NMR experiments.

3.4. High pefomzance liquid chromatography purification of cross-link The reaction mixture is concentrated and dialyzed after the photoreaction to

remove the salts and unreacted and photodamaged HMT from the sample. The purification of micromole quantities of cross-linked material away from other components in the reaction mixture was efficiently achieved by employing Cu reverse phase and anion exchange HPLC. The majority of the unmodified 8-mer was separated from psoralen-adducted DNA by an initial round of Cis reverse phase chromatography. In this system, both the cross-linked molecules and the pyrone-side adducted molecules co-elute (Fig. 5(a)). The psoralen-adducted molecules were collected and subjected to a second round of reverse phase HPLC. In the second round of chromatography, the amount of material loaded on the column was much smaller and separation was much more effective between the three species in the mixture. The MAp was contaminated with a small amount of cross-link (Fig. 5(b)). The fractions containing cross-linked molecules were pooled, concentrated and chromatographed using anion exchange HPLC (Fig. 5(c)). Anion exchange chromatography is especially useful in the separation of cross-linked molecules from single-stranded DNA. The higher charge density of a cross-linked molecule from its phosphate groups causes it to have significantly different chromatographic characteristics from the single-stranded DNA from which it was made. A final round of reverse phase chromatography was performed on the cross-linked DNA to desalt it. This completed the separation of all the components of the reaction mixture (Fig. 5(c)). The unmodified 8-mer was removed in the initial reverse phase HPLC steps and the various cross-link isomers were separated from each other and the single-stranded MAf in the anion exchange HPLC step. Utilizing this laser light source, we have prepared 4 pmol of purified cross-link formed between the DNA octanucleotide d(GCGTACGC) and Hh4T.

3.5. Preparation of pyrone-side adduct The third type of common psoralen DNA photochemical adduct is the MAp.

MAPS do not absorb light at wavelengths greater than 310 nm. Under typical UVA conditions they do not absorb light and go on to form cross-links. They are formed in relatively low yield during the forward photochemical cross-linking reaction. In fact, the MAp is not detected when the krypton laser is used to synthesize monoadducts at long wavelengths (data not shown). When the argon laser is used to cross-link between HMT and the 8-mer 5’-GCGTACGC-3’, pyrone-side monoadducted oligomers account for less than 5% of the total modified DNA. We have developed a method for producing micromole quantities of MAp using a base-catalyzed reversal of the furan-side cyclobutane ring of psoralen cross-linked DNA (see flow chart for the preparative scheme in Fig. 6).

73

r 8XL I

I

2.0 t

8Mpyt8Xl.

n

409 ,

, /

/ /

1.0 - / ,

, /

8.mer , ’ ,

20

TCN 15

0 6 10 15 20 26 30 35

(4 Mi” ,ner I”,ection

Fig. 5. (a), (b) Reverse phase Crs HPLC purification of &mer+HMT mixture irradiated with argon laser light. A Dynamax 60 A Cts column (21.4 cmX25 cm) was used in both rounds of purification. The flow rate was 8 ml min-’ fraction- *. The following elution profile was used: from 0 mm to 10 min an isocratic 10% MeCN in 0.1 M TRAA (pH 6.5) was used. The gradient was then changed from 10% to 20% MeCN over a period of 80 mm. (c) DEAR chromatography after two rounds of Cra HPLC. The column used was a Nucleogen 60-7 DEAR 60 A (4 mm X 125 mm). Buffer A was 0.02 M sodium acetate @H 5.5)+30% MeCN. Buffer B was 0.02 M Na acetate (pH 5.5) +30% MeCN + 1 M KCl. The KC1 gradient was started at 400 mM at the time of sample injection. The gradient was changed at the rate of 5 mM min-‘. The flow rate was 1 ml min-’ fraction-‘. Because of the large amounts of DNA loaded on the Cl8 and DEAE columns, we monitored the absorbance at 300 nm, so that the peaks would stay within the scale of reasonable sensitivity (less than 2) of our spectrophotometer.

We have previously found that under certain alkaline conditions the cross-link could undergo reversal without exposure to actinic W light to produce the unmodified DNA and the pyrone ring open and closed forms of the pyrone-side monoadducted DNA [23, 241. In contrast with photoreversal, base-catalyzed reversal of the cross-link yields only the pyrone-side monoadducted and unmodified DNA. No furan-side mono- adducted DNA was detected under the base-catalyzed conditions. The photoreversal with short wavelength W light of cross-linked double-stranded oligonucleotides yields mostly the furan-side monoadducted oligonucleotide [WI. The pyrone-side monoad-

Reverse Phase

Phase

Fig. 6. HhIT pyrone-side monoadducted oligonucleotide synthetic scheme: XL, cross-link.

HMT

0.1

t d

0.05

r 12r

8 mer 1

ner

t

I I I I I I 5

0 10 20 30 40 50 60 70 80 90

Minutes After Injection

Fig. 7. C,s reverse phase HPLC prolile and chromatogram of base-catalyzed reversal reaction of the Hh4T cross-linked 1%mer d(GAAGCTACGAGC) and 8-mer d(TCGATGCT).

ducted oligonucleotide is a minor product of this photoreversal reaction. Figure 7 is the chromatogram from the base-catalyzed reversal reaction of the 1Zmer cross-linked to the 8-mer. The reaction was performed by mixing HPLC-purified 12-mer cross- linked to the 8-mer with 0.1 N KOH and heating to 90 “C for 30 min. The reaction was neutralized with Tris-HCI and applied to a reverse phase HPLC column. This reaction should yield four products if it goes to completion: the unmodified 1Zmer and 8-mer, and the pyrone ring open forms of the 12-mer and 8-mer MAPS. There were two psoralen-monoadduct-containing peaks eluting from the column for each

75

strand. This can be explained as the pyrone-open and pyrone-closed forms of the monoadducts. The pyrone-ring-open form carries one more negative charge than does the pyrone-ring-closed form. The identity of the peaks in the chromatogram was determined by kinasing an aliquot of each fraction that corresponded to the peaks with y 32P ATP and then running the kinase reaction mixture out on a denaturing polyacrylamide gel. This is a synthetically useful reaction, and it is currently being exploited to produce micromole quantities of the MAp to the thymidine in the DNA octamer d(GCGTACGC).

3.6. Nuclear magnetic resonance spectroscopy We have synthesized sufficient quantities of the three main types of DNA-psoralen

adduct to begin NMR and X-ray crystallographic studies. These methods will enable us to study the structural changes in the DNA that allow recognition of damage by DNA repair enzyme systems. NMR studies are currently under way on both the

n .

,

I I I I I 1 I I

7.2 6.4 5.6 4.8 4.0 3.2 2.4 1.6 PPM

Fig. 8. 2D ‘H NOESY spectrum of HMT-d(GCGTACGC)2 at 25 “C (f,,,e=250 ms).

76

d(GCGTACGC) thymidine-HMT-thymidine cross-link and the thymidine-HMT MAf. NMR studies of the pyrone-side monoadducted oligonucleotide will commence in the near future. The furan-side and pyrone-side monoadducted DNA structures are of particular interest because a DNA duplex containing one of these monoadducts is thermodynamically more stable than an unmodified DNA duplex by about 1.7 kcal mol-r and 0.8 kcal mol-’ respectively [21]. Figure 8 shows a 2D NOESY spectrum of the HMT-d(GCGTACGC)* cross-link at 250 ms. Several features of the spectrum are consistent with cross-link formation. The H6 protons of the adducted thymines are shifted to higher field (about 5.5 ppm) relative to an H6 proton of an unadducted thymine (about 7.2 ppm). Cross-peaks between these adducted thymine H6 protons and both methyl hydrogen atoms and methylene hydrogen atoms of the psoralen indicate their close proximity. The weak cross-peaks observed between each thymine H6 proton and the Cl’ proton of its 3’ adenine neighbor locate the psoralen directly between the AT base pairs. The NOESY and COSY spectra obtained for the HMT- cross-linked molecule are of sufficient quality and resolution to enable a quantitative evaluation of the cross-peaks to obtain distance information. Further, the results from the double-quantum-filtered COSY experiments allow us to measure torsional angles for the ribose rings. Similar 2D NMR spectra have been acquired for the HMT-d(GCGTACGC)* MAf. Results to date indicate that the psoralen is, indeed, intercalated in the DNA and located between the AT base pairs. A full report on the NMR spectroscopy studies and structure determination results of these adducts will be published elsewhere.

4. Discussion

The use of laser techniques with HPLC purification for the large-scale preparation of psoralen-DNA adducts is a recent advance in our laboratory. This methodology is a significant improvement over the use of conventional arc lamps and preparative denaturing gel electrophoresis. HPLC purification methods have many advantages over denaturing PAGE. HPLC enables one to purify multimicromole quantities of DNA rapidly, because the columns have large binding capacities relative to PAGE slab gels, and because the separation can be based on both differences in hydrophobicity and charge density and not just the size and shape of the molecule. Adducted oligomers that are longer than 25 nucleotides are very inefficiently separated by reverse phase HPLC. The increased length of the oligo neutralizes the relative hydrophobicity differences between the psoralen-adducted molecules and their parent unmodified oligomers. The nucleotide sequence of the DNA oligo also affects the retention time and elution pattern of the adducted oligos during a reverse phase separation. In general, the more homogeneous the starting oligos, the easier is the subsequent purification of the photoreactants and the greater is the final yield of the required products. The best results are obtained with oligomers purified by HPLC to homogeneity before they are photoreacted with psoralen.

The distribution of MAfs and cross-links in a photoreaction will depend on the rate of MAf formation relative to the rate at which the monoadduct goes on to cross- link. This rate is directly proportional to the extinction coefficient of HMT and the MAf at the wavelength at which the irradiation takes place, as represented by the following equations:

kl = I~141 0)

77

kz=Iu2& (2)

where k1 is the rate of monoadduct formation, kz is the rate of cross-link formation from MAf, I is the intensity of the light irradiating the sample, 4r and A are the quantum yields for MAf and cross-link formation respectively, l 1 and l z are the extinction coefficients for HMT and MAf respectively, NA is Avogadro’s number, and al and u2 are the absorption cross-section of I-BIT and MAf respectively. u is directly proportional to the extinction coefficient of the molecule at that wavelength. The quantum yield $~r for the formation of MAf 8-methoxypsoralen by irradiation at 397.5 nm is 0.0065, and the quantum yield & for the conversion of MAf to cross-link by irradiation at 341 nm is 0.028 [26]. The quantum yield for conversion of HMT MAf to cross-link has been measured to be 0.024 by irradiation at 334 nm [21]. The quantum yields for the formation of cross-link from MAf are the same for these two psoralens within experimental error. Therefore it is convenient to assume that the quantum yields for formation of monoadducts from the two psoralens will also be similar. The quantum yield for driving an MAf on to cross-link is four times larger than the quantum yield for the initial formation of MAf. This explains why MAf does not accumulate in the reaction mixture when an HMT+DNA solution is irradiated at 364 nm (argon laser) where both HMT and the MAf have relatively large extinction coefficients. The extinction coefficients for HMT and the HMT MAf at 389 nm were estimated from their UV absorption spectra. The extinction coefficient for HMT at this wavelength is about 40 (1 mol-’ cm-‘), and for the MAf less than 1 (1 mol-’ cm-‘). If a mixture of DNA+HMT is irradiated with light in a region of the absorption spectrum where the extinction coefficient for the MAf is substantially lower than the extinction coefficient for the free psoralen (as in the case of krypton laser), MAf will accumulate in the reaction mixture.

A solution structure derived from NMR data for the 8-mer d(S’-GGGTACCC- 3’) cross-linked with the psoralen 4’-aminomethyl4,5’,&trimethylpsoralen @MT) has been reported by Tomic et al. [27]. Structures derived from NMR data and initial computer modeling of this cross-linked molecule indicate a significant kink (bending into the major groove) and unwinding at the site of the damage. Subsequent computer modeling with the same data and an improved distance geometry-simulated annealing technique have given structures with a smaller degree of bending (unpublished data). All these new structures are consistent with the original NMR data set obtained by Tomic et 41. [271, suggesting limitations in distance geometry methods for arriving at unique structures when there are a relatively small number of constraints imposed on the modeling procedure. Gel migration assays have concluded that no significant bending of the DNA occurs at a psoralen cross-link [28, 291. The drug that we are using in this study is different from the positively charged psoralen AMT used by Tomic et al., in that HMT is uncharged. AMT is suspected of inducing additional modification in the structure of nucleic acids by the nature of its charge. HMT does not introduce this additional complication. It is our intention to measure the sugar-phosphate backbone angles in our family of HMT-adducted DNA molecules in order to constrain the modeling and energy minimization procedures further, and to clear up these apparent contradictions in the literature for the structures of psoralen-modified DNA.

78

Before the twin innovations of laser irradiation and HPLC purification, preparation of even 20 pg of any of these adducts was a massive undertaking. Psoralen-induced damage to DNA serves as an excellent model for understanding the structural motifs that cellular repair enzyme systems recognize. The ability to produce large quantities of the three main psoralen-adducted DNA molecules will greatly facilitate the study of human and bacterial repair systems, and arrested RNA polyrnerase ternary complexes.

Acknowledgments

We are thankful to Dr. David Cook for advice and suggestions during the course of the experiments and the laser training course and Mr. David Koh for oligonucleotide synthesis. We thank Professor R. Mathies and Dr. J. Ames for their help with the use of the krypton laser. The various suggestions and help of the members of the Hearst Laboratory are also appreciated. This work was supported by the Director, Office of Energy Research, Office of General Life Sciences, Molecular Biology Division (J.E.H.) and Office of Energy Research, Office of Health and Environmental Research (D.E.W.) of the US Department of Energy under Contract DE-ACO3-76SF00098 and National Institute of Health Grant GM 41911 to J.E.H. S.S.S. was supported by Grant NIEHS 07075-11 from United States Public Health Service. Instrumentation was provided by US Department of Energy Grant DE FGOS-86ER75281 and National Science Foundation Grants DMB86-09035 and BBS87-20134.

References

1 2

3

.7

8

9

10

A. -Sancar and G. B. Sancar, DNA repair enzymes, Annu. Rev. Biochem., 57 (1988) 29-67. J. T. Reardon, P. Spielmann, J.-C. Huang, S. S. Sastry and A. Sancar, Removal of psoraien monoadducts and crosslinks by human cell free extracts, Nucleic Acids Res., 19 (1991) 4623-4629. R. R. Sinden, J. 0. Carlson and D. E. Pettijohn, Torsional tension in the DNA double helix measured with trimethylpsoralen in living cells: analogous measurements in insect and buman ceils, CeN, 21 (1982) 773-783. A. J. Courey, S. F. Plon and J. Wang, The use of psoralen-modified DNA to probe the mechanism of enhancer action, Cell, 45 (1986) 567-574. J. M. Hybbard and J. E. Hearst, Computer modeling 16s ribosomal RNA, J. MoL Biol., 221 (1991) 889-907. G. D. Cimino, H. Gamper, S. T. Isaacs and J. E. Hearst, Psoralens as photoactive probes of nucleic acid structure and function: organic chemistry, photochemistry and biochemistry, Annu. Rev. Biochem., 54 (1985) 1151-1193. Y.-B. Shi, H. Gamper, B. V. Houten and J. E. Hearst, Interaction of Escherichiu cob RNA polymerase with DNA in an elongation complex arrested at a specific psoralen crosslink site, .I. Mol. BioL, 499 (1988) 277-293. Y.-B. Shi, H. Gamper and. J. E. Hearst, Interaction of T7 RNA poiymerase with DNA in an elongation complex arrested at a specific psoraien adduct site, Z. BioL Chem, 263 (1988) 527-534. S. S. Sastry &id J. E. Hearst, Studies on the interaction of T7 RNA polymerase with a DNA template containing a site-specifically placed psoralen cross-link I. Characterization of elongation complexes, .Z. MoL i&d., 221 (1991) 1091-1110. S. S. Sastry and J. E. Hearst, Studies on the interaction of T7 RNA polymerase with a DNA template containing a site-specifically placed psoraien cross-link. II. Stability and some properties of elongation complexes, J. Mol. BioL, 221 (1991) 1111-1125.

79

11 S. T. Isaacs, J. W. Tessman, K. C. Metchette, J. E. Hearst and G. D. Cimino, Post-PCR sterilization: development and application to an HIV-l diagnostic assay, Nucleic Aciak Res., 19 (1991) 109-116.

12 G. D. Ciio, K. C. Metchette, J. W. Tessman, J. E. Hearst and S. T. Isaacs, Post-PCR sterilization: a method to control carryover contamination for the polymerase chain reaction, Nucleic Acids Res., 19 (1991) 99-107.

13 R. Y. Dodd, G. Moroff, S. Wagner, M. H. Dabay, E. Dorfman, V. George, A. Ribeiro, J. Shumaker and L. E. Benade, Inactivation of viruses in platelet suspensions that retain their in vitro characteristics: comparison of psoralen-ultraviolet A and merocyanine 540~visible light methods, Traryksion, 31 (1991) 483-490.

14 R. L. Edelson, Photophoresis-present and future aspects, J. Photochem. Photobiol. B: Biol., 10 (1991) 165-171.

15 H. J. Alter, R. P. Creagan, P. A. Morel, G. P. Wiesehahn, B. P. Dorman, L. Corash, G. C. Smith, H. Popper and J. W. Eichberg, Photochemical decontamination of blood components containing hepatitis B and non-A, non-B virus, Lancet, 2 (1988) 1446-1450.

16 L. Lm, G. P. Wiesehahn, P. A. Morel and L. Corash, Use of 8-methoxypsoralen and long- wavelength ultraviolet radiation for decontamination of platelet concentrates, Blood, 4 (1989) 517-525.

17 G. D. Ciiino, Y.-B. Shi and J. E. Hearst, Wavelength dependence of the photoreversal of a psoralen-DNA crosslink, Biochemistry, 25 (1986) 30X3-3020.

18 R. R. Sinden and P. J. Hagerman, Interstrand psoralen cross-links do not introduce appreciable bends in DNA, Biochemistry, 23 (1984) 6299-6303.

19 Rainin, HPLC instrumentation and Supplies Catalog, 1991. 20 T. Maniatis, E. F. Fritsch and J. Sambrook, Molecular Cloning - A Laboratoy Manual, Cold

Spring Harbor Laboratory, Cold Spring Harbor, NY, 1982. 21 Y.-B. Shi and J. E. Hearst, Thermostability of double-stranded deoxyribonucleic acids: the

effects of covalent additions of a psoralen, Biochemistry, 25 (1986) 5895-5902. 22 Y.-B. Shi and 3. E. Hearst, Thewavelength dependence for the photoreactions of DNA-psoralen

monoadducts. 1. Photoreversal of monoadducts, Biochemtity, 26 (1987) 3792-3798. 23 A. T. Yeung, B. Jones and C. T. Chu, Photoreactivities and thermal properties of psoralen

cross-links, Biochemistry 27 (1988) 3204-3210. 24 Y.-B. Shi, H. P. Spiehnann and J. E Hearst, Base-catalyzed reversal of a psoralen-DNA

crosslink, Biochemistry, 27 (1988) 5174-5178. 25 Y.-B. Shi and J. E. Hearst, Thewavelength dependence for the photoreactions of DNA-psoralen

monoadducts. 1. Photoreversal of monoadducts, Biochemistry, 26 (1987) 3786-3792. 26 J. W. Tessman, S. T. Isaacs and J. E. Hearst, Photochemistry of the furan-side 8-methox-

ypsoralen-thymidine monoadduct inside the DNA helix. Conversion to diadduct and to pyrone-side monoadduct, Biochemistry, 24 (1985) 1669-1676.

27 M. Tomic, D. Wemmer and S. H. Kim, Structure of psoralen cross-linked DNA in solution by nuclear magnetic resonance, Science, 238 (1987) 1722-1725.

28 R. R. Sinden and P. J. Hagerman, Interstrand psoralen cross-links do not introduce appreciable bends in DNA, Biochemistry, 23 (1984) 6299-6303.

29 T. E. Haran and D. M. Crothers, Phased psoralen cross-links do not bend the DNA double helix, Biochemisti, 27 (1988) 6967-6971.