Bioconjtgate Chem. 1001, 2, 431-434
431
Photooxidation of Oligodeoxynucleotides by U022+ Aubrey R. Hill, Jr. and Leslie E. Orgel’ The Salk Institute for Biological Studies, P.O.Box 85800, San Diego, California 92186-5800. Received June 28, 1991
Photooxidation and consequent cleavage of oligodeoxynucleotides by the uranyl ion is greatly enhanced in the presence of a terminal 5’- or 3’-phosphate group. This enhanced cleavage is confined to nearby phosphodiester bonds in the case of a 5’-phosphate, but is less localized in the case of a 3’-phosphate. Several attempts to use a complementary oligodeoxynucleotide to direct a uranyl “warhead” against an oligodeoxynucleotide target were unsuccessful. Our results are most easily explained if we suppose that uranyl ions form coordination complexes with terminal phosphates and that, on photoexcitation, coordinated uranyl ions extract a hydrogen atom from the CH bond of a nearby deoxyribose residue.
INTRODUCTION
Nll: d5’[CACCCA ACT CGI3’
Photochemical oxidation of polydeoxynucleotides by uranyl ions (UOz2+) has recently been introduced as a method of footprinting DNA (Nielsen et al., 1988; Jeppesen & Nielsen, 1989). In the absence of associated proteins, the efficiency of cleavage of backbone phosphodiester bonds is similar for single- and double-stranded DNA and is insensitive to base sequence. The binding of protein to internal sequences of double-stranded DNA protects some regions of the DNA from cleavage and, in addition, can induce extensive cleavage at specific sites close to the protein. Photochemical oxidation of alcohols by U022+involves direct attack of the excited ion on the alcohol (Rabinowitch & Belford, 1964) while other similar footprinting methods depend on the production of diffusible radicals (Hertzberg & Dervan, 1984). It seemed interesting, therefore, to explore in detail the oxidation of short oligodeoxynucleotides by UOz2+. EXPERIMENTAL PROCEDURES
Materials. Reagent-grade chemicals were used throughout. Uranyl nitrate hexahydrate was purchased from Mallinckrodt; adenosine 5’-[y-32P]triphosphate, triethylammonium salt (-3000 Ci/mmol), and cytidine 5’-[(r32P]triphosphate, triethylammonium salt (-3000 Ci/ mmol), were from Amersham; alkaline phosphatase from calf intestine (molecular biology quality) was from Boehringer Mannheim Biochemicals; T4 polynucleotide kinase was from New England Biolabs; terminal deoxynucleotidyl transferase was from Pharmacia LKB Biotechnology; and phosphodiesterase I, type VII, was from Sigma Chemical Co. The following oligodeoxynucleotides were synthesized on an automated DNA synthesizer and purified by HPLC on RPC-5 at pH 12 using a perchlorate gradient as previouslydescribed (Bridson and Orgel, 1980). 1 Abbreviations used: HPLC, high-performance liquid chromatography; TBE, Tris/borate/EDTA buffer, pH 8.0; pN is the 5’-phosphate of the nucleoside N; pNll is the 5’-phosphate of the oligodeoxynucleotide Nll; Nllp is the 3’-phosphate of the oligodeoxynucleotide Nll; NllprC is the oligonucleotide obtained by adding one riboC residue to the 3’-terminus of the oligodeoxynucleotide Nll. Corresponding derivatives of other oligodeoxynucleotides are represented in a similar way. * Corresponding author.
1043-180219 112902-043 1$02.50/0
N16: d5’[CACAAT TCC ACA CAA CI3‘ N3,:d5’[TCGTAT GTT GTG TGG AAT TGT GAG CGG ATA ACA A T T TI3’
Methods. Gel electrophoresis was carried out on 0.75 mm thick 20% polyacrylamide denaturing gels cast and run in 1 X TBE buffer. Conversion of oligodeoxynucleotides to 5’-[32P]phosphateswas carried out using T4 polynucleotide kinase (Chu & Orgel, 1985). Conversion of oligodeoxynucleotides to 3’- [32P]phosphateswas achieved by adding a single 32P-labeled C residue with terminal deoxynucleotidyl transferase (Deng& Wu, 19831,purifying by HPLC, and then carrying out oxidation-elimination (Weith & Gilham, 1967). Hydrolysis of oligodeoxynucleotides with venom phosphodiesterase was carried out in 10 mM Tris buffer at pH 7.5. Irradiations were performed at room temperature using Philips TL40W/03 fluorescent lamps with Super Actinic Radiation (A = 420 nm, 30-nm band width). Solutions were sterilized by filtration through 0.2 pm filters. Reactions were carried out in sterilized 0.65-mL presiliconized, RNase-free microfuge tubes (National Scientific) irradiated from below at a distance of 4 in. from the light source. Reaction volumes were usually 10 pL, and the concentration of oligodeoxynucleotides was in the range of 1-10 nM. Solutions of uranyl nitrate were prepared frequently, since they tend to decompose on storage even in the dark. The compositions of the reaction mixtures are indicated in the legends to the figures. RESULTS AND DISCUSSION
Preliminary experiments with 5’- [32P]-N3,showed that, in the absence of complexing anions, the rate of photooxidation of oligodeoxynucleotides and the nature of the photoproducts depends in a complicated way on the concentration of the UOz2+ion. In Figure 1we illustrate a typical series of experiments using a range of UOz2+ concentrations from 0.4 to 250 pM. At the higher concentrations of U0z2+the reaction is relatively inefficient. The regular ladder of bands (lanes 1and 2) suggests that all internucleotide bonds are attacked at a comparable rate. When the U022+ concentration is reduced (lanes 3 and 4), the reaction becomes much more efficient and the nature of the products changes-most of the cleavage occurs close to the 5’-end of the oligomer. At even lower 0 1991 American Chemical Society
Hill and Orgel
432 Bloconlugate Chem., Vol. 2, No. 6, 1991 1
PN37 -
2
3
4
1
5
t
2
3
4
5
6
7
8
PNll BPB
BPB -
Figure 1. Autoradiogram of a 20% denaturing gel showing concentration-dependence of the photooxidation of 5’- [32P]-N37 by U022+ in 10 mM Tris.HC1 (pH 7.5) after a 2-h irradiation. U022+concentrations are as follows: lane 1,250 p M ; lane 2,50 pM; lane 3,lO pM; lane 4 , 2 pM; lane 5,0.4 pM.
concentrations of U022+,the same pattern of products is obtained, but the reaction is slower (lane 5). The unusual concentration dependence that we observe is presumably due to the formation of polynuclear complexes. The addition of complexing agents including inorganic phosphate, inorganic polyphosphates, nucleoside monophosphates, and nucleoside polyphosphates to reaction mixtures containing high concentrations of U022+leads to a change in the pattern of products qualitatively similar to that observed on lowering the concentration of U022+. The nature of the complexingagents,however, significantly and reproducibly influences the relative amounts of different products. A typical experiment using 5’- [32P]N11 as substrate and a variety of complexing agents is illustrated in Figure 2. These findings are consistent with the work of Jeppesen and Nielsen (1989),who noted that the addition of citrate, a ligand that chelates the U022+ ion, leads to efficient photochemistry. We adopted their protocol and used a solution containing 10 mM Tris.HC1 (pH 7.5),0.25 mM U022+,and 0.25 mM citrate (pH 7.5) unless otherwise noted. The time course of the photooxidation of 5’- [32P]-N37 in the presence of citrate is illustrated in Figure 3, lanes 1-6. Extensive destruction occurs after only 5 min to give labeled inorganic phosphate and low molecular weight labeled phosphate-containing products. The corresponding reaction in the presence of uridine 5’-monophosphate is illustrated in Figure 3, lanes 7-12. Decomposition is slower and the pattern of products is slightly different. Since low molecular weight products are already present after short irradiation times, they must be formed by a single cleavage close to the labeled 5’-phosphate group and not as a result of extensive random degradation. The chemical nature of the 32P-labeledphotoproducts released from a 5’- [32P]-~lig~de~xynucle~tide cannot be determined from experiments of the kind described here, since the amounts of material photolyzed are very small. However, an estimate of the molecular sizes can be obtained using gel electrophoresis.
Figure 2. Autoradiogram of a 20% denaturing gel showing the effect of complexing agents on the photooxidation of 5’-[32P]-N11 after a 2-h irradiation in 10 mM Tris.HC1 (pH 7.5) with 0.25 mM U022+and 0.25 mM complexing agent. Complexing agents are as follows: lane 1,PA; lane 2, pC; lane 3, pG; lane 4, pU; lane 5, ADP; lane 6, CDP; lane 7, GDP; lane 8, UDP. 1 PN37
2
3
4
5
6
7
8
9
10 11
12
-a
BPB -
Figure3. Autoradiogram of a 20 % denaturing gel showingtimedependence of photolysis of 5’-[32P]-N37in the presence of citrate or pU in 10 mM Tris.HC1 (pH 7.5) with 0.25 mM UOz2+and 0.25 mM citrate (lanes 1-6) or 0.25 mM pU (lanes 7-12). Irradiation times: lanes 1and 7, time 0; lanes 2 and 8, 5 min; lanes 3 and 9,15 min; lanes 4 and 10,30 min; lanes 5 and 11,60 min; lanes 6 and 12, 120 min.
The fastest moving band seen on gels (for example in Figure 1)is shown to be inorganic phosphate by comparison with authentic material. In lanes 2 and 3 of Figure 4 we compare the ladder obtained by venom phosphodiesterase digestion of 5’-[32P]-Nllwith the pattern of products obtained by the U022+ photooxidation of the same oligomer (lane 4). One of the three major organic photolysis products moves faster than pC, one moves at about the same speed, and one moves somewhat slower, but still substantially faster than pCpA. Presumably the fastest
Bbcmjt@ate chem.,Vol. 2, No. 6, 1991 433
PhotooxIdatian of Oligodeoxynucleotklesby U02* 1
‘-
2
3
1
4
2
3
4
5
6
PNll
-PNll I - BPB
or -
NllP
BPB .
- Pi Figure 4. Autoradiogram of a 20%denaturing gel showing the comparison of producb formed by venom phosphodiesterase treatment of 5’-[32P]-N11with those formed by U022+photooxidation. Lane 1,5’-[32P]-Nl1,10 mM Tris.HC1 (pH 7.5); lane 2, same as lane 1 but incubated for 15 min with 1 X 10-6 units of enzyme at 37 “C; lane 3, same as lane 2 but with 2 X units of enzyme; lane 4, 10 mM Tris.HC1 (pH 7 3 , 0.25 mM U022+, 0.25 m M citrate, 5-min irradiation.
moving band corresponds to pCp or a bis-monophosphate related to pCp. The band with intermediate mobility probably corresponds to a monophosphate of C. The slowestmoving band is either a derivative of pC or a dimer related to pCpAp. It seems clear that the major products of photolysis of pNl1 are no longer than dimers, and hence that photooxidation occurs very close to the 5’-terminus. The patterns of labeled products obtained from pN11 (Figure 4, lane 4) and pN37 (Figure 1) are similar. Three strong bands are visible on the gel, corresponding to short oligodeoxynucleotides and very little material closer to the starting material. The results obtained with 5’-[32P]N16 are somewhat different (data not shown). In addition to the characteristic strong bands corresponding to short oligomers,which coincide exactly with those obtained from 5’- [32P]-Nllsince the 5’-terminal residues are C followed by A in each case, there is a substantial production of material of intermediate molecular weight. The experiments described above strongly suggest that the U022+ associates specifically with the terminal 5’phosphate group of each of the oligomers pN11, PNl6, and pN37 and then photooxidizes residues close to the 5’terminus of the nucleotide, thus releasing labeled inorganic phosphate and short labeled oligodeoxynucleotides and their oxidation products. In addition, some complexing must occur at an internal site on PN16 to account for the production of longer oligodeoxynucleotide photoproducts from that oligomer. To test this hypothesis we synthesized and photooxidized 3’-phosphate-labeled oligodeoxynucleotides. The results of the experiments using terminally radiolabeled 5’-[32P]-N11, N I ~ - ~ ’ - [ ~ ~ and P ] ,N11-[32P]-rC are shown in Figure 5. NllprC, which does not contain a phosphomonoester group, is not significantly photolyzed under our standard conditions (lane 6). N11p, on the other hand,
Figure 5. Autoradiogram of a 20%denaturing gel showing the photolysis of pNll (lanes 1 and 2), Nllp (lanes 3 and 41, and NllprC (lanes 5 and 6) by U022+after a 5-min irradiation. Lanes 1,3, and 5, 10 mM Tris-HC1(pH 7.5); lanes 2,4, and 6,lO mM Tris.HC1 (pH 7.5),0.25 mM U022+, and 0.25 mM citrate.
undergoes extensive breakdown to give a complex pattern of products, including a substantial amount of material of intermediate molecular weight (lane 4). In this behavior it differs significantly from pN11, which gives shorter products predominantly (lane 2). This experiment strongly suggests that a terminal phosphomonoester provides a complexing site for U022+ and consequently greatly increases the efficiencyof photooxidation. It also suggests that, while a 5’-phosphate directs oxidation to close-by sugar residues, a 3’-phosphate enhances less localized oxidation. The products of photooxidation of N16P and Nl6prC were generally consistent with this interpretation (data not shown). The oxidation of NlsprC yielded few short products, but substantial amounts of longer oligomers, again suggesting that there is a coordination site for UOz2+ somewhere in the interior of the molecule. Nl6P gave the most complicated pattern of products that we have seen, presumably because there are two coordination sites, one a t the 3’-terminus and one internal, and each produces cleavage at a substantial number of different internucleotide bonds. We carried out many experiments attempting to use the 5’-[32P]-37mer as target and a derivative of the complementary oligomer PN16 as an antisense reagent to direct a U022+ “warhead” to internal sites on the target (Figure 6). None of these experiments succeeded; the oxidation of the 37mer produced the same pattern of products whether it was free in solution or hybridized to a 16mer carrying a complexing group at its 5’-terminus. (See caption to Figure 6 for a list of the complexing groups used.) We confirmed in independent experiments that PN16 hybridizes to 5’- [32P]-37merunder the conditions of our experiments. The photooxidation of alcohols by U0z2+has been studied extensively (Rabinowitch & Belford, 1964). The reaction is initiated by the transfer of a hydrogen atom from a carbon atom of the alcohol to one of the 0 atoms
Hill and Orgel
Bioconjugate Chem., Vol. 2, No. 6, 1991
434
y3y"
J
3'
3' 5'
1 1 1 1 1 1 1 1 1 1 1 1
that, influenced by the bound protein, forms an unusually strong coordination site for the U O P ion (Jeppesen & Nielsen, 1989). It is not clear whether coordination is to backbone phosphodiester groups, bases, or both. ACKNOWLEDGMENT
This work was supported by Grant No. GM13435-23 from The National Institutes of Health. We thank Sylvia Bailey for manuscript preparation. Figure 6. Conformation that we anticipated for an oligodeoxynucleotide target and a complementary oligomer carrying a U022+"warhead" attached to a coordinating group X. The complexing groups X used unsuccessfully in this study were inorganic phosphate, cystamine, ethylenediaminetetraacetic acid, diethylenetriaminepentaacetic acid, ethylenediamine, pentaethylenehexamine.
of the U0z2+ion. Our results suggest that U0z2+attached to a 5'-terminal phosphate group attacks the terminal sugar residue to liberate inorganic phosphate or attacks one of the two terminal sugar residues to liberate mono- and possibly dinucleotide derivatives. Coordination of a second phosphate-containing molecule such as uridine 5'monophosphate to the U0z2+ion somewhat changes the conformation of the 5'4erminus of the oligodeoxynucleotide and hence alters the proportion of different oxidation products that are formed. A U0z2+ ion attached to a phosphate ion at the 3'-end of an oligodeoxynucleotide acts less discriminatingly by attacking residues more distant along the backbone. Sequence-dependent cleavage due to attachment of the U0z2+ion to internal sequences in the oligodeoxynucleotide is suggested by our findings with pN16. Highly specific cleavage of DNA in the neighborhood of a bound RNA polymerase molecule has been reported. It presumably arises in the same way, that is by coordination of a U0zZ+ ion to a region of the nucleic acid molecule
LITERATURE CITED Bridson, P. K., and Orgel, L. E. (1980) Catalysis of accurate poly(C)-directed synthesis of 3'-5'-linked oligoguanylates by Zn2+. J. Mol. Biol. 144, 567-511. Chu, B. C. F., and Orgel, L. E. (1985) Nonenzymatic sequencespecific cleavage of single-stranded DNA, Proc. Natl. Acad. Sci., U.S.A. 82, 963-961. Deng, G-r., and Wu, R. (1983) Terminal transferase: use in the tailing of DNA and for in vitro mutagenesis. In Methods in Enzymology (R. Wu, L. Grossman, and K. Moldave, Eds.) Vol. 100, pp 96-116, Academic Press, New York. Hertzberg, R. P., and Dervan, P. B. (1984) Cleavage of DNA with methidiumpropyl-EDTA-iron(I1): Reaction conditions and product analyses. Biochemistry 23, 3934-3945. Jeppesen, C., and Nielsen, P. E. (1989) Uranyl mediated photofootprinting reveals strong E . coli RNA polymerase-DNA backbone contacts in the +10 region of the DeoPl promoter open complex. Nucleic Acids Res. 17, 4947-4956. Nielsen, P. E., Jeppesen, C., and Buchardt, 0. (1988) Uranyl salts as photochemical agents for cleavage of DNA and probing of protein-DNA contacts. FEBS Lett. 235, 122-124. Rabinowitch, E., and Belford, R. L. (1964) Spectroscopy and Photochemistry of Uranyl Compounds, The Macmillan Company, New York. Weith, H. L., and Gilham, P. T. (1961) Structural analysis of polynucleotides by sequential base elimination. The sequence of the terminal decanucleotide fragment of the ribonucleic acid from bacteriophage f2. J.Am. Chem.SOC. 89,5413-5474.
