ABSTRACT
A monoclonal antibody (TDM-2) specific to a UV-induced cyclobutane pyrimidine dimer (T[cis-syn]T) has previously been established; however, the immunization had used UV-irradiated calf-thymus DNA containing a heterogeneous mixture of photoproduct sites. We investigated here the structural requirements of antigen recognition by the antibody using chemically synthesized antigen analogs. TDM-2 bound with cis-syn, but not trans-syn thymine dimer, and could bind strongly with four nucleotide analogs in which the cis-syn pyrimidine dimer was located in the center. Antigen analogs containing abasic linkers at the 5'- or 3'-side of the cis-syn cyclobutane pyrimidine dimer were synthesized and tested for binding to TDM-2. The results indicated that TDM-2 recognizes not only the cyclobutane ring but also both the 5'- and 3'-side nucleosides of the cyclobutane dimer. Furthermore, it was proved that either the 5'- or 3'-side phosphate group at a cyclobutane dimer site was absolutely required for the affinity to TDM-2. The antibody showed a strong binding to single stranded DNA but indicated little binding to double stranded DNA.
Ultraviolet light (UV) induces mutations in DNA, which are detected predominantly in dipyrimidine sites as cyclobutane pyrimidine dimers, pyrimidine [6-4] pyrimidone and Dewar photoproducts. These lesions cause mutations during the replication process (1 ). There are cis-syn [c,s] and trans-syn [t,s] cyclobutane pyrimidine dimer isomers (Fig. 1 ). These are photoproducts from the photo [2+2] cycloaddition of the 5,6-double bond of two adjacent pyrimidine nucleotides. The cis-syn pyrimidine dimers have anti glycosyl conformations, and in contrast, the 5'-side thymidine of the trans-syn pyrimidine dimers is in the syn glycosyl conformation. Usually, the cis-syn pyrimidine dimers are the major photoproducts of UV-irradiation, although trans-syn type pyrimidine dimers are generated in single stranded DNA (ssDNA). These photoproducts induce mutations in complementary strands during replication. Cis-syn pyrimidine photoproducts (T[c,s]C) have the ability to cause C -> T transitions derived from replicative bypass (2 ). We also reported that in in vivo experiments, cis-syn type cyclobutane thymine dimers (T[c,s]T) induced T -> A transversions on the 3'-side thymine of the dimer and trans-syn type (T[t,s]T) T -> A transversions on the 5'-side thymine of the dimer (3 ).
There are enzymes to repair thymine dimers (4 ) and [6-4] photoproducts (5 ). Antibodies specific for the cyclobutane or [6-4] photo products have been constructed to investigate the relationships between the generation of these UV-induced photo products and their biological effects (6 -8 ). Baan and colleagues obtained a monoclonal antibody that binds DNA containing a cyclobutane thymine dimer by immunization of mice with the tetranucleotide, GpT[c,s]TpG (6 ). The antibody bound thymine dimer-containing tetranucleotides, but not those with a photodimer only d(T[c,s]T). Antibodies binding to [6-4] photoproducts were established using dinucleotides of [6-4] photoproducts as the hapten (9 ). Antibodies that catalyze the photo-induced reverse reaction of a uracil cyclobutane dimer to normal dipyrimidine have also been reported (10 ).
Nikaido and colleagues have established a variety of monoclonal antibodies that bind with either cyclobutane, [6-4] or Dewar pyrimidine dimers (7 ,8 ,11 ). These monoclonal antibodies were obtained by immunization with UV-irradiated oligodeoxynucleotides and were characterized using DNA irradiated under controlled UV-irradiation conditions. These monoclonal antibodies have proved to be powerful tools for investigating cellular repair systems (12 ).
All oligodeoxynucleotides were synthesized using phosphoramidite units purchased from Perkin-Elmer Applied Biosystems Inc. Cis and trans cyclobutane thymine dimers were inserted in oligodeoxynucleotides by coupling the cyclobutane dimer amidite blocks as reported (13 ,14 ). Oligodeoxynucleotides containing the biotin or propyl phosphate group at the 3'-ends were synthesized using Biotin TEG CPG (GLEN research) or 3'-spacer C3 CPG (GLEN research) as the 3'-support. For the propyl phosphate group at the 5'-terminus, a propandiol phosphoroamidite block was synthesized as reported (15 ). pro[middot]pT[c,s]T and T[c,s]Tp[middot]pro were prepared from UV-irradiation of prop[middot]TpT and TpTp[middot]pro. All oligodeoxynucleotides were purified by reverse phase HPLC using columns of Inertsil ODS-3, 8 mm i.d. * 300 mm (GL Sciences) or [mu]-Bondasphere 4.6 mm i.d. * 250 mm (Waters). If necessary, anion-exchange HPLC was carried out using columns of TSK gel DEAE-2SW, 4.6 i.d. * 250 mm (Tosoh).
cs23, 5'd(ATCGCGACGAAT[c,s]TAAGCAGCGCT)3'; ts23, 5'd(ATCGCGACGAAT[t,s]TAAGCAGCGCT)3'; competitors consisting of four and two bases, 5'd(AT[c,s]TA)3', 5'd(GT- [c,s]TG)3', 5'd(CT[c,s]TC)3', 5'd(TT[c,s]TT)3', 5'd(T[c,s]T)3'. AA23, 5'd(AGCGCTGCTTAATTCGTCGCGAT)3'; mis2, 5'd(A- GCGCTGCTT
The ELISA was performed according to a standard procedure. A flat-bottom 96 micro well plate (Nunc) was precoated by incubation with 1% protamine sulfate (50 [mu]l/well) at 37oC for 2 h (11 ). After the wells were washed with sterile water and dried, 50 [mu]l (10 ng/[mu]l) oligodeoxynucleotide solution (cs23, ts23 or dsDNAs) were added to the wells, and the plate was incubated at 37oC for 20 h. The plate was dried and washed with PBS containing 0.05% Tween20 (Nakarai; 100 [mu]l/well *5) (PBS-T). PBS containing 2% skim milk (Yukijirushi Nyugyo) was added to the plate (200 [mu]l/well), and the plate was incubated at 37oC for 1 h to prevent non-specific binding of the antibody. The plate was washed with PBS-T (200 [mu]l/well *5) and the monoclonal antibody, which was diluted in PBS containing 2% skim-milk, was added to the wells (100 [mu]l/well). The plate was incubated at 37oC for 1 h. The plate was washed with PBS-T (150 [mu]l *5) and then was incubated with goat anti-mouse immunoglobulin G (IgG; H+L) conjugated with peroxidase (100 [mu]l/well) at 37oC for 1 h. After the plate was washed with PBS-T (150 [mu]l *3) and citrate-phosphate buffer, pH 5.0 (150 [mu]l *2), the substrate solution (0.04% o-phenylene diamine, 0.007% H2O2 in citrate-phosphate buffer) was added to the plate (100 [mu]l/well), and the plate was incubated at room temperature for 15 min. To stop the reaction, 2 M sulfuric acid was added (50 [mu]l/well), and the absorbance at 492 nm was measured using an Immuno Reader Model 2550 EIA (BioRad).
The 96 microwell plates were prepared as described previously (9 ,16 ). Avidin (Organon Teknika Co.) dissolved in PBS was dispensed into a 96 microwell plate (5 [mu]g/50 [mu]l per well), which was covered with parafilm and incubated at 4oC overnight. After the plate was washed with PBS (200 [mu]l/well *3), 3'-biotinylated cs23 dissolved in PBS (20 nM) was added to the plate (50 [mu]l/well), which was covered and incubated at 4oC for 2 h. The plate was washed with PBS-T (100 [mu]l/well *5). The antibody binding was measured as described above.
TDM-2 was dissolved in PBS (4 ml), at the concentration at which TDM-2 bound to 3'-biotinylated cs23 to indicate ~0.5 A492 unit, and competitors were dissolved in PBS (180 [mu]l). An equal volume (180 [mu]l) of the antibody solution was added to the competitor solution (180 [mu]l) and mixed gently. Aliquots of 100 [mu]l were added to wells of immobilized cs23-Bio. The binding of TDM-2 was measured using the same procedure as the ELISA with the protamine coated plate. Percent inhibitions were calculated as follows (7 ). Percent inhibition = A - A1/Ao - A1 *100, where Ao is the absorbance at 492 nm without competitors, A is the absorbance with competitors, and A1 is the background absorbance without antibody and competitors.
Gel mobility shift assays were carried out as previously reported (17 ). The 5'-end labeled cs23 (0.1 [mu]M) and complementary strand (AA23, mis2, mis4 or mis6; 0.2 [mu]M) were dissolved in a PBS buffer solution (1.9 mM NaH2PO4, 8.1 mM Na2HPO4, 140 mM NaCl, 25 [mu]l), denatured at 90oC for 2 min, and then immediately transferred to an ice bath. The solution containing dsDNAs was heated at 80oC for 2 min and was annealed by cooling to room temperature. The solution (10 [mu]l) was removed and mixed with TDM-2 and calf thymus DNA. The PBS solution contained 50 nM DNA duplexes, 0.05 [mu]g calf thymus DNA and 0.51 [mu]M TDM-2. After the solution was incubated at 37oC for 1 h, 30% glycerol (5 [mu]l) was added to the aliquot, and 8% native polyacrylamide gel (acrylamide:bisacrylamide, 29:1) electrophoresis was carried out at 24oC. The gel was dried and the percentages of complex formation were measured by a bioimaging analyzer (FUJIX BAS2000).
Cis-syn and trans-syn type cyclobutane thymine dimer blocks were chemically synthesized by the reported procedures (13 ,14 ), and the oligodeoxynucleotides cs23 and ts23 were synthesized (sequences shown in Materials and Methods). These oligonucleotides contained the cyclobutane dimer at the center. The binding to the antibody (TDM-2) was assayed by the enzyme-linked immunoassays (ELISAs) (7 ). TDM-2 bound with cs23 in an antibody concentration dependent manner, but did not recognize ts23 as the antigen (Fig. 2 ). UV irradiation of d(TpT) induces almost equal amounts of cis-syn and trans-syn type cyclobutane dimers as the photoproducts. On the other hand, cis-syn type pyrimidine dimers are generated as the major product in oligonucleotides exposed to UV. Since the haptens were prepared by UV-irradiation of oligonucleotides, they contained the cis-syn type of cyclobutane dimers as the major products, and so probably the antibody specific to the cis-syn type dimer was elicited.
To investigate the base preference in the antigen recognized by TDM-2, four antigen analog bases (5'NpT[c,s]TpN3'; N = A, G, U, C) were synthesized. These had the cis-syn type photodimer at the center, and a competitive ELISA against cs23 was performed. The 3'-end of cs23 was covalently attached to the biotin and was immobilized on the plate coated with avidin (9 ,17 ). All of the tetramers, NpT[c,s]TpN, showed the same inhibitory effects as cs23. However, the cyclobutane dimer, d(T[c,s]T) only had a slight inhibitory effect (Table 1 ). These results suggest that the recognition size of TDM-2 ranges in the four bases containing the cyclobutane dimer, and the composition of the nucleotide flanking the photodimer does not affect the binding of the antibody. Furthermore, the various antigen analogs shown in Figure 3 were synthesized to investigate whether the antibody recognizes the nucleosides and the phosphates on each side of the dimer, T[c,s]T. Analogs of NpT[c,s]TpN were tested, in which either N was replaced with propandiol (pro) phosphate as an abasic analog but retained the phosphate diester, which has the same negative charge as that in an oligodeoxynucleotide. The results are shown in Figure 4 and Table 1 .
Table 1
The analog pro[middot]pT[c,s]TpA, in which only the 5'-side nucleoside was replaced with propandiol, inhibited the binding of TDM-2 to cs23 to the same extent as ApT[c,s]TpA. The analog ApT[c,s]Tp[middot]pro, which had a propandiol at the 3'-side of the cyclobutane dimer, also competed with cs23 for the binding to the antibody; however, the inhibition by ApT[c,s]Tp[middot]pro was ~3-fold lower than that of pro[middot]pT[c,s]TpA. These results suggest that the antibody seems to bind to the 3'-side nucleoside to the cyclobutane dimer more tightly than the 5'-side nucleotide. Escherichia coli DNA photo-lyase also reversed T[c,s]TpT more efficiently than TpT[c,s]T (18 ). The thymine dimer {d(T[c,s]T)} adopts a different structure from that in DNA (19 ). Therefore, when the 3'-side nucleoside is attached to the thymine dimer, rather than the 5'-side nucleoside, the dimer may form a structure similar to that in oligonucleotides. The analog, pro[middot]pT[c,s]Tp[middot]pro, which has propandiols at both ends, showed an ~20-fold weaker binding ability than ApT[c,s]TpA or pro[middot]pT[c,s]TpA, and was ~6-fold weaker than ApT[c,s]Tp[middot]pro. The 5'-side nucleoside might also be important in maintaining the shape of the thymine dimer for binding to TDM-2. It is not clear why pro[middot]pT[c,s]Tp[middot]pro was poorly recognized, as compared with pro[middot]pT[c,s]TpA and ApT[c,s]Tp[middot]pro. The structure of T[c,s]T may affect the recognition, and the deoxyriboses of the nucleosides flanking T[c,s]T may also be recognized by TDM-2.
Although pro[middot]pT[c,s]Tp[middot]pro was recognized by TDM-2 less efficiently, as compared with the tetramer analogs, it has a higher affinity than the dimer {d(T[c,s]T)}. To investigate the effect the phosphate groups neighboring the cyclobutane dimer on the binding to TDM-2, T[c,s]Tp[middot]pro and pro[middot]pT[c,s]T lacking either the 5'- or 3'-side phosphate were synthesized (Fig. 3 ). In the competitive ELISA, neither T[c,s]Tp[middot]pro nor pro[middot]pT[c,s]T could inhibit TDM-2 binding sufficiently (Fig. 4 and Table 1 ). The lack of either the 5'- or 3'-phosphate group drastically decreased the affinity to the antibody. However, these competitors harboring a phosphate group still maintained a larger affinity than the dimer d(T[c,s]T) alone. These results indicate that both the 5'- and 3'-end phosphates participate in recognition of the epitope by TDM-2. Perhaps there are weak electrostatic interactions between the negatively charged phosphate groups and the positively charged amino acid residues in the antibody. The antibody is specific for the cyclobutane ring of T[c,s]T, because it binds neither a normal thymidine dinucleotide nor a trans-syn thymine dimer. The presence of the phosphate groups at both ends of the cyclobutane dimer seems to enhance the affinity of the antigen with the antibody, while the dimer d(T[c,s]T) without the flanking phosphates, might wobble in the binding pocket of the antibody.
The antibodies that were elicited by using only thymine dimers without flanking sequences {[6-4] photoproducts (9 ) or trans-syn thymine dimer analogs (10 )} were able to bind directly with the thymine dimers in the absence of neighboring nucleotides. However, TDM-2 was obtained from immunization with UV irradiated calf thymus ssDNAs, and it has been proved by ELISA experiments that not only the cyclobutane ring but also the neighboring nucleoside and phosphate groups enhance the affinity of the hapten to the antibody. The tertiary structure of the epitope binding site of TDM-2 might be different from that of other antibodies obtained from immunization with the smaller molecules.
TDM-2 was obtained by immunization using UV-irradiated ssDNAs, and it has been proved that the antibody recognized oligonucleotides with three or four base spans containing one cis-syn cyclobutane thymine dimer. Phosphate groups on both sides of these oligonucleotides were required for recognition by TDM-2. It was important to investigate whether TDM-2 was able to bind to the thymine dimer in a double stranded DNA (dsDNA).
In a previous report, protamine coated plates were used for the ELISA (7 ) and it is not clear whether the dsDNA dissociated into ssDNA. Such a dissociation of dsDNA could be detected in a gel mobility shift assay. The 5'-end labeled 23mer, cs23, was prepared and combined with a complementary strand (AA23) to form a dsDNA (cs23[middot]AA23), and gel mobility shift assays were performed. The results of the gel shift assays showed that TDM-2 could bind with a ssDNA, but not with a dsDNA, in the presence of large excess of antibody (Fig. 5 a). It was reported that both thymine bases in the cyclobutane dimer form T[middot]A base pairs with the opposing adenines, although the hydrogen-bonding interactions were reduced (19 ,20 ). TDM-2 might not recognize a cyclobutane dimer in the DNA helix because of the base pairs. To investigate the effect of base pairs at the cyclobutane dimer site, two adenine bases in the complementary strand were changed to guanines (mis2), and four and six bases in the neighboring positions were also changed (mis4, mis6, respectively). When cs23 was combined with mis2, TDM-2 showed slight binding with the cs23[middot]mis2 double strands. However, TDM-2 bound with a higher affinity to all double strands containing four and six base mismatches (cs23[middot]mis4, cs23[middot]mis6, Fig. 5 b). The hydrogen bonds between the cyclobutane thymine dimer and the opposite adenines in the cs23[middot]mis2 duplex seemed to be disrupted, and the cyclobutane dimers in the cs23[middot]mis4 and cs23[middot]mis6 duplexes may be exposed. Internal loops (bubble structures) seemed to be formed in these duplexes, and TDM-2 may be able to access the cyclobutane thymine dimers in these duplexes. TDM-2 showed high affinity with ApT[c,s]TpA but a low affinity with d(T[c,s]T) in the competitive ELISAs. Since TDM-2 bound to double strands containing four or six mispairs more efficiently than to a double strand in the presence of two mismatched pairs (cs23[middot]mis2), the results from the gel mobility shift assay can be explained. TDM-2 recognizes a narrow region of DNA containing the cyclobutane thymine dimer. After cs23 was incubated with TDM-2, the complementary strand (AA23) was added to the solution containing the cs23-TDM-2 complex. Then, the cs23-TDM-2 complex was disrupted and cs23[middot]AA23 duplex was formed (data not shown). This result suggests that TDM-2 could not bind to a dsDNA in the presence of the completely complementary strand.
TDM-2, obtained by immunization using UV-irradiated DNA, indicated specific binding to the cis-syn type cyclobutane thymine dimer, but not to the trans-syn type cyclobutane dimer. For TDM-2, binding requires not only the cyclobutane dimer but also the neighboring nucleosides. TDM-2 could bind with NpT[c,s]TpN efficiently. Either the 5'- or 3'-side nucleoside of NpT[c,s]TpN was changed to a propandiol linker as an abasic analog while maintaining a phosphate diester. From competitive ELISA using these analogs, it was proved that the 5'- and 3'-side nucleosides were important for the recognition by TDM-2. Particularly, the existence of the 3'-side nucleoside affected the binding to TDM-2. Both the 5' and 3'-phosphates of the cyclobutane dimer were important for the binding to TDM-2; the cyclobutane dimers lacking a phosphate at either end were almost inactive. In a gel mobility shift assay, TDM-2 did not indicate binding to a dsDNA. However, the antibody could bind to duplexes with mispairs near the cyclobutane dimer.
TDM-2 has been shown to specifically recognize a certain span of DNA containing a cis-syn thymine cyclobutane dimer. The procedures for using the antigen analogs described here will be important for recognition studies of photo damaged DNA by various enzymes and for the construction of new antibodies with various functions.
This work was supported by the Special Coordination Funds for Promoting Science and Technology of the Science and Technology Agency of the Japanese Government, and by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan. The authors thank Drs K. Shimada, K. Katou and T. Matsunaga for the gift of TDM-2, and H. Morioka and H. Kobayashi for helpful discussions.
Competitor IC50 (nM)
cs23 0.7
ApT[c,s]TpA 1.0
GpT[c,s]TpG 0.5
CpT[c,s]TpC 0.8
TpT[c,s]TpT 1.2
pro[middot]pT[c,s]TpA 0.9
ApT[c,s]Tp[middot]pro 3.4
pro[middot]pT[c,s]Tp[middot]pro 20
T[c,s]Tp[middot]pro 1.7 * 102
pro[middot]pT[c,s]T 3.0 * 103
d(T[c,s]T) 1.6 * 104 
REFERENCES