II-OLIGODEOXYHUCUOTIDES (II-DRA):
A HEW CHDIEIlIC lIUCLEIC ACID AllALOG
F. Morvan, B. Rayner and J.-L. Imbach Laboratoire de Chimie Bio-Organique, UA 488 CNRS Universite des Sciences et Techniques du Languedoc Place E. Bataillon 34060 Montpellier Cedex 1, France INTRODUCTION Nucleic acids have long been strategic targets for approaches to chemotherapy in view of their roles in replication, transcription and transla tion (1). The use of synthetic oligonucleotides, which bind specifically to complementary sequences of nucleic acids (RNA or DNA) .through base pairing, is now under extensive investigation. In principle, relatively short oligomers (20 bases or less) can specifically hybridize wi th DNA or RNA and thus be used for drug design strategies involving targeted interference of genetic expression. However, potential chemotherapeutic applications resul ting from sequence-specific hybridization require analogs that are resistant to degradation by various nucleases. Oligonucleotide analogs presenting modifications on the phosphate backbone (i.e., methylphosphonates (2-5) and phosphorothioates (6» have been introduced and have been shown to have good resistance to enzymemediated depolymerization. However, such modifications introduce asymmetric linkages and, as the synthesis is not stereocontrolled, lead to a mixture of 2n stereoisomers (where n is the number of these linkages), each of them presenting different binding capacities. In addition, in the case of methylphosphonate the replacement of the negatively charged oxygen atom by a neutral group could interfere with eventual enzyma tic in terac tions (ionic linkage) and al so wi th the hydrophilic properties of these oligomers. These phosphate backbone modifications have always been considered starting from "natural" 2'-deoxy-{:!-D-ribofuranosyl nucleosides as constituting synthons. But another possibility is to consider sugar-modified
Genetic Engineering, Vol. 12 Edited by 1.K. Set10w Plenum Press. New York. 1990
37
38
F. MORVAN, B. RAYNER, ANDJ.-L.IMBACH
nucleosides as building blocks and to synthesize the corresponding oligomers with natural phosphodiester linkages. Let us consider first the structure of a natural 2'-deoxy-BD-ribofuranosyl nucleoside (Figure 1). This molecule presents three chiral carbon atoms in I', 3' and 4' positions. In order to maintain the geome try of the sugar phosphate backbone wi th a 3', 4'-transorientation, configurations of 3' and 4' carbon atoms must be changed together. This modification leads to 2'-deoxy-Lribonucleosides. The oligomers of L-2'-dU have been shown to be much more resistant to snake venom phosphodiesterase than the oligomers of D-2'-dU, although no evidence was found for binding of a l8-mer of L-2'-dU to poly(dA) (7). Inversion of the I' carbon a tom leads to 2' -deoxy-Cl-D-ribofuranosyl nucleosides. Only Cl-oligodeoxyribonucleotides (Cl-DNA), consisting exclusively of Cl-anomeric deoxyribonucleotides, will be considered in this chapter. In 1973, U. Sequin (8), using Dreiding models, considered the possibility of Cl-oligonucleotides exhibiting secondary structure similar to that of the natural nucleic acids, featuring base pairing, base stacking and helix formation. This study predicted that an Cl-strand may form a helix duplex with a complementary B- or Cl-strand by base pairing, and the two strands should exhibit parallel and antiparallel polarity, respectively. Shortly thereafter, isomeric dithymidine monophosphates bearing one or two Cl-dTs were found to exhibit nuclease resistance (9). The aim of this work, initiated in 1985, was to set up efficient synthesis of Cl-oligodeoxynucleotides and to study their physical and biologi~al properties. SYNTHESIS In 1986 Cl-oligonucleotides (hexamers) were first synthesized in solution with the phosphotriester approach (10). However, in this primary publication only pyrimidine nucleosides were used, but the method was rapidly extended to the four usual bases (11) and to a solid phase approach (12,13). As the reactivity of an Cl-nucleoside is very close to that of a B-anomer, the methodology used for synthesizing an Cloligonucleotide is the same as for the usual B-one. The main problem arises from the preparation of a sufficient amount of the starting Cl-nucleoside synthons, which are not commercially available except for Cl-dT.
H~VO~l' 3~ OH
Figure 1.
Structure of deoxy-2'-B-ribose.
39
a-OLIGODEOXYNUCLEOTIDES (a-DNA)
The nucleoside a-dT can ei ther be ob tained in 16% overall yield following an anomerization procedure as described (14) or with a 47% yield through a direct condensation of l-chloro-3,5di-p- toluyl-2 '-deoxyr ibofuranose wi th bi s tr ime thyl s i1 Y1 thymidine followed by base-catalyzed deprotection (15). Similarly, the N-6-benzoyl-a-dC can be obtained in a 26% yield through a self-anomerization procedure starting with 3' ,5'diace tyl-N-6- benzoyl- {3-dC (10). The a-2' -deoxy-adenos ine derivative can be made as described (14) through a transglycosyla tion procedure or by a direc t condensa tion me thod wi th 1- chI oro-3, 5-bis-p-ni trobenzoyl-2' -deoxyribofuranose. In this latter case, the yield of the N-6-benzoyl-a-dA is 34% (16). In any case, a-dG is more difficul t to obtain. The literature reported only a very low yield of the desired a-nucleoside (17), but a transglycosylation procedure with the use of N-2 protected guanine and a pro tec ted cytosine did increase the yield to about 29% (18). At this stage one can see that a-nucleoside building blocks can be obtained in large amount with appropriate synthesis. The phosphotriester approach in solution was used first to synthesize "large" amounts (i.e., 20 mg) of pure aoligomers for biophysical and NMR studies. In two different papers (10,11), we have repor ted the physical data of all required phosphotriester synthons. In Figure 2 is drawn the structure of a phosphotriester intermediate that we use. We have noticed that no striking difference (reac tion time, forma tion of side produc ts) could be de tec ted during a-oligonucleotide synthesis when compared to that of natural {3-o1igonucleotides (10,11). The automated solid phase approach was designed using the appropriate a-nucleoside phosphoramidites. Two kinds of phosphate protecting groups were used, the cyanoethyl (12) and
DatrO~B
?--r\)---J
cneo-g-o
Tr
Cl
B :
6-N-benzoyladenine. 4-N-benzoylcytosine.
2-N-palmitoylguanine. thymine.
ene, 2-cyanoethyl; Dmtr, 4,4'dimethoxytrityl; Tr, trityl.
Figure 2.
Structure of an a-phosphotriester intermediate.
40
F. MORVAN, B. RAYNER, AND J.-L.IMBACH
the methyl (13). The necessary synthons were synthesized following the usual procedure and their physical da ta have been described (13). Fractosil 500 (12) and long chain alkylamino-CPG (13) were used as solid supports, and derivatized as described for the B-anomers. The amount of loaded nucleoside was 26 to 27 ~mol per gram in the latter case. It was possible to assemble aoligomers automatically with an automatic DNA synthesizer either with the same cycle as for a B-anomer (but only aoligonucleotides which contain A, C and T have been described) (12) or with a slightly modified one to take into account the poor solubility of the a-guanosine phosphoramidites (13). Using such methodology we have synthesized and described a 20-mer containing the four usual bases (13); it was obtained in 29% yield after purification with an average coupling yield of 98.3% and its primary structure was confirmed by Maxam and Gilbert sequence analysis. These data show that a-oligonucleotides of any length can be obtained with the usual a-oligonucleotide methodology and that the starting a building blocks are readily available in multigram amounts. STRUCTURE AND PHYSICO-CHEMICAL PROPERTIES Single-stranded a-Oligodeoxynucleotides The first a sequence which was examined, Le., a-[d(CCTTCC)], contained only pyrimidine residues (10) and was investigated by NMR spectroscopy. It was shown that the cytosine and thymine bases adopt an anti conformation at the glycosyl bonds. In addition, the deoxyriboses of the thymidines adopt an average conformation approximating C3, -~, while the cytidine furanose groups are close to C2, -~. Further studies on heterogeneous sequences a-[d(CATGCG)] and a-[d(CGCATG)] confirmed the anti conformation of the bases, the sugar rings being C3,exo (i9). It was then shown, based upon hypochromicity ~siderations, that the a-[d(CATGCG)] exhibits a higher level of base-base interaction than the corresponding B-strand. But this sequence is also able to self-anneal, thus complicating the interpretation (20). For instance, in the case of an homogenous a-oligothymidylate: a-[d(TphT], CD spectra indicate base-base interactions whose thermodynamic parameters are of the same order of magnitude as those of the B-[(pT)8] even though this interaction is formed with a different geometry (21). NMR studies on a-[d(CATGCG)] and on a-[d(CGCATG)] give some insight into the self-annealing of a-strands (20). We were able to detect base pairing due to self-annealing and to conclude that there was antiparallel self-recognition of these a-strands, whose thermodynamic stability was compatible with Watson-Crick rather than Hoogsteen base pairing. Once it was established that a-
a-OLIGODEOXYNUCLEOTIDES (a-DNA)
41
oligonucleotides are able to undergo strong antiparallel selfannealing, the important question of annealing to a complementary a-strand remained to be answered. Formation of a-a-Double Strands Much work dealing with this question allows one to state unambiguously that a-oligonucleotides are able to form stable duplexes with their a-counterparts. Both hypochromicity in thermal denaturation (10,22) and detection of base-paired imino protons in lH -NMR studies provide evidence for the annealing of a-[d(CCTTCC)] with a-[d(GGAAGG)]. Melting experiments (22) lead to the same conclusion for a-[d(GGAAGG)] and a-[d)CCTTCC)]. Fur the rmore , the annealing 0 f a- [ d ( CATGCG) ] with a- [ d (GTACGC) ] (19) and of a-[d(TCTAAAC)] with a-[d(AGATTTG)] (23) were demonstrated and studied in detail with the use of NMR. Applying l-D and 2-D strategies, the authors derived a right-helical structure for those complexes wi th the orientation of the bases at the glycosyl bond with respect to the sugar moiety being anti (19,23). Furthermore, the pucker of the sugar moiety has been found to be 2'-endo-3'-exo (19) or 3'-exo (23). As far as the polarity of thecomplexis concerned, the authors propose a parallel orientation of the two chains. Other work dealing with the annealing of a-oligothymidylate and its a-complement, with techniques different from that of NMR: photo-crosslinking reactions (24), circular dichroism spectra (21) and fluorescence (25), lead to the same conclusion that double helices are formed wi th parallel s'trands. Furthermore, it has been shown (22) that an a-hexamer forms a complex with its complementary a-sequence that is more stable than the corresponding a ,a-complex. This conclusion raises the question of the stability of the unnatural Cl,a-duplex compared to the natural a,a one. Stability of Cl-a-Oligonucleotide Duplexes The melting temperature of the duplex Cl-[(Tp)7T]: a-[d(pA)S] is reported to be 3l o C compared to 24 0 C for the corresponding a[(Tp)S]: a-[d(Ap)S] (21), suggesting that Cl,a-annealing is thermodynamically more stable than a ,a-anneal ing. On the other hand the a strand containing a stretch of nine thymidines flanked by two guanosines at each end, Cl-[d(G2T9G2)] hybridized with its complementary a-strand a-[d(C2A9C2)] less readily (Tm = 34 0 C) than the counterpart a-[d(G2T9G2)] (Tm = 40 0 C) (lS). This result suggests that Cl-guanosine may destabilize the Cl,a duplexes. However, mixing of Cl-[d(TCTAAAC)] with a-[d(AGATTTG)] produces a duplex whose melting temperature is 33 0 C (2 mM/2 mM) compared to 36 0 C for the corresponding a-[d(CAAATCT)]: a-[d(AGATTTG)] duplex (2.5 mM/2.5 mM), suggesting that Cl,a-annealing has a stability close to that of the a,a duplex (23). Detailed thermodynamics
42
F. MORVAN, B. RAYNER, AND J.-L.IMBACH
studies on the duplexes a-[d(CCT-TCC)]: S-[d(GGAAGG)] and a[d(GGAAGG)]:S-[d(CCTTCC)] lead to the conclusion that their stabilities depend upon the nature of the bases (purines or pyrimidines) involved in the building of the a-strand, at least for homogeneous purine or pyrimidine sequences (22). Comparison of the thermal denaturation of S-[d(CCTTCC)]: S-[d(GGAAGG)] and a-[d(CCTTCC)]: S[d(GGAAGG)] duplexes indicates that the a,Sduplex is more stable (Tma,S = 28 0 C) than the S,S-duplex (TmS,S = 19 0 ) in 1M NaCl. However, in the symmetrical situation where the purine strand is made of unnatural a-nucleotides, a-[d(GGAAGG)], the stability of the duplex is in favor of the S,S-system (Tma,S = l4 0 C for a-[d(GGAAGG)]: S-[d(CCTTCC)]). This result leads to a conclusion that the stability of the a ,S-anneal ing may be dependent upon base composi tion of the asequence involved in the formation of the unnatural duplex. It al so appears that base pairing formed wi th a-dG and S-dC may destabilize the a-S duplexes. The stability of the a,S duplexes raises the question of the annealing selectivity of such duplexes. Annealing Selectivity of a-Oligodeoxynucleotides Prel iminary resul ts obtained through mel ting exper imen ts showed the influence of the nature of a single central mismatch on a,S heteroduplex stability, in comparison with that observed in the corresponding S,S duplexes (Table 1) (18). The influence Table 1
A) ~-d(C2AAAAXAAAAC2
X
)
~-d(G2TTTTTTTTTG2
)
~-d(G2TTTTTTTTTG2
=A
33.9°C
6Tm
40.0°C
6Tm
= C
24.2°C
-9.7
27.5°C
-12.5
= G
25.9°C
-S.O
30.SoC
-9.2
= T
25.3°C
-S.6
2S.1°C
-11.9
)
B) ~-d(TTTTTTXTTTTTT)
X
= = = =
T C G A
poly rA 6Tm 43.5°C 36.6°C -6.9 31.6°C -11.9 -9.9 33.6°C
Melting temperature of A): a-DNA;S-DNA and B) a-DNA;S-RNA duplexes. The concen tra tion of each 01 igonucleo tide is 311M in buffer: O.OlM sodium cacodylate, O.lM NaCl, pH 7. lITm represents the variation in Tm between a duplex bearing a single mismatch and the corresponding perfectly matched duplex.
ex-OLIGODEOXYNUCLEOTIDES (ex-DNA)
43
of a l3-dX,
A HEW CHDIEIlIC lIUCLEIC ACID AllALOG
F. Morvan, B. Rayner and J.-L. Imbach Laboratoire de Chimie Bio-Organique, UA 488 CNRS Universite des Sciences et Techniques du Languedoc Place E. Bataillon 34060 Montpellier Cedex 1, France INTRODUCTION Nucleic acids have long been strategic targets for approaches to chemotherapy in view of their roles in replication, transcription and transla tion (1). The use of synthetic oligonucleotides, which bind specifically to complementary sequences of nucleic acids (RNA or DNA) .through base pairing, is now under extensive investigation. In principle, relatively short oligomers (20 bases or less) can specifically hybridize wi th DNA or RNA and thus be used for drug design strategies involving targeted interference of genetic expression. However, potential chemotherapeutic applications resul ting from sequence-specific hybridization require analogs that are resistant to degradation by various nucleases. Oligonucleotide analogs presenting modifications on the phosphate backbone (i.e., methylphosphonates (2-5) and phosphorothioates (6» have been introduced and have been shown to have good resistance to enzymemediated depolymerization. However, such modifications introduce asymmetric linkages and, as the synthesis is not stereocontrolled, lead to a mixture of 2n stereoisomers (where n is the number of these linkages), each of them presenting different binding capacities. In addition, in the case of methylphosphonate the replacement of the negatively charged oxygen atom by a neutral group could interfere with eventual enzyma tic in terac tions (ionic linkage) and al so wi th the hydrophilic properties of these oligomers. These phosphate backbone modifications have always been considered starting from "natural" 2'-deoxy-{:!-D-ribofuranosyl nucleosides as constituting synthons. But another possibility is to consider sugar-modified
Genetic Engineering, Vol. 12 Edited by 1.K. Set10w Plenum Press. New York. 1990
37
38
F. MORVAN, B. RAYNER, ANDJ.-L.IMBACH
nucleosides as building blocks and to synthesize the corresponding oligomers with natural phosphodiester linkages. Let us consider first the structure of a natural 2'-deoxy-BD-ribofuranosyl nucleoside (Figure 1). This molecule presents three chiral carbon atoms in I', 3' and 4' positions. In order to maintain the geome try of the sugar phosphate backbone wi th a 3', 4'-transorientation, configurations of 3' and 4' carbon atoms must be changed together. This modification leads to 2'-deoxy-Lribonucleosides. The oligomers of L-2'-dU have been shown to be much more resistant to snake venom phosphodiesterase than the oligomers of D-2'-dU, although no evidence was found for binding of a l8-mer of L-2'-dU to poly(dA) (7). Inversion of the I' carbon a tom leads to 2' -deoxy-Cl-D-ribofuranosyl nucleosides. Only Cl-oligodeoxyribonucleotides (Cl-DNA), consisting exclusively of Cl-anomeric deoxyribonucleotides, will be considered in this chapter. In 1973, U. Sequin (8), using Dreiding models, considered the possibility of Cl-oligonucleotides exhibiting secondary structure similar to that of the natural nucleic acids, featuring base pairing, base stacking and helix formation. This study predicted that an Cl-strand may form a helix duplex with a complementary B- or Cl-strand by base pairing, and the two strands should exhibit parallel and antiparallel polarity, respectively. Shortly thereafter, isomeric dithymidine monophosphates bearing one or two Cl-dTs were found to exhibit nuclease resistance (9). The aim of this work, initiated in 1985, was to set up efficient synthesis of Cl-oligodeoxynucleotides and to study their physical and biologi~al properties. SYNTHESIS In 1986 Cl-oligonucleotides (hexamers) were first synthesized in solution with the phosphotriester approach (10). However, in this primary publication only pyrimidine nucleosides were used, but the method was rapidly extended to the four usual bases (11) and to a solid phase approach (12,13). As the reactivity of an Cl-nucleoside is very close to that of a B-anomer, the methodology used for synthesizing an Cloligonucleotide is the same as for the usual B-one. The main problem arises from the preparation of a sufficient amount of the starting Cl-nucleoside synthons, which are not commercially available except for Cl-dT.
H~VO~l' 3~ OH
Figure 1.
Structure of deoxy-2'-B-ribose.
39
a-OLIGODEOXYNUCLEOTIDES (a-DNA)
The nucleoside a-dT can ei ther be ob tained in 16% overall yield following an anomerization procedure as described (14) or with a 47% yield through a direct condensation of l-chloro-3,5di-p- toluyl-2 '-deoxyr ibofuranose wi th bi s tr ime thyl s i1 Y1 thymidine followed by base-catalyzed deprotection (15). Similarly, the N-6-benzoyl-a-dC can be obtained in a 26% yield through a self-anomerization procedure starting with 3' ,5'diace tyl-N-6- benzoyl- {3-dC (10). The a-2' -deoxy-adenos ine derivative can be made as described (14) through a transglycosyla tion procedure or by a direc t condensa tion me thod wi th 1- chI oro-3, 5-bis-p-ni trobenzoyl-2' -deoxyribofuranose. In this latter case, the yield of the N-6-benzoyl-a-dA is 34% (16). In any case, a-dG is more difficul t to obtain. The literature reported only a very low yield of the desired a-nucleoside (17), but a transglycosylation procedure with the use of N-2 protected guanine and a pro tec ted cytosine did increase the yield to about 29% (18). At this stage one can see that a-nucleoside building blocks can be obtained in large amount with appropriate synthesis. The phosphotriester approach in solution was used first to synthesize "large" amounts (i.e., 20 mg) of pure aoligomers for biophysical and NMR studies. In two different papers (10,11), we have repor ted the physical data of all required phosphotriester synthons. In Figure 2 is drawn the structure of a phosphotriester intermediate that we use. We have noticed that no striking difference (reac tion time, forma tion of side produc ts) could be de tec ted during a-oligonucleotide synthesis when compared to that of natural {3-o1igonucleotides (10,11). The automated solid phase approach was designed using the appropriate a-nucleoside phosphoramidites. Two kinds of phosphate protecting groups were used, the cyanoethyl (12) and
DatrO~B
?--r\)---J
cneo-g-o
Tr
Cl
B :
6-N-benzoyladenine. 4-N-benzoylcytosine.
2-N-palmitoylguanine. thymine.
ene, 2-cyanoethyl; Dmtr, 4,4'dimethoxytrityl; Tr, trityl.
Figure 2.
Structure of an a-phosphotriester intermediate.
40
F. MORVAN, B. RAYNER, AND J.-L.IMBACH
the methyl (13). The necessary synthons were synthesized following the usual procedure and their physical da ta have been described (13). Fractosil 500 (12) and long chain alkylamino-CPG (13) were used as solid supports, and derivatized as described for the B-anomers. The amount of loaded nucleoside was 26 to 27 ~mol per gram in the latter case. It was possible to assemble aoligomers automatically with an automatic DNA synthesizer either with the same cycle as for a B-anomer (but only aoligonucleotides which contain A, C and T have been described) (12) or with a slightly modified one to take into account the poor solubility of the a-guanosine phosphoramidites (13). Using such methodology we have synthesized and described a 20-mer containing the four usual bases (13); it was obtained in 29% yield after purification with an average coupling yield of 98.3% and its primary structure was confirmed by Maxam and Gilbert sequence analysis. These data show that a-oligonucleotides of any length can be obtained with the usual a-oligonucleotide methodology and that the starting a building blocks are readily available in multigram amounts. STRUCTURE AND PHYSICO-CHEMICAL PROPERTIES Single-stranded a-Oligodeoxynucleotides The first a sequence which was examined, Le., a-[d(CCTTCC)], contained only pyrimidine residues (10) and was investigated by NMR spectroscopy. It was shown that the cytosine and thymine bases adopt an anti conformation at the glycosyl bonds. In addition, the deoxyriboses of the thymidines adopt an average conformation approximating C3, -~, while the cytidine furanose groups are close to C2, -~. Further studies on heterogeneous sequences a-[d(CATGCG)] and a-[d(CGCATG)] confirmed the anti conformation of the bases, the sugar rings being C3,exo (i9). It was then shown, based upon hypochromicity ~siderations, that the a-[d(CATGCG)] exhibits a higher level of base-base interaction than the corresponding B-strand. But this sequence is also able to self-anneal, thus complicating the interpretation (20). For instance, in the case of an homogenous a-oligothymidylate: a-[d(TphT], CD spectra indicate base-base interactions whose thermodynamic parameters are of the same order of magnitude as those of the B-[(pT)8] even though this interaction is formed with a different geometry (21). NMR studies on a-[d(CATGCG)] and on a-[d(CGCATG)] give some insight into the self-annealing of a-strands (20). We were able to detect base pairing due to self-annealing and to conclude that there was antiparallel self-recognition of these a-strands, whose thermodynamic stability was compatible with Watson-Crick rather than Hoogsteen base pairing. Once it was established that a-
a-OLIGODEOXYNUCLEOTIDES (a-DNA)
41
oligonucleotides are able to undergo strong antiparallel selfannealing, the important question of annealing to a complementary a-strand remained to be answered. Formation of a-a-Double Strands Much work dealing with this question allows one to state unambiguously that a-oligonucleotides are able to form stable duplexes with their a-counterparts. Both hypochromicity in thermal denaturation (10,22) and detection of base-paired imino protons in lH -NMR studies provide evidence for the annealing of a-[d(CCTTCC)] with a-[d(GGAAGG)]. Melting experiments (22) lead to the same conclusion for a-[d(GGAAGG)] and a-[d)CCTTCC)]. Fur the rmore , the annealing 0 f a- [ d ( CATGCG) ] with a- [ d (GTACGC) ] (19) and of a-[d(TCTAAAC)] with a-[d(AGATTTG)] (23) were demonstrated and studied in detail with the use of NMR. Applying l-D and 2-D strategies, the authors derived a right-helical structure for those complexes wi th the orientation of the bases at the glycosyl bond with respect to the sugar moiety being anti (19,23). Furthermore, the pucker of the sugar moiety has been found to be 2'-endo-3'-exo (19) or 3'-exo (23). As far as the polarity of thecomplexis concerned, the authors propose a parallel orientation of the two chains. Other work dealing with the annealing of a-oligothymidylate and its a-complement, with techniques different from that of NMR: photo-crosslinking reactions (24), circular dichroism spectra (21) and fluorescence (25), lead to the same conclusion that double helices are formed wi th parallel s'trands. Furthermore, it has been shown (22) that an a-hexamer forms a complex with its complementary a-sequence that is more stable than the corresponding a ,a-complex. This conclusion raises the question of the stability of the unnatural Cl,a-duplex compared to the natural a,a one. Stability of Cl-a-Oligonucleotide Duplexes The melting temperature of the duplex Cl-[(Tp)7T]: a-[d(pA)S] is reported to be 3l o C compared to 24 0 C for the corresponding a[(Tp)S]: a-[d(Ap)S] (21), suggesting that Cl,a-annealing is thermodynamically more stable than a ,a-anneal ing. On the other hand the a strand containing a stretch of nine thymidines flanked by two guanosines at each end, Cl-[d(G2T9G2)] hybridized with its complementary a-strand a-[d(C2A9C2)] less readily (Tm = 34 0 C) than the counterpart a-[d(G2T9G2)] (Tm = 40 0 C) (lS). This result suggests that Cl-guanosine may destabilize the Cl,a duplexes. However, mixing of Cl-[d(TCTAAAC)] with a-[d(AGATTTG)] produces a duplex whose melting temperature is 33 0 C (2 mM/2 mM) compared to 36 0 C for the corresponding a-[d(CAAATCT)]: a-[d(AGATTTG)] duplex (2.5 mM/2.5 mM), suggesting that Cl,a-annealing has a stability close to that of the a,a duplex (23). Detailed thermodynamics
42
F. MORVAN, B. RAYNER, AND J.-L.IMBACH
studies on the duplexes a-[d(CCT-TCC)]: S-[d(GGAAGG)] and a[d(GGAAGG)]:S-[d(CCTTCC)] lead to the conclusion that their stabilities depend upon the nature of the bases (purines or pyrimidines) involved in the building of the a-strand, at least for homogeneous purine or pyrimidine sequences (22). Comparison of the thermal denaturation of S-[d(CCTTCC)]: S-[d(GGAAGG)] and a-[d(CCTTCC)]: S[d(GGAAGG)] duplexes indicates that the a,Sduplex is more stable (Tma,S = 28 0 C) than the S,S-duplex (TmS,S = 19 0 ) in 1M NaCl. However, in the symmetrical situation where the purine strand is made of unnatural a-nucleotides, a-[d(GGAAGG)], the stability of the duplex is in favor of the S,S-system (Tma,S = l4 0 C for a-[d(GGAAGG)]: S-[d(CCTTCC)]). This result leads to a conclusion that the stability of the a ,S-anneal ing may be dependent upon base composi tion of the asequence involved in the formation of the unnatural duplex. It al so appears that base pairing formed wi th a-dG and S-dC may destabilize the a-S duplexes. The stability of the a,S duplexes raises the question of the annealing selectivity of such duplexes. Annealing Selectivity of a-Oligodeoxynucleotides Prel iminary resul ts obtained through mel ting exper imen ts showed the influence of the nature of a single central mismatch on a,S heteroduplex stability, in comparison with that observed in the corresponding S,S duplexes (Table 1) (18). The influence Table 1
A) ~-d(C2AAAAXAAAAC2
X
)
~-d(G2TTTTTTTTTG2
)
~-d(G2TTTTTTTTTG2
=A
33.9°C
6Tm
40.0°C
6Tm
= C
24.2°C
-9.7
27.5°C
-12.5
= G
25.9°C
-S.O
30.SoC
-9.2
= T
25.3°C
-S.6
2S.1°C
-11.9
)
B) ~-d(TTTTTTXTTTTTT)
X
= = = =
T C G A
poly rA 6Tm 43.5°C 36.6°C -6.9 31.6°C -11.9 -9.9 33.6°C
Melting temperature of A): a-DNA;S-DNA and B) a-DNA;S-RNA duplexes. The concen tra tion of each 01 igonucleo tide is 311M in buffer: O.OlM sodium cacodylate, O.lM NaCl, pH 7. lITm represents the variation in Tm between a duplex bearing a single mismatch and the corresponding perfectly matched duplex.
ex-OLIGODEOXYNUCLEOTIDES (ex-DNA)
43
of a l3-dX,
