Chemical Etiology of Nucleic Acid Structure: The α-Threofuranosyl-(3'→2') Oligonucleotide System

See allHide authors and affiliations

Science  17 Nov 2000:
Vol. 290, Issue 5495, pp. 1347-1351
DOI: 10.1126/science.290.5495.1347


TNAs [(l)-α-threofuranosyl oligonucleotides] containing vicinally connected (3′→2′) phosphodiester bridges undergo informational base pairing in antiparallel strand orientation and are capable of cross-pairing with RNA and DNA. Being derived from a sugar containing only four carbons, TNA is structurally the simplest of all potentially natural oligonucleotide-type nucleic acid alternatives studied thus far. This, along with the base-pairing properties of TNA, warrants close scrutiny of the system in the context of the problem of RNA's origin.

The strategy pursued in our work to establish a chemical etiology of nucleic acid structure (1) demands systematic screening of the base-pairing properties of potentially natural, sugar-based nucleic acid alternatives recruited from the structural neighborhood of RNA. The β-hexopyranosyl-(6′→4′) oligonucleotide analogs of RNA (2) derived from the hexose sugars allose, altrose, and glucose display base pairing far inferior to that of RNA with respect to both pairing strength and pairing-mode specificity (3). This behavior was interpreted to be a consequence of the steric bulk of fully hydroxylated hexopyranosyl sugar units (4). Consequently, the focus of our studies shifted toward potentially natural RNA alternatives derived from the sterically less bulky pentopyranose sugars. There it was found that not only the pyranosyl isomer of RNA (5), but a whole family of diastereomeric pentopyranosyl-(4′→2′) oligonucleotide systems (6) show Watson-Crick base pairing that is uniformly stronger than that of RNA itself.

The RNA analogs derived from tetrose instead of pentose sugar units were not considered to be candidates, because an oligonucleotide backbone that contains six covalent bonds per repeating mononucleotide unit (as RNA does) cannot be constructed with an aldosugar containing only four carbons (7). However, recent observations in the pentopyranosyl series changed that perspective: switching from a (4′→2′) pentopyranosyl to a (4′→3′) pentopyranosyl system was expected to result in the loss of base pairing due to shortening the phosphodiester bridge from six to five bonds. This was observed in the β-ribopyranosyl series, but not in the α-lyxopyranosyl series (8). In the latter base-pairing system, the vicinal (4′→3′) phosphodiester bridge assumes a diaxial conformation at the pyranose chairs. This led us to extend our investigations to the tetrose series, because α-threofuranosyl-(3′→2′) oligonucleotides may behave as conformational analogs of α-lyxopyranosyl-(4′→3′) oligonucleotides with regard to a quasi-diaxial positioning of their phosphodiester bridge at the furanose half-chairs (Scheme 1).

Figure 1

X-ray structure of (l)-α-threofuranosyl nucleoside derivatives (14). Torsion angles (max ± 0.5 ° are as follows): O-C2'-C3'-O: 164.9° in 3a, 158.6° in 3b, 165.7° in 3c, 2′,3′-dibenzoate, 131.5° in 3d, 160.2° in 3e, 2′,3′-dibenzoate. O-C1'-N9-C4: −169.3° in3a, −176.7° in 3e, 2′,3′-dibenzoate. O-C1'-N1-C2: −147.3° in3b, −166.4° in 3c, 2′,3′-dibenzoate, −169.28° in 3d. For the chemical formula's of3a–e, see Scheme 2.

Figure 2

Data of TNA duplexes under conditions as follows: 1.0 M NaCl, 10 mM NaH2PO4, 0.1 mM Na2EDTA, pH 7. (A) UV melting curve of the duplex t(A16) + t(T16) (c ≈ 5 μM + 5 μM), Heating curves of single strands are also shown. (B) Mixing curve (19) for the pairing between t(A16) and t(T16) demonstrating 1:1 stoichiometry, total c ≈ 3.6 μM, measured at 10°C. (C) Comparison of CD spectra of the individual strands t(A16) and t(T16) and the duplex t(A16) + t(T16) (c ≈ 1 μM at 30°C). (D) UV melting curves of (1:1)-mixtures of hexadecamer sequences complementary to each other in either antiparallel or parallel strand orientation, c = 10 and 20 μM, respectively. (E) UV melting curves andT m values of (1:1)-mixtures of hexadecamer sequences in intra- and intersystem cross-pairing in and between TNA and RNA (c ≈ 5 μM + 5 μM). (F) Temperature- dependent CD spectra of a hexadecamer duplex formed by cross-pairing between TNA and RNA (c≈ 5 μM + 5 μM; T = 0° to 90°C).

Scheme 1

Constitution, configuration, and conformation (with linearized backbone) of an (l)-α-threofuranosyl-(3′→2′) oligonucleotide strand in comparison with RNA. Equatorial substituents lie in averaged plane of pyranosyl chair, axial substituents perpendicular to it.

Here we describe the synthesis and base-pairing properties of TNA oligonucleotides containing the five canonical purine and pyrimidine nucleobases. The tetrose-based oligonucleotides indeed show efficient base pairing which is similar to that of pentose-based RNA with regard to specificity, strand orientation, and pairing strength. In addition, TNA oligonucleotides of the l-series are capable of cross-pairing with RNA and DNA. This is in contrast to all the previously studied potentially natural nucleic acid alternatives of the (6′→4′) hexopyranosyl and the (4′→2′) pentopyranosyl families, where base pairing—in systems where it occurs—is orthogonal to that of the natural nucleic acids (9).

The synthesis of TNA oligonucleotides follows the methodology we used earlier in other oligonucleotide series (4–6). Starting materials5a-e were prepared from (l)- threose (10) according to Scheme 2(11). All nucleosidations proceed with high diastereoselectivity to give α-nucleoside-2′,3′-dibenzoates 2a-e in high yields. Tritylation steps 3a-e4a-e show low regioselectivity and require chromatographic separations of the major 3′-tritylated isomers from their 2′ analogs. For the thymine and uracil members 4b and4c, where this separation gives the desired 3′ isomers in unsatisfactory yield, an alternative route has been developed which proceeds via intermediates 6b and 6c and7b and 7c and produces the 3′-tritylated derivatives 4b and 4c selectively and in good overall yield.

Scheme 2

Preparation of building blocks for the synthesis of (l)-α-threofuranosyl-oligonucleotides (11). Bz, benzoyl; Ac, acetyl; Dpc, diphenylcarbamoyl; BSA, N,O-bis(trimethylsilyl)acetamide; TMSOTf, trimethylsilyl trifluoromethanesulfonate; DMT, 4,4′-dimethoxy triphenylmethyl; AgOTf, silver trifluoromethane sulfonate; DMF, dimethyl formamide; HMPA, hexamethylphosphoramide; RT, room temperature. Numbers before reagents denote mole equivalents (mol-equiv.); % denotes yields of isolated products. 1R=Bz→2a: 1.1 6-N-benzoyl adenine, 2.0 BSA in CH3CN at 70°C for 1 hour, followed by 3.0 mol-equiv. SnCl4 at 70°C for 1.5 hours, 91%; 1 R=Bz (1 R=Bz, 1 R=Ac)→2b(2c, 2d): 1.1 thymine (1.0 uracil, 1.1 4-N-benzoyl cytosine), 2.0 (2.2, 2.5) BSA in CH3CN at 70°C for 1 hour, followed by 3.0 TMSOTf at 70°C for 1.5 (2, 1) hours, 92% (83, 94); 1R=Ac→2e: 2-N-acetyl-6-O-diphenylcarbamoyl guanine [for method see (12)], 2 BSA in CH2Cl2 at 70°C for 1 hour, followed by 0.91 R=Ac, 2 TMSOTf, in toluene at 70°C for 2.5 hours, 64%; 2a (2b, 2d,2e)→3a (3b, 3d,3e): 2.4 (2.3, 3.3, 8) NaOH in THF/MeOH/H2O 5:4:1, 0°C, 15 min (30, 30, 15), 85% (91, 96, 58);2c3c: in MeOH/H2O/Et3N 5:1:1, reflux for 3 hours, 98%; 3a (3b, 3c,3d)→4a (4b, 4c,4d): 1.3 DMTCl, 5 2,6-lutidine, 1.2 AgOTf in CH2Cl2/DMF 1:1 at 20°C for 5 (2, 5, 5) hours, followed by 0.1 (0.2, 0.1, 0.1) DMTCl, 0.1 (0.2, 0.1, 0.1) AgOTf at RT, overnight, 66% (25, 22, 45); 3e4e: 2.1 DMTCl, 6 lutidine in CH2Cl2/DMF 5:1 at RT for 3 days, 23%; 4a (4b, 4c,4d)→5a (5b, 5c, 5d): 1.1 P((i-Pr)2N)(OCH2CH2CN)Cl, 5.6 ethyldiisopropylamine in CH2Cl2 at RT, overnight, followed by 0.1 P((i-Pr)2N)(OCH2CH2CN)Cl for 2 hours, 64% (83, 74, 69); 4e5e: 2.5 P((i-Pr)2N)(OCH2CH2CN)Cl, 10N-ethyldimethylamine, CH2Cl2 at RT for 90 min, 67%. 3b (3c)→6b(6c): 2.1 diphenylcarbonate, 0.3 NaHCO3, HMPA at 150°C for 3 hours, 79% (92) [for method, see (13)]; 6b(6c)→7b (7c): 4.0 BzONa, 1.1 BzOH in HMPA at 150°C for 2 hours, 68% (77);7b (7c)→4b (4c): (a) 1.5 DMTCl, 6 2,4,6-collidine in CH2Cl2at RT for 16 (12) hours, 89% (96) (b) 2M NH3 in MeOH at RT for 24 hours, 95% (93).

X-ray structure analyses carried out on α-threofuranosyl mononucleoside derivatives containing the adenine, thymine, uracil, cytosine, or guanine nucleus (14) (Fig. 1) reveal a threofuranose conformation in which the two substituents at positions 2′ and 3′ indeed assume a quasi-diaxial orientation (torsion angles typically between 158° and 169°).

Table 1 summarizes the melting temperatures (Tm values ) (15), determined by ultraviolet (UV) spectroscopy, of TNA oligonucleotide duplexes as well as the thermodynamic data for their formation from corresponding single strands (16). Duplex formation was also characterized by temperature-dependent circular dichroism (CD) spectroscopy, and strand stoichiometry was confirmed by UV-spectroscopic mixing curves for selected examples (Fig. 2). Base pairing in TNA strictly demands antiparallel strand orientation (Fig. 2D). Duplex stabilities show a characteristic sequence-motif dependence: Duplexes with strands composed of regularly alternating purine-pyrimidine bases (Nos. 3, 4, and 5 in Table 1) are more stable than their isomers containing the base pairs as block oligomers (Nos. 1 and 2 in Table 1). This observation is consistent with the finding that base pairing is weakest in duplexes composed of homobasic purine and pyrimidine strands (Nos. 16 through 19 in Table 1).

Table 1

Tm values (15) in degrees Celsius and thermodynamic data of TNA duplexes determined under conditions as follows: c ≈ 5 μM + 5 μM, 1.0 M NaCl, 10 mM NaH2PO4, 0.1 mM Na2EDTA, pH 7. The labels RNA and DNA refer to duplexes consisting of RNA-RNA and DNA-DNA strands, respectively. Thermodynamic parameters were determined from plots of Tm−1versus lnc and for hairpin sequence byTm curve differenti- ation at c ≈ 5.0 μM [for methods see (15)]. ΔG, Gibbs free energy; ΔH, change in enthalpy (estimated experimental error for ΔH, ±5%); ΔS, change in entropy. For the automated synthesis of TNA-oligonucleotides see (16). Formation of hairpin (No. 15) is deduced from invariance of Tm (67.2° to 65.6°C) with variation in oligonucleotide concentration (2 to 111 μM).

View this table:

With strands containing mixed purine-pyrimidine sequences of moderate length, TNA duplexes can show a thermal (but not necessarily thermodynamic) stability that is comparable to that of RNA or DNA duplexes. This is exemplified by the data for the listed duplexes (Nos. 13, 14, 20, and 21 in Table 1) and is further documented in Fig. 2 andTable 2. In contrast, base pairing of shorter oligomers (e.g., Nos. 12 and 22 in Table 1), especially when they contain homobasic sequences (e.g., No. 16 in Table 1), gives rise to duplexes that are less stable than in RNA (Table 1). Hairpins seem to form as readily in the TNA series as they do in the natural series (No. 15 in Table 1).

Table 2

Tm values of duplexes A, B, C, and D formed by intra- and intersystem cross-pairing involving TNA, RNA, and DNA strands (conditions are as in Table 1). The color of the acronyms TNA, DNA, and RNA refer to the oligonucleotide sequences of the same color in the formulas of the duplexes A, B, C, and D shown at the bottom of the table. The labels 3′ and 2′ indicate strand orientation referring to TNA duplexes; for RNA and DNA duplexes, these labels must be replaced correspondingly by the labels 5′ and 3′.Tm values in the shaded diagonal refer to intrasystem cross-pairing.

View this table:

TNA cross-pairs efficiently with RNA and DNA. This is remarkable in view of the pronounced constitutional difference between the sugar backbones of TNA and the natural nucleic acids. The efficiency of the cross-pairing is documented by the UV melting curves and temperature-dependent CD spectra reproduced in Fig. 2, E and F, as well as by the overview of T m values given in Table 2for the four (non–self-complementary) base sequences A, B, C, and D. All these observations strongly suggest that the base-pairing mode of TNA is that of the Watson-Crick model.

Among the cross-pairing data is a striking characteristic difference in the behavior of homobasic versus heterobasic sequences. Whereas all intra- and intersystem combinations of the representative heterobasic hexadecamer sequences B and C give rise to comparable thermal duplex stabilities, these stabilities differ widely in corresponding combinations with the homobasic hexadecamers A16 and T16, depending on whether the all-purine or the all-pyrimidine sequence carries the TNA backbone in the TNA-RNA heteroduplex. The T m value of the t(A16) + r(T16) combination (T m = 76°C, c = 5 + 5 μM) markedly exceeds even that of the RNA-homoduplex r(A16) + r(T16) (T m = 62°C), in sharp contrast to the t(T16) + r(A16) combination, which is much less stable (T m = 28°C). The same phenomenon is also observed in the cross-pairing of TNA with DNA (20).

Not unexpectedly, TNA is much more stable toward hydrolytic cleavage of the phosphodiester linkage than RNA; under conditions (1.0 M NaCl; 0.25 M MgCl2, 0.1 M Hepes buffer, pH = 8, at 35°C) in which a half-life of about half a day was observed for r(U8), and one of about 4 days for pyranosyl-r(T8) (21), the TNA-oligonucleotide t(T8) remains unchanged over months. TNA's stability against hydrolytic decomposition may well be similar to that of DNA (22).

TNA is deemed a potentially natural nucleic acid alternative according to the criteria defined earlier (1). In this respect, the system differs from most artificial nucleic acid analogs that also cross-pair with the natural systems, but were constructed according to the demands of antisense technology (7, 23). Unlike the nucleic acid alternatives that we have studied previously, TNA could potentially serve as a template in nonenzymic template-directed formation of RNA sequences. This property remains to be experimentally tested (24). TNA also stands out among these alternatives—RNA included—with regard to its chemical prospects for constitutional self-assembly. The TNA structure allows for a special reactant economy in monomer formation that would have to be based on (C2 + C2 → C4) chemistry operating at the oxidation level of glycolaldehyde. We think that in an etiological context the full potential of such chemistry becomes apparent only if its range is not restricted to carbohydrate-phosphodiester types of target structures, but is extended to include all possible nitrogenous analogs of reactants, reactions, and reaction products (25). If cyanamide and cyanide were included as coreactants in such a reaction library, a whole family of potential base-pairing systems could be formed, including systems that contain alternative linkages between oligomer units, alternative nucleobases, and alternative backbone-nucleobase junctions (26). Thus, TNA may just be the oxygenous phosphodiester-type representative of a family of constitutionally as well as conformationally related nitrogenous systems that might all have the ability to communicate with RNA by cross-pairing. This possibility defines one direction for further experimentation (27).

  • * To whom correspondence should be addressed. E-mail: aeschen{at}


View Abstract

Navigate This Article