Synthesis and recognition by DNA polymerases of a reactive nucleoside, 1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide

Abstract

We report the synthesis of a new nucleoside, 1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide (dY^NH₂) as a reactive monomer for DNA diversification. The 5′-triphosphate derivative (dY^NH₂TP, 1) was evaluated in vitro as a substrate for several DNA polymerases. Primer extension reactions showed that dYNH₂TP was well tolerated by KF (exo^–) and Vent (exo–) DNA polymerases. One dYNH₂MP was incorporated opposite each canonical base with an efficiency depending on the template base (A ≈ T > G > C). Significant elongation after YNH₂ incorporation was observed independently of the YNH₂:N base pair formed. When the nucleobase YNH₂ was incorporated into synthetic oligodeoxynucleotides via the phosphoramidite derivative 11, it directed the insertion of natural bases as well as itself. The mutagenicity of dYNH₂TP was evaluated by PCR amplification using Vent (exo–) DNA polymerase. The triphosphate dY^NH₂TP was preferentially incorporated as a dATP or dGTP analogue and led to misincorporations at frequencies of ∼2 × 10^–2 per base per amplification. A high proportion of transversions with a large distribution of all possible mutations was obtained. The reactivity of the nucleobase YNH₂ within a template with several aldehydes was demonstrated.

INTRODUCTION

In vitro directed evolution of proteins can be achieved by techniques introducing randomised mutations into specific DNA fragments coding for proteins. The goal is to generate a mutant library containing the largest pool of point mutations. By varying the concentrations of Mn²⁺ and dNTPs, the mutation frequency during PCR could be controlled (up to 2%), but a significant bias for transitions over transversions was observed, A and T being mutated more frequently than C and G (1,2). A large number of nucleoside analogues have been proposed as universal bases (reviewed in 3), but they are not always substrates for DNA polymerisation and very few can be used as nucleoside triphosphates in PCR amplification. Most of the mutagenic nucleotides lead to transitions (4).

We have previously designed a nucleobase with imidazole moiety substituted at position 4 with a carboxamide group (dY; Fig. 1) which we predicted to act as an ambiguous base (5). Primer elongation reactions catalysed by KF (exo^–) and Taq DNA polymerase showed that this nucleobase directed the incorporation of canonical bases as well as itself, and was also incorporated opposite each of the four canonical bases as a nucleoside triphosphate (6,7). The mutagenic properties of dY as a 5′-triphosphate derivative (dYTP) was demonstrated during PCR amplification (8). The introduction of an alkyl chain (methyl or propyl) to the carboxamide extremity yielded monomers (dY^Me or dY^Pr, respectively) able to be incorporated and copied by DNA polymerases as efficiently as dYTP.

Figure 1.

Chemical structure of the simplified purine nucleosides dY and dY^NH2.

The anchoring of variable side motifs to this simplified purine, imidazole-4-carboxamide (Y), was further explored. The introduction of an amino group to the carboxamide extremity yields the nucleobase 4-imidazole hydrazide (Y^NH₂) (Fig. 1). By rotation around the hydrazide moiety and glycosidic bond, the nucleobase Y^NH₂ should pair with all the natural bases as well as with itself according to a base pairing scheme similar to that proposed for Y. Moreover, condensation between the hydrazino function and any aldehyde or ketone should permit to obtain in a single step a large variety of new nucleobases. Their ambiguous pairing properties or their ability to inhibit replication can be evaluated in primer extension reactions.

Here we describe a synthetic access to the 5′-O-triphosphate of 1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide (1) as well as to the phosphoramidite derivative 11. The ability of dY^NH₂TP to be incorporated into DNA was evaluated by primer extension reactions catalysed by commercially available DNA polymerases belonging to the A and B families. We also examined the polymerase recognition of Y^NH₂ placed in the template strand. The potential for mutagenicity was then evaluated by PCR amplification using dY^NH₂TP in place of each of the four canonical bases. The possible reactivity of the nucleobase into DNA was investigated using several aldehydes.

MATERIALS AND METHODS

General methods

¹H, ¹³C and ³¹P NMR spectra were recorded on a Bruker AC-300 instrument. Chemical shifts are given in p.p.m. (δ) relative to residual solvent peak in the case of DMSO-d6 or relative to TMS in the case of CDCl₃ for ¹H and ¹³C, and to 85% phosphoric acid as external standard for ³¹P. Electrospray mass spectra were recorded on a ES/MS platform (VG Biotech-Micromass) working in the negative mode with 50/50 CH₃CN/0.4% NH₄OH or 0.5% Et₃N in water (pH 10). Oligonucleotides were synthesised on an Expedite Millipore DNA synthesiser using standard β-cyanoethyl phosphoramidite chemistry on the 1 µmol scale. The oligonucleotides were separated from failure sequences by preparative HPLC on a Perkin Elmer system with a reverse phase column (C18) using a flow rate of 5.5 ml/min and a linear gradient of CH₃CN (A) in 10 mM triethylammonium acetate buffer (B) at pH 7.5 over 20 min. Reagents and enzymes were purchased from the following sources: the exonuclease-deficient Klenow fragment of Escherichia coli DNA polymerase I, T7 DNA polymerase (exo⁺), Vent (exo^–), T4 polynucleotide kinase and pyrophosphatase from New England Biolabs; Taq, Tli and Tth DNA polymerases from Promega; Pwo from Roche Diagnostics GmbH; Pfu from Stratagene; T7 Sequenase (exo^–) from Amersham; T4 DNA polymerase from Gibco; Ultra pure dNTPs from Amersham Pharmacia Biotech; and QIAquick PCR purification Kit from Qiagen.

Synthesis of the triphosphate and the phosphoramidite derivatives of the nucleobase Y^NH₂

Ethyl 1-[2-deoxy-5-O-(4,4′-dimethoxytrityl)-β-d-erythro-pentofuranosyl]-imidazole-4-carboxylate (3). Compound 2 [ethyl 1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-carboxylate] (0.53 g, 2.07 mmol) (5) in dry pyridine (15 ml) was treated with 4,4′-dimethoxytrityl chloride (0.78 g, 2.30 mmol). After stirring for 3 h at room temperature, the mixture was diluted with dichloromethane, washed in turn with aqueous sodium bicarbonate then water, and the organic layer dried (Na₂SO₄). The product was then purified by silica gel chromatography (dichloromethane/methanol). The resulting foam was dissolved in dichloromethane and then petroleum ether was added vigorously. The precipitated product was filtered off and dried in vacuo to give compound 3 as a white powder (0.84 g, 73%). ¹H NMR (CDCl₃) δ: 1.30 (t, 3H, CH₃), 2.45 (m, 2H, H2′ and H2″), 3.27 (dd, 1H, H5′, J = 4.6 Hz, J = 10.2 Hz); 3.37 (dd, 1H, H5′′); 3.78 (s, 6H, OCH₃), 4.12 (dd, 1H, H4′), 4.30 (q, 2H, CH₂), 4.55 (m, 1H, H3′), 6.02 (t, 1H, H1′, J = 6.3 Hz), 6.82 (d, 4H, H Arom. of DMT), 7.25–7.30 (m, 7H, H Arom. of DMT), 7.40 (m, 2H, H Arom. of DMT), 7.64 and 7.74 (each d, 2H, H2 and H5, J = 1.2 Hz). ¹³C NMR (CDCl₃) δ: 14.16 (CH₃), 41.57 (C2′), 55.04 (OCH₃ of DMT), 60.36 (CH₂), 63.60 (C5′), 72.12 (C3′), 86.13 and 86.17 (C4′ and C1′), 86.47 (Cq of DMT), 113.06 (C3 and C5 of DMT), 122.59 (C5 of Im), 126.78 (C4′ of DMT), 127.58–129.89 (C Arom. of DMT), 134.07 (C4 of Im), 135.33 and 135.40 (C1 of DMT), 136.21 (C2 of Im), 144.24 (C1′ of DMT), 158.39 (C4 of DMT), 162.46 (COOEt). MS (ES) m/z : 557.1 (M-H)^–.

1-[2-deoxy-5-O-(4,4′-dimethoxytrityl)-β-d-erythro-pento-furanosyl]-imidazole-4-hydrazide (4). Hydrazine hydrate (40 ml) was added to compound 3 (1.0 g, 1.79 mmol) in ethanol (4 ml). The solution was heated at 60°C until completion of the reaction (6 h). Dichloromethane was added to the mixture, the organic layer was washed with water (twice), dried (Na₂SO₄) and evaporated. Compound 4 (0.96 g, 96%) was used in the next step without further purification. MS (ES) m/z : 543.1 (M-H)^–.

N-Benzyloxycarbonyl-1-[2-deoxy-5-O-(4,4′-dimethoxytrityl)-β-d-erythro-pentofuranosyl]-imidazole-4-hydrazide (5). To a solution of compound 4 (0.96 g, 1.76 mmol) in dichloromethane (40 ml), triethylamine (0.30 ml, 2.15 mmol) and N-benzyloxycarbonyl succinimidate (0.62 g, 2.48 mmol) were added. After stirring for 6 h at room temperature, the mixture was diluted with dichloromethane, and washed in turn with aqueous sodium bicarbonate and water. The organic layer was dried (Na₂SO₄) and evaporated. Purification by silica gel chromatography (dichloromethane/methanol), followed by precipitation with petroleum ether gave compound 5 as a white powder (0.94 g, 81%). ¹H NMR (CDCl₃) δ: 2.30–2.50 (m, 2H, H2′ and H2″), 3.25 (dd, 1H, H5′), 3.35 (dd, 1H, H5′′), 3.78 (s, 6H, OCH₃), 4.08 (m, 1H, H4′), 4.45 (m, 1H, H3′), 5.18 (s, 2H, CH₂Ph), 5.97 (t, 1H, H1′, J = 6.3 Hz), 6.83 (d, 5H, H Arom. of DMT and NH), 7.20–7.40 (m, 14H, H Arom. of DMT and Ph), 7.57 and 7.67 (each d, 2H, H2 and H5, J = 1.2 Hz), 8.75 (m, 1H, NH). ¹³C NMR (CDCl₃) δ: 41.43 (C2′), 55.24 (OCH₃ of DMT), 63.74 (C5′), 67.76 (CH₂), 72.22 (C3′), 86.08 and 86.23 (C4′ and C1′), 86.63 (Cq of DMT), 113.26 (C3 and C5 of DMT), 120.70 (C5 of Im), 126.99–129.98 (C Arom. of DMT and Ph), 135.17 and 135.53 (C4 of Im and C1 of DMT), 135.71 (C2 of Im), 144.40 (C1′ of DMT), 156.27 (CONH), 158.58 (C4 of DMT). MS (ES) m/z : 677.3 (M-H)^–.

N-Acetyl-N′-benzyloxycarbonyl-1-(2-deoxy-3-O-acetyl-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide (6). Compound 5 (1.50 g, 2.22 mmol) in pyridine (14 ml) was treated with acetic anhydride (0.34 ml, 4.8 mmol). After stirring for 1 h at room temperature, methanol was added and solvents were removed under reduced pressure. The crude product was treated with 20 ml of 80% acetic acid for 20 min, and the solution was concentrated in vacuo and co-evaporated with toluene. Purification by silica gel chromatography gave compound 6 (1.23 g, 1.62 mmol) (73% yield). ¹H NMR (DMSO-d6) δ: 2.00 (s, 3H, CH₃), 2.30 (s, 3H, CH₃), 2.35–2.60 (m, 2H, H2′ and H2″), 3.50 (m, 2H, H5′ and H5′′), 3.55 (m, 1H, H4′), 5.00 (s, 2H, CH₂Ph), 5.10 (m, 1H, OH5′), 5.20 (s, 2H, H3′), 6.05 (dd, 1H, H1′), 7.25 (m, 5H, Ph), 7.90 and 7.92 (each s, 2H, H2 and H5), 9.12 (s, 1H, NH), 9.72 (s, 1H, NH).

N-Benzyloxycarbonyl-1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide 5′-O-phosphate (7). To a solution of compound 6 (0.33 g, 0.71 mmol) in pyridine (4 ml), a 1 M solution of 2-cyanoethylphosphate in pyridine (1.43 ml) and 1,3-dicyclohexylcarbodiimide (DCC) (0.91 g, 4.41 mmol) were added. After stirring for 2 days at room temperature, water was added to the mixture. The insoluble material was filtered off 1 h later. The filtrate was evaporated and treated with a solution of 1% sodium methylate in methanol (15 ml). After stirring for 2 h at room temperature, the mixture was neutralised by addition of resin Dowex (50WX8, H⁺ form). The filtrate was concentrated and purified by reverse phase HPLC to give compound 7 as the triethylammonium salt (0.18 g, 52%). ¹H NMR (D₂O) δ: 1.26 (t, H, CH₃), 2.60 (m, 2H, H2′ and H2″), 3.18 (q, H, CH₂), 4.05 (m, 2H, H5′ and H5′′), 4.25 (m, 1H, H4′), 4.65 (m, 1H, H3′), 5.22 (s, 2H, CH₂Ph), 6.27 (t, 1H, H1′, J = 6.4 Hz), 7.45 (m, 5H, H Arom.), 8.14 and 8.31 (each s, 2H, H2 and H5). ³¹P NMR (D₂O) δ: 0.83. MS (ES) m/z : 456 (M-H)^–.

N-Benzyloxycarbonyl-1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide 5′-O-triphosphate (9). A solution of DCC (0.22 g, 1.07 mmol) in t-butanol (4 ml) was added dropwise over 3 h to a refluxing solution of compound 7 (0.12 g, 0.26 mmol) in 6 ml of t-butanol/water (1/1) and morpholine (0.1 ml, 1.15 mmol). Water was added to the reaction mixture, the insoluble material was removed by filtration and the filtrate was evaporated in vacuo. Purification on a C18 column (H₂O/methanol) afforded the morpholidate derivative 8 as the DCU salt (0.18 g). ¹H NMR (D₂O) δ: 1.10–1.90 (DCU), 2.60 (m, 2H, H2′ and H2″), 3.43 (t, 4H, CH₂N), 3.77 (t, 4H, CH₂O), 3.94 (m, 2H, H5′ and H5′′), 4.20 (m, 1H, H4′), 4.64 (m, 1H, H3′), 5.21 (s, 2H, CH₂Ph), 6.23 (t, 1H, H1′, J = 6.4 Hz), 7.45 (m, 5H, H Arom.), 7.97 and 8.30 (each s, 2H, H2 and H5). ³¹P NMR (D₂O) δ: 8.20. The morpholidate 8 (0.17 g, 0.27 mmol) was dried by co-evaporations with toluene and a solution of pyrophosphoric acid (0.24 g, 1.39 mmol) in DMF (1 ml) containing tri-n-butylamine (0.30 ml, 1.28 mmol) was added. After 2 days, the mixture was diluted in 10 mM triethylammonium bicarbonate (TEAB) solution and purified by chromatography on DEAE cellulose (800 ml, 10 mM to 0.5 M TEAB) to give 9 as 3.5 eq. triethylammonium salt (130 mg, 50%). ¹H NMR (D₂O) δ: 2.50 (m, 1H, H2′), 2.60 (m, 1H, H2″), 4.15 (m, 2H, H5′ and H5′′), 4.25 (m, 1H, H4′), 4.70 (m, 1H, H3′), 5.20 (s, 2H, CH₂), 6.24 (t, 1H, H1′, J = 6.6 Hz), 7.45 (m, 5H, Ph), 8.00 and 8.07 (each s, 2H, H2 and H5). ³¹P NMR (D₂O) δ: –22.67 (t, J = 16.5 Hz), –10.76 (d, J = 16.6 Hz), –10.15 (d, J = 16.4 Hz). MS (ES) m/z : 618 (M-H)^–.

1-(2-Deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide 5′-O-triphosphate (1). Compound 9 as the triethylammonium salt (120 mg, 0.12 mmol) in water (5 ml) was hydrogenated in the presence of palladium (Pd) black (130 mg) at 4°C. After 2 h, the catalyst was removed by filtration and the solution was lyophilised. The residue was purified by reverse phase HPLC (0–15% A in B) to give 1 as the triethylammonium salt (48 mg). ¹H NMR (D₂O) δ: 2.55 (m, 1H, H2′), 2.65 (m, 1H, H2″), 4.17 (m, 2H, H5′ and H5′′), 4.25 (m, 1H, H4′), 4.75 (m, 1H, H3′), 6.24 (t, 1H, H1′, J = 6.6 Hz), 8.00 and 8.04 (each s, 2H, H2 and H5). ¹³C NMR (D₂O) δ: 9.02 (CH₃), 41.07 (C2′), 47.45 (CH₂), 66.22 and 66.30 (C5′), 71.93 (C3′), 86.66 and 86.78 (C4′), 87.65 (C1′), 121.72 (C5 of Im), 134.46 (C4 of Im), 138.68 (C2 of Im), 164.50 (CONH). ³¹P NMR (D₂O) δ: –22.58 (t, J = 19.8 Hz), –10.72 (d, J = 20 Hz), –10.14 (d, J = 19.7 Hz). MS (ES) m/z : 481 (M-H)^–.

N,N-Dimethylamino-ethylidene-1-(5-O-dimethoxytrityl-2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide (10). Compound 4 (0.43 g, 0.78 mmol) in dry DMF (8 ml) was treated with N,N-dimethylacetamide-dimethylacetal (0.3 ml, 1.9 mmol). After stirring for 3 h at room temperature, the mixture was concentrated in vacuo and co-evaporated with xylene. Purification by silica gel chromatography (CH₂Cl₂/MeOH) gave compound 10 as a white powder (0.42 g, 87%). ¹H NMR (CDCl₃) δ: 2.00 (s, 3H, CH₃), 2.27–2.45 (m, 2H, H2′ and H2″), 3.00 (s, 6H, CH₃), 3.23 (dd, 1H, H5′, J = 5.1 Hz, J = 10.1 Hz), 3.34 (dd, 1H, H5′′); 3.78 (s, 6H, OCH₃), 4.03 (dd, 1H, H4′), 4.01 (m, 1H, H3′), 5.96 (t, 1H, H1′, J = 6.1 Hz), 6.81 (d, 4H, H Arom. of DMT), 7.25–7.40 (m, 9H, H Arom. of DMT), 7.55 and 7.61 (each d, 2H, H2 and H5, J = 1.2 Hz), 8.98 (s broad, 1H, NH). ¹³C NMR (CDCl₃) δ: 13.28 (CH₃), 38.56 (CH₃), 41.32 (C2′), 55.25 (CH₃ of DMT), 64.03 (C5′), 72.09 (C3′), 85.76 and 86.03 (C4′ and C1′), 86.51 (Cq of DMT), 113.23 (C3 and C5 of DMT), 119.15 (C5 of Im), 126.92 (C4′ of DMT), 127.94–130.05 (C Arom. of DMT), 135.39–137.23 (C1 of DMT, C4 of Im, C2 of Im), 144.53 (C1′ of DMT), 158.55 and 158.62 (C4 of DMT, CONH), 164.12 (C=N). MS (ES) m/z : 612.1 (M-H)^–.

N,N-Dimethylamino-ethylidene-1-[5-O-dimethoxytrityl-3-(2-cyanoethyl-diisopropyl-phosphoramidite-2-deoxy-β-d-erythro-pentofuranosyl]-imidazole-4-hydrazide (11). 2-Cyanoethyl N,N-diisopropylchlorophosphoramidite (115 µl, 0.45 mmol) and N,N-diisopropylethylamine (260 µl, 1.52 mmol) were added to compound 10 (155 mg, 0.25 mmol) in anhydrous CH₂Cl₂ (5.3 ml). After stirring for 1 h at room temperature, the mixture was diluted with CH₂Cl₂ and successively washed with aqueous sodium bicarbonate and brine. The organic layer was dried (Na₂SO₄), concentrated and the crude product purified by flash chromatography on silica gel (AcOEt/CH₂Cl₂/TEA:45/45/10) affording compound 11 (155 mg, 88%). ³¹P NMR (CDCl₃) δ: 147.03 and 147.20.

Modified oligonucleotides. Oligonucleotides containing the modified base Y^NH₂ were synthesised using standard phosphoramide chemistry on 1 µmol scale. The coupling time for 11 (0.19 M in acetonitrile) was increased to 10 min. After DNA synthesis (trityl on), deprotection was performed with 33% aqueous ammonia at 55°C overnight. Oligonucleotides were purified by reverse phase HPLC, fractions containing the desired product were treated with 80% acetic acid for 20 min and then submitted to further purification by reverse phase HPLC. Analysis using negative-ion ESI-MS (0.5% TEA in water/acetonitrile) revealed experimental masses that are in agreement with theory. MS (ES) m/z for C₂₁₄H₂₇₄N₇₅O₁₃₅P₂₁ (5′-GCATY^NH₂GTCATAGCTGTTTCCTG) (expected 6707.4, found 6708) : 6746.10 (M+K-H), 6730.86 (M+Na-H), 6689.91 (M-17). MS (ES) m/z for C₂₁₅H₂₇₅N₇₄O₁₃₆P₂₁ (5′-ATTGY^NH₂GTCATAGCTGTTTCCTG) (expected 6722.4, found 6723) : 6759.06 (M+K-H), 6744.65 (M+Na-H), 6704.76 (M-17).

Primer extension reactions

5′-end-labelling of the primer (5′-CAGGAAACAGCTATGAC-3′, 1 µM) was carried out in 70 mM Tris–HCl pH 7.6, 10 mM MgCl₂ and 5 mM DTT by addition of 10 U of T4 polynucleotide kinase and [γ-³²P]ATP (10 Ci/mmol) in 50 µl final volume. The mixture was incubated for 30 min at 37°C, then for 15 min at 70°C and finally stored at 4°C. Annealing was realised in 30 µl final volume by incubating 6 µl of appropriate templates at 10 µM, 2 µl of primer at 10 µM and 10 µl of ³²P-end-labelled primer at 1 µM for 15 min at 75°C, followed by slowly cooling over 1 h to room temperature. Time-dependent incorporations were carried out at 37°C for T7 Sequenase, KF (exo^–), T4 or T7 DNA polymerase and at 50°C for thermostable DNA polymerases. A series of 12 µl reactions were performed for each DNA polymerase. Final concentrations used for elongation reactions were: primer-template, 100 nM; dNTPs, 10, 100 or 500 µM or 1 mM; and polymerase 0.003, 0.033 or 0.1 U/µl. Typical elongation reactions were initiated by mixing 6 µl of diluted DNA polymerase in 2× standard polymerase buffer and 2 µl of annnealed primer-template (600 nM) with 4 µl of various concentrations of dNTPs. Aliquots of elongation reactions (3 µl) were quenched in time intervals of 2, 15 and 30 min by adding an equal volume of loading buffer [0.02% (w/v) bromophenol blue, 0.02% (w/v) xylene cyanol FF, 50% (v/v) formamide, 50% (v/v) water]. Before loading onto a 20% denaturating polyacrylamide gel, samples were heated for 1 min at 75°C. After electrophoresis, gels were visualised by autoradiography.

DNA synthesis catalysed by Vent (exo^–) DNA polymerase

Hybrids were formed by mixing primer and templates as described above. Each primer extension reaction was started by mixing 12 µl of diluted Vent (exo^–) DNA polymerase (0.1 U/µl) in 2× standard buffer [1× ThermoPol reaction buffer: 10 mM KCl, 10 mM (NH4)₂SO₄, 20 mM Tris–HCl pH 8.8, 2 mM MgSO₄, 0.1% Triton X-100] and 4 µl of annnealed primer-template (600 nM) with 8 µl of dYNH₂TP (3 mM). Aliquots (3 µl) were taken after incubation at 50°C for 2, 15 and 30 min and quenched with the loading buffer. Further elongation was carried out by adding one of the canonical dNTPs depending on the template (1.5 µl at 10 mM) to the reaction mixture (15 µl). After additional incubation at 50°C, aliquots of elongation reactions were taken at time intervals of 2, 15 and 30 min and quenched. Products were separated by PAGE and visualised by autoradiography.

Modification of Y^NH₂ after its enzymatic incorporation into double stranded DNA

The modification of Y^NH₂ was performed at room temperature by adding a solution of benzaldehyde in CH₃COOH/MeOH (pH 4) to aliquots of the elongation reactions. Samples were heated for 1 min at 75°C before loading onto a 20% denaturated polyacrylamide gel. After electrophoresis, gels were visualised by autoradiography.

Mutagenesis by PCR amplification

The R67 gene coding for the type II dehydrofolate reductase (DHFR) was initially amplified from the parent plasmid pSUR67 (9) by PCR using primer RB (5′-CCCCATGGAACGAAGTAGCAATGAAGTCAG-3′) and primer RF (5′-GCGAATTCTTAGTTGATGCGTTCAAGCGCC-3′). The PCR product was digested and cloned into pTrc99A (Pharmacia) as a NcoI and EcoRI fragment to yield pTrcR67 (10). For each PCR amplification, reaction mixtures contained 10 mM KCl, 20 mM Tris–HCl pH 8.8, 10 mM (NH₄)₂SO₄, 2 mM MgSO₄, 0.1% Triton X-100, 5 ng pTrcR67, 100 pmol of each primer (RB and RF) and 0.05 U/µl Vent (exo^–) DNA polymerase in a final volume of 100 µl. Typically, dY^NH₂TP was used at 1 mM, one of the four canonical nucleotides at a variable concentration (0, 1.25, 2.5, 5, 10 or 20 µM) and the other three nucleotides at 200 µM. The cyclic parameters were: 95°C for 5 min; 50 cycles of 95°C for 30 s, 55°C for 30 s and 72°C for 10 min; and 72°C for 10 min. All reactions were hot-started. PCR products were analysed by electrophoresis (1.2% agarose gel). Following the mutagenesis reactions, 10% of the PCR products were re-amplified by PCR using primers RB and RF, 200 µM equimolar concentrations of the canonical bases and 0.025 U/µl of Vent (exo^–) DNA polymerase. The cyclic parameters were: 95°C for 5 min; five cycles of 95°C for 30 s, 55°C for 30 s; 15 cycles of 95°C for 30 s, 55°C for 30 s and 72°C for 1 min; and 72°C for 10 min.

Cloning and DNA sequencing of the amplified fragments

PCR fragments were separated from a 1.2% agarose gel, extracted with QIAquick gel purification kit, digested with NcoI and EcoRI, purified again with QIAquick and ligated to the expression vector pTrc99A digested with the same enzymes. After transformation (MG1655 cells) and phenotypic expression for 1 h at 37°C, plating an equal volume of the culture on ampicillin (100 µg/ml) plus trimethoprim (0.3 mM) LB plates yields the functional DHFR variants, while plating on ampicillin alone (100 µg/ml) yields the functional and defective DHFR variants; both plates contained IPTG (1 mM). After transformation, individual colonies were picked, grown up and sequenced (MWG).

RESULTS AND DISCUSION

Synthesis of the 5′-triphosphate of imidazole-4-hydrazide deoxyriboside

The synthetic route for 1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide-5′-O-triphosphate (1) is illustrated in Scheme 1. Nucleoside 2 was obtained by an enzymatic transglycosylation using N-deoxyribosyltransferase as previously reported (5). The 5′-dimethoxytritylation of 2 in pyridine afforded compound 3 in 73% yield. Treatment of 3 with a large excess of hydrazine hydrate at 60°C yielded compound 4, which was used in the next step without further purification. The benzyloxycarbonyl group was introduced by reaction of 4 with N-benzyloxycarbonyl succinimidate affording 5 in 81% yield. Attempts to acetylate selectively the 3′-hydroxyl group were unsuccessful. Acylation with acetic anhydride (1.1 eq.) in acetonitrile in the presence of triethylamine catalysed by 4-dimethylaminopyridine gave a mixture of two major products corresponding to the 3′-O-acetylated compound (24%) and the 3′-O,N-diacetylated compound (38%). Complete acetylation was rapidly achieved using acetic anhydride (2.2 eq.) in pyridine. Compound 6 was isolated after detritylation (73% yield from 5). 5′-Phosphorylation of 6 with 2-cyanoethylphosphate in pyridine in the presence of DCC (11), followed by treatment with 2% sodium methylate in methanol gave the 5′-monophosphate 7 (52% yield). The triphosphate 9 was obtained in two steps via the intermediate morpholidate 8 (12) and isolated after purification by DEAE-cellulose column chromatography with 50% yield. Deprotection of the hydrazino group was achieved by hydrogenolysis on palladium black. Finally, the deprotected triphosphate 1 was purified by reverse phase HPLC and characterised by ¹H,³¹P-NMR and mass spectrometries.

Scheme 1. Synthesis of 1. Reagents and conditions: (a) DMT-Cl, pyridine, RT, 3 h; (b) N₂H₄•H₂O, EtOH, 60°C, 6 h; (c) N-CBz-succinimidate, Et₃N, CH₂Cl₂, RT, 6 h; (d) Ac₂O, pyridine, RT, 1 h; (e) 2-cyanoethylphosphate, DCC, pyridine, RT, 2 days, then 2% MeONa in MeOH, RT, 2 h; (f) DCC, morpholine, t-BuOH/H₂O, reflux, 3 h; (g) H₃P₂O₇, n-Bu₃N, DMF, RT, 2 days; and (h) Pd (black), H₂, 2 h.

Synthesis of oligonucleotides containing the nucleobase Y^NH₂

The phosphoramidite derivative 11 was prepared according to Scheme 2. Protection of the hydrazino function of compound 4 by a N,N-dimethylacetamidine group afforded compound 10 (87% yield), which was converted to the phosphoramidite 11 by reaction with 2-cyanoethyl-N,N′-diisopropylaminochlorophosphoramidite in the presence of N,N-diisopropylethylamine in dry dichloromethane (88% yield). The synthesis of oligonucleotides containing this base at predetermined positions was achieved according to standard phosphoramide chemistry. After reverse phase HPLC purification, the purity of the oligomers was checked by PAGE and electrospray mass spectrometry.

Scheme 2. Synthesis of 11. Reagents and conditions: (a) (CH₃)₂NC(CH₃)(OCH₃)₂, DMF, RT, 3h; and (b) ClP(NiPr₂)(OCH₂CH₂CN), NEt₃, CH₂Cl₂, RT, 1 h.

Recognition of the nucleobase Y^NH₂ by DNA polymerases

The capability of dY^NH₂TP (1) to be incorporated into a DNA hybrid was evaluated by primer extension reactions catalysed by DNA polymerases. We tested several commercially available polymerases. Homopolymeric sequences at the 5′ end of the templates were designed to evaluate successive incorporation of the analogue opposite each canonical base. Relative amounts of canonical triphosphates and polymerase were adjusted for each primer extension reaction, in order to facilitate the incorporation of the analogue and minimise the formation of natural base mismatches (non-forcing conditions). Under these conditions, the following DNA polymerases T4, Pwo, Tli, T7 and T7 Sequenase were not able to incorporate this analogue opposite any natural base. Primer extension reactions catalysed by Pfu, Taq and Tth polymerases resulted in partial incorporation (30%) of dY^NH₂TP opposite T in the template (data not shown) and no significant incorporation was detected opposite the other canonical bases. KF (exo^–) and Vent (exo^–) polymerases were both able to incorporate dY^NH₂TP opposite all canonical bases. The incorporations of dY^NH₂TP and dYTP catalysed by Vent (exo^–) DNA polymerase as a function of time are illustrated in Figure 2. Both analogues were incorporated opposite A or G with a comparable efficiency (Fig. 2A and C). The pattern of incorporation differed in efficiency when these analogues were tested opposite C and T: insertion of dY^NH₂TP was more efficient opposite T than opposite C (Fig. 2B and D). In the conditions tested, there was no successive incorporation of dY^NH₂TP, but a band corresponding to primer + 2 opposite T was detected using dYTP. Similar results were observed using KF (exo^–).

Figure 2.

Time-dependent incorporation of dYTP and dY^NH2TP opposite A (A), C (B), G (C) and T (D). Concentrations used were 0.033 U/µl Vent (exo–) DNA polymerase, 100 nM primer-template and 10 µM dNTP. The reaction in the absence of dNTPs is realised as control.

We next examined the capacity of Vent (exo^–) DNA polymerase to further elongate the primer terminated by the analogue Y^NH₂ with a complementary natural base. Only full-length products were obtained with templates having five A, C or T (Fig. 3, lanes 3, 6 and 9), while a small amount of an intermediate product was detected opposite G (Fig. 3, lane 12). Under these conditions (high triphosphate concentration) limited incorporation of two dY^NH₂MP was detected opposite T and G.

Figure 3.

DNA synthesis using canonical or modified nucleoside triphosphates. Conditions: 0.033 U/µl Vent (exo–) DNA polymerase, 100 nM primer-template, 100 µM canonical complementary dNTP (lanes 1, 4, 7 and 10) or 1 mM dY^NH2TP (lanes 2, 5, 8 and 11) followed by 100 µM canonical complementary dNTP (lanes 3, 6, 9 and 12), 30 min reaction times. The first lane is the blank reaction in the absence of dNTPs.

These results showed that the triphosphate analogue was incorporated opposite each of the four canonical bases, with an efficiency depending on the nature of the base to be copied (A ≈ T > G > C). Incorporation of two successive dY^NH₂ residues was observed in the presence of high triphosphate concentration. Further extension with full-length DNA synthesis required the addition of canonical nucleotides.

In vitro replication of template Y^NH₂

An oligonucleotide (22mer) containing the nucleobase YNH₂ was used as a template in primer elongation reactions. Figure 4 illustrates the incorporation of canonical bases opposite Y^NH₂. dATP, dGTP and dTTP were incorporated opposite YNH₂ resulting in the formation of products at primer + 1 (primer + 2 in the case of dATP) whereas dCTP was less efficiently incorporated. When a mixture of the four canonical nucleotides was used, a full-length product was synthesised. Furthermore, dY^NH₂TP was inserted opposite the nucleobase Y^NH₂ (Fig. 5). No inhibition of replication was observed when dY^NH₂TP was used in combination with each of the four canonical bases (Fig. 5A compared with Fig. 4). When a mixture of dY^NH₂TP and the four canonical bases were used, full-length DNA synthesis occurred with some pauses depending on the capability of the polymerase to further elongate the Y^NH₂:N base pairs formed (Fig. 5B). The products distribution was similar to that obtained with canonical bases. The absence of inhibition is important since the analogue should be used in PCRs in great excess to compete with more efficient incorporation of natural triphosphates. Comparable results were obtained using KF (exo^–) (data not shown).

Figure 4.

Time-dependent incorporation of canonical nucleoside triphosphates opposite Y^NH2 in the template. Conditions: 0.033 U/µl Vent (exo–) DNA polymerase, 100 nM primer-template, 100 µM canonical dNTPs, 1 mM dY^NH2TP. Reaction conditions are described in the Materials and Methods. The first lane is the blank reaction in the absence of dNTPs.

Figure 5.

Evaluation of inhibitory effect of dY^NH2TP on the DNA synthesis. Conditions: 0.033 U/µl Vent (exo–) DNA polymerase, 100 nM primer-template, 1 mM dY^NH2TP in combination with one canonical dNTP at 100mM (A) or in combination with three or four canonical dNTPs at 100 mM (B).

Mutagenesis

Target DNA for PCR was the gene coding for the type II DHFR, encoded by the E.coli plasmid pTrcR67. No PCR product was detected when the analogue dY^NH₂TP totally substituted one of the four dNTPs. In order to force the incorporation of the analogue, its concentration was increased to 1 mM and a low concentration of the substituted dNTP (between 1.25 and 20 µM) was added. A PCR product could only be detected when dATP or dGTP was substituted by dY^NH₂TP. Since biased dNTP pools are known to be mutagenic in PCR, amplification without dY^NH₂TP was conducted as a control. Mutagenic PCR products were chased in a second round of PCR using only equimolar concentrations of canonical dNTPs. These PCR products were then cloned and screened for DHFR activity. R67 DHFR confers resistance to the antibiotic trimethoprim (trim), unlike its E.coli counterpart. Plating out R67 DHFR recombinants on trimethoprim plus ampicillin (amp) plates yields the functional DHFR variants, while plating on ampicillin alone yields the total collection of variants, functional and defective. Therefore the mutagenic effects of the reaction conditions were reflected in trim^R/amp^R ratios <1 (Table 1).

Table 1. Trimethoprim-resistant mutant frequencies resulting from mutagenic PCR with dY^NH2TP.

Reaction	dNTP/µM				trimR:ampR		trimR:ampR		dYNH2TP (mM)
	A	C	G	T	PCR control		PCR + dYNH2TP
					Ratio	%	Ratio	%
1	200	200	200	200	6048:7280	83	2903:4464	65	1
2	20	200	200	200	3034:3832	79	301:742	41	1
3	10	200	200	200	566:907	62	851:2096	41	1
4	5	200	200	200	133:137	97	56:82	68	1
5	2.5	200	200	200	191:231	82	103:136	76	1
6	200	200	20	200	997:1534	65	131:347	38	1
7	200	200	10	200	160:194	82	292:339	86	1
8	200	200	5	200	1254:1336	93	249:350	71	1
9	200	200	2.5	200	672:624	107	568:786	72	1

trim^R and amp^R, trimethoprim- and ampicillin-resistant colonies. The ratio trim^R:amp^R yields the proportion of functional variants following hypermutagenesis.

Two successive PCR amplifications using Vent (exo^–) and the four canonical nucleotides at the same concentration (200 µM) resulted in a spontaneous mutagenesis (trim^R/amp^R = 0.83) (reaction 1). When dY^NH₂TP was added to the four canonical bases, the ratio decreased to 0.65 implying that the analogue competed with the normal bases and induced some mutations. When dATP was present at a concentration of 20, 10 or 5 µM, dY^NH₂TP increased the number of mutations (reactions 2–4). Below 5 µM of dATP (reaction 5), no significant mutagenic effect could be assigned to dY^NH₂TP. When dY^NH₂TP substituted dGTP, an excess of mutations over the control was measured at 20 µM dGTP (reaction 6). No evidence of a significantly altered ratio was found when dGTP was added below 20 µM (reactions 7–9).

Clones obtained from reactions 2 and 6 were sequenced. The resulting mutations are given in Table 2. When dY^NH₂TP was used to replace dATP (reaction 2 + dY^NH₂TP) the mutation frequency was 2 × 10^–2 per base per amplification, while that obtained with biased dATP concentration (reaction 2 control) was 9 × 10^–3 per base per amplification. The major substitutions were A→G, T→C (30%), G→A, C→T (28%) and A→T, T→A (20%), with a slight excess of transitions (Ti) (58%) over transversions (Tv) (42%). A similar Ti/Tv ratio was obtained in the control reaction (61 and 39%, respectively), but with another composition. The use of substandard dATP concentration resulted in 35% of all possible A replacements (A→N), while the addition of dY^NH₂TP increased the percentage of A replacements (55%), indicating that dY^NH₂TP was accepted as an A mimic. The addition of dY^NH₂TP to substandard dGTP concentration raised the mutation frequency from 8 × 10^–3 (reaction 6 control) to 1.3 × 10^–2 (reaction 6 + dY^NH₂TP) per base per amplification. The major mutations were G→T, C→A (41%) and G→A, C→T (35%). If we consider all possible G replacements (G→N), a similar percentage was obtained using biased dGTP concentration with or without dY^NH₂TP (86% and 90%, respectively). Also, the incorporation of dY^NH₂TP as a G mimic could not be directly correlated with the type of mutations. However, the addition of dY^NH₂TP increased the total number of transversions (65% compared with 51%) with a higher proportion of A→T, T→A and G→C, C→G mutations. Among the 71 clones sequenced there was no insertion, but a small number of deletions were found. These appeared to be independent of the presence of dY^NH₂TP and could be attributed to the biased conditions.

Table 2. Base substitution frequency for hypermutagenised R67 DHFR.

Reactiona	Colonies sequenced	Nb. mut.b (del)	Mutation frequencyc	Ti/Tvd	Transition Nb (%)			Transversion Nb (%)
					A → G	G → A	A → T	A → C	G → C	G → T
					T → C	C → T	T → A	T → G	C → G	C → A
2 + dYNH2TP	22	105 (1)	2.0 × 10–2	61/44	32 (30%)	29 (28%)	21 (20%)	5 (5%)	4 (4%)	14 (13%)
2 control	12	26 (1)	0.9 × 10–2	16/10	6 (23%)	10 (38%)	3 (12%)	0 (0%)	2 (8%)	5 (19%)
6 + dYNH2TP	21	63 (6)	1.3 × 10–2	22/41	0 (0%)	22 (35%)	7 (11%)	2 (3%)	6 (10%)	26 (41%)
6 control	16	31 (4)	0.8 × 10–2	15/16	1 (3%)	14 (45%)	1 (3%)	1 (3%)	1 (3%)	13 (42%)

^aReactions are those given in Table 1.

^bTotal number of mutations (Ti + Tv) (number of deletions are given in parentheses).

^cMutation frequency is the number of mutations scored divided by the product of the number of clone sequences and the target length (237 bp).

^dNumber of transitions and transversions.

Amino acid sequences of sequenced clones 136–160 (reaction 2 + dY^NH₂TP) and 278–301 (reaction 6 + dY^NH₂TP) are given in Table 3. These data show that mutagenesis using dY^NH₂TP as replacement for dATP or dGTP yielded randomly distributed amino acid changes with no hotspot.

Table 3. Collections of R67 DHFR mutants.

Amino acid sequences are aligned to the R67 sequence using the one letter code. For each clone (indicated on the left), the number of amino acid (aa) and nucleotide (nu) substitutions are indicated on the right. –, in-frame stop codon (TAA, TAG or TGA); X, nucleotide deletion.

Reactivity of the hydrazine function toward aldehyde

The possible reaction of Y^NH₂ once it is incorporated into DNA with aromatic aldehydes of different sizes (benzaldehyde, trans-cinnamaldehyde and anthracene-9-carboxaldehyde) as well as with a fluorescent compound (fluorescamine) was investigated. Each of the four primer-templates was elongated in the presence of dY^NH₂TP in the conditions described previously. The elongation products were then treated with an excess of aldehyde in alcohol. Gel electrophoresis analysis shows that the coupling reactions were completed within 2 min as shown in Figure 6 when dY^NH₂TP was incorporated opposite T. We also checked that the nucleobase Y^NH₂ located within the primer could also be coupled with benzaldehyde (data not shown).

Figure 6.

Incorporation of dY^NH2TP opposite T followed by chemical modification of Y^NH2 at the 3′-extremity of the primer. The first lane is the blank reaction in the absence of dNTPs. Lanes 1–3 show the time-dependent incorporation of dY^NH2TP (100 µM) catalysed by Vent (exo^–) (0.033 U/µl). Coupling reaction was carried out with a 10 or 100-fold excess of fluorescamine (lanes 4 and 5), benzaldehyde (lanes 7 and 8), trans-cinnamaldehyde (lanes 10 and 11) or anthracene-9-carboxaldehyde (lanes 13 and 14). Mixtures of unmodified and modified samples are shown (lanes 6, 9, 12 and 15; noted mix).

CONCLUSION

We designed and synthesised 1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-hydrazide (Y^NH₂), which is related to the previously reported ambiguous base, 1-(2-deoxy-β-d-erythro-pentofuranosyl)-imidazole-4-carboxamide (Y) (5). This analogue was expected to pair with the canonical bases according to a base-pairing scheme analogous to that of Y. The presence of the hydrazino function allows to condense the nucleobase with various aldehyde or ketone to generate a family of nucleotides.

The effect of the hydrazino function attached to the carboxy-imidazole moiety on the recognition by DNA polymerases was examined. The new base possesses the characteristics of an ambiguous one, i.e. the ability to replace the canonical bases in DNA replication reactions both as a nucleoside triphosphate and as a template base. The mutagenic effect of the analogue was measured during PCR experiments when dY^NH₂TP replaced dATP or dGTP. The resulting substitution frequencies of 1.3–2 × 10^–2 per base per amplification are comparable with those obtained with other random mutagenesis procedures. However, the presence of dY^NH₂TP raises the proportion of transversions and allows to produce almost all possible mutations.

The specific reactivity of the nucleobase Y^NH₂ into DNA toward aromatic aldehydes of different size as well as fluorescamine was demonstrated. The presence of the hydrazino function permits the diversification of the imidazole motif by reaction with any aldehyde or ketone. Oligomer libraries could be generated by enzymatic incorporation of the hydrazide triphosphate catalysed by DNA polymerase, or by chemical synthesis using the phosphoramidite derivative, followed by modification of the hydrazide function. Oligomers containing YNH₂ or its derivatives should find applications in labelling or mutagenesis protocols.

Further work is now directed toward the development of a combinatorial approach aimed at producing, in one step, a large family of triphosphate derivatives that can be tested as possible substrates for DNA polymerases. The use of these compounds in primer extension reactions catalysed by DNA polymerases should permit the screening of new monomers having ambiguous or universal pairing properties, or acting as inhibitors of replication.