Synthesis of Oligonucleotides Carrying 5 ’ - 5 ’

--------------------------------------------------------------------------------------------------------There is considerable interest in coupling oligonucleotides to molecules and surfaces. Although aminoand thiol-containing oligonucleotides are being successfully used for this purpose, cycloaddition reactions may offer greater advantages due to their higher chemoselectivity and speed. In this study, copper-catalyzed 1,3-dipolar cycloaddition reactions between oligonucleotides carrying azido and alkyne groups are examined. For this purpose several protocols for the preparation of oligonucleotides carrying these two groups are described. The non-templated chemical ligation of two oligonucleotides via coppercatalyzed [3+2] cycloaddition is described. Using solid-phase methodology oligonucleotides carrying 5’-5’ linkages can be obtained in good yields. ---------------------------------------------------------------------------------------------------------Introduction.Recent years have seen an increasing demand for oligonucleotide conjugates, while the Human Genome Project has triggered a demand for oligonucleotide chips. Oligonucleotides carrying amino and thiol groups are the most commonly used intermediates in the preparation of oligonucleotide conjugates and DNA chips. This is due to the special reactivity of thiol and amino groups, which allows formation of specific covalent bonds thiols react with maleimido and bromoacetamido groups while aliphatic amino groups are reactive to active esters and isothiocyanates. Although these reactions are widely used, they are not completely chemoselective in aqueous solvents and hydrolysis occurs together with the desired coupling reaction, thereby lowering the efficiency of these reactions. This drawback has triggered the search for new chemoselective coupling reactions that may be used for the coupling of biomolecules in aqueous solvents. For example the DielsAlder reaction has been described for the preparation of oligonucleotide conjugates [1, 2] and for the immobilization of oligonucleotides [3, 4]. A further cycloaddition reaction of interest is the [3+2] copper-mediated cycloaddition [5] or “Click Chemistry” [6]. This particular reaction has had a major impact on biomolecular research, especially in peptide [7, 8] and protein chemistry [9]. In oligonucleotide research the development of applications using Click Chemistry has been slower. This probably reflects the difficulties involved in preparing oligonucleotides modified with azido and alkynyl groups and the possible role of Cu(I) in producing hydroxyl radicals that may damage DNA [10-12]. Nevertheless, Seo et al. have shown that this reaction is useful for the immobilization of oligonucleotides on a chip as a first step for DNA sequencing [13, 14]. In addition, the preparation of oligonucleotide-carbohydrate conjugates [15, 16] and the synthesis of bis-nucleosides [17] using Click Chemistry have been described. Recently, the azide-alkyne cycloaddition reaction has been used for the template-mediated chemical ligation of two oligonucleotides and for the intramolecular circularization of a single oligonucleotide [18] In this paper we seek to develop efficient protocols for the preparation of azido and alkyne groups. In addition we examine the use of the copper-catalyzed cycloaddition reaction for the preparation of oligonucleotide derivatives carrying a non-natural 5’-5’ phosphate bond in the middle of the molecule. If the two halves of the oligonucleotide are complementary, oligonucleotide derivatives of this type may form parallel-stranded duplexes [19]. If half of the hairpin is a polypurine sequence and the other half is a polypyrimidine sequence, the resulting parallel-stranded duplex will bind a polypyrimidine sequence by triple helix formation [20-22]. Our group has been especially interested in this type of oligonucleotides but until now the standard method for the preparation of parallelstranded duplexes has required the use of reversed phosphoramidites which are less efficient and more expensive [20-22]. The present study seeks to identify the optimal conditions for preparing parallel-stranded clamps. Results and Discussion.1. Synthesis of oligonucleotides carrying alkynyl groups. A general method for introducing functional groups at the 5’-end of oligonucleotides involves the reaction of 5’-amino-oligonucleotides with a compound carrying the desired functional group linked to a carboxylic acid. Using this approach, Seo et al. described the synthesis of oligonucleotides carrying a propargyl group at the 5’-end [13]. First, we used a variant of this method for the preparation of propargyl-oligonucleotides. Oligonucleotide 1 (T85’NH2) was synthesized on a 1 μmol scale. The phosphoramidite of 6-aminohexanol protected with the monomethoxytrityl [(MeO)Tr] group was used for introducing the amino group at the 5’-end. After removing the (MeO)Tr group, the resulting aminooligonucleotide-support was treated with succinimidyl N-propargyl glutariamidate [13], followed by cleavage and ammonia deprotection. Using this method, the propargyl oligonucleotide 2 was obtained in good yield and the mass spectrum of the purified compound was in agreement with the expected mass. Alternatively we used the phosphoramidite of 10-hydroxydecanoic acid Nhydroxysuccinimide ester for the introduction of the N-hydroxysuccinimide ester group at the 5’-end. Oligonucleotide sequences 4 (A8-5’COOH) and 6 (T8-5’COOH) were synthesized on a 1 μmol scale. The resulting 5’-carboxy-oligonucleotide-supports were treated with propargylamine followed by ammonia deprotection. Propargyloligonucleotides 5 and 7 were obtained in good yields and they were characterized by mass spectrometry. This method was simpler than that described above as the use of propargylamine avoided the need to prepare succinimidyl N-propargyl glutariamidate. As a further step in the simplification process, we prepare the phosphoramidite derivative of an alcohol carrying a terminal alkyne. Commercially available 5-hexyn-1-ol was reacted with chloro-N,N-diisopropylamino-O-(2-cyanoethoxy) phosphine yielding the desired phosphoramidite. This phosphoramidite was used to introduce an alkynyl group at the 5’-end of oligonucleotides 11 (GA-5’alkylnyl) and 12 (CT-5’alkynyl). Alkynyloligonucleotides 11 and 12 were obtained in excellent yields and they were characterized by mass spectrometry. No oxidation products resulting from the interaction of the iodine solution used in the DNA synthesizer and the alkyne function were observed. 2. Synthesis of oligonucleotides carrying azido groups. The azido group is not compatible with the phosphoramidite group because azido groups react with phosphites yielding phosphoramidates (Staudinger reaction [23]). For this reason azido groups need to be introduced in the oligonucleotide after the completion of the sequence. The preparation of oligonucleotides carrying azidonucleosides has also shown that azido groups attached to the nucleobases are stable to ammonia solutions only at room temperature, but not at higher temperature [24-26]. First we used the method described by Seo et al [13] for the preparation of oligonucleotides carrying 5-azido groups introducing some modifications. Oligonucleotide 1 (T8-5’NH2) was synthesized as described above. After the removal of the (MeO)Tr group the resulting amino-oligonucleotide-support was treated with 5-azidopentanoic acid Nhydroxysuccinimide ester [13]. The resulting support was treated with concentrated ammonia at room temperature to avoid azide decomposition [24-26]. Oligonucleotide 3 carrying an azido group at the 5’-end was obtained in good yields as determined by HPLC analysis. The purified product had the expected mass. A second protocol based on the iodination of the 5’-end followed by azide displacement was studied. Oligonucleotide 8 (CT) was synthesized on a 1 μmol scale and the last dimethoxytrityl [(MeO)2Tr] group was removed. The resulting support was treated with triphenoxymethylphosphonium iodide as described by Miller and Kool [27] to yield the iodo-oligonucleotide 9 (CT_I) and the resulting support was treated with sodium azide [11]. Finally, the support was treated with ammonia to yield the 5’-azido-oligonucleotide 10 (CT_N3) in excellent yields as determined by HPLC analysis. The purified oligonucleotide was characterized by mass spectrometry and enzymatic digestion using snake venom phosphodiesterase and alkaline phosphatase followed by HPLC analysis [25] showing the presence of 5-azido-2’,5’-dideoxycytosine. The success of the previous method suggested the need to prepare of a phosphoramidite to introduce the halohexyl group at the 5’-end of the oligonucleotide. We decided to study the potential use of the bromohexyl group in DNA as an intermediate group in the introduction of the azido group. In addition the hexyl linker would provide less steric hindrance to the azido group than the previous 5’-azido-2’-deoxynucleoside derivative. Starting from commercially available 6-bromohexanol, the phosphoramidite derivative was prepared. This phosphoramidite was introduced in the DNA synthesizer and incorporated into the CT oligonucleotide sequence 13. The support carrying the 5’-bromo oligonucleotide was treated with sodium azide and the resulting support was treated with concentrated ammonia at room temperature, giving the desired 5’-azido-oligonucleotide 14 (CT_N3hexyl) in good yields. The purified oligonucleotide had the expected mass. 3. Cu-catalyzed cycloaddition of azido-oligonucleotides and alkynyl-oligonucleotides. Next, the use of copper-catalyzed cycloaddition reactions to chemically ligate two oligonucleotides was studied. In order to find the optimal conditions for the coupling reaction, a small excess of T8-5’propargyl (2) was mixed with T8-5’azide (3) in the presence of either CuSO4/ascorbic acid or CuI as catalyst. Best results were obtained when a large excess (10-25 times excess) of copper catalyst was used (Figure 1). When the copper catalyst was only 0.1 equivalents, yields were between 10-15%. The length of the product of cycloaddition was confirmed by gel electrophoresis. Mass spectrometry gave a higher mass than expected probably due to the presence of copper ions that were not completely eliminated by HPLC purification. In order to facilitate the purification and removal of the copper ions we studied the Cucatalyzed cycloaddition reactions on the solid support. The solid support carrying oligonucleotide sequence T8-5’azide (3) was treated with 2 equivalents of T8-5’propargyl (2) using a large excess of CuI as catalyst. The resulting support was treated with concentrated ammonia. Analysis of the reaction by HPLC showed the formation of the expected product 15 as the major component (data not shown). Likewise the solid support carrying oligonucleotide sequence CT-N3hexyl (14) was reacted with T8-5’propargyl (7) and CuI to yield oligonucleotide 17 (Figure 2). This was also characterized by mass spectrometry and gel electrophoresis. Tris(benzyltriazolylmethyl)amine (TBTA) has been recommended as a copper ligand to enhance speed and prevent damage to the oligonucleotides [11, 12]. We compared the results using TBTA and CuI as catalysts in the reaction between CT-N3hexyl (14) and T85’propargyl (7) to yield oligonucleotide 16. Although the resulting chromatogram was slightly better when using TBTA, no great differences in the yield were observed. It is important to notice that the use of Click Chemistry in the oligonucleotide field is strongly influenced from the negative results described by Kanan et al [12]. These authors treated 17 pmols of an oligonucleotide with a large excess of copper sulphate/ ascorbic acid (more than 2000 time molar excess) and found approximately 50% degradation of the oligonucleotide after 10 min at room temperature [12]. These conditions are far away from the preparative work such as the study described here (150-300 nmols of oligonucleotide in the presence of 10-20 molar excess of copper). In our conditions we found only a slight degradation after 2-3 days of treatment as seen by the fast eluting peaks in front of the desired compound (Figure 2). Next, the solid support carrying oligonucleotide sequence T8-5’azide (3) was treated with A8-5’propargyl (5) and CuI (3 mg) as described above. After the reaction, the support was extensively washed to eliminate the copper ions. The desired oligonucleotide 17 was obtained in good yield. The purified oligonucleotide 17 was characterized by UV, mass spectrometry, gel electrophoresis and enzymatic digestion. CD spectra of the purified 17 show the presence of a parallel duplex structure as expected (Figure 3). When cycloaddition reactions were performed using oligonucleotides 11 and 12 carrying alkynyl groups the cycloaddition reaction resulted in low yields and the final product could not be isolated (data not shown). Nevertheless, reaction of 5-hexyn-1-ol with 5’-azidothymidine and oligonucleotides 11 and 12 with benzylazide gave the expected cycloaddition products (Figure 4). Most probably the 5-hexynyl group is not reactive enough to link two large molecules such as the oligonucleotides, but it may be used for linking small organic molecules to oligonucleotides. This result is in agreement with Click reactions involving alkynes without a neighbouring electron-withdrawing group [28, 29]. Conclusions. The synthesis of oligonucleotides carrying alkynyl and the synthesis of oligonucleotides carrying azido groups were carried out using three methods. We have demonstrated that the 5’-ends of two oligonucleotides can be chemically linked using a Cucatalyzed cycloaddition reaction with the following observations: A) An excess of copper ions is required. B) CuI is a more efficient catalyst than CuSO4/ascorbic acid. C) The use of the azido-oligonucleotide anchored still in the solid phase allows the efficient removal of the excess of copper ions. D) 5-Hexynyl groups are not reactive enough to produce the cycloaddition products between oligonucleotides. Oligonucleotides with a parallel duplex structure with 5’-5’ linkages are of interest for their triplex-forming properties [19-22]. The synthesis of these compounds by the linking of two parts is a considerable challenge. Previously we sought to link these two parts using thiol and maleimido groups but our attempts were unsuccessful. Using Click Chemistry, however, this synthesis has been possible. We believe that the power of this reaction will enable a large number of oligonucleotide conjugates to be synthesized. Acknowledgement. This study was supported by the Institute for Research in Biomedicine (IRB Barcelona), the Spanish Ministry of Education (NAN2004-09415-C05-03 and BFU2004-02048), the Generalitat de Catalunya (2005/SGR/00693), the Fundació La Caixa (BM04-52-0), the Instituto de Salud Carlos III (CIBER-BNN, CB06_01_0019) and the European communities (Nano-3D NMP4-CT2005-014006). M.A. thanks the Spanish Ministry of Education for a predoctoral fellowship.