WO1996041812A1

WO1996041812A1 - Oligonucleotide phosphorylation method and products

Info

Publication number: WO1996041812A1
Application number: PCT/GB1996/001360
Authority: WO
Inventors: George Gow Brownlee; Gordon Lowe; Ervin Fodor
Original assignee: Isis Innovation Limited
Priority date: 1995-06-09
Filing date: 1996-06-07
Publication date: 1996-12-27
Also published as: GB9511720D0

Abstract

An oligonucleotide having a terminal 5'-OH group is reacted with a phosphitylating agent or phosphorylating agent and then with a phosphate or pyrophosphate and an oligonucleotide having a terminal 5'-di- or triphosphate group is recovered. The resulting oligonucleotide having a terminal 5'-di- or triphosphate group may be capped and used for the study of influenza virus transcription.

Description

OLIGONUCLEOTIDE PHOSPHORY ATION METHOD AND PRODUCTS

INTRODUCTION

The control of transcription and replication of the RNA genome of influenza A viruses is of great interest not only because of its fundamental importance for the viral life-cycle but also because a fuller understanding of the mechanisms offer, in theory, the possibility of devising specific antiviral agents. A peculiarity of the influenza transcriptional mechanism is that mRNA synthesis is initiated at each of the 8 negative-stranded RNA segments by the endonucleolytic cleavage of host precursor mRNA in the nucleus of infected cells by a specific endonuclease activity of the influenza-specific RNA polymerase (for reviews see 1,2). This endonuclease recognizes capped ends of mRNA and cleaves predominantly but not exclusively at purine residues 9-15 nucleotide residues from the cap structure. The resultant capped oligonucleotide then acts as a primer by initiating influenza mRNA synthesis at the second or third nucleotide of the template virion RNA (3). A recent study has confirmed earlier data that priming by capped oligonucleotides can be uncoupled from the endonuclease activity of the influenza RNA polymerase (4).

Our chemical phosphorylation method may be ideal for the large-scale synthesis of short capped oligonucleotides. They are of great interest since short capped oligonucleotides with a 3'-phosphate end-group in the range of 4-9 nucleotides in length, can bind to the influenza RNA polymerase and specifically inhibit cap-dependent transcription in vitro (4). They are, therefore, potential specific antiviral compounds (4).

Our original purpose was to provide an improved method of preparing oligonucleotides which would be capped for the study of influenza virus transcription. Instead of synthesizing RNA in vitro by transcription of a suitable plasmid DNA to give δ'-triphosphorylated RNA, which is then capped by guanylyl transferase in the presence of GTP and S-adenosyl methionine (4), we used protected chemically synthesized oligoribonucleotides attached to a solid phase support as starting material for a chemical phosphorylation. After deprotection, capping was carried out enzymatically with guanylyl transferase. But the invention is of broader scope as is indicated below.

The phosphorylation method used here is adapted from the Ludwig & Eckstein (5) synthesis of thio-ATP analogues from adenosine using the phosphitylating agent 2-chloro-4H-1,3,2-benzodioxaphosphorin- 4-one (salicyl phosphorochloridite) and the subsequent reaction with pyrophosphate followed by oxidation with iodine. It had previously been adapted to the solid phase synthesis of 2'-O-methylated ATP from 2'-O- methylated adenosine attached to controlled pore glass (6).

THE INVENTION

This invention provides a method which comprises reacting an oligonucleotide having a terminal 5'-OH group with a phosphitylating agent or a phosphorylating agent and then with a phosphate or pyrophosphate and recovering an oligonucleotide having a terminal 5'-di- or triphosphate group.

The starting oligonucleotides may have the general formula a).

0

B represents a base, which may be free or preferably protected.

Z represents H or OH, which may be free or preferably protected, or alternatively alkyl or alkoxy e.g. methyl or methoxy. X represents an oligonucleotide (including oligoribonucleotide) chain, which may be protected at the base and 2'- positions. This oligonucleotide is preferably linked at its 3'-end to a solid phase.

The oligonucleotide may be composed of individual nucleotides or ribonucleotides or analogues thereof, provided that there is present a 5'-hydroxyl group capable of being phosphorylated. The length of the oligonucleotide is immaterial. The nature of the base protecting groups, which are preferably present although believed not necessary, is also not material to the invention. The starting oligonucleotide is reacted with a phosphitylating agent (based on trivalent phosphorus) or a phosphorylating agent (based on pentavalent phosphorus). Phosphitylating agents are described and exemplified in Nucleic Acids in Chemistry and Biochemistry pp 114-5 (eds G M Blackburn and M J Gait) IRL Press at OUP (1990). A preferred phosphitylating agent, used in the experimental section below, is salicyl phosphorochloridite. Phosphorylating agents are known in the art and can be used in its place. For example morpholidates (7), imidazolidates (8), phosphoramidates (9) or 8-quinolates (10) may be used. The reaction is performed under anhydrous conditions in a polar organic solvent system. This reaction gives rise to an intermediate shown, in the particular compound prepared in the experimental section below, as structure 2 in Figure 1 of the accompanying drawings.

After removal of the phosphitylating agent, this intermediate is reacted with a phosphate donor including pyrophosphate. The reaction is again performed under anhydrous conditions in a polar organic solvent system in which the phosphate is soluble. Various monomeric and oligomeric phosphates may be used. The use of simple orthophosphate is expected to give rise to products having a terminal diphosphate group. The use of pyrophosphate may give rise to products having a terminal di- or triphosphate group.

After removal of the phosphate reagent any phosphite groups may be oxidized to phosphate with a solution of aqueous iodine containing pyridine.

The product may have a structure as shown in b), in which n is 2 or 3 and X, Y, Z and B are previously defined. At this stage, protecting

groups may be removed if desired, and the phosphorylated product may be either left linked to a solid phase or recovered into solution. Products having a terminal 5'-di- or triphosphate group may then be capped enzymatically e.g. by reaction with guanylyl transferase (see below). This reaction can be performed under conditions well known in the art.

The phosphorylation and capping reactions described in the experimental section below may be represented by the following scheme shown in Figure 1 , where R, = cyanoethyl phosphotriester linked 0 oligoribonucleotide (base and 2'-protected) attached to controlled pore glass.

R₂ = t-butyldimethylsilyl or methyl. B = protected base. GTP = guanosine triphosphate SAM = S-adenosyl methionine

When a diphosphate intermediate 5 is obtained instead of the triphosphate intermediate 4, this is similarly capped to give the final capped product 6.

In another aspect this invention provides products of the general formula ppN_n or pppN_n, where N is any nucleotide or any nucleotide analogue and n is at least 2, e.g.4 to 13. Nucleotide analogues are compounds which are derivatives of mononucleotides by virtue of a base and/or ribose and/or phosphate modification of the standard four 2'- deoxy or ribonucleic acids. Furthermore to be useful in this invention they must be capable of being joined to other nucleotides to form an oligonucleotide chain. For example, nucleotides carrying a 2'-alkyl or 2'- alkoxy group are regarded as nucleotide analogues.

In another aspect the invention provides certain capped products of general formula m⁷GpppNmNm(N)_x where N is a nucleotide or nucleotide analogue residue, m is a 2'-O-methylated residue and x is equal or greater than 1.

Preferred capped products are those in which x is 2 to 7. As previously noted, these capped products are potentially valuable as influenza antiviral agents. Although the capped products described in the experimental section below have oligonucleotide chains ranging from 11 to 13 nucleotide residues, it will be immediately apparent that capped products with a chain containing 4 to 9 nucleotide residues can be made by exactly comparable routes.

Reference is directed to the accompanying drawings, in which:- Figure 1 is a reaction scheme illustrating the preparation and capping of oligonucleotides by the method of this invention; Figure 2 is a mass spectrograph; and Figure 3 is an electrophoresis gel autoradiograph.

MATERIALS AND METHODS

Synthesis and deprotection of oliqoribonucleotides

5'-Dimethoxytrityl 2'-O-t-butyldimethylsilyl cyanoethyl phosphoramidite protected nucleosides and CPG (controlled pore glass) 2-O-t-butyldimethylsilyl protected nucleosides were purchased from Peninsula Labs. These nucleosides were base protected as follows: A and C, benzoyl; G, isobutyryl. 2'-O- Methyl cyanoethyl phosphoramidite dimethylformamidine protected G was purchased from Glen Research. Oligoribonucleotide synthesis (0.2 - 1.0 μmole) was performed on an ABI 394 synthesizer using a standard RNA synthesis cycle with a 10 min coupling time with tetrazole. The following oligoribonucleotides were synthesized: GAAUACUCAAG, GmAAUACUCAAG (where Gm is 2'-O- methylguanosine), ACACUUGCUUUUG and U„. Automatic deprotection of the 5'-dimethoxytrityl group was carried out on the ABI machine. All other protecting groups were removed manually using procedures slightly modified from Lamond and Sproat (11). Either anhydrous ethanolamine, anhydrous ethanol (1:1) at 60°C for 3 hours or 0.88 sp. gr. ammonia, ethanol (1:3) at 60°C for 16 hours was used for base deprotection, hydrolysis of cyanoethyl groups and the succinyl linkage of the oligonucleotide to the CPG. After freeze-drying to remove ethanol and ammonia, the 2'-O-t-butyldimethylsilyl protecting group was removed with 0.25 ml 1 M tetrabutylammonium fluoride in THF (Aldrich, kept over molecular sieve, type 3A to reduce the water content (12)) for 24 hr at 30°C. After adding an equal volume of water, the fully deprotected oligonucleotide was desalted on a 3.5 ml Sephadex G-25 column (Pharmacia NAP-10, prepacked) in deionized water. The major A_260nπι fractions of 0.25 ml were pooled, freeze-dried and redissolved in 50 μl of 10 mM Tris-HCI, 1mM EDTA, pH 8.0.

Phosphorylation of oligoribonucleotides

Phosphorylation was carried out at room temperature on fully protected oligoribonucleotides attached to CPG after automated removal of the 5'-dimethoxytrityl group on the ABI synthesizer. The glass beads were then transferred to a small glass column (20mm x 6mm, internal diameter) containing a sinter to trap the beads (from Omnifit, Cambridge). The column could be maintained under near anhydrous conditions by an inlet septum through which reagents and dry solvents were injected via a 0.5 ml Hamilton syringe. A three-way side-arm inlet valve, connected to a N₂ cylinder via a drying tube containing silica gel, was connected manually when reagents or solvents were to be removed from the column to waste via another three-way tap. The following protocol is slightly modified from ref. 6 principally because of the much smaller scale of synthesis used here (1 μmole or less oligonucleotide have compared to 100 μmole nucleoside in reference 6). The oligonucleotide derivatized CPG (0.2 -1 μmole) in the reaction column was initially washed 3 times with dry acetonitrile and then 0.3 ml dry pyridine/dioxan (1:3) was added, followed immediately by 0.1 ml 1M 2-chloro-4f -1,3,2-benzodioxaphosphorin-4-one (salicyl phosphorochloridite, Aldrich) freshly made up in dry dioxan. The column was inverted 3 times to mix the solution with the solid support and left for 15 min. to form the bifunctional reactive intermediate 2. The phosphitylating agent was then removed and the column washed 3 times with dry dioxan followed by 3 times with dry acetonitrile. Then 0.3 ml 0.5 M tri-n- butylammonium pyrophosphate (5) in dry DMF was added followed immediately by 0.1 ml tri-n-butylamine. The column was again inverted 3 times to mix the reaction components. After 20 min the excess pyrophosphate was removed and the support washed 3 times with dry DMF followed by 3 times with acetonitrile. Oxidation was then performed with 0.3 ml iodine/H₂O/pyridine THF (3/2/20/75) for 10 min followed by washing 3 times with acetonitrile to remove excess iodine. Finally the oligonucleotide was dried by passing through a stream of N₂ for 10 min. After removal of the derivatized support from the reaction vessel, deprotection was carried out using ammonia/ethanol for base deprotection, followed by desilylation and desalting on NAP-10 columns and freeze- drying, as described (see above) for deprotection of oligoribonucleotides. This preparation is referred to as the crude phosphorylated oligonucleotide. For the preliminary work in which tri-n-butylamine phosphate replaced tri-n-butylamine pyrophosphate, the phosphate reagent was prepared as follows. 40 ml of 5% H₃PO₄ in deionised water was neutralised with n-butylamine until pH = 7.0. A further 1 ml of n-butylamine was added and the liquid evaporated to dryness in vacuo. Two further evaporation steps from 50 ml anhydrous ethanol were performed. The final residue was dissolved in 25 ml DMF (anhydrous) and stored at room temperature over molecular beads, type 3A.

31P NMR

Prior to ³¹P NMR, DMF was removed from the tri-n- butylammonium pyrophosphate by evaporation in vacuo and the oil redissolved in water. Residual DMF was removed by re-evaporation from water twice more. The aqueous pyrophosphate solution (about 0.2 M) was titrated to pH 10.0 with NaOH and compared with external standards of disodium hydrogen phosphate and tetra sodium pyrophosphate, also at pH 10.0, to avoid the significant variation in chemical shifts seen on ³¹P NMR caused by the different phosphate ionization states at different pH values. ³¹P NMR was performed on a Bruker AM250 instrument operating at 101.3 0 MHz, equipped with a 10 mm broad-band probe. Samples were placed in 8 mm diameter tubes which were then housed in 10 mm tubes for analysis, with the region between the two containing D₂O for field-frequency locking. Data were acquired at ambient probe temperatures (22°C) and spectra were referenced externally to 80% H₃PO₄ at O.Oppm. At least 3 compounds were present: pyrophosphate (-4.8 ppm, 51% yield), phosphate (3.7ppm, 22% yield), an unassigned doublet closely associated with pyrophosphate (-4.1 ppm, 19% yield) and another unassigned multiplet (-18.5 ppm, 8% yield).

Electrospray ionization mass spectrometry

Crude phosphorylated deprotected oligoribonucleotides (about 50 nmole derived from GmAAUACUCAAG or U^), were electrophoresed on a 20% polyacrylamide gel in 7M urea in 1 X Tris- borate-EDTA (TBE) buffer and the main optical product in each case (detected by UV shadowing) was cut out, crushed and eluted with 1.0 ml 0.25 M ammonium acetate for 16 hours at 4°C with gentle shaking. After desalting on NAP-10 columns (see above) in water, the oligonucleotides were dried in vacuo and redissolved in 0.2 ml water and desalted again to remove traces of cations which interfere in the mass spectrometry (13). The samples (about 0.5 nmole) were then dissolved in 50 μl water. An aliquot of each of these samples was diluted to a concentration of 10 pmol/μl in a solution of 50% aqueous methanol containing 1% triethylamine for analysis by mass spectrometry as described (14). Briefly, electrospray ionization spectra of these samples were acquired using a PE Sciex (Norwalk, CT) API III+ triple quadrupole mass spectrometer. Mass analysis was made using only Q1, (calibrated in negative ion mode with d(CCCCCC)), analysing over the mass range 450-1600. Capping and f³²P] labelling of phosphorylated oligonucleotides

Capping of the four crude phosphorylated oligonucleotides (about 0.2 nmole) to give m⁷G³²ppp-labelled oligonucleotides was achieved using 1 unit (u) guanylyl transferase (Gibco BRL) and 1 mM [α-³²P] GTP (3000 Ci/mmole, Amersham) in 0.05 M Tris- HCI (pH 7.8), 1.25 mM MgCI₂, 6 mM KCI, 2.5 mM DTT, 20 u human placental ribonuclease inhibitor (Promega), 0.1 mM S-adenosyl methionine in a 5 μl reaction volume for 1 hour at 37°C. In some experiments bovine serum albumin (0.4 μg) was added. The reaction products were analysed, or in preparative experiments purified, by electrophoresis on 20% polyacrylamide 7M urea gels. The major radioactive band was detected by autoradiography and eluted in 0.25 M ammonium acetate, as above. The eluate was centrifuged to remove gel pieces and the RNA precipitated from the supernatant with 3 volumes of ethanol in the presence of 2M ammonium acetate and 20 μg yeast carrier RNA.

Other Analytical Methods

P1 nuclease (Boehringer) digestion was carried out in 30 mM ammonium acetate pH 5.3, 10 mM ZnSO₄ (15) for 30 min at 37°C using 30 μg yeast RNA to establish the enzyme concentration needed for complete hydrolysis to mononucleoside 5'-phosphates, as judged by TLC on a Macherey-Nagel polygram SIL G/UV254 sheet (Camlab) developed using propan-2-ol:H₂O:ammonia (70:30:1) and detecting nucleotides under a UV lamp. Phosphorylated gel purified ACACUUGCUUUUG (2.5 μg) and non-phosphorylated ACACUUGCUUUUG were then digested with P1 nuclease in a 10 ml reaction volume and, after checking by TLC on an aliquot that the digestion was complete, were analysed by HPLC (Beckman Gold) using a Beckman Spherogel-TSK DEAE-5PW (10 μm particle size, 7.5 x 75 mm) anion-exchange column using a gradient from 10 mM to 0.5 0 M triethylamine acetate pH 6.8. P1 nuclease cleaves oligonucleotides and 2'-O-methylated residues to give mononucleoside 5'-phosphate end- products derived from internal nucleotides. An oligonucleotide with a 5'- triphosphorylated end group would degrade to give a pppN 5' end group

T2 RNase (Sigma) digestion of crude phosphorylated (30 μg) and control non-phosphorylated ACACUUGCUUUUG (30 μg) was carried out in 0.05 M ammonium acetate pH 4.5 with 2 u/ml enzyme for 2 hours at 37°C followed by analysis by HPLC on a DEAE anion- exchange column, as above. T2 RNase cleaves RNA giving mononucleoside 3'-phosphates from internal positions and pppNp as the 5' end group from oligonucleotides with a 5'-triphosphorylated end-group.

Tobacco acid pyrophosphatase (Epicentre from Cambio, Cambridge) was used, following the manufacturer's instructions, to digest m⁷G³²pppGmAAUACUCAAG (see above) for 30 min at 37°C analysing products by electrophoresis on 20% 7 M urea polyacrylamide gels. [gamma-³²P] ATP and [α-³²P] GTP were digested as controls giving labelled phosphate and GMP markers, respectively.

Calf intestinal phosphatase (Boehringer) digestion of 20% polyacrylamide 7M urea gel purified phosphorylated GmAAUACUCAAG was followed by phenol/chloroform and ether extraction of the aqueous layer. Labelling of an aliquot of the aqueous layer was carried out using [gamma-³²P] ATP and T4 polynucleotide kinase. Products were analysed by 20% polyacrylamide 7M urea gel electrophoresis, followed by autoradiography. A control labelling of crude non- phosphorylated GmAAUACUCAAG, and phosphorylated oligonucleotide without prior phosphatase treatment with [gamma-³²P] ATP and T4 polynucleotide kinase was done in parallel.

Transcription using influenza A virus RNA polymerase in vitro

Transcription was carried out by standard methods (16) using influenza A virus (X-31) cores (not micrococcal nuclease treated) as a source of RNA polymerase, except that [³²P]-labelled capped oligonucleotides (see above) were used as primers instead of ApG and no [α³²P] labelled nucleoside triphosphate was added to the reaction mixture. An equimolar mixture of 14 nucleotide-long synthetic RNA (5^* GGCCUGCUUUUGCU 3') mimicking the sequence at the 3' end of virion RNA and a 15 nucleotide-long synthetic RNA 5' AGUAGAAACAAGGCC 3' mimicking the 5' strand of influenza virion RNA (the so-called "RNA-fork") was used as a template (16). The labelled transcripts formed by incubating at 30°C for 2 hours were analysed by 18% polyacrylamide gel electrophoresis in 7M urea and products detected by autoradiography.

An Improved Tri-n-butylammonium pyrophosphate preparation.

³¹P-NMR spectroscopy of tri-n-butylammonium pyrophosphate prepared according to (5) showed significant degradation of pyrophosphate to phosphate as indicated by ³¹P NMR. Since this may have been due to the chromatographic step on Dowex 50 (H⁺) in the procedure and the known acid lability of pyrophosphate, the tri-n- butylammonium pyrophosphate was prepared in later experiments by a procedure not involving exposure to acid. Briefly, a Dowex 50 (pyridinium) column of 75 ml packed volume was prepared and a solution of tetrasodium pyrophosphate (2.23g in 50 ml H₂O) passed through the column followed by water. 10 x 10 ml fractions were collected and their pH measured, pooling the pH=4.0 fractions (70 ml). 40 ml ethanol was added followed by 10 ml tri-n-butylamine. The solution was freeze-dried and the syrup dissolved in 50 ml ethanol. A white precipitate (possibly residual tetrasodium pyrophosphate) was centrifuged and discarded and the supernatant rotary evaporated to dryness twice from 50 ml ethanol. Finally, the syrup was dissolved in 30 ml dry DMF and freeze-dried twice. The preparation was finally dissolved in =10 ml dry DMF to which 4A 0 molecular sieves were added and 0.2 ml dry tri-n-butylamine. ³¹P-NMR (see above) showed the preparation to be >99% pure with a very minor phosphate contaminant. We added tri-n-butylamine to the final preparation in an attempt to protect against degradation observed on storage, since fresh tri-n-butylammonium pyrophosphate (which gave a single ³¹P-NMR peak when analysed immediately after preparation), was observed to be partially degraded after 4 months storage at room temperature in the absence of added tri-n-butylamine.

Automation of the 5'-diphosphorylation reaction Initial experiments were attempted on an ABI 394 synthesizer, following the phosphorylation conditions described in Browπlee et al (1995) (25), to prepare ppAmAAUACUCAAG and ppAmAmAmUACUCAAG (both at the 200 nMole scale). A "phosphorylation cycle" was generated which was essentially an adaptation of a standard base addition synthesis cycle. These two new phosphorylated oligonucleotides were worked up and [³²P] labelled (see above) and used as primers in a transcription reaction with influenza virus RNA polymerase (see above).

Definitive automated phosphorylation was then done using AGCUGACCUUU and a modified reagent 1 (as suggested in 5) which proved to be fully soluble. This was pyridine (1 ml), DMF (4ml) and 1 M salicyl phosphorochloridite in dioxan (0.5 ml). Reagent 2 was 1.5 ml tri-n- butylammonium pyrophosphate, 0.375 ml tri-n-butylamine. UV shadowing of the phosphorylation product on polyacrylamide gel electrophoresis (see above) showed a faster moving product suggesting a successful reaction and m⁷GpppAGCUGACCUUU was successfully prepared by the capping enzyme, as above. RESULTS

Analysis of phosphorylation products

Seven oligoribonucleotides 11 to 13 residues long (see Materials & Methods) were synthesized on the ABI synthesizer using standard solid phase methods and the 5'-dimethoxytrityl protecting group removed. Phosphorylation was performed (see Materials & Methods) on the protected oligonucleotides, while still attached to the solid phase support, either in a separate apparatus or automatically on an ABI 374 synthesizer using the phosphitylating reagent salicyl phosphorochloridite, followed by reaction with pyrophosphate and oxidation with iodine (See Materials and Methods, Fig 1). After deprotection and desalting on Sephadex G-25, the products were analysed by electrophoresis on 20% polyacrylamide gels in 7M urea. In each case, the major product of phosphorylation (band X) migrated slightly faster than the non- phosphorylated control oligonucleotide. About 90% of the starting material was converted to this product (results not shown).

Gel purified band X (derived from phosphorylation of GmAAUACUCAAG) was initially characterized by treatment with calf intestinal phosphatase. Any free 5' OH groups generated by such phosphatase treatment were labelled by [gamma-³²P] ATP and T4 polynucleotide kinase followed by gel electrophoresis (see Materials & Methods). As controls, band X and a control non-phosphorylated GmAAUACUCAAG were labelled with gamma-³²P ATP and T4 polynucleotide kinase, without any prior phosphatase treatment. This showed that phosphatase treated band X gave rise to a labelled product with an electrophoretic mobility indistinguishable from 5'-[³²P]-labelled GmAAUACUCAAG. This suggested that band X had one or more phosphates added to the 5' end of the oligonucleotide. Similar results were obtained with band X from phosphorylated U₁ Molecular mass measurements resulting from electrospray ionization mass spectrometry of gel purified band X (i.e. phosphorylated GmAAUACUCAAG or phosphorylated U^), are shown in Table 1. Reasonable quality spectra were obtained (see Fig. 2 for an example) and, in each case, the predominant component corresponded to the addition of a monophosphate with lesser amounts of the diphosphate. No peaks corresponding to the triphosphates were detected, in either case, although our methods may have failed to detect triphosphates in <10% yield.

Further evidence for the presence of mono and di- phosphorylated oligonucleotides was obtained by complete enzymatic hydrolysis of phosphorylated ACACUUGCUUUUG using either P1 nuclease or T₂ RNase digestion followed by DEAE anion-exchange HPLC (see Materials & Methods). P1 nuclease gave a 5' end-product, ppA which was definitively identified at the precise elution time of marker ppA, in approximately 0.3 molar yield per mole of oligonucleotide, as estimated from integration of the pA and ppA peaks on the chromatogram. No product was detected in the position of marker pppA suggesting that 5'- triphosphorylated products were absent, or in too low yields (<10%) to be detected. The T₂ RNase products ppAp and pAp were tentatively identified, in approximately equal yields, from their HPLC elution positions which were close to, but not identical to, marker pppA and ppA.

Capping of phosphorylated oligoribonucleotides

Initially, labelling of phosphorylated oligoribonucleotides with [a-³ P]-GTP and guanylyl transferase (see Materials and Methods) was done to provide further evidence for the presence of 5'-di or triphosphorylated oligonucleotides, since this enzyme is known to accept either as a substrate (17). Capping of each of the 7 different crude phosphorylated oligonucleotides gave rise to a major radioactive product on a 20% polyacrylamide gel with minor faster and slower-moving products. The major capped products derived from phosphorylated GmAAUACUCAAG and phosphorylated U„ had electrophoretic mobilities just slower than 5'-[³²P]-labelled pGmAAUACUCAAG and pU,,, respectively (results not shown). Further evidence for the authenticity of these capped products was the failure to observe their synthesis if the corresponding non-phosphorylated control oligonucleotides were used as substrates for guanylyl transferase. Finally, the presence of a phosphodiester linkage in the labelled capped structure, m⁷G³²pppGmAAUACUCAAG, was confirmed by digestion with tobacco acid pyrophosphatase (see Materials and Methods). A labelled product, pm⁷G, was identified by its similar mobility to a marker labelled pG on 20% polyacrylamide gel electrophoresis.

Labelled capped oligoribonucleotides are primers for influenza RNA polymerase

We have confirmed earlier work (4) that m⁷GpppGmAAUACUCAAG acted as a primer for influenza RNA polymerase in vitro and have investigated its priming properties with mutant templates. We also study 6 more capped oligonucleotides allowing firstly, a comparison of oligonucleotides differing only in the 2'-O-methyl group (the Cap 1 structure). Secondly, we study priming with a capped oligonucleotide m⁷GpρpACACUUGCUUUUG (the 5' end of rabbit β-globin mRNA, but lacking the 6-methyl and 2-O-methyl groups on the first A residue and the partial 2'-O- methylation of the second residue-the Cap 2 structure (18)) since β-globin mRNA has been extensively studied as a substrate in a coupled endonuclease cleavage and primed transcription reaction with influenza RNA polymerase in vitro (3). Thirdly, we study the model compound m GpppUn. Fourthly, we study the unnatural (tri 2'-O-methyl) capped analogue m⁷GpppAmAmAmUACUCAAG, and the compound m⁷GppρAGCUGACCUUU. 0 Fig 3 shows that equal radioactive counts of four ³²P-labelled capped primers initially tested gave differing yields of transcripts on polyacrylamide gel electrophoresis with influenza RNA polymerase in response to the added model partially duplex "RNA fork" as a template (see Materials and Methods). The 2'-O-methylated m⁷GpppGmAAUACUCAAG primed transcription to give a transcription product (TP) in about 10 times the yield of the products derived from the same primer lacking the 2'-O-methyl group (compare lanes 3 and 6). The transcript primed by m⁷GpppGmAAUACUCAAG (lane 3) is a doublet, 23 and 24 nucleotides long, previously characterized by partial T, RNase sequencing, which initiated transcription at the third nucleotide of the added template and was partially elongated at its 3' end by the addition of a non-templated nucleotide (16). As previously noted (4), elongation of primers was not quantitative. A single nucleotide was added to the primer which was not dependent on added RNA template (see Fig 3, lanes 2, 5 & 8). This may reflect premature termination occurring on endogenous RNA template present in the influenza RNA polymerase cores. The rabbit β- globin mRNA capped oligonucleotide, m⁷GpppACACUUGCUUUUG, gave a triplet product (lane 9) in intermediate yield (36% of the 2'-O-methylated oligonucleotide) at a position on gel electrophoresis consistent with its priming on the second and third nucleotide of the added template (3). We have not been able to convincingly detect a specific product with m⁷GpppU₁₁ as a primer (results not shown). Further analysis of the ³²P labelled capped primers m⁷GpppAmAAUACUCAAG, m⁷GpppAmAmAmUACUCAAAG and m⁷GpppAGCUGACCUUU were carried out in a similar manner to the experiments in Figure 3 (results not shown). Both m⁷GpppAmAAUACUCAAG and m⁷GpppAmAmAmUACUCAAAG acted as primers for influenza transcription in vitro, although the latter gave a transcription product in somewhat higher yield than the former, (which was comparable to that derived from m⁷GpppGmAAUACUCAAG - see Figure 3). DISCUSSION

The chemical literature on the synthesis of nucleoside triphosphates from nucleosides is very extensive, but to our knowledge such methods have not previously been applied to oligonucleotides synthesized by modern solid phase methods, except for the addition of a single 5' or 3' monophosphate using the phosphorylation reagent, 2-[2- (4,4'- dimethoxytrityloxy)ethylsulphonyl]ethyl-(2-cyanoethyl)-(N,N- diisopropyl)-phosphoramidite (19). Multiple phosphate additions are not possible with this reagent. We wished to add a di- or triphosphate to an oligoribonucleotide to enable it to be subsequently capped enzymatically and then used as a primer for the influenza A virus RNA polymerase. We chose to leave all protecting groups intact except the 5' OH group and carry out a solid phase phosphorylation, assuming the protecting groups would favour solubility and minimise side reactions. Initially, we obtained satisfactory yields of 5'-diphosphorylated oligoribonucleotides but could not detect δ'-triphosphorylated products (Fig 1). The main evidence for diphosphates was from electrospray ionization mass spectrometry of 2 oligonucleotides, supported by evidence from degradation by P1 nuclease and T₂ RNase on a third. In all 3 cases, 5'-monophosphorylated oligonucleotides contaminated the 5'-diphosphorylated oligonucleotides. Successful enzymatic capping with guanylyl transferase supported the evidence for the presence of a 5'-diphosphate but could not exclude a 5'- triphosphate, since both end-groups are efficient substrates for guanylyl transferase (17).

It seemed unlikely that the initial phosphitylation by salicyl phosphorochloridite giving compound 2 (Fig 1) was to blame for the failure to detect the synthesis of δ'-triphosphorylated oligonucleotides, since little non-phosphorylated oligonucleotide (compound 1) remained after the reaction. Probably, the pyrophosphate catalysed double-displacement reaction giving rise to compound 3 was not optimal. ³¹P NMR analysis showed that the tri-n-butylammonium pyrophosphate reagent used was impure and contained a major phosphate contaminant as well as at least one other minor unidentified phosphate compound. Presumably these contaminants arose during the preparation of tri-n-butylammonium pyrophosphate from tetra sodium pyrophosphate (5). Thus phosphate, rather than pyrophosphate, might have preferentially reacted with compound 2 (Fig 1) giving rise to 5'-diphosphorylated oligoribonucleotides (compound δ). Preliminary experiments to test tri-n-butylammonium phosphate (δ), instead of pyrophosphate, for the synthesis of 5'- diphosphorylated oligoribonucleotides have been attempted and are promising.

Later, a purified form of tri-n-butylammonium pyrophosphate was prepared (see Materials and Methods) and was used to make phosphorylated oligonucleotides.

The δ'-diphosphorylated and any potential 5'- triphosphorylated oligonucleotides were substrates for the enzymatic capping with guanylyl transferase. Five of the seven [³²P]-labelled capped oligoribonucleotides tested primed the transcription of a short RNA product using influenza A virus RNA polymerase in vitro. As predicted from previous work using a coupled endonuclease and priming assay (20), the 2'-O-methylated capped oligonucleotide, m⁷GpppGmAAUACUCAAG, was more efficient than the same oligoribonucleotide without the 2'-O-methyl group. This is the first time that the effect of the 2'-O-methyl group has been critically tested in a priming assay, uncoupled from the effect it might have on the endonucleolytic activity of the RNA polymerase. The capped sequence, m⁷GpppACACUUGCUUUUG, though lacking the 6-methyl group, the 2'-O-methyl group on the first A residue and the partial 2'-O- methylation of the C residue (Cap 2) of authentic rabbit β-globin mRNA, was also reasonably efficient as a primer. Our failure to detect priming with m⁷GpppU₁₁, even although m⁷GpppGmC(U)_n (where n is an undefined large number) is reasonably active in a coupled endonuclease priming assay (21), may be explained by the lack of a 2'-O-methyl group, although other factors may be involved. The fact that an unnatural capped analogue m⁷GpppAmAmAmUACUCAAG could prime influenza transcription in vitro is a novel finding, suggesting the feasibility of the synthesis of other analogues as primers or inhibitors. Overall, our results confirm that 2'-O- methyl groups are important but not essential for priming (20).

As stated in the introduction our chemical phosphorylation method may be ideal for the large-scale synthesis of short capped oligonucleotides. They are of great interest since short capped oligonucleotides with a 3'-phosphate end-group in the range of 4-9 nucleotides in length, can bind to the influenza RNA polymerase and specifically inhibit cap-dependent transcription in vitro (A). They are, therefore, potential specific antiviral compounds(4). It is now relatively simple to systematically investigate the optimal length, methylation state and sequence for the specific inhibition of cap-dependent transcription. Analogues (eg phosphorothioates, or 2'-O-alkyl derivatives (22)) of these short primers can easily be chemically synthesized and tested as substrates for capping by guanylyl transferase. If such analogues can be capped and inhibit cap-dependent transcription by the influenza RNA polymerase in vitro, they would be candidates for an antiviral drug since they are known to be more resistant than unmodified oligoribonucleotides to RNase and other nuclease digestion in vivo. With this in view, it is encouraging that m⁷GpppAmAmAmUACUCAAG acts as a primer for influenza transcription. It may be that the synthesis of capped oligonucleotides with masking phosphate groups, perhaps using aryl phosphates (23), would increase membrane solubility of these potential antiviral compounds in vivo. 0 Finally, a more detailed study of phosphate or pyrophosphate addition using the automated solid phase method is warranted to establish conditions for optimal di- or triphosphate synthesis. Ultimately, a procedure for the total chemical synthesis of capped oligoribonucleotides is desirable (24). Moreover, the chemical phosphorylation described here should succeed equally well with deoxyribonucleotides so that other biological applications for the use of δ'- di- or triphosphorylated DNA may emerge, e.g. in the antisense field, either for basic research or in the biotechnology industry.

REFERENCES

1. Ishihama.A. and Nagata,K. (1988) Crit. Rev. Biochem., 23, 27-76.

2. Garcia-Sastre.A. and Palese.P. (1993) Ann. Rev. Microbiol., 47, 766-790. 3. Plotch.S.J., Bouloy.M., Ulmanen.l. and Krug.R.M. (1981)

Cell, 23, 847-868.

4. Chung.T.D.Y., Cianci,C, Hagen.M., Terry.B., Matthews, J.T.,

Krystal.M. and Colonno.R.J. (1994) Proc. Natl. Acad. Sci., USA, 91, 2372-

2376. δ. Ludwig.J. and Eckstein, F. (1989) J. Org. Chem., 64, 631-636.

6. Gaur.R.K., Sproat.B.S. and Krupp.G. (1992) Tetrahedron Letters, 33, 3301-3304

7. Moffatt.J.G. (1964) Canad. J. Chem., 42, 699-604.

8. Hoard,D.E. and Ott.D.G. (1965) J. Am. Chem. Soc, 87,

1785-1788.

9. Tomasz,J., SimoncsitsA, Kajtar.M., Krug.R.M. and Shatkin.A.J. (1978) Nucleic Acids Res., 5, 294δ-29δ7.

10. Takaku.H., Konishi.T. and Hata.T. (1977) Chem. Lett., 6δδ- 668. 11. Lamond.A.I. and Sproat.B.S. (1994) In Higgins.S.J. and Hames.B.D. (eds.), RNA processing - a practical approach, IRL Press, Oxford, Vol I. pp.103-140.

12. Hogrefe.R.I., McCaffrey, A. P., Borozdina.LU., McCampbell.E.S. and Vaghefi.M.M. (1993) Nucleic Acids Res., 21, 4739- 4741.

13. Limbach.P.A., Crain.P.F., and McCloskey.J.A. (1995) Current Opinion Biotechnol., δ, 96-102.

14. Limbach,P.A., Crain.P.F., and McCloskey.J.A. (1996) J. Am. Soc. Mass Spectrom. 6, 27-39.

16. Furuichi.Y., Morgan, M., Muthukrishnan.S. and Shatkin.A.J.

(1976) Proc. Natl. Acad. Sci. USA, 72, 362-366. 16. Fodor.E., Pritlove.D.C. and Brownlee.G.G. (1994) J. Virol.,

68, 4092-4096. 17. Venkatesan.S., Gershowitz.A. and Moss.B. (1980) J. Biol.

Chem., 2δδ, 903-908.

18. Lockard.R.E. and RajBhandary.U.L. (1976) Cell, 9, 747- 760.

19. Hom,T. and Urdea.M. (1986) Tetrahedron Letters, 27, 4705- 4708. 20. Bouloy.M., Plotch.S.J. and Krug.R.M. (1980) Proc. Natl.

Acad. Sci. USA, 77, 3962-3966.

21. Krug.R.M., Broni.B.A., LaFiandra.AJ., Morgan, M.A. and Shatkin.A.J. (1980) Proc. Natl. Acad. Sci. USA, 27, 6874-6878.

22. Sproat.B.S., Iribarren.A.M. Garcia, R.G. and Beijer.B. (1991) Nucleic Acids Res., 19, 733-738.

23. McGuigan.C, Pathirana.R.N., Balzarini.J., deClercq.E. (1993), J. Medicinal Chem., 36, 1048-1052.

24. Fukuoka.K., Suda.F., Suzuki, R., Takaku.H., Ishikawa.M. and Hata.T. (1994), Tetrahedron Letters, 35, 1063-1066. 25. Brownlee, G G, Fodor, E, Pritlove, D C, Gould, K G and

Dalluge, J J, (1995), Nucl. Acids Res., 23: 2641-2647.

Table 1

Molecular mass measurements by mass spectrometry

Compound m/z (charge) M_rfd M_r calc. pGmAAUACUCAAG 717.4 (-5), 896.7 (-4), 1195.7 (-3) 3591.0 3591.2 ppGmAAUACUCAAG 916.8 (-4), 1222.5 (-3) 3670.9 3671.2 pUn 482.7 (-7), 563.4 (-6), 676.3 (-5), 3386.1 3385.9 845.4 (-4) ppu„ 494.0 (-7), 576.7 (-6), 692.4 (-5), 3466.1 3465.9 865.5 (-4)

FIGURE LEGENDS

FIG 2

Mass spectrograph of phosphorylated GmAAUACUCAAG. M is pGmAAUACUCAAG and M₂ is ppGmAAUACUCAAG.

FIG 3

Capped oligoribonucleotides serve as primers for transcription by influenza virus RNA polymerase. In vitro transcription was carried out using viral cores as the source of RNA polymerase, a synthetic RNA template and [³²P]-labelled capped oligoribonucleotide primers (m⁷GpppGmAAUACUCAAG, lanes 1-3; m⁷GpppGAAUACUCAAG, lanes 4- 6; m⁷GpppACACUUGCUUUUG, lanes 7-9) as described in Materials and Methods. Lanes 1, 4 and 7 have equal radioactive amounts of primers incubated without RNA polymerase and added RNA templates; lanes 2,5 and 8, primers elongated in the absence of RNA template but in the presence of enzyme; lanes 3, 6 and 9, primers elongated by 14-15 nucleotides in the presence of RNA polymerase and added RNA templates. PR = primer. The relative yields of transcription products (TP) in lanes 3,6 & 9 were in the ratio 100:10:36 as measured by laser densitometry.

Claims

1. A method which comprises reacting an oligonucleotide having a terminal δ'-OH group with a phosphitylating agent or phosphorylating agent and then with a phosphate or pyrophosphate and recovering an oligonucleotide having a terminal δ'-di- or triphosphate group.

2. A method as claimed in claim 1 wherein the oligonucleotide is linked to a solid phase.

3. A method as claimed in claim 1 or claim 2, wherein the oligonucleotide is an oligoribonucleotide.

4. A method as claimed in any one of claims 1 to 3, wherein there is recovered an oligonucleotide having a terminal 5'-di- or triphosphate group which is then capped by reaction with guanylyl transferase.

5. A method as claimed in any one of claims 1 to 3, wherein there is recovered an oligonucleotide having a terminal 5'-diphosphate group.

6. A method as claimed in any one of claims 1 to 5, wherein the starting oligonucleotide is linked to a solid phase and the phosphorylated oligonucleotide is recovered into solution.

7. A method as claimed in any one of claims 1 to 6, wherein the starting oligonucleotide is selected from

GAAUACUCAAG

GmAAUACUCAAG

ACACUUGCUUUUG

U„ AmAAUACUCAAG Am Am Am U A C U C A A G and A G C U G A C C U U U where Gm is 2'-O-methylguanosine, and

Am is 2'-O-methyladenosine.

8. Products of the general formula ppN_n or pppN_n where N is a nucleotide or a nucleotide analogue residue and n is equal to or greater than 2.

9. Products as claimed in claim 8, wherein n is 4 to 9.

10. Products as claimed in claim 8 which are: ppp or ppGAAUACUCAAG ppp or ppGmAAUACUCAAG ppp or ppU„ ppp or ppACACUUGCUUUG ppp or ppAmAAUACUCAAG ppp or ppAmAmAmUACUCAAG ppp or ppAGCUGACCUUU.

11. Capped products which are m⁷GpppGAAUACUCAAG m⁷GpppU₁₁ m⁷GpppACACUUGCUUUG m⁷GpppAmAAUACUCAAG m GpppAmAmAmUACUCAAG m⁷GpppAGCUGACCUUU.

12. Capped products of the general formula m⁷GpppNmNm(N)_x where N is a nucleotide or nucleotide analogue residue, m is a 2'-O- methylated residue and x is equal or greater than 1.