A new universal linker for solid phase DNA synthesis

 1996 Oxford University Press

Nucleic Acids Research, 1996, Vol. 24, No. 14 2793–2798

A new universal linker for solid phase DNA synthesis Matthew H. Lyttle*, Derek Hudson and Ronald M. Cook Biosearch Technologies Inc., 40 Mark Drive, San Rafael, CA 94903, USA Received February 12, 1996; Revised and Accepted May 29, 1996

ABSTRACT A method is described as an alternative to the use of nucleoside pre-functionalized supports for DNA synthesis. The procedure should allow the generation of 3′-OH terminal moieties of any natural or modified DNA fragment using a single derivatized solid support material. The method utilizes 1-O-(4,4′dimethoxytrityl)-2O-succinoyl-3-N-allyloxycarbonylpropane immobilized on amino-propyl CPG followed by subsequent coupling of unit phosphoramidites. Work up is accomplished by removal of the 3-N-allyloxycarbonyl group [Pd(0) at 50
C for 15 min] followed by cleavage under very mild conditions (aqueous TEAA/NH3 buffer pH 10, room temperature) to release the desired product. The mechanism is believed to involve nucleophilic attack of the linker-derived amino group on the 3′-phosphate triester, followed by elimination of the desired product. DNA synthesis with the new support and with classical nucleotide synthesis supports have been performed, and the products shown to be identical. Further proof of product integrity was given by MALDI mass spectral studies and the efficacy of DNA primers made with the new support in PCR amplification. INTRODUCTION Conventional oligonucleotide synthesis utilizes solid supports prederivatized with nucleoside hemisuccinates. To simplify this strategy several groups (1–4) have described synthesis methods in which the 3′ terminus of a synthetic oligonucleotide is incorporated during the first addition of monomeric phophoramidite units. These alternative methods rely on a tetrahydropyran vicinal diol based linker system, in which the phosphate bond to the 3′-terminus of the desired DNA fragment is cleaved concomitant with cyclic phosphate formation with the diol. Although these procedures work well, the strongly basic conditions required are too harsh to be compatible with some nucleotide synthesis methods. For example, DNA synthesis involving biotinylated residues or base sensitive dyes may not be compatible with strongly basic conditions. Our methodology stems from solid phase chemistry previously used to produce 3′-[(hydroxylpropyl)amino] functionalized oligonucleotides by application of a FMOC-N-protected amino propane diol linker (5). In later work, Vu et al. compared a series of different base labile amine protecting groups (6). In this study, a small amount of 3′-hydroxyl oligonucleotide was observed as * To

whom correspondence should be addressed

an unwanted product with some of the more base labile amine masking groups. We surmised that generation of the amine while the 3′-phosphate was still protected as a triester could result in attack at the phosphorus atom by the nucleophilic nitrogen, with subsequent elimination of the 3′ hydroxy nucleotide from the cyclic intermediate formed. Vu et al. were able to eliminate this side reaction by using more base stable amine protecting groups to keep the nitrogen protected, presumably until all of the 3′ phosphate was hydrolyzed to the diester stage. We have found, conversely, that by using an amine protecting group which is removed under neutral conditions, the ‘side-reaction’ described by Vu can be used to generate pure 3′-hydroxynucleotides. MATERIALS AND METHODS Acetic acid, pyridine, methanol (MeOH), aqueous ammonia, potassium hydrogen phosphate, sodium sulfate, ammonium acetate and conc. hydrochloric acid were reagent grade from J. T. Baker; ethyl acetate (EtOAc), dimethylformamide (DMF), dichloromethane (DCM) and acetonitrile were Omnisolve grade from VWR. Tetrahydrofuran (THF), sodium hydroxide, sodium carbonate, sodium bicarbonate, N-methyl imidazole, diisopropylcarbodiimide (DIPCDI), triethylamine (TEA), acetic anhydride, lithium chloride and magnesium sulfate were from Aldrich. Hydroxybenzotriazole (HOBT) and benzotriazole-1-yl-oxy-Tris (dimethylamino) phosphonium hexafluorophosphate (BOP) were obtained from Chem Impex. Aminopropyl 1000 Å, 500 Å and nucleoside derivatized controlled pore glass (CPG) was obtained in house from Biosearch Technologies, Inc. DNA synthesis was performed on a Biosearch 8750 synthesizer, with Biogenex (San Leandro) DNA amidites and Clontech (Palo Alto) RNA amidites. All other DNA synthesis reagents were the same as those previously reported (7). Elemental analysis and mass spectra were performed by the University of California at Berkeley analytical services dept. and MALDI mass spectra of oligonucleotides were obtained by the University of Michigan Protein and Carbohydrate Structure Facility. Synthesis of the universal linker 3-N-(allyloxycarbonyl)-1,2-propane diol, 4 A solution of 10 g (109.8 mmol) of 3-amino-1,2-propane diol 2 was prepared in a mixture of 200 ml of water and 50 ml of THF. To this was added 10 g K2CO3, followed by dropwise addition, over 30 min, of 30 ml (34.08 g, 282.7 mmol) allyl chloroformate 3 in 100 ml of THF. The reaction was stirred for 2 h, with more K2CO3 added, as needed, to maintain a pH of 9–10. The mixture

2794 Nucleic Acids Research, 1996, Vol. 24, No. 14 was cautiously acidified to pH 4 with dilute HCl, and extracted twice with 200 ml EtOAc. The combined organic layers were dried over Na2SO4 and concentrated to 8 g (42% yield) of a clear oil. The oil was negative to a ninhydrin test whereas the starting material was strongly positive. 1H NMR, 360 mHz, CDCl3, d: 3.2 (dt, 1H), 3.3 (dt, 1H), 3.5 (dd, 1H), 3.6 (dd, 1H), 3.75 (m, 1H), 3.9 (broad s, 2H), 4.5(d, 2H), 5.2(dd, 2H), 5.3(dd, 2H), 5.9(m, 2H). Anal. calcd. for C7H13NO4: C, 47.99. H, 7.48. N, 8.00. Found: C, 47.65. H, 7.63. N, 7.39. 3-N-(allyloxycarbonyl)-1-O-(4,4′dimethoxytrityl)-2-propanol, 5 A solution of 8 g (45.7 mmol) of 4 dissolved in 100 ml of pyridine was reduced to an oil in vacuo. The oil was dissolved in 200 ml of pyridine, and 18 g (53 mmol) of 4,4′dimethoxytrityl chloride was added. The red mixture was stirred overnight. MeOH, 20 ml, was added to the mixture, and the solvent was removed in vacuo after 20 min. The residue was dissolved in 300 ml of EtOAc, and the organic phase was washed with water, 200 ml, and dried over MgSO4. The solution was filtered and reduced to a tar in vacuo. A 4 × 40 cm silica column was prepared with 1% v:v pyridine in DCM. The crude product was loaded onto the column and eluted with 1 l of this solvent, followed by 1 l of 1% MeOH–DCM, then 1 l of 2% MeOH–DCM. Fractions of 500 ml were collected, and those which contained pure product, Rf 0.73 (5% MeOH–DCM, aluminum backed silica plates) were pooled to give 7.4 g (31.5% yield) of an orange oil. M/e (relative intensity) 477 (M+, 6), 438(1), 303(100), 154(11), 136(10). 1H NMR, 360 mHz, CDCl3, δ: 3.2(m, 3H), 3.4(m,1H), 3.8(s, 6H), 3.9(m, 1H), 4.6(d, 2H), 5.6(broad s, 1H), 5.7(dd, 1H), 5.8(dq, 1H), 5.9(dq,1H), 6.8(s, 1H), 6.9(m, 4H), 7.2(m, 1H), 7.3(m, 6H), 7.4(d, 2H). High resolution mass spectrum: calc’d: 477.2151. Found: 477.2150. 1-O-(4,4′dimethoxytrityl)-2-O-succinoyl-3-N-allyloxycarbonyl propane 1 A solution of 5 g (10 mmol) of 5 in 100 ml of pyridine was reduced to an oil in vacuo. The oil was dissolved in 200 ml of pyridine, and 10 g (100 mmol) of succinic anhydride was added, along with 1 ml of N-methylimidazole. The mixture was allowed to stand overnight, after the solids were dissolved by swirling the flask. MeOH (20 ml) was added, and the solvents were removed in vacuo. A 4 × 40 cm silica column was prepared with 1% v:v triethylamine in DCM. The crude product was loaded onto the column and eluted with 1 l of this solvent, followed by 2 l of 1% MeOH–DCM, then 1 l of 4% MeOH–DCM, then 1 l of 6% MeOH–DCM. Fractions of 500 ml were collected, and those which contained pure product, Rf 0.27 (5% MeOH–DCM, aluminum backed silica plates) were pooled to give 5 g (84% yield) of 1 triethylammonium salt as an orange oil. 1H NMR, 360 mHz, CDCl3, δ: 1.25(t, 9H), 2.6(m, 4H), 3.1(q, 6H), 3.2(d, 2H), 3.3(m, 1H), 3.5(m, 1H), 3.8(s, 6H), 4.5(d, 2H), 5.0 - 5.3(m, 3H), 5.9(dq, 1H), 6.9(d, 4H), 7.2(dd, 1H), 7.3(m, 7H), 7.4(d. 2H), 8.6(d, 1H). Anal. calcd. for C38H50N2O9 . 1/2 H2O: C, 66.35. H, 7.47. N, 4.07. Found: C, 66.20. H, 7.87. N, 4.34. Immobilization of 1 on controlled pore glass (CPG) In a 125 ml erlenmeyer flask, 1 (1.0 g, 1.6 mmol) was dissolved in 22 ml of DMF. Hydroxybenzotriazole (HOBT), 150 mg (1.1 mmol) was added, and the mixture was swirled until this dissolved. Next, 0.2 ml (161 mg, 1.3 mmol) of DIPCDI was

added, followed immediately by 10 g of aminopropyl 1000 Å CPG. The mixture was allowed to stand overnight, whereupon preliminary testing of the loading gave 12 µmol/g. Another 1 g of 1 and 0.2 ml more DIPCDI were added, and the mixture was again allowed to stand overnight. The support was washed with two 100 ml portions of acetonitrile, and then excess amino groups on the support were acetylated with 100 ml of a mixture of 5:5:8:82 acetic anhydride:pyridine:N-methylimidazole:THF for 1 h. The support was then rinsed with two 100 ml washes of acetonitrile, two 100 ml washes of MeOH and two 100 ml washes of DCM. After overnight drying in vacuo the loading was determined to be 17 µmol/g. For 500 Å CPG, the following procedure gave good results. In a 125 ml erlenmeyer flask, 10 g of aminopropyl 500 Å CPG was slurried with ∼20 ml of acetonitrile, and a solution of 0.20 g (0.29 mmol) 1, 50 mg (0.4 mmol) of HOBT and 0.20 g (0.5 mmol) of BOP in 4 ml of 0.3 M N-methylmorpholine in acetonitrile was mixed and allowed to stand 5 min. The mixture was added to the slurry containing the CPG, and allowed to stand overnight after thorough mixing. The support was washed with two 100 ml portions of acetonitrile, and excess amino groups on the support were acetylated with 100 ml of a mixture of 5:5:8:82 acetic anhydride:pyridine:N-methylimidazole:THF for 1 h. The support was then rinsed with two 100 ml washes of acetonitrile, two 100 ml washes of MeOH and two 100 ml washes of DCM. After overnight drying in vacuo the loading was determined to be 24 µmol/g. Procedure for deprotection and cleavage of DNA from the universal support Automated DNA synthesis (8) is concluded with the 5′-DMT group either on or off, and a mixture of 25 mg tetrakistriphenylphosphine palladium(0), 50 mg ammonium acetate hydrate and 100 mg triphenyl phosphine in 1 ml THF is heated to 50
C for 2 min. About 200 ml of the yellow solution is taken up in a 1 ml syringe and ∼1/2 of this is passed into the oligonucleotide synthesis column containing the support bound nucleotide. The column, with syringe attached, is placed in a previously warmed 13 × 100 mm test tube and heated in an aluminum hot block at 50
C. After 10 min, the rest of the solution is forced through the column, and after 5 min the column is removed from the tube and washed with 5 ml acetonitrile, either on the DNA synthesizer or by syringe. Next, a solution of 1 ml 0.1 N TEAA, pH 8.5, is mixed with 40 µl of 3% aqueous ammonia, and 0.5 ml of this solution is taken up in a syringe. Over 2 h, this solution is pushed through the column in small increments, with the effluent collected in an eppendorf tube. The column is then further rinsed with 0.5 ml 50% acetonitrile in water, and the combined effluent is evaporated in vacuo. The residue is then subjected to concentrated ammonia for 5 h at 55
C, and evaporated for subsequent purification or analysis. Analysis of DNA Quantitation was performed at 254 nm with a Beckmann DU spectrophotometer. Reverse phase HPLC conditions The dried sample was dissolved in 20% acetonitrile in water, and 2–20 µl of this solution, depending on the concentration, was

2795 Nucleic Acids Acids Research, Research,1994, 1996,Vol. Vol.22, 24,No. No.114 Nucleic

Figure 1. Synthesis of universal linker 1.

injected onto a Perkin Elmer Reverse Phase 3.3 cm Cartridge. The column was eluted with a gradient of 100% A (0.1 N ammonium acetate) for 2 min, 100–90% A (B was 100% acetonitrile) over 10 min, then 90–50% A over 10 min, then back to 100% A over 1 min. The flow rate was 1 ml/min, with absorbance read at 260 nm. Ion paring HPLC conditions The dried sample was dissolved in 20% acetonitrile in water, and 2–20 µl of this solution, depending on the concentration, was injected onto a 16 × 40 mm long column of 7 micron Aspect (Biosearch Technologies Inc., San Rafael) polyethylene support. Buffer A was 2 mM tetrabutylammonium sulfate (TBAS) in 25 mM borate buffer, pH 6.5 containing 5% acetonitrile; B was 2 mM TBAS in 25 mM borate buffer, pH 10.5 containing 50% acetonitrile. The gradient was 1 min of 100% A, flow 1 ml/min; 2 min of 100% A at 2 ml/min, then a linear gradient to 100% B over 30 min at 2 ml/min, with absorbance read at 260 nm. RESULTS AND DISCUSSION We describe use of a solid support functionalized with a linker 1 which allows the incorporation, during automated DNA synthesis, of the 3′ terminus of the polynucleotide (through coupling of the corresponding phosphoramidite derivative), instead of by prederivatisation processes. The synthesis of the linker 1 is shown in Figure 1. The readily available, economical 3-amino-1,2 dihydroxy propane (2) was reacted with allyl chloroformate (3). The resulting diol (4) was then treated with 4,4′-dimethoxytrityl chloride (DMT-Cl) to give alcohol (5), which was then succinylated with succinic anhydride and N-methylimidazole in pyridine to afford 1. Attachment of 1 to the solid support (aminopropyl CPG) was accomplished with either diisopropylcarbodiimide (DIPCDI) and hydroxybenzotriazole (HOBt) in DMF solvent or hydroxybenzotriazole (HOBT) and benzotriazole-1-yl-oxy-Tris (dimethylamino) phosphonium hexafluorophosphate (BOP) with N-methylmorpholine in acetonitrile. Synthesis of deoxyoligonucleotides on this solid support was accomplished by the amidite method (8). Removal of the allyloxycarbonyl group from the immobilized oligonucleotide product was accomplished with tetrakis-triphenylphosphinepalladium(0), triphenylphosphine, and a saturated solution of ammonium acetate in THF at 55
C for 15 min. No loss of DNA

2795

from the support occurred during this treatment (as shown by DMT loading analyses before and after the allyl removal treatment). After allyl removal, the support was washed, and cleavage of the nucleotide from the solid support (to give a 3′-hydroxyl oligonucleotide) effected with 0.1 M triethylammonium acetate buffer and ammonia, pH 10, for 2 h at 20
C. The solution was then evaporated, and the residue treated with ammonia and heat in the usual way to remove side-chain protecting groups. The key to the successful application of the method lies in the generation of an amino functionality while the phosphate is still protected as a triester. This is permitted by palladium (0) mediated removal of the allyloxycarbonyl amino protecting group with ammonium acetate as the acceptor. Similar allyl deprotection conditions, with somewhat longer exposure times, have been previously reported by Hayakawa et al. (9). Other successful allyl deprotection chemistries have used morpholine (10), which might also eliminate the cyanoethyl phosphate groups which are crucial to the successful 3′-hydroxy nucleotide generation. For convenience, all manipulations of allyl removal and DNA cleavage can be performed while the support is still in the synthesis column. A solution containing the DNA can be collected in a small tube for drying and further treatment. Based on the studies of Nelson (5) and Vu (6), it is proposed that during the pH 10 buffer treatment following allyl removal, nucleophilic attack occurs at the 3′ phosphate atom by the nitrogen giving a pentacoordinate phosphorus intermediate which, subsequently, can expel either the desired 3′ hydroxyl nucleotide or eliminate the cyanoethyl group (Fig. 2). The intermediate formed (6) is also favorably disposed towards elimination of the oligonucleotide upon subsequent hydrolysis, since any other pathway would disrupt the six-membered ring. Indeed, stronger base treatment after the mild base treatment does liberate additional 3′-hydroxy DNA (Table 1), which is consistent with the proposed mechanism. The integrity of the products produced by the procedure was verified by HPLC and MALDI mass spectrometry (Figs 3, 4 and 5). First, a dinucleotide, 5′-CT-3′, was made on the new support by coupling T followed by C phosphoramidites in normal automated DNA synthesis, and cleaving the nucleotide with the 2 h buffered procedure. This product was compared with 5′-CT-3′ made by coupling C amidite onto conventional T-succinoylaminopropyl CPG, followed by the conventional support workup. Reverse phase HPLC (Fig. 3, DMT off products) showed that both syntheses produced the same major product (by coelution). Next, a series of four longer DNA fragments 15–25 base units in length were prepared with the new support. For these longer fragments, the yield of DNA obtained with the 2 h buffer treatment is 62–80% of what is obtained by direct ammonia cleavage of the CPG at a 200 nmol synthesis scale (Table 1). The purity of these DNA fragments matched that of the same fragments made with standard CPG in each case. Additional DNA was obtained by ammonia treatment of the support after the buffer treatment (Table 1). This material appeared to be as pure, by HPLC, as that obtained with the buffered cleavage (data not shown). When an accelerated cleavage procedure was used (10 min of conc. ammonia at room temperature, then heat) or other more aggressive treatments, HPLC and PAGE sometimes showed other species in these products. MALDI mass spectral analyses (Fig. 5) of a 16mer subjected to the palladium(0) mixture followed by ammonia and heat without the buffer treatment showed that 3′ phosphate diesters containing 3-amino-1,2-propane

2796 Nucleic Acids Research, 1996, Vol. 24, No. 14

Figure 4. Ion paring HPLC of 16mer ATACTTATCATGAGCC: A = nucleotide made with standard prederivatized CPG; B = universal support synthesis result; C = coinjection. Chart speed 3 cm/min.

Figure 2. Proposed cleavage mechanism.

Figure 3. 5′-CT-3′ reverse phase HPLC analysis. A = dinucleotide made with standard prederivatized CPG; B = dinucleotide made with universal linker; C = coinjection. Chart speed 1 cm/min.

diol were present, as would be expected if phosphate diesters were generated by loss of the cyanoethyl group at the 3′ terminus before the nitrogen can cyclize. The MALDI spectrum of the buffer cleaved 16mer did not show these impurities, and was very similar to that obtained with standard CPG. Pb2+ ions have been previously reported (3) as beneficial in accelerating the cleavage of a vicinal diol based universal linker, and recently lithium and sodium salts have also been used in this application [Yu, C. J. (1995) Abstract # 51, A.A.C.C. Nucleic Acids Conference, San Diego]. In our case, the cleavage reaction

Figure 5. MALDI Mass spectra of: (A) ATACTTATCATGAGCC made with standard CPG; (B) same sequence made with immobilized 1 with buffered cleavage; (C) same sequence made with immobilized 1 and ammonia only cleavage.

was also accelerated by the addition of lithium chloride to the TEAA/NH3 buffer, and DNA made with this procedure appeared to be identical, by HPLC, to fragments made with the 2 h buffer treatment. The cleavage reaction was finished after ∼30 min with the inclusion of 0.5 M LiCl in the buffer mixture, and the yield of DNA obtained was better. The added LiCl also allowed the cleavage reaction to occur with good yield at pH 8.5, whereas the buffer only conditions gave poor yields at this pH. These advantages are offset somewhat by the introduction of mineral salt into an otherwise completely volatile buffer system.

2797 Nucleic Acids Acids Research, Research,1994, 1996,Vol. Vol.22, 24,No. No.114 Nucleic

2797

Table 1. 200 nM DNA synthesis with 12 mg 17 µm/g functionalized universal support DNA sequence

Length

C.E.a

CGATCTGAATAGCTT

15

97.3

ATACTTATCATGAGCC

16

98.4

TCCACGTCATCGAGGTCATA

20

GATGAGTCCGTGTCCGTACAACTGG

25

NH4OH cleaveb

Buffer cleavec

Base washd

8.9

5.8

2.0

8.4

5.2

1.6

98.7

9.6

7.6

3.1

98.0

11.2

8.2

3.0

aCoupling

efficiency, 2nd and last DMT colors used, diluted and read at 489 nm. from 10 min ammonia at room temperature treatment of 12 mg of CPG. cODs from 2 h buffer treatment of 12 mg of CPG. dODs from 10 min ammonia at room temperature treatment of CPG after 2 h buffer treatment. bODs

Figure 7. Reverse phase HPLC traces of T monomer cleavage from universal support. A = T standard. B = T cleaved from universal support with 2 h buffer treatment. C = T cleaved from universal support with 10 min ammonia treatment. Chart speed 1 cm/min.

Figure 6. Electrospray mass spectrum of polar biproduct of DMT-T monomer cleavage from universal support.

Further insight into the mechanism of the cleavage reaction was gathered by model experiments. A study wherein DMT-T amidite was coupled onto the universal support, followed by the palladium cocktail treatment and mild base cleavage procedure gave DMT-T almost exclusively. This shows that strongly basic conditions are not needed to remove the 3′ phosphate. A small amount of a more polar product was isolated by column chromatography, and the mass spectrum (Fig. 6) of the material was consistent with a cyclic phosphoramidate containing T nucleoside as the other pendant phosphate ester alcohol. The presence of this product supports the mechanism presented above. The same T amidite functionalized CPG was detritylated and the products analyzed by reverse phase HPLC (Fig. 7). T nucleoside was observed in ∼95% purity. A more aggressive base treatment (conc. ammonia, 10 min at room temperature, then 55
C) gave less pure material (∼70%). The integrity of the 3′ hydroxy group in DNA synthesized by our method was further demonstrated by the PCR activity of primers made with the universal support. Two 25mer fragments, 5′-GATGAGTTCGTGTCCGTACAACTGG-3′ and 5′-GGTT-

ATCGAAATCAGCCACAGCGCC-3′ were synthesized with immobilized 1 and purified with a reverse phase cartridge (11). The two fragments define a 500 base pair (bp) segment of whole bacteriophage λ template DNA. These were effective, at a 1 µM final concentration, in PCR amplification (12) of the 500 bp segment after 25 cycles. PCR results were analyzed on a 1% agarose gel with ethidium bromide staining (data not shown). Finally, utility of immobilized 1 in RNA synthesis was explored by synthesizing 5′-U10-3′ and 5-T9U-3′ with both immobilized 1 and a conventional pre-derivatized U support. The 2′ hydroxyl groups were protected as their tert-butyl dimethylsilyl ethers. HPLC analysis showed that immobilized 1 gave low yields of impure products. The conditions for cleavage were the same as those given above for DNA cleavage (aqueous TEAA/ NH3 buffer, pH 10, room temperature). Controls made with the standard RNA synthesis methods (7) gave acceptable results. One explanation of the RNA results with immobilized 1 is that steric hindrance of the bulky 2-O-trialkylsilyl ether discourages initial attack of the amine onto the 3′-phosphate, and thus diminishes the rate of the cleavage reaction. This explanation lends further support to the proposed mechanism for DNA cleavage given above. CONCLUSION A method which allows for incorporation of the 3′-terminal residue of DNA during automated synthesis is presented. The new method will extend the art of oligonucleotide synthesis, as well as eliminate the need for four different DNA supports. In

2798 Nucleic Acids Research, 1996, Vol. 24, No. 14 contrast with previous universal supports, our method requires extraordinarily mild conditions for work-up. This utility may be optimally exploited when used in conjunction with hyper base labile DNA synthons (e.g. phenoxyacetyl or dimethylformamidine) which are now commercially available. A further application of our linker arm offers the possibility of preparing fully protected oligomeric blocks since the benign cleavage conditions do not disturb traditional phosphate and exocylic amine protection. REFERENCES 1 Koster, H. and Heyns, K. (1972) Tett. Lett. 16, 1531–1534. 2 Crea, R. and Horn, T. (1980) Nucleic Acids Res. 8, 2331- 2348. 3 Gough, G. R., Brunden, M. J. and Gilham, P. T. (1983) Tett. Lett. 24, 5321–5324.

4 Scott, S., Hardy, P., Sheppard, R. C. and McLean, M. J. (1994) In Epton, R. (ed.) Innovations and Perspectives in Solid Phase Synthesis. Mayflower Worldwide Limited, Birmingham, pp. 115–124. 5 Nelson, P.S., Frye, R. A. and Liu, E. (1989) Nucleic Acids Res. 18, 7187–7194. 6 Vu, H., Joyce, N., Rieger, M., Walker, D., Goldknopf, I., Hill, T., Jayaraman, K. and Mulvey, D. (1995) Bioconjugate Chem. 6, 599–607. 7 Wang, Y. Y., Lyttle, M. H. and Borer, P. N. (1990) Nucleic Acids Res. 18, 3347–3352. 8 Sinha, N. D., Biernat, J. and Koster, H. (1983) Tett. Lett. 24, 5843–5846. 9 Hayakawa, Y., Wakabayashi, S., Kato, H. and Noyori, R. (1990) J. Am. Chem. Soc. 112, 1691–1696. 10 Lyttle, M. H. and Hudson, D. (1992) in J. Smith and J. Rivier (eds) Peptides, Chemistry and Biology. Escom, Leiden, pp. 583–584. 11 Biosearch Technologies, Inc., reverse phase DNA purification cartridge and protocols. 12 Instructions and all other reagents for PCR were from the Perkin Elmer (ABI division, Foster City) Gene Amp PCR amplification kit.