Novel route to oligo(deoxyribonucleoside Xfy ... - BioMedSearch

3 downloads 66 Views 889KB Size Report
Nucleic Acids Research, Vol. 19, No. 21 5883- .... p-toluenesulfonic acid for deprotection of the 5'-hydroxyl function. .... Time (min). 1. a) Dichloroacetic acid in.
k.) 1991 Oxford University Press

Nucleic Acids Research, Vol. 19, No. 21 5883 -5888

Novel route to oligo(deoxyribonucleoside phosphorothioates). Stereocontrolled synthesis of P-chiral oligo(deoxyribonucleoside phosphorothioates) Wojciech J.Stec, Andrzej Grajkowski, Maria Koziolkiewicz and Bogdan Uznanski Polish Academy of Sciences, Centre of Molecular and Macromolecular Studies, Department of Bioorganic Chemistry, Sienkiewicza 112, Lodz 90-363, Poland Received August 23, 1991; Accepted September 13, 1991

ABSTRACT The synthesis and separation of diastereoisomerically pure 5'-O-DMT-nucleoside 3'-0-(2-thio-1 ,3,2-oxathiaphospholane) allows their use as synthons in DBUcatalyzed reaction with the 5'-hydroxyl function of solid-support-bound nucleoside moiety. Since this reaction is stereospecific (>99%), this novel method allows preparation of oligo(nucleoside phosphorothioates) (1) with predetermined chirality at each P-chiral internucleotide phosphorothioate centre.

INTRODUCTION The original observation of increased stability of

dialkyl(aryl)phosphorothioates towards nucleases (1) became the basis for the application of oligonucleotide phosphorothioates (1) in modulation of gene expression, as successfully demonstrated by Matsukura et al. (2) and Agrawal et al. (3). It has also been demonstrated recently that double-stranded phosphorothioate oligonucleotides are able to accumulate in cells more effectively than native double-stranded oligonucleotides, and to modulate gene expression in a specific manner, most probably by competition for binding of specific transcription factors, and in this way may provide antiviral, immunosuppressive, or other therapeutic effects (4). Phosphorothioate oligonucleotides prepared by currently available technology (5) consist of a mixture of 2n diastereoisomers, where n is the number of P-chiral phosphorothioate centers. Unfortunately, there are no available analytical tools for direct measurement of the reproducibility of preparing a diastereoisomeric population of phosphorothioate oligonucleotides for n>3. Since in the aforementioned biological studies phosphorothioate oligonucleotides are used in concentrations several-to-many orders of magnitude higher than their targets such as mRNA (when applied in an antisense manner) or regulatory proteins, one may assume that a target 'selects' the most appropriate population of diastereoisomers with regard to the chirality of phosphorothioate linkages. On the other hand, documented stereoselectivity of some nucleases towards dinucleoside (3', 5') phosphorothioates (1) suggests that stereocontrolled synthesis of phosphorothioate oligonucleotides with predetermined chirality at each P-stereogenic center might allow one to obtain certain

diastereoisomers which are most resistant towards nucleolytic enzymes, thus increasing their life-time, and in this way decreasing the required amount of xenobiotic material for a given

therapeutic effect. In addition to this practical aspect, stereo-defined phosphorothioate oligonucleotides could become uniquely informative tools for studies of mechanisms of their interaction with target macromolecules, and, in theory, their stereodifferentiated cell-uptake and, distribution, and metabolism can also be considered. Although there are several reports on the stereocontrolled preparation of dinucleoside phosphorothioates (6, 7), in only one case has there been successful preparation of a longer molecule, namely (RpRpRp)- and (SpSpSp)-tetra[thymidine phosphorothioate] (8). However, this later synthesis was performed in solution, and further application of the methodology to automated solid phase synthesis of longer oligonucleotides seems to be unlikely (8). B

DMTO

DMTO

0\

/

B

B

HO

X-----

A)°'''''

X/ \/\

/)

Xfy

s

deprotJcleav.

B,B = A!k, GiU, &, T

~~~~~~~~~~~~~~~

5884 Nucleic Acids Research, Vol. 19, No. 21 In this report we wish to present a novel approach to the chemical synthesis of phosphorothioate oligonucleotides (1). More importantly, this approach allows stereocontrolled synthesis of I and utilizes base-catalyzed nucleophilic substitution at a pentavalent phosphorothioyl center. It has been previously reported from this laboratory that alkylene [180]1,2-oxides react with unsymmetrical dialkyl phosphorothioates with full retention of configuration at phosphorus (9). In subsequent studies it was demonstrated that reaction of O-alkyl-O-aryl phosphorothioates with alkylene [180]1,2-oxides in [170]H20 proceeds stereospecifically with formation of 0alkyl-[160, 170, 180]phosphate (10). This stereochemical result was rationalized in terms of involvement of intermediates containing a 2-oxo-1,3,2-oxathiaphospholane ring system. The observed stereospecificity for the overall process prompted us to formulate a hypothesis in which nucleoside 3'-0[2-thio-1,3,2oxathiaphospholane] (2), if successfully separated into diastereoisomeric species, may undergo nucleophilic attack by the 5'-hydroxyl function of a 3'-blocked nucleoside (3) in a stereospecific manner. Subsequent elimination of ethylene sulfide should provide dinucleoside (3', 5')phosphorothioate (4). To check this hypothesis, we prepared compounds 2, by reaction of appropriately protected nucleosides with 2-N-diisopropylamino-1,3,2-oxathiaphospholane (5) (11) (bp 700 C/O. 1 mm Hg, 631p 141 ppm), in the presence of tetrazole as catalyst. Without isolation, the resultant nucleoside 3'-O-[1,3,2-oxathiaphospholanes] were oxidized by means of elemental sulfur, resulting in nucleoside 3'-0[2-thio-1,3,2-oxathiaphospholanes] (2). Subsequently, compounds 2 were each separated into individual diastereoisomeric species by means of column chromatography on silica gel. Diastereoisomeric purity was assessed by means of

HPTLC, HPLC and 31P NMR. Physicochemical characteristics of the diastereoisomers of 2 are presented in Table 1. In the case of 5'-DMT-thymidine 3'-0[2-thio-1,3,2-oxathiaphospholane] the 'fast' eluted isomer 2 (B =Thy) was treated with p-toluenesulfonic acid for deprotection of the 5 '-hydroxyl function. The resulting thymidine 3 '-0[2-thio- 1 ,3,2-oxathiaphospholane], upon treatment with triethylamine, underwent intramolecular cyclization to afford thymidine cyclic (3', 5')-Spphosphorothioate, ([Sp]-cTMPS) (12). In the same manner, the 'slow' eluted diastereomer of 2 (B=Thy) gave [Rp]-cTMPS (12). The results of these interconversions proved that intramolecular opening of the oxathiaphospholane ring is stereospecific and kindled hopes that intermolecular ring opening in 2 followed by elimination of ethylene sulfide could also be stereospecific. Reaction of 'fast' eluted 2 (B=Thy) with thymidine 3 (bound to solid support via the 3'oxygen) in acetonitrile in the presence of triethylamine appeared to proceed slowly. After 0.5 h, the product obtained via deprotection and cleavage from the support was isolated in only 5% yield (HPLC assay), but appeared to be [Sp]-dithymidylyl (3', 5') phosphorothioate (TpST, 4, B=B'=Thy, R'=OH), by comparison with authentic material (13). In the search for other catalysts, including Nmethylimidazole, 4-dimethylaminopyridine (DMAP), and some others, 1,8-diazabicyclo[5.4.0] undec-7-ene (DBU) appeared to be the catayst of choice. The same reaction as above when performed in the presence of a 300-fold molar excess of DBU gave (HPLC assay) the crude [Sp]TpsT (4) in 95% yield. Diastereoisomeric purity (d.p.) of the resultant product was Table 1. Physicochemical characteristics of separated diastereoisomers of 2 B

B

DMTO

I

/

TETRAZOLE 2. S8

1

DMI

0

(eq.2)

/

ABZ ABZ GiBu GiBu CBZ

FAST SLOW FAST SLOW FAST SLOW FAST SLOW

CBz

p_ N(Pr')2

T T S

S

31P NMRa

Rf"

Yield%

103.23 103.18 104.52 104.17 104.31 104.16 104.27 104.23

0.34 0.31 0.22 0.20 0.27 0.22 0.59 0.57

90 85 89 92

H3P04 as an ext. reference acetate as a developing solvent with HP/TLC silica gel 60 (Merck)

ain C6H6,

bEthyl I

DMT

=

4,4 -dimethoxytrityl-; B

=

A9t, GLBU, (, T

Table 2. Stereospecificities of the formation of dinucleoside (3', 5') phosphorothioates (4) in reaction according to eq. 1, with DBU as the catalyst Substrates 2 T T

ABz

AB?

cBz CBZ CBZ CBZ

GIBu GiBu

Figure 1. Schematic representation of oligo(nucleoside phosphorothioate) (1); = number of repeating units, and B Ade, Gua, Cyt, Thy.

n

=

d.p.a

4C

Product d.p.(HPLC)

FAST SLOW FAST SLOW FAST FAST SLOW SLOW FAST SLOW

100% 100% 100% 100% 100% 95 %b 100% 95 %b 100% 100%

[Sp]-d(TpST) [Rp]-d(TpST) [Sp]-d(ApSA) [Rp]-d(ApSA) [Sp]-d(CpSC) [Sp]-d(CpSC) [Rp]-d(CpSC) [Rp]-d(CpSC) [Sp]-d(GpSG) [Rp]-d(GpsG)

99.0% 100.0% 99.4% 99.5% 99.3% 95.0% 99.5% 95.0% 99.3% 98.0%

ad.p. = Diastereisomeric purity bPrepared by mixing the separated diastereoisomers 'All diastereoisomeric 4 were identified via HPLC by coinjections with genuine samples prepared according to Stec,W.J., Zon,G., Egan,W. and Stec,B. (1984) J. Am. Chem. Soc. 106, 6077-6079.

Nucleic Acids Research, Vol. 19, No. 21 5885

>99%. In the case of 'slow' 2 (B = Thy), under identical conditions, [Rp]-d(TpsT) was obtained in 95.5% yield, d.p. 99%. Results for the preparation of other dinucleoside (3', 5") phosphorothioates (4) are summarized in the Table 2. This first set of experiments had proved that: (1) reaction of 2 with the 5'-hydroxyl function of an appropriately protected nucleoside in the presence of DBU effectively leads to formation of dinucleoside phosphorothioates, and (2) this process is fully stereospecific. Table 3. Chemical steps for one synthesis cycle Step

Reagent or Solvent

1. a) Dichloroacetic acid in (2:98, V/V) b) Acetonitrile 2. a) Activated nucleotide in

acetonitrilea b) Acetonitrile 3. a) Acetic anhydride/lutidine in THF (10:10:80, v/v) N-methylimidazole in THF (16:84, v/v) b) Acetonitrile

Time (min)

Purpose

CH2Cl2 DETRITYLATION 1.5 2 ml 5 ml

5 ml

WASH COUPLING 10 WASH CAPPING

2

WASH

1

2 1

1 ml 1 ml 5 ml

aFor 1 umol synthesis scale, 2 M DBU in pyridine (150 yd) and 0.1 M 5'-ODMT-deoxynucleoside 3'-O-[2-thio-1,3,2-oxathiaphospholane (50 ,ul) in acetonitrile is used.

Automated synthesis of oligo(deoxyribonucleoside phosphorothioates) (1) based on the aforementioned results required the solution of a few problems. The first one was posed by the lability of the conventional LCA-CPG linker to DBU (14). Fortunately, it was demonstrated by Gait (15) that a sarcosinederived linker originally proposed by Brown (16) is resistant against DBU. The set of T, CBZ, GiBu, ABZ - bound to CPG via a sarcosinyl linker was therefore prepared and used for further solid-support syntheses. The next two tasks were posed by the fact that, in the process of ring opening of 2 followed by ethylene sulfide elimination, an unprotected internucleotide phosphorothioate function is formed, and this moiety could potentially complicate the next cycle of oligomer elongation by either branch formulation, via an unsymmetrical pyrophosphorodithioate, or acylations during the capping step. This latter event had been exemplified in earlier observations that trifluoroacetic anhydride effectively oxidizes neutral phosphorothioates (17, 18). We have proved experimentally that phosphorothioylation of dialkyl phosphorothioate by means of 2/DBU does not occur. It was also demonstrated that treatment of dithymidylyl (3',5')-phosphorothioate with acetic anhydride/DMAP for 60 minutes does not lead to conversion of phosphorothioate-to-phosphate. With these results, an initial protocol for automated synthesis of oligo(deoxyribonucleoside phosphorothioates) has been elaborated (Table 3). According to this protocol, the following oligonucleotides were 3

2 2

2

3

3 I2

22

E

c0

0

a

.4;

co

.~

-1

I] b

c

Figure 2. HPLC profiles of dinucleoside(3'-5')phosphorothioates 1.- dA; a) d[ApSA]: 1.- dA; b) d[ApSA]: 1.- dC; c) d[CpSC]: 1.- dC; d) d[CpSC]:

i4

9,_ d

(crude products): 2.- [Rp]-d[ApsA]; 2.- [Rp]-d[ApSA];

e

f

L

9

3.- [Sp]-d[ApSA]; 2.- [Rp]-d[CpSC]; 3.- [Sp]-d[CpSC]; 2.- [RpI-d[CpSC]; 3.- [Sp]-d[CpSC]; 2.- [Rp]-d[GpSG]; 3.- [Sp]-d[GpSG]; 1.- dG; e) d[GpSG]: 1.- dG; 2.- [Rp]-d[GpSG]; 3.- [Sp]-d[GpSG]; f) d[GpSG]: 1.- dT; 2.- [Rp]-d[TpST]; g) d[TpST]: 2.- [Sp]-d[TpST]; 1.- dT; h) d[TpST]: HPLC analysis was performed on ODS-Hypersil (5 A) column with the linear gradient of acetonitrile: 5-20% CH3CN/0. 1 M TEAB (triethylammonium bicarbonate); 0.75%/min, 1.5 ml/min.

5886 Nucleic Acids Research, Vol. 19, No. 21 compared with reference samples of the corresponding oligonucleotides prepared by the conventional phosphoramidite approach using stepwise sulfurization during each reaction cycle (5) (data not presented). HPLC examination has shown the identity of oligo(nucleoside phosphorothioates) obtained according to these two different methodologies. NMR investigation (31P NMR) has shown that the content of phosphates is c ca. 1%. Independently, oligomers 6, 7 and 8 have been prepared using separated diastereoisomers of 2. The correctness of the structure of each stereoregular oligonucleotide phosphorothioate was checked by enzymatic and electrophoretic methods (Figure 4). The use of two nucleolytic enzymes (svPDE - EC.3. 1.4.1 and nuclease P1 - EC.3. 1.30. 1) with known diastereoselectivity towards dinucleoside phosphorothioates allowed us to confirm the diastereoisomeric purity of stereoregular oligo(nucleoside phosphorothioates). While svPDE recognizes and hydrolyzes only Rp-isomers of dinucleoside phosphorothioates (20, 21), nuclease P1 is known to hydrolyze their Sp-isomers (22). Compounds designated as [Rp]-6 and [Sp]-6 were incubated with svPDE and, independently, with nuclease P1. HPLC analysis performed under conditions described in Table 4 revealed complete resistance of [Rp]-6 to nuclease P1 and its complete degradation in the presence of svPDE. In agreement with these results, the Sp-isomer of 6 appeared to be resistant to svPDE, but hydrolyzed in the presence of nuclease P1. These results proved high (above 97%) diastereoisomeric purity of [Rp]-6 and [Sp]-6. The same method of analysis was applied to Rp- and Spisomers of d[(CPS)4C] which were synthesized using mutually cross-contaminated (5 %) diastereoisomers of 2 (B=CBZ). In the case of these pentanucleotides, enzymatic degradation and HPLC

obtained: d[(CPS)4C](6),d[(TpS)]7T] (7), d[GpsGpsApSApSTpST_

psCpSC](8), d[TpsCpsGpsTpsCpsGpsCpsTpsGpsTpsCpsTpsCpsCpsGpsCpsTpsTpsCpsTpsTpsCpsCpsTpsGpsCpsCpSA](HIV antirev) (9) (19). Each synthesis was first performed using unseparated diastereoisomers of 2 which led to a random mixture of diastereoisomers of target oligonucleotides, and these were Table 4. HPLC characteristics of oligo(deoxyribonucleoside phosphorothioates) Comp. No. 6 7

8

5'-HOb [min]

5'-DMTa Sequence

[m]in

[Rp]-d[(CpS)4C] [Sp]-d[(CPS)4C] d[(CPS)4C] -random mixture [Rp]-d[(TPS)7T] [Sp]-d[(TpS)7T] d[(TPS)7T] -random mixture

20.40 20.70 20.00 20.00 19.00-20.50c

11.80 12.00 10.40- 11.30c 16.75 18.00 16.50 - 17.80c

sCPsC]

17.00

16.00

17.40

16.30

17.00- 19.0(f

15.70- 16.50c

17.50

15.70

17.7018.40c

[Rp]-d[GpsGpsApsApsTpsTp_

[Sp]-d[GpSGpSApsApsTpSTp_ sCPsC]

d[GpsGpsApsApsTpsTpsCps 9

C] - random mixture ANTI REVd -random mixture

aODS-Hypersil (5 i) column (4.7 mm x 30 cm) flow rate 1.5 ml/min. 5-30% CH3CN-0.1 M TEAB, t=20 min (exponent 0.25)

bColumn and conditions as above; flow rate 1.5 ml/min.; linear gradient acetonitrile: 5-20% CH3CN-0.1 M TEAB; 0.75%/min. cBroad multiple signal resulted from content of a complex mixture diastereoisomers. d28-mer d[TpsCps etc.] as in text.

of of

(0)

LI

it'

11

1

!I

.j h

I II I ,

--, I-

---

90.0

80.0

70.0

60.0

50.0

--I

40.0

30.0

1--

20.0

--------

10.0

PPM Figure 3. 31P

NMR

spectrum of crude 9 (5'-DMT ON); 5

in D20, mM solution H3Pi4

as an external standard

0.0

-10.0

Nucleic Acids Research, Vol. 19, No. 21 5887

b

a

c

d

e

f g

h

Figure 4. PAGE analysis on 25% gel (7M, TBE). Lanes: a) 7- mixture of diastereoisomers (phosphoramidite method - stepwise sulfuration); b) [Rp]-7; c) [Sp]-7; d) 8-mixture of diastereoisomers (phosphoramidite method - stepwise sulfuration); e) [Rp]-8; f) [Sp]-8; g) 9 - HIV anti-rev - mixture of diastereoisomers (oxathiaphospholane method); h) 9- HIV anti-rev - mixture of diastereoisomers (phosphoramidite method - stepwise sulfurization).

analysis revealed the presence of the internucleotide bonds with the opposite configuration (ca. 10%). Similar analysis and results apply to Rp- and Sp-isomers of 7 (data not presented). While in the enzymatic studies of [Rp]-6/[Sp]-6, [Rp]-7/[Sp]-7 and [Rp]-8/[Sp]-8 only one of two enzymes was used (svPDE or nPl), enzymatic analyses of the random mixture of diastereomers of 9 required simultaneous use of svPDE and P1 followed by the use of alkaline phosphatase. In this case, the mixture of nucleosides A, G, C, T was obtained as the final product. In conclusion, we would like to emphasize that we have established a new and effective method for synthesis of oligo(nucleoside phosphorothioates), which is based on novel chemistry, and is both simple and cost-effective. We have proven that this method, when utilizing separated diastereoisomers of 2, is stereospecific and leads to stereoregular oligonucleotides with predetermined chirality at each phosphorothioate center. Due to the known antiviral activity of anti-rev (9), which so far has been prepared and studied as the random mixture of diastereoisomers, the synthesis of stereoregular [all-Rp]-9 and [all-Sp]-9 is in progress with the purpose of evaluation of the problem of structure-activity relationship. One has to realize, I

I

I

LO

0

E LA

UL

0CD

V; I ,_

2

CD O

r.:

m

2

0

CD

LA)

CD %O

Ln

('3

C.

cl;

2

Ij

LO

0

LA * 0

3

a

b

c

d

e

f

Figure 5. HPLC analysis of the digestion mixtures obtained after the treatment of [Rp]-6 and [Sp]-6 with snake venom phosphodiesterase (svPDE), nuclease P1 (nPl) and alkaline phosphatase (AP). a) [Rp]-d[(CpS)4C] treated with svPDE/AP: 1.-dC; 2.-[Rp]-d[CpSC] b) [Rp]-d[(CpS)4C] treated with nPl: 1.-undigested substrate; for comparison [Rp]-d[(CpS)4] synthesized from 2 (d.p. 95%) and treated with svPDE/AP: 1.-dC; 2.-[Rp]-d[CpSCI; 3.-[Sp]-d[CpsC]; 4.-[Sp]-d[(CpS)4C]; c) d) [Sp]-d[(CPS)4C] treated with nPl/AP: 1.-dC; e) [Sp]-d[(CpS)4C] treated with svPDE/AP: 1.- undigested substrate; f) for comparison [Sp]-d[(CpS)4C] synthesized from 2 (d.p. 95%) and treated with nP1/AP: 1.-dC; 2.-[Rp]-d[CpSC]; 3.-[Rp]-d[(CPS)4C).

5888 Nucleic Acids Research, Vol. 19, No. 21 however, that the synthesis of 28-mer with a step yield of 95 % and stereospecificity of single coupling 99%, leads to product prepared in 25% yield, which theoretically may contain only 76% of the single diastereomer having predetermined chirality at each phosphorus atom.

ACKNOWLEDGEMENTS Authors are indebted to Ms A.Kazmierkowska for her technical assistance in separation of diastereoisomers of 2 and to Mr A.Wilk for recording NMR spectra. Discussions with Drs A.Okruszek and P.Guga, and the help of Dr G.Zon, Ms D.Clarke and T.Rivas (ABI) in preparation of the manuscript are highly appreciated. This project was financially assisted (in part) by Grant CPBR-3.13.4.

REFERENCES 1. Eckstein,F. (1979) Acc. Chem. Res. 12, 204-208. 2. Matsukura,M., Shinozuka,K., Zon,G., Mitsuya,H., Reitz,M., Cohen,J.S. and Broder,S. (1987) Proc. Natl. Acad. Sci. USA 84, 7706-7710. 3. Agrawal,S., Goodchild,J., Civeira,M.R., Thornton,A., Sarin,P.S. and Zamecnik,P.C. (1988) Proc. Natl. Acad. Sci. USA 85, 7079-7083. 4. Bielinska,A., Shivdasani,R.A., Zhang,L. and Nabel,G.J. (1990) Science 250, 997-1000. 5. Zon,G. and Stec,W.J. (1991) In Eckstein,F. (ed.), Oligonucleotides and Analogues: A Practical Approach, Oxford University Press, Oxford, in press 6. Fujii,M., Ozaki,K., Kume,A., Sekine,M.C. and Hata,T. (1986) Tetrahedron Lett. 26, 935 -938. Fujii,M., Ozaki,K., Sekine,M.C. and Hata,T. (1987) Tetrahedron 43, 3395-3407. 7. Cosstick,R. and Williams,D.M. (1987) Nucleic Acids Res. 15, 9921-9932. 8. Lesnikowski,Z.J. and Jaworska,M. (1989) Tetrahedron Lent. 30, 3821-3824. 9. Guga,P. and Stec,W.J. (1983) Tetrahedron Lett. 24, 2899-3902. 10. Okruszek,A., Guga,P. and Stec,W.J. (1987) J. Chem. Soc. Chem. Commun. 594-595. 11. Willson,M., Concalved,H., Boudjegel,H. and Burgada,R. (1975) Bull. Soc. Chim. Fr. Part 2, 615-620. 12. Zielinski,W.S. and Stec,W.J. (1977) J. Am. Chem. Soc. 99, 8365-8366. 13. Uznanski,B., Niewiarowski,W. and Stec,W.J. (1982) Tetrahedron Lett. 23, 4289-4292. 14. Balgobin,N. and Chattopadhyaya,J. (1987) Nucleosides and Nucleotides 6, 461-464. 15. Lehmann,Ch., Xy,Y.Z., Christidoulou,Ch., Tan,Z.K. and Gait,M.J. (1989) Nucleic Acids Res. 17, 2379-2390. 16. Brown,T., Pritchard,C.E., Turner,G. and Salisbury,S.A. (1989) J. Chem. Soc. Chem. Commun., 891-893. 17. Bruzik,K.S. and Stec,W.J. (1990) J. Org. Chem. 55, 6131-6135. 18. Helinski,J., Skrzypczynski,Z., Wasiak,J. and Michalski,J. (1990) Tetrahedron Len. 31, 4081-4084. 19. Matsukura,M., Zon,G., Shinozuka,K., Robert-Guroff,M., Shimada,T., Stein,C.A., Mitsuya,M., Wong-Stahl,F., Cohen,J.S. and Broder,S. (1989) Proc. Natl. Acad. Sci. USA 86, 4244-4248. 20. Burgers,P.M.J. and Eckstein,F. (1978) Proc. Natl. Acad. Sci. USA 75, 4798-4801. 21. Bryant,F.R. and Benkovic,S.J. (1979) Biochemistry 18, 2825-2829. 22. Potter,B.V.L., Connolly,B.A. and Eckstein,F. (1983) Biochemistry 22, 1369-1377.