Volume 7 Number 31979
Nucleic Acids Research
Fluoride ion promoted deprotection and transesterification in nucleotide triesters
Kelvin K.Ogilvie and Serge LBeaucage
Department of Chemistry, McGill University, Montreal, Quebec H3A 2K6, Canada
Received 16 April 1979
ABSTRACT
TetAabutyanmonium
Zuooide witt 'temove
tichto4oethiyt cyanoethyt A'oup6owm nuc2eotLde.. In addition to the deQiked nucteotZde ptodu.At.6 a eA tus inctading chain cteavage, pho6photuoAidate6 and thyumidne unit6 may be obtained depending on the condWtion46 cyanoethytated u6ed. FtLuoide i.on hah been used to 4ucceA66uUy exchange phenyt and tchtotoetihyt gkoup,6 6ot methyt, ethyt and butyt gwoup6 in nucteotLde tPie6te,. Thi- 'Lep,Le4entA a AapZd high yietd 'wute to a va'tiety o6 pho.6phate eAteAA. The ynQnthe46 o6 a novet nucteotide anatogue in which too chain4 ahLe bAidged ouugh theA phoi6p*ate6 i4 de6c.cbed. phkeny,
and
INTRODUCTION Since the successful adaptation of the triester method1 3 to the synthesis of oligonucleotides in solution it has become the most widely used method in this area. The general method has recently been used to prepare oligonucleotide sequences of biological importance.4 In addition nucleotide triesters have been used as probes of the structure5a and function of nucleic acids in biochemical experiments.5b,c It has therefore become of interest to develop procedures for the preparation of nucleotide triesters having a variety of groups as the third group at the phsophate. We have been exploring the technology of the triester method for some time. In this report we wish to describe in detail the use of tetrabutylannonium fluoride (TBAF) as a means of deprotecting phosphotriesters and also for the rapid production of triesters having a variety of groups on phosphate. We will also describe the synthesis of a novel nucleotide analogue where two strands are "bridged" through the phosphate esters of the two chains. Some of these results have appeared in preliminary form.6
C Information Retrieval Limited 1 Falconberg Court London Wl V 5FG England
805
Nucleic Acids Research DISCUSSION
One of the serious problems with the phenyl groups as phosphate protecting groups during nucleotide synthesis has been the removal of these groups at the end of the synthetic sequence (for a discussion of the problem see reference 7). The most serious aspect of the problem is base catalized chain-cleavage.7 With trichloroethyl groups deprotection occurs using various activities of zinc in a variety of solvents.8 While deprotection has generally appeared to be good to excellent in this area a deprotection procedure not involving metal ions was desired. We investigated the fluoride ion as a potential reagent for the deprotection of a variety of phosphate triesters. While fluoride ion rapidly deprotects phenyl, trichloroethyl and cyanoethyl triesters it often leads to chain cleavage to the extent of 5-15% in the case of dinucleotides. However the use of fluoride ion has allowed the synthesis of a number of nucleotide derivatives including mixed triesters of mononucleotides and nucl eoti de phosphofl uori dates.
Deprotection of Mononucleotide Triesters Using Fluoride Ion For the removal of phosphate protecting groups we investigated two our different solutions of fluoride ion. Reagent I was basically standard solution9 for the removal of alkylsilyl groups from nucleosides and nucleotides and consisted of 0.17M Bu4NF in anhydrous THF. Reagent II was similar to that employed by Narang10 and consisted of 0.17M Bu4NF in a solution of THF-pyridine-H20 (12:1:1). Reagent I requires only 30 min for complete reaction while Reagent II reactions are normally complete within 24 h. The two reagents often lead to different product distributions. In our early experiments with Reagent I we had not detected any internucleotide bond breaking. However some such cleavage actually occurs. The reason we missed it appears to be due to the fact that tetraalkylannonium salts have a marked influence on the chromatographic mobility of nucleotides. For example the direct application of a Reagent I solution containing a mixture of thymidine (10%), thymidine-3'-phosphate (Tp, 10%) and thymidylyl(3'-5')-thymidine (TpT, 80%) to Whatman 3 paper and development in solvent A often leads to no separation of components. However, if after evaporation of THF, the mixture is passed through a column of Dowex 50W-X8, the mixture in the eluant is cleanly separated on Whatman 3MM paper in solvent A.12 As will be shown below, fluoride ion treatment results in an average of nearly 10% chain cleavage at an internucleotide link. This observation of chain 806
Nucleic Acids Research cleavage caused by fluoride ion has been reported by others for a tetranucleotide.13 Such chain cleavage is also reported for the hydroxide ion hydrolysis of these same triesters.14 We have also verified the results and find that with amnonium hydroxidetnucleoside-phosphate cleavage occurs to the extent of 3.5% from la and 19% from lb. The mononucleotides studied are shown in Scheme 1 below and the results are collected in Table 1. It is interesting to note that for the bisphenyl esters (la,b) Reagent II leads to more phosphate-nucleoside cleavage than does Reagent I. Both reagents give excellent conversion (X80%) of la and lb into the fluorophosphate 2. This represents a very easy synthesis of the fluorophosphate which was identical to that prepared by the literature procedure15'16. In the case of the bistrichloroethyl esters lc and ld, yields of the fluorophosphate were around 35% from Reagent I and near 16% for Reagent II. However Reagent II produced reasonable (70%) yields of the monoester 3b. Both reagents produced around 9% of nucleoside-phosphate cleavage. Another interesting difference between Reagents I and II was the formation of 3',5'-cyclic phosphates (4) from lb and ld with Reagent I (10 and 5% respectively) while Reagent II gave no cyclic phosphate formation. Cyclic phosphate formation has also been implicated in the hydroxide catalyzed hydrolysis of lb and ld. The phosphate protecting groups can also be removed in t-butanol using cesium fluoride. Reactions are slow, usually requiring 30 h at room Table 1 Fluoride Ion Deprotection of Mononucleotide Triesters
Compound
MMT-Tp(¢)2 "1
Tp( ¢)2 if
MMT-Tp(TCE) 2 If
Tp(TCE)2 II
MWT-Tp($)2 W4T-Tr(TCE)2
Reagent I II I II I II I II CsF +BuOH CsF +BuOH
2 85 74 78 81 34 16 40 20 68 10
Products (%) 4 6 0 7 0 0 10 2.5 0 57 0 73 0 46 5 70 0 6 74
3(a orb)
5 9 19 12 16.5 9 11 9 10
807
Nucleic Acids Research
SCHEME 1
Th
RI
HO or
(b)
Reagent II
Th
0
Reagent I
(a)
HO+
oo
HOAc
0
0
O=P-OR'
O=P-OR'I
1H
1H
ORO R=MMT, Rl=¢
I (a) (b) R=H, R'=4 (c)
2
3 (a) R'+ (b) R[TCE
R=MMT, R'=TCE
(d) R=H, R'=TCE
HO
I0 0-
R-TPRI' -T-Si
1)
I
10r or II
OH 4
5
TpT (3'-5')+TpT (5'-5') 7
6
a,
R=MMT, R'=
b, R=H, R'=+ c," R=MMT, R'=TCE
d, R=H, R'=TCE e,
R=MtT, R'=CE
f, R=H, R'=CE
808
2)
HOAc
h
8
+
T 9
Nucleic Acids Research temperature for phenyl and 50 h for trichloroethyl groups. Using these conditions followed by acetic acid treatment and work-up, results in the conversion of la to 2 (68%) and 3a (6%). Compound lc leads to 2 (10%) and 3b
(74%). Deprotection of Dinucleotide Triesters Using Fluoride Ion Several dinucleotide triesters (Scheme 1) were treated with Reagents I and II to determine the degree of internucleotide chain cleavage. From Table 2 it can be seen that both reagents cause chain cleavage from phenyl and trichloroethyl triesters. Reagent II leads to slightly less chain cleavage (average 8%) than Reagent I (average 13%). Molecules having a free 5'hydroxyl group (6b and 6d) reacted with Reagent I to produce some rearranged dinucleotides having a 5'-5' linkage. Such rearrangements have also been observed in similar situations from hydroxide ion deprotection.17 This rearrangement does not occur with Reagent II. It is clear from these results and those reported by Reese13 that neither Reagent I nor Reagent II would be useful for the deprotection of oligonucleotide triesters bearing either a phenyl or trichloroethyl group at the phosphate.
Table 2 Fluorideelon Deprotection of Dinucleotide Triesters
Compound MMT-Tp+T-Si is
TpOT-Si II
MMT TpTCET-Si is
TpTCETSi "
MMT-TpCET-Si "1 TpCET Si "1
Reagent I II I II I II I II I II I II
7 87 92 62 93 88 92* 75 90* 64 100 48 100
Products (X) 8
24
10
Chainea 13 8 13 7 12 8 14 10
*yields are based on % of products; approximately 16% of the starting material remained intact at the end of 24 h. 809
Nucleic Acids Research With cyanoethyl triesters 6e and 6f, Reagent II gives a very clean deprotection to produce only the desired dinucleotide TpT (7). Thus the cyanoethyl triesters of oligonucleotides can be deprotected by fluoride ion in Reagent II. However Reagent I on the other hand gave a rather unusual result with compound 6f. The product (Scheme 2) consisted of the expected
0
Scheme 2
°I
HH
NCH2CH2CN
OMTO
HO
N
1) CH2=CH-CN 2) HOAc
OH
OH
9a
10
Tp(CE)T-Si
6f
1) TBAF 2) HOAc
TpT + TCE 7
11
+ TpTCE + TCE rCE 12
13
TpT (7, 48%) as well as three new nucleotide products. Two of these products (11 and 12) were dinucleotides of thymidine (total 40% of UV absorbing products) in which one of the thymidines possesses a cyanoethyl group on N-3. These compounds were not separable and their relative amounts were only determined by enzymatic degradation (snake venom gives T (9) from 12 and TCE (10) from 11; 9 and 10 are cleanly separable on chromatography). The third product was compound 13 which had a cyanoethyl group on each thymine unit. Apparently the acrylonitrile, which is produced when fluoride ion eliminates the cyanoethyl group from 6f, is able to alkylate the N-3 of the thymine residues in the nucleotide in the presence of fluoride ion. This alkylation occurs when thymidine derivatives are treated with acrylonitrile and TBAF. For example, 9ais converted to 3-cyanoethylthymidine (10) in 50% yield. This alkylation is quite similar to the base catalyzed addition of acrylonitrile to pyrimridines including pseudouridine which has been well documented.18,19 One additional feature that is of interest in the fluoride ion deprotections of nucleotides is shown by the following example. The compound Tp(OH)810
Nucleic Acids Research Tp(4)T was synthesized and treated with Reagent I. After 8 h the starting material was recovered intact. However, 0.2N NaOH (14 h) converted the compound to the free trinucleotide TpTpT. Clearly the presence of a charged phosphate one unit removed from the triester phosphate was sufficient to prevent removal of the phenyl group. Fluoride Ion Promoted Transesterification on Nucleotide Phosphotriesters. It was clear from the deprotection studies on mononucleotides that fluoride ion attacked the phosphate to produce phosphofluoridates and eliminate either phenoxide or trichloroethoxide. It seemed reasonable that these intermediate phosphofluoridates could be trapped in alcohol solvents resulting in a net exchange of phosphate protecting groups. This possibility was explored and the results on simple phosphates have been comnunicated.20 A preliminary report of a similar exchange in nucleotides has also been reported.6 The following discussion describes the detailed studies on mono and dinucleotides. The mechanism may involve intermediate phosphofluoridates or alternatively, in alcohol solution it is possible that fluoride ion through strong hydrogen bonding to the alcohols sufficiently increases the nucleophilicity of oxygen that the alcohol displaces the phenol or trichloroethanol groups directly. The phenyl group is much more easily exchanged than the trichloroethyl group on phosphate triesters. Thus compound 14, when dissolved in methanol, ethanol or n-butanol containing cesium fluoride (Scheme 3) is converted into SCHEME 3 CSF
R-Tp(R)2-2
R"OH
14 R-MMT, R'=4 15 R=DMT, R'=TCE R MMT-Tp T-Ac
_R or
R-Tp(R")2
16 R=MMT 17 R=H a) R"=CH3;
~~CSF C
R-Tp
R
18 R=DMT, R'=TCE 19 R=H, R'=TCE b) R"=Et; c) R"=Bu
R MMT-Tp T-Ac
ROH 20 a, R=4 b, R=TCE
21 a-c
the bisalkyl esters 16a-c in yields of 86%, 88% and 88% respectively. On the other hand with 15 one of the trichloroethyl groups is much more easily replaced than the second. Thus compounds 18a-c are obtained in yields of 71, 75 and 75% respectively when reaction conditions were either 50-530C for 811
Nucleic Acids Research 8-17 h or room temperature for 48 h. The fact that compounds 14 and 15 are so readily obtained means that through this exchange reaction a variety of mononucleotide triesters are available. Compounds 18 provide an easy route to mixed triesters or to a range of diesters since the trichloroethyl group can be removed without affecting the other alkyl groups. The exchange of phenyl triesters also occurs readily for dinucleotides. For example in 20, the phenyl group exchanges at room temperature to yield 21a-c in yields of 84%, 79% and 70%. Some loss of the acetyl group is also observed in these reactions and on average 7% of the deacylated derivatives of 20 were also obtained. The trichloroethyl group is much slower to exchange than phenyl as expected. In fact after 5 days at room temperature the degree of conversion of 20b to 21 is generally around 50%.
Synthesis of a Unique "Phosphate Bridged" Nucleotide. We felt that the phosphate exchange reaction provided a rare opportunity for the synthesis of novel nucleotide analogues. One particular goal was the synthesis of a compound such as 22. Such a structure possesses two nucleotide
R-TpR
T-R"
_
20 a, R=Tr, R'=¢, R"=Ac
R-T T-R"
(CH2)4
c, R=Tr, R'=-CH2CH2CH2CH2-OH, R"=Ac
R-TpT-R"
d, R=R"=H; R'=-CH2CH2CH2CH20H
22 a,
e, R=R"=TBDMS, R'=-CH2CH2CH2CH2OTBDMS
R=Tr,
R"=Ac
b, R=R"=H c,
R=R"=TBDMS
chains bridged via their phosphates. The synthesis of such a compound would serve as a model for the synthesis of other bridged nucleotides possessing complimentary strands. These structures would open the door to a number of interesting physical studies of these molecules as well as provide important structures for photochemical studies. The route to 22a was very simple. Compound 20a was dissolved in 1,4butanediol containing cesium fluoride and a 78% yield of 20c was obtained. This compound was deacetylated and detritylated to 20d which was fully characterized by 13C NMR (Table 3, including its tri TBDMS derivative 20e). Compound 20c was dissolved, along with a fourfold excess of 20a in a solution of t-butanol-DMF( 1:1) containing 10 equivalents of cesium fluorilde. After 812
Nucleic Acids Research 62 h a 30% yield of 22a was obtained. A molecular weight determination supported the assigned structure. In addition the compound was detritylated and deacetylated to 22b which was converted to its tetra TBDMS derivative 22c. Both 22b and 22c were characterized by 13C NMR. (Table 3, Figure 1) The data in Table 3 are consistent with the assigned structures. For simplicity the 5' nucleoside(s) at the 5'-end is (are) denoted as x and the other or 3'-end nucleoside(s) is (are) denoted as y. Thus the C2' carbons of 20(d and e) and 22(b and c) appear as a doublet and a singlet. Due to '3C-31P coupling it is clear that the doublet corresponds to C2'x while the singlet corresponds to C2' 2y When this synthesis is attempted except that 20a is reacted with ethylene glycol, the major product of the reaction is tritylthymidine. Presumably
Table 3 C
Chemical
Carbon 2 4 5 6 CH3 *1 'x
*1 'y 2'x 2'y 3'x
3'y *4'x
*4'y 5'x
5'y 1" 2"
3'" 411
Shift Data for Compounds 20d, 20e, 22b and 22c 20d(CD3OD) 152.2 166.2 111.9 137.9 12.6,12.5 86.6 86.1 39.5,39.4 40.4 80.4,80.2 71.7 87.2,86.9 86.1,85.8 62.2 70.0,69.8 68.9,68.7 28.1 ,27.8 29.5 62.5
22b(DMSO-d6) 150.6 163.9 109 .8g109 .9 135.8 12.2,12.0 84.1 83.8 **
**
78.6 ,78.3 70.0 85.5,85.2 84.5,84.1 61.0 67.3,67.1
±67.3967.1 26.1 ,25.8
20e(CDCI 3) 150.4 163.8 111.3 135.6 ,134.9 12.5 85.6,85.4 84.9 ,84.8 39.5,39.2 40.7 79.2,78.9 71.5 86.0,85.8 85.6 ,85.4 62.3 68.8,68.5 66.9,66.7 27.2,26.9 28.7 63.4
(6
values in
ppm)
22c(CDC1 3) 150.5,150.6,150.7 164.1 ,164.0,163.9 111.1,111.3 136.3,134.8 12.5 86.1 ,85.7 85.1 ,84.8 39.5,39.4 40.3 79.5,79.4 71.5 86.1 ,85.7 85.1 ,84.8 63.4 67.8,67 .5
Y'66.9 25.5,26.2
*Tentative assignments; **Obscured by solvent; tC-5'y and C-l" were superimposed; #Unresolved doublet. 813
Nucleic Acids Research Figure 1
(NH
HI~ R
R|
1
1a 2n 311 4a 0= -OCH2CH2CH2CH2OR
pi
bi
3
In
2
2
in
I
0- p -OCH2CH2CH2CHO-P=O
1H H H(N ~~~~~~~NH
H
0~~~~~~~~~~ 20d R=H . R=TSDMS
22 b c
R=H R=TBDMS
initial reaction occurs between 20a and ethylene glycol to produce the two carbon analogue of 20c. However such a compound apparently is not stable in the presence of fluoride ion. Presumably fluoride hydrogen bonds to the OH of the phosphate protecting group and it displaces a nucleoside froh the triester. A similar result has been observed by Smrt21 in the base catalysed decomposition of an analogous compound. GENERAL METHODS
Descending paper chromatography was carried out using Whatman 3 paper. The solvent systems employed were: solvent A, isopropyl alcohol-concentrated annonium hydroxide-water (7:1 :2); solvent B', n-butanol-ethanol-water (4:1:5, organic phase). The solvents were prepared on a volume basis. Thin-layer chromatography was carried out employing the ascending technique in closed jars which were not coated with absorbent paper. All thin-layer chromatography was run on Eastman chromagram sheets 13181 (formally 6060), silica gel with fluorescent indicator, or strips 10 cm x 2 cm. Thick-layer chromatography was carried out on glass plates (20 cm x 20 cm) coated with a 1-m thick layer of silica gel DSF-5 (Terrochem Laboratories). Paper electrophor814
Nucleic Acids Research esis was performed using Whatman 3MM paper in a Savant Flat Plate electrophoretic chamber will a Savant model HB power supply operated at 2000 V for 1 h; the solution was a triethylammonium bicarbonate buffer (0.05 M, pH 7.5), prepared by making 15.15 g triethylamine up to 31 volume with water and then bubbling 20 g carbon dioxide through the solution. Nucleosides and their derivatives were detected on paper chromatograms, thin- and thick-layer sheets using a uv light source (Mineralite, output "'254 nm). Compounds containing trityl or p-monomethoxytrityl groups were detected on chromatography by spraying with 10% perchloric acid solution and drying them in a stream of warm air. Imidazole was detected by placing the tlc in a jar containing iodine whereupon imidazole turns black very rapidly (within 3 min). All other general procedures and reagents were as previously described in detail.9 General Procedure for the Deprotection of Nucleotide Triesters by Reagent I and Reagent II A. Mononucleotides. The mononucleotide triester (la-d, 0.02 nmnole) was dissolved in Reagent I (0.18 ml TBAF soln + 0.5 ml THF) or Reagent II (0.18 ml of TBAF soln + 0.5 ml of THF:pyridine:water (8:1:1)). Reactions were stirred at room temperature for 30 min for Reagent I and 24 h for Reagent II. Water was added and the solvents removed at reduced pressure. For la and lc, the methoxytrityl group was removed at this stage with 80% HOAc at 850C for 30 min. Solvents were removed at reduced pressure and the residue was dissolved in water-ethanol (1:1) and passed through a column (lx35cm) of Dowex 50W-X8 (Na+ form) ion-exchange resin. The column was eluted with water (30-35 ml) and the total eluant was collected and concentrated at reduced pressure and applied Nucl eoto Whatman 3MM preparati ve sheets whi ch were devel oped i n Sol vent B' tide bands were el uted wi th water and yields determined spectrophotometri cal ly. Results are sumnarized in Table 1. Physical properties of nucleotide products are 1 i sted i n Tabl e 4. B. Dinucleotides. The conditions were similar to those in A except that 0.01 mmole of dinucleotide was used and products were isolated in solvent A. Results are recorded in Table 2. The results of Reagent I and MMT-TpCET-Si are discussed below. .
Deprotection of the Cyanoethyl Ester of Thymidylyl-(3'-5')-3 -t-Butyldimethylsi lyl thymidi ne (TpCET-Si) The protected compound TpCET-Si (6f) was treated with Reagent I under the standard conditions and was applied to Whatman preparative papers which were 815
Nucleic Acids Research Table 4 Chromatographic and Spectroscopic Data on Nucleotide Derivatives Compound*
A
TpF(g)
EtH (min) +ETp nm
f
TpTCE(X)
0.63 0.69 0.72
0.18 0.37 0.48
p
Nucleic Acids Research
Fluoride ion promoted deprotection and transesterification in nucleotide triesters
Kelvin K.Ogilvie and Serge LBeaucage
Department of Chemistry, McGill University, Montreal, Quebec H3A 2K6, Canada
Received 16 April 1979
ABSTRACT
TetAabutyanmonium
Zuooide witt 'temove
tichto4oethiyt cyanoethyt A'oup6owm nuc2eotLde.. In addition to the deQiked nucteotZde ptodu.At.6 a eA tus inctading chain cteavage, pho6photuoAidate6 and thyumidne unit6 may be obtained depending on the condWtion46 cyanoethytated u6ed. FtLuoide i.on hah been used to 4ucceA66uUy exchange phenyt and tchtotoetihyt gkoup,6 6ot methyt, ethyt and butyt gwoup6 in nucteotLde tPie6te,. Thi- 'Lep,Le4entA a AapZd high yietd 'wute to a va'tiety o6 pho.6phate eAteAA. The ynQnthe46 o6 a novet nucteotide anatogue in which too chain4 ahLe bAidged ouugh theA phoi6p*ate6 i4 de6c.cbed. phkeny,
and
INTRODUCTION Since the successful adaptation of the triester method1 3 to the synthesis of oligonucleotides in solution it has become the most widely used method in this area. The general method has recently been used to prepare oligonucleotide sequences of biological importance.4 In addition nucleotide triesters have been used as probes of the structure5a and function of nucleic acids in biochemical experiments.5b,c It has therefore become of interest to develop procedures for the preparation of nucleotide triesters having a variety of groups as the third group at the phsophate. We have been exploring the technology of the triester method for some time. In this report we wish to describe in detail the use of tetrabutylannonium fluoride (TBAF) as a means of deprotecting phosphotriesters and also for the rapid production of triesters having a variety of groups on phosphate. We will also describe the synthesis of a novel nucleotide analogue where two strands are "bridged" through the phosphate esters of the two chains. Some of these results have appeared in preliminary form.6
C Information Retrieval Limited 1 Falconberg Court London Wl V 5FG England
805
Nucleic Acids Research DISCUSSION
One of the serious problems with the phenyl groups as phosphate protecting groups during nucleotide synthesis has been the removal of these groups at the end of the synthetic sequence (for a discussion of the problem see reference 7). The most serious aspect of the problem is base catalized chain-cleavage.7 With trichloroethyl groups deprotection occurs using various activities of zinc in a variety of solvents.8 While deprotection has generally appeared to be good to excellent in this area a deprotection procedure not involving metal ions was desired. We investigated the fluoride ion as a potential reagent for the deprotection of a variety of phosphate triesters. While fluoride ion rapidly deprotects phenyl, trichloroethyl and cyanoethyl triesters it often leads to chain cleavage to the extent of 5-15% in the case of dinucleotides. However the use of fluoride ion has allowed the synthesis of a number of nucleotide derivatives including mixed triesters of mononucleotides and nucl eoti de phosphofl uori dates.
Deprotection of Mononucleotide Triesters Using Fluoride Ion For the removal of phosphate protecting groups we investigated two our different solutions of fluoride ion. Reagent I was basically standard solution9 for the removal of alkylsilyl groups from nucleosides and nucleotides and consisted of 0.17M Bu4NF in anhydrous THF. Reagent II was similar to that employed by Narang10 and consisted of 0.17M Bu4NF in a solution of THF-pyridine-H20 (12:1:1). Reagent I requires only 30 min for complete reaction while Reagent II reactions are normally complete within 24 h. The two reagents often lead to different product distributions. In our early experiments with Reagent I we had not detected any internucleotide bond breaking. However some such cleavage actually occurs. The reason we missed it appears to be due to the fact that tetraalkylannonium salts have a marked influence on the chromatographic mobility of nucleotides. For example the direct application of a Reagent I solution containing a mixture of thymidine (10%), thymidine-3'-phosphate (Tp, 10%) and thymidylyl(3'-5')-thymidine (TpT, 80%) to Whatman 3 paper and development in solvent A often leads to no separation of components. However, if after evaporation of THF, the mixture is passed through a column of Dowex 50W-X8, the mixture in the eluant is cleanly separated on Whatman 3MM paper in solvent A.12 As will be shown below, fluoride ion treatment results in an average of nearly 10% chain cleavage at an internucleotide link. This observation of chain 806
Nucleic Acids Research cleavage caused by fluoride ion has been reported by others for a tetranucleotide.13 Such chain cleavage is also reported for the hydroxide ion hydrolysis of these same triesters.14 We have also verified the results and find that with amnonium hydroxidetnucleoside-phosphate cleavage occurs to the extent of 3.5% from la and 19% from lb. The mononucleotides studied are shown in Scheme 1 below and the results are collected in Table 1. It is interesting to note that for the bisphenyl esters (la,b) Reagent II leads to more phosphate-nucleoside cleavage than does Reagent I. Both reagents give excellent conversion (X80%) of la and lb into the fluorophosphate 2. This represents a very easy synthesis of the fluorophosphate which was identical to that prepared by the literature procedure15'16. In the case of the bistrichloroethyl esters lc and ld, yields of the fluorophosphate were around 35% from Reagent I and near 16% for Reagent II. However Reagent II produced reasonable (70%) yields of the monoester 3b. Both reagents produced around 9% of nucleoside-phosphate cleavage. Another interesting difference between Reagents I and II was the formation of 3',5'-cyclic phosphates (4) from lb and ld with Reagent I (10 and 5% respectively) while Reagent II gave no cyclic phosphate formation. Cyclic phosphate formation has also been implicated in the hydroxide catalyzed hydrolysis of lb and ld. The phosphate protecting groups can also be removed in t-butanol using cesium fluoride. Reactions are slow, usually requiring 30 h at room Table 1 Fluoride Ion Deprotection of Mononucleotide Triesters
Compound
MMT-Tp(¢)2 "1
Tp( ¢)2 if
MMT-Tp(TCE) 2 If
Tp(TCE)2 II
MWT-Tp($)2 W4T-Tr(TCE)2
Reagent I II I II I II I II CsF +BuOH CsF +BuOH
2 85 74 78 81 34 16 40 20 68 10
Products (%) 4 6 0 7 0 0 10 2.5 0 57 0 73 0 46 5 70 0 6 74
3(a orb)
5 9 19 12 16.5 9 11 9 10
807
Nucleic Acids Research
SCHEME 1
Th
RI
HO or
(b)
Reagent II
Th
0
Reagent I
(a)
HO+
oo
HOAc
0
0
O=P-OR'
O=P-OR'I
1H
1H
ORO R=MMT, Rl=¢
I (a) (b) R=H, R'=4 (c)
2
3 (a) R'+ (b) R[TCE
R=MMT, R'=TCE
(d) R=H, R'=TCE
HO
I0 0-
R-TPRI' -T-Si
1)
I
10r or II
OH 4
5
TpT (3'-5')+TpT (5'-5') 7
6
a,
R=MMT, R'=
b, R=H, R'=+ c," R=MMT, R'=TCE
d, R=H, R'=TCE e,
R=MtT, R'=CE
f, R=H, R'=CE
808
2)
HOAc
h
8
+
T 9
Nucleic Acids Research temperature for phenyl and 50 h for trichloroethyl groups. Using these conditions followed by acetic acid treatment and work-up, results in the conversion of la to 2 (68%) and 3a (6%). Compound lc leads to 2 (10%) and 3b
(74%). Deprotection of Dinucleotide Triesters Using Fluoride Ion Several dinucleotide triesters (Scheme 1) were treated with Reagents I and II to determine the degree of internucleotide chain cleavage. From Table 2 it can be seen that both reagents cause chain cleavage from phenyl and trichloroethyl triesters. Reagent II leads to slightly less chain cleavage (average 8%) than Reagent I (average 13%). Molecules having a free 5'hydroxyl group (6b and 6d) reacted with Reagent I to produce some rearranged dinucleotides having a 5'-5' linkage. Such rearrangements have also been observed in similar situations from hydroxide ion deprotection.17 This rearrangement does not occur with Reagent II. It is clear from these results and those reported by Reese13 that neither Reagent I nor Reagent II would be useful for the deprotection of oligonucleotide triesters bearing either a phenyl or trichloroethyl group at the phosphate.
Table 2 Fluorideelon Deprotection of Dinucleotide Triesters
Compound MMT-Tp+T-Si is
TpOT-Si II
MMT TpTCET-Si is
TpTCETSi "
MMT-TpCET-Si "1 TpCET Si "1
Reagent I II I II I II I II I II I II
7 87 92 62 93 88 92* 75 90* 64 100 48 100
Products (X) 8
24
10
Chainea 13 8 13 7 12 8 14 10
*yields are based on % of products; approximately 16% of the starting material remained intact at the end of 24 h. 809
Nucleic Acids Research With cyanoethyl triesters 6e and 6f, Reagent II gives a very clean deprotection to produce only the desired dinucleotide TpT (7). Thus the cyanoethyl triesters of oligonucleotides can be deprotected by fluoride ion in Reagent II. However Reagent I on the other hand gave a rather unusual result with compound 6f. The product (Scheme 2) consisted of the expected
0
Scheme 2
°I
HH
NCH2CH2CN
OMTO
HO
N
1) CH2=CH-CN 2) HOAc
OH
OH
9a
10
Tp(CE)T-Si
6f
1) TBAF 2) HOAc
TpT + TCE 7
11
+ TpTCE + TCE rCE 12
13
TpT (7, 48%) as well as three new nucleotide products. Two of these products (11 and 12) were dinucleotides of thymidine (total 40% of UV absorbing products) in which one of the thymidines possesses a cyanoethyl group on N-3. These compounds were not separable and their relative amounts were only determined by enzymatic degradation (snake venom gives T (9) from 12 and TCE (10) from 11; 9 and 10 are cleanly separable on chromatography). The third product was compound 13 which had a cyanoethyl group on each thymine unit. Apparently the acrylonitrile, which is produced when fluoride ion eliminates the cyanoethyl group from 6f, is able to alkylate the N-3 of the thymine residues in the nucleotide in the presence of fluoride ion. This alkylation occurs when thymidine derivatives are treated with acrylonitrile and TBAF. For example, 9ais converted to 3-cyanoethylthymidine (10) in 50% yield. This alkylation is quite similar to the base catalyzed addition of acrylonitrile to pyrimridines including pseudouridine which has been well documented.18,19 One additional feature that is of interest in the fluoride ion deprotections of nucleotides is shown by the following example. The compound Tp(OH)810
Nucleic Acids Research Tp(4)T was synthesized and treated with Reagent I. After 8 h the starting material was recovered intact. However, 0.2N NaOH (14 h) converted the compound to the free trinucleotide TpTpT. Clearly the presence of a charged phosphate one unit removed from the triester phosphate was sufficient to prevent removal of the phenyl group. Fluoride Ion Promoted Transesterification on Nucleotide Phosphotriesters. It was clear from the deprotection studies on mononucleotides that fluoride ion attacked the phosphate to produce phosphofluoridates and eliminate either phenoxide or trichloroethoxide. It seemed reasonable that these intermediate phosphofluoridates could be trapped in alcohol solvents resulting in a net exchange of phosphate protecting groups. This possibility was explored and the results on simple phosphates have been comnunicated.20 A preliminary report of a similar exchange in nucleotides has also been reported.6 The following discussion describes the detailed studies on mono and dinucleotides. The mechanism may involve intermediate phosphofluoridates or alternatively, in alcohol solution it is possible that fluoride ion through strong hydrogen bonding to the alcohols sufficiently increases the nucleophilicity of oxygen that the alcohol displaces the phenol or trichloroethanol groups directly. The phenyl group is much more easily exchanged than the trichloroethyl group on phosphate triesters. Thus compound 14, when dissolved in methanol, ethanol or n-butanol containing cesium fluoride (Scheme 3) is converted into SCHEME 3 CSF
R-Tp(R)2-2
R"OH
14 R-MMT, R'=4 15 R=DMT, R'=TCE R MMT-Tp T-Ac
_R or
R-Tp(R")2
16 R=MMT 17 R=H a) R"=CH3;
~~CSF C
R-Tp
R
18 R=DMT, R'=TCE 19 R=H, R'=TCE b) R"=Et; c) R"=Bu
R MMT-Tp T-Ac
ROH 20 a, R=4 b, R=TCE
21 a-c
the bisalkyl esters 16a-c in yields of 86%, 88% and 88% respectively. On the other hand with 15 one of the trichloroethyl groups is much more easily replaced than the second. Thus compounds 18a-c are obtained in yields of 71, 75 and 75% respectively when reaction conditions were either 50-530C for 811
Nucleic Acids Research 8-17 h or room temperature for 48 h. The fact that compounds 14 and 15 are so readily obtained means that through this exchange reaction a variety of mononucleotide triesters are available. Compounds 18 provide an easy route to mixed triesters or to a range of diesters since the trichloroethyl group can be removed without affecting the other alkyl groups. The exchange of phenyl triesters also occurs readily for dinucleotides. For example in 20, the phenyl group exchanges at room temperature to yield 21a-c in yields of 84%, 79% and 70%. Some loss of the acetyl group is also observed in these reactions and on average 7% of the deacylated derivatives of 20 were also obtained. The trichloroethyl group is much slower to exchange than phenyl as expected. In fact after 5 days at room temperature the degree of conversion of 20b to 21 is generally around 50%.
Synthesis of a Unique "Phosphate Bridged" Nucleotide. We felt that the phosphate exchange reaction provided a rare opportunity for the synthesis of novel nucleotide analogues. One particular goal was the synthesis of a compound such as 22. Such a structure possesses two nucleotide
R-TpR
T-R"
_
20 a, R=Tr, R'=¢, R"=Ac
R-T T-R"
(CH2)4
c, R=Tr, R'=-CH2CH2CH2CH2-OH, R"=Ac
R-TpT-R"
d, R=R"=H; R'=-CH2CH2CH2CH20H
22 a,
e, R=R"=TBDMS, R'=-CH2CH2CH2CH2OTBDMS
R=Tr,
R"=Ac
b, R=R"=H c,
R=R"=TBDMS
chains bridged via their phosphates. The synthesis of such a compound would serve as a model for the synthesis of other bridged nucleotides possessing complimentary strands. These structures would open the door to a number of interesting physical studies of these molecules as well as provide important structures for photochemical studies. The route to 22a was very simple. Compound 20a was dissolved in 1,4butanediol containing cesium fluoride and a 78% yield of 20c was obtained. This compound was deacetylated and detritylated to 20d which was fully characterized by 13C NMR (Table 3, including its tri TBDMS derivative 20e). Compound 20c was dissolved, along with a fourfold excess of 20a in a solution of t-butanol-DMF( 1:1) containing 10 equivalents of cesium fluorilde. After 812
Nucleic Acids Research 62 h a 30% yield of 22a was obtained. A molecular weight determination supported the assigned structure. In addition the compound was detritylated and deacetylated to 22b which was converted to its tetra TBDMS derivative 22c. Both 22b and 22c were characterized by 13C NMR. (Table 3, Figure 1) The data in Table 3 are consistent with the assigned structures. For simplicity the 5' nucleoside(s) at the 5'-end is (are) denoted as x and the other or 3'-end nucleoside(s) is (are) denoted as y. Thus the C2' carbons of 20(d and e) and 22(b and c) appear as a doublet and a singlet. Due to '3C-31P coupling it is clear that the doublet corresponds to C2'x while the singlet corresponds to C2' 2y When this synthesis is attempted except that 20a is reacted with ethylene glycol, the major product of the reaction is tritylthymidine. Presumably
Table 3 C
Chemical
Carbon 2 4 5 6 CH3 *1 'x
*1 'y 2'x 2'y 3'x
3'y *4'x
*4'y 5'x
5'y 1" 2"
3'" 411
Shift Data for Compounds 20d, 20e, 22b and 22c 20d(CD3OD) 152.2 166.2 111.9 137.9 12.6,12.5 86.6 86.1 39.5,39.4 40.4 80.4,80.2 71.7 87.2,86.9 86.1,85.8 62.2 70.0,69.8 68.9,68.7 28.1 ,27.8 29.5 62.5
22b(DMSO-d6) 150.6 163.9 109 .8g109 .9 135.8 12.2,12.0 84.1 83.8 **
**
78.6 ,78.3 70.0 85.5,85.2 84.5,84.1 61.0 67.3,67.1
±67.3967.1 26.1 ,25.8
20e(CDCI 3) 150.4 163.8 111.3 135.6 ,134.9 12.5 85.6,85.4 84.9 ,84.8 39.5,39.2 40.7 79.2,78.9 71.5 86.0,85.8 85.6 ,85.4 62.3 68.8,68.5 66.9,66.7 27.2,26.9 28.7 63.4
(6
values in
ppm)
22c(CDC1 3) 150.5,150.6,150.7 164.1 ,164.0,163.9 111.1,111.3 136.3,134.8 12.5 86.1 ,85.7 85.1 ,84.8 39.5,39.4 40.3 79.5,79.4 71.5 86.1 ,85.7 85.1 ,84.8 63.4 67.8,67 .5
Y'66.9 25.5,26.2
*Tentative assignments; **Obscured by solvent; tC-5'y and C-l" were superimposed; #Unresolved doublet. 813
Nucleic Acids Research Figure 1
(NH
HI~ R
R|
1
1a 2n 311 4a 0= -OCH2CH2CH2CH2OR
pi
bi
3
In
2
2
in
I
0- p -OCH2CH2CH2CHO-P=O
1H H H(N ~~~~~~~NH
H
0~~~~~~~~~~ 20d R=H . R=TSDMS
22 b c
R=H R=TBDMS
initial reaction occurs between 20a and ethylene glycol to produce the two carbon analogue of 20c. However such a compound apparently is not stable in the presence of fluoride ion. Presumably fluoride hydrogen bonds to the OH of the phosphate protecting group and it displaces a nucleoside froh the triester. A similar result has been observed by Smrt21 in the base catalysed decomposition of an analogous compound. GENERAL METHODS
Descending paper chromatography was carried out using Whatman 3 paper. The solvent systems employed were: solvent A, isopropyl alcohol-concentrated annonium hydroxide-water (7:1 :2); solvent B', n-butanol-ethanol-water (4:1:5, organic phase). The solvents were prepared on a volume basis. Thin-layer chromatography was carried out employing the ascending technique in closed jars which were not coated with absorbent paper. All thin-layer chromatography was run on Eastman chromagram sheets 13181 (formally 6060), silica gel with fluorescent indicator, or strips 10 cm x 2 cm. Thick-layer chromatography was carried out on glass plates (20 cm x 20 cm) coated with a 1-m thick layer of silica gel DSF-5 (Terrochem Laboratories). Paper electrophor814
Nucleic Acids Research esis was performed using Whatman 3MM paper in a Savant Flat Plate electrophoretic chamber will a Savant model HB power supply operated at 2000 V for 1 h; the solution was a triethylammonium bicarbonate buffer (0.05 M, pH 7.5), prepared by making 15.15 g triethylamine up to 31 volume with water and then bubbling 20 g carbon dioxide through the solution. Nucleosides and their derivatives were detected on paper chromatograms, thin- and thick-layer sheets using a uv light source (Mineralite, output "'254 nm). Compounds containing trityl or p-monomethoxytrityl groups were detected on chromatography by spraying with 10% perchloric acid solution and drying them in a stream of warm air. Imidazole was detected by placing the tlc in a jar containing iodine whereupon imidazole turns black very rapidly (within 3 min). All other general procedures and reagents were as previously described in detail.9 General Procedure for the Deprotection of Nucleotide Triesters by Reagent I and Reagent II A. Mononucleotides. The mononucleotide triester (la-d, 0.02 nmnole) was dissolved in Reagent I (0.18 ml TBAF soln + 0.5 ml THF) or Reagent II (0.18 ml of TBAF soln + 0.5 ml of THF:pyridine:water (8:1:1)). Reactions were stirred at room temperature for 30 min for Reagent I and 24 h for Reagent II. Water was added and the solvents removed at reduced pressure. For la and lc, the methoxytrityl group was removed at this stage with 80% HOAc at 850C for 30 min. Solvents were removed at reduced pressure and the residue was dissolved in water-ethanol (1:1) and passed through a column (lx35cm) of Dowex 50W-X8 (Na+ form) ion-exchange resin. The column was eluted with water (30-35 ml) and the total eluant was collected and concentrated at reduced pressure and applied Nucl eoto Whatman 3MM preparati ve sheets whi ch were devel oped i n Sol vent B' tide bands were el uted wi th water and yields determined spectrophotometri cal ly. Results are sumnarized in Table 1. Physical properties of nucleotide products are 1 i sted i n Tabl e 4. B. Dinucleotides. The conditions were similar to those in A except that 0.01 mmole of dinucleotide was used and products were isolated in solvent A. Results are recorded in Table 2. The results of Reagent I and MMT-TpCET-Si are discussed below. .
Deprotection of the Cyanoethyl Ester of Thymidylyl-(3'-5')-3 -t-Butyldimethylsi lyl thymidi ne (TpCET-Si) The protected compound TpCET-Si (6f) was treated with Reagent I under the standard conditions and was applied to Whatman preparative papers which were 815
Nucleic Acids Research Table 4 Chromatographic and Spectroscopic Data on Nucleotide Derivatives Compound*
A
TpF(g)
EtH (min) +ETp nm
f
TpTCE(X)
0.63 0.69 0.72
0.18 0.37 0.48
p
