An unusual solvent effect on the regiochemical outcome (N-9 versus N-7) of guanine glycosylation using Robins' reagent (2-N-acetyl-6-O- diphenylcarbamoylguanine)

Article abstract of DOI:10.1016/S0040-4039(00)00423-8

An unexpectedly low N-9/N-7 regioselectivity was obtained when Robins reagent (2-N-acetyl-6-O-diphenylcarbamoylguanine) was coupled with a D- glucosamine derivative under trimethylsilyl trifluoromethanesulfonate activation. An unprecedented solvent effect (toluene versus dichloroethane) on the N-9/N-7 ratio was also observed in the same study. The use of 2-N- acetyl-6-O-benzylguanine to successfully overcome the above regioselectivity problem is described. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4039(00)00423-8

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 41 (2000) 3303–3307
An unusual solvent effect on the regiochemical outcome
(N-9 versus N-7) of guanine glycosylation using Robins’ reagent
(2-N-acetyl-6-O-diphenylcarbamoylguanine)
Adrian Wai-Hing Cheung,∗ Achyutharao Sidduri, Lisa M. Garofalo and
Robert A. Goodnow Jr.
Roche Research Center, Hoffmann-La Roche Inc., Nutley, New Jersey 07110, USA
Received 31 December 1999; accepted 3 March 2000
Abstract
An unexpectedly low N-9/N-7 regioselectivity was obtained when Robins’ reagent (2-N-acetyl-
6-O-diphenylcarbamoylguanine) was coupled with
a
D-glucosamine derivative under trimethylsilyl
trifluoromethanesulfonate activation. An unprecedented solvent effect (toluene versus dichloroethane) on
the N-9/N-7 ratio was also observed in the same study. The use of 2-N-acetyl-6-O-benzylguanine to successfully
overcome the above regioselectivity problem is described. © 2000 Elsevier Science Ltd. All rights reserved.
A novel series of glucosamine-based oligonucleotide analog 1 (GNA) was recently reported in our
laboratories (Fig. 1).1 These optically pure, conformationally constrained and amide-linked oligomers
were assembled from the nucleic acid building blocks (thymine 2a, cytosine 2b, adenine 2c and guanine
2d) on solid phase using N-Fmoc type peptide chemistry.2
In the synthesis of monomers 2a–2d, the key step is the coupling of D-glucosamine derivative 3
with modified nucleobases which works well for thymine 4a (80% yield), cytosine 4b (85% yield)
and adenine 4c (79% yield) (Fig. 2). However, the coupling of 3 with persilylated 2-N-acetylguanine
in acetonitrile gave a 2:1 mixture of the desired N-9 (4d, 36–43% yield) and N-7 isomers which required
chromatographic separation.1 Herein, we report our efforts to improve the regioselectivity and yield in
preparing guanine derivatives related to 4d.
The N-9 versus N-7 regioselectivity problem in guanine glycosylation is well documented.3 2-N-
Acetyl-6-O-diphenylcarbamoylguanine 5, introduced by Robins,4 has been used with much success in
the preparation of cyclic and acyclic nucleosides with excellent N-9 selectivity. We were surprised when
the coupling of 3 with persilylated 5 (bis(trimethylsilyl)trifluoroacetamide (BSA), ClCH2CH2Cl, 80°C)
under Robins’ conditions4 (trimethylsilyl trifluoromethanesulfonate (TMSOTf), toluene, 80°C, 2 h) gave
after chromatography 6% of N-9 isomer 6 and 14% of N-7 isomer 7 (Scheme 1). Significant amounts of
∗
Corresponding author. E-mail: adrian.cheung@roche.com (A. W.-H. Cheung)
0040-4039/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(00)00423-8
tetl 16703

3304
Fig. 1.
Fig. 2.
an inseparable mixture (∼20–30% yield) of compounds 8 and 9 (∼1:1 ratio) were also isolated from the
product mixture demonstrating the lability of the C-6 diphenylcarbamoyl group of 5 under the reaction
conditions. The low N-9/N-7 regioselectivity in the above coupling was in stark contrast to the excellent
N-9 regioselectivity of Robins’ reagent reported in the literature and in our own previous experience in
the ribose series. Even more surprisingly, when the TMSOTf-mediated coupling of 3 and persilylated 5
was carried out in dichloroethane instead of toluene, 45–55% yield of 7 was isolated without any trace of
6.5 This is, to the best of our knowledge, the only example in the literature involving the use of Robins’
reagent 5 in which a solvent switch led to a drastic change in the regiochemical outcome (N-9:N-7 ratio)
of guanine glycosylation.6–10
Scheme 1.
When N-7 isomer 7 was heated with TMSOTf in toluene at 80°C, no trace of N-9 isomer 6 could be
detected.11 Using tin(IV) chloride (SnCl4, 2 equiv., refluxing dichloroethane) in place of TMSOTf in the
coupling of 3 and 5 gave only N-7 isomer 9 (the C-6 diphenylcarbamoyl group was lost) in 40–50%

3305
yield.12 The N-9 assignment of 6 and N-7 assignment of 7 was established by analysis of 1H, 13C NMR
and UV spectra of the corresponding deprotected derivatives 10 and 11, respectively (Fig. 3). In addition,
an X-ray crystal structure of a derivative of 10 unambiguously confirmed its N-9 assignment.1
Fig. 3.
2-N-Acetyl-6-O-benzylguanine 12,13,14 in which the labile diphenylcarbamoyl group at C-6 of 5 is
replaced with a benzyl group, was prepared from 2-amino-6-chloropurine (step 1: NaH, BnOH, dioxane,
reflux; step 2: Ac2O, toluene, reflux; 72% over two steps). Compound 12 was persilylated and coupled
with 3 under TMSOTf activation in acetonitrile to give exclusively the desired N-9 isomer 13 in 30–40%
yield (Scheme 2). Dichloroethane was a suitable replacement for acetonitrile as solvent and about 35%
yield of 13 was isolated when the reaction was carried out in dichloroethane at 80°C. The glycosylation
reaction between 3 and 12 did not proceed at room temperature; running the reaction at 110°C (1,1,2,2-
tetrachloroethane as solvent) led to complete consumption of 3 but no trace of 13 was detected. The
N-9 linkage of 13 was confirmed by its exhaustive deprotection (step 1: H2, Pd/C, CH3OH; step 2:
Et3N, CH3OH, H2O; step 3: NH4OH, CH3OH, 60°C) to give 10 which was identical in all aspects to a
sample prepared from 6. Also, 13 was converted into 6 in two steps (step 1: H2, Pd/C, CH3OH; step 2:
Ph2NCOCl, (i-Pr)2EtN, pyridine) in 50% overall yield. Over 10 g of 13 were prepared in this manner and
were converted into guanine monomer 2d using similar procedures as previously described.1
Scheme 2.
It’s been reported by Imbach that only N-9 products (87% total yield) were obtained in the coupling
of peracetylated β-D-galactopyranose with Robins’ reagent 5.15 Since the reaction conditions used by
Imbach are identical to our conditions in the coupling of glucosamine derivative 3 with 5 (TMSOTf
catalysis in toluene at 80°C), this led us to postulate that the trifluoroacetamido group of 3 is likely
to be responsible for the failure of Robins’ reagent 5 to regioselectively give the N-9 isomer 6. This
dependence of the regiochemical outcome of guanine glycosylation (N-9 versus N-7) on the nature
of the sugar protecting group (2-O-acetyl versus 2-N-trifluoroacetamido) is precedented.16 While both
2-O-acetyl and 2-N-trifluoroacetamido groups could form the corresponding cyclic acyloxonium and
oxazolinium ions, respectively, N-trifluroacetamide group is believed to participate less strongly at C-
1.17 One possible scenario is that the coupling of 3 and 5 is irreversible (supported by the failure of 7 to
isomerize to 6 under TMSOTf activation, vide supra) and is under kinetic control. In dichloroethane, the
rate of N-7 isomer formation (giving 7) might be much higher than that of N-9 isomer formation (giving
6) and only 7 would be isolated.18 In toluene, the rate of N-9 isomer formation might become comparable
to that of N-7 isomer formation and a mixture of 6 and 7 would be formed (8 and 9 were formed through

3306
the loss of diphenylcarbamoyl group from 6 and 7).6b In the coupling of 3 and 12, N-9 isomer 13 might
be either the kinetic product (if the reaction is irreversible) or thermodynamic product (if the reaction is
reversible). In either case, 13 would be isolated as the exclusive product using either dichloroethane or
acetonitrile as solvent.
In summary, we encountered an unexpectedly low N-9/N-7 regioselectivity in the coupling of
Robins’ reagent (2-N-acetyl-6-O-diphenylcarbamoylguanine) with a D-glucosamine derivative and the
N-9:N-7 ratio was also strongly influenced by the solvent used in the glycosylation. 2-N-Acetyl-6-O-
benzylguanine was used successfully as a replacement of Robins’ reagent in the above glycosylation to
give selectively the desired N-9 product in either acetonitrile or dichloroethane in modest yield. Further
studies to probe the underlying cause of the above described solvent effect and investigation of the scope
in the use of 2-N-acetyl-6-O-benzylguanine are currently in progress and will be reported in due course.
Acknowledgements
The authors are grateful to Roche Physical Chemistry Department for spectroscopic measurements
and interpretations. Advice and encouragement from Drs. Christopher Exon, Masami Okabe, Ramakanth
Sarabu, Steve Tam and Steve Wolff are gratefully acknowledged.
References
1. Goodnow Jr., R. A.; Richou, A.-R.; Tam, S. Tetrahedron Lett. 1997, 38, 3195–3198.
2. Goodnow Jr., R. A.; Tam, S.; Pruess, D. L.; McComas, W. W. Tetrahedron Lett. 1997, 38, 3199–3202.
3. For leading references, see: (a) Robins, M. J.; Zou, R.; Hansske, F.; Madej, D.; Tyrrell. D. L. J. Nucleosides Nucleotides
1989, 8, 725–741. (b) Jenny, T. F.; Benner, S. A. Tetrahedron Lett. 1992, 33, 6619–6620. (c) Garner, P.; Ramakanth, S. J.
Org. Chem. 1988, 53, 1294–1298.
4. (a) Zou, R.; Robins, M. J. Can. J. Chem. 1987, 65, 1436–1437. (b) Robins, M. J.; Zou, R.; Guo, Z.; Wnuk, S. F. J. Org.
Chem. 1996, 61, 9207–9212.
5. Ref. 3a stated that “our procedure is very sensitive to moisture and has been found to proceed well only in anhydrous
toluene” but it was not further elaborated. Dichloroethane used in our studies was freshly distilled from calcium hydride
under argon.
6. (a) Ref. 3c stated that in glycosylation studies using 2-N-acetylguanine that “a variety of modifications involving solvent
composition, different catalysts, and nucleoside precomplexation were tried but did not increase the proportion of N-9
isomer significantly”. (b) For an example in which coupling with 2-N-acetylguanine gave a higher N-9:N-7 ratio when
a less polar solvent was used, see: Chamberlain, S. D.; Biron, K. K.; Dornsife, R. E.; Averett, D. R.; Beauchamp, L.;
Koszalka, G. W. J. Med. Chem. 1994, 37, 1371–1377.
7. We are aware of only two published examples involving persilylated 5 under TMSOTf activation (Robins’ conditions)
which did not use toluene as solvent. (a) Bednarski, K.; Dixit, D. M.; Mansour, T. S.; Colman, S. G.; Walcott, S. M.;
Ashman, C. Bioorg. Med. Chem. Lett. 1995, 5, 1741–1744. Methylene chloride was used as solvent, product yield was
25–30% and only N-9 regioisomers were formed. (b) Sells, T. B.; Nair, V. Tetrahedron 1994, 50, 117. Acetonitrile was
used as solvent, yield of N-9 product was 21% and it was not mentioned in the paper if any N-7 regioisomer was formed.
8. For an example involving the use of 5 in which trimethylsilyl perchlorate (TMSClO4) was used as activator and
dichloroethane was used as solvent, see: Herdewijn, P.; Van Aerschot, A.; Busson, R.; Claes, P.; De Clercq, E. Nucleosides
Nucleotides 1991, 10, 1525–1549. The yield of N-9 product was 55% and it was not mentioned in the paper if any N-7
regioisomer was formed.
9. For an example involving the use of 5 in which iodotrimethylsilane (TMSI) was used as activator and dichloroethane was
used as solvent, see: Tse, H. L. A.; Knight, D. J.; Coates, J. A. V.; Mansour, T. S. Bioorg. Med. Chem. Lett. 1997, 7,
1387–1392. The reaction regioselectively gave the N-9 product in 49% yield.
10. For an example involving the use of 5 in which the yield of N-9 glycosylation product depended strongly on the choice of
solvent (acetonitrile versus dichloroethane; 80% versus 40% yield), see: Janardhanam, S.; Nambiar, K. P. J. Chem. Soc.,
Chem. Commun. 1994, 1009–1010.

3307
11. See Ref. 4b for an example of a successful N-7 to N-9 isomerization.
12. For examples of selective N-9 guanine isomer formation using 5 activated by SnCl4, see: (a) Ref. 10. (b) Janardhanam, S.;
Nambiar, K. P. Tetrahedron Lett. 1994, 35, 3657–3660.
13. (a) Before our work described herein started, 12 was shown to couple with ribose derivatives under Robins’ conditions in
high yield and high N-9 selectivity. Li, W.-R.; Tam, S., Hoffmann-La Roche, unpublished results. (b) It was mentioned in
the footnotes of Ref. 4b that in the studies using 12 and other analogs, “difficulties were encountered during preliminary
investigations with preparation of and/or coupling” of these compounds.
14. After our work described herein was initiated, three reports concerning the use of 2-N-isobutyrl-6-O-benzylguanine in
glycosylation reactions were published. (a) Martin, P. Helv. Chim. Acta. 1995, 78, 486–504. (b) Martin, P. Helv. Chim.
Acta. 1996, 79, 1930–1938. (c) Martin, P. EP-626387, 1994.
15. El-Kattan, Y.; Gosselin, G.; Imbach, J.-L. J. Chem. Soc. Perkin Trans. 1 1994, 1289–1297.
We could not rule out the possibility that the change in chirality at the C-4 O-acetate group (glucose versus galactose) is
responsible/partially responsible for our observed anomalous behavior.
16. Ref. 3c.
17. (a) zu Reckendorf, W. M.; Wassiliadou-Micheli, N. Chem. Ber. 1970, 103, 1792–1796. (b) Wolfrom, M. L.; Conigliaro, P.
J. Carbo. Res. 1969, 11, 63–76.
18. For an example in which the N-7 isomer of a ribonucleoside was first formed which later on isomerized to give the N-9
isomer, see: Dudycz, L. W.; Wright, G. E. Nucleosides Nucleotides 1984, 3, 33–44.