Facilitation of addition-elimination reactions in pyrimidines and purines using trifluoroacetic acid in trifluoroethanol

Article abstract of DOI:10.1039/b308948g

S_NAr displacement reactions of 6-cyclohexylmethoxy-2- fluoropurine, 6-amino-2-butylsulfonyl-4-cyclohexylme-thoxypyrimidine and 2-amino-6-chloropurine with substituted anilines (e.g. the weakly nucleophilic 4-aminobenzenesulfonamide) are dramati

Full text of DOI:10.1039/b308948g

Facilitation of addition–elimination reactions in pyrimidines and purines
using trifluoroacetic acid in trifluoroethanol
Hayley J. Whitfield†, Roger J. Griffin, Ian R. Hardcastle, Andrew Henderson, Jerome Meneyrol,
Veronique Mesguiche, Kerry L. Sayle and Bernard T. Golding*
School of Natural Sciences – Chemistry, University of Newcastle upon Tyne, Bedson Building, Newcastle
upon Tyne, UK NE1 7RU
Received (in Cambridge, UK) 30th July 2003, Accepted 23rd September 2003
First published as an Advance Article on the web 14th October 2003
SNAr displacement reactions of 6-cyclohexylmethoxy-
2-fluoropurine, 6-amino-2-butylsulfonyl-4-cyclohexylme-
thoxypyrimidine and 2-amino-6-chloropurine with substi-
tuted
anilines
(e.g.
the
weakly
nucleophilic
4-aminobenzenesulfonamide) are dramatically accelerated
in the presence of trifluoroacetic acid and occur especially
efficiently in 2,2,2-trifluoroethanol solvent.
The classical method for introducing amino-substituents at the
2/4-positions of pyrimidines and the 2/6-positions of purines is
to react a precursor heterocycle containing a suitable leaving
group with an amine.1,2 This type of reaction occurs by an
addition–elimination (SNAr) mechanism, i.e. formation of a
tetrahedral intermediate that eliminates the leaving group
(Scheme 1).3,4 Although this method has been partly superseded
by the use of palladium-mediated substitutions,5,6 the direct
substitution method is still the mainstay of many syntheses of
substituted heterocycles. However, the reaction often fails with
poorly nucleophilic amines, e.g. anilines bearing electron-
withdrawing groups. In a programme seeking potent inhibitors
of cyclin-dependent kinases,7 we required purines 1 and
pyrimidines 2 substituted at C-2 with anilino groups.8 Of
particular interest were anilines with electron-withdrawing
substituents at C-4 (e.g. SO2NHR and CONHR) as these are
associated with high potency against CDK1 and CDK2.9 To
access the purines, the fluoro-substituted precursor 3 was
chosen, whereas for the pyrimidines the n-butylsulfonyl
precursor 4 was selected. We have found that the combination
of trifluoroacetic acid (TFA) as catalyst and 2,2,2-trifluoro-
ethanol (TFE) as solvent provides a remarkably effective way to
achieve nucleophilic substitutions in the substrates described,
even with anilines of low nucleophilicity.
In initial studies with 6-cyclohexylmethoxy-2-fluoro-9H-
purine 3, prepared by a Balz–Schiemann reaction on 2-amino-
6-cyclohexylmethoxy-9H-purine, reactions were performed
with a substituted aniline (e.g. 3-chloroaniline) in refluxing
butan-1-ol as solvent. In addition to product 1b, a significant by-
product from this reaction was 2-butoxy-6-cyclohexylmethoxy-
purine 5a. It was observed that the desired substitution reaction
appeared to accelerate at longer reaction times, and this was
ascribed to a catalytic effect of the hydrogen fluoride formed.
Substitution reactions of halopyridines with amines are known
to be accelerated by acidic catalysts [e.g. copper(II) sulfate].1a,2
A systematic study was therefore undertaken in which different
acids were added to the reaction mixture. It was found that for
a 0.2 M solution of substrate 3 in butan-1-ol at 120 °C
containing an excess of an aniline or a substituted aniline (e.g.
3-F, 3-Cl, 3-Br or 3-MeO; up to 7 mole equiv.), with and
without TFA, the formation of the product (1a–e) was
substantially accelerated in the presence of acid (0.4, 5 or 10
mole equiv. TFA). The results described were rationalised by
postulating that the presence of TFA generates both mono-
protonated 3 and protonated aniline at equilibrium with their
non-protonated precursors. Reaction of protonated 3 with non-
protonated aniline leads to a tetrahedral intermediate that loses
first a proton and then fluoride to give product 1b (Scheme 1).
The loss of fluoride requires that this anion is solvated, and this
can be achieved either by the use of a protic solvent or by an
added acid. In support of the proposed mechanism, increasing
the amount of aniline initially increased but eventually
decreased (at 7 mole equiv.) product yields, presumably by
buffering the acid. Weaker acids (e.g. acetic acid) were
ineffective as a catalyst.
Applying the principles developed with the anilines de-
scribed above to less nucleophilic anilines gave relatively poor
product yields, e.g. 27% of 6-cyclohexylmethoxy-2-(4-sulfa-
moylanilino)-9H-purine 1f from a reaction in which compound
3 (1 M in a 1 : 1 ethanol–glycerol mixture containing 1 mole
equiv. TFA) and 7 mole equiv. 4-aminobenzenesulfonamide
(sulfanilamide) was heated at reflux for 7 days. It was surmised
that performing the reaction in the more polar protic solvent
2,2,2-trifluoroethanol (TFE) would favour formation of the
charged tetrahedral intermediate (Scheme 1) and facilitate loss
of fluoride from this intermediate. Also, the formation of the by-
product 2-alkoxy-6-cyclohexylmethoxypurine should be sup-
pressed by the use of an alcohol of lower nucleophilicity.
Gratifyingly, the use of TFE with TFA provided optimum
conditions for achieving reactions of anilines with 3, even for
very weakly nucleophilic anilines. A further advantage of the
use of TFE was that all reactions were homogeneous in this
solvent. A series of experiments were performed for the reaction
of 3 with sulfanilamide in which the solvent and concentration
of TFA were varied. For a 0.1 M solution of 3 in TFE for 6 hours
at reflux, yields of 1f were 17 (0), 74 (0.5), 79 (1), 83 (2), 89 (3)
and 91% (5) [the numbers in brackets are mole equiv. TFA].
These data confirmed that TFE–TFA is the medium/acid of
choice and provided a reliable and efficient experimental
procedure.‡ Under these conditions, no product from displace-
ment of fluoride from 3 by TFE was detected, and we did not
observe N-trifluoroacetylation of the aniline, which was another
significant side-reaction for reactions performed in butan-1-ol.
Other solvents explored were propan-2-ol and pentan-2-ol,
Scheme 1 Reaction of fluoropurine 3 with 3-chloroaniline catalysed by
TFA [n.b. 3 protonated at N-1 and N-9 will also be present, with 3-H+ (N-1)
also able to react with 3-chloroaniline].
† This paper is dedicated to the memory of Hayley Whitfield
(1977–2001).
2802
CHEM. COMMUN., 2003, 2802–2803
This journal is © The Royal Society of Chemistry 2003

View Article Online
which reduced the yield of by-product 2-alkoxy-6-cyclohex-
ylmethoxypurine, but only gave ca. 60% 1f under conditions
where the use of TFE gave > 90% yield. Aprotic solvents and
mixtures of aprotic and protic solvents generally gave sig-
nificantly reduced yields of 1f, with the exception of acetonitrile
which gave comparable yields but with by-products.
Notes and references
‡
Typical procedures:
To a suspension of 6-cyclohexylmethoxy-2-fluoro-9H-purine (0.41 g,
1.64 mmol) and sulfanilamide (0.55 g, 3.19 mmol) in trifluoroethanol (4
cm3) was added trifluoroacetic acid (0.61 cm3, 8 mmol). The resulting
solution was boiled under reflux for 16 h, when LC-MS analysis showed
complete conversion of 6-cyclohexylmethoxy-2-fluoro-9H-purine to 1f.
The solvent was removed and the residue was taken up in methanol. The
methanolic solution was filtered through a bed of basic alumina. The filtrate
was concentrated and the residual solid was recrystallised from propan-2-ol
to afford compound 1f as an analytically pure white solid (69%).
To a suspension of 6-amino-2-n-butylsulfonyl-4-cyclohexylmethoxypyr-
imidine 4 (0.20 g, 0.61 mmol) and 4-aminobenzamide (0.17 g, 1.22 mmol)
in trifluoroethanol (4 cm3) was added trifluoroacetic acid (0.24 cm3, 3.05
mmol). The resulting solution was boiled under reflux for 2 h. The solvent
was removed and the solid was extracted into ethyl acetate and washed with
water. The organic layer was collected, dried (Na2SO4) and filtered. The
solvent was removed to yield a white solid. Recrystallisation from MeOH
yielded 2a as a white solid (0.13 g, 0.38 mmol, 62%).
We also found that Amberlyst 15 (polymer-supported
sulfonic acid), camphorsulfonic, hydrochloric, chloroacetic and
formic acid were not useful catalysts at comparable concentra-
tions to TFA. This suggests that for optimal catalysis the acid
strength should be sufficient to generate a viable concentration
of protonated 3 (Scheme 1), but not so strong that the amine
reaction partner is almost completely protonated. When benzy-
lamine was used as nucleophile with 3 in trifluoroethanol or
dimethyl sulfoxide, the formation of product 5b occurred
readily in the absence of TFA. However, addition of 5 mole
equiv. TFA to this reaction dramatically suppressed the
formation of 5b (none detectable after heating at reflux for 1
hour, under which conditions formation of 5b was nearly
quantitative in the absence of the acid). In this case, addition of
TFA clearly completely protonates and deactivates the more
basic benzylamine.
All new compounds gave spectroscopic data and elemental analyses in
accord with their assigned structure.
Similar results were obtained for the substrate 6-amino-2-n-
butylsulfonyl-4-cyclohexylmethoxypyrimidine
410
and
1 (a) B. C. Uff, in Comprehensive Heterocyclic Chemistry, ed. A. J.
Boulton and A. McKillop, Pergamon Press, Oxford, 1984, vol. 2, p. 360;
(b) B. C. Uff, in Comprehensive Heterocyclic Chemistry, ed. A. J.
Boulton and A. McKillop, Pergamon Press, Oxford, 1984, vol. 5, p.
561.
2 H. E. Mertel, in The Chemistry of Heterocyclic Compounds, ed. A
Weissburger, Interscience, New York, 1962, Pt. 2, p. 352.
3 E. Y. Sutcliffe and R. K. Robins, J. Org. Chem., 1963, 28, 1662.
4 G. B. Barlin and N. B. Chapman, J. Chem. Soc., 1965, 3017.
5 C. Enguehard, H. Allouchi, A. Gueiffier and S. L. Buchwald, J. Org.
Chem., 2003, 68, 4367 and references cited therein.
6 S. Ding, N. S. Gray, X. Wu, Q. Ding and P. G. Schultz, J. Am. Chem.
Soc., 2002, 124, 1594.
7 M. Knockaert, P. Greengard and L. Meijer, Trends Pharmacol. Sci.,
2002, 23, 417; Y.-T. Chang, N. S. Gray, G. R. Rosania, D. P. Sutherlin,
S. Kwon, T. C. Norman, R. Sarohia, M. Leost, L. Meijer and P. G.
Schultz, Chem. Biol., 1999, 6, 361.
8 I. R. Hardcastle, R. J. Griffin and B. T. Golding, Ann. Rev. Pharm.
Toxicol., 2002, 42, 325; C. Arris, F. T. Boyle, A. H. Calvert, N. J. Curtin,
P. Jewsbury, J. A. Endicott, A. E. Gibson, B. T. Golding, R. J Griffin,
L. N. Johnson, A. Laurie, N. K. Lembicz, D. R. Newell, M. E. M. Noble,
E. Sausville and R. Schultz, J. Med. Chem., 2000, 43, 2797–2804.
9 T. G. Davies, J. Bentley, C. E. Arris, F. T. Boyle, N. J. Curtin, J. A.
Endicott, A. E. Gibson, B. T. Golding, R. J. Griffin, I. R. Hardcastle, P.
Jewsbury, L. N. Johnson, V. Mesguiche, D. R. Newell, M. E. M. Noble,
J. A. Tucker, L. Wang and H. J. Whitfield, Nat. Struct. Biol., 2002, 9,
745.
2-amino-6-chloropurine, showing that the leaving groups n-
butylsulfonyl and chloride are also compatible with the TFE–
TFA milieu. Thus, heating compound 4 (0.15 M in TFE) at
reflux for 2 hours with 2 mole equiv. 4-aminobenzamide in the
presence of 5 mole equiv. TFA gave 4-(6-amino-4-cyclohex-
ylmethoxypyrimidin-2-ylamino)benzamide 2a (62% isolated
yield of recrystallised, analytically pure material). In a similar
manner, 4 with the appropriate aniline gave the corresponding
adduct 2b–g (yields of analytically pure product in the range
45–71%). 2-Amino-6-chloropurine in TFE did not react with
sulfanilamide in the absence of TFA, but on addition of 5 mole
equiv. of the acid, 77% of 2-amino-6-(4-sulfamoylanilino)-9H-
purine 6 was obtained after heating at reflux for 3 hours.
The reaction conditions described should be suitable for a
variety of nucleophilic substitutions in heterocyclic systems,
especially for weakly nucleophilic amines and for leaving
groups that require effective solvation in order to decay to
product from the tetrahedral intermediate. We have used this
methodology to prepare numerous purines and pyrimidines
related to structures 1 and 2 from a variety of mono- and di-
substituted anilines (cf. ref. 10). With strongly nucleophilic
amines (e.g. benzylamines), addition of TFA is detrimental
because the amine is fully protonated and therefore deacti-
vated.
10 K. L. Sayle, J. Bentley, F. T. Boyle, A. H. Calvert, Y. Cheng, N. J.
Curtin, J. A. Endicott, B. T. Golding, I. R. Hardcastle, P. Jewsbury, V.
Mesguiche, D. R. Newell, M. E. M. Noble, R. J. Parsons, D. J. Pratt, L.
Z. Wang and R. J. Griffin, Bioorg. Med. Chem. Lett., 2003, 13, 3079.
We thank Cancer Research UK (studentship to HJW),
BBSRC (AH), EPSRC (KLS), MRC and AstraZeneca Pharma-
ceuticals for support of this research.
CHEM. COMMUN., 2003, 2802–2803
2803