Synthesis of enantiomerically pure D-FDOC, an anti-HIV agent

Article abstract of DOI:10.1016/j.bmcl.2004.07.016

The d-enantiomer of FDOC was obtained in optically pure form via a tandem kinetic resolution/chiral salt crystallization protocol. In addition, conditions were developed that allowed the unwanted l-enantiomer to be racemized and recycled. The β-d-enantiomer of FDOC (2′,3′-dideoxy-5-fluoro- oxacytidine) exhibits potent anti-HIV-1 activity. It was obtained in optically pure form by employing a tandem kinetic resolution/chiral salt crystallization protocol. In addition, conditions were developed that allowed the unwanted butyrate ester of the l-enantiomer of FDOC to be racemized. This material could then be recycled in future resolutions.

Full text of DOI:10.1016/j.bmcl.2004.07.016

Bioorganic & Medicinal Chemistry Letters 14 (2004) 4991–4994
Synthesis of enantiomerically pure D-FDOC, an anti-HIV agent
Shuli Mao,a Martin Bouygues,a Christopher Welch,c Mirlinda Biba,c Jen Chilenski,c
Raymond F. Schinazib and Dennis C. Liottaa,
*
aDepartment of Chemistry, Emory University, 403 Cherry Logan Emerson Hall, 1521 Dickey Drive, Atlanta, GA 30322, USA
bDepartment of Pediatrics, Veterans Affairs Medical Center, Emory University School of Medicine, Decatur, GA 30033, USA
cMerck Research Laboratories, Division of Merck & Co., Inc., P.O. Box 2000, Rahway, NJ 07065, USA
Received 22 November 2002; accepted 7 July 2004
Available online 29 July 2004
Abstract—The b-D-enantiomer of FDOC (20,30-dideoxy-5-fluoro-oxacytidine) exhibits potent anti-HIV-1 activity. It was obtained in
optically pure form by employing a tandem kinetic resolution/chiral salt crystallization protocol. In addition, conditions were devel-
oped that allowed the unwanted butyrate ester of the L-enantiomer of FDOC to be racemized. This material could then be recycled
in future resolutions.
ꢀ 2004 Elsevier Ltd. All rights reserved.
The potential of 10,30-dioxolanyl nucleosides as anti-
HIV-1 drugs has been recognized for some time. Over
a decade ago, b-D-dioxolane-T was synthesized and
shown to possess modest anti-HIV activity in vitro with-
out acute cellular toxicity.1–3 In the mid 1990s, (ꢀ)-b-D-
2,6-diaminopurine dioxolane (DAPD), a water soluble
prodrug of (ꢀ)-b-D-dioxolane guanine (DXG), was
found to exhibit potent activity against HIV-1 and is
now in Phase II clinical trials.4–6 In the same time frame,
both enantiomers of 20,30-dideoxy-5-fluoro-oxacytidine
(FDOC) were synthesized by us. Although both enanti-
omers exhibited good potency against HIV, they both
appeared to be too toxic to be used clinically.7,8
as to whether the observed toxicity was inherent to D-
FDOC or resulted from the presence of small quantities
of its more toxic enantiomer. To answer this question,
we used preparative chiral chromatography to obtain
several grams of optically pure D- and L-FDOC, respec-
tively.9 With these two enantiomers in hand, we could,
for the first time, unambiguously evaluate their anti-
HIV-1 activity, cytotoxicity and resistance profile. The
results of these studies indicated that D-FDOC not only
showed excellent potency (EC50 and EC90 values in pri-
mary human lymphocytes infected with HIV-1LAI are
0.04 and 0.26lM, respectively) and low toxicity
(>100lM in uninfected primary human lymphocytes),
but also exhibited no cross resistance to 3TC, AZT or
nevirapine. In addition, molecular infectious clones con-
taining M184V or M41L/D67N/K70R/T215Y/K219Q
were as susceptible as wild type virus (WT-pNL4-3) to
D-FDOC. In primary mouse bone marrow cells, D-
FDOC showed no increase in lactic acid production
even at 300lM. In contrast, treatment with either L-
FDOC or ddC resulted in a >300-fold increase in lactic
acid production relative to untreated control.10 Further-
more, in HepG2 cells (5day assay), D-FDOC displayed
no toxicity when tested up to 100lM, whereas its
L-counterpart demonstrated significant toxicity at
1.4lM. While these data, taken in aggregate, clearly
indicate that our original toxicity determinations were
compromised by the presence of small quantities of
the toxic L-enantiomer, they also suggest that clinical
evaluations of D-FDOC should only be performed
with materials that contain very little, if any, of the
NH
O
N
NH
2
2
F
N
N
NH
N
N
HO
HO
HO
O
N
O
N
NH
2
O
O
O
O
O
O
Dioxolane-T
DAPD
D-FDOC
Subsequent to those reports, we discovered that the
sample of the less toxic D-enantiomer that was tested
actually contained 3–5% of its significantly more toxic
L-counterpart. This then raised the interesting question
*
Corresponding author. Fax: +1-404-712-8679; e-mail: dliotta@
emory.edu
0960-894X/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.07.016

4992
S. Mao et al. / Bioorg. Med. Chem. Lett. 14 (2004) 4991–4994
L-enantiomer. Herein, we describe a synthetic approach
that employs a tandem kinetic resolution/chiral salt crys-
tallization protocol for preparing the D-enantiomer of
FDOC in high enantiopurity. In addition, conditions
that allow for the racemization and recycling of the
unwanted butyrate ester of the L-enantiomer of FDOC
were developed.
resulting in the racemization of the chiral lactol acetate
prior to glycosylation. In an effort to substantiate our
complexation hypothesis, we employed a less oxaphilic
Lewis acid catalyst in the glycosylation reaction. Thus,
when 10 was allowed to react with the silylated base
and TMSOTf, under the standard Vorbruggen condi-
¨
tions, we obtained a 1:1 b:a mixture. Analysis of the
b-product by chiral HPLC indicated low levels of race-
mization (the observed ee were 90% and 97% when 1.0
and 0.5equiv of TMSOTf were used, respectively). Con-
sistent with this trend, when the N-glycosylation reac-
tion was carried out with 1.5equiv of TMSOTf, the
observed ee dropped to 85%. Even though this approach
allowed us to obtain D-FDOC in 97% ee, the tedious
chromatography required to separate the a- and b-iso-
mers, as well as the lower-than-desired enantiopurity
of the product, caused us to continue our search for
more attractive alternatives.
The key step in our previously reported synthesis of
FDOC was a b-selective glycosylation that was uniquely
mediated by titanium trichloroisopropoxide (Ti-
Cl3(OiPr)) (Scheme 1).11,12 Thus, when racemic acetate
1 was allowed to react with silylated 5-fluorocytosine,
2, in the presence of the oxaphilic Lewis acid, Ti-
Cl (OiPr), under Vorbruggen conditions,13 we observed
¨
3
b:a selectivity of >20:1, albeit only in moderate yield.
More recently, we have found that the efficiency of this
coupling reaction could be markedly improved (68%
yield, b:a selectivity of >20:1) by using 2equiv Ti-
Cl3(OiPr). The cause of this low yield in the original re-
port was largely due to the competitive formation of an
isomeric, ring opening byproduct 5.14 An insidious as-
pect of 5 is that it possesses the same exact mass as
FDOC butyrate. In addition, its two most diagnostic re-
sonances, the anomeric proton and the proton on C6 of
the cytosine ring, exhibit the same chemical shifts as
those of a isomer. Since this material was likely formed
by reaction of the product with HCl and isopropanol,
generated from TiCl3(OiPr) hydrolysis, scrupulous
exclusion of moisture should and did improve the yield.
We next attempted to further refine the enantiopurity of
the product obtained from our previously developed ki-
netic resolution [Pig Liver Esterase (PLE) from Sigma]
by using a subsequent fractional crystallization (Scheme
2). Thus, exposure of 3 and 4 to PLE in 20% acetonitrile/
water produced D-FDOC in 95% ee.18,19 Incubation of
this material with 1equiv L-tartaric acid in isopropanol
at room temperature for about 24h, resulted in the for-
mation of a crystalline precipitate. The crystals were fil-
tered, dissolved in water and neutralized with sodium
hydroxide. Evaporation of the water and subsequent
chiral HPLC analysis showed that the enantiomeric
excess of D-FDOC to be 99.7%.19 Interestingly, direct
resolution of the butyrates 3 and 4 by chiral salt crystal-
lization technique proved unsuccessful due to the small
solubility differences between the two chiral tartrates.
Other acids such as (R)-(ꢀ)-10-camphorsulfonic acid
and (R)-(ꢀ)-mandelic acid were also examined, but
L-tartaric acid gave the most satisfactory result.
Based on our earlier results we assumed that if an enan-
tiopure oxolactone had been employed, each of the
enantiomers of FDOC could be obtained in high enan-
tiopurity. Towards this end the chiral oxolactone 1015
(ee > 99%) was converted to the desired nucleoside using
the same process. However, chiral HPLC analysis16
showed the product to be completely racemic
(ee = 0%). In retrospect, this result is consistent with
the complexation hypothesis described above.7,11,17 In-
deed, Lewis acid complexation, while ensuring the good
selectivity, simultaneously renders the C–O bond labile,
Using this approach, we, for the first time, had in hand a
method for producing D-FDOC in 99.7% ee that did not
require the use of chiral preparative HPLC. However,
NH
NHTMS
2
NH
N
2
F
F
n-C H COO
N
3
7
N
F
N
n-C H COO
n-C H COO
3
7
3
7
O
O
N
O
N
2
OTMS
O
OAc
O
O
Cl
O
O
i-PrO
n-C H COO
Ti
3
7
TiCl3(OiPr), CH2Cl2
O
Cl
Cl
Racemic mixture
68%
3 and 4
1
Complex
5
Scheme 1.
O
NH
NH
2
2
O
F
1) L-tartaric acid,
F
n-C H COO
N
O
3
7
N
N
Pig Liver Esterase
iPrOH
HO
HO
N
+
3
+ 4
N
O
N
O
F
40%
2) NaOH
O
O
NH
2
O
O
80%
7, 90% ee
6, 95% ee
8, 99.7% ee
Scheme 2.

S. Mao et al. / Bioorg. Med. Chem. Lett. 14 (2004) 4991–4994
4993
NHTMS
NH
O
2
F
O
O
N
F
n-C H COO
N
O
3
7
N
O
n-C H COO
N
O
3
7
HMDS (excess)
(NH4)2SO4 (cat.)
N
OTMS n-C H COO
3 7
N
N
O
+
7
, 90% ee
F
2
N
O
F
NH
2
Racemic mixture
β :α >18:1
O
TiCl3(OiPr),TiCl4
NHTMS
3
9
, 90% ee
4
CH2Cl2
NHTMS
N
NHTMS
N
F
F
n-C H COO
n-C H COO
3
7
3
7
O
N
O
N
O
O
O
O
O
O
O
O
O
O
O
O
N
N
n-C H COO
3
7
Cl
Cl
i-PrO
i-PrO
Cl
i-PrO
NHTMS
Ti
Ti
Ti
10
F
Cl
Cl
Cl
Cl
Cl
Cl
Thermodynamically
Preferred
Scheme 3.
while we considered this to be an important achieve-
ment, we noted that during the kinetic enzymatic resolu-
tion step, half of the butyrate-FDOC, for example, the
unwanted L-enantiomer 7, was being wasted. We, there-
fore, sought to develop a method for racemizing 7,
thereby allowing it to be recycled in subsequent kinetic
resolution/fractional crystallization runs without pro-
ducing any a-isomers. Stating this another way, racemi-
zation requires the simultaneous scrambling of the 10S
and 40S stereocenters of 7 exclusively to produce an
equal population of its enantiomeric 10R,40R counter-
part. We reasoned that this could be achieved by form-
ing an all-trans titanium complex similar to the putative
complex produced in the glycosylation of 1 by 2 (vide
infra). Towards this end, the butyryl ester of L-FDOC,
7, was silylated with hexamethyldisilazane (HMDS)
and then allowed to react with 1equiv of 2 in the pres-
ence of 2equiv of TiCl3(OiPr). After 3h the reaction
was quenched and the optical purity of the products
was determined using chiral HPLC.20 No change was
observed! We hypothesized that, since the silylated
fluorocytosine was probably a poorer leaving group
than the acetate in 1, we might need a higher degree of
Lewis acidity to catalyze the desired process. After sev-
eral attempts we found that exposing silylated 9 to
1equiv of 2, 0.7equiv of TiCl4 and 2equiv of
TiCl3(OiPr) for 7h resulted in complete racemization
of the starting L-isomer with maintenance of good
b-selectivity (b:a > 18:1) (see Scheme 3).
the Department of Veterans Affairs (RFS). We thank
Dr. Shaoxiong Wu and Dr. Bing Wang from the Chem-
istry Department at Emory University for excellent
NMR assistance.
Supplementary data
Supplementary data associated with this article can
be found, in the online version, at doi:10.1016/
j.bmcl.2004.07.016.
References and notes
1. Norbeck, D. W.; Spanton, S.; Broder, S.; Mitsuya, H.
Tetrahedron Lett. 1989, 30, 6263.
2. Chu, C. K.; Ahn, S. K.; Kim, H. O.; Beach, J. W.; Alves,
A. J.; Jeong, L. S.; Islam, Q.; Roey, P. V.; Schinazi, R. F.
Tetrahedron Lett. 1991, 32, 3791.
3. Kim, H. O.; Shanmuganathan, K.; Alves, A. J.; Jeong, L.
S.; Beach, J. W.; Schinazi, R. F.; Chang, C.-N.; Cheng
Y.-C.; Chu, C. K. Tetrahedron Lett. 1992, 33, 6899.
4. Rajagopalan, P.; Boudinot, F. D.; Chu, C. K.; Tennant,
B. C.; Baldwin, B. H.; Schinazi, R. F. Antiviral Chem.
Chemother. 1996, 7, 65.
5. Chen, H.; Schinazi, R. F.; Rajagopalan, P.; Gao, Z.; Chu,
C. K.; McClure, H. M.; Boudinot, F. D. AIDS Res.
Human Retroviruses 1999, 15, 1625.
6. Furman, P. A.; Jeffrey, J.; Kiefer, L. L.; Feng, J. Y.;
Anderson, K. S.; Borroto-Esoda, K.; Hill, E.; Copeland,
W. C.; Chu, C. K.; Sommadossi, J.-P.; Liberman, I.;
Schinazi, R. F.; Painter, G. R. Antimicrob. Agents
Chemother. 2001, 45, 158.
7. Wilson, L. J.; Choi, W.-B.; Spurling, T.; Liotta, D. C.;
Schinazi, R. F.; Cannon, D.; Painter, G. R.; St Clair, M.;
Furman, P. A. Bioorg. Med. Chem. Lett. 1993, 3, 169.
8. (a) Liotta, D. C. U.S. Patent 5,276,151, 1994; (b) Liotta,
D. C.; Schinazi, R. F. U.S. Patent 5,852,027, 1998.
9. Preparative SFC separation was achieved with a Chir-
alpak AS column using 30% MeOH isocratic elution.
Using this approach we obtained D-FDOC > 99% ee
(retention time: 4.05min) and L-FDOC > 99% ee (reten-
tion time: 3.48min).
In conclusion, enantiomerically pure D-FDOC (>99.7%
ee) was obtained using a tandem kinetic enzymatic resolu-
tion/chiral salt crystallization technique. In addition, the
unwanted L-butyrate-FDOC could be completely race-
mized while maintaining excellent b selectivity. Based on
these approaches, we believe that optically pure D-FDOC
can be produced on a large scale, thereby permitting more
advanced pharmacological and toxicological studies.
Acknowledgements
10. Schinazi, R. F.; Mellors, J.; Erickson-Viitanen, S.; Ma-
thew, J.; Parikh, U.; Sharma, P.; Otto, M.; Yang, Z.; Chu,
C. K.; Liotta, D. C. Antiviral Ther. 2002, 7, S15.
This work was supported by NIH grants 2R37AI-28731
(DCL) and 1R37AI-41890 (RFS) and by a grant from

4994
S. Mao et al. / Bioorg. Med. Chem. Lett. 14 (2004) 4991–4994
11. Choi, W.-B.; Wilson, L. J.; Yeola, S.; Liotta, D. C. J. Am.
Chem. Soc. 1991, 113, 9377.
15. A generous sample of chiral, nonracemic oxolactone 10
was kindly provided by Dr. George Painter of Triangle
Pharmaceuticals, Inc (now Gilead Science).
16. Chiral HPLC analysis conditions: Chiralpak AS,
25 · 0.46cm, detection at 271nm, mobile phase: EtOH/
hexane = 15:85, flow rate: 1.2mL/min. The retention times
for D- and L-isobutyrate-FDOC are 17.8 and 20.3min.
17. Wilson, L. J.; Hager, M. W.; EI-Kattan, Y. A.; Liotta, D.
C. Synthesis 1995, 12, 1465.
12. The observed selectivity differed markedly from the results
obtained with other Lewis acids. For example, trimethyl-
silyl triflate (TMSOTf) gave no stereoselectivity, titanium
tetrachloride (TiCl4) caused significant product epimeri-
zation and titanium dichlorodiisopropoxide (TiCl2(OiPr)2)
lacked sufficient Lewis acidity to catalyze the glycosyl-
ation.
13. (a) Vorbruggen, H.; Krolikiewicz, K.; Bennua, B. Chem.
18. Hoong, L. K.; Strange, L. E.; Liotta, D. C. J. Org. Chem.
1992, 57, 5563.
19. Chiral HPLC analysis conditions: Chiralpak AS,
25 · 0.46cm, detection at 271nm, mobile phase: EtOH/
hexane = 20:80, flow rate: 0.9mL/min. The retention times
for D- and L-FDOC are 18.1 and 15.8min.
20. Chiral HPLC analysis conditions: Chiralpak AS,
25 · 0.46cm, detection at 271nm, mobile phase: EtOH/
hexane = 15:85, flow rate: 1.2mL/min. The retention times
for D- and L-butyrate-FDOC are 18.9 and 21.8min.
¨
Ber 1981, 114, 1234; Also, see: (b) Okabe, M.; Sun, R.-C.;
Tam, S. Y.-K.; Todaro, L. J.; Coffen, D. L. J. Org. Chem.
1988, 53, 4780.
14. 1H NMR (CDCl3, 400MHz): d 0.90 (t, 3H), 1.20 (d, 3H),
1.13 (d, 3H), 1.60 (sextet, 2H), 2.27 (t, 2H), 3.72 (m, 1H),
4.10 (q, 1H), 4.20 (q, 1H), 6.01 (dt, 1H), 7.46 (d, 1H). 13C
NMR (CDCl3, 100MHz): d 13.76, 18.41, 21.59, 22.92,
36.01, 63.97, 72.09, 80.93, 124.77, 136.96, 154.57, 158.45,
173.02. FAB MS (M + H)+ : 302.3.