Development of the carbocyclic nucleoside MDL 201449A: A tumor necrosis factor-α inhibitor

Article abstract of DOI:10.1021/op9701245

An efficient synthesis of (1S,4R)-(-)-4-tert-butyldimethylsilyloxy-2-cyclopentenyl acetate and (1R,4S)-(-)-4-tert-butyldimethylsilyloxy-2-cyclopentenol is described utilizing a furfuryl alcohol rearrangement, followed by a lithium aluminum hydride reduction with high facial selectivity and an efficient enzymatic resolution with pancreatin. Both of these intermediates were successfully utilized in the preparation of the carbocyclic nucleoside 9N-[(1′R,3′R)-trans-3′-hydroxycyclopentanyl]adenine hydrochloride, an agent which inhibits the formation of tumor necrosis factor-α.

Full text of DOI:10.1021/op9701245

Organic Process Research & Development 1998, 2, 357−365
Development of the Carbocyclic Nucleoside MDL 201449A: A Tumor Necrosis
Factor-r Inhibitor
Timothy J. N. Watson,* Timothy T. Curran,† David A. Hay,‡ Ramnik S. Shah, David L. Wenstrup, and Mark E. Webster
Hoechst Marion Roussel Research Institute, 2110 East Galbraith Road, Cincinnati, Ohio 45215-6300
Scheme 1. Retrosynthetic analysis
Abstract:
An efficient synthesis of (1S,4R)-(-)-4-tert-butyldimethylsilyl-
oxy-2-cyclopentenyl acetate and (1R,4S)-(-)-4-tert-butyldi-
methylsilyloxy-2-cyclopentenol is described utilizing a furfuryl
alcohol rearrangement, followed by a lithium aluminum hydride
reduction with high facial selectivity and an efficient enzymatic
resolution with pancreatin. Both of these intermediates were
successfully utilized in the preparation of the carbocyclic
nucleoside 9N-[(1′R,3′R)-trans-3′-hydroxycyclopentanyl]adenine
hydrochloride, an agent which inhibits the formation of tumor
necrosis factor-r.
Introduction
The overproduction of the proinflammatory cytokine,
tumor necrosis factor-R (TNF-R), has been directly linked
to multiple-organ diseases.1 Such diseases include septic
shock, adult respiratory distress syndrome, AIDS, inflam-
matory bowel disease, bacterial meningitis, malaria, multiple
sclerosis, atherosclerosis, and rheumatoid arthritis to name
a few.1 The carbocyclic nucleoside 9N-[(1′R,3′R)-trans-3′-
hydroxycyclopentanyl]adenine hydrochloride, (MDL 201449A,
1) has been shown to inhibit the production of TNF-R, thus
Retrosynthetically, one can envision a disconnection
between the N9 position of the adenine portion and the
cyclopentanol ring of 1, as shown in Scheme 1. The
coupling of adenine 2 to a differentially protected cyclopen-
tadiol A can be accomplished via SN2 displacement to give
the desired R,R chirality in product 1. Therefore, the 1,3-
disubstituted cyclopentane ring must be cis in nature, as
displayed in enantiomer A. These 1,3-cis-disubstituted
cyclopentane systems can be easily obtained through reduc-
tion of cis-2-cyclopentene-1,4-diol derivatives B. Previous
preparations of many cis-2-cyclopentene-1,4-diol derivatives
proved costly, lengthy, or difficult for scaling up.2
The first generation synthesis developed by Discovery
Chemistry centered on the preparation of the 2-cyclopentene-
1,4-diol derivative 5 shown in Scheme 2. Dicyclopentadiene
was cracked and captured with bromine to form a mixture
of the cis- and trans-1,4-dibrominated cyclopentene3 deriva-
tives 3. This was converted to the diol 5, which was
determined to consist of a 55:45 cis:trans mixture. This
mixture was desymmetrized enzymatically to produce the
desired (1R,4S)-(-)-4-acetoxy-2-cyclopentenol (6), which
was separated from the diacetate and trans diol via chro-
matography. The olefin was reduced catalytically to (1R,3S)-
(-)-1-acetoxycyclopentan-3-ol (7), and the free hydroxyl
group was converted to the mesylate 8. Coupling of
mesylate 8 with adenine 2 was carried out with sodium
hydride to give the acetoxy-protected carbocyclic nucleoside
having potential for the treatment of multiple inflammatory
diseases.1 It became necessary to develop a synthesis for
preparing larger quantities of MDL 201449A that would
provide 2 kg for preliminary toxicological studies.
* To whom correspondence should be addressed at Hoechst Marion Roussel,
Chemical Development, 2110 E. Galbraith Rd., Cincinnati, OH 45125.
† Present address: Wyeth-Ayerst Research, 401 N. Middletown Rd., Bldg.
222, Rm. 2148, Pearl River, NY 10965.
‡ Present address: Lilly Research Laboratories, Lilly Corporate Center, DC
4813, Indianapolis, IN 46285.
(2) (a) Curran, T. T.; Hay, D. A.; Koegel C. P.; Evans, J. C. Tetrahedron 1997,
53, 1983. (b) For a complete list of cis-2-cyclopentene-1,4-diol derivatives,
see ref 1 of ref 2a. (c) Watson, T. J. N. Chemical Development of MDL
201449A. Presentation at the Midwest Pharmaceutical Process Chemistry
Consortium, Cincinnati, OH, October 10, 1997. A copy of slides can be
obtained from Tim Watson: Hoechst Marion Roussel Research Institute,
2110 E. Galbraith Rd., Cincinnati, OH 45215-6300.
(1) (a) Edwards, C. K., III; Borcherding, S. M.; Zhang, J.; Borcherding, D. B.
The Role of Tumor Nercosis Factor-R in Acute and Chronic Inflammatory
Responses: Novel Therapeutic Approaches. In Xenobiotics and Inflamma-
tion; Schook, L. B., Laskin, D. L., Eds.; Acedemic Press: New York, 1994;
pp 97-136. (b) Borcherding, D. R.; Peet, N. P.; Munson, R.; Zhang, H.;
Hoffman, P. F.; Bowlin, T. L.; Edwards, C. K., III. J. Med. Chem. 1996,
39, 2615-2620.
(3) Owen, L. N.; Smith, P. N. Alicyclic Glycols 1952, 4035-4047.
10.1021/op9701245 CCC: $15.00 © 1998 American Chemical Society and Royal Society of Chemistry
Published on Web 09/12/1998
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
357

Scheme 2. First generation Discovery Chemistry synthesis
9. This was deprotected with acid, yielding MDL 201449A
(1) in an overall yield of less than 1% and a chemical purity
typically ranging from 90 to 95%.1,4
A critique of the discovery synthesis rendered the fol-
lowing development issues. The cracking of dicyclopenta-
diene to cyclopentadiene is commonly used to prepare cis-
2-cyclopentene-1,4-diol derivatives.2 Although the difficulties
attributed to this chemistry are not entirely impossible to
scale-up, we prefer to avoid them due to some safety issues
and patent questions. Bromination of the cyclopentadiene
is non-selective and thus yields a cis:trans mixture, of which
the trans portion is rendered useless for the preparation of
1. The 14% yielding enzymatic desymmetrization of diols
5 using pancreatin is less than desirable, and the use of
dichloromethane in the mesylation reaction has environmen-
tal constraints. The final deprotection with hydrochloric acid
in ethanol, however, was attractive because the product
crystallized from the reaction medium and could be easily
isolated via filtration.
Some analytical evaluations of the early development
samples revealed four major impurities (Figure 1). The
enantiomer 10 and the diastereomer 11 presumably come
from the contamination of (1R,3S)-(-)-1-acetoxycyclopen-
tan-3-ol (7) with the other possible isomers. Impurities 12
and 13 are the regioisomers generated during the coupling
reaction with adenine.2c,5a All four contributed to roughly
5% (best sample) of the total impurities in MDL 201449A
and could not be removed without tedious preparative
chromatography techniques.
Figure 1. Major impurities.
eliminate the need for preparative chromatography. The
details contained within pertain directly to our scale-up efforts
and issues encountered during the preparation of multigram
quantities of MDL 201449A.2,5,6
Results and Discussion
The route eventually developed to prepare a 1,4-cis-
disubstituted cyclopentene derivative is shown in Scheme
3.2a,c,6 Both (1S,4R)-(-)-4-tert-butyldimethylsilyloxy-2-cyclo-
pentenyl acetate (17) and (1R,4S)-(-)-4-tert-butyldimethyl-
silyloxy-2-cyclopentenol (18) have the potential to yield
Our efforts concentrated on developing a process that
would be cost-effective and easily scaled-up and would
(4) Borcherding, D. Discovery Chemistry, Hoechst Marion Roussel, private
communications.
(5) (a) Curran, T. T.; Barbuch, R. B.; Hay, D. A.; Vaz, R. J. Tetrahedron 1997,
53, 7181. (b) Curran, T. T.; Hay, D. A. Tetrahedron: Asymmetry 1996, 7,
2791.
(6) (a) For a more complete history on route selection and methodology, refer
to refs 2 and 5. (b) Curran, T.; Hay, D.; Evans, J. International Patent WO
96/30320, 3 October 1996.
358
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development

Scheme 3
Table 1. Furfuryl alcohol rearrangement
selectivity with hydride entering anti to the hydroxyl group.
Direct reduction of 14 gave low yields, poor isolation, and
moderate selectivity, suggesting the need for a protecting
group.2a,c A survey of protecting groups revealed the tert-
butyldimethyl silyl group (TBDMS) to have several advan-
tages.6 First, it helped add steric bulk to one face of the
molecule, thus increasing facial selectivity during reduction.
Second, it increased the overall lipophilicity, thus assisting
with isolation. For these reasons, along with time constraints,
the TBDMS protecting group was prepared under standard
conditions.
concn of furfuryl
alcohols (M)
yield
(%)
The reduction of cyclopentenone 15 was conducted using
lithium aluminum hydride (LAH)/lithium iodide (LiI) to give
the alcohol 16 with a 39:1:trace ratio of cis:trans:over-
reduction product resulting from 1,4 addition (followed by
1,2 addition).2a,c,6 Lithium iodide reduces the 1,4-addition
side product and ultimately permits an increase in reaction
temperature from -78 °C (difficult to attain at our facility)
to -30 °C, which is readily achievable in our reactor.2c All
Lewis acids substituted for LiI proved inferior.2a Although
not as efficient as the LAH/Et2O reducing conditions in terms
of cis:trans ratio,2a a toluene and tert-butyl methyl ether
(TBME) solvent system was developed to eliminate the
hazards associated with diethyl ether. tert-Butyl methyl ether
is an acceptable replacement for diethyl ether and is
necessary to solubilize the LAH in a toluene environment
(Table 2). The large-scale reduction eventually used 0.54
equiv of LAH and 2.1 equiv of LiI. Due to the apparent
handling hazards associated with LAH and desired anhydrous
LiI, these raw materials were purchased in preweighed
toluene-soluble bags. This is an excellent method of
charging hazardous and hygroscopic reagents to minimize
exposure and reduce safety issues.2c
Reduction of (()-4-tert-butyldimethylsilyloxy-2-cyclo-
pentenone (15) led to another interesting observation. Cy-
clopentenone 15 was found to be contaminated with roughly
17% of tert-butyldimethylsilanol from the use of excess tert-
butyldimethylsilyl chloride (TBDMSCl) in the protection
step. Although the silanol can be removed by tedious
vacuum distillation or an azeotrope with toluene, this was
not done on account of the small-scale reactions not being
effected from this contaminant. Surprisingly, tert-butyldi-
0.34
0.68
2.55
2.55a
53
35
17
20
a Alcohol added over 24 h.
MDL 201449A (1). Their simple synthesis was attractive
for preparing the first 2 kg of drug substance.
Following a procedure by Nanni et al.,2a,7 furfuryl alcohol
could easily be rearranged under acidic conditions to (()-
4-hydroxy-2-cyclopentenone (14), in which the oxygen
functionalities have the proper regioconfiguration. Our best
yield was 53% when the reaction was run under dilute
conditions;7 furthermore, after more experimentation,2a,c it
was found that the yield diminishes at higher concentrations
(Table 1). For the most part, the decreased yields were
attributed to higher amounts of a visible polymer formed
from the furfuryl subunit during the reaction. However, due
to the inexpensive furfuryl alcohol8 in conjunction with the
aqueous reaction medium, we proceeded at a concentration
of 2.5-2.8 M in a 30-gal reactor. A cost and time analysis
showed that this concentration maximizes the throughput for
the desired quantity of 14, which was necessary to expedite
the process. Due to time constraints, experimental design
to optimize or revise the reaction was not performed.
To provide the desired cis configuration in the cyclopen-
tane moiety, reduction of the ketone required high facial
(7) Nanni, M.; Ta-machi, H.; Moriyama-shi. Japanese Patent Disclosure Bulletin
No. 57-62236, 15 April 1982.
(8) Furfuryl alcohol cost for our campaign was $2.40/kg.
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
359

Table 2. Reduction of enone 15
and TBDMS-protected silanol 18 were obtained in high
enantiomeric excess, they contained some minor unknown
impurities (roughly 95% pure). When analyzed by chiral
gas chromatography and GC/MS, they were presumably
product related. How these would affect the overall purity
of 1 remained unknown.
The most direct path to 1 (route 1, Scheme 4) utilized
the TBDMS-protected silanol 18. While we believe the
TBDMS group added versatility, it remained to be deter-
mined if this was a sufficient replacement for the efficient
acetoxy group which Discovery Chemistry had shown could
be efficiently removed with hydrochloric acid to provide 1
from 9 (Scheme 2). Although the acetoxy-TBDMS-protected
cyclopentenediol 17 required a selective deprotection step
in the process, this chemistry was also proven successful in
preparing MDL 201449A (1). Both routes 1 and 2 from
Scheme 4 were evaluated and compared.
yield cis:trans:1,4
(%) reduction
conditions
1 equiv of LAH
1 equiv of LAH; 5 equiv
of LiI
Et2O, -78 °C 70 40:1:4
-20 °C
74 25:1:trace
79 24:1:trace
75 35-39:1:trace
0.5 equiv of LAH; 0.5 equiv -30 °C
of LiI
toluene/TBME 0.5 equiv
of LAH; 2 equiv of LiI
-30 °C
Table 3. Silanol contamination: reduction of enone 15
TBDMS-OH
cis:trans ratio
Route 1. Route 1 is summarized in Scheme 5. Hydro-
genation of the TBDMS-protected diol 18 using Raney nickel
revealed another obstacle: double bond migration of the
olefin. This migration produced ketone 19′ (Table 5) as a
side product, resulting in irreproducible and moderate yields
of desired 19. Minimization of the formation of ketone 19′
17-20% (∼0.5 equiv to LAH)
32%
0-2.1% (catalytic)
remove LiI (with 17% trimethylsilanol
35-39:1
28:1
20-24:1
26:1
would clearly affect the throughput of the process.
A
summary of some catalysts screened is shown in Table 5.2a
The catalyst of choice for this process proved to be nickel
boride.10 Nickel boride [P-2], generated from nickel acetate
and sodium borohydride in an alcoholic medium, gave
reproducible results in high yields with minimum olefin
migration. This catalyst was also attractive from a safety
perspective because it was significantly less pyrophoric than
the other catalysts that were screened.
methylsilanol clearly had an effect on the facial selectivity
observed during the LAH reduction (Table 3). One may
speculate that the tert-butyldimethylsilanol and LAH form
a new bulky reducing species. Why the 17% of tert-
butyldimethylsilanol gave the optimal cis:trans ratio is not
fully understood at this time.2a
A survey of resolving conditions for alcohol 16 ultimately
led to the selection of the acylation of the free hydroxyl group
with vinyl acetate and the enzyme pancreatin.2,5 Pancreatin
is readily available and is a relatively inexpensive enzyme.9
A screening of this enzyme in traditional solvents proved
that, by using TBME, both acetate 17 and alcohol 18 were
obtained in high enantiomeric excess (Table 4). The use of
THF as solvent yielded a undesirably slow resolution time
with diminished ee’s, while cyclohexane has flammability
liability which we wanted to avoid. The increase in
lipophilicity of the TBDMS alcohol 16 contributed to much
of the enzymatic resolution success in a wide array of ethers
and solvents. All changes of the protecting group gave
inferior results in the resolution and did not provide
exceptional optical purities as obtained in the silyl ether
series; such high optical purities of 17 and 18 permitted us
to investigate two routes to 1.2a,c Separation of acetate 17
from alcohol 18 was easily accomplished via plug chroma-
tography (see Experimental Section). The kilo-scale reaction
did not produce, in terms of yield, the same efficiency as
the small laboratory screenings because there were some
mixed fractions that were discarded. It should be noted that
acetate 17 and alcohol 18 possess different vapor pressures
which may afford a separation by distillation for future
supplies.2c
The mesylate of 19 was easily prepared in TBME
replacing the undesired dichloromethane used originally. The
coupling conditions of the mesylate of 19 with adenine 2
were restricted due to poor solubility.2a,c,5a Due to time
constraints, this coupling was eventually performed with
sodium hydride in dimethyl acetamide (DMA) to give the
TBDMS-protected MDL 201449A, 20. It was discovered
that this compound could be crystallized from methanol. This
crystallization removed the TBDMS-protected regioisomers
of 12 and 13 (Figure 1). Because of time constraints, a poor
coupling and crystallization yield of 34% was tolerated.
TBDMS deprotection of MDL 201449A 20 with hydrochlo-
ric acid proceeded without incident to provide MDL 201449A
(1), which crystallized from the reaction medium. The
overall purity of 1 was 96.8%, with the two major impurities
being enantiomer 10 and diastereomer 11. These could not
be removed via recrystallization of 1 or 20. In fact, the purity
profile of 20 directly dictated that of 1. Nevertheless, 2 kg
of 1 was prepared and delivered for preliminary toxicological
and safety studies.
Ultimately, the purification issue in route 1 was solved
when it was found that the (1R)-(-)-10-camphorsulfonate
(CSA) salt of 20 could be prepared and recrystallized to a
purity of >98% with no more than 1% each of 10 and 11
All of the intermediates in Scheme 3 were viscous oils.
Although acetoxy-TBDMS-protected cyclopentenediol 17
(10) (a) Rylander, P. N. Catalytic Hydrogenation in Organic Synthesis; Academic
Press: New York, 1979; pp 37-38. (b) Brown, C. A.; Ahuja, V. K. J.
Org. Chem. 1973, 38, 2226.
(9) Pancreatin (American Laboratories Inc.), $75/kg (8× USP specifications).
360
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development

Table 4. Pancreatin resolution screenings
yield (%) (% ee)
enzyme (wt equiv)
time (h)
solvent
THF
TBME
cyclohexane
reagents
17
18
pancreatin (3)
pancreatin (3)
pancreatin (3)
26
7
2
VAa (5 equiv), Et3N (0.7 equiv)
VA (5 equiv), Et3N (0.7 equiv)
VA (5 equiv)
46 (96)
46 (99)
50 (99)
48 (98)
50 (99)
44 (98)
a VA ) vinyl acetate.
Scheme 4
Scheme 5. Route 1
the scope of route 2. This deprotection problem was
eventually solved by selective silyl ether cleavage with
hydrochloric acid (1 equiv) at ambient temperature in ethanol.
Much to our delight, the acetyl alcohol 6 was found to
crystallize from TBME. This crystallization resulted in a
chemical purity of >99%, with an optical purity of >99%
ee.2a,c The acetyl alcohol 6 was the only solid encountered
in the protected cyclopentene-cyclopentanediol series.
Hydrogenation of the acetyl-protected cyclopentenediol
6, not suprisingly, displayed the same olefin migration as
the monoprotected TBDMS species (thus, shedding some
light on the moderate yields observed in the Discovery
Chemistry hydrogenation). Again, nickel boride proved to
be the catalyst of choice for this reduction. The mesylate
of alcohol 7 was coupled with adenine 2 using chemistry
analogous to route 1 to give acetyl-protected 9, which, in
turn, was plug chromatographed to minimize the N-3 isomer
12, followed by crystallization from MEK to effect the
removal of trace ammounts of the N-3 isomer and further
minimize acetyl isomers 12 and 13. Deprotection of 9 with
hydrochloric acid proved successful in ultimately preparing
107 g of MDL 201449A (1), with greater than 99% overall
purity. The value of the precoupling, crystalline intermediate
6 is clearly reflected in the overall purity of 1.
(Scheme 6). Interestingly enough, the (S) enantiomer of CSA
was also sufficient in carrying out this process. However,
it became clear that a crystalline intermediate prior to
coupling with adenine 2 could potentially eliminate the need
for this purification technique.
Route 2. Route 2 is summarized in Scheme 7. Selective
deprotection of the TBDMS group of 17 with tetrabutyl-
ammonium fluoride (TBAF) resulted in acetyl migration.
Prolonged reaction exposure times involved with scaling-
up resulted in excessive acetyl migration, thus forming more
of the enantiomer as well as the diacetate and diol, resulting
in diminished yields (table in Scheme 7).
Conclusion
Two successful routes to MDL 201449A were developed
from both products derived from an efficient enzymatic
resolution of TBDMS-protected cyclopentenediol 16. Both
routes eliminated preparative chromatography and proved
successful in providing larger amounts of MDL 201449A
with purities >98% and ee’s >99%. Route 1 utilized a salt
purification of the penultimate intermediate TBDMS-
Although the deprotection with TBAF was unacceptable,
it provided adequate amounts of acetoxy alcohol 6 to explore
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
361

Table 5. Hydrogenation summary of route 1
catalyst (wt %)
conditions
yield (%)
alcohol/ketone
Ra-Ni (10)
EtOH (40-55 psi)
70-88
70-80
75-80
95
>20:1 to 4:1 (irreproducible)
5% Pt/Al2O3 (10)
5% Pt/CaCO3 (10)
Ni2B-P2
EtOAc or EtOH (1 atm)
EtOAc or EtOH (1 atm)
EtOH (1 atm or 50 psi)
11:1
5:1
>9:1
Scheme 6. CSA purification
program, 100 °C for 2 min, then heated to 250 °C at 20
°C/min; column, HP-1 methyl silicone gum 10 m × 0.53
mm × 2.65 µm film thickness; tR of furfuryl alcohol ) 1.0
min, tR of 14 ) 3.6 min). A typical reaction time of 48 h
was required. The reaction mixture was cooled to 35 °C
and diluted with ethyl acetate (40 L), and the layers were
separated. The aqueous layer was extracted with ethyl
acetate (30 L), and the combined organic layers were
extracted with water (3 × 10 L). Aqueous extracts were
combined and concentrated at 60 °C/50 Torr to give an oil
weighing 3.6 kg. The oil was dissolved in THF (16 L), and
the resulting solution was dried with MgSO4 (500 g). The
drying agent was filtered off and the filter cake was washed
with THF (4 L). The filtrate was concentrated at 40 °C/50
Torr to give 2.6 kg of (()-4-hydroxy-2-cyclopentenone (14)
as a red oil (17%). 1H NMR (CDCl3): δ 7.6 (dd, 1H), 6.2
(d, 1H), 5.0 (dt, 1H), 3.6-3.2 (bs, -OH, 1H), 3.8 (dd, 1H),
2.2 (dd, 1H). Complete analytical data were compared
against the control sample reported in ref 2a.
(()-4-tert-Butyldimethylsilyloxy-2-cyclopentenone (15).
A mixture of THF (20 L), NEt3 (9.00 L, 64.5 mol), 14 (4.00
kg, 40.7 mol), and DMAP (100 g) was cooled under nitrogen
to 0 °C. tert-Butyldimethylsilyl chloride (5.85 kg, 38.8 mol)
was dissolved in THF (5 L) and added to the reaction mixture
at such a rate that the reaction temperature did not exceed 0
°C (total addition time was 0.75 h). The reaction mixture
was warmed to 25 °C, stirred for 16 h, and monitor by GC
(injector temperature, 190 °C; detector temperature, 270 °C;
program, 100 °C for 2 min, then heated to 250 °C at 20
°C/min; column, HP-1 methyl silicone gum 10 m × 0.53
mm × 2.65 µm film thickness; tR of 14 ) 3.6 min, tR of 15
) 2.6 min). The reaction mixture was cooled to 0 °C, and
H2O (22 L) was slowly added at such a rate that the
temperature of the reaction mixture did not exceed 25 °C.
A total of 16 L of 1 N HCl was added, and the mixture was
stirred for 30 min. The mixture was then extracted with
hexane (2 × 40 L). The organic layers were combined,
washed with 0.2 N HCl (20 L), saturated NaHCO3 (12 L),
and brine (12 L) and then dried over MgSO4. Drying agent
was filtered off and washed with hexane (5 L). Solvent was
removed at 40 °C/100 Torr, followed by distillation of
protected 20, whereas route 2 employed the crystalline
properties of acetyl-protected diol 6. The “cost of the
chirality” in MDL 201449A is lowered significantly since
furfuryl alcohol is efficiently converted into two useful
intermediates, 17 and 18, in high enantiomeric excess via
rearrangement, protection, stereoselective reduction, and
resolution.
Although this was a highly efficient two-route process
for preparation of MDL 201449A (1), ideally a single process
is desired. Furthermore, an initial cost analysis revealed that
TBDMSCl was responsible for 39% of the overall cost of
1.11 Developing a postresolution, recyclable intermediate and
replacing the TBDMS protecting group are two areas that
could greatly improve the process. Nevertheless, the syn-
thesis described herein provided an expeditious process to
larger quantities of MDL 201449A, along with providing
valuable insight into future development plans.
Experimental Section
General. NMR spectra were recorded on a Varian XL
300 and/or Varian Gemini-300 spectrometers at 300 MHz
for1H and at 75 MHz for 13C.
(()-4-Hydroxy-2-cyclopentenone (14). A mixture of
H2O (60 L), furfuryl alcohol (15.0 kg, 152.9 mol), and
KH2PO4 (750 g) was adjusted to a pH of 4.1 by addition of
H3PO4 and then heated to and maintained at 98 °C. The
reaction was monitored by gas chromatography (GC) (injec-
tor temperature, 190 °C; detector temperature, 270 °C;
(11) The cost analysis of TBDMSCl was performed in 1995. Since then, bulk
suppliers of TBDMSCl have lowered the price.
362
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development

Scheme 7. Route 2
product at 90 °C/6 Torr to give 6.6 kg (76% yield) of (()-
4-tert-butyldimethylsilyloxy-2-cyclopentenone (15) as a light
yellow oil. 1H NMR (CDCl3): δ 7.4 (dd, 1H), 6.2 (dd, 1H),
5.0 (dd, 1H), 2.7 (dd, 1H), 2.2 (dd, 1H), 0.95 (s, 9H), 0.05)
(s, 6H). 13C NMR (CDCl3): 206.4, 163.8, 134.4, 70.8, 44.9,
25.6, 18.0, -4.6 ppm. MS: m/z (M+) calcd 212, (M + H)
obsd 213. Complete analytical data were compared against
the control sample reported in ref 2a.
(1S,4R)-(-)-4-tert-Butyldimethylsilyloxy-2-cyclopen-
tenyl acetate (17) and (1R,4S)-(-)-4-tert-Butyldimethyl-
silyloxy-2-cyclopentenol (18). 16 (6.5 kg, 30 mol) and
TBME (40 L) were combined and stirred under a nitrogen
atmosphere. To the mixture were added Et3N (2.87 L, 20.6
mol), pancreatin (8× USP grade, 19.5 kg, 3 wt equiv), and
vinyl acetate (14 L, 151.9 mol). The mixture was stirred
for 7 h while the reaction was monitored using GC (injector
temperature, 190 °C; detector temperature, 270 °C; program,
100 °C for 2 min, then heated to 250 °C at 20 °C/min;
column, HP-1 methyl silicone gum 10 m × 0.53 mm × 2.65
µm film thickness; tR of 16 ) 2.3 min, tR of 17 ) 3.7 min,
tR of 18 ) 2.3 min). The pancreatin was filtered off and
washed with TBME (18 L), and the filtrate was concentrated
at 30 °C/50 Torr to give 9.0 kg of crude 17 and 18. The 9
kg was divided into thirds and chromatographed on 12 kg
of silica gel (Merck; 230-400 mesh), eluting first with 32
L of heptane, then with 5% ethyl acetate/heptane to collect
17, and finally with 50% ethyl acetate/heptane to collect 18
(TLC 5% ethyl acetate/heptane; Rf of 17 ) 0.5, Rf of 18 )
0.1). Both fractions were concentrated separately at 40 °C/
50 Torr to produce a total (from the three chromatographies)
of 2.72 kg of (1S,4R)-(-)-4-tert-butyldimethylsilyloxy-2-
cyclopentenyl acetate (17, 70% yield, 98% ee) and 2.5 kg
of (1R,4S)-(-)-4-tert-butyldimethylsilyloxy-2-cyclopentenol
(18, 77% yield, 98.4% ee). Note: conditions for determining
% ee of these materials can be found in refs 2a and 5b.
Compound 17. 1H NMR (CDCl3): δ 6.0 (bd, 1H), 6.0
(bd, 1H), 5.9 (bd, 1H), 5.4 (bt, 1H) 4.7 (bt, 1H), 2.8 (m,
1H), 2.0 (s, 3H), 1.6 (m, 1H), 0.9 (s, 9H), 0.1 (s, 6H). 13C
NMR (CDCl3): 170.8, 138.8, 131.1, 76.9, 74.8, 41.1, 25.8,
21.1, 18.1, -4.6 ppm. MS: m/z (M+) calcd 256, (M + H)
obsd 257. IR (neat): CdO 1653.
(()-cis-4-tert-Butyldimethylsilyloxy-2-cyclopentenol (16).
Toluene (48 L), LiI (8 kg, 59.7 mol), and LAH (600 g, 15.8
mol) were combined under a nitrogen atmosphere and cooled
to -30 °C. A solution of 15 (6 kg, 28.25 mol) in TBME
(28 L) was added to the mixture at such a rate that the
reaction temperature did not exceed -20 °C. The reaction
mixture was stirred at -30 °C for 3 h while the reaction
was monitored by GC (injector temperature, 190 °C; detector
temperature, 270 °C; program, 100 °C for 2 min, then heated
to 250 °C at 20 °C/min; column, HP-1 methyl silicone gum
10 m × 0.53 mm × 2.65 µm film thickness; tR of 15 ) 2.6
min, tR of 16 ) 2.3 min). The reaction was quenched with
saturated NH4Cl (20 L) at such a rate that the reaction
temperature did not exceed 10 °C. The aluminum salts were
filtered off and washed with toluene (6 L). The organic layer
was separated, and the aqueous layer was extracted with
toluene (6 L). All organic extracts were combined and
concentrated at 50 °C/100 Torr. The residue was dissolved
in ethyl acetate (8 L) and dried over MgSO4 (500 g). Drying
agent was filtered off and washed with ethyl acetate (2 L).
The filtrate was evaporated at 50 °C/100 Torr to give 6.4
kg of (()-cis-4-tert-butyldimethylsilyloxy-2-cyclopentenol
(16). This compound was used without further purification
(76% yield, 96% pure by GC; 37:1 cis/trans) (column,
CDX-â 10 m × 0.25 mm; 0.25 µm, T ) 100 °C). 1H NMR
(CDCl3): δ 6.0 (bd, 1H), 5.9 (bd, 1H), 4.7 (m, 1H), 4.6 (m,
1H), 2.7 (m, 1H), 2.2 (bm, 1H), 1.5 (m, 1H), 0.9 (s, 9H),
0.1 (s, 6H). 13C NMR (CDCl3): 128, 127, 176 (2C’s), 44,
26, 18, -4.7 ppm. Complete analytical data were compared
against the control sample reported in ref 2a.
Compound 18. 1H NMR (CDCl3): δ 6.0 (bd, 1H), 5.9
(bd, 1H) 4.7 (bt, 1H), 4.6 (bt, 1H), 2.6 (m, 1H), 1.8 (m, 1H),
0.8 (s, 9H), 0.0 (s, 6H) (-OH not visible). 13C NMR
(CDCl3): 136.8, 135.6, 75.1, 75.0, 44.6, 25.8, 18.1, -4.6
ppm. MS m/z (M+) calcd 214, (M + H) obsd 215. IR
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
363

(neat): -OH (b), 3352. Complete analytical data were
compared against the control sample reported in ref 2a.
(1R,3S)-(-)-1-tert-Butyldimethylsilyloxycyclopentan-3-
ol (19). A 5-gal autoclave was charged with nickel(II)
acetate tetrahydrate (136 g, 0.546 mol) and ethanol (6.6 L).
The mixture was stirred at room temperature while a 1 M
solution of NaBH4 in ethanol (550 mL) was added in one
portion. A solution of 18 (2.19 kg, 10.23 mol) in ethanol
(2.2 L) was added, and stirring was stopped. The unit was
purged three times with N2, followed by three times with
H2. The autoclave was charged to 200 psi with H2, and
stirring was reinitiated. After hydrogen uptake was complete
(4 h), the mixture was filtered through a prepared cake of
filter aid (500 g) and charcoal (35 g) and washed with ethanol
(4 L). The filtrate was evaporated at 40 °C/50 Torr to give
2.098 kg of (1R,3S)-(-)-1-tert-butyldimethylsilyloxycyclo-
pentan-3-ol (19, 95% yield, 92% pure by GC) (Column,
CDX-â 10 m × 0.25 mm × 0.25 µm, T ) 100 °C, tR of 19
) 9.89 min). 1H NMR (CDCl3): δ 4.4 (bt, 1H), 4.3 (bt,
1H), 3.0 (bs, 1H, -OH), 1.5-2.0 (m, 6H), 0.8 (s, 9H), 0.0
(s, 6H).
9N-[(1′R,3′R)-trans-3′-tert-Butyldimethylsilyloxycyclo-
pentanyl]adenine (20). A solution of 19 (1.73 kg, 8.00
mol), TBME (14 L), and Et3N were combined and cooled
to -5 °C. MsCl (744 mL, 9.6 mol) was added at such a
rate that the reaction temperature did not exceed 15 °C (total
addition time was 2 h). The reaction was monitored by GC
(injector temperature, 190 °C; detector temperature, 270 °C;
program, 100 °C for 2 min, then heated to 250 °C at 20
°C/min; column, HP-1 methyl silicone gum 10 m × 0.53
mm × 2.65 µm film thickness; tR of 19 ) 2.3 min, tR of
mesylate ) 5.6 min). The salts were filtered off and washed
with TBME (4 L), and then the filtrate was washed with
water (5 L). The organic layer was dried over MgSO4. The
drying agent was filtered off and washed with TBME (2 L),
and the filtrate was concentrated at 40 °C/50 Torr to give
2.34 kg of the mesylate, which was used without further
purification.
To a 10-gal reactor was charged NaH (384 g, 9.6 mol,
60% dispersion in mineral oil), DMA (12 L), and adenine 2
(1.521 kg, 11.2 mol). The solution was stirred vigorously
at ambient temperature until H2 evolution ceased. The
mixture was heated to 60 °C for 2 h and then cooled to
ambient temperature. The mesylate (2.34 kg, crude) was
added, and the mixture was heated and maintained at 60 °C
for 18 h (monitored by disappearance of mesylate on GC
and formation of product by HPLC; column, ChiralPak AD
250 × 4.6 mm; mobile phase, 85% pentane/15% IPA; flow
rate, 1 mL/min; tR of 20 ) 10.2 min). The reaction mixture
was cooled to room temperature and quenched with water
(24 L). The mixture was extracted with ethyl acetate/toluene
(3:1, 2 × 20 L), and the organic layers were combined and
washed with water (30 L) (NaCl may be added to help the
resulting emulsion disperse). The organic layer was sepa-
rated and dried over MgSO4 (2 kg). The drying agent was
filtered off and washed with ethyl acetate (2 L), and the
filtrate was concentrated at 50 °C/50 Torr to produce 2.24
kg of crude 20. The oily residue was dissolved in hot
methanol (4 L), and the solution was allowed to stand at
ambient temperature for 24 h. The crystals which formed
were filtered off and washed with heptane (2 L) to give 902
g of 9N-[(1′R,3′R)-trans-3′-tert-butyldimethylsilyloxycyclo-
pentanyl]adenine (20) in 34% yield. Complete analytical
data were compared against the control sample reported in
ref 5a.
9N-[(1′R,3′R)-trans-3′-Hydroxycyclopentanyl]ade-
nine Hydrochloride (1). 20 (2.73 kg, 8.20 mol) was
dissolved into 15 L of warm ethanol and polish-filtered into
a 30-gal reaction vessel. The mixture was cooled to room
temperature, and to this was added 2.7 L of 6 N HCl in
ethanol over a period of 1 min. The reaction mixture was
stirred for 14 min, and then 16 L of 1:1 heptane/ethanol was
added. The reaction mixture was cooled to and maintained
at 0 °C for 17 h. The white precipitate was filtered off via
centrifuge and washed with 8 L of heptane. The white solid
was allowed to air-dry for 5 h and then placed in a humidity
chamber overnight. This gave 1.98 kg of the stable mono-
hydrate of 9N-[(1′R,3′R)]-trans-3′-hydroxycyclopentanyl]-
adenine hydrochloride (1, MDL 201449A, 94% yield, 96.8%
pure) (HPLC: column, ChiralPak AD 250 × 4.6 mm; mobile
phase, 20% pentane/80% MeOH; flow rate, 1 mL/min; tR
of MDL 201449A ) 22 min). IR (KBr, cm-1): 3500-
3000,1690. MS: m/z (M+) calcd 219.62, obsd 219. Mp
(uncorrected) ) 245 °C. Anal. Calcd for C10H14N5O: C,
43.88; H, 5.89; N, 25.59. Found: C, 44.14; H, 5.78; N,
25.65. Complete analytical data were compared against the
control sample reported in ref 5a.
(1R)-(-)-10-Camphorsulfonic Acid (CSA) Salt Puri-
fication Procedure of 9N-[(1′R,3′R)-trans-3′-tert-Butyldi-
methylsilyloxycyclopentanyl]adenine (20). 20 (4.78 g,
0.014 mol) was dissolved in ethyl acetate (75 mL). (1R)-
(-)-10-Camphorsulfonic acid (3.0 g, 0.013 mol) was added,
and the mixture was stirred vigorously for 18 h. The solid
was filtered off to give 5.79 g of the salt that was 98.17%
pure by HPLC (column, ChiralPak AD 250 × 4.6 mm;
mobile phase, 85% pentane/15% IPA; flow rate, 1 mL/min;
tR of 20 ) 10.2 min). This was dissolved into CH2Cl2 (150
mL) and washed with 2 × 100 mL of 1 N NaOH and then
100 mL of water, and the organic layer was dried over
MgSO4. The drying agent was filtered off and washed with
40 mL of CH2Cl2, and the filtrate was removed at 30 °C/50
Torr to give 3.25 g of 9N-[(1′R,3′R)-trans-3′-tert-butyl-
dimethylsilyloxycyclopentanyl]adenine (20, (71% yield) that
was 98.3% pure by HPLC.
(1R,4S)-(-)-4-Acetoxy-2-cyclopentenol (6). A total of
115.4 g (0.45 mol) of 17, 500 mL (0.50 mol) of TBAF (1.0
M in THF), and 7 mL (0.05 mol) of Et3N were charged to
a 1-L three-necked flask equipped with stirrer and continuous
nitrogen purge. The reaction was monitored by TLC (30%
ethyl acetate/70% heptane; Rf of 17 ) 0.5, Rf of 6 ) 0.1;
the average reaction time was 30 min.) The reaction mixture
was evaporated onto silica gel (500 g) at 25 °C/50 Torr. The
material was loaded onto a column of 1 kg of silica gel and
eluted with 3:7 ethyl acetate/hexane. The fractions contain-
ing product were combined and concentrated at 35 °C/50
Torr to give a brown oil. The product was crystallized from
364
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development

250 mL of TBME and 250 mL of hexane at 0 °C for 24 h.
The crystals which formed were filtered off and washed with
1:1 TBME/hexane. The solid was air-dried to give 32 g
(50%) of (1R,4S)-(-)-4-acetoxy-2-cyclopentenol (6). Com-
plete analytical data were compared against the control
sample reported in ref 1b.
(1R,3S)-(-)-1-acetoxycyclopentan-3-ol (7). A mixture
of 11.2 g (45 mmol) of nickel acetate tetrahydrate, 42 mL
(42 mmol) of 1 M NaBH4 in ethanol, 167.3 g (1.18 mol) of
6, and 2.2 L of ethanol was charged to a 2-gal autoclave.
The reaction mixture was subjected to 100 psi H2 for 4 h.
After H2 uptake had ceased, the catalyst was filtered off on
a pad of Celite and charcoal. A total of 2 mL of Et3N was
added to the filtrate, and the solvent was removed at 50 °C/
50 Torr to give 147 g (87%) of (1R,3S)-(-)-1-acetoxycyclo-
pentan-3-ol (7). Complete analytical data were compared
against the control sample reported in ref 1b.
9N-[(1′R,3′R)-trans-3′-Acetoxycyclopentanyl]adenine (9).
A mixture of 180 g (1.25 mol) of 7, 209.1 mL (1.5 mol) of
Et3N, and 1.5 L of TBME was charged to a 3-L three-necked
flask equipped with stirrer, thermometer, and continuous
nitrogen purge. The reaction mixture was cooled to -10
°C, and 107 mL (1.38 mol) of MsCl was added at such a
rate that the reaction temperature did not exceed 10 °C. The
reaction was complete after 1 h (TLC: 10:90 ethyl acetate/
heptane; Rf of 7 ) 0.0, Rf of mesylate ) 0.70). The reaction
mixture was filtered, the filter cake was washed with 1 L of
TBME, and the filtrate was concentrated at 30 °C/50 Torr
to give 222.3 g of the mesylate.
A total of 202.7 g (1.5 mol) of adenine 2, 60 g (1.5 mol)
of NaH, and 1.5 L of DMA was charged to a 3-L three-
necked flask equipped with stirrer, thermometer, condenser,
and continuous nitrogen purge. The mixture was heated to
70 °C for 1 h and then cooled to 50 °C. A total of 222.3 g
of mesylate was added to the reaction mixture. The resulting
slurry was heated to and maintained at 70 °C for 4 h and
then cooled to room temperature and stirred for 18 h. The
reaction mixture was filtered through Celite, and the filtrate
was concentrated at 50 °C/50 Torr to give the crude product.
The product was purified by column chromatography (1 kg
silica gel column and eluting with 3:7 hexane/ethyl acetate).
The fractions containing product were combined and con-
centrated at 35 °C/50 Torr to give 180 g of crude product,
which was recrystallized two times from MEK (2 L/150 g).
The solid was filtered off, washed with MEK, and air-dried
to give 120 g (36%) of 9N-[(1′R,3′R)-trans-3′-acetoxy-
cyclopentanyl]adenine (9). Complete analytical data were
compared against the control sample reported in ref 1b.
9N-[(1′R,3′R)-trans-3′-Hydroxycyclopentanyl]ade-
nine Hydrochloride (1). A total of 120 g (0.46 mol) of 9,
700 mL of EtOH, and 25 mL of H2O was charged to a 3-L
three-necked flask equipped with stirrer, thermometer,
condenser, and continuous nitrogen purge. A total of 90 mL
of 6 N HCl in ethanol was added to the reaction mixture,
which was heated at 70 °C for 23 h. The reaction mixture
was cooled, and 1 L of 2:1 heptane/ethanol was added. The
mixture was stored at 0 °C for 24 h. The solid which
crystallized was filtered off, washed with 1 L of heptane,
and dried at 35 °C/100 Torr for 6 h. The product was
hydrated in a humid oven for 18 h to give 107 g (85%) of
9N-[(1′R,3′R)-trans-3′-hydroxycyclopentanyl]adenine hydro-
chloride, (1), with 99.5% purity (HPLC: column, ChiralPak
AD 250 × 4.6 mm; mobile phase, 20% pentane/80% MeOH;
flow rate, 1 mL/min; tR of MDL 201449A ) 22 min). Anal.
Calcd for C10H15NO2‚HCl (MW 273.72): C, 43.88; H, 5.89;
N, 25.59. Found: C, 44.18; H, 5.86; N, 25.53. Complete
analytical data were compared against the control sample
reported in ref 5a.
Received for review January 13, 1998.
OP9701245
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
365