A concise synthesis of the HIV-protease inhibitor nelfinavir via an unusual tetrahydrofuran rearrangement

Article abstract of DOI:10.1016/S0040-4039(00)01231-4

An efficient synthesis of nelfinavir 1 was developed. The synthesis features an unusual rearrangement of a 3-amidotetrahydrofuran into a functionalized oxazoline. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4039(00)01231-4

Tetrahedron Letters 41 (2000) 7017±7021
A concise synthesis of the HIV-protease inhibitor nel®navir
via an unusual tetrahydrofuran rearrangement
Scott E. Zook, Juliette K. Busse and Bennett C. Borer*
Chemical Development, Agouron Pharmaceuticals, Inc., 3565 General Atomics Ct., San Diego, CA 92121, USA
Received 28 April 2000; accepted 20 July 2000
Abstract
An ecient synthesis of nel®navir 1 was developed. The synthesis features an unusual rearrangement of a
3-amidotetrahydrofuran into a functionalized oxazoline. # 2000 Elsevier Science Ltd. All rights reserved.
The introduction of the HIV protease inhibitors has greatly improved the treatment of HIV
infection.1 Nel®navir mesylate 1a received FDA approval in March 1997 and is currently the
largest selling protease inhibitor. The combination of large dosing (ca. 2.5 g/day) and structural
complexity create a need for an ecient and cost-e€ective synthesis.
A retrosynthetic analysis of nel®navir (Scheme 1) shows that nucleophilic ring-opening of an
oxazoline 2 using thiophenol can directly a€ord nel®navir 1. Oxazoline 2 can be formed by the
coupling of an epoxide precursor 3 and the commercially available perhydroisoquinoline 4.2 In
the forward sense, this sequence has been carried out by Inaba and co-workers3 in their synthesis
of nel®navir.
We were interested in alternative syntheses of oxazolines of the type shown in structure 3. We
hypothesized that an intramolecular cyclization of a 3-amido-substituted tetrahydrofuran 5 using
an appropriate electrophile could generate oxazolines of the structure 6 (Scheme 2).
* Corresponding author.
0040-4039/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(00)01231-4

7018
Scheme 1.
Scheme 2.
There is some precedence for cyclizations of this type. McGarvey and co-workers4 have
reported that b-hydroxy-oxazolines can be converted under basic conditions into 3-amido-tetra-
hydrofurans and that there can be an equilibrium between the two compounds. Iwasaki, Tamaki
and co-workers5 have demonstrated the cyclization of a 2-endo-acetamidooxanorbornene 7 to
a€ord the oxazoline 8 (Scheme 3). While this was a promising lead, the ring strain and benzylic
nature of the tetrahydrofuran presumably facilitated the cyclization.
In order to test the cyclization on an appropriate substrate for the synthesis of nel®navir, the
amino alcohol 9a was required.
Scheme 3.
This compound was readily available from the meso-epoxide 10 using the Jacobsen asymmetric
desymmetrization via azidotrimethylsilane to a€ord the azide 11, followed by reduction and
hydrolysis6 (Scheme 4). We were able to prepare over 500 g of the free amine 9a using this
procedure.7 The hygroscopic free amine 9a could also be isolated as the HCl salt 9b or the
p-toluenesulfonate salt 9c.
The amino alcohol hydrochloride 9b was then converted to the amide 12,8 corresponding to the
amide sidechain of nel®navir, using the commercially available aryl acid chloride9 13 in an
average yield of 70%. The moderate yield was due to high water solubility and loss upon workup.

7019
Scheme 4.
In the cyclization reaction, we found that reaction of amide 12 with 1.1 equiv. of methane-
sulfonyl chloride and triethylamine (1.2 equiv.) at 0ꢀC followed by acetic anhydride (55 equiv.)
and sulfuric acid (2 equiv.) at room temperature a€orded a 1.3:1 ratio of the desired oxazoline 14
to the fused oxazoline 15 in a combined 49% yield (Scheme 5).
In order to reduce the amount of the fused oxazoline 15 being formed, we examined the
mesylation step. We found that it was advantageous to add the methanesulfonyl chloride ®rst,
followed by less than an equivalent amount of triethylamine (with respect to the methanesulfonyl
chloride). This reduced the amount of fused oxazoline to less than 5% (by HPLC). The formation
of oxazoline 15 was found to be base-induced. Indeed, when the reaction was run using standard
mesylation conditions (addition of methanesulfonyl chloride to compound 12 in the presence of 3
equiv. triethylamine), oxazoline 15 was isolated in a quantitative yield.10 Presumably, the reversed
order of addition prevented the build-up of base during the mesylation step. Attempts to isolate
the mesylate and subject it independently to the acetic acid/sulfuric acid rearrangement were
unsuccessful.
Scheme 5.

7020
We next addressed the moderate yield of amide 12 formation. Since the loss of yield appeared
to come from water solubility, the most straightforward approach would be a telescoped process.
We found that treatment of the p-toluenesulfonate salt 9c with 2.05 equiv. triethylamine and 1.06
equiv. of 3-acetoxy-2-methylbenzoyl chloride 13 gave an excellent in situ yield (by HPLC) of the
amide 12 (Scheme 6). This reaction mixture could then be treated directly with 2.5 equiv. of
methanesulfonyl chloride followed by 1.5 equiv. of triethylamine at 0ꢀC. Without workup, this
reaction mixture was then treated with 15 equiv. of acetic anhydride and 7.5 equiv. of conc.
sulfuric acid, allowing the temperature to rise to 35ꢀC. The ratio and amounts of acetic anhydride
and sulfuric acid were found to be necessary for complete oxazoline formation. A neutralization
and aqueous workup a€orded the oxazoline 14. Treatment of the crude oxazoline 14 with the
perhydroisoquinoline 4 in the presence of aqueous methanol and potassium carbonate gave the
adduct 2 (via the epoxide) which crystallized directly from the reaction mixture. A simple methyl-
isobutyl ketone reslurry a€orded an overall 72% yield of the oxazoline 2 from 9c in greater than
95% purity.11
Scheme 6.
The ring opening of oxazoline 2 by thiophenol occurred in 82% yield as described in Ref. 3,
a€ording nel®navir free base 1 in high purity.
In conclusion, this work represents a very concise and atom-ecient synthesis of nel®navir.
The total synthesis requires three reactors and two reslurries for puri®cation. There is only one
aqueous workup involved.

7021
References
1. Flexner, C. N. Engl. J. Med. 1998, 338, 1281±1292.
2. Available from the Aldrich Chemical Company, Inc.
3. Inaba, T.; Birchler, A. G.; Yamada, Y.; Sagawa, S.; Yokota, Y.; Ando, K.; Uchida, I. J. Org. Chem. 1998, 63,
7582±7583.
4. Wilson, K. J.; Sabat, M.; McGarvey, G. J. J. Org. Chem. 1993, 58, 6180±6181.
5. Yamazaki, H.; Horikawa, H.; Nishitani, T.; Iwasaki, T.; Nosaka, K.; Tamaki, H. Chem. Pharm. Bull. 1992, 40,
102±108.
6. Schaus, S. E.; Larrow, J. F.; Jacobsen, E. N. J. Org. Chem., 1997, 62, 4197±4199.
7. We obtained an average yield of 96% over 14 consecutive reactions with a 98% ee on the ®rst run and >99% ee
on the subsequent reactions. The azide reduction was performed using a mixture of Pd/C and PtO2/H2. The
catalyst was also recycled through the 14 runs.
25
D
8. Spectroscopic data for 12: mp 141ꢀC. ‰ꢀŠ +29.7 (c 1.0, MeOH). H NMR (300 MHz, DMSO-d6) ꢁ 2.03 (s, 3H),
1
2.25 (s, 3H), 3.47 (dd, J=1.8, 9.3 Hz, 1H), 3.55 (dd, J=2.8, 9.0 Hz, 1H), 3.80 (dd, J=4.3, 9.3 Hz, 1H), 3.93 (dd,
J=5.5, 9.0 Hz, 1H), 4.08±4.12 (m, 2H), 5.23 (d, J=3.8 Hz, 1H), 7.06±7.23 (m, 3H), 8.50 (d, J=6.6 Hz, 1H). 13C
(75 MHz, DMSO-d6) ꢁ 12.9, 20.9, 58.4, 71.2, 73.8, 75.1, 123.6, 125.2, 126.8, 127.9, 139.3, 149.6, 168.6, 169.3. Anal.
calcd for C14H17NO5: C, 60.21; H, 6.14; N, 5.02; found: C, 60.03; H, 6.10; N, 4.99.
9. Available from the Austin Chemical Company, Inc. 1565 Barclay Blvd. Bu€alo Grove, IL 60089, USA.
25
10. Spectroscopic data for 15: ‰ꢀŠ +18.2 (c 1.0, MeOH). 1H NMR (300 MHz, DMSO-d6) ꢁ 2.29 (s, 3H), 2.33 (s, 3H),
D
3.57 (dd, J=3.8, 11.1 Hz, 1H), 3.62 (dd, J=5.4, 9.6 Hz, 1H), 3.90 (d, J=9.6 Hz, 1H), 4.08 (d, J=11.0 Hz, 1H),
4.89 (dd, J=5.3, 7.6 Hz, 1H), 5.24 (dd, J=3.7, 7.7 Hz, 1H), 7.24 (dd, J=1.4, 8.0 Hz, 1H), 7.33 (dd, J=8.0, 8.0 Hz,
1H), 7.58 (dd, J=1.3, 7.6 Hz, 1H). 13C (75 Mhz, DMSO-d6) ꢁ 13.8, 20.9, 72.8, 73.8, 74.4, 82.7, 125.2, 126.9, 127.4,
128.8, 138.8, 150.0, 163.1, 169.4. Anal. calcd for C14H15NO4: C, 64.36; H, 5.79; N, 5.36; found: C, 64.15; H, 5.75;
N, 5.32.
11. Representative procedure: The amine salt 9c (25.0 g, 90.8 mmol) and acid chloride 13 (3-acetoxy-2-methylbenzoyl
chloride, 20.4 g, 95.9 mmol) were slurried in ethyl acetate (188 ml) at room temperature. With water bath cooling,
triethylamine (25.9 ml, 186 mmol) was added at a rate to keep the temperature below 25ꢀC. The slurry was stirred
at room temperature for 1 hour 45 min to give a suspension of 12. The mixture was then cooled in an ice/acetone
bath and methanesulfonyl chloride (17.6 ml, 227 mmol) was added in one portion. Triethylamine (19 ml, 136
mmol) was added dropwise at a rate to keep the internal temperature below 10ꢀC. Acetic anhydride (129 ml) was
added in one portion and the cooling bath was removed. Sulfuric acid (98%, 38 ml) was added in three portions at
15-min intervals. The mixture was stirred at room temperature for 17 hours. A suspension of sodium bicarbonate
(305 g) in 1 l of water was prepared. This was overlaid with ethyl acetate (250 ml). The reaction mixture from
above was added to the sodium bicarbonate slurry dropwise over 2 hours. The layers were separated and the
aqueous layer was washed with ethyl acetate (200 ml). The combined organic layers were washed with saturated
sodium bicarbonate (200 ml) and brine (200 ml). The organic layer was dried (MgSO4), ®ltered and evaporated to
give oxazoline 14 as an oil. Then 14 was dissolved in methanol (225 ml) and water (225 ml) was added. Potassium
carbonate (37.6 g, 272 mmol) and perhydroisoquinoline 4 (20.5 g, 86.1 mmol) were added sequentially. The
mixture was heated to 50ꢀC for 5 hours. Water (225 ml) was added to the slurry, which was allowed to cool to
room temperature. The solid 2 was ®ltered, washed with water and dried in a vacuum oven at 42ꢀC. The crude
yield of 2 was 45 g. The crude 2 was slurried at room temperature in methyl-iso-butyl ketone (400 ml) for 30 min
and ®ltered and washed with MIBK (100 ml). The puri®ed 2 was dried in a vacuum oven at 42ꢀC to constant
weight. The yield was 29.9 g, 72%. Spectral data for compound 2 was consistent with that reported in Ref. 3.