Synthesis of a long acting HIV protease inhibitor via metal or enzymatic reduction of the appropriate chloro ketone and selective zinc enolate condensation with an amino epoxide

Article abstract of DOI:10.1002/adsc.201300074

This paper describes a new convergent approach to the synthesis of an HIV protease inhibitor which was designed to be suitable in long acting formulations. Unique features in the synthesis include an asymmetric hydrogenation as well as enzymatic reduction

Full text of DOI:10.1002/adsc.201300074

FULL PAPERS
DOI: 10.1002/adsc.201300074
Synthesis of a Long Acting HIV Protease Inhibitor via Metal or
Enzymatic Reduction of the Appropriate Chloro Ketone and
Selective Zinc Enolate Condensation with an Amino Epoxide
Ioannis N. Houpis,a,* Renmao Liu,b Lin Liu,b Yanfei Wang,b Nengfa Dong,b
Xiangan Zhao,b Yan Zhang,b Tingting Xiao,b Youchu Wang,b Dominique Depre,b
Ulrike Nettekoven,c Michael Vogel,d Rob Wilson,d and Steve Collierd
a
Janssen Pharmaceutica, API Development. Turnhoutseweg 30, 2340 Beerse, Belgium
Fax : (+32)-14-60-575; phone (+32)-14-60-8614; e-mail: yhoupis@its.jnj.com
STA Pharmaceuticals, 288 Fute Zhong Rd., Shanghai 200131, Peopleꢀs Republic of China
Solvias AG, Rçmerpark 2, 4303 Kaiseraugst, Switzerland
Codexis, Penobscot Dr., Redwood City, CA 94063, USA
b
c
d
Received: February 8, 2013; Published online: May 15, 2013
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201300074.
Abstract: This paper describes a new convergent ap- ide and the diastereoselective coupling of the latter
proach to the synthesis of an HIV protease inhibitor with the zinc enolate of a suitable functionalized
which was designed to be suitable in long acting for- amide derivative.
mulations. Unique features in the synthesis include
an asymmetric hydrogenation as well as enzymatic Keywords: asymmetric ketone reduction; enzymatic
reduction of a key chloro ketone intermediate, to set ketone reduction; long acting HIV protease inhibi-
the threo stereochemistry in the corresponding epox- tors; zinc enolate addition to epoxides
Introduction
synthesis of this class of compounds, the need has in-
creased to produce such molecules at even lower
In the last few decades, the treatment of HIV has costs, thus making them accessible to a wider number
made significant advances. The most recent data indi- of patients, especially in developing countries. To ac-
cate that for those patients who comply with the cur- complish this, it has become imperative to devise
rent most successful triple combination therapy, their even better synthetic approaches to these life-saving
life expectancy, compared to that in the mid-1980s, medicines than the innovative methods in the litera-
has increased by an average of 15 years.[1] However, ture.
this is contingent on an early start and adherence to
In this paper, we describe our efforts towards the
the therapeutic regimen. The latter is a particularly convergent synthesis of clinical candidate 1, a protease
pressing problem and coupled with the significant inhibitor specifically designed to allow for long
cost of therapy it continues to be a challenge in con- acting-controlled release formulations.
trolling viral mutation.
In order to address these issues, long acting HIV
protease inhibitors have been considered and are
under development in our laboratories in collabora-
tion with Medivir. One promising discovery pro-
gram[2] has led to a number of potentially successful
clinical candidates. However, the significant structural
complexity remains a challenge to the initial success-
ful preparation (e.g., for clinical development) and
commercialization of such molecules.
Despite the fact that there are a large number of
publications[3] describing successful strategies for the
Adv. Synth. Catal. 2013, 355, 1829 – 1839
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1829

Ioannis N. Houpis et al.
FULL PAPERS
As mentioned above, a number of HIV protease in- approach via an enolate-epoxide opening (Scheme 1).
hibitors have been described in the literature in the In this synthetic strategy, the key features of the syn-
last twenty years with several of them possessing the thesis include the stereocontrolled preparation of the
peptidomimetic motifs described in Figure 1 (A, B,
Scheme 1. Retrosynthetic analysis for the synthesis of 1.
chloro ketone 4 and its challenging reduction to the
threo alcohol 5, and corresponding epoxide 3. As the
latter configuration has been such a challenge in the
literature[7] we decided early on to devise both enzy-
Figure 1. Motifs of HIV protease inhibitors and the pre-
ferred approach towards their synthesis.
matic and asymmetric hydrogenation methodologies
to ensure that the most economically efficient synthe-
sis is implemented from the beginning of the project.
C).[4] Access to these motifs was primarily achieved Finally, the strategic epoxide-enolate opening could
via a nucleophilic opening of epoxides I or II. The be carried out only by use of the Zn-diamine enolate
latter possesses the erythro stereochemistry which is complexes we have reported more than a decade
the most easily accessible via diastereoselective reduc- ago[8] and they have allowed us to form the key C C
ꢀ
tion of the corresponding chloro ketone (e.g., 4).[5] bond with high efficiency. To the best of our knowl-
Synthesis of the threo analogue is more challenging as edge this is the first time where such a synthetic strat-
we will discuss below.
egy was employed with the exception of Crixivan[9]
This epoxide-opening approach is highly successful where the more reactive glycosidyl tosylate was uti-
with heteroatom nucleophiles, such as amines, but it lized successfully. It is noteworthy that in our ap-
is less so with carbon nucleophiles (motif C, Figure 1), proach we can use the thiophene amino alcohol 22 to
such as enolates, due to the lack of reactivity of the control the absolute configuration in a similar fashion
epoxide. To overcome this problem, more circuitous to the now well-established aminoindanol in Crixivan
routes have been discovered (C)III+IV) where the
stereochemistry of the a center can be controlled by
employing kinetic or thermodynamic conditions in Results and Discussion
the alkylation of the lactone enolate depending on
the desired stereochemical outcome.[6]
The preparation of the epoxide 3 started from pro-
Our potential drug candidate (1) combines the tected tyrosine 6, followed by activation of the phenol
most challenging features in this class of molecules, function as the triflate 7 with triflic anhydride in the
namely the less accessible threo stereochemistry of presence of pyridine in high yield (Scheme 2). The ac-
the amino alcohol center and the three stereogenic- tivated intermediate was used without further purifi-
center array found in motif C. Nonetheless, instead of cation in the Suzuki reaction. Unfortunately the pub-
employing the more linear alkylation approach, we lished procedure[10] that worked well with simple aryl-
decided to construct the molecule using a convergent boronic acids failed to give any product with our pyri-
1830
asc.wiley-vch.de
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 1829 – 1839

Synthesis of a Long Acting HIV Protease Inhibitor via Metal or Enzymatic Reduction
Scheme 2. Preparation of the chloro ketone 4.
dylboronic acid derivative 8 and thus an investigation (3.5 equivalents) was added to 9 under a variety of
was undertaken to define the best ligand, solvent, conditions. Control experiments showed that the
base and catalyst precursors for this transformation. chloroacetic acid dianion itself was sufficient to cause
The study revealed that non-aqueous systems alone racemization in both the starting material and the
(MeOH, DME, toluene, DMF, dioxane) were not product. This problem was resolved by utilizing
suitable for the reaction, but upon addition of water, a milder base such as lithium hexamethyldisilazide to
a number of these same solvents resulted in useful deprotonate the carbamate to produce 9a. This depro-
conversions to the coupled product 9. The commer- tonation significantly slowed the epimerization of 9a
cially available, ferrocenyl based catalyst Cl2Pd
DCM and PdCl2-BINAP proved to be the most suc- the chloroacetic acid dianion caused essentially no
cessful and cost-effective catalysts. racemization and, after work up, the product was ob-
The best results were obtained when the reaction tained in 86–89% solution yield with >99.5% ee.
was run in DMF:H2O (2:1) at mild temperatures (60–
ACHTUNGTREN(NUNG dppf)- (and product 4); with this modification, addition of
The reduction of similar systems to 4 has been
658C) in the presence of 3 mol% of the catalyst and described extensively (Scheme 3, where Ar=H).[13]
1.5 equivalents of K2CO3. It is noteworthy that higher Thus, reduction of these unsubstituted derivatives
temperatures and larger amounts of base resulted in with NaBH4 and other hydride reagents proceeds
significant racemization and/or methyl ester hydroly- readily, except that the undesired erythro isomer is in-
sis. Interestingly, under these conditions homocou- variably produced. So, not surprisingly, hydride reduc-
pling of neither of the reaction partners was observed. tions of 4 under the conditions described in the litera-
Similarly, no reduction products were identified ture[13] resulted in good yields of the reduction prod-
ꢀ
ꢀ
(Ar OTf!Ar H). After aqueous work up, and re- ucts with ratios of 5:5a varying from 1:1 to 1:8 in
moval of the residual Pd by extraction with N-acetyl- favour of the undesired isomer. Although the threo
cysteine, the product was isolated in 82% yield by isomer has been prepared before with a variety of
crystallization from heptanes with 97% chemical methods,[13] we desired a more direct route than the
purity and >99.5% enantiomeric purity.
ones described thus far. In order to accomplish this,
The synthesis of analogues of chloro ketone 4, with we undertook one study employing asymmetric hy-
simpler aromatic substituents, has been performed in drogenation methodology and a parallel effort study-
a variety of ways in the literature,[11] most involving ing the enzymatic reduction of 4.
activation of the carboxylic acid corresponding to
In the asymmetric hydrogenation approach,[14] an
ester 9 followed by addition of a variety of chlorome- extensive screening program was initiated that evalu-
thylating reagents. When these methods were at- ated dozens of metal-ligand combinations that are
tempted, significant racemization was observed. normally useful in asymmetric ketone reductions such
Therefore, we decided to employ the methods devel- as the Josiphos, Taniaphos Walphos, and MeoBiphep
oped by Wang and co-workers[12] that utilized the class of ligands in combination with Rh, Ru and Ir
chloroacetic acid dianion directly on the methyl ester catalyst precursors.[15] Representative results for a few
9. Unfortunately, preliminary experiments showed sig- of these diverse classes of ligands (see Figure 2) are
nificant racemization (>15%) when the dianion shown in Table 1.
Adv. Synth. Catal. 2013, 355, 1829 – 1839
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1831

Ioannis N. Houpis et al.
FULL PAPERS
Scheme 3. Preparation of the epoxide 3.
was not optimized further as the substrate to catalyst
(S/C) ratio could never be improved beyond S/C=50.
Consequently we focused on the more promising
Rh-based catalysts in combination with the Mandy-
phos ligand 17 (Table 1, entries 7 and 8) which
showed higher reactivity and generated fewer by-
products in the initial screens.
An investigation into the factors affecting the reac-
tivity and selectivity of the catalyst (Table 2) revealed
that the steric demand of the solvent had the most
significant effect on the selectivity of the reduction.
Thus, tert-amyl alcohol gave the highest selectivity
(Table 2, entry 5) compared to smaller alcohols
(Table 2, entries 1 and 2). Non-protic solvents were
detrimental to the reaction while acid or amine addi-
tives showed no effect. The reaction could be scaled
up successfully on a multikilogram scale, but higher
pressures (ca. 100 bar) were required to achieve com-
mercially useful substrate to catalyst ratios (Table 2,
entry 7).
Not surprisingly, the reaction is air sensitive and at
the desired high S/C ratios the quality of the starting
chloro ketone was of paramount importance (Table 2,
entries 8 and 9). Even small amounts of contaminants
(e.g., chloride ion) caused incomplete conversions.
Moreover, relatively high pressures (80–100 bar) were
required to achieve good selectivity limiting the
choice of equipment that could be used for this trans-
Figure 2. Ligands used in the screening
Initial results rapidly showed that both ruthenium formation.
and iridium catalysts were not useful for this transfor-
For these reasons the enzymatic reduction using the
mation. The former gave predominantly the methyl ketoreductase family of enzymes, which employ
ketone by-product from reduction of the C Cl bond. NADH or NADPH as a cofactor, was studied.[16] Or-
ꢀ
Moreover, even in cases where useful amounts of 5 dinarily, the high cost of the cofactor would lead to
were formed (Table 1, entry 1) the selectivity was prohibitive costs for this transformation; however, the
rather low. Most iridium catalysts examined gave low cofactor can be regenerated in situ by the same
conversions (Table 1, entries 3 and 4) with one nota- enzyme using isopropyl alcohol as the stoichiometric
ble exception (Table 1, entry 2). However this system reductant (Figure 3).
1832
asc.wiley-vch.de
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 1829 – 1839

Synthesis of a Long Acting HIV Protease Inhibitor via Metal or Enzymatic Reduction
Table 1. Homogeneous hydrogenation of ketone 4.
Entry
Metal/L[a]
Ligand/Solvent
Temperature/Pressure[b]
Conversion/A [%] 5+5a[c]
Ratio 5:5a
1
2
3
4
5
6
7
8
9
N
E
10/i-PrOH
12/EtOH
16/EtOH
13/MeOH
15/TFE
14/MeOH
17/iPrOH
17/MeOH
11/MeOH
40/40
50/20
50/20
50/20
50/20
40/40
45/30
40/30
40/40
97%/54%[d]
98%/98%
22%/22%
18%/10%
100%/90%
72%/70%
99%/88%
79%/72%
98%/90%
73:27
80:20
84:16
54:46
20:80
52:48
88:12
82:17
60:40
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
A
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
[a]
For screening experiments an S/C ratio of 50 was used. M:L molar ratio was 1:1.
Temperature in oC and pressure in bar.
Conversion expressed as total of products/starting material. A [%] indicates the relative abundance of 5+5a in the
screening and is expressed as their total HPLC area % minus the by-products.
45% of the dechlorinated product was formed.
[b]
[c]
[d]
Figure 3. Enzymatic reduction.
Table 2. Optimization of the reduction of ketone 4 with 17 and [RhACTHNUGTRNEUNG(ndb)2]BF4 catalyst system.
Entry
Solvent/Additives[a]
Temperature [8C]/Time [h]
Pressure [bar]
S/C ratio.
Ratio 5:5a
1
2
3
4
5
6
7
8
EtOH
i-PrOH
40/18
45/29
40/20
40/20
40/20
30/20
30/20
30/20
30/20
30/20
30/20
30
30
40
40
40
80
100
80[c]
80[d]
80
50
50
50
50
85:15
88:12
89:11
NR[b]
95:5
94:6
95:5
95:5
90:10
95:5
i-PrOH/AcOH
i-PrOH/toluene
tert-amyl alcohol
tert-amyl alcohol
tert-amyl alcohol
tert-amyl alcohol
tert-amyl alcohol
tert-amyl alcohol/AcOH
50
200
300
300
200
200
200
9
10
11
tert-amyl alcohol/
(i-Pr)2NEt
80
94:6
[a]
10 equiv. of AcOH or (i-Pr)2NEt vs. Rh were added. The i-PrOH:toluene ratio was 1:1.
None of the desired products were produced although 4 was completely consumed.
The reaction was run in the presence of air. Conversion was 69%.
[b]
[c]
[d]
Starting material contained 2–3% of chloride ion. Conversion was 79%.
The initial evaluation tested 288 NADPH-depen- gave either low reactivity or high selectivity for the
dent enzyme mutants (3 sets of 96)[17] and was per- undesired diastereomer 5a (Table 3 entries 1–10).
formed by dissolving the substrate in isopropyl alco- Only two enzymes gave the desired 5 in excellent se-
hol-THF and exposing it to the enzyme and cofactor lectivity and although the conversion was not high, it
in pH 7 buffer. Some representative results are shown was sufficient to warrant further investigation
in Table 3. At first the results seemed rather disap- (Table 3 entries 11 and 12).
pointing as the majority of enzymes tested (>280)
Adv. Synth. Catal. 2013, 355, 1829 – 1839
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1833

Ioannis N. Houpis et al.
FULL PAPERS
Table 3. Enzymatic reduction of ketone 4 with Codexis ketoreductases.
Entry
Enzyme[a]
Temperature [oC]
pH
Conversion
5:5a
1
2
3
4
5
6
7
8
P1A1
P1A2
P1B1
P1B2
P1F4
P2D1
P2G9
P2H9
25
25
25
25
25
25
25
25
25
25
25
25
25
45
45
45
7
7
7
7
7
7
7
7
7
7
7
7
7
8.5
9
99.1%
94%
0.7:99.3
1:99
3.6:96.4
1:99
1:99
3:97
2:98
4:96
83:17
94:6
96.6:3.4
99:1
100:0
>99.5:0.5
>99.5:0.5
> 99.5:0.5
32.3%
92.9%
65.2%
33.1%
40%
18.8%
2%
4%
20%
22%
99.7%
>97%
> 97%
>97%
9
P3A2
P3B3
10
11
12
13
14[b]
15[b]
16[b]
P3C12
P3G6
CDX-022
CDX-022
CDX-022
CDX-022
9.5
[a]
Enzyme kits available from Codexis. Enzyme codes represent specific ketoreductase mutants. Conditions: 2.4 mg 4,
0.01 mg cofactor (NADP+ for entries 1–12, NAD+ for entry 13), 100 mg enzyme lysate, i-PrOH-THF-0.1M phosphate
buffer pH 7 (50 mL+15 mL+100 mL), overnight at 258C.
[b]
2 wt/wt% CDX-022, 0.5 wt/wt% NAD+, i-PrOH-0.1M triethanolamine buffer (10/90), overnight at 458C.
Finally, after careful investigation of mutants relat- ther adjustment of the pH was needed apart from
ed to the initial leads, enzyme CDX-022 was selected buffer additions. Indeed the enzyme performs well be-
for further development (Table 3, entry 13). The opti- tween pH 8 and 9.5 affording reproducible rates and
mization studies evaluated the substrate concentra- high yields of the desired compound in good chemical
tion, enzyme and cofactor loads, temperature, pH, co- yield (82%) and excellent diastereoselectivity
solvent amount and stirring rate.[18]
(>99.8:0.2).
These studies revealed the following trends. Opti-
The key intermediate, epoxide 3, can be obtained
mal substrate concentration was approximately 160 g from either the metal-catalyzed or enzymatic process
per L in a mixture of pH 9 buffer and isopropyl alco- in high yield by exposing 5 to aqueous potassium hy-
hol (5: 1). There was little variability in the rate of droxide in tert-amyl alcohol followed by crystalliza-
the reaction when 10–20 volume% of isopropyl alco- tion from heptanes (Scheme 3).
hol was used but incomplete reactions were observed
The optically pure amino alcohol (R,R-22), needed
with 5 volume% isopropyl alcohol. Interestingly, at for the convergent construction of 1, was prepared
this concentration the reaction mixture becomes more using the route shown in Scheme 4. Apart from being
viscous as the reaction nears completion necessitating essential for the biological activity of 1, this amino al-
an increase in the agitation rate (from 150 to cohol functions as an effective chiral auxiliary, when
250 rpm) to achieve complete conversions. The reac- it is incorporated in 2, determining the selectivity of
tion performs well with enzyme loads between 0.5 the addition of the enolate of 2 to the epoxide 3
and 5 weight% with no differences in the rate or con- (Scheme 5) in a similar fashion as the aminoindanol
version. To ensure robustness 1 weight% of the moiety in the synthesis of Crixivan. Thus the ketone
enzyme was used along with 0.5 weight% of NAD+. 18 was brominated selectively at the alpha aliphatic
Higher temperatures were tolerated well by the position, avoiding any aromatic bromination by-prod-
enzyme and resulted in increased reaction rates with ucts, by using CuBr2 in refluxing EtOAc to give rac-
the reaction requiring 6 h for completion at 508C. As 19. An NaBH4 reduction and Ritter reaction sequence
the enzyme denatured above that temperature, the re- afforded the racemic cis-amino alcohol 21. The opti-
action was performed at 40–458C. The pH was ex- cally active derivative (R,R-22) was obtained in good
pected to be a significant factor in the reaction; how- yield via a resolution with S-mandelic acid. The opti-
ever, the enzyme proved to be quite stable within cally active amide 2 was prepared via well established
a reasonable pH range (Table 3, entries 14–16). For methods.[19]
example, during scale-up, the initial value of pH 9,
The results for the critical coupling reaction that es-
upon mixing of the components of the reaction, tablished the last stereocenter of the molecule are
quickly fell to pH 8.4 as the reaction proceeded with- shown in Scheme 5 and Table 4. Attempts to repeat
out any meaningful drop in enzyme activity. No fur- the original literature procedure[20] of deprotonating 2
1834
asc.wiley-vch.de
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 1829 – 1839

Synthesis of a Long Acting HIV Protease Inhibitor via Metal or Enzymatic Reduction
Scheme 4. Preparation of the chiral amino alcohol 22.
Scheme 5. End-game coupling.
with n-BuLi followed by addition of 3 at low tempera- enolate at ꢀ788C. Warming up of the reaction mix-
ture failed to afford any product, confirming our sus- ture did result in the formation of the desired product
picion that 3 was too unreactive to be opened by an in about 40% yield. Even this modest yield was not
Table 4. Strategic coupling to produce 23.
Entry
Base[a]
Solvent
Temperature [oC]
Additive
Conversion/Yield [%][e]
1
2
3
4
5
6
7
8
n-BuLi
n-BuLi
THF
THF
THF
THF
THF
THF
MeTHF
DME
PhCH3
THF
ꢀ78
ꢀ25
ꢀ25
ꢀ25
ꢀ78
ꢀ25
ꢀ25
ꢀ25
ꢀ25
ꢀ25
ꢀ25
ꢀ5
–
–
–
–
–
–
–
–
0/0
80/40
30/14
0/0
KHMDS
tert-amylOK
LiHMDS
LiHMDS
LiHMDS
LiHMDS
LiHMDS
LiHMDS
LiHMDS
LiHMDS
LiHMDS
LiHMDS
0/0
89/59
74/58
35/27
77/58
92/67
77/42
99/80f)
48/25
30/15
9
–
10
11
12
13
14
ZnCl2
BF3-OEt2
ZnCl2-TMEDA
ZnCl2-TMEDA
ZnCl2-TMEDA
THF
THF[b]
THF[c]
THF[d]
ꢀ5
ꢀ5
[a]
The amide 2 was added to the base at ꢀ788C followed by 3 and warm up to the reaction temperature.
The base was added to a mixture of 2, 3 and 0.5 equiv. ZnCl2-TMEDA at ꢀ658C followed by warm up to ꢀ58C.
The base was added to a mixture of 2, 3 and ZnCl2-TMEDA at ꢀ308C followed by warm up to ꢀ58C.
The base was added to a mixture of 2, 3 and ZnCl2-TMEDA at ꢀ58C.
[b]
[c]
[d]
[e]
[f]
The yield was determined by quantitative HPLC against a standard.
Isolated yield.
Adv. Synth. Catal. 2013, 355, 1829 – 1839
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1835

Ioannis N. Houpis et al.
FULL PAPERS
peratures, inferior yields were obtained. It appeared
that now 3 was decomposing and we presume that
ꢀ
N H deprotonation needs to occur at lower tempera-
ture to avoid side reactions.
The reaction appears to be highly diastereoselective
with less than 0.5–1% of any of the undesired diaste-
reomer detectable by HPLC analysis. The various iso-
mers have been all synthesized independently by em-
ploying the erythro epoxide and/or the unnatural tyro-
Figure 4. By-products of lithiation of 2.
reproducible upon scale up and the reaction mixture sine isomer. The isomer at the alpha position to the
contained a number of impurities that made purifica- amide was prepared by employing thermodynamic al-
tion difficult. Control experiments showed that 2 was kylation conditions as described above (Figure 1: III+
not stable in the reaction medium upon warming IV and ref.[6]).
above ꢀ40 to ꢀ308C although 3 was stable under the
With the key intermediate 23 in hand, the synthesis
reaction conditions even at 08C overnight. Quenching of 1 proceeded uneventfully by deprotection of the
experiments confirmed our hypothesis that this insta- BOC group and coupling to the methyl carbamate
bility was the result of lithiation of the fluorobenzene tert-butylglycine as described previously and is
and thiophene moiety in 2 (Figure 4) at the higher described in detail in the Experimental Section.[2]
temperatures needed to open the epoxide. This com-
petition between the decomposition of 2 and its reac-
tivity towards 3 appears to be general in a variety of Conclusions
solvents (Table 4). Milder bases such as lithium hexa-
ACHTUNGTRENNUNGmethyldisilazide (LiHMDS) did appear to offer some In conclusion, we have succeeded in the preparation
benefit although the yield was still modest and depen- of 1, employing a convergent approach which utilized
dent on the rate of warming up of the reaction mix- the asymmetric reduction of the chloromethyl ketone
tures (Table 4, entries 5–9).
4 to afford the threo stereochemistry by using new en-
In order to facilitate the reaction, we sought to en- zymatic and asymmetric hydrogenation methodolo-
hance the reactivity of the epoxide by addition of gies. Moreover, we succeeded in the efficient con-
Lewis acids so that the enolate could effect the open- struction of the central core by the opening of epox-
ing of 3 at lower temperatures where the former is ide 3 with the Zn-diamine enolate of 2 in good overall
still stable. Al- and Ti-based Lewis acids were investi- yield and excellent selectivity.
gated first, but no reaction was observed at low tem-
perature while decomposition of 2 and 3 occurred
upon warming of the mixture. Finally (Table 4, en- Experimental Section
tries 10 and 11), addition of ZnCl2 showed some im-
All yields refer to isolated purified products unless other-
wise stated. All solvents used were of normal bulk solvent
purity and were used without further purification. Thorough
degassing was executed as described below.
provement but it was the combination of ZnCl2 with
TMEDA that afforded good yield and reproducible
reaction on 10-kg scale. We have utilized zinc-diamine
complexes in conjunction with enolates in the past
and found them to enhance significantly the reactivity
of the (ester) enolate while affording enhanced stabil-
ity at higher temperature.[21]
Preparation of Chloro Ketone 4
AHCTUNGTERN(NGUN i-Pr)2NH (74.7 g, 738.4 mmol) and THF (340 g) were
This trend appears to hold with the amide enolate
of 2. For example, control experiments showed that
the enolate is stable for up to 24 h in the presence of
0.5 equivalents of ZnCl2-TMEDA at ꢀ58C in THF. In
charged to a 1-L, 3-necked, round-bottom flask (R1) under
a nitrogen atmosphere. The mixture was cooled to 08C and
n-BuLi (204.0 g, 750.1 mmol) was added dropwise. The mix-
ture was cooled at ꢀ60 to ꢀ658C for 1 h. Then
the absence of the complex, the enolate of 2 is not ClCH2COOH (27.7 g, 293 mmol) was added dropwise as
a solution in THF (168 g) over 75 min so that the tempera-
stable above ꢀ308C as mentioned above. The best re-
ture is maintained at ꢀ60 to ꢀ658C. The mixture was stirred
at ꢀ60 to ꢀ658C for 2 h. To another nitrogen-inerted 2-L. 3-
necked, round-bottom flask (R2), 9 (45.3 g, 117.2 mmol) and
THF (252.0 g) were charged. The contents of R2 were
cooled to ꢀ108C and treated with LiHMDS (104.0 g,
117.2 mmol) dropwise over 20 min. The mixture, in R2, was
stirred for 1 h at ꢀ108C. The mixture in R1, was treated
with the solution in R2 over 35 min maintaining the internal
temperature in R1 between ꢀ5 and ꢀ108C. The resulting
sults were obtained by mixing 2 and 3 and solid
ZnCl2-TMEDA in THF at ꢀ658C followed by addi-
tion of LiHMDS. After 1 hour, the mixture was
warmed to ꢀ58C for 16 h to achieve complete conver-
sion and 80% isolated yield of 23 after heptanes-ethyl
acetate crystallization. Although the rate of warming
up was not critical to the success of the reaction, the
base did have to be added at lower temperature. If
the initial addition of the base occurred at higher tem- mixture in R2 was stirred for 1 h at ꢀ108C. Upon comple-
1836
asc.wiley-vch.de
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 1829 – 1839

Synthesis of a Long Acting HIV Protease Inhibitor via Metal or Enzymatic Reduction
tion of the reaction, as judged by HPLC analysis, the reac-
tion was quenched by adding a solution of acetic acid
(117.8 g, 1.96 mol) in THF (380.5 g) over 1 h at ꢀ10 to
ꢀ58C, followed by addition of water (200 g). The organic
layer was washed with saturated NaHCO3 solution (571.4 g)
at 0–108C, followed by saturated NaCl solution (340 g) at
20–258C. The organic layer was concentrated to ~150 mL,
isopropyl alcohol (742.9 g) was added and the mixture was
distilled under reduced pressure. This sequence was repeat-
ed 3 times until GC analysis indicated that the THF was re-
moved (<0.5% by volume). Precipitation of the product
could be observed during this process. The resulting slurry
was cooled to 10–208C and stirred for 16 h. The mixture was
filtered and washed with isopropyl alcohol (30 mL). The
cake was dried under vacuum at 308C for 17 h to afford 4 as
100 mL. n-Heptane (240 mL) was added dropwise, while
cooling to 0–58C, to form a suspension. The mixture was
stirred for 2 h at 0~58C; then filtered and washed with n-
heptane (60 mL). The cake was dried under vacuum at 458C
for 20 h to affordd 5 as a white solid; yield: 24.6 g (14.8
1
mmol). H NMR: (CDCl3, 400 MHz): d=8.38 (d, J=2.4 Hz,
1H), 7.79 (dd, J=2.4 Hz, J’=8.4 Hz, 1H), 7.48 (d, J=
8.0 Hz, 2H), 7.33 (d, J=7.6 Hz, 2H), 6.83 (d, J=8.4 Hz,
1H), 4.99 (d, J=8.4 Hz, 1H), 3.99 (s, 3H), 3.94 (m, 1H),
3.81 (m, 1H), 3.59 (d, J=6.4 Hz, 2H), 3.34 (s, 1H), 3.04 (m,
2H), 1.43 (s, 9H) 4.26 (d, J=16 Hz, 1H), 4.13 (dd, J=
16 Hz, 1H), 4.01 (s, 3H), 3.19 (m, 1H), 3.06 (m, 1H), 1.42
(s, 9H); 13C NMR (CDCl3): d=163.5, 156.1, 144.8, 137.3,
137.0, 136.2, 129.9, 129.8, 126.8, 110.7, 79.9, 71.5, 53.9, 53.5,
47.7, 38.0, 28.3; HR-MS: m/z=407.1721, calcd. for
C21H27ClN2O4 [M+H]+: 407.1732.
1
a white solid; yield: 40.82 g (100.8 mmol). H NMR (CDCl3,
400 MHz): d=8.38 (d, J=2.0 Hz, 1H), 7.79 (dd, J=2.4 Hz,
J’=8.4 Hz, 1H), 7.50 (d, J=8.0 Hz, 2H), 7.26 (d, J=8.0 Hz,
2H), 5.13 (d, J=6.4 Hz, 1H), 4.73 (d, J=6.4 Hz, 1H), 4.26
(d, J=16 Hz, 1H), 4.13 (dd, J=16 Hz, 1H), 4.01 (s, 3H),
3.19 (m, 1H), 3.06 (m, 1H), 1.42 (s, 9H); 13C NMR (CDCl3):
d=201.2, 163.7, 155.2, 144.9, 137.2, 136.9, 134.8, 129.8, 129.4,
127.0, 110.8, 80.5, 58.3, 53.5, 47.3, 37.1, 28.2; HR-MS: m/z=
405.1577, calcd. for C21H25ClN2O4 [M+H]+: 405.1576.
Condensation of Epoxide 3 with Amide 2
A nitrogen-inerted, 1-L, 3-necked, round-bottom flask was
charged with anhydrous ZnCl2 (18.9 g), TMEDA (16.2 g)
and anhydrous THF (2500 g). The mixture was stirred for
1 h at 20–308C. A solution of 2 in THF solution (neat 158 g,
64.6 wt%, 428 mmol) was charged under nitrogen followed
by cooling to ꢀ60 to ꢀ708C and stirring for 30 min. A solu-
tion of 3 in THF solution (neat 149 g, 64.6 wt%, 416 mmol)
was charged under nitrogen at ꢀ60 to ꢀ708C; followed by
LHMDS (1M solution, 988 g) added dropwise at ꢀ60 to
ꢀ708C. The mixture was stirred for 2 h at ꢀ60 to ꢀ708C,
then 10 h at ꢀ3 to ꢀ88C. After the reaction was finished, it
was quenched with 2N hydrochloric acid at ꢀ3 to ꢀ88C to
pH 7–8. Heptane (500 g) was charged and the mixture was
washed with water (1000 g). The aqueous layer was extract-
ed with THF (400 g) and heptane (100 g). The combined or-
ganic layer was concentrated to 1500 mL under vacuum
below 508C. n-Heptane (300 mL) was added in portions,
then the mixture was concentrated to 1500 mL under
vacuum below 508C. Ethyl acetate (200 g) was charged and
the mixture was concentrated to 600 mL under vacuum
below 508C. The mixture was cooled to 15–208C slowly and
stirred for 1 h. The suspension was filtered and washed with
n-heptane (200 g). The wet cake was dried under vacuum at
40–458C for 12 h to yield 23 as a white solid; yield: 242 g
(332.8 mmol). Due to rotomers, precise NMR assignments
were not useful. The final assignment and structure confir-
mation occurred at the final API (1). HR-MS: m/z=
730.3351, calcd. for C41H48FN3O6S [M+H]+: 730.3321.
Reduction of Chloro Ketone 4 via Asymmetric
Hydrogenation
[RhACHTUNGTRENNUNG(nbd)2]BF4 (760 mg, 2.04 mmol) and ligand 17 (2360 mg,
2.24 mmol) were placed in a Schlenk flask, under an argon
atmosphere (3 vacuum/argon cycles). Degassed 2-methyl-2-
butanol (500 mL) was added and the mixture was stirred for
30 min at room temperature. (S)-4 (82.5 g, 204 mmol) was
placed into the 300-mL autoclave (which had passed a tight-
ness test) under an argon atmosphere. Degassed 2-methyl-2-
butanol was added (1500 mL) under argon and the solution
was treated with the catalyst solution from the Schlenk flask
via cannula into the autoclave under a slight pressure of
argon. The mixture in the autoclave was flushed 3 times
with 5 bar of argon and 3 times with 5 bar of hydrogen. The
hydrogen pressure was set to 80 bar and the mixture was
heated to 308C. The resulting over-pressure was adjusted
down to 80 bar and the reaction mixture was allowed to stir
for 40 h, at which time HPLC analysis showed full conver-
sion. The autoclave was cooled, vented, flushed with 5 bar
of argon and unloaded. The crude product mixture of 5 was
crystallized from 2:1 heptane-MTBE or carried forward to
the next step; yield: ca. 70 g (173.4 mmol)).
Preparation of 1 via Deprotection of 23 and Con-
densation with (S)-2-(Methoxycarbonylamino)-
3,3-dimethylbutanoic
Enzymatic Reduction of Chloro Ketone 4
The chloro ketone 4 (30 g, 74.1 mmol) was charged under an
inert atmosphere into a 1-L, 3-necked flask, followed by
TEA buffer (triethanolamine solution in water, adjusted to
pH 8.5–9.5 with 2M HCl, 240 mL) and isopropyl alcohol
(45 mL). The mixture was stirred for 30 min at 25–308C and
A mixture of 23 (42.6 g, 58.43 mmol) and anhydrous isopro-
pyl acetate (680 mL) were charged in an inerted round-
bottom flask. HCl (45 g, 8.0 equiv., 12M in water) was
added dropwise over 30 min while cooling to 5–108C, and
the mixture was stirred for 2 h at that temperature. Water
(639 mL) was added over 1 h at 5–108C to form a 2-phase
solution and the mixture was stirred at 50–558C for 8 h.
After the reaction had finished, the mixture was separated
at 20–258C and 30% NaOH solution was added dropwise to
the aqueous layer over 1 h at 20–258C until the pH reached
+A
b-NAD CHTUNGTRENNUNG(NAD) (0.15 g) solution in TEA buffer (6 mL),
KRED CDX-022 (0.6 g) solution in TEA buffer (30 mL)
and TEA buffer (120 mlL) were added in sequence at 25–
308C. The mixture was stirred for 20 h at 40–458C. After
the reaction was finished, the mixture was extracted with
MTBE (2ꢂ150 mL) at 25–308C. The organic layer was fil-
tered over celite. The organic layer was concentrated to
Adv. Synth. Catal. 2013, 355, 1829 – 1839
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1837

Ioannis N. Houpis et al.
FULL PAPERS
8–9. The suspension was filtered and washed with water
(200 g). The wet cake was dried under vacuum at 40–508C
for 18 h to yield the free amine as a white solid which was
dissolved in DMF (500 mL) and was added to a separate
flask, containing (S)-2-(methoxycarbonylamino)-3,3-dime-
thylbutanoic acid (38.5 g,), HATU (77.3 g) and DMF
(800 mL) and Et3N (39.5 g). After 30 min the reaction was
judged complete by HPLC analysis and water (2.5 L) was
added at 25–308C. The suspension was stirred for 1 h at 25–
308C and filtered and washed with water (2 L). The wet
cake was suspended in ethanol (2 L) and the slurry stirred
for 3 h at 20–258C. Then the mixture was filtered again and
washed with ethanol (200 mL). The wet cake was dried
under vacuum for 12 h at 40–458C to afford 1 as a white
solid; yield: 26.97 g (35.44 mmol). 1H NMR: (DMSO,
400 MHz): d=8.41 (d, J=2 Hz, 1H), 7.94 (dd, J=2.4 Hz,
J’=4.8, 1H), 7.71 (d, J=8.8 Hz, 1H), 7.62 (d, J=8 Hz, 1H),
7.49 (d, J=8 Hz, 2H), 7.29 (d, J=8.4 Hz„ 2H), 7.25 (m,
2H), 7.15 (d, J=5.2 Hz, 1H), 7.11 (m, 2H), 6.89 (d, J=
8.8 Hz, 1H), 6.73 (d, J=8 Hz, 1H), 6.63 (d, J=4.8 Hz, 1H),
4.84 (d, J=5.6 Hz, 2H), 4.65 (d, J=4 Hz, 1H), 3.94 (m,
2H), 3.88 (s, 3H), 3.76 (d, J=3.6 Hz, 1H), 3.55 (m, 1H),
3.49 (s, 3H), 2.87 (m, 4H), 2.70 (m, 3H), 1.95 (m, 1H), 1.79
(m, 1H), 1.69 (m, 1H), 1.43 (m, 1H), 0.80 (s, 9H); 13C NMR
(CDCl3): d=174.4, 170.3, 163.3, 162.2, 159.8, 156.8, 144.8,
139.0, 137.7, 136.1, 135.7, 134.8, 131.7, 131.6, 130.2, 129.7,
128.5, 128.4, 127.6, 126.9, 126.3, 124.5, 122.8, 115.5, 115.3,
111.0, 69.0, 66.2, 63.2, 62.9, 55.4, 53.7, 51.8, 49.1, 43.1, 36.8,
34.1, 32.1, 29.2, 27.1, 20.7; HR-MS: m/z=761.3392, Calcd
for C41H49FN4O7S [M+H]+: 761.3379.
B. K. Handa, J. Kay, A. Krohn, R. W. Lambert, J. H.
Merrett, J. S. Mills, K. E. B. Parkes, S. Redshaw, A. J.
Ritchie, D. L. Taylor, G. J. Thomas, P. J. Machin, Sci-
ence 1990, 248, 358; b) W. Goehring, S. Gokhale, H.
Hilpert, F. F. Roessler, M. Schlageter, P. Vogt, Chimia
1996, 50, 532; c) S. W. Kaldor, V. J. Kalish, J. F. Davies
II, B. V. Shetty, J. E. Fritz, K. Appelt, J. A. Burgess,
K. M. Campanale, N. Y. Chirgadze, D. K. Clawson,
B. A. Dressman, S. D. Hatch, D. A. Khalil, M. B. Kosa,
P. P. Lubbehusen, M. A. Muesing, A. K. Patick, S. H.
Reich, K. S. Su, J. H. Tatlock, J. Med. Chem. 1997, 40,
3979; d) G. R. Painter, S. Ching, D. Reynolds, M. St.
Clair, B. M. Sadler, M. Elkins, R. Blum, R. Dornsife,
D. J. Livingston, J. A. Partaledis, S. Pazhanisamy, R. D.
Tung, M. Tisdale, Drugs Future 1996, 21, 347; e) L. A.
Sorbera, L. Martin, J. Castaner, R. M. Castaner, Drugs
Future 2001, 26, 224.
[6] A. K. Ghosh, S. P. McKee, W. J. Thompson, J. Org.
Chem. 1991, 56, 6500. Also see ref.[2] for a detailed de-
scription of the synthesis of 1 following this methodolo-
gy.
[7] Several methods have been described for the direct
and indirect synthesis of the threo isomer: a) G. Bold,
A. Fassler, H.-G. Capraro, R. Cozens, T. Klimkait, J.
Lazdins, J. Mestan, B. Poncioni, J. Rosel, D. Stover, M.
Tintelnot-Blomley, F. Acemoglu, W. Beck, E. Boss, M.
Eschbach, T. Hurlimann, E. Masso, S. Roussel, K.
Ucci-Stoll, D. Wyss, M. Lang, J. Med. Chem. 1998, 41,
3387; b) X. Rabasseda, J. Silvestre, J. Castaner, Drugs
Future 1999, 24, 375; c) H. Nogami, M. Kanai, M. Shi-
basaki, Chem. Pharm. Bull. 2003, 51, 702; d) P. Rad-
datz, A. Jonczyk, K.-O. Minck, C. J. Schmitges, J. Som-
broek, J. Med. Chem. 1991, 34, 3267; e) J. Barluenga, B.
Baragana, J. M. Concellon, J. Org. Chem. 1995, 60,
6696. Also see ref.[3] for more detailed references.
[8] T. Mase, I. N. Houpis, A, Akao, J. Dorziotis, K. Emer-
son, T. Hoang, T. Iida, T. Itoh, K. Kamei, S. Kato, Y.
Kato, M. Kawasaki, F. Lang, J. Lee, J. Lynch, P. Mali-
gres, A. Molina, T. Nemoto, S. Okada, R. Reamer, J. Z.
Song, D. Tschaen, T. Wada, D. Zewge, R. P. Volante,
P. J. Reider, K. Tomimoto, J. Org. Chem. 2001, 66,
6775.
References
[1] M. May, M. Gompels, V. Delpech, K. Porter, F. Post,
M. Johnson, D. Dunn, A. Palfreeman, R. Gilson, B.
Gazzard, T. Hill, J. Walsh, M. Fisher, C. Orkin, J. Ains-
worth, L. Bansi, A. Phillips, C. Leen, M. Nelson, J. An-
derson, C. Sabin, Br. Med. J. 2011, 343.
[2] B. Kesteleyn, K. Amssoms, W. Schepens, G. Hache, W.
Verschueren, W. Van De Vreken, K. Rombauts, G.
Meurs, P. Sterkens, B. Stoops, L. Baert, N. Austin, J.
Wegner, C. Masungi, I. Dierynck, S. Lundgren, D. Jçns-
son, K. Parkes, G. Kalayanov, H. Wallberg, ꢃ. Rosen-
quist, B. Samuelsson, K. Van Emelen, J. W. Thuring,
Bioorg. Med. Chem. Lett. 2013, 23, 310.
[9] D. Askin, K. K. Eng, K. Rossen, R. M. Purick, K. M.
Wells, R. P. Volante, P. J. Reider, Tetrahedron Lett.
1994, 35, 673.
[10] S.-C. Wen, J. A. Carlson, J. Org. Chem. 1992, 57, 379.
[11] For a comprehensive review on the synthesis of amino-
alkyl halomethyl ketones, see: M. R. Reeder, R. M.
Anderson, Chem. Rev. 2006, 106, 2828.
[3] For a detailed review on the industrial syntheses of
antiACHTUNGTRENNUNGretroviral medicines see: K. Izawa, T. Onishi,
Chem. Rev. 2006, 106, 2811.
[12] X. Wang, J. K. Thottathil, R. P. Polniaszek, Synlett 2000,
902.
[4] a) P. J. Ala, C.-H. Chang, in: Enzymes and Their Inhibi-
tion, (Eds. H. J. Smith, C. Simons), CRC Press LLC,
Boca Raton, FL, 2005, p 268; b) H.-G. Capraro, G.
Bold, A. Frꢄssler, R. Cozens, T. Klimkait, J. Lazdins, J.
Mestan, B. Poncioni, J. L. Rçsel, D. Stover, M. Lang,
Arch. Pharm. Pharm. Med. Chem. 1996, 329, 273;
c) W. J. Thompson, P. M. D. Fitzgerald, M. K. Holloway,
E. A. Emini, P. L. Darke, B. M. McKeever, W. A.
Schleif, J. C. Quintero, J. A. Zugay, T. J. Tucker, J. E.
Schwering, C. F. Homnick, J. Nunberg, J. P. Springer,
J. R. Huff, J. Med. Chem. 1992, 35, 1685.
[13] a) P. Chen, P. T. W. Cheng, S. H. Spergel, R. Zahler, X.
Wang, J. Thottathil, J. C. Barrish, R. P. Polniaszek, Tet-
rahedron Lett. 1997, 38, 3175; b) M.-N. Dufour, P.
Jouin, J. Poncet, A. Pantaloni, B. Castro, J. Chem. Soc.
Perkin Trans. 1 1986, 1895; c) K. Izawa, T. Onishi,
Chem. Rev. 2006, 106, 2811.
[14] Surprisingly, asymmetric hydrogenation methodology
has not been extensively used in the reduction of these
aminoalkyl halomethyl ketones. Transfer hydrogenation
methodology was successful in supplying the erythro
isomer with good selectivity, see: T. Hamada, T. Torii,
[5] a) N. A. Roberts, J. A. Martin, D. Kinchington, A. V.
Broadhurst, J. C. Craig, I. B. Duncan, S. A. Galpin,
1838
asc.wiley-vch.de
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 1829 – 1839

Synthesis of a Long Acting HIV Protease Inhibitor via Metal or Enzymatic Reduction
T. Onishi, K. Izawa, T. Ikariya, J. Org. Chem. 2004, 69,
7391.
[15] All the ligands and metal precursors are available from
Strem.
[16] a) R. N. Patel, A. Banerjee, C. G. McNamee, D. B.
Brzozowski, L. J. Szarka, Tetrahedron: Asymmetry
1997, 8, 2547; b) R. N. Patel, J. Am. Oil Chem. Soc.
1999, 76, 1275.
[19] Amide 2 was prepared by the literature protocols used
in the synthesis of Crixivan, see: C. H. Senanayake,
F. E. Roberts, L. M. DiMichele, K. M. Ryan, J. Liu,
L. E. Fredenburgh, B. S. Foster, A. W. Douglas, R. D.
Larsen, T. R. Verhoeven, P. J. Reider, Tetrahedron Lett.
1995, 36, 3993.
[20] See ref.[9]
[21] For example (see our previous work ref.[8]) an ester Li-
enolate, that was unstable above ꢀ308C, could be
made stable at 08C for 24 h by forming its Zn-diamine
complex. Yet that Zn-enolate reacted more than 10
times faster than the corresponding Li-enolate.
[17] These enzymes are available in ready-to-use kits by the
Codexis corporation.
[18] Detailed kinetic plots and evaluation of each parame-
ter will be reported in a separate communication.
Adv. Synth. Catal. 2013, 355, 1829 – 1839
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1839