A flexible, practical, and stereoselective synthesis of enantiomerically pure trans-5-oxohexahydropyrrolo[3,2-b]pyrroles (pyrrolidine-trans-lactams), a new class of serine protease inhibitors, using acyliminium methodology

Article abstract of DOI:10.1021/jo990306s

A flexible, practical, and stereoselective synthesis of enantiomerically pure trans-5-oxohexahydropyrrolo[3,2-b]pyrroles (pyrrolidine-trans-lactams) is described. The key reaction involves addition of Z-ketene acetal 24 to the acyliminium ion derived from 48. This reaction is mediated by BF₃. OEt₂ and introduces the 6S and 6aS stereocenters stereoselectively. The acyliminium precursor was prepared in four different ways: from racemic 2,4- diaminobutyric acid 8, from (R)-asparagine, from (R)-methionine, and via a crystallization-induced dynamic resolution of a salt of the racemic amine 56. (R)-Methionine is the preferred starting material for the preparation of enantiomerically pure material. The best conditions for addition of the ketene acetal to the acyliminium ion derived from 48 were determined by systematically screening a range of ketene acetals and Lewis acids. The best ketene acetal was Z-(1-ethoxy-3-methylbut-1-enyloxyl)triisopropylsilane 24. In this series, the bulk of the silyl group of the Z-ketene acetal can be correlated with increased 6S isopropyl product. Use of the E-ketene acetal does not lead to a significant change in stereoselectivity for the 6R isopropyl product. In contrast, variation of the Lewis acid has a considerable effect on the product stereochemistry. While BF₃·OEt₂ gives predominantly 6S,6aS product, AlCl₃ and TiCl₄ give predominantly mixtures of the 6R,6aS and 6S,6aS products and TMSOTf gives 6aR material with predominantly one unknown isopropy] isomer (trans-lactam numberings used). The synthesis can conveniently be carried out on a large scale to produce multigram quantities of the trans-lactam 28, which is a key precursor of pharmacologically active molecules such as 1, a selective and orally active human neutrophil elastase inhibitor. The overall chemical yield of 1 is 1.3%, corresponding to an average of >70% yield for each of the 14 steps, and the synthesis contains only one chromatographic purification.

Full text of DOI:10.1021/jo990306s

5166
J . Org. Chem. 1999, 64, 5166-5175
A F lexible, P r a ctica l, a n d Ster eoselective Syn th esis of
En a n tiom er ica lly P u r e
tr a n s-5-Oxoh exa h yd r op yr r olo[3,2-b]p yr r oles
(P yr r olid in e-tr a n s-la cta m s), a New Cla ss of Ser in e P r otea se
In h ibitor s, Usin g Acylim in iu m Meth od ology
Simon J . F. Macdonald,* Geoffrey D. E. Clarke, Michael D. Dowle, Lee A. Harrison,
Simon T. Hodgson, Graham G. A. Inglis, Martin R. J ohnson, Prit Shah,
Richard J . Upton,† and Steven B. Walls
Enzyme Chemistry 2 and Physical Sciences, GlaxoWellcome Medicines Research Centre,
Gunnels Wood Road, Stevenage SG1 2NY, U.K.
Received February 18, 1999
A flexible, practical, and stereoselective synthesis of enantiomerically pure trans-5-oxohexahydro-
pyrrolo[3,2-b]pyrroles (pyrrolidine-trans-lactams) is described. The key reaction involves addition
of Z-ketene acetal 24 to the acyliminium ion derived from 48. This reaction is mediated by BF3‚
OEt2 and introduces the 6S and 6aS stereocenters stereoselectively. The acyliminium precursor
was prepared in four different ways: from racemic 2,4-diaminobutyric acid 8, from (R)-asparagine,
from (R)-methionine, and via a crystallization-induced dynamic resolution of a salt of the racemic
amine 56. (R)-Methionine is the preferred starting material for the preparation of enantiomerically
pure material. The best conditions for addition of the ketene acetal to the acyliminium ion derived
from 48 were determined by systematically screening a range of ketene acetals and Lewis acids.
The best ketene acetal was Z-(1-ethoxy-3-methylbut-1-enyloxyl)triisopropylsilane 24. In this series,
the bulk of the silyl group of the Z-ketene acetal can be correlated with increased 6S isopropyl
product. Use of the E-ketene acetal does not lead to a significant change in stereoselectivity for the
6R isopropyl product. In contrast, variation of the Lewis acid has a considerable effect on the product
stereochemistry. While BF3‚OEt2 gives predominantly 6S,6aS product, AlCl3 and TiCl4 give
predominantly mixtures of the 6R,6aS and 6S,6aS products and TMSOTf gives 6aR material with
predominantly one unknown isopropyl isomer (trans-lactam numberings used). The synthesis can
conveniently be carried out on a large scale to produce multigram quantities of the trans-lactam
28, which is a key precursor of pharmacologically active molecules such as 1, a selective and orally
active human neutrophil elastase inhibitor. The overall chemical yield of 1 is 1.3%, corresponding
to an average of >70% yield for each of the 14 steps, and the synthesis contains only one
chromatographic purification.
In tr od u ction
lators such as Ventolin and Serevent used in the treat-
ment of asthma.
Serine proteases constitute a major class of enzymes
that are widely distributed in the human body and are
frequently implicated in life-threatening disease states.
Consequently, enormous research programs, many within
the pharmaceutical industry, are targeted at identifying
selective serine protease inhibitors. Despite these efforts,
there are only a few simple, tractable nonpeptidic tem-
plates for serine protease inhibition. However, research
at GlaxoWellcome has recently identified new templates
based on 5,5-trans-fused ring systems1sincluding pyr-
rolidine-trans-lactones2 and pyrrolidine-trans-lactams2s
which possess inhibitory activity against serine proteases
such as trypsin, chymotrypsin, thrombin, and cathepsin
G.2 The discovery that these templates also provide
effective inhibitors of human neutrophil elastase (HNE),2
combined with the knowledge that such inhibitors are
widely considered as potential therapy for respiratory
disease,3 was of interest to us at GlaxoWellcome as part
of a long-term search for effective therapies for respira-
tory diseases that has led to the discovery of bronchodi-
In chronic bronchitis, HNE is thought to play a pivotal
role in generating the excessive mucorrhea and mucus-
secreting cell hyperplasia characteristic of the disease.4
(1) O’Neill, M. J .; Lewis, J . A.; Noble, H. M.; Holland, S.; Mansat,
C.; Farthing, J . E.; Foster, G.; Noble, D.; Lane, S. J .; Sidebottom, P.
J .; Lynn, S. M.; Hayes, M. V.; Dix, C. J . J . Nat. Prod. 1998, 1328-31.
Weir, M. P.; Bethell, S. S.; Cleasby, A.; Campbell, C. J .; Dennis, R. J .;
Dix, C. J .; Finch, H.; J hoti, H.; Mooney, C. J .; Patel, S.; Tang, C. M.;
Ward, M.; Wonacott, A.; Wharton, C. W. Biochemistry 1998, 37, 6645.
Kelly, H. A.; Bolton, R.; Brown, S. A.; Coote, S. J .; Dowle, M.; Dyer,
U.; Finch, H.; Golding, D.; Lowdon, A.; McLaren, J .; Montana, J . G.;
Owen, M. R.; Pegg, N. A.; Ross, B. C.; Thomas, R.; Walker, D. A.
Tetrahedron Lett. 1998, 39, 6979-6982. Finch, H.; Pegg, N. A.;
McLaren, J .; Lowdon, A.; Bolton, R.; Coote, S. J .; Dyer, U.; Montana,
J . G.; Owen, M. R.; Dowle, M. D.; Buckley, D.; Ross, B. C.; Campbell,
C.; Dix, C.; Mooney, C.; Man-Tang, C.; Patel, C. Bioorg. Med. Chem.
Lett. 1998, 8, 2955-2960.
(2) (a) Macdonald, S. J . F.; Belton, D. J .; Buckley, D. M.; Spooner,
J . E.; Anson, M. S.; Harrison, L. A.; Mills, K.; Upton, R. J .; Dowle, M.
D.; Smith, R. A.; Molloy, C. R.; Risley, C. J . Med. Chem. 1998, 41,
3919-3922. (b) Macdonald, S. J . F.; Spooner, J . E.; Dowle, M. D. Synlett
1998, 1375-1377. (c) Macdonald, S. J . F.; Montana, J . G.; Buckley, D.
M.; Dowle, M. D. Synlett 1998, 1378-1380. (d) Macdonald, S. J . F.;
Mills, K.; Spooner, J . E.; Upton, R. J .; Dowle, M. D. J . Chem. Soc.,
Perkin Trans. 1 1998, 3931-3936. (e) Dowle, M. D.; Finch, H.;
Harrison, L. A.; Inglis, G. G. A.; J ohnson, M. R.; Macdonald, S. J . F.;
Shah, P.; Smith, R. A. WO 9736903 A1 971009.
† Physical Sciences.
10.1021/jo990306s CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/17/1999

Synthesis of Pyrrolidine-trans-lactams
J . Org. Chem., Vol. 64, No. 14, 1999 5167
These effects result in recurrent productive cough and
lung infection. We believe that inhibitors of HNE may
ameliorate these symptoms.
These criteria led to consideration of a synthesis based
on acyliminium chemistry. This paper describes our
success in using acyliminium chemistry to access either
enantiomer of pyrrolidine trans-lactams and the develop-
ment of a robust, flexible, and scalable synthesis. When
combined with the ability to use the trans-lactams for
inhibition of different serine proteases, and with the
pedigree that trans-lactams can be orally active, bioavail-
able molecules, it represents a breakthrough that unlocks
the enormous potential of this scaffold for exploitation.
This is demonstrated by our experience with HNE
inhibitors; the chemistry enabled an exploration of a new
area of medicinal chemistry that led to a potential lead
candidate 1 (GW311616A) and has also laid the founda-
tion for large-scale preparations of pyrrolidine trans-
lactams.
In previous publications,2 we have described the syn-
thesis and basic medicinal chemistry of some racemic
pyrrolidine-trans-lactones and lactams. Although potent
HNE inhibitors, these compounds were, however, meta-
bolically vulnerable in vivo. We have now designed
further pyrrolidine-trans-lactams that are potent, orally
active, and bioavailable in animal models5 as exemplified
by the crotonamido-trans-lactam 1. These compounds
have as a template the isopropylpyrrolidine-trans-lactam
2.
In this paper, we describe (1) the retrosynthetic plan-
ning; (2) the construction of the racemic acyliminium ion
precursor 15; (3) the synthesis of a racemic isopropyl
pyrrolidine trans-lactam 2; (4) the exploration of the
acyliminium chemistry; (5) the preparation of enantio-
merically pure isopropyl trans-lactam 28; and (6) the
dynamic chiral resolution of a key intermediate 56.
A key development in the medicinal chemistry program
was the introduction of the isopropyl group in 2 R to the
lactam carbonyl and in place of the n-propyl group in 3,
which had been present in previous series.2 This seem-
ingly straightforward structural change remained syn-
thetically elusive for a considerable time. Despite the
excellent stereoselectivity of our previous synthesis,2a it
proved unsuitable as a route for accessing quantities of
isopropyl derivatives 2 for our medicinal chemical pro-
gram. The necessary criteria for alternative syntheses
included the following: (i) an efficient route to isopropyl-
trans-lactams 2; (ii) the flexibility to introduce a range
of alternative groups in place of isopropyl or n-propyl;
and (iii) the option to prepare large quantities of enan-
tiomerically pure material for drug evaluation if required.
Normal scale-up criteria such as high-yielding reactions,
the avoidance of low-temperature steps (<-50 °C), or the
use of expensive reagents and ease of purification by
recrystallization or trituration (rather than chromatog-
raphy) were also sought. In the long term, these issues
would be critical to our ability to efficiently produce
quantities of high-purity drug candidate material.
Retr osyn th etic P la n n in g
Previous work in the pyrrolidine trans-lactone series2c
had demonstrated the suitability of acyliminium chem-
istry to access lactones. We regarded this as good
precedent for application to the trans-lactam series. Our
retrosynthetic plan (Scheme 1) initially disconnected the
lactam 2 to the ester 4, which would be prepared by the
reaction of the acyliminium ion 5 with a ketene acetal 6.
We expected from precedent2c in our work with the
analogous trans-lactones to be able to exclusively gener-
ate the trans isomer across C-2 and C-3 (as in 4), and
we believed that we would also be able to introduce the
isopropyl group stereoselectively by appropriate modifi-
cation of the ketene acetal 6. Thus, replacement of the
isopropyl group could, in principle, be achieved readily
by replacing 6 with a ketene acetal of choice. The
acyliminium ion 5 could be accessed from the 3-aminopyr-
rolidin-2-one 7, available readily from racemic 2,4-
diaminobutyric acid 8.
Once the validity of this approach had been estab-
lished, we planned to examine its use in the enantio-
merically pure series. In this regard, the configurational
stability of the C-3 stereocenter of the acyliminium ion
5 is critical as it controls the generation of the remaining
contiguous stereocenters.
There is little literature precedent for the reaction of
acyliminium ions containing nitrogen substituents at the
3-position (as in 5). One example is the reaction of the
pyrrolidone 9 where the 3-amino protecting group reacts
intramolecularly with the acyliminium ion6 (eq 1).
(3) HNE for respiratory diseases: (a) Stockley, R. A. Am. J . Respir.
Crit. Care Med. 1994, 150, S109-S113. (b) Vender, R. L. J . Invest.
Med. 1996, 44, 531-539. Reviews of HNE inhibitors: (c) Hlasta, D.
J .; Pagani, E. D. Ann. Rep. Med. Chem. 1994, 195. (d) Edwards, P. D.;
Bernstein, P. R. Med. Res. Rev. 1994, 14, 127. Leading references for
HNE inhibitors from pharmaceutical companies and academia follow.
Merck: (e) Finke, P. E.; Shah, S. K.; Fletcher, D. S.; Ashe, B. M.;
Brause, K. A.; Chandler, G. O.; Dellea, P. S.; Hand, K. M.; Maycock,
A. L.; Osinga, D. G.; Underwood, D. J .; Weston, H.; Davies, P.; Doherty,
J . B. J . Med. Chem. 1995, 38, 2449-2462. Zeneca: (f) Veale, C. A.;
Bernstein, P. R.; Bohnert, C. M.; Brown, F. J .; Bryant, C.; Damewood,
J . R.; Earley, J .; Feeney, S. W.; Edwards, P. D.; Gomes, B.; Hulsizer,
J . M.; Kosmider, B. J .; Krell, R. D.; Moore, G.; Salcedo, T. W.; Shaw,
A.; Silberstein, D. S.; Steelman, G. B.; Stein, M.; Strimpler, A.; Thomas,
R. M.; Vacek, E. P.; Williams, J . C.; Wolanin, D. J .; Woolson, S. J .
Med. Chem. 1997, 40, 3173-3181. Sterling Winthrop: (g) Hlasta, D.
J .; Ackermann, J . H.; Court, J . J .; Farrell, R. P.; J ohnson, J . A.; Kofron,
J . L.; Robinson, D. T.; Talomie, T. G.; Dunlap, R. P.; Franke, C. A. J .
Med. Chem. 1995, 38, 4687-4692. Hoechst: (h) Burkhart, J . P.; Mehdi,
S.; Koehl, J . R.; Angelastro, M. R.; Bey, P.; Peet, N. P. Bioorg. Med.
Chem. Lett. 1998, 8, 63-64. Wichita State University: (i) Kuang, R.;
Venkataraman, R.; Ruan, S.; Groutas, W. C. Bioorg. Med. Chem. Lett.
1998, 8, 539-544.
(1)
(4) (a) Sommeroff, C. P.; Nadel, J . A.; Basbaum, C. B.; Caughey, G.
H. J . Clin. Invest. 1990, 85, 682-689. (b) Thompson, A. B.; Daughton,
D.; Robbins, R. A.; Ghafouri, M. A.; Oehlerking, M.; Rennard, S. I. Am.
Rev. Respir. Dis. 1969, 140, 1527-1537.
(5) In-house models were developed by Robin A. Smith and col-
leagues. Details will be published in due course.
(6) Koot, W.-J .; Hiemstra, H.; Speckamp W. N. Tetrahedron: Asym-
metry 1993, 4, 1941-1948.

5168 J . Org. Chem., Vol. 64, No. 14, 1999
Macdonald et al.
Sch em e 1a
Sch em e 2a ,b
a
a
All structures are racemic. bKey: (a) AcCl, MeOH, reflux, 95%
All structures are racemic.
crude; (b) Dowex 2X8-400 (hydroxide form), MeOH, rt; (c)
CF3CO2Me, MeOH, rt, 67%; (d) n-BuLi, THF, -78 °C, then CBZCl,
-70 to -20 °C, 96% crude; (e) LiBH4, THF, -20 °C, then EtOH,
cH2SO4, 96% crude.
This precedent highlighted the need for the careful
selection of nitrogen protecting groups. While recognizing
the desirability of the 3-amino protecting group to direct
nucleophilic attack at C-2 from the opposite face,7 we
questioned the use of a tert-butylcarbamate protecting
group (as in 9) for our purposes, as the acid lability of
the tert-butyl group may facilitate the observed reaction.
Given these criteria, the trifluoroacetyl protecting group
was chosen. One concern, however, was the potential
lability of the trifluoroacetamide during reduction of the
pyrrolidone 7 (PG ) COCF3) to a suitable acyliminium
precursor.
to lithium borohydride in THF11 resulted in chemoselec-
tive reduction of the ring carbonyl to give a mixture of
the cis- and trans-alcohols, which were converted in situ
to a 1:1 mixture of cis- and trans-ethers 15 by addition
of ethanol and concentrated sulfuric acid. Addition of the
ethanol before the acid suppressed production of 16,
which was otherwise obtained by further reduction. The
cis- and trans-ethers 15 could be separated by flash
chromatography to give crystalline solids but were rou-
tinely used without purification.
This high-yielding route has been used to produce
hundreds of grams of the acyliminium precursor 15
without the requirement for chromatographic purifica-
tion.
Con str u ction of th e Ra cem ic Acylim in iu m
P r ecu r sor 15
The synthesis of the acyliminium precursor 15 starts
from commercially available racemic diaminobutyric acid
8 (Scheme 2), which can be converted into the pyrrolidone
12 in one step using hexamethyldisilazane.8 However, on
a larger scale, a two-step procedure was preferred. This
involved initially generating the methyl ester 11 under
standard conditions and was followed by treatment of the
intermediate with Dowex 2 resin (a strongly basic anion
exchanger) in methanol. After filtration, the filtrate was
reacted with trifluoromethyl acetate to give the crude
trifluoroacetamide 13. Recrystallization from water gave
pure 13 in 67% yield.9 Treatment of this acetamide 13
with 1 equiv of n-butyllithium in THF followed by benzyl
chloroformate selectively derivatized the lactam nitrogen
to give the imide 14 in 96% yield.10 Exposure of the imide
Th e Syn th esis of th e Ra cem ic
Isop r op ylp yr r olid in e-tr a n s-la cta m 2
With the availability of the ethers 15, the scene was
set for the key step of the synthesis (Scheme 3, conditions
1). Treatment of either the cis- or the trans-ether 15 with
the commercially available dimethylketene acetal 18 in
dichloromethane followed by boron trifluoride at 5 °C
provided the trans product 19 in 36% yield. This result
is consistent with the formation of the acyliminium ion
17 (which is likely to be in equilibrium with the dihy-
drooxazolium ion as shown in Scheme 3) as the reactive
1
intermediate. H NMR experiments (NOE experiments
and coupling constants) confirmed the trans geometry at
C2 and C3 of 19.2 Similarly, treatment of 15 with 4:1
E/ Z-isopropyltrimethylsilylketene acetal 20 gave the
trans products 21 (as confirmed by NOE experiments)
with a ratio of isopropyl diastereomers of ca. 1.3:1 â/R
(as determined by analytical HPLC). (In the racemic
(7) Thaning, M.; Wistrand L.-G. J . Org. Chem. 1990, 55, 1406-1408
and references therein.
(8) Pellegata, R.; Pinza, M.; Pifferi, G. Synthesis 1978, 614-616.
(9) Recrystallization is important to remove traces of unreacted
diaminobutyric acid 8, which severely affect the efficiency of derivation
to the benzyl carbamate 14.
(10) The lithium salt of 13 is preferred to the sodium salt due to its
substantially greater solubility in THF. n-Butyllithium is preferred
to lithium hexamethyldisilazide to avoid contamination of the product
with CBZNH2.
(11) Reduction of the imide 14 can be achieved with NaBH4 and
HCl in dioxan. However, excess borohydride is needed, and the reaction
is not homogeneous.

Synthesis of Pyrrolidine-trans-lactams
J . Org. Chem., Vol. 64, No. 14, 1999 5169
Sch em e 3a ,b
a
All structures are racemic. bConditions 1 (small scale): (a) Me2CC(OMe)OSiMe3 18, BF3‚OEt2, CH2Cl2, 5 °C to rt, 36%; (b) 4:1 Z:E-
iPrCHC(OEt)OSiMe3 20, BF3‚OEt2, CH2Cl2, 5 °C to rt, (81% if chromatographed); (c) K2CO3, H2O, EtOH, reflux; (d) t-BuMgCl, THF, -5
°C, 51% after chromatography for three steps. Conditions 2 (large scale): (b) Z-iPrCHC(OSiiPr3)OEt 24, BF3‚OEt2, CH2Cl2, 5 °C; (c) K2CO3,
H2O, EtOH, reflux; (d) t-BuMgCl, 1:1 THF/TMEDA, -5 °C, 20% after recrystallization for three steps.
Ta ble 1. Obser ved a n d P r ed icted a Key Cou p lin g
series as drawn, the â stereochemistry is defined as the
Con sta n ts for Isop r op yl tr a n s- a n d cis-La cta m s
isopropyl group drawn up and the R stereochemistry as
that drawn down). These isomers can be separated by
chromatography, but no purification was actually re-
quired at this stage. Instead, removal of the trifluoro-
acetyl group under basic conditions gave the amines 22.
Lactamization with tert-butylmagnesium chloride12 in
THF at -5 °C gave ca. a 1.3:1 mixture of the â- and
R-isopropyl lactams 2 and 23 in 51% yield for three steps.
27b
The relative stereochemistries of 2 and 23 were assigned
by 1H NMR experiments. They can be separated by flash
chromatography to give ca. 1.3:1 2â:23R, but one recrys-
tallization of the crude material from ethyl acetate gave
the more crystalline â-isomer 2 (the desired isomer for
our medicinal chemistry program) in analytical purity.
Our confidence in assigning the trans ring junction in
2
23
obsd
pred
obsd
pred
obsd
pred
J 6-6a
J 3a-6a
11
10
11.7
10.4
6.5
10.5
6.8
10.5
1
6
0.7
5.8
a
Predicted values were calculated using Macromodel 6.0,
Columbia University, 1996. (Mohamadi, F; Richards, N. G. J .;
Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.;.
Hendrickson, T.; Still, W. C. J . Comput. Chem. 1990, 11, 440).
1
2 and 23 from the H NMR spectra (vide supra)2a was
b
The predicted coupling contants using Macromodel for the other
derived from comparison of the coupling constants of the
ring junction protons (H6a and H3a) of the trans- and
cis-lactams (whose synthesis is described later) (Table
1). The stereochemistry of the isopropyl groups in 23 and
2 was established by the coupling constants between H-6
and H-6a, which are diagnostic because of the rigidity of
these ring systems (Table 1). From decoupling experi-
ments, the â-isopropyl lactam 2 has a coupling of ca. 11
Hz (corresponding to an angle between H-6 and H-6a of
ca. 180°) and the R-lactam 23 has a coupling of ca. 6.5
Hz (corresponding to an angle between H-6 and H-6a of
ca. 30°). Furthermore, these coupling constants for the
ring junctions and isopropyl stereochemistry are in good
agreement with those predicted by Macromodel and also
with those observed for the analogous trans- and cis-
lactones.2d
Having established the structures of the â- and R-iso-
propyl trans-lactams 2 and 23, we wished to establish
the stereochemistry of the isopropyl groups in the esters
21â and 21R. To this end, the isomers 21â and 21R (which
had been separated by chromatography) were converted
into the respective lactams 2 and 23 by the process just
described for the isomer mixture. This allowed the
isopropyl isomer are J 6-6a ) 5.6 Hz and J 6a-3a ) 6.3 Hz. The
difference in J 6-6a between the two isomers and the accuracy of
these predicted coupling constants support the assigned stereo-
chemistry of 27.
stereochemistry of the isopropyl group for 21â and 21R
to be assigned. At no point in the conversion of 21â and
21R to the respective lactams was the presence of the
other isomer detected. Although epimerization of the
isopropyl group is in theory possible under the reaction
conditions used in these conversions, it is considered
unlikely that 21R would epimerize to give the â-isopropyl
lactam 2 and that 21â would epimerize to give the
R-isopropyl lactam 23.
This synthetic route was a breakthrough in our me-
dicinal chemistry program since, for the first time, it
allowed us to synthesize a wide variety of â-isopropyl
analogues and confirmed, as had been predicted, their
superior biological profile over the corresponding n-propyl
analogues. Although this process generates ca. 40% of
the undesired R-analogue, the entire synthesis is notable
in that no chromatography is required.
Exp lor a tion of th e Acylim in iu m Ch em istr y
The main drawback of the above synthesis was the
(12) Borthwick, A. D.; Crame, A.; Exall, A.; Mason, A.; Pennell, A.
Tetrahedron Lett. 1999, 40, 3061-2.
production of ca. 40% of the undesired R-analogue 23.

5170 J . Org. Chem., Vol. 64, No. 14, 1999
Macdonald et al.
Ta ble 2. Rea ction of Keten e Aceta ls w ith th e
Ta ble 3. Rea ction of Acylim in iu m P r ecu r sor 15 (or 48)
a n d th e ter t-Bu tyld im eth ylsilylk eten e Aceta l 25 w ith
Lew is Acid s (a n d a Br on sted Acid )
Acylim in iu m P r ecu r sor 15 (All Str u ctu r es Ra cem ic)a
ketene acetal
ratio of
Lewis acid
21â:21Ra
Lewis acid
ratio of 21â:21Ra
BF3‚OEt2 3:1
Sn(OTf)2
Yb(OTf)3
MgBr2
no reactione
TiCl4
1:1b
no reactione
1:1c
no reactione
R1
R2
E/ Z ratio
ratio of 21â/ 21R
AlCl3
HCO2H
no reactiond
TMSOTf
26 6:1 isomer ratioe,f
Et
Et
Me
Me
Et
Et
Me
Et
SiMe3
20
25
4:1
1:12.5
1:2
1:10
1:12
1:10
1:12
>98:1
unknown
1.3:1
3:1
SitBuMe2
SitBuMe2
SitBuMe2
SiiPr3
a
The isomer ratios were determined by HPLC analysis of the
1:1b
crude products. b The cis isomers across C2-C3 were also produced
2:1b
i
(3:1 Pr isomer ratio); 4:1 21:26. c The cis isomers across C2-C3
d
were also produced (3:1 iPr isomer ratio); 1.3:1 21:26. The
24
3.5:1
2:1
SiiBu3
triisopropylsilylketene acetal 24 was used. e The homochiral ethers
48 were used instead of 15. f See Scheme 4.
SiiPr3
2.5:1c
1.4:1
1.4:1
SitBuMe2
SiEt3
Me
Sch em e 4
a
All results unless otherwise stated were obtained by analytical
HPLC of a sample of the crude reaction mixtures after dilution
b
with aqueous acetonitrile. These reactions did not proceed to
completion. c There was also some detrifluoroacetylated product
22 present in the crude reaction mixture.
The effect of varying the nature of the ketene acetal and
the Lewis acid was investigated in an attempt to improve
the â/R ratio in the production of 21.13 Given the avail-
ability of assigned pure 21R and 21â isomers, the â/R
ratio of isopropyl esters 21 from crude reaction mixtures
under different conditions could be readily determined
by analytical HPLC.
A range of ketene acetals were prepared14 and used in
the acyliminium reaction (Table 2).15 From this limited
exploration there appears to be little correlation between
the geometry of the double bond of the ketene acetal used
and the â/R ratio of the product 21. However, there is a
correlation between the steric influence of the silyl group
in the Z-ketene acetal and a preference for the â-product
21â. Thus, a 12:1 mixture of Z/ E-triisopropylsilylketene
acetal 24 gave a 3.5:1 ratio of 21â:21R.
We next examined variation of the Lewis acid (and a
Bronsted acid) (Table 3) using the 12.5:1 Z/ E-tert-
butyldimethylsilylketene acetal 25 and found that chang-
ing the Lewis acid can alter the â/R ratio. Thus, use of
boron trifluoride etherate gave a 3:1 ratio of 21â:21R. In
contrast, the use of titanium tetrachloride or aluminum
trichloride gave little selectivity for 21â over 21R. Formic
acid, magnesium(II) bromide, tin(II) triflate, and ytter-
bium triflate gave no reaction.
Trimethylsilyl triflate did not give the required trans
products 21, but instead, cis products 26 with a 6:1
isopropyl stereoisomer ratio in 44% yield were obtained.
To confirm the cis stereochemistry, the separated isomers
were converted into the cis-lactams 27 by hydrolysis of
the trifluoroacetamide with potassium carbonate in
aqueous ethanol (Scheme 4). Both isomers gave only the
â isopropyl cis-lactam 27, indicating that one of the
isomers had epimerized. 1H NMR experiments confirmed
the structure of 27 by observation of an NOE between
H6a and H3a and by comparison with the spectra of the
trans-lactams 2 and 23 (see Table 1).
Having identified a preferred ketene acetal 24 and
Lewis acid (BF3‚OEt2), the reaction was scaled up (Scheme
3, conditions 2). Thus, treatment of 15 and 3 equiv of 24
in DCM with 6 equiv of boron trifluoride etherate at 5
°C gave similar amounts of the trifluoroacetamides 21
and the amines 22, both having ca. 3.5:1 â/R ratio. It is
thought that the excess ketene acetal deprotects the
trifluoroacetamide. When less ketene acetal 24 was used,
unreacted starting material 15 was observed. Exposure
of the mixture of 21 and 22 to potassium carbonate in
1:1 water/ethanol, followed by an acid/base workup, gave
the crude amines 22 in 63% yield over the two steps.
Analytical HPLC showed the â/R ratio of 22 after workup
to be ca. 3.5:1.
After the considerable effort of discovering conditions
that would afford predominantly the â isomer, it was with
some dismay that we observed that lactamization of 22
with tert-butylmagnesium chloride in THF gave a 1:1
mixture of â/R trans-lactams 2 and 23. (This had not
previously been observed when lactamizing 1.3:1 22â/
22R). What was also noticed, however, was that with the
change in the â/R ratio, benzyl alcohol was also formed
in the reaction mixture. It became clear from analytical
HPLC of samples taken as the reaction proceeded that
the R isopropylamine 22R was converted into the R
isopropyl lactam 23 faster than the â isopropylamine 22â
was converted into the â isopropyl lactam 2. We postulate
that the reaction byproduct, magnesium ethoxide, attacks
(13) We have also explored the use of the boron enolates of isopropyl
valerate derivatives attached to a chiral auxiliary (see ref 28).
(14) Ireland, R. E.; Mueller, R. H.; Willard, A. K. J . Am. Chem. Soc.
1976, 98, 2868-2877. Oare, D. E.; Heathcock, C. H. J . Org. Chem.
1990, 55, 157-172. Ireland, R. E.; Wipf, P.; Armstrong, J . D. J . Org.
Chem. 1991, 56, 650-657. Otera, J .; Fujita, Y.; Fukuzumi, S. Synth.
Lett. 1994, 213-214.
(15) The (2-cyclopropyl-1-ethoxy-E-vinyloxy)trimethylsilane and the
(1-ethoxy-2-ethyl-1-butenyloxy)trimethylsilane react with 15 to give
the corresponding products in good yield.

Synthesis of Pyrrolidine-trans-lactams
J . Org. Chem., Vol. 64, No. 14, 1999 5171
Sch em e 5
A clue to an alternative or contributing explanation
for the observation that E-ketene acetals preferentially
give â rather than R isopropyl products comes from a
serendipitous observation that the E-ketene acetal (>98%
E), on standing in the dark for over a year, had isomer-
ized completely to the Z-ketene acetal together with some
decomposition to ethyl isovalerate. We cannot therefore
rule out the possibility that the E-ketene acetals, in the
presence of the Lewis acid, isomerize in situ to give the
Z-ketene acetals before reaction with the acyliminium
ion. This isomerization of E- to Z-ketene acetals has been
observed by Wilcox18 with trialkylammmonium perchlo-
rates and by Adam19 with trifluoromethyl ketones.
Whatever the reason for the Z-ketene acetal giving a
greater level of stereocontrol over the E-ketene acetal, it
is clear that the nature of the Lewis acid plays a crucial
role. It is not immediately apparent why BF3‚OEt2 gives
better stereocontrol than TiCl4 or AlCl3, although the
labilities of the metal-halogen bonds may be a factor.
While the BF3‚OEt2-mediated reaction is highly robust
and has been repeated many times, the parameters of
these reactions in terms of temperature, solvent, stoichi-
ometries (Lewis acid and ketene acetal), and the reaction
times in terms of their effect on the stereochemistries of
the products are not well understood.
F igu r e 1.
the benzyl carbamate of either the â isopropylamine 22â
and/or the â isopropyl lactam 2 in preference to the R
analogues, leading to a reduction in the yield of the
required â isopropyl lactam. This theory was supported
by mass spectroscopic evidence showing that ethyl car-
bamate trans-lactam was present in the crude reaction
mixture.16 We reasoned that by adding a metal chelator
to the reaction mixture we might reduce the propensity
for this side reaction. We were thus gratified to discover
that lactamization in a 1:1 mixture of TMEDA/THF17
preserves the â/R ratio in the lactam products 2 and 23
(although it does not completely suppress benzyl alcohol
formation). On a large scale, the desired trans-lactam 2
was obtained in 20% yield for three steps from the ethers
15. This yield is lower when compared with that obtained
previously (vide supra) and reflects principally loss of
material during purification of the lactam 2 by recrys-
tallization (as opposed to chromatography) and loss of
the amines 22 during the acid/base workup.
A possible explanation for the TMSOTf-mediated reac-
tion giving only cis products is shown in Scheme 5. After
N-silylation, the trifluoromethyl acetamide becomes more
susceptible to attack by the ketene acetal than the
acyliminium ion. This in effect then delivers a tethered
nucleophile to the iminium ion from the same face,
resulting in cis products. Desilylation occurs in workup.
Discu ssion of th e Ster eocon tr ol for th e Ad d ition
of Keten e Aceta ls to th e Acylim in iu m Ion
Med ia ted by Differ en t Lew is Acid s
Figure 1 shows the antiperiplanar (A and B) and the
synclinal (C and D) approaches of the E- or Z-silylketene
acetals to the acyliminium ion. Our results show that
either approach of the E ketene acetal is approximately
equally favored with BF3‚OEt2 as the Lewis acid, which
presumably indicates that the steric demands of the
isopropyl group and the silyloxy group are similar. In
contrast, with the Z-ketene acetal the antiperiplanar
approach is preferred due to reduced interactions be-
tween the OEt of the acetal and H3 of the iminium ion
Ap p lica tion of Th is Ch em istr y To P r ep a r e
En a n tiom er ica lly P u r e Isop r op yl tr a n s-La cta m
28
The capacity to prepare large quantities of isopropyl
trans-lactam greatly facilitated the discovery of potential
drug candidates in our medicinal chemistry program. The
next challenge involved the preparation of the enanti-
omers of these potential drug candidates (such as 1) to
determine their biological profiles and to assign their
absolute stereochemistry. We therefore separated the
racemic trans-lactam 2 into its enantiomers 28 and 29
by preparative HPLC20 and prepared analogues (such as
1). We then determined which enantiomer was the more
potent inhibitor of HNE, although its absolute stereo-
i
(as in A) over the Pr of the acetal and H3 of the iminium
ion (as in C). Increasing the steric demands of the silyl
i
group leads to restricted rotation of the Pr group and a
corresponding preference for A.
(16) A typical mass spectrum (thermospray) shows the presence of
desired product 23 and 2 - MH+ 303 (100%) and MNH4 320 (50%)
+
and peaks consistent with the corresponding ethyl carbamate MH+
+
241 (30%) and MNH4 258 (5%).
(17) Lactamization using t-BuMgCl in Et2O or solvent mixtures of
THF and Et2O fail to improve the â/R ratio.
(18) Wilcox, C. S.; Babston, R. E. J . Org. Chem. 1984, 49, 1451-3.
(19) Adam, W.; Wang, X. J . Org. Chem. 1991, 56, 7244-50.

5172 J . Org. Chem., Vol. 64, No. 14, 1999
Macdonald et al.
Sch em e 6
Sch em e 7a
a
Key: (a) CF3CO2Me, Et3N, MeOH, rt, 70%; (b) AcCl, MeOH,
-70 to -20 °C, 57%; (c) TsCl, pyridine, CH2Cl2, rt, 59%; (d) Rh on
Al2O3, H2, EtOH, rt, 23-26%; (e) n-BuLi, THF, -70 °C; then
CBZCl, rt, 55-70%.
chemistry remained unknown. To address this issue, we
decided to investigate whether the acyliminium chemis-
try described herein could be used to generate enantio-
merically pure material. If successful, it would allow
assignment of the absolute stereochemistry and also
allow access to greater quantities of enantiomerically
pure material for further biological evaluation.
The most obvious option was to repeat the chemistry
described previously (Scheme 3), starting from enantio-
merically pure diaminobutyric acid 30. We chose not to
pursue this option, primarily due to the high cost of
diaminobutyric acid21 and also because the biological
profiles of related compounds of known absolute stereo-
chemistry led us to predict that analogues of the 6S,-
6aS,3aR trans-lactam 28 would be those having the
greater biological activity. A retrosynthetic analysis
indicated that (R)-diaminobutyric acid would be the
required starting material, and this was not available
commercially.
pyridine gave, after chromatography, a 59% yield of the
nitrile 33.25 Hydrogenation of the nitrile 33 with rhodium
on alumina26 followed by an intramolecular cyclization
gave, after chromatography, a 26% yield of the cyclic
lactam 34 with a major byproduct being the trifluoro-
acetamide 35 (17% yield). Reaction of the lactam 34 with
1 equiv of n-butyllithium at -70 °C followed by addition
of benzyl chloroformate and warming to 0 °C or room
temperature gave yields from six experiments of 55-81%
of the desired imide 36. However, analysis of the products
36 by chiral HPLC showed that the enantiomer ratios
varied from R/S 50:50 to R/S 96:4. We did not investigate
extensively the origins of this loss in optical purity,
although clearly under certain conditions optical purity
may be maintained.
A batch of imide 37 (R/S 72:28) was progressed as
before (in 34% overall yield) to give the trans-lactams 28
(6S,6aS,3aR) and 29 (6R,6aR,3aS) in a ratio of 74:26,
suggesting that optical integrity is maintained during
this reaction sequence. Comparison of this synthetic
material by analytical chiral HPLC with the enantiomers
28 and 29, which had been separated by preparative
chiral HPLC, allowed their absolute stereochemistry to
be assigned:27 the stereochemistry of our potential lead
candidates was confirmed as 6S,6aS,3aR (as in 1).
Although this chemistry allowed us to assign absolute
stereochemistry, the tendency toward racemization dur-
ing the benzyl carbamate protection step and the low
yields led us to investigate (R)-methionine as an alterna-
tive starting material.28
After a thorough search of the literature, we decided
to investigate routes to enantiomerically pure trans-
lactam 28 from either asparagine or methionine via the
imide 31 (Scheme 6). Both enantiomers of asparagine and
methionine are commercially available, with (R)-aspar-
agine being cheaper than (R)-methionine.22 We thus
initially chose (R)-asparagine as our starting material.
An Ap p r oa ch to En a n tiom er ica lly P u r e Isop r op yl
tr a n s-La cta m 28 fr om (R)-Asp a r a gin e
(R)-Asparagine was converted into its trifluoroacet-
amide and then into the corresponding methyl ester 32
according to literature procedure23,24 (Scheme 7). Dehy-
dration of the primary amide using tosyl chloride and
La r ge Sca le Rou te to En a n tiom er ica lly P u r e
Isop r op yl tr a n s-La cta m 28 a n d Cr oton a m id e 1
fr om (R)-Meth ion in e
(20) The racemate 2 may be separated into its enantiomers by
preparative HPLC on a 2 in. Merck AD column, eluting with 15%
ethanol/heptane with a flow rate of 50 mL/min. The injection volume
is 25 mL, and the detector wavelength is 215 nm. The throughput of
material is limited by the solubility of 2 in the mobile phase (100 mg/
25 mL). The 6S,6aS,3aR enantiomer 28 elutes first and can be obtained
in satisfactory purity without recycling. The 6R,6aR,3aS enantiomer
elutes second and requires one to two recycles to obtain material of
satisfactory purity.
Construction of pyrrolidones from methionine is known
from the work of Freidinger et al.,29 and this method was
(25) There is a tosyl-derived byproduct formed in this reaction, and
its removal by chromatography also leads to degradation of some of
the product.
(21) (R,S)-2,4-Diaminobutyric acid dihydrochloride costs ca. £10/g.
(S)-2,4-Diaminobutyric acid dihydrochloride costs ca. £20/g from Ald-
rich Chemical Co.
(22) (R)-Asn costs £25/100 g, (S)-Asn costs £14/100 g, (R)-Met costs
£23/25 g, and (S)-Met costs £17/100 g from Aldrich Chemical Co.
(23) Curphey, T. J . J . Org. Chem. 1979, 44, 2805-2807.
(24) Maddaluno, J .; Corruble, A.; Leroux, V.; Ple, G.; Duhamel, P.
Tetrahedron: Asymmetry 1992, 1239-1242.
(26) Galan, A.; de Mendoza, J .; Prados, P.; Rojo, R.; Echavarren, A.
M. J . Org. Chem. 1991, 56, 452-454.
(27) Comparison of the circular dichroism spectra of the synthetic
material (mainly 28), the pure enantiomers 28 and 29 separated by
chiral HPLC, and the C-6 allyl analogues of 28 and 29 (whose absolute
stereochemistry had been determined by X-ray experiments) gave
consistent results reconfirming the assignment of absolute stereo-
chemistry.

Synthesis of Pyrrolidine-trans-lactams
J . Org. Chem., Vol. 64, No. 14, 1999 5173
Sch em e 8a
a
Key: (a) Boc2O, NaOH, H2O, 1,4-dioxan, rt, 100% crude; (b) Boc2O, pyridine, NH4HCO3, DMF, rt, 64-76%; (c) n-BuLi, THF, -70 °C
then CBZCl, -70 °C, 100% crude; (d) MeI, Me2CO or MeCN, rt, 77-88%; (e) Dowex 2X8-400 (hydroxide form), MeCN, rt, 67-76%; (f) 4
M HCl in 1,4-dioxan, rt then CF3CO2Me, NaHCO3, MeOH or CF3CO2Me, N-methylmorpholine, CH2Cl2, MeOH, 85-93%; (g) LiBH4, THF,
-20 °C then cH2SO4, EtOH, rt, 90-99% crude; (h) Z-iPrCHC(OSiiPr3)OEt 24, BF3‚OEt2, CH2Cl2; (i) K2CO3, H2O, EtOH, reflux; (j) t-BuMgCl,
THF, TMEDA, rt, 17-36% for three steps; (k) LiN(SiMe3)2, THF, -70 to 0 °C then recool to -70 °C, MeSO2Cl, 70-71%; (l) H2, Pd(OH)2,
EtOAc, 1,4-dioxan, rt, 96-100%; (m) 54, Me2NCH2CH2NdCdNEt‚HCl, 1-hydroxybenzotriazole, Et3N, MeCN, rt, 64%; after purification
1 M HCl, Et2O, CH2Cl2, rt, 87%. Alternatively 54, (COCl)2, DMF (cat.), CH2Cl2, rt then 53, NaHCO3, CH2Cl2, rt, 85%.
used as the basis for our preparation of enantiomerically
pure pyrrolidone 47. We wished ideally to introduce the
trifluoroacetyl group at the first step. However, due to
the known propensity30 of peptidic trifluoroacetamides
to cyclize to oxazolidinones and to racemize under basic
conditions, we decided to initially generate the tert-butyl
carbamate derivative of methionine and to replace this
protecting group with a trifluoroacetamide later in the
synthesis.
On a 1.7 M scale (250 g), (R)-methionine was protected
as its tert-butyl carbamate 41 and converted into the
primary amide 42 (Scheme 8).31 The yield after tritura-
tion was 64-76% over two steps.32 Conversion of the
primary amide 42 into the benzyl carbamate 45 was
achieved using 2 equiv of n-butyllithium (for complete
reaction) at -75 °C followed by addition of benzyl
chloroformate. The product 45 was obtained in quantita-
tive yield with greater than 96:4 R/ S chiral purity.33,34
The crude material was then treated with excess methyl
iodide in acetone or acetonitrile at room temperature to
give the sulfonium iodide 46 as a crystalline precipitate.35
Isolation gave a 77-88% yield of 46 over two steps.
(28) In this homochiral series, we also investigated briefly the
reaction of the ethers 37 with boron enolates of 39 in an attempt to
improve the selective introduction of the isopropyl group. There is
precedent that boron enolates can control the stereochemical introduc-
tion of substituents in a manner similar to our case. (See, for example:
Fuentes, L. M.; Shinkai, I.; Salzmann, T. N. J . Am. Chem. Soc. 1986,
108, 4675-4676. Shirai, F.; Chiba, T.; Nakai, T. Chem. Express 1992,
7, 141-144. Pilli, R.; Russowsky, D. J . Org. Chem. 1996, 61, 3187-
3190). Thus, treatment of 37 (R/S 72:28) with the boron enolate derived
from 38 gave predominantly one isomer presumed to be 39 in 17%
yield. 19F NMR of the crude mixture indicated an 83:14:3 mixture of
isomers. Attempted deprotection of the trifluoroacetamide failed, and
reaction of 39 with lithium ethoxide led to ring opening of the
oxazolidinone to give 40 in 60% yield rather than its displacement.
(31) Pozdnev, V. F. Tetrahedron Lett. 1995, 36, 7115-7118.
(32) Initially, the tert-butyl carbamate amide 42 was treated with
excess methyl iodide in ethyl acetate to give the sulfonium iodide 43.
Intramolecular cyclisation with sodium hydride gave the pyrrolidone
44 in 18% yield. After protecting the ring NH of 44 as its benzyl
carbamate derivative (25% yield), deprotection of the tert-butyl car-
bamate and reprotection as the trifluoroacetamide (32% yield), analyti-
cal chiral HPLC showed the resultant pyrrolidone 14 to be racemic.
(29) Freidinger, R. M.; Perlow, D. S.; Veber, D. F. J . Org. Chem.
1982, 47, 104-109.
(30) Bodansky, M. Principles of Peptide Synthesis; Springer-Ver-
lag: Berlin, 1984; p 161.
(33) An attempt to avoid the use of n-butyllithium by reaction of 42
with potassium carbonate, 18-crown-6, and benzyl chloroformate in a
variety of solvents gave little or low yields of product 45.

5174 J . Org. Chem., Vol. 64, No. 14, 1999
Macdonald et al.
Sch em e 9a
Cyclization of the sulfonium iodide 46 was achieved
initially on a small scale using potassium hydride in THF
and gave, after trituration with ether, a 42% yield of the
pyrrolidone 47 of >98% optical purity. To avoid the use
of potassium hydride on a large scale, we explored a
variety of bases such as lithium hexamethyldisilazide,
potassium carbonate, and hydroxide-exchanged Dowex
2X8-400 resin in a range of solvents such as acetonitrile,
acetone, THF, and water. Potassium carbonate in aceto-
nitrile gave good yields with little racemization in small-
scale experiments, but on scale-up some racemization of
47 was observed. The preferred cyclization conditions
were with ca. 1.3 wt equiv of Dowex resin in acetonitrile
at room temperature. This routinely gave crude product
47 with ca. 9:1 R/ S, which after one recrystallization
from cyclohexane/ethyl acetate afforded optically pure
product in 67-76% yields.35
a
Key: (a) TFA, CH2Cl2 then (+)-DPTT, rt, 96%; (b) cat. 3,5-
dichloro-2-hydroxybenzaldehyde, THF, rt, 81%; (c) CF3CO2Me,
N-methylmorpholine, CH2Cl2, MeOH, rt, 56%.
Swapping the tert-butyl carbamate of 47 for a trifluo-
roacetamide was achieved as follows. Deprotection of 47
was effected with 4 M hydrogen chloride in dioxan over
4 h at room temperature. After removal of the dioxan,
treatment of the residue with either sodium bicarbonate
and trifluoromethyl acetate in methanol or N-methyl-
morpholine and trifluoromethyl acetate in dichloro-
methane/methanol gave, without further purification,
85-93% yields of >98% optically pure 31.
The pyrrolidone 31 was converted into the trans-lactam
28 as shown in Scheme 8 using the conditions previously
described. As expected, the chirality was preserved
throughout. Analytical chiral HPLC of the crude trans-
lactam product showed 3:1 28:51 with less than 1% of
the undesired trans-lactam enantiomer 29. Recrystalli-
zation of the crude product 28 from ethyl acetate gave a
12:1 mixture of 28:51 with no other impurities. However,
the preferred procedure was to purify the crude material
by flash chromatography to remove all traces of 51, since
purification at the next stage was problematic.
The trans-lactam 28 was converted into the lead
candidate 1 in a straightforward manner. Deprotonation
of 28 with lithium hexamethyldisilazide followed by
addition of methanesulfonyl chloride gave the acyl sul-
fonamide 52 in 70-71% yield after trituration of the
crude product. The remaining 30% of the material was
mostly unreacted starting material 28 that could be
recycled. Removal of the benzyl carbamate with hydrogen
over palladium hydroxide on carbon gave a 96-100%
yield of the amine 53 as a crystalline white solid.
Treatment of the piperidinocrotonic acid36 54 with 1-(3-
dimethylaminopropyl)-3-ethylcarbodiimide hydrochlo-
ride, hydroxybenzotriazole, and 53 gave a 64% yield of
the free base of 1. Alternatively, treatment of 54 with
oxalyl chloride followed by the amine 53 avoided chro-
matographic purification and afforded 1 as the free base
in the much higher yield of 85%. The salt was prepared
with hydrogen chloride in ether/dichloromethane, and the
hydrochloride was recrystallized from 2-propanol/ethanol
to give material with less than 1% of any single impurity
by analytical HPLC.
Using this synthesis, 9.5 g of the elastase inhibitor 1
was obtained from a 250 g input of (R)-methionine. This
represents the yield obtained from a specific 14-step
synthetic run and does not incorporate the best yields
we have obtained for each step. It represents a 3.8 wt %
yield and a 1.3% chemical yield. If the best conditions
are used in each step, we estimate conservatively that a
ca. 3% chemical yield could be obtained. The entire
synthesis can be achieved in 25 working days (5 weeks)
and requires glassware no larger than 10 L and only a
single chromatographic purification. There are clearly
refinements that can be made to further improve the
efficiency of this process.
Th e Dyn a m ic Resolu tion of a Key In ter m ed ia te
Having previously established the absolute stereo-
chemistry of the desired trans-lactam 28 as that derived
from an R-amino acid of unnatural chirality, the bulk
availability of the amino acid as a starting material and
its cost were evaluated. Other options, such as an
alternative non-amino-acid starting material and intro-
duction of the optical purity by an appropriate chiral
reagent or resolution process were also considered. Of
these, examination of a dynamic resolution of an inter-
mediate used in the existing synthesis was very appeal-
ing because it would result in minimal deviation from
the synthetic route in place and would also provide the
opportunity to use an amino acid of natural chirality.
These are readily available in bulk at lower cost.
(34) Attempts to convert the acid 41 directly into the protected amide
45 via formation of the mixed anhydride and treatment with the
lithium anion of benzyl carbamate failed.
(35) Numerous experiments to effect S-alkylation and intramolecu-
lar cyclisation in a “one-pot” reaction by treatment with methyl iodide,
potassium carbonate, and a variety of polar solvents failed to give
superior yields of the pyrrolidone 47 compared to the two-step
procedure. Interestingly, warming 46 at 70 °C with or without K2CO3
in MeCN leads only to demethylation to give 45. Warming 46 in toluene
at 110 °C for 3 h gives 3:1 demethylation 45 to cyclization 47. Further
experiments to avoid methyl iodide by using trimethylsulfonium iodide
(cf. Abood, N. A.; Flynn, D. L.; Laneman, S. A.; Nosal, R.; Schretzman,
L. A. US patent 5,484,946, 1996) also failed to give acceptable yields
of product.
On the basis of the precedent of induced dynamic
resolution by crystallization,37 the racemic pyrrolidone
55 was deprotected and treated with 0.5 equiv of (+)-di-
p-toluoyltartaric acid (DPTT) to give the racemic salt 56
(Scheme 9). This material was dissolved in THF, and 4
mol % of 3,5-dichloro-2-hydroxybenzaldehyde was added.
After the mixture was stirred for 6 days, a crystalline
precipitate of 57 was obtained in 81% yield. Better yields
(36) The piperidinocrotonic acid 54 was readily prepared in good
yield by reacting ethyl 4-bromocrotonate with piperidine and potassium
carbonate in acetonitrile followed by hydrochloric acid hydrolysis of
the ester.
(37) Reider, P. J .; Davis, P.; Hughes, D. L.; Grabowski, E. J . J . J .
Org. Chem. 1987, 52, 955-957.

Synthesis of Pyrrolidine-trans-lactams
J . Org. Chem., Vol. 64, No. 14, 1999 5175
were obtained with THF as solvent than with methanol
or acetonitrile. Treatment of the salt 57 with N-methyl-
morpholine and methyltrifluoroacetate in dichloromethane/
methanol gave the trifluoroacetamide 31 as 9:1 R/ S from
chiral HPLC analysis. After one recrystallization, the
trifluoracetamide 31 was obtained optically pure. Given
that racemic methionine is dramatically cheaper ($18/
500 g) than (R)-methionine ($49/25 g), this dynamic
resolution process is highly attractive. Further develop-
ments incorporating this process on a large scale are
currently being evaluated.
interest is the stereocontrol in the acyliminium reaction,
the crystallization-induced dynamic resolution, and a
synthesis that only requires a single chromatographic
purification.
Ack n ow led gm en t. We are grateful to Tim Under-
wood, Claire Paterson, and Bob Boughtflower for HPLC
assistance, Ian Davidson for mass spectra, Steve Rich-
1
ards, Andrew McRiner, and Ailish Forristal for H NMR
spectra, Gianpaolo Bravi for obtaining the Macromodel
values, J onathan Goodacre for preparation of interme-
diates, Mike Anson, J ohn Whitehead, and Martin J ones
for useful discussions, and Keith Mills, Harry Finch, and
George Hardy for valuable comments during the prepa-
ration of this manuscript.
Con clu sion
This paper summarizes the investigation of a number
of routes to trans-lactams, leading to the discovery and
development of a synthesis of enantiomerically pure
trans-lactams that, with minor modification, can be
executed on a multikilogram scale. The process can
deliver high-purity compound 1 in bulk. Of particular
Su p p or tin g In for m a tion Ava ila ble: Full experimental
details. This material is available free of charge via the
Internet at http://pubs.acs.org.
J O990306S