General solid-phase synthesis approach to prepare mechanism-based aspartyl protease inhibitor libraries. Identification of potent cathepsin D inhibitors

Article abstract of DOI:10.1021/ja981812m

A general method has been developed for the expedient solid-phase synthesis of aspartyl protease inhibitors that do not incorporate any amino acids. The synthesis approach was designed to allow the introduction of diverse functionality at all four variabl

Full text of DOI:10.1021/ja981812m

J. Am. Chem. Soc. 1998, 120, 9735-9747
9735
General Solid-Phase Synthesis Approach To Prepare
Mechanism-Based Aspartyl Protease Inhibitor Libraries.
Identification of Potent Cathepsin D Inhibitors
Christina E. Lee, Ellen K. Kick,† and Jonathan A. Ellman*
Contribution from the Department of Chemistry, UniVersity of California, Berkeley, California 94720
ReceiVed May 26, 1998
Abstract: A general method has been developed for the expedient solid-phase synthesis of aspartyl protease
inhibitors that do not incorporate any amino acids. The synthesis approach was designed to allow the
introduction of diverse functionality at all four variable sites in inhibitors 1a,b using readily available precursors.
Representative, analytically pure inhibitors were obtained in 45-64% overall yields for the 12-step solid-
phase synthesis sequence that includes both stereoselective reduction and carbon-carbon bond forming steps.
A library of 204 spatially separate compounds was also prepared and screened against cathepsin D and resulted
in the identification of a number of potent small molecule inhibitors (Ki ) 0.7-3 nM). The described synthesis
approach enables the rapid identification of small molecule inhibitors to diverse aspartyl protease targets.
The general design principles are also applicable to other enzyme classes.
Introduction
potential of these library approaches will only be realized if a
full range of diverse functionality can be displayed at all variable
sites about the mechanism-based pharmacophore. In the first
report of a general solid-phase synthesis approach to target the
aspartyl proteases, we described a synthesis approach for
displaying functionality about a preformed hydroxyethylamine
isostere that had a fixed substituent at one of the potential sites
of variability.6 We have successfully used this chemistry to
identify low nanomolar small molecule inhibitors of cathepsin
D.7 Since our initial report, other researchers have reported solid-
phase synthesis routes to display diverse functionality about
mechanism-based pharmacophores to aspartyl proteases8 and
to metalloproteases.9 Herein we describe the first general
synthesis approach to incorporate diverse functionality at all
variable sites about a secondary alcohol pharmacophore targeting
aspartyl proteases. We further demonstrate the utility of this
method by the optimization of small molecule inhibitors
targeting cathepsin D.
Proteases, which catalyze the hydrolysis of amide bonds in
peptides and proteins, play essential roles in numerous biological
processes and are important therapeutic targets in a multitude
of diseases including cancer, viral, parasitic, and bacterial
infections, inflammation, and cardiovascular disease.1-3 One of
the most powerful strategies for designing inhibitors of proteases
is to employ a mechanism-based pharmacophore as a key
binding element that interacts with the active site of the protease.
Pharmacophores have been developed that target each of the
four major protease classes, which are defined according to the
mechanism of protease-catalyzed amide bond hydrolysis. Ex-
amples of mechanism-based pharmacophores used to target
proteases include secondary alcohols for aspartyl proteases,
ketone and aldehyde carbonyls for cysteine and serine proteases,
and hydroxamic acids for zinc metalloproteases.
The synthesis and evaluation of small molecule libraries
employing mechanism-based pharmacophores have begun to
have a significant impact upon the development of protease
inhibitors.4 These initial efforts demonstrated the utility of
designing a solid-phase synthesis about a pharmacophore-based
scaffold. However, these syntheses either were based on
peptidic libraries5 or only accessed a portion of the available
sites of diversity about the pharmacophore structure. The full
Library Design
The aspartyl proteases are a ubiquitous class of enzymes that
play an important role in mammals, plants, fungi, parasites, and
(6) Kick, E. K.; Ellman, J. A. J. Med. Chem. 1995, 38, 1427-1430.
(7) Kick, E. K.; Roe, D. C.; Skillman, A. G.; Liu, G.; Ewing, T. J. A.;
Sun, Y.; Kuntz, I. D.; Ellman, J. A. Chem. Biol. 1997, 4, 297-307.
(8) (a) Wang and co-workers have reported an approach to display
functionality about the diamino diol and diamino alcohol cores to target
HIV-1 protease. In this approach the P1 and P1′ side chains are invariant
(Figure 1). Wang, G. T.; Li, S.; Wideburg, N.; Krafft, G. A.; Kempf, D. L.
J. Med. Chem. 1995, 38, 2995-3002. (b) Rotella has recently reported
preliminary efforts toward a potential approach to introduce P1 side chain
diversity into pharmacophore-based inhibitors by attaching amino alcohols
onto solid support. However, methods to display functionality on the P′
side of the inhibitors were not demonstrated and the synthesis proceeded
in low yield. Rotella, D. P. J. Am. Chem. Soc. 1996, 118, 12246-12247.
(9) A number of research groups have reported strategies to prepare
inhibitors incorporating the hydroxamic acid pharmacophore motif by the
display of functionality upon support-bound hydroxylamine derivatives.
Whittaker, M. Curr. Opin. Chem. Biol. 1998, 2, 386-396.
† Present address: Bristol-Myers Squibb Co., P.O. Box 4000, Princeton,
NJ 08543.
(1) Babine, R. E.; Bender, S. L. Chem. ReV. 1997, 97, 1359-1472.
(2) Dunn, B. M. In Structure and Function of the Aspartic Proteinases:
Genetics, Structures, and Mechanisms; Plenum Press: New York, 1991;
Vol. 306, xviii, 585 pp.
(3) Takahashi, K. In Aspartic Proteinases: Structure, Function, Biology,
and Biomedical Implications; Plenum Press: New York, 1995.
(4) For recent reviews, see: (a) Whittaker, M. Curr. Opin. Chem. Biol.
1998, 2, 386-396. (b) Dolle, R. E. Mol.DiVersity 1997, 2, 223-236.
(5) (a) In seminal work in this area, Owens and co-workers incorporated
statine isosteres into a peptide library to identify HIV protease inhibitors.
Owens, R. A.; Gesellchen, P. D.; Houchins, B. J.; DiMarchi, R. D. Biochem.
Biophys. Res. Commun. 1991, 181, 402-408. (b) Recent examples of
peptidic libraries toward matrix metalloproteases are provided in ref 4.
S0002-7863(98)01812-5 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/05/1998

9736 J. Am. Chem. Soc., Vol. 120, No. 38, 1998
Lee et al.
Figure 1. Common isosteres that mimic the tetrahedral intermediate of amide bond hydrolysis as catalyzed by aspartyl proteases.
Figure 2. Synthesis strategy to incorporate R1, R2, and R3 functionality.6
retroviruses.2,3 Not surprisingly, aspartyl proteases have been
implicated in a number of disease states and serve as important
therapeutic targets. Most notably, HIV-1 protease inhibitors
have had a huge impact on the treatment of AIDS.10 Other
aspartyl proteases that may prove to be important therapeutic
targets include renin for the treatment of hypertension,11
cathepsin D in breast tumor metastasis,12,13 plasmepsins I and
II in malaria,14,15 and Candida albicans proteases in fungal
infection.16 The aspartyl proteases are endopeptidases that use
two aspartic acid residues to catalyze the hydrolysis of amide
bonds through a tetrahedral intermediate (Figure 1). Potent
inhibitors of the aspartyl proteases have been developed that
utilize a secondary alcohol as a stable mimetic of the tetrahedral
intermediate. The alcohol is incorporated within isosteres that
span the P1-P1′ region.17 Common isosteres include statine,
hydroxyethylene, and hydroxyethylamine (Figure 1).
To develop a general solid-phase approach to target aspartyl
proteases, we chose to display diverse functionality about the
hydroxyethylamine isostere (1a,b) (Figures 1 and 2). In contrast
to the other isosteres, the hydroxyethylamine motif readily
allows for the rapid display of diverse functionality at the P1′
site by straightforward amine alkylation. In addition, several
approved HIV protease drugs and clinical candidates are based
upon this structure.18 A key feature of our synthesis approach
is the attachment of the starting scaffold to the solid support
through the secondary hydroxyl group (Figure 2). Linkage
through the secondary alcohol is ideal because it is the only
inVariant part of the inhibitor structure and allows diVersity to
be displayed at all Variable sites of the inhibitor. In addition,
the linker serves to protect the alcohol functionality throughout
the synthesis sequence.
Although our ultimate goal was to develop a synthesis
sequence to incorporate diverse functionality at all four sites of
the hydroxyethylamine isostere (P1, R1, R2, and R3), we initially
reported a general method to introduce R1, R2, and R3
functionality (Figure 2).6 The pendant side chains, R1, R2, and
R3, were incorporated into inhibitor structures 1a,b, after linking
the scaffolds 3a,b to the solid support through the invariant
secondary alcohol. Both diastereomers were used to access final
products 1a and 1b because the preferred alcohol stereochem-
istry, S or R, respectively, depends on both the targeted aspartyl
protease and the overall inhibitor structure. The R1 functionality
was obtained through amine addition into intermediates 2a,b.
Acylation with a variety of acylating agents introduced R2
functionality. Azide reduction released the primary amine,
(10) Palella, F. J.; Delaney, K. M.; Moorman, A. C.; Loveless, M. O.;
Fuhrer, J.; Satten, G. A.; Aschman, D. J.; Holmberg, S. D. New Engl. J.
Med. 1998, 338, 853-860.
(11) Greenlee, W. J. Med. Res. ReV. 1990, 10, 173-236.
(12) Westley, B. R.; May, F. E. B. Eur. J. Cancer 1996, 32, 15-24.
(13) Rochefort, H.; Liaudet, E.; Garcia, M. Enzyme Protein 1996, 49,
106-116.
(14) Francis, S. E.; Gluzman, I. Y.; Oksman, A.; Knickerbockeer, A.;
Sherman, D. R.; Russell, D. G.; Goldberg, D. E. EMBO J. 1994, 13, 306-
317.
(15) Silva, A. M.; Lee, A. Y.; Gulnik, S. V.; Majer, P.; Collins, J.; Bhat,
T. N.; Collins, P. J.; Cachau, R. E.; Luer, K. E.; Gluzman, I. Y.; Francis,
S. E.; Oksman, A.; Goldberg, D. E.; Erickson, J. W. Proc. Natl. Acad. Sci.
U.S.A. 1996, 93, 10034-10039.
(16) Abad-Zapatero, C.; Goldman, R.; Muchmore, S. W.; Hutchins, C.;
Stewart, K.; Navaza, J.; Payne, C. D.; Ray, T. L. Protein Sci. 1996, 5,
640-652.
(17) Wiley: R. A.; Rich, D. H. Med. Res. ReV. 1993, 13, 327-384.
(18) West, M. L.; Fairlie, D. P. Trends Pharm. Sci. 1995, 16, 67-75.

Mechanism-Based Aspartyl Protease Inhibitor Libraries
J. Am. Chem. Soc., Vol. 120, No. 38, 1998 9737
Figure 3. Retrosynthetic strategy of introducing P1 side chain diversity.
Scheme 1a
which was treated with a variety of acylating agents for R3
incorporation. We have synthesized several libraries using this
chemistry and have identified potent inhibitors to the aspartyl
proteases cathepsin D (Ki ) 9-20 nM)7 and plasmepsins I and
II.19
A general solid-phase synthesis strategy designed to incor-
porate diverse functionality at the P1 side chain would allow
for rapid identification of potent inhibitors toward any aspartyl
protease and would enable us to further optimize inhibitors of
cathepsin D and plasmepsins I and II. The importance of
exploring P1 side chain diversity is clearly demonstrated by
Agouron’s development of their approved HIV-1 protease
inhibitor.20 By replacing the benzyl P1 side chain with the
methylthiophenyl side chain, a 10-fold improvement in potency
was observed. To probe the specificity of the P1 side chain
toward aspartyl proteases, a solid-support synthesis of scaffolds
2a,b with diverse P1 side chains is clearly preferred over the
time-consuming multistep preparation of numerous scaffolds
incorporating different P1 side chains. In our initial synthesis
efforts, we delayed developing a general sequence to introduce
diverse P1 side chains due to the difficulty of achieving this
goal. A synthesis approach quite different from previously
reported solution-phase isostere synthesis efforts would be
required due to the necessity of attaching the isostere to the
support through the secondary alcohol. In addition, introduction
of the P1 side chain with the natural S stereochemistry would
require both carbon-carbon bond formation and stereoselective
transformations on solid support.
a (a) LiOH, dioxane/H2O, 99%. (b) i-BuOCOCl, NMM, Me(OMe)NH‚
HCl, CH2Cl2, 99%. (c) 40% Aqueous acetic acid, 85%. (d) MMTCl,
DMAP, pyridine, 65%.
Results and Discussion
To prepare tertiary amides 6a,b, appropriate linkage to solid
support and a primary alcohol protecting group needed to be
selected. For our previously reported synthesis approach we
employed the acid-labile tetrahydropyranyl linkage.6,22 Unfor-
tunately, chelation-controlled reductions of THP-protected R-hy-
droxyketones proceed with poor diastereoselectivity. In contrast,
high diastereoselectivity is observed for reductions of R-hy-
droxyketones protected as ethers including benzyl and substi-
tuted benzyl ethers.23 Therefore, we elected to employ the acid
labile p-alkoxybenzyl linker. The mono-p-methoxytrityl (MMT)
group was selected to protect the primary alcohol, since mild
acid treatment would selectively remove the protecting group
without any cleavage of material from the resin. In addition,
trityl-based protecting groups provide convenient spectropho-
tometric methods for quantitating loading levels and for
confirming complete protecting group cleavage.
Herein, we report a solid-phase synthesis approach to access
diversity at all sites including the P1 site. We envisioned
introduction of the P1 side chain through a synthesis sequence
that would provide the previously employed azide intermediates
2a,b by functional group transformations upon alcohols 4a,b
(Figure 3). Alcohols 4a and 4b would be prepared by chelation
or nonchelation controlled reduction of ketones 5a and 5b,
respectively, to set the appropriate stereochemistry of the P1
side chain. The P1 side chain in ketones 5a,b would be
introduced by Grignard addition to corresponding tertiary amides
6a,b. Since the aspartyl proteases have a hydrophobic P1 pocket,
the synthesis is biased toward use of hydrophobic groups at
the P1 site.21 Grignard reagents are therefore ideal for the
introduction of a large set of diverse hydrophobic side chains
at this site.
To introduce P1 side chain functionality, we chose to develop
a Weinreb amide substrate that would provide a ketone product
after Grignard addition. Enantiomeric Weinreb amides 9a and
9b were prepared in four steps from the commercially available
(S)- and (R)-methyl esters of isopropylidene glycerate, 7a and
7b, respectively (Scheme 1). After saponification of the methyl
esters, activation of the resulting acids with isobutyl chlorofor-
mate and N-methylmorpholine was followed by treatment with
N,O-dimethylhydroxylamine hydrochloride to provide the cor-
responding Weinreb amides. After isopropylidene deprotection
(21) Almost every aspartyl protease (e.g., HIV protease, renin, cathepsins
D and E, plasmepsins I and II, pepsin, etc.) for which substrate specificities
have been performed prefers hydrophobic amino acid side chains at the P1
position. Two reported exceptions are the yeast aspartic protease 3 and the
related mammalian pro-opiomelanocortin converting enzymes that cleave
between basic residues to process prohormones. Loh, Y. P.; Cawley, N.
X.; Friedman, T. C.; Pu, L. P. In Aspartic Proteinases: Structure, Function,
Biology, and Biomedical Implications; Takahashi, K., Ed.; Plenum Press:
New York, 1995; pp 519-527.
(22) Thompson, L. A.; Ellman, J. A. Tetrahedron Lett. 1994, 35, 9333-
9336.
(23) Still, W. C.; McDonald, J. H. Tetrahedron Lett. 1980, 21, 1031-
1034.
(19) Haque, T. S. University of California at Berkeley, Berkeley, CA,
unpublished results.
(20) Kaldor, S. W.; Kalish, V. J.; Davies, J. F.; Shetty, B. V.; Fritz, J.
E.; Appelt, K.; Burgess, J. A.; Campanale, K. M.; Chirgadze, N. Y.;
Clawson, D. K.; Dressman, B. A.; Hatch, S. D.; Khalil, D. A.; Kosa, M.
B.; Lubbehusen, P. P.; Muesing, M. A.; Patick, A. K.; Reich, S. H.; Su, K.
S.; Tatlock, J. H. J. Med. Chem. 1997, 40, 3979-3985.

9738 J. Am. Chem. Soc., Vol. 120, No. 38, 1998
Lee et al.
Scheme 2a
Scheme 4a
a (a) Pyrrolidine, 95%. (b) 40% Aqueous acetic acid, 94%. (c)
MMTCl, DMAP, pyridine, 85%.
Scheme 5a
a (a) PPh3, CBr4, CH2Cl2. (b) 9a or 9b, NaH, Bu4NI, 18-crown-6,
THF, 2 h, 45 °C.
Scheme 3a
a (a) P1MgX, THF, 0 °C. (b) 1% p-TsOH, CH2Cl2. (c) TFA, CH2Cl2.
a (a) 15a or 15b, NaH, Bu4NI, 18-crown-6, THF, 2 h, 45 °C. (b)
P1MgX, THF, 0 °C.
with 40% aqueous acetic acid, primary alcohols 8a,b were
protected using 4-anisylchlorodiphenylmethane (MMTCl) and
(dimethylamino)pyridine (DMAP). The Weinreb amide prod-
ucts 9a,b were then attached to the support using (benzyloxy)-
benzyl bromide resin 10, which was readily prepared from
commercially available Wang resin by treatment with carbon
tetrabromide and triphenylphosphine.24 An 81% loading ef-
ficiency was achieved employing sodium hydride, tetrabuty-
lammonium iodide, and catalytic 18-crown-6 in THF for 2 h at
45 °C (Scheme 2).25
Grignard reagents were added to support-bound R-alkoxy
Weinreb amide 11a to introduce diversity at the P1 site. We
initially encountered two difficulties in obtaining clean Grignard
addition products. First, the aqueous Grignard workup sequence
needed to be modified for solid-phase chemistry. After filtering
away the excess Grignard reagent from the support, direct
addition of a mild acid solution resulted in some overaddition,
presumably due to breakdown of the tetrahedral intermediate
before all of the residual Grignard reagent had been quenched.
This problem was readily solved by consuming residual
Grignard reagent with addition of acetone before the quench
step. Second, Grignard addition and subsequent cleavage from
the solid support provided the desired ketone 12 contaminated
with 7-18% of methyl amide side product 13 depending on
the Grignard reagent that was employed (Scheme 3). Side
product 13 resulted from competitive N-O bond cleavage of
Weinreb amide 11a.26
of pyrrolidine amides 15a,b are straightforward and can readily
be performed on a large scale (Scheme 4). Pyrrolidine amides
15a and 15b were prepared in 76% overall yield in three steps
from commercially available (S)- and (R)-methyl esters of
isopropylidene glycerate, 7a and 7b, respectively. Treatment
of 7a,b with neat pyrrolidine followed by deprotection with 40%
aqueous acetic acid provided amides 14a,b. The only purifica-
tion in the sequence was performed after protection of the
primary alcohol with MMTCl and DMAP in pyridine. Amides
15a,b were coupled to support as previously described for
Weinreb amides 11a,b in 84% loading efficiency (Scheme 5).
Minimal (<2%) racemization28 was observed in the coupling
step as determined by cleavage of amide 16a from support
followed by derivatization with R-methoxy-R-(trifluoromethyl)-
phenylacetyl chloride (MTPACl) and 19F NMR analysis of the
diester product.29 Addition of Grignard reagents to support-
bound amide 16a and subsequent cleavage from the solid
support cleanly provided desired ketone 12 with no overalkyl-
ation product observed.
The next step was to set the stereochemistry of the P1 side
chain. We first chose to develop chemistry to access core
isostere 1a (Figure 2), which has the S secondary alcohol
stereochemistry, since we required this epimer for optimization
of our cathepsin D inhibitors (vide infra). Based on our pro-
posed synthesis strategy, isostere 1a would be derived through
the use of chelation controlled reduction of R,â-dialkoxy ketone
5a (Figure 3).30,31 Zinc borohydride (Zn(BH4)2) was selected
as the reducing agent because the reagent afforded higher
diastereoselectivity compared to less oxophilic reducing agents
such as lithium aluminum hydride. The support-bound ketone
17a was reduced with Zn(BH4)2 in diethyl ether:THF at -20
°C (Scheme 6). To determine reduction diastereoselectivity as
well as to ensure that minimal racemization had occurred during
the Grignard addition and reduction steps, secondary alcohol
In an effort to eliminate amide side product 13, we replaced
the N-methoxy-N-methyl (Weinreb) amide with a pyrrolidine
amide. Romea and co-workers have recently reported that
addition of Grignard reagents to R-alkoxy-substituted pyrrolidine
amides cleanly provides ketones with minimal overaddition, in
contrast to tertiary amides without R-alkoxy substituents.27
Presumably, chelation of the R-alkoxy group to the magnesium
stabilizes the tetrahedral intermediate. Preparation and loading
(24) Ngu, K.; Patel, D. V. Tetrahedron Lett. 1997, 38, 973-976.
(25) Altava, B.; Burgrete, M. I.; Luis, S. V.; Mayoral, J. A. Tetrahedron
1994, 50, 7535-7542.
(28) Extended reaction times can result in 10% racemization.
(29) Dale, J. A.; Mosher, H. S. J. Am. Chem. Soc. 1973, 95, 512-519.
(30) Iida, H.; Yamazaki, N.; Kibayashi, C. J. Org. Chem. 1986, 51,
3769-3771 and references therein.
(31) Pikul, S.; Raczko, J.; Ankner, K.; Jurczak, J. J. Am. Chem. Soc.
1987, 109, 3981-3987 and references therein.
(26) Armstrong has previously reported similar side products for Grignard
additions on solid support. Dinh, T. Q.; Armstrong, R. W. Tetrahedron
Lett. 1996, 37, 1161-1164.
(27) Martin, R.; Pascual, O.; Romea, P.; Rovira, R.; Urpi, F.; Vilarrasa,
J. Tetrahedron Lett. 1997, 38, 1633-1636.

Mechanism-Based Aspartyl Protease Inhibitor Libraries
J. Am. Chem. Soc., Vol. 120, No. 38, 1998 9739
Scheme 6a
Scheme 7a
a (a) Zn(BH4)2, Et2O:THF, -20 °C. (b) NosCl, 4-pyrrolidinopyridine,
CHCl3. (c) NaN3, DMF, 50 °C. (d) 1% p-TsOH, CH2Cl2. (e) NosCl,
pyridine, CHCl3. (f) TFA, CH2Cl2.
a (a) NaH, resin 10, Bu4NI, 18-crown-6, THF, 2 h, 45 °C. (b)
BnMgCl, THF, 0 °C. (c) l-selectride, THF, -78 °C to -20 °C. (d)
NosCl, 4-pyrrolidinopyridine, CHCl3. (e) NaN3, DMF, 50 °C. (f) 2%
p-TsOH, CH2Cl2. (g) NosCl, pyridine, CHCl3. (h) TFA, CH2Cl2.
18 (P1 ) Bn) was cleaved from the support to provide the
corresponding triol product, which was converted to the
triacetate using acetic anhydride and DMAP.32 A 90:10 to 85:
15 diastereoselectivity range was observed by GC analysis of
the triacetate. As expected, chelation-controlled reduction
performed on model substrates in solution provided similar
levels of diastereoselectivity. Zn(BH4)2 reductions performed
on ketone 17a with six different P1 side chains (benzyl,
m-phenoxybenzyl, â-naphthylmethyl, p-phenylbenzyl, p-bro-
mobenzyl, and isobutyl) provided diastereoselectivities ranging
from 90:10 to 80:20. The variability in diastereoselectivity
depended primarily on the batch of Zn(BH4)2 employed.
Conversion of secondary alcohol 18 to azide 19 required
considerable optimization. Mitsunobu reactions performed with
diphenyl phosphoryl azide, hydrazoic acid, or zinc azide
bispyridine complex proceeded poorly or resulted in significant
epimerization. We next investigated alcohol activation and
displacement. To achieve good conversion to the activated
alcohol, it was necessary to employ 4-nitrobenzenesulfonyl
chloride with 4-pyrrolidinopyridine as the catalyst and chloro-
form as the reaction solvent. Poor conversion was observed
with alcohol activation to the tosyl or mesyl alcohol, when the
reaction was performed in other solvents such as methylene
chloride, or when using other catalysts such as DMAP. Rapid
precipitation of DMAP salts occluded reaction sites on the resin,
which prevented good conversion from being achieved. After
primary alcohol activation, displacement with sodium azide was
accomplished in N,N-dimethylformamide at 50 °C. Minimal
competitive elimination of the nosyl alcohol (0-<5%) was
observed even for P1 side chains that contained acidic â-hy-
drogens (i.e., P1 ) CH2Ar). The MMT protecting group was
selectively removed using 1% p-toluenesulfonic acid (p-TsOH)
in CH2Cl2 (w/v). The primary alcohol was then activated to
obtain nosyl alcohol 2a using 4-nitrobenzenesulfonyl chloride
and pyridine in chloroform (Scheme 6). Intermediate 2a can
then be converted to the final inhibitor structures using the
previously reported synthesis sequence (Figure 2). To establish
the fidelity of the chemistry to this intermediate in the synthesis
sequence, nosyl alcohol 3a (P1 ) Bn) was cleaved from the
support and purified by column chromatography in 70% overall
yield based on the loading level of support-bound pyrrolidine
amide 16a.33
To have a completely general synthesis approach targeting
aspartyl proteases it was also necessary to access isostere 1b
(Figure 2), which has the R secondary alcohol stereochemistry.
This isostere is accessible by performing non-chelation-
controlled reduction upon ketone precursor 5b (Figure 3).
Reduction of ketone 5b with hindered reducing agents such as
L-selectride should provide 4b with high diastereoselectivity
based upon literature reports of highly diastereoselective reduc-
tions of R,â-dialkoxy ketones in solution.30,31,34,35 Unfortunately,
L-selectride reduction of ketone 17 (P1 ) Bn) resulted in poor
non-chelation-controlled diastereoselectivity (<60:40). Poor
diastereoselectivity for L-selectride reductions of related R,â-
dialkoxyketones in solution was also observed, indicating that
our initial findings were not due to a support effect. We
therefore investigated the â-hydroxyl protecting group strategy.
In solution-phase model studies, replacement of the MMT group
with the nonchelating but sterically hindered TIPS group also
resulted in poor reduction diastereoselectivity. In contrast, 90:
10 diastereoselectivity was observed when the less hindered
ethoxyethyl (EE) group was employed.
Accordingly, pyrrolidine amide 14b was treated with ethyl
vinyl ether and pyridinium p-toluenesulfonate in CH2Cl2 to
provide amide 20 (Scheme 7). Coupling to support was
accomplished using the conditions established to afford support-
bound Weinreb amide 11a. After Grignard addition with
benzylmagnesium chloride, the resulting support-bound benzyl
ketone 21 was reduced using L-selectride in THF to afford
alcohol 22 with 87:13 diastereoselectivity as determined by GC
analysis of the triacetate produced upon cleavage of the triol
from support followed by acetylation.36 Support-bound alcohol
22 is converted to the azide precursor 2b according to the
conditions used to prepare 2a. Although slightly stronger acidic
conditions, 2% p-TsOH in CH2Cl2 (w/v), were required for
complete removal of the ethoxyethyl group relative to the MMT
group, no cleavage of azide 23 from support was observed under
these reaction conditions. Removal of the ethoxyethyl group
and activation of the primary alcohol were performed as
previously described to obtain the desired nosyl alcohol
(34) Mori, Y.; Kuhara, M.; Takeuchi, A.; Suzuki, M. Tetrahedron Lett.
1988, 29, 5419-5422.
(35) Mead, K.; MacDonald, T. L. J. Org. Chem. 1985, 50, 422-424.
(36) Chelation-controlled reduction of ketone 21 with Zn(BH4)2 provided
the same level of diastereoselectivity that is observed for chelation controlled
reduction of the MMT-protected ketone 17.
(32) Chiral HPLC analysis (P1 ) Bn) showed that 5-6% racemization
had occurred to this point in the synthesis.
(33) When Weinreb amide 11a was employed, a slightly reduced overall
yield was observed (65%) in addition to 8.0% of the aforementioned methyl
amide side product.

9740 J. Am. Chem. Soc., Vol. 120, No. 38, 1998
Lee et al.
Figure 4. Library components.
intermediate 2b, which can be employed directly in the
previously developed synthesis sequence. To demonstrate a
high level of fidelity for the solid-phase chemistry, intermediate
3b (P1 ) Bn) was cleaved from the support and purified by
column chromatography in 68% overall yield based on the initial
loading level of pyrrolidine amide 20.
precedented in the literature.39,40 Therefore, we investigated the
use of a more THF-soluble source of magnesium, (Mg-
(anthracene)(THF)3), and lower Grignard concentrations.39,40 All
of the benzylic Grignard reagents were easily prepared from
gradual addition of the corresponding halide to the magnesium
anthracene THF complex. In solution-phase chemistry, one
problem with using Grignard reagents prepared from the
magnesium anthracene THF complex is the separation of
anthracene byproducts from desired products in the reaction
workup. In contrast, in the solid-support workup process, the
excess Grignard reagent along with any anthracene byproducts
are simply washed away from support-bound products.
After solid-support Grignard addition, a portion of resin for
each P1 side chain was cleaved at ketone 17a to determine
reaction completion. For the 17 Grignard reagents used in
library synthesis, the corresponding ketones were cleanly
obtained, as determined by NMR, except for 7-18% of methyl
amide side product 13 (Scheme 4). Two commercially available
Grignard reagents, neopentylmagnesium chloride and (2-methyl-
2-phenylpropyl)magnesium chloride, did not provide sufficiently
clean addition to Weinreb amide 11a and were not included in
the library synthesis. Conversion of all 17 ketone intermediates
to nosyl alcohols 2a were performed as earlier described. All
scaffolds were cleaved at intermediate 2a, and in every case,
the major product was determined by TLC and NMR to be the
desired alcohol 3a (1H NMR spectra supplied in the Supporting
Information).
Library Synthesis and Screening
Using the aforementioned chemistry, a library of 204
compounds with the general structure 1a was synthesized to
identify cathepsin D inhibitors with optimized side chains at
the P1 pocket (Figure 4). The crystal structure of cathepsin D
with pepstatin bound in the active site shows a large hydrophobic
P1 pocket that appears able to incorporate groups larger than
the benzyl group,37 which had been used in our initial library
syntheses toward cathepsin D.7 Seventeen different Grignard
reagents were chosen on the basis of structural features for
cathepsin D inhibition and/or to demonstrate the generality of
the chemistry. From our earlier libraries of cathepsin D inhibi-
tors, a representative set of the most active side chains at R1,
R2, and R3 were incorporated into the library of 204 compounds.
For the library synthesis, the Weinreb amide 11a was
employed because the pyrrolidine amide system had not yet
been developed. Since the methyl amide side product resulting
from Grignard addition to the Weinreb amide 11a (Scheme 3)
does not contain either the P1 or R3 side chain, the methyl amide
side product does not significantly inhibit cathepsin D.
Introduction of R1, R2, and R3 was performed as previously
reported (Scheme 8).7 The amine displacement to incorporate
R1 was performed in parallel in sealed vials with heating in
N-methylpyrrolidinone (NMP). The remaining reactions to
introduce R2 and R3 were performed in a spatially separate array
Of the 17 Grignard reagents used in the library, 5 were
commercially available and the remaining 12 were readily
prepared by one of two methods (Figure 4). The alkyl Grignard
reagents were synthesized from gradual addition of the corre-
sponding halide to activated magnesium turnings in diethyl
ether.38 In general, this method did not efficiently provide the
more activated, substituted benzylic-type Grignard reagents as
(38) Baker, K. V.; Brown, J. M.; Hughes, N.; Skarnulis, A. J.; Sexton,
A. J. Org. Chem. 1991, 56, 698-703.
(39) Gallagher, M. J.; Harvey, S.; Raston, C. L.; Sue, R. E. J. Chem.
Soc., Chem. Commun. 1988, 289-290 and references therein.
(40) Harvey, S.; Junk, P. C.; Raston, C. L.; Salem, G. J. Org. Chem.
1988, 53, 3134-3140 and references therein.
(37) Baldwin, E. T.; Bhat, N.; Gulnik, S.; Hosur, M. V.; Sowder, R. C.
I.; Cachau, R. E.; Collins, J.; Silva, A. M.; Erickson, J. W. Proc. Natl.
Acad. Sci. U.S.A. 1993, 90, 6796-6800.

Mechanism-Based Aspartyl Protease Inhibitor Libraries
J. Am. Chem. Soc., Vol. 120, No. 38, 1998 9741
Table 1a
Scheme 8a
a (a) R1NH2, NMP, 80 °C. (b) R2CO2H, HOAt, PyBOP, i-Pr2EtN,
NMP or R2NCO, NMP. (c) SnCl2, PhSH, NEt3, THF. (d) R3CO2H,
HOAt, PyBOP, i-Pr2EtN, NMP. (e) TFA, CH2Cl2.
using a 96-well filter apparatus.41 After acylation with R2
functionality, support-bound azide 24 was reduced using tin-
(II) chloride, thiophenol, and triethylamine in THF. The
resulting amine was then acylated with the R3 carboxylic acids.
The library of compounds was resubjected to the coupling
conditions for both R2 and R3 acylations to ensure complete
reactions. The compounds were then cleaved from the resin
using TFA:CH2Cl2 (1:1).
The library was screened for cathepsin D inhibition at 330,
100, 33, and 10 nM in a high-throughput fluorescence assay.7,42
Several of the most potent inhibitors were synthesized on larger
scale using the R-alkoxy pyrrolidine amide route, purified by
chromatography, and characterized by NMR, mass spectrometry,
and elemental analysis (Table 1). The overall yields of the
scaled-up inhibitors ranged from 45 to 64% for the 12-step solid-
phase synthesis sequence and were determined by the mass bal-
ance of purified final product compared to the initial loading
of starting pyrrolidine amide 16a (Schemes 5, 6, and 8). Com-
pounds for scale-up were predominantly chosen on the basis of
potency in the library screen and to provide SAR. In addition,
inhibitor 31 was synthesized on large scale to demonstrate the
synthesis generality using alkyl Grignard reagents. Inhibition
constants (Ki) were determined for each of these compounds
(Table 1).
As previously described (Schemes 6 and 7), the reduction
step is not completely stereoselective. Therefore, in library
screening, the inhibitors were contaminated with minor diaster-
eomeric byproducts that incorporate P1 side chains with the
unnatural R stereochemistry. To establish that the minor
diastereomers do not contribute to inhibition in library screening,
the Ki of 26, the minor diastereomer of the most potent inhibitor
25 (vide infra), was determined to be >5 µM, clearly demon-
strating the importance of side chain stereochemistry for aspartyl
protease inhibition.
Several inhibitors that were prepared on large scale had Ki
values of <3 nM (Table 1). The most potent compound, 25,
had a Ki of 0.7 nM. In comparison with the most potent
compound we had previously reported, the potency has increased
greater than 4-fold.43 For the most potent inhibitors from our
library (inhibitors 25, 27, and 28), the larger hydrophobic side
chains at P1 were preferred over the smaller benzyl side chain
(inhibitor 32).
Conclusions
In summary, we have reported a general synthesis strategy
that successfully allows for full diversification about the
a (*) The Ki for compound 26 was determined to establish that the
diastereomer with the unnatural stereochemistry at the P1 site does not
inhibit the enzyme. Since the minor diastereomer was isolated from a
diastereomeric mixture, the yield was not determined. (**) Inhibitors
32-35 with benzyl P1 side chains were synthesized and purified on
large scale in previously reported work.7
(41) Boojamra, C. G.; Burow, K. M.; Thompson, L. A.; Ellman, J. A. J.
Org. Chem. 1997, 62, 1240-1256.
(42) Krafft, G. A.; Wang, G. T. Methods Enzymol. 1994, 241, 70-86.
(43) A few of the compounds with the benzyl side chain at P1 (e.g.,
inhibitor 35) were assayed previously.7 Different batches of cathepsin D
isolated from human liver were used in the previous work and the work
reported here. This resulted in somewhat different Ki values. For example,
inhibitor 35 was determined to have a Ki value of 9 ( 2 nM in previously
reported work and was determined to have a Ki value of 3.0 ( 0.2 nM in
this study.
hydroxyethylamine aspartyl protease isostere using a 12-step
solid-phase synthesis sequence that includes both stereoselective
reduction and carbon-carbon bond forming steps. Using this
synthesis approach, a diverse array of hydrophobic functionality

9742 J. Am. Chem. Soc., Vol. 120, No. 38, 1998
Lee et al.
hexanes:ethyl acetate, then 67:33) to yield 11.1 g (94% for two steps)
of the product as a yellow oil. 1H NMR (300 MHz): δ 1.42 (s, 3H),
1.48 (s, 3H), 3.19 (s, 3H), 3.70 (s, 3H), 4.05 (m, 1H), 4.24 (m, 1H),
4.84 (apparent t, J ) 6.6, 1H). 13C NMR (101 MHz): δ 25.62, 25.69,
61.30, 66.73, 73.13, 110.43, 110.76, 171.15.
can readily be incorporated into the P1 side chain followed by
a wide range of diversity at the three other sites. By performing
the synthesis in a library format, several potent cathepsin D
inhibitors have been identified with optimized P1 side chains.
We are currently using the synthesis approach to optimize
binding and pharmacokinetic properties of inhibitors targeting
the malarial proteases, plasmepsins I and II. This synthesis
approach demonstrates the utility of designing a general, solid-
phase synthesis about a mechanism-based pharmacophore by
linking through an essential, invariant portion of the pharma-
cophore. In due course, we will report an extension of this
general design strategy to library synthesis approaches targeting
other enzyme classes including the cysteine and serine protease
classes.
(2S)-2,3-Dihydroxy-N-methoxy-N-methylpropanamide (8a).
A
solution of 21.0 g (0.111 mol) of (2S)-2,3-O-isopropylidene-N-methoxy-
N-methylpropanamide in 500 mL of 40% aqueous acetic acid was
stirred at 35 °C for 10 h. The solution was concentrated, and toluene
(5 × 50 mL) was added to form an azeotrope with acetic acid during
the concentration step. The crude yellow oil was purified by flash
chromatography (99:1 CH2Cl2:methanol, then 97:3 CH2Cl2:methanol)
to afford 14.1 g (85%) of the diol as a colorless oil. IR: 3381, 2942,
1650 cm-1 1H NMR (300 MHz): δ 3.23 (s, 3H), 3.68 (dd, J ) 5.1,
.
11.5, 1H), 3.70 (s, 3H), 3.84 (dd, J ) 3.2, 11.5, 1H), 4.48 (dd, J )
3.2, 5.1, 1H). 13C NMR (101 MHz): δ 32.32, 53.37, 61.30, 64.02,
172.14. Anal. Calcd for C5H11N1O4: C, 40.27; H, 7.43; N, 9.39.
Found: C, 40.46; H, 7.31; N, 9.15.
Experimental Section
(2S)-2-Hydroxy-N-methoxy-N-methyl-3-mono-p-methoxytrityl-
propanamide (9a). To a solution of 13.3 g (43.1 mmol) of 4-anisyl-
chlorodiphenylmethane (MMTCl) in 140 mL of pyridine at room
temperature was added a solution of 5.50 g (36.9 mmol) of 8a in 20
mL of pyridine followed by the addition of 0.225 g (1.84 mmol) of
DMAP. The reaction mixture was stirred for 4.5 h, diluted in 150 mL
of CH2Cl2, washed with 1.0 M citric acid (8 × 200 mL), saturated
NaHCO3 (2 × 150 mL), and brine (1 × 150 mL), dried over Na2SO4,
and concentrated. The crude yellow oil was purified by chromatography
(80:19:1 hexanes:ethyl acetate:triethylamine, then 67:32:1 hexanes:ethyl
acetate:triethylamine) to provide 10.2 g (66%) of 9a as a white solid.
1H NMR (400 MHz): δ 3.21 (s, 3H), 3.28 (dd, J ) 4.1, 9.4, 1H), 3.39
(dd, J ) 2.8, 9.4, 1H), 3.44 (s, 3H), 3.72 (d, J ) 8.2, 1H), 3.78 (s,
3H), 4.54 (m, 1H), 6.81 (m, 2H), 7.18-7.21 (m, 2H), 7.27-7.33 (m,
6H), 7.44 (m, 4H). 13C NMR (101 MHz): δ 55.18, 60.96, 65.38, 69.09,
85.97, 113.03, 126.83, 127.75, 128.44, 128.45, 130.30, 135.49, 144.21,
158.48. Anal. Calcd for C25H27N1O5: C, 71.24; H, 6.46; N, 3.32.
Found: C, 71.15; H, 6.29; N, 3.37.
(2S)-2,3-O-Isopropylidene-N,N-tetramethylenepropanamide. A
solution of 14.8 g (0.0924 mol) of 7a in 9.26 mL (0.111 mol) of
pyrrolidine was stirred at room temperature for 30 h. The solution
was concentrated to a yellow oil, and the crude product was carried on
to the next step without purification. However, the yellow oil can be
purified by chromatography (67:33 hexanes:ethyl acetate) to provide
the product as a colorless oil. 1H NMR (500 MHz): δ 1.38 (s, 3H),
1.41 (s, 3H), 1.80-1.98 (m, 4H), 3.46 (m, 3H), 3.66 (m, 1H), 4.14
(dd, J ) 6.9, 8.3, 1H), 4.28 (dd, J ) 6.5, 8.3, 1H), 4.59 (apparent t, J
) 6.7, 1H). 13C NMR (125 MHz): δ 23.33, 25.21, 25.33, 25.72, 45.75,
45.89, 65.97, 74.04, 109.92, 167.20. Anal. Calcd for C10H17N1O3:
C, 60.28; H, 8.60; N, 7.03. Found: C, 60.40; H, 8.38; N, 7.09.
General Methods. Materials were obtained from commercial
suppliers and employed without further purification unless otherwise
stated. Wang resin was purchased from NovaBiochem (San Diego,
CA). Sodium borohydride (NaBH4) was purchased from Alfa AESAR
(Ward Hill, MA). Anhydrous N,N-dimethylformamide (DMF) and
N-methylpyrrolidinone (NMP) were purchased from Aldrich (Milwau-
kee, WI). The following solvents were distilled under N2 from specified
drying agents: tetrahydrofuran (THF) and diethyl ether (Et2O) from
sodium/benzophenone ketyl, dioxane from sodium, and methylene
chloride (CH2Cl2) and pyridine from calcium hydride. Chloroform
(CHCl3) was passed through basic alumina prior to use. Flash column
chromatography was carried out with Merck 60 230-400-mesh silica
gel according to the procedure reported by Still.44 GC analysis was
performed on a Hewlett-Packard 5890 Series II gas chromatograph
using a Hewlett-Packard Ultra 2 column. Infrared spectra (IR) were
recorded with a Perkin-Elmer 1600 Series Fourier transform spectrom-
eter using KBr pellets or NaCl plates, and only partial spectral data is
listed. 1H and 13C NMR spectra were recorded using a UCB Bruker
AMX-300, AMX-400, AM-400, or DRX-500 spectrometer. 19F NMR
spectra were obtained with a UCB Bruker AM-400. All spectra were
obtained in CDCl3, and chemical shifts are recorded in parts per million
relative to the internal solvent peak. Coupling constants are reported
in hertz. Elemental analyses were performed by M-H-W Laboratories
(Phoenix, AZ).
The following procedures were employed to synthesize both S and
R isomers. Only one example is provided below.
(2S)-2,3-O-Isopropylidene-N-methoxy-N-methylpropanamide. To
a solution of 10.0 g (62.4 mmol) of the (S)-methyl ester of isopropyl-
idene glycerate (7a) in 250 mL of dioxane:H2O (1:1) at 0 °C was added
2.88 g (68.6 mmol) of lithium hydroxide monohydrate. The reaction
mixture was stirred for 12 h at 4 °C followed by the addition of 1.0 M
citric acid (70 mL) to acidify the reaction solution (pH ) 4). The
solution was saturated with NaCl and then washed with ethyl acetate
(8 × 75 mL). The organic extracts were dried over Na2SO4 and
concentrated to afford a yellow oil that was carried onto the next step.
1H NMR (300 MHz): δ 1.39 (s, 3H), 1.50 (s, 3H), 4.16 (dd, J ) 4.8,
8.8, 1H), 4.27 (dd, J ) 7.4, 8.8, 1H), 4.61 (dd, J ) 4.8, 7.4, 1H),
10.03 (broad s, 1H).
To a solution of 9.10 g (0.0624 mol) of the corresponding carboxylic
acid in 230 mL of CH2Cl2 at -11 °C (ethanol-ice bath) was added
14.4 mL (0.131 mol) of N-methylmorpholine followed by 8.09 mL
(0.0624 mol) of isobutyl chloroformate. The reaction mixture was
stirred for 15 min, changing from yellow to dark orange in color. To
the reaction solution was added 6.70 g (0.0690 mol) of N,O-
dimethylhydroxylamine hydrochloride at -11 °C. The mixture was
stirred for 8 h at -11 °C with warming to room temperature. The
reaction solution was poured into vigorously stirring H2O (150 mL).
The layers were separated, and the aqueous layer was extracted with
CH2Cl2 (2 × 75 mL). The combined organic layers were dried over
Na2SO4, concentrated, and purified by flash chromatography (80:20
(2S)-2,3-Dihydroxy-N,N-tetramethylenepropanamide (14a).
A
solution of 18.0 g (0.0904 mol) of (2S)-2,3-O-isopropylidene-N,N-
tetramethylenepropanamide in 300 mL of 40% aqueous acetic acid was
stirred with heating to 35 °C for 12 h. The solution was concentrated,
and toluene (5 × 50 mL) was added to form an azeotrope with acetic
acid during the concentration step. The crude yellow oil was purified
by flash chromatography (99:1 CH2Cl2:methanol, then 97:3 CH2Cl2:
methanol) to afford 13.1 g (89% for two steps) of 14a as a white solid.
1H NMR (500 MHz): δ 1.78-1.97 (m, 4H), 3.36 (m, 1H), 3.43 (m,
1H), 3.50 (m, 2H), 3.72 (dd, J ) 4.7, 11.8, 1H), 3.80 (dd, J ) 3.0,
11.8, 1H), 4.46 (m, 1H), 5.94 (broad s, 2H). 13C NMR (125 MHz): δ
23.43, 25.62, 45.86, 45.90, 63.62, 70.66, 170.22. Anal. Calcd for
C7H13N1O3: C, 52.82; H, 8.23; N, 8.80. Found: C, 52.65; H, 8.00; N,
8.71.
(R,R)-(+,+)-MTPA Diester.29 The following procedure was used
for small-scale preparation to determine the enantiopurity of starting
material pyrrolidine amide 14 and enantiopurity of pyrrolidine amide
14 after cleavage from support. To a solution of 4.0 mg (0.025 mmol)
of pyrrolidine amide 14 in 75 µL of CH2Cl2 was added 0.20 mL (2.5
mmol) of pyridine followed by 24 µL (0.13 mmol) of (R)-(+)-R-
methoxy-R-(trifluoromethyl)phenylacetyl chloride ((R)-(+)-MTPACl).
The reaction solution was stirred at room temperature for 8 h, at which
time 20 µL (0.20 mmol) of 3-(dimethylamino)-1-propylamine was
(44) Still, W. C.; Kahn, M.; Mitra, A. J. Org. Chem. 1978, 43, 2923-
2925.

Mechanism-Based Aspartyl Protease Inhibitor Libraries
J. Am. Chem. Soc., Vol. 120, No. 38, 1998 9743
added and was stirred for 5 min to convert excess (R)-(+)-MTPACl
into basic amide to be removed in acidic workup. The reaction mixture
was diluted in ether, washed with cold 1 M HCl, saturated Na2CO3,
and brine, dried over Na2SO4, and concentrated.
Determination of Enantiopurity of Starting Material Pyrrolidine
Amide 14a,b. Diastereomer S,R,R Prepared from Starting Material
(2S)-2,3-Dihydroxy-N,N-tetramethylenepropanamide (14a). 1H NMR
(400 MHz): δ 1.85 (m, 2H), 1.96 (m, 2H), 3.27 (s, 3H), 3.37-3.58
(m, 3H), 3.63 (s, 3H), 3.72 (m, 1H), 4.47 (dd, J ) 7.8, 11.9, 1H), 4.67
(dd, J ) 4.5, 11.9, 1H), 5.44 (dd, J ) 4.5, 7.8, 1H), 7.36-7.61 (m,
10H). 19F NMR (327 MHz): δ -9.13, -8.96. 19F NMR integration
showed 2.5% of the other diastereomer (R,R,R).45
Diastereomer S,R,R Prepared from 14a after Cleavage from the
Support. 19F NMR integration showed 2.8-3.3% of the other dia-
stereomer (R,R,R), indicating that <2.0% racemization occurred during
the support coupling step.
Diastereomer R,R,R prepared from Starting Material (2R)-2,3-
Dihydroxy-N,N-tetramethylenepropanamide (14b). 1H NMR (400
MHz): δ 1.85 (m, 2H), 1.95 (m, 2H), 3.35 (s, 3H), 3.33-3.53 (m, 3H),
3.50 (s, 3H), 3.63 (m, 1H), 4.59 (dd, J ) 8.1, 12.1, 1H), 4.79 (m, 1H),
5.47 (m, 1H), 7.38-7.59 (m, 10H). 19F NMR (327 MHz): δ -9.72,
-8.84. 19F NMR integration showed 1.5% of the other diastereomer
(S,R,R).
(2S)-2-Hydroxy-3-mono-p-methoxytrityl-N,N-tetramethylenepro-
panamide (15a). To a solution of 28.9 g (93.5 mmol) of MMTCl in
260 mL of pyridine at room temperature was added a solution of 13.5
g (85.0 mmol) of 14a in 40 mL of pyridine followed by the addition
of 0.610 g (5.00 mmol) of DMAP. The reaction mixture was stirred
for 7 h, diluted with 300 mL of CH2Cl2, washed with 1.0 M citric acid
(8 × 400 mL), saturated NaHCO3 (2 × 300 mL), and brine (1 × 300
mL), dried over Na2SO4, and concentrated. The crude yellow oil was
purified by chromatography (80:19:1 hexanes:ethyl acetate:triethyl-
amine, then 67:32:1 hexanes:ethyl acetate:triethylamine) to provide 31.2
g (85%) of 15a as a white solid. 1H NMR (400 MHz): δ 1.84 (m,
4H), 3.07 (m, 1H), 3.30 (m, 3H), 3.48 (m, 1H), 3.58 (m, 1H), 3.79 (s,
3H), 3.90 (d, J ) 7.7, 1H), 4.35 (m, 1H), 6.82 (m, 2H), 7.20-7.23 (m,
2H), 7.26-7.38 (m, 6H), 7.43 (m, 4H). 13C NMR (101 MHz): δ 23.85,
25.93, 45.83, 46.31, 55.20, 65.45, 69.45, 86.29, 110.46, 113.08, 126.91,
127.79, 128.40, 130.29, 135.32, 144.08, 144.12, 158.56, 170.58. Anal.
Calcd for C27H29N1O4: C, 75.15; H, 6.77; N, 3.25. Found: C, 75.26;
H, 6.88; N, 3.20.
General Solid-Phase Chemistry Methods. All solid-phase reaction
mixtures were stirred at the slowest rate. For the general solid-phase
workup procedure, excess reagent and the reaction solution were filtered
away from support-bound material using polypropylene cartridges with
70-µm PE frits (Spe-ed Accessories) attached to Teflon stopcocks.
Cartridges and stopcocks were purchased from Applied Separations
(Allentown, PA). The support-bound material was thoroughly washed
with various solvents as described below. If the support-bound material
was carried directly on to the next step, the last rinse was the solvent
of the next reaction (3×). For each solid-phase reaction in the synthesis
of intermediate 2, a general procedure based on 1.0 g of support-bound
material is provided. A specific example with appropriate characteriza-
tion for each general method is also reported.
(0.90 mmol) of tetrabutylammonium iodide (Bu4NI), 0.002 g (0.009
mmol) of 18-crown-6, and lastly 3.2 g of resin 10. The reaction mixture
was stirred at 45 °C for 2 h and then was transferred into one or more
polypropylene cartridges with 70-µm PE frits (Spe-ed Accessories)
attached to Teflon stopcocks for the workup process. The resin was
washed with THF (1 × 30 mL), THF:H2O (2:1) (3 × 30 mL), THF:
H2O (1:1) (3 × 30 mL), THF:H2O (1:2) (3 × 30 mL), THF:H2O (2:1)
(3 × 30 mL), THF (1 × 30 mL), DMF (3 × 30 mL), CH2Cl2 (5 × 30
mL), and MeOH (2 × 30 mL). The resin was dried under vacuum
and stored at -20 °C. The exact loading level of the resin was
determined by subjecting a portion of the resin (0.271 g) to MMT group
deprotection and the cleavage conditions reported below. The crude
material obtained from the cleavage step was purified by flash
chromatography (99:1 CH2Cl2:methanol, then 97:3 CH2Cl2:methanol)
to afford 0.021 g (0.141 mmol) of the corresponding Weinreb amide
diol (8a) as a colorless oil (0.52 mequiv/g, 81% loading efficiency based
on a 0.86 mequiv/g loading level of Wang resin).
Support-Bound Pyrrolidine Amide 16a. To a suspension of 0.650
g (27.1 mmol) of NaH in 65 mL of THF at room temperature was
added a solution of 11.7 g (27.1 mmol) of amide 15a in 30 mL of
THF. The suspension was stirred for 50 min followed by the addition
of 1.00 g (2.71 mmol) of Bu4NI, 0.007 g (0.03 mmol) of 18-crown-6,
and lastly 9.00 g of resin 10. The reaction mixture was stirred at the
slowest rate at 45 °C for 2 h and then was transferred into one or more
polypropylene cartridges with 70-µm PE frits (Spe-ed Accessories)
attached to Teflon stopcocks for the workup process. The resin was
washed with THF (1 × 30 mL), THF:H2O (2:1) (3 × 30 mL), THF:
H2O (1:1) (3 × 30 mL), THF:H2O (1:2) (3 × 30 mL), THF:H2O (2:1)
(3 × 30 mL), THF (1 × 30 mL), DMF (3 × 30 mL), CH2Cl2 (5 × 30
mL), and MeOH (2 × 30 mL). The resin was dried under vacuum
and stored at -20 °C. The exact loading level of the resin was
determined by subjecting a portion of the resin (0.460 g) to MMT group
deprotection and the cleavage conditions reported below. The crude
material obtained from the cleavage step was purified by flash
chromatography (99:1 CH2Cl2:methanol, then 97:3 CH2Cl2:methanol)
to afford 0.039 g (0.245 mmol) of the corresponding pyrrolidine amide
diol (14a) as a colorless oil (0.53 mequiv/g, 84% loading efficiency
based on a 0.86 mequiv/g loading level of Wang resin).
General Procedure for MMT Group Deprotection and Cleaving
Material from the Support.46 MMT group deprotection was ac-
complished using 1% (w/v) p-toluenesulfonic acid (p-TsOH) in CH2-
Cl2, which was prepared by dissolving p-TsOH in minimal MeOH
followed by dilution with CH2Cl2. To resin solvated in minimal CH2-
Cl2 was added 1% p-TsOH solution to provide a bright orange solution
indicating presence of mono-p-methoxytrityl cation. The resin was
subjected to 1% p-TsOH in CH2Cl2 (4 × 5 min) until most of the orange
solution had dissipated followed by washes with 3% MeOH in CH2-
Cl2 (3×) and CH2Cl2 (5×).
Cleavage of material from the support was performed as follows.
To resin solvated in minimal CH2Cl2 was added trifluoroacetic acid:
CH2Cl2 (1:1). The reaction mixture was stirred for 30 min followed
by filtration of the solution. The resin was washed with CH2Cl2 (5×).
The filtrate and washings were combined and immediately concentrated.
Toluene was added to form an azetrope with trifluoroacetic acid during
the concentration step.
Grignard Addition to Tertiary Amide (17). General Procedure
with Grignard Reagent Concentrations of >0.2 M. To 1.0 g of
support-bound tertiary amide 6 solvated in 10 mL of THF at 0 °C was
added P1MgX (0.20-0.25 M) (for preparation of Grignard reagents,
see Library Synthesis). The reaction mixture was stirred for 10 h at 4
°C. The reaction solution was removed and the resin was washed with
THF (1 × 15 mL), acetone (3 × 15 mL), 0.28 M hydrocinnamic acid
in THF (3 × 15 mL), DMF (3 × 15 mL), and CH2Cl2 (5 × 15 mL).
P1 ) Bn. To support-bound Weinreb amide 11a (1.0 g, 0.52 mmol)
solvated in 10 mL of THF at 0 °C was added 1.2 mL (2.4 mmol) of
2.0 M benzylmagnesium chloride in THF. The reaction mixture was
stirred at 4 °C for 8 h. After filtration of the reaction solution, the
resin was washed with THF (1 × 15 mL), acetone (3 × 15 mL), 0.28
p-(Benzyloxy)benzyl Bromide Resin (10).24 To 3.25 g (2.80 mmol,
loading 0.86 mequiv/g) of Wang (p-(benzyloxy)benzyl alcohol) resin
solvated in 27 mL of CH2Cl2 cooled to 0 °C was added 2.32 g (6.99
mmol) of carbon tetrabromide followed by a solution of 1.65 g (6.29
mmol) of triphenylphosphine in 5 mL of CH2Cl2. The reaction mixture
was stirred at room temperature for 3 h, washed with CH2Cl2 (6 × 30
mL), dried under vacuum, and employed in the coupling step within
48 h of preparation.
Support-Bound Weinreb Amide 11a. To a suspension of 0.22 g
(9.0 mmol) of NaH in 20 mL of THF at room temperature was added
a solution of 3.8 g (9.0 mmol) of amide 9a in 10 mL of THF. The
suspension was stirred for 50 min followed by the addition of 0.33 g
(45) The (S)-methyl ester of isopropylidene glycerate (7a), which was
obtained from Aldrich, was deprotected with 40% aqueous acetic acid and
treated separately with (R)-(+)-MTPACl and (S)-(-)-MTPACl to form the
S,R,R and S,S,S diastereomers, respectively. 19F NMR analysis indicated
that 2.3% of the (R)-methyl ester enantiomer was present.
(46) It should be noted that MMT deprotection and spectrophotometric
quantitation of the MMT cation provided a false, high loading level
presumably because some MMT material was entrapped in the resin.

9744 J. Am. Chem. Soc., Vol. 120, No. 38, 1998
Lee et al.
M hydrocinnamic acid in THF (3 × 15 mL), DMF (3 × 15 mL), and
CH2Cl2 (5 × 15 mL). A portion of the support-bound material was
deprotected (MMT group) and cleaved (see General Procedure for
MMT group deprotection and cleaving material from the support). The
crude oil was purified by column chromatography (99:1 CH2Cl2:
methanol, then 97:3 CH2Cl2:methanol) to provide benzyl ketone 12 as
Aldrich, originally contained 2.3% of the enantiomer. Therefore, during
our synthesis sequence, 5-6% racemization occurs. HPLC condi-
tions: Chiralcel OD column (Diacel Chemical Industries). Solvent:
99:1 hexanes:2-propanol. Flow Rate: 1.0 mL/min. Retention times
(major:minor) 14.2 min:17.2 min. Integration: 95:5 (major:minor).
General Procedure for Secondary Alcohol Activation and Azide
Displacement (19). To 1.0 g of support-bound secondary alcohol 18
solvated in 10 mL of CHCl3 was added 1.11 g (7.50 mmol) of
4-pyrrolidinopyridine followed by the addition of 1.11 g (5.00 mmol)
of 4-nitrobenzenesulfonyl chloride (NosCl) at room temperature. The
reaction mixture was stirred for 9 h. After filtration of excess reagent,
the resin was washed with CH2Cl2 (5 × 15 mL) and DMF (4 × 15
mL).
a colorless oil. IR: 3345, 1716, 1454, 1405 cm-1 1H NMR (400 MHz):
.
δ 3.84 (d, J ) 16.1, 1H), 3.90 (d, J ) 16.1, 1H), 3.94 (d, J ) 3.4, 2H),
4.31 (t, J ) 3.4, 1H), 7.19-7.35 (m, 5H). 13C NMR (101 MHz): δ
45.12, 63.47, 77.16, 127.29, 128.75, 129.49, 132.77, 207.87. Anal.
Calcd for C10H12O3: C, 66.65; H, 6.71. Found: C, 66.87; H, 6.83.
(2S)-2,3-Dihydroxy-N-methylpropanamide (13). 1H NMR (400
MHz): δ 2.89 (d, J ) 5.0, 3H), 3.84 (d, J ) 6.4, 11.3, 1H), 3.94 (t, J
) 3., 1H), 4.36 (dd, J ) 3.7, 6.4, 1H), 7.19-7.35 (m, 5H).
General Procedure with Grignard Reagent Concentrations of
<0.2 M. To 1.0 g of support-bound tertiary amide 6 solvated in
minimal THF at 0 °C was added 5 equiv of P1MgX (0.05-0.15 M).
The reaction mixture was stirred for 20 h at 4 °C. The reaction solution
was removed, and the resin was washed with THF (1 × 15 mL), acetone
(3 × 15 mL), 0.28 M hydrocinnamic acid in THF (3 × 15 mL), DMF
(3 × 15 mL), and CH2Cl2 (5 × 15 mL).
To 1.0 g of support-bound secondary nosyl alcohol solvated in 10
mL of DMF was added 0.650 g (10.0 mmol) of sodium azide. The
reaction mixture was stirred at 45 °C for 24 h. Following filtration of
the reaction solution, the resin was washed with DMF (2 × 15 mL),
DMF:H2O (1:1) (3 × 15 mL), DMF (2 × 15 mL), and CH2Cl2 (5 ×
15 mL). A portion of the resin was deprotected (MMT group) and
cleaved to determine reaction completion. The crude oil can be purified
by column chromatography (99:1 CH2Cl2:methanol, then 97:3 CH2-
Cl2:methanol) to provide the corresponding azide as a colorless oil. P1
Zinc Borohydride (Zn(BH4)2) Preparation.47 To a suspension of
2.32 g (61.3 mmol) of NaBH4 in 33 mL of Et2O at room temperature
was added 44 mL of a ZnCl2 solution (0.69 M) in Et2O. The ZnCl2
solution was prepared from recrystallized ZnCl2 which was fused
multiple times and heated in Et2O at reflux for 45 min. The suspension
was stirred for 48 h and could be stored at -20 °C for up to 3 weeks.
General Procedure for Zn(BH4)2 Reduction (18). To 1.0 g of
17a in 2.5 mL of THF at -20 °C was added 7.5 mL (3.0 mmol) of
0.40 M Zn(BH4)2 in Et2O. The reaction mixture was stirred at -20
°C for 20 h and then was allowed to warm to 0 °C over 1.5 h. Excess
reagent was filtered away, and the resin was washed with THF (1 ×
15 mL), ethanolamine:H2O:THF (10:2:88 by volume) (3 × 15 mL, 15
min per wash), DMF (3 × 15 mL), and CH2Cl2 (5 × 15 mL). A portion
of the support-bound material was deprotected (MMT group) and
cleaved to determine product purity and reaction completion. Due to
poor resin swelling in Et2O and the ability for resin to solidify or
crystallize at low temperatures, the resin was resubjected to the Zn-
(BH4)2 reduction conditions if the first subjection did not provide
complete reduction as indicated by the TLC of crude product from
cleavage. Crude product can be purified by chromatography (99:1 CH2-
Cl2:methanol, then 97:3 CH2Cl2:methanol) to provide the corresponding
) Bn. IR: 3381, 2930, 2109, 1456 cm-1 1H NMR (300 MHz): δ
.
2.95 (dd, J ) 7.9, 13.6, 1H), 3.07 (dd, J ) 6.2, 13.6, 1H), 3.57-3.80
(m, 4H), 7.26-7.35 (m, 5H).
General Procedure for Primary Alcohol Activation (2a) and
Cleavage of Support-Bound Material To Afford 3a. To 1.0 g of
support-bound secondary azide 19 solvated in minimal CH2Cl2 was
added 10 mL of 1% p-TsOH in CH2Cl2 (see General Procedure for
MMT group deprotection) to provide a bright orange solution indicating
the presence of mono-p-methoxytrityl cation. The orange solution was
filtered away from support-bound material. This step was repeated
three times until most of the orange solution had dissipated followed
by washes with 3% MeOH in CH2Cl2 (3×) and CH2Cl2 (5×). To the
deprotected support-bound material in 10 mL of CHCl3 was added 0.49
mL (6.0 mmol) of pyridine followed by the addition of 0.89 g (4.0
mmol) NosCl at room temperature. The reaction mixture was stirred
for 9 h. After filtration of excess reagents, the resin was washed with
CH2Cl2 (2 × 15 mL), DMF (3 × 15 mL), and CH2Cl2 (5 × 15 mL).
A portion (0.360 g) of support-bound nosyl alcohol 2a was cleaved,
and the crude oil was purified by column chromatography (99:1 CH2-
Cl2:methanol, then 97:3 CH2Cl2:methanol) to afford 0.046 g of 3a (P1
) Bn; 70% yield based on a 0.51 mequiv/g initial loading level of
16a) as a colorless oil. 1H NMR (400 MHz): δ 2.25 (d, J ) 6.7, 1H),
3.03 (m, 2H), 3.56 (dt, J ) 2.7, 7.5, 1H), 3.79 (m, 1H), 4.13 (dd, J )
4.7, 10.3, 1H), 4.21 (dd, J ) 7.0, 10.3, 1H), 7.21-7.35 (m, 5H), 8.09
(d, J ) 9.0, 2H), 8.39 (d, J ) 9.0, 2H).48
(2R)-3-(1-Ethoxy-1-ethyl)-2-hydroxy-N,N-tetramethylenepropan-
amide (20). To a solution of 15.0 g (0.940 mol) of (2R)-2,3-dihydroxy-
N,N-tetramethylenepropanamide (14b) in 450 mL of CH2Cl2 at 0 °C
was added 8.51 mL (0.890 mol) of ethyl vinyl ether followed by 3.0
g (0.012 mol) of pyridinium p-toluenesulfonate. The reaction solution
was stirred at room temperature for 6 h, washed with H2O (3 × 500
mL), saturated NaHCO3 (3 × 500 mL), and brine (2 × 500 mL), dried
over Na2SO4, and concentrated. The crude yellow oil was purified by
chromatography (80:19:1 hexanes:ethyl acetate:triethylamine, then 67:
32:1 hexanes:ethyl acetate:triethylamine) to provide 14.4 g (66%) of
20 as a colorless oil. NMR data are reported for a 1:1 mixture of
diastereomers. 1H NMR (500 MHz): δ 1.06 (t, J ) 7.1, 3H), 1.07 (t,
J ) 7.1, 3H), 1.17 (d, J ) 9.8, 3H), 1.18 (d, J ) 9.8, 3H), 1.72-1.89
(m, 8H) 3.32-3.62 (m, 16 H), 3.72 (d, J ) 7.4, 1H), 3.73 (d, J ) 7.4,
1H), 4.27 (m, 2H), 4.62 (q, J ) 5.3, 1H), 4.63 (q, J ) 5.3, 1H). 13C
NMR (101 MHz): δ 15.01, 19.38, 19.41, 23.66, 23.67, 25.78, 45.91,
45.93, 46.06, 46.08, 60.88, 60.93, 66.54, 66.66, 68.93, 69.19, 99.57,
99.59, 170.33, 170.40. Anal. Calcd for C11H21N1O4: C, 57.12; H,
9.15; N, 6.06. Found: C, 56.94; H, 8.98; N, 6.05.
triol as a white solid. P1 ) Bn. IR: 3355, 2930, 1457 cm-1 1H NMR
.
(300 MHz): δ 1.96 (broad s, 3H), 2.75 (dd, J ) 9.3, 13.8, 1H), 2.95
(dd, J ) 3.9, 13.8, 1H), 3.66 (ddd, J ) 3.7, 5.4, 9.1, 1H), 3.82 (dd, J
) 3.7, 11.3, 1H), 3.88 (dd, J ) 5.4, 11.3, 1H), 3.97 (ddd, J ) 3.9, 9.1,
9.3, 1H), 7.22-7.35 (m, 5H). 13C NMR (101 MHz): δ 39.52, 63.23,
73.39, 74.37, 126.68, 128.68, 129.32, 137.78. Anal. Calcd for
C10H14O3: C, 65.92; H, 7.74. Found: C, 66.11; H, 7.88.
Preparation of Triacetate. Crude triol product is directly employed
to form the corresponding triacetate. To a solution of crude triol product
(5.0 mg) was added 1 mL of acetic anhydride and DMAP (15 mg,
0.122 mmol). The reaction solution was stirred for 4 h followed by
concentration and purification by chromatography using a short plug
of silica gel (75:25 hexanes:ethyl acetate). P1 ) Bn. 1H NMR (400
MHz): δ 1.96 (s, 3H), 2.06 (s, 6H), 2.88 (dd, J ) 8.4, 14.2, 1H), 2.95
(dd, J ) 5.2, 14.2, 1H), 4.18 (dd, J ) 6.8, 12.2, 1H), 4.34 (dd, J )
3.2, 12.2, 1H), 5.18 (ddd, J ) 3.2, 4.9, 6.8, 1H), 5.32-5.37 (m, 1H),
7.18-7.31 (m, 5H). GC conditions: 180-220 °C; Ramp: 1 °C/min.
Pressure: 15 psi. Retention times (major:minor): 13.2:13.7 min.
Integration (major:minor): 90:10.
Determination of Enantiomeric Purity. To determine the enan-
tiomeric purity of the compounds after zinc borohydride reduction,
chiral HPLC analysis was performed. The diastereomers of the
triacetate derivatives reported above were separated by preparative
HPLC. Chiral HPLC of the major diastereomer revealed a 95:5
enantiomer ratio. As previously mentioned, the starting material, (S)-
methyl ester of isopropylidene glycerate (7a), which is obtained from
Support-Bound Pyrrolidine Amide. Preparation of the support-
bound pyrrolidine amide was accomplished using the conditions
(47) Nakata, T.; Tani, Y.; Hatozaki, M.; Oishi, T. Chem. Pharm. Bull.
1984, 32, 1411-1415.
(48) Liu, G. Ph.D. Dissertation, University of California at Berkeley,
Berkeley, CA, 1998.

Mechanism-Based Aspartyl Protease Inhibitor Libraries
J. Am. Chem. Soc., Vol. 120, No. 38, 1998 9745
described to prepare support-bound Weinreb amide 11 and pyrrolidine
amide 16. The exact loading level of the resin was determined by
cleaving a portion of support-bound material (0.330 g). Prior to
cleavage of the material from support, the ethoxyethyl group was
deprotected with 2% p-TsOH in CH2Cl2 (5 × 5 min) and washed with
3% MeOH in CH2Cl2 (3×) and CH2Cl2 (5×). The crude material
obtained from the TFA cleavage was purified by flash chromatography
(99:1 CH2Cl2:methanol, then 97:3 CH2Cl2:methanol) to afford 0.030 g
of 14b as a colorless oil (0.57 mequiv/g, 78% loading efficiency based
on a 0.86 mequiv/g loading level of Wang resin).
on a 0.57 mequiv/g initial loading level of pyrrolidine amide 20 to
support) as a colorless oil. 1H NMR (400 MHz): δ 2.32 (broad s, 1H),
2.80 (dd, J ) 8.6, 14.1, 1H), 3.06 (dd, J ) 4.3, 14.1, 1H), 3.69 (m,
1H), 3.78 (apparent dt, J ) 2.8, 6.5, 1H), 4.18 (dd, J ) 6.5, 10.5, 1H),
4.31 (dd, J ) 2.8, 10.5, 1H), 7.21-7.33 (m, 5H), 8.11 (d, J ) 8.9,
2H), 8.40 (d, J ) 8.9, 2H). 13C NMR (101 MHz): δ 36.86, 64.45,
70.76, 72.13, 124.59, 127.19, 128.77, 129.34, 130.44, 136.28, 141.08,
150.90.49
Library Synthesis. Grignard Reagent Preparation. The Grignard
reagents that were not commercially available were prepared by one
of two methods. A sample procedure is provided for each method. To
test the efficacy of the preparation, each Grignard reagent was added
to (2S)-2,3-O-isopropylidene-N-methoxy-N-methylpropanamide and the
reaction completion to the corresponding ketone was monitored by TLC.
The commercially available reagents included benzylmagnesium chlo-
ride, isobutylmagnesium chloride, phenethylmagnesium chloride, meth-
ylmagnesium bromide, and n-pentylmagnesium bromide.
Preparation from Magnesium Turnings. Magnesium turnings
were stirred under N2 for 48-72 h. To a vigorously stirring suspension
of 1.82 g (0.075 mol) activated magnesium turnings in 20 mL of Et2O
was very slowly added dropwise 6.98 mL (0.0500 mol) of cyclohexyl-
methyl bromide. Cyclohexylmethyl bromide was passed through a short
plug of basic alumina prior to use. After the addition of the first 2-3
drops of cyclohexylmethyl bromide, initiation of Grignard reagent
formation was observed by exotherm and dark gray color of the reaction
solution. The addition was complete after 2 h, and stirring was
continued at room temperature for 1 h. The Grignard reagent was
filtered away from remaining magnesium into Schlenk apparatus and
stored at -20 °C.
Grignard Addition to Pyrrolidine Amide (21). To the support-
bound pyrrolidine amide (1.0 g, 0.57 mmol) solvated in 10 mL of THF
at 0 °C was added 1.2 mL (2.4 mmol) of 2.0 M benzylmagnesium
chloride in THF. The reaction mixture was stirred at 4 °C for 8 h.
After filtration of the reaction solution, the resin was washed with THF
(1 × 15 mL), acetone (3 × 15 mL), 0.28 M hydrocinnamic acid in
THF (3 × 15 mL), DMF (3 × 15 mL), and CH2Cl2 (5 × 15 mL).
L-Selectride Reduction (22). To 1.0 g of support-bound ketone
21 in 5.0 mL of THF at -75 °C was added 5.0 mL (5.0 mmol) of 1.0
M L-selectride in THF. The reaction mixture was stirred at -75 to
-65 °C for 20 h and then was allowed to warm to -20 °C over 3 h.
Excess reagent was filtered away, and the resin was washed with THF
(1 × 15 mL), 0.28 M hydrocinnamic acid in THF (3 × 15 mL, 15 min
per wash), DMF (3 × 15 mL), and CH2Cl2 (5 × 15 mL). A portion
of the resin was deprotected (EE group) and cleaved to determine
product purity and reaction completion. Due to the tendency for the
resin to solidify or crystallize at low temperatures, the resin was
resubjected to the L-selectride reduction conditions if the first subjection
did not provide complete reduction as indicated by the TLC of crude
product from cleavage. Crude product can be purified by (99:1 CH2-
Cl2:methanol, then 97:3 CH2Cl2:methanol) to provide the corresponding
Preparation of Magnesium Anthracene THF Complex and
Grignard Reagent.39,40 To a suspension of 6.0 g (0.25 mol) of
magnesium turnings and 24.5 g (0.137 mol) of anthracene in 250 mL
of THF was added 0.41 mL of ethyl bromide. The reaction mixture
was stirred for 28 h at 60 °C. The white-yellow solution changed to
green and then orange over the course of the reaction. In an inert
atmosphere, the excess reagents were filtered away from desired orange
crystals and excess magnesium. The product was washed with THF
(2 × 70 mL) and dried under vacuum.
triol as a white solid. P1 ) Bn. IR: 3355, 2930, 1457 cm-1 1H NMR
.
(300 MHz): δ 2.06 (broad s, 3H), 2.84 (dd, J ) 8.2, 13.6, 1H), 2.93
(dd, J ) 5.5, 13.6, 1H), 3.59 (m, 1H), 3.74 (dd, J ) 4.4, 11.1, 1H),
3.79 (dd, J ) 3.7, 11.1, 1H), 3.86 (m, 1H), 7.22-7.34 (m, 5H). 13C
NMR (101 MHz): δ 40.05, 64.90, 72.51, 73.61, 126.58, 128.61, 129.40,
137.94.
Corresponding Triacetate. The triacetate was prepared using the
aforementioned method. P1)Bn. 1H NMR (400 MHz): δ 1.99 (s, 3H),
2.02 (s, 3H), 2.13 (s, 3H), 2.86 (d, J ) 7.0, 2H), 4.05 (dd, J ) 6.8,
11.8, 1H), 4.26 (dd, J ) 4.6, 11.8, 1H), 5.19 (ddd, J ) 3.7, 4.6, 6.8,
1H), 5.32 (dt, J ) 3.7, 7.0, 1H), 7.16-7.30 (m, 5H). GC conditions:
180-220 °C. Ramp: 1 °C/min. Pressure: 15 psi. Retention times
(major:minor): 13.1:13.7 min. Integration (major:minor): 13:87.
Procedure for Secondary Alcohol Activation and Azide Displace-
ment (23). To 1.0 g of support-bound secondary alcohol 22 solvated
in 10 mL of CHCl3 was added 1.11 g (7.50 mmol) of 4-pyrrolidinopy-
ridine followed by the addition of 1.11 g (5.00 mmol) of 4-nitroben-
zenesulfonyl chloride (NosCl) at room temperature. The reaction
mixture was stirred for 9 h. After filtration of excess reagent, the
resin was washed with CH2Cl2 (5 × 15 mL) and DMF (4 × 15
mL).
To 1.0 g of support-bound secondary nosyl alcohol solvated in 10
mL of DMF was added 0.650 g (10.0 mmol) of sodium azide. The
reaction mixture was stirred at 45 °C for 24 h. Following filtration of
the reaction solution, the resin was washed with DMF (2 × 15 mL),
DMF:H2O (1:1) (3 × 15 mL), DMF (2 × 15 mL), and CH2Cl2 (5 ×
15 mL).
Procedure for Primary Alcohol Activation (2b) and Cleavage of
Support-Bound Material To Afford 3b. To 1.0 g of support-bound
secondary azide 23 solvated in minimal CH2Cl2 was added 10 mL of
2% p-TsOH in CH2Cl2 (5 × 5 min) followed by washes with 3% MeOH
in CH2Cl2 (3×) and CH2Cl2 (5×). To the deprotected support-bound
material in 10 mL of CHCl3 was added 0.49 mL (6.0 mmol) of pyridine
followed by the addition of 0.89 g (4.0 mmol) NosCl at room
temperature. The reaction mixture was stirred for 9 h. After filtration
of excess reagents, the resin was washed with CH2Cl2 (2 × 15 mL),
DMF (3 × 15 mL), and CH2Cl2 (5 × 15 mL). A portion (0.470 g) of
support-bound secondary azide 23 was cleaved, and the crude oil was
purified by column chromatography (99:1 CH2Cl2:methanol, then 97:3
CH2Cl2:methanol) to afford 0.057 g of 3b (P1 ) Bn; 68% yield based
To a suspension of 3.06 g (7.31 mmol) of magnesium anthracene
THF complex in 55 mL of THF was very slowly added a solution of
1.24 g (6.09 mmol) of p-phenylbenzyl chloride in 5 mL of THF. The
solution changed from orange to dark green to gray over the addition
period. The reagent was tested and subsequently used in solid-phase
reactions.
Synthesis of Intermediates 2a. The library of 204 compounds was
synthesized according to the conditions reported above through
intermediate 2a. The support-bound Weinreb amide 11a that was used
for the library synthesis had an initial loading level of 0.28 mmol/g.
The resin was split into 17 batches of 0.7 g prior to the Grignard
addition. The synthesis was carried out in 50 mL round-bottom flasks
through intermediate 2a. The preparation of the Grignard reagents is
listed above. For each of the 17 ketones resulting from the Grignard
addition to Weinreb amide 11a, 45-50 mg of resin was cleaved. The
corresponding ketones were obtained in high purity by 1H NMR, except
for 7-18% of methyl amide side product 13. Approximately 20 mg
of resin was cleaved after zinc borohydride reduction and after
conversion to the azide 19 to determine reaction completion by TLC.
In several cases the support-bound material had to be resubjected to
the reaction with NosCl, 4-pyrrolidinopyridine in CHCl3, followed by
the azide displacement to obtain azide 19, due to incomplete reaction.
For each resin, 100-125 mg was also cleaved at intermediate 2a, and
in every case, the major product was determined by TLC and 1H NMR
to be the desired compound 3a (NMR data listed below).
NMR Data for Intermediates 3a. P1 ) benzyl. 1H NMR (400
MHz): δ 2.25 (d, J ) 6.7, 1H), 3.03 (m, 2H), 3.56 (apparent dt, J )
2.7, 7.5, 1H), 3.79 (m, 1H), 4.13 (dd, J ) 4.7, 10.3, 1H), 4.21 (dd, J
) 7.0, 10.3, 1H), 7.21-7.35 (m, 5H), 8.09 (d, J ) 9.0, 2H), 8.39 (d,
J ) 9.0, 2H).
(49) Kick, E. K. Ph.D. Dissertation, University of California at Berkeley,
Berkeley, CA, 1998.

9746 J. Am. Chem. Soc., Vol. 120, No. 38, 1998
Lee et al.
P1 ) Isobutyl. 1H NMR (400 MHz): δ 0.95 (d, J ) 3.2, 3H), 0.96
(d, J ) 3.2, 3H), 1.40 (m, 1H), 1.66 (m, 1H), 1.77 (m, 1H), 3.37 (m,
1H), 3.87 (m, 1H), 4.16-4.22 (m, 2H), 8.13 (d, J ) 8.6, 2H), 8.42 (d,
J ) 8.6, 2H).
with 3,4-methylenedioxyphenethylamine and 2,4-dichlorophenethyl-
amine. 3,4-Methylenedioxyphenethylamine was prepared from the
commercially available hydrochloride salt by basic extraction with
aqueous Na2CO3 and ethyl acetate. The reactions were performed with
1.0 M amine in NMP at 80 °C for 36 h in sealed vials. The resins
were washed with NMP (3×), THF (2×), CH2Cl2 (3×), and ether (1×)
and dried under vacuum overnight. The resins were then transferred
to a 96-well filter apparatus.41 Each well had approximately 15 mg of
resin, with each P1R1 combination placed in six wells for coupling to
the R2 and R3 side chains. Prior to each R2 or R3 coupling the resin
was rinsed twice with anhydrous NMP. Incorporation of N-phthaloyl-
â-alanine was carried out using a stock solution of 0.3 M acid, 0.3 M
HATU, and 0.9 M i-Pr2EtN in NMP overnight. Precipitation sometimes
occurs with reactions of N-phthaloyl-â-alanine acid and PyBOP, so
HATU was employed for this coupling. The acylation with (cyclo-
hexyl)isocyanate was carried out with 0.3 M isocyanate in NMP. The
resins were washed with NMP (4×), CH2Cl2 (3×), and THF (1×).
The azide reduction was accomplished using 0.2 M SnCl2, 0.8 M PhSH,
and 1.0 M Et3N in THF for 4 h. The resins were washed with THF:
H2O (1:1), THF (3×), and CH2Cl2 (3×). The R3 acids, 3-(2-
benzoxazolin-2-on-3-yl)propionic acid, 2-bromo-4,5-dimethoxybenzoic
acid, and 2,4-dichlorophenoxyacetic acid, were coupled using stock
solutions of 0.3 M carboxylic acid, 0.3 M PyBOP, 0.3 M HOAt, and
0.9 M iPr2EtN in NMP overnight. The resins were washed with NMP
(4×), THF (2×), and CH2Cl2 (3×). The resin was resubjected to the
coupling conditions for both R2 and R3 acylations to ensure completion
of the reaction. The compounds were cleaved from solid support into
2-mL well plates using TFA:CH2Cl2 (1:1) for 30 min followed by rinses
with CH2Cl2 (3×). The combined filtrates were concentrated to dryness
using a Jouan 10.10 centrifugation concentrator. Toluene was added
to form an azeotrope with TFA during the concentration step. The
resulting products were stored at -20 °C. Twenty-five compounds
from the library were picked at random and were analyzed by mass
spectrometry in a matrix of R-cyano-4-hydroxycinnamic acid on a
Perseptive Biosystems MALDI spectrometer. The expected molecular
ion peaks were obtained for 25 of the 25 compounds tested.
High-Throughput Cathepsin D Assay. A fluorometric high-
throughput assay for activity toward human liver cathepsin D (Cal-
biochem: San Diego, CA) was performed in 96-well microtiter plates.42
The peptide substrate (Ac-Glu-Glu(Edans)-Lys-Pro-Ile-Cys-Phe-Phe-
Arg-Leu-Gly-Lys(Dabcyl)-Glu-NH2) used in the assay has been previ-
ously reported (Km ≈ 4 µM).37 The assay was performed in DYNAT-
ECH Microfluor fluorescence microtiter plates, and readings were taken
on a Perkin-Elmer LS-50B with an attached Perkin-Elmer 96-well plate
reader. The excitation wavelength was 340 nm. A 340-nm interference
filter (Hoya, U-340) for excitation and a 430-nm cutoff filter for
emission were used. For the microwell-based assays the substrate
concentration was 5 µM and the cathepsin D concentration was 1 nM
in a 0.1 M formic acid buffer (pH ) 3.7). DMSO (10%) was used to
ensure complete dissolution of the inhibitors. The fluorescent unit
readings were taken at three time points (5, 10, and 15 min) within the
linear region of the substrate cleavage, and the percent activity of the
enzyme was determined by comparing the change of fluorescent units
(FU) for each well to the average change in FU for six control wells
without inhibitor. The library was screened at approximately 330 nM
of inhibitor. The concentration was based on the assumption that 50%
of the theoretical yield was obtained for each inhibitor. The library
was subsequently screened at 100, 33, and 10 nM for active inhibitors.
Synthesis of Potent Inhibitors. Several of the most potent
compounds were synthesized on the solid support following the
previously described method using the pyrrolidine amide 16a with a
0.51 mequiv/g initial loading level. Overall yields of the compounds
were determined by the mass balance of desired product after column
chromatography purification (100:0 CH2Cl2:methanol gradually increas-
ing polarity until 97:3 CH2Cl2:methanol). The 1H NMR is only listed
for the major amide rotamer for each compound.
P1 ) Phenethyl. 1H NMR (300 MHz): δ 1.89-2.05 (m, 2H), 2.25
(broad s, 1H), 2.64-2.86 (m, 2H), 3.26 (m, 1H), 3.89-3.93 (m, 1H),
4.15 (m, 2H), 7.16-7.33 (m, 5H), 8.08 (d, J ) 9.0, 2H), 8.39 (d, J )
9.0, 2H).
P1 ) Methyl. 1H NMR (400 MHz): δ 1.36 (d, J ) 6.7, 3H), 3.60
(m, 1H), 3.77 (m, 1H), 4.17 (d, J ) 5.4, 2H), 8.13 (d, J ) 8.9, 2H),
8.42 (d, J ) 8.9, 2H).
P1 ) n-Pentyl. 1H NMR (400 MHz): δ 0.90 (t, J ) 6.8, 3H), 1.31-
1.45 (m, 4H), 1.63-1.71 (m, 4H), 3.33 (m, 1H), 3.89 (m, 1H), 4.17
(m, 2H), 8.14 (d, J ) 8.9, 2H), 8.43 (d, J ) 8.9, 2H)
P1 ) Cyclohexylmethyl. 1H NMR (300 MHz): δ 0.86-0.96 (m,
2H), 1.11-1.27 (m, 3H), 1.38-1.47 (m, 2H), 1.56-1.76 (m, 6H), 2.2
(broad s, 1H), 3.38 (m, 1H), 3.86 (m, 1H), 4.12-4.20 (m, 2H), 8.12
(d, J ) 9.0, 2H), 8.41 (d, J ) 9.0, 2H).
P1 ) â-Naphthylmethyl. 1H NMR (300 MHz): δ 2.2 (broad s,
1H), 3.19 (m, 2H), 3.66 (apparent dt, J ) 2.7, 7.5, 1H), 3.80 (m, 1H),
4.11 (dd, J ) 4.9, 10.4, 1H), 4.20 (dd, J ) 7.0, 10.4, 1H), 7.32 (dd, J
) 1.8, 8.3, 1H), 7.48 (m, 2H), 7.68 (apparent s, 1H), 7.77-7.84 (m,
3H), 8.03 (d, J ) 9.0, 2H), 8.32 (d, J ) 9.0, 2H).
P1 ) r-Naphthylmethyl. 1H NMR (400 MHz): δ 3.46 (dd, J )
7.0, 13.7, 1H), 3.57 (dd, J ) 7.8, 13.7, 1H), 3.72 (m, 1H), 3.77 (m,
1H), 4.10 (dd, J ) 4.6, 10.4, 1H), 4.25 (dd, J ) 7.0, 10.4, 1H), 7.42
(m, 2H), 7.54 (m, 3H), 7.80 (m, 1H), 7.90 (m, 1H), 8.01 (d, J ) 8.8,
2H), 8.32 (d, J ) 8.8, 2H).
P1 ) 4-Methoxy-3-methylbenzyl. 1H NMR (400 MHz): δ 2.28
(s, 3H), 2.92-2.97 (m, 2H), 3.51 (m, 1H), 3.80 (m, 1H), 3.82 (s, 3H),
4.11 (dd, J ) 4.6, 10.3, 1H), 4.19 (dd, J ) 7.0, 10.3, 1H), 6.76 (dd, J
) 2.1, 8.1, 1H), 6.99 (m, 2H), 8.09 (d, J ) 9.0, 2H), 8.40 (d, J ) 9.0,
2H).
P1 ) p-Methoxyphenethyl. 1H NMR (300 MHz): δ 1.95 (m, 2H),
2.61-2.80 (m, 2H), 3.26 (m, 1H), 3.78 (s, 3H), 3.89 (m, 1H), 4.16 (m,
2H), 6.36 (d, J ) 8.6, 2H), 7.09 (d, J ) 8.6, 2H), 8.08 (d, J ) 9.0,
2H), 8.41 (d, J ) 9.0, 2H).
P1 ) m-Phenoxybenzyl. 1H NMR (300 MHz): δ 2.99 (m, 2H),
3.54 (m, 1H), 3.81 (m, 1H), 4.12 (dd, J ) 4.6, 10.4, 1H), 4.20 (dd, J
) 6.7, 10.4, 1H), 6.89-7.01 (m, 5H), 7.11 (m, 1H), 7.28-7.36 (m,
3H), 8.09 (d, J ) 8.6, 2H), 8.40 (d, J ) 8.6, 2H).
P1 ) 3,5-Dimethylbenzyl. 1H NMR (400 MHz): δ 2.30 (s, 6H),
2.95 (dd, J ) 7.4, 13.5, 1H), 2.96 (dd, J ) 7.6, 13.5, 1H), 3.54 (ddd,
J ) 2.6, 7.4, 7.6, 1H), 3.80 (m, 1H), 4.11 (dd, J ) 4.7, 10.3, 1H), 4.20
(dd, J ) 7.0, 10.3, 1H), 6.82 (s, 2H), 6.90 (s, 1H), 8.09 (d, J ) 8.8,
2H), 8.40 (d, J ) 8.8, 2H).
P1 ) 3-Benzylpropyl. 1H NMR (400 MHz): δ 1.65 (m, 4H), 2.65
(m, 2H), 3.33 (m, 1H), 3.86 (m, 1H), 4.15 (m, 2H), 7.16-7.31 (m,
5H), 8.11 (d, J ) 9.0, 2H), 8.41 (d, J ) 9.0, 2H).
P1 ) 4-tert-Butylbenzyl. 1H NMR (400 MHz): δ 1.31 (s, 9H), 2.20
(broad s, 1H), 3.00 (m, 2H), 3.53 (apparent dt, J ) 2.7, 7.5, 1H), 3.84
(m, 1H), 4.12 (dd, J ) 4.5, 10.3, 1H), 4.20 (dd, J ) 7.1, 10.3, 1H),
7.14 (d, J ) 8.2, 2H), 7.35 (d, J ) 8.2, 2H), 8.10 (d, J ) 8.9, 2H),
8.41 (d, J ) 8.9, 2H).
P1 ) p-Phenylbenzyl. 1H NMR (400 MHz): δ 2.25 (broad s, 1H),
3.08 (m, 2H), 3.60 (apparent dt, J ) 4.7, 10.2, 1H), 3.87 (m, 1H), 4.14
(dd, J ) 4.7, 10.2, 1H), 4.23 (dd, J ) 7.0, 10.2, 1H), 7.30 (d, J ) 8.0,
2H), 7.36 (m, 1H), 7.45 (m, 2H), 7.54-7.59 (m, 4H), 8.10 (d, J ) 8.6,
2H), 8.39 (d, J ) 8.6, 2H).
P1 ) 3,4-Methylenedioxybenzyl. 1H NMR (400 MHz): δ 2.19
(broad s, 1H), 2.94 (m, 2H), 3.49 (apparent dt, J ) 2.8, 10.2, 1H),
3.81 (m, 1H), 4.11 (dd, J ) 4.7, 10.3, 1H), 4.20 (dd, J ) 7.0, 10.3,
1H), 5.96 (s, 2H), 6.67 (d, J ) 7.6, 1H), 6.69 (s, 1H), 6.76 (d, J ) 7.6,
1H), 8.11 (d, J ) 8.9, 2H), 8.41 (d, J ) 8.9, 2H).
P1 ) p-Bromobenzyl. 1H NMR (400 MHz): δ 2.26 (broad s, 1H),
2.93-3.06 (m, 2H), 3.53 (apparent dt, J ) 2.8, 7.4, 1H), 3.81 (m, 1H),
4.12 (dd, J ) 4.8, 10.3, 1H), 4.21 (dd, J ) 6.9, 10.3, 1H), 7.11 (d, J
) 8.3, 2H), 7.46 (d, J ) 8.3, 2H), 8.10 (d, J ) 8.8, 2H), 8.41 (d, J )
8.8, 2H).
Inhibitor 25. From 0.338 g of resin was isolated 57 mg of pure
inhibitor 25 (46%) after chromatography. 1H NMR (400 MHz, CDCl3):
δ 2.65 (m, 2H), 2.88 (apparent t, J ) 7.7, 2H), 3.01 (apparent t, J )
6.9, 2H), 3.24 (m, 1H), 3.47 (m, 2H), 3.83-3.96 (m, 4H), 3.85 (s,
3H), 3.89 (s, 3H), 4.34 (apparent q, J ) 8.3, 1H), 4.66 (broad s, 1H),
Introduction of R1, R2, and R3. Each of the 17 support-bound
scaffolds 2a was split into two 0.2-g batches for amine displacement

Mechanism-Based Aspartyl Protease Inhibitor Libraries
J. Am. Chem. Soc., Vol. 120, No. 38, 1998 9747
6.71 (d, J ) 9.2, 1H), 6.84 (dd, J ) 1.7, 8.0, 1H), 6.93-7.00 (m, 5H),
7.05 (m, 1H), 7.05 (s, 1H), 7.07 (s, 1H), 7.16 (dd, J ) 2.1, 8.1, 1H),
7.23-7.30 (m, 3H), 7.34 (d, J ) 2.1, 1H), 7.71 (m, 2H), 7.83 (m, 2H).
Anal. Calcd for C44H40N3O8Cl2Br1: C, 59.41; H, 4.53; N, 4.72.
Found: C, 59.22; H, 4.76; N, 4.52.
(d, J ) 6.2, 3H), 0.93 (d, J ) 6.2, 3H), 1.39 (m, 1H), 1.66 (m, 2H),
2.59-2.71 (m, 2H), 2.89-2.93 (m, 2H), 3.27 (dd, J ) 2.1, 14.0, 1H),
3.45-3.56 (m, 2H), 3.78-3.83 (m, 2H), 3.85 (s, 3H), 3.86 (s, 3H),
3.89-3.94 (m, 2H), 4.16 (m, 1H), 4.37 (broad s, 1H), 6.59 (d, J ) 9.5,
1H), 6.97 (s, 1H), 7.10 (d, J ) 8.2, 1H), 7.13 (dd, J ) 2.0, 8.2, 1H),
7.16 (s, 1H), 7.30 (d, J ) 2.0, 1H), 7.67 (m, 2H), 7.78 (m, 2H). Anal.
Calc′d for C35H38N3O7Cl2Br1: C, 55.06; H, 5.02; N, 5.50. Found: C,
55.02; H, 5.18; N, 5.50.
Cathepsin D Assay. The cathepsin D assays for the compounds
that had been fully characterized were performed in a quartz cuvette
with a Perkin-Elmer LS-50B spectrometer (excitation ) 336 nM,
emission ) 490 nM). The same substrate as reported for the high-
throughput assay was used. A 0.1 M formic acid buffer (pH ) 3.7)
was used with a final concentration of 0.5 nM cathepsin D. In a typical
assay with a final volume of 600 µL, to 555 µL of buffer is added 20
µL of inhibitor in DMSO, followed by 5 µL of cathepsin D stock
(0.05% Triton X-100 in H2O). After a 3.5-min incubation time at 25
°C, 20 µL of substrate stock (2:1 H2O:DMSO) is added. The increase
of fluorescence intensity was recorded as a function of time. Assays
were performed in duplicate or triplicate at five to six inhibitor
concentrations for each Ki determination. Data were fitted by nonlinear
Inhibitor 26. 1H NMR (400 MHz): δ 2.00 (broad s, 1H), 2.69 (m,
2H), 2.94 (m, 4H), 3.12 (m, 1H), 3.55 (m, 4H), 3.81 (s, 3H), 3.87 (s,
3H), 3.96 (apparent t, J ) 7.2, 2H), 4.36 (m, 1H), 6.39 (d, J ) 8.3,
1H), 6.85 (m, 1H), 6.90-6.95 (m, 5H), 7.03-7.09 (m, 2H), 7.11 (s,
1H), 7.14 (dd, J ) 2.0, 8.1, 1H), 7.22-7.29 (m, 3H), 7.32 (d, J ) 2.0,
1H), 7.70 (m, 2H), 7.83 (m, 2H). Anal. Calcd for C44H40N3O8Cl2Br1:
C, 59.41; H, 4.53; N, 4.72. Found: C, 59.60; H, 4.70; N, 4.59.
Inhibitor 27. From 0.290 g of resin was isolated 48 mg of pure
inhibitor 27 (48%) after chromatography. 1H NMR (400 MHz, CDCl3):
δ 2.62 (apparent t, J ) 7.5, 2H), 2.82 (apparent t, J ) 7.6, 2H), 3.18-
3.25 (m, 3H), 3.40-3.47 (m, 2H), 3.57 (s, 3H), 3.85 (s, 3H), 3.91-
3.96 (m, 4H), 4.47 (apparent q, J ) 8.4, 1H), 4.76 (broad s, 1H), 6.69
(s, 1H), 6.92 (d, J ) 8.2, 1H), 6.95 (s, 1H), 7.04 (dd, J ) 2.1, 8.2,
1H), 7.29 (d, J ) 2.1, 1H), 7.40-7.45 (m, 3H), 7.68 (m, 2H), 7.71-
7.80 (m, 6H). Anal. Calcd for C42H38N3O7Cl2Br1: C, 59.52; H, 4.52;
N, 4.96. Found: C, 59.63; H, 4.67; N, 4.69.
Inhibitor 28. From 0.314 g of resin was isolated 55 mg of pure
inhibitor 28 (48%) after chromatography. 1H NMR (400 MHz, CDCl3):
δ 2.65 (m, 2H), 2.85 (apparent t, J ) 7.3, 2H), 3.08 (apparent t, J )
6.7, 2H), 3.23 (m, 1H), 3.44 (m, 1H), 3.57 (m, 1H), 3.75 (s, 3H), 3.86
(s, 3H), 3.94 (m, 4H), 4.39 (apparent q, J ) 8.3, 1H), 4.73 (broad s,
1H), 6.78 (d, J ) 9.2, 1H), 6.93 (s, 1H), 6.97 (s, 1H), 7.02 (d, J ) 8.2,
1H), 7.10 (dd, J ) 2.1, 8.2, 1H), 7.30 (d, J ) 2.1, 1H), 7.36-7.42 (m,
5H), 7.51-7.54 (m, 4H), 7.68 (m, 2H), 7.81 (m, 2H). Anal. Calcd
for C44H40N3O7Cl2Br1: C, 60.49; H, 4.62; N, 4.81. Found: C, 60.23;
H, 4.86; N, 4.58.
Inhibitor 29. From 0.325 g of resin was isolated 55 mg of pure
inhibitor 29 (46%) after chromatography. 1H NMR (400 MHz, CDCl3):
δ 2.64 (m, 2H), 2.86 (apparent t, J ) 7.1, 2H), 2.96 (m, 2H), 3.20 (m,
1H), 3.46 (m, 1H), 3.54 (m, 1H), 3.78 (m, 2H), 3.82 (s, 3H), 3.86 (s,
3H), 3.91 (m, 2H), 4.31 (apparent q, J ) 8.5, 1H), 4.73 (broad s, 1H),
6.73 (d, J ) 9.3, 1H), 6.85 (s, 1H), 6.96 (s, 1H), 7.03 (d, J ) 8.3, 1H),
7.14 (m, 2H), 7.16 (dd, J ) 2.2, 8.3, 1H), 7.32 (d, J ) 2.2, 1H), 7.37-
7.41 (m, 2H), 7.70 (m, 2H), 7.80 (m, 2H). Anal. Calc′d for
C38H35N3O7Cl2Br2: C, 52.08; H, 4.03; N, 4.79. Found: C, 52.28; H,
4.09; N, 4.60.
Inhibitor 30. From 0.307 g of resin was isolated 49 mg of pure
inhibitor 30 (45%) after chromatography. 1H NMR (400 MHz, CDCl3):
δ 2.65 (apparent t, J ) 7.2, 2H), 2.85 (apparent t, J ) 7.6, 2H), 2.91
(m, 2H), 3.12 (m, 1H), 3.42-3.48 (m, 2H), 3.63 (m, 1H), 3.88 (m,
1H), 3.94 (m, 2H), 4.16 (apparent q, J ) 8.1, 1H), 4.39 (d, J ) 14.5,
1H), 4.48 (d, J ) 14.5, 1H), 4.50 (broad s, 1H), 6.76 (d, J ) 8.8, 1H),
6.82 (m, 1H), 6.92 (m, 2H), 6.97 (m, 2H), 7.04 (m, 2H), 7.11-7.27
(m, 6H), 7.32 (m, 1H), 7.38 (d, J ) 2.5, 1H), 7.70 (m, 2H), 7.82 (m,
2H). Anal. Calc′d for C43H37N3O8Cl4: C, 60.79; H, 4.39; N, 4.95.
Found: C, 61.01; H, 4.40; N, 4.80.
regression analysis to the equation derived by Williams and Morrison:
50
V0
2Et
1/2
V )
Ki 1 +
+ It - Et 2 + 4Ki 1 +
Et
-
S
Km
S
Km
{ ( (
)
)
(
)
[
]
)
S
Km
Ki 1 +
+ It - Et
(
}
]
[
The Km for the substrate was determined to be 4 µΜ by using a
Lineweaver-Burke plot (the Km was previously determined by Krafft
to be 5 µM).37 The variables S, Et, and It are the concentrations of
substrate, active enzyme, and inhibitor, respectively.
Acknowledgment. Support from the NIH (R01GMS4051)
and the Burroughs Wellcome Foundation is gratefully acknowl-
edged. E.K.K. thanks the ACS Division of Medicinal Chemistry
and Parke-Davis Research for pre-doctoral fellowship. We also
thank Jack Kirsch for advice on kinetic analysis and use of a
fluorescence spectrophotometer and the W. M. Keck Foundation
for partial support of mass spectrometry instrumentation.
Structure searches were performed using the ACD provided by
MDL Information Systems, Inc., 14600 Catalina St., San
Leandro, CA 94577.
Supporting Information Available: 1H NMR spectra of
intermediates 3 for each P1 side chain are provided (18 pages
print). See any current masthead page for ordering information
and Web access instructions.
JA981812M
Inhibitor 31. From 0.366 g of resin was isolated 79 mg of pure
(50) Williams, J. W.; Morrison, J. F. Methods Enzymol. 1979, 63, 437-
467.
inhibitor 31 (64%) after chromatography. 1H NMR (500 MHz): δ 0.90