Identification of TNF-α inhibitors from a split-pool library based on a tyrosine-proline peptidomimetic scaffold

Article abstract of DOI:10.1016/S0960-894X(02)00877-6

The design and synthesis of a combinatorial library based on a 4-aryloxyproline scaffold with tyrosine as the aryl portion is described. The 1728 member library was prepared using the split-pool method to generate pools of compounds. Screening of the library components as mixtures followed by deconvolution led to the discovery of novel inhibitors of TNF-α induced apoptosis.

Full text of DOI:10.1016/S0960-894X(02)00877-6

Bioorganic & Medicinal Chemistry Letters 13 (2003) 205–208
Identification of TNF-ꢀ Inhibitors from a Split-Pool Library
Based on a Tyrosine-Proline Peptidomimetic Scaffold
Randy W. Jackson,* John C. Tabone and J. Jeffry Howbert
Department of Chemistry, Celltech R&D, Inc., 1621 220th Street SE, Bothell, WA 98021, USA
Received 19 July 2002; accepted 11 October 2002
Abstract—The design and synthesis of a combinatorial library based on a 4-aryloxyproline scaffold with tyrosine as the aryl portion
is described. The 1728 member library was prepared using the split-pool method to generate pools of compounds. Screening of the
library components as mixtures followed by deconvolution led to the discovery of novel inhibitors of TNF-a induced apoptosis.
# 2002 Elsevier Science Ltd. All rights reserved.
The design of combinatorial libraries for use as screen-
ing tools is now a firmly entrenched method in the field
of medicinal chemistry.1 Over the past decade or so
many strategies have been pursued that combine the
design and synthesis of combinatorial libraries with
biological screening efforts. One of the earliest approa-
ches consisted of performing a split-pool synthesis of a
library and assaying the subsequent mixtures for bio-
logical activity.2 This proved to be a very valuable
method, however it can be hampered by high levels of
false positives due in large part to additive effects arising
from the screening of mixtures. Nonetheless, a number
of viable medicinal chemistry hits have been discovered
using this method.1,2 As part of our program to gen-
erate a diverse collection of combinatorial libraries for
screening, we designed an aryloxy proline scaffold based
on a tyrosine-proline (Tyr-Pro) peptidomimetic onto
which we could easily append a range of diversity ele-
ments. Substituted 4-aryloxy prolines have been repor-
ted, but library scaffolds based on tyrosine as the aryl
portion are novel.3,4 We report herein the results of our
efforts including the synthesis of the scaffold, the gen-
eration of a 1728 member split-pool library and the
identification of discrete compounds possessing anti-
TNF-a activity.
(Scheme 1).5 The acid was then esterified using tri-
methylsilylethanol and EDCI/HOBT/DIEA. After
chromatography the TMSE ester was obtained in 53%
yield from hydroxyproline. The next step is a Mitsu-
nobu coupling utilizing a protected form of tyrosine to
generate the aryl ether.6 Initial efforts focused on utiliz-
ing the tert-butyl ester of Fmoc-Tyr but these reactions
invariably resulted in the isolation of a mixture of the
deprotected and Fmoc protected aryloxyprolines. It is
known that Mitsunobu conditions are sufficiently basic
for removal of Fmoc protecting groups,7 and our hopes
of overcoming the low yields due to these side reactions
were not realized. We next examined whether Tyr(tBu),
which is unprotected at the alpha amine, would suffice
as a substrate for the reaction. We were pleased to see
that addition of Tyr(tBu) to a THF solution of 2 in the
presence of triphenylphosphine and DEAD followed by
refluxing for 30min provided the coupled product in
acceptable yield. Purification of the product proved to
be difficult and flash chromatography afforded a mix-
The synthesis began with protection of l-trans-4-
hydroxyproline 1 as the Alloc carbamate using allyl
chloroformate under Schotten–Baumen conditions
Scheme 1. Reagents and conditions: (a) Alloc-Cl, Na2CO3, dioxane/
H2O; (b) TMSE-OH, EDCI, HOBT, DIEA, 53% from 1; (c) PPh3,
*Corresponding author at present address: ViroPharma, Inc., 405
Eagleview Blvd, Exton, PA 19341, USA. Fax:+1-610-458-7380;
e-mail: randy.jackson@viropharma.com
DEAD, Tyr(tBu), THF, 60 ꢀC, 30min; (d) Fmoc-Cl, NMM, CH Cl2;
2
(e) SiO2, toluene, 110 ꢀC, 26% from 2.
0960-894X/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00877-6

206
R. W. Jackson et al. / Bioorg. Med. Chem. Lett. 13 (2003) 205–208
ture of the aryloxyproline and Tyr(tBu) accompanied by
a small amount of dicarbethoxy hydrazine. This mixture
was then treated with Fmoc-Cl in the presence of NMM
followed by flash chromatography to afford a mixture
of the fully protected scaffold and Fmoc-Tyr(tBu).
Selective deprotection of the tert-butyl ester was
accomplished using a novel cleavage strategy developed
in these labs.8 Thus, treating a toluene solution of the
mixture with silica gel at reflux for 1.5 h resulted in
clean conversion to carboxylic acid 3. At this stage
removal of residual Fmoc-Tyr was uneventful via flash
chromatography, and the scaffold 3 was isolated in 26%
overall yield from 2.
with 5% sodium diethylthiocarbamate in DMF (2Â) to
fully remove the palladium containing material. The
proline amino group was then coupled with N-acetyl-
proline to afford the fully derivatized scaffold. Cleavage
with 90% TFA/H2O afforded the primary carboxamide
5 in 73% overall yield. Many of the cleaved inter-
mediates along with the final product from the solid-
1
phase synthesis were analyzed by TLC and H NMR
and shown to be comparable to the same compounds
prepared in solution using trimethoxybenzyl amide as a
Tentagel SRAM resin surrogate.11
The library synthesis was performed on an ACT
357 MPS automated synthesizer using a manual com-
bine/split procedure. The diversity elements are shown
in Figure 1. Following the protocol outlined in Scheme
2, 12 sidechains were attached at each of the three sites
to generate the 1728 member Tyr-Pro library. The ACT
357 uses a 36-well reactor block allowing us to perform
a split-pool strategy that resulted in the generation of 36
mixtures of 48 compounds each. The progress of the
synthesis was monitored by withdrawing small quan-
tities of resin, approximately 10–20 mg, randomly cho-
sen from 6 wells after addition of the first sidechain (1
compound per sample), and from a different set of 6
wells after the second coupling but before the second
pooling (4 compounds per sample). These samples were
cleaved with 90% TFA/H2O and the samples analyzed
by MS. In all cases the desired products were observed
as the major components, and no unprotected or
uncoupled products were observed. Analysis of the final
products was more complicated due to the number of
components in each mixture. Since the samples analyzed
during the course of the library synthesis verified the
successful attachment of the first two sidechains we were
most interested in determining the extent of conversion
for the final Alloc deprotection and amide bond forma-
With the scaffold in hand, we were poised to consider
the chemistry and the diversity elements that we
required for the synthesis of the library. Amide bond
formation is extensively utilized in combinatorial chem-
istry due both to the hardiness of the chemistry and the
vast diversity attainable from carboxylic acids and
amines, and we opted to pursue this chemistry to pre-
pare the library. The proposed chemistry was proto-
typed on solid phase in order to verify the robustness of
the chemistry and to identify any synthetic challenges
that may arise during library synthesis.
Solid phase prototyping began with attachment of the
scaffold (3 equiv based on resin loading) onto Tentagel
SRAM NH2 resin using a standard coupling cocktail of
3 equiv HATU9 and 7.5 equiv NMM in DMF (Scheme
2). Each step in the solid phase synthesis was performed
twice to ensure complete reaction. After attachment
onto the resin, the tyrosine amino group was depro-
tected and the liberated amine was coupled to 3-quino-
line carboxylic acid using the standard coupling
conditions to afford amide 4. Removal of the tri-
methylsilylethyl ester was accomplished with 2M TBAF
in THF. No effort was made to acidify the product, and
the resulting tetrabutylammonium carboxylate was
directly coupled with N,N-dimethylproline amide.
Removal of the Alloc group was carried out with
Pd(P(Ph)3)4 (0.3 equiv) and phenylsilane (15 equiv) in 9/
1 dichloromethane/ acetonitrile.10 The mixed solvent
system was necessary for adequate solvent detection by
the ACT 357 automated synthesizer used in the library
synthesis, and we found it necessary to wash the resin
1
tion. We investigated whether H NMR could provide
us with the necessary data. Twelve wells were analyzed,
Scheme 2. Reagents and conditions: (a) Tentagel SRAM NH2, HATU,
NMM, DMF; (b) 25% pip/DMF; (c) 3-quinolinecarboxylic acid, HATU,
NMM, DMF; (d) 2M TBAF, THF; (e) N,N-dimethylprolineamide,
HATU, NMM, DMF; (f) Pd(P(C6H5)3)4, PhSiH3, CH3CN/CH2Cl2; (g)
N-acetylproline, HATU, NMM, DMF; (h) 90% TFA/H2O, 73% overall.
Figure 1. Carboxylic acid and amine sidechains used for the library
synthesis.

R. W. Jackson et al. / Bioorg. Med. Chem. Lett. 13 (2003) 205–208
207
and indeed in all cases there was a complete absence of
alkene protons in the 5.10–5.40 ppm range indicating
the palladium catalyzed deprotection of the Alloc group
went to completion. Also, the absorptions due to the
scaffold and to the final sidechain, which all sample
members have in common, were by far the most promi-
nent absorptions in the spectra. We therefore concluded
that the library synthesis went as planned and was of
high quality.
compounds each. The screening data for the deconvo-
luted samples shows that a large percentage of the
activity is due to at most four compounds (Fig. 2B,
Wells 2, 14, 26). The four putative active components
varied only at the tyrosine amino group, and all pos-
sessed the dipentylamide moiety at the proline carbox-
ylate site indicating the need for a large hydrophobic
group at this site. The compounds were resynthesized in
solution, purified, and these screened in the apoptosis
assay. Three of the four purified compounds displayed
low micromolar activity in the assay (Fig. 2C). The lack
of significant discrepancy between the potency of the
crude library samples and the purified samples is an
additional indication of the library integrity. An IC50
was not determined for the pooled mixture from well 26.
The library was screened in a functional bioassay
designed to identify inhibitors of TNF-a signalling.
Addition of TNF-a to A549 cells in the presence of
actinomycin D causes the cells to undergo apoptosis.12
This is a critical part of cellular homeostasis,13 and if
left unchecked results in unwanted cell death and could
contribute to diseases such as sepsis or reperfusion
injury.14 TNF-a is also a pro-inflammatory cytokine,
and is known to play a role in a number of inflamma-
tory diseases such as rheumatoid arthritis, psoriasis, and
inflammatory bowel disease.15 To determine whether
the compounds could inhibit TNF-a, each of the 36
mixtures were screened in an apoptosis assay at 300 mM
total final assay concentration, resulting in each indivi-
dual compound being present at approximately 6 mM in
the mixture.12 The results of the assay are shown in
Figure 2. Many of the pools displayed significant inhib-
ition of apoptosis. The material derived from Pool 4
afforded the greatest inhibition, and chemical deconvo-
lution of that well was undertaken in order to identify
the active component(s). The deconvolution was carried
out on the ACT 357 and we were thus able to synthesize
24 discrete compounds along with 12 mixtures of 2
In summary, we have designed and synthesized a 1728
member split-pool combinatorial library based on a
novel tyrosine-proline peptidomimetic scaffold. Many
of the library pools were active in a TNF-a induced
apoptosis assay, and through one round of chemical
deconvolution of one of the pools we were able to
identify compounds 7, 8, and 10 as structurally novel
inhibitors of TNF-a signaling.
Acknowledgements
The authors would like to express their gratitude to
Devinder Singh and Kristen A. Kucera for performing
the biochemical studies, and also to Dr. Rich Kondrat
at the University of California, Riverside Mass Spec-
trometry Facility.
Figure 2. Screening results for the Tyr-Pro library—inhibition of TNF-a induced apoptosis: (A) Library pools 1–36, 300 mM total final assay con-
centration. (B) Deconvoluted compounds from Pool Number 4. Wells 1–24 are single compounds screened at 20 mM, wells 25–36 are mixtures of 2
compounds per well screened at 40mM total final assay concentration. (C) Structures of discrete active compounds and comparison of IC50’s between
the crude parallel synthesis material from solid phase and the resynthesized pure material.

208
R. W. Jackson et al. / Bioorg. Med. Chem. Lett. 13 (2003) 205–208
8. Jackson, R. W. Tetrahedron Lett. 2001, 42, 5163.
References and Notes
9. (a) Carpino, L. A. J. Am. Chem. Soc. 1993, 115, 4397.
(b) Carpino, L. A.; El-Faham, A. J. Org. Chem. 1995, 60,
3561.
1. Dolle, R. E. J. Comb. Chem. 2001, 3, 477, and references
therein.
2. (a) Furka, A.; Sebestyn, F.; Asgedom, M.; Dibo, G. Int. J.
Peptide Protein Res. 1991, 37, 487. (b) Lam, K. S.; Salmon,
S. E.; Hersh, E. M.; Hruby, V. J.; Kazmierski, W. M.; Knapp,
R. J. Nature 1991, 354, 82.
10. Dessolin, M.; Guillerez, M. G.; Thieret, N.; Guibe, F.;
Loffet, A. Tetrahedron Lett. 1995, 36, 5741.
11. Weygand, F.; Steglich, W.; Bjarnason, J.; Akhtar, R.;
Chytil, N. Chem. Ber. 1968, 101, 3623.
3. (a) Palkowitz, A. D.; Steinberg, M. I.; Thrasher, K. J.;
Reel, J. K.; Hauser, K. L.; Zimmerman, K. M.; Wiest, S. A.;
Whitesitt, C. A.; Simon, R. L.; Pfeifer, W.; Lifer, S. L.; Boyd,
D. B.; Barnett, C. J.; Wilson, T. M.; Deeter, J. B.; Takeuchi,
K.; Riley, R. E.; Miller, W. D.; Marshall, W. S. J. Med. Chem.
1994, 37, 4508. (b) Krapcho, J.; Turk, C.; Cushman, D. W.;
Powell, J. R.; DeForrest, J. M.; Spitzmiller, E. R.; Kar-
anewsky, D. S.; Duggan, M.; Rovnyak, G.; Schwartz, J.;
Natarajan, S.; Godfrey, J. D.; Ryono, D. E.; Neubeck, R.;
Atwa, K. S.; Petrillo, E. W., Jr. J. Med. Chem. 1988, 31, 1148.
(c) Rasmussen, P. H.; Ramanujam, S. H.; Berg, R. H. J. Am.
Chem. Soc. 1999, 121, 4738.
4. (a) Boldi, A. M.; Dener, J. M.; Hopkins, T. P. J. Comb.
Chem. 2001, 3, 367. (b) Poupart, M. A.; Cameron, D. R.;
Chabot, C.; Ghiro, E.; Goudreau, S. G.; Poirier, M.; Tsan-
trizos, Y. S. J. Org. Chem. 2001, 66, 4743.
5. Shibata, N.; Baldwin, J. E.; Jacobs, A.; Wood, M. E. Tet-
rahedron 1996, 52, 12839.
12. (a) Last-Barney, K.; Homon, C. A.; Faanes, R. B.; Mer-
luzzi, V. J. J. Immunol. 1988, 141, 527. (b) Briefly, 105 cells in
200 mL 10% FBS/RPMI antibiotic containing culture medium
were plated into 96-well round-bottom culture plates and
allowed to adhere for 6 h at 37 ꢀC in a 5% CO2 atmosphere.
The media was removed and 100 mL of RPMI+1 mg/mL
actinomycin-D was added to each well, followed by 90 mL of
test compound solution in 1% DMSO. This was incubated for
1 h. TNF-a (10 mL) was added to achieve a final assay con-
centration equal to its EC50 (normally 1 ng/mL) and the plates
incubated for 18 h. The media was aspirated from the plates
and 100 mL of 0.5% crystal violet in 20% methanol was
added. After 10min, the plates were rinsed with water to
remove excess stain, air dried, and absorbance measured at
590nm.
13. Chan, F. K. M.; Siegel, R. M.; Lenardo, M. J. Immunity
2000, 13, 419.
14. (a) Oberholzer, C.; Oberholzer, A.; Clare-Salzler, M.;
Moldawer, L. L. FASEB J. 2001, 15, 879. (b) MacLellan,
W. R.; Schneider, M. D. Circ. Res. 1997, 81, 137.
15. Locksley, R. M.; Killeen, N.; Lenardo, M. J. Cell 2001,
104, 487.
6. (a) Mitsunobu, O. Synthesis 1981, 1. (b) Bittner, S.; Assaf,
Y. Chem. Ind. (London) 1975, 281.
7. Campbell, D. A.; Bermak, J. C. J. Am. Chem. Soc. 1994,
116, 6039.