Synthesis of rigid trichostatin A analogs as HDAC inhibitors

Article abstract of DOI:10.1016/j.bmcl.2006.07.080

New inhibitors of histone deacetylase (HDAC) have been synthesized and evaluated for their activity toward non small lung cancer cell line H661. Their design is based on indanone (or tetralone) systems leading to trichostatin A (TSA) analogs with limited conformational mobility. Molecular modelization at the AM1 level revealed that the conformations of indane-based analogs and TSA bound to HDAC like protein are similar. The synthesis of these new analogs was achieved by alkylation of an appropriate indanone (or tetralone) to introduce the side chain bearing a terminal ester group, the latter being a precursor of hydroxamic acid and aminobenzamide derivatives. Hydroxamic acids with the TSA side chain were found to be the most active compounds and the presence of the dimethylamino group on the phenyl ring turned out to be essential to achieve low micromolar activities against H661 cancer cells.

Full text of DOI:10.1016/j.bmcl.2006.07.080

Bioorganic & Medicinal Chemistry Letters 16 (2006) 5339–5344
Synthesis of rigid trichostatin A analogs as HDAC inhibitors
a
a,
a
b
*
´
Cedric Charrier, Philippe Bertrand, Jean-Pierre Gesson and Joelle Roche
¨
a
`
´ ´
Laboratoire ‘‘Synthese et Reactivite des Substances Naturelles’’, UMR 6514, Universite de Poitiers et CNRS,
40 Avenue du Recteur Pineau, 86022 Poitiers, France
´
bInstitut de Physiologie et de Biologie Cellulaires, UMR 6187, Universite´ de Poitiers et CNRS, 40 Avenue du Recteur Pineau,
86022 Poitiers, France
Received 8 June 2006; revised 24 July 2006; accepted 25 July 2006
Available online 14 August 2006
Abstract—New inhibitors of histone deacetylase (HDAC) have been synthesized and evaluated for their activity toward non small
lung cancer cell line H661. Their design is based on indanone (or tetralone) systems leading to trichostatin A (TSA) analogs with
limited conformational mobility. Molecular modelization at the AM1 level revealed that the conformations of indane-based analogs
and TSA bound to HDAC like protein are similar. The synthesis of these new analogs was achieved by alkylation of an appropriate
indanone (or tetralone) to introduce the side chain bearing a terminal ester group, the latter being a precursor of hydroxamic acid
and aminobenzamide derivatives. Hydroxamic acids with the TSA side chain were found to be the most active compounds and the
presence of the dimethylamino group on the phenyl ring turned out to be essential to achieve low micromolar activities against H661
cancer cells.
Ó 2006 Elsevier Ltd. All rights reserved.
The eukaryotic DNA is structured in chromatin with
repetitive units called nucleosomes that not only provide
structural support for packing the nucleosomal DNA
but also are critical sites that control gene activation
and repression. At the center of the periodically assem-
bled units (about every 200 nucleotide base pairs) sealed
by histone H1 are the highly conserved histone proteins
H2A, H2B, H3, and H4 that form the octameric core
around which the DNA is folded. The assembly of
DNA in nucleosomal units is a fluid structure that
responds to the needs of the cells by remodeling regions
that are actively involved in gene transcription and
DNA replication. Regulation of histone acetylation,
methylation, and phosphorylation control chromatin
remodeling1. Modification of the R-amino group of spe-
cific lysines within histones by acetylation and deacetyla-
tion plays a crucial role in the transcriptional process
and deregulation can lead to cancer.2,3 This epigenetic
modification is regulated by two groups of enzymes
HAT (histone acetyl transferase) and HDAC (histone
deacetylase) with opposite activities.4,5 HDACs play
prominent roles in the transcriptional inactivation of
tumor suppressor genes. Many studies show that
inhibition of HDAC elicits anticancer effects in several
tumors by inhibition of cell growth and inducing cell
differentiation.6 In addition, recent data indicate that a
variety of proteins, including p53 a tumor suppressor
gene7 and a-tubulin involved in microtubule stability,8,9
are also substrates for HDACs. It is now clear that
many HDACs target non-histone proteins and have
functions beyond the alteration of chromatin struc-
tures.10 Therefore, inhibiting HDAC is an attractive
target for anticancer therapy.11–13
As a result of these findings, several programs for the
development of HDAC inhibitors (HDACI) as antican-
cer drugs have been initiated.14 Several natural com-
pounds have been isolated, like trichostatin A 115 (Fig. 1),
O
O
O
N
H
R
N
1 trichostatin A,R = H
2a trichostatin C, R = β-glucosyl
2b trichostatin D, R =α-glucosyl
H
N
O
O
N
H
H
Keywords: Histones deacetylases; Trichostatin A; Analogs; Anticancer
agents; Hydroxamic acids; Benzamides.
O
3 SAHA
*
Figure 1. Natural (1,2) and synthetic (3) HDACI.
Corresponding author. E-mail: philippe.bertrand@univ-poitiers.fr
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.07.080

5340
C. Charrier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5339–5344
Figure 2. TSA in HDLP protein (pdb 1C3R15).
a reversible HDACI,16 or synthesized, like SAHA 3,
actually in clinical trials (phase III).17
Structural requirements for HDAC inhibition have been
rationalized:18–20 a functional group able to chelate the
zinc atom at the bottom of the active site, a recognition
moiety interacting with the external surface of the pro-
tein, and a spacer linking these two moieties and fitting
the tubular geometry of the site (Fig. 2). This led to the
synthesis of a large number of more or less complex
compounds for SAR studies.21–26
Figure 4. Modeled conformations of TSA and of an indanone analog,
and crystallographic structures of TSA bound to HDLP (histone
deacetylase like protein) and HDAC8.
was carried out at the AM1 level31 (Fig. 4) and it was
found that the more stable conformer is rather close to
the computed and X-ray structures found for TSA in
the active site of HDAC8 or HDLP.18–20
HDACIs were classified in several groups: second gener-
ation hybrid polar compounds (HPCs) as SAHA 3,
benzamides, arylketones, benzo[de]isoquinolines, and
1,3-dioxanes. Another group is the cyclic hydroxamic
acid-containing compounds (CHAPs), where the recog-
nition part is a cyclic tetrapeptide.
The synthesis of these new analogs was implemented by
alkylation of an appropriate indanone (or tetralone,
Fig. 5) to introduce the unsaturated side chain bearing
a terminal ester group, the latter being a precursor of
hydroxamic acid and aminobenzamide derivatives.
TSA 1 itself has been synthesized in either racemic27 or
enantiopure28 form, and enantioselective preparation of
trichostatin D 229 has also been described. Biological
studies on TSA showed some toxicity and biological
instability limiting its clinical development: the hydroxa-
mic acid is hydrolyzed first to amide and then to carbox-
ylic acid, two inactive derivatives, and the amino group
is demethylated, giving metabolites rapidly eliminated in
biological systems.30 Furthermore, it should be consid-
ered that TSA may be sensitive to enolization and thus,
racemization leading to inactive compounds.
Indanone 4a is commercially available, tetralone 4b was
prepared from a-tetralone by methylation32 (79% over-
all yield), and fluoroindanone 4c was prepared from
fluorobenzene and 2-bromo-2-methylpropionylbromide
O
4a R=H, n=1
4b R=H, n=2
4c R=F, n=1
4d R=NMe2, n=1
( )n
R
Taking into account these observations, it seemed
worthwhile to explore the synthesis of rigid analogs of
TSA designed by linking the a position of the ketone
to the phenyl ring, leading to indanone (n = 1) or tetr-
alone (n = 2) derivatives (Fig. 3). Furthermore, other
modifications may be considered such ring substitution
or replacement of the hydroxamic acid by another Zn
chelating function.
Figure 5. Indanones and tetralone.
O
ii
R
N
N
7
For such analogs enolization will be impossible but the
flexibility of the phenyl ring with respect to the remain-
ing part of the molecule will also be suppressed. At this
stage, conformational analysis of the indanone analog
R
5 R=H
i
6 R=Me
iii
N NMe2
iv
4d
O
O
N
R= H, F, Me2 N
X= OH, 2-NH2Ph
n= 1,2
X
8
N
H
( )n
Figure 3. TSA analogs.
Scheme 1. Reagents and conditions: (i) CH2O, NaBH3CN, CH3CN,
98%; (ii) DDQ, THF, 90%; (iii) H2N–NMe2, pTsOH, Toluene, Dean–
Stark; (iv) LDA, MeI, ꢀ40°C; then H3O+, 78%.
R

C. Charrier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5339–5344
5341
O
O
O
O
O
O
OH
i
i
OEt
11a-d
12a-d, 85%
( )n
( )n
( )n
( )n
R
R
R
4a-b
9a, 82%
9b, 78%
ii
ii
O
O
O
H
H
v
O
iii
13a-d, 95%
O
H
( )n
10a, 68%
10b, 74%
( )n
iii
11a, 95%
11b, 90%
O
H
N
14a-d
H
H
( )n
OTHP
O
H
MeO
4c,d
O
R
iv
a : R=H, n=1
b : R=H, n=2
c : R=F, n=1
15a, 45%
15b, 54%
15c, 50%
15d, 27%
O
O
O
O
O
OH
N
iv
11c, 82%
11d, 80%
d : R=NMe2, n=1
( )n
( )n
H
R
R
16a, 85%
16b, 78%
16c, 40%
16d, 40%
Scheme 2. Reagents and conditions: (i) LDA, THF, ꢀ78 °C, isobu-
tyraldehyde; (ii) SOCl2, pyridine; (iii) SeO2, EtOH; (iv) a—LDA, THF,
ꢀ78 °C; b—H+.
N
NH2
H
a : R=H, n=1; b : R=H, n=2
c : R=F, n=1; d : R=NMe2, n=1
(AlCl3, CS2, quantitative yield).33 Attempts to directly
convert fluoroindanone to dimethylamino-2-methyl-
indanone 4d by nucleophilic aromatic substitution
(HMPA, 170 °C34 or Me2NH, NaH, DMSO, 120 °C35)
gave poor yields (<10%).
Scheme 3. Reagents and conditions: (i) Ph3P@CHCOOEt, toluene,
reflux, 12 h; (ii) LiOH, MeOH; (iii) H2N–OTHP, TBTU, DMF, NEt3;
(iv) CSA, CH2Cl2, MeOH; (v) H2N–Ph–NH2, EDC, THF.
Another strategy was used to prepare this material from
aminoindane 5 (Scheme 1). The latter was dimethylated
by reductive formylation (CH2O, NaBH3CN, CH3CN)
to give 6 which was then oxidized to 7 (DDQ, THF).36
Monomethylation of 7 (LDA, THF, CH3I) via its
N,N-dimethylhydrazone 8 finally afforded 4d (78% over-
all yield from 7).
O
ii
i
Br
4a
O
O
17
R
O
18, R = OEt
19, R = H
iii
O
O
iv
Introduction of the butadienoate chain was first carried
out from 4a,b via aldolization with isobutyraldehyde
(LDA, THF, ꢀ78 °C, Scheme 2). Aldols 9a,b were dehy-
drated (SOCl2, pyridine) to give alkenes 10a,b which
were oxidized to afford 11a,b as E isomers.37 A better
route was designed based on aldolization with 3-eth-
oxy-2-methyl-propenal as described by Wood,38 the
intermediate aldol being hydrolyzed in situ under acidic
conditions to afford 3-substituted 2-methyl-pronenal
derivatives. This methodology was first applied to 4b
and gave 11b in better yield (80%) as compared to the
first method (3 steps, 40% overall). Wood’s procedure
was then applied to fluoro and amino compounds 4c,d
to give the corresponding aldehydes 11c,d in good yield.
N
H
NH2
20
Scheme 4. Reagents and conditions: (i) LDA, THF, ꢀ78 °C, 85%; (ii)
LiCl, Bu4NCl, K2CO3, DMF, 80 °C, 65%; (iii) LiOH, MeOH, 60%;
(iv) H2N–Ph–NH2, EDC, THF, 40%.
A different butadienoate derivative was also prepared
with the aim of attempting double bond isomerization
to afford an alternative route toward 12a–d
(Scheme 4). Alkylation of indanone 4a with 2,3-dib-
romopropene afforded bromide 17 which was then
submitted to Heck coupling with ethyl acrylate. The
E dienoate 18 was hydrolyzed to acid 19, converted
as previously described to benzamide 20. Compound
19 could not be converted to the corresponding
hydroxamic acid and Pd catalyzed isomerization of
18 was not successful.
Wittig homologation (Ph3P = CHCOOEt, toluene, 12 h)
of aldehydes 11a–d gave the E,E-unsaturated esters 12a–
d (Scheme 3). After hydrolysis (LiOH, MeOH) to the
corresponding carboxylic acids 13a–d, hydroxamic acids
15a–d were obtained by treatment of 13a–d with the
known hydroxylamine O–THP39 ether (TBTU, DMF,
NEt3) giving 14a–d, followed by acid catalyzed removal
of the THP ether (CSA, CH2Cl2, MeOH). Aminobenza-
mides 16a–d were obtained by coupling 13a–d with 1,2-
diaminobenzene (EDC, THF, 64–85%).
All the hydroxamic and benzamide derivatives were
evaluated for their antiproliferative activity against
non small cell lung cancer H661 cells. IC50s were deter-
mined from two triplicate experiments by extrapolation

5342
C. Charrier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5339–5344
Table 1. Antiproliferative activities for TSA 1, SAHA 2, and TSA analogs
Entry
1
Compound
Inhibition IC50, lMa,b,c
<0.3a
0.1
Cell lines
H661
H129926
HCT11626
O
O
O
O
O
1
N
H
H
H
0.0013
N
H
N
2
3
4
5
3
>5a
7.24
1.93
H661
H129926
HCT11626
N
H
O
O
O
O
H
1.65b
H661
H661
H661
N
O
15a
15b
15c
H
O
H
3b
N
O
H
O
O
O
1.5b
N
H
H
F
O
O
O
N
H
H
6
7
15d
16a
<0.5b
0.2c
H661
H661
Me2N
O
O
O
O
3a
5a
7a
N
H
NH2
NH2
8
9
16b
16c
H661
H661
N
H
O
O
N
NH2
H
F
O
O
10
11
16d
20
2.5a
H661
H661
N
H
NH2
Me2N
O
O
>10
N
NH2
H
a Extrapolated IC50 from XTT tests at concentrations in range 1–10 lM.
b 1–5 lM.
c 0.025–1 lM.
to 50% proliferation (Table 1 and supplementary mate-
rial, Fig. 8). Values obtained were compared to previ-
ously reported data for TSA or SAHA.
Benzamides were found to be less active than the corre-
sponding hydroxamic acids, which is in agreement with
literature data. The tetralone analogs were found to be

C. Charrier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5339–5344
5343
half as potent than the corresponding indanones (Table
1, entries 3–8, compounds 15a,b–16a,b). Compound 20
appeared to be rather unstable and was not active (Table
1, entry 11).
Analogs not substituted on the phenyl ring or bearing a
fluoro substituent gave IC50s in the 1–5 lM range (Table
1, entries 5 and 9). On the other hand, the presence of a
dimethylamino group (Table 1, entries 6 and 10) appears
to be critical to achieve submicromolar activity.
This can be rationalized by analysis of the crystallized
form of HDLP with TSA which shows an interaction
between this amino group and Tyr91 (Tyr100 in
HDAC8) at the entry of the pocket, either with the car-
bonyl group of Tyr91 or by p–p stacking with the phenyl
ring (Fig. 6).
Then, compounds showing the best antiproliferative
activities were used for determination of acetylation of
histone H4 and tubulin using Western blot (Fig. 7).
The concentration of each compound was adjusted close
to the corresponding IC50 value (15c: 2.5 lM, 15b:
5 lM, 15d: 0.3 lM, TSA 1: 0.3 lM). At these concentra-
tions, 15b–d have good activities on acetylation of his-
tone H4. In particular, 15d shows a better response
than TSA at the same concentration. Evaluation on
tubulin acetylation was also determined for compounds
15b–d, with activities similar to those of TSA at the
given concentrations.
Figure 7. Western blots of: (A) histone H4 acetylation, (B) tubulin
acetylation, (C) a-tubulin for H661 cell treatment during 5 h 300 with:
1—1 (TSA) 0.3 lM, 2—control untreated cells, 3—15c 2.5 lM, 4—15b
5 lM, 5—15d 0.3 lM, (D) total protein extract of H661 cells stained by
Coomassie blue.
In conclusion, this work has demonstrated that con-
formationally constrained analogs of TSA, incorporat-
Acknowledgments
ing an indanone
nucleus
retain interesting
We thank MENRT, CNRS, and La Ligue Contre le
antiproliferative and HDAC inhibitory activities. The
dimethylamino group derivative compares favorably
to TSA, although being a racemic compound. Work
is in progress in our laboratory to prepare new substi-
tuted indanones as well as enantiomerically pure
compounds.
´
Cancer, Comite de Charente-Maritime for financial sup-
port, and Stephanie Sissoko for additional biological
´
evaluations.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2006.07.080.
References and notes
1. Chakraborty, S.; Senyuk, V.; Nucifora, G. J. Cell.
Biochem. 2001, 82, 310.
2. Cairns, B. R. Trends Cell Biol. 2001, 11, 15.
3. Lund, A. H.; van Lohuizen, M. Genes Dev. 2004, 18, 2315.
4. Kouzarides, T. EMBO J. 2000, 19, 1176.
5. Yang, X.-J.; Seto, E. Curr. Opin. Genet. Dev. 2003, 13,
143.
6. Marks, P. A.; Rifkind, R. A.; Richon, V. M.; Breslow, R.;
Miller, T.; Kelly, W. K. Nat. Rev. Cancer 2001, 1, 194.
7. Roy, S.; Packman, K.; Jeffrey, R.; Tenniswood, M. Cell
Death Differ. 2005, 12, 482.
Figure 6. TSA (pink) and an indanone analog (CPK) manually docked
in the active site of HDLP with Zn2+ (brown). Only Phe (green) and
Tyr (white) are shown with stacking of the vinylic methyl between
Phe141 and Phe198, and dimethylamino close to Tyr91 (produced
from 1C3R18 with MDL Chime plug in).
8. Hubbert, C.; Guardiola, A.; Shao, R.; Kawaguchi, Y.; Ito,
A.; Nixon, A.; Yoshida, M.; Wang, X.-F.; Yao, T.-P.
Nature 2002, 417, 455.
9. Matsuyama, A.; Shimazu, T.; Sumida, Y.; Saito, A.;
Yoshimatsu, Y.; Seigneurin-Berny, D.; Osada, H.; Koma-

5344
C. Charrier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5339–5344
tsu, Y.; Nishino, N.; Khochbin, S.; Horinouchi, S.;
Yoshid, M. EMBO J. 2002, 21, 6820.
10. Minucci, S.; Pelicci, P. G. Nat. Rev. Cancer 2006, 6, 38.
11. Johnstone, R. W. Nat. Rev. Drug Disc. 2002, 1, 287.
12. Somech, R.; Izraeli, S.; Simon, A. J. Cancer Treat. Rev.
2004, 30, 461.
13. Mai, A.; Massa, S.; Rotili, D.; Cerbara, I.; Valente, S.;
Pezzi, R.; Simeoni, S.; Ragno, R. Med. Res. Rev. 2005, 25,
261.
14. Monneret, C. Eur. J. Med. Chem. 2005, 40, 1.
15. Tsuji, N.; Kobayashi, M.; Nagashima, K.; Wakisaka, Y.;
Koizumi, K. J. Antibiot. 1976, 29, 1.
16. Yoshida, M.; Nomura, S.; Beppu, T. Cancer Res. 1987, 47,
3688.
17. Krug, L. M.; Curley, T.; Schwartz, L.; Richardson, S.;
Marks, P.; Chiao, J.; Kelly, W. K. Clin. Lung Cancer 2006,
7, 257.
18. Finnin, M. S.; Donigian, J. R.; Cohen, A.; Richon, V. M.;
Rifkind, R. A.; Marks, P. A.; Breslow, R.; Pavletich, N. P.
Nature 1999, 401, 188.
19. Vannini, A.; Volpari, C.; Filocamo, G.; Casavola, E. C.;
Brunetti, M.; Renzoni, D.; Chakravarty, P.; Paolini, C.;
De Francesco, R.; Gallinari, P.; Steinkuhler, C.; Di
Marco, S. Proc. Natl. Acad. Sci. U.S.A. 2004, 101,
15064.
24. Suzuki, T.; Nagano, Y.; Kouketsu, A.; Matsuura, A.;
Maruyama, S.; Kurotaki, M.; Nakagawa, H.; Miyata, N.
J. Med. Chem. 2005, 48, 1019.
25. Wu, T. Y. H.; Hassig, C.; Wu, Y.; Ding, S.; Schultz, P. G.
Bioorg. Med. Chem. Lett. 2004, 14, 449.
26. Remiszewski, S. W.; Sambucetti, L. C.; Bair, K. W.;
Bontempo, J.; Cesarz, D.; Chandramouli, N.; Chen, R.;
Cheung, M.; Cornell-Kennon, S.; Dean, K.; Diamantidis,
G.; France, D.; Green, M. A.; Howell, K. L.; Kashi, R.;
Kwon, P.; Lassota, P.; Martin, M. S.; Mou, Y.; Perz, L. B.;
Sharma, S.; Smith, T.; Sorensen, E.; Taplin, F.; Trogani, N.;
Versace, R.; Walker, H.; Weltcheck-Engler, S.; Wood, A.;
Wu, A.; Atadja, P. J. Med. Chem. 2003, 46, 4609.
27. Fleming, I.; Iqbal, J.; Krebs, E. P. Tetrahedron 1983, 39,
841.
28. Koseki, K.; Mori, K. Tetrahedron 1988, 44, 6013.
29. Hosokawa, S.; Ogura, T.; Togashi, H.; Tatsuta, K.
Tetrahedron Lett. 2005, 46, 333.
30. Elaut, G.; To¨ro¨k, G.; Vinken, M.; Laus, G.; Papeleu, P.;
Tourwe, D.; Rogiers, V. Drug. Metab. Dispos. 2002, 30,
1320.
31. Chem 3D tools (calculation from Mopac, Am1 method,
closed shield).
´
`
32. Henin, F.; M’Boungou-M’Passi, A.; Muzart, J.; Pete, J.-P.
Tetrahedron 1994, 50, 2849.
33. Maguire, A. R.; Papot, S.; Ford, A.; Touhey, S.; O’Con-
nor, S.; Clynes, M. Synthesis 2001, 1, 41.
34. Idoux, J. P.; Gupton, J. T.; Colo, C. Synth. Comm. 1982,
12, 907.
20. Somoza, J. R.; Skene, R. J.; Katz, B. A.; Mol, C.; Ho, J.
D.; Jennings, A. J.; Luong, C.; Arvai, A.; Buggy, J. J.; Chi,
E.; Tang, J.; Sang, B. C.; Verner, E.; Wynands, R.; Leahy,
E. M.; Dougan, D. R.; Snell, G.; Navre, M.; Knuth, M.
W.; Swanson, R. V.; McRee, D. E.; Tari, L. W. Structure
2004, 12, 1325.
35. Sircar, I.; Duell, B. L.; Cain, M. H.; Burke, S. E.; Bristol,
J. A. J. Med. Chem. 1986, 29, 2142.
21. Suzuki, T.; Ando, T.; Tsuchiya, K.; Fukazawa, N.; Saito,
A.; Mariko, Y.; Yamashita, T.; Nakanishi, O. J. Med.
Chem. 1999, 42, 3001.
36. Oikawa, Y.; Yonemitsu, O. J. Org. Chem. 1977, 42, 1213.
37. Rapoport, H.; Bhalerao, U. T. J. Am. Chem. Soc. 1971,
93, 4835.
22. Kim, Y. B.; Lee, K.-H.; Sugita, K.; Yoshida, M.;
Horinouchi, S. Oncogene 1999, 18, 2461.
38. McCann, E. E.; Janes, G.; Ortsey, C.; Wood, J. L.
Tetrahedron Lett. 1997, 38, 303.
23. Kapustin, G. V.; Fejer, G.; Gronlund, J. L.; McCafferty,
D. G.; Seto, E.; Etzkorn, F. A. Org. Lett. 2003, 5, 3053.
39. Patel, D. V.; Young, M. G.; Robinson, S. P.; Huniban, L.;
Dean, B. J.; Gordon, E. M. J. Med. Chem. 1996, 39, 4197.