Beta 3 agonists. Part 1: Evolution from inception to BMS-194449

Article abstract of DOI:10.1016/S0960-894X(01)00628-X

Screening of the BMS collection identified 4-hydroxy-3-methylsulfonanilidoethanolamines as full beta 3 agonists. Substitution of the ethanolamine nitrogen with a benzyl group bearing a para hydrogen bond acceptor promoted β₃ selectivity. SAR elucidation established that highly selective β₃ agonists were generated upon substitution of C_α with either benzyl to form (R)-1,2-diarylethylamines or with aryl to generate 1,1-diarylmethylamines. This latter subset yielded a clinical candidate, BMS-194449 (35).

Full text of DOI:10.1016/S0960-894X(01)00628-X

Bioorganic & Medicinal Chemistry Letters 11 (2001) 3035–3039
Beta 3 Agonists. Part 1: Evolution from Inception to
BMS-194449y
W. N. Washburn,* P. M. Sher, K. M. Poss, R. N. Girotra, P. J. McCann, A. V. Gavai,
A. B. Mikkilineni, A. Mathur, P. Cheng, T. C. Dejneka, C. Q. Sun, T. C. Wang,
T. W. Harper, A. D. Russell, D. A. Slusarchyk, S. Skwish, G. T. Allen, D. E. Hillyer,
B. H. Frohlich, B. E. Abboa-Offei, M. Cap, T. L. Waldron, R. J. George,
B. Tesfamariam, C. P. Ciosek, Jr., D. Ryono, D. A. Young, K. E. Dickinson,
A. A. Seymour, C. M. Arbeeny and R. E. Gregg
Bristol-Myers Squibb Pharmaceutical Research Institute, PO Box 4000, Princeton, NJ 08543, USA
Received 26 June 2001; accepted 7 September 2001
Abstract—Screeningof the BMS collection identified 4-hydroxy-3-methylsulfonanilidoethanolamines as full beta 3 agonists. Sub-
stitution of the ethanolamine nitrogen with a benzyl group bearing a para hydrogen bond acceptor promoted b3 selectivity. SAR
elucidation established that highly selective b3 agonists were generated upon substitution of Ca with either benzyl to form (R)-1,2-
diarylethylamines or with aryl to generate 1,1-diarylmethylamines. This latter subset yielded a clinical candidate, BMS-194449
(35).1 # 2001 Elsevier Science Ltd. All rights reserved.
The incidence of obesity and non-insulin dependent
diabetes mellitus (Type 2) is increasingat an alarming
rate in Western countries.2,3 Our goal was to utilize b3
adrenergic mediated thermogenesis to modulate obesity
and to lower plasma glucose and insulin levels thereby
Since large adult mammals including man do not have
defined BAT depots,7 our strategy was to utilize full
rather than partial b3 agonists to enhance the prob-
ability of elicitinga thermogenic response from what-
ever BAT is present in man. To minimize b1 and b2 side
effects, structural features were sought to reduce b1 and
b2 affinities (Ki) and intrinsic activity (IA). Ki for b1, b2,
and b3 and b3 IA were determined usingmembranes
isolated from CHO cells transfected with human b1, b2,
or b3 adrenergic receptors to insure that species depen-
dent differences in b3 responsiveness did not contribute
to clinical failure as was the case for BRL 37344 or
CL316243, for which rodent tissues had been employed
for SAR elucidation.3,8,9
4
amelioratingType 2 diabetes. Treatment of diabetic
rodents for 10–15 days with b3 agonists results in loss of
white adipose tissue (WAT), proliferation of brown
adipose tissue (BAT), and concurrent normalization of
elevated plasma glucose, insulin, nonesterified fatty acid
(NEFA), and triglyceride levels.5 In rodents, activation
of b3 receptors, localized almost exclusively on both
brown and white adipocytes, elevates c-AMP levels
thereby stimulatinglipolysis in WAT and upregulating
BAT specific genes.6 The increased expression of ther-
mogenin, a BAT specific protein that uncouples fatty
acid oxidation from oxidative phosphorylation, leads to
increased energy expenditure. Elevated NEFA con-
sumption necessitates increased glucose metabolism to
maintain homeostasis. The resultingdecrease in plasma
glucose leads to diminished insulin secretion and
peripheral insulin resistance.
Screeningof the Bristol-Myers Squibb compound col-
lection identified a subset of the b2 selective chemotype,
4-hydroxy-3-methylsulfonanilidoethanolamines
that
were full b3 agonists (1–8).10 Full adrenergic agonism of
this chemotype had been attributed to the bioisosteric
relationship between the left-hand portion of this class
and the catechol moiety of natural ligands adrenaline
and noradrenaline.11 Ki b3 for this chemotype was a
function of the ethanolamine N-substituent; alkyl moi-
eties strongly favored b2 whereas benzyl or phenethyl
promoted b3 particularly if the aryl ringbore a p-meth-
oxyl (see Table 1). Phenethyl containingderivatives 4–6
*Correspondingauthor. Fax: +1-609-818-3550; e-mail: william.
washburn@bms.com
ySee ref 1.
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00628-X

3036
W. N. Washburn et al. / Bioorg. Med. Chem. Lett. 11 (2001) 3035–3039
were particularly promisingas a startingpoint in view
of the 14-fold increase in selectivity for b3 relative to b2
upon progressively increasing the number of methoxy
substituents.
should bind in a specific fashion to fix the molecular
orientation; whereas, the loose interactions of the m-
chlorophenyl moiety of BRL 37344 and CL316243
would be more accommodating.
We anticipated that replacement of the methoxyl of 5
with an oxyacetic acid residue (10) would further
increase selectivity for b3 since we and others had con-
verted m-chlorophenylethanolamines to b3 selective
ligands by introduction of an appropriately oriented
anionic group.14 However, this modification diminished
b3 Ki 100-fold relative to 5; moreover, steric repulsion
was not the explanation since the methyl ester 9 bound
16-fold tighter than 10.15 Similar results were also
obtained for the corresponding m-substituted phenoxy-
acetic acid/ester (data not shown). We attributed these
failures to the difference in interactions of the left-hand
of these two chemotypes with the adrenergic receptors.
Hydroxysulfonanilides beinga catechol bioisostere
Before abandoningthis approach, 11, bearinga con-
formationally mobile valeric acid side chain as a Ca
substituent of 4, was synthesized to maximize the prob-
ability of achievingproper positioningof the anionic
charge (Table 2).16 Despite this feature, b3 affinity of 11
was diminished relative to that of 4; whereas, that of the
correspondingmethyl ester 12 was so enhanced that 12
became b3 selective. This effect was unique to the methyl
valerate side chain since the correspondinghydrocarbon
13, alcohol 14, and amide 15 were b2 selective. In the
course of exploringthe SAR of this a appendage, the
chain was cyclized to generate 16 and subsequently
aromatized to form 17. Although 16 bearinga cyclo-
hexyl moiety was b2 selective, 17 containinga planar
phenyl was b3 selective.
Table 1. Comparison of b3 AR agonist activity and selectivity versus
b1 and b2 for compounds 1–1012,13
Discovery that the 1,2-phenylethylamine moiety of 17
significantly enhanced b3 affinity of hydroxysulfonani-
lidoethanolamines was a critical advance. Small lipo-
philic or modestly polar para and/or meta substituents
of the a aryl ringincreased b1 and b2 affinity 2- to 4-
fold. Commensurate increases were found for b3 affinity
unless the para substituent was methoxy (18). The 10- to
12-fold further enhancement of b3 affinity due to the
para methoxyl of 18, 20, and 26 was not emulated by the
methyls of 22, chlorines of 23 or ethyls of 24. The para
methoxyl was the salient feature since an additional
meta methoxyl (20) modestly enhanced b3 binding
whereas ortho substitution (21) decreased b3 affinity
2-fold (Table 3).
Compd
R
b3 Ki b3 IA Selectivityb
(nM) (% act)
vs b1 vs b2
1a
i-Pr
t-Amyl
Adamantyl
4100 109
3700 125
11,000 110
2
0.6
0.3 0.03
2a
3a
15
2
4
4
2
22
0.04
0.07
0.3
1
0.25
2
4a
CH2CH2Ph
1100
480
91
93
5a
CH2CH2Ph-4-OMe
CH2CH2Ph-3,4-(OMe)2
CH2Ph
CH2Ph-4-OMe
CH2CH2Ph-4-OCH2CO2Me 3300
CH2CH2Ph-4-OCH2CO2H 49,000
6a
550 110
12,700
1550
7a
8a
95
79
77
67
45
93
9a
0.7 0.01
The similarity of this effect to that observed with methyl
ester 12 suggested that the effect was due to an appro-
priately oriented alkoxy fragment attached to a sp2 car-
bon. We suggest that the enhanced affinity is not steric
10a
2
18
20
0.06
1
11
BRL 37344
CL316243
36a
660
20,000
17,000
H
0.6 0.1
aRacemic mixture.
bSelectivity is defined as the ratio of b1 Ki or b2 Ki to b3 Ki.
Table 3. Comparison of b3 AR agonist activity and selectivity versus
b1 and b2 for compounds 18–2812,13
Table 2. Comparison of b3 AR agonist activity and selectivity versus
b1 and b2 for compounds 11–1712,13
Compd
X
b3 Ki
(nM)
b3 IA
(% act)
Selectivityb
Compd
R
b3 Ki
(nM)
b3 IA
(% act)
Selectivityb
vs b1
60
16
69
37
43
9
vs b2
vs b1
vs b2
0.3
18a
19a
20a
21a
22a
23a
24a
26
4-OMe
3-OMe
3,4-(OMe)2
2,4-(OMe)2
3,4-(Me)2
44
230
33
99
99
105
102
79
97
91
112
93
8
2
8
3
3
1
0.5
6
6
11a
12a
13a
14a
15a
16a
17a
28
(CH2)4CO2H
(CH2)4CO2Me
pentyl
(CH2)4CH2OH
(CH2)4CONHMe
cyclohexyl
Ph
(ꢀ) CH2Ph
87,000
570
320
57
89
77
72
104
80
133
1
26
8
10
7
7
5
0.5
0.5
0.5
0.7
3
70
480
133
414
520
11
1600
4300
360
3,4-(Cl)2
3,4-(Et)2
(R,R) 3,4-(OMe)2
(R,S) 3,4-(OMe)2
6
63
63
15
6
1500
0.3
27
320
aDiastereomeric mixture.
bSelectivity is defined as the ratio of b1 Ki or b2 Ki to b3 Ki.
aDiastereomeric mixture.
bSelectivity is the ratio of b1 Ki or b2 Ki to b3 Ki.

W. N. Washburn et al. / Bioorg. Med. Chem. Lett. 11 (2001) 3035–3039
3037
or electronic in origin, but rather stems from ability of
p-methoxy moiety of the a aryl ringto function as an H-
bond acceptor. Although proper spatial presentation of
two aryl rings are required to achieve full benefit, this
effect is also operative for N-benzyl- and N-phenethyl-
hydroxysulfonanilidoethanolamines since b3 affinity
increased 8- and 3-fold upon para methoxylation of
respectively 7 and 4 to generate 8 and 5. Thus, in this
instance, b3 selectivity was induced by increasing b3
affinity whereas b3 selectivity had been previously
achieved for BRL 37344, CL316243, and other related
agonists by introduction of an acidic moiety that
disfavored b1 and b2 binding.13
N-1,3-diphenyl-2-propylamine derivative 28 was b2
selective, presumably due to increased conformational
flexibility of the 1,3-diarylpropyl substituent. Conse-
quently, this series was not further pursued.
In contrast, since constraints inherent in the N-benzhy-
dryl moiety maintained b3 selectivity for 29, subsequent
synthetic efforts focused on the 1,1-diarylmethylamine
series. This effort, exploitingthe influence of hydrogen
bond accepting para substituents on activity at both the
b1 and b3 receptors, culminated with 34 and 35 (Table
4). Subsequently, incorporation of strongly electron
donatingsubstituents was shown to be undesirable.
Both 33 and 34 were more prone to undergo metabolic
cleavage and solvolysis resulting in formation of a non-
selective a and b adrenergic agonist, 4-hydroxy-3-
methylsulfonanilidoethanolamine11 36 (Table 1).
Methylation of Ca (Z=Me) was not beneficial since b1
and b2 affinity of 37 and 38 preferentially increased over
b3. In addition, the probability of acid catalyzed for-
mation of 36 was enhanced. As illustrated by 39
Knowingthat b3 agonists required a R configuration at
the hydroxylic center,10 the R,R and R,S diastereomers
26 and 27 comprising 20 were prepared by a route anal-
ogous to that outlined in Scheme 1, entailing sequential
alkylation of racemic 2-phenyl-1-(3,4-dimethoxy-
phenyl)ethyl amine with TBS protected (R)-iodohydrin
25, deprotection, and HPLC separation of diaster-
eomers.17 The b3 affinity for the R,R diastereomer 26
was 30-fold greater than that of the R,S isomer 27
thereby establishingthe preferred spatial orientation of
the rings. A similar pattern for b affinities and b3 IA was
observed for 17 other R,R/R,S pairs of diastereomers
(data not shown).18
(Z=CONH2), quaternizingC with an electronegative
a
moiety reduced affinities at all three b receptors, possi-
bly due to the inductive effect attenuatingthe ethanola-
mine basicity. The increase in b3 affinity and IA
correlated with hydrogen bonding capabilities of the
para substituent. Increasingsubstituent lipophilicity
modulated b1 IA as illustrated by generation of a weak
partial b1 agonist by replacement of the methoxyl of 34
with OCHF2 of 35.
Compounds 28 and 29 (Tables 2 and 4, respectively)
were prepared in an attempt to exploit the ability of a
neighboring aryl ring to promote b3 selectivity while
concurrently symmetrizingthe amine substituent to
reduce the number of chiral centers. Unfortunately, the
In vivo potency and the functional margin of separation
between b3 mediated effects versus b1 or b2 dependent
events were determined by iv administration of promis-
ing b3 agonists to ketamine anesthetized African green
Table 4. b1, b2, and b3 bindingaffinity and IA of compounds 29–
3912,13
Compd
R
Z
Ki b3 IA b3
Selectivity
IA b1
(nM)
(%)
(%)b
vs b1 vs b2
29a
30
31a
32
33
34
35
37
38
39
H
H
H
H
H
H
5800
44
43
78
70
96
100
77
79
94
106
5
5
CONMe2
CH2OMe
F
NHAc
OMe
OCHF2
OMe
OCHF2
OMe
980
2800
300
81
160
20
46
3
17
17
8
3
2
3
13
2
3
22
7
8
48
88
85
24
79
H
H
Me
Me
CONH2
81
220
2
10
Scheme 1. Reagents and conditions: (a) CH3SO2Cl, pyridine, 93%; (b)
CuBr2, EtOAc, ꢂCHCl3, 73%; (c) (R)-2-methyl-CBS-oxazaborolidine,
zole, DMAP, DMF, 70%; (f) CHClF2, 30% aq NaOH/iPrOH, 65 ꢂC,
84%; (g) NH2OH, 20 ꢂC, 95%; (h) Zn, HOAc; 90%; (i) diisopropyl-
ethylamine, THF, 140 ꢂC; NH4F, THF, 80%; (j) H2, Pd/C, methanol,
80%.
.
BH3 THF, ꢁ10 C, 86%; (d) NaI, acetone, 92%; (e) Et3SiCl, imida-
aRacemic mixture.
bb1 IA was determined by measuringthe acceleration in contraction of
spontaneously beatingguinea pigatria relative to that induced by
isoproterenol.19

3038
W. N. Washburn et al. / Bioorg. Med. Chem. Lett. 11 (2001) 3035–3039
Table 5. Response of African green monkeys to iv injection of 26 and
35a
References and Notes
1. (a) Prior preliminary communications of this work, see:
Washburn, W. N.; Girotra, R. N.; McCann, P. J.; Gavai, A.
V.; Mikkilineni, A. B.; Cheng, P.; Dejneka, T. C.; Sher, P. M.;
Sun, C. Q.; Wang, T. C.; Ryono, D.; Harper, T. W.; Russell,
A. D.; Slusarchyk, D. A.; Skwish, S.; Allen, G. T.; Hillyer, D.
E.; Frohlich, B. H.; Abboa-Offei, B. E.; Cap, M.; Waldron, T.
L.; George, R. J.; Tesfamariam, B.; Ciosek, C. P., Jr.; Young,
D.; Dickinson, K. E.; Seymour, A. A.; Arbeeny, C. M.; Gregg,
R. E. Abstract of Papers, MEDI-022, 216th National ACS
Meeting, Boston, MA, Aug 23–27, 1998; American Chemical
Society: Washington, DC, 1998. (b) Washburn, W. N., Giro-
tra, R. N., Sher, P. M., Mikkilineni, A. B., Poss, K. M.,
Mathur, A., Biacchi, G. S., Gavai, A. V., U.S. Patent
5,776,983, 1998; Chem. Abstr. 1995, 123, 2870.
2. Friedman, J. M. Nature 2000, 404, 632.
3. Amos, A. F.; McCarty, D. J.; Zimmet, P. Diabetic Med.
1997, 14, S7.
4. (a) Weyer, C.; de Souza, C. J. Drug Dev. Res. 2000, 51, 80.
(b) Bray, G. A.; Tartaglia, L. A. Nature 2000, 404, 672. (c)
Weyer, C.; Gautier, J. F.; Danforth, E., Jr. Diabetes Metab.
1999, 25, 11. (d) Kordik, C. P.; Reitz, A. B. J. Med. Chem.
1999, 42, 181. (e) Weber, A. E. Annu. Rep. Med. Chem. 1998,
33, 193.
Compd
Lipolysis
ED50 (mg/kg) onset of tachycardia decrease of serum K+
b1 Margin before
b2 Margin before
26
35
0.03
0.08
< 3
>6; <12
3
>60
aThe margin of separation was the ratio of the dose that produced the
onset (statistically significant) of a b1 or b2 event to the ED50 for
lipolysis.
monkeys. b1 agonist activity was reflected by tachy-
cardia; b2, by a decrease in serum K+ levels;20 b3, by an
increase in non-esterified fatty acids (NEFA). Failure to
see bluntingof lipolysis upon co-administration of the
b3 agonist and 0.1 mg/kg propranolol, a nonselective b1
and b2 antagonist, confirmed that the lipolytic response
of the most promising1,2-diarylethylamine 26 and 1,1-
diarylmethylamine 35 were b3 mediated.
The safety margin (Table 5) of the 1,1-diarylmethyl-
amine analogue 35 (BMS-194449) was superior than
that of the more potent 1,2-diarylethylamine 26. In this
primate model 35 appeared to be a partial b3 agonist
since a maximum NEFA elevation of 0.9 mEquiv/L
induced by 35 was less than the 1.3 mEquiv/L typically
induced by a full agonist such as 26.
5. Arbeeny, C. M.; Myers, D. S.; Hillyer, D. E.; Bergquist,
K. E. Am. J. Physiol. 1995, 268, E678.
6. Lowell, B. B.; Spiegelman, B. E. Nature 2000, 404, 652.
7. Himms-Hagen, J.; Ricquier, D. In Handbook of Obesity;
Bray, G. A., Bouchard, C., James, W. P. T., Eds.; Marcel
Dekher: New York, 1998; p 415.
The rat PK profile of 20 and 35 were not favorable; oral
bioavailability was less than 1–2%. Subsequent studies
with portal vein and bile duct cannulated rats revealed
that 70% of the drugadministered was converted pri-
marily to a mixture of two monoglucuronides of the
benzylic and phenolic hydroxyls alongwith minor
amount of the N-glucuronide of the sulfonamido moi-
ety. This transformation primarily occurred during
transit of the gut wall; however, hepatic glucuronidation
was also a factor. Although oxidative metabolism was a
very minor pathway, some N-dealkylation of 35 did
occur to generate the R enantiomer of 36. Despite this
poor prognosis for oral activity, we continued to pursue
this chemotype convinced that the combination of full
b3 IA and high b3 affinity offered the best opportunity
to ascertain whether b3 agonists could elicit a sustained
robust thermogenic response in man.
8. Cantello, B. C. C.; Smith, S. A. Drugs Future 1991, 16, 797.
9. Bloom, J. D.; Claus, T. Drugs Future 1994, 19, 23.
10. All compounds were evaluated usingthe truncated human
b3 adrenergic receptor reported by Emorine, L. T.; Marullo,
S.; Briend-Sutren, M. M.; Patey, G.; Tate, K.; Delavier-
Klutchko, C.; Strosberg, A. D. Science 1989, 245, 1118.
11. Larsen, A. A.; Gould, W. A.; Roth, H. R.; Comer, W. T.;
Uloth, R. H. J. Med. Chem. 1967, 10, 462.
12. Receptor bindingassays were run in triplicate with mem-
branes prepared from CHO cells expressingthe cloned human
receptor in the presence of 125I-iodocyanopindolol; b3 AR
agonist activity was assessed by measurement of cAMP accu-
mulation levels in CHO membranes expressinghuman b3
AR.13 IA was given as percent of the maximal stimulation
with isoproterenol.
13. Sher, P. M.; Fisher, L. G.; Skwish, S.; Michel, I. M.; Sei-
ler, S. M.; Washburn, W. N.; Dickinson, K. E. J. Med. Chem.
Res. 1997, 7, 109.
14. This approach exploited the findingthat b3 affinity is
unaffected by an appropriately oriented negative charge
whereas b1 and b2 affinities are diminished.13
Intravenous administration of 35 to 6 volunteers pro-
duced no separation between the onset of lipolysis and
b2 mediated prolongation of QT interval, suggesting that
lipolysis was either b2 mediated or that no separation
existed between doses producing b2 and b3 responses.21
Further studies with BMS-194449 were terminated.
15. Compounds 9–29 were prepared by sequentially treating
the correspondingamine with the previously described phena-
cyl bromide11 generated in step b of Scheme 1 for 30 min in
MeCN, followed by NaBH4 reduction in EtOH, and hydro-
genolysis over Pd/C in MeOH. Spectral data were fully con-
sistent with structures.
16. The amine precursor for 12 was obtained by sequential
treatment of cyclohexyl oxime dianion with (a) BnCl; (b) poly
H3PO4 at 130 ꢂC; (c) concd HCl/MeOH.
Acknowledgements
17. Absolute configuration of 27 determined by X-ray.
18. The 1,2-diarylethylamine series is the subject of the fol-
lowingpaper. Gavai, A. V.; Sher, P. M.; Mikkilineni, A. B.;
Poss, K. M.; McCann, P. J.; Girotra, R. N.; Fisher, L. G.;
Wu, G.; Bednarz, M. S.; Mathur, A.; Wang, T. C.; Sun, C. Q.;
Slusarchyk, D. A.; Skwish, S.; Allen, G. T.; Hillyer, D. E.;
We acknowledge the contributions of members of the
Bristol-Myers Squibb analytical chemistry department;
D. Strosbergfor transfected CHO cells; J. Gougoutas
for the X-ray based structure determination of 27; and
D. Floyd for suggesting use of OCHF2 as a substituent.

W. N. Washburn et al. / Bioorg. Med. Chem. Lett. 11 (2001) 3035–3039
3039
Frohlich, B. H.; Abboa-Offei, B. E.; Cap, M.; Waldron, T. L.;
George, R. J.; Tesfamariam, B.; Harper, T. W.; Ciosek, C. P.,
Jr.; Young, D. A.; Dickinson, K. E.; Seymour, A. A.;
Arbeeny, C. M.; Washburn, W. N. Bioorg. Med. Chem. Lett.
2001, 11, 3041.
19. Tesfamariam, B.; Allen, G. T. Br. J. Pharmacol. 1994, 112, 55.
20. Lipworth, B. J.; McFarlane, L. C.; Coutie, W. J.; McDe-
vitt, D. Eur. J. Clin. Pharmacol. 1989, 37, 297.
21. Unpublished work, communicated by P. Slugg, Bristol-
Myers Squibb, Princeton, NJ, USA.