Tandem optimization of target activity and elimination of mutagenic potential in a potent series of N-aryl bicyclic hydantoin-based selective androgen receptor modulators

Article abstract of DOI:10.1016/j.bmcl.2007.01.076

Pharmacokinetic studies in cynomolgus monkeys with a novel prototype selective androgen receptor modulator revealed trace amounts of an aniline fragment released through hydrolytic metabolism. This aniline fragment was determined to be mutagenic in an Ames assay. Subsequent concurrent optimization for target activity and avoidance of mutagenicity led to the identification of a pharmacologically superior clinical candidate without mutagenic potential.

Full text of DOI:10.1016/j.bmcl.2007.01.076

Bioorganic & Medicinal Chemistry Letters 17 (2007) 1860–1864
Tandem optimization of target activity and elimination of
mutagenic potential in a potent series of N-aryl bicyclic
hydantoin-based selective androgen receptor modulators
Lawrence G. Hamann,* Mark C. Manfredi, Chongqing Sun, Stanley R. Krystek, Jr.,
Yanting Huang, Yingzhi Bi, David J. Augeri, Tammy Wang, Yan Zou,
David. A. Betebenner, Aberra Fura, Ramakrishna Seethala, Rajasree Golla,
Joyce E. Kuhns, John A. Lupisella, Celia J. Darienzo, Laura L. Custer,
Jennifer L. Price, James M. Johnson, Scott A. Biller,
Robert Zahler and Jacek Ostrowski
Bristol-Myers Squibb, Pharmaceutical Research Institute, PO Box 5400, Princeton, NJ 08543-5400, USA
Received 20 December 2006; accepted 12 January 2007
Available online 27 January 2007
Abstract—Pharmacokinetic studies in cynomolgus monkeys with a novel prototype selective androgen receptor modulator revealed
trace amounts of an aniline fragment released through hydrolytic metabolism. This aniline fragment was determined to be muta-
genic in an Ames assay. Subsequent concurrent optimization for target activity and avoidance of mutagenicity led to the identifi-
cation of a pharmacologically superior clinical candidate without mutagenic potential.
Ó 2007 Elsevier Ltd. All rights reserved.
As men age, testosterone (T) levels decline and muscle
mass and strength erode in parallel.1 Over time this
can lead to frailty and ultimately necessitate the need
for assisted living.2 Testosterone replacement therapy
has shown significant clinical benefit at halting this
trend,3 though replacement therapy carries risks associ-
ated with poor selectivity for muscle versus prostate.4
Emerging selective androgen receptor modulators
(SARMs) have the potential to favorably impact this
selectivity equation and are consequently being actively
pursued by many groups.5
liability assays used to assess ‘drugability’. Although
compound 1 was well absorbed, highly bioavailable,
and metabolically stable in several preclinical species,
sensitive analytical detection methods allowed quantifica-
tion of trace amounts of 4-cyanonaphthylamine (2) in
urine after oral dosing of 1. This hydrolytic degradant
was not detectable in plasma of any species (at a lower
limit of quantitation of 1 ng/mL). However, approxi-
mately 0.003% and 0.1% of 2 (relative to the dose of par-
ent) were detected in the urine of dogs and cynomolgus
monkeys, respectively. Though the parent compound
tested negative in exploratory Ames assays, naphthyl-
amine 2 tested positive in these assays. As a mutagenic
metabolite would be undesirable for a drug candidate,
this finding prompted us to discontinue preclinical devel-
opment of this SARM in favor of more suitable agents.
We have recently reported the discovery of a novel series
of SARMs based on an N-aryl bicyclohydantoin scaf-
fold.6 Compounds in this class showed potent AR agonist
activity and >50-fold selectivity for stimulation of skeletal
muscle versus prostate growth in a standard rodent mod-
el. The prototype compound from this series also exhibit-
ed both favorable pharmacokinetic properties and an
encouraging profile on a standard battery of in-house
HO
O
H
N
N
CN
H2N
CN
O
Keywords: SARM; Androgen; Receptor; Mutagenicity; Aniline.
*
Corresponding author. Tel.: +1 203 677 6948; fax: +1 203 677
7884; e-mail: lawrence.hamann@bms.com
1
2
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.01.076

L. G. Hamann et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1860–1864
1861
respond to DNA damage detected by the transcription
complex by de-repression of SOS genes, resulting in
the expression of b-galactosidase, the gene product of
lacZ. Induction of SOS repair genes following DNA
damage is monitored by measuring b-galactosidase
activity, with cytotoxicity determined in parallel by con-
stitutively expressed alkaline phosphatase activity. This
medium throughput assay is formatted for 96-well
plates, has a typical turnaround time of two days, and
requires only 2 mg of test compound, compared with
30 mg and two weeks for a minimal Ames assay. Valida-
tion studies demonstrated that the SOS Chromotest
exhibits high concordance with the Ames test, with a
small number of false positive and negatives (for 121
proprietary and 46 commercial compounds, the overall
concordance was 87%).12 The sensitivity of the SOS
Chromotest for detecting mutagens is 79% and the spec-
ificity is 91%. Despite these limitations the SOS Chro-
motest has proved useful as an expedient first screen
for potential mutagens.
In seeking to obviate the potential for mutagenic metab-
olites arising from this otherwise promising chemotype,
we explored three separate strategies in parallel, in order
of increasing synthetic complexity: (1) optimization of
the aryl amine moiety of the bicyclohydantoin-based
compounds to eliminate mutagenic potential should
trace amounts be liberated in vivo; (2) modifications to
the core scaffold of these SARM compounds such that
an aryl amine moiety could not readily be released;
and (3) replacement of the aniline nitrogen with carbon
to exclude a formal aryl amine moiety from the mole-
cule. This report focuses on our efforts in the former
strategic avenue.7 Analysis of the X-ray co-crystal struc-
ture of compound 1 bound to the ligand binding domain
(LBD) of the AR6 (Fig. 1) suggested that the binding
energy from a p-edge-to-face interaction with the side
chain of Phe764 was mediated primarily through the
proximal aryl ring of the naphthyl moiety. As such, we
hypothesized that it would be possible to capture the
hydrophobic space-filling and electronic properties of
the distal aryl ring of the naphthyl system through sub-
stituents that would be predicted to mitigate known
physicochemical features correlated with mutagenic
potential.
The SOS Chromotest was used as part of a lead optimi-
zation scheme for concomitant elimination of the muta-
genic potential of the aryl amine moiety in this SARM
scaffold. As previously indicated, X-ray co-crystal struc-
tural analysis of advanced lead 1 suggested that an aryl
moiety in this region of the AR LBD was important for
binding and desired biological efficacy.6 A modified ver-
sion of DEREK, a rules-based software program for
predicting mutagenic potential, was used to guide the
selection of replacement aryl amines for synthesis.13,14
Aryl amines are not generally the principal mutagenic
species, but rather are precursors for metabolic conver-
sion to hydroxylamine and mutagenic nitroso species
(Fig. 2). Analyses of the factors contributing to the
mutagenicity of aryl amines indicate direct correlations
with the number of fused rings (hydrophobic surface
area, DNA intercalation), logP (hydrophobicity, cell
permeability), and inverse correlations with polarizabil-
ity (amine oxidation potential) and steric accessibility of
the amine group.15–18
In order to efficiently execute lead optimization of this
nature, we considered it essential to be able to assess
both the activity of the parent molecules on the target
of interest (AR functional assay) and the mutagenic po-
tential of the aryl amine fragment in parallel. Although
the Ames assay has long been viewed as the gold stan-
dard for determining mutagenic potential,8,9 compound
requirements and throughput are generally not amena-
ble to integration with rapid lead optimization cycles.
A more convenient approximation of genotoxic poten-
tial can be found in the SOS Chromotest assay.10
SOS repair refers to sparingly used, error-prone transla-
tional synthesis, a last resort to allow cell survival.11 The
SOS Chromotest employs a genetically engineered Esch-
erichia coli containing a lacZ reporter gene under tran-
scriptional control of a SOS repair gene. Bacteria
Assessments of mutagenic potential were integrated into
our lead optimization work flow, generally as follows:
aryl amines chosen with in silico guidance were incorpo-
rated into new SARM analogues which were tested in
cell-based AR transcriptional assays;6 amines leading
to sufficiently potent SARMs were then evaluated in
the SOS Chromotest; potent SARMs whose aryl amine
component was negative in the SOS Chromotest were
evaluated in vivo; adequate in vivo activity then trig-
gered verification of the lack of mutagenic potential of
the aryl amine in an Ames assay.
X
H
N
HO
X
X
N
O
H2N
n
n
n
Figure 1. X-ray co-crystal structure of 1 bound to the rat AR-LBD at
˚
2.0 A resolution. The helices and loops are colored magenta. The key
Figure 2. Aryl amine activation to mutagenic species and factors
affecting mutagenic potential: n, number of rings; X, electronic and
steric factors.
ligand binding residues are labeled and the dashed lines represent
potential hydrogen bonds.

1862
L. G. Hamann et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1860–1864
SARM analogues were prepared in a convergent fashion
as previously described (Scheme 1).6,19 Bicyclic hydanto-
in formation consisted of condensation of various aryl
isocyanates with cis-3-hydroxy-L-proline methyl ester.
Aryl isocyanates were derived from the corresponding
aryl amines, which were typically prepared through pro-
tection of a commercially available aryl amine as the
acetamide, selective bromination at the para position,
followed by Pd-mediated conversion to the correspond-
ing aryl nitrile using CuCN. Acetamide hydrolysis under
acidic conditions and subsequent treatment with phos-
gene provided the appropriate isocyanates. All final
compounds were purified to >98% HPLC purity and
1
characterized by H NMR and LC/MS.
Binding affinities and agonist activities were measured
for each SARM, and mutagenic potential for the respec-
tive aryl amine was assessed (Table 1). Because we had
previously determined that aryl group substituent toler-
ance for AR binding allowed at most a 1,2,3,4-substitu-
tion pattern, and that activity dropped off precipitously
with additional substitutions at the 5- or 6-positions
(data not shown), the scope of the current data set is
limited to this partially optimized chemotype space. Vir-
tually all of the analogues prepared in this series resulted
in potent binding affinity to the AR, consistent with
molecular modeling docking and X-ray structural data
suggesting that the hydroxy group, the aryl ring cen-
troid, and the nitrile nitrogen lone–pair interaction are
the key pharmacophoric determinants for receptor bind-
ing interactions. In contrast, the functional agonist
activity for this series was widespread, spanning more
than five orders of magnitude.
a,b
c,d,e
H2N
HN
Ac
Br
OCN
CN
R1
R2
O
R1
R2
R1
R2
3
4
5
CO2Me
H
HO
HO
H
NH
N
CN
6
N
R1
R2
f,g
O
7-31
Scheme 1. Preparation of N-aryl hydantoin SARM compounds.
Reagents and conditions: (a) Ac2O, AcOH, NaOH; (b) Br2; (c) CuCN,
Pd(dba)2, DMF; (d) HCl, EtOH; (e) phosgene, NaHCO3, CH2Cl2; (f)
i-Pr2NEt, toluene; (g) DBU, toluene, rt, 12 h.
Although simple N-substitution into the naphthyl ring
of 1 to give quinoline compound 7 obviated the aryl
Table 1. AR agonist potencies in functional assays and SOS chromotest assay results for reference compounds and SARMs 1 and 7–31
Compound
R1
R2
AR EC50 (nM)a Ki (nM) SOS Chromotest fold induction of
corresponding aniline fragment, S9 fractionb aniline fragmentc
Ames result for
2-Aminoanthracene
NAd
4-Nitroquinoline oxide NA
NA
NA
NA
NA
NA
NA
2.3
NA
NA
NA
3.2
21
5
11–2
Positive
Positive
Positive
Positive
Negative
NT
Aflatoxin B1
1
NA
23–24
Fused phenyl
2.02/1.58
1.05/1.03
NTe
7
Fused 5-pyridyl 6.2
H
8
H
>10,000
98
34
24
9
H
OMe
CF3
Cl
1.29/1.40
1.40/1.41
1.87/1.78
NT
Positive
Negative
Positive
NT
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
H
>10,000
2.8
345
7.8
1.5
48
H
H
Ph
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
F
H
Me
52
1.7
8.5
4.5
nd
NT
NT
NT
1.62/1.28
1.44/1.54
1.34/1.46
NT
CO2Me >10,000
NT
CF3
OH
OMe
F
3.6
2.9
nd
Negative
NT
NT
>10,000
487
1.7
3.7
1.3
2.1
1.2
8.0
4.3
3.0
NT
6.1
0.7
8.1
7.9
14
NT
1.39/1.32
1.46/1.35
1.31/1.24
NT
NT
Cl
Br
0.7
0.6
Negative
NT
NT
CN
F
2.2
47
NT
NT
NT
NT
F
Cl
Cl
Cl
1.0
NT
607
1.7
1.71/1.83
1.35/1.25
1.64/1.55
NT
NT
NT
Cl
Cl
OMe
Me
NT
NT
OMe Me
OMe CF3
221
1080
79
1.14/1.38
NT
1.42/1.13
NT
NT
Et
Et
H
CF3
36
3.3
NT
a Values are means of three experiments, standard deviation is given in parentheses; nd, not determined.
b Values are means of three experiments, P1.50-fold induction indicates a positive mutagenic response in this assay.
c TA 98 and TA 100 test strains of Salmonella typhimurium, P2-fold dose-related elevation in the mean revertant counts relative to DMSO controls.
d NA, not applicable.
e NT, not tested.

L. G. Hamann et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1860–1864
1863
ner, R. N.; Garry, P. J. Metabolism 1997, 46, 410; (b)
Harman, S. M.; Metter, E. J.; Tobin, J. D.; Pearson, J.;
Blackman, M. R. J. Clin. Endocrinol. Metab. 2001, 86,
724; (c) Feldman, H. A.; Longcope, C.; Derby, C. A.;
Johannes, C. B.; Araujo, A. B.; Coviello, A. D.; Bremner,
W. J.; McKinlay, J. B. J. Clin. Endocrinol. Metab. 2002,
87, 589.
amine mutagenic potential, this change resulted in an
order of magnitude loss in binding affinity, and so as a
general rule (also considering the inverse correlation of
mutagenic potential with the number of fused rings),
we chose to limit our SAR explorations to monocyclic
aryl substitutions into this scaffold. Both dramatic steric
and electronic effects were seen in response to substitu-
ent changes in these two positions. Substitution ortho
to the aryl nitrogen in this series typically resulted in
2. (a) Matsumoto, A. M. J. Gerontol. 2002, A, M76; (b)
Kaufman, J. M.; Vermeulen, A. Endocr. Rev. 2005, 26,
833.
1
atropoisomerism, readily observable by H NMR. This
3. (a) Snyder, P. J.; Peachey, H.; Hannoush, P.; Berlin, J.
A.; Loh, L.; Lenrow, D. A.; Holmes, J. H.; Dlewati, A.;
Santanna, J.; Rosen, C. J.; Strom, B. L. J. Clin.
Endocrinol. Metab. 1999, 84, 2647; (b) Ferrando, A.
A.; Sheffield-Moore, M.; Yeckel, C. W.; Gilkison, C.;
Jiang, J.; Achacosa, A.; Lieberman, S. A.; Tipton, K.;
Wolfe, R. R.; Urban, R. J. Am. J. Physiol. Endocrinol.
Metab. 2002, 282, E601; (c) Blackman, M. R.; Sorkin, J.
conformational restriction and presumed consequent
reduction in entropy appears to correlate well with ago-
nist potency in the cell-based assays (with the exception
of Cl-analogue 11), as compounds 8–12 suffer at least a
30-fold loss in potency, despite maintaining reasonable
binding affinity. Size limitations appeared to exist at
R2, as seen with compound 12. In the absence of coun-
terposing electronic factors, agonist potency is restored
by the addition of a simple methyl group at R1 (compare
10/16 and 11/20).
D.; Munzer, T.; Bellantoni, M. F.; Busby-Whitehead, J.;
¨
Stevens, T. E.; Jayme, J.; O’Connor, K. G.; Christmas,
C.; Tobin, J. D.; Stewart, K. J.; Cottrell, E.; St.Clair, C.;
Pabst, K. M.; Harman, S. M. J. Am. Med. Assoc. 2002,
288, 2282; (d) Bhasin, S.; Woodhouse, L.; Casaburi, R.;
Singh, A. B.; Mac, R. P.; Lee, M.; Yarasheski, K. E.;
Sinha-Hikim, I.; Dzekov, C.; Dzekov, J.; Magliano, L.;
Storer, T. W. J. Clin. Endocrinol. Metab. 2005, 90, 678;
(e) Page, S.; Amory, J. K.; Bowman, F. D.; Anawalt, B.
D.; Matsumoto, A. M.; Bremner, W. J.; Tenover, J. L.
J. Clin. Endocrinol. Metab. 2005, 90, 1502.
Electron donating groups appear to be poorly tolerated
throughout (as in 9, 17, 18, 26, 28, and 29), perhaps due
to attenuation of the strength of aryl p-edge/face interac-
tions with F764 in the AR LBD, as well as reduced
H-bonding capacity of the CN group. The presence of
two lipophilic and/or electron-withdrawing groups ap-
peared optimal at both R1 and R2, with halogen and
methyl groups generally optimal. All SARMs discov-
ered in this series with EC50’s < 5 nM follow this trend
(i.e., 11, 14, 16, 19–22, 24, and 27). The most potent
agonists (EC50 < 1 nM) were those where R1 = Me and
R2 = halogen (20 and 21).
4. (a) Basaria, S.; Dobs, A. S. Am. J. Med. 2001, 110, 563;
(b) Gruenwald, D. A.; Matsumoto, A. M. J. Am.
Geriat. Soc. 2003, 51, 101; (c) Rhoden, E. L.; Morgen-
taler, A. N. Eng. J. Med. 2004, 350, 482; (d) Liu, P. Y.;
Swerdloff, R. S.; Veldhuis, J. D. J. Clin. Endocrinol.
Metab. 2004, 89, 4789; (e) Hijazi, R. A.; Cunningham,
G. R. Ann. Rev. Med. 2005, 56, 117.
5. (a) Zhi, L.; Martinborough, E. Ann. Rep. Med. Chem.
2001, 36, 169; (b) Gao, W.; Bohl, C. E.; Dalton, J. T.
Chem. Rev. 2005, 105, 3352; (c) Segal, S.; Narayanan,
R.; Dalton, J. T. Expert Opin. Investig. Drugs 2006, 4,
377.
6. Sun, C. Q.; Robl, J. A.; Wang, T. C.; Huang, Y.; Driscoll,
J. E.; Lupisella, J. A.; Beehler, B. C.; Golla, R.; Sleph, P.
G.; Seethala, R.; Fura, A.; Krystek, S. R., Jr.; An, Y.;
Malley, M.; Sack, J. S.; Salvati, M. E.; Grover, G. J.;
Ostrowski, J.; Hamann, L. G. J. Med. Chem. 2006, 49,
7596.
7. Communication of our efforts in these other areas is the
subject of forthcoming reports.
8. Zeiger, E. Cancer Res. 1987, 47, 1287.
9. Mortelmans, K.; Zeiger, E. Mutation Res. 2000, 455, 29.
10. Hofnung, M.; Quillardet, P. Ann. N.Y. Acad. Sci. 1988,
534, 817.
11. Sutton, M. D.; Smith, B. T.; Godoy, V. G.; Walker, G. C.
Annu. Rev. Genet. 2000, 34, 479.
12. Quillardet, P.; Hofnung, M. Mutation Res. 1993, 297,
235.
13. Pearl, G. M.; Livingston-Carr, S.; Durham, S. K. Curr.
Top. Med. Chem. 2001, 1, 247.
14. Mattioni, B. E.; Kauffman, G. W.; Jurs, P. C.; Custer, L.
L.; Durham, S. K.; Pearl, G. M. J. Chem. Inf. Comput. Sci.
2003, 43, 949.
With respect to mutagenic potential, less clear SAR
trends emerged. Naphthylamine 2 was definitively muta-
genic in both SOS Chromotest and Ames assays. Con-
sistent with known aniline mutagenicity SAR, the aryl
amine components of several compounds without steric
encumbrance proximal to the amino group showed a
positive mutagenic response in one or both assays
(e.g., those from 9 and 11). Likely owing to the success
of aniline pre-selection criteria, most of the anilines test-
ed in the overall series were negative in the SOS Chro-
motest, and the generally high predictive power of this
assay was consistent within this group of anilines (with
the exception of that from 9). Though only a subset of
anilines were tested in Ames assays, no false positives
were observed in the SOS Chromotest, a desirable
feature in optimizing this parameter in concert with
pharmacological endpoints. Consequently, increasing
the throughput and decreasing the compound require-
ment for a mutagenicity screen allowed for the iterative
optimization of both efficacy and mutagenicity in paral-
lel, resulting in the rapid elimination of this unaccept-
able liability and identification of compound 20, which
advanced to human clinical trials.20,21
15. Mersch-Sundermann, V.; Schneider, U.; Klopman, G.;
Rosenkranz, H. S. Mutagenesis 1994, 9, 205.
16. Hatch, F. T.; Colvin, M. E. Mutation Res. 1997, 376, 87.
17. Shibutani, S.; Grollman, A. P. Mutation Res. 1997,
376, 71.
References and notes
18. Maran, U.; Karelson, M.; Katritzky, A. R. Quant. Struct.-
Act. Relat. 1999, 18, 3.
1. (a) Morley, J. E.; Kaiser, F. E.; Perry, H. M., 3rd; Patrick,
P.; Morley, P. M.; Stauber, P. M.; Vellas, B.; Baumgart-

1864
L. G. Hamann et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1860–1864
19. Sun, C.; Hamann, L.; Augeri, D.; Robl, J.; Huang, Y.-T.;
Wang, T.; Simpkins, L.; Holubec, A. PCT Int. Appl. WO
03096980, 2003, 176 p.
20. Ostrowski, J.; Driscoll, J. E.; Lupisella, J. A.; Manfredi,
M. C.; Beehler, B. C.; Bi, Y.; Sun, C.-Q.; Seethala, R.;
Golla, R., ; Sleph, P. G.; Fura, A.; Narasimhan, N.;
Krystek, S. R., Jr.; An, Y.; Kish, K.; Sack, J. S.;
Bonacorsi, S. J.; Cao, K.; Mookhtiar, K. A.; Grover, G.
J.; Hamann, L. G. Endocrinology 2007, 148, 4.
21. Wilson, E. M. Endocrinology 2007, 148, 1.