1948
J. E. Hempel et al. / Bioorg. Med. Chem. Lett. 26 (2016) 1947–1953
analog 9b, driven by cLogP reduction. Phenyl analog 9f indicated
steric disfavoring of large substituents. In marked contrast to the
favorable structure activity relationships (SAR) of the R1 function-
ality, substitution of the R2 thiophene in EGM1 showed little toler-
ance for alternate (hetero)aromatic and cycloalkyl amides with
major potency losses across the series (9g–n). Only 2-methylpiper-
idine 9m displayed an acceptable reduction in potency (ꢀ2-fold);
however, an LD50 of 16.7 lM precluded further investigation.
Therefore, bioisosteric replacement of the thiophene group for a
phenyl ring was seen as a prudent strategy going forward despite
the modest potency of 9g. Finally, analogs 9o, 9p, and 10a indi-
cated disfavored incorporation of polar atoms into the western
cyclohexyl ring.
Having established SAR for the most readily modifiable groups
of EGM1, we next sought to explore replacements for the cyclo-
hexylthiophene core, which we postulated could ameliorate inher-
ent EGM1 solubility limitations. Therefore, we intercepted the
Figure 1. Structures of FDA-approved Smo antagonists (1, 2), eggmanone (EGM1,
3), and the Gli antagonist GANT-61 (4).
Scheme 1 synthetic route with
a-aminoarylesters, including ben-
zothiophene 11 derived from three step conversion of 2-aminocy-
clohexylthiophene 6a by a protection, oxidation, and deprotection
strategy (Scheme 2). Benzothiophene 11 and readily available
methyl anthranilate were converted to EGM1-like structures fol-
lowing Scheme 1 steps b–e, providing analogs 12 and 14a–c.
Similar to analogs 14a–c, in which we removed the cyclohexyl
ring, arrival at the terminal thiophene core commenced by a mod-
ified Gewald reaction37 with 1,4-dithiane-2,5-diol (15, Scheme 3).
As in Scheme 1, conversion to dithiourea 17 was followed by
two-step cyclization with primary amines to yield cyclic thioureas
18 which were S-alkylated to provide EGM1 analogs 19a–c. In light
of biological activity presented in Table 2, we further explored the
role of the linker with 20a and 20b in which secondary and tertiary
alcohols were formed from ketone 19c by either reduction with
NaBH4 or Grignard addition with MeMgCl. Additionally, chloropy-
rimidinone 21 provided the necessary electrophilic center to inves-
tigate thioether modifications as the ether 22a or secondary amine
22b (Fig. 3).
with Gli antagonism (GANT-61, 4)26 and bromodomain inhibi-
tion27,28 emerging as two approaches. In addition to our studies
with EGM1, mounting evidence has linked PDE4 to Hh signaling
and tumorigenesis.29,30 Therefore, we viewed EGM1 as a starting
point for in vitro probe development toward an optimized down-
stream of Sufu Hh inhibitor; however, EGM1’s limited aqueous sol-
ubility and modest potency required improvement.
We initially targeted three points of modification of EGM1:
incorporation of polar atoms in the cyclohexyl ring, substitution
of the methylallyl functionality, and replacement of the pendant
thiophene. In linear fashion starting with cyclohexanone and
derivatives (5a–d), the tricyclic ring system was fashioned from
left to right starting with a Gewald reaction31 to provide the 2-
aminothiophene 6. In most cases, formation of the dithiourea 732
was followed by two-step cyclization with primary amines to yield
the cyclic thiourea 8; however, on smaller scales, direct formation
of thioureas with isothiocyanates and subsequent cyclization also
arrived at 8.33 S-alkylation under mild conditions installed alkyl
ketones to provide EGM1 analogs 9a–n and tetrahydropyran and
tetrahydrothiopyran analogs 9o and 9p. In the case of Y = NBoc,
further Boc deprotection gave piperidine 10a.
When choosing our primary assay for analog evaluation, consid-
eration was given to two factors: cellular reduction of Hh target
gene Gli1 transcription is directly linked to in vivo control of tumor
growth,10,12 and micromolar potency of the marketed PDE4 inhibi-
tor Roflumilast for Hh signaling inhibition (data not shown)
indicated partial contribution of PDE4 to EGM1’s observed Hh
inhibition. Therefore, we prioritized phenotypic analog evaluation
in the Gli-responsive luciferase reporter line TM3-Gli-Luc34
stimulated with 20 nM Smo agonist (SAG) while concurrently
monitoring non-specific cellular toxicity. Additionally, lipophilic
efficiency (LipE),35 a concurrent readout of a compound’s potency
(pEC50) and lipophilicity (cLogP) derived by subtracting the latter
from the former, was utilized to optimize solubility properties
(Table 1), allowing for straightforward tracking of compounds with
improvements in both parameters. Marketed Smo antagonists Vis-
modegib and Sonidegib displayed expected potency, while the EC50
of the Gli antagonist GANT-61 was slightly higher than the
reported ꢀ5
with the response of known Hh inhibitors, we profiled EGM1 and
noted a benchmark EC50 for EGM1 of 1.34 M and a corresponding
l
M EC50 in the Shh-LIGHT2 reporter line.36 Pleased
l
LipE of 0.73. Given that the low-nanomolar EC50s of 1 and 2
support LipEs of 4.33 and 3.02 respectively, we targeted a LipE
for optimized EGM1 analogs of >2 deriving from an EC50 of
<1 lM (pEC50 > 6). Replacement of the methylallyl group with
small alkyl and cycloalkyl substituents revealed broad tolerance
Scheme 1. Reagents and conditions: (a) S8, NCCH2CO2Me, Et3N, EtOH, rt, 16 h, 49–
80%; (b) CS2, NaOH, DMSO, H2O, rt, 1 h then Me2SO4, rt, 3 h, 72–93%; (c) R1NH2,
Et3N, CH3CN, 90 °C, 16 h; KOH, EtOH, H2O, 70 °C, 4 h, 17–72%, two steps; (d) R1NCS,
PhMe, 115 °C, 72 h; KOH, EtOH, H2O, 70 °C, 4 h, 16–80%, two steps; (e) XCH2C(O)R2,
Cs2CO3, CH3CN, rt, 3 h, 9–88%; (f) TFA, CH2Cl2, 0–25 °C, 3 h, 27%. X = Cl, Br.
and a resulting improvement of LipE to 1.40 with cyclopropyl