Difluorocyclobutylacetylenes as positive allosteric modulators of mGluR5 with reduced bioactivation potential

Article abstract of DOI:10.1016/j.bmcl.2016.11.014

Schizophrenia is a serious illness that affects millions of patients and has been associated with N-methyl-D-aspartate receptor (NMDAR) hypofunction. It has been demonstrated that activation of metabotropic glutamate receptor 5 (mGluR5) enhances NMDA receptor function, suggesting the potential utility of mGluR5 positive allosteric modulators (PAMs) in the treatment of schizophrenia. Herein we describe the optimization of an mGluR5 PAM by replacement of a phenyl with aliphatic heterocycles and carbocycles as a strategy to reduce bioactivation in a biaryl acetylene chemotype. Replacement with a difluorocyclobutane followed by further optimization culminated in the identification of compound 32, a low fold shift PAM with reduced bioactivation potential. Compound 32 demonstrated favorable brain uptake and robust efficacy in mouse novel object recognition (NOR) at low doses.

Full text of DOI:10.1016/j.bmcl.2016.11.014

Accepted Manuscript
Difluorocyclobutylacetylenes as Positive Allosteric Modulators of mGluR5
with Reduced Bioactivation Potential
Andrew P. Degnan, Darrell Maxwell, Anand Balakrishnan, Jeffrey M. Brown,
Amy Easton, Michael Gulianello, Umesh Hanumegowda, Melissa Hill-Drzewi,
Regina Miller, Kenneth S. Santone, Arun Senapati, Eric E. Shields, Digavalli
V. Sivarao, Ryan Westphal, Valerie J. Whiterock, Xiaoliang Zhuo, Joanne J.
Bronson, John E. Macor
PII:
S0960-894X(16)31154-4
DOI:
http://dx.doi.org/10.1016/j.bmcl.2016.11.014
BMCL 24409
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
2 October 2016
5 November 2016
7 November 2016
Please cite this article as: Degnan, A.P., Maxwell, D., Balakrishnan, A., Brown, J.M., Easton, A., Gulianello, M.,
Hanumegowda, U., Hill-Drzewi, M., Miller, R., Santone, K.S., Senapati, A., Shields, E.E., Sivarao, D.V., Westphal,
R., Whiterock, V.J., Zhuo, X., Bronson, J.J., Macor, J.E., Difluorocyclobutylacetylenes as Positive Allosteric
Modulators of mGluR5 with Reduced Bioactivation Potential, Bioorganic & Medicinal Chemistry Letters (2016),
doi: http://dx.doi.org/10.1016/j.bmcl.2016.11.014
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Difluorocyclobutylacetylenes as Positive Allosteric Modulators of
mGluR5 with Reduced Bioactivation Potential
Andrew P. Degnan,* Darrell Maxwell, Anand Balakrishnan, Jeffrey M. Brown, Amy
Easton, Michael Gulianello, Umesh Hanumegowda, Melissa Hill-Drzewi, Regina Miller,
Kenneth S. Santone, Arun Senapati, Eric E. Shields, Digavalli V. Sivarao, Ryan
Westphal, Valerie J. Whiterock, Xiaoliang Zhuo, Joanne J. Bronson, and John E. Macor
Bristol-Myers Squibb Co., Research and Development, 5 Research Parkway, Wallingford, CT 06492, United States
Abstract: Schizophrenia is a serious illness that affects millions of patients and has been associated
with N-methyl-D-aspartate receptor (NMDAR) hypofunction. It has been demonstrated that activation
of metabotropic glutamate receptor 5 (mGluR5) enhances NMDA receptor function, suggesting the
potential utility of mGluR5 positive allosteric modulators (PAMs) in the treatment of schizophrenia.
Herein we describe the optimization of an mGluR5 PAM by replacement of a phenyl with aliphatic
heterocycles and carbocycles as a strategy to reduce bioactivation in a biaryl acetylene chemotype.
Replacement with a difluorocyclobutane followed by further optimization culminated in the
identification of compound 32, a low fold shift PAM with reduced bioactivation potential. Compound
32 demonstrated favorable brain uptake and robust efficacy in mouse novel object recognition (NOR) at
low doses.
Schizophrenia is a debilitating illness which is thought to afflict approximately 1% of the world
population.1 The symptoms are categorized into three domains – positive (hallucinations, delusions),
negative (apathy, anhedonia), and cognitive (reductions in executive function and memory).2 First
discovered in the 1950s, the so-called first generation antipsychotics (e.g., haloperidol and
chlorpromazine) revolutionized the treatment of the positive symptom domains of schizophrenia, but
were associated with significant motor side effects. While more recently identified second-generation
antipsychotics are associated with reduced motor side effects, they too show limited utility with regard
to the treatment of the negative and cognitive symptom domains.
Metabotropic glutamate receptors (mGluRs) are a family of GPCRs that are activated by binding to
endogenous glutamate.3 The family is comprised of 8 receptors (mGluR1-8) that are further divided into
three groups (Groups I-III) based upon their sequence homology, agonist selectivity, and signal
transduction mechanism.4 Metabotropic glutamate receptor 5 (mGluR5) is a member of group I and is
activated by binding of glutamate to its large extracellular glutamate binding domain. The signal is
propagated into the neuron through coupling of Gq with phospholipase C, which gives rise to
phosphoinositide hydrolysis, elevation of intracellular calcium levels, and enhanced N-methyl-D-
aspartate receptor (NMDAR) function.5 NMDAR hypofunction has long been thought to play a role in

schizophrenia.6 One of the earliest pieces of data that supported this hypothesis was the observation that
NMDAR antagonists such as ketamine and phencyclidine recapitulated the positive, negative, and
cognitive symptoms of schizophrenia, which suggested the potential utility of NMDAR agonists to treat
all three symptom domains. Unfortunately, direct activation of the NMDA receptor leads to increased
neuronal calcium levels, which precipitates an excitotoxic cascade that leads to neuronal death. The
identification of positive allosteric modulators (PAMs) of mGluR5 thus offers an exciting ‘backdoor’
approach to NMDAR potentiation for the treatment of schizophrenia.7 Since PAMs sensitize the
receptor to signaling only in the presence of the orthosteric agonist, an mGluR5 PAM should offer the
opportunity to enhance NMDAR function in a way which maintains the spatial and temporal
components of glutamate signaling.
In an earlier paper, we disclosed our development candidate BMS-955829 (1, Scheme 1).8 Compound 1
had a number of desirable properties to support its selection. It was found to be a potent mGluR5 PAM,
possessed a very low fold shift (2.3), and was devoid of inherent agonist activity. This combination of
low fold shift and lack of agonist activity proved critical in identifying a compound with low potential
for seizures and neurotoxicity. Additionally, 1 had an excellent pre-clinical in vitro safety profile and
demonstrated acceptable safety margins in both 2-week rat and 5-day dog toxicity studies.
Unfortunately, during the 1-month pre-IND dog toxicology study, drug-induced liver injury was
observed at all doses (3 - 30 mg/kg QD). The hepatotoxicity manifested as increased serum levels of
alanine aminotransferase (ALT) and intrahepatic cholestasis with severity that increased in a dose-
dependent fashion. Metabolic profiling of 1 in liver microsomes indicated formation of dihydrodiol
metabolites and glutathione conjugates consistent with epoxidation of the unsubstituted phenyl, followed
by ring opening by water or glutathione (Scheme 1). While the site of dihydrodiol formation could not
be definitively localized to the phenyl group,9 it seemed the most likely site of metabolism based upon
its lipophilicity and electron density relative to the pyridine. It was reasoned that formation of such a
reactive metabolite in the liver could be the root cause of the dog liver toxicity, and it was against this
backdrop that we sought to identify a phenyl replacement that could reduce bioactivation relative to
compound 1.
Scheme 1. Proposed bioactivation pathway of compound 1 in rat liver S9 fractions.

We had previously explored the possibility of replacement of the phenyl group with heteroaromatics to
decrease lipophilicity.10 However, it was found that alkynes bearing heterocycles at both termini were
prone to NADPH-independent glutathione addition.11 Consequently, we turned our effort to the
identification of saturated phenyl replacements. We were inspired in this endeavor by the work of
Lovering and coworkers,12 whose careful analysis suggested that increasing the fraction of sp3 carbons
(Fsp3) was positively correlated with subsequent success in the clinic. To this end we prepared a small
library of phenyl replacements by methods that are directly analogous to those used to prepare 1.8 In an
earlier report,10 we demonstrated that fold shift was exquisitely sensitive to changes in the substitution
pattern of the difluorophenyl moiety. Indeed, compound 2 (Table 1), differing by a single fluorine, had
greatly increased in vitro efficacy (fold shift = 8.7). While our desired in vitro profile included a low
fold shift ( 3, vide supra), we opted to use compound 2 (rather than compound 1) as the starting point
for optimization. By starting with a high fold shift compound, we reduced the likelihood of losing PAM
functional activity due to minor perturbations in fold shift imparted by the newly installed aliphatic
substituent. The first compound prepared in our aliphatic series was 3. While substantially less potent
than 2, 3 retained useful levels of mGluR5 PAM activity and maintained binding affinity that was in
good agreement with the functional assay. A brief survey of alternate ring sizes (3-6) revealed
cyclobutane 5 to be the most potent (PAM EC50 = 8 nM), but microsomal stability was greatly
diminished. We also looked for opportunities to decrease lipophilicity by incorporating oxygen into the
ring (7-10). Unfortunately, all ethers showed unacceptable losses in potency.
Table 1
SAR studies: replacement of phenyl with aliphatic ringsa
mGluR5
mGluR5
binding
HLM/MsLM
(%remain) d
R
Compound
PAM EC50
(nM, (%Ymax)) b (Ki, nM) c
1.1 (134)
1.6
76/71
2
3
49 (200)
25 (152)
47
17
67/12
52/16
4

5
6
8 (38)
15
95
32/13
81/69
90 (42)
7
8
77 (155)
>3200
372
259
90/52
93/63
9
237 (67)
>3200
323
60/41
10
1667
100/98
amGluR5 functional and binding assays were performed by methods described in reference 8. bAll values
represent n≥2 replicates. cDisplacement of [3H]-MPEPy from human mGluR5. dPercentage remaining after 10
min incubation with human or mouse liver microsomes (HLM or MsLM).
Table 2.
SAR studies: impact of cyclobutane substitution on in vitro functional activity and oxidative stabilitya
mGluR5 PAM
EC50 (nM,
mGluR5
binding
(Ki, nM)c
HLM/MsLM
(%remain)d
R
Compound
(Ymax))b
5
8 (38)
15
64
32/13
80/74
11
31 (68)
12
13 (44)
30
83/57

13
14
15
16
17
18
>3200
>1600
38
5000
580
81/37
71/41
10/18
100/51
100/74
88/78
>3200
250 (58)
>1600
710
900
330 (114)
1900
19
300 (122)
490
95/81
amGluR5 functional and binding assays were performed by methods described in reference 8. bAll values
represent n≥2 replicates. cDisplacement of [3H]-MPEPy from human mGluR5. dPercentage remaining after 10
min incubation with human or mouse liver microsomes (HLM or MsLM).
Hypothesizing that the distal carbon of the cyclobutane was the most likely site of oxidative metabolism,
we focused our synthetic efforts on this position (Table 2). While all substituted cyclobutane analogs
proved to be less potent than the unsubstituted variant, substitution of the distal carbon did give
substantial improvements in microsomal stability in almost every case. We were most intrigued by the
profile of fluorocyclobutanes 11 and 12. Both showed useful levels of PAM potency, binding affinity
that was in good agreement with their functional potency, and substantial improvements to metabolic
stability. It is interesting to note that 13, the diastereomer of 12, was a silent allosteric modulator
(SAM), retaining nearly identical binding affinity, while lacking functional PAM activity. Ultimately,
the decision was made to advance 11 due to its increased in vitro PAM efficacy (Ymax = 68%) and
mouse microsomal stability (74% remaining after 10 min microsomal incubation) relative to 12.
Having replaced the phenyl group with a moiety that cannot undergo the same type of bioactivation
pathway, 11 was evaluated in an in vitro bioactivation study using an aroclor-induced rat liver S9
preparation. Unfortunately, 11 still showed substantial levels of bioactivation (~22%, Table 4).
However, analysis of the mass fragmentation pattern indicated that the primary site of bioactivation had
shifted from the biaryl acetylene of 1 to the fluorophenyl of 11. Having demonstrated a reduced
propensity of the difluorocyclobutane moiety toward bioactivation, we initiated the synthesis of a small

library of analogs in which substitution of the fluorophenyl ring was varied. Scheme 2 is representative
of the synthetic route used to prepare this series of compounds. (3,3-Difluorocyclobutyl)methanol (20)
was oxidized using Dess-Martin periodinane to yield aldehyde 21. The aldehyde was converted to the
corresponding terminal alkyne (22) by treatment with the Ohira-Bestmann reagent in the presence of
potassium carbonate. Both 21 and 22 were highly volatile, requiring isolation using careful distillation.
Fragments 22 and 23 were efficiently united in a Sonagashira reaction to afford 11.
Scheme 2. Reagents and conditions: a) Dess-Martin periodinane, CH2Cl2, 47%; b) Ohira-Bestmann
reagent, potassium carbonate, 0  20 oC, 49%; c) PdCl2(PPh3)2, PPh3, CuI, Et3N, 80 oC, 77%.
Table 3.
SAR studies: optimization of phenyl substitutiona
mGluR5 PAM
EC50 (nM,
mGluR5
binding
(Ki, nM)c
Fold
Shift
HLM/MsLM
(%remain)d
Compound
R
(Ymax))b
11
24
25
26
27
28
31 (68)
66 (118)
>3200
2.2
2.8
NA
5.9
10
64
30
80/74
85/76
84/70
79/59
60/23
NA/NA
3-F
3-Cl
47
3-OMe
4-F
35 (145)
18 (133)
>3200
88
200
67
4-Cl
NA
2,5-difluoro

29
30
31
>3200
69 (75)
60 (28)
NA
2.6
1.6
108
59
86/89
100/84
95/100
3,5-difluoro
2,4-difluoro
2-Cl, 4-F
36
130 (60)
(n=4)
1.3
13
32
2,3-dichloro
93/67
(n=2)
(n=2)
amGluR5 functional and binding assays were performed by methods described in reference 8. bAll values
represent n≥2 replicates. cDisplacement of [3H]-MPEPy from human mGluR5. dPercentage remaining after 10
min incubation with human or mouse liver microsomes (HLM or MsLM).
Whereas replacement of the fluorine of 11 with a chlorine (24, Table 3) gave a compound with a nearly
identical in vitro profile, replacement with a methoxy group (25) eliminated functional activity (SAM).
Given the attractive profile of 24, it too was evaluated in an aroclor-induced rat liver S9 preparation
(Table 4). Again, substantial bioactivation of the halophenyl was observed. Shifting the halide from the
meta- to the para- position (as in 26 and 27) gave an unacceptable increase in in vitro PAM efficacy
(Ymax, fold shift). Di-halo-substituted phenyls were also examined. As we had anticipated,
combination of the low fold shift difluorocyclobutane with the low fold shift 2,5-difluorophenyl of 1
gave a silent allosteric modulator (28). This result suggested additive effects of component fragments
on in vitro efficacy, and validated our strategy of initiating work in this series with a high fold shift basis
structure (2) to increase the range of phenyl replacements capable of delivering functionally-active
PAMs. While none of the dihalophenyl derivatives met our in vitro PAM potency (EC50) and efficacy
(fold shift) targets, we were intrigued by dichloro derivative 32. While the measured in vitro potency
was substantially higher than our targeted potency, 32 showed increased affinity in the binding assay
relative to other derivatives in the series. While there are many reports of mGluR5 PAMs that show
increased potency in functional assays relative to their binding affinity due to effects of allosteric
cooperativity,13 we were not aware of any examples with substantially higher binding affinity than
functional PAM potency. We became even more interested in this particular derivative when it proved
to be much more resistant to bioactivation relative to other derivatives examined (<4% total
bioactivation, Table 4). While its measured fold shift (1.3) was somewhat lower than our target profile,
the measured Ymax (60%) was more consistent with derivatives in the desired range (fold shift ~ 2-3).
Confounded by the binding and in vitro PAM potency and efficacy data, it was ultimately decided that
in vivo efficacy testing of 32 was required to understand the translation of its in vitro profile and to drive
subsequent compound progression decisions.
The novel object recognition (NOR) assay is a validated measure of episodic recognition that relies on a
rodent’s instinctive proclivity to explore novel objects at the expense of familiar ones.14 On day 1,
compound 32 was administered intraperitoneally to mice 60 minutes prior to training at doses ranging
from 0.1 – 3 mg/kg. On the second day, one of the familiar objects was replaced with a novel object and
time spent with each object was recorded. The results are illustrated in Figure 1. Mice dosed with 1 or
3 mg/kg of 32 showed significant reductions in time spent with the familiar object, indicating improved

recognition memory. Plasma concentrations were obtained using satellite animals and indicated total
plasma levels of 510 ± 70 nM ([brain] = 860 ± 120 nM) at the minimum efficacious dose (MED = 1
mg/kg). In order to assess the capacity of 32 to induce seizures, C57BL/6 mice were treated with doses
ranging from 1 – 30 mg/kg ip. Exposures were approximately dose proportional, achieving maximum
exposure at 30 mg/kg ([plasma] = 14900 ± 4000 nM; [brain] = 14100 ± 6900 nM). There were no
adverse events and all doses were well tolerated, suggesting a low potential for seizure induction. This
was in contrast with high fold shift PAMs such as 2, which induced seizures at 3 mg/kg.8,10
Table 4.
Aroclor-induced rat liver S9 incubations: extent and site of bioactivation
GSH
Conjugation
Dihydrodiol
via
Compound
Site of Bioactivation
Fsp3
oxidation
(%)
(%)
1
25
22
17
3
8
0.091
11
24
32
<1
<1
<1
0.30
0.30
0.30

Figure 1. In vivo efficacy: activity of 32 in mouse novel object recognition.
In summary, a phenyl replacement strategy was pursued in order to reduce bioactivation observed with
low fold shift mGluR5 PAM 1. These efforts led to the identification of the difluorocyclobutane as an
acceptable phenyl replacement with decreased propensity for bioactivation and decreased planarity
(fraction of sp3 carbon atoms (Fsp3) = 0.30). Further refinement of these leads led to discovery of 32, a
compound with excellent efficacy in a rodent model of cognition, a low potential for seizure induction,
and reduced propensity for bioactivation in vitro. Additional characterization of 32 will be disclosed in
due course.
Acknowledgements
We thank the Department of Chemical Synthesis for the scale-up of intermediate 23. We also thank Spectrix
and Discovery Analytical Sciences for final compound purification and characterizations.
Supplementary Material
Supplementary data (detailed experimental procedures for the preparation of 32, characterization data for
compounds 1–19 and 24-31) associated with this article can be found, in the online version, at
http://dx.doi.org/10.1016/j.bmcl.2016.xxx.
References and Notes
1. Perälä, J.; Suvisaari, J.; Saarni, S.I.; Kuoppasalmi, K.; Isometsä, E.; Pirkola, S.;Partonen, T.;
Tuulio-Henriksson, A.; Hintikka, J.; Kieseppä, T.; Härkänen, T.; Koskinen, S; Lönnqvist, J.
Arch. Gen. Psychiatry, 2007, 64, 19.

2. Van Os, J.; Kapur, S. Lancet, 2009, 374, 635.
3. Conn, P.J.; Pin, J.-P. Annu. Rev. Pharmacol. Toxicol. 1997, 37, 205.
4. Vanejevs, M.; Jatzke, C.; Renner, S.; Müller, S; Hechenberger, M.; Bauer, T.; Klochkova, A.;
Pyatkin, I.; Kazyulkin, D.; Aksenova, E.; Shulepin, S.; Timonina, O.; Haasis, A.; Gutcaits, A.;
Parsons, C.G.; Kauss, V.; Weill, T. J. Med. Chem. 2008, 51, 634.
5. Aniksztejn, L; Bregestovski, P.; Ben-Ari, Y. Eur. J. Pharmacol. 1991, 205, 327.
6. Coyle, J.T. Schizophrenia Bull. 2012, 38, 920.
7. Newell, K.A. Future Med. Chem. 2013, 5, 1471.
8. Yang, F.; Snyder, L.B.; Balakrishnan, A.; Brown, J.M.; Sivarao, D.V.; Easton, A.; Fernandes,
A.; Gulianello, M.; Hanumegowda, U.M.; Huang, H.; Huang, Y.; Jones, K.M.; Li, Y.-W.;
Matchett, M.; Mattson, G.; Miller, R.; Santone, K.S.; Senapati, A.; Shields, E.E.; Simutis, F.J.;
Westphal, R.; Whiterock, V.J.; Bronson, J.J.; Macor, J.E.; Degnan, A.P. ACS Med. Chem. Lett.
2016, 7, 289.
9. MS fragmentation pattern indicated bioactivation products were derived from reaction of the
diarylacetylene moiety. See table 4, entry 1.
10. Huang, H.; Degnan, A. P.; Balakrishnan, A.; Easton, A.; Gulianello, M.; Huang, Y.; Matchett,
M.; Mattson, G.; Miller, R.; Santone, K. S.; Senapati, A.; Shields, E. E.; Sivarao, D. V.; Snyder,
L. B.; Westphal, R.; Whiterock, V. J.; Yang, F.; Bronson, J. J.; Macor, J. E. Bioorg. Med. Chem.
Lett. 2016, 26, 4165.
11. Zhuo, X.; Huang, X.S.; Degnan, A.P.; Snyder, L.B.; Yang, F.; Huang, H.; Shu, Y.Z.; Johnson,
B.M. Drug Metab. Dispos. 2015, 43, 578.
12. Lovering, F.; Bikker, J.; Humblet, C. J. Med. Chem. 2009, 52, 6752.
13. Gregory, K.J., Noetzel, M.J.; Rook, J.M.; Vinson, P.N.; Stauffer, S.R.; Rodriguez, A.L.;
Emmitte, K.A.; Zhou, Y.; Chun, A.C.; Felts, A.S.; Chauder, B.A.; Lindsley, C.W.; Niswender,
C.M.; Conn, P.J. Mol. Pharmacol. 2012, 82, 860.
14. Monchi, O.; Petrides, M.; Petre, V.; Worsley, K.; Dagher, A. Wisconsin card sorting revisited:
Distinct neural circuits participating in different stages of the task identified by event-related
functional magnetic resonance imaging. J. Neurosci. 2001, 21, 7733.

Graphical abstract