An orally available, brain-penetrant CAMKK2 inhibitor reduces food intake in rodent model

Article abstract of DOI:10.1016/j.bmcl.2018.03.034

Hypothalamic CAMKK2 represents a potential mechanism for chemically affecting satiety and promoting weight loss in clinically obese patients. Single-digit nanomolar inhibitors of CAMKK2 were identified in three related ATP-competitive series. Limited optimization of kinase selectivity, solubility, and pharmacokinetic properties were undertaken on all three series, as SAR was often transferrable. Ultimately, a 2,4-diaryl 7-azaindole was optimized to afford a tool molecule that potently inhibits AMPK phosphorylation in a hypothalamus-derived cell line, is orally bioavailable, and crosses the blood–brain barrier. When dosed orally in rodents, compound 4 t limited ghrelin-induced food intake.

Full text of DOI:10.1016/j.bmcl.2018.03.034

Accepted Manuscript
An orally available, brain-penetrant CAMKK2 inhibitor reduces food intake in
rodent model
Daniel J. Price, David H. Drewry, Lee T. Schaller, Brian D. Thompson, Paul R.
Reid, Patrick R. Maloney, Xi Liang, Periette Banker, Richard G. Buckholz,
Paula K. Selley, Octerloney B. McDonald, Jeffery L. Smith, Todd W. Shearer,
Richard Cox, Shawn P. Williams, Robert A. Reid, Stefano Tacconi, Federico
Faggioni, Chiara Piubelli, Ilaria Sartori, Michela Tessari, Tony Y. Wang
PII:
S0960-894X(18)30223-3
https://doi.org/10.1016/j.bmcl.2018.03.034
BMCL 25688
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
11 December 2017
12 March 2018
13 March 2018
Please cite this article as: Price, D.J., Drewry, D.H., Schaller, L.T., Thompson, B.D., Reid, P.R., Maloney, P.R.,
Liang, X., Banker, P., Buckholz, R.G., Selley, P.K., McDonald, O.B., Smith, J.L., Shearer, T.W., Cox, R., Williams,
S.P., Reid, R.A., Tacconi, S., Faggioni, F., Piubelli, C., Sartori, I., Tessari, M., Wang, T.Y., An orally available,
brain-penetrant CAMKK2 inhibitor reduces food intake in rodent model, Bioorganic & Medicinal Chemistry
Letters (2018), doi: https://doi.org/10.1016/j.bmcl.2018.03.034
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.

Bioorganic & Medicinal Chemistry Letters
An orally available, brain-penetrant CAMKK2 inhibitor reduces food intake in
rodent model
Daniel J. Price a,*, David H. Drewryb, Lee T. Schallera, Brian D. Thompsonb, Paul R. Reidb, Patrick R.
Maloneyb, Xi Liangb, Periette Bankera, Richard G. Buckholza, Paula K. Selleyb, Octerloney B. McDonaldb,
Jeffery L. Smithb, Todd W. Shearerb, Richard Coxb, Shawn P. Williams a, Robert A. Reid a, Stefano
Tacconib, Federico Faggionib, Chiara Piubellib, Ilaria Sartorib, Michela Tessarib, and Tony Y. Wangb
aGlaxoSmithKline, 5 Moore Drive, Research Triangle Park, NC 27709, USA
bAll research described was performed as employees of GlaxoSmithKline. Current affiliations supplied in footnote 25.
ARTICLE INFO
ABSTRACT
Article history:
Received
Revised
Accepted
Available online
Hypothalamic CAMKK2 represents a potential mechanism for chemically affecting satiety and
promoting weight loss in clinically obese patients. Single-digit nanomolar inhibitors of
CAMKK2 were identified in three related ATP-competitive series. Limited optimization of
kinase selectivity, solubility, and pharmacokinetic properties were undertaken on all three series,
as SAR was often transferrable. Ultimately, a 2,4-diaryl 7-azaindole was optimized to afford a
tool molecule that potently inhibits AMPK phosphorylation in a hypothalamus-derived cell line,
is orally bioavailable, and crosses the blood-brain barrier. When dosed orally in rodents,
compound 4t limited ghrelin-induced food intake.
Keywords:
CAMKK2
Inhibitor
Satiety
2009 Elsevier Ltd. All rights reserved.
The detrimental implications of obesity on human health have
been well documented,1 and novel mechanisms for
therapeutically promoting weight loss would have significant
medical and social benefits. AMPK is a central integrator of
hormone and energy sensing in the hypothalamus, controlling
satiety via NPY expression. Central AMPK inhibition reduces
NPY-mediated feeding, however, peripheral inhibition of AMPK
promotes an undesirable anabolic phenotype for obese patients:
increased lipogenesis, cholesterol and glycogen synthesis,
gluconeogenesis, and a concomitant decrease in processes such
as fatty acid oxidation and serum glucose uptake.2 Intriguingly,
LKB1 is recognized as the predominant mediator of AMPK
phosphorylation in response to peripheral energy demand,3-4
whereas CAMKK2 has been identified as the hypothalamic
AMPK kinase that transduces Ca++-mediated ghrelin signaling,5
thus suggesting a mechanism to selectively inhibit hypothalamic
AMPK and NPY’s downstream orexigenic effect.
CAMKK2’s role in other processes and tissues is still being
elucidated, however. Whole and tissue-specific knock-out of
CAMKK2 have implicated it in glycemic control through a liver-
specifc, AMPK-independent mechanism,6 bone density via
osteoblast/osteoclast ratio,7 and long-term memory formation.8
We set out to identify an orally available, brain-penetrant
CAMKK2 inhibitor to probe the holistic impact of inhibiting
CAMKK2 with a therapy-relevant modality, with an immediate
goal of further substantiating the therapeutic hypothesis of
acutely reducing feeding.
Indeed, CAMKK2-null mice show a sustained reduction in
feeding on a high-fat diet and have resultingly lower body
weights, reduced adiposity, and improved glucose sensitivity
relative to their WT littermates.5 From a mechanistic standpoint,
CAMKK2-null mice are acutely resistant to ghrelin-induced food
intake, and eat less than their WT counterparts upon refeeding
after fasting, similar to NPY-depleted mice.5 These latter
observations are corroborated by pharmacological inhibition of
CAMKK2 using STO-609 delivered via i.c.v. administration.5
Figure 1. Three hits identified in focused screen.
A
fluorescent polarization-based assay of full length
CAMKK2 was established for primary screening.9 A screen of
~12K compounds focused on known ATP-competitive kinase

inhibitors and compounds with pharmacophoric features similar
to known kinase inhibitors identified several potent starting
points for a chemistry campaign (Figure 1). Potent inhibitors
were subsequently advanced to an ELISA-based assay
monitoring AMPK phosphorylation in the catalytic alpha subunit
at Thr172 in the mouse neuronal N39 cell line to measure
CAMKK2 activity in a hypothalamus-relevant cell model.10
Biochemical ortholog assays were deemed unnecessary based on
sequence identity.11
unsubstituted benzoic acid (1f) is roughly equipotent on
CAMKK2 and CK2 whereas the cyclopentyl benzoic acid (1a)
displays 1000x selectivity for CAMKK2 over CK2 (CK2 data
not shown). Unfortunately, the acids had poor cellular activity
(IC50 >1 M), perhaps reflecting poor cellular penetration.
Matched pairs from the initial screen suggested several acid
replacements (alcohol, methyl ether, ethyl sulfone, and N-linked
methyl sulfonamide) are poorly tolerated. Homologation of the
acid by one or two carbons is detrimental as well (data not
shown). The benzamide was confirmed as near equipotent with
the unsubstituted benzoic acid (1i-1l) as is the acid isostere
tetrazole 1g, whereas the methyl ester 1h showed reduced
potency. Unfortunately, substitution of an alkane adjacent to the
benzamide did not boost potency like with the acids (e.g. 1j vs
1p).
Our leads suggested high CAMKK2 potency is achievable,
though several properties would need to be optimized. Each of
the leads has been screened against 48-86 other kinases in dose
response, and is active on 15-30% of those with an IC50 <100
nM. Selectivity for CAMKK2 therefore required optimization,
which is perhaps not surprising as each template largely
conforms to a published generic pharmacophore for kinase
inhibition.12 As a correlation has been established relating
increasing aromaticity to poor solubility,13 we aimed to monitor
and optimize solubility using a high-throughput assay.14 Finally,
the properties of brain-penetrant molecules have also been well
described,15 and while some calculated properties of the 3 leads
(e.g. 66 ≤ tpsa ≤ 75, 367 ≤ mw ≤ 382, # hbond donors ≤ 2)16 are
similar to those seen in retrospective analyses of CNS-active
drugs, the benzoic acid of 1a and 3a is likely incompatible with
CNS exposure.
Table 1. 3,5-diaryl 7-azaindoles.
R1
R2
R3
Sola
(M)
150
333
>439
178
100
82
CAMKK2
pIC50
9.2
P-AMPK
pIC50
5.9
We expected from initial modeling that the 2-amino
pyrimidine of 3a and 7-azaindoles of 1a and 4a would bind the
so-called kinase hinge, albeit the latter two doing so in an
opposite orientation relative to each other. Furthermore, it was
anticipated that the benzoic acid of 1a and 3a and benzamide of
4a would interact directly with the catalytic lysine. The
pharmacophoric similarity of the molecules suggested that the
SAR would be transferable across series, and in particular that
the benzamide would be a suitable replacement for the benzoic
acid. We therefore pursued concerted optimization of the three
series together.
1a COOH
1b COOH
1c COOH
1d COOH
1e COOH
cyclopentyl
isobutyl
isopropyl
CH2CH3
CH3
H
H
H
H
H
H
8.3
8.4
5.9
7.7
7.3
1f
COOH
H
6.7
1g tetrazole
1h CO2Me
H
H
H
H
276
5
6.4
5.9
1i
1j
COOH
H
H
H
H
o-OPh
24
9
7.3
7.0
6.9
6.7
CONH2
o-OPh
1k COOH
1l CONH2
m-SO2CH3
177
m-SO2CH3 >115
1m CONH2 cyclopentyl o-OCH3
1n CONH2 cyclopentyl o-OH
1o CONH2 cyclopentyl o-OisoPr
1p CONH2 cyclopentyl o-OPh
1q CONH2 cyclopentyl o-OCF3
1r CONH2 cyclopentyl o-N(CH3)2
1s CONH2 cyclopentyl o-CH2CN
1
6
7.3
7.3
6.9
6.6
6.5
6.3
6.1
5.9
<5
5.4
<5
<5
<5
<5
Scheme 1. Reagents and conditions: (a) 1 eq Ar1B(OH)2, 0.064 eq
Pd(dppf)Cl2, 5.24 eq Na2CO3 in 1,4-dioxane, reflux, 105 °C, 3 days, (b) 2 eq
Ar2B(OH)2, 4.68 eq Na2CO3, 0.115 eq PdCl2 in 1,4-dioxane, 15 hrs, 100 °C,
(c) 3 eq K2CO3 in MeOH, 50 °C, 65 hrs.
52
2
2
Synthesis of 3,5-diaryl 7-azaindoles from commercially
available starting materials has been previously described17;
related conditions and reagents used here afforded rapid SAR
expansion (Scheme 1). A strong dependency for larger alkanes
ortho- to the benzoic acid was quickly identified (1a-1f in Table
1), a feature we hoped would benefit the other templates. While
placing a greasy group under the G-rich loop might appear to be
a promiscuous feature and increase potency on all kinases, we
found that larger substituents at this position and/or the induced
rotation of the acid were neutral or detrimental to the activity
96
12
1t
CONH2 cyclopentyl o-CF3
a Solubility
A 2.4-Å resolution crystal structure was obtained of the kinase
domain of CAMKK2 bound to 1b using a construct with a
deletion in the N-terminal domain necessary for crystallization
(149-465, 205-220).18 The assumed binding mode for the series
was confirmed (Figure 2), including direct interaction between
against many other kinases screened.
For example, the
the ligand carboxylic acid and the catalytic lysine, Lys194.
A

structured water molecule coordinated between the ligand,
Lys194 and Asp330 may explain some initial SAR, as modeling
suggests it may become frustrated in the presence of the ortho-
substitued benzamide. The large gatekeeper residue Phe267
precludes penetration deeper into the so-called inner hydrophobic
area. However, Leu269 at the edge of the hinge does provide an
opportunity to improve selectivity. Close van der Waals contacts
between the ligand and this residue can incur a steric bump and
loss of potency against the >60% of kinases that boast a large
residue (Phe, Tyr, Trp) at this position, a strategy that has been
exploited elsewhere.19
and selective CAMKK2 inhibitor. Once again however, the
benzoic acids showed poor cellular potency (2b-2d, 2f) and the
potency benefit of alkyl substitution is incompatible with the
terminal amide (2h-2j).
We turned our attention to the pyrimidine series, hopeful that
we could borrow SAR at several positions. Synthesis of 2-
anilino, 4-aryl pyrimidines has also been described,21 and
conditions used here represent only minor deviations (Scheme 3).
As expected, the unhindered benzoic acid and benzamide are
again equipotent (3a-3b in Table 3). It is interesting that some
kinases do not tolerate the acid as a lysine contact, perhaps
because of a clash with the DFG aspartate, conserved glutamate
on the C-helix, or intervening waters. For example, the amide is
~1000x more potent on CLK2 than the acid (data not shown).
This observation serves as a reminder that profound selectivity
between kinases can be obtained targeting perfectly conserved
features.
1
eq Ar1B(OH)2, 3eq
Figure 2. Crystal structure of CAMKK2 kinase domain deletion mutant
bound to 1b. Hydrogen bonds to ligand of ≤3 Å are shown. The N-terminal
domain has been clipped and only a partial molecular surface is shown for
clarity.20
Scheme 3. Reagents and conditions: (a)
K3PO4.3H2O, 0.03 eq Pd(PPh3)4 in 4:1 1,4-dioxane: water, 80 °C, overnight,
(b) 1 eq Ar2NH2, 1.2 eq K2CO3, 0.15 eq Pd2(dba)3, 0.15 eq BINAP in 1:1 1,4-
dioxane:EtOH, 100 °C, overnight.
Substitution at the ortho-position on the terminal phenyl
could theoretically access the space near Leu269. However,
despite modest potency improvements with small substitution
independent of polarity (1m-1t), selectivity improvements were
modest to nonexistent. We presumed that rotation about the
biaryl bond allows the ortho-substituent to orient away from the
hinge upon binding, thereby negating the intended effect on
selectivity.
We attempted to implement our aforementioned selectivity
strategy of targeting close contact with Leu269 by substituting at
the ortho-position on the aniline (3c-3e). Unlike the 7-azaindole,
bulk ortho- to the amine would sterically limit rotation of the
aniline torsion and thereby maximize the desired effect on
selectivity. Comparison of 3b and 3e, for example, reveals how
intolerant some kinases are of this substitution pattern (potencies
on GSK3, JNK1, and PI3K are shown as examples).
Table 2. 3,6-disubstituted, 7-azaindoles.
R1
R2
R3
Sola
CAMKK2
pIC50
5.7
P-AMPK
pIC50
(M)
2a COOH
2b COOH
2c COOH
2d COOH
2e COOH
H
H
H
cyclopentyl
cyclopentyl
isopropyl
H
>290
155
8.5
5.4
5.2
5.2
CN
7.8
Scheme 2. Reagents and conditions: (a) 1.2 eq mCPBA, 2.5 eq Et3N, 6 eq
TMSCN in DCM under N2, RT, 1 hr (b) 1 eq ICl in DCM, 20 °C, 1 hr, (c) 1.2
eq pTsCl, 0.02 eq Bu4NBr, in 20:1 DCM:NaOH, 20°C, 1 hr, (d) 1.2 eq
ArB(OH)2, 0.05 eq Pd(dppf)Cl2, 3 eq K2CO3 in 2.5:1 1,4-dioxane:water, 90
°C, 3 hr, (e) 1 eq NaOH in MeOH, 50 °C for 30 min (R=-CN) or 19 hrs (R=-
CONH2).
CN
>172
235
7.5
CN
5.6
2f
COOH
cyclopentyl
cyclopentyl
H
CH3
CONH2
H
>279
>205
7.5
<5
2g COOH
2h CONH2
5.9
We therefore attempted to more directly access the space near
Leu269 from the 6-position of the 7-azaindole, forgoing
substitution at the 5-position (Table 2). A representative
synthetic route is provided in Scheme 2. Our selectivity goals
were realized with this substitution pattern: compound 2c is
inactive on all other kinases screened (>40 kinases with IC50 >10
M, not shown), and as such, represents a remarkably compact
6.4
2i
2j
CONH2 cyclopentyl
CONH2 isopropyl
CN
7
5.9
CN
14
5.5
a Solubility

The abundance of aromatic rings makes it no surprise that
solubility is a major developability hurdle for this series; not even
the hydrophilic carboxylic acid can completely rescue the series’
inherent poor solubility (3a). Truncation of the terminal phenyl
resulted in a 10x loss in potency (3f), and did nothing to improve
the solubility of the benzamide-containing analogs. Like the 7-
azaindole series, the alkyl substitution adjacent to the amide did
not significantly help potency, though it did restore some
modicum of solubility (3g-3h), presumably disrupting packing in
the solid phase. With minimal room between the ligand and the
gatekeeper (Phe267), only conservative changes at the 5-position
of the pyrimidine were attempted. Indeed, the modest potency
benefit of the 5-fluoro substitution is lost with a larger halogen
(3h-3i).
Table 3. 2-anilino, 4-aryl pyrimidines.
R1
R2
R3
R4
R5
Sola
CAMKK2
pIC50
8.1
P-AMPK
pIC50
5.1
GSK3
pIC50
6.4
JNK1
pIC50
6.4
PI3K
pIC50
7.9
(M)
3a
3b
3c
3d
3e
COOH
CONH2
CONH2
CONH2
CONH2
H
H
H
H
H
H
H
F
phenyl
H
H
H
H
H
14
3
phenyl
phenyl
phenyl
phenyl
8.1
6.2
7.5
6.0
7.0
3
7.7
5.1
5.6
<5
<5
CH3
4
7.5
5.4
<4.6
<4.6
<5
<5
OCH3
7
7.9
<5
<5
5.1
3f
3g
3h
3i
CONH2
CONH2
CONH2
CONH2
COOH
COOH
COOH
H
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
H
H
H
H
H
H
H
H
H
F
<1
10
7.0
7.2
7.4
6.8
6.7
8.7
9.0
<5
<5
5.5
<5
<5
6.0
6.4
<4.6
<4.6
5.1
5.7
5.8
5.5
<5
<5
5.6
<5
isopropyl
isopropyl
isopropyl
H
11
Cl
H
H
F
5
4.7
5.2
5.4
5.8
5.2
3j
>444
>367
97
<4.6
4.8
5.3
5.5
5.5
3k
3l
isopropyl
isopropyl
5.1
aSolubility
Truncation of the terminal phenyl did benefit the solubility of the
acid containing analogs (3j-3l), and consistent with previous
SAR, the addition of a branched alkane adjacent to the acid
provided a strong potency improvement, again resulting in
extremely efficient and selective, albeit unprogressable,
CAMKK2 inhibitors (3k-3l).
dose pharmacokinetic (PK) properties of several acid-containing
analogs were measured, and our assumptions about the
limitations of these molecules were confirmed, revealing poor
absorption from oral dosing and/or limited CNS exposure from
IV dosing, with median brain concentrations typically < 10% of
that seen in plasma. On the other hand, molecules such as 3h did
show CNS penetration with a > 1:1 brain to plasma ratio,
however poor solubility precluded oral dosing.
Attempts to introduce solubilizing groups to the aniline of the
pyrimidine-benzamide series did improve solubility in some
cases where the terminal phenyl had been truncated (not shown),
however, potency remained flat and limited progression. At the
same time, our third hit (4a) had profoundly improved solubility
due to its pendant cationic amine, and also hinted that single-digit
nanomolar potencies could be achieved in the absence of a
benzoic acid moiety. Molecule 4a itself has an intrinsic
clearance near hepatic blood flow when dosed in rats
intravenously and is therefore insufficient as a tool. Screening of
existing analogs in the corporate collection (4b-4f) plus
additional chemistry efforts to expand SAR (4g-4m) revealed a
much higher potency threshold than had been seen thus far, with
potency not being exclusive to >10 M solubility (Table 4). An
established synthetic route22 formed the basis for rapid analog
creation (Scheme 4). Solubilizing groups at the solvent front are
robustly tolerated at a variety of positions, but with little effect on
potency. The combined improvements in potency and physical
Scheme 4. Reagents and conditions: (a) 1 eq Ar1B(OH)2, 9.92 eq
Na2CO3, 1.385 eq Pd(PPh3)4 in 1,4-dioxane, 120 °C, reflux, overnight, (b) 1
eq Ar2B(OH)2, 2 eq Na2CO3, 0.05 eq PdCl2 (dppf)-CH2Cl2 adduct in 1,4-
dioxane, 100 °C, 17 hrs, (c) 3 eq K2CO3, in MeOH, 50 °C, 4 hrs.
Neither series afforded truly potent inhibitors unless a
carboxylic acid was present as the catalytic lysine contact. Low-

properties within the series translated to cellular activities
surpassing 100 nM.
and moderate clearance and CNS exposure, albeit with only
modest AUC0-6hr when dosed orally (Table 5).
Again, introduction of a branched alkane adjacent to the
amide was equivocal on potency. While no head-to-head SAR
was generated with regards to central exposure in this series,
every ortho-alkyl benzamide tested had higher brain to plasma
ratios over every unsubstituted benzamide tested (data not
shown), encouraging us to continue to explore analogs with this
As we were now obtaining potencies better than 3 nM, we
began to reconsider if a loss in potency could be tolerated for
improved metabolic stability. The nitrile of 4q and 4r lost
significant potency relative to their amide analogs, 4j and 4l,
however, reintroduction of the o-isopropyl group restored much
of the activity (4s-4t). Furthermore, there appeared to be some
selectivity advantages to cyano substitution, eliminating all
substitution pattern.
Indeed, compounds such as 4o are
approaching tool compound status with potent cellular activity
Table 4. 2,4-diaryl 7-azaindoles.
R1
R2
R3
R4
Sola
(uM)
157
56
CAMKK2
pIC50
8.5
P-AMPK
pIC50
6.8
IKK1
pIC50
7.1
JNK1
pIC50
7.1
INSR
pIC50
6.5
4a
4b
4c
4d
4e
4f
CONH2
CONH2
CONH2
CONH2
CONH2
CONH2
H
H
H
H
H
H
H
H
H
H
H
H
p-CH2N(CH3)2
m-CH2N(CH3)2
8.4
6.9
7.1
6.2
6.2
p-CONH(CH2)3N(CH3)2
m-CONH(CH2)3N(CH3)2
p-NHSO2CH3
29
8.8
<5
7.1
6.9
6.8
8.6
5.4
6.9
6.2
5.5
5
8.7
6.5
7.5
7.2
6.5
m-NHSO2CH3
13
8.5
5.5
6.8
6.7
5.9
4g
4h
4i
CONH2
CONH2
CONH2
CONH2
CONH2
CONH2
CONH2
H
H
H
H
H
H
H
OCH3
CH3
CH3
CH3
CH3
CH3
CH3
H
16
18
18
2
8.5
8.8
8.1
8.8
8.7
8.8
8.7
6.1
6.7
5.7
7.5
6.6
6.8
6.8
7.0
6.9
5.5
6.8
6.9
6.7
6.8
6.5
6.2
5.2
6.6
6.7
6.8
5.8
5.9
5.4
<5
3-CH2-morpholine
4-CH2-morpholine
5-CH2-morpholine
3-O(CH2)2-morpholine
4-O(CH2)2-morpholine
5-O(CH2)2-morpholine
4j
4k
4l
17
11
17
5.5
5.9
5.4
4m
4n
4o
4p
CONH2
CONH2
CONH2
isopropyl
isopropyl
cyclopentyl
CH3
5-CH2-morpholine
4-O(CH2)2-morpholine
H
1
4
7
8.8
9.1
8.7
6.7
7.2
7.0
6.8
5.6
5.6
6.2
5.7
5.7
<5
6.4
6.1
CH3
OCH3
4q
4r
4s
4t
CN
CN
CN
CN
H
CH3
CH3
CH3
CH3
5-CH2-morpholine
2
7.1
7
5.6
5.1
6.4
6.3
<4.6
<4.5
5.6
<5
4.7
5.1
<5
<5
<5
<5
<5
H
4-O(CH2)2-morpholine
5-CH2-morpholine
9
isopropyl
isopropyl
1
7.9
8.2
4-O(CH2)2-morpholine
<1
<4.6
aSolubility
measureable activity on some kinases, such as INSR, Aurora A and DDR2 (IC50 > 10 M), and significantly reducing potency on
others, such as IKK1 and JNK1 (Table 4). Despite low aqueous solubility, molecule 4t could be formulated for oral dosing and
achieved our pharmacokinetic goals of reduced clearance, improved brain exposure (near 1:1 brain:plasma), and increased AUC0-6hr
(nearly 8-fold) relative to its benzamide counterpart 4o (Table 5).
We advanced 4t into an acute efficacy study. Sprague Dawley rats were pretreated with a 30 mg/kg p.o. dose of 4t or vehicle
alone 2 hours before food was offered.24 An hour after compound/vehicle dosing, animals were dosed i.c.v. with either 1 M
ghrelin or 3 L saline. Food intake was monitored for 4 hours. Figure 3 shows the expected dramatic increase in cumulative food
intake after 4 hours when ghrelin was administered in the absence of the CAMKK2 inhibitor (P = 1E-8), and a statistically
significant 40% reduction of ghrelin-induced feeding in the presence of 4t (P = 0.0004). Compound 4t did not reduce feeding in the
absence of ghrelin treatment, consistent with its mechanism of action.

Table 5. Pharmacokinetic properties of sample compounds.23
Cl
50
35
30
24
Dosep.o. (mg/kg) AUC0-6hr, p.o.
Cmaxp.o. (ng/ml) B:Pa
4o
4p
4s
4t
30
10
30
30
572
374
35
0.6
3.2
0.6
0.9
104
1662
4490
127
964
a Brain:Plasma ratio.
While we did not achieve all of our optimization goals in a single molecule, we believe that 4t represents the first orally
available, brain penetrant CAMKK2 inhibitor reported, and have demonstrated the pharmacological effect on satiety that was
expected based on the building literature around CAMKK2’s role in transducing ghrelin signaling in the hypothalamus.
Furthermore, we present some profoundly selective compounds that may serve as in vitro tools or starting points for new
optimization campaigns. As a final note, CAMKK1 activity was monitored throughout the optimization campaign and potency
tracked CAMKK2 closely, so all molecules reported here should be treated as CAMKK1/2 dual inhibitors.16
Figure 3. Cumulative grams of food intake per kilogram animal weight over 4 hours post ghrelin/saline treatment. Compound treated group contained 9
animals; vehicle treated 8 animals. Error bars reflect standard error of the mean based on pool standard deviation.
References and notes
1. Obesity: Preventing and Managing the Global Epidemic. WHO technical report series, 894. The World Health Organization, 2000.
2. Hardie, D.G., Ross, F.A., Hawley, S.A. Nat. Rev. Mol. Cell. Biol. 2012, 13, 251.
3. Sakamoto, K., McCarthy, A., Smith, D., Green, K.A., Hardie, G., Ashworth, A., Alessi, D.R. EMBO J. 2005, 24, 1810.
4. Shaw, R.J., Kosmatka, M., Bardeesy, N., Hurley, R.L., Witters, L.A., DePhino, R.A., Cantley, L.C. Proc. Natl. Acad. Sci. USA. 2004, 101,
3329.
5. Anderson, K.A., Ribar, T.J., Lin, F., Noeldner, P.K., Green, M.F., Muehlbauer, M.J., Witters, L.A., Kemp, B.E., Means, A.R. Cell Metab. 2008,
7, 377.
6. Anderson, K.A., Lin, F., Ribar, T.J., Stevens, R.D., Muehlbauer, M.J., Newgard, C.B., Means, A.R. Mol. Endocrinol. 2012, 26, 281.
7. Cary, R.L., Waddell, S., Racioppi, L., Long, F., Novack, D.V., Voor, M.J., Sankar, U. J. Bone Miner. Res. 2013, 28, 1599.
8. Peters, M., Mizuno, K., Ris, L., Angelo, M., Godaux, E., Giese, K.P. J. Neurosci. 2003, 23, 9752.
9. Using black 384-well plates at RT, 0.1 L of compound in DMSO solution was preincubated with 5 L of enzyme solution (40 mM HEPES, pH
7.2, 1 mM CHAPS, 1 mM DTT, 1 mM CaCl2, 4 nM full-length hCAMKK2, 1 M hCalmodulin) for 30 min, followed by incubation with 5 L
substrate mixture (40 mM HEPES pH 7.2,
1 mM CHAPS, 1 mM DTT, 10 mM MgCl2, 20 M ATP, 100 nM 5FAM-
AKPKGNKDYHLQTCCGSLAYRRR-amide, 1 mM CaCl2, 1 M hCalmodulin) for 120 min before addition of 10 L of 1:500 dilution IMAP
bead solution for 60-90 min and read-out in fluorescent polarization mode with excitation at 485 nM and emission at 530 nM using a 505 nM
dichroic lense.
10. 100 L of resuspended N39 cells were dispensed into 96-well tissue culture plates (~1000 cells/well) and incubated overnight at 37°C and 5%
CO2. 11 L of compound in DMSO was then incubated with cells for 80 min, followed by treatment with 12 L of 100 mM ionomycin. Media
was removed, and cells were incubated with a Cell Extraction Buffer for 30 min on ice before dilution with ice-cold MSD lysis buffer containing
protease and phosphatase inhibitors. Detection used a commercially available AMPK pT172 Immunoassay Kit supplied by Invitrogen
#KHO0651.
11. Mouse and rat CAMKK2 are >99% identical in the kinase domain, and identical in the ATP binding site (within 7 Å of ligand).
12. Aronov, A.M. Murcko, M.A. J. Med. Chem. 2004, 47, 5616.
13. Young, R.J., Green, D.V.S., Luscombe, C.N., Hill, A.P. Drug Discov. Today. 2011, 16, 822.
14. 5 L of a 10 mM stock solution of compound in DMSO was diluted to 100 L with a 10 mM phosphate buffered saline, pH 7.4, equilibrated for
1 h at RT, and filtered through Millipore MultiscreenHTS-PCF filter plates (MSSL BPC). The eluent was quantified by suitably calibrated flow
injection Chemi-Luminescence Nitrogen Detection (CLND). See Reference 13.
15. Pajouhesh, H., Lenz, G.R. NeuroRx. 2005, 2, 541.
16. CAMKK1 activities and some calculated and measured properties included in supplementary information.
17. Hong, S., Kim, J., Seo, J.H., Jung, K.H., Hong, S. Hong, S. J. Med. Chem. 2012, 55, 5337.
18. PDB code: 6CMJ; see Supplementary Information for details of crystallography. Additionally, a crystal structure of 1a bound to CAMKK2B has
recently been released by the Structural Genomics Consortium (pdb code: 6BKU), but at this time is unpublished.

19. Kothe, M., Kohls, D., Low, S., Coli, R., Rennie, G.R., Feru, F., Kuhn, C., Ding, Y. Chem. Biol. Drug. Des. 2007, 70, 540.
20. Image generated with The PyMol Molecular Graphics System, Version 1.6.0.0 Schrödinger, LLC.
21. Xu, Y., Foulks, J.M., Clifford, A., Brenning, B., Lai, S., Luo, B., Parnell, K.M., Merx, S., McCullar, M.V., Kanner, S.B., Ho, K. Bioorg. Med.
Chem. Lett. 2013, 23, 4072.
22. Adams, N.D., Adams, J.L., Burgess, J.L., Chaudhari, A.M., Copeland, R.A., Donatelli, C.A., Drewry, D.H., Fisher, K.E., Hamajima, T.,
Hardwicke, M.A., Huffman, W.F., Koretke-Brown, K.K., Lai, Z.V., McDonald, O.B., Nakamura, H., Newlander, K.A., Oleykowski, C.A.,
Parrish, C.A., Patrick, D.R., Plant, R., Sarpong, M., Sasaki, K., Schmidt, S.J., Silva, D.J., Sutton, D., Tang, J., Thompson, C.S., Tummino, P.J.,
Wang, J.C., Xiang, H., Yang, J., Dhanak, D. J. Med. Chem. 2010, 53, 3973.
23. 3 mg/kg i.v. dose. Compound formulated in 5% DMSO, 10% solutol®, and 85% saline. Reported parameters are averages from two male CD
rats for each dosing group.
24. All studies were conducted in accordance with the GSK Policy on the Care, Welfare and Treatment of Laboratory Animals and were reviewed by
the Institutional Animal Care and Use Committee either at GSK or by the ethical review process at the institution where the work was performed.
25. David H. Drewry: University of North Carolina, Structural Genomics Consortium, Center for Chemical Biology, Chapel Hill, NC 27599, USA.
Brian D. Thompson: Exela Pharma Sciences, 1245 Blowing Rock Blvd., Lenoir, NC 28645. Paul R. Reid: US Army Research Office, 800 Park
Offices Dr., Durham, NC, 27703. Patrick R. Maloney: Sanford Burnham Prebys Medical Discovery Institute, 6400 Sanger Rd., Orlando, FL
32827 . Xi Liang: MDDX Research & Informatics, 580 California St. Floor 16, San Francisco, CA, 94104. Paula K. Selley: Seattle Genetics,
21823 30th Drve S.E., Bothell, WA 98021. Octerlony B. McDonald: Retired . Jeffery L. Smith: No affiliation. Richard Cox: Kerr Health LTC,
8411 Garvey Dr., Raleigh, NC, 27616. Todd W. Shearer: Nuventra Pharma Sciences, 2525 Meridian Parkway, Suite 200, Durham, NC 27713.
Stefano Tacconi, Federico Faggioni, Ilaria Sartori, and Michela Tessari: Aptuit Center for Drug Discovery and Development, 37135 Verona,
Italy. Chiara Piubelli: Laboratory of Molecular Biology, Centre for Tropical Diseases, Sacro Cuore-Don Calabria Hospital, Via Don A.
Sempreboni, 5-37024 Negrar-Verona. Tony Wang: Kunming Biomed International, No. 1 Boda Rd. Yuhua District, Chenggong New Town,
Kunming, P.R. China 650600.

Graphical
Abstract
An orally available, brain-penetrant
CAMKK2 inhibitor reduces food intake in
rodent model
Leave this area blank for abstract info.
Daniel J. Price, David H. Drewry, Lee T. Schaller, Brian D. Thompson, Paul R. Reid, Patrick R. Maloney, Xi
Liang, Pierette Banker, Richard G. Buckholz, Paula K. Selley, Octerloney B. McDonald, Jeffery L. Smith, Todd
W. Shearer, Richard Cox, Shawn Williams, Robert A. Reid, Stefano Tacconi, Federico Faggioni, Chiara
Piubelli, Ilaria Sartori, Michela Tessari, and Tony Y. Wang