A new class of glycogen phosphorylase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2003.08.046

A new class of diacid analogues that binds at the AMP site not only are very potent but have ～10-fold selectivity in liver versus muscle glycogen phosphorylase (GP) in the in vitro assay. The synthesis, structure, and in vitro and in vivo biological evaluation of these liver selective glycogen phosphorylase inhibitors are discussed.

Full text of DOI:10.1016/j.bmcl.2003.08.046

Bioorganic & Medicinal Chemistry Letters 13 (2003) 4125–4128
A New Class of Glycogen Phosphorylase Inhibitors
Zhijian Lu,a,* Joann Bohn,a Raynald Bergeron,b Qiaolin Deng,a
Kenneth P. Ellsworth,b Wayne M. Geissler,b Georgianna Harris,b Peggy E. McCann,b
Brian McKeever,a Robert W. Myers,b Richard Saperstein,b Christopher A. Willoughby,a
Jun Yaob and Kevin Chapmana
aDepartment of Basic Chemistry, Merck Research Laboratories, Merck & Co., Inc., PO Box 2000, Rahway, NJ 07065, USA
bDepartments of Biochemistry and Metabolic Disorders, Merck Research Laboratories, Merck & Co., Inc., PO Box 2000,
Rahway, NJ 07065, USA
Received 21 February 2003; accepted 6 August 2003
Abstract—A new class of diacid analogues that binds at the AMP site not only are very potent but have ꢀ10-fold selectivity in liver
versus muscle glycogen phosphorylase (GP) in the in vitro assay. The synthesis, structure, and in vitro and in vivo biological eval-
uation of these liver selective glycogen phosphorylase inhibitors are discussed.
# 2003 Elsevier Ltd. All rights reserved.
It is known that excessive hepatic glucose production
(HGP) is a significant factor contributing to the diabetic
hyperglycemia. The liver is the major regulator of
plasma glucose levels in the post-absorptive state, and in
type 2 diabetics HGP is significantly higher than non-
diabetics.1,2 The liver produces glucose by gluconeo-
genesis and glycogenolysis. Type 2 diabetics have been
reported to display elevated gluconeogenic rates.2,3
Efforts to control HGP with gluconeogenesis inhibitors
have not been very successful except for metformin. On
the other hand, the reduction of glycogenolysis to
reduce HGP via inhibition of glycogen phosphorylase
has generated a lot of interest in recent years.4 In
humans, glucose is stored as glycogen in high con-
centrations in skeletal muscle and liver tissues, where
glycogen degradation is regulated by the control of the
activity of glycogen phosphorylase.
activity is absent. The patient is otherwise normal and
well developed except exhibits limited capability to per-
form strenuous exercise. Published data4 and our inter-
nal research5 have shown that the dimer interface
inhibitors display only marginal (2- to 3-fold) or no
selectivity between liver and muscle isozymes. On the
other hand, our research has found that a class of diacid
analogues not only are very potent but have about 10-
fold selectivity liver versus muscle in the in vitro assay.
One of the best compounds in the series, compound
11h,6 was evaluated in the glucagon challenge PD model
and it lowered glucose more than 50% at 30 mpk at
60 min after oral administration. X-ray crystallography
studies indicate that this class of compounds binds at
the AMP site. In this letter we discuss the synthesis,
structure, as well as in vitro and in vivo biological eval-
uation of these glycogen phosphorylase inhibitors.
Our research efforts have been focused on searching for
liver selective human glycogen phosphorylase inhibitors
to treat type 2 diabetic patients. We wanted to avoid
inhibiting the muscle glycogen phosphorylase due to
concern about the effect of McArdle’s disease.
The syntheses of these diacid analogues are summarized
in Schemes 1 and 2. Commercially available 3-amino-2-
naphthol was mixed with dimethyl 4-nitrophthalate and
potassium carbonate in DMF at 100 ꢁC for 12 h to
afford 2. Compound 2 reacted with the corresponding
acid chlorides to give amides 3. Amides 3 were then
either treated with phenyltrimethylsilane and iodine at
110 ꢁC or with aqueous NaOH to afford final diacids 4.
McArdle’s disease is a defect in glycogen metabolism
confined to muscle where the muscle phosphorylase
In a similar fashion, the substituted phenyl analogues
were prepared. Commercially available 4-hydroxy
*Corresponding author. Tel.: +1-732-594-4392; fax: +1-732-594-
9473; e-mail: zhijian_lu@merck.com
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.08.046

4126
Z. Lu et al. / Bioorg. Med. Chem. Lett. 13 (2003) 4125–4128
Table 1. SAR of naphthyl diacid analogues7
Compd
R
HLGP
(nM)
HMGP
(nM)
Ratio
GRAPH
(mM)
4a
4b
4c
4d
4e
4f
2880
23
167
1
12,300
140
844
3
4
6
5
3
6
6
6
7
7
>20
7.5
>20
0.3
Scheme 1. Synthesis of naphthyl diacids: (a) dimethyl 4-nitrophtha-
late, K2CO3, DMF, 100 ꢁC; (b) RCOCl, CH2Cl2, DIEA or RCO2H,
EDC, DIEA, CH2Cl2, 49–75%; (c) PhTMS, I2, 110 ꢁC or NaOH,
MeOH/H2O, 12–68%.
2
12
1.5
2
12
3.1
4g
4h
4i
10
4
57
3.5
29
2.8
12
80
2.2
potent than 4a for the liver with a selectivity ratio of 6.
Introducing a nitrogen atom at the a-position of the
phenyl carboxylic amide improves liver selectivity by
17-fold (4c vs 4a). Combining these features, the pyri-
dine analogue 4d,8 is the most potent compound
(IC50liver =1 nM) in the series with a slight drop of
selectivity (3-fold). Compared with 4d, the chloropyridyl
and the methoxypyridyl analogues are similarly potent
with improved selectivity. The methyl and the tri-
fluoromethyl analogues are 10-fold less potent than 4d,
while the ethyl analogue loses only 4-fold of potency
against the liver enzyme. All three of the alkyl analogues
display better selectivity than 4d (Table 1).
Scheme 2. Synthesis of phenyl diacids: (a) K2CO3, DMF, 100 ꢁC, 6;
66%; (b) H2, Pd/C, EtOH, 40 psi; HCl; 100%; (c) RCOCl, DIEA,
CH2Cl2 or RCO2H, EDC, DIEA, CH2Cl2; 18–43%; (d) PhTMS, I2,
110 ꢁC, 28–70%.
phthalic acid (5) was treated with thionyl chloride in
methanol under refluxing conditions to provide dime-
thyl 4-hydroxyphthalate (6) in good yield. 1-Fluoro
2-nitrobenzenes (7) were treated with 6 and potassium
carbonate in DMF at 100 ꢁC to yield ethers 8. Hydro-
genation of 8 with Pd/C at 40 psi gave amines which
were converted into hydrochloride salts 9. These amides
were coupled with the corresponding acids or acid
chlorides to afford amides 10. Final deprotection of 10
using phenyltrimethylsilane and iodine at 110 ꢁC gave
the desired products 11.
In a different SAR study, we attempted to determine the
effect of the substitutions on the central phenyl ring.
For a better comparison, we used the best amides from
the previous study. In the non-substituted phenyl series
11(a–e), we observed a similar trend: NO2>ClꢂOMe.
The nitropyridyl analogue 11a is very potent (3 nM) and
selective (8-fold). Other phenyl analogues (11b–c) are
about 10-fold less potent than the corresponding naph-
thyl analogues (4e–f). Interestingly the selectivity ratios
for the non-substituted phenyl analogues (11a–e) are 8-
to 12-fold better than the naphthyl analogues. But in the
cell based assay, only 11a and 11e show low mMinhibi-
tory activity. In the substituted phenyl series (11f–k), we
investigated the effect of the fluorine substitution at
Compounds 4a–i and 11a–k were evaluated in the
enzyme inhibition assay against human liver and muscle
glycogen phosphorylase and the cell based assay.7 The
results are outlined in Tables 1 and 2. Initially, we
investigated the SAR of 4 around the aromatic amides.
Phenyl amide 4a is a weak GP inhibitor to both liver
and muscle with a moderate selectivity (4.3-fold). The
3-nitrophenyl analogue 4b is about 100-fold more

Z. Lu et al. / Bioorg. Med. Chem. Lett. 13 (2003) 4125–4128
4127
Table 2. SAR of phenyl diacid analogues7 Table 3. A comparison among human, mouse and rat liver enzymes7
Compd
HLGP
(nM)
Mouse liver
GP (nM)
Rat liver
GP (nM)
11b
11c
11e
11f
11h
17.4
20
16
249
10
14
18
14
207
<9
16
24
19
302
<9
Compd
R
X
HLGP HMGP Ratio GRAPH
(nM)
(nM)
(mM)
11a
11b
11c
11d
11e
11f
11g
11h
11i
H
H
3
25
8
10
10
12
10
7
1
17
20
39
16
250
1
181
200
468
167
1860
9
8
13
15
3
H
H
H
30-F
30-F
30-F
30-F
40-F
40-F
3
9
0.3
0.3
1
10
11
13
78
60
6
111
109
647
10
8
Figure 1. X-ray crystallographic representation of the complex of
compound 11a with neighboring residues of rabbit muscle glycogen
phosphorylase.
11j
1.8
11
useful guide to correlate the animal study with the
potential human studies. Table 3 summarizes these data.
11k
8
To understand where and how these acids are bound to
the enzyme, an X-ray crystallographic analysis has been
performed. Given the similarity between the liver and
muscle enzymes and among the animal species, we chose
the readily available rabbit muscle enzyme. Compound
11a was co-crystallized with the rabbit muscle enzyme
(details to be published elsewhere).10 The crystal struc-
ture of the complex, solved to 2 A resolution, shows that
11a binds at the AMP site of the enzyme and the mol-
ecule bends to a V-shape (Fig. 1). The diacid forms a
number of stabilizing interactions with Arg 81, Arg 309
and Arg 310 and Tyr 155. The central aromatic ring
inserts into a lipophilic pocket containing Val 450, Gln
72 and Tyr 75. The pyridine ring of 11a fits into another
pocket formed by Trp 67 and Val 400, Ile 68 and Gln 71.
The nitro group appears to stack with the guanidinium
group of Arg 193. It is still not fully understood why the
pyridyl nitrogen is so important for the binding. It is
hypothesized that the nitrogen atom in the ring may
change the electron distribution of the aromatic ring
facilitating the beneficial interactions in the pocket.
Figure 1 shows the picture of the binding site.
various positions. As shown in Table 2, 30-fluoro phenyl
analogue 11f has moderate liver activity, but it displays
a low mMactivity in the cell based assay. If this trend
holds, the substituted pyridine analogues should be sig-
nificantly better. It turns out that these analogues dis-
play better cell activity (11g,h,i) while maintaining
enzymatic inhibitions. For example, compound 11g is 3
times more active both in the inhibition assay and in the
cell assay than 11a. In fact, compound 11g is another
excellent GP inhibitor (IC50=1 nM) along with 4d as
the results of our SAR studies have shown. The cell
based assay shows that these 30-fluorinated phenyl ana-
logues are 9- to 29-fold more active than the corre-
sponding
non-fluorinated
analogues
but
the
40-fluorinated analogues do not improve the cell based
activity.
In the light of these studies, we further investigated how
these acids would act on the mouse and rat liver
enzymes. Not surprisingly, all of the selected com-
pounds display similar activity among the human,
mouse and rat liver enzymes. The result might give us a
Compound 11h was chosen for further in vivo studies
based on its excellent cell inhibitory activity and reason-

4128
Z. Lu et al. / Bioorg. Med. Chem. Lett. 13 (2003) 4125–4128
3. Landau, B. R.; Wahren, J.; Chadramouli, V.; Schumann,
W. C.; Ekberg, K.; Kalhan, S. C. J. Clin. Invest. 1996, 98,
378.
4. (a) Martin, W. H.; Hoover, D. J.; Armento, S. A.; Stock,
I. A.; McPherson, R. K.; Danley, D. E.; Stevenson, R. W.;
Barrett, E. J.; Treadway, J. L. Proc. Natl. Acad. Sci. U.S.A.
1776, 98, 95. (b) Hoover, D. J.; Lefkowitz-Snow, S.; Burgess-
Henry, J. L.; Martin, W. H.; Armento, S. J.; Stock, I. A.;
McPherson, R. K.; Genereux, P. E.; Gibbs, E. M.; Treadway,
J. L. J. Med. Chem. 1998, 41, 2934. (c) Rath, V. L.; Ammirati,
M.; Danley, D. E.; Ekstrom, J. L.; Gibbs, E. M.; Hynes, T. R.;
Mathiowetz, A. M.; McPherson, R. K.; Olson, T. V.; Tread-
way, J. M. L.; Hoover, D. J. Chem. Biol. 2000, 7, 677. (d)
Oikonomakos, N. G.; Schnier, J. B.; Zographos, S. E.; Skam-
naki, V. T.; Tsitsanou, K. E.; Johnson, L. N. J. Biol. Chem.
2000, 275, 34566. (e) Bergans, N.; Stalmans, W.; Goldmann,
S.; Vanstapel, F. Diabetes 2000, 49, 1419.
5. (a) Chen, L.; Bergeron, R.; Deng, Q.; Ellsworth, K. P.;
Geissler, W. G.; Harris, G.; Myers, R. W.; Willoughby, C. A.;
Yao, J.; Chapman, K., in preparation. (b) Roseuer, K.; Ber-
geron, R.; Deng, Q.; Ellsworth, K. P.; Geissler, W. G.; Harris,
G.; Myers, R. W.; Willoughby, C. A.; Yao, J.; Chapman, K.,
in preparation.
6. Analytical data for 11h: 1H NMR (CD3OD, 500 MHz) d
8.55 (d, J=5.5 Hz, 1H), 8.06 (d, J=1.5 Hz, 1H), 7.15 (d,
J=8.5 Hz, 1H), 7.61 (dd, J=5.5, 1.5 Hz, 1H), 7.42 (dd, J=15,
8.5 Hz, 1H), 7.21 (d, J=2 Hz, 1H), 7.15 (t, J=8.5 Hz, 1H),
7.09 (dd, J=8.5, 2.5 Hz, 1H), 6.99 (d, J=8.5 Hz, 1H). LC–MS
(ES) 431 (M+1).
7. (a) Phosphorolysis of glycogen using recombinant human
liver or muscle enzyme. Glucose production was monitored
via enzyme linked assay involving phosphoglucomutase/glu-
cose dehydrogenase-mediated NADH production (ex. 340 nM,
em. 465 nM). (b) Glucagon-stimulated phosphorolysis using
isolated rat hepatocytes.
Figure 2. Effect of 11h on glucagon-induced glucose levels.
able pharmacokinetics.9 It was carried out in glucagon
challenge model. In the study, 11h was dosed in the
animals at 30 and 100 mpk. The plasma glucose levels of
the treated and untreated C57BI/6j mice were then
monitored. Sixty minutes later, glucagon was intro-
duced. As shown in Figure 2, in the treated group, the
glucose levels were reduced initially at both 30 and
100 mpk. Once dosed with glucagon, the glucose level in
the untreated animal was significantly higher than the
vehicle; on the other hand, the glucose levels of the
treated group increased only slightly. The glucose levels
were lower than that of the vehicle as well (Fig. 2). This
indicates that even dosed with glucagon compound 11h
is able to lower the glucose level significantly.
8. Analytical data for 4d: 1H NMR (CD3OD, 500 MHz) d
9.00 (s, 1H), 8.91 (d, J=6 Hz, 1H), 8.75 (d, J=5.5 Hz, 1H),
8.24 (dd, J=6, 3 Hz, 1H), 7.94 (d, J=7.5 Hz, 1H), 7.86 (d,
J=6.5 Hz, 1H), 7.70 (d, J=6.5 Hz, 1H), 7.59 (m, 1H), 7.56 (d,
J=2.5 Hz, 1H), 7.45 (s, 1H), 7.42 (m, 1H), 7.29 (dd, J=10,
3 Hz, 1H). LC–MS (ES) 496 (M+Na).
9. Pharmacokinetics data for 11h: Male C57BL/6J mice
were dosed iv at 1 mg/kg and orally at 2 mg/kg by gavage
(N=3 for both iv and po). Cmax=44 nM; t1=2 ¼ 60 min;
%F=4.1.
The in vivo selectivity study of compound 11h will be
the subject of another publication.11 In summary, we
have discovered a new class of glycogen phosphorylase
inhibitors that bind at the AMP site. One of the best of
these inhibitors has shown an ability to reduce plasma
glucose in vivo.
References and Notes
10. McKeever, B., in preparation.
11. Myers, R. W.; Bergeron, R.; Lu, Z.; Bohn, J.; Deng, Q.;
Ellsworth, K. P.; Geissler, W. G.; Harris, G.; Myers, R. W.;
Willoughby, C. A.; Yao, J.; Chapman, K., in preparation.
1. Consoli, A. Diabetes Care 1992, 15, 430.
2. Gerich, J. E. Horm. Metab. Res. 1992, 26, 18.