Phenylcyanoguanidines as inhibitors of glucose-induced insulin secretion from beta cells

Article abstract of DOI:10.1016/S0960-894X(01)00297-9

3,5-Disubstituted-phenylcyanoguanidines have been identified as activators of SUR1/Kir6.2 potassium channels and as potent inhibitors of insulin release from pancreatic beta cells in vitro.

Full text of DOI:10.1016/S0960-894X(01)00297-9

Bioorganic & Medicinal Chemistry Letters 11 (2001) 1749–1752
Phenylcyanoguanidines as Inhibitors of Glucose-Induced Insulin
Secretion from Beta Cells
Tina M. Tagmose,a John P. Mogensen,a Pia C. Agerholm,a Per O. G. Arkhammar,a,b
Philip Wahl,a Anne Worsaaea and J. Bondo Hansena,*
a
˚
Discovery and Development, Novo Nordisk A/S, Novo Nordisk Park, DK-2760, Maløv, Denmark
bBioImage A/S, Mørkhøj Bygade 28, DK-2860, Søborg, Denmark
Received 8 January 2001; revised 26 March 2001; accepted 23 April 2001
Abstract—3,5-Disubstituted-phenylcyanoguanidines have been identified as activators of SUR1/Kir6.2 potassium channels and as
potent inhibitors of insulin release from pancreatic beta cells in vitro. # 2001 Elsevier Science Ltd. All rights reserved.
The ATP regulated potassium (KATP) channels, which
are present in different tissues, for example smooth
muscle, the heart and in pancreatic beta cells are
important in coupling the effects of hormones and
transmitters and cell metabolism to cellular membrane
potential.1 KATP channel blockers have been developed
for treatment of diabetes whereas KATP channel activa-
tors are being investigated as potential drugs for the
treatment of, for example, hypertension urinary incon-
tinence, asthma and cardiac or neuronal ischemic
damage.2,3 Certain cyanoguanidines act as potent
potassium channel openers (PCOs) on vascular smooth
muscle to reduce blood pressure. These include N-pyr-
idyl derivatives (e.g., pinacidil and P1075; Fig. 1)4,5 and
chloro, cyano and nitro substituted phenyl cyanoguani-
dines.6 While pinacidil and P1075 have only minimal
effects on the KATP channels of pancreatic beta cells,
reducing the conformational flexibility of the aryl-cya-
noguanidine to form, for example, BPDZ 44, an
increase in potency with respect to inhibition of insulin
release from rat islets could be obtained.7 BPDZ 44, and
the even more potent analogue BPDZ 73,8 are related to
the 1,2,4-benzothiadiazine 1,1-dioxide, diazoxide, which
nonselectively opens KATP channels of smooth muscle
and beta cells to induce vasodilatation and inhibition of
insulin release. Diazoxide has been found to protect
beta cells in newly diagnosed patients suffering from
Type 1 diabetes,9 to potentiate weight loss in obese indi-
viduals10 and to be active in animal models of Type 2
diabetes,11,12 but are associated with several side effects.
To investigate the structural requirements of phenyl
cyanoguanidines as openers of beta cell KATP channels,
a series of compounds having a phenyl group sub-
stituted with electron withdrawing lipophilic groups and
different N-alkyl groups were prepared and studied for
effects on glucose stimulated beta cell membrane poten-
tial and insulin release versus relaxation of vascular
smooth muscle in vitro.
The method described by Yoshizumi et al. was applied for
the synthesis of 3,5-substituted phenyl-cyanoguanidines13
Figure 1. Activators of ATP regulated potassium channels.
*Corresponding author. Tel.: +45-4443-4857; fax: +45-4443-4547; e-mail: jbha@novonordisk.com
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00297-9

1750
T. M. Tagmose et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1749–1752
(Fig. 2). By this method, which involves reacting diphe-
nylcyanocarbonimidate with anilines and subsequently
treating the obtained cyano-O-phenylisourea with
alkylamines, the 3,5-disubstituted phenyl-cyanoguani-
dines derivatives are produced in moderate to good
yield.14
position was of major importance for the vascular
activity. Removing the methyl groups a to the nitrogen
one by one decreases the activity on aorta (2!1!3).
In the present series, all compounds, with one exception
(7) potently inhibited glucose stimulated insulin release.
The synthesized compounds (Table 1) were evaluated
for their ability to relax phenylepherine contracted rat
aorta rings15 and by their effects on glucose stimulated
insulin release from murine bTC6 cells16 and on murine
b-cell (bTC3) membrane potential as measured by
changes in DIBAC4(3) fluorescence.17 To further sub-
stantiate the effects of the compounds on insulin release,
selected compounds (6 and 7) were examined on freshly
isolated rat islets18 and on recombined human KATP
channels (Fig. 3).19
The inhibitory effect on insulin release was in general
accompanied by an effect on membrane potential of
bTC3 cells with the exception that compound 9 was
found to be a potent inhibitor of insulin release
(IC50=3.7 mM) with minimal effects on beta cell mem-
brane potential. This lack of correlation could be due to
non-specific interactions caused by the lipophilic com-
pound or interaction with other parts of the insulin
secretory process. In addition to being potent inhibitors
of insulin release, 3 and 10 also were tissue-selective. 6
and 7 were tested on freshly prepared rat islets and
found to be potent inhibitors of insulin release
(IC50=4.8 and 2.2 mM, respectively) both being
approximately as potent as diazoxide (IC50=6.1 mM)
(data not shown). To further evaluate the present com-
pounds, 6 (3–30 mM) was shown in a whole cell patch
clamp configuration to activate human SUR1/Kir6.2
KATP channels co-expressed in HEK 293 cells18 (Fig. 3)
whereas the compound (1–100 mM) did not activate
human SUR2A/Kir6.2 channels (data not shown). The
KATP channel blocker glibenclamide (1 mM) was able to
counteract the effects of 6 on the ion current through
the SUR1/Kir6.2 channels substantiating that 6 acts
through the KATP channels.
3,5-Bistrifluoromethylphenyl cyanoguanidines and 3,5-
dichlorophenyl cyanoguanidines, having the pinacidil
and P1075 alkyl side chains were found to inhibit insu-
lin secretion and to relax vascular smooth muscle. The
effects of several derivatives on rat aorta rings were as
potent as that of pinacidil (IC50=1.5 mM) and P1075
(IC50=0.25 mM). In contrast, the prepared monotri-
fluoromethylphenyl cyanoguanidine (5) was nearly
inactive both on aorta and beta cells. The 3-methoxy-5-
trifluoromethylphenyl cyanoguanidine (6) was con-
siderably less potent on vascular smooth muscle than on
murine beta cells and rat islets. Among the 3,5-disub-
stituted phenyl cyanoguanidines the branching in the a
Figure 2. Synthesis of phenyl-cyanoguanidines: (a) rt, 14 h; (b) R0-NH2, 75 ꢀC, 8 h.
Table 1. Effects of compounds 1–10 and reference KATP channel activators on vascular tissue and beta cells in vitro
Compounds
R, R
R0
Relaxation of rat
aorta rings
Inhibition of insulin release
bTC6 cells
Membrane potential
bTC3 cells
IC50 (mM)a
IC50 (mM)b
IC50 (mM)b
Efficacyb (%)
78 (7)
1
2
3
4
5
6
7
8
CF3,CF3
CF3,CF3
CF3,CF3
CF3,CF3
H, CF3
OCH3,CF3
Cl,Cl
CH(CH3)C(CH3)3
C(CH3)2CH2CH3
CH2CH2CH3
CH2CH2CH(CH3)2
CH(CH3)C(CH3)3
CH2CH2CH(CH3)2
CH(CH3)C(CH3)3
C(CH3)2CH2CH3
CH2CH2CH3
8.7
0.9
12.4
5.6
88
4.6 (1.2)
NT
5 (5.7)
NA
NA
0.9 (0.8)
0.5 (0.4)
>100
13 (6)
19 (14)
NA
36 (450)
4 (6)
77.4 (4.3)
58 (13)
77 (7)
29 (8)
79 (9)
86 (4)
57 (2)
50 (31)
72 (8)
23 (22)
2 (0.2)
>100
3.9 (3)
79
0.5
0.3
31
18.7
11.5
26.7 (2.9)
2.3 (3.3)
3.7 (4.7)
0.7 (0)
Cl,Cl
Cl,Cl
Cl,Cl
9
10
CH2CH2CH(CH3)2
Diazoxide
22.4 (15)
NA=not active; NT=not tested.
aValues are an average of two experiments.
bValues are means of at least two experiments, standard deviation is given in parentheses.

T. M. Tagmose et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1749–1752
1751
7. de Tullio, P.; Pirotte, B.; Lebrun, P.; Fontaine, J.; Dupont,
L.; Antoine, M.; Ouedraogo, R.; Khelili, S.; Maggetto, C.;
Masereel, B.; Diouf, O.; Podona, T.; Delarge, J. J. Med.
Chem. 1996, 39, 937.
8. Lebrun, B.; Arkhammer, P.; Antoine, M.; Bondo Hansen,
J.; Pirotte, B. Diabetologia 2000, 43, 723.
9. Bjork, E.; Berne, C.; Kampe, O.; Wibell, L.; Oskarsson, P.;
Karlsson, F. A. Diabetes 1996, 45, 1427.
10. Alemzadeh, R.; Langley, G.; Upchurch, L.; Smith, P.;
Slonim, A. E. J. Clin. Endocrinol. Metab. 1998, 83, 1911.
11. Standridge, M.; Alemzadeh, R.; Zemel, M.; Koontz, J.;
Moustaid-Moussa, N. FASEB J. 2000, 14, 455.
12. Aizawa, T.; Taguchi, N.; Sato, Y.; Nakabayashi, T.;
Kobuchi, H.; Hidaka, H.; Nagasawa, T.; Ishihara, F.; Itoh,
N.; Hashizume, K. J. Pharmacol. Exp. Ther. 1995, 275, 194.
13. Yoshiizumi, K.; Ikeda, S.; Nishimura, N.; Yoshino, K.
Chem. Pharm. Bull. 1997, 45, 2005.
14. Experimental and physical data for test compounds: N-(3,5-
(bistrifluoromethyl)phenyl)-N0-cyano-O-phenylisourea. A solu-
tion of diphenylcyanocarbonimidate (2 mmol, 476 mg), 3,5-
bis(trifluoromethyl)aniline (2 mmol, 458 mg) and triethyl-
amine (2 mmol, 202 mg) in dichloromethane (15 mL) was
stirred under nitrogen for 12 h. After concentration, the residue
was stirred with toluene (5 mL) for 2 h and the solid collected by
filtration giving 550 mg of N-(3,5-(bistrifluoromethyl)phenyl)-
N0-cyano-O-phenylisourea (73.6%); mp 190.5–191.5 ꢀC; 1H
NMR (DMSO-d6): 7.25 (m, 5H), 7.95 (s, 1H), 8.15 (s, 2H),
11.2 (s, 1H).
N-Cyano-N0 -(3,5-bis-(trifluoromethyl)phenyl)-N00 -(1,2,2-tri-
Figure 3. Effect of 6 on currents through human SUR1/KIR6.2
channels. (a) Representative experiment from HEK 293 cells stably
expressing human SUR1/Kir6.2 showing typical currents changes
upon application of 6 (30 mM) and glibenclamide (1 mM) plus 6 (30
mM). Currents were evoked by a 10 mV depolarizing pulse for 200 ms
every 10 s from a holding potential of ꢂ80 mV. (b) Dose–response
relationship for the current activated by increasing concentrations of
6. The current amplitude induced by each concentration was normal-
ized to the current induced by 30 mM 6 in each cell. Symbols and bars
indicate the mean and SE values, respectively. The number of the
observations at each point was three or four.
methylpropyl)guanidine (1).
A solution of N-(3,5-(bistri-
fluoromethyl)phenyl)-N0-cyano-O-phenylisourea (0.8 mmol,
300 mg), 2-amino-3,3-dimethylbutane (0.88 mmol, 0.09 g) and
triethylamine (0.88 mmol, 0.123 mL) in acetonitrile (2 mL)
was stirred for 8 h at 75 ꢀC. After concentration the residue
was purified by column chromatography (heptane/ethyl ace-
tate 2:1) to give the title compound (140 mg, 46%) as white
crystals. Mp 165.5–166.5 ꢀC; EI SP/MS: 380 (M+); H NMR
1
(CDCl3): 0.92 (s, 9H), 1.13 (d, 3H), 3.8 (m, 1H), 4.8 (br d, 1H),
7.74 (br s, 3H), 8.5 (br, 1H); MA calcd for C16H18F6 N4: C
50.53%, H 4.77%, N 14.73%. Found: C 50.48%, H 4.74%, N
14.45%.
2: Yield: 28%, mp 149–150 ꢀC; 3: yield: 28%, mp: 142.5–
143.5 ꢀC; 4: yield: 27%, mp 130–132 ꢀC; 5: yield: 63%, mp:
114.5–117 ꢀC; 6: yield: 81%, mp 105.5–108.5 ꢀC; 7: yield: 68%,
mp 158.5–160.5 ꢀC; 8: yield: 65%, mp 158.5–160 ꢀC; 9: yield:
23%, mp 141–143 ꢀC; 10: yield: 54%, mp 146.5–151.5 ꢀC.
15. Videbaek, L. M.; Aalkjaer, C.; Mulvany, M. J. J. Cardio-
vasc. Pharmacol. 1988, 12 (Suppl. 2), S23.
The present data suggest that by changing the structure
of pinacidil, which is nearly inactive on beta cells, to
3,5-disubstituted phenyl cyanoguanidines, it is possible
to identify compounds that are able to activate KATP
channels of pancreatic beta cells and to inhibit insulin
release. The N-alkyl side chain is important for potency
and selectivity as previously described by Yoshiizumi et
al.6 who found that 3,5-disubstituted phenyl cyano-
guandines having bulky alkyl substituents are potent
dilators of vascular smooth muscle, and by Pirotte et al.,
showing that in a series of 4H-pyrido[4,3-e]-1,2,4-thia-
diazine 1,1-dioxides the branching in the a position was
important for the inhibitory effects on insulin.7
16. Inhibition of glucose induced insulin release in bTC6:
bTC6 was cultured at 5ꢁ104 cells/well in microtiter plates in
DMEM+10%FCS, 1 g/L glucose, 1% Glutamax and 20 mM
Hepes for 3 days. Cells were washed twice with NN buffer
(NaCl, 114 mM; KCl, 4.7 mM; KH2PO4, 1.21 mM; MgSO4,
.
1.16 mM; NaHCO3, 25.5 mM; CaCl 2H2O, 2.5 mM; HEPES,
10 mM) supplemented with 0.1% BSA and incubated for 60
min in this buffer. All wells were aspirated and the cells incu-
bated for 3 h with NN buffer, 22 mM glucose and a serial
dilution of the compounds. 0.1 mM IBMX was added in order
to potentiate the glucose stimulated insulin release. A refer-
ence compound and a series of different glucose concentra-
tions without compound were used as references. The
supernatant from each well was harvested and insulin content
was measured by a competition Elisa. In brief, Elisa microtiter
plates were coated with anti-guinea pig IgG and incubated
overnight at 4 ꢀC in PBS. All plates were washed five times
with Washing buffer (PBS diluted 1:4 in H2O+0.05%
Tween20) and incubated 30 min in this buffer at rt. Wells were
aspirated and anti-insulin antibodies [polyclonal GP4 (NN)]
was added, followed by an incubation for 2 h at rt. Plates were
References and Notes
1. Aguilar-Bryan, L.; Bryan, J. Endocrinol. Rev. 1999, 20, 101.
2. Ashcroft, F. M.; Gribble, F. M. Diabetologia 1999, 42, 903.
3. Atwal, K. S. Curr. Med. Chem. 1996, 3, 227.
4. Petersen, H. J.; Nielsen, C. K.; Arrigoni-Martelli, E. J.
Med. Chem. 1978, 21, 773.
5. Nicholls, D. P.; Murtagh, J. G.; Scott, M. E.; Morton, P.;
Shanks, P. G. Br. J. Clin. Pharmacol. 1986, 22, 287.
6. Yoshiizumi, K.; Ideka, S.; Goto, K.; Morita, T.; Nishi-
mura, N.; Sukamoto, T.; Yoshino, K. Chem. Pharm. Bull.
1996, 44, 2042.

1752
T. M. Tagmose et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1749–1752
washed five times in washing buffer and test samples were
added together with Peroxidase-labelled insulin and incubated
for another 2 h at rt in Assay buffer (Washing buffer+0.5%
BSA). A series of standards of rat insulin were made to cover a
range of 1000 to 1 mg/mL insulin and the standards were
incubated with PO-insulin as well. Eventually, TMB substrate
was added to all wells and the enzyme reaction stopped after 5
min by adding H3PO4. Absorption was measured in an Elisa
reader and converted into ng/mL insulin. The results were
analyzed in Prism and expressed as EC50 (potency) and per-
centage inhibition (efficacy) of insulin release.
19. Electrophysiological recordings. The whole cell patch-
clamp technique was employed to measure the currents in a
HEK 293 cell line stably expressing the human SUR1/KIR6.2
receptor channel. The microelectrodes were pulled from bor-
osilicate glass and had resistances between 3 and 5 Mꢀ and
were filled with internal solution containing (in mM): 120 KCl,
1 MgCl2, 5 EGTA, 2 CaCl2, 20 Hepes, 0.3 NaADP, 5 K2ATP
(pH 7.3). The external recording solution was (in mM):
140 NaCl, 3 KCl, 1 CaCl2, 1 MgCl2, 20 mannitol, 10 Hepes
(pH 7.2). Currents were recorded on an EPC9 amplifier
(HEKA Electronic GmbH, Lambrecht, Germany) and filtered
at 2 kHz and sampled at 10 kHz. Data analysis was per-
formed using Pulse+PulseFit software (HEKA Electronic
GmbH, Lambrecht, Germany). Measuring the maximal cur-
rent induced by increasing concentrations of agonist pro-
duced agonist dose–response curves. Data were fitted to the
logistic equation: I=Imax/{1+(EC50/[6]n}, where I is the
current induced by 6. The parameters Imax (maximal current
at infinite 6 concentration), n (the Hill coefficient, and EC50
17. Effects on membrane potential of bTC3 cells were mea-
sured according to ref 8.
18. Inhibition of glucose induced insulin release in rat islets.
Islets were isolated by collagenase and gradient centrifugation
in Ficoll (40–13%). Isolated islets were incubated in bulk
overnight in RPMI, 10% FCS, 11 mM glucose. The islets were
handpicked and placed at 10 islets/microtiter well and cultured
overnight in DMEM, 10% FCS, and 3 mM glucose.
Essentially, the islets were tested as described for the bTC6
but with no addition of IBMX. The insulin content was mea-
sured in the same Elisa as used for the bTC6.
(concentration of 6 producing a current of 50% of Imax)
were determined by an iterative least-squares fitting routine
(Origen, MicroCal Software, MA, USA).