Aglycone exploration of C-arylglucoside inhibitors of renal sodium-dependent glucose transporter SGLT2

Article abstract of DOI:10.1016/j.bmcl.2008.07.109

Inhibition of sodium-dependent glucose transporter 2 (SGLT2), the transporter that is responsible for renal re-uptake of glucose, leads to glucosuria in animals. SGLT-mediated glucosuria provides a mechanism to shed excess plasma glucose to ameliorate dia

Full text of DOI:10.1016/j.bmcl.2008.07.109

Bioorganic & Medicinal Chemistry Letters 18 (2008) 4770–4773
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Aglycone exploration of C-arylglucoside inhibitors of renal
sodium-dependent glucose transporter SGLT2
*
Bruce A. Ellsworth , Wei Meng, Manorama Patel, Ravindar N. Girotra, Gang Wu, Philip M. Sher,
Deborah L. Hagan, Mary T. Obermeier, William G. Humphreys, James G. Robertson, Aiying Wang,
*
Songping Han, Thomas L. Waldron, Nathan N. Morgan, Jean M. Whaley, William N. Washburn
Research and Development, Bristol Myers Squibb Co., PO Box 5400, Princeton, NJ 08543-5400, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Inhibition of sodium-dependent glucose transporter 2 (SGLT2), the transporter that is responsible for
renal re-uptake of glucose, leads to glucosuria in animals. SGLT-mediated glucosuria provides a mecha-
nism to shed excess plasma glucose to ameliorate diabetes-related hyperglycemia and associated compli-
cations. The current study demonstrates that the proper relationship of a 40-substituted benzyl group to a
b-1C-phenylglucoside is important for potent and selective SGLT2 inhibition. The lead C-arylglucoside
(7a) demonstrates superior metabolic stability to its O-arylglucoside counterpart (4) and it promotes glu-
cosuria when administered in vivo.
Received 16 May 2008
Revised 25 July 2008
Accepted 28 July 2008
Available online 31 July 2008
Keywords:
Diabetes
Metabolism
Glucosidase
Metabolic diseases
NIDDM
Ó 2008 Elsevier Ltd. All rights reserved.
C-Glucoside
C-Arylglucoside
SGLT
SGLT2
Sodium-dependent glucose transporter
Hyperglycemia
Glucosuria
Phlorizin
Dapagliflozin
Glucosidase
Diabetes is a disease characterized by episodic bouts of hypergly-
cemia for which alternative complementary treatments are
needed.1 Attenuation of renal glucose recovery by inhibition of so-
dium-dependent glucose transporters (SGLTs) has been suggested
as a means to ameliorate hyperglycemia.2,3 Under normal condi-
tions, excess renal capacity exists to ensure complete glucose recov-
ery from the glomerular filtrate; however, inhibition of SGLT
diminishes this capability, resulting in glucosuria. Although the
more ubiquitously expressed SGLT1 plays a contributing role, cur-
rent evidence suggests that the primary effector for renal glucose
recovery is SGLT2, which is expressed on the luminal surface of the
renal proximal tubules.4 Selective SGLT2 inhibitors are desired since
inhibition of SGLT1, which is predominantly expressed in the gut to
absorb glucose and galactose, should produce gastrointestinal dis-
turbances.5,6 The risk of hypoglycemia due to selective SGLT2 inhibi-
tion is expected to be minimal, since the normal counter-regulatory
mechanisms would be maintained.
The O-arylglucoside natural product phlorizin 1 is a non-selec-
tive potent SGLT inhibitor (Fig. 1).7 Phlorizin protects diabetic ani-
mal models against hyperglycemia and associated glucose toxicity;
however, poor oral bioavailability due to b-glucosidase-mediated
metabolism in the gut necessitates subcutaneous administration.8
Structural modification of phlorizin by researchers at Tanabe Sei-
yaku led to the selective O-arylglucoside SGLT2 inhibitor T-
1095A (2).9 When administered po in rodent models of diabetes
as the 6-O-methylcarbonate prodrug T-1095, the resulting glucos-
uria induced by T-1095A, following liberation by liver esterases,
markedly ameliorated glycemic levels.3 Subsequently, we10 and
researchers at Kissei Pharmaceutical Co, Ltd11 have reported that
O-arylglucosides of o-benzylphenols are potent selective SGLT2
inhibitors. However, the susceptibility of 4, our lead compound
of this series, to b-glucosidase metabolism resulted in species-
dependent exposure and glucosuric efficacy.
* Corresponding authors. Tel.: +1 609 818 4965; fax: +1 609 818 3460
(B.A. Ellsworth), tel.: +1 609 818 4971; fax: +1 609 818 3460 (W.N. Washburn).
E-mail addresses: Bruce.Ellsworth@BMS.com (B.A. Ellsworth), William.
Washburn@BMS.com (W.N. Washburn).
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.07.109

B. A. Ellsworth et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4770–4773
4771
The b-C-arylglucosides depicted in Scheme 1 were synthesized
via the method of Czernecki and Ville17 Bihovsky et al.18 and Jara-
millo and Knapp,19 with the limitation that the aglycone function-
ality be resistant to the strongly basic and strongly acidic
conditions employed in this route. Bromodiarylmethanes, pre-
pared by methods reported in the literature,20 were lithiated and
then added to 2,3,4,6-tetra-O-benzylgluconolactone21 (8). Reduc-
tion of the resultant lactol generated predominantly b-linked glu-
cosides22 (formula II) that were deprotected by hydrogenolysis to
give compounds of formula I in which A is a methylene.
OH
O
HO
HO
O
O
HO
HO
O
HO
HO
O
H
O
H
X
HO
HO
O
O
HO
1 Phlorizin
2 T-1095A (X=O) 3 (X=CH2)
OMe
HO
Alternatively, as outlined in Scheme 2, the series of aryl gluco-
sides I in which A is a methylene could be obtained by conversion
of 2,3,4,6-tetra-O-benzyl-1-(3-bromo)phenyl-1-deoxyglucose (9)
to the corresponding aryl-trimethylstannane, followed by palla-
dium-catalyzed coupling to benzylhalides.23 The corresponding
biphenylglucosides I in which A is a bond were obtained as de-
picted in Scheme 2 by coupling of the versatile intermediate 9 with
arylboronic acids. Intermediate 9 was prepared by addition of
3-lithiobromobenzene to gluconolactone 8 followed by reduction
of the lactol.
Other aglycones were synthesized as outlined in Scheme 3; in
some cases, an unsaturated linker was carried through the synthe-
sis whereupon catalytic hydrogenation concomitantly deprotected
the glucosyl benzyl ethers and reduced the double bond. Lewis
acidic conditions were used to deprotect benzyl ethers in com-
pounds with aglycone functionality that were incompatible with
hydrogenolytic deprotection.24,25
The accompanying table (Table 1) summarizes the structure–
activity consequences upon alteration of the spacer moiety be-
tween the C-glucoside proximal and distal rings. Variation of the
spacer from one (7) to zero (10), two (11) or three methylenes
(12) reduced affinity ꢀ3-fold for the unsubstituted (R = H) and
m-methyl-substituted examples. In contrast, changing the methy-
lene spacer of 7a from one to zero (10c), two (11c), or to three
methylenes (12c) reduced the binding affinity of the p-methyl-
substituted analogs 13-, 19-, and 29-fold, respectively. The unique
HO
HO
O
X
HO
OH
NHEt
HO
HO
O
HO
HO
HO
O
6
4 (X=O) 5 (X=CH2)
R
7a R = 4-Me
7b R = H
7c R = 3-Me
HO
HO
HO
O
HO
Figure 1. Structures of SGLT inhibitors.
In an attempt to increase the metabolic stability of the glucosyl
linkage of O-arylglucoside 4, we synthesized C-benzylglucoside
5.12 This compound displayed a significant loss in SGLT2 activity
(75-fold) as compared to compound 4. Link et al. reported that a
similar modification of 2 to generate the carbon analog 3 produced
a >20-fold loss in potency.13 Together, these findings imply that the
isosteric replacement of the oxygen glucosidyl link with a methy-
lene greatly attenuated previously favorable ligand–protein inter-
actions. Possibly, the greater conformational freedom of 3 and 5
contributed to the reduction in SGLT2 affinity due to removal of
the conformational constraints imposed by the exo-anomeric
effect.14,15
Fortuitously, an alternative lead for C-glucoside-derived SGLT2
inhibitors surfaced upon characterization of 6, a minor C-arylg-
lucoside side-product that was generated during our SGLT2 pro-
gram.10 Of particular interest was the meta presentation of the
glucosyl and benzyl appendages of 6 rather than the typical ortho
presentation of O-glucoside-derived inhibitors. The promising
activity of 6 (EC50 SGLT2 = 1300 nM) and selectivity (>6-fold vs
SGLT1), despite the presence of polar substituents that had been
found to be unfavorable in the O-glucoside SAR, prompted the syn-
thesis of 7a, the counterpart of 4. The in vitro profile of 7a was ex-
tremely encouraging: SGLT2 EC50 = 22 nM; >600-fold selectivity vs.
SGLT1. The importance of a para substituent on the distal ring be-
came readily apparent upon comparison of 7a to the parent 7b or
meta isomer 7c thereby underscoring the role of the substituent to
properly orient the distal ring to achieve high affinity. A similar
bias for para substitution of the distal ring had been observed for
O-arylglucoside analogs of both 4 and dihydrochalcones reported
by Hongu et al.16
BnO
BnO
O
O
R
A
i) a,
BnO
R
BnO
A
BnO
8
BnO
BnO
BnO
O
II
ii) b
Br
H
R
A
HO
c
HO
HO
O
I
HO
Scheme 1. General route employed in the synthesis of C-arylglucosides. Reagents
and conditions: (a) BuLi, THF, ꢁ78 °C, then 8 (41–64%). (b) BF3OEt2, Et3SiH, CH3CN,
ꢁ30 °C (69–83%). (c) H2, Pd(OH)2 (33–95%).
These findings prompted a systematic study of meta-C-arylglu-
cosides with varying linkers to evaluate proper placement of the
distal aryl ring. The assumption was that high-affinity SGLT2 inhib-
itors require not only the distal aryl ring to bear a lipophilic substi-
tuent but also the distal ring that assumes an orientation such that
the lipophilic substituent can occupy a favorable binding pocket.
Therefore, to ensure that proper conclusions were drawn regarding
the consequences of introduction of a zero, one, two or three meth-
ylene spacer, three derivatives of each were prepared in which the
distal ring was either unsubstituted or bore a m-methyl or
p-methyl group.
R
Br
A
conditions c,d
(method 2A)
BnO
HO
a, b
BnO
BnO
BnO
O
HO
HO
O
8
or,
HO
conditions e,f,g
(method 2B)
I (A=bond, CH2)
9
Scheme 2. Synthesis of C-glycosyl bromo-aglycone 9 and conversion to compounds
of formula I. Reagents and conditions: (a) n-BuLi, m-dibromobenzene, THF, ꢁ78 °C;
(b) BF3OEt2, Et3SiH, CH3CN, ꢁ40 °C (59%, two steps). (c) RC6H4B(OH)2, Pd(PPh3)4,
Toluene/EtOH (3:1), Na2CO3, 80 °C, (90%). (d) BCl3, CH2Cl2, ꢁ78 °C (21–35%). (e)
Me3SnSnMe3, Pd(PPh3)4, toluene, 80 °C (63–75%). (f) RC6H4CH2Br, Pd(PPh3)4, THF
(49–63%). (g) H2, Pd(OH)2/C (33–60%).

4772
B. A. Ellsworth et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4770–4773
the corresponding O-glucoside 4.30 We attribute the greater in vi-
3A
a
b,c,d
R
tro stability of 7a and the diminished variability in glucosuric re-
sponse across species to 7a being impervious to glucosidase
cleavage (unlike 4). Further discussion of the in vitro SAR and in
vivo efficacy of 7a and analogs leading to the discovery of dapagli-
flozin31 will be a subject of a subsequent publication.
Conclusions: A combination of the meta-aryl presentation of the
salient structural elements of compound 6 and SAR considerations
of the o-benzylaryl-O-glucosides represented by 4 led to the dis-
covery of the C-arylglucoside 7a as a potent SGLT2 inhibitor. A
comparison of the profile of the C-glucoside 7 to that of 4 reveals
greater selectivity versus SGLT1, as well as enhanced metabolic
stability. C-Arylglucosides show enhanced glucosuric activity in
rats compared to O-arylglucosides that we attribute, in part, to
the metabolic stability of the aryl-glucosyl C–C bond. Further
exploration of the in vitro SAR around compound 7a and in vivo re-
sults will be reported in due course.
I (A=CH2CH2)
Br
Br
Br
PPh3
3B
e
R
f,g
I (A=CH2CH2CH2)
Br
Br
b,c,d
O
O
3C
h
b,c
i
I (A = O)
Br
OH
Br
O
3D
j
b,c
k
I (A = S)
Br
Br
Br
S
Scheme 3. Construction of aglycones and conversion to compounds of formula I.
Reagents and conditions: (a) KHMDS, THF, RC6H4CHO (45–99%). (b) n-BuLi, THF,
ꢁ78 °C, then 8. (c) BF3OEt2, i-Pr3SiH, CH3CN, ꢁ40 °C (28–77% two steps). (d) H2,
Pd(OH)2/C, EtOAc (40–70%). (e) NaOH, MeOH, RC6H4CHO (100%). (f) NaBH4, MeOH
(99%). (g) BF3OEt2, Et3SiH, CH3CN, ꢁ40 °C (73%). (h) 4-CH3C6H4Br, Cu(OAc)2,
pyridine, TEA, 4 Å MS, CH2Cl2 (40%). (i) BBr3, CH2Cl2, ꢁ78 °C. (j) n-BuLi, (4-MePhS)2,
THF, ꢁ78 °C (51%). (k) BF3OEt2, EtSH (30%).
Acknowledgment
We thank Dr. Robert Zahler for thoughtful comments during the
preparation of this manuscript.
Table 1
26C-Aryglucoside (I) SAR exploration of the aglycone spacing element (A) and distal
ring substituent (R)
References and notes
1. (a) Skyler, J. S. J. Med. Chem. 2004, 47, 4113; (b) Mizuno, C. S.; Chittiboyina, A.
G.; Kurtz, T. W.; Pershadsingh, H. A.; Avery, M. A. Curr. Med. Chem. 2008, 15, 61.
2. Rossetti, L.; Smith, D.; Shulman, G. I.; Papachristou, D.; DeFronzo, R. A. J. Clin.
Invest. 1987, 79, 1510.
3. Asano, T.; Ogihara, T.; Katagiri, H.; Sakoda, H.; Ono, H.; Fujishiro, M.; Anai, M.;
Kurihara, H.; Uchijima, Y. Curr. Med. Chem. 2004, 11, 2717.
4. Santer, R.; Kinner, M.; Lassen, C. L.; Scheneppenheim, R.; Eggert, P.; Bald, M.;
Brodehl, J.; Daschner, M.; Ehrich, J. H. H.; Kemper, M.; Li Volti, S.; Neuhaus, T.;
Skovby, F.; Swift, P. G. F.; Schaub, J.; Klaerke, D. J. Am. Soc. Nephrol. 2003, 14,
2873; Wright, E. M.; Turk, E. Pflugers Arch. Eur. J. Physiol. 2004, 447, 510.
5. Kasahara, M.; Maeda, M.; Hayashi, S.; Mori, Y.; Abe, T. Biochim. Biophys. Acta
2001, 1536, 141.
R
A
HO
HO
HO
O
HO
I
Compound
A
R
hSGLT2
Select. vs.
hSGLT1
Synthetic method
Scheme #
(overall yield, %)
EC50 (nM)
6. Pascual, J. M.; Wang, D.; Lecumberri, B.; Yang, H.; Mao, X.; Yang, R.; De Vivo, D.
C. Eur. J. Endocrinol. 2004, 150, 627.
7. McKee, F. W.; Hawkins, W. B. Physiol. Rev. 1945, 25, 255.
8. Malathi, P.; Crane, R. K. Biochim. Biophys. Acta 1969, 173, 245.
9. Tsujihara, K.; Hongu, M.; Saito, K.; Kawanishi, H.; Kuriyama, K.; Matsumoto, M.;
Oku, A.; Ueta, K.; Tsuda, K.; Saito, A. J. Med. Chem. 1999, 42, 5311.
7a
7b
7c
CH2
CH2
CH2
4-Me
3-Me
H
22
>600
ND
>50
>13
ND
>30
ND
ND
>20
ND
ND
>13
>15
>100
10
1 (42)27
2B (12)
2B (9)
510
190
623
1200
290
710
970
430
480
1200
630
540
69
10a
10b
10c
11a
11b
11c
12a
12b
12c
13
Bond
Bond
Bond
(CH2)2
(CH2)2
(CH2)2
(CH2)3
(CH2)3
(CH2)3
O
H
1 (11)
3-Me
4-Me
H
3-Me
4-Me
H
3-Me
4-Me
4-Me
4-Me
2A (19)
2A (11)
3A (50)
3A (11)
3A (7)
10. Washburn, W. N., unpublished results.
11. Fujikura, H., Fushimi, N., Nishimura, T., Tatani, K., Katsuno, K., Hiratochi, M.,
Tokutake, Y., Isaji, M. Glucopyranosyloxy Benzylbenzene Derivatives,
Medicinal Compositions Containing the Same and Intermediates for the
Preparation of the Derivatives. In PCT. Kissei Pharmaceutical Co., Ltd, JP, 2001
12. By analogy to the method outlined in Scheme 2, compound 5 was synthesized
in 2.5% overall yield by addition of 2-bromobenzylmagnesium bromide to
lactone 8.
3A (6)
3B (14)
3A (20)
3C (2)
14
1
S
3D (7)
13. Link, J. T.; Sorensen, B. K. Tetrahedron Lett. 2000, 41, 9213.
14. Kishi, Y. Pure Appl. Chem. 1993, 65, 771.
35
4
8
160
50
350
1
4
15. Wang, Y.; Goekjian, P. G.; Ryckman, D. M.; Miller, W. H.; Babirad, S. A.; Kishi, Y.
J. Org. Chem. 1992, 57, 482.
16. Hongu, M.; Tanaka, T.; Funami, N.; Saito, K.; Arakawa, K.; Matsumoto, M.;
Tsujihara, K. Chem. Pharm. Bull. 1998, 46, 22.
1a
2a
Note: aEC50 data from Oku et al.28
17. Czernecki, S.; Ville, G. J. Org. Chem. 1989, 54, 610.
18. Bihovsky, R.; Selick, C.; Giusti, I. J. Org. Chem. 1988, 53, 4026.
19. Jaramillo, C.; Knapp, S. Synthesis 1994, 1.
20. Wai, J. S.; Egbertson, M. S.; Payne, L. S.; Fisher, T. E.; Embrey, M. W.; Tran, L. O.;
Melamed, J. Y.; Langford, M.; Guare, J. P. J.; Zhuang, L.; Grey, V. E.; Vacca, J. P.;
Holloway, M. K.; Naylor-Olsen, A. M.; Hazuda, D. J.; Felock, P. J.; Wolfe, A. L.;
Stillmock, K. A.; Schleif, W. A.; Gabryelski, L. J.; Young, S. D. J. Med. Chem. 2000,
43, 4923.
21. Benhaddou, R.; Czernecki, S.; Farid, W.; Ville, G.; Xie, J.; Zegar, A. Carbohydr. Res.
1994, 260, 243.
22. Ellsworth, B. A.; Doyle, A. G.; Patel, M.; Caceres-Cortes, J.; Meng, W.;
Deshpande, P. P.; Pullockaran, A.; Washburn, W. N. Tetrahedron: Asymmetry
2003, 14, 3243.
advantage conferred by the single methylene of 7a is further con-
firmed by the respective 3- and 25-fold reduction in affinity upon
replacement with a sulfur (14) or oxygen (13) bridging atom. Sig-
nificant inhibition of human SGLT1 was not observed for any of the
C-arylglucosides tested. In particular, the demonstrably high level
of selectivity of 7a is expected to preclude gastrointestinal side
effects.
Upon iv administration to rats and mice at 1 and 0.3 mg/kg,
respectively, 7a produced maximum glucosuric levels of 230 and
600 mg/dL.29 In contrast, upon administration of O-glucoside 4 un-
der the same conditions, efficacy in rats was reduced ꢀ50-fold rel-
ative to that obtained in mice. C-Arylglucoside 7a was found to be
ꢀ100-fold more stable in the presence of rat liver microsomes than
23. Stille, J. K. Angew. Chem., Int. Ed. Eng. 1986, 25, 508.
24. Xu, J.; Egger, A.; Bernet, B.; Vasella, A. Helv. Chim. Acta 1996, 79, 2004.
25. Swayze, E. E.; Drach, J. C.; Wotring, L. L.; Townsend, L. B. J. Med. Chem. 1997, 40,
771.
26. The intracellular accumulation of the SGLT-selective glucose analog [14C]-
alpha-methyl glucopyranoside (AMG) by CHO cells expressing the human
SGLT2 or the human SGLT1 transporter was quantified in vitro in the presence
and absence of inhibitors, using the following conditions: Each inhibitor,

B. A. Ellsworth et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4770–4773
4773
dissolved in DMSO, was tested at 8 concentrations in the presence of 137 mM
NaCl and 10
M [14C] AMG, over a 120-min incubation in protein-free buffer.
29. Fasted male Sprague–Dawley rats or Swiss-Webster mice were anesthetized
with I.P. Ketamine:Xylazine (.001 mL/g), an abdominal incision made, and their
l
Percent inhibition of transport activity was calculated based on a comparison
of the activity of non-inhibited control cells treated with DMSO alone. The
response curve was fitted to an empirical four-parameter model to determine
the inhibitor concentration at half-maximal inhibition. The reported EC50 is the
aggregate result of triplicate dose–response determinations.
bladders cannulated with a 16-gauge catheter. Drug was administered
intravenously, and urine was collected over 60 min in 10 min intervals.
Glucose concentration in urine samples was determined by COBAS MIRA.
30. The oxidative metabolism of SGLT2 inhibitors was evaluated in liver
microsomes using the following conditions: Each BMS compound was
27. Characterization of compound 7a: 1H NMR (400 MHz, CD3OD)
d
7.27 (s,
dissolved in acetonitrile and tested at a concentration of 10 lM in the
1H), 7.23 (d, 2H, J = 4 Hz), 7.1 to 7.0 (m, 5H), 4.08 (d, 1H, J = 9 Hz), 3.91 (s,
presence of 1 mM NADPH and 1 mg/mL microsomal protein in 50 mM
potassium phosphate buffer. The final organic solvent content was 1%.
Incubations were conducted for various times, up to 60 min. Percent
turnover was calculated based on a comparison of the parent peak at time t
compared to that at time 0. Rates (nmol/min/mg) were calculated by taking the
2H), 3.90 to 3.85 (m, 1H), 3.70 to 3.65 (m, 1H), 3.37 to 3.30 (m, 4H), 2.27
(s, 3H). 13C NMR (100 MHz, CD3OD)
d 142.7, 140.8, 139.5, 136.5, 130.0,
129.8, 129.6, 129.5, 129.1, 126.6, 83.7, 82.2, 79.8, 76.4, 71.9, 63.2, 42.4,
21.0. HPLC: 99.0% pure, 6.56 min retention time. Gradient 0–100% B over
8 min. Solvent A: 10%MeOH/H2O + 0.2%H3PO4. Solvent B: 10%MeOH/
H2O + 0.2%H3PO4. Column: YMC S-5 C-18 4.6 ꢂ 50 mm. Flow rate: 2.5 mL/
min. Monochrome detection at 220 nm.
%
turnover and multiplying by the starting nanomoles of substrate in
incubation, then dividing by time of incubation and milligrams of protein.
31. Meng, W.; Ellsworth, B. A.; Nirschl, A. A.; McCann, P. J.; Patel, M.; Girotra, R. N.;
Wu, G.; Sher, P. M.; Morrison, E. P.; Biller, S. A.; Zahler, R.; Robertson, J. G.;
Wang, A.; Han, S.; Wetterau, J. R.; Janovitz, E. B.; Flint, O. P.; Whaley, J. M.;
Washburn, W. N. J. Med. Chem. 2008, 51, 1145.
28. Oku, A.; Ueta, K.; Arakawa, K.; Ishihara, T.; Nawano, M.; Kuronuma, Y.;
Matsumoto, M.; Saito, A.; Tsujihara, K.; Anai, M.; Asano, T.; Kanai, Y.; Endou, H.
Diabetes 1999, 48, 1794.