Glycosylated dihydrochalcones as potent and selective sodium glucose co-transporter 2 (SGLT2) inhibitors

Article abstract of DOI:10.1016/j.bmcl.2004.07.082

A series of glucose conjugates was synthesized and tested for inhibition of SGLT1 and SGLT2.

Full text of DOI:10.1016/j.bmcl.2004.07.082

Bioorganic & Medicinal Chemistry Letters 14 (2004) 5121–5125
Glycosylated dihydrochalcones as potent and selective
sodium glucose co-transporter 2 (SGLT2) inhibitors
Joseph Dudash, Jr.* Xiaoyan Zhang, Roxanne E. Zeck, Sigmond G. Johnson,
Geoffrey G. Cox, Bruce R. Conway, Philip J. Rybczynski and Keith T. Demarest
Johnson & Johnson Pharmaceutical Research and Development, L.L.C., 1000 Rt. 202, PO Box 300, Raritan, NJ 08869, USA
Received 19 May 2004; accepted 28 July 2004
Available online 28 August 2004
Abstract—A series of glucose conjugates was synthesized and tested for inhibition of SGLT1 and SGLT2. The core structure was
derived from compound 1a. Modification of the benzofuran moiety and 40-substituent of the phenyl ring in compound 1a improved
selectivity at SGLT2. Select compounds were compared to 1a in metabolic stability and in vivo efficacy studies.
Ó 2004 Elsevier Ltd. All rights reserved.
Diabetes mellitus is a polygenic disorder characterized
by chronic hyperglycaemia associated with a deficiency
in insulin action. In diabetic patients, one mechanism
for protection against the adverse effects of high plasma
glucose levels is the compensatory increase in urinary
glucose excretion.1 In the kidney, plasma glucose is con-
tinuously filtered in the glomerulus and then reabsorbed
in the proximal tubules by a class of transporters called
the sodium-glucose co-transporters (SGLTs).2 Renal
reabsorption of glucose is mediated by SGLT1 and
SGLT2,1 while intestinal absorption of glucose is prima-
rily mediated by SGLT1 and the facilitative glucose
transporter GLUT5.3
wanted side effects.7 We report herein a series of Phlor-
izin analogues with increased selectivity towards
SGLT2.
OH
O
O
OH
O
O
6'
O
4'
HO
OH
OH
OH
OH
OH
O
O
OR OH
OH OH
1a. R = H
Phlorizin
1b. R = CO2CH3
Phlorizin, a natural SGLTs specific inhibitor, provided
proof of concept in vivo by promoting glucose excretion
and lowering fasting and postprandial blood glucose
without hypoglycemic side effects in several animal mod-
els.4 Compound 1a and its carbonate prodrug 1b, which
are synthetic analogues of Phlorizin, have been shown to
inhibit glucose transport in brush border membrane ves-
icles by inhibition of both SGLT1 and SGLT2.5 Other
synthetic SGLT inhibitors have also been published as
potential treatments for diabetes.6 Considering that
SGLT1 is a high affinity/lowcapacity transporter and
SGLT2 is a lowaffinity/high capacity transporter, selec-
tive inhibition of SGLT2 should produce a therapeuti-
cally useful enhancement of urinary glucose excretion.
In addition, inhibition of SGLT1 may produce un-
Our goal was to synthesize compounds that would ad-
dress the structure–activity relationships of three regions
in compound 1a: the benzofuran portion, the 40-position
of the phenyl ring, and the 60-hydroxyl group of the phe-
nyl ring. As shown in Scheme 1,5a the appropriately sub-
stituted resorcinol 2 was treated with acetic anhydride
and then subjected to Fries rearrangement conditions
to give the ketone 3. For analogues where R1 = Cl and
R2 = H, the ketone 4 was commercially available. For
analogues where R1 = NBn2 and R2 = Me, the ketone
5 was prepared from pentane-2,4-dione and 2,2,6-tri-
methyl-[1,3]dioxin-4-one as reported in the literature.8
Monoglycosylation of 3–5 proceeded under phase
transfer conditions with 2,3,4,6-tetra-O-acetyl-a-D-glu-
copyranosyl bromide to give the peracetylated sugar
derivatives 6–8. Aldol condensation with the desired
aldehyde under saponifying conditions, followed by
*
Corresponding author. Tel.: +1 908 704 4861; fax: +1 908 203
8109; e-mail: jdudash@prdus.jnj.com
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.07.082

5122
J. Dudash, Jr. et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5121–5125
R1
O
O
R1
O
O
R3
R1
R1
O
d, e
c
a, b
R2
R2
OH
OH
OAc
OAc
O
R2
OH
R2
OH
O
OH OH
OAc OAc
2a. R1 = OH, R2 = H
2b. R1 = OH, R2 = Me
2c. R1 = OH, R2 = Et
2d. R1 = OH, R2 = iPr
2e. R1 = OH, R2 = Cl
2f. R1 = OH, R2 = F
9a - 9q
3a. R1 = OH, R2 = H
3b. R1 = OH, R2 = Me
3c. R1 = OH, R2 = Et
3d. R1 = OH, R2 = iPr
3e. R1 = OH, R2 = Cl
3f. R1 = OH, R2 = F
4. R1 = Cl, R2 = H
6a. R1 = OH, R2 = H
6b. R1 = OH, R2 = Me
6c. R1 = OH, R2 = Et
6d. R1 = OH, R2 = iPr
6e. R1 = OH, R2 = Cl
6f. R1 = OH, R2 = F
7. R1 = Cl, R2 = H
10a - 10c
11a - 11c
5. R1 = NBn2, R2 = Me
8. R1 = NBn2, R2 = Me
Scheme 1. Reagents: (a) acetic anhydride, pyridine; (b) AlCl3, chlorobenzene, 90°C; (c) 2,3,4,6-tetra-O-acetyl-a-D-glucopyranosyl bromide,
benzyltributylammonium chloride, K2CO3, CHCl3, H2O; (d) R3CHO, KOH, EtOH; (e) H2, Pd/C, DAMP, EtOH.
hydrogenation of the ensuing olefin gave the target com-
pounds 9–11.9
benzofuran to other bicyclic ring systems such as naph-
thalene (9a) or benzodioxane (9b) led to analogues with
similar SGLT2 inhibitory activity to 1a but with in-
creased selectivity for SGLT2 versus SGLT1. Other
bicyclic heterocycles gave analogues with similar poten-
cies (data not shown). The monocyclic 4-ethoxyphenyl
All compounds were screened in the SGLT cell base
functional assay.10 Modification of the benzofuran moi-
ety of 1a was explored first (Table 1). Changing from
Table 1. In vitro screening data for SGLT inhibitory activity and selectivity
R1
O
R3
R2
O
OH
OH
O
OH OH
Entry
1a
R1
R2
R3
SGLT2 Kia (lM)
0.004 0.001
0.010 0.004
SGLT1 Kia (lM)
0.058 0.004
1.325 0.034
Ratio (SGLT1/SGLT2)
OH
OH
Me
Me
14.5
O
9a
132.5
O
O
9b
OH
Me
0.009 0.001
1.461 0.432
162.2
9c
OH
OH
OH
OH
Cl
Me
Me
Me
Me
H
0.017 0.007
0.235 0.019
0.070 0.004
0.045 0.002
>1.7
1.573 0.395
4.383 0.705
1.384 0.006
1.080 0.273
>45
92.5
18.7
19.8
24.1
––
OEt
9d
9e
O
S
9f
10a
O
O
10b
Cl
H
>1.7
>45
––

J. Dudash, Jr. et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5121–5125
5123
Table 1 (continued)
Entry
R1
R2
H
R3
SGLT2 Kia (lM)
SGLT1 Kia (lM)
Ratio (SGLT1/SGLT2)
10c
Cl
>1.7
>45
––
OEt
11a
NH2
Me
0.520 0.008
9.897 2.626
19.1
O
O
11b
11c
NH2
NH2
Me
Me
0.297 0.020
1.283 0.300
16.15 3.791
13.52 1.348
54.3
10.5
OEt
a Ki values are mean SEM.
analogue (9c) gave a similar result. However, the mono-
cyclic furan-(9d), thiophene-(9e) and cyclohexane-(9f)
analogues were all less potent and less selective than
9b. Thus, bicyclic or 4-substituted phenyl rings found
in compounds 9a–c were the optimal replacements for
the benzofuran moiety in 1a. In this set, the best results
were obtained with the benzodioxane analogue 9b. This
compound is a potent SGLT2 inhibitor and 162-fold
selective for SGLT2 over SGLT1.
The importance of the 60-hydroxyl group was investi-
gated next. The hydroxyl group can act as either a
hydrogen bond donor (via intermolecular interactions
or intramolecular interaction with the ketone) or as a
Table 2. In vitro screening data for SGLT inhibitory activity and selectivity
OH
O
R3
R2
O
OH
OH
O
OH OH
Entry
R2
Et
R3
SGLT2 Kia (lM)
SGLT1 Kia (lM)
Ratio (SGLT1/SGLT2)
361.1
9g
0.018 0.005
6.50 1.42
O
O
9h
9i
9j
Et
0.015 0.004
0.034 0.007
0.150 0.021
6.28 0.99
8.39 2.01
8.59 0.21b
418.6
246.7
57.3b
Et
OEt
i-Pr
O
O
9k
9l
i-Pr
i-Pr
Cl
0.029 0.004
0.110 0.009
0.031 0.018
16.1 3.27
33.1 4.40
4.78 0.05
554
301
OEt
9m
154.2
O
O
9n
9o
Cl
Cl
0.029 0.010
0.087 0.008
3.75 0.45
4.64 0.25
129.3
53.3
OEt
9p
F
0.071 0.011
3.69 0.21
51.9
a Ki values are mean SEM.
b Maximal efficacy could not be achieved for compound 9j in this assay due to poor substrate solubility at higher concentrations.

5124
J. Dudash, Jr. et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5121–5125
OH
O
O
OH
O
O
OBn O
O
O
O
O
O
O
c
a, b
O
O
OH
OH
OH
OH
OH
OH
O
O
O
OH
OH OH
O
OH
12
H3CO
H3CO
O
O
9b
13
Scheme 2. Reagents: (a) BnBr, K2CO3, acetone; (b) methyl chloroformate, collidine; (c) Pd/C, H2 (30psi), EtOH.
hydrogen bond acceptor. It has been reported that con-
version of the hydroxyl group to a methoxy group re-
duced the inhibitory activity in vivo, while removal of
the hydroxyl group eliminated all in vivo activity.5c
Introduction of a chloro-group, which has no hydrogen
bond donor character but greater hydrogen bond accep-
tor character than a methoxy-group, led to analogues
10a–c with potency in the micromolar range. Introduc-
tion of an amino-group at the 60-position maintained
the hydrogen bond donor character and added greater
hydrogen bond acceptor character. This modification
led to analogues 11a–c that were less potent and less
selective than compounds 9a–c. Additional analogues
that are designed to address the role of substituents at
this position are in progress.
by intravenous administration, which was about 2-fold
more efficacious than compound 9b. Both compounds
9b and 13 induced only marginal urinary glucose excre-
tion at doses of 100mg/kg with oral administration,
probably due to poor drug exposure, as both com-
pounds 9b and 13 showed no bioavailability in standard
rat PK studies.
In summary, we have developed a series of SGLT inhib-
itors to further evaluate the SAR of compound 1a.
Modification of the benzofuran moiety and 40-substitu-
ent of the phenyl ring improved selectivity at SGLT2
with a slight decrease in potency, and increased human
microsomal stability compared to compound 1a. In
0
addition, these studies showthat the 6 -hydroxyl group
cannot be replaced with a chloro- or amino-group. The
SAR derived from these studies could be used in the
search for a novel series of SGLT2 inhibitors. Progress
will be reported in due course.
The role of substitution at the 40-position was examined
with a series of compounds that encompassed alkyl and
halogen groups (Table 2). Increasing the size of the sub-
stituent from methyl to ethyl or isopropyl gave com-
pounds 9g–l, which were uniformly less potent than
9a–c but more selective for inhibition of SGLT2. The
benzodioxane analogue 9k was the most selective with
a subtype Ki ratio of 554, and it should be noted that
the benzodioxane ring system maintained potency in this
series across a range of changes in other regions.
References and notes
1. Deetjen, P.; Baeyer, H. V.; Drexel, H. Textbook of
Nephrology, 3rd ed.; Massry, S. G., Glassock, R. J.,
Eds.; Williams & Wilkins: Baltimore, 1995; Vol. 1, pp
90–94.
2. Yano, H.; Seino, Y.; Imura, H. Diabetes Front. 1990, 1,
417.
3. Ferraris, R. P. Biochem. J. 2001, 360, 265.
4. (a) Review: McKee, F. W.; Hawkins, W. B. Physiol. Rev.
1945, 25, 255; (b) Alvarado, F.; Crane, R. K. Biochim.
Biophys. Acta. 1962, 56, 170; (c) Toggenburger, G.;
Kessler, M.; Semenza, G. Biochim. Biophys. Acta. 1982,
688, 557.
5. (a) Tsujihara, K.; Hongu, M.; Saito, K.; Kawanishi, H.;
Kuriyama, K.; Matsumoto, M.; Oku, A.; Ueta, K.; Tsuda,
M.; Saito, A. J. Med. Chem. 1999, 42, 5311; (b) Hongu,
M.; Funami, N.; Takahashi, Y.; Saito, K.; Arakawa, K.;
Matsumoto, M.; Yamakita, H.; Tsujihara, K. Chem.
Pharm. Bull. 1998, 46, 1545; (c) Hongu, M.; Tanaka, T.;
Funami, N.; Takahashi, Y.; Saito, K.; Arakawa, K.;
Matsumoto, M.; Yamakita, H.; Tsujihara, K. Chem.
Pharm. Bull. 1998, 46, 22; (d) Tsujihara, K.; Hongu, M.;
Saito, K.; Inamasu, M.; Arakawa, K.; Oku, A.; Matsu-
moto, M. Chem. Pharm. Bull. 1996, 44, 1174.
6. (a) Ohsumi, K.; Matsueda, H.; Hatanaka, T.; Hirama, R.;
Umemura, T.; Oonuki, A.; Ishida, N.; Kageyama, Y.;
Maezono, K.; Kondo, N. Bioorg. Med. Chem. Lett. 2003,
13, 2269; (b) Nishimura, T.; Fujikura, H.; Fushimi, N.;
Tatani, K.; Katsuno, K.; Isaji, M. WO03/000712, 2003;
Chem. Abstr. 2003, 138, 49945; (c) Nishimura, T.; Fush-
imi, N.; Fujikura, H.; Katsuno, K.; Komatsu, Y.; Isaji, M.
The electronic character of the 40-substituent was also
investigated. The electron-withdrawing chloro-group
was introduced in analogues 9m–o. These compounds
were comparable in potency to the isopropyl series 9j–l,
but were not as selective, perhaps due to the decrease
in steric bulk. Following this trend, the smaller 40-
fluoro-substituent 9p had comparable potency to the
ethyl analogue 9g, but showed even less selectivity for
inhibition of SGLT2.
Select compounds were tested for human liver micro-
somal stability, indicative of a compoundÕs first-pass lia-
bility.11 While compound 1a showed a metabolic half-
life of 24min, compounds 9b, 9c and 9i all exhibited
microsomal stability greater than 45min. Compound
9b and its methyl carbonate prodrug 13 (Scheme 2) were
evaluated for their ability to induce urinary glucose
excretion in male Zucker Diabetic Fatty (ZDF) rats.12
When a 3mg/kg dose was administered intravenously,
compound 9b induced excretion of 608mg of urinary
glucose, indicating this compound had inhibitory activ-
ity at SGLT2 in vivo. However, compound 1a induced
excretion of 1189mg of urinary glucose at the same dose

J. Dudash, Jr. et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5121–5125
5125
WO02/068439; Chem. Abstr. 2002, 137, 232854; (d)
Fujikura, H.; Fushimi, N.; Nishimura, T.; Tatani, K.;
Katsuno, K.; Hiratochi, M.; Tokutake, Y.; Isaji, M.
WO01/68660; Chem. Abstr. 2003, 135, 242456; (e) Wash-
burn, W. N.; Ellsworth, B.; Meng, W.; Wu, G.; Sher, P.
M. US03/0114390; Chem. Abstr. 2003, 139, 36736.
cific for sodium-dependent glucose transporters. After
2h the labelled cells were washed three times with ice-
cold PBS. Cells were then solubilized and Na-dependent
14C-AMG uptake was quantified by measuring
radioactivity.
11. Tests were conducted at Absorption Systems Exton PA.
Incubation at 37°C with test compound at 5lM and 1mg
protein/mL human liver microsomal prep. Half life was
determined by measuring the percent parent compound
remaining. In this system, values above 30min were
considered acceptable.
7. Unwanted side effects due to SGLT1 inhibition may
include flatulence and bloating, a consequence of altered
carbohydrate absorption from the intestine. It is also
possible that SGLT1 inhibition may have unwanted effects
in target tissues other than the intestine. SGLT1 expres-
sion has been reported to be very high in the heart. The
functional consequence of SGLT1 expression in the heart
remains unknown. See: Zhou, L.; Cryan, E. V.; DÕAndrea,
M. R.; Belkowski, S.; Conway, B. R.; Demarest, K. T.
J. Cell. Biochem. 2003, 90, 339.
8. Eiden, F.; Patzelt, G. Archiv. Pharm. 1985, 318, 328.
9. All compounds provided satisfactory spectral data (1H
NMR, LCMS) and were homogeneous by TLC.
10. Cell-based assay for sodium-dependent glucose transport:
Cell-based functional screens were conducted to assess
SGLT inhibitors. CHO-K1 cells overexpressing human
SGLT2 or SGLT1 were used. Cells were treated with
compound in the absence or presence of NaCl for 15min.
Cells were then labelled with 14C-a-methylglucopyrano-
side (AMG)––a nonmetabolizable glucose analogue spe-
12. Male Zucker Diabetic Fatty (ZDF) rats (7–8weeks) were
obtained from Charles River. Animals were maintained on
a 12-h light/dark cycle in a temperature-controlled room.
Animals were given ad libitum access to food (standard
rodent diet Purina 5008) and water. Animals were fasted
for 12h prior to initiation of the experiment. On the
morning of the experiment, animals were administered
vehicle (0.5% methylcellulose) or compound by oral
gavage (2mL/kg). For intravenous dosing, animals
received either vehicle (10% Solutol) or compound
(2mL/kg). After 1h, animals received an oral glucose
challenge (4mL/kg of 50% solution) and were immediately
placed in metabolism cages. Animals were given free
access to water and urine was collected for 4h. Urinary
glucose was quantified using the Trinder reagent (Sigma).