Development of disulfide-derived fructose-1,6-bisphosphatase (FBPase) covalent inhibitors for the treatment of type 2 diabetes

Article abstract of DOI:10.1016/j.ejmech.2020.112500

Fructose-1,6-bisphosphatase (FBPase), as a key rate-limiting enzyme in the gluconeogenesis (GNG) pathway, represents a practical therapeutic strategy for type 2 diabetes (T2D). Our previous work first identified cysteine residue 128 (C128) was an important allosteric site in the structure of FBPase, while pharmacologically targeting C128 attenuated the catalytic ability of FBPase. Herein, ten approved cysteine covalent drugs were selected for exploring FBPase inhibitory activities, and the alcohol deterrent disulfiram displayed superior inhibitory efficacy among those drugs. Based on the structure of lead compound disulfiram, 58 disulfide-derived compounds were designed and synthesized for investigating FBPase inhibitory activities. Optimal compound 3a exhibited significant FBPase inhibition and glucose-lowering efficacy in vitro and in vivo. Furthermore, 3a covalently modified the C128 site, and then regulated the N125–S124–S123 allosteric pathway of FBPase in mechanism. In summary, 3a has the potential to be a novel FBPase inhibitor for T2D therapy.

Full text of DOI:10.1016/j.ejmech.2020.112500

Journal Pre-proof
Development of Disulfide-derived Fructose-1,6-bisphosphatase (FBPase) Covalent
Inhibitors for the Treatment of Type 2 Diabetes
Yi-xiang Xu, Yun-yuan Huang, Rong-rong Song, Yan-liang Ren, Xin Chen, Chao
Zhang, Fei Mao, Xiao-kang Li, Jin Zhu, Shuai-shuai Ni, Jian Wan, Jian Li
PII:
S0223-5234(20)30472-4
DOI:
https://doi.org/10.1016/j.ejmech.2020.112500
EJMECH 112500
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 28 December 2019
Revised Date: 13 May 2020
Accepted Date: 25 May 2020
Please cite this article as: Y.-x. Xu, Y.-y. Huang, R.-r. Song, Y.-l. Ren, X. Chen, C. Zhang, F. Mao, X.-
k. Li, J. Zhu, S.-s. Ni, J. Wan, J. Li, Development of Disulfide-derived Fructose-1,6-bisphosphatase
(FBPase) Covalent Inhibitors for the Treatment of Type 2 Diabetes, European Journal of Medicinal
Chemistry, https://doi.org/10.1016/j.ejmech.2020.112500.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
©
2020 Elsevier Masson SAS. All rights reserved.

Development of Disulfide-derived Fructose-1,6-bisphosphatase
FBPase) Covalent Inhibitors for the Treatment of Type 2 Diabetes
(
a,
1
b,
1
b,
1
b
a
Yi-xiang Xu, Yun-yuan Huang, Rong-rong Song, Yan-liang Ren, Xin Chen,
a
a
a
a
c,
b,
Chao Zhang, Fei Mao, Xiao-kang Li, Jin Zhu, Shuai-shuai Ni, * Jian Wan, * and
a,
Jian Li *
a
State Key Laboratory of Bioreactor Engineering, Shanghai Key Laboratory of New
Drug Design, School of Pharmacy, East China University of Science and Technology,
1
30 Mei Long Road, Shanghai 200237, China
b
Key Laboratory of Pesticide & Chemical Biology (CCNU), Ministry of Education,
College of Chemistry, Central China Normal University, Wuhan 430079, China
c
Cancer Institute, Longhua Hospital Shanghai University of Traditional Chinese
Medicine, 725 South Wan Ping Road, Shanghai 200032, China.
1
These authors contributed equally to this work.
*
To whom correspondence should be addressed jianli@ecust.edu.cn,
jianwan@mail.ccnu.edu.cn, nss1106@126.com.

Abatract
Fructose-1,6-bisphosphatase (FBPase), as a key rate-limiting enzyme in the
gluconeogenesis (GNG) pathway, represents a practical therapeutic strategy for type 2
diabetes (T2D). Our previous work first identified cysteine residue 128 (C128) was an
important allosteric site in the structure of FBPase, while pharmacologically targeting
C128 attenuated the catalytic ability of FBPase. Herein, ten approved cysteine
covalent drugs were selected for exploring FBPase inhibitory activities, and the
alcohol deterrent disulfiram displayed superior inhibitory efficacy among those drugs.
Based on the structure of lead compound disulfiram, 58 disulfide-derived compounds
were designed and synthesized for investigating FBPase inhibitory activities. Optimal
compound 3a exhibited significant FBPase inhibition and glucose-lowering efficacy
in vitro and in vivo. Furthermore, 3a covalently modified the C128 site, and then
regulated the N125-S124-S123 allosteric pathway of FBPase in mechanism. In
summary, 3a has the potential to be a novel FBPase inhibitor for T2D therapy.
Key words: Diabetes; FBPase; Disulfiram; Covalent inhibitors; Drug repurposing

1
. Introduction
Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder that
characterized by the elevated levels of blood glucose (hyperglycemia) [1]. Prolonged
hyperglycemia leads to various diabetic complications, such as retinopathy,
nephropathy, neuropathy, and cardiovascular diseases, which are the major causal
factors for mortality in patients with diabetes [2-6]. The report from the International
Diabetes Federation (IDF) disclosed that the number of people of ages 20-79 affected
by diabetes has reached a total of 415 million in 2017 and is expected to rise to 629
million by 2045 worldwide [7]. Insulin resistance and relative insulin insufficiency
are one of the main causes for hyperglycemia [8-10]. In view of this, current
antidiabetic drugs focus on decreasing the blood glucose levels of T2DM patients as
close to those of normal ones as possible via relieving insulin resistance, augmenting
insulin secretion, or directly decreasing the blood glucose [11, 12]. Moreover, many
studies revealed that the hepatic GNG is overactivated continuously in T2DM patients,
which leads to fasting and postprandial hyperglycemia [13, 14]. Therefore, targeting
the GNG process for directly regulating the level of blood glucose represents a logical
and promising strategy for the development of antidiabetics, which provides an
important complementary for the research and discovery of antidiabetic drugs.
FBPase is a key rate-limiting enzyme in the GNG pathway, which catalyzes the
hydrolysis of fructose-1,6-bisphosphate transfering to both fructose-6-phosphate and
inorganic phosphate. This special functional position of FBPase in the GNG pathway
endows it with powerful attractiveness as a potential drug target, as inhibiting FBPase
can block the conversion of 3-carbon precursors to glucose without affecting
glycogenolysis, glycolysis, and the tricarboxylic acid cycle directly [15]. In addition,
some FBPase inhibitors have been confirmed to decrease the blood glucose levels in
both T2DM animal models and patients [16-21], suggesting that FBPase inhibitors
have the potential to be a novel class of antidiabetics.
Structurally, FBPase is a homotetramer type protein that co-regulated by two
endogenous indicators fructose 2,6-bisphosphate (substrate analogue of FBPase) and
adenosine monophosphate (AMP) [22]. At present, three types of FBPase inhibitors

have been reported (Fig. S1), which are divided into competitive inhibitors that
interacting with the substrate site including arabinose 1,5-bisphosphate [23],
uncompetitive inhibitors that interacting with subunit interface site including
anilinoquinazoline and quinoline derivatives [24,25], and non-competitive inhibitors
that interacting with the AMP allosteric site including non-phosphorus-based
inhibitors [26-33] and phosphorus-based AMP mimics [16-19, 34-37]. Among them,
two AMP mimics CS-917 and MB07803 have entered into the clinical studies of
diabetes mellitus [20, 21]. However, current efforts to explore FBPase inhibitors as
antidiabetic drug candidates were still lack of breakthrough progress, while directly
targeting AMP site was also difficult to develop the sufficiently therapeutic strategy
because of its high dependence on electrostatic binding interactions and functional
modulation of other enzymes involved in important anabolic and catabolic pathways
[
18]. Therefore, further researches for discovering novel efficient target of FBPase
and related inhibitors retain important significance.
Fig. 1. Structures of CS-917 and MB07803.
Our previous study has identified a new allosteric site (C128) on FBPase that
covalently modified and further attenuated the bind affinity of substrate FBP with
FBPase. Based on the results above, several nitrostyrene-derived compounds were
developed as novel FBPase covalent inhibitors by using an electrophilic
fragment-based screen strategy, which confirmed that targeting the allosteric site
C128 can attenuate the catalytic function of FBPase effectively, thereby inhibiting the
GNG pathway in vitro [38]. However, the drawbacks including poorly antidiabetic
effect in vivo and lowly structural modification potentiality, limited the further
exploration of these compounds, whereas drug repurposing as a practical strategy

overcomes those inferiorities effectively. In view of this, we first screened a series of
known cysteine covalent drugs via an enzymatic assay to explore novel FBPase
inhibitors, which has the potential to covalently bind to the C128 site in mechanism. A
drug approved for relieving alcohol dependence, disulfiram (01), was identified as a
good FBPase inhibitor in the assay. Based on the structure of lead compound
disulfiram, we designed and synthesized a series of disulfiram analogues. In particular,
derivative 3a not only showed significant FBPase inhibitory activity in vitro and in
vivo, but exhibited moderate glucose-lowering efficacy in ICR mice model and
diabetic db/db mice model as well. In addition, pharmacological mechanism
researches demonstrated that 3a might covalently modify the C128 site and suppress
the FBPase activity through regulating the N125-S124-S123 allosteric pathway, which
was consistent with the allosteric mechanism reported by our previous studies. These
results suggested that the C128 site of FBPase was a potential drug target for the
development of antidiabetic drugs.
2
. Results and discussion
2
.1. Design and synthesis
In this study, ten known old cysteine covalent drugs, including mercaptopurine,
furazolidone, disulfiram, carmofur, lipoic acid, afatinib, phosphonomycin, eflornithine,
warfarin, and phenoxybenzamine, were first selected to screen the FBPase inhibitory
activities in an enzyme-based assay. The alcohol deterrent disulfiram was found to
inhibit FBPase activity at a single digit micromolar IC value (IC = 1.78 μM, Fig.
5
0
50
S2, Supporting Information), while the glucose-lowering efficacy of disulfiram had
been confirmed by the previous research (its antidiabetic mechanism failed to be
elucidated) [39], suggesting that disulfiram definitely deserves to be further
investigated as a lead compound in the domain of medicinal chemistry. To obtain
more potent FBPase inhibitors and explore the structure-activity relationship (SAR) of
the disulfiram analogs, chemical modifications were performed in four series: (1) To
investigate the various skeletons responsible for FBPase inhibition, we removed or
replaced the linker that contained the disulfide and thioamide groups in the structure
of disulfiram, and produced the derivatives of series I (1a-e). (2) By replacing the

symmetrical terminal N-ethyl substituents of disulfiram with various steric and polar
alkyl, aromatic, heterocyclic, or aromatic heterocyclic groups, twenty-two compounds
(2a-v) were designed and synthesized as series II to explore the inhibitory effects. (3)
The thioacetamide moiety of disulfiram was cyclized to enrich the scaffold types in
this study and yielded the symmetric compound 3a and its analogues (compounds
3
a-n, series III). (4) The asymmetric disulfide derivatives (compounds 4a-q, series IV)
were designed and synthesized based on the structure of series III.
Fig. 2. Structural modification strategies and four series of compounds derived from
lead compound 01.
The synthesis of analogues 1a-e was shown in Scheme 1, and the procedure
generally followed the method reported by Kapanda [40]. Diethylamine, carbon
disulfide, and triethylamine were reacted to generate diethylcarbamodithioate, which
was
diethylcarbamothioic chloride to obtain compounds 1a, 1c, and 1e, respectively. Then,
c was sulfured to yield compound 1d, while compound 1b was synthesized through
treated
with
diiodomethane,
2-chloro-N,N-diethylacetamide,
and
1
the oxidation of disulfiram. The synthesis of target compounds 2a-v was started using
commercially available amino materials and was shown in Scheme 2. The synthesis
technique generally followed the modified method reported by Liu [41].

Scheme 1. Synthetic Route to Generate Compounds 1a-e. Reagents and conditions: (a)
carbon disulfide, triethylamine, acetonitrile, rt, 16 h; (b) diiodomethane, acetonitrile,
rt, 12 h; (c) N,N-diethylchloroacetamide, acetonitrile, rt, 12 h; (d) P S / Al O , 1,4 -
4
10
2
3
dioxane, reflux, 1 h; (e) diethylthiocarbamoyl chloride, acetonitrile, rt, 12 h; (f)
NaClO (5% in water), acetonitrile, rt, 12 h.
Scheme 2. Synthetic Route to Generate Compounds 2a-v. Reagents and conditions: (a)
(
i) CS , KOH, THF, rt, 12 h; (ii) I in MeOH, rt, 1 h.
2 2
Compounds 3a-d and 3k-l were synthesized by the oxidation of commercially

available thiols in the condition of H O or I (Scheme 3). Compounds 3e-g were
2
2
2
obtained commercially, and compounds 3h-j were obtained from our in-house library
41]. The synthetic procedures of the asymmetric disulfide derivatives 4a-n were
[
illustrated in Scheme 4. Most of the derivatives were synthesized by the nucleophilic
substitution of thiazole, benzothiazole, and thiadiazole (or substituted thiadiazole)
thiol with substituted arenesulfenyl chloride compounds in ethyl ether under very mild
condition [42]. Compounds 4a-b and 4n-o were synthesized via DDQ oxidation, and
compound 4e-f was synthesized by the Ohio-disulfide interchange reaction [43, 44].
Scheme 3. Synthetic Route to Generate Compounds 3a-n. Reagents and conditions:
(
a) H O , MeOH, rt, 0.5 h; or I , DCM, rt, 1 h; (b) HNO , Cu, HCl/H O, -5 °C to
2 2 2 2 2
b
c
5
5 °C, 3 h. Obtained commercially. Obtained from our in-house library.

Scheme 4. Synthetic Route to Generate Compounds 4a-q. Reagents and conditions:
(
a) DDQ, DCM, 0 °C, 1 h; (b) SO Cl , DCM, 0 °C, 1 h; (c) Et O, rt, 12 h; (d) CHCl ,
2 2 2 3
rt, 0.5 h.
In total, fifty-eight disulfide derivatives (1a-e, 2a-v, 3a-n, and 4a-q) were
designed and synthesized. Chemical structures are shown in Tables 1-2. These
derivatives were synthesized according to the routes outlined in Schemes 1-4, and the
1
synthetic procedures and characterization techniques ( H NMR, HRMS spectroscopy,
1
3
and C NMR of representative compounds) were used to confirm the structures of the
target compounds described in the experimental section. All target compounds were
confirmed to have > 95% HPLC purities.
2
.2. Inhibitory activity of human liver FBPase in vitro
To evaluate the T2DM therapeutic efficacy of these disulfide derivatives,
inhibitory activities against the recombinant human liver FBPase were determined,
while AMP, the natural allosteric inhibitor to FBPase, was used as the reference
compound. The related IC values or the inhibitory ratio of tested derivatives at the
50
concentration of 10 μM are shown in Table 1-2. Initial studies are involved in

modifying the central disulfide bond (linker) and the two characteristic thioamides in
the structure of lead compound 01. The results are summarized in Table 1. In general,
the replacement of disulfide bond with various mimic groups remarkably decreased
FBPase inhibitory activities (compounds 1a-e), while the replacement of substituents
at N-atom of thioamides with suitable building groups such as various alkyl,
cycloalkyl, aromatic, heterocyclic, or aromatic heterocyclic groups also had obvious
effects on the FBPase inhibitory activity (compounds 2a-v). Moreover, compared with
the derivatives with the replacement of bulk groups, those derivatives with small
steric substituents (2a, 2h-j, and 2l) presented superior the FBPase inhibition.
Especially, the inhibitory activity of compound 2j (IC = 0.13 μM) was elevated by
5
0
1
0-fold compared to that of 01 (IC = 1.78 μM).
50
Next, the thioacetamide moiety of disulfiram was fused to yield 3a based on the
strategy of scaffold hopping, which unexpectedly exhibited superior FBPase
inhibitory activity (IC50 = 0.22 μM). In view of this, further modification was
performed for increasing the FBPase inhibition based on the structure of hit
compound 3a. As illustrated in Table 2, the replacement of the thiazole moiety with
thiophene, 1-methylimidazole, cyclohexyl and phenyl groups decreased the potency
obviously (3a vs 3c-g), whereas the introduction of a nitrogen atom in the thiazole
moiety was tolerated (IC = 0.56 μM for 3b), suggesting that the nitrogen atom and
5
0
sulfur atom might play a crucial role in the binding mode of compounds with FBPase.
Fusing an aryl group or a cycloalkyl group into the thiazole skeleton did not affect the
inhibitory activity of the compound against FBPase (3a vs 3h-j). Moreover, a similar
tendency that bulk groups decreased the potency could be seen in the analogues of 3h,
meaning that the introduction of electron-withdrawing or electron-donating groups
weakened the potency of 3h (3m-n vs 3h). Considering that compounds 3a-b and 3h
displayed superior inhibitory effects against FBPase, thiazole, benzothiazole, and
thiadiazole (or substituted thiadiazole) thiol were selected to generate the asymmetric
disulfide derivatives by reacting with the alkyl or substituted phenylthiol groups.
Notably, this series of compounds displayed good FBPase inhibitory activities (IC =
5
0
0
.5~8 μM, except for 4a), but there was no obvious relationship between the

inhibitory activity and the types of substituents at benzene ring. Especially, compound
j inherited the inhibitory activity of 3a (IC = 0.6 μM for 4j).
4
5
0

Table 1
Structures and FBPase Inhibitory Activities of Compounds 1a-e, 2a-v.
a
a
HL (IC , μM)
HL (IC , μM)
5
0
50
Compd.
X
R1
Compd.
X
R1
(
inhibition at 10 μM)
(inhibition at 10 μM)
0
1
1
a
1.78 ± 0.38
2i
8.78 ± 0.60
N
> 10 (12.45%)
> 10 (2.59%)
> 10 (3.22%)
2j
2k
2l
0.13 ± 0.01
> 10 (29.36%)
1.20 ± 0.20
1
b
1
c

1
d
> 10 (21.35%)
> 10 (17.98%)
2.69 ± 0.18
2m
2n
2o
2p
2q
2r
> 10 (36.17%)
> 10 (39.94%)
> 10 (39.74%)
> 10 (47.59%)
> 10 (26.08%)
> 10 (26.25%)
1
e
2
a
2
b
> 10 (13.56%)
> 10 (47.79%)
> 10 (35.73%)
2
c
2
d

2
e
> 10 (32.40%)
2s
> 10 (47.62%)
2
f
> 10 (3.09%)
> 10 (28.52%)
2t
2u
2v
> 10 (37.72%)
> 10 (9.87%)
> 10 (38.38%)
2
g
2
h
2.28 ± 0.24
2.30 ± 0.10
N
AMP
a
HL, human liver FBPase; results are expressed as the mean ± SD of at least three independent experiments.

Table 2
Structures and FBPase Inhibitory Activities of Compounds 3a-n, 4a-q.
a
a
HL (IC , μM)
HL (IC , μM)
5
0
50
compd.
R2
R3
compd.
4c
R2
R3
(
inhibition at10 μM)
(inhibition at 10 μM)
3
a
0.22 ± 0.04
2.05 ± 0.19
3
b
0.56 ± 0.07
4d
2.05 ± 0.10
7.60 ± 0.34
3.79 ± 0.12
4.44 ± 0.60
2.50 ± 0.24
12.23 ± 1.70
3
c
4e
(50.61%)
3
d
> 10 (9.65%)
> 10 (12.39%)
> 10 (10.37%)
4f
3
e
4g
3
f
4h

3
g
> 10 (17.98%)
0.31 ± 0.09
0.30 ± 0.03
0.47 ± 0.04
0.49 ± 0.05
4i
4j
6.64 ± 0.45
0.60 ± 0.08
1.25 ± 0.15
2.52 ± 0.07
2.20 ± 0.06
3
h
3
i
4k
4l
3
j
3
k
4m
[45]
3
l
3.23 ± 0.63
4n
1.55 ± 0.15

3
m
> 10 (44.66%)
> 10 (18.45%)
4o
2.68 ± 0.45
3
n
4p
4q
1.29 ± 0.10
1.94 ± 0.37
4
a
> 10 (34.34%)
5.39 ± 0.38
4
b
a
HL, human liver FBPase; Results are expressed as the mean ± SD of at least three independent experiments.

Table 3
Cytotoxicity of 2j, 3a-b, 3h-k, and 4j on Human Hepatocellular Carcinoma Cell (HepG2) and Glucose-Lowering Effects in Fasted ICR mice.
a
b
a
b
compd
HepG2 (IC , μM)
glucose-lowering (%)
time , h
compd
HepG2 (IC , μM)
glucose-lowering (%)
time , h
5
0
50
2
j
> 100
75.08 ± 1.42
> 100
35
54
32
23
6
4
4
6
3i
3j
3k
4j
20.08 ± 2.15
> 100
31
50
26
31
6
8
8
6
3
a
3
b
h
> 100
3
> 100
64.54 ± 7.12
a
b
Results are expressed as the mean ± SD of at least three independent experiments.
Time of maximum glucose lowering.

2
.3. SAR studies
As shown in Table 1-2, the results revealed the noteworthy observations of the
SARs for compounds 1a-e, 2a-v, 3a-n, and 4a-q: (1) As for disulfide derivatives
containing thioamides, the disulfide linker and thioamides were indispensable
pharmacophores for the FBPase inhibitory activity, while the replacement or
modification of these groups decreased the inhibition of derivatives dramatically
(01 > 1a-e). (2) There was an extremely rigid requirement for the steric hindrance of
the symmetrical terminal substituents at N-atom of thioamides, whereas a smaller
substituent is tolerable (2a, 2h-j, 2l > 2b-g, 2k, 2m-v). (3) The disulfide derivatives
with thiazole group were synthesized by the cyclization of the thioacetamide moiety
of disulfiram, and most of the derivatives showed good inhibitory activities (3a-b and
3
h-l). Among these disulfide derivatives, N, S-containing aromatic rings showed
better inhibitory activity against FBPase than only N-containing aromatic rings or
S-containing aromatic rings (3a-b > 3c-d). (4) The asymmetric disulfide derivatives
with thiazole, benzothiazole and thiadiazole (or substituted thiadiazole) inherited the
FBPase inhibitory activities of corresponding symmetric disulfide derivatives. In total,
due to the good performance achieved by the structural modification, we obtained
eight derivatives (2j, 3a-b, 3h-k, and 4j) with significant inhibitory activities at the
rang of hundred nanomolar for further biological study.
2
.4. Preliminary glucose-lowering effects in vivo
We first introduced the ICR mice model for preliminary investigating the
glucose-lowering effects of the aforementioned compounds, which are widely applied
in the pharmacology researches including anti-diabetes [46-48]. Given that GNG is
the main source of endogenous glucose after fasting for 12 h in a normal mouse,
which provides the maximum sensitivity to the effects of a GNG inhibitor [34], the
mice were fasted for approximately 12 h before the intraperitoneal administration of
tested compound (50 mg/kg for 2j, 3h–k, and 4j; 25 mg/kg for 3a–b). As shown in
Table 3, most of the disulfide compounds elicited a significant decrease (> 30%) in
the fasting plasma glucose levels. In particular, 3a reduced the blood glucose by 54%,

and the concentration of glucose reached 2.9 mM at 4 h (Fig. S3, Supporting
Information). We further investigated the glucose-lowering effect of compound 3a by
oral administration, but 3a failed to show enough efficacy in fasting ICR mice (Fig.
S4, Supporting Information), thus intraperitoneal administration was adopted for
further antidiabetic effect study in vivo. In addition, the cytotoxicity evaluation of
eight compounds (2j, 3a-b, 3h-k, and 4j) was performed in HepG2 by using a Cell
Counting Kit-8 assay. As shown in Table 3, the majority of the tested compounds
displayed a low cytotoxicity after their incubation with the cells for 48 h (IC > 50
5
0
μM). Therefore, based on good FBPase inhibitory activity (IC = 0.22 μM), satisfied
5
0
glucose-lowering effect in vivo (54% reduction in blood glucose at 25 mg/kg), and
low cytotoxicity (IC50 = 75.08 μM in HepG2), compound 3a was chosen for the
further biological evaluation and covalent interaction mechanism study.
2
.5. Glucose–lowering effects in hepatocytes
Compound 3a displayed a significant pharmacodynamic effect in the ICR mice
that encouraged us to confirm whether the glucose-lowering effect was caused by
GNG suppression. Considering that GNG is one of the endogenous glucose
production pathways in the liver, FBPase inhibitors theoretically decrease the glucose
output in hepatocytes by suppressing the GNG pathway. Therefore, we first performed
a glucose production assay in primary mouse hepatocytes for evaluating the GNG
inhibition ability of compound 3a. As illustrated in Fig. 3A, 3a displayed the
dose-dependent inhibition of the glucose output after the incubation with mouse
primary hepatocytes for 6 h. The maximal glucose production inhibition (~90%) was
achieved at 100 μM. Moreover, Fig. 3B showed that the viability of cells with the
treatment of 3a was over 90%, indicating that 3a had little cytotoxicity in the mouse
primary hepatocytes. These results suggested that the reduced glucose production in
the primary hepatocytes might originate from the GNG inhibition by derivative 3a
rather than the structural cytotoxicity of 3a.

Fig. 3. Effect of compound 3a on the glucose output (A) and cytotoxicity (B) in
primary mouse hepatocytes after incubation for 6 h. Data are shown as the mean ±
SEM: *p < 0.05, **p < 0.01, ***p < 0.001 vs DMSO. Met: Metformin.
2
.6. GNG inhibition in ICR mice
Encouraged by the positive GNG inhibition results of 3a in the primary mouse
hepatocytes, we further investigated whether 3a could inhibit GNG in vivo. GNG
involves the production of glucose from non-carbohydrate precursors such as glycerol
and lactate during fasting. Therefore, inhibiting the FBPase activity logically leads to
the accumulation of blood lactate and decline of blood glucose levels (Fig. 4A). As
shown in Fig. 4B, the administration of 3a (25 mg/kg, i.p.) to 12 h-fasted ICR mice
resulted in the significant inhibition of the FBPase activity in the liver (p < 0.001 vs
vehicle), indicating that 3a not only could inhibit the activity of human FBPase, but
rodent FBPase, which is supported by the high amino acid sequence identity (84.9%)
of them (Fig. S5, Supporting Information). As expected, the lactate levels were
elevated but only modestly elevated (by < 2-fold, below 5 mM, the concentration of
lactic acidosis) in the 3a-treated group in comparison with that of vehicle group, while
the glucose level was decreased by 57%. Consequently, all the above-mentioned
results indicated that compound 3a reduced blood glucose levels by targeting FBPase
and inhibiting the GNG process and represented a promising lead compound for the
development of new antidiabetic drugs.

Fig. 4. Inhibition of GNG in the ICR mice. (A) Gluconeogenic pathway. (B) FBPase
activity of the liver freshly isolated from the ICR mice fasted for 12 h and treated with
3
a (25 mg/kg, i.p.) or the vehicle for 4 h. (C) Change in the gluconeogenic pathway
product: blood glucose. (D) Change in the gluconeogenic pathway precursor: blood
lactate. Data are shown as the mean ± SEM: *p < 0.05, **p < 0.01, ***p < 0.001 vs
vehicle, N = 6.
2
.7. Oral glucose tolerance test (OGTT) in the ICR mice and type 2 diabetic db/db
mice.
Considering that compound 3a could significantly inhibit GNG process in vitro
and in vivo and lower the glucose levels in the fasted ICR mice, compound 3a was
further evaluated its glucose tolerance capacity. An oral glucose tolerance test (OGTT)
was performed using ICR mice and type 2 diabetic db/db mice, and metformin, a
first-line drug for the treatment of diabetes which was usually used as positive control
for exploring the pharmaceutical effect of FBPase inhibitors in cellular and animal

model [18, 49], was used as a positive control. The 12 h-fasted ICR mice or db/db
mice were administered metformin (250 mg/kg, i.p.), 3a (12.5 mg/kg, i.p.), 3a (25
mg/kg, i.p.), and the vehicle 60 min before the oral administration of glucose (2 g/kg
for ICR mice, 1.5 g/kg for db/db mice). In the ICR mice, the glucose test showed that
the blood glucose levels dramatically increased in the control group, whereas
compound 3a at doses of 12.5 and 25 mg/kg significantly suppressed the blood
glucose levels, especially 30 min after glucose was administered (p < 0.001, Fig. 5A),
and reduced the blood glucose AUC0−120min by 24.2% and 22.7% compared with the
vehicle group (p < 0.05, Fig. 5B). The pharmacodynamic effect of 3a in the db/db
mice was weaker than that in the ICR mice at the same dose, which might be caused
by the elevated FBPase in the liver of db/db mice [50]. As shown in Fig. 5D, only
compound 3a at dose of 25 mg/kg significantly improved the glucose tolerance
capacity by decreasing the AUC0−120min by 10.4% compared with the vehicle group (p
<
0.05). Taken together, these data suggested that 3a could moderately improve
glucose homeostasis both in ICR mice and diabetic db/db mice. In addition, the
further long-term safety study indicated that compound 3a showed potential toxicity
at high dose (Fig. S6, Supporting Information).

Fig. 5. Effect of a single administration (i.p.) of compound 3a on the blood glucose
levels. (A) and (B) Blood glucose level changes in the OGTT and AUC of blood
drawn between 0 and 120 min in the ICR mice (n = 5, male). Compound 3a,
metformin, or the vehicle was administered to 12 h-fasted ICR mice 60 min prior to
the oral glucose test (2 g/kg). (C) and (D) Blood glucose level changes in the OGTT
and AUC of blood drawn between 0 and 120 min in the db/db mice (n = 6).
Compound 3a, metformin, or the vehicle was administered to the 12 h-fasted db/db
mice 60 min prior to the oral glucose test (1.5 g/kg). Data are presented as the mean ±
SEM: *p < 0.05, **p < 0.01, ***p < 0.001 vs vehicle. Student’s t test.
2
.8. Compound 3a inhibits FBPase by covalently binding to a cysteine residue
To further investigate the binding and inactivating steps of the FBPase inhibition
of 3a, the time-dependent inhibition (TDI) method was introduced to examine the
inhibition of FBPase by 3a [51]. The enzymatic activity was measured over time up to

4
5 minutes. As shown in Fig. 6, the preincubation of 3a apparently influenced the
IC values of the inhibitory activity against FBPase, which decreased from 0.28 to
5
0
0
.09 μM. The results suggested that covalent bond was formed between 3a and
FBPase, and 3a was an irreversible covalent inhibitor of FBPase.
Fig. 6. Time-dependent inhibition of FBPase by 3a.
The C128 residue is the only cysteine residue that affects the catalytic activity of
FBPase in our previous work [38]. Herein, we proceeded to perform a cysteine
mutation assay to confirm whether 3a targeted the C128 residue. On the premise of
retaining its activity maximally, all seven cysteine residues of FBPase were mutated
one by one. Seven mutants were generated, including C38S, C92S, C116S, C128S,
C179S, C183S, and C281S. The inhibitory activity of 3a against these mutants was
summarized in Table 4. All the mutants maintained the same enzymatic activity as the
wild-type FBPase meaning that these cysteine residues were not crucial for the
catalytic function of FBPase. Compound 3a largely lost its activity toward C128S
(
IC > 500 μM) but not to other six mutants. These data suggested that 3a exhibited
50
its FBPase inhibitory activity via covalently binding with the C128 residue.
Table 4
The IC of compound 3a against cysteine mutation and WT FBPase.
5
0
M
W
Vmax (U/mg)
4.0 ± 0.2
2.7 ± 0.2
1.6 ± 0.1
1.6 ± 0.2
2.7 ± 0.2
3a IC (μM) 3a (IC / IC 50)
50
50
WT
0.27 ± 0.06
0.31 ± 0.08
0.13 ± 0.02
0.19 ± 0.02
> 500
1
C38S
C92S
C116S
C128S
1.1
0.5
0.7
> 1850

C179S
C183S
C281S
2.3 ± 0.1
3.8 ± 0.2
3.6 ± 0.1
0.16 ± 0.02
0.21 ± 0.03
0.18 ± 0.02
0.6
0.8
0.7
a
Results are expressed as the mean ± SD of at least three independent experiments.
To further verify the covalent binding reaction between 3a and FBPase, we
conducted UV-visible spectroscopy experiments using the Ellman's reagent. In
general, the maximum absorption peak of either compound 3a or FBPase was
detected at a wavelength of 282 nm or 283 nm, respectively (Fig. 7A, B). In parallel,
we incubated 3a with FBPase for 5 min and investigated the changes in the absorption.
In contrast with the maximum absorption peaks observed for FBPase and compound
3
a, a new maximum absorption peak located at 306 nm was found after 3a was
incubated with FBPase (Fig. 7C). It was reassuring to note that this peak belonged to
thiazole-2-thiol (Fig. 7D), meaning that the disulfide bond of 3a was destroyed, and a
new covalent bond formed between FBPase and the half fragment of 3a. These data
indicated that 3a could covalently bind with the C128 residue in FBPase, resulting in
the strong inhibitory activity of 3a for FBPase.
Fig. 7. UV-visible spectra analysis. (A) Compound 3a. (B) FBPase. (C) FBPase

treated with 3a (D) Thiazole-2-thiol. Reaction system: Tris buffer (50 mM, pH 7.4),
0
.8 mM MgCl , 5 min.
2
2
.9. Covalent modification by 3a affects the catalytic activity of FBPase.
As abovementioned, our data demonstrated that compound 3a modulated the
enzymatic activity of FBPase by covalently binding to C128. To further explore the
covalent modification between 3a and C128, molecular docking was performed using
Autodock 4.0 and the X-ray crystallographic structure of FBPase (PDB ID: 4H46)
were obtained from the Protein Data Bank. Remarkably, the structure of 3a was
suitable to the pocket near the C128 residue (Fig. 8A), and the sulfur atom of the
disulfide group of 3a was located within a 3.9 Å distance of the sulfur atom of C128,
making it possible for a disulfide bond to form between them (Fig. 8B) [52]. A strong
hydrogen bond between the N-atom of 3a and C128-SH (1.9 Å) played an important
role in positioning and stabilizing the ligand inside the cavity of C128, which also
explained why 3a exhibited a better inhibitory potency than the compound 3c with a
substituted thiophene group (IC50 = 0.22 μM for 3a; IC = 12.23 μM for 3c). In
50
addition, the thiazole moiety of 3a established a face-to-face stacking π-π interaction
with the Y258 residue (4.3 Å). The docking study demonstrated a reasonable binding
mode between 3a and FBPase. However, it remained unclear how the covalently
modifying the residue by 3a affected the catalytic function of FBPase. The
abovementioned mutation assay results displayed that the C128S mutant maintained
the catalytic function of FBPase, meaning that the residue C128 failed to directly
affect the FBPase catalysis activity. Our previous study demonstrated that there exists
an N125-S124-S123 pathway (hydrogen bond network) between C128 site and the
catalytic site of FBPase (Fig. 8C) [38]. Therefore, we validly speculated that the
covalent binding of 3a with C128 affects the catalytic activity of FBPase through the
same N125-S124-S123 allosteric signaling pathway. The possible mechanism was
that the steric-hindrance effect between the sulfur atom of the thiazole group in 3a and
the carbonyl group of N125 (2.7 Å) led to N125 moving away from C128, destroying
the N125-S124-S123 conformation, and then S123 moved far from the catalytic site,
destroying the hydrogen bond between S123 and FBP which was necessary for

catalysis (Fig. 8D). To verify this mechanism, we performed the residue mutation near
the C128 residue, generating six mutants, namely S124A, N125A, D127A, R243A,
R254A, and Y258A As shown in Table 5, the inhibitory activity of 3a against N125A
mutant decreased by 926-fold compared to that of the WT, suggesting that N125A
played a crucial role in the N125-S124-S123 allosteric pathway and the covalent
modification of C128 residue affected FBPase catalytic activity via N125-S124-S123.
In addition, when the Y258A mutation was performed without the other mutations,
the IC value of 3a was elevated by 11.5-fold, demonstrating that the π-π interaction
5
0
between the thiazole moiety of 3a and Y258 benefited the binding stability.
Fig. 8. Binding mode of compound 3a and FBPase. (A) Surface representations of the
binding mode of compound 3a within the C128 pocket of FBPase. (B) 3D docking
model of 3a with FBPase (PDB ID: 4H46); (C) The R-state crystal structure of
FBPase and the residues nearby the C128 pocket (PDB ID: 1EYI); (D) Proposed
inhibitory mechanism of 3a against FBPase. Compound 3a is illustrated using blue
sticks and FBP using green sticks. The π-π stacking interactions are highlighted by
yellow dashed lines, the hydrogen bonds are shown with red dashed lines, and the
distance between two atoms is shown with black dashed lines. The oxygen atom is
represented in red, nitrogen atom in blue, and sulfur atom in yellow.
Table 5