Discovery of a novel inhibitor of NAD(P)+-dependent malic enzyme (ME2) by high-throughput screening

Article abstract of DOI:10.1038/aps.2013.189

Aim: Malic enzymes are oxidative decarboxylases with NAD⁺ or NAD(P)⁺ as cofactor that catalyze the conversion of L-malate to pyruvate and CO₂. The aim of this study was to discover and characterize a potent inhibitor of hu

Full text of DOI:10.1038/aps.2013.189

Acta Pharmacologica Sinica (2014) 35: 674–684
© 2014 CPS and SIMM All rights reserved 1671-4083/14 $32.00
www.nature.com/aps
npg
Original Article
Discovery of a novel inhibitor of NAD(P)+-dependent
malic enzyme (ME2) by high-throughput screening
2,
2,
Yi WEN1, #, Lei XU2, #, Fang-lei CHEN2, Jing GAO1, Jing-ya LI2, , Li-hong HU , Jia LI
*
*
*
1School of Pharmacy, Jiangsu University, Zhenjiang 212013, China; 2The State Key Laboratory of Drug Research, Shanghai Institute of
Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China
Aim: Malic enzymes are oxidative decarboxylases with NAD+ or NAD(P)+ as cofactor that catalyze the conversion of L-malate to pyruvate
and CO2. The aim of this study was to discover and characterize a potent inhibitor of human NAD(P)+-dependent malic enzyme 2
(ME2).
Methods: Recombinant human ME2-His-Tag fusion protein was overexpressed in E coli and purified with Ni-NTA resin. A high-
throughput screening (HTS) assay was developed to find ME2 inhibitors. Detergent Brij-35 was used to exclude false positives. The
characteristics of the inhibitor were analyzed with enzyme kinetics analysis. A thermal shift assay for ME2 was carried out to verify the
binding of the inhibitor with the enzyme.
Results: An HTS system for discovering ME2 inhibitors was established with a Z′ factor value of 0.775 and a signal-to-noise ratio (S/N)
of 9.80. A library containing 12 683 natural products was screened. From 47 hits, NPD387 was identified as an inhibitor of ME2. The
primary structure-activity relationship study on NPD387 derivatives showed that one derivative NPD389 was more potent than the
parent compound NPD387 (the IC50 of NPD389 was 4.63±0.36 μmol/L or 5.59±0.38 μmol/L, respectively, in the absence or presence
of 0.01% Brij-35 in the assay system). The enzyme kinetics analysis showed that NPD389 was a fast-binding uncompetitive inhibitor
with respect to the substrate NAD+ and a mixed-type inhibitor with respect to the substrate L-malate.
Conclusion: NPD389 is a potent ME2 inhibitor that binds to the enzyme in a fast-binding mode, acting as an uncompetitive inhibitor
with respect to the substrate NAD+ and a mixed-type inhibitor with respect to the substrate L-malate.
Keywords: malic enzyme; ME2; inhibitor; natural product; NPD389; high-throughput screening; structure-activity relationship; enzyme
kinetics; thermal shift assay
Acta Pharmacologica Sinica (2014) 35: 674–684; doi: 10.1038/aps.2013.189; published online 31 March 2014
amino acids. NADPH, which is necessary for the maintenance
Introduction
Malic enzymes (ME) are oxidative decarboxylases that cata- of intracellular glutathione, acts as a coenzyme in many bio-
lyze the conversion of L-malate to pyruvate and CO2 and use chemical reactions in vivo. In tumor cells, glutaminolysis is
NAD+ or NAD(P) + as a cofactor[1–3]. In mammals, there are upregulated to meet the needs of NADPH and pyruvate[5].
three malic enzyme isoforms with different cofactor specifici- ME2 catalyzes the conversion from malate to pyruvate and
ties: the cytosolic NADP+-dependent ME1, the mitochondrial NAD(P)H. Therefore, ME2 plays an important role in gluta-
NAD(P)+-dependent ME2, and the mitochondrial NADP+- minolysis[6, 7]
.
dependent ME3. Although ME2 can use either NAD+ or
Previous studies indicate that ME2 activity increases as cells
NADP+ as a cofactor, it has been demonstrated that this progress towards neoplasia in a rat tracheal epithelial line[8],
enzyme favors NAD+ as a cofactor under physiological condi- and similar findings have been reported for Morris hepato-
tions[4].
mas[9]. The knockdown of endogenous ME2 levels impairs the
Pyruvate is an intracellular hub in the conversion of three proliferation of K562 cells, leads to apoptosis in K562 cells and
major nutrient substrates: carbohydrates, fatty acids and suppresses tumor growth in vivo[4]. Silencing ME2 severely
impairs the growth of the HCT-116 and U2OS cancer cell lines.
# These authors contributed equally to this work.
To whom correspondence should be addressed.
E-mail jli@mail.shcnc.ac.cn (Jia LI);
lhhu@simm.ac.cn (Li-hong HU);
jyli@mail.shcnc.ac.cn (Jing-ya LI)
Received 2013-09-11 Accepted 2013-12-01
Furthermore, overexpression of ME2 enhances HCT-116 and
*
U2OS cell growth. It is known that ME2 is involved in the
regulation of p53 during tumor metabolism, senescence and
growth[6]. Thus, ME2 is believed to be a potential anticancer
target. In light of the scarcity of ME2 inhibitors[10], establishing

www.chinaphar.com
Wen Y et al
npg
675
a high-throughput screening system for the discovery of ME2 then stored at -70°C until use.
inhibitors may lay the foundation for novel anti-tumor drug
discovery.
pH optimization
In this study, we developed a molecular-level high- An activity assay for ME2 was established using NAD+ and
throughput screening assay for ME2 inhibitors. We screened L-malate as substrates; the NADH catalyzed by ME2 was
12 683 natural products and identified a novel ME2 inhibi- monitored by measuring absorbance at a wavelength of 340
tor, NPD387. From a structure-activity relationship study, nm.
we obtained a more potent derivative, NPD389, and we then
investigated the characteristics of NPD389.
We prepared nine assay buffers including 50 mmol/L Bis-
Tris, 50 mmol/L HEPES, 50 mmol/L Citric acid, 10 mmol/L
MgCl2, 12 mmol/L L-malate, and 1 mmol/L NAD+, with dif-
ferent pH values ranging from 4.0 to 8.0. The reactions were
started by adding 37 nmol/L of ME2, and the absorbance of
Materials and methods
Materials and instruments
The plasmid pRH281-ME2 was a kind gift from Prof Liang NADH at 340 nm was monitored for 4 min using a Spectra
TONG of Columbia University (New York, NY, USA). The Max 340 PC 384 microplate reader.
Escherichia coli strains BL21-CodonPlus (DE3) and JM109 were
purchased from Stratagene (La Jolla, CA, USA) and Promega The Km and Kcat of ME2
(Madison, WI, USA), respectively. 3-Indoleacrylic acid (IAA), To calculate the Km of NAD+, the reactions were started by
β-nicotinamide adenine dinucleotide hydrate (NAD+) and adding 15 nmol/L ME2 to enzyme reaction mixtures that
SYPRO orange protein gel stain were purchased from Sigma contained 50 mmol/L MES pH=6.5, 10 mmol/L MgCl2, 24
Aldrich (St Louis, MO, USA). Ni-NTA His-Bind Resin was mmol/L L-malate, and different concentrations of NAD+. To
obtained from Merck Millipore (Billerica, MA, USA). L-malate determine the Km of L-malate, the enzyme reaction mixtures
was obtained from MP Biomedicals LLC (Santa Ana, CA, contained 50 mmol/L MES pH=6.5, 10 mmol/L MgCl2, 1
USA). The other reagents and solvents used in the experi- mmol/L NAD+, 15 nmol/L ME2, and different concentrations
ments were of analytical grade.
of L-malate. For all reactions, absorbance at 340 nm was moni-
The Spectra Max 340 PC 384 microplate reader was from tored for 4 min using a Spectra Max 340 PC 384 microplate
Molecular Devices (Sunnyvale, CA, USA). The Fisher Scien- reader. Kcat was calculated as Vmax/[E concentration]; its units
tific Sonic Dismembrator Model 500 was from Bio Logics, Inc are 1/s. The formula to calculate Km has been described previ-
(Manassas, VA, USA). The transparent, 384-well, medium ously[11]
protein-binding plates were from PerkinElmer (Seattle, WA,
.
USA). The SAGIAN core integrated robotic system was from The IC50 and Ki of ATP
Beckman Coulter (Fullerton, CA, USA). The Light Cycler® 480 The enzyme reaction mixtures contained 50 mmol/L MES
System was from Roche (Basel, BS, Switzerland).
pH=6.5, 10 mmol/L MgCl2, 3 mmol/L L-malate, 0.2 mmol/L
NAD+, 133 nmol/L ME2, and different concentrations of ATP.
The IC50 was calculated using Prism 5 software (Graph Pad,
Expression and purification of ME2
The plasmid pRH281-ME2 was transformed into E coli BL21- San Diego, CA, USA) from the non-linear curve fitting of the
CodonPlus (DE3) cells for expression. BL21-CodonPlus (DE3) percent inhibition (% inhibition) versus the inhibitor concen-
cells containing the recombinant plasmid were grown in 1 L tration [I] using the following equation: % Inhibition=100/
of Luria-Bertani (LB) medium in the presence of ampicillin (1+[IC50/[I]]k), where k represents the Hill coefficient and
(100 mg/L) at 37°C with agitation at 250 rounds per minute. IC50=Ki (1+[S]/Km), where [S] represents the concentration of
Protein expression was induced at 18°C and 180 rounds per substrate. When [S] is equal to Km, then Ki=IC50/2.
minute by adding 400 μmol/L of 3-Indoleacrylic acid (IAA)
when the cultures reached an optical density of 0.4–0.6 at 600 Z’ factor and S/N of HTS assay of ME2
nm (OD600). After 15 h of induction, the cells were harvested Both DMSO and ATP were utilized on each plate to calculate
by centrifugation for 5 min at 4°C and 6000 rounds per min- the Z’ factor. The enzyme reaction mixtures contained 50
ute, washed with phosphate-buffered saline (PBS) buffer, and mmol/L MES pH=6.5, 10 mmol/L MgCl2, 3 mmol/L L-malate,
resuspended in lysis buffer (20 mmol/L Tris-HCl pH=7.5, 0.2 mmol/L NAD+, 133 nmol/L ME2, and 1 mmol/L ATP or
200 mmol/L NaCl, 1 mmol/L β-mercaptoethanol, 1% Triton 4% DMSO. The quality control parameter (Z’ factor) was cal-
X-100, 1 mmol/L PMSF, and Cocktail). The cells were lysed culated in real time. The Z’ factor was defined in terms of four
by sonication for 6 min on ice. After centrifugation at 12000 parameters: the means and standard deviations of both the
rounds per minute for 20 min, the supernatant was incubated positive (p) and negative (n) controls (μp, δp, and μn, δn, respec-
with Ni-NTA Resin for 1.5 h at 4 °C. After washing with tively); Z’=1-3(δp+δn)/(μn–μp). In our assay, ATP and DMSO
lysis buffer containing 50 mmol/L imidazole, the proteins were the positive and negative controls, respectively.
were eluted with lysis buffer containing 250 mmol/L imid-
The signal-to-noise ratio (S/N) is the mean of the positive
azole, dialyzed with buffer (20 mmol/L Tris-HCl pH=7.5, signal from reactions containing DMSO (ie, 100% enzyme
200 mmol/L NaCl, 1 mmol/L β-mercaptoethanol, 1 mmol/L activity) over the mean of the background in the reactions
EDTA, and 10% glycerol) at 4°C to remove the imidazole and treated with known inhibitors. The background is determined
Acta Pharmacologica Sinica

www.nature.com/aps
npg
Wen Y et al
676
by measuring the replicon levels in wells treated with ATP at in the absence or presence of the inhibitor (NPD389 or ATP)
1 mmol/L[12]
.
at various concentrations; when L-malate was the variable,
the concentration of NAD+ was fixed at 1 mmol/L, and when
HTS screening for ME2 inhibitors and exclusion of false positive NAD+ was the variable, the concentration of L-malate was
hits using Brij-35
fixed at 12 mmol/L. The kinetic data were analyzed using
A total of 12 683 natural products collected from different Graph Pad Prism 5.00 and presented as the mean±SD from at
resources (deposited in Chinese National Compound Library) least three independent experiments unless otherwise speci-
with wide structural diversity were randomly screened. One fied.
microliter of a stock solution of each compound in DMSO was
transferred to a well of transparent 384-well medium protein- Thermal shift assay of ME2 in combination with inhibitors
binding plates; this yielded a final compound concentration To investigate the binding between ATP and ME2, we used an
of 40 μg/mL and 4% DMSO in a final reaction volume of 25 assay buffer containing 50 mmol/L MES pH=6.5, 10× dilution
μL. The assay buffer contained 50 mmol/L MES pH=6.5, 10 Sypro Orange, 2.5 μmol/L ME2, and varying concentrations
mmol/L MgCl2, 3 mmol/L L-malate, and 0.2 mmol/L NAD+. of ATP. The thermal shift assay was conducted using a Light
The reactions were initiated by adding 133 nmol/L of ME2, Cycler 480 System. Assay buffer was added to the wells of
and the absorbance of 340 nm was monitored for 4 min using the 384-well LightCycler 480 Multi-well Plate. The plate was
a Spectra Max 340 PC 384 microplate reader. We used ATP as heated from 30 to 80°C with a heating rate of 0.11°C/s. The
a positive control and DMSO as a negative control to evalu- fluorescence intensity was measured with Ex/Em: 465/580 nm.
ate our HTS system. Both the negative and positive controls
To evaluate the potential binding between NPD389 and
were added to each plate for the calculations. The inhibition ME2, we used an assay buffer containing 50 mmol/L MES
effect of each compound was represented by the percentage pH=6.5, 10× Sypro Orange, 2.5 μmol/L ME2, and varying con-
of the slope of the linear portion of its kinetic curve relative to centrations of NPD389. To study the influence of NAD+ on the
that of the negative control. The quality control parameters binding between NPD389 and ME2, we used buffer containing
[coefficient of variation (CV) and Z’ factor] were calculated in 20 mmol/L NAD+ in the same assay. To determine the effect
real time[13]. All screening operations were performed using of L-malate on the binding of the inhibitor to the enzyme, 30
a SAGIAN core integrated robotic system (Beckman Coulter, mmol/L L-malate was added to the thermal shift assay buffer.
Fullerton, CA, USA).
To calculate the 50% inhibition concentration (IC50), we per- Statistical analysis
formed inhibition assays with 133 nmol/L ME2, 50 mmol/L All data shown are representative of three independent experi-
MES pH=6.5, 10 mmol/L MgCl2, 3 mmol/L L-malate, and 0.2 ments. The data are presented as the mean±SD.
mmol/L NAD+; the inhibitors were diluted to around the esti-
mated IC50 values. To exclude false-positive results from HTS
screening, all the hits were tested again using the same assay Establishment of a high-throughput assay to identify ME2
Results
in an assay buffer containing 0.01% (w/v) Brij-35.
inhibitors
We used E coli strain BL21-CodonPlus to overexpress ME2.
Because the recombinant human ME2 protein contains a His-
NPD389 is a more potent inhibitor than NPD387
To elucidate the SARs between ME2 and the derivatives, we tag, Ni-NTA His-binding resin was applied to purify the
tested the inhibitory strength of the derivatives of NPD387. recombinant protein. After washing with 10, 50, and 100
The reaction buffer contained 133 nmol/L ME2, 50 mmol/L mmol/L imidazole solutions, the target proteins were obtained
MES pH=6.5, 10 mmol/L MgCl2, 3 mmol/L L-malate, and 0.2 by elution with 250 mmol/L imidazole solution (Figure 1A)
mmol/L NAD+; the inhibitors were diluted to around the esti- and dialyzed at 4 °C to remove the imidazole. SDS-PAGE
mated IC50 values. We also included 0.01% (w/v) Brij-35 in the indicated that the mass of the protein was approximately 60
reactions to exclude false positives.
kDa, which is consistent with previously published results[7].
The enzyme was purified 142-fold with a yield of 16% from
Determination of the characteristics of inhibition of ME2 by whole lysate, and had a specific activity of 1652.25±11.69
NPD389
Units·min-1·mg-1 of protein (Table 1).
To investigate whether the inhibition of ME2 by NPD389 was
The pH value of a solution has a great effect on enzyme
reversible, incubation solution containing 1.3 μmol/L ME2 activity. To obtain the maximal activity of ME2, we optimized
with 70 μmol/L NPD389 was dialyzed to dissociate the inhibi- the pH value. We measured enzyme activity in buffers with
tor from the enzyme for different periods of time, and then the pH values ranging from 4.0 to 8.0. ME2 had the highest activ-
enzyme was dialyzed into the assay buffer (which included ity at a pH of 6.5 (Figure 1B). We then determined the Km of
the substrates) to initiate the reaction. In the time-independent NAD+ and L-malate in this enzymatic system. The Km and Kcat
inhibition experiment, ME2 at a concentration of 1.3 μmol/L of L-malate were 5.47±0.40 mmol/L and 35.6 s-1, respectively,
was pre-incubated with 70 μmol/L NPD389 on ice for the whereas the Km and Kcat of NAD+ were 0.22±0.02 mmol/L
indicated times, and then added to the assay system. For the and 55.12 s-1, respectively (Figure 1C and 1D); these values
inhibitory characterization study, the assay was conducted are consistent with the previously reported Kms of L-malate
Acta Pharmacologica Sinica

www.chinaphar.com
Wen Y et al
npg
677
Figure 1. Establishment of a high-throughput screening system to identify inhibitors of ME2. (A) SDS-PAGE analysis of purified ME2 separated using
a 10% polyacrylamide gel and stained with Coomassie Brilliant Blue. M, protein marker. Lanes 1–8 are the precipitate, supernatant, flow-through,
and elution fractions with 10, 50, 100, 250, and 250 mmol/L imidazole, respectively. (B) Optimization of the pH of the screening system. (C, D)
Determination of Km and Kcat of L-malate (C) and NAD+ (D). (E) Dose-response curve of inhibition of ME2 by ATP. (F) Determination of the Z′ factor and
S/N of the HTS system for ME2 inhibitors. Error bars represent SD. n=3.
and NAD+ (8.39±1.09 mmol/L and 0.35±0.02 mmol/L, respec- respective Km values.
tively[14]).
It is important to optimize the concentrations of the sub- Determination of IC50 and Ki of the physiological inhibitor ATP
strates before eslishing a high-throughput screening system. ATP, the physiological inhibitor of ME2, was utilized to evalu-
When a substrate concentration less than the Km value is used, ate the stability and quality of our HTS assay. The IC50 of ATP
the screening system can easily identify competitive inhibitors. in our assay system was determined to be 414±23.92 μmol/L
When a substrate concentration greater than the Km value is (Figure 1E). Because ATP is a competitive inhibitor with
used, the screening system can much more easily identify regard to NAD+, the concentration of substrate applied in our
uncompetitive inhibitors. Non-competitive inhibitors of the assay system should be equal to Km, so the Ki of ATP was cal-
enzyme will not be greatly affected by the substrate concen- culated to be 0.2 mmol/L using the formula IC50=Ki(1+[S]/Km);
tration[15, 16]. To acquire as many novel inhibitors of ME2 as this value was close to the Ki value of ATP of 0.2±0.02 mmol/L
possible, we used 0.2 mmol/L NAD+ and 3 mmol/L L-malate published previously[7].
in the HTS assay; both of these concentrations are close to the
Acta Pharmacologica Sinica

www.nature.com/aps
npg
Wen Y et al
678
Table 1. Summary of ME2 purification process from BL21-CodonPlus.
Fractions
Total protein
(mg)
Specific activity
(Units/mg prot)
Total activity
(Units)
Purification
(Fold)
Yield
(%)
Supernatant
Flow-through
Elution (0)
Elution (50)
Elution (100)
Elution (250)
Elution (–)
1331.09±31.45
1049.77±71.85
9.46±0.41
11.62±0.82
0.28±0.01
1.66±0.09
3.19±0.13a
15468.63±1091.47
294.64±8.35
1.00
0.02
0.14
0.27a
44.90a
99.65a
142.19
100
2
15.70±0.83
0
50.52±3.09
0.79±0.02
1.52±0.04
161.27±6.38a
414.62±5.69a
1764.86±25.00a
2511.43±17.76
1a
521.73±7.16a
1158.04±16.40a
1652.25±11.69
3a
11a
16
1.52±0.04
Note: 1 Unit presents 1 nmol NADH product per minute; Data marked with symbol “a” represents the catalytic activity determined including imidazole
in the elution buffer; Elution (0–250) means the elution fraction using buffer containing 0–250 mmol/L imidazole; The fraction of Elution (–) was the
dialyzed fraction of Elution (250) from the imidazole.
Validation of the HTS assay
After the hits were validated and dose-response curves were
Because a large-scale screen would require large amounts of generated, a compound named NPD387 with a novel struc-
materials and compounds, we began by carrying out a pilot ture (Figure 2A) (chemical name: 3,3’,4,4″,6’-pentahydroxy-
screen to assess the quality of the HTS assay. The Z’ factor has [1,1’:4’,1″-terphenyl]-2’,5’-dione) was identified as a moderate
been proposed for reference in high-throughput screening to
represent the response of an inhibitor or activator in a particu-
lar assay. If the value of the Z’ factor exceeds 0.5, it indicates
that the assay is viable[17–21]. We used ATP as the positive con-
trol and DMSO as the negative control to evaluate our HTS
system. The Z’ factor of our HTS system was calculated to be
0.775 (Figure 1F), which indicates that our assay is appropriate
for high-throughput screening.
The second criterion that can be used to measure the robust-
ness of an assay is the signal-to-noise ratio (S/N). S/N indi-
cates the mean of the positive signal over the mean of the
background[22]. In our system, S/N was calculated to be 9.80,
suggesting that our ME2 assay has robust and reproducible
window for the inhibitor discovery.
Discovery of a novel inhibitor of ME2 by HTS
A total of 12683 natural products from the Chinese National
Compound Library were screened. We obtained 47 natural
products as hits from the primary screening at a concentration
of 40 μg/mL; the hit rate was 0.37%. We then determined the
dose-response curves of these 47 inhibitors.
HTS is a primary technique for the identification of novel
leads in drug discovery and chemical biology. Unfortunately,
it is susceptible to false-positive hits. One common cause of
false-positives is the accumulation of organic molecules into
colloidal aggregates, which inhibit enzymes nonspecifically[23]
.
Detergents such as Brij-35 are often used in drug discovery
research to rule out the presence of promiscuous small mol-
ecules and non-specific inhibitors that act by aggregating in
solution or through undesirable precipitation in aqueous assay
buffers[24, 25]. As expected, we excluded 32 promiscuous com-
pounds using 0.01% Brij-35 in the subsequent dose-response
curve determination, leaving 15 ME2 inhibitors with different
structures. The final hit rate of the HTS assay was therefore
0.12%.
Figure 2. NPD387 is a novel inhibitor of ME2. (A) The structure of
NPD387. (B) Dose-response curve of inhibition of ME2 by NPD387. (C)
Dose-dependent inhibition of ME2 by NPD387 with 0.01% Brij-35. Error
bars represent SD. n=3.
Acta Pharmacologica Sinica

www.chinaphar.com
Wen Y et al
npg
679
Table 2. Screening structural analogs of NPD387.
IC50 (μmol/L)
Compound
R1
R2
R3
No Brij-35
0.01% Brij-35
NPD387
H
11.84±0.57
18.27±1.46
CAS0006-E009
NPLC1171
*
#
*
#
*
#
2.13±0.20
48.63±2.36
31.02±1.54
–
NPLC1174
NPD389
Me
H
–
–
4.63±0.36
5.59±0.38
Bq
H
H
H
H
H
8.34±0.71
7.32±2.31
Embelin
52.40±3.88
110.68±4.77
E-20
E-38
H
H
–
–
H
184.61±6.97
103.78±5.60
E-39
E-40
E-42
E-47
E-48
E-57
H
H
H
H
H
H
69.23±3.09
–
–
–
–
–
–
–
–
H
H
158.74±8.24
23.17±1.50
17.43±1.17
E-59
H
H
18.59±1.25
34.09±3.28
–, represented the IC50 more than 200 μmol/L. *, represented the structure of CAS0006-E009, please refer to Figure 3. #, represented the structure
of NPLC1171, please refer to Figure 3.
inhibitor of ME2, with IC50 values of 11.84±0.57 μmol/L and but NPLC1174 exhibited no potency, which indicates that the
18.27±1.46 μmol/L in the absence and presence of 0.01% Brij- 2,5-dihydroxyl groups are indispensable for the activity of
35, respectively (Figure 2B, 2C, Table 2).
the compound. In addition, when the carbonyl groups of the
2,5-dihydroxyl benzoquinone skeleton were reduced to phe-
nolic hydroxyl, the compound displayed decreased activity
SARs between ME2 and derivatives
After the identification of NPD387 as a lead compound in (CAS0006-E009). Therefore, the 2,5-dihydroxyl benzoquinone
the search for inhibitors of ME2, we synthesized a series of skeleton is essential for the ME2 inhibitory activity of these
NPD387 derivatives and measured their inhibition of ME2 to compounds.
determine their SARs (Figure 3). Bq and Embelin were pur-
In addition, we tested a series of 3-alkyl-substituted
chased from Alfa Aesar (Shanghai, China) and used without 2,5-dioxo-benzoquinones and 3,6-dialkyl substituted
any further purification, and the other derivatives were pre- 2,5-dioxo-benzoquinones. We found that the activities of these
pared as described previously[26] and as detailed in Scheme 1. compounds were decreased to different extents. Specifically,
The preliminary SARs are shown below.
the C3 alkyl chain-substituted compounds showed remarkable
It is clear that the original compound, NPD387, and one decreases in activity, and the C3,C6 dialkyl-substituted com-
derivative, NPD389, showed strong ME2 inhibitory activities, pounds exhibited a complete loss of activity. Among these
Acta Pharmacologica Sinica

www.nature.com/aps
npg
Wen Y et al
680
inhibition of ME2.
Among the compounds we screened, NPD387, NPD389
and Bq displayed the strongest inhibition of ME2. The com-
mon feature of these three compounds is their 2,5-dihydroxy
benzoquinone skeleton. By comparing the IC50s of NPD387
and NPD389 with that of Bq, we found that substituted
phenyl groups at C3 and C6 had some impact on inhibitory
activity towards ME2. Specially, 4-OMe-substituted phenyl
groups at C3 and C6 could improve the activity (NPD389), but
4-OH substituted phenyl groups at C3 and C6 could weaken
the interaction between the compound and ME2 (NPD387),
decreasing the inhibitory activity of the compound.
NPD387
NPLC1174
NPD389
Bq
Embelin
E-20
E-38
E-39
E-40
E-42
R1=H, R2=3,4-(OH)2C6H3, R3=4-OHC6H4
R1=Me, R2=R3=4-OHC6H4
R1=H, R2=R3=4-OMeC6H4
R1=R2=R3=H
R1=H, R2=n-C11H23, R3=H
R1=H, R2=n-C9H19, R3=n-C3H7
R1=H, R2=n-C3H7, R3=H
R1=H, R2=n-C9H19, R3=CH2C6H5
R1=H, R2=n-C5H11, R3=n-C5H11
R1=H, R2=n-C9H19, R3=n-C5H11
R1=H, R2=n-C8H17, R3=n-C8H17
R1=H, R2=n-C4H9, R3=H
Characterization of NPD389 as a more potent ME2 inhibitor
Through structural modification, we generated NPD389,
which is a more potent inhibitor of ME2 than NPD387 (Figure
4A). The IC50 of NPD389 was 4.63±0.36 μmol/L or 5.59±0.38
μmol/L when 0.01% Brij-35 was absent from or present in the
assay system, respectively (Figure 4B and 4C, Table 2).
E-47
E-48
E-57
E-59
R1=H, R2=CH2C6H4-4-C6H5, R3=H
R1=H, R2=CH2C6H4-4-OC6H5, R3=H
We further investigated the mode by which NPD389 inhib-
its ME2. NPD389 is a reversible inhibitor of ME2, as shown
by the restored activity of ME2 when the dialysis time was
extended (Figure 5A). The unchanged inhibitory rate over
incubation time indicated that NPD389 binds with ME2 in
a fast-binding mode (Figure 5B). NPD389 is a mixed-type
competitive inhibitor for the substrate L-malate, as shown
by the increased Km values and decreased Vmax values that
are seen when the concentration of inhibitor is increased; the
Lineweaver-Burk plot of different concentrations of NPD389
intersected in the second quadrant of the reciprocal plot (Fig-
ure 5C). Moreover, NPD389 is an uncompetitive inhibitor for
the other substrate NAD+ because both the Km and Vmax values
decreased when the concentration of inhibitor was increased,
and the Lineweaver-Burk plots of four different inhibitor con-
centrations showed parallel lines (Figure 5D). In this enzyme
kinetic characterization study, our data showed that ATP (the
positive control) is a competitive inhibitor with respect to
NAD+ (Figure 5E), consistent with previously reports[7].
Figure 3. The structures and formulae of derivatives of NPD387.
Validation of NPD389 binding to ME2 through thermal shift
assay
compounds, only compounds E-57 and E-59 retained some
inhibitory activity. This phenomenon indicates that the pres-
ence of flexible alkyl groups at C3 or C6 is incompatible with
The fluorescence-based thermal shift assay is a general method
for the identification of inhibitors of target proteins from com-
pound libraries. This method is based on the energetic cou-
Scheme 1. The synthesis of derivatives of NPD387. Reagent: (a) n-BuLi, THF, -10°C; (b) R1, Br, -10 °C to RT; (c) n-BuLi, THF, -10°C; R2, Br, -10°C to RT; (d)
1,4-dioxane, 6 mol/L HCl, air, reflux.
Acta Pharmacologica Sinica

www.chinaphar.com
Wen Y et al
npg
681
ME2 with NPD389. The result indicated that there was a dose-
dependent shift Tm in the presence of NAD+. However, there
was no significant difference when the assay included only
L-malate or neither substrate (Figure 6B). This result demon-
strated that NPD389 is an uncompetitive inhibitor of ME2 with
respect to NAD+ because the shift in the Tm of ME2 induced by
NPD389 occurred when the substrate NAD+ was present.
Discussion
Evidence from previous studies has supported the concept
that ME2 may be a promising target for cancer therapy. No
inhibitor of human ME2 has been reported previously except
for the physiological inhibitor ATP; this lack of an inhibitor
makes it difficult to validate the role of ME2 in tumorigenesis
and to develop cancer therapies based on ME2. The purpose
of this study was to discover a novel inhibitor of human ME2.
Since being developed approximately 20 years ago, HTS
has become a key technique in drug discovery. HTS pro-
vides a fast, effective, and convenient way to test hundreds
of thousands of structurally diverse compounds for their
ability to modulate disease-relevant targets and to identify
novel compounds with interesting biological activities; this
approach yields molecules that can be optimized into drugs.
In this study, an HTS assay was established to identify small-
molecule inhibitors of ME2. The average Z’ factor of this assay
was 0.775, which demonstrates satisfactory HTS quality con-
trol.
Through large-scale screening and structural modifica-
tion, we identified NPD389 as the first potent ME2 inhibitor.
Inhibition of ME2 by NPD389 is independent of incubation
time, which indicates that NPD389 is a fast-binding inhibitor
of ME2. We also demonstrated that NPD389 is a mixed type
inhibitor with respect to L-malate and an uncompetitive inhib-
itor with respect to NAD+ using a double reciprocal plot.
A key problem in HTS is the prevalence of non-specific
inhibitors. These molecules, which have peculiar properties,
act on unrelated targets and dominate the results of screen-
ing experiments. Part of the explanation for these nonspecific
effects is that some organic molecules form large colloid-like
Figure 4. NPD389 was identified as a more potent inhibitor of ME2
than NPD387. (A) The structure of NPD389. (B) Dose-response curve of
inhibition of ME2 by NPD389. (C) Dose-dependent inhibition of ME2 by
NPD389 with 0.01% Brij-35. Error error bars represent SD. n=3.
pling between ligand binding and protein unfolding. Using
an environmentally sensitive fluorescent dye (Sypro Orange)
to monitor protein thermal unfolding, the ligand-binding
affinity can be assessed from the shift in the unfolding temper-
ature (ΔTm) obtained in the presence of ligands relative to that
obtained in the absence of ligands[27, 28]. A shift in the melting
temperature (Tm) of an enzyme during incubation with a given
test compound indicates that the test compound specifically
binds to the enzyme.
After optimization with ATP as positive control inhibitor,
we used a 10× dilution of Sypro Orange and 2.5 μmol/L ME2
protein in the thermal shift assay; under these conditions, a
perfect melting curve could be obtained. As shown in Figure
6A, there was a dose-dependent shift in Tm with the increasing
concentrations of ATP, and a ΔTm of 4.26±0.18°C was calcu-
lated by comparing Tm in the absence of ATP to that with 500
μmol/L of ATP, which was close to its IC50 value (Figure 6A).
This thermal shift result confirmed that ATP is a competitive
inhibitor of ME2 with respect to NAD+.
aggregates that sequester and thereby inhibit enzymes[23]
.
Detergents such as Brij-35 and Tween are often used in drug
discovery to weed out promiscuous small molecules and non-
specific inhibitors that cause aggregation in solutions and
undesirable precipitation in aqueous assay buffers[24, 25]. Com-
pounds CAS0006-E009 and NPLC1171 might aggregate in a
solution, interfering with binding between the enzyme and the
substrate and thereby resulting in remarkable differences in
IC50 values measured with or without Brij-35.
To further investigate the interaction between the inhibitor
and the enzyme, we used a thermal shift assay to test the com-
bination of inhibitors and ME2. The fluorescence-based ther-
mal shift assay is a general method for identifying inhibitors
of target proteins from compound libraries. It has the advan-
tages of simple operation, short experimental cycle, and ease
of further evaluation of compounds. The thermal shift assay
result showed that there was a dose-dependent shift in Tm as
We then determined whether there was a shift in the Tm of
Acta Pharmacologica Sinica

www.nature.com/aps
npg
Wen Y et al
682
Figure 5. The mode of inhibition of ME2 by NPD389. (A) The effect
of NPD389 on enzyme activity after incubation with the enzyme and
subsequent dialysis. (B) Time-independent inhibition of ME2 by NPD389.
(C) Double-reciprocal plot with changes in the concentrations of the
inhibitor NPD389 and the substrate L-malate. (D) Double-reciprocal
plot with changes in the concentrations of the inhibitor NPD389 and
the substrate NAD+. (E) Double-reciprocal plot with changes in the
concentrations of the inhibitor ATP and the substrate NAD+. Error bars
represent SD. n=3.
the concentration of ATP increased. After testing the combi- inhibitor with respect to NAD+. Our study illustrated that the
nation of NPD389 and ME2, we found that there was a dose- substrate of an enzyme should not be omitted in a thermal
dependent increase in Tm in the presence of NAD+. However, shift assay when the inhibitor is uncompetitive.
the change in the Tm of ME2 with NPD389 disappeared when
In summary, we identified a novel 2,5-dihydroxyl benzoqui-
there was no NAD+ in the system. These results showed that none skeleton as an inhibitor of ME2 and identified a potent
NPD389 binds to the complex of ME2 and NAD+ and that derivative of this inhibitor, NPD389, from a preliminary SAR
NAD+ is necessary for the binding of NPD389 to ME2. These study. A kinetic characterization study demonstrated that
results also confirmed that NPD389 was an uncompetitive NPD389 is a fasting-binding inhibitor of ME2 and acts as an
Acta Pharmacologica Sinica

www.chinaphar.com
Wen Y et al
npg
683
Figure 6. The effect of ATP and NPD389 on the Tm of ME2 probed with Sypro Orange. (A) The effect of varying concentrations of ATP on Tm. (B) The
effect of NPD389 on Tm. From left to right, the groups are: without NAD+ and L-malate; in the presence of L-malate; and in the presence of NAD+,
respectively. Error bars represent SD. n=3.
uncompetitive inhibitor with respect to NAD+ and a mixed-
fumarate. Structure 2002; 10: 951–60.
8
9
Wasilenko WJ, Marchok AC. Malic enzyme and malate dehydrogenase
activities in rat tracheal epithelial cells during the progression of
neoplasia. Cancer Lett 1985; 28: 35–42.
Sauer LA, Dauchy RT, Nagel WO, Morris HP. Mitochondrial malic
enzymes. Mitochondrial NAD(P)+-dependent malic enzyme activity
and malate-dependent pyruvate formation are progression-linked in
Morris hepatomas. J Biol Chem 1980; 255: 3844–8.
type competitive inhibitor with respect to L-malate.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China (81125023 and 81021062), the National
Science and Technology Major Projects for Major New
Drugs Innovation and Development (2013ZX09507002),
and the Shanghai Commission of Science and Technology
(11DZ2292200 and 13DZ2290300). We also thank Yan-yun FU
and Bei-ying QIU for their help in editing the manuscript.
10 Hou S, Liu W, Ji D, Zhao ZK. Efficient synthesis of triazole moiety-
containing nucleotide analogs and their inhibitory effects on a malic
enzyme. Bioorg Med Chem Lett 2011; 21: 1667–9.
11 Cleland WW. Statistical analysis of enzyme kinetic data. Methods
Enzymol 1979; 63: 103–38.
12 Mondal R, Koev G, Pilot-Matias T, He Y, Ng T, Kati W, et al. Develop-
ment of a cell-based assay for high-throughput screening of inhibitors
against HCV genotypes 1a and 1b in a single well. Antiviral Res 2009;
82: 82–8.
13 Zhang JH, Chung TD, Oldenburg KR. A simple statistical parameter for
use in evaluation and validation of high throughput screening assays.
J Biomol Screen 1999; 4: 67–73.
Author contribution
Jia LI, Li-hong HU, Jing GAO, and Jing-ya LI designed the
study; Yi WEN and Lei XU performed the experiments and
prepared the manuscript; and Fang-lei CHEN synthesized the
derivatives of NPD387.
14 Hung HC, Chien YC, Hsieh JY, Chang GG, Liu GY. Functional roles of
ATP-binding residues in the catalytic site of human mitochondrial
NAD(P)+-dependent malic enzyme. Biochemistry 2005; 44: 12737–
45.
15 Cornish-Bowden A. Why is uncompetitive inhibition so rare? A
possible explanation, with implications for the design of drugs and
pesticides. FEBS Lett 1986; 203: 3–6.
References
1
Hsu RY, Lardy HA. Pigeon liver malic enzyme. II. Isolation, crystalliza-
tion, and some properties. J Biol Chem 1967; 242: 520–6.
Loeber G, Infante AA, Maurer-Fogy I, Krystek E, Dworkin MB. Human
NAD(+)-dependent mitochondrial malic enzyme. cDNA cloning,
primary structure, and expression in Escherichia coli. J Biol Chem
1991; 266: 3016–21.
2
3
4
5
6
7
Xu Y, Bhargava G, Wu H, Loeber G, Tong L. Crystal structure of human
mitochondrial NAD(P)+-dependent malic enzyme: a new class of
oxidative decarboxylases. Structure 1999; 7: R877–89.
Ren JG, Seth P, Everett P, Clish CB, Sukhatme VP. Induction of
erythroid differentiation in human erythroleukemia cells by depletion
of malic enzyme 2. PLoS One 2010; 5: e12520.
Romero-Garcia S, Lopez-Gonzalez JS, Baez-Viveros JL, Aguilar-Cazares
D, Prado-Garcia H. Tumor cell metabolism: an integral view. Cancer
Biol Ther 2011; 12: 939–48.
Jiang P, Du W, Mancuso A, Wellen KE, Yang X. Reciprocal regulation
of p53 and malic enzymes modulates metabolism and senescence.
Nature 2013; 493: 689–93.
Yang Z, Lanks CW, Tong L. Molecular mechanism for the regulation of
human mitochondrial NAD(P)+-dependent malic enzyme by ATP and
16 Dixon M. The determination of enzyme inhibitor constants. Biochem
J 1953; 55: 170–1.
17 Zhang JH, Chung TD, Oldenburg KR. A simple statistical parameter for
use in evaluation and validation of high throughput screening assays.
J Biomol Screen 1999; 4: 67–73.
18 Sui YX, Wu ZJ. Alternative statistical parameter for high-throughput
screening assay quality assessment. J Biomol Screen 2007; 12:
229–34.
19 Zhang XD, Espeseth AS, Johnson EN, Chin J, Gates A, Mitnaul LJ, et
al. Integrating experimental and analytic approaches to improve data
quality in genome-wide RNAi screens. J Biomol Screen 2008; 13:
378–89.
20 Zhang XD. A pair of new statistical parameters for quality control in
RNA interference high-throughput screening assays. Genomics 2007;
Acta Pharmacologica Sinica

www.nature.com/aps
npg
Wen Y et al
684
89: 552–61.
21 Zhang XD. Novel analytic criteria and effective plate designs for
25 Walters WP, Namchuk M. Designing screens: how to make your hits a
hit. Nat Rev Drug Discov 2003; 2: 259–66.
quality control in genome-scale RNAi screens. J Biomol Screen 2008; 26 Gan X, Jiang W, Wang W, Hu L. An approach to 3,6-disubstituted
13: 363–77.
2,5-dioxybenzoquinones via two sequential Suzuki couplings. three-
step synthesis of leucomelone. Org Lett 2009; 11: 589–92.
22 Mondal R, Koev G, Pilot-Matias T, He YP, Ng T, Kati W, et al. Develop-
ment of a cell-based assay for high-throughput screening of inhibitors 27 Pantoliano MW, Petrella EC, Kwasnoski JD, Lobanov VS, Myslik J, Graf
against HCV genotypes 1a and 1b in a single well. Antiviral Res 2009;
82: 82–8.
E, et al. High-density miniaturized thermal shift assays as a general
strategy for drug discovery. J Biomol Screen 2001; 6: 429–40.
23 McGovern SL, Helfand BT, Feng B, Shoichet BK. A specific mechanism 28 Lo MC, Aulabaugh A, Jin G, Cowling R, Bard J, Malamas M, et al.
of nonspecific inhibition. J Med Chem 2003; 46: 4265–72.
24 Feng BY, Shelat A, Doman TN, Guy RK, Shoichet BK. High-throughput
assays for promiscuous inhibitors. Nat Chem Biol 2005; 1: 146–8.
Evaluation of fluorescence-based thermal shift assays for hit identifi-
cation in drug discovery. Anal Biochem 2004; 332: 153–9.
Acta Pharmacologica Sinica