Novel inhibitors of nicotinamide-n-methyltransferase for the treatment of metabolic disorders

Article abstract of DOI:10.3390/molecules26040991

Nicotinamide-N-methyltransferase (NNMT) is a cytosolic enzyme catalyzing the transfer of a methyl group from S-adenosyl-methionine (SAM) to nicotinamide (Nam). It is expressed in many tissues including the liver, adipose tissue, and skeletal muscle. Its expression in several cancer cell lines has been widely discussed in the literature, and recent work established a link between NNMT expression and metabolic diseases. Here we describe our approach to identify potent small molecule inhibitors of NNMT featuring different binding modes as elucidated by X-ray crystallographic studies.

Full text of DOI:10.3390/molecules26040991

molecules
Article
Novel Inhibitors of Nicotinamide-N-Methyltransferase for the
Treatment of Metabolic Disorders
Aimo Kannt 1 , Sridharan Rajagopal 2, Mahanandeesha S. Hallur 2, Indu Swamy 2, Rajendra Kristam 2,
Saravanakumar Dhakshinamoorthy 2, Joerg Czech 3, Gernot Zech 3, Herman Schreuder 3 and Sven Ruf 3,
*
1
Fraunhofer Institute for Translational Medicine and Pharmacology-ITMP, Theodor-Stern-Kai 7,
60596 Frankfurt am Main, Germany; aimo.kannt@itmp.fraunhofer.de
2
Jubilant Biosys Limited, #96, Industrial Suburb, 2nd Stage Yeshwanthpur, Bangalore 560022, India;
Sridharan.Rajagopal@jubilanttx.com (S.R.); Mahanandeesha.Hallur@jubilantbiosys.com (M.S.H.);
Indu.Swamy@jubilantbiosys.com (I.S.); Rajendra.Kristam@jubilantbiosys.com (R.K.);
Saravanakumar.Dhakshinamoorthy@jubilantbiosys.com (S.D.)
Sanofi-Aventis Deutschland Gmbh, Industriepark Hoechst, 65926 Frankfurt am Main, Germany;
Joerg.Czech@sanofi.com (J.C.); Gernot.Zech@sanofi.com (G.Z.); Herman.Schreuder@sanofi.com (H.S.)
Correspondence: Sven.Ruf@sanofi.com
3
*
Abstract: Nicotinamide-N-methyltransferase (NNMT) is a cytosolic enzyme catalyzing the transfer
of a methyl group from S-adenosyl-methionine (SAM) to nicotinamide (Nam). It is expressed in
many tissues including the liver, adipose tissue, and skeletal muscle. Its expression in several
cancer cell lines has been widely discussed in the literature, and recent work established a link
between NNMT expression and metabolic diseases. Here we describe our approach to identify
potent small molecule inhibitors of NNMT featuring different binding modes as elucidated by X-ray
crystallographic studies.
ꢀꢁꢂꢀꢃꢄꢅꢆꢇ
ꢀꢁꢂꢃꢄꢅꢆ
Keywords: nicotinamide-N-methyltransferase; small molecule inhibitor; nicotinamide; metabolic
Citation: Kannt, A.; Rajagopal, S.;
Hallur, M.S.; Swamy, I.; Kristam, R.;
Dhakshinamoorthy, S.; Czech, J.;
Zech, G.; Schreuder, H.; Ruf, S. Novel
Inhibitors of Nicotinamide-N-
disease; X-ray
1. Introduction
Methyltransferase for the Treatment
of Metabolic Disorders. Molecules
2021, 26, 991. https://doi.org/
10.3390/molecules26040991
Nicotinamide-N-methyltransferase (NNMT) is a cytosolic enzyme that catalyzes the
transfer of a methyl group from S-adenosyl-methionine (SAM) to nicotinamide (Nam),
yielding S-adenosyl-homocysteine (SAH) and 1-methylnicotinamide (MNam) [
expressed in most tissues including the liver, adipose tissue, and skeletal muscle [
also in several human cancers [
systemic knockout in mice fully prevents MNam formation [
excreted via urine or further metabolized to 1-methyl-2-pyridone-5-carboxamide (Me2PY)
or 1-methyl-4-pyridone-5-carboxamide (Me4PY) in a reaction catalyzed by aldehyde oxi-
dase [7].
1
,
2
3
]. It is
], and
Academic Editor: Manoj K. Pandey
Received: 15 December 2020
Accepted: 6 February 2021
4
]. NNMT is the major Nam-metabolizing enzyme, and
]. MNam can either be
5,6
Published: 13 February 2021
Publisher’s Note: MDPI stays neu-
tral with regard to jurisdictional claims
in published maps and institutional
affiliations.
Higher NNMT expression and activity in white adipose tissue of mice [
mans [ 10] has been associated with obesity, insulin resistance and type-2 diabetes (T2D).
Plasma, serum or urinary MNam levels were found to correlate with obesity or T2D in
individuals from different geographic regions [10 12]. Interventions to improve metabolic
8] and hu-
9,
–
health such as exercise and bariatric surgery were shown to decrease adipose NNMT ex-
pression and plasma MNam levels in obese individuals with insulin resistance or T2D [10].
Single-nucleotide polymorphisms of NNMT have been associated with BMI [13], hyperho-
mocysteinemia [14], non-alcoholic steatohepatitis [15], and hyperlipidemia [16].
Reduction in hepatic and adipose tissue NNMT with an antisense oligonucleotide led
to lower weight gain, a decrease in relative fat mass, lower plasma insulin levels and im-
Copyright: © 2021 by the authors. Li-
censee MDPI, Basel, Switzerland. This
article is an open access article distributed
under the terms and conditions of the
Creative Commons Attribution (CC BY)
license (https://creativecommons.org/
licenses/by/4.0/).
proved glucose tolerance in mice on a high-fat diet (HFD) [
6,8]. Likewise, treatment with the
selective NNMT inhibitors 5-amino-1-methylquinoline [17] or 6-methoxynicotinamide [
5]
Molecules 2021, 26, 991. https://doi.org/10.3390/molecules26040991
https://www.mdpi.com/journal/molecules

Molecules 2021, 26, 991
2 of 22
resulted in reduced body weight, higher insulin sensitivity, and improved glucose tolerance
in mice with diet-induced obesity (Figure 1).
Figure 1. Known NNMT inhibitors.
Additional types of NNMT inhibitors that have recently been reported include other
nicotinamide-related compounds such as 6-methylaminonicotinamide [18], bisubstrate-
like inhibitors based on adenosyl scaffolds [19–23], and alpha-chloroacetamide-based
compounds covalently binding to a cysteine residue in the SAM-binding pocket of the
catalytic site [24].
Here we describe two novel types of NNMT inhibitors: (A) The 3-methyl-4-phenylpyrazole
series represented by (
center and is the first non-adenosyl bisubstrate inhibitor series with nanomolar potency.
(B) A series of tricyclic compounds represented by ( ), ( ) with high selectivity for NNMT
and strong inhibitory activity on the isolated enzyme (Figure 2).
1), which occupies both the Nam and SAM pockets of the catalytic
2
3
3
2
1
Figure 2. HTS hits identified from Sanofi compound collection.
Both series were identified in a high-throughput screening (HTS) of the Sanofi com-
pound collection and were not published previously. The inhibition of human NNMT
(hNNMT) enzymatic activity by these compounds (IC50 (hNNMT) = 0.26
µM (1); 1.6 µM
(2
) and 0.18 M ( )) was confirmed after re-synthesis. Therefore, we considered these series
µ
3
as good starting points for the further exploration of NNMT as a therapeutic target in
different indications. All compounds act as competitive inhibitors either of nicotinamide
alone (2,3) or of nicotinamide and SAM simultaneously (1), as will be further elucidated by
our investigations. Km values for nicotinamide and SAM were reported as 400 and 1.8 µM,
respectively [2].

Molecules 2021, 26, 991
3 of 22
2. Results and Discussion
2.1. Chemistry
Compound (1) was synthesized starting from 3-bromo-4-methyl pyrazole. Coupling
of phenyl boronic acid with the pyrazole derivative in the presence of Pd catalyst yielded
phenyl pyrazole (1A) in 84% yield. Thermal ring opening of the epoxide with phenyl-
pyrazole afforded intermediates (1B) (minor isomer) and (1C) (major isomer) in 80% overall
yield. Subsequent hydrolysis of the ester (1B) gave the acid (1D), and an amide coupling
with the piperazine substituted isoquinoline (1E) resulted in the desired final product
(1). Other derivatives of these series were synthesized following this general procedure
(Scheme 1). All compounds of this series contain a chiral center and were profiled as
racemic mixtures.
K2CO3, Pd(OAc)2,
MeOH, 90 °C, 20 h
(seal tube)
CATAXIUM A,
(80 %)
1,4-dioxane-H2O
100 °C, 12 h
(84 %)
1E
NaOH, MeOH:THF:H2O
EDC.HCl, HOBt, TEA,
DCM, r.t., 16 h
(17 %)
r.t., 12 h
(78 %)
1
Scheme 1. Synthesis of NNMT inhibitor (1).
Our hit structure ( ) is accessible by two synthetic steps starting from tetrahydroquino-
2
line: the thiourea (2A), generated from tetrahydroquinoline, undergoes a cyclization to the
desired product (2) in the presence of bromine (Scheme 2).
Br2
KSCN
CHCl3, r.t.
CH2Cl2, 70 °C
(16%)
HCl
2
2A
Scheme 2. Synthesis of NNMT inhibitor (2).
The chemistry for the generation of lead (3) starts from 8-aminoquinoline, which
is hydrogenated first in the pyridine part to give (3A). Afterwards, a cyclization to the
five-membered heterocycle occurs with BrCN to give (3B) and a final alkylation with
ethylbromide delivers the target compound (3) (Scheme 3).

Molecules 2021, 26, 991
4 of 22
PtO2, H2,
AcOH
BrCN/MeOH,
H2O
K2CO3
Dioxane,50 oC,
overnight,
48 h,
(80%)
80 oC, 2 h,
(65%)
(39.6%)
HBr
3A
Scheme 3. Synthesis of NNMT inhibitor (3).
Synthesis of compound ( ) has been accomplished starting from quinoline-8-carboxylic
3B
3
4
acid. After formation of the corresponding N-methyl-carboxamide (4A), the pyridine ring
of the quinoline system was reduced over PtO2 to give (4B), followed by reduction of the
carboxamide with LAH to afford (4C). Treatment of the resulting diamine (4C) with BrCN
results in a ring closure to the desired product (4) (Scheme 4).
EDCl, HOBt, DIEA,
PtO2, H2, HOAc,
LAH, THF,
reflux, 2 h
MeNH2, HCl, DMF,
r.t., overnight
r.t., overnight
(96%)
Crude taken to
next step
Crude taken to
next step
4B
4A
MeOH, THF,
50 0C, 3 h,
(25%)
4C
4
Scheme 4. Synthesis of NNMT inhibitor (4).
2.2. Structural Biology
We crystallized hNNMT following a protocol from the Structural Genomics Con-
sortium (SGC), which is available under DOI:10.2210/pdb2iip/pdb (see Section 3). As
was described by Peng et al [25], NNMT possesses a class I AdoMet-dependent methyl-
transferase (MTase) core fold, comprising a central seven-stranded
both sides by -helices. In addition to the MTase core, there are two extra
N-terminus, which, together with the -hairpin Y203-S212, form a cap over the active site.
β-sheet, flanked on
α
α
-helices at the
β
Two more helices are inserted within W107-A134.
Figure 3a–e shows the observed electron densities observed with our inhibitors (1–4)
and (33) after crystallization with hNNMT. In all cases, the electron density maps are
sufficiently well defined to allow an unambiguous fitting of the corresponding inhibitor.
2.3. Binding Mode of the High Throughput Screening Hits
We started our co-crystallization efforts with the HTS hit (
experiments of hNNMT with ( ) alone were not successful, but DSF experiments with
) and the cofactor (reaction product) SAH indicated a strong additive stabilization and
co-crystallization experiments with both SAH and (2) were successful.
The resulting crystal structure shows that inhibitor ( ) is bound in a flat, rather
2). Co-crystallization
2
(2
2
hydrophobic cavity. Its “bottom” is formed by the side chains of Tyr204, Tyr24, and Ala247,
while the “top” is formed by the side chains of Leu164, Ala168, Ala198, and Tyr242. Also
present in the “top” is the carboxylate group of Asp197, which binds the amide nitrogen
of the natural substrate nicotinamide. The “sides” of the cavity are more hydrophilic and
contain the O
makes a hydrogen bond to the imine nitrogen of (
γ
atoms of Ser201 and Ser213, as well as the hydroxyl group of Tyr20, which
). In addition, the C atom of Asp197 is
2
β
present in the side of the cavity, while its carboxylate group points away from the cavity.
An overview of these interactions is given in Figure 4.

Molecules 2021, 26, 991
5 of 22
As was mentioned above, the imine of (
2
) makes a hydrogen bond with the hydroxyl
group of Tyr20. It also makes a van der Waals contact (d = 3.3 Å) with the sulfur of SAH.
This is the position where the methyl group of the co-substrate S-adenosyl-L-methionine
(SAM) would be transferred by the enzyme to the substrate nicotinamide.
As is shown in Figure 4b, inhibitor (2) binds in the binding pocket for the substrate
nicotinamide [25]. The imine residue of inhibitor (
2
) is near the position of the pyridine
nitrogen of nicotinamide, the nitrogen of which accepts a methyl group during normal
catalytic turnover of the enzyme. The carboxamide group of nicotinamide interacts with
the side chains of Asp197 and Ser213. Around this position, inhibitor (2) only has aliphatic
and aromatic carbons.
The crystal structure of the hNNMT complex with the HTS hit (3) (Figure 5) shows
a very similar binding mode as for inhibitor ( ). The NH group of inhibitor (
2
3) forms a
hydrogen bond with the side chain of Tyr20 and also with the main chain oxygen atom of
Leu164, and there are van der Waals contacts of the inhibitor with the side chains of Ser201
and Ser213.
Figure 3. Omit maps, contoured at 2
σ
of
h
NNMT in complex with various inhibitors. (a) (2); 2.33 Å,
(b
) ( ); 2.81 Å, ( ) ( ); 2.28 Å, ( ) (33); 2.8 Å, (e 4
3
c
1
d
) ( ); 2.9 Å. Omit maps are calculated by subtracting
calculated X-ray data, where the ligand has been omitted from the model, from the observed X-ray
data. Anything that is present in the observed, but not in the calculated X-ray data (e.g., the omitted
ligand) will show up as positive density in the resulting difference map. Since no information of the
ligand has been used to calculate these maps, model bias will be minimal.

Molecules 2021, 26, 991
6 of 22
Figure 4.
S-adenosyl-homocysteine (SAH) in yellow, and hNNMT in green. (
(blue; pdb entry 3rod [ ]) and inhibitor ( ) (purple; this study), as well as a ribbon presentation of the NNMT protein from
the inhibitor (2) complex (green).
(
a) Crystal structure showing the interactions of inhibitor (
2) with hNNMT. Inhibitor (2) is shown in purple,
b) Comparison of the binding mode of nicotinamide
4
2
Figure 5. Crystal structure showing the interactions of (3) with hNNMT.
Finally, the crystal structure of the complex of hNNMT with the HTS hit compound
(
1
) shows that (
compound ( ) fits well in the electron density maps and its binding mode is well defined.
As will be discussed below, no SAH is bound to hNNMT since the SAH binding pocket
is partially occupied by compound ( ) (Figure 6b), making it impossible for NNMT to
simultaneously bind SAH and the inhibitor.
1) occupies both the SAH and Nam binding pockets. As shown in Figure 3c,
1
1

Molecules 2021, 26, 991
7 of 22
Figure 6.
binding mode of SAH (yellow), (
the bottom-center, a large cavity is visible, below the Nam binding site.
(
a
) Crystal structure showing the interactions of compound
1 (blue) with hNNMT. (b) Comparison with the
1
) (blue) and ( ) (yellow) in the ( ) complex. The surface of the protein is shown in grey. In
2
2
The 3-methyl-4-phenyl-pyrazole group of inhibitor (
binding pocket which is occupied by ( ) in the hNNMT complex of inhibitor (
) occupies both the SAH- and the Nam binding pockets. As with other NNMT inhibitors,
1) is located in the nicotinamide
2
2
); inhibitor
(1
the methyl group of the 3-methyl-4-phenyl-pyrazole group points into a large cavity below
the Nam binding site (see Figure 6b). With the exception of the piperazine nitrogens, all
heteroatoms of the inhibitor are involved in hydrogen bonds with the hNNMT (Figure 6a).
2.4. Optimization of a Bisubstrate-Like Inhibitor Series Based on Compound 1
To improve the potency and understand the mode of binding in compound (
focused library of compounds was synthesized, as shown in Figure 7 and Table 2. In
), we replaced the hydroxyl and methyl group of the 2-hydroxy-2-methyl-propanoic
carboxamide linker in ( ) with hydrogen and found a six-fold reduction in potency on
hNNMT and a 2.5-fold reduction in potency on mouse NMMT (mNNMT). In ( ), only
1), a
(
6
1
9
the R3 methyl group in the linker area was replaced with hydrogen and the result was a
three-fold reduction in hNNMT potency (Table 2). Based on these observations, we decided
to leave the linker region unchanged for further studies.
Figure 7. General formula of NNMT inhibitors described in Tables 1 and 2.

Molecules 2021, 26, 991
8 of 22
Table 1. Overview of NNMT inhibitors (14)–(24) and human/rat/mouse liver microsome (HLM/RLM/MLM) data.
HLM/RLM/MLM
% Remaining at 1 h
Compound Number
14
R1
R2
R3
R4
R5
hNNMT IC50 [µM]
mNNMT IC50 [µM]
OH
Me
F
H
0.49
3.48
<5/NA/<5
15
16
17
OH
OH
OH
Me
Me
Me
F
F
F
H
H
H
0.79
1.1
>3
5.1
1.63
NA
<5/<5/<5
<5/<5/<5
27/<5/8
18
19
OH
OH
Me
Me
F
F
H
H
>30
>10
NA
NA
35/<5/30
NA

Molecules 2021, 26, 991
9 of 22
Table 1. Cont.
R5
HLM/RLM/MLM
% Remaining at 1 h
Compound Number
20
R1
R2
R3
R4
hNNMT IC50 [µM]
mNNMT IC50 [µM]
OH
Me
F
H
H
>3
NA
18/NA/<5
21
22
OH
OH
Me
Me
F
F
0.25
>30
1.64
NA
20/<5/<5
16/31/<5
H
23
24
OH
OH
Me
Me
F
F
H
H
0.15
>10
1.0
<5/NA/<5
<5/NA/<5
NA

Molecules 2021, 26, 991
10 of 22
Table 2. Overview of NNMT inhibitors (1)–(13) and human/rat/mouse liver microsome (HLM/RLM/MLM) data.
HLM/RLM/MLM
% Remaining at 1 h
Compound Number
1
R1
R2
R3
R4
R5
hNNMT IC50 [µM]
mNNMT IC50 [µM]
OH
Me
H
H
0.26
0.84
NA
6
H
H
H
H
1.67
2.2
NA
7
8
9
OH
OH
OH
Me
Me
H
F
F
F
H
H
H
0.30
>10
0.88
0.44
NA
2.43
30/<5/<5
<5/<5/<5
NA

Molecules 2021, 26, 991
11 of 22
Table 2. Cont.
R5
HLM/RLM/MLM
% Remaining at 1 h
Compound Number
10
R1
R2
R3
R4
hNNMT IC50 [µM]
mNNMT IC50 [µM]
OH
Me
F
F
H
0.29
>3
7/NA/<5
11
12
OH
OH
Me
Me
F
0.2
1.87
<5/NA/<5
NA
H
OMe
0.008
0.0008
13
OH
Me
F
H
>3
NA
73/22/22

Molecules 2021, 26, 991
12 of 22
Substitution of the phenyl ring on the pyrazole moiety with fluoro in the meta position
7) delivered a two-fold increase in mNNMT enzymatic potency: however, difluoro sub-
(
stituted derivative (10) showed a mNNMT IC50 of >3
µM. The 6-substituted isoquinoline
compound ( ) completely lost hNNMT activity. Addition of a fluoro atom at the 6-position
8
of the isoquinoline moiety (11) retains potency for both hNNMT and mNNMT. Interestingly,
substitution of a methoxy as R5 in (12) resulted in a nanomolar potency for both hNNMT
and mNMMT (Table 2). Even though these compounds showed reasonable potency for
both hNNMT and mNNMT, their metabolic stability was insufficient for further profiling.
We envisaged that one reason for the lack of metabolic stability is related to oxidation of
the isoquinoline moiety. The N-oxide of isoquinoline (13) showed improved metabolic
stability, but its potency for hNNMT and mNNMT was >3 µM.
Based on the crystal structure of hNNMT, co-crystallized with (
1), we predicted the
binding mode of (12) using the Glide docking module of the Schrodinger 2016-2 release [26
]
to guide further optimization. Derivative (12) binds by replacing the SAH: its isoquinoline
group occupies the same position as the adenine of SAH and the piperazine residue acts
as a replacement for the ribose moiety of SAH. The substituted pyrazole sub-unit of (12
binds into the nicotinamide pocket and is connected to the other part of the molecule
by a 2-hydroxy-2-methyl-propionic carboxamide-based linker. Inhibitor (12) shows polar
interactions with the backbone amino group of Val-143, the carbonyl after the piperazine
demonstrates interactions with the side-chain of Asn-90, the hydroxyl group interacts with
the hydroxyl side-chain of Tyr-204, and the pyrazole nitrogen interacts with the phenol-OH
in Tyr-20. The ligand interaction diagram (Figures 8 and 9) also shows aromatic interactions
between the isoquinoline ring and Tyr-11, between the pyrazole ring and Tyr-20, and Tyr-
204 engages both pyrazole and terminal phenyl rings. Compound (12) is more potent
)
than the HTS hit (1), owing to a better interaction of its pyrazole group with the hydroxyl
group of Tyr-20 and the additional hydrophobic contacts of its methoxy group with the
surrounding residues such as Ala-247 and perhaps a water molecule bridging the methoxy
and surrounding polar residues such as Ser-201 and Ser-213. Since the binding mode
prediction uses the binding site of NNMT devoid of any explicit water molecules, such
water molecule interactions cannot be visualized.
Figure 8. Predicted binding mode of compound 12 (magenta), along with the crystal structure of compound
1 (green),
co-crystalized in hNNMT. The blue dashes are the polar interactions of compound 12 with the binding site amino acids. The
figure was made using PyMOL v0.99.

Molecules 2021, 26, 991
13 of 22
Figure 9. Ligand interaction diagram showing aromatic interactions of compound 12. This figure was made using the
Ligand Interaction Diagram module of Schrodinger/Maestro v2016-2.
Based on this docking analysis, we came to the conclusion that the 4-(para-methoxy-
phenyl)-pyrazol moiety present in (12) already delivered a very good fit into the nicoti-
namide binding pocket of NNMT and might not require additional optimization at this
stage. After analysis of liver microsomes data of several close analogs in Table 2, metabolic
stability clearly emerged as a very important parameter for further optimization. This
observation was not completely unexpected, as alicyclic amines such as piperazines are
well-established targets for metabolizing enzymes such as P450s or monoamine oxidases
(MAOs). [27] In addition, metabolism on the isoquinoline ring has to be considered. There-
fore, our strategy for further optimization focused on the identification of a metabolically
stable replacement of the 5-piperazinyl-isoquinoline moiety binding into the SAH pocket.
With such a replacement available to us, we expected a combination with the optimized
pyrazole residue from (12) to deliver a new NNMT inhibitor suitable for in vivo investiga-
tions.
Our X-ray crystallographic data and docking studies guided us to keep the key ni-
trogen interactions with Val143 intact, while exploring different heteroaryl ring systems.
Interestingly, quinolinone (21) showed reasonable potency, whereas the corresponding
isoquinolinone derivative (22) was not active. However, the metabolic stability of com-
pounds (21) and (23) did not improve. As can be seen from Figure 7 and Table 1, major
structural changes in this region are not allowed. The activity difference between (21) and
(24) also nicely confirms the general requirement of a heteroaromatic ring in this region of
our NNMT inhibitor series.
In our next round of optimizations, we replaced the piperazine ring system with dif-
ferent heterocyclolakyl ring systems (Table 3). Replacement of piperazine with piperdine

Molecules 2021, 26, 991
14 of 22
(
25) led to a two-fold reduction in hNNMT and mNNMT potency compared to (21). Re-
placement with azetidine, pyrrolidine, or homopiperazine ring systems (26,27,28) resulted
in a substantial loss of potency. Interestingly, (30) with a tertiary hydroxyl group on the
piperidine ring was inactive whereas (31) with an isoquinoline ring showed a three-fold
increase in potency. Even though (31) was active, the stability in liver microsomes was
insufficient for further profiling. As already observed with compound (13), the correspond-
ing N-oxide compound (32) showed improved metabolic stability for both humans and
mice, but completely lost NNMT inhibition.
Taking all results presented so far into consideration, we did not identify a suitable
replacement of the 5-piperazinyl-isoquinoline moiety featuring improved metabolic sta-
bility in combination with sufficient in vitro potency. Therefore, we had no clear pathway
forward for further optimization of our highly active NNMT inhibitor (12), and we decided
to focus our activities on different chemical series available to us.
2.5. Combination of Our Bisubstrate-Like Inhibitor Series with the HTS Leads 2 and 3
Based on the X-ray crystallographic data of our hits (
1) and (2), the question arose as
to whether the introduction of an appropriate 2-hydroxy-2-methyl-propionic carboxamide-
based linker on (
strate inhibitor.
2) is able to transform our nicotinamide pocket binder into a true bisub-
In our case, we saw the synthetic feasibility of transferring the corresponding structural
motif directly to the imino nitrogen of compound ( ) delivering compound (33) in the
2
process (Scheme 5). A subsequent X-ray analysis of (33) confirmed the binding mode as a
bisubstrate inhibitor (Figure 10).
(2)
Hybrid between
(1) and (2)
(33)
(1)
Scheme 5. Rationale design of the bisubstrate inhibitor (33).

Molecules 2021, 26, 991
15 of 22
Table 3. Overview of NNMT inhibitors (25)–(32) and human/rat/mouse liver microsome (HLM/RLM/MLM) data.
Compound Number
Structure
hNNMT IC50 [uM]
mNNMT IC50 [uM]
HLM/MLM
25
26
0.75
>3
17/<5
>10
NA
12/<5
27
>30
NA
NA

Molecules 2021, 26, 991
16 of 22
Table 3. Cont.
hNNMT IC50 [uM]
Compound Number
Structure
mNNMT IC50 [uM]
HLM/MLM
28
>3
>30
NA
29
30
>30
NA
<5/<5
>10
NA
<5/<5

Molecules 2021, 26, 991
17 of 22
Table 3. Cont.
hNNMT IC50 [uM]
Compound Number
Structure
mNNMT IC50 [uM]
HLM/MLM
31
0.11
0.56
23/<5
32
>30
NA
72/32

Molecules 2021, 26, 991
18 of 22
Figure 10. Crystal structure showing the binding mode of (33) (cyan) to hNNMT. (
a) Comparison with the crystal structure
of inhibitor (1) (blue). (b) Comparison with the crystal structure of the SAH and compound (2) complex (yellow).
As shown in Figure 10a, the binding mode in the SAH pocket of the part that is
common between (33) and ( ) is identical. The position of ( ) (Figure 10b) is slightly shifted
1
2
with respect to the position of the tricyclic in (33), but also here the positions are very
similar, showing that the fusion between the two inhibitor classes was successful.
Further profiling on human and mouse NNMT enzyme revealed that a bisubstrate
binding motif does not necessarily translate into an improved inhibitory activity (Table 4).
This may be related to the fact that a bisubstrate inhibitor has to compete with two sub-
strates, SAM and Nam, which may be more difficult to achieve than when competing with
a single substrate, especially when the second substrate, SAM, has a high affinity to the
enzyme (Km = 1.8 µM, compared to 400 µM for Nam, [2]). We also decided against further
modification of the linker as a viable optimization strategy. Based on our crystallographic
data, we obtained the correct linker length allowing the important interactions with Val-
143 and Tyr-204, and we also observed no obvious strains within the bound ligand. In
addition, our results obtained with inhibitors (
linker modifications.
6) and (9) advised against any additional
Table 4. hNNMT and mNNMT enzymatic activity of inhibitors (2) and (33).
Compound
hNNMT IC50 [µM]
mNNMT IC50 [µM]
(2)
(33)
1.3
2.5
2.3
3.0
2.6. Optimization of a Series of Tricyclic Compounds Binding to the Nam Pocket Only
In addition to the missing potency increase observed with the introduction of bisub-
strate binding, we expected the 5-piperazinyl-isoquinoline residue in (33) to have com-
parable metabolic lability issues, as observed with the previous lead series around (1).
Therefore, we decided to focus further studies on NNMT inhibitors binding into the Nam
pocket of NNMT only.
As a first strategy to maximize the space filling in this pocket, we went for ring
enlargement of the five membered ring in (
outlined in Scheme 6.
3) and synthesized (4) using a synthetic strategy

Molecules 2021, 26, 991
19 of 22
enlargement of 5-
membered ring
(3)
Scheme 6. Rational design of inhibitor (4) based on X-ray data of inhibitor (3).
(4)
Our crystal structure showed that the binding of (
) and ( ). The binding mode of ( ) and an overlay of the binding modes of all three Nam
binding analogs mentioned here is given in Figure 11.
4) is very similar to the binding of
(2
3
4
Figure 11.
(a
) Superposition of the crystal structures of the three Nam analogs (
2
) (yellow), (
3
) (green) and (4) (magenta).
The binding mode of the three inhibitors is very similar. (
b) Crystal structure showing the binding mode of (
4) in hNNMT.
Our ring enlargement strategy resulted in a clear improvement in in vitro potency, as
can be seen in Table 5.
Table 5. hNNMT and mNNMT enzymatic activity of inhibitors (3) and (4).
Compound
hNNMT IC50 [µM]
mNNMT IC50 [µM]
(3)
(4)
0.2
0.07
0.5
0.1
In addition, the metabolic lability of (
4) was in an acceptable range from a drug dis-
covery point of view and no CYP3A4 inhibition liabilities were detected. Based on these
data, ( ) became a starting point for additional Structure-Activity Realtionship (SAR) explo-
4
rations and in vivo profiling studies, which will be discussed in more detail in subsequent
publications.
3. Materials and Methods
Crystallographic Studies
Human NNMT with a His-tag and thrombin protease site, carrying the following
mutations—K100A/E101A, E103A—was produced following a protocol from the Structural
Genomics Consortium (SGC) (DOI:10.2210/pdb2iip/pdb).
However, in contrast to the published protocol, our NNMT could only be concentrated
to ~0.8 mg/mL. Thermal shift measurements showed that addition of the cofactor SAH

Molecules 2021, 26, 991
20 of 22
◦
and the inhibitor A000268687A led to an increase in melting temperature of 7 C, indicating
that these ligands stabilized NNMT. We then concentrated NNMT in the presence of SAH
and A000268687A and could reach a concentration of 6 mg/mL. This concentration was
lower than the concentration published by the SGC (17.2 mg/mL), but it was sufficient for
crystallization experiments.
Human NNMT was crystallized using the following conditions: a protein solu-
tion with 6 mg/mL NNMT, 50 mM Tris/HCl, pH 8.0, 1 mM DTT, 86 µM S-adenosyl-
L-homocysteine (SAH), 0.95 mM A000268687A and 5% v/v glycerol was equilibrated at
room temperature in a hanging drop setup against 2.2 M ammonium sulfate with 0.1 M
HEPES/Na, pH 7.6. Small crystals appeared after 1–2 weeks. Cocrystals of the other
inhibitors were obtained using similar procedures.
The crystals were cryoprotected with 25% ethylene-glycol and flash frozen in liquid
nitrogen. Data were collected at beamline PX-III of the Swiss Light source (SLS) in Villigen,
Switzerland. Data were processed with the autoproc script by Clemens Vonrhein [28] from
the global phasing consortium. The structure was solved by molecular replacement with
Phaser [29] using a single monomer from pdb entry 2iip as a starting model. Refinement
was done using Buster. [30] The final statistics are given in Table S1 of the supplement.
The space group is P1 and the cell-dimensions are around 46, 62, and 108 Å with
γ
α, β
and
around 92◦, 98◦, and 112◦. There are four independent molecules in the asymmetric
unit. This crystal form is the same as the publishe◦d crystal form (pdb entry 2iip) of 46.0,
◦ ◦
62.0, and 107.1 with α, β and γ 82.4 , 81.9 , and 68.3 , where the γ of entry 2iip corresponds
to 180◦ − our γ.
Our structures are identical within experimental error to the 2iip structure where
the root-mean-square (r.m.s.) deviations of 0.2–0.3 Å between the independent molecules
within our crystal are the same as the r.m.s. deviations with the four independent molecules
of the 2iip structure. As shown in Figure 3, the inhibitors are generally well defined in the
electron density maps and they could be fitted unambiguously.
The structures have been deposited at the PDB under accession numbers 7BKG, 7BLE,
7NBJ, 7NBM, and 7NBQ.
4. Conclusions
We have identified the first non-adenosyl based small molecule bisubstrate inhibitor
(
1) of NNMT and additional NNMT modulators (2) and (3) utilizing the nicotinamide
binding pocket alone. The binding mode of these inhibitors has been elucidated by X-ray
crystallography and we used this information for the design of the novel inhibitors (33
and (4).
)
Our rational design towards the bisubstrate inhibitor (33) starting from the fragment-
sized nicotinamide pocket binder ( ) revealed no beneficial impact on in vitro inhibitory
2
activity by addressing both binding sites of the NNMT enzyme simultaneously. In addition,
our bisubstrate inhibitors posed a significant challenge for optimization towards acceptable
metabolic stabilities while maintaining sufficient in vitro inhibitory potency.
As a consequence, we focused on the optimization of a chemical series binding only
into the nicotinamide pocket of NNMT. By using detailed knowledge on the binding mode
generated by several X-ray studies, we were able to identify (
inhibitor with a two-digit nano-molar potency on our target. Further studies on the SAR
around ( ) and characterization of the corresponding chemical series are in progress and
4), a small molecule NNMT
4
will be published as a separate article.
Supplementary Materials: The following are available online at, Table S1: Crystallographic Data and
Refinement Statistics; Experimental Procedures S1: Synthesis of compounds (
Procedures S2: Fluorescence-based NNMT assay; Experimental Procedures S3: Metabolic stability in
human and mouse liver microsomes; Figure S1: High-resolution MS and NMR spectra of ( ); Figure
S2: High-resolution MS and NMR spectra of ( ); Figure S3: High-resolution MS and NMR spectra of
); Figure S4: High-resolution MS and NMR spectra of ( ); Figure S5: High-resolution MS and NMR
spectra of (33).
1) to (33); Experimental
1
2
(3
4

Molecules 2021, 26, 991
21 of 22
Author Contributions: Conceptualization, A.K., S.R. (Sven Ruf); Methodology, H.S. (X-ray); Investi-
gation: G.Z., J.C., M.S.H., I.S.; Writing—Original Draft, A.K., S.R. (Sven Ruf), H.S., S.R. (Sridharan
Rajagopal); Writing—Review & Editing, A.K., S.R. (Sven Ruf), H.S., G.Z., S.R. (Sridharan Rajagopal);
Visualization: H.S., R.K.; Supervision: A.K., S.R. (Sven Ruf), S.D. All authors have read and agreed to
the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data is contained within the article or supplementary material.
Acknowledgments: We thank Alexander Liesum for crystallizing hNNMT and mounting and
freezing the crystals for data collection and Thomas Langer for conceiving and performing the
thermal shift experiments.
Conflicts of Interest: The authors declare no conflict of interest.
Sample Availability: Samples of the compounds are not available from the authors.
References
1.
Pissios, P. Nicotinamide N-Methyltransferase: More Than a Vitamin B3 Clearance Enzyme. Trends. Endocrinol. Metab. 2017, 28,
340–353. [CrossRef] [PubMed]
2.
Aksoy, S.; Szumlanski, C.L.; Weinshilboum, R.M. Human liver nicotinamide N-methyltransferase. cDNA cloning, expression,
and biochemical characterization. J. Biol. Chem. 1994, 269, 14835–14840. [CrossRef]
3.
4.
GTEx Consortium. The genotype-tissue expression (GTEx) project. Nat. Genet. 2013, 45, 580–585. [CrossRef]
Ulanovskaya, O.A.; Zuhl, A.M.; Cravatt, B.F. NNMT promotes epigenetic remodeling in cancer by creating a metabolic methyla-
tion sink. Nat. Chem. Biol. 2013, 9, 300–306. [CrossRef]
5.
6.
7.
8.
9.
Kannt, A.; Rajagopal, S.; Kadnur, S.V.; Suresh, J.; Bhamidipati, R.K.; Swaminathan, S.; Hallur, M.S.; Kristam, R.; Elvert, R.; Czech,
J.; et al. A small molecule inhibitor of Nicotinamide N-methyltransferase for the treatment of metabolic disorders. Sci. Rep. 2018
8, 3660. [CrossRef] [PubMed]
,
Brachs, S.; Polack, J.; Brachs, M.; Jahn-Hofmann, K.; Elvert, R.; Pfenninger, A.; Bärenz, F.; Margerie, D.; Mai, K.; Spranger, J.; et al.
Genetic Nicotinamide N-Methyltransferase (Nnmt) Deficiency in Male Mice Improves Insulin Sensitivity in Diet-Induced Obesity
but Does Not Affect Glucose Tolerance. Diabetes 2019, 68, 527–542. [CrossRef] [PubMed]
Aoyama, K.; Matsubara, K.; Okada, K.; Fukushima, S.; Shimizu, K.; Yamaguchi, S.; Uezono, T.; Satomi, M.; Hayase, N.; Ohta, S.;
et al. N-methylation ability for azaheterocyclic amines is higher in Parkinson’s disease: Nicotinamide loading test. J. Neural.
Transm. 2000, 107, 985–995. [CrossRef]
Kraus, D.; Yang, Q.; Kong, D.; Banks, A.S.; Zhang, L.; Rodgers, J.T.; Pirinen, E.; Pulinilkunnil, T.C.; Gong, F.; Wang, Y.C.; et al.
Nicotinamide N-methyltransferase knockdown protects against diet-induced obesity. Nature 2014, 508, 258–262. [CrossRef]
[PubMed]
Lee, Y.H.; Nair, S.; Rousseau, E.; Allison, D.B.; Page, G.P.; Tataranni, P.A.; Bogardus, C.; Permana, P.A. Microarray profiling of
isolated abdominal subcutaneous adipocytes from obese vs non-obese Pima Indians: Increased expression of inflammation-related
genes. Diabetologia 2005, 48, 1776–1783. [CrossRef] [PubMed]
10. Kannt, A.; Pfenninger, A.; Teichert, L.; Tönjes, A.; Dietrich, A.; Schön, M.R.; Klöting, N.; Blüher, M. Association of nicotinamide-N-
methyltransferase mRNA expression in human adipose tissue and the plasma concentration of its product, 1-methylnicotinamide,
with insulin resistance. Diabetologia 2015, 58, 799–808. [CrossRef] [PubMed]
11. Liu, M.; Li, L.; Chu, J.; Zhu, B.; Zhang, Q.; Yin, X.; Jiang, W.; Dai, G.; Ju, W.; Wang, Z.; et al. Serum N(1)-Methylnicotinamide Is
Associated With Obesity and Diabetes in Chinese. J. Clin. Endocrinol. Metab. 2015, 100, 3112–3117. [CrossRef]
12. Salek, R.M.; Maguire, M.L.; Bentley, E.; Rubtsov, D.V.; Hough, T.; Cheeseman, M.; Nunez, D.; Sweatman, B.C.; Haselden, J.N.;
Cox, R.D.; et al. A metabolomic comparison of urinary changes in type 2 diabetes in mouse, rat, and human. Physiol. Genomics.
2007, 29, 99–108. [CrossRef]
13. Bañales-Luna, M.; Figueroa-Vega, N.; Marín-Aragón, C.I.; Perez-Luque, E.; Ibarra-Reynoso, L.; Gallardo-Blanco, H.L.; López-
Aguilar, I.; Malacara, J.M. Associations of nicotinamide-N-methyltransferase, FTO and IRX3 genetic variants with body mass
index and resting energy expenditure in Mexican subjects. Sci. Rep. 2020, 10, 11478. [CrossRef] [PubMed]
14. Souto, J.C.; Blanco-Vaca, F.; Soria, J.M.; Buil, A.; Almasy, L.; Ordoñez-Llanos, J.; Martín-Campos, J.M.; Lathrop, M.; Stone, W.;
Blangero, J.; et al. A genomewide exploration suggests a new candidate gene at chromosome 11q23 as the major determinant of
plasma homocysteine levels: Results from the GAIT project. Am. J. Hum. Genet. 2005, 76, 925–933. [CrossRef] [PubMed]
15. Sazci, A.; Ozel, M.D.; Ergul, E.; Aygun, C. Association of nicotinamide-N-methyltransferase gene rs694539 variant with patients
with nonalcoholic steatohepatitis. Genet. Test Mol. Biomark. 2013, 17, 849–853. [CrossRef] [PubMed]

Molecules 2021, 26, 991
22 of 22
16. Zhu, X.J.; Lin, Y.J.; Chen, W.; Wang, Y.H.; Qiu, L.Q.; Cai, C.X.; Xiong, Q.; Chen, F.; Chen, L.H.; Zhou, Q.; et al. Physiological Study
on Association between Nicotinamide N-Methyltransferase Gene Polymorphisms and Hyperlipidemia. Biomed. Res. Int. 2016, 45,
50. [CrossRef] [PubMed]
17. Neelakantan, H.; Vance, V.; Wetzel, M.D.; Wang, H.L.; McHardy, S.F.; Finnerty, C.C.; Hommel, J.D.; Watowich, S.J. Selective and
membrane-permeable small molecule inhibitors of nicotinamide N-methyltransferase reverse high fat diet-induced obesity in
mice. Biochem. Pharmacol. 2018, 147, 141–152. [CrossRef] [PubMed]
18. Ruf, S.; Hallur, M.S.; Anchan, N.K.; Swamy, I.N.; Murugesan, K.R.; Sarkar, S.; Narasimhulu, L.K.; Putta, V.P.R.K.; Shaik, S.;
Chandrasekar, D.V.; et al. Novel nicotinamide analog as inhibitor of nicotinamide N-methyltransferase. Bioorg. Med. Chem. Lett.
2018, 28, 92–925. [CrossRef]
19. Van Haren, M.J.; Taig, R.; Kuppens, J.; Toraño, J.S.; Moret, E.E.; Parsons, R.B.; Sartini, D.; Emanuelli, M.; Martin, N.I. Inhibitors
of nicotinamide N-methyltransferase designed to mimic the methylation reaction transition state. Org. Biomol. Chem. 2017, 15,
6656–6667. [CrossRef]
20. Babault, N.; Allali-Hassani, A.; Li, F.; Fan, J.; Yue, A.; Ju, K.; Liu, F.; Vedadi, M.; Liu, J.; Jin, J. Discovery of Bisubstrate Inhibitors of
Nicotinamide N-Methyltransferase (NNMT). J. Med. Chem. 2018, 61, 1541–1551. [CrossRef]
21. Policarpo, R.L.; Decultot, L.; May, E.; Kuzmicˇ, P.; Carlson, S.; Huang, D.; Chu, V.; Wright, B.A.; Dhakshinamoorthy, S.; Kannt,
A.; et al. High-Affinity Alkynyl Bisubstrate Inhibitors of Nicotinamide N-Methyltransferase (NNMT). J. Med. Chem. 2019, 62,
9837–9873. [CrossRef]
22. Gao, Y.; van Haren, M.J.; Moret, E.E.; Rood, J.J.M.; Sartini, D.; Salvucci, A.; Emanuelli, M.; Craveur, P.; Babault, N.; Jin, J.; et al.
Bisubstrate Inhibitors of Nicotinamide N-Methyltransferase (NNMT) with Enhanced Activity. J. Med. Chem. 2019, 62, 6597–6614.
[CrossRef] [PubMed]
23. Chen, D.; Li, L.; Diaz, K.; Iyamu, I.D.; Yadav, R.; Noinaj, N.; Huang, R. Novel Propargyl-Linked Bisubstrate Analogues as
Tight-Binding Inhibitors for Nicotinamide N-Methyltransferase. J. Med. Chem. 2019, 62, 10783–10797. [CrossRef]
24. Lee, H.-Y.; Suciu, R.M.; Horning, B.D.; Vinogradova, E.V.; Ulanovskaya, O.A.; Cravatt, B.F. Covalent inhibitors of nicotinamide
N-methyltransferase (NNMT) provide evidence for target engagement challenges in situ. Bioorg. Med. Chem. Lett. 2018, 28,
2682–2687. [CrossRef]
25. Peng, Y.; Sartini, D.; Pozzi, V.; Wilk, D.; Emanuelli, M.; Yee, V.C. Structural Basis of Substrate Recognition in Human Nicotinamide
N-Methyltransferase. Biochemistry 2011, 50, 7800–7808. [CrossRef]
26. Friesner, R.A.; Banks, J.L.; Murphy, R.B.; Halgren, T.A.; Klicic, J.J.; Mainz, D.T.; Repasky, M.P.; Knoll, E.H.; Shelley, M.; Perry, J.K.;
et al. Glide: A new approach for rapid, accurate docking and scoring. 1. method and assessment of docking accuracy. J. Med.
Chem. 2004, 47, 1739–1749. [CrossRef]
27. Bolleddula, J.; DeMent, K.; Driscoll, J.P.; Worboys, P.; Brassil, P.J.; Bourdet, D.L. Biotransformation and bioactivation reactions of
alicyclic amines in drug molecules. Drug Metab. Rev. 2014, 46, 379–419. [CrossRef]
28. Vonrhein, C.; Flensburg, C.; Keller, P.; Sharff, A.; Smart, O.; Paciorek, W.; Womack, T.; Bricogne, G. Data processing and analysis
with the autoPROC toolbox. Acta Crystallogr. Sect. D Biol. Crystallogr. 2011, 67, 293–302. [CrossRef] [PubMed]
29. McCoy, A.J.; Grosse-Kunstleve, R.W.; Adams, P.D.; Winn, M.D.; Storoni, L.C.; Read, R.J. Phaser crystallographic software. J. Appl.
Crystallogr. 2007, 40, 658–674. [CrossRef] [PubMed]
30. Bricogne, G.; Blanc, E.; Brandl, M.; Flensburg, C.; Keller, P.; Paciorek, W.; Roversi, P.; Sharff, A.; Smart, O.S.; Vonrhein, C.; et al.
Buster; Global Phasing Ltd.: Cambridge, UK, 2011.