Fragment-Based Design of a Potent MAT2a Inhibitor and in Vivo Evaluation in an MTAP Null Xenograft Model

Article abstract of DOI:10.1021/acs.jmedchem.1c00067

MAT2a is a methionine adenosyltransferase that synthesizes the essential metabolite S-adenosylmethionine (SAM) from methionine and ATP. Tumors bearing the co-deletion of p16 and MTAP genes have been shown to be sensitive to MAT2a inhibition, making it an attractive target for treatment of MTAP-deleted cancers. A fragment-based lead generation campaign identified weak but efficient hits binding in a known allosteric site. By use of structure-guided design and systematic SAR exploration, the hits were elaborated through a merging and growing strategy into an arylquinazolinone series of potent MAT2a inhibitors. The selected in vivo tool compound 28 reduced SAM-dependent methylation events in cells and inhibited proliferation of MTAP-null cells in vitro. In vivo studies showed that 28 was able to induce antitumor response in an MTAP knockout HCT116 xenograft model.

Full text of DOI:10.1021/acs.jmedchem.1c00067

pubs.acs.org/jmc
Article
Fragment-Based Design of a Potent MAT2a Inhibitor and in Vivo
Evaluation in an MTAP Null Xenograft Model
Claudia De Fusco,* Marianne Schimpl,* Ulf Börjesson, Tony Cheung, Iain Collie, Laura Evans,
Priyanka Narasimhan, Christopher Stubbs, Mercedes Vazquez-Chantada, David J. Wagner,
Michael Grondine, Matthew G. Sanders, Sharon Tentarelli, Elizabeth Underwood, Argyrides Argyrou,
James M. Smith, James T. Lynch, Elisabetta Chiarparin, Graeme Robb, Sharan K. Bagal,
and James S. Scott
Cite This: https://doi.org/10.1021/acs.jmedchem.1c00067
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: MAT2a is a methionine adenosyltransferase that
synthesizes the essential metabolite S-adenosylmethionine (SAM)
from methionine and ATP. Tumors bearing the co-deletion of p16
and MTAP genes have been shown to be sensitive to MAT2a
inhibition, making it an attractive target for treatment of MTAP-
deleted cancers. A fragment-based lead generation campaign
identified weak but efficient hits binding in a known allosteric
site. By use of structure-guided design and systematic SAR
exploration, the hits were elaborated through a merging and
growing strategy into an arylquinazolinone series of potent MAT2a
inhibitors. The selected in vivo tool compound 28 reduced SAM-
dependent methylation events in cells and inhibited proliferation of
MTAP-null cells in vitro. In vivo studies showed that 28 was able to induce antitumor response in an MTAP knockout HCT116
xenograft model.
target protein. The co-deletion of MTAP gene and resulting
sensitization to MAT2a inhibition represent a tremendous
opportunity to circumvent this issue and provide an option
for therapeutic intervention. MTAP-deficient cancer cells are
unable to metabolize methylthioadenosine (MTA) as part of
the methionine salvage pathway7 and thereby accumulate
intracellular MTA, a potent and selective inhibitor of the
methyl transferase PRMT5.4 This inhibition renders these
cancer cells vulnerable to further PRMT5 inhibition either
directly or indirectly through inhibition of the upstream
metabolic enzyme MAT2a.
Few MAT2a inhibitors were known until very recently.
Methionine analogues have been reported as inhibitors of
SAM synthesis,8,9 but being amino acid mimetics and having
low potency have limited their applicability. The first drug-
like MAT2a inhibitor described in the literature was PF-9366
(Figure 1C).10 Its binding mode was characterized by X-ray
INTRODUCTION
■
The methionine adenosyltransferases MAT1a and MAT2a
synthesize the key cofactor S-adenosylmethionine (SAM)
from methionine and ATP and play an important role in
cellular growth and survival.1 In adults, MAT1a is primarily
expressed in the liver and plays an important role in
metabolizing the bulk of dietary methionine. In contrast,
MAT2a is mainly expressed in extrahepatic tissues and its
expression in the liver is primarily limited to cases of cellular
dedifferentiation, such as in fetal liver, or upon liver injury or
hepatocellular carcinoma (HCC).2 A link between MAT2a
and other cancers was demonstrated when a shRNA screen in
390 cancer cell lines identified MAT2a knockdown, along
with knockdown of PRMT5 and PRMT5 cofactors, as
synthetic lethal in MTAP-deficient cancer cells.3,4 Marjon and
colleagues demonstrated in vivo that an inducible knockdown
of MAT2a reduced tumor growth in an MTAP null MCF7
tumor model.5
Received: January 14, 2021
The MTAP gene is colocated on chromosome 9p21 with
the p16 gene (MTS1/CDK4I/CDKN2A), and homozygous
co-deletion of MTAP and p16 genes occurs in 80−90% of
tumors with p16 deletion.6 p16/CDKN2 deletion is among
the most frequent cancer-driving mutations, but targeting a
loss of function mutant is not trivial due to the lack of a
© XXXX The Authors. Published by
American Chemical Society
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX
A

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
Figure 1. MAT2a is a functional homodimer with two catalytic sites and a site of allosteric regulation located at the dimer interface. (A) The
two MAT2a monomers are depicted as cartoon in blue and gray. At the interface, the reaction products S-adenosylmethionine (SAM) and (β-γ-
imido)triphosphate are captured in one of the two active sites (sticks with yellow carbons). Binding of the carboxy-terminal peptide of MAT2b
(purple cartoon) to the dimer interface leads to allosteric regulation of SAM biosynthesis (PDB code 4ndn, chains A, B, E). (B) Surface view of
MAT2a. Gray shading depicts cavities at the dimer interface, and the largest of these harbor the active and allosteric sites. (C) Recently
reported allosteric inhibitors of MAT2a.
Figure 2. Fragment binding induces changes in the allosteric site. (A) Fragment hits 1 and 2 and their MAT2a binding affinities as determined
by SPR. (B) Crystal structure of 2 (PDB code 7bhs, 1.1 Å) shows two molecules of the ligand (green sticks and yellow surface) occupying the
symmetric site at the dimer interface. Overlay of the MAT2b-bound state (purple sticks and turquoise surface) illustrates the cryptic pocket
accessed by small-molecule ligands. (C, D) Close-up view of the binding site and interactions formed by triazinone hit 1 (cyan, PDB code 7bhr,
1.1 Å) and quinazoline hit 2 (green, PDB code 7bhs, 1.1 Å) with Arg313.
crystallography, which revealed that PF-9366 acted as an
allosteric inhibitor binding at the interface of two MAT2a
subunits, at a site of regulatory control otherwise occupied by
the regulatory subunit MAT2b (Figure 1).11 The authors
showed target engagement for PF-9366 and that the
compound was able to modulate SAM synthesis in cancer
cells with an IC50 of 1.2 μM. However, this did not translate
to inhibition of cell proliferation. The upregulation of MAT2a
gene expression as a feedback mechanism to MAT2a
inhibition was suggested to be the cause of lack of translation.
B
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
However, given the weak cell potency of PF-9366, the
question remained if a more potent compound would be able
to overcome the effect of the feedback loop and MAT2a
upregulation. In 2017 Agios disclosed their in vivo tool AGI-
25696 (Figure 1C) and its related chemical series in a
number of patents,12−14 and in the following year, the first
clinical trial with a MAT2a inhibitor, AG-270, was opened,
enrolling patients with advanced solid tumors or lymphomas
with homozygous MTAP deletion.
This paper describes a fragment-based approach to
discovering new MAT2a inhibitors. Several marketed drugs
and a growing number of clinical candidates have been
generated by fragment-based drug discovery, which is based
on the identification of small but efficient hits through highly
sensitive and robust biophysical methods.15,16 One of the
underlying principles is the efficient sampling of chemical
space by screening molecules with restricted molecular
weight. Consequently, the most productive ligand−protein
interactions and binders with high ligand efficiency can be
identified.17 With high quality binders in hand and relying
heavily on structure-based design, it is possible via strategies
of fragment merging or linking to combine desirable features
and to generate more complex and higher affinity
compounds. Hits can also be expanded by adding atoms or
functional groups to target new interactions and/or fill
available subpockets (growing).18 Evolution of the hits
through merging, linking, and growing strategies can
effectively provide advanced leads while closely controlling
the physicochemical properties.
occupying one-half of the binding site (Figure 2B). Ligand
binding caused the side chain of Phe333 to rotate by 180°,
reshaping the site from a short and wide pocket to a long and
narrow cleft. In its new orientation, Phe333 partially occluded
solvent access (Figure S1). Moreover, the position vacated by
this side chain formed a cryptic pocket occupied by the
ethoxy group of 1 and the chlorine of 2 (Figure 2B−D).
Both fragments formed a bidentate interaction with Arg313
(Figure 2C,D), which assumed slightly different conforma-
tions in the two crystal structures. The smaller compound, 1,
protruded into the cryptic pocket with its ethoxy substituent,
while the dimethylamine pointed toward solvent. There was
an additional contact to a water molecule in the subsite that
harbors a His or Phe side chain in the MAT2b complex with
the dimeric MAT2a (“aromatic side chain pocket”, Figure
2B). Compound 2 displaced this water and occupied the
subsite with a phenyl moiety while positioning a chlorine in
the cryptic pocket. Given that this site ordinarily harbors a
phenylalanine side chain, it offers predominantly lipophilic
contacts.
It was important to determine whether binding of
compounds at the allosteric site translated to inhibition of
enzyme activity. The hits were, therefore, further charac-
terized in an enzymatic functional assay; the assay followed
the formation of free phosphate (Experimental Section),
which is a product of the ATP dependent MAT2a-catalyzed
reaction converting methionine to SAM (eq S1).
Despite its small size and modest affinity, 1 showed
inhibition with an IC50 of 206 μM, which correlated well with
its binding affinity, while 2 was surprisingly inactive in this
assay.
A biophysical fragment screen against MAT2a identified
two highly efficient hits binding in the allosteric site. These
fragments were merged to generate a new core characterized
by higher ligand efficiency and able to inhibit the catalytic
activity of MAT2a. A growing strategy was then adopted,
resulting in compound 28, which is a potent and orally
bioavailable compound that was used to investigate the effect
of MAT2a inhibition in vivo.
Initial SAR and Fragment Merging. Initial exploration
of triazinone 1 showed a very tight SAR, with close analogues
having no affinity for MAT2a. Key compounds are
summarized in Table 1. Methylation of the triazinone NH
Table 1. Initial SAR Exploration around Hit 1
RESULTS AND DISCUSSION
■
Fragment Screening and Hits. The hit finding strategy
comprised three parallel screening approaches: high through-
put thermal shift/differential scanning fluorimetry assay of the
AstraZeneca fragment library, surface plasmon resonance
(SPR) of a high solubility subset of the library, and a
cocktail-based crystallographic screen of a different subset.
Each of these methods yielded a number of hits that were
validated in the screening cascade (data not shown). The
most potent and efficient hits originated from the SPR screen
and were triazinone 1 and quinazoline 2 with respective
affinities of 250 μM and 6.2 μM in SPR and high ligand
efficiency (LE) of 0.38 for both. Given the low log D7.4
(0.23) of compound 1, its lipophilic ligand efficiency (LLE)
was a promising 3.4, and overall 1 was deemed a good
starting point for a hit-to-lead campaign (Figure 2A).
Quinazoline 2 was a highly lipophilic compound, with a
log D7.4 of 4.1 and a LLE of 1.1. Crystal structures were
obtained of MAT2a in complex with the fragment hits,
showing that they bound in the same site as the regulatory
subunit MAT2b. This site has two identical halves, being
located at the dimer interface and on the rotational 2-fold
axis of cyclic C2 symmetry (Figures 1 and S1). While
MAT2b binds once per dimer in a 1:2 molar ratio, small-
molecule ligands bound with 1:1 stoichiometry, each ligand
a
b
c
b
d
compd
R
log D7.4
Kd (μM)
LE
IC50 (μM)
LLE
3.4
1
3
4
H
Me
Ph
0.23
0.66
1.1
250
>200
40
0.38
206
>1000
307
0.32
b
2.4
a
log D7.4 determined by shake flask method. All Kd and IC50 data are
the mean values of at least n ≥ 3 independent measurements. Each
has a SD 0.3 log units unless otherwise noted. See Supporting
c
Information for details. Ligand efficiency (LE) = 1.4·pKd/(heavy
d
atom count). Lipophilic ligand efficiency (LLE) = pIC50 − log D7.4.
in 3 led to loss of binding, likely due to the perturbation of
the water network and the inefficient filling of the aromatic
side chain pocket. Conversely, a much bigger phenyl
substituent in compound 4 was tolerated and led to a ∼6-
fold increase in binding affinity (Kd = 40 μM vs 250 μM of
the initial hit), but unexpectedly, this did not translate into an
increase in functional activity.
Following these initial findings, a different core was
proposed based on the overlay of the two fragment hits 1
and 2 (Figure 3). This new core, 2-quinazolinone 5, not only
C
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
Figure 3. Merging strategy from hits 1 and 2 to quinazolinone 5. Binding mode confirmed by crystallography (PDB code 7bht, 1.1 Å) shows
that key interactions are maintained, notably the bidentate hydrogen bonding to Arg313 and hydrophobic stacking with Phe333. Compound 5
is visibly nonplanar as seen in the electron density (blue mesh shows 2Fo − Fc map contoured at 1σ); see also Figure S3.
maintained the binding affinity of the most potent of the two
hits but also had a much better ligand efficiency due to its
small size (LE = 0.49 vs 0.38 of each of the two hits).
Moreover, 5 had a significant improvement in inhibitory
potency of 8.3 μM and given its log D7.4 of 1.3 had a
promising LLE of 3.8. The quinazolinone was found to bind
as expected, maintaining the bidentate interaction of 1 with
Arg313 and occupying more fully the lipophilic cryptic
pocket where the ethoxy substituent of 1 was located. The
Phe333 side chain displaced from this site participated in a
hydrophobic stacking interaction with the ligand, concom-
itantly shielding the compound from solvent. The ligand
participated in two additional π-stacking interactions with the
side chains of Phe20 and Trp274. The crystal structure also
clearly showed a distinct divergence from planarity in this
compound, with the dimethylamine group on the one hand,
and the carbonyl O atom on the other, being on opposite
sides of the extended plane of the phenyl ring (Figure 3).
This substituent effect on aromatic ring conformation
appeared to originate from steric interaction between the
aromatic CH in position 5 and the alkylamine substituent in
position 4.
The binding site of 5 was analyzed computationally by
characterizing putative water sites using the software
WaterMap19,20 (Figure S2). Three water sites, corresponding
to the crystallographic waters seen near N1, were determined
computationally to have excess free energies ΔG significantly
greater than 0 compared to bulk water (i.e., “unstable”). The
water site nearest N1, at about 3.4 Å distance, had the
greatest calculated excess free energy (ΔG = +6.8 kcal/mol)
and was potentially displaceable by substitution off the N1
position.
Table 2. Key SAR around the Quinazolinone Core and
Alternative 4-Substituents
a
compd
R
log D7.4
IC50 (μM)
LLE
3.8
5
6
7
8
N(Me)2
1.3
0.4
1.7
1.6
1.2
1.3
2.0
2.7
2.4
8.3
>100
>100
55
>100
>100
>100
>100
>100
H
NHMe
NH2
NHAc
OMe
OPh
Ph
2.7
9
10
11
12
13
a
Legend as in Table 1.
reasons, ranging from disrupting the water network around
the Arg313, steric clashes, and/or the inability of the
compounds to adopt a nonplanar conformation similar to
that of compound 5 (Figure 3).
Further investigation focused on exploring the tertiary
amine SAR, and the results are summarized in Table 3.
Compound 14 bearing a methylethylamine substituent
showed an improvement in both potency and LLE compared
to 5 (IC50 = 0.74 μM, LLE = 4.4 vs IC50 = 8.3 μM, LLE =
3.8). In an attempt to make an interaction with Glu342, a
base was appended to the ethyl substituent (15), but this
change abolished all activity. An N-linked phenyl substituent
was not tolerated (16), while N-benzyl 17 had some residual
activity but was ∼10-fold weaker than the starting point and
with a much lower LLE. An ethyl linker to the Ph resulted in
a complete loss of potency (18). Cyclic amines were then
investigated, and while the azetidine 19 was inactive, the
larger pyrrolidine substituent in 20 had a similar activity to 5.
A contributing factor to this piece of SAR may be the
peculiar substituent effect on the conformation of the
quinazolinone ring system. Quantum-mechanical calculations
SAR Understanding of the Quinazolinone Core and
Fragment Growing. With this new core in hand, the first
step was to understand the key pharmacophoric elements.
The chlorine and the dimethylamine substituents were both
found to be essential for activity, with compounds 6 and 7
being inactive (Table 2). Methylamine 8 lost activity (IC50
=
55 μM), and primary amine 9 was completely inactive.
Acetylating the amine (compound 10) was not tolerated. The
O-linked compounds (11 and 12) and the phenyl analogue
13 were also inactive. This could be due to a combination of
D
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
coplanar substituents (Figure S3). The observed out-of-plane
conformations of 5 and 20 may therefore contribute favorably
to their binding to MAT2a. Despite comparable activity to
compound 5, pyrrolidine 20 had a higher log D7.4 and a
slightly lower LLE of 3.5. Both increasing the size of the ring
to six atoms (21) or substitution of the 5-membered ring
(22, 23, and 24) did not improve potency.
Table 3. Fragment Growth from the Tertiary Amine in
Position 4
a
Given the encouraging results from expanding triazinone 4
into the aromatic side chain pocket (Table 1), substitution of
the endocyclic N became the following area of exploration
with data summarized in Table 4.21 Me-substituted 25 was
∼20-fold weaker than the starting point, consistent with the
result obtained by methylating the triazinone (3 vs 1). As
stated before, this was believed to be due to perturbation of
the water network and the formation of a high-energy void.
In order to compensate for the loss of water-mediated
interactions with the pocket, it was envisaged that a hydroxyl
group at the right distance would be able to form a hydrogen
bond with the backbone carbonyl of Gly273. Gratifyingly,
compound 26 was found to be equipotent with 5, but given
its lower log D7.4, its LLE increased to 4.4 compared to 3.8
for 5, suggesting that the compound was able to efficiently
displace the water and form a further productive interaction
within the pocket. The crystal structure confirmed this
predicted interaction, with the compound adopting two
distinct conformations that positioned the hydroxyl within
c
hydrogen bonding distance of Gly273 (Figure 4A). The Pr
substituent in 27 was also tolerated but did not bring any
advantage in terms of potency or efficiency. In the original
triazinone core the H to phenyl modification had led to a
∼6-fold improvement in affinity but not to an increase in
biochemical potency. Conversely, in this elaborated scaffold,
the addition of a Ph substituent led to 28 with a biochemical
potency of 22 nM and a LLE of 5.6, bringing an almost 400-
fold increase in potency compared to the parent 5. The
crystal structure showed the additional ring effectively
displacing two water molecules and interacting with Phe333
in a CH−π interaction. The greatly enhanced potency and
efficiency of 28 can be rationalized in terms of the good
complementarity between the ligand and the pocket as well
as the low rotational freedom of the ligand in the unbound
state. Also, the displacement of two “unstable” water
molecules (Figures S2 and 4B) likely contributed to its
potency gain. From further analysis of the crystal structure, a
small lipophilic pocket was identified and a methyl
substituent in the meta position was envisaged to fill it.
The resulting compound 29 had a biochemical potency of 16
nM, similar to 28, but because it was more lipophilic, the
LLE was not improved. Given the water network present in
the aromatic side chain pocket in the absence of the Ph, it
was hypothesized that a nitrogen in the ring could stabilize
this network by interacting with one of the waters. 3-Pyridyl
31 was found to be optimal with an IC50 of 21 nM and an
impressive LLE of 6.9. The crystal structure (Figure 4D)
showed a perfect overlay with the phenyl analogue 28 and a
water-mediated interaction between the pyridine N of the
ligand and the protein. The electron-poor nature of the
pyridine likely increased the ability of the proton in position
5 to form a weak CH-hydrogen bond with the Gly273
carbonyl group, all contributing to the potency and the
efficiency of 31.
a
Legend as Table 1.
suggested that the active compounds 5 and 20 had energy
minima that involve a clear divergence from planarity of the
quinazolinone ring system and substituents being partially
out-of-plane (as observed by X-ray for 5 bound to MAT2a
and described above; Figure 3), while the inactive 19 had an
essentially flat predicted minimum-energy conformation with
Activity in Cell and Combination of SAR. A selection
of compounds were profiled in a cellular assay which
E
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
a
Table 4. Growth from Position 1
a
b
Legend as Table 1. Standard deviation is 0.35.
measures symmetric dimethyl arginine (SDMA) levels.
Symmetric arginine methylation is a downstream measure
of MAT2a activity through PRMT5, which is inhibited by
being starved of its substrate SAM. The phenyl quinazolinone
28 showed a potent IC50 of 25 nM in this assay. Given the
promising effect in cells, the following step was to relate
MAT2a inhibition to the ability of the compounds to inhibit
cellular proliferation. Gratifyingly, compound 28 was able to
inhibit proliferation of HCT116 MTAP knockout cells with
an IC50 of 250 nM. Pyridine 31 showed a similar cellular
potency and proliferation inhibition for a lower log D.
A final round of optimization combined the N-4-ethyl,
identified from compound 14, with the optimal aryl groups
from 28 and 31 generating 32 and 33, respectively. Both
these compounds showed single digit nanomolar potency in
cells and proliferation inhibition potencies of 140 nM.
Pharmacokinetics. In vitro DMPK properties of com-
pounds 28 and 3133 were profiled. These four compounds
displayed moderate to high solubility, high permeability, and
low efflux, and they also showed low to moderate human
hepatocyte intrinsic clearance (CLint) (Table 6) which
generally tracked with log D (Table 5). The compounds
were screened for inhibition against a panel of five
cytochrome P450 isoforms (CYP1A2, CYP2C9, CYP2C19,
CYP2D6, and CYP3A4) in human liver microsomes. Only
weak inhibition of CYP1A2 and CYP3A4 was observed for
compounds 32 and 33 (Table 6) with all others showing no
activity (IC50 > 30 μM). No compound showed activity
against the hERG ion channel up to 40 μM.
Low to moderate CLint values were observed in rat
hepatocytes which corresponded to low to moderate in vivo
rat clearance (noncompartmental) (Table 7). Bioavailability
in rat (Table 7) was moderate to high (57−100%) consistent
with high absorption expected given the in vitro DMPK
properties of these compounds. The volume of distribution
was generally moderate and consistent with the neutral ion
class (0.56−2.1 L/kg). 28 and 31 were selected to be
F
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
Figure 4. Optimization of the substituent occupying the aromatic side chain subsite. (A) The hydroxyethyl substituent in 26 forms a hydrogen
bond with the backbone carbonyl of Gly273. Two alternative conformations were observed in the crystal structure, shown as light and dark gray
sticks (PDB code 7bhu, 1.2 Å). (B) The phenyl substitution in 28 (pink) brings an almost 400-fold potency boost compared to 5 (orange).
The superposition of both crystal structures shows that the phenyl group displaces two water molecules. The side chain of Phe333 undergoes a
slight rotation, maximizing a CH−π interaction (PDB code 7bhv, 1.2 Å). (C) 29 (yellow, PDB code 7bhw, 1.2 Å) fills a small hydrophobic side
pocket with a methyl substituent. (D) 31 (green, PDB code 7bhx, 1.1 Å) features a pyridine to capture a water interaction. A weak CH-
hydrogen bond to Gly273 is indicated by gray dashes.
Table 5. Combination of Substituents in Positions 1 and
4
prolonged plasma half-life and lowered Cmax. Conversely, 31
displayed rapid absorption from both sc and po admin-
istration displaying similar profiles due to higher solubility as
solution formulations (Figure 5). Due to the favorable PK
profile of sc dosing 28, this compound and formulation were
progressed to target engagement and efficacy studies. 50 mg/
kg sc dosing of 28 was utilized in target engagement and
efficacy studies after observing similar exposure and
prolonged half-life at both 50 mg/kg and 100 mg/kg sc
after repeat dosing (Figure S4).
In Vivo Target Engagement and Efficacy. In vivo
profiling of 28 was carried out through an efficacy study in an
HCT116 MTAP knockout xenograft tumor model. 50 mg/kg
of 28 was administered once daily sc to female Ncr nude
mice and resulted in an average reduction of tumor SAM
concentration by 74% compared to vehicle control (0.5% w/v
HPMC/0.1% w/v Tween 80) (Figure 6A). Consequently, a
significant antitumor response was observed leading to tumor
stasis (Figure 6B, left). However, while compound 28 was
tolerated over the course of the 28 day treatment, mice did
exhibit a maximum body weight loss of 8% (Figure 6B,
right).
a
IC50
cell IC50
(μM)
prolif IC50
(μM)
compd
R
X
log D7.4 (μM) LLE
28
31
32
33
Me
Me
Et
C
N
C
N
2.0
0.82
2.5
0.022
0.021
0.013
0.018
5.6
6.9
5.4
6.5
0.025
0.017
0.0063
0.0052
0.25
0.33
0.14
b
Et
1.2
0.14
a
b
Legend as Table 1. Average of two repeats.
evaluated in mouse pharmacokinetic studies to understand
suitability for pharmacodynamic studies.
Unfortunately, in vitro mouse metabolism was much higher
than rat or human for these compounds, which correlated
with much higher mouse in vivo clearance when compared
with rat (Table 7). Due to the higher mouse clearance, both
subcutaneous (sc) dosing and oral (po) dosing of 28 and 31
were evaluated at high doses to enable in vivo efficacy studies
(Figure 5). The lower solubility of 28 relative to 31 (103 vs
>1000 μM) allowed for a suspension SC formulation (0.5%
w/v hydroxypropylmethylcellulose/0.1% w/v Tween 80)
which displayed dissolution rate limited kinetics that
Synthesis of Selected Compounds. Quinazolinone 5
was prepared by SNAr reaction on 2,4,7-trichloroquinazoline
5a with dimethylamine, followed by hydrolysis with acetic
acid (Scheme 1).
Quinazolinones 28 and 31−33 were synthesized as shown
in Scheme 2. Commercial 4-chloro-2-fluorobenzamide was
G
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
Table 6. In Vitro Human DMPK Properties
IC50 (μM)
human hep CLint
(μL/min)/106 cells
human
Caco2
solubility
e
e
f
compd
CYP3A4
CYP1A2
hERG
a
b
c
d
% free
Papp/ER
μM
28
31
32
33
11
<1
18
3.4
13
64
8.1
35
29/0.97
64/1.6
73/0.86
44/2.2
103
>1000
220
>30
>30
>30
24
>30
>30
18
>40
>40
>40
>40
930
>30
a
b
Rate of metabolism ((μL/min)/106 cells) determined from DMSO stock solution in isolated hepatocytes diluted to 106 cells/mL. Determined
c
from DMSO stock solution by equilibrium dialysis in 10% human plasma supplied by BioreclamationIVT. Compounds were incubated at 10 μM
in cultured Caco2 cells. Intrinsic permeability was measured in units of ×10−6 cm/s in the apical direction in the presence of transport inhibitors
quinidine (50 μM), sulfasalazine (20 μM), and benzbromarone (30 μM) with an apical/basolateral pH of 6.5:7.4, respectively. Efflux ratio is
reported as the ratio of basolateral (B to A) to apical (A to B) transport in the absence of any inhibitors and an apical/basolateral pH of 7.4:7.4.
Solubility of compounds in aqueous phosphate buffer at pH 7.4 after 24 h at 25 °C from DMSO stock solution. Inhibition of cytochrome P450
enzymes IC50. Inhibition of the hERG tail current was measured using a plate-based planar patch clamp system (Syncropatch).
d
e
f
a
Table 7. In Vitro and in Vivo Rodent DMPK Properties
compd
species
CLint ((μL/min)/106 cells)
% free
Cl ((mL/min)/kg)
Vss (L/kg)
T1/2 (h)
F (%)
28
31
32
33
28
31
rat
rat
rat
rat
mouse
mouse
4.6
1.3
13
3.9
67
10
45
3.7
20
13
56
16
19
12
6.6
109
62
0.85
2.1
0.43
0.56
1.0
1.2
3.6
1.1
3.4
0.25
3.7
100
100
100
57
25
79
14
2.2
a
Compound was dosed in rat intravenously at 0.5 and orally at 1 mg/kg. The formulation was 5% DMSO/95% hydroxylpropyl β-cyclodextrin
(30% w/v) at a volume of 1 mL/kg and 4 mL/kg, iv and po, respectively.
a
Scheme 1. Synthesis of 5
a
Reagents and conditions: (a) dimethylamine, THF, rt, 2 h; (b) acetic
acid, 100 °C, 12 h, 29% over two steps.
Intermediate 34 was carried through to the next step,
together with an oxalamide byproduct impurity (approx-
imately 20−25%) identified after purification. A KHMDS-
mediated cyclization of 34 was used to generate the
quinazolinone ring (intermediates 35a and 35b). PyBOP
coupling with the corresponding amine completed the
synthesis of the desired compounds 28, 31, 32, and 33 in
respectable yields (50−92%).
Figure 5. Mouse plasma concentrations of 28 po (blue circle), 28 sc
(red square), 31 po (green triangle), and 31 sc (purple triangle) at
100 mg/kg.
treated with oxalyl chloride to form the isocyanate in situ.
The reaction was then cooled to room temperature and the
corresponding aniline added to form acyl ureas 34a and 34b.
Figure 6. Subcutaneous dosing of 28 at 50 mg/kg QD: (A) reduction in tumor SAM; (B) antitumor response (left) and body weight change
(right).
H
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
a
Scheme 2. Synthesis of 28, 31, 32, and 33
a
Reagents and conditions: (a) oxalyl chloride, DCE, 80 °C, 30 min; (b) aniline, rt, 1 h, 62−75% over two steps; (c) KHMDS, THF, −78 °C, 1 h,
then rt, 3.5 h, 67−80%; (d) PyBOP, DBU, amine, MeCN, rt, 2.5 h, 50−92%.
1
LCMS by UV absorbance and H NMR and are ≥95% for screening
compounds.
CONCLUSIONS
■
A series of MAT2a inhibitors was developed originating from
weak fragment hits through a structure-based merging
strategy. This led to quinazolinone 5 which was a low
molecular weight, potent, and efficient compound and formed
the basis for further optimization. By use of structure-based
design, fragment growing led to compounds with over 1000-
fold improvement in potency over the original fragment hits
together with excellent physicochemical properties. Arylqui-
nazolinones 28 and 31−33 showed inhibition of PRMT5-
mediated symmetric dimethylation of arginine (SDMA) and
antiproliferative activity in MTAP null cells. Compound 28
was identified as a promising in vivo tool that exhibited
reduction in tumor SAM levels and antitumor response in an
MTAP knockout HCT116 xenograft model.
4-Chloro-2-fluoro-N-(phenylcarbamoyl)benzamide (34a).
To a stirred suspension of 4-chloro-2-fluorobenzamide (10.0 g,
57.6 mmol) in DCE (200 mL) at rt was added oxalyl dichloride
(5.36 mL, 63.4 mmol). The resultant suspension was heated to 80
°C for 30 min. The reaction mixture was allowed to cool to rt, and
aniline (10.5 mL, 115 mmol) was added to the reaction mixture (on
addition a suspension formed). The resultant reaction mixture was
stirred at rt for 1 h. The reaction mixture was cooled using an ice
bath and was then quenched by carefully pouring into a saturated
aqueous solution NaHCO3 (1 L). This was stirred for approximately
10 min. The precipitate was collected by filtration, washed with
water (1 L), and dried under vacuum to afford 4-chloro-2-fluoro-N-
(phenylcarbamoyl)benzamide 34a (75% pure; 16.8 g, 99%) as a
1
white solid. H NMR (400 MHz, DMSO): 7.12 (tt, J = 7.8, 1.1 Hz,
1H), 7.31−7.4 (m, 2H), 7.41−7.48 (m, 1H), 7.52−7.6 (m, 2H),
7.63 (dd, J = 10.3, 1.9 Hz, 1H), 7.68−7.77 (m, 1H), 10.37 (s, 1H),
11.06 (s, 1H). m/z: ES+ [M + H]+ 293.
EXPERIMENTAL SECTION
■
7-Chloro-1-phenylquinazoline-2,4(1H,3H)-dione (35a). To a
stirred white suspension of 4-chloro-2-fluoro-N-(phenylcarbamoyl)-
benzamide 34a (75% pure; 16.8 g, 43.0 mmol) in anhydrous THF
(100 mL) under a nitrogen atmosphere at −78 °C was added
KHMDS (126 mL, 1 M in THF solution, 126 mmol) dropwise. On
addition, the reaction mixture turned an orange color. The reaction
mixture was stirred at −78 °C for 1 h and then warmed to rt for 3.5
h. The reaction mixture was slowly quenched with saturated
aqueous NH4Cl solution. A precipitate was formed to give a thick
suspension, which was filtered and washed with water to afford the
crude product. The crude product was stirred in approximately 200
mL of water for 30 min. The white suspension was filtered, washing
the solids with water, dried under vacuum at 50 °C to afford 7-
chloro-1-phenylquinazoline-2,4(1H,3H)-dione 35a (85% pure; 14.7
Unless otherwise stated, commercially available reagents were used
as supplied. Preparative reverse phase HPLC was performed on an
Agilent 1290 Infinity II Preparative system equipped with a SQ MS
detector (multimode ESI/APCI source), with a Waters CSH C18
OBD column (5 μm silica, 30 mm diameter, 100 mm length, flow
rate of 50 mL/min) using decreasingly polar mixtures of water
(containing 0.10.3% aqueous ammonium) or water (containing
0.1% formic acid) and acetonitrile as eluents. Preparative SFC
purification was performed on either a Sepiatec P100 SFC system or
Waters Prep 100 SFC system equipped with QDa MS detector,
using the chromatographic conditions as detailed in corresponding
experimental data. Proton magnetic resonance spectra were
determined using a Bruker Avance 500 (500 MHz) and Bruker
Avance 400 (400 MHz) instruments and were determined in
CDCl3, DMSO-d6, or MeOH-d4. Measurements were taken at
ambient temperature unless otherwise specified. The following
abbreviations have been used: s, singlet; d, doublet; t, triplet; q,
quartet; m, multiplet; dd, doublet of doublets; ddd, doublet of
doublets of doublets; dt, doublet of triplets; tt, triplet of triplets; bs,
broad signal. End products were also characterized by mass
spectrometry following liquid chromatography (LCMS or UPLC).
Reverse-phase C18 silica was used with a flow rate of 1 mL/min,
and detection was by electrospray mass spectrometry and by UV
absorbance recording a wavelength range of 220−320 nm. Analytical
UPLC was performed on CSH C18 reverse-phase silica, using a
Waters Acquity UPLC CSH C18 column with dimensions 2.1 mm
× 50 mm and particle size 1.7 μm). Gradient analysis was employed
using decreasingly polar mixtures as eluent, for example, decreasingly
polar mixtures of water (containing 0.1% formic acid or 0.1%
ammonia) as solvent A and acetonitrile as solvent B. A typical 2 min
analytical UPLC method would employ a solvent gradient over 1.3
min, at approximately 1 mL/min, from a 97:3 mixture of solvents A
and B respectively to a 3:97 mixture of solvents A and B. The
reported molecular ion corresponds to the [M + H]+ unless
otherwise specified. For molecules with multiple isotopic patterns
(Br, Cl, etc.) the reported value is the one obtained for the lowest
isotope mass unless otherwise specified. Purities were assessed using
1
g, 80%) as a white solid. H NMR (400 MHz, DMSO): 6.32 (d, J =
1.8 Hz, 1H), 7.31 (dd, J = 8.4, 1.9 Hz, 1H), 7.42−7.5 (m, 2H),
7.53−7.68 (m, 3H), 8.01−8.08 (m, 1H), 11.79 (s, 1H). m/z: ES−
[M − H]− 271.
7-Chloro-4-(dimethylamino)-1-phenylquinazolin-2(1H)-one
(28). DBU (11.0 mL, 73.9 mmol) was added dropwise to a stirred
suspension of 7-chloro-1-phenylquinazoline-2,4(1H,3H)-dione 35a
(85% pure, 6.77 g, 21.1 mmol), PyBOP (16.5 g, 31.7 mmol), and
dimethylamine (26.4 mL, 2 M in THF, 52.8 mmol) in acetonitrile
(110 mL) at rt. The resultant suspension was stirred at rt for 2.5 h.
Additional PyBOP (4.00 g, 10.6 mmol), dimethylamine (7.0 mL, 2
M in THF, 14 mmol), and DBU (3.0 mL, 20 mmol) were added,
and the reaction mixture was left stirring at rt for 30 min. The
reaction mixture was concentrated to dryness. The crude mixture
was diluted with DCM and washed sequentially with water and
saturated brine. The organic layer was dried with a phase separating
cartridge, filtered, and concentrated to afford the crude product as a
brown oil (approximately 31 g). The crude product was dissolved in
acetonitrile (80 mL). A fine white precipitate started to crash out.
The suspension was slowly passed through a phase separator
cartridge into a stirred flask with water (800 mL). Upon addition a
solid precipitated. This was carefully filtered and the solid washed
with water. The solid was dried under vacuum at 50 °C to afford
the crude product (6.78 g) as a light brown solid. The crude sample
I
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
(6.78 g) was dissolved to a concentration of 113 mg/mL in 60 mL
of DCM/DMF/DMSO (1:1:1), heated, sonicated, and filtered. The
resulting solution was purified by preparative SFC (column, YMC
Diol, 30 mm × 250 mm, 5 μm, mobile phase of 15−25% MeOH +
0.1% NH3/85−75% scCO2 over 7 min, 1.5 min flush at 50%
MeOH; flow rate of 100 mL/min; BPR, 120 bar; column
temperature of 40 °C). Fractions containing the desired product
were concentrated to dryness to afford 7-chloro-4-(dimethylamino)-
1-phenylquinazolin-2(1H)-one 28 (5.80 g, 92%) as an off white
dryness to afford 7-chloro-4-(dimethylamino)-1-(pyridin-3-yl)-
quinazolin-2(1H)-one 31 (0.945 g, 86%) as a white solid. 1H
NMR (400 MHz, DMSO): 3.32 (s, 6H), 6.38 (d, J = 2.1 Hz, 1H),
7.23 (dd, J = 8.8, 2.1 Hz, 1H), 7.64 (ddd, J = 8.1, 4.8, 0.7 Hz, 1H),
7.84 (ddd, J = 8.1, 2.5, 1.6 Hz, 1H), 8.06 (d, J = 8.8 Hz, 1H), 8.54
(dd, J = 2.5, 0.7 Hz, 1H), 8.71 (dd, J = 4.8, 1.5 Hz, 1H). 13C NMR
(126 MHz, DMSO, 37 °C) δ: 41.4, 108.7, 113.9, 120.7, 124.7,
129.6, 134.4, 137.2, 137. 7, 146.0, 149.4, 150.0, 153.4, 163.6. m/z:
ES+ [M + H]+ 301. HRMS (ESI) for C15H13ON4Cl (MH+)
calculated 301.08507; found 301.08453.
1
solid. H NMR (400 MHz, DMSO): 3.31 (s, 6H), 6.33 (d, J = 2.1
Hz, 1H), 7.19 (dd, J = 8.8, 2.1 Hz, 1H), 7.26−7.35 (m, 2H), 7.45−
7.56 (m, 1H), 7.60 (ddd, J = 7.7, 6.5, 1.3 Hz, 2H), 8.04 (d, J = 8.8
Hz, 1H). 13C NMR (126 MHz, DMSO, 37 °C) δ: 41.3, 108.6,
114.0, 120.3, 128.5, 129.0, 129.4, 129.9, 137.4, 137.6, 146.2, 153.4,
163.5. m/z: ES+ [M + H]+ 300; HRMS (ESI) for C16H14ClN3O
(MH+) calculated 300.08982; found 300.08911.
7-Chloro-4-(ethyl(methyl)amino)-1-phenylquinazolin-
2(1H)-one (32). DBU (0.148 mL, 0.99 mmol) was added dropwise
to a stirred suspension of 7-chloro-1-phenylquinazoline-2,4(1H,3H)-
dione 35a (85% pure; 95 mg, 0.28 mmol), PyBOP (220 mg, 0.42
mmol), and N-methylethanamine (0.061 mL, 0.71 mmol) in
acetonitrile (1.0 mL) at rt. The resultant suspension was stirred at
rt for 2.5 h. Additional N-methylethanamine (0.061 mL, 0.71
mmol), PyBOP (220 mg, 0.42 mmol), and DBU (0.15 mL, 0.99
mmol) were added, and the resultant reaction mixture was stirred at
rt for 2 h. The reaction mixture was concentrated to dryness, and
the crude residue was diluted with DCM. The organic layer was
washed sequentially with water and saturated brine, dried with a
phase separating cartridge, filtered, and evaporated to afford the
crude product as a brown oil (753 mg). The crude sample (753
mg) was dissolved in DMSO (6 mL) and was purified by
preparative HPLC (Waters CSH C18 OBD column, 30 mm ×
100 mm id, 5 μm particle size), using decreasingly polar mixtures of
water (containing 0.3% NH3 aq) and MeCN as eluents, followed by
a further step using decreasingly polar mixtures of water (containing
0.1% formic acid (aq)) and MeCN as eluents. Fractions containing
the desired product were concentrated to dryness to afford 7-chloro-
4-(ethyl(methyl)amino)-1-phenylquinazolin-2(1H)-one 32 (44 mg,
4-Chloro-2-fluoro-N-(pyridin-3-ylcarbamoyl)benzamide
(34b). To a stirred suspension of 4-chloro-2-fluorobenzamide (7.50
g, 43.2 mmol) in DCE (150 mL) at rt was added oxalyl dichloride
(4.02 mL, 47.5 mmol). The resultant suspension was heated to 80
°C for 45 min. The reaction mixture was allowed to cool to rt, and
pyridin-3-amine (8.13 g, 86.4 mmol) was added. The reaction
mixture was then stirred at rt for 4 h. The reaction mixture was
cooled using an ice bath, and the resultant suspension was filtered,
washing the solids with DCM. The solid was suspended in a mixture
of MeOH/DCM (1:1). The suspension was filtered and the solid
dried under vacuum to afford 4-chloro-2-fluoro-N-(pyridin-3-
ylcarbamoyl)benzamide 34b (85% pure; 9.26 g, 73%) as a white
1
solid. H NMR (400 MHz, DMSO): 7.36−7.5 (m, 2H), 7.64 (dd, J
= 10.3, 1.9 Hz, 1H), 7.74 (t, J = 8.0 Hz, 1H), 8.10 (ddd, J = 8.4,
2.6, 1.5 Hz, 1H), 8.35 (dd, J = 4.8, 1.4 Hz, 1H), 8.79 (d, J = 2.1 Hz,
1H), 10.46 (s, 1H), 11.21 (s, 1H). m/z: ES+ [M + H]+ 294.
7-Chloro-1-(pyridin-3-yl)quinazoline-2,4(1H,3H)-dione
(35b). To a stirred solution of 4-chloro-2-fluoro-N-(pyridin-3-
ylcarbamoyl)benzamide 34b (85% pure; 9.26 g, 26.8 mmol) in
anhydrous THF (80 mL) under a nitrogen atmosphere at −78 °C
was added potassium bis(trimethylsilyl)amide (69.4 mL, 1 M in
THF, 69.4 mmol) dropwise. On addition the reaction mixture
turned a dark brown color. The reaction mixture was stirred at −78
°C for 1 h and then warmed to rt for 18 h. The reaction mixture
was slowly quenched with saturated aqueous NH4Cl solution during
which a solid precipitated to give a thick suspension, which was
filtered and washed with acetonitrile to afford the crude product
(13.4 g) as an off white solid. The crude product was stirred in
water (100 mL) for 30 min. The white suspension was filtered,
washing the solids with water and diethyl ether, dried under vacuum
at 50 °C to afford 7-chloro-1-(pyridin-3-yl)quinazoline-2,4(1H,3H)-
1
50%) as a white solid. H NMR (500 MHz, DMSO, 37 °C) 1.32 (t,
J = 7.0 Hz, 3H), 3.30 (s, 3H), 3.71 (q, J = 7.0 Hz, 2H), 6.34 (d, J =
2.1 Hz, 1H), 7.20 (dd, J = 8.8, 2.1 Hz, 1H), 7.27−7.35 (m, 2H),
7.47−7.57 (m, 1H), 7.54−7.65 (m, 2H), 7.97 (d, J = 8.8 Hz, 1H).
13C NMR (126 MHz, DMSO, 37 °C) δ: 12.1, 38.6, 47.0, 108.6,
114.0, 120.4, 128.5, 129.0, 129.1, 129.9, 137.4, 137.6, 146.3, 153.4,
163.0. m/z: ES+ [M + H]+ 314. HRMS (ESI) for C17H16ON3Cl
(MH+) calculated 314.10547; found 314.10486.
7-Chloro-4-(ethyl(methyl)amino)-1-(pyridin-3-yl)-
quinazolin-2(1H)-one (33). DBU (0.12 mL, 0.80 mmol) was
added dropwise to a stirred suspension of 7-chloro-1-(pyridin-3-
yl)quinazoline-2,4(1H,3H)-dione (80% pure; 125 mg, 0.370 mmol),
PyBOP (247 mg, 0.480 mmol), and N-methylethanamine (0.047
mL, 0.55 mmol) in acetonitrile (3.0 mL) at rt. The resulting yellow
solution was stirred at room temperature for 2 h. Additional DBU
(0.12 mL, 0.80 mmol) was added, and the reaction mixture was
warmed to 50 °C for 10 min. The reaction mixture was
concentrated to dryness, and the crude residue was diluted with
DCM. The organic was washed sequentially with water and
saturated brine, dried with a phase separating cartridge, filtered,
and concentrated to dryness to afford the crude product (600 mg).
The crude product was dissolved in DMF (2.0 mL) and was
purified by preparative HPLC (Waters CSH C18 OBD column, 30
mm × 100 mm id, 5 μm particle size), using decreasingly polar
mixtures of water (containing 0.3% NH3 aq) and MeCN as eluents.
Fractions containing the desired compound were concentrated to
dryness to afford 7-chloro-4-(ethyl(methyl)amino)-1-(pyridin-3-yl)-
1
dione 35b (80% pure; 6.16 g, 84%) as a white solid. H NMR (400
MHz, DMSO): 6.40 (d, J = 1.7 Hz, 1H), 7.34 (dd, J = 8.4, 1.9 Hz,
1H), 7.68 (ddd, J = 8.1, 4.8, 0.8 Hz, 1H), 7.98 (ddd, J = 8.1, 2.5,
1.6 Hz, 1H), 8.06 (d, J = 8.4 Hz, 1H), 8.68 (dd, J = 2.5, 0.7 Hz,
1H), 8.76 (dd, J = 4.8, 1.5 Hz, 1H), 11.76 (s, 1H). m/z: ES+ [M +
H]+ 274.
7-Chloro-4-(dimethylamino)-1-(pyridin-3-yl)quinazolin-
2(1H)-one (31). DBU (1.91 mL, 12.8 mmol) was added dropwise
to a stirred suspension of 7-chloro-1-(pyridin-3-yl)quinazoline-
2,4(1H,3H)-dione 35b (80% pure; 1.25 g, 3.65 mmol), PyBOP
(2.85 g, 5.48 mmol), and dimethylamine (4.57 mL, 2 M in THF,
9.13 mmol) in acetonitrile (30 mL) at rt. The resulting yellow
solution was stirred at rt for 18 h. The reaction mixture was
concentrated to dryness. The crude mixture was diluted with DCM
and washed sequentially with water and saturated brine. The organic
layer was dried with a phase separating cartridge, filtered, and
concentrated to dryness to afford the crude product as an orange oil
(approximately 4 g). The sample was dissolved in DMSO and
purified by preparative SFC (column, YMC Cyano 30 × 250 mm, 5
μm, mobile phase of 10−12% MeOH + 0.1% NH3/9088% scCO2
over 8 min, 1.5 min flush at 50% MeOH + 0.1% NH3; flow rate of
100 mL/min; BPR of 120 bar; column temperature of 40 °C).
Fractions containing the desired product were concentrated to
1
quinazolin-2(1H)-one 33 (69.0 mg, 60%) as a white solid. H NMR
(400 MHz, DMSO): 1.31 (t, J = 7.1 Hz, 3H), 3.32 (s, 3H), 3.72 (q,
J = 7.0 Hz, 2H), 6.38 (d, J = 2.1 Hz, 1H), 7.24 (dd, J = 8.8, 2.1 Hz,
1H), 7.64 (ddd, J = 8.1, 4.8, 0.8 Hz, 1H), 7.85 (ddd, J = 8.1, 2.5,
1.6 Hz, 1H), 7.99 (d, J = 8.8 Hz, 1H), 8.55 (dd, J = 2.5, 0.7 Hz,
1H), 8.71 (dd, J = 4.8, 1.5 Hz, 1H). 13C NMR (126 MHz, DMSO,
37 °C) δ: 12.0, 38.7, 47.0, 108.8, 114.0, 120.8, 124.7, 129.3, 134.4,
137.2, 137.6, 146.0, 149.4, 150.1, 153.4, 163.1. m/z: ES+ [M + H]+
315. HRMS (ESI) for C16H15ON4Cl (MH+) calculated 315.10072;
found 315.09995.
J
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
Surface Plasmon Resonance. SPR experiments were per-
formed using a Biacore 4000, Biacore T200, or Biacore 8K
instrument (Cytiva) at 25 °C. A series S sensor chip SA (Cytiva)
was docked into the system in 20 mM Bicine-Na, pH 7.5, 100 mM
KCl, 5 mM MgCl2, 0.1 mM TCEP, 0.05% (w/w) Tween-20, and
2% (v/v) DMSO. The sensor chip surface was conditioned with 3 ×
60 s injections of 50 mM NaOH/1 M NaCl at 30 μL/min before
capturing biotinylated Avi-MAT2A protein diluted to 100 nM in
running buffer on the active flow cells (3000−5000 RU captured).
All remaining biotin binding sites were blocked on the active and
reference surfaces with a 60 s injection of 100 μM amine-PEG2-
biotin (ThermoScientific) in running buffer. Compounds were
tested over a suitable dose−response range by either multicycle
kinetics or single-cycle kinetics (slowly dissociating compounds).
Biochemical Assay. Inhibition of MAT2a activity was
determined using BIOMOL GREEN Reagent (Enzo Life Sciences
BML-KI102) which measures the free phosphate released during the
assay. For protein production, see the Supporting Information.
Assays were performed in 384-well plates (Greiner 781101) in 20
μL volume, 1% (v/v) final DMSO concentration and with the
following buffer conditions: 50 mM Tris-HCl (pH 7.4), 5 mM
MgCl2, 100 mM KCl, 1 mM DTT, and 0.01% Brij23. Twelve point
half-log compound concentration−response curves, with a top
concentration of 100 μM (1 mM for fragments), were generated
from 10 mM (100 mM for fragments) stocks of compounds
solubilized in DMSO using an Echo 555 (Labcyte Inc., Sunnyvale,
CA). 10 μL of enzyme (MAT2a and inorganic pyrophosphatase)
and 10 μL of substrate (methionine and ATP) were added to the
compound plate, and reaction was incubated at room temperature
for 90 min. Final concentrations in the reaction mix were 40 nM
MAT2a, 1 μg/mL inorganic pyrophosphatase, 20 μM methionine,
and 70 μM ATP. Enzyme reaction was stopped by addition of 40
μL of BIOMOL GREEN reagent to each well and incubated for 20
min. Absorbance at 620 nm was read using an Envision plate reader.
IC50 values were calculated using a Genedata Screener (Genedata
AG, Basel, Switzerland).
SDMA Imaging. HCT116 KO-MTAP cells were cultured in cell
media composed of McCoys media (Sigma no. M8403), 10% (v/v)
fetal calf serum, and 1% (v/v) L-glutamine. After harvesting, cells
were dispensed into black, 384-well Costar plates (no. 3712,
Corning) to give 2000 cells per well in a total volume of 40 μL of
cell media. Test compounds and reference controls were dosed
directly into the cell plates using a Labcyte Echo 555 acoustic
dispenser following a 12 point half-log compound concentration−
response with a top concentration of 30 μM.
The cell plates were then incubated for 48 h at 37 °C before
being fixed by the addition of 40 μL of 8% paraformaldehyde in
PBS/A (4% final concentration), followed by a 10 min rt incubation.
Plates were then washed twice with 150 μL of PBS using a BioTek
ELx406 platewasher, permeabilized for 10 min with 20 μL/well
0.1% saponin in PBS, washed again, and blocked with 20 μL/well
2% BSA in PBS-T (Sigma no. A8022) for 1 h at rt.
After normalization to maximum signal/top (vehicle control) and
minimum signal/bottom (30 μM of PRMT5 inhibitor GSK-
3326595,22 CAS no. 1616392-22-3), data were fitted to a four-
parameter (Hill) IC50 model.
Proliferation Assay. HCT116 KO MTAP cells were cultured in
cell media composed of McCoys media (Sigma no. M8403), 10%
(v/v) fetal calf serum, and 1% (v/v) L-glutamine. After harvesting,
cells were dispensed into black, 384-well Costar plates (no. 3712,
Corning) to give 400 cells per well in a total volume of 40 μL of
cell media. Test compounds and reference controls were dosed
directly into the cell plates using a Labcyte Echo 555 acoustic
dispenser following a 12 point duplicate half-log compound
concentration−response with a top concentration of 30 μM.
A day 0 plate is created in parallel but not dosed, after plating 4
μL of Alamar Blue (Thermo no. DAL1100) and incubating for 3 h
at 37 °C, 5% CO2. Following incubation, plates are read using an
EnVision plate reader with fluorescence excitation wavelength of
540−570 nm (peak excitation is 570 nm), fluorescence emission at
580−610 nm (peak emission is 585 nm).
The dosed cell plates were then incubated for 5 d at 37 °C before
adding Alamar blue and being read in the Envision following the
same protocol as for first plate. Data were analyzed, and IG50 values
were calculated using Genedata Screener software by normalizing to
plate 0.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.jmedchem.1c00067.
■
sı
Supplementary figures and table; extended methods
(PDF)
Molecular formula strings (CSV)
Accession Codes
Crystal structures of MAT2a small molecule complexes have
been deposited in the PDB under the following accession
codes: 1 (7bhr), 2 (7bhs), 5 (7bht), 26 (7bhu), 28 (7bhv),
29 (7bhw), 31 (7bhx). Authors will release the atomic
coordinates and experimental data upon article publication.
AUTHOR INFORMATION
Corresponding Authors
■
Claudia De Fusco − Discovery Sciences, R&D, AstraZeneca,
Cambridge CB4 0WG, United Kingdom; orcid.org/
0000-0001-8807-4976; Email: claudia.defusco@
gmail.com
Marianne Schimpl − Discovery Sciences, R&D, AstraZeneca,
Cambridge CB4 0WG, United Kingdom; orcid.org/
0000-0003-2284-5250; Email: marianne.schimpl@
astrazeneca.com
Primary anti-SDMA histone 4 antibody (Milipore no. 07-947)
was diluted 1:1000 in PBS-T + 0.05% BSA, 20 μL added per well,
and plates were incubated at 4 °C overnight. Cell plates were
washed 3× with 200 μL of PBS/T, and then 20 μL of 1:500
dilution in assay buffer of Alexa Fluor 488 goat anti-rabbit IgG
secondary antibody (Thermo no. A11008,), with a 1:1000 dilution
of Hoechst 33342, was added per well. Following a 1 h incubation
at rt, plates were washed 3× with 200 μL of PBS/T, and 40 μL of
PBS without Ca, Mg, and Na Bicarb (Gibco no. 14190-094) was
added per well.
Stained cell plates were covered with black seals and then read on
the Cell Insight imaging platform (Thermo Scientific), with a 10×
objective. The primary channel (Hoechst blue fluorescence 405 nm,
BGRFR_386_23) is used to autofocus and to count the number of
events (this will provide information about cytotoxicity of the
compounds tested). The secondary channel (Green 488 nM,
BGRFR_485_20) measures SDMA staining. Data were analyzed,
and IC50 values were calculated using Genedata Screener software.
Authors
Ulf Börjesson − Discovery Sciences, R&D, AstraZeneca,
Gothenburg SE-431 83, Sweden
Tony Cheung − Oncology R&D, AstraZeneca, Boston,
Massachusetts 02451, United States
Iain Collie − Discovery Sciences, R&D, AstraZeneca,
Cambridge CB4 0WG, United Kingdom
Laura Evans − Oncology R&D, AstraZeneca, Cambridge
CB4 0WG, United Kingdom
Priyanka Narasimhan − Discovery Sciences, R&D,
AstraZeneca, Cambridge CB4 0WG, United Kingdom
Christopher Stubbs − Discovery Sciences, R&D,
AstraZeneca, Cambridge CB4 0WG, United Kingdom
Mercedes Vazquez-Chantada − Discovery Sciences, R&D,
AstraZeneca, Cambridge CB4 0WG, United Kingdom
K
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
David J. Wagner − Oncology R&D, AstraZeneca, Boston,
Massachusetts 02451, United States; orcid.org/0000-
0002-5805-930X
Michael Grondine − Oncology R&D, AstraZeneca, Boston,
Massachusetts 02451, United States
symmetric dimethyl arginine; SFC, supercritical fluid
chromatography; SNAr, aromatic nucleophilic substitution;
SPR, surface plasmon resonance; t1/2, half-life; Vss, volume of
distribution at steady state
Matthew G. Sanders − Oncology R&D, AstraZeneca,
Cambridge CB4 0WG, United Kingdom
Sharon Tentarelli − Oncology R&D, AstraZeneca, Boston,
Massachusetts 02451, United States
Elizabeth Underwood − Discovery Sciences, R&D,
AstraZeneca, Cambridge CB4 0WG, United Kingdom
Argyrides Argyrou − Discovery Sciences, R&D, AstraZeneca,
Cambridge CB4 0WG, United Kingdom; orcid.org/
0000-0003-3141-9122
James M. Smith − Oncology R&D, AstraZeneca, Cambridge
CB4 0WG, United Kingdom; orcid.org/0000-0002-
6750-265X
REFERENCES
■
(1) Lu, S. C.; Mato, J. M. S-adenosylmethionine in liver health,
injury, and cancer. Physiol. Rev. 2012, 92 (4), 1515−1542.
(2) Ramani, K.; Mato, J. M.; Lu, S. C. Role of methionine
adenosyltransferase genes in hepatocarcinogenesis. Cancers 2011, 3
(2), 1480−1497.
(3) McDonald, E. R., III; de Weck, A.; Schlabach, M. R.; Billy, E.;
Mavrakis, K. J.; Hoffman, G. R.; Belur, D.; Castelletti, D.; Frias, E.;
Gampa, K.; Golji, J.; Kao, I.; Li, L.; Megel, P.; Perkins, T. A.;
Ramadan, N.; Ruddy, D. A.; Silver, S. J.; Sovath, S.; Stump, M.;
Weber, O.; Widmer, R.; Yu, J.; Yu, K.; Yue, Y.; Abramowski, D.;
Ackley, E.; Barrett, R.; Berger, J.; Bernard, J. L.; Billig, R.;
Brachmann, S. M.; Buxton, F.; Caothien, R.; Caushi, J. X.; Chung,
F. S.; Cortés-Cros, M.; deBeaumont, R. S.; Delaunay, C.; Desplat,
A.; Duong, W.; Dwoske, D. A.; Eldridge, R. S.; Farsidjani, A.; Feng,
F.; Feng, J.; Flemming, D.; Forrester, W.; Galli, G. G.; Gao, Z.;
Gauter, F.; Gibaja, V.; Haas, K.; Hattenberger, M.; Hood, T.;
Hurov, K. E.; Jagani, Z.; Jenal, M.; Johnson, J. A.; Jones, M. D.;
Kapoor, A.; Korn, J.; Liu, J.; Liu, Q.; Liu, S.; Liu, Y.; Loo, A. T.;
Macchi, K. J.; Martin, T.; McAllister, G.; Meyer, A.; Mollé, S.;
Pagliarini, R. A.; Phadke, T.; Repko, B.; Schouwey, T.; Shanahan, F.;
Shen, Q.; Stamm, C.; Stephan, C.; Stucke, V. M.; Tiedt, R.;
Varadarajan, M.; Venkatesan, K.; Vitari, A. C.; Wallroth, M.; Weiler,
J.; Zhang, J.; Mickanin, C.; Myer, V. E.; Porter, J. A.; Lai, A.; Bitter,
H.; Lees, E.; Keen, N.; Kauffmann, A.; Stegmeier, F.; Hofmann, F.;
Schmelzle, T.; Sellers, W. R. Project DRIVE: A Compendium of
Cancer Dependencies and Synthetic Lethal Relationships Uncovered
by Large-Scale, Deep RNAi Screening. Cell 2017, 170 (3), 577−592.
(4) Kryukov, G. V.; Wilson, F. H.; Ruth, J. R.; Paulk, J.; Tsherniak,
A.; Marlow, S. E.; Vazquez, F.; Weir, B. A.; Fitzgerald, M. E.;
Tanaka, M.; Bielski, C. M.; Scott, J. M.; Dennis, C.; Cowley, G. S.;
Boehm, J. S.; Root, D. E.; Golub, T. R.; Clish, C. B.; Bradner, J. E.;
Hahn, W. C.; Garraway, L. A. MTAP deletion confers enhanced
dependency on the PRMT5 arginine methyltransferase in cancer
cells. Science (Washington, DC, U. S.) 2016, 351 (6278), 1214−8.
(5) Marjon, K.; Cameron, M. J.; Quang, P.; Clasquin, M. F.;
Mandley, E.; Kunii, K.; McVay, M.; Choe, S.; Kernytsky, A.; Gross,
S.; Konteatis, Z.; Murtie, J.; Blake, M. L.; Travins, J.; Dorsch, M.;
Biller, S. A.; Marks, K. M. MTAP Deletions in Cancer Create
Vulnerability to Targeting of the MAT2A/PRMT5/RIOK1 Axis.
Cell Rep. 2016, 15 (3), 574−587.
James T. Lynch − Oncology R&D, AstraZeneca, Cambridge
CB4 0WG, United Kingdom
Elisabetta Chiarparin − Oncology R&D, AstraZeneca,
Cambridge CB4 0WG, United Kingdom; orcid.org/
0000-0002-2998-1346
Graeme Robb − Oncology R&D, AstraZeneca, Cambridge
CB4 0WG, United Kingdom; orcid.org/0000-0002-
4531-4375
Sharan K. Bagal − Oncology R&D, AstraZeneca, Cambridge
CB4 0WG, United Kingdom
James S. Scott − Oncology R&D, AstraZeneca, Cambridge
CB4 0WG, United Kingdom; orcid.org/0000-0002-
2263-7024
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.jmedchem.1c00067
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version
of the manuscript.
Notes
The authors declare the following competing financial
interest(s): At the time of writing, all authors were employees
of, and owned shares in, AstraZeneca PLC.
ACKNOWLEDGMENTS
■
(6) Zhang, H.; Chen, Z.-H.; Savarese, T. M. Codeletion of the
genes for p16INK4, methylthioadenosine phosphorylase, interferon-
α1, interferon-β1, and other 9p21 markers in human malignant cell
lines. Cancer Genet. Cytogenet. 1996, 86 (1), 22−28.
(7) Albers, E. Metabolic characteristics and importance of the
universal methionine salvage pathway recycling methionine from 5′-
methylthioadenosine. IUBMB Life 2009, 61 (12), 1132−1142.
(8) Lombardini, J. B.; Coulter, A. W.; Talalay, P. Analogues of
Methionine as Substrates and Inhibitors of the Methionine
Adenosyltransferase Reaction. Mol. Pharmacol. 1970, 6 (5), 481−
499.
The authors gratefully acknowledge Nin Guan’s valued
contribution to preparing samples for mass spectrometry
analysis; Crystal Brown’s significant effort in carrying out
compound tolerability and efficacy studies; Jason Breed’s help
with crystallographic data collection; Eva Lenz and Paul
Davey for compound purity checks; and Ian Storer’s
contributions regarding the manuscript.
ABBREVIATIONS USED
■
(9) Lombardini, J. B.; Sufrin, J. R. Chemotherapeutic potential of
methionine analogue inhibitors of tumor-derived methionine
adenosyltransferases. Biochem. Pharmacol. 1983, 32 (3), 489−495.
ATP, adenosine triphosphate; AUC, area under the curve;
CL, clearance; DCM, dichloromethane; DMPK, drug
metabolism and pharmacokinetics; ER, efflux ratio; hep,
hepatocyte; KHMDS, potassium bis(trimethylsilyl)amide; LE,
ligand efficiency; LLE, ligand lipophilic efficiency; MAT,
methionine adenosyltransferase; MDCK, Madin−Darby ca-
nine kidney; MTA, methylthioadenosine; Papp, apparent
permeability; po, per os (oral dosing); PyBOP, benzotriazol-
1-yl-oxytripyrrolidinophosphonium hexafluorophosphate;
SAM, S-adenosylmethionine; sc, subcutaneous; SDMA,
(10) Quinlan, C. L.; Kaiser, S. E.; Bolanos, B.; Nowlin, D.;
̃
Grantner, R.; Karlicek-Bryant, S.; Feng, J. L.; Jenkinson, S.;
Freeman-Cook, K.; Dann, S. G.; Wang, X.; Wells, P. A.; Fantin,
V. R.; Stewart, A. E.; Grant, S. K. Targeting S-adenosylmethionine
biosynthesis with a novel allosteric inhibitor of Mat2A. Nat. Chem.
Biol. 2017, 13 (7), 785−792.
(11) Murray, B.; Antonyuk, S. V.; Marina, A.; Van Liempd, S. M.;
Lu, S. C.; Mato, J. M.; Hasnain, S. S.; Rojas, A. L. Structure and
L
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
pubs.acs.org/jmc
Article
function study of the complex that synthesizes S-adenosylmethio-
nine. IUCrJ 2014, 1, 240−249.
(12) Konteatis, Z. D.; Sui, Z.; Travins, J. M.; Ye, Z. Pyrazolo[1,5-
a]pyrimidin-7(4H)-ones as inhibitors of cellular metabolic processes
and their preparation. Patent US20180079753 A1, 2018.
(13) Konteatis, Z. D.; Li, M.; Reznik, S. K.; Sui, Z.; Travins, J. M.
Preparation of azaheterobicyclic inhibitors of MAT2A and methods
of use for treating cancer. Patent WO2020139992 A1, 2020.
(14) Konteatis, Z. D.; Li, M.; Liu, P.; Medeiros, M.; Reznik, S. K.;
Sui, Z.; Travins, J. M.; Popovici-Muller, J.; Zhou, S.; Ma, G.
Preparation of heterobicyclic compounds as inhibitors of MAT2A
and methods of use for treating cancer. Patent WO2019191470 A1,
2019.
(15) Erlanson, D. A.; Fesik, S. W.; Hubbard, R. E.; Jahnke, W.;
Jhoti, H. Twenty years on: the impact of fragments on drug
discovery. Nat. Rev. Drug Discovery 2016, 15 (9), 605−619.
(16) Jacquemard, C.; Kellenberger, E. A bright future for fragment-
based drug discovery: what does it hold? Expert Opin. Drug
Discovery 2019, 14 (5), 413−416.
(17) Boyd, S. M.; de Kloe, G. E. Fragment library design:
efficiently hunting drugs in chemical space. Drug Discovery Today:
Technol. 2010, 7 (3), e173−e180.
(18) Cheron, N.; Jasty, N.; Shakhnovich, E. I. OpenGrowth: An
Automated and Rational Algorithm for Finding New Protein
Ligands. J. Med. Chem. 2016, 59 (9), 4171−4188.
(19) Abel, R.; Young, T.; Farid, R.; Berne, B. J.; Friesner, R. A.
Role of the active-site solvent in the thermodynamics of factor Xa
ligand binding. J. Am. Chem. Soc. 2008, 130 (9), 2817−2831.
(20) Young, T.; Abel, R.; Kim, B.; Berne, B. J.; Friesner, R. A.
Motifs for molecular recognition exploiting hydrophobic enclosure
in protein-ligand binding. Proc. Natl. Acad. Sci. U. S. A. 2007, 104
(3), 808−813.
(21) As this manuscript was in preparation a patent application
(Alam, M.; Cleary, L.; Fleury, M.; Pei, Z.; Steel, R.; Sutton, J.; Knox,
J. E.; Newby, Z. E. R. 2-Oxoquinazoline derivatives as methionine
adenosyltransferase 2A inhibitors and their preparation. Patent
WO2020123395 A1, 2020) from Ideaya was published on the
chemical series described in this paper.
(22) Duncan, K. W.; Chesworth, R.; Boriack-Sjodin, P. A.;
Munchhof, M. J.; Jin, L.; Penebre, E.; Barbash, O. I. Prmt5
Inhibitors and Uses Thereof. Patent WO2016022605 A1, 2016.
NOTE ADDED AFTER ASAP PUBLICATION
A new reference 11 was added May 5, 2021.
■
M
https://doi.org/10.1021/acs.jmedchem.1c00067
J. Med. Chem. XXXX, XXX, XXX−XXX