Property Focused Structure-Based Optimization of Small Molecule Inhibitors of the Protein-Protein Interaction between Menin and Mixed Lineage Leukemia (MLL)

Article abstract of DOI:10.1021/acs.jmedchem.5b01305

Development of potent small molecule inhibitors of protein-protein interactions with optimized druglike properties represents a challenging task in lead optimization process. Here, we report synthesis and structure-based optimization of new thienopyrimidine class of compounds, which block the protein-protein interaction between menin and MLL fusion proteins that plays an important role in acute leukemias with MLL translocations. We performed simultaneous optimization of both activity and druglike properties through systematic exploration of substituents introduced to the indole ring of lead compound 1 (MI-136) to identify compounds suitable for in vivo studies in mice. This work resulted in the identification of compound 27 (MI-538), which showed significantly increased activity, selectivity, polarity, and pharmacokinetic profile over 1 and demonstrated a pronounced effect in a mouse model of MLL leukemia. This study, which reports detailed structure-activity and structure-property relationships for the menin-MLL inhibitors, demonstrates challenges in optimizing inhibitors of protein-protein interactions for potential therapeutic applications.

Full text of DOI:10.1021/acs.jmedchem.5b01305

Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES
Article
Property focused structure-based optimization of small
molecule inhibitors of the protein-protein interaction
between menin and Mixed Lineage Leukemia (MLL)
Dmitry Borkin, Jonathan Wesley Pollock, Katarzyna Kempinska, Trupta Purohit, Xiaoqin Li, Bo Wen, Ting
Zhao, Hongzhi Miao, Shirish Shukla, Miao He, Duxin Sun, Tomasz Cierpicki, and Jolanta Grembecka
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.5b01305 • Publication Date (Web): 08 Jan 2016
Downloaded from http://pubs.acs.org on January 11, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Property focused structure-based optimization of small molecule inhibitors of the protein-
protein interaction between menin and Mixed Lineage Leukemia (MLL)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a
a
a
a
b
Dmitry Borkin , Jonathan Pollock , Katarzyna Kempinska , Trupta Purohit , Xiaoqin Li ,Bo
b
b
a
a
b
b
a
Wen , Ting Zhao , Hongzhi Miao , Shirish Shukla , Miao He , Duxin Sun , Tomasz Cierpicki ,
a,*
Jolanta Grembecka
a
Department of Pathology, University of Michigan, Ann Arbor, MI, 48109, USA
b
College of Pharmacy, University of Michigan, Ann Arbor, MI, 48109, USA
*
Corresponding author;
Jolanta Grembecka, PhD
Assistant Professor, Department of Pathology, University of Michigan
1150 West Medical Center Dr, MSRB I, Room 4510D
Ann Arbor, MI, 48108
eꢀmail: jolantag@umich.edu
Tel. 734ꢀ615ꢀ9319
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Abstract
Development of potent small molecule inhibitors of proteinꢀprotein interactions with optimized
drugꢀlike properties represents a challenging task in lead optimization process. Here, we report
synthesis and structureꢀbased optimization of new thienopyrimidine class of compounds, which
block the proteinꢀprotein interaction between menin and MLL fusion proteins that plays an
important role in acute leukemias with MLL translocations. We performed simultaneous
optimization of both activity and drugꢀlike properties through systematic exploration of
substituents introduced to the indole ring of lead compound 1 (MIꢀ136) to identify compounds
suitable for in vivo studies in mice. This work resulted in the identification of compound 27 (MIꢀ
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
538), which showed significantly increased activity, selectivity, polarity and pharmacokinetic
profile over 1 and demonstrated a pronounced effect in a mouse model of MLL leukemia. This
study, which reports detailed structureꢀactivity and structureꢀproperty relationships for the
meninꢀMLL inhibitors, demonstrates challenges in optimizing inhibitors of proteinꢀprotein
interactions for potential therapeutic applications.
2
ACS Paragon Plus Environment

Page 3 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Introduction
The proteinꢀprotein interaction between menin and MLL (Mixed Lineage Leukemia) fusion
1
proteins plays a critical role in acute leukemias with translocations of the MLL gene. The MLL
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
leukemias constitute about 5ꢀ10% of acute leukemia cases in adults, both AML (acute myeloid
2
3
leukemia) and ALL (acute lymphoblastic leukemia), and ~70% of acute leukemias in infants.
4
ꢀ6
Patients with MLL leukemias do not respond well to currently available treatments,
emphasizing the need for more effective therapies.
The MLL fusion proteins, which result from the chromosomal translocations of the MLL
7,8
gene, retain the Nꢀterminal fragment of MLL (about 1400 amino acids) that is involved in the
1
,9
proteinꢀprotein interaction (PPI) with menin. Importantly, the meninꢀMLL interaction plays a
1
,10
crucial role in the MLLꢀfusion protein mediated leukemogenic transformation. Menin is
required by MLL fusion proteins for enhanced proliferation and differentiation block of leukemic
1
,11
cells. We previously demonstrated that the Nꢀterminal fragment of MLL binds with a high
9
,12,13
affinity to the large central cavity on menin
and validated that this proteinꢀprotein
12,14ꢀ16
interaction is amenable to small molecule inhibition.
All these findings imply that the
meninꢀMLL proteinꢀprotein interaction represents an attractive molecular target to develop new
targeted agents as a potential therapeutic strategy for MLL leukemia.
Despite an important role of proteinꢀprotein interactions (PPIs) in human diseases,
1
7
including cancer, targeting PPIs with small molecules is still considered a challenging task due
to large contact areas, lack of wellꢀdefined pockets and flexibility of residues at PPI
1
8,19
interfaces.
Furthermore, development of cellꢀpermeable and orally bioavailable PPI
2
0
inhibitors with optimized drugꢀlike properties represents an additional challenge. Despite these
21ꢀ25
limitations, there are a number of successful examples of small molecules blocking PPIs.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
More importantly, PPI inhibitors have been successfully advanced to clinical trials, including
2
6,27
small molecules activating apoptotic pathways, such as Bclꢀ2 inhibitors and p53 activators,
providing evidence that PPI inhibitors might serve as novel therapeutics.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Our own work led to the identification of two distinct classes of meninꢀMLL inhibitors,
1
4,15,28
which demonstrated relatively potent activity in MLL leukemia cells,
but were not
optimized for in vivo studies. More recently, we developed a new class of meninꢀMLL inhibitors
by linking the cyanoindole ring with the thienopyrimidine core through the aminopiperidine
16
linker, Figure 1a and compound 1 (MIꢀ136) in Table 1. However, this class, represented by
the lead compound 1 (IC = 31 nM, Table 1), requires further improvement of cellular activity
5
0
and drugꢀlike properties, including metabolic stability, selectivity and aqueous solubility, to
develop compounds with potential therapeutic value. In this study we report structureꢀbased
optimization of compound 1 performed by systematic exploration of substituents introduced on
the indole ring of 1, together with extensive characterization of their in vitro potency, cellular
activity, selectivity, and microsomal stability to provide structureꢀactivity and structureꢀproperty
relationships. For several selected compounds we also determined the pharmacokinetic profile
(PK) in mice. This work resulted in the identification of compound 27 (MIꢀ538), which showed
significantly improved activity, selectivity, and PK profile over 1 and demonstrated a
pronounced effect in a mouse model of MLL leukemia. Interestingly, we discovered that
simultaneous incorporation of multiple substituents optimized for each site on the scaffold does
not always positively affects the activity or drugꢀlike properties of these compounds, proving an
additional level of complexity in lead optimization. This study illustrates the difficulties and
challenges in optimizing drugꢀlike properties of PPI inhibitors and emphasizes the need for
4
ACS Paragon Plus Environment

Page 5 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
extensive medicinal chemistry efforts combined with structural studies for successful
optimization of such compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Results and Discussion
Compound 1 represents a valuable pharmacophore for development of menin-MLL
inhibitors.
We recently developed a new class of small molecule inhibitors of the meninꢀMLL interaction
1
6
represented by compound 1, Figure 1a, Table 1. This compound strongly binds to menin (K =
d
24 nM), potently inhibits menin interaction with MLL (IC = 31 nM), Table 1, Supplementary
50
Figure 1, and blocks proliferation of MLL leukemia cells at subꢀmicromolar concentrations
(GI = 0.55 µM in the MLLꢀAF9 cells, Table 1), validating that it represents a valuable
5
0
pharmacophore for meninꢀMLL inhibitors. In addition, compound 1 did not show substantial offꢀ
target activity when profiled on a panel of kinases and GPCRs, Supplementary Table 1 and 2.
Compound 1, however, requires further improvement of drugꢀlike properties, including: (i)
potency and selectivity towards MLL leukemia cells (Selectivity Index for 1, SI < 6, Table 1),
(ii) metabolic stability (T1/2 = 6.8 min in mouse liver microsomes), (iii) pharmacokinetic profile
(PK) and (iv) polarity (clogP = 4.3), implying the need for additional medicinal chemistry
efforts.
The crystal structure of the menin in complex with 1 revealed that it binds to the MLL
binding site on menin and occupies the F9 and P13 pockets, with the 2ꢀcyanoindole ring
extending beyond the P13 pocket to form a hydrogen bond with the side chain of Trp341, Figure
1a. Based on the crystal structure we concluded that additional substitutions on the 2ꢀ
cyanoindole ring of 1 could be explored by introducing substituents at positions 1, 3, 4 and 6
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
substituents R1ꢀR4, Figure 1b). Substitutions at R2 and R4 were not tested before, while only a
1
6
limited set of substituents at R1 and R3 were probed in our recent study, providing a good
rational for much more systematic structureꢀactivity relationship (SAR) combined with structureꢀ
property relationship studies.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Development of analogues with substitutions at indole nitrogen
The crystal structure of the menin in complex with compound 1 revealed that the indole
nitrogen in 1 is involved in a direct hydrogen bond with the side chain of Glu363 on menin
(Figure 1a). This nitrogen atom is partly solvent exposed, suggesting that substituents of
different size should be tolerable at this position. Furthermore, the close proximity of the side
chains of Glu363 and Glu366 (Figure 1a) supports exploration of polar groups at R1 for
hydrogen bond formation. To identify optimal R1 substituents we focused on introducing a
variety of substituents harboring hydrogen bond donor groups at R1, Table 1, with the goal to
improve polarity while retaining or improving the potency of the new analogues.
First, we tested the hydroxyethyl group at R1 (compound 2), which was designed to form
a hydrogen bond with one of the glutamic acid residues on menin, but no activity improvement
was observed as compared to 1 (Table 1). We next designed compounds with 1,2ꢀpropanediol
functional group or its derivatives (compounds 3-(RS) (MIꢀ352), 3-(R), 3-(S), 4-(R), 4-(S), 5, 6),
Table 1), to further increase capabilities for hydrogen bond formation with Glu363 and/or
Glu366. We found that neither compound 3-(RS), which represents a racemic mixture, nor
individual enantiomers of this compound (3-(R) and 3-(S)), demonstrate better inhibitory activity
than 1, Table 1. To understand the molecular basis of inhibitory activity we determined the
crystal structure of menin in complex with 3-(RS), Figure 2a, Supplementary Table 3.
6
ACS Paragon Plus Environment

Page 7 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Interestingly, we observed that both enantiomers of 3-(RS) bind to menin in the same manner
with the only difference being a distinct orientation of the secondary hydroxyl group, Figure 2a.
This hydroxyl group is solvent exposed in both enantiomers, but only in one of them (3-R) it
forms a hydrogen bond with Glu363, while the terminal hydroxyl group forms a hydrogen bond
with Glu366 in both enantiomers of 3-(RS) , Figure 2a. We next explored substitutions on the
secondary hydroxyl group (compounds 4-(R), 4-(S) and 6) and found almost no effect on the
inhibition of the meninꢀMLL interaction, as compared to 3-(RS), Table 1, while alkylation of the
terminal hydroxyl group (compound 5) resulted in ~3ꢀfold reduced inhibitory activity. Overall,
the series of R1 substitutions with the 1,2ꢀpropanediol group or its derivatives did not improve
inhibitory activity over 1 despite the formation of additional hydrogen bonds with menin by
some of these compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Incorporation of the acetamide group at R1 in compound 7 (MIꢀ389), which results in
formation of hydrogen bonds with both Glu363 and Glu366 as validated by the crystal structure
of the complex (Figure 2b), also does not improve inhibition of the meninꢀMLL interaction
(Table 1). We then introduced the αꢀmethylated acetamide group at R1 (compound 8), which
showed about 10ꢀfold reduced inhibitory activity likely due to a limited space to accommodate
an additional methyl group at this region of the binding site, Table 1.
Out of aromatic substituents at R1 (compounds 9 and 10, Table 1), we have recently
explored a few 5ꢀmember heterocycles harboring hydrogen bond donor groups, out of which the
16
4
ꢀmethylpyrazole derivative (compound 10) showed best inhibitory activity (IC = 23 nM).
50
Here, we followed up on these findings and introduced the 4ꢀmethyltriazole substituent at R1
compound 9) to further increase polarity of these compounds. Importantly, we were able to
(
retain strong inhibitory activity of compound 9, while improving polarity represented by a
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
significantly lower clogP value when compared to 1, Table 1. Overall, despite the fact we did
not improve substantially the in vitro inhibitory activity by substituting at R1 of 1, the majority
of these compounds represent very potent inhibitors of the meninꢀMLL interactions with the IC50
values at the range of 25 nM ꢀ 45 nM, Table 1. Importantly, these new analogues are more polar
than 1, reflected by the clogP values below 3.5 for most of them, which affects their cellular
activity and other drugꢀlike properties (see below).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In the next step we assessed cellular activities and microsomal stabilities of the series of
meninꢀMLL inhibitors with various R1 substituents. Interestingly, despite similar in vitro
activity, these compounds demonstrate substantially different cellular activity in MLL leukemia
cells (ranging from 0.2 µM to 4 µM), Table 1, likely associated with differences in their polarity
(clogP values, Table 1) affecting cellular permeability, as assessed in the Cacoꢀ2 membrane
permeability studies (Supplementary Table 4). Likewise, compounds 1ꢀ10 exhibit different
level of selectivity towards MLLꢀAF9 cells over HMꢀ2 control cell line (reported as selectivity
index, SI, ranging from 5 ꢀ 32), Table 1. Furthermore, the microsomal stability of these
compounds in mouse liver microsomes also varied substantially, Table 1. For example,
incorporation of the hydroxyethyl (2) or 1,2ꢀpropanediol (3-(RS), 3-(R), 3-(S)) substituents at R1
substantially increases polarity (clogP < 3.5) and metabolic stability (T1/2 = 18ꢀ60 min), however
the effect of these compounds on proliferation of the MLLꢀAF9 cells (GI > 1 µM) is weaker
5
0
than for 1, Table 1. Introduction of the methyl (4-(R)) or isopropyl (6) groups at the secondary
hydroxyl substantially improves GI values (e.g. GI = 0.18 µM for 6), however these
5
0
50
compounds are metabolically unstable (T1/2 < 3 min). Interestingly, 4-(R) has markedly
improved cellular permeability over 3-(RS), Supplementary Table 4, despite very small
difference in their structures, Table 1. The efflux ratio was >20ꢀfold lower for 4-(R) as
8
ACS Paragon Plus Environment

Page 9 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
compared to 3-(RS) (Supplementary Table 4), with both apical to basolateral (A→B) and
basolateral to apical (B→A) permeability markedly improved for 4-(R). Accordingly, cellular
activity of 4-(R) is almost 10ꢀfold increased over 3-(RS) despite similar IC values for these
5
0
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
compounds, Table1, emphasizing that increased cellular permeability correlates with cell growth
inhibition. The analogue with the acetamide group at R1, compound 7, shows cellular activity
(
GI = 0.78 µM in MLLꢀAF9 cells) similar to 1, however this compound has substantially
50
improved selectivity (SI = 34) and metabolic stability over 1, Table 1, both are likely associated
with its higher polarity. Compound 7 also showed low cellular permeability (efflux ratio > 90,
Supplementary Table 4) reflected by the modest cellular activity of this compound, Table 1.
Finally, similarly to the compound with 4ꢀmethylpyrazole at R1 (10), the analogue with 4ꢀ
methyltriazole (9) has improved cellular activity (GI = 0.33 µM), selectivity (SI = 30),
5
0
microsomal stability as well as polarity when compared to 1, Table 1. Therefore, substitutions
on the indole nitrogen are very valuable to modulate both cellular activity and drugꢀlike
properties for this class of meninꢀMLL inhibitors. Overall, we concluded that hydroxyethyl
(present in 2), acetamide (present in 7), 4ꢀmethylpyrazole (present in 10) and 4ꢀmethyltriazole
(present in 9) represent the most valuable substituents at R1 due to either superior effect in MLL
leukemia cells (9 and 10) or improved drugꢀlike properties (e.g. polarity for 2 and 7).
Activity and property optimization of analogues substituted at positions 3, 4 or 6 of indole
Analysis of the crystal structure of the meninꢀcompound 1 complex revealed that carbon
at position 3 of the indole ring (position R2, Figure 1b) approaches the backbone of menin,
which limits the possibility of substituting at this position. Nevertheless, since this position has
not been explored before, we introduced a few small substituents (R2 substituents) at this
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
position, namely fluorine (compound 11), amino (compound 12) and methyl (compound 13),
Table 2. The inhibitory activity of these compounds was, however, 3ꢀ4 fold reduced as
compared to 1, Table2, consistent with their relatively weak cellular activity in the MLLꢀAF9
cells. These results support that position 3 on the indole ring of 1 should remain unsubstituted.
We have recently explored several substituents at position 4 of the indole ring (R3
substituent, Table 2) and concluded that small hydrophobic groups represent the best
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
6
substituents at this position, with the methyl group resulting in ~2ꢀfold improved activity
(compound 14, MIꢀ463, IC = 15 nM, Table 2), likely due to hydrophobic contacts with the
50
aromatic ring of Tyr323, Figure 1a. Here, we completed the series of R3 substituted analogues
by incorporating chorine (compound 15, Table 2) at this position to assess whether it might
represent an even more favorable substituent than a methyl group. The in vitro inhibitory activity
of 15 (IC = 22 nM) was slightly improved over 1, but the effect on proliferation of the MLL
5
0
leukemia cells was much weaker (GI > 1 µM), Table 2. Therefore, we concluded that methyl
50
group represents the best substituent at position 4 of the indole ring.
Finally, we also probed position 6 on the indole ring (R4 substituents, Table 2), which
was not explored previously for substitutions. The crystal structure of the meninꢀcompound 1
complex revealed that this position is partly solvent exposed and that relatively small polar
groups could likely enhance interꢀ and intramolecular interactions (Figure 1a, 1b). Indeed,
incorporation of the hydroxyl in compound 17 or methoxy group in compound 18 (MIꢀ336) at
R4 resulted in about 1.5ꢀfold improved inhibitory activity over 1, Table 2. More importantly,
these two substituents have led to a significant improvement of growth inhibition in the MLL
leukemia cells, in particular compound 17, which showed almost 3ꢀfold better GI value (GI =
5
0
50
0.18 µM) than 1, Table 2. Further increase in the size of R4 (e.g. compound 19 with the ethoxy
1
0
ACS Paragon Plus Environment

Page 11 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
group) or introduction of fluorine (compound 16) resulted in the reduction of inhibitory activity,
Table 2.
To understand the molecular basis of the activity of these compounds we determined the
crystal structure of 18 with menin, Supplementary Table 3. The structure revealed an
intramolecular hydrogen bond between the oxygen of the methoxy group and nitrogen in the
piperidine ring, which likely stabilizes the conformation of 18 required for effective binding,
Figure 3a. The same interaction is expected for compound 17 harboring a hydroxyl group at R4.
In summary, based on the in vitro and cellular activity, we concluded that hydroxyl and methoxy
groups represent optimal substituents at R4, and we selected both of them for preparation of
multiꢀsubstituted derivatives of 1 (see below). The metabolic stability and polarity (particularly
for 18) still need to be improved for compounds with R4 substituents (Table 2), but these
liabilities may be compensated by the presence of other substituents on the indole ring (see
below).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Activity of analogues with multiple substitutions on indole ring
Having established the optimal substituents at different positions of the indole ring, we
then assessed whether simultaneous incorporation of these substituents into the scaffold would
provide further improvement of activity and drugꢀlike properties. For this purpose, we prepared a
number of compounds to probe combinations of the best R1 substituents (hydroxyethyl,
acetamide, 4ꢀmethylpyrazole and 4ꢀmethyltriaozle) with the methyl group as an optimal
substituent at R3 and/or two best R4 substituents (hydroxyl or methoxy groups), Table 3. First,
we prepared the R3, R4 diꢀsubstituted analogues (compounds 20 and 21), Table 3. Although
neither compound showed improved in vitro inhibitory activity over 1, however, they
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
demonstrate improved cellular activity in MLLꢀAF9 cells, Table 3, with compound 21 showing
over 5ꢀfold improved cell growth inhibition (GI = 100 nM) as compared to 1, Table3.
50
Interestingly, both 20 and 21 showed relatively high to moderate cellular permeability in Cacoꢀ2
cells, Supplementary Table 4, which could contribute to strong growth inhibition in the MLL
leukemia cells, Table 3.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The subsequent set of analogues (compounds 22, 23, 24) contained a hydroxyethyl
substituent at R1, a methyl at R3 with R4 remaining either unsubstituted or substituted with
hydroxyl or methoxy groups, Table 3. None of these compounds, however, showed substantially
increased activity, either in vitro or in MLL leukemia cells, Table 3. Similarly, two analogues
with an acetamide group at the indole nitrogen (compounds 25 and 26) did not demonstrate
significantly improved in vitro or cellular activity over 1, Table 3. We next synthesized and
tested compounds with aromatic substituents at the indole nitrogen, either 4ꢀmethylpyrazole
(compounds 27ꢀ31) or 4ꢀmethyltriazole (compounds 32ꢀ34), Table 3. Notably, all these
compounds showed strong inhibitory activity, with the IC values for inhibition of the meninꢀ
5
0
MLL interaction in the range of 12 ꢀ 24 nM, Table 3, representing an improvement over 1. We
solved the crystal structure of menin in complex with the most potent 31 (MIꢀ574), which
harbors three substituents on the scaffold (Figure 3b). The structure shows that incorporation of
several substituents on the indole ring does not affect the binding mode of 31 to menin as
compared to 1. The structure also reveals two additional hydrogen bonds, one intramolecular
formed by the methoxy group of 31 and an additional intermolecular between the pyrazole ring
and Glu366, Figure 3b, which are likely responsible for ~2.6ꢀfold improved activity of this
compound over 1.
1
2
ACS Paragon Plus Environment

Page 13 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Importantly, simultaneous substitution at multiple sites of indole resulted in compounds
with a pronounced effect on the proliferation of MLL leukemia cells, with the most potent
compound 27 (MIꢀ538), which showed GI = 83 nM in MLLꢀAF9 cells, representing almost 7ꢀ
5
0
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
fold improvement over 1, Table 3. Compound 27 binds to menin with low nanomolar affinity
(K = 6.5 nM), as assessed by Isothermal Titration Calorimetry (Figure 4a), representing ~3.6ꢀ
d
1
6
fold improvement in the binding affinity as compared to 1 (K = 23.6 nM). Therefore, potent
d
cellular activity of 27 originates from the improved binding affinity to menin and possibly
increased cell membrane permeability (e.g. improved permeability over 7 with a similar IC50
value), Supplementary Table 4. Interestingly, 27, which is a double substituted analogue with
methylpyrazole at R1 and hydroxyl at R4, has improved cellular activity by 2ꢀ3 fold over
monosubstituted compounds 10 or 17, which harbor only one of these substituents, Tables 1ꢀ3.
Further addition of the methyl group at R3 resulting in compound 30 leads to reduced activity in
MLL leukemia cells by more than 2ꢀfold, Table 3. These results demonstrate that simultaneous
incorporation of multiple substituents to the scaffold can result in improved cellular activity over
monosubstituted analogues, however these substitutions are not fully additive as cellular activity
of triple substituted analogues is not improved over diꢀsubstituted analogues, Table 3.
Multi-substituted analogues demonstrate favorable drug-like properties
In addition to activity, other properties of the analogues of compound 1 were affected by
simultaneous introduction of multiple substituents. For example, substantial improvement in
selectivity towards MLL leukemia cells, reflected by increase in the SI index, was observed for
compounds with multiple substituents on the indole ring, Table 3. While 1 and the majority of
monoꢀsubstituted analogues have SI < 30, Tables 1 and 2, additional substitutions on the indole
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
led to increased selectivity, resulting in a number of compounds with SI > 50, Table 3. For
instance, compounds with no substituent at R1 but substituted at R3 and R4 (20 and 21),
analogues with acetamide at R1 (25 and 26) as well as 27 with 4ꢀmethylpyrazole at R1 and
hydroxyl at R4 showed best selectivity, Table 3. Increase in the selectivity index can be
attributed to both, the increase in cellular activity towards MLL leukemia cells as well as a
reduced effect in the control cell line HMꢀ2 (e.g. for 25 and 26, Table 3). Although cellular
selectivity was rather difficult to predict upon a particular modification in the scaffold, in general
the presence of an additional polar group improves selectivity, Table 3. Exceptions are
compounds 33 and 34 due to their weaker effect in MLLꢀAF9 cells over 32.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Another important property we attempted to optimize was the polarity of these
compounds, as the clogP value for the parent compound 1 is about 4.3, implying that it might
need further improvement. Indeed, the majority of compounds with multiple substitutions on
indole had lower clogP values than 1, Table 3, including the compound with the most potent
activity in MLL leukemia cells, 27 (clogP = 3.7), which harbors the hydroxyl group at R4. This
demonstrates that it is possible to simultaneously improve polarity, cellular activity and
selectivity for this class of meninꢀMLL inhibitors.
Finally, one liability of the lead compound 1 is its relatively poor metabolic stability in
the murine liver microsomes (T1/2 = 6.8 min), Table 1, reflected by unsatisfactory PK profile
(see below). By introducing multiple substituents to the indole ring, we were able to substantially
improve metabolic stability of analogues of compound 1, with a number of compounds showing
increase in T1/2 to over 30 min, Table 3. This includes compounds 20, 21, 31, which also
demonstrated strong effect on cell proliferation (GI of ~200 nM or better in MLLꢀAF9 cells)
5
0
and good selectivity (SI > 50). The analogue with the most potent cellular activity, compound 27,
1
4
ACS Paragon Plus Environment

Page 15 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
has about two fold increase in microsomal stability (T1/2 = 14 min) over 1, Table 3.
Nevertheless, due to its strongest activity in MLL leukemia cells, high selectivity to MLL
leukemia cells (SI > 100) and improved polarity (clogP = 3.7), 27 was chosen for
pharmacokinetic studies in mice together with several other compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pharmacokinetic studies of menin-MLL inhibitors in mice
We selected 27 and several other compounds (20, 21, 31), which showed the best cellular
activity in MLLꢀAF9 cells (GI < 0.2 µM), high level of selectivity to MLL leukemia cells (SI >
5
0
50) and good microsomal stability (T1/2 > 30 min) and performed in vivo pharmacokinetic (PK)
studies in mice. The PK studies were also performed for the parent compound 1. Table 4
summarizes the PK parameters for all compounds tested using both intravenous (iv, 15 mg/kg
dose) and oral (po, 30 mg/kg dose) administration. Importantly, these studies demonstrate
substantially improved PK profile for 27 over 1 and other compounds tested, Table 4.
Compound 27 demonstrated markedly improved exposure (area under the curve, AUC, values),
Cmax (maximum compound concentration) in the blood plasma as well as the lowest value of
clearance (Cl ), Table 4. The halfꢀlife (T ) for 27 was reduced as compared to 1, but still
obs
1/2
remained relatively long (~1.6 h). To investigate the possible reason for reduced halfꢀlife for 27,
we investigated phase II metabolism and found that this compound undergoes glucuronidation,
consistent with the presence of the hydroxyl group in the structure (Supplementary Figure 2).
Importantly, 27 has also high oral bioavailability (~50%), similarly to 1 and 21, Table 4. Out of
the other compounds tested in PK studies, 31 with the triꢀsubstituted 2ꢀcyanoindole ring
demonstrated good exposure, but this compound had very poor oral bioavailability, Table 4.
Two other compounds, 20 and 21, with R3, R4 diꢀsubstitutions on the 2ꢀcyanoindole ring did not
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
show substantial improvement in the PK profile over parent compound 1, and 20 had even
reduced exposure and increased clearance, Table 4. These results demonstrate that not all
substitutions on the 2ꢀcyanoindol ring of 1 result in additive effects on drugꢀlike properties,
including PK profile, and multiple combinations of substituents need to be tested to identify
compounds with the desired activity and properties. These studies also showed a lack of
correlation between in vitro microsomal stability and in vivo PK profile, as exemplified for
compound 20 when compared to 1 and for 27 with superior PK but relatively short T1/2 in mouse
liver microsomes, Tables 3 and 4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Compound 27 demonstrates on target mechanism of action in MLL leukemia cells
Based on strong effect on proliferation of the MLLꢀAF9 leukemia cells (GI = 83 nM),
5
0
excellent selectivity and best PK profile (Table 4), 27 was selected for further characterization in
MLL leukemia models. First, we assessed the effect of this compound in human MLL leukemia
cell lines, MV4;11 and MOLM13, harboring two different MLL translocations (MLLꢀAF4 and
MLLꢀAF9, respectively), and observed a pronounced inhibition of cell growth (GI = 170 ꢀ 270
5
0
nM), Figure 4b. Importantly, 27 showed no effect (up to 6 µM) on growth of the control cell
lines HLꢀ60 and HMꢀ2, which do not harbor MLL translocations, demonstrating good selectivity
towards MLL fusion protein transformed cells (Figure 4b). Inhibition of the meninꢀMLL
interaction is expected to result in the differentiation of MLL leukemia cells and downregulation
1,14
of MLL fusion target genes. Indeed, we observed pronounced differentiation of MLLꢀAF9
cells upon treatment with 27 as reflected by cell morphology change (decreased nuclei to
cytoplasm ratio, multiꢀlobed nuclei and highly vacuolated cytoplasm) we well as marked
increase in the expression level of Itgam, a differentiation marker of myeloid cells (Figure 4c,
1
6
ACS Paragon Plus Environment

Page 17 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
d). Finally, treatment with 27 resulted in strong downregulation of expression of Hoxa9 and
Meis1 genes, the downstream targets of MLL fusion proteins required for their leukemogenicity,
Figure 4e. About 100 nM of 27 was sufficient to reduce by ~50% Hoxa9 expression in MLLꢀ
AF9 cells and even more pronounced effect was seen on Meis1 expression, consistent with the
strong effect of this compound on proliferation of these cells, Figure 4b. All these findings
further confirm onꢀtarget effects and specific mechanism of action of 27 meninꢀMLL inhibitor,
suggesting that it represents a valid candidate for in vivo studies in a mouse model of MLL
leukemia.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 27 demonstrates in vivo efficacy in a xenograft model of MLL leukemia
Encouraged by the favorable PK profile as well as the pronounced and specific effect of 27 in
MLL leukemia cells, we selected this compound for in vivo efficacy studies in a mouse xenograft
model utilizing MV4;11 human MLL leukemia cells injected subcutaneously into CB17 SCID
mice. Treatment with 27 (45 mg/kg, once daily via ip) was initiated when tumor size reached
3
~
100 mm and was continued for two weeks, Figure 5. Importantly, treatment with 27 resulted
in a pronounced, about 80% reduction in the MV4;11 tumor volume (Figure 5), without causing
substantial signs of toxicity reflected by less than 10% reduction of the body weight (not shown).
These results validate that 27 is not only a potent inhibitor of MLL leukemia cell proliferation
but can also effectively inhibit the growth of MV4;11 tumors in mice. These data demonstrate a
strong potential for this class of meninꢀMLL inhibitors for further development towards
therapeutically useful compounds.
Chemistry
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
All inhibitors 1ꢀ34 contain the Nꢀ(piperidinꢀ4ꢀyl)ꢀ6ꢀ(2,2,2ꢀtrifluoroethyl)thieno[2,3ꢀd]pyrimidinꢀ
4ꢀamine moiety (39) that was synthesized in five steps starting from the 4ꢀ
2
9
trifluorobuteraldehyde, Scheme 1, with the key step of Gewald thiophene synthesis.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Condensation of thiophene with triethyl orthoformate gave pyrimidinone 37, which was then
converted to 38. Nucleophilic substitution of 38 with Bocꢀprotected aminopiperidine and
subsequent deprotection provided 39 as hydrochloride salt. The process was optimized to
provide high yields of 39 with no chromatography purification, resulting in overall yield of 45%
over 5 steps after recrystallization of the final product.
Synthesis of 1 and 11 started from commercially available 5ꢀmethylindoleꢀ2ꢀcarboxylic
acid, Scheme 2. The 3ꢀsubstituted indole was obtained by reaction of 40a with 1ꢀchloromethylꢀ
4ꢀfluoroꢀ1,4ꢀdiazoniabicyclo[2.2.2]octane bis(tetrafluoroborate) as a source of electrophilic
fluorine, providing 40b. The carboxylic acid functional group in 40a and 40b was then converted
to nitrile by dehydratation of amide obtained from acid chloride. The Boc protection of indole
nitrogen smoothly gave 42a,b that were subjected to freeꢀradical bromination. Bromoꢀderivatives
were used for coupling reaction with 39 followed by Boc group deprotection resulting in 1 and
11, Scheme 2.
For preparation of 7 and 8, we explored the possibility of introducing substituents on
indole nitrogen using compound 1 as a substrate, Scheme 2. Indeed, indole 1 was deprotonated
using Cs CO as a weak base and directly alkylated with bromoacetamide or 2ꢀ
2
3
bromopropionamide to form 7 and 8, respectively, although in moderate to poor yields.
Interestingly, similar conditions with other alkylating reagents did not yield the desired products.
Therefore we applied other methodology to afford compounds 2ꢀ6, 9, 10, 23ꢀ34 as described
below.
1
8
ACS Paragon Plus Environment

Page 19 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Synthesis of 2 was performed according to Scheme 3. Alkylation of indole 41a
proceeded efficiently with NaH as a base producing the hydroxyethyloindole 44, which was then
subjected to freeꢀradical bromination, followed by alkylation and deprotection reactions to afford
final product 2. Similar approach was employed for synthesis of 1,2ꢀpropanediol derivatives of 3,
where starting materials were either racemic mixture or enantiomerically pure alcohols 45,
Scheme 4. After protection of hydroxyl groups in diol 45 with tertꢀbutyldimethylsilyl ethers
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(TBS) the resulting intermediates reacted with indole 41a to obtain alkylated compound 46.
Further bromination, alkylation and deprotection reactions provided racemic compound 3 as well
as its enantiomers, Scheme 4. Synthesis of individual enantiomers of 4 required selective
methylation of secondary hydroxy group, which was accomplished by removal of both TBS
protecting groups from 46 followed by exclusive protection of primary hydroxyl group with a
bulky tertꢀbutyldiphenylsilyl ether (TBDPS) and methylation of a secondary alcohol to afford
47. Subsequently, the bromination, alkylation and deprotection reactions were performed to yield
enantiomerically pure 4ꢀ(S) and 4ꢀ(R).
To synthesize compound 12, the carboxylic acid 48 was used as a starting material,
Scheme 5. The carboxylic group of 48 was converted to nitrile via standard amidation and
dehydration reactions, followed by reduction of the nitro group with tin chloride (II) to afford 49
in four steps. By heating 49 with paraformaldehyde and KCN in the presence of zinc chloride
(II), the Nꢀcyanomethyl compound 50 was obtained, which was subjected to Boc protection with
four equivalents of Boc O and simultaneous cyclization to afford 51. Freeꢀradical bromination
2
followed by nucleophilic substitution with building block 39 and deprotection resulted in
compound 12, Scheme 5.
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Inhibitors bearing methyl groups at positions 3 or 4 of indole required different synthetic
procedure, in which reductive amination was employed to couple aldehydes 60aꢀd with 39,
Scheme 6. Synthesis of 13 started from reaction of 2ꢀoxobutyric acid and hydrazine 52 to
produce 53, which after hydrolysis was converted to nitrile via amide to obtain indole derivative
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
0
59a. To synthesize aldehydes 60bꢀd we used Hemetsberger indole synthesis. First, acids 56a,b
were converted to azidoacrylates 57a,b. To obtain 57c, direct iodination of phenol 54 at para
position was performed after protection of the hydroxyl group followed by conversion of the
intermediate to aldehyde with simultaneous deprotection of phenol, which was then alkylated to
yield 55 and further converted to afford 57c. Compounds 58aꢀc were obtained by refluxing of
azidoacrylates 57aꢀc, resulting in products that precipitated out upon cooling. Subsequently, the
2ꢀcyanoindole compounds 59aꢀd were obtained after conversion of esters 53 or 58a-c to nitriles,
Scheme 6. Aldehydes 60a,c,d were synthesized by adapting the procedure described by Moyer
3
1
et. al., while aldehyde 60b was prepared by freeꢀradical dibromination and hydrolysis of haloꢀ
derivative in the presence of silver carbonate. Reductive amination with 39 went efficiently to
produce 14, 15, 21 with good yields except for 13 (28% yield). Subsequently compound 20 was
obtained from 21 by demethylation, while 25 and 26 were produced by direct alkylation of 14
and 21, respectively.
Compounds 16ꢀ19 were synthesized using the procedure shown in Scheme 7. First, the
azidoacrylates 62a,b were synthesized from commercially available aldehydes 61a,b. Then, the
32
rhodium trfluoroacetate catalyzed reaction was adapted as a cyclization method, resulting in
63a,b, which were subsequently converted to nitriles and protected with Boc group. Synthesis of
compounds 16 and 18 was performed by bromination followed by nucleophilic substitution with
2
0
ACS Paragon Plus Environment

Page 21 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3
9 and removal of Boc protecting group. Compound 17 was obtained after demethylation of 18
with boron tribromide that was further alkylated with ethyl iodide to get final product 19.
The synthesis of compounds 23, 24, 27ꢀ34, with multiple substituents on indole ring is
depicted in Scheme 8. To prepare alcohol 24, the hydroxyethyl substituent was introduced by
alkylation of 60c with trityl protected bromoethanol to afford 69. Pyrazole derivatives 27ꢀ31
were synthesized from commercially available ester 65, which was first protected with trityl
group, then reduced to alcohol 66 and transformed to bromide 67 which was used immediately
for alkylation of aldehydes 60a, 60c or 68b, resulting in 70, 71 and 72. The triazol moiety was
introduced through alkylation of 60c via click chemistry to obtain 73. Reductive amination and
subsequent deprotection afforded the desired products 24, 28, 29, 31 and 34, Scheme 8. Finally
demethylation of 24, 28, 31 and 34 resulted in 23, 27, 30 and 33, respectively.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Conclusions
Development of very potent small molecule inhibitors of proteinꢀprotein interactions with
optimized drugꢀlike properties, including cellular permeability and pharmacokinetic profile, is
considered a challenging task due to the increased molecular weight of such compounds often
1
8,20
required to achieve high potency.
Here, we report optimization of a thienopyrimidine class of
compounds, which block the proteinꢀprotein interaction between menin and MLL fusion proteins
1
as a potential therapeutic strategy for MLL leukemia. The difficulty in developing meninꢀMLL
3
12
inhibitors partly arises from the large size of the MLL binding site on menin (over 5,000 Å ),
implying that compounds with a relatively large molecular weight might be required to
effectively block this PPI, complicating the optimization of their drugꢀlike properties.
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In this study, we focused on structureꢀbased optimization of our new class of meninꢀMLL
1
6
inhibitors, represented by the lead compound 1, with the goal to identify compounds with
improved activity and drugꢀlike properties and to select compound(s) for characterization in
biological experiments. Based on the crystal structure of the meninꢀcompound 1 complex, we
designed, synthesized and extensively evaluated series of novel analogues, by systematic
exploration of substituents on the cyanoindole ring of 1. The most potent monoꢀ and multiꢀ
substituted analogues demonstrated up to a few fold increase in in vitro activity over the lead
compound 1. Furthermore, we discovered that incorporation of substituents at least at two
positions of the 2ꢀcyanoindole ring was required to substantially increase both cellular activity
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
and metabolic stability (e.g. compounds 20, 21, 27, 29, 32, Table 3), with ~7ꢀfold increased GI50
value for 27. Interestingly, favorable substituents introduced simultaneously at three positions of
the cyanoindole ring did not have an additive effect on cellular activity, and in fact we observed
a reduced effect on proliferation of the MLL leukemia cell for such compounds (e.g. triꢀ
substituted compound 30 has weaker cellular activity than diꢀsubstituted compounds 20 or 27)
and on their metabolic stability (e.g. compound 30 as compared to 20). Furthermore, we found
that poor cellular permeability (e.g. high efflux ratio) for some of the compounds reported here
(
e.g. for 7, 3-(RS) and 25) likely leads to the discrepancy in their in vitro and cellular activity.
Other effects, including possible variations in the binding kinetics, stability in cells and solubility
of compounds as well as the need to compete with the full length MLL or MLL fusion proteins
in cells might also contribute to the observed differences between in vitro and cellular activity.
Overall, these results highlight difficulties in lead optimization process, in particular for
inhibitors targeting proteinꢀprotein interactions.
2
2
ACS Paragon Plus Environment

Page 23 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Our efforts led to the discovery of several meninꢀMLL inhibitors with greatly improved
potency and drugꢀlike properties, out of which 27 displayed: (1) strong in vitro inhibition of the
meninꢀMLL interaction, IC = 21 nM, and the most potent binding affinity to menin (K = 6.5
5
0
d
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
nM) obtained for small molecule meninꢀMLL inhibitor reported to date; (2) high potency in
inhibiting proliferation of MLLꢀAF9 transformed cells (GI = 83 nM), demonstrating the best
5
0
cellular activity observed for any meninꢀMLL inhibitor reported to date; (3) specific mechanism
of action in MLL leukemia cells; (4) superior, over 100ꢀfold, selectivity to the MLL leukemia
cells over control cell line; (5) improved polarity as compared to parent compound 1 (clogP = 3.7
for 27); (6) greatly improved PK profile, including high oral bioavailability; (7) pronounced
efficacy in vivo in mice model of MLL leukemia. Overall, the development of 27 represents a
great achievement for this new class of meninꢀMLL inhibitors. In fact this compound
16
demonstrates a significant improvement over our recently reported MI-463 and MI-503 in
terms of cellular activity in MLLꢀAF9 cells, selectivity to MLL leukemia cells and polarity
(clogP value), Tables 2 and 3, making it an excellent candidate for development of a
therapeutically useful compound. Furthermore, 27 has a favorable ligand efficiency index (LE =
3
3
0.28), which is above the average value of 0.24 reported for PPI inhibitors. Compound 27
discovered in this study constitutes a valuable addition to our meninꢀMLL inhibitor toolbox to
investigate the role of meninꢀMLL interaction in physiological and disease state and for potential
therapeutic applications. Overall, this study demonstrates that property optimization of inhibitors
targeting proteinꢀprotein interactions represents a complex and multidimensional problem, with
an outcome difficult to predict when a particular set of substituents is explored. The difficulties
in optimizing PPI inhibitors are at least partly associated with the relatively large molecular
weight of these compounds (e.g. most of the compounds reported here have M > 500 Da) and
w
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
25
rather high hydrophobicity often required for potent binding to the PPI interfaces. Extensive
medicinal chemistry efforts strongly supported by structureꢀbased design are required to
successfully conduct such studies to identify potent PPI inhibitors with optimized
physicochemical properties and for advancing such compound to future potential therapeutics.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Experimental Section
Expression and purification of menin
9
,12
The expression and purification of menin have been described previously.
Biochemical characterization of menin-MLL inhibitors
Inhibition of the meninꢀMLL interaction by small molecules was assessed by fluorescence
9
,14
polarization (FP) assay using the protocol described previously. Briefly, the fluoresceinꢀ
labeled MLL (MBM1) peptide at 10 nM, menin at 100 nM and varying concentrations of
compounds were used for IC determination in the FP buffer (50 mM Tris, pH 7.5, 50 mM
5
0
NaCl, 1 mM DTT). Compounds (5% final DMSO concentration) were added to the meninꢀMLL
peptide complex and incubated for 1 h before changes in fluorescence polarization were
measured using the PHERAstar microplate reader (BMG).
Isothermal Titration Calorimetry
Menin was extensively dialyzed at 4 °C against ITC buffer (50 mM phosphate, pH 7.5, 50 mM
NaCl, 1mM βꢀmercaptoethanol) and degassed prior to measurement. Compound 27 was
dissolved in DMSO and diluted with the ITC buffer to final concentration (60 ꢀM, 5% DMSO).
Protein solution (6 ꢀM) was adjusted to contain 5% DMSO final concentration. The titrations
were performed using a VPꢀITC titration calorimetric system (MicroCal) at 25 °C. The
2
4
ACS Paragon Plus Environment

Page 25 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
calorimetric cell, containing menin was titrated with 27 injected in 10 µl aliquots. Data was
analyzed using Origin 7.0 (OriginLab) to obtain K and stoichiometry.
d
Crystallization of menin complexes with small molecule inhibitors
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For coꢀcrystallization experiments 2.5 mg/mL menin was incubated with small molecule
inhibitors at 1:3 molar ratio. Crystals were obtained using the sitting drop technique at 10 ºC by
1
2
applying the procedure described previously. Prior to data collection, crystals were transferred
into a cryoꢀsolution containing 20% PEG550 MME and flashꢀfrozen in liquid nitrogen.
Crystallographic data collection and structure determination
Diffraction data for menin and menin complexes were collected at the 21ꢀIDꢀD and 21ꢀIDꢀF
beamlines at the Life Sciences Collaborative Access Team at the Advanced Photon Source. Data
3
4
were processed with HKLꢀ2000. Structures of the complexes were determined by molecular
replacement using MOLREP with the apoꢀstructure of human menin (PDB code: 4GPQ) as a
3
5
36
search model in molecular replacement. The model was refined using REFMAC, COOT and
3
7
38
the CCP4 package. Validation of the structures was performed using MOLPROBITY and
39
ADIT. Details of data processing and refinement are summarized in Supplementary Table 3.
Coordinates and structure factors for meninꢀinhibitor complexes have been deposited in the
Protein Data Bank under the following codes: 5DB0 (meninꢀcompound 3-(RS)), 5DB1 (meninꢀ
compound 18), 5DB2 (meninꢀcompound 7) and 5DB3 (meninꢀcompound 31).
Compound profiling
Profiling of compound 1 was outsourced to the Contract Research Organizations (CROs). Kinase
profiling was performed by Nanosyn (http://www.nanosyn.com/productsandservices/screeningꢀ
andꢀprofiling/) and GPCR profiling was performed by Millipore
(http://www.emdmillipore.com/).
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Caco-2 studies
The Cacoꢀ2 membrane permeability studies were performed at Cyprotex (www.cyprotex.com).
Viability assays
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The MLLꢀAF9 transformed murine bone marrow cells (BMCs) were prepared as described
4
0
previously. MV4;11, MOLMꢀ13 and HLꢀ60 cells were cultured in RPMIꢀ1640 medium with
10% FBS, 1% penicillin/streptomycin and NEAA (Nonꢀessential amino acid) solution. For
5
5
5
viability assay, MOLMꢀ13 (1x10 /mL), MV4;11 (1x10 /mL), HLꢀ60 (2x10 /mL) human
4
leukemia cells as well as MLLꢀAF9 and HMꢀ2 murine bone marrow cells (2.5x10 /mL) were
plated (1 mL/well), treated with compounds or 0.25% DMSO and cultured at 37 ᵒC for 7 days.
Media were changed at day four with viable cell number restored to the original concentration
and compounds were reꢀsupplied. 100 µL of cell suspension were transferred to 96 well plates
for each sample in quadruplicates. A Vybrant MTT cell proliferation assay kit (Molecular
Probes) was employed. Plates were read for absorbance at 570 nm using a PHERAstar BMG
microplate reader. The experiments were performed 2ꢀ3 times in quadruplicated with calculation
of mean and standard deviation for each condition.
Real-Time PCR
Effect of meninꢀMLL inhibitors on expression level of Hoxa9 and Meis1 was assessed by Realꢀ
1
2
time quantitative PCR (qRTꢀPCR) using the protocol described previously.
Cytospin/Wrigth-Giemsa staining
Mouse bone marrow cells transduced with MLLꢀAF9 were plated in 12ꢀwell plates (1 mL/well)
5
at an initial concentration of 3x10 /mL cells, treated with compounds or 0.25% DMSO and
1
4
incubated at 37 °C in a 5% CO incubator. Cytospins were prepared as described previously at
2
designated time points.
2
6
ACS Paragon Plus Environment

Page 27 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Microsomal stability studies
The metabolic stability was assessed using CDꢀ1 mouse liver microsomes. 1 µM compounds
were incubated with 1.3 mg/mL microsomes and 1.7 mM cofactor NADPH in 0.1 M phosphate
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
buffer (pH =7.4) containing 3.3 mM MgCl at 37 °C. The DMSO concentration was less than
2
0.1% in the final incubation system. At 0, 3, 5, 10, 15, 30, 45 and 60 min of incubation, the
reactions were stopped by adding 3ꢀfold excess of acetonitrile containing 100 ng/mL of internal
standard for quantification. The collected fractions were centrifuged at 3000 g for 20 min to
collect the supernatant for LCꢀMS/MS analysis, from which the amount of compound remaining
was determined. The natural log of the amount of compound remaining was plotted against time
to determine the disappearance rate and the halfꢀlife of tested compounds.
Pharmacokinetic studies in mice
All animal experiments in this study were approved by the University of Michigan Committee on
Use and Care of Animals and Unit for Laboratory Animal Medicine (ULAM).
The pharmacokinetics of meninꢀMLL inhibitors was determined in female C57BL/6 mice
following intravenous (iv) dosing at 15 mg/kg and oral dosing (po) at 30 mg/kg. Compounds
were dissolved in the vehicle containing 25% (v/v) DMSO, 25% (v/v) PEGꢀ400 and 50% (v/v)
PBS. Serial blood samples (50 µL) were collected over 24 h, centrifuged at 15,000 rpm for 10
min and saved for analysis. Plasma concentrations of the compounds were determined by the
LCꢀMS/MS method developed and validated for this study. The LCꢀMS/MS method consisted of
an Agilent 1200 HPLC system and chromatographic separation of tested compound was
achieved using an Agilent Zorbax ExtendꢀC18 column (5 cm x 2.1 mm, 3.5 µm; Waters). An AB
Sciex QTrap 3200 mass spectrometer equipped with an electrospray ionization source (ABIꢀ
Sciex, Toronto, Canada) in the positiveꢀion multiple reaction monitoring (MRM) mode was used
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 81
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
for detection. All pharmacokinetic parameters were calculated by noncompartmental methods
using WinNonlin® version 3.2 (Pharsight Corporation, Mountain View, CA, USA). Similar
procedure was applied for detection of glucuronidation of compound 27 in blood plasma after iv
administration of the compound (15 mg/kg).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In vivo efficacy studies
Freshly thawed MV4;11 cells (ATCC) were cultured in RPMIꢀ1640 medium with 10% FBS and
Penicillin/Streptomycin. On the day of injection, cells were resuspended in serum free medium at
8
6
1x10 cells/mL and mixed 1:1 with Matrigel. 5x10 cells (100 µL) were injected into the right
flank of 4ꢀ6 week old female BALB/c nude mice. Treatment started when tumor size has reached
3
~
100 mm . Mice were randomly grouped with each group containing 8 mice. Vehicle or
compounds were administrated once daily at designated doses using ip injections for two weeks.
Body weight and tumor sizes were monitored three times a week. The tumor size was calculated
2
as: tumor volume = a x b /2 (a, long diameter; b, short diameter). All animal studies were
performed under the guidelines of the University of Michigan Committee for Use and Care of
Animals.
General Chemistry
All solvents and reagents were used as obtained from commercial sources unless otherwise
1
indicated. The H NMRs were taken on Bruker Avance III 600 MHz or Varian MR400.
Chemical shifts are reported in ppm relative to tetramethylsilane or residual solvent signal. The
mass measurements were determined on a Micromass LCT timeꢀofꢀflight mass spectrometer
using positive mode and electrospray ionization. The exact mass measurements were determined
on Agilent QꢀTOF timeꢀofꢀflight mass spectrometer using positive ion mode and electrospray
ionization. The melting points were determined on PerkinꢀElmer DSCꢀ7 differential scanning
calorimeter and are not corrected. Analytical TLC was performed on Merck TLC aluminum
plates precoated with F254 silica gel 60 (UV, 254 nm, and iodine). Infrared (IR) spectra were
2
8
ACS Paragon Plus Environment

Page 29 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
recorded on PerkinꢀElmer FTꢀIR Spectrum BX on neat powder. The purity analysis of final
compounds was determined on Shimadzu Prominence HPLC system (20 series: binary pump,
UV/Vis at 254 nm, heated column compartment 30 °C), using Restek Ultra C18 (5 µm) 150 x
4.6 mm column. The solvents were programmed to run at gradient starting 20% CH CN in water
3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
to 80% during 8 min run. If not indicated, the purity of all final compounds was >95% as
determined by HPLC via integration of UV spectra at 254 nm. All target compounds 1ꢀ34 were
tested as hydrochloride salts unless otherwise stated. The standard procedure for HCl salt
formation included: dissolving in minimum volume of methanol, addition of 1 eq of 1 M HCl in
water, and thorough drying. Hydrochloride salts of inhibitors exist as a mixture of rotamers in
solution of either DMSO or MeOH with approximate ratio of 10:1, NMR is reported for the
major rotamer.
6-(2,2,2-trifluoroethyl)thieno[2,3-d]pyrimidin-4(3H)-one (37). 4,4,4ꢀtrifluorobuteraldehyde 20
g (158 mmol), cyanoacetamide 13.44 g (158 mmol), sulfur 5.12 g (158 mmol) and 27 mL (190
mmol) of Et N were stirred in 160 mL of DMF for 24 h. Solvent was evaporated under reduced
3
pressure. The 2ꢀaminoꢀ5ꢀ(2,2,2ꢀtrifluoroethyl)thiopheneꢀ3ꢀcarboxamide intermediate was
obtained as an oil, which was dissolved in a mixture of 110 mL of triethylorthoformate and 80
mL of acetic acid and refluxed for 4 h (monitored by TLC, EtOAc). Excess of volatiles was
removed under reduced pressure to form oily solid. 200 mL of hexane:EtOAc (1:1) was added.
After stirring for 30 min the precipitate was filtered off, washed with 100 mL of hexane:EtOAc
1
(
1:1) and dried on filter affording 28 g of 37, 76% yield. H NMR CD OD (600 MHz): δ 12.6
3
1
3
(
br s, 1H), 8.14 (s, 1H), 7.42 (s, 1H), 4.07 (q, J = 11.0 Hz, 2H). C NMR CD OD (150 MHz): δ
3
164.5, 157.01, 146.1, 128.4, 124.6, 123.5, 33.6 (q, J = 31.5 Hz).
-chloro-6-(2,2,2-trifluoroethyl)thieno[2,3-d]pyrimidine (38). 28 g (119 mmol) of 37 was
4
added to 70 mL of POCl with 2 drops of DMF. The heterogeneous mixture was refluxed until
3
all solid dissolved and no starting material was observed on TLC (3 h) and then evaporated to
remove the excess of phosphorus oxychloride. The residue was quenched carefully with ice and
saturated ammonia solution with additional external cooling (CAUTION: highly exothermic
reaction). 300 mL of chloroform was added and the mixture was transferred in a separatory
funnel. Water phase was extracted with additional 100 mL of chloroform. Combined organic
extracts were evaporated (no drying) and the residual material was used in the next step without
1
further purification, 28 g of 38, 92% yield. H NMR CDCl (600 MHz): δ 8.86 (s, 1H), 7.39 (s,
3
2
9
ACS Paragon Plus Environment