Structure-based design of high-affinity macrocyclic peptidomimetics to block the menin-mixed lineage leukemia 1 (MLL1) protein-protein interaction

Article abstract of DOI:10.1021/jm3015298

Menin is an essential oncogenic cofactor for mixed lineage leukemia 1 (MLL1)-mediated leukemogenesis through its direct interaction with MLL1. Targeting the menin-MLL1 protein-protein interaction represents a promising strategy to block MLL1-mediated leukemogenesis. Employing a structure-based approach and starting from a linear MLL1 octapeptide, we have designed a class of potent macrocyclic peptidomimetic inhibitors of the menin-MLL1 interaction. The most potent macrocyclic peptidomimetic (MCP-1), 34, binds to menin with a K_i value of 4.7 nM and is >600 times more potent than the corresponding acyclic peptide. Compound 34 is also less peptide-like and has a lower molecular weight than the initial MLL1 peptide. Therefore, compound 34 serves as a promising lead structure for the design of potent and cell-permeable inhibitors of the menin-MLL1 interaction.

Full text of DOI:10.1021/jm3015298

Article
pubs.acs.org/jmc
Structure-Based Design of High-Affinity Macrocyclic
Peptidomimetics to Block the Menin-Mixed Lineage Leukemia 1
(MLL1) Protein−Protein Interaction
Haibin Zhou,† Liu Liu,† Jing Huang,‡,§ Denzil Bernard,† Hacer Karatas,† Alexandro Navarro,† Ming Lei,‡,§
and Shaomeng Wang*,†
‡
†Comprehensive Cancer Center and Departments of Internal Medicine, Pharmacology, Medicinal Chemistry, Howard Hughes
§
Medical Institute, Biological Chemistry, University of Michigan, 1500 E. Medical Center Drive, Ann Arbor, Michigan 48109-0934,
United States
S
* Supporting Information
ABSTRACT: Menin is an essential oncogenic cofactor for
mixed lineage leukemia 1 (MLL1)-mediated leukemogenesis
through its direct interaction with MLL1. Targeting the
menin−MLL1 protein−protein interaction represents a
promising strategy to block MLL1-mediated leukemogenesis.
Employing a structure-based approach and starting from a
linear MLL1 octapeptide, we have designed a class of potent
macrocyclic peptidomimetic inhibitors of the menin−MLL1
interaction. The most potent macrocyclic peptidomimetic
(MCP-1), 34, binds to menin with a Ki value of 4.7 nM and is
>600 times more potent than the corresponding acyclic
peptide. Compound 34 is also less peptide-like and has a lower
molecular weight than the initial MLL1 peptide. Therefore, compound 34 serves as a promising lead structure for the design of
potent and cell-permeable inhibitors of the menin−MLL1 interaction.
A recent study8 has shown that menin interacts with MLL1
via a short bipartite motif, called MLL1MBM (menin-binding
motif of MLL1), spanning the region from residue 6 to residue
INTRODUCTION
■
Chromosomal translocations in the mixed lineage leukemia
gene 1 (MLL1) are observed in >70% of infant leukemia, 5−
10% of acute myeloid leukemia in adults, and in therapy-related
25 (RWRFPARPGTTGGGGGGGRR). MLL1MBM consists of
a highly conserved octameric peptide MLL16−13
leukemia in patients previously treated with topoisomerase II
inhibitors.1 Rearrangements of the MLL1 gene lead to the
(RWRFPARP), a nonstructural polyglycine linker and a
double-arginine motif (R24R25) (Figure 1).8 The cocrystal
fusion of the N-terminus of MLL1 protein to more than 60
functionally diverse partner proteins.2 The resultant MLL1-
structure of menin complexed with MLL1MBM shows that their
interaction is mostly mediated through the conserved octameric
fusion proteins provoke abnormal expression of HOX genes,
which is pivotal in leukemia pathogenesis.3 Menin, a product of
MLL1 peptide, which adopts a cyclized conformation stabilized
by an intramolecular hydrogen bond between the side chain of
the MEN1 gene mutated in the human cancer syndrome known
as multiple endocrine neoplasia type 1 (MEN1), associates with
the extreme N-termini of all the MLL1 fusion proteins and is
Arg8 and the backbone carbonyl of Pro13 (Figure 1). Mutation
data7,8 has also suggested that the binding affinity between
menin and MLL1MBM is dominated by hydrophobic contacts
essential for the recruitment of MLL1 oncoproteins to target
genes during leukemogenesis.4−7
A potential therapeutic strategy for MLL1-rearranged
leukemia is to block the interaction between menin and
through residues Phe9, Pro10, Ala11, and Pro13 of MLL1MBM
and surface hydrophobic pockets in menin (Figure 1).
A class of thienopyrimidine compounds was recently
discovered by a high-throughput screening approach as small-
MLL1, thus inhibiting recruitment of MLL1 fusion proteins to
molecule inhibitors of the menin−MLL1 protein−protein
interaction.9 It was also shown that these inhibitors of the
menin−MLL1 interaction reverse the oncogenic activity of
MLL1 fusion proteins in leukemia cells.9 The crystallographic
structure of a compound from this class bound to menin shows
target loci. It has been shown that expression of short MLL1
peptides containing the menin-binding motif of MLL1
sequence disrupts the in vivo menin−MLL1 interaction and
inhibits the growth of MLL1-AF9 transformed cells.5 Thus,
design of small-molecule inhibitors of the menin−MLL1
interaction represents a plausible approach for the development
of new therapies for the treatment of MLL1 rearranged
leukemia.
Received: October 19, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
value of 22 nM, inhibits cell proliferation, suppresses Hoxa9
expression and induce differentiation in MLL1-rearranged
leukemia cells.10
To probe the MLL1 binding site in menin and facilitate the
design of high-affinity small-molecule menin−MLL1 inhibitors,
we report herein our structure-based design of potent
macrocyclic peptidomimetic inhibitors of the menin−MLL1
interaction. Our efforts have led to the discovery of compound
34, which binds to menin with a Ki value of 4.7 nM and has a
much reduced molecular weight and peptidic characteristics as
compared to the initial MLL1 peptide.
Chemistry. All the linear peptides were synthesized on an
ABI 433 peptide synthesizer using Fmoc chemistry. Rink amide
resin was used as the solid support and yielded amide capping
at the C-terminus, and all the linear peptides were capped with
an acetyl group at the N-terminus.
For the synthesis of the macrocyclic compounds, ring-closing
metathesis11 (Scheme 1) or lactamization reaction (Scheme 2)
was employed to form the ring.
In Scheme 1, the general method for the synthesis of final
compounds IV (compounds 7, 9, 12−15, and 17−35 in Tables
3−6) are outlined. Briefly, intermediates I with a carboxylic acid
at the C-terminus were prepared on an ABI 433 peptide
synthesizer using proline preloaded 2-chlorotrityl chloride resin
as the solid support, followed by the cleavage of the peptides
from resin by treatment with 0.5% TFA in CH2Cl2. Acids I
were coupled to corresponding amines (CH2CH-
(CH2)nNH2) to generate intermediates II. Ring closures of
intermediates II were achieved by ruthenium-catalyzed olefin
metathesis reaction11 to give III, whose CC double bond was
Figure 1. Crystal structure of menin (surface) complexed with
MLL1MBM (cyan) (PDB ID: 3U85). Intramolecular hydrogen bond in
MLL1MBM is shown with dashed lines.
that these compounds bind to the same site as the MLL1
peptide but have a different binding mode.10 The most potent
compound reported from this class binds to menin with a Ki
a
Scheme 1. General Method for the Synthesis of Macrocyclic Peptidomimetics Using RCM Reaction
a
Reagents and conditions: (a) (i) ABI 433 peptide synthesizer, (ii) 0.5% TFA in CH2Cl2; (b) CH2CH(CH2)nNH2, HATU, HOAt, DIEA; (c)
Grubbs Catalyst (first generation), CH2Cl2; (d) (i) H2, Pd/C, (ii) TFA:TES:H2O (18 mL:0.5 mL:1 mL) or diethylamine then TFA:TES:H2O (18
mL:0.5 mL:1 mL). The structures for all final compounds are depicted in Tables 3−6.
B
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
a
Scheme 2. General Method for the Synthesis of Macrocyclic Peptidomimetics Using Lactamization Reaction
a
Reagents and conditions: (a) (i) ABI 433 peptide synthesizer, (ii) 0.5% TFA in CH2Cl2; (b) (i) HATU, HOAt, DIEA, (ii) TFA:TES:H2O (18
mL:0.5 mL:1 mL). Structures of compounds 10, 11, and 16 are outlined in Tables 3 and 4.
a
Scheme 3. Synthesis of 37
a
Reagents and conditions: (a) (i) ABI 433 peptide synthesizer, (ii) 0.5% TFA in CH2Cl2; (b) (i) VII, HATU, HOAt, DIEA, (ii) diethyl amine; (c)
(i) 5-FAM.SE, DIEA, (ii) TFA:TES:H2O (18 mL:0.5 mL:1 mL).
reduced by catalytic hydrogenation, followed by removal of the
protecting groups to yield final compounds IV. The acyclic
compound 8 was synthesized using a similar strategy as that for
intermediate II (Supporting Information Scheme S1).
Final compounds VI (compounds 10, 11, and 16 in Tables 3
and 4) were synthesized through lactamization as shown in
Scheme 2. Proline preloaded 2-chlorotrityl chloride resin was
used as the solid support on ABI 433 peptide synthesizer. TFA
in CH2Cl2 was used to cleave the peptides from the resin,
which also led to removal of the protecting groups at the
terminal amines to give intermediates V. Cyclization of
intermediates V were achieved using HATU and HOAt by
coupling the acid at the C-terminus and amine at the N-
terminus. Removal of the protecting groups yielded final
compounds VI.
carboxy fluorescein succinimide ester (5-FAM, SE) to the side
chain of the lysine of the immobilized linear peptide which was
synthesized on ABI 433 peptide synthesizer. The fluorescent
labeled tracer 37 was prepared as shown in Scheme 3.
Intermediate VII was made by the procedure used for
intermediates I. Acid VII was coupled to amine VIII, followed
by removal of the Fmoc group to give IX. Treatment of IX with
5-FAM, SE fluorophore in the presence of DIEA, followed by
removal of the protecting groups, yielded 37.
RESULTS AND DISCUSSION
■
Determination of the Minimal Motif in MLL1 for High-
Affinity Binding to Menin. The crystal structure of
MLL1MBM complexed with menin showed that the octameric
peptide MLL16−13 motif at the N-terminus of MLL1MBM sits in
a well-defined surface pocket in menin.8 The crystal structure
also showed that the seven glycines in the peptide have minimal
As shown in Scheme S2 (Supporting Information),
fluorescent labeled tracer 36 was synthesized by tethering 5-
C
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
contacts with menin.8 Indeed, removal of these seven glycines
in this peptide had minimal effect on the binding affinity to
menin.8
Linear peptide 2 has a relatively small size (eight residues)
and high affinity (Ki = 71.1 nM) to menin and was thus
employed as the starting point for our structure-based design of
macrocyclic peptidomimetics. In the cocrystal structure of the
menin−MLL1MBM complex, the octapeptide MLL16−13 adopts
a cyclized conformation, stabilized by an intramolecular
hydrogen bond formed between the side chain of Arg8 and
the backbone carbonyl of Pro13 (Figure 1). We designed and
synthesized a macrocyclic compound 7 by replacing the side
chain of Arg8 with a 7-carbon aliphatic amine in 2 and coupling
the amino group to the carboxylic acid group in Pro13 to form
an amide (Figure 2). A 7-carbon linker was selected, initially,
because there are seven atoms between the backbones of Arg8
and Pro13 in 2, including the hydrogen in the hydrogen bond
(Figure 2). Peptide 8, an acyclic counterpart of 7, was
synthesized to distinguish the effect of cyclization from any
additional interaction introduced by the linker (Figure 2).
Compound 7 binds to menin with a Ki value of 23.8 nM and
is 3 times more potent than the acyclic, wild-type peptide 2
(Figure 2). In contrast, the acyclic control compound, 8, has a
Ki value of 2170 nM, 91 times less potent than compound 7
(Figure 2). This indicates that the intramolecular hydrogen
bond has a significant contribution to the high binding affinity
of compound 2, and cyclization in compound 7 can successfully
replace this intramolecular hydrogen bond to effectively
constrain the conformation to mimic the bound conformation
of 2 to menin.
To understand the structural basis for its high binding affinity
to menin, we determined the cocrystal structure of 7 complexed
with menin (Figure 3a). The coordinates for the cocrystal
structure for compound 7 in complex with menin have been
deposited into the Protein Data Bank (PDB ID: 4I80). Detailed
data collection and refinement statistics are provided in Table
2.
In this cocrystal structure, the alkyl linker of compound 7 has
no direct interaction with the protein; however, cyclization
preorganizes the key residues to interact with the menin
protein. Compared with the menin−MLL1MBM costructure, the
backbone of 7 occupies the same position as that of MLL16−13
in MLL1MBM and thus captures all the critical interactions
observed in the menin−MLL1MBM complex (Figure 3b). The
unusual backbone dihedral geometry of Pro10 in MLL1 causes
its intrusion, together with the phenyl ring of Phe9 in MLL1,
into the deep hydrophobic well formed by the side chain of
residues Leu175, Leu177, Ala182, Phe238, Tyr276, and Met278
of menin (Figure 3a). The side chain of Ala11 interacts with a
shallow hydrophobic pocket in menin, while the backbone
carbonyl forms a hydrogen bond with Tyr323. The pyrrolidine
To evaluate the binding contribution of the C terminal
double arginines to menin, we synthesized peptide 1 (Ac-
RWRFPARPGTGRR-NH2) and a shorter peptide 2 (Ac-
RWRFPARP-NH2). In our optimized fluorescence-polarization
(FP) based competitive binding assay, 1 binds to menin with a
Ki value of 2.9 nM, whereas 2 has a Ki value of 71.1 nM, being
25-fold less potent than 1 (Table 1). Consistent with the
previous results,8 our data showed that the double arginine
motif makes a significant contribution to binding.
Table 1. Binding Affinities of Truncated MLL1 Peptides to
Menin
compd
sequence
IC50 SD (nM) Ki SD (nM)
1
Ac-RWRFPARPGTGRR-
NH2
11.4 3.3
2.9 1.3
2
3
4
5
6
Ac-RWRFPARP-NH2
Ac-RWRFPAR-NH2
Ac-WRFPARP-NH2
Ac-RFPARP-NH2
Ac-FPARP-NH2
232 54
85300 5830
1060 142
71.1 19.5
25500 1740
311 55
9570 2830
140000 4448
2850 821
41800 1330
We next determined the minimal sequence of 2 for high-
affinity binding to menin. Removal of the C-terminal residue,
Pro13, in 2 afforded 3, which binds to menin with a Ki value of
25500 nM and is >300-fold less potent than 2. This dramatic
loss in binding affinity upon removal of Pro13 is consistent with
its extensive hydrophobic contacts with menin and the
formation of an intramolecular hydrogen bond between the
backbone carbonyl of Pro13 and the side chain of Arg8.
Sequential deletions of N-terminal residues Arg6, Trp7, and
Arg8 in 2 resulted in 4, 5, and 6, which have Ki values of 311,
2850, and 41800 nM, respectively (Table 1). Compounds 4, 5,
and 6 are thus 4-, 40-, and 588-fold less potent than 2,
respectively. The binding data thus show that, consistent with
previously published mutagenesis data and cocrystal struc-
ture,7,8 each of these four residues (Arg6, Trp7, Arg8, Pro13)
makes a significant contribution to the binding of 2 to menin.
Structure-Based Design of Macrocyclic Peptidomi-
metics. As compared to their linear counterparts, macrocyclic
molecules can achieve higher affinities and specificities to their
target proteins through a conformationally constrained,
preorganized ring structure.12,13 We have therefore designed
and synthesized macrocyclic peptidomimetics starting from
linear MLL1 peptides.
Figure 2. Design of macrocycle 7 and acyclic control compound 8.
D
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 3. Crystal structure of menin (gray) complexed with compound 7 (yellow). (a) Overall binding mode for 7 with residues forming the binding
pocket shown as sticks. (b) Superposition of the crystal structure of 7 (yellow) onto the crystal structure of MLL1MBM (cyan). Hydrogen bonds are
indicated with dashed black lines. Figures were prepared with Pymol.
Table 2. Crystallographic Data Collection and Refinement
Statistics
Table 3. Structures and Binding Affinities of Macrocyclic
Peptidomimetics with Varying Length of Linker
menin−7
data collection
space group
P41212
cell dimensions
a, b, c (Å)
141.379, 141.379, 92.879
90, 90, 90
α, β, γ (deg)
resolution (Å) (high-resolution shell)
Rmerge (%) (high-resolution shell)
I/σ (high-resolution shell)
completeness (%) (high-resolution shell)
redundancy (high-resolution shell)
Refinement
100−3.10 (3.21−3.10)
9.8 (29.1)
41.7 (8.5)
92.4 (73.1)
25.5 (20.1)
resolution (Å)
44.7−3.1
16307
compd
linker length (n + 3)
IC50 (nM)
Ki (nM)
no. of reflections
2
N/A
232 54
78.1 11.7
7390 1160
63.3 12.1
514 28
71.1 19.5
23.8 7.2
2170 372
17.5 2.7
R
work/Rfree (%)
no. of atoms
protein
B-factors
protein
rms deviations
bond lengths (Å)
bond angles (deg)
21.49/24.54
7
7
N/A
6
8
3895
9
10
11
12
13
5
136
6
107.89
4
11000 1050
25.0 3.4
144 20
3320 359
6.9 1.0
8
0.009
1.246
9
42.5 10.2
ring of Pro13 is sandwiched between the side chains of Tyr319
and Tyr323 in menin (Figure 3a). The positively charged side
chain of Arg12 is involved in charge−charge and hydrogen
bond interactions with the negatively charged Glu359 and
Glu363 residues in menin (Figure 3a). Another hydrogen bond
is seen with the backbone carbonyl of peptide Arg6 and
Asn244. Superposition of the bound conformations of 7 and
MLL1MBM (Figure 3b) clearly shows that our cyclization
strategy does not affect the bound conformation but reinforces
it.
We further investigated the effect of the macrocyclic ring size
in compound 7 for binding to menin by varying the length of
the alkyl linker, and the results are shown in Table 3. While
compound 9 with a 6-methylene linker binds to menin with an
affinity similar to 7, compound 10 with a 5-methylene linker
and compound 11 with a 4-methylene linker are 6 and 139
times less potent than compound 7, respectively. These results
indicate that linkers shorter than the 6-methylene linker may
significantly distort the conformation of these macrocycles from
that desired for optimal binding to menin.
Compound 12, with an 8-carbon linker, binds to menin with
a Ki value of 6.9 nM and is 3 times more potent than 7.
However, 13 with a 9-carbon linker is 2 times less potent than 7
(Table 3). Thus, macrocyclic compound 12 is 10 times more
potent than the acyclic, wild-type peptide 2 and 300 times more
potent than the acyclic control compound 8. These data show
that compound 12 in this series of macrocycles has the optimal
ring size for effective interaction with menin.
Optimization of Compound 12. Although compound 12
has high affinity to menin, it consists of eight amino residues.
E
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
We next sought to reduce its size and peptidic characteristics.
Because residues Arg6 and Trp7 in 12 are not part of the
macro-ring system and are not critical for maintaining the ring
conformation for binding to menin, we replaced these two
residues with NH2, NHCOCH3, or H, which resulted in 14, 15,
and 16, respectively (Table 4). Compounds 14, 15, and 16
six residues, it is an attractive lead structure for further
optimization.
Further Optimization of Compound 14. We next
modified 14 in an effort to enhance its binding affinity to
menin. To identify the regions of the binding pocket most
likely to contribute to improved binding affinity, we determined
the hydrophobic hotspots on the menin surface by molecular
dynamics simulations (MD) using isopropyl alcohol and
phenol independently as cosolvents (Figure 4). Previous
studies have shown the ability of cosolvent simulations to
effectively map the hotspot regions as well as identifying
protein conformations relevant to ligand binding.14−16
Using two different cosolvents for MD simulations of menin,
we mapped convergent regions of favorable hydrophobic
interactions that encompassed the Phe9, Pro10, and Ala11 in
MLL1MBM for binding to menin. Because the Pro10 residue is
critical to maintain the conformation of this macrocycle, we
focused our modifications on Phe9 and Ala11 to enhance the
hydrophobic interactions between these two residues in 14 and
menin.
Table 4. Structures and Binding Affinities of Compounds 14,
15, and 16
The cocrystal structure of 7 complexed with menin (Figure
3) showed that the methyl side chain group of Ala11 interacts
with a small hydrophobic pocket in menin, and modeling
analysis (Figure 4) showed that there is room to accommodate
a slightly larger group at this site. We therefore designed and
synthesized a series of analogues of 14 in which we employed
different hydrophobic groups to probe the small binding pocket
occupied by the methyl group in Ala11. As shown in Table 5,
compound 17 with an additional methyl group on the α-carbon
of Ala11, is 3 times more potent than 14, while compound 18,
which contains an ethyl group, is 16 times less potent than 14.
On the other hand, compound 19, which contains a three-
membered ring, is slightly more potent than 17.
Encouraged by the data for 19, we synthesized 20 and 21,
with a cyclobutyl or cyclopentyl group. Compound 20 binds to
menin with a Ki value of 26.7 nM, which is twice as potent as
19 and 7 times more potent than 14. However, compound 21
is 8 times less potent than 14 and 62 times less potent than 20.
Thus the cyclobutyl group at this position was determined to
be optimal for binding to menin.
bind to menin with Ki values of 197, 586, and 1840 nM and are
29, 85, and 267 times less potent than 12, respectively (Table
4). These data show that Arg6 and Trp7, although not critical,
contribute to the binding of 12 to menin. Interestingly, 14,
which has a positively charged amino group, is 3 times more
potent than 15 and 9 times more potent than 16, indicating
that a positively charged group at this site can enhance binding
to menin, presumably due to the highly acidic nature of the
binding cavity of menin. Because compound 14 consists of only
Figure 4. Hydrophobic hot spots (mesh) in the MLL1 peptide (cartoon) binding groove detected from cosolvent isopropyl alcohol (a) and phenol
(b) mapping method based on 16 ns MD simulations with the menin (1−386) crystal structure (surface). MLL1 residues in the hotspot regions are
shown in stick.
F
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
pocket for the phenyl group can accommodate larger
hydrophobic moieties at the ortho and meta positions, the
electronic properties of the substituent may also be important.
The protein wall along one side of the phenyl binding site is
lined by several acidic residues, and the strong electronegativity
of fluorine may enhance polarization of the ring, leading to
better binding affinity. To further investigate the effect of F-
substitution at different positions, we synthesized 32 with F-
substitution at the ortho-position and 33 with F-substitution at
the para-position. 32 binds to menin with a Ki value of 27.5
nM, similar to that of 20, while 33 binds to menin with a Ki
value of 123 nM and is 5 times less potent than 20. Because F-
substitution at the meta-position improves the binding affinity
to menin, we synthesized and tested compound 34 with two F-
substitutions at the two meta-positions on the phenyl ring.
Compound 34 binds to menin with a Ki value of 4.7 nM and is
thus 6 times more potent than 20 (Figure 5). To test the
stereospecificity, we changed the 3,5-difluoro-L-Phe9 in 34 to
3,5-difluoro-D-Phe9, which resulted in 35, an epimer of 34.
Compound 35 binds to menin with a Ki value of >100 μM and
is >10000 times less potent than 34 (Figure 5). Modeling
studies suggested that changing the stereochemistry projects
the phenyl residue outside the pocket and also disrupts other
interactions with the protein, leading to very poor binding to
menin compared to 34 (Supporting Information).
Table 5. Structures and Binding Affinities of Compounds
with the Modified Ala11
Development and Optimization of Competitive Bind-
ing Assays. To determine quantitatively and accurately the
binding affinities of our designed compounds to menin protein,
we developed and optimized sensitive fluorescence polarization
(FP)-based competitive binding assays. All IC50 and Ki values
reported here are averages and standard deviations (SDs) of at
least three independent experiments.
First, on the basis of peptide 1, we designed and synthesized
a fluorescent labeled tracer, 36, containing a 14-residue MLL1
segment, a spacer (Ahx-Ahx), a lysine residue, and a
fluorophore tethered to the amino group of the side chain in
the lysine residue (Figure 6a). The Kd value of 36 to menin was
determined to be 15.9 nM (Figure 6b). Using 36 and menin
protein, we established a competitive binding assay and used
this assay to evaluate the binding affinities of our initial
compounds.
During the course of our work, some of our designed
macrocyclic peptidomimetics were found to have very high
binding affinities to menin, exceeding the lower limits of the
initial competitive binding assay. To accurately determine the
binding affinities of our designed compounds, we designed and
synthesized a new tracer, 37, based on the macrocyclic
peptidomimetic 34 (Figure 6a). The Kd value of 37 to menin
was determined to be 1.4 nM (Figure 6c). Using 37 and menin
protein, we reoptimized our FP-based competitive binding
assay. All the binding data reported in the present study were
obtained using the optimized assay.
The cocrystal structure of menin complexed with compound
7 showed that Phe9 is nested in a hydrophobic pocket of menin
(Figure 3), and our hotspot analysis also indicated that this
region is favorable for hydrophobic interactions (Figure 4). We
therefore made extensive modifications of the Phe9 residue in
20 to explore this hydrophobic pocket (Table 6). Replacement
of the phenyl group with a cyclohexyl group resulted in 22,
which is 32 times less potent than 20. Introduction of a
chlorine substituent at the ortho- or meta- or para-position of
the phenyl ring in 20 yielded 23, 24, and 25, which bind to
menin with Ki values of 30, 44.6, and 549 nM, respectively.
These data clearly show that a p-Cl substituent on the phenyl
group is detrimental while either an o- or m-Cl substituent has
minimal effect.
We then focused on modifications at the m-position of the
phenyl ring in Phe9 and synthesized compounds containing
substituents with varying size and electronic property. The
resulting compounds with a methyl, trifluoromethyl, ethyl,
isobutyl, or phenyl substituent are 12, 27, 50, 434, and 524
times less potent than compound 20, respectively (Table 6).
However, compound 31, with a m-F substitution, binds to
menin with a Ki value of 6.8 nM and is 4 times more potent
than 20 (Table 6).
SUMMARY
■
In this study, we first determined the minimal MLL1 segment
needed for high-affinity binding to menin. Our systematic
truncation studies showed that peptide 2, which contains an 8-
residue segment of MLL1, binds to menin with a high affinity
(Ki value of 71.1 nM) and further shortening from either the N-
or the C-terminus leads to a major decrease in the binding
affinity.
The surprisingly similar affinities of 20, 23, and 24, and the
improved binding of 31, suggested that while the binding
Utilizing the cocrystal structure of MLL1MBM complexed with
menin, we designed and synthesized a series of macrocyclic
G
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Table 6. Structures and Binding Affinities of Compounds with the Modified Phe9
peptidomimetics based upon peptide 2 through cyclization of
the side chain of Arg8 and backbone carbonyl group of Pro13.
Compound 12 with an 8-carbon cyclization linker has a Ki value
of 6.9 nM to menin, 10 times more potent than its acyclic, wild-
type counterpart. Determination of a cocrystal structure of 7, a
potent macrocyclic analogue of 12, in complex with menin,
established the structural basis for its high affinity binding to
menin and assisted our further structure-based design and
optimization.
Our extensive optimization of compound 12 resulted in a
highly potent macrocyclic peptidomimetic 34. Compound 34
binds to menin with a Ki value of 4.7 nM and is >600 times
more potent than the corresponding acyclic peptide and 15
times more potent than the initial starting linear peptide 2.
Importantly, compound 34 has much reduced molecular weight
and peptidic characteristics as compared to 2. Further
optimization of 34 may ultimately yield a class of potent and
cell-permeable small-molecule inhibitors of the menin−MLL1
interaction as potential new therapeutics for the treatment of
acute leukemia with MLL1 rearrangements.
EXPERIMENTAL SECTION
■
Figure 5. (a) Competitive binding curves and Ki values of compounds
2, 5, 34, and 35 to menin as determined using an FP-based binding
assay. (b) Chemical structures of 34 and 35.
General Chemistry Information. Unless otherwise stated, all
reagents and solvents were used as supplied without further
purification. The final products were purified by a C18 reverse
phase semipreparative HPLC column with solvent A (0.1% of TFA in
water) and solvent B (0.1% of TFA in CH3CN) as eluents. The
H
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 6. (a) Chemical structures of fluorescent labeled tracers 36 and 37. (b) Binding isotherm of 36 to menin. (c) Binding isotherm of 37 to
menin.
immobilized amino acid residues on the resin were protected as
follows: Arg(Pbf), Trp(Boc), Lys(Mtt), Homolys(Boc). The purity
was determined by Waters ACQUITY UPLC, and all the final
compounds were >95% pure. The characterization of the compounds
was accomplished by HRMS (ESI+) (Agilent Q-TOF Electrospray).
1H NMR and 13C NMR spectra of the representative compounds were
acquired at a proton frequency of 300 MHz, and chemical shifts are
reported in parts per million (ppm) relative to an internal standard.
These data are provided in the Supporting Information.
Synthesis of Linear Peptides in Table 1. The linear peptides
were synthesized with an ABI 433 peptide synthesizer using Fmoc
chemistry. Rink amide resin was used as the solid support yielding
amide capping at the C-terminus. All the peptides were capped with
acetyl at the N-terminus. TFA:TES:H2O (18 mL:0.5 mL:1 mL)
cleavage cocktail was used to cleave the peptides from resin, which also
led to removal of the protecting groups. The cleavage solution was
evaporated, and the crude product was precipitated with diethyl ether
followed by HPLC purification.
mg, 0.15 mmol, 3 equiv), and DIEA (35 μL, 0.20 mmol, 4 equiv) in
THF (10 mL) was stirred at room temperature for 2 h then
concentrated. Purification of this residue by HPLC afforded
compound II.
A solution of bis(tricyclohexylphosphine)ben-zylidine ruthenium-
(IV) dichloride (Grubbs First Generation Catalyst) (5.6 mg, 0.0068
mmol, 0.2 equiv) in DCM (30 mL) was degassed then added to a
solution of peptide II (0.034 mmol, 1 equiv) in DCM (30 mL) under
N2. The resulting solution was stirred at room temperature for 3 h
before filtering through Celite and concentrated. Purification of the
residue afforded III.
To a solution of peptide III (0.021 mmol, 1 equiv) in MeOH (10
mL) was added 10% Pd−C (20 mg). The solution was stirred under 1
atm of H2 at room temperature for 0.5 h before being filtered through
Celite and concentrated. The residue was treated with a mixture of
TFA:TES:H2O (18 mL:0.5 mL:1 mL) for 10 min at room temperature
and evaporated. Purification of the crude product by HPLC afforded
IV.
General Procedures for Synthesis of VI (Scheme 2).
Compound V was prepared using a similar procedure as that used
for compound I in Scheme 1. A solution of V (0.05 mmol, 1 equiv) in
THF (20 mL) was added dropwise to a stirred solution of HATU (38
mg, 0.1 mmol, 2 equiv), HOAt (14 mg, 0.1 mmol, 2 equiv), and DIEA
(35 μL, 0.2 mmol, 4 equiv) in THF (10 mL) under N2. The resulting
solution was stirred at room temperature for 3 h before being
concentrated. The residue was purified and then was treated with a
General Procedures for Synthesis of Final Compounds IV
(Scheme 1). Peptide I was synthesized on an ABI 433 peptide
synthesizer using proline preloaded 2-chlorotrityl chloride resin as the
solid support. The peptide was cleaved from the resin with 0.5% TFA
in CH2Cl2, yielding a carboxylic acid at the C-terminus. The cleavage
solution was evaporated, and the crude product was purified by HPLC.
A solution of acid I (0.05 mmol, 1 equiv), the corresponding amine
(0.15 mmol, 3 equiv), HATU (57 mg, 0.15 mmol, 3 equiv), HOAt (20
I
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
mixture of TFA:TES:H2O (18 mL:0.5 mL:1 mL) for 10 min at room
temperature and evaporated. Purification of the residue by HPLC
afforded VI.
Crystallographic Studies. The expression and purification of
human menin protein and crystallization methods are the same as
described previously.8 Briefly, crystallization of menin with 7 was
achieved by sitting-drop diffusion with a well solution containing 100
mM Tris-HCl (pH 7.0), 200 mM MgCl2, and 2.3 M NaCl. The
menin−7 complex structure was determined by multiwavelength
anomalous dispersion to a resolution of 3.1 Å. Data collection and
refinement statistics are given in Table 2.
Computational Methods. The menin−MLL1MBM complex
(PDBID: 3U85)8 was used to extract the MLL1-binding domain
(residues 1−386) and used for hotspot detection using cosolvent
simulations in Amber18 with the ff99SB force field.19 Then 16 ns
simulations were performed in each 20% v/v isopropyl alcohol/water,
and 10%v/v phenol/water cosolvent media following the protocol
described previously.14−16 Initial alignment of saved conformations
was performed using the ptraj utility from the Amber suite. Helices
comprising α11 (residues 277−289), α12 (residues 298−312), and
α13 (residues 318−331) were used for alignment. Subsequently,
hotspots were determined based on grid occupancy of cosolvent atoms
and used for design strategy.
Synthesis of 36 (Supporting Information Scheme S2). The
immobilized peptide, Ac-SRWRFPARPGTGRR-Ahx-Ahx-K(Mtt) (0.1
mmol), on Rink amide resin was achieved on an ABI 433 peptide
synthesizer. The resin was then treated with 1% TFA in CH2Cl2 (10
mL) to remove the 4-methyltrityl (Mtt) protecting group at the lysine
side chain, and this was followed by overnight treatment with 5-
carboxy fluorescein succinimide ester (5-FAM, SE) (71 mg, 0.15
mmol, 1.5 equiv) and DIEA (52 μL, 0.3 mmol, 3 equiv) in DMF. The
peptide was cleaved from the resin and followed by HPLC purification.
Synthesis of 37 (Scheme 3). Compound VII was prepared using
a similar procedure as that used for I. A solution of carboxylic acid VII
(45 mg, 0.04 mmol, 1 equiv), VIII (43 mg, 0.04 mmol, 1 equiv),
HATU (30 mg, 0.08 mmol, 2 equiv), HOAt (11 mg, 0.08 mmol, 2
equiv), and DIEA (35 μL, 0.20 mmol, 5 equiv) in THF (10 mL) was
stirred at room temperature overnight and then concentrated. Then
the residue was dissolved in CH3CN (10 mL) and treated with
diethylamine (82 μL, 20 equiv) for 10 min. The reaction mixture was
concentrated and purified by HPLC to give IX.
To a solution of IX (20 mg, 0.01 mmol, 1 equiv) in CH2Cl2 (3 mL),
5-FAM, SE (9.6 mg, 0.02 mmol, 2 equiv), and DIEA (7 μL, 0.04
mmol, 4 equiv) were added, and the resulting solution was stirred for 3
h before being concentrated. The residue was treated with a
TFA:TES:H2O (18 mL:0.5 mL:1 mL) cocktail for 10 min and
concentrated and purified to yield 37 (8 mg, 12% in four steps).
FP-Based Binding Assays. Both fluorescence and fluorescence
polarization were measured using the Tecan Infinite M-1000 plate
reader (Tecan U.S., Research Triangle Park, NC) in Microfluor 1, 96-
well, black, round-bottom plates (Thermo Scientific).
The Kd values of the fluorescent probes (36 and 37) to menin
protein were determined by monitoring the total FP of mixtures
composed of each fluorescent probe at a fixed concentration and the
menin protein with increasing concentrations up to full saturation.
Then 3 nM of 36 or 2 nM of 37 and menin protein were added to
each well to a final volume of 100 μL in the assay buffer (100 mM
potassium phosphate, pH 7.5, 100 μg/mL bovine γ-globulin, 0.02%
sodium azide [Invitrogen], 2% DMSO, and 0.005% of Triton X-100
for 36 saturation assay or 0.02% of Tween-20 for 37 saturation assay,
respectively). Plates were mixed and incubated at room temperature
for 60 min with gentle shaking to ensure equilibrium. FP values were
measured at an excitation wavelength of 485 nm and an emission
wavelength of 530 nm using the Infinite M-1000 plate reader (Tecan
U.S., Research Triangle Park, NC) in Microfluor 1, 96-well, black,
round-bottom plates (Thermo Scientific). Kd values were then
calculated by fitting the sigmoidal dose-dependent FP increases as a
function of protein concentrations using Graphpad Prism 5.0 software
(Graphpad Software, San Diego, CA).
Ki values of tested compounds were determined in dose-dependent
competitive binding experiments. Mixtures of 2 μL of the tested
compound with different concentrations in DMSO and 98 μL of
preincubated menin/36 or menin/37 complex with fixed concen-
trations in the assay buffer (100 mM potassium phosphate, pH 7.5,
100 μg/mL bovine γ-globulin, 0.02% sodium azide, with 0.005%
Triton X-100 for assay using 36 and 0.02% Tween-20 for assay using
37) were added into assay plates and incubated at room temperature
with gentle shaking for 60 min. Final concentrations of menin protein
and 36 were 30 and 3 nM, and menin protein and 37 were 3 and 2
nM, respectively, and final DMSO concentration was 2%. Negative
controls containing menin/36 or menin/37 complex only (equivalent
to 0% inhibition), and positive controls containing free 36 or 37 only
(equivalent to 100% inhibition) were included in each assay plate. FP
values were measured as described above. IC50 values were determined
by nonlinear regression fitting of the sigmoidal dose-dependent FP
decreases as a function of total compound concentrations using
Graphpad Prism 5.0 software (Graphpad Software, San Diego, CA). Ki
values of tested compounds to the menin protein were calculated using
the measured IC50 values, the Kd values of probes to menin, and the
concentrations of menin and probes in the competitive assays.17
ASSOCIATED CONTENT
■
S
* Supporting Information
Synthesis of 8 and 36, HRMS and NMR spectral data and
UPLC purity analysis for designed compounds, and computa-
tional modeling of menin complexes with compounds 34 and
35. This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*Phone: 734-615-0362 Fax: 734-647-9647 E-mail: shaomeng@
umich.edu.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Use of the Advanced Photon Source was supported by the US
Department of Energy, Office of Science, Office of Basic Energy
Sciences, under contract no. DE-AC02-06CH11357.
ABBREVIATIONS USED
■
MLL, mixed lineage leukemia; MLL1MBM, menin-binding motif
of MLL1; MEN1, multiple endocrine neoplasia type 1; FP,
fluorescence-polarization; Fmoc, fluorenylmethoxycarbonyl;
DIEA, N,N-diisopropylethylamine; 5-FAM, SE, 5-carboxyfluor-
escein succinimidyl ester; Ahx, 6-aminohexanoic acid; Mtt, 4-
methyltrityl; TFA, trifluoroacetic acid
REFERENCES
■
(1) Krivtsov, A. V.; Armstrong, S. A. MLL translocations, histone
modifications and leukaemia stem-cell development. Nature Rev.
Cancer 2007, 7, 823−833.
(2) Meyer, C.; Kowarz, E.; Hofmann, J.; Renneville, A.; Zuna, J.;
Trka, J.; Ben Abdelali, R.; Macintyre, E.; De Braekeleer, E.; De
Braekeleer, M.; Delabesse, E.; de Oliveira, M. P.; Cave, H.; Clappier,
E.; van Dongen, J. J.; Balgobind, B. V.; van den Heuvel-Eibrink, M. M.;
Beverloo, H. B.; Panzer-Grumayer, R.; Teigler-Schlegel, A.; Harbott, J.;
Kjeldsen, E.; Schnittger, S.; Koehl, U.; Gruhn, B.; Heidenreich, O.;
Chan, L. C.; Yip, S. F.; Krzywinski, M.; Eckert, C.; Moricke, A.;
Schrappe, M.; Alonso, C. N.; Schafer, B. W.; Krauter, J.; Lee, D. A.;
Zur Stadt, U.; Te Kronnie, G.; Sutton, R.; Izraeli, S.; Trakhtenbrot, L.;
Lo Nigro, L.; Tsaur, G.; Fechina, L.; Szczepanski, T.; Strehl, S.;
Ilencikova, D.; Molkentin, M.; Burmeister, T.; Dingermann, T.;
J
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Klingebiel, T.; Marschalek, R. New insights to the MLL recombinome
of acute leukemias. Leukemia 2009, 23, 1490−1499.
(3) Daser, A.; Rabbitts, T. H. Extending the repertoire of the mixed-
lineage leukemia gene MLL in leukemogenesis. Genes Dev. 2004, 18,
965−974.
(4) Yokoyama, A.; Somervaille, T. C.; Smith, K. S.; Rozenblatt-
Rosen, O.; Meyerson, M.; Cleary, M. L. The menin tumor suppressor
protein is an essential oncogenic cofactor for MLL-associated
leukemogenesis. Cell 2005, 123, 207−218.
(5) Caslini, C.; Yang, Z.; El-Osta, M.; Milne, T. A.; Slany, R. K.; Hess,
J. L. Interaction of MLL amino terminal sequences with menin is
required for transformation. Cancer Res. 2007, 67, 7275−7283.
(6) Chen, Y. X.; Yan, J.; Keeshan, K.; Tubbs, A. T.; Wang, H.; Silva,
A.; Brown, E. J.; Hess, J. L.; Pear, W. S.; Hua, X. The tumor suppressor
menin regulates hematopoiesis and myeloid transformation by
influencing Hox gene expression. Proc. Natl. Acad. Sci. U. S. A. 2006,
103, 1018−1023.
(7) Grembecka, J.; Belcher, A. M.; Hartley, T.; Cierpicki, T.
Molecular basis of the mixed lineage leukemia−menin interaction:
implications for targeting mixed lineage leukemias. J. Biol. Chem. 2010,
285, 40690−40698.
(8) Huang, J.; Gurung, B.; Wan, B.; Matkar, S.; Veniaminova, N. A.;
Wan, K.; Merchant, J. L.; Hua, X.; Lei, M. The same pocket in menin
binds both MLL and JUND but has opposite effects on transcription.
Nature 2012, 482, 542−546.
(9) Grembecka, J.; He, S.; Shi, A.; Purohit, T.; Muntean, A. G.;
Sorenson, R. J.; Showalter, H. D.; Murai, M. J.; Belcher, A. M.; Hartley,
T.; Hess, J. L.; Cierpicki, T. Menin−MLL inhibitors reverse oncogenic
activity of MLL fusion proteins in leukemia. Nature Chem. Biol. 2012,
8, 277−284.
(10) Shi, A.; Murai, M. J.; He, S.; Lund, G.; Hartley, T.; Purohit, T.;
Reddy, G.; Chruszcz, M.; Grembecka, J.; Cierpicki, T. Structural
insights into inhibition of the bivalent menin−MLL interaction by
small molecules in leukemia. Blood 2012, 120, 4461−4469.
(11) Blackwell, H. E.; Grubbs, R. H. Highly efficient synthesis of
covalently cross-linked peptide helices by ring-closing metathesis.
Angew. Chem. Int. Ed. 1998, 37, 3281−3284.
(12) Driggers, E. M.; Hale, S. P.; Lee, J.; Terrett, N. K. The
exploration of macrocycles for drug discoveryan underexploited
structural class. Nature Rev. Drug Discovery 2008, 7, 608−624.
(13) Marsault, E.; Peterson, M. L. Macrocycles are great cycles:
applications, opportunities, and challenges of synthetic macrocycles in
drug discovery. J. Med. Chem. 2011, 54, 1961−2004.
(14) Yang, C.-Y.; Wang, S. Computational Analysis of Protein
Hotspots. ACS Med. Chem. Lett. 2010, 1, 125−129.
(15) Yang, C.-Y.; Wang, S. Hydrophobic Binding Hot Spots of Bcl-xL
Protein−Protein Interfaces by Cosolvent Molecular Dynamics
Simulation. ACS Med. Chem. Lett. 2011, 2, 280−284.
(16) Yang, C.-Y.; Wang, S. Analysis of Flexibility and Hotspots in
Bcl-xL and Mcl-1 Proteins for the Design of Selective Small-Molecule
Inhibitors. ACS Med. Chem. Lett. 2012, 3, 308−312.
(17) Nikolovska-Coleska, Z.; Wang, R.; Fang, X.; Pan, H.; Tomita,
Y.; Li, P.; Roller, P. P.; Krajewski, K.; Saito, N. G.; Stuckey, J. A.;
Wang, S. Development and optimization of a binding assay for the
XIAP BIR3 domain using fluorescence polarization. Anal. Biochem.
2004, 332, 261−273.
(18) Case, D. A.; Darden, T. A.; Cheatham, T. E., III; Simmerling, C.
L.; Wang, J.; Duke, R. E.; Luo, R.; Crowley, M.; Walker, R. C.; Zhang,
W.; Merz, K. M.; Roberts, B.; Wang, J.; Hayik, S.; Roitberg, A.; Seabra,
́
G.; Kolossvary, I.; K. F..Wong; Paesani, F.; Vanicek, J.; Wu, X.; Brozell,
S. R.; Steinbrecher, T.; Gohlke, H.; Yang, L.; Tan, C.; Mongan, J.;
Hornak, V.; Cui, G.; Mathews, D. H.; Seetin, M. G.; Sagui, C.; Babin,
V.; Kollman, P. A. AMBER 10; University of California: San Francisco,
2008.
(19) Wang, J. C., P.; Kollman, P. A. How well does a restrained
electrostatic potential (RESP) model perform in calculating conforma-
tional energies of organic and biological molecules? J. Comput. Chem.
2000, 21, 1049−1074.
K
dx.doi.org/10.1021/jm3015298 | J. Med. Chem. XXXX, XXX, XXX−XXX