Beyond U0126. Dianion chemistry leading to the rapid synthesis of a series of potent MEK inhibitors

Article abstract of DOI:10.1016/j.bmcl.2004.01.012

Employing phenylmalonitrile dianion chemistry, a large number of analogues of MEK inhibitor lead SH053 (IC₅₀=140 nM) were rapidly synthesized leading to single digit nM inhibitors, displaying submicromolar AP-1 transcription inhibition in COS-7 cells. Compound 41, exhibiting a MEK IC ₅₀=12 nM showed ip activity in a TPA-induced ear edema model with an ED₅₀=5 mg/kg.

Full text of DOI:10.1016/j.bmcl.2004.01.012

Bioorganic & Medicinal Chemistry Letters 14 (2004) 1483–1486
Beyond U0126. Dianion chemistry leading to the rapid synthesis of
a series of potent MEK inhibitors
John Wityak, Frank W. Hobbs, Daniel S. Gardner, Joseph B. Santella, III,
Joseph J. Petraitis, Jung-Hui Sun, Margaret F. Favata, Andrea J. Daulerio,
Kurumi Y. Horiuchi, Robert A. Copeland, Peggy A. Scherle, Bruce D. Jaffe,
James M. Trzaskos, Ronald L. Magolda, George L. Trainor and John V. Duncia*
Bristol-Myers Squibb Pharmaceuticals Research Institute, PO Box 4000, Princeton, NJ 08543-4000, USA
Received 22 September 2003; revised 19 December 2003; accepted 10 January 2004
Abstract—Employing phenylmalonitrile dianion chemistry, a large number of analogues of MEK inhibitor lead SH053 (IC50=140
nM) were rapidly synthesized leading to single digit nM inhibitors, displaying submicromolar AP-1 transcription inhibition in
COS-7 cells. Compound 41, exhibiting a MEK IC50=12 nM showed ip activity in a TPA-induced ear edema model with an
ED50=5 mg/kg.
# 2004 Elsevier Ltd. All rights reserved.
1. Introduction
U0126 readily undergoes cyclization and therefore is
not a stable entity.4 We believed that keeping the mole-
cule symmetric was not important and that most likely,
both vinylogous cyanamides were not required for good
binding affinity. Since the vinylogous cyanamide moiety
is flat, we thought that we could replace it with a simple
benzene ring, leading to molecule SH053 which
exhibited good binding affinity.10,11
Steroids exert their anti-inflammatory effect via inhibition
of the transcription factors AP-1 and NF-kB, both of
which are regulators of the immune response genes.1
Unfortunately, steroids also interact with the gluco-
corticoid response elements (GREs) in gene promoters
resulting in the enhancement of transcription,2 which in
turn leads to undesirable side effects.3 Inhibition of AP-1
and/or NF-kB without interaction with the GREs
would constitute an ideal anti-inflammatory drug.
In a previous letter,4 we disclosed the discovery of
U0126, a compound which functionally antagonizes AP-1
transcriptional activity via upstream inhibition of MEK5
(MAP kinase kinase or MAPKK), a dual specificity
kinase in the mitogen-activated protein kinase (MAPK)
cascade.5 MEK inhibition thus appears to be an attractive
anti-inflammatory6 mechanism to pursue. Recently,
MEK inhibitors have been shown to display anticancer
properties as well.7ꢀ9
We further hypothesized that SH053 is less potent than
U0126 because it lacks U0126s right-most phenyl ring
which most likely binds in a hydrophobic pocket. One
of the first set of compounds to be made to test this
hypothesis involved the addition of a phenoxy group to
SH053 (compounds 1, 2, and 3 in Table 1). Compound
2, to our delight, exhibited an increased affinity for
MEK (IC50=0.04ꢁ0.01 mM; AP-1 IC50=3.03ꢁ0.9
mM). Replacement of the oxygen linker of 2 with a car-
bonyl group leads to decreased MEK inhibition (4).
However, the methylene and CH(OH) linkers yield
compounds of equivalent MEK inhibitory potency (7
and 9 versus 2). N-Methyl amides 10 and 11 were weak
inhibitors. Moving the amino group from the ortho to the
para position weakens potency (4 versus 5; 7 versus 8).
* Corresponding author. Tel.: +1-609-252-3123; fax: +1-609-252-
7569; e-mail: john.duncia@bms.com
0960-894X/$ - see front matter # 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.01.012

1484
J. Wityak et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1483–1486
Table 1. SH053 analogues
No.
X
Y
MEK IC50 (uM)a AP-1 IC50 (uM)
SH053 2-NH2
H
2-O-Ph
3-O-Ph
4-O-Ph
3-CO-Ph
3-CO-Ph
4-CO-Ph
3-CH2-Ph
3-CH2-Ph
0.14ꢁ0.05
0.51ꢁ0.01
0.04ꢁ0.01
0.93 (n=1)
0.41ꢁ0.06
1.81 (n=1)
0.54ꢁ0.20
0.04ꢁ0.01
0.45 (n=1)
0.03ꢁ0.01
4.01 (n=1)
9.20ꢁ4.0
—
1
2
3
4
5
6
7
8
2-NH2
2-NH2
2-NH2
2-NH2
4-NH2
2-NH2
2-NH2
4-NH2
3.03ꢁ0.9
—
6.17
—
3.13
3.2
38.8
9
10
11
2-NH2 3-CH(OH)-Ph
2-NH2 3-CO-N(CH3)-Ph
4-NH2 3-CO-N(CH3)-Ph 45% @ 10 uM
2.07ꢁ0.8
—
—
Synthesis of the CO and CH(OH) linked diaryl
intermediates was rather lengthy, especially when
the appropriate benzophenone was not commercially
available (Scheme 1). Fortunately, we discovered a
novel synthetic sequence involving dianion chemistry
(see Section 2), thus permitting the rapid synthesis of
the benzhydrol analogues summarized in Table 2.
Scheme 1. (a) THF, 10 ꢂC, 97%; (b) MnO2, CH2Cl2, (99% when using
2-pyridinecarboxaldehyde, for example); (c) NBS, benzoyl peroxide,
CCl4, reflux, 70%; (d) Et4NCN, CH2Cl2, reflux, 65%; (e) LDA,
2-ClBnSCN, 42%; (f) 2-aminothiophenol, Et3N, 25 ꢂC, 37%; (g)
NaBH4, MeOH, 0 ꢂC, 57%; (h) 2-aminothiophenol, Et3N, 25 ꢂC, 59%;
(i) TFA, NaBH4 pellets, 25 ꢂC, 51% (CAUTION: do not use
powdered NaBH4—explosion hazard); (j) 2-aminothiophenol, Et3N,
25 ꢂC, 57%.
In Table 2 we find that placement of a polar and/or H-
bonding functionality at the para-position of the Z=Ph
group leads to greater MEK and AP-1 inhibition. Thus,
para-NO2 (30) and para-CN (33) yield molecules that
exhibit MEK IC50’s of 8 and 7 nM, respectively, and
AP-1 IC50’s of 0.3 mM each, the latter being almost 7-
fold better than that of 9, and 30-fold better than that of
SH053. The carbomethoxy group (38) leads to lower
potency compared to the NO2 and CN groups. Of the
pyridine analogues (39–41), the 4-isomer again shows
the greatest affinity. The furans appear to be less potent,
but the thiophenes appear to be equipotent to pyridine
41 (44–47). Enlargement of the Z=Ph group of 9 by
replacement with Z=naphthyl (12, 13) and benzodiox-
anyl (14) leads to loss in potency most likely due to
steric hindrance. Addition of a simple methyl group
to Z=Ph also has the same effect (16, 18, 20). The same
is true for a CF3 group (48, 49). Extension of the Z=Ph
by the insertion of a methylene group decreases potency
(25). Removal of the phenyl group or replacement by
cyclohexyl leads to decreased potency (23 and 22,
respectively). When Y¼H, decreased binding is
observed (26 and 27). Compound 50 was an inter-
mediate to 51, which along with 42, contain groups
which are charged at physiological pH. Both show
extremely weak AP-1 IC50’s most likely due to poor cell
penetration. On the other hand, polar, but neutral
groups such as NO2, CN, CO2Me (30, 33, and 38) all
show good cell penetration with the most potent AP-1
IC50 values in Table 2.
2-OMe (37), and H (35), are all less potent in the 4-CN
series. The same is true for the 4-NH2 group in other
series as well (4 versus 5; 9 versus 15; 16 versus 17; 18
versus 19; 20 versus 21).
Overall, the SAR closely followed that established in the
U0126 series,4 suggesting that these two chemotypes
might bind at the same site on MEK. In binding kinetics
studies with U0126, it was found that the binding to
MEK was noncompetitive with either ERK or ATP.6
To determine the mechanism of MEK inhibition by our
new series of inhibitors, the effects of varying con-
centrations of 41 on the enzyme velocity at varying
substrate concentrations was measured, and the data
analyzed using classical Michaelis–Menten kinetics.12
Noncompetitive inhibition with respect to ERK and
uncompetitive inhibition with respect to ATP was
demonstrated for 41. Hence, 41 displayed significant
affinity for MEK only when ATP was bound to the
enzyme.
The ability of 41 to displace 3H-labeled U0126 from
MEK in the presence and absence of ATP was also
investigated. In the presence of ATP, a dose-dependent
Modifications on the left aromatic ring of the molecule
were investigated as well. The 2-OH group was about as
potent as the 2-NH2 group (compare 36 and 33; 29 and
28) with respect to MEK IC50’s, but 2-fold less potent
for AP-1 inhibition. The other groups, 4-NH2 (34),
3
displacement of H-U0126 by 41 was observed. How-
ever, when ATP was excluded from the buffer, 41 failed
to displace the radioligand, consistent with binding
selectively to the ATP-replete enzyme. The data suggest

J. Wityak et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1483–1486
Table 2. Benzhydrol MEK inhibitors and analogues
1485
No.
X
Y
Z
MEK IC50 (uM)
AP-1 IC50 (uM)
9
2-NH2
2-NH2
2-NH2
2-NH2
4-NH2
2-NH2
4-NH2
2-NH2
4-NH2
2-NH2
4-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-OH
H
H
H
H
H
H
H
H
H
H
H
H
H
CH3
H
Ph
CH3
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Ph
2-Napthyl
1-Napthyl
1,4-Benzo-dioxan-6-yl
Ph
0.034ꢁ0.006
0.076ꢁ0.007
0.047ꢁ0.008
0.025ꢁ0.004
0.570 (n=1)
0.024ꢁ0.008
0.659ꢁ0.008
0.057ꢁ0.036
0.503 (n=1)
0.050ꢁ0.033
0.390ꢁ0.039
0.056ꢁ0.030
0.183ꢁ0.000
0.340 (n=1)
0.170ꢁ0.019
2.220 (n=1)
0.682ꢁ0.200
0.017 (n=1)
0.025ꢁ0.008
0.008ꢁ0.000
0.008ꢁ0.002
0.013ꢁ0.006
0.007ꢁ0.002
0.053ꢁ0.006
0.053 (n=1)
0.008ꢁ0.004
2.081ꢁ0.000
0.014ꢁ0.003
0.099ꢁ0.030
0.025ꢁ0.012
0.012ꢁ0.003
0.330 (n=1)
0.044ꢁ0.005
0.323ꢁ0.110
0.055ꢁ0.009
0.017ꢁ0.003
0.017 (n=1)
0.053ꢁ0.029
0.057ꢁ0.027
0.255ꢁ0.000
0.035ꢁ0.001
2.07ꢁ0.8
3.25ꢁ0.0
2.59ꢁ0.6
2.10ꢁ0.1
10.7
2.69ꢁ1.5
3.67ꢁ1.5
2.4
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
2-CH3–Ph
2-CH3–Ph
3-CH3–Ph
3-CH3–Ph
4-CH3–Ph
4-CH3–Ph
Cyclohexyl
CH3
N/A
3.0
N/A
4.51ꢁ1.1
66.7
40
CH3
Bn
Ph
Ph
19.9
N/A
N/A
1.6
3.22ꢁ0.9
0.30ꢁ0.0
0.30ꢁ0.0
1.08ꢁ0.6
0.32ꢁ0.1
1.13ꢁ0.2
1.2
3-NO2–Ph
3-NO2–Ph
4-NO2–Ph
2,4-di-NO2–Ph
3-CN–Ph
4-CN–Ph
4-CN–Ph
4-CN–Ph
4-CN–Ph
4-CN–Ph
2-NH2
2-NH2
2-NH2
2-NH2
4-NH2
H
2-OH
0.64
N/A
2-OMe
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
2-NH2
4-(COOMe)–Ph
2-Pyridyl
3-Pyridyl
0.29ꢁ0.2
11.8ꢁ2.0
2.72ꢁ0.2
1.40ꢁ0.5
50
4-Pyridyl
4-(N-Me)Pyridinium Iꢀ
Ph-F5
2-Furyl
0.87ꢁ0.2
12.2
3-Furyl
7.02ꢁ1.4
2.22ꢁ0.2
2.4
5.7ꢁ0.0
3.97ꢁ0.3
6.88
Thiophen-2-yl
Thiophen-3-yl
3-CF3–Ph
4-CF3–Ph
4-[N2-Ph3C-Tetrazol-5-yl]Ph
4-[Tetrazol-5-yl]Ph N(CH2CH2OH)3
25
that there is a common binding site for U0126 and 41,
however, the accessibility of this binding site appears to
be modulated by the binding of ATP. We may infer
therefore that the vinylogous cyanamide portions of 41
and U0126 most likely overlap and bind in the same
site. The pyridyl portion of 41, however, probably binds
in a different site from U0126 which is accessible only in
the presence of ATP.
41 is selective for MEK, displaying IC50 values of >1
mM for the related MAP kinases MKK3 and MKK4.13
Due to its superior solubility properties, pyridine 41 was
tested for in vivo anti-inflammatory activity against
phorbol ester (TPA) mediated ear edema in the
mouse.14 When administered intraperitoneally as a
solution in 1:1 PEG:EtOH, the ED50 was 5 mg/kg.
For U0126, ATP binding had a minimal effect on its
affinity for MEK. In contrast, 41 had a significantly
greater affinity for MEK when ATP was bound to the
enzyme. Conformational changes are thought to med-
iate the phosphotransfer activity of MEK and other
members of the MAP kinase family. The present data
suggest that ATP mediated conformational changes of
the protein may also play a significant role in the bind-
ing of certain inhibitors as well, such as 41. Compound
2. Chemistry
Initial lead compounds were synthesized by the methods
outlined in Scheme 1. The key cyanation step (e) was done
by the method of Cava.15 Ethers 1–3 were synthesized in a
analogous fashion from the commercially available phen-
oxybenzyl cyanides. Amides 10 and 11 were made similarly
from the corresponding N-methyl-N-phenyltoluamides.

1486
J. Wityak et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1483–1486
and benzodioxanyl replacements (13 and 14) lowered
affinity. The N-methyl amides (10 and 11) being 2 atom
linkers could also be less potent due to size limitations
imposed by this terminal aromatic pocket. Cell pene-
tration is required for MEK inhibition and thus charged
compounds display poor AP-1 transcription inhibition.
With the replacement of a vinylogous cyanamide group
of U0126 with a benzene ring, we have increased both
chemical stability and potency. Some of the compounds
described in Tables 1 and 2 have been found to be stable
in aqueous buffer at pH=1 whereas U0126 is not. Further
work will be aimed at replacing the remaining cyanamide
moiety. The in vitro results translated into in vivo anti-
inflammatory activity seen in the TPA mediated ear
edema mouse model.
Scheme 2.
References and notes
1. Cato, A.; Wade, E. BioEssays 1996, 18, 371.
2. (a) Gehring, U. Biochem. Sci. 1987, 12, 399. (b) Fuller,
P. J. FASEB J. 1991, 5, 3092. (c) Mangelsdorf, D. J.;
Thummer, C.; Beato, M.; Herrlich, P.; Schutz, G.; Umesono,
K.; Blumberg, B.; Kastner, P.; Mark, M.; Chambon, P.;
Evans, R. M. Cell 1995, 83, 835.
3. Avery, M. A.; Woolfrey, J. R. In Burger’s Medicinal
Chemistry and Drug Discovery, 5th ed.; Wolf, M. E., Ed.;
John Wiley and Sons, Inc.: New York, 1997; p 286.
4. Duncia, J. V.et al. Bioorg. Med. Chem. Lett. 1998, 8, 2839.
5. Herrlich, P.; Ponta, H. Trends Endocrinol. Metab. 1994, 5,
341.
6. Favata, M. F., et al. J. Biol. Chem. 1998, 273, 18623.
7. MEK inhibition in Ras & Raf transformed cells: Dudley,
D. T.; Pang, L.; Decker, S. J.; Bridges, A. J.; Saltiel, A. R.
Proc. Natl. Acad. Sci. 1995, 92, 7686.
8. MEK inhibition in colon tumors: Sebolt-Leopold, J. S.;
Dudley, D. T.; Herrera, R.; Van Becelaere, K.; Wiland,
A.; Gowan, R. C.; Tecle, H.; Barrett, S. D.; Bridges, A.;
Przybranowski, S.; Leopold, W. R.; Saltiel, A. R. Nature
Medicine 1999, 5, 810.
9. MEK inhibition in tumor cell lines: Berger, D.; Dutia, M.;
Powell, D.; Wu, B.; Wissner, A.; Boschelli, D. H.; Floyd,
M. B.; Zhang, N.; Torres, N.; Levin, J.; Du, X.; Wojcie-
chowicz, D.; Discafani, C.; Kohler, C.; Kim, S. C.; Feld-
berg, L. R.; Collins, K.; Mallon, R. Bioorg. Med. Chem.
Lett. 2003, 13, 3031 and references therein.
10. The compounds in this letter were screened using a
constitutively active form of MEK: Mansour, S. J.; Mat-
ten, W. T.; Hermann, A. S.; Candia, J. M.; Rong, S.;
Fukasawa, K.; Vande Woude, G. F.; Ahn, N. G. Science
1994, 265, 966.
Scheme 3. Dianion chemistry. (a) LDA, 0 ꢂC to 25 ꢂC, benzene, 80%;
(b) 2 equiv n-BuLi, ꢀ70 ꢂC, THF, 4-pyridinecarboxaldehyde, 54%; (c)
2-aminothiophenol, Et3N, THF, 25 ꢂC, 62%.
In an attempt to replace the sulfur atom of SH053 with
a carbon, we initially tried to add benzyl Grignard to
phenylmalononitrile (Scheme 2). Even with 7 equiv of
Grignard reagent, starting material phenylmalononitrile
was obtained unchanged. This meant that the mal-
ononitrile anion protects both nitriles from nucleophilic
attack.
Using this ‘anionic protection’ we found that we could
generate a dianion of phenylmalononitrile (Scheme 3)
and quench it with a wide variety of aldehydes and
ketones. Subsequent reaction with aryl thiols led to a
library of CH(OH) and CR(OH) linker analogues,
which are summarized in Table 2. Thus, Cava cyanation
leads to malononitrile 62 (Scheme 3). Deprotonation of
the malononitrile proton followed by halogen–metal
exchange at ꢀ70 ꢂC with two equivalents of n-BuLi
yields dianion 63 as an orange slurry. Immediate
quenching with, for example, 4-pyridine-carbox-
aldehyde yields 64. Vinylogous cyanamide formation
yields compound 41. The dianion forming reaction was
scaled up to 0.15 moles for the synthesis of 41.
11. The compounds in this letter were assayed for potency
against AP-1 transcription activity in COS-7 cells as
delineated in reference 6.
12. Copeland, R. A. Enzymes: A Practical Introduction to
Structure, Mechanism and Data Analysis; VCH/Wiley:
New York, 1996; pp 187–223.
13. For MKK3 and MKK4, coupled assays were used as
described in reference 6.
14. TPA treatment activates the Ras pathway; the TPA ear
edema assay was performed as described in: Jaffee, B. D.,
et al. Biochemical and Biophysical Research Communi-
cations 2000, 268, 647.
3. Conclusion
We found that one of the vinylogous cyanamides of
U0126 could be effectively replaced by a benzene ring,
leading to the discovery of SH053. Attachment of an
additional phenyl ring with a para-H-bonding sub-
stituent increases the potency by 1 order of magnitude
over that of SH053. The binding pocket for this addi-
tional aromatic ring is limited in size, since naphthalene
15. Davis, W. A.; Cava, M. P. J. Org. Chem. 1983, 48, 2774.