Structure activity relationships of anthranilic acid-based compounds on cellular and in vivo mitogen activated protein kinase-5 signaling pathways

Full text of DOI:10.1016/j.bmcl.2018.05.029

Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Structure activity relationships of anthranilic acid-based compounds on
cellular and in vivo mitogen activated protein kinase-5 signaling pathways
Suravi Chakrabartya,1, Darlene A. Monlishb,1, Mohit Guptaa, Thomas D. Wrightb, Van T. Hoangc,
Mitchel Fedakd, Ishveen Chopraa, Patrick T. Flahertya,⁎, Jeffry Madurad, Shankar Mannepellid,
Matthew E. Burowc, Jane E. Cavanaughb
a
Department of Medicinal Chemistry, School of Pharmacy, Duquesne University, 410A Mellon Hall, 600 Forbes Ave., Pittsburgh, PA 15282, United States
Department of Pharmacology, School of Pharmacy, Duquesne University, 410A Mellon Hall, 600 Forbes Ave., Pittsburgh, PA 15282, United States
Department of Medicine-Section of Hematology and Medical Oncology, Tulane University, New Orleans, LA 70118, United States
Department of Chemistry, Duquesne University, 410A Mellon Hall, 600 Forbes Ave., Pittsburgh, PA 15282, United States
b
c
d
Cancer comprises of a group of heterogeneous diseases; increasingly
it is recognized that upregulation of mitogenic pathways occur early
during the process of disease progression. The mitogen-activated pro-
tein kinase (MAPK) signaling cascade is a complex web of signaling
events that mediate appropriate intracellular responses to external
mitogenic events.1 This highly branched signaling cascade possesses a
singular parallel processing stage where each ERK (extracellular signal-
regulated kinase) is phosphorylated uniquely by its corresponding MEK
(MAPK/ERK kinase). Enzymatically, MEKs belong to the serine/threo-
nine kinase2 family. A recent metabolomics study suggests that the
biochemical outcome resulting from this three-tiered kinase cascade
strategy is a more rapid yet adaptable cellular response. Genomic
analysis indicates that this 3-teired kinase cascade strategy3,4 is care-
fully conserved across species and represents a signaling strategy of
sustained biological relevance. Cross-talk between the pathways is
guided by a variety of mechanisms including controlled sub-cellular
localization of kinase isoforms, activation by enzyme-catalyzed phos-
phorylation events, and protein-protein association directly between
the kinases as well as associations involving platform proteins as well.
In a recent review of the MAPK pathways,5 careful analyses was
directed at the MEK1/ERK1 and MEK2/ERK2 pathway. The role of
other MEK pathways, specifically the MEK5/ERK5 pathway remains
less than thoroughly understood due to the lack of small molecule
probes.5
higher selectivity.6 Additionally, the capacity for ERK5 to exist either as
cytosolic or nuclear protein and to have multiple post-translational
phosphorylation states (pT219, pY221, pS720, and pT733)11 compli-
cates the understanding of the effects of the MEK5/ERK5 cascade by
targeting downstream components such as ERK5. For example, phos-
phorylation of ERK5 at THR732 is associated with translocation to the
nucleus.12,13 This is not a MEK5 mediated event: MEK5 phosphorylates
Thr21912 of the TEY sequence (219-221).
To date, ERK5 is the only known substrate for MEK5; therefore,
targeting this unique pathway at MEK5 may minimize side effects and
reduce toxicity by sparing other MAP kinase events critical for the
homeostasis and survival of healthy cells. The MEK5/ERK5 pathway
mediates effects of several oncogenes,7 including Ras and Src but fur-
ther understanding of this cascade is required.14–17 The MEK5/ERK5
pathway is constitutively active in numerous cancer types.1 MEK5 is
overexpressed in over 50% of tumors analyzed and ERK5 is elevated in
over 20% of patients. Significantly, elevated ERK5 expression reduces
disease-free survival time of breast cancer patients compared to patients
with lower expression levels of this kinase.18,19 The development of
small molecule modifiers of the MEK5/ERK5 MAPK pathway is neces-
sary to understand the MEK5/ERK5 pathways in various cell types and
to explore potential drug intervention strategies.
Development of inhibitors of the MEK1/2 isoform employed X-ray
crystal based compound design preceding from the discovery that ac-
tive substituted N,N-diphenylanilines do not bind at the ATP site, but
instead bind to a hydrophobic site proximal to the ATP-associated
Mg2+ ion site near the C helix.20,21 This site has been exploited ex-
tensively in the development of MEK1/2 inhibitors and has generated
two recently FDA-approved drugs: mekinist® (trametinib) and cotellic®
(cobimetinib). This site is now termed the type III inhibitor binding
site2,22 and its presence and relevance to drug design has been con-
firmed in other protein kinases.23 Ligand binding to the type III site
A compelling case has been presented for the role of MEK5 in var-
ious cancers.6–8 Major techniques to analyze contributions of each
MEK/ERK pathway entail a combination of small molecule probes as
well as gene manipulation strategies. These approaches have limita-
tions. Gene manipulations are often cell-line specific. The best-char-
acterized inhibitor of the MEK5/ERK5 pathway is XMD8-92.9–11 This
compound is an ATP-site inhibitor of ERK5. Interception at earlier step
in the kinase cascade could potentially offer a greater impact and
⁎
Corresponding author.
E-mail address: flahertyp@duq.edu (P.T. Flaherty).
These authors contributed equally to this work.
1
https://doi.org/10.1016/j.bmcl.2018.05.029
Received 14 December 2017; Received in revised form 11 May 2018; Accepted 12 May 2018
0960-894X/©2018ElsevierLtd.Allrightsreserved.
Pleasecitethisarticleas:Chakrabarty,S.,Bioorganic&MedicinalChemistryLetters(2018),https://doi.org/10.1016/j.bmcl.2018.05.029

S. Chakrabarty et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Fig. 2. Phylogram analysis of MEK isoforms; MEK5 is most closely related to
MEK1 and MEK2.
Fig. 1. Schematic of halogen bond from the 4′-iodo atom 1 (PD325901) to the
lone pair of the acyl of Val in 3EQC.24
prevents phosphorylation of a given MEK by the associated upstream
MEKK (MEK-kinase) and consequently prevents activation of a given
MEK. Considerable effort has been directed toward developing type III
MEK1/2 inhibitors. For all MEK1/2 inhibitors in the literature there is a
rigorous requirement for a halogen substitution on the terminal arene at
the para position. The 2-F, 4-I arene substitution pattern exists in the
original compound 1, PD325901 found in the X-ray crystal structure21
of MEK 1 (PDB ID: 3EQC), as well as for all FDA-approved MEK1/2
inhibitors (trametinib and cobimetinib). In most pre-clinical SAR stu-
dies the 4-iodo moiety is stated as essential for MEK1/2 activity.21 The
3.11 Å distance from the lone pair of electrons from the backbone
amide acyl of Val127 to the iodine atom in the X-ray structure 3EQC24
displayed in Fig. 1 is consistent with a halogen bond.25–29
A 4′-iodine atom on the terminal arene is essential for MEK1/2
activity. Additional examination of 3EQC indicates that the iodine atom
fills a large hydrophobic pocket. This hydrophobic pocket is non-
identical between MEK1 and the pocket proposed by the homology
model of MEK5 (see below). The central arene ring is tolerant of wide
variability for MEK1 inhibitors. Extension of the central arene ring to a
benzo-fused heterocycle has also been explored.30 These heterocyclic
ring variations have displayed a capacity to act as a hydrogen bond
partners with Ser212 on the C-helix. Side-chain variations from the
central benzamide have successfully been explored to yield potent in-
hibitors of MEK130 with an emphasis on designing better PK properties
than present in 1. An analog design approach informed by both existing
MEK1 SAR and structure-informed design based on multiple MEK5
homology models explored substitution patterns on the arene rings and
the amide sidechain with the goal of optimizing MEK5 potency and
selectivity.
There has been significant analysis of the structure of MEK1; there
are 39 X-ray crystal structures of MEK1 with resolutions ranging be-
tween 2.93 Å and 1.8 Å with 50% obtained at 2.5 Å resolution or better.
There is less known about the other isoforms. MEK2 is represented by 1
structure. MEK6 is represented by 3 structures containing the kinase
domain. MEK7 is represented by 5 structures containing the kinase
domain. The kinases MEK3, MEK4, and MEK5 do not have any X-ray
crystal structures with kinase domains deposited in the PDB. There are
three structures for wild type and mutant N-terminal PB1 domains of
MEK5 deposited (residues 5–10 and 16–130)31, but none contain the
kinase domain found in residues 166–409. The MEK5 structures de-
posited examined trimeric protein-protein interactions between MEK5
fragments, ERK5, and MEKK3. A sequence homology tree, shown in
Fig. 2, identified MEK5 as being most similar to MEK1 and MEK2 when
compared to all possible MEK isoforms.31
Fig. 3. ATP and Type III binding sites of the MEK1 crystal structure (PDB ID:
3EQC) superimposed on a homology model of MEK5. ATP is shown as space
filling and 1 is shown as a stick representation in the proposed type III binding
site.
Multiple models representing the DFG-in and the DFG-out conforma-
tions, as well of the various phosphorylated forms were examined. Both
Accelrys Discovery Studio and MOE 2009.10 were used to examine the
proteins and ligand interactions. Docking was conducted with side-
chain flexibility for residues lining and adjacent to the proposed type III
binding domain (See Fig. 3).
The N,N-diphenylaniline moiety was selected as the core structure
to be elaborated because it exists in the PD325901 structure, provides
rapid synthesis of analogs, and permits structural variation at locations
that interact with the proposed type III pocket. The modifications ex-
amined are presented in Fig. 4.
Desired biological consequences of structural modification include
both PD and PK effects. Specifically, providing either ionization or in-
creased hydrogen bonding on the side chain lowers the cLogP and
provides the potential to interact the more polar lip of the P-loop. This
region is normally associated with ATP triphosphate transfer and is the
same location were the glycerol portion of 1 binds to MEK1. This region
in the MEK5 homology models also displays a very polar domain with
subtle variations that could offer differential affinity for MEK5 versus
MEK1 or MEK2. Specifically Asp283 of the MEK5 HRD sequence is
proposed to correlate with Asp 190 of the HRD sequence of MEK1.
Although these regions are close in the overlay of the two models, they
differ by at least 2 Å and form the bottom of the ATP binding site ad-
jacent to the Type III binding pocket. Also, the C-helices near the ATP
binding site are non-identical; the MEK5 Tyr315 residue of the YVGT
Several homology models were constructed on the structure of
MEK1 (3EQC).32 The bound ligand is a type III inhibitor. The canonical
beta isoform33–35 of MEK5 (Uniprot: Q13163) with 448 amino acid
residues was chosen for sequence to be used for the homology model.
2

S. Chakrabarty et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Fig. 4. Side chain variation and amide variations and rationale.
Fig. 5. Proposed side chain variations.
corresponds to the MEK1 Phe223 residue of the FVGT sequence.
Variations of basic amine placement and alteration of flexibility were
proposed to explore this area. Given the flexibility of the P-loop geo-
metry relative to the lower α-helical domain, multiple variations were
examined as shown in Fig. 4. Sidechain variations examined also in-
cluded synthetic precursors (amides, acids, tert-butyl carbamates) and
characterized side-products (See Fig. 5).
Prior literature36 indicated that the 3 and 4 fluorine atoms of the
central arene ring, present on the lead compound PD325901, func-
tioned as more than just electron-withdrawing groups. It has been
proposed that the 4-fluorine atom, can form an electrostatic or hy-
drogen/halogen bond interaction with the Ser212 of the C-helix in the
MEK1 structure. With an eye toward eventual replacement of the cen-
tral arene, an initial survey of the requirement and utility of 3,4-di-
fluorine substitution on the central arene for MEK5 inhibition was
conducted and is presented in Fig. 6.
Terminal arene variations were prepared to survey the need and
utility of fluorine and iodine atom substitution for MEK5 inhibition. Of
particular note, the 4′-iodine atom presented multiple sub-ideal che-
mical features: high molecular weight (126 Da), large hydrophobic
area, and potential photo-instability. Despite these disadvantages, the
4′-iodophenyl moiety is scrupulously retained in all compounds in ad-
vanced MEK1 inhibitors and both FDA approved MEK1/2 inhibitors.
Point variations in Fig. 7 were examined with the goals of decreasing
MEK1/2 activity and increasing MEK5 activity.
Internal hydrogen bonding from the aniline proton to the acyl
oxygen lone pair of the amide is predicted by deposited X-ray crystal
structures of MEK1/2 type III inhibitors. These structures display an
unvarying presence of secondary amines at this position. The geometry
of the heavy atoms identified in the deposited X-ray crystal structures is
Fig. 6. Central unsubstituted aryl derivatives and associated right aryl and side-
chain variants.
consistent with internal hydrogen bonding resulting in a 6-membered
ring as shown in Fig. 8. It is unclear if this internal bond from the
aniline NH to the acyl lone pair is necessary or useful for MEK5 in-
hibition. The N-methyl-N,N-diphenylaniline analogs, compounds
21–24 in Fig. 9 are unable to participate in internal H-bonding and
were prepared to examine the contribution of internal hydrogen
bonding toward MEK1 and MEK5 activity.
The synthesis of these compounds proceeded by initial formation of
the N,N-diphenylaniline followed by subsequent conversion of the
3

S. Chakrabarty et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Fig. 9. N-Methyl derivatives designed to explore the necessity of internal H-
bonding.
Preferential substitution occurs at the ortho-position as identified in
the literature. Davis et al.45 proposed the lithium carboxylate of 25
coordinates with and directs the lithium salt of 26 towards the ortho-
position of the carboxylate to facilitate nucleophilic attack at the 2-
carbon bearing a fluorine shown as 27 in Scheme 1. The 2-carbon of
compound 25 is the most electron deficient carbon atom due to the
attached fluorine atom, an adjacent fluorine atom, and most sig-
nificantly the ortho carboxylate. Inductive effects are decreased by the
inverse square of distance encouraging selective nucleophilic attack at
the 2-carbon. This method has been used previously to prepare com-
pounds that were co-crystallized in the structure of MEK1 confirming
the regioselectivity of the addition. The 1H NMR and physical proper-
ties for the current synthesis were identical with the literature.45 This
general approach was subsequently explored for ring variations as de-
scribed below.
The acid, 2, was readily converted into the corresponding acid
chloride in 2 h at ambient temperature by oxalyl chloride and catalytic
DMF in dichloromethane as shown in Scheme 1. The crude acid
chloride, 28, was added directly to an excess of the appropriate amine.
Synthesis of the desired unsubstituted central arene benzoic acids
32–33 were not formed by SNAr chemistry from the lithium salt of the
aniline due to absence of fluorine atoms and resultant decrease in
electrophilicity of the 2-carbon of the benzoate. An Ullmann coupling
between 2 and iodobenzoic acid, 29, and amines 26, 30, and 31 gave
the desired acids 32–34 in acceptable isolated yield (typically ∼ 50%).
Recent Ullmann coupling variations47–51 recommended conditions
shown in Scheme 2. The best conditions used were copper iodide, po-
tassium carbonate, and a mixed solvent system of 9 to 1 DMF to water
with microwave irradiation.
Conversion of the benzoic acids 32–34 to the corresponding amides
was achieved with oxalyl chloride/dimethyl formamide route as pre-
viously described followed by excess amine to give the desired amides
12 and 13. Compound 14 could not be prepared efficiently by the acid
chloride method. Use of standard DIC (diisopropylcarbodiimide) cou-
pling gave 14 in acceptable yield. A rationalization for the failure of the
acid chloride strategy is probable self-condensation; the absence of
halogens in compound 34 renders the nitrogen atom less electron de-
ficient than previous N,N-diphenylanilines examined.
Fig. 7. Right aryl derivatives and associated side-chain variants.
benzoic acid to the desired amide. N,N-Diphenylaniline are typically
synthesized with a copper-catalyzed Ullmann coupling37–44 or in the
case of a 3,4-difluoro substitution pattern on the central ring, they can
be prepared by directed nucleophilic attack of the lithium salt of the
terminal aniline. This attack into the strongly electron deficient central
arene ring was explored by Davis45 and found to proceed well with
analogous strongly electron-deficient rings. For the central arene
without halogens, an Ullmann coupling was successfully employed.
Conversion of the benzoic acid to the desired amide could generally be
achieved by routine conversion to the intermediate acyl chloride with
oxalyl chloride/DMF then use of excess amine. This proceeded well
when the arenes contained halogens, but in the case of unsubstituted
arenes, this strategy failed. This was attributed to self-condensation of
the less-electron deficient anilines with the acyl chloride. In such cases,
lower temperature carbodiimide couplings46 were successfully used.
Synthesis of the 3,4-difluoro-N,N-diphenylaniline core, 2, at a multi-
gram scale was previously achieved using the lithium salt of the
terminal aniline displacement shown in Scheme 1 and developed by
Davis et al.45
Desired compounds retaining the central 3,4-difluoro-substitution
pattern with variations on the terminal arene as well as the side chain
variants were effectively prepared by SNAr of an appropriately sub-
stituted lithium amide. Fewer electron-withdrawing groups on the
terminal aniline did raise the pKa, but lithium amide was an adequate
Fig. 8. Internal hydrogen bonding results in a more conformationally restricted analog (compound 2, left); the N-methyl in compound 21 prevents internal hydrogen
bonding and results in a less conformationally restricted structure.
4

S. Chakrabarty et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Scheme 1. Synthesis of amides using the acid chloride method from the lithium salt of aniline 26.45
Scheme 2. Synthesis of central unsubstituted aryl derivatives and associated right aryl and side-chain variants.
5

S. Chakrabarty et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Scheme 3. Synthesis of right aryl derivatives and associated side-chain variants.
base to abstract the aniline proton and direct an ortho SNAr attack on
25.
pERK5: 88.4 kDa) permits efficient analysis of inhibition of the various
MEK/ERK cascades simultaneously. More rigorous confirmation of en-
gagement at the level of MEK and more significantly as type III in-
hibitors will require additional studies on compounds identified in
these initial screens. The compounds were initially examined in tripli-
cate at a single concentration of 10 μM on EGF-stimulated triple-nega-
tive breast cancer MDA-MB-231 cells. Controls included DMSO as the
blank representing 100% activity, and compound activity is represented
as percent decrease in activity normalized to the DMSO blank. EGF
stimulates both the MEK1/2 cascades as well as the MEK5 cascade and
acted as a control to confirm intact signaling cascades. Standard com-
pounds included U0126, a compound known to inhibit MEK1 and other
kinases not including MEK5. U0126 is not competitive with ATP al-
though the precise site of U0126 inhibition remains unconfirmed.
XMD8-92 was originally prepared the Gray lab10 and is an ATP-site
inhibitor of ERK5 and used as the standard for inhibition of the MEK5/
ERK5 cascade. Results are presented in Table 1.
Compounds 2–11 represent point variations of the sidechain with
the rest of the structure unaltered. With the exception of the diethyl
amide, 5, and to a lesser extent the ethylene-lined tertiary amine, 9, all
showed good activity at inhibiting the product of MEK1/2: pERK1/2. Of
the side chain variations very few showed appreciable activity at MEK5.
Compounds 5 and 11 display a slight activation of MEK5 activity above
100 activity (0% inhibition) by a repeatable, yet not fully understood
mechanism. Overall, this does indicate that MEK1/2 versus MEK5 se-
lectivity may be possibly achieved by structural variation in this do-
main, the path toward increasing MEK5 activity selectively is not ob-
vious with compounds 2–11 alone.
Subsequent amide formation was achieved either by conversion to
the acyl chloride then addition of the amine to yield the amide in the
case of 17 where ammonia was used, or were prepared with carbodii-
mide couplings in the case of 18–20. The use of EDC facilitated product
isolation (See Scheme 3).
Attempts to effect conversion from the N,N-diphenylaniline 2 or 3 to
the corresponding tertiary aniline 22 via an Eschweiler-Clarke conver-
sion failed. Alkylation of the N,N-diphenylaniline with methyl iodide
and sodium hydride in DMF was also unsuccessful. Reductive alkylation
of the aniline precursor, 2-fluoro-4-iodoaniline (26), preceded in ac-
ceptable yield to give the mono N-methylated product, 35, shown in
Scheme 4.52,53 This secondary aniline was then added into 2,3,4-tri-
fluorobenzoic acid (25) using the standard SNAr lithium amide dis-
placement approach of Davis.45 Conversion to the desired final com-
pounds followed the previously described acyl chloride route (See
Scheme 5).
The use of intact cells rather than isolated enzymes was selected as
the primary screen for several reasons. First, native signaling cascades
require multiple kinase events initiated by an external ligand binding
event; this initial event can be reproduced reliably in vitro with the use
of EGF. Notably triple-negative breast cancer cells can be used directly
and may be predictive of activity in actual tumors. Second, MEK1/2 or
MEK5 must be activated by dual phosphorylation to its active form to
be catalytically active. This would require either in vitro conversion of
the wild-type MEK to the active form by a second enzyme (MEKK) or
the use of a constitutively active MEK5βDD mutant. Although affinity
for the isolated enzyme is potentially useful information, the unavail-
ability of isolated MEK5 protein of either variety (MEK5 or MEK5DD),
but more significantly extensive protein-protein interactions31 of MEK5
with ERK5 recommend an isolated enzyme may not perfectly represent
in vivo conditions. Consequently, the cellular assay is proposed as better
representing actual physiological conditions. The significant molecular
weight difference of the final antibody-identifiable phospho proteins
pERK1, pERK2, and pERK5 (MW: pERK1: 43.4 kDa: pERK2: 41.4 kDa;
For the non-halogen central ring variations, all were considerably
less active both at MEK1/2 and MEK5. The 4-N-methyl piperazine de-
rivative, 14, although not a direct comparison of 11, did have the best
activity at inhibition of MEK5 and did have the best selectivity for
MEK5 versus MEK1/2 inhibition for the compounds examined so far. If
any conclusion can be draw from compounds 12–14 it would be that
MEK5 can tolerate a less polarized central arene better than MEK1/2.
Terminal arene variations 12–20 explored both terminal arene
Scheme 4. Monomethylation reaction of 26 then synthesis of acid 21 by lithium amide displacement method.
6

S. Chakrabarty et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Scheme 5. Synthesis of side chain variants of N-Methyl derivatives.
variations and side chain variations with the central ring held constant
with a 3,4-difluorophenyl substitution pattern. Activity was variable
across these series. When the carboxylate side chain compounds 2, 15,
and 16 are compared, the point omission of the 4-iodo generates
compounds that are half as active at inhibiting the MEK1/2 cascade and
are devoid of MEK5 inhibition. The amides present a more complicated
profile when the terminal arene is modified. For the primary amides 3,
17, and 18, MEK1/2 activity is only moderately reduced with the
omission of the 4-iodo. Removal of both the 2-fluoro atom and the 4-
iodo atom generates much weaker MEK1/2 inhibitors, approximately
one third of the fully adorned lead, compound 3. For the primary
amides, MEK5 activity is weak across the board. The piperazine of the
terminal arene series 10, 11, 19, and 20 identify compounds 20 as the
most potent and most selective at MEK5. The high activity at MEK5 of
the fully dehalogenated terminal arene 20 bearing a cationic charge
contrasts to the neutral primary amide series and the anionic carboxylic
series with the equivalent terminal arene substitution pattern. This
suggests that the cationic piperazine, in the case of MEK5 inhibitors,
may contribute more than halogen interactions of the terminal ring.
Preventing internal hydrogen bonding by replacing the aniline N-H
with an aniline N-methyl has differential effects on MEK1/2 versus
MEK5 activity. Comparison of 2 with 21 indicates that if the carbox-
ylate is prevented from internal hydrogen/ionic bonding the compound
is much more potent at inhibiting of MEK1/2 than MEK5. The relatively
Table 1
MEK1/2 and MEK inhibition data.
Compound
Side chain (R1R2X)
Central ring (R3)
Terminal ring (R4)
Nitrogen substitution (R5)
Decrease in pERK1/2
Decrease in pERK5
2
3
4
5
6
7
8
9
HO
H2N
Me2N
Et2N
MeHN
MeO
EtHN
Me2N(C2H4)MeN
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
–
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
2-F
H
H
H
H
H
H
H
H
98.5
96.8
87.3
5.5
99.6
98.9
98.6
27.9
70.9
99
20.1
59
0.2
0*
20.4
13
8.5
9.4
8.4
0*
10
11
12
BOC-Piperazine
Piperazine
H2N
H
H
H
0*
39
13
14
15
16
17
18
19
20
21
22
23
24
4-methylpiperazine
4-methylpiperazine
HO
HO
H2N
H2N
4-methylpiperazine
4-methylpiperazine
HO
H2N
–
–
2-F,4-I
–
2-F
–
2-F
–
2-F
–
2-F,4-I
2-F,4-I
2-F,4-I
2-F,4-I
–
H
H
H
H
H
H
H
H
Me
Me
Me
Me
33.5
29.3
31.7
43.5
80.2
35.3
37.1
0*
67.1
73
50.9
0*
30.9
71
0*
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
3-F,4-F
–
0*
24.5
14.8
56.1
82.4
18.4
0*
91.6
27.2
0
MeO
4-methylpiperazine
–
Blank
PD 0325901
U0126
XMD8-92
0
99.7
99.6
50.1
95.9
43
95.2
0* = moderate (∼20%) increase of pERK identified.
7

S. Chakrabarty et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
exemplified by compound 20 (cLogP = 3.6). The differential activity of
tertiary aniline compounds are also consistent with diverging activity
profiles for sidechains with a positive versus a negative charge. The
positioning of a cationic amine in a favorable orientation (by internal
hydrogen bonding) appears more significant than halogen bonding
from the terminal arene. This raises the possibility that the pharma-
cokinetically unfavorable 4-iodine atom may be dispensable for MEK5
activity in rotationally-restricted compounds with a cationic side chain.
In contrast, freely rotating N,N-diphenylaniline carboxylates seem to
favor MEK5 selectivity but this will require additional validation.
A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.bmcl.2018.05.029.
References
Fig. 10. Compound 3 (SC-1-151) inhibits MDA-MB-231 tumorigenesis in vivo.
1. Drew BA, Burow ME, Beckman BS. BBA. 2012;1825:37.
2. Endicott JA, Noble MEM, Johnson LN. Annu Rev Biochem. 2012;81:587.
3. Tian T, Harding A. Cell Cycle. 2014;13:2379.
4. Harding A, Tian T, Westbury E, et al. Curr Biol. 2005;15:869.
5. Kim EK, Choi E-J. Arch Toxicol. 2015;89:867.
high activity of the ester 23 requires additional examination to un-
derstand increased activity of 23 selectively for MEK5 when contrasted
to the activity of 21. Given similar MEK1/2 activity between 12 and 23,
it is unlikely to result from differential transport. Piperazine 24 com-
pared to piperazine 11 was much weaker for MEK1/2. The MEK5 ac-
tivity was less for compound 24 than 20. This remarkable loss of ac-
tivity could arise from a lack of directionality of the cation when
internal hydrogen bonding is prevented by the aniline methyl.
Piperidine derivatives with an internal hydrogen bond (compounds 20,
14, and 11) may represent a very favorable placement of the cationic
nitrogen. When compound 20 was overlaid on the ligand of 3EQC or on
the proposed ligand binding site of the MEK5 homology models, the
basic amine of the piperazine rotated through a region of space that was
4 Angstroms within the aspartic acid carboxylates of the HRD sequences
of MEK1 and MEK5, Asp190 and Asp283 respectively. Additionally the
N-methylpiperazine when directed by internal H-bonding and the
pyamidalized amide nitrogen can exist within H-bonding interaction (3
Å) of the Tyr315 reside of the YVGT of MEK5 or the corresponding
Phe223 of MEK1.
6. Simoes AES, Rodrigues CMP, Borralho PM. Drug Discov Today. 2016;21:1654.
7. Giurisato E, Tournier C, Giurisato E. Cell Cycle. 2016;15:619.
8. Hoang VT, Yan TJ, Cavanaugh JE. Cancer Lett (NY., NY, US). 2017;392:51.
9. Anon. Cancer Discovery 2014, 4, 1363.
10. Deng X, Yang Q, Kwiatkowski N, et al. ACS Med Chem Lett. 2011;2:195.
11. Wang J, Li Q, Wang C, et al. Biochem Pharmacol (Amsterdam, Neth.). 2016;99:132.
12. Honda T, Obara Y, Yamauchi A, et al. PLoS ONE. 2015;10 e0117914/1.
13. Raviv Z, Kalie E, Seger R. J Cell Sci. 2004;117:1773.
14. English JM, Pearson G, Hockenberry T, et al. J Biol Chem. 1999;274:31588.
15. Kamakura S, Moriguchi T, Nishida E. J Biol Chem. 1999;274:26563.
16. Barros JC, Marshall CJ. J Cell Sci. 2005;118:1663.
17. Lochhead PA, Gilley R, Cook SJ. Biochem Soc Trans. 2012;40:251.
18. Hsieh FC, Cheng G, Lin J. Biochem Biophys Res Commun. 2005;335:292.
19. Montero JC, Ocana A, et al. PLoS ONE. 2009;4:e5565.
20. Ohren JF, Chen H, Pavlovsky A, et al. Nat Struct Mol Biol. 2005;12:278.
21. Sebolt-Leopold JS, English JM. Nature (London, UK). 2006;441:457.
22. Gavrin LK, Saiah E. MedChemComm. 2013;4:41.
23. Palmieri L, Rastelli G. Drug Discovery Today. 2013;18:407.
24. Tatake RJ, O'Neill MM, Kennedy CA, et al. Biochem Biophys Res Commun.
2008;377:120.
25. Kolar MH, Hobza P. Chem Rev (Washington, DC, US.). 2016;116:5155.
26. Ford MC, Ho PS. J Med Chem. 2016;59:1655.
Testing MEK inhibitors for efficacy in a tumor xenograft was es-
sential to confirm any validity to the idea that MEK and perhaps MEK5
may be relevant for actual tumors. Compound 3 was selected as it was
one of the earliest compounds identified to display significant MEK5
inhibition. Severe Combined Immune Deficient SCID/beige female mice
with a MDA-MB-231/matrigel xenograft were treated with either
DMSO or compound 3 (SC-1-151) (25 mg/kg) for 31 days. Tumor vo-
lume in the group treated with compound 3 was significantly lower
than that of DMSO-treated animals at day 30 post cell injection
27. Persch E, Dumele O, Diederich F. Angew Chem Int Ed. 2015;54:3290.
28. Wilcken R, Zimmermann MO, Lange A, et al. J Med Chem. 2013;56:1363.
29. Grant SK, Lunney EA. Chem Biol (Cambridge, MA, US). 2011;18:3.
30. Heald RA, Jackson P, Savy P, et al. J Med Chem. 2012;55:4594.
31. Glatz G, Gógl G, Alexa A, et al. J Biol Chem. 2013;288:8596.
32. Fischmann TO, Smith CK, Mayhood TW, et al. Biochemistry. 2009;48:2661.
33. English JM, Vanderbilt CA, Xu S, et al. J Biol Chem. 1995;270:28897.
34. Khokhlatchev A, Xu S, English J, et al. J Biol Chem. 1997;272:11057.
35. Cameron SJ, Abe J-I, Malik S, et al. J Biol Chem. 2004;279:1506.
36. Spicer JA, Rewcastle GW, Kaufman MD, et al. J Med Chem. 2007;50:5090.
37. Bariwal J, Van der Eycken E. Chem Soc Rev. 2013;42:9283.
38. Beletskaya IP, Cheprakov AV. Organometallics. 2012;31:7753.
39. Vachhani DD, Kumar A, Modha SG, et al. Eur J Org Chem. 2013;2013:1223.
40. Pericherla K, Jha A, Khungar B, et al. Org Lett. 2013;15:4304.
41. Majumdar KC, Ganai S. Tetrahedron Lett. 2013;54:6192.
(29.10 mm3
2.29 SEM and 49.48 mm3
3.44 SEM, respectively, p-
value = 0.0001), implicating the role of dual MEK1/2 and MEK5 in-
hibition on tumor growth (See Fig. 10).
In summary, there does appear to be the possibility to develop MEK
isoform-selective small molecules that are active against intact MEK/
ERK signaling cascades and may have relevance to in vivo tumors. There
appears to be divergent structure-activity correlations for compounds
with carboxylate, neutral, or cationic resides on the side chain. These
ionic effects appear to predominate over terminal arene hydrophobic
and halogen contributions to activity. Specifically in the N-methyl-
bound piperazine series, compound 20, displayed both potency and
selectivity for inhibition of the MEK5/ERK5 cascade despite having no
halogens on the terminal arene. This suggests that altering hydrophobic
properties of the lead compound 3 (cLogP = 4.4) may be consistent
with improving both PD and PK properties in a parallel direction as
42. Kaswan P, Pericherla K, Kumar A. Synlett. 2013;24:2751.
43. Zhou F, Guo J, Liu J, et al. J Am Chem Soc. 2012;134:14326.
44. Tan G, Yang Y, Chu C, et al. J Am Chem Soc. 2010;132:12231.
45. Davis EM, Nanninga TN, Tjiong HI, et al. Org Process Res Dev. 2005;9:843.
46. El-Faham A, Albericio F. Chem Rev (Washington DC US). 2011;111:6557.
47. Gaedda TM, Kawanishi Y, Miyazawa A. Synth Commun. 2012;42:1259.
48. Vera-Luque P, Alajarin R, Alvarez-Builla J, et al. Org Lett. 2006;8:415.
49. Pellon RF, Martin A, Mesa M, et al. J Chem Res. 2006:527.
50. D'Angelo ND, Peterson JJ, Booker SK, et al. Tetrahedron Lett. 2006;47:5045.
51. Martin A, Pellon RF, Mesa M, et al. J Chem Res. 2005:561.
52. Teichert A, Jantos K, Harms K, et al. Org Lett. 2004;6:3477.
53. Barluenga J, Bayon AM, Asensio G. J Chem Soc Chem Commun. 1984:1334.
8