Optimization of allosteric MEK inhibitors. Part 2: Taming the sulfamide group balances compound distribution properties

Article abstract of DOI:10.1016/j.bmcl.2015.11.004

Recently, we had identified an unexplored pocket adjacent to the known binding site of allosteric MEK inhibitors which allowed us to design highly potent and in vivo efficacious novel inhibitors. We now report that our initial preclinical candidate, featuring a phenoxy side chain with a sulfamide capping group, displayed human carbonic anhydrase off-target activity and species-dependent blood cell accumulation, which prevented us from advancing this candidate further. Since this sulfamide MEK inhibitor displayed an exceptionally favorable PK profile with low brain penetration potential despite being highly oral bioavailable, we elected to keep the sulfamide capping group intact while taming its unwanted off-target activity by optimizing the structural surroundings. Introduction of a neighboring fluorine atom or installation of a methylene linker reduced hCA potency sufficiently, at the cost of MEK target potency. Switching to a higher fluorinated central core reinstated high MEK potency, leading to two new preclinical candidates with long half-lives, high bioavailabilities, low brain penetration potential and convincing efficacy in a K-Ras-mutated A549 xenograft model.

Full text of DOI:10.1016/j.bmcl.2015.11.004

Bioorganic & Medicinal Chemistry Letters 26 (2016) 186–193
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Optimization of allosteric MEK inhibitors. Part 2: Taming the
sulfamide group balances compound distribution properties
b
a
c
b
Ingo V. Hartung a, , Stefanie Hammer , Marion Hitchcock , Roland Neuhaus , Arne Scholz ,
⇑
Gerhard Siemeister b, Rolf Bohlmann a, Roman C. Hillig d, Florian Pühler b
a Medicinal Chemistry, Bayer HealthCare AG, Global Drug Discovery, Muellerstraße 178, 13353 Berlin, Germany
b Therapeutic Research Group Oncology, Bayer HealthCare AG, 13353 Berlin, Germany
c Research Pharmacokinetics, Bayer HealthCare AG, 13353 Berlin, Germany
d Structural Biology, Bayer HealthCare AG, 13353 Berlin, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Recently, we had identified an unexplored pocket adjacent to the known binding site of allosteric MEK
inhibitors which allowed us to design highly potent and in vivo efficacious novel inhibitors. We now
report that our initial preclinical candidate, featuring a phenoxy side chain with a sulfamide capping
group, displayed human carbonic anhydrase off-target activity and species-dependent blood cell accu-
mulation, which prevented us from advancing this candidate further. Since this sulfamide MEK inhibitor
displayed an exceptionally favorable PK profile with low brain penetration potential despite being highly
oral bioavailable, we elected to keep the sulfamide capping group intact while taming its unwanted off-
target activity by optimizing the structural surroundings. Introduction of a neighboring fluorine atom or
installation of a methylene linker reduced hCA potency sufficiently, at the cost of MEK target potency.
Switching to a higher fluorinated central core reinstated high MEK potency, leading to two new preclin-
ical candidates with long half-lives, high bioavailabilities, low brain penetration potential and convincing
efficacy in a K-Ras-mutated A549 xenograft model.
Received 29 September 2015
Revised 31 October 2015
Accepted 2 November 2015
Available online 3 November 2015
Keywords:
Allosteric MEK inhibitor
Oncology
Structure-based drug design
Human carbonic anhydrase
Blood–plasma ratio
Ó 2015 Elsevier Ltd. All rights reserved.
After decades of research of the canonical Ras–RAF–MEK–ERK
pathway,1 the BRAF (V600E) inhibitor vemurafenib (ZelborafÒ,
Plexxikon/Roche)2 and the allosteric MEK inhibitor trametinib
(MekinistÒ, GSK/Japan Tobacco)3 were recently approved for the
treatment of BRAF (V600E) driven metastatic melanoma. Gratify-
ingly, many years of basic research and drug discovery were finally
translated into a significant benefit for patients suffering from one
of the deadliest forms of cancer. In addition to being testimony for
the advancement of targeted cancer therapeutics, both drug
approvals are also prime examples of how first generation clinical
development compounds inform many aspects of second genera-
tion drug discovery.
Second generation programs are often considered to be less
challenging than first generation projects. Quite to the contrary,
these projects have to deliver clinical candidates which not only
fulfill the traditional PK, PD, safety, toxicological and galenical
requirements for clinical development, but in addition have to
address specific shortcomings of predecessor compounds. For
example, the first generation of advanced clinical MEK1/2
inhibitors, namely PD325901 and AZD6244, did not yield convinc-
ing clinical efficacy most likely due to insufficient target engage-
ment.4 In addition, PD325901 was burdened by ocular toxicity
and likely CNS-mediated adverse effects leading to discontinuation
of monotherapy clinical studies.5 In addition, it has been recently
reported that activated ERK phosphorylates and inhibits CRAF
kinase and that the inhibition of ERK signaling by first-generation
allosteric MEK inhibitors, such as PD325901, relieves ERK-depen-
dent feedback inhibition of CRAF and induces MEK phosphoryla-
tion in most cells.6 A new generation of MEK inhibitors, namely
GDC-0623 and RO5126766, differ from their predecessors in dis-
rupting this MEK feedback phosphorylation which may translate
into superior efficacy.7
Taking these findings into account, the second generation of
allosteric MEK inhibitors needs to maintain the exquisite target
selectivity of its predecessors while surpassing their PK profile,
more specifically by reducing brain penetration and at the same
time securing continuous target inhibition upon once-daily oral
dosing. In addition, effects on MEK feedback phosphorylation have
to be monitored.
In our previous communication,8 we have described the
structural evolution of a novel and highly potent series of allosteric
⇑
Corresponding author.
E-mail address: ingo.hartung@bayer.com (I.V. Hartung).
http://dx.doi.org/10.1016/j.bmcl.2015.11.004
0960-894X/Ó 2015 Elsevier Ltd. All rights reserved.

I. V. Hartung et al. / Bioorg. Med. Chem. Lett. 26 (2016) 186–193
187
MEK inhibitors. Using PD325901 as a starting point, truncation of
its hydroxamic ester headgroup (known to be a metabolic liability)
was combined with incorporation of alkyl and aryl ethers at the
neighboring C6 ring position (Fig. 1). Whereas alkoxy side chains
did not yield inhibitors with sufficient levels of potency, specifi-
cally substituted aryloxy groups gave compounds which fulfilled
this goal. Sulfamide 1 was identified as a highly potent MEK
inhibitor with nanomolar cell potency against BRAF (V600E) as
well as Ras-mutated cell lines, high metabolic stability and result-
ing long half-lives in rodent and non-rodent species. Sulfamide 1
was efficacious in BRAF as well as K-Ras driven xenograft models
and, despite being orally bioavailable, displayed a much lower
brain/plasma exposure ratio than PD325901.
While fulfilling our target profile for a best-in-class next gener-
ation MEK inhibitor, in-depth in vivo profiling of sulfamide 1 in
non-rodent species identified a compound-inherent liability which
prevented us from further advancing this compound toward
clinical development. Additional data for sulfamide 1 and a rational
optimization program which culminated in the identification of
two new candidates which were devoid of this liability are
described in this communication.
Newly synthesized compounds were profiled in an enzymatic
COT–MEK cascade assay and in cell proliferation assays employing
A375 cells, harboring a BRAF (V600E) mutation, and HCT116 cells,
harboring a K-Ras G13D mutation.9 All presented data are average
values of at least two independent measurements. In line with pre-
vious publications, A375 cells were found to be significantly more
sensitive to MEK inhibition than HCT116 cells.10 In order not to
limit the scope of this program to BRAF-mutated tumor entities,
we set the goal of also achieving nanomolar potency in assays with
Ras-mutated tumor cell lines.
As reported previously, 6-phenoxy-substituted benzamides
with NH-linked functionalities at the 30-position of the phenoxy
side chain have been identified as highly potent MEK inhibitors.8
Employing a sulfamide capping group, as in compound 1 (Fig. 1,
Table 1), provided nanomolar potency even in the less sensitive
HCT116 proliferation assay.11 Sulfamide 1 showed moderate-to-
high bioavailabilities in all investigated species (62% in rats, 84%
in mice, 85% in dogs) and long half-lives (32 h in rats, 34 h in mice,
58 h in dogs). Most notably, an exceptionally low brain/plasma
exposure ratio after iv dosing to mice was found. We have corre-
lated brain/plasma exposure ratios qualitatively to total polar sur-
We were well aware of potential liabilities arising from incorpo-
rating an exposed and unsubstituted sulfamide group into our MEK
inhibitor lead series. Inhibition of human carbonic anhydrase (hCA)
by complexation of its active site Zn2+ ion with a SO2NH2 group has
been known for at least a decade.14 From a previous clinical
program within our company,15 we were alerted to the fact that
binding to hCA may lead to compound accumulation in red blood
cells and thereby to exposure variability in patients. Therefore,
inhibitory potency in an in vitro hCA assay and blood–plasma
ratios were closely monitored for all relevant sulfamide analogs.
Sulfamide 1 indeed possessed a submicromolar IC50 value in our
hCA2 assay;16 however, the blood–plasma ratio in mice (0.6) did
not give any initial hint of compound accumulation in red blood
cells. Surprisingly, we later learned that sulfamide 1 displayed
species-dependent blood–plasma ratios. For non-rodents (dogs,
humans), blood–plasma ratios >3 were measured. We did not
investigate causes for this species-dependence as we had no rodent
CA assay available at that time.
Compound accumulation in red blood cells does not manifest a
concern for clinical development per se. However, taking the high
potency and long half-life of our candidate into account, and
expecting a small therapeutic window for highly potent long-act-
ing MEK inhibitors, we considered the likelihood of interindividual
exposure variations in humans due to red blood cell accumulation
to be unacceptable for a next generation best-in-class MEK
inhibitor.17 We therefore decided to embark on a further optimiza-
tion program to remove the unwanted hCA potency while
retaining the beneficial characteristics of sulfamide 1 such as sub-
micromolar potency in HCT116 cells, long half-life and low brain
penetration potential.
Replacement of the sulfamide group was readily dismissed as a
path forward. Sulfamide 1 was originally identified as a metabolite
of the dimethylated analog 2 (Table 1).18 While substituted
sulfamides such as 2 (and several analogs with functionalized alkyl
substituents; data not shown) were found to be highly potent MEK
inhibitors, metabolic instability prevented us from advancing any
of these compounds. In general, dealkylation to the mono- or
unsubstituted sulfamide group was identified as the dominating
metabolic pathway. Thereby, metabolism to pharmacologically
active compounds would not only complicate the assessment of
PK/PD relationships but ultimately lead back to sulfamide 1 and
its hCA issue. Higher alkyl sulfonamides (e.g., compound 3, Table 1)
were found to be as potent as the sulfamide 1 and showed high
metabolic stability and long half-lives (e.g., 28 h in rats for ethyl
sulfonamide 3). The lower polarity of sulfonamides versus sul-
famides (e.g., TPSA 110.5 Å2 for 3 vs 136.5 Å2 for 1) led to measur-
able brain exposure levels after iv dosing to mice.
face area (TPSA) values (as
a measure of polarity-driven
permeability limitations) and P-glycoprotein (Pgp) recognition
(as one well-known mechanism for preventing brain penetration
by active efflux).12 We did not find any evidence of Pgp-mediated
efflux of sulfamide 1; however, analogs from our series with a TPSA
value of 130–140 Å2 consistently possessed low brain penetration
potential in mice while retaining sufficient bioavailability after oral
dosing.13
Therefore, we decided to retain the unsubstituted sulfamide
moiety of our previous candidate 1 and to tame its undesirable
hCA affinity. Initially, we focused our optimization efforts on
capping
group
headgroup right side
OH
O
OH
H2N
H2N
O
PD325901
F
S
O
H
N
HN
O
6
O
HN
Ref. 8
3’
O
F
4’
I
H
H
N
1
3
C6 side chain
F
6
core
I
F
4
Bayer MEK Inhibitor Lead
F
1
Figure 1. Structural evolution of 6-(aryloxy)benzamide MEK inhibitors.8

188
I. V. Hartung et al. / Bioorg. Med. Chem. Lett. 26 (2016) 186–193
H2N
F
O
H2N
F
O
F
F
H
N
F
b)
a)
H2N
+
H2N
O
OH
I
I
F
F
R
O
S
H2N
O
H2N
H2N
F
O
R
F
H
H
H2N
O
N
c)
HN
O
N
I
R
I
F
F
Scheme 1. General synthesis of the sulfamides listed in Table 1. Reagents and conditions: (a) LiHMDS, THF, 0 °C–rt, direct crystallization from crude reaction mixture, 22%;
(b) Cs2CO3, DMF, 50 °C; (c) (i) ClSO2NCO, DMAc (cat.), CH2Cl2, HCO2H, 40 °C; (ii) aniline intermediate, DIPEA, DMAc, rt. Abbreviations: DIPEA = diisopropylethylamine,
DMAc = dimethylacetamide, LiHMDS = lithium hexamethyldisilazide.
introducing substituents into the neighboring 20- and 40-positions
to block access of the sulfamide group to the active site Zn2+ ion
of hCA. Sulfamides with substituents adjacent to the sulfamide
group were synthesized as described for our initial candidate 1
from 2,4,6-trifluorobenzamide (Scheme 1).6
Lithium hexamethyldisilazide promoted nucleophilic substitu-
tion allowed introduction of the right-side 2-fluoro-4-iodoaniline.
Installation of the second ortho substituent was best accomplished
by using cesium carbonate as base at slightly increased tempera-
ture. In general, high levels of regioselectivity were achieved under
these conditions. In most cases, unprotected aminophenols were
used which allowed for direct capping with chlorosulfonyl iso-
cyanate in the final step of our synthetic sequence.19 Our synthetic
route was extraordinarily efficient and allowed us to easily screen
many commercially available aminophenols as potential linkers.20
A selection of the newly synthesized analogs is compiled in Table 1
(compounds 4–15).
next (Table 2). All shown analogs were accessible by adopting
the general synthetic route as detailed in Scheme 1.
Moving the sulfamide group to the 20-position (compound 16)
led to a significant drop in enzymatic as well as cellular potency.
Of note, hCA2 potency was unaffected by this structural change.
Introduction of an additional substituent into the 30-position
(e.g., 17 and 18) further reduced target potency and thereby did
not constitute a viable path forward.
Replacing the inner sulfamide nitrogen of 1 by a methylene
group (compound 19) reduced potency and polarity while affecting
hCA2 potency only marginally. Amino sulfonamide 20 was pursued
because it retained the desired level of polarity while disrupting
the hCA pharmacophore. We were pleased to find that this com-
pound was still a moderately potent MEK inhibitor, although only
a micromolar IC50 value was achieved in the HCT116 proliferation
assay. As expected, hCA2 potency was above the upper limit of our
assay. Blood and plasma concentrations were closely matched both
in mice and humans. Regrettably, compound 20 showed no oral
bioavailability in rats, despite being a low clearance compound,
most likely due to the marked increase in basicity of the capping
moiety and resulting permeability limitations. Benzyl sulfamide
21 showed a much more desirable overall profile with borderline
cellular potency, hCA2 activity sufficiently reduced to achieve bal-
anced mouse and human blood–plasma ratios and good PK charac-
teristics (data not shown). Introduction of a methyl group to the
benzylic position (compound 22) was pursued as an option to rein-
state higher cellular potency, but failed to be successful. In sum-
mary, benzyl sulfamide 21 was identified as an analog being
devoid of blood cell accumulation. As for fluoro sulfamide 4, tam-
ing hCA affinity was offset by lower target potency. In turn, the
project team now faced the challenge to compensate for the
reduced side-chain affinity by optimizing the remainder of the
molecule.
Early in our optimization program, variations of the right-side
aniline had been pursued without identifying any potency improv-
ing modifications. Therefore, we focused on the central benzamide
core. Comparing our sulfamides 4 and 21 to PD325901 (Fig. 1) pre-
sented us with an immediate option for optimizing the core of our
lead series. The additional 3-fluoro substituent of PD325901 had
been initially omitted to streamline synthetic access to 6-substi-
tuted benzamides. Now, we decided to reintroduce an additional
fluoro substituent.
Applying direct CH functionalization to lead compounds (also
referred to as late-stage lead diversification) has attracted much
attention in recent years as an exceptionally efficient way to access
new test compounds from already available leads.21 As a quick
entry into higher fluorinated analogs of our MEK inhibitor lead ser-
ies, we applied direct fluorination conditions (Scheme 2).
Introduction of one fluorine atom into the 40-position (com-
pound 4) led to a marked drop in target potency. Of note, hCA
potency was also significantly reduced and, most importantly, this
compound no longer displayed a shift of the mice versus human
blood–plasma ratio. Installing a second fluorine atom into the 20-
position (compound 5) further deteriorated both MEK and hCA
potency. Similarly, increasing the size of the 40-substituent (com-
pounds 6–9) yielded incremental decreases in MEK potency. Link-
ing the 40-substituent to the proximal sulfamide nitrogen
(compound 10) did not give rise to a sufficiently potent inhibitor.
Interestingly, moving the blocking substituent from the 40-posi-
tion to the 20-position provided a completely different picture.
Installing a methyl group at the 20-position furnished sulfamide
11, which was by far the most potent MEK inhibitor in our program
at that point in time. Unfortunately, hCA2 potency was unaffected
by this substituent which resulted in an unchanged 5-fold shift
between mice and human blood–plasma ratios. Larger substituents
at the 20-position (compounds 12 and 13) paralleled these findings.
Sulfamides 14 and 15 are included in Table 1 as further instruc-
tive examples outlining the steepness of structure–activity rela-
tionships (SAR) for linker substitutions; we did not expect to
influence hCA binding by substituents not adjacent to the sul-
famide moiety.
In summary, 40-fluorinated sulfamide 4 was identified as the
most promising new analog with hCA inhibitory potency suffi-
ciently suppressed by the introduced fluoro substituent. While
taming the sulfamide, this structural modification at the same time
induced an unacceptable deterioration of MEK potency.
As an alternative to introducing neighboring substituents,
changing the position of the sulfamide group or introducing
methylene groups into the sulfamide appendage were pursued

I. V. Hartung et al. / Bioorg. Med. Chem. Lett. 26 (2016) 186–193
189
Table 1
Blocking strategy to reduce hCA affinity by sterical congestion
H2N
*
O
F
H
N
I
F
Compound
Side chain
In vitro IC50 (nM)
TPSA (Å2)
hCA2 (
lM)
Blood/plasma ratio mouse/human
MEK1a
14
A375 (BRAF)
HCT116 (K-Ras)
H
2'
H2N
O
N
O
S
S
*
1
2
3
4
4
180
136.5
113.8
110.5
136.5
0.53
—
0.6/3.8
—
O
4'
6'
H
N
N
O
O
21
15
94
18
155
*
O
H
N
O
S
<30
90
287
—
—
*
O
O
H
H2N
O
N
O
S
*
2510
3.8
0.6/0.7
O
F
F
H
H2N
O
N
O
S
5
118
374
—
136.5
10
—
*
O
F
H
H2N
O
N
O
O
S
*
*
6
7
262
345
180
483
—
—
136.5
136.5
—
—
—
—
O
H
N
H2N
O
S
O
Cl
H2N
O
H
N
O
O
S
*
8
9
592
731
977
—
—
145.8
145.8
—
—
—
—
O
H2N
O
H
N
O
S
*
1900
O
O
O
H2N
S
O
O
N
10
127
339
—
137.0
—
—
*
O
H
N
H2N
O
O
S
S
11
12
15
18
2
6
9
136.5
136.5
0.7
0.7
1.6/7.6
1.8/18
*
*
O
Cl
Br
H
N
H2N
O
O
87
O
H
N
H2N
O
O
O
O
13
14
S
S
S
32
8
152
—
136.5
—
—
—
—
—
*
*
*
O
H
H2N
O
N
899
570
O
H
H2N
O
N
15
941
922
—
—
—
—
O
O
a
Lower detection limit of this assay: IC50 of ꢀ5–15 nM.

190
I. V. Hartung et al. / Bioorg. Med. Chem. Lett. 26 (2016) 186–193
Table 2
Variation of the sulfamide position and introduction of methylene groups
H2N
*
O
F
H
N
I
F
Compound
Side chain
In vitro IC50 (nM)
TPSA (Å2)
136.5
hCA2 (
l
M)
Blood/plasma ratio mouse/human
MEK1a
14
A375 (BRAF)
HCT116 (K-Ras)
H
2'
H2N
N
O
S
S
*
1
4
180
0.53
0.6/3.8
O
O
O
4'
6'
O
H2N
NH
16
17
18
81
107
379
65
138
218
2400
136.5
136.5
136.5
0.5
—
—
—
—
O
O
O
*
*
*
O
O
S
H2N
F
NH
—
O
O
S
S
H2N
NH
—
—
H2N
O
*
19
20
32
11
37
88
3500
1800
124.5
136.5
1.3
—
O
O
H2N
H
N
O
>10
2.1/1.6
S
*
O
S
O
O
O
O
O
21
22
37
83
35
24
757
731
136.5
136.5
2.8
—
0.8/1.1
—
H2N
N
H
*
*
O
O
S
H2N
N
H
a
Lower detection limit of this assay: IC50 of ꢀ5–15 nM.
O
S
HN
O
S
HN
O
S
HN
H2N
O
H2N
O
5
F
O
F
N
O
F
H2N
O
O
O
H
N
H
O
F
1
H
N
O
a)
+
I
F
I
3
4 F
4F
I
F
23 (8%)
3
24 (3%)
Scheme 2. Direct fluorination of ethyl sulfonamide 3. Reagents and conditions: (a) fluoropyridinium trifluoromethanesulfonate (2 equiv), 1,1,2-trichloroethane, 120 °C.
Ethyl sulfonamide 3 was selected as starting material because
we had gram quantities available and did not expect interference
of fluorinating reagents by the sulfonamide group. Gratifyingly,
we were able to isolate both regioisomeric difluorinated benza-
mide cores from a single fluorination reaction. On first inspection,
the isolated yields of 8% (23: 3,4-F2 core) and 3% (24: 4,5-F2 core)
may be considered unacceptably low; however, it must be taken
into account that this approach allowed us to assess both new
cores within days instead of embarking on multistep route syn-
thetic endeavors. Thereby, the benefits clearly outweighed the dis-
advantage of low isolated yields.22
Table 3
Influence of the core fluorination pattern on potency
Compound
In vitro IC50 (nM)
A375 (BRAF)
MEK1a
HCT116 (K-Ras)
3
23
24
15
12
15
<30
2.4
4
287
65
81
a
Lower detection limit of this assay: IC50 of ꢀ5–15 nM. For structures of com-
pounds 23 and 24, see Scheme 2.

I. V. Hartung et al. / Bioorg. Med. Chem. Lett. 26 (2016) 186–193
191
As can be seen from Table 3, introduction of the second fluorine
substituent into the core of 3 increased cellular potency more than
4-fold (e.g., IC50 60–80 nM vs 287 nM in the HCT116 cell prolifera-
tion assay). The sulfonamide with the corresponding 3,4,5-trifluo-
rinated core was found to be almost as potent as 23 and 24 with
the difluorinated cores (data not shown). Further profiling in PK
assays identified the 3,4-difluorinated core (as in sulfonamide 23
and in PD325901) as the most promising core. Therefore, we
decided to combine this difluorinated core with our preferred side
chains.
Several synthetic routes were pursued for these new target
compounds. Adopting our previously used general route
(Scheme 1) necessitated using 2,3,4,6-tetrafluorobenzamide as
starting material. Unfortunately, base-promoted introduction of
phenols was not ortho-selective, as before. Instead, the additional
3-fluoro substituent activated the fluorine atom at the 4-position,
thereby leading to not readily separable mixtures of regioisomers
with the desired regioisomer as only the minor component. There-
fore, a more efficient and selective route was needed. One solution
to circumvent the regioselectivity issue is outlined in Scheme 3.
In order to capitalize on hidden symmetry, we employed 3,4,5-
trifluorophenylboronic acid as starting material and changed the
order of side-chain installations. First, the C6 side chain was intro-
duced by an Ullmann-type coupling of the phenol with the boronic
acid in the presence of copper(II) acetate. Subsequent metalation
and CO2 quenching installed the C1 carboxylic acid group. Intro-
duction of the aniline side chain under basic conditions delivered
the desired N-arylaniline, albeit in low isolated yield.23 Transfor-
mation of the carboxylic acid into the corresponding benzamide
and final Boc deprotection delivered the target compound 28. Data
for compounds containing the combination of the 3,4-difluoro core
with preferred and most instructive C6 side chains are compiled in
Table 4.
ing cellular potency data. Already established SAR trends for side
chains were unaffected by switching to the higher fluorinated core
(e.g., compound 26 being equally potent to compound 25). Most
importantly, employing the 40-fluorophenyl sulfamide side chain
or the benzyl sulfamide side chain provided us with two com-
pounds (27 and 28) which fulfilled our optimization goals. Antipro-
liferative potency was once again in the range of our previous
candidate (compare to Table 1). In addition, hCA2 potency was
reduced for both compounds 27 and 28 by 10-fold relative to 1,
which translated into balanced human and mouse blood–plasma
ratios. Compounds 29 and 30 showcase that we were able to coun-
terbalance reduced side-chain affinity in several, but not all, cases
by the higher fluorinated core. Sulfamide 31 was finally prepared
for solely academic reasons to prove the additivity of potency-
increasing effects: By combining the most potent side chain (ana-
log 11, Table 1) with the most potent core (compound 23,
Scheme 2), we achieved an IC50 of 2 nM in the HCT116 proliferation
assay, which is 100-fold more potent than our previous candidate 1
and PD325901.
Compounds from this series of allosteric MEK inhibitors as well
as hybrids between our series and published MEK inhibitors were
investigated with regard to their effects on MEK feedback phos-
phorylation and were found to differ significantly from the first
generation of MEK inhibitors, namely PD325901. These results
and a discussion of structural features likely responsible for mod-
ulating MEK feedback phosphorylation will be reported
elsewhere.24
Nevertheless, for advanced optimization projects, potency is
rarely the decisive selection criterion as was the case in our project
where sulfamides 27 and 28 were selected based on their overall
most favorable profile.
Since our novel sulfamides 27 (BAY-866) and 28 (BAY-438)
were unchanged with respect to overall polarity and we had not
introduced new metabolic liabilities, we did not expect major
changes in their PK profiles compared to our previous candidate
1. Indeed, both new sulfamides showed the same favorable PK pro-
file as their predecessor (Table 5) with low clearance, long half-life
and high bioavailability in rats. In addition, no relevant compound
exposure was seen in brain tissue after iv dosing to mice.
Combining the improved core with the unchanged sulfamide
side chain of our previous candidate 1 provided the expected gain
in antiproliferative potency. Several of the newly synthesized ana-
logs reached the lower detection limit of the biochemical MEK
assay (IC50 values below 20 nM). Therefore the true difference in
potency between those analogs became only apparent by compar-
O
O
O
O
S
OH
B
S
HN
NH
HN
NH
F
F
a)
b)
d)
OH
HO
F
F
O
O
O
+
O
O
F
F
O
O
O
O
S
S
HO
O
HO
O
HN
NH
HN
NH
F
H
c)
F
F
O
O
N
O
O
O
O
F
I
F
H2N
F
F
I
O
O
O
O
S
H2N
O
S
H2N
O
HN
NH
F
H2N
NH
F
H
N
H
e)
1
O
O
N
O
O
6
I
F
I
F
3
4
F
F
28
Scheme 3. Synthesis of sulfamide 28 with a 3,4-difluorinated benzamide core. Reagents and conditions: (a) Cu(OAc)2 (1 equiv), pyridine, rt, 20%; (b) LDA (5.5 equiv), CO2
(excess), ꢁ78 °C–rt, 61%; (c) LiHMDS (7.5 equiv), THF, 0 °C, 19%; (d) CDI, NH3, DMF, rt, 36%; (e) TFA, CH2Cl2, rt, 85%. Abbreviations: CDI = 1,10-carbonyldiimidazole,
LDA = lithium diisopropylamide, LiHMDS = lithium hexamethyldisilazide, TFA = trifluoroacetic acid.

192
I. V. Hartung et al. / Bioorg. Med. Chem. Lett. 26 (2016) 186–193
Table 4
Compounds containing a difluorinated benzamide core with a preferred C6 side chain
H2N
*
O
F
H
N
I
F
F
Compound
Side chain
In vitro IC50 (nM)
TPSA (Å2)
hCA2 (
lM)
Blood/plasma ratio mouse/human
MEK1a
13
A375 (BRAF)
HCT116 (K-Ras)
H
H2N
O
N
O
S
25
26
2
50
136.5
113.6
—
—
—
—
*
O
H
N
O
N
O
S
10
14
6
42
*
O
H
N
H2N
O
O
S
*
27 BAY-866
13
277
136.5
4.6
0.8/1.0
O
F
O
O
S
O
*
28 BAY-438
21
22
4
124
303
136.5
124.5
5
0.8/0.8
0.8/1.1
H2N
N
H
*
H2N
O
O
O
S
O
29
11
2.2
F
H
N
H2N
O
S
30
31
28
14
32
2950
2
136.5
136.5
—
—
—
—
*
*
O
F
H
H2N
O
N
O
S
O
0.2
a
Lower detection limit of this assay: IC50 of ꢀ5–15 nM.
Table 5
PK data of key compounds
Table 6
Cell proliferation data of key compounds
Compound
Rat PKa
Vss (L/kg)
Mice PK
Compound
IC50 (nM)
Clblood
(L/kg/h)
t1/2 (h)
F (%)
Brain/plasma
A375
(BRAF) (BRAF)
COLO205
HepG2
HCT116
(K-Ras)
A549
MCF7
ratiob
(N-Ras)
(K-Ras)
PD325901
AZD6244
1
BAY-866 (27)
BAY-438 (28)
0.5
1.7
0.2
2.0
0.5
1.9
5.4
4.1
32
26
32
104
24
62
84
78
0.11
PD325901
AZD6244
BAY-866
(27)
BAY-438
(28)
13
31
13
5
69
42
5
32
2
185
3000
277
166
1750
314
9550
>10,000
>10,000
0.06
0.03
0.02
0.07
<0.02
<0.02
<0.02
<0.02
4
9
5
124
185
>10,000
a
Dosing for PD325901: 0.5 mg/kg iv/1 mg/kg po; dosing for AZD6244: 0.5 mg/kg
iv/5 mg/kg po; dosing for 1: 1 mg/kg iv/1 mg/kg po; dosing for 27 and 28: 0.5 mg/kg
iv/1 mg/kg po.
b
AUC(brain)/AUC(plasma) for 0–3 h after 5 mg/kg iv dosing.
sulfamides showed significantly lower brain penetration behavior
than PD325901 in mice iv dosing studies (Table 5). This likely
reflects the higher polarity and higher hydrogen bond donor count
of 27 and 28 and thereby impaired passive diffusion via the blood–
brain barrier, which is known to be more restrictive to polar com-
pounds than the GI barrier.25
Our two new candidates were subsequently profiled in a larger
panel of cell lines (see Table 6 for a selection of the data). Benzyl
sulfamide BAY-438 (28) was as potent as PD325901, while the
40-fluorophenyl sulfamide BAY-866 (27) was slightly less potent.
As expected, no antiproliferative activity was seen in cell lines
devoid of Ras–RAF–MEK–ERK pathway deregulation. In line with
these findings, we did not see relevant off-target activities in selec-
tivity panels (e.g., kinase panels).
As for our previous candidate 1, we did not identify any proof of
Pgp-mediated active efflux in pharmacokinetic (e.g., Caco-2 assay)
or pharmacological studies as a reason for the low brain penetra-
tion behavior. For example, pERK inhibition measurements in a
matched pair of cell lines (HeLa-MaTu vs the Pgp-expressing
HeLa-MaTu-ADR cell line) showed a significant IC50 shift for
AZD6244 (3.7 lM in HeLa-MaTu-ADR vs 14 nM in HeLa-MaTu),
proving that AZD6244 is a strong Pgp substrate which prevents
access across the blood–brain barrier. For PD325901, low nanomo-
lar IC50 values for pERK inhibition were measured in both cell lines.
Our novel sulfamides BAY-866 and BAY-438 closely matched the
profile of PD325901, with a smaller than 5-fold shift in IC50 values
between the two cell lines (e.g., 1.8 nM in HeLa-MaTu and 8.7 nM
in HeLa-MaTu-ADR for sulfamide 28). At the same time, both
Both novel sulfamides showed strong efficacy in A549 mice
xenograft studies (Fig. 2). In line with our in vitro data, BAY-438

I. V. Hartung et al. / Bioorg. Med. Chem. Lett. 26 (2016) 186–193
193
Efficacy in A549 xenografts
125
100
75
50
25
0
vehicle, po qd
BAY-866 (27) 2 mg/kg po qd
T/C = 0.21
BAY-438 (28), 1 mg/kg po qd
T/C = 0.13
Compound 1, 1 mg/kg po qd
T/C = 0.15
0
10
20
30
days after implantation
Figure 2. In vivo A549 xenograft study with sulfamides BAY-866 (27) and BAY-438 (28) in nude mice.
11. Cell proliferation inhibition data were correlated in selected cases to
measurements of pERK inhibition in the respective cell lines (Mesoscale
assay set-up) to rule out off-target cytotoxicity. As pERK inhibition IC50 values
in general paralleled proliferation IC50 values the former are—for the sake of
clarity—excluded from this manuscript.
was slightly more potent (maximal efficacy reached @ 1 mg/kg oral
daily dosing) than BAY-866 (2 mg/kg orally once daily).
In conclusion, we were able to tame hCA potency of sulfamide-
containing allosteric MEK inhibitors by structural changes in the
direct vicinity of the sulfamide group. State-of-the-art MEK
potency was subsequently reinstated by introducing an additional
fluorine atom into the central benzamide core. A streamlined com-
pound profiling tree with high predictability of pharmacological
and PK in vitro assays for in vivo performance, in combination with
short and flexible synthetic routes, secured fast learning cycles.
This enabled selection of two new preclinical candidates within
6 months after identification of the hCA liability of the initial can-
didate. Both candidates fulfilled the target profile for next genera-
tion allosteric MEK inhibitors. Both compounds have been
advanced to non-rodent safety and toxicological profiling.
12. (a) Bagal, S. K.; Bungay, P. J. ACS Med. Chem. Lett. 2012, 3, 948; (b) Di, L.; Rong,
H.; Feng, B. J. Med. Chem. 2013, 56, 2.
13. Traditionally, a TPSA value of 120 Å2 is considered to be a cut-off for securing
sufficient passive diffusion through the GI barrier. In the case of sulfamide 1
and related analogs, intramolecular hydrogen bonds between the amide
headgroup and the neighboring ether and amine linkages have to be taken into
account as they reduce the exposed polar surface area. Therefore, we have not
compared TPSA data to compounds which do not feature this intramolecular
H-bonding network, for example, PD325901.
14. Ceruso, M.; Vullo, D.; Scozzafava, A.; Supuran, C. T. Bioorg. Med. Chem. 2013, 21,
6929. and references cited therein.
15. Lücking, U.; Jautelat, R.; Krüger, M.; Brumby, T.; Lienau, P.; Schäfer, M.; Briem,
H.; Schulze, J.; Hillisch, A.; Reichel, A.; Wengner, A. M.; Siemeister, G.
ChemMedChem 2013, 8, 1067.
16. Inhibitory potency against human carboanhydrase isozymes 1 (hCA1) closely
paralleled data for hCA2, albeit at a lower potency level (e.g., IC50 4.8 lM for
sulfamide 1). For the sake of clarity, hCA1 data are omitted from this
manuscript.
References and notes
17. Sulfamide-containing dual Raf–MEK inhibitors based on a coumarin core were
reported recently by researchers from Chugai without any mention of hCA
findings: Aoki, T.; Hyohdoh, I.; Furuichi, N.; Ozawa, S.; Watanabe, F.;
Matsushita, M.; Sakaitani, M.; Ori, K.; Takanashi, K.; Harada, N.; Tomii, Y.;
Tabo, M.; Yoshinari, K.; Aoki, Y.; Shimma, N.; Iikura, H. Bioorg. Med. Chem. Lett.
2013, 23, 6223.
18. Identification of clinical candidates as metabolites of initial lead compounds, a
process which has been coined as pre-metabolization, is not uncommon in lead
optimization programs: Clader, J. W. J. Med. Chem. 2004, 47, 1.
19. In selected cases, tert-butoxycarbonyl-protected aminophenols were used,
with subsequent deprotection under acidic conditions.
20. Gratifyingly, aniline intermediates (devoid of the sulfamide capping group)
could already be used for SAR assessment. MEK inhibitory potency trends for
aniline intermediates paralleled SAR for the fully functionalized sulfamides
(data not shown), albeit on a lower potency level. Thereby, we were able to
screen many more aminophenols than indicated in Tables 1 and 2 and discard
low potency linkers already at the aniline intermediate step.
21. (a) Wencel-Delord, J.; Glorius, F. Nat. Chem. 2013, 5, 369; (b) Masood, M. A.;
Bazin, M.; Bunnage, M. E.; Calabrese, A.; Cox, M.; Fancy, S.-A.; Farrant, E.;
Pearce, D. W.; Perez, M.; Hitzel, L.; Peakman, T. Bioorg. Med. Chem. Lett. 2012,
22, 1255.
22. In general, we consider an isolated yield of 10% in a final-stage-diversification
reaction as a more than acceptable outcome as this yield equates to a eight-
step sequence with 75% yield for each individual step, while reducing time and
resource investments by more than 95%.
23. The low isolated yield did not result from low regioselectivity of the addition
reaction but rather from low product recovery upon column chromatographic
purification due to limited solubility in organic solvents.
24. Hartung, I. V.; Pühler, F.; Neuhaus, R.; Scholz, A.; Siemeister, G.; Geisler, J.;
Hillig, R. C.; von Ahsen, O.; Hitchcock, M. ChemMedChem 2015. in press.
25. Wager, T. T.; Hou, X.; Verhoest, P. R.; Villalobos, A. ACS Chem. Neurosci. 2010, 1,
435.
1. (a) Roberts, P. J.; Der, C. J. Oncogene 2007, 26, 3291; (b) Montagut, C.; Settleman,
J. Cancer Lett. 2009, 283, 125.
2. Bollag, G.; Tsai, J.; Zhang, J.; Zhang, C.; Ibrahim, P.; Nolop, K.; Hirth, P. Nat. Rev.
Drug Disc. 2012, 11, 873.
3. (a) Gilmartin, A. G.; Bleam, M. R.; Groy, A.; Moss, K. G.; Minthorn, E. A.;
Kulkami, S. G.; Rominger, C. M.; Erskine, S.; Fisher, K. E.; Yang, J.; Zappacosta, F.;
Annan, R.; Sutton, D.; Laquerre, S. G. Clin. Cancer Res. 2011, 17, 989. See also:;
(b) Abe, H.; Kikuchi, S.; Hayakawa, K.; Iida, T.; Nagahashi, N.; Maeda, K.;
Sakamoto, J.; Matsumoto, N.; Miura, T.; Matsumura, K.; Seki, N.; Inaba, T.;
Kawasaki, H.; Yamaguchi, T.; Kakefuda, R.; Nanayama, T.; Kurachi, H.; Hori, Y.;
Yoshida, T.; Kakegawa, J.; Watanabe, Y.; Gilmartin, A. G.; Richter, M. C.; Moss, K.
G.; Laquerre, S. G. ACS Med. Chem. Lett. 2011, 2, 320.
4. In the meantime, AZD6244 has been moved into phase III studies for the
treatment of K-Ras-mutated NSCLC in combination with docetaxol, after
identification of a superior compound formulation.
5. Haura, E. B.; Ricart, A. D.; Larson, T. G.; Stella, P. J.; Bazhenova, L.; Miller, V. A.;
Cohen, R. B.; Eisenberg, P. D.; Selaru, P.; Wilner, K. D.; Gadgeel, S. M. Clin. Cancer
Res. 2010, 16, 2450.
6. (a) Hatzivassiliou, G.; Haling, J. R.; Chen, H.; Song, K.; Price, S.; Heald, R.; Hewitt,
J. F. M.; Zak, M.; Peck, A.; Orr, C.; Merchant, M.; Hoeflich, K. P.; Chan, J.; Luoh, S.-
M.; Anderson, D. J.; Ludlam, M. J. C.; Wiesmann, C.; Ultsch, M.; Friedman, L. S.;
Malek, S.; Belvin, M. Nature 2013, 501, 232; (b) Ishii, N.; Harada, N.; Joseph, E.
W.; Ohara, K.; Miura, T.; Sakamoto, H.; Matsuda, Y.; Tomii, Y.; Tachibana-
Kondo, Y.; Iikura, H.; Aoki, T.; Shimma, N.; Arisawa, M.; Sowa, Y.; Poulikakos, P.
I.; Rosen, N.; Aoki, Y.; Sakai, T. Cancer Res. 2013, 73, 4050.
7. Caunt, C. J.; Sale, M. J.; Smith, P. D.; Cook, S. J. Nat. Rev. Cancer 2015, 15, 577.
8. Hartung, I. V.; Hitchcock, M.; Pühler, F.; Neuhaus, R.; Scholz, A.; Hammer, S.;
Petersen, K.; Siemeister, G.; Brittain, D.; Hillig, R. C. Bioorg. Med. Chem. Lett.
2013, 23, 2384.
9. For detailed assay descriptions, see: (a) Rudolph, J.; Dumas, J.; Li, Y.; Auclair, D.;
Lobell, M.; Hitchcock, M.; Hartung, I.; Koppitz, M.; Brittain, D.; Puehler, F.;
Petersen, K.; Guenther, J. PCT Patent WO 2008/138639; (b) Hitchcock, M.;
Hartung, I. PCT Patent WO 2009/129938.
10. Solit, D. B.; Garraway, L. A.; Pratilas, C. A.; Sawai, A.; Getz, G.; Basso, A.; Ye, Q.;
Lobo, J. M.; She, Y.; Osman, I.; Golub, T. R.; Sebolt-Leopold, J.; Sellers, W. R.;
Rosen, N. Nature 2006, 439, 358.