Discovery of Rogaratinib (BAY 1163877): a pan-FGFR Inhibitor

Article abstract of DOI:10.1002/cmdc.201700718

Rogaratinib (BAY 1163877) is a highly potent and selective small-molecule pan-fibroblast growth factor receptor (FGFR) inhibitor (FGFR1–4) for oral application currently being investigated in phase 1 clinical trials for the treatment of cancer. In this publication, we report its discovery by de novo structure-based design and medicinal chemistry optimization together with its pharmacokinetic profile.

Full text of DOI:10.1002/cmdc.201700718

DOI: 10.1002/cmdc.201700718
Full Papers
Discovery of Rogaratinib (BAY 1163877): a pan-FGFR
Inhibitor
Marie-Pierre Collin,*[a] Mario Lobell,[a] Walter Hꢀbsch,[a] Dirk Brohm,[a] Hartmut Schirok,[a]
Rolf Jautelat,[a] Klemens Lustig,[c] Ulf Bçmer,[d] Verena Vçhringer,[b] Mꢁlanie Hꢁroult,[b]
Sylvia Grꢀnewald,[b] and Holger Hess-Stumpp[b]
Rogaratinib (BAY 1163877) is a highly potent and selective
small-molecule pan-fibroblast growth factor receptor (FGFR) in-
hibitor (FGFR1–4) for oral application currently being investi-
gated in phase 1 clinical trials for the treatment of cancer. In
this publication, we report its discovery by de novo structure-
based design and medicinal chemistry optimization together
with its pharmacokinetic profile.
Introduction
Fibroblast growth factors (FGFs) and their transmembrane tyro-
sine kinase receptors play an important role in tissue homeo-
stasis during embryonic development and adulthood.[1] It has
been found that FGF signaling is affected by multiple mecha-
nisms: gene amplifications, activating mutations, chromosomal
translocations, single nucleotide polymorphisms and aberrant
splicing at the post-transcriptional level.[2] Because of the
strong link between aberrant FGFR signaling and carcinogene-
sis, FGFR inhibition appears to be an innovative approach for
new cancer therapies.
Figure 1. Selected pan-FGFR inhibitors under investigation in clinical trials.
We set out to identify a novel chemotype for selective inhib-
ition of the FGFR kinases with a favorable drug metabolism
and pharmacokinetic (DMPK) profile for oral application. The
selectivity versus VEGFR-2 was an important optimization pa-
rameter in order to improve the tolerability related to VEGFR-
2-mediated side effects. The FGFR inhibitor project started
with a de novo structure-based design approach.
Several specific orally bioavailable small-molecule inhibitors
of FGFR are currently in clinical development (Figure 1).[3] The
acylaminopyrazole AZD4547 is a pan-FGFR inhibitor (in vitro
IC50 values of 0.2, 2.5, and 1.8 nm against FGFR-1, FGFR-2, and
FGFR-3, respectively) exhibiting selectivity against VEGFR-2
(also known as KDR, IC50 =24 nm). It is under investigation in
phase II/III clinical trials for the treatment of squamous cell
lung cancer.[4]
In 2015 Hillisch et al. described how computational chemis-
try has significantly and positively impacted the clinical devel-
opment pipeline of the pharmaceutical industry in general,
and Bayer in particular.[6] In that report, we referred to success-
ful examples of de novo structure-based design leading to clin-
ical candidates. Herein we report one of such yet unpublished
examples, namely the lead discovery and optimization pro-
gram leading to rogaratinib, a potent and selective FGFR1–4
inhibitor.
JNJ-42756493 is a quinoxaline pan-FGFR inhibitor (FGFR1–4
IC50 ꢀ1 nm; VEGFR kinase selectivity ~20-fold) with ongoing
development in a phase I/II clinical study for the treatment of
urothelial cancer.[5]
[a] Dr. M.-P. Collin, Dr. M. Lobell, Dr. W. Hꢀbsch, Dr. D. Brohm, Dr. H. Schirok,
Dr. R. Jautelat
Results and Discussion
Department of Medicinal Chemistry, Drug Discovery, Bayer AG, Postfach
101709, 42096 Wuppertal (Germany)
Design of 5-benzothiophene-pyrrolotriazine
E-mail: marie-pierre.collin@bayer.com
At the time this work was performed X-ray crystal structures of
the FGFR kinase domain had only been published for isoform 1
and not for isoforms 2, 3, or 4. Between the four isoforms
there are no mutations within 5 ꢂ of the ATP binding site
except for the Y566C mutation in FGFR4. For this reason it was
seen as acceptable to do all modeling with one of the pub-
lished FGFR1 X-ray crystal structures. The starting point of the
structure-based design approach was a WaterMap analysis of
the FGFR-1 ATP binding site (PDB ID: 2FGI, active DFG-in
[b] Dr. V. Vçhringer, Dr. M. Hꢁroult, Dr. S. Grꢀnewald, Dr. H. Hess-Stumpp
Therapeutic Research Group, Oncology, Drug Discovery, Bayer AG (Germa-
ny)
[c] Dr. K. Lustig
Research Pharmacokinetics Group, Drug Discovery, Bayer AG (Germany)
[d] Dr. U. Bçmer
Lead Discovery, Drug Discovery, Bayer AG (Germany)
Supporting information and the ORCID identification number(s) for the
author(s) of this article can be found under:
https://doi.org/10.1002/cmdc.201700718.
ChemMedChem 2018, 13, 1 – 10
1
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
kinase form).[7] WaterMap is a software tool from Schrçdinger,
which performs mapping of the locations and thermodynamic
properties of water molecules that solvate a protein binding
site.[8,9] Displacement of high-energy water molecules by a
ligand should contribute to its binding affinity. One can there-
fore use the locations of high-energy water molecules in a cal-
culated water map to guide de novo ligand design as illustrat-
ed in Figure 2. ATP-competitive kinase inhibitors usually form
region. In an effort to displace many of the unfavorable high-
energy water molecules calculated by WaterMap, a 7-alkoxy-
benzothiophene head group in the 5-position of the pyrrolo-
triazine was proposed. The five-membered thiophene ring pro-
vides exactly the desired vectors needed to position the
phenyl ring of the benzothiophene in a central position in the
back pocket and enables the alkoxy substituent to enter a
deep subpocket, forming a hydrogen bond to the backbone
NH of Asp641. An additional piperidinyl group was attached to
enhance water solubility, leading to compound 1 (Figure 2).
Compound 1 was synthesized and tested in the biochemical
FGFR-1 assay with low ATP (10 mm) and high ATP (2 mm) con-
centrations, yielding IC50 values of 48 nm (low ATP) and 115 nm
(high ATP). Some selectivity was observed versus VEGFR-2 with
an IC50 value of 823 nm at low ATP concentration (Table 1). In
our cellular FGFb-stimulated HUVEC proliferation assay, com-
pound 1 showed an IC50 value of 1.1 mm. The already appeal-
ing biochemical and cellular FGFR inhibitory potency of our
first compound combined with some selectivity toward VEGFR-
2 encouraged us to further optimize this lead structure.
Medicinal chemistry optimization
Starting with the optimization of the head group, analysis of
the FGFR-1 crystal structure indicated two areas for potential
improvement. Introducing a lipophilic substituent at C5 of the
benzothiophene group should fill a hydrophobic back pocket,
fully displacing the highest-energy hydration site, indicated by
a red arrow in Figure 2. Based on the above-mentioned hy-
pothesis, the 5-methylbenzothiophene
2 was synthesized
(Table 1). In vitro testing showed that despite the absence of
the C7 methoxy group, introduction of a methyl group at posi-
tion C5 of the benzothiophene maintained FGFR-1 potency
(compound 2, IC50 =86 nm). 5-Fluoro (3), 5-trifluoromethyl (4),
and 5-trifluoromethoxy (5) analogues were significantly
weaker. Elongation of the methoxy group promised to replace
an additional high-energy hydration site indicated by the
orange arrow in Figure 2B. However, replacement of the me-
thoxy group by a trifluoroethoxy substituent (6) resulted in de-
creased rather than the anticipated increased activity (Table 1).
Later attempts with an ethoxy or fluoroethoxy group in
combination with a different tail group also led to derivatives
with decreased activity (discussed below in greater detail). Fur-
ther variation around the benzothiophene moiety demonstrat-
ed that benzothiophene was the most favorable substituent.
Indeed, its replacement by a naphthyl (8) or benzofuran (9)
group led to lower FGFR-1 activity. Finally, combination of the
5-methyl and 7-methoxy substituents resulted in a favorable
additive effect on FGFR activity (compound 10: IC50 =4 nm)
and represented a major breakthrough in the lead optimiza-
tion.
Figure 2. A) Chemical structure of compound 1. B) Top view of ATP binding
site in FGFR-1 (green C atoms, semitransparent grey surface) with docked
compound 1 (pink C atoms) and overlayed hydration sites calculated by Wa-
terMap. Only the energetically most unfavorable hydration sites with the
highest calculated free energy (DG>2 kcalmolꢁ1) are shown as spheres col-
ored from green (near 2 kcalmolꢁ1) to red (near 7 kcalmolꢁ1), the two high-
est-energy hydration sites are indicated by a red (7.2 kcalmolꢁ1) and orange
(6.1 kcalmolꢁ1) arrow. Predicted hydrogen bonds between the inhibitor and
the protein are shown as yellow broken lines. C) Same as panel B, but side
view with front parts of the protein removed for clarity. The side chain of
Glu571 is shown without a semitransparent grey surface in the upper right-
hand corner.
one to three hydrogen bonds to the so-called kinase hinge
region, which offers a triad of potential hydrogen bonding
partners via backbone carbonyl acceptor, NH donor, and car-
bonyl acceptor. In a first design step, a virtual in-house collec-
tion of various heterocyclic scaffolds were docked for evalua-
tion as potential hinge binders. Among several possible heter-
ocyclic core structures, the aminopyrrolotriazine core was se-
lected as the most promising scaffold. It provided hinge bind-
ing with two hydrogen bonds and suitable attachment vectors
to place a head group into the hydrophobic back pocket of
FGFR-1 and a solubilizing tail group directed toward a solvent
To optimize the tail group, we tried to exploit a structural
difference in the ATP binding site. Glutamic acid 571 (Glu571)
of the FGFR-1 ATP binding site is exchanged for a threonine in
the corresponding position in VEGFR-2. We hoped to create a
beneficial ligand interaction with Glu571 to enhance FGFR-1
potency and VEGFR-2 selectivity. The insertion of a methylene
&
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
2
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 1. Selected RA substituents and their effect on potency toward FGFR-1 and VEGFR-2.
Compd
RA
FGFR-1
IC50 [nm][a]*
VEGFR-2
IC50 [nm][b]#
Compd
RA
FGFR-1
IC50 [nm][a]*
VEGFR-2
IC50 [nm][b]#
1
2
115
86
823
116
6
7
838#
319
n.d.
152
3
4
5
567
940
n.d.
n.d.
n.d.
8
9
526
3740
4
273
>20000
62
3770#
10
[a] Determined by biochemical FGFR-1 activity inhibition assay. [b] Determined by biochemical VEGFR-2 activity inhibition assay. See Supporting Informa-
#
tion for further details. *High ATP=2 mm, Low ATP=10 mm; n.d.=not determined.
group between the pyrrolotriazine core and a piperazine tail
group was predicted to allow the desired hydrogen bond in-
teraction with the side chain of Glu571 (Figure 3).
logue of piperazine 11 had weaker FGFR-1 activity. A straight-
forward combination of the most promising head and tail
groups afforded our lead candidate compound 18 (Table 3).
Potency, selectivity, and key pharmacokinetic properties of
compound 18 are listed in Tables 3 and 4. The pyrrolotriazine
18 is a potent FGFR-1 and FGFR-3 inhibitor with good cellular
potency. The VEGFR-2 selectivity of this compound was also
improved. Docking of compound 18 in FGFR-1 is illustrated in
Figure 3. In vivo pharmacokinetic investigations of pyrrolotria-
zine 18 in rat showed low clearance, but only low to moderate
bioavailability (Table 4).
In fact, compound 11, with a 7-piperazinylmethyl substituent
retained FGFR potency and diminished VEGFR-2 activity rela-
tive to its piperidine counterpart 1 (Table 2). Broad derivatiza-
tion of the tail group (Table 2) furnished further interesting
compounds with a 3-pyrrolidine (12) or a morpholinomethyl
moiety (13). Compound 15, a more flexible open-chain ana-
During the lead optimization phase, the head group was fur-
ther investigated. Replacing the benzothiophene group by a
thiophene carboxylic acid or its corresponding ethyl ester
(Table 5, compounds 23 and 24) resulted in a significant loss of
potency.
Compound 20 showed good FGFR potency paired with high
VEGFR-2 selectivity; however, in vitro hepatocyte clearance was
very high (3.4 Lhꢁ1 kgꢁ1), probably due to conjugative phase II
metabolism involving the hydroxy group. To address the lipo-
philic back pocket, the 5-chlorobenzothiophene analogue 25
was synthesized and surprisingly did not quite achieve the ac-
tivity level of our lead compound 18 (compound 25 IC50 =
35 nm). Elongation of the C7 substituents did not improve the
FGFR potency (compounds 26, 27, and 28) in contradiction
with modeling predictions based on the WaterMap analysis.
We hypothesized that the basic secondary amine moiety of
compound 18 (exp. pKa =9.1) was responsible for the active
Figure 3. Side view of ATP binding site in FGFR-1 (green C atoms, semitrans-
parent grey surface) with docked compound 18 (pink C atoms). Predicted
hydrogen bonds between the inhibitor and the protein are shown as yellow
broken lines. The carboxylate moiety of Glu571 is shown without a semi-
transparent grey surface in the upper right-hand corner. Front parts of the
protein have been removed for clarity.
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
3
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 2. Selected RB substituents and their effect on FGFR-1 and VEGFR-2 potency.
Compd
RB
FGFR-1
IC50 [nm][a]*
VEGFR-2
IC50 [nm][b]#
Compd
RB
FGFR-1
IC50 [nm][a]*
VEGFR-2
IC50 [nm][b]#
1
115
823
14
214
109
11
12
93
88
3030
215
15
16
376
588
1100
n.d.
13
124#
1300
17
629
n.d.
[a] Determined by biochemical FGFR-1 activity inhibition assay. [b] Determined by biochemical VEGFR-2 activity inhibition assay. See Supporting Informa-
#
tion for further details. *High ATP=2 mm, Low ATP=10 mm; n.d.=not determined.
Table 3. FGFR and VEGFR-2 potency of the lead candidate 18.
FGFR-1 (high ATP*)[a]
FGFR-3 (low ATP#)[b]
VEGFR-2 (low ATP#)[c]
10 nm
25 nm
846 nm
Biochemical assay IC50
Proliferation assay IC50
FGFb-stimulated HUVEC proliferation[d]
VEGF-stimulated HUVEC proliferation[e]
Selectivity ratio: VEGF/FGFb
36 nm
2.0 mm
56
[a] Determined by biochemical FGFR-1 activity inhibition assay. [b] Determined by biochemical FGFR-3 activity inhibition assay. [c] Determined by biochem-
ical VEGFR-2 activity inhibition assay. [d] Inhibitory potency of endothelial cell growth stimulated by FGFb. [e] Inhibitory potency of endothelial cell growth
#
stimulated by VEGF. See Supporting Information for further details. *High ATP=2 mm, Low ATP=10 mm.
hypothesis was supported by the high ERs of compounds 30
Table 4. Rat in vivo pharmacokinetic parameters after intravenous (i.v.)
and oral (p.o.) application of lead candidate 18.[a]
and 31 (Table 6, ERs of >28 and 13, respectively). As a conse-
quence, tail groups with decreased basicity were conceived. In-
troduction of a carbonyl group in the piperazine ring or re-
placement of the piperazine moiety by an aminopyrrolidinone
group would fulfill this criterion and was in agreement with
modeling considerations (Figure 4). Surprisingly, N-acetylation
of the piperazine ring also led to a quite active and more per-
meable compound (37: Caco-2 Papp(A!B) =241 nmsꢁ1; ER=1).
The rather small loss of activity of the N-acetyl analogue 37 rel-
ative to piperazine 18 or piperazinone 38 indicates that the hy-
pothesized hydrogen bond interaction with the side chain of
Glu571 is not essential to achieve good potency. These struc-
tural alterations improved the permeation profile (decrease in
ER; 38: Caco-2 Papp(A!B) =314 nmsꢁ1; ER=2) but also slightly
i.v. [1 mgkgꢁ1
]
p.o. [3 mgkgꢁ1
]
CLbl
Vss,pl
t1/2
0.45 Lhꢁ1 kgꢁ1
5.5 Lkgꢁ1
12 h
AUCnorm
t1/2
F
0.78 kghꢁ1 Lꢁ1
12 h
22%
[a] 5 mLkgꢁ1; vehicle: plasma/DMSO=99:1 (i.v.: 15 min infusion) or Solu-
tolꢄ/ethanol/water=40:10:50 (p.o.); CLbl =clearance blood, Vss,pl =volume
of distribution at steady state in plasma, t1/2 =half-life, AUCnorm =dose-
normalized (1 mgkgꢁ1) area under the curve, F=absolute oral bioavaila-
bility. See Supporting Information for further details.
transporter efflux in the Caco-2 assay (Papp(A!B) =38 nmsꢁ1
efflux ratio (ER)=4) explaining the low oral bioavailability. This
,
&
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
4
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 5. Selected RA substituents and their effect on FGFR-1 and VEGFR-2 potency.
Compd
18
RA
FGFR-1
VEGFR-2
Compd
22
RA
FGFR-1
VEGFR-2
Compd
26
RA
FGFR-1
VEGFR-2
IC50 [nm][a]* IC50 [nm][b]#
IC50 [nm][a]* IC50 [nm][b]#
IC50 [nm][a]* IC50 [nm][b]#
10
227
12
846
>20000
>20000
962
8190
n.d.
n.d.
n.d.
203
241
144
447
n.d.
n.d.
n.d.
n.d.
19
23
27
20
24
>20000
28
21
165
n.d.
25
35
2540
29
[a] Determined by biochemical FGFR-1 activity inhibition assay. [b] Determined by biochemical VEGFR-2 activity inhibition assay. See Supporting Informa-
#
tion for further details. *High ATP=2 mm, Low ATP=10 mm; n.d.=not determined.
Table 6. Selected RB substituents and their effect on FGFR-1 and VEGFR-2 potency as well as cellular activity, representative examples out of 368 analogues
in total.
Compd
RB
FGFR-1
IC50 [nm][a]*
FGFR
Prolif.
IC50 [nm][b]
VEGFR-2
IC50 [nm][c]#
Compd
RB
FGFR-1
IC50 [nm][a]*
FGFR
Prolif.
IC50 [nm][b]
VEGFR-2
IC50 [nm][c]#
18
30
31
10
36
48
66
846
388
35
36
37
95
20
40
21
127
74
221
795
114
1.1
2.2
5930
32
221
n.d.
n.d.
38
29
370
231
33
34
20
18
22
197
39
40
287
37
n.d.
166
n.d.
527
217
1100
[a] Determined by biochemical FGFR-1 activity inhibition assay. [b] Inhibitory potency of endothelial cell growth stimulated by FGFb. [c] Determined by bio-
chemical VEGFR-2 activity inhibition assay. See Supporting Information for further details. *High ATP=2 mm, #Low ATP=10 mm; n.d.=not determined.
decreased the FGFR potency (Table 6). The most promising
substituents were the aminopyrrolidinone 33 and the piperazi-
none 38 (Figure 4).
In parallel, examination of the crystal structure with docked
compound 18 revealed that the cavity around C6 of the pyrro-
lotriazine core structure could tolerate an additional substitu-
ent with polar or lipophilic properties. C6-substituted ana-
logues have their FGFR-1 activity significantly increased both
in the biochemical and cellular assays (Table 7). Lipophilic sub-
stituents can improve potency via hydrophobic interactions
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
5
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
with polar substituents (e.g., 41, 50, 51, 54, 55, 60–62, 74) also
demonstrated high FGFR potency, thus corroborating the pre-
vious modeling hypothesis (Figure 6). Unfortunately, combin-
ing the side chains at C6 with the piperazine tail group led to
a decrease in permeability and an increase in efflux (e.g., 41–
43, 49–52, 68; Table 7) relative to the lead structure 18.
As next step, the non-basic groups at C7 were combined
with a variety of C6 substituents, hoping to overcome the per-
meation/efflux issue, in line with our hypothesis that basic
amine moieties should be avoided. The combination of small
ether side chains at the 6-position with the 7-piperazinonyl-
methyl residue afforded the most promising compounds.
Among these derivatives, rogaratinib (75; Table 8) was chosen
as the clinical candidate. Rogaratinib showed improved FGFR
potency both in the biochemical and cellular proliferation
assay, as well as a significantly enhanced permeation profile
over the lead candidate 18. Rogaratinib inhibits FGFR1–4
kinase activity in the nanomolar range. In HUVEC proliferation
Figure 4. Docking images of the tail groups of compound 18 (A) and two
non-basic cyclic amide moieties (B and C) capable of maintaining the hydro-
gen bonding donor interaction to Glu571 as proposed by modeling. Predict-
ed hydrogen bonds are shown as yellow broken lines. The carboxylate
moiety of Glu571 is shown without a semitransparent grey surface in the
upper right-hand corner.
with the protein (Figure 5, supported, for example, by com-
pounds 42–49 in Table 7). On the other hand, compounds
Table 7. Representative selection of derivatives bearing side chains at C6 of the pyrrolotriazine and effect on FGFR potency and permeation (Caco-2).
Compd
RC
FGFR-1
IC50 [nm][a]
FGFR
Prolif.
IC50 [nm][b]
Caco-2
Papp(A!B)
Compd
RC
FGFR-1
IC50 [nm][a]
FGFR
Prolif.
IC50 [nm][b]
Caco-2
Papp(A!B)
[nmsꢁ1]/ER[c]
[nmsꢁ1]/ER[c]
41
42
43
44
ꢁCH2OH
ꢁCH2OMe
ꢁCH2OEt
ꢁCH2OPr
3.3
0.9
0.7
1.3
37
27
10
32
9/35
8/33
12/27
n.d.
60
61
62
63
ꢁCH2NH2
ꢁCH2NHMe
ꢁCH2NHEt
ꢁCH2NMe2
3.1
3.0
6.9
22
255
255
213
n.d.
n.d.
n.d.
n.d.
n.d.
45
46
47
48
ꢁCH2OBu
18
1.3
0.9
0.7
n.d.
121
57
n.d.
n.d.
n.d.
n.d.
64
65
66
67
3.2
1.7
2.1
8.6
85
n.d.
n.d.
n.d.
n.d.
ꢁCH2OiPr
107
157
307
110
ꢁCH2NEt2
49
50
ꢁCH2OCH2CF3
1.6
2.8
55
8/9
68
69
ꢁCH2NHCH2CH2OH
1.1
3.4
365
171
<2/>23
ꢁCH2OCH2CH2NH2
375
<5/>9
n.d.
51
52
53
ꢁCH2OCH2CH2OH
ꢁCH2OCH2CH2OMe
0.9
2.1
0.5
78
<1/>168
4/61
70
71
72
5.1
2.8
2.1
305
96
n.d.
n.d.
n.d.
140
865
n.d.
83
54
55
56
57
58
59
ꢁCH2OCH2CO2H
ꢁCH2OCH2CONH2
ꢁCH2OPh
ꢁMe
5.2
1.0
61
3.7
2.2
26
2740
n.d.
n.d.
47
25
126
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
73
74
ꢁCH2NHCH2CO2H
ꢁCH2NHAc
13
1.0
360
n.d.
n.d.
n.d.
ꢁEt
ꢁCl
[a] Determined by biochemical FGFR-1 activity inhibition assay, ATP=2 mm. [b] Inhibitory potency of endothelial cell growth stimulated by FGFb. [c] Trans-
port rate of compound across Caco-2 cell monolayer; Papp(A!B) =transport in apical to basolateral direction; ER=efflux ratio (A!B/B!A). See Supporting
Information for further details; n.d.=not determined.
&
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
6
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 5. Side view of ATP binding site in FGFR-1 (green C atoms, semitrans-
parent grey surface) with docked compound 43 (pink C atoms). Predicted
hydrogen bonds are shown as yellow broken lines. Front parts of the pro-
tein have been removed for clarity. The ethoxy group is predicted to make
beneficial interactions with a hydrophobic patch, which is formed by the b-
sheet groove of the P-loop (also called glycine-rich loop) and the side chains
of Phe489 and Val492.
Figure 6. Side view of ATP binding site in FGFR-1 (some selected parts of
the protein are shown as a stick model with green C atoms and a semitrans-
parent grey surface including yellow labels, the rest as a solid grey surface)
with docked compound 55 (stick model with pink C atoms). Predicted hy-
drogen bonds are shown as yellow broken lines. Front parts of the protein
have been removed for clarity. The primary amide is predicted to make hy-
drogen bonds to the side chains of Asn628 and Asp641. These two side
chains and the backbone carbonyl group of Arg627 form a triad of potential
hydrogen bond acceptors available for interaction with suitably placed hy-
drogen bond donors of inhibitors.
assays, rogaratinib potently inhibits FGFb-stimulated HUVEC
proliferation and demonstrates selectivity against VEGF-stimu-
lated growth (VEGF/FGFb selectivity ratio=28). Rogaratinib has
a good pharmacokinetic profile after intravenous and oral ap-
plication. The A!B permeation is 130 nmsꢁ1 with a low efflux
ratio (ER=4). In vivo, rogaratinib showed low clearance and a
moderate bioavailability in both rat and dog (Table 9).
Synthesis of rogaratinib (BAY 1163877)
Rogaratinib was prepared from the corresponding 6-methoxy-
methylpyrrolotriazine 83 (Scheme 1). It was planned to intro-
duce the oxo-piperazine group by reductive amination and the
Table 8. FGFR potency, VEGFR-2 selectivity, and DMPK parameters of rogaratinib (75).
FGFR-1 (high ATP*)[a]
FGFR-1 (low ATP#)[a]
12
15
<1
19
33
FGFR-2 (low ATP#)[b]
Biochemical assay IC50 [nm]
FGFR-3 (low ATP#)[c]
FGFR-4 (low ATP#)[d]
VEGFR-2 (low ATP#)[e]
120
Inhib. [%]
Kd [nm]
@100 nm 75
FGFR-1
97
97
96
99
19
1.6
5.0
7.8
7.6
–
Competition binding assay[f]
FGFR-2
FGFR-3
FGFR-4
VEGFR-2
FGFb-stimulated HUVEC proliferation[g]
VEGF-stimulated HUVEC proliferation[h]
Selectivity ratio: VEGF/FGFb
16 nm
453 nm
28
Proliferation assay IC50
[a] Determined by biochemical FGFR-1 activity inhibition assay. [b] Determined by biochemical FGFR-2 activity inhibition assay at MerckMillipore (see Sup-
porting Information for further details). [c] Determined by biochemical FGFR-3 activity inhibition assay. [d] Determined by biochemical FGFR-4 activity in-
hibition assay. [e] Determined by biochemical VEGFR-2 activity inhibition assay. [f] Active site-directed competition binding assay at DiscoverX.[10] [g] Inhibi-
tory potency of endothelial cell growth stimulated by FGFb. [h] Inhibitory potency of endothelial cell growth stimulated by VEGF. See Supporting Informa-
#
tion for further details. *High ATP=2 mm, Low ATP=10 mm.
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
7
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
change with n-butyllithium. Quenching with paraformaldehyde
at room temperature led to hydroxymethylpyrrole 81 in 69%
yield (three steps). Treating the pyrrole 81 with a hydrogen
chloride solution in dioxane not only cleaved the tert-butylcar-
bamate group, but also converted the hydroxymethyl group
into the corresponding chloromethyl 82. This very reactive in-
termediate 82 could be directly converted into the methoxy-
methylpyrrolotriazine 83 in a one-pot, two-step reaction se-
quence by treating 82 with methanol, followed by formami-
dine acetate and potassium phosphate at reflux temperature
(49%, two steps).
Table 9. DMPK parameters of rogaratinib (75).
i.v.[a]
p.o.[a]
0.5 mgkgꢁ1
5 mgkgꢁ1
CLbl
Vss,pl
t1/2
0.78 Lhꢁ1 kgꢁ1
0.54 Lkgꢁ1
0.9 h
AUCnorm
t1/2
F
0.96 kghꢁ1 Lꢁ1
rat
4.5 h
46%
0.5 mgkgꢁ1
1 mgkgꢁ1
CLbl
Vss,pl
t1/2
0.36 Lhꢁ1 kgꢁ1
1.2 Lkgꢁ1
3.1 h
AUCnorm
t1/2
F
1.2 kghꢁ1 Lꢁ1
dog
3.8 h
35%
[a] 5 mLkgꢁ1
; vehicle: plasma/DMSO=99:1 (rat i.v.: 15 min infusion);
In parallel, the boronic acid 89 was synthesized in five linear
steps (Scheme 2). The synthesis started with the hydrogenation
of nitro anisole 84 to afford aniline 85 using palladium on acti-
vated charcoal under an atmosphere of hydrogen. Aniline 85
PEG400/ethanol/water/HCl (1m)=20:5:74.5:0.5 (dog i.v.: 15 min infusion);
PEG400/ethanol/water=40:10:50 (rat p.o.: gavage); PEG400/ethanol/
water=20:5:75 (dog p.o.: gavage); CLbl =clearance blood, Vss,pl =volume
of distribution at steady state in plasma, t1/2 =half-life, AUCnorm =dose-
normalized (1 mgkgꢁ1) area under the curve, F=absolute oral bioavaila-
bility.
Scheme 2. a) H2, Pd/C, THF, RT; b) 1. NaNO2, HCl, NaOAc, 08C, 2. KCS2OEt,
water, 808C, 3. KOH, EtOH, reflux; c) bromoacetaldehyde diethylacetal,
Cs2CO3, DMF, RT, 66% (three steps); d) polyphosphoric acid, chlorobenzene,
reflux, 49%; e) 1. nBuLi, THF, ꢁ708C, then B(OiPr)3, 2. aq. NaOH (2n), 80%.
was converted into the corresponding diazonium salt, which
was treated with a hot aqueous solution of potassium ethyl-
xanthogenate. Finally, saponification with potassium hydroxide
led to the desired thiophenol 86, which was immediately alky-
lated with bromoacetaldehyde diethylacetal in the presence of
cesium carbonate to obtain the more stable thioether 87 in an
overall yield of 66% from nitroanisole 84. Thioether 87 was cy-
clized to the corresponding benzothiophene 88 using poly-
phosphoric acid in chlorobenzene at elevated temperature
(49% yield). Finally, deprotonation of benzothiophene 88 at C2
with n-butyllithium in THF, reaction with triisopropylborate and
subsequent hydrolysis afforded boronic acid 89 in a good yield
of 80%.
Scheme 1. a) HCl, dioxane, 908C, 41%; b) chlorosulfonylisocyanate, acetoni-
trile, DMF, 08C, 96%; c) NBS, DMF, ꢁ58C, 89%; d) 1. MeMgBr, 2-Me-THF,
ꢁ608C, 2. nBuLi, ꢁ608C, 3. paraformaldehyde, RT, 69%; e) 4m HCl/dioxane,
RT; f) formamidine acetate, K3PO4, MeOH, reflux, 49% (two steps).
benzothiophene via a Suzuki coupling. This synthetic route
was realized as depicted in Schemes 1–3, and comprises ten
linear steps in a convergent synthesis.[11] The pyrrolotriazine
core structure 83 was prepared in 6 linear steps (Scheme 1).
The synthesis started with an acid-catalyzed condensation re-
action of the commercially available 2,5-dimethoxytetrahydro-
furan (76) with tert-butyl carbazate (77) affording, after crystal-
lization, the protected aminopyrrole 78 in a yield of 41%.
Introduction of the nitrile group at C1 by electrophilic sub-
stitution using chlorosulfonylisocyanate followed by intramo-
lecular elimination afforded cyanopyrrole 79 (96%). Bromina-
tion at C4 of pyrrole 79 with N-bromosuccinimide was per-
formed at 08C to increase the desired regioselectivity and pre-
vent overreaction (dibromination), thus leading to bromo-pyr-
role 80 in a yield of 89%. The hydroxymethyl group was then
installed in a subsequent three-step one-pot reaction. At
ꢁ608C, the NH of bromide 80 was first deprotonated with
methylmagnesium bromide, followed by lithium–bromide ex-
The final assembly started with a Vilsmeier formylation of 83
at C7 to afford aldehyde 90 (26%, Scheme 3). The moderate
yield of the reaction was mainly due to the strong acidic reac-
tion conditions which led to the partial cleavage of the me-
thoxy group. Aldehyde 90 was then reacted with N-bromosuc-
cinimide at 08C to afford C5-bromo analogue 91 in 94% yield.
The coupling of bromide 91 with the boronic acid 89 under
thoroughly optimized Suzuki–Miyaura conditions afforded the
corresponding benzothiophene 92 in a yield of 51%. The best
coupling conditions were obtained with the X-Phos precata-
lyst.[12] Other catalysts, such as the corresponding S-Phos pre-
&
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
8
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Keywords: cancer · structure-based design · fibroblast growth
factor receptor (FGFR) · inhibitors · medicinal chemistry
[1] D. M. Ornitz, N. Itoh, WIREs Dev. Biol. 2015, 4, 215–266.
[2] R. Ronca, A. Giacomini, M. Rusnati, M. Presta, Exp. Opin. Ther. Targets
2015, 19, 1361–1377.
[3] a) S. Saichaemchan, W. Ariyawutyakorn, M. Varella-Garcia, Curr. Mol.
Med. 2016, 16, 40–62; b) N. Hallinan, S. Finn, S. Cuffe, S. Rafee, K.
O’Byrne, K. Gyle, Cancer Treat. Rev. 2016, 46, 51–62; c) C. Hierro, J.
Rodon, J. Tabernero, Semin. Oncol. 2015, 42, 801–819; d) L. Korsensky,
D. Ron, Semin. Cell Dev. Biol. 2016, 53, 101–114; e) G. Liang, G. Chen, X.
Wei, Y. Zhao, X. Li, Cytokine Growth Factor Rev. 2013, 24, 467–475;
f) B. V. S. S. Kumar, L. Narasu, R. Gundla, R. Dayam, J. A. R. P. Sarma, Curr.
Pharm. Des. 2013, 19, 687–701; g) M. V. Dieci, M. Arnedos, F. Andre, J. C.
Soria, Cancer Discovery 2013, 3, 264–279; h) J. Wesche, K. Haglund,
E. M. Haugsten, Biochem. J. 2011, 437, 199–213.
Scheme 3. a) POCl3, DMF, 608C, 26%; b) NBS, DMF, 08C; 94%; c) 89, X-Phos
precatalyst, CsF, THF, water, 608C, 51%; d) piperazinone, NaBH(OAc)3, AcOH,
MeOH, 608C, 14%.
[4] P. Gavine, L. Mooney, E. Kilgour, A. P. Thomas, K. Al-Kadhimi, S. Beck, C.
Rooney, T. Coleman, D. Baker, M. J. Mellor, A. N. Brooks, T. Klinowska,
Cancer Res. 2012, 72, 2045–2056.
[5] a) J. Tabernero, R. Bahleda, R. Dienstmann, J. R. Infoante, A. Mita, A. Ital-
iano, E. Calvo, V. Moreno, B. Adamo, A. Gazzah, B. Zhong, S. J. Platero,
J. W. Smit, K. Stuyckens, M. Chatterjee-Kishore, J. Rodom, V. Peddareddi-
gari, F. R. Luo, J.-C. Soria, J. Clin. Oncol. 2015, 33, 3401–3408; b) R. Dien-
stmann, R. Bahleda, B. Adamo, J. Rodon, A. Varga, A. Gazzah, S. Platero,
H. Smit, T. Perera, B. Zhong, K. Stuyckens, Y. Elsayed, C. Takimoto, V. Ped-
dareddigari, J. Tabernero, F. Roger Luo, J. C. Soria, American Association
for Cancer Research (AACR) 106th Annual Meeting, April 5–9, 2014
(San Diego, CA, USA), Abstract CT326; c) P. R. Angibaud, L. Mevellec, G.
Saxty, C. Adelinet, R. Akkari, V. Berdini, et al., American Association for
Cancer Research (AACR) 106th Annual Meeting, April 5–9, 2014 (San
Diego, CA, USA), Abstract 4748.
catalyst, gave lower yields or no conversion. The optimization
of the base was also important to prevent proto-deboronation
of 89 during the reaction. Cesium fluoride was identified as
the most suitable base for our purpose. Finally, reductive ami-
nation of aldehyde 92 with piperazine-2-one in the presence
of sodium triacetoxyborohydride as reducing agent led to ro-
garatinib (75) in 14% yield.
[6] A. Hillisch, N. Heinrich, H. Wild, ChemMedChem 2015, 10, 1958–1962.
[7] M. Mohammadi, S. Froum, J. M. Hamby, M. C. Schroeder, R. L. Panek,
G. H. Lu, A. V. Eliseenkova, D. Green, J. Schlessinger, S. R. Hubbard,
EMBO J. 1998, 17, 5896–5904.
[8] R. Abel, T. Young, R. Farid, B. J. Berne, R. A. Friesner, J. Am. Chem. Soc.
2008, 130, 2817–2831.
Conclusions
Using a computational chemistry approach to generate lead
compounds, we have presented the discovery of rogaratinib
(75), a novel small-molecule inhibitor of FGFR1–4. Rogaratinib
has good DMPK properties. It demonstrated tumor growth re-
duction in pre-clinical models bearing different FGFR altera-
tions both in mono- and combination therapy.[13] Rogaratinib is
currently being evaluated in phase I clinical trials (NCT
01976741).
[9] T. Young, R. Abel, B. Kim, B. J. Berne, R. A. Friesner, Proc. Natl. Acad. Sci.
USA 2007, 104, 808–813.
[10] M. A. Fabian, W. H. Biggs III, D. K. Treiber, C. E. Atteridge, M. D. Azi-
mioara, M. G. Benedetti, T. A. Carter, P. Ciceri, P. T. Edeen, M. Floyd, J. M.
Ford, M. Galvin, J. L. Gerlach, R. M. Grotzfeld, S. Herrgard, D. E. Insko,
M. A. Insko, A. G. Lai, J.-M. Lꢁlias, S. A. Mehta, Z. V. Milanov, A. M. Velas-
co, L. M. Wodicka, H. K. Patel, P. P. Zarrinkar, D. J. Lockhart, Nat. Biotech-
nol. 2005, 23, 329–336.
[11] D. Brohm, M. Hꢁroult, M.-P. Collin, W. Hꢀbsch, M. Lobell, K. Lustig, S.
Grꢀnewald, U. Bçmer, V. Voehringer (Bayer Pharma Aktiengesellschaft),
Int. PCT Pub. No. WO2013087578, 2013.
[12] T. Kinzel, Y. Zhang, S. L. Buchwald, J. Am. Chem. Soc. 2010, 132, 14073–
14075.
Experimental Section
All experimental data are shown in the Supporting Information.
[13] a) M. Hꢁroult, P. Ellinghaus, C. Sieg, D. Brohm, S. Grꢀnewald, M.-P. Collin,
U. Bçmer, M. Lobell, W. Hꢀbsch, M. Ocker, S. Ince, R. Jautelat, H. Hess-
Stumpp, M. Brands, K. Ziegelbauer, American Association for Cancer Re-
search (AACR) 106th Annual Meeting, April 5–9, 2014 (San Diego, CA,
USA), Abstract 1739; b) M. Hꢁroult, M. Ocker, C. Kopitz, D. Zopf, A. Hꢅ-
gebarth, K. Ziegelbauer, S. Ince, P. Ellinghaus, American Association for
Cancer Research (AACR) 107th Annual Meeting, April 18–22, 2015 (Phil-
adelphia, PA, USA), Abstract 772; c) M.-P. Collin, M. Lobell, D. Brohm, W.
Hꢀbsch, R. Jautelat, S. Jaroch, M. Brands, M. Hꢁroult, S. Grꢀnewald, U.
Bçmer, H. Hess-Stumpp, K. Ziegelbauer, American Association for
Cancer Research (AACR) 108th Annual Meeting, April 16–20, 2016 (New
Orleans, LA, USA), Abstract 4332.
Acknowledgements
The authors thank their co-workers for their dedication and skill-
ful work. The contributions by service functions of the various in-
stitutes are also gratefully acknowledged. Furthermore, we thank
the clinical development team for continuous support of this
project.
Manuscript received: November 16, 2017
Revised manuscript received: January 20, 2018
Conflict of interest
Accepted manuscript online: && &&, 0000
Version of record online: && &&, 0000
The authors declare no conflict of interest.
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
9
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

FULL PAPERS
M.-P. Collin,* M. Lobell, W. Hꢀbsch,
D. Brohm, H. Schirok, R. Jautelat,
K. Lustig, U. Bçmer, V. Vçhringer,
M. Hꢁroult, S. Grꢀnewald, H. Hess-Stumpp
In phase 1: We disclose the discovery
and chemical structure of rogaratinib
(BAY 1163877), a highly potent and se-
lective pan-FGFR inhibitor currently
being assessed in clinical trials for the
treatment of cancer. We describe the
structure–activity relationship and the
pharmacokinetic profile of the benzo-
thiophenyl-pyrrolotriazine structure
class.
&& – &&
Discovery of Rogaratinib (BAY
1163877): a pan-FGFR Inhibitor
&
ChemMedChem 2018, 13, 1 – 10
www.chemmedchem.org
10
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!