Potent and Orally Bioavailable Inverse Agonists of RORγt Resulting from Structure-Based Design

Article abstract of DOI:10.1021/acs.jmedchem.8b00783

Retinoic acid receptor related orphan receptor γt (RORγt), has been identified as the master regulator of T_H17-cell function and development, making it an attractive target for the treatment of autoimmune diseases by a small-molecule approach. Herein, we describe our investigations on a series of 4-aryl-thienyl acetamides, which were guided by insights from X-ray cocrystal structures. Efforts in targeting the cofactor-recruitment site from the 4-aryl group on the thiophene led to a series of potent binders with nanomolar activity in a primary human-T_H17-cell assay. The observation of a DMSO molecule binding in a subpocket outside the LBD inspired the introduction of an acetamide into the benzylic position of these compounds. Hereby, a hydrogen-bond interaction of the introduced acetamide oxygen with the backbone amide of Glu379 was established. This greatly enhanced the cellular activity of previously weakly cell-active compounds. The best compounds combined potent inhibition of IL-17 release with favorable PK in rodents, with compound 32 representing a promising starting point for future investigations.

Full text of DOI:10.1021/acs.jmedchem.8b00783

Subscriber access provided by AUSTRALIAN NATIONAL UNIV
Article
Structure-based design leads to potent and
orally bioavailable inverse agonists of ROR#t
Frank Narjes, Yafeng Xue, Stefan von Berg, Jesper Malmberg, Antonio Llinas, Roine I.
Olsson, Johan Jirholt, Hanna Grindebacke, Agnes Leffler, Nafizal Hossain, Matti Lepistö,
Linda Thunberg, Hanna Leek, Anna Aagaard, Jane McPheat, Eva L. Hansson, Elisabeth
Back, Stefan Tångefjord, Rongfeng Chen, Yao Xiong, Ge hongbin, and Thomas Hansson
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00783 • Publication Date (Web): 10 Aug 2018
Downloaded from http://pubs.acs.org on August 10, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Structureꢀbased design leads to potent and orally
bioavailable inverse agonists of RORγt
†
§
†
†
‡
Frank Narjes,* Yafeng Xue, Stefan von Berg, Jesper Malmberg, Antonio Llinas, Roine I.
†
ǂ
ǂ
ǂ
†
†
Olsson, Johan Jirholt, Hanna Grindebacke, Agnes Leffler, Nafizal Hossain, Matti Lepistö,
≠
≠
§
¶
¶
Linda Thunberg, Hanna Leek, Anna Aagaard, Jane McPheat, Eva L. Hansson, Elisabeth
¶
§
^
^
^
†
Bäck, Stefan Tångefjord, Rongfeng Chen, Yao Xiong, Ge Hongbin, Thomas G. Hansson
†
‡
ǂ
Medicinal Chemistry, DMPK and Bioscience, Respiratory, Inflammation and Autoimmunity,
IMED Biotech Unit, AstraZeneca, Gothenburg, SEꢀ43183 Mölndal, Sweden.
§
¶
Structure, Biophysics &FBLG and Mechanistic Biology and Profiling, Discovery Sciences,
IMED Biotech Unit, AstraZeneca, Gothenburg, SEꢀ43183 Mölndal, Sweden.
≠
Early Product Development, Pharmaceutical Sciences, IMED Biotech Unit, AstraZeneca,
Gothenburg, SEꢀ43183 Mölndal, Sweden.
^
Pharmaron Beijing Co., Ltd., Taihe Road BDA, Beijing, 100176, P.R. China
Eꢀmail: Frank.Narjes@astrazeneca.com
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT: Retinoic acid receptorꢀrelated orphan receptor γt (RORγt), has been identified as
the master regulator of T 17 cell function and development, making it an attractive target for the
H
treatment of autoimmune diseases by a small molecule approach. Herein we describe our
investigations on a series of 4ꢀarylꢀthienyl acetamides, which were guided by insights from Xꢀ
ray coꢀcrystal structures. Efforts in targeting the coꢀfactor recruitment site from the 4ꢀaryl group
on the thiophene led to a series of potent binders, with nanomolar activity in a primary human
TH17 cell assay. The observation of a molecule of DMSO binding in a subꢀpocket outside the
LBD inspired the introduction of an acetamide into the benzylic position of these compounds.
Hereby, a hydrogen bond interaction of the introduced acetamide oxygen with the backbone
amide of Glu379 was established. This greatly enhanced the cellular activity of previously
weakly cell active compounds. The best compounds combined potent inhibition of ILꢀ17 release
with favorable PK in rodents with compound 32 representing a promising starting point for
future investigations.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Introduction
The nuclear receptor retinoic acid receptorꢀrelated orphan receptor gamma(RORγ, NR1F3,
RORc) exists in two isoforms, with one isoform (RORγ, RORc1) widely expressed in a variety
of tissues, and the expression of the second isoform (RORγt, RORc2) restricted to the thymus
1
and cells of the immune system. RORγt is the key transcription factor for function and
+
development of CD4 Tꢀhelper 17 (T 17) cells and related ILꢀ17 producing immune cells such
H
1
ꢀ5
as innate lymphoid 3 cells and γδ T cells. TH17 cells are characterized by the production of the
cytokines ILꢀ17A, ILꢀ17F and ILꢀ22 and play a key role in host defence against extracellular
6
ꢀ7
pathogens. Dysregulation of the T 17 pathway has been implicated in the pathology of a
H
ACS Paragon Plus Environment
2

Page 3 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
variety of autoimmune disorders such as psoriasis, Crohn’s disease, multiple sclerosis and
7
ꢀ12
rheumatoid arthritis.
Antibodies targeting ILꢀ17, its receptor ILꢀ17RA, or the cytokines
involved in the inflammatory pathway such as ILꢀ23, have shown clinical efficacy in psoriasis
1
3ꢀ16
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
and related autoimmune disorders.
Nuclear receptors are ligandꢀinducible transcription factors, whose function can be modulated
by binding of small molecules to the ligandꢀbinding domain (LBD) of the receptor, thereby
inducing conformational changes to the coꢀfactor recruitment binding site, ultimately affecting
17ꢀ19
gene transcription.
RORγt has been shown to be active in the absence of ligands, but sterol
2
0ꢀ23
derivatives were described to affect receptor function, and to control T 17 differentiation.
.
H
The crystal structures of two sterol derivatives, agonist 25ꢀhydroxychesterol and inverse agonist
digoxin, provided insights into the molecular mechanisms leading to the different functional
2
4ꢀ25
effects.
Efforts to discover orally bioavailable inverse agonists of RORγt, as an alternative to antibody
26ꢀ29
treatment, has generated a variety of different structural classes.
Compounds affecting the
receptor by binding to the LBD are not expected to exhibit isoform selectivity, since the two
isoforms of RORγ differ only at their Nꢀterminus, and share the same DNA and ligand binding
3
0ꢀ40
41
42ꢀ43
domains.
Recent clinical candidates from these efforts include GNEꢀ3500, GSK805,
3
0
44
GSK2981278 and VTPꢀ43742 (Figure 1). GSK2981278, whose structure has not been
disclosed, was developed for the topical treatment of psoriasis, but failed to show efficacy in a
4
5
phase 1 study, whereas Vitae Pharmaceuticals recently reported positive results for VTPꢀ43742
from a Phase 2a study for the same indication. Oral administration of over four weeks was found
4
6
to be safe and efficacious, resulting in a 24% reduction in Psoriasis Area Severity Index.
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
GNE-3500
GSK805
VTP-43742
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Selection of published inverse agonists of RORγt.
We have recently disclosed a series of benzoxazepine based modulators, which we eventually
39
abandoned due to metabolic instability of the compounds. The hit identification approaches we
4
7
conducted to find new starting points included Xꢀray based fragment screening and the
characterization of compounds from the literature. Among these were derivatives 1 and 2,
48ꢀ49
previously described by scientists from GSK.
Table 1. Profile of compounds 1ꢀ3.
Cl
O
SO Et
SO Et
2
Cl
2
Cl
S
O
S
O
X
N
N
H
H
1
2
(X = N)
(X = CH)
3
a
b
c
IC SPA
IC FRET
50
IC ILꢀ17 Cell
50
50
d
e
Cmpd
logD
clogP
LLE
[
ꢀM]
[ꢀM] (% eff.)
0.022 (ꢀ87)
0.023 (ꢀ82)
0.18 (ꢀ96)
[ꢀM] (% eff.)
0.18 (ꢀ75)
0.006 (ꢀ85)
>3
1
2
0.009
0.002
0.11
>4.2
4.5
5.5
6.0
4.0
2.6
2.6
3.0
3
>4.1
a
b
Displacement of tritiated ligand from human RORγt LBD (geometric mean n ≥ 2). Inhibition or activation of the
recruitment of SRC1ꢀderived coactivator peptide (NCOA1 aa 677ꢀ700) to human RORγt LBD (geometric mean n ≥
c
2
); negative % efficacy (% eff.) values denote inverse agonism. Inhibition of ILꢀ17 production from human primary
T 17 cells (geometric mean of n ≥ 3); negative values of % efficacy signify inhibition of ILꢀ17 production.
H
ACS Paragon Plus Environment
4

Page 5 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
d
e
Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at pH 7.4. Ligand lipophilic efficiency
defined as SPA pIC – clogP.
5
0
Both compounds were potent binders in a radioligand binding assay (SPA) and behaved as
inverse agonists in a FRET assay, showing functional suppression of the recruitment of a
coactivator peptide derived from steroid receptor coactivator 1 (SRC1) to the RORγt LBD. The
activity of 2 observed in the FRET assay was about 10ꢀfold lower compared to the binding assay,
which we ascribed to the conditions of the FRET assay, placing a limit on the lowest IC that
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
can be determined at around 0.02 ꢀM. They efficiently suppressed ILꢀ17A production from
primary human T 17 cells, with thiophene 2 being 30ꢀfold more potent. Compound 2 is highly
H
lipophilic, and pharmacokinetic properties in mouse were characterized by high plasma clearance
4
9
and low oral bioavailability. Compound 3 was prepared as part of an early effort to understand
the SAR and reduce lipophilicity. It was about 60ꢀfold less active in the SPA assay, but we were
intrigued by the observation, that despite the lack of the benzoate moiety, 3 behaved as an
inverse agonist in the FRET assay. Even though 3 exhibited subꢀmicromolar binding potency,
and efficacy in the FRET assay, it was not able to suppress ILꢀ17 release from primary human
TH17 cells.
In this paper we describe our SAR efforts around 3, which were guided by insights from Xꢀray
crystal structures on the binding mode of compounds as the series evolved. These resulted in a
class of compounds, which efficiently suppressed ILꢀ17 production in primary human cells and
had favorable PK properties in rodent species.
Results and Discussion
We succeeded in soaking compound 1 into the apoꢀcrystals of RORγtꢀLBD, tethered to an
2
4, 39
SRC2 derived peptide, representing the agonist state of the receptor (Figure 2).
Recently, the
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
50ꢀ52
coꢀcrystal structures of 1 and of a closely related agonist, were published.
Despite the
different crystallization conditions, the key interactions of the compounds inside the ligand
binding domain are very similar. These include the key hydrogen bond interaction of the sulfone
oxygens with Arg367 and Gln286, and the amide moiety with the backbone carbonyl of Phe377.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
53
The amide bond is coꢀplanar with the thiazole ring, with a stabilizing 1,5ꢀO···S interaction. The
metaꢀchloro substituent of the 4ꢀaryl residue points to Ser404 (3.5 Å), and the side chain of
Met365 has moved significantly (about 3.5 Å) with respect to the apoꢀstructure.
Figure 2. a) Xꢀray structure of 1 (green) binding inside the RORγt LBD (gray) (2.30 Å, PDB:
5
NI5). Compound 1 and key residues of the RORγt LBD are shown as sticks. b) Overlay of 1
(green) and 25ꢀhydroxycholesterol (25ꢀHC, grey), showing the interaction of 25ꢀHC with the
water molecule (dotted sphere) and the TyrꢀHisꢀlock.
An overlay with 25ꢀhydroxycholesterol (Figure 2b) shows that benzoyl group at the 5ꢀposition
of the thiazole fills the hydrophobic pocket near helix 12 only partially, and does not interact
with the soꢀcalled TyrꢀHisꢀlock, composed of His479 on helix 11 and Tyr502 on helix 12. These
two residues which form part of the activation function 2 domain (AF2), the interaction site for
ACS Paragon Plus Environment
6

Page 7 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
coꢀregulatory proteins. We observed a conserved water near Trp317 and the HisꢀTyrꢀlock, which
in the structure of the agonist 25ꢀhydroxycholesterol interacts with the 25ꢀhydroxy group, and
2
4
stabilizes the agonist conformation. Stabilization of the AF2 domain by hydrophobic
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
interactions was suggested as an explanation for the agonism of thiazole amides and keto
50, 52
amides.
For the inverse agonism of 1, a ‘waterꢀtrapping mechanism’ was proposed, whereby
the release of the water molecule trapped at the AF2 domain in the hydrophobic environment is
51
energetically favorable, leading to destabilization of helix 12.
The Xꢀray structures described above were not published when we started our work, but the
results in Table 1 pointed to a key role of the lipophilic benzoyl group for potency and activity in
the cell assay. Assuming that the binding mode of 3 would closely resemble that of 1, we
hypothesized that it should be possible to combine cell potency and achieve drugꢀlike properties
by using appropriate substituents appended to the thiopheneꢀlinked aryl group of 3.
A focused library of compounds around the aryl substituent in the 4ꢀposition of the thiophene
was prepared (Table 2). Addition of a cyano group in the 5ꢀposition of the 4ꢀaryl group of 3
resulted in a 2ꢀfold gain in potency. Moving the chlorine to the 2ꢀposition gave equipotent 5,
which exhibited a lower partition coefficient compared to 4, whereas compound 6, with the
cyano group into the 4ꢀposition, was less active. Trisubstituted derivative 7, bearing a methoxy
group in place of the chlorine, gained another 2ꢀfold in potency. Removal of the fluorine was
tolerated, leading to 8, with an improved LLE value compared to 3.
Another set of interesting data were provided by the orthoꢀsubstituted pyridyl analogues 9 and
1
0. Methoxy derivative 9 exhibited diminished binding activity with respect to 3, but maintained
functional activity. A bulkier benzyl ether as in 10 led to a 20ꢀfold gain in binding affinity,
however 10 behaved as an agonist in the FRET assay, efficiently enhancing recruitment of the
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
coꢀactivator peptide. This indicated that our initial hypothesis of being able to modulate receptor
function by proper substitution from the aryl group was at least partly correct. Attempts to
further increase polarity of 9 as in pyrimidine 11 resulted in a significant loss in potency.
Compound 12 combined the potency enhancing nitrile moiety with the methoxy group and
similar to 8.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 2. Initial SAR around the 4ꢀaryl group.
a
b
IC SPA
[ꢀM]
IC FRET
50
5
0
c
d
Cmpd
Ar
logD / LLE
[ꢀM] (% eff.)
3
4
5
6
3ꢀClꢀPh
0.11
0.18 (ꢀ96)
>4.1 / <2.9
>4 / <3.3
3.8 / 3.5
3ꢀClꢀ5ꢀCNꢀPh
2ꢀClꢀ5ꢀCNꢀPh
2ꢀClꢀ4ꢀCNꢀPh
0.05
0.05
1.16
0.09 (ꢀ72)
0.06 (ꢀ78)
0.78 (ꢀ90)
3.7 / 2.2
*
7
8
0.02
0.02
0.03 (ꢀ84)
3.9 / 3.8
3.6 / 4.1
*
0.03 (ꢀ74)
0.11 (ꢀ82)
*
9
0.41
0.02
1.85
3.6 / 2.8
>4 / <3.7
2.8 / 2.9
*
*
10
+95% @ 3.3 ꢀM
1
1
1
2
0.70 (ꢀ63)
*
0.03
0.04 (ꢀ64)
3.5 / 4.0
a
b
Displacement of tritiated ligand from human RORγt LBD (geometric mean n ≥ 2). Inhibition or activation of the
recruitment of SRC1ꢀderived coactivator peptide (NCOA1 aa 677ꢀ700) to human RORγt LBD (geometric mean n ≥
2); negative % efficacy (% eff.) values denote inverse agonism; positive values denote agonism and were
ACS Paragon Plus Environment
8

Page 9 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
c
determined at a single concentration. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at
d
pH 7.4. Ligand lipophilic efficiency defined as SPA pIC – logD.
50
Concurrent with the optimization of the aryl group, we also explored the possibility to lower
lipophilicity by introducing changes to the aryl sulfone moiety. We considered groups, which
could maintain some of the interactions in the arginineꢀrich region. A sulfoxide and especially
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
54
the sulfoximine moiety fulfilled these requirements, but the corresponding compounds 13 and
1
4 lost 40ꢀ and 10ꢀfold activity in the SPA assay, respectively, whereas in the FRET assay a 10ꢀ
or 4ꢀfold loss was observed (Table 3). A methylsulfonamide as in 15 did not offer any reduction
in lipophilicity, but maintained potency with respect to 12. Replacing the phenyl sulfone
aromatic ring with a sulfonylated piperidine as in 16 also led to a considerable loss of activity,
which we believe to be due to the different geometries of the saturated ring compared to the
aromatic ring.
Table 3. SAR around the sulfonyl head group.
a
b
Cmpd
X
R
IC SPA [ꢀM]
IC FRET [ꢀM]
5
0
50
c
d
logD / LLE
(% eff.)
13
N
CH
N
1.29
0.34
0.07
3.25
0.44 (ꢀ74)
3.4 / 2.5
2.8 / 3.7
3.5 / 3.6
3.2 / 2.3
*
*
*
*
1
1
4
5
0.12 (ꢀ72)
0.04 (ꢀ69)
3.25 (ꢀ87)
16
N
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
b
Displacement of tritiated ligand from human RORγt LBD (geometric mean n ≥ 2). Inhibition or activation of the
recruitment of SRC1ꢀderived coactivator peptide (NCOA1 aa 677ꢀ700) to human RORγt LBD (geometric mean n ≥
2
); negative % efficacy (% eff.) values denote inverse agonism; positive values denote agonism and were
c
determined at a single concentration. Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
d
pH 7.4. Ligand lipophilic efficiency defined as SPA pIC – logD.
50
To better understand the tight SAR around the sulfone moiety, we performed a solvent analysis
55
of the apo and the complex structures with 1, using 3DꢀRISM. The results indicated the
presence of several water molecules in this arginine subꢀpocket. Four of these are hydrophobic
(unstable) and were replaced by the ethyl group, whereas two hydrophilic (stable) water
molecules were each replaced by one of the oxygen atoms of the sulfone upon binding of the
ligand (Figure 3). Accordingly, the two oxygen atoms retain favorable polar interactions with the
protein. As a result, the ethyl sulfone group seemed to provide the best configuration to replace
the different solvent positions and provide binding affinity of the ligand. This would also explain
the loss of potency for sulfoxide 13 and sulfoximine 14.
ACS Paragon Plus Environment
10

Page 11 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 3. Solvent analysis of RORγtꢀLBD in complex with 1 using 3DꢀRISM. The arylsulfone
group of 1 and key residues of the RORγt LBD are shown as sticks. Solvent positions in the apoꢀ
pocket are shown as spheres (hydrophobic/unstable waters in orange, hydrophilic/stable ones in
cyan).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
We tested some of the more active compounds for their ability to inhibit ILꢀ17A production in
primary human T 17 cells (Table 4). We found that all compounds were inactive at
H
concentrations of up to 3.3 ꢀM, apart from 7 and 10. In line with their behavior in the functional
FRET assay, inverse agonist 7 inhibited ILꢀ17A production with an IC50 of about 0.7 ꢀM,
whereas agonist 10 enhanced it to 38% over the baseline of the cell assay at a concentration of
®
3
.3 ꢀM. We noticed some effect on cell viability in the CellꢀTiterꢀGlo readꢀout for both
compounds at 3 ꢀM. This made the interpretation of the cell data problematic, as there was less
than 10ꢀfold difference to the IC observed for suppression of ILꢀ17A production. Based on the
5
0
good binding potency and functional activity in the FRET assay, we were expecting to see a
more profound response on ILꢀ17 production, and investigated if the lack of activity was due to
issues related to cell permeability or efflux. Results from the Caco2ꢀassay indicated that
compared to 2, all our inverse agonists showed improved permeability. Compound recovery was
only 40ꢀ50%, which was due to their relatively high lipophilicity (see Table 1ꢀ3), low solubility
and high affinity to plasma proteins. This can lead to unspecific binding or precipitation, making
data interpretation difficult. Compounds 7, 8 and 12 had similar permeability and moderate
efflux, whereas sulfonamide 15 demonstrated lower permeability and higher efflux. We also
measured the intrinsic permeability of these compounds in an assay, where the Caco2 cells are
incubated with a cocktail of compounds inhibiting the most common transporters present in the
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
56
cell, including PgP. The intrinsic permeability values for all inverse agonists, apart from 2,
were high, indicating that these compounds should be able to reach the interior of the cells.
Based on the data, low permeability or efflux do not seem explain the lack of cell potency. Our
conclusion was that the binding or coꢀfactor recruitment assays using the truncated form of the
receptor in a simplified system, are not always predictive of the more complex biology of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
inhibition of ILꢀ17 production in a T 17 cell, where RORγ is in its native state and interacting
H
with its endogenous ligands and coꢀfactors.
Table 4. Cell assay activity and physicochemical data for selected analogues.
b
c
a
a
Caco2 Papp
AtB / BtA/ efflux
Caco2
d
IC ILꢀ17 Cell
[ꢀM] (% eff.)
CTG [ꢀM]
hPPB
Solubility
[ꢀM]
50
Cmpd
Intrinsic P0
(1E cm/s)
e
(% eff.)
ꢀ6
ꢀ6
(% free)
(
1E cm/s)
0
.006 ± 0.0003
ꢀ85)
<0.001 / 0.6 / >324
52% rec.
2
7
8
> 3 (40%)
<0.5
37
<0.2
0.17
<0.10
<0.40
(
1
.8 / 7.1 / 3.9
0.7 ± 0.5 (ꢀ57)
>3 (60%)
ꢀ
4
1% rec.
1
.6 / 12 / 7.6
>3.3
+38% @ 3.3 ꢀM
>3.3
79
ꢀ
0.10
0.03
0.09
0.43
0.42
0.34
4
1% rec.
ꢀ
1
0
>3 (30%)
2
.6 / 17.1 / 5.8
4% rec.
.65 / 26 / 40
5% rec.
12
15
ꢀ
ꢀ
65
4
0
>3.3
67
<0.1
0.34
5
a
Inhibition of ILꢀ17 production from human primary T 17 cells (geometric mean of n ≥ 3); negative values of %
H
efficacy signify inhibition of ILꢀ17 production; positive values a potentiation under the assay conditions. Effect of
®
compound on cell viability or proliferation was assessed in the same assay with CellꢀTiterꢀGlo (CTG) measuring
b
c
total ATP levels. Permeability measured in Caco2 cells at pH=7.4/7.4. Permeability measured in Caco2 cells
d
incubated at pH=6.5 with a cocktail of inhibitors of common transporters. Binding to human plasma protein
e
determined by equilibrium dialysis. Determined from DMSO stock solutions in aqueous buffer at pH 7.4.
ACS Paragon Plus Environment
12

Page 13 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
We could obtain Xꢀray coꢀcrystal structures of the inverse agonist 8 and agonist 10 by soaking
the compounds into the apo crystals of RORγtꢀLBD (Figures 4), proving that the compounds
bind to the receptor and are genuine modulators of RORγt.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Overlay of 1 and 8 (2.45 Å, PDB: 5NI7), 8 and 10 (1.94 Å, PDB: 5NI8), respectively,
with key interactions to the RORγtꢀLBD. a) Compounds 1 (green), 8 (cyan) and key residues of
the RORγ LBD are shown as sticks. b) Compounds 8 (cyan) and 10 (magenta) and key residues
of the RORγt LBD are shown as sticks.
Both compounds showed an almost perfect overlap for the sulfone head group and the core
heterocycle. The same key interactions as described above for 1, were established. For
compound 8, the nitrile on the aryl moiety pointed towards the sideꢀchain oxygen of Ser404,
indicative of a rather long hydrogenꢀbond interaction of 3.2 Å, similar to the metaꢀCl of the 4ꢀ
arylygroup of 1. The 2ꢀmethoxy group is too short to interact with the AF2 domain, whereas the
benzyl group in agonist 10 filled the pocket around the HisꢀTyr lock, replacing the conserved
water observed in the structures of 1 and 8. This hydrophobic interaction is thought to stabilize
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the agonist conformation of the AF2 domain and similar observations were published recently
3
9, 50, 52, 57, 58
for other RORγ agonists.
Based on the Xꢀray structure of 10, we expected that increasing the size of the alkoxy group
would eventually lead to a clash with the HisꢀTyrꢀlock, and therefore to inverse agonism.
Increasing the size of the alkoxy substituent from methoxy as in 9 to ethoxy as in 17 and
subsequently to phenoxy as in 19 resulted in an increase in binding affinity and a gradual switch
in function from inverse agonism to agonism in the FRET assay (Table 5), with 19 behaving like
benzyl ether 10. Pushing the phenyl group out further by chain elongation of 10 did not lead to
inverse agonism, but derivatives 20 and 21 behaved as agonists. We hypothesized that the
flexibility of these substituents allowed them to be accommodated in the binding pocket,
avoiding the steric clash with the AF2 domain. Rigidifying the propyl linker of 21 with a double
bond supported this hypothesis, since cinnamyl ether 22 was a potent inverse agonist in the
FRET assay.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Introduction of a substituent to the paraꢀposition of the benzyl group of 10 led to the expected
change in the mode of action. Ester 23 as well as benzonitrile 24 behaved as inverse agonists.
Moving the cyano residue to the metaꢀposition of the phenyl ring as in 25 led to a switch to
agonistic behavior in the FRET assay.
Table 5.
[a]
[b]
[c]
IC SPA
IC FRET [ꢀM] IC ILꢀ17 Cell
CTG [ꢀM] RH
1
2
50
50
50
Cmpd
R
R
[d]
[
ꢀM]
(% eff.)
[ꢀM] (% eff.)
(% eff)
Clint
9
OMe
OBn
H
H
0.41
0.11 (ꢀ82)
n.t.
n.t.
13
10
0.020
+95% @ 3.3 ꢀM
+38% @ 3.3 ꢀM
>3 (30%)
128
ACS Paragon Plus Environment
14

Page 15 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
2
2
2
2
2
2
2
7
8
9
0
1
2
3
4
5
6
OEt
H
H
H
H
H
H
H
H
H
H
H
H
H
0.15
0.08 (ꢀ72)
ꢀ47% @ 1 ꢀM
n.a.
>10
28
OPr
0.059
0.046
0.029
0.017
0.011
0.014
0.01
ꢀ13% @ 3.3 ꢀM
+42% @ 3.3 ꢀM
+84% @ 3.3 ꢀM
+17% @ 3.3 ꢀM
0.029 (ꢀ90%)
0.021 (ꢀ98)
3 (40)
3 (50)
3 (70)
3 (30)
3 (40)
>10
80
OPh
n.a.
196
125
104
45
O(CH
O(CH
)
)
Ph
Ph
+26% @ 1ꢀM
>3.3 ꢀM
2
2
2
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
OCH
OCH
OCH
OCH
CH=CHPh
0.025 (ꢀ78)
0.035 (ꢀ92)
0.10 ± 0.08
2
2
2
2
Phꢀ4ꢀCO
Phꢀ4ꢀCN
Phꢀ3ꢀCN
Me
>300
286
281
28
2
0.03 (ꢀ57)
>1 (50)
0.3 (50)
>10
0.019
0.088
0.007
0.033
0.018
+112% @ 3.3 ꢀM +5% @ 1 ꢀM
OPhꢀ4ꢀCN
OCH Phꢀ4ꢀOCF
OPhꢀ4ꢀOCF
ꢀ18% @ 3.3 ꢀM
0.03 (ꢀ68)
0.19 ± 0.13 (ꢀ79)
0.025 ± 0.09 (ꢀ68)
0.144 ± 0.12 (ꢀ65)
0.10 ± 0.0 (ꢀ64%)
27
2
3
3 (40%)
>10
40
2
2
8
9
3
ꢀ19% @ 3.3 ꢀM
0.05 (ꢀ44)
20
OCH
OCH
CF
CF
>10
17
2
2
3
30
3
CN 0.002
0.03 (ꢀ49%)
0.065 ± 0.03 (ꢀ66%) >10
7.5
a
b
Displacement of tritiated ligand from human RORγt LBD (geometric mean n ≥ 2). Inhibition or activation of the
recruitment of SRC1ꢀderived coactivator peptide (NCOA1 aa 677ꢀ700) to human RORγ LBD (geometric mean n ≥
2
); negative % efficacy (% eff.) values denote inverse agonism; positive values denote agonism and were
c
determined at a single concentration. Inhibition of ILꢀ17 production from human primary T 17 cells (geometric
H
mean of n ≥ 3); negative values of % efficacy signify inhibition of ILꢀ17 production; positive values a potentiation
under the assay conditions. Effect of compound on cell viability or proliferation was assessed in the same assay with
d
−1
CellꢀTiterꢀGlo® measuring total ATP levels; n.a.: not active at 3.3 ꢀM; n.t.: not tested. Rat hepatocyte (ꢁL min
−
1
(
million cells) ) metabolic intrinsic clearance.
Functional behavior in the FRET assay was mirrored in the cell assay, but the observed activity
on ILꢀ17A secretion was again obstructed by effects on cell viability for several compounds.
Exceptions were cinnamyl ether 22 and ester 23 were at least a 100ꢀfold window between cell
activity and toxicity was observed.
The added lipophilicity, compounds 17ꢀ25 have measured logD7.4 of around 4 or greater, came
at the price of decreased metabolic stability, as the data from rat hepatocytes demonstrated.
Removal of the metabolically labile methylene in benzyl ether 24 led to phenylether 26 with a
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
lower logD7.4 value (3.6) and improved stability compared to either 19 or 24. Compound 26
behaved as an inverse agonist in the cell assay, where it exhibited an IC of 190 nM and no sign
5
0
of cytotoxicity. Further SAR around the 4ꢀphenyl substituent eventually identified the
trifluoromethoxy group as an optimal group in the paraꢀposition. Benzylether 27 combined
potent inhibition of ILꢀ17A production with a 100ꢀfold window over cytotoxicity and reasonable
stability in rat hepatocytes. The corresponding phenylether 28 showed a further improvement in
metabolic stability. It behaved like a weak partial inverse agonist in the FRET assay, as 26, but
suppressed ILꢀ17A secretion with an IC of 144 nM in the cell assay, with no effect on cell
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
viability.
We were also able to identify an inverse agonist with a short alkoxy sideꢀchain by converting
the ethoxy group of 17 to a bulkier trifluoroethoxy group. Compound 29 was a potent inhibitor
of ILꢀ17 production with no sign of cytotoxicity. Adding the nitrile group to the 5ꢀposition
resulted in 30, showing a 10ꢀfold increase in binding potency, an improvement in metabolic
stability and a slight increase in cell potency. The maximum achievable inhibition in the cell
assay reached a plateau at 66%, and most compounds in this series, apart from ester 23, did not
completely abolish ILꢀ17A production.
We succeeded in soaking agonist 25 into the apoꢀcrystal (Figure 5), and the only difference
with respect to 10 concerned the orientation of the benzyl moiety. In compound 25, the phenyl
ring was tilted in comparison to 10, filling a space between helices 11 and 7 and stabilizing the
agonist form. Structurally related inverse agonists such as 23, 24 or 27 would be expected to
push against the AF2ꢀdomain in either conformation, which was confirmed by the coꢀcrystal
structure of 23, described below. The sidechain of 30 is too short to interact directly with the
ACS Paragon Plus Environment
16

Page 17 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
helix 11/12 interface, but it seemed to disturb the conserved water, when we modelled it into the
5
1
structure. Release of this water might destabilize the AF2 domain, as recently described for 1.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Ligand interaction near the HisꢀTyr lock based on the structure of 25 (1.82 Å, PDB:
NIB). Compounds 10 (magenta), 25 (yellow) and 30 (grey, modelled based on 8) and key
5
residues of the HisꢀTyr lock of RORγt LBD are shown as sticks. The ligands are only shown
partially for clarity. The conserved water position is shown as dotted sphere in red, and the
position of the CF3 in 30 is shown as dotted sphere in cyan.
The coꢀcrystal structure of 23 with the native RORγt construct in the absence of coꢀactivator
3
9
peptide, showed that the major interactions of 23 were the same as illustrated for 10 and 25,
except for the benzoate ester group. In the agonist form, this group would clash with His479, as
hypothesized above and indicated in Figure 6a. Consequently, part of helix 11 and the entire
helix 12 are disordered and are missing from the electron density in the structure.
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. Ligand interaction compound 23 in RORγtꢀLBD (2.37 Å, PDB: 6FGQ). a) Overlay of
agonists 10 (magenta) and 25 (yellow) with inverse agonist 23 (orange). Key residues of the Hisꢀ
Tyr lock of RORγt LBD are shown as sticks. The ligands are only shown partially for clarity.
The conserved water position is shown as a dotted sphere in red. Helix 12 and part of helix 11
are disordered (indicated in the dashed circle) when compound 23 is bound. The putative steric
clash between 23 and the side chain of His479 (based on 10 and 25) is shown as a red star
(
~1.4Å in between). b) Overlay of RORγt complexes of digoxin (PDB 3B0W, blue), 10
magenta), and 23 (orange).
(
An overlay with the inverse agonist structure of digoxin showed a steric clash of ester 23 with
His479, which in the digoxin structure is pulled in by the ligand, establishing a hydrogenꢀbond to
2
5
one of its sugar hydroxy groups. This in turn “pushes” away part of helix 11 and all
downstream residues. Overall, 23 seems to achieve the inverse agonist effect in a similar way,
but without pulling in His479. Although we were not able to obtain a structure, we believe that
phenylethers 26 and 28 exert their mode of action by a related mechanism. In summary, cell
active agonists and two different types of inverse agonists can be obtained from the same
ACS Paragon Plus Environment
18

Page 19 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
template by the introduction of relatively minor structural changes to the alkoxy substituent.
3
9, 52, 57ꢀ59
Similar results have been reported for a variety of diverse RORγ modulators.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7: Structure overlay of RORγt in complex with compound 8 and PDB 5G45 (fragment
bound). a): Compound 8 is shown as sticks with carbons in cyan, and the bound fragment and
DMSO are shown with carbons in blue (DMSO depicted as sticks and spheres). b): surface view
to illustrate the channel (“entrance”) to the LBP. The tethered SRC2 peptide is shown as a grey
helix.
At this point, we had identified compounds 27ꢀ30, which demonstrated potent inhibition of ILꢀ
1
7A release in primary human cells and showed good metabolic stability. However, they were
still rather lipophilic, and attempts to introduce further polarity had failed. Fortunately, we were
4
7
able to capitalize on a finding from our Xꢀray fragment screen. In a number of structures we
had noted the presence of a DMSO site, which was near a conserved water position. The binding
of DMSO occurred in a subꢀpocket (a channel or “entrance” to the pocket) and its oxygen atom
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
formed a hydrogen bond to the backbone NH of Glu379 (Figure 7). We saw this as an
opportunity to endow our inhibitors with polar functionality. Modeling indicated that the
carbonyl oxygen atom of an acetamide attached to the benzylic position of the phenylacetamide
moiety would be able to mimic that interaction. We incorporated the acetamide into a set of
compounds containing diverse 4ꢀaryl residues, and found that this change was tolerated in all
compounds investigated, leading to similar activity in SPA and FRET assays with respect to the
parent compounds (Table 6). A reduction in lipophilicity was also achieved, accompanied by a
higher stability in rat hepatocytes. To our surprise we found that compounds 31 and 32, the
analogues of 12 and 7, respectively, also acquired activity in the cell assay, efficiently
suppressing ILꢀ17 production without an effect on cell viability. Compound 32 was about one
order of magnitude more active than 31, with an IC of 57 nM. The unsubstituted amine 33 was
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50
significantly less potent in the binding and the cell assay, indicating that the acetyl residue in 32
had formed the desired interaction.
Table 6. Incorporation of an acetamide residue in 4ꢀarylꢀthienyl acetamides.
[a]
[b]
[c]
[d]
IC SPA
IC FRET
IC ILꢀ17 Cell
[ꢀM] (% eff.)
Clint
LogD
/LLE
50
50
50
Cmpd Ar
R
[e]
[
ꢀM]
[ꢀM] (% eff.)
RH
*
3
1
NHAc 0.014
NHAc 0.009
0.067 (ꢀ53)
0.99 (ꢀ55)
4
5
3.3 / 4.8
*
*
32
0.063 (ꢀ72)
0.452 (ꢀ62)
0.057 (ꢀ80)
1.69 (ꢀ62)
3.3 / 4.8
3.2 / 3.9
3
3
NH
2
0.075
11
ACS Paragon Plus Environment
20

Page 21 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3
3
4
5
NHAc 0.066
0.055 (ꢀ52)
0.64 (ꢀ57)
10.4
3.4 / 3.7
*
O
CF3O
N
*
*
NHAc 0.006
NHAc 0.007
0.075 (ꢀ60)
0.072 (ꢀ60)
0.032 (ꢀ77)
0.049 (ꢀ66)
9.1
<1
4.6 / 3.7
3.7 / 4.4
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
36
*
3
3
3
7
8
9
NHAc 0.007
NHAc 0.003
NHAc 0.27
+23% @ 3.3 ꢀM +10% @ 3.3 ꢀM 24
>4 / <4.1
4.7 / 3.9
4.7 / 1.9
O
O
CF3O
CF3O
N
*
*
0.034 (ꢀ72)
3.53 (ꢀ90)
0.011 (ꢀ81)
0.023 (ꢀ83)
12
18
N
a
b
Displacement of tritiated ligand from human RORγt LBD (geometric mean n ≥ 2). Inhibition or activation of the
recruitment of SRC1ꢀderived coactivator peptide (NCOA1 aa 677ꢀ700) to human RORγt LBD (geometric mean n ≥
2
); negative % efficacy (% eff.) values denote inverse agonism; positive values denote agonism and were
c
determined at the maximum tested concentration. Inhibition of ILꢀ17 production from human primary T 17 cells
H
(
geometric mean of n ≥ 3); negative values of % efficacy signify inhibition of ILꢀ17 production; positive values a
potentiation under the assay conditions. Effect of compound on cell viability or proliferation was assessed in the
d
−1
same assay with CellꢀTiterꢀGlo® measuring total ATP levels. Rat hepatocyte intrinsic clearance (ꢁL min (million
−1
e
e
cells) ). Distribution coefficient between 1ꢀoctanol and aqueous phosphate buffer at pH 7.4. Ligand lipophilic
efficiency defined as SPA pIC – logD.
5
0
This assumption was ultimately confirmed by a coꢀcrystal structure, which defined the (R)ꢀ
isomer of 32 as the bound species (Figure 8). The interactions formed inside the RORγt LBD are
very similar to those described for compound 8 (see Figure 4), with the addition of a hydrogen
bond formed between the carbonyl oxygen of the acetamide moiety and the NH of Glu379. As
seen in the structure of 8, the methoxy residue is too short and does not interact with the helix
11/12 interface. There is also no conformational change of the receptor apparent upon binding of
the acetamide with respect to 8, and based on this structure we cannot explain the improvement
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
in cell potency of 31 and 32. Hydrogen bonding to Glu379 was also described recently for 6ꢀ
3
5, 60
substituted quinolines and bisaryl carboxamides.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Comparison of RORγtꢀcomplex with 32 (1.84 Å, PDB: 6ESN, carbons in yellow)
with the observed DMSO (PDB 5G45, carbons in blue) site in Xꢀray screen. a): The position of
DMSO as observed in the fragment structure. b): The binding mode of 32 depicted in the same
orientation as in (a). The newly introduced carbamate group superimposes well with the DMSO
position and engages in an Hꢀbond with the backbone NH of Glu379.
Different from 31 and 32, which bear a simple aryl group in the 4ꢀposition of the thiophene,
acetamides 34ꢀ36, with the more elaborate benzyl or aryl ether linkages, did not show a gain in
cell potency with respect to their parent matched pairs. Cell potency remained unaltered for 35
and 36, whereas for 34 a slight 3ꢀfold loss in potency, compared to 26, was observed. The
acetamide substitution also did not seem to affect the function of the compounds, since benzyl
ether 37 still behaved as an agonist.
ACS Paragon Plus Environment
22

Page 23 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Compound 35 was separated into its enantiomers 38 and 39, with the former being about 100ꢀ
fold more active in the binding and the FRET assay. Based on an Xꢀray structure of 32, 38 was
assigned the (R)ꢀconfiguration, and 39 the (S). Surprisingly, the latter inhibited ILꢀ17 release in
the cell assay with the same potency as 38. We hypothesized that the stereocenter might
racemize under the assay conditions, and investigated for 38 and 39. The compounds were stable
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
at pH 5, 7.4 and 8 at 37 °C over two days, but racemized in the medium of the T 17 assay,
H
explaining the activity observed for 39 (Table 7). In view of these results, it cannot be excluded
that some racemization occurs also under the conditions of other in vitro or in vivo assays, and
this should be considered when using these compounds.
Table 7. Racemization of 37 and 38 under different conditions.
[a]
[a]
[a]
[b]
Time
h]
pH 5
38
pH 7.4
38
pH 8
38
Xꢀvivo
38
[
39
39
39
39
1ꢀ9
99:1
99:1
99:1
nd
99:1
99:1
99:1
99:1
nd
99:1
99:1
99:1
99:1
nd
99:1
99:1
99:1
2
4
9
4
8
6
99:1
99:1
nd
99:1
99:1
nd
99:1
99:1
nd
60:40
50:50
50:50
80:20
70:30
55:45
a
The buffer solution (10 ml) and DMSO (0.5 ml) were added to the compound (5 mg), and the resulting slurry
was stirred at 37 °C. A sample of the slurry (20 ꢀl) was taken at the indicated time points, diluted with methanol (80
b
ꢀ
l) and analyzed by supercritical fluid chromatography (SFC). The compound (5 mg) was dissolved in the Xꢀvivo
medium (5 ml) and stirred at 37 °C. A sample of the solution (100 ꢀl) was added to dichloromethane (1 ml) and
shaken for 10 seconds. The organic layer was transferred to a vial, evaporated to about 100 ꢀl and analyzed by SFC
(
Column: Chiralpak AS (3 ꢀm, 150 x 4.6 mm I.D.), mobile phase: 25% ethanol in carbon dioxide at 120 bar;
temperature 40 °C, flow rate 3.5 ml/min (Rt = 1.6 min and Rt = 2.2 min)).
1
2
Based on their favorable cell potency and metabolic stability in rat hepatocytes, we further
profiled compound 28 and three matched pairs with and without the acetamide residue (Table 8).
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
The introduction of the acetamide residue raised the polar surface area by about 30 Å , lowered
logP and increased ligand efficiency (see also Table 6).
The compounds tested were not active on RORα in a timeꢀresolved FRET assay conducted in
agonist mode. When tested in the inverse agonist mode, compounds 30, 32 and 38 showed
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
micromolar activity, with 36 being the most active with an IC of 0.6 ꢀM. No activity could be
50
detected for 27, 32, 36 and 38 in a RORβ binding assay. Solubility was low for all analogues,
except for 32, where the acetamide improved solubility significantly with respect to 7. All
compounds showed very high affinity for plasma proteins, and moderate stability in human
hepatocytes. Relatively high activity on the hERG channel was noticed for 7 and 30, when tested
at a fixed concentration of 11 ꢀM, which was decreased to acceptable levels in the corresponding
acetamides 32 and 36.
Table 8. Profile of selected compounds.
NC
*
*
O
Ar
*
N
*
R
NHAc
28
H
7
NHAc
32
H
27
NHAc
38
H
30
NHAc
36
Cmpd
Mw
562.6
98
458.5
97
515.6
130
576.6
98
633.7
131
509.5
107
566.6
140
2
PSA (Å )
[b]
clogP /LLE
4.5 / 3.1
2.4 / 5.3
1.6 / 6.5
4.6 / 3.8
3.7 / 4.9
2.5 / 6.2
1.7 / 6.4
RORα ag. / inv. ag.
>
33 / >33
nd
nd
nd
>33 / 19 >33 / >33
>33 / 9.4
>10
>33 / 6.2
nd
>33 / 0.6
>10
[a]
IC [ꢀM]
50
[c]
RORβ IC [ꢀM]
>10
13
>10
0
5
0
hERG
% inhib. @ 11 ꢀM)
1
4
43
8
47
3
(
ACS Paragon Plus Environment
24

Page 25 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
[
d]
Solubility [ꢀM]
2.5
0.6
0.2
200
1.3
0.7
0.3
0.1
0.2
3.6
0.2
[e]
hPPB [%free]
<0.1
<0.02
0.06
[f]
Cl RH
int
2
0
11.3
5.6
5.0
5.3
40
8.6
8.9
8.5
7.5
5.9
<1
nd
ꢀ6
(
ꢀL/min/10 )
[f]
Cl HH
int
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1.9
ꢀ6
(
ꢀL/min/10 )
[g]
Caco2
Low
recovery
1.8 / 7.1 / 0.3 / 18 /
Low
<0.001 /
Low
AtB / BtA / efflux
nd
nd
ꢀ6
3.9
50
recovery 42 / >1500 recovery
(
10 cm/s)
[h]
Rat PK
Cl (ml/min/kg), t1/2 (h) 24, 1.6
V (L/kg), F (%)
5.5, 1.2
0.48, 85
92
2.5, 3.0
0.34, 18
19
46.5, 1.4
3.8, 10
30
12, 4.3
2.4, <1
<1.2
0.8, 4.4
0.23, 36
36
2.8, 45
67
ss
Fabs
a
Timeꢀresolved (TR)ꢀFRET assay in agonist/inverse agonist mode using a peptide derived from PGC1α 130ꢀ154.
b
d
c
Ligand lipophilic efficiency defined as SPA pIC50 – clogP. Scintillation proximity binding assay (SPA).
e
Determined from DMSO stock solutions in aqueous buffer at pH 7.4. Bindung to human plasma protein
f
−1
−1
determined by equilibrium dialysis. Rat or human hepatocyte intrinsic clearance (ꢁL min (million cells) ).
g
h
Permeability measured in Caco2 cells at pH=7.4/7.4. Compounds dosed to male Han Wistar rats (n = 2); dose iv
0
.5 mg/kg, solution in 5% DMSO, 95% SBEꢀBꢀCD (30% w/v) in water; dose oral 1 mg/kg, solution in 33% DMA,
33% TEG, 34% water or suspension in 0.5% HPMC, 0.1% Tween in water.
All compounds, apart from 36, were profiled for their pharmacokinetic behavior in Han Wistar
rats (Table 8). Plasma clearances for the nonꢀacetamide containing compounds 7, 27, 28 and 30
decreased from high for the benzyl ether 27 (46.5 ml/min/kg) to very low for 7 and 30 (<1
ml/min/kg), in line with the intrinsic clearance observed in rat hepatocytes.
Compounds in this series were quite lipophilic, some of them reaching the upper limit of our
logD assay, making it difficult to explore correlations based on the LogD values. However, using
2
a calculated logP value (ClogP) a decent linear correlation (r =0.8) is found between metabolic
stability in rat and lipophilicity. The volume of distribution values were surprisingly low for
compounds 7, 30 and 32. For such lipophilic and neutral compounds a volume of around 1 L/kg
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
is expected. The reason for such low volumes of distribution are not known. Derivative 7 had the
highest oral bioavailability of the compounds tested, while 27, 28 and 30 demonstrated
encouraging oral bioavailability with fraction absorbed (Fabs) equal to or above 30%. Acetamide
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2 exhibited lower clearance in rat and in human hepatocytes, but also a significantly diminished
oral bioavailability compared to 7. The same was observed for the pair 27 and 38. Introduction of
the acetamide residue clearly had a positive impact on activity and stability, but the added
polarity had a negative impact on the oral bioavailability. Caco2 data for the matched pair 7 and
3
2 indicate, that compared to its precursor 7, 32 displayed significant efflux, which could explain
its decreased bioavailability.
Conclusion
In conclusion, we have obtained a series of inverse agonists of the nuclear receptor RORγt by
starting from the Xꢀray coꢀcrystal structure of the thienyl acetamide 1. Simplification of 2 led to
3
, and 8, both which was not active in a primary human T 17 cell assay, despite good binding
H
potency and functional activity in the FRET assay of 8. The SAR investigations concentrated on
the aryl group in the 4ꢀposition of the thiophene, and were guided by available Xꢀray coꢀcrystal
structures of modulators bound in the RORγt LBD. Functional activity in the cell assay seemed
to depend on how groups fill the volume towards the cofactor recruitment site. The insights
gained from these Xꢀray coꢀcrystal structures will therefore be useful for the design of novel
RORγt modulators. Larger aromatic groups as in 27 or 28, interacting directly with the helix
1
1/12 interface, or smaller trifluoroethyl group as in 29 or 30, proved to be effective in
suppressing ILꢀ17 production in the primary human cells. The compounds also exhibited good
pharmacokinetic profiles in rat.
ACS Paragon Plus Environment
26

Page 27 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In several Xꢀray structures we had observed a DMSO molecule establishing a hydrogenꢀbond
to the proteinꢀbackbone. Based on this observation, we designed an acetamide residue into our
molecules to mimic this interaction. This resulted in compounds, which were potent in the cell
assay, had lower lipophilicity and showed improved metabolic stability. However, the added
polarity impacted negatively on permeability, resulting in lower oral bioavailability in rats.
Fortunately, the acetamide substitution endowed weakly active molecules such as 7 with activity
in the cell assay, leading to a reduction in molecular complexity compared to 2 or 27. Compound
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2, devoid of the larger aromatic functional triggers, combined favorable cell potency with good
metabolic stability, selectivity against the hERG channel and the related ROR isoforms, and
encouraging oral bioavailability in rat. This key compound presented the starting point for
further investigations in this series, leading ultimately to the discovery of AZD0284 (structure to
be disclosed), which is currently in phase Ib clinical trials for the treatment of psoriasis.
Chemistry
For the broader investigation of the SAR of the 4ꢀthienyl aryl substituent, a library of
compounds were prepared by Suzuki Miyaura coupling using commercially available boronic
acids with the acetamide 43, which in turn is readily available from coupling of amine 41 with 4ꢀ
sulfonylꢀphenylacetic acid 42a (Scheme 1). Compound 43 also served as a precursor for the
preparation of two advanced intermediates, fluoropyridine 44 and boronic ester 45 (Scheme 2).
a
Scheme 1. Synthesis of diverse arylꢀsubstituted 4ꢀthienyl acetamides from intermediate 43
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
42a
b
40
41
43
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Ar
Ar
4
5
6
7
8
2-Cl-5-CN-Ph-
3-Cl-5-CN-Ph-
2-Cl-4-CN-Ph-
2-MeO-3-F-5-CN-Ph-
2-MeO-5-CN-Ph-
9
2-MeO-3-Py-
10 2-BnO-3-Py-
c
12 2-MeO-5-CN-3-Py-
17 2-EtO-3-Py-
4
- 12, 17
a
Reagents and conditions: (a) TFA, DCM; (b) EDC, 4ꢀDMAP, DCM, 53%; (c) ArB(OH) or
2
Arpin, PdCl (dppf), K CO , dioxane/H O (4:1), 80 °C.
2
2
3
2
Coupling of 43 with 2ꢀ(fluoropyridꢀ3ꢀyl)boronic acid or bisꢀpinacolatoꢀdiboron yielded the
advanced intermediates 44 and 45, respectively. Treatment of 44 with the appropriate alcohol
under basic conditions furnished compounds 18ꢀ25 and 29 (Scheme 2). The boronic ester 45 was
coupled with halopyridines 46bꢀd under palladiumꢀcatalyzed Suzuki reaction conditions to
provide compounds 26ꢀ28.
a
Scheme 2. Synthesis of arylꢀsubstituted 4ꢀthienyl acetamides from intermediates 44 and 45
ACS Paragon Plus Environment
28

Page 29 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
b
43
4
4
1
1
8 (R = Pr)
9 (R = Ph)
20 (R = (CH ) Ph)
23 (R = (CH Ph-CO Me
2 2
24 (R = CH Ph-4-CN)
2
2
2
25 (R = CH Ph-3-CN)
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
21 (R = (CH ) Ph
2 3
29 (R = OCH CF )
2
3
2
2 (R = (CH CH=CHPh)
2
c
43
d
4
5
2
6 (R = Ph-4-CN)
2
7 (R = CH Ph-4-OCF )
2
3
2
8 (R = Ph-4-OCF3)
4
4
4
6a (R = Ph-4-I, X = Br)
e
f
6b (R = Ph-4-CN, X = Br)
6c (R = CH Ph-4-OCF , X = I)
2
3
46d (R = Ph-4-OCF , X = I)
3
a
Reagents and conditions: (a) 2ꢀ(fluoropyridinꢀ3ꢀyl)boronic acid, PdCl (dppf), K CO ,
2
2
3
t
dioxane/H O (4:1), 100 °C, 64%; (b) ROH, KO Bu, THF, 80 °C; (c) PdCl (dppf), B Pin , KOAc,
2
2
2
2
dioxane, 100 °C, 54%; (d) 46b, c or d, PdCl (dppf), K CO , DME/H O (5:1), 130 °C,
2
2
3
2
t
microwave; (e) 46a: 4ꢀiodophenol, KO Bu, THF, microwave 140 °C, 41%; 46c: (4ꢀ
trifluoromethoxy)phenyl)ꢀmethanol, NaH, DMF, 80 °C, 49%; 46d: (4ꢀtrifluoromethoxy)phenol,
NaH, DMF, 80 °C, 70%; (f) CuCN, Lꢀproline, DMF, 100 °C, 52%.
Several compounds were prepared by coupling of 4ꢀarylꢀ2ꢀaminothiophens to the rightꢀhand
phenylacetic acid moiety, which is exemplified for compound 30 in Scheme 3. The synthesis of
compounds 11, and 13 ꢀ16 followed similar routes and is described in the supporting
information. The preparation of building block 48a followed the synthetic routes laid out in
Scheme 2. Removal of the Boc protecting group from 48a could be done with either TFA or with
TBSOTf. When we used TBSOTf less decomposition of the resulting aminothiophene 49a was
observed, which was used directly in the coupling step to give 30.
ACS Paragon Plus Environment
29