3-Substituted Quinolines as RORγt Inverse Agonists

Article abstract of DOI:10.1016/j.bmcl.2019.04.021

We have previously reported the syntheses of a series of 3,6-disubstituted quinolines as modulators of the retinoic acid receptor-related orphan receptor gamma t (RORγt). These molecules are potent binders but are high molecular weight and they exhibited poor solubility at pH 2 and pH 7. This manuscript details our efforts at improving physical chemical properties for this series of compounds by increasing the diversity at the 3-position (i.e. introducing heteroatoms and lowering the molecular weight). These efforts have led to molecules which are potent binders with improved solubility.

Full text of DOI:10.1016/j.bmcl.2019.04.021

Accepted Manuscript
3
-Substituted Quinolines as RORγt Inverse Agonists
Virginia M. Tanis, Hariharan Venkatesan, Maxwell D. Cummings, Michael
Albers, J. Kent Barbay, Krystal Herman, David A. Kummer, Cynthia Milligan,
Marina I. Nelen, Rachel Nishimura, Thomas Schlueter, Brian Scott, John
Spurlino, Ronald Wolin, Craig Woods, Xiaohua Xue, James P. Edwards, Anne
M. Fourie, Kristi Leonard
PII:
S0960-894X(19)30241-0
https://doi.org/10.1016/j.bmcl.2019.04.021
BMCL 26388
DOI:
Reference:
To appear in:
Bioorganic & Medicinal Chemistry Letters
Received Date:
Revised Date:
Accepted Date:
11 February 2019
5 April 2019
11 April 2019
Please cite this article as: Tanis, V.M., Venkatesan, H., Cummings, M.D., Albers, M., Kent Barbay, J., Herman, K.,
Kummer, D.A., Milligan, C., Nelen, M.I., Nishimura, R., Schlueter, T., Scott, B., Spurlino, J., Wolin, R., Woods,
C., Xue, X., Edwards, J.P., Fourie, A.M., Leonard, K., 3-Substituted Quinolines as RORγt Inverse Agonists,
Bioorganic & Medicinal Chemistry Letters (2019), doi: https://doi.org/10.1016/j.bmcl.2019.04.021
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

3
-Substituted Quinolines as RORt Inverse Agonists
a,
a,
b,
Virginia M. Tanis *, Hariharan Venkatesan *, Maxwell D. Cummings *, Michael
c
b
d
a
b
Albers , J. Kent Barbay , Krystal Herman , David A. Kummer , Cynthia Milligan , Marina I.
b
a
c
a
b
a
Nelen , Rachel Nishimura , Thomas Schlueter , Brian Scott , John Spurlino , Ronald Wolin ,
a
d
a
d
b
Craig Woods , Xiaohua Xue , James P. Edwards , Anne M. Fourie , Kristi Leonard
a
Discovery Product Development and Supply, Janssen Research and Development, 3210
Merryfield Row, San Diego, CA 92121, United States
b
Discovery Product Development and Supply, Janssen Research and Development,
Welsh and McKean Roads, Spring House, PA 19477, United States
c
Department of Research, Phenex Pharmaceuticals AG, Heidelberg, Germany
d
Discovery Immunology, Janssen Research and Development, 3210 Merryfield Row, San
Diego, CA 92121, United States
*
Corresponding authors. E-mail addresses: vtanis@its.jnj.com (V.M. Tanis),
hvenkate@its.jnj.com (H. Venkatesan), mcummin1@its.jnj.com (M.D. Cummings).
Declarations of interest: none.
Keywords:
RORt; retinoic acid-related orphan nuclear receptor gamma t; Inverse agonist; Th17; IL-
1
7
1

Abstract:
We have previously reported the syntheses of a series of 3,6-disubstituted quinolines as
modulators of the retinoic acid receptor-related orphan receptor gamma t (RORt). These
molecules are potent binders but are high molecular weight and they exhibited poor solubility at
pH 2 and pH 7. This manuscript details our efforts at improving physical chemical properties for
this series of compounds by increasing the diversity at the 3-position (i.e. introducing
heteroatoms and lowering the molecular weight). These efforts have led to molecules which are
potent binders with improved solubility.
The retinoic acid-related orphan receptors (RORs) are members of the nuclear receptor
family of transcription factors. This family is comprised of three members: ROR, ROR and
1
ROR, each with multiple isoforms, that are expressed in a variety of tissues. The RORt
isoform is expressed in a few distinct cell types of the immune system, and is the key
transcription factor driving differentiation of Th17 cells, and transcription of proinflammatory
1
,2,3a
cytokines IL-17A, IL-17F and IL-22.
RORt also drives transcription of IL-23R, through
which IL-23 signals to promote proliferation, maintenance and stabilization of Th17 cells, and
3
b
expression of IL-17A and additional effector cytokines in multiple cell types. The role of the
4
Th17 / IL-17 pathway in the pathogenesis of psoriasis has been amply demonstrated by the
5
efficacy of biologics targeting IL-23, IL-17A, or the IL-17RA receptor. Small molecule
modulation of RORt provides a potential opportunity for oral treatment of psoriasis, and other
2

6
7
8
autoimmune disorders such as multiple sclerosis, psoriatic arthritis and inflammatory bowel
9
disease.
The potential for therapeutic inhibition of Th17 differentiation and IL-17 production with
1
0-13
oral small molecules has resulted in intense drug discovery activity,
and entry of selected
previously described
1
2
14a-15a
molecules into clinical development. Reports from our laboratories
1
5a
discovery of quinoline-based RORt inverse agonists. As previously noted, quinoline 1 (Figure
®16
1
) is potent in a ThermoFluor (TF, K = 9.2 nM) thermal stabilization binding assay and
d
2
1
cellular reporter assay (IC = 73 nM (99% inhibition at 6 M)) and was found to be a full
50
inverse agonist, however, the equilibrium aqueous solubility of 1 at pH 2 and pH 7 is sub-
optimal (<1 M). Structure-activity relationships (SAR) and variations in the mode of
modulation (agonists, inverse agonists, and neutral antagonists) established during optimization
of the substituents at the 2-, 3-, 4- and 6-positions on the quinoline core have been
1
4a,15a
described.
The impact of modest changes at the quinoline 3-, and 6-positions can be seen in
15b
compound 2. The incorporation of an additional heteroaryl group at the tertiary alcohol carbon
of the 6-position and the replacement of the pyrazole moiety with a CF substituent resulted in a
3
reduction in lipophilicity, as measured by cLogP (6.0 vs 5.1), improved binding affinity, cellular
1
7a
potency and metabolic stability (defined using human and rat extraction ratios (hER / rER) as
high (ER > 0.7), moderate (ER 0.3 – 0.7) and low (ER < 0.3)) and a large increase in pH 2
solubility. Based on these findings we elected to explore additional changes at the quinoline 3-
and 6-positions with the goal of maintaining potency and improving solubility. Herein, we report
a series of compounds with changes to the substituents on the 3- and 6-positions of the quinoline
core of 1 and 2. This investigation led to compounds that are partial and full inverse agonists
1
8
with nanomolar potency and improved solubility.
3

Figure 1. Structures of advanced leads 1 and 2.
We prepared analogs with smaller and more polar groups at the 3-position of the
quinoline to lower the lipophilicity and improve the pH 2 and pH 7 solubility. Our syntheses of
3
-alkyl substituted congeners of 1 and 2 were designed to allow facile 6-position alteration.
Despite numerous approaches (not reported) to introduce the 3-position substituent at a late
stage, it was found that the routes shown in Schemes 1-4 that introduce this group early were
most useful. Using appropriately substituted commercial malonic acids we prepared 6-bromo-3-
1
9
alkyl-2,4-dichlorosubstituted quinolines as shown in Scheme 1. Alternatively a 6-iodo-4-
hydroxy-2(1H)-quinolone was prepared from the readily available 5-iodo anthranilate and the
alkyl substituent was incorporated via an aldol addition and subsequent Hantzsch ester reduction
to afford 3-alkyl-4-hydroxy-2(1H)-quinolones. The target 6-iodo-2,4-dichloro congener was then
1
9
prepared (Scheme 2) by treatment with POCl . A target 3-CH CF -2,4-dichloro quinoline
3
2
3
intermediate was separately prepared as outlined in Scheme 3. Acylation of a 5-
bromoanthranilate with 3,3,3-trifluoro-propanoyl chloride and subsequent Knoevenagel
condensation gave a 4-OH-2(1H)-quinolone which afforded the related dichloride upon reaction
19
with POCl3.
4

Scheme 1: The preparation of 3-alkyl-6-bromo quinolines. Reagents and reaction conditions: (a)
-methylpropanedioic acid, 2-(2-methylpropyl)-propanedioic acid or 2-cyclopentylpropanedioic
2
acid, POCl3.
Scheme 2: The preparation of 3-alkyl-6-iodo quinolines. Reagents and reaction conditions: (a)
AcCl, DCM, NaHCO followed by KOtBu, THF; (b) cyclopropaneacetaldehyde, 4,4-
3
difluorocyclohexaneacetaldehyde or tetrahydro-2H-thiopyran-4-acetaldehyde 1,1-dioxide,
Hantzsch ester, pyridine; (c) POCl3.
Scheme 3: Reagents and reaction conditions: (a) 3,3,3-trifluoro-propanoyl chloride, DCM,
NaHCO followed by KOtBu, THF; (b) POCl .
3
3
The conversion of the 6-Br/I quinolines to final target molecules was accomplished as
shown in Scheme 4. The 2-methoxy group is selectively introduced by treatment with NaOMe in
either MeOH or PhCH , and the 6-position side chain was incorporated by halogen-metal
3
2
0
19
exchange (n-BuLi, THF) and reaction with the ketone of choice to yield alcohols 3-9.
Scheme 4: The preparation of 3-alkyl quinolines 3-9. Reagents and reaction conditions: (a)
NaOMe, MeOH or PhCH ; (b) n-BuLi, THF followed by (2,6-dimethyl-3-pyridinyl)(1-methyl-
3
1
H-1,2,3-triazol-5-yl)-methanone.
Further modifications at the 3-position of the quinoline moiety, in the form of 3-alkoxy,
were obtained by the practical route presented in Scheme 5. 5-Bromo-anthranilate acylation with
benzyloxy-acetyl chloride followed by Knoevenagel condensation (KHMDS) provided the 3-
benzyloxy-4-OH-2(1H) quinolone. Halogenation (POCl ) and subsequent reaction with NaOMe
3
afforded the 2-methoxy-3-benzyloxy-4-chloro-quinoline which provided the target 6-position
5

2
0
substitution after metal-halogen exchange and reaction with desired ketones. The substituent at
the 3-position is added at the terminus of the sequence by 3-position de-benzylation followed by
19
a Mitsunobu reaction. This sequence was unsuccessful in the case of the desired 3-O-iPr
quinoline, therefore this entity was prepared as illustrated in Scheme 6 from isopropyl i-PrO-
1
9
acetate.
Scheme 5: General 3-alkoxy quinoline synthesis route. Reagents and reaction conditions: (a) 2-
amino-5-bromobenzoic acid methyl ester, Et N, DCM; (b) KHMDS, THF; (c) POCl , 2,6-
3
3
lutidine; (d) NaOMe, MeOH or PhCH ; (e) n-BuLi, THF followed by (2,6-dimethyl-3-
3
pyridinyl)(1-methyl-1H-1,2,3-triazol-5-yl)-methanone;
(f)
H2,
Pd/C,
MeOH;
(g)
cyclopropanemethanol, 2,2-difluorocyclopropanemethanol, tetrahydro-2H-pyran-4-ol, 2-
thiazolemethanol or 2-pyrimidinemethanol, DIAD, PPh , THF.
3
Scheme 6: The synthesis of 3-O-iPr quinoline 10. Reagents and reaction conditions: (a) NaOiPr,
iPrOH; (b) 2-amino-5-bromobenzoic acid methyl ester, KHMDS, THF; (c) POCl ; (d) NaOMe,
3
MeOH or PhCH ; (e) n-BuLi, THF followed by (2,6-dimethyl-3-pyridinyl)(1-methyl-1H-1,2,3-
3
triazol-5-yl)-methanone.
4
5
4
5
Our final targets were 3-CH -NR R amines and a 3-C(O)NR R amide. The amines were
2
prepared as shown in Scheme 7 via a benzylic bromide generated from a 3-methyl quinoline,
which was prepared in Scheme 1. Radical bromination of the methyl group (NBS, ABCN)
4
5
followed by reaction with the amine of choice (R NHR , iPr NEt, CH Cl ) provided a 6-bromo-
2
2
2
6

4
5
3
-CH NR R intermediate. Metal halogen exchange and subsequent reaction with the selected
2
2
0
19
ketone led to the target 3-aminomethyl quinolines. Compounds 28, 29 and 30 were prepared
in a similar fashion, as shown in Scheme 8, through the reaction of the lithiate with an
19
appropriate ketone where the azetidine ring nitrogen was protected with a Boc group.
Deprotection of the Boc group followed by treatment with acetyl chloride under basic conditions
provided the desired products.
4
5
Scheme 7: The preparation of 3-CH NR R quinolines. Reagents and reaction conditions: (a)
2
NBS, ABCN, CCl ; (b) 3-(trifluoromethyl)-azetidine, 4-fluoropiperidine, 4,4-difluoropiperidine,
4
4
-(trifluoromethyl)-piperidine or 4-(1H-pyrazol-1-yl)-piperidine, iPrNEt , DCM; (c) n-BuLi,
2
THF followed by desired ketones.
Scheme 8: The preparation of compounds 28, 29 and 30. Reagents and reaction conditions: (a)
n-BuLi, THF followed by desired ketones; (b) TFA, DCM; (c) Et N, AcCl, DCM.
3
To further explore functionality at the 3-position with varying polarity, an amide (21) was
introduced, as shown in Scheme 9. The commercially available iodo-anthranilate was reacted
with dimethyl malonate under basic conditions to directly give the product of amide formation
and then underwent Knoevenagel condensation to provide the desired 3-carboethoxy quinolone.
Halogenation using POCl and methyl ether formation were followed by ester hydrolysis and
3
7

amide formation employing EDCI / HOBt. The tertiary amide then underwent smooth metal-
halogen exchange and reaction with a pyridyl-triazolyl ketone to furnish amide 21.
4
5
Scheme 9: The preparation of a 3-C(O)NR R quinoline 21. Reagents and reaction conditions:
a) dimethyl malonate, NaOMe, MeOH; (b) POCl ; (c) NaOMe, PhCH ; (d) LiOH, THF, iPrOH,
(
3
3
H O; (e) 4-(trifluoromethyl)piperidine, EDCI, HOBt, Et N, DMF; (f) n-BuLi, THF followed by
2
3
(2,6-dimethyl-3-pyridinyl)(1-methyl-1H-1,2,3-triazol-5-yl)-methanone.
All the compounds depicted in the synthesis schemes above are single enantiomers,
obtained by separation using chiral SFC / HPLC and the data reported below are derived from
the more potent enantiomer in the TF assay. The absolute configuration of the final compounds
was not determined.
Table 1. Binding, cell-based functional activity, metabolic stability and solubility data of 3-
alkyl-4-chloro-2-methoxyquinolines.
®
ThermoFluor
binding to
RORt cell-
based reporter
assay IC50
Microsomal
Stability
Human /
Rat
Equilibrium
Solubility
pH 2 / 7
(M)
Cmpd
#
X
cLogP
RORt LBD,
(nM)
K (nM)
d
(% inhibition
Extraction
8

at 6 M)
Ratio
3
4
5
2.6
2.9
4.1
63
2.9
412
327 (93)
45 (100)
844 (90)
>1 / >0.9
0.9 / 0.9
0.9 / 0.9
>400 / <4
395 / <4
>400 / <1
6
7
3.6
3.8
1.3
1.4
30 (98)
14 (98)
>1 / 0.9
0.9 / 0.9
>400 / <1
273 / <4
8
9
4.7
1.5
0.1
2000 (66)
0.9 / 0.9
0.9 / 0.7
>400 / <1
387 / 13
a
110
280 (68)
a
17b
RORt cell-based reporter assay run at Phenex Pharmaceuticals
Our approach was to take a potent compound (2) and modify one portion of the molecule,
by keeping the quinoline 6-position constant and changing the 3-position substitution to lower
lipophilicity and improve the solubility. The simple 3-methyl compound 3 (Table 1), with a
modest cLogP (2.6), is decidedly smaller and much less lipophilic than compounds 1 and 2 (vide
supra), however it did show moderate binding (K = 63 nM) and cellular activity (IC = 327
d
50
nM). Despite the small size of the substituent at the 3-position, compound 3 is an inverse agonist
18
and is fully efficacious. The 3-trifluoroethyl compound 4 is potent in both binding and cellular
assays and is fully efficacious. Enlarging the 3-position substituent to an isobutyl (5) results in a
dramatic loss in binding affinity and cellular potency. Interestingly, the closely related 3-
9

cyclopropylmethyl substituted compound (6) regains potency similar to 4 and is full efficacious.
Cycloalkyl substituents such as 5-membered ring 7 directly attached to the quinoline ring
maintain binding and cellular potency (Kd = 1.4 nM, cellular IC = 14 nM), however a 6-
5
0
membered ring, when attached to the 3-position by a methylene spacer, such as the 4,4-
difluorocyclohexyl (8) and the cyclic sulfone (9), are tolerated in the TF assay but lose potency
in the cell-based assay and exhibit partial inverse agonist activity. Compounds 3-9 all have
reduced cLogP and have demonstrated improved pH 2 solubility compared to benchmark
compound 2, however, pH 7 solubility remains difficult to achieve across the series with only
sulfone 9 (cLogP = 1.5) showing measurable solubility (13 M). Unfortunately, compounds 3-9
also showed poor metabolic stability, as measured by human and rat microsomal extraction ratios
(ER).
Table 2. Binding, cell-based functional activity, metabolic stability and solubility data of 3-
alkoxy-4-chloro-2-methoxyquinolines.
Microsomal
RORt cell-
based reporter
®
ThermoFluor
binding to
Stability
Human /
Rat
Extraction
Ratio
Equilibrium
Solubility
pH 2 / 7
(M)
Cmpd
#
X
cLogP
2.8
assay IC (nM)
5
0
RORt LBD,
(% inhibition
K (nM)
d
at 6 M)
1
0
6.1
29 (102)
>1 / >0.9
>400 / <1
1
0

1
1
1
2
2.9
2.3
3.8
8.2
20 (87)
33 (86)
>1 / 0.9
>400 / <1
371 / <4
>1 / >0.9
1
1
1
3
4
5
2.0
1.6
31
14
38 (-80)*
193 (61)
0.9 / 0.7
0.9 / 0.9
0.9 / 0.8
>400 / <4
>400 / 41
>400 / 336
1.2
262
326 (-129)*
*
EC ; Compounds are agonists
50
Table 2 depicts a series of analogs that possess an alkoxy substituent at the 3-position
with the goal of identifying substituents that could further improve solubility. The introduction of
the alkoxy-oxygen enabled the study of compounds with a lower cLogP value (viz 5 cLogP = 4.1
and 10 cLogP = 2.8) versus the alkyl equivalents of Table 1. The alkoxy-oxygen as well as
additional heteroatoms in the heterocyclic (14) and heteroaryl groups (13, 15) enabled the cLogP
values associated with the compounds of Table 2 to be in a modest cLogP (1.2-2.9) range despite
increases in the size of the substituent. Alkoxy compounds with small alkyl substituents were
potent in the TF binding assay (10, 11, 12 K = 6.1, 3.8, 8.2 nM respectively). Compounds
d
possessing additional heteroatoms in the C3 alkoxy sidechain (13, 14, 15) were less potent in the
TF assay, with Kd values of 14, 31, and 262 nM respectively. Functional activity in the cell-
based assay was highly variable across this subset of analogs. Alkoxys 10, 11, and 12 exhibit
potent suppression of RORt driven transcriptional activity (29, 20 and 33 nM), with 10 being a
fully efficacious inverse agonist while 11 and 12 are partial inverse agonists. Incorporating a
larger group, such as in compound 14, resulted in a loss of potency in the cell-based assay (IC =
50
1
93 nM) and the compound was also a partial inverse agonist. Compound 13 had similar binding
1
1

affinity as 14, but demonstrated agonistic activity in the cell-based assay (-80% inhibition) with
an EC of 38 nM. The slightly larger heteroaromatic analog 15 has 8.5-fold less cellular potency
5
0
than 13, and it was also an agonist (EC = 326 nM, -129% inhibition). Despite poor metabolic
5
0
stability for compounds 10 – 15, they all have reduced cLogP, and the lowest cLogP compounds
14, 15, cLogP < 1.6) have demonstrated improved pH 2 and pH 7 solubility.
(
Table 3. Binding, cell-based functional activity, metabolic stability and solubility data of 3-
aminomethyl-4-chloro-2-methoxyquinolines and 3-amido-4-chloro-2-methoxyquinolines.
RORt cell-
ThermoFluor based reporter
Microsomal
Stability
Human /
Rat
Extraction
Ratio
®
Equilibrium
Solubility
pH 2 / 7
(M)
Cmpd
#
binding to
assay IC50
(nM)
X
cLogP
RORt LBD,
K (nM)
(% inhibition
at 6 M)
d
1
1
6
7
2.5
2.9
3.9
46
41 (72)
0.9 / 0.9
0.9 / 0.9
>400 / 8
>6000 (14)
>400 / 100
1
8
2.5
5.7
19 (54)
0.9 / 0.9
>400 / <4
1
2

1
9
3.3
1.5
22 (99)
0.9 / 0.8
>1 / 0.7
>400 / <1
2
2
0
1
2.1
2.2
47
122 (87)
>400 / 30
>400 / 60
a
357
386 (66)*
0.8 / 0.8
a
17b
*
% Inhibition at 3 M; RORt cell-based reporter assay run at Phenex Pharmaceuticals or Janssen
Compounds 16-21 (Table 3) represent a series of analogs in which the tertiary alcohol
substituents were fixed with varying amine- or amide-containing groups at the 3-position of the
quinoline towards improving aqueous solubility at pH 2 and pH 7. As expected, the addition of
amines at the 3-position resulted in an increase in solubility at pH 7 for several compounds (16,
1
4a-b
1
7 and 20). However, as previously reported by ourselves and others,
structural modifications
to this region of the ligand binding domain can result in large differences in cellular efficacy,
despite similar binding affinities. This is clearly evidenced by the comparison of 17 and 20
which share near identical binding affinity but differ dramatically in cell-based efficacy (%
inhibition of 72, 54 and 99% respectively), suggesting that increasing steric bulk in this region of
the molecules is required for full inverse agonism. Amide 21 shows a dramatic loss in binding
affinity (> 200-fold) and an 18-fold loss in cellular activity compared to 19, however both pH 2
and pH 7 solubility were improved. It proved challenging to discern a clear SAR for 3-
1
substituents of this chemotype in the context of cellular efficacy, and to find a balance between
potency and pH 7 solubility.
1
This aspect discussed extensively in reference 14a.
1
3

To better understand the differences in cellular efficacy of the abovementioned
compounds, we determined the crystal structure of the complex of compound 25 (Table 4) bound
to the RORT LBD to 2.9Å resolution. With respect to the binding mode, we considered
compound 25 to be representative of the analogs described in this manuscript, especially for
those in Tables 3 and 4. Notably, compound 25 contains the same 4-CF -piperidine moiety as
3
compound 19 (Table 3). When bound to the RORt LBD, compound 25 occupies the central
sterol binding pocket, consistent with previous reports from our group describing RORt LBD
14a,15a
complexes with this chemotype.
In the published 25-hydroxycholesterol (25-HC) agonist
complex the terminal hydroxyl of 25-HC forms part of a hydrogen bond network including a
2
2
bound water molecule, His479 in helix 10/11 and Tyr502 of helix 12 (Figure 2). This
interaction is an important contributor to the orientation of helix 12, which in turn is a key
structural element of the binding site occupied by the co-activator peptide in the agonist
complex. The correct orientation of these structural elements is critical to activation function 2
2
3
(
AF-2). In the complex with compound 25, the methylene-linked trifluoromethylpiperidine
moiety overlays with the hydroxy-bearing side-chain of 25-HC, impinging on the region of the
above-mentioned interaction network described for the 25-HC agonist complex (Figure 2). This
intrusion of the methylene-linked trifluoromethylpiperidine moiety causes a shift of His479 and
leads to disorder in the C-terminal end of helix 10/11 as well as all of helices 11’ and 12. Such
disruptions of AF-2 are expected to preclude co-activator binding. Thus, suitably sized
methylene-linked 3-substituents sterically disrupt the co-activator binding site of the RORt
1
4a,15a and references therein
LBD.
1
4

Figure 2. Structural effect of compound 25 binding to the RORt LBD. Bound compound 25 is
shown in orange color-by-atom (N in blue, O in red, S in yellow, F in cyan, Cl in green); RORt
LBD from the complex with compound 25 is shown as orange ribbon with selected sidechains
22
shown in orange color-by-atom; 25-HC from the published 25-HC complex (PDB ID 3L0L) is
shown in purple color-by-atom; bound co-activator peptide from 3L0L is shown in purple
ribbon; RORt LBD from 3L0L is shown as white ribbon or white color-by-atom; a key
structural water molecule from 3L0L is shown as a red sphere; the difference in the position of
His479 between 3L0L and the complex with compound 25 is highlighted by the red arrow; for
clarity, only selected secondary structural elements and sidechains are shown. (Deposited as
PDB ID 6NAD; crystallography details as described in refs 14a, 15a and 22).
1
5

The observed differences in functional activity for these aminomethyl substituted analogs
(Table 3) appear consistent with the proposed mechanism outlined above for compound 25, with
these groups sterically clashing with one or more structural elements of the LBD, acting through
allosteric disruption of the co-activator binding site rather than through direct competition for a
binding site. These mechanistic details were discussed extensively in an earlier report from our
14a
group on this chemotype. Moieties that meet a specific size requirement may clash and thus
elicit the required structural effect. For ligands that meet or exceed this size criterion, variations
in affinity may be due to binding interactions not observed (due to disorder) in the measured
crystal structures, or some other effect. The maximal efficacy data for compounds 16-20 is
generally consistent with this speculation (Table 3). Modeling based on overlays (not shown), as
described in Figure 2, indicates that compounds 19 and 20 extend furthest toward the agonist
2
4
position for helix 10/11, and these compounds stand out as showing full inverse agonism in the
cellular assay. Similarly, modeling suggests that compounds 16-18 clash to lesser extents with
the agonist conformation of helix 10/11, and these compounds show varying sub-maximal
effects. Due to the amide linkage, compound 21 is not directly comparable with compounds 16-
2
0. Conformational analysis for the 3-position substituent of compounds 19 and 21 indicates that
the full sets of accessible conformers span similar spaces, although some members of the set for
compound 21 tend to clash with additional regions of the binding site (results not shown),
consistent with the reduced affinity observed for compound 21 (Table 3).
Table 4. Binding, cell-based functional activity, metabolic stability and solubility data of 4-
chloro-2-methoxyquinolines.
1
6

RORt cell-
based reporter
Microsomal
Stability
®
ThermoFluor
binding to
Equilibrium
Solubility
pH 2 / 7
(M)
Cmpd
#
cLogP
assay IC (nM) Human / Rat
5
0
RORt LBD,
(% inhibition
Extraction
Ratio
K (nM)
d
at 6 M)
a
2
2
2
3
3.3
3.3
0.02
1.3
5.1 (101)
0.9 / 0.8
0.7 / 0.6
>400 / <1
>400 / <4
a
12 (100)*
2
4
1.9
2.3
18 (103)
0.9 / 0.8
>400 / 31
a
2
2
2
5
6
7
4.1
3.9
3.9
0.5
0.3
1.1
14 (101)*
0.7 / 0.4
0.7 / 0.5
0.7 / <0.2
>400 / <4
>400 / <1
>400 / <1
12 (100)
61 (96)
a
2
8
2.6
3.3
50 (94)*
0.7 / 0.3
>400 / <4
a
2
3
9
0
0.4
1.2
18.8
4.8
166 (95)
0.8 / 0.4
ND
a
313 (87)
0.5 / <0.2
>400 / >400
a
3
3
1
2
3.3
2.7
0.6
2.4
25 (87)
0.8 / 0.5
0.8/0.4
>400 / 283
>400/7
121 (97)
b
a
3
3
2.2
1.2
116 (96)
0.9/0.7
>400/4
1
7

a
17b
b
*
% Inhibition at 2 M; RORt cell-based reporter assay run at Phenex Pharmaceuticals or Janssen ; C-3 substituent =
Since compound 19 displayed improved binding affinity, cell-based potency and full
efficacy, we explored a series of analogs that maintained the methylene-linked trifluoro-4-
methylpiperidine 3-substituent of compound 19 and modified the 6-position substituents to
investigate whether solubility at pH 7 and metabolic stability could be improved. These
compounds (22-33, Table 4) afforded a fifty-fold range in binding affinities and consistently
showed maximal or near maximal efficacy in the cell-based reporter assay, and their measured
equilibrium solubility was >400 M at pH 2. Replacement of the dimethyl pyridine heterocyclic
group present in 22 with a m-CF pyridine ring (compound 23) resulted in an improvement in
3
metabolic stability, but offered no benefit in pH 7 solubility. As previously seen, improvement in
pH 7 solubility could be accomplished by lowering cLogP, as in compound 24 (cLogP = 1.9),
which incorporates a dimethyl imidazole group, but this group did not improve metabolic
stability compared to 19. Compounds 23, 25 and 26, where one of the aryl substituents is a m-
CF pyridine, had improved microsomal stability relative to compound 22 especially in rat liver
3
microsomes. Further improvement in rat microsomal stability was obtained with compound 27
wherein the m-CF pyridine ring in 26 was replaced with a p-CF pyridine ring. Similar
3
3
enhancement in rat microsomal stability was observed with compound 28 wherein the imidazole
ring in 27 was replaced with a small saturated heterocycle (N-acetyl azetidine). A three to four-
fold reduction in cell-based activity was observed by replacing the p-CF pyridine ring with a
3
triazole, as in 29. However, replacing the triazole ring in 29 with a dimethyl imidazole ring, as in
3
0, afforded a compound with excellent solubility (>400 M) at both pH 2 and pH 7 and an
enhancement in human microsomal stability (hER = 0.5). Replacement of the N-acetyl azetidine
1
8

group with a tetrahydropyran afforded compound 31 which maintained solubility at pH 2 and pH
7
, but resulted in a loss of human microsomal stability. We speculate that the favorable solubility
3
of 31, despite its higher cLogP, is due to increased sp character. In addition to asymmetric
tertiary alcohols, we wanted to investigate whether an achiral molecule would impart solubility
and improve microsomal stability, while maintaining potency. Compound 32, where the two aryl
groups are identical, showed improved rat microsomal stability relative to compound 24, but lost
cellular potency and pH 7 solubility. We explored other alkyl trifluoromethyl-containing amines
at the C-3 position of the quinoline, keeping this asymmetric tertiary alcohol substituent fixed.
Representative compound 33 showed good efficacy in the cell-based reporter assay, and pH 2
solubility, but lacked a balance in properties with poor human microsomal stability and no
improvement in pH 7 solubility.
We also tested a subset of compounds described in this paper in a relevant human Th17
primary cell assay (Table 5) which measures inhibition of IL-17A production under conditions
1
4a
that favor Th17 cell differentiation as reported previously.
All compounds tested displayed
full inverse agonism and similar potency in both assays with a slight reduction in efficacy in the
Th17 cell assay (denoted in parenthesis in the table and ranging from 85% to 91% inhibition of
IL-17) establishing correlation between the two assays.
Table 5. Functional activity correlation between the transfected cell-based reporter assay and
relevant human Th17 primary cells.
RORt cell-based
Human Th17 assay
Compound
#
reporter assay IC (nM)
IC (nM)
5
0
50
(
% inhibition at 6 M)
(% inhibition @ 1.2 M)
1
2
73 (99)
3.9 (100)
29 (102)
95 (87*)
3.3 (89)
10 (85)
1
0
1
9

2
2
2
3
4
7
12 (100)
18 (103)
61 (96)
8 (89)
12 (91)
17 (85)
*
% Inhibition at 6 M
Table 6. Comparison of Rat Pharmacokinetic Data for Compounds 1 and 27.
Compound
IV (1 mg/kg)
PO (5 mg/kg)
#
CL
Vss
t1/2
Cmax
t1/2
F
(mL/min/kg)
(L/kg)
(h)
ng/mL
(h)
(%)
1
11.5
7.9
4.3
5.7
5.6
9.1
262
553
5.6
8.5
42
77
2
7
Compound 27 was further evaluated in rat PK (Table 6) due to its’ cellular potency and
favorable microsomal stability in rats. This compound exhibits low clearance, a high steady state
volume of distribution, longer half-life, higher PO Cmax (>2-fold) and higher oral bioavailability
relative to the comparator molecule 1. The in vitro metabolic stability data (rER <0.2) for
compound 27 and 1 (rER = 0.2) are in agreement with the observed in vivo clearance.
In summary, we methodically investigated small to moderate sized groups with buried
polarity at the 3-position of the quinoline ring to achieve the goal of maintaining potency and
improving solubility. We demonstrated that reducing lipophilicity by incorporating heteroatoms
generally led to >400 M pH 2 solubility and in some instances improved pH 7 solubility. The 4-
CF -piperidine group within 19 not only improved solubility, but also maintained full inverse
3
2
0

agonist activity in the cell-based assay by sterically disrupting the co-activator binding site of the
RORt LBD, in agreement with other crystal structures we have previously reported in this
series. We incorporated additional modifications at the 6-position of the quinoline to improve
liver microsomal stability. It is noteworthy that the changes detailed in Table 4 at the 6-position
of the quinoline resulted in the successful identification of compounds that maintained full
inverse agonist activity, with improvements in both pH 7 solubility (compounds 24, 30-32),
microsomal stability (compounds 23, 25-28, 30) and good rat pharmacokinetic properties (Table
5
, compound 27). Advanced profiling of compounds from the quinoline tertiary alcohol series,
including efficacy in a CIA model of arthritis, will be described in future reports.
Acknowledgements: We would like to acknowledge Tara Haarlander and the Discovery
®
Sciences group for generating ThermoFluor , metabolic stability and solubility data, Jingxue Yu,
Rosa Luna-Roman, Discovery Sciences and Phenex Pharmaceuticals for generating cellular
reporter assay data, and Aimee De Leon-Tabaldo, Karen Ngo and Patrick Wilkinson for
generating Th17 data.
1
2
3
. Jetten AM. Nucl Recept Signal. 2009;7:e003.
. Ivanov II, McKenzie BS, Zhou L, et al. Cell. 2006;126:1121–1133.
. (a) Castro G, Liu X, Ngo K, De Leon-Tabaldo A, Zhao S, Luna-Roman R, Yu J, Cao T, Kuhn
R, Wilkinson P, Herman K, Nelen MI, Blevitt J, Xue X, Fourie A, Fung-Leung WP. PLoS
One. 2017;12(8): e0181868. (b) Hawkes JE, Yan BY, Chan TC, Kreuger JG J. Immunol. 2018;
2
4
5
01(6);1613.
. Elloso MM, Gomez-Angelats M, Fourie AM. J Leukoc Biol. 2012;92:1187–1197.
. (a) Rizvi S, Chaudhari K, Syed BA. Nat Rev Drug Discov. 2015;14:745–746. (b) Campa M,
Mansouri B, Warren R, Menter A. Dermatol. Ther. 2016;6;1-12. (c) Fragoulis GE, Siebert S,
McInnes IB. Annu. Rev. Med. 2016;67;27.1-27.17.
6
. Tabarkiewicz J, Pogoda K, Karczmarczyk A, Pozarowski P, Giannopoulos K. Arch. Immunol.
Ther. Exp. 2015;63;435-449.
7
8
9
. Komiyama Y, Nakae S, Matsuki T, et al. J Immunol. 2006;177:566–573.
. Mease, PJ. Curr Opin Rheumatol. 2015;27;127-133.
. Sarra M, Pallone F, Macdonald TT, Monteleone G. Inflamm Bowel Dis. 2010;16:1808–1813.
2
1

1
1
1
1
1
1
0. Cyr P, Bronner SM, Crawford JJ. Bioorg. Med. Chem. Lett. 2016;26:4387–4393.
1. Bronner SM, Zbieg JR, Crawford JJ. Expert Opin. Ther. Pat.2017;1:101-112.
2. Pandya VB, Kumar S, Sachchidanand, Sharma R, Desai RC. J. Med. Chem. 2018;61:10976-
0995.
3. Zhang Y, Luo XY, Wu DH, Xu Y. Acta Pharmacol Sin. 2015;36:71–87.
4. (a) Kummer DA, Cummings MD, et al. Bioorg. Med. Chem. Lett. 2017;27;2047-2057. (b)
Rene, O, Fauber BP, et al. ACS Med. Chem. Lett. 2015;6;276-281.
5. (a) Barbay JK, Cummings, MD et al. Bioorg. Med. Chem. Lett. 2017;27;5277-5283. (b)
Leonard KA, Barbay JK, Edwards, JP, et al. US Patent 0107094 A1; 2014.
6. (a) Pantoliano MW, Petrella EC, Kwasnoski JD, et al. J. Biomol. Screen. 2001;6;429-40. (b)
Matulis D, Kranz JK, Salemme FR, Todd MJ. Biochemistry 2005;44;5258-5266.
7. (a) For assay details, see: Barbay JK, Edwards JP, Kreutter KD, et al. PCT Int. Appl.
1
1
1
WO 2015057203. (b) The construct containing full-length RORt was transfected to the cells,
which is different from the assay previously described in reference 17(a), where the construct
containing only RORt LBD was used.
1
1
8. Partial inverse agonism is defined as <90% inhibition in the RORt cell-based reporter assay.
9. Leonard KA, Barbay K, Edwards JP, Kreutter KD, Kummer DA, Maharoof U, Nishimura R,
Urbanski M, Venkatesan H, Wang A, Wolin RL, Woods CR, Fourie A, Xue X, Cummings MD,
McClure K, Tanis VM. WO 2015057626.
2
2
2
2
2
0. For a description of the syntheses of desired target ketones see: references 14 and 15(a).
1. Soroosh P, Wu J, Xue X et al. Proc. Natl. Acad. Sci. 2014;111;12163-12168.
2. Jin L, Martynowski D, Zheng S, Wada T, Xie W, Li Y. Mol. Endocrinol. 2010;24;923-929.
3. Nagpal S, Friant S, Nakshatri H, Chambon P. EMBO J. 1993;12;2349–2360.
4. Based on modeling of idealized fully extended conformations for each of the relevant
moieties, positioned as observed for compound 25 (results not shown).
2
2

3
-Substituted Quinolines as RORt Inverse Agonists
Equilibriu
m
Solubility
pH 2 / 7
ThermoFluo
Cmpd r binding to
®
.
RORt LBD,
Kd (nM)
(
M)
1
9.2
1.1
<1 / <1
27
>400 / <1
>
>
400 /
400
30
4.8
We have previously reported the syntheses of a series of 3,6-disubstituted quinolines as
modulators of the retinoic acid receptor-related orphan receptor gamma t (RORt). These
molecules, such as 1, are potent binders but are high molecular weight and they exhibited poor
solubility at pH 2 and pH 7. This manuscript details our efforts at improving physical chemical
properties for this series of compounds by increasing the diversity at the 3-position (i.e.
introducing heteroatoms and lowering the molecular weight). These efforts have led to molecules
such as 27 and 30 which are potent binders with improved solubility.
2
3