Retinoic acid receptor-related orphan receptor γt (RORγt) is a nuclear receptor associated with the pathogenesis of autoimmune diseases. Allosteric inhibition of RORγt is conceptually new, unique for this specific nuclear receptor, and offers advantages over traditional orthosteric inhibition. Here, we report a highly efficient in silico-guided approach that led to the discovery of novel allosteric RORγt inverse agonists with a distinct isoxazole chemotype. The the most potent compound, 25 (FM26), displayed submicromolar inhibition in a coactivator recruitment assay and effectively reduced IL-17a mRNA production in EL4 cells, a marker of RORγt activity. The projected allosteric mode of action of 25 was confirmed by biochemical experiments and cocrystallization with the RORγt ligand binding domain. The isoxazole compounds have promising pharmacokinetic properties comparable to other allosteric ligands but with a more diverse chemotype. The efficient ligand-based design approach adopted demonstrates its versatility in generating chemical diversity for allosteric targeting of RORγt.
Retinoic acid receptor-related orphan receptor γt (RORγt) is a nuclear receptor associated with the pathogenesis of autoimmune diseases. Allosteric inhibition of RORγt is conceptually new, unique for this specific nuclear receptor, and offers advantages over traditional orthosteric inhibition. Here, we report a highly efficient in silico-guided approach that led to the discovery of novel allosteric RORγt inverse agonists with a distinct isoxazole chemotype. The the most potent compound, 25 (FM26), displayed submicromolar inhibition in a coactivator recruitment assay and effectively reduced IL-17a mRNA production in EL4 cells, a marker of RORγt activity. The projected allosteric mode of action of 25 was confirmed by biochemical experiments and cocrystallization with the RORγt ligand binding domain. The isoxazolecompounds have promising pharmacokinetic properties comparable to other allosteric ligands but with a more diverse chemotype. The efficient ligand-based design approach adopted demonstrates its versatility in generating chemical diversity for allosteric targeting of RORγt.
The nuclear receptor (NR)
RORγt has emerged as an important therapeutic target in recent
years because of its important role in both cancer and autoimmune
disease. Inhibition of RORγt is a promising therapeutic strategy
for the treatment of prostate cancer because it stimulates androgen
receptor (AR) gene transcription.[1,2] However, RORγt
is most prominently targeted for inhibition because of its essential
role in promoting T helper 17 (Th17) cell differentiation.[3−5] Th17 cells produce the cytokine IL-17 which is strongly implicated
in the pathogenesis of autoimmune diseases[6] such as psoriasis,[7] multiple sclerosis,[8] and inflammatory bowel disease.[9] Disrupting the Th17/IL-17 pathway using IL-17 monoclonal
antibodies (mAb) is a successful therapeutic strategy, with three
mAbs approved for the treatment of plaque psoriasis: secukinumab (Cosentyx),[10] brodalumab (Siliq),[11] and ixekizumab (Taltz).[12] Inhibition
of RORγt with small molecules to disrupt the Th17/IL-17 pathway
has been the focus of much research in recent years,[13−20] with several compounds having progressed to clinical trials.[2]RORγt contains a hydrophobic ligand
binding pocket located within a ligand binding domain (LBD) that is
highly conserved across the NR family.[21] However, its transcriptional activity is not dependent on ligand
binding because the apo protein retains the C-terminal helix 12 (H12)
in a conformational state that allows for partial recruitment of coactivator
proteins.[22,23] Although formally an orphan receptor with
no proven endogenous ligands, RORγt is responsive to binding
of naturally occurring cholesterol derivatives. Hydroxycholesterols
have been shown to be effective agonists that stabilize H12 in such
a way to further promote coactivator binding.[24] In contrast, digoxin (1, Figure ) is an inverse agonist that stabilizes H12
in a conformation that is unsuitable for coactivator binding but promotes
corepressor binding, thus leading to diminished gene transcription.[25] Numerous synthetic inverse agonists are also
known, including T0901317 (2, Figure ).[26] In all these
cases, the ligands target the same orthosteric ligand binding pocket
(Figure ).
Figure 1
Orthosteric
and allosteric RORγt ligand binding sites are shown by overlay
of the crystal structures of RORγt LBD in complex with orthosteric
inverse agonist 2 (orange, PDB code: 4NB6) and allosteric
inverse agonist 3 (blue, PDB code: 4YPQ). The structures
of the orthosteric inverse agonist 1 and allosteric inverse
agonist 4 are also shown.
Orthosteric
and allosteric RORγt ligand binding sites are shown by overlay
of the crystal structures of RORγt LBD in complex with orthosteric
inverse agonist 2 (orange, PDB code: 4NB6) and allosteric
inverse agonist 3 (blue, PDB code: 4YPQ). The structures
of the orthosteric inverse agonist 1 and allosteric inverse
agonist 4 are also shown.NR orthosteric ligand binding pockets are the target
for numerous and highly effective drug molecules.[27] Nevertheless, the highly conserved nature of this pocket
across the NR family has led to issues associated with selectivity
and mutation-induced resistance. Furthermore, dosing levels must be
appropriate to compete with endogenous ligands. Molecules that target
allosteric binding sites on NRs could circumvent such problems, for
example because of the chemical uniqueness of the pocket and the absence
of a competitive endogenous ligand. Such allosteric compounds are
therefore extremely valuable for both drug discovery and chemical
biology applications.[28−30] The discovery that the potent RORγt inverse
agonists MRL-871 (3, Figure )[31] and later 4(32) target a previously unreported
allosteric binding site within the RORγt LBD was therefore highly
significant. These ligands were observed to directly interact with
the activation function loop between H11 and H12 (AF-2 domain), thus
forcing H12 to adopt an unusual conformation that prevents coactivator
recruitment (Figure ).[31]Allosteric modulation of RORγt
has enormous potential as a novel therapeutic strategy, but the examples
of ligands that unambiguously target the allosteric pocket have been
limited to compounds based on closely related chemotypes containing
indazole or imidazopyridinecores.[28] As
an example, indazoles 3 and 4 displayed
promising in vivo activity,[33,34] but challenges remain,
such as PPARγ cross-activity and pharmacokinetic (PK) profiles,
for which novel chemotypes are needed.[15] In order to better exploit the strategy of allosteric modulation
for therapeutic purposes, there is thus an urgent need to identify
novel chemotypes targeting the allosteric site. In this study, we
report the design, synthesis, and evaluation of a novel class of RORγt
allosteric inverse agonists. The novel chemotype, discovered by in
silico-guided pharmacophore screening and optimization, is based on
a trisubstituted isoxazolecore that, following efficient optimization
of two substituents, led to the discovery of a submicromolar inverse
agonist. Protein X-ray crystallography and biophysical data unambiguously
proved the designed allosteric mode of action. The compounds effectively
inhibit cellularIL-17a expression and thus constitute valuable leads
in the development of treatments for autoimmune diseases. To the best
of our knowledge, our highly efficient in silico-guided approach is
the first example of a medicinal chemistry program to overtly identify
and develop a novel chemotype that targets the RORγt allosteric
site.
Results and Discussion
In Silico Pharmacophore Screen
In
order to identify novel chemotypes for chemical optimization, we used
the crystal structure of RORγt LBD in complex with 3 as the basis for an in silico 3D pharmacophore screen against virtual
compound libraries. An analogous scaffold hopping approach had been
used previously to identify similar scaffolds to 3 such
as the potent inverse agonist thienopyrrazole 5 (Figure ), although an allosteric
mode of action was not proven.[35] We created
a 3D pharmacophore hypothesis based on the crystal structure of 3 bound to the allosteric pocket using Phase (Schrödinger
2017–2).[36,37] Six structural features of 3 known to be important for activity were incorporated in
the hypothesis: the three six-membered aromatic rings, an anionic
group, and two hydrophobic substituents (Figure ). This hypothesis was used to interrogate
a virtual library of 289,174 compounds from the Asinex Gold–Platinumcollection of drug-like molecules.[38] Compounds
matching at least four out of the six pharmacophore features were
deemed to be a good hit. These were ranked using the “Phase
Screen Score” with higher scores indicating a better alignment
with the hypothesis. The Phase Screen scores for 3 and 5 were used as contextual references. The four highest ranking
hit structures were all found to be based around the same trisubstituted
isoxazole scaffold with 6 returned as the best match
(Figure ). This same
scaffold was present in 13 of the top 30 hits. However, in each case
we noted that only four out of six pharmacophore features were matched.
Therefore, we designed two virtual ligands, 7 and 8, that incorporated five and six of the features, respectively.
As expected, this led to improved Phase Screen Scores (Figure ), and these compounds were
therefore selected as initial targets for experimental investigation.
Figure 2
3D Pharmacophore
screening identifies a compound class with a novel isoxazole-based
chemotype for experimental evaluation. The structural features of 3 incorporated into the pharmacophore hypothesis are indicated:
orange = aromatic rings, green = hydrophobic groups, and red = anionic
group.
3D Pharmacophore
screening identifies a compound class with a novel isoxazole-based
chemotype for experimental evaluation. The structural features of 3 incorporated into the pharmacophore hypothesis are indicated:
orange = aromatic rings, green = hydrophobic groups, and red = anionic
group.
Exploratory Structure–Activity Relationship
Study
Isoxazoles 7 and 8 were synthesized
via [3 + 2] dipolar cycloaddition of a nitrile oxide (generated in
situ from the oxime chloride 9a) and a commercially available
alkyne.[39] The regiochemistry of the resulting
trisubstituted isoxazole esters 10 was confirmed by 2D-NMR
experiments (key HMBC correlations are highlighted in Scheme ). Ester hydrolysis followed
by amidecoupling of tert-butyl-4-amino benzoate
via the respective acid chloride, and finally deprotection of the tert-butyl ester furnished the target compounds in an efficient
manner (Scheme ).
Scheme 1
Synthesis of Trisubstituted Isoxazoles 7 and 8
Key HMBC correlations
used to confirm the regiochemistry of 10a and 10b are shown. The 13C-NMR signals for the C-5 carbons are
distinctively downfield at 175 and 173 ppm, respectively. Reagents
and conditions: (a) NH2OH·HCl, NaOH (aq), EtOH, rt,
18 h, 83%; (b) NCS, DMF, 60 °C, 18 h, 86%; then (c) alkyne, NEt3, THF, 80 °C, 4 h, 69% (10a), 80% (10b); (d) LiOH, EtOH, H2O, 70 °C, 8 h, 84%
(11a), 95% (11b); and (e) (i) SOCl2, 50 °C, 2 h; (ii) tert-butyl-4-amino benzoate,
NEt3, CH2Cl2, 45 °C, 6 h; and
(iii) TFA, CH2Cl2, rt, 18 h, 42% (7), 69% (8).
Synthesis of Trisubstituted Isoxazoles 7 and 8
Key HMBC correlations
used to confirm the regiochemistry of 10a and 10b are shown. The 13C-NMR signals for the C-5carbonsare
distinctively downfield at 175 and 173 ppm, respectively. Reagents
and conditions: (a) NH2OH·HCl, NaOH (aq), EtOH, rt,
18 h, 83%; (b) NCS, DMF, 60 °C, 18 h, 86%; then (c) alkyne, NEt3, THF, 80 °C, 4 h, 69% (10a), 80% (10b); (d) LiOH, EtOH, H2O, 70 °C, 8 h, 84%
(11a), 95% (11b); and (e) (i) SOCl2, 50 °C, 2 h; (ii) tert-butyl-4-amino benzoate,
NEt3, CH2Cl2, 45 °C, 6 h; and
(iii) TFA, CH2Cl2, rt, 18 h, 42% (7), 69% (8).To determine if
the compounds showed a functional response in terms of RORγt
affinity for a coactivator, 7 and 8 were
tested in a time-resolved FRET (TR-FRET) coactivator recruitment assay.[31] Remarkably, both compounds inhibited coactivator
recruitment in a dose-dependent manner. The phenyl derivative 8 was found to be significantly more potent than the methyl
derivative 7: half-maximum inhibitory concentrations
(IC50) of 53.5 ± 2.9 μM for 8 compared
to >100 μM for 7. In line with previous reports, 3 and 5 were determined to be significantly more
potent with an IC50 of 7.8 ± 0.5 nM and 425 ±
61 nM, respectively (Table ).
Table 1
Structure-Activity Relationships around
the C-4 Isoxazole Positiona
cmpd
IC50 (μM)
Glide score
3
0.0078 ± 0.0005
– 14.576
5
0.425 ± 0.061
– 13.109
7
>100
– 13.372
8
53.5 ± 2.9
– 14.184
11b
>100
– 10.130
14
>100
n.d.
15
>100
– 13.724
16
73.9 ± 3.4
– 12.995
17
91.1 ± 4.6
– 14.308
18
8.76 ± 0.48
– 12.020
19
9.60 ± 0.60
– 14.012
20
>100
– 13.550
21
30.9 ± 1.3
– 13.519
22
62.6 ± 4.4
– 13.003
TR-FRET IC50 values
(μM) and respective Glide docking scores are shown. TR-FRET
data is recorded in triplicate; values are representative of >3
repeated experiments.
TR-FRET IC50 values
(μM) and respective Glide docking scores are shown. TR-FRET
data is recorded in triplicate; values are representative of >3
repeated experiments.In view of these highly promising TR-FRET results
with the in silico derived compounds around the trisubstituted isoxazole
scaffold already showing activity, phenyl isoxazole 8 was selected as the focus of a subsequent structure–activity
relationship (SAR) study focusing on the isoxazoleC-4 position. As
such, a small library of 11 derivatives was synthesized using carboxylic
acid 11b as the cornerstone intermediate (Scheme ) and evaluated using the coactivator
recruitment assay (Table ). While limited in size, this SAR study indicated that a
benzoic acid-containing substituent at the C-4 position was essential
for potency: examples bearing no C-4 substitution (11b), a para-benzoate (14), or a methylenecarboxylic acid (15) showed much reduced potency compared
to the initial hit. Moving the acid moiety to the meta-position (16) or adding a meta-fluoro substituent (17) somewhat lowered the activity. However, the insertion
of a single methylene unit between the amide and benzoic acid moieties
(18) led to a 6-fold increase in potency compared to
the initial hit. The corresponding amine (19) displayed
similar activity. Finally, reversing the relative positions of carbonyl
and nitrogencomponents of the amide bond (20–22) did not result in a corresponding increase in potency.
Scheme 2
Synthesis of C-4 Isoxazole Derivatives
Reagents and conditions:
(a) (i) SOCl2, 50 °C, 2 h; (ii) NH2R, NEt3, CH2Cl2, 45 °C, 6 h, 27–87%;
(b) LiOH, MeOH, H2O, 70 °C, 8 h, 43–99%; (c)
(i) SOCl2, 50 °C, 2 h; (ii) MeNH(OMe), NEt3, CH2Cl2, rt, 6 h; (iii) LiAlH4,
THF, 0 °C, 30 min, 65%; (d) (i) ethyl-4-aminobenzoate, AcOH,
MeOH, reflux, 24 h; (ii) NaCNBH3, MeOH, reflux, 12 h, 31%;
(e) (i) DPPA, t-BuOH, 85 °C, 18 h; (ii) TFA,
CH2Cl2, rt, 8 h, 59%; (f) (i) monomethyl terephthalate,
SOCl2, 50 °C, 2 h; (ii) 13, NEt3, CH2Cl2, 76%; (g) (i) methyl-4-formyl benzoate,
AcOH, MeOH, reflux, 24 h; (ii) NaCNBH3, MeOH, reflux, 18
h, 43%; and (h) methyl-4-(chlorosulfonyl)benzoate, pyridine, 60 °C,
24 h, 71%.
Synthesis of C-4 Isoxazole Derivatives
Reagents and conditions:
(a) (i) SOCl2, 50 °C, 2 h; (ii) NH2R, NEt3, CH2Cl2, 45 °C, 6 h, 27–87%;
(b) LiOH, MeOH, H2O, 70 °C, 8 h, 43–99%; (c)
(i) SOCl2, 50 °C, 2 h; (ii) MeNH(OMe), NEt3, CH2Cl2, rt, 6 h; (iii) LiAlH4,
THF, 0 °C, 30 min, 65%; (d) (i) ethyl-4-aminobenzoate, AcOH,
MeOH, reflux, 24 h; (ii) NaCNBH3, MeOH, reflux, 12 h, 31%;
(e) (i) DPPA, t-BuOH, 85 °C, 18 h; (ii) TFA,
CH2Cl2, rt, 8 h, 59%; (f) (i) monomethyl terephthalate,
SOCl2, 50 °C, 2 h; (ii) 13, NEt3, CH2Cl2, 76%; (g) (i) methyl-4-formyl benzoate,
AcOH, MeOH, reflux, 24 h; (ii) NaCNBH3, MeOH, reflux, 18
h, 43%; and (h) methyl-4-(chlorosulfonyl)benzoate, pyridine, 60 °C,
24 h, 71%.
In Silico Docking Directs Secondary SAR Study
In order to further improve the potency of our compounds, we next
explored the SAR at the isoxazoleC-5 position. For this, molecular
docking (Glide, Schrodinger 2017–2)[40,41] was used to select, with an attention to synthetic resource, C-5
substituents that were optimal for allosteric binding and therefore
activity. For the study, a single C-4 substituent, the amine of compound 19 was chosen based on its experimental activity and in silico
docking score (Table , vide infra). A virtual library of 84 C-5 analogues was enumerated
using the open-source ChemT software.[42] This library was docked against the allosteric site of RORγt
as defined by the X-ray crystal structure of 3 in complex
with the RORγt LBD.[31] A single docking
pose was returned for each virtual ligand, and these were ranked using
the “Glide Score”, an empirical measure of binding enthalpy.[43] We contextualized these scores by comparison
to those of compounds with known activity. The results (summarized
in Table , see Supporting Information for full information)
indicated that smaller heteroaromatic moieties at the C-5 position
would improve allosteric binding of the isoxazole ligands relative
to 19, heteroatoms at the 2-postion were predicted to
be optimal, for example, furan 23 and thiophene 24 (Table ). The introduction of a hydrogen-bond donor on the ring (specifically
at the 3-position) was predicted to be even more beneficial: docking
poses indicated that an additional hydrogen-bonding interaction with
the backbone of helix 4 might be possible (e.g., pyrrole 25, Table , Figure ). Bulkier substituents
were predicted to be detrimental for binding (e.g., naphthyl 26). To explore the predicted effect of a hydrogen-bond donating
group further we interrogated a designed subset of ligands in the
same docking experiment (see the Supporting Information). None of these ligands showed an improved Glide score compared
to pyrrole 25. However, we noted that 3-hydroxyl substitution
of the C-5phenyl ring (27) was predicted to significantly
enhance binding relative to 19. To validate our findings
experimentally, we selected a cross section of five derivatives for
synthesis (i.e., 23–27).
Table 2
Structure-Activity Relationships around
the C-5 Isoxazole Positiona
TR-FRET IC50 values
(μM) and respective Glide docking scores are shown. TR-FRET
data is recorded in triplicate; values are representative of >3
repeated experiments.
Figure 3
In silico modeled
docking pose of 25 (green) overlaid with crystal structure
of RORγt with 3 (orange) (PDB code: 4YPQ). For 25, the potential additional hydrogen bond with the RORγt H4
backbone is indicated.
In silico modeled
docking pose of 25 (green) overlaid with crystal structure
of RORγt with 3 (orange) (PDB code: 4YPQ). For 25, the potential additional hydrogen bond with the RORγt H4
backbone is indicated.TR-FRET IC50 values
(μM) and respective Glide docking scores are shown. TR-FRET
data is recorded in triplicate; values are representative of >3
repeated experiments.
Docking-Guided C-5 SAR Study
To expedite
the synthesis of isoxazole analogues with various C-5 and C-4 substituents,
we redesigned our synthetic approach. It was envisaged that 5-bromo-4-carboxy
isoxazole intermediate 30 would enable later stage introduction
of the desired C-5 substituents via palladium-mediated cross-coupling
chemistry. Introduction of C-4 substituents by manipulation of a carbonyl
functional group (as developed previously) would then be possible
(Scheme ).
Scheme 3
Retrosynthetic
Analysis of Trisubstituted Isoxazole 28 Allowing for
Late-Stage Diversification
The intermediate 30 was prepared
using analogous methodology to that used previously. In this case
it was necessary to isolate nitrile oxide 33 prior to
[3 + 2] cycloaddition with alkynyl bromide 32.[44] An efficient cycloaddition reaction led to an
essentially quantitative recovery of a 7:3 mixture of 5-bromoisoxazole 30a and 4-bromoisoxazole 30b as determined by 1HNMR (Scheme ). This result was in close alignment with literature examples that
indicated the 5-bromo isomer would predominate.[44] The mixture of regioisomers was purified by recrystallization
from hot n-heptane resulting in the isolation of
a 97:3 regiomeric mixture (43% recovery) that was employed in subsequent
steps. Assignment of the 5-bromoisoxazole 30a as the
major regioisomer was confirmed by 2D-NMR analysis of downstream products
and by synthesis via an independent route (see the Supporting Information).
Scheme 4
Synthesis of Isoxazole C-5 Analogues 23–27.
“R” groups
are defined in Table . Reagents and conditions: (a) NBS, AgNO3, Me2C(O), rt, 20 h, 80%; (b) (i) NCS, DMF, 60 °C, 18 h, (ii) NEt3, THF, rt, 30 min, 85%; (c) THF, 80 °C, 4 h, 30a 43%; (d) RB(pin), Pd(dppf)Cl2, DME, 85 °C, 8 h,
39–58%; (e) (i) LiAlH4, THF, 0 °C →
rt, 2 h, then (ii) DMP, CH2Cl2, rt, 8 h, 51–96%;
(f) tert-butyl-4-amino benzoate, MeOH, AcOH, reflux,
24 h then (ii) NaBH4, EtOH, 85 °C, 2–6 h, 16–24%;
(iii) TFA, CH2Cl2, rt, 18 h, 23, 24, 26, 48–73%; and (g) methyl-4-amino
benzoate, MeOH, AcOH, reflux, 24 h then (ii) NaBH4, MeOH,
reflux, 2–4 h, 16–19%; and (iii) LiOH, MeOH, H2O, 70 °C, 8 h, 25, 57%, 27, 99%.
Synthesis of Isoxazole C-5 Analogues 23–27.
“R” groups
are defined in Table . Reagents and conditions: (a) NBS, AgNO3, Me2C(O), rt, 20 h, 80%; (b) (i) NCS, DMF, 60 °C, 18 h, (ii) NEt3, THF, rt, 30 min, 85%; (c) THF, 80 °C, 4 h, 30a 43%; (d) RB(pin), Pd(dppf)Cl2, DME, 85 °C, 8 h,
39–58%; (e) (i) LiAlH4, THF, 0 °C →
rt, 2 h, then (ii) DMP, CH2Cl2, rt, 8 h, 51–96%;
(f) tert-butyl-4-amino benzoate, MeOH, AcOH, reflux,
24 h then (ii) NaBH4, EtOH, 85 °C, 2–6 h, 16–24%;
(iii) TFA, CH2Cl2, rt, 18 h, 23, 24, 26, 48–73%; and (g) methyl-4-amino
benzoate, MeOH, AcOH, reflux, 24 h then (ii) NaBH4, MeOH,
reflux, 2–4 h, 16–19%; and (iii) LiOH, MeOH, H2O, 70 °C, 8 h, 25, 57%, 27, 99%.The desired substituents were introduced at the
C-5 position by way of a Suzuki cross-coupling with a pinacol boronate[45] (to give intermediates 34–38) before conversion of the C-4ester to an aldehyde (39–43) and reductive amination (Scheme ). The lability
of the 5-bromo group under the conditions for ester reduction dictated
the order in which the synthesis steps were performed. Hydrolysis
of the benzoic methyl ester to the free acid yielded the desired compounds 23–27.In order to explore the SARaround the isoxazoleC-5 position, the five analogues prepared in
this second synthesis campaign were evaluated using the HTRF coactivator
recruitment assay (Table ). We were gratified to observe that furan 23 gave a 9-fold improvement in potency compared to phenyl 19. By comparison, thiophene 24 was slightly less potent.
Most significantly, pyrrole 25, which also showed the
most beneficial Glide score, was 36-fold more potent than 19 and with an IC50 value lower than the putative allosteric
modulator 5. These results were in excellent agreement
with the in silico Glide scores
obtained (Table ),
and the improvements in potency are a notable step toward emulating
the high potency of indazole 3 (Figure A). As predicted, the bulky naphthyl group
of 26 was detrimental for activity such that no IC50 curve could be fitted. The phenol derivative 27 showed a small improvement in potency compared to 19. For this more bulky group at the C-5 position, compared to pyrrole 25, the potential for additional hydrogen bonding, as indicated
in the docking study, is thus not strongly expressed.
Figure 4
Biochemical RORγt
assay data for 25, 3, and 5. (A) Dose–response curves from the TR-FRET coactivator recruitment
assay; (B–E) dose–response curves from the competitive
TR-FRET coactivator recruitment assay with fixed concentrations of
cholesterol (0, 0.25, and 1.0 μM); and (F) dose–response
curves from the ligand displacement HTRF assay using 44 (G) as an allosteric probe.
Biochemical RORγt
assay data for 25, 3, and 5. (A) Dose–response curves from the TR-FRET coactivator recruitment
assay; (B–E) dose–response curves from the competitive
TR-FRET coactivator recruitment assay with fixed concentrations of
cholesterol (0, 0.25, and 1.0 μM); and (F) dose–response
curves from the ligand displacement HTRF assay using 44 (G) as an allosteric probe.
Mode-of-Action Studies
The allosteric
mode-of-action for the novel lead compound 25 was first
explored using a competitive TR-FRET coactivator recruitment assay
against fixed concentrations of cholesterol (an orthosteric agonist).
If an allosteric ligand and cholesterol bind in a noncompetitive manner
at different sites on the RORγt LBD then the IC50 of the allosteric ligand should be independent of cholesterolconcentration.
By contrast, ligands competing for the same binding site should show
a cholesterol-dependent activity profile whereby increasing cholesterolconcentration should result in a corresponding increase in IC50 of the competing ligand.[31] In
our assay, increasing concentrations of 25 perturbed
coactivator recruitment in the absence of cholesterol with an IC50 value of 247.8 ± 17.7 nM. Interestingly, increasing
concentrations of cholesterol indeed resulted not in an increase but
in a further decrease in the IC50 value for 25 with a concomitant sharpening of the Hill slope (Figure B and Table ). This result provides strong evidence not
only for an allosteric mode-of-action but also for cooperative behavior
between orthosteric and allosteric ligand binding. The same profile
was observed for 5 (Figure C), providing the first evidence that this
compound also modulates RORγt activity in an allosteric fashion.
Indazole 3 also exhibited this behavior (Figure D). By comparison, the IC50 value for the orthosteric inverse agonist 1 increased as the concentration of cholesterol increased (Figure E). Collectively,
our competitive assay data provided strong evidence that 25 functioned as an allosteric inverse agonist.
Table 3
IC50 and Hill Slope Values
Observed in the Competitive TR-FRET Cofactor Recruitment Assay
0 μM
cholesterol
0.25 μM cholesterol
1.0 μM cholesterol
compound
IC50 (nM)
Hill slope
IC50 (nM)
Hill slope
IC50 (nM)
Hill slope
25
247.8 ± 17.7
–0.77 ± 0.04
138.0 ± 5.9
–0.86 ± 0.03
94.1 ± 3.3
–1.01 ± 0.03
5
547.3 ± 60.1
–0.74 ± 0.06
299.5 ± 18.0
–0.87 ± 0.04
268.9 ± 18.8
–0.90 ± 0.05
3
12.7 ± 0.6
–0.97 ± 0.04
9.4 ± 0.3
–1.04 ± 0.03
7.8 ± 0.2
–1.20 ± 0.03
1
7012 ± 588
–0.76 ± 0.05
33620 ± 1649
–0.77 ± 0.03
85400 ± 4276
–1.01 ± 0.06
To further confirm the allosteric mode-of-action for 25 on RORγt, we used an orthogonal assay to directly
probe for allosteric ligand binding, as opposed to measuring indirect
effects on coactivator recruitment. This assay used the previously
described AlexaFluor647-labeled MRL-871 derivative 44 (Figure G), which
upon binding to RORγt shows fluorescent emission as a result
of FRET from an anti-His terbium cryptate antibody donor.[32] The results of this experiment indeed corroborated
the data obtained from the competitive cofactor recruitment assay
(Figure F): the isoxazole 25 displaced the allosteric probe 44 with an
IC50 = 117.5 ± 8.5 nM, which was lower than that of 5 (IC50 = 180.0 ± 17.5 nM). As expected, indazole 3 was highly potent (IC50 = 17.3 ± 1.4 nM).Indazole 3 had previously been shown to be selective
for RORγt over other NRs (>100-fold), with only minor cross-activity
on PPARγ.[31] To give an indication
of the cross-reactivity of the isoxazole series on PPARγ, an
HTRF coactivator recruitment assay was performed with compounds 3, 5, and isoxazoles 19 and 23–27. 3 and 5 show IC50 values of 7.2 μM and 14.7 μM, respectively,
for PPARγ (vs 7.8 nM and 425 nM for RORγt) (Table ), meaning that they show some
cross-reactivity to PPARγ but still are 923- and 35-fold selective
for RORγt. 25 and all other compounds of the isoxazole
series result in only weak to no PPARγ inhibition (IC50 values >50 μM), indicating that the isoxazole scaffold
leads to favorably low PPARγ cross-reactivity. Thus, these data
indicate that the novel class of allosteric isoxazole inverse agonists
features potential as efficacious and selective RORγt inverse
agonists.
Table 4
IC50 Values Observed in
the Competitive TR-FRET Cofactor Recruitment Assay with PPARγ
compound
IC50 (μM)
3
7.2 ± 0.8
5
14.7 ± 1.0
19
78.6 ± 5.6
23
>100
24
>100
25
99.3 ± 6.4
26
>100
27
>100
Crystallography
Co-crystallization
studies were performed for the most potent isoxazole 25 with the RORγt-LBD, to provide molecular insights in the ligand–receptor
interaction. Crystals grew in a P622 space group and diffracted to a resolution
of 1.61 Å (Table S6). In the experimental
electron density map, clear density for compound 25 is
observed in the allosteric site, formed by helices 4, 5, 11, and 12
(Figure A, Figure S2). The compound binds to this allosteric
site in a similar orientation as 3 (Figure B), as was predicted by our
docking studies (Figure ). The 2,6-disubstituted phenyl ring common to both 3 and of 25 is located in the exact same part of the
binding pocket (Figure B). Moreover, hydrogen-bonding interactions between the carboxylic
acid group and the main-chain amidehydrogen atoms of A497 and F498,
as well as with the side chain of residue Q329, are also evident in
both structures. Unique to 25 is the pyrrole ring, which
is oriented to allow a hydrogen bond interaction with the main-chain
carbonyls of residues L353 and K354 (Figure C). The isoxazole scaffold also allows a
deeper penetration of this compound toward helix 4 of RORγt.
In the case of isoxazole 25, the AF-2 loop of the protein
and the allosteric ligand are positioned slightly further apart as
compared to 3 (Figure B). These structural data provide clear evidence for
the allosteric binding of 25 to RORγt in an orientation
that was predicted with remarkable accuracy in the docking study (Figure S3) but with specific additional molecular
effects resulting from the novel isoxazole scaffold and pyrrole based
substition pattern.
Figure 5
Co-crystal structure of RORγt with compound 25 (PDB code: 6SAL). (A) The tertiary structure of RORγt bound
to 25 (stick representation). The final 2Fo–Fc
electron density map of 25 is shown as an isomesh contoured
at 1σ; (B) overlay of the
crystal structure of RORγt bound to 25 and RORγt
bound to 3 (PDB code: 5C4O); and (C) zoom-in on the allosteric pocket
of RORγt showing the interactions between 25 and
the protein.
Co-crystal structure of RORγt with compound 25 (PDB code: 6SAL). (A) The tertiary structure of RORγt bound
to 25 (stick representation). The final 2Fo–Fc
electron density map of 25 is shown as an isomesh contoured
at 1σ; (B) overlay of the
crystal structure of RORγt bound to 25 and RORγt
bound to 3 (PDB code: 5C4O); and (C) zoom-in on the allosteric pocket
of RORγt showing the interactions between 25 and
the protein.
Isoxazole 25 Inhibits IL-17a Expression in
EL4 Cells
EL4 is a murine lymphoblast cell line that constitutively
expresses RORγt. Because RORγt promotes IL-17a production,
an effective means to determine the cellular activity of RORγt
inverse agonists is to measure the reduction in IL-17a mRNA expression
levels by quantitative reverse transcriptase PCR (RT-PCR). To this
end, EL4 cells were treated with 10 μM of 3, 25, and 23 for 24 h before IL-17a mRNA levels were measured
(Figure ). The most
potent isoxazole in vitro, 25, significantly reduced
IL-17a mRNA expression 27-fold, while the weaker inverse agonist 23 showed a smaller reduction (3.6-fold) compared to the DMSOcontrol. As expected, 3 led to the most significant decrease
in IL-17a expression (48-fold) which was in line with previous reports.
This result demonstrates that the allosteric modulation of RORγt
by optimized trisubstituted isoxazoles leads to an effective cellular
response, correlating with the biochemical protein binding data and
which is known to be beneficial for the treatment of autoimmune disease.[10−12]
Figure 6
IL-17a
mRNA expression in EL4 cells treated with ligand 3, 25, and 23 (10 μM, 24h) or DMSO. The level
of IL-17a expression was normalized to that of the GAPDH expression.
All data are expressed as the mean ± s.d. (n = 3). The relative gene expression was calculated by the 2–ΔΔCt (Livak) method using the DMSO control as calibrator.
IL-17a
mRNA expression in EL4 cells treated with ligand 3, 25, and 23 (10 μM, 24h) or DMSO. The level
of IL-17a expression was normalized to that of the GAPDH expression.
All data are expressed as the mean ± s.d. (n = 3). The relative gene expression was calculated by the 2–ΔΔCt (Livak) method using the DMSOcontrol as calibrator.
Absorption, Distribution, Metabolism, and
Excretion (ADME) Profile
To further assess the potential
of 25 and isoxazole analogues such as 23 and 8, we investigated the ADME profile of these compounds
and compared them to indazole 3 (Table ). The isoxazolecompounds showed favorable
profiles compared to 3 in terms of chemical stability,
solubility, and permeability through artificial plasma membranes (PAMPA).
A metabolic stability study with human liver microsomes indicated
that the 4-methylamino isoxazoles 23 and 25 were more liable to phase 1 metabolism compared to indazole 3, which showed good stability. 23 and 25 showed promising phase 2 stability. In blood plasma, while
inferior to 3, the stability of 23 and 25 was acceptable, although all these compounds showed high
levels of binding to plasma proteins. Pleasingly, the 5-phenyl-4-amido
isoxazole 8 showed a good ADME profile, with comparable
microsomal stability to 3 and reduced plasma protein
binding. This likely indicates that further optimization of the C-4
and C-5isoxazole substituents has the potential to produce candidate
molecules with desirable in vivo efficacy.
Table 5
ADME Properties for Ligands 3, 8, 23, and 25
microsomal
stability
compound
chemical stability
(% remain)
solubility (μM)
PAMPA (% flux)
phase 1 (CLint, μL/min/mg)
phase 2 (%
remain)
plasma stability (% remain)
plasma protein binding (% bound)
3
81.0
390
23.7
–1.2
47.1
100
99.9
8
95.4
490
60.1
–0.1
100
99.9
97.8
23
100
392
50.4
43.2
92.8
86.5
100
25
95.3
411
33.6
20.7
69.8
85.9
99.9
Conclusions
To summarize, we report
the design, synthesis, and early optimization of a novel class of
RORγt allosteric inverse agonists. The chemotype of the central
aromatic ring system differs significantly from all the other fused
bicyclic ring systems reported thus far. To identify this novel, more
diverse, molecular scaffold, we used the crystal structure of 3 bound to the RORγt allosteric site as the basis for
a 3D pharmacophore screen against a virtual compound library. Rational
design steps led to the discovery of the in silico designed hit 8, which already featured a modest inhibition of transcriptional
coactivator recruitment to the RORγt LBD and served as a starting
point for further optimization in a SAR campaign. A second and highly
efficient iteration of lead optimization was guided by in silico docking.
Through the synthesis of just five derivatives (Table ), this process delivered 25 (FM26), a submicromolar allosteric inverse agonist.
It is highly noteworthy that there was a strong correlation between
the Glide dockings scores and the RORγt biochemical activity
within this new class of isoxazole. Whereas screening approaches do
not overtly identify allosteric ligands, our rational scaffold hopping
approach is much more targeted, with less demand on experimental resource.
Overall, the discovery workflow adopted, with a central role for structure-driven
in silico screening and optimization, showed to be highly effective
and might have wider application in expediting NR allosteric drug
discovery.Competitive coactivator recruitment and ligand binding
assays were used to confirm the allosteric mode-of-action, with concomitant
cooperative RORγt binding with an inverse agonist. This was
also shown for thienopyrrazole 5, having not previously
been disclosed. The cocrystal structure of 25 in complex
with the RORγt LBD unequivocally proved the allosteric binding
mode, via a similar mechanism to 3 and was impressively
similar to the initially docked structure of 25 in RORγt.
The cocrystal of 25 with the RORγt LBD revealed
a number of unique interactions and structural RORγt modifications
that bring forward intriguing insights and new lines of exploration
regarding RORγt allosteric ligand binding, selectivity, and
affinity optimization, which are currently explored. Compound 25 was shown to significantly reduce IL-17a mRNA expression
in EL4 cells and to have a promising ADME profile. These factors highlight
the potential of this new isoxazole-based ligand class and overt targeting
of the RORγt allosteric site to deliver effective treatments
for autoimmune diseases.
Experimental Section
Pharmacophore Screening
The receptor–ligand
complex structure (PDB code: 4YPQ) was prepared using the Protein Preparation Wizard
within Maestro (version 2017–2, Schrödinger LLC, New
York, NY, USA) (default parameters). A 3D pharmacophore model for 3 bound to the allosteric pocket of RORγt LBD was created
using Phase (version 2017-2, Schrödinger LLC, default hypothesis
settings). Energy minimized 3D ligand conformations for each molecule
to be investigated were generated using the Ligand Preparation wizard
within Maestro (default parameters). These were screened against the
hypothesis whereby up to 50 ligand conformations were generated for
each molecule. A hit was returned for compounds that matched 4 out
of 6 pharmacophore features, and these were ranked using the Phase
Screen Score. The structure and ranking for the top 30 hits identified
from the Asinex Gold Platinum library can be found in the Supporting Information.
Molecular Docking Studies
The receptor–ligand
structure (PDB code: 4YPQ) was prepared as described above. A receptor grid centered on the
bound ligand was created using Glide (version 2017-2, Schrödinger
LLC). All parameters were kept as the default. Ligand libraries were
either enumerated in SMILES format using the open-access Chem-T software
or generated manually (see the Supporting Information). Ligands were prepared using the Ligand Preparation wizard as described
above. Ligands were docked using Glide (version 2017-2, Schrödinger
LLC) in standard precision mode with flexible ligand sampling. The
predicted binding modes of all ligands were ranked according to their
Glide Score (see Supporting Information for selected examples).
General Chemistry
All nonaqueous
reactions were performed under an argon atmosphere unless otherwise
stated. Water-sensitive reactions were performed in oven-dried glassware,
cooled under argon before use. Solvents were removed in vacuo using
a Büchi rotary evaporator and a diaphragm pump. THF and CH2Cl2 were dried and purified by means of a MBRAUN
Solvent Purification System (MB-SPS-800). Anhydrous DMF was obtained
in SureSeal bottles from Sigma-Aldrich. All other solvents used were
of chromatography or analytical grade and supplied by Biosolve or
Sigma-Aldrich. Commercially available starting materials were obtained
from Sigma-Aldrich, Fluka, Acros, Alfa-Aesar, or Fluorochem and were
used without further purification unless stated. TLC was carried out
on aluminum-backed silica (Merck silica gel 60 F254) plates supplied
by Merck. Visualization of the plates was achieved using an ultraviolet
lamp (λmax = 254 nm), KMnO4, anisaldehyde,
or ninhydrin. Column chromatography was either performed manually
using silica gel (60–63 um particle size) or using an automated
Grace Reveleris X2 chromatograph with prepacked silicacolumns supplied
by Buchi/Grace (40 μm particle size). LC–MS analysis
was carried out with a system comprising a Thermo Fischer LCQ Fleet
Ion Trap Mass Spectrometer and C18 Jupiter SuC4300A 150 × 2.0
mm column using a gradient of 5–100% MeCN in water (+ 0.1%
HCOOH) over 15 min. The purity of the samples was assessed using a
UV detector at 254 nm. Unless otherwise stated all final compounds
were >95% pure as judged by HPLC. GCMS analysis was performed on
a Phenomenex Zebron ZB-5MS 30 m × 0.25 mm × 0.25 mm column
with a gradient of 80 °C for 1 min to 300 °C for 1 min with
a rate of 30 °C/min in helium gas connected to a GCMS-QP2010
Plus Quadrupole Mass Spectrometer. High resolution mass spectra (HRMS)
were recorded using a Waters ACQUITY UPLC I-Class LC system coupled
to a Xevo G2 Quadrupole Time of Flight (Q-tof) mass spectrometer.
Proton (1H) and carbon (13C) NMR spectral data
were collected on a 400 MHz Bruker Cryomagnet or 400 MHz Varian Gemini.
Chemical shifts (δ) are quoted in parts per million (ppm) and
referenced to the residual solvent peak. Coupling constants (J) are quoted in Hertz (Hz) and splitting patterns reported
in an abbreviated manner: app. (apparent), s (singlet), d (doublet),
t (triplet), q (quartet), and m (multiplet). Assignments were made
with the aid of 2D COSY, HMQC, and HMBC experiments.
General Procedure for Ester Hydrolysis
LiOH·H2O (5.0 equiv) was added to a suspension
of ester (1.0 equiv) in a 4:1 mixture of MeOH-H2O (0.2
M). The reaction mixture was heated to 70 °C until TLC analysis
indicated complete consumption of the starting material. MeOH was
removed in vacuo, and the resulting aqueous mixture was acidified
to pH 3 using 10% v/v aqueous HCl and extracted with a 9:1 mixture
of CH2Cl2-MeOH (× 5). The combined organic
phase was dried (MgSO4), filtered, and concentrated in
vacuo to furnish an ester which was purified as described.
General Procedure for Amide Coupling
Carboxylic acids (1.0 equiv) were dissolved in SOCl2 (50
equiv) and heated to 50 °C for 2 h. The excess SOCl2 was removed in vacuo to furnish an acid chloride intermediate that
was immediately dissolved in CH2Cl2 (0.1 M).
To this was added NEt3 (3.0 equiv), the appropriate amine
or aniline (1.5 equiv), and DMAP (0.1 equiv), and the reaction mixture
was stirred at reflux for 18 h. The reaction mixture was diluted with
saturated aqueous NH4Cl solution and extracted with ethyl
acetate (3×). The combined organic phase was washed with brine,
dried over MgSO4, filtered, and concentrated in vacuo.
The crude product was purified by flash column chromatography (SiO2) using the specified eluent.
General Procedure for tert-Butyl Ester Deprotection
Esters (1.0 equiv) were treated
with a 20% trifluoroacetic acid solution in CH2Cl2 (0.2 M). The reaction mixture was stirred at the specified temperature
for the specified amount of time and then concentrated in vacuo. The
crude product was purified as indicated.
General Procedure for Suzuki Coupling
Under an inert atmosphere, the pinacol boronate (2.0 equiv), Cs2CO3 (2.0 equiv), and Pd(dppf)Cl2 (0.1
equiv) were added to a solution of bromide 30a (1.0 equiv)
in degassed DME. The reaction mixture was heated at 85 °C for
8 h, cooled to room temperature, diluted with saturated aqueous NH4Cl, and extracted with EtOAc (3×). The combined organic
phase was dried (MgSO4), filtered, and concentrated in
vacuo. The crude product was purified by flash column chromatography
using the specified eluent.
General Procedure for Conversion of Esters
to Aldehydes
LiAlH4 (1 M in THF, 1.0 equiv) was
added dropwise to a solution of ester (1.0 equiv) in THF (0.2 M) at
0 °C. The reaction mixture was warmed to room temperature and
stirred until TLC analysis indicated complete consumption of the starting
material. The reaction mixture was cooled to 0 °C, quenched by
the addition of saturated aqueous NH4Cl solution, and extracted
with EtOAc (3×). The combined organic phase was dried (MgSO4), filtered, and concentrated in vacuo. The resulting intermediate
product was dissolved in CH2Cl2 (0.2 M). To
this was added Dess-Martin Periodinane (1.5 equiv), and the reaction
mixture was stirred at room temperature until TLC analysis indicated
complete consumption of the intermediate. The reaction mixture was
quenched by the addition of 10% aqueous Na2S2O3 solution and extracted with CH2Cl2 (3×). The combined organic phase was washed with saturated
aqueous NaHCO3 and H2O, dried (MgSO4), filtered, and concentrated in vacuo to furnish the title compound
which was purified as described.
General Procedure for Reductive Amination
The chosen amine or aniline (1.0 equiv) was added to a solution
of the appropriate aldehyde (1.0 equiv) and AcOH (0.1 equiv) in MeOH
or EtOH (0.25 M). The reaction mixture was heated at reflux for 24
h and then concentrated in vacuo. The intermediate imine was isolated
by flash column chromatography using the specified eluent and then
dissolved in MeOH or EtOH (0.2 M), cooled to 0 °C (ice), and
treated with NaBH4 (5.0 equiv). The reaction mixture was
held at the specified temperature until TLC analysis indicated complete
consumption of the imine. The solvent was removed in vacuo, and the
mixture was dissolved in CH2Cl2 and washed with
water. The aqueous phase was further extracted with CH2Cl2 (2×), dried (MgSO4), filtered, and
concentrated in vacuo. The crude product was purified by flash column
chromatography (SiO2) using the specified eluent.
Hydroxylamine hydrochloride (3.95 g, 57.0
mmol, 1.2 equiv) was suspended in EtOH (20 mL) and 10% w/v aqueous
solution of NaOH (20 mL) was added such that the final pH of the resulting
solution was < pH 9. 2-Chloro-6-(trifluoromethyl)benzaldehyde (9.88
g, 47.4 mmol, 1.0 equiv) was then added as a solution in EtOH (20
mL) and the mixture was stirred at room temperature for 18 h. The
reaction mixture was diluted with H2O (100 mL) and extracted
with CH2Cl2 (3 × 100 mL). The combined
organic phase was dried (MgSO4), filtered, and concentrated
in vacuo to furnish 9 (8.77 g, 83%) as a colorless solid
which was used without further purification. R = 0.45 (4:1 c-hexane-EtOAc); 1HNMR (400 MHz, CDCl3): δ (ppm) 8.97 (1H, s, N=CH), 8.36 (1H, s, NOH), 7.67–7.64
(2 H, m, H-3, H-5), 7.45 (1 H, app. t, J = 8.0, H-4); 13CNMR (100 MHz, CDCl3): δ (ppm) 145.3 (N=CH), 135.7 (C-2), 133.6 (C-3), 131.3 (q, J = 31.2, C-6), 130.2 (C-4), 129.0 (C-1), 125.1 (q, J = 5.5, C-5), 123.22 (q, J = 274.2, F3C); LC–MS (ESI): calcd for C8H6ClF3NO [M + H]+: 224.00, observed: 224.00,
LC Rt: 5.82 min.
N-Chlorosuccinamide
(5.22 g, 39.1 mmol, 1.0 equiv) was added to a solution of hydroxylamine 9 (8.74 g, 39.1 mmol, 1.0 equiv) in DMF (80 mL). The reaction
mixture was stirred at 60 °C for 18 h then diluted with H2O (150 mL) and extracted with Et2O (3 × 100
mL). The combined organic phase was washed with H2O (3
× 100 mL) and brine (100 mL), dried (MgSO4), filtered,
and concentrated in vacuo to furnish 9a (9.10 g, 95%
purity, 86%) which was used immediately in the next step without further
purification. R = 0.42
(4:1 c-hexane-EtOAc); 1HNMR (400 MHz, CDCl3): δ (ppm) 8.50 (1H, s, NOH), 7.68–7.66
(2 H, m, H-3, H-5), 7.53 (1 H, app. t, J = 8.1, H-4).
Carboxylic acid 11b (2.0 g,
5.44 mmol,1.0 equiv) was dissolved in SOCl2 (10.0 mL, 138
mmol, 25 equiv) and heated to 60 °C for 2 h. Excess SOCl2 was removed in vacuo, and the intermediate acid chloride
was immediately dissolved in CH2Cl2 (20 mL)
and cooled to 0 °C. To this was added NEt3 (2.27 mL,
16.3 mmol, 3.0 equiv) and N,O-dimethylhydroxylamine hydrochloride (0.580 g, 5.98 mmol, 1.1 equiv). The reaction
mixture was allowed to warm to room temperature with stirring over
16 h before being quenched by the addition of saturated aqueous NH4Cl solution and extracted with CH2Cl2 (3 × 20 mL). The combined organic phase was dried (MgSO4), filtered, and concentrated in vacuo. The resulting Weinreb
amide was dissolved in THF (20 mL) and cooled to 0 °C. To this
was added LiAlH4 (1 M in THF, 2.72 mL, 2.72 mmol, 0.5 equiv),
and the reaction mixture was stirred at 0 °C for 1 h before being
quenched by the addition of saturated aqueous NH4Cl solution
(20 mL) and extracted with EtOAc (3 × 20 mL). The combined organic
phase was dried (MgSO4), filtered, and concentrated in
vacuo to furnish a crude product which was purified by automated flash
column chromatography, eluting with a gradient of 10–30% EtOAc
in n-heptane, to furnish 12 (1.25 g,
65%) as a colorless solid. R = 0.65 (1:1 c-hexane-EtOAc); 1HNMR (400 MHz, CDCl3): δ (ppm) 9.93 (1 H, s, CHO), 8.06–8.04
(2 H, m, PhH-ortho), 7.77 (1 H, d, J = 7.9, ArH-3
or ArH-5), 7.74 (1 H, d, J = 7.8, ArH-3 or ArH-5),
7.68–7.59 (4 H, m, H-4, PhH); 13CNMR (100 MHz,
CDCl3): δ (ppm) 182.8 (CHO), 174.6
(C-5), 160.8 (CO2Et), 158.6 (C-3), 136.3
(ArC-2), 133.2 (ArC-3), 132.7 (PhC-quart.), 132.0 (q, J = 30.9, ArC-6), 131.3 (ArC-4), 129.5 (PhC-ortho), 128.9 (PhC-meta),
125.9 (ArC-1), 125.7 (PhC-para), 125.0 (ArC-5), 123.0 (q, J = 274.9, F3C), 116.2 (C-4);
LC–MS (ESI): calcd for C17H10ClF3NO2 [M + H]+: 352.03, observed: 352.08,
LC Rt: 7.28 min.
NEt3 (0.760 mL, 5.44 mmol, 1.1
equiv) and diphenylphosphoryl azide (1.06 mL, 4.95 mmol, 1.0 equiv)
were added to a prewarmed solution of acid 11b (1.82
g, 4.95 mmol, 1.0 equiv) in t-BuOH (18 mL) at 50
°C. The reaction mixture was then heated to 85 °C for 18
h after which it was cooled to room temperature and diluted with 1
M aqueous HCl (50 mL) and extracted with EtOAc (3 × 30 mL). The
combined organic phase was washed with saturated aqueous NaHCO3 (100 mL) and brine (100 mL), dried (MgSO4), filtered,
and concentrated in vacuo. The crude product was purified by automated
flash column chromatography, eluting with a gradient of 0–20%
EtOAc in n-heptane, to furnish a carbamate (1.41
g, 65%) as a colorless solid. Trifluoroacetic acid (3.0 mL) was added
to a solution of the carbamate (1.13 g, 2.58 mmol, 1.0 equiv) in CH2Cl2 (9.0 mL). The reaction mixture was heated at
reflux for 4 h then cooled to room temperature and concentrated in
vacuo. The crude product was dissolved in CH2Cl2 (100 mL) and washed with saturated aqueous NaHCO3 (100
mL) and water (100 mL), dried (MgSO4), filtered, and concentrated
in vacuo to furnish 13 (0.798 g, 91%) as a pale yellow
solid that was used without further purification. R = 0.56 (1:1 c-hexane-EtOAc); 1HNMR (400 MHz, CDCl3): δ (ppm) 7.86 (1 H, d, J = 7.0, ArH-3 or ArH-5), 7.79–7.76 (2 H, m, PhH-ortho),
7.61 (1 H, app. t, J = 8.0, ArH-3 or ArH-5), 7.54–7.49
(3 H, m, PhH), 7.41 (1 H, app. t, J = 8.0, ArH-4),
2.97 (2 H, s, NH2); 13CNMR
(100 MHz, CDCl3): δ (ppm) 164.5 (C-5), 154.7 (C-3),
137.0 (ArC-2), 133.4 (ArC-3), 131.4 (PhC-quart.), 129.1 (PhC-ortho),
128.9 (9ArC-4), 128.3 (PhC-para), 126.5 (ArC-5), 126.0 (q, J = 4.0, ArC-1), 125.2 (PhC-meta), 122.9 (q, J = 274.4, CF3), 110.4 (C-4); LC–MS
(ESI): calcd for C16H11ClF3N2O [M + H]+: 339.04, observed: 339.08, LC Rt: 7.12 min.
NEt3 (5.80 mL, 41.5 mmol,
1.2 equiv) was added dropwise to a solution of imidoyl chloride 9a (8.90 g, 34.6 mmol, 1.0 equiv) in THF (110 mL). A white
precipitate formed immediately. The resulting suspension was stirred
vigorously at room temperature for 30 min and then filtered through
a pad of SiO2 that was subsequently washed with THF (250
mL). The solution was concentrated in vacuo to furnish 33 (8.25 g, 99%) as a colorless solid which was used immediately. R = 0.33 (9:1 c-hexane-EtOAc); 1HNMR (400 MHz, CDCl3): δ (ppm) 7.70 (1 H,
d, J = 8.2, H-3 or H-5), 7.67 (1 H, d, J = 8.2, H-3 or H-5), 7.54 (1 H, app. t, J = 8.2,
H-4).
Ester 38 (0.173 g, 0.340 mmol)
was treated according to the General Procedure for conversion of esters
to aldehydes. The crude product was purified by flash column chromatography,
eluting with a gradient of 0–10% EtOAc in n-heptane, to furnish aldehyde 43 (0.123 g, 80%) as a
colorless solid. R =
0.34 (9:1 n-heptane-EtOAc); 1HNMR (400
MHz, CDCl3): δ (ppm) 9.94 (1 H, s, CHO), 7.77 (1 H, d, J = 8.0, ArH-3 or ArH-5), 7.74
(1 H, d, J = 8.0, ArH-3 or ArH-5), 7.63–7.59
(2 H, m, phenol H-4 and phenolH-5), 7.51 (1 H, app. t, J = 2.0, phenolH-2), 7.46 (1 H, app. t, J = 8.0,
ArH-4), 7.11 (1 H, ddd, J = 8.2, 2.0, 1.0, phenolH-6), 1.02 (9 H, s, Si(CH3)2C(CH3)3), 0.27 (6 H, s, Si(CH3)2C(CH3)3); 13CNMR (100 MHz, CDCl3): δ (ppm) 182.8 (CHO), 174.4 (C-5), 158.5 (C-3), 156.6 (phenolC-3), 136.4 (ArC-2),
133.2 (ArC-3), 132.0 (q, J = 31.6, ArC-6), 131.3
(ArC-4), 130.6 (phenolC-5), 126.8 (ArC-1), 126.0 (phenolC-1), 125.0
(q, J = 5.1, ArC-5), 124.6 (phenolC-6), 123.0 (q, J = 274.5, CF3), 122.0 (phenolC-4), 120.3 (phenolC-2), 116.3 (C-4), 25.6 (Si(CH3)2C(CH3)3), 18.4 (Si(CH3)2C(CH3)3), −4.2 (Si(CH3)2C(CH3)3); LC–MS (ESI): calcd for C23H24ClF3NO3Si [M + H]+: 482.11, observed: 482.17, LC Rt: 9.45
min.
Biophysical Assays
RORγt-LBD Expression and Purification
(Used for TR-FRET Assays)
A pET15b expression vector encoding
the human RORγt LBD (residues 265–518) with an N-terminal His6-tag was transformed by heat shock
into BL21(DE3) E. coli cells. Single
colonies were used to inoculate precultures of 8 mL LB-media containing
100 μg/mL ampicillin. After overnight incubation at 37 °C,
each preculture was transferred to 1L TB media supplemented with ampicillin
(100 μg/mL) and incubated at 37 °C until an OD600 nm = 1.0 was reached. Protein expression was then induced with 0.5
mM isopropyl-b-d-thiogalactoside (IPTG), and cultures
were incubated for 16 h at 18°C. The cells were collected by
centrifugation and suspended in lysis buffer (300 mM NaCl, 20 mM TrisHCl
pH 8.0, 20 mM imidazole, 1 mM TCEP, 10% v/v glycerol, complete, EDTA-free
Protease Inhibitor Cocktail tablets (1 tablet/50 mL lysate) and benzonase
(0.1 μL/1 mL)). After lysis using a C3 Emulsiflex-C3 homogenizer
(Avestin), the cell lysate was cleared by centrifugation at 4 °C
and the protein was purified via Ni2+ affinity column chromatography.
Fractions containing the protein of interest were combined and dialyzed
against 150 mM NaCl, 20 mM Tris HCl pH 8.0, 5 mM DTT, and 10% v/v
glycerol.
TR-FRET Coactivator Recruitment Assay
Assays were conducted using 100 nM N-terminal
biotinylated SRC-1 box2 peptide (Biotin-N-PSSHSSLTARHKILHRLLQEGSPSD-CONH2) and 20 nM His6-RORγt-LBD or 100 nM His6-PPARγ-LBD in buffer containing 10 mM HEPES, 150 mM
NaCl, 5 mM DTT, 0.1% BSA (w/v), and 0.1 mM CHAPS, pH 7.5. A terbium
labeled anti-His antibody (CisBio Bioassays, 61HISTLA) and D2-labeled
streptavidin (CisBio Bioassays, 610SADLA) were used at the concentrations
recommended by the supplier. In the case of PPARγ, the assay
was performed in the presence of 1 μM rosiglitazone, in order
to initially activate PPARγ. Compounds (dissolved in DMSO) were
titrated using a 2× dilution series in Corning white low volume,
low binding, 384-well plates at a final volume of 10 μL. The
final DMSOconcentration was 2% v/v throughout. The plate was incubated
at room temperature for 30 min and centrifuged before reading (excitation
= 340 nm; emission = 665 and 620 nm) on a Tecan infinite F500 plate
reader using the parameters recommended by CisBio Bioassays. The data
were analyzed with Origin Software. The dose–response curve
was fitted represented bywhere y is the FRET ratio, A1 is the bottom asymptote, A2 is the top asymptote, p is the Hills slope, and x is the ligand concentration. Where dose–response
curves did not reach a bottom asymptote, this was fixed at the value
of the negative control. (Data recorded in triplicate; error shown
is standard deviation from the mean; curves are representative of
>3 repeated experiments).
Competition TR-FRET Coactivator Recruitment
Assay
Competition assays were performed in an analogous fashion
to that described above only in the presence of fixed concentrations
of cholesterol: 0 μM (DMSO), 0.25 μM, 1.0 μM, such
that the final concentration of DMSO remained at 1.2% v/v.
Ligand Binding TR-FRET Assay
Assays
were conducted using 100 nM Alexa647-labeled MRL-871 and 20 nM His6-RORγt-LBD in buffer as described above. A terbium-labeled
anti-His antibody (CisBio Bioassays, 61HISTLA) was used at the concentrations
recommended by the supplier. The assay was carried out in Corning
black low volume, low binding, 384-well plates at a final volume of
10 μL in the same manner as described above.
Protein X-ray Crystallography
RORγt-LBD Expression and Purification
(Used for Crystallography)
A pET15b expression vector was
ordered from GenScript encoding for the RORγt LBD (residues
265–507) containing a C455H mutation (RORγtC455H) and
a C-terminal His-tag. The plasmid was transformed by heat shock into
BL21(DE3) E. coli cells. A single colony
was used to start three precultures of 24 mL LB medium containing
100 μg/mL ampicillin. After overnight incubation at 37 °C,
each preculture was transferred to 2 L of 2× YT medium supplied
with 0.05% antifoam SE-15 (Sigma-Aldrich). These cultures were incubated
until they reached an OD600 = 0.6. Protein expression was
induced by adding 0.25 mM IPTG. The temperature was decreased to 15
°C, and protein expression proceeded overnight. The cells were
collected by centrifugation at 10.000 RCF for 10 min at 4 °C.
The resulting 30 g of cell pellet was dissolved in lysis buffer (20
mM Tris, 500 mM NaCl, 2 mM TCEP, 0.1% Tween20, 10% glycerol, 10 cOmplete
Protease Inhibitor Cocktail tablets (Roche), and 25 U/mL Bezonase
Nuclease (Millipore), adjusted to pH = 8.0). After cell lysis using
an Emulsiflex-C3 homogenizer (Avestin), the cell lysate was cleared
by centrifugation at 40.000 RCF for 40 min at 4 °C, and the supernatant
was loaded on a 5 mL Ni-NTA Superflow cartridge (QIAGEN) pre-equilibrated
with buffer A (20 mM Tris, 500 mM NaCl, 2 mM TCEP, 0.1% Tween20, and
10% glycerol). The column was washed with 10 CVs of buffer A supplied
with 20 mM and sequentially with 10 CVs of Buffer A supplied with
50 mM imidazole. The protein was eluted from the resin using an eight
column volumes elution buffer (buffer A supplied with 200 mM imidazole).
The purified protein was then dialyzed overnight to buffer A containing
1.2 U of restriction-grade thrombin (Millipore) per milligram of purified
protein to remove the His-tag. Next, the protein mixture was concentrated
using an Amicon Ultra centrifugal filter with a 10-kDa cutoff (Millipore)
and loaded on a Superdex 75 pg 16/60 size-exclusion column (GE Life
Sciences) using 20 mM Tris, 100 mM NaCl, and 5 mM DTT (adjusted to
pH = 8.0) as a running buffer. The flow-through was collected as 3
mL fractions which were analyzed using Q-ToF LC–MS. The fractions
containing RORγtC455H were combined and concentrated to a final
concentration of 11.1 mg/mL. The concentrated protein sample was then
aliquoted, flash-frozen, and stored at −80 °C.
X-ray Crystallography
The RORγtC455H
solution (11.1 mg/mL) was mixed with 2 equiv of 25 and
incubated on ice for 1 h. Next, the sample was centrifuged at 20.000
RCF for 20 min at 4 °C to remove precipitated proteins. MRC-2
well crystallization plates (Hampton Research, sitting drop) were
prepared using a Mosquito pipetting robot (TTP Labtech). Well-diffracting
crystals were obtained by mixing 0.9 ul of protein solution with 0.3
μL of 1.6–2.0 M ammonium sulfate and 0.1 M Tris (pH =
8.5). The well was filled with 80 μL precipitant solution, and
plates were placed at 20 °C. Crystals could be observed after
1 h of incubation and grew to their final size overnight. The crystals
were cryoprotected by transferring the crystals briefly to a solution
containing 1.2 M AmSO4, 0.1 M Tris (pH = 8.5), and 25% glycerol before
being flash cooled in liquid N2. Diffraction data were
collected at 100 K at the P11 beamline of the PETRA III facility at
DESY (Hamburg, Germany) and processed using the CCP4 suite (version
7.0.075).[46] DIALS was used to integrate
and scale the data.[47] The data was phased
with PHASER using 5C4O as a molecular replacement model and ligand
restraints of 25 were generated with AceDRG.[48,49] Sequential model building and refinement were performed with COOT
and REFMAC, respectively.[50,51] PyMOL (version 2.2.3,
Schrödinger) was used to make the figures.[52] The structure of RORγtC455 in complex with 25 was deposited in the protein data bank (PDB) under code 6SAL.
Quantitative IL-17a mRNA RT-PCR Assay
EL4 cells (Sigma-Aldrich) were grown in DMEM (Gibco) with 10% FBS.
At 24 h after the cells were seeded onto a 12-well plate, the cells
were incubated with 10 μM compound (from 10 mM stock in DMSO)
or DMSO for 24 h and activated with phorbol 12-myristate 13-acetate
(PMA, 50 ng/mL; Sigma-Aldrich) and ionomycin (1 μg/mL; Sigma-Aldrich)
for 5 h. The cells were then collected, and RNA was isolated using
a RNeasy Mini Kit (Qiagen) and reverse transcribed using the iScript
Advanced cDNA Synthesis Kit (Bio-Rad). Quantitative RT-PCR was performed
to analyze mRNA levels of mouseIL-17a levels (in triplicate) using
SYBR green technology (Bio-Rad) on a CFX Real-Time System (Bio-Rad).
The following primer assays were purchased from Bio-Rad: IL-17a (qMmuCID0026592)
and Gapdh (qMmuCED0027497). The level of IL-17a mRNA expression was
normalized to that of Gapdh expression. The relative gene expression
was calculated by the 2–ΔΔCt (Livak)
method using the DMSOcontrol as calibrator. (Data recorded in triplicate;
error shown is standard deviation from the mean; data are representative
of >3 repeated experiments).
Absorption, Distribution, Metabolism, and
Excretion Experiments
Chemical Stability
Chemical stability
was determined by incubating test compounds at a final concentration
of 2 μM in aqueous buffer at pH 7.4 for 1, 7, and 24 h, respectively.
The percentage of remaining compound (% remain) in relation to the
zero time point was calculated following LC–MS-based measurement
of sample aliquots of each time point.
Kinetic Solubility
Aqueous solubility
of compounds was determined by spectrophotometrical measurement of
the kinetic solubility of a 500 μM compound solution in aqueous
buffer pH 7.4 compared to a solution in the organic solvent acetonitrile
after 90 min of vigorous shaking at room temperature.
PAMPA
Permeability through artificial
membranes (PAMPA) was performed at an initial concentration of 500
μM of the compound in the donorcompartment. After an incubation
period of 20 h, absorption of the receiver wells was measured by spectrophotometry
and permeation was calculated by normalization of the compound flux
across a blank filter.
Microsomal Stability Phase I
Metabolic
stability under oxidative conditions was measured in liver microsomes
from different species by LC–MS-based measuring of depletion
of compound at a concentration of 3 μM over time up to 50 min
at 37 °C. On the basis of compound half-life t1/2, in vitro intrinsic clearance CLint was
calculated.
Microsomal Stability Phase II
Metabolic
stability under conjugative conditions was measured in the glucuronidation
assay by LC–MS-based determination of % remaining of selected
compounds at a concentration of 5 μM in incubations with liver
microsomes supplemented with UDPGA for 1 h at 37 °C.
Plasma Stability
Plasma stability
was measured by LC–MS-based determination of % remaining of
selected compounds at a concentration of 5 μM after incubation
in 100% plasma obtained from different species for 1 h at 37 °C.
Plasma Protein Binding
Assessment
of plasma protein binding was measured by equilibrium dialysis by
incubating plasma with the compound of interest at a concentration
of 5 μM for 6 h at 37 °C followed by LC–MS-based
determination of final compound concentrations.
Authors: Richard A Friesner; Jay L Banks; Robert B Murphy; Thomas A Halgren; Jasna J Klicic; Daniel T Mainz; Matthew P Repasky; Eric H Knoll; Mee Shelley; Jason K Perry; David E Shaw; Perry Francis; Peter S Shenkin Journal: J Med Chem Date: 2004-03-25 Impact factor: 7.446
Authors: R M V Abreu; H J C Froufe; P O M Daniel; M J R P Queiroz; I C F R Ferreira Journal: SAR QSAR Environ Res Date: 2011 Jul-Sep Impact factor: 3.000
Authors: Christopher Lock; Guy Hermans; Rosetta Pedotti; Andrea Brendolan; Eric Schadt; Hideki Garren; Annette Langer-Gould; Samuel Strober; Barbara Cannella; John Allard; Paul Klonowski; Angela Austin; Nagin Lad; Naftali Kaminski; Stephen J Galli; Jorge R Oksenberg; Cedric S Raine; Renu Heller; Lawrence Steinman Journal: Nat Med Date: 2002-05 Impact factor: 53.440
Authors: Garib N Murshudov; Pavol Skubák; Andrey A Lebedev; Navraj S Pannu; Roberto A Steiner; Robert A Nicholls; Martyn D Winn; Fei Long; Alexei A Vagin Journal: Acta Crystallogr D Biol Crystallogr Date: 2011-03-18
Authors: Junjian Wang; June X Zou; Xiaoqian Xue; Demin Cai; Yan Zhang; Zhijian Duan; Qiuping Xiang; Joy C Yang; Maggie C Louie; Alexander D Borowsky; Allen C Gao; Christopher P Evans; Kit S Lam; Jianzhen Xu; Hsing-Jien Kung; Ronald M Evans; Yong Xu; Hong-Wu Chen Journal: Nat Med Date: 2016-03-28 Impact factor: 53.440
Authors: Femke A Meijer; Annet O W M Saris; Richard G Doveston; Guido J M Oerlemans; Rens M J M de Vries; Bente A Somsen; Anke Unger; Bert Klebl; Christian Ottmann; Peter J Cossar; Luc Brunsveld Journal: J Med Chem Date: 2021-05-19 Impact factor: 7.446
Authors: Rens M J M de Vries; Femke A Meijer; Richard G Doveston; Iris A Leijten-van de Gevel; Luc Brunsveld Journal: Proc Natl Acad Sci U S A Date: 2021-02-09 Impact factor: 11.205
Authors: Femke A Meijer; Maxime C M van den Oetelaar; Richard G Doveston; Ella N R Sampers; Luc Brunsveld Journal: ACS Med Chem Lett Date: 2021-03-08 Impact factor: 4.345
Figure 1
Figure 2
Scheme 1
Scheme 2
Figure 3
Scheme 3
Scheme 4
Figure 4
Figure 5
Figure 6