Optimizing a Weakly Binding Fragment into a Potent ROR?t Inverse Agonist with Efficacy in an in Vivo Inflammation Model

Article abstract of DOI:10.1021/acs.jmedchem.8b00529

The transcription factor ROR?t is an attractive drug-target due to its role in the differentiation of IL-17 producing Th17 cells that play a critical role in the etiopathology of several autoimmune diseases. Identification of starting points for ROR?t inverse agonists with good properties has been a challenge. We report the identification of a fragment hit and its conversion into a potent inverse agonist through fragment optimization, growing and merging efforts. Further analysis of the binding mode revealed that inverse agonism was achieved by an unusual mechanism. In contrast to other reported inverse agonists, there is no direct interaction or displacement of helix 12 observed in the crystal structure. Nevertheless, compound 9 proved to be efficacious in a delayed-type hypersensitivity (DTH) inflammation model in rats.

Full text of DOI:10.1021/acs.jmedchem.8b00529

Subscriber access provided by UNIV OF DURHAM
Article
Optimizing a weakly binding fragment into a potent ROR#t
inverse agonist with efficacy in an in vivo inflammation model
David Carcache, Anna Vulpetti, Joerg Kallen, Henri Mattes, David Orain, Rowan Stringer, Eric Vangrevelinghe,
Romain M Wolf, Klemens Kaupmann, Johannes Ottl, Janet Dawson King, Nigel G. Cooke, Klemens
Hoegenauer, Andreas Billich, Juergen Wagner, Christine Guntermann, and Samuel Hintermann
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b00529 • Publication Date (Web): 10 Jul 2018
Downloaded from http://pubs.acs.org on July 10, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Optimizing a weakly binding fragment into a potent
RORγt inverse agonist with efficacy in an in vivo
inflammation model
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
David Carcache,†,# Anna Vulpetti,†,#,* Joerg Kallen,§,# Henri Mattes,† David Orain,† Rowan
Stringer,ǁ Eric Vangrevelinghe,† Romain M. Wolf,† Klemens Kaupmann,‡ Johannes Ottl,§ Janet
Dawson,‡ Nigel G. Cooke,† Klemens Hoegenauer,† Andreas Billich,‡ Juergen Wagner,† Christine
Guntermann,‡ and Samuel Hintermann†,*
†Global Discovery Chemistry, §Chemical Biology & Therapeutics, ǁPK Sciences, and
‡Autoimmunity, Transplantation and Inflammation, Novartis Institutes for BioMedical Research,
4002 Basel, Switzerland
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 47
1
2
3
4
5
6
7
8
9
ABSTRACT: The transcription factor RORγt is an attractive drugꢀtarget due to its role in the
differentiation of ILꢀ17 producing Th17 cells that play a critical role in the etiopathology of
several autoimmune diseases. Identification of starting points for RORγt inverse agonists with
good properties has been a challenge. We report the identification of a fragment hit and its
conversion into a potent inverse agonist through fragment optimization, growing and merging
efforts. Further analysis of the binding mode revealed that inverse agonism was achieved by an
unusual mechanism. In contrast to other reported inverse agonists, there is no direct interaction
or displacement of helix 12 observed in the crystal structure. Nevertheless, compound 9 proved
to be efficacious in a delayedꢀtype hypersensitivity (DTH) inflammation model in rats.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
INTRODUCTION
The T helper 17 (Th17) cell pathway has been strongly linked to autoimmunity; indeed
therapeutic interventions with inhibitors of different components of the pathway such as ILꢀ17A,
ILꢀ17RA, and ILꢀ23 have shown beneficial effects in multiple autoimmune diseases. Due to its
role in the differentiation of Th17 cells, the RORγ2 (also called RORγt) isoform of the nuclear
hormone receptor retinoic acid receptorꢀrelated orphan receptor C (RORC) is an additional
attractive drugꢀtarget.1ꢀ6 Besides thymusꢀ and Tꢀcellꢀspecific RORγt the widely expressed
RORCvar1 (RORγ1) is also known. Due to the structural identity of the ligandꢀbinding domains
of RORγ1 and RORγt, lowꢀmolecular weight inhibitors will inevitably target both isoforms.
Many RORγt inhibitors have been described covering a large diversity of chemical structures.7ꢀ10
For some, inhibition of inflammation in in vivo rodent models of autoimmune disease has been
reported.2ꢀ5,12ꢀ14 A number of compounds such as VTPꢀ43742,8 JNJꢀ3534, AZDꢀ0284 and JTEꢀ
451 have advanced into clinical studies. For the most advanced candidate VTPꢀ43742, Vitae
ACS Paragon Plus Environment
2

Page 3 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Pharmaceuticals reported the successful outcome of a fourꢀweek proofꢀofꢀconcept trial in
psoriasis patients but recently, clinical development of the compound was stopped.15
During early hitꢀfinding efforts to identify starting points for RORγt inhibitors, it became
apparent that most of the hits had poor physicochemical properties. This may not be surprising
considering the rather large and hydrophobic ligand binding pocket (LBP) of RORγt. However,
there are also regions of the LBP where polar interactions are possible, such as the ‘sulfate
pocket’.16 Our aim was to screen for fragments targeting these more favorable regions. Here we
report the identification of a weak binder from a biophysical screen and describe how this hit was
converted into a potent RORγt inverse agonist with an unusual indirect mode of action. This tool
compound was tested in a model of delayedꢀtype hypersensitivity in rats (proofꢀofꢀconcept
study) and demonstrated efficacy, thus confirming the functional activity of this novel binding
mode.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
RESULTS AND DISCUSSION
FRAGMENTꢀBASED SCREENING
To complement classical highꢀthroughput biochemical and cellular screening approaches, a
series of fragmentꢀbased screening (FBS) campaigns were carried out to discover novel starting
points for medicinal chemistry optimization. The most attractive hits discovered by screening a
proprietary fragment library with Differential Static Light Scattering (DSLS) were tested in coꢀ
crystallization trials with RORγt in order to obtain structural information on their binding modes.
Among the 13 hits successfully coꢀcrystallized, phenylꢀacetamide 1 with an IC50 value of 216
ꢀM in an MS affinity assay was identified as a preferred starting point for structureꢀbased
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 47
1
2
3
4
5
6
7
8
9
design, despite its moderate LE (0.26) and LLE (0.23). The Xꢀray coꢀcrystal structure of 1 bound
to the RORγt (Y264ꢀK518) construct and a RIP140 coꢀactivator peptide1 (receptorꢀinteracting
protein 140), solved at 1.80 Å resolution, revealed that 1 sits in the middle of the large LBP well
anchored by two hydrogen bonds (Hꢀbonds). It makes a direct Hꢀbond with the backbone
carbonyl of F377 as well as waterꢀmediated Hꢀbonds with the carbonyl backbone of H323 and
the sideꢀchain nitrogen of Q286. In addition the phenyl ring stacks nicely below F378 in an edgeꢀ
toꢀface fashion. Finally, the chlorine substituent of 1 points towards the sideꢀchain of C320
forming hydrophobic contacts. Based on the analysis of the crystal structure, we concluded that
this fragment could provide an attractive starting point for fragment growing and merging
approaches towards other regions of the large RORγt binding cavity. These regions are referred
as the ‘polar end region’ (or ’sulfate pocket‘), the ‘unexplored back pocket’, the ‘hydrophobic
hot spot’ and the ‘business end’ as reported previously.16
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
4

Page 5 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1: Crystal structure of RORγt (Y264ꢀK518) in complex with 1 and RIP140 peptide
(depicted in violet) (PDB access code = 6FZU).
IN SILICO ACTIVE SITE ANALYSIS
A short description and visualization of the structural characteristics of each of these subꢀpockets
is provided in this section, along with the SiteMap analysis of compound 217 (Figure 2) for which
we had solved the Xꢀray structure earlier. The Schrödinger SiteMap tool18 provides a wealth of
information in the form of computed properties and graphical contour maps, distinguishing
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 47
1
2
3
4
5
6
7
8
9
regions in a binding pocket that are favorable for occupancy by hydrophobic, Hꢀbond donor and
Hꢀbond acceptor ligand groups. The hydrophobic (depicted in yellow), donor (blue) and acceptor
(red) maps for the structure are shown in Figure 2. The pocket is almost completely enclosed and
mainly hydrophobic. This has been recognized as a challenge to balance potency and favorable
physicochemical properties in a single ligand. Opportunities for polar contacts are found in the
’polar end region’, in the region underneath the βꢀsheet portion and at the bottom of the
‘unexplored backꢀpocket’. The ‘polar end region’ is lined by the R367 side chain and the
backboneꢀNH of L287. These groups interact favorably with the SO2 motif of the ligand 2. The
βꢀsheet portion of the protein closes the pocket from the top and projects the backbone carbonyl
of F377 and the backbone NH of E379 into it (not shown in Figure 1). The backbone carbonyl of
F377 makes an Hꢀbond with the NH of the amide of 2 while the CO of the amide makes a water
mediated interaction with the sideꢀchain of Q286 and with the backboneꢀCO of H323. We called
the backꢀpocket shown in Figure 2 ‘unexplored’ because there were no ligand or fragment bound
crystal structures available at that time occupying this region. Later, 1,3ꢀdihydroꢀ2,1,3ꢀ
benzothiadiazole 2,2ꢀdioxide inverse agonists were described that occupy parts of the same subꢀ
pocket close to Ser404 with the sideꢀchain of M365 adopting a different orientation.19 This subꢀ
pocket is surrounded by the side chains of hydrophobic amino acids: L362, M365, V376 and
I400. However, the most buried region of this pocket is polar due to the presence of the S404
side chain. The ‘hydrophobic hot spot’ region was labelled as such due to the yellow
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
hydrophobic isoꢀcontour map (visible at ꢀ1.0 kcal/mol) but also based on the observation that
most of the coꢀcrystallized fragments present hydrophobic moieties filling this region. The side
chains of the hydrophobic residues lining the ‘hydrophobic hot spot’ region are F388, I397, I400
and F401. The ‘business end’ region points towards helices 11 and 12. This portion of RORγt is
ACS Paragon Plus Environment
6

Page 7 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
considered more flexible, as revealed by various crystal structures with agonists, antagonists and
inverse agonists.
7
8
9
All the fragments solved by crystal structure determination were bound to an agonist
conformation of the RORγt (Y264ꢀK518) construct with the RIP140 peptide bound as
exemplified for ligand 2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2: Crystal structure of RORγt (Y264ꢀK518) in complex with 2 (PDB ID code 6G05) and
RIP140 peptide (not shown). The binding pocket generated by SiteMap is shown as a gray
surface. The ‘site points’, shown in white, cover a large region comprising various subꢀpockets
labelled in red. The hydrophobic and Hꢀbond donor and acceptor maps are shown in yellow (ꢀ1.0
kcal/mol isoꢀcontour), blue (ꢀ10 kcal/mol), and red (ꢀ10 kcal/mol), respectively.
MEDICINAL CHEMISTRY STRATEGY
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 47
1
2
3
4
5
6
7
8
9
Based on our understanding of the molecular mechanism for RORγt inverse agonism, which we
have recently reported,16 it was anticipated that fragment 1, which binds into the central region of
the RORγt binding cavity, would be devoid of any functional activity. It was recognized early on
that growing the fragment further towards the ’business end‘ from the ethoxy substituent of 1
would be a preferred strategy to achieve RORγt inverse agonism.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1: Medicinal chemistry strategy towards the identification of the potent RORγt inverse
agonist 9
The aim of our initial medicinal chemistry strategy was to determine the SAR of the ethoxy
substituent, and explore the feasibility of attaching larger substituents to occupy the ‘business
end’ and demonstrate functional activity (Scheme 1). The cyclopropoxy analog 3 resulted in a
moderate 3ꢀfold improvement of the binding affinity for RORγt, whereas the isopropoxy
compound 4 displayed a greater than 10ꢀfold improvement in binding affinity compared to 1
ACS Paragon Plus Environment
8

Page 9 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(Table 1). The trifluoroisopropoxy substituent in compound 5 was shown to further improve the
binding affinity, likely due to an enhanced interaction with the ‘hydrophobic hot spot’ region, but
this was achieved at the expense of ligand efficiency and was not pursued further. Encouraged by
these initial results, the benzyl ether 6 was designed with the aim of extending to the ‘business
end’ and achieving functional activity. Both enantiomers were prepared and compounds 6 and 7
showed comparable binding affinity for RORγt in the MS reporter assay whilst only the (S)ꢀ
enantiomer 6 showed first hints of inverse agonism in a FRET assay (Table 2).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. Fragment growing towards the ‘business end’
MS reporter
Compound
R
IC50 [µM]a
Et
216 ± 15
1
cPr
iPr
69 ±10
19 ± 2
9 ± 1
3
4
5
6
6 ± 1
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 47
1
2
3
4
15 ± 1
7
5
6
7
8
9
a Binding assay: Values were calculated from two replicates (8ꢀpoint dose response). See
Supporting Information for assay details.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Having achieved functional activity by growing the fragment towards the business end, we
decided to use only the functional assay for the further optimization. The Affinity MS reporter
assay was geared towards a very robust and direct quantification of compound binding to the
target pocket in an affinity window of 0.7 µM to 2 mM whereas the functional assay was
designed to assess compounds with high affinities (< 10 µM).
Next, we wanted to replace the central phenyl ring with a heterocycle since the Nꢀ(4ꢀ
hydroxyphenyl)acetamide motif presents a high risk to form an iminoquinone species upon
oxidative metabolism. Thus, the 3ꢀchloropyridine analog 8 was synthesized and shown to
maintain similar potency. Gratifyingly, this compound displayed full inverse agonism.
Table 2. Fragment merging towards the potent RORγt inverse agonist 9
FRET IC50
Compound
X
R
R1
[nM]a
ACS Paragon Plus Environment
10

Page 11 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CH
CH
N
Me
Me
Me
4733 ± 2333
> 10000
6
7
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1146 ± 118
0.6 ± 0.7
27 ± 2
8
N
9
N
10
11
N
3.1 ± 0.9
a Functional TR FRET assay: Values were calculated from at least two independent
measurements (8ꢀpoint dose response).25
Whilst 8 represented the first full RORγt inverse agonist in this series, its potency on RORγt was
clearly two to three orders of magnitude too weak to be considered as a drug candidate. A
fragment merging approach was used to further evolve the series. Among the many Xꢀray
structures which were available inꢀhouse, our attention turned to compound 2,17 which occupies
the ’polar end subꢀpocket’ with a large substituent. By overlaying the crystal structure of 2 with
the crystal structure of fragment 1, both in complex with RORγt, the Nꢀacetyl moiety in both
compounds perfectly overlaps. The NH acts as Hꢀbond donor to the carbonyl backbone of F377,
while the acetyl CO accepts a waterꢀmediated interaction with H323 and Q286 (Figure 3). The
‘polar end’ region offers many productive interactions with the sulfone moiety, and increased
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 47
1
2
3
4
affinity was expected by growing into this region. The SO2 group of 2 makes Hꢀbond
5
6
interactions with both the sideꢀchain of R367 and the backbone NH of L287.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3: Overlay of the crystal structure of RORγt (Y264ꢀK518) in complex with 1 (green) and
2 (white) and RIP140 peptide (not shown). The LBP is represented as cavity (atomꢀbased
colored). The coordinates for 1 and 2 have been deposited in the PDB databank (PDB access
codes = 6FZU, 6G05).
This merging strategy was successful and led to the identification of compound 9 (Scheme 1) as
a breakthrough RORγt inverse agonist with subꢀnanomolar potency in the FRET assay. To better
ACS Paragon Plus Environment
12

Page 13 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
understand the influence of the configuration at the benzylic position on potency, the (R)ꢀ
enantiomer 10 was also prepared and showed a 45ꢀfold lower potency compared to the (S)ꢀ
enantiomer, suggesting that the absolute configuration of the benzylic substituent is critical for
RORγt potency. The corresponding desꢀmethyl analog 11 turned out to be a weaker inverse
agonist, confirming the importance of occupying the so called ‘hydrophobic hot spot’ (Figure 1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CHEMISTRY
A threeꢀstep synthetic sequence was developed for the preparation of compounds described here.
This sequence relied on a key nucleophilic aromatic substitution (SNAr). As shown in Scheme 2,
starting from commercially available 4ꢀhalogenoꢀnitroarenes (12 and 13), the 4ꢀhalogeno group
was displaced with an alcohol nucleophile, generated with sodium or potassium hydride, to form
the etherꢀintermediates. Béchamp reduction of the nitro group with iron provided the
corresponding anilines and 3ꢀaminopyridines (14 – 17). These aromatic amines were then treated
with acetyl chloride to obtain the Nꢀacetyl analogs (3 and 5 ꢀ8) or coupled with the
corresponding aryl acetic acids to give the final compounds 9 – 11, 18 and 19.
Scheme 2. General synthetic routes for the preparation of compounds 3 – 7, 8 – 11, 18 and 19
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 47
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ANALYZING THE MODE OF INHIBITION OF COMPOUND 9
A coꢀcrystal structure of 9 bound to the RORγt (Y264ꢀK518) construct, solved at 1.90 Å
resolution, revealed a surprising orientation for the benzyl ether side chain (Figure 4). Instead of
occupying the ‘businessꢀend region’, as assumed in our original design, the side chain is located
in the ‘unexplored back pocket’ formed by L362, M365, V376, I400 and S404. In the crystal
structure of 9, the sideꢀchain of M365 is shifted towards the front, and this movement creates an
enlarged cavity which perfectly accommodates the phenethoxy substituent. The phenyl ring is
located between the sideꢀchains of M365 and I400 forming many van der Waals contacts.
Interestingly, a superposition of the coꢀcrystal structures for 9 and the original fragment 1 (see
Figure S1) shows that the (5ꢀchloropyridin)acetamide and (3ꢀchlorophenyl)acetamide moieties
ACS Paragon Plus Environment
14

Page 15 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
superpose very well (largest differences of ca. 0.4 Å). This confirms that the fragment growing
towards both the ‘polar end’ and the ‘unexplored pocket’ did not change the original fragment
binding mode.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 47
1
2
3
4
5
6
7
8
9
Figure 4: Chain A of crystal structure of RORγt (Y264ꢀK518) in complex with 9 and RIP140
peptide (not shown). There are four complexes in the asymmetric unit. The distance between
NE2ꢀH479 and OHꢀY502 in the four chains ranges from 2.8 to 3.3 Å (corresponding to medium
to weak Hꢀbond energies). The coordinates have been deposited in the PDB databank (PDB
access code = 6G07).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In order to gain further insight into the contributions of each part of molecule 9 to the binding
potency, compounds 18 and 20 (Scheme 3) were tested for their binding affinity to RORγt.
Scheme 3. Truncated versions and a pyridyl analog of 9
Compound 18, which is the result from merging the pyridine analog of the original hit 1 with the
ethylsulfonylphenylacetamideꢀfragment, displayed a slightly lower gain of potency compared to
the merging of 8 with the same fragment (i.e. compound 9). The phenylacetamide 20 did not
show any measurable affinity for RORγt up to 500 µM thereby confirming the importance of
occupying at the same time the ‘polar end region’, the unexplored back pocket’ and the
‘hydrophobic hot spot’. To better understand the role of the ‘hydrophobic hot spot’, compound 9
ACS Paragon Plus Environment
16

Page 17 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and the desꢀmethyl analog 11 were selected to characterize their binding kinetics (kon and koff,
see Table 3). Binding kinetics was experimentally measured using a reporter displacement assay
as described.20 In summary, the proximity between a suitable reporter and the protein results in
the emission of an optical signal. Compounds that bind to the same site as the reporter probe
displace the probe, causing signal diminution. Reporter displacement is measured over time after
addition of compounds at various concentrations for deriving Kd, kon and koff.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 3: Binding kinetics of compounds 9 and 11
Compound Kd [nM] kon [1/s 1/M] koff [1/s] Residence
time [min]
Targeted MD Action ‘S’
[energy mol-1 time -1];
(exit pathway)
35.5
5.28·103
1.31·104
1.88·10ꢀ4 89
6.1·10ꢀ3
16.47 (a); 15.49 (b)
7.10 (a); 7.20 (b)
9
465.3
3
11
Compounds 9 and 11 show slow on and off binding kinetics. This finding prompted us to explore
the most probable ligand unbinding pathways by carrying out targeted Molecular Dynamics
(MD) simulations21 for both 9 and 11 to evaluate whether the in silico calculated ligand
resistance to dissociation would qualitatively correlate with the experimental koff. Molecular
dynamics studies revealed the two most probable unbinding pathways (Figure 5a) which
supported the findings of others in the nuclear receptor research field.22 Analysis of the dynamics
of the protein without 9 bound with respect to the two lowest energy unbinding pathways for 9,
named pathway (a) and (b), are shown in Figure 5B.
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 47
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5: A) Predicted exit pathways for 9 from the RORγt (Y264ꢀK518) complex. The
trajectories are represented by dots connecting 1 psꢀspaced MD frames of the atom on which the
targeted MD was acting (i.e. the aromatic carbon bearing the amide substituent). The protein is
colored in rainbow from blue (Nꢀterminus) to red (Cꢀterminus); B) RMS (Root Mean Squared)
fluctuations of Cα atoms for all residues, recorded after the initial heatꢀup and equilibration
phase. The purple and blue lines result from the targeted MD with the lowest S values for the
pathway (a) and (b), respectively. The green line represents the RMS fluctuations of Cα atoms
for all residues calculated on the RORγt structure in absence of 9 bound. The arrows in the plot
highlight those regions in the receptor where strong deviations are observed.
The unbinding pathway (a) is oriented towards the solvent exposed region delimited by the helix
1 – helix 2 loop and helix 4. The ligand takes advantage of the intrinsic flexible zones of the
ACS Paragon Plus Environment
18

Page 19 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
receptor protein to move in and out. The RORγt protein structure in absence of 9 also shows a
natural Cα fluctuation observed during the unconstrained molecular dynamics simulation in the
region of the protein between residue 280 and residue 294 (Figure 5B). During unbinding along
the pathway (a) the SO2 group of 9 remains close to the R367 side chain up to about 450 ps then
the distance of the S of SO2 from the position assumed in the crystal structure increases.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The unbinding along the pathway (b) occurs between the end of helix 7 and 11. The helix 3 ꢀ
helix 4 loop, helix 7 and helix 11 show high Cα fluctuations during the dynamics simulation as
the ligand passes close to them (Figure 5B). When unbinding along pathway (b) the SO2 group
shifts towards the ‘business end’ of the pocket (at about 370 ps time).
Interestingly, the ligand resistance to dissociation calculated via targeted MD simulations (i.e.
the overall ‘action’ S value to extract the ligand21) was found to qualitatively correlate with
experimental koff. Table 3 reports the lowest S values for the unbinding pathways calculated 20
times for both compounds 9 and 11.
The kon in the range of 103 to 104 is slower than the diffusion limiting rate (107/108), most likely
due to the fact that the pocket is rather closed and shielded from the solvent. Therefore a
conformational change of the protein must occur for the ligand to bind to the protein. The same
reasoning applies for the koff. However small changes, such as the removal of a methyl, has a
small effect on the kon but a larger effect on the koff. The kon of 9 is 2.5ꢀtimes slower than the one
for 11, whereas the koff of 9 is 32ꢀtimes slower which translates to a 13ꢀfold improvement of the
binding affinity. The reduced flexibility and the more spherical shape of the benzyloxy portion of
9, through the additional methylꢀgroup, could be responsible for this change in binding kinetics
and is well described by our simulation.
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 47
1
2
3
4
5
6
7
8
9
Our lead molecule 9 was further characterized by ITC in order to determine affinity and the
thermodynamic signature of the binding process and this yielded a Kd of approximately 8 nM. A
more precise measurement (even for reverse titration) was precluded by the low solubility.
Binding is enthalpy driven (ꢁHo = ꢀ17.3 kcal/mol), correlating with the formation of several
intermolecular Hꢀbonds and the entropic contribution to binding is unfavorable (ꢀTꢁSo = 6.3
kcal/mol). As discussed, the protein must undergo large conformational changes for
accommodating the ligand, this large entropy reduction could partially derive from the reduced
flexibility of the protein upon binding of the ligand.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The crystal structure of 9 was used as the basis for an in silico solvent analysis with the 3D
variant of the ‘reference interaction site model’, 3DꢀRISM.23,24 This application computes a timeꢀ
averaged distribution of water density and binding desolvation penalty maps. The in silico
generation of solvent organization in apo and ligand bound crystal structures is a useful approach
to provide insight for structureꢀbased ligand modifications. This is a complementary approach to
the previously described SiteMap binding site characterization. The 3DꢀRISM results for the
crystal structure, with or without 9 bound, are shown in Figure 6.
Figure 6. 3DꢀRISM results for the crystal structure of RORγt in complex with 9 (deposited with
PDB access code = 6G07) and 19 (modeled). (A) Predicted water oxygen atom positions
ACS Paragon Plus Environment
20

Page 21 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
superimposed with the crystal structure of RORγt in complex with 9. The waters observed in the
crystal structure (highlighted by red arrows) are in concordance with the 3DꢀRISM predicted
water oxygen density (shown in solid blue, iso level 4). The waters placed based on the 3Dꢀ
RISM results are shown as yellow spheres. (B) Predicted water oxygen atom positions calculated
for the protein in the absence of compound 9 bound. Compound 9 is shown in black as a
reference, but was not used during the calculation. More ‘stable’ waters, located in polar protein
environments (purple surface) are shown in green; more ‘unstable’ waters, located in lipophilic
protein environments (green surface) are shown in red. (C) Binding model of 19. The 3DꢀRISM
predicted water oxygen density on the complex is shown in solid blue (iso level 4). The predicted
waters mediating the interaction between 19 and the protein are shown as sticks.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6A shows a remarkable agreement between the 3DꢀRISM predicted waters for the
complex with the crystallographic water location at the proteinꢀligand interface. Figure 6B shows
the 3DꢀRISM predicted waters in the absence of the ligand, which is shown in black as a
reference, but was not used in the calculation. The placed waters are color coded according to
their ability to be displaceable (increasing from green to red). Our water analysis mainly focused
on two regions of the binding site: the ‘unexplored back pocket’ offering the possibility of
forming direct or waterꢀmediated interactions with S404 (not discussed) and the ‘polar end’
region showing the possibility of removing the sulfone moiety to form waterꢀmediated
interactions. In order to improve the solubility and permeability of compound 9, we explored the
latter opportunity by designing compound 19 (see Scheme 3) with the aim of forming waterꢀ
mediated interactions between the pyridine nitrogen of the ligand and the side chain of R367 and
the NH of the backbone of L287 (Figure 6C). This modification indeed led to a more soluble and
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 47
1
2
3
4
5
6
7
8
permeable derivative (compound 19: aq. solubility at pH 6.8 = 28 µM; log PAMPA = ꢀ3.7;
compound 9: aq. solubility at pH 6.8 < 4 µM; log PAMPA = ꢀ4.2) albeit with 30ꢀfold lower
potency compared to 9.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Based on its exceptional potency in the FRET assay, compound 9 was selected for testing in a
number of RORγtꢀdependent cellular assays and excellent potency was confirmed (Table 4). In
order to determine the selectivity of 9 against the related transcription factors RORα,β and
against other nuclear hormone receptors, we examined the agonistic and antagonistic effects in
cellular Gal4 reporter gene assays or biochemical TRꢀFRET assays. Compound 9 did not show
any activity against FXR, AR, ER, PR, PXR, GR, PPAR, LXRα/β, RXR (data not shown) and
the related nuclear receptor RORα (Table 4). The compound was only weakly active against
RORβ resulting in partial inhibition in the micromolar range (Table 4), demonstrating that 9 is
highly selective.
Table 4: in vitro pharmacological profile of compound 9
IC50 = 0.6 nM
TRꢀFRET RORγtꢀLBD RIP140 coꢀfactor
peptide recruitment25
IC50 = 14 nM
RORγt transduced HUT78 Tꢀcells: inhibition
of IL17 secretion25
RORα,β,γtꢀLBDꢀGal4 luciferase reporter gene
assays26
RORα
IC50 = > 10000 nM
ACS Paragon Plus Environment
22

Page 23 of 47
Journal of Medicinal Chemistry
1
2
3
4
RORβ
IC50 = 4113 nM; 54% inhibition
5
6
7
IC50 = 17 nM
RORγt
8
9
Human Th17 cell polarization25
Rat Th17 cell polarization25
Human whole blood25
IC50 = 23 nM
IC50 = 110 nM
IC50 = 255 nM
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 9 is highly lipophilic, poorly soluble at neutral pH, and highly permeable across
PAMPA and Cacoꢀ2 monolayers (Table 4). Intrinsic clearance values in hepatic microsomes
ranged between 31 to 48 ꢀL⋅minꢀ1⋅mgꢀ1, the molecule was stable after a 2 hour incubation in
either rat plasma or rat intestinal and hepatic S9 fractions. Compound 9 is > 99% plasma protein
bound in rat and mouse plasma. A pharmacokinetic evaluation of compound 9 was undertaken in
male SpragueꢀDawley rats (1 mg/kg i.v.; 3 mg/kg by oral gavage) yielding an i.v. halfꢀlife of 1.3
hours, blood clearance of 76 ml/min/kg and a volume of distribution of 5.3 l/kg. After oral
administration of a crystalline suspension, the maximum concentration in plasma (Cmax = 36 nM)
was reached after 2.7 hours and oral bioavailability was 10%.
Table 4: in vitro profiling data for 9
LogD(7.4)
> 5.7
Equilibrium solubility (pH 6.8)
0.5 mg/L / 1 ꢀM
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 47
1
2
3
4
– 4.2 cm⋅sꢀ1
LogPAMPA
5
6
7
8
9
Cacoꢀ2 Papp AꢀB / Papp BꢀA (x10ꢀ6 cm/sec)
Liver microsomes (rat/human/mouse) CLint
(ꢀL⋅minꢀ1⋅mgꢀ1)
7.5 ⋅10ꢀ6 / 4.7⋅10ꢀ6
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
48 / 31 / 46
Plasma protein binding (mouse/rat) (%)
>99 / >99
Based on this encouraging profile, 9 was selected for an in vivo proofꢀofꢀconcept study. We
wanted to confirm that this novel binding mode and mechanism of action would indeed lead to
the desired pharmacodynamic activity in a mechanistic model before embarking on further
optimization of physicochemical properties. We decided to test the compound at a rather high
dose of 50 mg/kg with b.i.d. dosing. This dose was selected based on past experience with the
goal to achieve the maximal possible efficacy.
PHARMACOLOGY
Based on our experience with RORγt inverse agonist pharmacology, we decided to profile 9 in a
DTH model in rats using an immunization/challenge protocol. Lewis rats were immunized with
methylated bovine serum albumin (mBSA) as the antigen in the presence of complete Freund’s
adjuvant (CFA). Two weeks later, the DTH response was elicited by a single injection of mBSA
into the right ear and swelling was monitored for 48 hours. Immediately prior to the antigen
ACS Paragon Plus Environment
24

Page 25 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
challenge compound 9 was dosed by oral gavage at 50 mg/kg twice daily throughout the study.
To determine the involvement of antiꢀILꢀ17A in this DTH model, we neutralized this cytokine
by a single injection of an antiꢀILꢀ17A antibody (10 mg/kg) before the mBSA antigen challenge.
Blockade of ILꢀ17A by the antibody inhibited ear swelling by approximately 50% compared to
control animals. Treatment of rats with the RORγt inhibitor 9 resulted in a significant reduction
of the ear swelling score and was comparable to the ILꢀ17A antibody treated group. The
inhibitory effect was already evident after 6 hours and persisted until the end of the study (Figure
7A).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To assess the effect of RORγt inhibition on proꢀinflammatory cytokine levels, inhibition of ILꢀ6
cytokine expression in ear punch homogenates was assessed. Compound 9 treatment led to an
approximately 50% reduction of ILꢀ6 production at the end of the DTH experiment (Figure 7B),
whereas treatment with the antiꢀILꢀ17 antibody led to a slightly higher degree of inhibition of ILꢀ
6 expression. Compound 9 reduced the frequencies of ILꢀ17A producing cells in ex vivo mBSA
recall assays using auricular draining lymph node cells compared to untreated controls (up to 30
%), however statistical significance was not reached most likely due to the high variability in the
control and a ‘nonꢀresponder’ in the treatment group, respectively (Figure 7C, p = 0.0625
unpaired t test). Exposures of 9 at the end of the experiment (2h after last dose) reached 1237 ±
598 nM in blood and 2305 ± 1251 nM in lymph nodes. The rather high interꢀanimal variability
might be due to subꢀoptimal physicochemical properties (e.g., the poor solubility) which might
influence oral absorption.
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 47
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7: A) Compound 9 attenuates mBSAꢀinduced DTH responses in female Lewis rats. Rats
were immunized with mBSA/CFA and two weeks later, the right ear of the animals was
challenged with mBSA in 5 % glucose, while the left ear was injected with vehicle (5 %
glucose). Starting just prior to antigen challenge, 9 was administered twice daily via oral gavage
at 50 mg/kg for 2 days and ear swelling was monitored. As an active comparator, a single bolus
injection of an antiꢀILꢀ17 antibody (10 mg/kg) prior to antigen challenge was used. The mean ±
SEM of the swelling ratios between antigenꢀchallenged and vehicle injected ears are shown (n =
5). B) Inhibition of ILꢀ6 production in ear homogenates after treatment with compound 9 or with
ACS Paragon Plus Environment
26

Page 27 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
an antiꢀILꢀ17 antibody. All groups n = 5. **, P < 0.01 Dunnett´s test. C) Draining lymph node
cells were prepared, stimulated ex vivo with mBSA for 24 hours and frequencies of IL17ꢀ
producing cells determined by ELISpot. Individual data and mean + SEM are depicted (n = 4ꢀ5).
Data is representative from two independent studies.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CONCLUSION
In summary, we described the identification of the RORγt inverse agonist 9 derived from a
weakly binding fragment hit. Subsequent fragment growing and merging efforts led to a highly
potent inverse agonist, efficiently blocking ILꢀ17 production in primary cells and in a human
whole blood assay. Surprisingly, when compound 9 was coꢀcrystallized with RORγt the benzylic
substituent did not sterically clash with helixꢀ12 as hypothesized. Instead, the compound binds
into the ‘unexplored back pocket’ and maintains the ‘TyrꢀHis lock’ (hydrogen bond between
NE2ꢀH479 and OHꢀY502) which typically characterizes the agonist conformation. The hydrogen
bond is present in all four chains observed in the crystal structure ranging from 2.8 to 3.3 Å. The
crystal structure of 9 and the water solvation analysis are consistent with the hypothesis that the
mechanism of RORγt inverse agonism might be due to trapped (‘unstable’, ‘displaceable’) water
molecules in a hydrophobic environment at the ‘business end’ which can be released into bulk
solvent when Y502 in helix 12 moves away from its agonist position (Figure S1). Without 9
bound, these water molecules are free to exchange with bulk solvent via the opening between
Q286 and H323 (i.e. helix 12 and Y502 do not have to move away, see Figure 2 and pathway (a)
Figure 5). Interestingly, we have observed such trapped ‘unstable’ water molecules for different
chemical series of inverse agonists (not necessarily with a moiety in the back pocket), providing
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 47
1
2
3
4
5
6
7
8
9
an indirect destabilization of helix 12.16 For these ‘nonꢀsteric clash inverse agonists’, coꢀ
crystallization with a coꢀactivator peptide and helix 12 in agonist position is still possible.
Different protein conformations in solution with respect to those observed by crystal structure
could support the hypothesized destabilization of helix 12, a potential novel functional
mechanism of RORγt inverse agonism. Another hypothesis to explain the inverse agonism
mechanism of 9 relies on a possible crossꢀtalk of helix 7 (lining the ‘back pocket’) and helix 11
(bearing the H479 residue) which are in direct contact (Figure S2). It might be conceivable to
think that the binding of 9 even if far from the ‘business end’ could impact the dynamics of the
protein in that region of the LBP and thereby its function. Thus the helix 11 flexibility could be
influenced by the binding of the ligand in the ‘unexplored back pocket’ and indirectly lead to
different protein conformations in solution where the H479/Y502 Hꢀbond is either weakened or
not present.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In addition to crystal structure elucidation, binding of compound 9 was further characterized by
its kinetics and thermodynamic features. The enclosure of the pocket is very likely responsible
for the slow on and off kinetics measured for compound 9, suggesting that protein
conformational changes have to occur upon ligand association and dissociation. Minimal
structural modification (removal of one methylꢀgroup) showed substantial effects on the kinetics
which was reproduced by molecular dynamics simulations. ITC measurement showed that the
binding of 9 leads to a rigidified complex characterized by an entropy loss which is
overcompensated by a large enthalpy gain consistent with several hydrogen bond interactions
(e.g. in the ‘polar end’ pocket). Finally, we have shown that compound 9 efficiently attenuated
the inꢀvivo ear swelling response in an mBSA induced DTH model in rat. Likewise, ILꢀ6
expression was reduced in ear homogenates and the frequency of ILꢀ17 producing cells were
ACS Paragon Plus Environment
28

Page 29 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
attenuated in an ex vivo antigen specific recall assay. Overall, the reported data clearly
demonstrate that compound 9 is an attractive RORγt inhibitor with an unusual indirect mode of
inhibition. Further exploration of how different back pocket substituents influence the
physicochemical properties is not completed yet and results will be reported in due course. Since
the environment in the back pocket is more polar in RORγt (S404) compared to RORα and
RORβ (where the corresponding amino acids are glycine and alanine respectively) we expect
that adding polarity in this region will be tolerated. This would lower the high lipophilicity of the
series and should help to identify analogs with improved solubility and increased metabolic
stability.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
EXPERIMENTAL SECTION
General Experimental Information. Chemicals were purchased from commercial sources and
were used without further purification. Reactions were magnetically and mechanically stirred
and monitored by thinꢀlayer chromatography (TLC) on Merck silica gel 60 F254 glass plates
(visualized by UV fluorescence at l = 254 nm) or analytical Waters Acquity UPLC instrument
equipped with PDA detector, Waters Acquity SQD mass spectrometer and Waters Acquity HSS
T3 1.8 mm2.1 V50 mm column. Peak detection is reported at full scan 210–450 nm eluting with
a gradient composed of water/0.05 %formic acid/3.75 mm ammonium formate and
acetonitrile/0.05 %formic acid. Mass spectrometry results are reported as the ratio of mass over
charge. The purity of new compounds was > 95 %, as determined by 1HꢀNMR and LC–MS after
chromatography. Flash column chromatography was performed with Biotage SNAP silica gel
cartridges, eluting with distilled technicalꢀgrade solvents on an Isolera Four apparatus from
ACS Paragon Plus Environment
29