Structure-Activity and Structure-Conformation Relationships of Aryl Propionic Acid Inhibitors of the Kelch-like ECH-Associated Protein 1/Nuclear Factor Erythroid 2-Related Factor 2 (KEAP1/NRF2) Protein-Protein Interaction

Article abstract of DOI:10.1021/acs.jmedchem.9b00279

The KEAP1-NRF2-mediated cytoprotective response plays a key role in cellular homoeostasis. Insufficient NRF2 signaling during chronic oxidative stress may be associated with the pathophysiology of several diseases with an inflammatory component, and pathway activation through direct modulation of the KEAP1-NRF2 protein-protein interaction is being increasingly explored as a potential therapeutic strategy. Nevertheless, the physicochemical nature of the KEAP1-NRF2 interface suggests that achieving high affinity for a cell-penetrant druglike inhibitor might be challenging. We recently reported the discovery of a highly potent tool compound which was used to probe the biology associated with directly disrupting the interaction of NRF2 with the KEAP1 Kelch domain. We now present a detailed account of the medicinal chemistry campaign leading to this molecule, which included exploration and optimization of protein-ligand interactions in three energetic "hot spots" identified by fragment screening. In particular, we also discuss how consideration of ligand conformational stabilization was important to its development and present evidence for preorganization of the lead compound which may contribute to its high affinity and cellular activity.

Full text of DOI:10.1021/acs.jmedchem.9b00279

Subscriber access provided by UNIV AUTONOMA DE COAHUILA UADEC
Article
Structure-Activity and Structure-Conformation Relationships of Aryl Propionic
Acid Inhibitors of the Kelch-like ECH-associated protein 1 : Nuclear factor
erythroid 2–related factor 2 (KEAP1:NRF2) protein-protein interaction
Tom D. Heightman, James Callahan, Elisabetta Chiarparin, Joseph E. Coyle, Charlotte M. Griffiths-Jones,
Ami S. Lakdawala, Rachel McMenamin, Paul Neil Mortenson, David Norton, Torren Peakman, Sharna
J. Rich, Caroline Richardson, William L. Rumsey, Yolanda Sanchez, Gordon Saxty, Henriette Willems,
Lawrence Wolfe, Alison J.-A. Woolford, Zining Wu, Hongxing Yan, Jeffrey K Kerns, and Thomas G. Davies
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00279 • Publication Date (Web): 11 Apr 2019
Downloaded from http://pubs.acs.org on April 12, 2019
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 76
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
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 76
1
2
3
4
5
6
7
8
9
Structure-Activity and Structure-Conformation Relationships of Aryl
Propionic Acid Inhibitors of the Kelch-like ECH-associated protein 1 :
Nuclear factor erythroid 2–related factor 2 (KEAP1:NRF2) protein-protein
interaction
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
Tom D. Heightman1, James F. Callahan2 Elisabetta Chiarparin1, Joseph E. Coyle1, Charlotte
,
Griffiths-Jones1, Ami S. Lakdawala2, Rachel McMenamin1, Paul N. Mortenson1, David
Norton1, Torren M. Peakman1, Sharna J. Rich1, Caroline Richardson1, William L. Rumsey2
,
Yolanda Sanchez2 Gordon Saxty1, Henriëtte M. G. Willems1, Lawrence Wolfe III2 Alison J.-
,
,
A. Woolford1, Zining Wu2, Hongxing Yan2 Jeffrey K. Kerns2* Thomas G. Davies1*
,
,
1Astex Pharmaceuticals, 436 Cambridge Science Park, Cambridge, CB4 0QA, UK
2GlaxoSmithKline, 1250 South Collegeville Road, Collegeville, PA 19426-0989, USA
*Corresponding authors
1
ACS Paragon Plus Environment

Page 3 of 76
Journal of Medicinal Chemistry
1
2
3
4
ABSTRACT
5
6
7
8
9
The KEAP1/NRF2-mediated cytoprotective response plays a key role in cellular homoeostasis.
Insufficient NRF2 signalling during chronic oxidative stress may be associated with the
pathophysiology of several diseases with an inflammatory component, and pathway activation
through direct modulation of the KEAP1-NRF2 protein-protein interaction is being
increasingly explored as a potential therapeutic strategy. Nevertheless, the physicochemical
nature of the KEAP1-NRF2 interface suggests that achieving high affinity for a cell-penetrant
drug-like inhibitor might be challenging. We recently reported the discovery of a highly potent
tool compound which was used to probe the biology associated with directly disrupting the
interaction of NRF2 with the KEAP1 Kelch domain. We now present a detailed account of the
medicinal chemistry campaign leading to this molecule, which included exploration and
optimization of protein-ligand interactions in three energetic “hot-spots” identified by fragment
screening. In particular, we also discuss how consideration of ligand conformational
stabilization was important to its development, and present evidence for preorganization of the
lead compound which may contribute to its high affinity and cellular activity.
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
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 76
1
2
3
4
INTRODUCTION
5
6
7
8
9
The protein KEAP1 (Kelch-like ECH-associated protein 1) plays a key role in the cellular
response to oxidative stress by regulating ubiquitination of the detoxification transcription
factor NRF2 (Nuclear factor erythroid 2–related factor 2)1-4. KEAP1 consists of three domains:
a N-terminal Broad-complex, Tramtrack and Bric-a-Brac (BTB) domain, which serves as an
adaptor for the E3 ubiquitin ligase Cul3/Rbx1; an Intervening Region (IVR) or BACK domain;
and a C-terminal Kelch repeat domain, which is a recognition module for NRF25-8. Under basal
conditions, NRF2 is sequestered and ubiquitinated by KEAP1/Cul3/Rbx1, before being
degraded by the proteasome. Increased levels of oxidative or electrophilic stress result in
covalent modification of key cysteine residues in the BTB and IVR domains, leading to
dissociation of Cul3, and/or loss of productive NRF2 binding9-12. The KEAP1-mediated
ubiquitination of NRF2 is perturbed and increasing levels of free NRF2 lead to activation of
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
cytoprotective genes under its transcriptional control13,14
.
Inadequate NRF2-mediated signalling during chronic oxidative stress is associated with the
pathophysiology of multiple diseases with an inflammatory component. As a result, KEAP1 is
now well recognized as an important target for therapeutic intervention in a range of
indications including respiratory, cardiovascular, renal and neurodegenerative conditions15-21
,
and activation of NRF2 via covalent targeting of reactive cysteines on the BTB and IVR
domains of KEAP1 is a clinically validated approach22-25. Recently however, direct antagonism
of the Kelch-NRF2 protein-protein interaction (PPI) has been increasingly explored as an
alternative point for therapeutic intervention which may offer a complementary profile26-29
.
NRF2 binds two Kelch domains in the context of dimeric KEAP1 through high affinity
(“ETGE”) and low affinity (“DLG”) motifs which interact with a pocket at the centre of a β-
propeller fold (Figure 1a), and loss of one or both of these interactions has been shown to
prevent the productive engagement of NRF2 with KEAP1, with subsequent activation of the
3
ACS Paragon Plus Environment

Page 5 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
anti-oxidant transcriptional program30-34. Nevertheless, the presence of multiple charged
arginine residues in the NRF2-binding site associated with the tight binding of the ETGE motif,
and the large and relatively solvent exposed nature of this PPI binding site suggest that
achieving high affinity for a cell-penetrant drug-like molecule might be challenging29,35
.
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
We recently reported the application of fragment-based drug discovery (FBDD) to the Kelch
domain of KEAP1, and outlined how this approach was instrumental in developing a
nanomolar affinity lead, KI-696 (Figure 1a), which combined tight and selective binding to the
Kelch domain with drug-like physicochemical properties, and which activates the NRF2
pathway in cell-based and animal models36. We now present a detailed account of the
medicinal chemistry campaign leading to this lead molecule and highlight how consideration
of free and bound ligand conformations were important for its development.
Figure 1
4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 76
1
2
3
4
RESULTS
5
6
7
8
Insights from fragment screening and medicinal chemistry strategy
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
As described previously36 fragment hits (exemplified by 1 –3) were observed to cluster within
three main subsites or “hot-spots” in the KEAP1 Kelch-domain NRF2-binding site (Figure 1b),
which we refer to here as the “Acid” ,“Planar Acceptor” and “Sulfonamide” pockets. The
fragment hit 4-chlorophenylpropionic acid 1, binding in the Acid pocket, was selected as a
useful “anchor” fragment, since it provides a variety of synthetically accessible vectors for
exploring the NRF2 binding site as a whole. In particular, it provides suitable vectors for
accessing the other two subsites identified from the fragment screen, and the overall strategy
employed was to grow from fragment 1 by introduction of functionality which could
recapitulate patterns of interactions formed between fragment hits and KEAP1 in these other
hot-spots (Figure 1c)36. Target affinity was assessed using a fluorescence polarization (FP)
assay which examined the ability of compounds to displace a TAMRA-labelled NRF2-derived
peptide; a subset of compounds was also selected for affinity determination by ITC. As potency
was increased, compounds were additionally progressed to an MTT assay in the lung epithelial
cell line BEAS-2B, which measures enzymatic activity of the NRF2-controlled oxidoreductase
NQO1 (see Experimental Procedures).
SAR in the “Planar Acceptor” pocket
Inspection of the KEAP1 co-structure with hit fragment 1 (Figure 2a) suggested the benzylic
position would provide a suitable vector along which to grow towards the “Planar Acceptor”
pocket occupied by fragment 2, and to introduce functionality to recapitulate the stacking and
hydrogen-bonding interactions with Gln 530 and Ser 555 observed in this subsite36. A 3-
5
ACS Paragon Plus Environment

Page 7 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
hydroxyphenyl group was initially introduced at this position to achieve this (compound 4;
Table 1), but did not give a measurable increase in affinity, likely due to the mismatch of the
hydroxyl presenting both hydrogen bond acceptor and donor functionality where a double
acceptor motif would be preferred as observed with fragment 2. In contrast, replacement of this
ring system with the hydroxy pyridine 5 gave the first analogue with measurable potency in
the fluorescence polarization assay, suggesting that this compound binds as the pyridone
tautomer. The co-crystal structure of 5 with the Kelch domain confirmed the anticipated
binding mode, in which the pyridine ring stacks with Tyr 525, and the pyridone carbonyl
accepts hydrogen bonds from Gln 530 and Ser 555 (Figure 2b) as observed for fragment hit 2.
Subsequent methylation of the nitrogen to fix the pyridone tautomer provided compound 6 with
an FP IC50 value of 260 µM.
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
We hypothesised that interactions in the Planar Acceptor pocket might be enhanced by
replacing the pyridone with a fused bicyclic heterocycle, such that adjacent heteroatoms would
accept the two hydrogen-bonds from Gln 530 and Ser 555, as well as increasing -stacking
with Tyr 525. This approach afforded a 4-5-fold enhancement in potency in the methyl
benzotriazole 7, which formed the expected pattern of stacking and polar interactions (Figure
2c)36. The methyl benzotriazole proved the most potent among the isosteric heterocycles tested:
the pyridotriazole 8 and benzisoxazole 9 both showed approximately 5-fold weaker binding
compared with 7, possibly due to differences in hydrogen bond acceptor strength or -stacking
complementarity with Tyr 525. Other heterocycles without the ability to undergo the double
acceptor interaction, such as the indazole 10 or imidazopyridine 11, also showed weaker
binding.
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 76
1
2
3
4
5
6
Table 1 Structure activity relationships for 3-substituted 4-chlorophenylpropionic acids. FP
IC50 displacement of NRF2 peptide from KEAP1 Kelch protein by fluorescence polarization
assay. Kd for direct binding to KEAP1 Kelch protein by ITC.
7
8
9
O
HO
R
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
Cl
Compound
1†
R
FP IC50/µMa ITC Kd/µMb
>1000
-
H
(n=2)
HO
HO
O
>1000
-
4
5
6
(n=1)
40% @ 300
-
(n=2)
N
260
(n=2)
-
N
N
61
(n=4)
N
7†
59
N
N
N
N
N
170
(n=2)
N
N
(R)-7
-
47
(n=3)
(S)-7
-
N
O
330
(n=1)
8
9
-
N
N
N
66% @ 300
-
(n=2)
H
N
57% @ 300
10
11
-
N
(n=1)
N
47% @ 300
-
(n=1)
N
aGeometric mean of independent determinations. Single determination. Note that ITC data were measured for
b
the racemate. †Data for compounds 1 and 7 is taken from Davies et al36.
7
ACS Paragon Plus Environment

Page 9 of 76
Journal of Medicinal Chemistry
1
2
3
4
Figure 2
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
SAR in the “Sulfonamide” pocket
Having established the methyl benzotriazole as an efficient binder in the Planar Acceptor
pocket, we next sought to exploit the “Sulfonamide” binding pocket identified by fragment 3
(Figure 1c), through growth ortho to the chloro substituent36. Extension of 7 in this manner
with either a methyl (12) or hydroxymethyl (13) group did not afford additional potency (Table
2); however, the co-structure of 12 confirmed the desired orientation for the growth vector,
with the methyl pointing towards the Sulfonamide pocket rather than into the tunnel region at
the centre of the Kelch domain (Figure 2d). This suggested that attachment of a sulfonamide
via nitrogen to a methylene carbon should allow the sulfonamide oxygens to occupy similar
positions to those in the structure of fragment 3. Indeed, the N-methyl methanesulfonamide 14
achieved a potency enhancement of approximately 18-fold compared with 7, and as described
previously36, the co-crystal structure of 14 confirmed the intended binding mode, in which one
sulfonamide oxygen accepts a hydrogen bond from the side chain of Ser 602 (Figure 2e). This
structure also highlighted an opportunity for further potency gain, as the sulfonyl methyl group
8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 76
1
2
3
4
5
6
7
8
9
is oriented towards the aromatic ring of Tyr 334 and the lipophilic edge of the binding site.
Replacement of this methyl group with a phenyl (15) gave a further 13-fold affinity
enhancement, taking the IC50 value to 270 nM36. As previously reported, the co-structure of
the benzenesulfonamide analogue 15 confirmed a similar binding mode to that of 14, while its
additional phenyl ring is well aligned to engage in -stacking and lipophilic interactions with
Tyr 334 (Figure 2f)36. SAR to explore alternative aromatic substituents in this region was
relatively flat, with replacement by electron poor (pyridine 16) and electron rich
(dimethylpyrazole 17) heterocycles having only small effects on affinity. Replacement with a
saturated ring system (cyclohexyl 18) reduced affinity, potentially indicating the importance of
-stacking and shape complementarity in addition to lipophilicity for optimal interaction in this
area of the site.
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
Only the (S)-enantiomers of compounds prepared as racemates were observed to bind by X-
ray crystallography, but separation of the enantiomers of 7 showed only a modest eudysmic
ratio (Table 1), and a co-structure of the single (R)-enantiomer of 7 with KEAP1 could also be
obtained, indicating that the methylene linker possesses enough flexibility to allow key
pharmacophore points to be reached without a major energetic penalty. Interestingly, X-ray
crystallographic analysis of Kelch soaked with (R)-7 also revealed clear electron density for a
second ligand binding event adjacent to the canonical pose, interacting with the lipophillic edge
of the NRF2 site and co-binding with a molecule of DMSO from the soaking solution (Figure
3). For our previously published KEAP1 co-structures with bis-aryl compounds, we had
observed residual electron density at this position which we speculated to arise due to the
binding of a second ligand molecule at the high concentrations employed during the soaking
experiments36. In these cases however, the low occupancy and high levels of disorder (as
evident from weak and broken electron density) precluded the fitting of these additional sites
with confidence. As discussed later, the chloro-phenyl moiety for the second copy of (R)-7,
9
ACS Paragon Plus Environment

Page 11 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
and the co-bound molecule of DMSO may highlight additional, albeit weaker hot-spots for
favourable interaction which were not evident from the main fragment screen.
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
Inspection of the co-structure of 15 suggested that the ligand might be bound in an energetically
unfavourable conformation, due to the relatively close approach of the chloro substituent and
one of the sulfonamide oxygens (rCl…O = 3.7 Å) (Figure 4a). We hypothesised that replacement
of the chloro-substituent with a methyl group might relieve this unfavourable intramolecular
contact, as this change would replace the negative electrostatic potential at the edge of the
chloro with a positive electrostatic potential arising from the C-H bond polarization in the
methyl group (Figure 4b). A 3-dimensional conformational search on a reduced model system
(Figure 4c), suggested that with the chloro substituent, the bound conformation is
approximately 2 kcal mol-1 higher in energy than the gas-phase global minimum. By contrast,
the global energy minimum for the methyl analogue is the same as (or very close to) the bound
conformation (see Supporting Information). Although the full potential energetic benefit of this
change was not realized, presumably due to additional competing factors, this optimization
nevertheless led to a small increase in target affinity (compound 19). In addition, this
10
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 76
1
2
3
4
5
6
7
8
9
compound also gave the first measurable activity in a cell-based assay, with stimulation of
NQO1 activity in the human lung epithelial cell line, BEAS-2B (Table 2). A methoxy group
was subsequently introduced at the benzotriazole 7-position with the aim of increasing the
hydrogen bonding potential of the ring nitrogens through electron donation, and to add
additional surface contacts36, and this afforded an additional small increase in target affinity
and cellular potency (compound 20).
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
11
ACS Paragon Plus Environment

Page 13 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
Table 2 Structure activity relationships for substituted phenyl methylbenzotriazole propionic acids. FP
IC50 displacement of NRF2 peptide from KEAP1 Kelch protein by fluorescence polarization assay; Kd
for direct binding to KEAP1 Kelch protein by ITC; BEAS-2B EC50 in NQO1 MTT assay.
O
9
HO
R3
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
R1
N
N
N
R2
FP
ITC
BEAS-2B
Compound R1
R2
H
R3
H
IC50/µMa Kd/µMb EC50/µMc
44
(n=2)
12
13
Cl
Cl
Cl
-
-
-
-
-
140
(n=2)
OH
H
3.4
(n=3)
14†
NMeSO2Me
H
H
4.0
0.27
(n=3)
15†
16
Cl
Cl
NMeSO2Ph
0.67
-
-
-
O O
S
0.58
(n=2)
H
H
H
N
N
O O
S
0.32
(n=1)
17
18
Cl
Cl
-
-
-
-
N
N
N
O O
S
2.4
(n=16)
N
0.13
(n=3)
3.4
(n=8)
19
20
Me
Me
NMeSO2Ph
H
0.11
0.069
(n=10)
0.85
(n=13)
NMeSO2Ph
OMe
0.044
O O
S
0.047
(n=12)
1.6
(n=14)
21
22
Me
Me
OMe
OMe
0.035
N
O
O O
S
0.037
(n=12)
0.57
(n=14)
N
-
N
12
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 76
1
2
3
4
5
6
7
8
9
O O
S
0.26
(n=10)
1.4
(n=14)
23
24
Me
Me
OMe
OMe
-
-
N
O
O O
S
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
0.020
(n=10)
0.030
(n=14)
N
O
O O
S
0.18
(n=6)
2.9
(n=6)
N
25
H
OMe
-
O
aGeometric mean of independent determinations. Single determinations. Note that ITC data for all compounds
were measured for racemates. cGeometric mean of independent determinations. †Data for compounds 14 and 15
is taken from Davies et al36.
b
Figure 4
As the sulfonamide 20 appeared to interact with most of the polar and -stacking features in
the NRF2-binding pocket (Figure 5a), we considered approaches other than optimizing non-
covalent interactions to further enhance the affinity and cellular potency. Ligand
conformational rigidification can be a favourable strategy to achieve enhanced binding affinity
and this approach has been used recently for other PPI inhibitors37. The co-structure of KEAP1-
13
ACS Paragon Plus Environment

Page 15 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
20 suggested geometric and steric suitability for cyclization of the phenyl sulfonamide to form
the fused 7-membered benzoxathiazepine 21, potentially pre-organizing this lower part of the
molecule in the bound conformation to give increased affinity. The co-structure of 21 with
Kelch confirmed the expected binding mode (Figure 5b), with only minor perturbation to the
canonical interactions; however, cyclization led to a void between the oxathiazepine and
benzotriazole in the bound conformation of 21 (Figure 5c), which may be energetically
unfavourable. To address this, we therefore explored the introduction of small lipophilic
substituents (eg methyl) to both oxathiazepine and thiadiazepine rings, which modelling
suggested would partially fill this unoccupied space (compounds 22-24).
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
Substitution at the position alpha to the oxathiazepine oxygen was predicted to be optimal for
space-filling, and addition of methyl at this location gave the largest increase in affinity (24).
Subsequent KEAP1 co-soaks with 24, which contained a mixture of four diastereomers,
revealed preferential binding of the (R,S) isomer. A subsequent study of the separate
14
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 76
1
2
3
4
5
6
7
8
9
stereoisomers of 24 (Table 3) confirmed (R,S) as the highest affinity configuration (29 (KI-
696); Kd 1.3 nM; BEAS-2B EC50 12 nM)36, consistent with minimal perturbation from the
binding mode of 21, and occupation of the space between oxathiazepine and benzotriazole by
the “pendant” pseudo-equatorial methyl (Figure 5d-e). Interestingly this methyl also occupies
the pocket region where the chloro-phenyl binds for the second copy of (R)-7, highlighting the
energetic benefits of space-filling and lipophilic interactions in this part of the binding site
(Figure 5f).
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 Structure-activity relationships for stereoisomers of 24. FP IC50 displacement of NRF2 peptide
from KEAP1 Kelch protein by fluorescence polarization assay; Kd for direct binding to KEAP1 Kelch
protein by ITC; BEAS-2B EC50 in NQO1 MTT assay.
O
HO
O
36
N
N
N
N
S
35
O
O
O
Configuration
FP
ITC
BEAS-2B
EC50/µMc
Compound
IC50/µMa
Kd/µMb
@35
@36
0.010
(n=12)
0.042
(n=10)
0.024
0.011
(n=12)
0.76
(n=12)
0.083
(n=20)
26
27
R
R/S
-
S
R
R
R/S
R
-
-
28
(n=18)
95%@15nM
(n=3)
29†
S
0.0013
0.012
b
c
aGeometric mean of independent determinations. Single determination. Geometric mean of independent
determinations. †Data for compound 29 is taken from Davies et al36.
15
ACS Paragon Plus Environment

Page 17 of 76
Journal of Medicinal Chemistry
1
2
3
4
NMR solution conformational studies for free ligands
5
6
7
8
9
In addition to relieving the possible energetic penalty associated with the void observed for the
KEAP1-bound 21, we hypothesised that introduction of the pendant C35 methyl (Figure 6a)
might favour a preorganized conformation of the free ligand through a “hydrophobic collapse”
between the benzotriazole heterocycle, the central phenyl and the oxathiazepine as observed in
the bound state; it could also lead to local conformational restriction of the seven-membered
ring due to the preference to maintain the methyl group in an pseudo-equatorial position. Both
modes of preorganization might be expected to lead to an increase in affinity over more
conformationally flexible analogues on entropic grounds. In addition, we speculated that the
large increase in cellular potency observed for 29 compared to 21 might in part arise due to
increased cell permeability which could result from reduced molecular flexibility38.
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
1
1
To examine the validity of this hypothesis experimentally, H 1D and 2D H-1H ROESY
spectra were acquired in aqueous solution for 20, 21 and 29. An initial observation from the 1H
1D spectra was that the diastereotopic protons on the phenyl-sulfonamide methylene linker (19′
and 19″, Figure 6) gave widely separated chemical shifts for 29, with a Δδ of 0.59 ppm,
potentially indicating restricted rotation around C19-C29 and/or C19-N37 bonds. This
contrasts with the singlet resonances observed for the methylene protons for des-Me 21, and
uncyclized 20 (Figure 6b), typical of averaging of the electronic environment due to ligand
flexibility.
16
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 76
1
2
3
4
Figure 6
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
To preclude the possibility that the large Δδ between 19′ and 19″ for 29 might be due to
1
chemical inequivalence, a H 1D spectrum was also acquired for the simpler analogue 25,
which retains the oxathiazepine methyl stereocentre, but was anticipated to be more
conformationally flexible due to the loss of the tolyl methyl. For this compound, a smaller Δδ
(0.19 and 0.21 ppm for the diastereoisomers) was observed between the methylene protons
(Figure 6b), suggesting that the larger Δδ observed between 19′ and 19″ in 29 is not solely a
result of their diastereotopic character. Further evidence for a preferred conformation for the
methylene linker in 29 was supported by the pattern of ROEs to/from 19′ and 19″, which are
consistent with the placement of 19′ and 19″ in different environments as observed in the bound
pose with KEAP1 (Figure 6c and Supporting Information Figure 1a). The pattern of ROEs
between the benzoxathiazepine ring and the central phenyl ring (specifically with proton 15) is
also consistent with 29 adopting a conformation in which the relative orientation of these parts
17
ACS Paragon Plus Environment

Page 19 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
of the molecule is close to that observed when bound to KEAP1 (Figure 6c and Supporting
Information Table 1).
7
8
9
For the des-Me analogues (21 and 25), a weakening of the analogous ROEs (Supporting
Information Table 1) suggests a greater conformational flexibility for this part of the molecule,
with a greater range of conformations adopted in solution compared to 29. Finally, for 29,
distinct axial (3.66, dd, J=15.0, 10.4 Hz) and equatorial (2.87, d, J=15.0 Hz) signals were
observed for 20-H2 in the seven-membered ring, consistent with a single pseudo-chair
conformation with the pendant methyl pseudo-equatorial as observed for the bound structure.
In contrast, for the des-Me 21, averaged signals were observed for the protons on the seven-
membered ring, suggesting a high degree of conformation mobility and rapid exchange
between axial and equatorial positions (Figure 6b).
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
For 29, weak ROEs between the pendant methyl and the benzotriazole methyl and the methoxy
were also observed, consistent with a “collapsed” conformation similar to that observed bound
to KEAP1 (Figure 6c and Supporting Information Figure 1b). Further evidence for this comes
from a small shielding (0.1 ppm) of the pendant methyl (3-H3) in 29 compared to 25 (Figure
6b), in agreement with a population of a collapsed conformation with the pendant methyl over
the face of the benzotriazole. In addition, shielding (0.34 ppm) of 29 20-Heq compared to 25 is
observed (Figure 6b), in agreement with this hydrogen lying over the central phenyl ring as
observed in the conformation bound to KEAP1.
Taken together, these observations are consistent with the hypothesis that introduction of the
pendant methyl leads to increased preorganization for 29, potentially through a combination of
local conformation restriction, and hydrophobic collapse. In contrast, both the des-Me analogue
21, and the uncyclized 20 exhibit features suggesting a greater degree of mobility and a lack of
preferred conformation, consistent with their lower affinity. The comparison with 25, which is
18
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 76
1
2
3
4
5
6
7
8
9
also considerably less potent in both FP and BEAS-2B assays compared to 29 (Tables 2 and
3), suggests that the presence of the tolyl methyl group on the central phenyl ring is also a key
structural element in reducing the number of conformations around the methylene group, and
in increasing the population of the bioactive conformation.
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
Synthesis
The 3-hydroxyphenyl analogue 4 was synthesized according to Scheme 1. Addition of the
Grignard reagent 30 to 4-chlorobenzonitrile followed by hydrolytic workup afforded the
benzophenone 31. Next, a Wittig reaction using trimethyl phosphonoacetate followed by nickel
borohydride reduction of the resulting unsaturated ester gave the bisarylpropionate ester 32,
which was demethylated using boron tribromide to give the acid 4.
The preparation of hydroxypyridine 5 and methylpyridone 6 mirrored the procedure for 4, as
shown in Scheme 2. Addition of 4-chlorophenylmagnesium bromide to the cyanomethoxy
pyridine 33 followed by a Wittig reaction and nickel borohydride reduction gave the
methoxypyridyl ester 34. This was either doubly demethylated using 4M HCl to give 5, or
rearranged to the N-methylpyridone with methyl iodide followed by basic ester hydrolysis to
give 6.
Synthesis of the N-methyl benzotriazole 7 in racemic form was described previously36. The
enantiomers were separated using chiral HPLC to provide (R)-7 and (S)-7.
The triazolopyridine analogue 8 was prepared according to Scheme 3. The triazolopyridyl
carboxylic acid 35 was converted to the Weinreb amide 36 by HBTU coupling, and treatment
with 4-chlorophenyl magnesium bromide gave the ketone 37. A similar sequence to Schemes
1-2 involving a Wittig reaction, conjugate reduction and ester hydrolysis gave 8.
19
ACS Paragon Plus Environment

Page 21 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The benzisoxazole 9 was prepared by a more concise route outlined in Scheme 4. Oxidative
cyclisation of ketone 39 with ammonia followed by NCS gave the bromo benzisoxazole 4039
which was converted to the boronate ester 41 under standard conditions. Next, a rhodium
catalysed conjugate addition40 to methyl 4-chlorocinnamate followed by saponification gave
the bisaryl propionic acid analogue 9 directly. The indazole 10 and imidazopyridine 11 were
prepared similarly from the corresponding boronate esters.
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 4-chloro-3-methyl analogue 12 was prepared according to Scheme 5. Here, the
benzotriazolyl aldehyde 42 was converted to the cinnamate 43 by a Wittig reaction, upon which
rhodium catalysed conjugate addition of the pinacol boronate ester, followed by saponification
gave 12.
The synthesis of compounds 13-15 was described previously36. The pyridyl sulfonamide 16
was prepared by conversion of the sulfonyl chloride 44 to the methylsulfonamide 45, which
was then subjected to a Mitsunobu alkylation with the hydroxymethyl compound 4636,
followed by saponification to give 16 (Scheme 6). Sulfonamides 17 and 18 were prepared by
a similar sequence starting from their corresponding sulfonyl chlorides.
Compound 19 containing a methyl group in place of the Cl in the preceding compounds was
prepared by a complementary route outlined in Scheme 7. First the cinnamate 4336,was coupled
with 3-hydroxymethyl-4-methyl phenyl boronic acid using rhodium catalysis to give
intermediate 51, and then a Mitsunobu coupling with N-methyl benzenesulfonamide followed
by saponification gave 19. The methoxy-substituted triazole analogue 20 was prepared via an
analogous sequence starting from the corresponding methoxybenzotriazolyl acrylate ester36.
Syntheses of the fused 7-membered ring analogues 21-24 are outlined in Scheme 8. Compound
21 was prepared by a Mitsunobu coupling of commercial benzoxathiazepine 52 with the
hydroxymethyl intermediate 5336, followed by saponification of the ester. Compound 22 was
prepared similarly: reaction of 2-fluorobenzenesulfonamide with 2-(methylamino)ethanol
20
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 76
1
2
3
4
5
6
7
8
9
followed by cyclization under Mitsunobu conditions gave the benzthiadiazepine 56, which was
again coupled with 53 under Mitsunobu conditions followed by saponification. The isomeric
methylated benzoxathiazepine analogues 23 and 24 were prepared by reaction of 2-
fluorobenzenesulfonyl chloride (58) with 2-amino-1-propanol or 1-amino-2-propanol
respectively, followed by cyclization under microwave conditions to give 60 and 63
respectively, which were then coupled with 53 and saponified as above.
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 desmethyl analogue 25 was prepared according to Scheme 9, by coupling the
methoxybenzotriazolyl acrylate ester under rhodium catalysis to 3-hydroxymethylphenyl
boronic acid, followed by Mitsunobu coupling with the methylated benzoxathiazepine 63 and
subsequent saponification.
The stereoisomers 26-29 were prepared by analogy to compound 24, following the route
outlined in Scheme 8 and using the appropriate enantiomer of 1-amino-2-propanol to set the
configuration at C35, followed by chiral SFC separation to isolate single diastereoisomers at
C36.
21
ACS Paragon Plus Environment

Page 23 of 76
Journal of Medicinal Chemistry
1
2
3
4
Scheme 1
5
6
O
7
O
OMe
8
9
THPO
MgBr
HO
O
(b,c)
(a)
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
Cl
Cl
32
30
31
(d)
O
OH
HO
Cl
4
a) 4-Cl-PhCN, THF, 75 °C, then HCl (aq), 45%. b) Trimethyl phosphonoacetate, KOtBu,
THF, 0 °C, 69%. c) NiCl2.6H2O, NaBH4, MeOH, 0 °C, 88%. d) BBr3, DCM, -42 °C to RT,
34%.
Scheme 2
O
O
OH
OMe
HO
O
CN
O
(a-c)
(d)
N
N
N
Cl
Cl
33
34
5
(e)
O
OH
O
N
Cl
6
a) 4-Cl-PhMgBr, THF, 75 °C. b) Trimethyl phosphonoacetate, KOtBu, THF, 0 °C. c)
NiCl2.6H2O, NaBH4, MeOH, 0 °C, 69% over 3 steps. d) 4M HCl, 1,4-dioxane, 60 °C, 34%.
e) Neat MeI, 100 °C, workup then 2M NaOH, MeOH, 50%.
22
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 76
1
2
3
4
Scheme 3
5
6
7
8
O
O
O
OH
N
(c,d)
(a,b)
N
N
OH
9
N
N
N
N
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
N
N
Cl
Cl
37
8
35
a) MeONHMe, HBTU, N-methylmorpholine, DCM, DMF, 47%. b) 4-Cl-PhMgBr, THF, 0
°C, 64%. c) Trimethyl phosphonoacetate, KOtBu, THF, 0 °C, 44%. d) NiCl2.6H2O, NaBH4,
MeOH, 0 °C, then NaOH, MeOH, 9% over two steps.
Scheme 4
O
HO
Br
Br
B
O
(b)
O
(a)
O
N
N
O
41
40
39
O
O
OH
OMe
Cl
(c, d)
Cl
O
N
9
a) NH3, MeOH, then NCS, K2CO3, THF, 21%. b) (Bpin)2, KOAc, 1,4-dioxane, PdCl2dppf, 90
°C, 76%. c) [RhCl(cod)]2, 1,4-dioxane/water, Et3N, 95 °C, 67%. d) 1M LiOH, MeOH, 7%.
23
ACS Paragon Plus Environment

Page 25 of 76
Journal of Medicinal Chemistry
1
2
3
4
Scheme 5
5
6
7
8
O
CO2Me
O
HO
N
(b, c)
(a)
H
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
N
N
N
N
Cl
N
N
N
N
12
43
42
a) Trimethyl phosphonoacetate, KOtBu, THF, 0 °C, quant. b) [RhCl(cod)]2, 1,4-
dioxane/water, Et3N, 95 °C. c) 1M LiOH, MeOH, 41% over 2 steps.
Scheme 6
O
O
HO
HO
Cl
N
N
N
HO
46
Cl
N
O O
S
O O
S
(b, c)
N
N
(a)
N
N
Cl
N
N
O
H
S
O
N
16
45
44
a) MeNH2, DCM/THF, 98%. b) ADDP, tBu3P, THF. c) 1M LiOH, MeOH,13% over 2 steps.
Scheme 7
O
(HO)2B
O
CO2Me
HO
N
HO
MeO
(b)
(a)
N
N
N
N
N
N
N
N
N
S
HO
O
19
51
O
43
a) [RhCl(cod)]2, 1,4-dioxane/water, Et3N, 95 °C, 45%. b) PhSO2NHMe, ADDP, nBu3P, THF,
then 3M NaOH (aq), 30%.
24
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 76
1
2
3
4
Scheme 8
5
6
O
NH
S
O
O
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
52
O
F
N
N
N
OEt
SO2NH2
(a, b)
(c, d)
(c, d)
N
NH
O
O
O
S
HO
N
N
O
53
N
OH
O
54
(e,f)
56
R2
R1
X
O
N
S
NH
O
O
S
O
F
O
SO2Cl
21
22
23
24
X=O, R1=R2=H
60
X=NMe, R1=R2=H
X=O, R1=Me, R2=H
X=O, R1=H, R2=Me
O
58
NH
S
O
O
63
a) MeNHCH2CH2OH, 130 °C microwave, 87%. b) DIAD, PS-PPh3, THF, 73%. (c) 2-Amino-
1-propanol or 1-amino-2-propanol, K2CO3, THF, quant. (d) KOtBu, DMSO, microwave
cyclization, quant. (60); 88% (63). e) ADDP, Bu3P or DIAD, PS-PPh3, THF. f) NaOH (aq.
2M), MeOH.
25
ACS Paragon Plus Environment

Page 27 of 76
Journal of Medicinal Chemistry
1
2
3
4
Scheme 9
5
6
O
7
8
9
NH
O
O
S
O
(HO)2B
O
CO2Et
N
N
63
N
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
HO
OEt
OH
MeO
N
(b)
(a)
MeO
N
O
O
N
N
N
N
N
S
HO
O
25
66
65
O
a) [RhCl(cod)]2, 1,4-dioxane/water, Et3N, 150 °C, 55 %. b) DIAD, PS-PPh3, THF, then 1M
NaOH (aq.), MeOH, 100 °C, 38 %.
26
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 76
1
2
3
4
DISCUSSION AND CONCLUSIONS
5
6
7
8
9
We have described here a detailed account of the medicinal chemistry campaign which led to
the identification of KI-696 (29), a highly potent inhibitor of the KEAP1 Kelch-NRF2
interaction, and activator of the NRF2 pathway in cell-based assays, primary cells and in
vivo21,36. As outlined previously, fragment screening identified three distinct “hot-spots” within
the NRF2 binding pocket of KEAP1, providing information which was used to progress a weak
fragment hit (Kd >1 mM) to molecules with nanomolar affinity for KEAP136. Of particular
importance was evidence that non-charged moieties can bind close to the region occupied by
the NRF2 Glu 82 side-chain, and this insight assisted in the development of a highly potent
lead containing only a single acidic functionality, allowing good cell permeability to be
achieved36.
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 structure-driven elaboration from millimolar fragment to nanomolar lead described in this
study highlights two important approaches to structure-guided affinity improvement. The first
is exploiting the insight into favourable intermolecular interactions afforded by fragment
screening. The prediction of protein-ligand energetics ab initio remains challenging and
identifying and targeting specific polar interactions to provide gains in affinity can be
particularly difficult due to competition with solvent binding. Recapitulating patterns of
protein-ligand interactions known empirically to improve affinity is a useful strategy for
addressing this limitation; for example, through exploitation of SAR and binding modes from
previous hit matter, or through virtual screening driven by interaction fingerprints41-44
.
Fragment screening is in theory well suited to give such information, but in practice, it is
unusual to observe a collection of hits that provide an energetic mapping of the whole binding
site of interest. Exploitation of weaker interactions throughout a target’s binding pocket
(“warm-spots”) can also be an important means to build potency45, but such sites may not
always be readily detectable using standard fragment screening protocols. In this study
27
ACS Paragon Plus Environment

Page 29 of 76
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
however, three discrete subsites were identified through X-ray fragment screening, providing
direct insight into multiple molecular recognition points, and allowing a rapid and efficient
improvement in affinity. Further insight into favourable interaction sites was also provided by
observation of an additional lower occupancy ligand binding event at high soak concentrations,
as well as the co-binding of DMSO, which could be considered to represent an additional,
“ultra-low” molecular weight fragment.
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
A second important route to increasing affinity is through designs which seek to stabilize
energetically unfavourable aspects of the bound ligand pose, and/or to pre-organize the ligand
in the bound conformation. This approach is particularly beneficial in the context of inhibiting
protein-protein interactions, because it may allow affinity gains to be achieved without addition
of molecular weight and/or polar surface area. This has been exemplified recently by a series
of IAP PPI inhibitors in which a folded conformation was stabilized in solution by an apparent
attractive interaction of complementary partial charges between a nitrogen lone pair and a
carbonyl carbon37.
In this study, we believe the driving force for the folded conformation of 29 arises from the
hydrophobic effect, rather than explicitly attractive contacts (such as van der Waals dispersion
or CH- interactions) for two reasons. Firstly, systematic studies using Wilcox balance systems
have shown that van der Waals forces between hydrocarbon motifs are weak, such that solvent-
solvent interactions have a dominant role in driving the association of apolar groups46.
Secondly, the distance between the methyl and centroid of the triazole ring system of Kelch-
bound 29 is rather long at approximately 4 Å, compared to CSD mean distances for similar
methyl CH- interactions (3.5 Å),47 although we note that this distance may be influenced by
the interaction of 29 with the protein, with a slightly closer approach of methyl and
benzotriazole possible in the unbound state. Interestingly, a potentially analogous methyl-
28
ACS Paragon Plus Environment