Discovery of AZD9833, a Potent and Orally Bioavailable Selective Estrogen Receptor Degrader and Antagonist

Article abstract of DOI:10.1021/acs.jmedchem.0c01163

Herein we report the optimization of a series of tricyclic indazoles as selective estrogen receptor degraders (SERD) and antagonists for the treatment of ER⁺breast cancer. Structure based design together with systematic investigation of each region of the molecular architecture led to the identification of N-[1-(3-fluoropropyl)azetidin-3-yl]-6-[(6S,8R)-8-methyl-7-(2,2,2-trifluoroethyl)-6,7,8,9-tetrahydro-3H-pyrazolo[4,3-f]isoquinolin-6-yl]pyridin-3-amine (28). This compound was demonstrated to be a highly potent SERD that showed a pharmacological profile comparable to fulvestrant in its ability to degrade ERα in both MCF-7 and CAMA-1 cell lines. A stringent control of lipophilicity ensured that 28 had favorable physicochemical and preclinical pharmacokinetic properties for oral administration. This, combined with demonstration of potent in vivo activity in mouse xenograft models, resulted in progression of this compound, also known as AZD9833, into clinical trials.

Full text of DOI:10.1021/acs.jmedchem.0c01163

Subscriber access provided by CORNELL UNIVERSITY LIBRARY
Drug Annotation
Discovery of AZD9833, a potent and orally bioavailable
selective estrogen receptor degrader and antagonist.
James S. Scott, Thomas A. Moss, Amber Balazs, Bernard Barlaam, Jason Breed, Rodrigo Carbajo,
Elisabetta Chiarparin, Paul Davey, Oona Delpuech, Stephen Fawell, David Fisher, sladjana
gagrica, Eric Gangl, Tyler Grebe, Ryan David Robert Greenwood, Sudhir Hande, Holia Hatoum-
Mokdad, Kara Herlihy, Samantha Hughes, Thomas A. Hunt, Hoan Huynh, Sophie Janbon, Tony
Johnson, Stefan Kavanagh, Teresa C. M. Klinowska, Mandy Lawson, Andrew Lister, Stacey Marden,
Dermot McGinnity, Christopher Morrow, J. Willem M. Nissink, Daniel H. O Donovan, Bo Peng,
Radek Polanski, Darren Stead, Stephen Stokes, Kumar Thakur, Scott Throner, Michael Tucker,
Jeffrey G. Varnes, Haixia Wang, David Wilson, Dedong Wu, Ye Wu, Bin Yang, and Wenzhan Yang
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c01163 • Publication Date (Web): 10 Sep 2020
Downloaded from pubs.acs.org on September 10, 2020
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 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
Wilson, David; AstraZeneca Mereside
Wu, Dedong ; AstraZeneca Pharmaceuticals
Wu, Ye; AstraZeneca Pharmaceuticals LP
Yang, Bin; AstraZeneca Pharmaceuticals LP,
Yang, Wenzhan; AstraZeneca Pharmaceuticals LP,
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 92
1
2
3
4
5
6
7
8
9
Discovery of AZD9833, a potent and orally bioavailable
selective estrogen receptor degrader and antagonist.
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
.
James S. Scott,* Thomas A. Moss,* Amber Balazs, Bernard Barlaam, Jason Breed, Rodrigo J.
Carbajo, Elisabetta Chiarparin, Paul R. J. Davey, Oona Delpuech, Stephen Fawell, David I. Fisher,.
Sladgana Gagrica, Eric T. Gangl, Tyler Grebe, Ryan D. Greenwood, Sudhir Hande, Holia Hatoum-
Mokdad, Kara Herlihy,. Samantha Hughes, Thomas A. Hunt, Hoan Huynh, Sophie L. M. Janbon,&
Tony Johnson, Stefan Kavanagh, Teresa Klinowska, Mandy Lawson, Andrew S. Lister, Stacey
Marden,$ Dermot F. McGinnity, Christopher J. Morrow, J. Willem M. Nissink, Daniel H.
O’Donovan, Bo Peng, Radoslaw Polanski,. Darren S. Stead, Stephen Stokes, Kumar Thakur, Scott
R. Throner, Michael J. Tucker, Jeffrey Varnes, Haixia Wang, David M. Wilson, Dedong Wu,$ Ye
Wu, Bin Yang, Wenzhan Yang.$
 Oncology R&D, AstraZeneca, Cambridge CB4 0WG, United Kingdom.

Oncology R&D, AstraZeneca, R&D Boston, 35 Gatehouse Drive, Waltham, Massachusetts 02451,
United States.
. Discovery Sciences R&D, AstraZeneca, Cambridge CB4 0WG, United Kingdom.
$
Advanced Drug Delivery, Pharmaceutical Sciences, R&D, Boston, MA, United States
&
Early Chemical Development, Pharmaceutical Sciences, R&D, Macclesfield, United Kingdom
 Oncology Safety, Clinical Pharmacology and Safety Sciences, R&D, AstraZeneca, Cambridge, United
Kingdom
1
ACS Paragon Plus Environment

Page 3 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
ABSTRACT
7
8
9
Herein we report the optimization of a series of tricyclic indazoles as selective estrogen receptor
degraders (SERD) and antagonists for the treatment of ER+ breast cancer. Structure based
design together with systematic investigation of each region of the molecular architecture led to
the identification of N-[1-(3-fluoropropyl)azetidin-3-yl]-6-[(6S,8R)-8-methyl-7-(2,2,2-trifluoroethyl)-
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
6,7,8,9-tetrahydro-3H-pyrazolo[4,3-f]isoquinolin-6-yl]pyridin-3-amine (28). This compound was
demonstrated to be a highly potent SERD which showed a comparable pharmacological profile to
fulvestrant in its ability to degrade ER in both MCF-7 and CAMA-1 cell lines. A stringent control of
lipophilicity ensured that 28 had favourable physicochemical and preclinical pharmacokinetic properties
for oral administration. This, combined with demonstration of potent in vivo activity in mouse xenograft
models, resulted in progression of this compound, also known as AZD9833, into clinical trials.
INTRODUCTION
Breast cancer is the most prevalent cancer in women, accounting for 24% of all cancer incidence and 15%
of all cancer related deaths in women, according to the Global Cancer Statistics 2018 analysis.1 Of these,
~70% of patients express the hormone receptors (HR) estrogen receptor α (ERα) and/or ERα’s
downstream target, progesterone receptor (PR), without the expression of human epidermal growth factor
receptor 2 (HER2). These are classed as HR+, HER2-.2 Anti-hormonal therapies targeting ERα, either
indirectly with aromatase inhibitors which reduce the levels of estradiol, or directly with the selective
estrogen receptor modulator tamoxifen, have been the mainstay of treatment for HR+, HER2- breast
cancer patients for the last 40+ years.3 However, there remains a significant unmet clinical need, with
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 92
1
2
3
4
5
~10%-40% of patients receiving adjuvant anti-hormonal treatment developing advanced breast cancer
within 20 years4 and a 25% 5-year survival rate for patients with advanced disease.5
6
7
8
9
Whilst anti-hormonal therapies remain effective in the advanced setting, resistance emerges. Resistance
mechanisms include ligand independent activation of ERα through point-mutations of ERα,6,7 activation
of signaling pathways that can activate ERα in the absence of estradiol,8 or changes in expression of ERα
co-factors that convert ERα modulators into agonists.9 However, in all these scenarios, the cell maintains
a reliance on ERα, and effectively targeting ERα remains a valid therapeutic strategy. Selective ERα
degraders (SERDs), which both degrade and antagonize ERα, are one approach to continue to target
ERα.10
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 date, the only clinically approved SERD is fulvestrant (Faslodex), which can degrade and completely
antagonize ERα in all settings tested11 and has proven superior to the aromatase inhibitor anastrozole in
clinical trials.12 However, fulvestrant is administered clinically via monthly intramuscular injection,
which limits the maximum administrable dose to 500 mg.13 Therefore, there has been much interest in
developing orally bioavailable SERDs to further enhance the benefit to patients brought through this class
of ERα targeting agents. A number of potential oral SERD candidates have been advanced into clinical
trials with selected examples shown in Figure 1. These have been grouped into two families according to
the charge class of the side chain projecting from the aryl group. The first group is comprised of acrylic
acids with examples including GDC-081014 (1a) from Aragon/Seragon/Genentech, AZD949615 (1b) from
AstraZeneca, LSZ10216 (1c) from Novartis and G1T4817 (1d) from the University of Illinois at
Chicago/G1 therapeutics. The second group includes basic degraders as exemplified by RAD190118 (2a)
20
from Radius and GDC-092719 (2b) from Aragon/Seragon/Genentech. SAR439859 (2c) from Sanofi
also contains a basic side chain attached to the aryl group but notably contains an acidic group in the core
making this an example of a zwitterionic SERD.
Acrylic acid containing oral SERDs, exemplified by 1a and 1b were only able to degrade ERα in certain
cell lines and had the potential to agonize ER in uterine models and some breast cancer cell lines.21,22 This
divergence in chemotype from fulvestrant could theoretically lead to cofactor changes converting the
3
ACS Paragon Plus Environment

Page 5 of 92
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
drugs to ER agonists in the clinic, in a similar manner to tamoxifen. Some SERDs containing a basic
group, such as 2b, have been able to address this issue, degrading ERα to the same extent as fulvestrant in
all cell lines tested and giving no sign of ER agonism in breast cell lines.22
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. Selected examples of oral SERDs.
OH
O
OH
OH
O
OH
O
O
F
F
F
H
N
F
O
O
N
F
F
Cl
S
S
HN
O
HO
HO
N
1a
F
N
1b
1c
1d
F
O
NH
O
Cl
N
F
N
O
F
O
OH
Cl
HO
HO
HO
2c
2a
2b
O
As part of our ongoing research efforts at AstraZeneca, we have previously reported a series of tricyclic
indazole acid SERDs23 (3 as an exemplar), themselves derived from a series of tetrahydroisoquinoline
(THIQ) phenols (Figure 2).24 We had also described the discovery of a novel tricyclic indole core and the
optimization to the clinical candidate 1b.15 In subsequent work on this core, we also investigated
switching between the charge classes with the replacement of the acrylic acid with a basic group (4 as an
exemplar).25 Interestingly, this compound has also arisen independently from the research efforts of other
groups.19c, 26
During the course of our research efforts we became aware that 1b, whilst an excellent degrader in certain
ER+ cell lines (e.g. MCF-7 which showed 87% protein degradation vs fulvestrant by Western blot) did
not degrade to the same extent in other cell lines (e.g. CAMA-1 proving to be a particularly challenging
line with only 28% degradation vs fulvestrant). Indazole acid 3 showed a similar profile to 1b with better
degradation being observed in MCF-7 (84%) than CAMA-1 (56%). In contrast, basic compound 4
4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 92
1
2
3
4
5
showed a more balanced degradation profile between the cell lines with MCF-7 (77%) and CAMA-1
(69%).
6
7
8
9
RESULTS AND DISCUSSION
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 this paper, we describe our efforts to amalgamate both strands of work to realise a basic series of
tricyclic indazole SERDs with good degradation in both MCF-7 and CAMA-1 cell lines. Our design
strategy was predicated on maximizing cellular potency (pIC50 >9.5) whilst controlling lipophilicity
(logD7.4 <3) so as to minimize general risks associated with basic lipophilic compounds (e.g. hERG,
phospholipidosis, off-target pharmacology) and maximise our chances of identifying a candidate with
favourable drug like properties (e.g. good solubility, high permeability, high metabolic stability) as
summarized in Figure 2.
Figure 2. Medchem strategy for basic series of tricyclic indazole SERDs.
O
OH
F
F
N
Investigate conformational
restriction to maximise
degradation
N
O
O
LogD control
F
F
O
F
H
N
F
N
N
N
F
F
pKa & LogD control
property modulation
F
F
HN
HN
5
3
4
N
N
Our first example of a basic indazole, compound 5, showed very similar pharmacology to previously
reported indole 4 (pIC50 9.9 (Dmax 95%) for 5 vs pIC50 10.1 (Dmax 96%) for 4). The lipophilicity of 5 was
high (logD7.4 4.4) but measurable, whereas for 4 this was out of the range of our standard assay. The
hypothesis that the indazole was intrinsically more polar than the indole core was supported by lower
plasma protein binding (0.79% fraction unbound for 5 vs 0.23% for 4). The higher aqueous solubility (33
M for 5 vs <3 M for 4) also provided encouragement to embark on an optimization campaign and we
therefore developed chemical routes towards these motifs (5-40) as shown in Schemes 1-10.
5
ACS Paragon Plus Environment

Page 7 of 92
Journal of Medicinal Chemistry
1
2
3
4
Synthetic chemistry
5
6
7
8
9
Drawing on our previous experience,23 we initially chose to construct the tricyclic scaffold from a
Pictet-Spengler cyclisation on an aniline substrate, followed by late-stage formation of the
indazole ring. This approach required us to synthesize a range of anilines bearing a diversity of
N-alkyl substituents (Scheme 1). To this end we adopted two complementary approaches. In
the first, treatment of 2,6-dibromotoluene 41a with n-BuLi in THF at −78 oC, followed by addition
of cyclic sulfamidate 42 gave the aliphatic amine 43 after Boc deprotection. N-Alkylation could
be achieved by heating this amine with the requisite alkyl triflate and DIPEA in 1,4-dioxane to
give compounds 44. Buchwald amination with benzophenone imine, followed by hydrolysis with
aqueous acid afforded the anilines 45. In order to facilitate later-stage variation of the alkyl
chain, we also adopted an approach that introduced the aniline at an earlier stage and
circumvented repeated Buchwald aminations. 3-Bromo-2-methyl aniline was first protected as
the dimethyl pyrrole. With the protected aniline 41b in hand, an analogous approach was
adopted where the alkyl chain was introduced via ring-opening of the cyclic sulfamidate to give
46. Following Boc deprotection and removal of the pyrrole protecting group with hydroxylamine,
the diamine 47 could be selectively alkylated on the aliphatic nitrogen by heating with the alkyl
triflate as per the approach described above. In a slight variation of the method, the fluoro-
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
6
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 92
1
2
3
4
5
6
7
8
9
cyclopropyl analogue was accessed by amide formation on the diamine with HATU and
subsequent reduction of the amide by heating with borane solution in THF to afford the required
amine 45c.
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
Scheme 1. Routes to anilines bearing a diversity of N-alkyl substituents
2
R1 R R3
41a X=Br
(c)
(b)
H2N
(a)
N
H
45a R1=Me; R2=Me; R3=F
45b R1=Me; R2=Me; R3=H
45c R1=F; R2-R3=-CH2-CH2-
Br
R1
Br
Br
O
O
O
43
45d R1=F; R2-R3=-CH2OCH2-
44
HN
3R2
S
+
NBoc
45e R1=F; R2=F; R3=Me
R
H2N
(f)
45f R1=F; R2=F; R3=CH2OSi(Ph)2tBu
X
H2N
45g R1=F; R2=F; R3=H
H2N
45
41
42
45h R1=F; R2=F; R3=F
(d)
(e)
NH2
N
47
46
41b X=2,5-dimethylpyrrole
aReagents and Conditions: (a) n-BuLi, THF, −78 oC to 0 oC, 1 h, then 4 N HCl/dioxane, RT, 1 h, 60%; (b)
alkyl triflate, DIPEA, 1,4-dioxane, 90 oC, 63-74% or isobutyrylaldehyde, Na(OAc)3BH, THF, 0 oC, 56%;
(c) benzophenone imine, Pd2dba3, Rac-BINAP, NaOtBu, toluene, 90 oC, then 1 N aq. HCl, 71-85%; (d) n-
BuLi, THF, −78 oC to 0 oC, 1 h, then 4 N HCl/dioxane, RT, 4 h; e) NH2OH, NH2OH.HCl, EtOH, reflux.
o
84% over 2 steps; (f) alkyl triflate, DIPEA, 1,4-dioxane, 90 C, 44-100% or 1-fluorocyclopropane-1-
carboxylic acid, HATU, Et3N, DMF, RT, 61%, then BH3.THF, THF, 65 oC, 82%.
The anilines 45 were then subjected to a Pictet-Spengler cyclization with the requisite 4-bromo
aldehydes to furnish the cyclized products 48 as the desired trans isomers.27 The cyclization
was found to proceed more smoothly with a small amount of water present in the AcOH; it was
7
ACS Paragon Plus Environment

Page 9 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
hypothesized that the water helps to hydrolyse the imine formed from condensation of the
aniline with the aldehyde, allowing cyclization to occur. Indazole formation could then be
achieved by treatment with NaNO2 in a mixture of propionic acid and water under cooling.28 The
indazoles were then protected with THP to give 49, which allowed installation of the azetidine
moiety by a palladium-catalyzed ether formation facilitated by the RockPhos 3rd generation Pd-
precatalyst.29 Finally, removal of the THP protecting group under acidic conditions afforded the
tricyclic indazoles 5-13, 29 and 30 (Scheme 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
Scheme 2. Pictet-Spengler approach to O-linked tricyclic indazolesa
5 R1,R2=Me; R3=F; X=F; Y=CF (from 45a, 48a, 49a)
F
N
6 R1,R2=Me; R3=F; X=OMe; Y=CH (from 45a, 48b, 49b)
7 R1,R2=Me; R3=H; X=F; Y=CF (from 45b, 48c, 49c)
Br
8 R1,R2=Me, R3=H; X=H; Y=CH (from 45b, 48d, 49d)
9 R1,R2=Me; R3=H; X=OMe; Y=CH (from 45b, 48e, 49e)
10 R1,R2=Me; R3=H; X=H; Y=N (from 45b, 48f, 49f)
11 R1=F; R2-R3=-CH2CH2-; X=OMe; Y=CH (from 45c, 48g, 49g)
O
Br
Y
Y
Y
X
R1
(d,e)
X
X
(b,c)
R1
(a)
R1
R1
3 R2
12 R1=F; R2-R3=-CH2CH2-; X=H; Y=N (from 45c, 48h, 49h)
13 R1=F; R2-R3=-CH2OCH2-; X=OMe; Y=CH (from 45d, 48i, 49i)
HN
N
3R2
N
3R2
N
3 R2
R
R
H2N
R
R
THP
N
29 R1,R2=F; R3=H; X=H; Y=N (from 45g, 48j, 49j)
H2N
HN
N
30 R1,R2,R3=F, X=H; Y=N (from 45h, 48k, 49k)
N
45
49
48
aReagents and Conditions: (a) aldehyde, AcOH (1% H2O), 90 oC, then aq. HCl or NH2OH, 44-100%; (b)
o
NaNO2, propionic acid, H2O, −20 C, 17-75%; (c) 3,4-dihydro-2H-pyran, PTSA hydrate, DCM, reflux,
53-99%; (d) 2-[3-(fluoromethyl)azetidin-1-yl]ethanol, RockPhos 3rd generation Pd pre-catalyst, Cs2CO3,
toluene, 90 oC, 42-78%; (e) 4 N HCl dioxane, RT, 37-78%.
Starting from the THP protected tricyclic indazoles bearing a 4-bromo aryl substituent 49g, the acyclic
amine was introduced by Pd-catalyzed ether formation with the N-Boc protected ethanolamine (step a).
8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 92
1
2
3
4
5
6
7
8
9
Removal of both the Boc and THP protecting groups (step b) gave the desired acyclic ether 14. The oxy-
linked azetidine was found to be a challenging substrate to install by Pd-catalyzed coupling. Instead, the
bromo intermediate 49g was first converted to the phenol 51 via the pinacol boronate 50 (steps d, e).
Mitsonobu reaction with the N-Boc protected hydroxy-azetidine was then used to install the ether linkage
(step f) then deprotection of both the Boc and THP protecting groups followed by N-alkylation (steps h, i)
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
furnished the oxy-linked azetidine 15 (Scheme 3). To introduce the acyclic amino linker on the
pyridyl substrate (intermediate 49h’, intermediate before THP protection in the synthesis of
49h), Buchwald amination catalyzed by BrettPhos 3rd generation precatalyst30 installed the unprotected
acyclic diamine selectively on the primary amine. In this instance, it was possible to perform the coupling
reaction directly on the unprotected tricyclic indazole (R = H) to give 17. Amino-azetidines were also
accessed via Buchwald amination. For the methoxyphenyl substrate 49g, formation of the N-Boc
protected amino-azetidine occurred in excellent yield (91%); then simultaneous deprotection of both the
Boc and THP groups with HCl followed by N-alkylation with 3-fluoroiodopropane furnished the N-
alkylated azetidine 16. For the pyridyl substrate 49h, the amination was carried out using an amino-
azetidine with the fluoropropyl chain already pre-installed to give 18 after deprotection of the THP group.
Scheme 3. Variation of the O- and N-linked basic substituenta
9
ACS Paragon Plus Environment

Page 11 of 92
Journal of Medicinal Chemistry
1
2
3
4
F
N
R1
H
N
F
Z
O
5
6
7
8
Z=O (f,h,i)
Z=NH (g,h,i)
OMe
(a,b)
OMe
OMe
F
N
F
F
N
N
THP
N
N
9
HN
HN
49g R1=Br
50 R1=B(Pin)
51 R1=OH
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
(d)
(e)
15 Z=O
16 Z=NH
14
F
H
N
N
Br
F
HN
HN
N
N
(j)
N
(c)
F
N
F
F
N
N
R
N
HN
N
HN
49h R= THP
49h' R=H
N
17
18
N
aReagents and Conditions: (a) tert-butyl (3-fluoropropyl)(2-hydroxyethyl)carbamate, RockPhos 3rd
generation Pd pre-catalyst, Cs2CO3, toluene, 90 oC, 92% (from 49g); b) HCl/dioxane, RT, 29%; c) N1-(3-
fluoropropyl)ethane-1,2-diamine. 2HCl, BrettPhos 3rd generation Pd pre-catalyst, NaOtBu, dioxane, 50
oC, 18% (from 49h’); d) (B(Pin))2, Pd(dppf)Cl2, KOAc, dioxane, 80 C, 46%; e) Aq. H2O2, 1 M NaOH,
o
o
o
THF, 5 C, 63%; f) tert-butyl 3-hydroxyazetidine-1-carboxylate, DEAD, PPh3, toluene, 110 C, 75%
(from 51); g) tert-butyl 3-aminoazetidine-1-carboxylate, BrettPhos 3rd generation Pd pre-catalyst,
o
Cs2CO3, dioxane, MW, 100 C, 91% (from 49g); h) HCl/dioxane, RT, 36-87%; i) 1-fluoro-3-
iodopropane, DIPEA, NMP, RT, 18-39%; j) 1-(3-fluoropropyl)azetidin-3-amine, BrettPhos 3rd generation
Pd pre-catalyst, NaOtBu, dioxane, 90 oC, then DCM, TFA, RT, 10% (from 49h).
With this methodology in hand, we next investigated variation of the N-alkyl group on the piperidine ring.
As we established during our previous investigations, the alkyl group could be varied according to
Scheme 1; following formation of the N-alkyl diamines 45, Pictet-Spengler cyclisation to 48 followed by
indazole formation furnished the tricyclic ring (Method A) which could be THP protected to give 49. The
amino-azetidine could then be installed by Buchwald amination to give compounds 19-28 (Scheme 4).
10
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 92
1
2
3
Scheme 4. Pictet-Spengler approach to N-linked tricyclic indazoles (Method A)a
4
5
6
7
8
9
19 R1,R2=Me; R3=F; X=OMe; Y=CH (from 45a, 48b, 49b)
20 R1,R2=Me; R3=F; X=H; Y=N (from 45a, 48l, 49l)
21 R1,R2=F; R3=Me; X=OMe; Y=CH (from 45e, 48m, 49m)
22 R1,R2=F, R3=Me; X=H; Y=N (from 45e, 48n, 49n)
23 R1,R2=F; R3=CH2OH; X=OMe; Y=CH
N
F
Br
HN
Br
(from 45f, 48p, 49p with R3=CH2OSiPh2tBu)
24 R1,R2=F; R3=CH2OH; X=H; Y=N
(from 45f, 48q, 49q with R3=CH2OSiPh2tBu)
25 R1,R2=F; R3=H; X=OMe; Y=CH (from 45g, 48o, 49o)
26 R1,R2=F; R3=H; X=H; Y=N (from 45g, 48j, 49j)
27 R1,R2,R3=F; X=OMe; Y=CH (from 45e, 48r, 49r)
28 R1,R2,R3=F, X=H; Y=N (from 45e, 48h, 49h)
Y
Y
Y
X
R1
X
X
(b,c)
(d,e,f,g)
R1
(a)
R1
R1
3 R2
HN
N
3R2
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
3R2
N
3R2
R
R
H2N
R
R
THP
N
H2N
HN
N
N
45
49
48
aReagents and Conditions: (a) aldehyde, AcOH (1% H2O), 90 C, then aq. HCl or NH2OH, 16-68%; (b)
o
o
NaNO2, propionic acid, H2O, −20 C, 34-75%; (c) 3,4-dihydro-2H-pyran, PTSA hydrate, DCM; (d) if
R3=CH2OSiPh2tBu, 1 M TBAF, THF, RT, 71%; (e) 1-(3-fluoropropyl)azetidin-3-amine or tert-butyl 3-
aminoazetidine-1-carboxylate, BrettPhos 3rd generation Pd pre-catalyst, NaOtBu or Cs2CO3, dioxane, 27-
88%; (f) HCl / dioxane, RT, 40-92%; (f) (when tert-butyl 3-aminoazetidine-1-carboxylate used in step
(e)) 1-fluoro-3-iodopropane, DIPEA, NMP, RT, 29-51%.
As our interest in compound 28 grew, we sought an improved route that would circumvent the indazole
formation step, as we envisaged that this could be challenging to carry out on multigram scale. To this
end we found that when a combination of a 2-pyridyl aldehyde and a poly-fluorinated N-alkyl sidechain
were employed, Pictet-Spengler cyclisation was possible on the pre-formed indazole (Method B). The
indazole starting material was accessed by lithiation of 4-bromo indazole 52 and subsequent reaction with
the cyclic sulfamidate 42. Boc deprotection and selective alkylation of the aliphatic amine with the alkyl
triflate afforded the N-trifluoroethyl alkylated indazole 53. Pictet-Spengler cyclisation in a toluene / TFA
mixture then afforded the trans tricyclic indazole 54. Finally, the amino-azetidine was added in the usual
manner through Buchwald amination. Pleasingly we found that the Buchwald amination could proceed
without the need to protect the indazole provided that an excess of base was used, circumventing the need
for a protection / deprotection sequence. Through these improvements the route towards 28 was reduced
from 9 synthetic steps to only 4 steps and provided 28 in an overall yield of 30% (Scheme 5).
11
ACS Paragon Plus Environment

Page 13 of 92
Journal of Medicinal Chemistry
1
2
3
Scheme 5. Pictet-Spengler approach on pre-formed indazole to access 28 (Method B)a
4
5
6
7
N
F
Br
HN
8
9
N
N
N
N
Br
O
O
O
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
(d)
(c)
(a,b)
F
F
S
F
F
F
F
HN
NBoc
N
+
F
N
H
F
F
HN
HN
HN
53
N
N
42
28
52
54
N
aReagents and Conditions: (a) n-BuLi, tert-butyl (R)-4-methyl-1,2,3-oxathiazolidine-3-carboxylate 2,2-
o
dioxide THF, −78 C, 57% then HCl / dioxane, DCM, RT, 57%. Isolated as .2HCl salt. (b) 2,2,2-
trifluoroethyl trifluoromethanesulfonate, K2CO3, DCM, reflux, 93%; (c) 5-bromopicolinaldehyde,
toluene, TFA, 80 C, 75%; (d) 1-(3-fluoropropyl)azetidin-3-amine, BrettPhos 3rd generation Pd pre-
o
catalyst, NaOtBu, dioxane, 55 oC, 74%.
We next sought to access 1-methyl substituted quaternary tricyclic indazoles. Pictet-Spengler cyclisations
with ketones are typically more challenging and often require specialised conditions when compared to
aldehydes.31 After initial unsuccessful attempts to close the ring in this way, an alternative approach was
developed. We envisaged that the quaternary centre could be formed by addition of suitable nucleophilic
species to the iminium ion of the pre-cyclised system. Thus, the THP protected tricyclic indazole 49j was
treated with cerium(IV) ammonium nitrate in acetonitrile/water to form the iminium ion, which could
then be reacted with methylmagnesium bromide to form the quaternary centre of 55 as a 3:2 mixture of
diastereoisomers in low yield.32 Removal of the THP protecting group and subsequent Buchwald
amination and separation of the isomers afforded the desired product 31 (Scheme 6).
Scheme 6. Oxidation / Grignard approach to the quaternary substituted core 31a
12
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 92
1
2
3
4
5
6
7
8
9
F
N
Br
Br
N
HN
N
N
(a)
(b,c)
N
N
F
F
N
F
F
THP
F
N
N
THP
N
F
HN
N
49j
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
55
31
N
aReagents and Conditions: (a) cerium(IV) ammonium nitrate, MeCN/H2O, RT, then MeMgBr, THF, -78
oC, 14%, 3:2 mixture of diastereoisomers; (b) HCl / dioxane, MeOH, RT, 28%, 3:2 mixture of
diastereoisomers; (c) 1-(3-fluoropropyl)azetidin-3-amine, BrettPhos 3rd generation Pd pre-catalyst,
NaOtBu, dioxane, 65 oC, SFC separation, 12%.
Introduction of geminal-dimethyl substitution into the saturated ring could be achieved from the
THP-protected 4-bromo-indazole 56 according to our previously reported procedure.23 Lithium-
halogen exchange with n-BuLi followed by quenching with dimethyloxirane and BF3.Et2O
afforded the tertiary alcohol 57. Ritter reaction with chloroacetonitrile gave 58 which following
cleavage of the amide with thiourea furnished the gem-dimethyl amine 59. The saturated ring
was then constructed with a Pictet-Spengler cyclization using 2-pyridyl-4-bromobenzaldehyde
on the unalkylated amine to give 60. THP protection of the racemic indazole gave 61; then
alkylation with the difluoroethyl triflate and DIPEA afforded 62. Subsequent Buchwald amination
reaction installed the amino-azetidine and deprotection of the THP group and chiral separation
of the enantiomers afforded enantiopure 32 (Scheme 7). In the case of 2-trifluoroethyl triflate,
alkylation of the sterically congested cyclic amine proved to be challenging. Thus, in a related
13
ACS Paragon Plus Environment

Page 15 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
approach, the more accessible gem-dimethyl amine could instead be alkylated, and the
resulting product taken through the same reaction sequence.
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
Scheme 7. Pictet-Spengler route to gem-dimethyl substituted tricyclic indazole 32 and 33a
O
Cl
Br
H2N
N
H
HO
F
F
HN
(i)
(a)
(c)
(b)
F
N
N
N
N
HN
N
N
N
N
H
N
H
THP
THP
63
58
56
57
59
(d)
(j)
F
N
F
N
Br
Br
Br
HN
HN
N
N
N
N
(g,h)
N
N
(f)
N
N
(k,h)
F
NH
F
F
F
F
N
F
F
F
F
X
N
THP
N
N
F
HN
N
HN
HN
N
60 X=NH
61 X=NTHP
N
64
32
62
33
(e)
o
aReagents and Conditions: (a) n-BuLi, 2,2-dimethyloxirane, BF3.Et2O, THF, −78 C, 1 h, 62%; (b) 2-
chloroacetonitrile, AcOH, H2SO4, RT, 48 h, 59%; (c) thiourea, EtOH, AcOH, reflux, 16 h, 83%; (d) 5-
o
bromopicolinaldehyde, toluene, TFA, 150 C, MW, 72%; (e) 3,4-dihydro-2H-pyran, PTSA hydrate,
DCM, 50 oC, 95%; (f) 2,2-difluoroethyl trifluoromethanesulfonate (0.4 M in DCM), DIPEA, dioxane, 60
oC, 81%; (g) 1-(3-fluoropropyl)azetidin-3-amine, BrettPhos 3rd generation Pd pre-catalyst, NaOtBu,
o
dioxane, 60 C, then DCM, TFA, RT, 68%; (h) SFC chromatography; (i) 2,2,2-trifluoroethyl
trifluoromethanesulfonate (0.32 M in chloroform), DIPEA, dioxane, 65 oC, 53%; (j) 5-
bromopicolinaldehyde, toluene, TFA, 100 oC, 25%; (k) 1-(3-fluoropropyl)azetidin-3-amine, BrettPhos 3rd
generation Pd pre-catalyst, NaOtBu, dioxane, 80 oC, 30%.
In order to access the analogous indole core it was necessary to develop a new route drawing on our
experience from the indazole series (Scheme 8). 1-Bromo-2-methyl-3-nitrobenzene 65 was treated with
potassium hydroxide and formaldehyde to afford the homologated hydroxy compound 66. Following
14
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 92
1
2
3
4
5
6
7
8
9
nitro reduction, the free aniline 67 was protected as a 2,5-dimethylpyrrole and the hydroxy group with a
4-methoxybenzyl to give fully protected bromide 68. We then employed our previously established
methodology to install the alkyl-amino unit via lithium-halogen exchange and ring opening of the cyclic
sulfamidate. Following removal of both the N-Boc and pyrrole protecting groups, the diamine 69 was
alkylated through formation of the trifluoroacetamide and subsequent reduction with borane solution in
THF. Alkyl diamine 70 successfully underwent Lewis-acid mediated Pictet-Spengler cyclisation to afford
the cyclic intermediate 71, following removal of the PMB protecting group with HCl in dioxane. Indole
formation was then achieved through iridium catalyzed cyclisation of the amino-alcohol33 to give 72.
Finally, the amino-azetidine moiety was installed through Buchwald-amination to give 34.
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 8. Pictet-Spengler route to tricyclic indole 34a
(a)
(c,d)
(b)
(e,f,g)
NH2
PMB
O2N
Br
O2N
Br
N
Br
H2N
H2N
Br
O
N
65
67
Br
66
68
69
OH
O
OH
PMB
F
Br
HN
F
HN
(h,i)
(j,k)
(l)
(m)
F
N
F
N
H2N
N
N
F
F
O
N
F
PMB
F
F
70
N
F
F
H2N
71
F
F
HN
OH
HN
72
34
aReagents and Conditions: (a) paraformaldehyde, KOH, DMA, RT, 72%; (b) ammonium formate, Zn,
o
MeOH, RT, 96%; (c) hexane-2,5-dione, PTSA hydrate, toluene, 100 C, 98%; (d) PMBCl, NaH, NaI,
DMF, RT, 80%; (e) n-BuLi, tert-butyl (4R)-4-methyl-1,2-oxathiolane-3-carboxylate 2,2-dioxide, THF, -
o
o
78 C, 61%; (f) HCl / dioxane, MeOH, RT, 57%; (g) Aq. NH2OH, NH2OH.HCl, EtOH, 50 C, 36%; (h)
o
ethyl 2,2,2-trifluoroacetate, DIPEA, MeOH, RT, quant. (i) BH3.THF, THF, 65 C, 77%; (j) 5-
o
bromopicolinaldehyde, Yb(III)Tf, MeCN / H2O, 70 C, 77%; (k) HCl / dioxane, MeOH, RT, 79%; (l)
pentamethylcyclopentadienyl iridium dichloride, K2CO3, toluene, 100 oC, 80%; (m) 1-(3-
fluoropropyl)azetidin-3-amine, BrettPhos 3rd generation Pd pre-catalyst, NaOtBu, dioxane, 50 oC, 62%.
15
ACS Paragon Plus Environment

Page 17 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Variation of the azetidine moiety was achieved by making use of the common tricyclic indazole
intermediate 54, the synthesis of which is described in Scheme 5. Buchwald amination proceeded
smoothly with the quaternary methyl azetidine to give 73, after which the Boc group could be removed
under acidic conditions and the azetidine alkylated in the usual manner to give 36. The aniline of 28 could
also be homologated under reductive amination conditions using paraformaldehyde and NaCNBH3 to
afford the N-Me compound 37 (Scheme 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
Scheme 9. Variation of the azetidine substituenta
NBoc
N
F
N
F
R1
Br
HN
HN
N
N
N
N
N
N
N
N
N
(a)
(b)
F
F
F
F
F
F
F
F
F
F
F
F
HN
HN
HN
HN
N
54
73
N
N
36
R1=H 28
R1=Me 37
N
(c)
aReagents and Conditions: (a) Amine, BrettPhos 3rd generation Pd pre-catalyst, NaOtBu, dioxane, 57%;
(b) TFA, DCM, RT, then 1-fluoro-3-iodopropane, DIPEA, DMF, RT, 75%; (c) paraformaldehyde,
NaCNBH3, AcOH, iPrOH, 70 oC, 72%.
Finally, we wished to re-investigate the amine component. Using the same tricyclic bromide 54,
Buchwald amination proceeded with the CbZ protected acyclic diamine to give 74, after which the
protecting group could be removed by hydrogenation to give 35. Buchwald amination was successful
with Boc protected (S)-amino-pyrrolidine on unprotected indazole 54 to give 75 but with the Pd-catalysed
ether formation with (S)-hydroxy-pyrrolidine required THP-protected indazole 49k to give 76. Boc
deprotection and subsequent reaction with 3-fluoro iodopropane afforded the N-alkylated products 38 and
40. Finally, in the case of amino-pyrrolidine 39, reductive amination on the aniline 38 with formaldehyde
and NaCNBH3 afforded the N-methylated product.
16
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 92
1
2
3
4
5
6
Scheme 10. Variation of the N-linked basic substituenta
7
8
R1
BOC
N
F
N
9
N
Br
NH
X
X
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
F
N
N
N
(c (X=NH))
(d (X=O))
N
N
N
N
38 X=NH
39 X=NMe
40 X=O
(e)
(a)
F
F
F
F
F
N
F
F
F
F
F
F
F
Z
N
HN
Z
N
HN
N
N
74 R1=CbZ
35 R1=H
75 Z=NH; X=NH
76 Z=NTHP; X=O
38 X=NH
39 X=NMe
54 Z=NH
49k Z=NTHP
(f)
(b)
aReagents and Conditions: (a) benzyl (2-aminoethyl)(3-fluoropropyl)carbamate, BrettPhos 3rd generation
Pd pre-catalyst, NaOtBu, dioxane, RT, 58% (from 54); (b) Pd/C, H2, EtOH, RT, 80%; (c) (X=NH): tert-
butyl (S)-3-aminopyrrolidine-1-carboxylate, BrettPhos 3rd generation Pd pre-catalyst, NaOtBu, dioxane,
80 C, 81% (from 54); (d) (X=O): tert-butyl (S)-3-hydroxypyrrolidine-1-carboxylate, RockPhos 3rd
o
o
generation Pd pre-catalyst, Cs2CO3, toluene, 90 C, 70% (from 49k); (e) DCM, TFA, or HCl / dioxane,
MeOH, RT, then 1-fluoro-3-iodopropane, DIPEA, DMF, RT, 14-69%; (f) paraformaldehyde, NaCNBH3,
AcOH, MeOH, 50 oC, 19%.
Initial optimization: control of lipophilicity and pharmacology
Our initial optimization began with a matrix of indazoles altering the aryl ring substitution and varying
the alkyl chain (Table 1). Compounds were evaluated in an ER binding assay and cellular degradation
assay in MCF7 cells as described previously.23 The SAR in our program was primarily driven by the
cellular assay but ER degradation in MCF-7 and CAMA-1 cells was also determined by Western blot for
key compounds. Based on previously reported SAR with the indazole acids, we were pleased to find that
with the 2-FiBu sidechain (R1=A), the switch from the 2,6-difluorophenyl (Core W) to the 2-
methoxyphenyl (Core Y) maintained potency (c.f. 5 vs 6). Importantly, this was associated with a
reduction in lipophilicity (logD7.4 -0.8) and a corresponding increase in fraction unbound and reduction
in rat hepatocyte clearance. In the case of the Bu sidechain (R1=B), deletion of the fluorine substituents
i
17
ACS Paragon Plus Environment

Page 19 of 92
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
(Core X) allowed lipophilicity reduction (c.f. 7 vs 8), and addition of the 2-methoxyphenyl (Core Y)
provided our first compound 9 with logD7.4 <3. Importantly, investigation of heteroaryl rings revealed
that a 2-pyridyl (Core Z) 10 was tolerated in terms of potency. An additional strategy to lower the
lipophilicity involved modifying the alkyl substituent with the formal linkage of the two methyl groups of
the 2-FiBu sidechain to form a cyclopropyl ring (R1=C). Comparison of the matched pair 11 with 6 (core
Y) revealed significant lipophilicity lowering (logD7.4 -0.8) with no effect on potency. Compound 12
with this alkyl sidechain also displayed good potency at low lipophilicity (logD7.4 2.6). Further
investigation highlighted the fluorooxetane (R1=D) as a group that could also be used to access highly
potent, low lipophilicity compounds such as 13.
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. Data for aryl ring and alkyl chain variation.
R1
F
F
F
F
A =
N
N
N
N
O
O
F
O
O
B =
C =
D =
N
N
F
F
O
R1
R1
R1
R1
F
F
N
N
N
HN
HN
HN
HN
N
N
N
N
Y
Z
W
X
O
Core R1
ER
ER DR
pIC50
LogD7.4 Human % Rat/Human % deg vs Fv
Cpd
bind
hepatocyte
CLint
c
d
f
(LLE)
Free
MCF-7/CAMA-1
(Dmax
a
pIC50
(µL/min/106
b
(%))
e
cells)
5
W
Y
A
9.0
8.5
9.9 (95)
9.8 (95)
4.4 (5.5)
3.6 (6.2)
0.8
6.5
88/8
84/66
69/74
-/-
6
7
A
B
28/12
g
9.1
8.7
W
X
9.7 (99)
9.8 (99)
4.6 (5.1)
3.4 (6.4)
0.4
11
74/9
66/8
g
-/-
8
B
18
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 92
1
2
3
4
5
6
7
8
9
Y
Z
B
B
8.1
9.3 (98)
9.4 (99)
2.4 (6.9)
3.2 (6.2)
52
19
9/3
94/74
-/-
g
8.0
8.0
10
41/8
g
h
79/88
75/87
79/75
9.9 (96)
9.4 (98)
11
12
13
Y
Z
C
C
D
2.8 (7.1)
2.6 (6.8)
2.7 (6.9)
21
34
25
20/7
38/8
42/-
g
8.9
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
g
8.4
Y
9.6 (95)
aER binding based on n  2 with SEM within 0.2 units, unless otherwise stated; bER degradation
based on n  2 with SEM within 0.2 units, unless otherwise stated and degradation (%Dmax
)
c
relative to fulvestrant ; logD7.4 determined by shake flask method with a buffer/octanol volume
d
ratio of 100:1 (LLE = ER DR pIC50-logD7.4); Determined from DMSO stock solution by
e
equilibrium dialysis in 10% human plasma supplied by Quintiles; Rate of metabolism
(µl/min/106 cells) determined from DMSO stock solution in isolated hepatocytes diluted to 1x106
f
cells/mL; % degradation of ER relative to fulvestrant (Fv) in MCF-7 and CAMA-1 cells as
g
determined by Western blot at 0.1 µM; n = 1; hSEM = 0.23.
With compounds in the desired lipophilicity space, we turned our attention to optimization of the basic
group fixing the alkyl chain as the 2-FcPr substituent (R1=C from Table 1). The SAR was found to be
particularly steep in this region with most changes resulting in poorer potency or degradation efficiency.
One aspect that did prove productive however was shifting the connectivity of the azetidine ring. Acyclic
base 14 (core S) represents a ring-opening of the azetidine and resulted in a significant reduction in the
extent of degradation (c.f. 14 vs 11 Dmax 82% vs 96%). Re-construction of an azetidine ring proximal to
the aryl ring (core T) restored the degradation efficiency (c.f. 15 vs 11 Dmax 97% vs 96%) albeit with
increased lipophilicity. Switching the linkage from an ether to an amine (16) maintained potency and
19
ACS Paragon Plus Environment

Page 21 of 92
1
Journal of Medicinal Chemistry
2
3
degradation efficiency whilst lowering lipophilicity and giving a compound with low clearance in both rat
4
5
and human hepatocytes.
6
7
8
9
Similar SAR was observed with the pyridyl ring; the acyclic base 17 (Core U) with an amine linker had
slightly lower degradation efficiency (Dmax 93%) with cyclisation to form an azetidine 18 (Core V)
restoring this (Dmax 96%) and giving a compound in our target zone of lipophilicity. Importantly, the
degradation measured by Western blot in both MCF-7 and CAMA-1 lines was generally higher for this
arrangement of the azetidine than the motif described in Table 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
a
Table 2. Data for basic group variation in the methoxyphenyl and pyridine series.
H
N
F
N
H
N
F
N
F
F
L
L
L
L
N
N
N
N
O
O
F
F
F
F
N
N
HN
HN
HN
HN
N
N
N
N
S
U
T
V
Core
L
ER
ER DR
pIC50
(Dmax
(%))
LogD7.4 Human % Rat/Human % deg vs Fv
Cpd
bind
pIC50
(LLE)
Free
hepatocyte
CLint
MCF-7/CAMA-
1
(µL/min/106
cells)
b
c
-/-
8.1
9.5 (82)
14
S
O
2.4 (7.1)
18
9/-
b
90/90
9.4 (97)
15
16
17
18
T
T
O
8.1
8.1
3.5 (5.9)
2.4 (7.1)
2.0 (7.0)
2.5 (6.7)
11
34
-
16/-
7/2
NH
NH
NH
9.5 (96)
9.0 (93)
9.2 (96)
89/84
-/-
b
8.0
U
V
22/-
24/5
8.2
26
90/102
aLegend as Table 1; bn = 1; cSEM = 0.29.
20
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 92
1
2
3
Structure and conformation of early leads
4
5
6
7
8
9
Both compounds 14 and 16 (Table 2) show comparable binding potency and high cellular pIC50, but
significant enhancement in maximum degradation (Dmax) was observed for 16 (96% vs 82%). We
investigated the conformation of the para-linked base as well as the behavior of the ortho-methoxy
substituted ring by NMR. Solution conformational analysis of 16 by NMR shows that the para-linked
basic group is able to rotate; however, unlike the fully flexible linear chain of 14, the base is more
restricted in the space it can access, as a result of the lower number of rotatable bonds (Figure 3a). The
NMR analysis highlighted that the aryl ring of 16 is rotationally restricted by the presence of the methoxy
group which is oriented below the plane of the molecular core. The orientation of this group corresponds
to the one seen in the protein-ligand crystal structure, where the methoxy group fills a small pocket
towards the back of the binding site (Figure 3a,b). The same orientation of the methoxy group was
observed previously for tricyclic indazoles carrying an acrylic acid group rather than a basic one, with the
group filling the same pocket.23 The N-alkyl side chain is conformationally rigid with the hydrogens in the
α-position and the fluorine pointing away from the phenyl ring, again corresponding closely to the
binding mode observed in the protein-ligand crystal structure (Figure 3a,c).
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 binding of the [5.6.6] indazole core of the molecule is in line with the binding mode seen
previously.23 The tricyclic core is bound in a deep pocket that is mostly lined with hydrophobic residues.
The nitrogens of the tricyclic indazole interact with a conserved water and form short hydrogen-bonding
contacts with Glu353, which itself forms a water-mediated salt bridge to Arg394. Unlike the previously
reported tricyclic [6.5.6] indole core of 1b, the tricyclic indazole lacks the indole NH hydrogen bond
donor on the central ring that interacts with the backbone carbonyl of Leu346.15 However, the distance of
the carbonyl oxygen from the protein backbone to the central ring of the tricyclic indazole is short (3.6Å),
and this is not an unfavourable contact.
21
ACS Paragon Plus Environment

Page 23 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Figure 3 (a) Solution conformation of 16. Flexibility is illustrated with thin-wire depictions of the basic
group which is capable of adopting different conformations in solution; (b) X-ray crystallography of 16
(cyan, pdb: 6zoq) bound to the ERα ligand binding domain construct with aligned structure of tamoxifen
(white, pdb: 5aav) for reference; (c) Aligned protein-ligand crystal structures of 16 and 18 (magenta, pdb:
6zos). The loop connecting helices 11 and 12 is tagged in b and c, as well as key residues.
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 closer analysis of the protein-ligand complexes of 16 and 18, which have ortho-methoxy phenyl and
pyridyl ring substituents, shows a close alignment of the core binding modes, and similar placements of
the basic group (Figure 3c). The orientation of the pyridyl group in 18 cannot be discerned in the crystal
structure, and was observed to be freely rotatable in solution by NMR. The protein construct used in these
structures has a L536S mutation in the loop connecting helix 11 (H11) and helix 12 (H12) that biases its
conformation and that of the connected H12 towards the antagonist orientation.34 The fluorinated terminal
group of the basic azetidine is positioned close to H12. In the case of compound 16, the base forms a
strong hydrogen-bonding contact to the backbone carbonyl of Val533 (Figure 3c, distance 2.9 Å), which
resides in the flexible loop connecting helices 11 and 12. The structure of 18 suggests that the base is
oriented towards the acidic side chain of Asp351, forming again a strong contact (distance 2.8 Å). The
proximity of the basic group to H12 and its ability to stabilize the conformation of a flexible loop by
forming close hydrogen-bonded contacts may suggest that this plays a role in the process that ultimately
leads to recognition of ERα by the cellular degradation machinery.
22
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 92
1
2
3
4
5
6
7
8
9
In addition to the solution and protein-ligand structures, a small molecule crystal structure of 16 was also
obtained to allow unambiguous confirmation of the chemical structure and stereochemistry. Single crystal
analysis of 16 revealed a sesqui-tert-butyl methyl ether (TBME) solvate form that crystallized in a non-
symmetric monoclinic C2 space group (Figure 4). The ortho-methoxy phenyl group showed a similar
orientation to that observed in the solution and protein-ligand structures. The terminal fluoropropyl chain
was orientated as a low energy anti-conformer with a staggered conformation which is in line with the
solution conformation measured by NMR. The N-alkyl group in the piperidine ring is found in a pseudo
equatorial position while the methyl group reflects a pseudo axial position. Intermolecular hydrogen
bonding was observed in the crystal lattice between the indazole NH (N22) and the N atom of the
azetidine ring (N27) and also between the anilinic NH (N7) and the indazole N atom (N23) constructing
a two-dimenstional network in the crystal lattice.
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)
(b)
Figure 4 (a) Molecular conformation of 16 from single crystal structure; (b) Crystal packing of 16 in the
unit cell (the disordering TBME molecules are shown as ball and stick models).
Physicochemical and DMPK profiling of 16
23
ACS Paragon Plus Environment

Page 25 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The favorable potency and human in vitro clearance characteristics of 16 prompted us to profile
this further as an exemplar of the series (Table 3). In terms of its physicochemical properties, 16
was highly soluble in aqueous media (>900 µM) but had modest intrinsic cellular permeability
(Papp 1.8x10-6 cm/s) with significant potential for being a substrate for active transport (efflux
ratio = 28) as measured in a Caco-2 assay. Compound 16 showed high fraction unbound levels
in plasma protein binding assays across species (>10% free drug) and was di-basic with both
the azetidine and piperidine (measured pKa(BH+) 8.7 and 7.8 respectively) having the potential to
protonate at physiological pH. Profiling against a high throughput human liver microsome panel
of five cytochrome P450 enzymes showed a clean profile against 3 isoforms (CYP1A2,
CYP2C19, CYP2C9) but activity against CYP2D6 (IC50 1.2 µM) and CYP3A4 (IC50 11 µM).
Additionally, the compound was shown to have the potential to interact with the hERG ion
channel (IC50 9.1 µM) which became a focus of subsequent optimization efforts.
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. Physicochemical & in vitro DMPK data for 16.
Intrinsic
Caco2
Mu/Rat/Dog
% Free
CYP 2D6 CYP 3A4 hERG
(µM)d (µM)d (µM)e
c
Sol (µM)b
pKa(BH+)
Papp
(efflux
24
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 92
1
2
3
ratio)a
4
5
6
1.8 (28)
>900
39/51/22
8.7/7.8
1.2
11
9.1
7
8
9
aCompounds were incubated at 10 M in cultured Caco-2 cells. Intrinsic permeability was
measured in units of x10-6 cm/s in the apical direction in the presence of transport inhibitors
quinidine (50 M), sulfasalazine (20 M), and benzbromarone (30 M) withan an
apical:basolateral pH of 6.5:7.4, respectively. Efflux ratio is reported as the ratio of basolateral
(B to A) to apical (A to B) transport in the absence of any inhibitors and an apical:basolateral pH
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
b
o
of 7.4:7.4; Solubility of compounds in aqueous phosphate buffer at pH 7.4 after 24 h at 25 C
c
from DMSO stock solution; pKa measured using UV absorption (Sirius) in 0.1 M potassium
o
d
e
chloride at 20 C; Inhibition of cytochrome P450 enzymes IC50. Inhibition of the hERG tail
current was measured using a plate-based planar patch clamp system (Syncropatch).
We also obtained in vivo rat pharmacokinetic data (Table 4). In rats, 16 was found to have very
high clearance (125 mL/min/kg) and exhibited poor bioavailability (3%), likely due to first pass
clearance by the liver. The volume of distribution was assessed at 71 L/kg and resulted in a
long half-life of 11 hours. This was consistent with the dibasic nature of the compound as
discussed above. While a large volume of distribution would aid in achieving a desirable long
25
ACS Paragon Plus Environment

Page 27 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
half-life in human, it would potentially carry risks of toxicity under chronic administration
scenarios (e.g. phospholipidosis). Accordingly, our strategy became to reduce the volume of
distribution by attenuating the di-basicity whilst maintaining the low human in vitro clearance
characteristics.
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
Table 4. Summary of pharmacokinetic parameters of 16 after IV and PO dosing in rat.
Species Clint
Dosea
CL
Vdss
t1/2
(h)
%F
(µL/min/106 IV/PO
(mL/min/kg)
(L/kg)
cells)
7
(mg/kg)
0.5/1
Rat
125
71
11
3
aCompound was dosed in rat intravenously at 0.5 and orally at 1 mg/kg. The formulation was
5% DMSO:95% hydroxylpropyl -cyclodextrin (30% w/v) at a volume of 1 mL/kg and 4 mL/kg,
IV and PO respectively.
Final optimization: Identification of a candidate
We aimed to reduce the volume of distribution by lowering the basicity of the piperidine ring nitrogen
through aliphatic fluorination of the side chain in both the methoxyphenyl (Core Q) and pyridyl (Core R)
series (Table 5). Formal opening of the cyclopropyl ring gave the 2-FiBu sidechain (A) present in 1b with
similar levels of potency for 19 and 20 but increased lipophilicity as expected. Using the 2,2-
difluoropropyl sidechain (B) allowed us to maintain potency whilst reducing the pKa of the piperidine in
21 and 22. For 21, the lipophilicity was undesirably high and to ameliorate this we introduced a hydroxy
26
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 92
1
2
3
group on the terminal methyl (C). This decreased potency (pIC50 -0.7 and -0.2) whilst dramatically
4
5
6
7
8
9
lowering the lipophilicity (logD7.4 -1.2 and -0.9) for matched pairs 23 vs 21 and 24 vs 22 respectively.
Deletion of the hydroxymethyl substituent gave the 2,2-difluoroethyl sidechain (D) which gave highly
potent compounds 25 and 26 with the pyridyl (Core R) having lower lipophilicity than the
methoxyphenyl (Core Q). Addition of a third fluorine atom to the ethyl sidechain (E) resulted in a modest
rise in lipophilicity and exquisitely potent compounds 27 and 28. Pyridyl compounds 26 and 28 matched
both our potency and lipophilicity criteria and with low turnover in human hepatocytes.
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
Table 5. Data for alkyl group variation in the 3-fluoropropyl azetidine base series.
N
F
N
F
R1
D =
HN
HN
F
F
A =
F
F
N
N
O
B =
C =
E =
R1
R1
F
F
F
F
F
N
OH
HN
HN
F
N
N
Q
R
Core R1
ER
ER DR
LogD7.4
(LLE)
Human % Rat/Human % deg vs Fv
Cpd
bind
pIC50
pIC50
(Dmax
(%))
Free
hepatocyte
CLint
MCF-
7/CAMA-1
(µL/min/106
cells)
19
Q
R
Q
R
A
A
B
B
8.2
9.7 (96)
9.5 (96)
3.2 (6.5)
3.2 (6.3)
14
10
8/-
86/89
-/-
b
8.5
20
21
22
26/-
15/5
49/-
8.6
10.4 (98) 3.7 (7.1)
4.3
24
80/101
b
102/100
8.4
9.5 (98)
2.9 (6.6)
b
-/-
8.3
23
Q
C
9.7 (98)
9.3 (99)
2.5 (7.2)
2.0 (7.3)
23
20/-
24
25
R
Q
C
D
8.2
8.7
64
12
16/-
104/105
99/100
10.0 (99) 3.2 (6.8)
14/6
27
ACS Paragon Plus Environment

Page 29 of 92
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
26
27
28
R
Q
R
D
E
E
8.4
9.5 (99)
10.2 (98) 3.6 (6.6)
9.8 (99) 2.9 (6.9)
2.4 (7.1)
40
6.0
23
22/5
12/5
23/6
95/98
b
95/110
8.3
8.6
94/98
aLegend as Table 1; bn = 1.
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
With 26 and 28 looking particularly promising, the corresponding compounds with the original basic
group 29 and 30 were made for comparison (Table 6). These proved to have very similar potency,
lipophilicity and properties (c.f. 26 vs 29 and 28 vs 30). Further profiling however revealed important
differences in their pharmacology in terms of the ability of the compounds to degrade ER across different
cell lines. Each of the four compounds (26, 28-30) were effective in degrading ER in the MCF-7 cell line
but compounds 29 and 30 were notably less effective in the CAMA-1 cell line.
a
Table 6. Data for O-linked azetidine compounds.
F
N
O
N
F
N
R1
F
HN
N
P
Core R1
ER
ER DR
pIC50
(Dmax
(%))
LogD7.4
(LLE)
Human % Rat/Human % deg vs Fv
Cpd
bind
pIC50
Free
hepatocyte
CLint
MCF-
7/CAMA-1
(µL/min/106
cells)
b
97/88
91/80
8.6
29
30
P
P
H
9.3 (98)
9.4 (96)
2.4 (6.9)
3.0 (6.4)
49
22
36/5
45/3
c
8.9
F
aLegend as Table 1; bSEM = 0.34; cSEM = 0.21.
This finding was mirrored across the larger data set as shown in Figure 5.35 Compounds with the two-
carbon link to the azetidine (A=NH/O) could sometimes achieve >90% degradation in the MCF-7 cell
28
ACS Paragon Plus Environment