Discovery of A-971432, an orally bioavailable selective sphingosine-1-phosphate receptor 5 (S1P5) agonist for the potential treatment of neurodegenerative disorders

Article abstract of DOI:10.1021/acs.jmedchem.5b00928

S1P₅ is one of 5 receptors for sphingosine-1-phosphate and is highly expressed on endothelial cells within the blood-brain barrier, where it maintains barrier integrity in in vitro models (J. Neuroinflamm. 2012, 9, 133). Little more is known about the effects of S1P₅ modulation due to the absence of tool molecules with suitable selectivity and drug-like properties. We recently reported that molecule A-971432 (Harris et al., 2010) (29 in this paper) is highly efficacious in reversing lipid accumulation and age-related cognitive decline in rats (Van der Kam, et al., AAIC 2014). Herein we describe the development of a series of selective S1P₅ agonists that led to the identification of compound 29, which is highly selective for S1P₅ and has excellent plasma and CNS exposure after oral dosing in preclinical species. To further support its suitability for in vivo studies of S1P₅ biology, we extensively characterized 29, including confirmation of its selectivity in pharmacodynamic assays of S1P₁ and S1P₃ function in rats. In addition, we found that 29 improves blood-brain barrier integrity in an in vitro model and reverses age-related cognitive decline in mice. These results suggest that S1P₅ agonism is an innovative approach with potential benefit in neurodegenerative disorders involving lipid imbalance and/or compromised blood-brain barrier such as Alzheimer's disease or multiple sclerosis.

Full text of DOI:10.1021/acs.jmedchem.5b00928

Subscriber access provided by SIMON FRASER UNIV BURNABY
Article
Discovery of A-971432, an Orally Bioavailable Selective
Sphingosine-1-Phosphate Receptor 5 (S1P5) Agonists for
the Potential Treatment of Neurodegenerative Disorders
Adrian D Hobson, Christopher M Harris, Elizabeth van der Kam, Sean C Turner, Ayome Abibi, Ana L
Aguirre, Peter Bousquet, Tegeste Kebede, Donald B Konopacki, Gary Gintant, Youngjae Kim, Kelly Larson,
John Maull, Nigel S Moore, Dan Shi, Anurupa Shrestha, Xiubo Tang, Peng Zhang, and Kathy K Sarris
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.5b00928 • Publication Date (Web): 28 Oct 2015
Downloaded from http://pubs.acs.org on November 2, 2015
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 free 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 accessible to all
readers and 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.
Journal of Medicinal Chemistry 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 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of Aꢀ971432, an Orally Bioavailable
Selective Sphingosineꢀ1ꢀPhosphate Receptor 5
(S1P5) Agonists for the Potential Treatment of
Neurodegenerative Disorders
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
Adrian D. Hobson,*a Christopher M. Harris,a Elizabeth L. van der Kam,c Sean C. Turner,c
Ayome Abibi,a,d Ana L. Aguirre,b Peter Bousquet,a Tegest Kebede,a Donald B. Konopacki,a
Gary Gintant,b Youngjae Kim,a Kelly Larson,b John W. Maull,a Nigel S. Moore,a Dan Shi,a
Anurupa Shrestha,b Xiubo Tang,e Peng Zhange and Kathy K. Sarris.b
a AbbVie Bioresearch Center, 381 Plantation Street, Worcester, Massachusetts 01605, United States
b AbbVie, Inc., 1 North Waukegan Road, North Chicago, Illinois 60064, United States
c AbbVie Deutschland GmbH & Co KG, Knollstrasse 50, 67061, Ludwigshafen, Germany
d Now at Cellular Biology, R&D, Firmenich, Inc, Alexandria Center 450 E 29th, New York, NY 10016
e
Shanghai ChemPartner Co. Ltd. Building 10, 998 Halei Road, Zhangjiang HiꢀTech Park, Pudong New Area,
Shanghai, China, 201203
KEYWORDS Selective S1P5 agonist, Neurodegenerative Disorders, Aꢀ971432
ABSTRACT S1P5 is one of 5 receptors for sphingosineꢀ1ꢀphosphate and is highly expressed on
endothelial cells within the bloodꢀbrain barrier, where it maintains barrier integrity in in vitro
models.1 Little more is known about the effects of S1P5 modulation due to the absence of tool
molecules with suitable selectivity and drugꢀlike properties. We recently reported that molecule
Aꢀ9714322 (29 in this paper) is highly efficacious in reversing lipid accumulation and ageꢀrelated
cognitive decline in rats.3 Herein we describe the development of a series of selective S1P5
agonists that led to the identification of compound 29, which is highly selective for S1P5 and has
excellent plasma and CNS exposure after oral dosing in preꢀclinical species. In order to further
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 37
1
2
3
4
5
6
7
8
9
support its suitability for in vivo studies of S1P5 biology, we extensively characterized 29,
including confirmation of its selectivity in pharmacodynamic assays of S1P1 and S1P3 function
in rats. In addition, we found that 29 improves bloodꢀbrain barrier integrity in an in vitro model
and reverses ageꢀrelated cognitive decline in mice. These results suggest that S1P5 agonism is an
innovative approach with potential benefit in neurodegenerative disorders involving lipid
imbalance and/or compromised bloodꢀbrainꢀbarrier, such as Alzheimer’s disease or Multiple
Sclerosis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
INTRODUCTION
The pleiotropic signaling factor sphingosineꢀ1ꢀphosphate is recognized by a family of 5 receptors
called S1P1, S1P2, S1P3, S1P4 and S1P5.4 Synthetic agonists of S1P1 drive lymphocyte
sequestration and immune suppression.5 Most notably, the proꢀdrug FTY720 (Fingolimod)
reduces relapse rate and lesion activity in multiple sclerosis, and it also results in cardiovascular
events.6 In addition to agonizing S1P1, the phosphorylated form of FTY720 activates S1P3, S1P4
and S1P5.5 Many other chemotypes of S1P1 agonists interact with multiple S1P receptors,
especially S1P5.7 S1P5 is most highly expressed in the central nervous system, particularly on
oligodendrocytes and brain endothelium.1,8 Activation of S1P5 on brain endothelial cells
enhances barrier integrity and reduces transendothelial migration of monocytes in vitro.1 There is
indirect evidence that some of the efficacy of Gilenya in models of multiple sclerosis is from a
direct, neuroprotective effect within the CNS.9 Based on the findings above, we hypothesized
that a selective S1P5 agonist may be efficacious for neurodegenerative disorders. If so, the lack
of S1P1ꢀdriven immune suppression and cardiovascular effects would be advantageous. There
are previous reports of selective S1P5 agonists,10 but our profiling indicates these have
insufficient pharmacodynamic properties to establish S1P5 target validation in vivo (see Results
section). Herein we describe the optimization of the first S1P5 agonists with the selectivity and
drugꢀlike properties to interrogate S1P5 biology in vitro and in vivo.
RESULTS AND DISCUSSION
To test the hypothesis that a selective S1P5 agonist would be efficacious for neurodegenerative
disorders it would require a compound with the requisite potency as an S1P5 agonist, selectivity
profile, primarily at the other S1P family receptors and PK properties to enable once daily oral
dosing. However, at the offset of this project there were no such selective S1P5 agonists reported
in the literature and so a medicinal chemistry program was initiated to identify such a compound
to enable POC studies. Cognizant of Sir James Black’s famous comment that the best way to
find a new drug is to start with an existing one, we decided to leverage the experience from our
sphingosineꢀ1 phosphate receptor agonist program. This was a very successful collaboration
with the University of Virginia based on the S1P1 chemical matter discovered by Macdonald et
al at UVA.11 As part of this project we identified12 cyclopentyl amino alcohol compound 1.
This compound is phosphorylated on the hydroxyl of the amino alcohol in vivo affording a
highly potent S1P1 receptor agonist 2. This meant that to enable in vitro SAR in this drug series
the amino alcohols had to be phosphorylated in an additional synthetic step prior to screening.
To facilitate a more rapid screening cascade a replacement for the amino alcohol “head group”
was therefore desired.
ACS Paragon Plus Environment
2

Page 3 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Structures of literature S1P1 agonists
Scientists at Merck13 had elegantly shown that azetidineꢀ3ꢀcarboxylic acid was a suitable direct
agonist surrogate for amino alcoholꢀbased prodrugs such as those reported previously.
Incorporation of the azetidine onto a highꢀthroughput screening hit afforded selective and orally
bioavailable S1P1 receptor agonist 3. Not only would the azetidine carboxylic acid afford
compounds that could be screened directly in our in vitro assays, it was ideal for exploration via
parallel synthesis. It was envisaged that libraries would be constructed from a reductive
amination of the azetidine and substituted benzaldehydes in the presence of polymer supported
sodium cyanoborohydride in the centralized High Throughput Chemistry group at AbbVie.
Scheme 1. Reductive amination of an aldehyde with azetidineꢀ3ꢀcarboxylic acid using
polymer supported NaCNBH3
Prior to library synthesis both a) the robustness of the reductive amination procedure and b) the
tolerance for replacement of the cyclopentyl amino alcohol with azetidine carboxylic acid needed
to be confirmed. From the ongoing S1P1 program data showed that the alkyloxy tail of
compound 2 drove selectivity between S1P1and S1P3. Based on this observation commercially
available benzaldehydes with a para alkyl chain of varying length were sourced through Aldrich
Market Select14 to test the reductive amination step. Aldehydes were selected that had both
carbon and ether group attachment of the chain to the aryl ring. Reductive amination of the
aldehyde with azetidineꢀ3ꢀcarboxylic acid was performed using polymer supported NaCNBH3
with the target compounds purified using massꢀdirected HPLC.15 To confirm the tolerance for
the cyclopentyl amino alcohol replacement with azetidine carboxylic acid, 4ꢀ
(octyloxy)benzaldehyde was included in this study as this would afford the direct crossover
analogue, compound 6.
This approach generated extensive chemical matter, some of which had failed to progress in the
S1P1 project due to a lack of potency on S1P1 and/or selectivity against S1P3 due to its
implication in bradycardia.16 Extensive data mining identified approximately 500 S1P1 project
compounds that had failed these potency and/or selectivity criteria. This set was screened in an
S1P5 radioꢀligand binding assay to identify S1P5ꢀselective agents. This screen, along with
existing S1P1 and S1P3 selectivity data, afforded a great launching point for further optimization
shown in Table 1.
Table 1. SAR and selectivity of compounds 4 ꢀ 10
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 37
1
2
3
4
5
6
7
8
9
S1P1 RLB
IC50 (ꢀM)
S1P3 RLB
IC50 (ꢀM)
S1P5 RLB
IC50 (ꢀM)
S1P5/S1P1
selectivity
Compound
n
R
ꢀ
ꢀ
ꢀ
>30
<0.001
0.018
0.005
0.002
0.048
1.18
>10
0.003
0.040
0.023
0.197
5.29
>10
ꢀ
1
1
2
ꢀ
<0.001
ꢀ
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
7
6
5
4
3
2
1
H
ꢀ
4
H
0.001
0.005
0.009
0.009
0.010
0.003
5
5
H
0.4
5
6
H
7
H
>100
>300
>1000
8
H
3.03
ꢀ
9
CH3
3.46
>10
10
The compounds were screened in assays for S1P1, S1P3 and S1P5 as a representative set of the 5
sphingosine receptors and this identified some striking SAR. Compound 1 clearly demonstrated
the requirement for phosphorylation of the amino alcohol for activity with the compound being
inactive at both S1P1 and S1P3. In stark contrast, phosphorylated compound 2 was shown to be
extremely potent at S1P1, S1P3 and S1P5. Compounds 4 and 5 with decyl and nonyl alkyl chains
respectively were equipotent at both S1P1 and S1P3, with compound 5 also being potent at S1P5.
Compound 6 is the azetidine analogue of compound 1 with an octyl alkyl chain and maintained a
high level of potency at S1P1 and S1P5 but was more than 40 fold selective over S1P3 versus the
other 2 S1P receptors tested. This trend continued with compound 7 with a heptyl alkyl chain.
This compound maintained its potency at both S1P1 and S1P5 but was now >500ꢀfold selective
versus S1P3. Of particular note was compound 8 with a hexyl alkyl chain. This compound
maintained a high level of potency against S1P5 but now >1000 fold selective versus S1P3 and
had introduced a selectivity window of around 130 versus S1P1. Based on the encouraging data
with compound 8, compound 9 with a pentyl alkyl chain was only screened against S1P1 and
S1P5. The compound again maintained a high level of potency against S1P5 but now the
selectivity versus S1P1 was > 300 fold. Compound 10 demonstrated that substitution on the
alkyl chain was tolerated with the compound maintaining a high potency at S1P5 with an
increased selectivity versus S1P1 of >1000. These compounds were determined to be full
agonists in a GTPγS activation assay, using S1P5ꢀtransfected HEK cell membranes. EC50 values
were similar between GTPγS and radioꢀligand binding formats, and so compound optimization
was driven by binding data, with occasional confirmation in the GTPγS assay.
These initial compounds designed around compound 1 had identified a critical structural link to
selectivity amongst the S1P family of receptors with an alkyl chain length of 5 or 6 optimal for
both S1P5 potency and selectivity. Having identified a selectivity pharmacophore for S1P5, this
information was used to design follow up libraries.
As there was no structural information available on the S1P5 receptor to drive compound design
it was planned to use parallel synthesis to generate a pharmacophore of the target. To enable this
compound 6 was divided into 4 regions shown in Figure 2. Libraries would then be designed to
sequentially interrogate these regions using parallel synthesis. Using the SAR from these
ACS Paragon Plus Environment
4

Page 5 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
libraries compounds combining the best structural motifs would then be designed and
synthesized by dedicated synthesis.
6
7
8
O
CO2H
9
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
Tail Region Spacer Aryl ring Head Group
Figure 2. Regions of Compound 6 selected for SAR investigation
Three main criteria were used for selection of the monomers for the libraries. The first was their
commercial availability from Aldrich Market Select. Secondly the choice of monomers was
based on the Topliss Tree17. Monomers were selected to interrogate the effect of both electronics
(halogen, methoxy) and steric constraints (methyl, ethyl) on both S1P5 binding and selectivity
across the S1P family of receptors. The final criteria was the calculated physicochemical
properties of the resulting compounds based on medicinal chemistry guidelines like drugꢀlike
properties,18 Golden Triangle19 and 3/75.20
(a) Drugꢀlike
(b) Golden Triangle
(c) 3/75 Rule
Figure 4. Representative examples from each library design showing (a) excellent
calculated physicochemical properties (b) good projected clearance and oral absorption
and (c) predicted reduced risk of toxicity
For clarity only a single representative of each library is shown in Figure 4. From the radar plot
visualization it is clearly evident that all the compounds have excellent predicted drugꢀlike
properties. The radar plot data is also shown in tabular form below (Table 2). All the
compounds were also predicted to have both good clearance and oral absorption using the
Golden Triangle. In addition the compounds all have calculated PSA and cLogP that suggest a
reduced risk of toxicity based on the 3/75 rule.
Table 2. Calculated physicochemical properties of compounds 10, 29, 31, 43 and 60
Compound cLogP PSA Mol wt HBD
HBA
NAR NRB FSP3
1.6
2.8
50
50
277.4
366.2
1
1
4
4
1
2
7
6
0.56
0.28
10
29
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 37
1
2
3
4
5
6
7
8
9
2.2
3.0
2.5
50
41
50
331.8
363.4
383.3
1
1
1
4
3
4
2
2
2
6
7
7
0.28
0.35
0.32
31
43
60
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 first library was designed to explore the head group region of compound 6. As lead
compound 1 contained the larger cyclopentyl ring, analogues with both pyrrolidine and
piperidine replacements of azetidine were considered. It was also recognized that the larger
piperidine ring afforded multiple opportunities to vary the carboxylic acid group orientation and
analogues exploring this included in the library.
Table 3. SAR and selectivity of compounds 15 ꢀ 21
O
R
S1P5 RLB
IC50 (ꢀM)
Compound
R
0.075
11
12
13
14
15
16
0.026
0.125
1.210
0.016
0.028
Racemic pyrrolidineꢀ3ꢀcarboxylic acid analogue 11 suggested that ring expansion of the
azetidine ring was tolerated while the (R)ꢀisomer 12 revealed important information on the
desired stereochemistry of the carboxylic acid group. This was further exemplified with the
piperidineꢀ3ꢀcarboxylic acid analogues 13 and 14, the (R)ꢀisomer being more than 10 times more
potent than the (S). Compounds incorporating piperidineꢀ4ꢀcarboxylic acid afforded the most
potent analogues. Compound 15 had an IC50 of 16 nM and substituted piperidine 16 only
slightly less potent.
Next a library was designed to expand the SAR of the spacer and tail regions of compound 6.
The library interrogated the steric limits by reducing the length of the spacer region between the
tail and the aryl ring proximal to the azetidineꢀ3ꢀcarboxylic acid. These compounds would
afford more conformationally restricted analogues that might have improved oral
ACS Paragon Plus Environment
6

Page 7 of 37
Journal of Medicinal Chemistry
1
2
bioavailability.21 A series of phenyl and phenoxy benzaldehydes was sourced again though
Aldrich Market Select and reacted with azetidine carboxylic acid.
3
4
5
6
7
Table 4. SAR and selectivity of compounds 17 ꢀ 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
S1P1 RLB
IC50 (ꢀM)
S1P3 RLB S1P5 RLB
S1P5/S1P1
selectivity
Compound
R
IC50 (ꢀM)
IC50 (ꢀM)
3,4ꢀDiClPh
4ꢀMePh
4.41
7.21
0.656
>10
0.023
200
>1000
>200
>24
17
18
19
20
21
22
>10
>10
ꢀ
0.007
4ꢀEtPh
0.003
3ꢀCF3Ph
3,5ꢀDiClPh
4ꢀClPhO
0.409
9.81
>10
>10
ꢀ
0.148
66
0.789
>12
Compounds 17, 18 and 19 with either a 4ꢀsubstituted or 3,4ꢀdisubstituted aryl ring directly
attached to the 1ꢀbenzylazetidineꢀ3ꢀcarboxylic acid were potent at S1P5 with at least a 200ꢀfold
window over S1P1. Surprisingly the potency at S1P5 for compound 20 with a trifluoromethyl
substituent in the 3ꢀposition was greatly reduced. This also resulted in much lower selectivity
against S1P1. Compound 21 was included to vary the substitution pattern around the aryl ring
and while tolerated resulted in a large reduction in potency at S1P5. Introduction of an oxygen
atom linker between the 2 aryl rings was not so promising. Compound 22 based on a biphenyl
ether had greatly reduced potency at S1P5.
The previous library established that both straight alkyl chains, like in compound 6, and more
sterically demanding groups, like aryl in compound 18, were tolerated off the paraꢀposition of 1ꢀ
benzylazetidineꢀ3ꢀcarboxylic acid. As a result a follow up library was planned that combined
these SAR observations. Incorporating the key observation from the compounds in Table 1 with
regard to overall chain length and S1P5 selectivity, it was recognized that the total chain length
must be maintained at 6 to 7 atoms. Consequently a library of benzyl ethers was designed and
synthesized.
Table 5. SAR and selectivity of compounds 23 ꢀ 29
S1P1 RLB
IC50 (ꢀM)
S1P3 RLB
S1P5 RLB
S1P5/S1P1
selectivity
Compound
R
IC50 (ꢀM)
IC50 (ꢀM)
H
>10
>10
ꢀ
0.077
0.035
>100
151
23
24
2ꢀMe
5.26
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 37
1
2
3
4
5
6
7
8
9
3ꢀCl
3ꢀCF3
4ꢀF
1.78
1.06
ꢀ
ꢀ
0.010
0.004
0.009
0.004
0.006
174
248
ꢀ
25
26
27
28
29
ꢀ
ꢀ
2,4ꢀCl
3,4ꢀCl
0.835
0.362
ꢀ
225
60
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
>10
It was pleasing that compound 23, the unsubstituted analogue, was as expected potent at S1P5
while being more than a 100ꢀfold selective against both S1P1 and S1P3. Compounds 24, 25, 26
and 27 having a monoꢀsubstituted aryl ring all had improved potency at S1P5 and increased
selectivity versus S1P1 (for 27 S1P1 data was not generated). Similarly the 2,4ꢀdichloro analogue
28 had both impressive S1P5 potency and selectivity versus S1P1. Compound 29 with a 3,4ꢀ
dicloro substitution motif displayed a very interesting profile. The compound was essentially
inactive at S1P3 while having more than a 50 fold window over S1P1.
Suitable benzaldehydes to examine substitution in both the tail and aryl regions simultaneously
were then considered. However, the commercial availability of such suitably substituted
benzaldehydes was very limited and only a few analogues were secured.
Table 6. SAR and selectivity of compounds 30 ꢀ 34
S1P1 RLB S1P5 RLB
IC50 (ꢀM) IC50 (ꢀM)
S1P5/S1P1
selectivity
Compound
R1
R2
2ꢀMe
2ꢀCl
H
4.17
4.23
0.12
0.22
6.23
0.008
0.006
0.009
0.005
0.115
>500
>900
13
30
31
32
33
34
H
3ꢀOMe
3ꢀNO2
3ꢀOMe
2,4ꢀCl
3,4ꢀCl
2ꢀCl
44
54
Compounds 30, 31, 32, 33 and 34 displayed good to excellent potency at S1P5 with a wide
selectivity window over S1P1 ranging from 10 to >900 fold. However, with the limited
commercial availability of suitable benzaldehydes to enable an extensive parallel synthetic
approach it was difficult to draw any meaningful SAR information from these analogues. To
afford the SAR from substitution of both aryl rings dedicated synthesis of specific examples was
deemed necessary and pursued later in the project (Table 9)
For exploration of the spacer region of compound 6 no suitable benzaldehydes were available
from commercial sources and so dedicated synthesis of the key compounds would need to be
conducted. The first structural modification to be considered in the spacer region was
replacement of the oxygen in the linker between the two aryl rings with carbon. Such a
replacement was known to be tolerated for our S1P1 agonists and so it was considered likely it
ACS Paragon Plus Environment
8

Page 9 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
would also be at least tolerated here. Protection of 4ꢀcyano benzaldehyde 35 with ethylene
glycol afforded the dioxolane 36 which underwent a Grignard reaction with commercially
available benzyl bromide 37 to yield advanced intermediate 38. Subsequent deprotection of the
dioxolane under acidic conditions afforded the free aldehyde 39 which was reacted with
azetidine carboxylic acid under the standard conditions with sodium cyano borohydride.
Hydrogenation of ketone 40 in the presence of palladium on carbon gave the desired compounds
41.
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. Synthetic route toward 1,2ꢀdiphenylethane analogues 42 and 43
Reagents and conditions: (a) ethane-1,2-diol, PTSA, PhMe; (b) Mg, I2, Et2O; (c) 36, Et2O; (d) 10%
HCl, THF, reflux; (e) azetidine-carboxylic acid, NaCNBH3, MeOH; (f) 1% H2SO4/EtOH, Pd/C, H2.
Surprisingly while compound 42 with the unsubstituted aryl ring to the azetidine head group
maintained good potency at S1P5 it was now equipotent at S1P1. This was in contrast to
compound 23 which while having a similar potency at S1P5 displayed a selectivity window
versus S1P1 of >100ꢀfold. Pleasingly introduction of a metaꢀelectron withdrawing group on the
distal aryl ring again established a high selectivity versus S1P1, compound 43 with R = CF3
being over 250ꢀfold selective.
Table 7. SAR and selectivity of compounds 42 and 43
S1P1 RLB S1P5 RLB
IC50 (ꢀM) IC50 (ꢀM)
S1P5/S1P1
selectivity
Compound
R
H
0.002
0.221
0.016
0.001
1
42
43
CF3
260
Having established the synthetic route to compound 40, which contains a ketone in the linker
region, additional ketone containing analogues were pursued. As ketones, particularly aryl
ketones, are not considered drugꢀlike molecules this would provide useful internal data around
this functional group. The key aldehyde intermediates 39 were synthesized in sufficient quantity
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 37
1
2
3
4
5
to enable analogues with both azetidineꢀ3ꢀcarboxylic acid and pyrrolidineꢀ3ꢀcarboxylic acid to
be secured.
6
7
Table 8. SAR and selectivity of compounds 44 ꢀ 51
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
Compound
R1
R2
S1P5 MFB IC50 (ꢀM)
Butyl
0.002
44
Butyl
Pentyl
0.013
0.007
0.005
0.032
0.277
0.008
0.008
45
46
47
48
49
50
51
Pentyl
Ph
Ph
3,4ꢀDiClPh
3,4ꢀDiClPh
For analogues 44 and 46 with the alkyl chain introduction of the carbonyl moiety had little
impact compared to the corresponding analogues with the oxygen in the linker 8 and 7
respectively. Compounds 45 and 47 also demonstrated that increasing the ring size of the amino
acid had little effect on potency. The SAR was more striking with the analogues containing an
aryl ring off the linker. Compound 48 was slightly less potent than the corresponding desꢀ
carbonyl compound 42 whereas compound 49 with the larger pyrrolidineꢀ3ꢀcarboxylic acid was
>15ꢀfold less potent. Varying the size of the ring of the amino acid head group had little impact
on either of the 3,4ꢀdichloro substituted compounds 50 and 51 which both showed similar
potency to the corresponding ether linked compound 32. Such SAR suggests that investigation
of the amino acid head group and other structural modifications would need to be done in
parallel to identify the optimal design.
The expeditious use of parallel synthesis had furnished a large amount of structural information
that drove both S1P5 binding and selectivity across the S1P family of receptors. However, this
information was generated in the 4 discreet regions of compound 6 that had been used to enable
library design. The final set of compounds was designed by combining this SAR on to the
benzyl ether scaffold of compound 23.
ACS Paragon Plus Environment
10

Page 11 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To explore simultaneous substitution on both aryl rings of the benzyl ether scaffold a synthesis
was designed starting from the appropriately substituted benzyl alcohols 52 and 4ꢀcyano phenols
53. These components could be coupled efficiently using a Mitsunobu reaction. The resulting
cyano compounds 54 were reduced to the desired aldehydes 55 using DIBAL. Introduction of
the azetidineꢀ3ꢀcarboxylic acid head group was effected under the standard conditions to afford
the desired compounds 56.
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 3. Synthetic route toward (benzyloxy)benzene analogues 57 ꢀ 63
Reagents and conditions: (a) Ph3P, DIAD, THF; (b) DIBAL, PhMe; (c) azetidineꢀ3ꢀcarboxylic
acid, NaCNBH3, MeOH.
Compound 57 suggested that substitution on the aryl ring proximal to the azetidine carboxylic
acid may drive selectivity over S1P1 while having a slightly negative affect on S1P5 potency.
Disappointingly introduction of halogens to both aryl rings furnished flat SAR. 3,4ꢀ
Dichlorophenyl containing compounds 58 and 59 had demonstrated essentially the same profile
as the desꢀfluoro parent compound 29. In the 3ꢀtrifluoromethylphenyl containing compounds it
was a similar story with fluoroꢀsubstituted compounds 60 and 61 and chloroꢀsubstituted
compounds 62 and 63 showing little differentiation from the desꢀhalo parent compound 26.
While introduction of a halogen had not greatly improved potency or selectivity it was pleasing
to see that its effect on the cLogP of the compounds was not too detrimental. As a result halogen
incorporation was identified as a strategy to block possible sites of metabolism on the proximal
to the amino acid head group should this be required later in development.
Table 9. SAR and selectivity of compounds 26, 29, 57 ꢀ 63
S1P1 MFB S1P5 RLB
IC50 (ꢀM) IC50 (ꢀM)
S1P5/S1P1
selectivity
Compound
R1
R2
cLogP
H
3,4ꢀCl
H
0.362
>10
0.006
0.041
0.005
0.005
60
244
90
2.8
1.6
3.1
3.1
29
57
58
59
3ꢀF
2ꢀF
3ꢀF
3,4ꢀCl
3,4ꢀCl
0.472
0.217
46
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 37
1
2
3
4
5
6
7
8
9
H
3ꢀCF3
3ꢀCF3
3ꢀCF3
3ꢀCF3
3ꢀCF3
1.06
0.68
0.004
0.002
0.002
0.002
0.001
248
340
250
130
160
2.2
2.5
2.5
3.1
3.1
26
60
61
62
63
2ꢀF
3ꢀF
2ꢀCl
3ꢀCl
0.401
0.256
0.109
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
Due to the limited commercial availability of alkyl substituted 4ꢀhydroxy phenols an alternative
synthetic route was required to investigate methyl analogues of compound 23. Mitsunobu
reaction of appropriately substituted benzyl alcohols 52 with 4ꢀbromoꢀ3ꢀmethylphenol 64 gave
key bromo intermediates 65 which were subsequently converted to compounds 66 by treatment
with copper(I) cyanide. Treatment with DIBAL gave aldehydes 67 which were reacted with
azetidineꢀ3ꢀcarboxylic acid under the usual conditions to afford target compounds 68.
Scheme 4. Synthetic route toward substituted 1ꢀ(4ꢀ(benzyloxy)ꢀ2ꢀmethylbenzyl)azetidineꢀ3ꢀ
carboxylic acids 69 and 70
Reagents and conditions: (a) Ph3P, DIAD, THF; (b) CuCN, NMP; (c) DIBAL, PhMe; (d)
azetidineꢀ3ꢀcarboxylic acid, NaCNBH3, MeOH.
Introduction of the methyl substituent resulted in dramatic SAR. While both compounds 69 and
70 were observed to be more potent at S1P1 than their central aryl ring desꢀmethyl analogues 26
and 29 respectively, they were significantly more potent at S1P5. This increased selectivity
versus S1P1 is also apparent when considering compound 30. Compound 30 has an
unsubstituted distal aryl ring and the 2ꢀmethyl substituent on the central aryl ring as in 69 and 70.
SAR from this set of compounds suggests the importance of substitution on the central aryl ring
for driving selectivity while substitution on the distal aryl ring drives potency, at both S1P1 and
S1P5.
Table 10. SAR and selectivity of compounds 26, 29, 30, 69 and 70
S1P1 MFB IC50 S1P5 RLB IC50
(ꢀM) (ꢀM)
S1P5/S1P1
selectivity
Compound
R1
R2
ACS Paragon Plus Environment
12

Page 13 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3ꢀCF3
3,4ꢀCl
H
H
1.06
0.362
4.17
0.004
0.006
0.008
0.001
0.001
248
60
26
29
30
69
70
H
Me
Me
Me
>500
400
150
3ꢀCF3
3,4ꢀCl
0.333
0.166
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 complete the SAR around the spacer region of compound 6 substitution from the benzylic
carbon was explored. Introduction of a methyl was chosen as this would greatly affect the
conformation of the resulting compound and hence alter the trajectory of the tail aryl ring
relative to the head group. The synthesis was initially conducted with 1ꢀphenylethanol 71 using
the typical 3ꢀstep procedure. A Mitsunobu reaction with 4ꢀcyanophenol 72 afforded compound
73 which was then treated with DIBAL to furnish aldehyde 74. Subsequent reaction with
azetidineꢀ3ꢀcarboxylic acid gave the desired racemic compound 75. The enantiomerically pure
(R)ꢀisomer 76 was secured using the same synthetic sequence, but using (S)ꢀ1ꢀphenylethanol.
Scheme 5. Synthetic route toward substituted 1ꢀ(4ꢀ(benzyloxy)ꢀ2ꢀmethylbenzyl)azetidineꢀ3ꢀ
carboxylic acids 75 and 76
Reagents and conditions: Ph3P, DIAD, THF; (b) DIBAL, PhMe; (c) azetidineꢀ3ꢀcarboxylic acid,
NaCNBH3, MeOH.
Unfortunately substitution off the benzylic carbon was at best tolerated and is likely detrimental.
Both the racemate 75 and pure enantiomer 76 had a reduced selectivity window versus S1P1 and
neither was more potent than the H analogue 23.
Table 11. SAR and selectivity of compounds 23, 75 and 76
S1P1 RLB S1P5 RLB
S1P5/S1P1
selectivity
Compound
R
IC50 (ꢀM)
IC50 (ꢀM)
H
>10
0.077
>100
>39
23
75
Racemic Me
>10
0.255
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 37
1
2
3
4
(R)ꢀMe
3.87
0.108
35
76
5
6
7
8
The project goal was to identify a compound with the requisite potency as an S1P5 agonist,
selectivity profile, primarily at the other S1P family receptors and PK properties to enable once
daily oral dosing. The expeditious use of parallel synthesis in combination with the dedicated
synthesis of key singletons had facilitated extensive in vitro profiling which enabled the
construction of a comprehensive pharmacophore around both S1P5 potency and selectivity S1P
family of receptors. To permit the additional in vivo profiling larger compound inventories
would be necessary. Due to project resource constraints it was only possible to scale up a limited
number of analogues and so 5 compounds were selected. The compound selection for additional
profiling was based on the in vitro data using potency at S1P5 and selectivity versus S1P1 with
examples selected to ensure the broadest coverage of chemical matter. The compounds selected
for scale up are shown in Figure 3: compound 10 with an alkyloxy chain, compound 29 with a
substituted distal aryl ring, compound 31 with a substituted central aryl ring, compound 43 with
an ethyl linker and compound 60 with substitution on both aryl rings. It should be noted that in
addition to this series with the azetidine carboxylic acid head group, three additional compound
series of S1P5 selective antagonists were identified and are currently being developed.22
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. Structures of 5 compounds selected for scale up to enable additional profiling
The 5 compounds were profiled in a highꢀthroughput ADME panel to assess solubility, plasma
protein binding, permeability and inhibition of CYP enzymes. All the compounds had good
solubility and free fraction. Of note was compound 10 with the branched alkyl chain, that
exhibited very low level of plasma protein binding. Despite the incorporation of the azetidineꢀ3ꢀ
carboxylic acid head group, the compounds displayed reasonable permeability as assessed by the
PAMPA assay. Pleasingly none of the compounds were identified as inhibitors of the 4 most
common CYP enzymes. All the compounds were then tested in the micronucleus assay and
found to be nonꢀclastogenic.
Table 12. Highꢀthroughput ADME data for compounds 10, 29, 31, 43 and 60
Inhibition (ꢀM)
Solubility
(ꢀM)
PPB (%
bound)
PAMPA
(cmsꢀ1)
Compound
Cyp1A2
>20
Cyp2C9
>20
Cyp2D6 Cyp3A4
10
29
46
37
3.2
>20
>20
70
95
2.7
>20
>20
>20
>20
ACS Paragon Plus Environment
14

Page 15 of 37
Journal of Medicinal Chemistry
1
2
3
31
43
60
61
73
70
86
88
87
4.2
5.3
9.4
>20
>20
16.7
>20
>20
>20
>20
>20
>20
>20
>20
>20
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
Screening PK was then conducted with the compounds being dosed at 10 mpk po in female
Balb/C mice (29 was tested in CD1 mice). As predicted by the physicochemical properties, all
the compounds displayed excellent PK characteristics. The compounds exhibited good t1/2
ranging from 3.3 to 5.7 hours, with large maximum concentrations achieved in an hour or less.
Most notable was the very large AUC observed with heptanone analogue 43 of 127,000
ng*h/mL.
Table 13. Mouse PK for compounds 10, 29, 31, 43 and 60
t1/2 (hr)
Cmax (ng/mL)
Tmax (ng/mL)
Compound
AUC (ng*h/mL)
10
29*
31
43
60
3.3
5.7
4.1
4.9
4.6
11,800
2,500
4,230
17,900
6,480
0.4
-
17,300
35,000
16,800
127,000
1.3
0.83
1.0
37,700
* 29 was tested in CD1 mice
The comprehensive data package generated for compounds 10, 29, 31, 43 and 60 showed them
all to be excellent candidates for in vivo studies. All the compounds were potent S1P5 agonists
and selective across the S1P family of receptors. No issues had been identified from the HTꢀ
ADME screen and pleasingly all the compounds had excellent solubility. The PK profiles of the
compounds were similarly all suitable for advancing to the planned in vivo studies. From these 5
compounds it was decided to advance compound 29 for largeꢀscale synthesis to enable further
characterization. The selection of compound 29 was based on it having the longest t1/2 which it
was thought would be the most appropriate for future studies. It should be noted however that all
5 compounds have profiles suitable for advancing in vivo.
While 4ꢀ((3,4ꢀdichlorobenzyl)oxy)benzaldehyde 79 was commercially available due to the high
cost it was synthesized from more readily available starting materials. A mixture of 3,4ꢀ
dichlorobenzyl chloride 77, 4ꢀhydroxybenzaldehyde 78, potassium carbonate and acetone was
refluxed overnight to afford 4ꢀ(3, 4ꢀdichlorobenzyloxy)benzaldehyde 79 in 79% yield after
filtration and washing. Reaction with azetidineꢀ3ꢀcarboxylic acid proceeded smoothly in the
presence of sodium cyanoborohydride in methanol at room temperature. The product 29 was
again collected by filtration in 60% yield.2
Scheme 6. Large Scale Synthetic Route for 1ꢀ(4ꢀ((3,4ꢀdichlorobenzyl)oxy)benzyl)azetidineꢀ
3ꢀcarboxylic acid 29
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 37
1
2
3
4
5
6
Cl
Cl
Cl
O
b
Cl
Cl
a
O
Cl
O
H
+
CO2H
Cl
H
N
HO
7
O
77
78
79
29
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
Reagents and conditions: (a) K2CO3, acetone, reflux; (b) Azetidineꢀ3ꢀcarboxylic acid,
NaCNBH3, MeOH.
Potency and selectivity SAR was developed with radioꢀligand binding assays. To confirm S1P5
agonism, 29 was profiled in two in vitro functional assays. FTY720ꢀphosphate was used to
define 100% agonism, as it gave more consistent results than the lipid S1P. In a GTPγS assay
using membranes from S1P5ꢀtransfected HEK cells, 29 induced full agonism and had an EC50 =
5.7 nM (Figure 5a). In a forskolinꢀinduced cAMP inhibition assay in S1P5ꢀtransfected CHO
cells, 29 was also a full agonist, and had an EC50 = 4.1 nM (Figure 5b). By comparison,
FTY720ꢀphosphate had an EC50 of 3.4 and 11 nM in the GTPγS and cAMP assays, respectively.
Figure 5. In vitro functional S1P5 assays. (a) GTPγS binding in membranes from S1P5ꢀ
transfected HEK cells. (b) Inhibition of forskolinꢀinduced cAMP in S1P5ꢀtransfected CHO
cells.
Compound 29 was profiled in receptor binding assays against the other S1P receptors and found
to have no measurable binding up to 10 ꢀM on S1P2, S1P3, and S1P4 (data not shown). At S1P1,
29 has weak potency (0.36 ꢀM). To address agonism versus antagonism at S1P1, we first used
an S1P1 GTPγS assay to demonstrate a complete lack of S1P1 activation up to 10 ꢀM 29 (data
not shown). We then determined 29 to be a competitive antagonist at S1P1 with a KB = 0.4 ꢀM
in the same system (Supporting information). Lastly, we confirmed a lack of S1P1 activity by
dosing 29 in SpragueꢀDawley rats and monitoring lymphopenia. We found no change in
lymphocyte counts at 5 hours after a single oral dose of 3, 10, 30, or 100 mg/kg (Figure 6).
Plasma exposure was doseꢀproportional and attained 15,000 ng/ml at the highest dose. These
data clearly show 29 is devoid of S1P1 agonism up to the highest doses tested.
ACS Paragon Plus Environment
16

Page 17 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Selectivity of 29 against S1P1 is confirmed by a lack of lymphopenia. Spragueꢀ
Dawley rats were administered 10, 30, or 100 mg/kg of 29, or vehicle (0.5% HPMC/0.02%
Tween 80) by oral gavage. Five hours later, blood was collected via cardiac puncture
under isoflurane anesthesia. Mean number of lymphocytes +/ꢀ sem are reported (n = 4).
Plasma exposures were determined to be 1,800, 5,000 and 15,000 ng/mL respectively
In rodents, S1P3 agonism results in bradycardia,16 although in humans these effects are likely
mediated by S1P1.23 To confirm selectivity against S1P3 in vivo, we evaluated 29 in
cardiovascular safety studies. In anesthetized SpragueꢀDawley rats, 29 induced no significant
change in heart rate during sequential 30 minute IV infusions at 3, 10, and 30 mg/kg (Figure 7),
achieving plasma concentrations of 1,840, 8,660 and 20,300 ng/mL respectively. Overall, the 10
mg/kg dose had minimal hemodynamic effects, and the 30 mg/kg dose produced a 16% increase
in mean arterial pressure (not statistically significant) and a 10% decrease in cardiac contractility.
These studies confirm in vivo that 29 has no activity at S1P3, and when interpreted in the context
of efficacy studies to be disclosed separately, also demonstrates a favourable safety margin.
While the lack of lymphopenia and bradycardia responses from 29 is implied by its in vitro
profile, these data demonstrate for the first time a lack of any S1P1–related activity in an S1P5
agonist and support the use of 29 in differentiating S1P1 versus S1P5 biology in vivo.
Figure 7. 29 does not cause bradycardia in anesthetized rats. 29 was administered by 3
sequential 30 minute IV infusions at 3, 10 and 30 mg/kg. No effects on heart rate were
observed through the highest mean exposure of 20.3 ug/mL. Similarly, no statistically
significant effects on mean arterial pressure were observed. Cardiac contractility (dP/dt50)
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 37
1
2
3
4
5
was decreased 10% (change from baseline) during the end of the third infusion period
(P<0.05)
There are two previous reports of S1P5ꢀselective compounds.10 Our in vitro profiling of these
molecules showed significantly weaker potency than the agonists we describe. In addition, we
were unable to formulate these molecules in a manner resulting in any measurable in vivo
exposure.
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
In order to confirm compound 29 was also selective outside the S1P receptor family it was
profiled in 2 large screening panels. 29 was found to have an IC50 >10 ꢀM at all 129 kinases of
an AbbVie internal panel (Supporting information). Of the 79 receptors tested on the CEREP24
panel, 29 only bound to 2 members at a threshold of >50%, 5ꢀHT2a (75%) and DA transporter
(63%) (Supporting information).
Compound 29 was screened in the Ames test and found to be nonꢀclastogenic. It was negative in
the Alarm NMR assay25 suggesting little likelihood of thiol reactivity, for example with
glucuronide. It was also profiled in a dofetilide binding assay and had an IC50 of 77 ꢀM.
Pharmacokinetic analyses (Table 14) of 29 in mouse, rat, dog and cynomolgus monkey indicates
high oral bioꢀavailability, high exposure, low clearance, a long halfꢀlife and at least where tested
in rodents, good brain exposure. Despite moderate to high plasma protein binding, total brain
exposure of as little as 20 ng/g brain tissue results in full efficacy, in studies to be discussed in
subsequent publications.26 These data indicate 29 can achieve sufficient exposure to saturate
S1P5 in peripheral tissues as well as the central nervous system.
Table 14. Pharmacokinetics of 29 in multiple species
IV
PO
Cmax
Protein
Binding
(%)
Dose
(mg/kg)
Sample
Analyzed
Species
Tma
x
(hr)
AUC
(ng*h/m
L)
t1/2
(hr)
AUC
(ng*h/mL)
CL
(L/h/kg)
Vss
(L/kg)
t1/2
(hr)
F
(%)
(ng/mL)
BALB/C
Mouse
2
2
Plasma
Brain
93
7.6
9.8
8,500
0.24
n.d.
1.9
7.4
10
2.0
300
4,800
1,600
57
3,200
(Cmax=133
ng/nL
BALB/C
Mouse
n.d.
n.d.
2ꢀ24
43
56
SD Rat
SD Rat
1
2
Plasma
Brain
93
9.0
6,400
n.d.
0.16
n.d
1.3
14
15
4.3
8
400
120
8,700
3,100
>100
n.d.
99.5
n.d.
n.d.
Beagle
Dog
1
1
Plasma
Plasma
96
97
9.3
3.5
12,000
6,400
0.09
0.16
1.2
10
1.5
1.7
690
650
11,000
5,500
92
86
Cyno
Monkey
0.82
6.7
n.d. = not determined
Compound 29 underwent a 14 day SpragueꢀDawley rat tolerability study at 10 and 100 mg/kg,
po, qd. One of three females in the 100 mg/kg group was euthanized on Day 8 due to 22%
ACS Paragon Plus Environment
18

Page 19 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
weight loss, the other females had no weight changes at either dose. All other animals survived
to the scheduled necropsy with no adverse clinical signs. Body weights of males on Day 14 were
10% less than vehicle controls at both doses. There were increased liver weights in high dose
groups (20% for male, 30% for female) but no corresponding increase in serum ALT. There
were no remarkable changes in weights of spleen or thymus, in clinical chemistry, or in
hematology, including no changes in lymphocytes. The exposure AUC was 43ꢀ54 ꢀg*hr/mL at
10 mg/kg, and 670ꢀ820 ꢀg*hr/mL at 100 mg/kg, and was the same at day 2 and at day 14.
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
Due to the paucity of pharmacology data around S1P5, we profiled 29 in a panel of 41
multiplexed in vivo screening assays at Melior Discovery.27 This panel was designed to
efficiently query many biological systems, and should be interpreted as hypothesisꢀgenerating,
not as ‘goldꢀstandard’ versions of these assays. 29 was dosed in CD1 mice at 0.3, 3 and 30
mg/kg, qd, po. 29 roughly doubled food intake over a 4 day observation period, but did not
result in any change in body weight, compared to vehicle (data not shown). Effects were
smallest at the midꢀdose. Additional hits that did not achieve significance and/or a dose response
include a) antiꢀconvulsant activity in response to pentylenetetrazole challenge, (b) a decrease in
MCPꢀ1 following pulmonary antigen challenge, and (c) an increase in urination frequency
(Supporting information). No additional hits were found in the other 36 endpoints evaluated.
S1P5 agonism does not appear to have pleiotropic effects.
S1P5 is highly expressed in brain endothelium, and its activation has been shown to enhance
barrier function in an in vitro model of bloodꢀbrain barrier integrity. We confirmed this role for
S1P5 by showing that 29 increases electrical resistance of hCMEC/D3 cells in vitro at 1 ꢀM
(Figure 8). This finding confirms previously published data on the impact of S1P5 agonism on
endothelial barrier integrity.
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 37
1
2
3
4
5
6
Figure 8. 29 impacts transendothelial electrical resistance as assessed through ECIS in
hCMEC/D3 cells; DMSO (n = 4) or 1 ꢀM 29 (n= 2) was added to the cell layer and
resistance (ohm.cm2) was assessed over 75h
7
As S1P5 is highly expressed in the brain, we explored the impact of S1P5 activation on a
functional readꢀout, namely reversal of ageꢀrelated cognitive deficits in male C57BL6J mice.
After 21 days of treatment with 29, cognitive function was assessed by spontaneous alternation
in the T maze, 225 min after the final dose (estimated Tmax of the compound). Compound 29
reversed the cognitive deficits with an exposure of 41.6 ± 2.5 ng/mL (mean + SEM) and 28.3 ±
1.9 ng/g (mean + SEM)(Figure 9). Further experiments confirmed the proꢀcognitive impact of
29 in a doseꢀdependent manner and that 0.1 mg/kg is the lowest effective dose3.
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 9. 29 reverses ageꢀrelated cognitive decline as seen in the spontaneous alternation
assay in the Tꢀmaze in aged male C57BL6J mice after subꢀacute treatment. 29 (0.1 mg/kg)
was administered for 21 days, qd. Testing was performed 225 minutes after the final dose.
Oneꢀway ANOVA: ** p < 0.05 in comparison to young vehicleꢀtreated animals, ^ p < 0.05
in comparison to aged vehicleꢀtreated animals
CONCLUSIONS
Azetineꢀ3ꢀcarboxylic acid was recognized as a bioisostere for the synthetically challenging
phosphorylated cyclopentyl amino alcohol. This enabled the application of parallel synthesis at
AbbVie’s centralized High Throughput Chemistry group to rapidly secure the desired
compounds. Diligent data mining of these compounds lead to the discovery of the first S1P5
agonists that are clearly selective within the S1P receptor family. These leads were optimized
further using parallel synthesis and dedicated synthesis affording a comprehensive
pharmacophore for S1P5 potency and selectivity. Multiple leads with excellent oral exposure
were identified with Aꢀ971432 (29) selected for extensive evaluation. It was confirmed with 29
that the in vitro selectivity against S1P1 and S1P3 translated to a lack of lymphopenia and
bradycardia, respectively, in rats. 29 was then shown to have highly favorable properties in both
kinase and receptor liability panels, Cyp inhibition assays, AMES test, micronucleus test, rat
cardiovascular safety, 14ꢀday rat tolerability and the Melior Discovery panel of 41 in vivo
endpoints. Finally it was demonstrated that 29 enhanced brain endothelial integrity, and
previously presented data showed an impact on ageꢀrelated cognitive decline and lipid
accumulation in the central nervous system.3 We will show in subsequent publications that 29 is
highly efficacious in preꢀclinical models of neurodegenerative diseases.26 Compound 29 is
available through an MTA for research purposes.
ACS Paragon Plus Environment
20

Page 21 of 37
Journal of Medicinal Chemistry
1
2
3
4
EXPERIMENTAL SECTION
5
6
7
8
The general synthetic methods used in each General Procedure follow and include an illustration
of a compound that was synthesized using the designated General Procedure. Compounds of the
present invention were synthesized and their activity assayed as described below. Unless
otherwise stated, reagents were purchased from Sigma Aldrich, Acros, Alfa Aesar or the Sigma
Aldrich Custom Packaged Reagent service. Reagent/reactant names given are as named on the
commercial bottle or as generated by IUPAC conventions or CambridgeSoft® ChemDraw Ultra
9.0.7. Compound names are generated by IUPAC conventions or CambridgeSoft® ChemDraw
Ultra 9.0.7.
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
Analytical data is included within the procedures below, in the illustrations of the general
procedures, or in the tables of examples. Unless otherwise stated, all 1H and 13C NMR data were
collected on a Varian Mercury Plus 400 MHz or a Bruker AVIII 300 MHz or a Bruker 500 MHz
instrument; chemical shifts are quoted in parts per million (ppm). HPLC analytical data are
either detailed within the experimental or referenced to the table of LCMS and HPLC conditions,
using the lower case letter in Table 16.
Table 15. List of HPLC methods
Method
HPLC Conditions
Analytical LCMS was performed on a Waters ZMD mass spectrometer and Alliance
HPLC system running MassLynx 3.4 and Openlynx 3.4 software. The ZMD mass
spectrometer was operated under positive APCI ionization conditions. The HPLC
system comprised a Waters 2795 autosampler sampling from 96ꢀwell plates, a
Waters 996 diodeꢀarray detector and Sedere Sedexꢀ75 evaporative light scattering
detector. The column used was a Phenomenex Luna CombiꢀHTS C8(2) 5ꢁm 100Å
(2.1mm × 30mm). A gradient of 10ꢀ100% acetonitrile (A) and 0.1% trifluoroacetic
acid in water (B) was used, at a flow rate of 2.0mL/min (0ꢀ0.1 min 10% A, 0.1ꢀ2.6
min 10ꢀ100% A, 2.6ꢀ2.9 min 100% A, 2.9ꢀ3.0 min 100ꢀ10% A. 0.5min postꢀrun
delay.
a
Analytical LCMS was performed on either a Waters LCT Premier mass
spectrometer and 1525 Binary HPLC Pump running MassLynx 4.0 and OpenLynx
4.0 software or a Waters Quattro Ultima mass spectrometer and Agilent 1100 HPLC
system controlled by MassLynx 4.0 and OpenLynx 4.0 software. The gradient was
5ꢀ60% B in 1.5 min then 60ꢀ95% B to 2.5 min with a hold at 95% B for 1.2 min (1.3
mL/min flow rate). Mobile phase A was 10mM ammonium acetate, mobile phase B
was HPLC grade acetonitrile. The column used for the chromatography is a 4.6x50
mm MACꢀMOD Halo C8 column (2.7 µm particles). Detection methods are DAD
and ELSD detection as well as positive/negative electrospray ionization.)
b
c
Analytical LCMS was performed on Agilent 1200 HPLC/6100 SQ System. Mobile
Phase: A: Water (0.05 % TFA) B: Acetonitrile (0.05 % TFA); Gradient Phase: 5 % ꢀ
95 % in 1.3 min; Flow rate: 1.6 mL/min; Column: XBridge, 2.5 min; Oven Temp.
50 °C.
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 37
1
2
3
4
5
6
7
8
9
Analytical LCMS was performed on a Waters ZMD mass spectrometer and Alliance
HPLC system running MassLynx 3.4 and Openlynx 3.4 software. The ZMD mass
spectrometer was operated under positive APCI ionization conditions. The HPLC
system comprised a Waters 2795 autosampler sampling from 96ꢀwell plates, a
Waters 996 diodeꢀarray detector and Sedere Sedexꢀ75 evaporative light scattering
detector. The gradient was 5ꢀ60% B in 1.5 min then 60ꢀ95% B to 2.5 min with a
hold at 95% B for 1.2 min (1.3 mL/min flow rate). Mobile phase A was 10mM
ammonium acetate, mobile phase B was HPLC grade acetonitrile. The column used
for the chromatography is a 4.6x50 mm MACꢀMOD Halo C18 column (2.7 µm
particles). Detection methods are diode array (DAD) and evaporative light
scattering (ELSD) detection as well as positive/negative electrospray ionization.
d
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
1ꢀ(4ꢀ(Hexyloxy)benzyl)azetidineꢀ3ꢀcarboxylic
acid
(8).
A
mixture
of
4ꢀ
(hexyloxy)benzaldehyde (1950 mg, 9.45 mmol), methanol (20 mL) and AcOH (6 drops) was
stirred at ambient temperature for 3 days. The solvent was removed under reduced pressure and
the residue dissolved in Methanol (20 mL) and Methanol (20 mL). The resulting solution was
divided in to 12 X 2 mL fractions and subjected to purification molecular ion directed prepꢀ
HPLC. The combined fractions were evaporated to dryness and dried in vacuo at 60 ℃ for 4
hours. This afforded 1ꢀ(4ꢀ(hexyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (683 mg, 2.34 mmol,
24.8 % yield) as an offꢀwhite solid. LCMS (Table 16, Method d). Rt = 1.96 min.; MS m/z:
1
292.16 (M+H)+. Purity 100%. H NMR (400 MHz, DMSO) δ ppm 6.85ꢀ6.80 (m, 2H), 7.17ꢀ
7.11 (m, 2H), 3.91 (t, J = 6.52 Hz, 2H), 3.43 (s, 2H), 3.29 (t, J = 7.61 Hz, 2H), 3.10 (t, J = 7.20
Hz, 2H), 2.99 (t, J = 7.34 Hz, 1H), 1.75ꢀ1.62 (m, 2H), 1.47ꢀ1.23 (m, 6H), 0.87 (ddd, J = 7.14,
3.25, 1.55 Hz, 3H). HRMS: Found 291.1835 C17H25NO3 requires 291.1834.
Compound 4 ꢀ 7 and 9 ꢀ 35 were synthesized in a similar manner using the appropriate
commercially available aldehyde and azetidineꢀ3ꢀcarboxylic acid unless otherwise stated
following this general procedure. In a 20 mL vial a solution of the aldehyde monomer (1.2eq)
dissolved in dichloromethane (1.1 mL) was added, followed by the addition of azetidineꢀ3ꢀ
carboxylic acid (29mg, 1eq.) dissolved in dichloromethane (0.4 mL), followed by acetic acid
(3eq.) dissolved in dichloromethane(0.4mL), followed by MPꢀCyanoborohydride resin(3eq.)
The mixture was shaken at room temperature for 5 hours.. The reaction was checked by LCMS
and concentrated to dryness. The residue was dissolved in 1:1 DMSO:MeOH and purified by
preparative HPLC on a Phenomenex Luna C8(2) 5 um 100Å AXIA column (30mm × 75mm). A
gradient of acetonitrile (A) and 0.1% trifluoroacetic acid in water (B) was used, at a flow rate of
50mL/min (0ꢀ0.5 min 10% A, 0.5ꢀ6.0 min linear gradient 10ꢀ100% A, 6.0ꢀ7.0 min 100% A, 7.0ꢀ
8.0 min linear gradient 100ꢀ10% A). Samples were injected in 1.5mL DMSO:MeOH (1:1). An
Agilent 1100 Series Purification system was used, consisting of the following modules: Agilent
1100 Series LCMSD SL mass spectrometer with APIꢀelectrospray source; two Agilent 1100
Series preparative pumps; Agilent 1100 Series isocratic pump; Agilent 1100 Series diode array
detector with preparative (0.3mm) flow cell; Agilent activeꢀsplitter, IFCꢀPAL fraction collector /
autosampler. The makeꢀup pump for the mass spectrometer used 3:1 methanol:water with 0.1%
formic acid at a flow rate of 1 mL/min. Fraction collection was automatically triggered when the
extracted ion chromatogram for the target mass exceeded the threshold specified in the method.
The system was controlled using Agilent Chemstation (Rev B.10.03), Agilent A2Prep, and Leap
FractPal software, with custom Chemstation macros for data export.
characterized by MS and LCMS (Table 1, Method a).
Products were
ACS Paragon Plus Environment
22

Page 23 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1ꢀ(4ꢀ(Decyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (4).
Synthesized from aldehyde 4ꢀ
(decyloxy)benzaldehyde. LCMS (Table 16, Method d) Rt = 2.36 min.; MS m/z: 348.2. (M+H)+.
1
Purity 100%. H NMR (400 MHz, Pyridine) δ ppm 7.35 (d, J = 8.65 Hz, 2H), 7.03ꢀ6.98 (m, 2H),
3.89 (t, J = 6.49 Hz, 3H), 3.64ꢀ3.42 (m, 7H), 1.69 (dd, J = 8.29, 6.78 Hz, 3H), 1.38 (s, 3H), 1.29ꢀ
1.11 (m, 14H), 0.82 (dd, J = 7.79, 5.99 Hz, 4H). HRMS: Found 347.2457 C21H33NO3 requires
347.2460.
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
1ꢀ(4ꢀ(Nonyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (5).
Synthesized from aldehyde 4ꢀ
(nonyloxy)benzaldehyde. LCMS (Table 16, Method d). Rt = 2.48 min.; MS m/z: 334.3
1
(M+H)+. Purity 99%. H NMR (400 MHz, CDCl3) δ 7.36 – 7.30 (m, 2H), 6.91 – 6.84 (m, 2H),
4.16 (t, J = 8.0 Hz, 2H), 4.11 (s, 2H), 3.99 (t, J = 9.6 Hz, 2H), 3.93 (t, J = 6.5 Hz, 2H), 3.44 –
3.31 (m, 1H), 2.05 (s, 3H), 1.81 – 1.69 (m, 2H), 1.41 (d, J = 7.1 Hz, 2H), 1.33 – 1.19 (m, 8H),
0.92 – 0.83 (m, 3H). HRMS: Found 333.2304 C20H31NO3 requires 333.2304.
1ꢀ(4ꢀ(Octyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (6).
Synthesized from aldehyde 4ꢀ
(octyloxy)benzaldehyde. LCMS (Table 16, Method d) Rt = 2.00 min.; MS m/z: 320.2 (M+H)+.
1
Purity 96%. H NMR (400 MHz, CDCl3) δ ppm 7.36 (d, J = 8.65 Hz, 2H), 6.89 (d, J = 8.68 Hz,
2H), 3.46ꢀ3.26 (m, 1H), 1.78 (dd, J = 14.69, 6.73 Hz, 2H), 1.55ꢀ1.41 (m, 2H), 1.34 (ddd, J =
14.64, 10.47, 2.94 Hz, 8H), 0.91 (t, J = 6.92 Hz, 3H), 4.11 (s, 4H), 4.07ꢀ3.87 (m, 4H). HRMS:
Found 319.2147 C19H29NO3 requires 319.2147.
1ꢀ(4ꢀ(Heptyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (7). Synthesized from aldehyde 4ꢀ
(heptyloxy)benzaldehyde. LCMS (Table 16, Method d). Rt = 2.07 min.; MS m/z: 306.3
1
(M+H)+. Purity 100%. H NMR (400 MHz, CDCl3) δ 7.37 – 7.31 (m, 2H), 6.91 – 6.84 (m, 2H),
4.17 (d, J = 7.7 Hz, 2H), 4.10 (s, 2H), 4.02 – 3.88 (m, 4H), 2.07 (s, 2H), 1.77 (dt, J = 14.5, 6.7
Hz, 2H), 1.43 (td, J = 9.7, 8.6, 4.5 Hz, 2H), 1.38 – 1.23 (m, 6H), 0.95 – 0.84 (m, 3H). HRMS:
Found 305.1991 C18H27NO3 requires 305.1991.
1ꢀ(4ꢀ(Pentyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (9). Synthesized from aldehyde 4ꢀ
(pentyloxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.36 min.; MS m/z: 278.1 (M+H)+.
Purity 100%. HRMS: Found 277.1678 C16H23NO3 requires 277.1678.
1ꢀ(4ꢀ(Isopentyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (10). Synthesized from aldehyde 4ꢀ
(isopentyloxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.22 min.; MS m/z: 278
(M+H)+. Purity 100%. HRMS: Found 277.1698 C16H23NO3 requires 277.1678.
1ꢀ(4ꢀ(Hexyloxy)benzyl)pyrrolidineꢀ3ꢀcarboxylic acid (11). Synthesized from aldehyde 4ꢀ
(hexyloxy)benzaldehyde and pyrrolidineꢀ3ꢀcarboxylic acid. LCMS (Table 16, Method a) Rt =
1.35 min.; MS m/z: 306.4 (M+H)+.
(R)ꢀ1ꢀ(4ꢀ(Hexyloxy)benzyl)pyrrolidineꢀ3ꢀcarboxylic acid (12).
Synthesized from 4ꢀ
(hexyloxy)benzaldehyde and (R)ꢀpyrrolidineꢀ3ꢀcarboxylic acid. LCMS (Table 16, Method a) Rt
= 1.35 min.; MS m/z: 306.4 (M+H)+.
(R)ꢀ1ꢀ(4ꢀ(Hexyloxy)benzyl)piperidineꢀ3ꢀcarboxylic acid (13). Synthesized from aldehyde 4ꢀ
(hexyloxy)benzaldehyde and (R)ꢀpiperidineꢀ3ꢀcarboxylic acid. LCMS (Table 16, Method a) Rt
= 1.37 min.; MS m/z: 320.5 (M+H)+.
(S)ꢀ1ꢀ(4ꢀ(Hexyloxy)benzyl)piperidineꢀ3ꢀcarboxylic acid (14). Synthesized from aldehyde 4ꢀ
(hexyloxy)benzaldehyde and (S)ꢀpiperidineꢀ3ꢀcarboxylic acid. LCMS (Table 16, Method a) Rt =
1.36 min.; MS m/z: 320.4 (M+H)+.
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 37
1
2
3
4
5
6
1ꢀ(4ꢀ(Hexyloxy)benzyl)piperidineꢀ4ꢀcarboxylic acid (15). Synthesized from aldehyde 4ꢀ
(hexyloxy)benzaldehyde and piperidineꢀ4ꢀcarboxylic acid. LCMS (Table 16, Method a) Rt =
1.35 min.; MS m/z: 320.4 (M+H)+.
7
8
9
1ꢀ(4ꢀ(Hexyloxy)benzyl)ꢀ3ꢀmethylpiperidineꢀ4ꢀcarboxylic acid (16).
Synthesized from
aldehyde 4ꢀ(hexyloxy)benzaldehyde 3ꢀmethylpiperidineꢀ4ꢀcarboxylic acid. LCMS (Table 16,
Method a) Rt = 1.40 min.; MS m/z: 334.4 (M+H)+.
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
1ꢀ((3',4'ꢀDichloroꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀyl)methyl)azetidineꢀ3ꢀcarboxylic acid (17). Synthesized
from aldehyde 3',4'ꢀdichloroꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀcarbaldehyde. LCMS (Table 16, Method a) Rt =
1.45 min.; MS m/z: 336 (M+H)+.
1ꢀ((4'ꢀMethylꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀyl)methyl)azetidineꢀ3ꢀcarboxylic acid (18). Synthesized from
aldehyde 4'ꢀmethylꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀcarbaldehyde. LCMS (Table 16, Method a) Rt = 1.35 min.;
MS m/z: 282 (M+H)+.
1ꢀ((4'ꢀEthylꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀyl)methyl)azetidineꢀ3ꢀcarboxylic acid (19). Synthesized from
aldehyde 4'ꢀmethylꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀcarbaldehyde. LCMS (Table 16, Method a) Rt = 1.34 min.;
MS m/z: 296 (M+H)+.
1ꢀ((3'ꢀ(Trifluoromethyl)ꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀyl)methyl)azetidineꢀ3ꢀcarboxylic
acid
(20).
Synthesized from aldehyde 3'ꢀ(trifluoromethyl)ꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀcarbaldehyde. LCMS (Table
16, Method a) Rt = 1.26 min.; MS m/z: 336 (M+H)+.
1ꢀ((3',5'ꢀDichloroꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀyl)methyl)azetidineꢀ3ꢀcarboxylic acid (21). Synthesized
from aldehyde 3',5'ꢀdichloroꢀ[1,1'ꢀbiphenyl]ꢀ4ꢀcarbaldehyde. LCMS (Table 16, Method a) Rt =
1.40 min.; MS m/z: 336 (M+H)+.
1ꢀ(4ꢀ(4ꢀChlorophenoxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (22). Synthesized from aldehyde
4ꢀ(4ꢀchlorophenoxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.27 min.; MS m/z: 318
(M+H)+.
1ꢀ(4ꢀ(Benzyloxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (23). Synthesized from aldehyde 4ꢀ
(benzyloxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.16 min.; MS m/z: 298 (M+H)+.
1ꢀ(4ꢀ((2ꢀMethylbenzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (24).
Synthesized from
aldehyde 4ꢀ((2ꢀmethylbenzyl)oxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.30 min.;
MS m/z: 310 (M+H)+.
1ꢀ(4ꢀ((3ꢀChlorobenzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (25).
Synthesized from
aldehyde 4ꢀ((3ꢀchlorobenzyl)oxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.22 min.;
MS m/z: 332 (M+H)+.
1ꢀ(4ꢀ((3ꢀ(Trifluoromethyl)benzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (26). Synthesized
from aldehyde 4ꢀ((3ꢀ(trifluoromethyl)benzyl)oxy)benzaldehyde. LCMS (Table 16, Method a) Rt
= 1.28 min.; MS m/z: 366 (M+H)+.
1ꢀ(4ꢀ((4ꢀFluorobenzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (27).
Synthesized from
aldehyde 4ꢀ((4ꢀfluorobenzyl)oxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.25 min.;
MS m/z: 316 (M+H)+.
1ꢀ(4ꢀ((2,4ꢀDichlorobenzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (28). Synthesized from
aldehyde 4ꢀ((2,4ꢀdichlorobenzyl)oxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.41
min.; MS m/z: 364 (M+H)+.
ACS Paragon Plus Environment
24

Page 25 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
6
1ꢀ(4ꢀ((3,4ꢀDichlorobenzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (29). Synthesized from
aldehyde 4ꢀ((3,4ꢀdichlorobenzyl)oxy)benzaldehyde. LCMS (Table 16, Method a) Rt = 1.37
min.; MS m/z: 366 (M+H)+.
7
8
9
1ꢀ(4ꢀ(Benzyloxy)ꢀ2ꢀmethylbenzyl)azetidineꢀ3ꢀcarboxylic acid (30).
Synthesized from
aldehyde 4ꢀ(benzyloxy)ꢀ2ꢀmethylbenzaldehyde. LCMS (Table 16, Method a) Rt = 1.32 min.;
1
MS m/z: 312 (M+H)+. H NMR (300MHz, d4ꢀMethanol) δ 7.45ꢀ7.43(2H, d), 7.40ꢀ7.37(2H, t),
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
7.34ꢀ7.31(2H, t), 6.98(1H, s), 6.94ꢀ6.92(1H, q), 5.12(2H, s), 4.42(2H, s), 4.35ꢀ4.29(4H, m), 3.73ꢀ
3.66(1H, m), 2.42(3H, s)
1ꢀ(4ꢀ(Benzyloxy)ꢀ2ꢀchlorobenzyl)azetidineꢀ3ꢀcarboxylic acid (31).
Synthesized from
aldehyde 4ꢀ(benzyloxy)ꢀ2ꢀchlorobenzaldehyde. LCMS (Table 16, Method a) Rt = 1.01 min.; MS
1
m/z: 332 (M+H)+. Purity 99%. H NMR (300MHz, d4ꢀMethanol) δ 7.51ꢀ7.49(1H, d), 7.46ꢀ
7.44(2H, d), 7.41ꢀ7.38(2H, q), 7.35ꢀ7.32(1H, q), 7.21ꢀ7.20(1H, d), 7.08ꢀ7.06(1H, q), 5.15 (2H,
s), 4.45(2H, s), 4.22ꢀ4.18(4H, m), 3.46ꢀ3.41(1H, m). HRMS: Found 331.0974 C18H18NO3Cl
requires 331.0975.
1ꢀ(4ꢀ((2,4ꢀDichlorobenzyl)oxy)ꢀ3ꢀmethoxybenzyl)azetidineꢀ3ꢀcarboxylic
acid
(32).
Synthesized from aldehyde 4ꢀ((2,4ꢀdichlorobenzyl)oxy)ꢀ3ꢀmethoxybenzaldehyde. LCMS (Table
16, Method a) Rt = 1.27 min.; MS m/z: 396 (M+H)+.
1ꢀ(4ꢀ((3,4ꢀDichlorobenzyl)oxy)ꢀ3ꢀnitrobenzyl)azetidineꢀ3ꢀcarboxylic acid (33). Synthesized
from aldehyde 4ꢀ((3,4ꢀdichlorobenzyl)oxy)ꢀ3ꢀnitrobenzaldehyde. LCMS (Table 16, Method a)
1
Rt = 1.35 min.; MS m/z: 411 (M+H)+. H NMR (400 MHz, CD3CN) δ 7.77 (d, J = 2.1, 1H),
7.65 (d, J = 1.9, 1H), 7.56 (d, J = 8.3, 1H), 7.51 (dd, J = 2.2, 8.6, 1H), 7.44 ꢀ 7.36 (m, 1H), 7.21
(d, J = 8.6, 1H), 5.20 (s, 2H), 3.59 (s, 2H), 3.45 (t, J = 7.6, 2H), 3.30 (t, J = 6.7, 2H), 3.24 (dd, J =
7.2, 13.9, 1H).
1ꢀ(4ꢀ((2ꢀChlorobenzyl)oxy)ꢀ3ꢀmethoxybenzyl)azetidineꢀ3ꢀcarboxylic acid (34). Synthesized
from aldehyde 4ꢀ((2ꢀchlorobenzyl)oxy)ꢀ3ꢀmethoxybenzaldehyde. LCMS (Table 16, Method a)
Rt = 1.27 min.; MS m/z: 360 (M+H)+.
4ꢀ(1,3ꢀDioxolanꢀ2ꢀyl)benzonitrile (36). A mixture of 4ꢀformylbenzonitrile 35 (13.1 g, 0.1 mol),
ethylene glycol (62 g, 1 mol), pꢀtoluenesulfonic acid monohydrate (1.9 g, 0.01 mol) in 150 mL
of toluene were refluxed overnight. After cooling the mixture to room temperature, it was added
to 200 mL of iceꢀcold water and stirred for 15 min. The organic layer was dried over with
Na2SO4, and the solvent was removed in vacuo.
Purification on silicaꢀgel column
chromatography (PE/EA from 10:1 to 4:1) afforded 4ꢀ(1,3ꢀdioxolanꢀ2ꢀyl)benzonitrile 36 as a
white solid (14.2 g, yield 81%). LCMS (Table 16, Method c) RT 0.66 min; m/z 176.7 (M+H)+.
4ꢀ(2ꢀPhenylacetyl)benzaldehyde. Under N2, to magnesium turnings (792 mg, 32.98 mmol) and
I2 (7 mg) in Et2O (5 mL), was added a solution of (bromomethyl) benzene (3.76 g, 21.99 mmol)
at room temperature. After stirring for 1 h, the mixture was cooled to 0~15°C. A solution of 4ꢀ
(1,3ꢀdioxolanꢀ2ꢀyl)benzonitrile 38 (2.89 g, 16.5 mmol) in Et2O (10 mL) was added dropwise,
then refluxed for 1 h. The solution was cooled to room temperature, and treated with iceꢀwater.
Subsequently, aqueous 5 M HCl was added, whereby two clear phases were formed. Then the
organic phase was separated and the aqueous phase was extracted with EtOAc. The combined
organic phase was washed with sat. NaHSO3 and sat. NaHCO3, dried (Na2SO4) and
concentrated in vacuo to get the crude 1ꢀ(4ꢀ(1,3ꢀdioxolanꢀ2ꢀyl)phenyl)ꢀ2ꢀphenylethanone. The
crude 1ꢀ(4ꢀ(1,3ꢀdioxolanꢀ2ꢀyl)phenyl)ꢀ2ꢀphenylethanone was dissolved in THF (20 mL) and
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 37
1
2
3
4
5
6
7
8
9
10% HCl (30 mL) added to the solution. The reaction mixture was refluxed for 16 h, then cooled
to room temperature. EtOAc was added and the organic layer was (Na2SO4). After removal of
solvent, it was purified by silicaꢀgel column chromatography (PE/EA from 50:1 to 10:1) to
afford 4ꢀ(2ꢀphenylacetyl)benzaldehyde as a white solid (1.1 g, yield 43%). LCMS (Table 16,
Method c) RT 1.57min; m/z 225.1 (M+H)+.
1ꢀ(4ꢀ(2ꢀPhenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic
acid
(41).
4ꢀ(2ꢀ
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
Phenylacetyl)benzaldehyde (133 mg, 0.59 mmol) was added to a stirred solution of azetidineꢀ3ꢀ
carboxylic acid (60 mg, 0.59 mmol) and HOAc (107 mg, 1.78 mmol) in 10 mL MeOH. The
mixture was heated to 40 C. After 15 min, NaCNBH3 (110 mg, 1.78 mmol) was added in a
single portion and stirred at 40 °C overnight. After acidifying with 1 M HCl, purification by
prepꢀHPLC afforded 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)pyrrolidineꢀ3ꢀcarboxylic acid (38.7, yield
1
21%). LCMS (Table 16, Method c) RT 1.23 min; m/z 310.2 (M+H)+. H NMR (300MHz, d4ꢀ
Methanol) δ 8.16ꢀ8.14 (2H, d), 7.61ꢀ7.59 (2H, d), 7.32ꢀ7.24 (5H, m), 4.51ꢀ4.50 (2H, s), 4.39ꢀ
4.36 (4H, m) and 4.35ꢀ4.33 (2H, m).
1ꢀ(4ꢀPhenethylbenzyl)azetidineꢀ3ꢀcarboxylic
acid
(42).
1ꢀ(4ꢀ(2ꢀ
Phenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 41 (200 mg, 0.65 mmol)was dissolved in 1%
(V/V) H2SO4/EtOH (50mL), to which Pd/C (10 mg) was added and the hydrogenation reaction
carried out at room temperature for 32 h. After completion of the reaction, the catalyst was
filtered off and ethanol solution neutralized with NaOH, followed by distillation of the solvent.
HPLC afforded 1ꢀ(4ꢀphenethylbenzyl)azetidineꢀ3ꢀcarboxylic acid (100mg , 52%). LCMS (Table
1
16, Method c) RT 1.35min; m/z 296.2 (M+H)+. H NMR (300MHz, d4ꢀMethanol) δ 7.24ꢀ7.23
(2H, d), 7.18ꢀ7.16 (2H, d), 7.13ꢀ7.10 (2H, t), 7.05ꢀ7.03 (2H, t), 7.02 (1H, s), 4.25 (2H, s), 4.19ꢀ
4.18 (2H, m), 4.17ꢀ4.15 (2H, m), 3.22ꢀ3.20 (1H, m), 2.87ꢀ2.84 (2H, m) and 2.82ꢀ2.79 (2H, m).
1ꢀ(4ꢀ(3ꢀ(Trifluoromethyl)phenethyl)benzyl)azetidineꢀ3ꢀcarboxylic acid (43).
1ꢀ(4ꢀ(3ꢀ
(Trifluoromethyl)phenethyl)benzyl)azetidineꢀ3ꢀcarboxylic acid was synthesized using the same
procedure as 1ꢀ(4ꢀphenethylbenzyl)azetidineꢀ3ꢀcarboxylic acid starting from 1ꢀ(bromomethyl)ꢀ3ꢀ
(trifluoromethyl)benzene. LCMS (Table 16, Method c) RT 0.63 min; m/z 364.2 (M+H)+. Purity
1
96%. H NMR (300MHz, d4ꢀMethanol) δ 7.35ꢀ7.28 (4H, m), 7.09ꢀ7.08 (2H, d), 7.02ꢀ7.00 (2H,
d), 3.49 (2H, s), 3.43ꢀ3.40 (2H, t), 3.25ꢀ3.23 (2H, d), 3.12ꢀ3.06 (1H, m), 2.90ꢀ2.87 (2H, t) and
2.83ꢀ2.79 (2H, t). HRMS: Found 363.1444 C20H20NO2F3 requires 363.1446.
1ꢀ(4ꢀHexanoylbenzyl)azetidineꢀ3ꢀcarboxylic acid (44). Prepared using the same procedure as
for 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 51 using 1ꢀbromopentane. LCMS
1
(Table 16, Method c) RT 1.32 min; m/z 290.2 (M+H)+. H NMR (300MHz, d4ꢀMethanol) δ
8.07ꢀ8.05 (2H, d), 7.59ꢀ7.57 (2H, d), 4.37 (2H, s), 4.16ꢀ4.14 (4H, m), 3.43ꢀ3.40 (1H, m), 3.06ꢀ
3.03 (2H, t), 1.76ꢀ1.70 (2H, m), 1.40ꢀ1.35 (4H, m) and 0.96ꢀ0.93 (3H, t).
1ꢀ(4ꢀHexanoylbenzyl)pyrrolidineꢀ3ꢀcarboxylic acid (45). Prepared using the same procedure
as for 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 51 using 1ꢀbromopentane and
pyrrolidineꢀ3ꢀcarboxylic acid. LCMS (Table 16, Method c) RT 1.32 min; m/z 304.3 (M+H)+.
1H NMR (300MHz, d4ꢀMethanol) δ 8.06ꢀ8.04(2H, d), 7.66ꢀ7.64(2H, d), 4.38ꢀ4.31 (2H, q), 3.48ꢀ
3.46 (1H, m), 3.33ꢀ3.31 (2H, m), 3.25ꢀ3.08 (2H, m), 3.06ꢀ3.03 (2H, t), 2.32ꢀ2.26 (2H, m), 1.74ꢀ
1.70 (2H, m), 1.40ꢀ1.37 (4H, m) and 0.96ꢀ0.93 (3H, t).
1ꢀ(4ꢀHeptanoylbenzyl)azetidineꢀ3ꢀcarboxylic acid (46). Prepared using the same procedure as
for 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 51 using 1ꢀbromohexane. LCMS
1
(Table 16, Method c) RT 1.39 min; m/z 304.3 (M+H)+. H NMR (300MHz, d4ꢀMethanol) δ
ACS Paragon Plus Environment
26

Page 27 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
8.08ꢀ8.06 (2H, d), 7.59ꢀ7.58 (2H, d), 4.38 (2H, s), 4.18ꢀ4.12 (4H, m), 3.45ꢀ3.40 (1H, m), 3.06ꢀ
3.03 (2H, t), 1.75ꢀ1.70 (2H, m), 1.42ꢀ1.31 (6H, m) and 0.94ꢀ0.92 (3H, t).
6
7
8
9
1ꢀ(4ꢀHeptanoylbenzyl)pyrrolidineꢀ3ꢀcarboxylic acid (47). Prepared using the same procedure
as for 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 53 using 1ꢀbromohexane and
pyrrolidineꢀ3ꢀcarboxylic acid. LCMS (Table 16, Method c) RT 1.40 min; m/z 318.3 (M+H)+.
1H NMR (300MHz, d4ꢀMethanol) δ 8.04ꢀ8.02 (2H, d), 7.67ꢀ7.65 (2H, d), 4.39ꢀ4.33 (2H, q), 3.49
(1H, s), 3.37ꢀ3.32 (2H, m), 3.27ꢀ3.23 (2H, m), 3.04ꢀ3.01 (2H, t), 2.37ꢀ2.23 (2H, m), 1.73ꢀ1.67
(2H, m), 1.44ꢀ1.37 (2H, m), 1.36ꢀ1.31 (4H, m) and 0.93ꢀ0.90 (3H, t).
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
1ꢀ(4ꢀ(2ꢀPhenylacetyl)benzyl)pyrrolidineꢀ3ꢀcarboxylic acid (49). Prepared using the same
procedure as for 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 53 using pyrrolidineꢀ3ꢀ
carboxylic acid. LCMS (Table 16, Method c) RT 1.24 min; m/z 324.2 (M+H)+. 1H NMR
(300MHz, d4ꢀMethanol) δ 8.17ꢀ8.15(2H, d), 7.68ꢀ7.66(2H, d), 7.34ꢀ7.30 (2H, t), 7.29ꢀ7.27(2H,
d), 7.26ꢀ7.22(2H, t), 4.53ꢀ4.47(2H, t), 4.38(2H, s), 3.77ꢀ3.76(6H, m), 2.45ꢀ2.34(2H, d).
1ꢀ(4ꢀ(2ꢀ(3,4ꢀDichlorophenyl)acetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid (50). Prepared using
the same procedure as for 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 51 using 4ꢀ
(bromomethyl)ꢀ1,2ꢀdichlorobenzene. LCMS (Table 16, Method c) RT 1.39 min; m/z 378.1
1
(M+H)+. H NMR (300MHz, d4ꢀMethanol) δ 8.05ꢀ8.03(2H, d), 7.53ꢀ7.51(2H, d), 7.37ꢀ7.36 (2H,
d), 7.10ꢀ7.08(1H, m), 4.40(2H, s), 4.31(2H, s), 4.27ꢀ4.22(4H, m), 3.62ꢀ3.59(1H, m).
1ꢀ(4ꢀ(2ꢀ(3,4ꢀDichlorophenyl)acetyl)benzyl)pyrrolidineꢀ3ꢀcarboxylic acid (51).
Prepared
using the same procedure as for 1ꢀ(4ꢀ(2ꢀphenylacetyl)benzyl)azetidineꢀ3ꢀcarboxylic acid 51
using 4ꢀ(bromomethyl)ꢀ1,2ꢀdichlorobenzeneand pyrrolidineꢀ3ꢀcarboxylic acid. LCMS (Table
16, Method c) RT 1.40 min; m/z 392.1 (M+H)+. 1H NMR (300MHz, d4ꢀMethanol) δ 8.14ꢀ
8.13(2H, d), 7.66ꢀ7.65(2H, d), 7.48ꢀ7.47(2H, m), 7.23ꢀ7.21(1H, q), 4.36ꢀ4.28(2H, q), 3.42ꢀ
3.41(1H, m), 3.30ꢀ3.11(4H, m), 2.32ꢀ2.24(2H, m).
4ꢀ(Benzyloxy)ꢀ3ꢀfluorobenzonitrile. Under N2, DIAD ( 0.32 g, 1.54 mmol) was treated with
Ph3P (0.41 g, 4.54 mmol) at 0 °C in dry THF(10 mL). The mixture was stirred until there was a
precipitate. Then 3ꢀfluoroꢀ4ꢀhydroxybenzonitrile (0.2 g, 1.46 mmol) and benzyl alcohol (0.17 g,
1.54 mmol) were added at the same time. The mixture was warmed to room temperature, and
stirred overnight. The reaction mixture was concentrated and purified by silicaꢀgel column
chromatography (PE/EA = 4:1) to afford 4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzonitrile as a white solid
(0.28 g, yield 84%). GCꢀMS RT 10.83 min; m/z 227.1 (M).
4ꢀ(Benzyloxy)ꢀ3ꢀfluorobenzaldehyde. 4ꢀ(Benzyloxy)ꢀ3ꢀfluorobenzonitrile (645 mg,2.84 mmol)
was dissolved in toluene (10 mL) and cooled to 0 °C. A portion of 1M DIBAL (4.55 mmol, 4.55
mL) in hexane was added dropwise under N2. The solution was further stirred for 1 hour at 0 °C.
Chloroform (12 mL) was then added followed by 10% HCl (30 mL), and the resulting solution
stirred at room temperature for 1 h. The organic layer was separated, washed with distilled
water, dried with Na2SO4 and filtered. After removal of the solvent, the residue was purified by
silicaꢀgel column chromatography (PE/EA = 4:1) to afford 4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzaldehyde
as a white solid (0.62 g, yield 95%). LCMS (Table 16, Method c) RT 1.26 min; m/z 231.1
(M+H)+.
1ꢀ(4ꢀ(Benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀcarboxylic acid (57).
4ꢀ(Benzyloxy)ꢀ3ꢀ
fluorobenzaldehyde (653 mg, 2.84 mmol) was added to a stirred solution of azetidineꢀ3ꢀ
carboxylic acid (287 mg, 2.84 mmol) and HOAc (536 mg, 8.52 mmol) in 5 mL MeOH. The
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 37
1
2
3
4
5
6
7
8
9
mixture was heated to 40 ℃. After 15 min, NaCNBH3 (512 mg, 8.52 mmol) was added in a
single portion and stirred at 40 ℃ overnight. After acidifying with 1M HCl, the residue was
purified by PrepꢀHPLC affording 1ꢀ(4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀcarboxylic acid
1
(571 mg, yield 63%). LCMS (Table 16, Method c) RT 0.95min; m/z 316.2 (M+H)+. H NMR
(300MHz, d4ꢀMethanol): δ 7.44ꢀ7.43 (2H, d), 7.39ꢀ7.36 (2H, m), 7.33ꢀ7.30 (1H, m), 7.28ꢀ7.25
(1H, m), 7.23ꢀ7.22 (1H, d), 7.20ꢀ7.18 (1H, m), 5.19 (2H, s), 4.32 (2H, s), 4.29ꢀ4.25(4H, m) and
3.68ꢀ3.63 (1H, m).
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
1ꢀ(4ꢀ((3,4ꢀDichlorobenzyl)oxy)ꢀ2ꢀfluorobenzyl)azetidineꢀ3ꢀcarboxylic acid (58). Prepared
using the same procedure as for 1ꢀ(4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀcarboxylic acid 60
using 4ꢀ(bromomethyl)ꢀ1,2ꢀdichlorobenzene and 2ꢀfluoroꢀ4ꢀhydroxybenzonitrile. LCMS (Table
1
16, Method c) RT 1.10min; m/z 384.1 (M+H)+. H NMR (300MHz, d4ꢀMethanol) δ 7.52 ( 1H,
s), 7.45ꢀ7.44 (1H, d), 7.36ꢀ7.32 (1H, t), 7.29ꢀ7.27 (1H, d), 6.87ꢀ6.84 (2H, q), 5.04 (2H, s), 4.33
(2H, s), 4.27ꢀ4.20 (4H, m) and 3.59ꢀ3.56 (1H, m).
1ꢀ(4ꢀ((3,4ꢀDichlorobenzyl)oxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀcarboxylic acid (59). Prepared
using the same procedure as for 1ꢀ(4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀcarboxylic acid 60
using 4ꢀ(bromomethyl)ꢀ1,2ꢀdichlorobenzene. LCMS (Table 16, Method c) RT 1.11min; m/z
1
384.1 (M+H)+. H NMR (300MHz, d4ꢀMethanol) δ 7.64 (1H, s), 7.57ꢀ7.55 (1H, d), 7.41ꢀ7.40
(1H, d), 7.31ꢀ7.29 (1H, d), 7.24ꢀ7.23 (2H, q), 5.21 (2H, s), 4.31 (2H, s), 4.22ꢀ4.21 (4H, d) and
3.50ꢀ3.33 (1H, m).
1ꢀ(2ꢀFluoroꢀ4ꢀ((3ꢀ(trifluoromethyl)benzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (60).
Prepared using the same procedure as for 1ꢀ(4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀ
carboxylic acid 59 using 1ꢀ(bromomethyl)ꢀ3ꢀ(trifluoromethyl)benzene and 2ꢀfluoroꢀ4ꢀ
hydroxybenzonitrile. LCMS (Table 16, Method c) RT 1.08min; m/z 384.2 (M+H)+. Purity 97%.
1H NMR (300MHz, d4ꢀMethanol) δ 7.75 ( 1H, s), 7.72ꢀ7.70 (1H, d), 7.65ꢀ7.63 (1H, d), 7.60ꢀ7.57
(1H, t), 7.45ꢀ7.42 (1H, t), 6.98ꢀ6.96 (2H, q), 5.22 (2H, s), 4.42 (2H, s), 4.36ꢀ4.29 (4H, m) and
3.70ꢀ3.63 (1H, m). HRMS: Found 383.1145 C19H17NO3F4 requires 383.1145.
1ꢀ(3ꢀFluoroꢀ4ꢀ((3ꢀ(trifluoromethyl)benzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (61).
Prepared using the same procedure as for 1ꢀ(4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀ
carboxylic acid 60 using 1ꢀ(bromomethyl)ꢀ3ꢀ(trifluoromethyl)benzene. LCMS (Table 16,
Method c) RT 1.08min; m/z 384.2 (M+H)+. 1H NMR(300MHz, d4ꢀMethanol): δ 7.76 (1H, s),
7.73ꢀ7.71 (1H, d), 7.65ꢀ7.64 (1H, d), 7.61ꢀ7.58 (1H, t), 7.31ꢀ7.28 (1H, m), 7.27ꢀ7.25 (1H, d),
7.23ꢀ7.21 (1H, m), 5.23 (2H, s), 4.33 (2H, s), 4.30ꢀ4.26 (4H, m) and 3.69ꢀ3.63 (1H, m).
1ꢀ(2ꢀChloroꢀ4ꢀ((3ꢀ(trifluoromethyl)benzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (62).
Prepared using the same procedure as for 1ꢀ(4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀ
carboxylic acid 57 using 1ꢀ(bromomethyl)ꢀ3ꢀ(trifluoromethyl)benzene and 2ꢀchloroꢀ4ꢀ
1
hydroxybenzonitrile. LCMS (Table 16, Method c) RT 1.12min; m/z 400.1 (M+H)+. H NMR
(300MHz, d4ꢀMethanol) δ 7.66 (1H, s), 7.62ꢀ7.61 (1H, d), 7.55ꢀ7.53 (1H, d), 7.50ꢀ7.47 (1H, t),
7.43ꢀ7.42 (1H, d), 7.15ꢀ7.14 (1H, d), 7.01ꢀ6.98 (1H, q), 5.12 (2H, s), 4.44 (2H, s), 4.30ꢀ4.23 (4H,
m) and 3.62ꢀ3.58 (1H, m).
1ꢀ(3ꢀChloroꢀ4ꢀ((3ꢀ(trifluoromethyl)benzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (63).
Prepared using the same procedure as for 1ꢀ(4ꢀ(benzyloxy)ꢀ3ꢀfluorobenzyl)azetidineꢀ3ꢀ
carboxylic acid 57 using 1ꢀ(bromomethyl)ꢀ3ꢀ(trifluoromethyl)benzene and 3ꢀchloroꢀ4ꢀ
1
hydroxybenzonitrile. LCMS (Table 16, Method c) RT 1.12min; m/z 400.1 (M+H)+. H NMR
(300MHz, d4ꢀMethanol) δ 7.82 (1H, s), 7.77ꢀ7.76 (1H, d), 7.67ꢀ7.65 (1H, d), 7.63ꢀ7.60 (2H, q),
ACS Paragon Plus Environment
28

Page 29 of 37
Journal of Medicinal Chemistry
1
2
3
4
5
7.42ꢀ7.40 (1H, q), 7.27ꢀ7.26 (1H, d), 5.33 (2H, s), 4.36 (2H, s), 4.33ꢀ4.28 (4H, m) and 3.72ꢀ3.66
(1H, m).
6
7
8
9
1ꢀBromoꢀ4ꢀ(3,4ꢀdichlorobenzyloxy)ꢀ2ꢀmethylbenzene. Under N2, DIAD (908 mg, 4.49 mmol)
was treated with Ph3P (1.18 g, 4.49 mmol) at 0 °C in dry THF(20 mL). The mixture was stirred
until there was a precipitate. Then 4ꢀbromoꢀ3ꢀmethylphenol (0.8 g g, 4.23 mmol) and 3,4ꢀ
dicholorobenzyl alcohol (795 mg, 4.49 mmol) were added at the same time. The mixture was
warmed to room temperature, and stirred overnight. After concentrating, purification by silicaꢀ
gel column chromatography (PE/EA = 4:1) afforded 1ꢀbromoꢀ4ꢀ(3,4ꢀdichlorobenzyloxy)ꢀ2ꢀ
methylbenzene as a white solid (0.94 g, yield 64%). GCꢀMS: m/z 345.9 (M), RT: 13.96 min. H
NMR (300MHz, CDCl3): δ 7.52 (1H, s), 7.46ꢀ7.44 (1H, d), 7.42ꢀ7.40 (1H, d), 7.25ꢀ7.23 (1H, q),
6.85ꢀ6.84 (1H, d), 6.66ꢀ6.64 (1H, q), 4.97 (2H, s) and 2.36 (3H, 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
1
4ꢀ(3,4ꢀDichlorobenzyloxy)ꢀ2ꢀmethylbenzonitrile.
A
mixture of 1ꢀbromoꢀ4ꢀ(3,4ꢀ
dichlorobenzyloxy)ꢀ2ꢀmethylbenzene (0.94 g, 207 mmol) and CuCN (0.56 g, 6.2 mmol) in Nꢀ
methylꢀ2ꢀpyrrolidone (10 mL) was heated to 160 °C for 32 h. After cooling slightly, the warm
mixture was washed with 5% NaCN and extracted with ether. The combined extract was washed
with 5% NaCN, water, brine and dried (Na2SO4). After removal of solvent in vacuo, the residue
was purified by silicaꢀgel column chromatography (PE/EA = 10:1) to afford 4ꢀ(3,4ꢀ
dichlorobenzyloxy)ꢀ2ꢀmethylbenzonitrile as a white solid (496 mg, yield 63%). LCMS (Table
16, Method c) RT 1.46 min; m/z 292.0 (M+H)+.
4ꢀ(4,5ꢀDichloroꢀ2ꢀmethylbenzyloxy)benzaldehyde.
4ꢀ(3,4ꢀdichlorobenzyloxy)ꢀ2ꢀ
methylbenzonitrile (644 mg, 2.2 mmol) was dissolved in toluene (35 mL) and cooled to 0 ° C. A
portion of 1M DIBAL (3.5 mmol, 3.5 mL) in hexane was added dropwise under N2. The
solution was stirred for an hour at 0 °C. CHCl3 (40 mL) was then added followed by 10% HCl
(30 mL), and the solution was stirred at room temperature for 1 h. The organic layer was
separated, washed with distilled water, dried (Na2SO4) and filtered. After removal of the solvent,
the residue was purified by silicaꢀgel column chromatography (PE/EA = 4:1) to afford 4ꢀ(4,5ꢀ
dichloroꢀ2ꢀmethylbenzyloxy)benzaldehyde as a white solid (345 mg, yield 53%). LCMS
Method c) RT 1.85 min; m/z 295.0 (M+H)+.
1ꢀ(2ꢀMethylꢀ4ꢀ((3ꢀ(trifluoromethyl)benzyl)oxy)benzyl)azetidineꢀ3ꢀcarboxylic acid (69).
Prepared
using
the
same
procedure
as
for
1ꢀ(4ꢀ(3,4ꢀdichlorobenzyloxy)ꢀ2ꢀ
methylbenzyl)azetidineꢀ3ꢀcarboxylic acid 75 using 1ꢀ(bromomethyl)ꢀ3ꢀ(trifluoromethyl)benzene.
LCMS (Table 16, Method c) RT 1.43min; m/z 380.2 (M+H)+.1H NMR (300MHz, d4ꢀMethanol)
δ 7.59(1H, s), 7.56ꢀ7.54 (1H, d), 7.48ꢀ7.46 (1H, d), 7.44ꢀ7.41 (1H, t), 7.19ꢀ7.17 (1H, d), 6.82
(1H, d), 6.82ꢀ6.76 (1H, q), 5.03 (2H, s), 4.22 (2H, s), 4.06ꢀ4.05 (4H, d), 3.30ꢀ3.29 (1H, m) and
2.25(3H, s).
1ꢀ(4ꢀ(3,4ꢀDichlorobenzyloxy)ꢀ2ꢀmethylbenzyl)azetidineꢀ3ꢀcarboxylic acid (70).
4ꢀ(4,5ꢀ
dichloroꢀ2ꢀmethylbenzyloxy)benzaldehyde (345 mg, 1.17 mmol) was added to a stirred solution
of azetidineꢀ3ꢀcarboxylic acid (118 mg, 1.17 mmol) and HOAc (211 mg, 3.51 mmol) in 5 mL
CH3OH. The mixture was heated to 40 °C. After 15 min, NaCNBH3 (218 mg, 3.51 mmol) was
added in a single portion and stirred at 40 °C overnight. After being acidified with 1M HCl prep
HPLC afforded 1ꢀ(4ꢀ(3,4ꢀdichlorobenzyloxy)ꢀ2ꢀmethylbenzyl)azetidineꢀ3ꢀcarboxylic acid 75
1
(201 mg, yield 45%). LCMS (Table 16, Method c) RT 1.47 min; m/z 380.1 (M+H)+. H NMR
(300MHz, d4ꢀMethanol) δ 7.49ꢀ7.48 (1H, d), 7.42ꢀ7.40 (1H, d), 7.26ꢀ7.24 (1H, q), 7.22ꢀ7.20 (2H,
ACS Paragon Plus Environment
29