Structure-activity relationships of 1-alkyl-5-(3,4-dichlorophenyl)-5-{2-[3- (substituted)-1-azetidinyl]-ethyl}-2-piperidones. Part 2: Improving oral absorption

Article abstract of DOI:10.1016/j.bmcl.2005.05.134

A series of piperidone analogues of 1b-q, seeking replacements for the polar sulfamide moiety in clinical candidate UK-224,671 1a, possessing reduced H-bonding potential as a strategy to improve oral absorption, were prepared. These studies led to the successful identification of 1n, which demonstrated equivalent pharmacology and metabolic stability to 1a, and greatly improved oral absorption as assessed in rat PK studies.

Full text of DOI:10.1016/j.bmcl.2005.05.134

Bioorganic & Medicinal Chemistry Letters 15 (2005) 3957–3961
Structure–activity relationships of 1-alkyl-5-(3,4-
dichlorophenyl)-5-{2-[3-(substituted)-1-azetidinyl]-ethyl}-2-
piperidones. Part 2: Improving oral absorption
Donald S. Middleton,a,* A. Roderick MacKenzie,a Sandra D. Newman,a Martin Corless,a
Andrew Warren,a Allan P. Marchingtona and Barry Jonesb
aDepartment of Discovery Chemistry, Pfizer Global Research and Development, Sandwich, Kent CT13 9NJ, UK
bDepartment of Pharmacokinetics, Dynamics and Metabolism, Pfizer Global Research and Development, Sandwich,
Kent CT13 9NJ, UK
Received 22 March 2005; revised 17 May 2005; accepted 17 May 2005
Available online 20 July 2005
Abstract—A series of piperidone analogues of 1b–q, seeking replacements for the polar sulfamide moiety in clinical candidate
UK-224,671 1a, possessing reduced H-bonding potential as a strategy to improve oral absorption, were prepared. These studies
led to the successful identification of 1n, which demonstrated equivalent pharmacology and metabolic stability to 1a, and greatly
improved oral absorption as assessed in rat PK studies.
Ó 2005 Elsevier Ltd. All rights reserved.
1. Introduction
tandem, we sought to identify alternative, metaboli-
cally stable replacements for the N-cyclopropylmethyl
piperidone substituent, which could confer greater
potency. These studies led to the successful identifica-
tion of 1n, which possessed both equivalent pharma-
cology and metabolic stability to 1a and greatly
improved oral absorption as assessed in rat PK
studies.
Tachykinins and their receptors have been postulated to
play a significant role in the pathophysiology of a wide
variety of diseases including gastrointestinal disorders,1
emesis,2 chronic pain,3 depression4 and asthma.5 Our
own long-standing efforts in this field have focused on
investigating the role of neurokinin-2 (NK2) antagonists
in urological disorders.6 In an earlier communication
from these laboratories,6c we reported the discovery of
UK-224,671 1a, (Fig. 1), as a potent selective antagonist
of the NK2 receptor. Progression of this compound into
human clinical trials, however, revealed the compound
to possess poor oral bioavailability,7 as a consequence
of poor oral absorption.8
2. Chemistry
Test compounds 1a–f, h–n, and p–q were prepared from
homochiral (S)-piperidone (2)9 by the general method in
Scheme 1, as described previously.9,10 Alkylation of pip-
eridone (2) with the appropriate alkyl halide or tosylate
as leaving group, followed by acid-mediated deprotec-
tion of the dioxolane group, yielded the aldehyde.
Reductive amination using NaBH(OAc)3 and the appro-
priate 3-substituted azetidine gave targets 1a–f, h–n, and
p–q.
In this communication, we report a successful strategy
to improve the oral absorption of this piperidone series
while retaining high functional potency against the hu-
man NK2 receptor.
Our strategy (Fig. 1) focused on replacement of the
polar (high H-bond count), primary sulfamide. In
Test compounds 1g and o were also prepared by the
methods previously described10 as shown in Scheme 2.
Treatment of ester (3)10 with amine (4)11 at 100 °C gave
the amide (5). Conversion of the primary alcohol to the
mesylate and intramolecular displacement with the
*
Corresponding author.
don.s.middleton@pfizer.com
Fax:
+44
1304
651987; e-mail:
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.05.134

3958
D. S. Middleton et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3957–3961
Figure 1. Revised medicinal chemistry strategy to improve oral absorption in piperidone series (1).
3. Results and discussion
All binding, functional and human liver microsomal sta-
bility screening were performed using the methods previ-
ously reported.6c
Previous communications from these laboratories
described the discovery of potent NK2 antagonist,
UK-224,671.6c Progression of this compound into hu-
man clinical trials revealed the compound to possess
long terminal half-life (28 h) and low clearance (2–
3 ml/min/kg).7 Oral bioavailability in human, however,
was found to be very low (5–10%), indicative of very
poor absorption, likely caused by a combination of
intrinsically poor permeability and transporter-mediated
(P–gp) efflux.8
Scheme 1. Synthesis of 1-alkyl-5-(3,4-dichlorophenyl)-5-{2-[3-(substi-
tuted)-1-azetidinyl]-ethyl}-2-piperidones 1a–f, h–n, p,q. Reagents and
conditions: (a) R–X (X = Br or –OTs), KOH, DMSO, 16 h, rt; (b) 5 N
HCl, THF, reflux, 4 h; (c) 3-(substituted)azetidine, NaBH(OAc)3,
AcOH, NEt3, THF, rt.
Our medicinal chemistry objectives in seeking a clinical
replacement for UK-224,671 were therefore to (1) retain
the excellent potency and selectivity; (2) retain high met-
abolic stability, as assessed in human liver microsomes
(HLM, T1/2 > 120 min), as a strategy to minimise first-
pass in vivo metabolism; and (3) significantly improve
intrinsic transcellular permeability, as assessed in a
Caco-212 model for passive GI transit (target: Caco-2
>20%/h), to minimise any potential impact of P–gp
affinity in this series.
We made a decision to remain in the piperidone series
from which UK-224,671 originated, as this template of-
fered excellent potency, general selectivity and metabolic
stability. Much work has been published on physico-
chemical descriptors for oral absorption13 and prelimin-
ary assessment of this series (MW 500–550 kDa,
relatively low H-bond donor (HBD) and acceptor
count, moderate, positive, c log P) did not suggest that
its physicochemistry should be incompatible with oral
absorption. However, while we did not regard the
HBD count in 1a as excessively high, (HBD 2), we felt
this, combined with the exposed nature of the polar pri-
mary sulfamide moiety, high topological polar surface
Scheme 2. Synthesis of 1-alkyl-5-(3,4-dichlorophenyl)-5-{2-[3-(substi-
tuted)-1-azetidinyl]-ethyl}-2-piperidones 1g and o. Reagents and con-
ditions: (a) (4), 100 °C, 16 h; (b) MsCl, CH2Cl2, NEt3, rt; (c) NaH,
THF, reflux; (d) O3, MeOH, Me2S; (e) 3-(N-morpholino)azetidine (1g)
or 3-(4-fluoropiperidinyl)azetidine (1o), NaBH(OAc)3, AcOH, NEt3,
THF, rt.
amide anion, generated using sodium hydride, gave
piperidone (6). Ozonolysis of the terminal olefin and
reductive amination of the resulting aldehyde with
appropriate 3-substituted azetidine in the presence of
NaBH(OAc)3 gave target compounds 1g and o.
2
˚
area (TPSA) (98 A ), and relatively high MW (545),
might explain the origin of the poor permeability ob-
served in man.
We, therefore, targeted modulation of TPSA and HBD
count in 1a to improve intrinsic permeability as the prin-
cipal focus of our design programme, recognising that a
The synthesis of all 3-substituted azetidines has been
reported previously.9,10

D. S. Middleton et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3957–3961
3959
likely consequence of this strategy would also be to in-
crease lipophilicity, again, consistent with our goal of
improving permeability.
by morpholine analogue 1c (Table 1) (HBD 0, MW
2
466 kDa, TPSA 36 A ), which showed greatly improved
˚
permeability in the Caco-2 assay [A–B, 12%/h], relative
to 1a (A–B 1%/h), while retaining good stability in the
HLM assay (T1/2 120 min). Unfortunately, 1c, demon-
strated low potency in h-NK2 binding studies (pIC50
We were reassured that high permeability, coupled to
good HLM stability, could be achieved in this series
Table 1. Human, NK2 binding, functional (RPA) rabbit pulmonary artery, human liver microsomal half-life and Caco-2 flux permeability data for
1-alkyl-5-(3,4-dichlorophenyl)-5-{2-[3-(substituted)-1-azetidinyl]-ethyl}-2-piperidones 1a–1q
Compounda
R
X
NK2
(pIC50
RPA
c log P
HLM
(T1/2, min) weight
Molecular TPSA HBD Caco-2
2
(A )
b
(pA2)b (log D,
pH 7.4)
(%/h)
(A–B/B–A)
˚
)
(kDa)
UK-224,671c
1a
8.4
7.8
8.9
8.0
2.2 (1.7) >120
545
98
2
0
1/18
1b
1c
1d
1e
1f
2.3 NT
573
466
480
509
544
521
509
545
622
558
559
557
561
537
482
497
76
NT
7.1
8.2
7.8
8.1
9.5d
9.1
8.1
8.9
9.0
8.9
8.5
9.4
10.0d
8.0
8.6
8.7
8.7
8.2
8.5
9.6
9.0
8.9
9.0
9.7
10.1
9.7
9.2
2.5 (2.3) 120
36
47
36
73
36
36
36
73
53
47
44
27
27
27
27
0
2
0
0
0
0
0
0
2
1
0
0
0
0
0
12/18
f
1.4
2.4
2.2
4.1
4.2
3.9
3.3
NT
NT
<10
< 1
14
2/3
34 ER = 1
10 ER < 2
NT
1g
1h
1i
12/16
33/35
6/12
80
1j
NT
1k
1l
2.5 (0.1) NT
2.4 (2.2) NT
< 1
2/12
1m
1ne
1o
1p
1q
3.8
4.1
4.7
3.1
3.6
NT
>120
2
NT
>35 ER = 1
60/60
60/60
NT
NT
8.6
8.8
85
NT
a All compounds are single (S)-enantiomers.
b All assay determination Pn = 2 (each experiment performed in triplicate). NT = not tested; ER = efflux ratio (ratio of apical-basolateral
(A–B):basolateral-apical (B–A)).
c pA2 (human bladder) = 8.8.
d pKi.
e pA2 (human bladder) = 8.0. human NK1 binding (IM9 cells) IC50 > 10 lM.
f Papp value (106 cm/s). This value is indicative of very poor transcellular permeability.

3960
D. S. Middleton et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3957–3961
7.1). Introduction of polar, 4-aminopiperidine function-
ality 1d gave a >10-fold increase in binding potency
(h-NK2 pIC50 8.2) relative to 1c (h-NK2 pIC50 7.1); how-
ever, the compound was poorly fluxed in the Caco-2 as-
say. 4-Ethoxypiperidine analogue 1e, which possessed no
HBD functionality, showed excellent Caco-2 flux (A–B,
34%/h). Similarly, piperazine sulfonamide 1f, which also
possessed no HBD groups, displayed significantly higher
permeability in the Caco-2 assay than 1d (A–B, 10%/h)
from poor permeability in the Caco-2 assay (A–B, 1%/
h and 2%/h, respectively), consistent with the emerging
SAR that any HBD functionality in this region was gen-
erally not compatible with high permeability in the
Caco-2 assay. Simple oxidation of the primary alcohol
in 1l to the ketone 1m was also found to retain potency.
Excitingly, replacing the 4-hydroxy group in 1l with a 4-
fluoro substituent 1n gave a compound with excellent
potency (pIC50 9.4, RPA pA2 9.2, human bladder pA2
8.0), coupled with high permeability in the Caco-2 assay
(A–B, >35%/h). In addition, 1n retained high metabolic
stability in the HLM assay (T1/2 > 120 min), the high
metabolic stability of 1n being particularly noteworthy,
given its significantly increased lipophilicity (c log P
4.1) relative to 1a (c log P, 2.3). Similarly, very high per-
meability was observed with 4-fluoropiperidine ana-
logues 1o and p (A–B 60%/h), although with highly
variable metabolic stabilities (HLM; 1o, T1/2 2 min; 1p,
T1/2 85 min). Attempts to further improve the potency
and stability profile through incorporating N-cyclopro-
pyl ethyl substitution (1q) were also investigated; howev-
er, this compound did not possess our overall target
profile.
2
2
˚
˚
despite both higher TPSA (73 A vs 47 A ) and MW
(544 kDa vs 480 kDa) than 1d.
These data suggested that minimising HBD count, in
particular, was key to achieving good permeability in
this series. Building on these observations, simple dime-
thylation (1b) (Table 1) of the primary sulfamide in 1a as
a strategy to remove the two HBD groups was under-
taken. Unfortunately, this resulted in an unacceptable
drop in potency in both the human NK2 binding assay
(pIC50 7.8) and the rabbit pulmonary artery (RPA)
functional screen (pA2 8.0).
Increasing the lipophilicity of the piperidone N-substitu-
ent in the 3-(N-morpholino)azetidine series as a strategy
to increase potency,6c 1g and h, did result in an increase
in activity against the human NK2 receptor (pKi 9.5 and
pIC50 9.1, respectively). However, in both instances,
these piperidone N-substitutions led to a reduction in
metabolic stability, relative to the N-cyclopropylmethyl
analogue 1c (1g, HLM T1/2 < 1 min; 1h, HLM, T1/2
14 min).
Analogue 1n was progressed into rat pharmacokinetic
studies (Table 2) to assess whether the in vitro perme-
ability improvement over 1a seen in the Caco-2 assay
translated into improved absorption in vivo. Excitingly,
1n exhibited greatly improved oral absorption (>80%)
relative to prototype clinical candidate 1a (<20%), con-
sistent with the observed excellent Caco-2 permeability.
This, coupled to its high metabolic stability in HLM
studies, equivalent to 1a which showed low clearance
in human, made 1n highly attractive. Following broader
profiling, 1n, UK-290,795 was nominated for clinical
development.
Metabolic route profiling in this series revealed that
CYP3A4-mediated oxidation at the 4-position of the
N-cyclohexyl ring in 1h was a likely major metabolic
pathway. Blocking this metabolically vulnerable site by
difluorination (1i) resulted in a significant improvement
in both metabolic stability relative to 1h (1h, T1/2 14 min;
1i, T1/2 80 min) and a further improvement in permeabil-
ity in the Caco-2 assay (A–B, 33%/h). Unfortunately,
this 4,4,-difluoro substitution resulted in a 10-fold
reduction in potency in the h-NK2 binding assay (1i,
pIC50 8.1; 1h, pIC50 9.1). As a strategy to further im-
prove potency in this series, we then evaluated a range
of alternative 3-azetidine substituents 1j–n, all designed
to possess lower HBD count and/or TPSA than 1a.
Piperazine sulfonamide (1j) was found to be potent in
the NK2 binding assay (pIC50 8.9) and functionally in
the RPA screen (pA2 9.0). Encouragingly, 1j like its N-
cyclopropylmethyl analogue 1f displayed encouraging
Caco-2 permeability (A–B 6%/h), despite its relatively
In summary, a detailed analysis of the origin of the poor
human pharmacokinetics observed with UK-224,671 led
to a revised medicinal chemistry design strategy, target-
ing compounds with reduced HBD potential and re-
duced TPSA in this high MW series, to improve
transcellular permeability. Metabolic route profiling of
potent, highly permeable, but metabolically vulnerable
N-cyclohexylmethyl analogue (1h) led to the design of
N-(4,4-difluorocyclohexyl)methyl substitution, to block
CYP3A4-mediated oxidative metabolism, leading to po-
tent compounds with significantly enhanced metabolic
stability.
2
˚
high MW (622 kDa) and TPSA (73 A ).
Exploration of the 3-azetidine substituent SAR led
to the discovery that while HBD groups in this region
conferred potency, the presence of even a single HBD
generally compromised permeability in this series.
4-Amino piperidine (1k) and 4-hydroxypiperidine (1l)
were potent in the RPA assay; however, both suffered
Table 2. Comparative Caco-2 permeability, HLM stability and oral rat PK data for UK-224,671 1a and UK-290,795 1n
Compound
Caco-2 (%/h) (A–B/B–A)
Rat PK (oral absorption) (%)a
HLM (T1/2, min)
Human in vivo clearance (ml/min/kg)
UK-224,671 (1a)
UK-290,795 (1n)
1/18
>35 (ER = 1)
<20b
>80
>120
>120
<3b
—
a UK-224,671 (10 mg/kg), UK-290,795 (20 mg/kg).
b See Ref. 7.

D. S. Middleton et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3957–3961
3961
2. Gale, J. D.; OÕNeill, B. T.; Humphrey, J. M. Exp. Opin.
Ther. Patents 2001, 11, 1354.
3. Sakurada, T.; Sakurada, C.; Tan-No, K.; Kisara, K. CNS
Drugs 1997, 8, 436.
4. Stout, S. C.; Owens, M. J.; Nemeroff, C. B. Ann. Rev.
Pharm. Toxicol. 2001, 41, 877.
5. Rogers, D. F. Exp. Opin. Ther. Patents 2001, 11, 1097, and
references cited therein.
6. (a) Bushfield, M.; Metcalfe, M.; Naylor, A. M. Br. J.
Pharmacol. 1995, 115, 875; (b) MacKenzie, A. R.;
Marchington, A. P.; Middleton, D. S.; Newman, S. D.;
Selway, C. N.; Terrett, N. K. Bioorg. Med. Chem. Lett.
2003, 13, 2211; (c) MacKenzie, A. R.; Marchington, A. P.;
Middleton, D. S.; Newman, S. D.; Jones, B. C. J. Med.
Chem. 2002, 45, 5365.
HBD score generally proved a better predictor of perme-
ability than TPSA values in this series. Poor permeabil-
ity was successfully overcome by incorporation of a 4-
fluoro-substituent (1n) in place of 4-amino- or 4-hy-
droxy-functionality on the piperidine ring. Analogue
1n displayed excellent potency and metabolic stability.
Evaluation of 1n in rat PK showed this compound to
possess greatly enhanced oral absorption relative to
UK-224,671.
Our continuing studies in this area will be the subject of
future communications from these laboratories.
7. Beaumont, K.; Harper, A.; Smith, D. A.; Abel, S.
Xenobiotica 2000, 30, 627.
Acknowledgments
8. Beaumont, K.; Harper, A.; Smith, D. A.; Bennett, J. Eur.
J. Pharm. Sci. 2000, 12, 41.
The authors wish to thank Mr. Kevin Beaumont and
Dr. Mark Bushfield for advice and acknowledge the able
assistance of Mrs. J. Banks, Mrs. D. Davey, S. Broad-
bent and Mrs. M. Stanley for ligand binding and func-
tional assays and Mrs. Joanne Bennett for Caco-2
permeability studies. We also wish to thank C. Howard,
Mr. Chris Maxwell and Mr. Chris Luckhurst for
compound preparation. We also thank the staff of the
Structural and Separation Sciences Department for
analytical and spectroscopic data.
9. MacKenzie, A. R.; Marchington, A. P.; Meadows, S. D.;
Middleton, D. S. US Patent 5 963 923, 1996.
10. MacKenzie, A. R.; Marchington, A. P.; Middleton, D. S.;
Meadows, S. D. US Patent 6 262 075, 2001.
11. Amine (4) was prepared in three steps. Treatment of
cyclopentanone with cyclopropyllithium (prepared
from tert-BuLi and cyclopropyl bromide) gave the
tertiary alcohol, which was then converted to the
azide using TFA and NaN3 in toluene. Reduction to
the primary amine was effected using LiAlH4 and
toluene/THF.
12. Artursson, P.; Palm, K.; Luthman, K. Adv. Drug Delivery
Rev. 1996, 22, 67.
References and notes
13. Lipinski, C. A. Drug Discovery Today: Technologies 2004,
1, 337.
1. Lecci, A. Trends Pharmacol. Sci. 2001, 22, 551.