Oxygenated analogues of UK-396082 as inhibitors of activated thrombin activatable fibrinolysis inhibitor

Article abstract of DOI:10.1016/j.bmcl.2009.11.029

A suitable inhibitor of activated thrombin activatable fibrinolysis inhibitor (TAFIa) has the potential to be a novel treatment for thrombosis. The TAFIa inhibitor UK-396082 (1) was used as a starting point to seek more potent analogues. With knowledge of encouraging human pharmacokinetics and toleration for the clinical candidate (1), the programme continued to seek structure-activity relationships (SAR) that could positively impact on both potency and half-life, and therefore the projected dose of any future nominated clinical agent. A series of oxygenated analogues based on compound 1 were prepared to evaluate changes in pharmacology, selectivity and pharmacokinetics.

Full text of DOI:10.1016/j.bmcl.2009.11.029

Bioorganic & Medicinal Chemistry Letters 20 (2010) 92–96
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Oxygenated analogues of UK-396082 as inhibitors of activated thrombin
activatable fibrinolysis inhibitor
a
a
a
b
Dafydd R. Owen a, , David J. Bull , Mark E. Bunnage , Melanie S. Glossop , Robert J. Maguire ,
*
Ross S. Strang a,
a Pfizer Global Research and Development, Sandwich Laboratories, Ramsgate Road, Sandwich CT13 9NJ, UK
b Pfizer Global Research and Development, Groton Laboratories, Groton, CT 06340, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
A suitable inhibitor of activated thrombin activatable fibrinolysis inhibitor (TAFIa) has the potential to be
a novel treatment for thrombosis. The TAFIa inhibitor UK-396082 (1) was used as a starting point to seek
more potent analogues. With knowledge of encouraging human pharmacokinetics and toleration for the
clinical candidate (1), the programme continued to seek structure–activity relationships (SAR) that could
positively impact on both potency and half-life, and therefore the projected dose of any future nominated
clinical agent. A series of oxygenated analogues based on compound 1 were prepared to evaluate changes
in pharmacology, selectivity and pharmacokinetics.
Received 30 September 2009
Revised 6 November 2009
Accepted 9 November 2009
Available online 13 November 2009
Keywords:
Thrombin activatable fibrinolysis inhibitor
TAFI
Ó 2009 Elsevier Ltd. All rights reserved.
Carboxypeptidase U
Carboxypeptidase inhibitors
Thrombosis is defined as the obstruction of a blood vessel
through the formation of a clot (thrombus). Given risk factors for
thrombosis such as diet, exercise and smoking (as opposed to ge-
netic or natural imbalances), it is not surprising that thrombotic
diseases represent one of the most common causes of mortality
and morbidity in the developed world. Thrombin activatable fibri-
nolysis inhibitor (TAFI) is an inactive 60-kDa glycoprotein zymo-
gen found in human plasma.1 Through the action of the enzyme
thrombin (from the coagulation cascade), TAFI is converted from
its inactive form to the activated TAFIa. Activated TAFIa is an unsta-
ble, zinc dependent basic carboxypeptidase that significantly
inhibits fibrinolysis. Inhibition of activated TAFI represents a
potentially subtle modulation of the clotting and lysis balance,
without resorting to direct action on the coagulation cascade
(as traditionally tackled by inhibitors of thrombin or factor Xa).
Ideally, this subtlety may imply a reduced risk of side effects when
using a TAFIa inhibitor, in comparison to other potential mecha-
nisms for treating thrombotic disease. The role of the zinc metallo-
protease TAFIa in coagulation is to remove C-terminal lysine
residues from the surface of fibrin—a key clot constituent. This
reduction in C-terminal lysine concentrations has the downstream
effect of reducing the local concentration of plasmin at the site of
the clot. Plasmin is anticoagulant in its properties. The action of
TAFIa serves to limit plasmin recruitment at the site of the forming
thrombus, thus an inhibitor of TAFIa should be antithrombotic.
The TAFIa inhibitor UK-396082 1 (Fig. 1) has previously been
disclosed as a clinical candidate from these laboratories.2 It was
rationally designed as a C-terminal lysine mimetic and unsurpris-
ingly possesses structural elements common to that amino acid
(position of the side-chain basic centre, free carboxylic acid).
The imidazole serves as a rare example of the functional group
being used as a ligand for zinc in a metalloprotease inhibitor.3 The
zinc ion is a key component of the enzyme in its role during pep-
tide cleavage and hence small molecule inhibitors of TAFIa are
likely to feature some sort of zinc ligating functionality.4
The attractive characteristics of compound 1 are its small size
and polarity. Alongside the pharmacological design requirements
of seeking to mimic a C-terminal lysine, we deliberately sought
to stay within low molecular weight, polar drug space. This strat-
TAFIa Ki = 16nM
Ligand Efficiency (LE) 0.62
NH2
MWt = 239
N
OH
N
i.v. PK Cl(ml/min/kg) Vd(L/kg) T1/2 (h)
O
Human
Dog
1.5
3.1
0.5
0.35
1.9
4
1.4
2.1
UK-396082
Rat
10.2
* Corresponding author. Tel.: +44 1304648445.
E-mail address: dafydd.owen@pfizer.com (D.R. Owen).
Present address: Novartis AG, Basel, CH-4002, Switzerland.
1
Figure 1. UK-396082 (1) profile.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.11.029

D. R. Owen et al. / Bioorg. Med. Chem. Lett. 20 (2010) 92–96
93
egy successfully translated to paracellular oral absorbtion, passive
R
N
O
O
renal clearance and limited any hepatic metabolic processes for
compound 1. This drug design space has further attractions. The
physicochemical properties of the compound also limit the likeli-
hood of off-target pharmacology, hERG activity and potentially,
compound-mediated toxicity.5 Compound 1 had no plasma protein
binding and a volume of distribution around that of body water in
vivo. Given this chemotype physicochemistry, design hypotheses
beyond 1 were limited to improving TAFIa inhibition and com-
pound pharmacokinetics. Small changes in either parameter would
likely have a significant impact on any future candidate’s predicted
dose.
i.
N
N
PMB
N
PMB
R
O
N
OH
O
ii.
N
O
R
N
R
N
iii.
O
O
N
N
N
N
PMB
PMB
O
O
The ligand efficiency of compound 1 is notably high (0.62).6 De-
spite ‘only’ being 16 nM in its primary pharmacology,7 this is
achieved using just seventeen heavy atoms. However, with such
an efficient lead and some molecular weight in hand before the
small, polar drug space characteristics were compromised, there
appeared to be sufficient flexibility in as yet, unexplored design
space for the template. The hypotheses investigated in this Letter
are around the effects of changes in the basic side-chain and the
imidazole N-1 substituent of 1. We initially sought to explore the
effect of an oxygen linking atom through compound 2 (Fig. 2).
Compound 1 was initially discovered by moving from an N-linked
basic side-chain (3) to a C-linked alternative (Fig. 2).2 This change
had achieved the vital 10-fold increase in potency, making the
nomination of 1 as a clinical candidate possible.
Undeterred by the apparently negative effect of a heteroatom at
this position in the side-chain (based on our knowledge of the SAR
relating to nitrogen), compound 2 was synthesised by the route
shown in Scheme 1. The general synthetic strategy allowed for a
variety of imidazole carboxaldehydes to be condensed in an al-
dol/dehydration sequence with a protected morpholinone enolate.
Alkene reduction, deprotection and ring-opening liberated the eth-
anolamine side-chain and carboxylic acid. Single enantiomers were
prepared by chiral HPLC separation of intermediate morpholinone
racemates, before the deprotection/ring-opening sequence.
This O-linked change covered several sub-hypotheses. Lacking
the H-bond donor of the previously studied secondary amine (3),
oxygen was thought to be more carbon-like in its properties. In
contrast to carbon, oxygen did offer the potential for intramolecu-
lar hydrogen bonding to the terminal primary amine through act-
ing as a weak acceptor, thus bringing in the potential for a
conformationally-driven optimisation of potency. It also had a role
to play in moderating the pKa of local ionisable groups, primarily
the carboxylic acid and the side-chain primary amine. In placing
the oxygen as a heteroatom with a b-relationship to the primary
amine, the pKa of this group could be reduced by two units. Equally
iv.
NH2
OH
R
N
R
N
v., vi.
O
O
N
N
N
PMB
O
O
Scheme 1. Reagents and conditions: (i) LDA, THF, À78 °C; (ii) MsCl, Et3N, THF, 0–
40 °C; (iii) 10% Pd/C, 50 psi, EtOH; (iv) Chiral HPLC; (v) CAN, CH3CN/H2O; (vi) 6 N
HCl, 110 °C.
group more acidic. This oxygenated analogue (2) was therefore cal-
culated to be less basic and more acidic, but with a similar d-pKa
between the zwitterionic functionality, when compared to 1.8
The oxygenated analogue 2 proved to be 2.5 times more potent
than the equivalent methylene containing lead compound 1. This
is clearly reflected in its improved ligand efficiency as the potency
enhancement has cost no extra heavy atoms. With a molecular
weight of only 241 used in attaining a 6.7 nM compound, it was felt
that up to another 60 mass units could be used in seeking out fur-
ther potency enhancement without compromising the small, polar
status of the series and the potential for paracellular oral
absorbtion.
Phenylation of the N-1 position of the imidazole (4) did achieve
a further potency enhancement when replacing the n-propyl group
of 2. This proved to be the most potent compound made in the
whole oxygenated series at 3.5 nM (Fig. 3). Addition of further lipo-
philicity at the para position of the phenyl group in compound 5
maintained, but did not improve potency.
Cyclic (6), or homologated (7) aliphatic lipophilicity at the imid-
azole N-1 added nothing over the initial prototype 2. Changing the
torsional angle of the phenyl-imidazole bond through an ortho-
substituent as in compound 8 cost around 10-fold in activity over
compound 4. In the case of 6, as with arylated examples 4 and 5,
the structural changes only served to erode selectivity over human
pancreatic carboxypeptidase B (Human pCPB). This was the only
having a heteroatom-
a
to a carboxylic acid should render the
Human
CPB
(nM)
Amine
pKa
Acid
pKa
TAFIa
δ-pKa
Template
Entry
X
CH2
O
LE
Ki (nM) 7
(calc.) 8
(calc.) 8 (calc.) 8
1
2
16
0.62
0.65
206
238
10.7
8.9
3.9
2.2
1.5
6.8
6.7
8.3
NH2
N
6.7
X
OH
N
O
3
NH
150
0.55
1160
9.8
Figure 2. The effect of O and NH insertion into compound 1 (UK-396082).

94
D. R. Owen et al. / Bioorg. Med. Chem. Lett. 20 (2010) 92–96
Human
pCPB
Ki (nM)
TAFIa
Ligand
Efficiency
Configuration
*
Compound
Entry
MWt
275
6
Ki (nM) 7
NH2
OH
N
O
4
3.5
5.7
0.58
0.47
27
49
S
S
N
*
O
NH2
5
331
N
O
OH
N
*
O
NH2
N
6
7
8
9
8.2
15
0.50
0.59
0.47
0.53
27
360
2720
265
S
+/-
+/-
S
309
255
289
290
O
OH
N
*
O
O
NH2
OH
N
O
N
*
NH2
N
65
O
OH
N
*
O
O
O
N
N
NH2
OH
5.8
O
N
*
NH2
OH
N
10
11
12
7.5
407
0.53
0.48
-
240
S
+/-
R
289
255
241
O
N
*
H
N
N
O
>10000
>10000
OH
N
*
O
O
NH2
OH
N
>1000
O
N
*
Figure 3. Structure–activity relationships within the series.
selectivity monitored for the series as Cerep BioPrintÒ screening
had indicated no other selectivity issues for this polar chemotype.
The loss of Human pCBP selectivity could be countered relatively
efficiently through the stereo-defined positioning of a methyl
group in the basic ethanolamine side-chain. Compounds 9 and 10
illustrate the point with TAFIa inhibition maintained at sub-
10 nM while disfavouring human pCPB activity. The methyl group
was responsible for improving selectivity over human pCPB >30-
fold. The toleration of a pyridyl group in compound 9 is also of note
as a more polar isostere of the phenyl N-1 substituent in 4. Changes

D. R. Owen et al. / Bioorg. Med. Chem. Lett. 20 (2010) 92–96
95
Dose
Cl
Vd
T1/2
Structure
Entry
Study
ppb
0%
Foral
-
(mg/kg)
(ml/min/kg)
(L/kg)
(h)
NH2
Rat i.v.
2
13.4
3.1
2.7
N
2
O
OH
N
Rat oral
Rat i.v.
Rat oral
Hu. i.v.
5
2
-
10.2
-
-
2.1
2.1
2.4
4
9.8%
O
1.9
-
-
10.6%
-
NH2
OH
N
1
0%
5
N
O
10
1.5
0.5
Figure 4. Pharmacokinetic comparison of compounds 1 and 2.
to the primary amine at the side-chain terminus were costly. A
simple methylation reduced activity almost 100-fold. Absolute
stereochemistry was also important as seen in compound 12. This
enantiomer of the leading oxygenated analogue 2 was completely
inactive against TAFIa.
The somewhat surprising conclusion to this SAR study was that
we were unable to significantly enhance potency beyond 6 nM
despite having up to six lipophilic heavy atoms to use, before
compromising our small, polar drug properties. On assessing the
optimal combination of potency, selectivity and the potential
for good oral bioavailability/pharmacokinetics, the programme
decided to profile compound 2 further.
Compound 2 was assessed in both iv and oral rat pharmacoki-
netic studies ( Fig. 4). In the iv study (2 mg/kg), the compound
had a relatively low clearance of 13.4 ml/min/kg and had no plas-
ma protein binding. The clearance was higher than expected. It was
three times higher than the glomerular filtration rate for rat that
would have indicated passive renal clearance for the compound.
However, the volume of distribution of 3.1 l/kg was also higher
than expected for such a small, polar zwitterion. The rat iv half-life
was 2.7 h. Oral pharmacokinetic studies at a dose of 5 mg/kg re-
vealed an oral bioavailability of ꢀ10% for compound 2. This overall
profile compared favourably with the equivalent rat data for clini-
cal candidate UK-396082 (1) which had a lower volume of distri-
bution, lower clearance and an iv half-life of 2.1 h.
error for those assays and models. Small changes in pharmacoki-
netics and primary pharmacology would have a significant effect
on clinical dose for these particular compounds. These would likely
manifest themselves in further clinical trials. Surprisingly, efforts
to enhance potency through addition of lipophilicity at the N-1 po-
sition of the imidazole made little impact, although addition of a
methyl group in the side-chain proved to be a good strategy for
regaining human CPB selectivity when required. These compounds
represent some of the most potent and selective TAFIa inhibitors
disclosed to date.4 In conjunction with reliable rat oral bioavail-
ability and a useful rodent half-life, compound 2 has significant po-
tential as a pre-clinical tool for further understanding the utility of
TAFIa inhibition as a mechanism for treating thrombosis or for
other potential emerging uses in non-cardiovascular indications.10
References and notes
1. Nesheim, M.; Bajzar, L. Thromb. Haemost. 2005, 3, 2139.
2. Bunnage, M. E.; Blagg, J.; Steele, J.; Owen, D. R.; Allerton, C.; McElroy, A. B.;
Miller, D.; Ringer, T.; Butcher, K.; Beaumont, K.; Evans, K.; Gray, A. J.; Holland, S.
J.; Feeder, N.; Moore, R. S.; Brown, D. G. J. Med. Chem. 2007, 50, 6095.
3. Lee, K. J.; Joo, K. C.; Kim, E.-J.; Lee, M.; Kim, D. H. Bioorg. Med. Chem. 1997, 5,
1989.
4. Bunnage, M. E.; Owen, D. R. Curr. Opin. Drug Discovery Dev. 2008, 11, 480.
5. Price, D. A.; Blagg, J.; Jones, L.; Greene, N.; Wager, T. Exp. Opin. Drug Met. Toxicol.
2009, 5, 921.
6. Hopkins, A. L.; Groom, C. R.; Alex, A. Drug Discovery Today 2004, 9, 430.
7. Details of assay for TAFla inhibition (i) TAFI activation: Human TAFI
The carbon to oxygen switch in going from 1 to 2 had been
designed to potentially moderate the primary amine basicity and
alter the carboxylic acid pKa. This could potentially affect the com-
pound’s half-life through small changes in volume of distribution,
but with minimal impact on clearance and oral bioavailability.
Although exemplified in a very different chemotype, the ethanol-
amine side-chain has previously been employed by these laborato-
ries to extend compound half-life in the discovery of the calcium
channel antagonist amlodipine.9 From the rat iv pharmacokinetic
data, compound 2 has the potential for a comparable human
half-life to C-linked compound 1. Given the chemotype physico-
chemistry, dramatic changes in half-life are unlikely through the
structural changes made (compounds have a low volume of distri-
bution, low clearance and no plasma protein binding). The mea-
sured change in the rat pharmacokinetic data for compound 2
compared to compound 1 falls within experimental error.
In conclusion, compound 2 has a comparable pre-clinical profile
to previously disclosed clinical candidate 1. The apparent small
improvements made in primary pharmacology and pharmacoki-
netics through the design of compound 2 fall within experimental
(recombinant or purified) was activated by incubating 20
(360 g/ml) with 10 l of human thrombin (10NIH units/ml), 10
thrombomodulin (30 g/ml), 6 l calcium chloride (50 mM) in 50 l
HEPES (N-10 [2-hydroxyethyl]piperazine-N-[2-ethanesulfonic acid]) buffer
containing 150 mM sodium chloride and 0.01% TWEEN 80 (polyoxyethylene-
sorbitan monooleate) pH 7.6 for 20 min at 22 °C. At the end of the incubation
l
l of stock solution
l of rabbit
l of 20 mM
l
l
l
l
l
period, thrombin was neutralised by the addition of 10
Arg chloromethyl ketone) (100 nM). The resulting TAFla solution was stored on
ice for 5 min and finally diluted with 175 of HEPES buffer. (ii) Ki
determination (TAFla): A number of different dilutions of the test compound
in water were made up. To 20 l of each dilution were added 150 l of HEPES
buffer and 10 l of TAFla, which was then pre-incubated for 15 min at 24 °C. To
each dilution was then added 20 l furylacryloyl-alanyl-lysine (FAAL) at a
ll of PPACK (D Phe-Pro-
l
l
l
l
l
l
standard concentration. Substrate turnover was measured by reading the
absorbance of the reaction mixture at 330 nm every 15 s for 30 min. The
reaction was performed at 24 °C and samples were mixed for 3 s prior to each
absorbance reading.
A graph of %inhibition against test compound
concentration was then plotted; from which was calculated the IC50 value.
The Ki value was then calculated using the Cheng–Prusoff equation. Two
controls, positive and negative, were used to check the accuracy of the results
in each case. For the first control, the assay was performed as above, but with
20 ll of water rather than a dilution of the test compound. This showed
minimal inhibition. For the second control, the assay was performed as above,
but with an effective amount of a non-specific carboxypeptidase inhibitor
rather than a dilution of the test compound. This showed maximal inhibition.
When the two controls did not demonstrate minimal and maximal inhibition

96
D. R. Owen et al. / Bioorg. Med. Chem. Lett. 20 (2010) 92–96
respectively, the results were discounted and the test compound was
substituents, where the accuracy is usually better than 0.5 pKa units. In order
to achieve this accuracy, ACD/pKa DB uses its own internal databases and
algorithms. For structures with important through conjugation, sometimes
larger differences from experimental values are observed. Each calculation is
supported by its 95% confidence limits and a detailed report of how it has
been carried out, including Hammett-type equation(s), substituent constants,
and the literature references where available.
reanalysed. Using the above assay, the compounds of the examples
illustrated in the Letter were found to be potent and selective inhibitors of
TAFla.
8. Calculated pKa’s quoted in the Letter were derived from
a commercially
available software package. ACD/pKa DB is a programme that calculates accurate
acid-base ionisation constants (pKa) at 25 °C and zero ionic strength in aqueous
solutions for organic structures. The accuracy of calculations is usually better
than 0.2 pKa units except for very complex structures or poorly-characterised
9. Smith, D. A.; Jones, B. C.; Walker, D. K. Med. Chem. Rev. 1996, 16, 243.
10. Bajzar, L.; Jain, N.; Wang, P.; Walker, J. B. Crit. Care Med. 2004, 32, S320.