Amidinohydrazones as guanidine bioisosteres: Application to a new class of potent, selective and orally bioavailable, non-amide-based small-molecule thrombin inhibitors

Article abstract of DOI:10.1016/S0960-894X(99)00632-0

We describe a new class of potent, non-amide-based small molecule thrombin inhibitors in which an amidinohydrazone is used as a guanidine bioisostere on a non-peptide scaffold. Compound 4 exhibits nM inhibition of thrombin, is selective for thrombin, and

Full text of DOI:10.1016/S0960-894X(99)00632-0

Bioorganic & Medicinal Chemistry Letters 10 (2000) 1±4
Amidinohydrazones as Guanidine Bioisosteres: Application to a
New Class of Potent, Selective and Orally Bioavailable,
Non-amide-Based Small-Molecule Thrombin Inhibitors
Richard M. Soll,* Tianobao Lu, Bruce Tomczuk, Carl R. Illig, Cynthia Fedde,
Stephen Eisennagel, Roger Bone, Larry Murphy, John Spurlino and
F. Raymond Salemme
3-Dimensional Pharmaceuticals, Eagleview Corporate Center, Exton, PA 19341, USA
Received 11 October 1999; accepted 2 November 1999
AbstractÐWe describe a new class of potent, non-amide-based small molecule thrombin inhibitors in which an amidinohydrazone
is used as a guanidine bioisostere on a non-peptide sca€old. Compound 4 exhibits nM inhibition of thrombin, is selective for
thrombin, and shows 60 and 23% bioavailability in rabbits and dogs, respectively. Crystallographic analysis of 4 bound to throm-
bin con®rmed the amindinohydrazone binding mode. # 1999 Published by Elsevier Science Ltd. All rights reserved.
A key strategy in new antithrombotic therapy has been
directed towards the discovery of small molecule inhi-
bitors of the coagulation cascade, in particular the ser-
ine protease thrombin, an enzyme which plays a pivotal
role in both ®brin generation and platelet activation.1
Recent years have witnessed a number of important
developments with respect to the discovery of potent
and selective inhibitors possessing improved pharma-
cokinetic characteristics in comparison to preceding
generations of such agents.2,3 However, for long-term
anticoagulation, the search for direct antithrombins
with good oral bioavailability and long half-life still
remains a signi®cant challenge.
of the series since it has been appreciated that the
antithrombotic performance of thrombin inhibitors is
dependent inter alia upon a proper balancing between
oral absorption, volume of distribution, and protein
binding, all of which may be linked to the lipophilicity
of the molecule.7
We report here further structure-based elaboration of
this series through replacement of the simple guanidine
grouping with an amidinohydrazone motif. This inter-
change was inspired by the oral activity of the guanidino-
containing a2-agonist guanabenz (3),8 the depressed
pKa guanabenz (8.1),9 and the potential for maintaining
key hydrogen-bonding interactions with Asp-189 of
thrombin's S1 binding pocket. This novel replacement
has led to a potent, selective, and orally bioavailable,
non-amide-based thrombin inhibitor series.
In previous papers we evaluated a novel series of achiral
thrombin inhibitors exempli®ed by 1 and 2.4±6 These
compounds exhibited potent and selective inhibition of
thrombin and are distinguished by the structural sim-
plicity of the series. However, proper in vivo evaluation
of the most interesting compounds, such as 2 (Ki
4.6 nM), was limited by poor solubility. Further, it was
desirable to diversify the properties of this series in two
critical ways through (a) modulation of the pKa of the
highly basic guanidino grouping given the poor in
vivo pro®les of known guanidino-containing thrombin
inhibitors and (b) variance of the overall lipophilicity
Compounds 4±9 were prepared as follows. Orcinol was
®rst derivatized to monosulfonate 21 using the appro-
priate sulfonyl chloride (ca. 1 equiv) for 4±5, 7±11, 13±
15, 18 and 19 in a biphasic mixture of saturated bicar-
bonate and Et2O (26±76% yield). For 6, 5-chloro-2-
methoxybenzenesulfonyl chloride was used (satd
NaHCO3, THF, n-Bu2O (50:38:12)) followed by cata-
lytic hydrogenation (H2 (1 atm), 10% Pd/C, MeOH) to
provide 21 in 63% yield. Monosulfonates 21 were con-
verted in the presence of excess diol to 23 (3±4 equiv) by
(a) standard Mitsunobu10 chemistry (PPh3 (1.5 equiv)/
DEAD (1.5 equiv)) for 14 and 19 or (b) Mitsunobu
*Corresponding author. Tel.: +1-610-458-6056; fax: +1-610-458-
8249; e-mail: soll@3dp.com
0960-894X/00/$ - see front matter # 1999 Published by Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(99)00632-0

2
R. M. Soll et al. / Bioorg. Med. Chem. Lett. 10 (2000) 1±4
modi®cation11 for compounds 5±13, 15 and 18 (PBu3
(1.5 equiv)/ADDP (1.5 equiv) (39±100% yield)). In the
case of 4, compound 22 was ®rst prepared using mono-
benzylated propanediol in the Mitsunobu reaction and
then deprotected to 23 (H2 (1 atm), 10% Pd/C, MeOH
(71% yield)). Intermediate 23 was oxidized (SO3 pyri-
dine complex/DMSO/DIEA/CH2Cl2)12 typically at 0 ꢀC
to ambient temperature (47±100% isolated yields); the
aldehydes for quinoline and pyridine analogues 11 and
14 were advanced into the next reaction without pur-
i®cation. Conversion of 24 into the corresponding ami-
dinohydrazones was accomplished with aminoguanidine
nitrate. Compounds 5, 7, 8, 12, 13, 15, 16, 18 and 19
were isolated as nitrate salts (2 equiv in ethanol; 58±
98% yields). Compounds 4 and 10 were isolated as HCl
salts after HCl treatment of the free base. Compounds
9, 11, 13 and 20 were isolated as HOAc salts after
HOAc treatment of the free base and/or chromato-
graphy through a silica gel SPE column using CH2Cl2/
MeOH/HOAc mixtures as eluants.
Ki's trypsin: 7 (59 mM), 8 (17 mM), 15 (23 mM), 13
(45 mM), 16 (14.8 mM), 19 (67 mM) and 20 (17 mM).
Exchanging the guanidino group of 1 (Ki 13 nM) with
an amidinohydrazone (4; Ki 8.3 nM) produced a potent
thrombin inhibitor. Critical for optimal potency is the
amidinohydrazone side chain length of 3 carbon atoms.
Reduction of the side chain from 3 carbons to 2 carbons
or elongation to 4 carbons signi®cantly decreased
potency (6 versus 18 and 19). An attempt to ®ll the
hydrophobic S1 pocket with a cyclopropyl ring resulted
in 4-fold less activity (10 versus 20), presumably due to
steric clashes with the protein.
The sulfonate group, which provides a rigid, directional
link from the central orcinol sca€old to thrombin's dis-
tal aromatic binding pocket contributes to the potency
of this series (11 versus 16). Replacing the phenolic ether
linkage with a carbon atom resulted in a pronounced
decrease in potency, possibly due to an increase in con-
formational ¯exibility and side chain conformational
change (4 versus 17).
Ether analogue 16 was prepared in 4 steps: (a) benzyla-
tion of orcinol (NaH, DMF then 2-tri¯uorobenzyl
chloride (41% yield)), (b) coupling with 1,3-propanediol
under Mitsunobu conditions10 (85% yield), (c) oxida-
tion11 (85% yield), and (d) amidinohydrazone formation
(77% yield; nitrate salt). Analogue 17 was prepared in
six steps from 3,5-dihydroxybenzaldehyde: (a) MeOH,
PPh3, DEAD, THF, 0 ꢀC (25%), (b) ole®nation with 26
(LDA in cyclohexane, THF, 78ꢀC (28% yield)), which
in turn was prepared from bromide 25 (PPh3, CH3CN,
re¯ux, 2 d), (c) catalytic reduction (H2, 10% Pd/C, THF)
(d) sulfonylation (2-Cl-PhSO2Cl, DIEA, cat. DMAP,
CH2Cl2 (77% yield for 2 steps), (e) oxidation12 (15%
yield) and (f) amidinohydrazone formation (obtained as
HCl salt after HCl treatment of free base; 75% yield).
Analysis of compounds 1 and 2, crystallographically
bound to thrombin, had shown that the projection of
the substituent ortho to the sulfonyl group in the aryl
binding pocket was directed towards solvent4 thereby
presenting additional opportunities for property diver-
si®cation. In the current series, a wide range of pendant
group functionality spanning electron donating/with-
drawing and hydrophilic/hydrophobic moieties is
allowed and produced a host of compounds with Ki's
substantially less than 50 nM. In addition, the 5-chlor-
othienyl (12; Ki 9.2 nM) and 8-quinolinyl analogue (15;
Ki 4.7 nM) groups are well accommodated in the aryl
binding pocket, but less potent are the 1-naphthyl ana-
logue (13; Ki 44 nM) and the hydrophilic 3-pyridyl
analogue (14; Ki 36 nM).
The compounds of the present study were evaluated for
thrombin inhibition and cross-screened for selectivity
against a panel of the serine proteases such as chymo-
trypsin, elastase, plasmin, and trypsin.4 The thrombin
inhibition data is presented in Table 1. Counter-screen-
ing concentrations varied between 1 and 27 mM and for
those compounds exhibiting activity at screening con-
centrations, Ki's were determined. Only trypsin and/or
chymotrypsin activity were observed at high concentra-
tions (Ki's chymotrypsin: 7 (88 mM), and 17 (13.7 mM);
Further insights were provided by comparison of 4 and
1 crystallographically bound to thrombin as shown in
the overlays in Figures 1 and 2.13 The interactions of the
S1 pocket are maintained by the amidinohydrazone
group, however the hydrogen bond is now more evenly
split between the terminal nitrogens than was the case in
the guanidino substituent. The increased planarity of
the amidinohydrazone group leads to a slight overall
displacement of the remainder of the molecule such that

R. M. Soll et al. / Bioorg. Med. Chem. Lett. 10 (2000) 1±4
Table 1. Inhibition of thrombin by amidinohydrazones 4±20
3
Structure
Thrombin Ki (nM)
1
4
13‹1.6
8.3‹2.5
5
6
9.4‹0.67
11‹9.9
Figure 1. Overlap of crystal structures of compounds 4 (gray) and 1
(magenta) bound to thrombin; surface view.
7
8
9
11‹1.5
6.0‹3.0
20‹7.6
10
11
12
9.1‹4.8
4.4‹2.5
9.2‹5.0
13
44‹3.8
Figure 2. Overlap of crystal structures of compounds 4 (gray) and 1
(magenta) bound to thrombin; focus on S1 pocket.
14
15
36‹10
edge-to-face interaction of 4 to tryptophan 215 of the
distal aromatic binding pocket has increased (3.5 A) in
comparison to 1 (3.2 A). Thus, the nearly equivalent
Ki's for the two compounds suggest that the loss in
positive van der Waal's packing energy due to the dis-
placement of the P2 and aryl binding groups is more
than compensated for by the loss of rotational freedom
due to the amidinohydrazone group.
4.7‹1.8
16
17
420‹220
990‹240
Pertinent physical properties of 4 and 10 along with
pharmacokinetic and pharmacodynamic characteristics
(evaluated in dogs) are summarized in Table 2. The
amidinohydrazone motif substantially reduced the pKa
of a guanidino group.14 Compounds 4 and 10 also
possess more ``drug-like'' lipophilicities,15 which may be
re¯ected in the oral bioavailability of these compounds
(23 and 21% for 4 and 10, respectively, as determined
using doses of 6 mg/kg i.v. and 30 mg/kg p.o. in 20%
2-hydroxypropyl-b-cyclodextrin formulations). The
18
19
20
>1000
1300‹60
40‹3.7

4
R. M. Soll et al. / Bioorg. Med. Chem. Lett. 10 (2000) 1±4
Table 2. Physical and pharmacokinetic characteristics in dogs of 4 and 10
Compound
pKa
log P
t1/2 (h) dog plasma
F (%)
Cmax (oral) (mM)
TCTa (s)
Cl (mL/min/kg)
Vss (K/kg)
4
10
8.71
8.84
2.34
2.11
21
19
23
21
1.8
0.9
34‹3a
101‹23b
39
79
3.5
23
aBaseline TCTs: a21‹2; b24‹1.
In summary, the present work demonstrates that the
amidinohydrazone motif is an e€ective replacement of
the highly basic guanidino functionality. Key hydrogen-
bonding interactions of the guanidine are preserved
while o€ering opportunities for improving overall in
vivo anticoagulant properties and performance through
reduction of pKa. Additional evaluation, diversi®cation
and optimization of this novel class of achiral, non-
amide-based thrombin inhibitors are in progress and
will be reported in due course. The results and observa-
tions of the present work are broadly applicable to
other potential targets involving such basic group inter-
actions and are under active investigation.
References and Notes
1. Chi, L.; Rogers, K. L.; Uprichard, A. C. G.; Gallagher, K.
Exp. Opin. Invest. Drugs 1997, 6, 1591.
2. Wiley, M. R.; Fisher, M. J. Exp. Opin. Ther. Patents 1997,
7, 1265.
3. Vacca, J. P. Annu. Rep. Med. Chem. 1998, 33, 81.
4. Lu, T.; Tomczuk, B.; Illig, C. R.; Bone, R.; Murphy, L.;
Spurlino, J.; Salemme, F. R.; Soll, R. M. Bioorg. Med. Chem.
Lett. 1998, 8, 1595.
Figure 3. Bioavailability of compound 4 in rabbits.
5. Illig, C.; Eisennagel, S.; Bone, R.; Radzicka, A.; Murphy,
L.; Randle, T.; Spurlino, J.; Jaeger, E.; Salemme, F. R.; Soll,
R. M. Med. Chem. Res. 1998, 8, 244.
6. Lu, T.; Soll, R. M.; Illig, C. R.; Bone, R.; Murphy, L.;
Spurlino, J.; Salemme, F. R.; Tomczuk, B. E. Bioorg. Med.
Chem. Lett. 1999, 10, 79.
7. Tucker, T. J.; Lumma, W. C.; Lewis, S. D.; Gardell, S. J.;
Luccas, B. J.; Baskin, E. P.; Woltmann, R.; Lynch, J. J.; Lyle,
E. A.; Appleby, S. D.; Chen, I. W.; Dancheck, K. B.; Vacca, J.
P. J. Med. Chem. 1998, 40, 1565.
8. Wendt, R. L., In Pharmacology of Antihypertension Drugs,
Scriabine A., Ed.; Raven: New York, 1980, pp. 99±111.
9. Shearer, C. M. Anal. Pro®les of Drug Substances 1986, 15,
319.
stability of the compounds in dog plasma were com-
parable (t1/2 21 h and t1/2 19 h, respectively). Pharma-
codynamically, 4 e€ected a maximal 1.6-fold increase in
TCT at 2 h (30 mg/kg, p.o.). Compound 10 maximally
prolonged clotting time 4-fold at 1 h (30 mg/kg, p.o.)
and produced measurable anticoagulation at 8 and 24 h
(1.6- and 1.2-fold, respectively). Given the high clear-
ance and wide distribution of these compounds, both
compounds produced only marginal anticoagulation
e€ects as measured by TCT prolongation.
10. Mitsunobu, O. Synthesis 1981, 1.
11. Tsunoda, T.; Yamamiya, Y.; Ito, S. Tetrahedron Lett.
1993, 34, 1639.
12. Godfrey, J. D.; Gordon, E. M.; Von Langen, D. J. Tetra-
hedron Lett. 1987, 28, 1603.
13. Crystallographic data for compound 4: R=18.3% at 1.9 A
resolution.
14. pKa of l-arginine guanidine: 13.2 The Merck Index; 10th
ed.; Windholz, M. Ed.; Merck & Co., 1983, p. 797.
15. Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P.
Adv. Drug Delivery Rev. 1997, 23, 3.
Compound 4 was further evaluated for bioavailability
in rabbits by measuring plasma concentration of drug
substance using doses of 6 mg/kg i.v. and 30 mg/kg p.o.
in rabbits (Fig. 3). Compound 4 was highly orally bio-
available in rabbits (F 60%). Micromolar level of com-
pound was observed at 8 h (termination of experiment).
Pharmacodynamically, thrombin clotting times (TCTs)
were prolonged to 1.9 over controls (50‹6.4 s at 60
min versus 26‹4.8 s baseline level) at 30 mg/kg p.o.