Discovery of benzothiazole guanidines as novel inhibitors of thrombin and trypsin IV

Article abstract of DOI:10.1016/j.bmcl.2012.05.046

In a project to find novel neutral P1 fragments for the synthesis of thrombin inhibitors with improved pharmacokinetic properties, fragments containing a benzothiazole guanidine scaffold were identified as weak thrombin inhibitors. WaterLOGSY (Water-Ligan

Full text of DOI:10.1016/j.bmcl.2012.05.046

Bioorganic & Medicinal Chemistry Letters 22 (2012) 4839–4843
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of benzothiazole guanidines as novel inhibitors of thrombin
and trypsin IV
⇑
Michael Karle , Wolfgang Knecht, Yafeng Xue
AstraZeneca R&D Mölndal, SE-43183 Mölndal, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
In a project to find novel neutral P1 fragments for the synthesis of thrombin inhibitors with improved
pharmacokinetic properties, fragments containing a benzothiazole guanidine scaffold were identified
as weak thrombin inhibitors. WaterLOGSY (Water-Ligand Observed via Gradient SpectroscopY) NMR
was used to detect fragments binding to thrombin and these fragments were followed up by Biacore
A100 affinity measurements and enzyme assays. A crystal structure of the most potent compound with
thrombin was obtained and revealed an unexpected binding mode as well as the key interactions of the
fragment with the protein. Based on these results, the structure-based design and synthesis of a small
series of optimized novel substituted benzothiazole guanidines with comparatively low pKa values was
accomplished. Testing of these compounds against human trypsin I and human trypsin IV revealed unex-
pected inhibitory activity and selectivity of some of the compounds, making them attractive starting
points for selective trypsin inhibitors.
Received 16 March 2012
Revised 10 May 2012
Accepted 10 May 2012
Available online 17 May 2012
Keywords:
Thrombin
Trypsin IV
Benzothiazole guanidine
WaterLOGSY
Fragment screening
Ó 2012 Elsevier Ltd. All rights reserved.
The control of the blood coagulation process presents one of the
major issues in the treatment of thrombosis (i.e., the formation of a
blood clot in a blood vessel) and the prevention of stroke.1 Warfa-
rin and heparin are still widely used as anticoagulants despite se-
vere side effects and complicated handling. Patients who are
prescribed warfarin have to be monitored very closely during the
whole treatment. Both substances inhibit a variety of factors which
are involved in the blood coagulation cascade. Thrombin as one of
these factors has a key function in platelet aggregation and poly-
merization of fibrin. It is therefore an attractive target for direct
inhibition. Recent examples of direct thrombin inhibitors are hiru-
din, argatroban,2 melagatran3 and dabigatran.4 Early work focused
function forms here a strong bidentate salt bridge with the
Asp189 carboxylate which is the major reason for the affinity to-
wards thrombin. Benzamidine itself has a pKa of 11.6 and thrombin
inhibitors with such a basic P1 group will be mainly protonated
under physiological conditions which makes it unlikely for these
compounds to be absorbed from the GI tract. These substances suf-
fer therefore from low bioavailability. In case of the direct throm-
bin inhibitors melagatran and dabigatran, the amidine function is
masked and the drugs are administered in form of the prodrugs
ximelagatran and dabigatran etexilate which are readily orally ab-
sorbed. Hence, the discovery of novel neutral P1 groups, which
provide adequate potency and selectivity, is central to the synthe-
sis of new direct thrombin inhibitors. Proteases belonging to the
human trypsinogen family show a substantial overlap in substrate
specificity and susceptibility to inhibitors, for example, melaga-
tran, with thrombin.6 A longstanding research interest of ours is
trypsin IV, a serine protease belonging to the human trypsinogen
family, that can cleave protease-activated receptors 1 and 2 to in-
duce inflammation, hyperalgesia and pain.6,7 However, so far lack
of trypsin IV specific small molecular inhibitory compounds hin-
dered evaluation of this potential interesting target in inflamma-
tion and pain. Therefore developing starting points for specific
inhibitors of trypsin IV is of increasing interest, especially those
that would discriminate between trypsin IV and other trypsin-like
proteases. In this work, NMR-based fragment screening8–11 against
thrombin identified benzothiazole guanidine as a lead scaffold
from a library of modified potential P1 fragments. Since the aim
of this project was not the generation of full inhibitors but the
on inhibitors derived from the tripeptide
D-Phe-Pro-Arg which
binds into the thrombin active site with proline fitting into the
S2 pocket and arginine residing in the S1 pocket.5 Typically, direct
thrombin inhibitors possess a P1 group, filling the specific S1 pock-
et. Very often such P1 groups are considerably basic like arginine in
D
-Phe-Pro-Arg, since one of the main interactions can be ascribed
to an interaction of the basic functional group with the carboxylic
acid of the aspartate residue Asp189 sitting deep into the S1 pock-
et. Such P1 groups are generally associated with unfavourable
pharmacokinetic properties and the issues related to these P1
groups cause major problems in inhibitor development so far.
Benzamidine is quite often used as a strong binder in the S1 pocket
and thus as P1 group in thrombin inhibitors. The basic amidine
⇑
Corresponding author. Tel.: +46 31 7761000.
E-mail address: Michael.Karle@astrazeneca.com (M. Karle).
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.05.046

4840
M. Karle et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4839–4843
focus on neutral P1-groups we decided to further examine the po-
tential of this scaffold for fragment generation. A crystal structure
of thrombin with the most potent inhibitor of this set assisted the
structure based design and synthesis of a second set of compounds
containing the benzothiazole guanidine motif, aiming for frag-
ments with lower molecular complexity and lower pKa. Additional,
screening of these compounds against human trypsin I and human
trypsin IV revealed inhibitory activity and selectivity of some of the
compounds, making them attractive starting points for selective
trypsin inhibitors. The pKa of guanidine is very sensitive to substi-
tution. For Oxyguanidines a comparatively low pKa of 7.0–7.5 was
measured regarding the guanidine pKa of 13–14. Thrombin inhibi-
tors with this rather neutral P1 group showed an acceptable per-
meability coefficient in the human Caco-2 monolayer assay.12 In
the case of benzothiazole guanidine, an even lower pKa of 5.7 has
been determined experimentally.13 The absorption potential of
substances with such a P1 group can also be estimated acceptable
regarding the pH range 5.5–7 of the lower GI tract resulting in a
higher proportion of uncharged species. Therefore, incorporation
of the benzothiazole guanidine motif into thrombin and trypsin
inhibitor design may lead to inhibitors with improved pharmacoki-
netic properties and unique specificity.
An interesting approach in the search for less basic or neutral P1
fragments for thrombin inhibitor design is illustrated by the incor-
poration of a p-chlorophenoxy acetamide P1 substituent to afford
potent thrombin inhibitors.14,15 Lacking a basic functionality, there
is no longer a direct interaction between Asp189 and the P1 group.
Instead, H-bonding of the acetamide side chain to Gly219 in the S1
pocket as well as a hydrophobic interaction of the chloro group
with the aromatic system of Tyr 228 located at the bottom of the
S1 pocket seems to be responsible for the potency of such a frag-
ment. Thus, it was hypothesized that attachment of suitable side
chains to fragments fitting into the thrombin S1 pocket could lead
to P1 fragments with increased affinity or selectivity towards
thrombin. Based on this, a set of 14 neutral and weakly basic po-
tential P1 fragments was chosen for the synthesis of a library of
modified fragments. The fragments were alkylated with in total
12 different side chains. After purification, a set of 114 modified
fragments was obtained and the whole library, including the
unmodified fragments, was submitted for WaterLOGSY NMR frag-
ment screening.16,17 WaterLOGSY (Water-Ligand Observed via Gra-
dient SpectroscopY) represents a powerful NMR technique to
detect ligands binding to macromolecules. Bulk water magnetisa-
tion is transferred via bound water molecules in the protein bind-
ing site to the ligand. The method proved to be fast and sensitive
requiring only a small amount of non-deuterated protein and li-
gand. Competition experiments with melagatran and benzamidine
were performed with binding fragments to confirm binding in the
thrombin binding site and the S1 pocket. From our library, four
compounds were identified showing affinity towards thrombin,
whereof three compounds featured the benzothiazole guanidine
scaffold (data not shown). Among those fragments, 1, 2 and 3 with
benzothiazole guanidine as core scaffold (Fig. 1) showed promising
activity and attractive possibilities for the development of further
lead structures. Compared to the unsubstituted benzothiazole gua-
nidine 1, which showed only weak binding to thrombin in the
WaterLOGSY experiment and a 13% inhibition of thrombin enzyme
activity at 5.3 mM compound concentration, attachment of an
ethyl acetate side chain led to a significantly increased potency.
Fragment 3 is slightly less active than 2, nevertheless 3 was shown
to be more potent than the unsubstituted derivative (Table 1). The
primary screening against thrombin using WaterLOGSY NMR
proved to be a fast and robust method to determine binding to
the protein and to confirm binding to the active site as well as to
provide the possibility to rank the compounds according to their
affinity. Nevertheless, KD values calculated from data collected by
WaterLOGSY NMR fragment screening showed a large variation
and this technique was not suitable to obtain quantitative binding
data. Also compound 1 only showed minor inhibition at high con-
centration in a thrombin activity assay. We therefore tested plas-
mon surface resonance technique to determine direct binding
data for all other compounds, which proved to be a much more
reliable method in our case. For 2 and 3, KD values of 95 and
145 lM, respectively, were measured using a Biacore A100 and hu-
man thrombin (Table 1). When these measurements were followed
up with enzymatic assays for thrombin and additionally for tryp-
sins, compound 2 inhibited trypsin I and trypsin IV more than
thrombin, while compound 3 showed certain selectivity for trypsin
IV.
The X-ray structure of 2 with thrombin revealed an unexpected
binding mode and indicates the principal interactions of the frag-
ment with the protein (Fig. 2). The structure has been deposited
with PDB (code 3PO1). Preliminary docking studies on compound
1 predicted the guanidine part to be oriented deep into the S1
pocket, developing the expected interaction with Asp189. Instead,
the whole molecule is flipped 180° with the guanidine part direc-
ted towards the carboxyl group of Glu192. The hydrogen bond dis-
tance here is 2.6 Å and compared to the thrombin crystal structure
with melagatran, the carboxyl function of Glu192 is shifted about
2 Å towards the guanidine function, which reflects a significant
interaction with the ligand. The guanidine group is placed in-be-
tween Glu192 and Gly216, demonstrating also a weak hydrogen
bond to Gly216. With the phenol directed down towards the S1
pocket, the hydroxy function can interact with Asp189 (distance
2.7 Å). A similar interaction has been reported for hydroxy benzo-
thiophene based thrombin inhibitors.18 Further weaker interac-
tions can be seen both between the ester carbonyl and Gly216
and between the ethoxy oxygen and the imidazole ring in His57.
Figure 1. Benzothiazole guanidine fragments 1–7.

M. Karle et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4839–4843
4841
Table 1
Affinity constants and inhibition constants
Compound
Thrombin (Biacore) KD
M)
Thrombin (enzyme assay) Kic
M)
Trypsin I (enzyme assay) Kic
M)
Trypsin IV (enzyme assay) Kic
M)
pKa
(l
(l
(l
(l
1
2
3
4
5
6
7
n.d.
See footnotec
481 58
No effecta
198 47
No effect
647 51
No effect
No effecta
No effect
8.9b
No effecta
168 16
1094 390
476 49
No effect
No effecta
794 107
13.1b
9.6d
5.1d
9.0
5.7
5.5
95 4 (n = 4)
145 46 (n = 4)
No effect
No effect
No effect
2731 2314
No effect
No effect
No effecta
No effect
95 25
5.4
No effect
78 11 (n = 3)
6.0
4-Aminobenzamidine
dihydrochloride
13.1d
The affinity constants for thrombin measured with the Biacore A100 are given as KD SD. The number of measurements is given in brackets. Kic and standard errors were
calculated from IC50 values and their standard errors. The standard errors were reported by the fitting software Sigma Plot 8 from the global IC50 fit to all data.
n.d.—not done.
No effect—no KD or IC50 could be determined.
a
Precipitates at concentrations >1 mM.
Data from [5].
b
c
Compound 1 at 5.3 mM did inhibit 13% of thrombin activity using an alternative less affine thrombin substrate (pyroGlu-Pro-Arg-pNA HCL at 300 lM) instead of 50 lM
S2238.
d
Calculated from ACD/pKa DB.
Figure 3. X-ray structure of 2 in thrombin, electrostatic surface of the protein.
Orientation of the ethyl acetyl side chain in the S2 subpocket.
way to link this fragment with the rest of the inhibitor. Potent P2
fragments which we used as proline replacement in our project
possessed a methyl group pointing also into the above mentioned
hydrophobic subpocket and were unsuitable to be connected with
2 due to an obvious overlap of both molecules. It was therefore
desirable to find novel P1 fragments which allow the P2 fragment
to reach into the subpocket. Introduction of other functional
groups utilizing further interaction alternatives of the ligand with
the S1 pocket was supposed to increase potency or to at least com-
pensate potency loss if the side chain is no longer present. Regard-
ing possible toxic effects of some phenols19 it was also preferable
to avoid the alcohol function. Electron withdrawing substituents
at the benzene ring should lower the pKa. Thus, benzothiazole
guanidines 4, 5 and 6 were designed and synthesized (Fig. 1). To
address the question whether a substitution at the 4-position is
tolerated, the 4-methyl derivative 7 was also synthesized. Besides
attachment at the ring nitrogen, the 4-position is, according to
docking studies, also regarded to be favourable to link P2 groups.
The introduction of the chloro group should examine the possibil-
ity to establish a further hydrophobic interaction, for instance with
the aromatic system of Tyr228. As expected, the measured pKa val-
ues for the guanidine group of 4–7 were considerably lower
compared to 3 (pKa 9.0), ranging from 5.4 to 6.0. We have to
mention that those smaller fragments could adopt both possible
Figure 2. X-ray structure of 2 in thrombin. Hydrogen bonds are represented as
white dotted lines.
It is hypothesized that hydrophobic interaction also plays a crucial
role in the increased affinity for 2 (Fig. 3). The ethyl ester group is
situated in a non-polar S2 subpocket formed by Trp60d, Tyr60a,
Leu99 and His57. The branching point of the side chain can be
responsible for the unexpected orientation flip with the guanidine
part interacting with Glu192 instead of Asp189. Other derivatives
of 1, N-alkylated with different larger alkyl acetamides were also
synthesized. Of these, only the ethyl acetamide substituted benzo-
thiazole guanidine 3 showed activity (data not shown), which
demonstrates the spatial limitations of the subpocket. The weaker
affinity of 3 compared to 2 can be ascribed to an increased rigidity
changing the ester bond with the amide bond resulting in a slightly
disfavoured position of the ethyl acetamide part in the subpocket.
Also the differences in lipophilicity of the side chain may be crucial
regarding the rather hydrophobic character of the subpocket. In
the search of further active fragments, the obtained crystal struc-
ture was used for the structure based design of a second set of frag-
ments based on the benzothiazole guanidine scaffold. Identifying a
potent thrombin inhibitor which contains a P1 fragment based on
the benzothiazole guanidine scaffold involves finding an optimal

4842
M. Karle et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4839–4843
Scheme 1. Reaction conditions: (a) TBDMSCl, Imidazole, DCM, rt; (b) FmocSCN, DCM, rt; (c) morpholine, DCM, rt; (d) ANMe3Br3 (A = Ph or PhCH2), DCM; (e) PhCH2NMe3Br3,
DCM, 15:16 = 5:1.
orientations in the S1 pocket, depending on a future connection to
P2 fragments. The second set of substituted benzothiazole guani-
dines was analyzed by WaterLOGSY NMR fragment screening
against thrombin and affinity measurements using a Biacore
A100. Though WaterLOGSY NMR indicated very weak binding of
these fragments to thrombin, we could not show binding of these
compounds to thrombin nor inhibition of thrombin in the enzy-
matic assay (Table 1). However, compound 4 inhibited trypsin I
and IV while compound 7 showed a selective inhibition of trypsin
IV. Thus, based on these fragments, novel lead structures as trypsin
inhibitors can be developed with an additional opportunity to in-
crease selectivity towards trypsin IV.
Benzothiazole guanidines 2 and 3 were part of a library of func-
tionalised 2-guanidino-6-hydroxy benzothiazole 1. Alkylation of
the benzothiazole guanidine scaffold was achieved by Mitsunobu
reaction of the di-Boc protected precursor of 1 with suitable alco-
hols and subsequent Boc-deprotection. Benzothiazole guanidines
4–7 were prepared as outlined in Scheme 1 and Scheme 2. A key
step in this synthesis was the preparation of substituted 2-amin-
obenzothiazoles 11a, 11b, 15 and 16 (Scheme 1). Starting from
anilines 8a, 8b and 12, condensation with Fmoc-isothiocyanate
in dichloromethane (DCM) at rt provided the Fmoc-protected
thiourea. Subsequent Fmoc-deprotection with morpholine gave
thioureas 10a, 10b and 14. Cyclisation of the thioureas using ben-
zyltrimethylammonium tribromide provided 4-, 5-, 6- and 7-
substituted 2-aminobenzothiazoles 11a, 11b, 15 and 16. Di-Boc
protected 2-guanidinobenzothiazoles 18a, 18b, 19a, 19b and 21
were obtained by treatment of the corresponding 2-aminobenzo-
thiazoles with N,N0-di-Boc-N0-triflylguanidine 17 as guanidinyla-
tion reagent20,21 (Scheme 2). Deprotection of the Boc-group with
trifluoroacetic acid (TFA) in DCM yielded target compounds 1, 5
and 6. Protected guanidines 18a and 18b were deprotected using
a mixture of TFA and HF in DCM to give 4 and 7.
Fragment screening techniques are powerful tools to find new
weak thrombin inhibitors with comparatively low molecular
weight. In this work, WaterLOGSY fragment screening proved to
be successful to identify binding fragments. Binding data could
then be obtained using the plasmon surface resonance technology
and enzymatic assays. To design potent thrombin inhibitors
starting from such fragments with comparatively low affinity, it
is essential to know the binding mode of these fragments. The
crystal structure of the most potent compound with thrombin re-
vealed a rather unexpected binding mode. So far, the key interac-
tion of strong basic P1 fragments containing the amidine motif is
Scheme 2. Reaction conditions: (a) Dioxane, NEt3, microwave heating 110 °C; (b) TFA, HF, DCM, rt; (c) TFA, DCM, rt.

M. Karle et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4839–4843
4843
interaction of the amidine with Asp189. In our case we could dem-
References and notes
onstrate that the guanidine group in benzothiazole guanidines
does not necessarily behave in the same way. Instead, the guani-
dine group can also interact with Glu192 and the molecule can
adopt a binding mode where the molecule is flipped 180° com-
pared to the expected binding mode. This behaviour seems to be
triggered by attachment of suitable side chains. Based on this,
our work resulted in the synthesis of further weak thrombin inhib-
itors, with a stronger focus on lower molecular complexity and
lower pKa which is considered to be favourable for increased bio-
availability. However, this has to be proved for full inhibitors
employing our fragments as P1-group. This approach led to a sig-
nificant drop in potency against thrombin. However testing of
the compounds against another emerging target protein, trypsin
IV revealed activity and selectivity of some of the compounds mak-
ing them attractive starting points for further work. Therefore,
finding an optimal way to incorporate these novel fragments into
suitable scaffolds can contribute to the future synthesis of neutral
thrombin and also trypsin IV inhibitors.
1. Turpie, A. G. G. Eur. Heart J. 2008, 29, 155.
2. Maragall, G. E. J. B. S. Drugs Today 2006, 42, 223.
3. Francis, C. W.; Berkowitz, S. D.; Comp, P. C.; Lieberman, J. R.; Ginsberg, J. S.;
Paiement, G.; Peters, G. R.; Roth, A. W.; McElhattan, J.; Colwell, C. W. N. Eng. J.
Med. 2003, 349, 1703–1712.
4. Eriksson, B. I.; Smith, H.; Yasothan, U.; Kirkpatrick, P. Nat. Rev. Drug Disc. 2008,
7, 557.
5. Das, J.; Kimball, S. D. Bioorg. Med. Chem. 1995, 3, 999.
6. Ceppa, E. P.; Lyo, V.; Grady, E. F.; Knecht, W.; Grahn, S.; Peterson, A.; Bunnett, N.
W.; Kirkwood, K. S.; Cattaruzza, F. Am. J. Physiol. Gastrointest. Liver Physiol. 2011,
300, 1033.
7. Knecht, W.; Cottrell, G. S.; Amadesi, S.; Mohlin, J.; Skaregarde, A.; Gedda, K.;
Peterson, A.; Chapman, K.; Hollenberg, M. D.; Vergnolle, N.; Bunnett, N. W. J.
Biol. Chem. 2007, 282, 26089.
8. Hajduk, P. J.; Boyd, S.; Nettesheim, D.; Nienhaber, V.; Severin, J.; Smith, R.;
Davidson, D.; Rockway, T.; Fesik, S. J. Med. Chem. 2000, 43, 3862.
9. Pellecchia, M.; Sem, D. S.; Wüthrich, K. Nat. Rev. Drug Disc. 2002, 11, 211–219.
10. Rees, D. C.; Congreve, M.; Murray, C. W.; Carr, R. Nat. Rev. Drug Disc. 2004, 3,
660.
11. Stockman, B. J.; Dalvit, C. Prog. Nucl. Magn. Reson. Spectrosc. 2002, 41, 187.
12. Tomczuk, B.; Lu, T.; Soll, R. M.; Fedde, C.; Wang, A.; Murphy, L.; Crysler, C.;
Dasgupta, M.; Eisennagel, S.; Spurlino, J.; Bone, R. Bioorg. Med. Chem. Lett. 2003,
13, 1495.
13. Taylor, P. J.; Wait, A. R. J. Chem. Soc., Perkin Trans. 2 1986, 1765.
14. Sanderson, P. E. J.; Cutrona, K. J.; Dyer, D. L.; Krueger, J. A.; Kuo, L. C.; Lewis, S.
D.; Lucas, B. J.; Yan, Y. Bioorg. Med. Chem. Lett. 2003, 13, 161.
15. Tucker, T. J.; Brady, S. F.; Lumma, W. C.; Lewis, S. D.; Gardell, S. J.; Naylor-Olsen,
A. M.; Yan, Y.; Sisko, J. T.; Stauffer, K. J.; Lucas, B. J.; Lynch, J. J.; Cook, J. J.;
Stranieri, M. T.; Holahan, M. A.; Lyle, E. A.; Baskin, E. P.; Chen, I.-W.; Dancheck,
K. B.; Krueger, J. A.; Cooper, C. M.; Vacca, J. P. J. Med. Chem. 1998, 41, 3210.
16. Dalvit, C.; Fogliatto, G.; Stewart, A.; Veronesi, M.; Stockman, B. J. Biomol. NMR
2001, 21, 349.
Acknowledgements
We thank Karin Kaspersson for excellent technical assistance
with the Biacore A100. We also want to thank Per-Olof Eriksson
for performing and evaluating all NMR WaterLOGSY experiments
and Ulrik Jurva for pKa measurements.
17. Dalvit, C.; Pevarello, P.; Tatò, M.; Veronesi, M.; Vulpetti, A.; Sundström, M. J.
Biomol. NMR 2000, 18, 65.
18. Sall, D. J. et al J. Med. Chem. 2000, 43, 649.
Supplementary data
19. Rajni Garg, S. K.; Hansch, C. Med. Res. Rev. 2001, 21, 73.
20. Feichtinger, K.; Sings, H. L.; Baker, T. J.; Matthews, K.; Goodman, M. J. Org. Chem.
1998, 63, 8432.
Supplementary data (experimental procedures for the synthesis
of 1–7, Biacore and enzymatic assay protocols and crystallographic
data) associated with this article can be found, in the online ver-
sion, at http://dx.doi.org/10.1016/j.bmcl.2012.05.046.
21. Feichtinger, K.; Zapf, C.; Sings, H. L.; Goodman, M. J. Org. Chem. 1998, 63, 3804.