(±) cis-bisamido epoxides: A novel series of potent FXIII-A inhibitors

Article abstract of DOI:10.1016/j.ejmech.2015.05.019

A novel class of potent FXIII-A inhibitors containing a (±) cis-bisamido epoxide pharmacophore is described. The compounds display highly potent inhibition of FXIII-A (IC₅₀ Combining double low line 5-500 nM) in an in vitro assay. In contrast to other types of previously described covalent transglutaminase inhibitors, the bis-Amido epoxides exhibited no measurable reactivity with glutathione, therefore possibly rendering this class of compounds suitable for future in vivo investigations. Additionally, the compounds show selective inhibition for FXIII-A against the cysteine protease, cathepsin S although they proved to have similar potency with a closely related transglutaminase, TGII, to that observed for FXIII-A.

Full text of DOI:10.1016/j.ejmech.2015.05.019

European Journal of Medicinal Chemistry 98 (2015) 49e53
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Research paper
( ) cis-bisamido epoxides: A novel series of potent FXIII-A inhibitors
Craig A. Avery a, Richard J. Pease b, Kerrie Smith b, May Boothby b, Helen M. Buckley b,
,
*
Peter J. Grant b, Colin W.G. Fishwick a
a School of Chemistry, University of Leeds, Leeds LS2 9JT, UK
b Leeds Institute of Cardiovascular and Metabolic Medicine, University of Leeds, Leeds LS2 9JT, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
A novel class of potent FXIII-A inhibitors containing a ( ) cis-bisamido epoxide pharmacophore is
described. The compounds display highly potent inhibition of FXIII-A (IC50 ¼ 5e500 nM) in an in vitro
assay. In contrast to other types of previously described covalent transglutaminase inhibitors, the bis-
amido epoxides exhibited no measurable reactivity with glutathione, therefore possibly rendering this
class of compounds suitable for future in vivo investigations. Additionally, the compounds show selective
inhibition for FXIII-A against the cysteine protease, cathepsin S although they proved to have similar
potency with a closely related transglutaminase, TGII, to that observed for FXIII-A.
Received 9 December 2014
Received in revised form
11 May 2015
Accepted 13 May 2015
Available online 16 May 2015
Keywords:
Thrombosis
© 2015 Elsevier Masson SAS. All rights reserved.
FXIII-A
Bis-amido epoxides molecular modelling
1. Introduction
would provide a mild and safe therapeutic approach for condi-
tions such as thrombosis, atherosclerosis and coronary heart dis-
FXIII is the final enzyme involved in the blood clotting process,
its major function being to introduce covalent cross links between
adjacent fibrin strands in a developing blood clot and covalently
ease [6]. Numerous in vitro [6] and several in vivo [7,8] trauma
induced thrombosis models have provided strong evidence that
this approach could provide a useful adjunct to current treatment
protocols.
attaching inhibitors of fibrinolysis, thus imparting
a vastly
increased stability and lifetime to the clot [1,2]. Plasma FXIII cir-
culates as a tetramer composed of two A domains and two B do-
mains. Cellular FXIII however is composed only of the two A chains,
suggesting the B domains are necessary for transportation of the
catalytic A regions in blood plasma. Several crystal structures of
human zymogen FXIII have been solved and consistently reveal the
presence of a C314, H373, D396 triad [3]. Zymogen FXIII has no
transglutaminase activity as the catalytic triad is buried deep in the
core domain of the enzyme surrounded by two barrel domains,
which block the entrance to the active site of the enzyme [4].
Conversion into the active transglutaminase (FXIII-A) is mediated
by thrombin and Ca2þ and involves a large conformational change,
which removes the barrel domains from the core region thus
allowing the substrates access to the active site [5].
There have only been a small number of non-peptidic FXIII-A
inhibitors identified to date. Two notable examples are the natural
products alutacenoic acids A and B [9], and the imidazo[1,2-d]
[1,2,4]-thiadiazoles [10], such as compound 1. FXIII-A is inhibited
irreversibly by these molecules via attack of the active site cysteine,
C314, on electrophilic moieties present in the inhibitors. The
mechanism of FXIII-A-catalysed fibrin cross-linking suggests that
the presence of a competing amine substrate will impede the for-
mation of fibrin cross-links. Lorand et al. thus established that
primary amines that can serve as substrates for the enzyme will
become incorporated into fibrin [11]. The same result can be ob-
tained via the use of pseudo-acceptor moieties, using glutamine
isosteres [12]. Several examples of these inhibitors have been
discovered for FXIII-A. In addition, a number of FXIII-A inhibitors
e.g. the peptide tridegin [13], ZG-1400 [14] and natural product
cerulenin 2 [15], were discovered through screening of natural
product libraries (Fig. 1). The inhibitors of FXIII-A currently re-
ported are severely lacking in many of the key criteria for successful
in vivo application and thus none have been investigated as po-
tential therapeutics. Therefore, there is a need to identify potent
and selective small molecule inhibitors for this enzyme that may
Factor XIII-A inhibitors do not prevent fibrin clot formation;
however they do inhibit the production of cross-linked, polymeric
clots. It has thus been proposed that selective inhibition of FXIII-A
* Corresponding author.
E-mail address: C.W.G.Fishwick@leeds.ac.uk (C.W.G. Fishwick).
http://dx.doi.org/10.1016/j.ejmech.2015.05.019
0223-5234/© 2015 Elsevier Masson SAS. All rights reserved.

50
C.A. Avery et al. / European Journal of Medicinal Chemistry 98 (2015) 49e53
Fig. 1. Examples of small molecule inhibitors of FXIII-A. Both compounds inhibit FXIII-A via reaction with the active site cysteine C314.
offer potential for therapeutic lead development. Here we report
the identification of a new class of FXIII-A inhibitors which exhibit
excellent inhibitory potency and hold potential for future devel-
opment for thrombosis therapy.
aid in the design of analogues. For these in silico studies, in the
absence of a crystal structure of the activated form of the enzyme, a
model of FXIII-A was created from the published zymogen crystal
structure (1GGT.pdb) [4], by manually deleting the ‘blocking’ barrel
domains (involving deletion of 229 residues between W500 e P729
inclusive) to reveal the active site. Cerulenin was then manually
positioned into this active site model of FXIII-A. Two constraints
were placed on the positioning of the epoxide moiety; firstly the
epoxide oxygen was placed in close proximity to the backbone NeH
of C314 as a primary ‘oxyanion hole’ residue [3]. Secondly, the
carbon atom immediately adjacent to the primary amide unit
within cerulenin was placed within close proximity to the thiol
moiety of C314, thus allowing the attack/ring opening of the
epoxide moiety to occur. Following this process, the complex was
minimised whilst freezing the protein atoms and including a
constraint such that the distance between C314 and the electro-
philic carbon of the epoxide was maintained at a distance of 2.4 Å to
simulate the necessary positioning of the thiol group immediately
prior to epoxide ring opening. This process led to the identification
of a binding mode very similar to that observed between cerulenin
and FAS, which has been previously characterised using X-ray
crystallography [16]. In particular, the FAS binding mode of cer-
ulenin reveals that inhibition involves attack by the active site
cysteine thiol on the epoxide ring whilst initially bound to the
enzyme as a non-covalent complex. After reaction, the resulting
oxyanion is stabilised via H-bonding with backbone NeH's within
the ‘oxyanion hole’ of the enzyme. The binding mode predicted
2. Results and discussion
Having studied a number of electrophilic moieties as covalent
‘warheads’ in the course of our FXIII-A research, it was apparent
that many of them were susceptible to reaction with endogenous
nucleophiles such as glutathione, which may compromise their
potential for in vivo application. Thus intracellular glutathione is
present at millimolar concentrations, but even if it were desirable
to selectively inhibit plasma FXIII-A with membrane impermeant
reagents, glutathione is present at micromolar concentrations in
plasma. However, cerulenin 2 which contains an epoxide moiety as
its electrophilic centre displays moderate inhibition of FXIII-A and
is unaffected by a large excess of glutathione. In particular, per-
forming the assay in the presence of 1 mM of reduced GSH affected
the IC50 only marginally, decreasing it from 4 mM to 18 mM (Fig. 2).
Cerulenin has previously been shown to inhibit the fatty acid
synthase enzyme (FAS) complex, which involves covalent bond
formation via attack of an active site cysteine thiol on the epoxide
moiety within the natural product with subsequent ring opening
[16]. In light of the interesting inhibitory properties displayed by
cerulenin in the presence of FXIII-A, we were keen to explore the
possible binding mode of this natural product with a view to
designing variants that would be amenable to synthetic manipu-
lation in order to enable SAR studies. Therefore, assuming that a
similar cysteine-based epoxide opening mechanism to that
observed with FAS was present, molecular modelling was per-
formed in order to deduce a likely binding pose for the pre-covalent
attack complex of cerulenin within the FXIII-A active site and to also
herein supports
a similar mechanism for FXIII-A inhibition
involving attack by C314 at the 2-position (bearing the primary
carboxamide) on the epoxide ring of cerulenin. Furthermore, the
modelling suggested that replacing the 3-keto unit within cer-
ulenin with an amide moiety was predicted to enhance binding to
FXIII-A in the initial complex via formation of an additional H-bond
from this amide NeH to the backbone carbonyl of Y372. Addi-
tionally, replacement of the alkenyl chain in cerulenin with an ar-
omatic ring was predicted to result in a possible
p-stacking
Cerulenin (FXIIIa assay)
interaction with Y372 and, again, potentially enhancing binding of
the inhibitor within the ‘pre-attack’ complex (Fig. 3).
150
cerulenin
In light of the predictions from the modelling studies, it was
desirable to prepare a small series of potential FXIII-A inhibitors
based upon a cis-bisamido epoxide motif. The modelling had
indicated that the absolute stereochemistry within cerulenin (2R,
3S) and, by analogy, within the designed bis-amido epoxides, as
shown above (Fig. 3) was important and that stereoisomeric ep-
oxides would bind less efficiently. However, in order to expedite
synthesis and biological evaluation of molecules corresponding to
the designed inhibitors, it was decided to prepare the inhibitors as
racemates.
cerulenin + GSH (20uM)
cerulenin + GSH (1mM)
100
50
0
-2
-1
0
1
2
3
log conc (M x 10-6)
Initially, synthesis of compounds 7 and 8 was achieved through
the intermediate succinimides 5 and 6, following ring opening
with ammonia. Succinamides 5 and 6 were readily obtained in two
steps from diacid 3 (Scheme 1).
Unfortunately this approach proved to have limited scope and
was therefore discontinued in favour of an alternative approach
involving amide coupling to amido acid 9. Thus, anhydride 4 was
Fig. 2. Dose response curve for cerulenin induced FXIII-A inhibition. Cerulenin (black
curve) inhibits with an IC50 of 4 M. GSH addition at 20 M (blue curve) and 1 mM (red
m
m
curve) are also shown. The IC50 is only marginally changed by the addition of GSH as
indicated by the close overlap of the dose response curves. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of
this article.)

C.A. Avery et al. / European Journal of Medicinal Chemistry 98 (2015) 49e53
51
Fig. 3. (A) Predicted binding mode of epoxide 16 within the FXIII-A active site. (B) Schematic diagram of the predicted polar contacts between epoxide 16 and FXIII-A and the
proposed mechanism of inhibition by epoxide 16.
Scheme 1. Synthesis of ( ) cis-bisamido epoxides, Reagents and conditions: a) TFAA, DCM, quantitative b) 4-butyl aniline, ether, followed by Ac2O, AcONa, 31e41% (over 2 steps) c)
NH3 (aq), MeOH, 67e75%.
ring opened with hexamethyldisilizane to produce acid 9. Amide
coupling with 1-hydroxybenzatriaozle monohydrate and 1-ethyl-
3-(3-dimethylaminopropyl) carbodiimide methiodide was then
performed with a number of amines (Scheme 2) to produce a small
library of inhibitors 10e20, which displayed good potency towards
FXIII-A (Table 1).
much weaker inhibition (with the exception of compound 14), than
compounds 11e17. The weak inhibition produced by epoxide 10 can
be interpreted as an entropic effect on binding the conformation-
ally flexible butyl chain to the FXIII-A active site. The remainder of
this series of inhibitors display potencies in broadly the same range
(0.1e0.5
mM) with two notable exceptions. Firstly, unsubstituted
The inhibition appeared to be irreversible as the activity of
enzyme that had been exposed to the inhibitors could not be
restored following exhaustive dialysis, consistent with irreversible
formation of an enzyme-inhibitor covalent bond resulting from
attack of C314 onto the epoxide moiety of the inhibitors. Addi-
tionally, the biological data revealed some telling insights into the
nature of the bis-amido epoxide binding mode. In particular, in-
hibitor 10, which contains a butyl chain, exhibits relatively weak
inhibition. Although containing a side chain modelled, in terms of
overall length, on cerulenin itself, butyl-based system 10 exhibits
aniline derivative 14 induced much weaker inhibition (0.72
m
M) of
FXIII-A. This indicates that substituents on the phenyl ring are
playing a role in binding to FXIII-A. Secondly, the 3-phenoxy, and 3-
chloro-4-bromo analogues 13 and 16 are at least ten-fold more
potent (IC50 ¼ 4 nM) than any other derivative assayed. The chloro-
substituent at the 3-position of 16 is evidently important in pro-
ducing a substantial increase in affinity, as emphasised by com-
parison to 4-bromo analogue 15 which has an IC50 of 130 nM. It is
interesting to note that 3,4-dimethyl analogue 17 produced
comparatively weak inhibition of FXIII-A, thus suggesting that the
Scheme 2. Alternative synthesis of ( ) cis-bisamido epoxides, reagents and conditions: a) Hexamethldisilazine, THF, 55% b) RNH2, EDCꢀMeI, HOBtꢀH2O, 19e69%.

52
C.A. Avery et al. / European Journal of Medicinal Chemistry 98 (2015) 49e53
Table 1
sulfonamide derivative 18 was prepared, in which the side chain
possesses a similar structure to that within the previously reported
thiadiazole based inhibitor 1 [10]. The sulphonamide-containing
amine precursor used for the synthesis of inhibitor 18, was pre-
pared from amine 22 as shown in Scheme 3. Additionally, analogue
21, based around an extension of phenoxy derivative 11, via the
formation of an amide at the 4-position, was targeted. Specifically,
amide 25 was synthesised as depicted in Scheme 3 and was then
coupled to acid 9 using the standard conditions, to yield molecule
21.
However, sulfonamide 18 showed similar potency to the simpler
derivatives suggesting that either the compound does not effec-
tively occupy the FXIII-A pocket or that the flexibility in this longer
side chain imparts an entropic penalty on binding, thus lowering
the overall affinity relative to the increase in size. Analogue 21 was
also found to possess similar potency to the simpler analogues
(Table 1). Perhaps in this case, a slightly more flexible system is
required in order to allow the ‘tail’ portion of the inhibitors to
effectively fill the binding pocket, although as indicated by sul-
phonamide 18, highly flexible alkyl chains may show relatively
modest inhibition.
In vitro activity data of bis-amido epoxide derivatives toward plasma Factor XIII-A.
Compound
R
IC50
FXIII-A
0.22 0.13
0.055 0.011
0.048 0.053
0.0040 0.053
0.72 0.13
(mM)
TGII
10
11
12
13
14
15
16
17
18
19
20
21
4-Bu
4-OPh
3-Cl
3-OPh
H
4-Br
3-Cl, 4-Br
3,4 di-Me
4-CH2CH2NHSO2(2-NO2Ph)
4-(4-NO2C6H4O)
4-(4-NH2C6H4O)
4-(C6H4)-4-NHCO(4-F C6H4)
0.15 0.092
0.046 0.070
0.048 0.054
0.050 0.067
1.1 0.043
0.13 0.019
0.0040 0.00021 0.0096 0.018
0.057 0.031
0.11 0.021
0.090 0.043
0.43 0.0097
0.086 0.073
0.10 0.016
0.13 0.025
0.088 0.024
N.D
0.048 0.029
0.15 0.092
IC50 values were generated by non linear regression of percent inhibition vs log
inhibitor concentration for at least four different concentrations.
Gratifyingly, as for cerulenin itself (Fig. 1), it was found that the
FXIII-A inhibitory activity of all of the bisamido epoxides did not
significantly change in the presence of GSH (See Supporting
Information), underlining the stability of these systems in the
presence of simple thiols. Further studies, (not shown), similarly
indicated a low rate of spontaneous hydrolysis of the epoxide to the
diol in neutral aqueous solution.
halogen groups present in inhibitor 16 are more suitable sub-
stituents than alkyls for producing potent inhibitors of FXIII-A. In
addition, the potent inhibition observed for 3-phenoxy derivative
13 suggests that larger groups at the 3-position of the arylamide
moiety of these inhibitors are well tolerated within the FXIII-A
active site. The corresponding acids of the carboxamide in-
hibitors, obtained from direct opening of anhydride 4 with the
corresponding amines, were also assayed against FXIII-A and
3. Selectivity
In order to assess the selectivity of the inhibitors, they were
assayed against two other types of cysteine-based enzymes. Firstly,
all compounds were tested against the evolutionarily distant
enzyme cathepsin S, in order to probe the ability of the epoxides to
inhibit other cysteine-protease inhibitors. Gratifyingly, all com-
pounds showed negligible inhibition of cathepsin S at concentra-
showed no activity at 100 mM (data not shown), underlining the
importance of the primary amide group and reinforcing the
modelling which predicts that the primary amide group is making
H-bonding contacts within the active site (Fig. 3). Additionally, ion
pair repulsion between the C314 thiolate and the carboxylate pre-
sent in the acidic compounds could also contribute to this lack of
potency.
Modelling had also indicated that the rather elongated binding
pocket of FXIII-A may be able to accommodate longer side-chains
than those within molecules 10e17, 19 and 20. Therefore,
tions as high as 100
mM, which equates to around 10,000 fold
selectivity for FXIII-A. Having established that the inhibitors show a
level of selectivity for the active site of FXIII-A, we wished to
determine the extent of this selectivity via comparison of the FXIII-
A inhibitory activity with that shown with transglutaminase II
(TGII), which is a transglutaminase with a remarkably close
Scheme 3. Synthesis of anilines required to prepare derivative 18 and 21, reagents and conditions: a) 2-Nitrobenzenesulfonyl chloride, NEt3, DCM, 44% b) 4-fluorobenzoyl chloride,
pyridine followed by H2 Pd/C, MeOH, 50% over 2 steps.

C.A. Avery et al. / European Journal of Medicinal Chemistry 98 (2015) 49e53
53
structural similarity to FXIII-A [17]. However, as shown in Table 1,
References
the inhibitors also displayed similar levels of potency to TGII, in
keeping with the very high structural similarities between the two
enzymes. Further studies may achieve selectivity by targeting a
binding cleft specific to TGII, but alternatively since FXIII-A is the
only transglutaminase expressed at measurable levels in plasma,
specificity might also be achieved by confining the inhibitor to this
compartment.
[1] S. Matacic, A.G. Loewy, The identification of isopeptide crosslinks in insoluble
fibrin, Biochem. Biophys. Res. Commun. 30 (1968) 356e362.
[2] L. Lorand, H.H. Ong, B. Lipinski, N.G. Rule, J. Downey, A. Jacobsen, Lysine as
amine donor in fibrin crosslinking, Biochem. Biophys. Res. Commun. 25 (1966)
629e637.
[3] L.C. Pedersen, V.C. Yee, P.D. Bishop, I. Letrong, D.C. Teller, R.E. Stenkamp,
Transglutaminase factor XIII uses proteinase-like catalytic triad to crosslink
macromolecules, Protein Sci. 3 (1994) 1131e1135.
[4] V.C. Yee, L.C. Pedersen, I. Letrong, P.D. Bishop, R.E. Stenkamp, D.C. Teller,
Three-dimensional structure of a transglutaminase: human blood coagulation
factor XIII, Proc. Natl. Acad. Sci. U. S. A. 91 (1994) 7296e7300.
[5] L. Lorand, A.J. Gray, K. Brown, R.B. Credo, C.G. Curtis, R.A. Domanik,
P. Stenberg, Dissociation of the subunit structure of fibrin stabilizing factor
during activation of the zymogen, Biochem. Biophys. Res. Commun. 56 (1974)
914e922.
[6] L. Lorand, Research on clot stabilization provides clues for improving
thrombolytic therapies, Arterioscler. Thromb. Vasc. Biol. 20 (2000) 2e9.
[7] E.M. Leidy, A.M. Stern, P.A. Friedman, L.R. Bush, Enhanced thrombolysis by a
factor XIIIa inhibitor in a rabbit model of femoral artery thrombosis, Thromb.
Res. 59 (1990) 15e26.
[8] R.J. Shebuski, G.R. Sitko, D.A. Claremon, J.J. Baldwin, D.C. Remy, A.M. Stern,
Inhibition of factor XIIIa in a canine model of coronary thrombosis: effect on
reperfusion and acute reocclusion after recombinant tissue-type plasminogen
activator, Blood 75 (1990) 1455e1459.
[9] H. Kogen, T. Kiho, K. Tago, S. Miyamoto, T. Fujioka, N. Otsuka, K. Suzuki-
Konagai, T. Ogita, Alutacenoic acids A and B, rare naturally occurring cyclo-
propenone derivatives isolated from fungi: potent non-peptide factor XIIIa
inhibitors, J. Am. Chem. Soc. 122 (2000) 1842e1843.
[10] R. Leung-Toung, T.F. Tam, J.M. Wodzinska, Y.Q. Zhao, J. Lowrie, C.D. Simpson,
K. Karimian, M. Spino, 3-Substituted Imidazo[1,2- d ][1,2,4]-thiadiazoles: a
novel class of factor XIIIa inhibitors, J. Med. Chem. 48 (2005) 2266e2269.
[11] J. Lorand, T.R. Pilkington, L. Lorand, Inhibitors of fibrin cross-linking: relevance
for thrombolysis, Nature 210 (1966) 1273e1274.
4. Conclusions
Molecular modelling of the initial binding mode of a natural
product-based FXIII-A inhibitor cerulenin, indicated that simple
substitution of a ketone for an arylamide moiety could yield ana-
logues which were predicted to be more potent but which would
also be more synthetically accessible, and a number of these cis-
bisamido epoxides were readily prepared using a simple synthetic
procedure. Assaying of these compounds revealed that a significant
increase in inhibitory potency could be achieved (up to IC50 ¼ 4 nM)
compared to that displayed by cerulenin (5 mM). A limited study of
selectivity revealed that, whilst these inhibitors appear to be se-
lective for FXIII-A compared to the unrelated cysteine protease
cathepsin S, they could not differentiate between FXIII-A and TGII,
which are closely related members of the transglutaminase family.
As such differentiation is desirable in a clinical setting, more work is
needed in order to optimise the structures of these inhibitors as
potential antithrombotic drug leads.
[12] L. Lorand, I. Simpson, C.H.J. Chou, Thiolester substrates for transamidating
enzymes: studies on fibrinoligase, Proc. Natl. Acad. Sci. U. S. A. 69 (1972)
2645e2648.
[13] S. Finney, L. Seale, R.T. Sawyer, R.B. Wallis, Tridegin, a new peptidic inhibitor
of factor XIIIa, from the blood-sucking leech Haementeria ghilianii, Biochem. J.
324 (1997) 797e805.
[14] R. West, T. Martinez, H. Franklin, P. Bishop, B. Rassing, PCT Int. Appl. (1996)
WO 9640671.
[15] A.A. Tymiak, J.G. Tuttle, S.D. Kimball, T. Wang, V.G. Lee, A simple and rapid
screen for inhibitors of factor XIIIa, J. Antibiot. (Tokyo) 46 (1993) 204e206.
[16] P. Johansson, B. Wiltschi, P. Kumari, B. Kessler, C. Vonrhein, J. Vonck,
D. Oesterhelt, M. Grininger, Inhibition of the fungal fatty acid synthase type I
multienzyme complex, Proc. Natl. Acad. Sci. U. S. A. 105 (2008) 12803e12808.
[17] I. Komaromi, Z. Bagoly, L. Muszbek, Factor XIII: novel structural and functional
aspects, J. Thromb. Haemost. 9 (2011) 9e20.
Acknowledgement
The authors would like to thank Heart Research UK for a Stu-
dentship (to C.A.A).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.ejmech.2015.05.019.