Novel Furin Inhibitors with Potent Anti-infectious Activity

Article abstract of DOI:10.1002/cmdc.201500103

New peptidomimetic furin inhibitors with unnatural amino acid residues in the P3 position were synthesized. The most potent compound 4-guanidinomethyl-phenylacteyl-Arg-Tle-Arg-4-amidinobenzylamide (MI-1148) inhibits furin with a K_i value of 5.5pM. The derivatives also strongly inhibit PC1/3, whereas PC2 is less affected. Selected inhibitors were tested in cell culture for antibacterial and antiviral activity against infectious agents known to be dependent on furin activity. A significant protective effect against anthrax and diphtheria toxin was observed in the presence of the furin inhibitors. Furthermore, the spread of the highly pathogenic H5N1 and H7N1 avian influenza viruses and propagation of canine distemper virus was strongly inhibited. Inhibitor MI-1148 was crystallized in complex with human furin. Its N-terminal guanidinomethyl group in the para position of the P5 phenyl ring occupies the same position as that found previously for a structurally related inhibitor containing this substitution in the meta position, thereby maintaining all of the important P5 interactions. Our results confirm that the inhibition of furin is a promising strategy for a short-term treatment of acute infectious diseases. Anti-infectious furin inhibitors: New peptidomimetic furin inhibitors with unnatural amino acid residues in the P3 position have been synthesized. The most potent compound, 4-guanidinomethylphenylacteyl-Arg-Tle-Arg-4-amidinobenzylamide (MI-1148), inhibits furin with a K_i value of 5.5pM. The results confirm that the inhibition of furin is a promising strategy for a short-term treatment of acute infectious diseases.

Full text of DOI:10.1002/cmdc.201500103

DOI: 10.1002/cmdc.201500103
Full Papers
Novel Furin Inhibitors with Potent Anti-infectious Activity
Kornelia Hardes,[a] Gero L. Becker,[a] Yinghui Lu,[b] Sven O. Dahms,[c] Susanne Kçhler,[d]
Wolfgang Beyer,[d] Kirsten Sandvig,[e] Hiroyuki Yamamoto,[f] Iris Lindberg,[f] Lisa Walz,[g]
Veronika von Messling,[g] Manuel E. Than,[c] Wolfgang Garten,[b] and Torsten Steinmetzer*[a]
New peptidomimetic furin inhibitors with unnatural amino
acid residues in the P3 position were synthesized. The most
potent compound 4-guanidinomethyl-phenylacteyl-Arg-Tle-
Arg-4-amidinobenzylamide (MI-1148) inhibits furin with a Ki
value of 5.5 pm. The derivatives also strongly inhibit PC1/3,
whereas PC2 is less affected. Selected inhibitors were tested in
cell culture for antibacterial and antiviral activity against infec-
tious agents known to be dependent on furin activity. A signifi-
cant protective effect against anthrax and diphtheria toxin was
observed in the presence of the furin inhibitors. Furthermore,
the spread of the highly pathogenic H5N1 and H7N1 avian in-
fluenza viruses and propagation of canine distemper virus was
strongly inhibited. Inhibitor MI-1148 was crystallized in com-
plex with human furin. Its N-terminal guanidinomethyl group
in the para position of the P5 phenyl ring occupies the same
position as that found previously for a structurally related in-
hibitor containing this substitution in the meta position, there-
by maintaining all of the important P5 interactions. Our results
confirm that the inhibition of furin is a promising strategy for
a short-term treatment of acute infectious diseases.
Introduction
Furin belongs to the family of proprotein convertases (PC),
which comprises nine members in humans. Furin and the
other six furin-like proteases, PC1/3, PC2, PACE4, PC4, PC5/6,
and PC7, cleave their substrates after paired, and in rare cases
after single, basic residues. The preferred cleavage motif of
furin substrates is Arg-Xaa-(Lys/Arg)-ArgflXaa, containing an
additional Arg residue in the P4 position. In contrast, the other
two PCs, SKI-1 and PCSK9, prefer paired hydrophobic residues
in the P1 and P2 positions.[1,2]
Furin is a type I transmembrane serine protease and is ubiq-
uitously expressed in vertebrates and invertebrates. The pro-
tein cycles from the trans-Golgi network (TGN) through the en-
dosomal system to the cell membrane and back. The physio-
logical function of furin is the activation of precursor forms of
numerous receptors, hormones, and cell surface proteins. Stud-
ies with furin knockout mice revealed that furin expression is
critical to embryogenesis. The complete knockout of the fur
gene is not tolerated and results in embryonic death at
day 11.[3,4] In contrast, mice with a liver-specific conditional
furin knockout are viable, which suggests sufficient redundan-
cy between the furin-like PCs in adulthood.[5] In addition, the
specific inhibition of furin by polyarginine inhibitors is well tol-
erated in adult animals.[6,7] Besides its physiological functions,
furin also contributes to the pathology of numerous diseases.
It activates certain viral and bacterial pathogenic proteins and
is involved in tumorigenesis, atherosclerosis, and diabetes as
well as neurodegenerative disorders.[1,2,8]
[a] Dr. K. Hardes, Dr. G. L. Becker, Prof. T. Steinmetzer
Institute of Pharmaceutical Chemistry, Philipps University
Marbacher Weg 6, 35032 Marburg (Germany)
E-mail: steinmetzer@uni-marburg.de
[b] Dr. Y. Lu, Prof. W. Garten
Institute of Virology, Philipps University
Hans-Meerwein-Strasse 2, Marburg (Germany)
[c] Dr. S. O. Dahms, Dr. M. E. Than
Protein Crystallography Group
Leibniz Institute for Age Research—Fritz Lipmann Institute (FLI)
Beutenbergstrasse 11, 07745 Jena (Germany)
[d] S. Kçhler, Dr. W. Beyer
Institute of Environmental and Animal Hygiene, University of Hohenheim
Garbenstrasse 30, 70599 Stuttgart (Germany)
[e] Prof. K. Sandvig
Department of Biochemistry and Centre for Cancer Biomedicine
Institute for Cancer Research, The Norwegian Radium Hospital
Montebello, 0310 Oslo (Norway)
Several viruses possess fusogenic surface glycoproteins that
are cleaved and thereby activated by furin or related PCs. Ex-
amples are the precursors of fusion proteins of Orthomyxoviri-
dae (e.g., influenza virus), Paramyxoviridae (e.g., measles,
canine distemper virus), Filoviridae (Ebola and Marburg virus),
and Flaviviridae (e.g., dengue, West Nile virus).[9] Furthermore,
furin activates various bacterial toxins including the protective
antigen (PA) precursor of anthrax toxin secreted by Bacillus
anthracis. Furin also cleaves and thereby activates diphtheria
toxin, Pseudomonas exotoxin A (PEA), and shiga toxin. There-
fore, the inhibition of furin could be a suitable approach for
the development of new short-term therapies in acute infec-
[f] Dr. H. Yamamoto, Prof. I. Lindberg
Department of Anatomy and Neurobiology, University of Maryland
Baltimore, Maryland 21201 (USA)
[g] L. Walz, Dr. V. von Messling
Veterinary Medicine Division, Paul-Ehrlich-Institute
Federal Institute for Vaccines and Biomedicines
Paul-Ehrlich-Strasse 51–59, 63225 Langen (Germany)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500103. It contains a scheme for the
1
synthesis of inhibitor 9, the H NMR and 13C NMR spectra of inhibitors 18
and 19, cytotoxicity assay with selected inhibitors in MDCK cells, and the
synthesis of the fluorogenic substrate Phac-Arg-Val-Arg-Arg-AMC.
ChemMedChem 2015, 10, 1218 – 1231
1218
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tious disease.[1] Cell-culture experiments and first proof-of-con-
cept studies in living animals indicate a therapeutic benefit of
furin inhibitors, for example, in the survival rate of mice treated
intraperitoneally with PEA or against anthrax-induced toxemia
in rats.[6,7] Moreover, autologous vaccines have been prepared
from tumor cells sampled from patients by incorporation of
a plasmid encoding granulocyte-macrophage colony-stimulat-
ing factor and a short hairpin RNAi targeting furin. The result-
ing knockdown of furin in these cells leads to nearly complete
inhibition of the activation of transforming growth factors b1
and b2, which thereby reduces immune suppression and in-
creases the antitumoral immune response.[10] Injection of these
cells was found to be beneficial in patients with various types
of advanced cancers,[11] including hepatocellular carcinoma.[12]
Moreover, no long-term toxicity has been observed with these
vaccines. Taken together, these studies suggest that despite
important roles of furin-like PCs in essential physiological pro-
cesses, its inhibition could be a suitable strategy for the treat-
ment of numerous diseases.
furin could be solved, which provides the first structural ex-
planation of the improved potency of P3 Tle inhibitors. For se-
lected compounds, and especially for MI-1148, a significant an-
tiviral and antibacterial effect was found in various cell culture
assays, as described below.
Results
Synthesis and inhibitory potency
In a previous study we had incorporated all of the proteino-
genic amino acids into the P3 position of our inhibitor scaffold,
with a neutral phenylacetyl (Phac) group as the P5 residue,
and found that furin preferred compounds with Val and Ile res-
idues over those with a Leu residue.[22] Therefore, starting from
Gly inhibitor 1 (Ki =44.3 nm) we performed a stepwise addition
of single methylene groups on the P3 side chain, and this re-
sulted in the compounds shown in Scheme 1. Notably, all of
the provided inhibition constants (Ki) against furin were newly
determined for this work by using the substrate Phac-Arg-Val-
Various types of synthetic furin inhibitors have been devel-
oped during the last years,[8,13,14] for example, pure peptide-
based derivatives such as nona-d-Arg-amide (D9R)[15] and non-
peptidic small molecules including the guanylated analogues
of 2,5-dideoxystreptamine.[16] Our focus is directed to the de-
velopment of substrate–analogue inhibitors containing decar-
boxylated arginine mimetics as P1 residues, whereby the stron-
gest inhibitory potency was obtained after incorporation of
a 4-amidinobenzylamide (4-amba) group.[17] This approach was
further elaborated by other groups using different P1 residues
and N-terminal elongations.[18,19] Further improved potencies in
the low picomolar range have been achieved by combination
with basic P5 residues.[20] Crystal structures of these inhibitors
in complex with human furin have revealed numerous key in-
teractions contributing to the strong in vitro potency of these
analogues.[21] During previous optimizations of the P3 position,
using all of the proteinogenic amino acid residues, we ob-
served significant differences in the inhibitory potencies after
incorporation of structurally closely related residues. The stron-
gest inhibition was found for analogues containing the Cb-
branched residues Val and Ile, whereby the direct Cg-branched
Leu derivative was approximately 15 times less potent.[22] This
surprising result was difficult to understand, because the crys-
tal structure of mouse furin in complex with the irreversible in-
hibitor decanoyl-Arg-Val-Lys-Arg-CMK (CMK=chloromethyl
ketone) reveals that the P3 Val side chain is directed into the
solvent and is not involved in any specific contacts with
furin.[23]
Arg-Arg-AMC (KM =5.40Æ0.35 mm, kcat/KM =1.04”105 mÀ1 sÀ1
;
for the synthesis, see the Supporting Information). This ex-
This result encouraged us to explore the P3 position of this
inhibitor scaffold in more detail. The incorporation of a set of
nonproteinogenic P3 residues provided a series of inhibitors
with a broad range of potency. Interestingly, using tert-leucine
(Tle, also named 3-methylvaline or tert-butylglycine) led to
a fourfold increase in potency relative to that of Val. The com-
bination of a P3 Tle with basic P5 groups provided the most
potent furin inhibitors known so far. For the best compound,
4-guanidinomethyl-phenylacteyl-Arg-Tle-Arg-4-amidinobenzyl-
amide (MI-1148, 19), a crystal structure in complex with human
Scheme 1. Stepwise modification of the P3 residue within the scaffold of the
furin inhibitor Phac-Arg-P3-Arg-4-Amba, whereby only the P3 side chains,
the abbreviation for the amino acids used, and the Ki values (in nm) against
furin are shown. The previously published inhibitors containing the protei-
nogenic amino acids Gly, Ala, Val, Leu, and Ile[22] are included as a reference
and for completeness. The asterisk indicates the side chain carbon atom
connected to the Ca atom of the P3 residue.
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1219
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Analytical characterization of the inhibitors and their potency against furin, PC1/3, and PC2.
No.
R
P3
HPLC
[min]
MS
Ki [nm][a]
furin
Residual activity [%][a]
PC1/3
calcd
found
PC2
at 1 mm
at 0.1 mm
at 1 mm
[d]
1[b]
2[b]
3[b]
4[b]
5[b]
6
7
8
9
10
11
12
13
14
15
16
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Gly
Ala
Val
Leu
Ile
21.4
21.8
24.4
27.5
27.1
22.3
32.7
23.0
26.8
28.8
25.1
26.0
32.2
27.7
25.3
28.9
636.4
650.4
678.4
692.4
692.4
664.4
770.5
664.4
692.4
706.4
678.4
690.4
732.5
692.4
710.4
718.4
319.3[c]
326.4[c]
679.4
347.3[c]
347.3[c]
333.3[c]
771.5
333.2[c]
347.4[c]
354.3[c]
679.3
44.3Æ6.0
80.6Æ1.6
61.2Æ1.8
7.4Æ0.4
46.6Æ3.0
3.9Æ0.3
87.6Æ1.0
84.9Æ2.6
36.7Æ1.2
1.4Æ0
–
–
86.5Æ3.7
74.8Æ2.4
62.9Æ1.1
78.0Æ0.9
73.8Æ2.2
97.6Æ2.1
93.0Æ4.0
85.4Æ1.9
52.6Æ1.5
87.6Æ2.4
90.6Æ1.7
77.5Æ3.6
90.9Æ4.9
86.8Æ3.3
64.8Æ1.0
89.5Æ0.8
[d]
18.2Æ1.5
0.66Æ0.22
7.95Æ0.09
0.67Æ0.25
408Æ55.3
64.7Æ2.3
27.1Æ0.8
[d]
–
16.3Æ1.3
[d]
Aib
–
–
–
[d]
1-Adg
2-Abu
Tle
Tba
Nva
Cpa
Cha
Nle
[d]
2.06Æ0.83
0.17Æ0.04
1.98Æ0.63
3.67Æ1.1
9.3Æ0.4
[d]
17.4Æ0.8
40.2Æ3.1
23.6Æ1.0
68.0Æ2.9
51.3Æ1.8
2.8Æ0.2
34.8Æ0.2
–
–
–
–
[d]
[d]
691.4
733.4
693.6
711.2
1.83Æ0.08
7.13Æ0.93
6.16Æ2.68
0.25Æ0.01
7.14Æ0.47
[d]
[d]
–
Pen
Chg
17.3Æ1.8
[d]
719.5
–
17[b]
(MI-0701)
Val
Tle
Tle
Tle
Tle
19.7
19.1
20.4
19.8
20.3
749.5
721.5
763.5
721.5
763.5
375.9[c]
722.4
764.3
722.4
764.3
0.0076Æ0.0013[e]
0.0224Æ0.0021[e]
0.0055Æ0.0003[e]
0.0361Æ0.0001[e]
0.0068Æ0.0005[e]
1.3Æ0.1
0.9Æ0.5
0.9Æ0.6
2.0Æ0.2
1.1Æ0.7
6.4Æ0.9
4.2Æ0.8
2.9Æ0.6
13.3Æ0.4
5.2Æ0.7
35.5Æ1.0
63.8Æ2.4
27.1Æ1.2
64.7Æ5.4
38.2Æ1.1
18
19
(MI-1148)
20
21
[a] Data represent the meanÆSD obtained for at least three independent measurements. [b] These compounds were previously published.[20,22]
[c] [M+2H]2+ À1. [d] Not determined. [e] Ki values were determined under tight-binding conditions.
2
plains small differences with our previous data, for which, for
instance, a Ki value of 40 nm was determined for this Gly inhib-
itor.[22]
The addition of the first methyl group provided Ala inhibitor
2; its potency was improved by a factor of roughly two relative
to that of inhibitor 1. Addition of the subsequent methyl
group had the strongest positive influence; a nearly 10-fold
improvement in the inhibition constant (2.06 nm) was achieved
by the incorporation of 2-Abu (8), which lacks only one termi-
nal methyl group in comparison to Val. In contrast, the incor-
poration of the sterically hindered Ca-dialkylated Aib residue
resulted in a dramatic drop in affinity. Further elongation of
the 2-Abu side chain by incorporation of norvaline (Nva, 11) or
noleucine (Nle, 14) reduced the affinity, whereas the addition
of a second methyl group at the Cb-atom provided three-
times more potent Val inhibitor 3. A nearly identical affinity
was found for Ile inhibitor 5, which confirmed our previous re-
sults.[22] Going from Nva to the Cg-branched Leu clearly re-
duced efficacy, whereas potency was restored both by cycliza-
tion of the Leu side chain leading to cyclopropylalanine (Cpa)
All compounds were synthesized according to a previously
described strategy,[17,22] and their analytical data are summar-
ized in Table 1. Briefly, the protected P5–P2 segment was pre-
pared on 2-chlorotritylchloride resin with a standard 9-fluore-
nylmethoxycarbonyl (Fmoc) solid-phase peptide synthesis
(SPPS) protocol. This segment was cleaved under mild acidic
conditions from the resin, maintaining the side chain protec-
tion, followed by coupling of 4-amidinobenzylamine·2HCl in
solution, and then final removal of all protecting groups. In
the case of inhibitors with an N-terminal aminomethylphenyl-
acetyl residue, the final conversion into the analogous
guanidinomethyl derivative was performed by treatment with
commercially available 1H-pyrazole-1-carboaxamidine·HCl[24]
(Scheme 2).
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1220
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
value for most-potent inhibitor 19 [(5.5Æ0.3) pm] was slightly
improved relative to that of Val analogue 17, the difference in
the inhibition constants is less than that of the inhibitors with
a neutral P5 residue.
Moreover, these inhibitors were also tested against the relat-
ed proteases PC1/3 and PC2. Owing to the limited amount of
available enzyme, only residual activities, at constant inhibitor
concentrations of 1 mm, were determined. As expected on the
basis of previous studies,[20] strong inhibition of PC1/3 was
found, especially for the inhibitors containing a basic-substitut-
ed P5 residue. With all compounds showing less than 10% re-
sidual activity, additional measurements at a 10-fold reduced
inhibitor concentration were performed. A strong correlation
between furin and PC1/3 inhibition was found. Within the pro-
teinogenic residues, Val and Ile were preferred by PC1/3,
whereas a slight preference for Ile was found. A similar poten-
cy was obtained for penicillamine analogue 15, and nearly
twofold stronger PC1/3 inhibition was determined for Tle in-
hibitor 9. Improved affinity was found for derivatives with
basic P5 groups; compound 19 is also the most potent PC1/3
inhibitor. A relatively weak affinity was observed against PC2.
For best compound 19, a residual activity of approximately
27% was determined at a concentration of 1 mm, although
only a minor improvement was found after incorporation of
basic P5 residues. Owing to its preferred potency for furin
among the PCs tested, the structure of inhibitor MI-1148 (19)
in complex with furin was determined.
Scheme 2. Synthesis of inhibitors 18 and MI-1148 (19). a) Loading of 2-chlor-
otrityl chloride resin with Fmoc-Arg(Pbf)-OH, DIPEA (4 equiv), dry CH2Cl2,
2 h; b) Fmoc-SPPS, double couplings with the amino acid or Boc-amino-
methylphenylacetic acid (4 equiv), HBTU (4 equiv), HOBt (4 equiv) and DIPEA
(8 equiv); c) cleavage from the resin with 1% TFA in CH2Cl2 (3”) for 30 min
and removal of the solvent in vacuo; d) 4-amidinobenzylamine·2HCl
(1.5 equiv), PyBOP (1.7 equiv), 6-Cl-HOBt (4.5 equiv), and DIPEA (10 equiv) in
DMF, 2 h; e) removal of the protecting groups with TFA/TIS/H2O (95:2.5:2.5
v/v/v), 3 h, precipitation in cold diethyl ether, purification by preparative
HPLC; f) 1H-pyrazole-1-carboxamidine·HCl (5 equiv) in 1m Na2CO3, 24 h. All
final compounds were purified by reverse-phase preparative HPLC and were
obtained as lyophilized TFA salts.
Crystallization of human furin in complex with MI-1148 (19)
Initial furin crystals in complex with the previously described
inhibitor I-1 (3-guanidinomethyl-Phac-Arg-Val-Arg-4-amba, Ki =
8 pm[20]) were grown, and the inhibitor was exchanged with
tighter binding analogue 19, essentially as described before.[21]
The complex crystals diffracted up to 2.15 Š resolution by
using synchrotron radiation and gave rise to a dataset of good
quality (Table 2). Refinement of the structure resulted in final
R/Rfree factors of 18.6/21.9% with 95.3% of the residues in the
most favored regions, 4.7% in the allowed regions, and 0% in
disallowed regions of the Ramachandran plot[25] and good
stereochemistry.
inhibitor 12 and by incorporation of tert-butylalanine (Tba, 10).
To our delight, relative to Val inhibitor 3, nearly fourfold stron-
ger furin inhibition was obtained for Tle derivative 9, which
contains a space-filling quaternary Cb-atom at the P3 side
chain. As a close analogue of Tle, penicillamine was also intro-
duced and provided similarly potent inhibitor 15. Moreover,
the mercapto group of this inhibitor could serve as a handle
for further modifications, for example, by alkylation. A strong
drop in affinity was obtained after incorporation of hydropho-
bic 1-adamantylglycine (7), although this residue also contains
a quaternary Cb-atom. Reduced potency similar to that found
for Leu inhibitor 4 was obtained for Cha analogue 13 and Chg
derivative 16.
On the basis of these results, four additional inhibitors were
prepared by combining the preferred Tle residue with the
basic-substituted phenylacetyl residues in the P5 position
(Table 1). As described for Val analogue 17 (MI-0701),[20] all of
these Tle derivatives possess very strong affinity against furin
and were characterized as tight-binding inhibitors. Relative to
potency of the P5 aminomethyl-substituted derivatives,
a higher potency in the single digit picomolar range was
found for the P5 guanidinomethyl analogues. Although the Ki
The occupancy of the novel inhibitor within the highly nega-
tively charged active site cleft of all six molecules within the
asymmetric unit of the orthorhombic crystals was confirmed
by a strong difference electron density, which clearly showed
the molecular variations between compound 19 and the ini-
tially cocrystallized analogue I-1, as described below (Fig-
ure 1A). Correspondingly, complete exchange of the inhibitor
was achieved, as expected from the extended soaking proce-
dure.
Inhibitor 19 structurally varies from I-1 at two positions; the
P3 residue is Tle instead of the proteinogenic Val, and the gua-
nidinomethyl substitution at the P5 phenyl ring resides at the
para position instead of the meta position. Correspondingly
and as expected, the overall binding of inhibitor 19 is very sim-
ilar to that of I-1 (Figure 1A).[21] The tetrabasic inhibitor is com-
plementary in charge to the strongly acidic catalytic cleft. Its
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1221
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tional shifts of the two methyl groups are almost within the
error of crystallographic structures, such that no final conclu-
sions can be drawn as to the structural reason for the ob-
served gain in affinity of P3-Tle relative to that of P3-Val.
Table 2. Data collection and refinement statistics (PDB-ID: 4RYD).
Data collection statistics
Wavelength [Š]
Unit cell parameters
a, b, c [Š]
0.918
P212121
141.65, 152.70, 168.40
50.0–2.15 (2.28–2.15)
13.5 (69.3)
9.1 (2.0)
98.3 (97.7)
669363/194799
Resolution range[a] [Š]
Rmerge[a] [%]
Binding of noncovalent inhibitors to human furin stabilizes
the enzyme
I/sI[a]
Completeness[a] [%]
Observations: total/unique
To compare the inhibitory strength of various inhibitors with
the conformational stability of the respective inhibitor–enzyme
complexes, we determined their melting temperature, Tm, in
respective thermofluor assays.[26] Plotting Tm as a function of
pKi, we observed a clear correlation (Figure 2), which suggests
a similar binding mode for these inhibitors. However, there are
small deviations from linearity with these inhibitors. This is
readily explained, as not all contacts that increase the interac-
tion between a given inhibitor and the enzyme, which thus
lowers the Ki, do so at the same time and to the same extent
also stabilize the enzyme–inhibitor complex. In general, how-
ever, a tighter binding inhibitor results in higher Tm and in-
creased conformational stability and hence better crystallizabil-
ity of the respective inhibitor–enzyme complexes.[21] The deter-
mination of Tm can, however, only be suggestive for inhibitory
constants and does not replace their measurement.
Refinement statistics
Resolution range[a] [Š]
Rwork/Rfree
Non-hydrogen atoms, total
Protein/inhibitor/other
B factors [Š2]
50.0–2.15 (2.25–2.15)
18.6 (26.8)/21.9 (30.7)
23774
21332/330/2112
Overall/Wilson plot
Protein/inhibitor/other
RMSD bond length [Š]
RMSD bonded B-factors [Š2]
26.0/29.9
25.3/21.4/33.8
0.0076
2.41
[a] Highest resolution shell is given in parentheses.
P1 4-amba residue inserts deeply into the S1 pocket, where it
forms tight hydrogen bonds with the furin residues and
through bridging water molecules also contacts the PC-specific
Ca-2 ion.[21] The P2 side chain interacts predominantly with
Asn192 and Asp154 of the furin surface. Furthermore, two
strong b-sheet-like H-bonds between the carbonyl groups and
the amide groups of the P3 Tle residue and Gly255 of furin
contribute to inhibitor binding, and the extended hydrogen-
bonding network at the joined S4/S5 pocket is basically re-
tained. Interestingly, the 4-guanidinomethyl group at the P5
phenyl ring makes similar strong hydrogen bonds with the
main-chain carbonyl group of Val231, the side-chain carboxyl-
ate group of Glu236, and to the conserved bridging water mol-
ecule #58, which in turn provides identical contacts to Asp233,
Glu236, and Ala267 of the protein surface, as observed for the
complex with I-1.[21] As a consequence, the phenyl ring of the
P5 residue shifts towards the protein surface by approximately
1 Š (Figure 1B).
Antibacterial and antiviral activities in cell culture
We previously demonstrated an inhibitory effect of selected
compounds on the propagation of highly pathogenic avian in-
fluenza virus (HPAIV) in Madin–Darby canine kidney (MDCK)
cells by using the H7N1 fowl plague virus (FPV) strain and ana-
lyzed their protective effect on Vero cells against intoxication
by shiga toxin. In the present study, we investigated the influ-
ence of our new inhibitors on the protection of cells against
anthrax and diphtheria toxin and analyzed the propagation of
the canine distemper virus wildtype strain 5804PeH[27] as well
as the spread of two HPAIV strains, including FPV and H5N1
KAN-1, in cell culture.
Inhibition of anthrax toxin action
Replacement of the P3-Val residue in I-1 with the more
bulky Tle residue in inhibitor 19 does not result in any major
conformational alterations. Two of its three side-chain methyl
groups are oriented similarly to that observed for P3-Val (Fig-
ure 1C). A slight counterclockwise horizontal rotation (basically
towards the viewer) of the P3 moiety by roughly 58 results,
however, in a shift of these two methyl groups by approxi-
mately 0.5–0.8 Š, which possibly gives rise to stronger van der
Waals contacts with the phenyl ring of P1 and Cg of Glu257.
The third methyl group also makes van der Waals contacts
with the inhibitor backbone, mainly with the amide hydrogen
atom, but also with the carbonyl oxygen atom of P2. Corre-
spondingly, the P3-Tle moiety seems to fit better into the trian-
gular crevice formed by the phenyl ring of P1, the inhibitor
backbone, and the Glu257 side chain of the protein. The differ-
ences in contact are, however, small, and the observed posi-
Anthrax is a serious bacterial infection caused by Bacillus an-
thracis and primarily affects herbivores.[28] The secreted anthrax
toxins are bipartite toxins consisting of the protective antigen
(PA) and either the edema factor (EF) or the lethal factor (LF).
The intoxication process begins with the binding of the precur-
sor protein PA83 to specific receptors on the cell surface. PA83
contains a multibasic sequence that is cleaved by furin or
furin-like PCs.[29] The activated PA then forms a heptameric
complex, which can bind up to three molecules of LF and/or
EF and is then internalized by clathrin-mediated endocytosis.
In the acidic endosomes, this complex forms a pore and ena-
bles the translocation of LF and EF into the cytosol. LF is
a Zn2+ protease that cleaves and thereby inactivates mitogen-
activated protein (MAP) kinase kinases, which leads to lethal
disruption of the MAP kinase signaling pathway. EF is a Ca2+
and calmodulin-dependent adenylate cyclase that mediates
-
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1222
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the intracellular increase of cyclic
adenosine
monophosphate
(cAMP), which leads to edema
formation.[30] Therefore, inhibi-
tion of furin-mediated PA activa-
tion might represent a supple-
mental approach to the conven-
tional therapy against anthrax of
antibiotics or the recently ap-
proved
mab.[31,32]
antibody
Raxibacu-
To test the protective effect of
selected inhibitors against an-
thrax toxin, macrophages were
treated with constant concentra-
tions of PA and LF in the pres-
ence of various inhibitors,
whereby D9R (Ki =1.3 nm against
furin)[15] served as a reference
(Figure 3). At concentrations be-
tween 0.05 and 100 mm, the in-
hibitors showed no cytotoxic ef-
fects. Surprisingly, only a weak
effect was obtained for the refer-
ence D9R, although significant
protection was found in a previ-
ous study.[15] Strong and concen-
tration-dependent
protection
was found for all other com-
pounds. The highest efficacy was
produced by inhibitors 17, 21,
and 19, all containing a guanidi-
nomethyl substitution at the P5
phenyl ring. These data reveal
a clear correlation between the
Ki values against furin and the
protective effects in cell culture.
Inhibition of diphtheria toxin
action
Corynebacterium diphtheriae can
cause severe infections in
humans, especially in children
and older people. This may lead
to complications such as myo-
carditis and peripheral neuropa-
thy.[33] Experiments in cell culture
revealed that furin cleaves diph-
theria toxin between Arg193
and Ser194 into two fragments,
called A and B, linked by a disul-
fide bridge, and thereby acti-
vates it.[34] Fragment A inhibits
protein biosynthesis by adeno-
sine diphosphate (ADP)-ribosyla-
tion of elongation factor 2,
Figure 1. Structure of human furin in complex with 4-guanidinomethyl-phenylacteyl-Arg-Tle-Arg-4-amidinobenzy-
lamide (MI-1148, 19). A) Stereo representation of the active site cleft (surface representation) with bound inhibitor
as a ball-and-stick model in gray. The molecular surface was colored by the calculated electrostatic potential rang-
ing from À20 kTeÀ1 (red) to +20 kTeÀ1 (blue). The residues of the inhibitor are numbered (P1–P5) and its kicked
Fo–Fc difference electron density omit map (green mesh) is contoured at 3.5 s. B) In the stereo panels, the Ca
carbon trace of the protease is given as a cartoon representation (pale yellow). The inhibitor is shown in dark
gray (ball-and-stick model), and the important furin residues are shown in cyan (stick model). Selected water mol-
ecules are shown as red spheres. Important interactions are highlighted with black dashes. The aligned structure
of I-1 (3-guanidinomethyl-Phac-Arg-Val-Arg-4-amba) bound to furin (PDB-ID: 4OMC[21]) is given as stick model in
green. C) Stereo representation of the space-filling properties of the Tle residue at the P3 position. The inhibitor is
shown in dark gray (ball-and-stick model) and the important furin residues are shown in yellow (cartoon, stick
model). The van der Waals radii of selected atoms are illustrated by dotted spheres. Yellow dots belong to protein
atoms, gray dots to inhibitor atoms, and magenta dots to side-chain atoms of Tle at P3. The superimposed struc-
ture of I-1 bound to furin (PDB-ID: 4OMC[21]) is given as a stick model in green.
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1223
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Gain of global structural stability as a function of inhibitory
strength. Melting temperatures were determined on the basis of the fluores-
cence of the dye Sypro Orange by stepwise heating. The mean Tm ÆSD
resulting from three independent measurements is given for each inhibitor
as a function of Ki in a semilogarithmic plot. The horizontal dashed line
represents the Tm of the uninhibited enzyme at (54.6Æ0.0)8C.
Figure 4. Intoxication of Vero cells by unnicked diphtheria toxin (n=2).
A) Protection against diphtheria toxin by selected inhibitors (all at 100 nm
final concentration); B) Concentration-dependent protection of cells by in-
hibitors 17 and 19 against 0.1, 1, 10, and 100 ngmLÀ1 of unnicked diphtheria
toxin [control without inhibitor ( ), ID50 =0.23 ngmLÀ1; inhibitor 17: 2 nm
&
~
~
(
), ID50 =0.41 ngmLÀ1; 10 nm ( ), ID50 =1.75 ngmLÀ1; 50 nm (3),
ID50 =3.13 ngmLÀ1; inhibitor 19: 2 nm ( ), ID50 =0.49 ngmLÀ1; 10 nm ( ),
ID50 =1.98 ngmLÀ1, and 50 nm ( ), ID50 =3.38 ngmLÀ1; ID50 =half maximal
*
*
^
infectious dose].
dues showed significantly stronger protection. The best results
were achieved with Tle derivative 19 and Val analogue 17,
both containing guanidinomethyl substitution in the para posi-
tion of the P5 phenyl ring. The protection factors determined
for inhibitors 19 and 17 are 9.39Æ0.86 and 8.69Æ0.26, respec-
tively. On the basis of this prescreen, a second experiment
with varying concentrations of these two inhibitors was per-
formed (Figure 4B). As expected, concentration-dependent
protection was detected for both analogues, whereby at all
concentrations a slightly better effect was found for Tle inhibi-
tor 19. Notably, all inhibitors tested provided no protection
against already-nicked diphtheria toxin, which confirms that
the intoxication depends on the proteolytic activation of the
toxin.
Figure 3. Protection of murine macrophages against anthrax toxin by select-
ed inhibitors (n=3), determined by MTT assay. Cells were treated with
500 ngmLÀ1 PA and 100 ngmLÀ1 LF in the presence of various concentra-
tions of inhibitors.
which is responsible for the guanosine triphosphate (GTP)-
dependent translocation step on the ribosome.
Therefore, we screened selected inhibitors at a constant con-
centration of 100 nm against unnicked diphtheria toxin in Vero
cells. In this cell line, we also observed no cytotoxic effect at
the inhibitor concentrations used. We determined a protective
factor by calculating the ratio of the required toxin concentra-
tions inducing 50% cell death in the presence of the inhibitors
relative to the control without inhibitor (Figure 4A).
Inhibition of influenza virus propagation
Only negligible protection was found in the presence of D9R
(protective factor 1.12Æ0.19). An approximately twofold im-
proved effect was achieved with inhibitors 3 and 9, both con-
taining a neutral P5 residue. All inhibitors with basic P5 resi-
The HPAIVs induce severe systemic infections with multiorgan
failure in birds, which leads to 75–100% mortality within
10 days. Owing to close contact with infected animals or con-
taminated environment, the virus can also infect humans with
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1224
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
a high mortality rate. However, no transmission of these strains
among humans has been reported so far. Nevertheless, the
HPAIV strains could acquire the potential to provoke pandemic
flu by efficient transmission between humans.[35] For the repli-
cation of influenza viruses, the trimeric hemagglutinin precur-
sor protein (HA0) has to be cleaved by a host cell protease
into the subunits HA1 and HA2.[36] The HPAIV strains of sub-
types H5 and H7 possess a multibasic cleavage site, which is
activated by furin or the closely related PC5/6, and this makes
these PCs attractive targets for antiviral therapy.[37]
Therefore, selected inhibitors were investigated in a virus
spread assay in MDCK II cells, which provides semiquantitative
information about their influence on virus propagation. Also in
these cells no significant cytotoxicity (<10%) was observed at
the used inhibitor concentrations (Figure S5). Cells were infect-
ed by the HPAIV strains A/FPV/Rostock/34 (H7N1) and A/Thai-
land/KAN-1/2004 (H5N1) at indicated inhibitor concentrations
(Figure 5A,B). After 24 h, immunostaining of the nucleoprotein
indicates the efficacy of virus propagation, as described previ-
ously.[38,39]
A nearly identical influence of the inhibitors was observed
with both HPAIV strains. Compared to the control, the refer-
ence inhibitor D9R and 9 containing the neutral P5 residue
showed no effect on virus spread at the used concentrations.
An improved potency was found for inhibitor 18 containing
a p-aminomethyl substituent at the P5 phenyl ring and Tle in
the P3 position. Surprisingly, m-aminomethyl-substituted ana-
logue 20 showed only a negligible effect. Similar to that found
in the anthrax and diphtheria toxin assays, the best results
were obtained again with Tle derivative 19 and Val-containing
analogue 17 modified by an N-terminal p-guanidinomethyl
substituent. Both compounds significantly reduced virus
spread up to an inhibitor concentration of 5 mm.
Figure 5. Suppression of influenza virus spread by selected furin inhibitors.
MDCK II cells were infected with FPV (A) or KAN-1 virus (B) at an MOI of
0.001 and incubated without (control) or with various concentrations of indi-
cated inhibitors. At 24 h post infection, the cells were immunostained with
anti-FPV antibodies, whereas virus spread is indicated by gray spots.
Inhibition of canine distemper virus propagation
Canine distemper virus (CDV) belongs to the Paramyxoviridae
family and is closely related to measles and to a lesser extent
to mumps and parainfluenza viruses. In carnivores, its natural
hosts, CDV causes a severe acute disease with high mortality
and frequent neurologic complications. In the TGN, furin
cleaves the precursor fusion protein F0 in the mature form
consisting of disulfide bond linked subunits F1 and F2, which
is a prerequisite for the fusion of the viral envelope with the
plasma membrane of the target cell.[40,41] The inhibition of furin
or related PCs could thus be a suitable approach for the treat-
ment of paramyxoviruses with furin-activated fusion proteins.
To test this hypothesis, the effect of inhibitors 9, 17, 19, and
D9R on CDV propagation was examined in VeroSLAMdogtag
cells[42] by measuring the viral titers in the supernatant and in
the cellular fractions on days 1, 2, and 3 after infection
(Figure 6). The virus titers are expressed as the logarithmic
value of the dilution factor, at which 50% of the wells show
a cytopathic effect (log10 TCID50). For less effective inhibitors
D9R and 9, significantly reduced virus titers in the supernatant
and the cellular fractions were only seen at the highest con-
centration of 50 mm. In contrast, more potent furin inhibitors
17 and 19 with p-guanidinomethyl substitution at the P5
phenyl ring resulted in an antiviral effect starting at 0.5 mm.
Discussion
Upon starting our work on substrate–analogue furin inhibi-
tors,[17] we simply selected Val as the P3 residue, because it
was known to be well suited for that position on the basis of
various studies with peptidic chloromethyl ketones.[43] In
a second series, we tried to optimize the P3 position but could
not identify a better residue. Nearly identical potency was
found for an analogous Ile inhibitor, whereby the Cg-branched
Leu derivative was approximately 15-fold less potent.[22] To in-
crease our understanding of the influence of the P3 moiety,
a new inhibitor series with nonproteinogenic residues was pre-
pared. Very poor potency was found for Aib inhibitor 6. From
NMR spectroscopy studies with relatively short peptides it is
known that relative to Ala, the additional Ca substitution leads
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1225
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 6. Infection of VeroSLAMdogtag cells with CDV in the presence of the indicated furin inhibitors A) D9R (n=2), B) 9 (n=2), C) 17 (n=4), and D) 19
(n=4) at various concentrations. The determined log10 TCID50 values of the cell-free titers (left) and the cell-associated titers (right) were determined 24, 48,
and 72 h post infection. The significances (*, P<0.05; **, P<0.01, ***, P<0.001) were calculated by the two-way analysis of variance (ANOVA) by using the
Holm–Sidak method for multiple comparison versus a control group (without inhibitor) implemented in Sigmaplot 11.0.
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1226
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
to a further drastically reduced conformational space accessi-
ble for the Aib residue.[44] In consequence, the f and y torsion
angles in Aib preferably adopt values found in helical struc-
tures, which should disturb the typical antiparallel b-sheet
binding mode around the P3 residue determined for sub-
strate–analogue inhibitors in complex with furin[21,23] and other
subtilisin- or chymotrypsin-like proteases.[45,46]
analogue furin inhibitors we observed additive effects in po-
tency upon replacing one of the P5 to P1 groups with
a better-suited residue. It seems that the additional basic P5
anchor, which strongly improves the affinity of this inhibitor
type, largely compensates for the beneficial effect of the P3
Tle substituent, found only in the weaker inhibitors with a neu-
tral P5 group. Nevertheless, Tle derivative 19 is the most
potent synthetic, reversible furin in vitro inhibitor identified
thus far. Similar tendencies in structure–activity relationships
were also found for the inhibition of PC1/3 and PC2, whereas
PC2 was clearly less potently inhibited. In these studies, Tle de-
rivative 19 was also the strongest inhibitor, which suggests
a general beneficial effect of this P3 residue in PC inhibitors.
The evaluation of selected inhibitors in various cell culture
assays revealed that the efficacy of the Tle inhibitors was
slightly stronger than that of their direct Val counterparts in
most cases (see pairs 9 versus 3 and 19 versus 17). The best
results were obtained with Tle inhibitors 19 and 21 and Val an-
alogue 17, all containing guanidinomethyl substitution at the
P5 ring, which leads to Ki values <10 pm. Most of the other
reference compounds showed reduced efficacy in cell culture
and in some cases no effect. The differences found in the in vi-
tro furin inhibition nicely correlate with their efficacies in pro-
tecting cells against intoxication by anthrax and diphtheria
toxins. Moreover, relative to the inhibitor concentrations of our
previous studies with shiga toxin in HEp-2 cells,[20] much lower
inhibitor concentrations were sufficient to demonstrate signifi-
cant protective effects in these assays. This is understandable
for anthrax toxin, because the activation of PA is catalyzed by
cell surface-bound furin, which should be easily accessible to
exogenously added inhibitors. However, furin-catalyzed cleav-
age of diphtheria toxin occurs within the early endosomes.
Similar to the situation described for the endocytotic entry of
polybasic cell penetrating peptides or oligoarginine derivatives,
we assume that these multibasic PC inhibitors initially accumu-
late on the acidic cell surface and that this is followed by en-
docytotic uptake.[52–54]
In contrast, inhibitory potency was gradually improved by
adding one, two, and three methyl groups to the Cb atom of
Ala, and the highest affinity was obtained for Tle inhibitor 9. A
similarly strong furin inhibition was found for related penicilla-
mine inhibitor 15, which suggests that the quaternary Cb
group provides a major binding contribution. In contrast, poor
affinity was found for 1-Adg analogue 7, the third inhibitor
possessing this structural motif. Most likely, the sterically de-
manding adamantyl group prohibits proper binding of this in-
hibitor.
Interestingly, Tle is also used as the P3 residue in the ap-
proved hepatitis C virus (HCV) protease inhibitors boceprevir[47]
and telaprevir[48] and is incorporated twice in the HIV protease
inhibitor atazanavir.[49] The S enantiomer of Tle is available in
large quantities by reductive amination of trimethylpyruvic
acid catalyzed by leucine dehydrogenase. Its bulky, space-filling
tert-butyl side chain and concomitant hydrophobicity contrib-
utes to conformational control of peptide drugs and improves
its overall stability against enzymatic degradation.[50] A combi-
nation of spectroscopic studies revealed that tetrasubstitution
at the Cb atom preferentially leads to extended peptide con-
formations.[51] Comparison of the new crystal structure of inhib-
itor 19 with the previously solved furin structures in complex
with P3 Val analogues[21] revealed that the terminal methyl
groups of Val can be perfectly superimposed with two side-
chain methyl groups found in the Tle structure. For the addi-
tional third methyl group in Tle, we could not identify any spe-
cific contact with furin that could easily explain the improved
affinity. It is, however, involved in additional close intramolecu-
lar van der Waals contacts to the amide hydrogen atom (with
a distance to the nitrogen of 3.1 Š) and to the carbonyl
oxygen atom (3.5 Š) of the P2 Arg residue. Very similar intra-
molecular distances of the Tle side chain to the neighboring
P2 residue were found in the crystal structure of boceprevir in
complex with the HCV NS3-4A protease (PDB-ID: 2OC8).[47] Fur-
thermore, a similar fourfold improvement in the value of Ki
was observed for a Tle derivative relative to its Val analogue
during the development of boceprevir. These boceprevir re-
sults are reminiscent of the differences in potency found in our
furin inhibitor series, which suggests similar binding contribu-
tions of the Tle residue.[47] We speculate that these additional
intramolecular contacts stabilize the b-sheet-like backbone
conformation of the bound inhibitor, which thereby contrib-
utes to the enhancement in potency, or possibly by stabilizing
a more suitable preformed conformation in solution. However,
the fourfold-improved in vitro potency found for Tle inhibitor
9 over that of Val analogue 3 was not retained in generally
more potent inhibitors 19 and 17 that contain the p-guanidi-
nomethyl substituent at the P5 phenyl ring. This was surpris-
ing, because with most of our previously developed substrate–
Much higher inhibitor concentrations were required for in-
hibition of HPAIV spread in MDCK cells, for which the inhibitors
must reach furin residing in the TGN to block hemagglutinin
(HA) cleavage. Many polybasic compounds tend to accumulate
within the early endosomes, which can be further sorted to
the late endosomes and lysosomes for degradation by carriers
back to the plasma membrane or to the TGN.[55] Through the
latter mechanism, it is likely that a certain amount of the inhib-
itors reaches furin anchored in the TGN.
Although the furin-catalyzed activation of the CDV F0 pro-
tein also occurs in the TGN, more pronounced efficacy of our
inhibitors was seen against that virus, likely because its replica-
tion cycle is slower than that of influenza. For the best inhibi-
tor, Tle derivative 19, significantly reduced virus titers were de-
termined at a concentration of only 0.5 mm. The availability of
a small animal pathogenesis model for CDV will thus enable us
to immediately assess the in vivo efficacy of the next-genera-
tion furin inhibitors, designed to possess increased bioavailabil-
ity. Because of the high structural conservation of paramyxo-
virus F proteins, human viruses with furin-activated cleavage
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1227
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
sites will likely be inhibited to a similar extent. In addition to
measles and mumps viruses, which remain two of the most
important causes of vaccine-preventable childhood deaths,
this also includes the human parainfluenza and respiratory syn-
cytial viruses, which play an important role in infant mortality
as a result of respiratory infections and croup, considered pre-
disposing factors for asthma and chronic respiratory disease in
later life.
Experimental Section
Synthesis of inhibitors 18 and 19
General: Analytical HPLC experiments were performed by using
a Shimadzu LC-10A system (column: Nucleodur C18, 5 mm, 100 Š,
4.6”250 mm, Machery–Nagel, Düren, Germany) with a linear gradi-
ent of acetonitrile in water containing 0.1% trifluoroacetic acid
(TFA, 1–50% acetonitrile in 49 min for all final inhibitors and 40–
90% acetonitrile for protected intermediate 22, detection at l=
220 nm) at a flow rate of 1 mLminÀ1. The final inhibitors were puri-
fied to more than 95% purity (based on detection at l=220 nm)
by using preparative HPLC (pumps: Varian PrepStar Model 218 gra-
dient system, detector: ProStar Model 320, fraction collector: Varian
Model 701) by using a C8 column (Nucleodur, 5 mm, 100 Š, 32”
250 mm, Macherey–Nagel, Düren, Germany) and a linear gradient
Our data from the CDV and HPAIV studies in cell culture
demonstrate a significant advantage for compounds contain-
ing p-guanidinomethyl substitution at the P5 phenyl ring. In
contrast, poor activity was observed for inhibitor 9 containing
a neutral P5 group, although its in vitro Ki value of 0.17 nm
should be sufficient for furin inhibition. Therefore, we assume
that the major beneficial effect of the basic P5 residue is pri-
marily related to improved transport of the exogenously
added inhibitors to its target in the TGN.
of acetonitrile containing 0.1% TFA at a flow rate of 20 mLminÀ1
.
All peptides were finally obtained as TFA salts after lyophilization.
The molecular mass of the synthesized compounds was deter-
mined by using a QTrap 2000 ESI spectrometer (Applied Biosys-
tems, now Life Technologies, Carlsbad, CA). Elemental analyses
were performed by using an Elementar Vario MICRO analyzer (Ele-
mentar Analysensysteme GmbH, Hanau, Germany). 1H NMR and
13C NMR spectra were recorded by using a Jeol-ECX500 (Jeol Inc.,
Peabody, MA) and were referenced to internal solvent signals.
In the design of small-molecule drugs designed both to
bind to intracellular targets and to be orally available, it is
a common strategy to use nonpeptidic structures and to avoid
the presence or to reduce the number of highly charged
groups such as guanidines or amidines. However, it will be
challenging to achieve sufficient potency against furin-like PCs
with neutral drug molecules, although a few of these possess
micromolar potency.[14,56,57] Moreover, although excellent ami-
dine prodrugs have been developed,[58] there is no or only
little progress in the design of guanidine prodrugs.[59] There-
fore, it will likely be difficult to convert these tetrabasic inhibi-
tors into completely neutral analogues on the basis of only
prodrug strategies. All potent low-nanomolar furin inhibitors
presently known possess several basic residues, and such com-
pounds will be scarcely orally available. However, for parenteral
application it might represent no disadvantage for inhibitors
to possess more than two or three basic residues, especially
given that prior studies of oligobasic cell penetrating peptides
indicate enhanced cell entry.
The reagents for synthesis, including the standard Fmoc-protected
amino acids, coupling reagents, resins, and solvents were obtained
from Orpegen, Bachem, Iris Biotech GmbH, Fluka, Acros, or Aldrich.
3- and 4-tert-butoxycarbonyl (Boc)-aminomethylphenylacetic acids
were purchased from PolyPeptides.
4-Aminomethyl-Phac-Arg-Tle-Arg-4-amba·4TFA (18): The synthe-
sis was performed by manual standard Fmoc SPPS in a 200 mL re-
action vessel with frit starting with Fmoc-Arg(Pbf)-2-chlorotrityl
resin (2.5 g, 0.66 mmolgÀ1). Fmoc deprotection was performed
with 20% piperidine in DMF (5 and 20 min), Fmoc amino acids
(4 equiv), hydroxybenzotriazole (HOBt, 4 equiv), and N,N,N’,N’-tetra-
methyl-O-(1H-benzotriazol-1-yl)uronium
hexafluorophosphate
(HBTU, 4 equiv), and N,N-diisopropylethylamine (DIPEA, 8 equiv)
were used for coupling. For the coupling of Boc-4-aminomethyl-
phenylacetic acid only a twofold excess (i.e., 2 equiv) of the phe-
nylacetic acid derivate, HOBt, and HBTU in the presence of DIPEA
(4 equiv) was used. Protected peptide 22 (Scheme 2) was cleaved
from the resin under mild acidic conditions (80 mL 1% TFA in
CH2Cl2, 3”30 min) at room temperature. The cleavage mixture was
immediately neutralized with DIPEA after each step (3”1 mL). The
solvent was removed in vacuo (yellowish oil, HPLC: tR =28.93 min;
MS calcd: m/z=1194.58, m/z found: 1196.10 [M+H]+).
Conclusions
In summary, identification of tert-leucine as the most suitable
P3 residue provided highly potent furin inhibitors such as com-
pound 19 (MI-1148). Incubation experiments of inhibitors 17
and 19 in human plasma over a period of 10 h also revealed
the high stability of both compounds (see the Supporting In-
formation). Inhibitor MI-1148 possesses strong antiviral activi-
ties in the used influenza and canine distemper virus propaga-
tion models in cell culture and strongly protects cells against
anthrax and diphtheria toxins. Moreover, owing to their low pi-
comolar potency we expect that this inhibitor scaffold will be
able to tolerate a wide range of further modifications, which
hopefully will improve bioavailability or selectivity within the
PC family. Even after a strong drop in affinity it should still be
possible to obtain nanomolar furin inhibitors with sufficient
potency to be used as drug candidates to treat infectious
diseases.
The oily intermediate was dissolved in DMF (80 mL) and treated at
08C with 4-amidinobenzylamine·2HCl[17] (365 mg, 1.65 mmol), 6-Cl-
HOBt (840 mg, 4.95 mmol), benzotriazol-1-yl-N-oxytris(pyrrolidino)-
phosphonium hexafluorophosphate (PyBOP; 947 mg, 1.82 mmol),
and DIPEA (861 mL, 4.95 mmol). The mixture was stirred at 08C for
15 min and then at room temperature overnight. The solvent was
removed in vacuo, and the remaining intermediate was treated
with TFA/triisopropylsilane (TIS)/H2O (95:2.5:2.5 v/v/v, 20 mL) and
stirred for 3 h at room temperature. The mixture was precipitated
in cold diethyl ether, and owing to incomplete 2,2,4,6,7-pentame-
tyldihydrobenzofuran-5-sulfonyl (Pbf) removal, the cleavage proce-
dure was repeated. The crude intermediate was precipitated in di-
ethyl ether and dried. The precipitate was purified by using prepa-
rative HPLC and was lyophilized from water to afford a white solid;
yield: 1.23 g (63.2% based on initial resin loading); 1H NMR
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1228
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
(500 MHz, [D6]DMSO): d=0.87 (s, 9H), 1.34–1.64 (m, 6H), 1.64–1.76
(m, 2H), 2.98–3.17 (m, 4H), 3.45–3.58 (m, 2H), 4.00 (s, 2H), 4.16–
4.39 (m, 4H), 4.44 (dd, J=16.0, 6.3 Hz,1H), 6.82–7.62 (m, 15H),
7.71–7.87 (m, 4H), 8.12 (d, J=7.5 Hz, 1H), 8.23 (br. s., 3H), 8.41 (d,
J=8.0 Hz, 1H), 8.6 (t, J=6.16 Hz, 1H), 9.17–9.42 ppm (m, 4H);
13C NMR (125 MHz, [D6]DMSO): d=25.05, 25.14, 26.5, 28.6, 28.7,
34.3, 40.3, 40.4, 41.5, 41.7, 42.0, 52.3, 52.5, 59.5, 126.4, 127.2, 127.9,
128.6, 129.2, 132.0, 136.6, 145.7, 156.8, 165.4, 169.8, 170.2, 171.0,
171.4 ppm; MS: m/z (%): 722.1 [M+H]+; elemental analysis calcd
(%) for C43H59F12N13O12 (1178.00 gmolÀ1): C 43.84, H 5.05, N 15.46;
found: C 42.83, H 5.25, N 15.19; HPLC (l=220 nm): tR =19.12 min,
purity: 96.9%.
1
ÂÀ
Á
Ã
À
Á
=
2
*
*
*
2
Ki þ It À Et þ4  Ki  Et
2 Â Et
À Ki þ It À Et
ð1Þ
v ¼ v0 Â
The apparent Ki* was converted into the true Ki by using Equa-
tion (2):
*
Ki
Ki ¼
ð2Þ
S
1 þ K
M
Enzyme kinetic measurements with mPC1/3 and mPC2: The meas-
urements were performed in 96-well plates (Costar Corning 3365)
at 378C with a microplate reader (SpectraMaxM2, lex =380 nm and
lem =460 nm, Molecular Devices, LLC, Sunnyvale, CA). Each well
contained 40 mL of enzyme solution (for PC1/3:[61] ꢀ100 ngwellÀ1
in 100 mm sodium acetate buffer pH 5.5, containing 5 mm CaCl2,
0.1% Brij 35, 0.1% NaN3, and 0.01% BSA; for PC2:[62] ꢀ40 ngwellÀ1
in 100 mm sodium acetate buffer at pH 5, containing 2 mm CaCl2,
0.1% Brij 35, 0.1% NaN3, and 0.01% BSA). After the addition of in-
hibitor (5 mL), dissolved in the aforementioned buffers, the plate
was incubated for 20 min at room temperature. Then, the sub-
strate Pyr-RTKR-AMC (5 mL, Bachem, Bubendorf, Switzerland) dis-
solved in water (200 mm in the assay) was added to provide a total
assay volume of 50 mL. The release of free fluorescent 7-amino-4-
methylcoumarin (AMC) was measured over 30 min, and the slopes
of the progress curves were calculated by linear regression.
4-Guanidinomethyl-Phac-Arg-Tle-Arg-4-amba·4TFA (19): Inhibitor
18 (1 g, 0.85 mmol) was dissolved in 1m Na2CO3 (10 mL) and DMF
(5 mL). The mixture was treated with 1H-pyrazole-1-carboxamidi-
ne·HCl (1.25 g, 8.5 mmol) and stirred at room temperature. After
7 h, the reaction was complete, and the mixture was acidified by
the addition of TFA (pHꢀ3). The crude product was purified by
preparative HPLC, and the product was lyophilized from water to
afford
a
white solid; yield: 650 mg (0.533 mmol); 1H NMR
(500 MHz, [D6]DMSO) d=0.86 (s, 9H), 1.37–1.63 (m, 6H), 1.63–1.76
(m, 2H), 3.00–3.17 (m, 4H), 3.43–3.56 (m, 2H), 4.20–4.39 (m, 6H),
4.44 (dd, J=16.18, 6.16 Hz, 1H), 6.76–7.62 (m, 19H), 7.72–7.83 (m,
4H), 8.07–8.18 (m, 2H), 8.41 (d, J=8.02 Hz, 1H), 8.56–8.64 (m, 1H),
9.21–9.41 ppm (m, 4H); 13C NMR (125 MHz, [D6]DMSO) d=25.1,
25.2, 26.5, 28.5, 28.7, 34.3, 40.3, 40.4, 41.5, 41.7, 43.7, 52.3, 52.5,
59.5, 126.5, 127.1, 127.2, 127.9, 129.2, 135.2, 135.5, 145.7, 156.8,
156.9, 165.4, 169.8, 170.4, 171.0, 171.4 ppm; MS: m/z (%): 764,35
[M+H]+; elemental analysis calcd (%) for C44H61F12N15O12
(1220.03 gmolÀ1): C 43.32, H 5.04, N 17.22; found: C 42.21, H 5.31,
N 17.07; HPLC: tR =20.41 min, purity: 97%.
Expression, crystallization, and structure determination: Expression,
purification of human furin, and its crystallization in complex with
inhibitor I-1 (m-guanidinomethyl-Phac-Arg-Val-Arg-4-amidinobenzy-
lamid) were performed as previously described.[21] Shortly, the pro-
tease was expressed by transient transfection of human embryonic
kidney cells. Three chromatography steps including immobilized
metal affinity purification, immobilized inhibitor affinity chromatog-
raphy, and gel-permeation chromatography were applied for the
purification of furin.
The other inhibitors were prepared by an identical strategy;[17,20,22]
for the analytical data see Table 1.
Initially, crystals of human furin were grown in complex with I-1.
These crystals were washed with the crystallization solution (5”
0.5 mL) supplemented with 1 mm of compound 19 and were finally
soaked for 3 days. The crystals were transferred to soaking solution
supplemented with 15% ethylene glycol, flash cooled in liquid ni-
trogen, and used for diffraction data collection at the BESSY-II
beamline 14.1 of the Helmholtz-Zentrum Berlin (HZB).[63] Data proc-
essing was performed with XDS[64] (v.03/2013) and programs of the
CCP4-suite[65] (CCP4 v.6.3.0, CCP4 interface v.2.2.0). The Rfree set of
reflections was defined in thin shells with the DATAMAN soft-
ware.[66] CNS (v.1.3)[67] was used for initial rigid body refinement
with the structure of human furin (PDB-ID 4OMC[21]). Model build-
ing was performed with in COOT (v.0.6.2).[68] CNS (v.1.3) was used
for refinement by applying tight NCS restraints to the main chain
atoms of the protein. Refinement parameters of the p-guanidino-
methyl-Phac, Tle, and 4-amba moieties of the inhibitor, including
bond length, bond angels, dihedrals, and force constants, were de-
rived from small-molecule structures of the Cambridge structural
database.[69] MOLEMAN[70] was used for analysis of B-factors. Kicked
omit maps were calculated in PHENIX.[71] PYMOL (http://www.
pymol.org) was used for molecular graphics and structural align-
ments. Surface electrostatics were calculated with APBS[72] and
were visualized with the APBS PYMOL plugin.
Enzyme-based assays
Enzyme kinetic measurements with recombinant soluble human
furin: The measurements were performed in black 96-well plates
(Nunc, Langenselbold) at room temperature with a microplate
reader (Safire2, Tecan, Switzerland) at lex =380 nm and lem
=
460 nm. Each well contained 2 mL inhibitor solution (dissolved in
DMSO), 20 mL of Phac-Arg-Val-Arg-Arg-AMC as substrate (dissolved
in water, concentrations used in the assay: 5, 20 and 50 mm) and
160 mL buffer [100 mm 4-(2-hydroxyethyl)-1-piperazineethanesul-
fonic acid (HEPES), 0.2% Triton X-100, 2 mm CaCl2, 0.02% NaN3,
and 1 mgmLÀ1 bovine serum albumin (BSA), pH 7.0]. The measure-
ments were started by the addition of furin[15] (20 mL) solution. For
inhibitors with neutral P5 residues, the lowest inhibitor concentra-
tion in the assay was 10 times higher than the enzyme concentra-
tion. The measurements were performed for 30 min, and the
steady-state rates were calculated from the slopes of the progress
curves. The Ki values were obtained by fitting the data to the equa-
tion for classical reversible competitive inhibition, as described pre-
viously.[17] All data calculations were performed with Origin 8.1.
For tight-binding inhibitors with basic P5 residues Equation (1)[60]
was used, in which v0 is the velocity in the absence of an inhibitor,
It is the total inhibitor concentration, Et is the total enzyme concen-
tration, and Ki* is the apparent inhibition constant at the used sub-
strate concentration.
Thermal denaturation assays: Thermal denaturation analysis (also
called thermoflour or thermal shift assay[26]) was performed in
40 mm sodium borate, pH 7.5, 100 mm NaCl, 2 mm CaCl2, and 15”
Sypro-Orange (Life Technologies), adding 1.5 mm of the protease
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1229
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
and 5 mm of the inhibitors. The final reaction volume was 25 mL.
Melting curves were determined with an IQ5 realtime PCR cycler
(Biorad) by measuring the increase in the fluorescence at excitation
and emission wavelengths of 490 and 575 nm, respectively. Heat-
ing was performed in 0.58C steps with a subsequent dwell time of
15 s. The melting temperatures correspond to the inflection points
of the melting curves, which were determined with the data analy-
sis module of the IQ5 software (Bio-Rad, v2.1.97.1001). All measure-
ments were performed in triplicate.
Inhibition of canine distemper virus propagation: VeroSLAMdogtag
cells[42] were seeded in 12-well plates at 80% confluency. Eight
hours prior to infection, the culture medium was changed to
DMEM media containing the respective inhibitor at the appropriate
concentration or untreated medium. Cells were then infected with
CDV 5804PeH[27] at a MOI of 0.01, and supernatant and cell-associ-
ated samples were collected every 24 h for 3 days. The medium of
all wells was changed daily to assure the presence of the inhibitor
throughout the experiment. Titers were quantified by using the
limited dilution method and are expressed as 50 tissue culture
infectious doses (TCID50).
Assays in cell culture
Inhibition of anthrax toxin action: Murine macrophage J774A.1 cells
(200 mL, 5”105 cellsmLÀ1; German collection of microorganisms
and cell cultures, Braunschweig, Germany) were grown overnight
in wells of a 96-well plate (Becton Dickinson Labware, Heidelberg)
by using Roswell Park Memorial Institute (RPMI) medium/Dulbec-
co’s modified Eagle’s medium (DMEM) (1:1) containing 10% fetal
calf serum (FCS) at 378C and 5% CO2 as a seeding medium Then
the cells were treated with a prepared solution of protective anti-
gen (PA, 500 ngmLÀ1 in assay), lethal factor (LF, 100 ngmLÀ1 in
assay), and various inhibitor concentrations (0.05–100 mm in the
assay) dissolved in sample medium containing 95% DMEM and
5% FCS. PA and LF were obtained from List Biological Laboratories,
Inc., Campbell, USA. The cells were incubated at 378C and 5% CO2
for 3 h. Afterwards, 25 mL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphe-
nyltetrazolium bromide (MTT) solution [5 mgmLÀ1 in phosphate-
buffered saline (PBS)] was added to each well, and the cells were
incubated in the dark at 378C with 5% CO2 for 2 h. After cell lysis
with 90% isopropanol (v/v, 100 mL), 0.5% sodium dodecyl sulfate
(SDS) (w/v), and 25 mm HCl, the absorbance of each well was mea-
sured by using a microplate reader (Model 680, Biorad Laborato-
ries, Hercules, CA) at l=540 nm.[73] The resulting measurements
were normalized through subtraction of the mean of the blanks
(no cells, sample medium with toxin). The protection was estimat-
ed as a percentage of the medium control (cells with sample
medium devoid of toxin) with the toxin control (cells with sample
medium and toxin) as zero value [protection=(sampleÀtoxin con-
trol)/(medium controlÀtoxin control)”100].
Inhibition of diphtheria toxin action: The Vero cells used were main-
tained in DMEM.[74] The cells were seeded into 24-well plates at
a density of 5”104 cellswellÀ1 1 day before the experiment. The
cells were preincubated with or without the inhibitors (100 nm in
the prescreen and 2, 10, and 50 nm in case of the concentration
dependency) in leucine-free and serum-free medium for 30 min,
followed by the addition of various concentrations of unnicked
diphtheria toxin (Merck KGaA, Darmstadt, Germany; 0.1, 1, 10, and
100 ngmLÀ1 in the assay). After 3 h, the cells were incubated in
HEPES medium containing 1 mCimLÀ1 [3H]leucine (Hartmann Ana-
lytic, Braunschweig, Germany) without unlabeled leucine for
20 min. Then, the medium was removed, the cells were washed
with 5% trichloroacetic acid (2”), dissolved in 0.1m KOH, and the
radioactivity was measured.
Acknowledgements
The work of W.G. was supported by the Collaborative Research
Centre 593 of the German Research Foundation (DFG).
Keywords: antiviral agents · crystal structure analysis · furin ·
inhibitors · proprotein convertases
[1] G. Thomas, Nat. Rev. Mol. Cell Biol. 2002, 3, 753–766.
[2] N. G. Seidah, A. Prat, Nat. Rev. Drug Discovery 2012, 11, 367–383.
[3] A. J. Roebroek, L. Umans, I. G. Pauli, E. J. Robertson, F. van Leuven, W. J.
Van de Ven, D. B. Constam, Development 1998, 125, 4863–4876.
[4] J. W. Creemers, A. M. Khatib, Front. Biosci. 2008, 4960–4971.
[5] A. J. Roebroek, N. A. Taylor, E. Louagie, I. Pauli, L. Smeijers, A. Snellinx, A.
Lauwers, W. J. Van de Ven, D. Hartmann, J. W. Creemers, J. Biol. Chem.
2004, 279, 53442–53450.
[6] M. S. Sarac, A. Cameron, I. Lindberg, Infect. Immun. 2002, 70, 7136–
7139.
[7] M. S. Sarac, J. R. Peinado, S. H. Leppla, I. Lindberg, Infect. Immun. 2004,
72, 602–605.
[8] F. Couture, F. D’Anjou, R. Day, Biomol. Concepts 2011, 2, 421–438.
[9] A. Pasquato, J. Ramos da Palma, C. Galan, N. G. Seidah, S. Kunz, Antiviral
Res. 2013, 99, 49–60.
[10] N. Senzer, M. Barve, J. Kuhn, A. Melnyk, P. Beitsch, M. Lazar, S. Lifshitz,
M. Magee, J. Oh, S. W. Mill, C. Bedell, C. Higgs, P. Kumar, Y. Yu, F. Norvell,
C. Phalon, N. Taquet, D. D. Rao, Z. Wang, C. M. Jay, B. O. Pappen, G. Wall-
raven, F. C. Brunicardi, D. M. Shanahan, P. B. Maples, J. Nemunaitis, Mol.
Ther. 2012, 20, 679–686.
[11] N. Senzer, M. Barve, J. Nemunaitis, J. Kuhn, A. Melnyk, P. Beitsch, M.
Magee, J. Oh, C. Bedell, P. Kumar, D. D. Rao, B. O. Pappen, G. Wallraven,
C. Brunicardi, P. B. Maples, J. Nemunaitis, J. Vaccines Vaccin. 2013, 4,
209.
[12] J. Nemunaitis, M. Barve, D. Orr, J. Kuhn, M. Magee, J. Lamont, C. Bedell,
G. Wallraven, B. O. Pappen, A. Roth, S. Horvath, D. Nemunaitis, P. Kumar,
P. B. Maples, N. Senzer, Oncology 2014, 87, 21–29.
[13] A. Basak, J. Mol. Med. 2005, 83, 844–855.
[14] U. C. De, P. Mishra, P. R. Pal, B. Dinda, A. Basak, Colloquium Series on Pro-
tein Activation and Cancer 2012, 1, 1–76.
[15] M. M. Kacprzak, J. R. Peinado, M. E. Than, J. Appel, S. Henrich, G. Lipkind,
R. A. Houghten, W. Bode, I. Lindberg, J. Biol. Chem. 2004, 279, 36788–
36794.
[16] G. S. Jiao, L. Cregar, J. Wang, S. Z. Millis, C. Tang, S. O’Malley, A. T. John-
son, S. Sareth, J. Larson, G. Thomas, Proc. Natl. Acad. Sci. USA 2006, 103,
19707–19712.
[17] G. L. Becker, F. Sielaff, M. E. Than, I. Lindberg, S. Routhier, R. Day, Y. Lu,
W. Garten, T. Steinmetzer, J. Med. Chem. 2010, 53, 1067–1075.
[18] H. Gagnon, S. Beauchemin, A. Kwiatkowska, F. Couture, F. D’Anjou, C.
Levesque, F. Dufour, A. R. Desbiens, R. Vaillancourt, S. Bernard, R. Desjar-
dins, F. Malouin, Y. L. Dory, R. Day, J. Med. Chem. 2014, 57, 29–41.
[19] A. Kwiatkowska, F. Couture, C. Levesque, K. Ly, R. Desjardins, S. Beauche-
min, A. Prahl, B. Lammek, W. Neugebauer, Y. L. Dory, R. Day, J. Med.
Chem. 2014, 57, 98–109.
Inhibition of influenza virus propagation: MDCK II cells were seeded
in 24-well plates and grown to 100% confluency. Cells were infect-
ed with A/FPV/Rostock/34 (H7N1) or A/Thailand/1(KAN-1)/2004
(H5N1) at a multiplicity of infection (MOI) of 0.001 at 378C for 1 h.
After that, virus-containing media were removed and cells were
washed with PBS pH 7.0.[39] Fresh media containing inhibitors at in-
dicated concentrations were added to the cells. Cells were incubat-
ed in the presence and absence of inhibitors for 24 h at 378C. In-
fected cells were immunostained by using anti-FPV-antibodies.
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1230
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[20] G. L. Becker, Y. Lu, K. Hardes, B. Strehlow, C. Levesque, I. Lindberg, K.
Sandvig, U. Bakowsky, R. Day, W. Garten, T. Steinmetzer, J. Biol. Chem.
2012, 287, 21992–22003.
[21] S. O. Dahms, K. Hardes, G. L. Becker, T. Steinmetzer, H. Brandstetter, M. E.
Than, ACS Chem. Biol. 2014, 9, 1113–1118.
J. M. Brisson, D. Prelusky, W. Korfmacher, R. White, S. Bogdanowich-
Knipp, A. Pavlovsky, P. Bradley, A. K. Saksena, A. Ganguly, J. Piwinski, V.
Girijavallabhan, F. G. Njoroge, J. Med. Chem. 2006, 49, 6074–6086.
[48] S.-H. Chen, S.-L. Tan, Curr. Med. Chem. 2005, 12, 2317–2342.
[49] B. S. Robinson, K. A. Riccardi, Y. F. Gong, Q. Guo, D. A. Stock, W. S. Blair,
B. J. Terry, C. A. Deminie, F. Djang, R. J. Colonno, P. F. Lin, Antimicrob.
Agents Chemother. 2000, 44, 2093–2099.
[22] G. L. Becker, K. Hardes, T. Steinmetzer, Bioorg. Med. Chem. Lett. 2011, 21,
4695–4697.
[23] S. Henrich, A. Cameron, G. P. Bourenkov, R. Kiefersauer, R. Huber, I. Lind-
berg, W. Bode, M. E. Than, Nat. Struct. Biol. 2003, 10, 520–526.
[24] M. S. Bernatowicz, Y. Wu, G. R. Matsueda, J. Org. Chem. 1992, 57, 2497–
2502.
[50] A. S. Bommarius, M. Schwarm, K. Stingl, M. Kottenhahn, K. Huthmacher,
K. Drauz, Tetrahedron: Asymmetry 1995, 6, 2851–2888.
[51] F. Formaggio, C. Baldini, V. Moretto, M. Crisma, B. Kaptein, Q. B. Broxter-
man, C. Toniolo, Chem. Eur. J. 2005, 11, 2395–2404.
[25] G. N. Ramachandran, V. Sasisekharan, Adv. Protein Chem. 1968, 23, 283–
438.
[52] S. M. Fuchs, R. T. Raines, Cell. Mol. Life Sci. 2006, 63, 1819–1822.
[53] F. Milletti, Drug Discovery Today 2012, 17, 850–860.
[26] U. B. Ericsson, B. M. Hallberg, G. T. Detitta, N. Dekker, P. Nordlund, Anal.
Biochem. 2006, 357, 289–298.
[54] S. Katayama, H. Hirose, K. Takayama, I. Nakase, S. Futaki, J. Controlled Re-
lease 2011, 149, 29–35.
[27] V. von Messling, D. Milosevic, R. Cattaneo, Proc. Natl. Acad. Sci. USA
2004, 101, 14216–14221.
[28] T. C. Dixon, M. Meselson, J. Guillemin, P. C. Hanna, N. Engl. J. Med. 1999,
341, 815–826.
[29] S. Liu, M. Moayeri, S. H. Leppla, Trends Microbiol. 2014, 22, 317–325.
[30] J. A. Young, R. J. Collier, Annu. Rev. Biochem. 2007, 76, 243–265.
[31] A. W. Artenstein, S. M. Opal, Clin. Infect. Dis. 2012, 54, 1148–1161.
[32] S. Mazumdar, MAbs 2009, 1, 531–538.
[33] T. L. Hadfield, P. McEvoy, Y. Polotsky, V. A. Tzinserling, A. A. Yakovlev, J.
Infect. Dis. 2000, 181, S116–S120.
[55] B. D. Grant, J. G. Donaldson, Nat. Rev. Mol. Cell Biol. 2009, 10, 597–608.
[56] T. Komiyama, J. M. Coppola, M. J. Larsen, M. E. van Dort, B. D. Ross, R.
Day, A. Rehemtulla, R. S. Fuller, J. Biol. Chem. 2009, 284, 15729–15738.
[57] J. M. Coppola, M. S. Bhojani, B. D. Ross, A. Rehemtulla, Neoplasia 2008,
10, 363–370.
[58] P. Ettmayer, G. L. Amidon, B. Clement, B. Testa, J. Med. Chem. 2004, 47,
2393–2404.
[59] D. Schade, J. Kotthaus, L. Riebling, H. Muller-Fielitz, W. Raasch, O. Koch,
N. Seidel, M. Schmidtke, B. Clement, J. Med. Chem. 2014, 57, 759–769.
[60] J. W. Williams, J. F. Morrison, Methods Enzymol. 1979, 63, 437–467.
[61] Y. Zhou, I. Lindberg, J. Biol. Chem. 1993, 268, 5615–5623.
[62] N. S. Lamango, X. Zhu, I. Lindberg, Arch. Biochem. Biophys. 1996, 330,
238–250.
[63] U. Mueller, N. Darowski, M. R. Fuchs, R. Forster, M. Hellmig, K. S. Paithan-
kar, S. Puhringer, M. Steffien, G. Zocher, M. S. Weiss, J. Synchrotron
Radiat. 2012, 19, 442–449.
[64] W. Kabsch, Acta Crystallogr. Sect. D 2010, 66, 125–132.
[65] M. D. Winn, C. C. Ballard, K. D. Cowtan, E. J. Dodson, P. Emsley, P. R.
Evans, R. M. Keegan, E. B. Krissinel, A. G. Leslie, A. McCoy, S. J. McNicho-
las, G. N. Murshudov, N. S. Pannu, E. A. Potterton, H. R. Powell, R. J.
Read, A. Vagin, K. S. Wilson, Acta Crystallogr. Sect. D 2011, 67, 235–242.
[66] G. J. Kleywegt, T. A. Jones, Acta Crystallogr. Sect. D 1996, 52, 826–828.
[67] A. T. Brunger, Nat. Protoc. 2007, 2, 2728–2733.
[34] M. Tsuneoka, K. Nakayama, K. Hatsuzawa, M. Komada, N. Kitamura, E.
Mekada, J. Biol. Chem. 1993, 268, 26461–26465.
[35] C. J. Russell, R. G. Webster, Cell 2005, 123, 368–371.
[36] H. D. Klenk, W. Garten, Trends Microbiol. 1994, 2, 39–43.
[37] W. Garten, H. D. Klenk in Avian Influenza, Vol. 27 (Eds.: H. D. Klenk, M.
Matrosovich, J. Stech), Karger, Basel, 2008, pp. 156–167.
[38] E. Bçttcher, C. Freuer, T. Steinmetzer, H. D. Klenk, W. Garten, Vaccine
2009, 27, 6324–6329.
[39] E. Bçttcher-Friebertshäuser, C. Freuer, F. Sielaff, S. Schmidt, M. Eickmann,
J. Uhlendorff, T. Steinmetzer, H. D. Klenk, W. Garten, J. Virol. 2010, 84,
5605–5614.
[40] V. von Messling, D. Milosevic, P. Devaux, R. Cattaneo, J. Virol. 2004, 78,
7894–7903.
[41] A. Beineke, C. Puff, F. Seehusen, W. Baumgartner, Vet. Immunol. Immuno-
pathol. 2009, 127, 1–18.
[68] P. Emsley, B. Lohkamp, W. G. Scott, K. Cowtan, Acta Crystallogr. Sect. D
2010, 66, 486–501.
[42] V. von Messling, C. Springfeld, P. Devaux, R. Cattaneo, J. Virol. 2003, 77,
12579–12591.
[69] F. H. Allen, Acta Crystallogr. Sect. B 2002, 58, 380–388.
[70] G. J. Kleywegt, J. Mol. Biol. 1997, 273, 371–376.
[43] W. Garten, S. Hallenberger, D. Ortmann, W. Schäfer, M. Vey, H. Angliker,
E. Shaw, H. D. Klenk, Biochimie 1994, 76, 217–225.
[44] S. Raghothama, M. Chaddha, P. Balaram, Proc. Natl. Acad. Sci., India,
Sect. A 1996, 66, 33–44.
[45] J. J. Perona, C. S. Craik, Protein Science 1995, 4, 337–360.
[46] P. K. Madala, J. D. Tyndall, T. Nall, D. P. Fairlie, Chem. Rev. 2010, 110, PR1–
PR31.
[71] P. D. Adams, P. V. Afonine, G. Bunkoczi, V. B. Chen, I. W. Davis, N. Echols,
J. J. Headd, L. W. Hung, G. J. Kapral, R. W. Grosse-Kunstleve, A. J. McCoy,
N. W. Moriarty, R. Oeffner, R. J. Read, D. C. Richardson, J. S. Richardson,
T. C. Terwilliger, P. H. Zwart, Acta Crystallogr. Sect. D 2010, 66, 213–221.
[72] N. A. Baker, D. Sept, S. Joseph, M. J. Holst, J. A. McCammon, Proc. Natl.
Acad. Sci. USA 2001, 98, 10037–10041.
[73] T. Mosmann, J. Immunol. Methods 1983, 65, 55–63.
[47] S. Venkatraman, S. L. Bogen, A. Arasappan, F. Bennett, K. Chen, E. Jao,
Y. T. Liu, R. Lovey, S. Hendrata, Y. Huang, W. Pan, T. Parekh, P. Pinto, V.
Popov, R. Pike, S. Ruan, B. Santhanam, B. Vibulbhan, W. Wu, W. Yang, J.
Kong, X. Liang, J. Wong, R. Liu, N. Butkiewicz, R. Chase, A. Hart, S.
Agrawal, P. Ingravallo, J. Pichardo, R. Kong, B. Baroudy, B. Malcolm, Z.
Guo, A. Prongay, V. Madison, L. Broske, X. Cui, K. C. Cheng, Y. Hsieh,
[74] A. Kurmanova, A. Llorente, A. Polesskaya, O. Garred, S. Olsnes, J. Kozlov,
K. Sandvig, Biochem. Biophys. Res. Commun. 2007, 357, 144–149.
Received: March 5, 2015
Published online on May 14, 2015
ChemMedChem 2015, 10, 1218 – 1231
www.chemmedchem.org
1231
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim