%4/<+89/:?5,+(8'91'/4)524%4/<+89/:?5,+(8'91'/4)524
/-/:'2533549%4/<+89/:?5,+(8'91'/4)524/-/:'2533549%4/<+89/:?5,+(8'91'/4)524
%#83?"+9+'8). %#+6'8:3+4:5,+,+49+
"+A4+*!.'83')56.58+*+4:/A+9!5:+4:"+A4+*!.'83')56.58+*+4:/A+9!5:+4:
3/45).25857;/452/4+'9+*4./(/:5895,:.+5:;2/4;33/45).25857;/452/4+'9+*4./(/:5895,:.+5:;2/4;3
+;85:5>/4#+85:?6++:'225685:+'9++;85:5>/4#+85:?6++:'225685:+'9+
'3+9;84+::
+0'4 69+4/)'
'3'8'0#8/8'-.'<'4
"+1.'!'4).'2
58*54";:.+2
#++4+>:6'-+,58'**/:/54'2';:.589
5225=:./9'4*'**/:/54'2=5819':.::69*/-/:'2)533549;42+*;;9'83?8+9+'8).
!'8:5,:.+ 6+8':/549"+9+'8).#?9:+394-/4++8/4-'4*4*;9:8/'24-/4++8/4-533549
;84+::'3+9 69+4/)'+0'4#8/8'-.'<'4'3'8'0!'4).'2"+1.'";:.+258*54+8354+
44"-;?+4$'3+44?$'8''4+5;-2'9)8':.54458#).3/*:'3+9
&+44+89:85354':.'4;99/5"/)1#52'0'5-*'4'4*'<'8/#/4'"+A4+*!.'83')56.58+
*+4:/A+9!5:+4:3/45).25857;/452/4+'9+*4./(/:5895,:.+5:;2/4;3+;85:5>/4#+85:?6+
+:'225685:+'9+
%#83?"+9+'8).
.::69*/-/:'2)533549;42+*;;9'83?8+9+'8).
$./98:/)2+/9(85;-.::5?5;,58,8++'4*56+4'))+99(?:.+%#+6'8:3+4:5,+,+49+':
/-/:'2533549%4/<+89/:?5,+(8'91'/4)524:.'9(++4'))+6:+*,58/4)2;9/54/4%#83?"+9+'8).(?'4
';:.58/@+*'*3/4/9:8':585,/-/:'2533549%4/<+89/:?5,+(8'91'/4)524
;:.589;:.589
'3+9;84+::+0'4 69+4/)''3'8'0#8/8'-.'<'4"+1.'!'4).'258*54";:.+244"
+8354+$'3-;?+4$'8'+44?5;-2'9'4+54458)8':.'3+9#).3/*:
54':.'4&+44+89:853"/)1;99/55-*'4#52'0''4*#/4''<'8/
$./9'8:/)2+/9'<'/2'(2+':/-/:'2533549%4/<+89/:?5,+(8'91'/4)524.::69*/-/:'2)533549;42+*;
;9'83?8+9+'8).
A Refined Pharmacophore Identifies Potent 4-Amino-7-chloroquinoline-Based Inhibitors of the
Botulinum Neurotoxin Serotype A Metalloprotease
James C. Burnett,
†
Dejan Opsenica,
‡
Kamaraj Sriraghavan,
§
Rekha G. Panchal,
†
Gordon Ruthel,
|
Ann R. Hermone,
†
Tam L. Nguyen,
†
Tara A. Kenny,
†
Douglas J. Lane,
†
Connor F. McGrath,
†
James J. Schmidt,
|
Jonathan L. Vennerstrom,
§
Rick Gussio,
⊥
Bogdan A. Sˇolaja,*
,#
and Sina Bavari*
,|
SAIC-Frederick, Inc., Target Structure-Based Drug DiscoVery Group, Frederick, Frederick, Inc., National Cancer Institute at Frederick,
P.O. Box B, F.V.C. 310, Frederick, Maryland 21702, The Institute of Chemistry, Technology, and Metallurgy, NjegosˇeVa 12, YU-11001
Belgrade, Serbia, College of Pharmacy, 986025 UniVersity of Nebraska Medical Center, Omaha, Nebraska 68198, U.S. Army Medical
Research Institute of Infectious Diseases, Fort Detrick, 1425 Porter Street, Frederick, Maryland 21702, DeVelopmental Therapeutics Program,
P.O. Box B, F.V.C. 310, NCI Frederick, Frederick, Maryland 21702, and Faculty of Chemistry, The UniVersity of Belgrade, Studentski trg 16,
P.O. Box 158, YU-11001 Belgrade, Serbia
ReceiVed December 19, 2006
We previously identified structurally diverse small molecule (non-peptidic) inhibitors (SMNPIs) of the
botulinum neurotoxin serotype A (BoNT/A) light chain (LC). Of these, several (including antimalarial drugs)
contained a 4-amino-7-chloroquinoline (ACQ) substructure and a separate positive ionizable amine component.
The same antimalarials have also been found to interfere with BoNT/A translocation into neurons, via pH
elevation of the toxin-mediated endosome. Thus, this structural class of small molecules may serve as dual-
function BoNT/A inhibitors. In this study, we used a refined pharmacophore for BoNT/A LC inhibition to
identify four new, potent inhibitors of this structural class (IC
50
’s ranged from 3.2 to 17 µM). Molecular
docking indicated that the binding modes for the new SMNPIs are consistent with those of other inhibitors
that we have identified, further supporting our structure-based pharmacophore. Finally, structural motifs of
the new SMNPIs, as well as two structure-based derivatives, were examined for activity, providing valuable
information about pharmacophore component contributions to inhibition.
Introduction
Botulinum neurotoxins (BoNTs)
a
are the most potent of the
biological toxins; the lethal dose of BoNT serotype A (BoNT/
A) is estimated to be between 1 and 5 ng kg
-1
for humans.
1,2
As a result, these enzymes, which are responsible for the
paralysis associated with botulism, are listed as category A
(highest priority) biothreat agents by the Centers for Disease
Control and Prevention (CDC). BoNTs are easily produced and
may be delivered via food “spiking” and/or aerosol route.
2-6
Furthermore, as BoNTs are now used to treat a range of medical
conditions, and in many cosmetic applications,
3,7-14
they are
being produced in increasing quantities, making their misuse,
accidental overdosing, and/or instances of adverse side effects
15
more likely. Neither the currently available CDC BoNT equine
antitoxins, which can cause adverse anaphylaxis and serum
sickness,
16
nor experimental antibodies can counter these
enzymes once they are inside neurons. Furthermore, BoNT
intoxication can occur rapidly,
17
and individuals who have been
maliciously exposed to a BoNT(s), or have received an
accidental overdose, will most likely seek medical attention only
after clinical symptoms (i.e., muscle paralysis) manifest. At this
time, critical care mechanical ventilation is the only life-saving
option once diaphragm muscles cease to function. Yet, the
effects of internalized BoNTs can last for weeks,
18,19
rendering
such medical care impractical for wide scale application. By
comparison, small molecule (non-peptidic) inhibitors (SMNPIs)
could serve as post-intoxication “rescue” therapeutics and
prophylactics.
There are seven known BoNT serotypes (identified as A-F).
Each cleaves a component of the SNARE (soluble N-ethylma-
leimide-sensitive factor attachment protein receptor) com-
plex,
20,21
which facilitates the transport of acetylcholine into
neuromuscular junctions. BoNT serotypes A and E cleave
SNAP-25 (synaptosomal-associated protein (25 kDa)),
22
sero-
types B, D, F,
23
and G cleave VAMP (vesicle-associated
membrane protein),
24-27
and serotype C1 cleaves both SNAP-
25 and syntaxin 1.
28
X-ray crystal structures of BoNT holotoxins
29,30
show that
these enzymes are composed of a heavy chain (HC) and a light
chain (LC), which, following post-translational modification,
are connected by a disulfide bridge.
29
The LC is a zinc
metalloprotease. The HC transports the LC into the neuronal
cytosol via an acidic endosome.
In a previous publication,
31
we identified a range of novel,
structurally diverse, non-zinc chelating SMNPIs of the BoNT/A
LC. In particular, we showed that several compounds containing
the weakly basic (pK
a
≈ 8) ACQ substructure and a separate
ionizable amine component were among the most potent of the
inhibitors.
31
Additionally, we proposed binding modes for the
compounds and developed a common pharmacophore for
BoNT/A LC inhibition.
31
In a subsequent molecular dynamics
study, we demonstrated that conformationally flexible loops
surrounding the BoNT/A LC substrate binding cleft may reorient
to decrease the solvent accessibility of the cleft (as compared
to respective energy refined structures), while simultaneously
providing more hydropathically feasible binding contacts for
* To whom correspondence should be addressed. (B.A.S.) Phone: (+381-
11) 638-606. Fax: (+381-11) 638-061. E-mail: bsolaja@chem.bg.ac.yu.
(S.B.) Phone: (301) 619-4246. Fax: (301) 619-2348. E-mail:
sina.bavari@us.army.mil.
†
National Cancer Institute at Frederick.
‡
The Institute of Chemistry, Technology, and Metallurgy.
§
University of Nebraska Medical Center.
|
U.S. Army Medical Research Institute of Infectious Diseases.
⊥
Developmental Therapeutics Program.
#
The University of Belgrade.
a
Abbreviations: BoNTs, botulinum neurotoxins; BoNT/A, botulinum
neurotoxin serotype A; BoNT/A LC, BoNT/A light chain; SMNPIs, small
molecule (non-peptidic) inhibitors; ACQ, 4-amino-7-chloroquinoline.
2127J. Med. Chem. 2007, 50, 2127-2136
10.1021/jm061446e CCC: $37.00 © 2007 American Chemical Society
Published on Web 04/07/2007
This article is a U.S. government work, and is not subject to copyright in the United States.
SMNPIs.
32
The molecular dynamics studies were pivotal for
identifying a binding mode
33
for the potent pseudo-peptide
inhibitor 2-mercapto-3-phenylpropionyl-RATKML (mpp-RAT-
KML, K
i
) 330 nM
34
). In the same study, we also proposed
new pharmacophore components and constraints based on the
docked model of mpp-RATKML, and, using this information,
we identified more potent SMNPIs.
33
In this study, we build
on our previous work by describing how the refined pharma-
cophore
33
was used to discover new small molecule inhibitors
possessing the ACQ substructure and a separate ionizable,
aliphatic amine component.
Results and Discussion
Based on the identification of N,N-bis(7-chloroquinolin-4-
yl)diamines and five antimalarial drugs (amodiaquine, chloro-
quine, quinacrine, quinidine, and quinine) as SMNPIs of the
BoNT/A LC,
31
we have continued to pursue the identification
of new inhibitors of this structural class possessing the weakly
basic ACQ substructure and an ionizable amine; both of these
components have been found to be key to activity.
31
Further
supporting this approach are the findings of Deshpande et al.
35
and Sheridan et al.,
36
demonstrating that ACQ-based drugs,
which also possess a separate ionizable nitrogen atom (e.g.,
amodiaquine), delay BoNT/A induced muscle paralysis. In those
studies, it was hypothesized that the compounds act by interfer-
ing with toxin translocation, most likely by increasing the pH
of the toxin-mediated endosome.
35,36
This is a logical supposition
as ACQ antimalarials have been shown to target and increase
the pH of the acidic food vacuole of Plasmodium falciparum.
37-39
However, neither of the earlier studies reported LC inhibition
in vitro. As indicated above, we demonstrated that several of
the same antimalarials also directly inhibit BoNT/A LC met-
alloprotease activity;
31
however, they do so at higher concentra-
tions than reported in the muscle twitch assays used in the earlier
studies.
35,36
This comparison indicates that the observed muscle
twitch inhibition could not have resulted solely from inhibition
of the enzyme’s LC. Thus, the Deshpande et al.
35
and Sheridan
et al.
36
data, in combination with our previous findings, indicate
that such molecules may serve as dual-function countermeasures,
acting both as translocation inhibitors and directly against the
metalloprotease activity of the LC.
A Refined Pharmacophore Is Used To Identify New
BoNT/A LC Inhibitors Possessing Both the ACQ Substruc-
ture and a Separate Ionizable Amine. The first step in
identifying new SMNPIs was to use a recently refined/expanded
pharmacophore for BoNT/A LC inhibition
33
(Figure 1A) to
evaluate diverse target compounds possessing the ACQ sub-
structure. It is important to note that the eight-component
pharmacophore model presented in Figure 1A is a logical
extension of the six-component model shown in our original
work
31
and is part of an ongoing refinement of our pharma-
cophore for BoNT/A LC inhibition
32,33
(which is occurring as
more data become available). In this study, we wanted to identify
compounds containing as many pharmacophore features as
possible, but with the following specific components: (1) the
ACQ substructure (which incorporates pharmacophore compo-
nents A, a heteroatom associated with A, and C (Figure 1A)),
(2) an ionizable amine located 6.5-9.5 Å from the centroid of
the ACQ (pharmacophore component E (Figure 1A)), and (3)
at least one of the new components from our expanded
pharmacophore, either F
33
(a positive ionizable moiety located
11.7-16.7 Å from the quinoline centroid) or G
33
(a hydrophobic
moiety located 8.5-12.5 Å from the quinoline centroid) (Figure
1A). Subsequently, a congeneric series of 4-amino-7-chloro-
quinoline-cholate-acetates, 1-3, and a tri-ACQ derivative, 4,
were identified (see Scheme 1 for two-dimensional (2-D)
structures of 1-3 and Figure 2 for the 2-D structure of 4).
Pharmacophore mapping to 1-4 indicated that in each case
ACQ is plane A (Figure 1B-E), and the quinoline ring nitrogen
is the heteroatom associated with this plane. For 1-3, compo-
nent B is the central decahydronapthalene of the methyl cholate
acetate (rings B and C according to steroid ring designation)
(Figure 1B-D), while in 4 it is a second ACQ (Figure 1E).
Hydrophobic component C is the 7-chloro substituent of the
plane A ACQ for all inhibitors (Figure 1B-E), and pharma-
cophore component D is a methyl for 1 (Figure 1B), an acetyl
group for 2 and 3 (Figure 1C and D), and a chloro substituent
for 4 (Figure 1E).
As a second specific criterion, each of the molecules contains
a separate ionizable, aliphatic amine (Figure 1B-E), which
corresponds to pharmacophore component E (Figure 1A).
Finally, all of the molecules possess hydrophobic pharmacoph-
ore component G from the expanded/refined pharmacophore
33
(Figure 1A) (the third specified criterion, see above), which is
an acetyl group for 1 and 2 (Figure 1B and C), a methyl group
for 3 (Figure 1D), and a chloro substituent for 4 (Figure 1E).
Figure 1. Pharmacophore mapping to inhibitors 1-4. (A) The refined/
expanded pharmacophore for BoNT/A LC inhibition. Planar compo-
nents A and B are magenta rectangles. The dashed, green circle in plane
A represents a heteroatom. Hydrophobic components C and D are
shown as dashed, light blue circles. The positive ionizable pharma-
cophore component E is shown as a dashed, red circle. New pharma-
cophore component F,
33
also a positive ionizable moiety, is shown as
a dashed red circle, while new hydrophobic pharmacophore component
G is a dashed orange circle. (B, C, D, and E) Mapping of inhibitors
1-4, respectively, to the expanded pharmacophore. Distances between
components are in angstroms. Pharmacophore component colors are
as indicated in (A).
2128 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 9 Burnett et al.
As observed in Figure 1B-E, 1-4 all show a good fit with
respect to the distance constraints specified by the pharmacoph-
ore.
In Vitro Testing. Compounds 1-3 were initially examined
for percent inhibition of the BoNT/A LC at 50 µM concentra-
tions, while 4 was tested at a 20 µM concentration. Results from
these initial analyses indicated 90% enzyme inhibition by 1 and
3, 80% inhibition by 2, and 67% inhibition by 4. Subsequent
determination of IC
50
values revealed that SMNPIs 1-4 are
among the most potent inhibitors of the BoNT/A LC reported
to date,
33,40-42
possessing IC
50
values ranging from 3.2 to 17.0
µM (Table 1). In agreement with structurally distinct SMNPIs
reported by our research group (possessing K
i
values ranging
from 3.0 to 10.0 µM
33
), compounds 1-4 demonstrate that our
strategy of incorporating new pharmacophore features (i.e.,
pharmacophore components F or G),
33
in addition to original
components,
31,32
results in the identification of more potent
inhibitors.
Molecular Docking of SMNPIs 1-4 Demonstrates a
Consistency with Previous Inhibitor Binding Modes, Rein-
forcing Our Structure-Based Pharmacophore Approach.
Compounds 1-4 were docked in the BoNT/A LC substrate
binding cleft to determine if they would also engage in
intermolecular contacts that were comparable to those predicted
for other structurally diverse SMNPIs.
31-33
Similar to our
previously docked models of ACQ inhibitors,
31,32
the 7-chloro
Scheme 1
a
a
Reagents and conditions: (a) (i) MsCl, pyridine; (ii) 6, DMF; (b) 6, NaBH
3
CN, CH
3
CN; (c) allyl bromide,
i
Pr
2
NEt, NMP, 80 °C; (d) (i) O
3
, MeOH, -78
°C; (ii) Me
2
S; (e) 6, NaBH(OAc)
3
,CH
2
Cl
2
; (f) NaBH(OAc)
3
,CH
2
Cl
2
; (g) (i) MsCl, pyridine; (ii) NaN
3
, DMF; (h) PPh
3
, THF, H
2
O, 50 °C; (i) 4,7-
dichloroquinoline, PhOH, 120 °C.
Figure 2. Two-dimensional structures of 4-9.
Table 1. Inhibitory Potencies of 1-4
SMNPI IC
50
a
(µM)
1 10 ((0.80)
2 17 ((1.7)
3 7.0 ((1.0)
4 3.2 ((0.92)
a
IC
50
values for the SMNPIs were calculated from plots of concentration
versus inhibition (see Experimental Section for details). Results are the
averages of duplicate determinations.
Pharmacophore Identifies Inhibitors of Botulinum A Journal of Medicinal Chemistry, 2007, Vol. 50, No. 9 2129