Professional Documents
Culture Documents
Multitarget Ligands in Antibacterial Research: Progress and Opportunities
Multitarget Ligands in Antibacterial Research: Progress and Opportunities
Multitarget ligands in
antibacterial research: progress
and opportunities
1. Introduction
Stephen P East† & Lynn L Silver
2. DNA replication †
Evotec (UK) Ltd., Oxfordshire, UK
3. Cell wall inhibitors
4. Miscellaneous multitarget Introduction: Resistance to current antibacterial therapies is an inevitability
ligands that represents a significant global health concern. Bacteria have the capacity
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
5. Future opportunities to render all current drug treatments ineffective, which places a demand on
the drug discovery community to constantly develop new antibacterial
6. Expert opinion
agents. Compounds that inhibit multiple biological targets, often referred
to as multitarget ligands, are an inviting prospect in antibacterial research
because, although they will not solve the issue of resistance, they might
help to delay the onset.
Areas covered: This review covers some of the recent progress in identifying
new ligands that deliberately interact with more than one essential biological
target in bacteria. The two principal areas covered are inhibitors of DNA
replication and cell wall biosynthesis.
Expert opinion: Antibacterial programs for the design of multitarget ligands
present an important opportunity for production of antibacterial agents.
For personal use only.
1. Introduction
10.1517/17460441.2013.743991 © 2013 Informa UK, Ltd. ISSN 1746-0441, e-ISSN 1746-045X 143
All rights reserved: reproduction in whole or in part not permitted
S. P. East & L. L. Silver
Multitargeted chemotherapy [4,5] can be broadly divided Enzymes that function as part of the bacterial DNA synthesis
into two categories: i) combination therapies, that is, two or machinery play an essential role in bacterial cell survival [13].
more compounds that when taken together, either as a cock- Two enzymes that contribute to DNA replication are the
tail or a multicomponent drug, work synergistically but typi- bacterial type II topoisomerases: DNA gyrase and topoisom-
cally elicit their pharmacological response by acting on erase (topo) IV. These enzymes are responsible for controlling
For personal use only.
independent biological targets [6] and ii) single compound DNA topology at various stages of DNA replication. Both
therapies, also termed multiple ligands [4,7] or multitarget enzymes form heterotetrameric complexes, which in the case
ligands [8], that is, one compound that modulates the pharma- of DNA gyrase is an A2B2 complex and for topo IV is a
cological response of two or more relevant biological targets. complex consisting of two ParC and two ParE subunits. As
Multitarget ligands can be a product of two or more discrete a consequence of their similar structural homology, dual
molecules which, when separate, interact independently with inhibition of gyrase and topo IV has been enabled, albeit
quite different biological targets, but they have been joined serendipitously [14], and members of the quinolone (including
(via a linker) or fused together to form a single compound. fluoroquinolones) class of antibiotics have validated this
Indeed this approach has been used successfully in the search approach clinically for antibacterial infections [12]. The quino-
for new antibacterial agents and reviewed recently [9]. Alterna- lones inhibit the related GyrA and ParC subunits. The GyrB
tively, multitarget ligands can contain specific structural fea- and ParE subunits bind and process ATP and so the similarity
tures that are important in the formation of key binding in the ATP-binding site of these proteins provides an addi-
interactions with not just one but multiple biological targets tional opportunity for inhibition of two targets. Indeed dual
of interest. Morphy and Rankovic [4] have referred to these inhibitors of GyrB/ParE have been identified but currently
types of ligands as being ‘merged’. In the scope of antibacterial there are no approved GyrB/ParE drugs.
research such multitarget ligands have been identified, most
typically in cases where the biological targets are structurally 2.1 GyrA/ParC inhibitors
similar at the ligand-binding site and/or they have closely Despite a plethora of quinolone antibacterials in the market-
related substrates and it is these ligands that are the focus of place, emerging resistance and issues with toxicity have meant
this review. A possible advantage of these merged ligands that the search for alternative GyrA/ParC inhibitors continues
over the joined or fused ligands is that the resulting drugs unabated. Several compounds are progressing in the latter
might be expected to have a lower molecular weight and con- stages of clinical development and these have been covered
sequently might occupy more attractive chemical space in elsewhere [15] but there are others that are at a preclinical stage
terms of their physicochemical and DMPK properties [10]. or that have recently entered Phase I, and some of these
So the questions are: How have we identified multitarget developments will be covered here.
ligands in antibacterial research programs in the past and Pucci et al. [16] have detailed the in vitro and in vivo profile
how might we do it better in the future? For the former of Achillion’s advanced isothiazoloquinolone ACH-702
question, the leading classes of multitarget antibacterials are (Figure 1). This compound is active in Staphylococcus aureus
the b-lactams that inhibit penicillin-binding proteins (PBPs) topo IV decatenation assays (IC50 120 nM) and DNA gyrase
(e.g., carbapenems [11]) and the quinolones that inhibit supercoiling (IC50 680 nM) assays and shows superior
DNA topoisomerases [12], but the discovery that these drugs dual-targeting inhibition compared with benchmark
O O O O O O
F F
N OH N OH
NH
N N S N N
H2 N
H2N OCH3
H2N
OCH3
O
N OCH3
OH H
S N
N
N O N
S S F N
OH NC
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
Viquidacin GSK299423
Figure 1. Quinolone-related dual GyrA/ParC inhibitors and other prototypical topoisomerase inhibitors.
fluoroquinolones moxifloxacin, gemifloxacin and ciprofloxa- fluoroquinolone-resistant organisms. For example, a represen-
cin. It also retains some activity in vitro against single- and tative compound in this series, GC-061 (Figure 1), is active in
double-mutated gyrases that show resistance to the same set S. aureus (MIC range < 0.015 -- 0.5 µg/ml), which includes
of control compounds. Broad-spectrum antibacterial activity strains that are ciprofloxacin- and levofloxacin-resistant (i.e.,
against clinical isolates was shown with ACH-702 having a MIC ‡ 16 µg/ml). In Gram-negative organisms, GC-061
better profile than levofloxacin on Gram-positives including has excellent activity against Haemophilus influenzae
activity on fluoroquinolone nonsusceptible organisms. On (MICs < 0.015 µg/ml) but also shows some activity against
For personal use only.
Gram-negative organisms, ACH-702 has a similar profile to strains of Escherichia coli (median MIC 16 µg/ml), Pseudomo-
ciprofloxacin. Time-kill analysis revealed that ACH-702 is nas aeruginosa (median MIC 8 µg/ml) and other organisms.
bactericidal and resistant mutations were observed at frequen- Confirmation of dual-targeting with GC-061 was demon-
cies of < 10-10 at 4 MIC in strains of S. aureus and this strated in gyrase and topo IV in vitro assays with
represents a ~ 2 log units improvement over ciprofloxacin. IC50s £ 1.42 and £ 2.33 µM reported. Spontaneous resis-
Positive data on ACH-702 in two Gram-positive models of tance frequencies for compounds in this series are reported
infection have been reported. A dose-dependent effect in a to be low (i.e., ~ 10-11 in E. coli and S. aureus at 4 MIC)
methicillin-resistance S. aureus (MRSA) mouse thigh infec- which is suggestive but not conclusive of dual inhibition in
tion model was observed, with 5 mg/kg leading to stasis in a cellular context. In vivo efficacy of GC-061 was achieved in
infection after 24 h, and a 2.5-log reduction in colony a mouse thigh MRSA infection model when dosed orally at
forming units (CFUs) was observed after the same timeframe 30 mg/kg with ~ 5-log reduction in bacterial tissue burden
following a dose of 20 mg/kg. On the Achillion website [17], being observed after 24 h. The same compound showed
ACH-702 is also reported to be active on DNA primase, a ~ 3-log units reduction in CFUs over the same time period
which is another enzyme involved in bacterial DNA replica- in a mouse E. coli thigh infection model. These data are at
tion. Although no data demonstrating activity of ACH-702 least as good as ciprofloxacin (20 mg/kg i.v. dose) in the
on this target are available, DNA primase and the topoisomer- same models. GC-061 was selective over human topoisomer-
ases have been shown to share a homologous catalytic site, the ases (IC50 Topo II > 35 µM; IC50 Topo I > 100 µM), and
TOPRIM domain, which may underlie activity against all there are no apparent in vitro toxicity issues (e.g., hERG
three proteins [18]. Detailed structure--activity relationship IC50 60 µM, inactive at the highest concentration tested in a
studies that led to the identification of ACH-702 have been HepG2 cell line, a micronucleus genotoxicity assay and
described by Kim et al. [19]. cytochrome p450 assays).
Evolva are progressing broad-spectrum bactericidal In addition to the quinolone and related classes of antibac-
2-pyridones (termed the EV-035 series) [20,21] that are isosteric terial agents, other novel bacterial topoisomerase II inhibitors
to the traditional quinolones and related in structure to that target the GyrA/ParC subunits have been described in the
ABT-719 (Figure 1), a compound that reached Phase II literature. A prototype compound was viquidacin (NXL101,
clinical trials [22]. In silico modeling has suggested that the Figure 1) [23], and this antibacterial agent advanced into
substitution pattern on the EV-035 series is crucial for Phase I studies before issues associated with QT prolongation
compounds to adopt an overlapping but different binding halted its progression in the clinic. Several research programs,
mode to the quinolones. This subtle change in the ligands including GSK [24,25], AstraZeneca [26-28] and Achillion [29],
interaction with DNA gyrase/topo IV is probably responsible have all reported on chemotypes related to viquidacin with a
for the 2-pyridones showing good antibacterial activity against focus on addressing the cardiovascular toxicity side effects by
OCH3
tuning logD and pKa and monitoring hERG inhibition
N
in vitro. These chemotypes are of significant interest because
OH
bridges to the bound DNA substrate such that it intercalates
between base-pairs, thus preventing DNA cleavage. This
represents an alternative binding mode to the quinolones [32]
N
[29]
are not provided in all of the publications, the implication is
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
[28]
O
AZ
septicemia model.
N
[25]
Ref.
H
H2N O N donor NH groups on the ethyl ureas that interact with an
O O O O
Asp residue in the ATP pocket together with the N-1 of the
HO H H H OH
O benzimidazole that acts as an acceptor donor to a conserved
water molecule that also hydrogen bonds to the same Asp
Kibdelomycin residue. In a very recent patent disclosure [43], Vertex have
reported the latest generation dual-targeting benzimidazoles
OH with compound 4a (Figure 3) showing good enzymatic inhibi-
HO O
tion (S. aureus Gyr Ki 9 nM; S. aureus Topo IV Ki 12 nM)
OH and antibacterial activity particularly against Gram-positive
O
F strains (S. aureus MIC90 0.03 µg/ml (n = 67); S. pneumoniae
F OH O O
MIC90 0.015 µg/ml (n = 67); Enterococcus faecalis MIC90
O 0.06 µg/ml (n = 34)). The compound 4a is shown to be
For personal use only.
O
OH
O active in a mouse S. aureus kidney infection model and a
O OH
S. aureus neutropenic rat thigh model. The patent also descri-
2 bes the prodrug 4b which was presumably investigated due to
the low solubility of 4a and indeed 4b might be the preferred
delivery method for active ingredient 4a.
Figure 2. Naturally occurring/derived dual DNA gyrase and Recognizing the key inhibitor design features of the benz-
topo IV inhibitors. imidazole ethyl ureas, several other groups have published
research in which the benzimidazole scaffold has been inter-
or no cross-resistance with aminocoumarin-resistant bacterial changed with a number of other bicyclic and biaryl systems
strains were observed. The lack of kibdelomycin-resistant that have the potential to retain the critical H-bond acceptor
mutants found could be due to inhibition of other essential as well as the ethyl urea moiety [44]. The benzothiazoles pre-
functions above the MIC. sented by Biota are the most recent refinement of the ethyl
Hossion et al. described quercetin diacylglycosides as dual urea series [45]. Fine tuning of the substituents in the 5- and
DNA gyrase/topo IV inhibitors [36,37]. One of the most inter- 7-position of the benzothiazole scaffold provided compounds
esting analogs, compound 2 (Figure 2), was reported to be active such as example 5 in Figure 3 that has excellent activity on
in an E. coli gyrase supercoiling assay (IC50 0.78 µM) and Gram-positive pathogens (S. aureus MIC 0.06 µg/ml;
E. coli topo IV (IC50 9.2 µM) and S. aureus topo IV Streptococcus pyogenes MIC 0.06 µg/ml; E. faecalis MIC
(IC50 0.22 µM) decantenation assays. The compound is active 0.02 µg/ml) and a good pharmacokinetic profile with
on Gram-positive organisms (MICs 0.13 -- 0.5 µg/ml on > 65% oral bioavailability being achieved in rodents. Some
S. pneumoniae, methicillin-resistant S. aureus and vancomycin- Gram-negative activity has been observed (H. influenzae
resistant enterococci) but no activity on Gram-negative species MIC 0.5 -- 2 µg/ml); Moraxella catarrhalis (MIC
was observed (MICs > 128 µg/ml on P. aeruginosa and E. coli). < 0.125 µg/ml) suggesting that the compound class could be
Molecular docking studies postulated that the compounds suitable for the treatment of respiratory tract infections. The
bind to the ATP-binding site on GyrB, although no data in structures of the most advanced compounds have not been
ATPase assays were reported to confirm that this was the revealed but spontaneous resistance frequencies for lead com-
site of action of this inhibitor class. It should be noted that pounds in this class were reported as being < 10-9 in S. aureus
quercetin is a flavonoid that has been shown to inhibit a great and S. pneumoniae, which is consistent with a dual-
number of enzymes, eukaryotic, viral as well as bacterial [38] targeting mechanism of action. In vivo efficacy via oral
and it is unlikely that gyrase/topoisomerase inhibition is the administration with a compound from this series was demon-
sole or even major mechanism of action of quercetin in strated in models of S. aureus including a neutropenic thigh
Gram-positives. model. In a S. pneumoniae mouse model at a 30 mg/kg dose
RO
N N
N H N N H
N H N H
N N F N N
H O H
F O
N O
3 Vertex 4a R = H
4b R = P(=O)(ONa)2
HO2C
N N OCH3 N H
N N H
N N N
H
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
N H
S N N O
O N N
O
N HN
CF3
O
5 Biota 6 AstraZeneca
O–
N+
S
N H
N N
N
N
Cl
H
For personal use only.
H2N
7 Trius
(intravenous), 100% survival was noted in 9 days after inocu- Gram-negative organisms was provided. Indeed Manchester
lation. In vitro toxicology assessment was good with no appar- et al. have reported in a separate publication [47] that the
ent issues identified and compounds were reported to be AstraZeneca experience of optimizing Gram-negative
progressing through in vivo safety experiments. antibacterial activity in new compounds classes is difficult.
Scaffolds that do not contain an ethyl urea group but have Efficient efflux systems together with more complex
been identified using structural information have also been permeability barriers in Gram-negatives compared with
described recently. Manchester et al. [46] assembled a library Gram-positives might be responsible for the apparent lack
of fragments previously observed in ATP-competitive kinase of data for the azaindoles.
inhibitor programs at AstraZeneca. From a virtual screening Trius have utilized a combination of structure-based virtual
exercise using a S. pneumoniae ParE structure, they identified screens and literature-led identification of new GyrB/ParE
an azaindole scaffold as satisfying the key molecular interac- dual inhibitor scaffolds. A key design feature in the Trius
tions with the Asp and conserved water residue in ParE. This program was to rationally design inhibitors of Gram-negative
scaffold was further elaborated to achieve low nanomolar pathogens as well as Gram-positive organisms. This was
potency on both S. pneumoniae ParE and S. aureus GyrB achieved by building in some degree of flexibility to the
with example 6 in Figure 3 being one of the most potent ana- compounds to compensate for the differences in the Gram-
logs described (ParE IC50 3 nM; GyrB IC50 2 nM; S. aureus positive and Gram-negative protein structures, as well as opti-
MIC 0.1 -- 0.2 µg/ml; S. pneumoniae MIC < 0.02 µg/ml). mizing the compounds for Gram-negative entry and lack of
Antibacterial activity did not correlate at all with enzymatic efflux. One of the advanced Trius compounds, 7 (Figure 3), is
activity but some relationship between logD and antibacte- from a series of pyrrolopyrimidines. This compound has
rial activity on Gram-positive organisms was observed. sub-nanomolar binding affinity on E. faecalis (Ki < 0.3 nM)
The hypothesis presented was that compounds with and Francisella tularensis GyrB (Ki < 0.3 nM) and single
higher logD show improved permeability properties (e.g., digit nanomolar activity on ParE (E. faecalis Ki 8.6 nM;
compound 6, logD = 2.91) to traverse the bacterial cell F. tularensis Ki 1.7 nM; E. coli Ki 2.5 nM) [48]. The in vitro
membrane. No activity for the azaindole compounds on antibacterial activity for compound 6 is good against
O
MIC 2 µg/ml; E. coli 2 µg/ml; P. aeruginosa 4 µg/ml), and it
Cl
also has good coverage of biodefense pathogens. The com-
S
N
pound was reported to show activity in a S. aureus septicemia
NH
N
[60]
11
nd
pyrrolopyrimidine series did not necessarily show all the same
specific effect on DNA synthesis in bacterial cells. This
experiment illustrates (as noted in several places above) that
antibacterial activity is not always a consequence of targeted
MurC (E. coli) 52%@100 µM
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
3.
MurF (E. coli) 2 µM
O
HO
10
infections [49].
[55]
9
[53]
nd
Ref.
CO2H Br CO2H
O
NH NH
O S O OH
O O N
H
O
12 13 14
O NO2 O
OH OH
N B N B
H H
OH OH
F
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
15 16
have utilized the available X-ray structural data to perform based on the level of potency observed in the biochemical
pharmacophore-based virtual screens. assay, the compounds would not be expected to have a specific
Several publications reported on the use of MurD structures effect on bacterial cells.
for the virtual screen experiments. Perdih et al. [53] identified
benzene dicarboxylic acids that were postulated to act as gluta- 3.2 Penicillin-binding proteins
mic acid mimics in the catalytic site. The most interesting com- The b-lactam antibiotics were the first, and they are arguably
pound reported was compound 8 (Table 2), which shows the most widely, studied of all of the clinically prescribed
For personal use only.
in vitro activity on both MurD and MurE. Tomašić et al. [54] antibacterial agents. They can also be considered as the arche-
identified glutamic acid containing MurD inhibitors and these typal multitarget inhibitors because they covalently modify
compounds have subsequently evolved into dual MurD/E members of a family of enzymes known as the PBPs. PBPs
inhibitors [55] with compound 9 (Table 2), which bears some have a variety of functions in the latter stages of bacterial
structural resemblance to compound 8, showing inhibition of cell wall biosynthesis including transpeptidase activity (as
ligase activity on E. coli and S. aureus MurD and MurE. The mentioned above). Although b-lactam antibiotics have been
antibacterial activity of compound 9 on S. aureus (including very successful, one of the resistance mechanisms that have
an MRSA strain) and E. faecalis was modest and no activity emerged to this class of compounds is as a consequence of their
was seen on E. coli or P. aeruginosa up to 128 µg/ml. inherent reactivity. Bacteria express a family of enzymes termed
A common feature of compounds 8 and 9 is the rhodanine b-lactamases that inactivate b-lactams via a similar mechanism
scaffold and Tomašić et al. [56] have also published on a set of to the way that b-lactams inhibit PBPs. An established
5-benzylidenethiazolidin-4-ones with the trihydroxy deriva- method to overcome this path of resistance is to co-administer
tive, compound 10 (Table 2) showing activity on all four Mur a b-lactam antibiotic with a b-lactamase inhibitor; however,
ligases although the compound does not display antibacterial this combination approach can complicate the clinical develop-
activity on Gram-positive or Gram-negative strains up to ment. Recently, there has been much interest in developing
128 µg/ml. Rhodanines and other five-membered multihetero- reversible PBP inhibitors that do not contain the b-lactam func-
cyclic scaffolds have been the subject of much debate in terms tionality and hence may not be as susceptible to b-lactamase
of whether they are privileged structural elements or simply inactivation (although they might still bind to b-lactamases as
promiscuous hitters [57-59]. In the case of the rhodanine well as PBPs).
compound 9, the published ligand-protein X-ray structure [55] Turk et al. [61] have reported some biaryl sulfonamides and
is compelling evidence for a specific interaction on MurD. biaryl amides as noncovalent binders to PBPs. Compounds
In what seems to be a complementary virtual screen to such as those shown in compounds 12 and 13 (Figure 4) were
those described previously, Tomašić et al. [60] scrutinized the identified from a limited compound screen and show weak
X-rays structures of available E. coli and H. influenzae inhibition (IC50s ~ 100 -- 400 µM) of PBP2a (S. aureus),
MurC-F enzymes focusing on the residues in the ATP- PBP2x (S. pneumoniae) and PBP5fm (E. faecium). Some
binding sites. They concluded that the binding sites were antibacterial activity against Gram-positive organisms was
similar and subsequently performed a virtual screen using also reported (MICs 1 -- 256 µg/ml), although it was not
E. coli MurD. A selection of ligands was tested and weak determined whether the cellular inhibition observed was a
dual inhibitors of MurC and MurD were identified with consequence of PBP inhibition. The compounds were inactive
compound 11 (Table 2) being the best compound of this (MICs > 512 µg/ml) on Gram-negative organisms. Shilabin
initial set. No antibacterial activity was reported, although, et al. [62] have reported a series of 4-quinolones as pan-PBP
HO2C N
H H
OH
DNA replication H3CO DNA polymerases III: PolC and PolE [70]
O
OCH3
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
N N N N OCH3
H H
OCH3
Folate biosynthesis OH Dihydropteroate synthase (DHPS) [71]
N NO2 and dihydrofolate reductase (DHFR)
H2N N NH
inhibitors. Compound 14 (Figure 4) has Kis of 5 -- 220 µM observed. Despite these differences in the structural informa-
against PBP 1a/1b, PBP2, PBP3, PBP4 and PBP5/6 from tion it is reasonable to assume that this compound is also a
E. coli. No antibacterial activity up to 1 mM on a strain of reversible inhibitor.
E. coli was observed.
For personal use only.
Boronic acids have been investigated as PBP inhibitors. 3.3 Lipid A biosynthesis
Here the concept is that the boronic acid forms a tetrahedral Gram-negative bacteria have a layer of lipopolysaccharide
intermediate with a critical serine residue in the active site (LPS) on their outer membrane that functions as a protective
of the PBPs. This reaction is similar to what happens with coating to the cell. A constitutive ingredient of LPS is
b-lactam antibiotics and PBPs (and indeed b-lactamases) Lipid A which acts as the anchor for LPS to the cell mem-
but, in the case of the boronic acid inhibitors, the reaction is brane but it also functions as an endotoxin, mobilizing the
reversible because the boron-oxygen bond that is formed can innate immune system in humans and leading to sepsis.
be readily hydrolyzed. Structural information on how boronic Lipid A is crucial for the survival of Gram-negative bacteria
acids bind to PBPs is beginning to emerge and this should so an inhibitor of the one or more steps in the synthesis of
facilitate improved reversible inhibitors in the future. Lipid A could provide a new therapy. The first three steps
A series of N-alkyl boronates were reported by Woon in Lipid A biosynthesis are catalyzed by the enzymes LpxA,
et al. [63] and Contrerase-Martel et al. [64]. Compound 15 LpxC and LpxD. LpxC has garnered much attention from
(Figure 4) is one of the most interesting compounds in the drug discovery community [66] with ACHN-975 recently
this series and it was active on PBP1b (IC50 20 µM) and being the first LpxC inhibitor to enter human trials [67].
also shown to have activity on PBP2, PBP2a, PBP3, Jenkins and Dotson [68] have suggested that LpxA and
PBP4 together with sub-micromolar inhibition of the PBP LpxD might offer an opportunity for a dual-targeted
R39 (IC50 0.27 µM). Antibacterial activity on S. aureus therapy. These two enzymes are both acyltransferases
(MIC 32 µg/ml including MRSA strains) and other Gram- and have been shown to be inhibited by the dodecapeptide
positive organisms was demonstrated and X-ray structural RJPXD33 with Kd values measured as 22 µM (E. coli
studies suggested that compounds of this type can adopt LpxA) and 6.5 µM (E. coli LpxD), respectively. Clearly the
two different binding modes. In a more recent paper, development of a long peptide chain as a novel antibacterial
Zervosen et al. [65] have described related glycine boronic is likely to be challenging; however, improving the under-
acids, for example, compound 16 (Figure 4) with activity on standing of the minimum pharmacophoric features of the
S. pneumoniae PBP1b (IC50 26 µM), S. pneumoniae PBP2x peptide required for binding might present possibilities for
(IC50 138 µM) and Actinomadura R39 (IC50 0.6 µM). small molecule inhibition.
Curiously, X-ray data on compound 16 crystallized with
R39 revealed that the tetrahedral intermediate forms by inter- 4. Miscellaneous multitarget ligands
actions with two serine residues and a lysine residue in the
active site. This has been described as a tricovalent interaction; Other multitarget ligands have been reported in the literature
however, the ligand-protein complex of compound 16 with and some of the salient examples are included in Table 3
PBP1b revealed that only a monocovalent interaction is [69-71]. As there is little new relevant information that
merits inclusion in this review, the readers are directed to ligands have been identified for biological targets unrelated in
other articles that contain the key references and a more protein sequence or structure. The idea here is to find addi-
comprehensive summary of the work shown in Table 3 [1,3]. tional biological targets for ligands already known to inhibit
one enzyme of interest.
5. Future opportunities
6. Expert opinion
Multitarget ligands need to modulate the pharmacological
response of multiple enzymes. Looking at this challenge Multitarget ligands represent an approach to discovery of
simplistically, an ideal ligand would inhibit the enzymes of novel antibacterials that should yield drugs having a low
interest to a similar degree (perhaps within an order of magni- probability of selecting rapidly for high-level target-based
tude [4]), as determined by in vitro biochemical experiments. resistance. While all antibacterials will eventually fall to resis-
In reality, the situation is much more complex and there tance mechanisms, it is highly likely that antibacterial com-
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
will be many factors that ultimately determine what the pri- pounds targeting single enzymes will immediately select for
mary target is in vivo and these are not easy to predict. For pre-existing resistance mutations that arise at a measurable
example, ligand structure and enzyme-binding site structure rate per generation. Indeed, a standard method of identifying
have both been reported to influence the primary target of the mechanism of action of a novel compound is to select for
dual GyrB/ParE inhibitors [41] in different bacterial strains. such target-site mutations. Whether the rate of resistance seen
What is clear though is that true, multitarget ligands can in the laboratory will translate into therapeutic failure
reduce in vitro spontaneous resistance frequencies (selected depends upon (among other things) that rate, the bacterial
at high levels of compound) to frequencies below those for burden during the infection, the fitness and virulence of the
single mutations; sequential mutations in more than one resistant mutants and the possibility of reaching high enough
essential target would be necessary in order for the bacteria levels of compound at the site of infection to inhibit the
to develop high-level resistance to a multitargeted drug. If resistant organisms. While there are single-enzyme targeted
the inhibitor is well balanced, then the likelihood of single agents in clinical use, they are usually dosed as combinations
For personal use only.
same caveats as for kinases would apply in the search for novel For Gram-negatives, the double membrane with its orthogo-
antibacterial multitarget ligands. nal sieving properties presents a formidable barrier for which
Indeed, an important aspect of antibacterial discovery in no set of chemical rules for its passage have been formulated.
general, also applicable in the case of multitarget ligands, is Thus, although target structure can guide rational design of
that the mechanism of action at the whole-cell level of inhibitors, the physicochemical properties of the compound
enzyme inhibitors cannot be inferred from their activity on enabling cell entry must be considered for antibacterial
the isolated enzymes or, in the case of putative dual-targeting activity -- preferably during the optimization process. For
agents, from the inability to obtain mutants. Independent example, it is highly likely that inability to find membrane
proof of mechanism is needed [1]. With specific pharmaco- permeant compounds has stymied the discovery, despite
phoric features in ligands that target multiple enzymes, there much effort, of validated antibacterial inhibitors of the Mur
is a finite likelihood that other targets will be inhibited as ligases, mentioned above as potential multitarget enzymes,
well. Such extra activities might have implications for toxic- since their common binding sites are for UDP-MurNAc
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
ity. The possibility of nonselectivity is illustrated by the find- and ATP substrates, and may thus require nonpromiscuous
ings, noted above, with series of GyrB/ParE compounds in permeant diphosphate isosteres.
studies by Trius [48]. Even small changes in structure that Thus, antibacterial programs for the design of multitarget
had little effect on topoisomerase inhibition led to altera- ligands present an important opportunity for production of
tions in whole cell activity which proved to be due to inhibi- antibacterial agents with longevity due to slow development
tion of macromolecular synthesis in addition to DNA. For of resistance comparable to that seen with other successful
the GyrB/ParE inhibitors kibdelomycin [35] and querce- agents, such as the fluoroquinolones -- but much improved
tin [38], the available data does show inhibition of the topoi- over single-targeted agents for which resistance can appear
somerases but does not exclude (and even demonstrates) overnight (at least in the laboratory).
inhibition of other cellular processes at concentration near
the MIC. With the rhodanine inhibitors of Mur ligases, Acknowledgments
mentioned above, while there is crystallographic data
For personal use only.
supporting specific enzyme interaction, the selectivity (anti- The authors would like to thank R Law, J Heim, J Manchester,
bacterial activity being due solely to inhibition of the ligases) J Finn, J Palmer and D Haydon for their assistance preparing
remains unproven [56,57]. this manuscript.
A major obstacle to target-based antibacterial discovery is
the necessity for enzyme inhibitors to reach their target Declaration of interest
site [1]. This is not as critical for activity against extracytoplas-
mic targets of Gram-positives (such as b-lactams and SP East is an employee of Evotec (UK) Ltd, while LL Silver is
glycopeptides), but is important for periplasmic targets in an independent consultant in antibacterial discovery at LL
Gram-negatives and cytoplasmic targets of all pathogens. Silver Consulting, LLC.
Bibliography
Papers of special note have been highlighted as 10. Morphy R, Rankovic Z. The 23. Black MT, Stachyra T, Platel D, et al.
either of interest () or of considerable interest physicochemical challenges of designing Mechanism of action of the antibiotic
() to readers. multiple ligands. J Med Chem NXL101, a novel nonfluoroquinolone
2006;49:4961-70 inhibitor of bacterial type II
1. Silver LL. Challenges of antibacterial
11. Papp-Wallace KM, Endimiani A, topoisomerases.
discovery. Clin Microbiol Rev
Taracila MA, Bonomo RA. Antimicrob Agents Chemother
2011;24:71-109
Carbapenems: past, present, and future. 2008;52:3339-49
2. Walsh C. Molecular mechanisms that
Antimicrob Agents Chemother 24. Miles TJ, Barfoot C, Brooks G, et al.
confer antibacterial drug resistance.
2011;55:4943-60 Novel cyclohexyl-amides as potent
Nature 2000;406:775-81
12. Drlica K, Hiasa H, Kerns R, et al. antibacterials targeting bacterial type
3. Silver LL. Polypharmacology as an IIA topoisomerases. Bioorg Med
Quinolones: action and resistance
emerging trend in antibacterial discovery Chem Lett 2011;21:7483-8
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
32. Wohlkonig A, Chan PF, Fosberry AP, structure-activity relationships. a rich source of drug targets.
et al. Structural basis of quinolone J Med Chem 2008;51:5243-63 Crit Rev Biotechnol 2011;31:295-336
inhibition of type IIA topoisomerases . Medicinal chemistry summary of the 51. Silver LL. Does the cell wall of bacteria
and target-mediated resistance. benzimidazole ureas including a remain a viable source of targets for
Nat Struct Mol Biol 2010;17:1152-3 hypothesis on the importance of the novel antibiotics? Biochem Pharmacol
33. Oblak M, Kotnik M, Solmajer T. conformation of the small molecule for 2006;71:996-1005
Discovery and development of ATPase dual GyrB/ParE inhibition and in vivo
52. Smith CA. Structure, function and
inhibitors of DNA gyrase as antibacterial efficacy data.
dynamics in the mur family of bacterial
agents. Curr Med Chem 41. Grossman TH, Bartels DJ, Mullin S, cell wall ligases. J Mol Biol
2007;14:2033-47 et al. Dual targeting of GyrB and ParE 2006;362:640-55
34. Angehrn P, Goetschi E, Gmuender H, by a novel aminobenzimidazole class of
53. Perdih A, Kovač A, Wolber G, et al.
et al. A new DNA gyrase inhibitor antibacterial compounds.
Discovery of novel benzene
subclass of the cyclothialidine family Antimicrob Agents Chemother
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13
penicillin-resistant bacteria. PLoS One agents. Curr Med Chem 72. Škedelj V, Tomašić T, Mašič LP,
2011;6:e19418 2012;19:2038-50 Zega A. ATP-binding site of bacterial
62. Shilabin AG, Dzhekieva L, Misra P, 67. Achaogen announces all objectives met in enzymes as a target for antibacterial drug
et al. 4-Quinolones as noncovalent Phase 2 plazomicin complicated urinary design. J Med Chem 2011;54:915-29
Inhibitors of high molecular mass tract infections study and start of 73. Keiser MJ, Roth BL, Armbruster BN,
penicillin-binding proteins. ACS Med first-in-human study with ACHN-975. et al. Relating protein pharmacology by
Chem Lett 2012;3:592-5 Available from: http://www.achaogen. ligand chemistry. Nat Biotechnol
63. Woon ECY, Zervosen A, Sauvage E, com/news/148 2007;25:197-206
et al. Structure guided development of 68. Jenkins RJ, Dotson GD. Dual targeting 74. Durrant JD, Amaro RE, Xie L, et al.
potent reversibly binding penicillin antibacterial peptide inhibitor of early A multidimensional strategy to detect
binding protein inhibitors. ACS Med lipid A biosynthesis. ACS Chem Biol polypharmacological targets in the
Chem Lett 2011;2:219-23 2012;7:1170-7 absence of structural and sequence
homology. PLoS Comput Biol
Expert Opin. Drug Discov. Downloaded from informahealthcare.com by University of Sussex Library on 01/20/13