Iron is an essential nutrient for nearly all organisms and pathogenic bacteria must acquire iron from the host to support growth and virulence.4,5
However, the free iron concentration is extremely low in the host environment (≈10−24
M) due to the low solubility of Fe3+
and the presence of numerous iron-sequestering host proteins. Thus, to acquire this iron, pathogenic bacteria biosynthesize iron-chelating small molecule natural products called siderophores. These siderophores are secreted into the host milieu where their high affinities for Fe3+
allow them to ‘steal’ iron from host proteins. The iron–siderophore complexes are then recognized by specific receptors and actively transported back into the bacteria, where the iron is released.
A significant number of siderophores have been identified as virulence factors in pathogenic bacteria. For example, a siderophore-deficient mutant strain of Yersinia pestis
exhibits a >10,000-fold higher LD50
in mice than a corresponding siderophore producing strain.6
Further, a siderophore-deficient Mycobacterium tuberculosis
mutant exhibits a significantly reduced growth rate in a macrophage-like cell line compared to a wildtype strain.7
Thus, small molecules that inhibit siderophore biosynthesis represent an important new class of potential antibiotics.
2.1 Biosynthesis of siderophores by non-ribosomal peptide synthetases
Many siderophore biosynthetic pathways involve nonribosomal peptide synthetases (NRPS).8,9
These modular “megaenzymes” assemble amino acid building blocks in a stepwise fashion and introduce a variety of chemical modifications into the polypeptide products.10
The sequence and structure of the non-ribosomal peptide product is encoded by the order of dedicated domains within the NRPS (). Adenylation (Ad) domains catalyze the activation and transfer of specific amino acid building blocks onto a thiol moiety of a peptidyl carrier protein (PCP) or thiolation domain. This thiol is derived from a phosphopantetheinyl group that is installed by a phosphopantetheinyl transferase enzyme. Adenylation domains most commonly accept natural amino acid substrates, but can also specify other substrates, including non-proteinogenic amino acids (e.g. d
-alanine, 2,4-diaminobutyric acid, ornithine) and aryl acids (e.g.
salicylic acid, dihydroxybenzoates). Adjacent aminoacyl-S
-PCP intermediates are coupled by a condensation (C) domain to form a peptide bond. A variety of chemical modifications, including epimerization, methylation, reduction, and oxidation reactions are carried out by other NRPS domains or associated soluble enzymes. Iterative couplings catalyzed by downstream modules lead to a penultimate polypeptidyl-S
-PCP thioester intermediate, which is then released from the NRPS machinery by a terminal thioesterase domain through hydrolysis or cyclization. NRPS may also be associated intra- or intermolecularly with related polyketide synthetases in hybrid biosynthetic pathways.10
Notably, several siderophores have been shown recently to be biosynthesized by NRPS-independent pathways.11
Fig. 1 Peptide assembly by a non-ribosomal peptide synthetase. (a) An adenylation (Ad) domain catalyzes the activation of a specific carboxylic acid building block and acyl transfer onto a peptidyl carrier protein (PCP) domain. (b) A condensation (C) domain (more ...)
Inhibitors of a variety of enzymes involved in siderophore biosynthesis have been reported recently and are described in the following sections.
2.2 Inhibition of isochorismate synthase and salicylate synthase
The first gene in the biosynthetic operons of the Y. pestis
and M. tuberculosis
siderophore biosynthesis gene clusters encodes a salicylate synthase (Irp9 and MbtI, respectively) that converts chorismate to salicylic acid.8
Related enzymes convert chorismate to dihydroxybenozoic acids. These aryl acids are then accepted by NRPS adenylation domains and ultimately transformed into the ‘aryl cap’ seen in a variety of phenolic and catecholic siderophores ().8,9
The salicylate synthase reaction is proposed to proceed through a two-step mechanism ().12,13
First, nucleophilic addition of water to C2 of chorismate displaces the C4 hydroxyl group through a SN
2″ mechanism to generate isochorismate, which remains bound to the enzyme in a twist-boat conformation. This conformation facilitates the second step, in which a [1,5]-sigmatropic rearrangement yields pyruvate and salicylic acid.14
(a) Reactions catalyzed by salicylate synthases (SS, e.g. Irp9) and isochorismate synthases (IS, e.g. EntC). (b) Designed inhibitors of an isochorismate synthase (E. coli EntC) and a salicylate synthase (Y. enterocolitica Irp9).
Abell and coworkers have used this mechanistic information to design a series of Irp9 inhibitors that could potentially block yersiniabactin siderophore biosynthesis.15
These chorismate (substrate) and isochorismate (intermediate) analogs were tested against purified enzyme from the gastroenteric pathogen Y. enterocolitica
and several inhibitors with moderate activity were identified (e.g. 1
, ). These are the first reported inhibitors of a salicylate synthase and set the stage for further exploration of these inhibitor designs and the therapeutic potential of these targets.
Notably, this work was inspired by earlier studies of Bartlett and coworkers on transition state analog inhibitors of the E. coli
isochorismate synthase EntC.16
This enzyme has high homology to the salicylate synthase family and performs the first half-reaction of salicylate synthase to provide isochorismate (). Two additional enzymes (EntB and EntA) then convert isochorismate to the 2,3-dihydroxybenzoic acid building block used in enterobactin biosynthesis.
These inhibitors were designed to mimic the EntC SN2″ reaction transition state, which was proposed to involve a metal-coordinated structure with the nucleophile and leaving group in a syn orientation. Potent biochemical inhibitors were identified using this approach (e.g. 2, ). Thus, theseinhibitors may be useful for targeting biosynthetic pathways leading to aryl-capped siderophores.
The availability of these salicylate synthase and isochorismate synthase inhibitors sets the stage for their further evaluation in cellular assays for inhibition of enzymatic activity, siderophore biosynthesis, and bacterial growth. Recently reported crystal structures of two salicylate synthases, Irp9 and MbtI,14,17,18
provide new insights into the reaction mechanism and, combined with existing structure–activity relationship (SAR) information, should facilitate the design of additional inhibitors.
2.3 Inhibition of salicylic acid adenylation enzymes and of salicylate-derived siderophore biosynthesis
The NRPS-mediated biosynthesis of aryl-capped siderophores is initiated by aryl acid adenylation enzymes, which are generally soluble proteins that are not linked covalently with the remainder of the NRPS machinery.8,9
These enzymes select and activate aryl acid substrates and load them onto an aryl carrier protein (ArCP) domain. This process involves a two-step reaction mechanism (). In the first half-reaction, the aryl acid is adenylated to form an aryl-AMP intermediate, which remains non-covalently bound to the enzyme active site. In the second half-reaction, the aroyl group is transthioesterified onto the phosphopantetheine moiety of the ArCP domain. The aryl acid is then coupled with downstream building blocks (e.g.
amino acids), leading to the aryl-capped siderophore product.
Two-step reaction catalyzed by aryl acid adenylation enzymes, leading to aryl-capped siderophores.
NRPS adenylation domains and mechanistically-related adenylate-forming enzymes bind their cognate acyl-AMP intermediates 2−5 orders of magnitude more tightly than the corresponding carboxylic acid and ATP substrates.19-21
Thus, a variety of non-hydrolyzable analogs of the acyl-AMP intermediates can be used to inhibit these enzymes.22,23
Furthermore, the reported cocrystal structure of DhbE, a 2,3-dihydroxybenzoate adenylation enzyme, with its cognate aroyl adenylate intermediate, 2,3-dihydroxybenzoyl-AMP, can be used to facilitate inhibitor design.24
Notably, the aroyl adenylate is bound by DhbE residues that are highly conserved across all aryl acid adenylation enzymes.
Our group, in collaboration with Quadri and coworkers, reported the first inhibitor of salicylate adenylation enzymes that was designed using this mechanistic and structural information.25
, ) contains a comparatively stable N
-acylsulfamate moiety in place of the acylphosphate group in the corresponding salicyl-AMP reaction intermediate. This compound was shown to be a potent inhibitor of three salicylate adenylation enzymes used in the biosynthesis of yersiniabactin (Y. pestis
YbtE), mycobactin (M. tuberculosis
MbtA), and pyochelin (Pseudomonas aeruginosa
PchD) siderophores. Inhibition of YbtE was shown to be competitive with respect to ATP and non-competitive with respect to salicylate.
Table 1 Representative inhibitors of aryl acid adenylation enzymes. * IC50 value. nd = not determined.
Salicyl-AMS also inhibited Y. pestis
and M. tuberculosis
growth in iron-deficient media, which mimics the host environment and where bacterial growth is known to be siderophore dependent, with IC50
values of 51.2 μM and 2.2 μM, respectively (). Furthermore, siderophore production was shown to be inhibited in both organisms by radiometric TLC visualization of 14
C-salicylate-labeled siderophores. Importantly, the growth inhibitory effects were attenuated significantly in iron-rich media, in which bacterial growth does not require siderophore production. These additional experiments provide support for the mechanism of action of salicyl-AMS. Separately, Aldrich and coworkers have also shown that this compound is non-toxic to a mammalian cell line (P388 murine leukemia) at >200 μM concentration.26
Aldrich and coworkers have also described a large number of salicyl-AMS analogs with variations in the sulfamate,26,27
and aryl acid regions,29,30
providing a detailed SAR profile with respect to inhibition of MbtA and M. tuberculosis
growth. Biochemical potency can be increased slightly by replacement of the sulfamate with a sulfamide (4
, ), replacement of the ribosyl ring 4′-oxygen with a carbon, or omission of either the 2′- or 3′-hydroxyl groups. Docking analyses using a homology model based on the DhbE structure suggested that maintenance of a 3′-endo
ribose conformation is critical for binding.26,28
Importantly, an intramolecular hydrogen bond between the phenolic hydroxyl group and sulfamate nitrogen appears to be required for salicyl-AMS to adopt an appropriate pharmacophoric conformation.27
Along these lines, Bisseret and coworkers have reported an indolylphosphonamide analog of salicyl-AMS designed to enforce this conformation.31
Two analogs have been identified with slightly more potent growth inhibitory activity compared to salicyl-AMS, the sulfamide analog 4
() and 4-fluorosalicyl-AMS (not shown).26,30
Several other analogs have more potent or equipotent biochemical activity but exhibit greatly reduced cellular activity. Based on this information, Aldrich and coworkers have suggested that salicyl-AMS may be a substrate for an as yet unidentified transporter that mediates its uptake.28
The salicyl-AMS class is the first series of compounds demonstrated to inhibit siderophore biosynthesis and bacterial growth in cell culture assays. Further studies in animal infection models will be critical for evaluating the ability of these compounds to block bacterial virulence in vivo and will also provide key insights into the therapeutic potential of blocking siderophore biosynthesis as a new antibiotic strategy.
2.4 Inhibition of a 2,3-dihydroxybenzoic acid adenylation enzyme
Adenylation enzymes specific for 2,3-dihydroxybenzoic acid are used in the biosynthesis of a variety of catecholic siderophores known to be required for virulence in animal models, including enterobactin derivatives that are produced in several Gram-negative enteric bacteria ().8,9
Two 2,3-dihydroxybenzoate adenylation enzyme inhibitors, which are aroyl-AMP mimics, have been reported (5, 6, ). Marahiel and coworkers showed that 2,3-dihydroxybenzoyl-AMS (5
) is a potent inhibitor of DhbE, the adenylation enzyme from Bacillus subtilis
Callahan and coworkers have also explored a series of novel N
-acylhydroxamoyl adenylates, in which a nitrogen atom is inserted between the phosphate and acyl groups. The 2,3-dihydroxybenzoyl derivative (6) proved to be a potent inhibitor of EntE, the adenylation enzyme from Escherichia coli
The potency of this inhibitor is notable considering that the N
-acylhydroxamoylphosphate is ≈2 Å longer than the acylphosphate it replaces. While cellular assays with these compounds have not yet been reported, they demonstrate that non-hydrolyzable aroyl-AMP analogs may be useful for inhibiting a variety of additional siderophore biosynthesis pathways.
2.5 Inhibition of a 3,4-dihydroxybenzoic acid adenylation enzyme
Pathogenic B. anthracis uses an unusual 3,4-dihydroxybenzoate adenylation enzyme, AsbC, to synthesize a second siderophore, petrobactin. Strains of B. anthracis that lack the asb locus, and, thus, the ability to biosynthesize petrobactin, have reduced virulence in mice models.33 AsbC has homology to other NRPS-associated aryl acid adenylation enzymes,34 but the majority of the biosynthetic pathway is actually NRPS-independent.11 Using the sulfamate-based inhibitor design strategy described above, Sherman and coworkers have explored 3,4-dihydroxybenzoyl-AMS (7, ) as a small molecule inhibitor of AsbC.34 Interestingly, this compound exhibits much weaker inhibitory activity against this enzyme compared to structurally related inhibitors of other aryl acid enzymes described above. While the molecular basis for this difference awaits further investigation, this work demonstrates that small molecule inhibition of the petrobactin is, in principle, possible and further broadens the potential therapeutic range of siderophore biosynthesis inhibitors.
2.6 Selective inhibition of an amino acid adenylation domain
Many siderophores do not contain aryl acid-derived moieties. Indeed, this is true of most NRPS-derived natural products. However, amino acid adenylation domains are, by definition, found in all NRPS biosynthetic pathways and, as such, are attractive targets for small molecule inhibition. Indeed, Marahiel and coworkers have demonstrated that aminoacyl-AMS derivatives can be used to inhibit amino acid adenylation domains from B. brevis
gramicidin synthetase and B. subtilis
However, these compounds also inhibit aminoacyl-tRNA synthetases, which catalyze mechanistically identical reactions, with the PCP thiol replaced by a tRNA hydroxyl group as the final nucleophile.21,22
As the latter enzymes are used ubiquitously in ribosomal protein translation, simple aminoacyl-AMP analogs are unsuitable as antibiotics. Two approaches to avoiding this undesired cross-reactivity for aminoacyl-tRNA synthetases can be considered. First, aminoacyl-AMP analogs derived from non-proteinogenic amino acids should only inhibit the NRPS adenylation domains since there would be no corresponding aminoacyl-tRNA synthetases. This approach has been used successfully to target a d
-alanine adenylation domain and is discussed in Section 3.2 below.35
Alternatively, pronounced structural differences between amino acid adenylation domains and aminoacyl-tRNA synthetases can be exploited to design selective inhibitors. This approach has been used successfully to target a cysteine adenylation domain involved in Y. pestis
Our group, in collaboration with Quadri and coworkers, recognized that, although amino acid adenylation domains and aminoacyl-tRNA synthetases catalyze mechanistically identical reactions, the requisite aminoacyl-AMP intermediates are bound in drastically different conformations in available cocrystal structures (). In the structure of the phenylalanine adenylation domain (PheA) of gramicidin synthetase, phenylalanine and AMP ligands are observed in an overall cisoid conformation with respect to the amino acid and adenine moieties ().37
Examination of related structures of an aryl acid adenylation enzyme,24
long chain fatty acid synthetase,38
suggests that this general cisoid conformation is conserved across this enzyme superfamily. In contrast, a carbonyl-reduced analog of phenylalanyl-AMP is bound in a transoid conformation in a cocrystal structure with a phenylalanyl-tRNA synthetase ().40
Indeed, similar transoid conformations are observed in all available structures of ligand-bound aminoacyl-tRNA synthetases.
Fig. 5 (a) Crystal structure of a phenylalanine adenylation domain (PheA) and bound conformations of phenylalanine and AMP ligands. (b) Crystal structure of a phenylalanyl-tRNA synthetase (PheRS) and bound conformation of a phenylalaninyl-AMP ligand. (c,d) Macrocyclic (more ...)
Thus, we designed macrocyclic aminoacyl-AMP analogs 8
() to enforce the pharmacophoric cisoid conformation that is specific to NRPS amino acid adenylation domains.36
These macrocycles were shown to inhibit the cysteine adenylation activity of Y. pestis
yersiniabactin synthetase HMWP2 with affinities comparable to those observed for the corresponding linear aminoacyl-AMS inhibitors 9
(). Most importantly, in contrast to the linear inhibitors, these macrocycles did not inhibit aminoacyl-tRNA synthetases, as determined by in vitro
translation assays containing all 20 of these enzymes.
Further studies to explore the scope of adenylation domain inhibition and the cellular activity of these novel macrocycles are ongoing. Such compounds may have broad potential in inhibiting the biosynthesis of siderophores as well as other NRPS-derived natural products.
2.7 Covalent modification of an aryl carrier protein domain
Another potential set of targets for inhibition of siderophore biosynthesis are the carrier protein domains that accept acyl-AMP intermediates from adenylation enzymes/domains using a phosphopantetheine thiol nucleophile. Aldrich and coworkers have used a vinyl sulfonamide analog of salicyl-AMP (10
, ) to target covalent modification of this thiol in the ArCP domain of MbtB from M. tuberculosis
While this compound is a weak inhibitor of the salicylate adenylation enzyme MbtA, probably due to its inability to form the critical intramolecular hydrogen bond between the phenolic hydroxyl and the (carbon) α-position of the sulfonamide moiety,27
it has an appropriately positioned electrophilic center at the β-carbon to trap the MbtB ArCP thiol nucleophile, forming a stable thioether linkage (observed by MALDI-TOF-MS at 2 μM inhibitor concentration). This adduct also stabilized the MbtA–MbtB protein–protein interaction and, as such, has the potential to block two separate components of the mycobactin biosynthetic machinery.
MbtA adenylation enzyme-catalyzed covalent modification of the ArCP domain of MbtB using a vinyl sulfonamide analog of salicyl-AMP.
Notably, Burkart and coworkers have previously reported a related approach to trapping thiol nucleophiles in polyketide synthetase ketosynthase domains, using carrier proteins functionalized with electrophilic phosphopantetheine analogs.41
2.8 Inhibition of enterobactin C-glucosylation
Several Gram-negative enteric bacteria, including Salmonella
spp., E. coli
, and Klebsella pneumoniae
, produce C
-glucosylated variants of enterobactin (salmochelins), such as diglucosylenterobactin (). Walsh and coworkers have demonstrated that this C
-glucosylation modification allows the bacterial siderophores to evade sequestration by lipocalin 2, a protein that is secreted by mammalian cells as part of the innate immune response to infection.42
While the parent, non-glucosylated enterobactin–iron complex is bound tightly by lipocalin 2 (Kd
= 0.43 nM), the diglucosylated variant is not (Kd
> 1 μM), and remains available for use in bacterial iron acquisition. The machinery for C
-glucosylation of enterobactin and processing of the corresponding iron complexes is encoded by the iroA gene cluster in E. coli
. Introduction of this gene cluster into non-pathogenic E. coli
leads to a hypervirulent phenotype in a mouse infection model. Thus, the biosynthesis of C
-glucosylated enterobactins represents a potential antibiotic target.
-glucosylation is carried out by the IroB glycosyltransferase enzyme in E. coli
(). Walsh and coworkers have identified several substrate analogs 11
that are potent inhibitors of this enzyme.43
Interestingly, none of these bromoenterobactin derivatives is a substrate for IroB-catalyzed C
-glucosylation. All three inhibitors are competitive with enterobactin and form non-covalent complexes with IroB. These inhibitors will allow further evaluation of the therapeutic potential of inhibiting enterobactin C
-glucosylation in enteric bacteria.
(a) The 2,3-dihydroxybenzoyl moieties of enterobactin are iteratively C-glucosylated by IroB to form C-glucosylated enterobactin derivatives. (b) Bromoenterobactin analogs are potent inhibitors of IroB.