Article Cite This: Acc. Chem. Res. 2017, 50, 2510-2518
pubs.acs.org/accounts
Highly Potent Antibacterial Organometallic Peptide Conjugates Bauke Albada*,† and Nils Metzler-Nolte*,‡ †
Laboratory of Organic Chemistry, Wageningen University & Research, Stippeneng 4, 6708 WE Wageningen, The Netherlands Inorganic Chemistry I − Bioinorganic Chemistry, Ruhr University Bochum, Universitätsstrasse 150, D-44780 Bochum, Germany
‡
CONSPECTUS: Resistance of pathogenic bacteria against currently marketed antibiotics is again increasing. To meet the societal need for effective cures, scientists are faced with the challenge of developing more potent but equally bacteria-specific drugs. Currently, most efforts are directed toward the modification of existing antibiotics, but ideally, compounds with a new mode of action are required. In this Account, we detail our findings in the area of novel metal-based antibiotics. Our strategy is based on the modification of simple antimicrobial peptides (AMPs) with organometallic agents, resulting in organometallic AMPs (OM-AMPs). Since bacteria have most likely never encountered these synthetically prepared unnatural organometallic agents, we anticipated that such agents could well become potentiating players in the antibiotics arena. Moreover, exploiting some of the particular properties of metal complexes should also help to elucidate the mode of action of small cationic AMPs, the molecular details of which have remained elusive despite intensive efforts. Using standard Fmoc/tBu-based solid-phase peptide synthesis approaches, we have prepared various organometallic−peptide conjugates with covalently linked group 8 and 9 metallocenes (ferrocene, ruthenocene, osmocene, and cobaltocenium). As a starting point we took the (RW)3 antibacterial hexapeptide lead structure. After modifying the peptide sequence (generations 1 and 2), changing the nature and position of the organometallic group (generation 3), and optimizing the amino acid chirality (generation 5), we identified several organometallic antibacterial peptides that are currently among the most active synthetic AMPs (synAMPs) that have ever been prepared. Through these rational and systematic optimizations, we were able to increase the antibacterial activity of a short non-organometallic synAMP 18-fold to submicromolar activity, rivaling the activity of vancomycin (often the drug of last resort) against methicillin-resistant Staphylococcus aureus (MRSA). Moreover, by making use of the unique physicochemical properties of ruthenocene, we were able to determine the mode of action of these short AMPs in unprecedented detail. We propose that the OM-AMP integrates into the bacterial membrane and changes its biophysical properties, which ultimately results in detachment of vital enzymes for respiration and cell-wall biosynthesis such as specifically cytochrome c and MurG from their locations in the membrane. Further explorations of these small OM-AMP derivatives that are summarized in this Account include lipid substitution, multivalent display of metalated di- or tripeptides on a trivalent scaffold with different linkers, and increasing the metal-to-peptide ratio such that every tryptophan in the (RW)3 scaffold is eventually replaced by a metalated lysine. While initial experiments with our OM-AMPs for systemic applications were largely disappointing, these OM-AMPs turned out to be potent antibiotics for topical applications. In this sense, two applications are described as examples in this Account, namely, bacterial decontamination of wastewater by reverse osmosis membranes (coated with our OMAMPs by Cu-catalyzed azide−alkyne cycloaddition reaction) and synergistic activities of one of our synAMPs with colistin and tobramycin for the treatment of Pseudomonas aeruginosa infections that are associated with cystic fibrosis. mutation of this bacterial target.2 In light of this, it was anticipated that resistance against an antibacterial agent that targets a less-defined part of bacteria would be much harder to achieve. Specifically, membrane-targeting antimicrobial peptides (AMPs) hold great promise as the next generation of antibacterial agents.3 Unfortunately, however, this strategy also did not remain unchallenged, as the first case of resistance against the membrane-targeting colistin has recently emerged in the USA;4 resistance of Pseudomonas aeruginosa was already shown earlier.5 Therefore, the race to novel antibacterial agents that can
1. INTRODUCTION Bacterial resistance against marketed antibiotics is one of the largest threats that our society faces.1 Without countermeasures, projections put the number of annual deaths caused globally by drug-resistant infections to 10 million by 2050.1 Since the advent of antibacterial agents, which was marked by the launch of the organometalloid compound arsphenamine (the first antibiotic that was widely applied), resistance has occurred. This has forced scientists to develop new antibacterial agents, a need that is still present. However, as the mode of action of most currently applied antibiotics depends on the tight interactionon an atomic levelof functional groups in the antibiotic with the bacterial target (enzyme), resistance can be acquired by simple © 2017 American Chemical Society
Received: June 3, 2017 Published: September 27, 2017 2510
DOI: 10.1021/acs.accounts.7b00282 Acc. Chem. Res. 2017, 50, 2510−2518
Article
Accounts of Chemical Research
study and fine-tune biological interactions.8 Numerous possibilities for the preparation of organometallic−peptide conjugates by solid-phase peptide synthesis (SPPS) are now well-established. In most cases, an organometallic carboxylic acid is introduced by acylation of a peptide-bound amino group. Using highly optimized protocols,9 we were able to prepare a large number of OM-AMPs, allowing us to deduce detailed structure−activity relationships (SARs) and finally identify OM-AMPs with muchenhanced activity over the lead compound 1-L. Our first-generation OM-AMPs were composed of Phe, Arg, and Trp residues in combination with either ferrocene (Fc) or cobaltocenium (Cc+) (Chart 1).10,11 Unfortunately, these first OM-AMPs ever synthesized showed disappointingly high minimal inhibitory concentration (MIC) values of >50 μg/mL. (In general, highly active AMPs have values in the range of 0.1− 10 μg/mL.) However, a first glimpse of SAR for OM-AMPs evolved in that with the right balance between the OM fragment and the peptide sequence, much better values should be possible. For example, it appeared that a higher hydrophobicity of the conjugate, as readily evidenced by its longer retention time on a C18 column during reversed-phase (RP) HPLC analysis, leads to more active AMPs. The fact that these OM-AMPs were bactericidal and rather nonselective, killing both Gram-positive and -negative bacteria (i.e., S. aureus and P. aeruginosa) with similar (albeit still low) efficiencies was additionally encouraging. Following this initial study, we embarked on a more detailed assessment of the role of the organometallic moiety with respect to the antibacterial properties. For this we changed the peptide sequence to one that was inspired by the group of Svendsen,7 namely, the WRWRW peptide fragment.12 Surprisingly, some of the resulting OM-AMPs were more active against Gram-negative E. coli than against Gram-positive S. aureus (Table 1). Although it proved to be not straightforward to relate the different properties of the metallocene moieties (Figure 2), we were able to show that all of the group 8 metallocenoyl OM-AMPs were potent antibacterial agents, at least against Gram-positive species.13,14 Importantly, these small OM-AMPs did not lead to significant hemolysis of human red blood cells (hRBCs), nor did they show significant toxicity against three selected human cancer cell lines. At this point, we had identified several apparently nontoxic OMAMPs with promising MIC values, i.e., in the 1−10 μg/mL range. With these compounds in hand, we shifted our attention to more in-depth studies of the mode of action of these short synthetic AMPs (synAMPs), with the intention to use this knowledge to design even more active OM-AMPs and to explore the potential clinical applicability of these short peptides.
successfully counter the challenge of resistance continues, even with membrane-targeting entities. Realizing that most antibiotics are purely organic compounds, we turned our attention to metal compounds, and more specifically organometallic moieties. (Organometallic compounds are molecules that contain at least one metal−carbon bond.) Despite the fact that some AMPs require the presence of a metal ion for their activity (e.g., daptomycin, which requires Ca2+),6 we reasoned that the introduction of an exotic molecular fragment like an organometallic moiety, which has not been encountered by bacteria, would offer additional possibilities to modify the conjugates’ properties and thereby the activity and selectivity of the antibacterial agent. Also, the unique properties of organometallic moieties should enable new methods to study the mode of action of these antibacterial agents and may even evoke dual modes of action. Therefore, we focused on peptides with a known preference for lipid-bilayer interaction that are not large enough to penetrate the membrane (Figure 1). As a starting
Figure 1. Cross-section of a typical biomembrane composed of phosphatidylethanolamine (PE) and phosphatidylcholine (PC), with emphasis on the dimensions of the membrane and the RWRWRW-NH2 lead structure.
■
UNRAVELING THE MODE OF ACTION OF SMALL CATIONIC AMPS Different proposals for the mode of action of small AMPs were summarized by Zasloff.3 It was assumed early on that short amphipathic AMPs, especially those that are rich in Arg and Trp residues, interact with the bacterial membrane.15 Since the precise cause for bacterial death was never established conclusively, we undertook an in-depth analysis of the mode of action of these short cationic synAMPs.16 One key objective was to identify the primary cellular target for these synAMPs. Although we obtained hints that the bacterial membrane was the primary target for these amphipathic synAMPs, conclusive evidence remained elusive. However, using the fact that ruthenium has a high electron density and is normally absent in bacteria, we could visualize its primary location in bacteria using transmission electron microscopy (TEM). In addition, we
point, we chose the cationic RWRWRW-NH2 hexapeptide (1-L) (see Chart 1 for its chemical structure), which is easily synthesized and modified. The amino acid sequence was optimized with activity against both Gram-negative Escherichia coli and Gram-positive Staphylococcus aureus, but its mode of action remained elusive.7 Peptide 1-L thus seemed an ideal starting point to investigate whether and how much incorporation of an organometallic moiety might enhance the antibacterial activity of the conjugates.
2. ORGANOMETALLIC ANTIMICROBIAL PEPTIDES (OM-AMPS) The introduction of organometallic (OM) moieties into biological systems has proven to be a valuable approach to 2511
DOI: 10.1021/acs.accounts.7b00282 Acc. Chem. Res. 2017, 50, 2510−2518
Article
Accounts of Chemical Research
Chart 1. First-, Second-, and Third-Generation Organometallic AMP (OM-AMP) Conjugates and (upper right) the Structure of the Hexapeptide Lead Sequence; The Following Abbreviations Are Sometimes Associated with These SynAMPs: 1 = MP196, 3 = MP66, 4 = MP276, and 5 = BA234
Table 1. Overview of Representative Members of the Second (Entries 1−6) and Third (Entries 7−8) Generations of SynAMP Derivatives
Bold red = positively charged hydrophilic; bold black = neutrally charged lipophilic; n.a. = not active. bThese “older” MIC values were obtained in a different lab in Heidelberg, Germany. cItalic one-letter amino acid codes refer to D-amino acids.
a
capitalized on the ability of atomic absorption spectrometry to quantify even traces of metals, in our case the amount of ruthenium in subcellular fractions. The results proved that OMAMP 4 was primarily bound to the bacterial membrane (Figure 3) and that the effects observed in other biochemical studies were related downstream to this primary event. On the basis of all of our combined observations and biochemical studies, a novel mode of action for synAMPs was discovered (Figure 4). In the native state of the bacterium, a line of enzymes performs the required transformations to build up the bacterial cell wall. Many of these proteins are membranelocated or at least membrane-anchored, and hence, the product from one transformation is smoothly passed on to the next enzyme as its substrate. Upon administration of the synAMP, which interacts with or even partly covers the membrane, the
Figure 2. Physicochemical properties of group 8 metallocenes. The absence or presence of hydrogen-bond formation to the metal ion is indicated by the red and blue ovals around the metal center, respectively. Reprinted from ref 8. Copyright 2016 American Chemical Society.
2512
DOI: 10.1021/acs.accounts.7b00282 Acc. Chem. Res. 2017, 50, 2510−2518
Article
Accounts of Chemical Research
and 4 behaved identically; Table 1) or of a C8 lipid as a membrane anchor.17 This implies that the above-described mechanism of membrane protein displacement is not unique for a specific compound but indeed seems to be rather universal for this type of synAMP. Surprisingly, none of our studies revealed any effect of the redox properties of the ferrocenoyl moiety on the activity of the OM-AMP. We did not detect unregulated catalase or elevated levels of H2O2. In fact, when we substituted the FcC(O) moiety with a lipid with similar hydrophobicity, i.e., C6H13C(O) or C8H17C(O) (to be more precise, the lipophilicity of ferrocene carboxylic acid (C10H9Fe) is best compared with that of heptanoic acid, C7H15; see Chart 2 for structures), we observed very comparable antibacterial activities and hemolysis.18 On this basis, it can be safely concluded that the enhanced activity of our OM-AMPs was not caused by any redox activity of the organometallic fragment attached to the peptidic structure. Thus, it was unlikely that altering the redox potential of the ferrocene moiety would lead to more active antibiotics. Clearly, to prepare more active OM-AMPs other means were required.
Figure 3. Side-by-side comparison of TEM pictures of Bacillus subtilis treated with 4 (left) and untreated (right), showing the darkened membrane caused by the presence of the ruthenium-derivatized OMAMP. For this study, cells were otherwise unstained. Percentages indicate the amounts of ruthenium present in the subcellular fractions as analyzed by atomic absorption spectrometry.
biophysical properties of the membrane change. As a consequence, the affinity of membrane-bound proteins is reduced, and in this case the MurG protein that carries out the lipid I → lipid II transformation is the first one to drop off the membrane. In the end, the biosynthesis of cell-wall constituents is halted and lipid I accumulates, leading to inhibition of the cellwall biosynthesis. To counter the effects of these small membrane-binding synAMPs, bacteria rapidly upregulated proteins involved in the anabolism of aspartic acid (Asp) and glutamic acid (Glu) and excreted substantial amounts of Glu. Interestingly, adding Glu to the growth medium of Bacillus subtilis increases the MIC value of the synAMP 1 8-fold; this indicates that the bacteria excreted Glu as a protective agent against the membrane-targeting cationic synAMP 1. This was further elegantly proven by studying B. subtilis lacking mechanosensitive channels; these transmembrane tunnels are triggered by turgor-induced pressure on the lipid bilayer, and as a consequence, Glu is excreted. The channeldeficient bacteria accumulated Glu intracellularly by 170% and remained sensitive to synAMP 1. In addition to the Glu-release defense mechanism, the proteomic profile of synAMP 1-treated B. subtilis indicated that the peptide-induced membrane stress was countered by (i) adjustment of the lipid composition of the membrane by activating alternative fatty acid biosynthesis enzymes, (ii) stabilization of the membrane by overexpressing phage-shock (related) proteins, and (iii) enhancement of cellwall teichoic acid D-alanylation, which should prevent the synAMP from approaching the membrane. Importantly also, the antibacterial mode of action was not severely affected by the presence or absence of the organometallic moiety (i.e., peptides 1
Further Increasing the Potency of OM-AMPs
After successful elucidation of the mechanism of action of our synAMPs, we developed a program to further explore the SAR in this class of antibacterial agents and possibly obtain more active synAMPs by implementing the following strategies: (a) alteration of the type of metallocene (second and third generations of OM-AMPs); (b) a metallocene scan to probe the optimal position of the OM fragment (fourth generation) and, potentially, combinations of different metallocenes; (c) L-toD substitution scan of all amino acid residues (fifth generation); and (d) conjugation of several synAMPs to one scaffolding molecule, i.e., synthesis of multivalent synAMPs. a. Metallocene Alterations (Second- and Third-Generation SynAMPs). The first-generation synAMPs containing cobaltocenium (Cc+) and ferrocene (Fc) derivatives (e.g., peptides 2 and 3) already indicated that the antibacterial properties were severely affected by the type of metallocene that was attached to the synAMP. To further explore this, we attached metallocenoyl derivatives of ruthenocene (Rc) and osmocene (Oc) that can be conveniently prepared. Indeed, changing Fc to Rc resulted in an OM-AMP that was 8-fold more active against MRSA (Table 1, entries 6 and 7). Interestingly, attachment of the osmocenoyl derivative (Oc) to the WRWRW sequence halted
Figure 4. Interference of OM-AMP 4 with the machinery of bacteria that construct the cell wall. 2513
DOI: 10.1021/acs.accounts.7b00282 Acc. Chem. Res. 2017, 50, 2510−2518
Article
Accounts of Chemical Research
antibacterial properties were comparable to those of the unmodified derivative. Furthermore, by combining both Nterminal and side-chain labeling, we were able not only to increase the “organometallic load” on the OM-AMP but also to combine different OM fragments, i.e., Fc and Rc (Chart 3).
Chart 2. Positioning of the Metallocene: Chemical Formulas of the Fourth-Generation SynAMPs Used To Study the Effect of the Position of the Organometallic Ferrocenoyl Moiety on the Antibacterial, Hemolytic, and Cell-Viability Properties of the Conjugates
Chart 3. Multimetallocene Substitution: Chemical Formulas of Two OM-AMPS in Which Tryptophan Moieties Have Been Replaced by FcLys Amino Acid Residues
the trend and did not produce a more active OM-AMP (Table 1, entries 7 and 8). As already mentioned above, the enhanced antibacterial properties of our organometallic−peptide conjugates were mostly due to the lipophilic properties of the Fc moiety. Since the lipophilicities of Fc, Rc, and Oc are all comparable to one another, as inferred from the identical retention times on an RP-HPLC column, there has to be another factor to explain the origin of the 8−16-fold difference in the activities of the three different group 8 metallocene-containing synAMPs. We attribute the enhanced activities to the 6% larger size of Rc and Oc compared with Fc and the hydrogen-bondaccepting properties of ruthenium and osmium ions in Rc and Oc, respectively, a property that iron in Fc lacks.19,20 Although evidence was lacking at the time, we reasoned that the tightly orchestrated bacterial membrane was more severely disrupted when bacteria were exposed to Rc- and Oc-AMPs compared to Fc-AMP. b. Ferrocene Scan and Ferrocene−Ruthenocene Combinations. Having seen the added effect of an organometallic moiety on the activity of Arg-Trp-based OM-AMPs, we addressed the questions of whether a particular position of the organometallic moiety would yield significantly more active OMAMPs and whether combinations of Fc and Rc would be beneficial. A tailor-made Fmoc-protected ferrocene-modified lysine (FcLys) derivative was used to replace tryptophan in the RWRWRW sequence; the N-terminus was unmodified or acylated with ferrocenoyl or ruthenocenoyl moieties. Although the substitution of a planar tryptophan residue with a spherical FcLys moiety constitutes a major chemical alteration, the
Remarkably, whereas the activity against Gram-positive bacteria was not significantly improved, OM-AMPs with high activity against Acinetobacter baumannii were discovered: MIC values as low as 4.7 μM were obtained for several N-terminally Fcderivatized OM-AMPs (for comparison, the control peptide MP196 (1) had a MIC value of 21 μM). These densely metalated OM-AMPs showed broad-spectrum activities, with similar or lower values against S. aureus and MRSA, and only limited hemolysis (
