With the recent emergence of reports on resistant Gram-negative ‘superbugs’, infections caused by multidrug-resistant (MDR) Gram-negative bacteria have been named as one of the most urgent global health threats due to the lack of effective and biocompatible drugs. Here, we show that a class of antimicrobial agents, termed ‘structurally nanoengineered antimicrobial peptide polymers’ (SNAPPs) exhibit sub-μM activity against all Gram-negative bacteria tested, including ESKAPE and colistin-resistant and MDR (CMDR) pathogens, while demonstrating low toxicity. SNAPPs are highly effective in combating CMDR Acinetobacter baumannii infections in vivo, the first example of a synthetic antimicrobial polymer with CMDR Gram-negative pathogen efficacy. Furthermore, we did not observe any resistance acquisition by A. baumannii (including the CMDR strain) to SNAPPs. Comprehensive analyses using a range of microscopy and (bio)assay techniques revealed that the antimicrobial activity of SNAPPs proceeds via a multimodal mechanism of bacterial cell death by outer membrane destabilization, unregulated ion movement across the cytoplasmic membrane and induction of the apoptotic-like death pathway, possibly accounting for why we did not observe resistance to SNAPPs in CMDR bacteria. Overall, SNAPPs show great promise as low-cost and effective antimicrobial agents and may represent a weapon in combating the growing threat of MDR Gram-negative bacteria.
A group of pathogens responsible for the majority of hospital-acquired infections—commonly referred to as the ‘ESKAPE’ pathogens—have been named as one of the biggest threats to health as a result of their multidrug resistance1,
Antimicrobial peptides (AMPs) have been widely regarded as a promising solution to combat MDR bacteria8. Unlike conventional antibiotics, which act on specific intracellular targets9, AMPs interact with microbial membranes through electrostatic interactions and physically damage the bacterial morphology10. The nature of this antimicrobial mechanism renders bacteria less likely to develop resistance to AMPs. However, AMPs have had limited success in clinical settings (with the exception of a few promising candidates being evaluated as topical agents, such as omiganan11), primarily due to their high toxicity towards mammalian cells12. Recently, ring-opening polymerization (ROP) of α-amino acid N-carboxyanhydrides (NCAs) has been shown to be a versatile method for the synthesis of peptide polymers with antimicrobial properties13,14, even though NCA–ROP cannot rival the precise peptide sequences obtained via solid-phase peptide synthesis. Nevertheless, advances in NCA–ROP have provided a facile route for the synthesis of well-defined peptide polymers with complex macromolecular architectures, such as star polymer nanoparticles15,16. We, along with several other research groups, have recently reported studies that demonstrated the potential of these star-shaped peptide polymer nanoparticles in nanomedicine17, particularly in the fields of gene therapy18,19 and targeted drug delivery20.
In this study, we report the discovery of star-shaped peptide polymer nanoparticles consisting of lysine and valine residues, synthesized via NCA–ROP, as a new class of antimicrobial agents. These star nanoparticles were termed ‘structurally nanoengineered antimicrobial peptide polymers’ (SNAPPs). Unlike existing self-assembled antimicrobial macromolecules, which will dissociate to unimers below their critical micelle concentration21,22, SNAPPs are stable unimolecular architectures up to infinite dilution. We demonstrate that SNAPPs exhibit superior antibacterial activity against a range of clinically important Gram-negative bacteria, possess high therapeutic indices and display selectivity towards pathogens over mammalian cells. Crucially, we show that SNAPPs can combat CMDR A. baumannii infection in mice. We also show that SNAPPs have a multimodal antimicrobial mechanism involving disruption of the integrity of the OM, cytoplasmic membrane (CM) disruption, unregulated ion efflux/influx and induction of apoptotic-like death (ALD), which potentially accounts for the superior performance of SNAPPs and differs from that typically reported for most AMPs.
Results and discussion
Synthesis and characterization of SNAPPs
To demonstrate the potential of this new class of antimicrobial nanomaterial, SNAPPs in the form of 16- and 32-arm star peptide polymer nanoparticles S16 and S32, respectively, were synthesized via NCA–ROP (number-average molecular weight, Mn = 43.8 kDa (S16), 74.8 kDa (S32); hydrodynamic diameter, DH = 7.7 nm (S16), 13.5 nm (S32); Fig. 1, Supplementary Section ‘Synthesis and Characterization of SNAPPs’, Supplementary Figs 1–8 and Supplementary Table 1), in a similar fashion to that reported by us18. Inspired by naturally occurring AMPs where the mode of action has been partly attributed to a positively charged amphipathic structure8 and after taking into account various synthetic considerations, lysine and valine were selected as cationic and hydrophobic amino acids, respectively. Their monomeric NCA derivatives were randomly polymerized from a PAMAM dendritic core to form the star arms with a theoretical lysine-to-valine ratio of 2:1, which was selected to promote water solubility and an average degree of polymerization (DP) of 30 residues per star arm at complete monomer conversion.
In vitro antimicrobial properties of SNAPPs
We evaluated the antibacterial efficacy of SNAPPs by determining their minimum bactericidal concentrations (MBCs) against a range of Gram-positive (Streptococcus mutans and S. aureus) and Gram-negative (Escherichia coli, Pseudomonas aeruginosa, Klebsiella pneumoniae and A. baumannii) bacteria. The MBC is defined as the minimum drug concentration that causes quantitative cell death (see Methods, Supplementary Fig. 9 and Supplementary Table 2 for further clarification). The antimicrobial susceptibility assays were initially conducted in Mueller–Hinton broth (MHB), a nutrient-rich bacterial growth medium. Our initial studies, which focused on the two Gram-positive bacteria (S. aureus and S. mutans) and two Gram negative bacteria (E. coli and P. aeruginosa) showed that S16 and S32 had preferential activity towards the Gram-negative species (MBC < 1.4 μM; Table 1) over the Gram-positive strains (MBC > 1.8 μM) (Supplementary Table 3). It is worth noting that homolysine star-shaped peptide polymer nanoparticles, that is, those that do not contain valine residues, exhibited a higher MBC value (greater than threefold increase in MBC against E. coli), thus demonstrating the need for an amphipathic structure to effect antimicrobial activity.
In further studies using Gram-negative bacteria, S16 and S32 were also found to be effective against K. pneumoniae and A. baumannii, with S16 registering MBC values of 1.54 and 0.85 µM, respectively, and S32 showing similar MBC values of 0.83 and 0.79 µM. Furthermore, both S16 and S32 were found to be equally effective against CMDR clinical isolates of P. aeruginosa (FADDI-PA067) and A. baumannii (FADDI-AB156) (Supplementary Tables 4 and 5) as to drug-sensitive strains, yielding MBC values ranging from 0.85 to 1.61 µM (Table 1). We observed that the MBC values of SNAPPs against all bacteria tested were within a similar order of magnitude, which implied that the antimicrobial efficacies of S16 and S32 might not be species-specific for Gram-negative bacteria. This was surprising, as P. aeruginosa and K. pneumoniae possess low antibiotic susceptibility as a result of their thick extracellular capsules, which contribute to reduced cell permeability23 and are a resistance mechanism against antibiotics and AMPs24,25. It is noteworthy that the probability of the PAMAM dendrimer cores contributing to the antimicrobial efficacies of S16 and S32 in this study was discounted, as the PAMAM cores were found to be non-active against the range of bacterial species tested above (Supplementary Table 6).
We also conducted an antimicrobial resistance study to evaluate if resistance against SNAPPs could be generated easily. Following serial passaging of bacterial cells in the presence of sub-MBC levels of S16, we did not obtain de novo resistant mutants of wild-type and CMDR A. baumannii to S16, even after 600 generations of growth (over a period of 24 days) (Supplementary Fig. 10). The MBC values of S16 against these bacterial strains remained relatively constant throughout the experiment, despite the fact that the CMDR strain is genetically capable of mutation and rapid resistance acquisition. This result suggests that resistance to SNAPPs is not acquired easily.
As a comparator, we synthesized the linear analogue L to represent one arm of S16 and S32 (Supplementary Figs 11–13 and Supplementary Table 7). Interestingly, compared to SNAPPs, the linear analogue was poorly active against E. coli and S. aureus, with MBCs at least 40-fold higher than those of SNAPPs (Supplementary Table 8). We hypothesize that the star architecture has a significant effect on enhancing the antimicrobial action of random co-peptide polymers against bacteria. Similar to that observed by Yang and co-workers, we theorize that the improvement in activity is a result of the increased local concentration of charges provided by the nanostructure, thereby leading to greater ionic interactions with the bacterial membranes22. Additionally, bacterial membrane-induced peptide aggregation has been postulated to be a key factor of AMP efficacy, as it enables AMPs to achieve the high threshold concentrations needed for membrane disruption26. Extending this concept, we hypothesize that the star architecture affords a high local concentration of peptide mass, even in solution, before contacting bacterial cells, which may contribute to the enhanced efficacy of SNAPPs.
The antimicrobial activities of S16 and S32 were compared with several peptide-based antimicrobial agents known to be effective against Gram-negative pathogens, including magainin II, ovispirin and melittin (Table 1). The antimicrobial efficacies of these AMPs against certain Gram-negative pathogens have been well-documented; however, they (as is the case with most AMPs) tend to demonstrate selective activity towards certain bacterial strains, even within the Gram-negative family (Table 1). This is in direct contrast to SNAPPs which displayed effective equipotent activity against all of the Gram-negative species tested. Furthermore, SNAPPs were orders of magnitude more effective than these AMPs (Table 1).
To explore the antimicrobial efficacy of SNAPPs in different media and to evaluate their toxicity against mammalian cells, we repeated the assays using minimal essential medium (MEM), which is a defined medium for mammalian cells. Both S16 and S32 exhibited at least four times lower MBC values (<0.19 µM) against all bacterial species tested in MEM compared with MHB. In MHB, SNAPPs were found to aggregate to form particles with average DH values of ∼224.2 nm or greater, which is significantly larger in size than SNAPPs in MEM (DH = 7.7 and 13.5 nm for S16 and S32, respectively) (Supplementary Figs 6 and 7). Cryo-TEM images of S16 in MHB further confirmed that the star formed large aggregates (∼200–500 nm in diameter) (Supplementary Fig. 14). The formation of aggregates is consistent with previous studies that have reported that nutrient/ion-rich media often contain anionic peptide and protein fragments, which might bind non-specifically to cationic antimicrobials27, thereby causing aggregation and the formation of larger-sized particles in MHB, as observed here. This postulation was validated as we observed a similar phenomenon where aggregates of S16, predominantly sized at 125 nm, were formed in 1% fetal bovine serum, another protein-rich medium (Supplementary Fig. 15). The aggregation of SNAPPs with the medium contents could possibly shield the active components of SNAPPs, thereby reducing their potency. Nevertheless, both S16 and S32 still possessed high efficacies in MHB, especially compared to the lead AMPs.
Biocompatibility of SNAPPs
As a test of biocompatibility, the haemolytic activities of SNAPPs were investigated by incubating them with red blood cells at different nanoparticle concentrations. Both S16 and S32 (as well as the control homolysine star) had negligible haemolytic activity (>45 μM, Supplementary Table 9). Even at a very high concentration of >100 × MBC, the extent of haemolysis was well below 30% (Supplementary Fig. 16). Subsequently, we investigated the viability of two types of mammalian cells (human embryonic kidney (HEK293T) cells and rat hepatoma (H4IIE) cells) in response to SNAPPs. The therapeutic indices (TI) of SNAPPs ranged from 52 to 171 (Supplementary Table 10), generally higher than the TI of colistin28, which is currently being used as the last therapeutic option for MDR Gram-negative pathogens29.
In vivo efficacy of SNAPPs
The effectiveness of S16 in vivo was evaluated in a mouse peritonitis model, where the intraperitoneal (i.p.) dose of A. baumannii (2 × 108 cells in MEM) resulted in the establishment of widespread bacterial infection by 24 h (Fig. 2a). At 0.5, 4 and 8 h post-infection, mice were treated with either MEM (control), the antibiotic imipenem (40 mg kg–1) or S16 (8.3 mg kg–1). Similar to the imipenem-treated mice, treatment with S16 resulted in >5-log reduction in bacterial cell counts in the peritoneal cavity (Fig. 2b), quantitative (>99%) eradication of bacterial cells in blood (Supplementary Fig. 17a) and >3-log colony forming unit (c.f.u.) reduction in the spleen (Supplementary Fig. 17b). Additionally, all mice treated with either imipenem or S16 survived, with no signs of animal distress, whereas only 20% of the control/mock-treated mice survived after 24 h. A number of studies have found that antimicrobial agents enhance host cell innate immunity to bacteria in vivo30; in this study, the S16-treated group had enhanced neutrophil infiltrate in the peritoneal cavity, while the imipenem-treated group did not show any significant difference from the mock-treated group (Fig. 2d).
As S16 was effective in vitro against CMDR bacteria, we extended the peritonitis model by including the CMDR A. baumannii. Mice treated with S16 had significantly less bacteria in the peritoneal cavity (Fig. 2c), blood (Supplementary Fig. 18a) and spleen (Supplementary Fig. 18b) and higher numbers of neutrophils in the peritoneal cavity (Fig. 2e) compared with the imipenem- and mock-treated groups. Imipenem treatment had no effect on reducing bacteria levels in all tissues examined, and this was comparable to the mock-treated group (Fig. 2c and Supplementary Fig. 18). All mice treated with S16 survived with no signs of animal distress, but only 50% of the mock- or imipenem-treated mice survived the 24 h infection. The mechanism by which host defence peptides (HDPs) effect bacterial clearance in vivo is by neutrophil recruitment; however, unlike SNAPPs, HDPs often have poor direct antimicrobial activity31. S16 has both direct (bacterial) and indirect (via neutrophil recruitment) antimicrobial activities in vivo. Although the CMDR A. baumannii isolate used herein has been found to acquire resistance against the last-resort drug, colistin32, these results demonstrated that S16 is capable of treating CMDR A. baumannii and potentially other CMDR Gram-negative infections in vivo. To the best of our knowledge, this is the first report of a synthetic antimicrobial polymer having in vivo efficacy against a CMDR Gram-negative bacterial infection.
Preliminary mechanistic studies
To directly observe the interactions between an antimicrobial agent and bacterial cells, we conducted super-resolution fluorescence imaging using 3D structured illumination microscopy (3D-SIM). Sample images of untreated E. coli are provided in Fig. 3a and Supplementary Fig. 20a. Figure 3 shows the 3D-SIM images of E. coli (labelled red with lipid membrane FM4-64FX dye) incubated with the AF488-labelled S16 (Supplementary Fig. 19 and Supplementary Table 11) in MHB at a dose approximately equivalent to 0.5×, 1× and 2× the MBC of the fluorescently tagged SNAPP (MBCtagged). The action of antimicrobials on bacteria has been imaged using a range of microscopy techniques33,34, but this is one of the first instances where a clear visualization of the interaction between an antimicrobial agent and bacterial cells at the super-resolution level has been demonstrated. S16 (labelled green) was found to associate at certain sites on and in the bacteria depending on SNAPP concentration (Fig. 3b–h, Supplementary Figs 20 and 21 and Supplementary Videos 1 and 2). This is different to the membrane interactions of proline-rich AMPs, where the peptides localize uniformly around the E. coli membrane, as previously reported by us35. At 0.5 × MBCtagged, SNAPPs associate with the surface of the bacteria (Fig. 3b); however, at 1 × MBCtagged, we observed a high density of bacterial cells with either membrane-associated or internalized star peptide polymers (Fig. 3c–e). On the other hand, in our previous study on membrane-lytic proline-rich AMPs, we observed complete internalization and uniform localization of peptides throughout the cytosol of the bacterial cells35. We attributed this difference to the larger sizes of SNAPPs, which possibly inhibited the quantitative internalization of all membrane-bound SNAPP macromolecules even when the membranes were disrupted. At 2 × MBCtagged, there are clearly more membrane-associated or internalized SNAPPs per bacterial cell (Fig. 3f–h and Supplementary Fig. 20b–e). Orthogonal projections showed that the membrane-localized SNAPPs appear to aggregate and span the cell envelope (Supplementary Fig. 21). Furthermore, bacteria with internalized SNAPPs were observed to have distinct cell envelope perforations (Fig. 3h, Supplementary Figs 20d,e and 21b). When the experiments were conducted in MEM at 2 × MBCtagged, the results were similar to those observed in MHB at supra-MBC dosage, albeit with a larger extent of SNAPP internalization into the cells (Supplementary Fig. 22).
Based upon the fluorescence imaging studies conducted, we hypothesized that SNAPPs initially localize on the bacterial OM as a result of electrostatic interactions. This could cause areas of destabilization/fragmentation, leading to SNAPPs possibly translocating to and disrupting the CM, thus leading to cell death36. To investigate this hypothesis, we conducted a competitive inhibition assay with lipopolysaccharide (LPS). The co-incubation of LPS (from E. coli) with S16 was found to inhibit the ability of S16 to disrupt the membrane of E. coli cells in a dose-dependent manner (Supplementary Figs 23 and 24). This suggested that SNAPPs bind to LPS on the OM and could explain the selective antimicrobial activity of SNAPPs towards Gram-negative bacteria. However, as SNAPPs were still moderately active against the Gram-positive species tested, we postulate that the LPS–SNAPP interactions are preferred, due to the strong electrostatic interactions, but not highly specific to the extent where antimicrobial activity would be lost in the absence of LPS. Using flow cytometry, we demonstrated that the association of the AF488-tagged S16 with E. coli was linearly correlated with bacterial membrane disruption (Supplementary Figs 25 and 26). Furthermore, a time-course study showed that the action of S16 was rapid, as >90% of an E. coli cell population had disrupted membranes within 30 min, which correlated with complete population death (Supplementary Fig. 27). This fast kinetics of killing is similar to that observed for AMPs (ref. 37), which is most likely a distinctive feature of membrane-disruptive antimicrobials.
To investigate whether SNAPP-induced membrane disruption is a result of pore formation, as is the case with some AMPs, we conducted a dye release assay using dextran-loaded large unilamellar vesicles (LUVs) as a mimic for Gram-negative CM (ref. 38). The results suggested that SNAPPs do not cause membrane disruption via pore formation (Supplementary Fig. 28a). Another mechanism of membrane disruption is through unregulated ion movement, which we investigated using the LUV chloride ion (Cl−) transport assay39. Cl− ion efflux was found to increase with increasing concentrations of S16 (Supplementary Fig. 28b). Additionally, we performed membrane potential measurements on bacterial cells to determine the ability of SNAPPs to alter the membrane potential. It was observed that the treatment of E. coli cells with S16 induced mixed hyperpolarized and depolarized bacterial cell populations, with a shift towards a more depolarized population as concentration increased (Supplementary Figs 29 and 30). Taken together, these studies suggest that the interaction of SNAPPs with the CM may result in membrane perturbations that lead to unregulated ion movement and membrane potential dissipation.
A number of recent studies have shown that bacteria, like eukaryotic cells, have mechanisms of programmed cell death (PCD) that could be triggered under stressful conditions, such as membrane disruption40,41. Two major PCD pathways have been described in bacteria: (1) the ALD pathway mediated by recA and lexA genes and (2) the mazEF pathway40,41. Based on gene expression studies, S16 at 1 × MBC induced a ten- and sevenfold increase in recA and lexA, respectively, but no change in mazEF levels (Supplementary Fig. 31). These results, when observed under cell death conditions, suggested that S16 induced ALD responses in E. coli. We also observed the production of reactive oxygen species (ROS) following S16 treatment (Supplementary Fig. 32), which has been reported to be a characteristic of ALD (ref. 42). The induction of ALD in bacteria has been reported in previous studies to lead to cell lysis42. We also found that when ALD was inhibited by pre-treating E. coli with a translation inhibitor (doxycycline), the antimicrobial activity of S16 remained comparable to that when ALD was not inhibited (Supplementary Fig. 33). This suggested that the induction of the ALD pathway is not a prerequisite for SNAPP activity, but is likely to be either an event that coincides with early membrane disruption or a supplementary (but not essential) bactericidal mechanism. At 5 × MBC, S16 induced significantly less recA and lexA mRNA than that at 1 × MBC concentration (Supplementary Fig. 31). We speculate that at supra-MBC dosage, other killing mechanisms would dominate, thus leading to insufficient time for the expression of ALD pathway components. This agreed with our previous postulation that there might be multiple mechanisms involved in the antimicrobial action of SNAPPs.
Next, we used cryo-TEM to visualize the effect of SNAPP treatment on E. coli cell morphology. Before treatment, all cells showed intact OMs and CMs (Fig. 4a and Supplementary Fig. 34a–c). After treatment with S16 in MHB at its MBC, large aggregates—probably formed by aggregation between S16 and media contents (vide supra)—were observed around the cells (Fig. 4b–g and Supplementary Fig. 34d–g). The cell membranes of bacteria incubated with S16 appeared disrupted (Fig. 4b) and had pores that traversed the OM, peptidoglycan (PG) layer and CM (Fig. 4c and Supplementary Fig. 34d). Most bacterial cells had fragmented or perforated OMs (Fig. 4d,e and Supplementary Fig. 34e,f) and some cells appeared to be broken into isolated fragments (Fig. 4g and Supplementary Fig. 34g). These observations were in agreement with the flow cytometry data, which indicated that SNAPP association leads to membrane disruption. Cryo-TEM experiments conducted in MEM at supra-MBC dosage resulted in observations similar to when MHB was used at similar dosages; however, more drastic cell lysis was noted in the case of MEM (Supplementary Fig. 35).
Taken together, we postulate that SNAPPs have a multimodal mechanism of inducing bacterial cell death. Initially, SNAPPs bind via electrostatic interactions with LPS and the OM, leading to destabilized/fragmented areas. They then assemble and traverse the cell envelope driven by the transmembrane electrical potential (the interior/cytosol being negative), most probably causing membrane perturbations that result in unregulated transmembrane ion movement in the CM. These membrane disruption events induce ALD at low SNAPP concentrations, thus leading to cell lysis42. At high concentrations, SNAPPs rapidly cause cell lysis by direct disruption of the OM and CM. Collectively, fluorescence imaging, flow cytometry and cryo-TEM assays provided evidence for membrane association, membrane disruption, OM fragmentation/destabilization and cell lysis. Although the exact sequence of events requires further investigations, we can conclude that the succession of antimicrobial events, as depicted in Fig. 5, is substantially different from the action of a monomeric cationic AMP. A typical monomeric AMP is commonly thought to traverse the OM of a Gram-negative bacterial cell via self-promoted uptake, binds to and inserts itself into the anionic surface of the CM, then kills the bacteria by either membrane disruption (pore formation) or translocation across the CM and acting on internal targets12,43,44. In the case where cell death is caused by membrane disruption, loss of CM integrity was commonly thought to be the lethal event12,45,46. Cryo-TEM analysis of E. coli after incubation with melittin or ovispirin (which are well-characterized AMPs) confirmed that the AMPs disrupted the CM while leaving the OM intact (Supplementary Fig. 36). On the other hand, as demonstrated here, it is likely that S16 effects its antimicrobial action in a cascade manner and by first disrupting the physical integrity of the OM upon binding with LPS. We hypothesize that the difference in the mode of action between monomeric cationic AMPs and S16 might be attributed to the latter's structural architecture, as SNAPPs might be unable to translocate across the OM and CM without causing significant perturbation.
AMPs as a whole class of compounds kill bacteria by multiple mechanisms, but each specific AMP tends to kill bacteria by one major mechanism8. The multifaceted interactions shown here between SNAPPs and bacteria (in combination with indirect antimicrobial activity via neutrophil recruitment in vivo) are truly unique and this, to the best of our knowledge, has not been shown previously in a definitive manner for any one single AMP. Silver ions have been shown to enhance antibiotic activity by interacting with several cytosolic molecules, which corroborates our data suggesting that an antimicrobial agent may have more than one mode of action47. The equipotency of SNAPPs against all of the Gram-negative bacteria tested suggested that the multimodal mechanism of action is non-specific, and we postulate that this is why bacteria did not acquire resistance to star S16, even after 600 generations of growth in the presence of the agent. The successful demonstration of the antimicrobial efficacy of SNAPPs against CMDR A. baumannii in vivo as well as their biocompatibility will undoubtedly distinguish SNAPPs as a new class of antimicrobial agents, potentially capable of addressing the dearth of suitable drug candidates to combat Gram-negative pathogens resistant to conventional antibiotics.
Synthesis of l-lysine(Z)–NCA (Lys NCA)
Dried H–Lys(Z)–OH (1.24 g, 4.43 mmol) was added to anhydrous THF (25 ml) in an oven-dried two-necked round-bottomed flask under argon. Triphosgene (580 mg, 1.96 mmol) was dissolved in anhydrous THF (5 ml) and added to the H–Lys(Z)–OH suspension. The mixture was heated at 50 °C for 30 min with continuous stirring. The clear solution was allowed to cool to room temperature and added to anhydrous pentane (100 ml). The resulting precipitate was isolated via centrifugation and washed with anhydrous pentane (30 ml × 2). The resulting white solid was dried at ambient temperature in vacuo to afford Lys NCA, 0.910 g (81%). 1H NMR (400 MHz, d6-DMSO) δH 1.23–1.37 (m, γ-CH2, 2H), 1.37–1.45 (m, δ-CH2, 2H), 1.60–1.80 (m, β-CH2, 2H), 2.94–3.02 (m, ε-CH2, 2H), 4.40–4.43 (m, α-CH, 1H), 5.00 (dd, C6H5CH2-, 2H), 6.90 (s, cyclic NH, 1H), 7.30–7.39 (m, C6H5-, 5H).
Synthesis of dl-valine–NCA (Val NCA)
Dried dl-valine (1.24 g, 4.43 mmol) was dissolved in anhydrous THF (25 ml) in an oven-dried two-necked round-bottomed flask under argon. Triphosgene (580 mg, 1.96 mmol) was dissolved in anhydrous THF (5 ml) and added to the dl-val–THF suspension. The mixture was heated at 50 °C for 30 min with continuous stirring. The clear solution was allowed to cool to room temperature and precipitated with anhydrous pentane (100 ml), followed by washing with more anhydrous pentane (30 ml × 2). The resulting residue was dried at ambient temperature in vacuo to afford Val NCA, 0.956 mg (85%). 1H NMR (400 MHz, d6-DMSO) δH 0.91 (dd, CH3, 6H), 2.00–2.12 (m, CH, 1H), 4.32 (dd, cyclic CH, 1H), 9.06 (s, cyclic NH, 1H).
Synthesis of poly(Z-l-lysine–r-dl-valine)armPAMAM–(NH2)16,core star peptide polymer S16Z
Lys NCA (1.3 g, 4.19 mmol) and Val NCA (0.3 g, 2.1 mmol) were dissolved in anhydrous DMF (16 ml) and added via syringe to PAMAM–(NH2)16 (dried, 43 mg, 13.1 µmol) dissolved in anhydrous DMF (1 ml). After stirring for 24 h under argon, n-butyl alcohol (1 ml) was added and the mixture was stirred for a further 1 h. Precipitation of the concentrated peptide polymer solution into diethyl ether (3 × 40 ml), followed by isolation via centrifugation and drying (0.1 mbar), afforded (PZLL–r-PVal)armPAMAM–(NH2)16,core star peptide polymer S16Z as an off-white solid, 1.21 g (90%). 1H NMR (400 MHz, d6-DMSO) δH 0.67–0.89 (b, CH3, 6H), 1.11–1.77 (b, γ-CH2 + δ-CH2 + β-CH2, 6H), 1.84–2.00 (b, CH, 1H), 2.78–3.00 (b, ε-CH2, 2H), 4.06–4.40 (b, α-CH, 1H), 4.90–5.00 (b, C6H5CH2-, 2H), 7.00–7.44 (b, C6H5-, 5H), 7.60–8.30 (b, NH, 1H).
Synthesis of poly(Z-l-lysine–r-dl-valine)armPAMAM–(NH2)32,core star peptide polymer S32Z
Lys NCA (1.3 g, 4.19 mmol) and Val NCA (0.3 g, 2.1 mmol) were dissolved in anhydrous DMF (16 ml) and added via syringe to PAMAM–(NH2)32 (dried, 43 mg, 13.1 µmol) dissolved in anhydrous DMF (1 ml). After stirring for 24 h under argon, n-butyl alcohol (1 ml) was added and the mixture was stirred for a further 1 h. Precipitation of the concentrated peptide polymer solution into diethyl ether (3 × 40 ml), followed by isolation via centrifugation and drying (0.1 mbar), afforded (PZLL–r-PVal)armPAMAM-(NH2)32,core star peptide polymer S32Z as an off-white solid, 1.15 g (85%). 1H NMR (400 MHz, d6-DMSO) δH 0.67–0.89 (b, CH3, 6H), 1.11–1.77 (b, γ-CH2 + δ-CH2 + β-CH2, 6H), 1.84–2.00 (b, CH, 1H), 2.78–3.00 (b, ε-CH2, 2H), 4.06–4.40 (b, α-CH, 1H), 4.90–5.00 (b, C6H5CH2-, 2H), 7.00–7.44 (b, C6H5-, 5H), 7.60–8.30 (b, NH, 1H).
Synthesis of linear poly(Z-l-lysine–r-dl-valine) peptide polymer LZ
Lys NCA (0.5 g, 1.63 mmol) and Val NCA (117 mg, 0.82 mmol) were dissolved in anhydrous DMF (6 ml) and added via syringe to 4-methylbenzylamine (10.3 µl, 80.9 µmol). After stirring for 24 h under argon, n-butyl alcohol (1 ml) was added and the mixture was stirred for a further 1 h. Precipitation of the concentrated peptide polymer solution into diethyl ether (3 × 40 ml), followed by isolation via centrifugation and drying (0.1 mbar), afforded linear PZLL–r-PVal peptide polymer LZ as an off-white solid, 420 mg (81%). 1H NMR (400 MHz, d6-DMSO) δH 0.67–0.89 (b, CH3, 6H), 1.11–1.77 (b, γ-CH2 + δ-CH2 + β-CH2, 6H), 1.84–2.00 (b, CH, 1H), 2.78–3.00 (b, ε-CH2, 2H), 4.06–4.40 (b, α-CH, 1H), 4.90–5.00 (b, C6H5CH2-, 2H), 7.00–7.44 (b, C6H4-, 4H), 7.60–8.30 (b, NH, 1H).
General procedure for deprotection of peptide polymers
The peptide polymer was dissolved in TFA (200 mg ml–1) and 33% HBr in acetic acid was then added (20 ml g–1 peptide polymer). After 24 h stirring at room temperature, the mixture was precipitated into diethyl ether (ten times the volume of the reaction). The precipitate was isolated via centrifugation, redissolved in hydrochloric acid solution (0.2 M, 0.2 ml mg–1 peptide polymer) and dialysed against RO water for 4 days. The dialysed solution was lyophilized to obtain the deprotected SNAPP (for example, S16 and S32). S16 and S32: 1H NMR (400 MHz, d6-DMSO) δH 0.67–0.88 (b, CH3, 6H), 1.05–1.77 (b, γ-CH2 + δ-CH2 + β-CH2, 6H), 1.84–2.00 (b, CH, 1H), 2.61–2.83 (b, ε-CH2, 2H), 4.00–4.39 (b, α-CH, 1H), 7.60–8.30 (b, NH, 1H). L: 1H NMR (400 MHz, d6-DMSO) δH 0.67–0.88 (b, CH3, 6H), 1.05–1.77 (b, γ-CH2 + δ-CH2 + β-CH2, 6H), 1.84–2.00 (b, CH, 1H), 2.61–2.83 (b, ε-CH2, 2H), 4.00–4.39 (b, α-CH, 1H), 7.00–7.20 (b, C6H4-, 4H), 7.60–8.30 (b, NH, 1H).
Synthesis of AMPs (ovispirin, magainin II and melittin)
Ovispirin (NH2-KNLRRIIRKIIHIIKKYG-COOH), magainin II (NH2-GIGKFLHSAKKFGKAFVGEIMNS-CONH2) and melittin (NH2-GIGAVLKVLTTGLPALISWIKRKRQQ-CONH2) were chemically synthesized on a CEM Liberty microwave peptide synthesizer (Ai Scientific). The peptide-resins were assembled from Fmoc-Rink-AM SURE resin in the Fmoc/TBu mode of synthesis. For a 0.1 mmol reaction scale, Fmoc-deprotection was performed in two stages by initial treatment with 20% piperidine/0.1 M HOBt/DMF (vol/vol, 7 ml) under microwave radiation for 30 s (40 W, 40 °C), followed by filtration and a second addition of the above solution (45 W, 75 °C; 3 min). The peptide-resins were then rinsed with DMF (4 × 7 ml). Acylation, where required, was achieved by the addition of a solution containing amino acid (5 equiv., relative to reaction scale), 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) (5 equiv.) and N,N-diisopropylethylamine (DIEA) (10 eq) in DMF/NMP (7:1, vol/vol; 4 ml) to the Nα-deprotected peptide-resin, and the mixture was agitated under microwave radiation for 10 min (30 W, 75 °C, vessel under external chilled air flow). Dichloromethane (DCM) (5 × 2 min) was used to rinse the peptide-resins before the cleavage step. The peptide was cleaved from the resin support by the addition of TFA/triisopropylsilane (TIPS)/thioanisole/phenol/water (90:2.5:2.5:2.5:2.5, % vol/vol/vol/vol/vol; 5 ml) for 2.5 h, after which the combined cleavage filtrates were evaporated under nitrogen flow and the crude product was isolated by precipitation in cold ether (4 × 30 ml).
The crude peptide was purified using an Agilent 1200 series liquid chromatograph instrument (Agilent) equipped with an ultraviolet detector (model G1316A) and a Zorbax 300 SB-C18 reversed phase column (9.4 mm × 25 cm). Crude peptide analysis was achieved using a linear acetonitrile gradient in 0.1% TFA at a flow rate of 4 ml min–1 (linear gradient of 0–54% CH3CN over 15 min). Analysis of the purified peptide was performed using an Esquire HCT electrospray ionization-mass spectrometry system (Bruker Daltronics).
Bacterial cell culture
Freeze-dried cultures of E. coli (ATCC 25922), K. pneumoniae (ATCC 13883), A. baumannii (ATCC 19606), colistin and multi-drug resistant (CMDR) A. baumannii (FADDI-AB156), CMDR P. aeruginosa (FADDI-PA067) and S. aureus (ATCC 29213) were grown aerobically and maintained by passage at ambient temperature on horse blood agar (10% vol/vol defibrinated horse blood, 4.4% wt/vol Oxoid Blood Agar Base No. 2). P. aeruginosa (ATCC 47085) were cultured in a similar fashion, except at 37 °C. Freeze-dried cultures of S. mutans (Ingbritt strain) were grown anaerobically and maintained by passage at 37 °C on Todd Hewitt agar (3.6% wt/vol Oxoid Todd-Hewitt Broth, 1.5% wt/vol sucrose, 1.5% wt/vol Bacto agar, 0.8% wt/vol Oxoid yeast extract). For E. coli, K. pneumoniae, P. aeruginosa, CMDR P. aeruginosa, A. baumannii, CMDR A. baumannii and S. aureus, overnight cultures were made by transferring a colony (∼half a loop) from the agar plates to culture tubes containing sterilized Luria–Bertani broth (LB, 1% wt/vol bacto tryptone, 1% wt/vol NaCl, 0.5% wt/vol Oxoid yeast extract) (20 ml). Bacterial cultures were incubated overnight at 37 °C with aeration and without agitation, with the exception of CMDR P. aeruginosa, which was cultured at 37 °C with aeration and agitation (150 r.p.m.). The next day, for E. coli, K. pneumoniae, P. aeruginosa, CMDR P. aeruginosa and S. aureus, small aliquots (that is, 0.5–2 ml) were taken from the culture tubes, further diluted with LB (20 ml), and incubated for 3–4 h at 37 °C with aeration before use. All bacterial cultures were cultured without agitation, with the exception of P. aeruginosa and CMDR P. aeruginosa, which were cultured with shaking at 150 r.p.m. For A. baumannii, an aliquot of 0.5 ml was taken from the overnight culture tube, further diluted with LB (200 ml) and incubated overnight at 37 °C with aeration before use. With regards to S. mutans, several colonies (∼half a loop) from the agar plates were transferred to culture tubes containing sterilized Todd Hewitt broth (3.6% wt/vol Oxoid Todd Hewitt broth, 1.5% wt/vol sucrose, 0.8% wt/vol Oxoid yeast extract) (20 ml). The cultures were incubated overnight at 37 °C in the anaerobic chamber. After 24 h, a small aliquot (that is, 0.5 ml) was taken from the culture tubes, further diluted with media (∼200 ml) and incubated overnight at 37 °C in the anaerobic chamber before use.
Bacterial cell counting
A Cell Lab Quanta SC MPL flow cytometer was used to count the number of bacterial cells before use in assays. Cells were diluted with saline using an appropriate dilution factor and incubated with Syto 9 and propidium iodide (PI) (that is, 1 ml cell solution to 1 µl of each dye). Syto 9 stains the nucleic acids in all cells, while PI stains the nucleic acids in cells with damaged membranes. Using the Cell Lab Quanta SC software, the number of viable cells per ml (Syto 9-positive, PI-negative) was obtained.
Measurement of MBC
A dilution series of each compound was made by diluting test compound stock in media to a desired range of concentrations and a final volume of 100 µl in each well of a 96-well plate. Bacterial cells (which gave an optical density (OD) reading of ∼0.7 at 650 nm for E. coli, K. pneumoniae, P. aeruginosa and S. aureus, ∼0.5 at 650 nm for A. baumannii and ∼1.8 at 650 nm for S. mutans) were diluted to 2.5 × 106 cells ml−1 in media, and 100 µl of the bacteria solution was added to each well. The 96-well plate was then incubated at 37 °C for 90 min. For each well, microbial solution was diluted with saline (0.9% NaCl solution) using an appropriate dilution factor and placed on an agar plate (identical to that used for bacteria culture). For E. coli, K. pneumoniae, A. baumannii and S. aureus, the agar plates were incubated overnight at room temperature and then at 37 °C with aeration for 2 h. For P. aeruginosa and S. mutans, the agar plates were incubated at 37 °C, with the former being incubated overnight with aeration and the latter in an anaerobic chamber for 48 h. The numbers of c.f.u. were counted and expressed as c.f.u. ml−1. Positive controls consisting of cells without any treatment were used. Concentration-killing curves were plotted with c.f.u. ml−1 as a function of compound concentration, and linear regression analysis was used to determine the lowest concentration (MBC) at which the c.f.u. ml−1 becomes zero (Supplementary Fig. 9). A minimum of two independent experiments (biological replicates) of the assay were conducted, and two technical replicates were used in each experiment for each bacterium, compound and concentration. Data are expressed as mean ± standard deviation (s.d.) of the biological replicates and analysed using Student's t-test. Note that for E. coli, K. pneumoniae, P. aeruginosa and A. baumannii, two sets of experiments were performed, either using a nutritionally rich medium (MHB, 3.8% wt/vol Oxoid Mueller–Hinton agar) or MEM (136.9 mM NaCl, 10.1 mM Na2HPO4, 2.7 mM KCl, 1.8 mM KH2PO4, 0.2% wt/vol d-(+)-glucose). For S. aureus and S. mutans, the assays were conducted in nutritionally rich media only (Luria–Bertani broth and Todd–Hewitt broth, respectively).
Measurement of MIC
The MIC values of the PAMAM dendrimers were determined using a broth microdilution method. After the preparation of a dilution series of each compound and the addition of bacterial cells (note: the steps are identical to those taken for the measurement of MBC), the optical density readings of each well at 630 nm were measured as a function of time using a microplate reader (Multiskan Ascent, Pathtech). Positive controls containing cells alone were incorporated. Optical density was plotted against polymer concentration, and linear regression analysis was used to determine the lowest concentration (MIC) at which the optical density reading becomes zero. A minimum of two independent experiments (biological replicates) of the assay were conducted, and two technical replicates were used in each experiment for each bacterium, polymer and concentration. Data are expressed as mean ± s.d. of the biological replicates and analysed using Student's t-test.
The method used was adapted from that of Gullberg and co-authors48. Overnight cultures of A. baumannii cells (ATCC 19606 or FADDI-AB156) in LB broth were obtained from independent colonies grown on horse blood agar. The cells were then serially passaged by 400-fold dilution in 1 ml batch cultures every 24 h for 600 generations (∼25 generations of growth per serial passage) in MHB containing 1/10 of the MBC of S16 (for both strains). After every 100 generations of growth, the MBCs of S16 were obtained using cells that were serially passaged in the presence of the antimicrobial agent. As a control, MBCs were also obtained using cells serially passaged in fresh MHB alone.
Fresh sheep red blood cells (RBCs) were diluted 1:20 in PBS (pH 7.4), pelleted by centrifugation and washed three times in PBS (1,000g, 10 min). The RBCs were counted using a cell counter (Coulter Particle Counter Z series, Beckman Coulter) and diluted to a final concentration of 2 × 107 cells ml−1. Aliquots (100 µl) of the RBC solution were seeded into a V-bottomed 96-well plate containing 100 µl of test compound solution of varying concentrations (4–2,000 µg ml–1) and incubated in a humidified atmosphere containing 5% CO2 at 37 °C for 2 h. Following incubation, the 96-well plate was centrifuged (1,000g, 10 min) and aliquots (100 µl) of supernatant were transferred to a flat-bottomed 96-well plate. Haemoglobin release upon lysis of the RBCs was monitored at 405 nm using a microplate reader (PerkinElmer 1420 Multilabel Counter VICTOR3). Positive and negative controls for haemolysis were taken as RBCs lysed with 0.5% Triton X-100 (1:1 vol/vol) and RBC suspension in PBS, respectively. The percentage of haemolysis was calculated using the formulaThe percent haemolysis was plotted against peptide polymer concentration, and linear regression analysis was used to determine the haemolytic concentration needed to lyse 50% (HC50) of RBCs. Two independent runs of the assay were conducted and two replicates were used in each run for each compound and concentration.
Mammalian cell culture
Human embryonic kidney cells (HEK293T) were cultivated in ‘complete’ RPMI-1640 medium (supplemented with 5% FBS, 1 × GlutaMAX, 1× antibiotic-antimycotic and 1× MEM non-essential amino acids) in a humidified atmosphere containing 5% CO2 at 37 °C. Cells were seeded in a T75 flask (∼3 × 106 cells ml−1) and passaged twice a week before performing subsequent cell viability studies. Rat hepatoma cells (H4IIE) were cultivated in DMEM medium (supplemented with 10% FBS, 1× GlutaMAX and 1× penicillin-streptomycin) in a humidified atmosphere containing 5% CO2 at 37 °C. Cells were seeded in a T75 flask (∼3 × 106 cells ml−1) and passaged twice a week before performing subsequent cell viability studies.
Adherent HEK293T or H4IIE cells (obtained from the ATCC and throughout the course of the study were checked for mycoplasma contamination using the Mycoplasma stain kit Myc1, Aldrich) were grown to 80% confluence and trypsinized before assay. HEK293T and H4IIE cells were chosen for this study as they are standard cell lines used in toxicity studies. Cells were diluted 1:2 with ‘complete’ medium (RPMI-1640 for HEK293T cells or DMEM for H4IIE cells) and seeded in a 24-well plate (1 ml per well). The cells were incubated at 37 °C in 5% CO2 for 24 h until ∼95% confluence. The medium was removed. Varying concentrations of test compound (4–128 µg ml–1) were prepared, and 200 µl aliquots of each were added to the cells, after which the cells were incubated at 37 °C in 5% CO2 for 90 min. The cells were then collected and all well contents were transferred to round-bottomed polypropylene tubes (5 ml). The cells were washed with cold Dulbecco's phosphate-buffered saline (DPBS), then stained with YO-PRO-1 and PI (0.2 ml from a stock solution, whereby both dyes were diluted 1:1,000 in cold DPBS, per well) and incubated on ice for 20–30 min. The cells were analysed by flow cytometry (Cytomics FC 500 MPL System). Standard compensation was performed using single-colour stained cells. Negative controls using untreated cells were included. Two independent runs of the assay were conducted, and two replicates were used in each run for each test compound and concentration.
In vivo efficacy of SNAPP S16
All experiments involving animals were performed according to protocols approved by the University of Melbourne Biochemistry and Molecular Biology, Dental Science, Medicine, Microbiology and Immunology, and Surgery Animal Ethics Committee (project no. 1513489). Ten- to 14-week-old female C57BL/6 mice (weighing 23.2 ± 1.7 g, animals under 20 g were not used in this study) were used in all in vivo studies, with five animals per group. Experiments were conducted without randomization or blinded protocol. Using preliminary peritonitis infection data and a power analysis (using SPSS for Windows, version 12), a sample size ≥2 would be needed to detect a large effect size (d = 0.8) with 95% power, using a t-test between means with alpha at 0.01. After 1 week of quarantine, inoculation (t = 0) was performed by i.p. injection of 300 µl of 2 × 108 cells, delivered in MEM, of wild-type A. baumannii (ATCC 19606) or CMDR A. baumannii (FADDI-AB056) with a 25-gauge syringe. Two groups (n = 5 for ATCC 19606 and n = 5 for FADDI-AB056) received either SNAPP S16 (8.3 mg kg–1 per dose in MEM, which corresponds to 1.5 × in vitro MBC taking into account the average peritoneal/blood volume of mice) or imipenem (derived from the carbapenem antibiotic family and considered to be the most successful class of antibiotics in evading emerging antimicrobial resistance49, 40 mg kg–1 per dose in MEM) treatment 0.5, 4 and 8 h after introduction of the inoculums. An untreated control group was included. Signs of animal distress were monitored and mice that did not meet distress-related euthanasia criteria at t < 24 h were defined as ‘survived’. At t = 24 h, all mice were euthanized. Peritoneal washes were performed by injecting 3.0 ml of sterile MEM in the intraperitoneal cavity followed by a massage of the abdomen. Subsequently, the abdomen was opened and 3.0 ml of peritoneal fluid was recovered from the peritoneum for analysis of c.f.u. ml−1. Spleen of each mouse was removed and suspended in 5.0 ml MEM in a gentleMACS tube, which was then subjected to automatic dissociation (gentleMACS dissociator, Miltenyl Biotec). The peritoneal fluid and supernatant from the dissociation of spleen were serially diluted in saline. A 10 µl portion of each dilution was plated on horse blood agar plates and incubated overnight at 37 °C. For mice that were still alive directly before euthanization, blood was also taken from the heart for immediate plating on horse blood agar plates. Colonies were counted and expressed as c.f.u. ml−1, and viable bacteria cell counts in the peritoneal cavity (Fig. 2b,c), blood (Supplementary Figs 17a and 18a) and spleen (Supplementary Figs 17b and 18b) were compared with those of the control group at 24 h. The bacterial levels were statistically analysed using a one-way classification analysis of variance (ANOVA) and Student's t-test (SPSS for Windows, version 12). Data are expressed as mean ± s.d. of five biological replicates.
Fluorescent tagging of SNAPP S16 with Alexa Fluor 488
SNAPP S16 was dissolved in sodium bicarbonate buffer (0.1 M, pH 8.3) (2.5 mg ml–1), and Alexa Fluor 488 (AF488) carboxylic acid succinimidyl ester dissolved in DMSO (10 mg ml–1) was added (20 µl mg–1 of peptide polymer) The mixture was stirred for 1 h at room temperature and then passed through a gel separation column (PD MidiTrap G-25) to remove excess dye. The filtrate was lyophilized to afford the fluorescently tagged derivative, AF488-S16.
Sample preparation for imaging with 3D-SIM super-resolution microscopy
Sterilized chambered coverglasses were coated with poly-d-lysine (0.1 ml per well from a 0.1 mg ml–1 stock solution in DPBS) for 90 min. The excess poly-d-lysine was removed by washing with sterilized MilliQ water (2 × 0.5 ml) and the coverglasses were left to dry overnight in a sterile environment. E. coli cells (1.25 × 106 cells ml−1, prepared as described for the measurement of MBC) were incubated with AF488-S16 (8–256 µg ml–1) in a 96-well plate at 37 °C for 90 min. The cell suspension was then transferred to Eppendorf tubes (2 ml) and washed with Hank's balanced salt solution (HBSS) twice (5,000g, 10 min). FM 4-64FX dye (0.2 ml from a 5 µg ml–1 stock solution in HBSS) was added to the cell pellet and the cells were incubated on ice for 10 min with regular mixing. After incubation, the cells were washed with HBSS (5,000g, 10 min), resuspended in HBSS (0.5 ml) and then transferred to the chambered coverglass. Subsequently, the cells were washed in HBSS (800g, 10 min) then fixed in 2% wt/vol paraformaldehyde in PBS for 10 min at room temperature. The fixative was removed with HBSS (800g, 10 min) and HBSS (0.5 ml) was added to each well before imaging.
LPS inhibition assay
SNAPP S16 (50 µl) was incubated with LPS from E. coli O111:B4 (50 µl) in MEM in a 96-well plate at 37 °C for 1 h. E. coli cells (which gave an optical density reading of ∼0.7 at 650 nm) were diluted to 2.5 × 106 cells ml−1 in MEM and 100 µl of the bacteria solution was added to the S16-LPS mixture. The final concentration of S16 was kept at 4 µg ml–1, whereas the LPS concentration was varied from 2 to 1,000 µg ml–1. The 96-well plate was then incubated at 37 °C for 90 min. A 50 µl aliquot was taken from each well, transferred to a second 96-well plate, and 100 µl of saline and dye mixture (that is, saline with 0.1% SYTO 9 and 0.1% PI) was added. Each well in the second 96-well plate was analysed with a Cell Lab Quanta SC MPL flow cytometer (Beckman Coulter) to determine the percent of PI-positive cells. Two independent runs of the assay were conducted and two replicates were used in each run for each variation.
Kinetics of antimicrobial activity
SNAPP S16 (at a final concentration of 8 µg ml–1) was incubated with E. coli cells (at a final concentration of 1.25 × 106 cells ml−1) in MEM at 37 °C. Aliquots were taken at t = 0, 15, 30 and 90 min for analysis to determine the c.f.u. ml−1 (see procedure for the measurement of MBC) and percent of PI-positive cells (via flow cytometry using a Cell Lab Quanta SC MPL flow cytometer (Beckman Coulter)). An untreated control group was also included. Two independent assay experiments were conducted and two replicates were used in each experiment for each variation.
Preparation of large unilamellar vesicles for dye release (pore formation) and lucigenin (chloride ion transport) based assays
To represent a model of an E. coli cytoplasmic membrane, large unilamellar vesicles (LUVs) consisting of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol (POPG) at a 7:3 mole ratio38 were used to encapsulate 2 mM rhodamine dextran (70 kDa; RD-40) and 2 mM fluorescein dextran (4 kDa; FD-4) in 10 mM Tris and 5 mM NaCl buffer solution (pH 7.3) for the dye release assay or 2 mM lucigenin solution containing NaCl (100 mM) and sodium phosophate salt (10 mM, pH 7.3) for the chloride ion transport assay, using the LUV preparation method we have previously described50. The dye release assay was performed as previously described50 and the chloride ion transport assay conducted as described by Elie and co-authors39. LUVs were incubated (0.5 h for the dye release experiment or overnight for the chloride ion transport assay) with SNAPP S16 or control AMP maculatin 1.1 at lipid-to-peptide molar ratios ranging from 50:1 to 10,000:1. To afford complete dye release or chloride ion transport, control LUVs were treated with 0.5% vol/vol Triton X-100. All measurements were made with a Varian Cary Eclipse spectrophotometer using a 4 mm path-length quartz microfluorimeter cell (Starna) for the dye release experiment or a FLUOstar Optima plate reader (BMG Labtech) for the chloride ion transport assay. Dye release or chloride ion transport was presented as the percent of fluorescence of RD-70 and FD-4 or lucigenin, respectively, compared to the Triton X-100 control. Data are representative of two independent assays completed in duplicate.
Membrane potential assay
Membrane potential was determined by flow cytometry using a BacLight Bacterial Membrane Potential Kit (Invitrogen). When at low concentrations, the dye DiOC2(3) exhibits green fluorescence in all bacterial cells. The fluorescence shifts towards red emission as the dye molecules become more concentrated and self-associate in healthy cells that are maintaining a membrane potential. E. coli cells were inoculated to mid-log phase. Viable cells were then diluted to 2.5 × 106 cells ml−1 in PBS and added with different concentrations (0.5×, 1× and 2× MBC) of SNAPP S16. A fully depolarized control was provided by the addition to the untreated cells of the proton ionophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP) at a final concentration of 5 mM. Before a 1 h incubation at 37 °C, 30 mM DiOC2(3) was added to all samples. Membrane potential was determined by a Cell Lab Quanta SC MPL flow cytometer (Beckman Coulter) as the ratio of cells that exhibited red fluorescence (FL-3) to those that displayed green fluorescence (FL-1). Gates were drawn based on the untreated (polarized) and CCCP-treated (fully depolarized) controls. Data are representative of two independent assays completed in duplicate.
RNA extraction and reverse transcriptase polymerase chain reaction (RT–PCR) analysis for programmed cell death pathways
To determine if the star peptide polymers induced PCD, E. coli cells were incubated with S16 at 1× and 5× the MBC, after which the mRNA levels of recA, lexA and mazEF relative to control genes and untreated E. coli were determined by RT-PCR40,41,51. E. coli ATCC 25922 was grown overnight at 37 °C in LB broth and used to freshly inoculate LB broth (2% vol/vol inoculum), and was grown to mid-log phase (OD600 = 0.6) at 37 °C. A 200 µl aliquot of the cell suspension was mixed with 1 µl of Syto9 and 1 µl of PI and counted on the Cell Lab Quanta SC MPL flow cytometer (Beckman Coulter). After counting, cells were collected by centrifugation at 8,000g for 10 min at 4 °C, washed twice in MEM and finally resuspended at 2 × 106 cells ml−1 in MEM. A 500 μl aliquot of a stock solution of SNAPP S16 (final concentration of 1 × MBC or 5 × MBC) was added to 500 µl (final concentration of 1 × 106 cells ml−1) of E. coli cells. Following incubation (4 h), bacterial cells were collected by centrifugation at 8,000g for 10 min at 4 °C and immediately resuspended in 1 ml RNAprotect Bacteria Reagent (Qiagen).
Total RNA was extracted using the RNAprotect bacterial reagent and RNeasy kit (Qiagen). Cells were collected by centrifugation at 8,000g for 10 min at 4 °C and resuspended in 100 µl TE buffer (10 mM TrisCl, 1 mM EDTA, pH 8.0) containing 1 mg ml–1 lysozyme. After 5 min incubation at room temperature, 350 µl of buffer RLT was added and the solution mixed by vortexing. Ethanol (250 µl) was then added and the entire 700 µl was added onto an RNeasy spin column and centrifuged at 16,000g for 1 min. The membrane was washed with 700 µl of buffer RW1 followed by 500 µl of RPE buffer. A final centrifugation at 16,000g for 1 min was performed to dry the membrane. RNase free water (50 µl) was added to the membrane and the RNA was eluted by centrifugation at 16,000g for 1 min. RNA was quantified by absorbance (260 nm/280 nm) using a Nanodrop spectrophotometer (Thermo Scientific).
Extracted RNA was immediately DNase-treated using the TURBO DNA-free kit (Ambion). Briefly, 5 µg of RNA was combined with 2 µl of 10× DNase buffer, 1 µl of DNase and sufficient water to make up a 20 µl reaction volume. The reaction was incubated for 20 min at 37 °C, after which a further 1 µl of DNase was added and the reaction was incubated for another 20 min at 37 °C. Following this second incubation, 2 µl of DNase inactivation reagent was added. After 5 min incubation at room temperature with occasional mixing, the inactivation reagent was pelleted by centrifugation at 10,000g for 1 min and the supernatant collected.
Reverse transcription was performed using the iScript Reverse Transcription Supermix (Bio-Rad). Briefly, 1 µg (4 µl) of the DNase-treated RNA extract was combined with 4 µl of iScript master mix and 12 µl of RNase free water. The reverse transcription reaction was performed with a 5 min, 25 °C priming step, a 30 min, 42 °C extension step, and a 5 min, 85 °C inactivation step. A no-reverse transcription reaction was also set up using 1 µg (4 µl) of the DNase-treated RNA extraction combined with 16 µl of RNase free water.
RT–PCR was performed using iTaq Universal SYBR Green Supermix (Bio-Rad). Template cDNA (25 ng, 1 µl) was combined with 0.8 µl of forward primer (5 nM), 0.8 µl of reverse primer (5 nM) and 10 µl of SYBR Green Supermix. Primers used for the RT–PCR; recA (For) AGATCCTCTACGGCGAAGGT, (rev) CCTGCTTTCTCGATCAGCTT; lexA (For) GACTTGCTGGCAGTGCATAA, (rev) TCAGGCGCTTAACGGTAACT; MazEF-1 (For) CTTCGTTGCTCCTCTTGC, (rev) CGTTGGGGAAATTCACCG; 16SrRNA (For) TGTAGCGGTGAAATGCGTAGA, (rev) CACCTGAGCGTCAGTCTTCGT51. Thermal cycling was performed by 40 cycles of 95 °C for 15 s, 60 °C for 15 s and 72 °C for 15 s. A positive control of genomic DNA and a no-reverse transcription control were included in each cycling run. Cycling was performed on a Rotor Gene RG-3000A Thermal Cycler running Rotor-Gene V6.1 software (Corbett Research). Analysis of the PCR was performed using LinRegPCR software version 2015.3. Comparative cycle threshold (Ct) analysis was performed according to the method of Schmittgen and Livak52.
E. coli cells (which gave an OD reading of ∼0.7 at 650 nm) were diluted to 2.5 × 106 cells ml−1 in MEM and 100 µl of the bacteria solution was added to each well containing either MEM (untreated control) or the test compound(s) at the desired concentrations (100 µl). The 96-well plate was then incubated at 37 °C for 90 min. The cells were then stained with the CellROX Orange Reagent at a final concentration of 750 nM following the manufacturer's instructions and were incubated for 1 h at 37 °C. The cells were analysed on the Cell Lab Quanta SC MPL flow cytometer (Beckman Coulter) and the fluorescence from the CellROX Orange Reagent was measured on FL-3. A minimum of two independent experiments were conducted for the assay, and two technical replicates were used in each experiment. Data are expressed as mean ± s.d.
Effect of ALD inhibition on the membrane disruption ability of SNAPP S16
E. coli cells (which gave an optical density reading of ∼0.7 at 650 nm) were diluted to 2.5 × 106 cells ml−1 in MHB, and 5 ml of the bacterial cell solution was added to an equivolume of doxycycline hyclate (to yield a final concentration equivalent to its MIC of 0.5 µg ml–1). The mixture was incubated at 37 °C for 4 h and the cells were recovered via centrifugation (3,000g, 10 min) at the end of the incubation period. The recovered cells (100 µl per well) were then incubated at 37 °C for a further 90 min in the absence or presence of SNAPP (at 0.5× and 1 × MBC, 100 µl per well) in a 96-well plate. A 50 µl aliquot was taken from each well, transferred to a second 96-well plate, and 100 µl of saline and dye mixture (that is, saline with 0.1% of SYTO 9 and 0.1% of PI) was added. Each well in the second 96-well plate was analysed with a Cell Lab Quanta SC MPL flow cytometer (Beckman Coulter) to determine the percent of PI-positive cells. Two independent runs of the assay were conducted and two replicates were used in each run for each variation.
Star peptide polymer S16 (35 µg ml–1 in MHB; 15 and 35 µg ml–1 in MEM), melittin (64 µg ml–1 in MHB) or ovispirin (19 µg ml–1 in MHB) was incubated with E. coli cells (1.25 × 106 cells ml−1 in MHB or MEM, prepared as described in the measurement of MBC) at 37 °C for 90 min. After incubation, the cells were pelleted (10,000g, 20 min), washed with pre-filtered PBS and resuspended in the same buffer (∼10–30 µl). Subsequent steps were taken based on the protocol previously described53.
The data obtained were determined to be normally distributed. Homogeneity of variances was assessed using Levene's test (SPSS for Windows, version 12). Statistical analysis was also performed using a one-way classification of ANOVA and Student's t-test (two-tailed), where differences were regarded as statistically significant with probability P < 0.05.
G.G.Q. acknowledges financial support from the Australian Research Council under the Future Fellowship (FT110100411) scheme. E.C.R. acknowledges financial support from the Australian Government, Department of Industry, Innovation and Science. S.J.L. acknowledges the Australian Government for providing an International Postgraduate Research Scholarship (IPRS) and an Australian Postgraduate Award (APAInt). The authors thank the Advanced Fluorescence Imaging Platform at the Materials Characterisation and Fabrication Platform (The University of Melbourne) for instrument access. The authors thank B. Hibbs for assistance with the Delta Vision OMX V4 BLAZE, S. Lowe for technical laboratory assistance and J. Li for the CMDR bacterial strains.