Advances in the understanding of AMP structure/function and the development of AMP-related bioinformatics tools means that it is now possible to identify putative AMPs in genomes with high confidence [17–19]. In silico approaches can be used to rationally identify a shortlist of candidate AMPs for targeted experimental validation, rather than having to perform expensive and time-consuming screening experiments. Here we have undertaken the first genome-wide characterisation of myxobacterial ribosomally-synthesised AMPs and highlight 37 particularly promising candidate AMPs.
AMPs are generally classified according to their source, structure, activities and amino acid residues [35]. Although mammalian, marine, amphibian, plant, and insect-derived AMPs have been well characterized in the literature, studies on AMPs from microorganisms are limited [35]. Bacteriocins are one group of AMPs obtained from bacteria that have been investigated recently, with probiotic organisms that produce bacteriocins being studied as alternative therapeutic agents [36]. This study analysed eight complete genomes of different myxobacterial species using multiple in silico approaches to screen for potential AMP sequences and predict the properties of putative AMPs. Myxobacteria are well known for their antimicrobial lifestyle and their production of bioactive secondary metabolites, which have been exploited for several decades as therapeutic candidates, with notable success [15, 37]. However, virtually nothing is known about the AMPs produced by myxobacteria, despite the current antibiotic resistance crisis, which demands novel approaches and alternative drug candidates to be investigated. The in-silico analyses described here revealed over 600 putative AMP sequences, of which 117 were predicted to have antimicrobial activity against multiple species of bacterial pathogens.
Secondary metabolites of myxobacteria are by and large active primarily against Gram-positive bacteria. For instance, S. cellulosum produces several secondary metabolites, including sorangicins, disorazoles, chivosazol, sorangiolids, sulfangolids, etnangien and thuggacins. These metabolites have been well characterized structurally and assays of their antimicrobial activities have shown them to have a preponderance of activity against S. aureus, yeasts, fungi and mycobacteria, with limited activity against Gram-negative bacteria [37]. In contrast, the candidate AMP sequences of S. cellulosum So ce56 were predicted to have good activity against K. pneumoniae (39% of AMPs), E. coli (21%) and P. aeruginosa (20%), while only 2% were predicted to be active against S. aureus. Studies of the predatory activities of live myxobacteria against pathogenic bacteria on non-nutrient agar have demonstrated that the predators kill Gram-negative bacteria like K. pneumoniae, E. coli and Proteus mirabilis more effectively than Gram-positive bacteria like S. aureus or Staphylococcus epidermidis [14]. Therefore, it appears that the mechanisms of secondary metabolite action contrast with those of AMPs, making them worth investigating experimentally. Our invitro studies highlight that the MICs weren’t efficacious compared to known antibiotics against the five organisms tested, although some AMPs (Stig_213, Coral_AMP411, and So_ce_56_913) had high MICs (250 mg) against E.coli and B.subtilis. A possible reason would be due to the use of a crude product (~ 50% purity) of the synthesized AMP which is a limitation of the study. Therefore, a purified product (> 95%) of the synthesized AMP would give a better perception of their potency.
AMPs from mammalian sources and insects are naturally anti-inflammatory in addition to their antimicrobial role [38, 39]. All of the 117 ‘potent’ putative AMPs, were predicted to be potentially anti-inflammatory, with more than 90% having a probability above 0.6 (Table S2). The APD3 database contains 193 antiviral peptides of which only 12 were from bacterial sources, the rest being from mammalian, amphibian, plant, arthropod and marine sources. None of the AMP sequences from our study were predicted to have antiviral properties. However, secondary metabolites from myxobacteria have been reported to have potent antiviral activities [40]. Therefore, whether AMPs from myxobacteria generally don’t have antiviral properties, or whether further investigations into other myxobacterial genomes are likely to identify AMPs with these properties, is currently unclear. All 117 ‘potent’ putative AMPs were predicted to have antifungal properties, with nine having particularly high prediction scores (> 0.9). This is perhaps to be expected, as fungal cells are abundant in natural environments such as soils, where they are likely to be preyed upon by myxobacteria. Similarly, many secondary metabolites from myxobacteria have antifungal properties, mainly by inhibiting electron flow in the mitochondrial respiratory chain [41].
Cationic amphipathic short peptides (20–50 residues) with high hydrophobicity and net charge are generally good candidates for therapeutic agents but a lot of them are cytotoxic and fail to be developed further [18]. Cytotoxicity of AMPs is measured by their haemolytic properties on human erythrocytes and killing of lymphocytes [42]. Only four putative AMP sequences from S. cellulosum and one from M. xanthus were predicted to be cytotoxic, while 26 of the 672 putative AMP sequences across the eight genomes were predicted to be haemolytic. Several AMPs, predominantly from mammalian sources, have been reported to have anticancer properties by acting on the negatively charged phosphatidylserine moieties of cancer cell membranes [43]. Anaero2CP1_256, Anaero2CP1_262, Anaero2CP1_264 (from A. dehalogenans) Haliangium_och1055 (from H. ochraceum), Stig_715 and Stig_926 (from S. aurantiaca) are proposed to have anticancer properties. Stig_715 in particular, a 75-residue peptide with a molecular weight of 6,822, had a high prediction score (0.972) and should be prioritized for investigating further.
The pathogenesis of many infectious diseases is related to biofilm formation, which allows pathogens to escape the effect of antibiotics and immune system mechanisms [44]. Therefore, there is a need for drugs that can disrupt biofilms and kill the sessile bacteria therein, rather than just killing planktonic cells. Several AMPs reported in the literature can prevent biofilm formation, downregulating the genes encoding the quorum sensing factors which stimulate biofilm formation, or degrading pre-formed biofilms [44]. 55 of the 117 ‘potent’ myxobacterial putative AMPs were predicted to have antibiofilm properties with eight of them having particular high prediction scores (> 0.9). Of the remaining 62, 47 were not predicted to have antibiofilm properties in their native form but 15 of them could be mutated at various amino acid residue positions resulting in positive predictions. Therefore, there are some interesting candidate AMPs potentially produced by myxobacteria, that seem likely to be able to kill not only planktonic bacteria but also those embedded in biofilms, and they should be prioritized for experimental validation. Our invitro antibiofilm assays showed some promising results despite the MICs not being very efficacious. Biofilm inhibition and degradation assays showed that Coral_AMP411 and Myxo_mac104 had good antibiofilm activity against E.coli, K.pneumoniae and B.subtilis compared to known antibiotics. Again, using a purified product of the synthesized AMP will reveal insights about the antibiofilm activities of the AMPs.
Myxobacteria utilize several mechanisms as part of their predatory activity, including secretion of secondary metabolites, digestion with hydrolytic enzymes, contact dependent killing etc., [45]. Studies have also shown that secreted proteins and contact dependent killing are selective towards particular target bacteria [46]. In our study AMPs were predicted to have particularly good activity against Gram-negative organism, an activity reported by Arend et al. to be associated with cell contact dependent killing [46]. This suggests that it would be beneficial to investigate the functional roles of AMPs. To elucidate whether AMPs interact with other predatory processes, the myxobacterial potent AMPs were searched against the NCBI database using BLAST-P, and STRING employed to suggest functional annotations and associations. Of the 14 AMPs that had > 85% similarity (100% query cover) with known proteins, nine of them were found within hypothetical proteins, which were functionally associated with several other hypothetical proteins, which is encouraging as the novelty of such proteins could potentially be exploited for innovative therapeutics.
Two ‘potent’ putative AMPs, Myxo_ful154 and So_ce_56_340, had sequence similarity with transposase enzymes, suggesting they could have been acquired via mobile genetic elements and lateral transfer. Conversely, Stig_715 exhibited sequence similarity to the 30S ribosomal protein S20, which would be expected to have been transmitted vertically. Presumably, contemporary sets of AMPs will have been acquired from both mobile genetic elements and by linear descent. Supporting this interpretation, pangenome analysis of C. coralloides genomes showed that 25% (12/48) of AMP sequences were part of the core genome, but the majority were found in the accessory genome. A recent pangenome study of Corallococcus spp. isolates gave similar results [47], finding that 30% of genes within each genome formed the core genome, with 70% of each genome belonging to the accessory genome. In contrast, the same study found that only 10% of BGCs belonged to the core Corallococcus genome [47], suggesting that AMPs may be more highly conserved than BGCs in myxobacterial genomes, an observation which merits further study.
To further shortlist those putative AMPs with the greatest likelihood of potential application in the clinic, the 117 ‘potent’ putative AMPs were checked against CAMPR3 predictions and filtered according to predicted toxicity, length, charge, and hydrophobicity (Table 2). We propose that these sequences are particularly worthy of being synthesised and tested for activity against pathogenic organisms using purified products of the AMPs.