Human gut microbiome: A source of new antibiotic peptides

Study: Mining human microbiomes reveal an untapped source of peptide antibiotics. Image Credit: nobeastsofierce / Shutterstock.com

In a recent study published in Cell, researchers performed a computational analysis of human gut meta-genomes to identify candidate molecules with antibiotic potential.

Discovering new sources of antibiotics

Antimicrobial resistance is a major public health issue that contributes to the emergence of drug-resistant microorganisms, thereby increasing the risk of nosocomial infections. Short peptide molecules, which have been proposed as novel antibacterial medications, consist of short-length amino acid residues, vast sequence space, and non-specific modes of action.

To date, few reported cases of resistance against antimicrobial peptides (AMPs) have been published, which has led researchers to become increasingly interested in their potential clinical applications. Several AMPs, the most common of which include bacitracin, colistin, and polymyxin B, have already been successfully approved for clinical use.

Despite the advantages and success of AMPs, the identification of novel AMP candidates remains a challenge due to the time-consuming processes involved in these experiments. Furthermore, although recent advances in machine learning, genetic algorithms, and pattern recognition algorithms have led to the development of novel peptides, few of these approaches have involved mining proteomes and metagenomes.  

The human microbiome consists of many bacteria species capable of suppressing the growth of pathogens. In fact, recent studies have reported that the human microbiome encodes hundreds of thousands of multiple small-size open reading frames (smORF), few of which have been characterized. Thus, there is tremendous potential to harness the human microbiome to identify unexplored peptide sequences with antimicrobial activity.

About the study

Using computational techniques, the researchers assessed the antibacterial capabilities of 444,054 peptides from 1,773 human metagenomes listed in the Human Microbiome Project (HMP). To refine the list, 78 peptides were identified and assessed for their antibiotic activity against several high-priority pathogenic and gut commensal bacteria in vitro, following which five peptides were selected for in vivo evaluation.

The smORFinder was used to identify genes that encode proteins with high confidence. Antimicrobial peptides were selected based on high AmPEP scores, representation of the peptide’s family of origin, effective amino acid composition for chemical synthesis, and absence of hydrophobic clusters.

The Database of Antimicrobial Activity and Structure of Peptides (DBAASP) was utilized to examine the physicochemical properties of smORF-encoded peptides (SEPs), including their mechanisms of action, secondary structures, and cytotoxicity.

The minimum inhibitory concentration (MIC) of selected SEPs was identified to evaluate their anti-infective activities in murine models of skin abscesses and deep-seated thigh infections using Acinetobacter baumannii.

Ribonucleic acid sequencing (RNAseq) was used to investigate prevotellin-2 homolog transcription. Contigs comprising 323 SEPs from the first list were analyzed using BLASTn against the National Center for Biotechnology Information (NCBI) RefSeq nucleotide database.

MetaProdigal predicted 531,822 ORFs in hCom2 genomes, after which researchers used the database to explore molecular alignments. The antimicrobial effectiveness of the SEPs against 11 pathogenic strains, including ESKAPEE pathogens and 13 of the most prevalent members of the human gut microbiota, was also determined to investigate whether these peptides would target gut commensal microorganisms.

The secondary structures of active SEPs were analyzed using ColabFold and circular dichroism to determine whether these peptides could collectively target bacteria. Checkerboard experiments on SEP pairs generated from similar sites were also performed to determine molecular interactions’ fractional inhibitory concentration index (FICI). The ability of SEPs to permeabilize the outer membrane of gram-negative bacteria utilizing 1-(N-phenylamino)naphthalene (NPN) tests and 3,30-dipropylthiadicarbocyanine iodide (DiSC3-5) fluorophore was also assessed.

Study findings

A total of 323 candidate peptide-based antimicrobials encoded in smORFs exhibited significant antibacterial activity against therapeutically relevant pathogens both in vitro and in vivo. In vitro, 71% of the 78 synthesized smORF-encoded peptides demonstrated antibacterial activity. Prevotellin-2, a lead hit from Prevotella copri, exhibited antibacterial activity equivalent to polymyxin B in vivo.

Five SEPs were highly potent against A. baumannii infection in both mouse deep thigh infection and skin abscess models, which included prevotellin-2, fecalibacticin-3, staphylococcin-2, fusobacticin-2, and keratinobacin-2. SEPs, unlike AMPs and EPs, do not permeabilize the outer membranes of bacteria. Rather, these peptides depolarize the bacterial cytoplasmic membrane, a process distinct from conventional AMPs and EPs.

The 323 methionine-rich SEPs had fewer hydrophobic sequences near the periphery or in sparsely populated places, thus distinguishing them from peptides such as AMPs and EPs. SEPs tend to have a disordered structure, which does not impair their antibacterial action.

Conclusions

The current study identified 323 peptides from the human microbiome, 71% of which exhibited in vitro antibacterial activity. Prevotellin-2, which emerged as a lead candidate from subsequent in vitro and in vivo experiments, exhibited antibacterial activity comparable to that of polymyxin B without causing any significant toxicity.

Our pipeline presents a platform for the discovery of peptide antibiotics, including those that might spare commensals.”