Enterococcus faecium is a member of the gastrointestinal (GI) microbiome; however, hospitalized patients can be colonized with vancomycin-resistant Enterococcus faecium (VREfm) mainly through the fecal-oral route. Due to high levels of antibiotic exposure, VREfm can expand within the GI tract and ultimately cause invasive infections. Colonized patients can also shed VREfm into the hospital environment, resulting in widespread transmission within healthcare systems (see image).  VREfm poses a significant challenge for healthcare systems, prompting researchers to investigate how VREfm populations change over time.

At the University of Pittsburgh, we have the unique ability to investigate VREfm evolution in the healthcare setting through the Enhanced Detection System for Healthcare-Associated Transmission (EDS-HAT) surveillance system implemented at our center¹. EDS-HAT works by combining pathogen whole genome sequencing data and patient electronic health data to identify instances of hospital transmission events. This system also laid the foundation of our study by allowing us to go back in time to study the genomes of patient isolates and explore how common healthcare-associated pathogens, such as VREfm, have evolved over time within a single health system. Through the efforts by EDS-HAT, we had the ability to investigate and characterize a large and systematic collection of 710 VREfm isolates spanning 6 years.

We first investigated the population dynamics of the EDS-HAT collection, and observed that VREfm lineages changed over the study period. At the beginning of the collection, the VREfm population at our center was relatively diverse, with a variety of different lineages causing infections. However, by the end of the study, two emergent lineages dominated the collection. To hypothesize possible causes of such a drastic shift, we thought of common factors for lineage replacement/persistence in healthcare systems, such as antibiotic resistance and virulence factors. Although we did observe some modest enrichment of antibiotic resistance and putative virulence factors in emergent lineages, we reasoned that these factors alone probably would not cause the major population shifts we were observing. After performing additional comparative genomics analyses and coming up empty-handed, we decided to step away from the computer and head to the lab bench.

We started our functional investigations by asking a simple question – can emergent lineages inhibit the growth of historical lineages? To test this, we selected representative emergent and historical isolates and performed some spot killing assays. We observed that emergent isolates showed large zones of inhibition when spotted onto bacterial lawns of historical isolates. Next, we tried to identify what was causing this growth inhibition. We considered growth rate differences, nutrient depletion, and prophage expression, but our investigations into each of these possibilities came up short. So, we turned to the literature. Enterococcus are carriers of bacteriocins, which are antimicrobial peptides known to cause growth inhibition. When we screened our entire collection of VREfm isolate genomes for bacteriocin genes, we found a bacteriocin called T8 that was differential present in our collection and was enriched in emergent lineages. We then tested the inhibitory activity of isolates selected based on their bacteriocin content, and found that bacteriocin T8 was strongly associated with growth inhibition. This gave us a concrete hypothesis to move forward with.

To confirm bacteriocin T8 was the cause of growth inhibition, we transformed a plasmid encoding bacteriocin T8 and the corresponding immunity gene (strain pBAC) or an empty plasmid (strain pEV) into a plasmid-cured clinical E. faecium isolate. We selected this plasmid-free recipient strain as, in our experience, the high plasmid carriage in clinical VREfm isolates can cause transformation to be very difficult, if not impossible. Using our two experimental strains, one with the bacteriocin and one with the empty vector, we aimed to characterize the competitive advantage conferred by bacteriocin T8 in liquid culture. So, we mixed the two strains at different starting ratios, and observed that in all cases the bacteriocin T8-expressing strain rapidly outcompeted the empty vector strain. This result generated a new question: if bacteriocin T8 confers a competitive advantage in a test tube, does it also facilitate competition in the mammalian gut?

In the initial submission of our manuscript, we showed that mice infected with the pBAC strain had a longer and higher duration of colonization compared to mice colonized with the pEV strain. However, one of the reviewers helpfully pointed out that our findings would be more impactful if we investigated competition using native, bacteriocin-containing clinical isolates.  We therefore performed a second experiment where we co-infected mice with a 10:90 mixture of a bacteriocin T8-positive, emergent strain and a bacteriocin T8-negative, historical strain. Just as with the pBAC and pEV strains, we observed that the bacteriocin T8-encoding strain dramatically outcompeted the strain lacking the bacteriocin, further supporting our hypothesis that bacteriocin T8 facilitates VREfm lineage replacement.

Our results thus far suggest that bacteriocin T8 enhanced the ability of VREfm to compete within the gut microbiome, contributing to the population shift we observed within our healthcare center. However, we were eager to know if this phenomenon was  unique to our center or was part of a global phenomenon. To investigate this, we aimed to gather the largest, most representative collection of likely healthcare-associated VREfm that we could. So, we downloaded over 15,000 publicly available VREfm genomes that are available through the National Center for Biotechnology Information (NCBI). Similar to our local collection, in the global collection we found that the same lineages emerged and replaced historical lineages over a 20-year period. Further, we found bacteriocin T8 was enriched in global emergent lineages and increased in frequency over time. This analysis would not have been possible without access to a large genomic collection of VREfm genomes available through NCBI. Further, these results highlight the importance of sharing genomic data to enable larger studies that characterize high-risk pathogens on a global scale. We also noted that our global analysis was almost certainly biased towards countries with resources and infrastructure to enable genomic surveillance. This prompts the need to support the development and expansion of similar systems in under resourced settings.

In summary, we characterized both local and global VREfm population dynamics and in doing so, identified bacteriocin T8 production as a potential driver of lineage emergence in this pathogen. Our current working model suggests that the antibiotic-perturbed GI tracts of hospitalized patients might be colonized with multiple strains of VREfm. Those strains containing bacteriocin T8 can efficiently outcompete susceptible strains, resulting in further gastrointestinal tract expansion, invasive infections, and shedding and transmission within the hospital environment (see image). While bacteriocin T8 appears to play a key role in lineage replacement and GI colonization, further investigation is still be needed to uncover additional factors shaping VREfm evolution and to develop new therapeutic approaches for this difficult to treat pathogen.

The poster image was created with BioRender.

References

1. Sundermann, A. J. et al. Whole-Genome Sequencing Surveillance and Machine Learning of the Electronic Health Record for Enhanced Healthcare Outbreak Detection. Clin Infect Dis 75, 476–482 (2022).