Phages, pathogenicity and people

Like Comment
Read the paper

Our current paper may be found here:

What have we learned and why is it important?

A bacterium called Streptococcus pneumoniae, or the ‘pneumococcus’, is a leading cause of deaths from meningitis and pneumonia worldwide. Paradoxically, healthy children carry pneumococci in their nasopharynx soon after birth and throughout the early years of life.  We still do not fully understand what makes the pneumococcus a devastating pathogen in some people and yet causes no apparent harm to others.  

There are many examples of bacteria that are pathogenic because they harbour a bacteriophage or phage, a virus that infects bacteria.  The viral genes can be integrated within the bacterial genome, in which case the phage is called a ‘prophage’, and often those viral genes provide some benefit to the host bacteria. 

The first phage work in my research group happened fortuitously. Kelly Wyres, a doctoral student, was investigating mobile antimicrobial resistance determinants and discovered a prophage that contained the tetracycline resistance gene.  Prophage-encoded antimicrobial resistance genes had been described in other bacterial species but not pneumococci, so this was rather exciting1

It was known that pneumococci could harbour prophages, but there was little evidence on how often they did so, and limited in vitro data to demonstrate that pneumococcal prophages might have anything to do with pathogenicity. Therefore, the aim of our next phage study was to determine the prevalence and diversity of prophages among a dataset of around 500 pneumococcal genomes2.  This was a really challenging project – it transpired that there were loads of prophage sequences, the majority of which were novel and had to be carefully scrutinised. 

The analyses were performed by undergraduate students Caroline Harrold, Ben Edwards and Angus McDonnell, and doctoral student Andries van Tonder.  Early on we realised that existing phage screening tools frequently misidentified these pneumococcal prophages and so we developed our own tools and iteratively screened the dataset as we generated more and more new prophage data. We also selected a few pneumococci and tested these experimentally to see whether prophage genes could be expressed. Reza Rezaei Javan later joined the group as a doctoral student and analysed the RNA sequencing data we had generated in the laboratory.

Each one of the students made a significant contribution to that study and as such it is one of the papers I am most proud of having published2.  We revealed that prophage DNA was ubiquitous among pneumococcal genomes and identified nearly 300 different prophages, over 160 of which had not yet been reported.  We also demonstrated that prophage genes were expressed under experimental conditions, suggesting they might well have an important function among pneumococci (Figure 1).  Most importantly, there was indisputable evidence for an epidemiological association between specific prophages and pneumococcal genetic lineages.  This is important because it allows us to further investigate why that might be the case, and whether or not prophages contribute to the pathogenicity of a particular lineage.

There was more to do though.  I had spent much of one summer identifying 400 ‘partial prophages’ reported in our first paper.  These were viral sequences that could not encode fully functional prophages and yet often persisted for many decades among pneumococci.  Were they just prophage fragments left over from previous recombination or lysogenic events?  Why were they not purged from the genome? This was all rather curious and so I asked Reza to investigate these partial prophages further.

To cut a long story short, some of these partial prophages have the characteristics of satellite prophages, that is, they lack key viral genes and must rely on the bacterial host and another helper prophage for survival3. Intriguingly, Reza also discovered that a virulence-associated gene called vapE was found on a pneumococcal satellite prophage. This led to a collaboration with Prof Jerry Brown and Dr Elisa Ramos-Sevillano, who had developed mouse models of pneumococcal pneumonia and septicaemia.  Our two research groups, including Dr Asma Akter in my group, collaborated to test whether one of these vapE-containing pneumococcal satellite prophages was associated with virulence.  The in vivo work showed that if either vapE or the whole satellite prophage were knocked out, virulence was diminished. The mechanism by which this occurs remains to be determined, but these data provide evidence that these satellite prophages are probably quite important to the pneumococcus.

Finally, we expanded the prophage discovery component to include over 800 additional genomes of 69 non-pneumococcal Streptococcus species.  Reza developed a new bioinformatics tool to look for evidence of prophages and thus in total among all of the 1,300 genomes analysed in our two studies we identified over 400 full-length prophages and 350 satellite prophage genomes.  One surprising finding from the cross-species analyses was that sometimes the same prophages were found in unrelated streptococci, which suggests that prophages can move between different bacterial species and challenges the theory that prophages are bacterial species-specific3

Overall, this has been a fascinating programme of research.  Phage work in my research group began in 2012, but the first phages in pneumococci were reported over 40 years ago4.  We have really only scratched the surface of all there is to understand about streptococcal prophages and their impact on human and animal health. 

 What else have we learned?

  • Bright, enthusiastic and highly motivated students are a joy to work with. 
  • Manual curation is essential.  Detailed inspection of the sequence data by curious humans is indispensable.  Pipelines are only one tool to use in the process.
  • Hard work pays off.  Our main challenges in these studies were the novelty of the prophages and the volume of data to analyse.  Success demanded focused effort and ingenuity. 
  • Collaboration is vital.  Cooperation between the students to share their data and analyse the different prophage sequences was essential to progress.  There was a great deal of excitement as they discovered new prophages.  Jerry and Elisa applied their clinical and laboratory expertise to test our in silico findings in their in vivo model.  Elisa and Asma worked together to create the genetic mutants.

Figure 1. Evidence for prophage gene expression in one pneumococcal isolate: results of the RNA-seq experiment. Full-length prophages are depicted in (a) and (b), and (c) is a partial prophage sequence, which we later discovered is a satellite prophage. Rows indicate prophage genes and columns present differential expression levels at each of five time points of growth. Mitomycin C was added to the broth culture after 3 h of incubation to induce prophage gene expression. Statistically significant differential levels of expression (p<0.05) are marked with an asterisk.
Figure 2. Prophages identified among streptococcal genomes. An unrooted phylogenetic tree of all streptococcal prophage genomes identified in the dataset. Blue branches mark full-length prophages and red branches mark satellite prophages.

1. Wyres, KL et al. Evidence of antimicrobial resistance-conferring genetic elements among pneumococci isolated prior to 1974. BMC Genomics 14, 500 (2013).

2. Brueggemann, AB et al. Pneumococcal prophages are diverse, but not without structure or history. Sci Rep 7, 42976; doi: 10.1038/srep42976 (2017).

3. Rezaei Javan, R et al.  Prophages and satellite prophages are widespread in Streptococcus and may play a role in pneumococcal pathogenesis. Nat Commun 10, 4852 (2019).

4. McDonnell, M et al. ‘Diplophage’: a bacteriophage of Diplococcus pneumoniae. Virology 63, 577–582 (1975).

Angela Brueggemann

Professor of Infectious Disease Epidemiology, University of Oxford