Distinct signatures of diversifying selection revealed by genome analysis of respiratory tract and invasive bacterial populations (Shea et al, PNAS 2011)

Diversifying selection is a form of natural selection where intermediate values of a trait become less represented within a population, in favour of extreme values; a process that may subdivide a population between specialized niches and eventually lead to speciation. For instance, it can be theorized that a pathogen colonising several sites of the human body, where it is exposed to wildly different conditions and selective pressures, would have greater chances of survival by expressing a multitude of site-appropriate phenotypes than by reaching an adaptive compromise. While this strategy could be achieved through phenotypic plasticity, it could also result from genetically distinct strains of the pathogen.

Streptococcus pyogenes, also known as the group A streptococcus, or GAS, is a Gram-positive human bacterial pathogen. It is responsible for diseases such as impetigo, a localized skin infection, or pharyngitis, the streptococcal “sore throat”, both of which are mild superficial infections. The same bacterium is involved in a wide range of “invasive” infections, i.e. infections of sterile sites such as blood, which can be severe. On an experimental standpoint, S. pyogenes is a useful model for studying bacterial clonal evolution, because its strains exhibit relatively limited amounts of horizontal transfer across portions of the core genome. This is in contrast to bacterial species that frequently exchange genetic material, thus complicating phylogenetic inference.

The authors of this paper compare S. pyogenes strains found in superficial infections, more precisely in pharyngitis cases, to strains found in invasive infections. The authors enunciate several objectives:

• First, they want to extend our limited knowledge about the genomes of pharyngitis strains. Greater efforts have so far been expended to dissect the molecular basis of the more health threatening invasive infections.
• Secondly, they point out that very little is known about the precise genetic relationship between those two categories, and present their work as the first full genome analysis performed to address this issue. This analysis has been made possible thanks to high-thoughput DNA sequencing technologies.
• In particular, they want to test the widely accepted model, supported by epidemiologic studies, that most strains causing invasive infections arise from pharyngeal or other benign infections. In other words, do pharyngitis and invasive strains belong to the same genetic pool, provided they were collected from the same geographical location?
• Finally, they try to make sense of the genetic differences between pharyngitis and invasive strains in the light of diversifying selection. Can a link be made between the genomic sequences and the selective forces expected from the host oropharynx or sterile-site environments?

On the origin of data

During the tutorial session, we discussed the notions of convenience sampling and reusing material from previous studies. The work presented in this paper is based on eighty-six serotype M3 GAS pharyngitis strains collected from six regional laboratories across Ontario from 2002 to 2010, as well as on two hundred fifteen serotype M3 GAS invasive strains collected from the same location as part of a prospective population-based surveillance study of invasive GAS infections from 1992 to 2009. Those invasive infections include unequal numbers of soft tissue infections, bacteremias, lower respiratory infections, unknown invasive infections, septic arthritides, necrotizing fasciitis, meningitides, toxic shock syndrome cases, peritonitis and other unspecified invasive infections. We were unsure as to whether the different and long time periods involved, or the high number and diversity of invasive infections when compared to pharyngitis, should be seen as strengths or weaknesses for the pertinence of the paper. Some of our concerns on the matter resurfaced while we were discussing figure 4, as will be explained later.

The DNA sequence data obtained from those strains was mapped to the genome sequence of the M3 reference strain MGAS315 (NC_004070). A different but related experiment, also described in this paper, involved strains obtained from experimentally inoculated nonhuman primates [1].

Go figure

As with previous sessions of this tutorial, we organised our discussion on a figure-by-figure basis. We found most of the figures in this paper to be in a large part confusing and / or unconvincing:

• Figure 1 shows the distribution of Chi2 statistics for unique polymorphisms per gene. The corresponding Bonferroni-adjusted P values, to correct for multiple testing, are written next to the dots on the plot. The meaning of the x-axis is not indicated, making the figure difficult to understand.
• Figure 3 shows two unrooted neighbour-joining phylogenetic trees assembled from the complete list of all core biallelic SNPs. One corresponds to the eighty-six pharyngitis strains and the other to one hundred temporally matched invasive strains. The two trees were assembled completely independently from each other. The authors claim that their remarkably similar overall structure suggests common evolutionary histories.  First, we discussed whether or not it would have been possible to root the trees, and concluded that it probably would have been very difficult. We also had our doubts about the focus on SNPs that seems to appear through this paper. But most importantly, we didn’t see any striking similarity between the shapes of those two trees.
• Figure 4, a combined phylogenetic tree for pharyngitis and invasive strains, was much more convincing in regard to the last issue. Pharyngitis strains are not massed on one branch of the tree, nor are invasive strains. An invasive strain will often be closer to a pharyngitis strain than to another invasive strain, supporting the idea that the two kinds of strains belong to the same genetic pool. Still, we also had problems with figure 4. The meaning of “SC”, as in SC1 to SC10, is unclear. Assuming those represent ten different strain collections, it would further suggest that strains from the same geographical area are more closely related. But the figure should more explicitly indicate which strains belong to which collection and give us more information about those collections. Otherwise, we are left wondering, for example, why SC3, SC4 and SC7 are so close from each other. Another issue, resurfacing from earlier in our discussion, is that the tree was assembled from a lot more invasive strains than pharyngitis strains. In particular, there is no pharyngitis strain in the “SC3 & SC4” region of the tree. A reader could believe that pharyngitis strains arise from invasive strains rather than the other way around. Finally, the emm3.53 pharyngitis strains, described in the paper as recently emerged subclone lineages, are presented in a very crowded part of the figure, and there would have been space for a zooming lens.

We spent a lot of time on Figure 2, because it is very detailed and made of five smaller figures. The first three figures are a schematic of polymorphisms within the has operon promoter as well as the hasA, hasB and covS genes, with a distinction between polymorphisms found in pharyngitis strains and those found in invasive strains. Those genes and several others were identified by the authors as having a significant excess of allelic variation, i.e. greater than expected by chance alone, although we think they should have better defined what they meant by an excess of polymorphism. Changes observed from the reference genome were predicted to either jeopardize or upregulate the expression of hasA and hasB. As the paper explains, those two genes encode proteins essential for the synthesis of the antiphagocytic hyaluronic capsule [2]. The rest of Figure 2 shows how the authors tested the effect of several of these polymorphisms on hasA transcript levels, hyaluronic acid production and colony morphology. It would appear that pharyngitis strains lose their antiphagocytic capsule, believably because they don’t need it in the host oropharynx, because it is expensive to produce and because it reduces exchanges between the cell and its environment. Invasive strains, on the other hand, need an increased resistance to the human immune system [3]. Although this is a very clear figure that supports the authors’ hypotheses, some of us still had doubt regarding whether or not S. pyogenes really was an example of diversifying selection.

Conclusions

The authors make several conclusions at the end of their paper:

• It is the largest whole-genome comparative analysis of a bacterial pathogen to date.
• It is a genome wide investigation of S. pyogenes strains involved in upper respiratory tract infection.
• Invasive strains are genetically more similar to a given population of pharyngis strains than to invasive strains as a whole, confirming previous morphological observations.
• They didn’t identify a single highly prevalent genetic variant explaining the various diseases. Instead, an accumulation of rare variants would be involved, altering functions such as the CovR/S global gene regulatory system [4].

We also had questions regarding research funding, the ecology of S. pyogenes or the reason they excluded prophage sequences from their main study (although there is a paragraph on prophage content). We noticed that this paper was a direct submission with a prearranged editor, a fact that might explain how we were able to come up with so many questions and criticisms in just an hour long tutorial session.

In my opinion, this paper was nonetheless a very interesting read. It shows the possibilities of new sequencing technologies, as well as the kind of thinking that must be done in order to understand diseases and epidemics. It made for a lively tutorial session.

1. Virtaneva K, et al. (2005) Longitudinal analysis of the group A Streptococcus transcriptome in experimental pharyngitis in cynomolgus macaques. Proc Natl Acad Sci USA 102:9014–9019.
2. Dougherty BA, van de Rijn I (1994) Molecular characterization of hasA from an operon required for hyaluronic acid synthesis in group A streptococci. J Biol Chem 269:169–175.
3. Stollerman GH, Dale JB (2008) The importance of the group a streptococcus capsule in the pathogenesis of human infections: A historical perspective. Clin Infect Dis 46: 1038–1045.
4. Federle MJ, McIver KS, Scott JR (1999) A response regulator that represses transcription of several virulence operons in the group A streptococcus. J Bacteriol 181:3649–3657.

Shea, P., Beres, S., Flores, A., Ewbank, A., Gonzalez-Lugo, J., Martagon-Rosado, A., Martinez-Gutierrez, J., Rehman, H., Serrano-Gonzalez, M., Fittipaldi, N., Ayers, S., Webb, P., Willey, B., Low, D., & Musser, J. (2011). Distinct signatures of diversifying selection revealed by genome analysis of respiratory tract and invasive bacterial populations Proceedings of the National Academy of Sciences, 108 (12), 5039-5044 DOI: 10.1073/pnas.1016282108