E. coli and Salmonella: names, pathotypes and genomes
Lee Katz, Andrew Page and Nabil-Fareed Alikhan discuss why E. coli and Salmonella have overlapping, sometimes awkward naming systems. They compare clinical descriptions, genomic markers, evolutionary relationships and serotyping, explaining why these do not always identify the same groups. The discussion connects everyday terminology with practical questions about interpreting genomes and replacing laboratory tests with bioinformatics.
Listen to Episode 60: Whats in a name: Salmonella and E. Coli on SoundCloud
In this episode
What makes a microbe enteric?
Andrew opens by asking what an enteric organism actually is. Lee explains the term through its relationship to the gut, distinguishing bacteria that belong there from organisms that cause disease after reaching it. An enteric pathogen does not have to be a normal resident of the gut.
Questions about Clostridioides and SARS-CoV-2 test the boundaries of that definition. The hosts distinguish an organism being swallowed, surviving in the gut, causing disease there and being shed in stool. Their central point is that simply eating something, or detecting it in stool, does not by itself make it an enteric pathogen.
E. coli pathotypes describe different things
The E. coli discussion starts with the confusing collection of labels: ETEC, EPEC, STEC and UPEC. Nabil describes a history in which clinical presentations were recognised before their relationships to the organism were understood. Consequently, names may refer to disease, the site of infection or a particular virulence feature rather than a single evolutionary lineage. Shigella illustrates the mismatch: it was classified around dysentery but is discussed here as a subset of E. coli.
Some labels have useful genomic signals. For EPEC, the discussion centres on attaching and effacing lesions, type III secretion and the locus of enterocyte effacement. For Shiga-toxigenic E. coli, Shiga toxin is the defining feature discussed, with phages allowing its movement between lineages. Uropathogenic E. coli is less straightforward: a gut strain can cause a urinary tract infection without having the usual genomic hallmarks. Iron-acquisition systems and other genomic islands offer clues, not a universal classification.
Recombination, mixed pathotypes and MLST
E. coli’s large pangenome leads to a discussion of genetic exchange. Nabil contrasts recently expanded, relatively clonal pathogens such as ST131 with commensal E. coli that shuffle genetic material extensively. Andrew asks whether those commensals provide a reserve of potentially dangerous traits; Nabil treats that as an open question rather than a settled explanation.
The European outbreak discussed as occurring roughly ten years earlier illustrates the consequences of mixing traits: an enteroaggregative E. coli acquired Shiga toxin and combined features associated with different pathotypes. Seven-gene MLST does not generally resolve this complexity. Sequence types can be informative for particular established clones, but they are not reliable general predictors of pathology. Host differences add another layer: similar organisms can be associated with very different outcomes in different people.
Salmonella diversity and sampling bias
Andrew asks why Salmonella serovars appear more stable than E. coli pathotypes. Nabil points to more than 2,500 Salmonella serovars and distinguishes the familiar clinical groups from less-studied environmental diversity. The roughly 20 serovars most often encountered clinically can look like sharply separated, stable groups; isolates from rivers, reptiles and other settings can show a much messier picture.
Sampling matters. Nabil estimates that Typhimurium, Enteritidis and Typhi together account for perhaps two-thirds of the Salmonella genomes available at the time. The hosts also discuss historical lineages that may no longer be available to sample, making it difficult to reconstruct how today’s groups arose. Lee points to research from the people behind ANI on the continuity of species, in which sequencing ever-closer relatives raises the question of where species boundaries lie. An ANI primer accompanies the episode’s show notes.
Serotyping tools and the rules for names
Replacing laboratory phenotyping requires genomic methods that recover the information people already use. The hosts name ECTyper, SeqSero and SISTR while discussing in silico serotyping, and describe public-health publications supporting genome-based alternatives to laboratory serotyping.
They then examine why some Salmonella isolates have familiar names while others are described by long antigenic formulae. Outside subspecies enterica, they explain, shared antigenic formulae can group organisms with little else in common. The naming convention discussed therefore retains the formula rather than assigning new serovar names outside that subspecies.
Names also preserve different historical habits: some describe disease, while others refer to places where isolates were found. A place name does not establish geographical origin. Formatting is another source of frustration: in Salmonella enterica subspecies enterica serovar Typhimurium, the species and subspecies names are italicised, but the rank words and serovar name are not. The hosts mention returning to CDC nomenclature guidance when writing.
Stamp collecting and duplicated software
Nabil’s tweet about types of microbial bioinformatics papers provides a lighter interlude. Adapted from an xkcd comic, it jokes about sequencing collections and running BEAST for 92 days before choosing a relaxed-clock model because the tree looked attractive. Nabil stresses that much of the humour targets his own work.
The hosts also admit to duplicating existing functionality, including bedtools and MLST callers, and joke about the proliferation of SARS-CoV-2 pipelines. Andrew’s proposal to long-read sequence every available Salmonella and E. coli becomes premium stamp collecting: a humorous ending to a discussion that repeatedly returns to the value and limitations of more data.
Highlights
- 00:01:04 — Andrew asks what enteric means and why the distinction matters.
- 00:03:54 — Why E. coli has so many labels, including ETEC, EPEC, STEC and UPEC.
- 00:07:50 — Genomic clues to EPEC and other pathotypes, starting with attaching and effacing lesions.
- 00:09:46 — Clonal pathogens versus recombining commensal E. coli.
- 00:12:09 — Can seven-gene MLST predict E. coli pathotypes?
- 00:13:29 — Comparing the apparent stability of Salmonella serovars with E. coli diversity.
- 00:16:51 — Whether pathogen-focused sampling misses large parts of Salmonella diversity.
- 00:18:37 — Nabil explains his joke about types of microbial bioinformatics papers.
- 00:21:08 — Evidence for recovering serotyping information from genomes rather than laboratory tests.
- 00:22:10 — Why Salmonella outside subspecies enterica is described using antigenic formulae.
- 00:24:56 — How Salmonella naming moved between disease descriptions and place names.
- 00:26:45 — Why Salmonella serovar names should not be italicised.
In their own words
It’s not stamp collecting; it’s like high quality, the premium stamp collecting.
— Andrew Page, 00:20:36
Who is talking
- Lee Katz (host)
- Andrew Page (host)
- Nabil-Fareed Alikhan (host)
Tools and resources mentioned
ANI, seven-gene MLST, ECTyper, SeqSero, SISTR, BEAST, bedtools.
Questions this episode answers
Can an E. coli genome reveal its pathotype?
Some features provide strong clues, such as Shiga toxin for STEC or the locus of enterocyte effacement in the EPEC discussion. The hosts caution that clinical labels, genomic features and evolutionary lineages do not always coincide, particularly for urinary tract infections and disease severity.
Does seven-gene MLST predict E. coli pathogenicity?
Not reliably in general, according to the discussion. It can be informative for particular established clones, but mobile genetic elements and differences between hosts prevent sequence type from serving as a universal predictor of pathology.
Why do common Salmonella serovars look genetically stable?
The hosts describe familiar clinical serovars as relatively well-separated groups, while less-studied environmental isolates can look more variable. They discuss both biased sampling towards common pathogens and missing historical lineages as limitations on interpreting this pattern.
Which tools are mentioned for genome-based serotyping?
The hosts name ECTyper, SeqSero and SISTR when discussing replacing laboratory serotyping with genomic methods. They emphasise the need to reproduce information that laboratories and public-health users already rely on.
Should Salmonella serovar names be italicised?
No. The episode distinguishes italicised species and subspecies names from non-italicised serovar names, using the full Salmonella enterica subspecies enterica serovar Typhimurium name to illustrate the formatting problem.
This page was written from a machine transcript of the episode and checked against it. Listen to the episode for the whole conversation.
Show notes
Today we’re talking about getting your head around our favourite enteric microbes. E. coli and Salmonella. Why do they have some of the names they have? Primer on ANI: https://www.pnas.org/content/102/7/2567.short Wikipedia on Salmonella serovars: https://en.wikipedia.org/wiki/Kauffman%E2%80%93White_classification Nabil’s viral tweet: https://twitter.com/happy_khan/status/1387804862830809091?s=20