After Friday’s post, I’ve held off on writing much about the German E. coli outbreak, often referred to by its serotype, O104:H4, or as HUSEC041 (HUS stands for hemolytic uremic syndrome). Having had the weekend to digest some of the ongoing analysis and news reports, here are some additional thoughts:
1) The multilocus sequence type (MLST) of this outbreak is definitely ST678. This means this outbreak strain is related to an older strain found in 2001 that caused disease. A new, improved assembly released by BGI yields a perfect match to ST678. In addition, there is independent confirmation of this (pdf).
2) Ion Torrent-based sequencing has some errors that will require filtering. Both Nick Loman’s original assembly as well as the assembly of a different isolate had the same base pair change in the adk gene. The BGI assembly, which uses quality filtering that takes into account the idiosyncracies of Ion Torrent, lacks this change; the new BGI assembly also lacks the additional A observed in recA in Nick Loman’s assembly. The moral of the story is if you care about individual nucleotide changes (i.e., looking for SNPs), you have to be careful with these data.
3) It appears that HUSEC041 is very similar to E. coli 55989. According to David Studholme, HUSEC041 shares 99.69% nucleotide sequence identity over 96.07% of 55989’s chromosome. Keep in mind, this is only a comparison of the chromosome.
4) At this point in time, the best evidence is that HUSEC041 is an EAEC (enteroaggregative) strain which has acquired a Shiga-toxin encoding phage similar to that found in E. coli O157:H7. This is based on an analysis by Kat at Bacpathogenomics and Nico Petty.
5) This is a highly drug resistant strain. Its resistance profile:
Amoxicillin/Clavulanic acid R
Nalixidic Acid R
It also has a CTX-M-15 beta-lactamase, which confers resistance to most penicillin derivatives, except for the carbapenems. In English, any antibiotic that starts with “Cef-“, “Ceph-“, or ends with “-cillin” won’t work. What’s disconcerting about this is, if the recent reports are correct and the outbreak was initiated by contaminated bean sprouts, agricultural isolates now have many different drug resistance genes. Multidrug resistance is no longer confined to hospitals and other medical settings. Note however, most accepted treatments for Shiga-toxin producing infections do not involve antibiotics, as they are thought to stress the bacterium, leading to increased toxin production and more severe disease; resistance is not a problem for therapy in this particular case, despite claims to the contrary.
6) We need better genomes to understand the role of plasmids in HUSEC041. This is not to slight any of the sequencing efforts, but the current datatypes (e.g., Ion Torrent fragment reads) simply will not yield information about plasmids (plasmids are ‘mini-chromosomes’ that can move from strain to strain). This is critical for understanding the role of plasmids in the acquisition of antibiotic resistance in HUSEC041, especially since other resistance plasmids can be maintained by selective pressures other than antibiotic exposure. In one study, a resistance plasmid found in cattle was advantageous even in the absence of antibiotics because the plasmid enabled E. coli strain to use vitamin D. In essence, what we call a ‘resistance’ plasmid (because that’s the trait we look for) could be an ‘eat your vitamins’ plasmid, or have some other function, such as protection against bacterial viruses (‘phage’). We need much better assemblies to understand this, so we can develop better strategies for preventing the spread of strains with these plasmids.
7) The public health communications response has been less than stellar. We seem to have forgotten everything we learned from the H1:N1 flu outbreak. I’ve received several requests from science journalists and writers who know biology and who were extremely frustrated by the contradictory and unclear information they were receiving. This isn’t helping, and, in fact, probably contributed to an atmosphere of panic.
8) More on the misuse of language: the hybrid edition One of the things that freaked people out was the use of the word hybrid to describe this strain. For microbiologists (or this one, anyway), that word does mean something–but it doesn’t appear to describe what we’re seeing here. In Staphylococcus aureus (the bacterium responsible for ‘staph’ infections, and when resistant to methicillin is called MRSA), we do observe hybrids: several clones appear to have swapped ten to twenty percent of their genome. Not bits and pieces, but an entire continuous segment that spans hundreds of kilobases (it’s not clear how this happens, but the DNA signature is crystal clear). That is a hybrid. But with HUSEC041, it simply appears to have picked up a phage or bacterial virus (‘Stx’) that enables it to produce Shiga toxin (this phage is also found in E. coli O157:H7 and other STEC E. coli). It also has acquired smaller genomic segments from other strains. But all E. coli do this–in that sense, every E. coli strain is a hybrid.
That’s enough for now, but here are some good posts on this topic:
My contribution to the ‘HUSEC41-strains-are-not-that-new’ debate
E. coli O104:H4 in Europe–is it new?
A whole genome phylogeny including HUSEC041
New clues found in tracing the origin of the deadly E coli strain and an appeal for the sharing of additional data
Comparisons of E. coli TY2482 against previously sequenced E. coli genomes
E. coli update: sprouts as the culprit?
EHEC Genome Assembly
E. coli TY2482: strain-specific genes
The reason why this deadly E coli makes doctors shudder