The continuing evolution of a bacterial pathogen

Although Escherichia coli is a prototypic commensal bacterial species of the mammalian intestine and a laboratory workhorse for molecular biology, certain strains of this species are capable of causing significant human disease. The spectrum of disease caused by E. coli includes enteric/diarrheal disease, urinary tract infections, renal failure, and sepsis/meningitis (1). The different pathotypes of E. coli possess genes encoding a wide variety of virulence factors that are frequently encoded on mobile genetic elements such as bacteriophages, plasmids, and pathogenicity islands. E. coli strains can be serotyped as one of >10,000 possible combinations of O (LPS) and H (flagellar) antigens, but one serotype, O157:H7, has achieved particular notoriety as a cause of deadly outbreaks of food-borne illness throughout the world. The work of Manning et al. (2) in this issue of PNAS reveals significant new information on the evolution and genetic composition of E. coli strains belonging to the O157:H7 serotype and reports the emergence of a new variant that appears to cause more severe disease.

First recognized a quarter of a century ago, E. coli O157:H7 causes bloody and nonbloody diarrhea, hemorrhagic colitis, and hemolytic uremic syndrome (HUS), which is the leading cause of acute renal failure in children. One of the most potent toxins ever described, Shiga toxin, is a critical virulence factor in HUS, and different variants of this bacteriophage-encoded toxin, e.g., Stx1, Stx2, Stx2c, are found in this pathogen. The mechanism of Shiga toxin is similar to that of ricin and involves inhibition of protein synthesis in renal endothelial and other cells. Toxin produced in the intestine enters the circulation, resulting in direct and indirect effects on the kidney. Other important virulence factors include the type III secretion system encoded on the locus of enterocyte effacement (LEE) pathogenicity island and a variety of secreted effector proteins encoded in the LEE and elsewhere in the genome. Previous work from the Whittam laboratory (3) indicates that E. coli O157:H7 evolved from enteropathogenic E. coli serotype O55:H7, a cause of nonbloody diarrhea, in a stepwise manner through the sequential acquisition of phage-encoded Stx2, a large virulence plasmid, an rfb locus that converted the somatic O antigen from O55 to O157, and additional chromosomal mutations. Comparative genomic studies of several E. coli O157:H7 strains have revealed extensive genomic diversity related to the structure, position, and genetic content of bacteriophages and variability in the content of putative virulence genes, including those encoding adhesins, Shiga toxins, and non-LEE effector proteins. These developments have led Wick et al. (3) to conclude that “O157:H7 genomes are rapidly diverging and radiating into new niches as the pathogen disseminates.”

Widespread Outbreak Associated with Spinach

Although the first outbreaks of E. coli O157:H7 disease were associated with insufficiently cooked hamburgers, many subsequent outbreaks have been associated with the consumption of raw vegetables, including the large 1996 outbreak in Japan in which contaminated radish sprouts infected ≈8,000 persons. In 2006, a widespread outbreak affecting 204 persons in 26 states was linked to contaminated spinach. In this outbreak, 52% of cases were hospitalized, 16% had HUS, and one person died. The higher-than-typical rates of hospitalization and HUS fueled speculation that the implicated strain of E. coli O157:H7 had greater virulence potential than “typical” strains of the serotype. This outbreak forms the basis of the hypothesis of the present study by Manning et al. (2) who investigated genomic diversity of E. coli O157:H7 strains by SNP analysis in an effort to better appreciate the evolutionary dynamics of E. coli O157:H7 at a population level and to place the spinach outbreak strain within an evolutionary framework.

No specific therapeutic regimens have been developed for E. coli O157:H7.

Using comparative genomic sequencing microarrays previously developed by Zhang et al. (4) for E. coli O157:H7, Manning et al. (2) investigated SNPs in 96 loci of >500 strains dating back to the first human outbreak in 1982. They identified 39 SNP genotypes that differ at 20% of SNP loci and separated them into nine clades. The spinach outbreak strain corresponded to clade 8 as did another 2006 isolate from a lettuce-borne outbreak involving 71 cases. Manning et al. observed differences between clades in the frequency and distribution of Shiga toxin genes and in the type of clinical disease reported. In particular, they found that patients with HUS were significantly more likely to be infected with clade 8 strains. The 2006 spinach and lettuce outbreaks caused by clade 8 strains were notable for an average HUS rate of 13%, which is three times greater than the average HUS rate for 350 U.S. outbreaks between 1982 and 2002. Characterization of isolates from all 444 patients infected with E. coli O157 in Michigan since 2001 shows that the frequency of clade 8 strains has increased from 10% in 2002 to 46% in 2006 despite the steady decrease in total O157 cases identified via surveillance in this time period. Although the number of HUS cases in this collection was small (n = 11), HUS patients were seven times more likely to be infected with clade 8 strains than patients with strains from all other clades combined. Manning et al. conclude that “an emergent subpopulation of the clade 8 lineage has acquired critical factors that contribute to more severe disease.” They further assert that genomic content variability among the different clades likely explains differences in clinical and epidemiological characteristics.

Recent Emergence of a Clade

Their study provides new insights into the evolution of a bacterial pathogen that was rarely isolated before 1982. The strains characterized by Manning et al. (2) span the entire outbreak history of E. coli O157:H7, and the results of the analysis provide a roadmap for characterizing future isolates. The data showing the recent emergence of a clade that is associated with more severe disease is a particularly worrisome development because antibiotic usage is contraindicated for this pathogen. No specific therapeutic regimens have been developed for E. coli O157:H7, although supportive care guidelines have been improved so that mortality rates are low (5). Manning et al.'s conclusion that this emergent subpopulation of a more virulent clade has acquired critical virulence factors that contribute to more severe disease is certainly consistent with the data, although the present study does not identify what the critical virulence factor(s) may be. Clade 8 strains were more likely to have both the stx2 and stx2c genes when compared with other stx2c-positive clades, but Manning et al. conclude that the toxin–gene combination alone does not account for the variation in hospitalization and HUS rates by clade. Even the determination of the complete genome sequence of the clade 8 spinach isolate did not reveal an obvious explanation for increased virulence, although a number of differences were noted between the clade 8 genome sequence and the previously determined genome sequences of clades 1 and 3 strains. In contrast, the recent increase in the rate and severity of Clostridium difficile-associated disease in North America has been associated with a new strain characterized by a deletion in a putative negative regulator of toxins A and B (6). Our incomplete understanding of the pathogenesis of disease caused by E. coli O157:H7 and the lack of a suitable animal model contribute to the difficulty in explaining the potential increased virulence of clade 8 strains.

Although Manning et al.'s conclusion that clade 8 strains has acquired critical factors that contribute to more severe disease could be borne out in future studies, it should be noted that pathogen virulence factors represent only part of the equation in attempting to understand the outcome of infection. Differences in the frequency of HUS between outbreaks could also be a function of host factors such as immunity, age, and gastric acidity and also the concentration of the organism in the food. Expression of virulence genes in different food vehicles is also a theoretical factor that could influence clinical outcome. Although the molecular information presented in their study does not give any obvious insights into preventing transmission of E. coli O157:H7 infection, a task that is greatly complicated by the very low infectious dose (<100 colony-forming units), Manning et al. note the critical importance of further investigation of the bovine reservoir, agricultural practices in areas where livestock and produce fields coexist, and other factors involved in the transmission of disease.

The work of Manning et al. (2) reminds us that evolution of microbial agents is an ongoing process. E. coli O157:H7 is the most notorious of several different types of pathogenic E. coli that are “relentlessly evolving” (7). This study demonstrates the power of current molecular techniques, where the entire genome sequence of a bacterial pathogen can be determined to investigate a specific disease outbreak, but it also demonstrates the limitations of such techniques in relating the sequence information to the complex interaction of host and pathogen.