From May to July 2011, Germany was struck by the largest outbreak of hemolytic uremic syndrome (HUS) and bloody diarrhea caused by Escherichia coli ever reported [1]. A total of 3,816 cases were reported, 845 (22%) of which involved HUS. This high rate of HUS was the first indicator that the bacterial cause of illness was not a typical enterohaemorrhagic E. coli (EHEC) strain [2]. It was shown to be a highly pathogenic enteroaggregative E. coli (EAEC) strain, which in addition carried the EHEC genes for the classical HUS-associated Shiga toxin 2 [3]–[4]. Notwithstanding the timeliness of the German surveillance system (reporting occurred faster than required by law) for HUS and Shiga toxin-producing E. coli notifiable diseases [5], no less than 54 patients died. Eventually, sprouts were identified as the most likely outbreak vehicle [6]. The outbreak strain was resistant to beta-lactam antibiotics and third-generation cephalosporins and was partially resistant to fluoroquinolones, but sensitive to carbapenems and ciprofloxacin [1]. However, a Cochrane database review analyzing seven randomized controlled clinical trials of pediatric HUS associated with EHEC infections found no benefit of antibiotic treatment over supportive therapy alone [7]. Moreover, the use of antibiotics to treat Shiga toxin producing E. coli infections has been discouraged because it has been shown to release toxins. The genes for the Shiga toxin are actually not bacterial genes, but (bacterio)phage genes. When an E. coli bacterium gets infected with a temperate phage harboring a Shiga toxin gene, upon integration of the phage genome (as prophage) into the bacterial genome (lysogeny), it can be expressed by the bacterium, which then becomes pathogenic. Some types of antibiotics have been shown to induce the so-called SOS response in bacteria, i.e. a ubiquitous response to DNA damage, which induces phage replication and lytic cycle [8]. As such, the use of antibiotics may be helping Shiga toxin genes to spread. Phage-mediated transfer of bacterial virulence, fitness and antibiotic resistance is a negative (from the human point of view) consequence of bacterial phage coevolution.

Isolation and selection of E. coli O104:H4 phages were performed according to the methods described by Merabishvili et al. [15]. Considering laboratory safety in the handling of EHEC, all manipulations involving the O104:H4 outbreak strain were performed in a dedicated biosafetylevel 3 facility. A set of 16 phages known to infect E. coli was screened against the outbreak strain. The set included the well-known T4 phage, nine phages from the therapeutic phage library of the Eliava Institute of Bacteriophage, Microbiology and Virology in Tbilisi (Georgia) and six phages that were recently isolated from wastewater of the Queen Astrid Military Hospital in Brussels.

Only one out of 16 phages, myovirus GEC-3S (Fig. 1) from the Eliava collection, showed activity against the outbreak strain. In addition, two myoviruses, QAMH-FFP.1 and QAMH-NES.1, were newly isolated from the wastewater of the Brussels Military Hospital using the culture enrichment method with the original outbreak strain as the host. In contrast to phages QAMH-FFP.1 and QAMH-NES.1, phage GEC-3S could be propagated in E. coli strain K12 and this without significant decrease of titer or activity against strain O104:H4, which is not negligible in the context of an eventual production of therapeutic or prophylactic phage preparations. The non-pathogenic K12 strain is commonly used in Good Manufacturing Practices (GMP) and clean-room compliant preparation of medicinal products, whereas the handling of the O104:H4 strain requires strict biocontainment precautions. Phages QAMH-FFP.1 and QAMH-NES.1 showed no activity against the K12 strain. Of the ten tested pathogenic E. coli serotypes, only two – O104:H4 and O125:K70– were susceptible to phage GEC-3S (Table 1). The overall activity of the phage against 92 pathogenic and non-pathogenic E. coli strains proved to be 20.7% (19/92). Complete genome sequencing (Accession number HE978309) revealed that phage GEC-3S belongs to the genus of T4-like viruses with high similarity to the representatives of the RB49 group. Two hundred and seventy-six open reading frames were predicted and their functions were defined. At the nucleotide level, GEC-3S showed high homology to phi1, RB49 and JSE phages, respectively 94.6, 93.6 and 88.0%. The shared differences between GEC-3S and the other phages are spread throughout the whole genome and in particular in the regions between 39,000–40,000, 67,000–68,000, 100,000–110,000 and 115,000–125,000 bp, loci which mostly comprise ORFs coding for small hypothetical proteins and homing endonucleases (Fig. 2). Phage GEC-3S harbors six unique genes that are not present in the genomes of the other representatives of the RB49 group (phi1p263, RB49p263, EpJSE_00266) (Table 2). Four of these genes are ORFans, while the other two are coding homing endonucleases, extremely widespread enzymes in T-even-like phages according to available genomic data [22]. The only gene that is typical for the other three phages from this group, but is not present in GEC-3S is a hypothetical protein of 51 amino acids length (Table S1).