CJIE1-positive isolates were around 6 to 7-fold more adherent and 16 to 21-fold more invasive for INT-407 cells in culture than the isolate that did not carry this prophage. These values are close to those found when other genes have been examined; a similar 7-fold increase in adherence to INT-407 cells was found with a cj1461 (methyltransferase) mutant versus the wild-type strain . Mutants in waaF showed a 14-fold reduction in invasion of INT407 cells compared with the wild type strain . Disruption mutants of adhesin-encoding genes cadF and flpA exhibited a 72% and 62% reduction in adherence, respectively . Insertion mutagenesis of cj0588 encoding the TlyA product caused a significant reduction in adherence to Caco-2 cells in culture of C. jejuni strains 81–176 (decreased to 59% compared with wild type) and 81116 (reduced to 48% compared with wild type) . Results from our assays were quite similar to these studies, showing a 0.5 to 1.0 log reduction in adherence of the isolate without the CJIE1-family prophage (Table 2).
The presence of the prophage therefore makes a substantial contribution to the adherence of the lysogenized bacterium. Though the trend to much higher adherence by isolates carrying the prophage was clear in all experiments, the differences in the adherence of isolates with and without the prophage did not reach statistical significance. This was likely partly due to the inter-experimental variability in the adherence and invasion assays, which has been noted before  and appears to be a characteristic of the assay. Differences in adherence in vivo can be very significant even when cell culture assays demonstrate no difference between strains . It is critically important that the role of the prophage be assessed in a relevant animal model and with functional mutagenesis studies.
Invasion of Caco-2 cells was reduced in tlyA mutants to 56% and 31% of wild-type in C. jejuni strains 81–176 and 81116, respectively . The 16- to 21-fold difference in invasion detected in the isolates with and without the CJIE1-family prophage was similar to this but much less than the 50-fold reduction in invasion of INT-407 cells resulting from an insertion mutation of cj1461 . However, the cj1461 mutant also resulted in a motility defect, which is known to have profound effects on invasion [14, 15]. In contrast, no gross alterations in motility were seen in C. jejuni isolates with and without the prophage in the present study. The relative numbers of invaded bacteria expressed as a percentage of those adherent at 30 min post-inoculation was higher than seen by Christensen et al. . However, the differences between adherence and invasion of bacteria with and without the CJIE1-family prophage were consistent in all experiments, suggesting that whatever technical differences resulted in the higher %I/A values were also consistent. The measurable differences in adherence and invasion associated with prophage carriage found in this study appear to be substantiated. Of note, both strain 81–176 and 00–2538 had differences in adherence that were statistically significant when compared with isolate 00–2426. Furthermore, the levels of adherence and invasion expressed as percentage of input or inoculum counts was very similar to that found in other studies .
DNA sequencing of the CJIE1-1 prophage from isolate 00–2425  has demonstrated the presence of a few genes associated with the prophage that are likely not important for prophage structure, life cycle, or replication, ie. that appear to be cargo genes, in addition to a number of hypothetical proteins. Among the putative cargo genes are: the CJE0220 homolog, a DAM methylase; ORF3, a KAP family P loop domain protein; a CJE0256 homolog, dns, an extracellular DNase; ORFs 10 and 11 inserted in the early region of the prophage with no homology to any protein of known function within GenBank. We speculate that the effects of the CJIE1-1 prophage on cells in culture are mediated either by a novel effector or by a regulator of virulence genes or even general metabolism within the C. jejuni bacterial cell. Differences in protein expression between isolates with and without CJIE1 in iTRAQ experiments support this hypothesis (unpublished data).
No consistent or statistically significant differences in motility were found when comparing isolates with and without the prophage. The differences in adherence and invasion were therefore not directly the result of differences in motility, and were also not likely to be due to differences in gene content, other than the previously noted prophage genes, or growth rate.
The four isolates used were all obtained at the same time and in the same place during an outbreak of disease. They were the same subtype and had indistinguishable gene content as measured by comparative genomic hybridization DNA microarray analysis except for the fact that isolate 00–2426 lacked the CJIE1-family prophage. Though a consistent difference in growth rate was seen during mid-logarithmic phase between the isolate lacking the prophage and the three isolates carrying the prophage, this difference was extremely subtle. It does not seem likely that this degree of difference could be responsible for the differences seen in adherence and invasion.
It must be noted that the combination of microarray data and calculation of genome sizes does not prove absolutely that the four isolates have identical DNA sequences other than the presence or absence of CJIE1. Because the microarray had probes for genes from only two strains it is possible that other genes or DNA segments could be present. However, calculation of genome sizes from PFGE fragments sizes was done previously with a reasonable degree of accuracy, and the resulting data indicate that genomes of the isolates 00–2425 and 00–2544 carrying CJIE1 differed from 00–2426, which lacked CJIE1, by 39 kb . This constrains the variability that would be expected for the four genomes mainly to the presence or absence of the prophage and to DNA sequence changes arising from horizontal gene transfer. The use of three different, closely related isolates carrying the CJIE1 prophage at different places in the genome (as assessed through differences in PFGE patterns) should minimize the chances that differences arising from proximity of the prophage to specific genes, from point mutations, and from horizontal gene transfer. Whole genome sequencing of these isolates is planned for the near future and should provide unambiguous data regarding gene content and prophage location.
An unexpected observation unrelated to the investigation into prophages came from conducting growth curve experiments with C. jejuni for the first time. Very similar OD600 values were obtained for all four test strains after 48 h (early stationary phase) growth in initial experiments suggesting that, if differences existed between isolates, they were both quite subtle and quite growth phase-specific. Note that these subtle effects were visualized as occurring in mid-log phase (around 5 × 105 cfu/ml) as measured by plating growing cultures, and would likely not have been observed if growth were measured using spectrophotometry, as growth was not detectable at OD600 until cell density was between 5 × 107 to 1 × 108 cfu/ml (data not shown).
Molecular typing data and information about patient symptoms were available for a relatively large number of human and non-human isolates obtained through the C-EnterNet sentinel site surveillance system. Though there appeared to be some association of ORF11 with bloody diarrhea and hospitalization, this did not attain statistical significance. A further, somewhat puzzling, observation was that the presence in C. jejuni of CJIE1 in the absence of ORF11 appeared to reduce the frequency of some symptoms (Table 3). This was statistically significant for abdominal pain and fever, though caution should be used in interpretation of the statistical analysis because only a relatively small number of isolates fit into this category. It should be noted that not all patients for which isolates were available filled out questionnaires, and isolates were not available for all patients who filled out questionnaires. It would be of interest to add to the observations in this study over time and determine whether any of the apparent trends are supported by further data.
Carriage of both the prophage and of ORF11 was less frequent in most C. jejuni isolates from water, suggesting these elements do not have adaptive value for the organism in this environment. Further research is required to verify this observation and to determine whether this is associated with the biology of the organism or purely stochastic in nature. Differences in the proportion of isolates with and without the CJIE1 prophage between C. jejuni isolates from chicken, human, and bovine sources were either slightly statistically significant (chicken and bovine, P = 0.027) or not significant (chicken and human, human and bovine). There is no strong evidence that the prophage or ORF11 play a role in host adaptation or host specificity.
A striking difference in the frequency of carriage of both CJIE1 alone and of CJIE1 + ORF11 in both STs and in flaA SVR types suggests that the carriage of these elements may be specific to certain Campylobacter lineages, groups, or clones. Prophage CJIE1 + ORF11 was found at higher frequency in ST 8, 21, 48, and 982. STs 21 and 982 differ only by a single allele and ST 8 is included with ST 21 in clonal complex 21, while ST 48 differs at three alleles from ST 21 and four alleles from ST 982. Similarly, CJIE1 alone is found at higher frequency in ST 21, 42, 50, and 982, and a few other STs, while it is found in much lower frequency in ST 45 and several additional STs (Table 5). One possibility is that the carriage and transmission of the CJIE1 prophage may be strongly associated with a specific animal host or environmental niche. MLST types exhibit a host-specific signature of alleles acquired through homologous recombination during carriage and adaptation of Campylobacter within the host species . Studies in Finland indicate that the ST-45 clonal complex is significantly associated with chicken isolates, while the ST-21 and ST-48 clonal complexes are significantly associated with human isolates . Clonal complexes ST-21 and ST-42 are also among the lineages that predominate among C. jejuni isolates from cattle . Together this information might suggest that the CJIE1 prophage, like the host-specific MLST alleles, may be circulating in a subset of C. jejuni more closely associated with humans and cattle than with chickens. This finding supports the conclusions of Pittenger et al. , who determined that C. jejuni RM1221 variable genes – most of them of prophage origin – were more widely distributed in isolates from cattle and humans than from other sources. However, for CJIE1 it was apparent from the results presented in Table 4 that the prophage was present in a greater proportion of C. jejuni from chickens and swine manure than any other sources, though the number of isolates obtained from swine manure do not allow much confidence in that result. A great deal more research into the association of prophages and cargo genes carried by prophage elements is warranted.