Lamina propria macrophage phenotypes in relation to Escherichia coli in Crohn’s disease

We found that E. coli can be identified within LP macrophages in most CD patients using immunohistochemistry with an anti-E. coli specific polyclonal antibody. This concurs with previous studies that demonstrate E. coli within LP macrophages in CD using immunohistochemistry [5] and FISH [25] or within granulomas using LCM and nested PCR [10]. In future studies, further confirmation of the presence of E. coli within lamina propria macrophages in CD could be achieved using an alternative technique such as 16S rRNA PCR. Additionally, the presence of other bacteria within macrophages could be determined by extracting DNA from laser-captured macrophages, performing bacterial 16S rDNA sequencing and comparing any 16S rDNA sequences present to database reference bacterial sequences [26].

It is possible that the presence of intramacrophage E. coli in CD results from successful adherence to and invasion of the mucosa by AIEC, with subsequent survival and replication within LP macrophages. Certainly, in vitro, AIEC have been shown to possess properties that might facilitate this process [8, 9]. However, it is also possible that an innate defect of bacterial killing by LP macrophages contributes to E. coli persistence within macrophages, a concept which is also supported by the presence of macrophage cytokine defects and mutations in bacterial handling genes in CD [1, 27, 28]. We did not determine whether intra-macrophage E. coli were AIEC in our study because these E. coli cannot be distinguished morphologically. However, using culture of biopsies with gentamicin protection to isolate intracellular E. coli from the mucosa in CD, we have found that only a minority of intracellular isolates have adherent and invasive properties in vitro [12].

The lower prevalence (1/9) of E. coli-laden macrophages in UC is in keeping with previous reports of lower intramucosal bacteria in UC than CD [6, 29]. Meanwhile, the absence of E. coli-laden macrophages in all 18 healthy subjects illustrates that the mucosal immune system prevents bacterial persistence within the LP in health.

Real-time RT-PCR can provide a sensitive and accurate measurement of mRNA when performed in accordance with the MIQE guidelines. Normalisation of mRNA data remains a contentious issue. House-keeping genes such as GAPDH and β-Actin have been proposed as being stable appropriate reference genes for normalisation. Prior to initiation of this study, we performed preliminary experiments using GAPDH and β-Actin and found that these reference genes were unstable in these inflamed tissue samples. We also have previously published data demonstrating that internal reference genes may not be appropriate for normalisation of qPCR data for mRNA, especially when derived from tissue biopsies [30, 31]. However, the use of total RNA for normalisation has been demonstrated to be valid [24] and produce quantification results that are biologically relevant [30] as long as certain criteria are met [32, 33]. These are that the RIN is above 8 (which can be considered perfect total RNA for downstream applications [22]) which was confirmed using the Agilent 2100 Bioanalyzer in our study, and the use of small amplicons which minimise the variability caused by RNA degradation. Hence, our decision to use total RNA for normalisation in this study is valid, especially as we are reporting very large and characteristic differences in expression of mRNA for a range of cytokines between Crohn’s disease, healthy controls and UC that are not consistent with chance RNA degradation.

In healthy controls, cytokine and surface marker mRNA expression in macrophages from uninflamed colonic mucosa were low, in keeping with previous data on intestinal macrophages in health [34]. In inflamed CD mucosa, there was clear differentiation of macrophage cytokine and surface marker profiles according to E. coli carriage (E. coli-unladen; higher proinflammatory cytokines (TNFα, IL-23, IL-6, IL-8) and iNOS, and E. coli-laden; higher IL-10 and CD163, both characteristic features of regulatory macrophages). The phenotype of E. coli-unladen macrophages in CD is consistent with that of recently recruited CD14+ macrophages, which secrete high TNFα and IL-23 and are more numerous in active CD [16]. It is likely that these activated macrophages contribute significantly to inflammation and recruitment of other pro-inflammatory cells important in CD pathogenesis such as Th17 cells [16]. E. coli-unladen macrophages in uninflamed mucosa from CD patients with active disease expressed very low cytokine mRNA levels, similar to healthy controls, and are therefore likely to be inactive resident LP macrophages [15].

The observed phenotype (high IL-10, lower TNFα) of E. coli-laden macrophages in CD might represent an appropriate regulatory response to microbial encroachment, or may facilitate E. coli persistence, and thus contribute to pathogenesis. Supporting the former supposition, the immunoregulatory role of macrophages secreting IL-10 is well documented [35]. However, IL-10 also facilitates intracellular persistence of numerous microorganisms [35], possibly due to inhibition of autophagy [36]. Interestingly, in Whipple’s disease, Tropheryma Whipplei accumulate in duodenal macrophages which express a similar phenotype to the E. coli-laden macrophages in this study (high IL-10, CD163) which is thought to facilitate their persistence [37].

E. coli-laden macrophages were present and also expressed a high IL-10 phenotype in uninflamed mucosa in 3 of the 6 CD patients with active disease in whom paired inflamed/uninflamed biopsies were taken. This raises the possibility that E. coli access the mucosa at an early stage of CD pathogenesis rather than as a consequence of a disrupted inflamed mucosa. Of note, the presence of E. coli-laden macrophages correlated with endoscopic severity and higher pro-inflammatory cytokine mRNA expression of E. coli-unladen macrophages. This highlights further the dilemma of cause and effect as this may either be because the presence of E. coli in LP macrophages causes intestinal inflammation or that their presence is merely a consequence of inflammation.

In UC, E. coli-unladen macrophage cytokine and surface marker expression were elevated often to similar levels as in CD, however COX-2 mRNA was substantially lower in UC. This may be in keeping with a recent Danish study that reported the association of a COX-2 gene polymorphism (A-1195G variant allele) with UC but not CD [38]. The authors hypothesised that the mutation, which is associated with low COX-2 activity, may lead to increased UC susceptibility because of reduced prostaglandin synthesis as prostaglandins regulate mucosal inflammation. It would be of interest to correlate the prevalence of this mutation with macrophage COX-2 expression in UC in future work.