Characterisation of dendritic cells in the intestine

Due to the large surface area of the gut and its continual exposure to a wide variety of agents including dietary constituents, commensal bacteria and pathogens, the intestinal arm of the immune system has evolved to be the largest component of the immune system. It must be able to discriminate between harmless and harmful antigens, so that it can induce tolerance to harmless commensal, self or dietary antigens, but active immunity against pathogens. As the sentinels of the immune system, intestinal dendritic cells (DCs) are central to these processes, continually sampling antigen in the environment and migrating to the mesenteric lymph nodes (MLNs), where they present the antigen to naïve T cells and induce appropriate T cell responses. However the nature and functions of DCs in the intestine remains a topic of debate. Their characterisation has been hampered by the use of non-specific and overlapping markers which has led to intestinal DCs being confused with other cells of the mononuclear phagocyte system, especially macrophages (mφs) which vastly outnumber DCs in the intestinal mucosa. While considerable progress has been made in recent years with the identification of CD103 and CX3CR1 as mutually exclusive markers of DCs and mφs respectively, it has become assumed that CD103+ DCs are intrinsically tolerogenic and thus it remains unclear how DCs contribute to active immunity in the intestine. Furthermore it is unknown whether CD103 is sufficient to define all intestinal DCs, or whether bona fide CD103- DCs may also exist. Thus a major aim of my project was to develop methods that allowed precise characterisation of the mononuclear phagocytes in the intestinal lamina propria (LP) and examine the functions of phenotypically defined subsets. As part of this, I also examined the contribution of the inhibitory signalling receptor, signal regulatory protein alpha (SIRPα) co-expressed by CD11b+ DCs, in regulating intestinal DC function. In Chapter 3, I set out to examine the phenotype of mononuclear phagocytes in the small intestine lamina propria (SI LP). Initially I confirmed previous studies that CD103 and CX3CR1 were mutually exclusive markers of DCs and mφs respectively. This identified 2 populations of DCs separated on the basis of CD11b expression, together with two populations of mφs distinguished by their levels of CX3CR1. However, further experiments examining F4/80 expression in combination with the recently identified mφ-specific marker CD64 showed that the CX3CR1int CD103- MPs were heterogeneous. Although the majority were F4/80+CD64+CD11b+ mφs, I could also identify two additional populations that were F4/80-CD64- and could be separated on the basis of CD11b expression. I hypothesised these were DCs and this was supported by the fact that all 4 subsets of putative DCs could also be found amongst CD11c+MHCIIhi migratory DCs in the MLNs and in pseudo-afferent intestinal lymph. All the subsets also expressed genes and markers of DCs but not mφ and were dependent on Flt3L in vivo. Unlike CD64+ mφs, the DC subsets had no ability to phagocytose E. coli particles. Four similar subsets were also identified in the colonic LP, however the proportions of the subsets in this location were distinct from those seen in the SI LP. While the CD103+CD11b+ DCs were the main subset in the SI LP, in the colonic LP the CD103+CD11b- DCs dominated. Having identified two novel populations of genuine CD103- DCs in the intestinal LP, in Chapter 4 I went on to examine their origin. Previous reports had shown that CD103+ LP DCs were derived from DC-committed precursors (pre-DCs), whereas CD103- MPs were reported to be of monocyte origin. However as I had shown the CD103- MPs to include both DCs and mφs, it was necessary to re-examine their origin using appropriate gating strategies. Adoptive transfer of pre-DCs from the BM into resting WT mice generated all subsets of DCs in the LP, including the two novel populations of CD103- DCs I had identified. In contrast adoptive transfer of Ly6Chi monocytes into monocytopenic CCR2-/- recipients generated mφs exclusively. By comparing the small intestine, colon and spleen, I could show that the development of pre-DCs was determined by the tissue they entered, as the progeny took on the same subset profiles as seen in the endogenous DC populations. Thus the differentiation of pre-DCs appears to be driven by the local microenvironment. By tracking the appearance of donor-derived DCs over time, I could monitor their differentiation in situ. These studies and experiments using BrdU incorporation showed that all DC subsets turned over much more rapidly in vivo than mφs and that a significant proportion were actively dividing in situ. No clear differences suggesting a precursor-product relationship between any of the DC subsets could be seen in these kinetic experiments. To gain a better idea of how the DC subsets might develop, I also examined them in neonatal animals and examined the effects of administering broad-spectrum antibiotics. These studies demonstrated that the CD103+CD11b+ DCs were likely regulated by the presence of specific microbiota as they did not develop in the neonatal animals until day 7 after birth and were increased in proportion following administration of antibiotics. In Chapter 5, I examined how the DC populations might behave during inflammation, using DSS colitis, post-operative ileus and infection with Citrobacter rodentium as models. DSS-colitis caused considerable inflammatory infiltrate and the number of DCs was increased, however there were no subset specific differences. Post-operative ileus also caused inflammation characterised by monocyte and neutrophil infiltration, but had few effects on the DC populations. Infection with C. rodentium resulted in a selective increase in the number of CD103- DCs in the colonic LP, suggesting these may be involved in modulating the Th17 response which characterises the protective immune response in this infection. By transferring pre-DCs into colitic mice, I found that these still gave rise to all the DC subsets during inflammation. In Chapter 6, I examined the functions of the phenotypically defined subsets of LP DCs by pulsing them with ovalbumin (OVA) protein in vitro and culturing them with OVA-specific CD4+ or CD8+ T cells. Consistent with their expression of CD8α and XCR1, I found the CD103+CD11b- DCs to be the most efficient at cross-presenting antigen to naïve CD8+ T cells and they were also the most efficient inducers of IFNγ-producing CD4+ T cells. All populations of DCs could induce FoxP3+ TReg cells, but consistent with their ability to produce retinoic acid as measured by the ALDEFLUOR assay, the CD103+ DC subsets were most efficient at this. The CD103+CD11b- subset also expressed the TGFβ-activating integrin αvβ8. In contrast, induction of IL17a-producing CD4+ T cells was a function of CD103+CD11b+ and CD103-CD11b+ DCs, with the latter being the most efficient. In Chapters 7 and 8, I examined the role of the inhibitory molecule SIRPα in intestinal DC behaviour by examining the DC populations in SIRPα mutant (mt) mice, which have a truncated cytoplasmic domain and hence cannot signal intracellularly. Despite being expressed by most myeloid cells including all CD11b+ DC subsets and CD64+ mφs, SIRPα mt mice had a selective defect in the number of CD103+CD11b+ DCs in the LP and MLN. This correlated with a reduction in the number of Th17 cells in the LP of steady state SIRPα mt mice and these mice showed reduced levels of Th17 cell induction after antigen-specific immunisation and infection by C. rodentium. In parallel, SIRPα mt mice had impaired clearance of C. rodentium infection. T cells from SIRPα mt mice did not have an intrinsic defect in their ability to be polarised to the Th17 phenotype and CD103+CD11b+ DCs from SIRPα mt LP were fully capable of priming Th17 cells in vitro. They also produced normal amounts of the Th17 inducing cytokines IL1β and IL6. As total DCs from SIRPα mt mice produced lower than normal levels of these cytokines, these findings suggest that the defective induction of Th17 responses in vivo reflects the lower numbers of CD103+CD11b+ DC in these mice, rather than a defect in their function. However, preliminary studies investigating the role of CD103-CD11b+ DCs from these mice in the induction of Th17 responses suggests that lack of SIRPα signalling in these cells may also contribute to the reduced Th17 cell population observed in the LP. Having identified these defects in Chapter 7, in Chapter 8 I went on to examine how the SIRPα mutation might be affecting the homeostasis of CD103+CD11b+ DCs. By competitively transferring pre-DCs from WT and SIRPα mt mice into WT recipients, I found that the SIRPα mt precursors generated CD103+CD11b+ DCs in the LP more rapidly than WT pre-DCs. Although this seemed to be at odds with the reduced numbers of these DCs in the SIRPα mt intestine, further work revealed that the CD103+CD11b+ DCs from SIRPα mt LP expressed higher basal levels of CD86 than their WT counterparts, while Annexin V staining suggested that these DCs were apoptosing more rapidly in SIRPα mt MLN. As these results suggested that the lack of CD103+CD11b+ DCs in SIRPα mt mice might reflect increased activation and susceptibility to death because of the absence of SIRPα’s inhibitory signal, I examined whether this would also lead to more efficient migration into afferent lymph. However there were no differences in the proportions of CD103+CD11b+ DCs found in pseudo-afferent intestinal lymph from SIRPα mt and WT mice. Thus more rapid departure into lymph may be sufficiently marked to counterbalance the decreased population in the mucosa. Further experiments are clearly needed to test the hypothesis that lack of SIRPα mediated inhibitory signalling allows hyperactivation of CD103+CD11b+ DCs and subsequent poor survival, as well as to determine why it is restricted to this subset. Finally in Chapter 9, I examined the role of CCR2 in intestinal DC behaviour, aiming to follow up a recent report which had suggested that CD103- Zbtb46-expressing DCs in the LP were monocyte-derived based on their depletion following administration of an anti-CCR2 antibody. I found CCR2-/ mice had a selective reduction in the number of CD103-CD11b+ DCs in the LP and amongst migratory DCs in MLN, as well as amongst CD11b+ DCs in the spleen. Competitive WT:CCR2-/- BM chimeras experiments demonstrated that the requirement for CCR2 in DC development was cell intrinsic. Although antibody staining, fluorescent CCL2-uptake and RFP expression in CCR2+/RFP mice revealed different levels of CCR2 expression, each method showed there was a proportion of CD103-CD11b+ DCs expressing CCR2. The CCR2+ subset of CD103-CD11b+ DCs in LP expressed similar levels of the DC markers Zbtb46, CCR7 and Flt3 to the other bona fide intestinal DCs and were generated by adoptive transfer of pre-DCs, a proportion of which also expressed CCR2. A similar pattern of expression was found on CD103-SIRPα+ DCs isolated from the human colonic LP. The CCR2+CD103-CD11b+ DCs were the most efficient inducers of Th17 responses by naïve CD4+ T cells in vitro and CCR2-/- mice had reduced generation of Th17 responses after antigen specific immunisation and C. rodentium infection in vivo. This enhanced ability of CCR2+CD103-CD11b+ DCs to induce Th17 responses correlated with higher constitutive and LPS induced expression of IL12/IL23p40, which was presumed to be IL23. Together these findings reveal an entirely novel population of CCR2+ and CCR2-dependent, genuine DCs which have a selective ability to drive Th17 responses. Taken together, the data in this thesis represent an important step forward in our understanding of intestinal dendritic cells. By developing rigorous gating strategies, I have been able to distinguish accurately between the various populations of intestinal mononuclear phagocytes and have identified two novel populations of CD103- conventional DCs that are derived from DC committed precursors rather than monocytes as previously assumed. Furthermore, by showing there is a population of bona fide intestinal DC which expresses CCR2 and is dependent on this receptor in vivo, my project may help explain why previous studies have concluded that intestinal DCs may be derived from Ly6ChiCCR2+ monocytes. Importantly, the phenotypic subsets I identified appear to have functional consequences, although my work on SIRPα mt mice and on CCR2 expression have generated somewhat contradictory results on exactly how the different subsets of CD11b+ DCs contribute to Th17 polarisation. One explanation for this may be that these subsets are related members of a linear differentiation pathway, in which CD103-CD11b+ DCs precede the development of the CD103+CD11b+ subset. Further work on the nature of the precursors of these subsets at the single cell level will be required to shed light on this possibility, while it will also be important to investigate the nature of the local factors which determine the unique differentiation pathways of intestinal DCs