Proinflammatory Role of Monocyte-Derived CX3CR1int Macrophages in Helicobacter hepaticus-Induced Colitis

ABSTRACT Cells of the monocyte/macrophage lineage play important roles in the pathogenesis of inflammatory bowel diseases, but they are also present in the normal healthy intestine, where they are critical for maintaining homeostasis. It has been unclear whether the proinflammatory roles of intestinal macrophages reflect altered behavior of the existing resident cells, or whether they involve recruitment of a distinct cell type. Here, we have explored these ideas using the model of colitis induced by Helicobacter hepaticus in the context of neutralization or deletion of interleukin-10 (IL-10). Granulocytes and monocytes made up most of the inflammatory myeloid infiltrates found in the colon of H. hepaticus-infected colitic mice, rising to a peak within 2 weeks of H. hepaticus inoculation but taking several months to resolve completely. The inflammatory response was dependent on the combined presence of H. hepaticus and absence of IL-10 and was accompanied by increased production of inflammatory mediators such as IL-1β, tumor necrosis factor alpha (TNF-α), IL-6, and IL-23p19 by infiltrating myeloid cells, mostly relatively immature cells of the macrophage lineage that express intermediate levels of CX3CR1. In contrast, the population of mature CX3CR1hi macrophages did not expand as markedly during colitis, and these cells made little contribution to inflammatory mediator production. Taking into account their numerical dominance in the myeloid compartment, we conclude that newly recruited monocytes are the main source of proinflammatory mediators in colitis induced in the absence of IL-10 signaling and that altered behavior of mature macrophages is not a major component of this pathology.

are growing health problems in the developed world. Although recent advances have helped elucidate some of the mechanisms underlying IBD, treatment remains unsatisfactory and individual regimes are seldom effective for all patients. Thus, a better understanding of the processes and cells involved in the pathogenesis of IBD could help develop new targets for therapy.
Macrophages (Ms) have received considerable attention in recent years because of their potential roles in both steady-state and inflamed intestines (1)(2)(3). Resident Ms are abundant in the healthy intestine, where they are involved in clearance of apoptotic cells and play a crucial role in maintaining homeostasis, ingesting and killing commensal bacteria that cross the epithelial barrier (reviewed in reference 2). In contrast to other tissues, these processes do not provoke overt inflammation in the intestine, due to powerful control mechanisms that prevent local Ms from producing proinflammatory mediators in response to stimuli such as Toll-like receptor (TLR) ligands (4,5). However, findings from both IBD and experimental models of the disease have dem-onstrated that Ms are also crucial components of the inflammatory infiltrate (4,(6)(7)(8)(9), raising the possibility of exploiting them as therapeutic targets.
We and others have shown recently that intestinal Ms originate from Ly6C hi blood monocytes that continuously enter the colonic mucosa and differentiate locally through a series of intermediaries. Although present in only small numbers, cells with the phenotypic and morphological features of Ly6C hi monocytes can be found in the steady-state mucosa. However, major transcriptional differences exist between these and Ly6C hi monocytes found in blood, suggesting that the differentiation process occurs immediately after monocytes enter the colonic mucosa (6,10). Importantly, as monocytes mature, they progressively acquire anti-inflammatory properties, such as constitutive production of interleukin-10 (IL-10) and hyporesponsiveness to, e.g., TLR ligands (6,11,12). However, this physiological process is disrupted during inflammation (6,12,13), and understanding how this change in M behavior occurs would be an important advance in our knowledge of how to prevent and treat IBD. Studies of the colitis induced by feeding dextran sodium sulfate (DSS) suggest that the tissue pathology is paralleled by the accumulation of highly proinflammatory monocytes, whereas the fully differentiated Ms that remain do not alter their behavior and retain their anti-inflammatory characteristics (6,12). However, it is not clear if this pattern extends to other forms of intestinal inflammation, particularly under conditions where there are intrinsic defects in the mechanisms that normally control resident M function.
The IL-10 -IL-10 receptor (IL-10R) axis is an important brake on M activation, particularly in the intestine, where deletion of either the cytokine or its receptor leads to spontaneous onset of inflammation in association with hyperresponsiveness of Ms (13)(14)(15)(16)(17)(18)(19). Furthermore, humans with nonfunctional mutations in IL-10, IL10RA, or IL10RB develop severe enterocolitis within the first months of life (reviewed in reference 20). Here, we have examined the relationship between IL-10 and the differentiation of intestinal Ms in inflammation in more depth by exploring M behavior during colitis induced by inoculation of mice with Helicobacter hepaticus in the absence of IL-10 signaling (21,22). We report that H. hepaticus-induced colitis is characterized by the accumulation of proinflammatory CD11b ϩ myeloid cells that are hyperresponsive to TLR stimulation. Most of these cells are monocytes and their immediate progeny that express intermediate levels of CX3CR1 and that have not differentiated fully into anti-inflammatory, resident-type CX3CR1 hi Ms. We further show that the expansion of cells in the monocyte/M compartment is maintained for several months before returning to normal levels as the inflammation resolves.

Expansion of myeloid cells in the large intestine of H. hepaticus-infected colitic
To begin to characterize the innate immune response in the large intestine following H. hepaticus inoculation, wild-type (WT) and Il10 Ϫ/Ϫ mice were inoculated intragastrically (i.g.) with the bacterium, and the cellular composition of lamina propria (LP) cells from pooled ceca and colons was examined 2 weeks later, the time point at which pathology peaks in Il10 Ϫ/Ϫ hosts (23). As expected, H. hepaticusinfected Il10 Ϫ/Ϫ mice displayed enhanced cellularity of the large intestinal LP compared with uninfected controls (Fig. 1A, left panel), with greatly expanded proportions and numbers of CD45 ϩ hematopoietic cells (Fig. 1A, middle and right panels), consisting of B cells, CD4 ϩ T cells, and CD11b ϩ myeloid cells (Fig. 1B). Only minor increases were observed in the proportion and number of LP CD45 ϩ cells in H. hepaticus-infected WT mice (Fig. 1A), and the size of the CD11b ϩ myeloid compartment was unaffected in these hosts (Fig. 1B, right panel). More detailed analysis of the early cellular kinetics following H. hepaticus inoculation of Il10 Ϫ/Ϫ animals revealed that the numbers of total CD45 ϩ cells and of CD11b ϩ myeloid cells were significantly increased in H. hepaticusinfected Il10 Ϫ/Ϫ mice on day 5 postinfection (p.i.) and expanded steadily until day 11 p.i. (Fig. 1C). Thus, H. hepaticus-driven typhlocolitis in Il10 Ϫ/Ϫ mice is associated with massive infiltration by both lymphoid and myeloid cells.

Myeloid cells from H. hepaticus-infected Il10
؊/؊ mice display a proinflammatory phenotype. To understand how the expansion of the myeloid compartment might contribute to the colitis in Il10 Ϫ/Ϫ hosts, we examined the cytokine secretion profile of CD11b ϩ cells from uninfected and H. hepaticus-infected Il10 Ϫ/Ϫ mice. To this end, large intestinal LP cells were stimulated with the TLR ligand Pam3CSK4, and tumor necrosis factor alpha (TNF-␣) production was examined by intracellular cytokine staining (ICS). TLR ligation of LP cells from H. hepaticus-infected Il10 Ϫ/Ϫ animals resulted in an ϳ3-fold increase in the proportion of CD11b ϩ cells producing TNF-␣ compared with cells cultured in medium alone ( Fig. 2A, bottom panels), and this translated to a highly significant increase in the absolute number of TNF-␣ ϩ CD11b ϩ cells in the LP of infected Il10 Ϫ/Ϫ mice compared with uninfected controls (Fig. 2B). In contrast, the  proportion of TNF-␣ ϩ CD11b ϩ cells from uninfected Il10 Ϫ/Ϫ animals was unaffected by Pam3CSK4 stimulation and remained at levels similar to those seen in naive WT colon ( Fig. 2A, upper panels, and B).
To extend these analyses, we next purified CD11b ϩ cells from the large intestinal LP of uninfected and 2-week H. hepaticus-infected Il10 Ϫ/Ϫ animals by fluorescence- activated cell sorting (FACS) and cultured them overnight with lipopolysaccharide (LPS), Pam3CSK4, or soluble H. hepaticus antigen (SHelAg), before assessing the levels of a wider range of cytokines using enzyme-linked immunosorbent assay (ELISA) and FlowCytoMix. CD11b ϩ cells from H. hepaticus-infected Il10 Ϫ/Ϫ mice secreted larger amounts of IL-12p40, IL-6, and TNF-␣ after all forms of stimulation than did CD11b ϩ cells from uninfected Il10 Ϫ/Ϫ hosts (Fig. 2C). When analyzed directly ex vivo, FACSpurified CD11b ϩ cells from H. hepaticus-infected Il10 Ϫ/Ϫ mice also contained higher levels of Il12a and Il23a transcripts than did CD11b ϩ cells from uninfected controls, whereas the levels of Il12b accumulation were similar in the two populations (Fig. 2D). Thus, the expanded myeloid cell compartment in the large intestine of H. hepaticusinfected colitic Il10 Ϫ/Ϫ mice displays proinflammatory characteristics.

Composition of the myeloid compartment in H. hepaticus-infected Il10
؊/؊ mice. We next set out to explore what cell types accounted for the change within the CD11b ϩ compartment during H. hepaticus colitis. To do this, we exploited multiparameter flow cytometry and rigorous gating strategies that we have developed recently to characterize the myeloid compartment of the intestinal LP, allowing precise identification of monocytes, Ms, eosinophils, neutrophils, and dendritic cells (DCs) (6) (see Fig.  S1 in the supplemental material). We also omitted the Percoll gradient step during the purification to exclude the possibility of selective loss of individual cell types. These approaches confirmed marked changes in the composition of the myeloid cell compartment in the colon 2 weeks after H. hepaticus inoculation of Il10 Ϫ/Ϫ mice, compared with uninfected Il10 Ϫ/Ϫ mice or H. hepaticus-infected WT mice (Fig. 3A). Ly6G ϩ neutrophils and SSC hi (major histocompatibility complex class II-negative [MHCII Ϫ ]) eosinophils ( Fig. 3A, B, and C) accounted for a substantial part of the infiltration found in H. hepaticus-infected Il10 Ϫ/Ϫ mice, and no significant differences were seen in these granulocyte populations in H. hepaticus-infected WT mice or in uninfected Il10 Ϫ/Ϫ mice compared with uninfected WT mice (Fig. 3A, B, and C).
H. hepaticus-infected Il10 Ϫ/Ϫ mice also showed marked expansion in the numbers of nongranulocytic myeloid cells (Fig. 3A, D, and E), and we therefore explored the contribution of monocytes and Ms in more detail. To do this, we focused on cells expressing the pan-M markers F4/80 and/or CD64 (Fig. S1) and examined the monocyte/M differentiation continuum that we and others have defined in the colonic LP (6,24,25). This so-called "monocyte waterfall" consists of newly arrived Ly6C hi MHCII Ϫ monocytes, differentiating Ly6C ϩ MHCII ϩ monocytes, and more mature Ly6C Ϫ MHCII ϩ cells that include tissue-resident Ms (Fig. 3D). H. hepaticus-infected Il10 Ϫ/Ϫ mice showed marked increases in the proportions and absolute numbers of Ly6C hi MHCII Ϫ and Ly6C ϩ MHCII ϩ cells, which were increased by ϳ125-fold and 155-fold, respectively, compared with their numbers in uninfected Il10 Ϫ/Ϫ mice ( Fig. 3D and E). There was also a modest but significant increase in the number of more mature Ly6C Ϫ MHCII ϩ cells in H. hepaticus-infected Il10 Ϫ/Ϫ mice (Fig. 3E, right panel). H. hepaticus-infected WT mice, which do not develop colitis, showed some evidence of increased infiltration by Ly6C hi MHCII Ϫ , Ly6C ϩ MHCII ϩ , and Ly6C Ϫ cells, while uninfected Il10 Ϫ/Ϫ mice had slightly higher numbers of Ly6C ϩ MHCII ϩ cells than their naive WT counterparts; however, these differences were modest and did not attain statistical significance (Fig. 3E).
DCs were identified as CD11c ϩ MHCII ϩ CD64 Ϫ cells among LP leukocytes (26), and their numbers increased in H. hepaticus-infected mice compared with control Il10 Ϫ/Ϫ mice (Fig. S2A). The frequency of DCs among LP cells of H. hepaticusinfected Il10 Ϫ/Ϫ mice decreased compared to uninfected controls (data not shown), most likely reflecting the increase in other leukocytes. Finally, there were minor changes in the proportion of DC subsets identified on the basis of CD11b and CD103 expression, although none of these changes reached statistical significance (Fig. S2B).
Altered monocyte/M differentiation in H. hepaticus-infected colitic mice. The accumulation of monocytes that we found in H. hepaticus colitis is reminiscent of our own and other results from DSS-and T-cell-mediated colitis, where there appeared to be an arrest in the local differentiation continuum that normally generates anti-inflammatory, resident Ms (6,12,24). To examine whether a similar block was present during H. hepaticus-induced colitis, we used Cx3cr1 ϩ/gfp reporter mice, in which one allele of the Cx3cr1 gene has been replaced with the gene encoding green fluorescent protein (GFP) (27). This allows fully differentiated resident CX3CR1 hi Ms to be distinguished from cells in the earlier CX3CR1 int stages in the developmental continuum, some of which would have been included among the Ly6C Ϫ MHCII ϩ population that we defined earlier. In this way, we could examine the relative roles of resident and recently recruited Ms in H. hepaticus-induced inflammation, as well as explore how these cells behave during the resolution of pathology that occurs at later stages after bacterial inoculation (23). Because the Cx3cr1 ϩ/gfp reporter mice are IL-10 sufficient, we had to induce colitis by H. hepaticus inoculation plus weekly administration of anti-IL-10R monoclonal antibody (MAb) (22). Consistent with our studies in Il10 Ϫ/Ϫ mice, this resulted in massive accumulation of total leukocytes and CD11b ϩ myeloid cells in the colonic mucosa by day 14 p.i. (Fig. 4A and B). However, by day 41 p.i., both compartments had contracted significantly, and by day 77 p.i., they were both reduced almost to baseline levels ( Fig. 4A and B).
As in colitic Il10 Ϫ/Ϫ mice, the composition of the myeloid compartment was markedly altered in anti-IL-10R-treated H. hepaticus-infected Cx3cr1 ϩ/gfp mice, with a large expansion of granulocytes and monocyte-derived cells at day 14 p.i., followed by resolution at later times (Fig. 4C)    myeloid infiltrate contained large numbers of Ly6C hi MHCII Ϫ and Ly6C ϩ MHCII ϩ cells at day 14 p.i. (Fig. 4C to E). The Ly6C Ϫ MHCII ϩ population also expanded in number early in colitis, but most of this expansion was accounted for by CX3CR1 int cells (Fig. 4F), a population that we have shown previously to be a further intermediary stage in the local differentiation of monocytes (6). In parallel, there was a substantial decrease in the proportion of fully mature CX3CR1 hi Ms among the Ly6C Ϫ MHCII ϩ cells at this time (Fig. 4F), although their absolute numbers were increased and remained so throughout the experiment (Fig. 4G). By day 41 p.i., the proportions and numbers of CX3CR1 int cells among Ly6C Ϫ MHCII ϩ cells had contracted significantly, and by day 77 p.i., CX3CR1 hi cells had again come to dominate the Ly6C Ϫ mature M compartment (Fig. 4F and G). At this time, the numbers of Ly6C hi MHCII Ϫ and Ly6C ϩ MHCII ϩ cells had also returned to baseline levels (Fig. 4E). The numbers of Ly6G ϩ neutrophils and CD11c ϩ CD11b ϩ F4/80 Ϫ DCs were greatly increased by day 14 p.i. in anti-IL-10R-treated H. hepaticus-infected Cx3cr1 ϩ/gfp mice, before falling at later times in parallel with the reduction in monocytes and Ms ( Fig.  S3A and B). Interestingly, however, the numbers of colonic eosinophils remained high in these mice until the end of the experiment on day 77 p.i. (Fig. S3C). All these H. hepaticus-induced changes in myeloid cells were seen only when H. hepaticus-infected Cx3cr1 ϩ/gfp mice were also coadministered anti-IL-10R ( Fig. 4 and data not shown). Anti-IL-10R treatment alone of uninfected Cx3cr1 ϩ/gfp mice did not result in leukocyte accumulation in the colonic mucosa (Fig. S4), findings that are in agreement with those that we previously reported for uninfected WT mice (22).
Together, our results demonstrate that the acute phase of H. hepaticus-induced colitis is associated with accumulation of CX3CR1 int monocytes and early-stage Ms, consistent with the idea that there may be a block in the normal differentiation process.
Colonic CX3CR1 int monocytes/Ms are the major producers of proinflammatory cytokines during H. hepaticus colitis. Finally, we explored whether elicited CX3CR1 int cells were responsible for the production of proinflammatory mediators by myeloid cells during colitis, or if this reflected altered behavior of the more mature CX3CR1 hi Ms. Quantitative reverse transcription-PCR (RT-qPCR) analysis showed marked upregulation of mRNA transcripts for Il1b, Nos2 (inducible nitric oxide synthase [iNOS]), Il23p19, and Il12p35 in CX3CR1 int cells from anti-IL-10R-treated H. hepaticusinfected mice at the peak of inflammation on day 14 compared with CX3CR1 int cells from control mice (Fig. 5A to D). In contrast, colonic CX3CR1 hi Ms from anti-IL-10Rtreated H. hepaticus-infected mice showed only minor changes in mRNA levels of proinflammatory mediators (Fig. 5). Thus, taking into account their numerical dominance in the myeloid compartment, newly recruited CX3CR1 int monocytes/Ms form the predominant proinflammatory population in this model of colitis.

DISCUSSION
The results presented here underline the importance of the myeloid compartment in the inflammatory colitis that occurs in mice infected with Helicobacter hepaticus when IL-10-mediated signaling is absent. Using a variety of approaches to identify myeloid lineages, including Cx3cr1 ϩ/gfp mice, we show that an intense infiltrate of CD11b ϩ cells appears early during infection and that this is made up of neutrophils, eosinophils, Ly6C hi MHC -/ϩ monocytes, and CX3CR1 int Ms at the time of peak disease at 2 weeks after H. hepaticus inoculation. While this confirms other work (28), we also show here that this expansion of the monocyte/M compartment is sustained for up to 11 weeks after infection, by which time other parameters of inflammation, such as changes in myeloid cell subset composition, have returned to steady-state levels. Importantly, the inflammatory changes required both infection with H. hepaticus and neutralization of IL-10R signaling, with very few changes being seen in mice with H. hepaticus infection alone or loss of IL-10 alone. Similarly, the greatly heightened production of proinflammatory mediators in response to activation in vitro by intestinal myeloid cells was fully dependent on these factors operating together. These results confirm previous conclusions that H. hepaticus plays a crucial role in provoking intes-tinal inflammation in the absence of IL-10-mediated immunoregulation and that proinflammatory responses to this organism are normally restrained by IL-10 (21,22,29).
In previous work, we have shown that resident Ms in normal intestinal mucosa are replenished continuously by Ly6C hi monocytes that differentiate via a number of intermediary stages into mature, resident Ms (6,11,24). At early times after inoculation with H. hepaticus, this so-called monocyte/M "waterfall" expanded dramatically and consisted mostly of Ly6C hi MHCII -/ϩ monocytes and CX3CR1 int Ms, with less expansion of fully mature CX3CR1 hi Ms. Although these findings are consistent with previous work, e.g., in chemically induced colitis (6), and suggest that the normal process of monocyte/M differentiation is arrested, it is important to note that the absolute numbers of mature Ms were actually increased in our mice with H. hepaticus-induced colitis and remained so for the duration of the 11-week experiment. Thus, although the majority of infiltrating monocytes may be short lived in the colitic mucosa, a proportion of these cells may still develop into mature CX3CR1 hi Ms, even in the face of inflammation. How the Ms recruited under these conditions might contribute to the inflammation or its resolution and if they eventually acquire all the properties of the normal, resident population remain to be determined, as do the relative contributions of newly recruited and preexisting cells to the "resident" M compartment.
Although methods were not available to examine and track infiltrating monocytes directly following their arrival in the gut, we were able to explore whether the total mature M population was altered in function during H. hepaticus colitis. In steadystate intestine, resident Ms exhibit an anti-inflammatory phenotype characterized by constitutive production of IL-10 and low levels of TNF-␣, together with an inability to respond to stimuli such as TLR ligands, but it has been unclear whether these properties can change during inflammation (30). In contrast to their CX3CR1 hi counterparts, as we show here, cells within the CX3CR1 int compartment are actively proinflammatory, expressing much higher levels of mRNA for proinflammatory mediators such as IL-1␤, iNOS, IL-23, and IL-12 during H. hepaticus colitis than their control counterparts. In parallel, CX3CR1 hi Ms taken at the peak of inflammation showed no increased expression of IL-1␤ or IL-12 mRNA and only a modest increase in iNOS and IL-23 mRNA, suggesting some, but limited, ability to adopt a proinflammatory phenotype under these circumstances. In DSS colitis, we and others have shown that resident Ms do not acquire proinflammatory characteristics (6,12), but this has been reported in models of colitis where IL-10-mediated signaling is absent, with wider alterations in gene expression than we tested here (18). While this contrasting behavior in the presence and absence of IL-10 is yet to be confirmed in other models of inflammation, the results are consistent with recent evidence that IL-10 drives permanent, epigenetic silencing of proinflammatory genes (31). As a result, the absence of IL-10R-mediated signaling profoundly alters the genetic landscape of mature intestinal Ms, allowing them to respond to stimuli to which they are normally resistant (18). Whether IL-10 also controls other aspects of intestinal M differentiation is unclear, although it has been reported that their expression of the scavenger receptor CD163 is dependent on IL-10 (14). Arnold et al. showed reduced levels of CD206 on CX3CR1 hi Ms in H. hepaticus colitis (28), but it is not clear whether this reflects an intrinsic effect of IL-10 or is secondary to the effects of inflammation. That resident Ms retain a substantial proportion of their homeostatic properties in the absence of IL-10 is suggested by our finding that sustained expansion of this population is not accompanied by maintenance of H. hepaticus-dependent immunopathology. The colitis induced by H. hepaticus is dependent on IL-23 (22), and upregulation of this cytokine was one of the major changes that we and others observed in CX3CR1 int CD11b ϩ myeloid cells in this model (28). It has often been difficult to identify precisely the cell responsible for producing IL-23 in infection or inflammation, in great part because Ms and DCs share many phenotypic features such as CD11c, MHCII, CX3CR1, and CD11b. Therefore, it is important to note that by using rigorous gating strategies based on CD64 as a M marker, both we and Arnold et al. (28) show that IL-23 was derived from the M lineage, which is consistent with work in other models (7,32,33). Notably, however, we detected increased IL-12p35 and IL-23p19 transcript levels in elicited CX3CR1 int monocytes/Ms, suggesting that these cells produce both IL-12 and IL-23 during H. hepaticus-induced intestinal inflammation. Thus, we propose that the T H 17 response and subsequent phenotype switching of T H 17 cells into IFN-␥ ϩ IL-17A ϩ lymphocytes in H. hepaticus colitis (23) is driven by IL-23 derived from Ms, presumably activated directly by products of the organism. Nevertheless, it should be noted that both CD103 ϩ CD11b ϩ and CD103 Ϫ CD11b ϩ DCs from the intestine have been shown to be capable of driving T H 17 responses in vivo and in vitro, respectively, via production of IL-23 and/or IL-6 (26,(35)(36)(37). Thus, there may be flexibility between the M and DC lineages in their ability to stimulate T H 17 activity depending on the stimulus. Alternatively, these cell types may cooperate in driving such T-cell responses, perhaps with migratory DCs being responsible for priming T H 17 cells in the draining lymph node, while tissue-resident Ms may sustain the survival of these cells once they migrate back to the mucosa (38).
Together, our results show that elicited monocyte-derived CX3CR1 int macrophages form the predominant proinflammatory macrophage population in H. hepaticusinduced colitis. Given that these cells appear to derive from the same Ly6C hi blood monocytes as their homeostatic counterparts, identifying the factors that govern monocyte fate in the colon during homeostasis versus inflammation could provide new therapeutic targets for the treatment of IBD.

MATERIALS AND METHODS
Infection and antibody treatment of mice. Female C57BL/6 (B6) Il10 Ϫ/Ϫ , B6 WT, and B6 CD45.1 ϩ Cx3cr1 gfp/gfp mice (obtained from Jackson Laboratory) and B6 CD45.1 ϩ Cx3cr1 ϩ/gfp mice (generated by crossing Cx3cr1 gfp/gfp mice with B6 CD45.1 ϩ WT animals) were bred and maintained in an accredited specific-pathogen-free (SPF) facility, and the experiments were conducted in accordance with the UK Animals (Scientific Procedures) Act 1986 under a project license authorized by the UK Home Office and approved by the University of York Animal Welfare and Ethical Review Body. The mice tested negative for antibodies to specific murine viruses, were free of Helicobacter spp. as assessed by PCR, and were Ͼ6 weeks old when used.
LP cell isolation. For the experiments shown in Fig. 1 and 2, colon and cecum were pooled from individual mice and digested with Liberase Cl (0.42 mg/ml; Roche, Burgess Hill, UK) and DNase I (125 U/ml; Sigma-Aldrich, Gillingham, UK), followed by enrichment of lamina propria (LP) cells on 40/80% Percoll gradients as described previously (23). For the experiments shown in Fig. 3 to 5 and in the supplemental figures, colonic LP cells were isolated as described previously without the use of Percoll gradients (4,40).
Flow cytometric analysis and FACS. After blocking Fc receptors with anti-Fc␥RII/III MAb (2.4G2; BD Biosciences), LP cells were stained at 4°C with fluorochrome-conjugated antibodies (see Table S1 in the supplemental material) for 20 to 30 min in the dark before being washed in FACS buffer (2% fetal calf serum [FCS] and 1 mM EDTA in phosphate-buffered saline [PBS]) and then analyzed on a CyAn ADP (Beckman Coulter, High Wycombe, UK) or LSRII/FACSAria I (BD Bioscience) flow cytometer. Data were analyzed using FlowJo software (Tree Star Inc., OR, USA). Myeloid cell populations (as defined in figure legends) were sorted using a MoFlo Astrios or a FACSAria I sorter, with purities of Ͼ96%. For flow cytometric analyses on BD instruments, automatic compensation was performed in FACSDiva using UltraComp or OneComp beads together with fluorescence-minus-one controls.
Intracellular cytokine staining. Large intestinal LP cells (2 ϫ 10 6 /ml) from uninfected and 2-week H. hepaticus-infected Il10 Ϫ/Ϫ mice were cultured in medium alone or in the presence of 10 g/ml Pam3CSK4 (InvivoGen, Toulouse, France) for 4 h at 37°C with 10 g/ml brefeldin A during the last 3 h. Thereafter, cells were stained for surface markers (CD45 and CD11b) and intracellular TNF-␣ as described previously (23).
For cytokine RT-qPCR analysis in Fig. 2, FACS-purified CD45 ϩ CD11b ϩ cells were homogenized in TRIzol, and total RNA was isolated by chloroform extraction and reverse transcribed using SuperScript II and random hexamers. cDNA was amplified using SYBR green reagents and an ABI Prism RT-PCR system (Applied Biosystems). Cytokine expression levels for each individual sample (run in duplicates) were normalized to hypoxanthine phosphoribosyltransferase (HPRT) using threshold cycle (ΔC T ) calculations and the 7000 System SDS software (Applied Biosystems). For myeloid cell populations in Fig. 5, FACS-purified cells were lysed in RLT buffer (Qiagen) and homogenized using QIAshredders (Qiagen). RNA was then isolated and purified using an RNeasy microkit (Qiagen) and reverse transcribed using a high-capacity RNA-to-cDNA kit (Life Technologies) and random hexamers. cDNA was amplified using SYBR green reagents and an ABI 7900HT Prism RT-PCR system (Applied Biosystems). Cytokine expression levels for each individual sample (run in triplicates) were normalized to TATA-binding protein (TBP) using ΔC T calculations. Specific primer pairs are detailed in Table S2.
Statistical analysis. Multiple group comparisons were performed by one-way analysis of variance (ANOVA), while Student's t test and the Mann-Whitney test were used to compare two groups. Differences were considered statistically significant with a P value of Ͻ0.05.