Dendritic cells (DCs) and monocyte-derived macrophages (MΦs) are key components of intestinal immunity. However, the lack of surface markers differentiating MΦs from DCs has hampered understanding of their respective functions. Here, we demonstrate that, using CD64 expression, MΦs can be distinguished from DCs in the intestine of both mice and humans. On that basis, we revisit the phenotype of intestinal DCs in the absence of contaminating MΦs and we delineate a developmental pathway in the healthy intestine that leads from newly extravasated Ly-6Chi monocytes to intestinal MΦs. We determine how inflammation impacts this pathway and show that T cell-mediated colitis is associated with massive recruitment of monocytes to the intestine and the mesenteric lymph node (MLN). There, these monocytes differentiate into inflammatory MΦs endowed with phagocytic activity and the ability to produce inducible nitric oxide synthase. In the MLNs, inflammatory MΦs are located in the T-cell zone and trigger the induction of proinflammatory T cells. Finally, T cell-mediated colitis develops irrespective of intestinal DC migration, an unexpected finding supporting an important role for MLN-resident inflammatory MΦs in the etiology of T cell-mediated colitis.
The intestinal lamina propria (LP) contains cells that express high levels of CX3CR1, the receptor for the fractalkine chemokine [1, 2]. Based on their monocytic origin and on their inability to migrate to the mesenteric lymph nodes (MLNs) such CX3CR1hi cells have been defined as macrophages (MΦs) [1-3]. CX3CR1hi MΦs contribute to the intestinal LP homeostasis through the production of anti-inflammatory cytokines and the clearance of commensal bacteria that breach the epithelial barrier . In contrast, during intestinal inflammation, microenvironmental signals promote the differentiation of extravasated monocytes into proinflammatory MΦs with the ability to produce interleukin (IL)-12, IL-23, tumor necrosis (TNF)-α and inducible nitric oxide synthase (iNOS) [5-7]. However, little is known about the developmental trajectories that lead extravasated monocytes to either anti- or proinflammatory intestinal MΦs. This is primarily due to the fact that a surface marker permitting unequivocal identification of MΦs within the intestine and their distinction from dendritic cells (DCs) is lacking.
The interstitial DCs (Int-DCs) present throughout the LP derive from blood precursors known as pre-DCs . Under steady-state conditions, the Int-DCs found in the intestinal LP induce oral tolerance by carrying antigens originating from food or from harmless bacteria to the MLNs [8, 9]. The CD103+ Int-DCs found in the steady-state LP have the selective ability to express aldehyde dehydrogenase (ALDH) and thereby produce retinoic acid (RA). As a result, upon migration to MLNs they trigger the differentiation of naive CD4+ T cells specific for food and microbiota antigens into induced Foxp3+ regulatory T (iTreg) cells [10-13]. In contrast, the Int-DCs that develop in inflamed LP upon exposure to pathogens lose their capacity to generate iTreg cells and, upon migration to the MLNs, trigger the differentiation of naive, antigen-responsive CD4+ T cells into T helper type 1 (Th1) cells that are specific for the invading pathogen and produce mainly IFN-γ [9, 14].
We have previously reported that the monocyte-derived cells that are present in skeletal muscle express high levels of CD64, the high-affinity IgG receptor FcγR . By using the expression of CD64, we also succeeded in distinguishing MΦs from DCs in the LP of the large and small intestine and the MLNs of both mice and humans. We thus reinvestigated at a high resolution MF and DC development in the intestinal LP under healthy conditions and studied how it was affected by inflammation. In this process, we found that during T cell-mediated colitis a similar developmental pathway gives rise to inflammatory MΦs in both the intestinal LP and in the MLNs. Moreover, we made the highly unanticipated observation that T cell-mediated colitis unfolds in absence of a normal influx of LP-derived migratory DCs (Mig-DCs) and showed that the numerically dominant inflammatory MΦs that develop within the MLNs also contribute to the induction of proinflammatory CD4+ T cells during T cell-mediated colitis.
CD64 expression distinguishes MΦs from DCs
We sought to identify a surface marker that would allow the distinction of MΦs from DCs in the intestine and the MLNs. Using mice expressing an enhanced green fluorescent protein (EGFP) under the control of the gene coding for the CX3CR1 chemokine receptor (CX3CR1-EGFP mice; ), we found that all the CX3CR1hi MΦs of the LP of the small and large intestine expressed CD64 (Fig. 1A). Moreover, by combining CD64 and CD11c expression, we observed that the MHC class II (MHCII)-positive cells present in the LP of the small and large intestine could be readily separated into a CD11chiCD64− and a CD11c− to intCD64+ subset (Fig. 1B and C).
Ly-6Chi monocytes express CCR2, a chemokine receptor that promotes their egress from the bone marrow into the blood, and as a result CCR2-deficient (Ccr2−/−) mice show markedly reduced numbers of circulating Ly-6Chi monocytes and tissue MΦs . In contrast, DCs are not affected by CCR2 deficiency . Therefore, we used this differential CCR2 requirement to determine whether the presence of CD11chiCD64− and CD11c− to intCD64+ cells in the intestinal LP depended on CCR2 expression. Accordingly, mice coexpressing CD45.1 and CD45.2 were lethally irradiated and reconstituted with a 1 to 1 mixture of BM cells isolated from CD45.1+ wild-type (WT) mice and from CD45.2+ Ccr2−/− mice. These B6 (CD45.1) WT + B6 (CD45.2) Ccr2−/− → B6 (CD45.1-CD45.2) competitive chimeras were analyzed 8 weeks after BM transfer. As expected, Ly-6Chi blood monocytes were primarily composed of CD45.1+ Ccr2+ donor cells whereas neutrophils were comprised of almost equal percentages of CD45.1 and CD45.2 donor cells (Fig. 1E). CCR2-deficient and WT donor cells contributed equally to the CD11chiCD64− intestinal LP cells (Fig. 1B and C). The capacity to develop in a CCR2-independent manner combined with a specific absence in mice lacking the ligand for the Fms-like tyrosine kinase receptor 3 (Flt3L; Supporting Information Fig. 1A), strongly suggest that the CD11chiCD64− intestinal LP cells correspond to Int-DCs. In contrast, the CD11c− to intCD64+ intestinal LP cells showed percentages of CD45.1 and CD45.2 donor cells identical to those of Ly-6Chi blood monocytes (Fig. 1B and C). Consistent with the view that CD11c− to intCD64+ intestinal LP cells constitute monocyte-derived MΦs, they developed independently of Flt3L (Supporting Information Fig. 1A). A similar conclusion applied to the few CD11c− to intCD64+ cells found among the MHCII+ cells present in the MLNs of mice kept under specific pathogen-free conditions (Fig. 1D). Therefore, by combining CD64 and CD11c expression, it is possible to unequivocally distinguish CD11c− to intCD64+ MΦs from CD11chiCD64− DCs among the MHCII+ cells of the intestinal LP and of the MLNs without having to rely on the use of CX3CR1-EGFP reporter mice.
CD8α+-type and CD11b+-type intestinal DCs
Conventional DCs can be categorized into CD8α+- and CD11b+-type DCs [19, 20]. The CD11chiCD64− Int-DCs of the LP and their migratory counterparts present in the MLNs showed the same dichotomy and comprised CD8α+- and CD11b+-type DCs that can be unambiguously identified by their CD24+CD11b− and CD11b+CD64− phenotype, respectively (Supporting Information Fig. 1B). As expected, CD11b+-type DCs expressed CD172α (SIRPα) whereas CD8α+-type DCs were CD172α− (Supporting Information Fig. 1B). In contrast to CD64+ MΦs that lacked CD103 (data not shown), a substantial fraction of CD24+CD11b− and CD11b+CD64− DCs found in the LP and the MLNs expressed CD103 (Supporting Information Fig. 1C). Notably, CD24+CD11b− DCs were CX3CR1−, whereas CD11b+CD64− DCs were CX3CR1int (Supporting Information Fig. 1C), emphasizing that in the intestine CX3CR1 expression is not restricted to monocyte-derived cells (Fig. 1A). Likewise, F4/80 was expressed at high and intermediate levels on CD11c− to intCD64+ MΦs and CD11b+CD64− DCs, respectively (Supporting Information Fig. 1C). Therefore, by identifying MΦs on the basis of their CD11c− to intCD64+ phenotype and subsequently subdividing CD11chiCD64− DCs into CD24+CD11b− and CD11b+CD64− cells, it is possible to overcome several of the limitations previously encountered in the identification of MΦs and DC subsets in the intestinal LP and the MLNs.
CD64 marks MΦs in humans
To evaluate whether CD64 could be also used to identify intestinal MΦs in humans, we generated humanized mice and determined whether the HLA-DR+ cells found in their intestinal LP also contained CD64+ MΦs in addition to DCs. CD34+ human stem cells from cord-blood were transfected with GFP and then injected into NOD-scid-IL-2Rγ−/− (NSG) mice. Eleven weeks after reconstitution, the GFP+HLA-DR+ cells found in the intestinal LP could be readily divided into CD64− to low and CD64hi cells (Supporting Information Fig. 2A). The CD64− to low cells were composed of BDCA3+ and BDCA1+ cells, which represent the human equivalent of mouse CD8α+- and CD11b+-type DCs, respectively . In contrast, the CD64hi cells were low for both BDCA3 and BDCA1 and likely corresponded to LP MΦs (data not shown).
To support our conclusion that human intestinal MΦs expressed CD64, we analyzed colon biopsies obtained from patients suffering from ulcerative colitis. The HLA-DR+ LP cells found in healthy sections of the human colon contained BDCA3+ and BDCA1+ DCs as well as CD64hi cells that based on expression of CD14 likely corresponded to monocyte-derived cell (Supporting Information Fig. 2B). Interestingly, analysis of diseased tissue samples from the same patients showed a dramatic increase in the proportion of CD14+CD64hi MΦs among HLA-DR+ LP cells as compared with healthy tissue samples. Therefore, HLA-DR+ MΦs were CD64hi in both steady state and inflamed human colon whereas colonic DCs were CD64− to low.
Connecting recently extravasated Ly-6Chi monocytes to intestinal MΦs
The CD11b+ cells found in the LP of the small intestine comprised CX3CR1int and CX3CR1hi cells (Fig. 2A). In contrast to the CX3CR1hi cells that are only constituted of MΦs (Fig. 2B), the CX3CR1int cells were heterogeneous and contained CD11b+-type Int-DCs that can be readily gated out on the basis of their CD11chiCD64− phenotype (Fig. 2C). Analysis of the remaining, Int-DC-deprived, CX3CR1int cells on a Ly-6C-MHCII plot led to the identification of three populations (Fig. 2C). Population 1 (P1) resembled Ly-6Chi blood monocytes in terms of FSC-SSC profile (data not shown) and of its Ly-6Chi, CD64low, MHCII−, CD11c− to low, CX3CR1int, and CD11b+ phenotype (Fig. 2C and D). P1 cells in the LP can also be identified on the basis of their unique CD11b+Ly-6Chi phenotype (Fig. 2E). Population 2 (P2) was composed of Ly-6Cint to hiMHCII+ cells whereas population 3 (P3) had a Ly-6ClowMHCII+ phenotype identical to that of the CX3CR1high MΦs that are denoted as P4 cells (Fig. 2B and C). Note that the P3 and P4 cells can only be distinguished on the basis of their distinct levels of expression of the CX3CR1(EGFP) reporter (Fig. 2). Therefore, in WT mice, P3 and P4 cells cannot be distinguished and are considered below as a single Ly-6ClowMHCII+CD64+ P3/P4 population. The P2 cells “bridged” the diagonally opposite positions occupied by the P1 and the P3/P4 cells, a feature expected for developmental intermediates linking recently extravasated Ly-6ChiMHCII− monocytes to CX3CR1int and CX3CR1high MΦs.
When CD11b+ LP cells from WT mice were analyzed on a Ly-6C-CD64 plot they comprised Ly-6C−CD64− cells that were CCR2 independent and corresponded to CD11b+ Int-DCs (Fig. 3A). The remaining cells (red gate; Fig. 3A) strictly depended on CCR2 expression for their development and were thus of monocytic origin (Fig. 3A). They adopted a similar waterfall-shaped distribution on Ly-6C-CD64 and Ly-6C-MHCII plots (Fig. 3A), a feature consistent with the fact that CD64 and MHCII expression followed the same trend along the putative P1 → P2 → P3 → P4 developmental series. For the sake of brevity, the gate including the P1, P2, and P3/P4 cells is thus referred here as the “Mo-waterfall gate” and the waterfall-shaped distribution as the “Mo-waterfall” (Fig. 3B) A similar Mo-waterfall was also observed in the LP of the large intestine (Fig. 3C).
To validate the hypothesis that a precursor-product relationship exists between the P1, P2, and P3/P4 stages that composed the Mo-waterfall (specified by red arrows in Figure 3A, bottom panels), Ly-6Chi monocytes were labeled with CFSE and adoptively transferred into healthy Ccr2−/− mice. We used Ccr2−/− hosts because they have a selective reduction in circulating Ly-6Chi monocytes that should facilitate the engraftment of the transferred CFSE+ monocytes. Analysis of the CFSE+ donor cells found in the LP of the small intestine 12 h after transfer showed some increase in MHCII and CD64 expression (Fig. 3D). By 72 h after transfer, most donor cells expressed both MHCII and CD64 and had downregulated Ly-6C expression. A small fraction of the Ly-6ClowMHCII+ CD64+ donor cells also expressed CD11c. In contrast, when the CFSE-labeled Ly-6Chi monocytes were kept in culture, they did not change their original Ly-6ChiCD64lowMHCII− phenotype upon time (Fig. 3E). These data demonstrate that upon entry into the intestinal LP, Ly-6Chi monocytes sequentially give rise to the P1, P2, and P3/P4 populations and that this progression was accompanied by the upregulation of CX3CR1, MHCII, CD64, and CD11c molecules and by the downregulation of Ly-6C (Fig. 3A and B). Moreover, the cells traversing this developmental series maintained CD172α expression and, importantly for the rest of this study, did not express CCR7 (Supporting information Fig. 3B and C). Therefore, by combining CD64 expression with Ly-6C and MHCII expression it is possible to track the progressive differentiation of newly extravasated Ly-6Chi monocytes into MΦs within healthy intestinal LP (Supporting Information Fig. 3).
T cell-mediated colitis induces a massive differentiation of Ly-6Chi monocytes into MΦs
To analyze the effect of inflammation on the dynamics of MF and DC differentiation in the intestinal LP and the MLNs, we used a model of T cell-mediated colitis that relied on Cd3eΔ5/Δ5 mice that lack T cells but not B cells . Transfer of CD4+Foxp3− T cells into Cd3eΔ5/Δ5 mice allows their extensive proliferation and differentiation into effector CD4+ T cells that produce mainly IFN-γ and that home to the large intestine where they trigger a severe colonic inflammation, typified by weight loss, diarrhea, and rectal prolapse (Supporting Information Fig. 4; ). Using this model, we first analyzed the evolution of MF and DC populations in the LP during the development of T cell-mediated colitis. Under steady-state conditions, WT and Cd3eΔ5/Δ5 mice contained similar large numbers of CD64+ MΦs in the LP of the large intestine (Fig. 4A and data not shown). Upon transfer of CD4+ Foxp3− T cells into Cd3eΔ5/Δ5 mice, the numbers of CD64+ MΦs increased in a rapid and massive manner in the LP of the large intestine (Fig. 4A and B and Supporting Information Fig. 5A). The numbers of CD8α+- and CD11b+-type Int-DCs also increased although to a lesser extent (Fig. 4A and B). Therefore, as observed in patients suffering from ulcerative colitis (Supporting Information Fig. 2B), the colitis observed in Cd3eΔ5/Δ5 mice is also associated with a large increase in the proportion of CD64+ MΦs among MHCII+ LP cells.
The Mo-waterfall found in the inflamed LP differed from the one observed under healthy conditions in that it consisted primarily of Ly-6ChiMHCII− (P1) and Ly-6Cint to hiMHCII+ (P2) cells and yielded comparably few Ly-6ClowMHCII+ (P3/P4) cells (compare Fig. 3C and Fig. 4C). Consistent with this last observation, adoptive transfer of CFSE-labeled Ly-6Chi BM monocytes into colitic Cd3eΔ5/Δ5 × Ccr2−/− mice gave rise to CD64+ LP MΦs that had a predominant Ly-6Cint to hiMHCII+ phenotype (Fig. 4E). The same conclusion can be reached after marking the Ly-6Chi blood monocytes present in Cd3eΔ5/Δ5 mice undergoing T cell-mediated colitis via intravenous injection of latex beads . Four days after injection of latex beads, the bead-labeled cells showed a blunted waterfall-shaped distribution (Fig. 4D). Therefore, in mice undergoing colitis, blood monocytes are constantly recruited to the large intestine LP where they predominantly differentiate into Ly-6Cint to hiMHCII+CD64+ MΦs.
Colitis is associated with the differentiation of Ly-6Chi monocytes into MΦs in the MLNs
Unexpectedly, colitic Cd3eΔ5/Δ5 mice also displayed a rapid accumulation of CD64+ MΦs in the MLNs (Fig. 5 A and B and Supporting Information Fig. 5B). To determine whether these MΦs corresponded to LP-derived migratory MΦs, we determined if they expressed CCR7, a chemokine receptor required for the migration of intestinal DCs to the MLNs . In contrast to the LP-derived Mig-DCs, the MΦs found in the MLNs of colitic Cd3eΔ5/Δ5 mice were CCR7− (Fig. 5C). It is thus likely that the MΦs that accumulate in the MLNs of colitic Cd3eΔ5/Δ5 mice do not represent the migratory counterpart of intestinal MΦs and originate from blood monocytes that seeded the MLNs and differentiated in situ. Consistent with this view, after excluding CD11b+-type DCs, the remaining CD11b+ monocytic cells present in the MLNs of colitic Cd3eΔ5/Δ5 mice showed a waterfall-shaped distribution that consisted primarily of Ly-6ChiMHCII− and Ly-6Cint to hiMHCII+ cells and that resembled the blunted Mo-waterfall found in the LP of colitic mice (Fig. 5D). Moreover, the Ly-6ChiMHCII− cells had an FSC-SSC profile typical of undifferentiated blood monocytes and their progression through the Mo-waterfall was associated with an increase in both cell size and granularity (Fig. 5D). Consistent with this last observation, adoptive transfer of CFSE-labeled Ly-6Chi monocytes into colitic Cd3eΔ5/Δ5Ccr2−/− mice gave rise in the MLNs to CD64+ MΦs that were predominantly Ly-6Cint to hiMHCII+ (Fig. 5E). Importantly, in both the LP and the MLNs, the differentiation of adoptively transferred Ly-6Chi monocytes into CD64+ MΦs followed the same kinetics (compare Fig. 4E and 5E). In the case the MΦs found in the MLNs of colitic Cd3eΔ5/Δ5 mice will have represented migratory counterpart of intestinal MΦs, a protracted kinetic of reconstitution would have been expected in the MLNs as compared with that in the LP. Therefore, in Cd3eΔ5/Δ5 mice undergoing T cell-mediated colitis, Ly-6Chi blood monocytes are massively recruited to the MLNs where they differentiate into MΦs. The fact that they do not represent the migratory counterpart of intestinal MΦs is congruent with previous studies demonstrating that intestinal MΦs do not migrate to the MLNs under healthy or TLR-induced inflammatory conditions .
Ly-6Chi monocytes differentiate into iNOS-expressing inflammatory MΦs during T cell-mediated colitis
A fraction of the CD64+ MΦs that developed in the MLNs of colitic Cd3eΔ5/Δ5 mice that had received CD4+Foxp3− T cells became iNOS+ (Fig. 6A and B). iNOS+ MΦs also developed in the LP of the large intestine, but with slower kinetics compared with MLNs (Fig. 6A and B). The majority of iNOS+ MΦs present in the large intestine LP and the MLNs were Ly-6Cint to hiMHCII+ (Fig. 6C). In contrast, DCs remained iNOS− at all time points (Fig. 6A). Therefore, during T cell-mediated colitis, a fraction of the blood monocytes that are recruited to the large intestine LP and the MLNs differentiate into iNOS-expressing inflammatory CD64+ MΦs.
T cell-mediated colitis develops in mice with impaired migration of DCs to the MLNs
To evaluate the requirement of LP-derived Mig-DCs in the development of T cell-mediated colitis, we generated Cd3eΔ5/Δ5 × Ccr7−/− mice in which the CCR7-dependent migration of Int-DCs from the intestinal LP to the MLNs is impaired . Cd3eΔ5/Δ5 × Ccr7−/− mice and Cd3eΔ5/Δ5 control mice were reconstituted with CD4+Foxp3− T cells isolated from CCR7-sufficient mice to avoid adventitious effects due to the lack of CCR7 on T cells, and analyzed 4 weeks after transfer. As expected, while both types of mice had comparable numbers of Int-DCs in the LP of the large intestine (Fig. 7A), the MLNs of Cd3eΔ5/Δ5 × Ccr7−/− mice contained numbers of CD8α+-type and CD11b+-type Mig-DCs that were reduced fourfold and sevenfold respectively, as compared with that in the Cd3eΔ5/Δ5 mice (Fig. 7A). In contrast, the numbers of MLN-resident MΦs were comparable in both types of mice (Fig. 7A). Unexpectedly, preventing a normal influx of LP-derived Mig-DCs resulted in exacerbated T cell-mediated colitis as shown by the increased percentage of IFN-γ-producing effector T cells in the large intestine and the MLNs (Fig. 7B and C). This amplified inflammatory response was also associated with a more rapid weight loss (Fig. 7D). Therefore, in the Cd3eΔ5/Δ5model, T cell-mediated colitis unfolds irrespective of a normal influx of LP-derived Mig-DCs.
MLN-resident inflammatory MΦs induce IFN-γ-producing effector T cells in vitro
We next analyzed whether the numerically dominant inflammatory MΦs that develop within the MLNs can participate in the induction of colitogenic T cells. First, we assessed whether they were located within the T-cell zone of the MLNs, a prerequisite for the activation of naive CD4+ T cells. Although CD64 expression would have constituted the best way of localizing MLN-resident inflammatory MΦs, our fixation protocol prevented the use of this marker. Considering that iNOS+ inflammatory MΦs expressed high levels of CD11c and that iNOS expression is restrained to these cells in the MLNs of colitic mice (Fig. 6A and B), we relied on iNOS and CD11c coexpression to locate MLN-resident inflammatory MΦs. Confocal microscopy showed that iNOS+CD11c+ inflammatory MΦs were primarily found within the T-cell zone of the MLNs of colitic Cd3eΔ5/Δ5 mice (Fig. 6D). Moreover, MLN-resident inflammatory MΦs had a high phagocytic activity and lacked a DC-morphology (Supporting Information Fig. 6).
Next, we assessed whether the inflammatory MΦs that develop in the MLNs of colitic Cd3eΔ5/Δ5 mice were capable of converting naïve CD4+ T cells into effector T cells in vitro and we compared them with the LP-derived Mig-DCs that are also found in colitic MLNs. Considering that colitic MLNs still contains some tolerogenic ALDH+ Mig-DCs (data not shown) that may have led us to underestimate the IFN-γ-producing capacity of ALDH− Mig-DCs, LP-derived Mig-DCs were sorted into ALDH+ and ALDH− cells. The various sorted cell types were pulsed with an ovalbumin (OVA)-derived peptide and cocultured with OVA-specific OT-II CD4+ T cells. After 5 days of culture, OT-II T cells were analyzed for the production of IFN-γ and IL-17 (Fig. 6E). Surprisingly, among the analyzed MLN cells, inflammatory MΦs, together with ALDH− CD8α+-type Mig-DCs, possessed the strongest capacity to induce IFN-γ production by naive OT-II T cells. Therefore, the numerically dominant inflammatory MΦs that develop in the MLNs of colitic mice were located in the T-cell zone and excelled in the generation of Th1 effectors in vitro.
MLN-resident MΦs from colitic mice induce IFN-γ-producing effector T cells in vivo
To determine whether MLN-resident inflammatory MΦs were capable of triggering the differentiation of CD4+ T cells into IFN-γ-producing effector T cells in vivo, we developed the adoptive transfer experiment outlined in Fig. 8A. Consistent with previous data , when CD4+ T cells were labeled with CFSE and transferred into Cd3eΔ5/Δ5 host, two coincident and independent phenomena occurred . A fraction of the transferred CD4+ T cells proliferated very slowly, accounting for the CFSEhigh cells observed at day 8 after transfer (Fig. 8B, upper panels). In contrast, a fraction of them proliferated extensively, yielding a prominent population of CFSElow cells. These fast-proliferating, CFSElow CD4+ T cells were capable of producing IFN-γ and their generation was strictly dependent on interaction with MHCII molecules expressed on the surface of the antigen-presenting cells of the host (Fig. 8B, middle panels). It is those fast-proliferating T cells that are responsible for the induction of colitis when transferred in the absence of Treg cells. Antigenic peptides derived from the enteric bacteria and presented by MHCII molecules have been shown to be responsible for this TCR-driven fast proliferation whereas the slow proliferation is independent of MHCII molecules and IL-7-driven [26-30].
We took advantage of this model to determine whether upon transfer into Cd3eΔ5/Δ5 x MHCIIΔ/Δ mice, MLN-resident inflammatory MΦs isolated from MHCII-sufficient, colitic mice were capable of rescuing the fast proliferation of cotransferred CD4+ T cells and their differentiation into IFN-γ-producing effectors. As shown in Fig. 8B, transfer of CD64+ MΦs from MLNs of colitic mice was sufficient to trigger the fast proliferation of a fraction of the cotransferred CD4+ T cells and their differentiation into IFN-γ-producing effectors. The fact that we could only transfer relatively small numbers of CD64+ MΦs per mouse likely explained that the magnitude of the CD4+ T-cell responses was smaller than that observed after injecting T cells into MHCII-sufficient Cd3eΔ5/Δ5 mice. Regardless of this limitation, our data demonstrate that inflammatory CD64+ MΦs isolated from the MLNs of colitic mice can induce the proliferation and differentiation of CD4+ T cells into IFN-γ-producing effector T cells in vivo.
We have shown that in healthy and inflamed intestinal LP and MLNs, CD64 expression makes it possible to accurately identify MΦs and to delineate a developmental series — called the Mo-waterfall — that starts from extravasated Ly-6Chi monocytes and leads to intestinal MΦs. Using Ly-6C-MHCII and Ly-6C-CD64 plots, the corresponding stages of this monocyte-derived developmental series have been denoted as P1 (Ly-6ChiMHCII−CD64lowCX3CR1int LP monocytes), P2 (Ly-6Cint to hiMHCII+CD64lowCX3CR1int intermediate cells), P3 (Ly-6ClowMHCII+CD64+CX3CR1int MF), and P4 (Ly-6ClowMHCII+CD64+CX3CR1hi MF) (Supporting Information Fig. 3). Interestingly, we found that in humanized mice and humans, CD64 was highly expressed by HLA-DR+ cells that coexpressed CD14; a finding strongly suggesting that high CD64 expression marks cells of monocytic origin in humans. In support of this conclusion, a recent study demonstrated that the CD14hi and CD14lo subsets of HLA-DR+ MΦs were uniformly CD64+ in both healthy and inflamed human intestine, whereas MHCII+CD14− cells that include DCs were mostly CD64− . Therefore, CD64 can be used in both mice and human to distinguish MΦs from CD8α+/BDCA-3+-type and CD11b+/BDCA-1+-type DCs, and to study how MΦs originate from extravasated Ly-6Chi monocytes in healthy and inflamed conditions (Fig. 9). Although monocyte-derived MHCII+ cells found in the intestinal LP have been subdivided into CD11c− MΦs and CD11c+ monocyte-derived DCs (Mo-DCs) by some authors [, , ], CD11c shows a continuous density distribution from negative to low among monocyte-derived MHCII+ LP cells (Fig. 2 and 3), a finding emphasizing the difficulty in using CD11b to define discrete stages amongst monocyte-derived MHCII+ LP cells.
During T cell-mediated colitis, extravasated Ly-6Chi monocytes differentiate into CD64+MHCII+ inflammatory MΦs that are CCR7− and, consistent with results from Pabst et al. , cannot migrate to the MLNs. This characteristic prevents their participation in the activation of naïve T cells in the MLNs. However, giving their high MHCII expression, inflammatory MΦs can potentially present antigens to tissue-resident memory T cells and to the iTreg cells that reach the LP [32, 34]. Along that line, due to its high affinity for IgG, the expression of CD64 may possibly arm LP MΦs with IgG specific for previously encountered pathogens and allow them to efficiently initiate recall responses upon subsequent pathogen reexposure . Although the inflammatory conditions achieved during T cell-mediated colitis do not induce LP MΦs to migrate to MLNs, it remains possible that some bacterial adjuvants confer MLN-homing property to some LP MΦs as shown for muscle Mo-DCs .
Unexpectedly, T cell-mediated colitis was also found associated with a massive recruitment of blood Ly-6Chi monocytes in the MLNs, where they differentiated into inflammatory MΦs via a blunted Mo-waterfall similar to the one observed in the LP of inflamed intestine. These MLN-resident inflammatory MΦs did not display a dendritic morphology and were phagocytic, a hallmark of MΦs. In addition, they expressed high levels of MHCII molecules and were located in the T-cell zone, which are two prerequisites for proper interactions with naive CD4+ T cells. Consistent with these attributes, MLN-resident inflammatory MΦs were capable of generating IFN-γ-producing T cells in vitro and in vivo. Therefore, during T cell-mediated colitis, MLN-resident inflammatory MΦs likely synergize with LP-derived ALDH− Mig-DCs to induce the differentiation of naïve T cells into proinflammatory T cells.
Upon transfer of CD4+Foxp3− T cells, we showed that Cd3eΔ5/Δ5 × Ccr7−/− recipient mice developed an exacerbated colitis as compared with Cd3eΔ5/Δ5 recipient mice. This result is consistent with our hypothesis that MLN-resident inflammatory MΦs actively participate in the induction of T cell-mediated colitis irrespective of the presence of Mig-DCs and with the view that the RA-producing ALDH+ LP-derived Mig-DCs that are still present in colitic mice contribute to dampen colitic disease (Fig. 6E). It is important to stress that our results do not question a role for ALDH− LP-derived Mig-DCs in the generation of proinflammatory T cells during colitis. However, they differ from previous studies in that they reveal that the inflammatory MΦs that expand massively in the MLNs of colitic mice and outnumber Mig-DCs are also capable of inducing proinflammatory CD4+ T cells. It remains to be determined how the inflammatory MΦs that permanently reside in the MLNs gather enteric antigens in the absence of LP-derived Mig-DCs. Akin to the situation recently described for mice depleted of innate lymphoid cells , the lack of T cells in Cd3eΔ5/Δ5 × Ccr7−/− mice may favor the dissemination of commensal bacteria to the MLNs and their capture by the resident MΦs.
In conclusion, we have defined a novel gating strategy that is based on the differential expression of CD64 and that allows the unequivocal identification of MΦs in the intestine of mice and humans in both healthy and inflamed conditions (Fig. 9). On that basis, we demonstrated that T cell-mediated colitis is associated with a massive recruitment of monocytes to the intestinal LP where they differentiated into LP-resident inflammatory MΦs. Coincidentally, T cell-mediated colitis also induced the massive recruitment of monocytes to the MLNs and their differentiation into MLN-resident inflammatory MΦs capable of converting naïve T cells into effector Th1 cells. In the case functionally similar inflammatory MΦs develop in the MLNs of patients suffering from inflammatory bowel disease, they might constitute potential targets for the management of inflammatory digestive diseases.
Materials and methods
CX3CR1-EGFP , Foxp3-EGFP , Cd3eΔ5/Δ5 , OT-II , Ccr2−/− , CCR7−/− , and Cd3eΔ5/Δ5 × MHCIIΔ/Δ mice  were maintained on a C57BL/6 background and under specific pathogen-free conditions. All experiments were done in accordance with French and European guidelines for animal care.
DCs were isolated from the MLNs and the LP of the small and large intestine as described in .
Cells were stained and analyzed using a FACS LSRII system (BD Biosciences). Allophycocyanin-Cy7-conjugated anti-NK1.1 (PK136), anti-CD3 (17A2), anti-Ly-6G (1A8), anti-CD19 (6D5), and PE-conjugated anti-CD64 (X54–5/7.1) were all from Biolegend, allophycocyanin-conjugated anti-CCR3 (83103) was from R&D, PE-Cy7-conjugated anti-CD11c (N418), Alexa-700-conjugated anti-MHC Class II (I-A/I-E) (M5/114.15.2), PE-Cy5.5-conjugated anti-CD45.2 (104), allophycocyanin-conjugated anti-CD45.1 (A20), PE-Cy5-conjugated anti-CD24 (M1/69), and PE-Cy5-conjugated anti-CD5 (53–7.3) were all from eBioscience, Pacific-Blue-conjugated anti-CD11b (M1/70), PE- or biotin-conjugated anti-CD103 (M290), allophycocyanin-conjugated anti-CD172α (P84), FITC-conjugated anti-Ly6C (AL21), Pacific-Blue-conjugated anti-CD4 (RM4–5), biotin-conjugated anti-CCR7 (4B12) were all from BD Pharmingen. Biotin-conjugated antibodies were detected using streptavidin conjugated with Quantum-Dot605 (Invitrogen). Intracellular iNOS staining was performed using unlabeled anti-rabbit iNOS (M19, Santa Cruz) and allophycocyanin-conjugated anti-rabbit IgG (Invitrogen). ALDH+ cells were identified as described , and intracellular cytokines staining performed as described . Prior to DC and MF analysis, B cells, T cells, NK cells, eosinophils and neutrophils were systematically gated out using a “dump-channel” corresponding to cells positive for CD19, CD3, NK1.1, CCR3 or Ly-6G cells. Analysis was performed using FlowJo software (Tree Star, Inc.).
Assessment of phagocytosis
FACS-sorted DC and MF subsets were seeded on Alcian blue-treated coverslips and incubated with 5% FCS supplemented RPMI containing yellow green fluorescent 0.5 μm Fluoresbrite microspheres (Polysciences, Inc.) for 1 h at 37°C before washing, fixation, staining for MHCII and iNOS and analysis by confocal microscopy.
Generation of BM chimeras
Seven to –8-week-old B6 CD45.1 × CD45.2 mice were lethally irradiated with two doses of 550 rads, 5 h apart, and then injected i.v. with 2 × 106 BM cells. BM cells were obtained from femurs and tibias of WT B6 CD45.1 or of Ccr2−/− CD45.2 mice.
T cell-mediated colitis
T cells were purified by magnetic separation from spleens and LNs of Foxp3-EGFP mice using a CD4 negative isolation kit (Dynal). Cells were then stained with PE-conjugated anti-CD4 (RM4–5) antibody and CD4+Foxp3− T cells (2 × 106) were sorted and injected intravenously into Cd3eΔ5/Δ5 mice.
In vivo labeling of Ly-6Chi monocytes
Blood monocytes were labeled with latex beads as described .
Adoptive transfer of Ly-6ChiMHCII− monocytes
Bone marrow and spleens from RAG−/− mice were harvested. Cells were stained for CD11b, Ly6C, Ly6G, MHCII, CCR3, and CD11c. Ly-6Chi monocytes were sorted as CD11b+Ly-6G−CCR3−CD11c−MHCII−Ly-6Chi cells with a purity of more than 95% and labeled with CFSE (Molecular Probes). A 2 × 106 CFSE-labeled Ly-6Chi monocytes were transferred intravenously into healthy Ccr2−/− mice or Cd3eΔ5/Δ5 × Ccr2−/− colitic mice. In parallel 2 × 104 CFSE-labeled Ly-6Chi monocytes were maintained in vitro in complete medium in the presence of 1 × 106 WT BM cells.
Adoptive transfer of inflammatory CD64+ MΦs
Cd3eΔ5/Δ5 mice were injected i.v. with 2 × 106 CD4+Foxp3− T cells to induce colitis. Two weeks later, when colitis was largely installed, inflammatory CD64+ MΦs were isolated from the MLNs. Cd3eΔ5/Δ5 × MHCIIΔ/Δ hosts were first injected i.v. with 2 × 106 inflammatory CD64+ MΦs and then, 1 day later, with 2 × 106 CFSE-labeled CD4+ T cells. In parallel experiments, 2 × 106 CFSE-labeled CD4+ T cells were injected into Cd3eΔ5/Δ5 and Cd3eΔ5/Δ5 × MHCIIΔ/Δ. Eight days after T-cell transfer, CD4+ T-cell proliferation and cytokine production were evaluated.
Preparation of CFSE-labeled T cells
OT-II T cells were purified by magnetic separation from pooled spleen and LN using a CD4 negative isolation kit (Dynal). For CFSE labeling, purified cells were resuspended in PBS and labeled with 2.5 μM 5- and 6-carboxy-fluorescein diacetate succinimidyl ester (CFSE) (Molecular Probes) for 3 min at room temperature.
Culture of CFSE-labeled OT-II T cells
3 × 103 DCs or MΦs were cocultured with 2 × 104 CFSE-labeled OT-II T cells in 200 μL in the presence of ovalbumin (323–339) peptide (0.06 μg/mL). After 5 days of culture, cells were restimulated with PMA and ionomycin for 5 h before intracellular cytokine staining. Proliferation was measured by loss of CFSE staining.
MLNs were fixed with 3.2% paraformaldehyde for 1 h, washed in PBS, infused overnight in 35% sucrose, and frozen in Tissue-Tek OCT compound (Electron Microscopy Sciences, Hatfield, PA). After permeabilization for 5 min in PBS containing 0.5% saponin, 2% bovine serum albumin, 1% fetal calf serum, and blockade of Fc binding with 1% goat serum for 30 min, 12 μm cryostat tissue sections were labeled overnight at 4°C with rabbit anti-iNOS (Santa Cruz Biotechnology), hamster anti-CD11c (N418, Biolegend) and rat anti-CD45R (RA3–6B2, Biolegend) antibodies or control antibodies followed by incubation for 1 h at room temperature with secondary antibodies and SYTOX Blue for nuclei staining. Slides were mounted in Prolong Gold (Invitrogen) and observed with a Zeiss LSM 510 or a Zeiss LSM 780 confocal microscope (Carl Zeiss, Jena, Germany).
Characterization of DCs and MΦs from human colon biopsies
Colon biopsy specimens were obtained from patients with a history of ulcerative colitis during colonoscopy after informed consent and with the approval of the Ethics Committee of Erasmus University Medical Centre, Rotterdam. Endoscopic signs of inflammation were used to define the biopsies as inflamed. LP mononuclear cells were isolated from the biopsies as previously described .
Humanized mice experiments were developed and analyzed as described .
Comparative experiments were tested for statistical significance using the unpaired Student's t-test in GraphPad Prism software (version 4.0; GraphPad).
We thank M. Malissen, H. Luche, B. Lucas, G. Randolph, and L. Leserman for discussions. We thank M. Barad, P. Grenot, and A. Zouine for assistance with cell sorting and M. Fallet and M. Barad for assistance with confocal microscopy. This work was supported by CNRS, INSERM, European Communities Framework Program 7 (MASTERSWITCH Integrating Project; HEALTH-F2–2008-223404 and NANOASIT Euronanomed Project), AFM, FRM, ANR (Skin DCs) and by doctoral and postdoctoral fellowships from Ministère de la Recherche (S.T.), and a Marie Curie Fellowship from the European Communities (M.G., project number 237109). Y.R., C.R.E.S., and D.B. have funding from the Cancer Research UK. Y.R. is supported by Leukaemia and Lymphoma Research. L.F.P. is supported by a CNRS/INSERM ATIP/AVENIR program. A.M.M. and C.C.B. were supported by the Medical Research Council and Wellcome Trust (UK).
Conflict of interests
The authors declare no financial or commercial conflict of interest.