The origin and functional specialization of dermal macrophages in cutaneous infections have been little studied. In this paper, we show that a strain of Leishmania major (L. major Seidman [LmSd]) that produces nonhealing cutaneous lesions in conventionally resistant C57BL/6 mice was more efficiently taken up by M2-polarized bone marrow (BM)–derived macrophages (BMDMs) in vitro and by mannose receptor (MR)hi dermal macrophages in vivo compared with a healing strain (L. major Friedlin V1). Both in steady and in T helper type 1 (Th1) cell–driven inflammatory states, the MRhi dermal macrophages showed M2 characteristics. The dermal macrophages were radio resistant and not replaced by monocytes or adult BM-derived cells during infection, but were locally maintained by IL-4 and IL-10. Notably, the favored infection of M2 BMDMs by LmSd in vitro was MR dependent, and genetic deletion of MR or selective depletion of MRhi dermal macrophages by anti–CSF-1 receptor antibody reversed the nonhealing phenotype. We conclude that embryonic-derived, MRhi dermal macrophages are permissive for parasite growth even in a strong Th1-immune environment, and the preferential infection of these cells plays a crucial role in the severity of cutaneous disease.
The control of intracellular pathogens such as Leishmania major can be understood in the context of macrophages found in two broadly distinctive activation states, termed M1 and M2 (Gordon, 2003). M1 macrophages are induced by IFN-γ and microbial stimuli and have enhanced antimicrobial capacity, whereas M2 macrophages are permissive to intracellular microbial growth but can mediate type 2 immunity against helminth infection, as well as contribute to tissue repair. Recent studies have emphasized the plasticity of tissue macrophages with regard to their activation states and have suggested that their ontogeny and tissue-derived signals shape their functional specialization (Gautier et al., 2012; Lavin et al., 2014). As most studies have focused on steady-state conditions or sterile tissue injury, the question of whether tissue macrophages can be reprogrammed in infection-driven inflammatory settings has only rarely been addressed. Recently, in a sequential infection model involving nematodes and bacteria, reprogramming of the peritoneum-resident macrophage appeared limited compared with newly recruited monocyte-derived macrophages, suggesting that the origin of macrophages plays an important role in their functional adaptation (Rückerl et al., 2017). Little is known regarding the plasticity of dermis-resident macrophages and their relative contributions to antimicrobial immunity or to pathology in cutaneous infection.
We have described a model of nonhealing cutaneous leishmaniasis in C57BL/6 mice infected with the L. major Seidman (LmSd) strain that was isolated from a patient with nonhealing cutaneous lesions (Anderson et al., 2005). Paradoxically, the nonhealing cutaneous infections occur within a strong T helper type 1 (Th1) cell setting that reflects the immunological conditions associated with many chronic forms of cutaneous leishmaniasis in humans (Pirmez et al., 1993; Louzir et al., 1998). Multiple factors, including IL-10, IL-1β, and inflammasome activation, contribute to the pathogenesis of nonhealing infection with LmSd (Anderson et al., 2005; Charmoy et al., 2016). To date, however, it has not been possible to explain how these factors act in concert to promote a nonhealing phenotype in such a strong Th1 environment.
In the experiments reported here, we have identified a population of M2-like dermal macrophages that are present under steady-state conditions and that are preferentially infected by the LmSd strain in a mannose receptor (MR)–dependent fashion to promote nonhealing cutaneous disease. The dermal macrophages are not replaced by blood precursors during infection, but are locally maintained by IL-4 and IL-10 and retain M2 functionality despite the high levels of IFN-γ produced in the site. So far as we are aware, this is the first demonstration that the severity of cutaneous infection can be linked to the preferential targeting of dermis-resident macrophages.
MR mediates preferential uptake of nonhealing L. major strains by bone marrow (BM)–derived macrophages (BMDMs) in vitro
As previously described (Charmoy et al., 2016), infection of C57BL/6 mice with a low dose of 1,000 LmSd metacyclic promastigotes results in a nonhealing lesion that eventually ulcerates, leading to complete erosion of the ear dermis (Fig. S1). In contrast, the L. major Friedlin V1 (LmFn) strain produces a chronic, nonulcerative, nodular lesion that eventually heals. To determine whether the differences in clinical outcome might be reflected in their interactions with innate cells, the L. major strains were compared for infection and replication in M-CSF–induced BMDMs in vitro (Fig. 1 A). Metacyclic promastigotes of LmSd showed an approximately twofold higher infection than LmFn at 5 h postinfection (p.i.), which was maintained during 3 d of culture. When calculated as the mean number of parasites per infected cell, LmSd and LmFn replicated similarly. The amastigote stage of LmSd was also more efficiently taken up than LmFn amastigotes and with a more rapid kinetic compared with metacyclic promastigotes of either strain (Fig. 1 B). Their intracellular replication rates were again similar. The association between the differences in L. major uptake by BMDMs in vitro and the infection outcomes in vivo was reinforced by the analysis of a series of genetic hybrids (LmFnSd) generated from sexual crosses between LmFn and LmSd (Inbar et al., 2013). The progeny clones inherited either the healing or nonhealing parental phenotypes (Fig. 1, C and D), and there was a precise correlation of the nonhealing hybrids with their more efficient uptake by BMDMs (Fig. 1 E).
Various receptors can mediate phagocytosis of promastigotes and amastigotes by macrophages, including MR, fibronectin receptor, or complement receptor 3 (CR3; Ueno and Wilson, 2012). To address the possible role of either MR or fibronectin receptor in the increased infection by LmSd, we treated the BMDMs with BSA-mannose or RGDS (H-Arg-Gly-Asp-Ser-OH tetrapeptide found on the integrin-binding domain of fibronectin), respectively, before infection. Using metacyclic promastigotes, we observed a dose-dependent reduction of LmSd infection in BMDMs treated with BSA-mannose, whereas no effect on infection by LmFn was observed (Fig. 1 F). There was no effect of RGDS treatment on either strain. The inhibition of LmSd infection in BSA-mannose–treated BMDMs was also observed using amastigotes (Fig. 1 G). CR3- and C3-deficient BMDMs showed no reduction, but rather a slight increase in the favored uptake of both LmFn and LmSd, which maintained a twofold difference between strains (Fig. 1 H). Finally, LmSd uptake by BMDMs from MR-deficient mice (mrc1−/−) was significantly reduced, whereas uptake of LmFn was not affected (Fig. 1 I).
MR is well described as a marker of M2 macrophages (Murray, 2017). M-CSF–differentiated BMDMs expressed higher levels of MR in comparison to GM-CSF–induced BM dendritic cells (DCs; Fig. 1 J). Subsequent treatment with M1 stimuli (LPS or IFN-γ plus LPS) decreased MR expression in BMDMs, whereas treatment with IL-10, and especially with IL-4 plus IL-10, though not with IL-4 and/or IL-13 alone, enhanced MR expression (Fig. 1 K). Accordingly, the favored uptake of LmSd versus LmFn was lost when BMDMs were converted to M1 macrophages, and both strains were killed after 3 d of infection (Fig. 1 L). In contrast, stronger M2 polarization after stimulation with IL-4/-10 resulted in a pronounced difference in infection by LmSd versus LmFn, whereas both strains survived and replicated in these cells. Collectively, these results indicate that the higher infectivity of LmSd is caused by an MR-dependent uptake of parasites that promotes preferential infection of M2 BMDMs in vitro.
Identification and functional characterization of MRhi macrophages in the steady-state dermis
To identify cells expressing MR in the skin, we analyzed ear cells from naive mice (Fig. 2 A). Staining for Ly6C versus MR on CD11b+ cells that were negative for several lineage markers identified four distinct populations, designated P1–P4. High expression of MR was confined to the P4 population defined as Ly6CintMRhi, which comprises 30% of CD11b+ cells in the steady-state skin (Fig. 2, A and B). The P4 population was selectively labeled in vivo by a fluorescent reagent (Manocept–Alexa Fluor 488) containing mannose moieties (Fig. 2 C and Fig. S2, A and B; Azad et al., 2015). Applying an additional set of markers recently used to define subsets of myeloid cells present in the steady-state ear dermis (Tamoutounour et al., 2013), the P4 cells were CD64+CCR2low, identifying the population as dermal macrophages, of which 95% were MHCII− (Fig. S2 C). Applying these additional markers to cells within the P1–P3 gates identified the populations as inflammatory monocytes (P1), monocyte-derived DCS (moDCs; P2), and moDCs plus CD11b+Ly6C−CD64− dermal DCs (P3). P1 cells displayed morphological features of monocytes, whereas P2 and P3 populations showed a phagocyte morphology with larger cytoplasm and small processes, and P4 cells showed characteristics of mature macrophages with abundant cytoplasmic vacuoles and melanin granules (Fig. 2 D).
In addition to MR, several other previously described M2 markers (Hughes et al., 2008; Shaul et al., 2010; Huang et al., 2014; Thornley et al., 2014; Zagórska et al., 2014; Murray, 2017) were highly expressed in P4 recovered from the steady-state dermis, including CD36 (fatty acid translocase), CD209 (DC-SIGN), Colec12 (scavenger receptor), CD301 (CLEC10A, C-type lectin), MERTK (TAM receptor tyrosine kinase), Tim-4 (PtdSer recognition receptor), Tgfr2 (TGF-β receptor 2), Lyve-1 (angiogenic factor), and S1pr1 (Sphingosine-1-phosphate receptor 1; Fig. 2 E and Fig. S2 D). Of these, only Lyve-1 and S1pr1 were elevated in P1, and none were elevated in P2 or P3. Because the high expression of MERTK and Tim-4 suggested a role in apoptotic cell clearance, this function was compared after intradermal injection of CFSE-labeled apoptotic thymocytes. Most apoptotic cells were captured by P4 dermal macrophages within 10 min, whereas P1–P3 and CD11b+Ly6C−CD64− dermal DCs showed delayed uptake (Fig. 2 F). Recently, a subset of skin-resident macrophages was shown to continuously capture blood-borne macromolecules using protrusions across the endothelial junctions (Barreiro et al., 2016). We investigated the ability of P4 to phagocytose high–molecular weight FITC-dextran from the blood vessel lumen. Only the P4 dermal macrophages were able to rapidly capture the FITC-dextran (Fig. 2 G). P4 cells were also rapidly (within 3 min) and selectively labeled by intravenously injected anti-MR antibody (Fig. 2 H). Together, the results suggest that the P4 cells show M2-like phenotypes, highly phagocytic/trans-endothelial capability, and homeostatic functions related to the rapid clearance of apoptotic cells.
P4 dermal macrophages are not replaced by blood precursors during infection
After infection, the recruitment of monocytes was accompanied by increased numbers of other infiltrating cells, including eosinophils, natural killer cells, and T cells (Fig. 3, A and B). However, the number of P4 dermis-resident macrophages showed only a slight increase, suggesting that monocytes were not likely to differentiate into P4 during infection. To better address the ontogeny of P4 during infection, we infected cx3cr1-gfp mice with LmSd to track cells originating from GFP+ monocytes (Fig. 3 C). In contrast to P1–P3, for which 40–80% of the cells were GFP+ within the first week of infection, <10% of P4 was GFP+. Consistent with these results, we observed the differentiation of adoptively transferred CD45.1+GFP+ monocytes into P2 and P3 moDCs, but not to P4 dermal macrophages in CD45.2+ recipient mice infected with LmSd (Fig. 3 D). The fact that 40% of adoptively transferred GFP+ monocytes lost their GFP expression during their differentiation to moDCs explains why only 40–80% of P1–P3 were GFP+ in the cx3cr1-gfp mice (Fig. 3, C and D). Using BM chimeras, >90% of P4 remained of recipient origin in naive mice, even at 1 mo after BM transfer, whereas the other myeloid populations in the skin were >75% donor derived (Fig. 3 E). Moreover, P4 remained almost entirely of recipient origin during LmSd infection in these BM chimeras. Finally, we established parabionts between CD45.1+ and CD45.2+ congenic mice that were infected with LmSd, one ear in each partner, after 2 wk when their shared blood supply had been established (Fig. 3 F). After 2 wk of infection, all of the lymphoid and myeloid cells recovered from the ear dermis, with the exception of P4, showed exchange between partners, with 20–50% originating from the parabiotic partner. The P4 dermal macrophages were entirely maintained without input from blood precursors originating in the parabiotic partner. Together, these data strongly indicate that MRhi dermal macrophages represent a radio-resistant population that is not replaced by adult BM–derived progenitors during infection.
Preferential infection of P4 dermal macrophages by LmSd in vivo
Intracellular amastigotes were readily observed in sorted populations of P1–P4 cells recovered from RFP+ LmSd-infected ears (Fig. 4 A). In immunostained vertical sections of the ear dermis infected with LmSd-RFP (Fig. 4 B), we found MR+ cells containing melanin granules (Fig. 4 B, insets 3 and 6) in both the dorsal and ventral aspects of the ear pinnae. MR+ cells harboring RFP+ parasites were observed in inflamed regions of the section (Fig. 4 B, insets 1, 4, and 5) in contrast to the noninflamed periphery (Fig. 4 B, inset 2).
To determine whether the MRhi dermal macrophages were preferentially infected by LmSd, as would be predicted based on the in vitro findings involving BMDMs, we measured the frequencies of CD11b+ dermal populations and their relative levels of parasitization by LmSd-RFP and LmFn-RFP during the early stage of infection. The percentages of each myeloid population were comparable between these strains, with P3 being the predominant population at days 2 and 5 and increased frequencies of P1 monocytes and eosinophils by day 12 p.i. (Fig. 4 C). P4 was the predominant subset infected by LmSd, and it was the only population showing a significant difference in infection between the strains (Fig. 4 D) at days 2 and 5 p.i. when there was no difference in overall parasite burden (Fig. 4 E). After 12 d of infection, P1/P3, as well as P4, showed increased infection by LmSd that was associated with a 10-fold increase in parasite burden compared with LmFn (Fig. 4, D and E). At later time points, when the respective healing and nonhealing outcomes were well established (Fig. 4 F), a high percentage of P4 was infected with LmSd, although the P1–P3 populations were also well parasitized during the chronic stage (Fig. 4 G).
Nonhealing infection with LmSd is MR dependent
The nonconsequential role of MR in L. major infection involving a healing strain (LmFn) has been previously reported (Akilov et al., 2007). To evaluate the role of MR in the nonhealing response to LmSd, we infected MR-deficient mice and measured lesion development and parasite burdens. The mrc1−/− mice were as resistant to LmSd infection as the WT mice infected with LmFn with respect to lesion size, pathology score, and parasite load (Fig. 5, A and B). Infected in parallel were casp1/11−/− and il1r−/− mice that displayed the healing phenotypes previously described (Charmoy et al., 2016).
P2 and P3 expressed low but significant levels of surface MR (Fig. 2 B) and bound low but significant amounts of Manocept–Alexa Fluor 488 (Fig. 2 C and Fig. S2, A and B). To address the possible role of these populations in the MR-dependent, nonhealing outcome, we generated BM chimeras between CD45.1+ WT and CD45.2+ mrc−/− animals, taking advantage of the fact that P4 is the only radio-resistant population of MR-expressing cells in the skin (Fig. 3 E). We used the MR-independent gating strategy for identification of CD11b+Lin−CCR2loCD64hiLy6Cint dermal macrophages (Tamoutounour et al., 2013) to determine chimerism in these BM chimeras (Fig. S2 C). At 16 wk after irradiation and 12 wk p.i. in the CD45.2+ mrc−/− BM→CD45.1+ WT chimeras, most of the myeloid and lymphoid populations recovered from the inoculation site had been completely replaced by BM-derived cells of donor origin. The exceptions were the T cells and the P4 dermal macrophages that remained of 25 and 75% recipient origin, respectively (Fig. 5 C). Confining MR expression to these cells was sufficient to promote the nonhealing outcome (Fig. 5, D and E). In contrast, in the WT BM→mrc−/− chimeras, only the dermal macrophages remained of predominant recipient origin (75%), and confining the absence of MR to these cells was sufficient to reproduce the resistance phenotype observed in the mrc−/− mice. Thus, MR expression on dermal macrophages is both necessary and sufficient for the evolution of the nonhealing infection.
The selective depletion of dermal macrophages ameliorates nonhealing infection with LmSd
M-CSF and CSF-1 receptor (CSF-1R) signaling are required for dermal macrophage development (Ginhoux and Jung, 2014). As blockade of the CSF-1R has been reported to deplete resident macrophages in various tissues, including skin (MacDonald et al., 2010), we attempted to selectively deplete the P4 population using an antibody against mouse CSF-1R (M279). The M279 treatment increased circulating M-CSF levels caused by blocking receptor-mediated consumption (Fig. S3 A). We also observed an accelerated weight gain relative to controls, as previously reported (Fig. S3 B; Sauter et al., 2014). Before infection, 3-wk-long M279 treatment resulted in a virtually complete depletion of P4, indicating that M-CSF plays a critical role in maintaining the steady-state levels of these cells (Fig. 6 A). Although an approximate twofold reduction in each of the other dermal myeloid populations was also observed in the M279-treated naive mice, their numbers were rapidly reconstituted after infection, likely the result of infiltration by unaffected blood-derived cells (Fig. 6 B and Fig. S3 C). At 9 d p.i., only the P4 population remained depleted in the infected skin. The selective effect on P4 appears to depend on the differential requirement for M-CSF to maintain this population rather than differential receptor expression because other myeloid cells also express CSF-1R (Fig. S3 D). Finally, M279-treated animals were able to control their infections with LmSd and were as resistant as controls infected with LmFn with respect to lesion size, pathology score, and parasite load (Fig. 6, C and D).
P4 dermal macrophages maintain M2 functionality during infection
As previously described (Charmoy et al., 2016) and confirmed here using both tetramer staining and antigen restimulation of dermal and draining lymph node cells, Leishmania-specific T cell numbers and Th1 cytokine responses were found to be similar or higher in LmSd-infected compared with LmFn-infected mice (Fig. S4, A–C). Considering the strong production of IFN-γ and TNF-α in the LmSd-infected mice, we explored whether P4 dermal macrophages are able to maintain an M2 functionality during infection. Arginase 1 (Arg1) and RELM-α are well-described IL-4Rα−induced products that have proreparative functions (Allen and Sutherland, 2014). Βy intracellular staining of dermal cells from naive mice, we found no expression of iNOS or Arg1 in P1–P4 and only low expressions of Relm-α in P3 and P4 (Fig. 7 A). During the chronic stage of infection (9 and 12 wk p.i.), a high frequency of P1 and P2 and a lower frequency of P3, but virtually no P4, were iNOS+. In contrast, a significant percentage of P4 recovered from the LmSd-infected mice were Arg1+ and/or Relm-α+. A similar fraction of P2 and P3 from these mice were also Arg1+.
To elucidate the fate of the parasite inside the different populations recovered from infected ears, we sorted RFP+ P1–P4 and assayed parasite survival and growth ex vivo (Fig. 7 B). Only the P4 dermal macrophages supported the growth of the parasite during the 48 h of ex vivo culture. Growth arrest or persistence of the parasite was observed over this time frame in P1–P3. Collectively, P4 dermal macrophages maintain their M2 phenotype even in a strong Th1 environment, and they remain permissive to intracellular growth of L. major.
IL-4 and IL-10 are essential to maintain the MRhi P4 population during infection
Because IL-4 and/or IL-10 up-regulated MR on BMDM in vitro (Fig. 1 K), we were interested to know whether these cytokines might locally maintain MR expression on P4 in the face of the high levels of IFN-γ that would normally down-regulate MR (Schreiber et al., 1993). In IL-4– or IL-10–deficient naive mice, the numbers of MRhi P4 cells were not altered (Fig. S5 A). In infected mice, however, the absence of IL-4 and/or IL-10 resulted in a significant decrease in both the percentage and number of P4 relative to infected WT mice (Fig. 8 A). In contrast, the number of P1 and P2 was significantly increased, indicating the inhibitory effects of IL-4 and IL-10 on the recruitment or survival of monocytes and moDCs (Couper et al., 2008; Lazarski et al., 2013). Consistent with the lack of MR up-regulation on BMDM by IL-13 in vitro (Fig. 1 K), there was no change in the number or frequency of P1–P4 in IL-13−/−–infected mice, whereas the recruitment of eosinophils was significantly decreased (Fig. S5 B).
We determined whether the deficit in the P4 population was caused only by the down-regulation of MR expression or also by the loss of these cells as defined using MR-independent markers (Fig. S2 C; Tamoutounour et al., 2013). Compared with WT mice, IL-4 and/or IL-10 deficiency resulted in a significant decrease both in the percentage of the CD11b+ cells that were dermal macrophages and the level of MR expression on these cells (Fig. 8 B). Because IL-4 can induce local macrophage proliferation (Jenkins et al., 2011), we investigated whether there was IL-4–dependent expansion of P4 during LmSd infection (Fig. 8, C and D). On average, 5 and 1% of dermal macrophages in naive animals were Ki67 positive or BrdU incorporated, respectively, indicating a slow rate of self-renewal in steady state. At 12 d p.i., the P4 exhibited a two to threefold increase in the frequencies of Ki67+ and BrdU+ cells that were in each case abrogated in the IL-4–deficient mice, consistent with the small increase in the number of P4 cells during infection (Fig. 3 B). Notably, both IL-4Rα and IL-10Rα were expressed at higher levels on P4 than P1–P3 (Fig. 8 E). Collectively, during infection, IL-10 and to a lesser degree IL-4 are needed for MR expression, whereas IL-4 is needed to expand the population through local proliferation.
The LmSd infection outcomes were also monitored in the il4−/−, il10−/−, and il4/10−/− mice. As previously reported (Anderson et al., 2005), il10−/− mice had significantly lower parasite burdens than WT (Fig. 8 F). Interestingly, il4/10−/− mice had further decreased parasite counts compared with il10−/− mice. The role of IL-4 in promoting lesion development was apparent after 10 wk p.i., although the lesions still progressed in the il4−/− mice (Fig. 8 G). Despite the reduced parasite burdens, il10−/− and il4/10−/− mice showed comparable or exacerbated lesions compared with WT mice during the first 12 wk of infection, consistent with a critical role for IL-10 in modulating immune-mediated pathologies. However, lesions in all of the il10−/− and some of the il4/10−/− mice started to resolve or moderate after 12 wk p.i.
IL-1R and inflammasome activation contribute to P4 maintenance during infection
The IL-1β–inflammasome axis plays an essential role in the evolution of the nonhealing response (Charmoy et al., 2016). We were interested to understand how this innate response pathway might interplay with the disease-exacerbating mechanisms revealed in this study. We first determined that the source of IL-1β was not P4, but confined largely to P1–P3 (Fig. 9 A). We and others have previously reported that IL-1R signaling up-regulates IL-4 and IL-10 secretion and mRNA expression in L. major–infected mice (Kostka et al., 2006; Charmoy et al., 2016). In this study, the frequency of dermal CD4+ T cells producing IL-4 and/or IL-10, already low in the WT mice, was significantly reduced in casp1−/− and il1r−/− mice, whereas the frequency of IFN-γ+ T cells was increased (Fig. 9 B). Accordingly, the number of P4 dermal macrophages was significantly reduced in the LmSd-infected casp1−/− and il1r−/− mice (Fig. 9 C). The number of eosinophils was also reduced.
In this study, we explore the infection and activation states of myeloid subsets in the dermis of conventionally resistant C57BL/6 mice infected with a healing and nonhealing strain of L. major. We identify a population of MRhi dermal macrophages that is present in the steady-state dermis and is preferentially infected by LmSd in an MR-dependent manner to promote nonhealing cutaneous disease. The MRhi dermal macrophages are radio resistant and only minimally replaced by monocytes or adult BM–derived precursors during infection. Their self-renewal is M-CSF dependent in the steady state and requires IL-4 during infection. They maintain M2-associated markers and remain permissive to L. major growth despite the high levels of IFN-γ and TNF-α that activate monocyte-derived subsets for controlling parasites within the same tissue environment.
The MRhi population described here provides clear evidence that there is a substantial M2-like population present in the steady-state dermis. The population was identified by combining MR and Ly6C staining on CD11b+lin− cells and comparing them with the markers previously used to define dermal macrophages (CD64hiLy6CintCCR2loMHCII+/−; Tamoutounour et al., 2013). MR up-regulation on macrophages in response to type 2 cytokines provided one of the first markers for their alternative activation state (Stein et al., 1992; Martinez-Pomares et al., 2003). MR functions as a scavenger of blood-circulatory glycoproteins that are elevated in inflammation and wound healing (Lee et al., 2002). We showed that MRhi dermal macrophages rapidly scavenged high–molecular weight dextran from blood lumen, which was reported to be MR dependent and a characteristic of skin trans-endothelial, radio-resistant macrophages, termed STREAMs (Wollenberg et al., 2002; Barreiro et al., 2016). We also showed that the dermal macrophages efficiently captured apoptotic cells and up-regulated an additional set of M2-associated molecules, including PtdSer receptors, scavenger receptors, angiogenic factors, and C-type lectins. A related panel of molecules was found to be transcriptionally up-regulated in the dermal macrophages previously described (Tamoutounour et al., 2013). Altogether, these molecular and functional attributes suggest that the MRhi dermal macrophages have important homeostatic functions related to tissue repair and the resolution of inflammation. It is important to emphasize that the designation of the P4 population as M2-like encompasses macrophages with broadly antiinflammatory, reparative phenotypes and that their gene expression profiles are likely to be distinct from other designated M2 populations that have been conditioned by a unique set of mediators and microbial stimuli in vivo or in vitro (Murray et al., 2014).
Most peripheral tissues are populated early during embryogenesis by yolk sac– or fetal liver–derived macrophages that are self-renewed in situ (Ginhoux et al., 2010; Yona et al., 2013). Dermal macrophages are thought to be an exception to this ontogenic scheme because in the two studies addressing this question, a substantial fraction of these cells originated from the parabiotic partner (Jakubzick et al., 2013; Tamoutounour et al., 2013), implicating a contribution from blood precursors. In contrast, our data pertaining to the P4 population and using multiple approaches to study its renewal by monocyte-derived or adult BM–derived precursors indicate minimal replacement by these cells, even during the first weeks of infection when there were high numbers of monocytes recruited to the site. The inconsistencies in the extent of donor exchange in the parabiotic partners, comparing the prior studies with our own, likely reflect the steady-state versus infection conditions analyzed. Our study highlights the self-turnover of dermis macrophages that is IL-4 dependent during L. major infection. In this regard, the findings share some similarities with the first study to describe localized macrophage renewal, in which helminth infection in the pleural cavity resulted in IL-4–dependent proliferative expansion of resident macrophages with minimal recruitment of adult BM–derived cells (Jenkins et al., 2011). A key distinction of our study, however, is that the self-renewal of dermal macrophages and their maintenance as M2-like cells occurred not within the context of a helminth-driven, polarized type 2 response, but within the strong proinflammatory environment of the L. major–loaded dermis.
In this study, L. major infection induced an iNOSnegArg1hiRelm-αpos response in at least a fraction of the P4 population even as the Th1 response strongly up-regulated iNOS expression in the inflammatory monocytes and moDCs. Our results suggest that so long as the P4 population remained locally conditioned by IL-4 and IL-10 during infection, its size and M2 functionality such as MR expression were maintained. The up-regulated expression of IL-4Rα and IL-10Rα on the P4 population likely explains how these cells remain responsive to the relatively low levels of IL-4 and IL-10 produced in the site. In addition, there may be a close tissue association between the cells that produce these cytokines and the dermal macrophages that favors their compartmentalized response. A recent study showed that genetic deletion of TNF strongly up-regulated Arg1 and down-regulated iNOS in monocyte-derived inflammatory cells despite a prominent Th1 response during L. major infection, suggesting that the P4 dermal macrophages may have lost their responsiveness to TNF that would normally function to antagonize their alternative activation (Schleicher et al., 2016). Their functional specialization for apoptotic cell capture (Fig. 2 F) might also contribute to the signals that establish and maintain their M2 activation program. It was reported that coordinated signals involving type 2 cytokines and recognition of apoptotic cells were required to maintain the reparative function of tissue-resident macrophages during inflammation in the lung and the gut (Bosurgi et al., 2017).
In contrast to P4 dermal macrophages, TNF-α– and iNOS-producing dermal DCs of monocyte origin have been previously associated with the control of L. major infection (Wei et al., 1995; Stenger et al., 1996; De Trez et al., 2009). Our own findings in P1–P3 confirm the coexpression of iNOS and Arg1 in monocyte-derived cells from L. major–infected C57BL/6 mice reported previously (Schleicher et al., 2016) and are consistent with prior studies showing Arg1 production by M1-polarized cells in other infection models (El Kasmi et al., 2008; Qualls et al., 2010). The phenotype of P1–P3 populations may explain why these cells continue to be parasitized during the chronic phase of L. major infection and may contribute to the lack of parasite control in LmSd-infected mice.
The size of the P4 population was significantly decreased by the absence of IL-1R or caspase-1/11, consistent with the requirement for inflammasome-dependent IL-1β for the evolution of the nonhealing phenotype (Charmoy et al., 2016). We and others previously observed (Charmoy et al., 2016) and confirm here that the production of IL-4 and IL-10 is decreased and the production of IFN-γ is increased during infection in the absence of IL-1R or caspase 1/11. There is accumulating evidence that IL-1R signaling can promote Th2 development (Santarlasci et al., 2013). Of particular relevance to the current study, the conversion of monocyte-derived macrophages to an alternatively activated state with reparative function in the liver after infection with Listeria monocytogenes relied on an IL-1β–caspase-1–dependent pathway of IL-4 secretion by basophils (Blériot et al., 2015). We are currently investigating multiple possible cellular sources of IL-4, including eosinophils that infiltrate Leishmania-infected skin and can be potent producers of IL-4 (Rothenberg and Hogan, 2006).
Our data confirm that CSF-1 is a survival factor for dermal macrophages under steady-state conditions. Although the monocyte-derived populations were also partially depleted in the naive skin by the CSF-1R blockade, only the dermal macrophages remained significantly depleted in the infected skin. This is in agreement with the lack of effect of CSF-1R blockade on monocyte recruitment in other inflammatory models, including wound healing (MacDonald et al., 2010). The requirement of prolonged treatment of M279 for a complete depletion of P4 is consistent with a prior study (Sauter et al., 2014) and concurs with their slow rate of turnover, as shown in the Ki67 staining and BrdU incorporation. CSF-1R blockade is currently being used as a strategy to inhibit the tumor-promoting function of tumor-associated macrophages that is thought to rely on their M2 activated state (Pyonteck et al., 2013).
MR recognizes the mannose-containing moieties of a variety of microorganisms (Linehan et al., 2000). We showed that treatment of BMDMs with BSA-mannose significantly decreased LmSd infection, but not LmFn infection in vitro, and that stronger M2 polarization using IL-4 and IL-10 further up-regulated MR and augmented the favored uptake of LmSd versus LmFn. Differences in the oligosaccharide structures of the major surface lipophosphoglycans of LmSd and LmFn promastigotes have been described that might account for the differences in MR-mediated uptake (Sacks and da Silva, 1987). However, because the selective recognition extended to the amastigote stage that is largely devoid of lipophosphoglycan, the involvement of other mannose-containing glycoconjugates is indicated. Our finding that the in vitro behavior of LmSd×LmFn hybrids generated by sexual crosses precisely predicted their healing or nonhealing phenotypes in vivo reinforces the link between the differential targeting of MRhi macrophages and clinical outcome and offers the possibility to use a forward genetic approach to identify the parasite ligands involved.
MR-mediated infection of tissue-resident macrophages that maintain an M2 activation program is a possible immune evasive strategy used by other intracellular pathogens. For example, MR plays an important role in the phagocytosis of a virulent but not an attenuated strain of Mycobacterium tuberculosis (Schlesinger, 1993). Reagents that specifically target the CD206 macrophage MR in humans are being developed for diagnostic imaging and therapy to inhibit tumor progression (Cope et al., 2016). The clinical relevance of our findings can be inferred from the fact that the LmSd strain was isolated from a patient with chronic lesions and persisting organisms even after treatment and despite a strong cell-mediated immune response to Leishmania antigens (Neva et al., 1979). Our study provides a rationale for therapeutic targeting of MRhi dermis-resident macrophages for the treatment of chronic forms of cutaneous leishmaniasis.
Materials and methods
C57BL/6 mice were purchased from Taconic Farms. C57BL/6 IL1r1−/−, C57BL/6 Casp1/11−/−, C57BL/6 il4−/−, C57BL/10 il10−/−, C57BL/6 il13−/−, C57BL/6 il4−/−il10−/−, CD45.1+ B6.SJL, and CD45.1+cx3cr1gfp/+ (B6.SJL-Cd45a(Ly5a)/Nai×C57BL/6-CD45a-[KO]CX3CR1-[KI]EGFP F1) mice were obtained through a supply contract between the National Institute of Allergy and Infectious Diseases (NIAID) and Taconic Farms. C57BL/6J cr3−/− (itgamtm1Myd), C57BL/6J c3−/− (c3tm1Crr), and C57BL/6J il10gfp (il10Tm1.1Karp) mice were purchased from The Jackson Laboratory. C. Bosio (NIAID, National Institutes of Health, Bethesda, MA) provided C57BL/6 mrc−/− mice. All the mice used in these experiments were females and were maintained in the NIAID animal care facility under specific pathogen-free conditions, used under a study protocol approved by the NIAID Animal Care and Use Committee (protocol no. LPD 68E). All aspects of the use of animals in this research were monitored for compliance with the Animal Welfare Act, the Public Health Service Policy, the U.S. Government Principles for the Utilization and Care of Vertebrate Animals Used in Testing, Research, and Training, and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
In vivo injections
M279 (Amgen) is a rat IgG mAb, which blocks ligand binding to the CSF-1R. In this study, mice were treated with 200 µg M279 or rat IgG (Sigma-Aldrich) intraperitoneally three times a week for 9 wk. Manocept–Alexa Fluor 488 (Navidea Biopharmaceuticals) is a fluorescently labeled derivative of FDA-approved 99MTC-Tilmanocept targeting MR. Naive mice were injected with Manocept–Alexa Fluor 488 intradermally and intravenously and sacrificed 1 and 2 h later, respectively, for analysis. Naive mice were also injected intravenously with 1% (wt/vol) of 2-MD FITC-dextran (Sigma-Aldrich) or 10 µg of anti–mouse CD206 (C068C2; Biolegend) and sacrificed 10–30 and 3 min later, respectively. For measuring apoptotic cell uptake, 106 thymocytes were treated for 6 h with 1 µM dexamethasone (Sigma-Aldrich) and were intradermally injected into naive mice.
L. major infection and lesion measurements
LmSd (MHOM/SN/74/SD) and LmFn (MHOM/IL/80/Friedlin) were maintained as follows: promastigotes were grown at 26°C in medium 199 (M199) supplemented with 20% heat-inactivated FCS (Gemini Bio-Products), 100 U/ml penicillin, 100 µg/ml streptomycin, 2 mM l-glutamine, 40 mM Hepes, 0.1 mM adenine (in 50 mM Hepes), 5 mg/ml hemin (in 50% triethanolamine), and 1 mg/ml 6-biotin (M199/S). Parasites expressing an RFP (LmFn-RFP and LmSd-RFP) were grown using the identical culture medium supplemented with 50 µg/ml Geneticin (G418; GIBCO BRL). Infective-stage metacyclic promastigotes were isolated from stationary cultures (5–6 d) by density gradient centrifugation as described previously (Späth and Beverley, 2001). Mice were then inoculated with 1,000 or 200,000 metacyclic promastigotes in the ear dermis by intradermal injection in a volume of 10 µl. Lesion development was monitored weekly by measuring the diameter of the ear nodule with a direct-reading Vernier caliper (Thomas Scientific).
Processing of ear tissues and evaluation of parasite burden
Ear tissue was prepared as previously described (Belkaid et al., 2000). In brief, the two sheets of infected ear dermis were separated, deposited in DMEM containing 0.2 mg/ml Liberase TL–purified enzyme blend (Roche Diagnostics Corp.), and incubated for 1.5 h at 37°C. Digested tissue was processed in a tissue homogenizer for 3.5 min (Medimachine; Becton Dickinson) and filtered through a 70-µm cell strainer (Falcon Products). Parasite titrations were performed as previously described (Belkaid et al., 1998). In brief, tissue homogenates were serially diluted in 96-well, flat-bottom microtiter plates containing 100 µl M199/S. The number of viable parasites in each ear was determined from the highest dilution at which promastigotes could be grown out after 7–10 d of incubation at 26°C.
Immunolabeling and flow cytometry analysis
Single-cell suspensions were stained with a LIVE/DEAD Fixable Aqua Dead Cell Stain kit (Thermo Fisher) and incubated with an anti–Fc-γ II/III (CD16/32) receptor antibody (2.4G2; BD Biosciences) in PBS containing 1% FCS followed by fluorochrome-conjugated antibodies for 1 h on ice. The following antibodies were used for surface staining: FITC, APC/Cy7, and Brilliant Violet 421 anti–mouse Ly6G (1A8; Biolegend); APC/Cy7 anti–mouse Ly6C (HK1.4; Biolegend); PE/Cy7 anti–mouse CD11b (M1/70; Biolegend); FITC anti–mouse CD4 (GK1.5; Biolegend); PerCP/Cy5.5 and APC anti–mouse CD8-α (53-6.7; Biolegend); PerCP/Cy5.5, Brilliant Violet 421, and APC/Cy7 anti–mouse NK1.1 (PK136; Biolegend); Alexa Fluor 647 and APC anti–mouse CD206 (C068C2; Biolegend); PE and Brilliant Violet 421 anti–mouse Siglec-F (E50-2440; BD Biosciences); FITC and PE anti–mouse CD45.1 (A20; Biolegend); PerCP/Cy5.5 anti–mouse CD45.2 (104; Biolegend); PE anti–mouse IL-10Rα (1B1.3a; Biolegend); PE anti–mouse IL-4Rα (I015F8, Biolegend); PE anti–mouse CD36 (HM36; Biolegend); PE anti–mouse CD301 (LOM-14; Biolegend); PE anti–mouse DC-SIGN (902404; R&D Systems); PE anti–mouse TGFR2 (R&D Systems); biotin anti–mouse COLEC12 (R&D Systems) with PE-streptavidin (Biolegend); FITC anti–mouse MHCII (AF6-120.1; Biolegend); PE anti–mouse CCR2 (475301; R&D Systems); PerCP/Cy5.5 anti–mouse CD64 (X54-5/7.1; Biolegend); PerCP/Cy5.5 anti–mouse TCR-β (H57-597; Biolegend); PerCP/Cy5.5 anti–mouse B220 (RA3-6B2; Biolegend); PE anti–mouse Lyve-1 (223322; R&D Systems); PE anti–mouse MERTK (2B10C42; Biolegend); and PE anti–mouse Tim-4 (RMT4-54; Biolegend). For intracellular detection of cytokines, cells were cultured in the presence of 20 ng/ml PMA and 700 ng/ml ionomycin in the presence of GolgiStop protein transport inhibitor (BD Biosciences) for 4 h at 37°C. The staining of surface and cytoplasmic markers was performed sequentially. Cells were first incubated with a LIVE/DEAD Fixable Aqua Dead Cell Stain kit. They were stained for their surface markers, then fixed and permeabilized using Cytofix/Cytoperm (BD Biosciences), and finally stained for detection of intracellular molecules for 30 min on ice. The following antibodies were used for cytokine detection: PE anti–mouse IL-1β proform (NJTEN3; eBioscience); APC anti–mouse IL-2 (JES6-5H4; eBioscience); PE and PE/Cy7 anti–mouse IFN-γ (XMG1.2; eBioscience); PE anti–mouse IL-4 (11B11; eBioscience); PE and Brilliant Violet 421 anti–mouse IL-10 (JES5-16E3; eBioscience); PE anti-mouse TNF-α (TN3-19; eBioscience); APC anti–mouse Arginase1 (R&D Systems); Alexa Fluor 488 anti–mouse iNOS (CXNFT; eBioscience); and biotinylated anti–mouse Relm-α (Peprotech). For detection of antigen-specific CD4+ T cells, single-cell suspensions were stained with PE glycosomal phosphoenolpyruvate carboxykinase–tetramer (I-Ab L. major GPC 335-351 NDAFGVMPPVARLTPEQ) and PE and APC Clip-tetramer (I-Ab human CLIP 87-101 PVSKMRMATPLLMQA) for 1 h at room temperature before surface staining. The data were collected using FacsDIVA software and a flow cytometer (FacsCANTO II; BD Biosciences) and analyzed with FlowJo software (Tree Star). To measure in vivo cell proliferation, mice were intravenously injected with 2 mg BrdU (eBioscience) in PBS. After 3 h, intracellular staining for Ki67 and incorporated BrdU were performed using eBioscience kits according to the manufacturer’s instructions.
BMDM generation, infection, and in vitro treatment
Isolated femurs and tibia were flushed with PBS, and precursor cells were cultured in complete RPMI supplemented with 30% L929 cell–conditioned medium. After 7 d in culture, mature BMDMs were further incubated with M1- or M2-inducing stimuli (10 ng/ml IFN-γ, 50 ng/ml LPS, 20 ng/ml IL-4, 10 ng/ml IL-10, or 10 ng/ml IL-13; Peprotech) or harvested by washing with cold PBS, incubating on ice for 15 min, and pipetting extensively. 5 × 104 BMDMs/well were plated onto an 8-well Permanox chamber slide (Thermo Fisher) in complete RPMI and infected with metacyclic promastigotes and amastigotes at different multiplicities of infection (MOIs) with or without BSA-mannose (Vector Laboratories) and RGDS (Sigma-Aldrich) pretreatment. After infection, BMDMs were washed three times with PBS to remove free parasites and further incubated in complete RPMI for Wright-Giemsa staining (Thermo Fisher). To acquire amastigotes, infected BMDMs were lysed by passing through 25-gauge needles in amastigote media (50 mM glucose and 20 mM EDTA in PBS) and spun at 300 g for 10 min. Supernatant cells were pelleted and washed two times with amastigote media at 3,000 g for 15 min.
BM chimera generation
Recipient mice were irradiated with a single dose of 1,250 rads and reconstituted with donor BM. The harvested BM cells were injected intravenously at a dose of 2.5 × 106 cells/mouse in 100 µl PBS. Animals were maintained on trimethoprim-sulfamethoxazole (Hi-Tech Pharmacal) antibiotic water for up to 5 wk after irradiation. Hematopoietic reconstitution of all animals was verified by FACS staining of splenocytes and ear isolates.
Adoptive transfer of monocytes
For adoptive transfer, total monocytes from the BM of CD45.1+ cx3cr1gfp/+ mice were sorted on a monocyte isolation kit (Miltenyi Biotec) as EGFP+ to a purity of >95%. For homing experiments, cells were injected intravenously (2 × 106 cells) into C57BL/6 recipient mice at 7 d p.i. with 2 × 105 LmSd. Ear tissues were processed, and the homing of GFP+ monocytes was analyzed by flow cytometry at 1, 3, and 5 d after the injection.
7–8-wk-old female CD45.2+ C57BL/6 and CD45.1+ B6.SJL mice were matched for body weight and cohoused for 2 wk. Pairs were surgically connected as previously described (Wright et al., 2001).
LmSd-RFP–infected mouse ears were mounted in Tissue-Tec OCT (Sakura Finetek). 12-µm cryostat-cut sections were placed onto Superfrost Plus microscope slides (Thermo Fisher) and allowed to dry for 5 min. Before immunostaining, the slides were rehydrated in PBS for 5 min. For immunohistochemistry, the sections were blocked with 10% FCS in PBS and incubated with APC anti–mouse CD206 (C068C2; Biolegend). After 1 h of incubation (at 37°C temperature in a humidified chamber), the slides were washed in PBS for 10 min and stained with Hoechst 33342 (Thermo Fisher). The samples were analyzed with a confocal laser fluorescence microscope (SP8; Leica) and Imaris software (Bitplane).
The differences in values obtained for two different groups were determined using a nonparametric Mann-Whitney test. For comparisons of multiple groups, ANOVA followed by Dunn’s posttest was used. Analyses were performed using Prism 6.0f software (GraphPad).
Online supplemental material
Fig. S1 shows the healing and nonhealing cutaneous infection of C57BL/6 mice by LmFn and LmSd, respectively, in terms of lesion size and pathology score. Fig. S2 shows selective uptake of Manocept–Alexa Fluor 488 by P4 dermal macrophages, the comparison of P1–P4 populations with previously reported dermal myeloid populations (Tamoutounour et al., 2013), and histograms of M2 markers expressed on P1–P4 in a naive state. Fig. S3 analyzes effects of an extended blockade of CSF-1R signaling by anti-CSFR antibody (M279). Fig. S4 shows that LmFn and LmSd had no significant difference in both antigen-specific T cell development and cytokine production from T cells. Fig. S5 shows no difference in the number or the percentage of P4 in naive WT, IL-4−/−, IL-10−/−, and IL-4/10−/− mice or in infected IL-13−/− mice.
We thank Dr. Catherine Bosio for the mrc−/− mice; Dr. Owen Schwartz (NIAID) and Dr. Sundar Ganesan (NIAID) for help with image acquisition and analysis; and Dr. Calvin Eigsti (NIAID) for help with the cell sorting.
This work was supported in part by the Intramural Research Program of the NIAID, National Institutes of Health.
The authors declare no competing financial interests.
Author contributions: S.H. Lee and D.L. Sacks designed the study. S.H. Lee, M. Charmoy, A. Romano, A. Paun, M.M. Chaves, F.O. Cope, and D.A. Ralph performed experiments. S.H. Lee and D.L. Sacks analyzed the data and wrote the manuscript.