The lymphoid-associated interleukin 7 receptor (IL7R) regulates tissue-resident macrophage development

Hematopoietic stem cells (HSCs) are responsible for sustaining blood and immune cell production across the lifespan of the animal, under steady-state conditions, during infection, and following transplantation. However, recent findings have revealed that many tissue-resident immune cells are poorly generated or regenerated from adult HSCs, including subsets of tissue-resident macrophages (trMacs), such as microglia, epidermal Langerhans cells, liver Kupffer cells, and alveolar macrophages (Beaudin and Forsberg, 2016; Cool and Forsberg, 2019; Ginhoux et al., 2010; Gomez Perdiguero et al., 2015; Guilliams et al., 2013; Hashimoto et al., 2013; Hoeffel et al., 2015, 2012; Sawai et al., 2016; Yona et al., 2013). These trMacs are specialized macrophages that reside within tissues and have specific functions in tissue and immune homeostasis (Epelman et al., 2014; Ginhoux et al., 2010; Gomez Perdiguero et al., 2015; Guilliams et al., 2013; Hoeffel et al., 2012). Unlike classical adult monocyte-derived macrophages, which are recruited from circulation upon inflammation and have high turnover rates, trMacs are locally maintained by proliferation, independently of adult hematopoiesis (Ajami et al., 2007; Hashimoto et al., 2013; Hulsmans et al., 2017; Merad et al., 2002). Recent fate-mapping studies have revealed a fetal origin of specific trMacs (Epelman et al., 2014; Gomez Perdiguero et al., 2015; Hoeffel et al., 2015; Schulz et al., 2012; Yona et al., 2013), but the cellular and molecular mechanisms driving trMac establishment and expansion within fetal tissues are poorly understood.

Whether trMacs are seeded once from a common progenitor or in reiterative waves throughout development is under intense investigation. Accumulating evidence points towards extra-embryonic yolk sac (YS)-derived erythromyeloid progenitors (EMPs) as the initial cell of origin for macrophages that seed developing tissues, and the primary source of brain microglia (Gomez Perdiguero et al., 2015; Kierdorf et al., 2013). Microglia and other trMacs can differentiate directly from EMPs through either Myb-dependent or -independent pathways, likely through an EMP-derived intermediate (Hoeffel et al., 2015; Schulz et al., 2012). Macrophages derived from fetal liver (FL) precursors may replace or replenish certain fetal-derived trMac populations initially seeded by YS-derived cells, including lung alveolar macrophages, epidermal Langerhans cells and liver Kupffer cells (Guilliams et al., 2013; Hoeffel et al., 2015, 2012; Yona et al., 2013). However, both the precise cell of origin for FL precursors and their specific contribution to adult trMac compartments remain controversial, partly because of incomplete progenitor labeling and the difficulties in accurately tracking progeny in situ. Despite intense investigation of the mechanisms regulating macrophage differentiation from YS progenitors, including gene expression programs (Mass et al., 2016) and mechanisms of tissue seeding (Stremmel et al., 2018), comparatively less is known about the developmental mechanisms that regulate the establishment of tissue macrophages from later waves of hematopoietic cell production.

Here, our investigation of hematopoietic development using the Il7r-Cre lineage-tracing model (Schlenner et al., 2010) revealed unexpected and robust labeling of adult trMacs in multiple tissues. Examination of fetal myeloid development revealed a transient wave of IL7Rα expression in developing fetal macrophages that was dynamically regulated as macrophages differentiated within developing resident tissues. Both germline deletion and specific blockade of IL7R during the developmental window in which trMacs express IL7R confirmed the functional requirement for IL7R signaling during fetal trMac development. In contrast, IL7Rα was not expressed by YS EMPs and YS hematopoiesis was unperturbed in Il7r^−/− embryos. Together, these experiments reveal IL7R as a novel regulator of fetal macrophage development during late gestation.

The lymphoid-associated genes Flk2 (Flt3), Il7r, and Rag1 label cells with increasingly restricted lymphoid potential in adult hematopoiesis (Adolfsson et al., 2005; Forsberg et al., 2006; Igarashi et al., 2002; Kondo et al., 1997). In contrast, during fetal development all three markers identify oligopotent hematopoietic progenitors with both myeloid and lymphoid potential (Beaudin et al., 2016; Boiers et al., 2013). Although we and others have used the Flk2-Cre and Rag1-Cre models to track the contribution of fetal progenitors to trMac populations (Boiers et al., 2013; Epelman et al., 2014; Gomez Perdiguero et al., 2015; Hashimoto et al., 2013; Hoeffel et al., 2015), the contribution of IL7R-marked progenitors to the same populations has not been previously examined. To decipher the contribution of specific transient hematopoietic progenitors to adult trMacs, we compared adult trMac labeling across three lineage-tracing models: Flk2-Cre, Il7r-Cre and Rag1-Cre. We crossed mice expressing Rag1-Cre (Welner et al., 2009) or Il7r-Cre (Schlenner et al., 2010) to mTmG mice expressing a dual-color fluorescent reporter (Muzumdar et al., 2007), thereby creating ‘Rag1Switch’ and ‘IL7RSwitch’ models (Fig. 1A) analogous to the previously described FlkSwitch mouse (Beaudin et al., 2016; Boyer et al., 2012, 2011). In all models, all cells express Tomato (Tom) until Cre-mediated recombination results in the irreversible switch to GFP expression by that cell and all of its progeny (Fig. 1B).

We compared reporter expression in fetal-derived trMacs of all three models to reporter expression in adult bone marrow (BM)-derived circulating peripheral blood (PB) monocytes (CD11b^hiGr1^loSSc^lo; CD11b also known as ITGAM, Gr1 as Ly6G; SSc, side scatter parameter), the precursors of adult HSC-derived macrophages. As expected, Cre-driven GFP labeling of monocytes in IL7RSwitch and Rag1Switch mice was less than 5% (Fig. 1C) and paralleled the nominal labeling observed in adult HSCs and myeloid progenitors (Fig. S1A,B), as previously reported (Schlenner et al., 2010; Welner et al., 2009). Adult FlkSwitch mice exhibited high GFP labeling in PB monocytes (Fig. 1C), as we previously reported (Boyer et al., 2011). Intriguingly, examination of Cre-driven reporter switching in adult trMacs revealed distinct GFP labeling patterns across all three lineage-tracing models and across tissues. Within Langerhans cells (LCs) of the epidermis, around 20% of cells were labeled by GFP in the FlkSwitch model (Fig. 1D), consistent with previous reports (Gomez Perdiguero et al., 2015; Hoeffel et al., 2015). Low GFP labeling was observed in LCs of Rag1Switch mice (Fig. 1D), as described previously for skin and other tissue macrophages during fetal development (Boiers et al., 2013). In sharp contrast, up to 96% of LCs expressed GFP in IL7RSwitch mice (Fig. 1D). A similar pattern was observed for microglia; consistent with a pre-HSC origin; virtually no reporter switching was observed in adult microglia of FlkSwitch mice (Fig. 1E) and minimal microglia labeling was observed in Rag1Switch mice. Remarkably, however, GFP labeling of microglia was over 85% in IL7RSwitch mice (Fig. 1E).

Examination of reporter expression across lineage-tracing models in trMacs of the lung (alveolar macrophages, AMs) and the liver (Kupffer cells, KCs) yielded a different labeling pattern compared with the LCs and microglia (Fig. 1F,G). In FlkSwitch and Rag1Switch mice, GFP labeling in AMs and KCs was low and roughly comparable to LCs (∼8-19% GFP⁺ cells; Fig. 1D,F,G). Surprisingly, in IL7RSwitch mice, GFP labeling in AM and KC populations was substantially higher, ∼40%, compared with both FlkSwitch and Rag1Switch models (Fig. 1F,G). Although labeling in these two populations was considerably less than that of LCs or microglia in the same mice (∼85-96%), it was substantially higher than labeling of adult BM-derived myeloid populations (F4/80^loCD11b⁺; F4/80 also known as ADGRE1) within the same tissues (<5%) (Fig. S1C-E). Comparison of Cre-mediated labeling between circulating BM-derived monocytes (<4%; Fig. 1C) and trMacs (40-90%; Fig. 1D-G) in adult IL7RSwitch mice also revealed stark differences in labeling, supporting previous reports that adult BM-derived monocytes did not substantially contribute (at steady-state) to the trMac populations investigated, including microglia, LCs, KCs, and AMs (Gomez Perdiguero et al., 2015; Hashimoto et al., 2013; Hoeffel et al., 2012; Sheng et al., 2015). Similarly, other myeloid cells in the tissues, such as neutrophils, were also minimally labeled (Fig. S1F,G), consistent with a previous report (Schlenner et al., 2010). Despite substantial Il7r-Cre-mediated labeling, IL7Rα surface expression was undetectable in any adult trMacs surveyed (Fig. 1H). Il7r was also virtually undetectable in trMacs compared with BM Pro-B-cells (Fig. 1I). The robust labeling of adult trMacs in the IL7RSwitch model in the absence of message or surface expression therefore suggested that IL7Rα was expressed during an earlier window of macrophage development.

YS EMPs have been proposed as a cell of origin for fetal trMac development, either through direct differentiation or through the Myb-dependent generation of intermediate progenitors that seed the fetal liver (Gomez Perdiguero et al., 2015; Hoeffel et al., 2015, 2012; Mass et al., 2016; Schulz et al., 2012). To determine whether Il7r-Cre labeling of adult trMac populations initiated in YS EMPs during embryonic development, we examined Il7r-Cre-driven GFP labeling, IL7Rα surface expression and Il7r in YS EMPs and YS macrophages. YS EMPs (Ter119⁻Kit⁺CD41⁺; Ter119 also known as LY67; CD41 also known as ITGA2B) at embryonic day (E)9.5 were minimally labeled by Il7r-Cre, consistent with a complete absence of IL7Rα surface expression (Fig. 2A,A′). There was virtually no Il7r-Cre-induced reporter expression within the CD45⁺ (CD45 also known as Ptprc) fraction of the YS, which included myeloid cells expressing CD11b or F4/80 (Fig. 2B), and there was similarly no IL7Rα surface expression observed (Fig. 2B′). Il7r in E9.5 EMPs was detectable at very high Ct values (Fig. 2C,C′), and consequently we observed slightly more labeling in YS EMPs by E10.5 (Fig. 2A). However, increased reporter expression was not accompanied by surface expression (Fig. 2A′). Overall, labeling of YS EMPs was comparable to background labeling observed in adult hematopoiesis (Schlenner et al., 2010). To clarify whether IL7R expression regulated YS hematopoiesis, we examined the frequency of YS EMPs and YS macrophages in Il7r^−/− embryos, which have a germline deletion of IL7Rα (Peschon et al., 1994). Neither YS EMPs nor YS macrophage frequency was significantly different between wild-type (WT) and Il7r^−/− mice (Fig. 2D,E). Together, these data indicate that IL7R is not required for the generation of YS progenitors or YS macrophages, and suggest that the robust labeling of F4/80^hi trMacs by Il7r-Cre occurs later in trMac ontogeny.

As early YS progenitors and macrophages were minimally labeled by Il7r-Cre, we investigated labeling in putative progenitors slightly later in development. Labeling remained comparably low in both YS and aorta-gonad-mesonephros (AGM) Kit⁺CD41⁺ EMPs at E11.5, and IL7R surface expression was still undetectable in the EMPs within the FL (Fig. 2A,A′; Fig. S2A). Higher Il7r-Cre-driven labeling was observed in FL Lin-Kit⁺Sca1⁺ (KLS) progenitors by E12.5 (18%), but surface expression was still low (Fig. 2F,H). Labeling subsequently declined in E14.5 progenitors (Fig. 2G,H; P<0.001), and all labeling in E14.5 progenitors was confined to the CD150 (SLAMF1)-multipotent progenitor compartment (Fig. S2B). Consistent with labeling of upstream multipotent cells in the FL, comparable Il7r-Cre labeling was observed in all downstream progenitor and mature compartments (Fig. S2C-G), with the exception of significant labeling in committed lymphoid progenitors (Fig. S2D) and mature lymphoid cells (Fig. S2E). Labeled progenitors at E14.5 did not possess long-term multilineage reconstitution, confirming that they were not definitive HSCs (Fig. S2H). These data suggest that IL7R reporter expression in late (E10.5) YS and AGM partially labels a transient progenitor that contributes to FL hematopoiesis but is not a definitive HSC.

Despite low levels of labeling in YS and fetal liver progenitors, the Il7r-Cre labeling in adult trMacs (Fig. 1) was considerably higher compared with any progenitor we profiled across fetal development (Fig. 2H; Fig. S2). We reasoned that progenitor labeling alone could not entirely explain the labeling observed in those compartments. To resolve this discrepancy, we next evaluated Il7r-Cre-driven labeling and IL7Rα expression in myeloid-restricted precursors within peripheral tissues during FL stage development. We initiated our investigation at E12.5, the stage at which significant progenitor labeling by Il7r-Cre was initially observed (Fig. 2F,H). At this stage of development, macrophages derived from the extra-embryonic YS (F4/80^hiCD11b^lo) are already present in the tissues, but may be replaced by incoming FL-derived macrophage precursors or monocytes (F4/80^loCD11b^mid/hi; Fig. 3A; Fig. S3) (Hoeffel et al., 2015). F4/80^loCD11b^mid/hi macrophage precursors in fetal peripheral tissues are heterogeneous or low for Ly6c expression and express the fetal macrophage marker CD64 (FCGR1) (Fig. S4A-C; Hoeffel et al., 2015), suggesting their propensity to differentiate into macrophages.

Analysis of Il7r-Cre-driven reporter expression revealed two very clear patterns: first, GFP labeling at E12.5 was significantly higher in all F4/80^lo monocytes compared with F4/80^hi macrophages across all tissues (Fig. 3B, E12.5). Second, cross-tissue comparison revealed higher GFP labeling in F4/80^lo monocytes in brain, lung and epidermis compared with liver (Fig. 3B, E12.5). Across all F4/80^hi macrophages, GFP labeling was highest in the skin (25%; P<0.05 compared with liver F4/80^hi macrophages). A very comparable pattern emerged at E14.5; however, between E12.5 and E14.5, GFP labeling increased slightly across all F4/80^hi macrophage populations, but only significantly for the skin (Fig. 3C, Skin/Epi).

From late gestation into early neonatal development, the patterns of Il7r-Cre-mediated GFP labeling among F4/80^hi macrophages and F4/80^lo monocytes were strikingly different, both between cell types and between tissues (Fig. 3B,C). Beyond E17.5, GFP labeling in liver and lung was higher in F4/80^hi macrophages compared with F4/80^lo monocytes. This ‘switch’ reflected a significant increase in GFP labeling of F4/80^hi macrophages between E14.5 and postnatal day (P)0 and a concomitant decrease in F4/80^lo monocyte labeling (Fig. 3C), coincident with the HSC-dependent contribution to fetal monocytes later in gestation (Hoeffel et al., 2015; Yona et al., 2013). GFP labeling in lung and liver macrophages reached adult levels by P0 (Fig. 3B, Fig. 1F,G; P>0.5). The increase in labeling of F4/80^hi macrophages across fetal development suggested either that Il7r-Cre labeling was occurring cell-autonomously in F4/80^hi macrophages, or that Il7r-labeled fetal monocytes were gradually replacing or contributing to tissue macrophage populations in the lung and liver during fetal hematopoiesis.

In contrast to lung and liver, progression of GFP labeling within the skin and brain across ontogeny relayed a different pattern. F4/80^lo monocytes were barely detectable in the epidermis and brain after E14.5 (Fig. 3A, bottom right panels), as previously described (Hoeffel et al., 2015). GFP labeling of skin macrophages increased robustly from E17.5 to P0 (Fig. 3C, Skin/Epi), and continued to increase postnatally (Fig. 3B,C). Similarly, microglia labeling increased steadily across development, but had still not reached adult levels by P14 (Fig. 3B,C). The disparate GFP labeling that we observed across different tissues suggested a differential involvement of IL7R in the development of different trMac populations across ontogeny.

To gain insight into what was driving the observed pattern of Il7r-Cre-mediated GFP labeling within F4/80^lo monocyte and F4/80^hi macrophage populations across development, we probed Il7r levels beginning at E14.5 (Fig. 4A). Across tissues, F4/80^lo monocytes expressed low but consistently detectable levels of Il7r relative to fetal liver monocyte dendritic progenitors (MDPs), which are known to express Il7r (Hoeffel et al., 2015; Fig. 4A). Circulating peripheral blood Ly6c^hi monocytes also expressed detectable Il7r and were labeled by Il7r-Cre to a similar degree as liver F4/80^lo monocytes (Fig. S4C,D). Increased Il7r mRNA in peripheral tissue F4/80^lo monocytes corresponded with higher Cre-mediated GFP expression (Fig. 4A), as GFP labeling increased in monocytes once they had exited the liver at E14.5 (Fig. 3B). Lung and skin monocytes with the highest GFP labeling also expressed the highest levels of Il7r mRNA. In comparison, and consistent with significantly lower Cre-mediated GFP labeling, Il7r was virtually undetectable in F4/80^hi macrophages (Fig. 4A) at E14.5. Postnatally, at P14, Il7r levels continued to be negligible in macrophages of the lung and liver, but microglia and LCs expressed detectable Il7r (Fig. 4B), driving continued postnatal Cre-mediated recombination in those tissues. These data confirmed the fidelity of the Il7r-Cre model and revealed Il7r expression by fetal myeloid cells during development.

To ascertain the relationship between Il7r-Cre-mediated GFP labeling and IL7Rα surface expression in developing macrophages during fetal development, we profiled surface expression from E12.5 to P14. Considering the high degree of GFP labeling and expression of Il7r mRNA by F4/80^lo monocytes during fetal development, we expected to observe monocytes displaying surface IL7R. Unexpectedly, F4/80^lo monocytes never displayed detectable IL7Rα surface protein at any of the time points examined (Fig. 4C). Surprisingly, robust IL7Rα surface protein was instead observed on prenatal F4/80^hi macrophages in peripheral tissues (lung, brain and skin), whereas minimal surface expression was observed on F4/80^hi macrophages in the liver (Fig. 4D). IL7Rα surface protein on peripheral F4/80^hi trMacs was observed beginning at E12.5, and peaked at E17.5 (Fig. 4D, middle panel), despite significantly lower GFP labeling compared with F4/80^lo monocytes (Fig. 3B,C). The vast majority (>90%) of IL7Rα-expressing macrophages in the brain, lung and skin also co-expressed the common γ chain (CD132; also known as IL2RG; Fig. S5A), indicating expression of a functional IL7 receptor. By P0, macrophages in the liver and lung ceased to express IL7Rα, concomitant with a plateau in GFP labeling by birth (Fig. 3C). In contrast, some proportion of macrophages in the brain and epidermis continued to express IL7Rα protein at the surface postnatally, albeit at lower levels, and IL7Rα surface expression was detectable on epidermal macrophages as late as P14. Continued IL7Rα expression in these particular tissues was consistent with continued postnatal labeling by Il7r-Cre (Fig. 3C) and Il7r mRNA expression (Fig. 4B). These data therefore revealed tightly regulated and highly coordinated expression of IL7Rα in developing macrophages across different tissues.

We hypothesized that F4/80^lo monocytes upregulate Il7r as they exit the liver and migrate to and enter fetal tissues, where they differentiate into F4/80^hi macrophages. IL7Rα surface expression is then rapidly switched on as F4/80^lo monocytes differentiate into F4/80^hi macrophages. To test our hypothesis, we isolated F4/80^lo monocytes from the fetal liver at E14.5 and differentiated them into F4/80^hi macrophages ex vivo in the presence of macrophage colony-stimulating factor (M-CSF). F4/80^loCD11b^mid cells were sorted from FL based on Ly6c expression (Fig. 5A) and cultured with 20 ng M-CSF/ml for 5 days. FL monocytes expressed detectable Il7r (Fig. 4A), but IL7R surface expression was not observed (Fig. 5B-C′). M-CSF induced the differentiation of Ly6c^hiF4/80^loCD11b^hi to F4/80^hi CD11b^mid macrophages that expressed CD64 after only 1 day in culture, and macrophage differentiation plateaued at 3 days (Fig. 5C,E,E′). Differentiated F4/80^hi macrophages could also be differentiated from F4/80^lo monocytes by their larger size and higher cellularity complexity, as defined by forward and side scatter (Fig. S5C′). Remarkably, differentiated F4/80^hi cells also upregulated IL7Rα expression on the cell surface (Fig. 5C,C′; also see Fig. S5D). IL7Rα was co-expressed with the common gamma chain (Fig. 5D,D′; Fig. S5D). In contrast to Ly6c^hi cells, Ly6c^loCD11b^hi FL monocytes displayed limited differentiation into F4/80^hi macrophages, even after 5 days in culture with M-CSF (Fig. S5B). However, additional refinement of these populations by sorting for CD115-expressing cells enhanced macrophage differentiation from Ly6c^hi F4/80^hiCD11b^mid monocytes, and also resulted in macrophage differentiation from Ly6c^loCD115⁺F4/80^hiCD11b^mid cells (Fig. S5C). Both MDP and Ly6c^lo monocytes displayed significantly higher Cre-mediated labeling compared with Ly6c^hi monocytes (Fig. S4D), suggesting distinct developmental pathways. These data reveal that Ly6c^hiCD11b^hi cells in the fetal liver that express Il7r (Fig. 4A; Fig. S4E) but not surface IL7R surface protein (Fig. S4F) have the capacity to differentiate into F4/80^hi macrophages, and that this differentiation is accompanied by upregulation of surface IL7R expression.

The robust and dynamic expression of IL7R by monocyte precursors and developing tissue macrophages suggested that IL7R regulates fetal trMac development from precursors within fetal tissues. To test this hypothesis, we injected the highly specific IL7Rα monoclonal blocking antibody A7R34, or an IgG2A control, into pregnant mice during fetal development (Fig. 6A). Injection of A7R34 during pregnancy completely blocks IL7R signaling, as evidenced by deletion of Peyer's patches in developing embryos following a single injection (Hashizume et al., 2008; Yoshida et al., 1999). We injected pregnant WT mice with A7R34 or IgG2A control at E13.5 and E15.5, coincident with robust IL7Rα expression on fetal macrophages (Fig. 4B), and examined cellularity of macrophages in the epidermis, lung, liver and brain in neonates 9 days later (Fig. 6A). Macrophage cellularity in the epidermis, liver and lung was significantly reduced following this temporally limited blockade of IL7Rα signaling, whereas microglia were unchanged (Fig. 6B).

We further investigated trMac cellularity in Il7r^−/− mice. Although Il7r deletion did not affect YS hematopoiesis (Fig. 2), examination at P19 revealed that Il7r deletion drastically reduced cellularity of trMacs in all tissues (Fig. 6C, Fig. S6A). Il7r deletion did not affect cellularity of other tissue myeloid populations, such as neutrophils (Fig. S6B). By adulthood, differences in macrophage cellularity had normalized (Fig. S6C), despite sustained impairments in lymphocyte development (Fig. S6C′). Cellularity of trMacs was similarly normalized by adulthood following developmental IL7R blockade (Fig. S6D), as were B cells (Fig. S6D′). As trMacs are capable of self-maintenance across adulthood, these data suggest that IL7R plays a unique role in the establishment of these populations, but is not required for their homeostatic maintenance in adulthood. Together with the IL7R blocking experiments, these data indicate that IL7R plays a unique role in the establishment of trMacs during perinatal development.

Our investigation has revealed a completely novel role for IL7R in fetal myeloid development, and specifically in the generation of fetal-specified trMacs. Tracing the progeny of cells mapped by the lymphoid marker IL7Rα revealed that trMacs derived from fetal hematopoiesis were distinctly marked by IL7Rα expression history (Fig. 1). Our comprehensive analysis of Il7r-Cre labeling, Il7r mRNA levels, and surface expression concluded that trMac labeling was not due solely to derivation from an IL7Rα-labeled bipotent progenitor, nor was labeling acquired in adulthood. Instead, our developmental analysis revealed that labeling of trMacs by Il7r-Cre occurred as a result of dynamic stage- and tissue-specific expression of IL7R during macrophage development within tissues during fetal development (Figs 3 and 4). We showed that the CD11b^hiLy6c^hi fetal liver monocytes rapidly upregulated surface IL7R as they differentiated into F4/80^hi macrophages ex vivo (Fig. 5), and observed robust IL7R surface expression on developing macrophages during a limited window of fetal development. Using two different loss-of-function approaches, we further demonstrated that blocking IL7R signaling during the window of robust IL7R expression by developing trMacs impairs their establishment (Fig. 6). In adult BM hematopoiesis, IL7Rα expression exclusively marks lymphoid cells (Schlenner et al., 2010), and the many functions of IL7R signaling in lymphopoiesis are well-delineated (Fry and Mackall, 2005), including proliferation, differentiation and survival. IL7R may therefore regulate similar processes in developing macrophages, but the mechanisms and signaling pathways for IL7R in myeloid cells remain to be established. By revealing a role for IL7R in trMac development, our data identify a function of IL7R during fetal development that extends beyond regulation of the lymphoid lineage, contributing to accumulating evidence that fetal hematopoietic lineage specification is considerably less constrained than adult hematopoiesis (Beaudin et al., 2016; Mebius et al., 2001; Notta et al., 2016).

Common signaling factors, including dependency on the transcription factor PU.1 (also known as SPI1) (Schulz et al., 2012) and signaling downstream of the CSF-1 receptor (Chitu and Stanley, 2017), regulate macrophage development across multiple tissues, whereas other signals regulate and maintain tissue-specific identity and function (Guilliams et al., 2013; Mass et al., 2016; Scott et al., 2018). IL7R appears to function as a regulatory factor across different tissues, but may also have distinct roles in different tissues across ontogeny. Blocking IL7Rα function during a narrow fetal window (E13.5-E15.5) negatively affected cellularity of liver, lung and epidermal macrophages, with no effect on microglia (Fig. 6B). In contrast, germline deletion of IL7Rα in the Il7r^−/− mouse significantly impaired trMac cellularity across all tissues, including microglia, examined at P19 (Fig. 6C). The different effect of total knockout and transient blockade of IL7Rα on trMac development suggests that the window of dependency on IL7R signaling differs between different tissues. The requirement for IL7Rα during postnatal microglia development fits with recent reports of rapid changes in proliferation and apoptosis occurring postnatally in microglia (Askew et al., 2017; Nikodemova et al., 2015), and suggests the possibility that IL7Rα contributes to proliferation or survival of microglia in the transition between the postnatal period and adulthood. Although IL7Rα is not expressed by adult myeloid cells at homeostasis (Schlenner et al., 2010), there have been a handful of reports suggesting that IL7 can promote monocyte activation in the context of autoimmunity and infection (Chen et al., 2013; Gessner et al., 1993), and a bipotent IL7R⁺ lympho-myeloid progenitor has been identified that emerges in the context of malaria infection (Belyaev et al., 2010). Whether autoimmunity or other inflammatory conditions can promote the usage of fetal hematopoietic differentiation pathways to confer distinct functions on myeloid cells during disease states is an interesting hypothesis that remains be tested experimentally.

We used three different lineage-tracing models based on lymphoid markers associated with both increased lymphoid potential (Flk2, Rag1, IL7Rα; Fig. 1) (Boyer et al., 2019; Forsberg et al., 2006; Igarashi et al., 2002; Kondo et al., 1997) and overlapping expression in FL progenitors (Beaudin et al., 2016; Boiers et al., 2013) to trace trMac origin. The stark contrast in labeling between these models allows us to exclude the possibility that labeling of adult trMacs by Il7r-Cre reflected inheritance from an earlier lymphomyeloid progenitor. A clear example is epidermal LCs: if labeling of LCs was due to derivation from lymphomyeloid progenitors, LCs would be robustly labeled not only by Il7r-Cre, but also by Flk2-Cre and Rag1-Cre, as those genes are expressed in IL7R⁺ FL progenitors (Boiers et al., 2013). The high degree of Il7r-Cre labeling of trMacs was similarly not reflected in YS EMPs or YS macrophages (Fig. 2, Fig. S2) and IL7R deletion had no effect on YS EMP cellularity or the development of YS macrophages during embryonic hematopoiesis (Fig. 2D,E), but resulted in significant impairments later in trMac development (Fig. 6C). Together, these data suggest that although IL7R expression labels a multipotent fetal liver progenitor, expression of IL7R on developing tissues macrophages later in gestation regulates their development during later waves of macrophage seeding (De et al., 2018; Ferrero et al., 2018; Tan and Krasnow, 2016).

Our analysis of the IL7R knockout mouse and in response to physiological blockade of IL7R reveal IL7R as a novel regulator of tissue macrophage development. Functional inhibition of IL7R also affected lymphocyte cellularity (Fig. S6C′), as previously reported (Hashizume et al., 2008; Peschon et al., 1994; Yoshida et al., 1999). One caveat is that the impairments in trMac development may be secondary to impaired lymphocyte development. However, despite sustained impairments in lymphocyte development in the IL7R knockout (Fig. S6C′), trMac cellularity recovered by adulthood (Fig. S6C), arguing that lymphocyte impairment alone is not responsible for impaired trMac development. Deletion of IL7R had no effect on YS EMPs or YS macrophage generation, affirming that IL7R plays a more crucial role in fetal macrophage generation at later stages of hematopoiesis.

Fetal F4/80^hi macrophages expressed negligible Il7r message levels (Fig. 4A) yet expressed robust surface IL7R protein (Fig. 4D) and acquired Il7r-Cre labeling across fetal development (Fig. 3C). In lung and liver, for example, increased GFP labeling in F4/80^hi trMacs across fetal development did not reflect Cre-driven reporter expression, as trMacs did not express Il7r during fetal development (Fig. 4A). The most parsimonious explanation for this surprising discrepancy is the initiation of Cre recombination in IL7R-marked F4/80^lo macrophage precursors with subsequent inheritance of the floxed allele by F4/80^hi trMacs upon differentiation. F4/80^hi trMac differentiation is then coincident with the rapid translation of Il7r message into IL7Rα protein and surface display (Fig. 6D). Indeed, our ex vivo differentiation assay revealed that Ly6c⁺ FL monocytes expressing Il7r mRNA (Fig. 4C; Fig. S4E) differentiate into F4/80^hi macrophages that upregulate surface IL7R protein ex vivo (Fig. 5). In vivo, the rapid and dynamic regulation of IL7R protein surface expression may reflect a response to ligand in tissues, as surface expression of IL7R is regulated by IL7 availability (Clark et al., 2014; Wei et al., 2000). The lower surface expression of IL7R on fetal liver F4/80^hi macrophages compared with peripheral tissues (Fig. 4D) in vivo may also reflect surface regulation by immediate sources of IL7 ligand in the fetal liver (Namen et al., 1988).

By tracking GFP expression within fetal myeloid cells in the Il7r-Cre model, we observed the contribution or replacement of GFP-labeled F4/80^lo precursors to F4/80^hi trMacs in the lung and liver, as evidenced by increased labeling in F4/80^hi macrophages across ontogeny (Fig. 3C). Whereas trMacs of the lung and liver exhibited robust labeling (∼40%) that had plateaued at birth (Fig. 3A-C), Il7r mRNA expression (Fig. 4B,D; Mass et al., 2016) and Cre-mediated labeling of LC and microglia continued postnatally, leading to almost complete labeling by Il7r-Cre in adulthood. Although monocytes entering the epidermis and brain expressed both the highest levels of Il7r mRNA and had the greatest degree of Cre-mediated labeling, there was very little monocyte infiltration from E14.5 and beyond (Fig. 4A). Instead, the higher degree of labeling in the epidermis and brain could be attributed to higher and persistent expression of IL7Rα within trMacs peri- and postnatally, leading to increased GFP labeling (Fig. 3). Together, our analysis provides additional support for the contribution of FL F4/80^lo macrophage precursors to specific trMac compartments during late gestation (Guilliams et al., 2013; Hoeffel et al., 2015; Rantakari et al., 2016; Tan and Krasnow, 2016), and reveals how IL7R expression during macrophage differentiation regulates establishment of tissue macrophage compartments.

All animals were housed and bred in the AALAC accredited vivaria at UC Santa Cruz or UC Merced and group housed in ventilated cages on a standard 12:12 light cycle. All procedures were approved by the UCSC or the UC Merced Institutional Animal Care and Use (IACUC) committees. IL7Rα-Cre (Schlenner et al., 2010), Rag1-Cre (Welner et al., 2009) and Flk2-Cre (Benz et al., 2008) mice, obtained under fully executed Material Transfer Agreements, were crossed to homozygous Rosa26^mTmG females (Muzumdar et al., 2007) to generate ‘switch’ lines, all on the C57Bl/6 background. WT C56Bl/6 mice were used for controls and for all expression experiments. Adult male and female mice were used randomly and indiscriminately, with the exception of the FlkSwitch line, for which only males were used because of their high and uniform floxing efficiency. Similarly, mice for developmental analysis were used indiscriminately without knowledge of gender.

Mice were sacrificed by CO₂ inhalation. Gravid uteri removed, and individual embryos dissected. Adult liver, lung, brain and skin were isolated and treated with 1× PBS (+/+) with 2% serum, 0.2-1 mg/ml collagenase IV (Gibco) with or without 100 U/ml DNase1 for 20 min to 2 h. For adult epidermis isolation, ears were first incubated with 1× PBS (+/+) containing 2.4 mg/ml dispase (Gibco) to separate the epidermis, followed by 2 h incubation of the epidermis with 1/mg/ml collagenase. Following incubation, all tissues were passaged through a 16 g needle or a 19 g needle ten times, and then filtered through a 70 µm filter.

Cell labeling was performed on ice in 1× PBS with 5 mM EDTA and 2% serum. Antibodies used are listed in Table S1. Analysis was performed on BD FACS Aria II at University of California-Santa Cruz, and BD FACS Aria III and the University of California-Merced and analyzed using FlowJo.

Transplantation assays were performed as previously described (Beaudin et al., 2014; Beaudin et al., 2016; Smith-Berdan et al., 2015; Ugarte et al., 2015; Boyer et al., 2019). Briefly, sorted Tom⁺ or GFP+ KLS cells were isolated from IL7RαSwitch or Rag1Switch E14.5 fetal liver donors. WT recipient mice aged 8-12 weeks were sublethally irradiated (750 rad, single dose). Under isofluorane-induced general anesthesia, sorted cells were transplanted intravenously. Recipient mice were bled at 4, 8, 12 and 16 weeks post-transplantation via the tail vein and peripheral blood was analyzed for donor chimerism by means of fluorescence profiles and antibodies to lineage markers. Long-term multilineage reconstitution was defined as chimerism in both the lymphoid and myeloid lineages of >0.1% at 16 weeks post-transplantation.

E14.5 fetal livers from WT embryos were harvested and homogenized via trituration. CD11b⁺ cells were enriched by positive selection with CD11b biotin-conjugated antibodies and streptavidin microbeads, using LS columns (Miltenyi). F4/80^lo CD11b^mid cells were sorted based on Ly6c expression or CD115 expression and cultured in a 96-well plate (DMEM, 20% FBS, 1 mM sodium pyruvate, 10 mM HEPES, 0.1 mM 2-mercaptoethanol and 50 mg/ml primocin, Life Technologies). Cells were cultured in triplicate in the presence of 20 ng/ml M-CSF and analyzed at three different time points (days 1, 3 and 5) for the presence of F4/80^hi CD11b^mid macrophages and for surface expression of IL7R, common gamma chain (CD132) and CD64.

RNA isolation from sorted cells was accomplished using the RNeasy mini kit (Qiagen), and cDNA was reverse transcribed from RNA (High capacity cDNA reverse transcription kit, ThermoFisher Scientific). TaqMan probes (TaqMan Gene Expression Assay, ThermoFisher Scientific) were used for qRT-PCR analysis on a StepOnePlus Real-Time PCR System (ThermoFisher Scientific) in comparative CT mode. Samples were run in triplicate and were run with positive (CD43⁺ B220⁺ Pro B-cells), negative (Gr1⁺ CD11b⁺ BM monocytes), and no cDNA controls.

Number of experiments, n, and what n represents can be found in the legend for each figure. Statistical significance was determined by two-tailed unpaired Student's t-test. All data are shown as mean±s.e.m. representing at least three independent experiments.

We thank Drs Hans-Reimer Rodewald and Susan M. Schlenner for the IL7Rα^Cre strain; Drs Patricia Ernst and Terence Rabbitts for the Rag1^Cre strain; Dr T. Boehm for the Flt3^Cre strain; Bari Nazario and the UCSC Institute for the Biology of Stem Cells for flow cytometry support; and David Gravano and the UC Merced Stem Cell Instrumentation Foundry for flow cytometry support. CIRM Facilities awards CL1-00506 and FA1-00617-1 to UCSC.