XPR1 regulates fetal liver macrophage development, identity, and pyrenocyte clearance

Inorganic phosphate (Pi) is essential for diverse functions including fetal growth, development, and metabolism, and its dyshomeostasis can lead to abnormal bone mineralization or ectopic tissue calcifications (Xu et al., 2020; Berndt and Kumar, 2007). Pi homeostasis is tightly regulated by a small number of highly conserved transporter molecules; the phosphate importer family of solute carrier (SLC) SLC20 and SLC34 proteins (Forster et al., 2013); and the only known metazoan phosphate exporter xenotropic and polytropic retrovirus receptor 1 (XPR1) (Giovannini et al., 2013). Pi transporters are critical for all life, and deletion of either Slc20a1, Slc34a2, or Xpr1 in mice leads to growth defects and embryonic or neonatal lethality (Festing et al., 2009; Shibasaki et al., 2009; Xu et al., 2020). As the only phosphate exporter in multicellular organisms, little is known about the function(s) of XPR1. In humans, loss-of-function mutations in XPR1 cause primary familial brain calcification, a neurodegenerative disease where calcification of blood vessels in basal ganglia is a diagnostic criterion (Legati et al., 2015). XPR1 is also highly expressed in the developing murine central nervous system (Yao et al., 2017 and https://www.genepaint.org), and heterozygous deletion of Xpr1 (Xpr1^LacZ/wt) leads to brain vascular calcifications and microgliosis (Maheshwari et al., 2023). Recent reports have solidified the role of XPR1 as a phosphate exporter (Yan et al., 2024; Lu et al., 2024; Zhang et al., 2025; He et al., 2025), while non-export functions of XPR1 in regulating intracellular phosphate homeostasis have also been suggested (Bondeson, 2025). Interestingly, besides its role in regulating phosphate levels, Xpr1 in zebrafish is required for the development and differentiation of microglia and Langerhans cells (LCs) (Meireles et al., 2014). Whether XPR1 is also essential for the development of murine microglia and LCs or other macrophages remains unknown.

Virtually all organs in the body harbor macrophages, which contribute to innate immune defense by recognizing and phagocytosing invading pathogens. Moreover, tissue-resident macrophages (TRMs) perform important organ-specific homeostatic functions (Mass et al., 2023). Red pulp macrophages (RPMs) and Kupffer cells (KCs) are TRMs of the spleen and liver, respectively, where they are implicated in the recycling of iron-containing heme from red blood cells (Kovtunovych et al., 2010). Some types of macrophages also support erythropoiesis by providing developmental signals, shuttling iron-containing ferritin, and facilitating the enucleation of erythroblasts (Li et al., 2021). These populations are known as erythroblastic island macrophages (EBI-Macs) and are found in major hematopoietic organs, such as the adult bone marrow (BM) and the fetal liver (FL). The FL is the primary site of hematopoiesis during gestation from approximately embryonic day 12 (E12) to E15 (Crawford et al., 2010) and of erythropoiesis from E12.5 to E18.5 (Tada et al., 2006). At around E18, embryonic erythropoiesis moves from the liver to the spleen and persists there until early neonatal development, when it shifts to the BM (Wilding Crawford et al., 2010; Tada et al., 2006). Alterations in EBI-Mac function and/or development result in impaired erythropoiesis, leading to anemia and even embryonic lethality (reviewed in Li et al. [2021]). Although not considered EBI-Macs in the steady state, adult KCs and splenic RPMs can transform into EBI-Macs during erythropoietic stress (Sonoda and Sasaki, 2012).

In mice, the majority of TRMs develop during embryogenesis from erythromyeloid precursors in the yolk sac, and in adulthood, they self-renew locally with little to no input from the BM (Ginhoux and Guilliams, 2016; Cox et al., 2021). While most TRMs require colony-stimulating factor 1 receptor (CSF1R) signaling for their development and maintenance (Lelios et al., 2020), other factors that guide their organ-specific development and fate are starting to be discovered. A well-characterized niche is that of KCs in the liver, where expression of Id3, bone morphogenic proteins, and Notch signaling pathway ligands (Dll1 and Dll4) have been found to be critical for their genesis and identity (Bonnardel et al., 2019).

In this study, we investigated the potential role of XPR1 in the development of TRM populations in mice. Using Xpr1-deficient and conditional Xpr1 knock-out mice, we found that XPR1 was essential for the development, identity, and function of FL macrophages and RPMs, both involved in erythroblast maturation and erythrocyte production. The absence of FL macrophages led to an accumulation of interferon (IFN)-activated monocyte/macrophage-like cells in the FL that were unable to clear the expelled nuclei (pyrenocytes) of maturing erythrocytes. Together, these findings reveal a specific role for XPR1 in the development and function of EBI-Macs.

To analyze whether XPR1 plays a role in the development of murine TRMs, we performed flow cytometry on cells from embryonic organs at different time points during embryogenesis. As Xpr1-deficient mice are not viable (Xu et al., 2020; Maheshwari et al., 2023), we used Xpr1^LacZ/+ mice, which carry one functional Xpr1 allele and one non-functional allele containing a LacZ reporter (Fig. S1 A). Crossing Xpr1^LacZ/+ heterozygote mice resulted in no significant difference in the Mendelian ratio between Xpr1^+/+ and Xpr1^LacZ/LacZ embryos, suggesting that perinatal lethality is not due to an early abort during development (Fig. S1 B). We first examined the presence of brain macrophage populations, comprising parenchymal microglia and border-associated macrophages (BAMs), which reside in the meninges, choroid plexus, and perivascular spaces (Mildenberger et al., 2022). No differences in the frequencies of microglia (CX3CR1⁺CD206⁻) or BAMs (CX3CR1⁺CD206⁺) were detected in Xpr1^LacZ/LacZ embryos compared to littermate controls at E14.5, E16.5, and E18.5 (Fig. 1 A and Fig. S1 C). Total cell numbers per brain were slightly reduced at E18.5 (Fig. S1 C), likely due to the smaller overall size of Xpr1^LacZ/LacZ embryos (Xu et al., 2020). Analyses of LCs in the skin produced similar results: we found no significant differences in the frequencies and numbers of LC (F4/80^hi) precursors and only minor changes in Ly6C^hi monocytes in Xpr1^LacZ/LacZ embryos (Fig. 1 B and Fig. S1 D), demonstrating that the reported requirement for Xpr1 in LCs and microglia in the zebrafish model (Meireles et al., 2014) was not conserved in mice.

Conversely, in the embryonic liver, we noted a drastic reduction in the frequency and total cell numbers of F4/80^hi (CD11b^lo) FL macrophages in Xpr1^LacZ/LacZ embryos from E12.5 onward (Fig. S1 E and Fig. 1 C). As Xpr1^LacZ/LacZ embryos are smaller than wild-type (WT) and Xpr1^LacZ/+ control littermates (Xu et al., 2020), correlating with smaller liver weights (Fig. S1 F), we calculated the total cell numbers per milligram tissue at different time points (Fig. 1 C). F4/80^hi FL macrophages were significantly reduced compared to controls; furthermore, we observed a concomitant increase in the frequency and numbers of FL Ly6C^hi monocytes during development (Fig. 1 D).

Similar to the brain and skin, we observed no significant differences in embryonic macrophages in the heart, kidney, and lung when comparing WT, Xpr1^LacZ/+, and Xpr1^LacZ/LacZ embryos (Fig. S1 G). Overall, our data show that XPR1 is required for FL macrophage development but is dispensable for the embryonic development of other TRMs.

Given the observed growth defects and neonatal lethality in Xpr1-deficient embryos, we next sought to identify the specific Xpr1-expressing cell type necessary for FL macrophage development. Published single-cell RNA-sequencing (scRNA-seq) datasets from adult and FL demonstrate low but ubiquitous Xpr1 expression in all major liver cell populations (Fig. S2, A and B, data from Guilliams et al. [2022], Wang et al. [2020]). To investigate whether XPR1 is intrinsically required in hematopoietic cells, we used Vav1^iCre mice, which target the whole hematopoietic compartment, as early as E8.5 (de Boer et al., 2003; Padrón-Barthe et al., 2014). Crossing Vav1^iCre to Xpr1^fl/fl animals produced no viable Vav1^iCreXpr1^fl/fl offspring, indicating that, as with complete Xpr1 deficiency, deletion of Xpr1 in hematopoietic cells and their precursors leads to lethality, while maintaining normal Mendelian frequencies during embryogenesis (Fig. S2 C). Flow cytometry analysis of Vav1^iCreXpr1^fl/fl embryos at E16.5 revealed a profound loss of FL macrophages (Fig. 2 A). This was not due to downregulation of Tim4, as we also noticed the loss of FL macrophages using F4/80 and MerTK (Fig. S2 D). Unlike Xpr1^LacZ/LacZ embryos, we found no differences in embryo or liver weights between control and Cre⁺ embryos even at E18.5 (Fig. 2 B), suggesting that the growth defect observed in the total knock-out was not due to lack of Xpr1 expression in hematopoietic cells. We did not observe differences in yolk sac (YS) CSF1R⁺c-Kit⁺ erythromyeloid progenitors (EMPs) or CX3CR1⁺CD64⁺ YS macrophages at E10.5 (Fig. S2 E), suggesting that FL macrophages, but not other EMP-derived macrophages, are specifically affected in Vav1^iCreXpr1^fl/fl mice. To assess targeting efficiency, we sorted FL monocytes (Ly6C⁺CD11b⁺)—given that FL macrophages were absent—and liver endothelial cells (Lyve1⁺CD45⁻), shown to be partially targeted in Vav1^Cre mice (Georgiades et al., 2002), from E16.5 embryos (Fig. S2 F) and performed qRT-PCR. Sorted monocytes exhibited an almost complete loss of Xpr1 mRNA expression, while endothelial cells exhibited an ∼60% reduction (Fig. S2 G). Notably, liver endothelial cell numbers were not different between control and Vav1^iCreXpr1^fl/fl embryos (Fig. S2 H).

Similar to Xpr1-deficient embryos (Fig. 1 D), frequencies and numbers of FL monocytes (Ly6C^hiCD64⁺) were also significantly increased in E16.5 Vav1^iCreXpr1^fl/fl embryos (Fig. 2 C), suggesting that cell-intrinsic loss of Xpr1 in hematopoietic cells was responsible for the failure of macrophage development and an increase in monocytes. To determine whether liver monocyte and macrophage cell populations changed phenotypically, we performed high-dimensional flow cytometry on FL cells extracted from Xpr1^fl/fl and Vav1^iCreXpr1^fl/fl embryos at E16.5. Clustering on CD64⁺F4/80⁺ cells identified five different populations (Fig. 2, D–F). As previously observed, classical FL macrophages (Tim4^hiF4/80^hi) were absent in Vav1^iCreXpr1^fl/fl animals (Fig. 2, D and E). This analysis further revealed that the increase in CD11b⁺CD64⁺Ly6C^hi monocytes in Cre⁺ embryos was not due to an increase in Ly6C^hi FL monocytes themselves, but rather to the emergence of two alternate monocyte/macrophage-like (Mo/Mac) populations that were largely absent in control embryos (Fig. 2, D and E, labeled as Mo/Mac 1 and 2). These Mo/Macs expressed high levels of CD11b, CD64, CD34, and CD48 and intermediate levels of F4/80, Ly6C, and CD88 (Fig. 2 F and Fig. S2 I). Notably, they did not express the canonical FL macrophage/KC marker Tim4. We next analyzed FLs by immunofluorescence microscopy and found that, while Xpr1^fl/fl control livers contained a dense network of FL macrophages stained for CD64 and F4/80 (Fig. 2 G, left, insert 1 and 3) or IBA1 and TIM4 (Fig. 2 H), Vav1^iCreXpr1^fl/fl livers were almost completely devoid of F4/80⁺ and IBA1⁺TIM4⁺ cells (Fig. 2, G and H). In addition, Cre⁺ livers instead contained large numbers of spherical cells expressing high levels of CD64, reminiscent of monocytes (Fig. 2 G, right, insert 1). Moreover, Vav1^iCreXpr1^fl/fl FLs exhibited a marked increase in TUNEL signal in CD64⁺ cells, indicating elevated apoptosis in monocytes/macrophages lacking Xpr1 (Fig. S2, J and K). These data indicate that classical FL macrophages are largely absent in Vav1^iCreXpr1^fl/fl livers, which instead generate Mo/Macs that fail to acquire a characteristic liver macrophage morphology and phenotype.

The almost complete loss of canonical FL macrophages in Vav1^iCreXpr1^fl/fl embryos prompted us to investigate how XPR1 affects macrophage development. To address this question, we performed scRNA-seq on CD45⁺CD3⁻CD19⁻NK1.1⁻ cells extracted from E15.5 Xpr1^fl/fl and Vav1^iCreXpr1^fl/fl livers, revealing 16 cell clusters (Fig. 3 A). Cell frequencies were largely even between control and Cre⁺ samples except for clusters 6, 7, 10, 12, and 13 (Fig. 3 A and Fig. S3 A). Clusters 6 and 7 expressed Elane, Ms4a3, and Prtn3 and were both classified as granulocyte-monocyte progenitors (GMPs) (Fig. 3 A and Fig. S3 B). Upon closer inspection, the offset UMAP coordinates of clusters 6 and 7 appeared to be due to slightly increased expression of Ngp, Lyz2, Ifitm3, and Ifi27 in Xpr1^fl/fl (cluster 7) versus Vav1^iCreXpr1^fl/fl (cluster 6) (Fig. S3 C). Subsetting and reanalysis of these two clusters confirmed that the two populations were otherwise homogenous, however (Fig. S3 D). Cluster 12 expressed Epx and Prg3 and was classified as eosinophils (Fig. 3 A and Fig. S3 B). Although this cluster appeared to be over-represented in the Vav1^iCreXpr1^fl/fl sample, flow cytometry did not show a difference in eosinophils between control and Cre⁺ conditions (Fig. S3 E).

Cluster 13 was found exclusively in Xpr1^fl/fl, whereas cluster 10 was highly enriched in Vav1^iCreXpr1^fl/fl (Fig. 3 A and Fig. S3 A). Cluster annotation revealed these to be macrophages and monocytes, respectively (Fig. 3 A). To better understand how the loss of Xpr1 affects macrophage development, we further subsetted monocytes (clusters 1 and 10), macrophages (cluster 13), and the precursor populations in clusters 2 (MLP/MDP/GMPs), 3 (MLPs), and 0 (CDP/MDPs). This reanalysis revealed nine Seurat clusters, including a monocyte cluster (cluster 3) present in Vav1^iCreXpr1^fl/fl livers that was virtually absent in control livers (Fig. 3 B and Fig. S3 F). This cluster was enriched with IFN-regulated genes. We also noted the loss of cluster 5 in Cre⁺ compared to control samples; this cluster expressed high levels of Cd74, Hpgd, Id3, Ccr2, and Klf2 and is likely an FL macrophage precursor population (Fig. 3 B and Fig. S3 F). Lastly, we annotated cluster 8 (erythroid progenitors), which was rich in hemoglobin genes and genes involved in heme and iron catabolism/transport and was strongly increased in Cre⁺ livers. Comparing the monocyte and macrophage clusters using the E15 KC core gene list published by Bonnardel et al. (2019), we found that only bona fide FL macrophages/KCs (cluster 7) in control mice expressed high levels of Timd4, Clec4f, Cdh5, Mrc1, and Nr1h3 (Fig. 3 C). In contrast, monocytic cells from clusters 2, 3, and 6 expressed high levels of monocyte/macrophage-related genes like Ccr2, Ly6c2, S1006, and Itgam (Fig. 3 D). While previous studies demonstrated heterogeneity within FL macrophages (Mukherjee et al., 2021; Kayvanjoo et al., 2024), we found that all FL macrophages were absent in Vav1^iCreXpr1^fl/fl mice. Overall, these data indicated that loss of Xpr1 in Vav1^iCre-expressing cells resulted in dysregulated FL macrophage development.

We further compared our data to published KC/FL macrophage–associated transcription factors (Bonnardel et al., 2019) and found that core transcription factors such as Rxra and Zbtb4 were only expressed in fetal macrophages (cluster 7) in Xpr1^fl/fl mice. However, even though IFN-responsive monocytes (cluster 3) exhibited some expression of the liver macrophage transcription factors Maf, Spic, Nr1h3, and Id3, they failed to develop into true FL macrophages (Fig. 3 E).

To further investigate the hypothesis that loss of XPR1 affects macrophage development, we performed RNA trajectory inference using RNA velocity (Bergen et al., 2020). Compared to Xpr1^fl/fl control livers, the trajectory of cells in Vav1^iCreXpr1^fl/fl livers was directed towards the IFN-signature monocytes (cluster 3), suggesting an alternative developmental pathway compared to normal liver monocytes (clusters 2 and 6) or macrophages (cluster 7) (Fig. S3 G).

Lastly, we isolated FL cells from E14.5 control (Xpr1^fl/+ and Xpr1^fl/fl) and Vav1^iCreXpr1^fl/fl embryos and cultured them with colony-stimulating factor 1 (CSF-1) for 7 days to generate FL-derived macrophages (FLDMs). This showed that the generation of Vav1^iCreXpr1^fl/fl FLDMs was impaired compared to control livers (Fig. S3 H). Moreover, an F4/80⁻CD11b⁺ population was more pronounced in the Vav1^iCreXpr1^fl/fl sample, consistent with our in vivo findings of altered macrophage development (Fig. S3 H).

Altogether, these results indicate that the transcriptional program leading to the generation of FL macrophages was dysregulated in the absence of Xpr1 and, instead, an alternative monocyte/macrophage-like population developed.

To investigate whether loss of FL macrophages in Vav1^iCreXpr1^fl/fl embryos leads to altered erythropoiesis, we analyzed blood smears from E15.5 embryos. No difference in the frequency of nucleated erythrocytes was found between Cre⁺ and Cre⁻ embryos (Fig. 4 A). Equally, the number of red blood cells (RBCs), hemoglobin, hematocrit, blood cell volume, or hemoglobin per RBC were also unaltered between control and Vav1^iCreXpr1^fl/fl mice at E15.5 (Fig. 4 B). As a comparison, we depleted embryonic macrophages in WT embryos by administration of anti-CSF1R antibody to pregnant dams at E8.5 (Fig. S4 A). This direct depletion of embryonic macrophages led to increased nucleated RBCs and altered blood parameters, including lower RBC count, hemoglobin, and red cell distribution width (Fig. S4, B and C). These findings suggest that although XPR1 is involved in macrophage development, the loss of canonical FL macrophages in Vav1^iCreXpr1^fl/fl embryos does not affect the enucleation of erythroblasts.

While no difference in enucleation of RBCs was found, we next assessed whether the expelled nuclei (known as pyrenocytes) could be phagocytosed and cleared by macrophages in Vav1^iCreXpr1^fl/fl versus Xpr1^fl/fl embryos. Pyrenocyte nuclei are denser than normal nuclei due to nuclear condensation prior to expulsion, making them easily distinguishable by histology (Moras et al., 2017). E16.5 Vav1^iCreXpr1^fl/fl livers stained for CD64 and DAPI demonstrated large, dense DAPI⁺ nuclei clusters throughout the tissue (Fig. 4 C). Closer inspection of these clusters revealed nuclei engulfed by CD64⁺ cells (Fig. 4 C, insert), some of which contained upwards of 16 individual nuclei. Although we identified some nuclei-containing CD64⁺ cells in Xpr1^fl/fl control livers, these were less numerous and contained fewer nuclei (Fig. 4 C).

Expelled pyrenocytes are digested by cathepsins and DNase II in the lysosomes of EBI-Macs (Kawane et al., 2001). Importantly, DNase II-deficient FL macrophages do not clear pyrenocyte DNA, leading to STING-dependent type I IFN production (Ahn et al., 2012). Given our observation that Mo/Macs in Vav1^iCreXpr1^fl/fl livers have a type I IFN signature (Fig. 3 B and Fig. S3 F), we wondered if the accumulation of nuclei in these cells was due to an inability of the non-canonical Mo/Macs to digest pyrenocyte DNA. Investigating our scRNA-seq data, we found that, in contrast to Xpr1^fl/fl FL macrophages (cluster 7), Vav1^iCreXpr1^fl/fl IFN-activated Mo/Macs (cluster 3) expressed lower levels of Dnase2a and were comparable to levels found in other non-macrophage populations (Fig. 4 D). These non-canonical Vav1^iCreXpr1^fl/fl FL macrophages also expressed cathepsins at lower levels than Xpr1^fl/fl macrophages (Fig. 4 E). Altogether, these data indicate that XPR1 is required for the development of bona fide FL erythroblastic macrophages and, in their absence, pyrenocytes are not cleared by other macrophage/monocyte populations, leading to type I IFN signaling.

To circumvent the developmental role of XPR1 in FL macrophages, we generated Mrc1^CreXpr1^fl/fl animals, where Xpr1 is deleted in differentiated CD206⁺ macrophages but not in EMPs or in monocytes. CD206 (encoded by Mrc1) is expressed by most KCs in development and is also highly expressed by liver sinusoidal endothelial cells (Guilliams et al., 2022), which is supported by our scRNA-seq data, flow cytometry, and immunohistochemistry from E15.5 embryos (Fig. S5, A–C). Analysis of embryonic livers revealed largely unchanged macrophage frequencies between Mrc1^CreXpr1^fl/fl and Xpr1^fl/fl embryos (Fig. 5 A). However, a Tim4^lo population was present in Mrc1^CreXpr1^fl/fl mice in addition to the Tim4^hi cells seen in control embryos (Fig. S5 D). To check gene targeting efficiency, we sorted macrophages, Ly6C^hi monocytes, and endothelial cells from the FL of E15.5 embryos and performed qPCR. Xpr1 mRNA was significantly reduced in Mrc1^CreXpr1^fl/fl compared to Xpr1^fl/fl macrophages (Fig. 5 B). Similarly, Xpr1 expression was highly reduced in Mrc1^CreXpr1^fl/fl endothelial cells and only marginally reduced in Ly6C^hi monocytes (Fig. 5 B). We next assessed whether Mrc1^CreXpr1^fl/fl embryos displayed a similar impairment in pyrenocyte clearance as seen in Vav1^iCreXpr1^fl/fl embryos above. We indeed noted increased numbers of pyrenocyte clusters in Mrc1^CreXpr1^fl/fl livers (Fig. 5 C). Although the phenotype was less pronounced than in Vav1^iCreXpr1^fl/fl embryos, the findings nevertheless support the conclusion that XPR1 is intrinsically required in macrophages for efficient clearance of expelled nuclei during erythropoiesis. This was unlikely due to impaired phagocytic capacity per se, as FL macrophages from Mrc1^CreXpr1^fl/fl and Xpr1^fl/fl mice exhibited comparable uptake of pHrodo Red Zymosan bioparticles ex vivo (Fig. S5 E). However, similar to Vav1^iCreXpr1^fl/fl mice, Dnase2a expression was significantly reduced in Mrc1^CreXpr1^fl/fl FL macrophages at E15.5 (Fig. 5 D), suggesting a defect in the degradation of pyrenocytes.

Next, to better understand how Xpr1 deletion in FL macrophages affects their function, we sorted F4/80⁺ macrophages (including both Tim4^hi and Tim4^lo populations) from E15.5 Xpr1^fl/fl and Mrc1^CreXpr1^fl/fl FLs and performed bulk RNA-seq. Principal component analysis (PCA) of the sorted cells revealed a distinct separation of the genotypes (Fig. 5 E). 6,104 genes were significantly changed in macrophages from Mrc1^CreXpr1^fl/fl FLs when compared to Xpr1^fl/fl controls (Fig. 5 F and Fig. S5 F). Of these, 3,089 were significantly upregulated and 3,015 were downregulated. Among the differentially expressed genes (DEGs), we observed a profound reduction in FL/KC-identity genes including Vsig4, Clec4f, Timd4, and Cdh5 in the absence of Xpr1 (Fig. 5 G). Concurrently, IFN-stimulated genes Oasl1, Irf7, Rsad2, Iigp1, and Helz2 were increased in Mrc1^CreXpr1^fl/fl versus Xpr1^fl/fl macrophages. The source of type I IFNs in this model remains unclear, as Ifn gene expression was undetectable in the scRNA-seq data in the macrophages. Gene ontology analysis of genes upregulated in Mrc1^CreXpr1^fl/fl also revealed a strong association with the biological processes of “innate immune regulation” and “interferon signaling” (Fig. S5 G), consistent with the observed pyrenocyte accumulation and our findings in Vav1^iCreXpr1^fl/fl livers.

By comparing the top 50 DEGs of EBI-Macs (Li et al., 2019), we noted that although the majority of the EBI-Mac signature was lost in the absence of Xpr1 (Fig. 5 H), several FL macrophages/KC identity genes (such as Spic, Cmbl, Cadm1, and Nr1h3) (Sakai et al., 2019) were increased in Mrc1^CreXpr1^fl/fl macrophages. Spic, for instance, is essential for the differentiation of iron-recycling macrophages (Haldar et al., 2014); however, its increased expression in Mrc1^CreXpr1^fl/fl macrophages was insufficient to restore EBI-macrophage identity. These findings indicate that XPR1 is required to maintain the EBI-Mac transcriptional program.

In contrast to Vav1^iCreXpr1^fl/fl embryos, which die shortly after birth, Mrc1^CreXpr1^fl/fl mice survive into adulthood. We therefore investigated the fate of liver macrophages in this model, noting significantly reduced KCs in adult Mrc1^CreXpr1^fl/fl livers compared to control livers (Fig. 5 I). Among the remaining F4/80⁺ liver macrophages, we noted both Tim4^hi and Tim4^lo subsets in Mrc1^CreXpr1^fl/fl mice (Fig. S5 H), which correlated with our embryonic data.

Since Xpr1 was also deleted in endothelial cells in the Mrc1^CreXpr1^fl/fl model (Fig. 5 B), we sought to determine whether the reduction in KCs in adult Mrc1^CreXpr1^fl/fl animals was due to the intrinsic loss of XPR1 in macrophages rather than endothelial cells. To address this, we analyzed Siglec1^CreXpr1^fl/fl animals, in which Siglec1 (CD169)-expressing macrophages are targeted, but endothelial cells remain unaffected. These mice survived into adulthood and exhibited a significant reduction in KCs (Fig. S5 I), similar to Mrc1^CreXpr1^fl/fl animals. While XPR1 may have distinct roles in various cell types, including endothelial cells, these data strongly suggest a macrophage-intrinsic requirement for XPR1 in KC development and maintenance.

Given the overlapping transcriptional signatures and functional roles of FL macrophages and RPMs (Bonnardel et al., 2019), we investigated whether RPMs were affected in the Vav1^iCreXpr1^fl/fl model. Similar to FL macrophages at E15.5, we found a marked reduction of RPMs in Vav1^iCreXpr1^fl/fl spleens at E18.5 (Fig. 6 A), and we also observed the previously identified increase in monocytes in Vav1^iCreXpr1^fl/fl E18.5 spleens compared to control embryos (Fig. 6 B). These findings suggest that XPR1 might be universally required for the development and function of all EBI-Macs, or at least macrophages with functions involved in RBC development and iron recycling.

To address this hypothesis, we analyzed spleens and BM of adult Mrc1^CreXpr1^fl/fl mice. We observed a dramatic reduction of RPMs in adult Mrc1^CreXpr1^fl/fl compared to control animals (Fig. 6 C). Unlike the increase of splenic monocytes in Vav1^iCreXpr1^fl/fl embryos, however, monocytes in adult Mrc1^CreXpr1^fl/fl animals were unchanged compared to controls (Fig. 6 D). In the BM, we saw an overall reduction of F4/80^hiMerTK⁺ macrophages (Fig. 6 E, top), among which there was a complete loss of the CD163⁺Tim4⁺ subset (Fig. 6 E, bottom). CD163 is a hemoglobin-haptoglobin scavenger receptor that is highly expressed on EBI-Macs including RPMs and KCs (Li et al., 2019). Similar to what was observed in the spleen, monocytes were not affected in the BM of Mrc1^CreXpr1^fl/fl mice (Fig. 6 F). The maintenance of heart macrophages was not dependent on XPR1 (Fig. S5 J); however, we observed partial reductions in LCs, kidney, and alveolar macrophages in adult Mrc1^CreXpr1^fl/fl mice (Fig. S5, K–M), suggesting that these macrophages were also sensitive to the loss of XPR1. Overall, our data show that XPR1 is a novel factor required for the development and function of EBI-Macs.

Here, we show a requirement for XPR1 in EBI-Mac development and function. Deletion of Xpr1 in Vav1⁺ hematopoietic cells led to a complete loss of FL macrophages and splenic RPMs and the appearance of more undifferentiated, inflammatory Mo/Macs. Furthermore, deletion of Xpr1 in CD206-expressing cells resulted in altered FL macrophage transcriptional identity and function, and reduced KC, RPM, and BM macrophage numbers in adult mice. The absence of FL macrophages in anti-CSF1R antibody depletion models or gene-deficient mice (Maea^−/−, Tnfrsf11a^Cre/+Spi1^f/f) results in disturbed erythropoiesis (Soni et al., 2006; Kayvanjoo et al., 2024). Here, in the absence of XPR1 and/or FL macrophages, erythrocyte development appeared normal, suggesting that the alternate macrophages compensate for their absence. Indeed, the signals inducing erythroblast enucleation remain elusive but have been proposed to include soluble factors, cell-to-cell contact with macrophages, and erythroblast intrinsic factors such as Emp and Klf1 (reviewed in Moras et al. [2017]). Regardless of whether FL macrophages (or the alternate Mo/Macs) provide the enucleation signal, we demonstrate that XPR1 in FL macrophages is required for pyrenocyte clearance, which was impaired in our models. However, how Xpr1 expression is linked to genes associated with pyrenocyte clearance, including Dnase2, Ctsl, and Ctsd, remains to be determined.

Id3 is a lineage-determining transcription factor for KCs, and its absence leads to partial loss of these cells (Mass et al., 2016). In Vav1^iCreXpr1^fl/fl livers, we noted the loss of an Id3-expressing population that is likely an FL macrophage precursor. Although Id3 was expressed at high levels in the alternate Mo/Macs, this population failed to develop into true FL macrophages, suggesting that other factors in addition to Id3 drive the transcriptional KC program. Similarly, Maf, a transcription factor critical for terminal macrophage differentiation (Sakai et al., 2019), and expression of EBI-Mac genes (including Vcam1, Mrc1, Csf1r, and Sell) (Kusakabe et al., 2011), was highly expressed in the alternate Mo/Macs. Yet, we found no expression of these EBI signature genes in Maf-expressing alternate Mo/Macs, suggesting that XPR1 is a non-redundant factor for the FL macrophage/KC-specific transcriptional program. Furthermore, the alternate Mo/Macs lacking Xpr1 displayed a type I IFN-responsive gene signature. Whether these macrophages themselves produce type I IFNs, similar to Dnase2^−/− macrophages in the FL, which fail to degrade pyrenocyte DNA (Yoshida et al., 2004), remains to be shown.

We observed that deletion of Xpr1 in CD206-expressing KCs resulted in a macrophage population that was heterogenous for Tim4 in both embryos and adults. Tim4 is an apoptotic receptor for “eat-me” signals, is expressed early during FL macrophage/KC development, and is among the top KC markers. It was previously shown that, following the depletion of KCs, infiltrating monocytes quickly upregulate KC signature genes such as Clec4f, Nr1h3, Id3, and Spic; however, Timd4/Tim4 levels on repopulating macrophages did not reach pre-depletion levels even after 1 mo (Bonnardel et al., 2019; Sakai et al., 2019; Scott et al., 2016). This suggests that Tim4⁻ liver macrophages are monocyte-derived and/or immature KCs. Indeed, KC-specific deletion of Nr1h3, which encodes the nuclear cholesterol receptor and KC lineage-determining transcription factor Liver X receptor α (Sakai et al., 2019; Scott et al., 2018), led to a similar phenotype of Tim4⁺ and Tim4⁻ KCs. In contrast to our Mrc1^CreXpr1^fl/fl model, however, deletion of Nr1h3 in Clec4f-expressing KCs did not lead to reduced KC numbers in the liver (Sakai et al., 2019). Our data therefore indicate that XPR1 is required for the survival and homeostasis of fetal macrophages/KCs and that, in its absence, monocytes fail to fully develop into bona fide KCs.

The mechanism by which XPR1 promotes EBI-Mac development and programming remains unclear. XPR1 has only recently been shown to be a definitive phosphate exporter (Yan et al., 2024; Lu et al., 2024; Zhang et al., 2025; He et al., 2025). SPX-domain-containing proteins (such as eukaryotic phosphate signaling proteins and transporters including XPR1) in plants have been shown to interact with transcription factors (Wild et al., 2016). Furthermore, as high concentrations of extracellular phosphate are cytotoxic, increased accumulation of phosphate within Xpr1-deficient macrophages or their precursors could lead to apoptosis (Alexander et al., 2022). However, it is not clear why EBI-Macs in particular would be sensitive to dysregulated phosphate homeostasis. Whether the intracellular phosphate-sensing ability of XPR1 is required for its role in EBI-Mac development is also not resolved. We show that the absence of Xpr1 in uncommitted precursors leads to an arrest in the differentiation to EBI-Macs, causing them to acquire an alternate Mo/Mac program. A role for XPR1 as an EBI-Mac developmental and identity factor is also supported by our finding in Mrc1^CreXpr1^fl/fl mice, where deletion of Xpr1 in Mrc1/CD206-expressing cells led to a downregulation of canonical EBI-Mac genes. In addition to its function in EBI-Macs, we demonstrate that XPR1 also plays a role in the maintenance of AMs and kidney macrophages in adulthood. In zebrafish, the development of some TRMs was also shown to be impaired in Xpr1b mutants (Meireles et al., 2014), demonstrating a specific role for XPR1 in the macrophage lineage across species.

Taken together, we describe that XPR1 is critical for the differentiation, fate, and function of murine EBI-Macs. These data may open new avenues for investigating the function of XPR1 in mononuclear phagocytes and its role in supporting the EBI-Mac niche in development, health, and disease.

C57BL/6JRj mice were purchased from Janvier Labs. The Xpr1^LacZ (C57BL/6N-Xpr1^{tm1a(KOMP)Wtsi}) mouse strain was obtained through the Knock-Out First Consortium (KOMP repository) (MGI accession no. 4362650). These mice carry a transgenic LacZ-Neo cassette upstream of exon 2 of Xpr1, flanked by FRT sites (Fig. S1 A). Exon 2 is flanked by loxP sites. Heterozygous Xpr1^LacZ/+ mice carry one functional Xpr1 allele and one non-functional Xpr1 allele with a LacZ reporter, while homozygous Xpr1^LacZ/LacZ mice are complete knock-outs for Xpr1. Crossing a Xpr1^LacZ mouse to a Flp-recombinase mouse excises the LacZ-Neo cassette, resulting in a conditional Xpr1^fl strain (Fig. S1 A).

Mrc1^Cre mice were generated by inserting a transgenic construct into Exon 30 of the Mrc1 locus using CRISPR technology, as described previously (Van Hove et al., 2025). The targeting construct contained an internal ribosomal entry site, the codon-optimized iCre sequence, self-cleaving 2A peptide, and eYFP sequence and was synthesized by GenScript. Fertilized C57BL/6J oocytes were injected with linear-dsDNA repair targeting fragment (10 ng/μl) and Cas9 RNP (50 ng/μl). Subsequent litters were genotyped for the transgenic insertion and backcrossed to C57BL/6JRj mice for at least three generations. The whole insertion region, including surrounding genomic DNA, was sequenced to verify correct integration.

Vav1^iCre (de Boer et al., 2003), CD169^Cre (Karasawa et al., 2015), and Mrc1^Cre mice were bred in-house. All strains used were backcrossed onto the C57BL/6JRj strain for at least 10 generations. All Cre strains were used as heterozygotes unless otherwise indicated. Mice were kept in individually ventilated cages under specific-pathogen-free conditions with a 12-h light-dark cycle, under controlled temperature (21–24°C) and humidity (30–70%). All experimental animal procedures at the University of Zurich were performed in accordance with the Swiss Federal regulations and approved by the Cantonal Veterinary Office of Zurich.

2- to 5-month-old mice were mated overnight and separated early the next morning. The morning after mating was considered E0.5.

Pregnant dams were euthanized by CO₂ inhalation and the embryo-containing uterus was removed. For blood collection, embryos were decapitated, and the blood was collected into 3 μl of 0.5 M EDTA in a 24-well plate. Embryonic organs were removed under a stereoscopic dissection microscope. Organs were finely dissociated using scissors and digested in HBSS (with Ca²⁺/Mg²⁺) containing 2% FCS, 0.4 mg/ml Collagenase IV (Worthington), and 0.04 mg/ml DNAse I (Sigma-Aldrich) for 30 min in a shaking incubator at 37°C. Samples were washed in cold PBS and erythrocytes were lysed in RBC-lysis solution (155 mM NH₄Cl, 11 mM KHCO₃). Samples were washed with PBS and subsequently ready for flow cytometric staining.

Adult mice were euthanized by CO₂ inhalation and perfused with cold PBS. Mouse livers were dissociated with scissors and digested in HBSS (with Ca²⁺/Mg²⁺) containing 2% FCS, 1 mg/ml Collagenase A (Sigma-Aldrich), and 0.04 mg/ml DNAse I for 30 min in a shaking incubator at 37°C. Mouse spleens were dissociated with scissors and digested in HBSS (with Ca²⁺/Mg²⁺) containing 2% FCS, 0.4 mg/ml Collagenase IV, and 0.04 mg/ml DNAse I for 30 min in a shaking incubator at 37°C. Samples were washed in cold PBS and erythrocytes were lysed in RBC-lysis solution. Samples were washed with PBS and were subsequently ready for flow cytometric analysis.

E14.5 FLs were homogenized with a P1000 pipette and 2.5 × 10⁶ cells seeded into each well of a 6-well plate in DMEM supplemented with 10% FCS (Thermo Fisher Scientific), 100 U/ml pen/strep (Thermo Fisher Scientific), and 20 ng/ml CSF-1 (Peprotech). Cells were incubated at 37°C and 5% CO₂. On day 4 after seeding, an equal volume of fresh media supplemented with CSF-1 was added to each well. On day 7 after seeding, cells were washed in PBS, scraped off the plate and analyzed by flow cytometry.

Cells were incubated with anti-mouse CD16/32 (clone 93, 101310; BioLegend) and Zombie NIR Live/Dead fixable viability stain (BioLegend) for 20 min at 4°C in the dark to block the Fc-receptor and stain dead cells, respectively. Cells were then stained with fluorochrome-conjugated antibodies in FACS Wash (PBS containing 2% FCS and 2 mM EDTA) for 30 min at 4°C in the dark. Antibodies used were I-A/I-E (clone M5/114.15.2; PacificBlue, 107620; BioLegend; BV605, 107639; BioLegend; BV510, 107636; BioLegend), CD31 (clone 390, BUV805, 741949; BD; clone MEC13.3; PE, 102507; BioLegend), CD11b (clone M1/70, BUV737, 612800; BD), CD11c (clone N418, PE/Cy5.5, 35-0114-82; eBioscience; PE/Cy7, 117318; BioLegend; BV570, 117331; BioLegend; APC, 117310; BioLegend), CD45 (clone 30-F11, BUV395, 564279; BD; AF700, 103128; BioLegend), Ly6G (clone 1A8, BV650, 127641; BioLegend; BV785, 127645; BioLegend; V450, 560603; BD; BUV563, 565707; BD), Ly6C (clone HK1.4, BV711, 128037; BioLegend; APC, 128016; BioLegend; clone AL-21, FITC, 553104; BD), Sca1 (clone D7, BV510, 108129; BioLegend), CD48 (clone HM48-1, AF700, 103426; BioLegend), CD34 (clone SA376A4, BV421, 152208; BioLegend), CD115 (clone AFS98, PE/Cy7, 25-1152-82; eBioscience; APC, 17-1152-82; eBioscience), CD88 (clone 20/70, BV750, 747227; BD), CD90.2 (clone 30-H12, APC, 140312; BioLegend), CD49b (clone DX5, APC, 108910; BioLegend), CD62L (clone MEL-14, BUV737, 612833; BD; BV570, 104433; BioLegend), CD117 (clone 2B8, PE-EFluor610, 61-1171-82; eBioscience), CD3 (clone 17A2, APC, 100236; Biolegend), NK1.1 (clone PK136, BV711, 108745; BioLegend), B220 (clone RA3-6B2, APC, 103212; BioLegend), CD19 (eBio1D3, APC, 17-0193-82; eBioscience), CD150 (clone TC15-12F12.2, BV785, 115937; BioLegend), SiglecF (clone E50-2440, PE/CF594, 562757; BD), CD64 (clone X54-5/7.1, PE, 139304; BioLegend; BV421, 139309; BioLegend; BV711, 139311; BioLegend), F4/80 (clone BM8, PE/Cy5, 123112; BioLegend; BV510, 123135; BioLegend), CD169 (clone SER-4, BV605, 142413; Biolegend; AF488, 53-5755-82; eBioscience), CD163 (clone TNKUPJ, PerCP-EFluor710, 46-1631-82; eBioscience; PE, 12-1631-82; eBioscience), CD206 (clone C068C2, AF700, 141734; BioLegend; PE, 141706; BioLegend), Lyve1 (clone ALY7, eF450, 48-0443-82; eBioscience; AF488, 53-0443-80; eBioscience), CD38 (clone 90, PE/Dazzle594, 102729; BioLegend), CX3CR1 (clone SA011F11, BV605, 149027; BioLegend; BV510, 149025; BioLegend; PE/Dazzle594, 149013; BioLegend, Gr-1 (clone RB6-8C5, PE, 12-5931-83; eBioscience; BUV805, 741920; BD), MerTK (clone DS5MMER, SB780, 78-5751-82; eBioscience; PE/Cy7, 25-5751-82; eBioscience), Tim4 (clone F31-5G3, PE/Cy7, 130009; BioLegend; AF647, 130007; BioLegend; BV805, 749133; BD; BUV395, 745685; BD), Ter119 (clone Ter119, APC, 17-5921-81; eBioscience; BV650, 116235; BioLegend), VCAM1 (clone 429, FITC, 553332; BD), CD71 (clone RI7217, APC, 113819; BioLegend; clone C2, RB780, 755614; BD), and EPCAM (clone G8.8, PE/Cy7, 118216; BioLegend).

All flow cytometry data were acquired on a BD LSRII Fortessa, BD FACSymphony, or Cytek Aurora and analyzed with FlowJo v10 software (Tree Star). Flow cytometric cell sorting was performed on either a BD FACSAria III or a BD FACSymphony S6 using a 100 µm nozzle. Cells were sorted into 100% FCS before adjusting the volume with PBS. Sorted cells were pelleted by centrifugation, the supernatant aspirated, and the cells resuspended in RNA lysis buffer (Zymo Research).

Dead cells and doublets were excluded from all downstream analyses. Raw flow cytometry data were compensated using FlowJo, followed by transformation and normalization in R. Dimensionality reduction was performed using the UMAP algorithm. The FlowSOM clustering algorithm was used for population clustering. Cluster frequencies and heatmaps were generated in R.

Peripheral blood was analyzed on a Scil Vet ABC Plus blood counter. Blood smears were made by streaking 6 μl of blood along a glass slide using a coverslip. Blood smears were imaged on a Zeiss AxioScan Z1.

FL cells were isolated and stained for flow cytometry as described above. Following staining, cells were resuspended in RPMI with 10% FBS and incubated with pHrodo Red Zymosan BioParticles (P35364; Invitrogen) for 1.5 h at 37°C and 5% CO₂. After the incubation, cells were immediately analyzed by flow cytometry.

Livers from mouse embryos were fixed with 4% paraformaldehyde for 6 h at room temperature. Livers were rinsed in PBS, then cryoprotected in 30% sucrose/PBS for 48 h at 4°C. Samples were embedded in OCT (Sakura). Sections were cut at 10 or 20 µm thickness using a Hyrax C60 cryostat (Zeiss) and transferred onto Superfrost Plus slides (Thermo Fisher Scientific). For staining, sections were blocked and permeabilized by incubation in staining buffer (1% normal donkey serum or 3% normal goat serum, 0.5% Triton X-100, and 0.5% BSA in PBS) for 30 min at room temperature. Slides were incubated with primary antibodies overnight at 4°C in a staining buffer (3% normal goat serum in PBS). Primary antibodies used were F4/80 (clone BM8, 123101; BioLegend; clone CI:A3-1, MCA497A488; Bio-Rad), CD64 (clone AT152-9, MCA5997; Bio-Rad), IBA1 (ABIN2857032; Antibodies Online), TIM4 (AF2826; R&D Systems), CD31 (AF3628; R&D Systems), and CD206 (GTX42265; GeneTex). Samples were washed three times in PBS. For secondary antibody staining, slides were incubated for 2 h at room temperature with secondary antibody diluted in staining buffer. Secondary antibodies used were goat anti-rat AF488 (A11006; Thermo Fisher Scientific) and goat anti-rat AF647 (A21247; Thermo Fisher Scientific), donkey anti-rabbit AF488 (A32790; Thermo Fisher Scientific), donkey anti-goat AF555 (A21432; Thermo Fisher Scientific), and donkey anti-rat AF647 (A78947; Thermo Fisher Scientific). Slides were washed three times and sections mounted with DAPI-containing ImmunoSelect mounting medium (Dianova) or stained for 30 min at room temperature with Hoechst 33342 (1:5,000, H3570; Thermo Fisher Scientific before mounting with ImmunoSelect mounting medium (Dianova). Images were acquired on a Leica Stellaris 5 (objective: HC PL APO CS2, 20× magnification, 0.75 numerical aperture; detectors: Power HyD S; acquisition software: Leica LSX) or Zeiss LSM980 (objective: Plan-Apochromat, 10× magnification, 0.45 numerical aperture; camera: Zeiss Axiocam 506 mono; acquisition software: ZEN 3.3 Blue Edition) or Evident FV4000 (objective: UCPLFLN 20×, 0.7 numerical aperture or UPlanXApo 40×, 0.95 numerical aperture; acquisition software: cellSENS). All images were acquired at room temperature. Images were processed using FIJI v2.9 software.

TUNEL staining was performed using the ApopTag Red In Situ Apoptosis Detection Kit (S7165; Sigma-Aldrich) according to the manufacturer’s instructions for fluorescent staining of tissue cryosections. For co-staining with antibodies, the TUNEL staining was performed first, followed by the antibody staining as previously described. Images were acquired on an Evident Fluoview 4000 microscope and TUNEL mean intensity within the CD64⁺ area was quantified using QuPath (version 0.5.1) imaging analysis software.

E15.5 embryos were extracted from pregnant dams and FL cell suspensions prepared as described above. Embryos were genotyped and three Xpr1^fl/fl and three Vav1^iCreXpr1^fl/fl samples pooled. CD45⁺Lin^-(CD19⁻CD3⁻B220⁻Ter119⁻CD49b⁻CD90.2⁻) cells were sorted as described above. The quality of the single-cell suspensions was assessed using a hemocytometer under a Leica DM IL LED Fluo microscope, and the quantity was determined using an automated cell counter (LUNA-FX7; Logos). Approximately 17,000 cells per sample were loaded onto the 10x Chromium X platform. Library preparation followed the manufacturer’s guidelines specified in the Chromium Single-Cell 3′ Reagent Kits User Guide (v3.1 Chemistry Dual Index). For sequencing, the resulting libraries were processed on an Illumina NovaSeq 6000 SP Flow Cell. Sequencing parameters were set according to 10x Genomics recommendations, using paired-end reads with the following specifications: R1 = 28, i7 = 10, i5 = 10, R2 = 90. An average depth of around 50,000 reads per cell was achieved during sequencing.

Read alignment, cell-calling, and feature-barcode count matrix generation were performed using the 10x Genomics Cell Ranger v7.1.0 pipeline. Downstream analysis was performed on the feature-barcode count matrices using the R package Seurat v4.3.0. Cells were further filtered according to UMI count, feature count, and mitochondrial percentage using a median-absolute-deviation approach implemented in the scater package. Cells with >0.7 riboprotein content and those marked as doublets by the scDblFinder package were removed. Filtered data were log-normalized and scaled using sctransform implemented in Seurat, using the top 3,000 highly variable genes. Dimensional reduction was performed using PCA. The Louvain algorithm was applied with a resolution of 0.6 to cluster the cells using the first 15 PCs. The clustered cells were visualized in a two-dimensional space via UMAP of the same PCs.

Subclustered macrophages, monocytes, and precursors were normalized, scaled, and dimension-reduced by PCA. UMAP dimensionality reduction and clustering were performed with the first 30 PCs at a resolution of 0.5. Cluster-specific genes were identified with the FindAllMarkers function and clusters annotated manually using the ImmGen database and SingleR package. FL scRNA-seq data were downloaded from Wang et al. (2020).

Between 85,000 and 100,000 macrophages (CD45⁺F4/80⁺CD11b^lo) from Xpr1^fl/fl and Mrc1^CreXpr1^fl/fl E15.5 embryos were sorted (as described above) and total RNA extracted using RNeasy Micro RNA isolation kits (QIAGEN). Sample RNA quality was analyzed on an Agilent TapeStation, and samples with an RNA integrity number value over 7.2 were used for sequencing (median sample RNA integrity number was 8.3). cDNA libraries were prepared using 45 ng of total RNA. Samples were subjected to a paired-end sequencing run of 150 reads on an Illumina NovaSeq X Plus system.

Sample reads were quality control tested using FastQC. Read count matrices were generated with the salmon package using mouse reference genome GRCm39 to build the index. Transcript reads were annotated using AnnotationHub, and data were imported using the tximport package. RNA-seq analysis was performed with DESeq2. DEGs between control and Cre⁺ samples were identified using a false discovery rate of 0.05. Gene ontology pathway analysis was performed using the clusterProfiler package.

Lysed RNA from sorted cells was purified using the Quick-RNA Microprep Kit (Zymo Research) and reverse-transcribed using the Tetro cDNA Synthesis Kit (Meridian Bioscience). Quantitative real-time PCR was performed using SYBR Green mixes (Thermo Fisher Scientific) on a CFX384 thermal cycler (Bio-Rad). The Δ-Δ CT method was used to quantify gene expression. Primers used were 18S-F: 5′-GTA​ACC​CGT​TGA​ACC​CCA​TT-3′, 18S-R: 5′-CCA​TCC​AAT​CGG​TAG​TAG​CG-3′, Xpr1-F: 5′-CAG​GAC​CAG​GCA​CCT​TCA​GTT​G-3′, Xpr1-R: 5′-AAG​TGT​AGC​AAA​CCT​GCG​CTG​A-3′, Dnase2a-F: 5′-AAG​CCC​TGA​GCT​GCT​ATG​G-3′, and Dnase2a-R: 5′-ATA​CGT​CAG​TCC​CTT​TGG​AGT​A-3′.

Statistical analyses were calculated using GraphPad Prism v10 software. The statistical testing methodologies are described in the figure legends. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns = not significant.

Fig. S1 shows the cell frequencies and numbers of various monocyte and macrophage populations in Xpr1 mutant embryos during gestation. Fig. S2 shows the expression of Xpr1 among various liver cell populations, additional characterization of cell populations in Vav1^iCreXpr1^fl/fl embryos and immunofluorescent images of apoptotic cells in Vav1^iCreXpr1^fl/fl FLs. Fig. S3 shows extended data of the scRNA-seq analysis of Xpr1^fl/fl and Vav1^iCreXpr1^fl/fl FL cells, as well as additional flow cytometry data of eosinophils and alternate Mo/Macs in Vav1^iCreXpr1^fl/fl FLs. Fig. S4 depicts the characterization of FL macrophages and blood parameters in embryos depleted of macrophages using anti-CSF1R. Fig. S5 shows extended data of the RNA-seq analysis of Xpr1^fl/fl and Mrc1^CreXpr1^fl/fl liver macrophages, immunofluorescent imaging data showing the co-expression of CD206, CD31, and IBA1 in FLs and additional flow cytometry data characterizing macrophage populations in Mrc1^CreXpr1^fl/fl and Siglec1^CreXpr1^fl/fl animals.

The scRNA-seq data of E15.5 Xpr1^fl/fl and Vav1^iCreXpr1^fl/fl FLs are publicly available under GEO accession number GSE269382. The bulk RNA-seq data of E15.5 Xpr1^fl/fl and Mrc1^CreXpr1^fl/fl macrophages are publicly available under the GEO accession number GSE269322. Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Melanie Greter ([email protected]).

We thank the Flow Cytometry Facility (University of Zurich), the Center for Microscopy and Image Analysis (University of Zurich), the Laboratory Animal Services Center (University of Zurich), and the Functional Genomics Center (University of Zurich) for technical support. We would also like to thank Mirjam Pinzger and Nicole Puertas for technical support. We also wish to thank Dr. Daniel Ackerman from Insight Editing London for assistance reviewing the manuscript.

This work was supported by grants from the Swiss National Science Foundation (310030_184915 to Melanie Greter), the Swiss Cancer League (KFS-5533-02-2022 to Melanie Greter), and the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No. 819229 to Melanie Greter).

Author contributions: Sebastian A. Stifter: conceptualization, data curation, formal analysis, investigation, methodology, project administration, software, validation, visualization, and writing—original draft, review, and editing. Mitchell Bijnen: formal analysis, investigation, methodology, visualization, and writing—review and editing. Selma Tuzlak and Ekaterina Petrova: investigation, methodology, and writing—review and editing. Philipp Häne: investigation. Elsa Roussel: investigation. Hannah Van Hove: investigation. Kenichi Asano: resources and writing—review and editing. Burkhard Becher: conceptualization, funding acquisition, resources, supervision, and writing—review and editing. Annika Keller: conceptualization, resources, validation, and writing—review and editing. Melanie Greter: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, visualization, and writing—original draft, review, and editing.