Yolk-sac–derived macrophages regulate fetal testis vascularization and morphogenesis

Significance

A main requisite for organogenesis is the integration of vascular networks, which not only deliver oxygen and nutrients but also serve instructive roles in organ patterning that determine how organs develop into their final structure and function in the adult. Our previous work showed that vascularization is essential to induce formation of the primitive cord structures that give rise to the seminiferous tubules of the adult testis. Here we show that fetal macrophages are required to remodel the vasculature and refine organ compartments during differentiation of the gonad. This study reveals an underappreciated and likely vital role for macrophages in fetal organogenesis that may be relevant to the development of many organs.

Abstract

Organogenesis of the testis is initiated when expression of Sry in pre-Sertoli cells directs the gonad toward a male-specific fate. The cells in the early bipotential gonad undergo de novo organization to form testis cords that enclose germ cells inside tubules lined by epithelial Sertoli cells. Although Sertoli cells are a driving force in the de novo formation of testis cords, recent studies in mouse showed that reorganization of the vasculature and of interstitial cells also play critical roles in testis cord morphogenesis. However, the mechanism driving reorganization of the vasculature during fetal organogenesis remained unclear. Here we demonstrate that fetal macrophages are associated with nascent gonadal and mesonephric vasculature during the initial phases of testis morphogenesis. Macrophages mediate vascular reorganization and prune errant germ cells and somatic cells after testis architecture is established. We show that gonadal macrophages are derived from primitive yolk-sac hematopoietic progenitors and exhibit hallmarks of M2 activation status, suggestive of angiogenic and tissue remodeling functions. Depletion of macrophages resulted in impaired vascular reorganization and abnormal cord formation. These findings reveal a previously unappreciated role for macrophages in testis morphogenesis and suggest that macrophages are an intermediary between neovascularization and organ architecture during fetal organogenesis.

Mammalian testis morphogenesis is a highly orchestrated process involving pre-Sertoli cells, germ cells, interstitial/mesenchymal cells, and vascular endothelial cells (1), providing an ideal model system to study cellular interactions during fetal organ patterning. In the mouse, cells in the XY (male) gonad undergo extensive cellular rearrangements between embryonic day (E) 11.5 and E12.5 that lead to the formation of testis cords, the precursors of seminiferous tubules in the adult organ (2). Pre-Sertoli cells express sex-determining genes, such as sex determining region of chromosome Y (Sry) and sex determining region Y (SRY)-box 9 (Sox9) (3, 4), and traditionally have been considered the main driving force in generating testicular architecture. However, recent evidence from our laboratory and others suggests that endothelial and other interstitial mesenchymal cells also play critical roles in testis morphogenesis (5–8).

Mononuclear phagocytes (MPs) represent a diverse subset of the myeloid immune cell lineage which includes macrophages and dendritic cells. MPs play diverse roles in developmental and disease contexts (9, 10), likely acting through their control of vessel anastomosis, clearing of apoptotic cells and other cellular debris, secretion of cytokines and growth factors, and modulation of extracellular matrix (9, 11). MPs are nearly ubiquitous in adult organs throughout the body and are involved in the morphogenesis of multiple tissues, such as bone, mammary gland ducts, pancreatic islets, and eye vasculature (12–15). However, almost all these processes are postnatal tissue-remodeling events.

The origin and function of MPs during fetal organogenesis is poorly understood. Traditional models assumed that hematopoietic stem cells (HSCs) differentiate into monocytes, which circulate through peripheral blood and subsequently are recruited to target tissues via local secretion of cytokines, leading to their differentiation into monocyte-derived lineages, e.g., granulocytes, dendritic cells, or macrophages (reviewed in ref. 10). Definitive hematopoiesis (and the appearance of HSCs) occurs initially in the yolk sac, placenta, and aorta-gonad-mesonephros region (AGM) at different time points between E8.5 and E10.5. HSCs populate the fetal liver shortly thereafter (E11.5–E13.5), before the establishment of the bone marrow. However, “primitive” hematopoiesis, originating from unique yolk-sac–derived progenitor cells, takes place at around E7.5, before definitive hematopoiesis, and gives rise to hematopoietic cell types, including erythrocytes and macrophages, that migrate through the yolk-sac vasculature to colonize the fetus (16). Yolk-sac–derived primitive macrophages can contribute to adult hematopoiesis and adult cell types (17–19). In particular, lineage-tracing analyses showed that adult microglia (brain-specific macrophages) are derived exclusively from primitive yolk-sac precursors (17). Yolk-sac–derived macrophages are distinct from HSC-derived macrophages: Yolk-sac–derived macrophages are F4/80-bright, CD11b-dim, and myeloblastosis oncogene (Myb)- and FMS-like tyrosine kinase 3 (Flt3)-independent, whereas HSC-derived macrophages are F4/80-dim, CD11b-bright, and Myb- and Flt3-dependent (19). It is likely that the distinct origins of these cells impart unique identities and functions.

In the postnatal and adult testis, macrophages make up a large portion of the interstitial compartment (20) and are important for Leydig cell development and steroidogenic function (21, 22). However, macrophages have not been detected or functionally characterized during fetal testis morphogenesis. We show that yolk-sac–derived macrophages are the only major myeloid cell type in the fetal gonad during morphogenesis of the testis. These macrophages interacted with multiple cell types but were prominent near developing vasculature. Specific depletion of macrophages resulted in significant vascular and architectural abnormalities in the fetal testis. These studies uncover a previously unidentified role for gonadal–mesonephric macrophages in testis morphogenesis, consistent with a broader role for macrophages in fetal development and organogenesis.

Results

Macrophages Constitute the Major Myeloid Cell Type Present in the Gonad–Mesonephros Primordium During Initial Testis Differentiation.

To label macrophages during the initial stages of gonad formation, we used a well-characterized macrophage cell-surface marker, F4/80 (20, 23). At E9.5, before gonad specification, we did not detect macrophages in the presumptive gonadal region or along the migratory path germ cells travel to the urogenital ridge (UGR) (Fig. S1A). However, we detected numerous F4/80⁺ cells in the brain, consistent with previous reports of early yolk-sac–derived macrophages entering the head before that time (17, 19) (Fig. S1B). By E10.5, we detected F4/80⁺ cells in the brain (Fig. 1A) as well as in the UGR, which includes the gonad–mesonephros primordium (Fig. 1B). F4/80⁺ cells had a stellate morphology characteristic of macrophages, strongly expressed chemokine (C-X3-C motif) receptor 1 (Cx3cr1)-GFP at all stages examined (24) (Fig. 1 C and D and Fig. S1C), and were concentrated in the mesonephros near the gonad border at E10.5 (Fig. 1 B–D).

Fig. 1.

Macrophages are present in the fetal gonad primordium during initial gonad formation. Immunofluorescent images of E10.5 brain (A), E10.5 UGRs (B–D), and E11.5-E12.5 XY gonads (E–H). In all figures, colors of immunofluorescent markers are indicated. White dashed outlines indicate the gonad–mesonephros border in all figures. A′, B′, and G′ are higher-magnification images of the boxed regions in A, B, and G, respectively. Yellow dashed outlines in B and C denote nascent gonads. At E10.5 (A–D), F4/80⁺ macrophages were present in the brain and the UGR. Macrophages in the E10.5 UGR also expressed Cx3cr1-GFP (C) and were concentrated near the gonad (g)–mesonephros (m) border (D). (E–G) At E11.5, gonads contained IBA1⁺, Cx3cr1-GFP⁺, CSF1R⁺, and F4/80⁺ macrophages near the gonad–mesonephros border. GATA4 labels gonadal somatic cells. In F′, single-channel versions of the boxed region in F show independent macrophage markers. (H) CD45 staining revealed extensive colocalization with IBA1. The arrowhead in H points to a rare CD45⁺, IBA1⁻ cell. (Scale bars: 50 μm.) See also Figs. S1 and S2.

Although Cx3cr1-GFP is expressed in multiple immune cell types in adult tissues (24), flow cytometry experiments revealed that >95% of Cx3cr1-GFP⁺ fetal gonad cells were double-positive for the macrophage markers F4/80 and CD11b (25) (Fig. S1D), demonstrating that Cx3cr1-GFP is an effective and specific marker of gonadal–mesonephric macrophages during fetal stages. Quantitative RT-PCR (qPCR) analysis of purified GFP⁺ gonadal–mesonephric cells obtained via fluorescence-activated cell sorting (FACS) also showed a significant (∼1,000-fold) enrichment of the macrophage markers EGF-like module containing, mucin-like, hormone receptor-like sequence 1 (Emr1) (which encodes F4/80) and colony stimulating factor 1 receptor (Csf1r) (also called “c-fms”) (26) relative to GFP⁻ cells (Fig. S1 E and F).

At E11.5, F4/80- and Cx3cr1-GFP–expressing cells near the gonad border increased in number and also expressed CSF1R and the microglial/macrophage marker ionized calcium binding adaptor molecule 1 (IBA1) (also called “AIF1”) (27, 28) (Fig. 1 E and F), consistent with a macrophage identity. Macrophages were found occasionally within the gonad at this stage but still were concentrated near the gonad–mesonephros boundary at E11.5 and E12.5 (Fig. 1 E–G).

At E12.5, ∼90% of cells positive for the pan-leukocyte marker CD45 coexpressed IBA1 by immunofluorescence (Fig. 1H), suggesting that macrophages are the major MP cell type present. Flow cytometry analysis of gonadal CD45⁺ cells revealed a minor MP population (9.2% of CD45⁺ cells) immunoreactive for Gr-1 (Fig. S2A), a marker for granulocytes such as neutrophils (29). We saw virtually no immunoreactivity via flow cytometry (3.1% of CD45⁺ cells) for the dendritic cell marker CD11c (also called “ITGAX”) (30) (Fig. S2A), in contrast to hematopoietic fetal liver, where higher numbers of both CD45/CD11c⁺ dendritic cells (25.9%) and CD45/Gr-1⁺ granulocytes (16.5%) were present (Fig. S2B). Finally, immunofluorescence analyses showed virtually no lymphoid cells (B cells or T cells) present in the gonad during initial testis differentiation (Fig. S2 C–J).

Gonadal–Mesonephric Macrophages Arise from Primitive Yolk-Sac–Derived Hematopoietic Progenitors.

To determine whether gonadal–mesonephric macrophages are derived from the yolk sac, we delivered a pulse of 4-hydroxytamoxifen at E7.5 to induce activity of Csf1r-CreER [which is active in yolk-sac cells at this stage (31)] and to label primitive yolk-sac progenitors permanently with a Cre-responsive Rosa-Tomato red fluorescent lineage tracer (32). We subsequently assessed Tomato⁺ cells at E14.5, a stage when tissue-resident macrophages are readily distinguishable as being yolk-sac–derived or HSC-derived by their relative levels of F4/80 and CD11b expression (19). As a positive control we examined fetal brains, which are colonized by primitive yolk-sac–derived macrophages at around E8.5–E9.5 (17). As expected, we detected extensive Tomato labeling throughout the brain, which colocalized with F4/80 (Fig. 2A). Flow cytometry confirmed that Tomato efficiently labeled F4/80⁺ and CD11b⁺ macrophages (Fig. 2 G–I). We did not detect any Tomato labeling of Cre⁻ littermate controls in any organs we assayed (Fig. 2 B, D, and F).

Fig. 2.

Gonadal macrophages arise from primitive yolk-sac–derived progenitors. (A–F) Immunofluorescence images of fetal brains (A and B), livers (C and D), and testes (E and F) from E12.5 Csf1r-CreER; Rosa-Tomato (A, C, and E) and littermate control embryos carrying Rosa-Tomato, but no Cre recombinase (B, D, and F) injected at E7.5 with 4-hydroxytamoxifen. A′–F′ are higher-magnification images of the boxed regions in A–F, respectively. Expression of Tomato red fluorescent protein was detected only in Csf1r-CreER; Rosa-Tomato embryos (A, C, and E) but not in control littermates (B, D, and F) exposed to 4-hydroxytamoxifen in utero. Tomato-positive macrophages (A′, C′, and E′, arrowheads) were observed in all three organs. Occasionally, Tomato was observed in F4/80⁻ cells in gonads and other tissues (arrows). (Scale bars: 50 μm.) (G) Flow cytometric analysis of Tomato⁺ cells from a representative litter of E14.5 Csf1r-CreER; Rosa-Tomato embryos injected at E7.5 with 4-hydroxytamoxifen. Tomato⁺ cells from fetal brain (Left) and testis (Center) contained mostly F4/80-high, CD11b-intermediate cells, but livers (Right) showed a shift toward F4/80-intermediate, CD11b-high cells. Numbers next to polygonal gates indicate the percent of Tomato⁺ cells within the adjacent gate. (H) Graph shows the percent of Tomato⁺ cells expressing F4/80 and CD11b in XY brain (n = 8 brains total), testis (n = 16 testes), and liver (n = 8 livers) from three independent litters. Colors of bars in H correspond to the population whose percentages are labeled in the same color in G. (I) Graph shows percent efficiency of Tomato labeling in F4/80⁺ cells by flow cytometry in E14.5 brain, testis, and liver in the litters analyzed in H. Data in H and I are represented as means ± SEM. Lowercase letters (a versus b or y versus z) indicate statistically different values (P < 0.05). See also Fig. S3.

In the fetal liver, Tomato labeled a smaller percentage of F4/80⁺ cells and more F4/80⁻ cells relative to the brain and testis (Fig. 2 C and I). Flow cytometry confirmed this result and revealed a marked increase in the number of CD11b-bright cells (as opposed to CD11b-intermediate cells) in the liver (Fig. 2 G and H). This result is consistent with an enrichment of HSC-derived macrophages in the liver (19), suggesting a more complex origin for fetal liver macrophages. kit oncogene (C-KIT), a marker of HSCs, stained some Tomato⁺ cells in the fetal liver, but many were C-KIT⁻, perhaps suggesting they had differentiated and no longer expressed C-KIT (Fig. S3 A and B).

In the fetal testis, Tomato labeling was extensive and overlapped with F4/80⁺ cells (Fig. 2E). Flow cytometry revealed that the expression profiles of testis and brain macrophages were virtually identical, with an enrichment of F4/80-bright, CD11b-intermediate cells (Fig. 2 G and H), consistent with yolk-sac–derived macrophages rather than HSC-derived macrophages (19). This result suggests that, like brain microglia (17), gonadal macrophages are derived almost exclusively from primitive yolk-sac progenitors, with a negligible contribution of HSC-derived cells that arise later in development from the yolk sac, AGM, or placenta.

Consistent with this hypothesis, purified gonadal macrophages expressed lower levels of the HSC-associated genes Myb and Flt3 (19) relative to purified fetal liver macrophages (Fig. S3C). These data agree with our previous gonad microarray analysis (33), in which the yolk-sac macrophage–associated gene spleen focus forming virus proviral integration oncogene (Sfpi1; also called “Pu.1”) was detected in hematopoietic-derived cells but the HSC-associated genes Myb and Flt3 were not [Fig. S3D (33)].

Fetal Gonadal Macrophages Are M2-Like Macrophages in Active Cell Cycle.

E12.5 Cx3cr1-GFP gonadal macrophages were nearly 100% positive for a marker of active cell cycle, antigen identified by monoclonal antibody Ki 67 (MKI67) (Fig. 3A), and some macrophages expressed the M-phase marker phospho-histone H3 (pHH3) (Fig. S4 A and B). These data are consistent with the idea that, once populated from yolk-sac progenitors, fetal macrophages proliferate locally to increase their numbers rather than recruiting from other hematopoietic sources such as the AGM. Adult testes, in contrast to fetal testes, contained macrophages that were MKI67⁻ (Fig. 3B), indicating that certain cell-cycle properties are unique to fetal or adult testicular macrophages.

Fig. 3.

Fetal gonadal macrophages are in active cell cycle and are in M2 activation status. Immunofluorescence images of E12.5 fetal (A and D–G), and adult (B) testes. A′, B′, and D′–G′ are higher-magnification images of the boxed regions in A, B, and D–G, respectively. Fetal gonadal macrophages (A) are positive for MKI67 (arrowhead in A′; Inset is MKI67 channel alone for the GFP⁺ cell indicated by the arrowhead), unlike adult testis macrophages (B and arrowhead in B′). CD68 marks macrophages in B. (C) Graph shows the number of F4/80⁺ macrophages per single 450 × 336 μm confocal optical section in the testis–mesonephros region of E10.5–E13.5 wild-type CD-1 embryos (each stage contains ≥12 gonads). Data in C are represented as means ± SEM. (D–G) Fetal testis macrophages show low expression levels of the M1-associated markers CD86 (D) and MHCII (E) but are strongly positive for the M2-associated markers CD206 (F) and C-MAF (G). C-MAF also is expressed in interstitial cells of the fetal gonad (8). Insets in D′ and E′ are CD86- and MHCII-only channels, respectively, for macrophages indicated by arrowheads. (Scale bars: 50 μm.) See also Figs. S4 and S5.

Although we found that fetal gonad macrophages were in active cell cycle, we did not find any appreciable apoptosis of macrophages during fetal testis development. However, we did observe that activated (cleaved) Caspase-3⁺ cells were regularly engulfed by testis macrophages (Fig. S4 C and D; see also below). Consistent with our MKI67 and pHH3 data suggesting that macrophages proliferate in situ and show negligible cell death, we found a steady increase in the number of macrophages observed in the fetal testis between E10.5 and E13.5 (Fig. 3C).

Macrophages are characterized by M1/M2 activation status, defined by specific cell-surface markers and by the secretion of a characteristic subset of proinflammatory or anti-inflammatory cytokines (34). M1 macrophages traditionally are associated with proinflammatory, cytotoxic responses, whereas M2 macrophages are anti-inflammatory and are linked to tissue remodeling and angiogenesis. The expression of two markers of M1 macrophages, CD86 and major histocompatibility complex, class II (MHCII), was very low based on fluorescent antibody staining of fetal testes (Fig. 3 D and E). In contrast, two M2 macrophage markers, CD206 (also called “MRC1”) and avian musculoaponeurotic fibrosarcoma AS42 oncogene homolog (C-MAF) (34), were strongly expressed in virtually 100% of IBA1⁺ and F4/80⁺ cells (Fig. 3 F and G). C-MAF is a transcription factor that mediates the immunosuppressive activity of the M2-associated anti-inflammatory cytokine interleukin 10 (IL10) (35), and its expression in macrophages is consistent with an M2 status. c-Maf also was reported recently to be enriched in yolk-sac–derived F4/80-bright macrophages (19), consistent with our lineage tracing results. These markers were expressed in distinct patterns in the fetal brain and liver (Figs. S2 and S5 A–D): Similar to gonadal macrophages, brain macrophages were M2-like, but fetal liver macrophages were more M1-like, with stronger expression of CD86 and weak expression of CD206 (Fig. S5 A–D). In addition, qPCR analysis of FACS-purified Cx3cr1-GFP cells from E12.5 XY brains, gonads, and livers revealed that, similar to brain macrophages and in contrast to liver macrophages, gonadal macrophages expressed lower levels of the M1-associated gene interleukin 12 (Il12) and higher levels of the M2-associated gene arginase 1 (Arg1) (Fig. S5 E and F). All these results suggest that gonadal macrophages are M2-like, a state that is associated with vascular and tissue remodeling (11, 36).

Fetal Gonadal Macrophages Are Localized near Developing Vasculature in the Nascent Testis.

By E10.5, macrophages were localized specifically to the developing vascular plexus along the gonad–mesonephros border and to vascular branches between the mesonephric ducts (Fig. 4 A and B). Between E11.5 and E13.5, the peak of vascular remodeling and neovascularization of the testis, large numbers of macrophages accumulated in the gonad–mesonephric region (Fig. 4 C–G). They were located near the vascular plexus and along the coelomic surface artery and its associated side branches, often wrapping long processes around endothelial cells (Fig. 4 F and G).

Fig. 4.

Fetal macrophages are intimately associated with nascent vasculature in the developing testis. Immunofluorescence images of E10.5 (A and B), E11.5 (C and D), E12.5 (E and F), E13.5 (G) fetal testes, and E11.5 (H–K) cultured fetal testes. (A and B) At E10.5 a vascular plexus (vp, yellow bracket in A) already is visible between the GATA4⁺ gonad (containing SOX2⁺ germ cells) and the neighboring mesonephros, and fetal macrophages concentrated near nascent vasculature are starting to invade the gonad (arrows in B). The asterisk in A shows mesonephric ducts marked by diffuse SOX2 staining. (C and D) By E11.5, macrophages are enriched in the vascular plexus (vp) near the gonad–mesonephros border and near vascular sprouts in the gonad (arrows in D). Kdr-mCherry labels endothelial cells (42). (E–G) By E12.5 (E and F) and E13.5 (G), an increase in the number of macrophages is evident, concentrated in the vascular bed at the gonad border and near the newly formed testis vasculature (arrowheads in E and G). Inset in G is a higher-magnification image of the coelomic vessel region denoted by the arrowhead. (H and I) Relative to control E11.5 cultured testes (H), vascular-depleted testes cultured in VEGFR TKI II (I) do not show an increase in apoptosis of nonendothelial cells. The arrow in H indicates the normal formation of a coelomic vessel. Arrowheads in I indicate dying vascular clumps in the absence of VEGF signaling. (J and K) Vascular-depleted testes show altered localization of macrophages relative to controls. (Scale bars: 50 μm.) (L) The graph shows fold change in gene expression in E11.5 and E12.5 vascular-depleted testes cultured in VEGFR TKI II (removed from mesonephros) relative to cultured control testes for markers of germ cells (Oct4), Sertoli cells (Sox9), endothelial cells (Cdh5), and macrophages [Emr1 (F4/80) and Csf1r]. Expression of Vegfa was unaffected. +, 0.05<P < 0.1; *P < 0.05; **P < 0.005. Also see Fig. S6.

Testis macrophages expressed a number of endothelial markers including Neuropilin 1 (NRP1), which is a vascular endothelial growth factor (VEGF) coreceptor that normally is expressed in endothelial cells but also has been implicated as a marker of macrophages involved in neovascularization of tumors (36–38) (Fig. S6A); endothelial-specific receptor tyrosine kinase (TEK; also called “TIE2”) (Fig. S6B), an Angiopoietin receptor expressed in macrophages associated with tumor neovascularization (36–38); and Tie2-Cre (Fig. S6C), as we previously reported using a Cre-responsive GFP reporter strain (39). Tie2-Cre is not expressed in hematopoietic cells of the yolk sac (40), suggesting that the expression of Tie2-Cre is not merely the result of the labeling of early hematopoietic progenitors but reflects de novo Tie2 expression within differentiated macrophages.

To investigate whether macrophages are principal recruiters of endothelial cells, we used Cx3cr1-GFP to FACS-purify macrophages and determined their expression of three proangiogenic factors, vascular endothelial growth factor A (Vegfa), angiopoietin 1 (Angpt1), and angiopoietin 2 (Angpt2) relative to the GFP⁻ population of the gonad and mesonephros. GFP⁺ macrophages expressed detectable, albeit low, levels of Vegfa and Angpt1 but not of Angpt2 (Fig. S6D). However, macrophage expression of Vegfa (and Angpt1) was lower than that of the nonmacrophage cell populations of the gonad/mesonephros (Fig. S6 D and E), suggesting, in agreement with our previous report (7), that macrophage-derived VEGF is not the major source of VEGF in the gonad. By immunofluorescence, we detected bright VEGFA punctae within gonadal macrophages (Fig. S6F). These punctae could represent autonomously produced VEGFA; alternatively, macrophages might sequester or relay VEGFA from the environment.

Gonad Macrophage Number and Localization Require Vasculature.

Given our observation that testis macrophages are found consistently in close proximity to blood vessels, we hypothesized that vasculature is required for the maintenance, recruitment, or migration of fetal testis macrophages. To test this hypothesis, we depleted testis vasculature in our whole-organ culture system (41) using a VEGF receptor small-molecule inhibitor, VEGFR tyrosine kinase inhibitor (TKI) II (5) and observed the effects on macrophage number and gene expression. We achieved robust vascular depletion, visualized via dramatically reduced anti-platelet/endothelial cell adhesion molecule 1 (PECAM1) endothelial cell immunofluorescence (Fig. 4 H and I), and a 90–97% reduction in cadherin 5 (Cdh5) (also called “VE-Cad”) testis expression (relative to control cultured littermate testes) via qPCR (Fig. 4L). These effects were observed in fetal testes cultured at both E11.5 and E12.5. We saw no evidence of apoptosis or reduction in germ cells based on PECAM1 and cleaved Caspase-3 staining (Fig. 4 H and I). Consistent with this result, the expression of germ cell [POU domain, class 5, transcription factor 1 (Pou5f1); also called “Oct4”] and Sertoli cell (Sox9) markers was unchanged based on qPCR (Fig. 4L). These data indicate that overall tissue health was maintained in vascular-depleted organ explants.

When we assessed macrophage number and gene expression in vascular-depleted testes, we saw that, although macrophage number in the overall UGR (gonad plus mesonephros) was unaffected, there were fewer macrophages in vascular-depleted gonads, particularly in regions where the newly formed coelomic vasculature normally would reside (Fig. 4K). By qPCR, we saw a 55–65% reduction in both Emr1 (F4/80) and Csf1r expression (Fig. 4L) in the testis, although this effect was more significant when gonads were cultured at E12.5 than at E11.5. This greater reduction likely occurs because more macrophages are normally present in the gonad by E12.5 (Fig. 4 C and E), so a reduction in gonad-specific macrophage number or gene expression upon vascular depletion is more robust at later stages. These data suggest that macrophage localization and/or migration into the gonad are dependent on VEGF-mediated vascularization of the fetal testis.

Gonadal Macrophages Promote Tissue Clearance and Remodeling to Regulate Morphogenesis.

In gonad development, germ cells sometimes are lost in the mesonephros en route to the gonad. Although it was assumed that these cells die by apoptosis, our analyses revealed that, at E11.5 and E12.5, germ cells remaining in the mesonephros tended to cluster and nearly always were associated with or engulfed by macrophages: We observed SOX2⁺ nuclei completely contained within F4/80⁺ cells (Fig. 5A). These germ cells were apoptotic, because we often observed C-KIT⁺/Caspase-3⁺ cells engulfed by macrophages (Fig. 5B). These results suggest that macrophages are involved in the clearance of mislocated germ cells during initial gonad formation.

Fig. 5.

Fetal gonadal macrophages play roles in tissue clearance and remodeling to regulate tissue morphogenesis. Immunofluorescence images of E11.5 (A and D) and E12.5 (B, C, and E–I) fetal testes. A′–F′ are higher-magnification images of the boxed regions in A–F, respectively. (A) At E11.5, SOX2⁺ germ cells in the mesonephros are engulfed by macrophages. A number of these engulfed germ cells (C-KIT⁺) are apoptotic (cleaved Caspase-3⁺ cells; arrowheads in B′). Panels to right of B′ are Caspase-3-only and C-KIT-only channels for the image in B′. (C and C′) AMH⁺ Sertoli cells remaining in the interstitium (i) outside the testis cords (tc) are surrounded by macrophages. Dashed outlines in C′ denote testis cord boundaries. (D and E) Macrophages (arrowheads in D′) surround vascular sprouts in the region of the gonad–mesonephros border (dashed line in D) as well as in gonad surface vasculature (arrowheads in E). The Inset to right of E′ is a higher-magnification image of the cell denoted by the open arrowhead in E indicating internalized vascular material. (F and G) A number of apoptotic PECAM1⁺ endothelial cells (F and F′) are engulfed by Cx3cr1-GFP⁺ macrophages (G). (H) Engulfed vascular material is colocalized with macrophage phagolysosomal marker CD68 within a NRP1⁺ macrophage (outlined). (I) By E12.5 macrophages also are within testis cords (closed arrowheads). The Insets to the right of I are higher-magnification images of the macrophage–germ cell interaction denoted by open arrowhead in I. (Thin scale bars: 50 μm; thick scale bars: 12.5 μm.) Also see Fig. S7.

In normal gonad morphogenesis, Sertoli cells, like germ cells, are rarely found outside testis cords. At E12.5, when testis cords have formed completely, we observed macrophages within the interstitial compartment completely surrounding anti-Müllerian hormone (AMH)-positive cellular material (Fig. 5C), suggesting that macrophages also engulf or phagocytize AMH-expressing Sertoli cells that fail to be incorporated into cords during initial testis morphogenesis.

During the stages when extensive vascular remodeling is taking place in the XY gonad (E11.5–E12.5), we often observed kinase insert domain protein receptor (Kdr)-mCherry⁺ (also known as “Flk” or “Vegfr2”) endothelial cells (42) completely engulfed by macrophages, both in the mesonephric vascular plexus and in the surface testis vasculature (Fig. 5 D and E). We commonly observed apoptotic endothelial cells, especially within the mesonephric vascular plexus (Fig. 5F), and these dying endothelial cells (PECAM1⁺/Caspase-3⁺ cells) also were engulfed by macrophages (Fig. 5G). Kdr-mCherry⁺ material contained within macrophages colocalized with CD68 (also called “macrosialin”), a marker of macrophage-specific lysosomes and endosomes of the phagocytic pathway (43, 44) (Fig. 5H), suggesting that macrophages may physically break down the vascular plexus to free individual endothelial cells, which then can migrate from the mesonephros into the gonad (39).

Although the localization of macrophages is interstitium-specific at postnatal stages (20), we found a large number of macrophages within testis cords in the fetal gonad (Fig. 5I). These macrophages had a much more stellate morphology than their interstitial counterparts and were intercalated between Sertoli cells and germ cells, making contact with multiple cells in the cord. Localization within cords was observed only during fetal stages. By birth, macrophages were restricted to the interstitium and were excluded from testis cords (Fig. S7), similar to their localization in adult testes (20).

Macrophages Are Required for Vascular Breakdown and Testis Cord Morphogenesis.

To determine whether macrophages play a critical role in vascularization and testis morphogenesis, we used the Rosa-eGFP-Diphtheria Toxin A (Rosa-eGFP-DTA) system (45), under the control of Cx3cr1-Cre to deplete macrophages specifically. To confirm that Cx3cr1-Cre is active in early macrophages, we used a Cre-responsive farnesylated-GFP (Rosa-fGFP) reporter (46) as a lineage tracer and fluorescent reporter to detect Cre-expressing cells. At E10.5, GFP⁺ cells were detected in double-positive embryos within the brain, yolk sac, and UGR, and ∼95% of these cells expressed F4/80 (Fig. 6 A–C), whereas Cre⁻ littermates showed no GFP expression (Fig. 6D). Similar results were obtained using an independent Cre-responsive Rosa-Tomato lineage tracer (32) (Fig. 6E). In these experiments, Tomato was not detected in Sertoli cells, germ cells, or vasculature (Fig. 6F), demonstrating that Cx3cr1-Cre is an effective and specific Cre line for targeting fetal macrophages.

Fig. 6.

Cx3cr1-Cre is specific to fetal macrophages and can be used effectively for targeted macrophage ablation. (A–H) Immunofluorescence images of E10.5 brain (A), E10.5 yolk sac (B), E10.5 UGR (C and D), E12.5 XY gonads (E and F), and E13.5 XY gonads (G and H). C′–F′ are higher-magnification images of the boxed regions in C–F, respectively. Cx3cr1-Cre drives the expression of Rosa-fGFP in brain microglia (A), yolk-sac macrophages (B), and UGR macrophages (C) but not in control Cre⁻ littermates (D). Arrowheads in A and B indicate rare GFP⁻ F4/80⁺ cells. (E) Independent verification of Cx3cr1-Cre by a Rosa-Tomato reporter also shows effective Cre activation specifically in E12.5 gonadal macrophages. (F) Tomato is not detected in SOX9⁺ Sertoli cells or in PECAM1⁺ germ cells and vasculature. (G and H) Maximum-intensity projection images of 10 optical sections equally spaced through the entire Z-plane of the testis show that Rosa-DTA driven by Cx3cr1-Cre results in near-complete ablation of gonadal macrophages. (I) Quantification of total F4/80⁺ cells in a 375 × 375 × 50 μm region of E13.5 control (pooled wild-type, +; Rosa-DTA, and Cx3cr1-Cre; + genotypes, n = 5) and macrophage-depleted Cx3cr1-Cre;Rosa-DTA fetal testes (n = 4). Data are represented as means ± SEM. (Scale bars: 50 μm.)

E13.5 Cx3cr1-Cre; Rosa-eGFP-DTA (macrophage-depleted) embryos showed no significant gross anatomical abnormalities and were recovered in expected numbers from litters at E13.5. The Cx3cr1-Cre; Rosa-eGFP-DTA system consistently exhibited nearly complete ablation (∼95%) of macrophages (Fig. 6 G–I). Although a robust, coherent coelomic artery was present in controls, fewer endothelial cells migrated into the coelomic region of the macrophage-depleted E13.5 Cx3cr1-Cre; Rosa-eGFP-DTA gonads, and the coelomic vessel was poorly organized with fewer endothelial cells (Fig. 7 A–D). Reciprocally, the mesonephric vascular plexus was enlarged in depleted animals and showed a dramatic increase in the overall vasculature remaining in the mesonephros (Fig. 7 E–H).

Fig. 7.

Macrophages are required for testis-specific vascularization and morphogenesis. Immunofluorescence images of E13.5 control +;Rosa-DTA (A, C, E, G, and I), E13.5 macrophage-depleted Cx3cr1-Cre;Rosa-DTA (B, D, F, H, and J), E12.5 control (K, M, and O), and E12.5 macrophage-depleted (L, N, and P) testes. C, D, G, and H are maximum-intensity projection images of 10 optical sections spaced equally through the entire Z-plane of the testes shown in A, B, E, and F, respectively; A, B, E, F, and I–P are single confocal optical sections. Although control testes had a robust, coherent surface coelomic artery (A and C), macrophage-depleted testes exhibited an unorganized surface vasculature (B and D). Macrophage-depleted testes also showed more vasculature remaining in the mesonephros near the gonad border (E–H) and abnormal testis cords (marked by AMH) (I and J) relative to control littermate testes. Inset in J is a higher-magnification image of the boxed region showing an aberrant Sertoli cell cluster outside the testis cord boundaries (dashed lines in Inset). E12.5 control (K, M, and O) and E12.5 macrophage-depleted (L, N, and P) testes show similar levels of MKI67 (K and L), pHH3 (M and N), and Caspase-3 (O and P) expression, in particular within PECAM1⁺ endothelial cells (arrowheads in K–N). (Thin scale bars: 50 μm; thick scale bars: 12.5 μm.)

Wild-type E13.5 testes had well-established cords that were of regular size and spacing (Fig. 7I). Although gross partitioning of interstitial cells from Sertoli and germ cells took place in macrophage-ablated samples, cords were irregularly branched and fused and often failed to reach the coelomic surface region of the gonad (Fig. 7J). There was some variability in the severity of cord architecture defects, although abnormalities were consistent. Interestingly, small clusters of three to five Sertoli cells were detected outside the boundaries of testis cords (Fig. 7J, Inset), consistent with an absence of the surveillance function of testis macrophages.

Although the distribution of vasculature was abnormal in macrophage-depleted testes, the overall number of endothelial cells was grossly unaffected. To assess the direct effects of macrophage depletion on testis cell types, and on endothelial cells in particular, we examined cell-cycle activity (by MKI67 and pHH3 expression) and apoptosis (by cleaved Caspase-3 expression) in macrophage-depleted testes. In Cx3cr1-Cre; Rosa-DTA fetal testes we saw normal expression of MKI67 and pHH3 in PECAM1⁺ endothelial cells (Fig. 7 K–N). We also noticed no dramatic increase or decrease of Caspase-3 expression within the vasculature or in the testis overall (Fig. 7 O and P). We assume that the few macrophages remaining after depletion are sufficient to maintain a level of cell clearance required by the low level of cell death that occurs during testis development (see Fig. 4H). Overall, these data suggest that fetal testis macrophages are required for vascular remodeling but not for cell-cycle activity or survival of endothelial cells.

Discussion

The Presence of Macrophages in the Fetal Testis.

Although macrophages are known to populate the interstitium of the postnatal and adult rat testis (20, 47), little is known about their fetal origins. Although some reports have documented the presence of macrophages in the fetal rat testis as early as E16.5–E19 (47, 48), there has been no investigation of the role of macrophages during the initial steps of fetal testis organogenesis. Here we demonstrated that differentiated macrophages expressing well-characterized markers such as F4/80 and IBA1 are detectable in the UGR at the initial stages of gonad specification (E10.5). Because our data suggested that the great majority of macrophages during initial gonad differentiation (E10.5–E12.5) are localized to the mesonephric region near the gonad border, it is possible that macrophages were not detected in previous studies because the mesonephros was not examined. However, a substantial increase in numbers of macrophages also was observed within the gonad by E12.5–E13.5, likely because of the proliferation of resident cells. Testicular macrophages have been reported in many diverse nonmammalian species, including swan, catfish, and lizard (49–51). Thus, it is possible that macrophages are present in the nascent gonad in all vertebrates and may have evolutionarily conserved roles.

Yolk-Sac–Derived Primitive Macrophages Play Diverse Roles in Fetal and Adult Organogenesis.

Traditional models of hematopoiesis are centered around the dogma that HSCs give rise to precursors that patrol the peripheral blood and enter tissues that secrete macrophage recruitment or differentiation factors (16). Although this model of definitive HSC-based hematopoiesis has dominated the fields of immunology and hematology for decades, recent lineage-tracing experiments have revealed that primitive hematopoiesis in the yolk sac plays a significant role in both fetal and adult myelopoiesis (17–19).

In this study, we demonstrated that primitive yolk-sac hematopoietic progenitors give rise to gonadal–mesonephric macrophages. This result was surprising, given the proximity of the UGR to the AGM region, where a subset of definitive HSC progenitors arises. Other groups have shown that a subset of other organ-specific macrophages (19) and microglia in the brain (17) also are derived exclusively from primitive yolk-sac progenitors. Although our data are consistent with a yolk-sac primitive hematopoietic origin for gonad macrophages, we cannot formally rule out the possibility that a subset of gonad macrophages also is derived from the yolk sac at later stages, or from other sites of definitive hematopoiesis such as the AGM or placenta. However, we observed that the efficiency of tamoxifen labeling with Csf1r-CreER;Rosa-Tomato is virtually identical in gonad macrophages and brain macrophages (microglia), the latter of which are derived exclusively from yolk-sac–derived primitive hematopoietic progenitors (18). If a subset of gonad macrophages comprised definitive macrophages arising later in development (i.e., at E8.5 or later), then we would see decreased labeling efficiency in the gonad relative to the brain, similar to observations in the liver. However, that was not the case, suggesting that gonad macrophages, like microglia, are solely primitive in origin. The fate of fetal gonadal macrophages is unclear; however, a significant subset of macrophages in multiple adult organs is yolk-sac–derived (20). Our data show that testis macrophages are present throughout fetal development, at least until birth, and may persist until adulthood in the testis, where they could play roles such as promoting the development of adult Leydig cells (23).

Although the brain microenvironment imparts some unique properties to microglia, their phenotype likely is attributable in part to their yolk-sac origins. Similar to microglia, which are involved in anastomosis of blood vessels in the brain (36), gonadal macrophages are involved in vascular remodeling. It is possible that the primitive hematopoietic cells in the yolk sac are predisposed to take part in blood vessel formation and patterning and may have other unique roles during organ formation, as well.

Roles for Macrophages in Fetal Organogenesis.

Growing evidence suggests that macrophages have critical functions in tissue development apart from their classic phagocytic functions, in particular in promoting angiogenesis and vascular remodeling in postnatal organs (9). A well-characterized example is the refinement of the hyaloid vasculature in the mouse postnatal eye, in which WNT ligands secreted by macrophages are critical for the breakdown, remodeling, and patterning of blood vessels (52).

In contrast to postnatal contexts and disease states, there are few documented examples of the role of macrophages in fetal organogenesis. In the fetal brain, macrophages are involved in anastomosis during vascular remodeling, because they associate with endothelial tip cells to promote the fusion of adjacent vascular branches (36). One previous study reported that late-stage E18.5 colony stimulating factor 1 (Csf1)-deficient fetuses, which showed reduced numbers of macrophages, had decreased insulin cell mass and abnormal pancreatic islet morphogenesis (12). However, whether fetal macrophages affect vascularization of the pancreas or initial pancreatic islet differentiation was not shown.

In this report, we demonstrated previously unidentified, essential roles for macrophages in fetal organogenesis, using an effective macrophage-specific cell ablation model: a Cre-responsive DTA line driven by a macrophage-specific Cx3cr1-Cre. Although previous reports of macrophage-depleted models such as Csf1 mutants found a modest number of macrophages remaining in fetal organs, often rescued by maternal circulating CSF1 (53), our model specifically and effectively targets macrophages in a cell-autonomous manner, avoiding the confounding effects in earlier mutant studies as well as potential secondary toxic side effects of diphtheria toxin administration in the Cre-inducible diphtheria toxin receptor system (54).

Macrophages are known to be involved in the clearance of apoptotic bodies and cells in development (55). Such cytotoxic activities usually are associated with M1 macrophages. However, M2 macrophages can maintain high phagocytic activity in certain contexts (56). Our results revealed that germ cells that fail to migrate into the gonad but remain in the mesonephros at E11.5–E12.5 nearly always are surrounded and engulfed by macrophages. Because germ cells can give rise to teratomas or other tumors in an inappropriate environment, they may be a priority for clearance by macrophages. The unusual morphology and clustering of these germ cells may reflect the initiation of an apoptotic program, part of which induces phagocytic activity of macrophages. In addition, Sertoli cells almost never are observed outside the testis cords after E12.5 because of the activity of macrophages that normally act to refine testis cord structures by pruning mislocated Sertoli cells. Accordingly, in macrophage-ablated testes, clusters of Sertoli cells were readily found outside of testis cords. Although our data point to a primary function of macrophages in vascular remodeling during testis organogenesis, future studies are required to determine definitively if macrophages play critical, vascular-independent roles in tissue remodeling of the testis, i.e., if they can influence the behavior of Sertoli cells and/or interstitial cells directly to promote testis cord morphogenesis.

Although Sertoli cells traditionally have been considered the key cell type driving the formation of testis cord structures, they are not sufficient for testis cord morphogenesis. The vasculature plays an instructive role in testis morphogenesis (5–8, 39). The first step of this process, the breakdown of the mesonephric vascular plexus, is dependent on gonadal–mesonephric macrophages. Although the signals that drive vessel remodeling and testis neovascularization still are not well-defined, Vegf signaling is required, and several lines of investigation implicated Sertoli cells and/or interstitial cells as the source of VEGFA (5, 7). In other contexts, such as in tumor vascularization, macrophages have been implicated as a source of VEGFA production (11). In this study, we showed that macrophages in the fetal gonad also express Vegfa and, like tumor-associated macrophages (36–38), express the angiogenic-associated receptors NRP1 and TEK/TIE2. Macrophages may use these cell-surface receptors to sequester pools of the endothelial attractant factors VEGF or ANGPT and present them to endothelial cells to guide neovascularization of tumors. Consistent with this idea, VEGFA is sequestered inside gonadal macrophages, perhaps influencing the activity, behavior, or migration of endothelial cells during testis formation. However, our data in this study show that gonadal macrophages are not a major source of Vegfa mRNA, suggesting that other interstitial cells produce the majority of VEGFA required for endothelial cell proliferation and survival. Our results showing that overall endothelial cell number is not affected in macrophage-depleted gonads are consistent with this idea. One possibility is that macrophage-derived VEGFA is a local source of VEGFA that influences the migratory path of endothelial cells during fetal testis vascular remodeling.

Our data indicate that macrophages play an important role during de novo morphogenesis of testis cords by controlling vascularization and tissue pruning. This unexpected role for fetal macrophages may be a developmental mechanism that is used commonly during neovascularization and morphogenesis of many fetal organs.

Materials and Methods

Mice.

CD-1 mice (Charles River) were used for wild-type expression studies. Cx3cr1-GFP mice (Cx3cr1^tm1Litt) (24) are publicly available at Jackson Laboratories and are maintained on a mixed background of C57BL/6J (B6) and CD-1. Kdr-mCherry (also called “Flk1::myr-mCherry”) (42), maintained on a CD-1 background, were a gift from M. Dickinson (Department of Molecular Physiology and Biophysics, Baylor College of Medicine). Csf1r-CreER [Tg(Csf1r-cre/Esr1*)^1Jwp] (31) is a tamoxifen-inducible macrophage Cre line maintained on a mixed FVB/CD-1 background. Rosa-Tomato [Gt(ROSA)26Sor^{tm14(CAG-tdTomato)Hze}] (32), available from Jackson Laboratories, is maintained on a B6 background. Cx3cr1-Cre (also called “MW126-Cre”), a constitutive macrophage/microglial BAC-Cre line generated by the GENSAT project (57), was a gift from M. D. Gunn (Departments of Immunology and Medicine, Duke University Medical Center) and is maintained on a 129/B6 mixed background. Rosa-eGFP-DTA mice [Gt(ROSA)26SOR^tm1(DTA)Jpmb] (45) are maintained on a B6 background and are publicly available from Jackson Laboratories. Farnesylated-GFP reporter mice (Rosa26R-CAG-fGFP) (46) were a gift from F. Wang (Department of Cell Biology, Duke University Medical Center) and are maintained on a B6 background. Tie2-Cre (also called “Tek-Cre”) [Tg(Tek-cre)^1Ywa] mice (40) were a gift from R. Lang (Divisions of Pediatric Ophthalmology and Developmental Biology, Cincinnati Children's Hospital Medical Center) but are publicly available from Jackson Laboratories. Mice were housed in accordance with National Institutes of Health guidelines, and experimental protocols were approved by the Institutional Animal Care and Use Committee of Duke University Medical Center and/or Cincinnati Children’s Hospital Medical Center.

Yolk-Sac Macrophage Lineage Tracing.

To label yolk-sac progenitors specifically, we followed a previously described protocol using Csf1r-CreER (19), except that we used a Rosa-Tomato Cre–responsive lineage tracer (32). Csf1r-CreER males were crossed to Rosa-Tomato females, and pregnant females were injected i.p. at E7.5 with 75 μg/g 4-hydroxytamoxifen (Sigma-Aldrich #H6278) and s.c. with 75 μg/g progesterone (Sigma-Aldrich #P0130). Pregnant females were euthanized at the required stage of pregnancy, and embryos were processed for immunofluorescence or flow cytometry as described below.

Immunofluorescence.

Whole-mount immunofluorescence was performed on whole embryos, fetal gonads and yolk sacs at E13.5 and younger as previously described (8), and on cryosections of brains, livers, and adult testes as previously reported (58). Primary antibodies used for immunofluorescence are listed in Table S1. Alexa-488– and Alexa-647–conjugated secondary antibodies (Molecular Probes/Life Technologies) and Dy-Lite 488 donkey anti-chicken and Cy3 donkey or goat anti-rabbit/rat secondary antibodies (Jackson ImmunoResearch) were used at 1:500. Nuclei were stained with DAPI (Sigma) or Hoechst 33342 (Molecular Probes/Life Technologies), but nuclear stains are shown as “DAPI” in all figure panels. Samples were imaged on a Leica SP2 confocal microscope or Nikon EclipseTE2000-E microscope equipped with an OptiGrid structured illumination imaging system.

Flow Cytometry Analysis.

Fetal gonads, livers, and brains were disassociated by 10-min, 2-min, and 3-min incubations, respectively, in 0.25% Trypsin-EDTA (Gibco/Life Technologies #2520056). Livers and brains were ruptured mechanically before trypsin treatment. After several washes in PBS, tissues were resuspended in 800 μL PBS plus 4 μL DNase I (Promega #M6101). Gonads were pipetted vigorously to disrupt tissue and were passed several times through a 27-gauge needle (BD Biosciences #309623). Cell suspensions were filtered through a cell strainer cap (Falcon/Corning Life Sciences #352235).

After a 10-min incubation in anti-CD16/32 (FCgammaIII/II, clone 2.4G2, a gift from S. Thornton, Division of Rheumatology, Cincinnati Children's Hospital Medical Center), antibody mixes were added and incubated for 20 min. Antibodies and reagents used are listed in Table S2. Cells were washed, and flow cytometry was performed on either a BD Biosciences LSRII flow cytometer or a BD Biosciences Fortessa flow cytometer. All data were analyzed using FACS Diva software (BD Biosciences). Statistical analyses were performed via a two-tailed Student t test.

FACS Purification of Fetal Macrophages.

Macrophages were purified from Cx3cr1-GFP heterozygous embryos obtained from crossing Cx3cr1-GFP heterozygous males to CD-1 females. GFP expression in embryos was determined by fluorescence microscopy prior to dissection. Cells were disassociated and prepared for FACS as previously described (33). Cell purification was performed on a Beckman Coulter MoFlo XDP cell sorter. Cells were collected into PBS in a 1.5-mL microcentrifuge tube, pelleted by a 5-min spin at 3,500 × g at 4 °C, and stored at −80 °C until RNA extraction was performed.

PCR.

RNA extraction, cDNA synthesis, and qPCR were performed as previously described (58), except that Fast SYBR Green Master Mix (Applied Biosystems/Life Technologies #4385610) was used. Expression levels were normalized to Gapdh. Primers used for qPCR analysis are listed in Table S3.

Vascular-Depletion Gonad Cultures.

Whole-gonad droplet cultures were performed on wild-type CD-1 E11.5 or E12.5 testes as previously described (41), except cultures were performed for 42–44 h. One gonad of each pair was cultured in a 30-μL droplet of control medium (DMEM/FBS/antibiotic mixture); the contralateral gonad was cultured in a 30-μL droplet of medium containing 1.5 μg/mL VEGF receptor Tyrosine Kinase Inhibitor II (VEGFR TKI II; Calbiochem/EMD Millipore #676481–5MG). This concentration was sufficient to deplete vasculature effectively but did not result in increased apoptosis in nonendothelial cells or significantly affect Sertoli/germ cell numbers or gene expression. VEGFR TKI II targets both KDR (VEGFR2) and FLT-1 (VEGFR1). Sex of embryos was confirmed by a PCR-based method (59). Three independent experiments were performed for each time point (E11.5 or E12.5), and within each experiment multiple XY gonads (n = 3–6) were removed from the mesonephros and pooled for RNA extraction. Statistical analyses were done using a two-tailed Student t test.