Two distinct origins for Leydig cell progenitors in the fetal testis

Abstract

During the differentiation of the mammalian embryonic testis, two compartments are defined: the testis cords and the interstitium. The testis cords give rise to the adult seminiferous tubules, whereas steroidogenic Leydig cells and other less well characterized cell types differentiate in the interstitium (the space between testis cords). Although the process of testis cord formation is essential for male development, it is not entirely understood. It has been viewed as a Sertoli-cell driven process, but growing evidence suggests that interstitial cells play an essential role during testis formation. However, little is known about the origin of the interstitium or the molecular and cellular diversity within this early stromal compartment. To better understand the process of mammalian gonad differentiation, we have undertaken an analysis of developing interstitial/stromal cells in the early mouse testis and ovary. We have discovered molecular heterogeneity in the interstitium and have characterized new markers of distinct cell types in the gonad: MAFB, C-MAF, and VCAM1. Our results show that at least two distinct progenitor lineages give rise to the interstitial/stromal compartment of the gonad: the coelomic epithelium and specialized cells along the gonad-mesonephros border. We demonstrate that both these populations give rise to interstitial precursors that can differentiate into fetal Leydig cells. Our analysis also reveals that perivascular cells migrate into the gonad from the mesonephric border along with endothelial cells and that these vessel-associated cells likely represent an interstitial precursor lineage. This study highlights the cellular diversity of the interstitial cell population and suggests that complex cell-cell interactions among cells in the interstitium are involved in testis morphogenesis.

INTRODUCTION

The primordial gonad arises from the proliferation of the coelomic epithelial cell layer overlying the mesonephros, starting at around E10.5 (Karl and Capel, 1998; Schmahl et al., 2000). Between E10.5 and E11.5, the gonad is undifferentiated and morphologically indistinguishable between the sexes, but upon expression of the dominant Y-linked Sry gene in XY gonads, molecular and morphological changes occur rapidly (Gubbay et al., 1990; Hacker et al., 1995). Sertoli cells express Sry and its downstream target, Sox9, and begin to corral germ cells into testis cords, the precursors of adult seminiferous tubules. Given their central role in sex determination, Sertoli cells have been assumed to be the driving force in testis cord formation. However, other cell types in the testis are also required for testis morphogenesis. Vascular endothelial cells migrate into the gonad from the neighboring mesonephros during this time (Combes et al., 2009; Cool et al., 2008; Coveney et al., 2008; Martineau et al., 1997) and form a hallmark male-specific coelomic vessel, with branches that track through the interstitial space between testis cords. Migrating endothelial cells, presumably attracted to VEGFA signals coming from gonadal tissue, are required for cord formation (Baltes-Breitwisch et al., 2010; Bott et al., 2010; Bott et al., 2006; Combes et al., 2009; Cool et al., in press). However, more recent analysis suggests that the cord-forming influence of the endothelium is mediated indirectly through vascular-interstitial cell interactions (Cool et al., in press). This work is consistent with work in other organs, such as the prostate, gut, and mammary gland, which increasingly highlights the importance of the stromal compartment during morphogenesis and neoplastic growth (Cunha et al., 1983; Duluc et al., 1997; Sakakura et al., 1976).

The interstitium of the testis harbors the steroidogenic Leydig cell lineage and other uncharacterized cell types. However, the origin and role of the interstitial cell population has not been carefully studied and is poorly understood. Previous studies indicated that cell divisions in the coelomic epithelium give rise to both Sertoli and non-Sertoli cells (Karl and Capel, 1998; Schmahl et al., 2000), but distinct interstitial cell types were not identified in those studies. Other studies reported that mesonephric tissue could migrate into the gonad and give rise to interstitial and Leydig cells (Buehr et al., 1993; Merchant-Larios and Moreno-Mendoza, 1998; Nishino et al., 2001; Val et al., 2006). However, more recent efforts to identify migrating Leydig cells in organ recombination experiments have been unsuccessful (Combes et al., 2009; Cool et al., 2008).

The female counterpart of the Sertoli cell is the pre-granulosa (follicle) cell, marked by the forkhead transcription factor FOXL2 (Albrecht and Eicher, 2001; Cocquet et al., 2002; Loffler et al., 2003). However, there is no direct parallel to the interstitial compartment of the testis. In the case of the ovary, the non-granulosa cell population includes the steroidogenic theca cells that surround follicles. As steroidogenesis does not occur until near or after birth (Mannan and O'Shaughnessy, 1991), a female fetal steroidogenic cell precursor may not exist or may be maintained in a naive state. Uncharacterized stromal cells occupy the space between follicles in the ovarian cortex and in the medulla of the ovary, which harbors the major vessels and nerves in the adult organ. The stromal population is likely complex and may differ between the cortex and medulla, but has not been characterized.

To precisely determine the origin and fate of interstitial cells in the developing gonad, molecular markers of different cell types are required. However, it has proven difficult to find cell-type-specific interstitial markers. One report focusing on testis peritubular myoid cells (a specific interstitial cell type) demonstrated that fetal interstitial cells uniformly express the same genes and myoid gene expression only restricts later during testis morphogenesis (Jeanes et al., 2005), suggesting that interstitial cells are initially a homogeneous population. However, several other studies implied that distinct cell populations might already exist within the nascent interstitial compartment (Cool et al., in press; Tang et al., 2008). The identification of different cell types within the interstitial population could be critical to understanding multiple aspects of testis morphogenesis, such as the development of the Leydig cell lineage, formation of testis cord architecture, and maintenance of progenitor populations in this compartment.

In this study, we performed a detailed analysis of the interstitial compartment in the developing mouse gonad. Our results show that, from the earliest stages of gonad formation, the interstitium is a diverse cell population that intimately interacts with vasculature during organogenesis. We characterized the localization of two Maf transcription factors (MAFB and C-MAF) and a cell-surface molecule (VCAM1) as early markers for distinct interstitial cell types in male and female gonads. We show that two cell populations that are likely multipotent, the coelomic epithelium and gonad-mesonephros border cells, give rise to steroidogenic Leydig cells of the testis. Early interstitial cells are intimately associated with invading blood vessels. A subset of perivascular cells expresses a marker for Leydig cell progenitors and likely represents an interstitial precursor population. Our work defines the origins of various interstitial cell types and positions the field towards elucidating their unique roles during morphogenesis.

MATERIALS AND METHODS

Mouse lines

CD-1 mice (Charles River) were used to investigate the expression patterns of MAFB, C-MAF, and VCAM1, and for organ culture experiments. To detect Sertoli cells, a Sox9-CFP transgenic line was utilized (Kim et al., 2007). Mafb-GFP mice (Moriguchi et al., 2006) were used in live imaging and interstitial cell labeling experiments. We obtained similar results when Mafb-GFP was crossed to CD-1 or C57BL/6J (B6) backgrounds. For endothelial cell tracing experiments, a Z/EG Cre-responsive GFP reporter strain (Novak et al., 2000) on a mixed background (129S1/SvImJ; B6) was crossed to an endothelial cell-specific Flk1-Cre line (Motoike et al., 2003) maintained on a B6 background. Both Wnt4 (Vainio et al., 1999) and Fgf9 (Colvin et al., 2001) mutant strains were maintained on a B6 background. Smaa-YFP mice (Cool et al., 2008) were obtained from J. Lessard and are carried on a CD-1 genetic background. Ubiquitous eGFP-expressing mice used for recombination experiments were Tg(CAG-EGFP)B5Nagy/J (Jackson Labs), maintained on a FVB/NJ background.

Timed matings were performed with the day vaginal plugs are detected considered E0.5. For accurate embryo staging, tail somite (ts) number was counted at the time of dissection. Using this method, E10.5 corresponds to 8 ts, E11.5 to 18 ts and E12.5 to 30 ts (Hacker et al., 1995).

Immunofluorescence and antibodies

Samples were dissected in PBS and fixed in 4% paraformaldehyde overnight at 4°C. After several washes in PBS+0.1% Triton X-100 (PBTx), samples were incubated in blocking solution (PBTx+10%FBS+10%BSA) for 1–2 hours at room temperature or overnight at 4°C.

Primary antibodies used were: anti-PECAM1 (BD Pharmingen, 1:250), anti-LHX9 (a gift from T. Jessell, 1:5000), anti-MAFA (Bethyl Labs, 1:2000), anti-MAFB (Bethyl Labs, 1:2000), anti-C-MAF (Bethyl Labs, 1:2000), anti-3β-HSD (a gift from K. Morohashi, 1:2000) (Baba et al., 2008), anti-FOXL2 (Novus, 1:200), anti-VCAM1 (R&D, 1:2000), anti-GATA4 (Santa Cruz, 1:100), anti-CD11b (BD Pharmingen, 1:250), anti-SF1 (a gift from K. Morohashi, 1:2000), and anti-BrdU (Accurate Chemical, 1:200).

Primary antibodies were diluted in blocking solution and applied to samples overnight at 4°C. Fluorescent secondary antibodies were applied for 4–5 hours at room temperature or overnight at 4°C. Cy5-, Cy3-, or Cy2-conjugated secondary antibodies (Jackson ImmunoResearch, 1:500 for Cy5 and Cy3 and 1:200 for Cy2) or Alexa 647-, Alexa 555-, or Alexa 488-conjugated secondary antibodies (Molecular Probes, 1:500) were used for immunofluorescence. DAPI (Sigma-Aldrich) or Syto13 (Invitrogen) was used to stain nuclei. Samples were mounted on slides using 2.5% DABCO (Sigma-Aldrich) in 90% glycerol and imaged on a Leica SP2 confocal microscope.

Gonad explant culture

Gonad-mesonephros complexes were cultured in 1.5% agar blocks at 37°C with 5% CO₂. Culture medium was Dulbecco’s Minimal Eagle Medium (DMEM) containing 10% fetal bovine serum and 50µg/ml of ampicillin.

GFP recombination cultures

Whole genital ridges were removed from E11.5 embryos, which were sexed by PCR for presence of X and Y chromosomes (Clapcote and Roder, 2005). After separating the gonad and mesonephros, wild-type XY gonads and XX or XY eGFP-expressing mesonephroi were recombined on agar blocks as previously described (Martineau et al., 1997). In some cultures, the technique was modified so that the nascent vasculature in the gonad-mesonephros border region was not disturbed during gonad-mesonephros separation. All recombinations were then cultured for 48–64 hours and processed for immunostaining.

Ectopic FGF9 experiments

Wild-type E11.5 CD-1 gonad-mesonephros complexes were dissected and sexed via PCR methods. Affi-Gel Blue Gel beads (Bio-Rad Laboratories #153–7301) were incubated for 3 hours at room temperature with either recombinant human FGF9 (R&D #273-F9–025, 16 µg/ml) or 10% BSA in culture medium, and washed 5 times in culture medium before placement on gonad surface. A single FGF9- or BSA-coated bead was placed onto the surface of the gonad, cultured for 48 hours, and processed for immunostaining.

Aflibercept injections

Wild-type CD-1 E11.3-E11.5 embryos (16–18 ts) were dissected from the uterus without rupturing the umbilical cord or major blood vessels in the yolk sac. Glass microinjection needles were front-loaded with rhodamine-lectin or a mix of rhodamine-lectin and Aflibercept (also called VEGF-Trap; Regeneron, 38 µg/ml) (Holash et al., 2002). Embryos were injected in the right ventricle while the heart was still beating to allow injected reagents to perfuse into the gonad-mesonephros vascular network. After 15–20 minutes, gonad/mesonephric complexes were dissected and inspected using a fluorescence microscope to determine whether dye was effectively delivered. After 24 hours in culture, gonads were fixed and processed for immunofluorescence (Cool et al., in press).

MitoTracker labeling

To label coelomic epithelial cells, MitoTracker Orange CMTMRos (Invitrogen #M7510, 500 nM) was pipetted onto the surface of Mafb-GFP gonad-mesonephros complexes and incubated for 45 minutes. After several washes in culture medium, gonads were cultured for 24–48 hours. As a control, several gonads were removed and imaged after 1 hour of culture to verify that only the surface coelomic epithelial cells were initially labeled. Since Mafb-GFP is a homozygous lethal mutation (Moriguchi et al., 2006) and may affect gonad development, Mafb-GFP heterozygous males were crossed to CD-1 females to ensure that all labeled embryos selected for experiments were heterozygous.

Time lapse live imaging

All Mafb-GFP imaging experiments were performed on a Leica SP2 confocal microscope, as described previously (Coveney et al., 2008). Gonads were cultured for 17–20 hours, using standard culture conditions, and 6–10 Z-stacks were collected every 6–10 minutes. Movies were constructed using maximum intensity projections at each time point. Mafb-GFP heterozygous males were crossed to CD-1 females to ensure that no labeled embryos were homozygous mutants.

BrdU labeling

Wild-type CD-1 embryos were dissected at E12.0 (around 24 ts) and cultured for 2 hours using a droplet method (Maatouk et al., 2008), in 30 µl droplets containing 3.125 µg/ml BrdU (Sigma). Samples were immediately fixed for 1 hour at room temperature in a 30% 50 mM glycine/70% ethanol fixative, washed several times in PBS, and subsequently treated with 2M HCl for 30 minutes at room temperature. Samples were then washed several times in PBTx and processed for immunofluorescence using an anti-BrdU antibody. Sexing of embryos was performed using PCR.

RESULTS

Interstitial cells remain in a progenitor state during initial testis morphogenesis

Prior to the expression of Sry, cells in the nascent gonad (thought to be mostly derived from the coelomic epithelium) express the homeobox gene Lhx9 (Birk et al., 2000; Cocquet et al., 2002; Mazaud et al., 2002; Wilhelm and Englert, 2002), which is required for gonad primordium development and has been described as a marker of an undifferentiated, progenitor state (Birk et al., 2000; Mazaud et al., 2002). In XX gonads, we do not see downregulation of LHX9 at E11.3 or E11.7, as LHX9 is uniformly expressed throughout the somatic gonad (Fig. 1A, A’ and 1B, B’). This observation is consistent with the later differentiation of FoxL2-expressing granulosa cell precursors in the female gonad (Loffler et al., 2003). In E12.3 XX embryos, LHX9 is still expressed throughout the gonad, although expression is strongest in the coelomic epithelium and is slightly downregulated in the center of the gonad, presumably in pre-granulosa cells as they differentiate (data not shown).

Figure 1. Interstitial cells maintain an undifferentiated status during early testis formation.

Immunofluorescence images of embryonic gonads. Color of markers is indicated beneath panels. PECAM1 labels germ cells and vasculature. Black and white panels show LHX9 expression alone. White dashed lines indicate the gonad-mesonephros border. In XX gonads, LHX9 is uniform throughout the tissue at both E11.3 (15 ts) (A, A’) and E11.7 (20–21 ts) (B, B’). In E11.3 XY gonads (C, C’), LHX9 is downregulated in the central region where Sertoli cells have started to differentiate (green bracket). By E11.7 (20–21 ts), LHX9 expression in XY gonads (D, D’) is restricted to the coelomic epithelium, scattered cells in the center of the tissue, and in cells lining the gonad-mesonephros border (arrowheads). Co-labeling with a Sertoli cell marker (Sox9-CFP) shows that LHX9 and Sox9 expression are mutually exclusive both at E11.7 (E, E’) and E12.3 (F, F’). Green dashed line in E’ denotes Sertoli cell-interstitial cell boundary. g, gonad; m, mesonephros. Scale bar in A represents 50 µm in all panels.

By contrast, in XY gonads, Sertoli cells, the first male-specific cell type in the developing gonad, are specified by their expression of the male sex-determining gene Sry between E10.5 and E12.5. Sry is expressed in a center to pole pattern, and activates its direct target Sox9 similarly (Albrecht and Eicher, 2001; Bullejos and Koopman, 2001; Moreno-Mendoza et al., 2003; Schepers et al., 2003). The center-to-pole wave of activation of Sox9 is associated with a concomitant downregulation of LHX9 in a center-to-pole fashion (Fig. 1C, C’ and D, D’). Most somatic cells that have not committed to the Sertoli lineage express LHX9 during early testis formation, including progenitor cells in the coelomic epithelium, cells underneath the epithelium (“coelomic domain”) adjacent to nascent Sertoli cells (Fig. 1E, E’), and a few scattered LHX9-positive cells that persist throughout the middle of the XY gonad and along the highly vascularized gonad-mesonephros border region (Fig. 1D, D’, arrowheads). In E12.3 XY gonads, LHX9 expression is further restricted to the coelomic domain and scattered cells in the interstitial compartment (Fig. 1F, F’). These cells may represent a progenitor population that can give rise to heterogeneous cell types in the interstitium. To further study this population and its role during testis morphogenesis, we sought molecular markers that definitively distinguish various early interstitial cell types.

MAF family members are dynamically expressed in interstitial precursors

In Drosophila gonad development, traffic jam (tj) regulates cell adhesion interactions during morphogenesis (Li et al., 2003). tj encodes a large Maf basic leucine transcription factor with multiple mammalian orthologs, the closest of which are MAFA, MAFB, and C-MAF. Using immunofluorescence and confocal microscopy, we determined the expression pattern of MAFA, MAFB, and C-MAF in the mouse gonad between E11.5 and E13.5, stages during which sexual differentiation and initial gonad morphogenesis take place.

MAFA expression was restricted to a minor subset of cells in the male gonad and was rarely observed in the female gonad between E11.5-E13.5 (Supplementary Fig. 1A, B and data not shown). MAFA was occasionally observed in testis cords, but in Sox9-negative cells. These cells are likely macrophages, as MAFA expression overlapped with expression of the monocyte/macrophage marker CD11b (Mac1) (Supplementary Fig. 1C). However, we cannot exclude the possibility that MAFA-positive cells are being phagocytosed by macrophages.

In contrast, MAFB showed a dynamic expression pattern in the embryonic gonad. At E11.5, MAFB was observed in the gonad-mesonephros border region and in a subset of somatic cells throughout the gonad in both sexes (Fig. 2A, B). Early MAFB expression in both sexes was most often associated with nascent vasculature in the gonad-mesonephros complex (Fig. 2A–D, arrows). Sexually dimorphic expression was observed in the coelomic domain of the gonad, just beneath the coelomic epithelium. Expression of MAFB in this domain was seen in scattered cells only in XY gonads (Fig. 2A, C, arrowheads). By E12.5, MAFB expression expanded significantly in XY gonads to include most somatic cells that lie outside testis cords (the interstitial cell population), whereas expression was still restricted to cells immediately adjacent to blood vessels in XX gonads (Fig. 2E, F). By E13.5, MAFB expression contracted to a subset of somatic interstitial cells in the testis (Fig. 2G), while XX expression was unchanged (Fig. 2H).

Figure 2. Maf transcription factors are expressed dynamically in the gonad.

Fluorescent antibody immunostainings for MAFB (A–H) and C-MAF (I–P) in XY and XX gonads between E11.5 and E13.5 (green). In all panels, anti-PECAM1 (purple) labels germ cells and vasculature. Dashed lines indicate the border between gonad and mesonephros. In XY gonads, MAFB is expressed in the coelomic domain of the gonad underneath the coelomic epithelium (A, C, arrowheads), and is expressed in cells along the highly vascularized gonad-mesonephros border in both sexes (A–D, arrows). Between E12.0-E13.5, MAFB expression expands in the interstitial compartment and then restricts in XY gonads (C, E, G), but the expression pattern does not change significantly in XX gonads (D, F, H). C-MAF is expressed in very few cells within the gonad at E11.5 (I, J), but is expressed within the gonad in both sexes at E12.0 and is enriched in the gonad-mesonephros border region (K, L, arrows). By E13.5, C-MAF expands in the interstitial compartment in males (M, O), but the expression pattern is constant in females (N, P). Scale bar in A represents 50 µm in all panels, except G (100 µm), K and O (25 µm).

C-MAF expression at E11.5 was less widespread than MAFB, and was restricted to the mesonephros and a very few gonadal cells in both sexes (Fig. 2I, J). By E12.5, similar to MAFB, C-MAF became sexually dimorphic. C-MAF expression was increased in interstitial cells, while expression in XX gonads was limited to vascular-associated cells (Fig. 2K–N). C-MAF was strongly expressed in the gonad-mesonephros border region in both sexes (Fig. 2K, L, arrows. Interestingly, expansion of C-MAF was complementary to MAFB: sexually dimorphic C-MAF expansion started from the mesonephric domain of the gonad and proceeded towards the coelomic domain, whereas MAFB started from the coelomic domain and expanded toward the mesonephric border. By E13.5, C-MAF was expressed in many (but not all) interstitial cells in the testis, while expression in XX gonads was restricted to vascular-associated cells (Fig. 2O, P).

MAF family members are expressed in unique interstitial cell populations

To determine the relationship between MAFB- and C-MAF-expressing cells and the Sertoli cell lineage, we stained Sox9-CFP transgenic gonads (Kim et al., 2007) for MAFB and C-MAF. Starting at the earliest stages of gonad expression, both MAFB (E11.5) and C-MAF (E12.0) expression were mutually exclusive of Sox9-CFP, (Fig. 3A, A’, B, B’), demonstrating that they are markers of early interstitial cells.

Figure 3. Mafs label unique interstitial/stromal cell lineages.

Immunostainings of XY (A–E) and XX (F–G) embryonic gonads. Color of each marker is indicated in each panel. Dashed lines denote gonad-mesonephros border or boxed area of magnification. A’, B’, C’, F’, and G’ are higher magnification images of the boxed regions in A, B, C, F, and G. Anti-PECAM1 staining is in blue in all panels. At E11.5 (A, A’) and E12.0 (B, B’), MAFB and C-MAF expression are mutually exclusive with Sox9-CFP at both coelomic and gonad-mesonephros border regions (arrowheads), and mark early interstitial cell fate. Note that Sox9-CFP is not restricted to the nucleus in Sertoli progenitors. By E13.0, C-MAF and Mafb-GFP show extensive overlap (C, C’), but by E14.5, Mafb-GFP is enriched in Leydig cells (D) and C-MAF is enriched in non-Leydig interstitial cells (E). In females, Mafb-GFP and C-MAF colocalize (F, F’) in cells that do not express the pre-granulosa marker FOXL2 (G, G’). Scale bar in A represents 50 µm in A, F and G, 17 µm in A’, 100 µm in B and C, 13 µm in B’, 34 µm in C’, D, and E. 22 µm in F’ and G’.

Using a Mafb-GFP knock-in transgenic line as a reporter for Mafb expression (Moriguchi et al., 2006), we determined whether Mafb and C-MAF were expressed in overlapping cell populations. At E13.0, Mafb and C-MAF showed significant but incomplete overlap (Fig. 3C, C’), suggesting that they either mark different interstitial populations or Mafb has begun to restrict by this point and is no longer expressed in some C-MAF-positive cells. At later stages, Mafb and C-MAF became restricted to unique, mutually exclusive interstitial cell types. By E14.5, Mafb-GFP was observed in 3β-HSD-positive cells (Fig. 3D), a marker of differentiated steroidogenic Leydig cells. In contrast, C-MAF expression was detected exclusively in Mafb-GFP-negative cells (Fig. 3E), suggesting that C-MAF is localized to non-Leydig interstitial cells. This expression pattern suggests that MAFB and C-MAF could act in concert at earlier stages of gonad development, but later segregate to distinct interstitial lineages.

To determine whether MAFB also labels the non-supporting cell lineage in the XX gonad, we investigated the identity of MAFB-positive cells in the female, using FOXL2 as a marker of pre-granulosa cells in the XX embryonic gonad (Cocquet et al., 2002; Loffler et al., 2003). In all embryonic stages we investigated, MAFB and C-MAF expression overlapped and were mutually exclusive with FOXL2 expression (Fig. 3F, F’ and 3G, G’), indicating that MAFB and C-MAF may also label a paralogous cell type in the ovary. Given the steroidogenic nature of MAFB-expressing cells in the testis and the neuroendocrine expression of Mafs in the pancreas (Nishimura et al., 2006), it is possible that the Maf proteins label a precursor of the female steroidogenic theca cell lineage.

MAF expression is associated with nascent vasculature in the gonad

MAFB and C-MAF were expressed in the highly vascularized gonad-mesonephric border region (Figs. 2 and 3). Our laboratory and others have shown that invading endothelial cells from the mesonephros are required for testis morphogenesis (Buehr et al., 1993; Combes et al., 2009; Cool et al., 2008; Martineau et al., 1997; Merchant-Larios et al., 1993; Tilmann and Capel, 1999). However, it is not clear exactly how blood vessels interact with the gonad on a cellular level to drive the formation of testis structures.

At E11.5 Mafb-GFP expression was highly enriched at sites where invading endothelial cells enter the gonad from the mesonephros (Fig. 4A, A’). MAFB/C-MAF-positive cells adjacent to endothelial cells in both sexes express Vascular Cell Adhesion Molecule-1 (VCAM1) and Smooth muscle alpha actin-YFP (Smaa-YFP) (Fig. 4B, B’ and Supplementary Fig. 2), both of which are expressed in perivascular smooth muscle cells (Briscoe et al., 1992; Cool et al., 2008; Gabbiani et al., 1981; O'Brien et al., 1993). In Mafb-GFP- and VCAM1-positive perivascular cells throughout the gonad and mesonephric border region, LHX9 was frequently co-expressed (Fig. 4C, C’ and 4D, D’), suggesting that these perivascular cells were in an undifferentiated state. The border region cells did not express GATA4 or SF1 (Supplementary Fig. 3), general markers for gonadal somatic cells at this stage, suggesting that they represent a novel population of interstitial cells in the gonad that may have a distinct origin.

Figure 4. MAFB expression is associated with nascent vasculature in the gonad.

Immunofluorescence of embryonic gonads. Color of each marker is indicated in each panel. Dashed lines denote gonad-mesonephros border or boxed area of magnification. A’, B’, C’, and D’ are higher-magnification images of the boxes in A, B, C, and D, although A’ is in a different optical plane than A. At E11.5 (A, B), Mafb-GFP and VCAM1 are enriched in gonad-mesonephros border cells (A, arrowheads) intimately associated with endothelial cells (A’, B’, arrows) branching into the gonad from the mesonephros. These border cells express LHX9 and likely represent progenitor cells (C, D, arrowheads in C’ and D’). At E10.5 (E), MAFB is limited to expression in the GATA4-negative region of the gonad-mesonephros complex. (F, G) Flk1-Cre; Z/EG lineage tracing of endothelial cells dissected at E12.5 shows that MAFB-positive cells do not overlap with GFP, therefore are not directly derived from the endothelial cell lineage in either sex. g, gonad; m, mesonephros. Scale bar in A represents 30 µm in A, 13 µm in A’, 50 µm in B, C, D, F, and G. 14 µm in B’, and 21 µm in C’ and D’, and 25 µm in E.

Prior to vascularization of the gonad, labeled by GATA4 (somatic cells) and PECAM1 (germ cells) expression, MAFB-positive cells are observed only in the mesonephros and at a few sites along the mesonephric border (Fig. 4E). Given their close association with blood vessels, we investigated whether Mafb-expressing cells in the gonad are directly derived from the endothelium. To test this hypothesis, we used a genetic method to lineage trace endothelial cells using an endothelial cell-specific Flk1-Cre (Motoike et al., 2003) to drive permanent GFP expression via a Z/EG reporter (Novak et al., 2000). This technique efficiently labels blood vessels in the gonad, yet we do not observe MAFB expression in >95% of GFP-positive cells in either sex (Fig. 4F, G), indicating that vascular-associated MAFB-expressing cells are not directly derived from Flk1-positive endothelial/perivascular cells.

To further assess the relationship between vasculature and the interstitial cell population, we asked whether the induction of ectopic (male-like) vasculature was sufficient to induce male-specific patterns of MAFB expression in XX gonads. Adding exogenous FGF9 to XX gonads has been shown to recruit ectopic vasculature (Colvin et al., 2001). FGF9-coated beads placed on the surface of XX gonads efficiently recruited blood vessels, and MAFB expression near those vessels was significantly expanded relative to control BSA-coated beads (Fig. 5A, B), and similar to what is seen in the interstitium of wild type males. However, we cannot definitively determine whether additional MAFB-positive cells migrated into the gonad along with ectopic vasculature, or the ectopic vasculature induced new MAFB expression or increased proliferation of existing MAFB-positive cells. Furthermore, FGF9-bead-treated XX gonads exhibit 3β-HSD-expressing cells, indicating that Leydig cells have been induced to differentiate (Supplementary Fig. 4). Interestingly, ectopic Leydig cells are associated with recruited vasculature, even at long distances from the bead, suggesting that the vasculature, and not solely FGF9, is responsible for Leydig cell differentiation. However, we also see ectopic SOX9 expression in close proximity to the bead in FGF9-treated XX gonads (Supplementary Fig. 4 and (Kim et al., 2006)), so there is a possibility that the SOX9-positive Sertoli-like cells are acting as an intermediary by inducing MAFB or 3β-HSD expression in this experiment.

Figure 5. Vasculature is sufficient to induce MAFB expression and is required to expand the MAFB population.

Immunostainings of XX (A–D) and XY (E–H) embryonic gonads. A, B, E, and F are E11.5 gonads cultured for either 24 hours (E, F) or 48 hours (A, B). C, D, G, and H are E12.5 gonads. Color of each marker is indicated. Dashed lines denote gonad-mesonephros border. The presence of ectopic vasculature (A, B, arrow) recruited to an FGF9-coated bead cultured atop an XX gonad is accompanied by a dramatic increase in MAFB expression (A, B, arrowhead). Wnt4 XX mutants, which exhibit male-like coelomic vasculature, also show a significant increase in the number of MAFB-expressing cells (C, D). Blocking vasculature in an XY gonad via the anti-angiogenic reagent Aflibercept results in a decrease in the MAFB-positive population in the coelomic domain (E, F). An Ffg9 mutation, which blocks male development and male-specific vascularization, leads to strong downregulation of MAFB (G, H). Scale bar in A represents 50 µm in all panels.

These experiments suggest that the vasculature directly regulates the MAFB population. To test this idea further, we used a genetic model, in which Wnt4 mutation causes male-like coelomic vasculature to form in XX gonads (Jeays-Ward et al., 2003). In XX Wnt4 mutants, the domain of MAFB-positive cells is similar to the one in XY gonads (Fig. 5C, D). Interestingly, MAFB-expressing cells partition the gonad into testis cord-like domains that are not observed in wild-type XX gonads.

To complement our ectopic vasculature experiments, we also tested whether blood vessels are necessary for MAFB expression. To inhibit vascularization of the XY gonad, we utilized a blocking antibody for VEGFA, Aflibercept, which is an anti-angiogenic agent (Holash et al., 2002). Aflibercept can also act by blocking the function of Placental Growth Factor (PLGF). However, microarray analysis on sorted gonadal cell populations suggests that Plgf is not expressed in the gonad during gonad vascularization (E11.5 and E12.5) (data not shown). Thus, Aflibercept is likely acting to block gonadal vasculature specifically through its inhibitory effects on VEGFA-dependent signaling.

Injecting Aflibercept and a fluorescent lectin into the beating hearts of E11.5 embryos delivers the reagents to the gonad-mesonephric border and effectively blocks the invasion of endothelial cells into the XY gonad, which prevents the formation of the hallmark coelomic vessel and leads to a reduction in interstitial cells (Cool et al., in press). In Aflibercept-treated gonads, the coelomic MAFB expression domain is significantly reduced relative to controls injected with lectin alone (Fig. 5E, F). However, MAFB expression is not completely eliminated. This residual expression of MAFB in Aflibercept-treated XY gonads is likely due to cells expressing MAFB in the coelomic domain before, and likely independently of, vascular invasion (see Fig. 2A, arrowhead). This data indicates that, even in the absence of the vasculature, there is underlying sexual dimorphism in the specification of this lineage.

We see more dramatic results with Aflibercept treatment using VCAM1 as a marker for perivascular interstitial cells, which, unlike MAFB, is not expressed in the coelomic domain prior to vascularization and is initially only expressed in vessel-associated cells near the gonad-mesonephros border (see Supplementary Fig. 2). VCAM1 expression in the coelomic domain of Aflibercept-treated XY gonads is severely disrupted relative to control gonads and often cannot be detected (Supplementary Fig. 5A–D). In these samples, VCAM1 is only observed near the gonad-mesonephros border in groups of cells that were likely associated with vascular sprouts already present when the treatment was initiated (see Fig. 4B and 4D). These results indicate that the perivascular VCAM1/MAFB double-positive cells are distinct from coelomic MAFB single-positive cells in that perivascular cells have a greater requirement for vasculature to induce their migration or proliferation. Consistent with a role for vasculature in influencing proliferation, BrdU experiments reveal that VCAM1-expressing cells are highly proliferative at E12.0 (Supplementary Fig. 5E–F), at the time when vascularization and gonad morphogenesis are taking place.

As a genetic method to disrupt vasculature in the gonad, we used an Fgf9 mutation in which general testis development, including coelomic vessel formation, is blocked in XY gonads (Colvin et al., 2001; Schmahl et al., 2004), and a female-type vascular pattern is present. In these mutants, we also fail to observe any coelomic MAFB expression. Instead, MAFB expression is female-like in perivascular cells in the central and mesonephric domains of the gonad (Fig. 5G, H), suggesting that the sexually dimorphic pattern of MAFB expression is downstream of Fgf9, as are most patterns of testis organogenesis.

Leydig cells are derived from both coelomic epithelial and mesonephric border region precursors

Previous work from our laboratory demonstrated that both Sertoli and interstitial cells are derived from divisions of the coelomic epithelium prior to E11.5 (18 tail somites, ts) (Brennan et al., 2003; Karl and Capel, 1998; Schmahl et al., 2000). Divisions of the epithelium after 18 ts no longer give rise to Sertoli cells, but progeny of later divisions were not identified due to lack of cell-type-specific reagents.

We performed MitoTracker Orange labeling of E11.5 (after 18 ts) gonads to track the fate of cells derived from later divisions of coelomic epithelial cells. Control XX and XY gonads cultured for 1 hour after MitoTracker incubation were fluorescently labeled only in the outermost layer of cells comprising the coelomic epithelium (Fig. 6A, A’), while those cultured for 24 hours showed significant labeling of multiple cell layers deep into the gonad (Fig. 6B, B’). Consistent with previous reports, Sertoli cells adjacent to germ cells in XY gonads were not MitoTracker positive, and only Mafb-GFP-positive interstitial cells were labeled. These results are consistent with the hypothesis that later-stage proliferation of the coelomic epithelium (after 18 ts) gives rise exclusively to interstitial cells (Brennan et al., 2003; Karl and Capel, 1998; Schmahl et al., 2000). Mafb-GFP-positive cells deeper in the gonad were negative for the label, suggesting that these cells arose earlier than the initial MitoTracker labeling or were derived from another source.

Figure 6. Leydig cells can arise from coelomic epithelial-derived precursors.

Immunofluorescence images of cultured embryonic XY gonads with MitoTracker-labeled coelomic epithelia. Color of each marker is indicated in each panel. Dashed lines indicate gonad-mesonephros border or boxed region of magnification. A’, B’, C’, and D’ are higher-magnification images of the boxed regions in A, B, C, and D. One hour after MitoTracker is applied to the coelomic surface of the E11.5 gonad, only the outermost layer of coelomic epithelial cells is labeled (A). Mafb-GFP expression lags behind anti-MAFB antibody staining (see Fig. 2), hence little GFP is seen in the coelomic domain at this stage. After 24 hours in culture, MitoTracker dye is observed 5–10 cell layers into the gonad, and labels Mafb-GFP-positive interstitial cells (B). After 48 hours in culture, MitoTracker labeling is found even deeper into the gonad, and some 3β-HSD-positive Leydig cells retain the label (C’, yellow arrowheads) while others are MitoTracker negative (C’, green arrowhead). After 24 hours in culture, VCAM1, a specific marker of perivascular and gonad-mesonephric border cells, labels some cells in the coelomic domain that are MitoTracker-negative (D). Scale bar in A represents 50 µm in A, B, C, and D, 12 µm in A’ and B’, 23 µm in C’, and 13 µm in D’.

One of the most important cell types that arises within the interstitium is the Leydig cell, which secretes testosterone necessary for virilization of the embryo. The origin of Leydig cells remains unclear. Multiple hypotheses exist for the source of Leydig cells, including the coelomic epithelium and the mesonephros (Brennan et al., 2003; Buehr et al., 1993; Karl and Capel, 1998; Merchant-Larios and Moreno-Mendoza, 1998; Schmahl et al., 2000; Val et al., 2006). We performed 48-hour MitoTracker Orange tracking experiments on 18–19 ts XY gonads, allowing for differentiation of Leydig cells, identified by an anti-3β-HSD antibody. After 48 hours of incubation, MitoTracker-labeled cells were found deep within the gonad, but only in Mafb-GFP positive interstitial regions (Fig. 6C, C’), consistent with 24-hour labeling experiments. 3β-HSD antibody staining revealed that the majority of Leydig cells in the coelomic domain of the gonad were MitoTracker-positive (Fig. 6C, C’), indicating that at least some Leydig cells are derived from divisions of the coelomic epithelium.

Interestingly, MitoTracker-negative Leydig cells were found in areas of strong MitoTracker-positive interstitium, suggesting that some Leydig cells are not derived from the coelomic epithelium and may arise from other regions, such as from the Mafb-positive cells along the gonad-mesonephros border. We combined MitoTracker labeling with VCAM1 antibody staining, which labels only perivascular cells near the mesonephric border in XX and XY gonads between E11.5-E12.0 (see Fig. 4 and Supplementary Fig. 2). VCAM1-positive cells were rarely MitoTracker positive (Fig. 6D, D’), even when they were in an area of strong MitoTracker labeling, suggesting that these cells migrated into the coelomic domain of the gonad from an independent source.

To investigate whether cells migrate into the gonad from the mesonephric border region, we performed live imaging of Mafb-GFP gonads during the early stages of endothelial migration (E11.5-E12.5). Consistent with our culture experiments, we often observed GFP-expressing cells at the gonad-mesonephros border region migrating as far as the coelomic domain of the gonad, likely contributing to the interstitium throughout the gonad (Fig. 7).

Figure 7. Cells from the gonad-mesonephros border region migrate toward the coelomic domain between E11.5-E12.0.

Still-frame images from 7 hours of time-lapse imaging of a Mafb-GFP XY gonad, beginning at E11.5 (18 ts). Dashed line indicates gonad-mesonephros border. Times elapsed are as indicated in upper-right corner of each panel. Arrowheads follow a GFP-expressing cell or group of GFP-expressing cells that migrate from the gonad-mesonephric border to the coelomic domain of the gonad. g, gonad; m, mesonephros. Scale bar represents 100 µm in all panels.

An additional method to assess cell migration between the mesonephros and the gonad is a GFP organ recombination assay, where a wild type gonad is cultured adjacent to a GFP-labeled mesonephros to detect GFP-positive cells that migrate into the gonad. For these assays, the gonad has typically been severed from the mesonephros along the gonad/mesonephric border. Using this method, endothelial migration between the mesonephros and the XY gonad is easily detected. However, additional cell types have rarely been reported (Combes et al., 2009; Cool et al., 2008; Martineau et al., 1997) (Fig. 8A). When we performed GFP recombination culture assays using the standard protocol, we also detected only GFP-positive endothelial cells in the migrating population and failed to observe any GFP-expressing interstitial cells (Fig. 8B). However, based on the evidence from live imaging experiments above, we speculated that separation of the gonad and mesonephros along this border disrupts vascular sprouts or cellular interactions that act as the conduit for a population of MAFB/VCAM1 positive cells moving into the gonad from the bed at the mesonephric/gonad border.

Figure 8. Gonad-mesonephros border cells migrate into the gonad and give rise to Leydig cells.

(A) A cartoon depicting modification of previous GFP recombination experiments. Dashed lines diagram position of gonad-mesonephros recombination in old (gray dashed lines) and new (green dashed lines) experiments. The color of each marker is indicated in panels B–D’. Using the unmodified protocol (B), only endothelial cells are found in the migrating population. In the modified protocol (C–D), which keeps vascular sprouts and perivascular cells intact in the GFP-positive mesonephros portion, both endothelial cells (C’, arrows) and VCAM1-positive interstitial cells (C’, arrowheads) are detected in the migrating population. When cultures are collected after 64 hours, some GFP-positive cells from the mesonephros are positive for 3β-HSD (D’ arrowheads) indicating that they differentiate as Leydig cells (D, D’). Scale bar in B represents 50 µm in B, 100 µm in C and D, 49 µm in C’, and 44 µm in D’.

To determine whether perivascular cells along the gonad-mesonephric border could migrate into the gonad and give rise to a subset of interstitial cells, we modified the GFP recombination assay such that the nascent vascular sprouts and associated perivascular cells in the gonad-mesonephros border remained intact in the mesonephric component of the recombination cultures (Fig. 8A). In these experiments we consistently found GFP-expressing VCAM1 positive interstitial cells migrating into the gonad (Fig. 8C, C’). Interestingly, these GFP-positive cells were usually closely associated with GFP-positive endothelial cells, suggesting that the two cell types migrated into the gonad together. In conjunction with our vascular blocking experiments, these results suggest that intact vascular sprouts are required for these cells to migrate into the gonad from the mesonephric border region.

We also determined if these migrating cells could give rise to Leydig cells, consistent with the hypothesis that gonad-mesonephric cells comprise an additional progenitor population. After 64 hours in culture, which would allow for cells to differentiate into Leydig cells, we observed GFP-positive 3β-HSD-expressing cells (Fig. 8D, D’). This indicates that migrating cells from near the gonad-mesonephric border represent a second progenitor population that can also give rise to Leydig cells.

DISCUSSION

The initial formation of the embryonic testis relies upon specific interactions between germ cells, pre-Sertoli cells, endothelial cells, and interstitial cells. While germ cells, pre-Sertoli cells, and endothelial cells are well-characterized by multiple unique markers, the origin and heterogeneity of the early interstitial population is much less well understood, and distinct populations have not been identified previously. Here we identify distinct populations arising from both the coelomic epithelium and the mesonephric border, including Leydig cells, non-Leydig progenitors, and perivascular cells.

As differentiated Leydig cells rarely proliferate (Orth, 1982), they are believed to arise from a precursor population. We show here that Leydig cells arise from two progenitor populations in the fetal testis: the coelomic epithelium and vascular-associated cells in the gonad-mesonephric border region. LHX9 marks both of these undifferentiated cell populations in the early gonad (Mazaud et al., 2002). We show that LHX9 expression is retained in a subset of vascular-associated cells near the gonad-mesonephros border that do not express the gonad somatic markers GATA4 and SF1 (Fig. 4 and Supplementary Fig. 3). In the adult rat testis, spindle-shaped cells closely associated with the vasculature proliferate after diethylstilbestrol (DES) treatment, which depletes adult testes of differentiated Leydig cells. These perivascular cells have been proposed to be the adult Leydig progenitors (Davidoff et al., 2004; Ge et al., 2006). Further experiments will be required to determine whether the adult Leydig population arises from LHX9 perivascular progenitors that originate in fetal life as described in this study.

Molecular heterogeneity of the early interstitial cell compartment

Research on fetal testis development has mostly focused on Sertoli cells, given their central role in sex determination and multiple Sertoli-specific molecular markers that are available. Due to a lack of molecular tools, the interstitial/stromal compartment of the nascent gonad is less well understood. A recent study sought to obtain a specific marker for peritubular myoid cells, and discovered that candidate genes were all expressed ubiquitously in the interstitium (Jeanes et al., 2005), suggesting that the early interstitial cells are a homogeneous population. In this study, we have identified heterogeneous cell types in the interstitium, based on expression of the transcription factors MAFB and C-MAF and the cell-surface protein VCAM1. These markers may prove useful tools to further our understanding of cells in the nascent gonad that influence gonad morphogenesis and give rise to critical lineages, such as Leydig cells.

MAFB is expressed in the coelomic domain of the testis at the earliest stages of interstitial development (E11.5). Its domain of expression expands, and then becomes restricted to Leydig cells. C-MAF expression also progressively expands in the interstitium with a delay relative to MAFB, and then becomes restricted to non-Leydig cells. At E12.5-E13.0, virtually all interstitial cells transiently express both MAFB and C-MAF. Between E13.5-E14.5, expression of the Maf transcription factors segregates between MAFB-positive Leydig cells and C-MAF-positive non-Leydig cells. Similarly, in the pancreas MAFB is initially expressed in the neuroendocrine precursor lineage and MAFA and MAFB are later segregated to pancreatic beta and alpha cells, respectively (Nishimura et al., 2006).

It is not yet clear whether Mafs regulate cell adhesion molecules in a similar manner to their Drosophila homolog traffic jam (Li et al., 2003). We have shown MAFB expression overlaps with VCAM1, which is expressed in perivascular cells that migrate into the gonad and give rise to a subset of interstitial cells. VCAM1 and its binding partner, integrin alpha4beta1/VLA-4 are both expressed in male somatic gonad cells (Jaspers et al., 1995; Nef et al., 2005) and have been implicated in the condensation of mesenchyme in multiple organs (Jaspers et al., 1995). Given its expression pattern, VCAM1 and perivascular cells may also play a role in cell adhesion and cellular rearrangement in the interstitial mesenchyme during gonad morphogenesis.

Whereas MAFB labels both coelomic domain-derived cells and perivascular cells, VCAM1 is initially restricted to perivascular cells that arise from the gonad-mesonephros border and is not expressed in the coelomic domain. It is unclear what molecular differences may exist between these two cell types, as our experiments have shown that both cell types can give rise to Leydig cells. Given that VCAM1-positive cells also express Smaa-YFP (Supplementary Fig. 2), which later on in development is enriched in peritubular myoid cells (PMC) that encase testis cords (Cool et al., 2008), it is possible that some VCAM1-positive cells represent PMC precursors. However, during early stages of gonad development (prior to E14.5), Smaa-YFP is ubiquitously expressed in the interstitium (Cool et al., 2008), so the significance of early Smaa/VCAM1 co-expression is difficult to assess. We still have not discovered a marker that is specific to PMCs prior to the time at which Smaa becomes highly enriched in that cell lineage.

Cells in the interstitium arise from both the coelomic epithelium and the mesonephros

Previous work from our laboratory demonstrated that daughter cells of the coelomic epithelium give rise to the Sertoli cell lineage only during a narrow time window prior to E11.5, but give rise to non-Sertoli cells throughout all time points tested (Karl and Capel, 1998; Schmahl et al., 2000). In this study, we confirmed these results using a specific marker for the interstitial lineage, MAFB. Previously, we (and others) reported that non-endothelial cell types of mesonephric origin were rarely detected in organ recombination assays (Combes et al., 2009; Cool et al., 2008; Martineau et al., 1997). However, based on images of early endothelial sprouts into the gonad co-stained with antibodies against MAFB or VCAM1, we hypothesized that the ability to detect migration of this cell type might be impaired by the dissection procedure. GFP recombination cultures that leave these vascular sprouts and perivascular cells intact reveal significant migration of cells from this border region along the endothelium throughout the gonad. Time-lapse live imaging of Mafb-GFP gonads support this conclusion, as cells from the gonad-mesonephros border region move along vessels and enter the coelomic domain. Further experiments to clarify the source of the attractive signal and the nature of this migration (e.g., if it only occurs along vascular tracks), are needed to understand this morphogenetic process.

One point of debate in the field of testis and ovary organogenesis has been the origin of steroidogenic cells. Hypotheses for Leydig cell origin have included the coelomic epithelium and the mesonephros (Buehr et al., 1993; Karl and Capel, 1998; Merchant-Larios and Moreno-Mendoza, 1998; Nishino et al., 2001; Schmahl et al., 2000; Val et al., 2006). In this study, we have used MitoTracker to label the coelomic epithelium and follow its fate during the time window of testis morphogenesis. We observe that the daughter cells of the coelomic epithelial divisions give rise to 3β-HSD-positive Leydig cells. However, we have also observed that cells from the gonad-mesonephros border migrate into the coelomic domain of the gonad and give rise to an additional subset of Leydig cells. These results are interesting in light of a recent study that characterized a group of adrenal-like cells in the interstitium of the testis that are derived from the gonad-mesonephros border region (Val et al., 2006). It remains to be seen whether Leydig cells derived from different cellular sources vary in their steroidogenic or other cellular properties.

Interdependent relationships between vasculature, Sertoli cells, and multiple interstitial lineages are required for testis morphogenesis

The formation of testis architecture from a bipotential, undifferentiated gonad primordium requires an orchestrated effort among many cell types. A previous view in the field is that the differentiation of the Sertoli cell is the driving force behind testis formation, but such a unidirectional perspective may not faithfully represent how the process takes place. The presence of the Sertoli cell is certainly a requirement for constructing the testis, given that the Sertoli cell is the scaffold for testis cord structure and also is an important signaling center, but it is not sufficient for creating testis architecture. Previous work from our lab and others has recently demonstrated that the action of VEGFA and vascular endothelial cells is required for Sertoli cells to organize into testis cords (Baltes-Breitwisch et al., 2010; Bott et al., 2010; Bott et al., 2006; Combes et al., 2009; Cool et al., in press). However, there are precedents in the literature for aspects of testis development that can proceed independently of Sertoli cell differentiation. A male-like coelomic vessel can form in a XX Wnt4 mutant gonad (Jeays-Ward et al., 2003), while steroidogenic Leydig-like cells can be induced in the ovary via exogenous FGF9 treatment or overexpression of the Hedgehog pathway (Barsoum et al., 2009; Schmahl et al., 2004). Interestingly, we have recently shown that Vegfa is expressed in the interstitial cell population (Cool et al., in press), further reinforcing the idea that multiple interdependent relationships are involved in testis morphogenesis.

In this study, we have revealed additional roles for vasculature and perivascular cells. Our results show that the expansion of the MAFB population in the testis is dependent on the presence of male-specific vasculature, in agreement with a previous report from our laboratory demonstrating a role for vasculature in promoting proliferation of the interstitial population (Cool et al., in press). Consistent with this data, we observe that perivascular cells proliferate rapidly between E11.5-E12.5, as many VCAM1-expressing cells in the gonad are BrdU-positive (Supplementary Fig. 5). Our data also shows that the vasculature is required for VCAM1-positive interstitial cells to migrate into the testis and to contribute to Leydig cells, and perhaps also to Leydig progenitors and other interstitial cell types. This suggests that blood vessels serve an instructive role for both proliferation and migration of interstitial cells in the nascent testis.

This system shows many parallels with other organ systems that require mesenchymal-vascular or epithelial-vascular interactions during morphogenesis, such as lung, adipose tissue, bone, pancreas, liver, and neuronal stem cells (Del Moral et al., 2006; Fukumura et al., 2003; Gerber et al., 1999; Jacquemin et al., 2006; Lammert et al., 2001; Matsumoto et al., 2001; Shen et al., 2004). In these cases, disruption of endothelial cells results in either a failure of gene expression or cellular rearrangements necessary for organ structure and function. In the case of endochondral bone formation, endothelial cell invasion into the growth plate is associated with recruitment of chondroclasts and osteoblasts which are necessary for apoptosis of chondrocytes, extracellular matrix degradation, and subsequent ossification (Gerber et al., 1999). However, in most instances, it is not known whether vascular invasion serves as a conduit for diverse cell populations (or vice versa) or how vascular-associated cells play an active role during organ morphogenesis.

It seems clear that the role of the interstitial cell population has been underappreciated in the field of gonad development. How this diverse mesenchymal population interacts with the vasculature and with other cell populations during testis morphogenesis is a topic for future studies with broad implications for organogenesis in other systems.

CONCLUSIONS

This work represents a significant advance toward understanding the origins of the various lineages in the interstitium, and positions the field to clarify the unique roles of these multiple cell types. We have shown that steroidogenic Leydig cells are derived from proliferation of the coelomic epithelium and also from specialized, undifferentiated perivascular cells in the gonad-mesonephros border region. Our results also reveal that, in contrast to previous reports, interstitial progenitor cells migrate from the mesonephric border region into the gonad, along with endothelial cells that originate in the mesonephros. We have characterized a new role for the vasculature in forming a conduit for mesenchymal cells to migrate into the gonad. Our data conclusively shows that the early interstitium is a heterogeneous cell population. Interdependent relationships among multiple, diverse cell types may be critical to organogenesis.

Supplementary Material

01. Supplementary Figure 1. MafA is expressed in macrophages in the embryonic gonad.

Immunofluorescent images of embryonic gonads. Color of each marker is indicated. Thick dashed lines in A and B indicate gonad-mesonephros border or boxed area of magnification. A’ is a higher-magnification image of the boxed region in A. Thin dashed lines in A’ and C denote testis cord boundaries. MAFA is expressed in a small subset of cells in the E13.5 testis (A) and ovary (B). Co-labeling of MAFA with Sox9-CFP and an anti-CD11b antibody (a macrophage marker) shows that MAFA-positive macrophages are present in E13.5 XY (C) gonads (arrowhead), although MAFA expression is also detected in other non-Sertoli cells within testis cords (arrow). Scale bar in A represents 100 µm in A and B, 29 µm in A’, and 17 µm in C.

02. Supplementary Figure 2. Maf-expressing perivascular cells also express Smaa-YFP and VCAM1.

Immunofluorescent images of E11.8 (21 ts) fetal XY (top) and XX (bottom) gonads.The color of each marker is indicated. Dashed lines indicate gonad-mesonephros border. Arrowheads highlight examples of Maf-expressing cells that express other perivascular markers. Scale bar in top left panel represents 50 µm in all panels.

03. Supplementary Figure 3. Gonad-mesonephros border cells represent a unique, undifferentiated gonadal cell population.

Immunofluorescent images of E11.5 (18 ts) XY gonads. The color of each marker is indicated. Dashed lines denote gonad-mesonephros border or boxed area of magnification. A’ and B’ are higher-magnification images of the boxed regions in A and B. MAFB-positive (A) and VCAM1 -positive (B) border region cells do not express the gonadal somatic markers GATA4 or SF1 (arrowheads in A’ and B’). Scale bar in A represents 50 µm in A and B, and 17 µm in A’ and B’.

04. Supplementary Figure 4. Vascularization of XX gonads by FGF9 treatment induces Leydig cell differentiation.

Immunofluorescent images of E11.5 XX (A, B, D, E) or XY (C, F) gonads cultured for 48 hours in the presence of an FGF9-coated (B, E) or BSA-coated control (A, C, D, F) bead. The color of each marker is indicated. Dashed lines denote gonad-mesonephros border or location of bead relative to gonad. XX control samples do not exhibit any 3β-HSD or SOX9 expression (A, D), while XX FGF9-treated samples show 3β-HSD and SOX9 staining near the bead (B, E). Arrowhead in B points to ectopic Leydig cells observed at a location distant from the bead. Germ cells, which are less efficiently labeled by PECAM1, are present, but not visible in B because PECAM1 labeling of the recruited vasculature is too high. XY control samples show extensive 3β-HSD and SOX9 staining (B, E). Scale bar in A represents 50 µm in all panels.

05. Supplementary Figure 5. Vascularization is required for migration and expansion of the VCAM1-positive population.

Immunofluorescent images of E11.5 (17–18 ts) XY (A, B) and XX (C, D) gonads cultured for 24 hours +/− Aflibercept. The color of each marker is indicated. Dashed lines denote gonad-mesonephros border or boxed area of magnification. E’ and F’ are higher magnification images of the boxed regions in E and F. While control XY gonads possess a large VCAM1-positive coelomic domain (A, yellow bracket), Aflibercept-treated XY gonads show a dramatic decrease in size of the coelomic domain with few VCAM1-expressing cells (compare bracket in A and B). BrdU-labeling of E12.0 XY (E, E’) and XX (F, F’) gonads shows that many interstitial VCAM1-positive cells (arrowheads) are BrdU-positive. Scale bar in A represents 50 µm in A–E and F, and 23 µm in E’ and F’.