Vascular-mesenchymal cross-talk through Vegf and Pdgf drives organ patterning

Abstract

The initiation of de novo testis cord organization in the fetal gonad is poorly understood. Endothelial cell migration into XY gonads initiates testis morphogenesis. However, neither the signals that regulate vascularization of the gonad nor the mechanisms through which vessels affect tissue morphogenesis are known. Here, we show that Vegf signaling is required for gonad vascularization and cord morphogenesis. We establish that interstitial cells express Vegfa and respond, by proliferation, to endothelial migration. In the absence of vasculature, four-dimensional imaging of whole organs revealed that interstitial proliferation is reduced and prevents formation of wedge-like structures that partition the gonad into cord-forming domains. Antagonizing vessel maturation also reduced proliferation. However, proliferation of mesenchymal cells was rescued by the addition of PDGF-BB. These results suggest a pathway that integrates initiation of vascular development and testis cord morphogenesis, and lead to a model in which undifferentiated mesenchyme recruits blood vessels, proliferates in response, and performs a primary function in the morphogenesis and patterning of the developing organ.

Endothelium-derived signals are required for the development and maintenance of many vertebrate organs. Information about vascular influences on organ budding, tissue-specific cell-type specification, and generation of progenitor niches have come from seminal work on the liver, pancreas, and nervous system (1–3). Despite the broad implications of this research, the mechanisms through which endothelial cells influence tissues have been difficult to identify. In several organs, specialized progenitor cells associate with the vasculature and maintain their proliferative status through contact with the ECM that shrouds vessels (4). However, the dynamics and cellular response of less specialized cells that mediate organ morphogenesis are not understood with the same clarity.

Endothelial cells influence testis cord morphogenesis in the embryonic mouse gonad (5, 6). The gonad is a uniquely powerful model in which to study the role of vasculature during organ morphogenesis because of the ability to culture and image whole organs during the coincident process of vascularization and epithelial morphogenesis. In the lung, the endothelium is reported to interact with the airway epithelium to induce septae formation in the distal airways (7). However, the potential importance of the mesenchyme was not investigated in that, despite the extensive literature supporting mesenchymal–epithelial interactions as a primary force during lung morphogenesis. In the gonad, Sertoli cells are currently assumed to attract migrating endothelial cells and initiate the hallmark patterning of testis cords, although there is no direct evidence that this is the case.

Here we report that cross-talk between the endothelium and nonspecialized mesenchymal cells drives testis morphogenesis. Vegfa is expressed specifically by the undifferentiated mesenchyme. Neutralizing antibodies against Vegf reveal a requirement during the initial steps of testis vascularization. By using real-time imaging of whole organs, we demonstrate that the primary cell type affected by endothelial migration is the mesenchyme itself. In the absence of vasculature, interstitial proliferation is reduced and wedge-like structures of mesenchyme that partition the gonad into cord-forming domains do not form. We propose that the endothelium does not directly regulate epithelialization, but promotes mesenchyme aggregation as a primary morphogenetic force. When the endothelial cell adhesion molecule vascular endothelial (VE)-cadherin was blocked with BV13, a less severe effect on mesenchymal proliferation was observed. However, mesenchymal proliferation was rescued by the addition of PDGF-BB to XY gonads treated with VEGF Trap or BV13. This leads to a model in which undifferentiated mesenchyme recruits blood vessels, proliferates in response, and performs a primary function in the morphogenesis and patterning of the developing organ.

Results

Vegfa and Its Receptors Are Expressed in the Gonad.

Endothelial migration into XY gonads begins by embryonic day (E) 11.5. Major vascular remodeling of the XY circulation occurs at approximately E12.0 and continues during the next 12 to 24 h (8). Expression of Vegfa was visualized using Vegfa-lacz mice at E12.0 (Fig. 1 A and B). LacZ was widely expressed throughout XX and XY gonads, but with subtle differences (Fig. 1 A′ and B′). In XX organs, Vegfa was expressed throughout much of the gonad and parts of the mesonephros. Interestingly, expression was absent in the coelomic domain (Fig. 1B′, brackets) and enriched along the gonad–mesonephros border (Fig. 1B′, arrowhead). In XY gonads, Vegfa was expressed strongly in the coelomic domain, but appeared at low levels in cells along the mesonephric border (Fig. 1A′, brackets, arrowhead). Domains of Vegfa expression are consistent with the sexually dimorphic vascular stability along the mesonephric border. In XX organs, this vascular bed remains intact, whereas the same vessels in XY urogenital ridges dissociate, giving rise to individual endothelial cells that migrate to the coelomic surface of the gonad (9).

Fig. 1.

Vegfa and its receptors are expressed in XX and XY gonads. Whole-mount E12.0 Vegfa-lacZ gonads stained with X-gal (blue). (A) X-gal staining of XY E12.0 gonads shows broad expression throughout the gonad including the coelomic domain. Expression of Vegfa extends to the surface of the gonad (A′, brackets) and is reduced at the gonad/mesonephros border (A′, arrowhead). (B) XX gonads also express Vegfa, although expression is absent from the coelomic domain (B′, brackets) and is enriched along the gonad/mesonephros border (B′, arrowhead). (C) VEGFA receptor expression was examined relative to PECAM-1, which labels both germ and endothelial cells in the gonad. Both NRP1 (D) and Flk1-mCherry (E) localize specifically to PECAM-1–positive endothelial cells (F). (G) Schematic representation of the early gonad shows the expression domain of various transgenic reporter lines expressed by gonadal subpopulations at E12.5. Key regions of the XY gonad including the coelomic domain (CD, brackets) and coelomic vessel (CV, arrow) are also indicated. (H) mRNA was extracted from FACS-sorted populations and expression of Vegfa was compared across XY cell types after normalizing to expression in whole E12.5 XY gonads. Bar colors represent analysis of specific populations as indicated in the schematic (G). Pink bar indicates Vegfa expression levels of whole XX gonads relative to whole XY gonads (*P < 0.05, **P < 0.005, ***P < 0.0005).

VEGFA is a secreted ligand and signals primarily through three receptor tyrosine kinases: VEGFR1 (Flt-1), VEGFR2 (Flk1), and NRP1. Of these receptors, FLK1 is the most critical for activation of downstream signaling, and mutation of this receptor prevents endothelial specification and patterning by VEGFA (10). In the E12.5 XY gonad, the vascular marker CD31 (PECAM-1) labels both endothelial cells and the germ line (Fig. 1C). These cell types are easily distinguished based on their morphology and localization within the gonad. Antibodies against NRP1 and a Flk1-mCherry reporter line reveal robust expression of both of these receptors in the microvasculature of the gonad (Fig. 1 D and E). Flk1 expression and NRP1 staining colocalized with PECAM-1 specifically on cells comprising the microvasculature of the gonad, and in particular the large male-specific coelomic vessel (Fig. 1F). We never found Sertoli or germ cells that were positive for NRP1 or Flk1. Our findings are consistent with widespread expression of these receptors on endothelial cells.

Vegfa Expression Is Differentially Regulated in XY Gonads.

VEGFA was previously reported in Sertoli cell cytoplasm with only faint expression in germ and interstitial cells (11). However, Vegfa-lacZ did not appear to be enriched in testis cords based on whole-mount staining (Fig. 1A). To determine which cells expressed Vegfa, we isolated individual populations of cells from the embryonic testis (Fig. 1G). By E12.5, the XY gonad is segregated into two major cellular compartments: the testis cords, containing Sertoli cells and germ cells; and the interstitium, comprising a heterogeneous population of mesenchyme that surrounds endothelial cells. By using transgenic mice expressing specific fluorescent tags for each cell type and FACS, we isolated Sox9-ECFP–positive Sertoli cells, Oct4-EGFP–positive germ cells, αSma-EYFP–positive interstitial cells, and Flk1-mCherry–positive endothelial cells (Fig. 1G). RNA was extracted from each positive fraction, and expression of Vegfa was measured by using quantitative RT-PCR (qRT-PCR) normalized to whole XY gonad cDNA to identify gonadal populations enriched for Vegfa expression. Surprisingly, at E12.5, Vegfa was not detected in Sertoli cells (Sox9-ECFP–positive). Instead, we found that interstitial cells (αSma-EYFP–positive) were enriched for Vegfa. Comparison of XX and XY whole gonads also revealed an approximately twofold enrichment of Vegfa transcripts in males (Fig. 1H).

Posttranscriptional modifications of Vegfa transcripts constitute an additional possibility for sex-specific regulation. In the Vegfa-lacZ reporter line, lacZ was inserted into the 3′ UTR of Vegfa and does not provide information about the numerous posttranscriptional Vegfa splice variants. Previous studies examined posttranscriptional regulation of Vegfa in the gonad but never compared XX and XY gonads for sexually dimorphic isoforms (11). To determine whether sex-specific isoforms of Vegfa are present in XX versus XY gonads, we used nested RT-PCR. Predominant isoforms of Vegfa-164 and Vegfa-120 were detected in both sexes, although a rare variant, Vegfa-144, was specific to XY gonads at all time points between E11.5 and E13.5 (Fig. S1A, white arrowhead, E12.5 shown). Consistent with qRT-PCR results, all isoforms of Vegfa were detected in αSma-EYFP–positive interstitial cells and not Sox9-ECFP–positive cells (Fig. S1B). The 144-kDa isoform is thought to behave similarly to VEGFA-164 as a result of its intermediate affinity for ECM and potent signaling interactions with FLK1. Expression analysis of Vegfa suggested that expression domain variation, expression level differences, and sex-specific splicing may all contribute to the differential regulation of vascular recruitment between the XX and XY gonad.

Inhibiting VEGF Blocks Sex-Specific Vascular Migration.

Traditional genetic ablation of Vegfa is complicated by the severe, systemic, and early phenotypes resulting from the loss of this growth factor (12). To overcome these obstacles, we took advantage of a well characterized and pharmacologically specific intervention, VEGF Trap (aflibercept; Regeneron), which contains critical domains of VEGFR1 and VEGFR2 that bind and inhibit secreted VEGF (13). Importantly, VEGF Trap does not broadly affect RTK signaling, and distinguishes the role of VEGF from FGF and PDGF, which also have critical functions during sex determination and the initiation of testis development.

To deliver VEGF Trap to the gonad, we developed a method to introduce inhibitors into the systemic circulation of the embryo and then explanted the organ for prolonged culture (Fig. 2A and SI Materials and Methods). The success of each injection was monitored in individual organs by coinjecting fluorescent lectins and imaging immediately after dissection (Fig. 2A′). Strong fluorescence in the vasculature at the border of the gonad and mesonephros suggested that drug delivery was efficient. Control organs were uninjected or injected with fluorescent dye without VEGF Trap. Coelomic vessel formation and endothelial migration were robust in control organs injected with lectin alone (Fig. 2B, arrowhead).

Fig. 2.

Inhibition of VEGF blocks vascular remodeling in the gonad but not male-specific lineage specification. (A) VEGF Trap and rhodamine lectins were delivered to the gonad by injection into the embryonic heart. (A′) Injected gonads were explanted to culture and the delivery of fluorescent lectins was evaluated for each sample. Dotted line indicates gonad/mesonephric boundary in A′ and B–D. (B) After 24 to 36 h of culture, control gonads, injected only with rhodamine lectins, showed normal development of a coelomic vessel (arrowhead) and germ cell aggregation inside testis cords. (C and D) In XY organs injected with VEGF Trap, male-specific vasculature was very limited or absent. (C) The presence of some endothelial cells in injected samples correlated with sporadic cord-like structures (arrowhead/dotted lines). (D) Robust delivery of VEGF Trap completely blocked male-specific vascular development and testis morphogenesis. (E–J) Markers of distinct gonadal cell types were specified but mislocalized after vascular inhibition (Right) compared with controls injected with lectins alone (Left). (E–J) Dotted lines define the surface epithelium and bars indicate the coelomic domain in I and J. (E and F) SOX9-positive (red) and AMH-positive (green) Sertoli cells did not aggregate into testis cords. (G and H) 3β-HSD–positive Leydig cell (red) localization was severely disrupted after Vegf inhibition. (G′) In controls, Leydig cells (red) are in close proximity to PECAM-1–positive (green) endothelial cells (arrow). (H′) Injection of VEGF Trap randomized Leydig localization. (I) αSMA-EYFP–positive (green) interstitial protrusions typically extend into the gonad at regular intervals and surround testis cords. (J) Inhibiting VEGF blocked extension of interstitial protrusions between testis cords.

Injection with VEGF Trap followed by in vitro culture for 24 to 36 h resulted in a strong blockade of male-specific vascular development in most gonads (Fig. 2 C and D). When all vasculature was eliminated from XY gonads, there were no regions of germ cell aggregation and male morphogenesis failed completely (Fig. 2D). A limited number of PECAM-1–positive endothelial cells were observed in some XY gonads after injection with VEGF Trap (Fig. 2C, arrowhead). Interestingly, the sporadic presence of these endothelial cells correlated with limited regions of germ cell aggregation and presumptive testis cord formation (Fig. 2C, dotted lines). Together these data conclusively demonstrate the requirement for endothelial migration into the XY gonad and the tight relationship between endothelial cells and male-specific morphogenesis.

Endothelial Migration Does Not Control Sertoli or Leydig Cell Specification, but Patterns Testis Cord Morphogenesis.

The failure of cord formation after VEGF Trap injection suggested that vascular migration may be required for promoting development of Sertoli cells, which are believed to be the primary cell type regulating cord formation. To address this possibility, we first compared the differentiation of Sertoli cells before and after blocking the vasculature. After injections with lectin alone, control organs developed normal testis cords containing SOX9 and AMH-positive Sertoli cells within, surrounded by unlabeled interstitial space. Injection of VEGF Trap by E11.25 resulted in robust inhibition of cord formation. However, AMH/SOX9-positive Sertoli cells were specified in the absence of the vasculature (Fig. 2 E and F).

The male-specific Leydig cell lineage (3β-HSD positive) was also specified in the absence of the vasculature, although these cells were found throughout XY gonads instead of concentrated in a defined interstitial space (Fig. 2 G and H). Fetal Leydig cells are typically located in close proximity to the vasculature, as seen in control XY gonads (Fig. 2G′, arrowhead). Interestingly, both Sertoli and interstitial progenitor cells arise from early (i.e., before E11.5) divisions of steroidogenic factor 1-positive cells in the coelomic epithelium (14). Subsequently (after E11.5), the coelomic domain gives rise to a population of steroidogenic factor 1-negative somatic cells, which are uncharacterized and have not been suggested to play a primary role during early morphogenesis of the testis. Although blocking vascular development did not inhibit Sertoli or Leydig cell specification, the coelomic domain of the gonad (Fig. 2 E–J, below dotted lines), which typically consists of several layers of mesenchymal cells above the condensing Sertoli cells (Fig. 1G, brackets), was significantly reduced in injected samples (Fig. 2 F, H, and J).

Vascular Migration Is Required for Interstitial Expansion.

The reduced size of the coelomic domain and the failure of testis cords to form in the absence of blood vessels, indicated that a primary function of the vasculature was to promote somatic development and patterning of the gonad. To investigate the dynamics of the interstitial population more carefully, we injected VEGF Trap or lectin alone into αSma-EYFP–positive embryos, which express EYFP throughout the interstitial mesenchyme. Gonads from embryos that received control injections developed a clear interstitial space marked by the dense accumulation of EYFP-positive cells and a thick layer of somatic cells surrounding the coelomic vessel (Fig. 2I, bar). However, in VEGF Trap-injected samples, interstitial cells were severely reduced and did not coalesce into distinct compartments between testis cords. Consistent with our previous observations, EYFP-positive cells were mostly restricted to the shrunken coelomic domain (Fig. 2J, bar).

To further characterize somatic cell development in the gonad, four-dimensional confocal microscopy was used to visualize how αSma-EYFP–positive interstitial cells respond in both the presence and absence of VEGF-mediated vascular migration. At E11.5, cells throughout the coelomic epithelium express heterogeneous levels of αSma-EYFP in both XX and XY gonads. Live imaging of XY αSma-EYFP gonads showed that interstitial cells rapidly expand between E11.5 and E12.5 (Fig. 3 A–D and Movie S1). After approximately 6 h of culture, αSma-EYFP is expressed by several layers of cells in the coelomic domain (Fig. 3A). Between 12 and 18 h after initiation of the cultures, imaging revealed that the interstitium extends throughout the gonad and a clear distinction between testis cords and the interstitial space can be made (Fig. 3 B and C). Live imaging showed that during this time period, cells initiate EYFP expression as well as migrate into the interior of the gonad from the coelomic domain (Movie S1). Testis cord domains (Fig. 3D, asterisk) occupy dark areas in stark contrast to the brightly labeled αSma-EYFP–positive interstitium after 24 h of culture.

Fig. 3.

Interstitial expansion fails after VEGF block. (A–D) αSma-EYFP–positive gonads were imaged in real time to characterize interstitial dynamics. (E–H) Littermates were injected with VEGF Trap and imaged in parallel. (A and B) After 6 to 12 h of culture, control organs display increased αSma-EYFP expression (Movie S1). (C and D) Within 18 to 24 h, controls develop clear testis cord domains that are segregated from the interstitium (asterisk). (E–H) Cord structures never appear and αSma-EYFP–positive cells do not expand into the interior of the gonad in the absence of robust vascular remodeling (Movie S2).

VEGF Trap injection into αSma-EYFP XY embryos revealed a severe defect in the expansion of the interstitium (Fig. 3 E–H and Movie S2). At the outset of culture, treated and untreated gonads appear nearly identical (Fig. 3 A and E). Expression of EYFP in the coelomic domain of both organs is abundant before the initiation of testis cord morphogenesis. Twelve hours after injection with VEGF Trap, it is evident that expansion of the interstitium is defective compared with controls (Fig. 3 B and F). Within the 24-h period of culture, in the absence of vasculature, interstitial cells expand only modestly and do not penetrate into the gonad (compare Fig. 3H with Fig. 3D; Movies S1 and S2). To clarify the dynamics of endothelial and mesenchymal cell interactions during organ patterning, we performed live imaging on embryos positive for Flk1-mCherry and αSma-EYFP (Movie S3). This revealed that the presence of endothelial cells preceded expansion of the mesenchyme into the interior of the testis from both the surface and gonad/mesonephros border.

Endothelial Migration Is Required for Male-Specific Proliferation.

Movies S1 and S3 indicate that expansion and penetration of αSma-EYFP–positive cells from the coelomic domain into the gonad occurred in part by migration away from the coelomic domain. However, another likely explanation for the failure of the expansion of the coelomic domain is that interstitial cells failed to proliferate in the absence of the vasculature. Proliferation is required for morphogenesis of XY gonads, although the direct regulatory mechanisms are not well understood (15).

To compare active proliferation of gonadal cells in the coelomic domain in the presence or absence of the coelomic vasculature (Fig. 4 A and B, boxed areas), we stained with antibodies against phosphorylated histone H3 (pHH3). Confocal Z-stacks of individual organs were condensed into maximum intensity projections and dividing cells within the indicated domain were manually counted. This eliminated variation in focal plane and assured that the observed decrease in proliferation was reflective of the entire coelomic domain. Proliferation in XY gonads is enriched in somatic cells throughout the coelomic domain after 24 h of culture (approximately E12.5; Fig. 4A). In addition to divisions in the coelomic epithelium, somatic cells throughout the domain surrounding the coelomic vessel are highly proliferative (Fig. 4A, arrowheads). These results are consistent with previously published data carefully characterizing sex-specific proliferation patterns (14). In the absence of endothelial cells resulting from VEGF Trap injection, fewer cells were dividing throughout XY gonads based on pHH3 staining (Fig. 4B).

Fig. 4.

Vascular migration is necessary and sufficient for interstitial proliferation. (A–G) Male-specific proliferation was quantified by counting pHH3-positive dividing cells (green) in control and cases in which vasculature is blocked or misregulated. (A–C) Gonads at E11.25 to E11.5 were cultured for 24 h. Endothelial cells were visualized by using PECAM-1 (red). (A and B) Proliferation in the coelomic domain (CD; boxes designated with dotted lines) was compared between WT and VEGF Trap-injected embryos. (A) In WT XY gonads, proliferation is abundant and up-regulated in somatic cells within the epithelium and surrounding vessels (arrowheads). (B) Blocking vascular development inhibits male-specific somatic cell proliferation in the CD. (C) BV13-treated gonads develop disorganized vasculature with reduced proliferation in the CD. (D) Quantification of pHH3-positive cells showed reduced proliferation after VEGF Trap injection and BV13 treatment. (E–G) To determine if vasculature was sufficient to induce proliferation, XX Wnt4^−/− mutants were compared with littermate controls. Blue nuclei are false colored, and indicate PECAM-1/pHH3 double-positive proliferating germ cells not included in this analysis. In XY Wnt4^+/− gonads (E), proliferation is higher than in XX Wnt4^+/− littermates (F). (G) However, XX Wnt4^−/− gonads develop male-specific vasculature and exhibit proliferation levels similar to XY littermates and significantly higher than XX gonads, suggesting the presence of vasculature is sufficient to drive proliferation. Quantification of these results is presented in D (***P < 0.0005).

Endothelial cells migrate into the XY gonad as individual cells, then coalesce to form the coherent coelomic vessel (9). In previous reports, disruption of endothelial adhesion disrupted cord formation (6). However, the mechanism mediating the effect on morphogenesis was unclear. We investigated whether endothelial cell adhesion and vessel maturation are required to stimulate proliferation of mesenchymal cells. Blocking endothelial cell adhesion by treating E11.5 cultured gonads with an antibody against VE-cadherin (BV13) resulted in a significant decrease in proliferation in the coelomic domain after 24 h of culture, suggesting that vessel coalescence is required after migration of individual endothelial cells, but before the induction of interstitial proliferation (Fig. 4C). Proliferation was quantified after VEGF Trap injection (n = 13; P = 0.0001) and BV13 treatment (n = 13; P = 0.0005) to confirm that reduced proliferation caused by both treatments was significant (Fig. 4D).

Proliferation analysis after inhibition of vascular migration and disrupted endothelial adhesion demonstrated that both of these processes were critical to induce male-specific expansion of the coelomic domain. To test whether a male-specific coelomic vessel was sufficient to induce somatic proliferation, XX Wnt4^−/− gonads were analyzed. In XX Wnt4^−/− gonads, an ectopic vessel reminiscent of the male-specific coelomic vessel forms (16). Comparison of XY Wnt4^+/−, XX Wnt4^+/−, and XX Wnt4^−/− littermates revealed increased proliferation in the coelomic domain of XX Wnt4^−/− mutants with a particular increase in cells adjacent to the ectopic vasculature (Fig. 4 D–G). Proliferating germ cells found in the XX coelomic domain were identified as PECAM-1/pHH3–positive cells (i.e., blue cells) and were not counted in our analysis. XX Wnt4^−/− gonads had similar levels of somatic proliferation to XY Wnt4^+/− controls, which is significantly more than XX controls (n = 5; P = 0.001; Fig. 4 D–G). This result suggests that ectopic endothelial cell migration is sufficient to induce proliferation of nearby mesenchymal cells.

Vessel-Derived PDGF Promotes Proliferation.

XY Pdgfrα mutants resemble BV13-treated gonads: despite their ability to recruit endothelial cells to the coelomic domain, they show defects in the later stage of male-specific somatic proliferation (17). Moreover, expression analysis of XX Wnt4^−/− gonads showed up-regulation of several Pdgf pathway components in the presence of ectopic vasculature, although Sertoli cells do not persist (18). This suggested that Pdgfrα/β, which are both expressed in the interstitium, are activated by vascular migration in the absence of Sertoli cells. Mutation of Pdgfrß does not result in overt gonad phenotypes. It is possible that Pdgfrα compensates for the loss of Pdgfrß given coexpression on interstitial cells (17). Pdgfa is expressed by Sertoli cells and originally proposed to activate Pdgfrα and induce XY-specific proliferation (17). However, gonad morphogenesis is normal in Pdgfa^−/− mutants, suggesting the involvement of an alternate ligand (19).

We used qRT-PCR to measure changes in Pdgfa and Pdgfb expression in XY gonads after BV13 or VEGF Trap treatment. Under both conditions the expression of Pdgfa was unchanged compared with controls (Fig. 5A). Stable expression of Pdgfa after disrupted vascular development is consistent with our observation that Sertoli specification is unaffected, as shown by normal levels of Sox9. In contrast, we found that Pdgfb expression was significantly reduced after injection with VEGF Trap (Fig. 5A). Interestingly, treatment with BV13 resulted in a slight reduction in Pdgfb expression despite constant expression of the endothelial cell receptor Tie2. Although Fgf9 was another attractive candidate mitogen, given its well characterized role in promoting testis development (20, 21), we found that expression of Fgf9 was invariant in the presence of both inhibitors (Fig. 5A). Thus Fgf9 is unlikely to be mediating this effect. Using our panel of gonadal populations isolated by FACS, we confirmed that Pdgfa is expressed predominantly by Sertoli cells and Pdgfb is specific to endothelial cells (Fig. 1G and Fig. S2). This expression analysis suggests that Pdgfb expression reflects the presence of mature vessels in the gonad and correlates with the ability of endothelial cells to induce mesenchyme proliferation.

Fig. 5.

PDGF-BB rescues defective vascular development. (A) Expression of Pdgfb, but not Pdgfa or Fgf9, was reduced after VEGF Trap and BV13 treatment. (B–D) Reduced proliferation after VEGF Trap injection was rescued by addition of rPDGF-BB to cultures. All organs were stained with PECAM-1 (red), NRP1 (blue), and pHH3 (green) in order to assess vascular remodeling and proliferation. (B and C) Injected cultures showed a loss of proliferation relative to controls. (D) Proliferation was restored in injected cultures treated with rPDGF-BB. (E) Although addition of rPDGF-BB to WT gonads had no significant effect on proliferation, quantification of pHH3-positive cells confirmed that the significant decrease in proliferation after VEGF Trap injection, and to a lesser extent BV13, was rescued by the addition of rPDGF-BB (*P < 0.05, ***P < 0.0005).

Based on these results, we tested whether PDGF-BB could compensate for disrupted vasculature. We added recombinant PDGF-BB (rPDGF-BB) to VEGF Trap and BV13 cultures. Addition of rPDGF-BB to culture media increased the number of pHH3-positive cells in the coelomic domain after VEGF Trap treatment compared with organs cultured in unsupplemented media and restored proliferation to near WT levels (n = 6; P = 0.004; Fig. 5 B–E). Addition of rPDGF-BB to BV13 cultures resulted in a modest but significant increase in proliferation that was commensurate with the milder proliferative phenotype (n = 8; P = 0.01). Testis cord aggregation was not evident in gonads in which proliferation was rescued, which may reflect the absence of other factors associated with or induced by vessels. Live imaging of mesenchyme and endothelial cells captured the intimate relationship of these cell types as they collectively form the interstitial space (Movie S3). The mesenchyme may be unable to establish an interstitial compartment without the continued influence of the vasculature. These results clarify the context of Pdgf signaling in the fetal testis and also establish PDGF-BB as a critical signal downstream of vascular development during organ morphogenesis.

Discussion

Dimorphic development of the gonad is a paradigm for organ growth and patterning. Shortly after the bipotential window, the XY mouse gonad begins a remarkable 24 h in which male-specific vascular remodeling and de novo cord morphogenesis occur. Like other internal mammalian organs, the rapid developmental timeframes combined with inaccessibility of organs during morphogenesis have made the integration of concurrent events difficult. However, the inaccessibility and temporal limitations have been overcome in the gonad by the advent of whole-organ culture and simultaneous live imaging, which allows us to integrate cell behavior and molecular mechanisms in real time. We have used these tools to investigate vessel dynamics in an organ undergoing morphogenesis. Experiments perturbing vascular development demonstrate a link between Vegfa-guided endothelial migration and interstitial proliferation, which suggest an endothelial–mesenchymal feedback loop necessary for organ morphogenesis.

Endothelial cells have emerged as a critical influence on developing tissues. Endothelial–progenitor cell interactions are a mechanism by which vessels promote tissue development and maintenance. In the few characterized cases, vascular-derived ECM establishes a niche for progenitor cells (22). In the adult testis, several populations of specialized cells are closely associated with blood vessels, although mechanisms underlying putative functional relationships are unknown (23). In these cases, disruption of vessel–progenitor cell interactions prevented the differentiation of specific cell types, but did not affect gross morphology of the organs. Our data do not support a requirement for the vasculature during early specification of Sertoli or fetal Leydig cells, although we cannot preclude homeostatic relationships at later stages of development.

In addition to progenitor cell interactions, vessels have a less well characterized role during organ formation and recovery after injury. In endodermal organs, mature vessels influence surrounding pancreatic tissue by providing growth and survival signals (24). Although previous work established vessels as a source of mitogens, relating this to a mechanism that affects mammalian organogenesis and patterning has proven difficult.

The gonad is an ideal model system in which to investigate how the initial steps of vascularization occur in a developing organ. In this study, we show that vascular recruitment and interstitial proliferation are interdependent during tissue morphogenesis. We show that the Vegf pathway is a primary mediator of gonad vascular remodeling and acts upstream of VE-cadherin mediated vascular adhesion. We previously showed that PDGF receptor α (PDGFRα) activation is required for proliferation but not initial vascularization of the XY gonad (17). Here we provide evidence that Pdgf signaling lies downstream of the establishment of endothelial migration and adhesion during testis morphogenesis. Specifically, addition of rPDGF-BB rescued reduced proliferation after vascular perturbation, which suggests that this endothelial-derived signal plays a critical role during PDGFRα activation. The addition of recombinant growth factors can give misleading results. However, the similar phenotype of the Pdgfrα mutants, and the fact that levels of Pdgfb are altered in the absence of the vasculature, supports the hypothesis that endothelial cells promote proliferation by secreting PDGF-BB.

Although interactions between remodeling endothelial cells and PDGFR-positive mesenchyme are common in the cardiovascular system, they are not well understood as a primary morphogenetic force during tissue growth and patterning. Our results support a model in which endothelial cells and α-smooth muscle actin–positive mesenchyme maintain a dynamic relationship central to testis morphogenesis. Our results suggest that Vegf and Pdgf initiate a feed-forward relationship among endothelial cells and gonadal mesenchyme. However, it remains unclear what mechanisms coordinate ingression of the interstitium into the gonad. Two-color live imaging revealed that the mesenchyme expands along vessels to form physical “wedges” that divide the nascent tissue into testis cords and suggests that adhesion between the cell types is an important factor. However, these observations do not preclude the existence of a morphogen or secreted factor that may coordinate the dynamics of the various cell types. Clarifying mechanisms of endothelial–mesenchymal interaction and their cumulative effect on Sertoli epithelialization will be an important area of investigation moving forward.

The field of sex determination has revealed that Sertoli cells mediate the formation of testis cords. As in many tissues, the idea that a highly specific cell type orchestrates development is appealing. Our work promotes a different model, and suggests that the segregation of epithelial cells into testis cord structures is driven by changes in the mesenchymal population rather than by cell-autonomous forces within epithelial cells. The gonadal mesenchyme uniformly expresses many smooth muscle markers, including PDGFRα/β, α-smooth muscle actin, and various ECM constituents. We suggest that this cell type, common to most organs and tumors, plays a much more critical role during tissue development and repair than we have previously appreciated. Understanding the dynamic relationship between these cells and the vasculature will further our understanding of normal tissue growth and patterning in addition to identifying effective targets for control of vascularization during neoplastic growths.

Materials and Methods

Standard protocols were used for timed matings, tissue collection, β-gal staining, FACS sorting, heart injections, and immunocytochemistry. These and a description of mouse lines used are provided in SI Materials and Methods. All experiments were conducted in accordance with the principles and procedures outlined in the National Institutes of Health Guidelines for the Care and Use of Experimental Animals and approved by the Duke University Institutional Animal Care and Use Committee.

Live Imaging.

All imaging experiments were performed on a Zeiss LSM510 or LSM710 as previously described (9). Z-stacks were collected every 10 to 15 min. All movies are maximum intensity projections at each time point.

Organ Culture.

E11.5 genital ridges were cultured in 1.5% agar blocks at 37 °C with 5% CO₂/95% air. Organs were cultured in Dulbecco minimal Eagle medium supplemented with 10% FBS and 50 μg/mL ampicillin (25). BV13 (gift from Elisabetta Dejana, Milan University, Milan, Italy) was used at 24 to 48 μg/mL as previously described (6, 26). rPDGF-BB (R&D Biosystems) was added to media with a final concentration of 50 ng/mL as previously described (17).