Morphometric analysis of testis cord formation in Sox9-EGFP mice

Abstract

Sox9-EGFP knock-in mice were generated to label Sertoli cells and visualize testis cord formation during development. Confocal microscopy and morphometric analysis of developing cords were performed. Serial histological sections were used for three-dimensional cord reconstruction. Initially, gonad length decreased from embryonic day (E) 11.5 to E13.5, but increased thereafter, while gonad width doubled every 12 hours from E11.5 through E14.5. At E12.5, the average number of cords was 12.5, whereas this decreased to 10.4 at E13.5 and E14.5. Cord number at a given time point varied between gonads and influenced dimensions. The initial cords that formed were complex and branches were common. Time-lapse imaging revealed an intricate behavior of the Sertoli-germ cell mass and cellular exchange between connected neighboring cords. These results suggest that cord formation is a highly dynamic process that subsequently becomes refined to establish the final number of seminiferous tubule precursors.

Introduction

In the mouse, the gonad primordium arises as a thickening of the coelomic epithelium of the urogenital ridge at embryonic day (E) 10.5 (tail somite stage 8 [TS8]), which coincides with the arrival of the primordial germ cells (Karl and Capel, 1998). At E11.5 (TS18), undifferentiated, bipotential gonads that are morphologically identical in XX and XY embryos, contain a presumably unorganized mixture of germ cells and somatic cells. These gonads are in intimate contact with the mesonephros, the transient fetal organ in which the precursors to the male and female reproductive tract, as well as the urinary system, arise. The first event towards sex differentiation is activation of the sex-determining region of the Y chromosome (Sry) gene in pre-Sertoli cells of XY gonads around E10.5 (Sinclair et al., 1990; Hacker et al., 1995; Koopman et al., 2001). This results in upregulation of Sox9 at E11.5 before Sry is downregulated at E12.5 (TS 25). Subsequently, differentiation of the somatic component and progression of the gonad toward the male developmental path occurs (See (Kanai et al., 2005) for review).

At least 5 different cell types, excluding connective tissue and germinal epithelium, contribute to the adult testis. These are Sertoli, germ, peritubular myoid (PTM), endothelial and Leydig cells, each with a very specific function and presumed origin. Sertoli cells are the main structural component of the testis cords in the developing male gonad. Differentiation of Sertoli cells at around E11.5 coincides with proliferation of Sox9-expressing cells (Schmahl et al., 2000; Schmahl and Capel, 2003), resulting in an overall size increase of the gonad. As differentiation proceeds, migration of mesenchymal cells from the neighboring mesonephros is triggered by unknown signals from Sertoli cells (Buehr et al., 1993; Martineau et al., 1997). This mesenchymal migration appears to be crucial for testis differentiation because blocking migration inhibits cord formation (Buehr et al., 1993; Tilmann and Capel, 1999; Combes et al., 2008). Interestingly, pairing gonads with mesenchymal tissue other than that from the mesonephros, such as limb bud mesenchyme, also leads to cord formation (Moreno-Mendoza et al., 1995). Previously, migrating mesenchymal cells were believed to be the origin of PTM cells (Buehr et al., 1993; Martineau et al., 1997). However, recent findings suggest that PTM cells originate from within the gonad (Cool et al., 2008), while endothelial cells migrate from the mesonephros (Combes et al., 2008). The origin of fetal Leydig cells remains an open question (Habert et al., 2001; Griswold and Behringer, 2009).

At E13.5, the testis cords are formed and aligned perpendicular to the length of the mesonephros, in which the Wolffian and Mullerian duct, as well as mesonephric tubules have developed. The cords consist of Sertoli cells that surround the germ cells and separate them from the mesenchymal interstitium. Sertoli cell cytoplasm extends toward the lumen to surround the germ cells to maintain a nurturing environment (Dym, 1977). This relationship between Sertoli and germ cells will persist throughout adulthood, when Sertoli cells play an essential role in spermatogenesis.

How the initial homogeneous mixture of somatic and germ cells in the fetal gonad remodel and become organized into the stereotypical cords of the testis are poorly understood. Few studies have employed live-imaging techniques to follow the dynamic changes that occur during testis differentiation and cord formation (Coveney et al., 2008). Knowledge of these morphometric changes over time will serve as a foundation for subsequent mechanistic studies of testis development.

SOX9 is a transcription factor that is upregulated by SRY in Sertoli cell precursors between E10.5 and E11.5 in the mouse and is essential for cord formation (Kent et al., 1996; Morais da Silva et al., 1996; Chaboissier et al., 2004; Barrionuevo et al., 2006). In addition, Sox9 is sufficient for testis development and production of fertile male mice in the absence of Sry (Vidal et al., 2001; Qin and Bishop, 2005). Therefore, SOX9 is useful marker for Sertoli cells.

We have generated a Sox9-Enhanced Green Fluorescent Protein (EGFP) knock-in mouse line. In the current study, we evaluated the expression of EGFP in the developing gonad to confirm its correlation with endogenous SOX9 expression. Subsequently, this line was used for morphometric analysis to characterize the three-dimensional (3D) structure of the gonad, specifically the Sertoli-germ cell mass as cords develop, and describe how it changes over time during testis differentiation. Serial sections supplied data for reconstruction of individual testis cords. These studies quantify the highly dynamic nature of testis cord formation.

Results

EGFP expression in Sertoli cells of embryonic and adult testes

Timed matings were established to document EGFP expression in the gonads of Sox9-EGFP knock-in mice. When urogenital ridges were dissected at E10.5, no EGFP expression was observed under fluorescence microscopy in male or female gonads. At E11.5 (TS18), the gonads could be distinguished as a separate entity by a clear border between the mesonephros and the gonadal tissue that appeared lighter in color and more uniform in texture under bright field stereomicroscopy. EGFP expression, however, was absent or very low. In gonads where EGFP was expressed, it was primarily in the center area of the anterior-posterior axis of the gonad, fading towards the poles and often in small patches along the length. EGFP rapidly appeared and increased thereafter (as observed in real-time live imaging organ culture, data not shown) and was always present at detectable levels at TS19 on E11.5 (Fig. 1A). At E12.0 (TS22) and later, EGFP expression was observed along the entire length of the gonad and was very robust (Fig. 1B, C). It remained high through E16.5 (Fig. 1D–F), and in adult seminiferous tubules (data not shown). Upon closer examination, the green fluorescing tissue had a granular appearance, with black dots representing the Sertoli cell nuclei and germ cells (Fig. 1). EGFP was never detected in embryonic female gonads, however in both sexes the area around the mesonephric tubules and the Wolffian and Mullarian ducts fluoresced.

FIG. 1.

Sox9-EGFP testes at different stages of development. Parameters for morphometric analysis at E11.5 (A), 12.5 (B) and 14.5 (C) are shown. The Sertoli and germ cell mass is green and is surrounded by a layer of EGFP-negative tissue. The edges of the gonads are indicated by the lines at the end of the arrows. At E15.5 (D) and 16.5 (E, F) coiling of elongated, single cords are visible (arrowheads). At E16.5 (E, F) coils are arranged in a wave-like pattern (visible from two different sides) and the early epididymis (D, E, arrows) and Wolffian duct (F, arrow) are fluorescing. The testis cords run parallel into the developing rete testis (A, B, oval) that connects to the epididymis. L = length; W = width; P = proximal cord diameter; D = distal cord diameter; R = rete testis length. Cord numbers are in yellow and each dot represent a single visible protrusion (B) or cord (C). Scale bar = 200 µm.

Immunostaining for SOX9 protein labeled nuclei of Sertoli cells in E13.5 (Fig. 2A, E) and adult (not shown) testes. EGFP fluorescence was preserved through histological processing and found in the Sertoli cell cytoplasm (Fig. 2B, F). The merged image indicated that the endogenous SOX9 protein and EGFP were co-localized in Sertoli cells, confirming that EGFP expression in the testicular cords marks Sertoli cells (Fig. 2D, H). Therefore, the Sox9-EGFP knock-in mouse line effectively labels Sertoli cells and can be used to visualize testis cords in the developing gonad.

FIG. 2.

Immunofluorescent analysis of Sox9-EGFP testis at E13.5. SOX9 immunostaining (red) marks the nuclei of Sertoli cells (A, E). EGFP (green) is visible in cell cytoplasms inside the testis cords (B, F). Cell nuclei are stained with DAPI (blue; C, G). Merged images shows that Sertoli cells co-express SOX9 and EGFP (arrowheads; D, H). Scale bar = 50 µm.

Morphometric dynamics of developing testes

No differences were found between left and right gonads for any of the variables (gonad length, width, rete testis length, cord diameter proximal and distal from the mesonephros, and the total number of cords) that were measured at any time point examined (p>0.05, data not shown). Therefore, left and right gonads were pooled for subsequent analyses. The parameters that were measured are indicated in Fig. 1. Results are presented in Fig. 3A. Gonad length decreased by 11% from E11.5 to E13.5 and then increased so that there was no difference between gonad length at E11.5 and E14.5. Gonad width increased by 100% from E11.5 through E12.5 and 76% from E12.5 to E13.5, and 38% E13.5 to E14.5. In total, the gonad width increased by 526 µm (335%) over the three days. The rete testis was only measured at E13.5 and E14.5, because defined loop-like cords were only established at E13.5 and therefore no rete testis was present at earlier time points. The rete testis became apparent at E13.5 once loops were fully formed and distinguishable. The length of the rete testis increased by was 59 µm (8%) between E13.5 and E14.5.

FIG. 3.

Measurements of gonad length, width, rete testis length and cord number count over time (A, B). Proximal cord diameter (PCD) and distal cord diameter (DCD) of E13.5 and E14.5 gonads are compared (C). Columns with different letters (a–d) are statistically different (P<0.05).

Cord numbers were counted as the number of cords proximal to the mesonephros, where cords entered the rete testis as observed in 2D images. At E12.5, the mean number of cords was 12.5 ± 3.6 (Fig. 3B). The mean number of cords was reduced to 10.4 ± 2.7 and 10.4 ± 2.1 at E13.5 and E14.5, respectively. Mean cord diameter was only measured at E13.5 and 14.5, when complete loop-like cords surrounded by interstitium were present. Proximal cord diameter (PCD) and distal cord diameter (DCD) increased 24% and 31%, respectively, over this 24-hour period (Fig. 3C). Furthermore, the testis cords increased in diameter from proximal to distal with respect to the mesonephros (30% at E13.5 and 37% at E14.5).

Stage-specific changes of Sertoli-germ cell mass in developing testes

A series of static 3D images and time-lapse movies were acquired, reconstructed and analyzed to characterize the changes of the gonadal tissue during the period of cord formation. At TS18, gonads were a simple, cylindrical shape connected along its longitudinal axis to the mesonephros. Sox9-EGFP expression was very faint (when visible) and detected along the length of the gonad, but initially was weaker towards the anterior and posterior poles and sometimes patchy.

The first images were taken at TS19, after bright fluorescence was present along the entire length of the gonad (Fig. 4A). The XY gonads had a very simple, seemingly indifferent morphology when compared with XX gonads and except for green fluorescence in early testes, the two genotypes were indistinguishable. The edges of the green Sertoli-germ cell mass (SGCM) had a relatively smooth edge both proximally and distally from the mesonephros (Fig. 4A).

FIG. 4.

Sox9-EGFP gonads at different stages of development, showing gross morphology at tail somite stages TS19 to TS32 and E14.5. Sertoli cells are labeled green. Morphological changes of the gonad begin with the appearance of small protrusions (arrows) and non-fluorescing areas (arrowhead) that persist and change through the developmental stages (B–E). Proximal and distal refers to the position in the gonad in relation to the mesonephros. Scale bar = 100 µm.

At TS21, the testis increased in thickness and the first signs of differentiation of the SGCM appeared, with the distal edge first obtaining an uneven, serrated appearance, followed by the formation of small protrusions of the SGCM and distinct, non-fluorescing areas forming inside the tissue mass (Fig. 4B, Movie 1A).

Another 9–12 h later, at TS24, distinct protrusions had formed on the distal side, with separations between protrusions reaching deep into the SGCM (Fig. 4C, Movie 1B). These protrusions were counted as testis cords during morphometric analyses, because they are the structures that will eventually give rise to testis cords. Large, defined, non-fluorescing areas were still visible within the SGCM, and the borders of these non-fluorescing areas appeared more distinct.

During the next 24 hours, from ~TS25-32 (~E13.5), protrusions significantly increased in length, while the non-fluorescing areas became larger and elongated, refining individual testis cords. It appeared as though these areas eventually fused with other interstitial tissue (Fig. 4D, Movie 1C). Thereafter, the cords became longer and increased in diameter. Individual cords had a loop-like structure and ran parallel to adjacent cords (Fig. 4E, Movie 1C). The proximal regions of the cords connected to a single cord-like structure that ran parallel to the mesonephros where the rete testis develops and was visible when isolated testes were cut longitudinally, laid flat, and imaged from both inside and outside (Fig. 5).

FIG. 5.

Isolated testes at E14.5 cut open longitudinally on the distal side through the cords and laid flat for imaging from the outside (A) and inside (C), and a whole gonad positioned so that the rete testis faced the objective (B). Ovals indicate the location where individual cords (dots) merge to form the single cord-like structure (arrow) that runs parallel to the mesonephros in the developing rete testis.

The first indication of cord coiling was noted at E15.5, when the loop-like cords have elongated further (based on the increase in overall gonad width) and started deviating from the straight, parallel arrangement distally from the mesonephros to create bends in the cord (Fig. 1D). At the proximal end, the cords entered the rete testis. This distal coiling increased and became more organized to form a seemingly coordinated wave-like pattern of neighboring cords at E16.5 (Fig. 1E, F), while the rete testis remained apparently unchanged (Fig. 1E). The Wolffian duct also expressed EGFP (Fig. 1F), especially in the forming caput epididymis that was visible as a highly coiled tubule immediately adjacent to the rete testis (Fig. 1E). In addition, the Wolffian duct appeared to be connected to the cord-like structure in the developing rete testis region at E15.5 and 16.5 because there was continuous fluorescence between these two structures (Fig. 1D, E).

Throughout differentiation, the SGCM was surrounded by a layer of unlabeled, non-EGFP expressing tissue of 4–6 cell layers thick, presumably the future tunica albuginea and developing interstitium (data not shown).

Number of cords affects testis morphometrics

For all three time points at which cords were counted (E12.5-E14.5), there was variation in cord number. To determine the effect of cord number on other variables at a single stage (E13.5 or E14.5), gonads were grouped into two classes: those with 10 or less cords and those with 11 or more cords. This was not done for E12.5, because the final number of cords at this stage are not established yet. Data is presented in Fig. 6. At E13.5, only testis width was significantly different between the two groups, with testis width being greater (564 ± 13.0 µm) when fewer cords are present than when 11 or more cords are present (533 ± 4.1 µm; P<0.05; Fig. 6A). At E14.5, however, testis length (1285 ± 11 µm), width (725 ± 12.2 µm) and PCD (93 ± 2.6 µm) was greater (P<0.05) in gonads with 10 or less cords when compared with those that had 11 or more cords (1192 ± 35, 638 ± 7.4, and 80 ± 2.6 µm, respectively, Fig. 6B).

FIG. 6.

Measurements of E13.5 (A) and E14.5 (B) gonads with 10 or less cords compared with those from gonads with 11 or more cords. Bars with different letters within variable groups are statistically different (p<0.05). PCD = proximal cord diameter, DCD = distal cord diameter.

Cord arrangements vary between testes

The arrangement of cords of E14.5 testes was studied among different gonads. Several interesting cord morphologies were noted. The most prevalent gonadal morphology was a simple, linear cord arrangement where individual cords ran parallel to one another and each cord extended the whole width of the gonad with the exception of the unlabeled surrounding tissue beneath the germinal epithelium (Fig. 7A). Parallel cords appeared to slightly “fan out” towards the posterior and anterior poles of the testis, and cords close to the poles were shorter than those in the center of the gonad. In some cases cord buds were noted, which were significantly shorter than neighboring cords (Fig. 7B). Sometimes two or more cords originated from the rete testis, but were fused together to form a single cord as it extended distally (Fig. 7C, D). The reverse was also observed, where a single cord originating from the rete testis was branched to form more than one cord distally (Fig. 7C). Furthermore, cords were noted that were connected with a neighboring cord by a bridge, while the number of cords at both the proximal and distal sides remained the same (Fig. 7C). More than one of these phenomena were noted in the same gonad (Fig. 7C). Formation of these connections between cords was visible in time-lapse movies (Movie 1C). These observations were made only from the outside surface of the gonad, because the thickness of the organ does not permit the laser to penetrate deep enough to image internal structures. To evaluate the inner morphology of the fetal testis, serial histology sections were generated and analyzed.

FIG. 7.

Sox9-EGFP testes at E14.5 with different cord arrangements. Simple linear arrangement (A), cord bud (B, arrowhead), fused cords (C, asterix), branched cord (C, arrowhead), a cord bridge (C, arrow) and multiple fused cords (D, asterix). Scale bar = 200 µm.

Testis cords branch to fill the inner volume of the early testis

Two individual, neighboring testis cords from the center of the gonad were followed through serial sections by outlining the cords from the distal to the proximal end of the gonad, through the entire organ in a color-coded manner (Fig. 8A). These images were used to reconstruct the two cords to obtain a lateral view (Fig. 8B, C). Although only the results for two cords are presented, similar findings were obtained for other cords at E13.5 and E14.5. Cords had a circular outer region representing the distal loop structure, which were observed under fluorescence microscopy, and an inner region, which branched off from the loop and extended through its center. This branch ran either perpendicular to the mesonephros (Fig. 8B), or at an angle (Fig. 8C) before it merged again at the rete testis. This inner part could also split to form two different branches before joining the rete that connected adjacent cords. Separate cords could not be distinguished at the most proximal part of the SGCM, because the cords merged to form the rete testis. In some cases, neighboring cords fused before entering the rete testis, which corresponds to the above descriptions of different cord morphologies (Fig. 7). The 3D reconstruction of cords was performed by extrapolation of section images. These are presented in wire frame for the yellow-labeled cord (Fig. 8D) and iso-surface format for the red-labeled cord (Fig. 8E). Movie 2A and B show animations of these reconstruction models.

FIG. 8.

Serial histological sections and 3D reconstruction of E13.5 mouse testis from the distal end through the rete testis toward the mesonephros. Two separate cords were outlined in red and yellow in the serial sections (A) and followed from the distal to proximal side (B, C) until no more germ cells were noted, which marked the end of the cords at the rete tesis. Cords were then reconstructed into models (D, E) of the two single cords by mathematical interpolation to show the complex nature and branching of individual cords. The yellow cord (D) is presented in wire frame format and corresponds to Movie 2A while the red cord (E) is showing the solid isosurface and corresponds to Movie 2B. The actual cords extended further toward the mesonephros, but data was truncated where cords fused on the side of the rete testes and individual cords became indistinguishable.

Discussion

We have generated Sox9-EGFP knock-in mice to mark Sertoli cells of the developing testis. We first characterized the expression pattern of EGFP in the gonad starting at E11.5 to determine if it corresponded to Sox9 expression in mouse testes. Kent et al. (1996) and Morais de Silva et al. (1996) detected low levels of Sox9 in both XY and XX gonads in E10.5 mouse urogenital ridges. However, Sox9-EGFP expression was only detected at E11.5 (TS18). We suspect that Sox9 transcription at E10.5 may not be at sufficient levels to detect EGFP fluorescence. The fluorescent protein was initially detected starting around TS18 along the length of the morphologically indifferent XY gonad and often apparently originating from the center while the poles are still not expressing the fluorescent protein. This finding is consistent with previous reports that show that both Sry and Sox9 exhibit center-to-pole expression patterns at initiation of expression in the embryonic testes (Bullejos and Koopman, 2001; Bullejos and Koopman, 2005). This is also true for Sox8 and Amh, that are expressed shortly after Sox9 at TS25 (Schepers et al., 2003). Furthermore, when Sry-EGFP gonads were analyzed, EGFP expression also showed this center-to-pole expression and was detected shortly after Sry is first expressed (Albrecht and Eicher, 2001).

Fluorescent intensity greatly increased after initial detection and persisted throughout testicular development and adulthood, whereas no fluorescence was ever detected in female gonads. This corresponds with previous reports of up-regulation of Sox9 expression patterns at the onset of testis differentiation in mice (Kent et al., 1996; Morais da Silva et al., 1996; Sekido et al., 2004). EGFP expression could therefore be used to differentiate between male and female gonads before sex-specific morphologies are apparent.

We found faint fluorescent protein expression in the anterior part of the mesonephros in the location of the mesonephric tubules from E11.5 through E14.5. Morais da Silva et al. (1996) also reported this in whole-mount in situ hybridization sections and according to the authors, this was reminiscent of a Dax1 positive group of cells present in both sexes after E12.5. In contrast, Kent et al. (1996) did not report Sox9 expression in the mesonephroi of either sexes by whole-mount in situ hybridization at E11.5, but reported expression in the mesonephric tubules in the male at E12.5. In addition, the Wolffian duct also had faint fluorescence indicative of Sox9 activity, which was similar to findings in mice (Kent et al., 1996) and in marsupials (Pask et al., 2002).

After translation, EGFP remains in the cytoplasm (Okabe et al., 1997). Furthermore, EGFP fluorescence persists after formaldehyde fixation (Chalfie et al., 1994), which allows immunostaining of other cellular components while the EGFP remains visible. Co-localization of EGFP and SOX9-positive nuclei confirmed that in the gonad EGFP is expressed only by Sox9-expressing Sertoli cells and that green fluorescence was a Sertoli cell-specific indicator. These Sox9-EGFP mice should provide a useful marker of Sertoli cells when crossed with mice carrying mutations that affect testis development.

The first notable morphological change was an increase in gonad thickness. This supported similar findings where, although no clear organizational changes were notable, the gonad increased in size from E11.5 to E12.0 (Karl and Capel, 1995). The entire SGCM is surrounded by ~4–6 layers of unlabeled cells that proliferate during this stage of development. The gonadal size increase, however, is not attributed to an increase of this layer, since we did not notice an increase in the number of cell layers between the germinal epithelium and testis cords. Instead, we propose that Sertoli cell proliferation (Schmahl et al., 2000; Schmahl and Capel, 2003), migration of mesenchymal cells into the gonad (Buehr et al., 1993; Martineau et al., 1997) and gathering of interstitial cells around the forming cords towards the inside of the gonad, are the main causes for the increase in gonadal volume. It is also possible that the Sertoli and germ cells at the anterior and posterior ends of the gonads migrate or contracts towards the center by surrounding tissue which would contribute to the thickness increase and explain the decrease in gonad length that is observed from E11.5 to E13.5. Interestingly, it has been reported that there is a temporary cease in Sertoli cell proliferation between TS18 and 23 (Schmahl et al., 2000), which is the time when Sertoli cell rearrangement into cords begin. Therefore, the increase in gonadal width is likely attributed to rearrangement of the Sertoli cells and other cell populations already present and not to an increase in Sertoli cell numbers, at least for the above-mentioned window of time. After TS23, however, proliferation of Sertoli cells is resumed, which then contributes to formation and increase in cord length.

A recent study described the formation of germ cell clusters on the ventral and dorsal sides of the gonad prior to cord formation (Combes et al., 2008), which appear to coincide with the origin of the small protrusions initially noticed in our investigation. This may lead to the conclusion that clustering of germ cells are responsible for the protrusions at the distal edge and initiate cord formation. However, germ cells are not essential for cord formation (McCoshen, 1983). Further investigation is required to determine whether this clustering is intrinsic to the germ cells or from external signals or forces exerted on them by surrounding cells.

Two different types of tissue migration have been described during development. The first is a mass migration where the cells maintain cohesive contact with one another while moving as a unit, such as convergent extension (Keller et al., 2000). The second type involves the migration of individual cells or small groups of cells through the extracellular matrix and requires the loss of cell contact (Hay, 1995). Based on the behavior of the SGCM we observed, it appears that the tissue moves as a unit, which would suggest the first type of migration. Interestingly, the SGCM consists of two different cell types, which would make this a unique case of mass migration. Another interesting question is how much of the tissue mass reshaping is due to intrinsic tissue movement and how much is due to external forces exerted by the surrounding interstitium and endothelial network. The latter would be a more passive process as far as the SGCM is concerned, while it would require active processes by the surrounding tissue.

Adult seminiferous tubules are responsible for producing millions of spermatozoa per day, which requires an enormous Sertoli epithelial tubule surface to nurture the germ cells used during spermatogenesis. It is crucial that the testis cords elongate and become coiled to increase the Sertoli cell surface and maximally use the available testis volume. This is typical for mammals, although variation between species and individuals exists (Lennox and Logue, 1979; Wing and Christensen, 1982; Gaytan et al., 1986; Sinha Hikim et al., 1988; Neves et al., 2002; Franca and Godinho, 2003; Almeida et al., 2006). Our cord reconstruction data suggests that a morphogenetic mechanism exists in the mouse that significantly increases cord volume through branches of outer loops that run through the interior of the gonad and maximizes volume usage. It would be interesting to know if more such branching mechanisms for increasing the seminiferous surface exist in other species.

The number of cords that formed in each gonad varied between gonads at the same developmental stage. Furthermore, a few cords were eliminated between E12.5 to E13.5, either by fusing with neighboring cords or perhaps by shrinking back into the rete testis. This reduction in cords could be seen in vitro during live, time-lapse imaging. When gonads were grouped according to number of cords, it became clear that cord number affects gonad dimensions. Serial section of gonads was therefore performed to compare measurements in gonads with different cord numbers. After noticing the complexity of the cords, we decided to use these sections to reconstruct the cords instead.

Initially, it was thought that cords are simple loops or arches that run parallel to one another and connect at the rete testis, as was described for the rat (Clermont and Huckins, 1961). To our surprise, very few (if any) of the cords that we examined consisted of a single, simple loop. Instead, they branched off to extend cords through the inside of the loops. Furthermore, in one example, the inner cord branched further as it approached the rete testis, resulting in four different cord endings at the rete testis. We expect much variation between different cords in their shape and branches, and this is likely also affected by cord number and gonad dimensions. Reconstruction data also differs from findings in rats by Clermont and Huckins (1961), in which they described “inner” and “outer” cords as “simple arches”. We never saw inner cords that had distal ends underneath those of outer cords at the stages examined (E13.5 and E14.5). Instead, we sometimes saw slight variation in the length of adjacent cords, but these were still visible side by side in a single section. In the same study, the total number of cords reported for rats was more than double what we counted for mice when cords were first distinguishable. They did, however, report cord branching, which corresponds with our findings.

The development of the rete testis and vas efferens have not been described in the mouse embryo. Although more detailed characterization is needed, our study shows that the sex cords merge into a single, longitudinal cord-like structure running parallel to the mesonephros in the site where the rete testis develops. This structure appears to shorten relative to the testis length and by E15.5 is connected to the vas efferens through an unknown number of ducts. Interestingly, fluorescence in the Wolffian duct continued through the early epididymis and appeared to connect with the longitudinal cord-like structure in the forming rete testis. This suggests that as early at E15.5 the future seminiferous tubules are connected to the immature male reproductive tract through what will become the adult vas efferens. In adult mice and rats there are two efferent channels that connect the rete testis with the caput epididymis (Cooper and Jackson, 1972). Future studies will focus on the anatomy of these early ducts to determine the number of ducts or cords that make up the vas efferens in the embryo.

Our current data suggests that very early in gonadal development, interaction between differentiating Sertoli cells, unidentified non-Sertoli cells and migrating cells from the mesonephros causes a rapid rearrangement of gonadal cell types into testis cords with surrounding interstitium. It is known that mesenchymal migration into the gonad is triggered by signals from within the developing testis (Martineau et al., 1997). It is also known that embryonic Sertoli cells secrete chemotactic factors, such as fibroblast growth factor 9, platelet derived growth factor, hepatocyte growth factor and neurotropin-3, all of which are seemingly involved in mesonephric cell migration and possibly vascularization by affecting the migrating endothelial cells (Cupp and Skinner, 2005). We hypothesize that the gonadal cell rearrangement is accomplished by such initial signals from the differentiating Sertoli cells, followed by mechanical constriction of the SGCM by surrounding and migrating cells, which act as a mold and reshape the SGCM into what we know as testis cords. Future studies will focus on identifying the mode of cellular relocation of the different cell types that are involved in forming the differentiated embryonic testis.

Experimental Procedures

Mice

Sox9-EGFP knock-in mice were generated by introduction of an IRES-EGFP-pA cassette into the 3’ untranslated region of the endogenous Sox9 gene by gene targeting in mouse embryonic stem cells and blastocyst injection (Nagy et al., 2003). Apparently, this knock-in does not significantly affect Sox9 function because Sox9-EGFP heterozygotes and homozygotes were viable and fertile, whereas Sox9 null heterozygotes die at birth (Bi et al., 2001) and Sox9 null homozygotes die around E11.5 (Chaboissier et al., 2004). Details of the generation of these mice will be described in detail elsewhere (Akiyama, in preparation). Sox9-EGFP homozygous males were generated on a C57Bl/6J-129/SvEv-Swiss mixed genetic background. All subsequent experiments were performed by mating Sox9-EGFP homozygous males with Swiss females (Taconic, NY) and embryonic genotype was confirmed by fluorescence.

Regents and media

All reagents were purchased from Gibco (Invitrogen Corporation, Grand Island, NY), unless stated otherwise. Organ culture medium consisted of Dulbecco’s modified Eagle’s medium (with 4.5 g/L D-glucose, without L-glutamine, sodium pyruvate and phenol red) containing 10% fetal bovine serum, 2 mM GlutaMax, 1 mM sodium pyruvate, 0.1 mM β-mercaptoethanol (Sigma-Aldrich Inc., St. Louis, MO), 10 uL/mL MEM non-essential amino acids, 100 IU penicillin and 0.1 mg/mL streptomycin. Urogenital ridges were dissected in medium supplemented with 20 mM Hepes buffer (Sigma-Aldrich).

Immunofluorescence

SOX9 protein in embryonic and adult testes was detected by immunofluorescence as described (Poche et al., 2008). Heterozygous and wild-type E13.5 gonads (with mesonephroi) and adult testes were dissected, rinsed in Ca²⁺ and Mg ²⁺-free Dulbecco’s phosphate buffered saline (DPBS) and fixed for 30 min (embryonic) or 60 min (adult) in 4% paraformaldehyde (PFA). After dehydration in 15% and 30% sucrose in DPBS, tissue was infiltrated with 50% O.C.T compound (Tissue-Tek, Sakura Finetek USA Inc., Torrance, CA) in 30% sucrose-DPBS and frozen in 100% O.C.T. 12 µm sections were labeled with rabbit anti-SOX9 antibody (Chemicon International, Temecula, CA; 1:200 dilution) paired with a goat anti-rabbit Alexa-Fluor 546-conjugated secondary antibody (Molecular Probes, Eugene, OR; 1:400). Mounting medium contained 1.5 µg/mL 4’,6 diamidino-2-phenylindole (DAPI; Vectashield, Vector Laboratories, Inc., Burlingame, CA) and sections were imaged on a PerkinElmer spinning disc laser confocal (SDLF) microscope (PerkinElmer Life and Analytical Sciences, Inc., Waltham, MA) using 405 (blue for DAPI), 568 (red for AF 546), or 488 nm (green for EGFP) lasers.

Static imaging and morphometric analysis

Sox9-EGFP heterozygous embryos were dissected at E11.5, 12.5, 13.5, 14.5, 15.5 and 16.5. For E11.5, embryos were sexed by fixing and staining amnion cells with 1% Toluidine blue to observe the presence (XX) or absence (XY) of a condensed chromatin body (Palmer and Burgoyne, 1991). Gonads (with adjacent mesonephroi attached) were placed in dissection medium in a poly-d-lysine coated glass bottom culture dish (MatTek corporation, Ashland, MA) and imaged on a spinning disc laser confocal microscope using 4× and 10× objectives. Images were processed and measured using UltraVIEW ERS ImageSuite Software (PerkinElmer Life and Analytical Sciences, Inc.). Data collected included gonad length (all stages), gonad width (all stages), rete testis length (E12.5-E14.5), PCD, DCD and the total number of cords entering the rete testis. For PCD and DCD, the three longest cords of each gonad were measured and the average values were used. For counting cell layers surrounding the testis cords, frozen sections of Sox9-EGFP testis was performed, stained with 4', 6-diamidino-2-phenylindole (DAPI) and imaged. Testes of E14.5 embryos were cut longitudinally through the distal loop-like cord structures and imaged from the inside out and outside in to observe cords as they entered the developing rete testis.

Statistical analyses

Statistical analyses were performed using GraphPad Instat software (version 3.0; GraphPad Software, San Diego, CA) and P-values smaller than 0.05 was considered significant. Initially, left and right side gonads were compared based on the mentioned variables using unpaired, two-way t-tests (parametric) and Mann-Whitney tests (nonparametric). Because left and right side gonads were similar for all variables, data from the two sides were pooled for further comparisons. Differences in measurements at different time points, as well as cord number groups were compared by unpaired t-tests (parametric) and Mann-Whitney tests (nonparametric) for two groups, or by one-way analysis of variance (ANOVA) with Bonferonni multiple comparisons test (parametric) or Kruskal-Wallis (nonparametric) tests for more than two groups.

Organ culture and time-lapse imaging

A previously described organ culture system for in vitro differentiation of mouse gonads (Martineau et al., 1997) was modified to accommodate time-lapse imaging requirements. Briefly, gonads were cultured at the air-medium interface using a Millicell tissue culture plate insert (Millipore Corporation, Billerica, MA) in a 35 mm poly-d-lysine coated glass bottom culture dish. A high viscosity grease ring was inserted around the culture well to form a fluid chamber at the outer edges of the culture dish. This was filled with medium to increase humidity and help prevent medium evaporation. The control gonads were cultured similarly in a humidified incubator at 37 °C in 5% CO₂. Time-lapse imaging was performed on a PerkinElmer spinning disc laser confocal microscope at 37 °C and 5% CO₂ in a humidified environmental chamber using the 488 nm laser. Images were acquired at 100× magnification (10× objective) with 60–80% laser power and 500 msec exposure. The Z-plane distance was 4–6 µm. For creation of movies, data was processed using ImageSuite software.

Serial sections and cord reconstruction

Gonads from E13.5 (n=2) and E14.5 (n=2) were fixed, embedded in paraffin, serially-sectioned at 3 µm thickness, and stained with H&E. High resolution images that allowed identification of basement membranes and germ cells inside the cords were acquired every 9–15 µm (every 3–5 sections). Testis cords were outlined using Adobe Illustrator software (Adobe Systems inc., San Jose, CA) in different colors to allow visualization of neighboring cords. Two neighboring cords from an E13.5 gonad were chosen for further processing and 3D reconstruction. A 2D image was constructed from the outlined images using Adobe Illustrator. To generate a 3D reconstruction, the sections from the original image were extracted using Adobe Photoshop CS2. A custom written MATLAB program (Mathworks, Inc., Natick, USA) based on a linear algorithm was used to generate sections to interpolate between the extracted sections. An isosurface defined by a pixel intensity threshold (Imaris 5.0.3, Bitplane) was used to construct a 3D model from the image sections.

Supplementary Material

Movie 1A. MOVIE 1A.

Sox9-EGFP testis at E11.5 cultured and imaged for 15 h at 30 min. intervals. The mesonephros is to the right and the distal edge to the left of the green SGCM. Initially the distal edge appears smooth, but become serrated after approximately 6 h when single Sertoli cell extensions become visible, followed by more defined small protrusions of the cell mass into the surrounding tissue. The gonad also becomes progressively thicker while EGFP intensity increases.

Movie 1B. MOVIE 1B.

Sox9-EGFP testis at E12.0 cultured and imaged for 18 h at 20 min. intervals. The mesonephros is at the bottom and the distal edge at the top of the green SGCM. Small protrusions on the distal edge become more defined and elongate while the length of the gonad decreases.

Movie 1C. MOVIE 1C.

Sox9-EGFP testis at E12.5 cultured and imaged for 43.5 h at 30 min. intervals. The mesonephros is at the top and the distal edge at the bottom of the green SGCM. Well-defined, small protrusions enlarge and fuse to form large cord-like structures, while the cord number decreases over time. The gonad length also decreases, while the width increases as cords elongate. Non-fluorescing areas form in the SGCM, that seemingly merge with one another to contribute to the interstitial spaces between the cords.

Movie 2A. MOVIE 2A.

Animation of a 3D reconstruction model showing the morphology of single testis cord A (wire frame format using Imaris 5.0.3) (Fig. 10 D).

Movie 2B. MOVIE 2B.

Animation of a 3D reconstruction model showing the morphology of single testis cord B (isosurface through equal intensity points using Imaris 5.0.3) (Fig. 10 E).