REPRODUCTIONRESEARCH

Initiation of testicular tubulogenesis is controlled by neurotrophic tyrosine receptor kinases in a three-dimensional Sertoli cell aggregation assay

Kathrin Gassei1, Jens Ehmcke1,2 and Stefan Schlatt1,2 1Department of Cell Biology and Physiology, Center for Research in Reproductive Physiology, University of Pittsburgh School of Medicine, 3500 Terrace Street, Pittsburgh, Pennsylvania 15261, USA and 2Institute of Reproductive and Regenerative Biology, Centre of Reproductive Medicine and Andrology, Domagkstraße 11, 48149 Mu¨nster, Germany Correspondence should be addressed to S Schlatt; Email: [email protected]

Abstract

The first morphological sign of testicular differentiation is the formation of testis cords. Prior to cord formation, newly specified Sertoli cells establish adhesive junctions, and condensation of somatic cells along the surface epithelium of the genital ridge occurs. Here, we show that Sertoli cell aggregation is necessary for subsequent testis cord formation, and that neurotrophic receptors (NTRKs) regulate this process. In a three-dimensional cell culture assay, immature rat Sertoli cells aggregate to form large spherical aggregates (81.36G7.34 mm in diameter) in a highly organized, hexagonal arrangement (376.95G21.93 mm average distance between spherical aggregates). Exposure to NTRK inhibitors K252a and AG879 significantly disrupted Sertoli cell aggregation in a dose-dependent manner. Sertoli cells were prevented from establishing cell–cell contacts and from forming spherical aggregates. In vitro-derived spherical aggregates were xenografted into immunodeficient nude mice to investigate their developmental potential. In controls, seminiferous tubule-like structures showing polarized single-layered Sertoli cell epithelia, basement membranes, peritubular myoid cells surrounding the tubules, and lumen were observed in histological sections. By contrast, grafts from treatment groups were devoid of tubules and only few single Sertoli cells were present in xenografts after 4 weeks. Furthermore, the grafts were significantly smaller when Sertoli cell aggregation was disrupted by K252a in vitro (20.87 vs 6.63 mg; P!0.05). We conclude from these results that NTRK- regulated Sertoli–Sertoli cell contact is essential to the period of extensive growth and remodeling that occurs during testicular tubulogenesis, and our data indicate its potential function in fetal and prepubertal testis differentiation. Reproduction (2008) 136 459–469

Introduction upon specification of Sertoli cells in the male gonad remains poorly understood. The development of the testis during ontogenesis presents The family of neurotrophic growth factors has been one of the earliest events of sex determination and is proposed to be involved in testicular development dependent on the sex-specific expression of Sry in XY (Wheeler & Bothwell 1992, Levine et al. 2000). Nerve primodial gonads (Gubbay et al.1990, Sinclair et al. 1990). (NGF) was the first member of the family to Initially, pre-Sertoli cells in the undifferentiated genital be found, followed by 3 (NTF3), neurotro- ridge aggregate and establish tight contacts with each phin 5 (NTF5), and brain-derived neurotrophic factor other, increase in size, and gradually deposit basal laminar (BDNF; Reichardt 2006). The factors act through the components around these cell aggregates (Magre & Jost specific high-affinity neurotrophic tyrosine kinase 1991, Merchant-Larios & Taketo 1991). Primordial germ receptors NTRK1 (for NGF, formerly known as trkA), cells become enclosed within these aggregates during the NTRK2 (for NTF5 and BDNF,formerly known as trkB), and process of cell aggregation. Subsequently, the migration of NTRK3 (for NTF3, formerly known as trkC). Additionally, precursor cells of other somatic testicular cell lineages all factors share the common low-affinity neurotrophic such as myoid peritubular,endothelial, and interstitial cells p75/LNGFR. Recently, neuro- from the mesonephros is essential for the formation of testis trophic growths factors were reported to participate in cords (Capel 2000). Migration of mesonephric cells into the development of various non-neural organs, such as the the gonad is exclusive to XY gonads and does not occur in dermatome, kidney, and ovary (Sainio et al. 1994, Brill XX gonads (Buehr et al. 1993). Regulation of the et al.1995, Dissen et al. 1995). In the testis, Ngf mRNAwas morphogenetic events leading to testis cord formation first detected in spermatocytes and early spermatids in the

q 2008 Society for Reproduction and Fertility DOI: 10.1530/REP-08-0241 ISSN 1470–1626 (paper) 1741–7899 (online) Online version via www.reproduction-online.org Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access 460 K Gassei and others adult rodent testis where it acts as a survival factor for (Hadley et al. 1985, 1990), and to partially mature maturing spermatozoa (Olson et al.1987, Ayer-LeLievre when cultured on extracellular matrix components et al.1988). At the onset of meiosis, DNA synthesis in stage (Suarez-Quian et al.1984, Tung & Fritz 1987). We recently VIII and IX preleptotene spermatocytes is thought to be showed that three-dimensional extracellular matrix gel under autocrine control through NGF (Parvinen et al. in combination with xenografting stimulates in vitro 1992). By contrast, stage-dependent secretion of andro- morphogenesis of seminiferous tubule-like structures gen-binding protein by Sertoli cells is also regulated by from immature Sertoli cells (Gassei et al. 2006). Using NGF (Lo¨nnerberg et al. 1992), suggesting a paracrine role this novel approach to mimic testicular morphogenesis, of in the testis. Neurotrophin receptors were we conducted preliminary investigations to explore the localized to Sertoli cells in the prepubertal testis (Djakiew relevance of neurotrophins, FSH and activin on aggre- et al. 1994), whereas transient p75/LNGFR receptor gation of early postnatal Sertoli cells. We then focused our expression in late-meiotic spermatocytes and early investigations on neurotrophin signaling as a key element spermatids was found in the adult testis (Seidl et al. of Sertoli–Sertoli cell contact and tubulogenesis. 1996). NGF was also found in the embryoinc testis and it was suggested that neurotrophins might be involved in mesenchymal–epithelial transitions that occur during Results morphogenesis (Wheeler & Bothwell 1992, Russo et al. Characterization of testicular cell types after enzymatic 1995). On the other hand, NTF3 secreted by Sertoli cells digestion was suggested to act as a paracrine factor that stimulates mesonephric cell migration at this stage of gonad Testicular single cells were isolated from 7-day-old rats development (Levine et al. 2000). This is in contrast to and subjected to immunostaining in order to identify reports that show NTF3 expression exclusive to peritubular different cell types in the preparation (Fig. 1). In cell cells during embryonic development and early postnatal suspensions from three consecutive isolation procedures, Sertoli (AMH, 88.47%G3.5 positive cells), peritubular life (Russo et al. 1995). G b myoid (IGHMBP2, 16.44% 2.1), and Leydig cells Members of the transforming growth factor 1 (TGFB1) G superfamily of morphogens have been shown to (CYP11A1, 0.27% 0.1) were identified and quan- orchestrate testicular development (for review, see titatively assessed. An additional experiment included a Itman et al. 2006). Members of the superfamily include the TGFB1 subfamily, the activin subfamily, the bone morphogenetic proteins subfamily, and the more distantly related superfamily members, glial-derived neurotrophic factor and anti-Mullerian hormone (AMH), which act as key factors during testicular development (Griswold 1998, Buageaw et al. 2005, Fouchecourt et al. 2006). In particular, activin stimulates Sertoli cell proliferation during early postnatal life (Boitani et al. 1995, Schlatt et al. 1999). Activin action in the reproductive system is modulated by its antagonist follistatin, which is known to be present in the rat testis at this age (Buzzard et al. 2004). Activin thereby contributes to the establishment of the proper testicular size that is directly depending on Sertoli cell proliferation (Atanassova et al. 2005). In turn, Sertoli cell mitotic activity is modulated by follicle-stimulating hormone (FSH)-mediated contact inhibition (Schlatt et al. 1996). Thus, the regulatory network of activin, follistatin and FSH is a center piece of testicular development and function. At the cellular level, FSH activates different signal transduction cascades in Sertoli cells, including cAMP-mediated activation of protein kinase A Figure 1 Characterization of the single-cell suspension. (A) After (PRKACA), the MAP kinase pathway, and the phospha- sequential enzymatic digestion, Sertoli cells represented the main cell tidylinositol 3-kinase, C2 domain containing, a poly- type in the cell suspension, as confirmed by positive staining for anti- Z peptide (PIK3C2A) pathway (Walker & Cheng 2005). Mullerian hormone (AMH). Scale bar 100 mm. (B) A small fraction of peritubular myoid cells was observed (IGHMBP2). Scale barZ10 mm. During postnatal testis development, Sertoli cells (C) Only very few Leydig cells remained in the cell suspension proliferate and maintain a general undifferentiated state. (CYP11A1). Scale barZ100 mm. (D) Some germ cells were present in Up to postnatal day 10, these immature Sertoli cells retain the cell suspension as identified by DDX4 immunofluorescence their potential to reorganize into testis cords in vitro staining. Scale barZ50 mm.

Reproduction (2008) 136 459–469 www.reproduction-online.org

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access Analysis of in vitro Sertoli cell aggregation 461 germ cell-specific marker (DDX4, 25.04%G2.74). Peritubular myoid (15.8%G1.31) and Leydig cells (0.32%G0.12) were present in quantities similar to those observed before. Negative controls where the primary antibody was omitted prior to incubation with the fluorescent-labeled secondary antibody did not show specific staining (data not shown).

Formation of Sertoli cell spheres in three-dimensional culture The morphological cascade of Sertoli cell aggregation in three-dimensional culture has been described by us before (Gassei et al.2006). We observed rapid cell migration within hours after plating (Fig. 2). Live-imaging experi- ments showed that 4 h after plating, single cells assembled into cord-like structures (Fig. 2C). The cascade was completed after 72 h when cultures were terminated for quantitative analysis. Sertoli cell aggregation led to the formation of large spherical aggregates (Fig. 2G). These spheres showed clear edges due to refraction and could be easily distinguished in the Matrigel in a typical hexagonal arrangement (Fig. 2H). No differences between cultures using standard Matrigel or growth factor-reduced Matrigel were observed (data not shown). Two characteristics of spherical aggregates were measured to quantify Sertoli cell aggregation in control and treatment groups. As a measure of Sertoli cell aggregation, we first determined the diameters of spherical aggregates to gain information on aggregate size. Second, the distance between spherical aggregates was considered a measure for Sertoli cell migration through the Matrigel. The spatial arrangement of Sertoli cell aggregates provides information about the gradual attractive or repulsive signals established within the three-dimensional culture matrix. Overall, the three-dimensional culture system and the applied computer-based analysis of Sertoli cell clusters proved to be applicable to assess reliable parameters. Small areas of the micrograph were out of focus due to the concave meniscus of the three-dimensional Matrigel and were excluded from analysis. Data derived for both parameters were statistically tested in order to evaluate the validity and the repeatability of Figure 2 Morphological cascade of Sertoli cell aggregation in three- the assay. The coefficient of variation (CV) for data derived dimensional culture. (A) After 1.5 h of plating, Sertoli cells are attached from five independent control experiments (no treatment) to the culture matrix. Scale barZ100 mm. (B) After 2.5 h, Sertoli cells Z ranged from 5.5 to 7.9% (diameter) and 0.5 to 5.4% start to aggregate. Scale bar 100 mm. (C) Sertoli cells form a network of cord-like structures after 3.5 h in culture. Scale barZ100 mm. (distance). The low CV values obtained from this experi- (D) The network of cord-like structures is not present after 16 h, when ment showed that the assay could be repeated with high Sertoli cells further aggregate to form clusters. Scale barZ100 mm. reliability and validity with only little dispersion of data. (E) The migration of Sertoli cells through the Matrigel is terminated after 23 h. Spherical aggregates show cytoplasmic processes. Scale barZ100 mm. (F) Aggregation and compaction of Sertoli cell spheres is Antagonizing neurotrophin signaling, but not activin or mainly completed after 44 h. Scale barZ100 mm. (G) The morpho- FSH action, disturbs Sertoli cell aggregation logical cascade is completed after 72 h. Sertoli cell aggregates form compact spherical aggregates that do not change over longer culture In qualitative preliminary experiments inhibition of periods (16!). Scale barZ100 mm. (H) Spherical aggregates show a NTRK signaling caused interruption of the morpho- highly organized distribution within the Matrigel after finalizing logical cascade of Sertoli cells. In addition to complete migration and aggregation (4!). Scale barZ500 mm. www.reproduction-online.org Reproduction (2008) 136 459–469

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access 462 K Gassei and others cell aggregation in control experiments, two different demonstrate a morphogenic effect, and antagonizing patterns of Sertoli cell aggregation were observed in FSH action could not disturb Sertoli cell aggregation. this experiment. Partially inhibited Sertoli cell aggrega- Similarly, antagonizing activin action by supplementing tion led to the formation of loose Sertoli clusters that cultures with follistatin did not prevent Sertoli cells from were smaller than spherical aggregates but still showed aggregating or stimulate enhanced aggregation (Table 1). some degree of organization (Fig. 3C). By contrast, We therefore decided to further investigate the role of high doses of inhibitor completely disturbed Sertoli cell neurotrophins during Sertoli cell aggregation. aggregation. Sertoli cells attached to the Matrigel but remained as single cells throughout the culture period NTRK3 involvement in Sertoli cell aggregation (Fig. 3D). Addition of exogenous NTF3 did not enhance Sertoli Application of the NTRK inhibitor K252a caused a cell aggregation. Furthermore, exogenous FSH did not dose-dependent disruption of Sertoli cell aggregation (Table 1, panel B). In cultures treated with 1 nM K252a, cell migration was partially inhibited. In these cultures, Sertoli cell clusters were predominantly observed after 3 days of culture (Fig. 3C). Higher doses of the inhibitor (5 and 10 nM) completely inhibited Sertoli cell aggregation and only single cells could be observed (Fig. 3D). For quantitative analysis, samples that showed spherical aggregates or Sertoli cell clusters in more than 40% of all captured frames were measured, whereas samples with mainly single cells present in the culture were qualitatively recorded and not subjected to quantitative analysis (Table 2). Accordingly, the cultures supplemented with 1 nM K252a were ana- lyzed and compared with no-treatment and vehicle controls (Fig. 4A). Aggregate size averaged to 81.36 mm(G7.34 mm S.E.M.) diameter in controls, and the distance between Sertoli cell spheres averaged to 376.95 mm(G21.93 mm S.E.M.). By contrast, K252a in concentrations below the IC50 value (3 nM) caused partial Sertoli cell inhibition and the formation of cell clusters with a reduced diameter of 39.93 mm(G3.31 mm S.E.M.), representing a decrease of 50.92%. Similarly, Sertoli cell clusters also showed a highly significant decrease in distance (68.57%) and averaged to 118.47 mm(G6.29 mm S.E.M.; P!0.001). The overall hexagonal arrangement of Sertoli cell clusters was still established in the cultures containing 1 nM K252a, but was not observed in the cultures incubated with above IC50 concentration (5 and 10 nM; Table 2). Recent studies have suggested a role of NTF3 in testicular cord formation in organ cultures (Levine, Cupp et al. 2000). We therefore tested whether the effect of K252a in vitro was due to the suppression of neuro- trophin 3 signaling. We investigated the influence of Figure 3 Disruption of Sertoli cell migration and aggregation by the NTRK inhibitors K252a and AG879. (A) No-treatment control. Sertoli exogenous NTF3 in the cultures and the effects of cell aggregates after 3 days of culture form large spherical aggregates. neutralizing anti-NTF3 antibodies on Sertoli cell aggre- (B) Vehicle control with 0.5% DMSO. No influence on Sertoli cell gation in vitro. To attempt a higher degree of Sertoli cell aggregation was observed. (C) Incubation with 1 nM K252a partially aggregation, the cultures were supplemented with 10, inhibits Sertoli cell aggregation. (D) Incubation with 5 nM K252a has a 50, or 100 ng/ml neurotrophin 3 (NTF3). We did not find dramatic inhibitory effect on Sertoli cell aggregation. No spherical a stimulating effect of NTF3 on Sertoli cell aggregation, aggregates were observed after 3 days of culture. (E) Similarly, Sertoli cell aggregation was partially inhibited by 5 mM AG879. (F) Twenty and the appearance of spherical aggregates was similar micro molar AG879 disrupted the morphogenetic cascade of Sertoli to no-treatment controls. Furthermore, exogenous NTF3 cell cord formation in vitro. Occasionally, small clusters were observed was not able to rescue Sertoli cell aggregation upon after 3 days of culture (All scale barsZ50 mm). treatment with K252a (Fig. 4B). In addition, incubating

Reproduction (2008) 136 459–469 www.reproduction-online.org

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access www.reproduction-online.org

Table 1 Qualification and quantification of Sertoli cell aggregation.

NTF3 K252a FSH Anti-FSHR antibody

10 ng/ml 50 ng/ml 100 ng/ml 1 nM 5 nM 10 nM 10 ng/ml 50 ng/ml 100 ng/ml 1:100 1:500 1:1000 (A) Preliminary tests – qualitative analysis n 66 6666666666 Stimulatory or K K K CCCCCCK K K KKK inhibitory effect

Neutralizing anti-NTF3 Neutralizing Neutralizing anti- K252a antibody anti-NTRK3 antibody Tyrphostin AG879 NGF antibody Follistatin

1:20 1:200 50 100 500 1 nM 5 nM 10 nM 1:100 1:250 1:500 1 mg/ml 10 mg/ml 5 mM10mM20mM 000 000 ng/ml ng/ml ng/ml (B) Treatment with potential aggregation inhibiting compounds – quantitative analysis n 10 10 10 2 2 2 3 3 2 2 2 2 2 4 4 4 Inhibitory effect C CC CCC K K K K K C CC CCC K K K K K

Neutralizing anti-NTRK3 antibodyC K252aC50 ng/ml NTF3 K252aC100 ng/ml NTF3 100 ng/ml NT3 nlssof Analysis 1 nM 5 nM 10 nM 1 nM 5 nM 10 nM 1 mg/ml 10 mg/ml (C) Rescue experiments – quantitative analysis Downloaded fromBioscientifica.com at10/02/202101:46:22PM n 33322233 nvitro in Aggregation P N N P N N Y Y

Minus (K) and plus (C) signs denote intensity of stimulation or inhibition respectively. P, partial; N, no; Y, yes. etl elaggregation cell Sertoli Reproduction (2008) 136 459–469 463 via freeaccess 464 K Gassei and others

Table 2 Evaluation of dose-dependent inhibition of Sertoli cell The survival rate of hosts was 100%. Xenografts could be morphological cascade. recovered from all hosts (nZ24 for controls, nZ23 for 1 nM K252a, nZ24 for 5 nM K252a) and graft weights K252a Tyrphostin AG879 Inhibitor were significantly decreased in treatment groups Concentration 1 nM 5 nM 10 nM 5 mM10mM20mM (P!0.05). The grafts from the control group (DMSO) G Areas with 0% 0% 0% 10% 0% 0% averaged to 20.87 mg ( 2.16 S.E.M.), whereas the grafts spheres from the low-concentration treatment group (1 nM) Areas with 92.98% 23.50% 19.15% 75% 38.10% 20% averaged to 9.92 mg (G1.82 S.E.M). Grafts from both clusters groups were vascularized. By contrast, the grafts from the Areas with 7.02% 76.50% 80.85% 15% 61.90% 80% G single cells high-concentration group (5 nM; 6.63 mg 0.75 S.E.M.) showed fewer blood vessels and appeared clear (Fig. 6). Histological sections of the grafts revealed that Sertoli cell cultures with a neutralizing anti-NTF3 antibody or cell aggregates developed into seminiferous tubule-like neutralizing anti-NTRK3 antibody did not affect cell structures after 4 weeks in the host (Fig. 7A, control). migration and aggregation (Fig. 4C and D). Polarized epithelial Sertoli cells were observed. Further- more, peritubular myoid cells were present on the NTRK1 signaling is ivolved in Sertoli cell aggregation outside of tubule-like structures. A basement membrane surrounding tubule-like structures was present, as well as Although K252a is a competent inhibitor of high-affinity lumen formation (Fig. 7B). The grafts derived from Sertoli neurotrophin receptors, it acts in a non-specific manner cells treated with 1 nM K252a showed similar tubular and has been shown to inhibit NTRK3 as well as NTRK1 organization, but the grafts were smaller and contained and NTRK2. In a second set of experiments, we therefore fewer tubule-like structures (data not shown). In contrast, examined the action of NGF on Sertoli cell aggregation. complete inhibition of Sertoli cell aggregation by 5 nM Similar to the results obtained for exogenous NTF3, K252a prior to xenografting caused complete absence of supplementation of cultures with NGF did not stimulate seminiferous tubule-like structures in xenografts Sertoli cell aggregation (Fig. 5B). However, when Sertoli (Fig. 7C). Grafts were encapsulated similar to controls, cells were cultured in the presence of the specific NTRK1 but Matrigel patches contained only few single cells that antagonist tyrphostin AG879, the same dose-dependent were mainly aligned along the host-derived capsule effect on Sertoli cell inhibition was observed as described (Fig. 7D). for K252a (Fig. 3E and F). Quantitative analysis revealed a At the time of graft removal, body weights and seminal highly significant decrease in spherical aggregate size by vesicle weights from the castrated recipients were 46.69% (32.21 mmG2.86 vs 60.42 mmG3.62; P!0.001) recorded. All hosts appeared healthy and we did not find in cultures treated with 5 mM AG879 (Fig. 5A). In addition, differences in body weight between treatment groups. the cultures appeared increasingly unorganized and Notably, seminal vesicles were significantly larger in distances between cell clusters were reduced from control animals than in animals carrying grafts after in vitro 318.10 mmG17.87 in controls to 116.56 mmG11.45 inhibitor treatment (Table 3), thus indicating steroidogenic after treatment with the inhibitor (P!0.001). These results activity in controls, but not in the treatment group. present a significant decrease in cell cluster distance by 63.36%. In cultures treated with concentrations equal or above the IC50 value (10 mM) of tyrphostin AG879 (10 mM and 20 mM), single Sertoli cells were mainly present as a Discussion result of a complete inhibition of Sertoli cell migration and In this study, we developed a quantifiable cell culture aggregation (Table 2). Although these results point to the assay for Sertoli cell aggregation that models early involvement of NTRK1 in Sertoli cell aggregation, testicular organogenesis. We then used this assay to experiments using neutralizing anti-NGF antibody examine the relevance of NGF, NTF3, FSH, and activin surprisingly did not affect Sertoli cell aggregation in this on Sertoli cell aggregation. assay (Fig. 5B). Our results indicate a regulatory role of high-affinity neurotrophic tyrosine kinase receptors during Sertoli cell aggregation. Recent studies on explants of the undiffer- Tubule formation in xenografts depends on NTRK- entiated gonadal primordium have suggested a potential mediated Sertoli cell aggregation role of neurotrophin-mediated signaling during testicular Sertoli cell aggregates from three-dimensional cell development (Levine et al. 2000). In accordance with cultures treated with K252a were xenografted to host this, we observed a compelling dose-dependent anti- mice in order to test the necessity of Sertoli cell aggregation effect on Sertoli cells of neurotrophic aggregation for testicular differentiation and the develop- tyrosine kinase receptor inhibitors (K252a and AG879), ment of seminferous tubule-like structures. As controls, indicating that Sertoli cell aggregation underlies Sertoli cell aggregates treated with DMSO were grafted. regulation via NTRK signaling.

Reproduction (2008) 136 459–469 www.reproduction-online.org

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access Analysis of in vitro Sertoli cell aggregation 465

By contrast, Sertoli cell aggregation could not be inhibited by follistatin, the naturally occuring binding protein for activin. From this result, we conclude that signaling through activin-specific receptor subunits does not contribute to Sertoli cell aggregation. Similarly, addition of FSH or blocking of FSH receptor binding did not interfere with Sertoli cell aggregation in vitro, indicating that Sertoli cells aggregate independent of FSH signaling via cAMP and PRKACA. This result is surprising because of the known regulatory effects of activin and FSH on postnatal Sertoli cell proliferation and tubular outgrowth. One possible explanation could be our preliminary observation that Sertoli cells do not proliferate in three-dimensional Matrigel culture and only resume mitosis upon xenografting. On the contrary, we show that NTRK-mediated aggregation of immature Sertoli cells is crucial for subsequent differentiation of Sertoli cells and tubule formation in xenografts. Sertoli cells that were prevented from aggregating in vitro by K252a supplementation did not develop into seminiferous tubule-like structures in xenografts. As previously shown, although three-dimen- sional culture does support Sertoli cell in vitro morpho- genesis, differentiation is arrested after short-term culture. However, Sertoli cells resume maturation upon xeno- grafting to an immunodeficient host, leading to the formation of seminiferous tubule-like structures (Gassei et al. 2006). Similarly, Sertoli cell aggregates from control groups showed a mature phenotype in xenografts and were arranged in tubule-like structures. By contrast, tubule-like structures were absent in xenografts after exposure of Sertoli cells to NTRK inhibitors. We conclude from these results that Sertoli–Sertoli cell adhesive contacts within aggregates are mandatory for subsequent testis cord formation and tubulogenesis. K252a and AG879 have been used in previous studies reporting a role of NTF3 in testis cord formation (Levine et al. 2000, Cupp et al. 2003). To examine whether NTF3 could be accounted for the inhibition of in vitro Sertoli cell aggregation by K252a, some Sertoli cell cultures were supplemented with a combination of K252a and NTF3 at the time of plating. NTF3 did not rescue the

Figure 4 Quantitative analysis of aggregate size and distances between Sertoli cell aggregates upon K252a treatment. (A) Effect of K252a on Sertoli cell aggregate size (diameter) and distance. At a concentration of 1 nM K252a, both parameters were significantly decreased when compared with no-treatment and vehicle controls (DMSO). (B) The inhibitory effect of 1 nM K252a could not be rescued by simultaneously incubating cells with 100 ng/ml NTF3. (C) Neutralizing anti-NTF3 antibody did not show an effect on Sertoli cell aggregation when added to three-dimensional cultures in low, medium, or high concentrations. Spherical aggregates were observed in all cultures. (D) Supple- mentation of cultures with anti-NTRK3 antibody alone or in combination with NTF3 did not affect the formation of spherical aggregates (Different lettering indicates significant statistical differences between treatment groups as revealed by one-way ANOVA or Kruskal– Wallis one-way ANOVA on ranks analysis). www.reproduction-online.org Reproduction (2008) 136 459–469

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access 466 K Gassei and others

Figure 6 Distribution of graft weights after 4 weeks in the three treatment groups. The range of graft weights is shown as dot blot for each group (blue circles). Mean and S.D. are denoted on the right (yellow circles). Statistically significant differences are indicated by different letters (a and b) and were identified by Kruskal–Wallis one- way ANOVA on ranks analysis.

than the Ki values of 18–24 nM for PRKACA, PRKCC, and PKG inhibition. In addition, previous studies using organ cultures to investigate the chemotactic role of neuro- trophins during testis cord formation were performed with 100 nM K252a, and in these studies, no inhibitory effect on PRKACA by K252a was observed (Levine et al. Figure 5 Quantification of the inhibitory effect of tyrphostin AG879 on 2000). K252a and AG879 had strikingly similar, dose- Sertoli cell aggregation. (A) Tyrphostin AG879 at a concentration of dependent inhibitory effects on Sertoli cell aggregation m 5 M significantly inhibited Sertoli cell migration and aggregation when using two different concentrations close to the within the three-dimensional culture system. (B) Neutralizing anti-NBF antibody did not show an effect on Sertoli cell aggregation when added respective IC50 values (3 nM for K252a and 10 mM for to three-dimensional cultures in low or high concentrations. Spherical AG879). Partial inhibition of Sertoli cell aggregation was aggregates were observed in all cultures (Different lettering indicates observed at concentrations below this threshold, significant statistical differences between treatment groups as revealed whereas higher concentrations completely inhibited by one-way ANOVA). Sertoli cell aggregation. This implies that the morpho- logical effects observed for K252a and AG879 are due to the inhibition of NTRKs. inhibitory effect of K252a. This is in correspondence to In summary, the combined in vitro/in situ strategy to the lack of a stimulatory effect of NTF3 alone. To more recapitulate testicular tubulogenesisallowsforthe specifically inhibit NTF3 action, we then used neutraliz- detailed study of the morphogenesis of the somatic ing antibodies against NTF3 and NTRK3 to inhibit Sertoli component of the testis in a reliable and quantifiable cells in culture. In these experiments, we did not see an fashion. We demonstrate that neurotrophic receptor inhibitory effect, and no differences in aggregation tyrosine kinases NTRK1 and NTRK3 are important for the patterns could be observed between treated cells and in vitro aggregation of immature rat Sertoli cells, which untreated controls. However, since K252a and AG879 occurs in a manner similar to that observed in E13–E13.5 also have been suggested to exert inhibitory effects on embryonic gonads (Magre & Jost 1991). It remains NTRK1, we tested NGF and neutralizing anti-NGF unclear whether this action is mediated by NGF or NTF3 antibody in our cultures, but did not detect a measurable specifically or independent of the ligands. effect on Sertoli cell behavior. In conclusion, we have shown that Sertoli cell K252a has been shown to act as a competitive aggregation is a crucial event during early testicular inhibitor for PRKACA and PRKCC, and thereby might development and an indespensable prerequisite for inhibit Sertoli cell aggregation via various signaling subsequent testis cord formation and testicular develop- pathways. To rule out the inhibitory effects other than ment. Failure of Sertoli cells to make contact with one those on NTRKs, we used K252a at concentrations lower another results in the disturbance of testis cord formation.

Reproduction (2008) 136 459–469 www.reproduction-online.org

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access Analysis of in vitro Sertoli cell aggregation 467

A sequential enzymatic digestion protocol was used to obtain testicular single-cell suspensions enriched for Sertoli cells (Schlatt et al. 1996). Upon dissection of the testes, the tunica albuginea was removed and the tissue digested to small tubule fragments with 1 mg/ml collagenase I (no. C2674; Sigma) and 5 mg/ml DNAse (15 U/ml, no. 104132; Roche Applied Science) in digestion medium. Digestion medium consisted of Dulbecco’s Minimum Essential Medium (DMEM, 4.5 g glucose/ml) mixed 1:1 with Ham’s F12 (Sigma N6658) and supplemented with 1% MEM non-essential amino acids (100!; BioWhittaker 13-114E), 100 IU/ml penicillin, and 100 mg/ml streptomycin (100!; Cellgro 30- 002-CI). To remove interstitial and peritubular cells, tubule fragments were allowed to settle by unit gravity for 5 min before discarding the supernatant. A second digestion step was performed similarly by incubating tubule fragments with 1 mg/ml collagenase I and 5 mg/ml DNAse in combination with 1 mg/ml hyaluronidase (Sigma H-3506). The resulting cell suspension was centrifuged at 200 g for 10 min and the pellet was resuspended in culture medium consisting of low- glucose DMEM (1 g/ml), 1% MEM nonessential amino acids, 100 IU/ml penicillin, and 100 mg/ml streptomycin solution. Figure 7 Development of xenografts from K252a-treated Sertoli cell Cell numbers were determined using a Bright-Line hemato- cultures. Resin-embedded semi-thin sections (4 mm) were stained with the periodic acid/Schiff’s reagent method. (A) Low-power micrograph of cytometer (Hausser Scientific no. 3100). a xenograft from the control group. Scale barZ100 mm. (B) High-power Identification of different cell types in the cell suspension micrograph of the xenograft shown in (A). Sertoli cells (black arrows) was assessed by immunocytochemistry. The cells were plated differentiated to form seminiferous cord-like structures with lumen onto poly-L-lysine-coated cover slips for 15 min and formalin (asterisk) and basement membrane. Peritubular myoid cells are aligned fixed (4% in DPBS) for 15 min. After washing with 1! DPBS, along tubules (arrowheads). Vascularization (v) is observed close to the the cells were incubated with 10% goat serum diluted in Z tubules. Scale bar 50 mm. (C) Xenografts did not contain tubule-like dilution buffer (1! TBS with 0.1% BSA) for 30 min at room structures when Sertoli cells were treated with 5 nM K252a. Scale barZ temperature. The cells were immunostained with antibodies 100 mm. (D) High-power micrograph of the xenograft in (C). Sertoli (black arrow) and peritubular cells (arrowhead) are present in Matrigel raised against AMH (Sertoli cell specific; cat. no. sc-6886; residues but remain unorganized and as single cells. Scale barZ50 mm. 1:50 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), a- immunoglobulin mu-binding protein 2 (IGHMBP2; peritubular myoid cell marker; Sigma no. Materials and Methods A2547; 1:2000 dilution), cytochrome P450, family 11, Isolation of primary Sertoli cells subfamily a, polypeptide 1 (CYP11A1, Leydig cell marker; Chemicon no. AB1244; 1:500; Millipore, Billerica, MA, USA), Seven-day-old CD rats were used as donors in order to yield a and DEAD (Asp-Glu-Ala-Asp) box polypeptide 4 (DDX4; single-cell suspension of undifferentiated Sertoli cells. At this germ cell marker; Abcam no. ab-13840; 1:100; Abcam Inc., age, only a relatively small fraction of premeiotic germ cells is Cambridge, MA, USA). Antibodies were incubated over night present in the testis, which greatly facilitates Sertoli cell at 4 8C. For the anti-DDX4 antibody, heat-mediated antigen isolation. Rat pups were provided by Charles River Labora- retrieval in citrate buffer (pH 6) prior to antibody incubation tories, Inc. (Wilmington, MA, USA). All procedures were in was necessary to obtain optimal immunostaining. Negative compliance with the University of Pittsburgh Guidelines for the controls were performed in all experiments by omitting Care and Use of Laboratory Animals. primary antibodies to ensure antibody specificity. The cover slips were washed thoroughly with 1! TBS (3!5 min) and Table 3 Xenografting after treatment with K252a. incubated with goat anti-mouse IgG (for IGHMBP2 detection), goat anti-rabbit IgG (DDX4 and CYP11A1 detection), or Seminal Total no. Treatment No. of vesicle Grafts per of grafts chicken anti-goat IgG (for AMH detection) at a 1:100 dilution group recipients weights recipients recovered for 60 min at room temperature. Secondary antibodies were conjugated to AlexaFlour488 chromophor. Cover slips were Control 351(G8.5)a 631 (DMSO 0.25%) then washed thrice for 5 min with TBS and nuclei were 1 nM K252a 3 21 (G6.7)b 624stained with 0.5 mg/ml DAPI stain for 5 min. After washing, 5 nM K252a 3 12 (G1.5)b 624cover slips were mounted with Vectashield medium for fluorescence microscopy. For each marker, w1000 cells a,bDifferent superscripts indicate statistically significant differences between treatment groups (P!0.05) as detected by one-way ANOVA from three independent experiments were counted to and Holm–Sidak post hoc test. determine the percentage of positively stained cells. www.reproduction-online.org Reproduction (2008) 136 459–469

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access 468 K Gassei and others

Three-dimensional cell culture evaluated was the distance from each aggregate subjected for diameter determination to the nearest, similarly identified Approximately, 1!106 cells were plated onto 300 ml recon- spherical aggregate. The straight line measurement tool was stituted extracellular matrix gel (Matrigel, BD Biosciences, used by measuring the distance between the midpoints of the Bedford, MA, USA; no. 354234, diluted 1:1 with culture neighboring aggregates. All absolute values obtained for each medium) on 24-well culture plates. The cells were cultivated in parameter were combined and averaged for each culture (see 8 a humidified incubator at 35 C in an atmosphere containing Table 1 for treatment groups and repeats). Data were expressed 5% CO . In pilot experiments, growth factor-reduced Matrigel 2 as meansGS.E.M., and data sets were statistically analyzed with (BD Biosciences) was used to test whether growth factors that SigmaStat software using one-way ANOVA and, if applicable, a naturally occur in Matrigel might influence Sertoli cell subsequent pairwise multiple comparison post hoc test. The CV behavior in vitro. was determined for both parameters (diameter of aggregates Qualitative and quantitative experiments using a variety of and distance between aggregates) in the case of negative supplements were performed as depicted in Table 1. The cells controls (no treatment) as a measure of dispersion and to were treated at the time of plating. Compounds used in this validate the cell culture-based assay. study included human recombinant neurotrophin-3 (Sigma no. N1905), neutralizing anti-NTRK3 antibody (Sigma no. T2450), anti-neurotrophin 3 antibody (Chemicon AB1532), receptor Xenografting tyrosine kinase inhibitors K252a (Sigma no. K1639), tyrphostin Adult male nude mice (strain: nu/nu) were obtained from AG879 (Sigma T2067), recombinant NGF (Sigma no. N6009), Charles River Laboratories and served as hosts for cell culture and neutralizing anti-NGF antibody (Sigma no. N6655). For grafts. Castration was performed at the time of grafting to assess control experiments, the cells were either cultured with culture androgen levels in hosts upon graft development by means of medium alone (no-treatment control) or DMSO was added to seminal vesicle weights. Sertoli cells were prepared as the medium in comparable concentrations to the treatment described above and three treatment groups were defined. As groups (vehicle control). Additionally, cultures were also vehicle controls, the cells were incubated with 0.25% DMSO. supplemented with recombinant human follicle stimulating The low-concentration group was supplemented with 1 nM hormone (FSH, from the National Hormone Program; kindly K252a inhibitor, in contrast to the high-concentration group provided by Dr A F Parlow, Torrance, CA, USA), anti-FSHR that was treated with 5 nM K252a at the time of cell plating. antibody (clone H-90), and activin-antagonizing follistatin Sertoli cells were cultured for 9 days in extracellular matrix gel (Sigma no. F2177) to control whether signaling via cAMP and were then injected s.c. in the back of hosts using an 18 G second messenger (through FSH) or through activin specific injection needle. Each host received six injections of 250 ml TGFB1 receptor subunits might interfere with Sertoli cell extracellular matrix gel. For each treatment group, three hosts migration and aggregation. All compounds were used at were grafted (18 grafts total per treatment group). The grafts concentrations recommended by the supplier or as suggested were allowed to develop for 4 weeks. Hosts were killed by by data from previous research to ensure optimal stimulation of exsanguinations under deep anesthesia and the grafts were Sertoli cells in culture. For clarity reasons, amounts of growth removed from the inner surface of the back skin. Grafts were factors, reagents, and gonadotrophins are compiled in Table 1. fixed in Bouin’s fixative overnight and then transferred to 70% ethanol for resin embedding. Body weights, seminal vesicle weights, as well as graft numbers and weights were recorded Live imaging of cell cultures in Matrigel and for each animal. quantitative analysis Phase contrast micrographs of spherical Sertoli cell aggregates Histology in Matrigel were captured with an Olympus IX71 inverted microscope equipped with a Retiga 4000R digital camera Xenografts were embedded in resin (Technovit 7100; Heraeus (QImaging, Surrey BC, Canada) and Northern Eclipse imaging Kulzer, Hanau, Germany) and 4 mm semi-thin sections were software (MVIA Inc., Monaca, PA, USA). prepared. The sections were stained with the periodic The spatial arrangement of spherical cell aggregates was acid/Schiff’s reagent method followed by hematoxylin counter- analyzed after 72 h of culture. Phase contrast micrographs staining for histological evaluation. were taken at 16! magnification of four to six randomly selected fields from each culture. In samples that showed few clearly defined aggregates, all structures were analyzed. For Declaration of interest samples with a more scattered appearance, cell clusters were The authors declare that there is no conflict of interest that randomly selected using a grid overlay. Ten grid points in the could be perceived as prejudicing the impartiality of the center of the micrograph were used for analysis and clusters at research reported. these grid points were assigned to measurements. Data were collected for two qualitative parameters. First, we determined the diameters of spherical aggregates and clusters. Funding The diameter was measured using the straight line measure- ment of the Northern Eclipse software at the largest point of the This work was financially supported by the start-up funds from aggregate. The second parameter that was quantitatively the University of Pittsburgh School of Medicine, and NIH

Reproduction (2008) 136 459–469 www.reproduction-online.org

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access Analysis of in vitro Sertoli cell aggregation 469 grant U54 HD 008610, Center grant, project 1 (S S), and a Itman C, Mendis S, Barakat B & Loveland KL 2006 All in the family: TGF- doctoral scholarship from the Ernst Schering Research beta family action in testis development. Reproduction 132 233–246. Levine E, Cupp AS & Skinner MK 2000 Role of neurotropins in rat embryonic Foundation (K G). testis morphogenesis (cord formation). Biology of Reproduction 62 132–142. Lo¨nnerberg P, So¨der O, Parvinen M, Ritzen EM & Persson H 1992 Beta- influences the expression of androgen-binding References protein messenger ribonucleic acid in the rat testis. Biology of Atanassova NN, Walker M, McKinnell C, Fisher JS & Sharpe RM 2005 Reproduction 47 381–388. Evidence that androgens and oestrogens, as well as follicle-stimulating Magre S & Jost A 1991 Sertoli cells and testicular differentiation in the rat hormone, can alter Sertoli cell number in the neonatal rat. Journal of fetus. Journal of Electron Microscopy Technique 19 172–188. Endocrinology 184 107–117. Merchant-Larios H & Taketo T 1991 Testicular differentiation in mammals Ayer-LeLievre C, Olson L, Ebendal T, Hallbo¨o¨k F & Persson H 1988 Nerve under normal and experimental conditions. Journal of Electron growth factor mRNA and protein in the testis and epididymis of mouse Microscopy Technique 19 158–171. and rat. PNAS 85 2628–2632. Olson L, Ayer-LeLievre C, Ebendal T & Seiger A 1987 Nerve growth factor- Boitani C, Stefanini M, Fragale A & Morena AR 1995 Activin stimulates like immunoreactivities in rodent salivary glands and testis. Cell and Sertoli cell proliferation in a defined period of rat testis development. Tissue Research 248 275–286. Endocrinology 136 5438–5444. Parvinen M, Pelto-Huikko M, So¨der O, Schultz R, Kaipia A, Mali P, Brill G, Kahane N, Carmeli C, von Schack D, Barde YA & Kalcheim C 1995 Toppari J, Hakovirta H, Lo¨nnerberg P, Ritzen EM et al. 1992 Expression Epithelial-mesenchymal conversion of dermatome progenitors requires of beta-nerve growth factor and its receptor in rat seminiferous neural tube-derived signals: characterization of the role of neurotrophin- epithelium: specific function at the onset of meiosis. Journal of Cell 3. Development 121 2583–2594. Biology 117 629–641. Buageaw A, Sukhwani M, Ben-Yehudah A, Ehmcke J, Rawe VY, Reichardt LF 2006 Neurotrophin-regulatedsignalling pathways. Philosophical Pholpramool C, Orwig KE & Schlatt S 2005 GDNF Family receptor Transactions of the Royal Society of London. Series B, Biological Sciences alpha1 phenotype of spermatogonial stem cells in immature mouse 361 1545–1564. testes. Biology of Reproduction 73 1011–1016. Russo MA, Giustizieri ML, Farini D & Siracusa G 1995 Expression of Buehr M, Gu S & McLaren A 1993 Mesonephric contribution to testis neurotrophin receptors in the developing and adult testis. Italian differentiation in the fetal mouse. Development 117 273–281. Journal of Anatomy and Embryology 100 (Supplement 1) 543–551. Buzzard JJ, Loveland KL, O’Bryan MK, O’Connor AE, Bakker M, Hayashi T, Sainio K, Saarma M, Nonclerq D, Paulin L & Sariola H 1994 Antisense Wreford NG, Morrison JR & de Kretser DM 2004 Changes in circulating inhibition of low-affinity nerve growth factor receptor in kidney cultures: and testicular levels of inhibin A and B and activin A during postnatal power and pitfalls. Cellular and Molecular Neurobiology 14 439–457. development in the rat. Endocrinology 145 3532–3541. Schlatt S, de Kretser DM & Loveland KL 1996 Discriminative analysis of rat Capel B 2000 The battle of the sexes. Mechanisms of Development 92 Sertoli and peritubular cells and their proliferation in vitro: evidence for 89–103. follicle-stimulating hormone-mediated contact inhibition of Sertoli cell Cupp AS, Kim GH & Skinner MK 2000 Expression and action of mitosis. Biology of Reproduction 55 227–235. neurotropin-3 and nerve growth factor in embryonic and early postnatal Schlatt S, Zhengwei Y, Meehan T, de Kretser DM & Loveland KL 1999 rat testis development. Biology of Reproduction 63 1617–1628. Application of morphometric techniques to postnatal rat testes in organ Cupp AS, Uzumcu M & Skinner MK 2003 Chemotactic role of neurotropin culture: insights into testis growth. Cell and Tissue Research 298 3 in the embryonic testis that facilitates male sex determination. Biology 335–343. of Reproduction 68 2033–2037. Seidl K, Buchberger A & Erck C 1996 Expression of nerve growth factor and Dissen GA, Hirshfield AN, Malamed S & Ojeda SR 1995 Expression neurotrophin receptors in testicular cells suggest novel roles for of neurotrophins and their receptors in the mammalian ovary is neurotrophins outside the nervous system. Reproduction, Fertility and developmentally regulated: changes at the time of folliculogenesis. Development 8 1075–1087. Endocrinology 136 4681–4692. Sinclair AH, Berta P, Palmer MS, Hawkins JR, Griffiths BL, Smith MJ, Djakiew D, Pflug B, Dionne C & Onoda M 1994 Postnatal expression of Foster JW, Frischauf AM, Lovell-Badge R & Goodfellow PN 1990 A gene nerve growth factor receptors in the rat testis. Biology of Reproduction 51 from the human sex-determining region encodes a protein with 214–221. homology to a conserved DNA-binding motif. Nature 346 240–244. Fouchecourt S, Godet M, Sabido O & Durand P 2006 Glial cell-line- Suarez-Quian CA, Hadley MA & Dym M 1984 Effect of substrate on the derived neurotropic factor and its receptors are expressed by germinal shape of Sertoli cells in vitro. Annals of the New York Academy of and somatic cells of the rat testis. Journal of Endocrinology 190 Sciences 438 417–434. 59–71. Tung PS & Fritz IB 1987 Morphogenetic restructuring and formation of basement Gassei K, Schlatt S & Ehmcke J 2006 De novo morphogenesis of membranes by Sertoli cells and testis peritubular cells in co-culture: seminiferous tubules from dissociated immature rat testicular cells in inhibition of the morphogenetic cascade by cyclic AMP derivatives and by xenografts. Journal of Andrology 27 611–618. blocking direct cell contact. Developmental Biology 120 139–153. Griswold MD 1998 The central role of Sertoli cells in spermatogenesis. Walker WH & Cheng J 2005 FSH and signaling in Sertoli cells. Seminars in Cell and Developmental Biology 9 411–416. Reproduction 130 15–28. Gubbay J, Collignon J, Koopman P, Capel B, Economou A, Mu¨nsterberg A, Wheeler EF & Bothwell M 1992 Spatiotemporal patterns of expression of Vician N, Goodfellow P & Lovell-Badge R 1990 A gene mapping to the NGF and the low-affinity NGF receptor in rat embryos suggest functional sex-determining region of the mouse Y chromosome is a member of a roles in tissue morphogenesis and myogenesis. Journal of Neuroscience novel family of embryonically expressed genes. Nature 346 245–250. 12 930–945. Hadley MA, Byers SW, Suarez-Quian CA, Kleinman HK & Dym M 1985 Extracellular matrix regulates Sertoli cell differentiation, testicular cord formation, and germ cell development in vitro. Journal of Cell Biology Received 22 February 2008 101 1511–1522. First decision 27 March 2008 Hadley MA, Weeks BS, Kleinman HK & Dym M 1990 Laminin promotes formation of cord-like structures by Sertoli cells in vitro. Developmental Revised manuscript received 2 June 2008 Biology 140 318–327. Accepted 24 July 2008

www.reproduction-online.org Reproduction (2008) 136 459–469

Downloaded from Bioscientifica.com at 10/02/2021 01:46:22PM via free access