Gene trap analysis of germ cell signaling to Sertoli cells: NGF-TrkA mediated induction of Fra1 and Fos by post-meiotic germ cells

Germinal differentiation in the adult mammalian testis is a highly ordered process. Throughout its progression from spermatogonia to meiosis and spermiogenesis, a germ cell remains in intimate contact with a Sertoli cell. There is general agreement that the many morphological and functional changes that germ cells undergo during spermatogenesis are controlled by Sertoli cells (reviewed by Griswold, 1995). Several biochemical properties of Sertoli cells have a cyclic variation suggesting a critical role of the associated spermatogenic cells. A new cycle is initiated when germ cells have reached a defined stage of differentiation, before the completion of the ongoing cycle, thus generating the classical pattern of stages, each with a defined assortment of differentiation steps (Leblond and Clermont, 1952). Microdissection of isolated tubule segments has demonstrated the stage-specific expression of a series of proteins by the Sertoli cell, including growth factors, cytokines and proteases (reviewed by Parvinen, 1993). These notions imply a continuous cross-talk between the Sertoli cell and the associated germ cells, in which, at each maturation stage, specific signaling systems trigger the synthesis and/or the release of the required Sertoli factors.

Our knowledge of the mediators of Sertoli-germ cell interactions remains, however, imperfect. Owing to its complex architecture, the seminiferous epithelium has been a difficult area for biological studies, and many of the modes of signaling between the germ cells and their somatic partner remain to be identified at the molecular level. In vitro analysis conducted on simplified culture systems may offer useful alternatives. We previously described one such system, based on the properties of a Sertoli differentiated cell line, 15P-1. These cells express a series of Sertoli-specific genes, and form with male germ cells in cocultures multicellular complexes which support the progression of pachytene spermatocytes to the haploid state (Rassoulzadegan et al., 1993; Vincent et al., 1998).

We took advantage of the ability of 15P-1 cells to interact with male germ cells to devise a general strategy based on gene trapping (Friedrich and Soriano, 1991) for the identification of genes whose expression in Sertoli cells is regulated by germ cells and/or defined effector molecules. The method, largely used in ES cell lines, involves the selection of drug-resistant clones after transfer of a promoter-less construct composed of a splice acceptor in front of a gene (βgeo), which encodes a protein with both β-galactosidase and neo^r activities. Integration into the intron of an expressed gene in the correct orientation is predicted to create a fusion transcript. If the coding sequences in both messengers are in frame, an active protein is produced. After transfer of βgeo into 15P-1 by a retroviral vector, 2,000 independent drug-resistant clones, each one of them corresponding to a distinct integration of the transgene, were selected in geneticin-containing medium. The resulting library could then be screened for clones in which β-galactosidase activity, reflecting that of the upstream cellular promoter, would be either up- or down-modulated upon application of the stimulus of interest (addition of germ cells, of cell fractions, of soluble factors). Isolation of the chromosomal sequences upstream of the integrated transgene could then identify the responsive cellular genes, and further characterization of their expression and regulation could be conducted.

As a test of the general usefulness of the system, we conducted a first screen on 700 clones by measuring β-galactosidase activity with and without an overnight preincubation with total germ cells. Seven clones were identified in which the activity of the trapped promoter is either up- or down-modulated. One of them, designated 11F7, exclusively responded to the addition of purified spermatids and to that of nerve growth factor (NGF). A member of the neurotrophin family, NGF is essential for the development and maintenance of sensory and sympathetic neurons of the peripheral nervous system (reviewed by Levi-Montalcini et al., 1996). In the established line PC12, it induces the expression of a series of characteristic neuronal properties (Greenberg et al., 1985; Kruijer et al., 1985). It is also present in the adult testis. Although still a somewhat controversial issue, its production has been assigned by several reports to the germinal component, predominantly the round spermatids, in the form of an unprocessed precursor (Ayer et al., 1988; Parvinen et al., 1992; Chen et al., 1997). It was shown to increase the viability of Sertoli cells in culture, to stimulate DNA synthesis and cell proliferation in isolated seminiferous tubules, and to increase the levels of androgen binding protein (Lonnerberg et al., 1992; Parvinen et al., 1992; Chen et al., 1997). The two receptors identified in nerve cells are also expressed in the testis, namely the tyrosine kinase TrkA and p75^NTR, a member of the TNF receptor family. Their precise localization, however, has been the subject of somewhat divergent reports (Ayer et al., 1988; Parvinen et al., 1992; Russo et al., 1994; Russo et al., 1996; Seidl et al., 1996; Chen et al., 1997). Our own results, in agreement with the conclusions of Seidl et al. (Seidl et al., 1996), confirmed the expression in Sertoli cells of TrkA, but not of p75^NTR.

We found that, in the 11F7 gene trap clone, the βgeo cassette was inserted in a 5′ intron of the ‘Fos-related’ gene Fra1. With Fos, Jun and a series of other proteins, Fra1 is a constituent of the AP1 complex (Cohen and Curran, 1988; Cohen et al., 1989). Fra1 and Fos thus appear as candidates for central regulatory functions, that will in turn modulate the expression of other genes. One example, which is not likely to remain unique, is provided by a previous study concluding that induction of the transcription of Fos by Follicle Stimulating Hormone leads to the secondary activation of the transferrin promoter (Chaudhary et al., 1996). To access this central regulatory node, and in a more general way, to validate the gene trap approach as a means to the identification of regulated genes in complex paracrine networks, we further analyzed Fra1 and Fos induction by NGF in Sertoli cells.

15P-1 and the gene trap derivatives were grown at 32°C in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (GIBCO-BRL Life Technologies, Paisley, UK). For NGF treatment, 15P-1 cells were cultured for 24 hours before experimentation in the absence of serum, with 0.2% bovine serum albumin (Fatty acid free, Sigma A-6003, St Louis, MO, USA). NGF was obtained from Promega, Madison, WI, USA (mNGF 2.5S G5142), and the phosphorylation inhibitor PD 098,059 from Sigma.

After removal of the tunica albuginea of testes of C57BL/6 × DBA/2 F1 adult mice, the seminiferous tubules cut into small pieces were transferred in DMEM, 0.5% bovine serum albumin, 20 mM Hepes, pH 7.4. Large aggregates were removed by decantation. DNase I and collagenase A (Boehringer Mannheim) were added at final concentrations of 100 μg/ml and 1 mg/ml, respectively. After incubation at 32°C for 20 minutes, cells were pelleted by low speed centrifugation, washed twice, resuspended in the same medium, and passed through a filter with 40 μm-pores (Falcon, Becton-Dickinson, Lincoln Park, NJ, USA) pushed by the piston of a 5 ml syringe.

Total germ cells prepared from twenty mice were loaded in the Beckman (Palo Alto, CA, USA) elutriation rotor JE-5.0, at a flow rate of 7 ml/minute and a constant speed of 2,000 rpm. Cells were collected in 11 fractions of 400 ml each, obtained by changing the flow rate from 7 to 50 ml/minute at constant speed. Fraction purity was checked by microscopic analysis after Hoechst 33253 staining.

Optimal conditions for the coculture of the 15P-1 gene trap clones with germ cells were identical to those previously determined for the progression of pachytene spermatocytes through meiosis (Vincent et al., 1998). Briefly, the three most important variables were (1) to use 15P-1 cells in active exponential growth phase, (2) to add the germ cells to a suspension of 15P-1 cells after trypsinization, (3) to plate the cell mixture onto a glass substrate (Lab-Tek Chamber Glass Slides, Nunc Inc, Naperville, IL, USA).

Primary Sertoli cell cultures were established from 3 week-old mice according to published methods (Steinberger and Jakubowiak, 1993). They were maintained for 20 hours in DMEM with 10% fetal calf serum at 32°C. Attached germ cells were removed by hypotonic treatment (20 mM Tris-HCl, pH 7.4, for 2 to 5 minutes at 20°C) and the remaining Sertoli cells were washed with culture medium before further treatments. Before exposure to NGF, the cells were maintained for 24 hours in medium containing 0.5% fetal calf serum.

The enzymatic activity was measured in cell extracts by using the Galacto-Light Kit (Tropix, Bedford, MA, USA) according to the manufacturer’s instructions and was detected in situ by staining in 1.0 mg/ml X-Gal, 2 mM MgCl₂, 5 mM potassium ferricyanide and 5 mM potassium ferrocyanide in phosphate-buffered saline (pH 7.4).

Experiments were performed with Fugene 6 according to the supplier’s instructions (Boehringer Mannheim).

Quantitation of RNA amounts in cell extracts was done by comparison with that of Hprt reverse transcripts amplified as an internal standard in the same reaction mixture. Total RNA (1 μg) prepared using the Total RNA Isolation Kit (Boehringer Mannheim) was reverse transcribed using the MuLV reverse transcriptase according to the supplier’s instructions. PCR amplification was performed with the Taq DNA polymerase (Boehringer Mannheim), using the following oligonucleotide primers (polymerization conditions indicated in parenthesis). Fra1: 5′-gaccagactccgagaggc-3′ and 5′-gataggccagaggtcggg-3′ (94°C, 30 seconds: 30 cycles, 59°C, 30 seconds, 72°C, 30 seconds). Fos: 5′-caacgccgactacgaggc-3′ and 5′-cttctgccgatgctctgcg-3′ (94°C, 30 seconds: 30 cycles 59°C, 30 seconds, 72°C, 30 seconds). TK: 5′-accgagacctggccacc-3′ and 5′-ccataggtgaagatctccc-3′ (94°C, 30 seconds: 30 cycles 56°C, 30 seconds, 72°C, 30 seconds). TrkA: 5′-gaatgtgacgtgctgggc-3′ and 5′-gcgcagagacggtgctg-3′ (94°C, 30 seconds: 30 cycles 57°C, 30 seconds, 72°C, 30 seconds). TrkB: 5′-cctggctgaagtggcatg-3′ and 5′-cacgatgcctggagaagg-3′ (94°C, 30 seconds: 30 cycles 57°C, 30 seconds, 72°C, 30 seconds). TrkC: 5′-caccctgacgtgcattgc-3′ and 5′-gttgccacgcaccacaaac-3′ (94°C, 30 seconds: 30 cycles 57°C, 30 seconds, 72°C, 30 seconds). p75^NTR: 5′-ctgcctggacagtgttac-3′ and 5′-ccaagatggagcaatagac-3′ (94°C, 30 seconds: 30 cycles 55°C, 30 seconds, 72°C, 30 seconds). Hprt: 5′-gctggattacattaaagcactg-3′ and 5′-aagggcatatccaacaacaaac-3′ (94°C, 30 seconds: 20 cycles 60°C, 30 seconds, 72°C, 30 seconds). For the Hprt internal controls, a sample was taken after only 5 minutes of reverse transcription. Image analysis and densitometric measurements were performed using Adobe Photoshop 5.5 (Adobe Systems Inc., San Jose, CA, USA).

The RACE technique (5′ rapid amplification of cDNA ends; Frohman et al., 1988) was used for the characterization of sequences located at the 5′ end of the βgeo insert in the gene trap clones. First strand cDNA synthesis was performed from 2 μg of total RNA, using an oligonucleotide primer in the 5′ region of the β-geo insert (5′-gggcctcttcgctattacgc), with the AMV reverse transcriptase and deoxynucleotide solution according to the manufacturer’s instructions (Boehringer Mannheim). The first strand cDNA was then purified from non incorporated nucleotides and primers by the High Pure PCR Product Purification Kit (Boehringer Mannheim), in which the elution buffer was replaced by 10 mM Tris-HCl buffer, pH 8. Terminal transferase was then used to add a homopolymeric A-tail to the 3′ end of the cDNA. The tailed cDNA was amplified by PCR using a second primer in the β-geo sequence located upstream of the first one used (5′-atgtgctgcaaggcgattaag-3′) and an oligo-dT anchor primer provided by the manufacturer. The resulting cDNA was further amplified using a third nested oligonucleotide primer in the β-geo sequence (5′-agggttttcccagtcacgacg-3′) with the same anchor primer. 5′ RACE PCR products were cloned in the pTAG vector using LigATor Kit (R&D System, Abingdon, UK). Sequencing of the cloned cDNAs on both strands was performed using the DNA cycling sequencing kit (Perkin Elmer, Foster City, CA, USA) according to the manufacturer’s instructions.

Cells were lyzed for 15 minutes on ice in TNET solution (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM EDTA, 1% Triton X-100) containing protease inhibitors (100 μM PMSF, 1 μM leupeptin, 1 μM pepstatin A) and phosphatase inhibitors (5 mM NaF, 2 mM Na ortho-vanadate). The lysate was then centrifuged to remove cellular debris (15 minutes, 14,000 rpm, 4°C). Protein concentration was estimated by the BCA reaction (Pierce, Rockford, IL, USA). Four micrograms of proteins in Laemmli buffer were loaded onto a 10% acrylamide gel, electrophoresed, and transferred onto nitrocellulose. Phosphorylated ERK1 and ERK2 proteins were revealed with anti-active Map kinase antibodies (dilution 1/10,000, Promega, Madison, WI, USA) and anti-rabbit-HRP secondary antibodies (Bio-Rad, Hercules, CA, USA), using ECL reagent (Amersham Life Science, Buckinghamshire, UK).

In order to tag expressed loci in the established Sertoli cell line 15P-1, cells were infected with the Rosaβgeo retrovirus (Friedrich and Soriano, 1991) and seeded in 96-well plates in the presence of a relatively low concentration of geneticin (200 μg/ml). After a maximum of one month of selection with media changes every fifth day, the resistant cells were expanded. The initial seeding density had been adjusted to a value such that the frequency of the selected event was less than one per well. We generated in this way a library of about 2,000 neo^r clones. Equivalent number of cells from each well were then seeded in quadruplicate 96-well plates and preserved by freezing. A first round of screening conducted on 700 clones by culturing 2 series of plates in parallel, respectively with and without addition of total mouse germ cells at a ratio of 50 to 1 relative to 15P-1 neo^r cells. β-Galactosidase activity was determined in lysates prepared after overnight incubation.

As shown in Fig. 1 for a representative sample, β-galactosidase activity in a large majority of the clones either was not changed, or showed only limited variations after exposure to germ cells. Six clones for which a greater than 2-fold increase and one for which a 4-fold decrease in activity were registered were kept for further studies. Induction factors initially observed were limited to a 3-to 8-fold range (Fig. 1B). It was, however, subsequently found that the culture conditions in 96-well plates, convenient for large-scale screening, are in fact less than optimal. Greater variations in enzymatic activity were recorded (Fig. 2A) when cultures were performed under the conditions that we had independently determined as optimal for the recognition by 15P-1 of germ cells (Rassoulzadegan et al., 1993; Vincent et al., 1998). Important variables are the actively growing state of 15P-1, the culture substrate, glass being more efficient than plastic, and the way the initial mixtures of germ cells and 15P-1 cells are prepared (see Materials and Methods).

In order to ascertain that the whole procedure generated clonal derivatives, three isolates were subcloned by limiting dilution in a 96 well-plate, and for each one, five subclones were tested again for the effect of germ cells on β-galactosidase activity, with concordant results in each series (data not shown). Definitive proof of clonality was, however, provided by the structure of the transgene, once established by Southern blot analysis and 5′-RACE extension (see below), evidencing in each case a unique integration of the transgene.

Identification of the trapped cellular promoters was performed by 5′-RACE amplification, cloning and sequencing of the region 5′ of the β-geo splice acceptor in the fusion transcripts. Results for the seven clones are summarized in Table 1. The analysis was inconclusive for only one of them (3.3), for which only transgene sequences were repeatedly amplified, suggesting that insertion had occurred immediately 3′ of a cellular promoter. Among the other six genes, data banks searches showed one unknown gene and five whose sequences were either completely identical to a known mouse gene, or more than 90% identical to a human or a rat gene whose murine homologue was not present in the data base. Whenever the genomic structure of the gene was available, it was clear that integration of the provirus had occurred in the first intron of the gene. In all cases it resulted in a fused RNA in frame with the upstream exon of the gene. Subsequent studies were then focused on clone 11F7, in which the transgene was inserted in the Fra1 gene, of specific interest since modulation of expression of the AP1 complex would be expected to mediate secondary events in the Sertoli-germ cell interaction.

We determined the response of 11F7 cells to the addition of purified fractions of germ cells prepared by elutriation centrifugation (Meistrich, 1977). This technique allows essentially for the fractionation to a purity of 80 to 90% of three types of germ cells, the pachytene spermatocytes, round spermatids and elongated spermatids. Neither pachytene spermatocytes nor elongated spermatids had any effect on β-galactosidase synthesis, whereas exposure to round spermatids resulted in a 30-to 40-fold increase in activity (Fig. 2B). On the basis of this result, we searched for a possible effect of NGF, based on previous studies which identified NGF as a likely mediator in the function of the seminiferous epithelium. Produced by post-meiotic germ cells, NGF is assumed to interact with receptors on the surface of the Sertoli cells. RT-PCR determinations (not shown) confirmed that NGF mRNA was present in the round spermatid fractions. β-galactosidase activity in 11F7 extracts, was found to increase in a dose-dependent manner after overnight incubation in the presence of NGF (Fig. 2C). In situ staining for β-galactosidase activity evidenced a strongly positive reaction in 11F7 cells that had been incubated overnight in the presence of either total germ cells (50 germ cells per 11F7 cell) or NGF (100 ng/ml) (Fig. 2D). On the other hand, cell culture medium conditioned by overnight incubation of total germ cells had no effect on the level of β-galactosidase in 11F7 extracts (not shown). A likely, although trivial explanation is that this negative result is due to an insufficient concentration of NGF in the conditioned medium. It may also reflect a more complex situation, with factors secreted by other types of germ cells modulating the response to NGF of the Sertoli cell, a point that will clearly require further studies.

Further analysis demonstrated that the effect of NGF on the expression of the Fra1 promoter is indeed a property of Sertoli cells, and not a unique property of the gene trap clone 11F7, thereby validating the gene trap approach as a way to detect modulation of gene expression in the Sertoli-germ cell cross talk. Semi-quantitative RT-PCR determination in extracts prepared from 15P-1 cells and from freshly established primary Sertoli cell cultures with and without addition of the neurotrophin demonstrated in all cases an increased amount of Fra1 mRNA after overnight exposure to NGF (Fig. 3).

In fibroblasts, Fra1 expression induced by growth factors follows that of Fos after a delay of about 60 minutes (Cohen and Curran, 1988). Furthermore, NGF induces Fos expression in PC12 cells with the same kinetics as serum and growth factors (Cohen and Curran, 1988; Bartel et al., 1989). We were therefore led to ask (i) whether Fos is induced by NGF in Sertoli cells, and (ii) what are the time courses of Fra1 and Fos expression. As shown in Fig. 4, the answer to the first question was that Fos showed a discrete peak of expression within the first 1 hour after NGF induction. Fra1 showed a more complex profile, with a clearly biphasic kinetics (Fig. 4). A first burst of expression concomitant with Fos expression, about 1 hour after NGF addition, was followed by a return to the initial basal level of expression and a late increase to the higher values recorded in previous experiments after overnight exposure, starting at about 16 hours and peaking around 20 hours after NGF addition. By contrast, after serum induction in parallel 15P-1 cultures, the expected early induction of Fra1 was detected during the first hour, but RNA then remained at a low and roughly constant level for the following 24 hours (Fig. 4B). This delayed increase may result from a requirement for the intermediary activation of other gene(s). A clear prediction would be that protein synthesis inhibitors should prevent the late increase in Fra1 RNA, and indeed preliminary results indicate that addition of cycloheximide 15 minutes before that of NGF completely suppressed the increase in Fra1 RNA 24 hours later.

The two receptors p75^NTR and TrkA respond to NGF with different affinities and the concentrations used in our experiments were sufficient to activate both the low affinity p75^NTR and the high affinity TrkA. To determine which receptor was involved in Fra1 regulation, and in view of the somewhat controversial results reported in the literature, we first proceeded to an analysis of their expression in 15P-1 cells, primary Sertoli cell cultures and whole testis, by RT-PCR assays using oligonucleotide primers specific either for the tyrosine kinase domain common to the Trk family, or for the p75^NTR receptor (Fig. 5). Expression at messenger level of at least one of the Trk receptors was evidenced in whole testis RNA, in primary culture Sertoli cells and in 15P-1 cells. p75^NTR mRNA, on the other hand, was detectable neither in Sertoli nor in 15P-1 cells, although it was clearly present in total testis extracts. To further precise the Trk receptor(s) present in Sertoli cells, we designed couples of oligonucleotide primers specific for TrkA, TrkB and TrkC mRNAs, located in the 5′ parts of their respective sequences. Results indicated that the three genes are all expressed in the testis but that TrkA is the only one transcribed in Sertoli cells.

The conclusion that TrkA signaling was responsible for Fra1 induction was independently confirmed by the observation of high levels of Fra1 RNA in transfected cells expressing a constitutively activated form of the receptor. The Trk5 oncogenic mutant results from a deletion of 50 amino acid residues in the extracellular domain of TrkA (Coulier et al., 1990). Experiments were performed both on 15P-1 cells transiently expressing Trk5 and neo^r after transfection and on pools of clones selected in geneticin medium. RT-PCR assays demonstrated in all cases elevated levels of Fra1 expression in the absence of NGF stimulation (Fig. 6).

One of the main pathways of signal transduction by tyrosine kinase receptors is the ERK1-ERK2 MAP kinase kinase pathway, required for the induction by NGF of the differentiation of PC12 cells (Pang et al., 1995). We thus determined whether the ERK1-ERK2 kinases were activated by assaying the degree of phosphorylation of the two proteins after exposure to NGF (Fig. 7A). Western blot experiments were performed with antibodies specific to the phosphorylated forms of the proteins. A somewhat irregular background of phosphorylation was observed at time 0, clearly related with the fact that, unlike the standard fibroblast cell lines often used in this type of experiment, Sertoli cell cultures are not completely arrested in the absence of serum. In spite of that, an increase of the phosphorylated form was consistently observed upon NGF addition, with a peak about 30 minutes after addition of NGF. The same results were observed when freshly isolated Sertoli cells were used in the same conditions (Fig. 7B), except that the maximum of phosphorylation was reached after only ten minutes of treatment with NGF. This difference in kinetics between two widely different types of cells in culture could not be explained at this stage, but it was highly reproducible. These results were confirmed by using the MAP kinase kinase pathway inhibitor PD 098,059 (Alessi et al., 1995). Treatment of 15P-1 cells with the inhibitor 30 minutes before NGF addition to the culture medium completely suppressed Fra1 mRNA accumulation (Fig. 7C).

A primary aim of this work was to establish a convenient tool for the identification of some of the genes involved in a complex intercellular signalisation network such as the cross-talk between a Sertoli cell and its associated germ cells at their successive differentiation stages. This method relies on the fact that one of the partner cells is established as a differentiated cell line. A library of gene trap clones can then be generated, each one corresponding to the random integration of the βgeo reporter in a chromosomal ‘trapped’ locus, different in each clone. A simple enzymatic assay will then quantitatively reveal the activity of the promoter, and thus, its possible modulations under a set of predetermined experimental conditions. The method has obvious inborn limits. First, inducible genes will be identified only if they are already expressed at a basal level in the absence of stimulation. This basal level may, however, be quite low, and still, be sufficient to confer resistance to concentrations of geneticin not exceeding 200 μg/ml. A second limit is the work load involved, which limits the number of genes analyzed. Screening several hundred, if not thousands of genes is possible, but these numbers remain far below the range explored by the microarray technology (Lipshutz et al., 1999). Beside a much lower cost, however, the interest of the gene trap approach is to provide us from the start with useful ‘knocked in reporters’ for the genes of interest. It is also important that, unlike other methods (microarray analysis, substractive hybridization, differential display), it does not bias towards genes with high expression levels. A third series of questions may be raised regarding the possibility of artefacts. No cell line will ever be a totally faithful reproduction of the physiology of a given cell, due to the genetic alterations associated with immortalization as well as to the rather alien environment of the cell culture, but some of them like the 15P-1 line used in the present work, maintain a significant part of the in vivo phenotype. Subsequent studies, either in the mouse or on primary cultures of freshly explanted cells, should then validate the conclusions drawn from the analysis of the gene trap clones.

The exon trapping method had initially been developed to establish patterns of gene expression during embryonic development (Friedrich and Soriano, 1991). Here we show that the same method can be applied to differentiated cells. By using only a fraction of the library, a series of seven genes were identified whose expression is either up- or down-regulated during an overnight coculture of Sertoli cells with total germ cells. Several of them appear of interest for further studies, as their variations can be related to a biological feature of germinal maturation. For instance, the induction of a non muscle myosin gene may be related with the movement of the maturing germ cells, in all likelihood driven by the Sertoli cell, from the periphery to the central part of the tubule, and at some stages in the opposite direction. Similarly, modulation of Type III TGFβ receptor expression may have a functional significance. A recent study (Lewis et al., 2000) established a possible role of this receptor in the sensitivity of the cell to inhibin, itself a regulator of Sertoli cell proliferation (Matzuk et al., 1992; Lopez et al., 1999).

As a first step, and with the primary aim to validate the gene trap approach, we have focused our studies on the induction of Fra1 and Fos, two obvious candidates for a central role in transcriptional regulations. We determined that induction of Fra1 in Sertoli cells was triggered by the post-meiotic round spermatids, with NGF as a potential mediator. Function of NGF in the testis has been previously indicated (Lonnerberg et al., 1992; Parvinen et al., 1992; Chen et al., 1997). Induction by NGF of the production by 15P-1 cells of antimicrobial proteins of the defensin family (Grandjean et al., 1997) had established the presence of a receptor, now identified as TrkA, whose role in the induction of Fra1 was further confirmed by its constitutive high level of expression in cells expressing the activated oncogenic derivative Trk5.

In Sertoli cells as in the neuronal cell line PC12, TrkA activation leads to the downstream activation of the ERK1-ERK2 kinase kinases (Pang et al., 1995). The resulting time course of Fra1 activation appears, however, different in Sertoli cells from that in neuronal and fibroblast cells. In serum-stimulated fibroblasts, as well as in NGF-treated PC12 cells, Fra1 induction is an early event, although it is delayed by 30 to 60 minutes relative to Fos induction. In Sertoli cells, we also observed an early induction to a relatively modest level, peaking 1 hour after the addition of NGF, and concomitant with Fos induction. More strikingly, a second wave of transcription, of a larger amplitude, begins between 12 and 16 hours, reaching a maximum at about 20 hours. This late increase was not seen after serum induction in 15P-1 cells (Fig. 4), and it has not been reported in other cell types, after induction either by serum (fibroblasts, PC12) or by NGF (PC12). These features clearly differentiate the effects of NGF on 15P-1 Sertoli cells from those observed after serum stimulation in various cell lines, including 15P-1 (Fig. 4B) and mouse fibroblasts (data not shown). Establishing whether this delayed effect is due to a cascade of induction with the intermediary activation of other gene(s) will be the subject of further studies. The ‘knock in’ βgeo reporter provides a convenient assay for the expression of Fra1 and a tool for subsequent studies on the transduction of the NGF signal in Sertoli cells. Further studies will then be required to evaluate the physiological function of NGF during germinal differentiation in vivo. Establishing the patterns of expression of the neurotrophin in Sertoli cells during germinal differentiation will require a precise in situ analysis. One element suggestive of a physiological function of Fra1 in the control of spermatogenesis is the specific expression pattern observed in the testis of a seasonal breeder, the European red fox (Cohen et al., 1993). In the mouse, in situ analysis of Fra-1 expression in the adult testis is made difficult by the complex and compact structure of the seminiferous epithelium. Preliminary data (not shown) on the first wave of spermatogenesis in the testis of the young mouse indicated that Fra-1 transcription is not turned on before its completion. This result is consistent with induction in Sertoli cells in response to post-meiotic stimuli. It is clear that definite answers on the in vivo function of NGF will depend on the availability of a targeted mutation of the gene. Since, however, the Ngf^– homozygous genotype is lethal at an early developmental stage (Crowley et al., 1994), a conditional mutation is required, which, ideally, would affect the gene exclusively during spermatogenesis. One possible way is offered by the transgenic ‘TAMERE’ mice which express the Cre recombinase during meiosis (Hérault et al., 1998; Vidal et al., 1998). This strategy, however, requires the Ngf locus to be first modified to include properly positioned copies of the target sequence of the recombinase (LoxP sites).

We thank Dr Dionisio Martin-Zanca for the generous gift of Trk5 mutant DNA, and Dr Barry Rosen for helpful comments and editorial help. The expert technical assistance of Mireille Cutajar, Yann Fantei and Christel Lust is gratefully acknowledged. This work was made possible by grants to F.C. from the Association pour la Recherche sur le Cancer, France.