Retinoic Acid Receptor Alpha Is Essential in Postnatal Sertoli Cells but Not in Germ Cells

Retinoic acid signaling is indispensable for the completion of spermatogenesis. It is known that loss of retinoic acid nuclear receptor alpha (RARA) induces male sterility due to seminiferous epithelium degeneration. Initial genetic studies established that RARA acts in Sertoli cells, but a recent paper proposed that RARA is also instrumental in germ cells. In the present study, we have re-assessed the function of RARA in germ cells by genetically ablating the Rara gene in spermatogonia and their progenies using a cell-specific conditional mutagenesis approach. We show that loss of Rara in postnatal male germ cells does not alter the histology of the seminiferous epithelium. Furthermore, RARA-deficient germ cells differentiate normally and give rise to normal, living pups. This establishes that RARA plays no crucial role in germ cells. We also tested whether RARA is required in Sertoli cells during the fetal period or after birth. For this purpose, we deleted the Rara gene in Sertoli cells at postnatal day 15 (PN15), i.e., after the onset of the first spermatogenic wave. To do so, we used temporally controlled cell-specific mutagenesis. By comparing the testis phenotypes generated when Rara is lost either at PN15 or at embryonic day 13, we show that RARA exerts all of its functions in Sertoli cells not at the fetal stage but from puberty.


Introduction
Spermatogenesis, the process that allows for the formation of spermatozoa, consists of three distinct phases: (i) the proliferative phase, during which the spermatogonia stem cells divide and differentiate to maintain both germ cell production and stem cell renewal; (ii) the meiotic phase, during which the spermatocytes undergo two consecutive divisions to produce haploid spermatids; and (iii) the spermiogenesis phase, during which the spermatids differentiate into spermatozoa. Occurring within the seminiferous epithelium of the testis, spermatogenesis is supported by somatic Sertoli cells. Both germ cells and Sertoli cells are impacted upon changes in vitamin A metabolism or its signaling pathway ( [1,2] and references therein). For instance, vitamin A deficiency in rodents induces an arrest of spermatogonia differentiation, resulting in the progressive depletion of more differentiated germ cells. Additionally, Sertoli cells lose their cyclical changes in morphology and gene expression [3,4]. Importantly, administration of all-trans retinoic acid (ATRA) to vitamin Tg(Stra8-cre) 1Reb and Tg(Sox9-cre/ER T2 ) 1Msan ATTTGCCTGCATTACCGGTC ATCAACGTTTTCTTTTCGGA 350 Gt(ROSA)26Sor tm1(EYFP)Cos (excised allele) AAGGGAGCTGCAGTGGAGTA TGGTGCAGATGAACTTCAGG 620 * +, L2, and ∆ correspond to wild-type, floxed, and excised allele of Rara gene, respectively.
In Stra8-Cre Tg/0 mice, the recombinase is first expressed at PN3 in a subset of undifferentiated spermatogonia and in all differentiating spermatogonia [20,23]. Both mutant and control males (bearing and free of Tg(Stra8-cre) 1Reb transgene, respectively) were generated from the same breeding pairs, in the same litters.
In Sox9-CreER T2 Tg/0 mice, the recombinase is expressed in SOX9-positive cells throughout development and adulthood [21], including the postnatal Sertoli cells [24], where it becomes functional only upon TAM treatment. Experimental males (bearing and free of Tg(Sox9-cre/ER T2 ) 1Msan transgene) were generated from the same breeding pairs, in the same litters. Pups were administered TAM (50 mg/kg body weight) by intraperitoneal injections, three times every second day, between PN5 and PN15. Both those bearing and those free of the transgene received TAM, precluding therefore that the observed abnormalities at later stages are due to this drug [25]. TAM (T5648, Sigma-Aldrich, 38070 Saint-Quentin-Fallavier, France) was dissolved in ethanol at a concentration of 100 mg/mL and further diluted in sunflower oil to a concentration of 10 mg/mL before administration.

Histology, Stage Frequencies, and Synchronization Factor
For histology, testis and cauda epididymis samples were fixed in Bouin's fluid for 16 h and embedded in paraffin. Histological sections (5 µm thick) were stained with hematoxylin and eosin (HE) or with periodic acid-Schiff (PAS). Testes from at least three mice per genotype were analyzed at each time point.
In the seminiferous epithelium, the different generations of germ cells form cellular associations of fixed composition (called epithelial stages). Twelve epithelial stages (I-XII) can be identified in the mouse [26,27]. The determination of epithelial stages' frequencies was performed on PAS-stained sections from 3-to 9-month-old mutants and control mice (bearing and free of Tg(Stra8-cre) 1Reb transgene, respectively). Sections were scanned in a Hamamatsu NanoZoomer 2.0-HT scanner, and 200 to 300 seminiferous tubules per mouse were analyzed using the NDP.view2 software. The identifying features were as follows: stages I-III, small unstained or weakly stained proacrosomal granule; stage IV, acrosomic granule forming an indentation and beginning to flatten; stages V-VI, acrosomic system forming a straight PAS-positive line, with the angle subtended by the acrosome being lower than or equal to 120; stages VII-VIII, acrosome forming a cap that covers more than one-third of the nucleus; stage IX, oblong spermatid nuclei; stage X, elongating spermatid head displaying a sharp angle between its ventral and caudal surface; stage XI, spermatid nucleus thinner, more elongated and stained more intensely; stage XII, presence of meiotic metaphases, meiotic anaphases, and/or secondary spermatocytes. Testes from 10 controls, 3 heterozygotes, and 3 of each mutant type were analyzed. One entire testis cross-section was scanned for each animal.
The comparison of the stage frequencies was assessed using the synchronization factor as described by van Beek and Meistrich [28], for which we developed a program [29]. Statistical analysis was performed using a one-tailed Student's t-test.

Immunohistochemistry (IHC) and In Situ Hybridization (ISH)
Testes were fixed by intracardiac perfusion of ice-cold 4% (w/v) paraformaldehyde dissolved in phosphate-buffered saline (PBS) and then kept in the same fixative overnight at 4 • C, washed in PBS, dehydrated, and embedded in paraffin. Sections (5 µm thick) were stored at 4 • C. To assess the specificity of the anti-RARA antibodies or the antisense Rara probe, testes from Rara Ser−/− and/or Rara-knockout mutants [12,19] were used as "negative" controls.
For IHC, sections were processed for antigen retrieval for 20 min at 120 • C in a pressure cooker. They were rinsed in PBS and then incubated in a humidified chamber for 16 h at 4 • C, with the primary antibodies diluted in PBS containing 0.1% (v/v) Tween 20 (PBST). After rinsing in PBST (three times for 3 min each), detection of the bound primary antibodies was performed for 45 min at room temperature in a humidified chamber using dye-conjugated secondary antibodies. The sections were then counterstained with 4 ,6diamidino-2-phenylindole (DAPI) to label nuclei and with Alexa Fluor 488-conjugated peanut agglutinin (lectin from Arachis hypogaea) to label the acrosomal systems. Antibodies used in the IHC, working dilution, and related procedure for antigen retrieval are listed in Table 2. Testes from 4 mice per genotype were analyzed at each time point (PN15, PN21, and 4 months old). IHC experiments were performed in triplicate. Protocol 1: 10 mM sodium citrate buffer (pH = 6.0); Protocol 2: 10 mM tris base, 1 mM EDTA, and 0.05% (v/v) Tween 20 (pH = 9.0); Protocol 3: 0.05% citraconic anhydride buffer (pH = 7.4). * These antibodies recognize epitopes which are not RARA (see below); ** these antibodies recognize epitopes which are not RARA, as illustrated in Figure S1.
For ISH, the BaseScope v2-RED Reagent kit was used according to the manufacturer's instructions (Advanced Cell Diagnostics, ref. 323900). Briefly, deparaffinized sections were treated with hydrogen peroxide for 10 min, washed in distilled water, and boiled at 100 • C for 10 min in the target retrieval reagent. Protease IV was then applied for 15  were then washed in the appropriate buffer and subjected to a series of signal amplifications (AMP1 to AMP8). Hybridization signals were detected using the chromogenic Fast RED-B/Fast RED-A reagent. The sections were counterstained for 3 min with 12.5% (v/v) Harris hematoxylin, and then for 20 s in 0.02% (v/v) ammonium hydroxide, both diluted in distilled water. Four samples per genotype at PN15 were analyzed in duplicate.

Counts of Sertoli Cells Expressing RARA
Sertoli cell counts were performed on at least 10 seminiferous tubule cross-sections from the testes of PN21 TAM-treated mice (bearing and free of Tg(Sox9-cre/ER T2 ) 1Msan transgene). Testes from 4 mice per genotype at PN21 were analyzed. The proportion of RARA-expressing Sertoli cells was expressed as a percentage of GATA4-positive Sertoli cells. Statistical analysis was performed using a one-tailed Student's t-test, having verified for equal variance after arcsine transformation of the percentages.

Spermatogenesis Is Not Altered in Males Lacking RARA in Germ Cells
In the first approach ( Figure 1A), we bred Stra8-Cre Tg/0 females with Rara L2/L2 males. Then, Stra8-Cre Tg/0 /Rara +/L2 females were mated with Rara L2/L2 /R26R-EYFP Tg/0 males to generate the F2 offspring. This was performed because the Tg(Stra8-cre)1Reb transgene is not functional in the ovary [20], thereby avoiding deletion of the floxed alleles in the germ cells and transmission of null, excised (∆) alleles of Rara to the F2 offspring [31]. Genotyping of these F2 mice showed that the L2 alleles were detected in DNA extracted from tail biopsies of Tg(Stra8-cre)1Reb-positive mice, while the excised (∆) allele of Rara was not detected, as expected ( Figure 1B). Loss of RARA in germ cells could not be evidenced by IHC because (i) all the commercially available antibodies directed against RARA that we tested lack specificity ( Figure 2 and Figure S1) and (ii) the RPalpha(F) antibody [30] displays low sensitivity and did not allow for detecting RARA in germ cells [8]. Thus, Cre-mediated ablation of the Rara gene in germ cells of 3-month-old males was assessed by visualizing the expression of the excised Gt(ROSA)26Sor tm1(EYFP)Cos reporter transgene on testis sections. In agreement with previous reports [2,16,17], we found that EYFP was robustly expressed in all germ cells of Stra8-Cre Tg/0 /Rara L2/L2 /R26R-EYFP Tg/0 males, from the pachytene spermatocyte to the elongated spermatid stages. Expression of EYFP was weaker in the earlier stages of germ cell differentiation ( Figure 1C). Moreover, EYFP was not detected in somatic cells, as expected. Assuming that the expression of the reporter faithfully reflects the excision of Rara alleles, loss of RARA occurred in all germ cells. Our breading procedure yielded both control (Rara L2/L2 /R26R-EYFP Tg/0 ) and mutant (Stra8-Cre Tg/0 /Rara L2/L2 /R26R-EYFP Tg/0 ) males, hereafter referred to as Rara Germ+/+ and Rara Germ−/− mice, respectively.

Figure 2.
Two commercially available antibodies directed against RARA lacked specificity in IHC experiments. IHC analysis on histological sections of testes from 8-week-old wild-type (WT) and Rara-knockout mice (Rara-null). (A) The sc-551 antibody recognized an epitope (red signal) which was distinct from RARA because it was similarly detected in the cytoplasm of Sertoli cells of WT and Rara-null mice. Sertoli cell nuclei were identified by using an antibody directed against the GATA4 transcription factor (green signal). (B) The sc-525796 antibody recognized a nuclear epitope (red signal) which was distinct from RARA because it was detected in the nuclei of germ cells (arrowheads) and Sertoli cells (arrows) of WT and Rara-null mice. The right panels show enlargements of the boxes in the left panels. Nuclei were counterstained with DAPI (blue signal). Star indicates interstitial signal due to the use of a mouse monoclonal antibody. Scale bar: 80 μm for left panels; 20 μm for right panels.
Quite surprisingly, in view of the pathological phenotype described in Rara-cKO mutants [16], Rara Germ−/− males (n = 3) were fertile. When mated with wild-type females, they sired 10 litters for a total of 78 pups. In comparison, Rara Germ+/+ males (n = 11) sired 21 litters for a total of 146 pups. This was not statistically different from the reproductive capabilities of the Rara Germ−/− males (p = 0.25). Testis sections of 3-month-old Rara Germ+/+ (n = 10) and Rara Germ−/− males (n = 3) were stained using PAS and analyzed. In both Rara Germ+/+ and Rara Germ−/− males, the cellular associations characterizing the twelve stages of the seminiferous epithelium cycle (I to XII) were readily identified. Moreover, in agreement with the fertility of Rara Germ−/− males, their seminiferous epithelium also displayed normal histology ( Figure 1D), with the twelve stages of the cycle present in normal proportions ( Figure 3A). Accordingly, the mean synchronization factors were similar ( Figure 3B). These data indicate that Rara Germ−/− males display normal spermatogenesis, as opposed to Rara-cKO mutants [16].  The lack of abnormalities in the testis of Rara Germ−/− males compared to Rara-cKO males [16] led to questioning the efficacy of Rara gene excision in germ cells of Rara Germ−/− males [31]. Since the recombination of a given floxed sequence by Cre depends on its genomic context [32], one could figure out that the Gt(ROSA)26Sor tm1(EYFP)Cos transgene was accessible for Cre-mediated recombination, while the Rara locus was not. Thus, in the second approach (Figure 4A), we bred Stra8-Cre Tg/0 /Rara +/L2 F1 males with Rara L2/L2 /R26R-EYFP Tg/0 females in order to transmit one excised (∆) allele of Rara to the F2 offspring. In fact, the sperm of Stra8-Cre Tg/0 /Rara +/L2 F1 males transmitted either the Rara + or the Rara ∆ allele because of the expression of Cre in their germ cells thanks to the Tg(Stra8-cre)1Reb transgene. This breeding scheme reduced the amount of floxed Rara alleles that Cre had to excise in the germ cells of F2 males and additionally presented the advantage of being identical to the one set up by Peer et al. in their study [16]. Genotyping of tail biopsies from F2 males showed that the excised (∆) allele of Rara was present in some animals, even in absence of the Tg(Stra8-cre)1Reb transgene ( Figure 4B). This observation confirmed that excision of Rara had occurred in the sperm of the Stra8-Cre Tg/0 /Rara +/L2 F1 father, who transmitted a null (∆) allele to its progeny. The Stra8-Cre Tg/0 /Rara ∆/L2 /R26R-EYFP Tg/0 F2 males (hereafter named Rara ∆/Germ− ) were considered mutants (i.e., lacking RARA in germ cells). Importantly, F2 males without the Tg(Stra8-cre)1Reb transgene were obtained from the same breeding pairs and were used as controls (hereafter named Rara ∆/L2 males).
Here again, Rara ∆/Germ− males (n = 5) were fertile. When mated with wild-type females, they sired 14 litters, with an average of 7.6 pups per litter, a proportion similar to the one obtained from control males (see above). Most importantly, out of the 185 pups born from this breeding, 179 (i.e.,~94%) were Rara ∆/+ and 6 (i.e., 6%) were Rara L2/+ . This finding indicates that recombination of the Rara locus by Cre was efficient in almost 94% of the germ cells produced by Rara ∆/Germ− males. In agreement with this finding, a Mendelian proportion of pups sired by Rara ∆/Germ− males were EYFP-positive when observed under ultraviolet light illumination ( Figure 4C). Thus, even though we could not evidence RARA loss by IHC (Figure 2 and Figure S1), we concluded that Rara ∆/Germ− mice were actually deprived of RARA in 94% of their germ cells. Next, we carefully analyzed the histology of the testes from 9-month-old Rara ∆/Germ− males (n = 3) ( Figure 4D). The full representation of the germ cell associations that are characteristic of the 12 stages of the seminiferous epithelium cycle was present in similar proportions in testes from Rara ∆/Germ− and Rara ∆/L2 males ( Figure 3A). Their mean synchronization factor was accordingly similar to that of controls ( Figure 3B). Altogether, these experiments show that Rara is efficiently inactivated in germ cells without causing any defect in spermatogenesis.
This finding clearly contradicts the proposal that RARA plays an essential role in germ cells [16]. While looking for an explanation for such a discrepancy, we noticed that the loxP-flanked allele of Rara (Rara fl allele), used to generate Rara-cKO males, still contains a neomycin resistance mini-gene in intron #3, the orientation of which is opposite to that of the Rara gene [16]. Often, the insertion of a mini-gene in an intron interferes with the normal functioning of the targeted allele, generating a so-called "hypomorphic" allele driving diminished gene expression (reviewed in [33,34]). One can easily conceive that the expression of Rara driven by the Rara fl allele is not optimal, a possibility that cannot be assessed because males without Rara fl alleles were used as controls, unfortunately [16]. If the Rara fl allele is actually hypomorphic, one can hypothesize that Rara-cKO males not only lack RARA in germ cells, but they also suffer from RARA insufficiency in all other cell types, including Sertoli cells. In keeping with this hypothesis, it is worth noting that males lacking RARA in Sertoli cells opportunely display vacuolization of their seminiferous epithelium and sloughing of immature germ cells [12], just like Rara-cKO males do [16].

RARA Exerts Most, If Not All, of Its Functions on Spermatogenesis in Postnatal Sertoli Cells
Aside from germ cells, RARA is strongly expressed in Sertoli cells [8,12]. The fate of these cells is established during embryonic development, at the time of sex determination. In the mouse, this occurs at around E12.5 [35]. Then, Sertoli cells proliferate prior to puberty and terminally differentiate to become postmitotic at puberty, at around PN15 [36]. RARA expression in Sertoli cells starts as early as E13.5 [37] and persists throughout adulthood [8]. Ablation of Rara in Sertoli cells from E13.5 (Rara Ser−/− mutants) [12] does not allow for discriminating the functions exerted by RARA during the fetal period and at later stages of development. It was recently proposed that both testosterone production by Leydig cells and proliferation of Sertoli cells are controlled by RARA during the fetal period [37]. As some of the defects displayed by Rara Ser−/− mutants (e.g., vacuolization of the seminiferous epithelium, loss of germ cells) may have arisen as a consequence of altered Sertoli cell proliferation or reduced testosterone synthesis by Leydig cells during fetal life, determining whether prenatal expression of RARA in Sertoli cell is sufficient to ensure normal spermatogenesis remained an important question.
In this context, we aimed to assess the role of RARA in spermatogenesis in postnatal Sertoli cells by inducing ablation of Rara using a tamoxifen-inducible Cre-loxP mutagenesis system. We chose to ablate the Rara gene before Sertoli cells became postmitotic but after the onset of the first spermatogenic wave (i.e., between PN5 and PN15). To do so, Rara L2/L2 mice were crossed with Sox9-CreER T2 Tg/0 transgenics, thereby generating both Rara L2/L2 and Sox9-CreER T2 Tg/0 /Rara L2/L2 experimental males from the same breeding pairs ( Figure S2A). They were all treated with TAM, as recommended [25], three times between PN5 and PN15 and analyzed at several time points as indicated ( Figure 5A).
To evaluate if TAM-treated Sox9-CreER T2 Tg/0 /Rara L2/L2 males actually lacked RARA in their Sertoli cells, we first performed ISH analyses. They showed that Rara mRNA could not be detected in their testes (n = 4) at PN15 (i.e., between 4 and 6 days after the last TAM administration). The number of dots detected in this situation was almost identical to that observed in the testes of Rara Ser−/− mutants ( Figure 5B). Accordingly, TAM-treated Sox9-CreER T2 Tg/0 /Rara L2/L2 males are hereafter referred to as Rara Ser−/−PN15 mice. Further IHC analyses were performed combining the use of anti-RARA [8,30] and anti-GATA4 antibodies for detecting Sertoli cells. In all the Rara Ser−/−PN15 mutants analyzed at PN15 (n = 4) and PN21 (n = 4), the seminiferous tubules contained many Sertoli cells devoid of RARA, but some others were still expressing it ( Figure 5C and Figure S2B). Quantification on histological sections revealed that 35 ± 11% of Sertoli cells remained RARA-positive at PN21 ( Figure S2C). Importantly, these Sertoli cells were always scattered across the seminiferous tubules, with no case where all the Sertoli cells of a given tubule still expressed RARA. Thus, ablation of the Rara alleles occurred in about 65% of the Sertoli cells from the Rara Ser−/−PN15 mutants, and no seminiferous tubule escaped Cre-mediated recombination. The reason RARA remains present in some Sertoli cells is unclear. It is possible that the protein half-life is long enough to make RARA persist for a few hours, even in the absence of its mRNA [38].
Most importantly, despite the persistence of RARA in some Sertoli cells, all the Rara Ser−/−PN15 mutants were infertile (n = 19). To uncover the reason for their infertility, we performed histological analyses at 3.5 months after the last TAM treatment ( Figure 6). Testis sections from these males (n = 12) showed that mature spermatids failed to align at the luminal side of the seminiferous epithelium and were often retained within the epithelium (insets in Figure 6B,C). Many seminiferous tubule sections, distributed through-out the testis, displayed large vacuoles ( Figure 6E,F) or were depleted of germ cells ( Figure 6F). Accordingly, the caudal epididymis contained few spermatozoa but an unusually high number of round immature germ cells, suggesting extensive desquamation of cells from the seminiferous epithelium ( Figure 6H,I). The degree of vacuolation and tubule degeneration varied between Rara Ser−/−PN15 mutants. About 80% of them (i.e., 10 out of 12) were considered severely affected; more than half of their seminiferous tubule sections contained more than two vacuoles ( Figure 6C,F,I). The remaining 20% were mildly affected; about one-fifth of their seminiferous tubules sections contained one or two vacuoles ( Figure 6B,E,H). These alterations were not regionalized to a particular area but were distributed across the testes. Most importantly, despite the persistence of RARA in some Sertoli cells, all the Rara Ser−/−PN15 mutants were infertile (n = 19). To uncover the reason for their infertility, we performed histological analyses at 3.5 months after the last TAM treatment ( Figure 6). Testis sections from these males (n = 12) showed that mature spermatids failed to align at the luminal side of the seminiferous epithelium and were often retained within the epithelium (insets in Figure 6B,C). Many seminiferous tubule sections, distributed throughout the testis, displayed large vacuoles ( Figure 6E,F) or were depleted of germ cells ( Figure  6F). Accordingly, the caudal epididymis contained few spermatozoa but an unusually high number of round immature germ cells, suggesting extensive desquamation of cells from the seminiferous epithelium ( Figure 6H,I). The degree of vacuolation and tubule degeneration varied between Rara Ser−/−PN15 mutants. About 80% of them (i.e., 10 out of 12) were considered severely affected; more than half of their seminiferous tubule sections contained more than two vacuoles ( Figure 6C,F,I). The remaining 20% were mildly affected; about one-fifth of their seminiferous tubules sections contained one or two Interestingly, the histological abnormalities displayed by our Rara Ser−/−PN15 mutants are similar in all respects to those we described in Rara Ser−/− mice, including the variability between individuals [12]. This finding indicates that (i) heterogeneity in the severity of the abnormalities is not linked to the proportion of Sertoli cells that have lost RARA, and (ii) RARA produced by Sertoli cells until PN15 is dispensable for the proper development of the seminiferous epithelium. In this context, our study casts doubt on the proposal that RARA is instrumental to Sertoli cell proliferation during the fetal period [37]. vacuoles ( Figure 6B,E,H). These alterations were not regionalized to a particular area but were distributed across the testes. Interestingly, the histological abnormalities displayed by our Rara Ser−/−PN15 mutants are similar in all respects to those we described in Rara Ser−/− mice, including the variability between individuals [12]. This finding indicates that (i) heterogeneity in the severity of the abnormalities is not linked to the proportion of Sertoli cells that have lost RARA, and (ii) RARA produced by Sertoli cells until PN15 is dispensable for the proper development of the seminiferous epithelium. In this context, our study casts doubt on the proposal that RARA is instrumental to Sertoli cell proliferation during the fetal period [37].
Several genes normally display a periodical pattern of expression in Sertoli cells at given stages of the seminiferous epithelium cycle [39,40]. In the present study, we investigated the periodical expression of selected genes in Sertoli cells through IHC analysis of 4-month-old testes using anti-GATA-1 [41] and anti-androgen receptor (AR) [42] antibodies. As expected, GATA1 expression in Sertoli cells was high at stage VII and very low at stage XII of the seminiferous epithelium cycle in control testes ( Figure 7A). In Rara Ser−/−PN15 Several genes normally display a periodical pattern of expression in Sertoli cells at given stages of the seminiferous epithelium cycle [39,40]. In the present study, we investigated the periodical expression of selected genes in Sertoli cells through IHC analysis of 4-month-old testes using anti-GATA-1 [41] and anti-androgen receptor (AR) [42] antibodies. As expected, GATA1 expression in Sertoli cells was high at stage VII and very low at stage XII of the seminiferous epithelium cycle in control testes ( Figure 7A). In Rara Ser−/−PN15 mutants, GATA1 was detected in Sertoli cells at all stages of the seminiferous epithelium cycle ( Figure 7B). Along the same lines, AR expression was robust in Sertoli cells at stage V and weak at stage XII of the control testes ( Figure 7C), but it was high all along the cycle of the seminiferous epithelium in Rara Ser−/−PN15 mutants ( Figure 7D). We conclude from these experiments that the loss of RARA in Sertoli cells during puberty abrogates their periodical expression of GATA1 and AR. Interestingly, this occurs even though RARA appears not lost in all Sertoli cells of a given seminiferous tubule. As a similar phenomenon is observed in a germ cell-deficient environment [1], it is possible that abrogation of the periodical expression of genes in Sertoli cells of Rara Ser−/−PN15 mutants is linked to their loss of germ cells. Our present data thus suggest that RARA maintains the periodical activity of Sertoli cells rather than initiating it during fetal development, as we previously proposed [12]. cycle ( Figure 7B). Along the same lines, AR expression was robust in Sertoli cells at stage V and weak at stage XII of the control testes ( Figure 7C), but it was high all along the cycle of the seminiferous epithelium in Rara Ser−/−PN15 mutants ( Figure 7D). We conclude from these experiments that the loss of RARA in Sertoli cells during puberty abrogates their periodical expression of GATA1 and AR. Interestingly, this occurs even though RARA appears not lost in all Sertoli cells of a given seminiferous tubule. As a similar phenomenon is observed in a germ cell-deficient environment [1], it is possible that abrogation of the periodical expression of genes in Sertoli cells of Rara Ser−/−PN15 mutants is linked to their loss of germ cells. Our present data thus suggest that RARA maintains the periodical activity of Sertoli cells rather than initiating it during fetal development, as we previously proposed [12]. In the control, GATA1 expression (red signal) was strong in Sertoli cell nuclei at stage VII and at the background level at stage XII of the seminiferous epithelium cycle. In the Rara Ser−/−PN15 testis, GATA1 was uniformly expressed at all epithelial stages. Sertoli cell nuclei were identified using an antibody directed against SOX9 (green signal in Sertoli cells). (C,D) In the control, AR expression (red signal) was strong in Sertoli cell nuclei at stage V and weak at stage XII. In the Rara Ser−/−PN15 testis, AR was uniformly expressed at all epithelial stages. Nuclei were counterstained by DAPI (blue signal). In (A-D), the acrosomal system is identified by Alexa Fluor 488conjugated peanut agglutinin (green signal), allowing proper staging. The dotted lines indicate the periphery of seminiferous tubules. Legend: E, elongated spermatids; M, spermatocytes in metaphase; PR and P, preleptotene and pachytene spermatocytes, respectively; R, round spermatids; S, Sertoli cells; S2, type 2 spermatocytes; Z, zygotene spermatocytes. Roman numerals designate stages of the seminiferous epithelium cycle. Scale bar: 70 μm. In the control, GATA1 expression (red signal) was strong in Sertoli cell nuclei at stage VII and at the background level at stage XII of the seminiferous epithelium cycle. In the Rara Ser−/−PN15 testis, GATA1 was uniformly expressed at all epithelial stages. Sertoli cell nuclei were identified using an antibody directed against SOX9 (green signal in Sertoli cells). (C,D) In the control, AR expression (red signal) was strong in Sertoli cell nuclei at stage V and weak at stage XII. In the Rara Ser−/−PN15 testis, AR was uniformly expressed at all epithelial stages. Nuclei were counterstained by DAPI (blue signal). In (A-D), the acrosomal system is identified by Alexa Fluor 488-conjugated peanut agglutinin (green signal), allowing proper staging. The dotted lines indicate the periphery of seminiferous tubules. Legend: E, elongated spermatids; M, spermatocytes in metaphase; PR and P, preleptotene and pachytene spermatocytes, respectively; R, round spermatids; S, Sertoli cells; S2, type 2 spermatocytes; Z, zygotene spermatocytes. Roman numerals designate stages of the seminiferous epithelium cycle. Scale bar: 70 µm.

Conclusions
RARs are nuclear transcription factors that, once activated by ATRA, regulate the expression of target genes, which in turn control various processes, among which is spermatogenesis. As Rara-knockout mice display a pathological testis phenotype and RARA is expressed in both somatic and germ cells of the testis, a strong research effort has been directed towards the identification of the cell types in which RARA is operational [12,13,16,17,43]. Our present study adds to this effort. Firstly, by showing that ablation of RARA in germ cells has no effects on testis integrity and functioning, it invalidates the recent proposal that RARA has dual functions in germ cells, balancing proliferation and differentiation of spermatogonia and controlling genome integrity during meiosis. Secondly, it shows that ablation of RARA in Sertoli cells, starting from PN15, recapitulates the complete set of abnormalities induced by knocking out Rara in all cells. Thus, it demonstrates that RARA exerts all its functions in Sertoli cells only, and at puberty but not at fetal stages. Although being a challenging and difficult task, it is now necessary to reveal more specifically the gene networks and mechanisms that RARA governs in Sertoli cells to allow proper spermatogenesis. Animal models with controlled ablation of Rara in Sertoli cells as we presented here could be useful in such future studies.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cells11050891/s1. Figure S1: Four commercially available antibodies directed against RARA recognize epitopes which are not RARA and are therefore not suitable for IHC experiments. Figure S2: Generation of mice in which Rara deletion is induced in Sertoli cells after birth, and quantitative assessment of RARA loss.  Institutional Review Board Statement: The animal study protocol was approved by the Institutional Ethics Committee of the IGBMC (protocol code APAFIS#28483-2020120115253832).