Abstract Study question Is all-trans retinoic acid (ATRA) instrumental to male germ cell differentiation in the mouse beyond the A1 spermatogonia cell stage? Summary answer While ATRA is useless to meiosis initiation, it is crucial to allow germ cells progressing beyond the mid-pachytene stage of meiosis. What is known already In mammals, ATRA is instrumental to spermatogenesis. It is synthesized by retinaldehyde dehydrogenases (RALDH), and then it binds to and activates nuclear receptors (RAR) either within the ATRA-synthesizing cells or in the neighboring cells. We established previously that ATRA synthesized by Sertoli cells initiates differentiation of A aligned into A1 spermatogonia during the 1st spermatogenic cycle. A controversy exists as to the requirement of ATRA in meiosis and spermiogenesis initiations. Some authors propose that ATRA is indispensable to both, while others claim that ATRA is useless to meiotic initiation but essential to spermiogenesis. Study design, size, duration In order to determine the role of ATRA in meiosis and spermiogenesis, we have generated mutant mice lacking all retinaldehyde dehydrogenase (RALDH) activities either in Sertoli cells or in the whole seminiferous epithelium (i.e., in both Sertoli and germ cells). We then compared in these two types of mutants the fate of spermatocytes generated during the first wave of spermatogenesis, following A1 spermatogonia differentiation at post-natal day 4. Participants/materials, setting, methods To generate these mutants, transgenic mice in which the Cre recombinase expression is driven by Stra8 or Amh promoters (active in undifferentiated spermatogonia and Sertoli cells, respectively) were crossed with mice bearing loxP-flanked alleles of Raldh1, Raldh2 and Raldh3. The animals bearing the Amh-Cre transgene lack ATRA only in Sertoli cells, while those bearing both the Amh-Cre and Stra8-Cre transgenes lack ATRA in the whole seminiferous epithelium. Main results and the role of chance In mice lacking RALDH in the whole seminiferous epithelium, we show that differentiated spermatogonia initiate meiosis properly, despite the evident lack of ATRA. However, the generated spermatocytes die by apoptosis at the mid-pachytene stage. They display signs of altered meiotic recombination (i.e., increased methylation of histone H3 on lysine 4, increased phosphorylation of histone H2AX variant, increased number of RAD51 foci), without abnormal chromosomal pairing, as assessed by SYCP1 and SYCP3 distributions. In addition, they lose TEX14 expression and inter-cytoplasmic connections, thus forming syncytia. When ATRA is missing in seminiferous epithelium, the Sertoli cells also are affected. They display an altered architecture, a disrupted distribution of connexin 43, and fail to maintain a functional blood-testis barrier. These defects are possibly related to a cell-autonomous, abnormal, increase of SRC kinase phosphorylation. We finally demonstrate that ATRA given at the preleptotene/leptotene stage is able to rescue all of these defects and to allow normal meiotic progression and spermiogenesis. Altogether, these results indicate that ATRA is required during the early step of meiotic prophase I, not to initiate meiosis by inducing STRA8 expression, but to allow proper chromosome recombination, survival of spermatocytes beyond the mid-pachytene stage and maintenance of Sertoli cell functioning. Limitations, reasons for caution These results were obtained using a mouse model and therefore the relevance for human spermatogenesis remains to be demonstrated. Wider implications of the findings Pharmacological studies support a role for ATRA in spermatogenesis in humans. Whether testicular impairment of ATRA synthesis or functioning might be a cause of azoospermia or oligozoospermia remains to be explored in clinical research. According to our study, ATRA emerges as a potential candidate for developing treatments in infertile patients. Trial registration number N.A.