The active form of vitamin A is retinoic acid that is synthesized in precise cellular locations by a two-step mechanism. First, retinol which is the circulating form of the vitamin is oxidized in a reversible reaction to retinal by retinol dehydrogenase (RDH10). Retinal is then oxidized to retinoic acid in a nonreversible reaction by one of 3 retinal dehydrogenases known as ALDH1A1, ALDH1A2, and ALDH1A3 (1).

ALDH1A2 and ALDH1A3 are required during fetal development. ALDH1A2^−/− mice die during embryonic development and ALDH1A3^−/− mice die shortly after birth (2, 3). However, ALDH1A1^−/− mice develop normally (4). In humans, ALDH1A1 mRNA is found in the liver, kidney, testis, brain, lung, red blood cells, and lens of the eye while ALDH1A2 mRNA is found in the testis, uterus, and skeletal muscle, and ALDH1A3 mRNA is localized in the prostate, trachea, intestine, and testis (5). Clearly all three ALDH1A enzymes contribute to RA synthesis during postnatal life. All together these studies underscore the tissue-specific central roles that ALDH1A enzymes play in animal physiology, and the vital significance of obtaining information concerning the expression and essential nature of the activity of these enzymes in human tissues.

In the mouse testis, retinoic acid is essential for the progression of undifferentiated spermatogonia A to become differentiating spermatogonia A1 and enter into spermatogenesis (6). In the absence of retinoic acid, undifferentiated spermatogonia never begin this progression (7). We have previously shown that deletion of the RDH10 gene in Sertoli cells alone will inhibit the progression of undifferentiated spermatogonia and also the deletion in both germ cells and Sertoli cells blocks this progression (8). Enzyme inhibitors have been used to eliminate the activity of all 3 aldehyde dehydrogenases and using the CRElox P approach all 3 Aldh1a genes have been deleted only in germ cells, only in Sertoli cells and in both cell types (9, 10). Both of these approaches have shown that the retinal dehydrogenases are essential for spermatogenesis and both enzymes are present in germ cells and Sertoli cells. Aldh1a1 is most highly expressed in the Sertoli cells and Aldh1a2 and Aldh1a3 are expressed primarily in the germ cells but all 3 enzymes appear to be expressed at some level in both cell types (11).

It has been known that global deletion of the Aldh1a1 gene in mice has little effect and does not significantly alter fertility (4). Recent studies have shown that the knockout of Aldh1a2 alone or the simultaneous knockout of Aldh1a1-3 in germ cells has little effect on successful spermatogenesis and fertility (9, 12). However, the simultaneous knockout of Aldh1a1-3 in Sertoli cells does not allow the undifferentiated A spermatogonia to progress to differentiating A1 spermatogonia (9). In this study we have broadened these previous observations by examining the effect on spermatogenesis of leaving the Aldh1a3 gene intact. We started with a mouse mutant with a global deletion in Aldh1a1. From that genotype we used the CRE-loxP system to remove the Aldh1a2 gene in germ cells and/or Sertoli cells and globally in all cells. In addition, and we have included sperm counts and fertility studies.

Using the Aldh1a1^-/-, Aldh1a2^fl/fl mice as our starting point we first wanted to see whether the presence of Aldh1a3 altered the results from the previous studies of Teletin et al. (9). Their data showed that the deletion of all 3 Aldh1a genes in Sertoli cells blocked spermatogenesis at the conversion of A spermatogonia to A1 spermatogonia in mice. However, if these mice were injected with retinoic acid once, the block was removed, and spermatogenesis proceeded normally and continuously. They also showed that spermatogenesis was normal with the deletion of all 3 Aldh1a genes in germ cells alone. They concluded that RA from Sertoli cells was necessary for the initial A to A1 conversion of spermatogonia but that RA from germ cells could maintain the process. If RA synthesis was normal in Sertoli cells the presence of RA synthesis in germ cells was not necessary. We created Aldh1a1^-/-, Aldh1a2^fl/fl mice under control of AMH Cre or Stra8 Cre to produce deletions of only Aldh1a1 and Aldh1a2 in Sertoli cells or germ cells, respectively. Aldh1a1^-/- mice have essentially normal spermatogenesis and we previously showed that Aldh1a2^-/- mice also have normal spermatogenesis (12). However, the deletion of both of these two genes in Sertoli cells or germ cells recapitulated the results from deletion of all 3 genes reported by Teletin et al. (9). We also found that deletion of Aldh1a1 and Aldh1a2 genes in Sertoli cells completely blocked spermatogenesis at the conversion of A spermatogonia to A1 and that this block could be overcome by a single injection of RA ( Figure 1 ). The knockout of genes coding for both enzymes in germ cells had no effect on sperm production. The results based on histology ( Figure 1 ) were reflected in testis weights and number of sperm detected in the cauda epididymis. In breeding studies all of the crosses that had detectable sperm in the cauda produced normal litter numbers and sizes. So, the production of RA by Sertoli cells from either or both Aldh1a1 or Aldh1a2 enzymes is required for the initiation of spermatogenesis and the presence of Aldh1a3 enzymes cannot compensate.

In order to examine the effect of globally deleting Aldh1a1 and Aldh1a2 we created Aldh1a1^-/-, Aldh1a2^fl/fl mice with an inducible ERT-Cre, and then utilized the injection of tamoxifen to activate the CRE activity. We used several protocols for the injection of tamoxifen. We injected once per day for 3 consecutive days starting at 8 days of age or once per day for 5 days at 21 days of age. In both protocols we found that spermatogenesis appeared to be little affected since the sperm counts per cauda epididymis were near normal and most of the mice fathered litters of near normal size. However, if we combined the protocols and injected tamoxifen for 3 consecutive days at 8 days of age and then repeated the injections for 5 days at 21 days of age, we found that by 4 to 5 months of age when the mice were analyzed the sperm counts went to zero and no litters were produced in breeding trials. Apparently, neither of the 2 individual tamoxifen injection regimes were sufficient to eliminate all ALDH1A2 activity. An alternative protocol where on the day of birth and on 1 day of age the males were injected with tamoxifen produced more robust results. Under this protocol none of the males produced any sperm throughout their lifetime. Some of these mice treated with tamoxifen on day of birth and day 1 after birth were raised to 23 days of age and injected with a bolus of RA. The histology of the testes of these animals were examined 42 days after the injection of RA but the block at the conversion of A spermatogonia to A1 spermatogonia was still in place and no sperm were produced ( Figure 1 ). This was expected since there is no source of RA from the germ cells or Sertoli cells to support spermatogenesis as was also seen for Aldh1a1 ^-/-, Aldh1a2 ^fl/fl, Stra8-Cre+, Amh-Cre+ ( Table 1 line 6).

The mice with globally deleted Aldh1a1 and Aldh1a2 from either tamoxifen injection protocol that resulted in aspermatogenesis were viable and appeared normal in all respects with the exception of the testis ( Table 2 ). A detailed pathological examination was not done but the mice were routinely aged to 4 months and some were left for over 6 months and showed normal body weight and no obvious pathologies.

The action of retinoic acid (RA) is required for normal spermatogenesis in rodents and possibly all mammals (15). We have previously shown that RA is synthesized locally in pulses along the seminiferous tubules (16). These pulses are required for the transition of undifferentiated A spermatogonia into A1 spermatogonia and into the differentiation pathway (7). The location of these pulses corresponds to the onset of spermatogenesis and the initiation of the cycle of the seminiferous epithelium. In the absence of RA there is no cycle, and no germ cells advance beyond undifferentiated spermatogonia. It has been established that the pulse of retinoic acid is a result of the localized synthesis of retinal by retinol dehydrogenase 10 (RDH10) and the conversion of retinal to retinoic acid by 3 aldehyde dehydrogenases designated ALDH1A1, ALDH1A2 and ALDH1A3 (8, 9, 17, 18). Both the Sertoli cells and the germ cells have the capacity to synthesize RA (9).

Deletion of either the Aldh1a1 gene or Aldh1a2 gene alone has no major consequences to spermatogenesis or the mice. Teletin et al. (9) used a Cre-Lox P approach to eliminate all 3 Aldh1a genes from Sertoli cells or from germ cells or from both cell types. From these experiments they determined that RA from Sertoli cells was essential to begin the first wave of germ cell development. Deletion of all 3 genes from germ cells had no effect on spermatogenesis. However, in the Sertoli cell specific triple gene deletion, if RA was present during the first wave in the form of a single injection, spermatogenesis proceeded normally and was continuous suggesting that the germ cell RA was sufficient to maintain spermatogenesis once it had been initiated. We addressed these studies using a different genetic approach where we left the Aldh1a3 gene intact. While there are only low levels of ALDH1A3 in the testis we wanted to determine if it was sufficient to maintain spermatogenesis.

Our cell specific deletions of only Aldh1a1 and Aldh1a2 recapitulated the results from Teletin et al. (9) who deleted all 3 Aldh1a genes. Deletion of these 2 genes and maintenance of the Aldh1a3 gene in Sertoli cells completely blocked spermatogenesis unless an injection of RA was made. Thus, the presence of ALDH1A3 alone is not sufficient to maintain spermatogenesis. In addition, deletion of these two genes in germ cells had no effect on sperm production. RA synthesized in Sertoli cells is sufficient to initiate and maintain spermatogenesis while RA from germ cells can only maintain spermatogenesis after it has been initiated.

Because of the absolute requirement of RA for spermatogenesis, it has been proposed that inhibition of the synthesis or the action of RA could be a possible approach for contraceptive development (10, 11, 19). Blocking RA synthesis with an aldehyde dehydrogenase inhibitor or use of RA analogs that inhibit the action of retinoic acid receptors (RAR) have been shown to block spermatogenesis (10, 20–22). However, given the prevalence of the RA signaling system in biology and its absolute requirement in embryogenesis there was serious concern about developing a contraceptive approach for the testis that could have serious consequences for other organ systems. Previously it has been shown that global deletions of the genes for Cyp26A1 and Cyp26b1, the enzymes involved in RA homeostasis, lead to increased concentrations of RA in several organs, reduced lifespan, failure to gain weight, and fat atrophy (23). So, increased RA concentrations in adult mice led to severe physiological consequences. Therefore, one of the goals of this study was to examine the viability of mice after global deletion of 2 of the 3 Aldh1a genes and a decreased ability to synthesize RA. We found that the global deletion of Aldh1a1 and Aldh1a2 had no apparent effect on the gross viability of the mice. Teletin et al. (9), only reported data on the testis cell specific deletion of all 3 genes coding for ALDH1A enzymes so in our studies it is possible that ALDH1A3 was able to compensate in some tissues other than the testis. Nonetheless, inhibitors targeting ALDH1A1 and ALDH1a2 would certainly act as effective contraceptive compounds while not affecting gross viability. While we did not examine the physiopathology of potentially affected systems such as the immune system these results are significant in attesting to the feasibility of a RA focused contraceptive approach.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

The animal study was reviewed and approved by WSU IACUC.

Experimental protocols were done by TT and the experiments were planned by TT and MG.

Supported by HD 10808 from NIH.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer KR declared a shared affiliation with the authors to the handling editor at the time of review.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.