The present study has identified a novel mechanism that could potentially play a key role in up-regulating testosterone production by rat fetal LC during and after the critical masculinization programming window (MPW) 
. The mechanism we propose is a progressive age-related reduction in expression of COUP-TFII in fetal LC, which effectively removes repression by a competitor with SF-1 for binding to overlapping sites in the promoter region of steroidogenic enzyme genes. The present results show that this mechanism is perturbed by exposure to three separate factors in the rat, some of which may be relevant to the human and other species. This mechanism could also partly account for the previously unexplained ‘paracrine’ regulation of fetal LC steroidogenesis during the MPW in rats and humans 
. Our proposal would add another tier of evidence for the COUP-TF family being active repressors of key elements in the male reproductive system, ranging from luteinizing hormone expression (LHβ
) through LH receptor expression (LHR
) to fetal LC function (this study). Our results also identify, for the first time, a primary mechanism by which phthalates, such as DBP, inhibit steroidogenesis by fetal LC in the rat (but not in the mouse). Our studies show that the COUP-TFII mechanism is present in human fetal LC, but whether it plays a role analogous to that which we propose in the rat will depend on further studies.
Our results provide convincing time course and dose-response evidence that exposure to DBP, DES and to a lesser extent dexamethasone, prevent the normal time-dependent down-regulation of nuclear COUP-TFII that occurs in fetal LC in the rat and which is associated temporally with expansion of LC cytoplasmic volume (which harbors the steroidogenic organelles) and increase in ITT. The temporal changes fit with the demonstration that DBP-induced down-regulation of SF-1-dependent LC-specific genes first emerges at e17.5 
, consistent with this being the earliest age at which COUP-TFII expression in LC in control animals is down-regulated, a change prevented by DBP-treatment. We show that treatment-induced changes in LC nuclear COUP-TFII expression are, in all instances, associated with inverse changes in ITT and with altered expression of SF-1-dependent LC-specific genes that have shared/overlapping SF-1 and COUP-TFII response elements in their promoter regions (). In contrast, expression of the LC steroidogenic gene (3β-HSD
) that does not have an overlapping SF-1 and COUP-TFII response element in its promoter, was unaffected by DBP exposure, as was expression of Amh
in Sertoli cells; in the latter case, there are separate SF-1 and COUP-TFII response elements in the promoter of Amh
(), but in any case COUP-TFII was never expressed in Sertoli cells in our studies. Therefore, we show a robust association between the LC-specific expression of COUP-TFII, reduced ITT and the down-regulation of steroidogenic genes that have overlapping SF-1 and COUP-TFII response elements.
Our identification of altered COUP-TFII expression in fetal rat LC as a mechanism underlying suppression of ITT resulting from experimental treatments (DBP and/or Dex or DES) is based on showing a consistent inverse association
between the percentage of fetal LC expressing COUP-TFII in their nuclei and ITT levels. This association does not in itself prove ‘cause and effect’. The ideal way of proving this would be to over-express COUP-TFII in fetal LC and show this reduces testosterone production. Such studies have been done with adult-derived bovine steroidogenic cells via transfection and shown to result in reduced steroidogenesis and expression of StAR
, as found in the present association studies. Numerous studies have shown that the mechanism underlying such effects involves competition between COUP-TFII and SF-1 for binding to an overlapping response element in the promoter region of genes encoding steroidogenic enzymes 
, as proposed for the present studies in fetal rat LC. Unfortunately, our studies using viral transfection of ex vivo
cultured rat fetal LC with COUP-TFII resulted in cell death (unpublished data), and there are also inherent problems with the culture of fetal LC, which rapidly lose their steroidogenic function 
. Therefore, this direct approach was not an option for us. We therefore decided on two alternative approaches to provide stronger evidence for causation, one involving re-induction of COUP-TFII in rat fetal LC (by DBP treatment) after its age-related loss, and the second involving parallel studies in the mouse in which DBP had been shown by others to be incapable of suppressing steroidogenesis and the expression of SF-1-dependent genes 
For the first approach, we exposed pregnant rats to DBP at a time in gestation (from e19.5–e20.5) when COUP-TFII had already switched off in the majority of fetal LC. This ‘late window’ DBP treatment resulted in re-induction of COUP-TFII expression in most of the fetal LC and an associated reduction in ITT at e21.5, consistent with our mechanistic proposal. In our mouse studies we confirmed that DBP exposure had no effect on ITT, nor was there induction/maintenance of COUP-TFII expression in fetal LC. However, exposure of pregnant mice to DES, rather than DBP, did result in profound suppression of ITT and a corresponding increase in the percentage of fetal LC expressing COUP-TFII, a change that paralleled that found for DES in the rat. The degree of suppression of ITT induced by DES was notably larger than that induced by DBP (in the rat), a difference probably explained by a parallel reduction in LH drive to the LC due to reduced LHR
expression. This raises the possibility that LH secretion, which is initiated at ~e18.5 in the rat and increases progressively thereafter 
, might be involved in switching off the expression of COUP-TFII in fetal LC and that DBP causes its steroidogenic effects by suppressing LH. As we were unable to measure fetal LH in blood, we could not test this possibility directly, but existing data suggests it is an unlikely explanation for our findings. First, it would fail to explain why the effects of DBP on ITT and steroidogenic enzyme expression in rats are first detectable at e17.5 (this study and 
), an age prior to the production of LH in the rat 
. Second, in vitro
studies using rat fetal testis cultures show that phthalate metabolites inhibit testosterone production regardless of the absence or presence of LH in the culture media 
. Nevertheless, even if DBP did suppress LH, it would appear that this suppression then results locally in a failure of COUP-TFII to switch off normally in fetal LC, which would still represent the causal mechanism within the LC. Moreover, if DBP exposure should inhibit fetal LH secretion, it is likely to involve a similar mechanism to that which we propose for the fetal LC, as COUP-TFII has been shown to competitively antagonize SF-1-induced LHβ
expression in the adult pituitary gland 
We considered reverse causation as an alternative explanation for our findings, namely that because reduced ITT was found in every instance in which there was abnormal maintenance/induction of COUP-TFII expression in fetal LC, then the former could be driving the latter. We consider this unlikely, because in complete androgen receptor knockout (ARKO) mice the fetal LC at e18.5 are predominantly immunonegative for COUP-TFII and, second, in the rat most fetal LC do not express the androgen receptor and are thus not directly androgen-responsive 
Our analyses of COUP-TFII expression in fetal LC used confocal microscopy and identification of LC by cytoplasmic staining for 3β-HSD. We were able to do this because, unlike the other SF-1-regulated LC steroidogenic genes, expression of 3β-HSD
was unaffected in any of our treatment groups. Use of high resolution tiled images of complete fetal testis cross-sections allowed us to identify fetal LC unequivocally and to specifically assess the presence or absence of COUP-TFII expression in individual LC. Since COUP-TFII is abundantly expressed in other cell types in the fetal testis, especially in non-Leydig interstitial cells, whole testis measurements such as the analysis of total testicular COUP-TFII
mRNA expression would not be meaningful, and, indeed, we found no effect of DBP-exposure on overall COUP-TFII
mRNA expression in the fetal rat testis. We saw no evidence for altered COUP-TFII expression in the non-Leydig interstitial cells in the fetal testis, and these cells did not affect our analyses because these were focused only on identifiable fetal LC (ie cells expressing 3β-HSD in their cytoplasm). We chose an antibody dilution for detection of COUP-TFII immunoexpression that discriminated between immunonegative LC in controls and immunopositive LC in DBP-exposed animals. In reality, we think it likely that this distinction represents profound down-regulation, rather than complete absence, of COUP-TFII immunoexpression in the nuclei of late gestation fetal LC in controls, based on titration studies with the COUP-TFII antibody (Fig. S6
Based on the age-related change in COUP-TFII immunoexpression in fetal human LC in the present studies, the mechanism which we propose for COUP-TFII in the rat may apply to the human, but more detailed studies are needed to support this possibility. This does not imply that each of the treatment effects shown to affect this mechanism in the rat will apply to the human, as our preliminary data is that in the human, as in mice, DBP neither affects steroidogenesis 
nor COUP-TFII expression (our unpublished data), at least in a xenograft model system. Nevertheless, as we show that three separate factors can maintain/increase nuclear expression of COUP-TFII in fetal rat LC with associated decreases in fetal ITT, it suggests that the COUP-TFII mechanism is potentially vulnerable to a wider range of factors.
In conclusion, our results all point strongly towards COUP-TFII expression being a key (negative) regulator of steroidogenesis within fetal LC during and after the critical period for masculinization in the rat, and potentially in the human. Thus, lifting of steroidogenic repression by COUP-TFII, rather than direct stimulation of steroidogenesis by paracrine factors, could be the primary LH-independent mechanism responsible for increasing testosterone production to induce masculinization. Perturbation of this novel pathway is clearly linked via our DBP studies in the rat to downstream TDS disorders. We show that this pathway can be impacted by factors other than DBP, for example via the stress hormone axis (glucocorticoids) and by estrogens. We consider it likely that other factors (eg other environmental chemicals) also target this pathway. The present findings suggest new pathways by which lifestyle factors in combination with environmental chemicals could exert adverse effects and lead to TDS disorders.