In the present study, we investigated the effect of DHA treatment on bovine GCs in culture. For the first time, we showed that the addition of 10 μM and 50 μM DHA increased cellular proliferation, that the addition of 20 μM and 50 μM DHA increased progesterone secretion and that the addition of 10 μM and 20 μM DHA increased estradiol secretion. We also performed experiments to decipher the possible mechanisms of DHA action on GC functions. We showed that DHA increased PCNA, StAR and steroidogenic enzyme HSD3B1 and CYP11A1 protein expression and MAPK14 phosphorylation. We also reported, for the first time, the mRNA and protein expression of FFAR4 in bovine GCs and showed that FFAR4 activation by the FFAR4 agonist TUG-891 led to a similar increase in cellular proliferation and MAPK14 phosphorylation as with DHA treatment. However, TUG-891 treatment did not lead to increases in steroid secretion.
DHA increased in vitro granulosa cell steroidogenesis
With regard to steroidogenesis, we reported here that DHA increased secretion of both progesterone and estradiol. This finding is consistent with increases in the protein expression level of the steroidogenic enzymes, HSD3B1, CYP11A1 and of the carrier StAR that we observed. Indeed, StAR protein is essential for the initial and rate-limiting step in steroid biosynthesis, namely transfer of hydrophobic cholesterol from the outer mitochondrial membrane to the inner mitochondrial membrane (reviewed in [
54]), the location of another enzyme crucial to steroid biosynthesis, CYP11A1. This enzyme converts cholesterol into pregnenolone [
55]. The HSD3B1 enzyme is a required step in the conversion of pregnenolone into progesterone [
56] or in theca cells to further obtain androstenedione [
57], the latter being a necessary precursor in estradiol synthesis. Moreover, such stimulating effects of n-3 PUFAs on steroidogenesis have already been described for ovine GCs in vitro, involving both progesterone and estradiol secretion [
51]. The increase in progesterone and estradiol observed after DHA treatment is also consistent with our previous results, which showed a stimulatory effect of DHA on progesterone secretion from bovine cumulus cells, originating from GCs [
19]. N-3 PUFAs have also been shown to increase progesterone secretion from ovine theca cells in vitro [
58] and follicular fluid progesterone concentration (from follicles 4–7 mm diameter) in vivo, in ewes fed with a n-3 PUFA-enriched diet [
51]. Furthermore, in agreement with the increase in StAR protein level measured in the present study, Waters et al. previously showed that a n-3 PUFA-enriched diet led to increased
StAR gene expression in another bovine reproductive compartment, the endometrium [
59]. In mouse Leydig tumor cells, inhibition of PTGS2 activity (an enzyme required for prostaglandin biosynthesis) is known to facilitate cAMP-induced steroidogenesis via increased StAR protein expression [
60]. N-3 PUFAs are known to be potent inhibitors of PTGS2 activity [
61]; this may explain the increase in StAR protein expression reported in the present study, following DHA treatment of GCs. The increased estradiol effect observed after DHA treatment indicates that DHA might also affect cytochrome P450 aromatase and/or 17β-HSD protein expression levels, as both of these enzymes are necessary for conversion of androstenedione (supplied in our culture media) to estradiol [
62]. However, this hypothesis would require further investigation. Finally, examination of steroidogenic enzyme activities would clarify whether DHA affects expression levels only or also the activities of these enzymes.
DHA mechanisms of action through FFAR4
For the first time, we showed in the present study that FFAR4 is expressed in bovine granulosa cells. To our knowledge, granulosa expression of FFAR4 has not yet been reported in other species. Only a few published papers have reported FFAR4 expression in bovine tissues, namely adipose tissue [
63,
64]. In other species, FFAR4 is expressed in adipose tissue, spleen and intestine in pigs [
65], while in human and rodents, its expression is more widely reported in additional tissues (immune cells, pancreatic cells, lung, muscle, liver and placenta) [
42,
66‐
69]. We demonstrated here that FFAR4 is localized in the vicinity of the cellular membrane in GCs, as expected for a G protein-coupled receptor. This expression of FFAR4 in bovine GCs indicates that long chain fatty acids, including DHA, could act on GCs, at least in part, via this receptor. However, it should be noted that not all cultured GCs were shown to express FFAR4.
Activation of FFAR4 by TUG-891 (1 and 50 μM) mirrored the effect of DHA on GC proliferation, but not on progesterone and estradiol secretion, suggesting that DHA effects on cell proliferation could occur through FFAR4 activation. Indeed, TUG-891 is a specific and potent agonist of FFAR4 [
41,
70]. Thus, the effects observed after TUG-891 treatment could be related to FFAR4 activation. It has already been reported that DHA could exert effects through FFAR4 stimulation in adipocytes [
71], in macrophages [
72] or in cardia cells [
73]. Of note, a linear dose response was not reported in the present study, notably concerning GC proliferation (DHA 20 μM). This observation could suggest that the DHA dose response followed a U-shaped (for example) response pattern [
74]. It is also possible that, at higher concentration, other DHA mechanisms could occur, potentially counteracting the FFAR4 effect on proliferation.
Moreover, several effects, such as the ability of DHA to inhibit responsiveness of macrophages to endotoxin, inhibition of IκB kinase phosphorylation, IκB phosphorylation and degradation, and inhibition of production of TNF, IL-6 and MCP-1, that have already been demonstrated, were abolished in FFAR4 knockdown cells [
28,
75]. In our study, we observed an increase in MAPK14 phosphorylation in GCs after DHA supplementation (10 and 50 μM), and a similar increase after TUG-891 treatment (1, 10 and 50 μM). As shown in the published literature, MAPK14 (p38 MAPK) may be involved in follicle stimulating hormone (FSH)-induced estradiol secretion in rat and mouse GCs [
37,
38], but was also demonstrated to be involved in GC proliferation in a culture model of hamster pre-antral follicles stimulated with transforming growth factor β1 (TGFβ1) [
76]. Thus, these results suggest that FFAR4 activation leads to the activation of this MAPK14 signaling pathway, which might play a role in DHA effects on bovine GC proliferation. However, additional experiments would be needed to demonstrate that DHA could activate FFAR4 in bovine GCs, involving investigation of, for example, plasmid transfection, FFAR4 overexpression and inhibition, and intracellular calcium measurement.
No effect of TUG-891 treatment was reported on GC steroidogenesis, suggesting that DHA effects on progesterone and estradiol secretion are independent of FFAR4 activation. Several other mechanisms of action have already been proposed for DHA: modification of cellular membrane lipid composition (fluidity, raft formation); regulation of eicosanoid production (for example, prostaglandins and leukotrienes); transcription factor expression (
PPARG and
SREBF1, for example) or signaling pathways, such as NFκB [
26,
28,
30]. In this study, we evidenced a significant increase in cellular DHA level from total lipids after 1 μM, 10 μM and 50 μM DHA treatment. It is thus possible that DHA affected steroidogenesis by one of these mechanisms and not by activating FFAR4.
The study of candidate gene expression that was performed here showed no modification of canonical genes (
PPARG,
PPARA,
SREBF1) following DHA treatment, except for
PPARA with the highest DHA dose used, DHA50 μM. Similar absence of effects on
PPARG have already been reported in other studies, as in this study [
77]. Nevertheless, TUG-891 treatment has already been reported to modify
PPARG transcription after several days of culture in 3T3L1 [
71]. In this study, we analyzed GC gene expression after 8 h of TUG-891 stimulation, which might be too brief a period to enable transcription modifications to occur immediately after FFAR4 stimulation. The only differences observed in gene expression are upregulations of
GPX4 (glutathione peroxidase 4),
NFκB (nuclear factor kappa B) and
PPARA (peroxisome proliferator-activated receptor alpha) expression after TUG-891 treatment (10 μM). GPX4 is a phospholipid hydroperoxidase that protects cells against membrane lipid peroxidation. Such a mechanism has already been described for DHA in murine hippocampal cells [
78]. By increasing
GPX4 expression, DHA is able to protect the cell from oxidative damage resulting of non-enzymatic peroxidation of membrane phospholipids. It is possible that this mechanism is associated with FFAR4 activation, as we found a similar increase in
GPX4 expression after TUG-891 treatment. DHA is able to bind
PPARG and
PPARA and can consequently increase insulin sensitivity [
26]. Through binding to
PPARG, DHA could inhibit activation of NFκB, a key transcription factor involved in inflammatory pathways [
26]. Moreover, the inhibitory effect of DHA on
NFkB can also occur via the FA receptor FFAR4 [
28]. The increased
NFkB mRNA expression reported after DHA 50 μM and TUG 10 μM in the present study is thus surprising and NFκB activation, which was not investigated in the present study, should be studied.
Overall, despite functional differences between DHA-treated and control GCs, there were no huge changes in gene expression. In order to further investigate the effects and mechanisms of action of FFAR4 activation, experiments involving primary GC culture should be replaced by those involving GC line cultures; such cultures would enable FFAR4 overexpression and inactivation, while maintaining a GC phenotype. It is also possible that a global transcriptomic approach should be envisioned after DHA treatment in order to have a broad picture of the potential mechanisms involved. Indeed, the investigation of other genes, such as PTGS2, might have shown differences that could explain StAR regulation, for example.
Of note, the culture system used in this experiment is serum free and prevent the differentiation of granulosa cells generally occurring after about 15 h culture. It also enable to treat cells with the precise concentrations of DHA (which would be already present in culture medium if we had used serum). In the culture system used in the present paper, GC are maintaining a round shape even when platted, at least till 48 h culture (the latest endpoints in this paper), and not a fibroblastic-like shape, meaning they did not differentiate as much as when serum is used. On the other hand, this culture system also presented some disadvantages. Indeed, primary cell cultures require freshly isolating GCs for each culture, and therefore they can exhibit huge variation in response to treatment, depending on the batch of ovaries used. In this context, the use of a high number of independent culture can be compulsory, as in the present work. Moreover, in this culture system, in order to be able to set up endpoints before 48 h culture, we chose to treat cells at the beginning of the culture, meaning that both platted and non-platted cells were treated. A similar number of living cells is cultured in each well, with a varying proportion of dead cells (assessed by trypan blue staining and cell counting). Concerning proliferation assay, as floating cells are removed before the assay, only thymidine incorporated in platted cells is measured (most of floating cells being dead cells in our cell culture system). Concerning data on steroidogenesis, steroid secreted in the culture medium by both platted and non-platted cells during the cell culture are measured after 48 h. Steroid concentration are normalized by protein concentration in each well, meaning by protein amount of platted cells only. We estimated that the proportion of platted cells to viable floating cells is the same in each well for a specific batch of cells. This normalization might biased the steroid results as the concentration of steroid for a viable amount of cells might be slightly overestimated due to the normalization taking into account only platted cells. Nevertheless, we believe that this normalization would not affect differences observed between conditions.