Synthesis of DHA (omega-3 fatty acid): FADS2 gene polymorphisms and regulation by PPAR a

– In humans, in several biological systems, in particular the nervous system, the FADS2 gene transcribes D 6-desaturase, which is the rate-limiting enzyme for converting a -linolenic acid into docosahexaenoic acid (an n-3 fatty acid). The peroxisome proliferator-activated receptor a (PPAR a ) modulates the transcription of FADS2 gene by interacting with a second transcription factor: the retinoid X receptor a (RXR a ).ThesetranscriptionfactorstaketheformofaPPAR a -RXR a heterodimerandaremodulated by the ligands that modify their respective structures and enable them to bind to the peroxisome proliferator responseelement(PPRE)locatedinthepromoterregionofthe FADS2 gene.Freeestradiolinducestheactivation ofPPAR a via twopathways(i)transcriptionthroughgenomicactionmediatedbyanestrogenreceptor;(ii)anon-genomic effect that allows for phosphorylation and activates PPAR a via the ERK1/2-MAPK pathway. Phosphorylation is an on/off switch for PPAR a transcription activity. Since D 6-desaturase expression is retro-inhibited by free intracellular DHA in a dose-dependent manner, this position paper proposes an original hypothesis: if DHA simultaneously binds to both phosphorylated PPAR a and RXR a , the resulting DHA-PPAR a P-RXR a -DHA heterodimer represses FADS2 gene via PPRE. The retinoic acids-RAR a -RXR a -DHA heterodimerwouldnotdissociatefromcorepressorsandwouldprevent coactivators from binding to FADS2 .We speculate that SNPs, which are mostly located on PPRE, modulate the binding af ﬁ nities of DHA-PPAR a P-RXR a -DHA heterodimer to PPRE. The DHA-PPAR a P-RXR a -DHA heterodimer ’ s greater af ﬁ nity for PPRE results in a decreased production of D6D and DHA. FADS2 promoter polymorphism would increase the competition between DHA and other ligands, in accordance with their concentrations and af ﬁ nities.

There is a local competition between the substrates. The reactivity of the enzyme depends on the concentration of available substrates and their affinity with D6D. This is usually expressed as the enzyme's Km (Michaelis constant), an inverse measure of affinity (Ivanetich et al., 1996;Rodriguez et al., 1998). The conversion rate is also dependent on tissue type.
Several studies examining n-3 fatty acids have demonstrated the transcription of FADS2. Dietary n-3 fatty acid intake regulates D6D expression (Cho et al., 1999), but conversion to DHA is limited (Burdge, 2006). D6D expression is subject to feedback regulation by dietary polyunsaturated fatty acids (PUFAs) (Nara et al., 2002). Several studies in animals and in vitro models suggest that D6D expression and/or activity is regulated by steroid hormones (Childs et al., 2008), as well as by insulin in the liver (Cho et al., 1999). More 18:3n-3 is converted to 20:5n-3 and 22:6n-3 in women than in men (Burdge, 2006). However, the mechanisms regulating FADS2 transcription have not been well established to date. This position paper aims to highlight the possible mechanisms involved À active molecules and transcription factors À while also considering the impact of single-nucleotide FADS2 promoter polymorphisms.
2 Regulation of the FADS2 gene Peroxisome proliferator-activated receptor a (PPARa) modulates transcription of the FADS2 gene  by interacting with a second transcription factor, retinoid X receptor a (RXRa) (Tang et al., 2003). PPARa is stabilized through heterodimerization with RXRa, and there is a rapid turnover of excess protein unpaired with RXRa, unless it is bound by an appropriate ligand (Hirotani et al., 2001). 17bestradiol treatment upregulates PPARa mRNA expression and PPARa protein content (Campbell et al., 2003).
Like most nuclear receptors, PPARa is organized into functional domains called A/B, C, D and E/F. The A/B domain supports an activation function that is independent of ligand binding (Activation Function-1: AF-1). The central region, which contains the DNA-binding domain (domain C), contains the best-conserved nuclear receptor: a highly conserved twofinger zinc structure that allows interaction with the DNA molecule at specific sites. The C-terminal region, which contains domain E, is the second-best conserved region and contains the E domain. This region is the repository of numerous functions, in particular, the ligand-binding-dependent activation of genes (activation function 2 (AF-2) (Bugge and Mandrup, 2010).
PPARa is a phosphoprotein. Phosphorylation is an on/off switch for transcriptional PPARa activity. The phosphorylation of serine residues (at Ser 6, 12 and 21) (Barger et al., 2001) takes place in the A/B domain. It affects intramolecular communication, since the phosphorylation status of the A/B domain affects the activity of the E/F region of the protein (Burns and Vanden Heuvel, 2007).
These transcription factors take the form of a PPARa-RXRa heterodimer, located within the nucleus and are modulated by ligands that modify their tertiary structures, and enable them to bind to the peroxisome proliferator response element (PPRE) located in the promoter region of the FADS2 gene. The PPAR contains a direct repeat sequence 1 (DR1) that assists in the regulation of human FADS2 gene transcription (Tang et al., 2003). The heterodimer bond to the PPRE modulates the transcription of the gene under highly specific conditions. Several fatty acids prevalent in vivo, including palmitoleic acid (de Souza et al., 2017;Bolsoni-Lopes et al., 2013); oleic acid; linoleic acid (LA); a-linolenic acid (ALA); arachidonic acid (ARA) (Popeijus et al., 2014;Kliewer et al., 1997); eicosapentaenoic acid (EPA); and docosahexaenoic acid (DHA) (Diep et al., 2002;Deckelbaum et al., 2006;Dziedzic et al., 2018) function as ligands for PPARa at concentrations consistent with those found in human serum. 8-HETE (8-hydroxyeicosatetraenoic acid) is also a natural agonist of PPARa (Yu et al., 1995). This eicosanoid metabolite is a lipoxygenase product of arachidonic acid. It is believed to have tissue concentrations that are too low to be a ligand for PPARa.
In their free form, unlinked to binding proteins, these ligands compete with PPARa in accordance with their relative concentrations and dissociation constants.
Alternatively, transcription factor Elk-1, an ETS domain transcription factor encoded by the ELK1 gene, could be one of the proteins that binds to the promoter region of the FADS2 gene (Lattka et al., 2010). Elk-1 is known to play a role in transcription regulation by forming ternary complexes with serum response factor (SRF) on the serum response elements (SRE) of gene promoters (Hipskind et al., 1991). Elk-1 contains a MAPK-binding motif (Yang et al., 1998). Its activation and nuclear translocation require phosphorylation of serine 383-389, located in the C-terminal region of Elk-1, via the ERK-MAPK pathway, which results in a conformational change to Elk-1 (Lavaur et al., 2007). In silico, phosphorylated ELK-1 would act as a coactivator.
Does biological sex affect the regulation of the FADS2 gene? Observational evidence suggests that in populations that consume low levels of n-3 highly unsaturated fatty acids, women have higher blood DHA levels than men (Kitson et al., 2010). Women of reproductive age are known to convert more ALA into DHA than men (Burdge and Wootton, 2002). Estrogens cause higher DHA concentrations in plasma cholesteryl esters in women than in men, by upregulating synthesis from ALA. This difference is independent of dietary differences. It has also been suggested that estradiol may increase the activity of the desaturation pathway because DHA synthesis has been shown to be almost 3 times greater in women who take oral contraceptive pills that contain ethinylestradiol than in women who do not, while testosterone stimulus decreases DHA status (Giltay et al., 2004). This difference in conversion appears to be associated with estrogen and some evidence indicates that the expression of enzymes, including desaturases, involved in synthesizing DHA from ALA is higher in females. PPARa may mediate an estrogenassociated effect. However, because estrogens are weak ligands for PPARa, estrogen-mediated increases in PPARa activity probably occur through an indirect mechanism involving membrane-bound estrogen receptors (Kitson et al., 2010). What is the relationship between the regulation of the FASD2 gene by estradiol and by PPARa?
3 PPARa-RXRa regulation mechanisms governing the FADS2 gene As we mentioned above, at least two key molecules are involved in the regulation of the FADS2 gene: free 17bestradiol, when not bound to its binding protein SHBG (sex hormone-binding globulin) À when bound to SHBG it is not biologically activeand PPARa.
Free estradiol induces PPARa activation via two pathways. First, by transcription, through its genomic action on the PPARa gene, which is mediated by an estrogen receptor (ER) (Campbell et al., 2003). Then, via its non-genomic effect, mediated by ERa, which allows phosphorylation and activates PPARa via the extracellular signal-regulated kinase/mitogenactivated protein kinase pathway (ERK1/2-MAPK) (Moreno et al., 2010), as we will show below. This is also true of ELK-1.
Free estradiol has a synergistic effect with insulin-like growth factor 1 (IGF-1) (Majou, 2015). For example, several studies have shown this interaction in different brain regions ( Garcia-Segura et al., 2006;Varea et al., 2010;Park et al., 2014) and in breast cancer cells (Song et al., 2010). The anabolic action of IGF-1 is mediated by the IGF-1 receptor (IGF-1R). When IGF-1 binds to the IGF-1R, it causes a conformational change to the receptor, inducing the autophosphorylation of tyrosine residues (Hubbard and Till, 2000). This leads to the recruitment of insulin receptor substrates (IRS-1-4), which in turn phosphorylate the tyrosine residues of the IRSs. Estradiol stimulates the rapid activation of the IGF-1R through phosphorylation via ERa, and induces the formation of a ternary protein complex comprising of phosphorylated Shc protein, ERa and IGF-1R (Song et al., 2004). Adapter protein Shc generally acts by activating MAPK and phosphoinositide-3-kinase/Akt signaling pathways (Gu et al., 2000;Vindis et al., 2003). The tyrosine phosphorylation of Shc is mediated by the tyrosine kinase Fyn (Src family kinase) associated with caveolin-1 (scaffolding protein). Upon integrin ligation, Fyn is activated and binds to Shc via its SH3 domain. Shc is subsequently phosphorylated at tyrosine 317 (Wary et al., 1998). IGF-1R activation stimulates MAPK kinases and consequently phosphorylation of ERK1/2. Activation of ERK1/2 may in turn lead to phosphorylation of ERa (Kato et al., 1995;Russo et al., 2002). This process leads to the activation of two main downstream signaling pathways: phosphoinositide-3-kinase (PI3K)/Akt and MAPK cascades (Zheng et al., 2000). Thus, free estradiol induces MAPK activation (Kahlert et al., 2000) and PPARa phosphorylation. This is also true of ELK-1 (Duan et al., 2001) (Fig. 2).
The activated PPARa function is then regulated by its ligands, initially by binding to RXRa. Each PPARa and RXRa ligand modifies the tertiary structure of each transcription factor, creating a specific PPARa-RXRa heterodimer conformation that modulates binding to the FADS2 gene. The change in ligand binding domain conformation (AF-2) following ligand binding induces the recruitment of coregulators. In the absence of a ligand or in the presence of an antagonist, PPARa interacts with nuclear receptor corepressors (NCoR1/2), which block coactivators from binding and prevent the formation of the transcriptional complex (Dowell et al., 1999). This repression is reversible. When ligands (all-trans retinoic acid, 9-cis-retinoic acid) (Allenby et al., 1993) are bound by RAR (the retinoic acid receptor), this promotes the dissociation of the PPRE-NCoR complex and the subsequent interaction of PPRE with one or more coactivator proteins including the cAMP response element-binding protein (CREB)/p300 (Dowell et al., 1997), and steroid receptor coactivator 1 (SRC-1) (Gocke et al., 2009).
Given that the expression of D6-desaturase is retroinhibited by free intracellular DHA (Matzusaka et al., 2002;Bewicz-Binkowska et al., 2019), we propose the following original hypothesis: when DHA simultaneously binds to phosphorylated PPARaP and RXRa, the DHA-PPARaP-RXRa-DHA heterodimer represses the FADS2 gene via PPRE. Thus, depending on its concentration and binding affinities with PPARa and RXRa, DHA could block other competing ligands from accessing the two transcription factors. Both blockage and inhibition are reversible (Fig. 3). They can be overcome by competition between free DHA not bound to its binding protein called FABP (fatty acid binding protein), which is tissue-dependent (the FABP-bound fraction is not biologically active) and (i) free MUFAs (palmitoleic acid, oleic acid) or PUFAs (LA, ALA, EPA, ARA), not bound to FABP, on PPARa; (ii) free 9-cis-retinoic acid, not bound to its binding protein (retinoic acid-binding protein), on RXRa. Indeed, in the absence of 9-cis-retinoic acid, DHA diminishes FADS2 gene expression, and, in its presence, DHA upregulates its expression, especially in astrocytes (Dziedzic et al., 2018). At 75mM, DHA represses PPARa transactivation almost completely, in a dose-dependent manner, and acts as an antagonist (Lee and Hwang, 2002); conversely, DHA activates at lower doses (Popeijus et al., 2014).
Under basal conditions, the RARa-RXRa heterodimer represses the transcription of target genes by recruiting corepressor complexes on RARa, such as NCoR1/2 (Farboud et al., 2003). When a ligand binds to RARa, it causes a change to the receptor's conformation. These changes cause a transition from transcriptional repressor activity to an activator state through the loss of corepressors and the recruitment of coactivator complexes, such as CBP/p300 and SRC-1. In the RARa-RXRa heterodimer, RXRa acts as a transcriptionally silent partner to RARa while it is active in the RXRa-RXRa homodimer. This phenomenon is known as the "subordination" of RXRa to RARa (Le Maire et al., 2019): (i) in the RARa-RXRa heterodimer, an RARa agonist alone allows for the transactivation of target genes; (ii) an RXRa agonist (9-cis-RA) alone is unable to dissociate corepressors from the complex, preventing RXRa/coactivator binding; (iii) the binding of retinoic ligands to RARa and RXRa induces a binding synergy with a coactivator. Above a certain concentration, we suggest that DHA would both bind to PPARa and replace 9-cis-RA in binding with RXRa. In such a case, RXRa-DHA would form one heterodimer with PPARa-DHA, and another with RARa-retinoic acids. However, the retinoic acids-RARa-RXRa-DHA heterodimer would not dissociate in the presence of corepressors and would prevent binding with coactivators. Thus, above a certain concentration (Popeijus et al., 2014), DHA retro-inhibits the FADS2 gene, limits translation into D6D and the synthesis of DHA (Fig. 3).
It is interesting to note that the PPARa-RXRa regulation mechanisms governing the FADS2 gene are similar to those governing the human SCD gene. Stearoyl-CoA desaturase (SCD) is a lipid-metabolic enzyme that catalyses the insertion of a double bond between carbon atoms 9 and 10 in the saturated fatty acids palmitic acid and stearic acid to generate the mono-unsaturated fatty acids palmitoleic acid and oleic acid respectively (Fig. 1). SCD is a key, rate-limiting enzyme in the cellular synthesis of MUFAs from saturated fatty acids. A functional PPRE has been identified in the promoter of the SCD gene (Rakhshandehroo et al., 2010) and its gene expression is regulated by PPARa (Miller and Ntambi, 1996;Hebbachi et al., 2008).
DHA lowers the SCD gene's protein expression and mRNA expression levels (Bellenger et al., 2004;On et al., 2019). All-trans-retinoic acid and 9-cis-retinoic acid increase SCD mRNA expression in a dose-dependent manner (Miller et al., 1997;Samuel et al., 2001;Mahesh et al., 2016). All these elements suggest that SCD regulation appears to be mediated by PPARa-RXRa and RARa-RXRa heterodimers with inhibition by DHA and activation by retinoic acids.
Thus, DHA seems to retro-regulate the FADS2 gene in two ways: (i) directly, by inhibiting the expression of the FADS2 gene, (ii) indirectly, by inhibiting the SCD gene. This repression downregulates SCD and decreases palmitoleic acid and oleic levels; these two MUFAs are PPARa ligands in competition with DHA.
Some studies suggest that D6D expression is dually regulated by SREBP-1c and PPARa (Matsuzaka et al., 2002). However, SREBP-1c's mechanism of action remains under debate. The transcription factor, sterol regulatory elementbinding protein (SREBP)-1c, controls lipogenesis in the liver (Shimano et al., 1999), but also in astrocytes (Tabernero et al., 2002) among other tissues. It regulates the expression of several genes that encode enzymes involved in fatty acid synthesis, in particular stearoyl-CoA desaturase (SCD)-1. SCD-1 catalyzes the rate-limiting step in biosynthesis (Kim et al., 2002). SCD-1 has a specific affinity for palmitic and stearic acid. Palmitoleic acid is synthesized from palmitic acid (Fig. 1) directly by desaturation (SCD-1) (likewise, oleic acid is synthesized from stearic acid). Palmitoleic acid is regarded as a lipid-controlling hormone, or "lipokine", because it modulates lipid metabolism. This characteristic (regulatory signal) distinguishes palmitoleic acid from oleic acid (Cao et al., 2008). Otherwise, palmitoleic acid functions as a ligand for PPARa and increases its transactivation (Popeijus et al., 2014). Thus, we can assume that the regulation of D6D expression by SREBP-1c is carried out indirectly via SCD-1 and the synthesis of palmitoleic acid. This regulation would be upstream of PPARa and complementary to this transcription factor (Fig. 4).
SREBP-1c is bound to two other proteins: SCAP (SREBPcleavage-activating protein), a polytopic membrane protein that serves as an escort for both SREBPs (Sakai et al., 1997) and Insig-1 (insulin-induced gene 1). The dissociation of Insig-1 from the SREBP-SCAP complex allows the complex to migrate from the endoplasmic reticulum to the Golgi apparatus, where SREBP is cleaved by S1P and S2P (site-1 and À2 protease) À two enzymes that are activated by SCAP. The cleaved SREBP-1c then migrates to the nucleus and acts as a transcription factor to bind to the sterol regulatory element (SRE). The retention of the SREBP-SCAP complex in the endoplasmic reticulum depends on whether it binds to the endoplasmic reticulum resident proteins, Insig-1. By binding with SREBP-1c, Insig-1 inhibits these proteins depending on the level of Insig-1. Insig-1 is a critical regulator of SREBPs (Yang et al., 2002;Yabe et al., 2002) (Fig. 4).
4 Promoter polymorphisms of the FADS2 gene and their consequences Several significant associations between FADS genotypes and long-chain polyunsaturated fatty acids have been confirmed in diverse types of human tissue (erythrocytes, plasma, skin, breast milk, etc.), demonstrating that FADS2 gene cluster polymorphisms are major regulators of the synthesis of this type of fatty acid. One study (Ameur et al., 2012) examined the genotyping of the FADS cluster in five European cohorts, as well as the genomic data available from human populations, archaic hominids and more distant primates. The results show that modern humans have two haplotypes (groups of alleles at different loci on the same chromosome) for the FADS cluster: A and D, defined by 28 SNPs. These two haplotypes differ considerably in their capacity to synthesize long-chain fatty acids. In the two families of fatty acidsomega-3 and omega-6haplotype D is strongly associated with lower levels of precursors to the synthesis of fatty acids (a-linolenic acid, linoleic acid) and higher levels of EPA, DHA and arachidonic acid. This indicates that this haplotype is more effective at converting precursors. People who are homozygous for haplotype D have 24% more DHA and 43% higher levels of arachidonic acid than homozygotes for haplotype A. Several studies have shown a close correlation between several single nucleotide polymorphisms (SNPs) in the FADS1 and FADS2 genes and the concentrations of omega-3 and -6 fatty acids (Schaeffer et al., 2006;Xie and Innis, 2008;Rzehak et al., 2009;Glaser et al., 2011). Homozygous carriers of different minor alleles have higher levels of desaturase substrates (a-linolenic acid, linoleic acid) and lower levels of desaturation products (DHA, EPA, arachidonic acid) (Glaser et al., 2011). This suggests the reduced expression of desaturases in these polymorphisms (Moltó-Puigmarti et al., 2010). Conversely, several studies suggest that some alleles of several SNPs in the FADS2 gene cluster are associated with higher D6D activity (homozygous carriers of the minor allele of rs1535) (Harsløf et al., 2013). Nwankwo et al. (2003) demonstrated that the insertion of a nucleotide in the transcriptional regulatory region of the human FADS2 gene resulted in D6D deficiency and decreased FADS2 transcription (minor allele rs968567). Promoter-reporter assays demonstrated a 6-fold decrease in promoter activity in the regulatory region of the polymorphic FADS2 variant compared with the normal gene, confirming the functional relevance of the insertion mutation to diminished expression of the gene. Lattka et al. (2010) showed that the FADS2 promoter region surrounding SNP rs968567 exhibits promoter activity, which increases when the major C allele of SNP rs968567 is replaced by the minor T allele. This effect could be caused by the allelespecific differential binding affinity of transcription factors.
Given the hypothesis that the DHA-PPARaP-RXRa-DHA heterodimer represses the FADS2 gene via its PPRE and the results of these studies, we speculate that SNPs, especially those on PPRE, modulate the binding affinity of DHA-PPARaP-RXRa-DHA heterodimer on PPRE. If this affinity is increased, it will take less DHA to bind to PPARa and inhibit the transcription of FADS2 gene, and vice versa with decreased affinity. Lower production of D6D and DHA result from greater DHA-PPARaP-RXRa-DHA heterodimer affinity for PPRE (Lattka et al., 2010). FADS2 promoter polymorphism would enhance competition between DHA and other ligands, depending on their concentrations and affinities (dissociation constants Kd).

Conclusion
The DHA concentration level is highly dependent on the catalytic capacity of D6D, a key enzyme in DHA synthesis. The regulation of the FADS2 gene by PPARa-RXRa and their ligands is compatible with the recruitment of coactivators and corepressors. DHA causes dose-dependent genetic feedback inhibition of the FADS2 promoter.
Mutations in the FADS2 gene can have positive or negative effects on D6D levels. The polymorphism of this gene appears to have been essential in allowing the Homo genus to adapt to its diet and evolve (Majou, 2018). The geographic distribution of FADS2 genetic variants now differs dramatically between continents. These differences in the capacity to synthesize DHA might contribute to health disparities. For example, over the past 10 years or so, studies have increasingly highlighted the relationship between DHA deficiency and certain pediatric neuropathologies, such as hyperactivity, learning difficulties (Milte et al., 2012), mental retardation (Neggers et al., 2009), epilepsy (Emory University Health Sciences Center, 2004) and autism (Bent et al., 2009). In the elderly, DHA depletion is an aggravating factor in the etiology of Alzheimer's disease (Majou, 2015).
It is essential to understand the regulation modalities of this gene, particularly in relation to its various alleles. Some SNPs in the FADS2 gene are considered an aggravating factor in DHA deficiency. The dose-dependent feedback inhibition caused by DHA in the DHA-PPARa-RXRa-DHA heterodimer could be a key factor for people with a genetic DHA deficiency. DHA depletion caused by FADS gene polymorphisms should be compensated entirely or partially by dietary DHA intake.