The Preservation of PPARγ Genome Duplicates in Some Teleost Lineages: Insights into Lipid Metabolism and Xenobiotic Exploitation

Three peroxisome proliferator-activated receptor paralogues (PPARα, -β and -γ) are currently recognized in vertebrate genomes. PPARγ is known to modulate nutrition, adipogenesis and immunity in vertebrates. Natural ligands of PPARγ have been proposed; however, the receptor also binds synthetic ligands such as endocrine disruptors. Two paralogues of PPARα and PPARβ have been documented in teleost species, a consequence of the 3R WGD. Recently, two PPARγ paralogue genes were also identified in Astyanax mexicanus. We aimed to determine whether the presence of two PPARγ paralogues is prevalent in other teleost genomes, through genomic and phylogenetic analysis. Our results showed that besides Characiformes, two PPARγ paralogous genes were also identified in other teleost taxa, coinciding with the teleost-specific, whole-genome duplication and with the retention of both genes prior to the separation of the Clupeocephala. To functionally characterize these genes, we used the European sardine (Sardina pilchardus) as a model. PPARγA and PPARγB display a different tissue distribution, despite the similarity of their functional profiles: they are unresponsive to tested fatty acids and other human PPARγ ligands yet yield a transcriptional response in the presence of tributyltin (TBT). This observation puts forward the relevance of comparative analysis to decipher alternative binding architectures and broadens the disruptive potential of man-made chemicals for aquatic species.

In general, PPAR paralogues display a characteristic gene expression pattern, modulating different metabolic processes; specifically, PPARγ is highly expressed in tissues with active lipid metabolism [14], mostly due to its involvement in the regulation of adipogenesis and lipid storage as well as lipid metabolism and energy homeostasis [15]. In addition to this, other PPARγ-related processes have also been described. With respect to teleosts, PPARγ has been associated with the regulation of nutrition [16,17], adipogenesis [13,18], immunity [19,20] and was shown to be targeted and disrupted by environmental chemicals [6,16,21].
In the present work, we provide an exhaustive examination of PPARγ gene repertoire in different teleost fish species. Through comparative genomic and phylogenetic analyses, we show that the PPARγ gene duplicated in the ancestor of teleosts, and that both paralogues were retained in Salmoniformes, Clupeiformes, Siluriformes, Characiformes and Esociformes taxa. Furthermore, expression profiles and functional characterization of European sardine's (Sardina pilchardus) PPARγA and PPARγB is provided. Despite the differences in PPARγA and PPARγB gene expression patterns across several tissues of sardine, similar ligand-binding profiles were found, including activation by TBT, previously unreported within teleosts.

Phylogenic Analysis
A phylogenetic analysis was performed to establish the orthology of novel PPARγ sequences. Amino acid sequences from teleosts belonging to several taxonomic orders were collected from GenBank and Ensembl. In addition, PPARα and PPARβ amino acid sequences of a reptile (Anolis carolinensis), an amphibian (Xenopus tropicalis), two Actinopterigii (D. rerio and Lepisosteus oculatus), a bird (Gallus gallus) and two mammals (Homo sapiens and Mus musculus) were used to root the phylogenetic tree. The gathered amino acid sequences were aligned to identify and exclude partial and/or low-quality sequences for further analyses, resulting in a final dataset of 68 PPAR amino acid sequences (accession numbers in Table S1). A MAFFT alignment was generated using MPI Bioinformatics Toolkit [32]. Columns containing 50% gaps were stripped, resulting in a final alignment with 483 positions for phylogenetic analysis. Maximum Likelihood phylogenetic analysis was performed in PhyMl V3.0 [33] using JTT + G + I + F model as the best evolutionary model, as determined by Smart Model Selection (SMS) [34]. The branch support for the phylogenetic tree was calculated using 1000 bootstraps. The resulting tree was visualized in FigTree V1.3.1 (available at http://tree.bio.ed.ac.uk/software/figtree/, accessed on 15 February 2021).

Isolation and Cloning of PPARγ Gene Paralogues of the European Sardine
Liver tissue from a single specimen of European sardine was collected and preserved in RNAlater at −20 • C. Total RNA was then extracted using Illustra RNAspin Mini RNA Isolation Kit (GE Healthcare, Chicago, IL, USA), following manufacturer's instructions. Isolated RNA was treated with RNase-free DNase I and eluted with RNase-free water. After RNA quality and quantity assessment using a microplate spectrophotometer (Take 3 and Synergy HT Multi-Mode Microplate Reader, Biotek, Agilent, Santa Clara, CA, USA), 500 ng of liver RNA was used for cDNA synthesis with iScript TM cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA), considering the manufacturer's recommendations.
Two gene sequences of PPARγ-like were identified in a liver transcriptome dataset of S. pilchardus [39]. Two sets of specific primers, including regions for XbaI and KpnI restriction enzymes, were designed to isolate the ligand binding domain (LBD) of both paralogous genes in sardine. Phusion Flash PCR Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) and specific primer pairs (PPARγA Fw-5' TTAATCTAGATTGGCG-GAGGTCTCGG 3' and PPARγA Rv-5' GAGCGGTACCCTAGTACAAGTCTCTGATGATC 3'; PPARγB Fw-5' AAATTCTAGACTGGGTGAAATTGCCAGG 3' and PPARγB Rv-5' AAATGGTACCCTAGTAGAGGTCCCTCATGA 3') were then used in a polymerase chain reaction (PCR) that comprised an initial step at 98 • C for 10 s followed by 40 cycles at 98 • C for 1 s, 58 • C (PPARγA) or 56 • C (PPARγB) for 5 s, and 72 • C for 15 s, with a final extension step for 60 s. At the end, PCR products were loaded onto a 2% agarose gel, stained with GelRed, and those corresponding to the predicted size were cut and purified with NZYGelpure kit (NZYTech, Lisbon, Portugal).
PCR products and the cloning vector pBIND (AF264722; Promega, Madison, WI, USA) were then digested with XbaI and KpnI (NZYTech) and ligated with T4 ligase (Promega) to produce two GAL4 DBD/PPARγ LBD hybrid proteins, one for each sardine PPARγ gene sequence, featuring the DNA binding domain (DBD) of GAL4 and the LBD of one of the PPARγ paralogues. The produced hybrid protein will conditionally bind to the upstream activation sequences (UAS) of a second experimental vector containing a luciferase reporter gene (pGL4.31; DQ487213). Besides the hybrid protein, the pBIND vector also expresses Renilla luciferase, used to normalize firefly luciferase activity. The plasmid sequences were confirmed by Sanger sequencing (Eurofins Genomics, Ebersberg, Germany).
For luciferase-based reporter transactivation assays, COS-1 cells were seeded 24 h before transfection in 24-well culture plates at a density of 2 × 10 5 live cells/mL in DMEM with phenol red supplemented as described before. The next day, cells were transfected with 500 ng of pBIND and 1000 ng of pGL4.31, using Lipofectamine 2000 reagent (Thermo Fisher Scientific), in Opti-MEM reduced-serum medium (Gibco, Carlsbad, CA, USA), following the manufacturer's instructions. After 5 h of incubation, transfection medium was removed and cells were exposed to 1 mL of DMEM without phenol red (PAN-Biotec), previously supplemented with 10% of charcoal-treated fetal bovine serum (PAN-Biotec) and 1% of penicillin/streptomycin, containing 1 µL of each test compound (explained in detail below). Dimethyl sulfoxide (DMSO; Sigma-Aldrich, St. Louis, MO, USA) was used as solvent control, at a final concentration of 0.1% per well. After 24 h of exposure, cells were lysed and firefly luciferase (pGL4.31) and Renilla luciferase (pBIND) activities were determined using a Dual Luciferase Assay System kit (Promega, Madison, WI, USA), considering manufacturer's recommendations, in a Synergy HT Multi-Mode Microplate Reader.
All assays were repeated three times independently, with two technical replicates per condition being performed each time. Beyond the European sardine PPARγA and PPARγB, these assays were also performed with human PPARγ (kindly provided by Ana Capitão) for comparison purposes, and to guarantee the reliability of the results. . Two dilutions (50% and 25%) of a mixture composed of seven FAs (FA mix; 42.9 mM ARA, 42.9 mM EPA, 42.9 mM DHA, 42.9 mM oleic acid, 42.9 mM ALA, 42.9 mM γ-linolenic acid and 11.4 mM palmitic acid) were also tested. All the stock solutions and respective dilutions were prepared in DMSO, with its final concentration being 0.1% in each well.

Statistical Analysis
To determine the statistical significance of the transactivation results, firefly luciferase activity (reported by pGL4.31) was first normalized using Renilla luciferase activity (internal control for transfection efficiency expressed by pBIND). Final results were presented as fold-induction variations with respect to the solvent control (DMSO). Normalized values of the different replicates were Ln(Log e)-transformed, and a one-way ANOVA followed by Tukey's post hoc test were performed, using Sigma Stat (version 11, SPSS Inc.), to compare the results. A p-value (p) ≤ 0.05 was considered statistically significant. Results are presented as mean ± standard error of the mean (SEM).

Homology Modelling
The tridimensional models of the LBD of the European sardine PPARγA and PPARγB were deduced using a Swiss-Model homology modelling workspace in alignment mode [40,41]. Human PPARγ crystal structure 3WJ4 was selected as the scaffold template. The quality of the obtained protein structure model was estimated using sequence identity, GMQE (global model quality estimation) and QMEAN (qualitative model energy analysis) [42,43]. Models were visualized, inspected, and aligned to human crystal structure 3WJ4 in PyMOL v1.74 [44].

Phylogenetic and Synteny Analyses
The Maximum Likelihood phylogenetic analysis ( Figure 1) performed with 53 PPARγ sequences, 7 PPARα sequences and 8 PPARβ, revealed two different PPARγ clades: one including mammal, bird, reptile and amphibian PPARγ genes, and the other containing fish PPARγs. In the fish PPARγ group, the non-teleost fish spotted gar (L. oculatus) branched off from the teleost PPARγ clade. In agreement with the 3R WGD (third Round of Whole-Genome Duplication), two branches diverged in the teleost PPARγ group, corresponding to the PPARγ gene duplication in the last common ancestor of teleosts. Hence, both PPARγA and PPARγB paralogous genes are identified in some teleost orders, namely, Salmoniformes (Oncorhynchus mykiss, Oncorhynchus kisutch and S. salar), Clupeiformes (C. harengus and S. pilchardus), Siluriformes (I. punctatus), Characiformes (A. mexicanus and P. nattereri) and Esociformes (E. lucius). Furthermore, duplicate PPARγB genes (PPARγB1 and PPARγB2) were observed in the salmonids as result of the salmonid-specific fourth round (4R) of WGD ( Figure 1).  The retention or loss of PPARγ paralogous genes was also confirmed by examining PPARγA and PPARγB loci composition and neighborhood in 10 teleosts and in a nonteleost (L. oculatus) fish species (Figure 2). Synteny analysis showed that PPARγA and PPARγB neighboring genes are mostly conserved, even in teleosts retaining a single PPARγ paralogous copy (Figure 2).

Relative Gene Expression Analysis
We next examined the relative gene expression (RGE) of the two PPARγ paralogues using the European sardine as a model. RGE analysis was performed using 11 tissues (Figure 3). Our analysis shows that PPARγA is ubiquitously expressed in all tested tissues, contrarily to PPARγB that showed higher expression in midgut, liver and head kidney,

Relative Gene Expression Analysis
We next examined the relative gene expression (RGE) of the two PPARγ paralogues using the European sardine as a model. RGE analysis was performed using 11 tissues (Figure 3). Our analysis shows that PPARγA is ubiquitously expressed in all tested tissues, contrarily to PPARγB that showed higher expression in midgut, liver and head kidney, and residual expression in kidney, brain and pituitary gland and white muscle. No PPARγB expression was observed in spleen, red muscle, gonad, gills and branchial arches or caudal fin tissues. In general, the comparison of RGE among tissues, especially in kidney and brain and pituitary gland, suggests higher expression levels of PPARγA than PPARγB. However, in liver, head kidney and white muscle, both paralogues presented similar gene expression ( Figure 3 and Table S2). PPARγB expression was observed in spleen, red muscle, gonad, gills and branchial arches or caudal fin tissues. In general, the comparison of RGE among tissues, especially in kidney and brain and pituitary gland, suggests higher expression levels of PPARγA than PPARγB. However, in liver, head kidney and white muscle, both paralogues presented similar gene expression ( Figure 3 and Table S2).

Transactivation Assays
We next examined the ligand binding profile of European sardine PPARγA and PPARγB in the presence of proposed PPARγ ligands ( Figure 4). Tributyltin (TBT), a known obesogen shown to bind to several vertebrate PPARγ receptors, excluding teleosts [6,30], significantly activated both sardine PPARγ receptors when compared to the solvent control (DMSO), except at the lowest TBT concentration (10 nM). Regarding polyunsaturated FAs (200 µM) and FAs mixtures (50% and 25% diluted), no significant activation or repression of European sardine PPARγ paralogues was detected when compared to the solvent control. Similarly to FAs, neither rosiglitazone nor T0070909 (10 µM), agonist and antagonist of the human PPARγ, respectively, had a significant effect on the European sardine PPARγA or PPARγB modulation (Figure 4) relative to the DMSO. The human PPARγ LBD, used as control, was significantly activated by TBT (100 and 250 nM), FAs (200 µM and 50% and 25% dilution mixes) and rosiglitazone (10 µM), and repressed by T0070909 (10 µM), as previously described ( Figure S1).

Transactivation Assays
We next examined the ligand binding profile of European sardine PPARγA and PPARγB in the presence of proposed PPARγ ligands ( Figure 4). Tributyltin (TBT), a known obesogen shown to bind to several vertebrate PPARγ receptors, excluding teleosts [6,30], significantly activated both sardine PPARγ receptors when compared to the solvent control (DMSO), except at the lowest TBT concentration (10 nM). Regarding polyunsaturated FAs (200 µM) and FAs mixtures (50% and 25% diluted), no significant activation or repression of European sardine PPARγ paralogues was detected when compared to the solvent control. Similarly to FAs, neither rosiglitazone nor T0070909 (10 µM), agonist and antagonist of the human PPARγ, respectively, had a significant effect on the European sardine PPARγA or PPARγB modulation (Figure 4) relative to the DMSO. The human PPARγ LBD, used as control, was significantly activated by TBT (100 and 250 nM), FAs (200 µM and 50% and 25% dilution mixes) and rosiglitazone (10 µM), and repressed by T0070909 (10 µM), as previously described ( Figure S1).

Sequence Analysis and Homology Modelling
To further address the TBT activation profile, previously unreported in teleost fish, we deduced European sardine PPARγA and PPARγB tridimensional structures and examined the conservation of LBD residues reported to interact with TBT, through structural and sequence alignment, using the human PPARγ crystal structure (3WJ4) as a template ( Figure 5). Overall, the TBT-interacting amino acid composition of the teleost sequences are poorly conserved, as previously reported for D. rerio (zebrafish) and P. buchholzi (butterflyfish) [6]. The observed substitutions occur between amino acids with hydrophobic side chains, yet an increase in aromatic ring content is also noted, with smaller hydrophobic amino acids replaced by residues with aromatic hydrophobic side chains (Phe, Trp, or Tyr). While these substitutions maintain the hydrophobic core of the binding pocket, they impose distinct spatial and volume constraints within the binding cavity. For instance, as previously noted, Cys285Tyr replacement in zebrafish and butterflyfish leads to an aromatic ring protrusion into the pocket [6], similar to the Cys285Phe and Leu330Trp substitutions observed in European sardine PPARγA and PPARγB, respectively. More specifically, we examined the conservation status of a signature cysteine in helix 3 (Cys285, H3), with a sulfur-containing side chain shown to bind to the tin atom of TBT [45]. Curiously, this residue, critical for TBT-induced activation in other non-teleost species, is not conserved in the TBT-responsive European sardine PPARγA and PPARγB [6]. Despite the absence of a Cys residue in H3, additional sulfur-containing amino acids are overserved in the β-strand B3 of European sardine PPARγA and PPARγB (Met and Cys, respectively), but not in zebrafish. In the human PPARγ, TBT partial agonist activity was suggested to derive predominantly from interactions with residues in H3 and B3 [45].

Sequence Analysis and Homology Modelling
To further address the TBT activation profile, previously unreported in teleost fish, we deduced European sardine PPARγA and PPARγB tridimensional structures and examined the conservation of LBD residues reported to interact with TBT, through structural and sequence alignment, using the human PPARγ crystal structure (3WJ4) as a template ( Figure 5). Overall, the TBT-interacting amino acid composition of the teleost sequences are poorly conserved, as previously reported for D. rerio (zebrafish) and P. buchholzi (butterflyfish) [6]. The observed substitutions occur between amino acids with hydrophobic side chains, yet an increase in aromatic ring content is also noted, with smaller hydrophobic amino acids replaced by residues with aromatic hydrophobic side chains (Phe, Trp, or Tyr). While these substitutions maintain the hydrophobic core of the binding pocket, they impose distinct spatial and volume constraints within the binding cavity. For instance, as previously noted, Cys285Tyr replacement in zebrafish and butterflyfish leads to an aromatic ring protrusion into the pocket [6], similar to the Cys285Phe and Leu330Trp substitutions observed in European sardine PPARγA and PPARγB, respectively. More specifically, we examined the conservation status of a signature cysteine in helix 3

Discussion
During the evolution of vertebrates, polyploidization events have been proposed to provide the new genetic material for the emergence of novelties [46][47][48][49][50][51]. Although the exact duplication timings have fueled intense debate, the current consensus suggests one shared genome duplication (1R; first Round) between cyclostomes and gnathostomes, with independent duplication taking place in each lineage independently [52]. In teleosts, a 3R WGD (Whole-Genome Duplication) is proposed at the origin of teleosts [53]. Besides this last WGD, a 4R WGD was inferred in salmonids [53][54][55][56]. In the context of NR (Nuclear Receptor) gene evolution, WGD occurrences have been central components. Recently, Capitão and collaborators [6] suggested that PPAR-like sequences present in all vertebrate lineages result from WGD events that occurred before gnathostome radiation and after tunicate divergence, yielding the three PPAR paralogous genes. The consequences of the 3R WGD on PPAR gene numbers has also been previously established for both PPARα and PPARβ genes [8,11,[57][58][59][60][61]. Recently, Wafer and collaborators [13] identified two PPARγ paralogous genes in Mexican tetra (A. mexicanus), suggesting a pronounced loss  [45]) depicted as sticks (down). (B-D) Zebrafish PPARγ, sardine PPARγA and sardine PPARγB homology models with representation of LBD cavities (top), and detailed view of the LBP cavities with the corresponding interacting residues depicted as sticks (down). TBT molecule from crystal structure 3WJ4 is shown in magenta. Sequence identity (%), global model quality estimation (GMQE), QMean are indicated below each model. (E) Human PPARγ residues that interact with TBT and the corresponding residues found in Zebrafish PPARγ, sardine PPARγA and sardine PPARγB. The location of the residues in the structure is indicated (H, helix; B, β-strand). * Indicates a residue substitution in sardine PPARγB, resulting in an aromatic ring protrusion.

Discussion
During the evolution of vertebrates, polyploidization events have been proposed to provide the new genetic material for the emergence of novelties [46][47][48][49][50][51]. Although the exact duplication timings have fueled intense debate, the current consensus suggests one shared genome duplication (1R; first Round) between cyclostomes and gnathostomes, with independent duplication taking place in each lineage independently [52]. In teleosts, a 3R WGD (Whole-Genome Duplication) is proposed at the origin of teleosts [53]. Besides this last WGD, a 4R WGD was inferred in salmonids [53][54][55][56]. In the context of NR (Nuclear Receptor) gene evolution, WGD occurrences have been central components. Recently, Capitão and collaborators [6] suggested that PPAR-like sequences present in all vertebrate lineages result from WGD events that occurred before gnathostome radiation and after tunicate divergence, yielding the three PPAR paralogous genes. The consequences of the 3R WGD on PPAR gene numbers has also been previously established for both PPARα and PPARβ genes [8,11,[57][58][59][60][61]. Recently, Wafer and collaborators [13] identified two PPARγ paralogous genes in Mexican tetra (A. mexicanus), suggesting a pronounced loss of duplicated PPARγ in other teleost lineages. In the present work, by means of comparative genomics and phylogenetic analyses, we investigate the presence of PPARγ paralogous genes in an ample collection of ray-finned fish lineages. As expected, spotted gar (L. oculatus) presented a single PPARγ copy, since this species diverged prior to the 3R WGD event (Figures 1 and 2). Moreover, our analyses showed that two PPARγ paralogous genes ( Figure 2) were retained not only in Characiformes (tetras and piranhas), but also in Salmoniformes (salmons and trouts), Clupeiformes (herrings and sardines), Siluriformes (catfishes), and Esociformes (pikes) taxa (Figure 1). These observations suggest a wider retention of PPARγ paralogues than previously anticipated [13]. Additional salmonid-specific PPARγB duplicates were found in Atlantic salmon (S. salar), coho salmon (O. kisutch) and rainbow trout (O. mykiss), which resulted of the 4R WGD. The presence of both PPARγ paralogues in such lineages and the retention of a single PPARγ paralogue in the early-diverging teleost lineages such as Elopomorpha (e.g., eels) and Osteoglossomorpha (e.g., butterflyfishes), allows us to estimate that the retention of PPARγ duplicated genes occurred at the time of Clupeocephala divergence, approximately 185-190 Million years ago [62]. Thus, similarly to PPARα and PPARβ, PPARγ duplication resulted from 3R WGD, with episodes of gene loss shaping teleost PPARγ gene repertoire.
Human PPARs can be activated by both saturated and unsaturated dietary FAs, as well as lipids involved in intracellular signaling pathways [63][64][65]. When binding to these ligands, PPARs play an important physiological role, with PPARγ being crucial for lipid storage and adipogenesis [66]. To determine the binding profile of European sardine's PPARγA and PPARγB, we assessed the ability of transiently transfected mammalian cells to induce reporter gene transcription upon exposure to ligands able to bind and modulate human PPARγ (see Figure S1). Regarding FAs, no significant responses were observed with the European sardine PPARγ paralogues. This agrees with previous reports with plaice (P. platessa) and gilthead sea bream (Sparus aurata) PPARγs, also yielding no activation upon exposure to FAs [57]. In the chondrichthyan Leucoraja erinacea (little skate) on the other hand, FAs were shown to be potent inducers of PPARγ [6]. These disparate observations could be due to structural constraints in the binding cavity of teleost PPARγs. In fact, the deduced tridimensional structures of the European sardine, zebrafish and butterflyfish PPARγs show low residue conservation, displaying more amino acid residues with bulky aromatic side chains possibly affecting pocket size and architecture, whereas the little skate binding pocket is highly conserved when compared to human PPARγ [6]. Null activation profiles were obtained with the synthetic PPARγ agonist and antagonists, in agreement with previous reports on the modulation of zebrafish PPARγ by rosiglitazone [27]. Sardine PPARγ paralogues were also tested for their susceptibility towards TBT (tributyltin) binding. TBT is a currently restricted organotin compound, long used as an antifouling agent in general maritime industries [67][68][69]. In aquatic environments, exposure to TBT mediates endocrine disruption, causing an abnormal induction of male sex characters in females in gastropod mollusks [70,71] and fishes [72,73]. TBT is also considered an obesogen since it is able to bind and greatly induce PPARγ activity, disturbing lipid and energy metabolism in some vertebrates [6,74,75]. Crystallographic analysis of human PPARγ bound to TBT unraveled a partial agonist binding mode, with the ligand TBT interacting with a sulfur-containing cysteine residue (Cys285) on helix 3 (H3) and establishing hydrophobic interactions with the side chains of two residues (Val339 and Ile341) embedded on the β-strand (B3), inducing conformational changes that allow cofactor binding and transcription [45]. Cysteine residues have been shown to be crucial for the anchorage of TBT, not only in PPARs but also in the Retinoid X Receptor (RXR), through interaction of the tin atom with the highly reactive sulfhydryl group of the cysteine side chain [45,76]. In agreement, teleost PPARγs lacking such signature cysteine (e.g., plaice, zebrafish or butterflyfish) were shown to be unresponsive towards growing concentrations of TBT [6,30]. Furthermore, the substitution of the cysteine residue in human and little skate PPARγs abolished TBT-mediated reporter gene transcription in vitro [6,45]. Yet, both European sardine PPARγA and PPARγB were able to induce reporter gene transcription upon TBT exposure, despite lacking the conserved cysteine in H3 (Figures 4 and 5). This observation challenges the universality of the TBT-dependent PPARγ activation mechanism. Careful analysis of pocket residue composition highlights possible alternatives. In fact, sardine PPARγB exhibits an available cysteine residue in B3 to anchor TBT, whereas PPARγA has a methionine. Similarly to cysteine, methionine is a sulfur-containing amino acid, and despite its lower reactivity could also react with the tin atom [77]. In spite of the consensus on the crucial role of cysteine for TBT binding, unusual binding modes have been previously observed. For instance, in the Echinodermata Paracentrotus lividus, PPARγ is unresponsive to TBT despite the conservation of H3 cysteine, whereas the cysteine-to-tyrosine substitutions observed in teleosts such as zebrafish and in the mollusk Patella depressa yield null or repressive responses, respectively [4][5][6]. Alternative modes for TBT binding have also been proposed for RXR [78]. Thus, future structural studies should address these potential alternative binding architectures.
Although no differences were found in the ligand-binding profile of sardine PPARγA and PPARγB, relative gene expression analysis shows a distinct tissue distribution pattern, with the European sardine PPARγA displaying ubiquitous expression in all analyzed tissues, whereas PPARγB is only expressed in midgut, kidney, liver, head kidney, brain and pituitary gland, and white muscle ( Figure 3 and Table S2). In mammals, PPARγ expression is detected in the gastrointestinal tract (only during embryonic development), liver, kidney, heart, adipose tissues, skeletal muscle, placenta and lung. Further promoter usage and alternative splicing lead to the differential expression of PPARγ splice variants in each of these tissues reviewed in [3,79,80]. Thus, the retention of both PPARγ genes in various teleost lineages could be related, with a distinct spatial and temporal distribution.

Conclusions
In conclusion, this study indicates that 3R WGD led to PPARγ gene duplication, as previously described for PPARα and PPARβ paralogues. PPARγ paralogue retention was first reported in A. mexicanus (Characiformes), however, we suggest that PPARγ paralogue retention affected many other teleost lineages. In spite of the differential tissue distributions and expression patterns of sardine PPARγA and PPARγB, indicating tissue-specific functions, both PPARγ genes presented similar ligand preferences. Nevertheless, the modulation of sardine PPARγ paralogues by TBT challenges current paradigms and emphasizes the need to further address the mechanisms of ligand-binding and the susceptibility of teleost lineages towards organotin exposure.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/genes13010107/s1, Figure S1: In vitro transcriptional activation of Homo sapiens PPARγ, Table S1: Accession numbers of the sequences used to perform the phylogenic analysis, Table S2