Functional Evolution of Clustered Aquaporin Genes Reveals Insights into the Oceanic Success of Teleost Eggs Alba Ferré ,†,1 François Chauvigné ,†,2 Anna Vlasova,3 Birgitta Norberg ,4 Luca Bargelloni ,5 Roderic Guigó ,3,6 Roderick Nigel Finn ,* ,1,7 and Joan Cerdà * ,1 1 Institute of Agrifood Research and Technology (IRTA)-Institute of Biotechnology and Biomedicine (IBB), Universitat Autònoma de Barcelona, Barcelona, Spain 2 Institute of Marine Sciences, Spanish National Research Council (CSIC), Barcelona, Spain 3 Centre for Genomic Regulation (CRG), The Barcelona Institute of Science and Technology, Barcelona, Spain 4 Institute of Marine Research, Austevoll Research Station, Storebø, Norway 5 Department of Comparative Biomedicine and Food Science (BCA), University of Padova, Legnaro, PD, Italy 6 Universitat Pompeu Fabra (UPF), Barcelona, Spain 7 Department of Biology, Bergen High Technology Centre, University of Bergen, Bergen, Norway † These authors contributed equally to this work. *Corresponding authors: E-mails: joan.cerda@irta.cat; nigel.finn@uib.no. Associate editor: Dr. Camilla Whittington Abstract Aquaporin-mediated oocyte hydration is considered important for the evolution of pelagic eggs and the radiative suc­ cess of marine teleosts. However, the molecular regulatory mechanisms controlling this vital process are not fully understood. Here, we analyzed >400 piscine genomes to uncover a previously unknown teleost-specific aquaporin- 1 cluster (TSA1C) comprised of tandemly arranged aqp1aa-aqp1ab2-aqp1ab1 genes. Functional evolutionary analysis of the TSA1C reveals a ∼300-million-year history of downstream aqp1ab-type gene loss, neofunctionalization, and subfunctionalization, but with marine species that spawn highly hydrated pelagic eggs almost exclusively retaining at least one of the downstream paralogs. Unexpectedly, one-third of the modern marine euacanthomorph teleosts selectively retain both aqp1ab-type channels and co-evolved protein kinase-mediated phosphorylation sites in the intracellular subdomains together with teleost-specific Ywhaz-like (14-3-3ζ-like) binding proteins for co-operative membrane trafficking regulation. To understand the selective evolutionary advantages of these mechanisms, we show that a two-step regulated channel shunt avoids competitive occupancy of the same plasma membrane space in the oocyte and accelerates hydration. These data suggest that the evolution of the adaptive molecular regulatory features of the TSA1C facilitated the rise of pelagic eggs and their subsequent geodispersal in the oceanic currents. Article Key words: functional evolution, teleosts, aquaporin, trafficking, 14-3-3 proteins, oocyte hydration. Introduction marine elopomorph teleosts (the ancestors of extant eels, bonefishes, and tarpons) was the pelagic egg, which The most dramatic evolutionary radiation in vertebrate passively drifts in the oceanic currents (Finn and history occurred amongst the true spiny ray-finned fishes Kristoffersen 2007). Upon fertilization, these free-floating, (Euacanthomorphacea) in the late Cretaceous and early single-celled orbs of life become the caretakers of the gen­ Eocene (Maissey 1996; Friedman 2010). Fossil evidence re­ ome, which holds the past, present, and future of the lin­ vealed that this unprecedented radiation transpired pre­ eage. However, they are devoid of any of the organ systems dominantly in marine environments in a fraction of the or avoidance behaviors that develop in juveniles and time it took for birds and mammals to achieve such diver­ adults, yet must equally survive in the hyperosmotic sity (Maissey 1996; Friedman 2010). The sudden appear­ oceanic environment to ensure its continuance. Despite ance of the myriad of adult forms pertaining to the lack of parental care, the selective success of this repro­ hundreds of new families found in fossil sites around the ductive trait certainly became manifest in the euacantho­ world suggests that key traits associated with the adaptive morph marine teleosts, the majority of which regardless of marine radiation of euacanthomorph teleosts were also their systematic affinities, demersal or pelagic habitats, likely associated with an efficient mechanism of dispersal. coastal or oceanic distribution, or tropical or boreal ranges, In this regard, a key reproductive trait that may have inde­ spawn pelagic eggs (Ahlstrom and Moser 1981; Kendall pendently evolved for the first time in the Triassic in et al. 1984). Understanding the molecular adaptations © The Author(s) 2023. Published by Oxford University Press on behalf of Society for Molecular Biology and Evolution. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https:// creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Open Access Mol. Biol. Evol. 40(4):msad071 https://doi.org/10.1093/molbev/msad071 Advance Access publication March 22, 2023 1 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE associated with the evolution of such traits could therefore gene was noted in the gilthead seabream (Sparus aurata) shed light on how teleosts came to dominate the oceans. and African cichlids (Zapater et al. 2013; Finn et al. 2014). The pelagic nature of marine teleost eggs is considered In the seabream, what appeared to be exon ghosts in the to have arisen as an exaptation to the recolonization of 5´-promoter region of aqp1ab indicated that a duplicated seawater (SW) by freshwater (FW) ancestors (Fyhn et al. aqp1ab-type pseudogene existed in this locus (Zapater 1999; Finn et al. 2000; Finn and Kristoffersen 2007). In et al. 2013), while in the African cichlids, each species stud­ this regard, it has been shown that the oocytes of ovipar­ ied retained a tetra-exon copy of the aqp1ab gene in the ous marine teleosts that freely broadcast their eggs at same upstream locus (Finn et al. 2014). spawning undergo substantial hydration during the phase To decipher whether the aqp1ab upstream genes in tel­ of meiotic maturation (Finn and Fyhn 2010; Cerdà et al. eosts were the result of lineage-specific duplications or 2017). Since the yolk osmolality is the same as the body were inherited from a common ancestor, here we initially fluids of the adults and thus strongly hyposmotic to SW, conducted a gene-walking experiment in the seabream. the hydration provides the embryos with a vital pool of This revealed a complete tetra-exon aqp1ab-type gene, water until osmoregulatory cells and organs develop but when ectopìcally expressed in amphibian (Xenopus (Finn et al., 2002a). The level of oocyte hydration is related laevis) oocytes, the translated protein was not trafficked to the degree of proteolysis of vitellogenin-derived yolk to the PM. Surprisingly, however, phylogenetic analyses re­ proteins and is most pronounced in pelagic eggs, typically vealed that the novel seabream aqp1ab-type coding se­ resulting in water contents of >90% of the egg mass, thus quence (CDS) was more closely related to the African reducing the specific gravity below that of the ambient SW cichlid upstream aqp1ab CDS, as well as the previously and causing them to float (Finn 2007; Sullivan and Yilmaz studied Atlantic halibut (Hippoglossus hippoglossus) aq­ 2018). p1ab CDS, which was only functional in native or teleost A facilitated pathway for water entry into the oocyte oocytes, but not when expressed in frog oocytes was discovered almost two decades ago as a teleost- (Zapater et al. 2011). These observations suggested that specific, water-selective aquaporin (Aqp1o—now termed the upstream aqp1ab-type gene was likely derived from Aqp1ab) (Fabra et al. 2005). This channel is temporally in­ a common ancestral duplication, and that a trafficking fac­ serted into the oocyte plasma membrane (PM) during tor present in teleost oocytes was missing in amphibian meiotic maturation and maximal yolk proteolysis, and it oocytes. We, therefore, conducted a systematic phylogen­ is thus essential for the development of the pelagic egg etic survey amongst >400 piscine genomes to uncover the phenotype (Fabra et al. 2006; Zapater et al. 2011, 2013). ancient origin of a hitherto unknown teleost-specific aqp1 The Aqp1ab channel is now known to belong to a broader cluster (TSA1C), comprised of aqp1aa and two aqp1ab- superfamily of aquaporins (Agre et al. 1993), which in type genes. To understand the functional evolution of jawed vertebrates currently comprise 17 subfamilies the channels in the TSA1C, we selected distantly related (Aqp0-16) (Finn et al. 2014; Chauvigné et al. 2019). teleosts as models that retain one or both of the down­ Teleosts, however, retain an even larger repertoire of aqua­ stream aqp1ab-type genes. Using comparative transcrip­ porins, in part due to independent whole genome duplica­ tomics and functional experiments, we reveal how the tions (WGDs), but also due to more ancestral and intracellular subdomains of the TSA1C channels co-evolved lineage-specific tandem duplications (Tingaud-Sequeira with teleost-specific Ywhaz-like regulatory proteins (tyro­ et al. 2008, 2010; Cerdà and Finn 2010; Finn and Cerdà sine 3-monooxygenase/tryptophan 5-monooxygenase acti­ 2011; Finn et al. 2014; Yilmaz et al. 2020). As a result, aqua­ vation proteins or 14-3-3ζ-like) together with protein porin gene clusters became more prevalent in the actinop­ kinase-mediated phosphorylation pathways to differentially terygian lineage with up to seven identified to date (i.e., regulate their membrane trafficking. The findings provide aqp1aa-1ab; aqp3a1-3a2; aqp3b1-3b2; aqp8aa-8ab; new insight into the molecular adaptations facilitating the aqp8ba-8bb; aqp10aa-10ab; and aqp10ba-10bb) compared success of teleost eggs in the oceans. to two in the sarcopterygian lineage (AQP3-7 and AQP2-5-6 type clusters) (Finn et al. 2014; Yilmaz et al. 2020). The actinopterygian aqp1ab gene thus evolved Results and neofunctionalized as a downstream tandem duplicate of the aqp1aa paralog, which retains a more conserved Phylogenomics Uncovers an Ancient aqp1-type protein structure to tetrapod AQP1 (Tingaud-Sequeira Cluster in Teleosts et al. 2008; Zapater et al. 2011). An initial gene-walking experiment using gilthead seab­ The aqp1aa-1ab binary cluster has so far been detected ream genomic DNA uncovered four aquaporin-like exons in all teleosts studied to date, including members of the 219 nucleotides (nt) upstream of the previously identified elopomorph (e.g., eels), ostariophysan (e.g., zebrafishes, aqp1ab channel, and 8,144 nt downstream of the encoded carps, and catfishes), and euteleostean lineages such as aqp1aa gene (Fabra et al. 2005, 2006) (fig. 1A). The de­ protacanthopterygians (e.g., salmons) and euacanthomor­ duced primary structure and initial phylogenetic analysis phaceans (e.g., seabreams, seabasses, cichlids, flatfishes, in relation to 263 aqp1-type piscine CDS assembled from and pufferfishes) (Cerdà et al. 2017; Zhang et al. 2021). In 94 genomes and 42 transcriptomes revealed that an add­ two of these studies, however, a second aqp1ab-type itional aqp1ab-type channel exists in this locus (fig. 1B 2 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE FIG. 1. The origin of the TSA1C. (A) Genomic organization of the TSA1C in distantly related teleosts drawn to scale. Canonical aqp1aa, -1ab2, and -1ab1 genes consist of four exons (blue, green, and magenta bars, respectively), while pseudogenes vary from premature stop codons to highly degraded fragments (e.g., salmon†). In cypriniform teleosts (e.g., zebrafish) aqp1ab2 is absent, while in anguilliform teleosts (e.g., eel) aqp1ab1 is absent. (B) Bayesian majority rule consensus tree of the TSA1C coding sequences rooted with chondrichthyan aqp1. The tree is inferred from 20 million Markov Chain Monte–Carlo (MCMC) generations (nucmodel = 4by4, nst = 2, rates = gamma) of 239,282 aligned nucleotide sites aligned by codon. Support values shown at each node (colored dots) are Bayesian posterior probability ranges shown in the key. The scale bar indicates the expected rate of substitutions per site. (C ) Bayesian majority rule consensus tree of piscine aqp1 orthologs rooted with hyper­ oartian aqp01. The tree is initially inferred from 100 million MCMC generations (nucmodel = 4by4, nst = 2, rates = gamma) of 399,429 aligned 3 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE independent R4 allotetraploidization event (Berthelot et al. 2014; MacQueen and Johnston 2014; Xu et al. 2019). Conversely, the aqp1aa2 gene in soleid and cyno­ glossid flatfishes is not found in the genomes of all other families (n = 7) of flatfish searched but evolved in the ab­ sence of the aqp1ab1 gene, which is present in the gen­ omes of pleuronectid, paralichthyid, scophthalmid, and bothid flatfishes (fig. 1A and B and supplementary fig. S1, Supplementary Material online). In tongue sole (Cynoglossus semilaevis), aqp1aa2 is located on chromosome 15 (GenLoc 271 kb) and is not syntenic to the extant aqp1aa-aqp1ab2 cluster, which is located on chromosome 20 (GenLoc 8,758 kb). From this, we surmise that the aqp1aa2 gene of so­ leid and cynoglossid flatfishes is not an integral member of the TSA1C in these lineages. FIG. 2. The origins of multiple aquaporin-1 orthologs in fishes. Whole Interestingly, the initial phylogenetic analysis (fig. 1B) in­ genome duplication (polyploidy) events (R3, R4, and P), respectively, generated aqp1b ohnologs in teleosts, a paralogous TSA1C in paleo­ dicated that the absence of the aqp1ab1 gene is not re­ tetraploid salmonids and allotetraploid cyprinids and aqp1c2 in stricted to certain families of flatfishes, but it is also chondrostean sturgeons and paddlefishes. Tandem duplication (T ) absent from the genomes of other species, such as elopo­ gave rise to the teleost aqp1ab2 and aqp1ab1 paralogs. morph eels, rivulid killifishes, and perciform sticklebacks, Conversely, lineage-specific duplication (D) generated the aqp1aa2 or exists as a pseudogene in scombrid tunas and perciform paralogs in soleid and cynoglossid flatfishes. rockfishes (supplementary fig. S1, Supplementary Material online). Conversely, aqp1ab2 is absent from the genomes and supplementary fig. S1, Supplementary Material on­ of cyprinid fishes, including zebrafish (Danio rerio), gobiid line). Bayesian inference indicated that nonteleost species mudskippers, and tetraodontiform pufferfishes, or exists as such as cartilaginous, chondrostean, and holostean fishes a pseudogene in the genomes of some pleuronectid, labrid, encode a single aqp1 channel, while all teleosts have at moronid, and perciform teleosts (supplementary fig. S1, least one aqp1aa gene, and different species have one or Supplementary Material online). However, intact aqp1ab2 two copies of aqp1ab genes, which cocluster on a separate and aqp1ab1 genes are found in the genomes of clupeid, branch to the aqp1aa genes. In line with the convention of characiform, siluriform, cichliform, and pleuronectiform aquaporin nomenclature, which names genes in accord­ teleosts, suggesting that the TSA1C could have arisen in ance with the chronology of their discovery, we named the common ancestor of all clupeocephalan teleosts the second seabream aqp1ab-type channel and its ortho­ (Otomorpha and Euteleosteomorpha). Since recent stud­ logs in other species aqp1ab2, and renamed the original ies have revealed that aquaporin tandem duplicates of seabream aqp1ab channel (Fabra et al. 2005, 2006) and aqp3, aqp8, and aqp10 evolved in preteleostean lineages its orthologs in other species aqp1ab1. These analyses (Cerdà and Finn 2010; Finn et al. 2014; Yilmaz et al. thus uncover a novel ternary TSA1C comprised of 2020), we conducted a separate phylogenetic analysis of aqp1aa-aqp1ab2-aqp1ab1, in which the aqp1ab2 and aq­ aqp1-type genes and their syntenic loci assembled from p1ab1 genes, when present in teleost genomes, are closely the genomes of nonteleost fishes in relation to those of juxtaposed downstream of the aqp1aa gene (fig. 1A). the elopomorph, osteoglossomorph, and clupeid teleosts. The above analyses also revealed that some species of The results confirmed that lampreys retain a single teleost harbor a second type of aqp1aa gene, including so­ fused ortholog (aqp01) (Finn et al. 2014), while cartilagin­ leid and cynoglossid flatfishes, salmonids, and some cypri­ ous, sarcopterygian, and actinopterygian cladistian and ho­ nids. In each case, however, the second aqp1aa gene, lostean fishes retain a single aqp1-type ortholog which we term aqp1aa2, coclusters within these specific (supplementary fig. S3, Supplementary Material online). lineages, and can thus be explained based on lineage- However, two orthologs were detected in the chondros­ specific duplication. Thus aqp1aa2 in salmonids is consist­ tean sturgeons and paddlefishes, and four orthologs in­ ent with a fourth round (R4) of autotetraploidization, cluding one or two pseudogenes were identified in while that of selected cyprinids is consistent with an osteoglossomorph teleosts such as the Aba (Gymnarchus nucleotide sites aligned by codon (N = 494 taxa), with the Acanthomorphata segment of the tree re-inferred from a further 100 million MCMC generations (nucmodel = 4by4, nst = 2, rates = gamma) of 235,624 aligned nucleotide sites aligned by codon (N = 301 taxa). Whole genome duplication events (R3, R4) are indicated at relevant nodes. Support values and scale bar as in (B). (D) Syntenic arrangement of the TSA1C in selected fishes illustrating full (yellow backgrounds), partial (presence of pseudogenes), or no conservation of the cluster (both aqp1ab-type paralogs lost). Gene coding direction is indicated by the pointed end of each symbol with pseudogenes indicated by a white X on the symbol. Lost genes are indicated with an X. Both paralogons are shown for paleotetraploid salmonids. Asterisks indicate that some species within the group have aqp1ab-type pseudogenes. Red arrows highlight species selected for experimental studies. TPE: transposable element. 4 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE niloticus) or mormyrid elephantfishes (Mormyridae). Teleostei (fig. 2), in the near aftermath of the R3 event Conversely, three intact orthologs were identified in elopo­ ∼300 Ma (Amores et al. 2011). The TSA1C is thus ancient morph and clupeid teleosts. The phylogenetic topology of but has until now been masked due to differential gene the chondrostean duplicates shows that they form a bin­ loss in various lineages. To determine whether a retention ary cluster (posterior probability [pp] = 1.0) on a sister bias of either the aqp1ab2 or aqp1ab1 genes occurred dur­ branch to the single cladistian orthologs (pp = 1.0; ing teleost evolution, we conducted a systematic phylo­ supplementary fig. S3A, Supplementary Material online), genetic analysis of the aqp1-type genes (fig. 1C, and and are located on separate chromosomes (supplementary supplementary fig. S2 and S4, Supplementary Material on­ fig. S3B, Supplementary Material online). This is consistent line) and their syntenic loci (fig. 1D) assembled from >400 with the expected taxonomic position of Chondrostei, and piscine genomes. Bayesian inference of the aqp1aa, aqp1b, an independent paleopolyploidy events in the lineage and nonteleost aqp1 CDS, recovered the major groupings (Crow et al. 2012; Du et al. 2020). We, therefore, concluded of piscine diversity, including the majority (85%) of teleost that the chondrostean paralogs are not associated with the orders (Hughes et al. 2018), and confirmed the asymmetric TSA1C, and named them aqp1c1 and aqp1c2. retention of the aqp1aa gene in all teleosts, but only the The phylogenetic topology of the teleost orthologs also re­ aqp1b ohnolog in osteoglossomorph and elopomorph spe­ veals that one of the CDS assembled from the elopomorph cies. The tree further recovered the topology of the R4 du­ and osteoglossomorph genomes clusters on a sister branch plicates in salmonids and selected cyprinids and the to the aqp1aa paralogs (pp = 0.8; supplementary fig. S3A, aqp1aa2 duplicates in soleid and cynoglossid flatfishes. Supplementary Material online), yet each gene is located By contrast, a high number of pseudogenes of either aq­ on a separate linkage group to the aqp1aa-aqp1ab2-aqp1ab1 p1ab2 (n = 105; supplementary fig. S4A, Supplementary paralogs, which are all tandemly arranged in the TSA1C Material online) or aqp1ab1 (n = 72; supplementary fig. (supplementary fig. S3B, Supplementary Material online). S4B, Supplementary Material online) were identified, Based upon the semiconserved synteny, we concluded that with no evidence of either paralog in 41 species. The loss this novel aqp1-type ortholog found in elopomorph and os­ or fragmentation of aqp1ab-type genes is thus not related teoglossomorph teleosts represents the previously undetect­ to the systematic affinities of teleosts, but appears to have ed ohnolog (i.e., originating from R3, the teleost-specific occurred stochastically throughout teleost evolution ei­ WGD, Amores et al. 2011; Gasanov et al. 2021), and we, there­ ther in association with paleotransposon activity in salmo­ fore, named it aqp1b. This ohnolog was not found in any of nids and eels, or through mutation load in other species. the clupeocephalan genomes searched (n = 365), suggesting Nevertheless, 59 species from 16 orders of teleosts still re­ that it is probably lost in the vast majority (93%) of extant tain both the aqp1ab2 and aqp1ab1 paralogs within the teleost families. TSA1C. Since the biological significance of clusters of re­ The above analyses revealed that the genomes of elopo­ tained genes remains unclear (Albalat and Cañestro morph teleosts such as bonefishes, tarpons, and deep-sea 2016), we investigated whether aqp1ab-type gene reten­ eels retain the aqp1b ohnolog together with a binary clus­ tion could be related to the previously established vital ter of aqp1aa-aqp1ab2 paralogs. However, although function of oocyte hydration and the formation of pelagic catadromous anguillimorph eels also retain the aqp1aa- eggs (Fabra et al. 2005, 2006; Kagawa et al., 2011; Zapater aqp1ab2 cluster, the aqp1b ohnolog was not found in their et al. 2011, 2013). genomes and all elopomorph teleosts have seemingly lost the downstream aqp1ab1 tandem duplicate. Interestingly, identification of a Tcb1-type transposase in the putative Selective Retention and Expression of aqp1ab Genes aqp1ab1 locus suggests that gene deactivation and loss in Marine Pelagophil Teleosts may have been associated with paleotransposon activity. Data on the spawning habitat (SW or FW) and type of egg Conversely, with the exception of the Aba, which appears spawned, that is pelagic, benthic, or internal, were com­ to have a complete form of aqp1ab2, the genomes of piled for 425 species of teleosts (supplementary table S1, most osteoglossomorph teleosts still encode intact Supplementary Material online). Pelagic eggs were defined aqp1b and aqp1aa ohnologs, but have fragmented pseu­ as eggs that float freely without external buoyancy aids dogenes of the aqp1ab2 and aqp1ab1 tandem duplicates such as bubble nests or egg rafts. Benthic eggs included (supplementary fig. S3, Supplementary Material online). all eggs that sink in a nonlotic environment, including Amongst clupeid teleosts, the three extant aqp1-type those that are spawned on, or are attached to substrate, genes are all tandemly located as a ternary cluster that egg rafts, bubble nests, or are collected by mouth brooders. forms the TSA1C, with each of the aqp1aa, aqp1ab2, Internal eggs were defined as those that are not exposed to and aqp1ab1 paralogs phylogenetically clustering with the external environment, including those of ovovivipar­ the same topology and high pp as shown in figure 1B. ous and viviparous species, as well as species that incubate The only pseudogene detected was identified as a frag­ their eggs in brood pouches (e.g., seahorses and pipefishes). mented aqp1ab1 gene in the denticle herring (Denticeps The compiled data set reveals that approximately half clupeoides). (51%) of the species spawn in SW, and half (49%) spawn Taken together, the above analyses reveal that the in FW. However, as evident in figure 1C, there is a system­ TSA1C likely evolved at the root of the crown clade of atic bias, where the more basal lineages of the 5 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE noneuacanthomorph teleosts mostly (79%) spawn in FW, negative controls, and subsequently submitted to swelling while the majority (65%) of the modern euacanthomorph assays (fig. 3). Oocytes expressing Aqp1aa from zebrafish teleosts spawn in SW. An intriguing observation emerges (DrAqp1aa), European eel (AaAqp1aa), Atlantic salmon when examining the euacanthomorph species (n = 217) (SsaAqp1aa), gilthead seabream (SaAqp1aa), Atlantic hali­ in relation to the type of egg spawned and retention of but (HhAqp1aa), or Senegalese sole (SseAqp1aa) showed a aqp1ab-type genes. A higher proportion (60%) of the mar­ 6–11-fold increase in osmotic water permeability (Pf) with ine species from this group spawn pelagic eggs (pelago­ respect to the control oocytes after a hypoosmotic chal­ phils) as opposed to benthic (benthophils) (26%) or lenge, whereas oocytes injected with SseAqp1aa2 showed internally incubated eggs (14%) (supplementary fig. S5A, a Pf increment of ∼3 times (fig. 3A). However, with respect Supplementary Material online). Conversely, a greater frac­ to the controls, oocytes expressing DrAqp1ab1, SsAqp1ab1, tion (81%) of the FW species are benthophils as opposed SaAqp1ab1, or HhAqp1ab1 only showed ∼4-fold increase in to pelagophils (1%), and there is no significant difference Pf, while those expressing AaAqp1ab2, SaAqp1ab2, between the proportions of species in SW (14%) or FW HhAqp1ab2, SseAqp1ab2, or Aqp1ab2 from the stinging (18%) with internal egg development. Most striking is catfish (HfAqp1ab2), displayed even less swelling with that nearly all of the pelagophils (95%) retain at least one ∼1-fold increase in Pf (fig. 3B and C and supplementary of the aqp1ab2 or the aqp1ab1 genes, with a third of the fig. S7A and B, Supplementary Material online). species (33%) retaining both paralogs (supplementary fig. To investigate the intracellular trafficking to the X. laevis S5B, Supplementary Material online). This level of gene re­ oocyte PM of the TSAC1channels, we employed immuno­ tention is much higher when compared to that of the blotting and immunofluorescence microscopy using benthophils (57% retain 1 paralog; 12% retain both para­ paralog-specific or anti-HA antibodies. These experiments logs), and greatly exceeds that of species that incubate their revealed that teleost Aqp1aa polypeptides were constitu­ eggs internally, where no species retains both paralogs, and tively expressed at the PM (fig. 3D). In contrast, the 91% no longer possess the TSA1C. Quantitative analyses of Aqp1ab1 and -1ab2 channels were differentially retained the mRNA expression levels in different adult tissues further in the cytoplasm (fig. 3E and F), thus explaining the lower revealed that aqp1ab-type transcripts are highly enriched in Pf of oocytes expressing these paralogs compared to those the ovaries of marine pelagophil species, such as pleuronec­ injected with Aqp1aa. Double immunostaining of tids, soleids, and sparids, but not those of FW benthophils Aqp1ab1 or -1ab2 with the endoplasmic reticulum (ER) such as salmonids and cyprinids (supplementary fig. S6, marker protein disulfide isomerase (PDI) further revealed Supplementary Material online; Tingaud-Sequiera et al. that both channels colocalized with the PDI (fig. 3G and 2008). H and supplementary fig. S7C and D, Supplementary Material online), suggesting an impaired sorting of these Intracellular Trafficking of the TSA1C is Differentially aquaporins from the ER to the frog oocyte PM. Transient Regulated expression of flag-tagged Aqp1aa, -1ab1, or -1ab2 from dif­ To better understand the evolutionary basis of the above ferent teleosts in the human breast cancer cell line observations and to investigate how species that retain Michigan Cancer Foundation-7 confirmed the constitutive both aqp1ab1 and aqp1ab2 paralogs can avoid expression expression of teleost Aqp1aa in the PM, as well as the dif­ redundancy of their channel products, we functionally ferential retention of Aqp1ab1 and -1ab2 in the ER characterized each channel in the TSA1C from distantly re­ (supplementary fig. S8A–C, Supplementary Material on­ lated marine and FW teleosts. These species included line). However, these trials also showed a stronger coloca­ those that retain one of the downstream aqp1ab-type lization of Aqp1ab2 with the lysosomal-associated genes, such as the elopomorph European eel (Anguilla an­ membrane protein 1 compared to that of Aqp1ab1 guilla), the ostariophysan zebrafish, and stinging catfish (supplementary fig. S8D and E, Supplementary Material (Heteropneustes fossilis), the protacanthopterygian online), indicating that the Aqp1ab2 channels are also rap­ Atlantic salmon (Salmo salar) and the euacanthomorph idly redirected to lysosomal degradation. Senegalese sole (Solea senegalensis), and two euacantho­ morph marine teleosts that retain both aqp1ab-type Protein Kinases Regulate the Intracellular Trafficking genes, the Atlantic halibut and the gilthead seabream. of the TSA1C Channels These species also spawn different types of eggs with zeb­ Since intracellular trafficking of aquaporins is often rafish, salmon, and catfish producing benthic eggs, while regulated by protein kinase-mediated phosphorylation of sole, seabream, and halibut produce highly hydrated pela­ subdomain-specific residues (Nesverova and Törnroth- gic eggs. Horsefield 2019), we initially tested whether protein kinase To determine whether the TSA1C paralogs from each A (PKA) and C (PKC) activators, forskolin (FSK), and phor­ model species encode functional water channels, we bol 12-myristate 13-acetate (PMA), respectively, could used oocytes from X. laevis as standard surrogate experi­ regulate the trafficking of Aqp1aa, -1ab1, or -1ab2 mental systems for the fish oocytes (Chauvigné et al. from each model teleost species in X. laevis oocytes 2021). Oocytes were injected with the corresponding (supplementary fig. S9, Supplementary Material online). cRNAs, synthesized from nontagged or human influenza Subsequently, to identify the potential amino acid residues hemagglutinin (HA)-tagged cDNAs, or with water as involved, we conducted in silico searches for putative PKA- 6 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE FIG. 3. Functional characterization of Aqp1-like channels in teleosts. (A–C) Osmotic water permeability (Pf) of X. laevis oocytes injected with water (controls) or expressing WT Aqp1aa, -1ab1, or -1ab2 paralogs from different teleosts. All data points are presented as box and whisker plots/scatter dots with horizontal line (inside box) indicating median and outliers. The number of oocytes in each group is indicated on the left of each plot. Data were statistically analyzed by an unpaired Student’s t-test (**P < 0.01; ***P < 0.001; with respect to controls). (D–F) Immunolocalization (upper panels) and immunoblot (lower panels) of the different intact or HA-tagged channels (as indicated) in oocytes. Proteins were detected by using an anti-HA commercial antibody or paralog-specific antibodies. Molecular mass markers (kDa) are on the left. TM, total membrane; PM, plasma membrane. (G, H) Double immunostaining of oocytes expressing zebrafish or seabream Aqp1ab1, or European eel or seabream Aqp1ab2, and the ER marker PDI. The PM of oocytes is indicated by arrows (D–H), whereas the arrowheads (G and H) indicate colocalization of the channels with the PDI. Scale bars, 25 µm. or PKC-mediated phosphorylation sites in the intracellular to the presence of two Thr-Thr residues in loop D, whereas loop D and C-termini of the channels, followed by site- that of Aqp1ab2 was stimulated by PKC but only in the directed mutagenesis experiments (supplementary fig. elopomorph European eel (AaAqp1ab2). Activation of S10, Supplementary Material online). The data confirmed PKA promoted Aqp1ab1 PM trafficking in all the teleost that PKC positively regulates PM trafficking of human species tested, however, the involved phosphorylation AQP1, as well as the Aqp1aa orthologs from elopomorph sites could not be determined using current prediction al­ and cyprinid teleosts (AaAqp1aa and DrAqp1aa) through gorithms. The same positive effect of PKA on channel traf­ phosphorylation of conserved Thr residues in loop D. In ficking was only conserved in Aqp1ab2 from species that contrast, in modern euteleost lineages (SaAqp1aa, retain only this paralog, such as European eel and sole. SseAqp1aa, and SseAqp1aa2), this mechanism switches Interestingly, we also identified p38 mitogen-activated to negative regulation of Aqp1aa trafficking due to substi­ protein kiknase (MAPK) putative phosphorylation sites tution of the upstream Thr by a Val residue. These experi­ conserved in the Aqp1ab1 C-termini of most (74%) of ments also revealed that PKA mediates positive membrane the euteleost species examined (supplementary fig. S10F, trafficking of Aqp1aa orthologs from elopomorph and Supplementary Material online), which reduce PM traf­ cyprinid teleosts through phosphorylation of conserved ficking (Tingaud-Sequeira et al. 2008). Such p38 MAPK C-terminal Ser residues, but not in euacanthomorph spe­ sites are not found in the otophysan lineages of siluriform cies (SaAqp1aa and SseAqp1aa2). By contrast, the traffick­ and clupeiform teleosts that also retain intact aqp1ab1 ing of the Aqp1ab1 paralog was negatively regulated by genes. These data thus suggest that the intracellular traf­ PKC in the cyprinid and salmonid teleosts (DrAqp1ab1 ficking of the TSA1C is differentially regulated by PKA and SsaAqp1ab1), which surprisingly appeared to be linked and PKC, and also by p38 MAPK in the case of Aqp1ab1, 7 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE FIG. 4. Identification of teleost-specific YWHA proteins and regulation of Aqp1ab1 and -1ab2 trafficking. (A) Bayesian majority rule midpoint rooted consensus tree (3 million MCMC generations inferred from 278,404 nucleotide sites) from 337 gnathostomes YWHA coding sequences. Posterior probabilities are shown at selected nodes (see supplementary figs. S13 and S14, Supplementary Material online for fully annotated trees). X. laevis YWHAZ and flatfish ywhazLa and -zLb branches are respectively indicated. (B and C) Expression levels were assessed as reads per kilo base per million mapped reads (RPKM) of transcripts encoding YWHA proteins in the X. laevis and Atlantic halibut ovarian follicle tran­ scriptomes. (D, H ) Pf of X. laevis oocytes injected with water or expressing WT SaAqp1ab1 (D) or SaAqp1ab2 (H ), with or without co-injection of HA-tagged seabream YwhazLa or -zLb, respectively, and treated with DMSO, FSK (PKA activator), or PMA (PKC activator). (E, I ) Schematic re­ presentation of SaAqp1ab1 and -1ab2 monomers showing the six transmembrane domains, the five connecting loops (A-E), and the predicted YWHA-binding regions (underlined) in the C-termini (see also supplementary fig. S10, Supplementary Material online). Residues shaded in blue and red indicate putative PKA and p38 MAPK-phosphorylation sites, respectively. (F, J ) Pf of oocytes injected with water or expressing WT 8 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE but that the specific residues involved in these regulations amphibians have orthologs of the mammalian YWHAB, appear to have changed with the evolution of the different YWHAE, YWHAG, YWHAH, YWHAQ, and YWHAZ subfam­ lineages. ilies (fig. 4A, and supplementary fig. S13, Supplementary Material online). Teleost duplicates were identified in Novel Teleost Ywhaz-Like Binding Proteins Regulate each subfamily, except YWHAH, with the longest branch Aqp1ab1 and -1ab2 Trafficking lengths in relation to the single X. laevis orthologs noted Although the previous experiments could not determine for YWHAQ and YWHAZ. The YWHAQ cluster recovered the specific mechanisms involved in the trafficking regula­ the expected interrelationships of Chondrichthyes, tion of Aqp1ab1 and -1ab2 in some teleosts, experiments Actinopterygii, and Sarcopterygii, however, the duplicated using chimeric proteins indicated that the Aqp1ab1 and ywhaqa and ywhaqb orthologs were restricted to noneua­ -1ab2 C-termini are mechanistically associated with the canthomorph species, with only the ywhaqb gene found in cytoplasmic retention of the channels in amphibian the Euacanthomorphacea. By contrast, both of the teleost oocytes (supplementary fig. S11, Supplementary Material duplicates within the YWHAZ cluster are found in the online). Further in silico analysis of the C-termini of teleost Euacanthomorphacea but formed more distantly related Aqp1ab1 and -1ab2 revealed the presence of potential clusters external to the nonteleost actinopterygian, binding motifs for YWHA (14-3-3) regulatory proteins chondrichthyan, and sarcopterygian CDS. We, therefore, (supplementary fig. S10F and J, Supplementary Material named these novel teleost-specific duplicates Ywhaz-like-a online), which can bind membrane proteins in a (ywhazLa) and Ywhaz-like-b (ywhazLb). Using this initial PKA-mediated Ser/Thr phosphorylation-dependent man­ analysis as a framework, we searched the halibut and X. lae­ ner to regulate their function and subcellular sorting (Fu vis transcriptomes to establish which of the YWHA genes et al. 2000; Moeller et al. 2016; Prado et al. 2019). The are expressed in the respective ovarian follicles. The orthol­ YWHA interaction motif in the target protein usually con­ ogy of each identified transcript was confirmed by Bayesian tains an Arg residue that increases binding affinity (Rajan inference (supplementary fig. S14A–E, Supplementary et al. 2002; Shikano et al. 2005). Therefore, to investigate Material online), and revealed that the YWHAB, YWHAE, whether YWHA proteins could regulate teleost Aqp1ab1 YWHAQ, and YWHAZ paralogs are each expressed in the and -1ab2 trafficking, as well to establish which specific re­ X. laevis follicle with YWHAE showing the highest expression sidues could be involved in these mechanisms, we con­ (fig. 4B). In the halibut follicle, the ywhaqb, ywhazLa, and ducted a mutagenesis screen in which the C-terminal ywhazLb transcripts showed the highest expression, where­ Arg within the predicted YWHA motif was substituted as ywhaba, ywhaea, ywhaeb, and ywhagb were expressed at by Ala, and all potential phosphorylation sites in this sub­ relatively lower levels (fig. 4C). domain were separately replaced by Asp to mimic a con­ The above results highlighted YwhazLa and -zLb as can­ stitutive phosphorylation state (supplementary fig. S12, didate binding proteins that could have duplicated and Supplementary Material online). The results of these ex­ neofunctionalized to a degree that is specific to modern periments for the Aqp1ab1 channels appeared to be puz­ euacanthomorph teleosts. Thus, although X. laevis oocytes zling among species, but the data did suggest that binding express orthologous YWHA proteins that can facilitate the of Aqp1ab2 to a teleost-specific YWHA carrier protein heterologous membrane trafficking of the teleost Aqp1aa might be involved in the regulation of channel trafficking. and -1ab1 paralogs, amphibian YWHA proteins may no To identify teleost-specific YWHA paralogs that could longer possess the specificity required for trafficking the facilitate Aqp1ab1 and -1ab2 membrane trafficking, we Aqp1ab2 channel. To test this hypothesis, we conducted conducted a comparative transcriptomic analysis of further experiments using site-directed mutagenesis and Atlantic halibut postvitellogenic and X. laevis stage IV co-immunoprecipitation to detect aquaporin-YwhazL in­ ovarian follicles by RNA sequencing (supplementary teractions, where seabream YwhazLa and -zLb cRNAs table S2, Supplementary Material online). In parallel, we were co-expressed with those of Aqp1ab1 and -1ab2 of dis­ further constructed a phylogenetic tree of the vertebrate tantly related teleosts in X. laevis oocytes. The results sug­ YWHA-binding proteins with a view to finding teleost- gested that PM trafficking of teleost Aqp1ab1 channels specific duplicates that display long branch lengths and (DrAqp1ab1, SsaAqp1ab1, SaAqp1ab1, and HhAqp1ab1) thus potential neofunctionalization in relation to the am­ in response to PKA phosphorylation can be regulated by phibian orthologs. Bayesian inference of the CDS of the endogenous frog YWHA proteins, but that this mechan­ YWHA-encoding transcripts shows that teleosts and ism is more efficiently controlled by the binding to SaAqp1ab1 (F) or WT SaAqp1ab2 (J ), or mutant channels at specific residues potentially involved in YwhazL binding, and treated with FSK alone or in the presence of the p38 MAPK activator ANS. Control and aquaporin expressing oocytes were also co-injected with YwhazLa-HA or -zLb-HA as above. (G, K ) Co-immunoprecipitation of Aqp1ab1 and YwhazLa (G), or Aqp1ab2 and YwhazLb (K), in oocytes treated as in F and J, respectively. The IP using an anti-HA or anti-SaAqp1ab1 and -1ab2 specific antibodies (in G and K, respectively) were immunoblotted with appropriate antibodies (as indicated). Molecular mass markers (kDa) are on the left. In D, H, H, and J, data are presented as box and whisker plots with the number of oocytes in each group indicated on the left or below each plot. Data were analyzed by the unpaired Student’s t-test (*P < 0.05; **P < 0.01; ***P < 0.001; as indicated in brackets). In D and H, letters indicate significant differences in one-way analysis of the variance (ANOVA) (P < 0.05). ns, statistically not significant. 9 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE FIG. 5. Aqp1ab1 and -1ab2 PM trafficking and YwhazL interaction is regulated during seabream oocyte hydration in vivo and vitro. (A–C) Double immunostaining of Aqp1ab1 (green) and YwhazLa/b (red), and Aqp1ab2 (green) and YwhazLb (red), in ovarian follicles enclosing vitellogenic (Vit; A), hydrating (Hyd; B), or mature (Mat; C) oocytes. The granulosa (arrowheads) and theca (arrows) cells are counterstained with DAPI (blue). Abbreviations: vitelline envelope; o, oocyte; yg, yolk globule. Control sections are shown in (supplementary fig. S21B, Supplementary Material online). Scale bars, 10 µm. (D) Double immunostaining of Aqp1ab1 (red; arrowhead) and -1ab2 (green; arrow) in hydrating oocytes. Scale bars and abbreviations as in A–C. (G) Fluorescence quantitative estimation of the amount of Aqp1ab1, -1ab2, YwhazLa/b, and -zLb in the PM of oocytes during hydration. (F, G) Co-immunoprecipitation of Aqp1ab1 (F ) or -1ab2 (G) and YwhazLa and -zLb in Vit, Hyd, and Mat follicles. The IP using seabream Aqp1ab1 and -1ab2 specific antibodies (in F and G, respectively) were immunoblotted with appropriate antibodies as indicated. Phosphorylated forms of the channels are indicated by arrows, whereas aquaporin and YwhazL monomers are indicated by arrowheads. I, input; IgG or IgY, immunoglobulin G or Y; Ab, immunoprecipitating antibody. In F-G and J-K, molecular mass markers (kDa) are on the left. (H ) Immunostaining of Aqp1aa, -1ab1, and -1ab2 in postvitellogenic ovarian follicles treated with DMSO (control) or FSK in vitro. Labels as in A–C. Scale bar, 10 µm. (I ) Uptake of 3H2O by follicle-enclosed oocytes treated with DMSO or 3-Isobutyl-1-methylxanthine (IBMX)/ FSK. (J, K ) Co-immunoprecipitation of Aqp1ab1 (J ) or -1ab2 (K ) and YwhazLa and -zLb in postvitellogenic follicles treated or not with IBMX/FSK. Labels as in F-G. Data in E and I are presented as box and whisker plots with the number of follicles (E) or pools of ten follicles (I ) analyzed indicated by dots, and were analyzed by ANOVA (P < 0.05; different letters indicate significant differences for each protein) or an unpaired Student’s t-test (**P < 0.01), respectively. 10 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE the teleost-specific YwhazLa paralogs (fig. 4D and a relatively relaxed, low-specificity YWHA-binding requirement supplementary figs. S15A–F and S17A, Supplementary that has been maintained in species that retain only one of the Material online). In the cyprinid zebrafish, the data con­ channels, albeit with a slight binding preference for Aqp1ab1 to firmed the presence of two PKA mechanisms regulating YwhazLa. However, a highly constrained binding requirement Aqp1ab1 trafficking, one dependent on the phosphoryl­ of Aqp1ab2 to YwhazLb evolved in euacanthomorph species ation status of Ser261 in the C-terminus of the channel that retain both of the Aqp1ab-type channels. In addition, for YwhazLa binding, and another independent of the pre-euteleost Aqp1ab1 channels evolved YWHA- YwhazLa interaction and controlled by Ser263 phosphoryl­ independent phosphorylation mechanisms for trafficking to ation (supplementary figs. S12A and S15G and H, the PM, while the euteleost channels evolved recycling me­ Supplementary Material online). In contrast, in the marine chanisms mediated by p38 MAPK-phosphorylation. euacanthomorph teleost seabream (SaAqp1ab1) (fig. 4E–G), as well as in salmonids (SsaAqp1ab1) (supplementary figs. In Vivo Aqp1ab1 and -1ab2 PM Trafficking and S17B, Supplementary Material online), apparently, only YwhazL Binding is Regulated to Facilitate Oocyte one mechanism based on PKA phosphorylation of Hydration During Meiosis Resumption C-terminal Ser253 and Ser258, respectively, regulates the To investigate the physiological significance of YwhazL YwhazLa interaction and channel trafficking. These data regulation of the TSA1C trafficking, we determined the indicate that although salmonids only retain the in vivo dynamics of the subcellular localization of en­ Aqp1ab1 paralog as do cyprinids, the salmonid dogenous Aqp1aa, -1ab1, and -1ab2, and their interactions Aqp1ab1s have lost the second YwhazLa-independent with YwhazLa and -zLb proteins, during teleost oocyte hydra­ PKA regulatory mechanism for channel trafficking. We fur­ tion. For this, we employed the seabream as a model organ­ ther observed that the p38 MAPK-mediated phosphoryl­ ism for marine euacanthomorph teleosts (supplementary fig. ation sites within the YWHA-binding domain in the S19, Supplementary Material online) and paralog-specific Aqp1ab1 C-termini of salmonid and euacanthomorph tel­ antisera (supplementary fig. S20A and B, Supplementary eosts (Ser252-Ser253 and Ser254 in salmon and seabream, re­ Material online). spectively) prevent the YwhazLa interaction resulting in Immunofluorescence microscopy showed that the expres­ recycling of the channel (fig. 4E–G and supplementary sion of Aqp1aa was restricted to the theca cells surrounding fig. S17B and C, Supplementary Material online). vitellogenic ovarian follicles, and this localization did not For Aqp1ab2, we found that in the elopomorph eel traf­ change during oocyte hydration (supplementary fig. S21A, ficking of the channel is preferentially regulated by Supplementary Material online). In contrast, both Aqp1ab1 PKA-mediated YwhazLa binding, through phosphorylation and -1ab2 were localized in the most cortical region of the of C-terminal Ser241 (supplementary fig. S16A, C, E, ooplasm, while Aqp1ab2 was also detected in the granulosa and G, Supplementary Material online), as observed for and theca cells (fig. 5A). During oocyte hydration, Aqp1ab1 DrAqp1ab1. However, eel AaAqp1ab2 also retains a almost disappeared from the oocyte cortical cytoplasm and YwhazLa-independent mechanism for channel trafficking, accumulated in the most distal part of the microvilli crossing which is mediated by dual PKC phosphorylation of Ser255 in the vitelline envelope, whereas Aqp1ab2 became localized in the C-terminus (supplementary fig. S16G, Supplementary a more basal region of the microvilli (fig. 5B and D). In hy­ Material online) and of Thr148 in loop D (supplementary drated mature oocytes, Aqp1ab1 and -1ab2 signals were no figs. S10J and K, Supplementary Material online). In contrast, longer observed in the microvilli or were greatly reduced, re­ in all the euacanthomorph marine teleosts investigated, spectively (fig. 5C). Immunoblotting and double immunolo­ Aqp1ab2 trafficking is specifically (SaAqp1ab2 and calization experiments using seabream YwhazLa/b- or -zLb HhAqp1ab2) or partially (SseAqp1ab2) dependent on specific antibodies showed the colocalization of Aqp1ab1 YwhazLb interaction (fig. 4H and supplementary fig. S16B, and YwhazLa/b signals in the oocyte, as well as of those of D, and F, and S17D, Supplementary Material online). Thus, Aqp1ab2 and YwhazLb (fig. 5A–C), with their abundance in seabream, halibut, and sole, the interaction depends on in the PM increasing during the hydration process (fig. 5E). PKA phosphorylation of Thr262, Ser241-Ser242, and Ser242, re­ Concomitantly, increased phosphorylation of Aqp1ab1 and spectively (fig. 4I and J and supplementary fig. S17E and F, -1ab2 was only observed in follicles enclosing hydrating oo­ Supplementary Material online). PM targeting of cytes (supplementary fig. S19C and D, Supplementary SseAqp1ab2 is also regulated by PKA phosphorylation of Material online). However, co-immunoprecipitation revealed Thr233 outside of the predicted YWHA interacting region, that Aqp1ab1 already interacts with YwhazLa and -zLb in vi­ but this mechanism still depends on YwhazLa or -zLb tellogenic follicles, with the interaction being maintained dur­ binding, since inhibition of this interaction prevents ing oocyte hydration and terminated in mature follicles (fig. PKA-mediated trafficking of the channel to the PM 5F), whereas Aqp1ab2 only binds to YwhazLb in follicles con­ (supplementary fig. S17F, Supplementary Material online). taining hydrating oocytes (fig. 5G). The Ywhaqb paralog, also highly expressed in halibut ovar­ To confirm that PKA activation triggers Aqp1ab1 and ian follicles, did not have any positive effect on seabream -1ab2 phosphorylation and the YwhazL interactions in oo­ Aqp1ab1 or -1ab2 trafficking (supplementary fig. S18, cytes, we incubated isolated seabream vitellogenic ovarian Supplementary Material online). Therefore, our observations follicles in vitro in the presence or absence of FSK. These ex­ suggest that Aqp1ab-type trafficking regulation evolved from periments showed that FSK triggered the translocation of 11 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE FIG. 6. Seabream Aqp1ab1 trafficking is regulated by dual phosphorylation of the cytoplasmic loop D and C-terminus of the channel. (A) Schematic representation of the SaAqp1ab1 monomer showing the C-terminus regulatory sites for YwhazLa or -zLb binding (S253) and p38 MAPK phosphorylation (S254), and the amino acid sequence of loop D with the two predicted PKA phosphorylation sites (S151 and S156). (B) Pf of X. laevis oocytes injected with water or expressing SaAqp1ab1-WT or mutant channels in loop D (S151A and S156A), and treated with DMSO (control) or FSK. All oocytes were co-injected with YwhazLa-HA. Data are presented as box and whisker plots with the number of oocytes in each group indicated below the plot. For each construct, data were analyzed by the unpaired Student’s t-test (**P < 0.01: ***P < 0.001; as indicated in brackets). Data from oocytes exposed to FSK were separately analyzed by one-way ANOVA (P < 0.05; different letters in red color indicate significant differences). (C and D) Immunoblot of Aqp1ab1 and YwhazLa in protein extracts of TM and PM from oocytes treated as in A (C ), and of Aqp1ab1 co-immunoprecipitates (D). (E) Double immunostaining of Aqp1ab1 (red) and Aqp1ab2 (green) in X. laevis follicle-enclosed oocytes expressing WT Aqp1ab1 (red) or Aqp1ab1-S156A (red), and Aqp1ab2 (green), YwhazLa and -zLb, and treated with DMSO or FSK. gc, granulosa cells; ve, vitelline envelope; o, oocyte. Scale bar, 5 µm. (F ) Pf of oocytes expressing increasing amounts of Aqp1ab1 cRNA (0.05–20 ng) and fixed amounts of YwhazLa and -zLb (15 ng), in the presence or absence of Aqp1ab2 (20 ng) expression, and treated with FSK. The insert shows the Pf of oocytes injected from 0.05 up to 1 ng Aqp1a1 cRNA. (G) Immunoblot of Aqp1ab1, -1ab2, YwhazLa, and -zLb in TM and PM protein extracts from oocytes treated as in F. In C, D, and G, molecular mass markers are on the left. Aqp1ab1 and -1ab2 to the distal and proximal regions of the YwhazL interaction to facilitate channel trafficking to oocyte microvilli, respectively (fig. 5H and supplementary fig. the PM. However, the binding of Aqp1ab1 to YwhazL pro­ S22, Supplementary Material online), as observed in the in teins apparently does not induce channel shuttling to the vivo hydrating oocytes, and increased water permeability distal region of the oocyte microvilli. Rather, this mechan­ with respect to DMSO-treated (control) follicles (fig. 5I). ism appears to be associated with an additional phosphor­ Co-immunoprecipitation confirmed that in control follicles ylation of Aqp1ab1, indicating that a second PKA-mediated Aqp1ab1 was already phosphorylated and bound to both phosphorylation event of Aqp1ab1 may have evolved in pe­ YwhazLa and -zLb (fig. 5J), without the channel trafficking lagophil euacanthomorph teleosts. To investigate this hy­ to the distal region of the oocyte microvilli. However, pothesis, we re-examined the role of the two predicted when such distal membrane trafficking is activated by FSK, PKA phosphorylation sites in loop D of seabream Aqp1ab1 becomes more phosphorylated, while maintaining Aqp1ab1 (Ser151 and Ser156) in channel trafficking (fig. the interaction with YwhazLa and -zLb (fig. 5J). 6A), either of which is conserved in >95% of marine eua­ Phosphorylation of Aqp1ab2 was also increased after FSK canthomorphs that spawn pelagic eggs such as Atlantic stimulation, but in this case, the YwhazLb interaction was halibut (supplementary fig. S10F, Supplementary Material only detected in the presence of FSK, when the channel is online). Experiments using X. laevis oocytes showed that translocated to the basal region of oocyte microvilli (fig. 5K). the increment of the Pf of FSK-treated oocytes co- expressing SaAqp1ab1-S151A or -S156A plus YwhazLa was reduced when compared to oocytes injected with wild-type Dual Phosphorylation Generates a Two-Step (WT) SaAqp1ab1 and YwhazLa, the reduction being more Regulation of Aqp1ab1 Trafficking to Avoid prominent in the S156A mutant-expressing oocytes (fig. Competitive PM Occupancy During Oocyte 6B). Co-immunoprecipitation and immunoblotting corro­ Hydration borated that none of the mutants prevented the YwhazLa The previous data suggested that Aqp1ab1 and -1ab2 interaction (fig. 6C and D), but the relative amount of the phosphorylation in seabream oocytes mediates the S156A mutant in the oocyte PM was lower than that of 12 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE to the most distal region of the oocyte microvilli traversing the vitelline envelope, whereas the S156A mutant remained retained in the microvillar proximal region (fig. 6E). In con­ trast, SaAqp1ab2 was accumulated in the proximal region of the microvilli in response to FSK, regardless of the co- expression with the WT SaAqp1ab1 or the S156A mutant (fig. 6E). These data therefore suggest that when Aqp1ab1 is bound to YwhazLa or -zLb, a second PKA-mediated phos­ phorylation of a loop D Ser is the mechanism by which the channel is shuttled to the distal region of the oocyte micro­ villi, thus avoiding competitive PM spatial occupancy with Aqp1ab2 during oocyte hydration. To finally investigate the physiological relevance of non­ competitive membrane space occupancy of Aqp1ab chan­ nels during oocyte hydration, X. laevis oocytes were injected with increasing amounts of SaAqp1ab1 cRNA and fixed amounts of YwhazLa and -zLb, in the presence or absence of a constant dose of SaAqp1ab2, followed by Pf determination and immunoblot analysis after FSK treat­ ment. At low doses of SaAqp1ab1 cRNA, the Pf of oocytes correlated well with the protein expression levels and the accumulation of the channel in the oocyte PM, whereas at high doses of SaAqp1ab1 the Pf and the expression levels reached a plateau (fig. 6F and G), possibly indicating the saturation of the oocyte translation machinery. However, co-expression of SaAqp1ab2 with low doses of SaAqp1ab1 resulted in an increased oocyte Pf, which was no longer observed when high doses of SaAqp1ab1 were injected in the oocytes (fig. 6F). These data indicate that expression of Aqp1ab1 and -1ab2 in different regions of the oocyte PM augments the bulk rate of aquaporin- mediated water flow. A Model for the Regulation of Aqp1ab1 and -1ab2 FIG. 7. Proposed model for the regulatory pathways controlling Intracellular Trafficking During Oocyte Hydration Aqp1ab1 and -1ab2 intracellular trafficking during oocyte hydration in Marine Euacanthomorph Teleosts in the marine euacanthomorph teleost gilthead seabream. During vi­ tellogenesis, Aqp1ab1 is phosphorylated at Ser253 by PKA (or by an­ By integrating the functional data obtained in this study other kinase activated by PKA) to bind YwhazLa and -zLb, and the for the seabream, a new model for the molecular regula­ complex is transported to the oocyte cortical region, where it is re­ tion of Aqp1ab1 and -1ab2 intracellular trafficking during tained just below the oocyte PM. At this stage, Aqp1ab2 is not oocyte hydration in marine euacanthomorph teleosts that bound to YwhazLb and remains in the cytoplasm. When meiosis re­ retain both paralogs can be proposed (fig. 7). During vitel­ sumption and yolk hydrolysis are activated in oocytes by hormonal signals, the Aqp1ab1 channel complexed with YwhazLa or -zLb is logenesis, Aqp1ab1 is phosphorylated at Ser253 by PKA to phosphorylated again by PKA at Ser156 in loop D, which further shut­ bind YwhazLa and -zLb, and the complex is transported to tles Aqp1ab1 to the most distal region of the oocyte microvilli. At the oocyte cortical region, where it is retained just below this time, Aqp1ab2 is also phosphorylated by PKA at Thr262, which the oocyte PM. At this stage, Aqp1ab2 is not bound to drives the YwhazLb interaction and the trafficking of the channel YwhazLb and remains in the cytoplasm. When oocyte to the proximal region of the oocyte microvilli. This model may ex­ plain how marine euacanthomorph teleosts that retain a complete meiosis resumption and yolk hydrolysis are activated by TSA1C evolved subfunctionalized mechanisms involving protein ki­ hormonal signals (Fabra et al. 2006), the Aqp1ab1 channel nases and specific YwhazL proteins for the trafficking regulation of complexed with YwhazLa or -zLb is further phosphorylated the downstream Aqp1ab-type paralogs, such that competitive PM by PKA at Ser156 in loop D, which further shuttles Aqp1ab1 spatial occupancy is avoided and the bulk water flow is augmented to the most distal region of the oocyte microvilli. Activation during oocyte hydration. of PKA also phosphorylates Thr262 in Aqp1ab2, which drives the YwhazLb interaction and the trafficking of the channel the WT (fig. 6C). Immunostaining of intact X. laevis ovarian to the proximal region of the oocyte microvilli. After hydra­ follicles expressing YwhazL proteins and SaAqp1ab2, to­ tion, Aqp1ab1 is probably phosphorylated by p38 MAPK at gether with WT SaAqp1ab1 or the S156A mutant, and ex­ Ser254, which prevents the YwhazL interaction needed for posed to FSK, revealed that WT SaAqp1ab1 was trafficked continued trafficking, while also likely decreasing Aqp1ab1 13 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE FIG. 8. Evolution of phosphorylation sites regulating the membrane trafficking of vertebrate aquaporin-1 channels. (A) Aligned amino acid se­ quences of loop D and the carboxyl termini (CT) of Aqp1aa, -1ab1, and -1ab2 paralogs from distantly related teleosts showing the specific re­ sidues involved in PKA, PKC, and p38 MAPK regulation of channel trafficking as experimentally determined in this study using X. laevis oocytes. The YwhazLa and -zLb binding regions in Aqp1ab1 and -1ab2, respectively, are underlined. (B) Cartoon aquaporin structures indicating the intra­ cellular domains for the consensus sequences for loop D and the CT which are scaled according to residue prevalence and illustrate the con­ servation or substitution of PKA, PKC, and p38 MAPK phosphorylation sites in different vertebrate groups. The number of taxa (N ) used to calculate each consensus sequence is given. half-life so that the channel is directed into the degradation Discussion pathway (Moeller et al. 2016). Simultaneously, the inter­ The origin of the extraordinary diversity of modern marine action of Aqp1ab2 with YwhazLb is also terminated by a euacanthomorph teleosts has yet to be fully explained. yet unknown mechanism that dissociates YwhazLb binding, Since the earliest teleosts are estimated to have arisen to­ or due to a decreased synthesis of YwhazLb in the oocyte, ward the end of the Paleozoic Era ∼300 Ma in the after­ such that this paralog also ceases its trafficking to the PM. math of an R3 WGD (Hurley et al. 2007; Amores et al. The model shows how marine euacanthomorph teleosts 2011; Near et al. 2012; Hughes et al. 2018), it has been sug­ that retain a complete TSA1C evolved subfunctionalized gested that the additional gene repertoires and their subse­ mechanisms involving protein kinases and specific YwhazL quent neo-/subfunctionalization could have provided the proteins for the regulation of the trafficking of the down­ genetic material for the radiative success (Hoegg et al. stream Aqp1ab-type paralogs such that competitive PM 2004; Postlethwait et al. 2004). However, the large chrono­ spatial occupancy is avoided and the bulk water flow is aug­ logical gap (∼245 million years) between the R3 event and mented during oocyte hydration. the explosive radiation of the Euacanthomorphacea in 14 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE marine environments ∼55 Ma, together with the rediploi­ Although some species of FW teleosts also retain the dization of the genomes (Lien et al. 2016; Van de Peer et al. TSA1C, with aqp1ab-type gene expression occurring in ex­ 2017) are considered problematic for single causal events traovarian tissues (Cerdà and Finn 2010; Engelund and such as R3 (Santini et al. 2009). Indeed, a similar uncoupling Madsen 2014; Cerdà et al. 2017), our present and previous between WGD events and adaptation potential has been data (Tingaud-Sequeira et al. 2008; Zapater et al. 2011) noted for plants (Schranz et al. 2012). In this work, we un­ show that aqp1ab1 and -1ab2 transcript enrichment with­ cover an ancient TSA1C together with novel molecular in the ovaries of marine pelagophil species are orders of regulatory mechanisms that co-evolved to provide the ma­ magnitude higher than in the somatic tissues and that jority of marine euacanthomorph embryos with the water both channels are localized in the same PM of hydrating of life when no osmoregulatory systems exist. The process oocytes. Such a colocalization of closely related water- of oocyte hydration, which occurs during the final meiotic selective channels is unexpected, since it may imply redun­ maturation stages of oogenesis, is so efficient that the dant membrane space occupancy. This observation, spawned eggs typically achieve water contents of >90% together with the absence of membrane trafficking when so that they float pelagically for dispersal in the oceanic cur­ Aqp1ab2 was heterologously expressed in amphibian oo­ rents. In contrast to previous studies (Fabra et al. 2005, cytes, prompted us to investigate the intracellular traffick­ 2006; Kagawa et al. 2009; Zapater et al 2011), we show ing regulation of the TSA1C channels. here that the mechanism of oocyte hydration involves in By comparing the subdomains of the teleost TSA1C fact two tandemly duplicated aqp1ab-type genes that are channels to the Aqp1 orthologs of nonteleost vertebrates, selectively retained in species that spawn pelagic eggs. it became evident that the regulatory sites in the Aqp1 and The phylogenetic and synteny analyses of piscine aqp1-type Aqp1aa orthologs have remained relatively fixed through genes revealed that two of the three main lineages of extant time, while those of the Aqp1ab1 and −1ab2 channels teleosts (Elopomorpha and Osteoglossomorpha) still en­ have selectively changed during teleost evolution (fig. 8). code the R3 ohnologs (aqp1aa and aqp1b), and that the For example, PKC positively regulates the membrane traf­ teleost-specific aqp1ab-type genes only duplicated down­ ficking of the Aqp1/Aqp1aa orthologs through phosphor­ stream of the aqp1aa paralog to form the TSA1C. Since ylation of a conserved Thr in loop D in most vertebrate we did not find traces of the aqp1ab-type genes in the gen­ groups except some amphibians, Polypteriformes, omes of the preteleost actinopterygian lineages, yet ob­ Siluriformes, and Euteleostei, which encode a Val or Ala. serve that the osteoglossomorph teleosts still retain The Thr/Val substitution in Euteleostei resulted in the pseudogene remnants of aqp1ab1 and -1ab2, the tandem loss of the positive PKC-mediated regulation and the duplication event likely occurred soon after R3 and was emergence of a negative PKC-mediated regulation. thus ancient and specific for teleosts. Further, in contrast to Chondrichthyes and Sarcopterygii, The consequences of gene duplication have been widely many actinopterygian fishes evolved a C-terminal Ser for recognized, with stochastic silencing and loss occurring as positive regulation via PKA-mediated phosphorylation. the most prevalent outcome within a few million years Nevertheless, even in euacanthomorph species in which (Lynch and Conery 2000). This type of outcome fits the the Aqp1aa C-terminal PKA site is conserved, as in evolution of the aqp1b ohnologs, which are seemingly ex­ Senegalese sole, an alternative negative PKC-mediated tinct in all clupeocephalan teleosts. Conversely, gene pres­ regulation still exists, while the positive PKA-mediated ervation is associated with the acquisition of novel (neo-) regulation of the older lineages is also maintained. or split (sub-) functions (Force et al. 1999). In the present In contrast to the Aqp1/Aqp1aa, the two Aqp1ab-type context, we find that all three processes have continued to channels evolved PKA, PKC, and p38 MAPK regulatory impact the trajectories of aqp1ab-type genes throughout sites that are both dependent and independent of the ∼300 million years of teleost evolution. The data show teleost-specific YwhazL binding proteins (fig. 8). The iden­ that many lineages have lost or only retain fragmented tification of the novel YwhazLa and -zLb paralogs in the pseudogenes of both paralogs, while others have lost either present study extend recent findings reporting that ostar­ the aqp1ab1 or -1ab2 gene, yet many species still retain iophysan teleosts encode duplicated ywhaz genes of the both of the tandem duplicates. An overwhelming majority mammalian YWHAZ ortholog (Zhang et al. 2021). In this (95%) of species that spawn pelagic eggs in marine envir­ context, however, we note that the ywhazLa gene synteny onments retain at least one of the duplicates, with a third is only conserved between Otophysi and osteoglossomorph of the species still harboring complete copies of both para­ teleosts, while the ywhazLb gene is only syntenic within the logs. Conversely, much higher levels of gene loss are seen in Ostariophysi (e.g., zebrafish and catfishes). The novel species that spawn benthic eggs (43%) or incubate their ywhazLa and -zLb genes identified here cluster within the eggs internally (91%). In marine benthic eggs, oocytes hy­ vertebrate ywhaz clade, but are polyphyletic and are not drate to a significantly lower degree than pelagic eggs, syntenic between euacanthomorph and the more basal reaching water contents of 68–82% of egg wet mass teleost lineages. Their neofunctionalization can be inferred (Finn et al., 2002a, 2002b), while the eggs of seahorses from the longer branch lengths, and the mutagenesis ex­ and pipefishes, which are incubated in the male brood periments show their specific functional requirement for pouch that remains hyposmotic to SW (Ripley 2009), do PM trafficking of the Aqp1ab-type channels. For example, not hydrate (Finn, unpublished data). in the basal cohort of elopomorph teleosts, such as the 15 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE European eel, which spawn pelagic eggs and only retain the competitive space occupancy in the oocyte PM and an in­ Aqp1ab2 paralog, PKA-mediated positive regulation of crease in the rate of bulk water influx during hydration. membrane trafficking is dependent on C-terminal The hydration process is terminated though the evolution YwhazLa binding and two YwhazLa-independent of p38 MAPK-mediated phosphorylation sites in the PKC-regulated sites in loop D and the C-terminus, respect­ YwhazLa binding region of Aqp1ab1, which dissociates ively (fig. 8). In modern lineages of euacanthomorph species the interaction with YwhazLa, and through a different that retain both the Aqp1ab2 and -1ab1 paralogs, the mechanism also resulting in the release of the YwhazLb PKC-mediated regulation of Aqp1ab2 is lost, and mem­ from Aqp1ab2. Both mechanisms consequently induce brane trafficking became dependent on PKA-mediated the recycling of the two Aqp1ab-type channels from the phosphorylation of the same C-terminal Ser residues as in PM, so that >90% of the egg water content remains locked the elopomorph teleosts, but dependent on YwhazLb ra­ within the yolk cell due to the reduction in the oocyte per­ ther than YwhazLa binding. meability. When fertilized, the water-endowed eggs with a A more intricate regulation of membrane trafficking recombined allelic potential have a lower specific gravity evolved in the teleost Aqp1ab1 channels (fig. 8). Amongst than the surrounding SW and they thus float as passive pe­ cyprinid species, such as zebrafish, PKA-mediated phos­ lagic passengers in the oceanic currents, which geodisperse phorylation of different Ser residues in the C-terminus them to new horizons. positively regulates membrane trafficking, on both YwhaLa-dependent and independent manner, while PKC-mediated phosphorylation in loop D recycles the chan­ Materials and Methods nel. In salmonids, such as the Atlantic salmon, only the posi­ tive PKA-mediated regulation in the C-terminus that is Animals and Reagents dependent on YwhazLa binding is maintained, while new Fish were obtained from different aquaculture stations in C-terminal recycling sites emerge that are regulated by Spain and Norway, whereas adult X. laevis were purchased PKC (outside of the YwhazLa binding domain) and p38 from the Centre de Ressources Biologiques Xénopes MAPK (within the YwhazLa binding site). The p38 MAPK (University of Rennes, France). Details relating to the negative regulatory site in the YwhazLa binding region is care and use of fish and frogs are described in maintained in euacanthomorph teleosts, such as seabream, (supplementary text S1, Supplementary Material online). together with a positive PKA-mediated YwhazLa-dependent All reagents were purchased from Merck unless indicated site. However, two additional positive PKA-mediated regula­ otherwise. The commercial and custom-made antibodies tion sites appear in loop D of Aqp1ab1 that depend on the employed are described in (supplementary table S3, binding of the channel to YwhazLa. It is this latter two-step Supplementary Material online). regulation that generates a membrane trafficking shunt of Aqp1ab1 to the distal region of the microvilli, and thus the avoidance of competitive occupancy of the same membrane Phylogenetic, Syntenic, and Structural Analyses space as the Aqp1ab2 channel. In contrast, Aqp1ab2 is de­ Sequences were assembled from open-source databases, pendent on YwhazLb binding, lacks the loop D regulatory aligned, and converted to codons before analysis via step, and thus remains in the proximal region of the PM. Bayesian inference, with in silico sequence analyses per­ Our ex vivo experimental data also show that the Aqp1ab1 formed as described in (supplementary text S1, membrane shunt can be recapitulated in the surrogate X. lae­ Supplementary Material online). A list of aquaporin type, vis oocytes co-expressing Aqp1ab1, -1ab2, and the YwhaL spawning habitat, and egg type according to fishbase binding proteins, resulting in an increase in the bulk rate of (https://www.fishbase.se/) is provided in (supplementary water transport. table S1, Supplementary Material online). The syntenic The combined findings thus shed new light on the mo­ analyses of the aquaporin genes were conducted via lecular adaptations facilitating the success of teleost eggs tblastn searches of whole genome sequencing databases. in the oceans. To survive in the harsh hyperosmotic saline Scaled sequence logos illustrating the relative conservation environment, teleosts evolved adaptive solutions that pro­ of residues in the aquaporin loop D and C-terminus- vide their hyposmotic marine embryos with the water of subdomains were generated with Geneious v9.1.8 life prior to the development of homeosmotic regulatory (Biomatters Ltd, New Zealand). systems. The process involved the selective retention of duplicated aqp1ab-type water channel genes that are tan­ demly arranged in an ancient ternary cluster, the TSA1C. Transcriptome Shotgun Sequencing Each channel co-evolved with specific YwhazL regulatory Illumina RNA-seq library construction and HiSeq2500 se­ proteins and PKA-, PKC-, and p38 MAPK-mediated phos­ quencing were carried out to sequence the transccriptome phorylation pathways to regulate the hydration of the oo­ of Atlantic halibut and X. laevis postvitellogenic ovarian cytes. The most intricate solutions evolved in the highly follicles. The complete methods for the novo assembly of diverse marine euacanthomorph teleosts that retain the transcriptomes and the identification of YWHA ortho­ both of the Aqp1ab1 and -1ab2 channels. In such species, logs are described in (supplementary text S1, an innovative membrane shunt evolved that avoids Supplementary Material online). 16 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE Expression Constructs and Site-Directed Mutagenesis washed in water for the subtraction of the background sig­ The teleost aquaporin and YwhazL cDNAs employed in nal from the externally bound solute. this study, and their GenBank accession numbers, are indi­ cated in (supplementary text S1, Supplementary Material Protein Extraction and Immunoblotting online). The cDNAs were subcloned into the pT7Ts The total membrane (TM) and PM fractions of X. laevis oo­ (Deen et al. 1994) or pcDNA3 (Invitrogen) expression vec­ cytes (n = 10) and seabream ovarian follicles (n = 50) were tors, and in most cases fused with an HA or flag epitope tag isolated as described previously (Kamsteeg and Deen at the N- or C-terminus of the encoded proteins by using 2001). For the seabream follicles, the final pellet was resus­ polymerase chain reaction (PCR). Mutations into the loop pended in 1 × Laemmli sample buffer supplemented with D or C-terminal amino acid sequence of aquaporins were 1 mM NaF, 1 mM Na3VO4, and protease inhibitors. In introduced by using the QuikChange lightning kit (Agilent some experiments, whole protein extracts from seabream Technologies; 210518) following the manufacturer’s in­ follicles were obtained by homogenization in 500 µl of structions. Substitution of the Aqp1ab1 or -1ab2 complete radioimmunoprecipitation assay buffer (150 mM NaCl, C-terminus for that of Aqp1aa was carried out by PCR as 50 mM Tris pH 8.0, 1.0% Triton X-100, 0.5% sodium deox­ previously described (Tingaud-Sequeira et al. 2008). All ycholate, 0.1% SDS, supplemented with 1 mM NaF, 1 mM constructs were verified by DNA Sanger sequencing to Na3VO4, protease inhibitors, and 80 U benzonase) using a confirm that only the desired chimeras or mutations glass douncer, and further incubation on ice for 15 min. were produced. The samples were then centrifuged at 14,000 × g at 4 °C for 1 min, and the supernatant was mixed with 2 × Expression in X. laevis Oocytes and Swelling Assays Laemmli sample buffer and heated at 95 °C for 10 min. The cRNAs corresponding to the different constructs were In some cases, lysates were digested with calf intestine al­ synthesized in vitro with T7 RNA polymerase from kaline phosphatase (New England Biolabs Inc.; M0525) for XbaI-digested pT7Ts vector containing the cDNAs. The 1 h at 37 °C before Laemmli sample buffer addition. isolation and microinjection of oocytes (0.5–15 ng of Immunoblotting was subsequently carried out as previously cRNA depending on the construct), as well as the swelling described (Zapater et al. 2011). assays, were carried out as previously described (Deen et al. 1994; Chauvigné et al. 2011). The effect of PKC or PKA ac­ Co-Immunoprecipitation tivation on the oocyte Pf was tested by preincubating the X. laevis oocytes and seabream follicles were homogenized oocytes with 100 nM PMA for 30 min, or with 100 μM IBMX in the lysis buffer from the Pierce Crosslink Magnetic im­ for 1 h and subsequently with 100 μM FSK for 30 min, before munoprecipitated products (IP)/Co-IP Kit (Pierce; 88805) determination of oocyte swelling or protein extraction. The with a bouncer on ice, and centrifuged at 14,000 × g for effect of p38 MAPK activation was tested using 100 μM ani­ 1 min. An aliquot (10%) of the protein extract was col­ somycin (ANS). Control oocytes were treated with 0.1% of lected as the input of the IP and mixed with 2 × the drug vehicle dimethyl sulfoxide (DMSO). Laemmli sample buffer, whereas the remaining extract was incubated overnight at 4 °C under constant agitation In Vitro Incubation of Seabream Ovarian Follicles and with magnetic beads previously coated with anti-HA, Water Uptake Assays anti-SaAqp1ab1 or anti-SaAqp1ab2 antibodies. For rabbit Ovarian follicles at a different developmental stage were primary antibodies, the IP was carried out using the kit in­ collected from naturally spawning seabream females and dicated above, and final eluted proteins in 50 µl of elution the Pf of follicles at each stage was determined under buffer were mixed with 4 × Laemmli sample buffer sup­ hypertonic conditions as previously described (Fabra plemented with protease and phosphatase inhibitors. et al. 2006; Zapater et al. 2011). For in vitro incubations, For chicken antibodies, the IP was performed using groups of ten postvitellogenic follicles were placed in Dynabeads M-280 Tosylactivated (Invitrogen; 14203) fol­ 75% Leivovitz L-15 culture medium with L-glutamine and lowing the manufacturer’s instructions. The input and treated with 100 µM IBMX and 100 μM FSK for 1 h at immunoprecipitated samples were immunoblotted as in­ 18 °C, or with 0.1% DMSO (controls). After incubation, fol­ dicated above. licles were fixed for immunofluorescence microscopy or used for water uptake assays. For this, follicles were trans­ Immunofluorescence Microscopy and Quantification ferred to 200 µl ultrapure water containing 10 µCi [3H] Samples were processed and immunostained as previously water (1 mCi/ml) (American Radiolabeled Chemicals described (Chauvigné et al. 2018). Control sections were Inc., ART 0194A) for 20 min at room temperature. incubated without primary antibodies (when using com­ Follicles were washed twice in distilled water, dissolved mercial antibodies) or with antibodies preincubated with in 400 μl of 10% sodium dodecyl sulfate (SDS) for 1 h, the corresponding peptides used for immunization (for mixed with 4 ml of liquid scintillation cocktail (Perkin custom-made antibodies). Double immunostaining using Elmer LLC Ultima Gold; LLC6013326), and counted using rabbit antibodies against seabream Aqp1ab1 and either a Beckman Coulter scintillation counter. Some follicles YwhaLa/b or -zLb was performed by staining first for were exposed to [3H] water for 1 min and immediately YwhaLa/b or -zLb as described above using an antirabbit 17 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE Cy3-coupled secondary antibody. After fixing with 4% par­ Author Contributions aformaldehyde (PFA) for 1 h, slides were incubated with J.C. and R.N.F. designed the study. B.N. and L.B. provided the Aqp1ab1 antibody previously labeled with Zenon biological samples and genomic sequences. R.N.F. and J.C. Rabbit IgG Alexa Fluor 488 Labeling Kit (Invitrogen, carried out the phylogenetic and in silico analyses. A.F., Z25302) following manufacturer instructions. After 1 h, F.C., and M.C. performed wet laboratory experiments sections were washed and fixed again with PFA for and collected data. A.V., R.G., and R.N.F. carried out 15 min, incubated with 4′,6-diamidino-2-phenylindole RNA-seq analysis. A.F., F.C., R.N.F., and J.C. interpreted (DAPI; 1:10,000), and mounted with Fluoromount aqueous the results. J.C. acquired the funding and supervised the mounting medium (F4680). Immunofluorescence was ob­ work. J.C. and R.N.F. wrote the manuscript. All authors ap­ served and documented with a Zeiss Axio Imager Z1/ proved the final version of the manuscript. Apotome fluorescent microscope (Carl Zeiss Corp.). Conflict of Interest: The authors declare that they have Changes in Aqp1ab1, -1ab2, YwhazLa/b, and -zLb abun­ no competing interests. dance in the PM of seabream oocytes were evaluated by measuring the fluorescence intensity of the microvilli crossing the vitelline envelope with respect to the total Data Availability fluorescence (PM plus cytoplasm) using the ImageJ soft­ ware v.1.52o (https://imagej.nih.gov/ij/). For all treatments Data underlying this article are available in the article and or developmental stages, the images were acquired with in its online supplementary material. the same fluorescence exposure and similar orientation, so that the area of the PM and cytoplasm was equivalent. References Agre P, Sasaki S, Chrispeels MJ. 1993. Aquaporins: a family of water Statistics channel proteins. Am J Physiol. 265:F461. Comparisons between two independent groups were Ahlstrom EH, Moser HG. 1981. Systematics and development of made by the two-tailed unpaired Student’s t-test. The stat­ early life history stages of marine fishes: achievements during the past century, present status and suggestions for the future. istical significance among multiple groups were analyzed Rapp P-v Réun Cons int Explor Mer. 178:541–546. by one-way ANOVA, followed by Tukey’s multiple com­ Albalat R, Cañestro C. 2016. Evolution by gene loss. Nat Rev Genet. parison test, or by the nonparametric Kruskal–Wallis test 17:379–391. and further Dunn’s test for nonparametric post hoc com­ Amores A, Catchen J, Ferrara A, Fontenot Q, Postlethwait JH. 2011. parisons. Percentages were square root transformed previ­ Genome evolution and meiotic maps by massively parallel DNA ous analyses. Differences between proportions were tested sequencing: spotted gar, an outgroup for the teleost genome du­ plication. Genetics 188:799–808. via the critical z-value approach. Statistical analyses were Berthelot C, Brunet F, Chalopin M, Juanchich A, Bernard M, Noel B, carried out using GraphPad Prism v9.1.2 (226) (GraphPad Bento P, Da Silva C, Labadie K, Alberti A, et al. 2014. The rainbow Software). In all cases, statistical significance was defined trout genome provides novel insights into evolution after whole- as P < 0.05 (*), P < 0.01 (**), or P < 0.001 (***). genome duplication in vertebrates. Nat Commun. 5:3657. Cerdà J, Chauvigné F, Finn RN. 2017. The physiological role and regu­ lation of aquaporins in teleost germ cells. In: Yang B, editor. Advances in experimental medicine and biology. Vol. 969. Supplementary Material Aquaporins. Dordrecht: Springer. p. 149–171. Supplementary data are available at Molecular Biology and Cerdà J, Fabra M, Raldúa D. 2007. Physiological and molecular basis Evolution online. of fish oocyte hydration. In: Babin P, Cerdà J, Lubzens E, editors. The fish oocyte: from basic studies to biotechnological applications. Dordrecht: Springer. p. 349–396. Cerdà J, Finn RN. 2010. Piscine aquaporins: an overview of recent ad­ Acknowledgments vances. J Exp Zool. 313A:623–650. We thank Michael Dondrup (University of Bergen, Chauvigné F, Ferré A, Cerdà J. 2021. The Xenopus oocyte as an expres­ sion system for functional analyses of fish aquaporins. Methods Norway) and Per Gunnar Fjelldal (Institute of Marine Mol Biol. 2218:11–28. Research, Norway) for assistance in bioinformatic analysis Chauvigné F, Lubzens E, Cerdà J. 2011. Design and characterization of and collection of biological samples from Atlantic salmon, genetically engineered zebrafish aquaporin-3 mutants highly respectively. This work was supported by the Spanish permeable to the cryoprotectant ethylene glycol. BMC Ministry of Science and Innovation (MCIN/AEI/10.13039/ Biotechnol. 11:34. Chauvigné F, Parhi J, Ducat C, Ollé J, Finn RN, Cerdà J. 2018. The cel­ 501100011033), the European Regional Development lular localization and redistribution of multiple aquaporin para­ Fund (ERDF) “A way of making Europe” (European logs in the spermatic duct epithelium of a maturing marine Union), Grant no. AGL2016-76802-R (to J.C.), and the teleost. J Anat. 233:177–192. Norwegian Research Council (RCN) Grant no. 294768/ Chauvigné F, Yilmaz O, Ferré A, Fjelldal PG, Finn RN, Cerdà J. 2019. E40 (to R.N.F). F.C. and A.F. were supported, respectively, The vertebrate Aqp14 water channel is a neuropeptide- by the “Ramon y Cajal” programe (RYC-2015-17103) and regulated polytransporter. Comm Biol. 2:462. Crow KD, Smith CD, Cheng JF, Wagner GP, Amemiya CT. 2012. An in­ a predoctoral (BES-2014-068745) contract from Spanish dependent genome duplication inferred from Hox paralogs in the MCIN. R.N.F. was also supported by the University of American paddlefish - a representative basal ray-finned fish and Bergen (Norway). important comparative reference. Genome Biol Evol. 4:937–953. 18 Functional Evolution of a Teleost Aquaporin Cluster · https://doi.org/10.1093/molbev/msad071 MBE Deen PM, Verdijk MA, Knoers NV, Wieringa B, Monnens LA, van Os based on transcriptomic and genomic data. Proc Natl Acad Sci CH, van Oost BA. 1994. Requirement of human renal water chan­ U S A. 115:6249–6254. nel aquaporin-2 for vasopressin-dependent concentration of ur­ Hurley IA, Mueller RL, Dunn KA, Schmidt EJ, Friedman M, Ho RK, ine. Science 264:92–95. Prince VE, Yang Z, Thomas MG, Coates MI. 2007. A new time- Du K, Stöck M, Kneitz S, Klopp C, Woltering JM, Adolfi MC, Feron R, scale for ray-finned fish evolution. Proc R Soc B. 274:489–498. Prokopov D, Makunin A, Kichigin I, et al. 2020. The sterlet stur­ Kagawa H, Horiuchi Y, Kasuga Y, Kishi T. 2009. Oocyte hydration in geon genome sequence and the mechanisms of segmental redi­ the Japanese eel (Anguilla japonica) during meiosis resumption ploidization. Nat Ecol Evol. 4:841–852. and ovulation. J Exp Zool. 311A:752–762. Engelund MB, Madsen SS. 2014. Tubular localization and expression­ Kagawa H, Kishi T, Gen K, Kazeto Y, Tosaka R, Matsubara H, al dynamics of aquaporins in the kidney of seawater-challenged Matsubara T, Sawaguchi S. 2011. Expression and localization of Atlantic salmon. J Comp Physiol B. 185:207–223. aquaporin 1b during oocyte development in the Japanese eel Fabra M, Raldúa D, Bozzo MG, Deen PM, Lubzens E, Cerdà J. 2006. (Anguilla japonica). Reprod Biol Endocrinol. 9:71. Yolk proteolysis and aquaporin-1o play essential roles to regulate Kamsteeg EJ, Deen PM. 2001. Detection of aquaporin-2 in the plasma fish oocyte hydration during meiosis resumption. Dev Biol. 295: membranes of oocytes: a novel isolation method with improved 250–262. yield and purity. Biochem Biophys Res Commun. 282:683–690. Fabra M, Raldúa D, Power DM, Deen PM, Cerdà J. 2005. Marine fish Kendall AW, Ahlstrom EH, Moser HG. 1984. Early life history stages egg hydration is aquaporin-mediated. Science 307:545. of fishes and their characters. In: Moser HG, editor. Ontogeny and Finn RN. 2007. The maturational disassembly and differential prote­ systematics of fishes. Lawrence, KS: American Society of olysis of paralogous vitellogenins in a marine pelagophil teleost: a Ichthyologists and Herpetologists Special publ 1. p. 11–22. conserved mechanism of oocyte hydration. Biol Reprod. 76: Lien S, Koop BF, Sandve SR, Miller JR, Kent MP, Nome T, Hvidsten TR, 936–948. Leong JS, Minkley DR, Zimin A, et al. 2016. The Atlantic salmon Finn RN, Cerdà J. 2011. Aquaporin evolution in fishes. Front Physiol. genome provides insights into rediploidization. Nature 533: 2:44. 200–205. Finn RN, Chauvigné F, Hlidberg JB, Cutler CP, Cerdà J. 2014. The Lynch M, Conery JS. 2000. The evolutionary fate and consequences lineage-specific evolution of aquaporin gene clusters facilitated of duplicate genes. Science 290:1151–1155. tetrapod terrestrial adaptation. PLoS One 9:e113686. MacQueen DJ, Johnston I. 2014. A well-constrained estimate for the Finn RN, Fyhn HJ. 2010. Requirement for amino acids in ontogeny of timing of the salmonid whole genome duplication reveals major fish. Aquacult Res. 41:684–716.. decoupling from species diversification. Proc R Soc B Biol Sci. 281: Finn RN, Fyhn HJ, Norberg B, Munholland J, Reith M. 2000. Oocyte 20132881. hydration as a key feature in the adaptive evolution of teleost Maissey JG. 1996. Discovering fossil fishes. New York: Westview Press. fishes to seawater. In: Norberg B, Kjesbu OS, Taranger GL, Moeller HB, Slengerik-Hansen J, Aroankins T, Assentoft M, MacAulay Anderson E, Stefansson SO, editors. Proceedings of the 6th N, Moestrup SK, Bhalla V, Fenton RA. 2016. Regulation of the International Symposium on the Reproductive Physiology of Fish. water channel aquaporin-2 via 14-3-3θ and -ζ. J Biol Chem. Bergen: Institute of Marine Research and the University of 291:2469–2484. Bergen. p. 289–291. Near TJ, Eytan RI, Domburg A, Kristen KL, Moore JA, Davis MP, Finn RN, Kristoffersen BA. 2007. Vertebrate vitellogenin gene du­ Wainwright PC, Friedman M, Smith WL. 2012. Resolution of ray- plication in relation to the “3R hypothesis”: correlation to the finned fish phylogeny and timing of diversification. Proc Natl pelagic egg and the oceanic radiation of teleosts. PLoS One 2: Acad Sci U S A. 109:13698–13703. e169. Nesverova V, Törnroth-Horsefield S. 2019. Phosphorylation- Finn RN, Østby G, Norberg B, Fyhn HJ. 2002a. In vivo oocyte hy­ dependent regulation of mammalian aquaporins. Cells 8:82. dration in Atlantic halibut (Hippoglossus hippoglossus); pro­ Postlethwait J, Amores A, Cresko W, Singer A, Yan YL. 2004. teolytic liberation of free amino acids, and ion transport are Subfunction partitioning, the teleost radiation and the annota­ driving forces for osmotic water influx. J Exp Biol. 205: tion of the human genome. Trends Genet. 20:481–490. 211–224. Prado K, Cotelle V, Li G, Bellati J, Tang N, Tournaire-Roux C, Finn RN, Wamboldt M, Fyhn HJ. 2002b. Differential processing of Martinière A, Santoni V, Maurel C. 2019. Oscillating aquaporin yolk proteins during oocyte hydration in fishes (Labridae) that phosphorylation and 14-3-3 proteins mediate the circadian regu­ spawn benthic and pelagic eggs. Mar Ecol Progr Series. 237: lation of leaf hydraulics. Plant Cell 31:417–429. 217–226. Rajan S, Preisig-Müller R, Wischmeyer E, Nehring R, Hanley PJ, Force A, Lynch M, Pickett FB, Amores A, Yan YL, Postlethwait J. 1999. Renigunta V, Musset B, Schlichthörl G, Derst C, Karschin A, Preservation of duplicate genes by complementary, degenerative et al. 2002. Interaction with 14-3-3 proteins promotes functional mutations. Genetics 151:1531–1545. expression of the potassium channels TASK-1 and TASK-3. J Friedman M. 2010. Explosive morphological diversification of spiny- Physiol. 545:13–26. finned teleost fishes in the aftermath of the end-cretaceous ex­ Ripley JL. 2009. Osmoregulatory role of the paternal brood pouch for tinction. Proc R Soc B 277:1675–1683. two Sygnathus species. Comp Biochem Physiol. 154A:98–104. Fu H, Subramanian RR, Masters SC. 2000. 14-3-3 Proteins: structure, Santini F, Harmon LJ, Carnevale G, Alfaro ME. 2009. Did genome du­ function, and regulation. Annu Rev Pharmacol Toxicol. 40: plication drive the origin of teleosts? A comparative study of di­ 617–647. versification in ray-finned fishes. BMC Evol Biol. 9:194. Fyhn HJ, Finn RN, Reith M, Norberg B. 1999. Yolk protein hydrolysis Schranz ME, Mohammadin S, Edger PP. 2012. Ancient whole genome and oocyte free amino acids as key features in the adaptive evo­ duplications, novelty and diversification: the WGD radiation lag- lution of teleost fishes to seawater. Sarsia 84:451–456. time model. Curr Opin Plant Biol. 15:147–153. Gasanov EV, Jedrychowska J, Kuznicki J, Korzh V. 2021. Evolutionary Shikano S, Coblitz B, Sun H, Li M. 2005. Genetic isolation of transport context can clarify gene names: teleosts as a case study. BioEssays signals directing cell surface expression. Nat Cell Biol. 7:985–992. 43:e2000258. Sullivan CV, Yilmaz O. 2018. Vitellogenesis and yolk proteins, fish. Hoegg S, Brinkmann H, Taylor JS, Meyer A. 2004. Phylogenetic timing Encyclopedia of reproduction. 2nd ed. Elsevier Inc. p. 1–11. of the fish-specific genome duplication correlates with the diver­ Tingaud-Sequeira A, Calusinska M, Finn RN, Chauvigné F, Lozano J, sification of teleost fish. J Mol Evol. 59:109–203. Cerdà J. 2010. The zebrafish genome encodes the largest verte­ Hughes LC, Orti G, Huang Y, Sun Y, Baldwin CC, Thompson AW, brate repertoire of functional aquaporins with dual paralogy Arcila D, Betancur-R R, Li C, Becker L, et al. 2018. and substrate specificities similar to mammals. BMC Evol Biol. Comprehensive phylogeny of ray-finned fishes (actinopterygii) 10:38. 19 Ferré et al. · https://doi.org/10.1093/molbev/msad071 MBE Tingaud-Sequeira A, Chauvigné F, Fabra M, Lozano J, Raldúa D, Cerdà Zapater C, Chauvigné F, Norberg B, Finn RN, Cerdà J. 2011. Dual neo­ J. 2008. Structural and functional divergence of two fish functionalization of a rapidly evolving aquaporin-1 paralog re­ aquaporin-1 water channels following teleost-specific gene du­ sulted in constrained and relaxed traits controlling channel plication. BMC Evol Biol. 8:259. function during meiosis resumption in teleosts. Mol Biol Evol. Van de Peer Y, Mizrachi E, Marchal K. 2017. The evolutionary signifi­ 28:3151–3169. cance of polyploidy. Nat Rev Genet. 18:411–424. Zapater C, Chauvigné F, Tingaud-Sequeira A, Finn RN, Cerdà J. 2013. Xu P, Xu J, Liu G-J, Chen L, Zhou Z, Peng W, Jiang Y, Zhao Z, Jia Z, Sun Primary oocyte transcriptional activation of aqp1ab by the nu­ Y, et al. 2019. The allotetraploid origin and asymmetrical genome clear progestin receptor determines the pelagic egg phenotype evolution of the common carp Cyprinus carpio. Nat Commun. of marine teleosts. Dev Biol. 377:345–362. 10:4625. Zhang X, Yu P, Wen H, Qi X, Tian Y, Zhang K, Fu Q, Li Y, Li C. 2021. Yilmaz O, Ferré A, Nilsen F, Fjelldal PG, Cerdà J, Finn RN. 2020. Genome-wide characterization of aquaporins (aqps) in Unravelling the complex duplication history of deuterostome Lateolabrax maculatus: evolution and expression patterns during glycerol transporters. Cells 9:1663. freshwater acclimation. Mar Biotechnol. 23:696–709. 20