Lack of membrane sex steroid receptors for mediating rapid endocrine responses in molluscan nervous systems

Despite the lack of endogenous synthesis and relevant nuclear receptors, several papers have been published over the decades claiming that the physiology of mollusks is affected by natural and synthetic sex steroids. With scant evidence for the existence of functional steroid nuclear receptors in mollusks, some scientists have speculated that the effects of steroids might be mediated via membrane receptors (i.e. via non-genomic/non-classical actions) - a mechanism that has been well-characterized in vertebrates. However, no study has yet investigated the ligand-binding ability of such receptor candidates in mollusks. The aim of the present study was to further trace the evolution of the endocrine system by investigating the presence of functional membrane sex steroid receptors in a mollusk, the great pond snail (Lymnaea stagnalis). We detected sequences homologous to the known vertebrate membrane sex steroid receptors in the Lymnaea transcriptome and genome data: G protein-coupled estrogen receptor-1 (GPER1); membrane progestin receptors (mPRs); G protein-coupled receptor family C group 6 member A (GPRC6A); and Zrt- and Irt-like protein 9 (ZIP9). Sequence analyses, including conserved domain analysis, phylogenetics, and transmembrane domain prediction, indicated that the mPR and ZIP9 candidates appeared to be homologs, while the GPER1 and GPRC6A candidates seemed to be non-orthologous receptors. All candidates transiently transfected into HEK293MSR cells were found to be localized at the plasma membrane, confirming that they function as membrane receptors. However, the signaling assays revealed that none of the candidates interacted with the main vertebrate steroid ligands. Our findings strongly suggest that functional membrane sex steroid receptors which would be homologous to the vertebrate ones are not present in Lymnaea. Although further experiments are required on other molluscan model species as well, we propose that both classical and non-classical sex steroid signaling for endocrine responses are specific to chordates, confirming that molluscan and vertebrate endocrine systems are fundamentally different.


Introduction
Despite the lack of endogenous synthesis and relevant nuclear receptors (1-9), several papers have been published over the past 70 years showing that the physiology of mollusks is affected by natural and synthetic sex steroids occurring in the environment, though the robustness of the bioassay data in most of these papers has been questioned (10).Recent reviews tried to fill in the gaps in steroid signaling of mollusks and raised the idea that vertebrate sex steroids affect molluscan species via membrane receptors (i.e.non-genomic/ non-classical actions) (4,11).
Most efforts to prove this possible mode of action were performed by the investigation of hemocytes of Mytilus spp.
[reviewed by (4)].These works demonstrated that estradiol (E 2 ) rapidly induced lysosomal membrane instability, production of intracellular reactive oxygen species, lysozyme release and nitrogen-oxide production in Mytilus hemocytes; the underlying mechanisms, similar to those identified in mammalian cells, involved cytosolic Ca 2+ increase, activation of MAPKs and PKC, as well as phosphorylation of signal transducers and transcription factors (12-15).However, it is unknown whether E 2 binds to a specific membrane receptor in mollusks, if it cross-reacts with a receptor for other compound(s), or if there is a cognate sex steroid receptor at all.Furthermore, to our knowledge, no study has investigated the potential rapid effects of progesterone (P) and testosterone (T) on mollusks or conducted homology searches for their known membrane receptors in mollusks.Although recent studies have revealed sequences homologous to membrane progestin receptor beta (mPRb) and gamma (mPRg), two of the five vertebrate mPRs (16), in molluscan species such as the cuttlefish Sepia japonica (17), owl limpet Lottia gigantea, and Pacific oyster Crassostrea gigas (18), the ability of these receptor candidates to bind P has not been examined.
The aim of the present paper was to advance our understanding of the evolution of the endocrine system and to contribute to the ongoing debate about the role of vertebrate-type sex steroids in mollusks, by investigating the great pond snail (Lymnaea stagnalis).Lymnaea has long-served as a multipurpose molluscan model in neuroscience and neuroendocrinology [reviewed by (19)(20)(21)(22)(23)(24)], as well as in ecotoxicological studies related to endocrine disruption (25-28).To accomplish our aim, we searched for sequences in the Lymnaea transcriptome and genome data that show homology with the known vertebrate membrane sex steroid receptors that are not related to nuclear receptors: G protein-coupled estrogen receptor-1 (GPER1), five mPRs, G protein-coupled receptor family C group 6 member A (GPRC6A), and Zrt-and Irt-like protein 9 (ZIP9).We cloned the coding sequence of selected candidates and transiently expressed them in Human Embryonic Kidney (HEK) cells to assess their membrane expression and ability to bind vertebrate sex steroids.Our findings represent the first evidence indicating the absence of specific, evolutionarily conserved membrane receptormediated sex steroid signaling in a mollusk.

Experimental animals
5-month-old adult, mature specimens of Lymnaea were obtained from our laboratory-bred stocks.Snails were maintained in large plastic tanks containing 10 L oxygenated artificial snail water (composition in mM: 0.1309 NaHCO 3 , 0.0378 K 2 SO 4 , 0.4013 CaCl 2 .2H 2 O, 0.0390 Mg(NO 3 ) 2 .6H 2 O; pH=7.6) at 20°C (± 1°C) on a 12:12 h light:dark regime with natural wavelength light.Specimens were fed on lettuce ad libitum three times a week.

In silico searches and bioinformatic analysis
The neural transcriptome data obtained in our previous study (29) were used to search for sequences homologous to vertebrate membrane sex steroid receptors.For homology-searching, the relevant molluscan and vertebrate sequences were used as search queries (Supplementary Table 1).We also compared the obtained sequences with the public Lymnaea CNS transcriptome shotgun assembly (sequence read archive: #DRR002012 (30);) and genome data (31).Conserved domain search using NCBI CDD/SPARCLE (32,33) was first performed to check if the key regions were present in the deduced protein sequence.Phylogenetic analysis was made in the Molecular Evolutionary Genetics Analysis v7 software (34).Transmembrane domain prediction was performed with the TMHMM 2.0 and DeepTMHMM tools (35).

Plasmid constructions and transfection
Plasmid construction and transfection was according to the description of our previous study (36).The complete coding sequence of GPER1, mPR, and ZIP9 candidates was amplified with PrimeStar HS DNA Polymerase (#R010A, TaKaRa Bio, Japan).The applied gene-specific primers (designed with the SnapGene Viewer software (GSL Biotech)) with extensions to vector ends were as follows: 1) GPER1 candidate: AAT GCG GCC GCA TGG AGA CCA CAG TCG GGG (forward) and GCC TCT AGA CAG ATG GAC TGG TGG TTT CAG AAA GCC AT (reverse); 2) mPR candidate: AAT GCG GCC GCA TGT TGT TAC TGC CAG CAA CAT TGA G (forward) and GCC TCT AGA ATG GAT GAC CCC ATT TCT GTG ATG AA (reverse); 3) ZIP9 candidate: AAT GCG GCC GCA TGG ATG ATA TCT TGA CTC TCC TTT CTC TGT (forward) and GCC TCT AGA GTG CTT ATG ACC CAC AGC TAA AAA AAC A (reverse) (Integrated DNA Technologies, Belgium).
The amplicons were checked by gel electrophoresis, restricted with NotI -XbaI enzymes, purified by gel electrophoresis, and subcloned into the linearized pcDNA6-His-G16 plasmid [described in (37)] using T4 DNA ligase (#LGK-201, Toyobo, Japan).Since the coding sequence of GPRC6A candidate was too long to effectively amplify it, it was synthetized and ordered in pcDNA6-His-G16 plasmid as a ready-to-use vector (Eurofins Genomics).Codon usage optimized mPR and ZIP9 candidates in pcDNA6-His-G16 plasmids were synthesized by FASMAC (Kanagawa, Japan).After transformation and amplification in JM109 competent cells, colony PCR was carried out.Next, the plasmids were purified from the appropriate colonies with the QIAGEN Plasmid Midi Kit (#12143, Qiagen) and sequence verification by nucleotide sequencing was performed.
Finally, the constructed vectors were transiently transfected into genetically engineered HEK293MSR cells using the Lipofectamine 3000 Transfection reagent (#L3000001, Thermo Fisher Scientific) following the manufacturer's instructions.Transfection efficiency and expression for all candidates were evaluated by an immunodetection method as described below.
The real-time fluorescence assessment of Ca 2+ mobilization was performed using the FLIPR Calcium 5 Kit (#3808451, Molecular Devices, USA) following the manufacturer's instructions.All measurements were performed with a FlexStation III Multi Mode microplate reader (Molecular Devices).The data were obtained at least from two independent transfections for Ciona GALPR, all Lymnaea receptor candidates, and the empty vector.Ciona GALPR and empty vector samples were applied to the plate in two technical replicates, while all Lymnaea receptor candidate samples were applied in three technical replicates.

Statistical analysis
The data obtained during the second messenger assays were plotted using four-parameters logistics-based non-linear regression curve fit feature in the GraphPad Prism5 software (GraphPad, San Diego, USA).Significant differences at a given data point was analyzed with the OriginPro 2018 software (OriginLab Corp., USA).The normality of the datasets was investigated using the Shapiro-Wilk test and the homogeneity of variances between groups was checked with the Levene-test.The data were then analyzed using the two-sample t-test.

Results
Our homology-searching revealed candidate sequences for all vertebrate membrane sex steroid receptors which are not related to the nuclear ones (Supplementary Figure 1).After the initial bioinformatic analysis, including conserved domain analysis (Supplementary Figure 2), phylogenetics (Supplementary Figures 3-5), and transmembrane domain prediction (Supplementary Figure 6), four sequences were investigated further: one GPER1 candidate, one mPR candidate, one GPRC6A candidate, and one ZIP9 candidate.Based on the in silico analyses, the mPR and ZIP9 candidates appeared to be genuine homologs.Although the GPER1 and GPRC6A candidates were unlikely to be the counterparts of the vertebrate receptors, we did not exclude the possibility that, during evolution, these deduced proteins might have acquired the ability of recognizing steroids as well [as reported in the case of vertebrate ZIP9 (40)].
First, we investigated the expression and subcellular localization of the candidates by transiently transfecting the G16-fused coding sequences into HEK293MSR cells.In the positive control experiments (Supplementary Figure 7), the expression level of GFP indicated an appropriate transfection efficiency (Supplementary Figure 7A).Moreover, the G16-fused Ciona GALPR showed a high expression and localization at the plasma membrane (Supplementary Figure 7B).Immunocytochemical fluorescence confocal microscopic detection confirmed that the Lymnaea membrane sex steroid receptor candidates were localized to the cell surface of HEK293MSR cells (Figure 1).The expression of the GPER1 and GPRC6A candidates was high.
However, the expression of the mPR1 and ZIP9 candidates was very low and only a few cells showing immunosignal were detected.To evaluate the ligand-receptor interaction between the three main vertebrate sex steroids and the Lymnaea membrane sex steroid receptor candidates, we assessed the mobilization of intracellular Ca 2+ using G16-fused receptor candidates (41).In the positive control experiment (Supplementary Figure 7C), the Ciona GALP specifically induced the elevation of intracellular Ca 2+ in a concentration-dependent manner, as previously reported (37).In the case of the Lymnaea candidates, however, the assay showed that the respective ligand-receptor combinations failed to induce the elevation of intracellular Ca 2+ (Figure 2), indicating that these transmembrane proteins are not bona fide membrane sex steroid receptors.Both in the case of GPRC6A and ZIP9 candidates, there was an increase in response at the 10 -7 M concentration, but statistical analysis with two-sample t-test revealed no significant difference compared to the response of HEK cells expressing the empty vector.
We did not exclude the possibility that the mPR and ZIP9 homologues were unresponsive due to their low expression.Hence, we applied codon usage optimization for both sequences (Supplementary Figure 8) and investigated their expression and ligand-binding interaction again (Figure 3).Although the expression of both candidates was improved after the codon usage optimization (Figure 3A), they were still found to be unresponsive to vertebrate sex steroids (Figures 3B, C).

Discussion
There is a long-standing debate regarding whether vertebrate sex steroids can affect the physiology of mollusks.Sex steroids are present in tissues of a wide range of mollusks, and their concentration sometimes show an association with stages of gonad development or differences between tissues and sexes [reviewed by (1)].However, since the 2000s, it has become evident that mollusks possess a remarkable ability to absorb and retain vertebrate sex steroids for extended periods, even months (42-51).Consequently, their presence in molluscan tissues does not necessarily indicate an endogenous origin.Given that vertebrate sex steroids have been present in the aquatic environment -and thus in molluscan tissues -for a considerable part of molluscan evolutionary history, the question arises as to whether evolution has favored the development of signaling systems to recognize these compounds.
Sex steroid signaling in vertebrates is regulated via three possible pathways (Figure 4).First, through nuclear receptors localized in the cytoplasm or nucleus (Figure 4A) (52).Second, following post-translational modification, the nuclear receptors relocate to the plasma membrane to facilitate the rapid effects of sex steroids (Figure 4B) (52,53).Considering the absence of relevant functional nuclear receptors in mollusks, the nuclear receptor-mediation of sex steroids via these pathways can be excluded.Previous screenings of Lymnaea genome and transcriptome data did not identify functional nuclear sex steroid receptors either (2,31,54).We are careful to use the word 'functional' in this context, as molluscan genomes all contain a gene that was initially annotated as being a homologue of the vertebrate nuclear estrogen receptor (nER), although it is actually orthologous to the common ancestor of vertebrate estrogen and oxosteroid receptors (55).Moreover, all studies on mollusks to date, except one (56), have shown that molluscan nERs do not bind to E 2 and are, in fact, in an already activated state (57-59).Previously, the nER of the scallop was reported to be bind to E 2 (56).Nevertheless, the hybrid vector used in that study was made up of fish and scallop sequences, and the binding evidence was a two-fold increase at one dose of E 2 .Hence, we would argue that data are insufficient to refer to the scallop nER as a 'functional' E 2 receptor and support the proposal of (60) that sequences previously termed molluscan nER should be renamed NR3D.The third possible pathway involves rapid effects mediated by membrane receptors which are distinct from nuclear receptors (Figure 4C) (16,40,(61)(62)(63).Given that GPCRs trace their origins back into history over one billion years, whereas the original ancestral nuclear steroid receptor appeared about 500-600 million years ago (7), other works (4, 11) have very reasonably suggested that sex steroids affect the physiology of mollusks through this mechanism.
The rapid effects of E 2 in vertebrates are mediated by a single GPCR called GPER1 [for detailed reviews, see (62,63)].Our candidate was unresponsive to E 2 , suggesting that no GPER1 homolog is present in Lymnaea.The identified sequence likely belongs to the protostome allatostatin-C receptors, which are closely related to echinoderm somatostatin/allatostatin-C-type receptors and chordate somatostatin-type and opioid-type receptors (Supplementary Figure 3) (64, 65).Without assessing the robustness of previously reported rapid effects of E 2 on Mytilus hemocytes, we note that only the lysosomal membrane stabilitydecreasing effect has so far been properly verified.This classic 'immune response', as also referenced by the authors of the original study (4), led to the hypothesis the E 2 may act as a xenobiotic 'inflammatory agent' for mollusks.Consistent with the previous idea that recognition and deactivation of xenobiotics were significant evolutionary pressures for receptor development (66), we hypothesize that mollusks have an unknown specific membrane receptor, though not a GPER1 homolog, capable of binding E 2 and initiating a rapid immune response.
Rapid actions of P are implemented by five mPR subtypes, which are not GPCRs but belong to the progestin and adipoQ receptor (PAQR) family, in vertebrates [for recent detailed reviews, see (16,61)].The previous scenario was that mPRs are exclusive to Second messenger assay (Ca 2+ mobilization) for assessing the potential ligand-receptor interaction between vertebrate type-sex steroids and Lymnaea membrane sex steroid receptor candidates.Dose-response curve of E 2 in GPER1 candidate-expressing HEK cells (A), P in mPR candidate-expressing cells (B), and T in GPCR6A candidate or ZIP9 candidate-expressing HEK cells (C, D).Empty vector was transfected as the negative control (Mock).None of the candidates were responsive to the main steroid ligand.Data points are means ± SEM of two (A, B, D) or three (C) independent transfections.Confocal microscopic observation of immunostained HEK293 cells expressing codon usage optimized mPR and ZIP9 candidates fused with G16 protein (A) and functional receptor assay for assessing ligand-binding (B, C). (A) Expressed receptor candidates stained with anti-G16 antibody (green), nuclei stained with Hoechst 33342 (blue), and merged views of the transfectants.The candidates were localized at the plasma membrane.The expression of both candidates significantly improved due to the codon usage optimization.Bars = 50 mm.Dose-response curve of P in mPR candidate-expressing cells (B) and T in ZIP9 candidate-expressing HEK cells (C).Empty vector was transfected as the negative control (Mock).Both candidates were unresponsive to the main steroid ligand.Data points are means ± SEM of two independent transfections.Sex steroid signaling in vertebrates.(A) Classic nuclear pathway involving estrogen (ER), progesterone (PR), and androgen (AR) receptors, which are primarily located as monomers in the nucleus or as monomers bound to heat shock proteins (HSPs) in the cytoplasm.After steroid binding, dimerization, changes in receptor conformation (and nuclear localization), the receptors bind to specific steroid-response elements (SREs) to modulate gene transcription.(B) The membrane forms of ER, PR, and AR are tethered to the cell membrane and can also bind sex steroids activating G proteins (e.g., Gs) and/or kinases (e.g., PI3K/Akt pathway).(C) Membrane sex steroid receptors (GPCRs and non-GPCRs) which are unrelated to ER, PR, and AR can also bind sex steroids activating G proteins and kinases.Red X symbols mark vertebrate receptors which have no homologs in Lymnaea, while green/red X symbols indicate vertebrate receptors which have structural but, in terms of the sex steroid signaling, not functional homologs in Lymnaea.
chordates (67).However, a recent comprehensive genomic analysis of metazoans revealed homologs for mPRb and mPRg in bivalves, suggesting that the five vertebrate mPRs evolved from an ancestral metazoan mPRg (18).It is also noteworthy that mPRa, which has been shown to mediate most mPR functions in vertebrate cells (even when they express all isoforms), is absent in invertebrates.Our in silico revealed only one potential mPR candidate in Lymnaea.Conservation analysis of this sequence yielded similar results to those previously reported in mollusks (Supplementary Figures 9, 10) (17).However, our ligand-receptor interaction experiment showed that this candidate did not respond to P, suggesting the absence of an mPR homolog in Lymnaea.Adiponectin receptor, a member of PAQR family, was recently identified and showed to be involved in learning and memory formation of Lymnaea (68).Our identified sequence may also belong to the PAQR family (Supplementary Figure 4) and use adiponectin as its ligand, but further studies are required to confirm this assumption.Collectively, the present study supports the original theory that functional mPRs are exclusive to chordates.
Non-genomic effects of T in vertebrates are mediated by two proteins, GPRC6A (GPCR) and ZIP9 (non-GPCR) [reviewed by (40,69)].Our homology-searches revealed one candidate for both proteins in Lymnaea, but neither responded to T, suggesting the absence of mAR homologs in this species.GPRC6A is also known to be activated by L-a-amino acids (40).Based on our in silico analyses (Supplementary Figures 2, 5), we suppose that the identified Lymnaea sequence functions as a metabotropic glutamate receptor, although experimental validation awaits further studies.In vertebrate cells, ZIP9 has dual functions as a zinc transporter and as a mAR activating second messengers (40).Since Lymnaea ZIP9 was unresponsive to T, we hypothesize that its sole function is to participate in zinc homeostasis.
In summary, our findings strongly suggest that functional homologs of vertebrate membrane sex steroid receptors are not present in Lymnaea.Although further experiments are required on other molluscan model species as well, we propose that both classical and non-classical sex steroid signaling for endocrine responses are specific to chordates.In light of this, it remains an important and unanswered question how vertebrate steroids affect the physiology of mollusks.Vertebrate sex steroids are probably xenobiotic 'inflammatory agents' for mollusks and may be recognized by membrane receptors involved in rapid immune responses.Another potential target for endocrine disruption by vertebrate steroids in mollusks might be an ecdysone-like pathway.Ecdysone signaling was long thought to be exclusive to arthropods; however, studies from 2010 have demonstrated the presence of an orthologue of the EcR/NR1H receptor in other protostome taxa, including mollusks and annelids (70).In insects, 20hydroxyecdysone exerts both genomic and non-genomic actions (71).It is noteworthy that GPCR-mediated effects of 20hydroxyecdysone have also been described in mammals, demonstrating the activation of an endogenous signaling pathway by a clearly exogenous steroid (72).Our homology searches revealed Lymnaea sequences homologous to both insect nuclear and membrane (GPCR) ecdysone receptors (Supplementary Figure 11), paving the way for future studies aimed at investigating this potential target pathway.Our findings have implications not only for the fields of endocrinology and evolution but also for ecology: it is important to consider the signaling pathways and responsiveness of mollusks to vertebrate sex steroids when assessing the impact of these compounds on ecosystems.

FIGURE 1
FIGURE 1Confocal microscopic observation of immunostained HEK293MSR cells expressing the different Lymnaea membrane sex steroid receptor candidates fused with G16 protein.Expressed candidates stained with an anti-G16 antibody (green; top row), nuclei stained with Hoechst 33342 (blue; middle row), and merged views (bottom row) of the transfectants.The candidates were localized at the plasma membrane.The expression of the GPER1 and GPRC6A candidates were high, while the expression of the mPR1 and ZIP9 candidates were very low (only a few cells showing immunosignal were detected).Bars = 50 mm.