Parallel evolution of placental calcium transfer in the lizard Mabuya and eutherian mammals

Most reptiles are egg laying, but viviparity has evolved from an ancestral oviparity state at least 115 times within the squamates (lizards and snakes) (Blackburn, 2014). One of the most relevant anatomical and physiological outcomes in the transition to viviparity in squamates is the reduction or complete loss of the eggshell (Griffith and Wagner, 2017; Herbert et al., 2006a). The eggshell plays a key role as the main source of calcium uptake during embryonic development in crocodiles, turtles and some squamates (Stewart and Ecay, 2010); thus, its loss in viviparous lizards imposes a developmental constraint favoring, in principle, the conservation of the yolk as the sole resource of calcium. However, this constraint can be overcome by the deployment of the uterine epithelium as an alternative and direct source of calcium to support embryonic growth and development (Herbert et al., 2006a). The loss of eggshell, in contrast, allowed a closer interaction between the uterine epithelium of the mother and the embryonic tissues, facilitating the development of an exceptional diversity of placental morphologies and physiologies that fulfilled the nutritional functions carried out by the yolk. Lizard viviparity thus displays a remarkable convergence with the evolution of placentation in eutherian mammals (Blackburn, 2006; Griffith and Wagner, 2017).

Viviparous squamates are mostly lecithotrophic (the egg yolk provides the nutrients – including calcium ions – required for embryonic development), in which the uterine shell glands still deposit calcium in a thin eggshell that is lost as the embryo develops (Stewart and Blackburn, 2015). Therefore, maternal calcium provision to the embryo in squamates rests on a combination of lecithotrophic, eggshell and placental (placentotrophy) resources (Blackburn, 2000; Stewart and Ecay, 2010). Most of the nutrients in highly placentotrophic skink lineages such as Trachylepis ivensii, Chalcides chalcides and Mabuya spp. (Blackburn, 1993; Blackburn and Flemming, 2012; Ramírez-Pinilla et al., 2011) are transferred through morpho-physiological specializations of complex placentas. Among these skinks, lizards from the genus Mabuya show exceptional reproductive and developmental features. Mabuya exhibits a rudimentary eggshell that disappears very early during gestation (Jerez and Ramírez-Pinilla, 2003). Mabuya also shows a premature shutdown of vitellogenesis, which causes ovulation of microlecithal eggs that lack fatty yolk platelets, reminiscent of the situation in therian mammals (Gómez and Ramírez-Pinilla, 2004; Hernández-Franyutti et al., 2005; Vieira et al., 2010). Thus, the absence of potential calcium resources due to the reduction of the eggshell, together with the precocious shutdown of vitellogenesis, provides a scenario where the only source of calcium (and virtually all nutrients) in Mabuya, is direct transfer from the uterine tissues (Ramírez-Pinilla, 2006; Ramírez-Pinilla et al., 2011). From early gestation, the placenta of Mabuya transfers water, glucose, lipids (cholesterol, vitamin E and fatty acids), proteins (related to metabolism, signaling, progesterone synthesis) and some ions including potassium, sodium, magnesium and calcium, and supports processes such as gas exchange (Barbosa-Moyano et al., 2020; Duarte-Méndez et al., 2018; Hernández-Díaz et al., 2017; Ramírez-Pinilla et al., 2011; Wooding et al., 2010). Later in gestation, a more specialized allantoplacenta is established, with a more complex arrangement of the maternal–fetal tissues that gives rise to specializations like the placentome, paraplacentome, chorionic areolas, absorptive plaques (structures associated with the exchange of nutrients) and respiratory segments (Jerez and Ramírez-Pinilla, 2001, 2003; Leal and Ramírez-Pinilla, 2008; Blackburn and Flemming, 2009).

Uncovering the cellular and molecular tools used to circumvent similar physiological challenges imposed by the evolution of a novel organ for nutrient transfer in lizards and mammals has the potential to elucidate the recursive nature of evolutionary genetic mechanisms behind the emergence of evolutionary innovations, such as the repeated evolution of placental organs in amniotes. Calcium placentotrophy, for example, is one of the most important cellular mechanisms repurposed to fulfill the new physiological shifts required during the evolution of novel organs such as the placenta (Griffith and Wagner, 2017). Hence, the study of the mechanisms linked to the transference of calcium in lizards such as Mabuya would provide a thorough understanding of the functional and evolutionary innovations related to the evolution of placentotrophy within amniotes. We tested the hypothesis that the calcium transporters involved in the pathway for incorporating calcium from the uterine epithelium of the mother into the eggshell in oviparous and viviparous reptiles were co-opted in a novel placental organ for the transport of calcium directly from the uterine epithelium to the embryo in Mabuya sp. Therefore, we expected that the morphological specialization of the mature placenta of Mabuya sp. would lead it to express some of these calcium transporters, indicating a putative role in calcium transfer in the lizard placenta.

In mammals, the transport of calcium ions in the placenta, kidney and liver is mainly mediated by dedicated Ca²⁺ transporter proteins, such as the apical plasma membrane channels transient receptor potential cation channel subfamily v member 5 and 6 (TRPV5 and TRPV6), the cytosolic proteins calbindin-D9K (CABP9K) and calbindin-D28K (CABP28K), the basolateral plasma membrane calcium ATPase (PMCA) and the sodium–calcium exchanger protein NCX (Nijenhuis et al., 2005). Interestingly, some of these protein transporters are also involved in the transport of calcium during embryonic development in oviparous, viviparous lecithotrophic and placentotrophic lizards (Herbert et al., 2006b; Thompson et al., 2007; Stinnett et al., 2012).

Previous studies in placentotrophic lizards showed that CABP28K and PMCA have tissue-specific expression in the uterine epithelium, eggshell glands, omphaloplacenta and allantoplacenta (Herbert et al., 2006a,b; Stinnett et al., 2012). By contrast, expression of TRPV5 and TRPV6 has not been reported in the uterine epithelium or the extraembryonic membranes of any squamate species, whereas CABP9K was previously detected in the placentome and paraplacentome of Mabuya sp., suggesting that CABP9K also has a role in calcium transport in placentotrophic lizards (Wooding et al., 2010). However, as the antibodies used in the CABP9K immunohistochemical analyses were specific for mammalian proteins, it is as yet unclear whether the antibodies used previously could cross-react with a homologous protein in Mabuya sp.; hence, further studies are needed to validate the results.

As CABP28K, CABP9K, TRPV5, TRPV6 and PMCA are essential for calcium transfer in the placenta of eutherian mammals, and the mechanisms by which calcium is transported in placentotrophic squamates is fundamental for the understanding of the evolution and development of the placenta, we tested the hypothesis that the molecular mechanisms underlying calcium transference during Mabuya sp. pregnancy depends on similar transporter proteins to those seen in eutherian mammals. For this, we identified and analyzed the expression of cabp28k, cabp9k, trpv6 and pmca transcripts by RT-qPCR through different stages of gestation in different organs including the liver, kidney, oviducts, embryonic chambers and placental tissues, and demonstrated the immunolocalization of some of these transporters in placental tissues of Mabuya sp.

The animals used in this study were collected under the permit for the collection of wild specimens of biological diversity for non-commercial scientific research, resolution 47, 22 January 2015, from the Autoridad Nacional Ambiental given to the Universidad Industrial de Santander. Specimens were registered at the herpetological collection of the Museum of Natural History at the university: UIS-MNH-R-3892-3897, 3902, 3912-3914, 3920, 3922, 3926–3928. All work conducted with the animals was consistent with government guidelines on the ethical treatment of animals and all applicable regulations, and followed the considerations of The Herpetological Animal Care and Use Committee (HACC) (American Society of Ichthyologists and Herpetologists) (Beaupre et al., 2004).

Tissue samples were obtained from a Colombian population of Mabuya from the candidate species described by Pinto-Sánchez et al. (2015) as Mabuya sp. IV; we will refer here to Mabuya sp. IV as Mabuya sp. Fifteen adult females of Mabuya sp. were obtained from the municipalities Curití, Piedecuesta, San Gil and Floridablanca, in the Department of Santander, Colombia. Females were kept in the lab for a week with food and water ad libitum, euthanized by intrathoracic injection of lidocaine 2% v/v, dissected, and the reproductive and gestational stages determined. Tissues collected were: (1) empty oviducts from non-pregnant females (n=3); (2) recently ovulated eggs from pregnant females (n=3); (3) embryonic chambers (at early gestation) with embryos at blastula, gastrula and early neurula stages (n=3); (4) embryonic chambers (at mid gestation) with embryos at limb bud stages (n=3); (5) placental chambers (at advanced gestation) with fetuses with body pigmentation (n=3); and (6) the kidney and liver from one of the pregnant females with recently ovulated eggs. After determination of the female's reproductive stages, the embryos at limb bud stages (mid gestation) and the fetuses with body pigmentation (late gestation) were dissected out from the embryonic chambers. Oviducts, embryonic chambers, placental tissues, liver and kidneys were immersed in RNAlater (Qiagen), whereas placental tissues, embryos and fetuses were immersed in buffered paraformaldehyde 4% v/v (PFA) and then dehydrated in a gradual series of ethanol. The embryos and fetuses were used to classify the embryonic developmental stages under a stereo microscope according to the Dufaure and Hubert (1961) table.

Total RNA extraction was carried out with InviTrap^® Spin Universal RNA Mini Kit according to the recommendations of the manufacturer. RNA concentration was measured by spectrophotometry using a NanoDrop^® ND-1000. Purity was assessed with the optical density ratio at 260 nm/280 nm (≥1.8) and by denaturing gel electrophoresis analysis. To evaluate integrity, samples were mixed with an equal volume of 2× RNA loading dye (New England BioLabs^®, NEB^®) and heated at 65°C for 10 min; the gel was prepared by melting agarose in DEPC-treated water, 10× Mops running buffer (0.4 mol l⁻¹ Mops pH 7.0, 0.1 mol l⁻¹ sodium acetate, 0.01 mol l⁻¹ EDTA) and formaldehyde. Amplified bands were compared with the ssRNA ladder from NEB^®. Once the RNA sample had been checked, 1 μg of RNA per sample was reverse transcribed with ProtoScript II^® First Strand cDNA. A primer mix that contained both oligo dT and random primers was used to obtain a maximum number of cDNA transcripts.

We identified cabp28k, cabp9k, trpv6, pmca, β-actin and gapdh transcripts using the stand-alone blast tool (Altschul et al., 1990) from the nucleotide sequences of Mabuya sp. placental transcriptome assemblage previously deposited at the European Nucleotide Archive (ENA; https://www.ebi.ac.uk/ena) (Cornelis et al., 2017). Transcripts identified from the Mabuya sp. transcriptome were used to design specific primers in Primer3 and IDT software (Table 1) in order to perform real-time PCR (qPCR). qPCR was performed with 20 ng of cDNA using the Luna Universal qPCR Master Mix (M3003S, NEB^®) in a Mastercycler^® ep realplex 4 (Eppendorf) under the following conditions: denaturation at 95°C for 3 min, followed by 40 cycles of 95°C for 15 s, primer-dependent annealing temperature for 15 s, extension step at 72°C for 20 s and melting curve. The specificity and the lack of primer dimers was validated by agarose gel electrophoresis 2% (w/v) and melting curves. A no-template control was included to test possible contamination.

Gene efficiency was calculated from a 5 serial cDNA dilution standard curve, and relative gene expression was calculated with the 2^−ΔΔC method (Livak and Schmittgen, 2001) by taking β-actin as the reference gene. The relative expression of calcium transporter transcripts was analyzed in oviducts of non-pregnant females (used as a control before the onset of the placentation), where oviducts with oviductal eggs, early embryonic chambers, and placental tissues of embryos at mid and late gestation stages were the experimental samples. We also tested the expression of calcium transporters in the kidney and liver (as positive controls because of their vital role in calcium transport and absorption) from a pregnant female with recently ovulated eggs. Statistical significance of quantified transcripts between different tissues was obtained using a t-test in RStudio (http://www.rstudio.com/); results with P<0.05 were considered statistically significant. The qPCR analysis was done with three biological experimental samples from independent females for each experimental stage with four technical replicate assays each.

To further demonstrate the relevance of the calcium transporters in Mabuya sp., we explored the expression of their homologous counterparts from a list of annotated genes derived from RNA-seq studies in the uterine tissue of non-pregnant and pregnant females, of a variety of oviparous and viviparous lizards. Thus, we searched for these genes in previous reports from Chalcides ocellatus (Brandley et al., 2012), Pseudemoia entrecasteauxii (both female viviparous species at pigmented fetus stages), Lampropholis guichenoti and Lerista bougainvillii (both female oviparous species had eggs with embryos ending somite development and starting limb-bud stages; Griffith et al., 2016) to determine the presence or absence of expression of these calcium transporters.

CABP28K and CABP9K calcium transporter proteins were analyzed by immunofluorescence in the mature allantoplacenta (late stages of gestation) of Mabuya sp. Placental tissues were paraffin-embedded and sectioned at 5 µm thickness in a rotatory microtome (Leica Biosystems^®) onto Silane-coated slides. The sections were then deparaffinized in xylene and rehydrated consecutively in a gradient of ethanol to water. Antigen retrieval was performed using 10 mmol l⁻¹ sodium citrate buffer pH 6.0 and 0.05% Tween-20 for 20 min at 98°C and cooled down in ice water for 10 min. Unspecific binding was blocked with 10% normal serum (1:1000 dilution, sc-2043 Santa Cruz Biotechnology^®, SCBT^®) in 1× PBS with 0.4% Triton X-100 for 1 h at room temperature. Sections were washed 3 times in 1× PBS, and tissue slides were incubated overnight at 4°C with primary antibodies for CABP28K (1:100, sc-365360, SCBT^®) and CABP9K (1:100, anti-S100G antibody, ab4066, Abcam^®) in blocking buffer. After washing away the primary antibodies, samples were incubated with secondary antibodies anti-mouse Alexa Fluor^® 488 (sc-516606, SCBT^®) or anti-rabbit IgG (H+L), F(ab′)2 Fragment Alexa Fluor^® 594 (#8889, Cell Signaling Technology^®) correspondingly. Secondary antibody was washed away with 1× PBS and DAPI was used for nuclear staining. Negative controls were performed in the absence of primary antibodies. Adjacent tissue sections were also counterstained with Hematoxylin and Eosin stain to have a reference of the histomorphology of the placental region analyzed by immunofluorescence. Digital images were captured using Zeiss Axio scan Z1^® and were processed with Adobe Photoshop 7.0.

A molecular phylogenetic analysis was performed with the isolated transcripts from the placental transcriptome of Mabuya sp. (Cornelis et al., 2017). The predicted amino acid sequences for the transcripts of cabp28k, cabp9k, trpv6 and pmca were obtained based on the open reading frame (which generated a similar peptide to the structures reported for related species). The protein similarity of each protein sequence was compared with amniote species by using standard protein BLAST (against non-redundant protein sequences and ref-seq) and translated BLAST (tblastn) in the nucleotide collection (nr) and refseq-RNA databases on NCBI and Ensembl genome browsers (protein accession numbers are given in Table S1). Several searches were performed by BLAST (Altschul et al., 1990). A multiple protein alignment was performed using the MUSCLE algorithm in AliView (Larsson, 2014). The best-fit amino acid model replacement was found for each protein, according to the lower Akaike information criteria value with ModelTest-NG software (Darriba et al., 2020). Maximum likelihood (ML) analysis was derived from approximately 79 amino acid positions of CABP9K, 260 amino acid positions of CABP28K, 1220 amino acid positions of PMCA and 730 amino acid positions of TRPV6. ML inference trees were constructed with 1000 bootstrap percentages on RAxML (v.8.2) (Stamatakis, 2014) on the CIPRES Science Gateway (Miller et al., 2010), and displayed using iTOL software (Letunic and Bork, 2019).

We identified putative homologous sequences of cabp28k, cabp9k, trpv6 and pmca from a published transcriptome of the placenta of Mabuya sp. at the late stage of gestation (Cornelis et al., 2017). Our molecular phylogenetic reconstruction based on the predicted CABP28K, CABP9K, TRPV6 and PMCA amino acid sequences obtained from the placental transcripts confirmed the orthologous relationships of these transcripts with the same genes in other amniotes. This result is based on well-supported bootstrap values in the corresponding ML phylogenetic trees, which showed the position of each protein sequence from Mabuya sp. within the expected protein family (Fig. 1; Figs S1, S2 and S3). The analyses made by ML phylogenetic inference were rooted using protein sequences from calretinin, calcium binding protein 1, TRPV4 and sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA), for CABP28K, CABP9K, TRPV6 and PMCA phylogenetic trees, respectively.

CABP9K (also known as S100G) protein sequences in amniotes in public databases show that cabp9k is mainly annotated in mammalian genomes, but we also found annotated sequences for some birds, tuatara, the sand lizard, and the viviparous lizard Zootoca vivipara. Given that members of the S100 protein family exhibit a high degree of sequence similarity, we also included protein sequences from closely related protein families to better test the phylogenetic position of putative CABP9K sequences outside mammalian species. To do this, we incorporated sequences for S100A1, S100A3, S100A4, S100A6 and S100A7 proteins in the ML inference analyses. Interestingly, the phylogenetic analysis demonstrated that only S100G proteins from both Mabuya sp. and tuatara are nested within the S100G lineage (Fig. 1). In addition, we compared the primary amino acid sequences and annotated structural motifs from S100G proteins and compared the protein sequences of tuatara, birds and mammals with that in the placental transcriptome of Mabuya sp. Multiple sequence alignment showed that the S100G protein of Mabuya sp. shares the conserved S100 domain and the calcium-binding site with the protein sequences from other amniote species (Fig. 2). While doing these analyses, we also identified some S100G-like sequences which do not cluster within the S100G lineage, but are more closely related to other families, so they should be further revised (Fig. 1).

Transcripts that encode the proteins that carry calcium in the placenta of eutherian mammals are also expressed in placental tissues of Mabuya sp. Interestingly, our results show that the four transporters that we analyzed are differentially expressed at different time points, with cabp28k representing the most abundantly expressed transcript in placental tissues from mid to late gestation stages. Levels of cabp28k, cabp9k, trpv6 and pmca mRNA in the liver and kidney are high relative to levels in the oviducts of non-pregnant female Mabuya sp., with the kidney showing the highest expression for all calcium transporters analyzed in this study (Fig. S4). There is a common dynamic pattern of expression of cabp28k, cabp9k, trpv6 and pmca mRNA in oviducts of non-pregnant females, oviducts with oviductal eggs, embryonic chambers with embryos at early stages, and placental tissues at mid and advanced gestation. We found that all the transcripts encoding the calcium transporters analyzed show an overall increased expression during gestation, with a marked increase in expression of all of calcium transporters from mid to late pregnancy (Fig. 3; Table S2).

The expression of cabp9k and cabp28k is downregulated in the embryonic chambers at early gestation compared with the previous oviductal egg stage, in which the relative expression levels of both mRNAs are higher. However, cabp9k and cabp28k show a marked shift, with an upregulation of their expression in placental tissues specially from mid gestation. In particular, expression of cabp28k stands out during mid to late pregnancy, reaching a particularly high peak of expression in placental tissues at advanced pregnancy compared with the other transcripts (P<0.05; Fig. 3; Table S2). Additionally, cabp28k and trpv6 mRNA expression increases considerably during mid and advanced gestation, becoming two of the most highly expressed calcium transporters during advanced gestation in this study. Although cabp9k and pmca mRNA increase their abundance from early to mid gestation, neither of them shows changes in their relative expression between mid and late pregnancy in placental tissues. Taken together, our results revealed that the relative expression of calcium transporters follows a general trend of increase through pregnancy, which seems to correspond to the higher calcium requirements of advanced embryonic developmental stages.

Immunofluorescence detection of CABP28K protein in tissue sections of late-stage placenta of Mabuya sp. indicates CABP28K is expressed in the uterine epithelium and the chorionic epithelium of the placentome and paraplacentome, specifically in the giant binucleated chorionic cells (Fig. 4A,B). CABP9K protein, in contrast, shows high expression on the basal membrane of the uterine epithelium, the apical border and the microvillar surfaces of the giant cells in the chorionic epithelium, and the cytoplasm of the uterine epithelium in the placentome and paraplacentome; however, it was not located in the uterine glands (Fig. 4C,D). These results highlight the relevance of specific anatomically complex placental structures, such as the placentome and the paraplacentome, in processes related to the transfer and regulation of nutrients such as calcium during the most demanding stages (embryonic stages 37–40) in embryonic development during gestation.

The evolution of novel organs, like the placenta, has been proposed not to involve the introduction of new genes or entire gene networks, but rather the recruitment and the repurposing (co-option) of genes expressed elsewhere in the body (Griffith and Wagner, 2017). Our results support this hypothesis, showing that placental calcium transport in Mabuya sp. resulted from the reorganization of the expression of existing genes, rather than the introduction of novel genes, in a similar way to the role of genes implied in the evolution of viviparity (Gao et al., 2019).

Our work highlights the co-option of calcium transporter proteins ancestrally expressed in non-placental tissues in amniotes, which were redeployed to fulfill the physiological function of calcium transportation in the placenta of lizards and represents a critical step in the evolution of the complex placentation in Mabuya species. Hence, the acquisition of the calcium transport machinery used in the placentas of Mabuya sp. evolved in parallel to that in eutherian mammals by redeploying some of the same calcium transporter genes present in homologous tissues in the ancestor of amniotes.

Our study also implies that calcium transfer through the placenta of Mabuya sp. is stage dependent, being highly transcriptionally active at advanced stages of gestation when the fetuses are ready to incorporate calcium to support developmental processes such as bone mineralization and calcium fetal homeostasis (Ramirez-Pinilla et al., 2006; Ramírez-Pinilla et al., 2011). This inferred calcium transportation dynamic during pregnancy in Mabuya sp. follows a similar pattern to that in viviparous lecithotrophic squamates (e.g. Virginia striatula, Elaphe guttata and Saiphos equalis; Ecay et al., 2004; Fregoso et al., 2012; Linville et al., 2010) and placentotrophic species (Niveoscincus ocellatus, Pseudemoia pagenstecheri, P. entrecasteauxii and Pseudemoiaspenceri; Table 2; Herbert et al., 2006a,b; Stewart and Ecay, 2010; Stinnett et al., 2012), and even to that in eutherian mammals (e.g. humans, rats and cows). In mammals, there is an upregulation of the calcium transport genes and proteins in uterine and placental tissues, especially during the last period of gestation, a period that corresponds to an enhanced fetal growth (Koo et al., 2012; Sprekeler et al., 2012). Future studies in other highly placentotrophic lineages, such as Trachylepis ivensii and C. chalcides, will be important to evaluate whether the dynamic expression of the calcium transporters reported here shows a similar pattern in these species.

Calbindin proteins bind to and transport calcium, and they protect cells from high concentrations of Ca²⁺ by buffering excessive intracellular levels of free ions (Belkacemi et al., 2003). Consequently, the importance of CABP28K is reflected in the wide variety of species and tissues among amniotes that employ it as their main calcium transporter, including the uterine and chorionic epithelia of oviparous and viviparous squamates (Table 2), as well as the kidney, liver, heart, egg-shell glands, yolk sac, uterus and chorioallantoic membrane of chicks (Brionne et al., 2014; Sechman et al., 1994), and in the human cerebellum, pancreas, uterus, trophoblast cells and endometrium (where it regulates endometrial receptivity) (Yang et al., 2011).

The cabp28k transcript, encoded by the calbl1 gene, is also expressed in the uterine epithelium of non-pregnant and pregnant females of oviparous species such as L. bougainvillii and L. guichenoti, and viviparous lecithotrophic and placentotrophic species such as C. ocellatus and P. entrecasteauxii (Brandley et al., 2012; Griffith et al., 2016). Similar to the pattern expression in Mabuya sp., calb1 is also significantly differentially expressed among reproductive tissues and developmental stages. For example, calb1 is highly expressed in the uterus of the yolk sac placenta rather than the uterus of the chorioallantoic placenta in P. entrecasteauxii during late pregnancy (Griffith et al., 2016).

Expression of cabp28k and cabp9k in Mabuya sp. is higher in the recently ovulated eggs than in the subsequent stage. The upregulation of both calbindins at this point might be related to the regulation of oviductal muscle contraction during the embryonic implantation process, just as in mammalian species (Sadigh et al., 2019). Our immunochemical analyses in the mature placenta of Mabuya sp. showed the expression of CABP28K and CABP9K proteins in the placentome and paraplacentome, consistent with results reported by Wooding et al. (2010). Following these ideas, calcium transporters involved in the incorporation of calcium from the mother's uterine epithelium into the eggshell in oviparous and most non-matrotrophic reptiles have been co-opted in Mabuya sp. to support the uptake of calcium from the uterine lumen to be directly given to the embryo.

Calbindin protein transporters are selectively localized in embryonic and maternal tissues of amniotes. For example, CABP28K occurs in the yolk splanchnopleure and the chorioallantoic membranes in the oviparous snake E. guttata (Ecay et al., 2004), the viviparous lecithotrophic snake V. striatula (Fregoso et al., 2010) and the chorioallantoic membrane of the placentome, as well as the omphalopleure in the substantially placentotrophic lizard P. pagenstecheri (Stinnett et al., 2012); thus, it would be very interesting to study the expression of all the calcium transporter proteins in other highly placentotrophic species in situ.

In mammals, CABP28K and CABP9K occur in the cytotrophoblast and syncytiotrophoblast cells of humans in term placentas (Belkacemi et al., 2003, 2004), and in the trophoblastic giant cells and the intraplacental yolk sac of mice (Shamley et al., 1992, 1996). CABP9K protein is localized to the maternal caruncular epithelium, the fetal chorionic epithelium and the trophoblastic binucleated cells of the placentome in the bovine placenta (Sprekeler et al., 2012). Some locations of calbindin proteins in the placental tissues of mammals are similar to those in the mature placenta of Mabuya sp., but there are still differences such as the lack of detection of calbindin proteins in the yolk sac membrane in Mabuya sp., which suggests that the yolk sac does not have an important role in calcium transport via calbindin transporters. In summary, the expression of calbindin proteins in the mature placenta of Mabuya sp. is convergent with that in homologous placental tissues in other viviparous squamates and eutherian mammals.

The presence of CABP9K in extraembryonic and uterine epithelia of a viviparous skink was first suggested by Wooding et al. (2010), who detected CABP9K protein in the advanced placenta of Mabuya sp. by immunohistochemical analysis. Additionally, the relative quantification of the cabp9k expression among a variety of tissues, together with its histological localization in the placenta, gives essential information about its putative physiological role, highlighting its importance in evolutionary processes leading to lizard viviparity and placentation in clades outside eutherian mammals. Further studies are needed to understand the molecular functions of cabp9k in squamates. Molecular cloning of full-length candidate cabp9k into expression vectors amenable for gene expression in mammalian cells will allow the biochemical characterization of this protein, particularly its calcium-binding properties using mammalian cell lines, as well as the purification of the protein by affinity chromatography (i.e. by the use of His-tags) for in vitro analysis of its calcium-binding affinity and kinetics.

Additionally, comparative gene expression analyses in situ, in both extraembryonic membranes and uterine tissues from oviparous and viviparous squamates, will be fundamental to provide an evolutionary and comparative physiological context, to better understand the mechanisms that regulate calcium acquisition in oviparous lizards, and the specific innovations during the evolution of pregnancy in squamates. Future efforts in the isolation of cabp9k in other squamates could be guided by our results. Using small and highly conserved amino acid sequences derived from squamate cabp9k multiple sequence alignments will allow the design of degenerate DNA oligonucleotides for the isolation of fragments of cabp9k in other squamates by RT-PCR, which can then be extended into full-length sequences by rapid amplification of cDNA ends (RACE) PCR.

The high expression of cabp28k and cabp9k in oviducts of Mabuya sp. with recently ovulated eggs, and then their downregulation after the embryo forms an embryonic chamber (early pregnancy), suggests that they have an important involvement in the embryo's implantation process in the oviduct. Indeed, calbindin proteins in mammalian species as mouse, pig, rhesus monkey and humans work in the regulation of intracellular Ca²⁺ concentration in the endometrium during blastocyst implantation, although CABP28K and CABP9K expression seems to be specifically controlled in the different groups. This important upregulation protects the embryo from any hypercalcemic effect, and regulates the calcium concentration in the uterine cavity to further facilitate proper placental development and functioning (Lee et al., 2009; Luu et al., 2004a,b; Tatsumi et al., 1999).

cabp9k, also known as the s100 g gene, occurs in the genome of some birds, although its expression seems not to have been investigated yet in their reproductive organs or in any other reptile until now. Despite different genes from the s100 family (s100a1, a6, a10, a13 and a16, among others) being reported as differentially expressed in the uterine tissues of non-pregnant versus pregnant squamates (e.g. C. ocellatus, P. entrecasteauxii and Mabuya sp.; Brandley et al., 2012; Cornelis et al., 2017; Griffith et al., 2016), there is little information about the presence of s100g in reptiles. Surprisingly, our database mining and phylogenetic analysis only identified the s100g transcript in tuatara, the sand lizard, and Z. vivipara, apart from our reference sequence for Mabuya sp. This fragmentary distribution could be related to unique evolutionary processes that underlie the diversity of the s100 gene family. First, it is possible that functional redundancy in the S100 family has led to the evolutionary loss of some members, and other unknown functional genetic parameters might have played a role in the generation of a scatter distribution, as a result of the differential loss and conservation of S100 orthologs and paralogs across reptiles. Second, it is possible that the absence of S100G proteins in many lineages could represent false negatives, because of imperfect coverage or assemblage of squamate transcriptomes and genomes sequenced so far (Wheeler et al., 2016; Zimmer et al., 2013).

P-type ATPases comprise an extensive family of proteins involved in the active pumping of charged substrates across biological membranes (Axelsen and Palmgren, 1998). PMCA generally presents four isoforms encoded by the atp2b1, atp2b2, atp2b3 and atp2b4 genes, which are the major mechanism for calcium deposition in the formation of the calcareous eggshell in oviparous reptiles (Thompson et al., 2007) and chicks (Akins and Tuan, 1993). The oviparous lizard L. bougainvillii expresses elevated transcript levels of these four genes in the uterus of gravid and non-gravid females, with atp2b2 being the most expressed transcript in both tissues. Moreover, even though atp2b1, atp2b2, atp2b3 and atp2b4 all occur in the uterus of the chorioallantoic and the yolk sac placenta of viviparous P. entrecasteauxii, atp2b1 and atp2b4 transcripts are the most highly expressed in the uterus of the yolk sac placenta in advanced gestation (Griffith et al., 2016). The differences between the two species can be related to their different reproductive mode and embryonic nutrition pattern, as viviparous placentotrophic lizards such as P. spenceri and P. entrecasteauxii still have uterine shell glands that express PMCA to form a thin eggshell (Herbert et al., 2006b).

By contrast, the expression of the atp2b2, atp2b3 and atp2b4 transcripts does not appear to be as significant in the uterine and placental epithelia of Mabuya sp., and only atp2b2 is significantly expressed during advanced gestation (from supplementary data in Cornelis et al., 2017). These results might be related to the fact that in Mabuya sp. a calcareous eggshell is not formed, which is very different from the situation in most squamates. Moreover, it implies that calcium pumps are chiefly involved in the transport of calcium out of the uterine epithelial cells into the uterine lumen, where it is received to form the eggshell in oviparous and some viviparous species, whereas it is directly taken up by the extraembryonic membranes in lizards with complex placentas. Taking into account that calbindin proteins (28K and 9K) are not being expressed in the uterine glands of the placenta of Mabuya sp., it would be worthwhile evaluating the pattern of expression of PMCA protein in tissue sections, to determine whether it shares a similar pattern of expression to that in other placentotrophic lizards, just as it does in Pseudemoia species.

The epithelial calcium channels TRPV5 and TRPV6 are highly selective ion channels implicated in Ca²⁺ (re)absorption in all vertebrates; they play vital physiological roles in calcium homeostasis from fish gills to metanephric kidneys (Flores-Aldama et al., 2020). The two calcium channels exhibit similarities in many ways, as they share a high level of amino acid sequence identity, comparable functional properties, and similar mechanisms of regulation. Yet, their physiological contributions toward maintaining a systemic calcium balance are distinct (Flores-Aldama et al., 2020; Peng et al., 1999).

In fact, TRPV6 serves as an important rate-limiting step in the facilitation of calcium entry into cells and is widely expressed, e.g. esophagus, stomach, small intestine, colon, kidney, placenta, pancreas, prostate, uterus, salivary gland and sweat gland of mammals (Lee et al., 2009; Nijenhuis et al., 2005; Peng et al., 1999). The broader expression of TRPV6 indicates that its physiological role is not limited to the (re)absorption of calcium ions in kidneys and calcium homeostasis. Indeed, our results show that cabp28k and trpv6 transcripts are most highly expressed in placental tissues of Mabuya sp. during the last stages of gestation. This expression pattern is similar to what occurs in the uterine and extraembryonic membranes in pigs, mice, humans and bovines, in which the highest level of TRPV6 expression usually occurs in the last trimester of pregnancy (Lee et al., 2009; Sprekeler et al., 2012), indicating it plays an active role in calcium transport to the fetus (Lee et al., 2009).

The origin of trpv5 and trpv6 within tetrapods corresponded to independent duplication events, in amphibians, sauropsids and mammals (Flores-Aldama et al., 2020). The ancestor species of each lineage had a single copy of a trpv5/6 gene, and independent lineage-specific duplication events resulted in a repertoire of trpv5 and trpv6 genes in each lineage. Therefore, although our phylogenetic analysis demonstrates the relationship of Mabuya sp. trpv6 to the exclusion of other members of trpv (trpv4) families, a more comprehensive phylogenetic analysis in the future will help to clarify the identity of our trpv6 sequence in Mabuya sp. to the exclusion of trpv5.

In summary, our results demonstrate that maternal–fetal calcium transport in Mabuya sp. is mediated by CABP28K, CABP9K, PMCA and TRPV6 proteins from the beginning until the end of gestation, with an expression pattern that steadily increases through gestation. In this sense, it is similar to the situation in other viviparous squamates and eutherian mammals, but it also has some specific patterns. Considering the co-option of the genes involved in the transport of calcium, we can say that specializations such as the placentome and paraplacentome act as the major sites of calcium transport by bound transport proteins, and given the evidence obtained here and in other amniotes, we can infer the possible route by which calcium ions are provided to the embryo. First, calcium ions are released out of the uterine epithelial cells into the uterine lumen by PMCA pumps (usually present in the uterine epithelium and glands of viviparous lizards; Herbert et al., 2006b), then calcium ions enter the materno-fetal epithelium by membrane TRPV5/6 proteins (found in the uterine glandular epithelium, and the placental and the fetal chorionic epithelium in cows; Sprekeler et al., 2012), where they are taken by cytosolic calbindin proteins present in the uterine and chorionic cells of the placentome and the paraplacentome (CABP9K mainly mediates the apical and basal region, whereas CABP28K regulates the cytoplasm; Stinnett et al., 2012; Wooding et al., 2010) to finally be received by the embryo.

We are very grateful to J. L. Fuentes and the LMMA research group for their support, and to A. Gutiérrez who helped us in the collection of the specimens.