Identification and lateral membrane localization of cyclin M3, likely to be involved in renal Mg²⁺ handling in seawater fish

Abstract

The kidney of marine teleosts is the major site of Mg²⁺ excretion and produces urine with a high Mg²⁺ concentration. However, the transporters involved in Mg²⁺ excretion are poorly understood. The cyclin M (Cnnm; also known as ancient conserved domain protein) family comprises membrane proteins homologous to the bacterial Mg²⁺ and Co²⁺ efflux protein, CorC. To understand the molecular mechanism of Mg²⁺ homeostasis in marine teleosts, we analyzed the expression of the Cnnm family genes in the seawater (SW) pufferfish, torafugu (Takifugu rubripes), and the closely related euryhaline species, mefugu (Takifugu obscurus). Database mining and phylogenetic analysis indicated that the Takifugu genome contains six members of the Cnnm family: two orthologs of Cnnm1, one of Cnnm2, one of Cnnm3, and two of Cnnm4. RT-PCR analyses indicated that Cnnm2, Cnnm3, and Cnnm4a are expressed in the kidney, whereas other members are mainly expressed in the brain. Renal expression of Cnnm3 was upregulated in SW mefugu, whereas renal expression of Cnnm2 was upregulated in freshwater (FW) mefugu. No significant difference was observed in renal expression of Cnnm4a between SW and FW mefugu. In situ hybridization and immunohistochemical analyses of the SW mefugu kidney revealed that Cnnm3 is expressed in the proximal tubule, and its product localizes to the lateral membrane. When Cnnm3 was expressed in Xenopus laevis oocytes, whole cellular Mg²⁺ content and free intracellular Mg²⁺ activity significantly decreased. These results suggest that Cnnm3 is involved in body fluid Mg²⁺ homeostasis in marine teleosts.

seawater (sw) contains ∼53 mM Mg²⁺, a concentration ∼50 times higher than that in the plasma of marine teleosts. The bladder urine Mg²⁺ concentration of marine teleosts is 57–167 mM (4, 5, 29, 36), and urine-to-plasma ratios of Mg²⁺ can exceed 100. Renal tubular fluid secretion accounts for much or all initial urine production in glomerular or aglomerular marine teleosts, and the renal proximal tubule is the major site of fluid secretion and Mg²⁺ excretion (4, 5, 18, 33). In SW-acclimated rainbow trout (Oncorhynchus mykiss), Mg²⁺ load from glomerular filtration is ∼3.2 μmol·kg⁻¹·h⁻¹; however, excreted Mg²⁺ rises to ∼100 μmol·kg⁻¹·h⁻¹ as a result of tubular secretion (5). This is in contrast to teleosts in ion-poor freshwater (FW), in which the kidney produces hypo-osmotic urine for water elimination by reabsorbing ions, including Mg²⁺ (5, 8). In FW-acclimated rainbow trout, Mg²⁺ load from glomerular filtration is ∼11 μmol·kg⁻¹·h⁻¹, which is reduced to ∼2.9 μmol·kg⁻¹·h⁻¹ in final urine, indicating net Mg²⁺ reabsorption in the renal tubule (5).

Proximal tubules isolated from the winter flounder (Pseudopleuronectes americanus) generate a Mg²⁺ concentration of 27 mM in the tubule lumen when the medium contains 1 mM Mg²⁺ (3, 11). Mg²⁺ is secreted from the peritubular bath into the tubule lumen against electrochemical potentials; therefore, the mechanism of active Mg²⁺ transport must be powerful (5). In the urine of marine teleosts, urinary Mg²⁺ and Na⁺ concentrations are negatively correlated, suggesting that Mg²⁺ excretion is coupled with Na⁺ reabsorption (32). X-ray microanalysis of freeze-dried cryosections of the kidneys of the dogfish shark (Squalus acanthias) demonstrated the presence of small apical vacuoles containing high concentrations of Mg (over 200 mM) in the proximal tubular cells (17). Ion microscopy analysis of the kidneys of the SW killifish (Fundulus heteroclitus) demonstrated that Mg distribution in epithelial cells of the proximal tubule is diffuse and punctate (10). These observations suggest that transepithelial Mg²⁺ transport involves sequestration of Mg²⁺ in small apical vacuoles and its subsequent exocytosis at the apical membrane (5, 33). Studies of brush-border membrane vesicles isolated from the proximal tubules of the Mozambique tilapia (Oreochromis mossambicus; FW and SW) (7) and rainbow trout (FW; 14) indicate the presence of electrodiffusive Mg²⁺ transport consistent with Mg²⁺ channel-like activity. Mg²⁺ channel-like activity can facilitate tubular reabsorption, suggesting that the apical membrane is not the site of active Mg²⁺ secretion. Recently, a pufferfish homolog of a Mg²⁺ transporter, solute carrier family 41 member 1 (Slc41a1), was discovered to be localized in intracellular vacuoles of the proximal tubule (21) and implicated in vesicular Mg²⁺ accumulation and excretion; however, the molecular mechanisms of Mg²⁺ excretion remain poorly understood.

The cyclin M (Cnnm) family, also known as the ancient conserved domain protein (Acdp) family, consists of four proteins (Cnnm1–4 or ACDP1–4) in mammals (44, 45). The Cnnm family was first characterized as comprising mammalian homologs of the bacterial protein, CorC. CorC was initially identified as a Mg²⁺/Co²⁺ transport system in bacteria through genetic analysis of Salmonella typhimurium strains resistant to Co²⁺ (15). In Saccharomyces cerevisiae, MAM3 encodes a homolog of the Cnnm family whose expression levels directly correlate with the degree of Mn toxicity (47). Mammalian Cnnm2 is a plasma membrane protein comprising the following: an N-terminal signal peptide (cleaved by signal peptidases); an extracellular NH₂ terminus; three transmembrane domains; and an intracellular COOH terminus containing two cystathionine β-synthase (CBS) domains, a cyclin box motif, and a cyclic nucleotide (cNMP)-binding domain (12, 44, 45). In an investigation of Mg²⁺-dependent current in Xenopus laevis oocytes (16) and a complementary study of the Mg²⁺-deficient S. typhimurium strain MM281 (37), Cnnm2 was shown to be involved in Mg²⁺ transport. A later study on mammalian cells expressing Cnnm2 demonstrated Mg²⁺-sensitive Na⁺ current instead of Mg²⁺ current, suggesting that Cnnm2 is involved in Mg²⁺ sensing during the regulation of renal Mg²⁺ uptake (38). In a recent study, Cnnm4 expressed in human embryonic kidney (HEK) 293 cells exhibited an electroneutral Na⁺/Mg²⁺-exchange activity, which was involved in Mg²⁺ extrusion by the cells (46); similar Mg²⁺ efflux activity was also observed in HEK293 cells expressing Cnnm2 (19).

To elucidate the role of the Cnnm family in Mg²⁺ homeostasis in marine teleosts, we searched for a Cnnm family member with increased renal expression after SW acclimation. Two closely related Takifugu species, the SW tiger pufferfish (Takifugu rubripes; torafugu) and the euryhaline river pufferfish (Takifugu obscurus; mefugu) were used as animal models, and the renal expressions of Cnnm family members were compared between torafugu, SW-acclimated mefugu, and FW-acclimated mefugu (21, 24). The genome sequence of T. rubripes is available (1), enabling us to thoroughly identify members of the Cnnm family by genome database mining. In this paper, the expression of Cnnm family members was analyzed in torafugu and mefugu, and Cnnm3 was identified as a gene with upregulated renal expression after SW acclimation. Cnnm3 was highly localized to the lateral membrane of the proximal tubule and was associated with decreased cellular Mg²⁺ content when expressed in X. laevis oocytes. These results suggest that Cnnm3 is involved in Mg²⁺ efflux in the lateral membrane, and it stimulates paracellular Mg²⁺ excretion by the proximal tubule in marine teleosts.

MATERIALS AND METHODS

Animals.

Mefugu (T. obscurus) were purchased from a local dealer and held in 150-liter indoor tanks containing brackish water (3%-14% diluted SW) until use, as previously described (21, 24). For FW samples, mefugu were transferred to FW tanks and acclimated for 7–9 days before sample collection. For SW samples, mefugu in FW tanks were transferred to SW tanks and acclimated for 7–9 days. Torafugu (T. rubripes) were purchased from a local dealer and held in SW until use, as previously described (24, 31). The experimental animals were anesthetized by immersion in 0.1% ethyl m-aminobenzoate (MS-222; Tricaine; Nacalai Tesque, Kyoto, Japan) before death. The animal protocols and procedures were approved by the Institutional Animal Care and Use Committee of the Tokyo Institute of Technology, and the investigation conforms with the American Physiological Society (APS) Guiding Principles for the Care and Use of Vertebrate Animals in Research and Training.

Database analyses and construction of phylogenetic trees.

We obtained the sequences of torafugu genes from genome databases (http://genome.jgi-psf.org/ and http://www.ensembl.org/). Collected protein sequences were aligned using the CLUSTALW phylogenetic analysis service (http://www.ddbj.nig.ac.jp/) (42), and phylogenetic trees were constructed using MEGA5 software (http://www.megasoftware.net/) (40) based on the neighbor-joining method with 1,000 bootstrap replicates. Synteny analysis was performed using the Ensembl genome browser (http://www.ensembl.org/).

Semiquantitative reverse transcription polymerase chain reaction.

RNA was isolated using Isogen (Nippon Gene, Tokyo, Japan), and total RNA (5 μg) from each tissue was used for cDNA synthesis using SuperScript III (Invitrogen, Carlsbad, CA) to obtain 20 μl of cDNA solution, as previously described (27, 31). Synthesized cDNA was diluted to 160 μl with distilled water. For primer design, full-length or partial cDNAs of torafugu and mefugu Cnnms were isolated by RT-PCR from the brain (Cnnm1a, Cnnm1b, Cnnm2, Cnnm4a) and kidneys (Cnnm3, Cnnm4a) using the primers listed in Table 1 (DNA Data Bank of Japan/GenBank/European Molecular Biology Laboratory accession numbers: AB871954-AB871965).

Table 1.

List of primers used for polymerase chain reaction amplification

Gene	Species	Sequence	Remarks
Cnnm1a	Mefugu/Torafugu	TCCAGGAGCTCAAGTTCAACC	RT-PCR (S)
	Mefugu/Torafugu	GATGCGGTTCCTCTCGTT	RT-PCR (AS)
	Torafugu	CTCCCGTGTCCTTCAATAGTTCAGTCAG	Full-length cloning (S)
	Torafugu	TGCGTCAGTGCTCTTCCTCTCAGCCACAAT	Full-length cloning (AS)
	Mefugu	GCAGCAGGACTTCTCCATCTTC	Partial cloning (S)
	Mefugu	TGCGTCAGTGCTCTTCCTCTCAGCCACAAT	Partial cloning (AS)
Cnnm1b	Mefugu/Torafugu	TGACGCGTCTCCTAATGGTG	RT-PCR (S)
	Mefugu/Torafugu	CCAGTTTGGTGTCGCTGAAG	RT-PCR (AS)
	Torafugu	CTGAGCCTCACACGCATC	Full-length cloning (S)
	Torafugu	CGCGCACACACAGACTCTG	Full-length cloning (AS)
	Mefugu	CACGGAAACTACGTCCTGTGT	Partial cloning (S)
	Mefugu	CGCGCACACACAGACTCTG	Partial cloning (AS)
Cnnm2	Mefugu/Torafugu	ATCCACAGCACCCACCACTG	RT-PCR (S)
	Mefugu/Torafugu	ATGGGGTCCAGAGCCATGAG	RT-PCR (AS)
	Mefugu	TGTCACTGGCTGGTGTGTTAC	qPCR (S)
	Mefugu	CCGCAAAAAGCCCCTATCCAT	qPCR (AS)
	Mefugu/Torafugu	TCCATTCAGGTTCGCTGCCGTGCGGTTATC	Partial cloning (S)
	Mefugu/Torafugu	GGTCCACGAAAGCCAGGTCT	Partial cloning (AS)
Cnnm3	Mefugu/Torafugu	TCCTGTCCAGAGAAGTGGAAC	RT-PCR (S)
	Mefugu/Torafugu	AGGATCTGGACTCTGACCAAC	RT-PCR (AS)
	Mefugu/Torafugu	CGCAGCAGCCTGACCAAATCCAAT	Full-length cloning (S)
	Mefugu/Torafugu	CTATGATCCCTACGGGGTTTTTGTCACAG	Full-length cloning (AS)
	Mefugu	AGTGGAACACTTCAGCCCTC	qPCR (S)
	Mefugu	GTGGTTCCGCGTGTACAGAT	qPCR (AS)
	Torafugu	CCGTTAGAAACAACACGATCAG	5′ RACE, outer (AS)
	Torafugu	AGGTGGTAGAAGGTCTGCTC	5′ RACE, innner (AS)
	Torafugu	TCTGCCCGCTGACTATATCCC	3′ RACE, outer (S)
	Torafugu	TTCAAGCCTCCCGAAGGAGAAC	3′ RACE, innner (S)
	Mefugu/Torafugu	ATCCAATATGGTGGCCAGCTTGGCAG	Cloning (pGEMHE, pcDNA3) (S)
	Mefugu/Torafugu	TGCTACCCATGAGCCTCCTCTTCTGTG	Cloning (pGEMHE, pcDNA3) (AS)
	Mefugu/Torafugu	CAATATGGTGGCCAGCTTGGC	Cloning (p3xFLAG-CMV-14) (S)
	Mefugu/Torafugu	CAGAAGAGGAGGCTCATGGGAT	Cloning (p3xFLAG-CMV-14) (AS)
Cnnm4a	Mefugu/Torafugu	GGTCTTGGGTTTGGTAACGCT	RT-PCR (S)
	Mefugu/Torafugu	CGCGTGTGATCTTGACGAACT	RT-PCR (AS)
	Mefugu/Torafugu	CCCGAACTCAGTGAGCGAAACAAAC	Full-length cloning (S)
	Mefugu/Torafugu	CTTTATTTATAGGTGTGTTAATATGCGAGG	Full-length cloning (AS)
	Mefugu	AAGGTGTCCCCACAACTGCT	qPCR (S)
	Mefugu	AGGATCCGCAGCAGCACTT	qPCR (AS)
	Torafugu	GCACAAGCAGAAAAGAGATGAGG	5′ RACE, outer (AS)
	Torafugu	AGGGACTTCTTCTCCTCCAC	5′ RACE, innner (AS)
	Torafugu	CGCCCGATAAGAATAAGAGAGAC	3′ RACE, outer (S)
	Torafugu	TGCAGTTCGTCAAGATCACACG	3′ RACE, innner (S)
Cnnm4b	Mefugu/Torafugu	ACGCGTATCCCAGTGTACGA	RT-PCR (S)
	Mefugu/Torafugu	CCGCGCTGGTAAACGTAGTG	RT-PCR (AS)
	Mefugu/Torafugu	CATGCGGCTGGAGAAGAGCAACAAG	Partial cloning (S)
	Mefugu/Torafugu	CCGCGCTGGTAAACGTAGTG	Partial cloning (AS)
β-Actin	Mefugu/Torafugu	AGCGTGGGTACTCCTTCACTAC	RT-PCR (S)
	Mefugu/Torafugu	TCGTACTCCTGCTTGCTGATCC	RT-PCR (AS)
Gapdh	Mefugu/Torafugu	GGCCCAATGAAAGGCATTCT	qPCR (S)
	Mefugu/Torafugu	TGGGTGTCGCCGTTGAA	qPCR (AS)

S, sense primer; AS, antisense primer.

Semiquantitative RT-PCR was performed by GoTaq Green Master Mix (Promega, Madison, WI) using the primers listed in Table 1. Each reaction mix contained 6.25 μl of GoTaq Green Master Mix, 1.5 μl of each primer (10 μM), 0.5 μl of tissue cDNA solution, and 3.25 μl of nuclease-free water. The PCR conditions comprised 27 cycles of denaturation (94°C, 30 s), annealing (55°C, 30 s), and extension (72°C, 40 s). After PCR, 5 μl from each reaction mix was electrophoresed on a 1.5% agarose gel in Tris·HCl/acetic acid/ethylenediaminetetraacetic acid buffer. The gel was stained with 0.5 μg/ml ethidium bromide for 30 min, and the fluorescent image was analyzed using a Kodak Image Station 2000R (Eastman Kodak, Rochester, NY).

Quantitative real-time PCR.

For the quantification of mRNA levels by real-time PCR, total RNAs were extracted from the kidneys of mefugu acclimated to SW and FW (n = 3–4 for each group). RNA (5 μg) was used as a template for reverse transcription using oligo(dT) primers and the SuperScript III First-Strand Synthesis System (Invitrogen). After reverse transcription, the cDNAs were amplified using primers for Cnnm2, Cnnm3, Cnnm4a, and glyceraldehyde-3-phosphate dehydrogenase (Gapdh) (21). Reactions were performed via the SYBR Green method using a SYBR Premix Ex Taq II Kit (Takara Bio, Otsu, Japan) on a Thermal Cycler Dice Real Time System (Takara Bio). The optimized 25-μl PCR mixture contained SYBR Premix Ex Taq II (12.5 μl), 900 nM forward and reverse primers (Table 1), and template DNA (1 μl), and the reactions were performed in a 96-well plate (Takara Bio). The thermal cycling conditions comprised predenaturation for 5 min at 95°C, 40 cycles of denaturation at 95°C for 15 s, annealing at 57°C for 30 s, and final extension at 95°C for 30 s. Melting curve analysis by raising the temperature from 65 to 95°C at a rate of 0.1°C/s was used to verify specificity. For each assay, the threshold cycle value, defined as the PCR cycle at which the fluorescence signal increased above the background threshold, was determined to quantify each mRNA product. The reference gene gapdh was stably expressed. The experiments were performed in duplicate. Quantitative data were presented as means ± SE. Relative values for the expression levels of each mRNA were compared between FW and SW mefugu, and statistical significances (P values) were calculated by unpaired two-sided Student's t-test using GraphPad Prism software (GraphPad Software, San Diego, CA).

In situ hybridization.

The kidneys of SW mefugu were perfused and fixed with 10% neutral-buffered formalin, embedded in paraffin, and sectioned to a thickness of 6 μm. In situ hybridization histochemistry was performed as previously described (21). In brief, a 339-base pair fragment of mefugu Cnnm3 cDNA (nucleotides 67–405) was subcloned into the pGEM-T Easy Vector (Promega) and used as a template for the preparation of digoxigenin (DIG)-labeled cRNA probes (Roche Diagnostics Deutschland, Mannheim, Germany). Sections were hybridized using DIG-labeled cRNA probes at 60°C for 16 h, and washed several times at 60°C. The bound label was detected using nitro-blue tetrazolium/5-bromo-4-chloro-3′-indolyl phosphate solution, and the sample was counterstained with Kernechtrot (Nuclear Fast Red; Muto Pure Chemicals, Tokyo, Japan). Serial sections were stained with hematoxylin and eosin (Muto Pure Chemicals) or anti-eel Na⁺/K⁺-ATPase α-subunit rabbit serum, as previously described (25). Images were observed using a microscope (model Axioskop; Carl Zeiss AG, Oberkochen, Germany) and an AxioCam HRc camera (Carl Zeiss AG) and TOCO automatic virtual slide system (Claro, Hirosaki, Japan).

Antibody production.

Polyclonal antisera were prepared in rabbits immunized with BSA-conjugated synthetic peptides corresponding to parts of mefugu Cnnm3 (amino acid residues: CGSEEEKRVAKRLEP and EFSLFKPPEGEPKI).

Expression of cyclin M3 in Madin-Darby canine kidney cells.

For expression in mammalian cells, cDNA corresponding to the open reading frame of mefugu Cnnm3 was obtained by RT-PCR from the kidneys of SW-acclimated mefugu using KOD plus DNA polymerase (Toyobo, Tokyo, Japan) and the primers listed in Table 1, and subcloned into the pcDNA3 vector (Invitrogen). Madin-Darby canine kidney (MDCK) cells were cultured and transfected with plasmids encoding Cnnm3 (pcDNA3-Cnnm3) using FuGENE 6 transfection reagent (Roche Diagnostics, Mannheim, Germany), according to the manufacturer's protocol. For immunofluorescence experiments, 36 h after transfection, the cells were fixed and reacted with the antisera (1:1,000) for 1 h at room temperature. Subsequently, the cells were treated with Alexa Fluor 488-labeled anti-rabbit secondary antibody (1:2,000; Invitrogen) and Hoechst 33342 (100 ng/ml; Invitrogen) for 1 h at room temperature. In MDCK cells, anti-zona occludens-1 mouse monoclonal antibody (1:200, clone ZO1–1A12; Invitrogen) was used to visualize tight junctions. Preimmune serum was used for negative controls. Fluorescence images were acquired using a laser confocal microscope (model TCS SPE; Leica Microsystems Vertrieb, Wetzlar, Germany) and processed using Leica Application Suite Advanced Fluorescence software (Leica Microsystems Vertrieb).

Expression of cyclin M3 in HEK293 cells and Western blot analysis.

An expression vector for Cnnm3 with a C-terminal 3xFlag epitope tag was obtained by subcloning a PCR product using the primers listed in Table 1 into the p3xFLAG-CMV-14 Vector (Sigma-Aldrich, St. Louis, MO). HEK293 cells were cultured and transfected with plasmids encoding Cnnm3 or FLAG-tagged Cnnm3 using FuGENE 6 transfection reagent, according to the manufacturer's protocol. Two days after transfection, the cells were washed in PBS and lysed in PBS containing 1% Triton X-100 and protease inhibitor mixture at 4°C. Protein concentrations in the samples were determined using a bicinchoninic acid protein assay kit (Thermo Fisher Scientific, Rockford, IL). The proteins (10 μg of protein per lane) were separated by SDS-PAGE using a 10% polyacrylamide gel, and electroblotted on to a polyvinylidene difluoride (PVDF) membrane. After blocking in Tris-buffered saline with Tween 20 (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Tween 20) containing 5% nonfat milk for 1 h at room temperature, the PVDF membrane was incubated with anti-Cnnm3 rabbit serum (1:10,000), anti-FLAG monoclonal antibody M2 (3.5 μg/ml; Sigma-Aldrich), or preimmune rabbit serum (1:10,000) for 16 h at 4°C. Bound antibodies were detected using peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) and chemiluminescent substrate solution (Immobilon Western Chemiluminescent HRP Substrate; EMD Millipore, Billerica, MA), and the signals were captured using an LAS 4000 Mini biomolecular imager (GE Healthcare, Piscataway, NJ).

Immunohistochemistry.

SW mefugu kidney specimens were fixed with paraformaldehyde (PFA) solution, and frozen sections (6 μm) were prepared as previously described (21, 22). The sections were blocked with 10% FBS in PBS for 1 h at room temperature and then reacted with anti-Cnnm3 or preimmune sera (1:2,000) for 16 h at room temperature. Sections were then washed with PBS and incubated with a mixture of Alexa Fluor 488-labeled anti-rabbit secondary antibody (1:2,000), tetramethylrhodamine isothiocyanate (TRITC)-labeled phalloidin (0.15 μM; Sigma-Aldrich), and Hoechst 33342 (100 ng/ml) in PBS containing 10% FBS at 20°C for 1 h. After washing with PBS, the sections were mounted with a cover slip, and fluorescence images were obtained using a laser confocal microscope as described earlier.

Functional analyses of expression of cyclin M3 in X. laevis oocytes.

The entire coding region of mefugu Cnnm3 cDNA was inserted into the pGEMHE X. laevis expression vector. The plasmid was linearized with NotI, and cRNAs transcribed in vitro using the T7 mMessage mMachine Kit (Ambion, Austin, TX). X. laevis oocytes were dissociated using collagenase, as previously described (34) and injected with 50 nl of water or a solution containing cRNA at 0.5 μg/μl (25 ng/oocyte) using a Nanoject II Injector (Drummond Scientific, Broomall, PA). Oocytes were incubated at 16°C in OR3 medium and were studied 3–4 days after injection.

Intracellular localization of Cnnm3 was analyzed by immunohistochemistry. Oocytes were washed in ND96 medium, fixed in 2% PFA in 100 mM phosphate buffer (pH 7.4) at 4°C for 1 h, and washed in ND96. The ND96 medium contained 96 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, and 5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (pH 7.5). Frozen sections (6 μm) were prepared and stained as previously described (28). Anti-Cnnm3 rabbit serum (1:1,000) was used as the primary antibody. After washing with PBS, the sections were incubated with anti-rabbit Alexa Fluor 488-labeled secondary antibody (1:2,000), and fluorescence images were obtained using a laser confocal microscope as described above.

The extracellular [Mg²⁺]-dependent membrane current of oocytes was analyzed by the two-electrode voltage clamp method, as previously described (23). The current-voltage (I/V) relationships of cRNA or water-injected oocytes in the presence of different Mg²⁺ concentrations were analyzed. Each oocyte was perfused with ND96 solution and clamped at a holding potential of −60 mV in ND96. Subsequently, each oocyte was perfused with ND96 containing different Mg²⁺ concentrations. At each change of solution, oocytes were perfused in solution until the current stabilized, following which the I-V relationship was recorded. We prepared 65 Mg-ND96 by replacing 96 mM NaCl with 64 mM MgCl₂. We prepared 0 Mg-ND96 (Mg²⁺-free ND96) by replacing 1 mM MgCl₂ with 1.5 mM NaCl. Test solutions with differing Mg²⁺ concentrations were prepared by mixing 65 Mg-ND96 and 0 Mg-ND96. Mg²⁺-elicited currents were measured using ND96 containing 0, 1, and 20 mM Mg²⁺ and were calculated by subtracting I_{(no Mg²⁺)} from I_(Mg²⁺).

Total Mg²⁺ content in X. laevis oocytes was analyzed using inductively coupled plasma atomic emission spectrometry (ICP-MS; model ICPS-8100; Shimadzu, Kyoto, Japan). Oocytes injected with water or cRNA for mefugu Cnnm3 were washed in ND96 solution and incubated in ND96 solution for 1 h at 23°C. Subsequently, the oocytes were washed briefly with water and dried at 65°C for 1 h. The oocytes (five oocytes/tube) were digested with 1 ml of concentrated nitric acid in polytetrafluoroethylene tubes at 90°C for 30 min and 120°C for 3 h, in that order, and then left to dry completely at 90°C. The residues were dissolved in 0.08 M nitric acid containing 5 μg/ml Be, and Mg²⁺ concentration was measured using ICP-MS with Be as an internal standard. The average volume of oocytes was estimated on the basis of their diameters measured using a microscope under the assumption that the oocytes were spherical.

The intracellular free ion activity of oocytes was analyzed as previously described (35). Mg²⁺ activity (aMg_i) of oocytes was analyzed with Mg²⁺-selective microelectrodes, which were filled with magnesium ionophore II-cocktail A (Fluka) at the tip and backfilled with 1 mM MgCl₂ solution. aMg_i was measured as the difference between the Mg²⁺ electrode and a KCl electrode impaling the oocyte. The Mg²⁺ electrodes were calibrated with 1 and 0.1 mM MgCl₂ solutions containing 100 mM KCl, and were ∼1,000-fold selective for Mg²⁺ vs. Na⁺ and K⁺. The Mg²⁺ electrodes detected Ca²⁺ approximately eight-fold more sensitively than Mg²⁺, and we used Ca²⁺-free ND96 solution for a point calibration of each experiment. In the cytosol, the Ca²⁺ level is generally ∼100 nM (i.e., ∼10,000 times lower than that of Mg²⁺); thus, the aMg_i measured by the Mg²⁺ electrodes was unaffected by intracellular Ca²⁺.

The Mg²⁺ content and aMg_i of each oocyte are presented as means ± SE. Values were compared between Cnnm3- and water-injected oocytes, and the statistical significance (P values) was calculated by unpaired two-sided Student's t-test using the GraphPad Prism software.

RESULTS

Identification of members of the takifugu cyclin M family and phylogenetic analysis.

Phylogenetic analyses of six Cnnm genes in the genome database of torafugu (T. rubripes; Fig. 1) demonstrated that torafugu Cnnm genes comprise two Cnnm1, one Cnnm2, one Cnnm3, and two Cnnm4 orthologs. The numbers of Cnnm paralogs differed among teleosts (Table 2).

Fig. 1.

Phylogenetic analysis of the Cyclin M family. The tree was generated on the deduced amino acid sequences using the maximum likelihood method with CLUSTALW (42) and MEGA5 (40). Solid and open circles indicate branches corresponding to the teleost-specific third-round whole genome duplication (R3) and the linage-specific gene duplication, respectively. The numbers show bootstrap values (200 replications). The scale bar represents a genetic distance of 0.05 amino acid substitutions per site.

Table 2.

Cyclin M genes in the mammalian and teleost genomes

Species	Cnnm gene (Ensembl gene ID or GenBank/EMBL/DDBJ accession number)
Human	Chromosome 10		Chromosome 2
	CNNM1 (ENSG00000119946)	CNNM2 (ENSG00000148842)	CNNM3 (ENSG00000168763)	CNNM4 (ENSG00000158158)
Mouse	Chromosome 19		Chromosome 1
	Cnnm1 (ENSMUSG00000025189)	Cnnm2 (ENSMUSG00000064105)	Cnnm3 (ENSMUSG00000001138)	Cnnm4 (ENSMUSG00000037408)
Rat	Chromosome 1		Chromosome 9
	Cnnm1 (ENSRNOG00000016302)	Cnnm2 (ENSRNOG00000020113)	Cnnm3 (ENSRNOG00000016032)	Cnnm4 (ENSRNOG00000015886)
Cow	Chromosome 26		Chromosome 11
	Cnnm1 (ENSBTAG00000011956)	Cnnm2 (ENSBTAG00000012857)	Cnnm3 (ENSBTAG00000010846)	Cnnm4 (ENSBTAG00000019213)
Zebrafish	Chromosome 13		Chromosome 5
	Cnnm1 (1a) (ENSDARG00000078087)	Cnnm2a (ENSDARG00000061195)		Cnnm4a (ENSDARG00000074309)
	Chromosome 1		Chromosome 11
		Cnnm2b (ENSDARG00000078733)	Cnnm3 (ENSDARG00000074853)	Cnnm4b (ENSDARG00000077508)
Medaka	Chromosome 15			Ultracontig 107
	Cnnm1a (ENSORLG00000007759)			Cnnm4a (XM_004086453)
	Chromosome 1		Chromosome 12
	Cnnm1b (ENSORLG00000000036)	Cnnm2 (2b) (ENSORLG00000012918)	Cnnm3 (ENSORLG00000000826)	Cnnm4b (ENSORLG00000000803)
Stickleback	Group VI		Group XIII
	Cnnm1a (ENSGACG00000007529)			Cnnm4 (4a) (ENSGACG00000014606)
	Scaffold 369	Group IX	Group XIV
	Cnnm1b (ENSGACG00000000245)	Cnnm2 (2b) (ENSGACG00000018261)	Cnnm3 (BR001184)
Tetraodon	Chromosome 17		Chromosome 12
	Cnnm1 (1a) (ENSTNIG00000002757)			Cnnm4a (ENSTNIG00000015632)
	Chromosome 18		Chromosome 4
		Cnnm2 (2b) (ENSTNI G00000012089)	Cnnm3-1 (ENSTNIG00000013561) Cnnm3-2 (ENSTNIG00000013554)	Cnnm4b (ENSTNIG00000013551)
Takifugu	Scaffold 24		Scaffold 4
	Cnnm1a (ENSTRUG00000007603)			Cnnm4a (ENSTRUG00000015973)
	Scaffold 7891/102	Scaffold 62	Scaffold 27
	Cnnm1b (AB871955)	Cnnm2 (2b) (ENSTRUG00000014689)	Cnnm3 (ENSTRUG00000018580)	Cnnm4b (ENSTRUG00000018576)

Paralogous genes in the genomes of teleosts are generally of two types: paralogs generated by the teleost-specific third-round whole-genome duplication [known as “ohnologs” after R3 (9, 41)]; and paralogs generated by lineage-specific gene duplications.

To understand the evolutionary history of the Cnnm paralogs, synteny analyses were performed using the human, mouse, rat, cow, zebrafish, medaka, stickleback, and pufferfish genome databases (Tetraodon and Takifugu; data not shown). As in the human CNNM family (44), Cnnm1 and Cnnm2 are located on the same chromosome, whereas Cnnm3 and Cnnm4 are tandemly arranged on another chromosome in mammals (Table 2). Two paralogs of Cnnm1 (Cnnm1a and Cnnm1b) and Cnnm4 (Cnnm4a and Cnnm4b) were identified in the genome databases of teleosts, indicating that these are ohnologs after R3. Cnnm1b was not present in the genome databases of zebrafish and Tetraodon, and Cnnm4b was absent in that of the stickleback, suggesting that these ohnologs have been deleted in each lineage. Two ohnologs of Cnnm2 (Cnnm2a and Cnnm2b) were present in the genome database of zebrafish; however, only one Cnnm2 was found in those of medaka, stickleback, Tetraodon, and Takifugu. Synteny and phylogenetic analyses demonstrated that the single Cnnm2 gene in each teleost is the ortholog of zebrafish Cnnm2b. Lineage-specific tandem gene duplication of Cnnm3 was found in the Tetraodon genome database.

Tissue distribution and salinity-dependent expression of members of the takifugu cyclin M family.

The tissue distribution of torafugu (SW) and mefugu (SW and FW) Cnnms was analyzed by semiquantitative RT-PCR (Fig. 2). In torafugu and mefugu, Cnnm3 and Cnnm4a were highly expressed in various tissues, including the kidneys (Fig. 2, A and B, left, 27 cycles). Cnnm1a, Cnnm1b, Cnnm2, and Cnnm4b were primarily expressed in the brain. A high level of Cnnm2 expression was also observed in the kidneys of FW mefugu, but not in SW mefugu or torafugu.

Fig. 2.

Expression of the members of the cyclin M family in seawater and euryhaline pufferfishes. A and B: expression of cyclin M (Cnnm) family members in the seawater (SW) pufferfish (torafugu, Takifugu rubripes) and the euryhaline pufferfish (mefugu, Takifugu obscurus) using semiquantitative reverse transcription-PCR. β-actin was used as an internal control. C: real-time PCR quantification of Cnnm2, Cnnm3, and Cnnm4a expressions in the kidney of freshwater- and SW-acclimated mefugu. Values are expressed as means ± SE of the relative expression levels (*P < 0.05; n = 3 or 4 for each group). F, freshwater; S, seawater.

Quantitative real-time PCR analysis demonstrated that the renal expression of Cnnm3 in SW mefugu was 2.2 times higher than that in FW mefugu (P < 0.05, n = 4). In contrast, renal expression of Cnnm2 in FW mefugu was 4.2 times higher than that in SW mefugu (P < 0.05, n = 3 or 4). No significant difference was observed in the expression of Cnnm4a between cDNAs isolated from the kidneys of SW and FW mefugu.

Expression of cyclin M3 in the proximal tubule.

A DIG-labeled antisense cRNA probe for Cnnm3 produced strong signals in some renal tubules and weak signals in the remaining renal tubules and in a number of cells that do not constitute nephrons (Fig. 3, A and G). No hybridization was observed with sense probes (Fig. 3B). To identify nephron segments, serial sections were stained with anti-Na⁺/K⁺-ATPase antibody (Fig. 3, C and F), nonimmune rabbit serum as a negative control (Fig. 3D), or hematoxylin (Fig. 3E). Strong positivity was evident in cells in the proximal tubule, which were characterized by the presence of a brush border and shallow basolateral infoldings labeled by the anti-Na⁺/K⁺-ATPase antibody (Fig. 3, E–G) (25).

Fig. 3.

In situ hybridization analysis of cyclin M3 in the seawater mefugu kidney. Paraffin-embedded kidney sections were stained with digoxigenin-labeled antisense (A and G) and sense (B) cRNA probes for mefugu cyclin M3, anti-Na⁺/K⁺-ATPase antibody (C and F), nonimmune rabbit serum (D), and hematoxylin and eosin (E). Low- (A–D) and high- (E–G) magnification images are shown. PI, proximal tubule segment I; PII, proximal tubule segment II; D, distal tubule; CT, collecting tubule; CD, collecting duct; G, glomerulus. Scale bars: 50 μm.

Lateral membrane localization of cyclin M3 in the proximal tubule and in Madin-Darby canine kidney cells.

The polarized mammalian epithelial cell line MDCK was transfected with the Cnnm3 expression vector, and the localization of Cnnm3 was determined by indirect immunofluorescence using anti-Cnnm3 antiserum and optical sectioning microscopy. Cnnm3 was localized to the lateral membrane in MDCK cells (Fig. 4A). No signal was obtained from untransfected MDCK cells.

Fig. 4.

Localization of cyclin M3 in the proximal tubules of the seawater mefugu. A: renal immunolocalization of cyclin M3 (Cnnm3) in the seawater mefugu. Sections of mefugu kidney were stained with anti-Cnnm3 antiserum (green), phalloidin-tetramethylrhodamine isothiocyanate (red), and Hoechst 33342 (blue). B: Polarized distribution of Cnnm3 in Madin-Darby canine kidney (MDCK) cells. Anti-Cnnm3 antiserum (green), anti-zona occludens-1 antibody (red), and Hoechst 33342 (blue) were used to stain MDCK cells transiently transfected with or without an expression vector pcDNA3-Cnnm3. Confocal XY maximum projection image and XZ (vertical) sections are shown.

In frozen sections of the kidneys of SW-acclimated mefugu, strong immunoreactivity for Cnnm3 was detected at the lateral membrane of the proximal tubule (Fig. 4B). The proximal tubules were identified by staining the apical brush borders with the F-actin marker phalloidin-TRITC (36). No signals were observed in negative controls stained with preimmune serum.

Western blotting analysis of cyclin M3 expressed in HEK293 cells.

The specificity of anti-Cnnm3 antiserum was also confirmed by Western blot analyses of cultured mammalian cells exogenously expressing Cnnm3 (Fig. 5). Specific signals for the Cnnm3 monomer (∼85 kDa) and potential dimer (∼180 kDa) were detected by anti-Cnnm3 or anti-FLAG antibodies. A similar pattern was observed in the Western blot analysis of mouse Cnnm2 (12). The calculated molecular masses of Cnnm3 and FLAG-tagged Cnnm3 are 83 and 87 kDa, respectively. No signal was detected when preimmune serum was used for the analysis.

Fig. 5.

Western blot analysis of Cnnm3 in HEK293 cells. Proteins from HEK293 cells transfected with or without vectors expressing Cnnm3 or FLAG-tagged Cnnm3 were analyzed with anti-Cnnm3 antiserum, anti-FLAG monoclonal antibody, or preimmune rabbit serum.

Multiple sequence alignment of cyclin M3.

Figure 6 shows multiple sequence alignment of Cnnm3. Amino acid sequences of Cnnm3 in T. rubripes and T. obscurus share 100% identity; thus, they are termed Takifugu Cnnm3 here. Takifugu Cnnm3 has five hydrophobic regions homologous to the four transmembrane domains and one intermembrane domain of mouse Cnnm2 (12). Takifugu Cnnm3 also has a domain of unknown function (DUF21), two cystathionine β-synthase (CBS) domains, and a cNMP-binding domain, similarly to other members of the mammalian Cnnm family (12, 44). Amino acid sequences of the NH₂ terminus (1–175) and DUF21 domain (176–331) in Takifugu Cnnm3 were only 31% and 34% identical to those in human CNNM3, respectively. Conversely, the amino acid sequences of the CBS domain pairs (365–499) and COOH terminus (500–747) were 81% and 64% identical to those of human CNNM3, respectively.

Fig. 6.

Multiple sequence alignment of cyclin M3. Conserved residues are indicated by black boxes. Four predicted hydrophobic segments of human and Takifugu cyclin M3 (Cnnm3) are indicated by gray overlines and underlines, respectively. A hydrophobic segment homologous to the intermembrane domain of human Cnnm2 (12) is indicated by a gray dotted line. The domain of unknown function (DUF21), cystathionine β-synthase (CBS) domain, cyclin box motif, and cyclic nucleotide (CNMP)-binding domain of human Cnnm3 defined by Wang et al. (44) are indicated by black overlines. Cnnm3 of T. rubripes and that of T. obscurus share 100% identity in amino acid sequences.

Activity of cyclin M3 expressed in Xenopus laevis oocytes.

Immunohistochemical analysis of X. laevis oocytes injected with cRNA for mefugu Cnnm3 showed that the majority of exogenously expressed Cnnm3 localizes to the plasma membrane (Fig. 7A). Two-electrode voltage-clamp analyses observed no significant changes in membrane current between Cnnm3- and water-injected oocytes incubated in media containing 0, 1, and 20 mM Mg²⁺ (Fig. 7D). Whole Mg²⁺ content of Cnnm3-injected oocytes was 12.7 ± 0.14 nmol (n = 10), which was 9% lower than that of water-injected oocytes (13.9 ± 0.1 nmol, n = 10; P < 0.001). Using an estimated average oocyte volume of 1.05 μl, the calculated concentrations of whole Mg²⁺ were 12.1 and 13.2 mM in Cnnm3-injected and control oocytes, respectively (Fig. 7B). Intracellular free Mg²⁺ (aMg_i) of Cnnm3-injected oocytes measured using a Mg²⁺ electrode was ∼40% lower than that of control oocytes (P < 0.05, n = 5–7; Fig. 7C).

Fig. 7.

Reduced cellular Mg²⁺ content in Xenopus laevis oocytes expressing cyclin M3. A: surface expression of cyclin M3 (Cnnm3) in oocytes. Immunohistochemical analysis was performed on oocytes injected with cRNAs for Cnnm3 or water (control) using anti-Cnnm3 antibody. B: reduced basal Mg²⁺ content in Cnnm3 oocytes. Cnnm3 oocytes and control oocytes were incubated in ND96 saline for 1 h and cellular Mg²⁺ content was measured by inductively coupled plasma atomic emission spectrometry. *P < 0.001; n = 10 for each group. C: reduced intracellular free Mg²⁺ activity (aMg_i) in Cnnm3 oocytes. aMg_i of Cnnm3 and control oocytes were analyzed with Mg²⁺-selective microelectrodes in ND96 saline. **P < 0.05; n = 5–7 for each group. D: current voltage (I-V) relationships of Cnnm3 and control oocytes in solution containing various Mg²⁺ concentrations. Extracellular Mg²⁺-elicited current was not observed in either Cnnm3 or control oocytes.

DISCUSSION

In the present study, Cnnm3 was identified as a candidate Mg²⁺ transporter that mediates proximal tubular Mg²⁺ secretion in marine teleosts. Cnnm3 is highly expressed in the kidneys of pufferfishes, and renal expression of Cnnm3 significantly increases in euryhaline mefugu acclimated to SW, suggesting that Cnnm3 is involved in SW acclimation in pufferfishes. In the kidney, Cnnm3 is expressed in the proximal tubule, which is known to be the site of active Mg²⁺ secretion in marine teleosts (3, 4). In the proximal tubule, mefugu Cnnm3 is expressed in the lateral membrane, similarly to CNNM2 in the human kidney (38) and Cnnm4 in the mouse intestine (46). We were unable to detect Mg²⁺-elicited current, i.e., Mg²⁺-channel activity, in Cnnm3 oocytes; however, whole Mg²⁺ content and free intracellular Mg²⁺ activity are significantly reduced in Cnnm3-injected oocytes, suggesting that pufferfish Cnnm3 mediates or regulates electroneutral Mg²⁺ efflux. Single nucleotide polymorphisms in human CNNM3 loci are associated with serum Mg²⁺ levels (30), and the kidney is a tissue that highly expresses Cnnm3 in the mouse (45). These studies and our present data on pufferfishes strongly suggest that Cnnm3 is involved in renal Mg²⁺ handling in vertebrates. However, a recent study (19) showed that no significant Mg²⁺-efflux activity is observed in HEK293 cells expressing human CNNM3, whereas expression of mouse Cnnm2 and human CNNM4 in HEK293 cells causes significant Mg²⁺ efflux. Further analyses are required to clarify the role of Cnnm3 in Mg²⁺ extrusion by the cell.

In the genome databases of teleosts, at least one orthologous gene existed for each member of the mammalian Cnnm family. Tissue expression patterns of each member of the Cnnm family are similar between the mouse (45) and Takifugu species (torafugu and mefugu). Mouse Cnnm1 and Takifugu ohnologs are expressed mainly in the brain. Cnnm3 and Cnnm4 are widely expressed in both the mouse and Takifugu tissues. In Takifugu species, Cnnm4b is mainly expressed in the brain, whereas Cnnm4a is widely expressed except in specific tissues, including the brain, suggesting that the expression of the two Cnnm4 ohnologs diversified in teleosts after R3.

In the mouse, Cnnm2 is expressed in the brain and kidney (45). Conversely, in SW pufferfishes, Cnnm2 is primarily expressed in the brain, and only weak expression of Cnnm2 is evident in the kidney. In mefugu acclimated to FW, relatively high-level expression of Cnnm2 was evident in the kidneys. In the mouse, renal expression of Cnnm2 increases when they are fed a low-Mg²⁺ diet, and the expression of Cnnm2 in distal convoluted tubule cells increases when cultured in low-Mg²⁺ medium (16). In humans, CNNM2 is mutated in dominant familial hypomagnesemia (38) and is associated with serum Mg²⁺ levels (30), coronary artery disease (13), and hypertension (39). Arjona et al. (2) reported that renal and intestinal expression of Cnnm2 increases in zebrafish fed a Mg²⁺-deficient diet. These results suggest that Cnnm2 is commonly involved in Mg²⁺ retention in Mg²⁺-poor environments in both mammals and teleosts (8), and that the expression of teleost Cnnm2 is downregulated in Mg²⁺-rich environments, such as SW.

Perspectives and Significance

If Takifugu Cnnm3 mediates or regulates Mg²⁺ efflux across the lateral membrane of the proximal tubule, this is quite advantageous for SW fish for the following reasons: 1) the proximal tubules of SW fish have a weak lumen-negative transepithelial potential that drives paracellular cation secretion (5, 6); and 2) Cnnm3-mediated or -stimulated Mg²⁺ secretion may concentrate Mg²⁺ into the narrow intercellular space, increasing Mg²⁺ concentration there and enhancing paracellular Mg²⁺ secretion (Fig. 8).

Fig. 8.

Comparison of hypothetical transport models for Mg²⁺ in the kidneys of seawater fish and mammals. A: Mg²⁺ secretion by the proximal tubule of seawater fish. Two candidate Mg²⁺-efflux transporters, solute carrier family 41 member 1 (Slc41a1; 21), and cyclin M3 (Cnnm3) are localized to the intracellular vesicles and the lateral membrane, respectively. Although 2 Na⁺/Mg²⁺ exchange activity of each transporter has not yet been demonstrated in fish, human SLC41A1 and CNNM4 are reportedly 2 Na⁺/Mg²⁺ exchangers (26, 46). An apical Cl⁻ channel that produces lumen-negative transepithelial potential and a basolateral channel for Mg²⁺ loading have not yet been identified in the fish kidney. B: Mg²⁺ reabsorption by the thick ascending limb of Henle in mammals (20). Lumen-positive transepithelial membrane potential is maintained by the renal outer medullary K⁺ channel (ROMK; inwardly rectifying K⁺ channel Kir1.1) in the mammalian kidney and drives paracellular Mg²⁺ reabsorption. C: Mg²⁺ reabsorption by the distal convoluted tubule in mammals. Transient receptor potential (TRP) M6 and TRPM7 heteromers are considered to form an Mg²⁺ channel at the apical membrane (43). Cnnm2 is present in the basolateral membrane and is the strongest candidate for mediation or regulation of basolateral Mg²⁺ extrusion in the mammalian kidney (38).

It is believed that the transepithelial potential is insufficient for paracellular Mg²⁺ secretion against the ∼20-fold concentration gradient, but this conclusion was made under the assumption that Mg²⁺ concentration in the narrow intercellular space is equal to that in the peritubular fluid. When luminal Mg²⁺ is 27 mM and the transepithelial potential is 1.9 mV (lumen-negative; 5), Mg²⁺ is secreted paracellularly if the Mg²⁺ concentration in the narrow intercellular space exceeds 23 mM through Cnnm3-mediated or -stimulated Mg²⁺-efflux activity. The direction of paracellular movement of Mg²⁺ proposed here opposes that of the paracellular reabsorption of Mg²⁺ by the thick ascending limb of Henle in the mammalian kidney, where lumen-positive membrane potential is generated by an apical K⁺ channel (20) (Fig. 8B). Further detailed analyses of Cnnm3 function in the lateral membrane of the proximal tubule are required to confirm this hypothesis. Previously, Slc41a1 Mg²⁺ transporter was identified on vacuoles of the proximal tubule in mefugu (21), supporting the hypotheses that Mg²⁺ secretion is achieved using a vesicular exocytosis-mediated system in teleosts (10) and elasmobranch fish (17) (Fig. 8). It may be speculated that the proximal tubular cells of marine teleosts have multiple transport systems used to secrete Mg²⁺ into the lumen to deal with large system Mg²⁺ loads (from SW) and/or to more tightly control systemic Mg²⁺ homeostasis.

GRANTS

This work was supported by the Sumitomo Foundation (100535), Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (KAKENHI 22370029 and 26292113); the Global Center of Excellence Program of the Ministry of Education, Culture, Sports, Science and Technology of Japan; and the Program for Promoting the Enhancement of Research Universities at Tokyo Institute of Technology. Work in the Romero laboratory was supported by National Institutes of Health grants EY-017732 and DK-083007.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: Z.I., N.H., and A.K. conception and design of research; Z.I., N.H., H.I., T.U., Y.K., H.D., and A.K. performed experiments; Z.I., N.H., H.I., and A.K. analyzed data; Z.I., N.H., M.F.R., and A.K. interpreted results of experiments; Z.I., N.H., and A.K. prepared figures; Z.I., N.H., and A.K. drafted manuscript; Z.I., N.H., H.I., T.U., Y.K., H.D., M.F.R., S.H., and A.K. approved final version of manuscript; M.F.R., S.H., and A.K. edited and revised manuscript.