Molecular cloning and characterization of a novel splicing variant of the Kir3.2 subunit predominantly expressed in mouse testis

Abstract

1. One of the features of weaver mutant mice is male infertility, which suggests that Kir3.2, a G-protein-gated inwardly rectifying K⁺ channel subunit, may be involved in spermatogenesis. Therefore, we have characterized the Kir3.2 isoform in mouse testis using immunological, molecular biological and electrophysiological techniques.

2. Testicular membrane contained a protein that was recognized by the antibody specific to the C-terminus of Kir3.2c (aG2C-3). Its molecular mass was ≈45 kDa, which was smaller than that of Kir3.2c (≈48 kDa). The immunoprecipitant obtained from testis with aG2C-3 contained a single band of the 45 kDa protein, which could not be detected by the antibody to the N-terminus common to the known Kir3.2 isoforms (aG2N-2).

3. A novel alternative splicing variant of Kir3.2, designated Kir3.2d, was isolated from a mouse testis cDNA library. The cDNA had an open reading frame encoding 407 amino acids, whose molecular mass was calculated to be ≈45 kDa. Kir3.2d was 18 amino acids shorter than Kir3.2c at its N-terminal end, which was the only difference between the two clones. The 18 amino acid region possesses the epitope for aG2N-2.

4. In heterologous expression systems of both Xenopus oocytes and mammalian cells (HEK 293T), Kir3.2d either alone or with Kir3.1 exhibited G-protein-gated inwardly rectifying K⁺ channel activity.

5. Prominent Kir3.2d immunoreactivity in the testis was detected exclusively in the acrosomal vesicles of spermatids, while Kir3.1 immunoreactivity was diffuse in the spermatogonia and spermatocytes. These results indicate the possibility that the testicular variant of Kir3.2, Kir3.2d, may assemble to form a homomultimeric G-protein-gated K⁺ channel and be involved in the development of the acrosome during spermiogenesis.

G-protein-gated K⁺ (K_G) channels are activated by various inhibitory neurotransmitters via G proteins in neurons, endocrine cells and cardiac myocytes (North 1989; Hille 1992; Jan & Jan 1994; Wickman & Clapham 1995; Yamada et al. 1998). They are proposed to be hetero- or homotetrameric assemblies of Kir3.0 subunits. Four kinds of Kir3.0 subunits have been isolated from mammalian cDNA libraries so far. They are Kir3.1 (also termed GIRK1 and KGA; Kubo et al. 1993; Dascal et al. 1993), Kir3.2 (also termed GIRK2; Lesage et al. 1994, 1995; Tsaur et al. 1995), Kir3.3 (also termed GIRK3; Lesage et al. 1994) and Kir3.4 (also termed GIRK4 and CIR; Krapivinsky et al. 1995). It has been suggested that the neuronal K_G channel is composed of Kir3.1 and Kir3.2 subunits (Kofuji et al. 1995; Duprat et al. 1995; Slesinger et al. 1996; Velimirovic et al. 1996), while the cardiac K_G channel is composed of Kir3.1 and Kir3.4 subunits (Krapivinsky et al. 1995). Kir3.2 possesses at least three splicing isoforms, i.e. Kir3.2a, Kir3.2b and Kir3.2c (Lesage et al. 1994, 1995; Tsaur et al. 1995; Isomoto et al. 1996). It was recently shown that the K_G channels in rat cerebral cortex are assemblies of Kir3.1 and either Kir3.2a or Kir3.2c (Liao et al. 1996; Inanobe et al. 1999) and also that at least some of the K_G channels in dopaminergic neurons of rat substantia nigra are composed of Kir3.2a and Kir3.2c (Inanobe et al. 1999). In mouse pancreatic α cells, the K_G channel may be an assembly of Kir3.2c and Kir3.4 (Yoshimoto et al. 1999). Therefore, in various tissues the splicing variants of Kir3.2 may be specifically expressed and form K_G channels in various combinations with other Kir3.0 subunits. This may be important in allowing K_G channels to play differential functional roles in various tissues.

It was recently shown that a point mutation in the Kir3.2 gene is responsible for the abnormalities in the weaver mutant mouse (Patil et al. 1995). The weaver mutation causes alteration of GYG to SYG in the signature amino acid sequence in the H5 region of Kir3.2, which results in the loss of selectivity of K⁺ ions over Na⁺ ions in the weaver K_G channel (Kofuji et al. 1996; Slesinger et al. 1996). It was also shown that the weaver K_G channel is constitutively active (Kofuji et al. 1996; Slesinger et al. 1996; Tucker et al. 1996) and is insensitive to G protein regulation (Slesinger et al. 1996; Navarro et al. 1996). Probably due to these deficiencies in the weaver K_G channel, degeneration of dopaminergic neurons in substantia nigra (Schmidt et al. 1982) and mal-migration of granule cells occur in cerebellum (Rakic & Sidman 1973). In addition to these neurodegenerative defects, it is known that male homozygous weaver mice are sterile (Harrison & Roffler-Tarlov 1994). This suggests that Kir3.2 may be expressed in testis and may play a critical role in spermatogenesis.

In this study, we have examined Kir3.2 isoforms in mouse testis using immunological, molecular biological and electrophysiological techniques. We found a novel splicing isoform of Kir3.2 in testis and designated it Kir3.2d. Kir3.2d either alone or with Kir3.1 could form functional K_G channels in both Xenopus oocytes and mammalian cells. The Kir3.2d immunoreactivity appeared specifically at the acrosome of spermatids but not in either spermatogonia or spermatocytes. Thus, it is suggested that Kir3.2d participates in the development of the acrosome during spermiogenesis.

METHODS

Antibodies

Rabbit polyclonal antibodies (aG2N-2 and aG2A-5) were raised with synthetic peptides corresponding to amino acids 4–17 and 385–398, respectively, of Kir3.2a (Lesage et al. 1994), while the antibody named aG2C-3 was generated in guinea-pig against a synthetic peptide corresponding to amino acids 415–425 of Kir3.2c (Lesage et al. 1995). Their properties have been described previously in detail (Inanobe et al. 1999). In brief, aG2C-3 specifically recognizes Kir3.2c, while aG2A-5 detects both Kir3.2a and Kir3.2c. aG2N-2 recognizes all three variants, Kir3.2a, Kir3.2b and Kir3.2c, at their N-termini. To detect Kir3.1, we used the aG1C-1 antibody raised aginst a synthetic peptide corresponding to the C-terminus (amino residues 488–501) of rat Kir3.1 (Inanobe et al. 1995b).

Immunoblot and immunoprecipitation

Membrane fractions of testis and neocortex were prepared from Balb/c mice as described in Inanobe et al. (1995a). Balb/c mice were anaesthetized with sodium pentobarbital (60 mg kg⁻¹, i.p.) and tissues were removed. These procedures are in accordance with the guidelines for the use of laboratory animals of Osaka University Medical School. Decapsuled testes and cereberal cortex were homogenized with three volumes of the preparation buffer, which contained 20 mM Hepes-NaOH (pH 7·4), 1 mM Na-EDTA, 0·5 mM Na-EGTA, 1 mM dithiothreitol, 150 mM NaCl, 0·32 M sucrose and a protease inhibitor cocktail (1 mM phenylmethanesulfonyl fluoride, 0·2 mg ml⁻¹ benzamidine:HCl, and 5 μg ml⁻¹ each of pepstatin, leupeptin and chymostatin). After centrifugation of the homogenate at 150 000 g for 30 min in a Beckman TLA-100.4 rotor, the pellet was resuspended in about 5 volumes of the preparation buffer supplemented with 850 mM NaCl and then centrifuged under the same conditions. Precipitated membranes were suspended in the preparation buffer and stored at −80°C. Each Kir3.2 isoform was transfected into human embryonic kidney (HEK 293T) cells with the LipofectAMINE reagent (Life Technologies Inc.; Horio et al. 1997). Two days after transfection, HEK 293T cells grown on Petri dishes (9 cm diameter) were washed twice with 10 ml PBS and collected in 1 ml of the preparation buffer. The cell suspension was sonicated with a TOMY Ultrasonic Disruptor (UD-201) and the membrane fraction was obtained as described above. After the separation of the membrane proteins by SDS-PAGE, channel subunits were analysed using their antibodies at a concentration of 0·5–1 μg ml⁻¹ by immunoblotting (Inanobe et al. 1999). Immunoprecipitation and biotinylation of the membrane proteins were performed as described previously (Inanobe et al. 1995a).

Screening of a mouse testis cDNA library and DNA sequencing

A mouse testis cDNA library (Stratagene) was screened under high stringency conditions using Nco I- and Nhe I-digested mouse Kir3.2c (about 1 kb) as a probe; 5 × 10⁵ phage clones were screened with a [α-³²P]CTP-labelled probe. Hybridization was conducted in 5 × SSC (saline-sodium citrate buffer; SSC contained 0·15 M NaCl and 15 mM sodium citrate, pH 7·0), 50% (v/v) formamide, 0·1% (w/v) bovine serum albumin (BSA), 0·1% (w/v) polyvinylpyrolidone, 20 mM Na₂HPO₄ and 120 μg ml⁻¹ denatured salmon sperm DNA at 42°C for 12 h. Filters were washed three times with 0·1% SDS in 2 × SSC at 65°C for 15 min each and then exposed to X-ray films overnight at −80°C with intensifying screens. DNA sequencing was performed on both strands using an ABI Dye terminator cycle sequencing kit with an ABI PRISM 310 Genetic Analyzer (Applied Biosystems), according to the manufacturer's protocol. Under these conditions, 11 positive clones were isolated. Among them, 10 clones had the same open reading frame and the other was a partial clone. The GenBank accession number of mouse testicular Kir3.2 is AB029502.

Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis

Total RNAs from adult mouse testis and forebrain were isolated with the TRIZOL reagent (Life Technologies) and reverse-transcribed with the oligo(dT)_12-18 primer using SuperScript II RT (Life Technologies), according to the manufacturer's instructions. The cDNAs were amplified by PCR with LA-Taq polymerase (Takara). The sequences of the primers for the individual exons (see Fig. 2A) were as follows: g2ex1, gaatcttgctccgttccgagaga (forward); g2ex2, gaagaaatgacaatggccaagtt (forward); g2ex3, gtaaccggcttcttgtaccacgt (forward); g2ex4, agaagcttgatccctccactgca (forward); g2ex5-5′, aacgttgcacttcccatccttcc (reverse). PCR amplification was performed for 30 cycles at 94°C for 30 s, at 61°C for 30 s, and then at 72°C for 45 s. Amplified DNA fragments were separated in a 3% agarose gel. Subsequently, fragments were sequenced with an automatic sequencer (ABI PRISM 310 Genetic Analyzer).

Figure 2. Molecular cloning of testicular Kir3.2 isoform.

A, sequence of the 5′-region of testicular Kir3.2 isoform (Kir3.2d). Exon 3 and exon 5 correspond to the exons (AF040048 and AF040049) reported by Wei et al. (1998), whereas exon 4 is a putative exon which is only found in the Kir3.2d sequence. Small and large capitals represent the nucleotide sequence of 5′-non-coding and partial coding regions, respectively. The start codon for methionine is underlined. B, alignment of amino acid sequences of Kir3.2 isoforms. The testicular Kir3.2 isoform designated as Kir3.2d is composed of 407 amino acids, lacks the N-terminal portion common to the other isoforms, and possesses the same C-terminal sequence as that of Kir3.2c. Amino acids are represented in single letter code. C, structural comparison of four Kir3.2 mRNAs. The Kir3.2 gene is predicted to be composed of 8 exons including several splicing sites. Alternative splicing seems to produce a series of Kir3.2 isoforms. ATG and TAG/TGA indicate the positions of individual start and stop codons, respectively. Each hatched box displays the open reading frame. M1, M2 and H5 represent the positions of the two transmembrane domains and the pore region located in the same exon (exon 5). D, RT-PCR analysis for the 5′-structure of Kir3.2 isoforms from mouse testis and forebrain cDNAs. The cDNAs from mouse testis (odd numbered lanes) and forebrain (even numbered lanes) were subjected to PCR with the reverse primer g2ex5-5′, which is common to all Kir3.2 isoforms, and the forward primers g2ex1, g2ex2, g2ex3 or g2ex4, specific for each of individual exons 1–4, respectively, as indicated in Fig. 2C. PCR products were resolved by agarose gel electrophoresis and detected by staining with ethidium bromide. No DNA fragments were amplified with the template without treatment with reverse transcriptase (RT-; lane 9 and 10). Numbers to the left of the gels indicate the size markers in base pairs.

Electrophysiological functional analysis of testicular Kir3.2

Functional analyses of testicular Kir3.2 subunits were performed in Xenopus oocyte and HEK 293T cell expression systems. In Xenopus oocytes, the open reading frame of isolated testicular Kir3.2 was amplified by PCR to tag Bam HI and Hin dIII sites at either the 5′- or 3′-end, respectively, and subcloned into the pGEMHE vector (Liman et al. 1992). Mouse testicular Kir3.2, as well as clones of porcine m₂R (kindly provided by Dr T. Kubo, Kyoto University; see Kubo et al. 1986) and mouse Kir3.1 (Inanobe et al. 1995b), were transcribed in vitro with an mRNA capping kit (Stratagene). A female Xenopus laevis was anaesthetized by immersing in 0·2% tricaine (Sigma) solution for 20 min. Xenopus oocytes were surgically removed, defolliculated in 1 mg ml⁻¹ collagenase solution (Wako Pure Chemical; Shih et al. 1998) and injected with various amounts of cRNAs as indicated in the figure legends. After injection, the oocytes were incubated in a modified Barth's solution at 18°C with daily solution changes. Electrophysiological studies were undertaken 48–72 h later.

Two-electrode voltage-clamp recordings of the oocytes were performed using a commercially available amplifier (CEZ-1250; Nihon Kohden) and glass microelectrodes which, when filled with 3 M KCl, had resistances of 0·5-1·5 MΩ. The bath solution contained 90 mM KCl, 3 mM MgCl₂, 0·15 mM niflumic acid and 5 mM Hepes-KOH (pH 7·4). Currents were obtained with the voltage steps (1·2 s duration, delivered every 3 s) from a holding potential of 0 mV to potentials between −120 and +60 mV in 20 mV increments.

Single-channel analyses of Kir3.0 channels co-expressed with m₂R in HEK 293T cells were performed in the cell-attached patch configulation using a patch-clamp amplifier (Axopatch 200A, Axon Instruments). All clones were subcloned into an expression vector pcDNA3 (InVitrogen) and transfected with the LipofectAMINE reagent. The pipette solution contained 140 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂ and 5 mM Hepes-KOH (pH 7·4) and 1 μM acetylcholine, and the bathing solution was composed of 140 mM KCl, 5 mM EGTA, 2 mM MgCl₂ and 5 mM Hepes-KOH (pH 7·4).

All electrophysiological data from oocytes and HEK 293T cells were stored on video tape using a PCM data recording system and subsequently replayed for computer analysis (Patch Analyst Pro; MT Corporation, Hyogo, Japan). For the single-channel analysis, the data were low pass-filtered at 1 kHz by an 8-pole Bessel filter (Frequency Devices), and sampled at 5 kHz. The data were expressed as means ±s.e.m. Experiments were performed at room temperature.

Immunohistochemistry

Balb/c mice were anaesthetized with sodium pentobarbital (60 mg kg⁻¹, i.p.) and perfused transcardially with 4% (w/v) paraformaldehyde in 0·1 M sodium phosphate buffer (pH 7·4). The testes were dissected, post-fixed in the same solution at 4°C for 12 h, and then dehydrated in 30% (w/w) sucrose and 0·05% (w/v) NaN₃ in PBS at 4°C. Decapsuled testes were cut into 10 μm sections, mounted on glass slides and stored at −80°C. After dehydration with an air dryer at room temperature for 2 h, testicular sections were incubated with 0·3% NaN₃ and 0·3% (w/v) H₂O₂ in PBS at room temperature for 30 min in order to deactivate endogenous peroxidase activity, and then with the base solution at 4°C for 16 h. The base solution contained 5% BSA, 5% (v/v) normal goat serum, 0·2% (w/v) Triton X-100 and 0·05% NaN₃ in PBS. Sections were exposed to the individual primary antibodies (aG2A-5 and aG1C-1) at 0·1–1 μg ml⁻¹ in the base solution at 4°C for 48 h. After the incubation, sections were washed 5 times for 1 h each in PBS containing 0·1% Triton X-100. The sections were further incubated with biotinylated goat anti-rabbit IgG from the Vectastain ABC kit (Vector Laboratories), and the immunoreactivity was developed with H₂O₂, diaminobenzidine (DAB) and nickel ammonium. The slices were dehydrated serially with ethanol and xylene, mounted with Entellan neu (Merck) and examined with a Zeiss Axioskop microscope. The images were captured on a CCD camera (HC-2500; Fujifilm), stored digitally in a Macintosh 8600/250 computer, processed and printed with Adobe Photoshop software. No immunoreactivity could be detected when the antibodies were preabsorbed with the individual antigenic peptides (data not shown). Determination of the developmental stage of the seminiferous tubules was according to the criteria described in Russell et al. (1990).

RESULTS

Immunological analyses of Kir3.2 and Kir3.1 in mouse testis

Because Kir3.2 mRNA was found to be expressed in mouse testis (Patil et al. 1995; Wei et al. 1998; Schwarz et al. 1998), we first analysed by immunoblotting the Kir3.2 protein in the membrane fraction obtained from the testis of 12-week-old mice (Fig. 1A). A single intense signal at ∼45 kDa was developed with the aG2C-3 antibody (open arrowhead, lane 1). The antibody detected a protein of ∼48 kDa in the membrane fraction of HEK 293T cells transfected with Kir3.2c (filled arrowhead, lane 4), but did not detect any significant protein bands in those cells transfected with either Kir3.2a (lane 2) or Kir3.2b (lane 3). This confirmed that aG2C-3 was specific to the C-terminus of Kir3.2c among the Kir3.2 isoforms so far reported (Inanobe et al. 1999) and suggested that the 45 kDa protein in testicular membrane might be Kir3.2c. The 45 kDa band was also developed with the aG2A-5 antibody. The aG2A-5 antibody did not recognize any significant bands other than the 45 kDa band on the SDS-PAGE. The aG2B-2 antibody, an antibody specific to the C-teminus of Kir3.2b, did not detect any significant bands in the testicular membrane preparation (not shown). It was, however, noted that the molecular mass of the band detected by aG2C-3 in the testicular membrane (lane 1) was significantly smaller than that of Kir3.2c. The 45 kDa protein was not found in the membrane fraction obtained from prepubertal testes of 2-week-old mice (lane 5) but appeared in those from 8-week-old mice (lane 6). Therefore, it is suggested that expression of the 45 kDa protein is related to spermatogenesis.

Figure 1. Immunological analyses of mouse testicular Kir3.2 and Kir3.1 subunits.

A, membrane proteins obtained from adult mouse testis (12 weeks old, 20 μg per lane; lane 1) and HEK 293T cells (5 μg per lane) transfected with the Kir3.2a, Kir3.2b or Kir3.2c isoform (lanes 2–4, respectively) were immunoblotted with the aG2C-3 antibody specific to the C-terminus of Kir3.2c. Testicular membrane proteins from 2- and 8-week-old mice were also immunoblotted with aG2C-3 (A, right panel) and aG1C-1 antibodies (C, left panel), as indicated below each panel. B and C, Kir3.2 (B) and Kir3.1 subunits (C, right panel) were immunoprecipitated with aG2C-3 and aG1C-1 antibodies, respectively, from the detergent-extracted membrane proteins (350 μg) obtained from mouse testis (odd numbered lanes) and neocortex (even numbered lanes). Immunoprecipitated proteins, which were labelled with biotin prior to extraction from membranes, were detected with various probes, as indicated above each panel. Open and filled arrowheads and asterisks indicate the migration positions of testicular Kir3.2, Kir3.2c and IgG heavy chain proteins, respectively. Testicular Kir3.1 (open arrows in C) migrates faster than that in the neocortex (filled arrows in B and C). Numbers on the left side of each panel present the molecular mass of standard markers in kilodaltons.

To further characterize the 45 kDa protein in the testicular membrane, we compared the immunoprecipitants obtained with aG2C-3 from the membrane fractions of testis and cerebral cortex (Fig. 1A). The immunoprecipitant from testicular membrane contained only a single band of the 45 kDa protein (lane 1). On the other hand, that from the mouse neocortex contained two major bands at ∼65 and ∼48 kDa (lane 2). The 65 kDa protein was recognized by aG1C-1 (lane 8) and the 48 kDa by aG2A-5 (lane 6). This is consistent with the notion that some of the K_G channels in rat brain are heteromultimers of Kir3.1 and Kir3.2c subunits (Inanobe et al. 1999). The immunoprecipitant from testis did not contain any significant protein bands other than the 45 kDa protein, which was also recognized by aG2A-5 (lane 5). In contrast to the data from the neocortex, Kir3.1 immunoreactivity was not detected at all in the immunoprecipitant produced with aG2C-3 in membrane from the testis (lane 7). Therefore, it seemed likely that the 45 kDa protein did not co-assemble with any other Kir3.0 subunit protein in testicular membrane. When we compared the immunoprecipitants obtained with aG2N-2, which should recognize the N-termini of all three isoforms of Kir3.2, the 45 kDa protein was not detected in testis (lane 3), but appeared in Kir3.2c isolated from the neocortex (lane 4). These results indicate that the testicular 45 kDa protein may not possess the epitope detected by aG2N-2.

Because it has been shown that Kir3.1 mRNA is expressed in rat testis (Dixon et al. 1995), we examined the properties of Kir3.1 protein in mouse testis using the aG1C-1 antibody (Fig. 1A). Immunoblotting of testicular membranes with aG1C-1 produced a band at ∼60 kDa in both 2-week-old (lane 1) and 8-week-old (lane 2) mice. SDS-PAGE analysis of the immunoprecipitant obtained from testicular membrane with aG1C-1 also showed only a single band at ∼60 kDa (lane 3), while that from neocortex was composed of two major bands at ∼65 and ∼48 kDa (lane 4), the latter of which was recognized by aG2C-3 (lane 6). These results suggest that Kir3.1 also does not co-assemble with any other Kir3.0 subunit in testis.

Molecular cloning of the testicular Kir3.2 isoform

The testicular 45 kDa protein seemed to be a novel isoform of Kir3.2c, which possesses the specific C-terminus of Kir3.2c but may lack the N-terminus detected by aG2N-2. Therefore, we tried to isolate the cDNA of the putative variant of Kir3.2 from a mouse testis cDNA library (Fig. 2). We screened 5 × 10⁵ clones with a Kir3.2-specific probe, and then obtained 10 full length cDNAs with insert sizes of about 2·3 kb. The isolated cDNA possessed a 5′-non-coding region, which was different from those of Kir3.2a, Kir3.2b and Kir3.2c (Fig. 2A). There was a stop codon at 162 bp upstream of the predicted start codon for methionine in the 5′-non-coding region. The open reading frame was composed of 407 amino acids with a molecular mass calculated to be ∼45 kDa. Alignment of the deduced amino acid sequence of the clone with those of the three Kir3.2 isoforms indicated that the testicular clone was 18 amino acids shorter at the N-terminal end. This was the only difference between the testicular clone and Kir3.2c (Fig. 2A). We designated it Kir3.2d. Kir3.2d possessed the C-terminus detectable by aG2C-3 and lacked the N-terminal epitope for aG2N-2. These characteristics strongly suggest that Kir3.2d may encode the 45 kDa protein detected by aG2C-3 in testicular membrane.

Wei et al. (1998) proposed that the Kir3.2 gene consists of seven exons. We needed to add one more putative exon for the Kir3.2 gene between the third and fourth exons proposed by them (Fig. 2A), because we found a 105 bp sequence in Kir3.2d cDNA (EXON 4 in Fig. 2A) which was not included in the previously proposed exons. We looked for the presence of the proposed exon 4 by analysing the 5′-ends of Kir3.2 mRNAs from mouse testis and forebrain using the RT-PCR technique (Fig. 2A). The PCR reactions were carried out using the reverse primer (g2ex5-5′) constructed from exon 5, which is common to all Kir3.2 isoforms, and the sense primers (g2ex1-g2ex4) specific for each of putative exons 1∼4 (Fig. 2A).

The PCR reactions using the g2ex5-5′ primer and each of the exon 1- or exon 2-specific primers (g2ex1 or g2ex2, respectively) produced a single DNA fragment from the cDNAs of forebrain (lane 2 and lane 4, respectively, in Fig. 2A), but not from those of testis (lane 1 and lane 3, respectively). Sequencing of both bands from forebrain (lane 2 and lane 4) has revealed that their nucleotide sequences are identical to those of the 5′-regions of Kir3.2a, Kir3.2b and Kir3.2c, i.e. exons 1, 2 and 5. The amplification with the exon 3-specific sense primer (g2ex3) and the g2ex5-5′ primer from testis cDNAs yielded two bands (lane 5). One was a 270 bp fragment which was also detected in the PCR product from forebrain cDNAs. The other was a 375 bp fragment (the larger band in lane 5), which was not produced from forebrain cDNAs. With the exon 4-specific sense primer (g2ex4) and g2ex5-5′, the 272 bp DNA fragment was amplified from cDNAs of testis (lane 7), but not from those of forebrain (lane 8). The nucleotide sequences of the DNA fragment in lane 7 and the larger 375 bp fragment in lane 5 were identical to the 5′-region of Kir3.2d (exons 3, 4 and 5). These results indicate that Kir3.2 transcripts containing exon 4 were expressed in mouse testis but not in forebrain. Thus, it seems reasonable to postulate that the Kir3.2 gene is composed of at least eight putative exons.

The nucleotide sequence of the smaller 270 bp DNA fragment in lane 5 indicated that the fragment is composed of exons 3 and 5 and does not contain exon 4. This nucleotide sequence is identical to that at the 5′-end of Kir3.2 mRNA in testis designated Girk2C by Wei et al. (1998). The same DNA fragment was also identified in the PCR product from forebrain cDNAs (lane 6). These results suggest that the 5′-region for Kir3.2 mRNAs in testis exhibits two distinct types; one is composed of exons 3, 4 and 5 (as reported in this study) and the other consists of exons 3 and 5 as previously reported by Wei et al. (1998). Because neither exon 3 nor 4 possess the starting codon, translation of both types of mRNAs may start in exon 5. The starting codon in exon 5 corresponds to the fourth methionine of the other Kir3.2 isoforms (Fig. 2A).

Alternative splicing at the 3′-end of the open reading frame of Kir3.2 mRNA generates the three isoforms Kir3.2a, Kir3.2b and Kir3.2c (Lesage et al. 1994, 1995; Tsaur et al. 1995; Isomoto et al. 1996). The splicing at the 3′-end in the Kir3.2d isoform is the same as that in Kir3.2c. We could detect four types of 3′-end sequences of Kir3.2 mRNAs in both forebrain and testis using the RT-PCR technique (data not shown). The putative exons constituting these isoforms were exons 5 and 7 for the Kir3.2c and Kir3.2d type, exons 5, 7 and 8 for Kir3.2a type, exon 5 for the Kir3.2b type, and exons 5 and 6 for the Girk2C type (not shown). This is consistent with the report by Wei et al. (1998). The Girk2C clone identified by Wei et al. (1998) is proposed to comprise exons 3, 5 and 6 in Fig. 2C. However, we could not isolate the cDNA encoding Girk2C from the testis cDNA library in this study. On the other hand, Wei et al. (1998) have failed to identify Kir3.2d mRNA in their study.

Functional analysis of Kir3.2d in heterologous expression systems

The functional properties of Kir3.2d were examined in the Xenopus oocyte expression system (Fig. 3). The oocytes injected with Kir3.2d cRNA alone expressed an inwardly rectifying K⁺ (I_K(IR)) current, which was inhibited by Ba²⁺ (1 mM) added to the bath (Fig. 3Aa). The amplitude of the Ba²⁺-sensitive I_K(IR) was 2·7 ± 1·8 μA (n = 9) at the end of the voltage pulse to −120 mV. When the m₂-muscarinic receptor was co-expressed with Kir3.2d cRNA, the I_K(IR) was enhanced by acetylcholine (ACh, 1 μM), and the enhancement was inhibited by atropine (10 μM) (Fig. 3Ab and C). The ACh-induced I_K(IR) was rapidly activated and slowly decreased during hyperpolarizing voltage pulses (Fig. 3A). At the end of the pulse to −120 mV, its amplitude was 1·2 ± 0·84 μA (n = 9). Because, in this series of experiments, the oocytes injected with Kir3.1 and m₂-receptor cRNAs expressed only a small amount of ACh-induced I_K(IR) (Fig. 3A), it appeared that Kir3.5/XIR mRNA was not expressed in these oocytes (Hedin et al. 1996). Thus, the I_K(IR) expressed in the oocytes may be derived from a homomeric assembly of Kir3.2d subunits.

Figure 3. Functional expression of Kir3.2d.

A, whole-cell currents recorded from Xenopus oocytes injected with cRNAs for Kir3.2d alone (a), Kir3.2d and the m₂-muscarinic receptor (m₂R; b), and Kir3.2d, Kir3.1 and m₂R (c). In two-electrode voltage clamp, K⁺ currents were measured with the application of acetylcholine (ACh), atropine and Ba²⁺ to the bath solution at final concentrations of 1, 10 μM and 1 mM, respectively, as indicated above each trace. A combination of 600 ms test pulses to +60 and −60 mV was applied every 3 s from a holding potential of 0 mV. In order to obtain the current-voltage relations, currents were elicited by test pulses for 1·2 s each from −120 to 60 mV in steps of 20 mV. The bath solution contained 90 mM K⁺ and arrowheads indicate the holding currents at 0 mV. B, gating properties of Kir3.2d in oocytes. The ACh-induced currents from oocytes injected with the cRNAs indicated with each trace were obtained by subtraction of the basal currents from the currents recorded in the presence of ACh in order to clarify each gating kinetic. Subtracted traces were recorded at membrane potentials stepped from −120 to 60 mV in 20 mV increments. Arrowheads to the left of the traces represent the zero current levels. Scale bar is 0·3 μA for Kir3.2d + m₂R, Kir3.1 + m₂R and dH₂O, and 1·0 μA for Kir3.2d + Kir3.1 + m₂R. C, the current-voltage relationships present the current amplitudes at the end of each pulse in the presence or absence of agonist and Ba²⁺ for the set of oocyte injections (means ±s.e.m., n = 9–11) shown in A. Oocytes were injected with cRNAs for Kir3.2d (≈1·7 ng), and Kir3.1 and m₂R (≈17 ng). Da, single-channel recordings in HEK 293T cells expressing Kir3.2d + m₂R (left) and Kir3.2d + Kir3.1 + m₂R (right). Channel currents were recorded at various potentials in the cell-attached patch-clamp configuration. The pipette and bath solutions were symmetrical with 140 mM KCl to make the reversal potential 0 mV and the pipette solution contained 1 μM acetylcholine. Arrowheads to the left of the traces represent the zero current levels. Db, current-voltage (i-V_m) relationships showing that the channel has a strong inward rectifying property. Data points are given as means ±s.e.m. (n = 3 patches).

In Fig. 3Ac, oocytes were injected with the cRNA derived from Kir3.1 in addition to those from Kir3.2d and the m₂-receptor. The oocytes expressed a background I_K(IR) of 14 ± 3·9 μA at −120 mV (n = 11). When ACh (1 μM) was added to the bath, a large I_K(IR) was induced (9·0 ± 2·5 μA at −120 mV, n = 11) (Fig. 3Ac and C). Atropine (10 μM) inhibited all of the ACh-induced and some of the background I_K(IR). Ba²⁺ (1 mM) completely inhibited both components (Fig. 3Ac and C). The ACh-induced I_K(IR) was activated slowly during hyperpolarizing voltage pulses (Fig. 3A), which is a feature of K_G channels containing the Kir3.1 subunit (Lesage et al. 1995; Kofuji et al. 1995; Duprat et al. 1995; Krapivinsky et al. 1995; Isomoto et al. 1996; Slesinger et al. 1996; Velimirovic et al. 1996). Therefore, Kir3.2d may be able to assemble with Kir3.1 to form a heteromultimeric K_G channel, when co-expressed in Xenopus oocytes.

We further examined the properties of Kir3.2d expressed in a mammalian cell line, HEK 293T cells. Figure 3A shows the representative single-channel recordings of Kir3.2d and Kir3.2d + Kir3.1 expressed with the m₂-receptor. The channel currents recorded in the cells expressing Kir3.2d alone either with (Fig. 3Da) or without m₂-receptor (data not shown) exhibited spiky short openings with a mean open time of ∼0·5–1 ms (data not shown). This property has also been reported in previous studies of homomeric channels composed of the other Kir3.2 isoforms (Lesage et al. 1995; Kofuji et al. 1995; Duprat et al. 1995; Tucker et al. 1996; Velimirovic et al. 1996). The current-voltage relationship of the homomeric Kir3.2d channel showed a strong inward rectification with a unitary conductance of 32 ± 0·14 pS (n = 3) at negative potentials (Fig. 3Db). When Kir3.2d was co-transfected with Kir3.1 and the m₂-receptor, the channel current exhibited the same conductance (34 ± 0·30 pS, n = 3, Fig. 3Db) but markedly longer openings than that in cells expressing Kir3.2d alone (Fig. 3Da). An increase in open time duration is also one of the features o heteromeric K_G channels containing the Kir3.1 subunit (Lesage et al. 1995; Kofuji et al. 1995; Duprat et al. 1995; Velimirovic et al. 1996). These results indicate that, in mammalian cells, Kir3.2d can form either a functional homomeric K_G channel when expressed alone or a heteromeric channel with Kir3.1.

Distribution of Kir3.2d in mouse testis

We examined the immunohistochemistry of Kir3.2d and Kir3.1 in mouse testis at the age of 12 weeks (Fig. 4). Intense immunoreactivity to the aG2A-5 antibody could be found scattered in the middle layer of many, but not all, seminiferous tubules (Fig. 4A). Immunoreactivity to aG2C-3 also showed the same distribution while that to aG2N-2 was not detected (data not shown). At a higher magnification, Kir3.2d immunoreactivity was found to be concentrated in a granular pattern in spermatids at various stages of development (arrowheads; Fig. 4B–H), but was not found either in the diploid germ cells of spermatogonia and spermatocytes or in the haploid cells of the late phase spermatids. Several immunoreactive vesicles were found in step 2 spermatids (B). The vesicles converged and came into contact with the nucleus in step 4 spermatids (C) and gradually elongated to cover the apex of nucleus in step 5–10 spermatids (D–G). Kir3.2d immunoreactivity could be detected only at the apex of the acrosome in step 12 spermatids and it disappeared after step 13 of spermatid development (data not shown).

Figure 4. Immunolocalization of Kir3.2d and Kir3.1 in mouse testis.

Sections of mouse testis (10 μm) were immunostained with the primary antibodies aG2A-5 for Kir3.2d (A–G) and aG1C-1 for Kir3.1 (I–K). Overviews of staining for Kir3.2d (A) and Kir3.1 (I) present the seminiferous tubules at stages V and VII, respectively. Kir3.2d is found as a granular pattern in many seminiferous tubules (A). B–G, Kir3.2d immunoreactivity (arrowheads) is observed in the spermatids at developmental steps 2, 4, 6, 8, 9 and 10. Note the change of the shape of Kir3.2d localization (areas stained brown) during spermatid development (H). Numbers in H indicate the steps of spermatid development. Kir3.1 immunoreactivity is found at the periphery of every tubule (I). J and K, particles of Kir3.1 immunoreactivity could be detected in the diploid germ cells of intermediate and type B spermatogonia (B) as well as leptotene (L) and pachytene spermatocyte (P). Note the attenuation of Kir3.1 immunoreactivity through the proliferative and differentiating phases (L). Scale bars: A and I, 50 μm; B–G and J–K, 5 μm.

On the other hand, Kir3.1 immunoreactivity was found in the peripheral region of the seminiferous tubules (Fig. 4A). The most intense signal was detected in the cytoplasm of intermediate and type B spermatogonia (B in Fig. 4A). The immunoreactivity seemed to be attenuated during the progress of the proliferative and differentiating phases. Only a little immunoreactivity could be detected in the late pachytene spermatocytes (P in Fig. 4A). Thus, the immunoreactivities of Kir3.2d and Kir3.1 were distributed in the germ cells at distinct developmental stages (Fig. 4A and L). This observation is consistent with the immunoprecipitation study showing that Kir3.2d does not co-assemble with Kir3.1 and vice versa (see Fig. 1A and D).

DISCUSSION

In the present study, a novel splicing variant of Kir3.2, Kir3.2d, was isolated from a mouse testis cDNA library (Fig. 2). The deduced amino acid sequence of Kir3.2d is 18 amino acids shorter at the N-terminal end than that of Kir3.2c, which is the only difference between the two isoforms. This feature of the primary amino acid sequence of Kir3.2d explains the biochemical properties of the testicular Kir3.2 isoform in terms of its immunoreactivity to three region-specific anti-Kir3.2 antibodies, aG2C-3, aG2A-5 and aG2N-2, and the migration profile in SDS-PAGE (Fig. 1).

Although Kir3.2d lacked the N-terminal end common to the other isoforms of Kir3.2, it could form functional K_G channels either alone or with Kir3.1 (Fig. 3). We previously showed that oocytes injected with the cRNAs of Kir3.2c and the m₂-muscarinic receptor could not express any K⁺ currents, although the Kir3.2c protein seemed to be generated and transported to the cell membrane (Inanobe et al. 1999). Kir3.2c could, however, form functional K_G channels with Kir3.2a or Kir3.1. On the other hand, Kir3.2a expressed K_G channel activity by itself. Because Kir3.2a is 11 amino acids shorter at its C-terminal tail than Kir3.2c, and this is the only difference between the two isoforms (Fig. 2A), it was postulated that the 11 amino acids at the C-terminal tail of Kir3.2c prevent the homomeric Kir3.2c channel from being functional (Inanobe et al. 1999). Although Kir3.2d possesses the 11 amino acids which are found at the C-terminal end of Kir3.2c, it could form a functional K_G channel on its own. It may be the case therefore that both the N- and C-terminal ends of Kir3.2 isoforms play complex roles in the regulation of channel activity in K_G channels composed of homomeric assemblies of Kir3.2 isoforms. Further studies are needed to clarify how the N- and C-terminal ends interact to control the Kir3.2 channel activity.

Based on an examination of mouse brain mRNA using RT-PCR and an investigation, using PCR, of the intron/exon boundaries in a mouse genomic cDNA library using the primers for Kir3.2 isoforms, Wei et al. (1998) proposed that the Kir3.2 gene is composed of seven exons. In order to explain the 5′-non-coding region of Kir3.2d cDNA, we needed to add one more putative exon between the third and fourth exons. The existence of this putative exon 4 was confirmed with RT-PCR analysis of testicular mRNAs (Fig. 2A). Thus, the Kir3.2 gene appears to be composed of at least eight exons (Fig. 2A).

It seems likely that these eight putative exons are selected by diverse alternative splicing mechanisms and produce distinct Kir3.2 mRNAs (Lesage et al. 1994, 1995; Tsaur et al. 1995; Isomoto et al. 1996; Wei et al. 1998). Individual Kir3.2 isoforms seem to be expressed in a tissue-specific manner; Kir3.2a was reported to be specific to brain (Lesage et al. 1994), Kir3.2c was identified in brain and pancreas (Lesage et al. 1995; Tsaur et al. 1995), Kir3.2b was distributed ubiquitously (Isomoto et al. 1996), and Kir3.2d is expressed in testis (this study). The Kir3.2 isoforms can all form functional K_G channels heteromerically with Kir3.1, although we have found that the K_G channels in dopaminergic neurons of rat substantia nigra are composed of different Kir3.2 isoforms and do not contain Kir3.1 (Inanobe et al. 1999), and those in mouse pancreatic α cells are heteromers of Kir3.2c and Kir3.4 (Yoshimoto et al. 1999). Therefore, native K_G channels with a variety of properties can be generated according to the differential expression of Kir3.2 isoforms with the other Kir3.0 subunits, Kir3.1, Kir3.3 and Kir3.4, in individual cells. Such diversity should be important for the mediation by K_G channels of signalling by many different inhibitory neurotransmitter receptors, such as m₂-muscarinic, A₁-purinergic, D₂-dopaminergic, 5-HT₁-serotonergic, opioid, somatostatin and GABA_B receptors, in a variety of cells including neurons, endocrine cells and cardiac myocytes (North 1989; Hille 1992; Jan & Jan 1994; Wickman & Clapham 1995; Yamada et al. 1998). Further studies are needed to clarify the mechanism controlling splicing of Kir3.2 mRNAs in different tissues.

Using RT-PCR analysis from total testis RNAs, we could detect two different isoforms at the 5′-end of Kir3.2 mRNA and four at its 3′-end (Fig. 2A). It may be a feature of testis that the putative exons 3 and 4, but not exons 1 or 2, are selected to produce the isoforms of Kir3.2 mRNA, because we could not detect any PCR products from testis using g2ex1 or g2ex2 for sense primers (Fig. 2A and D). Because the isoforms at the 5′-end in testis start from the same ATG codon in exon 5, theoretically four kinds of Kir3.2 isoform proteins might be generated in mouse testis. However, we could detect only a single band of protein with aG2C-3 and aG2A-5 in testicular membrane. No significant protein band was detected by aG2B-2 (data not shown). Therefore, the isoform proteins possessing the C-termini of the Kir3.2a and Kir3.2b subunit types may not exist in significant amounts in testis. We could not examine the expression level of the isoform containing exon 6 (Girk2C in Wei et al. 1998) because we could not isolate the cDNA encoding Girk2C from the mouse testis cDNA library in this study. However, because all of ten cDNAs isolated from the library were Kir3.2d, Kir3.2d might be the predominant isoform of Kir3.2 in testis. Further studies are needed, however, to reach a final conclusion on the expression of the various Kir3.2 isoform proteins in testis.

Although Wei et al. (1998) failed to identify Kir3.2d in their analyses of testis mRNAs, they showed that the Kir3.2 mRNA detected by a probe for the 3′-non-coding region containing exon 7 (Kir3.2c and Kir3.2d type) is expressed in a subset of seminiferous tubules not only in wild-type mice but also in weaver mutants (Wei et al. 1998; Schwartz et al. 1998). On the other hand, they also showed that the mRNAs detected by the probes for the 3′-non-coding regions in exons 5, 6 or 8 (Kir3.2a, Kir3.2b and Girk2C types) distributed diffusely in testis. Thus, the Kir3.2 isoform containing exon 7 seems to be important for spermatogenesis. Exon 7 is included in both Kir3.2c and Kir3.2d (Fig. 2A), and Kir3.2c is not expressed in testis (Fig. 2A). Therefore, the isoform detected by the probe for exon 7 in Wei et al. (1998) may be the Kir3.2d isoform found in this study. Consistent with this, we found that Kir3.2d immunoreactivity locates only at the acrosomes of step 2–12 spermatids in the seminiferous tubules. Furthermore, in weaver mutant mice, it has been reported that development of spermatids is specifically impaired (Harrison & Roffler-Tarlov 1994), again in agreement with the distribution of Kir3.2d immunoreactivity. Thus, it is suggested that at the 3′-end, Kir3.2d, but not other Kir3.2 isoforms, may be responsible for the damage to spermatogenesis in weaver mouse testis. Therefore, it may be the case that Kir3.2d may play an important functional role in physiological spermatogenesis. Because Kir3.2-null mice are free from neurodegeneration and male infertility (Signorini et al. 1997), a gain-of-function weaver mutation in Kir3.2 may be responsible for the cellular damage in brain and testis. Further studies are needed to clarify the role of Kir3.2d at the acrosome not only in the normal development of spermatids but also in the degeneration of spermatids in weaver mutant mice.

In contrast to the localization of Kir3.2d in the haploid cells, Kir3.1 locates in the cytoplasm of the diploid germ cells. It is known that the homomeric K_G channel composed of Kir3.1 subunits is not functional (Krapivinsky et al. 1995; Hedin et al. 1996), but that heteromers of Kir3.1 and another Kir3.0 subunit form functional K_G channels. However, we failed to detect any Kir3.0 subunits with testicular Kir3.1 in this study. The cellular function of testicular Kir3.1 in the proliferative and differentiating phases of the germ cells thus also needs to be clarified.