Molecular Cloning and Functional Expression of the Mouse Gap Junction Gene Connexin-57 in Human HeLa Cells*

A new mouse connexin gene has been isolated that codes for a connexin protein of 505 amino acid residues. Based on the predicted molecular mass of 57.115 kDa, it has been designated connexin-57. Similar to most other mouse connexin genes, the coding region of connexin-57 is not interrupted by introns and exists in the mouse genome as a single-copy gene. Within the connexin family, this new gene shows highest sequence identity to porcine connexin-60 in the α group of connexins. The connexin-57 gene was mapped to a position on mouse chromosome 4, 30 centimorgans proximal to a cluster of previously mapped connexin genes. Low levels of connexin-57 mRNA were detected in skin, heart, kidney, testis, ovary, intestine, and in the mouse embryo after 8 days post coitum, but expression was not detected in brain, sciatic nerve or liver. In order to analyze gene function, the connexin-57 coding region was expressed by transfection in human HeLa cells, where it restored homotypic intercellular transfer of microinjected neurobiotin. Heterotypic transfer was observed between HeLa connexin-57 transfectants and HeLa cells, expressing murine connexin-43, -37, or -30.3. Double whole-cell voltage clamp analyses revealed that HeLa-connexin-57 transfectants expressed about 10 times more channels than parental HeLa cells. Voltage gating by transjunctional and transmembrane voltages as well as unitary conductance (∼27 picosiemens) were different from intrinsic connexin channels in parental HeLa cells.

Gap junctions in vertebrates are formed by subunit proteins of the connexin gene family. Six connexin proteins oligomerize and form a hemichannel (connexon). Between apposed membranes of adjacent cells, connexons interact with the opposite connexons to form gap junction channels (cf. Refs. 1 and 2). Functional gap junction channels in vertebrates are permeable to small molecules (Ͻ1 kDa), i.e. ions, metabolites, and second messenger molecules. Gap junctional communication has been suggested to contribute to metabolic cooperation, synchronization of cellular physiological activities, growth control, and regulation of development.
To date, 14 different connexin genes in the mouse and rat genome have been described (1,3,4). Many of these genes have been located on different chromosomes (5), but some of them have been assigned to the same chromosome (cf. Refs. 6 and 7). Individual members of the connexin (Cx) 1 family are designated according to the theoretical molecular mass of the protein (8) or according to a Greek nomenclature that is based on differences in the cytoplasmic loop (2,4,9). The similar membrane-spanning topology of connexin proteins has been revealed by limited proteolysis and site-directed antibodies to Cx32, Cx43, and Cx26 (10 -12). By these methods, together with hydropathy plots, we deduced that connexin proteins span the plasma membrane four times (M1-M4), with two extracellular loops (E1, E2) and three cytoplasmic regions (amino-and carboxyl-terminal regions, cytoplasmic loop). Within the connexin gene family, the four membrane-spanning regions, the amino-terminal region, and the first extracellular loop show the highest sequence identities. Major differences in sequence and length were found in the carboxyl-terminal region and the cytoplasmic loop. The transmembrane regions of a connexon are assumed to be involved in forming the channel pore. Furthermore, the two extracellular loops mediate docking of two connexin hemichannels, and the cytoplasmic domains of the connexins are thought to regulate channel voltage and chemical gating (13)(14)(15)(16). Most mammalian cells express one or more connexin genes in a cell type-specific manner. For functional characterization, the various cloned connexins have been expressed in Xenopus oocytes and/or cultured mammalian cells (cf. Ref. 17). These reconstitution experiments have shown that different connexin channels exhibit different permeabilities to tracer molecules (18) and show different unitary conductances (19). Reconstitution experiments with hemichannels composed of different connexins, leading to heterotypic gap junction channels, made it possible to distinguish compatible (functional) and incompatible (non-functional) combinations of connexons (18,19). Evidence for the existence of functional heteromeric channels (different connexin proteins in the same hemichannel) has been reported in chicken lens and for the liver-derived connexin-26 and -32 (20,21).
Several of the mouse connexin genes have now been deleted by homologous recombination (for reviews, see Refs. 1, 23, and 24). So far, the phenotypic abnormalities observed in these connexin-deficient mice were different from each other, although in most cases residual intercellular communication in * This work was supported by Deutsche Forschungsgemeinschaft Grant SFB 284 (Project C1) and a grant from the Fonds der Chemischen Industrie (to K. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s)AJ010741.
** To whom correspondence should be addressed: Inst. fü r Genetik, Universitä t Bonn, Römerstr. 164, 53117 Bonn, Germany. Tel.: 49-228-734210; Fax: 49-228-734263; E-mail: genetik@uni-bonn.de. the connexin-deficient cells could be demonstrated, probably due to the expression of other functional connexin channels. Thus, a more complete understanding of gap junction function in mammalian tissues requires the identification and characterization of the remaining connexin genes. Therefore, we searched for novel connexin genes in a mouse genomic library and found a new reading frame, which coded for a previously unknown connexin protein of 57.1 kDa molecular mass and, thus, was called mouse connexin-57. Here we describe the relationship of this gene to other connexins, its expression pattern in embryo and adult tissues, the chromosomal location of this gene, and its functional activity following transfection into human HeLa cells.

MATERIALS AND METHODS
Isolation of Genomic Mouse Cx57 DNA-We isolated 25 connexin homologous recombinant FixII phage clones by screening a 129/SvJ mouse genomic library (Stratagene, La Jolla, CA) with a mouse Cx26 probe (25). We used the procedure described by Hennemann et al. (26), to identify connexin genes in recombinant phages. Several of these phage clones subsequently failed to hybridize to any of the known connexin genes under stringent conditions of hybridization (50% formamide, 5ϫ SSC at 42°C). DNA of these latter phages was isolated using standard protocols (27). Southern blot hybridization of restricted phage DNA was performed under low stringency conditions (40% formamide, 5ϫ SSC at 38°C) using the mouse Cx26 probe. It was concluded that a 2.2-kb XbaI fragment contained sequences homologous to the mouse Cx26 probe. These fragments were subcloned in the pBluescriptII SKϩ vector (Stratagene, La Jolla, CA). Sequencing was performed on both strands by the modified chain termination method (28), using either vector-derived primers or appropriate primers derived from previous sequencing results. The amino acid sequence deduced from the longest open reading frame (i.e. mouse Cx57) was aligned with different connexin sequences, using the PCGene sequence analysis program (PC-Gene 6.8, IntelliGenetics, Mountain View, CA) and the HUSAR program (German Cancer Research Center, Heidelberg, Germany).
Southern and Northern Blot Analyses-Genomic DNA from livers of BALB/c mice was prepared according to a standard procedure (27). Restriction endonuclease-digested DNA (10 g) was electrophoresed in 0.7% agarose and blotted by alkaline transfer onto Hybond Nϩ membrane, following the manufacturer's directions (Amersham Pharmacia Biotech, Braunschweig, Germany). High stringency hybridization of the Southern blot was carried out overnight using a 531-bp BstXI Cx57 fragment (representing nucleotides 59 -590 in Fig. 2). The fragment was labeled with [␣-32 P]dCTP by random priming (Megaprime Labeling Kit, Amersham Pharmacia Biotech) to a specific activity of 0.2-1 ϫ 10 9 cpm/g of DNA. Filters were washed at high stringency (0.2ϫ SSC, 0.1% SDS at 60°C) and exposed to XAR-5 film for 1 day to 5 weeks.
Total RNA from mouse tissues was isolated with the TRIzol reagent according to the manufacturer's procedure (Life Technologies, Inc., Eggenstein, Germany). Total RNA from HeLa cells was prepared with the Qiagen RNeasy kit, as described by the company (Qiagen, Hilden, Germany).
Radioactive Northern blot analysis with total RNA (20 g) was carried out as described previously (6) with the mouse Cx57 XhoI/XbaI (1318 bp, positions 412-1745) fragment as probe. Non-radioactive Northern blot analysis with total RNA (10 g) was carried out with the RNA-labeling kit, as described by Roche Molecular Biochemicals (Mannheim, Germany). A mouse Cx57 DIG (digoxygenin)-labeled antisense cRNA as hybridization probe was generated by cloning the mouse Cx57 XhoI/ClaI DNA fragment (1023 bp, positions 412-1435) into XhoI/ ClaI-linearized pBluescript II SKϩ DNA. Plasmid DNA was linearized with the restriction enzyme XhoI, and the cRNA DIG-labeled antisense probe was prepared by in vitro transcription using the T3 RNA polymerase in the presence of digoxygenin-UTP. The amount of total RNA in different samples on Northern blots was standardized by hybridization to a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe (29). The hybridization signals were quantified by densitometric evaluation using Scan Pack version 14.1A27 (Biometra, Göttingen, Germany) RT-PCR Analysis-Reverse transcription (RT) and amplification of cDNA by polymerase chain reaction (PCR) were based with slight modifications on the Access RT-PCR system (Promega, Madison, WI). In order to avoid genomic DNA contamination in total RNA preparations, they were treated with RQ-DNase (Promega). Aliquots of 2 g of total RNA were incubated with 2 l of AMV buffer (5ϫ, Promega), 1 l of 25 mM MgSO 4 , 1 l of RQ-DNase (1 unit/l, Promega), 1 l of RNasin (40 units/l, Promega, RNase inhibitor) and diethyl pyrocarbonate (DEPC)-treated H 2 O to a total volume of 10 l and incubated at 37°C for 30 min. Then, RQ-DNase was inactivated at 75°C for 5 min. The RT reactions were performed with 10 l of DNA-free RNA solution by adding 3 l of AMV buffer (5ϫ, Promega), 2.5 l dNTP mix (10 mM each dATP, dCTP, dGTP, dTTP), 1 l of RNasin, 3 l of AMV-reverse transcriptase (10 units/l, Promega), 1 l of oligo(dT) 18  . The PCR products were analyzed by Southern blot hybridization to a mouse Cx57 probe (XhoI/XbaI fragment, see above). Using the mouse Cx57 RT-PCR protocol described above, we checked the RT reaction mixtures for genomic DNA contaminations by RT-PCR with ␤-actin primers. The mouse ␤-actin primers were (5Ј-primer) 5Ј-CGT GGG CCG CCC TAG GCA ACC-3Ј, and (3Ј-primer) 5Ј-TTG GCC TTA GGG TTC AGG GGG-3Ј (30). These primers led to amplification of a 243-bp segment of the cDNA and a 330-bp fragment of the genomic sequence.
Transfection-For transfection of HeLa cells, an NcoI/XbaI mouse Cx57 fragment (positions 1-1745) was cloned into the SmaI site of pBluescriptII SKϩ (Stratagene). From this plasmid, a KpnI/BamHI fragment containing the coding region of mouse Cx57 was cloned into the KpnI/BamHI site of the transfection vector pBEHpac18 (37), which contained the SV40 early promoter, a polyadenylation signal, and a gene that conferred resistance to puromycin.
HeLa cells were transfected with 20 g of DNA of the mouse Cx57 coding region in pBEHpac18, using the calcium phosphate transfection protocol of Chen and Okayama (38). Forty-eight hours after incubation with the DNA/calcium phosphate precipitate, 1 g/ml puromycin was added to the medium. Clones were picked after 3 weeks and grown under selective conditions. The clones were checked by Northern blot analysis.
Microinjection of Tracers-Glass micropipettes were pulled from capillary glass (World Precision Instruments Inc., Berlin, Germany) with a horizontal pipette puller (PD-5, Narashige, Tokyo, Japan) and backfilled with tracer solution. Tracers were injected iontophoretically (Iontophoresis Programmer model 160; World Precision Instruments Inc.). Dye transfer was examined, using an inverse microscope (IM35; Zeiss, Oberkochen, Germany) with fluorescent illumination (HBO100; Zeiss). During injection, cell culture dishes were kept on a heated block at 37°C.
Lucifer yellow CH (Molecular Probes, Eugene, OR) as 4% (w/v) in 1 M LiCl was injected by applying negative voltage for 10 s (I ϭ 20 nA). Cell-to-cell transfer was evaluated by fluorescent microscopy (Zeiss IM-35, filter set 9) 5-30 min after dye injection. Neurobiotin (N-2(2aminoethyl)-biotinamide hydrochloride; Vector Lab, Burlingame, CA) and rhodamine 3-isothiocyanate dextran 10S (Sigma) at concentrations of 6% and 0.4% (w/v) in 0.1 M Tris-Cl (pH 7.6) were iontophoretically injected by application of positive voltage for 10 s (I ϭ 20 nA). The transfer of tracer molecules was observed using filter set 15 (Zeiss) in the microscope. Five to thirty min after injection, cells were washed twice with phosphate-buffered saline (PBS), fixed for 10 min in 1% glutaraldehyde in PBS, washed twice with PBS, incubated in 2% Triton X-100/PBS for 2 h, washed three times with PBS, incubated with horseradish peroxidase-avidin D diluted 1:1000 in PBS (Vector Lab), for 90 min, washed three times with PBS, and incubated in 0.05% diaminobenzidine (Sigma), 0.003% hydrogen peroxide solution for 30 s to 2 min. The staining reaction was stopped by washing three times with PBS. The cell-to-cell transfer was quantified by counting the number of stained neighboring cells around the microinjected cell.
For assay of heterotypic coupling, one cell type was stained with DiI as described by Goldberg et al. (39) and co-cultivated with a 1000-fold excess of unstained cells expressing a different connexin gene. The cells were incubated 18 h before microinjection of neurobiotin tracer.
The dual voltage-clamp method was used to control the membrane potential of both cells individually and to measure the associated membrane and junctional currents (41,42). After the establishment of whole-cell patch-clamp conditions, voltage of the same amplitude was clamped in both cells (V 1 ϭ V 2 ), and non-junctional membrane currents (I m1 , I m2 ) were recorded individually for each cell. Changing the membrane potential in one cell induced transjunctional potential (V j ϭ V 2 Ϫ V 1 ) and associated junctional current, I j , equal in amplitude but opposite in polarity for both cells. The junctional conductance is determined as g j ϭ I j /(V 2 Ϫ V 1 ). Signals were recorded in parallel on videotape (Digital Data Recorder VR-100, Instrutech Corp.) and computer (A/D converter TL-1, Axon Instruments). Data acquisition and analysis were performed using Pclamp software (Axon Instruments) and Sigma Plot (Jandel Corp.).

Cloning of the Mouse Cx57
Gene-After screening a mouse genomic library of -FixII phages (Stratagene) with a mouse Cx26 probe using low stringency hybridization conditions, 3 of 25 independently isolated positive recombinant phage clones did not hybridize under stringent conditions to any of the known rodent connexin genes. Restriction and Southern blot analyses suggested that all three phage clones contained the same novel, potentially connexin-related gene. A 2.1-kb XbaI DNA subfragment of the recombinant phage clone, which hybridized to the Cx26 probe under low stringency conditions, was ligated in XbaI-linearized pBluesript II SKϩ, as indicated in Fig. 1. This XbaI fragment contained the complete reading frame of a new mouse connexin (Fig. 2). The open reading frame coded for a protein of 505 amino acids with a theoretical molecular mass of 57.115 Da. Following the nomenclature proposed by Beyer et al. (8), we designated this new connexin as mouse Cx57.
Analysis of Mouse Cx57 Amino Acid Sequence-The deduced amino acid sequence of Cx57 showed all the typical features of a connexin protein, i.e. four potential transmembrane regions (the third one has amphipathic features) predicted by the algorithm of Kyte and Doolittle (43), and could be aligned to topological domains of Cx32 and Cx26 that were previously deduced from studies with site-specific antibodies and limited proteolysis of the membrane embedded proteins (12,18). The pattern of cysteine residues in the two putative extracellular loops of the Cx57 protein included three cysteine residues in the sequence CX 6 CX 3 C and CX 4 CX 5 C, like in many other murine connexins.
The comparison of the overall amino acid identities between Cx57 and all other known mouse or rat connexins showed that Cx57 shared no significant relationship to any of these connexins (48 -32% identity range), but exhibited relatively high amino acid identity (i.e. 74%) to porcine Cx60 (44). By comparison of the amino acid sequence of the putative extracellular region 1, Cx57 was classified into the ␣-group of connexins (9). Table I lists the amino acid identities, according to predicted topological domains, between mouse Cx57 and porcine Cx60 (44), in comparison to mouse connexins Cx43, Cx40 (36), Cx37 (26), and Cx26 (6). Cx57 showed greater similarity to porcine Cx60 than to any other murine connexin protein sequence. A phylogenetic tree of all known murine connexins and porcine Cx60 supports this conclusion (Fig. 3).
Genomic Organization and Chromosomal Localization-Southern blot analysis of mouse genomic DNA showed that the Cx57 gene is present in the mouse genome as a single-copy gene (Fig. 4A). Under these conditions, single DNA fragments of 1.4, 1.1, and 1.0 kb were detected after digestion with HindIII, XhoI, and EcoRI. DNAs of the parental mice after two sets of genetic crosses were typed for restriction enzyme polymorphisms in the Cx57 sequence. For the M. spretus crosses, PstI produced fragments of 12.9 kb in DNA from parental NFS/N and C57/J mice and 9.7 kb in M. spretus, whereas no polymorphisms were detected in the M. musculus ϫ M. musculus crosses. Inheritance of the variant fragment in the M. spretus crosses placed this gene on the proximal chromosome 4 (Fig. 4B). We suggest the genetic symbol Gja-9 for designation of the mouse Cx57 gene, in accordance with the nomenclature used for connexin genes (cf. Refs. 45 and 46).
Expression of Mouse Cx57 mRNA in Human HeLa Cells-The expression pattern of Cx57 mRNA in total RNA from different mouse tissues was analyzed by Northern blot hybridization (Fig. 5, A and B) and by RT-PCR (Fig. 5, C-E). Cx57 mRNA was found in 10 dpc mouse embryos and in heart, intestine, testis, kidney, lung, and skin (adult and embryo), as well as in the keratinocyte-derived cell lines Hel30 and Hel37. The transcript signals on the autoradiographs were detected only after 5 weeks of exposure, suggesting low levels of Cx57 mRNA in these tissues and cells. For further examination we used the more sensitive RT-PCR technique and found, in addition, that Cx57 mRNA was also expressed in ovary and mouse embryos between 8.5 and 18.5 dpc (Fig. 5, C and D).
Functional Expression of Mouse Cx57 in Human HeLa Cells-For functional characterization of the Cx57 channel, we transfected human HeLa cells deficient in gap junctional communication (47) with Cx57 DNA (see "Materials and Methods"). The isolated stable Cx57 transfected HeLa cells were characterized by Northern blot analysis using total RNA and a Cx57 probe. Fig. 6 shows results of the nonradioactive Northern blot hybridization and illustrates that several of the HeLa Cx57 transfectants expressed high levels of Cx57 mRNA, in contrast to wild type HeLa cells.
Tracer Transfer through Homotypic and Heterotypic Cx57 Gap Junction Channels-Permeability of homotypic and het- erotypic junctions was examined by using neurobiotin (M r 287, net charge ϩ1) and Lucifer yellow (M r 448, net charge Ϫ2), as described under "Materials and Methods." No cell-cell transfer for any of these dye molecules was observed in parental HeLa cells. Homotypic transfer of microinjected neurobiotin between HeLa-Cx57 cells was detected in first order neighboring cells after 10 min, and in second order cells 30 min after microinjection (Fig. 7). No cell-cell transfer of Lucifer yellow was observed. Intercellular transfer of calcein was weak and visible only after 9 h of incubation. Among 13 heterotypic cell cocultures, tested for dye transfer, only 4 showed transfer of neurobiotin (see Table II). None of these combinations yielded transfer of microinjected Lucifer yellow or calcein. Neurobiotin transfer was evident in Cx57-Cx30.3, Cx57-Cx37, and Cx57-Cx43 heterotypic junctions.
V j and V m -sensitive Gating-Experiments were performed on 16 spontaneously preformed cell pairs by using the doublevoltage clamp method. Initially, the voltage in both cells was clamped to the same holding potential (0 Ϭ Ϫ60 mV). Junctional current, I j , was measured during periodic positive and/or negative pulses of transjunctional voltage, V j . All cell pairs tested were electrically coupled. In three cell pairs cytoplasmic bridges were identified (see below). In cell pairs connected by gap junctions, coupling conductance varied between 0.2 and 2.5 nS, with a mean value of ϳ1.5 nS (n ϭ 13). HeLa parental cell pairs typically demonstrated uncoupling or very weak coupling through only a few channels (47).
Dependence of junctional conductance on V j was measured by depolarizing or hyperpolarizing one cell long enough to establish steady state of g j , g j (ss). Fig. 8A illustrates an example of junctional current record and voltage protocol in cell 1, V 1 , and cell 2, V 2 . During depolarization of cell 1 with a voltage step of 50 mV, g j decreased about 10 times, exposing g min , and slowly recovered after hyperpolarization to the holding potential. Summarized data of g j (ss)-V j dependence are presented in Fig. 8B. Individual g j (ss) data were normalized to instantaneous g j , g j (ss)/g j (inst). Continuous lines show fitting of the experimental data to the Boltzmann equation, V o corresponds to V j with a halfmaximum of g j . Parameter A characterizes steepness of g j (ss) decay, and g min shows residual values of g j for negative and positive V j polarities. Fitting parameters were as follows: V oϩ ϭ 25 Ϯ 1 mV (mean Ϯ S.E., V o-ϭ Ϫ25 Ϯ 2 mV, A ϩ ϭ 0.2 Ϯ 0.3, A Ϫ ϭ 0.16 Ϯ 0.03, g minϩ ϭ 0.14 Ϯ 0.02, and g minϪ ϭ 0.20 Ϯ 0.03, where subscript ϩ or Ϫ corresponds to depolarization or hyperpolarization of cell 1, respectively. V o is close to that measured in Cx 31.1 (ϳ20 mV), Cx37 (ϳ17 mV), and Cx45 (ϳ20 mV) but significantly smaller than in Cx 26 (ϳ85 mV), Cx30.3 (ϳ60 mV), Cx40 (ϳ55 mV), Cx43 (ϳ45 mV), and 46 (ϳ45 mV). Numbers in parentheses refer to our published (42) and unpublished data for different HeLa transfectants. The difference among g minϩ and g minϪ indicates that V j gating is slightly asymmetric and more sensitive at positive V j values.
Surprisingly, we found that Cx57 exhibited g j dependence on the membrane potential, V m . In all four cell pairs we tested for g j Ϫ V m dependence, g j rose in response to simultaneous depolarization of both cells (see Fig. 8C). When both cells were depolarized from Ϫ20 mV to ϩ60 mV and then to ϩ70 mV, g j increased approximately 1.5 times. During both depolarization periods, g j rose with slow time constant (in tens of seconds) and slowly recovered during hyperpolarization to the holding potential.
Single Gap Junction Channel Conductance-Measurements were performed either by applying rectangular pulses or ramps with amplitude up to 120 mV to one of the cells. Fig. 9A illustrates an example of single-channel I j record during voltage steps in cell 2, V 2 , from the holding potential of Ϫ10 mV (see dotted line) to Ϫ130, Ϫ110, and Ϫ80 mV. All I j Ϫ V j dependence of the single channel was tested by using a ramp protocol (see Fig. 9B). Both cells were at the same holding potential (dotted line). Periodic voltage pulses of Ϯ20 mV and ramps of Ϯ80 mV were applied to cell 2. During the first bi-phasic voltage step and ramp the channel was open, exhibiting ϳ27 pS conductance. I j ϪVj dependence was almost linear when V 2 rose from Ϫ80 mV to ϩ80 mV. During the second positive pre-pulse, the channel was closed, presumably, to the residual state (see arrow). I j record during the second ramp shows that the channel was closed to the substate with conductance of ϳ4 pS (see dashed line). The frequency histogram in Fig. 9C summarizes single-channel conductance data measured in six experiments. Mean value of single-channel conductance was 27.4 Ϯ 0.3 pS (n ϭ 143). ␥(residual) was in the range of ϳ4 -6 pS.
Based on single-channel records, we estimated the number of gap junctions in HeLa-Cx57 transfectants. Cell pairs, exhibiting coupling conductance of ϳ1.5 nS, contained approximately 60 functional channels. This number is significantly higher than the number of functional channels between HeLa parental cell pairs, which varied between 0 and 10, but in most cases cell pairs were uncoupled. 2 Additionally, single-channel conductance in HeLa-57 transfectants was lower than ␥(open) measured in HeLa parental cells (about 40 pS). 2 Effect of Chemical Factors on g j -In 6 cell pairs from 16, we have tested the effect of heptanol (2 mM) on g j . Fig. 10 demonstrates typical uncoupling effect of heptanol in one of three cell pairs from three coupled via gap junctions. Three cell pairs with g j ϭ 15, 20, and 25 nS exhibited no V j -sensitive gating and heptanol had no significant effect on g j. These data indicate the presence in these cell pairs of cytoplasmic bridges (48). Arachidonic acid (10 Ϫ5 M) tested in one cell pair with g j ϭ 1 nS produced full uncoupling. During the washout period of 30 min, we observed very slow coupling recovery with random periods of single-channel activity of ϳ27 pS conductance (data not shown). The CO 2 effect was tested in two cell pairs and yielded full uncoupling during 1-2 min. Recovery of intercellular con-ductance during period of CO 2 was relatively slow and took 10 -20 min to reach 50% of control g j level (data not shown). Thus, Cx57 channels exhibited similar properties as other connexin channels, in response to these well established uncoupling factors. DISCUSSION The new mouse Cx57 gene and its derived protein show the structural features of ␣ class connexins, i.e. no intron in the coding region, three conserved cysteine residues in both of the putative extracellular loops, and relatively high amino acid sequence identity to other ␣ class connexins. We have located the Cx57 gene, Gja9, on mouse chromosome 4, to which also the genes for Cx30.3 (Gjb5), Cx31 (Gjb3), Cx31.1 (Gjb4), and Cx37 (Gja4) have been previously assigned (36,45). Although these previously mapped genes apparently form a cluster, the map location of Gja9 places it more than 30 centimorgans proximal to this cluster. Cx30.3, Cx31, and Cx31.1 are ␤ class connexins, mainly expressed in skin, whereas Cx37 is a member of the ␣ subgroup and highly expressed in endothelium.
Cx57 mRNA was found in the mouse embryo after 8 dpc and, at low level, in several tissues: heart, skin, kidney, lung, ovary, testis, but not in brain and sciatic nerve. Based on its relatively high amino acid identity and its similar expression in ovary, heart, kidney, lung, and intestine, at low mRNA abundance in total RNA, it is likely that mouse Cx57 and porcine Cx60 (44) are analogous genes in the two species. This expression pattern, however, does not lead to insights into the functional role of Cx57 in the tissues in which it is expressed. Expression of Cx43, Cx30.3, and Cx37 in ovaries has been previously reported (44, 49, 50). Simon et al. (50) had reported that Cx37-deficient mice suffer from female infertility, due to inhibited development of oocytes. Transfer of microinjected neurobiotin has been shown to occur between the oocyte and surrounding granulosa cells in wild type mice, but not in Cx37-deficient mice. Cx37 is expressed in wild type oocytes (50). It is not known to which connexin hemichannels the Cx37 hemichannels in oocytes dock for functional communication with granulosa cells. Cx57 could be a candidate, since we have shown that it can form heterotypic gap junctions with Cx37 hemichannels expressed in transfected HeLa cells. So far, it is not known in which cell type(s) Cx57 protein is expressed in mouse ovaries.
Are there peculiar features of the mouse Cx57 gene and its channel-forming protein? Compared with other connexin genes, the level of Cx57 mRNA was rather low in all tissues where we found a positive signal. It is possible that we have not identified the cell type that may express Cx57 mRNA at high abundance. Alternatively, coexpression of Cx57 with other connexins in the same cell type could lead to heterotypic channels (for example with Cx37 (Gja-4), Cx43 (Gja-1), Cx30.3 (Gja-5), as shown in this paper) or heteromeric channels (not studied in this paper) and, thereby, influence the pattern of functional A, examples of g j -V j dependence record. Instantaneous junctional conductance, g j (inst), was measured by applying periodic positive and negative pulses of 25 mV to cell 2. Voltage in cell 1 was constant and equal to the holding potential (see dashed line). During depolarization of cell 2 to 50 mV, g j decreased about 10 times and upon hyperpolarization to the holding potential slowly recovered to the control value. B, data summarizing g j (ss)-V j dependence, where individual g j (ss) data are normalized to g j (inst), g j (ss)/g j (inst). Continuous lines show fitting of the experimental data with the Boltzmann equation. C, example of I j , V 1 , and V 2 records illustrating V m dependence. When both cells were depolarized simultaneously from Ϫ20 mV to ϩ60 mV and then to ϩ70 mV, g j increased. Instantaneous I j was measured by applying voltage steps of Ϯ25 mV. connexin channels between the same or different cell types.
Our electrophysiological analysis revealed that Cx57 channels, expressed between transfected HeLa cells, exhibited low single-channel conductance, ␥(open) Ϸ 27 pS. Cx57 channels are relatively strongly gated by transjunctional voltage (Boltzmann V o Ϸ 25 mV) exhibiting fast transitions between open and residual states (see Fig. 9B). ␥(residual) was in the range of 4 -6 pS and the ratio ␥(residual)/␥(open) ϭ 5/27 Ϸ 0.18. This number is close to the value of g min (g minϩ ϭ 0.14 Ϯ 0.02 and g minϪ ϭ 0.20 Ϯ 0.03) and suggest that at V j values more than ϳ50 mV, almost all channels are in the residual state.
Cx57 demonstrated unique g j -V m dependence and g j rise during depolarization of cells. This property is well expressed in insect cells, where g j strongly decreased with depolarization (48,51,52). g j -V m dependence was unusual for the members of the connexin family and was reported only for Cx45 (53). The resting potential of cells during different periods of the cell cycle and during development might change significantly. This would potentially involve V m -sensitive gating in the regulation of cell-cell coupling.
The Cx57-expressing HeLa-transfected cells showed low intercellular permeability to neurobiotin and were not permeable to Lucifer yellow. Thus, Cx57 channels appear to be penetrable preferentially by small molecules. At present, it is not known why connexin channels of low single-channel conductance may be coexpressed between cells that are connected by other channels of higher unitary conductance.
In order to further analyze these questions, antibodies to the Cx57 protein need to be raised and used to study the cell type-specific expression pattern in relation to other connexins. Furthermore, mouse mutants with defects in the Cx57 gene should allow one to decipher the peculiar functional contribution of this new connexin gene to molecular physiology and/or development of the ovary and other tissues.