Gap junctions are plasma membrane structures that contribute to intercellular communication through channels that allow the direct exchange of ions, signaling molecules, and metabolites. The channels that form gap junctions are composed of structural components known as connexons, with the plasma membrane of each cell contributing one connexon to form a half channel. Each connexon comprises six protein subunits termed connexins. Connexins are members of a family of closely related proteins, found only in vertebrates, which possess a highly conserved amino acid sequence and membrane topology (Bruzzone et al., 1996a,b; Simon and Goodenough, 1998; Yeager et al., 1998; White and Paul, 1999).
Gap junctional communication is thought to contribute to the regulation of many cellular activities that are critical to the appropriate organization of the developing organism. Because gap junction–mediated intercellular communication is frequently established early in the developing embryo, it is likely that this form of communication is an important determinant for establishing correct tissue morphogenesis and pattern formation (Caveney, 1985; Kidder, 1987; Warner, 1992; Lo, 1996).
The zebrafish, Danio rerio, has recently emerged as a new model organism for the study of vertebrate development. A number of genetic strategies have been developed in the zebrafish making it possible to conduct saturation mutagenesis screens, and this, coupled with attributes such as transparency of the embryo and external fertilization, make the search for mutant phenotypes relatively straightforward (Mullins and Nusslein-Volhard, 1993; Mullins et al., 1994; Driever et al., 1996; Haffter et al., 1996; Solnica-Krezel, 1999). The rapid, early eye development of the zebrafish, coupled with its large size relative to the rest of the embryo, make the ocular lens an attractive structure with which to study the genetic control of embryonic morphogenesis (Schmitt and Dowling, 1994, 1996, 1999; Brockerhoff et al., 1995; Raymond et al., 1995; Fadool et al., 1997; Sullivan et al., 1997; Fini et al., 1997; Brockerhoff et al., 1998; Neuhauss et al., 1999).
The lens of the adult vertebrate eye solely comprises two cell types, with a simple cuboidal epithelium lining the anterior surface, and elongated fibre cells forming the bulk of the lens tissue. The surface epithelial cells are able to regulate the movement of ions and metabolites between the aqueous humor and the fibre cells. Because the lens is an avascular organ, it depends greatly on the anterior epithelium for ionic homeostasis, and without the provision of ions and metabolites, the crystallins, which are the major soluble proteins in the fibre cell cytoplasm, cannot remain in solution and aggregate to form cataracts (Goodenough, 1992; Mathias et al., 1997; Francis et al., 1999).
In the vertebrate lens, gap junctional coupling is present between fibre cells, between epithelial cells, and between the epithelial cells and the fibre cells (Paul et al., 1991; White et al., 1992; Rup et al., 1993; Jiang et al., 1994; Jiang and Goodenough, 1996; Rae et al., 1996; Gao and Spray, 1998; Ebihara et al., 1999; Berthoud et al., 1999; Dahm et al., 1999). Because there are only two types of cells in the lens expressing only three different types of connexins, the lens is an ideal structure to use for the study of intercellular communication. The lens is also a particularly attractive system with which to study the role of connexins in development because its avascularity compels it to rely so heavily upon gap junctional communication for maintaining normal growth and metabolic equilibrium.
There are three different connexins known to be expressed in the vertebrate lens. Connexin43 is expressed in the membranes of the epithelial cells (Beyer et al., 1989; Musil et al., 1990). The fibre cells of the lens express two connexins. These are rodent Cx46 (Paul et al., 1991) [orthologous with chicken Cx56 (Rup et al., 1993), bovine Cx44 (Gupta et al., 1994), human Cx46 (Mackay et al., 1999)]; and rodent Cx50 (White et al., 1992) [orthologous with human Cx50 (Church et al., 1995), chicken Cx45.6 (Jiang et al., 1994), and sheep Cx49 (Yang and Louis, 1995)]. The picture may be more complex than suggested above. Studies on cultured sheep lens epithelial cells (TenBroek et al., 1994, 1997), and developing chick and mouse lenses in vivo (Jiang et al., 1995; White et al., 1998), show that under these conditions, the epithelial cells can express the fibre type connexins in addition to Cx43. More recently, the expression of Cx50 has been shown in adult mammalian lens epithelial cells in vivo (Dahm et al., 1999).
We report the cloning and initial characterization of Cx44.1, the zebrafish orthologue of rodent and human Cx50. The zebrafish Cx44.1 gene codes for a 391 amino acid protein, with a predicted molecular mass of approximately 44.1 kDa. By using the LN54 radiation hybrid panel, the Cx44.1 gene was mapped to linkage group 1. In situ hybridization and Northern blot analyses revealed that the Cx44.1 gene is expressed in the lens of the developing embryo, and that expression begins during the early stages of lens morphogenesis. Expression of Cx44.1 in Xenopus oocyte pairs generated functional gap junctional channels with electrical properties similar to those of its chicken and rodent orthologues.
Results and Discussion
Isolation and Characterization of Cx44.1 Clones
The great degree of amino acid sequence conservation in certain segments of the connexins and the lack of introns in the coding region were used to facilitate the isolation of a zebrafish connexin gene. Degenerate primers matching conserved regions of the known vertebrate connexins were designed to PCR-amplify a partial connexin sequence from genomic DNA (Fig. 1). PCR with the degenerate primers produced a single, approximately 1,000-bp band, which was subsequently cloned, and sequenced. The largest predicted open reading frame was used to perform a BlastP search of GenBank (Altschul et al., 1997). This search revealed a high degree of sequence similarity to the lens connexins belonging to the Cx50 group, with a 53.6% amino acid identity when globally aligned with chicken Cx45.6 (Rup et al., 1993).
To obtain the full coding sequence, an 850-bp EcoRI fragment from the partial clone was used to screen 1 × 106 plaques of a zebrafish genomic library at high stringency. One positive clone that elicited a strong signal on the primary screen was followed through two additional rounds of screening.
To isolate the insert contained in the λ vector, two large PCR fragments (approximately 4 and 9 kb) were amplified using gene and phage-specific primers (Fig. 1). The 9-kb fragment contained the 3′ end of the coding region and 3′ flanking sequences. The 4-kb fragment contained the 5′ end of the coding region as well as 5′ flanking sequences.
The long PCR products were used to sequence 2,233 bases off both strands, starting from the centre of the partial PCR clone sequence. The longest open reading frame identified is 1,173 nucleotides long, coding for a 391 amino acid protein with a predicted molecular weight of 44,094 Da, leading to its designation, by convention, as zebrafish Cx44.1. As is common with connexin genes, the coding region did not appear to be interrupted by any introns (Paul et al., 1991; Haefliger et al., 1992; White et al., 1992; Rup et al., 1993; Jiang et al., 1994). The sequenced region included 433 bases upstream of the start codon, and 624 bases downstream of the termination codon. The Cx44.1 sequence has been deposited in GenBank under accession number AF288817.
Analysis of the amino acid sequence of zebrafish Cx44.1 reveals the highly conserved extracellular regions, transmembrane domains, and amino terminus characteristic of connexin proteins, and places zebrafish Cx44.1 in an orthologous group of connexins expressed primarily in the ocular lens fibre cells. In pair-wise alignments, the predicted amino acid sequence of Cx44.1 has 63.1, 63.7, 65.2, and 67.4% identity to mouse Cx50, human Cx50, sheep Cx49, and chicken Cx45.6, respectively, when globally aligned. A multiple amino acid sequence alignment of these sequences reveals the extensive sequence similarity between the lens connexins of this subgroup (data not shown). The high degree of overall sequence conservation shown by these alignments, particularly in the transmembrane segments and the extracellular loops, suggests that zebrafish Cx44.1 may adopt the typical connexin membrane topology. These sequence comparisons also show that the most divergent region is the cytoplasmic tail, and the lower predicted molecular weight for Cx44.1, as well as for chicken Cx45.6 (Jiang et al., 1994) when compared with the mammalian orthologues, is primarily a reflection of shorter cytoplasmic tails in the nonmammalian members. An increase in the length of the tail appears to have occurred in the course of evolution. It has been proposed that this disparity in length is nonfunctional, although the cytoplasmic domains are regions for certain modifications such as phosphorylation (Bennett et al., 1991). An understanding of the significance of these size differences, if any, awaits studies directly comparing the responses of these different connexins to modifiers such as kinases, phosphatases, and proteases.
The Cx44.1 protein also has the three conserved cysteine residues, common to all connexins, in each of the extracellular loops. Comparisons of connexin amino acid sequences with the amino acid sequences of other channel-forming proteins have suggested that an amphipathic domain in the third transmembrane region (K-X-X-X-E), present in all the connexins of this group, including Cx44.1, may contribute to the pore lining (Milks et al., 1988; Unwin, 1989). This hypothesis remains controversial, however, and more recent studies have suggested that a portion of the first transmembrane segment may also contribute to the lining of the channel (Zhou et al., 1997).
Mapping of the Cx44.1 Gene in the Zebrafish Genome
The LN54 radiation hybrid panel was used to map the position of the Cx44.1 gene in the zebrafish genome. This hybrid panel was constructed by fusing irradiated zebrafish AB9 cells with mouse B78 cells (Hukriede et al., 1999). To minimize the possibility that the results would be confounded by amplification from the mouse Cx50 gene present in the DNA from each hybrid, the primers were chosen such that the reverse primer corresponded to a portion of the 3′ untranslated region, where the similarity to the mouse sequence is very low. The appropriateness of the choice of primers was validated by the absence of a visible PCR amplification product from the control mouse parental cell line DNA. Analysis of the panel was carried out by a PCR assay of the DNA from the entire panel of 93 hybrid cell lines, and was performed twice. Each set of PCRs contained DNA from the zebrafish and mouse parental cell lines as positive and negative controls, respectively, as well as a 1:10 mixture of zebrafish and mouse parental cell line DNA to evaluate the sensitivity of the assay. The presence or absence of amplification products was scored as described above, and the resulting scoring vector was entered into the RHMAPPER program (Hudson et al., 1995). By this approach, the Cx44.1 gene mapped to linkage group 1 (LG 1), 2.43 centiRays (cR) from framework marker Z11154 (LOD of 13.3). In the LN54 panel, on average, 1 cR equals 148 kb. Recently, a syntenic relationship has been demonstrated between zebrafish LG1 and human chromosome 1 (Hukriede et al., 1999; Barbazuk et al., 2000). The mapping of the Cx44.1 gene to LG1 is consistent with these findings and provides additional evidence for a syntenic relationship between zebrafish LG1 and human chromosome 1.
Expression of Cx44.1
To characterize the expression pattern of Cx44.1, whole-mount in situ hybridizations were performed on embryos at various developmental stages. Two partially overlapping riboprobes were used to verify the specificity of the hybridizations. Probe A encompassed a large portion of the open reading frame, whereas probe B corresponded to the cytoplasmic tail and 3′ untranslated sequence. The staining patterns observed with the two probes were indistinguishable; therefore, the data from the two probes were combined and will be described together. A strong hybridization signal was observed in the ocular lenses of embryos at 24, 28, and 30 hpf (Fig. 2A). The specificity with which these probes elicited positive signals only in the lens correlates well with previous studies showing that the Cx44.1 orthologues are expressed exclusively in the lens (White et al., 1992; Jiang et al., 1994; Church et al., 1995; Yang and Louis, 1995). No hybridization signal was detected in embryos at 48, 72, 60, and 96 hpf. The lenses in 36 hpf embryos displayed variable staining intensities, ranging from quite weak to quite strong (Fig. 2C). Although the reasons for this variability are not clear, it may simply be a reflection of the heterogeneity in the rate of development between clutches of embryos. Embryos at 20 hpf were included in some of the experiments, but the lenses in these embryos had not yet developed beyond the placode stage, and no hybridization signal was detected (data not shown). Each experiment included control hybridizations with sense riboprobes. These probes never produced any signals above background (Fig. 2B and D). To verify the whole-mount results, and to examine the labelling pattern in more detail at the cellular level, some of the embryos were embedded and semithin sectioned following extended colour reaction times. Observations of the sections confirmed the expression results obtained from the whole-mounts. Strong expression levels were seen in the developing lens in the sectioned embryos at 24 and 36 hpf (depending on the batch of embryos) probed with the antisense riboprobe (Fig. 2E and G). The embryos hybridized with the sense probe were always negative for staining (Fig. 2F and H), as were those at 72 hpf probed with either antisense riboprobe (data not shown). These results suggest that Cx44.1 expression is restricted to the lens, and that expression is initiated just subsequent to its morphological appearance. These results also suggest that the features of the primary sequence of this group of connexins that impart the specialized properties required for function in the lens must have arisen relatively early in vertebrate evolution. Furthermore, the transcriptional regulators responsible for the lens-exclusive expression of these connexin genes must also have been in place in the last common ancestor of zebrafish and mammals, about 420 million years ago (Postlethwait et al., 1999).
Developmental changes in Cx44.1 transcript levels were also investigated by Northern blot analysis. Poly (A)+ RNA from embryos at 5 to 8 hpf (day 1), 24 to 28 hpf (day 2), 48 hpf (day 3), 72 hpf (day 4), 96 hpf (day 5), and 120 hpf (day 6) was probed with a 32P labelled Cx44.1 antisense riboprobe at high stringency. A single, approximately 3.8-kb mRNA was detected in all the samples except day 1 (Fig. 3A). Reprobing the blot with the control eF1α probe showed that intact RNA had been loaded in all lanes, in similar amounts (Fig. 3B). The relative abundance of Cx44.1 transcripts in the embryo appears to be very low since the Cx44.1 message was not detected in total RNA with random primed DNA probes. Also, while the probes used in Figure 3A and B were of similar specific activity, the Cx44.1 signal seen in Figure 3A required a 5-day film exposure, whereas the eF1α signal seen in Figure 3B required only a 2-min exposure.
Together, our in situ hybridization and Northern blot data indicate that the Cx44.1 gene becomes transcriptionally active in the lens at the earliest stages in its morphogenesis and then continues to be transcribed until at least the sixth day of embryogenesis. In fact, it is probable that transcription continues for the entire lifespan of the fish, because it has been shown that in vertebrates new lens fibre cells are produced throughout the life of the organism (Fini et al., 1997; Mathias et al., 1997). The expression of mouse Cx50 and chicken Cx45.6 during embryonic development has been investigated previously and our results are in general agreement with the results from these studies. Using immunofluorescence microscopy, Evans and colleagues (Evans et al., 1993) were unable to detect Cx50 in the developing mouse lens before the formation of the first primary lens fibre cells. With the appearance of the primary and then secondary fibre cells, Cx50 was detected between these cells. Similarly, in the chick, Cx45.6 was detected throughout embryonic development of the lens by fluorescence microscopy (Jiang et al., 1994, 1995). In addition, two different sized Cx45.6 transcripts were observed in the chick embryo, and the relative amounts of these two transcripts appeared to be developmentally regulated (Jiang et al., 1995). While not a consistent feature, we have occasionally seen a second, slightly smaller, Cx44.1 band on our Northern blots (data not shown). Clarification of whether the two bands are alternatively spliced transcripts or produced from different transcriptional start sites requires further analysis.
From our Northern blot experiments it is clear that Cx44.1 transcripts are present in the zebrafish embryo between 24 and 120 hpf, yet we never detected Cx44.1 transcripts by in situ hybridization beyond 36 hpf. This may simply be due to an inability of the in situ hybridization probes to penetrate the lens sufficiently in the whole-mount in situ hybridization protocol used. By 36 hpf, the zebrafish lens has begun to form a clearly distinguishable dense core or nucleus, surrounded by less dense outer layers of cells (Easter and Nicola, 1996). As the growth of the lens continues during embryonic development it becomes increasingly dense due to tight packing of the fibre cells, accompanied by extensive reduction in the volume of the extracellular space and high levels of crystallin proteins being expressed in the cytoplasm (Schmitt and Dowling, 1994; Easter and Nicola, 1996; Mathias et al., 1997).
One interesting and unique feature of many of the lens fibre connexins is that they possess unusually long transcripts relative to the size of their coding regions. Northern blot analyses have shown transcript sizes of 8.5 Kb for mouse Cx50 (White et al., 1992), 6.8 Kb for sheep Cx49 (Yang and Louis, 1995), 6.4 and 9.4 Kb for chicken Cx45.6 (Jiang et al., 1994), 8.0 Kb for chicken Cx56 (Rup et al., 1993), and 3.8 Kb for zebrafish Cx44.1 (present study). Most of the other connexins have a transcript size in the range of 1.5–2.5 Kb. This unusual length must be due to very extensive 5′ and 3′ untranslated regions. Long untranslated regions are frequently involved in translational control of some expression (Preiss and Hentze, 1999), which may be a particularly important mechanism for the control of gene expression in the lens since the fibre cells lose their nuclei as they mature (Mathias et al., 1997). Except for a good initiation of translation context sequence (the consensus sequence is GCCA/GCCATGG (Kozak, 1991), the Cx44.1 sequence is TCAACCATGG), whether the zebrafish Cx44.1 mRNA contains specific sequences involved in translational regulation is difficult to assess based on our data. The Cx44.1 transcript is approximately 3.8 kb, whereas we determined only about 2.2 kb of sequence from the genomic clone. We are, therefore, missing at least 1.6 kb of untranslated sequence from the mRNA. Furthermore, some of the 433 bases sequenced upstream of the start codon probably belong to an intron (there are potential splice acceptor sites located both 16 and 298 nucleotides upstream of the start codon), and analysis of the 624 bases sequenced downstream of the termination codon failed to reveal a poly (A) addition sequence. Clearly, isolation and sequencing of a Cx44.1 cDNA clone, or more extensive sequencing of our genomic clone is required before translational control of Cx44.1 can be fully addressed. Similarly, identification of Cx44.1 transcriptional regulatory elements awaits further sequencing of the upstream region (no promoter elements are present in the 433-bp segment upstream of the start codon).
The functional properties of Cx44.1 were analyzed in the paired Xenopus oocyte system (Dahl et al., 1987; Swenson et al., 1989). Before injection of mRNA or water, an oligonucleotide antisense to Xenopus Cx38 was injected into the oocytes to eliminate the possible contribution of this endogenous connexin to the observed conductance (Bruzzone et al., 1993). Following exogenous connexin mRNA injection, oocytes were then paired for measurements of junctional conductance. Table 1 summarizes the conductance data. While oocytes injected with water were not electrically coupled (0.03 ± 0.02 μS; n = 14), oocyte pairs injected with Cx44.1 were highly coupled (8.83 ± 3.55 μS; n = 18), showing that expression of Cx44.1 results in the formation of functional intercellular channels in oocytes. Similar levels of conductance were detected between pairs of oocytes expressing rat Cx32, a well-characterized member of the connexin family (Dahl et al., 1987; Swenson et al., 1989; Barrio et al., 1991). The current-voltage relationship for Cx44.1 channels is illustrated in Figure 4. Hyperpolarizing or depolarizing voltage steps of >10 mV evoked junctional currents that decreased relatively slowly with time, approaching steady state. The rate of this decay increased with increasing transjunctional voltage. At transjunctional potentials of ±50 mV, the current decay was well fit by a single order exponential with a time constant (τ) of 115 ± 12 ms (mean ± SD, n = 4). The current-voltage relationship observed for Cx44.1 is very similar to that of chicken Cx45.6 and mouse Cx50 (White et al., 1994; Jiang et al., 1995; Srinivas et al., 1999).
Table 1. Intercellular Conductance Induced by Zebrafish Cx44.1 in Paired Xenopus Oocytes
Gj [μS, mean ± SE]
Number of pairs
6.79 ± 1.87
0.03 ± 0.02
8.83 ± 3.55
The voltage dependence of Cx44.1 channels was further analyzed by plotting junctional conductance (Gj) as a function of transjunctional potential (Vj) (Fig. 4C). Gj values for steady-state junctional conductance (Gjss) were normalized to the maximal conductance measured at the lowest Vj (=10 mV). No fast gating effects (<5 ms) of voltage on these channels were observed (data not shown). In contrast, Gjss was sharply dependent on voltage. This plot was fitted to a Boltzmann relation (Spray et al., 1981) (Table 2). From this analysis, Cx44.1 channels displayed voltage sensitivity virtually identical to that of chicken Cx45.6 and mouse Cx50 (White et al., 1994; Jiang et al., 1995).
Table 2. Boltzmann Parameters of Zebrafish Cx44.1 and Its Mouse and Chicken Orthologs
The electrophysiological data clearly demonstrate that Cx44.1 induces communicating channels when expressed in paired Xenopus oocytes, and, therefore, is a functional member of the connexin gene family. Cx44.1 is more homologous with rodent Cx50 and chick Cx45.6 than with rodent Cx46 or chick Cx56 among the lens fiber connexins. Consistent with this sequence identity, the voltage gating properties (T, A, V0, and Gjmin) of the channels formed by Cx44.1 were more similar to those formed by mouse Cx50 and chick Cx45.6 (White et al., 1994; Jiang et al., 1994) than to those formed by rat Cx46 or chick Cx56 (Ebihara et al., 1995). Taken together, these data confirm that zebrafish Cx44.1 is a member of the Cx50 branch of the connexin family tree.
We report here the molecular cloning of a new connexin, zebrafish Cx44.1. Furthermore, we demonstrate that this connexin forms functional gap junction channels, and is expressed primarily in lens fiber cells during early lens development. We also show that the Cx44.1 gene resides on LG1.
The lens fibre specific group of connexins has been implicated in the etiology of some forms of cataracts. The ionic balance required by the metabolically inactive fibre cells to prevent the crystallin proteins from precipitating and forming cataracts is primarily regulated by membrane pumps and channels, and gap junctions (Mathias et al., 1997). Recently, a Cx46 knockout mouse was shown to develop nuclear cataracts, probably as a consequence of crystallin proteolysis (Gong et al., 1997, 1998). In addition, the locus for autosomal dominant congenital zonular pulverulent cataract on human chromosome 13q11 has been linked to three different mutations in Cx46 (Mackay et al., 1999; Rees et al., 2000). Similarly, a knockout of the Cx50 gene in mice leads to congenital zonular pulverulent cataracts (White et al., 1998), and in humans two different mutations in the Cx50 gene on chromosome 1q21.1 have been linked with congenital autosomal dominant zonular pulverulent cataracts (Shiels et al., 1998; Berry et al., 1999). With the discovery of zebrafish Cx44.1, and Cx43 (unpublished data), we can now analyse the roles that connexin genes play during lens development, as well as how these genes can influence the development of certain lenticular diseases, such as cataracts. The use of the zebrafish model for these studies will facilitate the identification of some of the various interacting components by genetic methods.
The breeding stock of adult zebrafish was obtained from local pet stores. Adult zebrafish maintenance and embryo collection were according to Westerfield (Westerfield, 1995).
Molecular cloning of Cx44.1 was accomplished by first PCR amplifying a portion of the coding sequence from genomic DNA with degenerate primers, and then using the amplified fragment to screen a genomic library according to standard protocols (Sambrook et al., 1989).
The Cx44.1 gene was mapped in the zebrafish genome by PCR analysis of the LN54 radiation hybrid (RH) cell panel (Hukriede et al., 1999). The 93 DNA samples representing this hybrid panel, as well as control DNA from each of the parental cell lines, were screened by PCR with gene-specific primers. The results were analyzed with the RHMAPPER software program by web submission of the raw data to http://mgchd1.nichd.nih.gov:8000/zfrh.beta.cgi.
Two different regions of Cx44.1 were used to make riboprobes for in situ hybridization. Plasmid DNA containing either probe A or probe B was linearized with the appropriate restriction enzymes, and in vitro transcribed DIG-labelled single stranded sense and antisense riboprobes were generated using T3, T7, or SP6 polymerases (Promega) in the presence of DIG-11-UTP (Boehringer Mannheim). Embryos were fixed in 4% paraformaldehyde in 0.1% Tween 20 in PBS (PBST) at 4°C overnight. The protocol by Harland (1991), with modifications from Lee (Lee et al., 1996), was followed for the in situ hybridization. Following the post-hybridization washes, the embryos were incubated with a preabsorbed, alkaline phosphatase-conjugated, sheep anti-DIG antibody (final dilution 1:4,000) (Boehringer Mannheim) in block solution for 4 hr at room temperature. The colour reaction was performed with 4.5 μl/ml nitrobluetetrazolium chloride (Boehringer Mannheim) and 3.5 μl/ml 5-bromo-4-chloro-3-indolyl-phosphate (Boehringer Mannheim). Embryos for whole-mount viewing were cleared and mounted in benzyl benzoate:benzyl alcohol (2:1; Sigma). Embryos to be sectioned were gradually dehydrated until they were immersed in 70% ethanol and then embedded in JB-4 infiltration resin (Polysciences) according to the manufacturer's instructions.
For Northern blots, total RNA was isolated with Trizol Reagent (Life Technologies). Poly (A)+ RNA was isolated from the total RNA using the Oligotex mRNA Mini Kit (Qiagen). Approximately 2 μg of poly (A)+ RNA from each stage was separated by formaldehyde/agarose gel electrophoresis and transferred to nylon membranes according to standard protocols (Sambrook et al., 1989). Radioactively labelled riboprobes were hybridized to the blots for 12 to 16 hr at 55°C. To assess the integrity of the RNA and the loading levels, the blot was reprobed with an elongation factor 1 α (eF1α) antisense riboprobe (plasmid was a gift from Karl Clark, University of Minnesota) under the same conditions. The blots were washed three times in 0.1× SSC and 0.1% SDS at 68°C and exposed to film.
For electrophysiological measurements in Xenopus oocytes, a 1,720-bp fragment, containing the entire Cx44.1 coding region and flanking sequences, was PCR amplified from genomic DNA and cloned into the pCS2+ vector (Turner and Weintraub, 1994). Capped Cx44.1 mRNA was transcribed in vitro with SP6 RNA polymerase using the mMessage mMachine kit (Ambion) according to the manufacturer's instructions.
Stage V–VI oocytes were removed from gravid Xenopus laevis females and processed for microinjection as described previously (Swenson et al., 1989). To suppress contributions to the measured currents from endogenous connexins, all oocytes were injected initially with an oligonucleotide antisense (2.5 ng/cell) to a sequence of Xenopus Cx38 mRNA (5′-CTG ACT GCT CGT CTG TCC ACA CAG-3′, (Bruzzone et al., 1993). After incubation for 24 hr at 18°C, antisense-treated oocytes were injected with 40 nl (4 to 6 ng/cell) of connexin cRNA, and then stripped of their vitelline membranes and paired. Water-injected oocytes served as controls.
We are grateful to Drs. M. Petkovich and B. Beckett for providing the zebrafish genomic library and for their suggestions and insights regarding the library screening. We are also indebted to Dr. Marc Ekker for providing the LN54 RH panel. This work was supported by a Natural Sciences and Engineering Research Council of Canada grant awarded to Dr. G. Valdimarsson, and grants from the National Eye Institute (EY-13163 and EY-02430) awarded to Drs. T.W. White and D.A. Goodenough.