Major loss of the 28-kD protein of gap junction in proliferating hepatocytes.

There is a reduction in the 28-kD gap junction protein detectable by immunofluorescence in livers of partially hepatectomized rats and in cultured hepatocytes stimulated to proliferate. By the coordinate use of antibodies directed to the hepatic junction protein (HJP28) and the use of a monoclonal antibody that recognizes bromodeoxyuridine (BrdU) incorporated into DNA, we have been able to study the relationship between detectable gap junction protein and cell division. Hepatocytes that label with BrdU in the regenerating liver and in cell culture show a significant reduction of HJP28. Cells that do not synthesize DNA, on the other hand, show normal levels and distribution of immunoreactive gap junction protein. We postulate that the quantitative changes in gap junction expression might play an important role in the control of proliferation in the liver.

Comparative characterization of the 21-kD and 26-kD gap junction proteins in murine liver and cultured hepatocytes.

Affinity-purified antibodies to mouse liver 26- and 21-kD gap junction proteins have been used to characterize gap junctions in liver and cultured hepatocytes. Both proteins are colocalized in the same gap junction plaques as shown by double immunofluorescence and immunoelectron microscopy. In the lobules of rat liver, the 21-kD immunoreactivity is detected as a gradient of fluorescent spots on apposing plasma membranes, the maximum being in the periportal zone and a faint reaction in the perivenous zone. In contrast, the 26-kD immunoreactivity is evenly distributed in fluorescent spots on apposing plasma membranes throughout the rat liver lobule. Immunoreactive sites with anti-21 kD shown by immunofluorescence are also present in exocrine pancreas, proximal tubules of the kidney, and the epithelium of small intestine. The 21-kD immunoreactivity was not found in thin sections of myocardium and adult brain cortex. Subsequent to partial rat hepatectomy, both the 26- and 21-kD proteins first decrease and after approximately 2 d increase again. By comparison of the 26- and 21-kD immunoreactivity in cultured embryonic mouse hepatocytes, we found (a) the same pattern of immunoreactivity on apposing plasma membranes and colocalization within the same plaque, (b) a similar decrease after 1 d and subsequent increase after 3 d of both proteins, (c) cAMP-dependent in vitro phosphorylation of the 26-kD but not of the 21-kD protein, and (d) complete inhibition of intercellular transfer of Lucifer Yellow in all hepatocytes microinjected with anti-26 kD and, in most cases, partial inhibition of dye transfer after injection of anti-21 kD. Our results indicate that both the 26-kD and the 21-kD proteins are functional gap junction proteins.

Specific permeability and selective formation of gap junction channels in connexin-transfected HeLa cells.

DNAs coding for seven murine connexins (Cx) (Cx26, Cx31, Cx32, Cx37, Cx40, Cx43, and Cx45) are functionally expressed in human HeLa cells that were deficient in gap junctional communication. We compare the permeabilities of gap junctions comprised of different connexins to iontophoretically injected tracer molecules. Our results show that Lucifer yellow can pass through all connexin channels analyzed. On the other hand, propidium iodide and ethidium bromide penetrate very poorly or not at all through Cx31 and Cx32 channels, respectively, but pass through channels of other connexins. 4,6 Diamidino-2-phenylindole (DAPI) dihydrochloride shows less transfer among Cx31 or Cx43 transfectants. Neurobiotin is weakly transferred among Cx31 transfectants. Total junctional conductance in Cx31 or Cx45 transfected cells is only about half as high as in other connexin transfectants analyzed and does not correlate exactly with any of the tracer permeabilities. Permeability through different connexin channels appears to be dependent on the molecular structure of each tracer, i.e. size, charge and possibly rigidity. This supports the hypothesis that different connexin channels show different permeabilities to second messenger molecules as well as metabolites and may fulfill in this way their specific role in growth control and differentiation of cell types. In addition, we have investigated the function of heterotypic gap junctions after co-cultivation of two different connexin transfectants, one of which had been prelabeled with fluorescent dextran beads. Analysis of Lucifer yellow transfer reveals that HeLa cells expressing Cx31 (beta-type connexin) do not communicate with any other connexin transfectant tested but only with themselves. Two other beta-type connexin transfectants, HeLa-Cx26 and -Cx32, do not transmit Lucifer yellow to any of the alpha-type connexins analyzed. Among alpha-type connexins, Cx40 does not communicate with Cx43. Thus, connexins differ in their ability to form functional heterotypic gap junctions among mammalian cells.

Transplacental uptake of glucose is decreased in embryonic lethal connexin26-deficient mice.

Mice that harbor a targeted homozygous defect in the gene coding for the gap junctional protein connexin26 died in utero during the transient phase from early to midgestation. From day 10 post coitum onwards, development of homozygous embryos was retarded, which led to death around day 11 post coitum. Except for growth retardation, no gross morphological alterations were detected between homozygous connexin26-defective embryos and wild-type littermates. At day 9 postcoitum, when chorioallantoic placenta started to function, connexin26 was weakly expressed in the yolk sac epithelium, between syncytiotrophoblasts I and II in the labyrinth region of the placenta, and in the skin of the embryo. At day 10 post coitum, expression of connexin26 in the placenta was much stronger than at the other locations. To analyze involvement of connexin26 in the placental transfer of nutrients, we have measured embryonic uptake of the nonmetabolizable glucose analogue 3-O-[14C]methylglucose, injected into the maternal tail vein. At day 10 post coitum, viable, homozygous connexin26-defective embryos accumulated only approximately 40% of the radioactivity measured in wild-type and heterozygous littermates of the same size. We conclude that the uptake of glucose, and presumably other nutrients as well, from maternal blood into connexin26-deficient mouse embryos was severely impaired and apparently not sufficient to support the rapid organogenesis during midgestation. Our results suggest that connexin26 gap junction channels likely fulfill an essential role in the transfer of maternal nutrients and embryonic waste products between syncytiotrophoblast I and II in the labyrinth layer of the mouse placenta.

Reduced cardiac conduction velocity and predisposition to arrhythmias in connexin40-deficient mice.

Intercellular channels of gap junctions are formed in vertebrates by the protein family of connexins and allow direct exchange of ions, metabolites and second messenger molecules between apposed cells (reviewed in [1-3]). In the mouse, connexin40 (Cx40) protein has been detected in endothelial cells of lung and heart and in certain heart muscle cells: atrial myocytes, cells of the atrial ventricular (AV) node and cells of the conductive myocardium, which conducts impulses from the AV node to ventricular myocyctes [3]. We have generated mice homozygous for targeted disruption of the Cx40 gene (Cx40-/-mice). The electrocardiograph (ECG) parameters of Cx40-/- mice were very prolonged compared to those of wild type (Cx40+/+) mice, indicating that Cx40-/- mice have lower atrial and ventricular conduction velocities. For 6 out of 31 Cx40-/- animals, different types of atrium-derived abnormalities in cardiac rhythm were recorded, whereas continuous sinus rhythm was observed for the 26 Cx40+/+ and 30 Cx40+/- mice tested. The expression levels of other connexins expressed in heart (Cx37, Cx43 and Cx45) were the same in Cx40-/- and Cx40+/+ mice. Our results demonstrate the function of Cx40 in the regulation and coordination of heart contraction and show that cardiac arrhythmogenesis can not only be caused by defects in the ion channels primarily involved in cellular excitation but also by defects in intercellular communication through gap junction channels. As the distribution of Cx40 protein is similar in mouse and human hearts, further functional analysis of Cx40 should yield relevant insights into arrhythmogenesis in human patients.

Unique and shared functions of different connexins in mice.

BACKGROUND: Connexins are the protein subunits of intercellular gap junction channels. In mammals, they are encoded by a family of at least 15 genes, which show cell-type-specific but overlapping patterns of expression. Mice lacking connexin43 (Cx43) die postnatally from obstruction of the right ventricular outflow tract of the heart. To discriminate between the unique and shared functions of Cx43, Cx40 and Cx32, we generated two 'knock-in' mouse lines, Cx43KI32 and Cx43KI40, in which the coding region of the Cx43 gene was replaced, respectively, by the coding regions of Cx32 or Cx40. RESULTS: Heterozygous mutants were fertile and co-expressed the wild-type and the corresponding recombinant allele in all tissues analyzed. Heterozygous Cx43KI32, but not Cx43KI40, mutant mothers were unable to nourish their pups to weaning age, possibly reflecting a defect in milk ejection. Homozygous mutant males were sterile because of extensive germ-cell deficiency. The ovaries of homozygous Cx43KI32 neonates exhibited all stages of follicular development and ovulation. The hearts of homozygous Cx43KI32 neonates showed mild morphological defects, but the cardiac morphology of homozygous Cx43KI40 neonates was relatively normal. Spontaneous ventricular arrhythmias were observed in most Cx43KI40 and some Cx43KI32 mutant mice, suggesting increased ventricular vulnerability in these mice. CONCLUSIONS: The postnatal lethality of Cx43-deficient mice was rescued in Cx43KI32 or Cx43KI40 mice, indicating that Cx43, Cx40 and Cx32 share at least some vital functions. On the other hand, Cx43KI32 and Cx43KI40 mice differed functionally and morphologically from each other and from wild-type mice. Thus, these connexins also have unique functions.

Transforming activity of DNA fragments from normal human lymphocytes results from spontaneous activation of a c-Ha-ras1 gene.

An activated human Ha-ras gene was present in a secondary NIH 3T3 transformant isolated after serial transfection of originally low-molecular-weight DNA fragments from normal human cells. This gene appeared to have acquired its transforming properties by a spontaneous mutation in codon 12 by substitution of a deoxythymidine residue for a deoxyguanosine residue. DNA rearrangements in the flanking sequences of the transferred Ha-ras gene were not involved in the activation of the protooncogene.

Incompatibility of connexin 40 and 43 Hemichannels in gap junctions between mammalian cells is determined by intracellular domains.

Murine connexin 40 (Cx40) and connexin 43 (Cx43) do not form functional heterotypic gap junction channels. This property may contribute to the preferential propagation of action potentials in murine conductive myocardium (expressing Cx40) which is surrounded by working myocardium, expressing Cx43. When mouse Cx40 and Cx43 were individually expressed in cocultured human HeLa cells, no punctate immunofluorescent signals were detected on apposed plasma membranes between different transfectants, using antibodies specific for each connexin, suggesting that Cx40 and Cx43 hemichannels do not dock to each other. We wanted to identify domains in these connexin proteins which are responsible for the incompatibility. Thus, we expressed in HeLa cells several chimeric gene constructs in which different extracellular and intracellular domains of Cx43 had been spliced into the corresponding regions of Cx40. We found that exchange of both extracellular loops (E1 and E2) in this system (Cx40*43E1,2) was required for formation of homotypic and heterotypic conductive channels, although the electrical properties differed from those of Cx40 or Cx43 channels. Thus, the extracellular domains of Cx43 can be directed to form functional homo- and heterotypic channels. Another chimeric construct in which both extracellular domains and the central cytoplasmic loop (E1, E2, and C2) of Cx43 were spliced into Cx40 (Cx40*43E1,2,C2) led to heterotypic coupling only with Cx43 and not with Cx40 transfectants. Thus, the central cytoplasmic loop of Cx43 contributed to selectivity. A third construct, in which only the C-terminal domain (C3) of Cx43 was spliced into Cx40, i.e., Cx40*43C3, showed neither homotypic nor heterotypic coupling with Cx40 and Cx43 transfectants, suggesting that the C-terminal region of Cx43 determined incompatibility.

Negative growth control of HeLa cells by connexin genes: connexin species specificity.

In order to examine whether different connexin gene species exert different degrees of tumor-suppressing activity, we characterized growth characteristics of a gap junction-deficient human cancer cell line, HeLa cells, before and after transfection with cDNA for three different connexins, connexin (cx) 26, cx 40, and cx 43. All transfected cell lines (3 clones transfected with the cx 26 gene, 2 clones with cx 40, and 1 with cx 43) showed establishment of gap junctional intercellular communication (GJIC). Two of the cx 26-transfected clones showed significantly slower growth compared with the parental HeLa cells. When transfectants were grown in soft agar, the three cx 26-transfected clones grew much less than the other transfectants and parent HeLa cells. When injected into nude mice, the two cx 26 clones which exhibited the highest amount of cx 26 transcript induced almost no tumors, whereas other transfectants, including the cx 26 clone which exhibited the lowest amount of cx 26 transcript, were tumorigenic. Among transfectants of various connexin genes, there was no good inverse correlation between their GJIC and tumorigenicity. GJIC levels were significantly higher in tumors induced in nude mice by clone cx 26 A and E transfectants. These results suggest that all of the connexin genes examined could induce recovery of GJIC of HeLa cells, but only the cx 26 gene exerts strong negative growth control on HeLa cells; thus, this connexin gene may have different functions from other connexin genes.

Cyclic adenosine monophosphate stimulates biosynthesis and phosphorylation of the 26 kDa gap junction protein in cultured mouse hepatocytes.

Hepatocytes prepared from 18-day-old mouse embryos were grown in serum-free medium and reached confluence after two days in culture. The total amount of the 26 kDa gap junction protein decreased in these cells during the first 24 h in culture and increased again between day 1 and day 3 more than 10-fold. At day 3 a half-life time of 2.5 to 3 h was determined for the 26 kDa protein by [35S]methionine incorporation and immunoprecipitation using affinity-purified anti-26 kDa. Incorporation of [32P]orthophosphate into the 26 kDa protein of cultured hepatocytes was found at serine residues (98%) and tyrosine residues (about 2%). The addition of dibutyryl cyclic adenosine monophosphate (db cAMP) to the culture medium at day 2 had two effects: After 15 min the extent of phosphorylation of the 26 kDa protein increased 2.7-fold whereas the total amount of the 26 kDa protein increased only 1.2-fold. After 3 h of incubation with db cAMP, a 2.5-fold increase of the 26 kDa protein was noticed which was accompanied by a 3.2-fold increase in phosphorylation of serine residues. The effects of db cAMP on phosphorylation of the 26 kDa protein could be augmented or mimicked by the addition of isoproterenol, theophylline or forskolin to the culture medium of hepatocytes. In extracts of rat hepatocarcinoma MH1C1 cells and dog kidney MDCK cells, a phosphorylated 26 kDa protein can be immunoprecipitated using anti-liver 26 kDa. These results demonstrate that the gap junction 26 kDa protein can be posttranslationally modified by cAMP-dependent phosphorylation.(ABSTRACT TRUNCATED AT 250 WORDS)

Differential expression of the gap junction proteins connexin45, -43, -40, -31, and -26 in mouse skin.

The expression of five different members of the gap junction multigene family, connexin (Cx)45, -43, -40, -31, and -26 was investigated in embryonic and adult mouse skin. For this purpose, polyclonal antibodies to Cx31 and Cx45 were raised by immunizing rabbits with fusion proteins of glutathione S-transferase and carboxy-terminal peptides including 65 amino acids of Cx31 or 138 c-terminal amino acids of Cx45, respectively. Here we describe characterization of the affinity-purified Cx31 antibodies in human HeLa cells, transfected with mouse Cx31 coding DNA, and in mouse keratinocyte-derived cell lines. In the epidermis of embryonic mice at day 19 of gestation Cx43 and -45 were detected in the basal layer, while the stratum spinosum showed expression of Cx43, -31 and -26. In the stratum granulosum we found expression of Cx31 and -26. In the epidermis of adult mice Cx43 and -31 were located similarly as in embryonic tissue, but Cx45 as well as Cx26 were not detected and in addition Cx40 was weakly expressed in the stratum basale. Furthermore, during hair development, Cx31 was detected in the inner epithelial root sheath and sebaceous glands of hair follicle. Cx43 and -40 were found in the outer epithelial root sheath and to a lesser extent in sebaceous glands. Cx31 was also demonstrated in Hel-37 and Hel-30, i.e. two related cell lines derived from mouse keratinocytes. Our results show that epidermal and follicular differentiation coincides with differential expression of five different connexin proteins, suggesting specific and coordinated function(s) of gap junctional communication during skin and hair development.

Cytoplasmic and cell surface structure of purified liver gap junctions revealed by freeze-drying.

After freeze-drying of purified liver gap junction plaques and vesicles the structural features of the inner and outer aspects of purified gap junctions were investigated. No structural details were seen on the cytoplasmic side of the connexions whereas on the cell surface side the connexions were organized in a paracrystalline pattern and exhibited a central depression or pore. We conclude that the central pore through each connexion varies in diameter along its length and that the closing site is located near the cytoplasmic face.

Expression of different connexin genes in rat uterus during decidualization and at term.

The expression of different connexin genes (cx26, cx32, cx37, cx43) that code for the protein subunits of gap junctions, was investigated in various uterine tissues during the estrous cycle of nonpregnant rats, in pregnant rats at decidualization and at term. Connexin gene expression was studied at the mRNA level by Northern blot hybridization and at the protein level by immunocytochemistry. In gap junctions from uterine epithelium, stroma, or myometrium, connexin 26 and/or connexin 43 are much more abundant than connexins 32 and 37. The expression of connexin 26 and 43 appears to be modulated by maternal steroid hormones. High expression of these connexins is found in developing decidual cells by day 7 to 8 post coitum; furthermore, coexpression of connexins 26 and 43 in myometrium is observed just before delivery on day 21 post coitum. In both the decidua and the myometrium, the connexin 26 protein appears to be distributed in lower abundance than connexin 43. In uterine epithelium only connexin 26 is expressed throughout all of the reproductive phases investigated. The enhanced expression of this gene correlates with higher levels of maternal estrogen both in the proestrus/estrus phase and at term. The distinct spatial and temporal pattern of expression of connexins 26 and 43 in different uterine tissues suggests a physiological role for these proteins during embryo implantation and subsequent contraction of the uterus at birth.

On the mechanisms of cell uncoupling induced by a tumor promoter phorbol ester in clone 9 cells, a rat liver epithelial cell line.

It is known that in Clone 9 (C9) cells, intercellular gap junctional communication (IGJC) is rapidly blocked by the tumor promoter phorbol ester, 12-O-tetradecanoylphorbol-13-acetate (TPA), but it recovers spontaneously a few hours later and becomes refractory to TPA (Yada et al., J. Membr. Biol. 88, 217-232 (1985)). We now report that gap junctions between C9 cells contain at least two junctional proteins, connexin26 (Cx26) and connexin43 (Cx43), and that the TPA-induced changes in IGJC correlate temporally to changes in the state of phosphorylation of Cx43. The latter changes were prevented by inhibition of protein kinase C. Phosphoamino acid analysis and two-dimensional tryptic peptide maps of 32P-labeled Cx43 showed that during the TPA-induced phosphorylation at least two of the phosphorylated forms of Cx43 were differentially phosphorylated in seryl residues as compared to control. TPA induced a drastic reduction in junctional conductance as well as a redistribution of unitary gap junction channel event sizes seen in control cells. These changes were associated with retrieval of Cxs from the plasma membrane. Reappearance of gap junctions formed by Cx43 but not by Cx26 accounted for the spontaneous recovery in IGJC. It is proposed that gap junctions between C9 cells contain two types of channels each formed by Cx43 or Cx26 and that they are differentially affected during the action of TPA.

Characterization of the gap junction protein connexin37 in murine endothelium, respiratory epithelium, and after transfection in human HeLa cells.

Affinity-purified antibodies to oligopeptides derived from two different regions of the carboxyterminus and cytoplasmic loop or to the last 103 C-terminal amino acids of mouse connexin37 (Cx37) were used to characterize expression of this gap junctional protein in endothelium of several murine tissues. Cx37 was expressed in endothelium of large blood vessels in brain, liver, kidney, spleen, heart, and lung, but not in capillaries. In addition, weak Cx37 immuno-signals were observed in lung respiratory epithelium of small bronchi and in alveolar epithelial cells of bronchioli. The ratios of Cx37 protein to Cx37 mRNA in adult and embryonic kidney as well as skin were 29-303-fold larger than in lung, suggesting that Cx37 mRNA was translated at different efficiencies in kidney and skin versus lung. Cx37 protein was more abundant in embryonic kidney and lung than in the corresponding adult tissues. After differential centrifugation of plasma membrane fractions in sucrose gradients, we found that Cx37-containing gap junctions in lung were much smaller than Cx32 and Cx26 aggregates from liver. HeLa cells were transfected with mouse Cx37 cDNA. In these cells, mouse Cx37 protein was phosphorylated mainly at serine, less at tyrosine, and very little at threonine residues. Three conductance states were resolved at 110, 240, and 315 pS.

Tyrosine and serine phosphorylation of the neural cell adhesion molecule L1 is implicated in its oligomannosidic glycan dependent association with NCAM and neurite outgrowth.

We have previously shown that a cis interaction between the cell adhesion molecules L1 and NCAM is mediated by N-linked oligomannosidic glycans carried by L1 and that this L1/NCAM association is involved in basal neurite outgrowth from early postnatal cerebellar neurons of mouse brain [R. Horstkorte et al., J. Cell Biol. 121, 1409-1421 (1993)]. Extending these earlier studies we investigated signal transduction mechanisms elicited by this molecular interaction. We show here that phosphorylation of L1 is reduced concomitant with reduced neurite outgrowth when the L1/NCAM interaction is inhibited by oligomannosidic glycopeptides. Similarly, when a peptide of the 4th immunoglobulin (Ig)-like domain of NCAM - representing part of NCAM's carbohydrate-binding site - was added to the culture medium of the cells, neurite outgrowth and phosphorylation of L1 was strongly reduced. No effect on neurite outgrowth and phosphorylation of L1 was observed when cells were maintained in the presence of a peptide comprising part of the 1st Ig-like domain of NCAM or in the presence of the peptide encoded by the variable alternative spliced exon (VASE), which is also located in the 4th Ig-like domain of NCAM. Furthermore, phosphorylation of tyrosine and serine residues of L1 is reduced when the L1/NCAM interaction at the cell surface of cerebellar neurons is perturbed. Our observations suggest that a signal transduction mechanism is implicated in basal neurite outgrowth in which both tyrosine and serine phosphorylation of L1 represent a possible proximal step. Some of these results were presented at the International Glycoconjugate Symposium in Seattle, USA [P. C. Heiland et al., Glycoconj. J. 12, 521(1995)].

Immunochemical and electrophysiological characterization of murine connexin40 and -43 in mouse tissues and transfected human cells.

Human HeLa or SkHep1 cells, defective in intercellular communication through gap junctions, were transfected with coding sequences of murine connexin40 (Cx40) and -43. The transfected cells were restored in gap junctional coupling as shown by 100-fold increased electrical conductance. When studied by the double whole-cell patch-clamp technique, Cx40 HeLa transfectants exhibited single channel conductances of gamma = 121 +/- 7 pS and gamma = 153 +/- 5 pS. They were voltage gated with an equivalent gating charge of z = 4.0 +/- 0.5 for a voltage of half-maximal inactivation U0 = 44 +/- 7 mV. The corresponding values of connexin43 (Cx43) HeLa transfectants are: gamma = 60 +/- 4 pS and gamma = 40 +/- 2 pS as well as z = 3.7 +/- 0.8 and U0 = 73 +/- 7 mV. Transfer of the dye Lucifer Yellow was always considerably lower in Cx40- than in Cx43-transfectants though their total junctional conductance was similar or even higher than for Cx43-transfectants. In order to characterize cell and tissue distribution as well as phosphorylation of connexin40 and -43 proteins, antibodies to C-terminal oligopeptides of these proteins were prepared and used for immunoblotting, immunoprecipitation, and immunofluorescence analysis of transfected cells where they exhibited the punctate pattern characteristic of gap junctions on contacting membranes. Phosphorylation of connexin40 was shown by immunoprecipitation from 32P-labeled, transfected SkHep1 cells. Analyses of protein distribution in tissues revealed that the amount of connexin40 detected in heart was higher than in lung which is the inverse of the level of connexin40 mRNA in these tissues, suggesting posttranscriptional control of expression. Connexin40 protein in adult mouse heart and skin is about 20-fold more abundant than in the corresponding embryonic tissue. Connexin43 in adult mouse heart appears to be more highly phosphorylated than in embryonic heart or in transfected human cells.

Gap junction connexin genes cx26 and cx43 are differentially regulated by ovarian steroid hormones in rat endometrium.

In rat endometrium, expression of gap junction connexin-26 (cx26) in the epithelium and cx43 in the uterine stroma is suppressed by progesterone before implantation. For further study of connexin gene regulation we analyzed expression of cx26, cx43, and cx32 in the endometrium of ovariectomized rats treated with different ratios of 17beta-estradiol (E2) and progesterone (P). A hormonal ratio of E2 to P that mimics conditions during pregnancy (0.1 microg E2 and 4 mg P) suppressed expression of cx26 and cx43. By changing the ratio to higher E2 levels (1 microg E2), cx26, in contrast to cx43, was not suppressed even by application of a high P concentration (10 mg). Time-course experiments supplying E2 alone led to an early gene response of cx26 within 3 h, whereas induction of cx43 transcripts was not detected until 14 h after E2 treatment. Simultaneous application of the antiestrogen ICI 182780 abolished E2-mediated induction of both connexins. No hormonal regulation of cx32 could be detected. As already shown for cx43 gene induction in the myometrium, E2-mediated induction of cx26 expression in the endometrium also required newly synthesized transcription factors. It can be concluded that only a hormonal ratio resembling conditions during pregnancy is able to suppress the expression of both cx26 and cx43 and that cx26 gene expression is induced earlier by E2 and is likely to be more sensitive to a shift in the E2 to P ratio than cx43.

Differential expression of gap junction connexins in endocrine and exocrine glands.

We have investigated the expression of three gap junction proteins and their corresponding mRNAs by secretory cells of a variety of endocrine and exocrine rat glands. By immunostaining cryostat sections (indirect immunofluorescence) with antibodies against connexins (Cx) 26, 32, and 43 and by hybridizing total glandular RNA (Northern blot) with cRNAs for these proteins, we have found that several endocrine glands (pituitary, parathyroid, pancreatic islets, and adrenal) express Cx43, variable levels of Cx26, and no Cx32, whereas several exocrine glands (lacrimal gland, salivary glands, pancreas, prostate, and seminal vesicle) express high levels of Cx32 and variable levels of Cx26, but no Cx43. Thus, different sets of proteins comprise the gap junctions of endocrine and exocrine glands. Together with the findings that an endocrine gland (thyroid) that discharges secretory products extracellularly before releasing them in the vascular compartment expresses both Cx43 and Cx32 and that an exocrine gland (preputial gland) that has a pheromonal role expresses Cx43, these observations suggest that the differential expression of gap junction connexins may be required to specify the endocrine or exocrine differentiation of a secretory cell.

Identification of a rat liver cDNA and mRNA coding for the 28 kDa gap junction protein.

By screening of a rat liver cDNA library with complex and deoxyinosine containing oligonucleotide probes a cDNA clone was isolated and shown by sequencing to code for the amino-terminal half of the rat liver 28 kDa gap junction protein. The insert hybridized to a 1.9 kb species from rat and mouse liver poly(A)+ RNA in Northern blot analysis. In embryonic mouse hepatocytes the amount of the 1.9 kb mRNA increased 3-fold between 24 and 96 h in culture. This correlates with the previously described increase of the 28 kDa gap junction protein under these conditions.