Cloning of a Gap Junctional Protein from Vascular Smooth Muscle and Expression in Two-cell Mouse Embryos*

construction of two mutant forms of connexin43, we show that the formation of cell-cell connections does not depend upon a predicted cytoplasmic domain of the protein within 98 residues of the carboxyl terminus.

Gap junctional proteins (connexins) form aqueous channels that enable direct cell-cell transfer of ions and small molecules. The distribution and conductance of gap junction channels in cardiac muscle determine the pattern and synchrony of cellular activation. However, the capacity for smooth muscle to restrict contractile events temporally and spatially suggests that cell-cell coupling or its regulation may be decidedly different in this tissue, We isolated a cDNA from vascular smooth muscle which encodes a connexin (iK 43,187) structurally homologous to cardiac connex-in43. Vascular smooth muscle connexin43 mRNA was expressed prominently in smooth muscle tissues, cultured vascular myocytes, and arterial endothelial cells. A model for functional expression of connexins was developed in two-cell B6D2 mouse embryos. Microinjection of in vitro transcribed vascular smooth muscle connexin43 mRNA was shown to be sufficient to induce intercellular coupling in previously uncoupled blastomeres. Through the construction of two deletion mutants of connexin43, we also show that the formation of cell-to-cell connections does not depend upon a predicted cytoplasmic region within 98 residues of the carboxyl terminus. Finally, the identification of connexin43 in smooth muscle and endothelial cells provides supporting evidence for the existence of heterocellular coupling between cells of the vascular intima.
Gap junctions are specialized regions of cell contact with a distinct morphological appearance which are prevalent in a number of diverse tissues (1, 2). These membrane structures contain aqueous channels connecting adjoining cells that enable direct cell-cell transfer of ions and small molecules (1, 2). The most widely held model of the gap junction channel is a hexameric configuration of identical protein subunits (connexins) surrounding a central pore which is aligned with an identical structure in the adjacent cell membrane (3,4). Despite the ultrastructural similarity of gap junctions among * This study was supported in part by National Institutes of Health Grants HL01451 (to J. A. L.) and HL01303 (to M. L. P.), Grantsin-Aid from the American Heart Association, the Krannert Charitable Trust, and Methodist Hospital of Indiana, Inc. 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 USC. Section 1734 solelv to indicate this fact. various tissues, both functional and immunological evidence has suggested that there is structural diversity of connexins from different tissues (5-8). Recently this reasoning has been confirmed by direct comparison of the sequences of cDNAs isolated from dissimilar tissues. The deduced amino acid sequences of connexins from heart (connexin43), liver (connexin32 and connexin26), and embryonic cells (connexin38) have demonstrated clearly the structural heterogeneity of these protein subunits (9-14). However, the functional correlates of this structural heterogeneity of the various connexins is only starting to be explored (15). The significance imparted by the expression of a particular connexin to the function of both an individual cell and the entire organ remains unresolved.
In cardiac and smooth muscle, gap junctions coordinate contraction by providing low-resistance pathways necessary for cell-to-cell propagation of action potentials. Nevertheless, the capacity for smooth muscle to restrict both temporally and spatially contractile events, such as during vasospasm or peristalsis, suggests that cell-cell coupling or its regulation may be decidedly different in this tissue than that which occurs in the heart. In this regard, Beyer et al. (9) have shown by Northern hybridization that connexin43, identified and isolated from heart tissue, is also expressed in the uterus, but that connexin32 is expressed in the stomach. Although gap junctions have been detected ultrastructurally and functionally in vascular smooth muscle (16), there have been no reports of the identity or structural characteristics of connexins in this tissue. The importance of gap junctions to smooth muscle function extends not only to the contractile activity of the blood vessel, but also to areas such as direct endothelialsmooth muscle signaling, angiogenesis, and modulation of smooth muscle cell phenotype (17)(18)(19)(20).
In this study, we constructed an oligonucleotide probe for screening a vascular smooth muscle cDNA library based on the observation that deduced protein sequences of connexins share highly conserved membrane-spanning domains. We report the nucleotide and deduced amino acid sequence of a vascular smooth muscle connexin virtually identical to cardiac connexin43 and provide evidence for mRNA expression in vascular smooth muscle, gastric smooth muscle, cultured vascular myocytes, and endothelial cells. We also demonstrate that microinjection of in vitro transcribed mRNA is sufficient to induce intercellular coupling in previously uncoupled twocell mouse embryos. Finally, through the construction of two mutant forms of connexin43, we show that the formation of cell-cell connections does not depend upon a predicted cytoplasmic domain of the protein within 98 residues of the carboxyl terminus. Sequence Analysis-Inserts identified in a bovine aortic smooth muscle X gtl0 library (21) were subcloned into the EcoRI restriction site of Bluescript plasmid (Stratagene). A full-length cDNA insert uncut at an internal EcoRI site was obtained by limited EcoRI digestion and also subcloned in Bluescript. Exonuclease III deletions were constructed and both strands sequenced by the dideoxy chain termination method (22) using Sequenase (United States Biochemical Corp., Cleveland, OH). Internal oligonucleotide primers were used to complete any gaps. Northern Hybridization-Total RNA isolated from fresh tissues or cultured vascular myocytes and endothelial cells (21, 22) was separated by formaldehyde gel electrophoresis and transferred to nitrocellulose membranes (Schleicher & Schuell). Duplicate lanes of each sample were stained with ethidium bromide for localization of the ribosomal subunits. A probe consisting of the LO-kb' EcoRl restriction fragment of GJ3 (containing -90% of the coding region for VSM connexin43) was gel purified and labeled (>lO' cpm/pg) by random priming (Promega Biotec, Madison, WI). Hybridization and washing of nitrocellulose filters was performed as described (22) with a final wash stringency of 0.25 X SSC at room temp. Autoradiograms were exposed at -70 "C using an intensifying screen.
In Vitro Transcription and Translation-mRNA was produced utilizing T7 RNA polymerase and capped with 5'7meGppp5'G following the directions of the manufacturer (Stratagene). In vitro translation of mRNA in the presence of [35S]methionine with rabbit reticulocyte lysate (Promega) was performed according to the manufacturer's recommendations. Protein products were subjected to denaturing discontinuous polyacrylamide gel electrophoresis (23) and visualized by autoradiography with the addition of Entensify (Du Pont-New England Nuclear) to the gel.
Embryo Microinjection-Female mice (B6D2F1; Harlan Sprague-Dawley) were superovulated with intraperitoneal injections of 5 IU of pregnant mare serum gonadotropin followed 48 h later by 5 IU of human chorionic gonadotropin. Females were paired with known fertile males and examined 24 h later for confirmation of mating. Fertilized eggs were recovered from the ampulla of the oviduct (24), and adherent cumulus cells were removed by brief treatment with 1 mg/ml of hyaluronidase (Sigma). Embryos were washed in TALP-HEPES medium (25) supplemented with 4 mg/ml of bovine serum albumin (THl) then manipulated on the stage of an inverted microscope in 20-~1 drops of THl under paraffin oil. Microelectrodes were fabricated from acid-washed glass that was rinsed further with diethylpyrocarbonate (0.2%, v/v) and then autoclaved to inactivate ribonucleases. The mRNA was synthesized in vitro and pressure-injected (200 kPa) through micropipettes with a tip diameter of 0.1-0.2 Wm. Embryos surviving injection (80%) were transferred to Whitten's medium (26) and cultured overnight at 37 "C in an atmosphere of 5% CO*, 5% 02, 90% NZ. After 20-24 h in culture, one blastomere of the two-cell embryos was pressure-injected with the fluorescent dye, Lucifer Yellow (Mr 457). The presence or absence of dye transfer to the uninjected cell was observed for ~45 min before scoring the results.

RESULTS
A X gtl0 cDNA library prepared from bovine aorta smooth muscle (21) was screened with an oligonucleotide probe (5'-CAGCCACACCTTCCCTCCAGCCGTGGAGTA-3') complementary to a portion of the first predicted transmembrane region of the published cDNA sequence of rat cardiac connexin43 (9). Three clones were identified from screening -2.5 x lo5 plaques. Each clone yielded insert fragments of -1.9 and -1.0 kb when excised from X phage DNA by EcoRI, likely the result of library amplification. Southern hybridization of the oligonucleotide probe with the two restriction fragments suggested that the l.O-kb fragment contained the coding region for the protein. The full-length, uncut insert obtained by limited EcoRI digestion of the X phage DNA was utilized for subsequent sequencing and expression studies. Fig. 1 shows the nucleotide and deduced amino acid sequence of clone GJ3. The insert consisted of a total of 2932 bases, with 10 bases upstream of the ATG codon and a poly(A) tail. An EcoRI restriction site was present at position 1039. Two polyadenylation signals (AATAAA) were identified in the 3'-untranslated region at positions 2162 and 2878. An open reading frame of 1149 base pairs encoded a protein containing 383 amino acids with M, of 43,187. The protein sequence exhibited extensive homology with connexin43 from rat cardiac muscle (9), differing by only nine amino acids ( Predicted transmembrane domains are underlined. for alanine at residue 349. Four of nine amino acid substitutions were found in relative proximity to the EcoRI restriction site which is notably absent in rat cardiac connexin43. The deduced amino acid sequence of VSM connexin43 has four predicted membrane-spanning regions (27) which were identical with those of rat cardiac connexin43 (9). The amino acid charge distribution within the first signal-anchor sequence predicts that the amino terminus of the protein faces the cytoplasmic surface of the membrane (28). Consequently, this orientation constrains the carboxyl terminus also to project into the cytosol, in agreement with models proposed for connexins from liver and heart (8, 29-31).
fer studied 20-24 h later when the embryos had reached a two-cell stage (85% of injected survivors). Fig. 4 illustrates the results: connexin43 mRNA-injected embryos developed cell-cell coupling whereas no dye transfer occurred in uninjetted embryos. Overall, positive dye transfer occurred in 58% of connexin43 mRNA-injected embryos versus none of 24 uninjected controls and only 3 of 20 buffer-injected controls. Statistical analysis of these groups showed that the incidence of coupling was significantly higher in the mRNA-injected embryos versus either of the control groups (p < 0.005 by Hybridization of total RNA from bovine aortic smooth muscle with a probe derived from the l.O-kb EcoRI restriction fragment of clone GJ3 showed a major mRNA species hybridizing at 3.3 kb and a second band at 1.8 kb (Fig. 3). The 3.3kb mRNA was -IO-fold more abundant than the 1.8kb species as determined by densitometric scanning. A similar pattern was observed in total RNA isolated from heart and gastric smooth muscle (Fig. 3), as well as in coronary artery, pulmonary artery, and carotid artery smooth muscle (not shown). Total RNA from cultured vascular myocytes and pulmonary arterial endothelial cells demonstrated the same prominent 3.3-and 1.8-kb signals. In agreement with previous studies (9), there was no hybridization signal from connexin43 detected in liver (Fig. 3). It is of note that similar levels of connexin43 mRNA were found in smooth muscle tissues, heart, and cultured endothelial cells, whereas they were -6fold higher in cultured vascular myocytes.
Although the 3.3-kb mRNA species identified on Northern blots would be predicted to represent connexin43, the identity of the 1.8-kb signal is unknown. Localization of the 28 S and 18 S ribosomal subunits on ethidium bromide-stained gels demonstrated the 18 S subunit to be reproducibly smaller than the identified 1.8-kb signal. In addition, there was no hybridization signal from ribosomal subunits detected in total RNA from liver (see Fig. 3). Finally, selection for poly(A+) RNA did not change appreciably the relative ratio of the 3.3kb to the 1.8-kb signals (not shown).
We determined whether mRNA coding for connexin43 is sufficient to induce intercellular coupling by exploiting the finding that mouse embryonic cells (B6D2F1 strain) normally are not coupled at the two-cell stage (see Table I  Fisher's exact test); there was no statistical significance to the small difference between buffer-injected uersus uninjected controls. The intensity of dye transfer varied widely (relative fluorescence of the two blastomeres was typically -1:lO) due partly to variation in the amount of injectate and partly to differences in the extent of coupling. Dye transfer was not immediate; Lucifer Yellow filled the injected cell within seconds, whereas minutes were required for fluorescence to be visible in the adjacent blastomere.
Cell division was not essential for development of cell-cell coupling. In a few experiments, VSM connexin43 mRNA was microinjected into each cell of two-cell embryos and the presence or absence of dye transfer determined later at the same stage. The results were similar to microinjection of mRNA in one-cell embryos; 33% showed intercellular transfer of dye uerms none of the controls. Tolerance of the embryos for double injections was poor (only 6 of 20 pairs survived 24 h), and continued development limited the number of embryos that could be studied at the two-cell stage.
We also investigated the functional significance of the large predicted carboxyl-terminal cytoplasmic region of VSM connexin43 with respect to induction of cell-cell coupling. Fig.  5A shows the deduced amino acid sequence of the mutant mRNA species that were produced from a Bluescript plasmid containing the l.O-kb EcoRI restriction fragment of VSM connexin43 cDNA. The plasmid was linearized with SmaI and transcribed utilizing the T7 promoter, producing a mRNA in which the coding region was deleted for the terminal 98 amino acids of connexin43. A second mRNA was produced by linearizing the plasmid at the XbaI site within the polylinker region of the plasmid. The mRNA transcript contained the coding region for amino acids l-346 from connexin43 followed by 8 amino acids coded by the vector. In uitro translation of the mRNAs with reticulocyte lysate yielded the protein products shown in Fig. 5B despite the absence of a termination codon and all 3'-untranslated sequence in the truncated mRNAs. The apparent molecular weights by SDS-polyacrylamide gel electrophoresis of the protein products of the fulllength and truncated connexin43 mRNAs were somewhat smaller than would be predicted to be encoded by the mRNAs.  Furthermore, with progressive deletion of regions of the car-boxy1 terminus of the protein, a presumed dimeric form of the connexin was evident on the gels. The discordance between the predicted molecular weight and the observed SDS-polyacrylamide gel mobility pattern of both the monomeric and dimeric forms of connexins has been previously reported (32,33). Injection of the truncated forms of connexin43 mRNAs into fertilized eggs also resulted in the development of intercellular coupling (Fig. 5C). Cumulative results are shown in Table I. There was no significant difference (p > 0.1) in the rate of induction of cell-cell coupling after injection of the different forms of connexin43 mRNA. These results suggest that the carboxyl-terminal 98 amino acids are unnecessary for the formation of open gap junction channels in early mouse blastomeres. DISCUSSION It has long been recognized that gap junctions are the membrane regions through which ionic currents pass from cell to cell during propagation of action potentials (34,35). The distribution and conductance of gap junction channels in cardiac and smooth muscle determines the pattern and syn- Yet gap junctions are also present in other cells which are inexcitable. In these tissues, there is evidence to suggest that gap junctions provide a mechanism for transmission of small cytoplasmic molecules between cells. However, it is unknown if different functional characteristics are shared by all gap junctions or are a reflection of intrinsic structural differences in the connexin subunits comprising the gap junctions.
Recent information from analysis of connexin cDNAs from heart, liver, and embryonic cells has provided a structural basis for suggesting that variations in particular domains of the protein may be the basis for different physiological functions. In this study we have isolated and characterized a cDNA from vascular smooth muscle that encodes a gap junction protein, and we have demonstrated the ability of in vitro transcribed mRNA to induce cellular coupling. The deduced amino acid sequence of this cDNA yielded a protein of M, 43,187 which is virtually identical to the sequence of rat cardiac connexin43 (9). Eight amino acid substitutions as well as one additional residue were identified in the deduced sequence of connexin43 from vascular smooth muscle. The substitutions were all confined to predicted cytoplasmic regions of the protein.
Since most of the substitutions are conservative, it seems probable that these differences are species-related and unlikely to be of functional significance. However, the extracellular and transmembrane domains of these two proteins were conserved absolutely, not surprising given the presumed role of these regions in membrane assembly and alignment with the connexin array across the gap. Yet the observation by others of single-channel activity after reconstitution of connexins in planar bilayers (36) suggests that the capacity for assembly into functional channels is not entirely dependent on alignment with an adjoining hemichannel. There must be additional factors in vivo that lead to localization of connexins in junctional membranes or their selective opening therein. We observed that none of the twocell embryos in which coupling was induced by connexin43 mRNA showed leakage of dye across open channels in the surface membrane. Given the propensity of connexins to form channels in lipid bilayers, it appears improbable on the basis of chance alone that connexins would have inserted only into the junctional membrane region of the blastomere. Furthermore, our initial studies with deletion mutants of connexin43 suggest that substitutions within 98 residues of the carboxyl terminus have little significance with regard to induction of functional cell-cell channels. However, it is not yet known whether point mutations within the carboxyl-terminal region result in alterations in the probability of channel opening or closure which may be of considerable importance to cellular regulation of gap junctional function. Northern analysis for expression of connexin43 identified a major 3.3-kb mRNA species and a minor 1.8-kb mRNA in vascular tissue, gastric smooth muscle, heart, cultured vascular myocytes, and endothelial cells. As predicted from other studies (9), connexin43 did not hybridize to RNA from liver, which has been shown to contain connexin32 and connexin26. The identity of the 1.8-kb hybridization signal to connexin43 cDNA is conjectural. A second polyadenylation signal identified within the 3'-untranslated region could predict a potential -2.6-kb message, but this was not evident on the blots. Furthermore, the absence of a hybridization signal in liver strongly suggests that the 1.8-kb mRNA is not connexin32 or connexin26. It is possible that the 1.8-kb signal represents alternative splicing of connexin43 mRNA or a novel gene product structurally similar to connexin43. The latter is an intriguing alternative in light of the recent isolation of a second connexin cDNA from liver (14). The demonstration of expression of connexin43 in both smooth muscle and endothelial cells provides supporting evidence for the suggested physiological presence of heterocellular coupling between these two cell types in the vascular intima (16, 17). Gap junction mediated intercellular communication within the vascular intima may potentially provide integration of the ostensibly distinct processes of contraction, growth control and hormonal responsiveness. Gap junctions have been implicated in the regulation of cellular growth and differentiation during embryological development (37). There is also evidence to suggest that cell-cell coupling between normal and transformed cells in culture inhibits growth of the transformed cells (38). Finally, gap junctions appear to play a role in the coordination of endothelial migration (39). Whether gap junctions could serve to regulate the process of phenotypic modulation, proliferation and migration of smooth muscle cells in the vascular intima are unknown. Our data demonstrate that connexin43 mRNA is expressed in vascular smooth muscle cells in both "contractile" and "synthetic" phenotypes. However, currently it is unknown if the significantly higher levels of connexin43 mRNA found in cultured vascular smooth muscle cells reflect a fundamental difference in cellular coupling with modulation to the "synthetic" phenotype.
Functional evidence has been obtained for regulation of gap junctional conductance by protons (40, 41), Ca*' (41, 42), Ca*+-calmodulin (43), and CAMP (44). However, it is unknown whether these mediators alter conductance by direct interaction with the connexin or indirectly through activation of specific kinases resulting in phosphorylation of connexins (45). It is noteworthy that VSM connexin43 does not contain a consensus phosphorylation sequence for CAMP-, cGMP-, or Ca*+-calmodulin-dependent protein kinases but does have number of serine residues near the carboxyl terminus (positions 297,365,366,369,370,373,374) which may be potential protein kinase C phosphorylation sites. In fact, a role for closure of gap junction channels by activators of protein kinase C recently has been demonstrated in lacrimal glands (46).
Induction of junctional connections between pairs of mouse blastomeres offers potential advantages as a model for study of gap junctional regulation. Previous studies have shown that pairs of Xenopus oocytes may be coupled by the protein products of mRNA coding for rat liver connexin32 (47), Xenopus connexin38 (13), rat cardiac connexin43 (33, 48), and combinations of two connexins (33,48). Paired oocytes coupled by gap junctional proteins provide an expression system with large durable cells that can he used to study regulation of junctional conductance, site-directed mutagenesis, etc. (47). However, the large size of Xenopus oocytes is disadvantageous for achieving an adequate voltage clamp throughout the cell with a point source of current and makes the recording of single channel events unlikely. The two-cell mouse embryo model offers the advantage of a mammalian expression system for study of gap junctional regulation. Furthermore, the -40-50 pm diameter of mouse two-cell blastomeres (one-tenth to one-twentieth the diameter of Xenopus oocytes) may permit the recording gap junctional channels in situ because input resistance is increased by the -lOOOfold smaller cell volume.