Gap junction structures: Analysis of the x-ray diffraction data

Models for the spatial distribution of protein, lipid and water in gap junction structures have been constructed from the results of the analysis of X-ray diffraction data described here and the electron microscope and chemical data presented in the preceding paper (Caspar, D. L. D., D. A. Goodenough, L. Makowski, and W.C. Phillips. 1977. 74:605-628). The continuous intensity distribution on the meridian of the X-ray diffraction pattern was measured, and corrected for the effects of the partially ordered stacking and partial orientation of the junctions in the X-ray specimens. The electron density distribution in the direction perpendicular to the plane of the junction was calculated from the meridional intensity data. Determination of the interference function for the stacking of the junctions improved the accuracy of the electron density profile. The pair-correlation function, which provides information about the packing of junctions in the specimen, was calculated from the interference function. The intensities of the hexagonal lattice reflections on the equator of the X-ray pattern were used in coordination with the electron microscope data to calculate to the two-dimensional electron density projection onto the plane of the membrane. Differences in the structure of the connexons as seen in the meridional profile and equatorial projections were shown to be correlated to changes in lattice constant. The parts of the junction structure which are variable have been distinguished from the invariant parts by comparison of the X-ray data from different specimens. The combination of these results with electron microscope and chemical data provides low resolution three- dimensional representations of the structures of gap junctions.     

Electron crystallographic methods for investigating gap junction structure

Gap junctions are clusters of closely packed intercellular membrane channels embedded in the plasma membranes of two adjoining cells. The central pore of the membrane channels serves as a conduit between cell cytoplasms for molecules less than 1000 Da in size. Advances in the purification of gap junctions and electron cryocrystallography and computer reconstruction techniques have produced new insights into the intercellular channel structure. Methods are described here for the purification of gap junction membranes, biochemical treatments to produce hemichannel layers ("split junctions"), assessment of the purity of gap junction preparations, electron cryomicroscopy, image processing and reconstruction, three-dimensional visualization, and interpretation. The critical step in electron crystallographic structure determination remains the isolation of crystalline material in sufficient and pure quantities for recording of electron microscope images. Along with sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting, the quality of gap junction purification is assessed using electron microscopy of negatively stained preparations. Electron microscopy is also used to assess the crystallinity of the purified gap junctions and split junctions. Electron cryocrystallography is a powerful technique for high-resolution structural characterization. Image processing is used to combine and enhance two-dimensional images. Electron crystallographic analysis is used to generate a three-dimensional structure from a set of electron micrographs. This three-dimensional information is extracted from a set of images recorded after tilting the specimen in the electron microscope stage and recombined using Fourier analysis techniques analogous to those used in X-ray crystallography. Computer modeling of the three-dimensional gap junction structures is a useful tool for analyzing hemichannel docking.     

Gap junction structures: V. Structural chemistry inferred from X-ray diffraction measurements on sucrose accessibility and trypsin susceptibility

X-ray diffraction patterns have been recorded from partially oriented specimens of gap junctions isolated from mouse liver and suspended in sucrose solutions of different concentration and thus of different electron density. Analysis of these diffraction patterns has shown that sucrose is excluded from the 6-fold rotation axis of the junction lattice for a length of about 100 Å. This indicates that the aqueous channel of the junctions is in the closed, high resistance state in these preparations. Mapping of the sucrose-accessible space in the junction indicates that the cross-sectional area of the channel entrance on the cytoplasmic side of the membrane could be up to five times larger than the area of the transmembrane channel. Sucrose does not penetrate more than 20 Å into the membrane along the channel. Apparently the aqueous channel, 8 to 10 Å in radius for most of its length, is narrowed or blocked by a small feature about 50 Å from the center of the gap. Very close interactions exist between the gap junction protein and the lipid polar head groups on the cytoplasmic surface of the membrane. In this region, the protein intercalates between the polar head groups. These results suggest that the gap junction protein may have a functional two-domain structure. One domain, with a molecular weight of about 15,000, spans one bilayer and half of the gap and is contained largely within a radius of 25 Å from the 6-fold axis. The second domain is smaller and occupies the cytoplasmic surface of the gap junction membrane. Trypsin digestion removes about 4000 M_rmr from the cytoplasmic surface domain of the junction protein. Most of the material susceptible to trypsin digestion is located more than 28 å from the 6-fold axis.     

COEXISTENCE OF GAP AND SEPTATE JUNCTIONS IN AN INVERTEBRATE EPITHELIUM

The intercellular junctions of the epithelium lining the hepatic caecum of Daphnia were examined. Electron microscope investigations involved both conventionally fixed material and tissue exposed to a lanthanum tracer of the extracellular space. Both septate junctions and gap junctions occur between the cells studied. The septate junctions lie apically and resemble those commonly discerned between cells of other invertebrates. They are atypical in that the high electron opacity of the extracellular space obscures septa in routine preparations. The gap junctions are characterized by a uniform 30 A space between apposed cell membranes. Lanthanum treatment of gap junctions reveals an array of particles of 95 A diameter and 120 A separation lying in the plane of the junction. As this pattern closely resembles that described previously in vertebrates, it appears that the gap junction is phylogenetically widespread. In view of evidence that the gap junction mediates intercellular electrotonic coupling, the assignment of a coupling role to other junctions, notably the septate junction, must be questioned wherever these junctions coexist.     

THE ISOLATION OF MOUSE HEPATOCYTE GAP JUNCTIONS : Preliminary Chemical Characterization and X-Ray Diffraction

A method is reported for isolating a preparation of hepatic gap junctions from the mouse. The method involves a collagenase digestion, treatment with the detergent Sarkosyl NL-97, and ultrasonication, followed by sucrose gradient ultracentrifugation. A run with 36 animals yields 0.1–0.5 mg protein. Electron microscopy with thin-sectioning and negative staining techniques reveals that the final pellet is a very pure preparation of gap junctions, accompanied by a small amount of amorphous contamination. Polyacrylamide-gel electrophoresis of sodium dodecyl sulfate (SDS)-solubilized material shows one major protein in the junction, with an apparent mol wt of 20,000, and two minor components. Thin-layer chromatography demonstrates one major and one minor phospholipid, and some neutral lipid. Low-angle X-ray diffraction of wet and dried specimens show reflections which index on an 86 A center-to-center hexagonal lattice, corresponding closely to electron microscope data. Dried specimens also show a lamellar diffraction, corresponding to the total profile thickness of the junction (150 A).     

Reflexive gap junctions. Gap junctions between processing arising from the same ovarian decidual cell

Ajacent processes on ovarian decidual cells were shown by electron microscopy to form gap junctions with one another. Micrographs of tissues preserved with lanthanum included in the fixative confirm the hexagonal array and 2-4 nm gap which characterize gap junctions. It is suggested that these gap junctions may play a role in the process of merocrine secretion from the peduncular processes of ovarian decidual cells. The term reflexive gap junction is introduced to describe gap junctions between adjacent processes from the same cell.     

Isolation and characterization of gap junctions from tissue culture cells.

The purification of membrane proteins in a form and amount suitable for structural or biochemical studies still remains a great challenge. Gap junctions have long been studied using electron microscopy and X-ray diffraction. However, only a limited number of proteins in the connexin family have been amenable to protein or membrane purification techniques. Molecular biology techniques for expressing large gap junctions in tissue culture cells combined with improvements in electron crystallography have shown great promise for determining the channel structure to better than 10 A resolution. Here, we have isolated two-dimensional (2D) gap junction crystals from HeLa Cx26 transfectants. This isoform has never been isolated in large fractions from tissues. We characterize these preparations by SDS-PAGE, Western blotting, negative stain electron microscopy and atomic force microscopy. In our preparations, the Cx26 is easily detected in the Western blots and we have increased expression levels so that connexin bands are visible on SDS-PAGE gels. Preliminary assessment of the samples by electron cryo-microscopy shows that these 2D crystals diffract to at least 22 A. Atomic force microscopy of these Cx26 gap junctions show exquisite surface modulation at the extracellular surface in force dissected gap junctions. We also applied our protocol to cell lines such as NRK cells that express endogenous Cx43 and NRK and HeLa cell lines transfected with exogenous connexins. While the gap junction membrane channels are recognizable in negatively stained electron micrographs, these lattices are disordered and the gap junction plaques are smaller. SDS-PAGE and Western blotting revealed expression of connexins, but at a lower level than with our HeLa Cx26 transfectants. Therefore, the purity and morphology of the gap junction plaques depends the size and abundance of the gap junctions in the cell line itself.     

Low resistance junctions in crayfish. Structural changes with functional uncoupling

Electrical uncoupling of crayfish septate lateral giant axons is paralleled by structural changes in the gap junctions. The changes are characterized by a tighter aggregation of the intramembrane particles and a decrease in the overall width of the junction and the thickness of the gap. Preliminary measurements indicate also a decrease in particle diameter. The uncoupling is produced by in vitro treatment of crayfish abdominal cords either with a Ca++, Mg++-free solution containing EDTA, followed by return to normal saline (Van Harreveld's solution), or with VAn Harreveld's solution containing dinitrophenol (DNP). The uncoupling is monitored by the intracellular recording of the electrical resistance at a septum between lateral giant axons. The junctions of the same septum are examined in thin sections; those of other ganglia of the same chain used for the electrical measurements are studied by freeze-fracture. In controls, most junctions contain a more or less regular array of particles repeating at a center to center distance of approximately 200 A. The overall width of the junctions is approximately 200 A and the gap thickness is 40-50 A. Vesicles (400-700 A in diameter) are closely apposed to the junctional membranes. In uncoupled axons, most junctions contain a hexagonal array of particles repeating at a center to center distance of 150-155 A. The overall width of the junctions is approximately 180 A and the gap thickness is 20-30 A. These junctions are usually curved and are rarely associated with vesicles. Isolated, PTA-stained junctions, also believed to be uncoupled, display similar structural features. There are reasons to believe that the changes in structure and permeability are triggered by an increase in the intracellular free Ca++ concentration. Most likely, the changes in permeability are caused by conformational changes in some components of the intramembrane particles at the gap junctions.     

Immunocytochemical localization of the main intrinsic polypeptide (MIP) in ultrathin frozen sections of rat lens

The in situ distribution of the 26-kdalton Main Intrinsic Polypeptide (MIP or MP 26), a putative gap junction protein in ocular lens fibers, was defined at the electron microscope level using indirect immunoferritin labeling of ultrathin frozen sections of rat lens. MIP was found distributed throughout the plasma membrane of the lens fiber cell, with no apparent distinction between junctional and nonjunctional membrane. MIP was not detectable in the basal or lateral plasma membrane of the lens epithelial cell, including the interepithelial cell gap junctions; nor was MIP detectable in the plasma membrane or gap junctions of the hepatocyte. Previous reports have indicated that the protein composition of the lens fiber cell junction differs from that of the hepatocyte gap junction. The evidence presented here suggests that the composition of the fiber cell junction and plasma membrane is also immunocytochemically distinct from that of its progenitor, the lens epithelial cell.     

Intercellular communication in normal and regenerating rat liver: a quantitative analysis

We have compared intercellular communication in the regenerating and normal livers of weanling rats. The electrophysiological studies were conducted at the edge of the liver, and we have found that here as elsewhere in the liver there is a dramatic decrease in the number and size of gap junctions during regeneration. The area of hepatocyte membrane occupied by gap junctions is reduced 100-fold 29-35 h after hepatectomy. By combining observations made with the scanning electron microscope with our freeze fracture data we have estimated the number of "communicating interfaces" (areas of contact between hepatocytes that include at least one gap junction) formed by hepatocytes in normal and regenerating liver. In normal liver a hepatocyte forms gap junctions with every hepatocyte it contacts (approximately 6). In regenerating liver a hepatocyte forms detectable gap junctions with, on average, only one other hepatocyte. Intercellular spread of fluorescent dye and electric current is reduced in regenerating as compared with normal liver. The incidence of electric coupling is reduced from 100% of hepatocyte pairs tested in control liver to 92% in regenerating liver. Analysis of the spatial dependence of electronic potentials indicates a substantial increase in intercellular resistance in regenerating liver. A quantitative comparison of our morphological and physiological data is complicated by tortuous pattern of current flow and by inhomogeneities in the liver during regeneration. Nevertheless we believe that our results are consistent with the hypothesis that gap junctions are aggregates of channels between cell interiors.     

Intercellular communication in a positional field. Ultrastructural correlates and tracer analysis of communication between insect epidermal cells.

The junctional membrane in the epidermal cells of the larval beetle (Tenebrio molitor L.) is comprised of macular gap junctions embedded in septate junctions. Ultrastructural and morphometric analysis of the distribution of gap junctions within the segmental epidermis suggests that this junction alone could account for the high electrotonic coupling recorded for the epidermal sheet. Analysis of the lanthanum-impregnated septate junction makes it doubtful that this junction serves as a communicating channel between beetle cells. A new model for the septate junction is presented in which pleated septa, less than 30 A thick, connect adjacent plasma membranes; the septa themselves are interconnected by two interseptal platforms that are coplanar with the plasma membranes. Iontophoretic injection of organic tracers into single epidermal cells suggests that only molecules of less than MW 1000 can transfer between cells through low-resistance junctions.     

A detergent-independent procedure for the isolation of gap junctions from rat liver

In this paper, the isolation of rat liver gap junctions from alkali-extracted rat liver plasma membranes is described. The purification is significantly more rapid than the commonly used detergent-based approaches and is subject to less variability. The gap junctions isolated by this method are comprised of a 27,000-Da polypeptide previously identified as the major gap junction polypeptide. The isolated gap junctions have the characteristic double-membrane organization and subunit structure observed in vivo. The protein yield is from 8 to 10 micrograms/g of liver (wet weight), about a 10-fold increase in recovery over that of earlier isolation procedures. With the availability of increased amounts of material, antibodies were raised to the liver gap junction polypeptide. Immunofluorescence localization of these antibodies on rat liver sections revealed a distribution consistent with that expected from electron microscopic analysis of liver thin sections. Double diffusion of antibody against solubilized gap junctions in detergent-containing gels resulted in the formation of precipitin arcs, suggesting response to multiple determinants. Antibody binding to the 27,000-Da gap junction polypeptide was demonstrated by immunoblot analysis of sodium dodecyl sulfate-polyacrylamide gels containing rat liver plasma membranes and isolated gap junctions. These results confirm the identification of the 27,000-Da polypeptide as the major protein component of gap junctions.     

Effects of diminished expression of connexin43 on gap junction number and size in ventricular myocardium

Gap junction number and size vary widely in cardiac tissues with disparate conduction properties. Little is known about how tissue-specific patterns of intercellular junctions are established and regulated. To elucidate the relationship between gap junction channel protein expression and the structure of gap junctions, we analyzed Cx43 +/- mice, which have a genetic deficiency in expression of the major ventricular gap junction protein, connexin43 (Cx43). Quantitative confocal immunofluorescence microscopy revealed that diminished Cx43 signal in Cx43 +/- mice was due almost entirely to a reduction in the number of individual gap junctions (226 +/- 52 vs. 150 +/- 32 individual gap junctions/field in Cx43 +/+ and +/- ventricles, respectively; P < 0.05). The mean size of an individual gap junction was the same in both groups. Immunofluorescence results were confirmed with electron microscopic morphometry. Thus when connexin expression is diminished, ventricular myocytes become interconnected by a reduced number of large, normally sized gap junctions, rather than a normal number of smaller junctions. Maintenance of large gap junctions may be an adaptive response supporting safe ventricular conduction.     

Gap junction structures. VIII. Membrane cross-sections.

Profiles of negatively stained gap junctions have been measured by grid sectioning. After normal levels of electron irradiation, the membrane thickness shrinks to about half that of unirradiated controls, but no shrinkage occurs in the hexagonal lattice plane. Even under low irradiation conditions, there is significant thinning of the membranes. Edge views, in which rows of connexons are aligned parallel to the beam, were obtained from grid sections, folds in normal negatively stained specimens, and sections of a positively stained specimen. Averaging these micrographs with the translational and mirror symmetry of the projected lattice image displays conserved and variable features in the stain distribution of different specimens. Variations in the relative amount of negative stain in the gap at the surfaces and in the channel are uncorrelated with the irradiation but appear to depend on the local staining conditions and the integrity of the connexons. The dimensions measured from previously unirradiated grid sections, folds, and positively stained sections are in accord with x-ray diffraction measurements. Radiation-induced shrinkage can be accounted for by mass loss principally from the membrane bilayer. Disordering of the surface structure appears to be correlated with the radiation sensitivity of the bilayer; in contrast, the gap structure is well preserved under a variety of conditions.     

Gap junction structures. IV. Asymmetric features revealed by low- irradiation microscopy

Micrographs of mouse liver gap junctions, isolated with detergents, and negatively stained with uranyl acetate, have been recorded by low-irradiation methods. Our Fourier-averaged micrographs of the hexagonal junction lattice show skewed, hexameric connexons with less stain at the threefold axis than at the six indentations between the lobes of the connexon image. These substructural features, not clearly observed previously, are acutely sensitive to irradiation. After an electron dose less than that normally used in microscopy, the image is converted to the familiar doughnut shape, with a darkly stained center and a smooth hexagonal outline, oriented with mirror symmetry in the lattice. Differences in appearance among 25 reconstructed images from our low-irradiation micrographs illustrate variation in staining of the connexon channel and the space between connexons. Consistently observed stain concentration at six symmetrically related sites approximately 34 A from the connexon center, 8 degrees to the right or left of the (1, 1) lattice vector may reveal an intrinsic asymmetric feature of the junction structure. The unexpected skewing of the six-lobed connexon image suggests that the pair of hexagonal membrane arrays that form the junction may not be structurally identical. Because the projected image of the connexon pair itself appears mirror symmetric, each pair may consist of two identical connexon hexamers related by local (noncrystallographic) twofold axes in the junctional plane at the middle of the gap. All connexons may be chemically identical, but their packing in the hexagonal arrays on the two sides of the junction appears to be nonequivalent.     

Structure and biochemistry of mouse hepatic gap junctions

A new method for the isolation of gap junctions from mouse liver is described. Particular attention has been directed to minimising the effects of proteolysis during isolation. The purified membrane fragments retain the typical morphological features found in junctions of the intact liver.The junctions show two major polypeptides upon polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulphate. The apparent molecular weights are 26,000 for the more abundant species and 21,000 for the minor component. Preliminary protein chemical characterisation by fingerprint analysis suggests that the two polypeptides are structurally related. While an in vivo origin of the 21,000 molecular weight species cannot be excluded, the sensitivity of the junction proteins to proteolytic degradation in vitro suggests that the 21,000 molecular weight molecule may be a breakdown product of the major component.Image reconstruction methods applied to micrographs of negatively stained isolated junctions show that the membrane contains a close-packed hexagonal lattice of components having marked 6-fold symmetry. It is suggested that these represent hexamers of the 26,000 molecular weight protein.Lipid analysis performed on gap junctions isolated by different procedures shows that the lipid composition is strongly affected by the detergents employed during the isolation. A large amount of phopholipid, but not cholesterol, can be extracted from the structure without affecting its gross morphology. This result suggests that cholesterol is tightly bound to the junction protein and may play a role in determining the structure of the gap junction.     

THE STRUCTURAL ORGANIZATION OF THE SEPTATE AND GAP JUNCTIONS OF HYDRA

The septate junctions and gap junctions of Hydra were studied utilizing the extracellular tracers lanthanum hydroxide and ruthenium red. Analysis of the septate junction from four perspectives has shown that each septum consists of a single row of hexagons sharing common sides of 50–60 A. Each hexagon is folded into chair configuration. Two sets of projections emanate from the corners of the hexagons. One set (A projections) attaches the hexagons to the cell membranes at 80–100-A intervals, while the other set (V projections) joins some adjacent septa to each other. The septate junctions generally contain a few large interseptal spaces and a few septa which do not extend the full length of the junction. Basal to the septate junctions the cells in each layer are joined by numerous gap junctions. Gap junctions also join the muscular processes in each layer as well as those which connect the layers across the mesoglea. The gap junctions of Hydra are composed of rounded plaques 0.15–0.5 µ in diameter which contain 85-A hexagonally packed subunits. Each plaque is delimited from the surrounding intercellular space by a single 40-A band. Large numbers of these plaques are tightly packed, often lying about 20 A apart. This en plaque configuration of the gap junctions of Hydra contrasts with their sparser, more widely separated distribution in many vertebrate tissues. These studies conclude that the septate junction may possess some barrier properties and that both junctions are important in intercellular adhesion. On a morphological basis, the gap junction appears to be more suitable for intercellular coupling than the septate junction.     

Comparative analysis of the major polypeptides from liver gap junctions and lens fiber junctions

Gap junctions from rat liver and fiber junctions from bovine lens have similar septilaminar profiles when examined by thin-section electron microscopy and differ only slightly with respect to the packing of intramembrane particles in freeze-fracture images. These similarities have often led to lens fiber junctions being referred to as gap junctions. Junctions from both sources were isolated as enriched subcellular fractions and their major polypeptide components compared biochemically and immunochemically. The major liver gap junction polypeptide has an apparent molecular weight of 27,000, while a 25,000-dalton polypeptide is the major component of lens fiber junctions. The two polypeptides are not homologous when compared by partial peptide mapping in SDS. In addition, there is not detectable antigenic similarity between the two polypeptides by immunochemical criteria using antibodies to the 25,000-dalton lens fiber junction polypeptide. Thus, in spite of the ultrastructural similarities, the gap junction and the lens fiber junction are comprised of distinctly different polypeptides, suggesting that the lens fiber junction contains a unique gene product and potentially different physiological properties.     

Gap junction structures after experimental alteration of junctional channel conductance

Gap junctions are known to present a variety of different morphologies in electron micrographs and x-ray diffraction patterns. This variation in structure is not only seen between gap junctions in different tissues and organisms, but also within a given tissue. In an attempt to understand the physiological meaning of some aspects of this variability, gap junction structure was studied following experimental manipulation of junctional channel conductance. Both physiological and morphological experiments were performed on gap junctions joining stage 20-23 chick embryo lens epithelial cells. Channel conductance was experimentally altered by using five different experimental manipulations, and assayed for conductance changes by observing the intercellular diffusion of Lucifer Yellow CH. All structural measurements were made on electron micrographs of freeze-fracture replicas after quick-freezing of specimens from the living state; for comparison, aldehyde-fixed specimens were measured as well. Analysis of the data generated as a result of this study revealed no common statistically significant changes in the intrajunctional packing of connexons in the membrane plane as a result of experimental alteration of junctional channel conductance, although some of the experimental manipulations used to alter junctional conductance did produce significant structural changes. Aldehyde fixation caused a dramatic condensation of connexon packing, a result not observed with any of the five experimental uncoupling conditions over the 40-min time course of the experiments.     

Freeze-fracture analysis of gap junction disruption in rat ovarian granulosa cells

The effects of chemical dissociation on rat ovarian granulosa cell gap junctions has been studied using freeze-fracture electron microscopy. Sequential exposure of granulosa cells within follicles to solutions containing 6·8 mM EGTA [ethylene-bis-(β-aminoethyl ether)-N,N′-tetra acetic acid] and 0·5 M sucrose results in extensive cellular dissociation of the follicular epithelium. Freeze-fracture replicas made from fixed, control or EGTA-treated ovarian follicles exhibit extensive gap junctions between granulosa cells that are characterized by a range of packing order of constituent P-face particles or E-face pits. In contrast, exposure to 0·5 M sucrose containing 1·8 mM EGTA for as little as 1 min results in a consistently close packing of particles or pits which is accompanied by splitting of gap junctions between granulosa cells. The process of junction splitting was studied in detail in replicas prepared from follicles treated sequentially for various periods of time with EGTA and sucrose solutions. Initially, large gap junctions lose their regular shape and fragment into numerous tightly packed aggregates of P-face particles or E-face pits which are separated by unspecialized areas of plasma membrane. Subsequent to junction fragmentation, individual junction plaques separate at sites of cell contact and generate hemijunctions that border the intercellular space, Hemijunctions undergo particle dispersion of the P fracture face which results in an increased density of large intramembrane particles; no corresponding change in E-face pits is discernible at this stage. Morphometric analysis of replicas of tissue undergoing junction splitting indicates that junctional surface area decreases to 10–20% of control levels during this same treatment and so further supports the qualitative observations on junction fragmentation. Viabilities of granulosa cells obtained by these techniques also agree with the sequence observed in the morphometric analysis of the replicas. Finally, within 15 min after placing ovaries in isotonic, Ca²⁺-containing salt solutions, gap junction reformation occurs by aggregation of particles at sites of intercellular contact. These sites are distinguished by the appearance of short surface protrusions or indentations on their respective P and E fracture faces. The data suggest a mechanism for EGTA-sucrose mediated cellular dissociation in the follicular epithelium in which gap junctional particles are free to move in the plane of the plasma membrane and may be re-utilized to form gap junctions in the presence of extracellular calcium.