Gap junction dynamics: reversible effects of divalent cations.

Reversible changes in gap junction structure similar to those previously seen to parallel electrical uncoupling (9, 33, 34) are produced by treating with Ca++ or Mg++ gap junctions isolated in EDTA from calf lens fibers. The changes, characterized primarily by a switch from disordered to crystalline particle packings, occur at a [Ca++] of 5 x 10(-7) M or higher and a [Mg++] of 1 x 10(-3) M or higher and can be reversed by exposing the junctions to Ca++- and Mg++-free EGTA solutions. Similar changes are obtained in junctions of rat stomach epithelia incubated at 37 degrees C in well-oxygenated Tyrode's solutions containing a Ca++ ionophore (A23187). Deep etching experiments on isolated lens junctions show that the true cytoplasmic surface of the junctions (PS face) is mostly bare, suggesting that the particles may not be connected to cytoskeletal elements. A hypothesis is proposed suggesting a mechanism of particle aggregation and channel narrowing based on neutralization of negative charges by divalent cations or H+.

Gap junction dynamics: reversible effects of hydrogen ions.

Reversible crystallization of intramembrane particle packings is induced in gap junctions isolated from calf lens fibers by exposure to 3 x 10(-7) M or higher [H+] (pH 6.5 or lower). The changes from disordered to crystalline particle packings induced by low pH are similar to those produced in junctions of intact cells by uncoupling treatments, indicating that H+, like divalent cations, could be an uncoupling agent. The freeze-fracture appearance of both control and low pH-treated gap junctions is not altered by glutaraldehyde fixation and cryoprotective treatment, as suggested by experiments in which gap junctions of both intact cells and isolated fractions are freeze-fractured after rapid freezing to liquid N2 temperature according to Heuser et al. (13). In junctions exposed to low pH, the particles most often form orthogonal and rhombic arrays, frequently fused with each other. A number of structural characteristics of these arrays suggest that the particles of lens fiber gap junctions may be shaped as tetrameres.

Bridges linking gap junction particles extracellularly: a freeze-etching rotary-shadowing study of split junctions.

The nature of the interaction between neighboring gap junction particles and the mechanism involved in particle crystallization are still unclear. We describe here interparticle bridge-like structures which could participate in the mechanism of gap junction particle aggregation and pattern determination. Gap junction membranes of rat liver, pulled apart by vascular perfusion with hypertonic sucrose, were freeze-fractured in deionized water, etched at - 100 degrees C for 8 min and rotary-shadowed at a 32 degrees angle. At the extracellular true surface of the junctions (ES-surface), the particles appear as 7.0 to 7.5 nm rings often resolvable into six radially arranged subunits. The particles appear linked to each other by filamentous bridges 1.5 to 2.2 nm thick and approximately 1.5 nm long. Some bridges (single bridges) directly interlink neighboring particles while other bridges (multiple bridges) are joined to a particle at one end and to the other bridges at the other end. Bridges are referred to as double or triple bridges if they result from the interaction of two or three bridges respectively. In particles which can be resolved into six subunits, the bridges appear to bind to the subunit tips. Stereo images indicate that the bridges lay in planes lower than the particle summits. The bridges could be either polypeptide chains of the main gap junction protein or peripheral proteins, but the unlikely possibility that they are a shadowing artifact cannot be entirely ruled out yet.

Reversible effects of heptanol on gap junction structure and cell-to-cell electrical coupling.

It has been shown that certain alkanols such as heptanol and octanol reversibly uncouple crayfish axons. We have studied the effects of heptanol on the electrical coupling in Xenopus embryo cells and on the gap junction particle distribution of stomach and pancreas epithelia. Cell-to-cell electrical communication in Xenopus embryos reversibly decreases 10-15 min after the beginning of heptanol superfusion to reach values of coupling ratio lower than 0.05. Particle density of rat stomach and pancreas gap junction reversibly increases after exposure to heptanol-Tyrode's solutions. In stomach, gap junction particles from isles with hexagonal patterns after 10 to 15 min; in pancreas crystallization occurs after 20 to 30 min. Optical diffractions are used to average particle spacings in treated and control junctions of pancreas.

Calmodulin colocalizes with connexins and plays a direct role in gap junction channel gating.

The direct calmodulin (CaM) role in chemical gating was tested with CaM mutants, expressed in oocytes, and CaM-connexin labeling methods. CaMCC, a CaM mutant with greater Ca-sensitivity obtained by replacing the N-terminal EF hand pair with a duplication of the C-terminal pair, drastically increased the chemical gating sensitivity of Cx32 channels and decreased their Vj sensitivity. This only occurred when CaMCC was expressed before Cx32, suggesting that CaMCC, and by extension CaM, interacts with Cx32 before junction formation. Direct CaM-Cx interaction at junctional and cytoplasmic spots was demonstrated by confocal immunofluorescence microscopy in HeLa cells transfected with Cx32 and in cryosectioned mouse liver. This was confirmed in HeLa cells coexpressing Cx32-GFP (green) and CaM-RFP (red) or Cx32-CFP (cyan) and CaM-YFP (yellow) fusion proteins. Significantly, these cells did not form gap junctions. In contrast, HeLa cells expressing only one of the two fusion proteins (Cx32-GFP, Cx32-CFP, CaM-RFP or CaM-YFP) revealed both junctional and non-junctional fluorescent spots. In these cells, CaM-Cx32 colocalization was demonstrated by secondary immunofluorescent labeling of Cx32 in cells expressing CaM-YFP or CaM in cells expressing Cx32-GFP. CaM-Cx colocalization was further demonstrated at rat liver gap junctions by Freeze-fracture Replica Immunogold Labeling (FRIL).

Lens junctions are communicating junctions.

Lens fibers are electrically coupled with each other and directly exchange dyes and metabolites. In most cells, this form of communication is mediated by gap junctions. Lens fibers lack typical gap junctions. The lens junctions, although morphologically similar to gap junctions, differ from them structurally, chemically and immunologically. Nevertheless, recent evidence suggests that indeed lens junctions are communicating junctions. The lens junction protein, MIP26, displays structural characteristics similar to other channel proteins. Once incorporated into liposomes it forms channels permeable to molecules as heavy as 1.5 kDa. Like other communicating junctions, lens junctions assume crystalline arrays and uncouple with Ca++. The liposome incorporated channels close with Ca++ and H+ in the presence of calmodulin (CaM). Partial loss of gating competency occurs after proteolytic cleavage of the C-terminal arm of MIP26. The need for a unique type of communicating junction in lens is unclear. A possibility is that this tissue has some special cell-to-cell transport requirements, in terms of size and/or charge of permeants, not shared by coupled cells of other tissues.

CO(2) sensitivity of voltage gating and gating polarity of gapjunction channels--connexin40 and its COOH-terminus-truncated mutant.

The CO(2) sensitivity of transjunctional voltage ( V(j)) gating was studied by dual voltage clamp in oocytes expressing mouse Cx40 or its COOH terminus (CT)-truncated mutant (Cx40-TR). V(j) sensitivity, determined by a standard V(j) protocol (20 mV V(j) steps, 120 mV maximal), decreased significantly with exposure to 30% CO(2). The Boltzmann values of control versus CO(2)-treated oocytes were: V(0) = 36.3 and 48.7 mV, n = 5.4 and 3.7, and G(j min) = 0.21 and 0.31, respectively. CO(2) also affected the kinetics of V(j)-dependent inactivation of junctional current ( I(j)); the time constants of two-term exponential I(j) decay, measured at V(j) = 60 mV, increased significantly with CO(2) application. Similar results were obtained with Cx40-TR, suggesting that CT does not play a role in this phenomenon. The sensitivity of Cx40 channels to 100% CO(2) was also unaffected by CT truncation. There is evidence that CO(2) decreases the V(j) sensitivity of Cx26, Cx50 and Cx37 as well, whereas it increases that of Cx45 and Cx32 channels. Since Cx40, Cx26, Cx50 and Cx37 gate at the positive side of V(j), whereas Cx45 and Cx32 gate at negative V(j), it is likely that V(j) behavior with respect to CO(2)-induced acidification varies depending on gating polarity, possibly involving the function of the postulated V(j) sensor (NH(2)-terminus).

Chimeric evidence for a role of the connexin cytoplasmic loop in gap junction channel gating.

Gap junction channels are regulated by gates that close upon exposure to 100% CO2, probably via an increase in intracellular Ca2+ concentration, [Ca2+]i. For defining connexin (Cx) domain(s) involved in gating, we have studied chemical and voltage gating sensitivities of channels made of Cx38, Cx32 or chimeras of the above, expressed in Xenopus oocytes. Cx38 channels are more sensitive to CO2 and voltage than those of Cx32. A 3-min exposure to 100% CO2 reduces Cx38 junctional conductance (Gj) to 0% of initial values at a maximum rate of 25%/min, whereas even a 15-min exposure to 100% CO2 reduces Cx32 Gj by approximately 50% at the slow rate of 9%/min. Of the various Cx32 mutants and Cx32/38 chimeras constructed, two chimeras (Cx32/38I and Cx32/38N) expressed functional channels. Upon exposure to CO2, channels made of Cx32/38I (Cx32 inner loop replaced with that of Cx38) reproduced precisely the uncoupling behavior of Cx38 channels in uncoupling magnitude and in both uncoupling and recoupling rates, whereas channels made of Cx32/38N (N-terminus replaced) behaved closer to Cx32 than to Cx38 channels. Cx38 channels were more voltage sensitive than those of Cx32, with V0, i.e., the transjunctional voltage at which voltage-sensitive conductance is half maximal = 35.3 and 59.5 mV, and n, i.e., equivalent gating charge = 3.3 and 2.1, respectively. Of the two chimeras, Cx32/38I channels were similar to Cx38 channels, with V0 = 40.6 mV, Gj min, i.e., the theoretical minimal normalized junctional conductance = 0.35 and n = 3.0, whereas Cx32/38 N channels displayed very low voltage sensitivity, with V0 = 84.8 mV, Gj min = 0.5 and n = 1.1. The data suggest that the inner loop plays a major role in pH and voltage gating sensitivity, but whether other domains also participate in the gating mechanism cannot be excluded.

Inhibition of calmodulin expression prevents low-pH-induced gap junction uncoupling in Xenopus oocytes.

The relationship among intracellular pH (pHi), -log10 intracellular Ca2+ concentration (pCai) and gap junctional conductance, the participation of Ca2+ stores, and the role of calmodulin in channel regulation have been studied in Xenopus oocytes, expressing the native connexin (Cx38), exposed to external solutions bubbled with 100% CO2. The time courses of pHi [measured with 2',7'-bis(2-carboxyethyl)-5,6-carboxyfluorscein (BCECF)], pCai (measured with the membrane-associated fura-C18) and junctional conductance (measured with a double voltage-clamp protocol) were compared. The data obtained confirm previous evidence for a closer relationship of junctional conductance with pCai than with pHi. Evidence for an inhibitory effect of intracellularly injected ruthenium red or 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid (BAPTA) on CO2-induced uncoupling, coupled to negative results with Ca2+-free external solutions, point to a low-pHi -induced Ca2+ release from internal stores, likely to be primarily mitochondria. The hypothesis proposing a participation of calmodulin in channel gating was tested by studying the effects of calmodulin expression inhibition by intracellular injection of oligonucleotides antisense to the two calmodulin mRNAs expressed in the oocytes. Calmodulin mRNA was permanently eliminated in 5h. The oocytes injected with the antisense nucleotides progressively lost the capacity to uncouple with CO2 within 72 h. The effect of CO2 on junctional conductance was reduced by approximately 60% in 24 h, by approximately 76% in 48 h and by approximately 93% in 72 h. Oocytes that had lost gating sensitivity to CO2 partially recovered gating competency following calmodulin injection. The data suggest that lowered pHi uncouples gap junctions by a Ca2+- calmodulin-mediated mechanism.

Is the chemical gate of connexins voltage sensitive? Behavior of Cx32 wild-type and mutant channels.

Connexin channels are gated by transjunctional voltage (Vj) or CO2 via distinct mechanisms. The cytoplasmic loop (CL) and arginines of a COOH-terminal domain (CT1) of connexin32 (Cx32) were shown to determine CO2 sensitivity, and a gating mechanism involving CL-CT1 association-dissociation was proposed. This study reports that Cx32 mutants, tandem, 5R/E, and 5R/N, designed to weaken CL-CT1 interactions, display atypical Vj and CO2 sensitivities when tested heterotypically with Cx32 wild-type channels in Xenopus oocytes. In tandems, two Cx32 monomers are linked NH2-to-COOH terminus. In 5R/E and 5R/N mutants, glutamates or asparagines replace CT1 arginines. On the basis of the intriguing sensitivity of the mutant-32 channel to Vj polarity, the existence of a "slow gate" distinct from the conventional Vj gate is proposed. To a lesser extent the slow gate manifests itself also in homotypic Cx32 channels. Mutant-32 channels are more CO2 sensitive than homotypic Cx32 channels, and CO2-induced chemical gating is reversed with relative depolarization of the mutant oocyte, suggesting Vj sensitivity of chemical gating. A hypothetical pore-plugging model involving an acidic cytosolic protein (possibly calmodulin) is discussed.

Is calmodulin involved in the regulation of gap junction permeability?

The cell-to-cell channels of gap junctions mediate the direct exchange of ions and small metabolites between neighboring cells. A number of studies have shown that these channels close when the intracellular free calcium or hydrogen concentration increases, the result being cell-to-cell uncoupling. Since most of the calcium-activated biological phenomena are mediated by calmodulin (CaM), an obvious question is whether or not CaM is involved in the mechanism of cell coupling regulation. Data from the present study, showing the inhibitory effects of a calmodulin blocker on electrical uncoupling in Xenopus embryo cells, suggest a possible CaM participation in the uncoupling mechanism.

Chemical gating of gap junction channels.

Chemical gating of gap junction channels is a complex phenomenon that may involve intra- and intermolecular interactions among connexin domains and a cytosolic molecule (calmodulin?) that may function as channel plug. This article focuses on the methodology we have employed for studying the molecular basis of chemical gating by lowered cytosolic pH. Our approach has combined molecular genetics and biophysics, using exposure to 100% CO(2) for assaying chemical gating efficiency. Chimeras of connexin 32 (Cx32) and connexin 38 (Cx38) and Cx32 mutants modified at residues of the cytoplasmic loop, the initial C-terminus domain, or both have been expressed in Xenopus oocytes, and channel expression and gating have been tested electrophysiologically by double voltage clamp. In addition, various channel forms, including homotypic, heterotypic, and heteromeric channel combinations, have been evaluated for chemical gating sensitivity.

Slow gating of gap junction channels and calmodulin.

Certain COOH-terminus mutants of connexin32 (Cx32) were previously shown to form channels with unusual transjuctional voltage (V(j)) sensitivity when tested heterotypically in oocytes against Cx32 wild type. Junctional conductance (G(j)) slowly increased by severalfold or decreases to nearly zero with V(j) positive or negative, respectively, at mutant side, and V(j) positive at mutant side reversed CO(2)-induced uncoupling. This suggested that the CO(2)-sensitive gate might be a V(j)-sensitive slow gate. Based on previous data for calmodulin (CaM) involvement in gap junction function, we have hypothesized that the slow gate could be a CaM-like pore plugging molecule (cork gating model). This study describes a similar behavior in heterotypic channels between Cx32 and each of four new Cx32 mutants modified in cytoplasmic-loop and/or COOH-terminus residues. The mutants are: ML/NN+3R/N, 3R/N, ML/NN and ML/EE; in these mutants, N or E replace M105 and L106, and N replace R215, R219 and R220. This study also reports that inhibition of CaM expression strongly reduces V(j) and CO(2) sensitivities of two of the most effective mutants, suggesting a CaM role in slow and chemical gating.

Calmodulin directly gates gap junction channels.

Cytosolic changes control gap junction channel gating via poorly understood mechanisms. In the past two decades calmodulin participation in gating has been suggested, but compelling evidence for it has been lacking. Here we show that calmodulin indeed is associated with gap junctions and plays a direct role in chemical gating. Expression of a calmodulin mutant with the N-terminal EF hand pair replaced by a copy of the C-terminal pair dramatically increases the chemical gating sensitivity of gap junction channels composed of connexin 32 and decreases their sensitivity to transjunctional voltage. The increased chemical gating sensitivity, most likely because of the higher overall Ca(2+) binding affinity of this mutant as compared with native calmodulin, and the decreased voltage sensitivity are only observed when the mutant is expressed before connexin 32. This indicates that the mutant, and by extension native calmodulin, must interact with connexin 32 before gap junctions are formed. Immunofluorescence data suggest further that this interaction leads to incorporation of native or mutant calmodulin into the connexon as an integral regulatory subunit.

Is the voltage gate of connexins CO2-sensitive? Cx45 channels and inhibition of calmodulin expression.

The sensitivity of Cx45 channels to CO2, transjunctional voltage ( V(j)) and inhibition of calmodulin (CaM) expression was tested in oocytes by dual voltage clamp. Cx45 channels are very sensitive to V(j) and close with V(j) preferentially by the slow gate, likely to be the same as the chemical gate. With a CO2-induced drop in junctional conductance ( G(j)), both the speed of V(j)-dependent inactivation of junctional current ( I(j)) and V(j) sensitivity increased. With 40-mV V(j)-pulses, the tau of single exponential I(j) decay reversibly decreased by;40% during CO2 application, and G(j steady state)/G(j peak) decreased multiphasically, indicating that both kinetics and V(j) sensitivity of chemical/slow V(j) gating are altered by changes in [H(+)](i) and/or [Ca(2+)](i). CaM expression was inhibited with oligonucleotides antisense to CaM mRNA. With 15 min CO2, relative junctional conductance ( G(jt)/ G(jt0)) dropped to 0% in controls, but only by;17% in CaM-antisense oocytes. Similarly, V(j) sensitivity was significantly lessened in CaM-antisense oocytes. The data indicate that both the speed and sensitivity of V(j)-dependent inactivation of the junctional current of Cx45 channels are affected by CO2 application, and that CaM plays a key role in channel gating.