Gap junctional intercellular communication and the regulation of connexin expression and function L.S. Musil and D.A. Goodenough Department
of Anatomy
and Cellular
Biology,
Harvard
Medical
School,
Boston,
Massachusetts,
USA
Current Opinion in Cell Biology 1990, 2:875-880
Introduction
tional communication in these and other complex processes remains to be elucidated.
Gap junctions are collections of transmembrane channels which directly link the cytoplasms of adjoining cells (reviewed in Bennett and Goodenough, Neurosci Res Prog Bull 1978, 16:373-486). Found in virtually all animal tissues, gap junctions provide an almost ubiquitous solution in metazoans to the problem of cell-to-cell transfer of low-molecular-weight substances without exposure to the extracellular environment. The gap junction channel is composed of a hemichannel (connexon) embedded in the plasma membrane of one ceU joined in mirror symmetry with a connexon in the apposing ceU membrane (reviewed by Caspar et al., In Modem Cell Biology edited by Hertzberg and Johnson. Alan R Liss, 1988, pp 117-133). Each connexon is an oligomer of six integral membrane protein subunits which delineate a 1.5 m-n diameter aqueous pore permeable to inorganic ions and small metabolites. The molecular cloning of one of these proteins in 1986 (Paul, J Cell Biol 1986, 103:123-134; Kumar and Gilula, J Cell Biol 1986, 103:767-776) has led to the discovery of a widely distributed family of related gap junction proteins, the connexins. The basic structural features of connexins were reviewed in this journal last year by Stevenson and Paul (Cum Opin Cell Bioll989, l&%-891> and more recently by Beyer et al. (I Membr Bioll990,116:187-194). In this review, the regulation of three members of the connexin family will be discussed: connexin32 (Paul, 1986; Kumar and Gilula 1986), connexin26 (Zhang and Nicholson, J Cell Biol 1989, 109:3391-3401), and connexin43 (Beyer et al, J Cell Bioll987, 105:2621-2629). While passage of current-carrying ions through gap junction channels is clearly important in the activity of electrically excitable tissues such as myocardium, smooth muscle, and nerve, the functions of gap junctions in other ceU types is more obscure. Gap junctions have been implicated in embryonic development, maintaining homeostasis in avascular systems (i.e. lens, vertebrate ovarian follicle), growth control, regulation of glandular secretion, and cellular differentiation (Guthrie and Gilula, Trends Neurosci 1989, 1232-16; Mehta et al. Cell 1986, 44187-196) [l]. However, the precise role of gap junc-
A critical feature of gap junctional intercellular communication is that it can be reversibly modulated in response to extra- and intracellular events. Experimentally, gap junction-mediated cell-to-cell communication is sensitive to a wide range of stimuli (reviewed in Spray and Bennett, Ann Rev Pbysiol1985,47:281-303). The determination of the mechanisms by which this bewildering array of effectors influences cell-to-cell communication is essential for understanding the role of gap junctional communication in non-electrically excitable cells. Some of these effecters (e.g. H+, Ca2+, voltage, and heptanol) reversibly alter junctional permeability on a very rapid time scale (seconds/minutes), presumably by ‘gating’ phenomena (reviewed by Bennett et al, In Mouk-rn Cell Biology. 1988, pp 287-304). In principle, gap junctional communication could also be regulated at the level of cell-cell adhesion, connexin gene transcription and/or translation, connexin assembly, or connexin degradation and/or reutilization. This review will focus on studies published in the last year that examine the control of gap junctionmediated intercellular communication by differential connexin gene expression and by post-translational covalent processing of connexins.
Regulation
of connexin
Rene expression
The recent development of connexin cDNA probes has led to new biochemical approaches to the study of gap junction regulation. We will discuss live weU characterized examples in which modulation of the expression of a particular connexin gene is likely to have a physiologically important effect on cell-to-cell communication.
Hormonal
effects
In pregnant mammals, the period immediately prior to labor (O-24 h) is invariably marked by an increase in morphologically and physiologically recognizable gap junctions between the smooth muscle cells of the uterus
Abbreviations pk-protein
kinase;
RSV-Rous
@ Current
sarcoma
Biology
virus;
TPA-12-O-tetradecanoyl
Ltd ISSN 0955-0674
phorbol-14-acetate.
875
876
Cell-@cell
contact
and extracellular
matrix
(myometrium), a phenomenon thought’to be necessary in synchronizing electrical and contractile activities in the uterine wall during delivery (reviewed by Cole and Garlield, In Gap Junctions edited by Bennett and Spray. Cold Spring Harbor Laboratory, 1985, pp 215-230). This elaboration of uterine gap junctions has been shown to be estrogen-dependent and requires synthesis of both RNA and protein, indicating that it involves gene expression (Garfield et al, Am JPbysioll980, 239:C217-228). A recent study by Risek et al. [2] provides compelling evidence that the pre-parturition changes in myometrial gap junctions are due to a marked increase in the expression of connexin43 (ul in the authors’ nomenclature). Risek et al report a dramatic and specific increase in steadystate connexin43 mRNA levels in the myometrium of term rats, reaching a maximum (5.5 times that of non-pregnant controls) the day before parturition. Anti-connexin43 immunostaining of gap junctional plaques in frozen sections of myometrium peaked 24 h later. Within a day after delivery, both the myometrial connexin43 mRNA and gap junction immunostaining levels rapidly declined. These temporal changes in connexin43 mRNA content and immunostaining were also observed in the stroma of the ovary, another steroid-sensitive organ, but not in the heart, demonstrating tissue-specific regulation of connexin43. Hormonal changes can also lead to a decrease in connexin gene expression. Around the time of ovulation, the interfollicular cell and follicle cell-oocyte gap junctions in mammalian and amphibian ovarian follicles are lost, as is functional oocyte-follicular cell coupling (Gilula et al., J Cell Biol 1978, 78~58-75; Vilain et al., Dev Growth Dif fer 1980, 22687691). In mammals, this downregulation of gap junctions results from the preovulatoty luteinizing hormone surge and can be induced in immature rats by gonadotrophin injection (Gilula et al., 1978) and in isolated intact ovarian follicles by luteinizing hormone and/or follicle-stimulating hormone exposure (Dekel et al, Dev Bioll981, 86:356362; Moor et al., Exp Cell Res 1980, 126:15-29). In Xenopus, binding of progesterone to receptors on the oolemma has an analogous role. Gimlich et al. (31 have shown that maturation of Xenopus oocytes in vitro with progesterone leads to a specific and profound decrease (to less than 9% of control values within 3 h) in the steady-state mRNA levels of al (a Xerzo pus homologue of mammalian connexin43); compatable results were obtained when ovulation was induced in vivo by injection with human chorionic gonadotrophin. It was suggested that the al gene product may form oocyte-follicular cell gap junctions in immature follicles and that its loss at ovulation is responsible for uncoupling communication between the two cell types. The physiological significance of this uncoupling is controversial, with evidence both supporting and refuting a causal role in releasing the oocyte from meiotic arrest (discussed in Wert and Larsen, Gamete Rex 1989, 22:143-162; and Racowsky et al, Eur J Cell Bioll989, 49:244-251).
Responses physiology
to disruption
of tissue architecture
and
Revel and his colleagues have demonstrated (Yee and Revel, J Cell Biol 1978, 78:554-564; Meyer et al., J Cell Bioll981,91:505-523) that partial hepatectomy in young rats leads to marked changes in gap junction structure and function in the remaining liver lobe. Beginning about 20 h postoperatively, the size and number of morphologically recognizable liver gap junctions starts to decrease sharply, accompanied by a loss (although not complete elimination) of gap junctional intercellular communication. However, by 48 h after partial hepatectomy, morphologically recognizable gap junction levels have returned to their control values. A similar temporary downregulation of gap junctions was documented by freezefracture analysis of rat liver after bile duct ligation, which was reversed when cholestasis was relieved (Metz and Bressler, Cell Tissue Res 1979, 199:257-270). Traub et al. (Proc Nat1 Acud Sci USA 1983, 80:755-759) [4] have demonstrated that these alterations in gap junction ultrastructure are associated with changes in connexin protein levels. Using quantitative immunoblot analysis of purified rat liver plasma membrane preparations with anti-connexin antibodies, they report that connexin32 (and, to a lesser extent, connexin26) protein levels fall and then rise after partial hepatectomy in a manner temporally consistent with the morphological changes observed by Revel and coworkers (Yee and Revel, 1978; Meyer et al., 1981). Connexin32 protein concentrations also decline after bile duct ligation and recover after duct recanalization. In the case of partial hepatectomy, there is evidence that the loss of connexin32 protein is preceded by a precipitous drop in steady-state connexin32 mRNA content (to 10% of control values by 10 h after operation), indicating regulation at the mRNA rather than at the protein level (Beer et al, Cancer Res 1988,48:161&1617). Recovery from either condition (by liver regeneration or bile duct recanalization) appears to restimulate connexin synthesis, although the mechanism by which this is accomplished is unknown. Similar loss and reacquisition of connexin mRNA and protein has been documented in cultured primary embyronic mouse hepatocytes (Traub et al., Eur J Cell Bioll987, 43:48-54; Heynkes et al, FEBS Lett 1986, 205:56-60) [4], prompting Dem-tietzel et al. (J Cell Biol 1987, 105:1925-1934) to suggest a relationship between the loss of gap junctions and ceU proliferation. However, it is clear that connexin32 gene expression can also be modulated in nondividing systems such as primary cultures of adult rat hepatocytes (Spray et al., J Cell Biol1987, 105:541-551) [ 51 and during in vivo cholestasis.
Chemically
induced
hepatocarcinogenesis
Several recent studies have reported a decrease (compared with normal liver) in steady-state levels of connexin32 mRNA and in immunohistochemically detectable
Gap junctional
connexin32 protein in chemically induced rat liver neoplasms generated by various protocols (Beer et al, Cancer Res 1988, 48:161&1617; Janssen-Timmen et al., Carcinogenesis 1986, 7:1475-1482) [6]. These results are consistent with the long-standing theory that transformed cells generally (although not always> have lower levels of cell-to-cell communication than their normal counterparts, and that this defect helps to release them from the growth-suppressing effects of surrounding normal cells (Ioewenstein, Biocbim Biopkys Acta 1979, 560:1-65X However, the variability in the extent to which connexin32 mRNA levels were suppressed in the liver tumors tested, and the lack of physiological data on their junctional communication potential, makes any conclusions concerning the role of reduced connexin32 expression in hepatocarcinogenesis premature. Conclusions
Judging from the rate at which new putative connexin gene sequences are being identified, it seems very likely that there are many as yet undiscovered members of the connexin gene family (Beyer et al, 1990). This, combined with the known occurrence of multiple connexlns in many tissues and even within the same cell type, makes it diihcult to interpret the functional consequences of changes in connexin gene expression and to demonstrate that the connexin under study is indeed the one responsible for an observed effect on gap junctional communication. Techniques capable of determining the contribution of individual connexin species to cell-cell communication (i.e. anti-connexin antibody blockade, antisense connexin mRNA expression) are impractical in many systems and have potential artifacts of their own (discussed in [7] and Warner, J Cell Sci 1988, 89:1-7). Thus, in most instances, one can only demonstrate a careful correlation between a change in connexin expression detected biochemically (including assessment of both connexin mRNA and total connexin protein levels) and a change in functional connexin expression detected physiologically, which cannot be taken as proof of a cause-and-effect relationship.
Regulation
of connexin
phosphorylation
Agonists and/or antagonists of cAMP-dependent protein kinase, protein kinase C, calmodulin/Ca*+ -dependent protein kinases, and various tyrosine kinases have been shown to modulate gap junctional communication in multiple systems (reviewed in Loewenstein, Biocbem Sot Sjnnp 1985, 50143-58; Spray et al, In Modern Cell Biology. 1988, pp 227-244). Moreover, increasing the intracellular level of a particular protein kinase (by transfection of cells with kinase-encoding cDNA or by introduction of put&d kinase molecules) alters gap junctional communication in the direction predicted from such kinase effector experiments (Wiener and Loewenstein, Nuture 1983, 305:433435; Iasater, Proc Natf Acud Sci USA 1987, 84:7319-7323) [8]. In some instances, gap junc-
intercellular
communication
Musil and Goodenough
tional communication is altered within an hour of protein kinase up- or downregulation, consistent with a direct effect on phosphorylation of pre-existing proteins. Recent studies from several groups have led to the intriguing conclusion that connexins themselves are substrates for protein kinases. Serine threonine Connexin32
phosphorylation
The first reports of connexin phosphorylation in intact cells demonstrated metabolic labeling of connexin32 with [3*P] orthophosphate in primary rodent hepatocytes. Addition of 8-bromo-cAMP raised gap junctional conductance by 50-75% and resulted in a 1.6-fold lncrease in incorporation of [3*P] into immunoprecipitable connexin32 within 30-60 min (Saez et al, Proc NatlAcud Sci USA 1986, 83:24732477), apparently without appreciably increasing the total amount of connexin32 protein (Traub et al, 1987). In both basal and stimulated states, phosphotylation of connexin32 occurred almost exclusively on serine residues. Isolated rat liver gap junctions could also be phosphotylated on serine by purified cAMP-dependent protein kinase (pkA) with low stoichiometty [between 0.025 (Saez et UC, Prcc Nutf Acud Sci USA1986,83:24732477) and 0.07 [9] moles of phosphate per mole of connexin321; the extent of phosphorylation of connexin32 in intact cells is unknown. On the basis of these results, Saez et al proposed a role for pkA-mediated phosphorylation of connexin32 in gap junctional communication (see Discussion section, however). Recently, Takeda et a~! (FEBS Lett 1987,210:2681- 2688) [9] have determined that connexin32 is also a substrate for another serine/threonine kinase, protein kinase C (pkC). They demonstrated that addition of 120-tetradecanoyl phorbol-14-acetate (TPA) to intact rat hepatocytes rapidly (within 15 min) increased phosphorylation of connexin32 on serine residues by approximately 50%, an effect mimicked by other activators of pkC but not by biologically inactive phorbol esters [ 91. Moreover, purified connexin32 could be phosphorylated in vitro by pkC in a Ca*+ -dependent manner to a maximal stoichiometty of 0.21 mole per mole. Unfortunately, these studies did not address whether activation of pkC decreased cell-cell coupling between primary hepatocytes, as it does in some, but not all, cell types (Chanson et a!, Am J Pkytiol 1988, 255:C699--C704). Thus, the relationship, if any, between pkC-catalyzed phosphotylation of connexin32 and gap junctional intercellular communication remains unclear. Much the same can be said for phosphorylation of connexin32 by calmodulin/Ca*+ _ dependent protein kinase II, which has been reported in an in vitro system but whose physiological significance is unknown (Saez et al., J Cell Bioll986, 103:73a).
Connexin43 Synthesized
both in vivo and in vitro as a single, - 42 kD species, connexin43 has been shown in many cell types
877
878
Cell-to-cell
contact
and extracellular
matrix
to be post-translationally converted to t&o forms with slightly increased molecular weights (PI and P2, in the nomenclature of Musil et al [ 101) by the addition of phosphate onto serine residues [ 11,121. Although the kinases involved in connexin43 phosphotylation have not been characterized, a preliminary account indicating phosphotylation of connexin43 by pkA in isolated mammalian heart gap junctions [ Pressler and Hathaway, Circulation 1987, 76(suppl IV) [18], and the presence of several potentially phosphotylatable serine residues in the cytoplasmic tail domain of connexin43 (Beyer et al, 1987), make pkA a possible candidate. It is weU established that raising cytoplasmic CAMP concentrations rapidly and reversibly increases gap junctional conductance between myocardial cells (De MeUo, Biccbim Bio @~~s~cta 1986,888:71-79); however, whether phosphoryfation of connexin43 is also affected has not been reported. It is also unclear whether pkC, which has been implicated in the control of gap junctional communication in connexin43expressing cells (Spray and Burt, Am J P&i01 1990,258:C195-C205), participates in phosphoryfation of connexin43 under physiological conditions. A potential functional relationship between connexin43 phosphorylation and gap junctional cell-cell comrnunication was suggested by the studies of Musil ef al [lo]. They demonstrated that two cell lines known to be severely deficient in gap junctional communication (mouse S180 and L929 cells) constitutively synthesized connexin43. However, expression of connexin43 in these cells differed from that in comrnunication-competent cells in two critical respects: connexin43 accumulated intracellularIy rather than in cell surface gap junctional plaques, and was not detectably phosphotylated to the mature P2 form despite being as metabolically stable as in communication-competent cells. Conversion of S180 cells to a communication-competent phenotype by transfecting them with a cDNA encoding the interceUular adhesion molecule LCAM led to phosphorylation of connexin43 to the P2 form; furthermore, cells that are usually communication-competent, treated with known inhibitors of gap junction permeability, no longer detectably processed connexin43 to the P2 form. Taken together, these results establish a strong correlation between the ability of cells to phosphorylate connexin43 to the P2 form and to mediate gap junctional communication, but do not prove that the two events are causally linked. The demonstration of connexins in cells lacking morphologically and physiologically recognizable gap junctions has three inter-related implications for the control of gap junction-mediated intercellular communication. First, it illustrates that a change in the number and/or size of gap junctions detected morphologically does not necessarily reflect a change in connexin gene expression, emphasizing the importance of ana@ing both the steady-state connexin protein and mRNA levels before such a claim is made. Second, the presence of connexin ~RNA and protein does not mean that a ceU possesses functioning intercellular channels. Third, it raises the possibility that at least in some instances a rise in intracellular second
messenger levels (e.g. CAMP) converts communicationdeficient cells to a communication-competent phenotype not by stimulating new gene expression but by inducing connexin phosphorylation. Tyrosine
phosphorylation
Infection of various ceU lines with the oncogenic Rous sarcoma virus (RSV) leads to a dramatic reduction in gap junction-mediated intercellular dye coupling. Studies using temperature-sensitive RSV mutants demonstrated that the decrease in junctional permeability was detectable within l5rni1-1 at the permissive temperature, did not require protein synthesis or involve a change in the number of morphologically recognizable gap junctions, and was equally rapidly reversible upon shift to the non-permissive temperature (Atkinson et al., J Cell Biol 1981, 91:573-578; Azarnia and Loewenstein, J Membr Bioll984, 82:191-205). These effects on ceUceU communication were also obsened in cells transfected with the v-STC oncogene of RSV and were shown to be critically dependent on the protein tyrosine kinase activity of the v-src gene product, pp60v-m (Chang et al, Proc Nat1 Acud Sci USA 1985,82:536&5364; Azarnia et al., Science 1988, 239398-401). It is therefore of great interest that connexin43 has recently been shown to be phosphotylated on tyrosine (as weU as on serine) residues in cells expressing ppGO~-m [ 11,131. A cause-and-effect relationship between tyrosine phosphorylation of connexin43 and the decrease in junctional permeability induced by p~60~.~C is suggested by the studies of Swenson et al. [13]. Using an expression system consisting of paired Xenopus oocytes programmed to synthesize connexin43, connexin32, and/or pp60v-sTc by microinjection of their respective mRNAs, Swenson et al. demonstrated that the degree of junctional coupling between connexin43-expressing oocyte pairs is reduced lOO-lOOO-fold in the presence of pp60v-TC. In contrast, junctional permeability between connexin32-expressing oocytes is only marginally affected by pp6Ov-TC (to approximately one-third that of oocytes injected with connexin32 mRNA alone). Phosphoamino acid analysis revealed that pp60v-rC induces tyrosine phosphorylation of connexin43 but not of connexin32, consistent with the presence of a tyrosine (residue 265) in the cytoplasmic tail domain of rat connexin43 that is lacking in rat connexin32. Mutation of Tyr2(j5 to a phenylalanine residue abolishes both tyrosine phosphotyfation of connexin43 and the inhibition of gap junctional communication by ppbO”-PC. The simplest interpretation of these results is that phosphorylation of connexin43 by pp6Ov-TC is clrectly responsible for the decrease in junctional permeability induced by RSV, although the involvement of additional pathways cannot be ruled out (see [13,14]).
Discussion
Recent research in the gap junction Eeld has shed considerable light on the subject of the regulation of gap
Cap iunctional
junction-mediated intercellular communication. Determining the mechanism by which a particular effector influences gap junctional communication can, however, be extremely complicated for several reasons. Some agents (e.g. CAMP and perhaps retinoids and pkC agonists) act on more than one level, and the same agent can have opposite effects on gap junctional communication in different cell types (Saez et al, 1986; Teranishi et al., Nu0.ue 1983, 301:243246) or in the same cell type at different concentrations [ 151. Furthermore, Merent connexins respond to junctional perturbants in distinct ways [13], and the signal transduction pathways used by various transcriptional control (Hoeffler et al, A401Endo crinoll989,3:868-880) and protein kinase (Cohen, EurJ Biochem 1985,151:439-448; Macara et al, Proc NatlAcad Sci USA 1984, 81:2728-2732) systems are closely interlinked and can interact on many levels. Among the many questions raised by these studies, the following two stand out as major topics for the future investigation.
How is connexin
gene expression
regulated?
Resolution of this issue clearly requires characterization of the untranslated regions of connexin genes. Recently, Miller et al (Biosci Rep 1988, 8:455-464) examined the structure of the rat connexin32 gene and found sequences homologous to CAMP response elements in the 5’ non-coding region. Although the functionality of these elements remains to be determined, it is possible that in at least some instances agents that elevate CAMP levels exert their long-term effects on gap junctional communication by increasing the transcription rate of the connexin32 gene. The apparent sensitivity of connexin gene expression to progesterone and estrogen [2] warrants a search for steroid hormone response elements in connexin genes as well. Differential regulation of two connexins in the same tissue [2,3] may therefore reflect heterogeneity in the types of transcriptional response elements present in different connexin genes.
What is the functional role of connexin phosphorylation in gap junctional communication? The observation that CAMP increases both connexin32
phosphotylation and gap junctional conductance in rat hepatocytes led Saez et al. (1986) to speculate that connexin phosphotylation is involved in the gating of gap junction channels. However, no cause-and-effect relationship between these two phenomena has been established in this or in any other system. Proof of an effect of connexin phosphorylation on channel gating will most likely require reconstitution of functional gap junction channels into an artificial bilayer and the demonstration that controlled connexin phosphorylation either up- or downregulates channel permeability. Phosphorylation of connexins could also influence gap junctional communication by affecting non-gating events such as connexin assembly, intracellular transport, or degradation. Evaluation of these possibilities awaits more detailed characterization of connexin biosynthesis, as well as identification
intercellular
communication
Musil and Coodenouah
and site-directed mutagenesis of the serine residues modilied in connexin32 and connexin43
Acknowledgements We thank Drs D Paul, R Bruzzone, K Swenson, R Gimlich. and C Codes for helpful discussions. This work was supported by grants #GM18974 and EYO2430 to DA Goodenough
Annotated reading l l *
references
and recommended
Of interest Of outstanding interest
1. *
CHANSON M, BRUZZONER, BOX0 D, MDA P: Etfects of nalcohols on junctional coupling and amylase secretion of pancreatic acinar cells. / Cell P&sioI 1989, 139:147-156. The ability of alcohols to block gap junctional permeability rapidly and reversibly between rat pancreatic acinar cells was shown to be independent of several known modulators of intercellular communication. Only those alcohols that inhibited gap junctional communication increased basal amylase secretion, and did so on a time scale compatible with a cause-and-effect relationship between the two events. It is proposed that modulation of cell-cell coupling may be a physiologically lmpcmant means of regulating digestive enzyme secretion. RISEK B, GUTHRE S, KUMAR NM, GILUM NB: Modulation of gap junction transcript and protein expression during preg nancy in the rat. / cell Biol 1990. 110:26%282. Steady-state mRNA levels for various connexins were examined in several organs in pregnant and non-pregnant tats. Results demonstrated spatial and temporal regulation of connexin expression throughout pregnancy, with examples of both tissue-specitic modulation of connexin43 (and connexin32) mRNA levels and differential regulation of two connexin transcripts in the same tissue. hnmunohistochemistry revealed parallel changes in the intensity of anti-connexin staining in the tissues examined.
2.
00
G~MUCHRL, KUMAR NM, Gttu~ NB: Ditferential regulation of the levels of three gap junction tuRNAs in Xen0pu.s embryos. J Cell Biol 1990, 100:597605. Two cDNAs encoding putative connexins were cloned from a Xenqbtu ovary cDNA library and were used for northern blot analysts of connexin expression during early Xen0pu.s development. One connexinlike mRNA was shown to be a maternal transcript that disappeared by the late gastrula stage, whereas the other was rapidly degraded upon oocyte maturation and did not reappear until organogenesis. Taken together with data on the developmental regulation of a previously described connexin32-like Xen0pu.r mRNA, these results demonstrate differential tempotal regulation and tissue distribution of co~exins during embryogenesis. 3.
l e
0, WOK J, DERMEIZEL R, BRUMMER F, HUISER D, WULECKEK Comparative characterization of the 21 kD and 26kD gap junction proteins in mttrine liver and cultured hepatocytes. J Celf Bid 1989, 108:103~1051. Alfinity-purified antibodies specific for either connexin32 or connexin26 were used to characterize expression of these co~exins in rodent hepatocytes biochemically, morphologically, and physiologically. Both connexins colocalized to the same gap junction plaques, had a similar degradation t-ate,were coordinately regulated after partial hepatectomy or in primary culture, and apparently mediated gap junctional permeability as assessedfrom anti-connexin32 and anti-connexin26 antibody junctional blockade experiments. They differed, however, in that connexin32 could be phosphotylated both in intact hepatocytes, and by purified protein kinase A in oifm, whereas connexin26 could not 4. l
TIME
879
880
Cell-t~ell
contact
and extracellular
matrix
SNZ JC, GREGORY WA, WATANABE T, DERMIEIZEL R HERTZBERG El REID L, BENMTT MQ SPRAY DC: cAhlP delays disappearance of gap junctions between pairs of rat hepatocytcs in primary culture. Am / Pkyiol 1989, 257:Cl-Cll. intercellular communication and morphologically recognizable gap junctions between adult rat hepatocytes were shown to decline sharpty 5-8 h after plating onto uncoated tissue culture plastic, a phenomenon tempody associated with a drop in connexin32 steady-state mRNA and protein levels. This decrease in connexin32 expression was delayed by about 8 h by the addition of membrane-permeable derhatives of CAMP; although the mechanism OF the cAMP effect is unclear, an increase in connexin32 mRNA stability and a decrease in removal of gap junction plaques from the cell surface were suggested. 5.
0
6. 0
Northern mRNA duced opment Whether decrease however,
FTIZGERAID DJ, MESNIL M, OYAMADA M, TSUDA H, 1~0 N, Ym H: Changes in gap junction protein (connexin32) gene expression during rat liver carcinogenesis. / Cell Bb%em 1989, 41:97-102. blot analysis revealed a decrease in connexin32 steady-state levels during chemically induced rat hepatocarcinogenesis. Reconnexin32 transcript levels were detected prior to the develof frank carcinoma and persisted after cell transformation. this change in connexin32 mRNA content is paralleled by a in connexin32 protein levels or in ce&ceU coupling was not, addressed.
7.
BEV~~ACQUA A, LOCH-CARUSO R, ERICKZQN RP: Abnormal dwelopment and dye coupling produced by antisense RNA to gap junction protein in mouse preimplantation embryos. Proc Nat1 Acud Sci US.4 1989, 865444-5448. Injection of antisense connexin32 RNA (but not control antisense RNA) into preimplanration mouse embryos led to the exclusion of injected cells from the compacted embryo, and delayed blastulation in compacted embryos. Demonstration that antisense (but not sense) connexin RNA inhibited dye coupling in compacted embryos after a 0.5-2.0 h delay was consistent w-itb a direct effect on connexin expression and a critical role for connexin32-mediated gap junctional communication in early embryogenesis. However, neither connexin32 protein levels nor possible interference with connexin43 translation were examined, allowing for alternative interpretations of the data. 0
8.
AREKIIANO RO, RNERA A, RAMON F: Protein phosphorylation and hydrogen ions modulate calcium-induced closure of gap junction channels. BiqYys J 1990, 57:363-367. Coupled giant lateral axons from crayfish were used to examine the role of Ca*+, H+, and protein phosphotylation in gap junction charnel regulation. CeU uncoupling was rapidly and reversibly stimulated by inuaceUukir perfusion with catalytically active protein kinase A in the presence of elaated Ca*+ whereas either agent alone had no effect on cell-cell communication. Cytoplasmic acidification could substitute for the phosphorylating cocktail, leading the authors to conclude that Ca*+ directly interacts with the gap junction channel and that its effect can be potentiated either by low intracellular pH or by phosphorylation of an as yet unknown protein. l
TAKEDA 4 SAHEKI S, SHIMAZU T, TAKELICHI N: Phosphorylation of the 27kDa gap junction protein by protein kinase C in vitro and in rat hepatocytes. J Eiochem 1989, 106:723727. PartiaUy dissociated primary rat hepatocytes were metabolically labeled with [3*P]orthophosphate and the phosphorylation of cormem monitored. Phosphorylation occurred exclusively on serine residues and was stimulated approximately 1.5.fold in the presence of pkC ago. nists but not by biologicaUy inactive phorbol esters. Both in intact hep. atocytes and in vih-0 pkC-stimulated phosphoryiation was largely con. Iined to a 10 kD cytoplasmic tail domain although the precise residues modified were not determined.
9. l
10. 00
Musn IS, CUNNINGHAM BA, EIDEU&N GM, GOODENOUGH DA Differential phosphorylation of the gap junction protein connexin43 in junctional communication -competent and deficient ceU lines. J Cell Biol November 1990, in press. Metabolic radiolabeling, immunoprecipitation, northern blot analysis, and imrnunohistochemisuy were used to demonstrate that certain cell
lines severely deficient in gap junctional communication constitutiveiy synthesize connexin43 but do not accumulate it on the ceU surface nor phosphoryiate it to its mature form. Manipulation of the junctional communication potential of various connexin43-expressing cell lines reveals a close correlation between the ability of cells to phosphorylate connexin43 on serine residues and to form functional gap junctions. 11. 0
CROW DS, BEYER EC, PALIL DI+ KOBE SS, bu AF: Phosphorylation of connexin43 gap junction protein in uninfected and Rous sarcoma virus-transformed mammalian fibroblasts. Mol Cell Biol 1990, 10:1754-1763. Metabolic labeling, immunoprecipitation, and phosphoamino acid analysis were used to demonstrate phosphorylation of connexin43 on serine residues in uninfected fibroblasts and on both serine and tyrosine in RSV-transformed cells. Pulse-chase analysis indicated that connexin43 in uninfected vole fibroblasts is irreversibiy dephosphorylated within 3Gi5mi1-1 of chase, whereas other investigators [ 10,121 observe no such loss of phosphate from connexin43 in several other ceU types. 12. a
MUSIL IS. BEYER EC, GOODENOUGH DA: Expression of the gap junction protein connexin43 in embryonic chick lens: molecular cloning, ultrastructural localization, and post-translational phosphorylation J Membr Biol 1990. 116:163175. The authors demonstrate very high sequence homology between rat and embryonic chick connexin43, localization of connexin43 in lens to epithelial cells, and in tlitr, processing of connexin43 from a M7. = 42 kD protein to a heterogeneous 44-46 k.D species by the post. translational addition of phosphate.
13.
SWENSON KI, PIWNICXWORMS H, MCNAMEE J, PAUL DL Tyrosine phosphorylation of the gap junction protein connexin43 accounts for the pp6Ov-m-induced inhibition of communication in Xenopus oocyte pairs. Cell Regulation 1990, in press. A paired Xenopuc oocyte expression system was used to demonstrate that the tyrosine kinase ppbov-m dramatically reduces cell-cell coupling mediated by connexin43 but not by connexin.32. Phosphoamino acid analysis revealed that pp6ov-m expression led to tyrosine phosphoryiation of connexin43, but not of connexin32. Site-directed mutagenesis of m265 of connexin43 eliminated both connexin43 tyrosine phosphory larion and downregulation of cell-cell communication in the presence of v-src, elegantly demonstrating that connexin phosphoryiation on ty rosine residues is required for ppGoy m.induced reduction of gap junctional permeability in this system. a*
14. a
HYRC tior.al
K, ROLE B: permeability
The action is modulated
of V-srr on gap by pH. J Cell Eiol
junc1990,
110:1217-1226. The ability of the protein tyrosine kinase pp60V.m to inhibit gap junctional communication in fibroblasts was shown to be counteracted either by cytoplasmic acidification (down to pH 6.75) or by treatment with the calcium antagonist 8-N, N-[diethylamino]octy-3,4,5trimethoxybenzoate (W-8). TMB-8 did not affect either intracellular pH or (in contrast to low pH) pp60v-m.induced tyrosine phosphory lation of general celhilar proteins, indirectty suggesting that pp60v-.m may decrease ceULceU communication by more than one mechanism and/or by some means other than by directly catalyzing tyrosine phos phoryiation of a gap junction protein. 15. 0
MEKTA PP, BERTRAM JS, ILXWENSTEIN WR: The actions of retinoids on cellular growth correlate with their actions on gap junctional communication. J Cell Biol 1989, 108:10531065. A carefully executed study documenting the effects of various retinoids on gap junctional communication, ceU viability, and growth of normal and v-mar transformed ceU Lines. Depending on the type and concentration of retinoid used and the ceU type examined, retinoids either inhibited or enhanced junctional permeability in a manner inversely proportional to ceU saturation density, consistent with the longstanding theoly that growth-controlling substances are transferred from ceU to ceU via gap junctions. Time course studies suggested that retinoids may upregulate gap junctional communication by atfecting gene expression but downregulate the same process by a diEerent, faster mechanism, possibly by directly inUuencing channel closure.
of Anatomy
and Cellular
Biology,
Harvard
Medical
School,
Boston,
Massachusetts,
USA
Current Opinion in Cell Biology 1990, 2:875-880
Introduction
tional communication in these and other complex processes remains to be elucidated.
Gap junctions are collections of transmembrane channels which directly link the cytoplasms of adjoining cells (reviewed in Bennett and Goodenough, Neurosci Res Prog Bull 1978, 16:373-486). Found in virtually all animal tissues, gap junctions provide an almost ubiquitous solution in metazoans to the problem of cell-to-cell transfer of low-molecular-weight substances without exposure to the extracellular environment. The gap junction channel is composed of a hemichannel (connexon) embedded in the plasma membrane of one ceU joined in mirror symmetry with a connexon in the apposing ceU membrane (reviewed by Caspar et al., In Modem Cell Biology edited by Hertzberg and Johnson. Alan R Liss, 1988, pp 117-133). Each connexon is an oligomer of six integral membrane protein subunits which delineate a 1.5 m-n diameter aqueous pore permeable to inorganic ions and small metabolites. The molecular cloning of one of these proteins in 1986 (Paul, J Cell Biol 1986, 103:123-134; Kumar and Gilula, J Cell Biol 1986, 103:767-776) has led to the discovery of a widely distributed family of related gap junction proteins, the connexins. The basic structural features of connexins were reviewed in this journal last year by Stevenson and Paul (Cum Opin Cell Bioll989, l&%-891> and more recently by Beyer et al. (I Membr Bioll990,116:187-194). In this review, the regulation of three members of the connexin family will be discussed: connexin32 (Paul, 1986; Kumar and Gilula 1986), connexin26 (Zhang and Nicholson, J Cell Biol 1989, 109:3391-3401), and connexin43 (Beyer et al, J Cell Bioll987, 105:2621-2629). While passage of current-carrying ions through gap junction channels is clearly important in the activity of electrically excitable tissues such as myocardium, smooth muscle, and nerve, the functions of gap junctions in other ceU types is more obscure. Gap junctions have been implicated in embryonic development, maintaining homeostasis in avascular systems (i.e. lens, vertebrate ovarian follicle), growth control, regulation of glandular secretion, and cellular differentiation (Guthrie and Gilula, Trends Neurosci 1989, 1232-16; Mehta et al. Cell 1986, 44187-196) [l]. However, the precise role of gap junc-
A critical feature of gap junctional intercellular communication is that it can be reversibly modulated in response to extra- and intracellular events. Experimentally, gap junction-mediated cell-to-cell communication is sensitive to a wide range of stimuli (reviewed in Spray and Bennett, Ann Rev Pbysiol1985,47:281-303). The determination of the mechanisms by which this bewildering array of effectors influences cell-to-cell communication is essential for understanding the role of gap junctional communication in non-electrically excitable cells. Some of these effecters (e.g. H+, Ca2+, voltage, and heptanol) reversibly alter junctional permeability on a very rapid time scale (seconds/minutes), presumably by ‘gating’ phenomena (reviewed by Bennett et al, In Mouk-rn Cell Biology. 1988, pp 287-304). In principle, gap junctional communication could also be regulated at the level of cell-cell adhesion, connexin gene transcription and/or translation, connexin assembly, or connexin degradation and/or reutilization. This review will focus on studies published in the last year that examine the control of gap junctionmediated intercellular communication by differential connexin gene expression and by post-translational covalent processing of connexins.
Regulation
of connexin
Rene expression
The recent development of connexin cDNA probes has led to new biochemical approaches to the study of gap junction regulation. We will discuss live weU characterized examples in which modulation of the expression of a particular connexin gene is likely to have a physiologically important effect on cell-to-cell communication.
Hormonal
effects
In pregnant mammals, the period immediately prior to labor (O-24 h) is invariably marked by an increase in morphologically and physiologically recognizable gap junctions between the smooth muscle cells of the uterus
Abbreviations pk-protein
kinase;
RSV-Rous
@ Current
sarcoma
Biology
virus;
TPA-12-O-tetradecanoyl
Ltd ISSN 0955-0674
phorbol-14-acetate.
875
876
Cell-@cell
contact
and extracellular
matrix
(myometrium), a phenomenon thought’to be necessary in synchronizing electrical and contractile activities in the uterine wall during delivery (reviewed by Cole and Garlield, In Gap Junctions edited by Bennett and Spray. Cold Spring Harbor Laboratory, 1985, pp 215-230). This elaboration of uterine gap junctions has been shown to be estrogen-dependent and requires synthesis of both RNA and protein, indicating that it involves gene expression (Garfield et al, Am JPbysioll980, 239:C217-228). A recent study by Risek et al. [2] provides compelling evidence that the pre-parturition changes in myometrial gap junctions are due to a marked increase in the expression of connexin43 (ul in the authors’ nomenclature). Risek et al report a dramatic and specific increase in steadystate connexin43 mRNA levels in the myometrium of term rats, reaching a maximum (5.5 times that of non-pregnant controls) the day before parturition. Anti-connexin43 immunostaining of gap junctional plaques in frozen sections of myometrium peaked 24 h later. Within a day after delivery, both the myometrial connexin43 mRNA and gap junction immunostaining levels rapidly declined. These temporal changes in connexin43 mRNA content and immunostaining were also observed in the stroma of the ovary, another steroid-sensitive organ, but not in the heart, demonstrating tissue-specific regulation of connexin43. Hormonal changes can also lead to a decrease in connexin gene expression. Around the time of ovulation, the interfollicular cell and follicle cell-oocyte gap junctions in mammalian and amphibian ovarian follicles are lost, as is functional oocyte-follicular cell coupling (Gilula et al., J Cell Biol 1978, 78~58-75; Vilain et al., Dev Growth Dif fer 1980, 22687691). In mammals, this downregulation of gap junctions results from the preovulatoty luteinizing hormone surge and can be induced in immature rats by gonadotrophin injection (Gilula et al., 1978) and in isolated intact ovarian follicles by luteinizing hormone and/or follicle-stimulating hormone exposure (Dekel et al, Dev Bioll981, 86:356362; Moor et al., Exp Cell Res 1980, 126:15-29). In Xenopus, binding of progesterone to receptors on the oolemma has an analogous role. Gimlich et al. (31 have shown that maturation of Xenopus oocytes in vitro with progesterone leads to a specific and profound decrease (to less than 9% of control values within 3 h) in the steady-state mRNA levels of al (a Xerzo pus homologue of mammalian connexin43); compatable results were obtained when ovulation was induced in vivo by injection with human chorionic gonadotrophin. It was suggested that the al gene product may form oocyte-follicular cell gap junctions in immature follicles and that its loss at ovulation is responsible for uncoupling communication between the two cell types. The physiological significance of this uncoupling is controversial, with evidence both supporting and refuting a causal role in releasing the oocyte from meiotic arrest (discussed in Wert and Larsen, Gamete Rex 1989, 22:143-162; and Racowsky et al, Eur J Cell Bioll989, 49:244-251).
Responses physiology
to disruption
of tissue architecture
and
Revel and his colleagues have demonstrated (Yee and Revel, J Cell Biol 1978, 78:554-564; Meyer et al., J Cell Bioll981,91:505-523) that partial hepatectomy in young rats leads to marked changes in gap junction structure and function in the remaining liver lobe. Beginning about 20 h postoperatively, the size and number of morphologically recognizable liver gap junctions starts to decrease sharply, accompanied by a loss (although not complete elimination) of gap junctional intercellular communication. However, by 48 h after partial hepatectomy, morphologically recognizable gap junction levels have returned to their control values. A similar temporary downregulation of gap junctions was documented by freezefracture analysis of rat liver after bile duct ligation, which was reversed when cholestasis was relieved (Metz and Bressler, Cell Tissue Res 1979, 199:257-270). Traub et al. (Proc Nat1 Acud Sci USA 1983, 80:755-759) [4] have demonstrated that these alterations in gap junction ultrastructure are associated with changes in connexin protein levels. Using quantitative immunoblot analysis of purified rat liver plasma membrane preparations with anti-connexin antibodies, they report that connexin32 (and, to a lesser extent, connexin26) protein levels fall and then rise after partial hepatectomy in a manner temporally consistent with the morphological changes observed by Revel and coworkers (Yee and Revel, 1978; Meyer et al., 1981). Connexin32 protein concentrations also decline after bile duct ligation and recover after duct recanalization. In the case of partial hepatectomy, there is evidence that the loss of connexin32 protein is preceded by a precipitous drop in steady-state connexin32 mRNA content (to 10% of control values by 10 h after operation), indicating regulation at the mRNA rather than at the protein level (Beer et al, Cancer Res 1988,48:161&1617). Recovery from either condition (by liver regeneration or bile duct recanalization) appears to restimulate connexin synthesis, although the mechanism by which this is accomplished is unknown. Similar loss and reacquisition of connexin mRNA and protein has been documented in cultured primary embyronic mouse hepatocytes (Traub et al., Eur J Cell Bioll987, 43:48-54; Heynkes et al, FEBS Lett 1986, 205:56-60) [4], prompting Dem-tietzel et al. (J Cell Biol 1987, 105:1925-1934) to suggest a relationship between the loss of gap junctions and ceU proliferation. However, it is clear that connexin32 gene expression can also be modulated in nondividing systems such as primary cultures of adult rat hepatocytes (Spray et al., J Cell Biol1987, 105:541-551) [ 51 and during in vivo cholestasis.
Chemically
induced
hepatocarcinogenesis
Several recent studies have reported a decrease (compared with normal liver) in steady-state levels of connexin32 mRNA and in immunohistochemically detectable
Gap junctional
connexin32 protein in chemically induced rat liver neoplasms generated by various protocols (Beer et al, Cancer Res 1988, 48:161&1617; Janssen-Timmen et al., Carcinogenesis 1986, 7:1475-1482) [6]. These results are consistent with the long-standing theory that transformed cells generally (although not always> have lower levels of cell-to-cell communication than their normal counterparts, and that this defect helps to release them from the growth-suppressing effects of surrounding normal cells (Ioewenstein, Biocbim Biopkys Acta 1979, 560:1-65X However, the variability in the extent to which connexin32 mRNA levels were suppressed in the liver tumors tested, and the lack of physiological data on their junctional communication potential, makes any conclusions concerning the role of reduced connexin32 expression in hepatocarcinogenesis premature. Conclusions
Judging from the rate at which new putative connexin gene sequences are being identified, it seems very likely that there are many as yet undiscovered members of the connexin gene family (Beyer et al, 1990). This, combined with the known occurrence of multiple connexlns in many tissues and even within the same cell type, makes it diihcult to interpret the functional consequences of changes in connexin gene expression and to demonstrate that the connexin under study is indeed the one responsible for an observed effect on gap junctional communication. Techniques capable of determining the contribution of individual connexin species to cell-cell communication (i.e. anti-connexin antibody blockade, antisense connexin mRNA expression) are impractical in many systems and have potential artifacts of their own (discussed in [7] and Warner, J Cell Sci 1988, 89:1-7). Thus, in most instances, one can only demonstrate a careful correlation between a change in connexin expression detected biochemically (including assessment of both connexin mRNA and total connexin protein levels) and a change in functional connexin expression detected physiologically, which cannot be taken as proof of a cause-and-effect relationship.
Regulation
of connexin
phosphorylation
Agonists and/or antagonists of cAMP-dependent protein kinase, protein kinase C, calmodulin/Ca*+ -dependent protein kinases, and various tyrosine kinases have been shown to modulate gap junctional communication in multiple systems (reviewed in Loewenstein, Biocbem Sot Sjnnp 1985, 50143-58; Spray et al, In Modern Cell Biology. 1988, pp 227-244). Moreover, increasing the intracellular level of a particular protein kinase (by transfection of cells with kinase-encoding cDNA or by introduction of put&d kinase molecules) alters gap junctional communication in the direction predicted from such kinase effector experiments (Wiener and Loewenstein, Nuture 1983, 305:433435; Iasater, Proc Natf Acud Sci USA 1987, 84:7319-7323) [8]. In some instances, gap junc-
intercellular
communication
Musil and Goodenough
tional communication is altered within an hour of protein kinase up- or downregulation, consistent with a direct effect on phosphorylation of pre-existing proteins. Recent studies from several groups have led to the intriguing conclusion that connexins themselves are substrates for protein kinases. Serine threonine Connexin32
phosphorylation
The first reports of connexin phosphorylation in intact cells demonstrated metabolic labeling of connexin32 with [3*P] orthophosphate in primary rodent hepatocytes. Addition of 8-bromo-cAMP raised gap junctional conductance by 50-75% and resulted in a 1.6-fold lncrease in incorporation of [3*P] into immunoprecipitable connexin32 within 30-60 min (Saez et al, Proc NatlAcud Sci USA 1986, 83:24732477), apparently without appreciably increasing the total amount of connexin32 protein (Traub et al, 1987). In both basal and stimulated states, phosphotylation of connexin32 occurred almost exclusively on serine residues. Isolated rat liver gap junctions could also be phosphotylated on serine by purified cAMP-dependent protein kinase (pkA) with low stoichiometty [between 0.025 (Saez et UC, Prcc Nutf Acud Sci USA1986,83:24732477) and 0.07 [9] moles of phosphate per mole of connexin321; the extent of phosphorylation of connexin32 in intact cells is unknown. On the basis of these results, Saez et al proposed a role for pkA-mediated phosphorylation of connexin32 in gap junctional communication (see Discussion section, however). Recently, Takeda et a~! (FEBS Lett 1987,210:2681- 2688) [9] have determined that connexin32 is also a substrate for another serine/threonine kinase, protein kinase C (pkC). They demonstrated that addition of 120-tetradecanoyl phorbol-14-acetate (TPA) to intact rat hepatocytes rapidly (within 15 min) increased phosphorylation of connexin32 on serine residues by approximately 50%, an effect mimicked by other activators of pkC but not by biologically inactive phorbol esters [ 91. Moreover, purified connexin32 could be phosphorylated in vitro by pkC in a Ca*+ -dependent manner to a maximal stoichiometty of 0.21 mole per mole. Unfortunately, these studies did not address whether activation of pkC decreased cell-cell coupling between primary hepatocytes, as it does in some, but not all, cell types (Chanson et a!, Am J Pkytiol 1988, 255:C699--C704). Thus, the relationship, if any, between pkC-catalyzed phosphotylation of connexin32 and gap junctional intercellular communication remains unclear. Much the same can be said for phosphorylation of connexin32 by calmodulin/Ca*+ _ dependent protein kinase II, which has been reported in an in vitro system but whose physiological significance is unknown (Saez et al., J Cell Bioll986, 103:73a).
Connexin43 Synthesized
both in vivo and in vitro as a single, - 42 kD species, connexin43 has been shown in many cell types
877
878
Cell-to-cell
contact
and extracellular
matrix
to be post-translationally converted to t&o forms with slightly increased molecular weights (PI and P2, in the nomenclature of Musil et al [ 101) by the addition of phosphate onto serine residues [ 11,121. Although the kinases involved in connexin43 phosphotylation have not been characterized, a preliminary account indicating phosphotylation of connexin43 by pkA in isolated mammalian heart gap junctions [ Pressler and Hathaway, Circulation 1987, 76(suppl IV) [18], and the presence of several potentially phosphotylatable serine residues in the cytoplasmic tail domain of connexin43 (Beyer et al, 1987), make pkA a possible candidate. It is weU established that raising cytoplasmic CAMP concentrations rapidly and reversibly increases gap junctional conductance between myocardial cells (De MeUo, Biccbim Bio @~~s~cta 1986,888:71-79); however, whether phosphoryfation of connexin43 is also affected has not been reported. It is also unclear whether pkC, which has been implicated in the control of gap junctional communication in connexin43expressing cells (Spray and Burt, Am J P&i01 1990,258:C195-C205), participates in phosphoryfation of connexin43 under physiological conditions. A potential functional relationship between connexin43 phosphorylation and gap junctional cell-cell comrnunication was suggested by the studies of Musil ef al [lo]. They demonstrated that two cell lines known to be severely deficient in gap junctional communication (mouse S180 and L929 cells) constitutively synthesized connexin43. However, expression of connexin43 in these cells differed from that in comrnunication-competent cells in two critical respects: connexin43 accumulated intracellularIy rather than in cell surface gap junctional plaques, and was not detectably phosphotylated to the mature P2 form despite being as metabolically stable as in communication-competent cells. Conversion of S180 cells to a communication-competent phenotype by transfecting them with a cDNA encoding the interceUular adhesion molecule LCAM led to phosphorylation of connexin43 to the P2 form; furthermore, cells that are usually communication-competent, treated with known inhibitors of gap junction permeability, no longer detectably processed connexin43 to the P2 form. Taken together, these results establish a strong correlation between the ability of cells to phosphorylate connexin43 to the P2 form and to mediate gap junctional communication, but do not prove that the two events are causally linked. The demonstration of connexins in cells lacking morphologically and physiologically recognizable gap junctions has three inter-related implications for the control of gap junction-mediated intercellular communication. First, it illustrates that a change in the number and/or size of gap junctions detected morphologically does not necessarily reflect a change in connexin gene expression, emphasizing the importance of ana@ing both the steady-state connexin protein and mRNA levels before such a claim is made. Second, the presence of connexin ~RNA and protein does not mean that a ceU possesses functioning intercellular channels. Third, it raises the possibility that at least in some instances a rise in intracellular second
messenger levels (e.g. CAMP) converts communicationdeficient cells to a communication-competent phenotype not by stimulating new gene expression but by inducing connexin phosphorylation. Tyrosine
phosphorylation
Infection of various ceU lines with the oncogenic Rous sarcoma virus (RSV) leads to a dramatic reduction in gap junction-mediated intercellular dye coupling. Studies using temperature-sensitive RSV mutants demonstrated that the decrease in junctional permeability was detectable within l5rni1-1 at the permissive temperature, did not require protein synthesis or involve a change in the number of morphologically recognizable gap junctions, and was equally rapidly reversible upon shift to the non-permissive temperature (Atkinson et al., J Cell Biol 1981, 91:573-578; Azarnia and Loewenstein, J Membr Bioll984, 82:191-205). These effects on ceUceU communication were also obsened in cells transfected with the v-STC oncogene of RSV and were shown to be critically dependent on the protein tyrosine kinase activity of the v-src gene product, pp60v-m (Chang et al, Proc Nat1 Acud Sci USA 1985,82:536&5364; Azarnia et al., Science 1988, 239398-401). It is therefore of great interest that connexin43 has recently been shown to be phosphotylated on tyrosine (as weU as on serine) residues in cells expressing ppGO~-m [ 11,131. A cause-and-effect relationship between tyrosine phosphorylation of connexin43 and the decrease in junctional permeability induced by p~60~.~C is suggested by the studies of Swenson et al. [13]. Using an expression system consisting of paired Xenopus oocytes programmed to synthesize connexin43, connexin32, and/or pp60v-sTc by microinjection of their respective mRNAs, Swenson et al. demonstrated that the degree of junctional coupling between connexin43-expressing oocyte pairs is reduced lOO-lOOO-fold in the presence of pp60v-TC. In contrast, junctional permeability between connexin32-expressing oocytes is only marginally affected by pp6Ov-TC (to approximately one-third that of oocytes injected with connexin32 mRNA alone). Phosphoamino acid analysis revealed that pp60v-rC induces tyrosine phosphorylation of connexin43 but not of connexin32, consistent with the presence of a tyrosine (residue 265) in the cytoplasmic tail domain of rat connexin43 that is lacking in rat connexin32. Mutation of Tyr2(j5 to a phenylalanine residue abolishes both tyrosine phosphotyfation of connexin43 and the inhibition of gap junctional communication by ppbO”-PC. The simplest interpretation of these results is that phosphorylation of connexin43 by pp6Ov-TC is clrectly responsible for the decrease in junctional permeability induced by RSV, although the involvement of additional pathways cannot be ruled out (see [13,14]).
Discussion
Recent research in the gap junction Eeld has shed considerable light on the subject of the regulation of gap
Cap iunctional
junction-mediated intercellular communication. Determining the mechanism by which a particular effector influences gap junctional communication can, however, be extremely complicated for several reasons. Some agents (e.g. CAMP and perhaps retinoids and pkC agonists) act on more than one level, and the same agent can have opposite effects on gap junctional communication in different cell types (Saez et al, 1986; Teranishi et al., Nu0.ue 1983, 301:243246) or in the same cell type at different concentrations [ 151. Furthermore, Merent connexins respond to junctional perturbants in distinct ways [13], and the signal transduction pathways used by various transcriptional control (Hoeffler et al, A401Endo crinoll989,3:868-880) and protein kinase (Cohen, EurJ Biochem 1985,151:439-448; Macara et al, Proc NatlAcad Sci USA 1984, 81:2728-2732) systems are closely interlinked and can interact on many levels. Among the many questions raised by these studies, the following two stand out as major topics for the future investigation.
How is connexin
gene expression
regulated?
Resolution of this issue clearly requires characterization of the untranslated regions of connexin genes. Recently, Miller et al (Biosci Rep 1988, 8:455-464) examined the structure of the rat connexin32 gene and found sequences homologous to CAMP response elements in the 5’ non-coding region. Although the functionality of these elements remains to be determined, it is possible that in at least some instances agents that elevate CAMP levels exert their long-term effects on gap junctional communication by increasing the transcription rate of the connexin32 gene. The apparent sensitivity of connexin gene expression to progesterone and estrogen [2] warrants a search for steroid hormone response elements in connexin genes as well. Differential regulation of two connexins in the same tissue [2,3] may therefore reflect heterogeneity in the types of transcriptional response elements present in different connexin genes.
What is the functional role of connexin phosphorylation in gap junctional communication? The observation that CAMP increases both connexin32
phosphotylation and gap junctional conductance in rat hepatocytes led Saez et al. (1986) to speculate that connexin phosphotylation is involved in the gating of gap junction channels. However, no cause-and-effect relationship between these two phenomena has been established in this or in any other system. Proof of an effect of connexin phosphorylation on channel gating will most likely require reconstitution of functional gap junction channels into an artificial bilayer and the demonstration that controlled connexin phosphorylation either up- or downregulates channel permeability. Phosphorylation of connexins could also influence gap junctional communication by affecting non-gating events such as connexin assembly, intracellular transport, or degradation. Evaluation of these possibilities awaits more detailed characterization of connexin biosynthesis, as well as identification
intercellular
communication
Musil and Coodenouah
and site-directed mutagenesis of the serine residues modilied in connexin32 and connexin43
Acknowledgements We thank Drs D Paul, R Bruzzone, K Swenson, R Gimlich. and C Codes for helpful discussions. This work was supported by grants #GM18974 and EYO2430 to DA Goodenough
Annotated reading l l *
references
and recommended
Of interest Of outstanding interest
1. *
CHANSON M, BRUZZONER, BOX0 D, MDA P: Etfects of nalcohols on junctional coupling and amylase secretion of pancreatic acinar cells. / Cell P&sioI 1989, 139:147-156. The ability of alcohols to block gap junctional permeability rapidly and reversibly between rat pancreatic acinar cells was shown to be independent of several known modulators of intercellular communication. Only those alcohols that inhibited gap junctional communication increased basal amylase secretion, and did so on a time scale compatible with a cause-and-effect relationship between the two events. It is proposed that modulation of cell-cell coupling may be a physiologically lmpcmant means of regulating digestive enzyme secretion. RISEK B, GUTHRE S, KUMAR NM, GILUM NB: Modulation of gap junction transcript and protein expression during preg nancy in the rat. / cell Biol 1990. 110:26%282. Steady-state mRNA levels for various connexins were examined in several organs in pregnant and non-pregnant tats. Results demonstrated spatial and temporal regulation of connexin expression throughout pregnancy, with examples of both tissue-specitic modulation of connexin43 (and connexin32) mRNA levels and differential regulation of two connexin transcripts in the same tissue. hnmunohistochemistry revealed parallel changes in the intensity of anti-connexin staining in the tissues examined.
2.
00
G~MUCHRL, KUMAR NM, Gttu~ NB: Ditferential regulation of the levels of three gap junction tuRNAs in Xen0pu.s embryos. J Cell Biol 1990, 100:597605. Two cDNAs encoding putative connexins were cloned from a Xenqbtu ovary cDNA library and were used for northern blot analysts of connexin expression during early Xen0pu.s development. One connexinlike mRNA was shown to be a maternal transcript that disappeared by the late gastrula stage, whereas the other was rapidly degraded upon oocyte maturation and did not reappear until organogenesis. Taken together with data on the developmental regulation of a previously described connexin32-like Xen0pu.r mRNA, these results demonstrate differential tempotal regulation and tissue distribution of co~exins during embryogenesis. 3.
l e
0, WOK J, DERMEIZEL R, BRUMMER F, HUISER D, WULECKEK Comparative characterization of the 21 kD and 26kD gap junction proteins in mttrine liver and cultured hepatocytes. J Celf Bid 1989, 108:103~1051. Alfinity-purified antibodies specific for either connexin32 or connexin26 were used to characterize expression of these co~exins in rodent hepatocytes biochemically, morphologically, and physiologically. Both connexins colocalized to the same gap junction plaques, had a similar degradation t-ate,were coordinately regulated after partial hepatectomy or in primary culture, and apparently mediated gap junctional permeability as assessedfrom anti-connexin32 and anti-connexin26 antibody junctional blockade experiments. They differed, however, in that connexin32 could be phosphotylated both in intact hepatocytes, and by purified protein kinase A in oifm, whereas connexin26 could not 4. l
TIME
879
880
Cell-t~ell
contact
and extracellular
matrix
SNZ JC, GREGORY WA, WATANABE T, DERMIEIZEL R HERTZBERG El REID L, BENMTT MQ SPRAY DC: cAhlP delays disappearance of gap junctions between pairs of rat hepatocytcs in primary culture. Am / Pkyiol 1989, 257:Cl-Cll. intercellular communication and morphologically recognizable gap junctions between adult rat hepatocytes were shown to decline sharpty 5-8 h after plating onto uncoated tissue culture plastic, a phenomenon tempody associated with a drop in connexin32 steady-state mRNA and protein levels. This decrease in connexin32 expression was delayed by about 8 h by the addition of membrane-permeable derhatives of CAMP; although the mechanism OF the cAMP effect is unclear, an increase in connexin32 mRNA stability and a decrease in removal of gap junction plaques from the cell surface were suggested. 5.
0
6. 0
Northern mRNA duced opment Whether decrease however,
FTIZGERAID DJ, MESNIL M, OYAMADA M, TSUDA H, 1~0 N, Ym H: Changes in gap junction protein (connexin32) gene expression during rat liver carcinogenesis. / Cell Bb%em 1989, 41:97-102. blot analysis revealed a decrease in connexin32 steady-state levels during chemically induced rat hepatocarcinogenesis. Reconnexin32 transcript levels were detected prior to the develof frank carcinoma and persisted after cell transformation. this change in connexin32 mRNA content is paralleled by a in connexin32 protein levels or in ce&ceU coupling was not, addressed.
7.
BEV~~ACQUA A, LOCH-CARUSO R, ERICKZQN RP: Abnormal dwelopment and dye coupling produced by antisense RNA to gap junction protein in mouse preimplantation embryos. Proc Nat1 Acud Sci US.4 1989, 865444-5448. Injection of antisense connexin32 RNA (but not control antisense RNA) into preimplanration mouse embryos led to the exclusion of injected cells from the compacted embryo, and delayed blastulation in compacted embryos. Demonstration that antisense (but not sense) connexin RNA inhibited dye coupling in compacted embryos after a 0.5-2.0 h delay was consistent w-itb a direct effect on connexin expression and a critical role for connexin32-mediated gap junctional communication in early embryogenesis. However, neither connexin32 protein levels nor possible interference with connexin43 translation were examined, allowing for alternative interpretations of the data. 0
8.
AREKIIANO RO, RNERA A, RAMON F: Protein phosphorylation and hydrogen ions modulate calcium-induced closure of gap junction channels. BiqYys J 1990, 57:363-367. Coupled giant lateral axons from crayfish were used to examine the role of Ca*+, H+, and protein phosphotylation in gap junction charnel regulation. CeU uncoupling was rapidly and reversibly stimulated by inuaceUukir perfusion with catalytically active protein kinase A in the presence of elaated Ca*+ whereas either agent alone had no effect on cell-cell communication. Cytoplasmic acidification could substitute for the phosphorylating cocktail, leading the authors to conclude that Ca*+ directly interacts with the gap junction channel and that its effect can be potentiated either by low intracellular pH or by phosphorylation of an as yet unknown protein. l
TAKEDA 4 SAHEKI S, SHIMAZU T, TAKELICHI N: Phosphorylation of the 27kDa gap junction protein by protein kinase C in vitro and in rat hepatocytes. J Eiochem 1989, 106:723727. PartiaUy dissociated primary rat hepatocytes were metabolically labeled with [3*P]orthophosphate and the phosphorylation of cormem monitored. Phosphorylation occurred exclusively on serine residues and was stimulated approximately 1.5.fold in the presence of pkC ago. nists but not by biologicaUy inactive phorbol esters. Both in intact hep. atocytes and in vih-0 pkC-stimulated phosphoryiation was largely con. Iined to a 10 kD cytoplasmic tail domain although the precise residues modified were not determined.
9. l
10. 00
Musn IS, CUNNINGHAM BA, EIDEU&N GM, GOODENOUGH DA Differential phosphorylation of the gap junction protein connexin43 in junctional communication -competent and deficient ceU lines. J Cell Biol November 1990, in press. Metabolic radiolabeling, immunoprecipitation, northern blot analysis, and imrnunohistochemisuy were used to demonstrate that certain cell
lines severely deficient in gap junctional communication constitutiveiy synthesize connexin43 but do not accumulate it on the ceU surface nor phosphoryiate it to its mature form. Manipulation of the junctional communication potential of various connexin43-expressing cell lines reveals a close correlation between the ability of cells to phosphorylate connexin43 on serine residues and to form functional gap junctions. 11. 0
CROW DS, BEYER EC, PALIL DI+ KOBE SS, bu AF: Phosphorylation of connexin43 gap junction protein in uninfected and Rous sarcoma virus-transformed mammalian fibroblasts. Mol Cell Biol 1990, 10:1754-1763. Metabolic labeling, immunoprecipitation, and phosphoamino acid analysis were used to demonstrate phosphorylation of connexin43 on serine residues in uninfected fibroblasts and on both serine and tyrosine in RSV-transformed cells. Pulse-chase analysis indicated that connexin43 in uninfected vole fibroblasts is irreversibiy dephosphorylated within 3Gi5mi1-1 of chase, whereas other investigators [ 10,121 observe no such loss of phosphate from connexin43 in several other ceU types. 12. a
MUSIL IS. BEYER EC, GOODENOUGH DA: Expression of the gap junction protein connexin43 in embryonic chick lens: molecular cloning, ultrastructural localization, and post-translational phosphorylation J Membr Biol 1990. 116:163175. The authors demonstrate very high sequence homology between rat and embryonic chick connexin43, localization of connexin43 in lens to epithelial cells, and in tlitr, processing of connexin43 from a M7. = 42 kD protein to a heterogeneous 44-46 k.D species by the post. translational addition of phosphate.
13.
SWENSON KI, PIWNICXWORMS H, MCNAMEE J, PAUL DL Tyrosine phosphorylation of the gap junction protein connexin43 accounts for the pp6Ov-m-induced inhibition of communication in Xenopus oocyte pairs. Cell Regulation 1990, in press. A paired Xenopuc oocyte expression system was used to demonstrate that the tyrosine kinase ppbov-m dramatically reduces cell-cell coupling mediated by connexin43 but not by connexin.32. Phosphoamino acid analysis revealed that pp6ov-m expression led to tyrosine phosphoryiation of connexin43, but not of connexin32. Site-directed mutagenesis of m265 of connexin43 eliminated both connexin43 tyrosine phosphory larion and downregulation of cell-cell communication in the presence of v-src, elegantly demonstrating that connexin phosphoryiation on ty rosine residues is required for ppGoy m.induced reduction of gap junctional permeability in this system. a*
14. a
HYRC tior.al
K, ROLE B: permeability
The action is modulated
of V-srr on gap by pH. J Cell Eiol
junc1990,
110:1217-1226. The ability of the protein tyrosine kinase pp60V.m to inhibit gap junctional communication in fibroblasts was shown to be counteracted either by cytoplasmic acidification (down to pH 6.75) or by treatment with the calcium antagonist 8-N, N-[diethylamino]octy-3,4,5trimethoxybenzoate (W-8). TMB-8 did not affect either intracellular pH or (in contrast to low pH) pp60v-m.induced tyrosine phosphory lation of general celhilar proteins, indirectty suggesting that pp60v-.m may decrease ceULceU communication by more than one mechanism and/or by some means other than by directly catalyzing tyrosine phos phoryiation of a gap junction protein. 15. 0
MEKTA PP, BERTRAM JS, ILXWENSTEIN WR: The actions of retinoids on cellular growth correlate with their actions on gap junctional communication. J Cell Biol 1989, 108:10531065. A carefully executed study documenting the effects of various retinoids on gap junctional communication, ceU viability, and growth of normal and v-mar transformed ceU Lines. Depending on the type and concentration of retinoid used and the ceU type examined, retinoids either inhibited or enhanced junctional permeability in a manner inversely proportional to ceU saturation density, consistent with the longstanding theoly that growth-controlling substances are transferred from ceU to ceU via gap junctions. Time course studies suggested that retinoids may upregulate gap junctional communication by atfecting gene expression but downregulate the same process by a diEerent, faster mechanism, possibly by directly inUuencing channel closure.
