FEBS Letters 588 (2014) 1470–1479

journal homepage: www.FEBSLetters.org

Review

Control of neuronal morphology and connectivity: Emerging developmental roles for gap junctional proteins Michael W. Baker, Eduardo R. Macagno Section of Cell and Developmental Biology, University of California, San Diego, CA 92093, USA

a r t i c l e

i n f o

Article history: Received 27 January 2014 Revised 10 February 2014 Accepted 12 February 2014 Available online 19 February 2014 Edited by Michael Koval, Brant E. Isakson, Robert G. Gourdie and Wilhelm Just Keywords: Synaptogenesis Tiling Dendrite Adhesion Innexin Connexin Homolog avoidance

a b s t r a c t Recent evidence indicates that gap junction (GJ) proteins can play a critical role in controlling neuronal connectivity as well as cell morphology in the developing nervous system. GJ proteins may function analogously to cell adhesion molecules, mediating cellular recognition and selective neurite adhesion. Moreover, during synaptogenesis electrical synapses often herald the later establishment of chemical synapses, and thus may help facilitate activity-dependent sculpting of synaptic terminals. Recent findings suggest that the morphology and connectivity of embryonic leech neurons are fundamentally organized by the type and perhaps location of the GJ proteins they express. For example, ectopic expression in embryonic leech neurons of certain innexins that define small GJ-linked networks of cells leads to the novel coupling of the expressing cell into that network. Moreover, gap junctions appear to mediate interactions among homologous neurons that modulate process outgrowth and stability. We propose that the selective formation of GJs between developing neurons and perhaps glial cells in the CNS helps orchestrate not only cellular synaptic connectivity but also can have a pronounced effect on the arborization and morphology of those cells involved. Ó 2014 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction The unique morphologies and patterns of connections made by neurons during development are thought to arise from an initial cell-type specific period of stereotypic outgrowth, largely under the control of molecular mechanisms that depend on intrinsic developmental programs, which is then followed by an extended period of growth in which cell–cell interactions help to sculpt it’s arbor into its final shape, size and participation in different synaptic networks. These cell–cell interactions include (1) cellular recognition, as mediated by molecules such as immunoglobulin superfamily proteins like DSCAMs and Turtle [1,2], (2) selective neurite adhesivity and repulsion, mediated by cell adhesion molecules, particularly proteins of the immunoglobulin, receptor protein tyrosine kinase and phosphatase, cadherin, and leucine-rich repeat families [3–7]; and (3) activity-dependent processes that help select and sculpt the synaptic terminals of the neuron [8,9]. Recent findings from a number of laboratories suggest that gap junction (GJ) proteins may be in a unique position to contribute to all three of these mechanisms. In this review, we highlight the roles played by GJ proteins as adhesion molecules, and as regulators of neuronal circuit formation in the developing nervous sysE-mail addresses: [email protected] (M.W. Baker), [email protected] (E.R. Macagno)

tem. In the latter role, we suggest, GJ protein arrays may act as surface recognition factors and not simply as conduits that provide for the exchange of electrical and small-molecule signals. Lastly, we present recent evidence that suggests that GJ proteins can have a fundamental role in determining the morphology of particular identified neurons. Among the key properties supporting a role for GJs in cell–cell recognition is that they (1) belong to large gene families, (2) that most cell types, including neurons express more than one type of GJ protein [10–12], (3) and that they are present at axon-to-axon and dendrite-to-dendrite points of contact between growing neurons [10,11]. There are around twenty-one connexins in mammals and at least thirty-seven in zebrafish [13,14]. Twenty-five innexins have been reported in Caenorhabditis elegans [15], and twenty-one different innexin genes in the medicinal leech Hirudo [12]. In contrast, only eight innexins appear to be present in Drosophila, but multiple splice isoforms (e.g. the shaking-B gene gives rise to five possible transcripts [16]), expands that number substantially. Gap junctions form when two hexamers in closely apposed membranes dock together selectively through their extracellular loops [17,18]. All gap junction proteins share a common topology, with four transmembrane domains connected by two extracellular loops and one cytoplasmic loop, leaving both amino and carboxyl termini in the cytoplasm. Studies using paired Xenopus oocytes have revealed that only ‘compatible’ gap junction proteins can

http://dx.doi.org/10.1016/j.febslet.2014.02.010 0014-5793/Ó 2014 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

form functional gap junctions and that the extracellular loops are critical for this docking process to occur [19–24]. By swapping the extracellular loop domains between different connexins, chimeric gap junctions can be assembled and, at least among some of the connexins, this docking is dependent on key amino acid residues along the second extracellular loop hydrogen bonding with a facing hemichannel [25,26]. 2. Gap Junctions as adhesion complexes There is now considerable evidence that GJs can act as adhesion mediators, providing not only an intercellular bridge for communication but also helping to shape adhesive interactions between cells. Some of this evidence comes from direct experimental results examining the roles of GJ proteins in cellular adhesion assays, and some from indirect evidence, which places GJ proteins in close association with other membrane adhesion proteins, including tight junction proteins and cadherins, scaffold proteins and the cellular cytoskeleton (for review see [27]). Cellular aggregation assays, in which mono-dispersed cells on a rotary shaker are examined for their ability to aggregate into cell clusters, are a useful measure of cell adhesivity. Cotrina et al. [28] used a short-term aggregation assay with C6-glioma and HeLa cells stably transfected with Cx43 or Cx32. They found that, although these two cell types do not usually adhere to each other, they do so when they both express the same connexin. Furthermore, this aggregation/adhesivity was found to be calcium independent, unlike the classical role documented for cadherins in tissue self-assembly [29–30]. Just how strong are connexin-mediated adhesive interactions? Studies using different cell lines suggest that it can be comparable to the well-established role of cadherins in cell and tissue assembly [31]. Carbenoxolone, a general GJ inhibitor, disrupts this self-assembly process, as does application of a blocking antibody recognizing the second extracellular loop of Cx43, both of which reduced cellular aggregate size, and critically, this inhibition is comparable to that produced by N-cadherin blocking antibodies [31]. It has long been recognized that hydrophobic interactions of apposing connexins provide for exceptionally strong binding. Denaturants and chaotropic salts such as urea are required to split established GJ channels into two hexamers on separate membranes [32,33]. Indeed, when cardiac tissue is dissociated by collagenase perfusion, the GJ hexamers are ripped from the plasma membrane of one myocyte and retained by its opposite neighbor rather than being split into their component membranes [34]. Do the innexins share this property of adhesivity? One recent report suggests that they do. Drosophila S2 cells are ideal for adhesion interaction assays since untransfected cells are non-adhesive. However, cells transiently transfected with wild-type leech innexins aggregate reliably into multicellular clusters when placed on a rotating shaker [35], suggesting that the innexins, like their vertebrate cousins, share a similar adhesive capacity. Moreover, when two cultures, each expressing a leech innexin, Inx1 or Inx6, are shaken together, they form innexin-specific clusters, indicating that innexin-based adhesivity is selective. There also exists considerable evidence that GJ proteins can function as part of larger protein complexes at the plasma membrane with roles in cellular adhesion. For example, cadherins can help control the trafficking of connexins to the plasma membrane. Firstly, treatment with antibodies against cadherins inhibits GJ-mediated dye transfer between cells in culture, whereas transgene expression of E-cadherin of poorly coupled cell lines increases GJ coupling [36–38]. Furthermore, the cytoplasmic loop of Cx43 has been shown to be necessary for this localization to occur [39], suggesting that connexon insertion and/or localization in the plasma membrane may be cadherin dependent.

1471

An analogous protein–protein interaction has been described for innexins and cadherin proteins in Drosophila epithelial tissues [40]. Fly Inx2 and 3 co-localize to puncta in the membranes of cells in the epidermis. In Inx2 mutants, Inx3 is mislocalized to the cytoplasm and conversely, Inx3 RNAi leads to the mislocalization of Inx2 and, critically, to the mislocalization of Drosophila cadherin, causing cell polarity defects in the epidermis [40]. Furthermore, a direct interaction between Inx2 and adherens junction proteins was identified by yeast two-hybrid analysis, and coimmunoprecipitation experiments using embryonic extracts have shown that Inx2, like Cx43, interacts via its cytoplasmic loop domain with the C terminus of Drosophila cadherin [41]. A third source of evidence for GJ proteins playing an adhesive role comes from studies implicating connexins in cellular migration and morphology. Perhaps the best studied is Cx43, which has been implicated in helping to control the morphology and migration of cells in a variety of tissues, including cardiac neural crest cells, CNS ventricular neuronal cell migration, wound-healing, epithelial cell and B lymphocyte cell migration and glioma invasivity [42–46]. For example, in the rat CNS, transplanted Cx43-expressing glioma cells disseminated freely throughout the brain parenchyma, whereas Cx43-deficient cells did not [43]. In another example, two GJ proteins, Cx43 and Cx26, have been shown to be expressed at the points of contact between migrating neurons in the mammalian cerebral cortex and radial glia cells, which are thin bipolar cells that extend from the inner ventricular surface to the pial surface of the cortex [47,48]. Acute down-regulation of Cx26 or 43, via electroporation of a short-hairpin RNA plasmid, impairs the migration to the cortical plate of neurons expressing the plasmid [49]. Most striking, cells expressing chimeric connexins capable of docking, but not functional channel formation, showed no change in their ability to migrate [49], suggesting that the role played by these GJ proteins does not require the formation of a functional pore. The role of GJs in this process is not yet fully understood, but at its’ simplest, it can be imagined that Cx43 and 26 function in one or more of the following steps in cellular migration; (1) adhesion site formation at the leading edge, (2) adhesion site stabilization or, (3) adhesion site removal at the trailing edge. Lastly, unlike the connexin or innexin’s cytoplasmic loop’s role with cadherin membrane localization [39], Cx43’s function in neuronal, fibroblast and lymphocyte migration appears to be dependent upon the C-terminal domain of the Cx43 [46,48]. In this regard, it may be important to consider that Cx43 has a tubulin-binding domain located in its C-terminus [50,51].

3. Innexons and connexons can act as ‘Lock and Key Recognition’ factors In addition to conferring adhesive properties, the second requirement for the hypothesized role of GJ proteins in helping to shape neuronal connectivity and morphology is that they need to function as cell–cell recognition molecules, helping to discriminate potential target cells from non-target cells by conferring a neuronal identity marker and thereby helping to build GJ-defined neuronal circuits. Considering the case of a cell expressing only two innexins or connexins, which is likely to be a low estimation for most cells including neurons (reviewed in [11]), and assuming that those two proteins are free to associate in a stochastic fashion, then the cell could express two different monomeric hemichannels and up to twelve different combinations of heteromeric hemichannels [52]. Given the numerous possibilities, and the large size of GJ gene families, it can easily be imagined that groups of neurons might display on their surfaces unique and shared hemichannels,

1472

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

which may help categorize potential pre- and postsynaptic partners. Paired Xenopus oocytes have been used widely as a heterologous system to examine the ability of different GJ proteins to form gap-junction channels. Such studies have highlighted the selectivity and discrimination among both innexins and connexins to form functional GJs. For example, Drosophila Inx2 can form homotypic channels among oocytes but Inx3 cannot [53]. However, combining Inx2 and Inx3 can form a channel with unique properties. Similarly, the fly innexin, Ogre, like Inx3, does not form intercellular channels independently, but when co-expressed with Inx2 also modifies intercellular conductance [54]. Ogre was discovered in a mutagenesis screen where it was found that mutant flies had small nervous systems; in particular, the optic lobes were significantly smaller than those of wild-type flies and the neural architecture was highly disorganized [55]. However, it remains unclear how prevalent these mixed-types of interactions are in animal tissues. Many studies have nevertheless inferred the presence of both monomeric and/or heteromeric heterotypic GJs [52]. In the absence of double-label immunoelectron microscopy, the demonstration that different connexins or innexins co-localize to the same GJ requires much more than the co-localization at the light-microscopic level of two fluorescently-tagged proteins. Thus, immunoprecipitation of hemichannels with antibodies specific to one GJ protein can be performed at the tissue level, potentially resulting in the subsequent detection of a different GJ protein. The reconstitution in vitro of innexons and connexons isolated from the animal may also allow a comparison and classification of GJ channel properties. Alternatively, gene knockout or dominant-negative channel effects can be used to infer the mixed composition of certain GJs. One exception may be with the presence of rectifying gap junctions, which are often seen between neurons of different classes. Rectification, or the property by which current flows preferentially through gap junctions in one direction, has been shown to result when gap junction proteins of different identity are located on different sides of a junction. This was first shown to be the case in the Drosophila giant fiber system. Different splice isoforms of the innexin shaking-B, are found in the giant fiber and its postsynaptic motor neurons [16,56]. An analogous heterotypic configuration has recently been demonstrated for connexins at a vertebrate electrical synapse. The Cx36 homologs in goldfish, Cx35 in found on presynaptic afferents and Cx34.7 on the postsynaptic Mauthner cell [57]. An example of where in vivo heteromeric hexamers have been reported is in the fly’s epidermis, where Inx2 and Inx3 are expressed in overlapping patterns [53], and mutation analysis of either GJ gene leads to cuticle deformation [40]. Immunoprecipitation experiments confirm that these two innexins form heteromers in the embryo and immunological membrane localization of innexin 2 was altered in innexin 3 mutants, and vice versa, indicating that Drosophila innexin heteromerization likely occurs between these two proteins in epithelial cells [40]. An example in C. elegans comes from body wall muscle cells, where genetic rescue and single, double and triple-mutations of different innexins indicated that at least six innexins contribute to the coupling seen between two identified muscle cells [58]. Furthermore, by comparing the junctional conductances of the loss of function mutants, it was discovered that for two sets of the innexins, comprised of two and four innexins respectively, their combined loss was equivalent to the loss of either one individually, suggesting that among the six innexins, two functionally distinct populations of GJs were being formed. Thus, since each muscle cell appears to express all six of the innexins, the GJs that form could be imagined to represent two or more innexin proteins assembling into homomeric/heterotypic, heteromeric/heterotypic or heteromeric/homotypic GJs [23,58].

Among chordates, mixed GJ channels have been isolated from different parts of the nervous system. In the mouse cochlea, Cx26 and Cx30 co-localize to the same plaques and their heteromeric assembly was confirmed by co-immunoprecipitation experiments [59]. Single site mutations in different connexins have also lead to the isolation of several GJ proteins with dominant-negative properties impairing not only their own assembly and membrane trafficking but also those of other connexins (e.g. [60,61], and therefore inferring the possibility for mixed hexamers. A similar observation for innexins in the embryonic leech is discussed below. 4. Leech neurons express multiple innexins, some panneuronally, and others by unique or smaller subsets of neurons or glial cells Most of the neurons that comprise the CNS of the medicinal leech are known to make select electrical synapses with some cells and not with others. Among the twenty-one innexin genes found in the Hirudo genome, fifteen are expressed by neurons and glial cells in the central and/or peripheral nervous systems [12]. Two of these (Inx1, Inx14), appear to be expressed pan-neuronally, while the other thirteen are expressed by unique, smaller subsets of cells that in some cases are known to define discreet coupled networks of cells [12,62,63]. For example, Inx6 expression was observed in only three embryonic neurons in each segmental ganglion, the S cell and the two Coupling Interneurons, which together form the S–CI network (Fig. 1C; [63–65]). Injection of Neurobiotin or Lucifer Yellow into a single S cell shows tracer-coupling to the local CIs and to the S–CI network in adjacent ganglia, through an axo-axonal junction between the large-caliber S axons, located in the middle of the interganglionic connectives [66]. In common with fast conducting escape circuits of many invertebrates and vertebrates, the Inx6 network propagates action potentials generated in S cells by sensory inputs along the ganglionic chain, activating motor neurons that produce contractions in longitudinal muscles and shorten the animal. Another example is the circuit defined by Inx2. Each segmental ganglion in the leech includes eight, dye and electrically-coupled, large glial cells, all of which express Inx2, an exclusively glial innexin [62,63]. Finally, quantitative PCR analysis of individual identified adult neurons detected the presence of different combinations of innexin genes. For example, the Retzius cells (Rz) express Inx1, 14, 17 and 19, while the mechanosensory P (Pressure) neurons express Inx1, 3, 6, 14, 17 and the Anterior Pagoda (AP) motor neurons express Inx1, 4, 14, 19 [12]. It is this combinatorial property of multiple innexin expression that we have speculated may underlie the participation of these neurons in different neuronal circuits. 5. Gap junctions with different innexins are heterogeneously distributed in leech neurons To assess the expression patterns of different neuronal innexins in individual embryonic leech neurons, Firme et al. [67] and Yazdani et al. [68] employed GFP and mCherry fusion transgenes, which were expressed randomly by using a gene gun to deliver DNA plasmids or by direct nuclear injection of these plasmids in selected identified neurons. When expressed in single neurons, these tagged innexins produced a punctal pattern of distribution that resembled immunological labeling, and varied in a stereotypical fashion depending on the innexin expressed and their ability to form mixed hexamers (Fig. 1A). For example, when the pan-neuronal innexins Inx1 and Inx14 were expressed together in the same cell, their puncta invariably co-localized within neuronal arbors, whereas other pairs of innexins displayed only partial mixing or none at all [68]. Inx1 and Inx6 fluorescent puncta were seen to

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

1473

Fig. 1. Leech innexin transgene expression can selectively alter neuronal connectivity. (A) A single touch (T) sensory neuron expressing the pan-neuronal innexin, Inx1. (B) Neurobiotin tracer coupling reveals normal coupling to a set of small-unidentified interneurons (arrowheads). (C–F) Ectopic expression of Inx6 by select neurons leads to their inclusion into the S–CI circuit (data taken from [67]). (C) Control Neurobiotin and Alexa 488 Dextran (MW 10 000) injection of a single S cell shows tracer-coupling only to the bilateral CIs, all of which normally express Inx6 [63]. (D) Transgene expression of Inx6 in a Leydig (LY) cell for 48 h followed by Neurobiotin injection reliably leads to it becoming tracer coupled to the S–CI circuit. The Leydig cell’s normal coupling with just its homolog is also detectable (arrowhead). (E) Ectopic expression of Inx6 in Rz neuron or a (T) cell (F) also leads to tracer-coupling with the S–CI network. Bar = 30 lm in A and B, 75 lm in C–D.

co-localize in many of the puncta associated with the primary branches of identified sensory neurons, though they usually did not mix together among peripherally located puncta found at the terminals of these neurons [68]. By contrast, the neuronal innexin Inx1 and the glial innexin Inx2 did not appear to co-localize at all when co-expressed. In the leech, with two notable exceptions presented below, over-expression or ectopic expression of a single innexin in single neurons did not appear to cause a change in cellular coupling [67]. A similar finding was also recently reached while overexpressing different single innexins in C. elegans muscle cells [58]. These observations support the premise that GJ formation may depend on more than any single innexin protein, at least in those tissues examined. A proline residue is found in the second transmembrane domain of all innexins and pannexins and many of the connexins [69]. Studies with connexins describe this proline as introducing a kink into the three-dimensional structure of the protein, and mutation of the proline to leucine in Cx26 has been shown to impart a dominant-negative effect on all connexons composed of the mutated GJ protein subunit [70–72]. Likewise, leech Inx1P149L (proline ? leucine at amino acid location 149) leads to a loss of transgene puncta when it is expressed in neurons, and to the uncoupling of Inx1-based gap junctional communication [35,68]. In addition, it blocked the formation of fluorescent puncta when co-expressed with wild type Inx1 or Inx14, but not Inx6. Thus, like the Cx26 mutant constructs [70–72], the Inx1PL149L appears to exert a trans-dominant-negative effect on other GJs, providing a useful tool for eliminating broad GJ channel functionality. Together with the results of the fluorescent transgene experiments, these findings reveal that leech innexins can selectively interact with one another to form GJ puncta (plaques) and that these puncta are heterogeneously located in arbors of neurons and glia cells. 6. Ectopic expression of Inx2 or Inx6 is sufficient to couple a neuron to an existing GJ-coupled network in the leech CNS When Inx6 or Inx2 transgenes were ectopically expressed in embryonic CNS neurons, they produced fluorescent puncta in their arbors and consistently became dye-coupled to the S–CI (Fig. 1E– H) or glial networks, respectively [67]. This rewiring in cellular

connectivity was long lasting, continuing into the juvenile and adult animal, and very robust. In contrast, overexpression of one of the pan-neuronal innexins (Inx1 or Inx14) in single cells also generated fluorescent puncta but consistently failed to alter the coupling network of that cell. Selectivity among cellular networks as a result of GJ protein expression has been demonstrated in many systems. For example, in the mollusk Clione, ectopic expression of a molluscan innexin gene has been shown to alter electrical coupling between identified neurons [73], and dye-passage experiments among retinal ganglion cells in Cx36 knock-out mice have shown that only some subtypes of ganglion cells lose their coupling [74]. However, because of the detailed cellular expression profile obtained for many of the identified cells in the leech ganglion [12,62,63], we have for the first time been able to show that such selectivity can be directly correlated with the expression of a particular innexin, Inx2 and Inx6, by a network of cells. While ectopic expression of Inx2 or Inx6 was sufficient to rewire individual neurons to the glial or S–CI network, respectively, what about the vast majority of circuits in an assembled collection of cells like those in the leech ganglion, many of which do not appear to have a unique innexin signature? One possibility is that other assembled circuits are specified by a combination of heterotypic GJ hemichannels, which help determine which neurons do and which do not couple to one another. The relatively simple model that GJ-defined synaptic circuits are determined by differential expression of leech innexins unique to particular neuronal pairs is likely to be an oversimplification. Nevertheless, our results with leech Inx2 and 6 suggest one of two possibilities: first, that they represent exceptions rather than general rules of synaptic circuit formation, or alternatively, that their uniqueness lies with the fact that they represent small dedicated cellular circuits within an experimentally well-defined small number of neurons (a leech ganglion contains 400 neurons) and were thus more easily revealed. 7. Gap junctions as regulators of subsequent chemical synaptic connectivity A common feature of early nervous system development is the widespread coupling of cells through GJs, many of which become uncoupled as the nervous system matures. Uncoupling has been

1474

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

seen at the time of axon outgrowth [75,76], as well as during axonal pathfinding at decision points, such as while interacting with guidepost cells (e.g. [77,78]) and sibling and homologous neurons and glial cells (e.g. [79–81]). In the developing optic system of the small crustacean, Daphnia, serial EM reconstructions revealed that during synaptogenesis, transient GJs were the first to form between developing neurons, and that these were only later replaced by chemical synapses in the optic lobe [83]. This sequence of events by which GJ electrical synapses precede the establishment of chemical synapses, has subsequently been observed in the central and peripheral nervous systems of animals from across the animal kingdom [77–82]. Moreover, a causal link between these transient GJs and synaptic circuit formation has been established. In Drosophila, mutations in either the ogre or shaking-B (neuronal) innexin genes produced impaired electrical and chemical synapses in the visual system, and transgenic rescue expression of these innexins in mutants restored chemical synaptic transmission [84]. By comparison, Cx36 knockout mice failed to establish chemical synapses in their olfactory bulbs [85]. In the leech, the necessity of early GJ coupling in establishing later chemical synapses was examined by exploiting a well-defined neuronal circuit involved in a local bending response [81,86]. Todd et al. [87], used single cell siRNAs injections to knockdown the pan-neuronal leech innexin, Inx1, in a sensory-motor neuron synapse before the establishment of chemical synapses between these cells. Electrophysiological and behavioral measures were then used to demonstrate that the expected chemical synapses failed to form on schedule, and they were still missing months later when the nervous system was fully mature. Finally, it is important to consider that the causal link from electrical to chemical synapses is not unidirectional. Chemical synaptic activity can also have a profound affect on the gating and regulation of electrical synapses [88,89]. A regulation that may play a role in sculpting neuronal arbors via activity-dependent mechanisms. 8. Evidence that GJ proteins can regulate neuronal morphology Electrical activity has been show to play a dominant role in helping to sculpt neuronal innervation and circuit formation. In the vertebrate CNS, this is perhaps best exemplified by studies of how electrical activity helps shape retinal ganglion cell axons growing to innervate central regions of the CNS (e.g. [90–92]). In the absence of electrical activity this neuronal arborization has been shown to be significantly more profuse [93–95]. A wealth of data suggests that electrical activity in developing networks influences the branching and stabilization of the axon terminals, and hence the formation of functional neural circuits [94]. In particular, it has been demonstrated that a competitive advantage goes to those cells that have multiple presynaptic inputs and synchronous electrical activity (e.g. [92]). Perturbing GJ signaling can also have significant affects on the morphology of developing neurons. In the developing neuromuscular system, the innervation of muscles is initially accomplished by several different motor axons but this is followed by a period of synapse elimination culminating in innervation by only a single motor axon. It has been shown that by reducing overall motor neuron activity this synapse elimination can be delayed, whereas increasing motor neuron activity accelerates it [96]. In this regard, Cx40 knockout mice were observed to have accelerated neuromuscular synapse elimination [97], This change was attributed to a loss of synchrony in activity at the neuromuscular junction, by decoupling the motor neurons at the level of the spinal cord. Additionally, GJ mutations have been shown to lead to various neuronal pathologies, from the disorganized and shrunken nervous systems produced by the Ogre mutation in flies [55], to morphological neuronal defects in Cx36 knockout mice [98]. However,

whether these states are the direct result of GJ signaling between pre and postsynaptic neurons or effects on hemichannel signaling is not known. Other CNS pathologies, such as seen in Charcot–Marie–Tooth disease, which results from mutations in Cx32, may be the result of secondary effects such as demyelination [98,99]. In other cases, connexin knockout studies have described surprisingly subtle or absent morphological changes to neurons. For example, horizontal cells in the mammalian retina are coupled by dendritic–dendritic GJs, and Cx57-deficient mice have significantly reduced horizontal cell receptive fields and impaired Neurobiotin coupling, but nevertheless, their morphologies appear to be unaffected [100]. Similarly, Cx36 is necessary for GJ coupling by most types of retinal ganglion cells, but knockout mice showed no overt morphological differences as compared with wild type mouse retinal ganglion cells [74]. As with many mutant models, the absence of defects may indicate compensatory mechanisms, but at least in these cases compensation was not brought about by simple replacement of the lost connexin, at least as judged by the loss of a functional GJ pore as revealed by tracer and electrical recordings. Nevertheless, GJ and hemichannel signaling is likely to be critical at many different stages in the life of a neuron and its progenitors, from stem and blast cell proliferation [101–103], cellular differentiation and migration [49,104,105] to cell death [106–108]. In particular, it is worth mentioning that in the developing brain, neuronal and glial coupling and hemichannel signaling has been suggested to directly regulate cell death/survival mechanisms [109]. Therefore, studies examining defects in animals from zygotic genetic knockout have to be carefully appraised. Although neuronal arbors display diverse branching patterns that suit their functions, a common feature among many axonal and dendritic arbors that innervate a common target is that they fill the space evenly, without clumping or overlapping of their branches, requiring some mechanism of self- and homolog recognition and avoidance [110]. As very specific recognition factors, might GJs be involved in these phenomena? Indeed, a recent report from our laboratory is the first to show that this is the case [35]. The leech’s anterior pagoda (AP) neuron, like several other identified motor neurons in the developing CNS, extends interganglionic projections which form functional GJs (passing not only current but also Ca2+ ions and small tracer molecules (such as intracellularly injected serotonin) with their oppositely-directed homologs and which are later retracted [111–114]; shown schematically in Fig. 2A). By contrast, another homologous pair, the S cell, the interneuron from the S–CI network (Fig. 1B), undergoes a similar interaction but their projections are stabilized, stop extending and are retained, connected by GJs, for the life of the animal [65,115]. Yet other neurons (e.g. T and P sensory neurons) do not form GJs between their projections, which bypass each other and grow into and out from the adjacent ganglia (Fig. 2C). Perhaps significantly, the three cells compared in Fig. 2 express different sets of innexins in the adult: AP motor neurons express Inx1, 4, 6, 14, 19; S neurons express Inx6 primarily, and probably also Inx1, 14; and P neurons express Inx1, 3, 6, 14, 17 [12,63]. A role for the AP–AP GJs in the connective nerves in shaping the motor neuron’s mature arbor was suggested by experiments looking at the effects of ablating one of the AP cells before retraction occurs. When killed, the projections of the extrasegmental homologs were observed to continue to grow and take over the vacated territory of the killed cell [111–114]. Interactions of this type that define the extent of a neuron’s arbor through reciprocal growthinhibiting contacts between homologs cells, have been termed homolog avoidance or neuronal tiling [76,116], and have been described in many animal systems, ranging from sensory and motor neurons of the same modality in the leech [116–118], to sensory neurons in Drosophila [119], and to cells in the different layers of the mammalian retina [120,121].

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

1475

Fig. 2. Gap junction interactions between homologous neurons in the developing ganglia. (A–C) Schematic of two ganglion illustrating three types of neuronal homolog development: (A) Embryonic AP neuron axons interact in the connective forming transient GJs between their axons (arrow), which are subsequently retracted (dashed lines). (B) S cells interact in the connective to form stable GJs (arrowhead) and the maintenance of their axons. (C) In contrast, the axons of the P cell do not form GJs and grow past one another. (D) Composite image of an AP neuron 7 days following expression of the empty short hairpin RNAi expression vector and (E) 8 days following expression of the shINX1 transgene. The connective axons have continued to extend, reaching the borders of the anterior and posterior ganglia (red arrowheads). Loss of INX1 GJ signaling in the AP neuron leads to sprouting by the extrasegmental untreated AP homolog. (F) Composite image of two adjacent ganglia. The shINX1 transgene was expressed in the top ganglion’s AP cell for 7 days after which time the extrasegmental ganglion’s ipsilateral AP neuron (bottom ganglion) was injected with Neurobiotin–dextran. (G) Higher magnification view of the shINX1 expressing AP neuron (EGFP) and the extrasegmental AP neuron (H; red label). Both the shINX1 expressing AP and the extrasegmental AP have continued to extend their connective axons into the adjacent ganglion and extend projections out the contralateral nerve roots (arrows). Bar = 200 lm in D–F, 50 lm in G and H (Data taken from [35]).

Recently, we sought to test the role played by these transient GJs in AP homolog avoidance by knocking down Inx1 selectively in single embryonic AP neurons by expressing an Inx1 hairpin siRNA (Fig. 2D–H; [35]). This manipulation resulted in the continued growth of the interganglionic projections of the experimental cell (mimicking cellular ablation) into and in some cases past neighboring ganglia, as well as towards peripheral targets in adjacent segments. Critically, continued growth also occurred by the untreated segmental homolog (Fig. 2G and H), indicating that GJ homolog avoidance is symmetrical. Such novel outgrowth was also observed to occur when knockdown was achieved by expressing the dominant negative Inx1P149L mutant of Inx1. Moreover, both transgenes abolished the normal dye coupling of the experimental AP to its homologs, indicating that Inx1 is necessary for normal GJ coupling to be established. Process retraction by the AP neuron is also abolished by expression of a blocked-pore mutation. A point mutation in Inx1 of a highly conserved leucine (AA 35) in the 1st transmembrane domain (conserved in all the innexins; [69]) to tryptophan allows the formation of normal looking puncta and cellular adhesion but abolishes detectable AP–AP tracer passage [35]. The same mutation in Drosophila innexin Shak-B leads to a reduction of GJ coupling by over 30-fold when expressed in Xenopus oocytes and paired with the wild type Shak-B innexon [122]. Expressing the Inx1L35W transgene in an AP neuron mimicked Inx1 RNAi, with the axons continuing their growth into and beyond neighboring ganglia [35]. This suggests that AP–AP GJs in the connective are likely functioning as mediators of an intercellular diffusible signal essential for process retraction, and not as purely adhesive contacts, but further exploration of this phenomenon is needed. 9. Do vertebrate connexins perform analogous homolog interactions? About half of the twenty-one mammalian connexins are expressed in the CNS [10] and there are some compelling similarities between their patterns of expression and possible roles during development with those of the leech innexins. The retina, for

example, is comprised of seven major cell types and, depending on the species, five to seven different expressed connexins [123]. A recent immunofluorescent labeling study showed that astrocytes in the rat retina alone may have GJs comprised of 1, 2, 3 or 4 different Cxs (Cx26, 30, 43, 45) and that these ratios vary dramatically with development and age [124]. Furthermore, many connexin-defined cellular networks have been described in the vertebrate retina, for example, among amacrine (Cx36), horizontal (Cx50) and retinal ganglion cells (Cx36; [74,125,126]). In the cortex, neurons, astrocytes, oligodendrocytes and macro and microglia also express unique, as well as common sets of connexins, and ‘‘permissive’’ connexin pairing combinations have been suggested to help define separate pathways for neuronal vs. glial GJ communication [10,11,57]. For example, Cx43 is strongly expressed in astrocytes [127]. What would happen, if Cx43 were conditionally and selectively expressed in neurons? Would this lead to the coupling of neurons and glia as seen in the leech ganglion? Regarding homolog avoidance, examples in the vertebrate brain are widespread; for example, in the retina, between the dendrites of retinal ganglion cells [120], horizontal cells [121] and bipolar cells [128], and in the cerebellum between Purkinje cells [129], and among astrocytes in the cortex and hippocampus [130]. However, in each case the role of GJs has not been fully addressed. The cerebellum, with only five easily identified major cell types and its organized and repetitive structure, provides an excellent model system for studying how mammalian brain circuits are formed during development. Purkinje cells form non-overlapping dendritic fields and time-lapse imaging of their growing dendrites in culture have shown that they display characteristic homolog avoidance (tiling), including the retraction of dendritic tips when contacting the tips of sibling cells [129]. During development, GJ dye-transfer of Neurobiotin between these dendro-dendritic contacts has been documented to occur, with dye coupling increasing within the first 2 weeks after birth and diminishing afterwards [131], a period coincident with morphological and synaptic maturation [132]. In support of a role for GJs in helping to control neuronal morphology, dendrites in the inferior olive of Cx36 knockout mice may

1476

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

Fig. 3. (A) Schematic representations of 6 leech GJ-defined networks (double-headed arrows signify non-rectifying GJs, rectifying are single-headed). Colors correspond to the cells in the table (B) and the boxes to the probable leech innexins involved. The AP–AP GJs (blue) could include Inx1, 4, 14, and 19; The P-AP include Inx1, 4, 14,17,19; Rz–Rz (brown) includes Inx1, 14, 17, 19; S–CI (purple) include Inx1, 6, 14; the glia (red) innexins are Inx2 and 3. All electrical connections can also be found in the adult ganglion except the P–AP connection, which becomes a chemical synapse. (C) Ectopic expression of Inx6 (light blue arrows) by the AP, Rz, P, T or glia cells leads to their inclusion in the CI–S network ([67]; unpublished observations). (D) Cartoon of potential GJ interactions between the bilateral AP neurons, the AP and the contralateral P neuron, and the AP and its extrasegmental homologs. We anticipate that different GJs proteins help to mediate each of these cell–cell interactions and that the innexons involved are spatially segregated within the AP and P neuron’s arbor.

not be as closely apposed in the absence of GJs [133]. At the light microscope level, olivary neurons, which normally display GJs between their dendritic spines within the olivary glomeruli [134,135], had thicker dendrites in the knockout mice and at the electron microscope level, abnormally wide interneuronal gaps between their dendrites [133]. Lastly, while the large innexins/pannexin and connexin gene families appear to have arisen independently during evolution [56,136,137], they share analogous functions across the invertebrate and vertebrate divide. This even extends to their protein domain interactions and regulation by other membrane and cytoplasmic proteins [39,41]. Thus, while the sufficiency of a GJ-based selection process in controlling neuronal connectivity and morphology may differ between invertebrates and vertebrates, there is considerable indirect evidence that connexins have adopted some of the same developmental roles as have been documented for innexins in invertebrates. 10. A model for GJ control of neuronal connectivity We have proposed a simplified model for how GJs might function to help regulate neuronal connectivity and morphology. To do this we have borrowed a number of findings and assumptions gathered from studies of innexins and connexins in other cell systems. First, based on cell culture studies using the biotinylation of surface membrane proteins or the uptake of dyes and tracers via

connexons and innexons, it is known that a significant amount of the total connexin and innexin protein found on the cell surface is in the form of unopposed hemichannels [68,109,138,139]. We therefore propose that hemichannels, comprised of single or multiple GJ proteins, diffuse laterally in the plasma membrane [140] of developing neurons, and based on their selective adhesivity are able to recognize and dock with the hemichannels of nearby adjacent neurons. Secondly, since leech neurons appear to express multiple innexin proteins, some of which can be unique, like Inx2 and Inx6 in the leech, and others which are common, we propose that assembled circuits in the leech ganglion are specified by a combination of heterotypic GJ hemichannels, which help determine which neurons do and which do not couple to one another. Thirdly, independent of direct signaling by the GJ pore, this docking helps provide for the stabilization of dynamic neuronal processes, anchoring the contacting membranes along with their internal cellular cytoskeleton. Fourthly, synaptic stabilization also occurs via an activity-dependent mechanism. As epitomized by the saying that ‘‘neurons which fire together, wire together’’, early GJs promote the stabilization of the contacting processes by synchronizing electrical activity which, through mostly still unidentified mechanisms, provides a competitive trophic growth factor advantage over the processes of other non-synchronous neuronal inputs. Lastly, certain GJs, particularly those that form between neuronal homologs, produce a signal, which can have the opposite effect, destabilizing the connecting processes and promoting neurite retraction.

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

11. Future directions To date, there has been only limited structural information available on the docking and formation of functional GJ channels. Detailed descriptions of the stoichiometries of actual GJs found in the CNS would help validate some of these ideas and provide details of how GJ protein hemichannels function as recognition molecules. It might be imaged, for example, that heteromeric hemichannels of two different innexins (and not three, four or six) might provide more stability/adhesivity because of the greater symmetry provided by an alternating protein extracellular loop structure [58]. With regard to the mapping of GJ proteins within the arbor of a neuron, the co-localization of GJ proteins to the same fluorescent puncta is only suggestive of heteromeric GJs or mixed innexons and connexons. Resolving the constituent proteins in a GJ hemichannel, in the absence of double-label immunoelectron microscopy, however, may be possible through indirect means. One possibility may lie with the bimolecular fluorescence complementation assay (BiFC), which relies on the reconstitution of an intact fluorescent protein when two complementary non-fluorescent fragments are brought together by a pair of closely interacting proteins [141]. For example, BiFC has been successfully used to document the close interaction of an innexin protein in C. elegans (UNC-9) with a 2-pass membrane protein (UNC-1) that helps control channel gating [142]. It will also be critical to determine if different GJ proteins or their hemichannels are differentially distributed within neuronal arbors. In such a fashion it might be imagined that a neuron can be joined to more than one, GJ-defined cellular network (Fig. 3D). For example, what makes the GJs involved in AP–AP axon retraction behavior different from the other GJs between homologs in the leech CNS? Are the GJs that are formed between the bilateral isosegmental APs different from those that form between the extrasegmental APs, or between sensory P cells and the AP neuron (Fig. 3D)? We expect that answering these questions will help delineate the developmental significance of multiple and differential innexin expression. Lastly, our work with AP homolog avoidance suggests that the repulsive signal requires an active GJ pore [35], but how this comes about is currently unknown, although a number of candidate pathways present themselves from the neuronal tiling literature including such things as co-receptor activation and calcium activated proteases [143,144]. Acknowledgement This work was supported by NSF Grant DBI-0852081 and IBN0446346 to E.R.M. References [1] Millard, S.S., Flanagan, J.J., Pappu, K.S., Wu, W. and Zipursky, S.L. (2007) Dscam2 mediates axonal tiling in the Drosophila visual system. Nature 447, 720–724. [2] Ferguson, K., Long, H., Cameron, S., Chang, W.T. and Rao, Y. (2009) The conserved Ig superfamily member Turtle mediates axonal tiling in Drosophila. J. Neurosci. 29, 14151–14159. [3] Baker, M.W., Rauth, S.J. and Macagno, E.R. (2000) Possible role of the receptor protein tyrosine phosphatase HmLAR2 in interbranch repulsion in a leech embryonic cell. J. Neurobiol. 45, 47–60. [4] Irie, F. and Yamaguchi, Y. (2002) EphB receptors regulate dendritic spine development via intersectin, Cdc42 and N-WASP. Nat. Neurosci. 5, 1117– 1118. [5] Williams, M.E., de Wit, J. and Ghosh, A. (2010) Molecular mechanisms of synaptic specificity in developing neural circuits. Neuron 68, 9–18. [6] Zipursky, S.L. and Sanes, J.R. (2010) Chemoaffinity revisited: dscams, protocadherins, and neural circuit assembly. Cell 143, 343–353. [7] Kolodkin, A.L. and Tessier-Lavigne, M. (2011) Mechanisms and molecules of neuronal wiring: a primer. Cold Spring Harb. Perspect. Biol. 3. pii: a001727. [8] Lichtman, J.W. and Purves, D. (1983) Activity-mediated neural change. Nature 301, 17–23.

1477

[9] Park, H. and Poo, M.M. (2013) Neurotrophin regulation of neural circuit development and function. Nat. Rev. Neurosci. 14, 7–23. [10] Nagy, J.I., Dudek, F.E. and Rash, J.E. (2004) Update on connexins and gap junctions in neurons and glia in the mammalian nervous system. Brain Res. Brain Res. Rev. 47, 191–215. [11] Bedner, P., Steinhäuser, C. and Theis, M. (2012) Functional redundancy and compensation among members of gap junction protein families? Biochim. Biophys. Acta 1818, 1971–1984. [12] Kandarian, B., Sethi, J., Wu, A., Baker, M., Yazdani, N., Kym, E., Sanchez, A., Edsall, L., Gaasterland, T. and Macagno, E. (2012) The medicinal leech genome encodes 21 innexin genes: different combinations are expressed by identified central neurons. Dev. Genes. Evol. 222, 29–44. [13] Eastman, S.D., Chen, T.H., Falk, M.M., Mendelson, T.C. and Iovine, M.K. (2006) Phylogenetic analysis of three complete gap junction gene families reveals lineage-specific duplications and highly supported gene classes. Genomics 87, 265–274. [14] Cruciani, V. and Mikalsen, S.O. (2007) Evolutionary selection pressure and family relationships among connexin genes. Biol. Chem. 388, 253–264. [15] Altun, Z.F., Chen, B., Wang, Z.W. and Hall, D.H. (2009) High resolution map of Caenorhabditis elegans gap junction proteins. Dev. Dyn. 238, 1936–1950. [16] Zhang, Z., Curtin, K.D., Sun, Y.A. and Wyman, R.J. (1999) Nested transcripts of gap junction gene have distinct expression patterns. J. Neurobiol. 40, 288– 301. [17] Dahl, G., Werner, R., Levine, E. and Rabadan-Diehl, C. (1992) Mutational analysis of gap junction formation. Biophys J. 62, 172–180. [18] Bruzzone, R., White, T.W. and Paul, D.L. (1994) Expression of chimeric connexins reveals new properties of the formation and gating behavior of gap junction channels. J. Cell Sci. 107, 955–967. [19] Meyer, R.A., Laird, D.W., Revel, J.P. and Johnson, R.G. (1992) Inhibition of gap junction and adherens junction assembly by connexin and A-CAM. Antibodies. J. Cell Biol. 119, 179–189. [20] Dahl, G., Nonner, W. and Werner, R. (1994) Attempts to define functional domains of gap junction proteins with synthetic peptides. Biophys J. 67, 1816–1822. [21] White, T.W., Bruzzone, R., Wolfram, S., Paul, D.L. and Goodenough, D.A. (1994) Selective interactions among the multiple connexin proteins expressed in the vertebrate lens: the second extracellular domain is a determinant of compatibility between connexins. J. Cell Biol. 125, 879–892. [22] Foote, C.I., Zhou, L., Zhu, X. and Nicholson, B.J. (1998) The pattern of disulfide linkages in the extracellular loop regions of connexin 32 suggests a model for the docking interface of gap junctions. J. Cell Biol. 140, 1187–1197. [23] Harris, A.L. (2001) Emerging issues of connexin channels, biophysics fills the gap. Q. Rev. Biophys. 34, 325–472. [24] Pfahnl, A., Zhou, X.W., Werner, R. and Dahl, G. (1997) A chimeric connexin forming gap junction hemichannels. Pflugers Arch. 433, 773–779. [25] Nakagawa, S., Gong, X.Q., Maeda, S., Dong, Y., Misumi, Y., Tsukihara, T. and Bai, D. (2011) Asparagine 175 of connexin32 is a critical residue for docking and forming functional heterotypic gap junction channels with connexin26. J. Biol. Chem. 286, 19672–19681. [26] Gong, X.Q., Nakagawa, S., Tsukihara, T. and Bai, D. (2013) A mechanism of gap junction docking revealed by functional rescue of a human-disease-linked connexin mutant. J. Cell Sci. 126, 3113–3120. [27] Hervé, J.C., Derangeon, M., Sarrouilhe, D., Giepmans, B.N. and Bourmeyster, N. (2012) Gap junctional channels are parts of multiprotein complexes. Biochim. Biophys. Acta 1818, 1844–1865. [28] Cotrina, M.L., Lin, J.H.C. and Nedergaard, M. (2008) Adhesive properties of connexin hemichannels. Glia 56, 1791–1798. [29] Hyafil, F., Morello, D., Babinet, C. and Jacob, F. (1980) A cell surface glycoprotein involved in the compaction of embryonal carcinoma cells and cleavage stage embryos. Cell 21, 927–934. [30] Takeichi, M., Atsumi, T., Yoshida, C., Uno, K. and Okada, T.S. (1981) Selective adhesion of embryonic carcinoma cells and differentiated cells by Ca2+ dependent sites. Dev. Biol. 87, 340–350. [31] Bao, B., Jiang, J., Yanase, T., Nishi, Y. and Morgan, J.R. (2011) Connexonmediated cell adhesion drives microtissue self-assembly. FASEB J. 25, 255– 264. [32] Manjunath, C.K., Goings, G.E. and Page, E. (1984) Detergent sensitivity and splitting of isolated liver gap junctions. J. Membr. Biol. 78, 147–155. [33] Ghoshroy, S., Goodenough, D.A. and Sosinsky, G.E. (1995) Preparation, characterization, and structure of half gap junctional layers split with urea and EGTA. J. Membr. Biol. 146, 15–28. [34] Severs, N.J., Slade, A.M., Powell, T., Twist, V.W. and Green, C.R. (1990) Integrity of the dissociated adult cardiac myocyte: gap junction tearing and the mechanism of plasma membrane resealing. J. Muscle Res. Cell Motil. 11, 154–166. [35] Baker, M.W., Yazdani, N. and Macagno, E.R. (2013) Gap junctiondependent homolog avoidance in the developing CNS. J. Neurosci. 33, 16673–16683. [36] Matsuzaki, F., Mège, R.M., Jaffe, S.H., Friedlander, D.R., Gallin, W.J., Goldberg, J.I., Cunningham, B.A. and Edelman, G.M. (1990) CDNAs of cell adhesion molecules of different specificity induce changes in cell shape and border formation in cultured S180 cells. J. Cell Biol. 110, 1239–1252. [37] Jongen, W.M.F., Fitzferald, D.J., Asamoto, M., Piccoli, C., Slaga, T.J., Gros, D., Takeichi, M. and Yamasaki, H. (1991) Regulation of connexin 43- mediated gap junctional intercellular communication by Ca2+ in mouse epidermal cells is controlled by E-cadherin. J. Cell Biol. 114, 545–555.

1478

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479

[38] Chakraborty, S., Mitra, S., Falk, M.M., Caplan, S.H., Wheelock, M.J., Johnson, K.R. and Mehta, P.P. (2010) E-cadherin differentially regulates the assembly of Connexin43 and Connexin32 into gap junctions in human squamous carcinoma cells. J. Biol. Chem. 285, 10761–10776. [39] Nambara, C., Kawasaki, Y. and Yamasaki, H. (2007) Role of the cytoplasmic loop domain of Cx43 in its intracellular localization and function: possible interaction with cadherin. J. Membr. Biol. 217, 63–69. [40] Lehmann, C., Lechner, H., Löer, B., Knieps, M., Herrmann, S., Famulok, M., Bauer, R. and Hoch, M. (2006) Heteromerization of innexin gap junction proteins regulates epithelial tissue organization in Drosophila. Mol. Biol. Cell 17, 1676–1685. [41] Bauer, R., Lehmann, C., Martini, J., Eckardt, F. and Hoch, M. (2004) Gap junction channel protein innexin 2 is essential for epithelial morphogenesis in the Drosophila embryo. Mol. Biol. Cell 15, 2992–3004. [42] Xu, X., Li, W.E., Huang, G.Y., Meyer, R., Chen, T., Luo, Y., Thomas, M.P., Radice, G.L. and Lo, C.W. (2001) Modulation of mouse neural crest cell motility by Ncadherin and connexin43 gap junctions. J. Cell Biol. 154 (2001), 217–230. [43] Lin, J.H., Takano, T., Cotrina, M.L., Arcuino, G., Kang, J., Liu, S., Gao, Q., Jiang, L., Li, F., Lichtenberg-Frate, H., Haubrich, S., Willecke, K., Goldman, S.A. and Nedergaard, M. (2002) Connexin 43 enhances the adhesivity and mediates the invasion of malignant glioma cells. J. Neurosci. 22, 4302–4311. [44] Simpson, K.J., Selfors, L.M., Bui, J., Reynolds, A., Leake, D., Khvorova, A. and Brugge, J.S. (2008) Identification of genes that regulate epithelial cell migration using an siRNA screening approach. Nat. Cell Biol. 10, 1027–1038. [45] Francis, R., Xu, X., Park, H., Wei, C.J., Chang, S., Chatterjee, B. and Lo, C. (2011) Connexin43 modulates cell polarity and directional cell migration by regulating microtubule dynamics. PLoS ONE 6, e26379. [46] Machtaler, S., Dang-Lawson, M., Choi, K., Jang, C., Naus, C.C. and Matsuuchi, L. (2011) The gap junction protein Cx43 regulates B-lymphocyte spreading and adhesion. J. Cell Sci. 124, 2611–2621. [47] Nadarajah, B., Jones, A.M., Evans, W.H. and Parnavelas, J.G. (1997) Differential expression of connexins during neocortical development and neuronal circuit formation. J. Neurosci. 17, 3096–3111. [48] Cina, C., Bechberger, J.F., Ozog, M.A. and Naus, C.C. (2007) Expression of connexins in embryonic mouse neocortical development. J. Comp. Neurol. 504, 298–313. [49] Elias, L.A., Wang, D.D. and Kriegstein, A.R. (2007) Gap junction adhesion is necessary for radial migration in the neocortex. Nature 448, 901–907. [50] Giepmans, B.N., Verlaan, I., Hengeveld, T., Janssen, H. and Calafat, J. (2001) Gap junction protein connexin-43 interacts directly with microtubules. Curr. Biol. 11, 1364–1368. [51] Dai, P., Nakagami, T., Tanaka, H., Hitomi, T. and Takamatsu, T. (2007) Cx43 mediates TGF-beta signaling through competitive Smads binding to microtubules. Mol. Biol. Cell 18, 2264–2273. [52] Cottrell, G.T. and Burt, J.M. (2005) Functional consequences of heterogeneous gap junction channel formation and its influence in health and disease. Biochim. Biophys. Acta 1711, 126–141. [53] Stebbings, L.A., Todman, M.G., Phillips, R., Greer, C.E., Tam, J., Phelan, P., Jacobs, K., Bacon, J.P. and Davies, J.A. (2002) Gap junctions in Drosophila: developmental expression of the entire innexin gene family. Mech. Dev. 113, 197–205. [54] Holcroft, C.E., Jackson, W.D., Lin, W.H., Bassiri, K., Baines, R.A. and Phelan, P. (2013) Innexins Ogre and Inx2 are required in glial cells for normal postembryonic development of the Drosophila central nervous system. J. Cell Sci. 126, 3823–3834. [55] Lipshitz, H.D. and Kankel, D.R. (1985) Specificity of gene action during central nervous system development in Drosophila melanogaster: analysis of the lethal (1) optic ganglion reduced locus. Dev. Biol. 108, 56–77. [56] Phelan, P., Goulding, L.A., Tam, J.L., Allen, M.J., Dawber, R.J., Davies, J.A. and Bacon, J.P. (2008) Molecular mechanism of rectification at identified electrical synapses in the Drosophila giant fiber system. Curr. Biol. 18, 1955–1960. [57] Rash, J.E., Curti, S., Vanderpool, K.G., Kamasawa, N., Nannapaneni, S., PalaciosPrado, N., Flores, C.E., Yasumura, T., O’Brien, J., Lynn, B.D., Bukauskas, F.F., Nagy, J.I. and Pereda, A.E. (2013) Molecular and functional asymmetry at a vertebrate electrical synapse. Neuron 79, 957–969. [58] Liu, P., Chen, B., Altun, Z.F., Gross, M.J., Shan, A., Schuman, B., Hall, D.H. and Wang, Z.W. (2013) Six innexins contribute to electrical coupling of C. elegans body-wall muscle. PLoS ONE 8, e76877. [59] Ahmad, S., Chen, S., Sun, J. and Lin, X. (2003) Connexins 26 and 30 are coassembled to form gap junctions in the cochlea of mice. Biochem. Biophys. Res. Commun. 307, 362–368. [60] Beahm, D.L., Oshima, A., Gaietta, G.M., Hand, G.M., Smock, A.E., Zucker, S.N., Toloue, M.M., Chandrasekhar, A., Nicholson, B.J. and Sosinsky, G.E. (2006) Mutation of a conserved threonine in the third transmembrane helix of alpha- and beta-connexins creates a dominant-negative closed gap junction channel. J. Biol. Chem. 281, 7994–8009. [61] Su, C.C., Li, S.Y., Su, M.C., Chen, W.C. and Yang, J.J. (2010) Mutation R184Q of connexin 26 in hearing loss patients has a dominant-negative effect on connexin 26 and connexin 30. Eur. J. Hum. Genet. 18, 1061–1064. [62] Dykes, I.M., Freeman, F.M., Bacon, J.P. and Davies, J.A. (2004) Molecular basis of gap junctional communication in the CNS of the leech Hirudo medicinalis. J. Neurosci. 24, 886–894. [63] Dykes, I.M. and Macagno, E.R. (2006) Molecular characterization and embryonic expression of innexins in the leech Hirudo medicinalis. Dev. Genes. Evol. 216, 185–197.

[64] Elliot, E.J. and Muller, K.J. (1981) Long-term survival of glial segments during nerve regeneration in the leech. Brain Res. 218, 99–113. [65] Muller, K.J. and Scott, S.A. (1981) Transmission at a ‘‘direct’’ electrical connection mediated by an interneurone in the leech. J. Physiol. (Lond) 311, 565–583. [66] Modney, B.K. and Muller, K.J. (1994) Novel synapses compensate for a neuron ablated in embryos. Proc. Biol. Sci. 257, 263–269. [67] Firme 3rd, C.P., Natan, R.G., Yazdani, N., Macagno, E.R. and Baker, M.W. (2012) Ectopic expression of select innexins in individual central neurons couples them to pre-existing neuronal or glial networks that express the same innexin. J. Neurosci. 32, 14265–14270. [68] Yazdani, N., Firme 3rd, C.P., Macagno, E.R. and Baker, M.W. (2013) Expression of a dominant negative mutant innexin in identified neurons and glial cells reveals selective interactions among gap junctional proteins. Dev. Neurobiol. 73, 571–586. [69] Phelan, P. (2005) Innexins, members of an evolutionarily conserved family of gap-junction proteins. Biochim. Biophys. Acta 1711, 225–245. [70] Suchyna, T.M., Xu, L.X., Gao, F., Fourtner, C.R. and Nicholson, B.J. (1993) Identification of a proline residue as a transduction element involved in voltage gating of gap junctions. Nature 1365, 847–849. [71] Ri, Y., Ballesteros, J.A., Abrams, C.K., Oh, S., Verselis, V.K., Weinstein, H. and Bargiello, T.A. (1999) The role of a conserved proline residue in mediating conformational changes associated with voltage gating of Cx32 gap junctions. Biophys. J. 76, 2887–2898. [72] Ambrosi, C., Boassa, D. and Pranskevich, J. (2010) Analysis of four connexin26 mutant gap junctions and hemichannels reveals variations in hexamer stability. Biophys. J. 98, 1809–1819. [73] Kelmanson, I.V., Shagin, D.A., Usman, N., Matz, M.V., Lukyanov, S.A. and Panchin, Y.V. (2002) Altering electrical connections in the nervous system of the pteropod mollusc Clione limacina by neuronal injections of gap junction mRNA. Eur. J. Neurosci. 16, 2475–2476. [74] Pan, F., Paul, D.L., Bloomfield, S.A. and Völgyi, B. (2010) Connexin36 is required for gap junctional coupling of most ganglion cell subtypes in the mouse retina. J. Comp. Neurol. 518, 911–927. [75] Goodman, C.S. and Spitzer, N.C. (1979) Embryonic development of identified neurones: differentiation from neuroblast to neurone. Nature 280, 208–214. [76] Kramer, A.P. and Kuwada, J.Y. (1983) Formation of the receptive fields of leech mechanosensory neurons during embryonic development. J. Neurosci. 3, 2474–2486. [77] Allen, F. and Warner, A. (1991) Gap junctional communication during neuromuscular junction formation. Neuron 6, 101–111. [78] Taghert, P.H., Bastiani, M.J., Ho, R.K. and Goodman, C.S. (1982) Guidance of pioneer growth cones: filopodial contacts and coupling revealed with an antibody to Lucifer Yellow. Dev. Biol. 94, 391–399. [79] Yuste, R., Peinado, A. and Katz, L.C. (1992) Neuronal domains in developing neocortex. Science 257, 665–669. [80] Peinado, A., Yuste, R. and Katz, L.C. (1993) Extensive dye coupling between rat neocortical neurons during the period of circuit formation. Neuron 10, 103–114. [81] Marin-Burgin, A., Eisenhart, F.J., Kristan Jr, W.B. and French, K.A. (2006) Embryonic electrical connections appear to pre-figure a behavioral circuit in the leech CNS. J. Comp. Physiol. A Neuroethol. Sens. Neural Behav. Physiol. 192, 123–133. [82] Szabo, T.M. and Zoran, M.J. (2007) Transient electrical coupling regulates formation of neuronal networks. Brain Res. 1129, 63–71. [83] Lopresti, V., Macagno, E.R. and Levinthal, C. (1974) Structure and development of neuronal connections in isogenic organisms: transient gap junctions between growing optic axons and lamina neuroblasts. Proc. Natl. Acad. Sci. U.S.A. 71, 1098–1102. [84] Curtin, K.D., Zhang, Z. and Wyman, R.J. (2002) Gap junction proteins expressed during development are required for adult neural function in the Drosophila optic lamina. J. Neurosci. 22, 7088–7096. [85] Maher, B.J., McGinley, M.J. and Westbrook, G.L. (2009) Experience-dependent maturation of the glomerular microcircuit. Proc. Natl. Acad. Sci. U.S.A. 106, 16865–16870. [86] Marin-Burgin, A., Kristan Jr, W.B. and French, K.A. (2008) From synapses to behavior: development of a sensory-motor circuit in the leech. Dev. Neurobiol. 68, 779–787. [87] Todd, K.L., Kristan Jr, W.B. and French, K.A. (2010) Gap junction expression is required for normal chemical synapse formation. J. Neurosci. 30, 15277– 15285. [88] Hatton, G.I. (1998) Synaptic modulation of neuronal coupling. Cell Biol. Int. 22, 765–780. [89] Arumugam, H., Liu, X., Colombo, P.J., Corriveau, R.A. and Belousov, A.B. (2005) NMDA receptors regulate developmental gap junction uncoupling via CREB signaling. Nat. Neurosci. 8, 1720–1726. [90] Stryker, M.P. and Harris, W.A. (1986) Binocular impulse blockade prevents the formation of ocular dominance columns in cat visual cortex. J. Neurosci. 6, 2117–2133. [91] Zhang, L.I., Tao, H.W., Holt, C.E., Harris, W.A. and Poo, M. (1998) A critical window for cooperation and competition among developing retinotectal synapses. Nature 395, 37–44. [92] Hua, J.Y., Smear, M.C., Baier, H. and Smith, S.J. (2005) Regulation of axon growth in vivo by activity-based competition. Nature 434, 1022–1026. [93] Reh, T. and Constantine-Paton, M. (1985) Eye-specific segregation requires neural activity in three-eyed Rana pipiens. J. Neurosci. 5, 1132–1143.

M.W. Baker, E.R. Macagno / FEBS Letters 588 (2014) 1470–1479 [94] Katz, L.C. and Shatz, C.J. (1996) Synaptic activity and the construction of cortical circuits. Science 274, 1133–1138. [95] Ruthazer, E.S., Akerman, C.J. and Cline, H.T. (2003) Control of axon branch dynamics by correlated activity in vivo. Science 301, 66–70. [96] Sanes, J.R. and Lichtman, J.W. (1999) Development of the vertebrate neuromuscular junction. Annu. Rev. Neurosci. 22, 389–442. [97] Personius, K.E., Chang, Q., Mentis, G.Z., O’Donovan, M.J. and Balice-Gordon, R.J. (2007) Reduced gap junctional coupling leads to uncorrelated motor neuron firing and precocious neuromuscular synapse elimination. Proc. Natl. Acad. Sci. U.S.A. 104, 11808–11813. [98] Kleopa, K.A., Orthmann-Murphy, J. and Sargiannidou, I. (2010) Gap junction disorders of myelinating cells. Rev. Neurosci. 21, 397–419. [99] Lundgaard, I., Osório, M.J., Kress, B., Sanggaard, S. and Nedergaard, M. (2013) White matter astrocytes in health and disease. Neuroscience, S0306–S4522. [100] Shelley, J., Dedek, K., Schubert, T., Feigenspan, A., Schultz, K., Hombach, S., Willecke, K. and Weiler, R. (2006) Horizontal cell receptive fields are reduced in connexin57-deficient mice. Eur. J. Neurosci. 23, 3176–3186. [101] Ishikawa, E.T. and Cancelas, J.A. (2012) Lack of communication rusts and ages stem cells. Cell Cycle 11, 3149–3150. [102] Hartfield, E.M., Rinaldi, F., Glover, C.P., Wong, L.F., Caldwell, M.A. and Uney, J.B. (2011) Connexin 36 expression regulates neuronal differentiation from neural progenitor cells. PLoS ONE 6, e14746. [103] Peiris, T.H. and Oviedo, N.J. (2013) Gap junction proteins: master regulators of the planarian stem cell response to tissue maintenance and injury. Biochim. Biophys. Acta 1828, 109–117. [104] Bentley, D. and Keshishian, H. (1982) Pathfinding by peripheral pioneer neurons in grasshoppers. Science 278, 1082–1088. [105] Hsieh, Y.W., Chang, C. and Chuang, C.F. (2012) The microRNA mir-71 inhibits calcium signaling by targeting the TIR-1/Sarm1 adaptor protein to control stochastic L/R neuronal asymmetry in C. elegans. PLoS Genet. 8, e1002864. [106] Cusato, K., Bosco, A., Rozental, R., Guimares, C.A., Reese, B.E., Linden, R. and Spray, D.C. (2003) Gap junctions mediate bystander cell death in developing retina. J. Neurosci. 23, 6413–6422. [107] Liu, T., Li, M., Zhang, Y., Pang, Z., Xiao, W., Yang, Y. and Luo, K. (2013) A role for Innexin2 and Innexin3 proteins from Spodoptera litura in apoptosis. PLoS ONE 8, e70456. [108] Coburn, C., Allman, E., Mahanti, P., Benedetto, A., Cabreiro, F., Pincus, Z., Matthijssens, F., Araiz, C., Mandel, A., Vlachos, M., Edwards, S.A., Fischer, G., Davidson, A., Pryor, R.E., Stevens, A., Slack, F.J., Tavernarakis, N., Braeckman, B.P., Schroeder, F.C., Nehrke, K. and Gems, D. (2013) Anthranilate fluorescence marks a calcium-propagated necrotic wave that promotes organismal death in C. elegans. PLoS Biol. 11, e1001613. [109] Nijhawan, D., Honarpour, N. and Wang, X. (2000) Apoptosis in neural development and disease. Annu. Rev. Neurosci. 23, 73–87. [110] Grueber, W.B. and Sagasti, A. (2010) Self-avoidance and tiling: mechanisms of dendrite and axon spacing. Cold Spring Harb. Perspect. Biol. 2, a001750. [111] Gao, W.Q. and Macagno, E.R. (1987) Extension and retraction of axonal projections by some developing neurons in the leech depends upon the existence of neighboring homologues. I. The HA cells. J. Neurobiol. 18, 43–59. [112] Gao, W.Q. and Macagno, E.R. (1987) Extension and retraction of axonal projections by some developing neurons in the leech depends upon the existence of neighboring homologues. II. The AP and AE neurons. J. Neurobiol. 118, 295–313. [113] Wolszon, L.R., Rehder, V., Kater, S.B. and Macagno, E.R. (1994) Calcium wavefronts that cross gap-junctions may signal neuronal death during development. J. Neurosci. 14, 3437–3448. [114] Wolszon, L.R., Gao, W.Q., Passani, M.B. and Macagno, E.R. (1994) Growth cone ‘‘collapse’’ in vivo: are inhibitory interactions mediated by gap junctions? J. Neurosci. 14, 999–1010. [115] McGlade-McCulloh, E. and Muller, K.J. (1989) Developing axons continue to grow at their tip after synapsing with their appropriate target. Neuron 2, 1063–1068. [116] Kramer, A.P. and Stent, G.S. (1985) Developmental arborization of sensory neurons in the leech Haementeria ghilianii. II. Experimentally induced variations in the branching pattern. J. Neurosci. 5, 768–775. [117] Blackshaw, S.E., Nicholls, J.G. and Parnas, I. (1982) Expanded receptive fields of cutaneous mechanoreceptor cells after single neurone deletion in leech central nervous system. J. Physiol. 326, 261–268. [118] Gan, W.B. and Macagno, E.R. (1995) Interactions between segmental homologs and between isoneuronal branches guide the formation of sensory terminal fields. J. Neurosci. 15, 3243–3253. [119] Grueber, W.B., Jan, L.Y. and Jan, Y.N. (2002) Tiling of the Drosophila epidermis by multidendritic sensory neurons. Development 129, 2867–2878. [120] Perry, V.H. and Linden, R. (1982) Evidence for dendritic competition in the developing retina. Nature 297, 683–685. [121] Huckfeldt, R.M., Schubert, T., Morgan, J.L., Godinho, L., Di Cristo, G., Huang, Z.J. and Wong, R.O. (2009) Transient neurites of retinal horizontal cells exhibit columnar tiling via homotypic interactions. Nat. Neurosci. 12, 35–43.

1479

[122] Depriest, A., Phelan, P. and Martha Skerrett, I. (2011) Tryptophan scanning mutagenesis of the first transmembrane domain of the innexin ShakingB(Lethal). Biophys. J. 101, 2408–2416. [123] Völgyi, B., Kovács-Oller, T., Atlasz, T., Wilhelm, M. and Gábriel, R. (2013) Gap junctional coupling in the vertebrate retina: variations on one theme? Prog. Retin. Eye Res. 34, 1–18. [124] Mansour, H., McColm, J.R., Cole, L., Weible 2nd, M., Korlimbinis, A. and ChanLing, T. (2013) Connexin 30 expression and frequency of connexin heterogeneity in astrocyte gap junction plaques increase with age in the rat retina. PLoS ONE 8, e57038. [125] Dedek, K., Schultz, K., Pieper, M., Dirks, P., Maxeiner, S., Willecke, K., Weiler, R. and Janssen-Bienhold, U. (2006) Localization of heterotypic gap junctions composed of connexin45 and connexin36 in the rod pathway of the mouse retina. Eur. J. Neurosci. 24, 1675–1686. [126] Janssen-Bienhold, U., Trümpler, J., Hilgen, G., Schultz, K., Müller, L.P., Sonntag, S., Dedek, K., Dirks, P., Willecke, K. and Weiler, R. (2009) Connexin57 is expressed in dendro-dendritic and axo-axonal gap junctions of mouse horizontal cells and its distribution is modulated by light. J. Comp. Neurol. 513, 363–374. [127] Ochalski, P.A., Frankenstein, U.N., Hertzberg, E.L. and Nagy, J.I. (1997) Connexin-43 in rat spinal cord: localization in astrocytes and identification of heterotypic astro-oligodendrocytic gap junctions. Neuroscience 76, 931– 945. [128] Lee, S.C., Cowgill, E.J., Al-Nabulsi, A., Quinn, E.J., Evans, S.M. and Reese, B.E. (2011) Homotypic regulation of neuronal morphology and connectivity in the mouse retina. J. Neurosci. 31, 14126–14133. [129] Fujishima, K., Horie, R., Mochizuki, A. and Kengaku, M. (2012) Principles of branch dynamics governing shape characteristics of cerebellar Purkinje cell dendrites. Development 139, 3442–3455. [130] Rouach, N., Koulakoff, A., Abudara, V., Willecke, K. and Giaume, C. (2008) Astroglial metabolic networks sustain hippocampal synaptic transmission. Science 322, 1551–1555. [131] Meller, K., Krah, K. and Theiss, C. (2005) Dye coupling in Purkinje cells of organotypic slice cultures. Brain Res. Dev. Brain Res. 160, 101–105. [132] van Welie, I., Smith, I.T. and Watt, A.J. (2011) The metamorphosis of the developing cerebellar microcircuit. Curr. Opin. Neurobiol. 21, 245–253. [133] De Zeeuw, C.I., Chorev, E., Devor, A., Manor, Y., Van Der Giessen, R.S., De Jeu, M.T., Hoogenraad, C.C., Bijman, J., Ruigrok, T.J., French, P., Jaarsma, D., Kistler, W.M., Meier, C., Petrasch-Parwez, E., Dermietzel, R., Sohl, G., Gueldenagel, M., Willecke, K. and Yarom, Y. (2003) Deformation of network connectivity in the inferior olive of connexin 36-deficient mice is compensated by morphological and electrophysiological changes at the single neuron level. J. Neurosci. 23, 4700–4711. [134] De Zeeuw, C.I., Hertzberg, E.L. and Mugnaini, E. (1995) The dendritic lamellar body: a new neuronal organelle putatively associated with dendro-dendritic gap junctions. J. Neurosci. 15, 1587–1604. [135] Sotelo, C., Llinás, R. and Baker, R. (1974) Structural study of inferior olivary nucleus of the cat: morphological correlates of electrotonic coupling. J. Neurophysiol. 37, 541–559. [136] Phelan, P., Bacon, J.P., Davies, J.A., Stebbings, L.A., Toodmand, M.G., Avery, L., Baines, R.A., Barnes, T.M., Ford, C., Hekimi, S., Lee, R., Shaw, J.E., Starich, T.A., Curtin, K.D., Sun, Y. and Wyaman, R.J. (1998) Innexins, a family of invertebrate gap-junction proteins. Trends Genet. 14, 348–349. [137] Yen, M.R. and Saier Jr., M.H. (2007) Gap junctional proteins of animals: the innexin/pannexin superfamily. Prog. Biophys. Mol. Biol. 94, 5–14. [138] Retamal, M.A., Cortés, C.J., Reuss, L., Bennett, M.V. and Sáez, J.C. (2006) Snitrosylation and permeation through connexin 43 hemichannels in astrocytes: induction by oxidant stress and reversal by reducing agents. Proc. Natl. Acad. Sci. U.S.A. 103, 4475–4480. [139] Samuels, S.E., Lipitz, J.B., Dahl, G. and Muller, K.J. (2010) Neuroglial ATP release through innexin channels controls microglial cell movement to a nerve injury. J. Gen. Physiol. 136, 425–442. [140] Johnson, R.G., Reynhout, J.K., TenBroek, E.M., Quade, B.J., Yasumura, T., Davidson, K.G., Sheridan, J.D. and Rash, J.E. (2012) Gap junction assembly: roles for the formation plaque and regulation by the C-terminus of connexin 43. Mol. Biol. Cell 23, 71–86. [141] Shyu, Y.J., Suarez, C.D. and Hu, C.D. (2008) Visualization of ternary complexes in living cells by using a BiFC-based FRET assay. Nat. Protoc. 3, 1693–1702. [142] Chen, B., Liu, Q., Ge, Q., Xie, J. and Wang, Z.W. (2007) UNC-1 regulates gap junctions important to locomotion in C. elegans. Curr. Biol. 17, 1334–1339. [143] Kanamori, T., Kanai, M.I., Dairyo, Y., Yasunaga, K., Morikawa, R.K. and Emoto, K. (2013) Compartmentalized calcium transients trigger dendrite pruning in Drosophila sensory neurons. Science 340, 1475–1478. [144] Mizumoto, K. and Shen, K. (2013) Interaxonal interaction defines tiled presynaptic innervation in C. elegans. Neuron 77, 655–666.