Dye-coupling visualizes networks of large-field motion-sensitive neurons in the fly

In the fly, visually guided course control is accomplished by a set of 60 large-field motion-sensitive neurons in each brain hemisphere. These neurons have been shown to receive retinotopic motion information from local motion detectors on their dendrites. In addition, recent experiments revealed extensive coupling between the large-field neurons through electrical synapses. These two processes together give rise to their broad and elaborate receptive fields significantly surpassing the extent of their dendritic fields. Here, we demonstrate that the electrical connections between different large-field neurons can be visualized using Neurobiotin dye injection into a single one of them. When combined with a fluorescent dye which does not cross electrical synapses, the injected cell can be identified unambiguously. The Neurobiotin staining corroborates the electrical coupling postulated amongst the cells of the vertical system (VS-cells) and between cells of the horizontal system (HS-cells and CH-cells). In addition, connections between some cells are revealed that have so far not been considered as electrically coupled.     

Gap junctional communication in the post-implantation mouse embryo.

We studied the extent of cell-to-cell communication via junctional channels in in vitro-implanted mouse blastocysts by monitoring ionic coupling and the spread of two injected low molecular weight dyes, fluorescein and Lucifer yellow. In the early attached embryos, both trophoblasts and cells of the inner cell mass (ICM) were ionically coupled to one another. Dye injections in either trophoblasts or ICM cells resulted in spread to the entire embryo. As older and more developed embryos were examined, the spread of injected dye was progressively more limited. In the most developed embryos examined, dye injected into a cell in the ICM region resulted in spread throughout the ICM but not into the surrounding trophoblast cells, while dye injected into a trophoblast cell did not spread to any other cell in the embryo. Simultaneous monitoring of ionic coupling and dye injections in embryos of intermediate stages in this transition revealed that the trophoblast and ICM cells were ionically coupled, even across the apparent boundary where no dye was observed to pass. In the latest stage embryos examined in which no injected dye was observed to move out of the ICM, ionic coupling was still observed between the cells of the ICM and the trophoblasts. Furthermore, in the more developed embryos, dye injected into the ICM region frequently was not transferred to all the cells of the ICM, thus suggesting a further compartmentalization of due spread within the ICM. Our observations that ionic coupling is more extensive than the detectable spread of injected dyes may perhaps reflect a reduced number of junctional channels. With fewer channels less dye would pass between cells, so that, together with continuous quenching, the transfer of injected dye would not be detectable. This partial segregation of cell-to-cell communication as indicated by the limited dye spread may parallel specific differentiation processes, in particular that of giant trophoblast, embryonic ectoderm and extraembryonic endoderm differentiation.     

Rhythmic coupling among cells in the suprachiasmatic nucleus

In mammals, the part of the nervous system responsible for most circadian behavior can be localized to a pair of structures in the hypothalamus known as the suprachiasmatic nucleus (SCN). Previous studies suggest that the basic mechanism responsible for the generation of these rhythms is intrinsic to individual cells. There is also evidence that the cells within the SCN are coupled to one another and that this coupling is important for the normal functioning of the circadian system. One mechanism that mediates coordinated electrical activity is direct electrical connections between cells formed by gap junctions. In the present study, we used a brain slice preparation to show that developing SCN cells are dye coupled. Dye coupling was observed in both the ventrolateral and dorsomedial subdivisions of the SCN and was blocked by application of a gap junction inhibitor, halothane. Dye coupling in the SCN appears to be regulated by activity-dependent mechanisms as both tetrodotoxin and the GABA(A) agonist muscimol inhibited the extent of coupling. Furthermore, acute hyperpolarization of the membrane potential of the original biocytin-filled neuron decreased the extent of coupling. SCN cells were extensively dye coupled during the day when the cells exhibit synchronous neural activity but were minimally dye coupled during the night when the cells are electrically silent. Immunocytochemical analysis provides evidence that a gap-junction-forming protein, connexin32, is expressed in the SCN of postnatal animals. Together the results are consistent with a model in which gap junctions provide a means to couple SCN neurons on a circadian basis.     

Electrical coupling and uncoupling of exocrine acinar cells

The electrical communication network in the mouse pancreatic acinar tissue has been investigated using simultaneous intracellular recording with two separate microelectrodes and direct microscopical control of the localizations of the microelectrode tips. All cells within one acinus were electrically coupled, and the coupling coefficient (the electrotonic potential change in a cell neighboring to the cell into which current is injected [V2] divided by the electrotonic potential change in the cell of current injection [V1]) between two cells near each other (less than 50 micron) was always close to 1. Cells farther apart (50-100 micron) were, in some cases, coupled; in other cases, there was no coupling at all. Coupling coefficients varied between 0 and 1. There was rarely electrical coupling over distances of more than 110 micron. Using microiontophoretic acetylcholine (ACh) application, it was possible to evoke almost complete electrical uncoupling of two previously coupled pancreatic or lacrimal acinar cells from different acini or within one acinus. The effects were fully and quickly reversible. While the ACh-evoked uncoupling in the pancreas was associated with membrane depolarization, ACh caused hyperpolarization in the lacrimal acinar cells. The uncoupling was associated with a very marked reduction in electrical time constant, indicating a reduction in input capacitance (effective surface cell membrane area). The concentrations of stimulants needed to evoke reduction in pancreatic cell-to-cell coupling were 1 micron for ACh, 0.14 nM for caerulein, and 3 nM for bombesin. These concentrations are smaller than those required to evoke maximal enzyme secretion.     

Embryonic electrical connections appear to prefigure a behavioral circuit in the leech CNS

During development, many embryos show electrical coupling among neurons that is spatially and temporally regulated. For example, in vertebrate embryos extensive dye coupling is seen during the period of circuit formation, suggesting that electrical connections could prefigure circuits, but it has been difficult to identify which neuronal types are coupled. We have used the leech Hirudo medicinalis to follow the development of electrical connections within the circuit that produces local bending. This circuit consists of three layers of neurons: four mechanosensory neurons (P cells), 17 identified interneurons, and approximately 24 excitatory and inhibitory motor neurons. These neurons can be identified in embryos, and we followed the spatial and temporal dynamics as specific connections developed. Injecting Neurobiotin into identified cells of the circuit revealed that electrical connections were established within this circuit in a precise manner from the beginning. Connections first appeared between motor neurons; mechanosensory neurons and interneurons started to connect at least a day later. This timing correlates with the development of behaviors, so the pattern of emerging connectivity could explain the appearance first of spontaneous behaviors (driven by a electrically coupled motor network) and then of evoked behaviors (when sensory neurons and interneurons are added to the circuit).     

A dye mixture (Neurobiotin and Alexa 488) reveals extensive dye-coupling among neurons in leeches; physiology confirms the connections

Although the neuronal circuits that generate leech movements have been studied for over 30 years, the list of interneurons (INs) in these circuits remains incomplete. Previous studies showed that some motor neurons (MNs) are electrically coupled to swim-related INs, e.g., rectifying junctions connect IN 28 to MN DI-1 (dorsal inhibitor), so we searched for additional neurons in these behavioral circuits by co-injecting Neurobiotin and Alexa Fluor 488 into segmental MNs DI–1, VI–2, DE–3 and VE–4. The high molecular weight Alexa dye is confined to the injected cell, whereas the smaller Neurobiotin molecules diffuse through gap junctions to reveal electrical coupling. We found that MNs were each dye-coupled to approximately 25 neurons, about half of which are likely to be INs. We also found that (1) dye-coupling was reliably correlated with physiologically confirmed electrical connections, (2) dye-coupling is unidirectional between MNs that are linked by rectifying connections, and (3) there are novel electrical connections between excitatory and inhibitory MNs, e.g. between excitatory MN VE-4 and inhibitory MN DI-1. The INs found in this study provide a pool of novel candidate neurons for future studies of behavioral circuits, including those underlying swimming, crawling, shortening, and bending movements.     

Functional gap junctions in the early sea urchin embryo are localized to the vegetal pole.

Using the whole-cell voltage-clamp technique we have studied electrical coupling and dye coupling between pairs of blastomeres in 16- to 128-cell-stage sea urchin embryos. Electrical coupling was established between macromeres and micromeres at the 16-cell stage with a junctional conductance (G(j)) of 26 nS that decreased to 12 nS before the next cleavage division. G(j) between descendants of macromeres and micromeres was 12 nS falling to 8 nS in the latter half of the cell cycle. Intercellular current intensity was independent of transjunctional voltage, nondirectional, and sensitive to 1-octanol and therefore appears to be gated through gap junction channels. There was no significant coupling between other pairs of blastomeres. Lucifer yellow did not spread between these electrically coupled cell pairs and in fact significant dye coupling between nonsister cells was observed only at the 128-cell stage. Since 1-octanol inhibited electrical communication between blastomeres at the 16- to 64-cell stage and also induced defects in formation of the archenteron, it is possible that gap junctions play a role in embryonic induction.     

Morphological and electrophysiological changes in mouse dorsal root ganglia after partial colonic obstruction

There is evidence that sensitization of neurons in dorsal root ganglia (DRG) may contribute to pain induced by intestinal injury. We hypothesized that obstruction-induced pain is related to changes in DRG neurons and satellite glial cells (SGCs). In this study, partial colonic obstruction was induced by ligation. The neurons projecting to the colon were traced by an injection of 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate into the colon wall. The electrophysiological properties of DRG neurons were determined using intracellular electrodes. Dye coupling was examined with an intracellular injection of Lucifer yellow (LY). Morphological changes in the colon and DRG were examined. Pain was assessed with von Frey hairs. Partial colonic obstruction caused the following changes. First, coupling between SGCs enveloping different neurons increased 18-fold when LY was injected into SGCs near neurons projecting to the colon. Second, neurons were not coupled to other neurons or SGCs. Third, the firing threshold of neurons projecting to the colon decreased by more than 40% (P < 0.01), and the resting potential was more positive by 4-6 mV (P < 0.05). Finally, the number of neurons displaying spontaneous spikes increased eightfold, and the number of neurons with subthreshold voltage oscillations increased over threefold. These changes are consistent with augmented neuronal excitability. The pain threshold to abdominal stimulation decreased by 70.2%. Inflammatory responses were found in the colon wall. We conclude that obstruction increased neuronal excitability, which is likely to be a major factor in the pain behavior observed. The augmented dye coupling between glial cells may contribute to the neuronal hyperexcitability.     

Dye and electrical coupling of endothelial cells in situ

Electron microscopic studies show that endothelial cells of pig coronary arteries are linked by gap junctions. We investigated the dye and electrical coupling of these junctions in a strip of pig coronary artery in vitro. The membrane potential of two neighbouring (about 0.2 mm) endothelial cells were simultaneously recorded with two microelectrodes. The fluorescent dye lucifer yellow was microiontophoretically injected through one of the microelectrodes. The endothelial cells in situ were dye and electrically coupled. The dye coupling extended parallel to the longitudinal axis of the arteries. We conclude that an electrical message like the bradykinin and substance P hyperpolarizations of the endothelial cells can be conveyed electrotonically by the endothelium along the longitudinal axis of arteries.     

Electrical Transmission between Mammalian Neurons is Supported by a Small Fraction of Gap Junction Channels

Electrical synapses formed by gap junctions between neurons create networks of electrically coupled neurons in the mammalian brain, where these networks have been found to play important functional roles. In most cases, interneuronal gap junctions occur at remote dendro-dendritic contacts, making difficult accurate characterization of their physiological properties and correlation of these properties with their anatomical and morphological features of the gap junctions. In the mesencephalic trigeminal (MesV) nucleus where neurons are readily accessible for paired electrophysiological recordings in brain stem slices, our recent data indicate that electrical transmission between MesV neurons is mediated by connexin36 (Cx36)-containing gap junctions located at somato-somatic contacts. We here review evidence indicating that electrical transmission between these neurons is supported by a very small fraction of the gap junction channels present at cell-cell contacts. Acquisition of this evidence was enabled by the unprecedented experimental access of electrical synapses between MesV neurons, which allowed estimation of the average number of open channels mediating electrical coupling in relation to the average number of gap junction channels present at these contacts. Our results indicate that only a small proportion of channels (~0.1?%) appear to be conductive. On the basis of similarities with other preparations, we postulate that this phenomenon might constitute a general property of vertebrate electrical synapses, reflecting essential aspects of gap junction function and maintenance.     

Cross-connections coordinate postural motoneuron activity in the crayfish abdomen

Intracellular recordings and dye injections were used to examine mutual coupling among slow abdominal postural motoneurons in the 4th abdominal ganglion in crayfish (Procambarus clarkii). Intracellular current injection into one motoneuron altered the spike firing rate of some of its synergists. Depending on the polarity of the injected current, the premotor effect on the synergists was excitatory or inhibitory. The magnitude of the effect was intensity dependent. No dye coupling was found among the motoneurons following injection of Lucifer yellow. The morphological basis of the coupling was examined by differential filling of motoneuron pairs, one with horseradish peroxidase and the other with Lucifer yellow. The stained motoneurons were simultaneously visualized under light microscopy to determine the proximity of their differently colored dendrites. It was thus possible to locate the site of the presumed monosynaptic contacts between them. Combined physiological and morphological evidence suggests that these neurons are mutually coupled, forming part of an integrative system for abdominal posture control in crayfish.     

Dye and electrical coupling of endothelial cells in situ

Electron microscopic studies show that endothelial cells of pig coronary arteries are linked by gap junctions. We investigated the dye and electrical coupling of these junctions in a strip of pig coronary artery in vitro. The membrane potential of two neighbouring (about 0.2 mm) endothelial cells were simultaneously recorded with two microelectrodes. The fluorescent dye lucifer yellow was microiontophoretically injected through one of the microelectrodes. The endothelial cells in situ were dye and electrically coupled. The dye coupling extended parallel to the longitudinal axis of the arteries. We conclude that an electrical message like the bradykinin and substance P hyperpolarizations of the endothelial cells can be conveyed electrotonically by the endothelium along the longitudinal axis of arteries.     

Pseudorabies Virus Infection Alters Neuronal Activity and Connectivity In Vitro

Alpha-herpesviruses, including human herpes simplex virus 1 & 2, varicella zoster virus and the swine pseudorabies virus (PRV), infect the peripheral nervous system of their hosts. Symptoms of infection often include itching, numbness, or pain indicative of altered neurological function. To determine if there is an in vitro electrophysiological correlate to these characteristic in vivo symptoms, we infected cultured rat sympathetic neurons with well-characterized strains of PRV known to produce virulent or attenuated symptoms in animals. Whole-cell patch clamp recordings were made at various times after infection. By 8 hours of infection with virulent PRV, action potential (AP) firing rates increased substantially and were accompanied by hyperpolarized resting membrane potentials and spikelet-like events. Coincident with the increase in AP firing rate, adjacent neurons exhibited coupled firing events, first with AP-spikelets and later with near identical resting membrane potentials and AP firing. Small fusion pores between adjacent cell bodies formed early after infection as demonstrated by transfer of the low molecular weight dye, Lucifer Yellow. Later, larger pores formed as demonstrated by transfer of high molecular weight Texas red-dextran conjugates between infected cells. Further evidence for viral-induced fusion pores was obtained by infecting neurons with a viral mutant defective for glycoprotein B, a component of the viral membrane fusion complex. These infected neurons were essentially identical to mock infected neurons: no increased AP firing, no spikelet-like events, and no electrical or dye transfer. Infection with PRV Bartha, an attenuated circuit-tracing strain delayed, but did not eliminate the increased neuronal activity and coupling events. We suggest that formation of fusion pores between infected neurons results in electrical coupling and elevated firing rates, and that these processes may contribute to the altered neural function seen in PRV-infected animals.     

Communication compartments in the gastrulating mouse embryo

We characterized the pattern of gap junctional communication in the 7.5-d mouse embryo (at the primitive streak or gastrulation stage). First we examined the pattern of dye coupling by injecting the fluorescent tracers, Lucifer Yellow or carboxyfluorescein, and monitoring the extent of dye spread. These studies revealed that cells within all three germ layers are well coupled, as the injected dye usually spread rapidly from the site of impalement into the neighboring cells. The dye spread, however, appeared to be restricted at specific regions of the embryo. Further thick section histological analysis revealed little or no dye transfer between germ layers, indicating that each is a separate communication compartment. The pattern of dye movement within the embryonic ectoderm and mesoderm further suggested that cells in each of these germ layers may be subdivided into smaller communication compartments, the most striking of which are a number of "box-like" domains. Such compartments, unlike the restrictions observed between germ layers, are consistently only partially restrictive. In light of these results, we further monitored ionic coupling to determine if some coupling might nevertheless persist between germ layers. For these studies, Lucifer Yellow was coinjected while ionic coupling was monitored. The injected Lucifer Yellow facilitated the identification of the impalement sites, both in the live specimen and in thick sections in the subsequent histological analysis. By using this approach, all three germ layers were shown to be ionically coupled, indicating that gap junctional communication is maintained across the otherwise dye-uncoupled "germ layer compartments." Thus our results demonstrate that partially restrictive communication compartments are associated with the delamination of germ layers in the gastrulating mouse embryo. The spatial distribution of these compartments are consistent with a possible role in the underlying development.     

Cultured astrocytes from a syncytium after maturation

The formation of functional gap junctions between astrocytes was investigated during differentiation of these cells in culture. Precursor cells of GFA (glial fibrillary acidic) protein-positive astrocytes were cultured in a chemically defined medium as a homogeneous population. These cells were rarely coupled to one neighbour, as revealed by electrical and dye coupling and never formed a large syncytium, as investigated by injection and spread of Lucifer Yellow. Differentiation with respect to GFA protein accumulation can be induced in these cells by culturing in horse serum-containing medium. The formation of functional junctions developed within 2 weeks in about 20% of the cells. Coupled cells formed a large syncytium. When the astrocytes were co-cultured with primary cerebellar cells (consisting predominantly of small neurons) after the switch to serum-containing medium the percentage of coupled astrocytes increased to about 65%. Again the coupled cells formed a large syncytium. Since no physical contact was possible between the astrocyte cultures and the primary cerebellar cells the stimulation of coupling had to be signalized by soluble factor(s).     

Loss of responsiveness of circular smooth muscle cells from the guinea pig ileum is associated with changes in gap junction coupling

Gap junction coupling and neuromuscular transmission to smooth muscle were studied in the first 4 h after preparations were set up in vitro. Intracellular recordings were made from smooth muscle cells of guinea pig ileum. Fast inhibitory junction potentials (IJPs) were small (1.3 ± 1.0 mV) in the first 30 min but increased significantly over the first 120 min to 15.8 ± 0.9 mV (n = 12, P < 0.001). Comparable increases in slow IJPs and excitatory junction potentials were also observed. During the same period, resting membrane potential depolarized from -58.8 ± 1.4 to -47.2 ± 0.4 mV (n = 12, P < 0.001). Input resistance, estimated by intracellular current injection, decreased in parallel (P < 0.05), and dye coupling, measured by intracellular injection of carboxyfluorescein, increased (P < 0.001). Input resistance was higher and dye coupling was less in longitudinal than circular smooth muscle cells. Gap junction blockers [carbenoxolone (100 μM), 18β-glycyrrhetinic acid (10 μM), and 2-aminoethoxydiphenyl borate (50 μM)] hyperpolarized coupled circular smooth muscle cells, reduced the amplitude of fast and slow IJPs and excitatory junction potentials, increased input resistance, and reduced dye coupling. Local application of ATP (10 mM) mimicked IJPs and showed comparable increases in amplitude over the first 120 min; carbenoxolone and 2-aminoethoxydiphenyl borate significantly reduced ATP-evoked hyperpolarizations in coupled cells. In contrast, synaptic transmission between myenteric neurons was not suppressed during the first 30 min. Gap junction coupling between circular smooth muscle cells in isolated preparations was initially disrupted but recovered over the next 120 min to a steady level. This was associated with potent effects on neuromuscular transmission and responses to exogenous ATP.     

Transegmental electrical coupling between adjacent intersegemental muscles in the tobacco hawkmoth (Manduca sexta)

The lateral intersegmental muscles of pharate adult tobacco hawkmoths (Manduca sexta), exhibited electrical coupling across the segmental boundary. The degree of electrical coupling was constant throughout adult development. These muscle fibres did not appear to be dye coupled in that neither cobalt ions nor the flourescent dye Lucifer Yellow CH passed between cells. Electrical coupling was unaffected by cellular acidification with CO₂. Data are presented which suggest that this electrical coupling may be through the extracellular space rather than some membrane specialization. It is further speculated that many invertebrates may show this form of electrical coupling due to the metameric architecture of certain skeletal muscles.     

Reduced junctional permeability at interrhombomeric boundaries.

Intercellular communication is considered to have a role during pattern specification processes in early embryonic development. This report analyzes the changing gap junctional communication properties of chick neuroepithelial cells depending on their position relative to the segmental partitions of the rhombencephalon. Intercellular electrical coupling and dye transfer were studied with microelectrode techniques. Neuroepithelial cells were electrically coupled irrespective of their location relative to interneuromeric boundaries. Iontophoretic injection of biocytin or Lucifer Yellow into single cells inside the rhombomeres was followed by transjunctional diffusion to the surrounding cells. In contrast, dye transfer was strictly limited when the diffusion zone contacted the cells forming the interneuromeric limits. Label injected into the boundary cells did not spread to other cells at all. Avian interrhombomeric boundaries are thus sites of reduced junctional permeability during early morphogenesis.     

Cut-loading: A Useful Tool for Examining the Extent of Gap Junction Tracer Coupling Between Retinal Neurons

In addition to chemical synaptic transmission, neurons that are connected by gap junctions can also communicate rapidly via electrical synaptic transmission. Increasing evidence indicates that gap junctions not only permit electrical current flow and synchronous activity between interconnected or coupled cells, but that the strength or effectiveness of electrical communication between coupled cells can be modulated to a great extent^1,2. In addition, the large internal diameter (~1.2 nm) of many gap junction channels permits not only electric current flow, but also the diffusion of intracellular signaling molecules and small metabolites between interconnected cells, so that gap junctions may also mediate metabolic and chemical communication. The strength of gap junctional communication between neurons and its modulation by neurotransmitters and other factors can be studied by simultaneously electrically recording from coupled cells and by determining the extent of diffusion of tracer molecules, which are gap junction permeable, but not membrane permeable, following iontophoretic injection into single cells. However, these procedures can be extremely difficult to perform on neurons with small somata in intact neural tissue.Numerous studies on electrical synapses and the modulation of electrical communication have been conducted in the vertebrate retina, since each of the five retinal neuron types is electrically connected by gap junctions^3,4. Increasing evidence has shown that the circadian (24-hour) clock in the retina and changes in light stimulation regulate gap junction coupling^3-8. For example, recent work has demonstrated that the retinal circadian clock decreases gap junction coupling between rod and cone photoreceptor cells during the day by increasing dopamine D2 receptor activation, and dramatically increases rod-cone coupling at night by reducing D2 receptor activation^7,8. However, not only are these studies extremely difficult to perform on neurons with small somata in intact neural retinal tissue, but it can be difficult to adequately control the illumination conditions during the electrophysiological study of single retinal neurons to avoid light-induced changes in gap junction conductance.Here, we present a straightforward method of determining the extent of gap junction tracer coupling between retinal neurons under different illumination conditions and at different times of the day and night. This cut-loading technique is a modification of scrape loading^9-12, which is based on dye loading and diffusion through open gap junction channels. Scrape loading works well in cultured cells, but not in thick slices such as intact retinas. The cut-loading technique has been used to study photoreceptor coupling in intact fish and mammalian retinas^{7, 8,13}, and can be used to study coupling between other retinal neurons, as described here.     

Modelling the Effects of Electrical Coupling between Unmyelinated Axons of Brainstem Neurons Controlling Rhythmic Activity

Gap junctions between fine unmyelinated axons can electrically couple groups of brain neurons to synchronise firing and contribute to rhythmic activity. To explore the distribution and significance of electrical coupling, we modelled a well analysed, small population of brainstem neurons which drive swimming in young frog tadpoles. A passive network of 30 multicompartmental neurons with unmyelinated axons was used to infer that: axon-axon gap junctions close to the soma gave the best match to experimentally measured coupling coefficients; axon diameter had a strong influence on coupling; most neurons were coupled indirectly via the axons of other neurons. When active channels were added, gap junctions could make action potential propagation along the thin axons unreliable. Increased sodium and decreased potassium channel densities in the initial axon segment improved action potential propagation. Modelling suggested that the single spike firing to step current injection observed in whole-cell recordings is not a cellular property but a dynamic consequence of shunting resulting from electrical coupling. Without electrical coupling, firing of the population during depolarising current was unsynchronised; with coupling, the population showed synchronous recruitment and rhythmic firing. When activated instead by increasing levels of modelled sensory pathway input, the population without electrical coupling was recruited incrementally to unpatterned activity. However, when coupled, the population was recruited all-or-none at threshold into a rhythmic swimming pattern: the tadpole “decided” to swim. Modelling emphasises uncertainties about fine unmyelinated axon physiology but, when informed by biological data, makes general predictions about gap junctions: locations close to the soma; relatively small numbers; many indirect connections between neurons; cause of action potential propagation failure in fine axons; misleading alteration of intrinsic firing properties. Modelling also indicates that electrical coupling within a population can synchronize recruitment of neurons and their pacemaker firing during rhythmic activity.