Fibroblast–myocyte electrotonic coupling: Does it occur in native cardiac tissue?

Abstract

Heterocellular electrotonic coupling between cardiac myocytes and non-excitable connective tissue cells has been a long-established and well-researched fact in vitro. Whether or not such coupling exists in vivo has been a matter of considerable debate. This paper reviews the development of experimental insight and conceptual views on this topic, describes evidence in favour of and against the presence of such coupling in native myocardium, and identifies directions for further study needed to resolve the riddle, perhaps less so in terms of principal presence which has been demonstrated, but undoubtedly in terms of extent, regulation, patho-physiological context, and actual relevance of cardiac myocyte–non-myocyte coupling in vivo. This article is part of a Special Issue entitled "Myocyte-Fibroblast Signalling in Myocardium."

Highlights

• •Electrical coupling of cardiomyocytes and fibroblasts is well-established in vitro
• •Whether such hetero-cellular coupling exists in vivo has been a matter of debate
• •We review the development of experimental and conceptual insight into the topic
• •Conclusion 1: hetero-cellular coupling in heart tissue has been shown in principle
• •Conclusion 2: extent, regulation, context, and relevance remain to be established

1. Heterocellular coupling between cardiac cells in vivo?

When we consider the cellular basis of cardiac function, we tend to focus on myocytes, ‘forgetting’ the seemingly silent majority of electrically non-excitable cells. These include endothelial, fat, immune and stem cells, but the largest sub-population is formed by connective tissue cells (fibroblasts). Non-excitable does not mean ‘not exciting’, however, as these cells are crucial for structural, biochemical, and electro-mechanical integrity of the heart [1–3].

1.1. ‘The’ cardiac fibroblast?

Until the 1990s, cardiac fibroblasts were largely considered to be of limited relevance beyond structural support (a bit like the traditional view on neuroglia). That changed with the discovery of fibroblast-mediated signalling two decades ago, and led to a step-increase in the number of publications on cardiac fibroblasts from tens or fewer to hundreds per year (roughly 99% of all cardiac fibroblast papers currently listed on Web of Science have been published after 1990). As a result of this surge, fibroblasts are now accepted as key contributors to development, adaptation, and disease-related remodelling of the heart (e.g. [4–7]). At the same time, we are still far from having comprehensive insight into the roles of connective tissue in the heart. In part, this is caused by the fact that ‘the’ cardiac fibroblast does not exist.

Cardiac fibroblasts originate from at least three progenitor populations. Firstly, fibroblasts arise from endocardium via epithelial-to-mesenchymal transformation [8]. Secondly, retroviral lineage tracing showed that fibroblasts also originate from the pro-epicardium, a group of embryonic progenitor cells known to give rise to the epicardium [9]. Thirdly, cardiac fibroblasts derive from bone marrow in normal development and as a contributor to the homeostatic maintenance of adult myocardium [10–12], but also during post-injury scar formation [13]. Fibroblasts therefore constitute a heterogeneous and dynamic population of cells, whose origin, regulation, and function in health [14] and disease [15], including their role in repair, impact on cardiomyocyte proliferation [16] and, possibly, their transformation into cardiac muscle cells [17], pose exciting challenges of high clinical relevance. In addition to heterogeneity related to their origin, cardiac fibroblasts respond to activation, such as during myocardial infarction, thorough phenotype transformation into myofibroblasts, which are regarded by some as a distinct cell type altogether (for recent reviews see [18–21]).

This review is focussed on exploring the presence, in vivo, of electrophysiologically relevant heterocellular connections between cardiac myocytes and connective tissue cells. Given that (i) the developmental origin and (ii) the precise state of cell-activation are not usually known or reported, and since (iii) neither of these aspects is of primary concern for exploring the principal question as to whether or not there is in vivo heterocellular coupling in the heart between the excitable muscle and the non-excitable connective tissue cells, we will refer to the latter as ‘fibroblasts’, aware of the inherent limitations of such abbreviated terminology.

1.2. Properties of fibroblasts in tissue and in a dish

For roughly half a century, the presence of electrotonic coupling between cardiac fibroblasts and myocytes has been a well-established fact in vitro. Since the mid-1960s, the synchronization of spontaneous contractile activity in isolated cardiomyocytes, interconnected solely by fibroblasts, has been noted and characterised using time-lapse microscopy, microelectrode recordings, and dye transfer studies [22–24]. More recently, using linearly structured cell cultures [25–28], optical mapping of voltage sensitive dyes has shown that fibroblast inserts can electrotonically bridge gaps between groups of myocytes that are up to 300 μm apart [29].

This ability of fibroblasts to act as long-distance conductors benefits from their high membrane resistance and relatively low capacitance [30]. When considering numerical parameter ranges for fibroblast electrophysiological properties, it is imperative to appreciate pronounced differences between cells in vivo and in vitro (even prior to electrophysiological remodelling of fibroblasts in cell culture [31]).

Cardiac fibroblasts in vivo form large sheet-like extensions, often with additional irregular folds, and elongated cytoplasmic processes [32,33] (not just in mammals; see on-line movie S1 of [34] with data from fish). Careful electron microscopy (EM) based reconstruction of an individual fibroblast in rabbit sino-atrial node revealed that it formed a membrane juxtaposition with a neighbouring myocyte covering 720 μm² [33]. As this reconstruction excluded some of the more distant fibroblast extensions, total surface area of this cell will have been 1500 μm², or more.

In contrast, fibroblasts, freshly-isolated from healthy myocardium, are rounded cells with initial diameters of about 7–9 μm [31,35,36]. EM data showed that they lack not only the typical membrane extensions, but also folds or membrane invaginations that otherwise could increase surface area [36]. Therefore, considering their near-spherical cell shapes, one can calculate the surface membrane area of freshly isolated fibroblasts as 150–250 μm², an order of magnitude less than in tissue. A probable reason for this discrepancy is the fact that cardiac cell isolation protocols tend to involve a combination of enzymatic digestion (to destroy connective tissue bonds) and mechanical agitation (to disturb tissue integrity), with the effect that surviving fibroblasts are likely to be the truncated non-myocyte fragments that contain a nucleus [37].

Currently, there is no data on the cell size distribution of fibroblasts in vivo, so we don't know how typical a surface area of 1500 μm² is for fibroblasts in the heart. As a ball-park value, though, it is in keeping with the observation that fibroblast membrane resistances in situ (0.5–1 GΩ [38,39]) are generally about an order of magnitude lower than in freshly isolated cells [40,41]. It stands to reason that fibroblast membrane capacitances in vivo (hard to quantify by direct electrophysiological means in these extended and mutually interconnected cells) exceed those of freshly isolated cells (typically 6–10 pF in fibroblasts isolated from healthy myocardium [31,41]). Even if that were by an order of magnitude as well, which is not inconceivable, it would still render fibroblast capacitances small compared to cardiac myocytes (values for ventricular cardiomyocytes, isolated from healthy tissue, range from about 150 pF in rabbit and ferret to 300 pF in rat [42]).

Fibroblasts have a relatively depolarised membrane potential (usually between - 10 and - 50 mV; in tissue at the less negative end of this range), whether recorded using sharp electrodes in situ [38,39], or using single [43,44] and dual [38,40] patch clamp in vitro. While precise membrane potential measurements with these direct (but invasive) electrophysiological techniques are challenging in individual cells when cell- and seal-resistances are in a similar ball-park, the above potential range is also evident from observations based on an indirect assessment of biological reporter systems (e.g. by monitoring fibroblast effects on cultured cardiomyocytes [45–48]). Therefore, in addition to potentially supporting conduction, cardiac fibroblasts can depolarize resting, electrotonically connected cardiac myocytes [38–40,45]. Quantitative computational modelling since the mid-1990s has suggested that this could modulate pacemaker rate and alter excitability of working cardiomyocytes [38,49] (for the recent state-of-the-art in cardiac myocyte–fibroblast interaction modelling, see [50,51]), and this has been corroborated in experimental studies using myocyte–fibroblast co-culture systems (e.g. [45–48]).

1.3. Scenarios of fibroblast electrophysiological integration

Electrical coupling of cardiac cells is generally thought to be mediated by gap junctions [52,53], although the possibility of direct cytosolic links through ‘tunnelling nanotubes’ has recently come to light [54]. Electrotonic effects of non-excitable cells on cultured cardiomyocytes can indeed be modulated by altering connexin expression [55].

Loss or gain of Connexin43 (Cx43) function in fibroblasts can be achieved by genetic approaches [56]. While Cx43 knockdown or knockout provides useful experimental tools for studies of electrical effects of fibroblasts on myocytes in vitro, this approach may entail complexities that confound matters in vivo. For example, reduced Cx43 expression appears to up-regulate fibrosis [57] and may be associated with mesenchymal tumour formation in the heart [58]. In contrast, fibroblasts that are genetically engineered to overexpress Cx43 appear to have more benign effects on cultured cardiomyocyte electrophysiology, potentially offering an anti-arrhythmogenic ‘electrotonic buffer’ [44]. These and other observations have stimulated research into the potential of turning fibroblasts from a bystander in, or even culprit of, cardiovascular disease, into a therapeutic target. Indeed, cell-therapeutic approaches using fibroblasts transfected to overexpress potassium channels have been used to reduce cardiac automaticity and prolong ventricular refractoriness in vivo [59].

Based on the topology of fibroblast–myocyte interactions, three hypothetical types of functionally-relevant coupling effects of fibroblasts on the electrophysiology of the intact heart can be proposed [60]. First, ‘zero-sided coupling’, i.e. fibroblasts not interacting with myocytes, will separate myocardial cells and/or layers, acting as an electrical insulator, as indeed will be the effect of the a-cellular part of cardiac connective tissue. This option is most in keeping with traditionally-held views, such as pointed out as early as 1906 by Keith and Flack in their original description of the human atrio-ventricular node [61]. Second, ‘single-sided coupling’ refers to fibroblasts connected to groups of myocytes that are themselves well-connected electrically. Here, electrotonic load, or buffer effects would dominate. Third, ‘double-sided coupling’ refers to fibroblast-connections that interlink myocytes that are not, or less directly, coupled electrically. These links are not necessarily afforded just by single cells, since cardiac fibroblasts are interconnected by homotypic gap junctions [33,62] that can support fibroblast–fibroblast electrotonic coupling in situ, as confirmed by transmission of hyperpolarizing impulses between fibroblasts, 40 μm apart, studied using double-barrelled microelectrodes in the atria of frog and rat [63,64], and via dye coupling studies in rabbit [65]. Double-sided coupling could allow cardiac fibroblasts to form conducting pathways of electrophysiological relevance at the tissue and organ level. If confirmed, this case would be of particular interest, as it would offer novel, therapeutically relevant, insight into post-ablation and post-infarction electrophysiological behaviour.

This being said, the presence and extent of fibroblast–myocyte electrotonic coupling in native myocardium (i.e. including observations in vivo and ex vivo, as long they pertain to cells in a morphologically intact tissue environment) remain controversial, forming the topic of this review. We will first re-cap the reasons for which one would doubt the presence of heterocellular coupling in vivo (‘Contra’), highlight why at least some of these doubts are not without limitations either (‘Contra-Contra’), review what is known in support of the presence of heterocellular coupling in the heart (‘Pro’), and finish by summarizing what should be done to resolve the riddle (‘Ergo’).

2. Contra

There are (at least) three common objections to the view that non-myocytes may be electrically coupled to muscle cells in native cardiac tissue.

2.1. Heterocellular gap junctions are not normally seen in histological studies

Generations of histologists have studied cardiac tissue. Intercalated discs at the juncture of cardiac myocytes were identified, using EM, in the early 1950s (e.g. [66]). In the mid-1960s, these disks were proposed to contain structures that link the cytosol of neighbouring cardiomyocytes [67] by bridging the gap at the junction of cells [68]. This suggestion was confirmed in the 1970s by purification of gap-junction plaques [69], identification in the 1980s of the first single gap-junctional protein capable of linking neighbouring cells via formation of inter-cellular ion channels [70], and followed up by detailed characterisation of multiple connexins (Cx), including the major isoform found in the heart, Cx43 [71]. This triggered significant interest in the role of cardiac gap junctions, including links to other cellular structures (such as the cytoskeleton [72]). Towards the end of the 1980s, Cx research received a significant boost from the introduction of immunological techniques to identify presence and location of Cx in various tissues of the body, including heart [73–77]. During the 1990s, Cx distribution studies opened up a treasure chest of insight into cardiac cell–cell interactions (for a selection of reviews from that time, see [78–90]) — yet none of the studies suggested a presence of heterocellular junctions between myocytes and non-myocytes in native cardiac tissue.

2.2. Heterocellular junctions are exceedingly rare, even when specifically chased

Following up on early reports on heterocellular coupling in vitro [22–24,52], and after detailed electrophysiological characterisation of heterocellular gap-junctional channels connecting freshly isolated or cultured cell pairs [30,40], the team of Jongsma set out to specifically search for gap junctions between myocytes and fibroblasts in native heart tissue. In a painstaking, serial sectioning based EM study of adult rabbit sino-atrial node, covering a total area of 7.7 mm², they observed just one gap-junction like contact between a fibroblast and a myocyte [33]. In contrast, small gap junctions between fibroblast pairs were common, while the analysed tissue volume was estimated to have contained ~ 10⁴ gap junctions between cardiac myocytes [33]. Thus, even targeted EM studies did not support the notion of a regular presence of heterocellular gap junctions in the heart.

2.3. Ockham's Razor: the heart works without invoking heterocellular coupling

William of Ockham, early 14^th century English scholar, stated that, if there are multiple possible explanations of a phenomenon, the simplest of them should be preferred. Cardiac electrophysiology can generally be explained, even in quantitative computational models [91–102], based solely on considering electrotonic coupling of myocytes. This shifts the burden of proof: if there is no need to invoke heterocellular coupling between myocytes and non-myocytes, one should not do so (never change a winning team).

3. Contra–Contra

3.1. Connexin-based coupling is possible in the absence of detectable gap junctions

Gap junctions, in particular between cardiac myocytes, lend themselves to relatively easy identification. Fibroblast gap junctions are significantly smaller [33], and thus are more easily overlooked. One should further note that early immuno-histochemical characterisation of connexin distribution in cardiac tissue did not tend to include any form of cell labelling. In this setting, intercalated discs between myocytes are a prominent feature, while the small ‘speckles’ of fluorescence, indicative of fibroblast-based gap junctions, could easily be regarded as background noise [103].

What's more, the absence of immuno-histochemically detectable gap junctions in confocal microscopy studies is not necessarily synonymous with a lack of connexin-based cell coupling, as shown for example in native arterial smooth muscle [104,105]. Unless super-resolution optical techniques are used, fluorescently labelled Cx will be detected only if spatially-clustered, involving at least 80–120 channels [106] (of note, smaller arrays, containing 15–20 Cx-channels, are regularly present in rat ventricular myocardium, as confirmed by EM [106]). Computer modelling studies suggest that as few as 10–30 connexin channels would provide sufficient coupling for electrophysiologically relevant effects of cardiac fibroblasts on myocytes [38]. Possible sites for such coupling could be finger-like extensions of cardiac fibroblasts into the cardiomyocyte basement membrane [32] — points of contact that may be more frequent in living (Fig. 1A), rather than the fixed and dehydrated tissue used in most histological analyses.

Fig. 1.

Fibroblast–myocyte interrelations in native cardiac tissue. A: Confocal laser scanning microscopy images of rabbit sino-atrial node live tissue, following diffusion-loading of the CellTracker dye CMFDA. Groups of sino-atrial node myocytes (centrally located ‘large’ uniform structures) are intermingled with cardiac fibroblasts (smaller cells with numerous fine processes towards the myocytes, see white arrows). Outline view left: 158 × 158 μm; detail view right: 30 × 30 μm (corresponds to the area near the arrow at the right edge of the outline). B: Extended structures, resembling membrane nanotubes (see white arrows), in adult mouse ventricular tissue. Confocal micrographs of cardiac tissue, labelled for WGA (membranes), vimentin (fibroblasts) and α-actinin (myocytes). Scale bar: 10 μm. From [32] in (A), and [54] in (B); with permission.

3.2. Connexin-based coupling may not be necessary for heterocellular cell junctions

So-called ‘tunneling nanotubes’ [107], membranous channels that are 50–200 nm wide and that can link cells over distances up to 300 μm in vivo [108], provide an alternative substrate for electrotonic coupling between different cell types. These structures may either form connexin-free direct cytosolic links [109], or involve Cx-coupling at the point of contact of two semi-tubes. The latter could occur ‘anywhere’ along the nanotubular connection, i.e. somewhere between apparently separate cells. Punctate immuno-labelling (in the absence of a cytosolic dye) would, in this setting, probably be regarded as background unrelated to intercellular connections [107]. Evidence for the presence of nanotube coupling between cardiac fibroblasts and myocytes has been shown in vitro and in vivo (Fig. 1B; [54]), although their relevance for cardiac electrophysiological integration remains to be established.

3.3. Ockham's Razor: special cases

The notion of being able to explain cardiac electrophysiology, based solely on homocellular myocyte coupling, is not without limitations. For example, atrial ablation lines (scars, created to interrupt uncontrolled excitation waves) often become ‘electrically transparent’ with time, due to re-emergence of functional conduction pathways [110]. While this may be due to incomplete lesioning, 10–20% of heart transplant recipients also show electrical coupling across the (unquestionably continuous) separation between donor and recipient tissue [111]. Aberrant electrical coupling can arise also across suture lines after operations to fix cardiac birth defects [112]. In all these cases, electrical conduction crosses scar tissue. Possible explanations include direct ‘touch-and-go’ interaction of surviving myocytes either side of a cut, de novo generation of cardiomyocytes that link the two tissue edges, or electrical conduction via non-myocardial cells such as fibroblasts [62]. Based on the histological appearance of post-transplantation scar tissue [113], the last option may offer the most straightforward explanation and, hence, be in keeping with Ockham.

4. Pro

4.1. Electrotonic fibroblast–myocyte coupling in native tissue

Fibroblasts in native heart tissue at rest have a membrane potential of - 10 to - 20 mV when electrically isolated from other cells [38]. They are electrically non-excitable (though subsequent to long-term cell culture, expression of the voltage-gated sodium channel Nav1.5 has been reported in human atrial fibroblasts [114]). Fibroblasts are excellent ‘followers’ of an electrotonically provided waveform. Thus, if electrically coupled to a cardiomyocyte, the fibroblast's membrane potential will show a myocyte-like action potential shape, albeit with a slowed upstroke and reduced amplitude (as illustrated in double whole-cell patch clamp experiments [40]). The ability of fibroblasts to passively convey action potentials, together with their high membrane resistance that supports ‘low-loss’ long-range signal conduction, makes fibroblasts a tantalising potential conduit for electrotonic signal transmission.

Unfortunately, this very behaviour also makes it nearly impossible to confidently explore cardiac fibroblast electrical integration in situ, using classic electrophysiological techniques [38]. While fibroblasts that are electrically isolated from cardiac muscle cells can be identified by their weakly polarised membrane potential and mechanically-induced polarizations in the rhythm of tissue contraction, those fibroblasts that are well-coupled (and, hence, at the centre of interest here) are indistinguishable from poorly impaled cardiomyocytes (Fig. 2; [38]). Nonetheless, indirect support for heterocellular connections in weakly coupled fibroblasts has been found (Fig. 2, centre; note that upon electro-mechanical uncoupling of the preparation, the mechanically induced component ‘a’ recedes, while ‘b’ and ‘c’ remain).

Fig. 2.

Membrane potential pairs (top and bottom), recorded simultaneously in spontaneously beating adult rat atrium using double-barrelled floating microelectrodes (tip-to-tip distance 40 μm) from putative fibroblasts (F, top) and cardiomyocytes (M, bottom). Left: F, electrically not connected to surrounding cardiomyocytes, showing a relatively depolarized membrane potential (here about - 20 mV) and the contraction-induced membrane potential change typical for these cells (label ‘a’). Middle: in some cases, presumably when weakly coupled to adjacent myocytes, F display an additional, probably electrotonically transmitted membrane depolarization (labelled ‘b’) while the myocyte membrane potential is positive to the diastolic potential in F, and a ‘capacitative’ spike (labelled ‘c’) believed to be caused by the synchronous depolarization of myocytes in the proximity of F. Right: recording of a potential waveform that would be commensurate with an F that is well-coupled to adjacent cardiomyocytes, where the electrotonically transmitted component mimics an action potential waveform of reduced amplitude and upstroke velocity (here on top of the atrial contraction-induced mechano-sensitive peak; note different potential scales for first two F potentials). Adapted from [38]; with permission.

4.2. Connexins at fibroblast–myocyte contact sites in cardiac tissue

Immuno-histochemical labelling studies for a range of cardiac connexins, combined with identification of cells on either side of connexin signals, suggest that Cx-localization at the point of contact between cardiac fibroblasts and myocytes is far from uncommon, both in healthy (Fig. 3 [62]) and post-infarct tissue (Fig. 4 [115]).

Fig. 3.

Connexin location relative to cardiac myocytes and connective tissue cells in native myocardium of healthy rabbit. Triple immunolabeling for myocytes (M; red: antimyomesin), fibroblasts (F; blue: antivimentin), and Cx43 or Cx40 (top and bottom rows, respectively; green: anti-Cx43/anti-Cx40) in ventricle, atrium, and atrioventricular node (AVN). Vertical arrows: homotypic myocyte contacts; horizontal arrows: homotypic non-myocyte connections; slated arrows: Cx at heterotypic cell contacts. Side-panels show 2.55 × zoomed views of the areas highlighted by dashed squares in the main images. Note significantly smaller size of Cx labels at homo- and heterotypic contact sites that involve non-myocytes. Scale bars: 20 μm. From [62], with permission.

Fig. 4.

Propagation of electrical activation into ventricular scar tissue. A: Epicardially recorded optical maps of trans-membrane potential changes (dye used: RH237) in Langendorff-perfused rabbit isolated heart, 8 weeks after induction of transmural left-ventricular infraction (see B, left panel). Electrical pacing (at site identified in the schematic on the left by a red dot, and in the photograph (i) of the infract area by a white asterisk) in the normal zone (NZ) gives rise to a conduced wave of electrical excitation, which is delayed by about 10 ms at the peri-infarct before invading the infarct zone (PZ, IZ; respectively); see crowding and thinning of activation isochrones (ii). Normalised trans-membrane potential shapes, recorded at the locations labelled d–h, are shown in (iii). Signals from inside the scar tissue (i.e. at sites g and h) persisted even after chemical ablation of surviving endocardial muscle layers (panel B). LA: left atrium, RA: right atrium; LV: left ventricle; red vertical line in (iii): time of pacing stimulus. B: Histological substrate of scar tissue in rabbit (left, 8 weeks post-infarct, used in the experimental studies shown in A) and sheep (border zone at 1 week post infract [middle]; infarct zone 2 weeks post-infarct [right]). The panel on the left (trichrome staining) illustrates the transmural nature of infarcts used in the optical mapping studies, including the presence of islands of surviving myocytes in the infarct zone, and a surviving endocardial tissue rim (lower edge of section) that was ablated in part of the study. The images in the centre and at the right (triple immuno-labelling for Cx43 [green], myomesin to identify myocytes by their striation, and vimentin to label fibroblasts [both red]) highlights Cx43 distribution, including presence at heterocellular contact sites at the infract border (middle) and in central zone myocyte islands. Vertical arrows: homotypic myocyte contacts; horizontal arrows: homotypic non-myocyte connections; slated arrows: Cx43 at heterotypic cell contacts. From [116] in (A) and from [117,115,1], respectively, in (B); with permission.

Of course, presence of a protein ‘in the right place’ does not confirm its functionality. Direct evidence for heterocellular coupling in native tissue has thus far been obtained only in rabbit sino-atrial node, where Lucifer yellow dye transfer between myocytes and fibroblasts was observed [65].

Clearly, further investigation is needed to explore developmental dynamics, anatomical distribution, physiological regulation, and response to disease progression or treatment attempts, but proof-of-concept for fibroblast–myocyte coupling in vivo has been established. A conceptual approach to myocardial function that excluded the possibility of heterocellular coupling would, therefore, be unnecessarily restricted.

4.3. Optical mapping of conduction inside scar tissue

Given the principal difficulty of distinguishing between trans-membrane potential waveforms of (active) cardiac myocytes and (passive) fibroblast-followers in normal heart tissue, cardiac scars offer an interesting alternative study subject, because the ratio of electrically active and passive cells is shifted significantly towards predominance of the latter. A fine example of such work is the 2007 study of Walker et al., investigating fully-transmural infarcts in adult rabbit left ventricle by optical mapping of voltage-sensitive dye signals (Fig. 4A). They observed propagation of cardiac excitation waves into scar tissue, even after chemical ablation of any surviving sub-endocardial muscle layers (Fig. 4B), performed to exclude possible contributions from these deeper myocyte-sheets to epicardially mapped potentials; [116]). The signals from within the scar resembled ventricular action potentials, albeit with slowed upstroke and reduced amplitude — very much in keeping with previous observations in neonatal rat fibroblast–myocyte cell pairs [40] and adult rat atrial tissue [38].

Subsequent work by other investigators confirmed that excitation waves can indeed invade and pass through cardiac scar tissue [118,119]. These studies further showed that the myocyte-like action potentials inside four-week old ventricular scar tissue in rabbit ventricle are not accompanied by significant cyclic changes in intracellular free calcium concentration [118]. As calcium transients are a signature activity of cardiomyocytes, the most straightforward explanation again is that non-myocytes passively display much of the electrical signals conducted inside scars [117], perhaps electrically integrating surviving myocyte islands that could act as active ‘repeater stations’ that maintain or boost electrical signal amplitude.

5. Ergo

The short response to the initial question “Fibroblast–myocyte electrotonic coupling: does it occur in native cardiac tissue?” is: ‘yes’. However, as will be clear from the above, this raises many more questions than it answers. Some of the aspects, relevant for further study, are summarised below.

5.1. Need targeted assessment of dynamic heterocellular interactions

Confirmation, in native myocardium, of functional fibroblast–myocyte coupling at the cellular level has been obtained thus far in healthy atrial tissue of a single species only [65]. Whether heterocellular interactions in vivo change during development, disease, and/or in response to therapeutic interventions, and whether they differ between species and/or cardiac tissue regions, is unknown. Given the presence of fractional differences in fibroblast origin [8–13], as well as disease-induced fibroblast phenotype transitions [120], and mechanical modulation of Cx coupling [121], it is reasonable to expect differences in heterocellular interactions of fibroblasts in diseased tissue, compared to control [122].

Equally unconfirmed is the functional relevance of heterocellular coupling in in vivo. For example, fibroblasts show cardiac contraction-related fluctuations in membrane potential [38], probably mediated by stretch-activated ion channels [123–125] and changes in calcium handling [126]. This may be functionally relevant beyond electrophysiology, integrating cardiac electrical and mechanical activity [127]. Another area of possible functional importance is related to the observation that, during myocardial infarction, fibroblasts increasingly express functional K_ATP channels in scar and border zone tissue [128]. This could be of relevance for arrhythmia prevention and treatment, and lends support to the suggestion that (pre-/post-)conditioning needs to consider effects on the whole heart, not just the cardiomyocytes [129].

In addition, it is important to realise that not all scars are created equal. Oxygen starvation during myocardial infarction, for example, will affect preferentially the metabolically more active cardiomyocytes, so that locally-surviving cells (with a bias towards non-myocytes) will contribute to scar tissue composition. In contrast, ablation by excess local energy delivery will destroy cells irrespective of their nature (but potentially dependent on their proximity to circulating normothermic ‘coolant’ in coronary vessels), so that the vast majority of cells forming the post-ablation scar will invade from intra- or extra-cardiac sources outside the ablation focus. Yet another setting is observed at surgical suture lines, where the scar is built de novo, in the absence even of any pre-existing extracellular matrix. Further variances in electrophysiological properties of atrial vs. ventricular fibroblasts [130] may contribute to dissimilar functional behaviours displayed by different cardiac scars, for example after atrial ablation or ventricular infarction. And last, but not least, scar tissue is highly dynamic, containing a range of non-myocyte cell- and phenotypes whose relative quantities, and Cx expression patterns [115], change over time with scar maturation.

Scars, of course, are but the tip of the iceberg and it will be important to develop an understanding of the roles that fibroblasts may have as electro-mechanical signalling hubs in normal, and diffusely or patchy fibrotic tissue. This will require a better understanding of whole-heart histo-anatomical features, benefitting from imaging-based 3D reconstruction and biophysically-detailed simulation [131–134], an area that has made significant progress in recent years in terms of inclusion of connective tissue effects on arrhythmogenesis [100,135–137] and defibrillation [138,139]. It is vital, though, not to fall victim of the ‘plausibility trap’ [140]: simply because modelling and reality ‘agree’ with one another does not mean that mechanisms invoked in a theoretical model are crucial, or indeed even involved at all. Key to progress is validation, and this requires development and application of novel experimental tools.

5.2. Need better tools

It is evident that, on the one hand, fibroblast–myocyte interactions in cardiac cell cultures overestimate Cx-coupling while, on the other hand, native ‘bulk’ tissue studies are associated with significant challenges in terms of electrophysiological identification and characterisation of fibroblast activity and coupling. This calls for intermediate level biological model systems, such as thin sections of live cardiac tissue. Cardiac tissue slices were originally established and used as an experimental model for biochemical and pharmacological research (e.g. [141–143], reviewed in [144]). By now, they have been employed in electrophysiology research, using extra- and intracellular potential recordings [145] and optical mapping techniques, including dual-dye (calcium and voltage) studies [146]. They have been tested for a range of cardiac tissue sources, from neonatal and adult rat to human [147]. Heart tissue slices, usually cut at a thickness of 300–400 μm to prevent oxygen starvation, allow prolonged maintenance of biological activity [145,148]. They lend themselves to improved correlation of observed electrophysiological signals to underlying histological substrates, both because they are thin enough for visual inspection, and since local signals are less affected by more distant tissue (e.g. far-field potentials for contact recordings, or depth-effects of photon scattering in optical mapping of bulk tissue [149]).

In addition to the use of pseudo two-dimensional yet organotypic models for observation, optical tools for targeted interference offer another exciting route to further progress. This has been exquisitely illustrated by Entcheva and colleagues, who transfected non-excitable cells with a genetically encoded light-sensitive ion channel (channelrhodopsin-2), which they used to pace electrotonically connected cardiomyocytes in vitro [150]. It would be interesting to combine studies like this with the in vivo engraftment of Cx43-expressing non-excitable cells in scar tissue, which has been found to strongly reduce infarct-related arrhythmias in mice [151]. In any case, cell-type specific targeting of optogenetic probes will open up significant potential for novel insight into the multicellular nature of cardiac electrical function.

5.3. Need a broader vision of cell coupling in the integrated heart

The need for a broader conceptual approach to heterocellular interactions is evident also from the observation that gap junctions are sites not only of electrical, but also mechanical integration [152]. Mechanical interaction of fibroblasts and myocytes could not be illustrated more vividly than in the five-day time-lapse movie accompanying Driesens' beautiful in vitro report on partial cell fusion of rat cardiomyocytes and fibroblasts [153]. The targeted contacting and mechanical pull of fibroblasts on myocytes may give rise to mechano-electric feedback effects of the former on the latter [154], and/or serve as a guide for fibroblast cell migration [155]. In addition, it illustrates the highly dynamic nature of the heterocellular contact points (Fig. 1) between cardiac fibroblasts and myocytes (and serves as a note of caution regarding apparent cell-type transitions, in particular in vitro, given the presence of cell fusion events).

In addition, the ‘perinexus’ [156], a specialised region surrounding the connexin-dominated part of cell junctions, may support non-electrotonic ephaptic coupling. Presence of this type of cell interaction has been proposed in the 1980s [157], and it has recently seen a revival in terms of interest and insight [158,159].

Finally, when considering the heart as a hetero-cellular organ, we should not stop at myocytes and fibroblasts. Electrophysiological effects of nerve, immune and fat cells, vascular smooth muscle, or endothelium (to name but a few), clearly desire a better understanding as well [2,3].

6. Heterocellular coupling between cardiac cells in vivo!

There is a growing array of new experimental tools at our disposal to explore heterocellular coupling in native cardiac tissue. In coming years it is likely that we will see these tools provide unexpected advances in our understanding of the electrical interplay between the multiple cell types that make up the heart. In particular, it is anticipated that there will be definitive answers on where and when myocytes and non-myocytes are capable of electrophysiologically-relevant coupling in vivo, and what the context and the relevance are of such interactions. Answers here will provide new insight into normal cardiac function and into mechanisms of disease development. Perhaps more importantly, an understanding of the nature and dynamics of heterocellular electrical communication in the heart could give rise to novel or improved therapeutic approaches. This may range from amendments to established procedures, such as atrial ablation, to the vision of modifying cardiac scar properties in a targeted manner to reduce arrhythmogenesis following infarction.

The key question to which we should move on, therefore, is: Fibroblast–myocyte electrical coupling: does it matter in native cardiac tissue?

Disclosures

RGG has modest equity in FirstString Research Inc., a company testing Cx43-based therapeutics for wound healing (co-founder, scientific advisory board member).