Molecular and cellular neurocardiology: development, and cellular and molecular adaptations to heart disease

Abstract

The nervous system and cardiovascular system develop in concert and are functionally interconnected in both health and disease. This white paper focuses on the cellular and molecular mechanisms that underlie neural–cardiac interactions during development, during normal physiological function in the mature system, and during pathological remodelling in cardiovascular disease. The content on each subject was contributed by experts, and we hope that this will provide a useful resource for newcomers to neurocardiology as well as aficionados.

Abbreviations

APD

action potential duration

β‐AR

β‐adrenergic receptor

AT II

angiotensin II

CaM

calmodulin

CaMKII

calmodulin‐dependent protein kinase II

CGRP

calcitonin gene related peptide

ChR2

channelrhodopsin‐2

ICG

intrinsic cardiac ganglia

NA

noradrenaline

nNOS

nitric oxide synthase

NGF

nerve growth factor

NPY

neuropeptide Y

PKA

protein kinase A

RyR

ryanodine receptor

SHR

spontaneously hypertensive rat

SR

sarcoplasmic reticulum

TH

tyrosine hydroxylase

VIP

vasoactive intestinal peptide

Neural–cardiac interactions: co‐maturation in development

Autonomic control of the heart occurs through transmission by sympathetic and parasympathetic neurons, which have a common neural crest origin. The anlagen of cardiac parasympathetic ganglia first appear around the middle of embryonic development in rodents, and vagal input is already present as the immature neurons migrate to positions at the dorsal atrium (Hildreth et al. 2009). Parasympathetic neurons and surrounding satellite glial cells mature postnatally in rodents, and the neurons double in size as they develop connections with the myocardium (Fregoso & Hoover, 2012). The rodent sympathetic–cardiac circuit is established in the late prenatal and early postnatal period as growing sympathetic fibres innervate the developing heart (Hildreth et al. 2009). This can be considered as a co‐maturation system in which signals from cardiac tissue regulate the growth, patterning and transmission properties of the innervating sympathetic neurons, while reciprocal signalling from the nerve fibres influence the maturation of cardiac myocytes and adult cardiac properties.

The role of cardiac target‐derived factors as regulators of neuronal maturation has been well established. Cardiac‐derived neurturin is essential for the survival of cardiac parasympathetic neurons (Hiltunen et al. 2000; Mabe & Hoover, 2009), and cardiac signals control multiple events in sympathetic neuron maturation. Trophic signals that include nerve growth factor, neurotrophin‐3, and cytokines regulate outgrowth of sympathetic axons within the heart, continued survival upon reaching the target (Chun & Patterson, 1977; Landis & Keefe, 1983; Lockhart et al. 2000; Glebova & Ginty, 2005), regulation of neurotransmitter phenotypes (Yamamori et al. 1989; Slonimsky et al. 2003), and synaptic transmission (Lockhart et al. 1997; Luther & Birren, 2006, 2009; Hasan et al. 2012). Additional signals from cardiac tissue regulate developmental transitions between different growth states in sympathetic neurons. For example, contact with cardiac tissue results in a change in the neuronal response to bone morphogenetic proteins (BMPs), providing a stop‐growth signal for sympathetic axons that have reached their target (Moon & Birren, 2008). Interacting actions of neurotrophic factors (Habecker et al. 2008; Lorentz et al. 2010) and Sema3a signalling (Ieda et al. 2007) are also likely to act in the patterning of sympathetic fibres within the heart. Overall, a rich array of heart‐derived signals drives the maturation and function of the postganglionic sympathetic system.

While less is known of the effects of innervating sympathetic fibres on cardiac development, the development of a number of cardiac properties are regulated by sympathetic signals. Co‐culture with sympathetic neurons regulates the expression of angiotensin II and atrial natriuretic peptide (ANP) in developing cardiomyocytes (Hunt et al. 1995; Hansson et al. 2001). Additionally, sympathetic innervation has been linked to functional alterations in a number of important cardiac ionic currents, including Na⁺, L‐type Ca²⁺, pacemaker, inward rectifier and transient outward K⁺ currents (Qu & Robinson, 2004). Recent studies also suggest that sympathetic interactions play an important developmental role in regulating cardiomyocyte maturation during the postnatal period.

Questions and controversies

Can we exploit the fact that myocyte proliferation in pathological situations is influenced by the autonomic nervous system with new therapeutics or interventional treatments in man?

Is there a component of inherited structural and electrophysiological heart conditions that arises from dysfunctional autonomic control during cardiac development?

Does abnormal autonomic cardiovascular development contribute to human essential hypertension?

One way that early sympathetic innervation regulates cardiomyocyte development, and the properties of the adult heart, is through the regulation of cell cycle withdrawal and determination of cardiomyocyte cell number (Kreipke & Birren, 2015). While some organs are able to replenish cell number throughout life, the regenerative capacity of cardiomyocytes is, at best, limited (Zak, 1973). Cardiomyocytes withdraw from the cell cycle shortly after birth and, for the most part, remain quiescent throughout an organism's life. Following cell cycle withdrawal, heart size increases predominantly via cellular hypertrophy, or the enlargement of individual cardiomyocytes (Li et al. 1996; Ahuja et al. 2007). Thus, the number of postnatal cardiomyocyte divisions, and the timing of the developmental transition between proliferative and hypertrophic growth in the young animal are significant factors for determining the total number of cardiomyocytes in the adult heart.

The rapid transition of cardiomyocytes from a proliferative to a hypertrophic growth state (Li et al. 1996) corresponds to the postnatal period of sympathetic ingrowth into the heart (Hildreth et al. 2009). Recent evidence has shown that neonatal sympathectomy results in smaller hearts in developing and adult rats, demonstrating that early sympathetic innervation has long‐term effects on heart development (Kreipke & Birren, 2015). In vitro experiments using neonatal cardiomyocytes grown in the presence or absence of innervating sympathetic fibres demonstrated that sympathetic β‐adrenergic signalling prolonged the period of postnatal cardiomyocyte proliferation, delaying the period of hypertrophic growth. While, ultimately, the size of individual cardiomyocytes was not altered in adult animals that were sympathectomized at birth, the total number of these cells was reduced in the sympathectomized animals (Kreipke & Birren, 2015). This work shows that early sympathetic signalling leads to the development of larger hearts with more cardiomyocytes in vivo and suggests that sympathetic signalling is part of a homeostatic system that promotes cardiomyocyte proliferation and acts to set the final number of cardiomyocytes in the adult heart.

Sympathetic regulation of postnatal cardiomyocyte proliferation may also have implications for heart regeneration following cardiac damage. The neonatal mouse heart undergoes regeneration following tissue damage, a transient capability that is greatly diminished in older animals (Porrello et al. 2011; Bryant et al. 2015). Recent studies have shown that disruption of peripheral nerves blocks this regeneration. In one study, pharmacological inhibition of cholinergic signalling prevented cardiomyocyte proliferation following neonatal resection of the mouse ventricle (Mahmoud et al. 2015). This suggests a role for parasympathetic innervation in heart regeneration, a model supported by the finding that vagal nerve ablation also decreased cardiomyocyte proliferation following cardiac injury via resection or myocardial infarction (Mahmoud et al. 2015). Interestingly, another study showed that chemical sympathectomy in young animals also blocked early regeneration of damaged ventricular tissue, while increasing scarring in the tissue (White et al. 2015). While additional work defining the relative roles of cholinergic and noradrenergic signalling in the regeneration process is needed, together this work defines a new role for cardiac innervation in regenerative processes. Genetic programmes associated with proliferative cardiomyocytes are reactivated during pathological hypertrophy (Komuro et al. 1988; Ahuja et al. 2007), suggesting that these reciprocal interactions have implications for later processes in the cardiac system.

Questions and controversies

What are the best molecular targets to ensure optimal neural remodelling to promote myocyte survival and minimize the risk of arrhythmia post myocardial infarction and during chronic heart failure?

Normal neurotransmission and injury‐induced plasticity in cardiac nerves

The mature heart is densely innervated, with parasympathetic, sympathetic and sensory neurons innervating regions of the myocardium and cardiac conduction system. Each of these populations of neurons has a distinct distribution and phenotype that is critical to its function. Cardiac nerves undergo significant changes in morphology and phenotype in disease. Changes in growth factor expression, oxidative stress, and inflammatory cytokines within the heart and vasculature contribute to neuronal remodelling. These issues have been reviewed in detail recently (Kimura et al. 2012; Fukuda et al. 2015; Gardner et al. 2016), and will be summarized here briefly.

Sensory nerves

Sensory innervation of the heart includes both mechanosensitive and chemosensitive nerve endings (reviewed by Hainsworth, 1991; Schultz, 2001). Mechanosensitive afferents transduce pressure changes in the atria and ventricles during the cardiac cycle, while chemosensitive nerve endings respond to endogenous substances including adenosine and H⁺ and chemicals like capsaicin (Longhurst et al. 2001). Most of the sensory neurons innervating the heart reside in the nodose–petrosal ganglia and dorsal root ganglia, although some neurons within the intrinsic cardiac ganglion are sensory afferents (Armour, 2008). Many cardiac sensory fibres that arise in dorsal root ganglia produce nitric oxide synthase (nNOS), substance P and/or calcitonin gene related peptide (CGRP), distinguishing them from nodose–petrosal neurons which lack all three (Hoover et al. 2008). CGRP and substance P levels are decreased in coronary artery disease and diabetes mellitus, and end stage heart failure (Taquet et al. 1992; Wang et al. 2012). The loss of neuropeptide content in diabetes corresponds with CGRP‐immunopositive cardiac sensory denervation and atrophic changes in the dorsal root ganglia (Ieda et al. 2006).

Parasympathetic nerves

Cardiac parasympathetic neurons reside in intrinsic cardiac ganglia (ICG), which have a diffuse distribution on the surface of the heart, are connected by an extensive plexus of nerve fibre bundles (O'Shea & Evans, 1985; Pauza et al. 2000; Ardell, 2001; Hoover et al. 2004, 2009; Mabe et al. 2006), and project to distinct regions of myocardium (Quan et al. 1999, 2002; Zarzoso et al. 2013). Cholinergic nerve fibres are most abundant in sinoatrial and atrioventricular nodes, atrial myocardium and the ventricular conducting system (Hoover et al. 2004). Stimulation of cardiac parasympathetic nerves projecting to these regions leads to acetylcholine (ACh) release, which can evoke prominent bradycardia, negative dromotropic responses and decreases in atrial contractility via activation of postjunctional M2 muscarinic receptors. Cholinergic innervation in less abundant in the ventricles, but studies in humans and other large mammals have shown that these nerve fibres can trigger significant negative inotropic responses mediated by muscarinic receptors (DeGeest et al. 1965; Xenopoulos & Applegate, 1994; Ardell, 2001; Lewis et al. 2001; Coote, 2013). Only two studies have evaluated vagal parasympathetic effects on ventricular contractility in small mammals (Takahashi et al. 2003; Nalivaiko et al. 2010). While the first study failed to detect a negative inotropic response to vagal stimulation, the second study provided definitive evidence for a small decrease in contractility in a majority of experimental preparations. Based on the slow onset of these responses, it was suggested that cholinergic inhibition might occur indirectly though attenuation of noradrenaline (norepinephrine) release and/or postjunctional signalling (Nalivaiko et al. 2010). Cardiac parasympathetic ganglia integrate information from a variety of sources, including cholinergic preganglionic neurons (McAllen & Spyer, 1976; Hoover et al. 2004, 2009; Mabe et al. 2006), presumptive noradrenergic nerve fibres (Leger et al. 1999; Hoard et al. 2008; Hoover et al. 2009) and peptidergic fibres (Calupca et al. 2000, 2001).

Studies of human ICG have demonstrated that detrimental changes can occur in cardiac disease. Morphological studies of ganglia from ischaemic hearts revealed pathological changes in about 35% of the neurons evaluated (Hopkins et al. 2000). Abnormalities included various inclusions and vacuoles in somata, neuronal enlargement, loss of dendrites, and neuronal degeneration with the infiltration of phagocytes. Recently, another study found significant hypertrophy of intrinsic cardiac neurons from patients with autopsy‐confirmed heart failure (Singh et al. 2013). In contrast, mouse models of diabetes, which have significantly impaired vagal control of the heart, do not exhibit any changes within cardiac parasympathetic neurons of the ICG (Mabe & Hoover, 2011). Decreased parasympathetic transmission in the heart, and its role in cardiovascular pathology, is discussed further in the section on ‘Neuromodulation and autonomic imbalance’.

Sympathetic nerves

The sympathetic neurons innervating the heart reside predominantly in the stellate ganglia (Norris et al. 1974; Kostreva et al. 1977; Pardini et al. 1990) and produce the transmitter noradrenaline (NA) and the co‐transmitter neuropeptide Y (NPY). Sympathetic axons innervate the atria, cardiac conduction system and ventricles, where they stimulate increased heart rate (chronotropy), conduction velocity (dromotropy), and contractility (inotropy) via NA activation of β₁‐adrenergic receptors. Sympathetic nerves are most dense in the atria and near the base of the ventricles with fewer nerves near the apex of the heart. Stimulation of the stellate increases contractility throughout the heart (Szentivanyi et al. 1967), but a transmural gradient of axons accompanies an epicardial to endocardial gradient in cardiac action potential duration that is important for normal activation and repolarization of the left ventricle (Antzelevitch et al. 1991; Nabauer et al. 1996; Brunet et al. 2004). Activation of cardiac β‐adrenergic receptors (β‐ARs) modulates myocyte repolarization by altering transmembrane currents and Ca²⁺ homeostasis (Thomas et al. 2004; Bers, 2008; Cutler et al. 2011), while it increases contractility by modulating calcium handling proteins so that Ca²⁺ release from the sarcoplasmic reticulum (SR) is increased (Valdivia et al. 1995; Marx et al. 2000; Shannon et al. 2000). The effects of sympathetic stimulation allow myocytes to meet increased cardiac demands during stress or exercise, but changes to sympathetic transmission in pathological conditions can contribute to cardiac dysfunction.

Cardiac sympathetic neurons can change their function in cardiovascular disease. Sympathetic hyperinnervation (Zhou et al. 2004; Hasan et al. 2006; Meloni et al. 2010) and denervation (Barber et al. 1983; Stanton et al. 1989; Dae et al. 1995; Li et al. 2004 b; Gardner & Habecker, 2013) have both been observed after myocardial infarction. Hyperinnervation has also been observed in compensatory cardiac hypertrophy during the transition to overt heart failure (Kimura et al. 2007; Lu et al. 2012), but NA synthesis is paradoxically reduced due to downregulation of tyrosine hydroxylase (TH), the rate‐limiting enzyme in NA synthesis (Kimura et al. 2007, 2010). In heart failure highly polysialylated neural cell adhesion molecule (PSA‐NCAM), which is a marker of immature neurons, is highly expressed in cardiac sympathetic neurons (Kimura et al. 2007). Further, many neurons in the stellate ganglia and left ventricle have increased expression of cholinergic markers such as choline transporter and choline acetyltransferase during heart failure. This is accompanied by decreased TH expression and thought to represent cholinergic transdifferentiation of cardiac adrenergic neurons into cholinergic neurons, which is observed in animal models of heart failure and in autopsy specimens from patients with heart failure (Kanazawa et al. 2010). Emerging evidence suggests that a similar cholinergic transdifferentiation occurs transiently after acute myocardial infarction (Olivas et al. 2016). Leukemia inhibitory factor (LIF) and other members of the interleukin‐6 (IL‐6) family, which can induce fetal gene expression (so‐called rejuvenation) in adult cardiomyocytes, are upregulated during heart failure (Kimura et al. 2007) and after myocardial infarction (Aoyama et al. 2000; Gritman et al. 2006). IL‐6 family cytokines secreted from the damaged myocardium act as negative modulators of sympathetic function by suppressing noradrenergic function (Kimura et al. 2007; Parrish et al. 2009) and inducing cholinergic differentiation (Kanazawa et al. 2010; Olivas et al. 2016) via a gp130 signalling pathway.

Nerve growth factor

Many of the injury‐induced changes in cardiac sympathetic and sensory nerves are driven by alterations in cardiac nerve growth factor (NGF). NGF supports the development and maintenance of cardiac sympathetic axons as well as the CGRP‐immunoreactive cardiac sensory nerves that richly innervate the epicardium and the ventricular myocardium (Ieda et al. 2006). Increased cardiac NGF is observed after myocardial infarction (Zhou et al. 2004; Hasan et al. 2006; Meloni et al. 2010; Gardner & Habecker, 2013) and in compensated cardiac hypertrophy (Ieda et al. 2004; Kimura et al. 2007), where it is associated with regional sympathetic hyperinnervation (Cao et al. 2000 b; Li et al. 2004 b; Zhou et al. 2004; Hasan et al. 2006; Kimura et al. 2012). Conversely, decreased cardiac NGF in experimental and clinical heart failure (Kaye et al. 2000; Qin et al. 2002; Kimura et al. 2010) and diabetes mellitus (Hellweg & Hartung, 1990; Schmid et al. 1999; Ieda et al. 2006) is associated with the loss of sympathetic and sensory fibres and reduced cardiac NA and CGRP levels (Taquet et al. 1992; Qin et al. 2002; Kristen et al. 2006; Kimura et al. 2007, 2010; Ieda & Fukuda, 2009; Kuehl & Stevens, 2012). Restoring NGF to the heart prevents sensory dysfunction in diabetes mellitus (Ieda et al. 2006). The regulation of NGF in the heart is not completely understood, but the hypertrophic factor endothelin‐1 increases NGF expression in cardiomyocytes (Ieda et al. 2004; Kimura et al. 2007), while mechanical stretch and α₁‐adrenergic stimulation attenuate myocyte NGF expression (Rana et al. 2009).

Neuromodulation and autonomic imbalance

Emerging evidence suggests that factors produced within the microenvironment of the heart, its vasculature and between neuronal populations can influence sympatho‐vagal balance. These neuromodulators can be intrinsic to sympathetic ganglia or vagal neurons (such as neuronal nitric oxide and its adaptor protein, CAPON), or may act as either autoinhibitory or crosstalk communicating messengers between neuronal populations (for example the co‐transmitters neuropeptide Y (NPY), galanin and vasoactive intestinal peptide (VIP)). Substances released from cardiac myocytes and the coronary vasculature, including natriuretic peptides and angiotensin II (AT II), may also influence neuronal function acting in a paracrine manner (see Fig. 1). These signalling pathways converge on neuronal calcium handling, transmitter release and/or re‐uptake mechanisms. The plasma levels and protein expression of many of these neuromodulators are increased in both animal models and patients with cardiovascular disease. However, their functional significance and the potential to exploit these pathways pharmacologically in disease states is yet to be established.

Figure 1. Schematic representation of the neuromodulatory pathways influencing the release of noradrenaline (NA) and acetylcholine (ACh) from postganglionic sympathetic and parasympathetic neurons .

Neuromodulators can arise from the coronary microvasculature (e.g. angiotensin II (AT II) and C‐type natriuretic peptide (CNP)), myocytes (B‐type natriuretic peptide (BNP)) as well as between neurons (neuropeptide Y (NPY), galanin (Gal), acetylcholine (ACh) and vasoactive intestinal peptide (VIP)) and within neurons (such as neuronal nitric oxide synthase (nNOS) and its activator protein, CAPON). Stimulatory pathways are represented by green arrows, and inhibitory pathways by red arrows. β, beta adrenergic receptor; M, muscarinic receptor.

Studying the actions of cholinergic neuromodulators is particularly challenging. It should be noted that whilst stimulation of the cervical vagus activates efferent preganglionic parasympathetic neurons, up to 70% of the fibres are sensory afferents (Berthoud & Neuhuber, 2000). Proximal stimulation of the cut cervical vagus in vitro will therefore cause retrograde stimulation of sensory afferents. Stimulation of the intact vagus nerve in vivo will cause both anterograde and retrograde stimulation of both afferent and efferent fibres. Recent work has shown that this has the potential to elicit a complex profile of mediators and modulators from both efferent and afferent fibres engaging multiple levels of the cardiac neuraxis to alter peripheral neural processing and central–peripheral neural interactions, including central drive (Ardell et al. 2015). With the introduction of implantable cardiac vagal nerve stimulators in patients with heart failure, it is important to understand the differences in cellular mechanisms by which this intervention may influence the heart, compared to application of exogenous neurotransmitters, or stimulation of the cervical vagus removed from the central nervous system, as these approaches may yield different results.

Intrinsic neuromodulators

Neuronal nitric oxide and its activator protein, CAPON

Neuronal nitric oxide synthase (nNOS) is localized in both intrinsic cardiac vagal neurons and postganglionic sympathetic neurons of the stellate ganglia. Nitric oxide acts via stimulation of soluble guanylate cyclase, to generate cGMP, leading in parasympathetic neurons to inhibition of phosphodiesterase 3 (PDE3) and an increase in cAMP–protein kinase A (PKA) dependent phosphorylation of N‐type calcium channels and release of acetylcholine (Herring et al. 2000, 2001 b; Herring & Paterson, 2001), whilst in sympathetic neurons stimulation of PDE2 and/or protein kinase G (PKG) leads to inhibition of calcium influx and NA release (Schwarz et al. 1995; Wang et al. 2007). Increasing nNOS expression using viral vectors can reverse impaired vagal (Heaton et al. 2007) and exaggerated sympathetic drive (Li et al. 2007) in the spontaneously hypertensive rat (SHR), and improve acetylcholine release and mortality acutely following myocardial infarction in the guinea pig (Dawson et al. 2008).

Genome wide association studies implicate a variant in the neuronal nitric oxide synthase adaptor protein (CAPON) in electrocardiographic QT variation and sudden cardiac death, both of which are strongly influenced by cardiac autonomic balance (Arking et al. 2006; Kao et al. 2009). Recently it was shown that CAPON resides in cardiac sympathetic neurons (Lu et al. 2015) and its expression, along with nNOS (Li et al. 2013 a), is decreased in the pre‐hypertensive SHR. Moreover, increasing its expression using a novel noradrenergic specific adenoviral vector increases nNOS activity and normalizes sympathetic neurotransmission. Many questions relating to the role of nNOS/CAPON in cardiac autonomic control remain.

Questions and controversies

How is nNOS localized within postganglionic neurons and is its activity influenced by calcium influx triggering neurotransmitter release at individual varicosities?

Is the main role of CAPON intracellular targeting of nNOS and/or to increase nNOS enzyme activity?

Are CAPON single nucleotide polymorphisms in neurons the trigger for sudden cardiac death in patients with and without QT disturbances?

Convergence of neuromodulators on calcium induced transmitter release and re‐uptake

The main trigger for the exocytotic release of neurotransmitter is the binding of calcium to synaptotagmin to activate the SNARE complex and cause fusion of the vesicular and cellular membranes (Sudhof & Rothman, 2009). The local calcium signal is a function of the opening of neuronal calcium channels and may also be influenced by calcium induced calcium release from the endoplasmic reticulum (ER) and/or mitochondrial buffering of intracellular calcium. The contribution of these calcium sources has been studied in postganglionic sympathetic neurons from the stellate ganglia at the whole cell level through patch clamping and fluorescence imaging. The pre‐hypertensive SHR for example has a larger intracellular calcium transient, associated with a higher ER calcium content, lower mitochondrial membrane potential and reduced mitochondrial buffering of intracellular calcium (Li et al. 2012). The whole cell calcium current is also larger in the pre‐hypertensive SHR (Lu et al. 2015). Moreover increased NA release (Shanks et al. 2013 b) and impaired NA re‐uptake (Shanks et al. 2013 a) is also observed in the SHR at this developmental stage which may contribute to a larger tachycardia during stimulation of the stellate ganglia and increased heart rate observed in vivo (Shanks et al. 2013 b).

Co‐transmitters as neuromodulators

Sympathetic co‐transmitters

Sympathetic nerves throughout the autonomic nervous system contain co‐transmitters such as ATP, neuropeptide‐Y (NPY) and galanin. Both NA and ATP have relatively short half‐lives as NA is taken back into the nerve terminal via the NA transporter and ATP is rapidly metabolized. Conversely whilst the half‐lives of NPY and galanin are significantly longer, they are released in much smaller quantities (Burnstock, 2009). The role of co‐transmitters may therefore be more important during conditions associated with high levels of adrenergic drive. For example, plasma NPY levels are elevated following myocardial infarction (Cuculi et al. 2013) and during left ventricular failure in humans where they correlate positively with one‐year mortality (Hulting et al. 1990; Ullman et al. 1994). In the presence of β‐blockers, which improve mortality in these conditions, both sympathetic release of NPY and galanin can produce crosstalk inhibition of vagal neurotransmission via the Y₂ and GALR₁ receptors in animal models resulting in a shift in autonomic balance (Herring et al. 2008, 2012). Moreover, NPY may also have a direct action on ventricular myocytes to increase susceptibility to ventricular arrhythmias (Herring, 2015) and also limit myocardial perfusion through vasoconstriction of the coronary vasculature (Shanks & Herring, 2013).

Parasympathetic co‐transmitters

Many parasympathetic neurons within the heart also co‐stain for different co‐transmitters including NPY, somatostatin, VIP and endogenous opioids such as dynorphins (Steele & Choate, 1994). Whether these co‐transmitters are released during physiological activation of the vagus nerve or change their expression profile in cardiovascular disease is not known. For example, inhibitors of VIP enhance vagally mediated bradycardia in the rat although high frequency stimulation is required to observe these effects (Hogan & Markos, 2006). Several cardiac opioid analogues have also been identified, including the heptapeptide methionine‐enkephalin‐arginine‐phenylalanine (MEAP) (Younes et al. 2000). MEAP has been shown to be vagolytic (Farias et al. 2001), probably via δ₂ receptors (Jackson et al. 2001), whilst low doses of MEAP can augment vagal bradycardia, via δ₁ receptors (Farias et al. 2003). Interestingly, repeated transient ischaemia to the sinoatrial node artery is associated with an increase in locally released MEAP and increased vagal bradycardia (Deo et al. 2009) whilst δ₁ receptor agonists significantly reduce infarct size in the rat (Schultz et al. 1998). Nevertheless, many questions still remain regarding the pathophysiological role of different neuropeptides.

Nitric oxide generated by nNOS has also been suggested to act as a co‐transmitter responsible for the cholinergic modulation of ventricular fibrillation threshold via NO acting independently of the muscarinic receptor (Brack et al. 2011). Early studies demonstrated that the anti‐fibrillatory effect of the vagus on the ventricle was abolished by the muscarinic receptor antagonist atropine (Corr & Gillis, 1974; Yoon et al. 1977; Vanoli et al. 1991). Muscarinic agonists such as oxotremorine and methacholine can also reduce ventricular arrhythmias following myocardial infarction (De Ferrari et al. 1992, 1993). However, in an isolated Langendorff perfused rabbit heart, recent work suggests the anti‐fibrillatory action of direct vagal stimulation appears to be preserved in the presence of atropine, abolished in the presence of hexamethonium (Brack et al. 2011), and dependent on the generation of nitric oxide by nNOS acting in a paracrine fashion (Ng et al. 2007; Brack et al. 2009). This seemed surprising given previous observations with atropine and muscarinic agonists in vivo, and the fact that NO generated by nNOS can facilitate acetylcholine release (Herring & Paterson, 2001). More recently, vagus nerve stimulation has been shown to be anti‐arrhythmic in the setting of ischaemia–reperfusion via effects on mitochondrial function with atropine abolishing this effect (Shinlapawittayatorn et al. 2014). Silencing vagal preganglionic neurons from the dorsal motor nucleus of the vagus using an optogenetic approach also limits infarct size during ischaemia–reperfusion via an atropine dependent pathway (Mastitskaya et al. 2012). Some beneficial changes in ventricular electrophysiology were still observed, suggesting there might be a paracrine action via NO, since this action was removed following NOS inhibition (Machhada et al. 2015). However, in this issue of The Journal of Physiology, Kalla et al. (2016) challenge the functional significance of the paracrine effect of NO. Whilst they demonstrate that the acetylcholine analogue carbamylcholine raises ventricular fibrillation threshold (VFT), they observed its effect was prevented by atropine, the nicotinic receptor antagonist mecamylamine, and inhibitors of neuronal nitric oxide synthase (nNOS) and soluble guanylyl cyclase (sGC). Moreover, the NO donor sodium nitroprusside could mimic the anti‐fibrillatory action of carbamylcholine, but its action was again ultimately dependent on muscarinic receptor stimulation, since all effects were blocked by atropine (Kalla et al. 2016).

Questions and controversies

What are the physiological roles of cardiac sympathetic and parasympathetic co‐transmitters?

Are they just neuromodulators or do they play a major role as co‐transmitters during cardiovascular disease?

Can these pathways be exploited therapeutically in patients receiving conventional pharmacological and interventional treatments?

Paracrine neuromodulators

Natriuretic peptides

The natriuretic peptide family is made up of at least three distinct peptides: atrial natriuretic peptide (ANP), brain derived natriuretic peptide (BNP), and C‐type natriuretic peptide (CNP) which share a disulphide linked 17 amino acid ring structure giving them their functionality. ANP is secreted from the atria in response to myocardial stretch producing natriuresis and vasorelaxation, whilst BNP is synthesized and secreted in the atria and ventricle, particularly during congestive heart failure where plasma levels are a prognostic indicator (Doust et al. 2005). CNP is the predominant natriuretic peptide of the nervous system (Sudoh et al. 1990) and is also found within vascular endothelium and the human ventricle (Wei et al. 1993). BNP is able to act directly on cardiac sympathetic neurons to reduce the intracellular calcium transient and release of NA during neuronal depolarization via a cGMP–PKG dependent pathway (Li et al. 2015). Conversely, exogenous BNP and CNP are also capable of directly augmenting vagal neurotransmission in a manner similar to NO via generation of cGMP (Herring et al. 2001 a). BNP may therefore be expected to result in a beneficial shift in cardiac autonomic balance towards sympathetic withdrawal and greater vagal tone during chronic heart failure, although cGMP dependent signalling may be limited by phosphodiesterase 2A (PDE2A) that is upregulated in this condition (Li et al. 2015). This may explain the lack of success of recombinant BNP (nesiritide), which has only been trialled in acute heart failure where results were disappointing (O'Connor et al. 2011).

Questions and controversies

Although a natriuretic peptide receptor agonist has been trialled unsuccessfully in acute decompensated heart failure, would their use as sympatholytic neuromodulating agents be beneficial in chronic stable heart failure if neuronal PDE2A levels can be controlled?

Ventricular myocytes as a source of acetylcholine

Rat ventricular myocytes express choline acetyltransferase (ChAT) and vesicular acetylcholine transporter, and the expression of these genes and the intracellular acetylcholine content can be modulated by muscarinic receptor inhibition (Kakinuma et al. 2009). This suggests that cardiomyocytes possess an acetylcholine synthesis system that is positively modulated by cholinergic stimuli and may therefore amplify the beneficial effects of vagal stimulation on the ventricles. In keeping with this, cardiac specific overexpression of ChAT in mice improves survival post myocardial infarction (Kakinuma et al. 2013). Whether such a system plays an important role in large mammals and humans is yet to be determined.

Angiotensin II

The renin–angiotensin system plays a key role in the regulation of blood pressure, circulating volume and neuronal function. In addition to the classical pathway via angiotensin converting enzyme (ACE) in the pulmonary vasculature, the presence of local cardiac ACE (Saito et al. 1987) raises the possibility of local AT II synthesis and influence on neurotransmission. Inhibitors of ACE and AT₁ receptors facilitate the heart rate response to peripheral vagal nerve stimulation (Takata et al. 2004) whilst AT II inhibits myocardial acetylcholine release (Kawada et al. 2007) via the AT₁ receptor, although the intracellular pathway potentially linking the AT₁ receptor to transmitter release has not been elucidated. Moreover, cardiac AT II levels are higher in the SHR where it potentiates NA release reinforcing the sympathetic hyperactivity (Herring et al. 2011). The use of ACE inhibitors post myocardial infarction (e.g. AIRE trial) and in chronic heart failure (e.g. SOLVD and CONSENSUS trials) is well established where they confer a mortality benefit.

Neurohumoral impact on cardiac adaptations in disease

Role of sensory afferent nerves in the diseased heart

The cardiac autonomic nervous system is composed of efferent and afferent nerves. In contrast to sympathetic innervation, which has well‐defined roles in the progression of cardiac disease, sensory innervation and its alteration in diseased hearts have not been elucidated sufficiently. Sensory nerves within the heart express the peptides substance P and CGRP, and those peptides have differing effects in cardiovascular disease. Substance P induces adverse myocardial remodelling (Melendez et al. 2011), while CGRP has cardioprotective effects in pathophysiological conditions including cardiovascular disease and diabetes mellitus (Brain & Grant, 2004; Chai et al. 2006) (Fig. 2). CGRP is also an important angiogenic factor for microvascular formation in the diseased heart (Li et al. 2013 b). Experimentally, CGRP is cardioprotective by activating the PI3K/Akt and ERK1/2 mitogen‐activated protein kinase (MAPK) kinase pathways and restoring Bcl‐2/Bax to inhibit apoptosis in cultured rat cardiomyocytes (Schaeffer et al. 2003; Ma et al. 2013; Umoh et al. 2014). CGRP may also have a protective function in heart failure through decreased inflammation, cell death and fibrosis.

Figure 2. Diagram of the pathophysiological interactions between the cardiac sensory nerves and diseased state .

CGRP can cause activation of multiple signalling pathways and has cardioprotective effects. Cardiovascular disease and autonomic neuropathy can lead to downregulation of CGRP, which causes inflammation, vasoconstriction, impairing angiogenesis, fibrosis and apoptosis, and may ultimately lead to heart failure or sudden cardiac death. CGRP, calcitonin gene‐related peptide; ROS, reactive oxygen species; PI3K, phosphatidylinositol‐3 kinase; Akt, serine/threonine‐specific protein kinase; ERK, extracellular signal‐regulated kinases; Bcl‐2/Bax, B‐cell lymphoma 2/Bcl‐2‐associated X protein.

Role of sympathetic efferent nerves in the diseased heart

The contributions of sympathetic dysfunction to cardiovascular pathology have been the subject of many reviews (Esler et al. 1997; Lefkowitz et al. 2000; Gourine & Gourine, 2014; Shen & Zipes, 2014; Fukuda et al. 2015; Gardner et al. 2016). Under normal physiological conditions, NA released from sympathetic nerves impacts cardiac ion channels and Ca²⁺ handling to enact prototypical chronotropic and inotropic responses. Briefly, NA binds to β‐ARs on cardiac myocytes leading to phosphorylation of several intracellular targets, including L‐type Ca²⁺ channels, K⁺ channels, phospholamban and ryanodine receptors (RyRs) (Bers, 2002). Phosphorylation of the L‐type Ca²⁺ channel (Cav1.2) increases current, which by itself would be expected to increase the cardiac action potential duration (APD). However, this increase is counterbalanced by the effect of β‐AR on delayed rectifier K⁺ currents, most notably an increase in rapidly activating I _Ks, but slowly activating I _Kr may also be involved (Kagan et al. 2002; Marx et al. 2002). The net effect is typically a shortening of the APD, which is required to accommodate fast heart rates during sympathetic activity.

Phosphorylation of phospholamban leads to increased sarco(endo)plasmic reticulum Ca²⁺‐ATPase (SERCA) activity and an increase in the Ca²⁺ content of the SR. Although increased SR Ca²⁺ content facilitates positive inotropy, SR Ca²⁺ overload can also occur, resulting in spontaneous opening of RyRs and Ca²⁺ release into the cytosol that is not in response to an action potential. This leads to Ca²⁺ extrusion from the cytosol via the Na⁺/Ca²⁺ exchanger (NCX), which produces a net inward current (Pogwizd et al. 2001). If the inward current is large enough, the cell membrane depolarizes and a triggered action potential occurs, also known as a delayed after‐depolarization (DAD).

Long term remodelling of these prototypical cardiomyocyte responses occurs as a result of chronic sympathetic hyperinnervation or denervation, and both of these conditions have been linked to increased risk of ventricular arrhythmias and sudden cardiac death (Cao et al. 2000 a,b; Liu et al. 2003; Li et al. 2004 b; Boogers et al. 2010; Nishisato et al. 2010; Vaseghi et al. 2012; Fallavollita et al. 2014). A loss of cardiomyocyte β‐AR responsiveness over time is a hallmark of sustained adrenergic stimulation and hyperinnervation. Changes in β‐AR sensitivity are probably due to altered expression of either β‐ARs or G‐protein receptor kinase 2 (GRK2, also known as βARK1) (Iaccarino et al. 1998; Yatani et al. 2006). In addition to decreased responsiveness to NA, APDs may be prolonged in hyperinnervated regions due to NGF‐induced downregulation of K⁺ currents (Heath et al. 1998). This increase in APD may also facilitate early after‐depolarizations (EADs), in which a secondary depolarization occurs before final repolarization. Conversely, chronic sympathetic denervation can lead to β‐AR supersensitivity, which results from upregulation of β‐ARs and increased responsiveness upon catecholamine binding (Vatner et al. 1985) due to downregulation of GRK2 (Yatani et al. 2006). Denervation also leads to a loss of I _to, which is responsible for the initial repolarization in phase 1 of the action potential (Bru‐Mercier et al. 2003; Bai et al. 2008). Gardner et al. recently showed in a murine model that reperfused infarcts remain denervated and display β‐adrenergic supersensitivity, calcium mishandling, and decreased I _to that are reversed by restoring the innervation (Gardner et al. 2015).

In addition to these cellular level responses, abnormally heterogeneous release of NA due to hyperinnervation or denervation leads to tissue level responses that also facilitate arrhythmia. Heterogeneous sympathetic nerve distribution may produce non‐uniform shortening of cardiac APD and increased dispersion of repolarization. Although non‐uniform APD shortening occurs even under normal sympathetic control (due to the normally heterogeneous sympathetic nerve distribution), Ajijola and colleagues found that following myocardial infarction, sympathetic stimulation further increased the dispersion of repolarization and altered activation and propagation patterns (Ajijola et al. 2013 b). These results were confirmed in human patients with myocardial infarction in which reflex sympathetic stimulation caused a 230% increase in dispersion compared to patients with structurally normal hearts (Vaseghi et al. 2012). Collectively, these increases in dispersion of repolarization increase the likelihood of conduction block and reentrant arrhythmias. Furthermore, localized release of adrenergic agonists (e.g. hyperinnervation) has been shown to induce propagating premature ventricular complexes (PVCs) and sustained focal ventricular tachycardia in intact hearts (Nash et al. 2001; Myles et al. 2012, 2015).

However, NA is not the only substance released by sympathetic nerves in the heart during pathological conditions. The impact of sympathetic NPY on coronary arteries was described previously, but the effect of ACh released from cardiac sympathetic nerves is not yet understood. Mice with heart failure and cholinergic transmission from cardiac sympathetic nerves had significantly improved survival and ventricular function compared with sympathetic nerve‐specific gp130‐deficient mice, which lacked sympathetic ACh (Kanazawa et al. 2010). This suggests that sympathetic cholinergic transdifferentiation might be an adaptive response that protects the heart. Consistent with this idea, cholinergic signalling via cardiac muscarinic acetylcholine receptors decreases the susceptibility of ventricular myocytes to chronic adrenergic stress, improving cardiac function (LaCroix et al. 2008). Furthermore, cardiomyocytes can also synthesize and release ACh, which is positively modulated by cholinergic stimuli. This suggests a positive feedback mechanism for ACh release, which may further any cardioprotective effects (Kakinuma et al. 2009). Similarly, augmenting parasympathetic nerve activity decreases mortality rates in heart failure, suggesting a beneficial effect of cholinergic signalling (Li et al. 2004 a). However, it remains controversial if release of ACh from sympathetic nerves, which are distributed differently from parasympathetic nerves within the myocardium, is a favourable or unfavourable event for cardiac performance and prognosis (DeMazumder et al. 2015).

Thus, the spatial and temporal innervation pattern and activity of sympathetic nerves directly impacts pathogenesis in the diseased heart (Fig. 3).

Figure 3. Crosstalk between cardiomyocyte and sympathetic nerve via humoral factors in diseased heart .

Failing cardiomyocytes induce NGF via an endothelin‐1 mediated pathway and LIF. NGF leads to hyperinnervation (anatomical modulation) and LIF leads to rejuvenation/cholinergic differentiation (functional modulation). This phenomenon shows the expression of catecholaminergic marker such as reduced TH and increased cholinergic (CHT, ChAT) and juvenile markers (PSA‐NCAM). PSA‐NCAM, polysialylated neural cell adhesion molecule; TrkA, tropomyosin‐related kinase A; NGF, nerve growth factor; LIF, leukemia inhibitory factor; LIFR, LIF receptor; TH, tyrosine hydroxylase.

Questions and controversies

Does sympathetic release of ACh protect the heart?

Does sympathetic ACh have the same effect in heart failure and acute myocardial infarction?

Neurohumoral regulation of calmodulin‐dependent protein kinase II in health and disease

The multifunctional Ca²⁺‐ and calmodulin‐dependent protein kinase II (CaMKII) is a serine/threonine kinase that is abundant in myocardium. CaMKII appears to augment myocardial performance in response to physiological stress, at least in part, by neurohumoral agonist stimulation. CaMKII catalyses phosphorylation events that activate many of the voltage‐gated sarcolemmal ion channels and Ca²⁺ homeostatic proteins orchestrating excitation–contraction coupling (Rokita & Anderson, 2012). The role of CaMKII‐catalysed phosphorylation of the type 2 intracellular Ca²⁺ release ryanodine receptor (RyR2) seems to be particularly important in augmenting release of myofilament activating Ca²⁺ from the SR during neurohumoral agonist stimulation (Grimm et al. 2015; Polakova et al. 2015) and during oxidant stress (Shirokova et al. 2014). In cardiac pacemaker cells, CaMKII activation contributes to fight or flight heart rate increases in response to physical activity and isoproterenol (isoprenaline) (Wu et al. 2009). Thus, CaMKII activation downstream to neurohumoral agonists acts to augment myocardial performance under physiological stress. However, a large body of evidence supports a view that CaMKII is also a nodal signal for promoting heart failure and arrhythmias (Swaminathan et al. 2012). Myocardial diseases are marked by sustained elevation of neurohumoral signalling and excessive CaMKII activity. Furthermore, CaMKII is activated by neurohumoral agonist pathways that are amongst the best clinically validated targets for heart failure therapeutics. Thus, it seems likely that high impact cardiovascular drugs, including β‐adrenergic receptor antagonists (β‐blockers), angiotensin II receptor blockers (ARBs), angiotensin converting enzyme (ACE) inhibitors and mineralocorticoid receptor (MR) antagonists are beneficial, at least in part, because they mitigate against CaMKII hyperactivation under conditions of sustained, pathological neurohumoral stress.

The CaMKII holoenzyme consists of two stacked hexamers and includes a total of 12 homologous or identical subunits (Chao et al. 2011). Each monomeric CaMKII subunit joins the holoenzyme by a C‐terminal association domain (Fig. 4). The kinase domain, containing the ATP binding pocket resides in the N‐terminus. The regulatory domain is interposed between the kinase and association domains and consists of an autoinhibitory region (N‐terminus) and a calmodulin (CaM) binding region (C‐terminus). Under resting conditions, CaMKII is largely inactive because the kinase domain engages the autoinhibitory region, which acts as a pseudosubstrate. However, when the concentration of intracellular Ca²⁺ ([Ca²⁺]_i) rises and binds to CaM the calcified CaM (Ca²⁺/CaM) embraces the CaM binding region and induces a conformational change that distorts the relationship between the catalytic domain and the autoinhibitory region, leading to disinhibition and activation of CaMKII. Brief, low frequency [Ca²⁺]_i elevations permit Ca²⁺/CaM unbinding and a return to autoinhibition. However, when [Ca²⁺]_i elevations are rapid or sustained CaMKII transitions between Ca²⁺/CaM‐dependent and Ca²⁺/CaM‐autonomous conformations by an autophosphorylation process (Kuret & Schulman, 1985; Schworer et al. 1986). Autophosphorylation of a conserved threonine at position 286 or 287 (Thr 287, numbering varies slightly by isoform) in the regulatory domain enhances the avidity of Ca²⁺/CaM binding (so called CaM trapping) (Meyer et al. 1992), favouring sustained CaMKII activation, but also prevents effective autoinhibition even after Ca²⁺/CaM unbinding, allowing for Ca²⁺/CaM autonomous CaMKII activity. More recently, we discovered a mechanism for Ca²⁺/CaM independent CaMKII activity by oxidation of a methionine pair (281/282) (Erickson et al. 2008). Oxidized CaMKII (ox‐CaMKII) does not support CaM trapping and is reversible by a methionine reductase (MsrA, methionine sulfoxide reductase A). Interestingly, CaMKII activation can be sustained at very low [Ca²⁺]_i in oxidative environments, suggesting that reactive oxygen species (ROS) sensitize the Ca²⁺/CaM dependence of CaMKII activation. Other post‐translational modifications to the regulatory domain shown to increase Ca²⁺/CaM autonomous CaMKII activity are O‐GlcNAcylation (at serine 280) (Erickson et al. 2013) and nitrosylation (at cysteine 290) (Gutierrez et al. 2013; Curran et al. 2014). Cysteine 290 may show more nitrosylation during catecholamine stimulation but other nitrosylation site(s) may lead to reduced CaMKII activity (Erickson et al. 2015). O‐GlcNAcylation is not presently known to be downstream to neurohumoral agonist stimulation. Thus, a rich variety of regulatory domain post‐translational modifications account for important aspects of CaMKII activation in response to neurohumoral agonist stimulation.

Figure 4. Neurohumoral activation of CaMKII .

Catecholamines, angiotensin II and aldosterone are upstream activators of CaMKII by inducing post‐translational modifications (PTM) to the N‐terminus of the regulatory domain (blue horizontal bar, see text). Catecholamine agonists promote phosphorylation of threonine 287 and nitrosylation of cysteine 290. Angiotensin II and aldosterone increase oxidant stress leading to oxidation of methionines 281 and 282. Hyperglycaemia can also activate CaMKII directly (by O‐Glycnacylation of serine 280) and by increasing reactive oxygen species and oxidizing methionines 281 and 282. The inactive CaMKII holoenzyme exists in a compact configuration. Intracellular Ca²⁺ increases lead to calcification of calmodulin (Ca²⁺/CaM) that binds to the C‐terminus of the regulatory domain. Ca²⁺/CaM binding induces CaMKII into an extended, active conformation that is inactivated after Ca²⁺/CaM unbinding. Post‐translational modifications to the regulatory domain keep CaMKII in a persistantly extended, active state even after Ca²⁺/CaM unbinding. ARB, angiotensin II receptor blockers; ACE, angiotensin converting enzyme; MR, mineralocorticoid receptor.

The role of Ca²⁺/CaM independent CaMKII activation appears to be important for the role of CaMKII in neurohumoral agonist induced myocardial disease. At this time, neurohumoral agonist initiated pathways are best understood to activate Ca²⁺/CaM autonomous CaMKII activity by increasing Thr 287 autophosphorylation or Met 281/282 oxidation. Infusion of the mixed β_1/2 receptor agonist isoproterenol is a model for catecholamine induced cardiomyopathy. In our hands, isoproterenol activates CaMKII and induces Ca²⁺/CaM independent CaMKII activity primarily by favouring Thr 287 autophosphorylation (Erickson et al. 2008). Features of isoproterenol induced myocardial toxicity include hypertrophy, cell death, fibrosis and depressed left ventricular contraction (Zhang et al. 2005; Yang et al. 2006). Mice with transgenic expression of a CaMKII inhibitory peptide (AC3‐I) (Zhang et al. 2005) or a protein kinase A (PKA) inhibitory peptide (PKI) (Zhang et al. 2013) are resistant to CaMKII activation and isoproterenol infusion‐induced myocardial injury. The precise nature of pathways for activating CaMKII in response to β‐adrenergic agonist stimulation remain controversial because some evidence supports exclusive, or near exclusive, CaMKII activation as a consequence of PKA‐induced intracellular Ca²⁺ mobilization (Zhang et al. 2013), while other studies support a role for the PKA‐independent, exchange proteins directly activated by cAMP (EPAC) pathway (Grimm et al. 2015).

Questions and controversies

Do β‐adrenergic receptors activate CaMKII solely through PKA, or does EPAC play a role?

Although detailed dose ranging studies are lacking, isoproterenol appears to be a more toxic myocardial agent compared to angiotensin II or aldosterone because these latter two agents do not induce substantial loss of left ventricular contraction in the absence of a ‘second hit’, such as myocardial infarction. We found that mice with myocardial targeted expression of a CaMKII inhibitory peptide, AC3‐I, were protected against isoproterenol infusion induced cardiomyocyte death (Yang et al. 2006), cardiomyopathy and myocardial infarction (Zhang et al. 2005), a condition where elevated neurohumoral signalling contributes to increased mortality. Mice with myocardial and mitochondrial CaMKII inhibition by transgenic expression of another CaMKII inhibitory peptide, CaMKIIN (Chang et al. 1998), are also protected against myocardial death after isoproterenol injection or myocardial infarction (Joiner et al. 2012), suggesting CaMKII actions in mitochondria contribute to catecholamine cardiotoxicity. CaMKIIδ knock‐out mice are also protected from pathological stress responses after aortic banding (Backs et al. 2009; Ling et al. 2009), a model of elevated afterload, where neurohumoral signalling probably contributes to adverse outcomes. Angiotensin II infusion, to approximate blood concentrations present in patients with heart failure, oxidizes CaMKII at Met 281/282 (ox‐CaMKII) leading to sinus node cell death (Swaminathan et al. 2011), inducing a heart rate disease phenotype resembling ‘sick sinus syndrome.’ Angiotensin II infusion increases susceptibility to rapid pacing induced atrial fibrillation and knock‐in mice lacking Met 281/282 on CaMKIIδ or knock‐in mice lacking a validated CaMKII phosphorylation site on RyR2 (serine 2814) are resistant to angiotensin II biasing of atrial fibrillation induction (Purohit et al. 2013). Aldosterone infusion, to achieve blood levels present in patients with heart failure after myocardial infarction, increases ox‐CaMKII and promotes myocardial rupture by enhancing ox‐CaMKII mediated expression of matrix metalloproteinase 9 in mice after myocardial infarction surgery (He et al. 2011). Taken together, these and other findings support a view that CaMKII is an important component of physiological and pathological pathways central to neurohumoral agonist signalling in myocardium.

Adapting novel molecular and cellular tools for neurocardiology research

Detailed studies of how whole‐heart function is altered by the activity of specific populations of neurons that innervate the heart provide insights into neurocardiac physiology as well as disease mechanisms and potential therapies. However, controlled measurements of the interplay between nerve activity and the myocardium can be challenging, particularly for in vivo animal studies and in excised perfused heart experiments. For example, in a hybrid perfused heart approach, parasympathetic neurons were activated by careful dissection and electrical stimulation of the left and right vagus nerve while sympathetic neurons were activated by electrical stimulation of the spinal canal at the second thoracic vertebra (Ng et al. 2001; Brack et al. 2004). This approach is powerful because it provides for controlled activation of the cardiac nerves in the absence of circulating compounds, although spinal cord stimulation activates both left and right sympathetic pathways, reducing left–right specificity (Brack et al. 2004). The approach has been primarily used in rabbits and could be difficult to implement in smaller species due to the additional steps required to prepare the heart for perfusion, isolate the nerve tissue, and position stimulation electrodes. For in vivo animal studies, electrodes are usually implanted around the vagus nerve for chronic parasympathetic stimulation (Li et al. 2004 a), but this has the limitation of potentially stimulating both efferent and afferent nerves. Exposing and electrically stimulating the stellate ganglia provides sympathetic stimulation in animals (Ajijola et al. 2013 a) while sympathetic cardiac neurons can be silenced by removing or ablating the stellate ganglia (Schwartz, 2014; Vaseghi et al. 2014).

In other studies, exogenous neurotransmitters, or their analogues, are administered to the coronary circulation to activate postsynaptic receptors to study the downstream effect of receptor activation on whole‐heart function (Liu et al. 1999; Lang et al. 2015). The neurotransmitters circulate through the myocardium to ubiquitously activate their receptors. Although this approach reveals, for example, important aspects of adrenergic receptor sensitivity and expression within the myocytes, it does not recapitulate neurotransmitter release and the physiological kinetics and spatially dependent concentration of transmitters released from cardiac neurons. This can be an important experimental limitation when the spatial arrangement of innervation (Freeman et al. 2014), disparate activity levels of nerve bundles (Ng et al. 2001), or altered autonomic tone (Thayer & Lane, 2007; Cauley et al. 2015) influences myocardial function.

Optogenetics is a powerful way to selectively target and activate cells in living tissue with unmatched spatiotemporal specificity using light‐actuated ion channels, such as channelrhodopsin‐2 (ChR2). Originally developed for neuroscience (Boyden et al. 2005), with ongoing developments in cardiac tissue (Ambrosi et al. 2014; Nussinovitch & Gepstein, 2015), optogenetics provides unparalleled specificity for cellular activation. This has created new opportunities for investigating the interaction between the cardiac nerves and the myocardium, overcoming previous approaches that required dissection and non‐specific electrical stimulation of bundles of heterogeneous nerve fibres. Precise cell‐type activation can be accomplished by regulating ChR2 expression using Cre‐Lox recombination and cell‐specific promoters. In a transgenic implementation of this system, a Cre responder animal would have a loxP‐flanked STOP cassette upstream of a ChR2‐enhanced yellow fluorescent protein (EYFP) fusion gene. This animal would then be mated with a Cre driver animal that expresses Cre recombinase under the control of a cell‐specific promoter such as α‐myosin heavy chain (cardiomyocytes) (Zaglia et al. 2015), tyrosine hydroxylase (TH, sympathetic neurons) (Wang et al. 2014; Wengrowski et al. 2015), or choline acetyltransferase (ChAT, parasympathetic neurons) (Kalmbach et al. 2012). In the progeny, ChR2‐EYFP would then have promoter‐specific expression only in targeted tissues.

This approach was recently used to study the neural control of parasympathetic tone in the brainstem of transgenic mice (Wang et al. 2014). ChR2 was expressed in noradrenergic cells using Cre‐Lox recombination and a TH promoter. Parasympathetic cardiac vagal neurons were labelled with the retrograde fluorescent tracer X‐rhodamine isothiocyanate (XRITC). Brains were then excised, sliced to reveal the locus coeruleus and cardiac vagal neurons, and noradrenergic (TH) neurons were photo‐stimulated to study their influence on the cardiac vagal neurons. Photo‐stimulated action potentials in the noradrenergic neurons increased the frequency of inhibitory postsynaptic currents in cardiac vagal neurons, providing a circuit substrate for the withdrawal of parasympathetic activity and increase in heart rate that occurs during stressful events. The selective activation of noradrenergic cells using optogenetics was a critical component in the success of these studies.

Questions and controversies

Could therapies that target the brainstem blunt autonomic imbalance that develops during heart failure?

In other studies, transgenic mice similar to those of Wang et al. (2014) were used for expression of ChR2 in noradrenergic (TH) neurons to elicit a cardiac β‐adrenergic response that originated from sympathetic nerve terminals within the heart (Wengrowski et al. 2015). Hearts were excised and perfused using a standard Langendorff approach. The epicardial surface was illuminated with blue light to photo‐stimulate noradrenergic neurons to generate a myocardial adrenergic response. Changes in contractile force, heart rate, and electrical activity were measured before and after photo‐stimulation. Results demonstrated pronounced increases in heart rate and contractile force during photo‐stimulation as well as recovery of prestimulation function after cessation of photo‐stimulation. Photo‐stimulation also shortened myocardial action potential duration, consistent with β‐adrenergic activation, and made hearts more susceptible to arrhythmia. Photo‐stimulation of sympathetic nerve terminals provided approximately half the increase in contractile force and heart rate that resulted from ubiquitous β‐adrenergic receptor activation using isoproterenol (5 μm) (Wengrowski et al. 2015).

Questions and controversies

Could a premature ventricular contraction be initiated by the simultaneous activation of a minimum number of sympathetic nerve terminals?

How would increased fibrosis and scar tissue during heart failure, or within an infarct, modulate that minimum number?

Optogenetics has also been instrumental in studies of cardiac ischaemia–reperfusion injury by providing for selective stimulation of specific populations of cardiac neurons. Using optogenetics to activate neurons in the brainstem, Mastitskaya et al. showed that cardioprotection provided by remote ischaemic preconditioning in rats is dependent upon the activity of brainstem vagal preganglionic neurons (Mastitskaya et al. 2012). Lenti‐ and adenoviral vectors were administered to specific locations within the dorsal motor nucleus of the vagus nerve (DVMN) for selective expression of a channelrhodopsin variant (ChIEF) (Lin et al. 2009) to activate neurons using light and a Drosophila allatostatin receptor (AlstR) to silence neurons using allatostatin (Tan et al. 2006). ChIEF and AlstR were selectively expressed in cholinergic vagal preganglionic neurons of the DVMN by driving expression with a PRSx8 promoter (Hwang et al. 2001), which is activated by Phox2, a transcription factor expressed in cholinergic vagal preganglionic neurons of the DVMN. Cardioprotection provided by remote preconditioning was abolished when the brainstem DVMN neurons were silenced with allatostatin. Interestingly, photo‐stimulation of the brainstem DVMN neurons before ischaemia and during reperfusion (in the absence of remote preconditioning) effectively mimicked the effect of remote preconditioning. This work has provided new insights into the mechanisms of preconditioning, which appear to be strongly mediated by vagal activity and cardiac muscarinic receptors.

Questions and controversies

How should optogenetic technologies evolve to enable new therapies for human heart disease?

Is the temporal control provided by optogenetics always necessary for cardiac therapies or could a drug delivered through the bloodstream also selectively activate an expressed neural actuator (DREADDS, for example) to achieve the same therapeutic result?

Selective activation as well as selective inhibition of neuron populations using new technologies, such as optogenetics and designer receptors exclusively activated by designer drugs (DREADDs), provides many exciting opportunities for investigating neurocardiac physiology. This includes the potential for unravelling fundamental functions in cardiac control that have largely eluded complete understanding, such as how parasympathetic neurons inhibit sympathetic activity at the level of the brainstem and cardiac ganglia. It is likely that technologies such as optogenetics will also facilitate new targets of opportunity for restoring autonomic balance to improve cardiac function in many diseases. For example, futuristic therapies for restoring autonomic balance may include implantable devices (Xu et al. 2014) that provide in vivo illumination using low power micro‐LEDs (Kim et al. 2010) with wireless control to activate neurons that express light‐activated ion channels.

Additional information

Competing interests

None declared.

Funding

This work was supported, in part, by National Institutes of Health (NIH) Grants HL079031, HL096652, HL070250 and HL071140 (M.E.A.), HL068231 and HL093056 (B.A.H.), HL111600 (C.M.R.) and GM107949 (D.B.H.), and by the British Heart Foundation (N.H., D.J.P.).