Towards an understanding of the function of the intrinsic cardiac ganglia

Prior to the 1950s, the ability of the heart to respond to challenges was perceived primarily in terms of the intrinsic properties of the myocardium, but two captivating studies published at about mid-century inaugurated a re-examination of this issue. The finding that the diameter of the canine heart decreased during exercise relative to rest (Rushmer et al. 1959) was contrary to the then current understanding of Starling's Law and, shortly thereafter, the importance of the extrinsic (i.e. autonomic neural) control of cardiac function became more apparent. A second signal observation was Alexander's description of the ‘pressor’ and ‘depressor’ regions of the brainstem (see Gebber, 1990). Our knowledge of the central nervous control of autonomic function has expanded dramatically since that time, but, until recently, little attention had been directed to the role of the neurons residing at the heart itself. In fact, it is still not widely appreciated that these ‘intrinsic cardiac ganglia’ include neuronal types (e.g. sensory and interneurons) other than classic parasympathetic neurites so that the potential exists for intracardiac reflex control of the heart's function. The article by Thompson et al. (2000) in this issue of The The Journal of Physiology takes a key step in testing the reality of the concept of such a ‘heart brain’ (Randall et al. 1996). The canine intrinsic cardiac ganglia typically cluster within fatty pads at discrete locations on the heart. The concentration of ganglia studied by Thompson and colleagues was previously designated by its anatomical location within the ‘pulmonary vein fat pad’, but it is now generally referred to as the ‘right atrial ganglionated plexus’ (RAGP). This change in nomenclature reflects the growing evidence that these neurons serve in ways that transcend their classically held role as simple relay stations for activity emanating from the central nervous system. The location on the heart of this particular cluster of neurons is advantageous experimentally since the encompassing fatty tissue can be visualized from a ventral approach and stabilized sufficiently to allow extracellular recordings of the activity of individual neurons, even in the beating heart. Thompson and his colleagues simultaneously recorded the activity of neurons from two separate locations within the fat pad during the control state and during ventricular epicardial application of substances known to activate sensory neurites. Their most novel and exciting observation is that the activity of two populations of neurons within this ganglionated plexus displayed intervals of coupled behaviour every 15–30 s, with each interval lasting up to 10 s. These episodes occurred both during control and during epicardial application of veratridine. Thus they have demonstrated that these neurons are capable of functioning interdependently. The observations of Thompson and his co-workers could in theory be explained by one or both of two mechanisms: (1) the two pools of neurons in question received, and responded to, a common input(s), or (2) the neurons inter-communicated via synaptic connections. These investigators analysed the coupling between the two populations of neurons over time by computing a centred moving average of each population's instantaneous activity over 2 s periods; these two signals were then continuously cross-correlated over 5 min intervals using sliding windows of 10 s duration. This analysis is sensitive to any co-ordination of activity between the two signals within this window, and provides a continuous computation that ranges from +1 (i.e. demonstrating tight direct coupling) through zero (i.e. no demonstrable coupling) to −1 (i.e. demonstrating tight inverse coupling). As such, it effectively tests whether the two populations of neurons receive common input. Thompson and colleagues report coupling that attained peak values of 0·88 or larger. The mathematical algorithm does not specifically examine the phase relationship between the two signals. Assuming that any coupling attributable to the second mechanism, above, would entail a time offset introduced by the synaptic processes, this cross-correlation computation would be relatively mute as to any intraganglionic neural network mechanisms. More involved mathematical procedures would have to be utilized to discriminate effectively between these two mechanisms. The question naturally occurs as to the specific physiological function of the various concentrations of intrinsic cardiac ganglia that have been identified. Is it possible, for example, that some of the integrative control of the heart normally ascribed to the brain might, at least to some degree, be assumed by intracardiac reflexes?Smith (1999) has shown in the pig that input from the vagus nerve and from cardiopulmonary nerves converges on individual intrinsic cardiac neurons, thereby documenting the functional infrastructure for an interaction between the sympathetic and parasympathetic systems at the heart itself. Indeed, surgical removal of the canine posterior atrial ganglionated plexus (which, like the RAGP, contains neurons that project to the sinoatrial node) markedly attenuates a parasympathetic modulation of sympathetic effects on the cardiac pacemaker (Randall et al. 1998). Another particularly attractive hypothesis is that the intrinsic cardiac ganglia are responsible for a beat-by-beat co-ordination of sinoatrial and atrioventricular nodal function. In a broader sense, Armour et al. (1998) have proposed a provocative model which posits that cardiodynamics are regulated via a hierarchy of nested feedback loops, including loops restricted to the intrinsic cardiac ganglia. If so, we may be seeing the incipient convergence of concepts formulated 50 years ago, perhaps even bringing us towards a new understanding of how intrinsic and extrinsic mechanisms interact to control and co-ordinate cardiac function.