|Home | About | Journals | Submit | Contact Us | Français|
Neuronal networks are highly plastic and reconfigure in a state-dependent manner. The plasticity at the network level emerges through multiple intrinsic and synaptic membrane properties that imbue neurons and their interactions with numerous nonlinear properties. These properties are continuously regulated by neuromodulators and homeostatic mechanisms that are critical to maintain not only network stability and also adapt networks in a short- and long-term manner to changes in behavioral, developmental, metabolic, and environmental conditions. This review provides concrete examples from neuronal networks in invertebrates and vertebrates, and illustrates that the concepts and rules that govern neuronal networks and behaviors are universal.
Behaviors emerge through the complex integration of molecular, cellular, and synaptic properties involving components of the peripheral and central nervous systems. Neuronal inputs from the periphery and other areas of the central nervous system (CNS) converge onto central pattern generators (CPGs), centrally located neuronal networks that establish basic patterns of activity. As demonstrated for the mammalian respiratory network, activity patterns may emerge through the integration of different central neuronal networks interacting with a CPG located within the pre-Bötzinger complex (PreBötC) (Smith et al. 1991; Guyenet and Wang 2001; Tan et al. 2008; Bouvier et al. 2010; Pagliardini et al. 2011; Schwarzacher et al. 2011). The centrally generated activities are further processed at the level of motoneurons (Parkis et al. 1995; Gabriel et al. 2011). In the case of mammalian networks, various motor nuclei contribute to the generation of the overall behavior. The resulting behavior is shaped by numerous biomechanical and cellular properties of the muscles (Dickinson et al. 2000; Biewener 2006; Higham et al. 2011; Roberts et al. 2011) that not only determine the overall motor output but also set the behavioral conditions that generate proprioceptive information, which is fed back to the CNS (Buschges and Manira 1998; Borgmann et al. 2009). Although, this review focuses on the generation of motor behaviors, “higher” brain functions follow many of the same principles of network integration. One of these principles is that there is no neuronal network that is hard-wired. Short- and long-term plasticity of synaptic and intrinsic membrane properties, as well as neuromodulation, play critical roles at all levels of neuronal integration.
Many insights into the functioning of the CNS were gained from studying small neuronal networks of invertebrates. These networks exhibit similar degrees of complexity as those of neuronal networks located within the mammalian nervous system and thus they continue to serve as mechanistically amenable models. One of the key lessons learned was that a functional network is not defined by a predetermined set of intrinsic and/or synaptic properties. Instead, these properties emerge through a dynamic interplay between the cellular and subcellular elements of a neuronal network (Turrigiano et al. 1994, 1995; Thoby-Brisson and Simmers 1998; Golowasch et al. 1999). Applied to the mammalian nervous system this discovery led to the formulation of homeostatic plasticity and synaptic scaling (Turrigiano 2008).
Another important finding: computational simulation of a crustacean network revealed that an identical neuronal output can emerge from a large number of network configurations with very different synaptic and intrinsic properties (Prinz et al. 2004). This discovery led to the important proposal that the network output is more tightly regulated than are the cellular and subcellular elements that give rise to the expected output. Experimental studies showed that the intrinsic properties of crustacean neurons could vary 2- to 7-fold, but still produce a similar network output (Schulz 2006; Grashow et al. 2010). These discoveries have not been appreciated fully by researchers studying mammalian neuronal networks but, like the discovery of homeostatic plasticity, these findings have equally important implications for mammalian neurophysiology. These concepts also provide important mechanistic insights into neurological disorders.
In this review, we discuss concepts that contribute to the plasticity of neuronal networks by the tuning of intrinsic cellular properties, neuromodulation, scaling of synaptic transmission, and the state dependency of network configurations.
Early studies of networks viewed nerve cells mostly as passive structures. An influential hypothesis by McCulloch and Pitts (1943) proposed that neuronal networks consist of logical neuronal elements with a set threshold that are capable of performing complex information transformations (also see Perkel 1988). The complexity of network functions was thought to arise primarily from the complexity of the underlying connectivity. At that time, this idea was very radical and opened the way to the “possibility of a true science of the brain” (Perkel 1988). In its extreme, this hypothesis led to the notion that a complete understanding of how neurons are interconnected will allow us to predict how neurons generate activity patterns and even higher-level cognitive processes. This “connectionism” was among the first mechanistic approaches that could potentially allow “physical sciences to conquer psychological and related sciences” as reviewed by Donald Perkel (1988). The field of “connectionism” received renewed interest as first efforts were made to obtain a complete “connectome” of the mammalian nervous system (Lichtman et al. 2008).
The first connectivity studies were experimentally performed on small (mostly invertebrate) neuronal networks that give rise to rhythmic motor activities, the so-called CPG. Many of these studies were inspired by the half-center model proposed by Graham-Brown (1911, 1914). According to this model, rhythmicity arises from synaptic interactions between two reciprocally organized groups of inhibitory neurons. Additional tonic excitatory synaptic drive provides the required excitability. Reciprocally organized synaptic inhibitory interactions were identified in the neuronal networks of numerous invertebrates (Harris-Warrick and Marder 1991; Marder and Calabrese 1996; Sharp et al. 1996; Ramirez et al. 1998) and mammals (Ballantyne and Richter 1984; Ogilvie et al. 1992; Duffin et al. 2000; Rybak et al. 2006; Abdala et al. 2009).
While reciprocal inhibitory synaptic connections may contribute to the generation of alternating network activities, our increased understanding of the dynamic nature of network interactions resulted in a much more complex picture. Indeed, the idea that connectivity is the major determinant of a network’s output pattern needed to be significantly revised after early intracellular recordings demonstrated that neurons are not simple logical elements. Neurons were found that are capable of intrinsically bursting, both in invertebrates (Treistman and Levitan 1976; Russell and Hartline 1978) and vertebrates (Wong and Prince 1978; Llinas and Sugimori 1980; Dekin et al. 1985). It is now well established that bursting neurons are present in the majority of networks across the majority, if not all, animal species (Guckenheimer et al. 1997; Ramirez et al. 2004; Ramcharan et al. 2005; Soto-Trevino et al. 2005; Llinas and Steriade 2006; Marcuccilli et al. 2010; Selverston 2010; Mrejeru et al. 2011). In all of these networks, bursting interacts with synaptic mechanisms and contributes to the nonlinearities of network functions (Ramirez et al. 2004). This principle is illustrated in Fig. 1.
In the locust’s flight system, intrinsic bursting differentially amplifies functionally different synaptic inputs. Bursting amplifies only proprioceptive input from the tegula, a wing afference, but not synaptic input from the CPG. As a consequence, synaptic activity generated by the CPG is filtered out (Fig. 1A and B; Ramirez and Pearson 1991a, 1991b). This has important functional consequences; in the flying animal these proprioceptive inputs will establish the timing of wing elevation because the tegula is activated before the CPG itself (Ramirez and Pearson 1991b). The intrinsic bursting is conditional, and induced by the biogenic amine octopamine (Ramirez and Pearson 1991a, 1991b) and during flight when octopamine levels are elevated (Orchard et al. 1993; Ramirez and Pearson 1993). Consequently, activation of the tegula evokes bursting and wing elevation only in the flying, but not quiescent animal (Ramirez and Pearson 1993). Neurons that possess intrinsic bursting properties are also involved in the initiation and amplification of recurrent synaptic activity in multiple systems, including the neocortex (Stuart and Sakmann 1995; Schwindt and Crill 1999; van Drongelen et al. 2006), the hippocampus (Sanabria et al. 2001), the spinal cord (Darbon et al. 2004), and the respiratory network (Ramirez et al. 2004). In the respiratory network, bursting can also be conditional, and it is induced e.g., by norepinephrine, an amine very similar to octopamine (Viemari and Ramirez 2006). However, the ability to intrinsically burst is just one of numerous intrinsic membrane properties that interact with synaptic properties. A host of intrinsic membrane properties were discovered early, such as synaptic facilitation or depression, spike-frequency adaptation, and rebound bursting (Bullock 1959). Within half-center networks various intrinsic membrane properties play critical roles (Perkel and Mulloney 1974), of which the activation of the hyperpolarization-activated cation current (Ih-current) or the calcium-dependent potassium currents (KCa2+) are particularly important for establishing transitions in phase (Wang and Rinzel 1993; Sharp et al. 1996).
These intrinsic mechanisms greatly influence the type of output generated by a given synaptic network. Dependent on the specifics of the intrinsic and synaptic properties, the same synaptic architecture can generate alternating rhythmic activity and also synchrony or stochastic activities (Sharp et al. 1996; Sorensen and DeWeerth 2007). Thus, without detailed information on the timing, polarity, and strength of all synaptic currents, as well as all the underlying intrinsic membrane currents, it is almost impossible to predict the output that can be generated by any given network. Thus, the “simple connectionism,” which assumed that networks are composed of logical circuit elements that follow simple threshold rules, has been replaced by the much broader concept that considers networks as multi-layered and composed of highly nonlinear, dynamically regulated and modulated, neuronal elements with numerous intrinsic membrane properties. Because of this complexity, it becomes exceedingly difficult to predict the rich repertoire of activities that can emerge from any given network at any given state. In this review, we describe some of the principles that characterize and maintain the dynamic integration of the multitude of synaptic, intrinsic properties of membranes and cells. We hypothesize that there is not a single rhythm-generating principle that can be made responsible for any given rhythmic activity pattern, but rather that reciprocal synaptic mechanisms, recurrent excitation, rebound excitation, and intrinsic bursting are just a small sample of the rich repertoire of extracellular and intracellular mechanisms that govern the generation of activity patterns and ultimately of behavior.
Following the realization that networks and neurons, as well as their ionic, cellular, and subcellular components, are plastic, it became obvious that there must also be homeostatic mechanisms that regulate, maintain, and fine tune the cellular and network mechanisms that maintain stable levels of activity. One of the first concrete examples of homeostatic regulation was found in crustacean stomatogastric neurons. These neurons lose their rhythmic firing pattern once they are acutely isolated in organotypic culture (Thoby-Brisson and Simmers 1998), but they regain rhythmicity over a 3-day time course by increasing inward current densities of sodium and calcium (Turrigiano et al. 1994, 1995). These findings were formative for mammalian neuroscience and led to an increased understanding of how neurons regulate their intrinsic excitability under control conditions and during learning and memory (Woody et al. 1991; Aou et al. 1992). Deprivation of activity leads to increased inward sodium currents and decreased outward potassium currents in neocortical neurons (Desai et al. 1999). Activity regulates the strength of the Ih-current as shown in sensory-deprived animals. Under these conditions, dendritic Ih-currents decrease in layer-5 pyramidal neurons, which leads to an increased intrinsic excitability (Breton and Stuart 2009). In conscious cats, an increased excitability of motor cortical neurons is associated with the acquisition of the eye-blink reflex (Woody et al. 1991; Aou et al. 1992).
The alterations of these currents might not only determine the firing threshold, but also the firing pattern of a neuron (Turrigiano et al. 1994). Changes in a cell’s conductance can lead to switches between tonic and burst firing. These changes are typically mediated by neuromodulators that make these intrinsic properties “conditional” (Llinas and Sugimori 1980; Dickinson and Nagy 1983; Arshavsky Yu et al. 1985; Satterlie 1985; Llinas and Yarom 1986; Arbas and Calabrese 1987; Wallen and Grillner 1987; Hounsgaard and Kiehn 1989; Ramirez and Pearson 1991b; Godwin et al. 1996; Luo and Perkel 2002; Llinas and Steriade 2006). Changes in the firing threshold or firing mode are associated with either increases in inward currents (Llinas and Steriade 2006), or a downregulation of outward currents (Sanabria et al. 2001; Beck and Yaari 2008).
The regulation of inward and outward currents is an important mechanism in fine-tuning the properties of cellular firing (Llinas 1988; Desai et al. 1999; Marder and Prinz 2002; Aizenman et al. 2003; MacLean et al. 2003; Schulz et al. 2006), and it can change postnatally. A clinically relevant example is the regulation of endogenous pacemaking in dopaminergic neurons of the substantia nigra. Although this autonomous activity is very similar at different stages of postnatal development, this activity depends on Cav 1.3 calcium channels in adult mice, while in juvenile mice it is dependent on the Ih-current (mediated by HCN channels). When the calcium channels are pharmacologically inhibited for more than 1h in slices from adult mice, neurons reactivate the juvenile mechanism for pacemaking, which depends on HCN channels (Chan et al. 2007).
Homeostatic regulation determines also the differential distribution of ion channels along the dendrites as demonstrated for cortical layer-5 pyramidal neurons that possess long apical dendrites (Beck and Yaari 2008). In these neurons, distal dendrites exhibit an increased expression of the noninactivating Ih-current (Berger et al. 2001; Lorincz et al. 2002). This distribution contributes to the differential integration of dendritic and somatic inputs (Tsay et al. 2007).
The notion that neurons can adapt their activity through multiple homeostatic mechanisms and thereby maintain stability in function is now well established (Turrigiano et al. 1998; Desai et al. 1999; Davis and Bezprozvanny 2001; Desai et al. 2002; Maffei et al. 2004, 2006; Maffei and Turrigiano 2008; Turrigiano 2008; Koch et al. 2010). Self-tuning mechanisms often involve activity-dependent plasticity located at the synaptic terminals of a variety of systems. In cultured cortical neurons, synaptic scaling of both excitation and inhibition can occur in response to deprivation of activity (Turrigiano et al. 1998; Turrigiano 2008). The best-studied form of “synaptic scaling” is characterized by a global increase in the accumulation of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors at various synapses (O'Brien et al. 1998; Wierenga et al. 2005) while maintaining the relative synaptic strength between the individual connections.
The direct correlation between internal calcium levels [Ca2+]i and cellular activity presents a possible intracellular mechanism that underlies homeostatic plasticity (Turrigiano 2008). Consistent with this hypothesis, blockade of calcium channels induce up-scaling at excitatory synaptic terminals (Ibata et al. 2008). Moreover, pharmacological blockade of calcium-dependent kinases (i.e., the CaMK-family) prevents the effects of deprivation of activity on excitatory synaptic transmission (Thiagarajan et al. 2002). Intracellular calcium levels as a measure of cellular activity are thought to regulate various forms of plasticity (Lisman et al. 2002; Zhang and Linden 2003; Malenka and Bear 2004; Grubb and Burrone 2010) but other second-messenger pathways and molecules have also been implicated as mediators of homeostatic synaptic plasticity (Rutherford et al. 1998; Stellwagen and Malenka 2006; Aoto et al. 2008; Koch et al. 2010). Some of these molecules are part of the inflammatory pathway and include prostaglandin-E2 (PGE2) and tumor necrosis factor-α (TNF-α) (Stellwagen and Malenka 2006; Koch et al. 2010; Lee et al. 2010b; Steinmetz and Turrigiano 2010). Stellwagen and Malenka (2006) demonstrated that the TNF-α mediates homeostatic plasticity derives from glia rather than from neurons. However, TNF-α signaling may not be essential for inducing, but rather for maintaining, synaptic up-scaling in cortical neurons (Steinmetz and Turrigiano 2010).
There is increased evidence that homeostatic plasticity may also be involved in promoting pathophysiological states (Houweling et al. 2005; Trasande and Ramirez 2007; Avramescu and Timofeev 2008; Koch et al. 2010). PGE2, the major reaction product of the cyclooxygenase-2 enzymes (COX-2) acutely inhibits network activity in the neocortex, which could have important clinical consequences in the context of epilepsy (Koch et al. 2010). While synaptic excitation is inhibited if applied acutely, chronic exposure to PGE2 causes a presynaptic increase in excitatory synaptic transmission (Koch et al. 2010). This homeostatic response is illustrated in Fig. 2 for experiments that were performed in organotypic slice preparations.
Under control conditions, the cultured neocortical network generates spontaneously up-and-down states. One spontaneously occurring upstate is exemplified in Fig. 2A. The synaptic drive potential of the upstate is highlighted in brown. Forty-eight hours following the suppression of up-and-down activity caused by PGE2, the network generates upstates with significantly enhanced amplitudes (Fig. 2C). A similar homeostatic response can be obtained by activity deprivation achieved by the blockade of sodium channels with tetrodotoxin (TTX) (Fig. 2B). In the examples shown in Fig. 2, the homeostatic responses to activity deprivation were associated with a significantly enhanced synaptic drive.
The regulation of activity-dependent plasticity through inflammatory pathways is also relevant in the context of cardio-respiratory control, where it could be relevant in the context of sleep apnea, a disorder associated with recurrent, intermittent episodes of hypoxia. The COX-2 pathway is activated by hypoxia through hypoxia-induced factor-1α (HIF-1α) (Lee et al. 2010a), and PGE2 is known to directly affect the network generating respiratory rhythm (Hofstetter et al. 2007). Acute intermittent hypoxia (AIH), but not sustained hypoxia leads to the so-called long-term facilitation (LTF) of ventilation (Dwinell et al. 1997; Olson et al. 2001). Various forms of LTF have been described. AIH induces LTF at the level of motor neurons which results in a persistent (≤90min) increase in respiratory amplitude of phrenic motor activity (Millhorn et al. 1980; Turner and Mitchell 1997; Baker and Mitchell 2000; Olson et al. 2001). The signaling cascade triggered by intermittent hypoxia has been proposed to involve intermittent release of serotonin and activation of PKC by a reactive oxygen species (ROS) (MacFarlane et al. 2009) that stimulates new brain-derived neurotrophic factor (BDNF) production and the consequent activation of TrkB receptors. This cascade of events leads to increased density of postsynaptic glutamate receptors, increased efficacy of glutamatergic synaptic transmission and ultimately increased amplitude of respiratory motor output (Baker-Herman et al. 2004). However, intermittent hypoxia induces also a persistent increase in frequency discharge (≤90min) from the PreBötC (Blitz and Ramirez 2002). This long-term frequency modulation appears to be responsible for the increase in frequency also seen at the behavioral level. LTF also occurs at the level of sensory receptors. Subsequent to chronic intermittent hypoxia (CIH) conditioning the carotid bodies also results in long-term plastic changes in response to AIH (Peng et al. 2003). These observations imply that while motor LTF occurs with AIH alone, sensory LTF of the carotid bodies to AIH is conditional. Sensory LTF of the carotid bodies involves ROS-mediated oxidative stress and HIF-1α (Peng et al. 2006). Both HIF-1α and -2α appear to be important to the redox regulation of the carotid bodies and their responsiveness to hypoxia (Peng et al. 2011). Thus, the interplay between both HIF-1α and -2α seems to be an essential factor dictating the long-term physiological behavior of the carotid bodies in response to acute fluctuations in O2 with, and without, CIH conditioning. Understanding how intermittent hypoxia influences the multiple components responsible for the control of breathing demonstrates an important lesson about the integration of neuronal networks and the subsequent behavior from such integration. Behavioral plasticity, in this case, LTF of ventilation, does not involve only one form of plasticity in one area of the nervous system, but rather is the result of several forms of plasticity occurring at multiple levels of integration from sensory neurons, to central networks to motor neurons.
The concept that the same neuronal network can assume different configurations that give rise to different forms of rhythmic activities was first established in the neuronal networks of invertebrates (Dickinson and Nagy 1983; Flamm and Harris-Warrick 1986; Nusbaum and Marder 1989; Harris-Warrick and Marder 1991; Weimann et al. 1991; Marder and Calabrese 1996; Ramirez 1998). In these cases, the same neurons can switch their allegiance from one rhythmic pattern to another activity pattern (Dickinson et al. 1990; Meyrand et al. 1991; Ramirez 1998). As exemplified by Fig. 3, the same identified neurons located within the so-called subesophageal ganglion of locusts (Fig. 3B) are rhythmically active in phase with expiration in the quiescent animal (Fig. 3A) and activated in phase with the flight rhythm in the flying animal (Fig. 3C). Although activated in phase with the flight rhythm, these neurons are part of the generator of the respiratory rhythm because they can reset the respiratory rhythm when activated intracellularly (Fig. 3D). The switch in the rhythmic alliance seems to contribute to the reconfiguration of ventilatory behavior during the transition from quiescence to the metabolically much more demanding flight behavior (Ramirez 1998).
Similar principles have also been demonstrated for mammalian neuronal networks, in particular, for the respiratory network (Lieske et al. 2000). Although, the generation of breathing involves neurons and networks widely distributed throughout the CNS, including areas within the medulla, pons, cerebellum, amygdala, neocortex, and hypothalamus, one area seems to be very critical for the generation of these activities; located within the ventrolateral medulla, the so-called PreBötC is both essential (Smith et al. 1991; Ramirez et al. 1998; Wenninger et al. 2004; Tan et al. 2008) and sufficient to generate different forms of respiratory activities (Smith et al. 1991; Lieske et al. 2000; Gray et al. 2010).The same PreBötC neurons are activated during very distinct types of respiratory activities that include eupneic, gasp, and sigh activities (Lieske et al. 2000).
Although all these activities are respiratory in function, they serve different behavioral roles. Eupneic activity is seen during “normal breathing” and it provides the predominant ventilatory drive under normal conditions. Sigh activity occurs much less frequently and is an important mechanism preventing atelectasis (Reynolds 1962; Bendixen et al. 1964), but it is also activated during times of relief (Soltysik and Jelen 2005; Vlemincx et al. 2009, 2011); sighs constitute an important arousal mechanism (Wulbrand et al. 2008). Gasping is activated during severe hypoxia and becomes an important mechanism of autoresuscitation, and is the last chance for arousal from severe hypoxic conditions (Fewell 2005; Thach 2008). Activity maps of the isolated PreBötC reveal that sigh and eupneic activities (Fig. 4A), as well as gasps, are generated concurrently within the same area of the PreBötC (Lieske et al. 2000). The rhythmic activities characterizing sigh and eupneic activities seem to emerge by way of the differential activation of different synaptic and intrinsic membrane properties (Lieske and Ramirez 2006a, 2006b; Tryba et al. 2008). Generation of a sigh is critically dependent on P-type calcium currents and involves the activation of group III mGluR 8 receptors, while other metabotropic glutamate receptors are critical for the generation of eupneic activity (Lieske and Ramirez 2006a, 2006b). Intrinsic properties of membranes seem to be as critical as different activation of synaptic mechanisms in mediating the configurations of this network. As shown in Fig. 4B and C, even after synaptic isolation from the network, individual pacemaker neurons are capable of generating two types of bursting mechanisms. The two intrinsic bursting types are differentially affected by the same neuromodulators that affect eupneic and sigh activities. Activation of muscarinic receptors inhibits the small intrinsic bursting and eupneic activities, while large-amplitude intrinsic bursting is significantly accelerated, as is the generation of sighs (for more detail see Tryba et al. 2008).
The neurons of the PreBötC possess a wide variety of ionic conductances that contribute to the respiratory rhythm. The persistent sodium current (INAP), and the calcium-activated nonselective cation current (ICAN) are important for shaping the intrinsic excitability of respiratory neurons (Ramirez et al. 1997; Del Negro et al. 2002; Pena et al. 2004; Del Negro et al. 2005; Pace et al. 2007; Rubin et al. 2009). These inward currents are critical for the generation of intrinsic bursting properties that distinguish two types of respiratory pacemakers: cadmium-sensitive (CS) and cadmium-insensitive (CI) pacemaker neurons (Thoby-Brisson and Ramirez 2000; Pena et al. 2004; Del Negro et al. 2005). Cadmium-sensitive pacemaker neurons cease to burst in the presence of CdCl2, which blocks all voltage-dependent calcium channels or flufenamic acid (FFA) which inhibits the ICAN. Bursting in most CI pacemaker neurons persists in CdCl2, but is abolished by riluzole (RIL), a blocker of the persistent sodium current. These inward currents also play important roles in amplifying synaptic transmission (see also Fig. 1). At the network level, pharmacological blockade of either INAP or ICAN alone does not abolish the rhythmic activity in the PreBötC, suggesting that either of these currents alone is not essential for the generation of network bursting (Pena et al. 2004; Del Negro et al. 2005). Yet, a combination of FFA and RIL is sufficient to block the respiratory rhythm in vitro (Tryba et al. 2006) and in vivo (Pena and Aguileta 2007).
The network’s configuration and its dependency on different rhythm-generating mechanisms changes under exposure to hypoxic conditions. An initial augmentation phase is characterized by a decrease in synaptic inhibition, which has been observed both in vivo and in vitro (Richter et al. 1991; England et al. 1995; Volker et al. 1995; Ramirez et al. 1998; Thoby-Brisson and Ramirez 2000). The depression of synaptic inhibition contributes to the reconfiguration of the respiratory network by altering the discharge pattern of a variety of neurons. Inspiratory neurons change from an augmenting to a decrementing burst discharge, while expiratory neurons become tonically active and postinspiratory neurons begin to discharge in phase with inspiration (Ramirez et al. 1997). This reconfiguration at the level of the CPG is associated with distinct changes at the level of motor output. Some expiratory motor neurons seem to loose rhythmic activity, while neurons of the hypoglossus nucleus (XII) exhibit a massive increase of amplitude in rhythmic bursts both in vivo and in vitro (Ramirez et al. 1997; Telgkamp and Ramirez 1999). Within the PreBötC, CAN-current-dependent pacemaker neurons cease to burst during this reconfiguration. The cessation of burst discharge in these neurons plays an important role in the transition from normal breathing to gasping. Gasping is characterized by the persistence of the activity of persistent-sodium-dependent pacemaker neurons (Pena et al. 2004).
The transition from normal breathing to gasping is associated with a change in the relative functional importance of different ionic conductances. During gasping, but not while in the normal respiratory state, the network relies only on the persistent sodium current (INAP) and can be blocked in vitro and in vivo by Riluzole (Pena et al. 2004; Paton et al. 2006; Pena and Aguileta 2007). The dependency of gasping on one particular mechanism of generating rhythm may explain why gasping is more prone to failure than is normal breathing. Children that die of SIDS have normal breathing but gasping is disturbed (Poets et al. 1991), which is significant as gasping is an important arousal mechanism (Thach 2008). Bursting in persistent sodium-dependent pacemaker neurons, as well as gasping, are dependent on aminergic mechanisms (Tryba et al. 2008); these are also disturbed in SIDS (Kinney et al. 2009).
The above discussion illustrates that plasticity and the dynamic regulation of neuronal network activity are universal properties of invertebrates’ and vertebrates’ nervous systems. The dynamic regulation of intrinsic, synaptic, and modulatory properties occurs continuously and disturbance of these processes can be detrimental to the network, to the behavior and ultimately to the organism.
The National Institutes of Health (R01 HL/NS-60120 and P01 HL 090554-01).