Homeostatic plasticity of excitability in crustacean central pattern generator networks

Abstract

Plasticity of excitability can come in two general forms: changes in excitability that alter neuronal output (e.g. long-term potentiation of intrinsic excitability) or excitability changes that stabilize neuronal output (homeostatic plasticity). Here we discuss the latter form of plasticity in the context of the crustacean stomatogastric nervous system, and a second central pattern generator circuit, the cardiac ganglion. We discuss this plasticity at three levels: rapid homeostatic changes in membrane conductance, longer-term effects of neuromodulation on excitability, and the impacts of activity-dependent feedback on steady-state channel mRNA levels. We then conclude with thoughts on the implications of plasticity of excitability for variability of conductance levels across populations of motor neurons.

Figure 1. Activity-dependent plasticity of intrinsic excitability in cardiac ganglion motor neurons of the crab, Cancer borealis

A. Effects on excitability of LC motor activity after exposure to 4AP, which blocks the A-type current in these cells. Pacemaker neurons of the network were silenced with TTX in this experiment, and activity elicited with DC current injections. 4AP shifts these cells from tonic spiking to a state of hyperexcitation, a burst potential. Within 60 minutes, intrinsic compensation occurs that restores excitability towards baseline concomitant with an increase in High-Threshold K⁺ (HTK) currents made up of I_BKKCa and delayed rectifier currents. Representative recordings of the increase in I_A after TEA exposure are shown for control (dark green) and one hour after 4AP exposure (light green) [18]. B. A working model for homeostatic plasticity of intrinsic excitability via compensation between I_A and I_BKKCa [18]. Baseline excitability (1) is disrupted by the block of I_A with 4AP, causing an alteration in the excitability of the cell and its output in the network, in this case rapid initial firing frequency (2). This change results in the release of calcium from intracellular stores as well as influx through voltage-dependent calcium channels (3) that increases activity of the calcium-dependent phosphatase calcineurin (4). Calcineurin activity induces an increase in I_KCa (5), restoring the excitability of the cell and re-establishing motor neuron activity in the intact network (6).

Figure 2. A working model for how tonic nM DA acts to enable activity-dependent regulation of I_H in LP neurons of the STG of the lobster, Panulirus interruptus

The graph represents I_H activity dependence in 5nM DA, with the line indicating the idealized fold change observed in response changes in LP duty cycle. The inset shows a working model for a mechanism regulated by tonic dopamine exposure. Dopamine receptor binding activates protein kinase A (PKA) which phosphorylates an unknown target (in blue) resulting in an increase in G_H. PKA also targets calcineurin, a calcium-dependent phosphatase. Calcineurin activation then works through an unknown substrate to reduce G_H in LP cells. The phosphorylation state of the unknown protein is superimposed upon the G_H activity-dependence curve in 5nM dopamine. Intracellular calcium concentration presumably is lowest at −100 and highest at +50% change in the LP duty cycle, and the gradient below the graph tracks subsequent calcineurin activation. PKA activity is constant because dopamine is constitutively present (red bar).

Figure 3. Activity-dependent feedback regulates correlated channel mRNA levels in pacemaker cells: modeling and experimental evidence

A. A simple biochemical scheme for activity-dependent channel expression. Channel mRNAs are produced at a rate α_m that depends on a Ca²⁺-dependent factor, T, and degraded at rate β_m. Functional channel proteins are produced at a rate α_g from mRNAs and degraded at a rate β_g. This scheme is equivalent to an integral controller. Deviation from the target calcium concentration ([Ca²⁺]_tgt) is accumulated in the mRNA (m) concentration (shaded region), which causes a change in ion channel expression (g). B. Membrane potential and conductance density evolution of a self-regulating neuron implementing integral control for its seven voltage-dependent conductances (fast sodium, g_Na; slow Ca²⁺, g_CaS; transient Ca²⁺, g_CaT; A-type/transient potassium, g_KA; Ca²⁺-dependent potassium, g_KCa; delayed-rectifier potassium, g_Kd; hyperpolarization-activated mixed-cation, g_H). A total of 20 independent runs are shown with mean in bold. Voltage traces for an example neuron at the stages indicated are shown above the plot. C. Scatter plots of conductance distributions at steady state of the 20 neurons from the runs shown in panel B. D. Scatterplots for correlated mRNA levels for six ion channel mRNA pairs (CbCaV1, g_CaS; CbCaV2, g_CaT; CbCaV3, g_Ca-LVA (low voltage activated); shal, g_A; shab, g_Kd; Cb-IH, g_H) from PD neurons from STGs of N = 13 crabs (Cancer borealis). The effects of activity on these correlated mRNAs for a representative pair (CbCaV1-shal) is denoted with a dashed red box in the control PD neuron matrix, and the corresponding model conductances also noted with a dashed red box in panel C. These four experimental groups effectively decouple modulation from activity, as described in the text. In the two conditions where activity is preserved (Control and Decentralized+Pilocarpine), correlations are maintained. When activity is limited (Decentralized), even in the presence of a modulator agonist (Pilocarpine+TTX), the correlation is not maintained. These data indicate that activity-dependent feedback is at least partly responsible for correlated channel transcript levels in PD neurons.

Figure 4. A schematic of how variability could arise in STG neurons

Following neurogenesis and differentiation, developmental tuning processes result in conductance profiles in individual neurons that give rise to appropriate network output. These developmental tuning processes can result in conserved patterns of conductance expression, in which across individuals the networks converge on a common conductance profile, leading to minimal variability in the population. Alternatively, developmental tuning creates conductance variability, whereby different networks in different individuals achieve convergent output through unique tuning pathways – leading to variability in the population. Once appropriate output is achieved, then over the lifetime of the animal homeostatic compensation maintains the output. Because every individual experiences distinct environmental conditions, variability arises in the population as a result of homeostatic feedback. If after development these networks had very similar solutions, then this mature feedback compensation is the origin of variability in the population through generation of conductance variability. However, if developmental tuning had already resulted in variable solutions to network output, then variability measured in adult populations is either the result of maintenance of tuned conductances, or continued evolution of conductance variability due to ongoing homeostatic compensation.