Persistent modulatory actions and task switching in the feeding network of Aplysia

Abstract

The activity of multifunctional networks is configured by neuromodulators that exert persistent effects. This raises a question, does this impact the ability of a network to switch from one type of activity to another? We review studies that have addressed this question in the Aplysia feeding circuit. Task switching in this system occurs ‘asymmetrically’. When there is a switch from egestion to ingestion neuromodulation impedes switching (creates a ‘negative bias’). When there is a switch from ingestion to egestion the biasing is ‘positive’. Ingestion promotes subsequent egestion. We contrast mechanisms responsible for the two types of biasing and show that the observed asymmetry is a consequence of the fact that there is more than one set of egestive circuit parameters.

Introduction

Like many neural networks the feeding network in Aplysia is multi-functional. It generates both ingestive and egestive motor programs. To a large extent this is a consequence of the fact that activity is configured by modulatory neurotransmitters (for review see [1] and for very recent characterizations of feeding peptides see [2,3]). This is not an unusual arrangement; in fact, all networks appear to be subject to neuromodulation [4]. Although this has been most extensively documented in motor networks, it is becoming apparent that neuromodulation plays a role in configuring network activity in many other contexts such as human cognition [5].

The induction of a persistent state in the feeding network is an advantage under some circumstances. For example, when Aplysia encounter patches of seaweed they generate bouts of ingestive reponses [6]. Additionally, however they are capable of task switching. For example, if they ingest a piece of food that is attached to a substrate the attached food can be egested and another piece of seaweed ingested [7].

Task switching is a common phenomenon and even occurs during the execution of innate behaviors (e.g., [8,9]). Mechanistic studies of how task switching occurs have primarily focused on circuit level questions, e.g. how they are triggered by changes in afferent input [10–13]. A question less commonly addressed is, can the persistence of an internal state impact the ability of a network to task switch quickly? Studies of the feeding circuit of Aplysia that speak to this issue are described below.

The feeding circuit is multi-functional

As Aplysia animals feed, they adapt to changes in the external environment and modify behavior [14,15]. For example, one type of ingestive response (a bite) can be converted to a second type of ingestive response (a swallow) if an animal successfully grasps food. These types of adjustments in ongoing behavior are often mediated by dynamic changes in afferent input.

This review will focus on a further complication; the fact that the network operates in at least two different modes [1]. Namely, it can generate a bout of ingestive activity or a bout of egestive activity. How this is achieved has been the subject of much investigation and it has become apparent that these two network configurations are to a large extent induced via actions of modulatory neurotransmitters [1].

State changes in the feeding network

Some neuromodulator induced alterations of the Aplysia feeding circuit are long-term and are important for learning and memory (recent description of this include [16–23]). This review will focus on modifications that are not as long lasting, but persist for minutes. On this time scale persistence produces a change in the internal state of the animal in that responses to external stimuli are impacted until modulatory effects dissipate [24]. The state change creates a ‘bias’ in that there is an increase in the probability that a certain type of response will be generated. In another mollusc, Lymnaea, biasing can even cause satiated animals to reject stimuli that are palatable to food deprived animals [25]**. When state changes are induced, afferent input can be integrated over a relatively long time scale and brief ‘distractions’ can be ignored (also see [26]**).

Internal states are not always a product of neuromodulator release. For example, in other contexts recurrent connections can be important [27,28]*. Comparisons of state induction across species have however noted that in most cases persistent neuromodulation does play an important role [29]. An issue we address in this review is how does this impact the ability of the network to task switch?

Asymmetries in task switching in the feeding network

Even in mammals it has become apparent that feeding as a whole can be surprisingly complex. It can be fragmented, and include behaviors that are not strictly related to food intake [30]**. Studies of switches between ingestive and egestive activity in the isolated feeding network of Aplysia have shown that there is a pronounced asymmetry. After a bout of egestive activity it is not immediately possible to generate ingestive activity, i.e., negative biasing is observed [31] (Fig. 1A). In contrast, after a bout of ingestive activity, the biasing is positive [32] (Fig. 1B). A stimulus that would trigger an intermediate response in a naïve preparation triggers a fully egestive response. Asymmetries in task switching have also been described in other systems (e.g., [33,34]). A question addressed in the feeding network is, how do they arise?

Figure 1.

(A) Negative biasing. If an ingestive input to the CPG is activated under control conditions a weakly ingestive (or intermediate) cycle of activity is generated (light green) [31]. In contrast, if the same input is used to trigger a cycle of activity after egestion the cycle triggered is egestive (dark red) [31]. Negative biasing is a consequence of the fact that the induction of egestive activity produces a persistent increase in the excitability of the egestive interneuron B20 [36]. Consequently, B20 is ‘overactive’ when there is an attempt to trigger an ingestive cycle of activity. (B) Positive biasing. If an egestive input to the CPG is activated under control conditions a weakly egestive (or intermediate) cycle of activity is generated (light red) [31]. In contrast, if the same input is used to trigger activity after a bout of ingestive activity the cycle triggered is egestive (dark red) [32]. Positive biasing is a consequence of the fact that the induction of ingestive activity produces a persistent increase in the excitability of the egestive interneuron B65 [32]. B65 is, however, actively inhibited so it does not fire during ingestion. When a cycle of activity is subsequently triggered by stimulating the egestive input to the CPG B65 is excited and the latent effect on its excitability becomes apparent. B65 fires and cycles of activity are egestive (dark red) [32].

Negative biasing

Negative biasing is perhaps what would be expected when a task switch occurs after a persistent state has been created. In the feeding network it is a consequence of the fact that modulators are released during bouts of egestive activity [1] which produce persistent increases in the excitability and synaptic outputs of an egestive interneuron (the cell B20) [35].

These persistent modifications make B20 overactive when there is a subsequent attempt to generate ingestive motor programs [36] (Fig. 1A).

A recent study demonstrated that there is an additional ‘anticipatory’ circuit modification that also contributes to negative biasing [37]*. During bouts of egestive activity an outward current is induced in a motor neuron that is required for all types of feeding motor programs (the neuron B8). B8 is active during egestion, but only because it receives excitatory synaptic input. Thus, induction of the outward current is counterproductive during egestion. Because the current persists, the B8 excitability is still low when there is a subsequent attempt to task switch and B8 fires at a reduced frequency. This is now a ‘desirable’ effect since it contributes to negative biasing by making motor programs less ingestive. Although apparently counterproductive circuit modifications have not been widely reported, a recent computational study of the feeding network suggests that they may be more common than is currently recognized [38]*.

In summary, studies of egestive-ingestive task switching in Aplysia have characterized mechanisms that can impede task switching and lead to negative biasing. These mechanisms operate at both the motor neuron and interneuron level. At the interneuron level, negative biasing occurs when there is a persistent increase in the excitability of a neuron that promotes egestive activity. This neuron is then overactive when there is a subsequent attempt to task switch.

Positive biasing

In the other direction the biasing that is observed is positive, i.e., ingestive activity promotes the subsequent induction of at least a short bout of egestive activity [32]. To a large extent, positive biasing results from the fact that B20 is not the only egestive interneuron. A second interneuron, B65, can serve a similar function[39,40]. When ingestive motor programs are repeatedly induced, there is a persistent peptide-induced increase in the B65 excitability [32] (Fig. 1B). Since B65 is an egestive interneuron, it might be expected that this would disrupt ongoing ingestive activity. This does not happen because B65 is actively inhibited. As a result, the excitability increase has no immediate effect. However, when there is a subsequent switch to egestive activity B65 is excited rather than inhibited and the latent effect on excitability becomes apparent [32]. B65 fires at an elevated frequency and motor programs are highly egestive (Fig. 1B).

In essence, the asymmetry in task switching is a consequence of what has been referred to as network degeneracy, i.e., there is more than one set of circuit parameters that will produce a certain type of motor activity [41,42]. In particular, the egestive interneuron that patterns activity during positive biasing (B65) is not the same as the interneuron that patterns activity during negative biasing (B20) [43]. B65 differs from B20 in that during ingestive behavior it receives excitatory and inhibitory input in parallel. Previous descriptions of this type of arrangement have established that it can be important for generating sequences of behavior. For example, it is an important part of the suppression hierarchy that drives grooming in Drosophila [44–46]. Research in the feeding system extends these findings in that it shows that in addition to playing a role in ordering behavior, latent excitation can also play a role in determining parametric features of the behaviors in a sequence.

Positive biasing presumably occurs when animals make temporary alterations in feeding movements [7]. They would be analogous to the adjustments that vertebrates make when they alter locomotion to compensate for obstacles in the environment [47]. Although switching involves a cessation of one behavior and initiation of another, it differs from ‘quitting’ in that it is likely to be followed by a return to ingestion (rather than a period of inactivity). Thus, behavior is maintained despite the switching. A further question that has been addressed in the feeding network is, how does this occur?

Maintenance of feeding behavior during task switching

Data suggest that the ability of the feeding circuit to maintain activity is facilitated by its modular design. In particular, some modules in the feeding network are ‘behavior independent’, and are utilized during all types of motor programs [48]. This is the case for the neurons that move the radula back and forth in the buccal cavity. Radula protraction precedes radula retraction during both ingestive and egestive motor programs [49,50]. Consequently, a modification of the protraction (or retraction) circuitry will similarly impact all types of motor programs.

Because radula protraction occurs first, the processes that determine its induction constitute the ‘decision to feed’. These processes have been the subject of much investigation [51,52] and it has become apparent that the neuron B63 plays a key role. When ingestive and egestive inputs to the feeding CPG are activated, modulators are released that exert divergent effects that are important for selectively configuring activity (e.g., [35,53–55]). Additionally, however a recent study demonstrated that ingestive and egestive modulators also act convergently [56]*. Both types of substances increase the B63 excitability, a circuit modification that promotes program induction [57,58]. This suggests that as task switching occurs modulatory effects on B63 will be additive. Thus, activity is likely to be maintained by a type of feedforward excitation that does not depend on the nature of the behavior induced.

Retention of the ingestive state during task switching

A final question that we address here is, can a state be maintained during a brief switch to a different type of motor activity? Experiments that explored this issue in the feeding circuit focused on an experimentally advantageous motor neuron (the neuron B48). B48 is selectively active during ingestion, and there are persistent increases in its excitability when an ingestive state is induced [59,60]. Excitability changes persist for ~ 40 minutes (Fig. 2A) [60]. During this time task switching is possible, i.e., activation of an egestive input to the feeding CPG produces an immediate decrease in the B48 excitability and firing frequency (Fig. 2C) [60]. To determine whether the original (ingestive) state is retained during switches to egestive activity, excitability was measured after allowing time for the effects of egestive stimulation to dissipate (Fig. 2B, 2C) [60]. Excitability was still elevated (Fig. 2C). Thus, in B48, the ingestive state can be retained during a bout of egestive activity.

Figure 2.

(A) Ingestive state in neuron B48. Stimulation of an ingestive input to the feeding CPG (green bar) produces an increase in the excitability of B48 that persists for ~ 40 minutes (green plot to the right of the bar) [60]. As shown in the inset the increase is mediated by the induction of an inward sodium current [54,60]. The induction of the current is not PKA dependent and presumably results from direct gating of the channel by cAMP. (B) Egestive state in B48. Stimulation of an egestive input to the feeding CPG (red bar) produces a shorter lasting decrease in B48 excitability (red plot to the right of the bar) [60]. As shown in the inset the decrease is mediated by the induction of an outward potassium current [60]. (C) Retention of the ingestive state during ingestive-egestive task switching. Immediately after an ingestive-egestive switch (green/red bar) excitability is depressed (black plot to the immediate right of the bars) [60]. As shown in the top inset this is presumably a consequence of the induction of both the inward sodium and outward potassium current [60]. After effects of egestive input activation have dissipated excitability is still elevated and is the same as the excitability at a matched time point when is no switch (overlapping black and light green plots), i.e., the ingestive state is retained [60]. As shown in the bottom inset this is a consequence of the persistent induction of the inward sodium current [54,60]. The excitability plots are after Fig. 3 in [60].

Mechanistic experiments have demonstrated that increases in B48 excitability are mediated by the induction of a cAMP-dependent inward current (Fig. 2A, inset) [54,60]. Data strongly suggest that this current is directly gated by cAMP and is similar to a current characterized in other invertebrate neurons [61–66]. Retention of the ingestive state occurs as a result of the persistence of the cAMP signalling. (It does not, however, require PKA induction [54,60] as is commonly the case [9,67,68].)

When there is a switch to egestive activity the B48 firing frequency (and excitability) decrease (Fig. 2C). This occurs because an outward potassium current is induced (along with the inward current) (Fig. 2C, top inset) [54,60]. Thus, there is a brief period of time when there an overall decrease in net current. The outward current is not as persistent as the inward current. Effects of its induction wear off while the inward current is still present. Consequently, there is a return to a situation where the net current is inward (Fig. 2C, bottom inset). Thus, there is an interaction between persistent modulation and relatively short-lasting inhibition. This type of interaction also occurs in Drosophila where it determines the duration of a behavior (copulation) [69]. The feeding system data make an additional point. As long as the brief inhibition does not negatively impact the original modulatory effect it will be possible to return to the original state after a task switch.

In summary, studies of ingestive-egestive task switching have characterized a mechanism that can promote rather than impede task switching. This type of positive biasing occurs because the induction of ingestive activity has a latent effect. In addition to increases in the excitability of interneurons that promote ingestive activity, there is also an increase in the excitability of an egestive interneuron. Under physiological conditions positive biasing presumably occurs during ingestive behavior and allows animals to make brief adjustments that allow them to cope with difficulties that are encountered when they try to pull food into the buccal cavity. In this situation it is presumably beneficial for the original state to be maintained despite the task switching.

Concluding Remarks

Although task switching has been most extensively studied in the context of motor control it has been described in other contexts. For example, it occurs when humans are performing cognitive tasks [5,70]. Often task switching results in some type of biasing, i.e., the execution of the behavior that occurs after the switch is impacted by the behavior that occurs before the switch. Biasing can be either negative or positive. Negative effects often have undesirable consequences and are referred to as the ‘cost’ of task switching. It has been argued that mechanistic studies that seek to determine how costs arise are valuable since they provide insight into strategies that can be used to minimize them (e.g., [70]). Work in experimentally advantageous model systems such as the feeding system of Aplysia are likely to make an important contribution in this regard.

Highlights.

• Neuromodulators induce persistent ingestive and egestive states that impact subsequent task switching

• Negative biasing is observed after an egestive-ingestive switch

• Positive biasing is observed after an ingestive-egestive switch

• The asymmetry in biasing is a consequence of network degeneracy (there is more than one set of egestive circuit parameters)