Neuromodulator-induced temperature robustness in a motor pattern: a comparative study between two decapod crustaceans Free

Temperature changes are a particular challenge for the nervous system, as they can disrupt the well-balanced weighting of intrinsic and synaptic ionic conductances that are required to maintain neuronal function and behavioural performance (Robertson and Money, 2012; Harding et al., 2019; Marder and Rue, 2021; Robertson et al., 2023). Poikilothermic aquatic animals are especially prone to temperature changes because they do not actively maintain their body temperature. They live in an environment with high heat transfer and are especially susceptible to climate change-induced global warming and the resulting heatwaves (Reid et al., 2009; Somero, 2010, 2012; Mann et al., 2018; Oliver et al., 2018). Nevertheless, many vital behaviours in these animals function across a wide temperature range, making them ideal for studying the mechanisms of temperature resilience. Recent studies have revealed that some neural circuits exhibit intrinsic robustness to acute temperature perturbations arising from cellular and synaptic ionic conductances (Alonso and Marder, 2020). Other neuronal circuits rely less on intrinsic properties but gain robustness through external neuromodulatory influences. In particular, the gastric mill central pattern generator in the stomatogastric nervous system of the Jonah crab Cancer borealis (Stein, 2017) shows very little intrinsic robustness against heating. A change of only a few degrees Celsius can terminate the gastric mill motor pattern. We have previously shown that this termination is due to an increase in membrane leak conductance with temperature in the lateral gastric (LG) neuron, which eliminates spike activity and shunts LG bursting (Städele et al., 2015; DeMaegd and Stein, 2021). As the LG neuron is a member of the gastric mill half-centre circuit that drives the gastric mill rhythm, the rhythm terminates. However, we have also shown that neuropeptide modulation enables thermal robustness and allows normal neuronal functioning at much more elevated temperatures. Specifically, additional substance P-related peptide CabTRP Ia (Cancer borealis tachykinin-related peptide Ia) enables neurons in the gastric mill central pattern generator to produce bursts of action potentials at elevated temperatures in C. borealis. CabTRP Ia is released from a pair of descending projection neurons (the modulatory commissural neuron 1, MCN1), and there are indications that MCN1 activity increases with temperature (Städele et al., 2015; Städele and Stein, 2022). CabTRP Ia rescues neuronal activity by activating a voltage-gated inward current (I_MI) that counterbalances temperature-induced increases in leak currents. As neuromodulator release from descending projection neurons is a common process to control neuronal activity in many circuits and across taxa, we hypothesized that neuropeptide-mediated temperature robustness is not distinctive to C. borealis. This study aimed to test this hypothesis by comparing the effects of temperature and peptide neuromodulation on LG neurons in Dungeness crabs (Cancer magister) and green crabs (Carcinus maenas) with findings from previous studies on C. borealis. Each species inhabits different environments and experiences different temperature fluctuations from the previously studied C. borealis.

Adult Cancer magister Dana 1852 were purchased at a local store in Illinois and kept in filtered, aerated artificial seawater at 11°C until used. Adult Carcinus maenas (Linnaeus 1758) crabs were provided by Dr Markus Frederich (University of New England, USA) and Adam Smith (Perfect Limit Green Crabs, Peabody, MA, USA). Carcinus maenas crabs were kept in filtered, aerated artificial seawater at 20°C until used. Fig. 1A gives an overview of the species used, where they originated, and the temperatures they typically encounter in their habitat.

Established protocols were used. After ice anaesthesia for 20–40 min, the stomatogastric nervous system was isolated according to Gutierrez and Grashow (2009), pinned out in a silicone-lined (Wacker) Petri dish, and continuously superfused with physiological saline (10°C). We worked with fully intact and decentralized stomatogastric nervous system (STNS) preparations. In the latter, the stomatogastric ganglion (STG) was separated from the commissural ganglia (CoGs) by transecting the paired inferior oesophageal nerves (ions) and superior oesophageal nerves (sons) (Fig. 1B).

The animal species used in this study are not subject to ethics approval at Illinois State University. We nevertheless adhered to general animal welfare considerations regarding humane care and use of animals. Water quality, salinity and temperature were monitored daily. Crabs were euthanized by chilling on ice, which is a method recognized as acceptable under the AVMA guidelines for the euthanasia of aquatic invertebrates.

Cancer magister saline was composed of (in mmol l⁻¹): 440 NaCl, 26 MgCl₂, 13 CaCl₂·2H₂O, 11 KCl, 10 Trisma base, 5 maleic acid, pH 7.4–7.6 (Sigma-Aldrich). Carcinus maenas saline was composed of (in mmol l⁻¹): 410 NaCl, 11 KCl, 11 CaCl₂·2H₂O, 21 MgCl₂·6H₂O, 30 NaHCO₃, 11 Trizma base, 5.1 maleic acid, pH 7.4–7.6 (Sigma-Aldrich). CabTRP Ia (GenScript) was added to the saline in some experiments. Solutions were prepared from concentrated stock solutions immediately before the experiment. Stock solutions were stored at −20°C in small quantities. Measurements were taken after 15 min wash in.

Extracellular action potentials were recorded, filtered and amplified with an AM Systems amplifier (model 1700). We recorded from several gastric mill motor nerves, including the lateral gastric nerve (lgn), which contains the LG axon, the dorsal gastric nerve (dgn), which contains the dorsal gastric (DG) axon, the medial ventricular nerve (mvn) and the lateral ventricular nerve (lvn) to monitor pyloric and gastric mill rhythm activity (Fig. 1B). Intracellular recordings were obtained from STG cell bodies using 20–30 MΩ glass microelectrodes (Sutter 1000 puller, 0.6 mol l⁻¹ K₂SO₄+20 mmol l⁻¹ KCl solution, or cytoplasm-matched solution: 20 mmol l⁻¹ NaCl, 15 mmol l⁻¹ Na₂SO₄, 10 mmol l⁻¹ Hepes, 400 mmol l⁻¹ potassium gluconate, 10 mmol l⁻¹ MgCl₂; Hooper et al., 2015). Signals were filtered and amplified through an Axoclamp 900A amplifier (Molecular Devices) in bridge or two-electrode current clamp mode. Files were recorded, saved, and analysed using Spike2 software at 10 kHz (version 7.18; CED, Cambridge, UK) and a Power 1401 (CED). Input resistance was measured using hyperpolarizing current pulses (−2 nA, 2 s duration). Membrane potential voltage deflections were measured in steady state (after 2 s).

To elicit gastric mill rhythms in decentralized nervous system preparations, we extracellularly stimulated the axons of descending projection neurons in the part of the transected ion that remained connected to the STG (marked ‘stim’ in Fig. 1B) at 10°C. Both ions were tonically stimulated for 200 s with the same frequency (Master-8 stimulator, AMPI; 1 ms pulse duration). In C. borealis, the lowest activation threshold of the ion elicits action potentials in the MCN1 projection neuron (Coleman et al., 1995; Bartos and Nusbaum, 1997). Tonic MCN1 stimulation induces a specific version of the gastric mill rhythm that has been thoroughly characterized in previous studies (Coleman et al., 1995; Stein et al., 2007; Hedrich et al., 2011). In the two species studied in this paper, the presence of LG postsynaptic potentials (PSPs) during ion stimulation was used to confirm that the activation threshold was reached. We adjusted the activation threshold separately for each ion by slowly increasing stimulation voltage until PSPs were obtained in the LG neuron. Once the activation threshold was reached, it was maintained at all temperatures. It is unclear whether the axonal projections in the two species studied here are identical to those in C. borealis. However, stimulating at subthreshold amplitudes did not elicit any response of the pyloric or gastric mill neurons, suggesting that, as in C. borealis, threshold stimulation recruited an MCN1-like descending projection that innervates the LG neuron. Stimulation frequency was increased in 1 Hz steps until a gastric mill rhythm was observed at 10°C (i.e. the threshold frequency).

Preparations were continuously superfused with physiological saline. Temperature was manipulated inside a petroleum jelly well around the STG. The well thermally isolated the STG from the rest of the nervous system. The temperature inside and outside the well was controlled independently with two saline superfusion lines, cooled by separate Peltier devices. Temperature was continuously measured within a few millimetres of the STG and the CoGs with separate temperature probes (Voltcraft 300 K). We selectively altered STG temperature while the surrounding nervous system was kept constantly at ∼10°C. Temperature was changed by ∼1°C min⁻¹ unless otherwise mentioned. With intracellular recordings, the temperature was changed by ∼1°C 6–7 min⁻¹ to avoid losing the recording. Measurements were taken in steady state, i.e. after keeping the temperature constant for at least 2 min.

Figures were prepared with CorelDraw X7 (Corel Cooperation), Excel 365 (Microsoft) and SigmaPlot (version 15, Jandel Scientific). Analyses were carried out using Spike2. Unless stated otherwise, data are presented as means±s.d. Alternatively, individual data points for each animal are given. Significant differences are stated as *P<0.05, **P<0.01, ***P<0.001.

We tested the influence of temperature on the gastric mill rhythm by controlling the temperature of the saline superfusion to the STG. At the same time, upstream areas, including the paired CoGs, were kept at a constant temperature (∼10°C) (Fig. 1B). The gastric mill rhythm is a biphasic, episodic pattern, and in C. borealis is known to be driven by the half-centre oscillations of Interneuron 1 (Int1) and the LG neuron (Fig. 1C) (Stein et al., 2007). LG and Int1's cell bodies are in the STG, but rhythmic gastric mill activity requires modulatory input from descending projection neurons in the upstream CoGs (Fig. 1B,C). These projection neurons can be spontaneously active but are strongly activated by a diverse set of sensory pathways (Nusbaum and Beenhakker, 2002; Beenhakker and Nusbaum, 2004; Blitz et al., 2004, 2008; Hedrich et al., 2009; Diehl et al., 2013; Nusbaum, 2013; Stein, 2017). In C. borealis, they can also be selectively stimulated to elicit specific versions of the gastric mill rhythm (Coleman et al., 1995).

Our previous data from C. borealis demonstrated that LG, without which the gastric mill rhythm cannot function, stops producing rhythmic bursts when the STG temperature is raised by only a few degrees Celsius (Städele et al., 2015; DeMaegd and Stein, 2021; Städele and Stein, 2022). In a first set of experiments in C. magister, we tested the effects of temperature on spontaneously occurring gastric mill rhythms. As previously reported in C. borealis (Städele et al., 2015; DeMaegd and Stein, 2021; Städele and Stein, 2022), these rhythms stopped when the STG was selectively heated (Fig. 2A).

Spontaneous gastric mill rhythms can vary in activity patterns because multiple descending projection neurons may contribute to their initiation. To investigate the temperature effects on the rhythm better, we transected the axons of all projection neurons (see Materials and Methods) and stimulated both ions. In C. borealis, this elicits a distinct and well-described version of the rhythm (Bartos et al., 1999; Hedrich et al., 2011) that is driven by the paracrine release of CabTRP Ia from MCN1 onto the LG neuron (Stein et al., 2007), in addition to providing excitatory synaptic input via an electrical synapse (Stein et al., 2022). We recorded intracellularly from LG in these experiments. As in C. borealis, transecting the axons of the projection neurons terminated ongoing spontaneous gastric mill rhythms, and stimulating the ion consistently elicited rhythmic activity in the LG neuron when the STG was kept at 10°C (Fig. 2B). The elicited rhythm strongly resembled the MCN1-type gastric mill rhythm of C. borealis, as evidenced by the burst patterns and subthreshold membrane potential oscillations (Coleman et al., 1995). Increasing the temperature reversibly terminated rhythmic bursting in the LG neuron (Fig. 2B). This was consistently the case at 14°C.

In C. maenas, no spontaneous gastric mill rhythms were observed (N=19). However, stimulation of the ion at 10°C elicited a gastric mill rhythm (Fig. 2C), albeit less reliably than in C. borealis and C. maenas. In 8 of 12 experiments, no gastric mill rhythm could be elicited. Failure to elicit gastric mill rhythms occasionally also occurred in C. magister, but only in less than 10% of preparations. The inability to elicit a gastric mill rhythm in C. maenas was not due to a failure of the stimulation. Clear PSPs from each stimulus pulse were seen in the LG neuron, even when no rhythm was elicited, indicating that the MCN1-like projection neuron was being activated. In addition, our stimulation had a clear excitatory effect on the pyloric rhythm, further suggesting that the stimulation was successful. One of the reasons for these failures to activate the gastric mill rhythm could have been that we were stimulating at the wrong temperature. We thus changed the STG baseline temperature to various temperatures between 10 and 20°C in preparations that did not show a gastric mill rhythm at 10°C. However, no gastric mill rhythm could be started in these preparations in the tested temperature range. Finally, increasing the stimulation frequency did not establish a gastric mill rhythm either.

In those preparations where a gastric mill rhythm could be elicited, the rhythm was stable and showed similar activity patterns to those in C. magister (Fig. 2C) and as previously reported in C. borealis (Coleman et al., 1995). Increasing the STG temperature by 4°C ended burst generation in the LG neuron and stopped or disrupted the rhythm (N=4, Fig. 2C).

To investigate the cellular effects of heating the STG on the LG neuron, we recorded intracellularly from the LG soma and measured the changes in resting membrane potential, synaptic input and action potential amplitude. In C. magister, the resting membrane potential hyperpolarized when the STG temperature was increased (Fig. 3A). Concurrently, action potential amplitude diminished with higher temperature. Similarly, the amplitude of the PSP elicited by each stimulus pulse, i.e. by each MCN1-like action potential, diminished.

We found similar results in C. maenas. As the temperature was increased, the LG neuron resting membrane potential hyperpolarized (Fig. 3B). Concurrently, the amplitudes of its action potentials and the PSPs of the MCN1-like projection neuron diminished.

We have previously shown that in C. borealis, MCN1 activity increases with higher temperature, and that increasing MCN1 firing frequency can rescue the gastric mill rhythm at high temperatures. We have also shown that high MCN1 firing frequencies correlate with ongoing gastric mill rhythm at high temperatures (Städele et al., 2015; DeMaegd and Stein, 2021; Städele and Stein, 2022). In addition, when MCN1 activity is maintained at its cold activity level with a slow spike rate, supplementing the released neuropeptide with bath-applied CabTRP Ia selectively to the STG also rescue the gastric mill rhythm at high temperatures. Fig. 3C shows an original recording of the C. magister LG neuron, with rhythmic bursting at 10°C and after the rhythm crashed at 14°C. Increasing ion stimulation frequency restored LG bursting and rescued the rhythm (N=7 of 7 tested animals). Similarly, when CabTRP Ia (0.1 µmol l⁻¹) was bath applied at 14°C instead of increasing ion stimulation frequency, the rhythm was also restored (Fig. 3D, N=7 of 7 tested animals). This suggests that the activity of descending projection neurons (and the release of neuropeptides from them) is also critical for temperature resilience of the gastric mill rhythm in this species. We noted that in no case was the application of CabTRP Ia alone sufficient to elicit a rhythm. Maintaining ion stimulation was always necessary.

In C. maenas preparations where gastric mill rhythms could be elicited, higher ion stimulation frequencies also restored a crashed rhythm at high temperatures (Fig. 4A; N=3). Unfortunately, because of the small number of successful preparations, we could not test the effects of CabTRP Ia on the gastric mill rhythm in this species. However, we tested whether higher ion stimulation frequencies or CabTRP Ia application could elicit gastric mill rhythms in preparations where ion stimulation did not elicit a rhythm in the first place. This was not the case (N=8 for stimulation, N=3 for CabTRP Ia).

To understand the cellular cause of the gastric mill rhythm crash at high temperature, we turned back to the cellular properties of the LG neuron. The LG neuron cell body and neuropil processes do not support regenerative action potentials (Raper, 1979). The diminishment of LG action potential and PSP amplitude thus suggested that LG input resistance diminished with higher temperature. To test this, we measured LG input resistance in two-electrode current clamp. In both species, input resistance was lower at higher temperature (Fig. 4B). Because previous data from C. borealis had shown that CabTRP Ia modulation increases the neuronal length constant by counterbalancing low input resistance (DeMaegd and Stein, 2021), we measured the effect of CabTRP Ia on input resistance in a subset of the experiments (Fig. 4B, orange). CabTRP Ia (0.1 µmol l⁻¹, selectively applied to the STG) restored input resistance at 14°C to values indistinguishable from saline values at 10°C in both crab species.

We tested whether the observed changes in input resistance are sufficient to explain the crash of the rhythm at high temperature. For these experiments, we reduced input resistance at cold temperature using dynamic clamp and measured whether this reduction stopped the rhythm. Conversely, we increased input resistance at high temperature to test whether this increase was sufficient to maintain the rhythm at that temperature. These recordings were carried out in two-electrode dynamic clamp, with one electrode measuring voltage and the other injecting current. Changes in input resistance were matched to those measured during temperature changes in the same experiment (see Materials and Methods).

Fig. 4C (top) shows an original recording of the gastric mill rhythm in C. magister in four conditions: at 10°C, at 10°C with dynamic clamp-reduced input resistance, at 14°C with no dynamic clamp manipulation, and at 14°C with dynamic clamp-increased input resistance. At 10°C, a regular gastric mill rhythm was observed. Increasing the temperature to 14°C terminated the rhythm. The decrease in input resistance observed during this temperature change was sufficient to terminate the gastric mill rhythm when applied through dynamic clamp at 10°C. Conversely, when input resistance was increased by the same amount at 14°C, the rhythm was restored. Fig. 4D provides the quantitative data for four successful experiments. However, in contrast to C. borealis, changing input resistance by the amount suggested by the temperature increase was not always successful. In four other experiments (not shown), the rhythm at 10°C was not terminated by the change in input resistance. Conversely, the rhythm could also not be restored at 14°C. Instead, LG continued to produce bursts, albeit often with fewer action potentials and at lower frequencies. However, we noted that when we introduced larger changes in input resistance, the rhythm at 10°C could be terminated in all preparations, and restored at 14°C. Consequently, in these preparations, by increasing the temperature we were able to terminate the rhythm, but increasing the input resistance by the temperature-induced amount did not rescue the rhythm. This could suggest that additional cellular or synaptic mechanisms contributed to the heat-induced termination of the rhythm.

The main finding of our study is that, as in the previously tested C. borealis, the intrinsic permissive (‘stable’) temperature range of the gastric mill rhythm in C. magister and C. maenas was small (3–4°C). This was the case even though these species experience larger temperature fluctuations in their habitat (Fig. 1A), and their pyloric rhythm was maintained during all temperature challenges we applied. In C. maenas, for example, the gastric mill rhythm crashed more than 20°C below the pyloric rhythm (Stein et al., 2023). This appears to suggest that the animals cannot chew food at elevated temperatures, given that the gastric mill rhythm no longer functions. However, this does not seem to be the case. Adult C. maenas, for example, survive and presumably feed and digest well in habitats with temperatures well beyond the 14°C at which the gastric mill rhythm crashed at threshold stimulation in our experiments. Their primary habitat is the intertidal zone (Crothers, 1968), where they are exposed to the most extreme fluctuations in ocean temperature. Although they originate from the temperate waters of Europe (Klassen and Locke, 2007), they recently experienced a worldwide expansion and survive average temperatures ranging from 0°C to more than 35°C (reviewed by Young and Elliot, 2020).

The daily temperature variations they experience easily exceed the 4°C increase that was sufficient to crash the gastric mill rhythm. Accordingly, the temperature limits for feeding are estimated to be 5°C and 25°C (McGaw and Whiteley, 2012; McGaw and Curtis, 2013), with acclimation and selection pressures likely to play an essential role in achieving this broad range of feeding temperatures (Cuculescu et al., 1998). We also regularly observed feeding in our tanks at 20°C, even though the gastric mill rhythm failed at 14°C, and our dissections suggest that ingested food had been processed. Similar arguments can be made for C. magister and C. borealis (the original species in which the thermal protection of CabTRP Ia was found; Städele et al., 2015), although the temperature fluctuations these species experience are probably smaller than in C. maenas. They are found in benthic, subtidal and intertidal zones and show strong acclimation and thermotactic behaviours (Prentice and Schneider, 1979; Lewis and Ayers, 2014).

In C. borealis, in vivo recordings have shown that gastric mill rhythms can occur at temperatures higher than the crash temperature during threshold ion stimulation (Städele et al., 2015). Our study shows a possible solution for the apparent contradiction between small intrinsic and considerable behavioural temperature robustness: the gastric mill rhythm was maintained at more elevated temperatures when extrinsic modulatory activity was increased, or additional neuropeptide was bath applied. This suggests that temperature robustness in the gastric mill circuit is highly dependent on extrinsic modulatory input in these species. However, temperature robustness also requires increased neuromodulator release when temperature rises. There is evidence for such an effect in C. borealis: MCN1 is the descending projection neuron that releases CabTRP Ia (Christie et al., 1997), and its firing frequency has been shown to increase with temperature in at least two conditions: spontaneous activity (Städele et al., 2015) and during sensory-induced gastric mill rhythms (Städele and Stein, 2022). Although the definitive test to determine whether higher temperatures result in increased CabTRP Ia release is still pending, these studies strongly suggest this.

Model predictions have suggested potential mechanisms that explain the difference in temperature robustness between pyloric and gastric mill rhythms. The pyloric central pattern generator relies upon the activity of a continuously active pacemaker and strong inhibitory feedback within the pyloric circuit (reviewed by Stein, 2017). Activity in pyloric follower neurons is facilitated by a rebound from inhibition, and inhibitory feedback from follower neurons to the pacemaker contributes to pyloric cycle frequency control (Thirumalai et al., 2006). Weakened inhibition makes it susceptible to failure (Selverston et al., 2000). The pyloric circuit thus shows features of a so-called escape circuit. Modelling studies predict that escape circuits possess a higher intrinsic robustness to temperature perturbations and rely less on extrinsic modulation (Morozova et al., 2022). In contrast, in the gastric mill circuit, LG cannot escape the Int1 inhibition but is instead released (disinhibited) when Int1 is inhibited by the pyloric pacemakers (Marder et al., 1998; Bartos et al., 1999). Release circuits are predicted to be more intrinsically sensitive to rising temperatures, but also to continue to function at these high temperatures when extrinsic neuromodulation is available. In that modelling study, the ionic current evoked by CabTRP Ia (I_MI) was tested, and its previously demonstrated effect of opposing leak currents and amplifying oscillations in feedback circuits (either synaptic or cell intrinsic) was shown to be a likely candidate for enabling temperature robustness (Morozova et al., 2022).

A consistent effect of elevated temperature in the two species we tested was the hyperpolarization of the resting membrane potential and a decrease in input resistance. The latter led to a shunt of PSPs and action potentials, as evidenced by their diminished amplitude at elevated temperatures. Temperature effects on input resistance are well described in other systems as well. For example, the input resistance in goldfish Mauthner cells is consistently smaller after acclimation to elevated temperatures (Szabo et al., 2008). Changes in input resistance are known to play an essential role in network oscillations (Cymbalyuk et al., 2002; Blethyn et al., 2006; Koizumi et al., 2008; Zhao et al., 2010), regulation of excitability (Rekling et al., 2000; Brickley et al., 2007) and switches in activity states of neurons (Gramoll et al., 1994). Our results suggest that in all tested crab species so far, including C. borealis (DeMaegd and Stein, 2021), input resistance contributes to the termination of the rhythm at elevated temperatures. However, our dynamic clamp experiments also suggest that additional factors may add to the termination of the rhythm in C. magister because changing the input resistance was insufficient to recreate the temperature effects in all experiments. Additional factors may include effects on synaptic properties, voltage-gated ionic currents, and other neurons that are part of the gastric mill central pattern generator. The detailed connectivity of the gastric mill circuit in C. maenas and C. magister is unknown. However, in both species, LG neuron membrane potential trajectory during the gastric mill rhythm and its synaptic inputs look very similar to the well-characterized gastric mill circuit of C. borealis (Coleman et al., 1995; Stein et al., 2007). Yet, we observed a few differences. Most notably, in C. magister, there were slow periodic subthreshold hyperpolarizations during failed rhythms (Fig. 3C) that were absent in C. maenas (Fig. 4A) and C. borealis (Coleman et al., 1995; Stein et al., 2007).

The identity of the descending projection neuron that activates the LG neuron during ion stimulation has also not been confirmed in C. maenas and C. magister. It is known that synaptic connectivity between homologous neurons can vary between species, as can the identity and actions of peptide co-transmitters (summarized in Stein et al., 2016). Surprisingly, however, increasing the descending projection neuron's activity through ion stimulation restored the rhythm in all preparations. This indicates that whatever differences exist between species and contribute to the LG neuron temperature crash can be counterbalanced by descending modulation. Strikingly, the application of MCN1's peptide co-transmitter CabTRP Ia was able to rescue the C. magister gastric mill rhythm at elevated temperature, and we suspect that in vivo, the additional peptide is provided by higher projection neuron activity. A remote possibility is that temperature effects on the ionic current elicited by CabTRP Ia (I_MI) could also contribute to temperature robustness. Its conductance may increase with temperature, boosting temperature robustness by counterbalancing negative leak conductance (Zhao et al., 2010), thus restoring gastric mill oscillations. However, how much I_MI changes with temperature remains to be determined. Surprisingly, our dynamic clamp experiments succeeded in only half of the experiments in C. magister. In the other half, the injected conductance was insufficient to mimic the temperature effects, and the gastric mill rhythm continued, albeit with weaker bursts. This may reflect space-clamp problems in our approach. The LG neuron in C. magister is a very large cell. We estimate its cell body to be >100 µm, with a primary neurite of at least the same length. This could interfere with the ability of our (somatic) clamp to reach the distant dendritic processes of the cell where synaptic input arrives and the spike initiation site where action potentials are generated. This may be particularly the case at elevated temperatures because input resistance drops (Fig. 4B), suggesting that the length constant of the neuron shortens. This effect was present in LG of C. borealis (DeMaegd and Stein, 2021).

Counterbalancing detrimental temperature effects on the nervous system is critical for maintaining neuronal function and is often essential for survival. This is particularly true when circuits that drive vital body functions are affected. Central pattern-generating circuits are ubiquitously found in the nervous systems of all animals. It is striking how much even small temperature changes can alter the output of these vitally essential circuits. Hyperthermia during fever or heat stroke can, for example, cause failure of the respiratory pattern generator and induce apnoea (Tryba and Ramirez, 2004). Generally, it is assumed that the permissible temperature range a neuronal circuit displays is related to the temperature fluctuation it experiences (Stein and Harzsch, 2021). For homeotherms, including mammals, the permissive range is predicted to be small, given the relatively small variations in body temperature that occur daily or monthly. In contrast, the permissive range is expected to be much more significant for poikilotherms, particularly those that live in habitats with large temperature fluctuations. Recent studies have provided some evidence for this, showing that the pyloric rhythm of intertidal species possesses a more extensive permissive range than in deeper-ocean species (Soofi et al., 2014; Stein et al., 2023). This suggests that neuronal mechanisms to counterbalance temperature effects are more sophisticated in animals that experience large temperature fluctuations in their environment. It also indicates that neuronal activity in homeotherms may be more susceptible to temperature changes. This becomes particularly important when neuronal circuits are studied at temperatures that are very different from the actual internal temperature of the animal, as is common practice in many fields. For example, the pyloric and gastric mill central pattern generators have been studied for over six decades, and they have provided much insight into the functioning principles of neuronal pattern generation and its modulation. However, most studies were conducted at an agreed-upon constant temperature that did not account for the temperatures the nervous system is exposed to in vivo. As a consequence, the finding that the intrinsic temperature robustness of the pyloric rhythm is large (Tang et al., 2010) and that of the gastric mill rhythm is small (Städele et al., 2015) was initially controversial. It seemed contradictory that two parts of the same neuronal system (the STNS) would respond differently to temperature changes. Our findings resolve this contradiction: the lack of intrinsic robustness of the gastric mill rhythm is rectified by the effects of extrinsic modulation, which largely enhances temperature robustness. The pyloric and gastric mill central pattern generators thus serve as an extreme example of how fundamentally different neuronal circuits can respond to temperature changes. They are also a reminder that dramatic shifts in circuit activity may occur when circuits are exposed to changing temperatures and modulatory conditions.

We would like to thank Dr Markus Frederich (University of New England, ME, USA) and Adam Smith (Perfect Limit Green Crabs, Peabody, MA, USA) for the Carcinus maenas used in this study. We would also like to thank Dr Margaret DeMaegd (New York University) for providing some of her input resistance data.