Ocean acidification affects acid-base physiology and behaviour in a model invertebrate, the California sea hare (Aplysia californica).

Behavioural impairment following exposure to ocean acidification-relevant CO2 levels has been noted in a broad array of taxa. The underlying cause of these disruptions is thought to stem from alterations of ion gradients ( HC O 3 - / C l - ) across neuronal cell membranes that occur as a consequence of maintaining pH homeostasis via the accumulation of HC O 3 - . While behavioural impacts are widely documented, few studies have measured acid-base parameters in species showing behavioural disruptions. In addition, current studies examining mechanisms lack resolution in targeting specific neural pathways corresponding to a given behaviour. With these considerations in mind, acid-base parameters and behaviour were measured in a model organism used for decades as a research model to study learning, the California sea hare (Aplysia californica). Aplysia exposed to elevated CO2 increased haemolymph HC O 3 - , achieving full and partial pH compensation at 1200 and 3000 µatm CO2, respectively. Increased CO2 did not affect self-righting behaviour. In contrast, both levels of elevated CO2 reduced the time of the tail-withdrawal reflex, suggesting a reduction in antipredator response. Overall, these results confirm that Aplysia are promising models to examine mechanisms underlying CO2-induced behavioural disruptions since they regulate HC O 3 - and have behaviours linked to neural networks amenable to electrophysiological testing.

Background

Ocean acidification is occurring at rates not observed in the last 300 million years. Average global oceanic CO2 levels are projected to increase from current levels of approximately 400 to approximately 940 µatm CO2 by the end of the century and approximately 1900 µatm CO2 by the year 2300 unless the rate of CO2 emissions is substantially curtailed [1-3]. This rapid rate of change has made predicting the sensitivity of organisms to future predicted CO2 levels a major focus of climate change research. Early studies focused heavily on calcifying invertebrates, reporting widespread impacts to calcification and growth [4]. Fish exposed to CO2 have exhibited alterations to mitochondrial pathways, intestinal base secretion and otolith growth [5-7].

In addition, impaired behaviour following CO2 exposure has been reported in more than approximately 130 studies to date in marine organisms at ocean acidification-relevant CO2 levels (less than 1900 µatm CO2). The majority of these studies have focused on marine fish, noting impairments to various endpoints including vision, olfaction, lateralization and learning [8-11], reviewed in [12]. Examination of behavioural disturbances has also been extended to invertebrates, where negative effects on predator defence behaviours have been observed [13-16].

The underlying cause of these behavioural disruptions is thought to result from the compensatory mechanism that allows fish and some active invertebrates to maintain pH homeostasis when exposed to elevated CO2. Following the onset of CO2 exposure, animals that are acid–base ‘regulators’ counter an initial drop in blood pH through the retention and/or uptake of . This process allows acid–base regulators to correct pH to pre-exposure levels; however, both and PCO2 remain elevated [12,17,18]. This compensation mechanism is generally related to how ‘active’ an organism is, as higher metabolic rates (O2 consumption) necessitate higher rates of CO2 excretion [19]. The accumulation of in extracellular fluids is usually coupled with an equimolar decrease in Cl− [18,20]. The resulting changes in extracellular and intracellular and Cl− are thought to alter behaviour by attenuating the movement of these ions through the primary receptor (GABAA) responsible for background inhibitory responses in the vertebrate and invertebrate nervous system [11,21,22]. Thus, strong acid–base regulators with the ability to accumulate are hypothesized to be most at risk for behavioural disturbances [11].

Nilsson and colleagues [11] first implicated GABAA receptor involvement in behavioural disruptions by treating CO2-impaired animals with gabazine, a GABAA receptor antagonist. This treatment was found to reverse CO2-induced behavioural changes. Similar subsequent studies have continued to provide evidence for the involvement of GABAA receptors using gabazine or muscimol (GABAA receptor agonist) in fish [8,10,23-28] and in some invertebrates [13,29]. While this methodology has been seminal in providing a parsimonious explanation for altered behaviour and GABAA-receptor involvement in CO2-induced disruptions, future studies would benefit from two important considerations. First, although this proposed mechanism hinges on changes that occur following CO2 compensation, few studies have measured acid–base parameters in a marine species while also measuring behaviour [30-35]. Such measurements would be especially important in invertebrates, where there is more inherent variation in acid–base regulatory ability [19,36,37]. For example, sea urchins (Paracentrotus lividus) retain to defend pH, while mussels (Mytilus edulis) do not, and experience an acidosis when exposed to the same CO2 level (1480 µatm CO2) [38]. Second, although crucial in implicating the GABAA receptor, immersing an animal in seawater containing a GABAA receptor pharmacological agent lacks resolution in targeting specific behaviours and could induce effects on unintended targets [23]. In addition, there has been little exploration of potential alternative or additional mechanisms in CO2-induced behavioural disruptions [39,40]. Finally, although the majority of CO2 behavioural studies are performed on fish, the vertebrate nervous system is complex, making it difficult to link a particular behaviour to specific neural networks.

To address these limitations, we propose that future research examining the behavioural impacts of CO2 would benefit from identifying a model organism well-suited for both acid–base balance and neurophysiological studies. The ideal study organism would meet three criteria: (1) a simple and well-mapped nervous system, (2) reproducible behavioural assays, and (3) an acid–base ‘regulator’ profile, with the ability to accumulate to defend pH. The California sea hare (Aplysia californica), referred to herein as ‘Aplysia’, is widely known to meet the first two criteria perfectly and has been used for decades as a biomedical research model to study the cellular basis of learning [41].

Since the ability to acid–base regulate has been linked to behavioural disruptions, measuring the baseline CO2 acid–base response in Aplysia is a necessary step in assessing their feasibility as a model for CO2 behavioural research. The first objective of the present study was to examine acid–base parameters in haemolymph from Aplysia exposed to either control (approx. 400), 1200 or 3000 µatm CO2. Since Aplysia are not sessile invertebrates, it was hypothesized that they would exhibit an acid–base ‘regulator’ profile, and actively retain to defend pH following CO2 exposure. The second objective of this study was to examine the impacts of elevated CO2 on two simple behaviours with well-characterized neural networks [42], the righting reflex and the tail-withdrawal reflex. Righting is important for orientation and reattachment to substrate, while the tail-withdrawal reflex is an antipredator response that activates muscles used in escape responses [42-44]. Elevated CO2 was expected to alter behaviour, as noted in previous studies. Notably, the chosen behavioural assays and CO2 levels are environmentally relevant for Aplysia living in the intertidal zone of the North American Pacific coast [45]. Ultimately, this study marks the first step in assessing Aplysia as a potential model for future studies of CO2-induced behavioural disruptions in marine organisms, including exploration of the GABAA hypothesis in addition to potential alternative mechanisms.

Material and methods

Animal care and experimental exposure

Aplysia (Aplysia californica), hatchery-reared from egg masses of wild-caught animals, were provided by the National Resource for Aplysia (National Institute of Health Grant P40OD010952) at the University of Miami Rosenstiel School of Marine and Atmospheric Science. Prior to use in experiments, Aplysia were fed ad libitum with red macroalga Gracilaria ferox and Agardhiella subulata [46] and were kept in 16 l tanks with a seawater flow rate of approximately 1.3 l min−1 at approximately 15°C.

During experimentation, Aplysia were acclimated to either control (400), 1200 or 3000 µatm CO2 for acid–base (n = 2–3 tank replicates, 3–5 animals/tank) and behavioural experiments (n = 4–7 tanks, 2–5 animals/tank). These acclimations were performed in 16 l tanks with flow-through seawater (0.6 l min−1, 15°C). Animals were exposed for either 4 or 11 days to each CO2 level. Since day of exposure (4 versus 11) did not significantly impact any measured endpoint (see below), exposure duration is referred to as 4–11 days throughout the manuscript. These time periods have previously been sufficient to reach a stable accumulation for CO2 compensation [47]. In addition, 4 days is close to the 5-day exposure period previously demonstrated to induce behavioural disruptions in other invertebrates [13,14]. Animals were permitted to feed on the first day of the exposure but food was subsequently withheld approximately 96 h prior to experimental testing. Animals that experienced 11-day exposures were subjected to the same approximately 96 h fasting period. Animals remained immersed in seawater throughout the experimental period. Animals used in experiments were approximately 10–11 months of age and weighed 90–110 g (electronic supplementary material, table S1).

Seawater CO2 manipulation

Desired PCO2 levels were achieved using a CO2 negative feedback system as previously described (Loligo Systems, Denmark) [6]. First, a standard curve was made by determining the relationship between known gas standards and seawater pH. Using this relationship, a pH set-point corresponding to each desired PCO2 level was calculated, and 100% CO2 was slowly bubbled into flow-through, aerated tanks to achieve the chosen PCO2 level. The pH electrode and meter (WTW Sentix H electrode and 3310 meter) corresponding to each experimental tank were connected to CapCTRL software that delivered CO2 using solenoid valves controlled by a DAQ-M digital relay instrument (Loligo Systems). Validation of desired PCO2 values was achieved using pHNBS and total CO2 (TCO2) and was performed approximately two times per experiment. Measurements of pHNBS were recorded multiple times per week using an independent pH electrode and meter (Radiometer PHC3005 electrode, ThermoFisher Orion Star A221 meter). A Corning 965 CO2 analyser (Corning Diagnostics) was used to measure TCO2. To calculate PCO2 and titratable alkalinity (TA), values of pHNBS and TCO2, were entered into CO2SYS [48]. Calculated PCO2 values for control, 1200 and 3000 μatm CO2 are reported in electronic supplementary material, table S2. Temperature and salinity were measured approximately three times per week (WTW 3310 meter and TetraCon 325; electronic supplementary material, table S2).

Objective 1: Haemolymph acid–base balance and ion measurements

Extracellular haemolymph was sampled by inserting a 500 µl gas-tight glass syringe (Hamilton) towards the posterior and alongside the foot of the animal and gently withdrawing fluid. Haemolymph was measured immediately for extracellular pH (pHe) using a custom glass chamber fitted around a needle pH microsensor attached to pH-1 Micro meter (Loligo Systems). The pH microsensors were pre-calibrated from the manufacturer and were corrected after verification with a known pHNBS value from sterile seawater. This sterile seawater was used to flush out the pH chamber in between sample measurements and was measured using Radiometer PHC3005 pH electrode attached to a ThermoFisher Orion Star A221 meter. Haemolymph from the same animal was measured for TCO2 (Corning 965, Corning Diagnostics). and PCO2 were calculated from TCO2 and pH using the Henderson–Hasselbach equation using an established solubility constant (αCO2) and dissociation constant (pK) for carbonic acid [49].

Objective 2: General behavioural assay protocols

For all behavioural assays, animals were gently removed from their experimental tank and placed carefully in the bottom of test tanks in water at their respective acclimation PCO2 level. Tanks were 16 l and had a depth of 16 cm. In all assays, animals were given a 5 min acclimation time to become accustomed to the test tank prior to commencing behavioural tests. All assays were recorded on video and the experimenter was blind to the experimental treatment both during experiments and video analyses. In some cases, animals were tested in one of the two behaviour assays on the 4th day of exposure, returned to acclimation tanks, then tested on the 11th day for the second behaviour. The order of behaviours tested was altered. Based on previous studies, even repeated stimuli or noxious stimuli do not elicit long-term memory formation (animals retested on day 7) [50-53]. Accordingly, there was no reason to suspect that the mild stimulus in the present study would impact animals receiving a second behavioural test. In both assays, animals remaining in a contracted state for more than 1 min or animals that inked during tests were eliminated from analyses as ‘non-participators’. In previous studies, extreme stress has been shown to lead to tachycardia and suppressed reflex activity [54], and inking is considered a ‘high-threshold, all or none’ behaviour [55]. This criteria resulted in removal of five control, seven 1200 µatm and eight 3000 µatm animals from the tail-withdrawal assay. One animal was removed from the 1200 µatm and the 3000 µatm treatments during the righting assay.

Righting behavioural assay

Protocols followed those outlined in a previous study [42]. Following the 5 min acclimation period, the animal was gently lifted to the top of the water column and released while on its side. The start time of the reflex occurred the moment the animal made contact with the bottom of the tank. The time from bottom contact to when the animal returned to an upright position and initiated its first crawling was recorded as righting time. The assay was performed in triplicate with a rest period of 5 min between trials [42,43]. The data was summarized for each individual as the mean of the triplicate measurements for the reflex time.

Tail-withdrawal behavioural assay

Protocols followed those outlined in previous studies [42,56]. Following the acclimation period, the animal was carefully lifted off the test tank bottom, and gently held by the experimenter as close to the tank bottom as possible without allowing the animal to adhere to the bottom (approx. 1 cm). At this point, the length from the tip of the tail to the top of the head was measured and recorded as the resting length, using a transparent ruler lying next to the animal in the bottom of the tank. The animal was then placed on the bottom of the tank and a blunted 20G needle was pressed onto the tip of the animal's tail (approx. 50–70° angle) for one second to depress the tissue against the test tank bottom to a depth approximately half the thickness of the tail. This depression caused the tail to withdraw and represented the starting time of the reflex. At maximal contraction, the total length of the animal from the tip of the tail to the top of the head was noted using the ruler. Relaxation of the tail to approximately 50% of the original tail length signified the end of the reflex. The reflex was measured in triplicate with rest intervals of 10 min between each trial [42].

Statistical analysis

Linear mixed effect (LME) models were used to test for responses to CO2 exposure levels for the time to complete the righting reflex and the time to complete the tail-withdrawal reflex. These models included CO2 level and day of exposure as fixed factors, and tank as a random factor. Tukey's post hoc tests with a Holm-adjusted p-value was used to compare means between CO2 exposure levels. The righting time and the tail-withdrawal reflex time data were log-transformed prior to analysis. A general linear model was used in instances where inclusion of tank as a random factor in mixed models resulted in overfit, using treatment and day as fixed factors. This applied to the per cent of the tail withdrawn in the tail-withdrawal reflex, haemolymph pHe, haemolymph and haemolymph PCO2. The per cent of the tail withdrawn in the tail-withdrawal reflex was arcsin transformed prior to statistical analysis. All models were conducted in R v. 3.5.2 [57] using the lme4 and lmerTest packages [58,59], and post hoc testing was conducted using the multcomp package [60]. Significance was determined at p < 0.05 for all tests and all values are presented as means ± s.e.m. Figures were made using SigmaPlot 13.0 and presented as treatment means pooled across days of exposure, since day of exposure was not significant in any test.

Results

Physiological measurements

For all parameters, the day of testing was not significant, so results are presented across CO2 treatments. Haemolymph pHe was significantly affected by CO2 exposure (figure 1a; F2,29 = 5.94, p = 0.007), but was not affected by the day of testing (F1,29 = 0.248, p = 0.622). Accordingly, post hoc comparisons on the effect of CO2 exposure on haemolymph pHe reflected pooled values across days of exposure. Aplysia exposed to CO2 for 4–11 days showed a significant reduction in haemolymph pHe at 3000 (t = −2.736, p = 0.021), but not at 1200 µatm CO2 when compared with controls (figure 1a; t = 0.373, p = 0.712). As expected, Aplysia showed evidence of a compensatory response via the accumulation of (figure 1b). was significantly affected by CO2 exposure (F2,27 = 157.53, p < 0.001), but was not affected by day of testing (F1,27 = 2.15, p = 0.15). Post hoc comparisons revealed significant differences in between all three CO2 levels (figure 1b, all p < 0.001). pCO2 increased significantly with CO2 exposure (F2,27 = 87.41, p < 0.001), but was also not affected by day of testing (F1,27 = 0.036, p = 0.850). Post hoc testing revealed significant differences in pCO2 between all three CO2 levels (figure 1c, all p < 0.001).

Figure 1.

Haemolymph (a) pHe, (b) and (c) PCO2 in Aplysia (Aplysia californica) exposed to either control (400), 1200 µatm CO2 or 3000 µatm CO2 for 4–11 days. Values are reported as means ± s.e.m; n = 10–11. Means that share the same letter are not significantly different (p < 0.05).

The relationship between haemolymph and PCO2 exposure was not perfectly linear, which probably accounts for incomplete pH compensation at 3000 µatm CO2 (electronic supplementary material, figure S1).

Behavioural responses

Aplysia exposed to CO2 displayed no difference in the time to right when compared with control animals (figure 2a; F1,13 = 0.411, p = 0.533). The day of testing did not affect the righting response (F1,12 = 0.002, p = 0.964). Tail-withdrawal time was significantly affected by increased CO2 exposure (F2,15 = 4.52, p = 0.029), but was not affected by the day of testing (F1,16 = 0.04, p = 0.84). Accordingly, post hoc comparisons on the effect of CO2 exposure on tail-withdrawal reflex time reflected pooled values across days of exposure. Animals exposed to 1200 and 3000 µatm CO2 relaxed their tail approximately 36–37% faster than control animals (figure 2b; z = −2.521, p = 0.027, z = −2.612, p = 0.027, respectively). High CO2-exposed groups did not show a significant difference from one another (z = 0.134, p = 0.893). The percentage of body length withdrawn following tail depression exhibited no significant differences with treatment or day (figure 2c; treatment: F1,49 = 2.58, p = 0.12, day: F1,49 = 1.36, p = 0.25).

Figure 2.

Behavioural analysis in Aplysia (Aplysia californica) exposed to either control (400), 1200 µatm CO2 or 3000 µatm CO2 for 4–11 days. (a) Righting reflex (n = 13–16), (b) tail-withdrawal reflex (TWR) amount of time to relax the tail to 50% of original length and (c) TWR percentage of starting body length withdrawn following tail touch (n = 19, 17, 16 for control, 1200 µatm CO2 and 3000 µatm CO2, respectively). All values are reported as means ± s.e.m. Means that share the same letter are not significantly different (p < 0.05).

Discussion and conclusion

Aplysia exposed to elevated CO2 (1200 and 3000 µatm CO2) were able to accumulate significantly higher levels of in haemolymph following a 4–11 day exposure period (figure 1b). This compensatory effort led to complete pH defence at 1200 µatm CO2, an ocean acidification-relevant level close what is predicted globally by year 2100 (940 µatm CO2 under business as usual [2]) (figure 1a; 1200 µatm CO2). Of the two behavioural responses tested, tail withdrawal was impacted by high CO2 exposure, as hypothesized, whereas righting was not (figure 2).

It has long been known that invertebrates show more inherent variation in acid–base regulatory ability than fish. Generally, active invertebrates tend to show a stronger buffering capacity, while less active invertebrates may experience metabolic suppression associated with a decline in pH [19,37]. In studies addressing acid–base status of invertebrates at ocean acidification-relevant CO2 levels (less than approximately 2000 µatm CO2), sea urchins (Paracentrotus lividus, Echinometra mathaei, Tripneustes ventricosus) [36,38,61-63], Arctic spider crabs (Hyas araneus) [35], velvet swimming crabs (Necora puber) [64] and shore crabs (Carcinus maenas) [33] all accumulate to correct an acidosis. In contrast, blue mussels (Mytius edulis) [38], king scallops (Pecten maximus) [34], northern sea urchins (Strongylocentrotus drobachiensis) [65], sea stars (Asteria rubens, Leptasterias polaris) [33,66], slate pencil sea urchins (Eucidaris tribuloides) [36] and Arctic spider crabs at higher CO2 levels (3000 µatm CO2) [35] show incomplete or an absence of accumulation that is often insufficient in maintaining pH during high CO2 exposure. The diversity in acid–base responses to CO2 seen among invertebrates offers a fruitful avenue for studies of the mechanistic underpinnings of disturbed behaviour. Responses in animals showing regulatory and non-regulatory responses can be studied in the same species using Aplysia. It is clear that they regulate pH at lower CO2 levels (1200 µatm CO2) but cannot maintain this response at higher CO2 levels (3000 µatm CO2) (figure 1a; electronic supplementary material, figure S1).

Similar to acid–base regulatory ability, the behavioural responses of invertebrates have been variable. In the present study, CO2 exposure did not alter the self-righting response of Aplysia (figure 2a). This mirrors self-righting results of CO2-exposed gastropod molluscs (Gibberulus gibbosus) [13] and sea stars (Asteria rubens) [33]. Some studies have noted a faster righting time with elevated CO2, in brittlestars (Ophiura ophiura) at higher CO2 levels (corresponding to pH 7.3) [67], and in the Chilean abalone (Concholepas concholepas) [68]. In one case, righting time has been shown to increase with elevated CO2 exposure in a marine gastropod (Margarella antarctica) [69], and there was a trend of an increase in the cone snail (Conus marmoreus; p = 0.052) [70]. The tail withdrawal, a defence mechanism elicited by Aplysia, showed a significant decrease in reflex time at elevated CO2 levels (figure 2b), taking more time to relax the tail to half its original length after maximum contraction, but showed no change to the magnitude of the response (% body length contracted; figure 2c). Animals exposed to elevated CO2 relaxed their tail approximately 37% faster compared with control animals. The decrease in the timing of the tail-withdrawal reflex could suggest a decline in antipredator response or increased boldness, findings that have been observed across taxa [9,13,14,16]. For example, the marine snail G. gibbosus jumped away from a predator cue less frequently and with increased latency when exposed to elevated CO2 [13]. Similarly, flight behaviour of the black turban snail (T. funebralis) was reduced with increasing CO2, albeit at higher CO2 levels corresponding to a pH of 7.1 [15].

Given the ubiquity of CO2-induced behavioural disruptions across taxa, a common neural mechanism has been proposed, where altered and Cl− ion gradients resulting from efforts to maintain pH homeostasis are presumed to change the function of the GABAA receptor [11]. Despite the proposed link between acid–base regulatory ability and behavioural disruptions in marine organisms, this study represents one of few that have measured both parameters in the same species at ocean acidification-relevant CO2 levels [30-35]. Although Aplysia experienced an acidosis at 3000 µatm CO2, they were still able to accumulate at both CO2 levels. Based on extracellular measurements, this change could alter neuronal gradients and possibly explained shortened time to tail relaxation. However, in other invertebrate studies, sea stars unable to elevate [32,33] and crabs able to elevate both showed no difference in righting [35]. Scallops showing an acidosis with very limited accumulation showed significant impacts on clapping performance [34]. The lack of consistency across studies and in Aplysia in the current study may seem difficult to reconcile. The source of variation could stem from a number of factors including differential intracellular pH regulation or behavioural compensatory mechanisms. In addition, these variations may reflect that certain behaviours are not GABA-mediated. It is clear that the field would benefit from more measurements of acid–base parameters in species showing behavioural disruptions to help resolve these discrepancies.

Although the involvement of the GABAA receptor was not directly tested in the present study, GABAA receptor involvement in CO2-induced behavioural disruptions have been demonstrated in some fish [8,10,23-28] and invertebrates [13,29]. These studies have largely implicated GABAA receptor involvement using whole-animal exposure to pharmacological agents targeting GABAA. This method has been fundamental in establishing the proposed mechanism, but lacks resolution in targeting specific mechanisms responsible for a given behavioural disturbance. Since Aplysia accumulate and show a significant behavioural disruption at both tested CO2 levels, they are an ideal candidate for obtaining a better understanding of mechanisms underlying CO2 behavioural impairment. Findings from the present study combined with decades of research examining the electrophysiological basis of learning means that methods to link well-characterized neural networks to specific behaviours are already established. For many reflexes, including the CO2-impacted tail-withdrawal reflex, the reflex can be elicited in in vitro preparations where the specific neural network for a given reflex is isolated from the animal [56]. In addition, Aplysia neurons are large and amenable to patch clamp techniques, where individual cells and/or specific transporters can be investigated. While the specific role of the GABAA is not well-studied in the context of the CO2-impacted tail-withdrawal reflex, gamma-aminobutyric acid (GABA) has been localized to certain areas in the pedal ganglia [71], which innervates the tail [42,72]. Furthermore, Aplysia neurons from a number of regions including the pleural ganglia (also involved in the tail-withdrawal reflex), have shown both excitatory and inhibitory currents with the application of GABA and were found to be reactive to GABAA receptor antagonists [73].

In summary, we believe all of the advantages of using Aplysia as a biomedical research model for learning could be applied to ocean acidification research. Aplysia meet three important criteria (1–3). In addition to simple and well-mapped nervous systems (1), there are established and reproducible behavioural assays (2) that can be applied to examine all major forms of learning including habituation, sensitization, classical conditioning and operant conditioning [74]. Most importantly, the present study demonstrates that Aplysia accumulate at an ocean acidification-relevant CO2 level (3). These three criteria allow for further exploration of the proposed link between acid–base regulatory ability and behaviour, including detailed testing of GABAA hypothesis.