The primary purpose of this study was to examine the plausibility of inflammation-mediated learning and memory dysfunction in L. stagnalis. The concept of inflammation and inflammation-mediated oxidative stress as agents of neural dysfunction and degeneration has become well-established in the context of mammalian nervous systems [5, 7–11, 17–25, 40]. However, despite their prominence as neurobiological model systems of learning and memory, remarkably little is known about neuroinflammatory aspects of learning and memory dysfunction in gastropods. Recent findings implicating non-enzymatic lipid peroxidation in PLA2-dependent LTM failure in aging L. stagnalis[6, 27] led us to hypothesize involvement of the snails’ immune system in learning and memory dysfunction. The current results support that hypothesis, even though many questions about the identity, modes and loci of action of the signalling processes involved remain to be answered. This conclusion hinges on the following observations: 1- systemic delivery of the immune activator laminarin induces a transient increase in H2O2 release from circulating haemocytes within 1-3 hrs after injection that dissipates within 24 hrs; 2- intracoelomic injection of laminarin 1 hr before the start of behavioural conditioning induced LTM dysfunction; 3- intracoelomic injection of laminarin 24 hrs before the start of behavioural conditioning had no impact on LTM function; 4– co-administration of laminarin with putative PLA2 inhibitor aristolochic acid or the putative COX inhibitor indomethacin negated laminarin’s effect on LTM; 5- Neither aristolochic acid nor indomethacin inhibited laminarin-induced haemocyte- mediated H2O2 release; 6- intracoelomic injection of laminarin did not cause obvious chemosensory or motor deficits; 7- laminarin injection did not affect intermediate term memory (ITM) function; 8- exposure of isolated CNS to laminarin did not affect electrical activity of interneuron CGC. Taken together these results support a model in which the haemocyte’s inflammatory response disrupts LTM formation by recruiting neuronal PLA2-mediated processes. As will be discussed below, such a model is consistent with our recent finding that PLA2 plays a pivotal role in LTM impairment associated with oxidative stress and old age in L. stagnalis.
We propose that the inflammatory response from circulating haemocytes is a primary cause in laminarin’s behavioural effects. We base this opinion on the following evidence. First, the H2O2 release data leaves no doubt that circulating haemocytes are among the targets of laminarin in our experiments. Second, neither the chemosensory response to amyl-acetate, nor the behavioural response to sucrose application were affected by laminarin 1 hr after its injection, a time point where no measurable increase haemocyte H2O2 release had occurred yet. Third, laminarin applied to isolated CNS preparations had no effect on spontaneous spiking activity and resting membrane potential of the Cerebral Giant Cell, one of the interneurons of the feeding circuit that is crucial to the formation of appetitive LTM and that receives chemosensory synaptic input and maintains widespread synaptic connections throughout the circuit. Although it is difficult if not impossible to test all-inclusively for the absence of drug effects, these results together provide quite compelling evidence that laminarin has no substantial direct neuronal effects in at least the first 1-1.5 hr after treatment. Further investigations are required to assess whether, on the longer run, laminarin may affect components of the nervous system like perhaps the professional phagocytes known to reside in the CNS .
The present results show striking parallels with a previous report that aristolochic acid corrects experimental oxidative stress-induced LTM failure in L. stagnalis as well as age-associated LTM impairment . In other words, PLA2 appears to lie at the root of oxidative stress-dependent, inflammation-induced and age-associated LTM impairment paradigms in the L. stagnalis model system. We interpret these parallels as evidence of a close causal relation between these phenomena. This interpretation resonates well with a rapidly growing literature that associates PLA2 with (neuro)inflammation, deregulation of lipid metabolism, and cell-, neuronal- and cognitive dysfunction in humans and other mammals [9, 19–24].
We assume that aristolochic acid’s ability to restore LTM impairment in the current inflammation model reflects the drug’s actions at the level of the nervous system. This idea arises from previous evidence demonstrating that PLA2 inhibition with aristolochic acid reverses, within minutes, electrophysiological phenomena induced by experimental oxidative stress in isolated CNS preparations thought to contribute to LTM impairment as well as similar phenomena in the brains isolated from LTM impaired old snails [6, 27]. The idea of a neural locus of aristolochic acid’s restorative actions is further supported by the present finding that the drug does not affect laminarin-induced haemocyte H2O2 release, a result consistent with prior literature emphasizing phospholipase C (PLC)/protein kinase C (PKC)-dependent signalling pathways in this process [35, 36]. Parenthetically, assuming the above reasoning is true, the question of identity of the signalling intermediate(s) between haemocytes and neurons arises. There is probably more than one answer to this question. However, based on the uncanny parallels between the experimental oxidative stress model of LTM impairment that identifies non-enzymatic lipid peroxidation as a probable activator of neuronal PLA2[6, 27] and the current data, we consider oxidizers released by activated haemocytes are plausible candidates.
At this point it is important to recognize the importance of ROS and RNS signalling in synaptic plasticity and memory formation including the induction of LTM in the very model system we used here [5, 15, 19, 42–44]. Specifically, NO-signalling is particularly relevant during the first 5 hours of LTM consolidation in the current LTM paradigm [42, 44]. Since L. stagnalis hemocyte’s release substantial amounts of NO together with H2O2 during their respiratory burst [35, 36], the possibility that haemocyte-derived NO interferes directly with critical NO-dependent steps in appetitive LTM formation exists.
The finding that the putative COX inhibitor indomethacin, but not the putative LOX inhibitor NDGA, partially reversed laminarin’s inhibition of appetitive LTM suggests COX involvement in this process. Although this notion is consistent with the metabolic interdependence of COX and PLA2, this conclusion needs to be treated with some care because, as discussed in more detail below, reservation’s about indomethacin’s COX specificity. Nevertheless, the notion of COX involvement in inflammation-induced learning impairment is consistent with the enzyme’s long-standing stature as a pivotal producer of pro-inflammatory mediators [17, 18, 45] and not without precedent in the context of synaptic transmission and plasticity, learning and memory formation [13–16, 19, 46–53]. Very little, if anything, is known about inflammatory and/or neuro-modulatory implications of COX activity in L. stagnalis. Considering growing evidence implicating PLA2 and COX-metabolites in cognitive function and dysfunction in mammals [4, 5, 7, 46–53], the current results, although not conclusive, do warrant further investigation into this matter.
Haemocyte activation: in vivo immune challenge vs. in vitro immune challenge
Our data support the conclusion that a single intracoelomic injection of laminarin provokes an increase in haemocytic H2O2 release within 1-3 hours after injection that dissipates within 24 hrs of the injection. Whereas these results are in general agreement with those obtained in in vitro studies of L. stagnalis haemocytes, they also reveal important differences . Particularly, whereas H2O2 release increases within 30 to 60 minutes after in vitro stimulation of isolated haemocytes with 5 mg/ml laminarin, our results indicate that the same may take up to three hours when laminarin is injected into the animal’s body cavity. Considering the much more complex biological and pharmacokinetic conditions prevailing in the whole animal, a somewhat slower and smaller in vivo response does not seem to be out of the ordinary. Most importantly, although we cannot pinpoint the exact time course of the response, our data do support the notion that the impact of laminarin injection on haemocyte H2O2 release is gone by the time LTM tests are performed 24 hrs after training.
Limitations: alternative targets of laminarin and drug specificity?
As any other study relying on pharmacological interventions, our results might be affected by non-specific actions of the drugs we used. Aristolochic acid has a long track record as broad spectrum PLA2 inhibitor [54–56]. Other than an extensive literature on the compound’s genotoxic effects that are associated with its ability to form aristolactam-DNA adducts [57, 58], we are not aware of reports of non-specific activities. It seems unlikely that DNA-adduction is a factor of importance in the present study. An observation strengthening our believe in aristolochic acid’s relative specificity is the finding that the compound does not interfere with laminarin’s ability to activate haemocytes, a process that involves a major phospholipid-dependent signalling component in the form of protein kinase C- (PKC), phospholipase C- (PLC), extracellular signal-regulated kinase- (ERK) and MAPK/ERK kinase (MEK)-dependent signalling cascade . Moreover, no effect of aristolochic acid on electrophysiological, behavioural and biochemical facets were observed in L. stagnalis without any form of pro-oxidant present . While we have no compelling reason to suspect aristolochic acid’s pharmacological activities, indomethacin warrants more extensive scrutiny. For instance, like other NSAIDs, indomethacin may modify the activity of various serine/threonine kinases [59, 60]. However, electrophysiological studies on L. stagnalis neurons revealed no evidence of indomethacin effects on PKC nor protein kinase A (PKA) signalling at the concentrations used here . Moreover, we show that indomethacin does not affect laminarin induced H2O2 release from haemocytes, which as noted before, is a PKC-dependent process. Thus, it seems unlikely that the effects of indomethacin reported here are due to interference with PKA and/or PKC signalling. However, several reports indicate that indomethacin, in addition to its ability to inhibit COX, can inhibit several types of PLA2 enzymes [55, 62–64]. No information is available about the type of PLA2 underlying the current observations. Thus we cannot exclude the possibility that indomethacin’s effects reported here are in part due to the inhibition of PLA2. However, that uncertainty does not in any way invalidate our central conclusion that laminarin-mediated haemocyte activation induces PLA2-dependent LTM dysfunction.
Differential susceptibility of ITM and LTM to systemic immune challenge
Our results show that non-aversive appetitive ITM in contrast to its LTM counterpart is not significantly compromised by systemic injection of laminarin. There are a number of noteworthy observations to make from this finding. First, it provides additional evidence that relatively short term exposure to laminarin itself does not compromise the neural systems and processes involved in the detection and execution of chemosensory-induced feeding responses. Second, it points towards independence of the neural substrates ITM and LTM and as such reiterates recent evidence by Marra et al. indicates that non-aversive appetitive ITM and LTM formation in this model follows a parallel rather than serial causal pathway . Third, the current evidence that LTM is more sensitive to disruption of immune-related processes than ITM takes on a particularly intriguing perspective in view of a growing body of literature associating (neuro)inflammation with age-associated cognitive and neurological dysfunction and our own work showing LTM but not ITM to be impaired in old L. stagnalis in both learning and memory paradigms we use in our research [6, 26].