Discussion

These data provide the first evidence that loss of ASIC1A has age-dependent effects on acquisition and recall of both cued and contextual fear memory. Our key findings are that: 1) Freezing during acquisition is lower in Asic1a_-/-_ mice across all ages tested, 2) Cued fear memory tested at 24 h and 15 days after training (D15) is impaired in post-weaning Asic1a_-/-_ mice, but is not impaired in P17 Asic1a_-/-_ mice, and 3) Context fear memory is largely impaired in Asic1a_-/-_ mice when tested at 48 h and at 16 days after training (D16). Interestingly, effects of ASIC1A disruption on freezing increased with age.

Fear memory acquisition in Asic1a_-/-_ mice was increasingly impaired with age, suggesting a developmental role of ASIC1A in fear learning. This idea is consistent with reports that ASIC1A is expressed very early in development, as early as embryonic day 12 (Alvarez de la Rosa et al. 2003). The BLA and hippocampus, key sites for fear learning and fear memory recall, robustly express ASIC1A (Wemmie et al., 2003, Coryell et al., 2007, Zha et al., 2006). It largely remains unclear if ASIC1A expression in these sites is developmentally regulated, though one paper reported a doubling in ASIC1A membrane expression in the hippocampus between the second and 12th postnatal week (Ma et al. 2019). If ASIC1A levels do increase in these areas in the timeframe assessed here, it could explain the progressively greater impairments in fear memory acquisition and cued fear memory recall in Asic1a_-/-_ mice. While we did not assess ASIC1A protein levels in this study, it would be an important that future studies look at the developmental time course of ASIC1A expression in the BLA and hippocampus to see if ASIC1A protein expression throughout development parallels the behavioral effects we described here.

These age-dependent deficits may also reflect changes in glutamate receptors across development. ASIC1A has been implicated in synaptic plasticity, with ASIC1A disruption impairing LTP in both the amygdala and hippocampus (Du et al., 2014; Chiang et al. 2015; Wemmie et al. 2002). Glutamate receptors are essential for LTP generation and fear conditioning (Maren, 1999, Maren, 2005). Two receptor subunits that have been critically implicated in fear memory acquisition, NR2B (Zhao et al., 2005, Zhang et al., 2008, Rodrigues et al., 2001) and GluA1 (Humeau et al., 2007, Feyder et al., 2007, Rumpel et al., 2005), are both developmentally regulated in the BLA. NR2B expression in the BLA decreases by almost half between P17 and adulthood, whereas GluA1 phosphorylation at Ser831, which is key for short-term memory formation, decreases by more than 80% in the same time period (Bessieres et al. 2019). Interestingly, ASIC1A disruption has been reported to alter expression and function of these glutamate receptor subunits. Asic1a_-/-_ mice have been reported to have decreased postsynaptic NR2B expression, and decreased NMDA receptor activity in striatal neurons (Ma et al., 2019, Yu et al., 2018). Moreover, altered GluA1 membrane trafficking, GluA1 phosphorylation and GluA1 protein expression have been reported in Asic1a_-/-_ mice (Li et al., 2019, Mango et al., 2017, Yu et al., 2018). Although these effects of ASIC1A disruption on glutamate receptors have not yet been evaluated in the fear circuit, they suggest a plausible mechanism for the age-dependent effects we observed. At younger ages, higher levels of NR2B expression and GluA1 Ser831 phosphorylation may partially compensate for the loss of ASIC1A, and this compensation may diminish as levels of these subunits decline.

Despite their impaired acquisition, Asic1a_-/-_ mice trained at P17 showed unimpaired cued fear memory recall when tested at 24 h and at 15 days after training (D15). This suggests the presence of a compensatory mechanism after acquisition that facilitates cued fear memory consolidation, leading to normal cued fear memory recall. This age-specific mechanism appears to be specific to cue-evoked fear memory, as contextual fear memory recall is impaired in Asic1a_-/-_ mice trained at P17, suggesting separable underlying processes.

A possible site where compensation might occur, is the BLA. Stabilization of new dendritic spines in the amygdala is essential for memory consolidation (Holtmaat et al., 2008, Muñoz-Cuevas et al., 2013). Multiple studies have implicated ASIC1A in spine dynamics (Zha et al., 2006, Kreple et al., 2014). Notably, there seem to be region-specific effects, with ASIC1A in the nucleus accumbens thought to oppose the formation of thin and stubby spines (Kreple et al. 2014), whereas increasing ASIC1A expression in the hippocampus increased dendritic spine density (Zha et al. 2006). Importantly, the role of ASIC1A in dendritic spine dynamics in the amygdala has not yet been assessed, though it is possible that altered spine dynamics in Asic1a_-/-_ mice might contribute to a compensatory mechanism after acquisition. If there were increased formation of dendritic spines or decreased turnover in the amygdala of Asic1a_-/-_ mice trained at P17, this could explain the milder impairment of cued fear memory acquisition and enhance consolidation. If this mechanism is age-dependent, and either absent or diminished in mice trained at older ages, it would explain the deficit we see in cued fear memory recall at 24 h after training at all older ages.

Cue-evoked freezing in mice trained at P17 declined significantly between 24 h and 15 days after training (D15), regardless of genotype, but remained stable in mice trained at all older ages. This might reflect developmental differences in synaptic pruning. Spine turnover is high at P17 (Holtmaat et al., 2005, Runge et al., 2020), and the preservation of spines plays an important role for memory maintenance after initial consolidation (Basu and Lamprecht, 2018). Though direct comparisons in spine stability between P17 and P23 are not available, there is a general trend of increasing spine stability with age, raising the possibility that low spine stability at P17 contributes to the decline in cue-evoked freezing.

In contrast to cued fear memory recall, contextual fear memory recall at 48 h was largely impaired in Asic1a_-/-_ mice at all ages. Follow-up analysis showed that male Asic1a_-/-_ mice trained at P17 showed no evidence for retention of contextual fear memory at 48 h after training. However, female Asic1a_+/+_ mice showed significant retention of contextual fear memory after training, and female Asic1a_-/-_ mice and male Asic1a_+/+_ mice trended in the same direction. At all older ages, both sexes and genotypes retained contextual fear memory at 48 h after training. It has been reported that Pavlovian fear conditioning before P17 does not to induce stable context-evoked freezing in mice when tested 24 h after training and the hippocampus is essential for the development and retention of contextual fear memory (Akers et al. 2012). Our results might indicate that hippocampal development necessary to retain contextual fear memory might be delayed in Asic1a_-/-_ males, resulting in a lack of retention of contextual fear memory in these mice at 48 h after training. Our results in mice trained at P17 also suggest a developmental dissociation between the retention of context-evoked and cue-evoked freezing. Interestingly, a similar developmental dissociation between cued and contextual fear memory recall has been reported for rats (Rudy, 1993, Rudy and Morledge, 1994).

Contrasting the stability of cue-evoked freezing in all mice trained post-weaning, context evoked fear memory tended to decline between 48 h and 16 days after training in all mice trained before P42. Regional differences in maturation of the circuits for cued and contextual fear memory could possibly explain this developmentally later onset of stability of context-evoked freezing – relative to cue-evoked freezing. In line with this a delayed onset of context evoked fear memory – relative to cue evoked fear memory – has been reported for rats (Burman et al. 2014). While no study tested parameters identical to ours, it has been found that retention of 30-day memory for both, cue and context, emerges at P25 (Samifanni et al. 2021). One possible explanation for the relative instability of contextual fear memory in younger mice could be the greater levels of pruning seen at younger ages and the differences in stability between cued and contextual fear memory could reflect differences in pruning in their respective underlying circuits (Holtmaat et al., 2005, Runge et al., 2020, Koss et al., 2014, Parato et al., 2019).

Studying sex-effects was not the primary objective of this study, and thus experimental groups were not necessarily powered to study this. Despite this, we did observe some sex-dependent effects. It is interesting that sex-genotype interactions for cued and contextual fear memory emerged in mice trained at P42, which may reflect hormonal changes occurring as mice reach sexual maturity around this time (Bell, 2018). While sex differences in fear conditioning have been reported, they are complex, and the reported results are equivocal (Bauer, 2023). So far sex differences in ASIC1A expression have not yet been studied.