Results

Baseline freezing is largely normal in Asic1a-/- mice across development

Asic1a_-/-_ and Asic1a_+/+_ mice trained at P17, P23, or P42 showed similar baseline freezing on training day (P17: t(28.8) = 1.011, p = 0.320; P23: t(8.9) = 1.594, p = 0.146; P42: t(3.5) = 1.191, p = 0.309) (Fig. 1A, B, D; Table 1). Contrasting this, Asic1a_-/-_ mice trained at P29 showed reduced baseline freezing compared to Asic1a_+/+_ mice (t(7.1) = 2.400, p = 0.047) (Fig. 1C; Table 1). In mice trained at P83 we saw a trend towards lower baseline freezing for Asic1a_-/-_ mice (t(30.0) = 1.769, p = 0.087) (Fig. 1E; Table 1). Given the small numerical differences in baseline freezing, the biological significance of those numerical differences in baseline freezing remains unclear. We saw no significant sex-effects nor significant sex-by-genotype interactions at any age. The model which probed for effects of age on baseline freezing did not yield significant results (t(104.6) = −0.985, p = 0.327) and there was no age-by-genotype interaction (t(65.8) = −0.622, p = 0.536).

Age at training | Training baseline freezing | Training from 1stcue onset | Cued fear memory test 24 hrs. after training | Cued fear memory test 15 days after training | Contextual fear memory test 48 hrs. after training | Contextual fear memory test 16 days after training
 | Asic1a+/+ | Asic1a-/- | p | Asic1a+/+ | Asic1a-/- | p | Asic1a+/+ | Asic1a-/- | p | Asic1a+/+ | Asic1a-/- | p | Asic1a+/+ | Asic1a-/- | p | Asic1a+/+ | Asic1a-/- | p
P17 | 18.0 ± 24.1 | 12.2 ± 13.9 | 0.320† | 63.4 ± 23.4 | 50.9 ± 20.4 | 0.004* | 58.1 ± 36.6 | 58.8 ± 35.7 | 0.922 | 29.3 ± 33.1 | 13.8 ± 21.9 | 0.143 | 43.6 ± 31.4 | 18.4 ± 24.3 | 0.009*† | 14.5 ± 14.0 | 6.5 ± 8.8 | 0.036*†
P23 | 3.0 ± 8.8 | 0.1 ± 0.3 | 0.146† | 50.8 ± 12.0 | 32.4 ± 11.7 | 0.001* | 75.6 ± 30.3 | 51.7 ± 31.4 | 0.019* | 75.5 ± 23.6 | 34.4 ± 22.1 | < 0.001* | 30.3 ± 22.3 | 7.9 ± 12.8 | 0.001*† | 16.8 ± 12.6 | 6.7 ± 6.1 | 0.003*†
P29 | 1.3 ± 1.4 | 0.2 ± 0.6 | 0.047*† | 57.7 ± 12.3 | 35.2 ± 15.4 | < 0.001* | 86.0 ± 19.6 | 46.3 ± 23.1 | 0.003* | 70.5 ± 27.3 | 36.6 ± 26.3 | 0.021* | 43.0 ± 23.2 | 21.7 ± 17.8 | 0.002*† | 21.2 ± 26.0 | 8.3 ± 13.0 | 0.132†
P42 | 0.8 ± 0.1.1 | 0.4 ± 0.8 | 0.309† | 49.1 ± 15.0 | 25.0 ± 18.6 | 0.014* | 68.8 ± 25.5 | 29.7 ± 22.3 | 0.007 | M:81.7 ± 12.3F:72.4 ± 20.8 | M:15.5 ± 14.1F:39.1 ± 18.1 | M: < 0.001*F: 0.005* | M:37.4 ± 14.0F:19.8 ± 11.6 | M:5.8 ± 4.6F:12.3 ± 14.1 | M: < 0.001*†F:0.144† | 20.2 ± 11.1 | 6.4 ± 10.7 | 0.051†
P83 | 1.1 ± 2.8 | 0.1 ± 0.3 | 0.087† | 43.6 ± 12.2 | 10.6 ± 8.1 | < 0.001* | 73.9 ± 14.6 | 15.1 ± 15.5 | < 0.001* | 69.6 ± 16.3 | 11.2 ± 9.7 | < 0.001* | M:16.2 ± 12.7F:42.6 ± 17.0 | M:11.5 ± 21.0F:7.7 ± 11.7 | M:0.318†F:0.009†* | 18.8 ± 10.8 | 2.7 ± 5.1† | < 0.001*†

The effect of ASIC1A disruption on fear memory acquisition increases with age

Asic1a_-/-_ mice trained at all ages showed lower freezing during fear memory acquisition than Asic1a_+/+_ mice (P17: t = (27.6)= 3.161, p = 0.004; P23: t(8.6) = 4.580, p = 0.001; P29: t(30.0) = 4.594, p < 0.001; P42: t(5.9) = 3.428, p = 0.014; P83: t(30.0) = 10.350, p < 0.001) (Fig. 1A-E, Table 1). A sex effect was observed for mice trained at P83, females freezing more than males (t(30.0) = −3.209, p = 0.003), but there was no genotype-by-sex interaction, so data are not separated by sex. In both genotypes, freezing during fear memory acquisition tended to decrease with age, however this decrease was only significant in Asic1a_-/-_ mice (Asic1a_-/-: t(65.7) = −5.16, p < 0.001; Asic1a+/+_: t(45.6) = −1.70, p = 0.096), resulting in increased genotype difference at older ages (genotype-by-age interaction: t(56.4) = 2.025, p = 0.048; estimated difference in mice trained at P17: 13.4%; estimated difference in mice trained at P83: 32.2%) (Supplementary Fig. 2).

To ensure animals of all age groups learned, we compared baseline freezing on training day to freezing during acquisition. For all ages freezing during acquisition significantly differed from baseline freezing indicating that all animals acquired fear memory (P17: Asic1a_-/-_ t(15.5) = 5.735, p < 0.001, Asic1a_+/+: t(15.5) = 6.841, p < 0.001; P23: Asic1a-/-_ t(18.6) = 9.777, p < 0.001, Asic1a_+/+: t(5.8) = 16.603, p < 0.001; P29: Asic1a-/-_ t(4.4) = 9.063, p = 0.001, Asic1a_+/+: t(6.2) = 16.707, p < 0.001; P42: Asic1a-/-_ t(3.7) = 5.172, p = 0.008, Asic1a_+/+: t(5.7) = 8.686, p < 0.001; P83: Asic1a-/-_ t(7.6) = 5.928, p < 0.001, Asic1a_+/+_: t(3.67) = 19.544, p < 0.001).

ASIC1A disruption increases shock reactivity in adult females

As fear memory acquisition was impaired in Asic1a_-/-_ mice, we were wondering if this acquisition deficit is due to reduced ability to feel the foot shock. To probe this possibility, we evaluated shock reactivity at each developmental stage at which we performed fear conditioning, using motion in response to the first foot shock. At P17, there was a trend towards increased motion in Asic1a_-/-_ mice, but there was no significant genotype effect was observed in mice trained from P17 to P42 (P17: t = (27.5)= 1.909, p = 0.07; P23: t(10.2) = 0.307, p = 0.765; P29: t(5.7) = −0.518, p = 0.624; P42: t(2.6) = −0.818, p = 0.481) (Supplementary Fig. 3A-D). In mice trained at P83, we saw a significant genotype-by-sex interaction (t(28.8) = 2.509, p = 0.018), reflecting a higher shock reactivity in female, but not male, Asic1a_-/-_ mice compared to their Asic1a_+/+_ counterparts (females: t(9.3) = 2.915, p = 0.017, males: t = (8.4)= 0.589, p = 0.571) (Supplementary Fig. 3E, F). Due to these results, we performed sex-specific analysis on the effect of training age on shock reactivity. For male mice, our model showed increased shock reactivity with increasing age (t(30.6) = 6.662, p < 0.001), but no genotype-by-age interaction (t(40.3) = −1.382, p = 0.175), indicating that both genotypes had a similar increase in shock reactivity. For female mice, our model also showed increased shock reactivity with increasing age (t(38.8) = 8.579, p < 0.001), but in contrast to our result for males, we saw a genotype-by-age interaction (t(45.1) = −3.573, p < 0.001). This increasing shock reactivity might result in higher freezing during fear memory acquisition, and thus decrease genotype difference during acquisition.

In mice trained post-weaning, loss of ASIC1A impairs cue-evoked fear memory at 24 h after training

Despite the deficit in fear memory acquisition that we saw in Asic1a_-/-_ mice trained at P17, both genotypes froze similarly during cue presentation at 24 h after training (t(32.9) = −0.099, p = 0.922) (Fig. 2A, Table 1). In contrast, Asic1a_-/-_ mice trained at P23 and beyond froze less than Asic1a_+/+_ mice during cue presentation at 24 h after training (P23: t(18.1) = 2.568 p = 0.019; P29: t(6.5) = 4.562, p = 0.003; P42: t(5.8) = 4.080, p = 0.007; P83: t(6.3) = 9.261, p < 0.001) (Fig. 2B-F, Table 1). For mice trained at P83 we also saw a significant sex-effect, with females freezing more than males (t(29.6) = −2.224, p = 0.034). Freezing during cue presentation 24 h after training declined with age in Asic1a_-/-_ mice (t(64.9) = −4.653, p < 0.001), but not in Asic1a_+/+_ controls (t(41.7) = 0.762, p = 0.450), a difference that resulted in a significant genotype-by-age interaction: t(67.0) = 3.714, p < 0.001). While Asic1a_-/-_ mice trained at P17 froze similarly as Asic1a_+/+_ mice (estimated difference 1.1%), Asic1a_-/-_ mice trained at P83 froze dramatically less (estimated difference 59.2%) (Supplementary Fig. 4). Data on pre-cue freezing can be found in Supplementary Table 3.

Fig. 2: Age-specific effects of Asic1a disruption on cue-evoked fear memory. Cue-evoked freezing in Asic1a knockout (-/-, blue) vs. wild-type (+/+, red) mice measured at 24 h (hrs.) and 15 days post-training for mice trained at: P17 (A); P23 (B); P29 (C); P42, reported by sex: P42 males (D) and P42 females (E); and P83 (F). For mice trained at P42, males and females are shown in different graphs to highlight a significant sex-by-genotype interaction at 15 days after training. No sex-by-genotype interaction was detected for 24hrs after training. For 24hrs. after training, the respective male-specific p-value (WT m vs. KO m) is 0.002, and the respective female-specific p-value (WT f vs. KO f) is 0.072. Mouse numbers per experiment and sex distribution are the same as reported in Fig. 1.*p < 0.05, **p < 0.005, ***p < 0.001.

In mice trained post-weaning, ASIC1A disruption impairs cue-evoked fear memory at 15 days after training

Asic1a_-/-_ and Asic1a_+/+_ mice were also tested for cue-evoked fear memory recall at 15 days after training (D15). In mice trained at P17, we saw no genotype effect at D15 (t(33.7) = 1.500, p = 0.143), which is in line with our result at 24 h after training (Fig. 2A, Table 1). In contrast, we did see a genotype effect for older mice: Asic1a_-/-_ mice trained at P23, at P29, or P83 froze less upon cue presentation on D15 than Asic1a_+/+_ mice (P23: t(42.0) = 5.661, p < 0.001; P29: t(6.5) = 3.040, p = 0.021; P83: t(30.0) = 14.658, p < 0.001) (Fig. 2B, C, F, Table 1). This genotype effect paralleled the results at 24 h after training. For mice trained at P23 or at P83 we saw a sex effect, as females froze more than males (P23: t(42.0) = −2.265, p = 0.029; P83: t(30.0) = −3.430, p = 0.002). For mice trained at P42, we saw a significant genotype effect in both sexes (males: t(11.2) = −7.057, p < 0.001; females: t(10.4) = −3.562, p = 0.005) (Fig. 2D, E, Table 1) and there was a significant sex-by-genotype interaction (t(29.8) = 2.780, p = 0.009), indicating that the genotype effect was larger in males than in females. In light of the sex-by-genotype interaction seen in mice trained at P42, we evaluated the effects of training age on cue-evoked fear memory at 15 days after training separately for each sex. For females we saw a significant genotype-by-age interaction (t(39.1) = 2.342, p = 0.024) while males exhibited a trend towards a genotype-by-age interaction (t(49.4) = 1.730; p = 0.090) (Supplementary Fig. 5A-B). Data on pre-cue freezing can be found in Supplementary Table 3.

When we were assessing the stability of cue-evoked fear memory in mice trained at P17, we saw a significant decline of cue-evoked freezing from 24 h to 15 days after training in both genotypes (Asic1a_+/+_ mice: t(27.6) = −4.082, p < 0.001; Asic1a_-/-_ mice: t(21.0) = −6.026, p < 0.001) (Fig. 2A, Table 1) with no significant difference in decline between genotypes (t(33) = −1.405; p = 0.169). In mice trained at P23, P29 or P83, within-genotype freezing at those timepoints remained stable (P23 Asic1a_+/+_ mice: t(8.9) = 0.049, p = 0.962; P23 Asic1a_-/-_ mice: t(17.8) = −1.033, p = 0.315; P29 Asic1a_+/+_ mice: t(6.4) = −1.946, p = 0.097; P29 Asic1a_-/-_ mice: t(5.8) = −1.197, p = 0.278; P83 Asic1a_+/+_ mice: t(4.91) = −0.915, p = 0.402: P83 Asic1a_-/-_ mice: t(7.38) = −0.676, p = 0.520) (Fig. 2B, C, F, Table 1), and there was also no difference in the stability of cue-evoked freezing between genotypes (P23: t(13.8) = −0.784, p = 0.447; P29: t(6.1) = 0.379, p = 0.718; P83: t(5.7) = 0.298, p = 0.776). Despite of the sex-by-genotype interaction we saw for freezing at 15 days after training in mice trained at P42, we saw no sex-by-genotype interaction for the stability of cue-evoked freezing between 24 h and 15 days after training. There was no significant change in freezing between genotypes between 24 h and 15 days after training (t(5.1) = −0.727, p = 0.500) and within-genotype freezing between those timepoints also remained stable (Asic1a_+/+_ mice: t(6.05) = 0.679, p = 0.522; Asic1a_-/-_ mice: t(4.14) = −0.330, p = 0.758). Due to the sex-by-genotype interaction at D15, freezing at 24 h and D15 are plotted by sex for mice trained at P42 (Fig. 2D, E).

Loss of ASIC1A has age-and sex-specific effects on context-evoked freezing tested at 48 h after training

Context-evoked freezing, tested at 48 h after training, was significantly lower in Asic1a_-/-mice trained from P17 to P29 (P17: t(32.7) = 2.793, p = 0.009; P23: t(16.63) = 3.894, p = 0.001; P29: t(30.0) = 3.481, p = 0.002) (Fig. 3A-C, Table 1). In mice trained at P42, we saw a sex-by-genotype interaction (t(29.8) = 2.834, p = 0.008) indicating, lower freezing in Asic1a-/-_ males (t(11.39) = −5.192, p < 0.001), but not in Asic1a_-/-_females (t(10.57) = −1.58, p = 0.144) compared to same-sex wildtypes (Fig. 3D, E, Table 1). In mice trained at P83, we also saw a significant sex-by-genotype interaction (t(28.3) = −2.118, p = 0.043) indicating a genotype effect in females, but not in males (males: t(8.9) = −1.057, p = 0.318, females: t(9.8) = −3.232, p = 0.009) (Fig. 3F, G, Table 1). Due to the sex-by-genotype interaction seen in mice trained at P42 and at P83, we evaluated the effects of training age on context-evoked fear memory separately for each sex. Neither for males nor for females did we see a genotype-by-age interaction (males: t(53.9) = −1.139, p = 0.260; females: t(37.3) = 1.541, p = 0.132) (Supplementary Fig. 6A-B).

Fig. 3: Age-specific effects of Asic1a disruption on contextual fear memory. Context-evoked freezing in Asic1a knockout (-/-, blue) vs. wild-type (+/+, red) mice at 48 h (hrs.) and 16 days post-training for mice trained at: P17 (A); P23 (B); P29 (C); P42, by sex: P42 males (D) and P42 females (E); and P83, by sex: P83 males (F) and P83 females (G). For mice trained at P42 and at P83, males and females are shown in different graphs due to significant sex-by-genotype interactions at 48 h after training. There were no sex-by- genotype interactions 16 days after training. For P42, the respective male-specific p-value (WT m vs. KO m) is 0.002 and the respective female-specific p-value (WT f vs. KO f) is 0.072. For P83 the respective male-specific p-value (WT m vs. KO m) is 0.002 and the respective female-specific p-value (WT f vs. KO f) is < 0.001. Mouse numbers per experiment and sex distribution are the same as reported in Fig. 1. *p < 0.05, **p < 0.005, ***p < 0.001.

Given the small difference between baseline freezing at day zero and freezing upon re-exposure to the training context at 48 h after training (Table 1) we performed follow up-analysis to probe if mice of each age group, genotype and sex retained memory about the training context by comparing baseline freezing at training day with context freezing at 48 h after training. This analysis revealed that in mice trained at P17 only Asic1a_+/+_ females showed a significant difference between baseline freezing at day zero and contextual fear memory recall at 48 h after training (t(37.8) = −3.29, p = 0.002), whereas a trend in this direction was observed in Asic1a_+/+_ males and Asic1a_-/-_ females (Asic1a_+/+_ males: t(44.0) = −1.92, p = 0.062; Asic1a_-/-_ females: t(44.1) = −1.79, p = 0.080) and no statistical difference was observed in Asic1a_-/-_ males (t(36.5) = 0.15, p = 0.882). At all other ages mice of both sexes and genotypes showed freezing above baseline levels when tested 48 h after training, indicating that they retained contextual memory (Supplementary Table 4).

Loss of ASIC1A has age-specific effects on contextual fear memory tested at 16 days after training

Asic1a_-/-_ and Asic1a_+/+_ mice were tested for the stability of contextual fear memory at 16 days post-training (D16). Asic1a_-/-_ mice trained at P17, P23 or P83 exhibited reduced context-evoked freezing when tested 16 days after training (P17: t(31.7) = 2.185, p = 0.036; P23: (t(42.0) = 3.095, p = 0.003; P83: t(30.0) = 6.415, p < 0.001) (Fig. 3A, B, F, G, Table 1). However, Asic1a_-/-_ mice trained at P29 showed no significant difference in freezing compared to age-matched Asic1a_+/+_ mice at 16 days after training (t(6.5) = 1.721, p = 0.132) (Fig. 3C, Table 1), which contrasts with the reduced, context-evoked freezing this age group showed at 48 h after training. Asic1a_-/-_ mice trained at P42 showed a strong trend towards lower freezing compared to Asic1a_+/+_ mice trained at the same age (t(6.3) = 2.414, p = 0.051) (Fig. 3D, E, Table 1). We saw a trend for lower freezing with increasing age at training (t(107.5) = −1.697, p = 0.093) and a significant interaction between age at training and genotype (t(60.1) = 2.13, p = 0.037), indicating lower freezing in Asic1a_-/-_ mice with increasing age at training upon re-exposure to the training context at 16 days after training (D16) (Supplementary Fig. 7).

In Asic1a_+/+_ mice trained at P17, context-evoked freezing declined significantly between 48 h and 16 days after training (t(19.9) = −3.795, p = 0.001) (Fig. 3A). Asic1a_-/-_ mice trained at P17 trended in the same direction (t(18.7) = −1.883, p = 0.075). Further analysis showed no genotype effect on the extent of this decline (t(30.3) = 1.593, p = 0.122). In mice trained at P23, context-evoked freezing between 48 h and 16 days after training significantly declined in Asic1a_+/+_ mice (t(6.37) = −2.759, p = 0.031) (Fig. 3B) and we saw a trend for a decline of context-evoked freezing in Asic1a_-/-_ mice (t(18.4) = −0.059, p = 0.095). There was also a trend for a genotype effect on the extent of decline of context-evoked freezing (t(12) = 1.802, p = 0.097) between 48 h and 16 days after training. In mice trained at P29, context-evoked freezing between 48 h and 16 days after training declined significantly in both genotypes (Asic1a_+/+: t(6.5) = −3.390, p = 0.013; Asic1a-/-: t(6.4) = −2.602, p = 0.038) (Fig. 3C), with no genotype difference in the extent of decline (t(6.5) = 0.280, p = 0.788). In mice trained at P42, we saw a trend towards a genotype-by-sex interaction for the change in context-evoked freezing between 48 h and 16 days after training (t(29.8) = −1.797, p = 0.083). Follow-up analysis on the stability of context-evoked freezing in each sex did not reveal a genotype effect (males: t(11.3) = 1.216, p = 0.249; females: t(10.5) = −1.036, p = 0.323) (Fig. 3D, E). Given this lack of effect, we also assessed both sexes together and found no significant change in freezing between genotypes between 48 h and 16 days after training (t(4.5) = 0.120, p = 0.910) and within-genotype freezing between those timepoints also remained stable (Asic1a+/+_ mice: t(5.45) = −1.453, p = 0.201); Asic1a_-/-_ mice t(3.52) = −1.548, p = 0.201). In mice trained at P83, though there was a sex-by-genotype interaction we saw for freezing at 48 h after training, we found no sex-by-genotype interaction for the stability of context-evoked freezing between 48 h and 16 days after training. There was no significant change in freezing between 48 h and 16 days after training in Asic1a_+/+_ versus Asic1a_-/-_ mice (t(4.9) = −0.226, p = 0.803) and within-genotype freezing between those timepoints also remained stable (Asic1a_+/+_ mice: t(3.67) = −1.427, p = 0.233); Asic1a_-/-_ mice t(7.58) = −1.989, p = 0.084). Due to the sex-by-genotype interaction at 48 h, freezing at 48 h. and D16 are visualized separately for each sex for mice trained at P42 and at P83.