Npas2 cKO and lighting condition have effects on long-term recognition memory
Novel object recognition (NOR) tasks were performed to assess whether NPAS2 loss in the liver affected cognitive function.
Memory performance was measured by discrimination index (DI), which either showed a positive preference for novelty, or a negative preference for familiarity. Mixed-effects analysis revealed a significant effect of genotype (F(1,80 = 6.141, p = 0.0153) and a significant interaction between genotype and lighting condition (F(2,80) = 6.548, p = 0.0023) (Fig. 1A). DI’s revealed almost identical preferences for novelty between cKO and fl/fl controls in LD and DD, which were also comparable across both lighting conditions, however, cKO showed a significantly higher mean DI than fl/fl controls in LL, (Šídák’s post-hoc, p = 0.0001) (Fig. 1A) which is indicative of superior recognition memory. Intra-genotype differences revealed no significant difference in fl/fl memory performance across lighting conditions but did show that cKO had a significantly higher DI in LL than in DD (Tukey’s post-hoc, p = 0.0394) and LD (Tukey’s post-hoc, p = 0.0181) (Fig. 1A). Mean total interaction time, where interaction with both the novel and the familiar object were combined, revealed no significant differences in total test interactivity between genotypes, though a significant effect of lighting condition was seen (F(1,29) = 4.194, p = 0.0497) (Fig. 1B).
Both cKO and fl/fl mice displayed significantly more interaction with the novel object than the familiar object in all conditions (LD: Šídák’s post-hoc, fl/fl: p = 0.0034, cKO: p = 0.0003, DD: Šídák’s post-hoc, both: p < 0.0001, LL: Šídák’s post-hoc, fl/fl: p = 0.0008, cKO: p = 0.0226) (Fig. 1C, D and E). This increased interaction with the novel object was significantly above chance in both genotypes, under all lighting conditions (LD: fl/fl t(15) = 5.893, p < 0.0001, cKO t(13) = 6.103, p < 0.0001), (DD: fl/fl t(14) = 7.554, p < 0.0001, cKO t(12) = 7.342, p < 0.0001), (LL: fl/fl t(13) = 2.837, p = 0.0140, cKO t(13) = 10.50, p < 0.0001). No spatial bias or object preferences were noted in sample trials (Figure S2). Taken together these data suggest a light condition specific improvement of long-term recognition memory performance in NPAS2 cKO mice, that is unlikely to be explained by increased total object interaction time.
Locomotor activity and circadian rhythmicity was not altered in hepatic Npas2 -/- mice
To assess whether differences in recognition memory performance could be confounded by altered general locomotor activity, distance travelled and activity patterns were measured between the two genotypes. Locomotor activity recordings of distance travelled were taken in 10-minute bins during all three lighting treatments. No significant differences were noted in the average hourly activity patterns of each genotype in LD, DD or LL conditions. In LD, for both fl/fl controls and cKO, onset of activity begun almost as soon as the lights were turned off at ZT12 and activity ended immediately after the lights were turned back on at ZT0. Locomotor activity in both the light phase and the dark phase was comparable between the same sex of opposing genotypes. Though, both female groups demonstrated increased locomotor activity compared to males throughout (Fig. 2A). Following the onset of DD, all groups became generally more active across a 24-hour period, with activity dispersing into the subjective day. Here, fl/fl females displayed increased distance travelled in comparison to cKO females (Tukey’s post-hoc, p = 0.0470) (Fig. 2B). After the induction of LL, fl/fl females displayed reduced distance travelled in comparison to DD (Tukey’s post-hoc, p < 0.001), yet this effect was not seen in cKO females or males of either genotype (Fig. 2C and D). Three-way ANOVA analysis revealed a significant effect of lighting condition on locomotor activity (F(2, 138) = 7.875, p = 0.0006) (Fig. 2C and D). Additionally, there was a significant genotype X sex X lighting condition interaction(F(2, 138) = 3.373, p = 0.0371) and a significant effect of sex alone (F(1, 138) = 101.5, p < 0.0001). Genotype did not have a significant effect on locomotor activity. Actograms generated using this distance travelled data can be found in the supplementary data (Figure S4, S5, S6, S7).
No significant differences were seen in circadian period, phase or amplitude in hepatic Npas2 -/- mice
Locomotor activity recordings were then examined using FFT-NLLS cosine analysis to yield period, phase and amplitude estimates. This was to establish whether poorer recognition memory performance in cKO mice during LL conditions could be explained by altered circadian rhythmicity. Mixed-effects analysis revealed lighting condition significantly impacted circadian period (F(2,87) = 24.02, p < 0.0001), but genotype did not (F(1,87) = 0.1936, p = 0.6610). DD shortened circadian period in both genotypes (Šídák’s post-hoc, fl/fl: p = 0.0003, cKO: p = 0.0107) and LL elongated circadian period in both genotypes (Šídák’s post-hoc, fl/fl: p = 0.0002, cKO: p = 0.0049) (Fig. 3A).
FFT-NLLS cosine analysis also revealed that circadian phase was significantly influenced by lighting condition (mixed effects analysis, F(1.575,44.89) = 61.82, p < 0.0001)), but not by genotype (F(1,30) = 0.1896, p = 0.6664). Both genotypes demonstrated a significant phase advance in DD (Šídák’s post-hoc, fl/fl: p = 0.0002, cKO: p < 0.0001), and a significant phase delay in LL (Šídák’s post-hoc, both p < 0.0001) (Fig. 3b). Similarly, rhythm amplitude was significantly impacted by lighting condition (mixed effects analysis, F(2,60) = 9.047, p = 0.0004), but not by genotype (F(1,30) = 0.8636, p = 0.3604). Both fl/fl and cKO mice demonstrated reduced amplitude in LL compared to LD (Šídák’s post-hoc, fl/fl: p = 0.0031, cKO: p = 0.0475) (Fig. 3c). Thus, all core circadian parameters remain unchanged despite the loss of NPAS2 in the liver, suggesting hepatic NPAS2 is not required for global rhythmicity, as hypothesized.
Anxiety phenotypes in hepatic Npas2 -/- mice
Next, to assess if recognition memory differences between fl/fl and cKO mice may be due to an anxiety phenotype following hepatic NPAS2 loss as has been demonstrated previously in a global Npas2-/- model, both groups were subjected to open field arena and light-dark box anxiety trials. No anxiety phenotype was seen in any parameter in either anxiety paradigm (Figures S5, S6).
Genotype variation in recognition memory and sexual dimorphism in fl/fl control mice
While previous literature is variable, some studies have shown object recognition memory can vary by sex18,19,20. Since variation in recognition memory performance in LL between genotypes was not explained by altered anxiety phenotypes, or activity levels, an exploratory analysis of sex on DI was performed. A significant interaction sex X genotype X lighting condition interaction was seen on DI (mixed effects analysis, F(2,74) = 4.256, p = 0.0178). There were also significant lighting condition X sex (F(2,74) = 3.442, p = 0.0372), lighting X genotype (F(2,74) = 7.804, p = 0.0008) interactions and an effect of genotype (F(1,74) = 7.463, p = 0.0079). Notably, post-hoc analysis revealed a significant difference in LL DI between cKO females and fl/fl females, where cKO females showed significantly better recognition memory (Tukey’s post-hoc, p = 0.0153) (Fig. 4A).
In addition, total interaction time was also assessed for an impact of sex. There was no significant interaction between sex, genotype and lighting condition (F(2,47) = 0.4964, p = 0.4399). However, an effect of sex was seen (F(1,28) = 11.03, p = 0.0025). Males appeared to interact with the novel object test more than females. Despite this, no individual effects in any lighting condition were seen between sexes, and no effect of genotype was noted (Fig. 4B). Furthermore, no differences in total test interaction were observed between cKO females and fl/fl females in any lighting condition. Thus, any variance in memory performance between cKO and fl/fl females is unlikely to be attributable to altered engagement with the task.
Sexual dimorphism in circadian period in conditions of constant light
Circadian parameters were also then analysed to assess whether recognition memory changes observed may be due to sex specific changes in circadian period and phase. No interaction effect of sex X genotype X lighting condition was seen. However, there was and a significant effect of lighting condition on circadian period (F(1.492, 61.16) = 28.44, p < 0.0001) and a significant interaction between lighting condition and sex (mixed effects analysis, F(2,82) = 8.106, p = 0.0006). Most notably, period was increased in females compared to males irrespective of genotype (Fig. 5A).
An effect of lighting condition was found on phase (mixed effects analysis, F(1.588, 42.88) = 54.15, p < 0.0001), but there was no effect of sex or interactions with sex (Fig. 5B).
Central circadian gene expression in hepatic Npas2 -/- mice
Next in order to examine potential mechanisms of NOR effects identified transcript levels of core circadian genes in the frontal cortex were analysed between genotypes and sex groups using RT-qPCR. This was measured at ZT0 and ZT12, to additionally assess if cyclical gene expression was maintained in hepatic Npas2 cKO mice. Previous findings in global Npas2 KO mice have noted dampened oscillatory gene expression in some circadian targets1,2. A significant effect of time point X genotype X sex was seen on frontal cortical Clock expression (Three-way ANOVA, F(1,40) = 6.535, p = 0.0145), as well as a significant interaction between time point and sex (F(1,40) = 6.434, p = 0.0152). Clock expression was seen to be antiphase between fl/fl females and fl/fl males. Fl/fl males had a significantly higher central Clock gene expression at ZT12 than fl/fl females (Tukey’s post-hoc, p = 0.008). In addition, reduced expression was seen in both cKO sexes, and a significant effect of genotype was noted (F(1,40) = 8.161, p = 0.0068) (Fig. 6A).
Frontal cortical expression of Bmal1 at ZT0 and ZT12 revealed significant effects of sex (Three-way ANOVA, F(1,39) = 9.345, p = 0.0040) and time point (F(1,39) = 8.493, p = 0.0059), as well as a significant interaction between sex and time point (F(1,39) = 13.37, p = 0.0008). As with frontal cortical Clock expression, Bmal1 expression appeared to be antiphase between fl/fl males and fl/fl females. At ZT12, central Bmal1 expression was significantly higher in fl/fl males compared to fl/fl females (Tukey’s post-hoc, p = 0.0027). cKO males displayed a similar pattern of Bmal1 expression as fl/fl males, peaking at ZT12, but fl/fl females appeared to show no changes in expression over the two time points (Fig. 6B).
Finally, frontal cortical expression of Rev-erbβ at ZT0 and ZT12 was assessed. All genotype and sex groups displayed in phase Rev-erbβ expression, with peaks at ZT12. A significant effect of time point was noted (Three-way ANOVA, F(1,44) = 6.956, p = 0.0115), as well as a significant effect of sex (F(1,44) = 5.957, p = 0.0187). Importantly, a significant interaction between sex and time point was seen (F(1,44) = 5.453, p = 0.0242). Both cKO and fl/fl females display reduced ZT12 Rev-erbβ expression compared to male counterparts (Tukey’s post-hoc, fl/fl: p = 0.0037) (Fig. 6C). This suggests males have stronger Rev-erbβ oscillatory gene expression. No effect of genotype was noted on Rev-erbβ expression, though fl/fl males demonstrated higher levels at ZT12 than cKO males (Tukey’s post-hoc, p = 0.0134).
