Cognitive Deficits in Female GPR158 Knockout Mice Revealed by Automated Home‐Cage Testing

1. Introduction

GPR158 is a G protein coupled receptor (GPCR) that has been heavily implicated in cognition and affective states. GPR158 is found highly expressed in the central nervous system (CNS), including in the prefrontal cortex, nucleus accumbens, and hippocampus [1, 2, 3]. With its high expression in the brain, GPR158 has been shown to mediate synaptic transmission, neuronal excitability, and structural plasticity [4, 5]. This has been closely examined within the hippocampus where GPR158 acts as a specific synaptic organizer within hippocampal neurocircuits [6]. In the CA3 region, GPR158 plays a role in establishing synaptic functioning of developing mossy fiber‐CA3 synapses. It was shown that the loss of GPR158 not only disrupts morphology but also reduces synaptic strength at this synapse. GPR158 expression in the CA1 pyramidal neurons has been shown to regulate intrinsic excitability, and decreases in GPR158 levels disrupt dendritic architecture leading to reduced dendritic length and branching complexity [7]. GPR158 plays a role in spatial learning as mice deficient in GPR158 have been shown to display impairments in the acquisition and retrieval of spatial memory in the Morris Water Maze (MWM) test and show deficits in the novel object recognition task (NOR) [7, 8] As a major component for establishing hippocampal connections, GPR158 may also play a role in other hippocampal‐dependent learning tasks.

Studies have also implicated GPR158 signaling in stress‐induced depression, and maintenance of social interaction [8, 9, 10]. GPR158 levels are directly regulated by stress exposure via a corticosteroid‐mediated mechanism due to the presence of glucocorticoid response elements in the promoter region of GPR158 [11]. Mice subjected to chronic stress paradigms such as unpredictable chronic mild stress (UCMS) and physical resistant stress (PRS) exhibit an upregulation of GPR158 levels in the prefrontal cortex [3]. Elevated GPR158 levels have also been observed in the hippocampus after the stress of an immediate shock [12]. Increased GPR158 levels have been shown to influence various behavioral responses including depressive‐like responses in the tail suspension test and forced swim test, whereas GPR158 knockout (KO) mice display antidepressive and anxiolytic behaviors [3].

Stress exposure can lead to both adaptive and maladaptive responses and can affect cognitive performance based on the extent and duration of the stressor. Many cognitive tests for rodents induce stress intrinsically from the test itself or extrinsically from experimenter intervention such as repetitive experimenter handling of mice, new environmental factors, testing during rest period, and single‐housed animals [13, 14]. In addition, traditional behavioral testing often necessitates the isolation of individuals, which has been shown to decrease spatial learning, synaptic plasticity, and hippocampal neurogenesis [15]. Social interaction in a group setting has been found to alleviate stress‐induced cognitive and memory impairments in response to restraint stress and fear conditioning [16, 17]. Given the direct involvement of GPR158 in stress responses, it is unclear if this effect influences learning and memory processes.

Given the implication of GPR158 in cognitive processing, this study will further characterize GPR158 knockout mice in a series of learning and memory paradigms. As stress exposure and sociability can affect cognitive ability, we opted to use an automated home cage that monitors group‐housed behaviors. Testing within a home cage environment in a group setting has been achievable as each mouse is implanted with a radio frequency identification (RFID) and certain activities can be continuously monitored long‐term. In this report, we examine the effect of GPR158 on spatial learning and memory, cognitive flexibility, and working memory in a longitudinal, social, and minimally stress‐inducing environment to assess a more naturalized role of GPR158 in cognition.

2. Materials and Methods

3. Results

We subjected female and male GPR158^−/− and GPR158^+/+ mice to the Barnes Maze test to assess spatial learning and memory. During the 16 acquisition trials (with 4 trials/day), mice learned to locate and enter the escape tunnel located below the target hole on the platform. During acquisition training, female mice showed a significant effect over the trial sessions, indicating that latencies decreased over training sessions (Figure 1A; trial F_{(15, 300)} = 11.65, p < 0.0001). However, female GPR158^−/− mice were found to have a longer latency to enter the escape tunnel during the initial trials than GPR158^+/+ mice (genotype F _{(1, 20)} = 1.280, p = 0.2714; trial × genotype F _{(15, 300)} = 1.865, p < 0.05). Female GPR158^−/− mice spent significantly less time in the reward area compared to GPR158^+/+ mice (Figure 1B; day F _{(3, 60)} = 18.59, p < 0.0001; genotype F _{(1, 20)} = 4.383, p < 0.05; day × genotype F _{(3, 60)} = 0.3926, p = 0.758). In contrast, male GPR158^+/+ and GPR158^−/− mice showed similar latency times during the acquisition trials of the Barnes Maze (Figure 1C; day F _{(5, 315)} = 3.817, p < 0.0001; genotype F _{(1, 21)} = 0.07028, p = 0.79; day × genotype F _{(15, 315)} = 1.333, p = 0.1807) and time spent in the reward area (Figure 1D; day F _{(3, 69)} = 16.64, p < 0.0001; genotype F _{(1, 23)} = 1.586, p = 0.2205; day × genotype F _{(3, 69)} = 0.4807, p = 0.697). Over this acquisition period, the traces revealed that GPR158^+/+ and GPR158^−/− mice used different strategies to find the escape tunnel (Figure 1E), and we therefore analyzed the cognitive performance by identifying whether mice learned to use a spatial strategy approach comparing both male and female cohorts. Over the acquisition period, the traces revealed that male and female GPR158^+/+ mice showed the development of spatial strategies with increasing cognitive performance, whereas female and male GPR158^−/− mice's performance did not change over the four‐day period (Figure 1F; day F _{(3, 723)} = 95.63, p < 0.0001; genotype F _{(1, 45)} = 6.68, p = 0.013; sex F _{(1, 45)} = 0.04, p = 0.84; sex × genotype interaction F _{(1, 45)} = 2.80, p = 0.100). During the probe trial, the escape tunnel was removed and the time spent in the reward quarter was calculated. We found that female GPR158^−/− mice spent significantly less time in the reward quarter compared to female GPR158^+/+ mice, whereas there was no difference between male GPR158^+/+ and GPR158^−/− mice (Figure 1G; genotype F _{(1, 45)} = 7.94, p = 0.007; sex F _{(1, 45)} = 1.54, p = 0.022; genotype × sex F _{(1, 45)} = 2.97, p = 0.0914).

We monitored activity from group‐housed GPR158^−/− and GPR158^+/+ mice using IntelliCages under a schedule of free adaptation, nose‐poke adaptation, place learning and reversal followed by a patrolling paradigm (Figure 2). Activity was recorded for each mouse by a RFID transponder. Initially, we assessed the free adaptation period where mice could move freely for 3 days and had access to all 8 water bottles. As expected, peaks of activity were observed during the dark periods of female and male GPR158^−/− and GPR158^+/+ mice with minimal activity occurring during the light periods (Figure 3A). We observed that the number of corner visits decreased over the days for each group, showing habituation to their new environment regardless of sex or genotype (Figure 3B; day F _{(2, 80)} = 58.06, p < 0.0001, genotype, F _{(1, 40)} = 0.6092, p = 0.4397; sex F _{(1, 40)} = 0.091, p = 0.765; sex × genotype F _{(1, 40)} = 0.032, p = 0.8578). The total number of licks over this period did not change with no effect of sex but a significant effect of genotype was identified (Figure 3C; day F _{(2, 80)} = 2.485, p = 0.0897; sex F _{(1, 40)} = 3.010, p = 0.09; genotype F _{(1, 40)} = 4.517, p = 0.0398). Further comparison revealed that male GPR158^+/+ mice had a significantly higher number of licks than male GPR158^−/− mice (p < 0.0045). For the latency to visit the first corner, there is a significant effect of time but not an effect of genotype or sex (Figure 3D, day F _{(2, 80)} = 7.087, p < 0.001; genotype F _{(1, 40)} = 0.269, p = 0.607; sex F _{(1, 40)} = 1.958, p = 0.169; sex × genotype F _{(1, 40)} = 0.5283, p = 0.4716). Overall, the data from the free adaptation phase show that both GPR158^+/+ and GPR158^−/− mice habituate to the new environment over the 3‐day exposure.

Following the free adaptation phase, mice were exposed to 4 days of nose‐poke adaptation where mice learned to perform a nose‐poke to gain access to the water bottles. For the number of visits, there was a significant effect of day with no effect of sex or genotype (Figure 4A; day F _{(3, 120)} = 9.181, p < 0.001; genotype F _{(1, 40)} = 1.196, p = 0.2806; sex F _{(1, 40)} = 0.1129, p = 0.7386; sex × genotype F _{(1, 40)} = 4.155, p = 0.0482). There is a decrease in the nose‐pokes over time with no effect of sex or genotype as each group decreased the number of nose‐pokes during these sessions (Figure 4B; day F _{(3, 120)} = 12.40, p < 0.0001; genotype F _{(1, 40)} = 1.557, p = 0.2194; sex F _{(1, 40)} = 0.2442, p = 0.623). Across the 4‐day session, lick counts did not vary significantly with day (Figure 4C; day F _{(3, 120)} = 0.8278, p = 0.481) but there were robust main effects of sex and genotype (sex F _(1,40) = 28.53, p < 0.001, genotype F _{(1, 40)} = 6.614, p = 0.013). Post hoc comparison showed that male GPR158^+/+ mice exhibited a higher number of licks than male GPR158^−/−, indicating that the genotype effect on licking behavior was driven largely by increased licking in wildtype males. Altogether, both GPR158^+/+ and GPR158^−/− learned to nose‐poke to gain access to water.

Next, we tested the mice in a place learning paradigm where mice are assigned access to two corners (correct corners) in which they could access water in response to a consecutive 5 s nose‐poke. The other corners are accessible for a mouse to visit but will not have access to the bottles (incorrect corners). The place error was analyzed using a three‐way repeated‐measures ANOVA with days as a within‐subject factor and sex and genotype between‐subject factors. There was a significant main effect of day (F _{(3, 120)} = 18.72, p < 0.001) indicating that error rates changed across sessions (Figure 5A). The main effects of sex and genotype were not significant (sex F _(1,40) = 1.09, p < 0.30; genotype F _{(1, 40)} = 0.54, p = 0.47) but a significant sex × genotype interaction was detected (F _(1,40) = 4.37, p = 0.043). Comparison within females showed a difference between the genotypes (p < 0.05), where the incorrect response did not change over time for GPR158^−/− mice but a decrease was observed for GPR158^+/+ mice. Both male GPR158^+/+ and GPR158^−/− displayed a decrease in the incorrect responses over time. The total number of licks over this period did not change (Figure 5B; F _{(3, 120)} = 2.426, p = 0.07). While there was no effect of sex, a significant effect of genotype was identified (sex F _{(1, 40)} = 13.95, p = 0.0006; genotype F _{(1, 40)} = 3.070, p = 0.0874). Further comparison revealed that male GPR158^+/+ mice had a significantly higher number of licks during place preference than GPR158^−/− mice (p = 0.0045). We analyzed the behavioral transition categories of correct–correct (C–C) nose‐pokes over incorrect–incorrect (I–I) or incorrect–correct/correct–incorrect (C–I) nose‐pokes during initial and final days (Figure 5C). The ART repeated‐measures analysis revealed a highly significant behavioral transition category (F _{(2, 216, 3)} = 205.01, p < 2.2 × 10⁻¹⁶), indicating that the mean numbers of transitions differed significantly among the three behavioral transition categories (C–C, C–I, and I–I). During the initial phase, mice exhibited the greatest mean number of correct‐to‐incorrect (C–I) transitions (mean = 64.2), followed by consecutive correct (C–C) transitions (mean = 27.2) and consecutive incorrect (I–I) transitions (mean = 20.7). A significant behavioral transition category × phase interaction was also detected (F _{(2, 216, 3)} = 7.13, p = 0.001), indicating that the pattern of behavioral transitions changed between the initial and final phases. Specifically, the mean number of consecutive correct transitions (C–C) increased from 27.2 during the initial phase to 34.6 during the final phase (7.4 transitions), whereas correct‐to‐incorrect transitions (C–I) decreased from 64.2 to 55.8 (−8.4 transitions) and consecutive incorrect transitions (I–I) decreased from 20.7 to 14.3 (−6.4 transitions). Together, these changes indicate a shift toward more consistent correct spatial choices over time. No significant main effects of sex, genotype, or phase were detected. However, a significant sex × genotype interaction was observed (F _{(1, 169, 5)} = 4.77, p = 0.030), indicating that the influence of genotype differed between males and females. To identify the source of this interaction, we performed post hoc comparisons using estimated marginal means from the ART model to compare genotypes within each sex and sexes within each genotype. These analyses showed that the overall pattern of behavioral transitions differed significantly between female GPR158^−/− and GPR158^+/+ mice (p = 0.024), whereas no genotype difference was detected among males (p = 0.366). Similarly, the overall pattern of behavioral transitions differed significantly between female and male GPR158^−/− mice (p = 0.008), whereas no significant sex difference was observed among GPR158^+/+ mice (p = 0.444). Although the sex × genotype × behavioral transition category interaction approached significance (p = 0.064), neither this nor any higher‐order interaction involving phase reached statistical significance.

The reverse period immediately followed where the water bottles that were originally accessible were inaccessible, and the inaccessible corners were accessible. We analyzed the place error and found a main effect of day (Figure 5D; F _(3,120) = 14.97, p < 0.0001), indicating a change in performance across days. There were no significant effects of sex or genotype. Comparison of females revealed a significant interaction of time x genotype (p < 0.05) in which the percentage of incorrect responses did not change over days for GPR158^−/− mice but GPR158^+/+ mice had fewer incorrect responses during the reversal period. There was no significant difference in the percentage of incorrect responses between male GPR158^+/+ and GPR158^−/− mice as both groups exhibited a decrease in the number of incorrect responses (p > 0.05). The total number of licks over this period did not change but a significant effect of sex was found (Figure 5E, day F _{(3, 120)} = 1.99, p = 0.1191; sex F _{(1, 40)} = 12.60, p = 0.001). Male GPR158^+/+ mice had significantly more licks than GPR158^−/− mice (p < 0.05). Next, we assessed the change in responses during the initial and final phase of the reversal sessions (Figure 5F). The ART repeated‐measures analysis revealed a highly significant main effect of behavioral transition category (F _{(2, 216, 2)} = 267.66, p < 2.2 × 10⁻¹⁶), indicating that the mean numbers of transitions differed significantly among the three behavioral transition categories (C–C, C–I, and I–I). A significant behavioral transition category × phase interaction was also detected (F _(2,216,3) = 11.21, p = 2.34 × 10⁻⁵), indicating that the pattern of behavioral transitions changed between the initial and final phases. Specifically, the mean number of C–C increased from 27.4 during the initial phase to 34.6 during the final phase (7.2 transitions), whereas C–I decreased from 65.1 to 59.6 (−5.5 transitions) and I‐I decreased from 22.8 to 15.7 (−7.1 transitions). Together, these changes indicate a shift toward more consistent correct spatial choices over time. Although there were no significant main effects of sex, genotype, or phase, there were significant sex × genotype (F _{(1, 172, 4)} = 8.22, p = 0.0047) and sex × genotype × category (F _{(2, 216, 3)} = 4.47, p = 0.0125) interactions. Post hoc comparisons comparing genotypes separately showed that the transitions differed between female GPR158^−/− and GPR158^+/+ mice (p < 0.05) with wildtype mice exhibiting more C–C responses than knockout counterparts.

A patrolling paradigm was used to evaluate working memory, where the correct corner was programed to switch in a clockwise position with every correct response. Analyzing the place error, we found a significant main effect of time (F _{(2, 80)} = 21.38, p < 0.001) with no significant effect on sex or genotype (Figure 6A). We analyzed the change in behavioral categories (C–C, C–I, I–I) from Day 1 and Day 3 and found a significant effect in transition categories (Figure 6B; F _{(2, 216,1)} = 274.43, p < 2.2 × 10⁻¹⁶). There was only a slight change in C–C response from a mean of 9.6 to 10.2 (0.6 transitions), whereas C–I decreased from 51 during the initial phase to 44.8 during the final phase (−6.2 transitions), and I‐I response also decreased from a mean of 63.4 to 44.4 (−19 transitions). Post hoc comparison showed that each group's incorrect responses decreased over time (p < 0.05). Together, these results indicate a reduction in incorrect transition behaviors over the patrolling protocol.

4. Discussion

In our study we show that GPR158 knockout mice display deficits in spatial learning and cognitive flexibility using both standard and automated phenotyping paradigms. In the Barnes Maze test, both male and female GPR158^−/− mice showed deficits in the spatial strategies, with female GPR158^−/− showing more pronounced deficits as their latency to the tunnel, time spent in the RA, and time in reward quarters all showed deficits. Using an automated homecage system, we conducted a longitudinal study of group‐housed mice to monitor four different behavioral phases all in their homecage environment; free adaptation, nose‐poke adaptation, place preference, and patrolling. During the free adaptation period, GPR158^+/+ and GPR158^−/− exhibit similar levels of habituation and exploratory behavior in both male and female cohorts. There was also no difference between genotypes during the nose‐poke adaptation period. In the place preference test, we found that only female GPR158^−/− mice showed impaired spatial learning. Female GPR158^+/+ mice displayed significantly fewer incorrect responses over both the preference and reversal periods, whereas GPR158^−/− mice showed no change in the number of incorrect responses across these phases. In contrast, both male GPR158^+/+ and GPR158^−/− mice exhibit comparable place preference learning with a decreasing percentage of incorrect responses over time. Overall, female GPR158^−/− mice showed deficits in spatial learning regardless of whether testing occurs in isolation or a group environment.

Several studies have shown that GPR158^−/− mice show deficits in learning and memory paradigms as assessed with standard behavioral methods, including the Morris Water Maze and a traditional Barnes Maze [8, 24]. In addition, GPR158^−/− mice show deficits in object recognition, displaying difficulty distinguishing between novel and familiar objects in the novel object recognition test, as well as aberrant responses in the passive avoidance task [6]. Many of these studies have been conducted in only one sex, and comparative analysis between males and females has not been performed. While we found deficits in the spatial strategies that were used in the Barnes Maze for both male and female GPR158^−/− mice, only female knockout mice showed deficits in the place learning and reversal in the IntelliCage. This describes a sex difference between males and females GPR158^−/− mice towards spatial learning when testing in a group environment. Spatial learning tasks in the IntelliCage provide a less stressful testing environment with social enrichment [25]. Mice maintained under an enriched environment have shown increased performance of hippocampus‐dependent spatial working memory tasks [26]. Environmentally enriched mice showed enhanced spatial learning and reversal learning in the Morris Water Maze task, as well as enhanced LTP in the hippocampal dentate gyrus compared to reductions found in socially isolated mice [15]. Furthermore, in group testing environments, mice have the potential to use social observation and perseveration behaviors to make spatial choices. This has been observed with aging female mice, where mice used non‐spatial strategies to find a reward (drinking) corner in the IntelliCage [27]. These findings suggest that male GPR158^−/− mice may have compensated for spatial learning deficits by adopting alternative, nonspatial behavioral strategies during place learning and reversal in the IntelliCage.

The effect of GPR158 signaling on learning has been speculated to be due to its role in the structural organization and formation of the synapse. Loss of GPR158 affects the hippocampal architecture and synaptic organization as it is a pathway‐specific organizer and has been shown to mediate hippocampal‐dependent learning [7, 8]. It has been shown that GPR158^−/− mice have reductions in dendrite architecture, complexity, and electrophysiological properties and negatively affects hippocampal‐dependent learning in a Morris water maze [6, 8]. Similar to our observed deficits in spatial learning tasks, these deficits in spatial learning are likely contributed to by aberrant neurocircuitry in the hippocampus of the GPR158^−/− mice.

From a clinical perspective, GPR158 has been associated with Alzheimer's disease with studies demonstrating disruptive GPR158 levels in both the cortex of patients with Alzheimer's disease and animal models [28, 29, 30]. While these studies have shown an association, it has been difficult to investigate the signaling mechanism until recently as GPR158 has been classified as an orphan receptor. Glycine has been proposed to be the endogenous ligand for GPR158, demonstrating it as the metabotropic glycine receptor for this amino acid [31]. Activation of GPR158 by glycine has been shown to enhance the intrinsic excitability of medium spiny neurons in the striatum and in cortical neurons [31, 32] Besides glycine, osteocalcin, a bone‐derived hormone, has also been proposed as an endogenous ligand [8]. Interestingly, reduced levels of osteocalcin have been documented in cerebrospinal fluid of Alzheimer's disease patients, and in the aging population associated with cognitive deterioration, as well as mice deficient in osteocalcin exhibited impaired spatial learning and memory dependent on the hippocampus [33, 34, 35] Understanding how GPR158 is activated by these specific ligands and how it modulates spatial learning requires further exploration.

5. Conclusions

Together this study has demonstrated the value of an automated group home cage to characterize mouse models whose behavioral output may be sensitive to stress, environmental conditions, and social behaviors. Standard behavioral testing may induce multiple confounds and is only able to provide snapshots of individual behavior in a novel, unfamiliar environment. This longitudinal study minimizing these confounds in a social setting is necessary to illuminate a more naturalized role of GPR158 in cognition and may provide valuable information on the impact of confounds induced by traditional behavioral testing methods. By mitigating stress and social isolation during behavioral testing we have shown that the spatial deficits observed in female GPR158^−/− mice are due to cognitive deficits.

Funding

This work was supported by a startup fund from the University of Maryland Baltimore County.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.