A comparative analysis of kainate receptor GluK2 and GluK5 knockout mice in a pure genetic background

Kainate receptors (KARs) are members of the glutamate receptor family that regulate synaptic function in the brain. Although they are known to be associated with psychiatric disorders, how they are involved in these disorders remains unclear. KARs are tetrameric channels assembled from a combination of GluK1-5 subunits. Among these, GluK2 and GluK5 subunits are the major heteromeric subunits in the brain. To determine the functional similarities and differences between GluK2 and GluK5 subunits, we generated GluK2 KO and GluK5 KO mice on a C57BL/6 N background, a well-characterized inbred strain, and compared their behavioral phenotypes. We found that GluK2 KO and GluK5 KO mice exhibited the same phenotypes in many tests, such as reduced locomotor activity, impaired motor function, and enhanced depressive-like behavior. No change was observed in motor learning, anxiety-like behavior, or sociability. Additionally, we identified subunit-specific phenotypes, such as reduced motivation toward their environment in GluK2 KO mice and an enhancement in the contextual memory in GluK5 KO mice. These results revealed that GluK2 and GluK5 subunits not only function in a coordinated manner but also have a subunit-specific role in regulating behavior. To summarize, we demonstrated subunit-specific and common behavioral effects of GluK2 and GluK5 subunits for the first time. Moreover, to the best of our knowledge, this is the first evidence of the involvement of the GluK5 subunit in the expression of depressive-like behavior and contextual memory, which strongly indicates its role in psychiatric disorders.


Introduction
Kainate receptors (KARs) are members of the ionotropic glutamate receptor family, that are widely expressed in the central nervous system [1,2]. It was believed for a long time that KARs contributed less toward synaptic transmission than did other ionotropic glutamate receptors, such as AMPA receptors and NMDA receptors, both of which are essential for learning and memory [3,4]. However, KARs were found to be located at both pre-and post-synaptic sites, and it is now well accepted that they regulate the release of glutamate and GABA as well as the synaptic response [5][6][7]. KARs have been linked to psychiatric disorders such as schizophrenia, bipolar disorder, major depressive disorder, and autism [2,8]. In an early pharmacological study, [ 3 H] kainic acid binding sites were measured in the post-mortem brains of schizophrenia patients, and a 25-50 % increase in [ 3 H]kainic acid binding was observed in the medial frontal regions and the areas responsible for eye-movement [9]. The expression of mRNA in KAR subunits is also reduced in the orbitofrontal cortex of patients with schizophrenia [10][11][12]. Moreover, multiple genetic studies have associated genetic variants of KAR subunits with psychiatric disorders [13][14][15][16]. Therefore, there is a growing interest in the relationship between KARs and psychiatric disorders.
GluK4-GluK5 (KA1-KA2) subunits, which exhibit significant differences in terms of their spatiotemporal expression patterns in the brain [17]. Low-affinity subunits form functional homomeric KARs [18,19], while high-affinity subunits require any of the low-affinity subunits to generate functional KARs [20]. Among the various heteromeric subunit combinations, GluK2/GluK5 KARs are known to be the most abundant KARs in the brain [17,21]. A direct association between GluK2 and GluK5 subunits has been shown in CA3 pyramidal cells by co-immunoprecipitation analysis [21], and a single-molecule imaging of fluorescence-tagged GluK2 and GluK5 subunits has revealed that they assemble with a 2:2 stoichiometry [22]. Crystal structure analysis has shown that the N-terminal domains of GluK2 and GluK5 subunits are involved in the assembly of heterodimers [23,24]. Furthermore, electrophysiological studies performed in Xenopus oocytes have shown that the whole-cell currents activated by kainate or glutamate are several times larger in GluK2/GluK5 KARs than in GluK2 homomeric KARs [25]. Studies using HEK cells have also revealed that GluK2/GluK5 KARs have a markedly higher glutamate sensitivity compared to that of GluK2 homomeric KARs and are capable of a relatively slower deactivation [26]. Therefore, the presence of the GluK5 subunit in KARs helps broaden the functional and pharmacological profile of KARs. Additionally, studies using high-affinity GluK4 and GluK5 double knockout (KO) mice revealed that the KAR-mediated excitatory postsynaptic currents (EPSCs) were lost at mossy fiber-CA3 synapses, even though low-affinity subunits were expressed [27]. Furthermore, GluK2 KO (Grik2 tm1Sfh ) mice were less sensitive to kainate and mossy fiber kainate EPSCs were absent in these mice [28]. These observations strongly support the fact that heteromeric receptors consisting of high-and low-affinity subunits are functional as well as crucial in the brain.
Previous studies using GluK2 KO mice generated from embryonic stem (ES) cells derived from the 129/SvEv strain have revealed abnormal behavioral phenotypes, and these phenotypes varied depending on how they were backcrossed [28][29][30][31]. Contrastingly, GluK5 KO (Grik5 tm1Cont ) mice have only been used for electrophysiological analysis, and no behavioral examination has been conducted [32]. In this study, we generated GluK2 KO and GluK5 KO mice in a pure C57BL/6 N genetic background, a well-characterized inbred mouse strain, and investigated the impact of each deletion (GluK2 and GluK5) on various behaviors including anxiety, sociability, depression, and learning and memory. By comparing the two genotypes, we identified subunit-specific and common behavioral effects of these subunits.

Animals
GluK2 KO and GluK5 KO mice as well as GluK2/4 double knockout (DKO) mice were generated using the ES cell line RENKA derived from the C57BL/6N strain [33]. A detailed characterization of GluK2 KO and GluK5 KO mice was performed as reported previously [17]. The mice were housed in a standardized animal room (lights on 8 a.m. to 8 p.m.; room temperature 22 ± 2 • C), with access to food and water. The experimental protocols used throughout the study were approved by the Institutional Animal Care and Use Committee of the Niigata University (SA00466) and were performed in accordance with the Japanese regulations on animal experiments. Behavioral tests were carried out in 8-to 21-week-old male wild-type (WT), GluK2 KO (Grik2 − /− ), and GluK5 KO (Grik5 − /− ) littermates generated by heterozygous breeding, and were performed during the light phase (between 10 a.m. and 5 p.m.). Mice were handled for 3 days (3 min per day) before conducting the behavioral tests. After each trial, the apparatus was cleaned with hypochlorous water to prevent a bias due to olfactory cues.

Antibody
Anti-GluK2 (Synaptic Systems, Göttingen, Germany) and anti-GluK5 (Millipore Corporation, Bedford, MA, USA) antibodies were used for both western blotting and immunohistochemistry. The specificity of these antibodies have been determined in our previous report [17].

Western blotting
The standard western blot protocol was used as described previously [35]. Protein samples were separated by SDS-polyacrylamide gel electrophoresis and transferred to a supported nitrocellulose membrane (GE Healthcare Life Sciences, USA). The membrane was stained with Ponceau S (Sigma-Aldrich, USA) for protein detection, blocked with a blocking buffer (5 % non-fat milk in Tris Buffered Saline with Tween 20) for 1 h and probed with primary antibodies against GluK2 and GluK5. Membranes were then probed with horseradish peroxidase-conjugated secondary antibodies (GE Healthcare, UK) and visualized by an enhanced chemiluminescent substrate (SuperSignal West Dura Extended Duration Substrate, Thermo Fisher Scientific). Blots were quantified using the CCD-based Amersham Imager 680 system (GE Healthcare Life Sciences) and the intensity of bands was measured using NIH ImageJ.

Fixation and the preparation of histological sections
Brains were freshly obtained under deep pentobarbital anesthesia (Maruishi Pharmaceutical, Japan) and immediately frozen in powdered dry ice for the preparation of fresh frozen sections (20 μm). Before immunohistochemistry, fresh frozen sections were air-dried and fixed by dipping in 4 % paraformaldehyde in 0.1 M phosphate buffer (pH 7.2) for 15 min.

Immunohistochemistry
All immunohistochemical incubations were performed at room temperature. The sections were incubated with 10 % normal donkey serum for 20 min, followed by an incubation with a mixture of the following primary antibodies overnight: polyclonal rabbit anti-GluK2 antibody (1 μg/mL) and polyclonal rabbit anti-GluK5 antibody (1 μg/ mL). The sections were then incubated with a mixture of Alexa Fluor 488 labeled species-specific secondary antibodies for 2 h at a dilution of 1:1000 (Thermo Fisher Scientific). Images were taken with a confocal laser-scanning microscope (Zeiss LSM710; Carl Zeiss, Oberkochen, Germany).

Open-field test
The open field test was conducted to assess the spontaneous locomotor activity. It was carried out using a method similar to that reported previously [36]. The open-field chamber was made of a square platform with 50 cm (width) × 40 cm (height) walls (O'Hara & Co. Tokyo, Japan), and illuminated with a light intensity of 100 lx. Mice were placed in a corner of the field and left for 10 min to allow free exploration. During this period, the total movement distance and the time spent in the central region were recorded and automatically calculated using Image OFCR software (O'Hara & Co., Japan; see 'Image analysis for behavioral tests').

Balance beam test
Mice were placed at the end of the 1 m beam, which was resting 50 cm above the tabletop on two poles with a block box placed at the other end of the beam to mark the finish point. The performance of each mouse on the beam was quantified by measuring the number of paw slips occurred as the mouse traversed the beam.

Rotarod test
The rotarod test was performed to evaluate the locomotor ability and motor coordination [37]. The latency to fall from a rotating rod (diameter, 30 mm; acceleration, 4-67 rpm in 3 min) was measured. The test was performed for 7 days (two trials per day), and each trial was conducted for 3 min.

Light-dark transition test
The apparatus was consisted of a cage (21 cm × 42 cm × 25 cm high) divided into two equal chambers by a black partition containing a small opening (5 cm × 3 cm high) (O'Hara & Co., Japan). One chamber was made of a white plastic and was brightly illuminated at 225 lx (light chamber), whereas the other chamber was made of a black plastic and was not illuminated (dark chamber). Mice were placed in the dark chamber and allowed to move freely between the two chambers for 10 min. The time spent in each chamber, total number of transitions, latency to the first transition from the dark to light chambers, and total distance walked were recorded automatically using Image LD software (O'Hara & Co., Japan; see 'Image analysis for behavioral tests').

Three-chambered social behavior test
Crawley's social interaction test using three chambers was performed as described previously [38]. The apparatus was comprised of a rectangular box consisting of three chambers (O'Hara & Co., Japan). Each chamber measured 20 cm × 40 cm × 22 cm, and the dividing walls had small openings (5 cm × 3 cm) to allow free access into each chamber. Data acquisition and analysis were performed automatically using Image CSI (see 'Image analysis for behavioral tests'). One day before the test, the mouse was placed in the middle chamber and allowed free exploration through the entire apparatus for 10 min. In the sociability test, a stranger mouse (C3H strain; male; purchased from Charles River Laboratories, Yokohama, Japan) that had no previous contact with the tested mice was placed in a wire cage in one of the side chambers. The test mouse was placed in the middle chamber and allowed to explore all three chambers freely for 10 min. The amount of time spent around each wire cage and the total distance traveled were recorded automatically for 10 min.

Tail suspension test
Each mouse was suspended 30 cm above the floor by the tail in a chamber made of a white plastic (31 × 41 × 41 cm; O'Hara & Co., Japan). The behavior was recorded for 5 min. Immobility was measured using Image TS software according to a certain threshold (see Section, 'Image analysis for behavioral tests'). Immobility lasting for less than 2 s was not included in the analysis.

Fear conditioning test
Each mouse was placed in a transparent acrylic chamber (33 × 25 × 28 cm) with a stainless-steel grid floor (0.2 cm diameter, spaced 0.5 cm apart; O'Hara & Co., Japan) and allowed to explore freely for 3 min. Subsequently, a 55-dB white noise, which served as the conditioned stimulus (CS), was presented for 20 s. During the last 2 s of CS presentation, a foot shock (0.2 mA, 2 s), which served as the unconditioned stimulus (US), was presented. Two more CS-US pairings were presented with an inter-stimulus interval of 40 s. Twenty-four hours after the conditioning, a contextual test was conducted in the same chamber. Forty-eight hours after the conditioning, a cued fear memory was tested in a triangular chamber (33 cm × 33 cm × 32 cm high) made of an opaque white plastic and the mice were allowed to explore freely for 1 min. Subsequently, each mouse was given CS presentation for 3 min. In each session, the percentage of freezing was calculated automatically using Image FZ software (O'Hara & Co., Japan; see 'Image analysis for behavioral tests').

Image analysis for behavioral tests
The application software used for the behavioral studies (Image OFCR, LD, CSI, TS, and FZ) was based on the public domain NIH Image program (developed by the U.S. National Institutes of Health and available at http://rsb.info.nih.gov/nih-image/) and ImageJ program (http://rsb.info.nih.gov/ij/), which were modified for each test (available through O'Hara & Co., Japan).

Statistical analysis
Statistical analyses for the behavioral studies were conducted using GraphPad Prism7 (GraphPad Software Inc.). Data were analyzed using one-way analysis of variance (ANOVA), and Tukey's multiple comparison test was performed as appropriate. Values in the graphs are expressed as means ± standard error of the mean (SEM).

Expression patterns of GluK2 and GluK5 in the mouse brain
To examine the expression of GluK2 and GluK5 proteins during brain development, we prepared forebrain lysates from mice at P0, P3, P7, 2 w, 4 w, 8 w, 3 M and 6 M. We found that both GluK2 and GluK5 subunits were expressed at all ages, and there were age-related changes in their expression levels ( Fig. 1A and B). The expression of GluK2 was gradually decreased after P7 with a peak expression at P0-P3 (Fig. 1A). By contrast, the expression of GluK5 subunits was gradually increased (F (7,16) = 4.144, p = 0.0088, P0 vs. 4 w, p = 0.0312; P0 vs. 6 M, p = 0.0438; Fig. 1B). These results indicate that the amounts of GluK2 and GluK5 proteins are differentially regulated in the developing brain. We also examined the expression of GluK2 and GluK5 in various brain regions ( Fig. 1C and D). We found that GluK2 was broadly expressed in the mouse brain with distinct expression in the cortex, hippocampus and cerebellum, while GluK5 expression levels were high in the cortex, hippocampus, and striatum, but very low in the cerebellum.

Distribution of GluK2 and GluK5 in the hippocampus
The hippocampus, which plays a vital role in learning and memory, displays a very high density of binding sites for [ 3 H]kainate [39], and KARs are found to be highly enriched on the postsynaptic side of mossy fiber-pyramidal cell synapses in the CA3 region [40][41][42]. To determine the localization of GluK2 and GluK5 subunits in the hippocampus, we performed immunohistochemical analysis. Consistent with previous reports, both GluK2 and GluK5 were expressed in the mossy fiber-recipient layer of the CA3 region, i.e., the stratum lucidum ( Fig. 2) [43]. We also found a strong expressions of GluK2 in the molecular layer of the dentate gyrus (DG) and GluK5 in the stratum lacunosum-moleculare of the CA1 region as well as the inner molecular layer of the DG. This indicates that the distribution of GluK2 and GluK5 is not completely matched in the hippocampus and DG.

Locomotor activity and motor function in GluK2 KO and GluK5 KO mice
To determine if GluK2 KO and GluK5 KO mice display different behavioral phenotypes, we performed various behavioral tests. First, we carried out an open field test to assess the motor activity and function. We found that the distance traveled by GluK2 KO and GluK5 KO mice was significantly shorter than that traveled by WT mice (WT: 3339 ± 175 cm, GluK2 KO: 2298 ± 174 cm, GluK5 KO: 2308 ± 93 cm; F (2,57) = 14.55, p < 0.0001; WT vs. GluK2 KO mice, p < 0.0001; WT vs. GluK5 KO mice, p = 0.0001; Fig. 3A). Walking speeds of the KO mice were also reduced compared to that of the WT mice (WT: 5.6 ± 0.3 cm/s; GluK2 KO: 3.9 ± 0.2 cm/s, GluK5 KO: 3.8 ± 0.2 cm/s; F (2,57) = 14.56, p < 0.0001; WT vs. GluK2 KO mice, p < 0.0001; WT vs. GluK5 KO mice, p = 0.0001; Fig. 3B). Since both GluK2 KO and GluK5 KO mice exhibited reduced motor activity and function in the open filed test, we conducted the rotarod test to measure their motor coordination. Both GluK2 KO and GluK5 KO mice showed normal motor coordination and learning when we measured the latency to fall from a rotating rod for 7 days (Fig. 3C). In addition, we evaluated motor balance and motor coordination using a balance beam (Fig. 3D). The hind limbs of both GluK2 KO and GluK5 KO mice frequently slipped from the bars compared to those of the WT mice (WT: 0.7 ± 0.1, GluK2 KO: 4.6 ± 0.6, GluK5 KO: 3.2 ± 0.6; F (2,75) = 23.15, p < 0.0001; WT vs. GluK2 KO mice, p < 0.0001; WT vs. GluK5 KO mice, p = 0.0027; Fig. 3E). Since GluK2 is highly expressed in the cerebellum (Fig. 1C), which is important for motor control, and the expression is particularly high in the cerebellar granule cells ( Fig  S1A), we generated cerebellar granule cell-specific GluK2 KO (GluK2 GC  Immunohistochemical staining of GluK2 and GluK5 in the brain section of adult mouse hippocampus. Intense labeling of GluK2 was found in the CA3 mossy fiber and molecular layer of the DG. Intense labeling of GluK5 was found in the CA3 mossy fiber, LM, and in the inner layer of Mo. CA1-3, Cornu ammonis 1-3; DG, dentate gyrus; Or, stratum oriens; Py, pyramidal cell layer; Ra, stratum radiatum; LM, stratum lacunosum-moleculare; Mo, molecular layer; Gr, granule cell layer; Pl, polymorphic layer; Lu, striatum lucidum. Scale bar represents 500 μm. KO) mice by crossing GluK2-floxed mice with E3CreN mice [44] (Fig. S1B). We performed an open field test and a balance beam test, and found that GluK2 GC KO mice exhibited normal locomotor activity and normal motor balance/coordination ( Fig. S1C and S1D). These results suggest that the GluK2 expression in cerebellar granule cells is not responsible for the reduced locomotor activity and motor dysfunction seen in GluK2 KO mice.

Anxiety-related behavior in GluK2 KO and GluK5 KO mice
To determine whether GluK2 and GluK5 are involved in anxietyrelated behavior, we measured basal anxiety using an open field test and a light-dark transition test. There were no significant differences in the percentage of time spent in the central area of the open field between the three genotypes (WT: 8.9 ± 1.0 %; GluK2 KO: 11.2 ± 1.6 %; GluK5 KO: 10.0 ± 2.3 %; F (2,57) = 0.5353, p = 0.5884; Fig. 4B). Interestingly, when we analyzed the travel distance during 10 min of examination, we found that GluK2 KO mice exhibited a remarkably reduced travel distance in the first 2 min, which was not seen in GluK5 KO mice (1 min : F  Fig. 4C). After 3 min, both GluK2 KO and GluK5 KO mice showed a significantly reduced travel distance compared to that of the WT mice (Fig. 4C). In the lightdark transition test, we found a significant reduction in the total travel distance in both GluK2 KO Fig. 5B). We did not    Fig. 5D). The number of transition between the light and dark boxes was significantly decreased in both GluK2 KO and GluK5 KO mice, possibly due to reduced locomotor activity (WT: 39 ± 2.9; GluK2 KO: 18 ± 1.8; GluK5 KO: 24 ± 1.7; F (2, 48) = 20.87, p < 0.0001; WT vs. GluK2 KO mice, p < 0.0001; WT vs. GluK5 KO mice, p = 0.0001; Fig. 5E).

Depressive-like behavior in GluK2 KO and GluK5 KO mice
Depressive-like behavior was evaluated using the tail suspension test (Fig. 7). We chose a tail suspension test rather than a forced swim test to avoid any influence on motor activity. The immobility time of GluK2 KO mice was significantly longer compared to that of the WT mice during the entire time of suspension (1 min : F (2,40)

Contextual fear memories in GluK2 KO and GluK5 KO mice
We investigated fear expression in GluK2 KO and GluK5KO mice, using contextual and cued fear conditioning tests (Fig. 8). During the conditioning session, the travel distance of GluK2 KO mice was significantly shorter than that of the WT and GluK5 KO mice (20 s: F (2,42) Fig. 8B). However, there were no significant differences in the cued test between genotypes (pre tone: F (2,

Discussion
GluK2 and GluK5 are major heteromeric KAR subunits in the central nervous system. Here, we generated GluK2 KO and GluK5 KO mice in a pure C57BL/6N background and examined their functional differences using various behavioral paradigms. Both GluK2 KO and GluK5 KO mice showed motor dysfunction and depressive-like behavior. Interestingly, GluK2 KO mice were found to be less motivated to explore the experimental arena relative to WT and GluK5 KO mice. Furthermore, only GluK5 KO mice exhibited enhanced contextual memory.
In the current study, we performed western blot analysis of GluK2 and GluK5 subunits in P0 to 6-month-old mouse brains and in various brain regions. GluK2 subunit transcripts in rats have been shown to transiently increase in many brain regions during the perinatal period E17-P0. In adulthood, they are expressed abundantly in the hippocampal DG, CA3, and cerebellar granule cells [17,45]. In our western blot analysis, the protein expression of GluK2 subunit gradually decreased from 2 weeks of age with a peak at P3-7. Therefore, the expression profiles of mRNA previously reported in rats and the expression of GluK2 proteins in mice from our studies were not similar. Since no developmental analysis of GluK2 mRNA has been conducted in mice, these studies will help clarify this point. In contrast to GluK2 mRNA, GluK5 mRNA is abundantly expressed in many brain regions, including the cortex, hippocampus, and thalamus, and continues to be expressed during rat ontogeny from E14 to adulthood [45]. Interestingly, the expression of GluK5 protein increased from 2 weeks of age. Therefore, there may be a regulatory mechanism that suppresses GluK5 protein expression in the early stages. It has been reported that the kainate binding in the rat brain is not detectable until E14, even though transcripts are present at E12, indicating a time lag between the appearance of mRNA and the accumulation of proteins [45]. Further studies investigating the molecular mechanisms regulating the expression of KAR proteins are required. Although we found reduced expression of GluK2 subunit after 2 weeks, our previous study revealed that, in the hippocampal post synaptic density fraction of 7− 12 weeks-old mice, the relative abundance of GluK2 to GluK5 subunits is in the order of GluK2 > GluK3 > GluK5 > GluK4 [17]. Thus, even though GluK2 expression is reduced in the adulthood, GluK2 is likely to be the most abundant KAR subunit in the brain.
Western blotting and immunohistochemistry revealed that a strong expression of GluK2 in the cortex, hippocampus, and cerebellum. These results are consistent with findings from previous studies that examined the expression of KAR subunit mRNA in the mouse brain [17,46]. Furthermore, the expression of GluK2 in the cerebellum was particularly distinctive in the granule cells. Although GluK2/GluK5 heteromers are thought to be major subunits in the brain, we found very little expression of GluK5 in the cerebellum, suggesting the possibility of a subunit-specific role of GluK2 in the cerebellum. The histological analysis of GluK2 and GluK5 in the hippocampus also confirmed that the GluK2 KO showed a significantly higher percentage of immobility than did WT mice throughout the study. GluK5 KO showed significantly higher percentage of immobility in the minute 3. *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA with Tukey's post hoc test. (B) Percentage of total immobility time. There was a significantly higher immobility in GluK2 KO and GluK5 KO mice than in WT mice. WT, n=15; GluK2 KO, n=14; GluK5 KO, n=14; *p < 0.05, ***p < 0.001, one-way ANOVA with Tukey's post hoc test. All values presented are mean ± SEM. distribution of these two subunits are not completely similar. However, a strong expression of both the subunits was evident in the CA3 stratum lucidum but not in the stratum radiatum. This characteristic pattern of immunostaining in these two subunits was consistent with findings from previous reports [32,42,43]. Although GluK2 and GluK5 mRNA were likely to be expressed in the hippocampal CA1 regions [17], GluK2 protein expression in this region was not detectable in our histological analysis.
To understand the physiological role of GluK2 and GluK5 subunits, knockout mice were generated. GluK2 KO mice generated in the 129/ SvEv strain that was twice backcrossed with C57BL/6 mice exhibited reduced activity in the open field test and reduced spatial learning and memory in the Morris water maze test [28]. GluK2 KO mice originating from the C57BL/6:129S-mixed background and backcrossed with 129/SvEv mice for at least 10 generations showed abnormal behavioral phenotypes, such as hyper locomotor activity and increased social interaction in the open field test, anxiolytic behavior in both the open field and elevated plus maze tests, and lesser depressive-like behavior in the forced swim test [31]. In addition, GluK2 KO mice generated in the 129/SvEv strain, which was further backcrossed with C57BL/6 mice for at least 10 generations, showed altered spatial learning and memory in the Morris water maze test and T maze tests, as well as reduced social interaction [47]. Notably, our GluK2 KO mice with a pure C57BL/6N strain showed reduced locomotor activity in the open field test, normal levels of anxiety in the open field and light-dark transition tests, normal levels of sociability in the three chamber test, and a higher depressive-like behavior in the tail suspension test. These contradictory results in comparable behavioral tests revealed that there is a discrepancy in the phenotypes of GluK2 KO mice depending on their genetic background [ Table 1]. Supporting this hypothesis, it has been reported that 129/Sv mice and C57BL/6 mice exhibit differences in the basal behaviors such as anxiety [35,36], prepulse inhibition [38], social interaction [39], depressive-like behavior [36], and learning and memory [41,42]. To investigate the behavior of KAR KO mice, we generated GluK2 KO and GluK5 KO mice in a pure C57BL/6N genetic background without any backcrossing. These mice enabled us to compare the differences in their behaviors and to understand the physiological functions of GluK2, GluK5, and GluK2/GluK5 subunits containing KARs.
Both GluK2 KO and GluK5 KO mice showed reduced locomotor activity in the open field test. Even though the spontaneous locomotor activity in both KO mice was only reduced by up to 30 % in both KO mice and many behaviors were unaltered compared to WT mice, this phenotype affected our results and introduced some limitation into our . Total travel distance during the pre-tone of the conditioning test was reduced in GluK2 KO mice (left). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, one-way ANOVA with Tukey's post hoc test. Freezing responses on the conditioning tests were not significantly different between genotypes (right). (B) Contextual test. There was a significantly higher freezing response in GluK5 KO but not in GluK2 KO mice 24 h after conditioning. p = 0.013 compared to WT mice, p = 0.028 compared to GluK2KO mice, ns, not significant; *p < 0.05, one-way ANOVA with Tukey's post hoc test. (C) Cued test. There was no significant difference between genotypes during pretone and tone. WT, n=18; GluK2 KO, n=14; GluK5 KO, n=12. ns, not significant; one-way ANOVA. All values presented are mean ± SEM.

Spatial learning and memory
Morris water maze test (probe test) studies. Therefore, to evaluate the results correctly, we used various behavioral paradigms in each study. In the open field test, there were no significant differences in the time spent in the center area. In a show of anxiety-like behavior in the light-dark transition test, the transition number between light and dark boxes and the total travel distance were significantly decreased in both GluK2 KO and GluK5 KO mice, which was likely due to reduced basal locomotor activity found in these genotypes. However, the time spent in each box and the latency to the first transition were unaltered. These results are different from that of several previous reports, which have indicated an abnormal mood behavior in GluK2 KO mice [31,47]. Due to the altered locomotor activity found in most of the KAR KO mice, it is difficult to evaluate anxiety precisely. However, both GluK2 KO and GluK5 KO mice displayed similar phenotypes in terms of anxiety-related behavior.
Using three chambers, we found that both GluK2 KO and GluK5 KO mice exhibited normal social behavior, as determined by the noticing of the stranger mice and the traveling around their cage to seek their smell. This result was somewhat different from that of a previous study, wherein less sociability relevant to autism spectrum disorders was observed in GluK2 KO mice [47]. In this study, the authors manually counted nose contacts with a cylinder containing a female stranger mouse and the time (s) spent near the cylinder. Reduced sociability was observed in GluK2 KO mice. However, these phenotypes could not be reproduced when time spent around each cylinders were calculated. Additionally, when sociability was examined in the open field, the number of side-by-side events did not differ between GluK2 KO and WT mice, however, GluK2 KO mice exhibited a significantly higher frequency of sniffing [31]. From these data, we suspect that GluK2 KO mice exhibit normal sociability at least in the measurement of the time spent around the cages in the three-chamber test.
In the current study, we were able to demonstrate for the first time the specific behavioral phenotype of the GluK5 subunit. In the fear conditioning test, GluK5 KO mice showed significantly longer freezing times in the 24 h contextual test. It is well known that the hippocampus interferes with the conditioning of fear responses to the context but not to the cue, whereas the amygdala interferes with the conditioning to the cue. Therefore, GluK5 subunits in the hippocampus are thought to be responsible for contextual memory. In this study, we found that the GluK5 subunit was broadly expressed in the CA1 with a particularly enriched expression in the stratum lacunosum-moleculare (SLM). However, in the DG, their expression was limited to the inner molecular layer (IML). On the contrary, the GluK2 subunit was not expressed in the CA1 but broadly expressed in the inner molecular layer of the DG. Based on these results, we think that GluK5 subunits in the CA1 are more important for contextual memory. In agreement with our hypothesis, recent studies have revealed that hippocampal CA1 pyramidal neurons are critical for the formation and retrieval of episodic memories [48,49]. In addition, KAR auxiliary subunit NETO2 KO mice, which express lower levels of GluK5 and GluK2/3 at the synapse in the ventral hippocampus, showed higher fear expression and delayed extinction compared to those in WT mice in the fear conditioning test [50]. These phenotypes in NETO2 KO mice strongly corroborate our results of the fear conditioning test, suggesting that GluK5 is critical for the contextual conditioning memory. Since our fear conditioning tests revealed that GluK2 subunits do not relate to either contextual fear or cued memory, it is possible that other GluK5 heteromers such as GluK1/GluK5 and GluK3/GluK5 may be involved in the development and maintenance of contextual memory.
In addition, we found that the depressive-like phenotype was more severe in GluK2 KO mice than in GluK5 KO mice in the tail suspension test, suggesting that GluK2 is relatively more critical in regulating these behaviors. Supporting these observations, GluK2 KO mice exhibited a significant decrease in the exploration time in the first 2 min of the open field test, during the habituation of the three-chamber sociability test, and before the fear conditioning, neither of which were found in GluK5 KO mice. In the light-dark transition test, GluK2 KO mice also showed a trend of spending more time to get into the first transition compared to the time taken by GluK5 KO mice. These results indicate that GluK2 KO mice are less motivated to explore their environment, possibly due to an increased depressive-like behaviors. Therefore, a general motor disability identified in the GluK2 KO mice may not be caused solely by impaired motor function. Our findings revealed that there might be a different role for GluK2 and GluK5 subunits in depressive-like behavior and motivation. The hippocampus is the most commonly studied brain region for depression, and there is much evidence using animal models that depression occurs in the context of stress [51]. Stress is known to impair long-term potentiation in the CA3 together and facilitate long-term depression (LTD) and spike-timing-dependent LTD in the CA1 of the hippocampus [52,53]. KARs are thought to be linked to multiple stress disorders and major depressive disorder. Lowering of stress-related corticosteroid hormone leads to an increase in the expression of GluK2 and GluK3 mRNA in the rat hippocampus [54]. In this study, we found that GluK2 and GluK5 are co-expressed in the CA3 mossy fiber, where stress has been known to have a major impact. Therefore, the depressive-like behavior identified in our GluK2 KO and GluK5 KO mice may be due to the dysfunction of neural circuits originating from the CA3 regions where the GluK2/GluK5 heteromer is expressed. Since GluK2 and GluK5 are expressed in various brain regions such as the cerebral cortex, amygdala, and thalamus, further studies are needed to identify the association between KARs and depression. In a previous report, GluK2 KO mice that were backcrossed with the C57BL/6 background for more than 10 times exhibited enhanced spatial learning in the Morris water maze [47]. However, studies using a 129/Sv × C57BL/6 background showed a significant reduction in fear memory in the mice [29]. Interestingly, GluK2 KO mice in our study showed normal contextual and cued memory. These discrepancies of GluK2 KO mice in learning and memory need to be clarified in the future studies.
GluK2 is highly expressed in the cerebellum, and the motor learning ability was normal in GluK2 KO mice. This was consistent with a previous report on GluK2 KO mice maintained on a 129/Sv × C57BL/6mixed background [31]. However, we found that the locomotor activity, walking speed, and motor balance were all reduced in our GluK2 KO mice. Generation of GluK2 GC KO mice revealed that the reduced basal activity and motor coordination in GluK2 KO mice were not due to a defect in neurotransmission via GluK2 containing KARs in the cerebellum. In a recent study, a complete loss of KAR subunits in mice resulted in both functional and structural alterations of synapses in the striatum, and a dysfunction of the hindlimb-clasping response was one of their behavioral phenotypes [55]. High expression of GluK2 mRNA has also been shown in the striatum [56], and our western blot analysis indicated that both GluK2 and GluK5 proteins are highly expressed in the striatum. Therefore, altered neurotransmission in the striatum may have induced abnormal phenotypes, including reduced activity and hindlimb motor deficits in GluK2 KO and GluK5 KO mice. Further research is needed to identify the causes of motor deficits and reduced locomotor activity in these mice.

Conclusion
In this study, we generated GluK2 KO and GluK5 KO mice under a pure C57BL/6 N genetic background and compared their behaviors. GluK2 KO and GluK5 KO mice exhibited similar phenotypes in many tests, with reduced motor activity and depressive-like behaviors. GluK2 KO mice were less motivated to explore their environments, and GluK5 KO mice exhibited an enhancement in the contextual memory. GluK2/ GluK5 heteromers are thought to be the major KARs in the central nervous system; however, our results revealed that GluK2 and GluK5 subunits not only function concomitantly but also have a subunitspecific role in the behavior. Of note, this is the first evidence implicating GluK5 in the expression of depressive-like behavior and the development of contextual conditioning memory, which strongly supports the role of GluK5-containing KARs in psychiatric disorders.