All experiments in the study using mice were approved by Uppsala Animal Ethical Committee (C69/15, C77/14 and 5.8.18-09820/2018) and followed Swedish Legislation on Animal Welfare. The animals were kept in an animal facility with a controlled environment, with humidity (45–65%), temperature (20–24°C), and ventilation system, and a 12-h light/dark cycle with lights on at 7 a.m. Animals were group housed in open cages, with two to five animals per Green Line cage (501–530 cm²) separated by sex, with wood-chip bedding, a cardboard house, and paper as enrichment. Breeding was carried out in type III cages (800–820 cm²). Mice had daily supervision and ad libitum access to food (pellets R3, Lantmännen, Sweden) and water.

Heterozygous (B6Dnk;B6N-Slc38a10^{tm2a(EUCOMM)Wtsi}/H) and control mice were bought from the International Mouse Phenotyping Consortium (IMPC; Slc38a10, n.d.) and kept for in house breeding. The first heterozygous males from that breeding were bred with C57BL/6J females (Taconic M&B, Denmark). Further breeding was carried out from in house bred offspring.

In this paper, SLC38A10^−/− mice will be abbreviated as KO, SLC38A10^+/− as HET, and SLC38A10^+/+ as WT.

Ear biopsies were obtained after weaning (3–4 weeks of age) and incubated in 75 μl of 25 mM NaOH and 200 μM EDTA at 95°C for 40 min. After a cooling time of 5 min on ice, 75 μl of 40 mM Tris-HCl was added and the samples were then ready for PCR analysis. For the PCR, primers for the LacZ domain in the knockout construct (Figure 1A) and primers for the wild-type gene were designed.

Mated females were checked for a vaginal plug and, if present, that was determined as gestational day 0.5. At gestational day 15.5, the females was euthanized through cervical dislocation and embryos (16 WT, 25 KO) were dissected, photographed with a ruler, and decapitated. Measurements were made with ImageJ software. Weight data from the first 3 weeks of life were obtained from individual weighing of pups (males: 10 WT, 8 KO; females: 11 WT, 5 KO) at the same time every day, and was always carried out by the same person. Individual weighing of adult mice was carried out at the age of 10, 13, 20, and 25 weeks of age (males: 12 WT, 10 HET, 12 KO; females: 15 WT, 12 HET, 11 KO).

Two mice of the same genotype were housed per Green Line cage, in which a pre-weighed amount of food was placed. Every 24 h, during the light period, the amount of food left was weighed and thereafter refilled and weighed again, for five consecutive days. The amount of food eaten was divided with total weight of the two mice, hence each cage is one statistical unit. Only WT and KO mice were included, and they were 8–9 weeks of age (9 WT cages: 3 cages with males; 6 cages with females; 7 KO cages: 3 cages with males; 4 cages with females).

Both male and female mice were used, and behavior testing was carried out on WT, HET, and KO mice. Male mice were always tested prior to female mice, and cleaning was done between each mouse with 10% ethanol (unless otherwise stated). Mice were handled daily on the week before initiating the first behavioral test and always by the same person that performed the experiments (FL).

Handling protocol: The first 4 days of handling was carried out in the home room of the animals, with just a hand present in the cage for 5 min the first day. The same procedure was conducted during the second day, but without enrichment in the cage. During the third and fourth day of handling, the hand of the handler was placed in the cage for 2 min, without enrichment present in the cage, and each moused was lifted and put on the arm of the handler for 30 s. During the fourth day, each mouse was placed on the arm twice. During the fifth and final day of the handling protocol, the cages was placed in the experimental room for 30 min, and thereafter was each mouse handled in the same way as on the previous day.

Animals were habituated to the test room in their home cage (unless otherwise stated) 1 h prior to the start of each test. The elevated plus maze (EPM), Y-maze, and open field (OF) tests were recorded with the Ethovision system (version 11.0, Noldus Information Technology, Wageningen, the Netherlands). The same software was used for automatic tracking of the OF and EPM.

Animals were either going through the EPM and the Y-maze, or the OF, marble burying, rotarod, and the challenging beam walk, in respective order (Figure 1B). Mice used in the grip strength test were only used in that test. Littermate controls from heterozygous breeding were used for all tests, except for the grip strength test where KO and WT mice were derived from separate breedings. Mice were of age 7–9 weeks when they entered the first test, and it was at least 48 h between each test in case of repeated testing. All behavioral testing occurred during the light phase of the light/dark cycle.

Embryo size was analyzed with Mann-Whitney U-test and body weight data were analyzed with two-way ANOVA (repeated measures) and Bonferroni post-hoc tests (GraphPad Prism version 5.02, GraphPad software Inc., San Diego, California USA). A majority of the results from the behavioral tests were not normally distributed as shown by the Shapiro-Wilk normality test (GraphPad Prism, version 5.02), hence non-parametric statistics were used. The EPM, Y-maze, rotarod, challenging beam walk and grip strength tests were analyzed with the Kruskal–Wallis and Dunn's multiple comparisons test (GraphPad Prism, version 5.02). Longitudinal data sets (OF and food intake) were analyzed for main effects and interactions in R 4.0.4 (R Core Team, 2021) with the nparLD package (Noguchi et al., 2012), with genotype and sex as between-subject factors and time as within-subject factor. Significant effects and interactions were further examined in Statistica 13 (TIBCO Software Inc., Tulsa, OK, USA) with post-hoc Mann–Whitney U-test with continuity correction (between subject-dependent) and/or Wilcoxon's matched pairs test (within subject-dependent). Data are presented as means ± standard deviation (body weights) or medians with interquartile range (IQR; embryo size, food intake and all behavioral data), and the significance level used was 0.05 for all tests. Effect sizes were calculated for significant differences according to Fritz et al. (2012).

Length of KO (median 1.443 cm, IQR 1.39–1.587, n = 25) and WT (median 1.483 cm, IQR 1.423–1.562, n = 16) embryos at gestational day 15.5 was used as an indicator of body size in the embryonal stage (Figure 2A) where no significant difference was observed. The food intake of KO (n = 7) and WT (n = 9) mice did not show any differences between the genotypes (main effect of genotype p > 0.5; main effect of time p < 0.01; main effect of sex p > 0.05; no interactions found) (Figure 2B, Supplementary Table 1). However, a significant body weight difference with lower body weight in KO than WT mice was found from postnatal day (PND) 13 in males (Figure 2C), with main effect of genotype at p < 0.01, main effect of time at p < 0.001, and interaction genotype:time at p < 0.001. In females, a body weight difference could be seen from PND 28 in females (Figure 2D), with no main effects of genotype at p > 0.05, main effect of time at p < 0.001, and interaction genotype:time p = 0.001. Post-hoc analyses showed that an increasing PND age discloses the genotype effect. The body weight of HET mice was not studied until adult age wherein a difference between KO and HET could be observed for most time points in both sexes [Figure 2E (main effect of genotype at p < 0.001; main effect of time at p < 0.001; no interactions found) and Figure 2F (main effect of genotype at p < 0.001; main effect of time at p < 0.001; interaction genotype:time p < 0.05)], while no differences were observed between HET and WT mice. Post-hoc analysis revealed that genotypes had the same effect for all time points except for the age of 13 weeks in HET female mice, explaining the interaction seen from the repeated measures ANOVA. Mean weights and standard deviations for each genotype and age are summarized in Supplementary Table 2.

In the OF test, mice explored the arena for 60 min. The 60-min period was divided into 5-min time bins when analyzed (referred to as time-points), and medians and interquartile range of all groups are shown in Supplementary Table 3. Results are shown for male and female mice together for all parameters, except for total distance moved in the arena since it was influenced by sex (main effect of sex p < 0.01; Table 1). The main effect of time was significant for all parameters studied (p < 0.01; Table 1), while genotype had a main effect on frequency, duration, duration per visit, distance moved in central circle, and on duration in wall zone (p < 0.05; Table 1). Interactions were found for frequency and distance in the central circle (Sex:Genotype:Time, p < 0.05; Table 1). Statistics for each sex separately can be found for these parameters in Supplementary Table 3 and effect sizes can be seen in Supplementary Table 4. The most pronounced statistically significant genotype differences were seen between HET and WT mice, with KO mice following the same trend as HET mice. Relative to WT, SLC38A10-deficient mice showed increased amounts of visits to the central circle (at time-points 5–10, 10–15, 25–30, 30–35, 45–50, and 55-60 min; Figure 3A) along with an increased duration of time spent in the central circle (at 5–10, 10–15, 25–30, 30–35, 35–40, 45–50, and 55–60 min; Figure 3B, Supplementary Figure 1B), but not an increased duration per visit during the first half of the test (Figure 3C), indicating that SLC38A10 KO mice visited the central circle more often but did not spend more time there when visiting. For the second half of the test, SLC38A10-deficient mice had an increased duration per visit in the central circle than WT mice (at time-points 35–40, 40–45, 45–50, and 55–60 min; Figure 3C). The distance moved within the central circle was increased in HET and KO mice as a general trend, although it was only statistically significant for a few of these time points (at 5–10, 10–15, 30–35, 45–50, and 55-60 min; Figure 3D). Time spent in the wall zone was clearly affected by genotype, where both KO and HET mice spent less time than WT (Figure 3F, Supplementary Figure 1C). Effect sizes were calculated for statistically significant differences (Table 2), which indicated that the differences seen were either small or intermediate in size. Mean velocity in the central circle and total distance moved in the whole arena was not influenced by genotype (Figures 3E,G,H), although visual inspection of the graph indicates a trend toward a reduced velocity of SLC38A10-deficient mice in the central circle (Figure 3E). The frequency of rearing and grooming during the first 5 min of the OF test revealed no significant effect of genotype (Figures 3I,J). Fecal boli were counted after 60 min in the arena but showed no differences between the genotypes (Supplementary Figure 1A). These data together suggests that SLC38A10-deficient mice were more risk-taking than their WT littermates.

For 10 min, mice explored the EPM and frequency and duration in open and closed arms were analyzed along with center time. As shown in Figures 4A,B, the genotype did not affect the frequency of visits or the cumulative duration of time spent in open arms. Neither did the frequency of visits nor cumulative duration in the closed arms (Figures 4C,D) for each sex separately, but with sex collapsed HET mice displaying more visits to the closed arms than WT mice (Figure 4C). Duration in center was not affected by genotype (Figure 4E). The sum of visits to the center, open, and closed arms (Figure 4F) was analyzed as a measure of general activity, which, for the group with sex collapsed, showed an increased activity in HET mice compared to WT mice. Frequency in the center were found to be increased in HET mice with sex collapsed (Supplementary Table 5).

As a motor coordination test, the rotarod was used. After three consecutive days of training, mice had three trials on the test day from which the best trial was analyzed. As seen in Figure 5A, the genotype did not affect the performance in this test for either males or females. In the challenging beam walk test, a tapered beam with a mesh grid was used to study the mice's ability to coordinate their movement in a more challenging manner. The best trial out of three on the test day was analyzed for both sexes separately and together, but the KO or HET mice did not differ in their performance compared to WT mice (Figure 5B).

Spontaneous alternation is an innate behavior of mice, and is a trait that is used in the Y-maze to study spatial memory. The mice explored the Y-maze arena for 10 min and each visit to each arm was scored. The alternation score (Figure 5C) was not affected by either sex or genotype. The total number of entries into the arms was used as an activity measure but did not differ between the experimental groups (Figure 5D).

Repetitive behaviors were studied in the marble burying test in which mice were introduced to 18 glass marbles on woodchip bedding. For 15 min, mice investigated the marbles and marbles buried to a minimum of 75% were counted as buried ones. As observed in Figure 5E, either sex or genotype affected the number of buried marbles.

To test the grip strength, mice were placed on a grid and turned upside-down for a maximum of 3 min. No HET mice were included. No differences were found between KO and WT mice in male or female mice (Figure 5F).

Descriptive statistics for each of these tests can be found in Supplementary Table 6.

The SLC38A10 KO mouse showed a lower body weight than both the WT and the HET mice (Figures 2E,F), which agrees with previous data by IMPC (Dickinson et al., 2016; Slc38a10, n.d.). Since the focus of this study was to characterize the effects of SLC38A10 knockout on behavior, we did not investigate the lower body weight further. However, previous data (Bassett et al., 2012; Dickinson et al., 2016; Slc38a10, n.d.) indicate that the lower body weight is due to a decreased lean body mass. The normal weight of HET mice may indicate that one functional Slc38a10 allele is enough to maintain a normal body weight and, possibly, bone mineralization. Herein, the differences in body weight between KO and WT mice were first seen at postnatal day (PND) 13 in males, and not until PND 28 in females (Figures 2C,D). The delay in females is probably due to the lower body weight in females compared to males, which makes a larger number of animals needed to detect differences. Since just five animals were used in the female KO group, it is possible that this body weight difference could have been detected earlier if more animals were used. When looking at when SLC38A10 is starting to be expressed during development, there are annotations of it to be expressed as early as at Theiler stage 12 (e7.5-8.75) (Smith et al., 2019). The body weight difference was not observed until 24–39 days later (depending on sex) based on our data, which is probably due to the low number of animals per group (as previously discussed). In order to detect size and/or weight differences earlier than we did, a much bigger n would probably be needed. On another note, the breeding strategy for collection of pup weight data was with homozygous breeding couples, and we cannot exclude the possible effect of the low body weight in the female as a confounding factor in these results. Previous studies have suggested Slc38a10 to be important for cell volume regulation during development (Bassett et al., 2012).

Disruptions in function of other members of the SLC38 family, such as SNAT3 in mouse (Chan et al., 2016) and SLC38A11 ortholog in D. melanogaster (Aggarwal et al., 2020), also affect body weight, indicating the importance of SLC38 transporters.

In the OF, both KO and HET mice spent less time in the wall zone, accompanied by more visits, longer duration per visit, and longer distance moved in the central circle in comparison to WT mice (Figure 3). The percentage of cumulative time spent in the central circle during the whole test period (60 min) was 2.1, 4.2, and 4.1%, and for time spent in wall zone, 86, 76.3, and 72.7% for WT, HET, and KO mice, respectively (Supplementary Figures 1B,C). Thus, the results suggest a more explorative and risk-taking behavior relative to WT. This is further supported by the lower, but non-significant, mean velocity in the central circle in HET and KO mice. Since the total distance did not differ between the genotypes, it is implied that the activity itself was not the cause of the increased visits to the center. Previous results have shown that animals with increased risk-taking behavior display a similar pattern (Momeni et al., 2014). Thus, the results imply that SLC38A10 could be involved in processes affecting risk-taking and explorative behavior. Fecal boli are sometimes used as a measure of emotionality (Hall, 1934), with a positive correlation between defecation and emotionality, but no differences were observed in this study (Supplementary Figure 1A), supporting the interpretations of the above discussed data. However, when calculating effect sizes of the significant results from the OF test (Table 2), the effects seen are small to intermediate, indicating that the significant results seen might not cause any major behavioral shift in the HET and KO mice. However, the increased risk-taking behavior observed in the OF was not reflected in the EPM. Either frequency of visits or the cumulative duration spent on the open arms in the EPM differed between the genotypes, for either of the sexes separately or together (Figures 4A,B,D). However, a genotype effect was seen on visits to the closed arms in HET mice in the analysis with sex collapsed (Figure 4C). The same effect was seen for the activity parameter (sum of all frequencies), indicating a slightly increased activity in HET mice. In males, the large individual variation in open arm behavior of the WT group impacted on the analysis, which makes the interpretation hard. If visually inspecting the graphs, one can speculate that there might be a tendency to an increase of time spent on the open arms in SLC38A10-deficient female mice, but it needs further studies to establish if there is an effect of genotype or not in this test.

To draw any conclusions from this data is troublesome due to the lack of a clear phenotype. The data indicates that there is a possibility of a more explorative/risk-taking behavior in SLC38A10 mice, but data from the OF test and the EPM do not point in the same direction. However, it is possible that the OF and EPM do not measure the same aspects of emotionality, which have previously been shown by the lack of correlations between them (O'Leary et al., 2013). It has also been previously suggested that locomotor activity can influence anxiety-related behavior, where results regarding emotionality need to be interpreted with caution due to its high correlation to activity (Dawson and Tricklebank, 1995; Milner and Crabbe, 2008). In our data, we do not see a different activity pattern in SLC38A10-deficient mice compared to their WT littermates in the OF test (see Figures 3G,H), suggesting that the differences between the genotypes seen is rather an effect on emotionality than locomotor activity. HET mice in the EPM were more active than both WT and KO mice (Figure 4F), but no aspects of risk-taking/thigmotaxic behavior could be observed. In the OF test though, females were a bit more active than males (Figures 3G,H), but this was not affected by genotype. Sex differences in locomotion have been reported before (Broida and Svare, 1984).

Amino acid transporters of various kinds are known to be implicated in neurodegenerative and psychiatric diseases (Smith et al., 2001; Tǎrlungeanu et al., 2016; Guan et al., 2019; Hu et al., 2020), which makes gene characterization research like the study conducted here needed. The altered behavior in the OF test could be due to the changed amino acid homeostasis that most likely exists in mice deficient in SLC38A10. If one considers the hypothesis by Tripathi et al. (2019) that SLC38A10 might function as a transceptor within the cell, likely as a glutamine sensor, one could hypothesize that flawed glutamine sensing would lead to a disturbance in glutamine/glutamate/GABA levels. Due to the importance of these systems, it is likely that there are compensatory mechanisms in place to correct flaws within them, or that additional systems for sensing also exists, and hence give rise to a milder phenotype as the one we see in this study (Tautz, 1992; El-Brolosy and Stainier, 2017). However, the SLC38A10 protein has been associated with Alzheimer's disease, and a single nucleotide polymorphism (SNP) and differences in methylation of the gene have been, respectively, found in patients with frontotemporal dementia and schizophrenia (Ferrari et al., 2015; Montano et al., 2016; Hashimoto et al., 2019), suggesting a possible role of the SLC38A10 protein in cognitive function. Based on the results from this study and the studies earlier mentioned associating SLC38A10 to cognitive function, further studies regarding these behaviors would be of interest, e.g., a memory test such as the Barnes maze. A multivariate test approach, like the multivariate concentric square field™ test (Meyerson et al., 2006), could be an interesting alternative to further study the behavioral profile of the SLC38A10 KO mouse.

What is interesting to note is that HET mice are more similar to WT mice when comparing body weights (Figure 2). However, when looking at the behavior in the OF test, HET mice are behaving more like the KO mice. These data indicate that there might be enough of the SLC38A10 protein in HET mice to maintain a normal body weight, but not enough to not affect behavior in an OF test. One explanation for this could be that maintaining a normal body weight is more acute for the health of the animal, while the mild behavioral phenotype seen in this study is not as profound to affect the survival of the animal. In the long run, an increased risk-taking behavior could, of course, affect an individual's survival in the wild.

Regarding rearing, according to Sturman et al. (2018), unsupported and supported rears are to be seen as two distinct behavioral variables, and both should always be scored individually. In this study, the rearings scored were total rears, both unsupported and supported together. Hence no interpretation of the type of rearing occurring in each experimental group could be done. Rearing and stretch-attend-postures (SAPs) are behaviors that has been proven being more correlated between tests than the individual tests themselves (O'Leary et al., 2013), which was not scored for all tests (rearing) or at all (SAPs). It is also possible that mice who went through several behavioral tests might have been influenced by the previous test, which in turn could have influenced test results (McIlwain et al., 2001). However, by using the current study layout, we could decrease the number of animals needed. The inter-test interval of at least 48 h has before been proven to work just as well as one to two weeks in a specific behavior test battery (Paylor et al., 2006), which support the inter-test time in this study to be sufficient.