Detailed analysis of paternal knockout Grb10 mice suggests effects on stability of social behavior, rather than social dominance

Abstract Imprinted genes are highly expressed in monoaminergic regions of the midbrain and their functions in this area are thought to have an impact on mammalian social behaviors. One such imprinted gene is Grb10, of which the paternal allele is generally recognized as mediating social dominance behavior. However, there has been no detailed study of social dominance in Grb10 +/p mice. Moreover, the original study examined tube‐test behavior in isolated mice 10 months of age. Isolation testing favors more territorial and aggressive behaviors, and does not address social dominance strategies employed in group housing contexts. Furthermore, isolation stress impacts midbrain function and dominance related behavior, often through alterations in monoaminergic signaling. Thus, we undertook a systematic study of Grb10 +/p social rank and dominance behavior within the cage group, using a number of convergent behavioral tests. We examined both male and female mice to account for sex differences and tested cohorts aged 2, 6 and 10 months to examine any developments related to age. We found group‐housed Grb10 +/p mice do not show evidence of enhanced social dominance, but cages containing Grb10 +/p and wild‐type mice lacked the normal correlation between three different measures of social rank. Moreover, a separate study indicated isolation stress induced inconsistent changes in tube test behavior. Taken together, these data suggest future research on Grb10 +/p mice should focus on the stability of social behaviors, rather than dominance per se.


| INTRODUCTION
Imprinted genes are defined by their monoallelic, parent-of-origin dependent expression originating from differential epigenetic marks established in the germline. 1 This class of genes is highly expressed in the central nervous system and significantly impacts brain development and adult behaviors. 2 The paternally expressed copy of the imprinted gene Grb10 (growth factor receptor bound protein 10) is expressed in the developing and adult brain, and we have previously established a potential link to social dominance in mice with disruption of the paternally inherited allele (Grb10 +/p ). 3 Murine Grb10 is located on proximal chromosome 11 and encodes a cellular adapter protein belonging to the small Grb7/Grb10/Grb14 family. 4,5 This protein has an inhibitory effect on signaling through receptor tyrosine kinases, including the insulin receptor and insulin-like growth factor receptor. 6 Paternal Grb10 is highly expressed in the midbrain and hindbrain, including regions such as the ventral tegmental area, the substantia nigra pars compacta, the dorsal raphe nucleus, thalamus and hypothalamus, and is neuron-specific. 3,7 Male Grb10 +/p mice 10 months of age were previously reported to be significantly less likely to back down in the Lindzey tube test. This correlated with an elevated incidence of facial barbering in cages containing Grb10 +/p mutants. 3 Both measures are considered indicators of social dominance. [8][9][10] However, in the original study tube testing was not conducted within an animal's normal cage group, and also took place after mice were isolated for an extended period to determine whether the barbering was self-inflicted. 3 Social isolation impacts midbrain function and dominance-related behaviors, often through alterations in monoaminergic signaling. 11,12 In periods of isolation between 14 and 28 days, this includes alterations in tyrosine hydroxylase transcription, and over 3 months this includes changes in epigenetic marks and writer/eraser activity in the midbrain. 11,13 Even short periods alter signaling and connectivity. Acute social isolation over 24 hours potentiates synapses onto dopamine neurons in the dorsal raphe nucleus (DRN) and alters their glutamate receptor composition. 14 Furthermore, social rank itself impacts the subjective experience of isolation, as dominant mice are more sensitive to the behavioral effects of manipulating DRN dopaminergic activity through optogenetic activation and inhibition. 14 Here we systematically explore social dominance behavior of Grb10 +/p mice. We used convergent measures to assess dominance behavior in socially housed Grb10 +/p mice, including the stranger-and social-encounter Lindzey tube tests, the urine marking test, and characterization of barbering behavior. Both male and female cohorts were used to test for any sex differences. Also, cohorts at 2, 6 and 10 months of age were tested in a cross-sectional study designed to account for any differences that may develop with age. Given the extensive changes to midbrain synaptic function, monoaminergic signaling and epigenetic regulation induced by social isolation, we saw a need to determine whether the isolation period from the earlier experiment 3 impacted the tube test phenotype observed in Grb10 +/p mice. We therefore replicated the dominance testing of isolated Grb10 +/p mice 10 months of age to determine whether isolation stress was required to precipitate the phenotype. Our results indicate Grb10 +/p mice are not more dominant, but may show a social instability phenotype.

| Animals
All procedures were conducted in accordance with the requirements of the UK Animals (Scientific Procedures) Act 1986, under the remit of Home office license number 30/3375 with ethical approval at Cardiff University. Grb10 heterozygous knockout mice on a B6CBAF1/J background were previously created as described in Garfield et al 3 using a LacZ:neomycin gene-trap cassette interrupting exon 7. 3,7 This mouse colony was derived via embryo transfer from a colony in Bath and maintained on exactly the same mixed genetic background. Specifically, breeding stock was maintained with either a B6CBA F1/crl line from Charles River or with an in house mixed B6CBA F1/crl × B6CBA F1/J background. Experimental animals were generated by crossing wild-type (WT) breeding stock with the desired parent of origin heterozygous Grb10 +/− animal. Dams were placed in individual housing the week prior to full term. This measure was necessary to aid pre-weaning ear clip identification and genotyping of the behavioral cohorts. Mice were weaned between P19 and P28 and sorted into genotype-balanced social cages of 4 mice: 2 WTs, 2 Grb10 +/p for behavioral testing. Male mice were genotyped prior to weaning to enable the cage set-up. Females were weaned prior to genotyping and re-sorted into the appropriate set-up as soon as possible. Where possible, animals of the same birth litter were kept together.
All mice were housed in single-sex, environmentally enriched cages (cardboard tubes, shred-mats, chew sticks) of 1-5 adult mice per cage (except for isolation study detailed below). Cages were kept in a temperature and humidity controlled animal holding room (21 ± 2 C and 50 ± 10% respectively) on a 12-hour light-dark cycle (lights on at 7:00 hours, lights off at 19:00 hours). All subjects had ad libitum access to standard rodent laboratory chow and water. Cages were cleaned and changed once a week at a regular time and day of the week for minimal disruption. Cages were not cleaned during multiple day testing of the same dominance test, and were half-cleaned between tube testing and urine marking blocks.

| Behavioral testing
The 2, 6 and 10-month cohorts (but not the isolation cohorts) underwent dominance testing, in order, for: stranger tube test, social tube test and (males only) urine marking ( Figure 1). Behavioral testing was limited to a 4-week window to prevent age overlap with the other cohorts. Mice were handled as little as possible up until 1 week prior to the start of behavioral testing; then they were handled daily for 5 days before beginning testing. Testing was performed in a quiet room lit by a single indirect lamp bulb between 25 and 60 W. Match and cage numbers included in analysis for each behavioral test are reported in Tables 1-2 and 3 below. A "match" constitutes a Grb10 +/p vs WT encounter.

| Tube testing
The Lindzey tube test is an accepted measure of social dominance in mice and can be used to match subjects against strangers or cagemates. 8 The stranger encounter and social encounter tube tests were conducted under identical conditions. For the stranger test, unfamiliar opponents were chosen from different home cages and different litters. Any socially housed WT opponents were housed in genotypebalanced (2 WT, 2 Grb10 +/p ) cages. Opponent mice were simultaneously presented to either end of a Perspex tube (30.5 m × 3.5 cm or 30 cm × 2.5 cm depending on weight class). Opponents met in the middle of the tube and outcome was scored when one animal was forced to back out. Losers were counted as the first animal with all four feet out of the tube. No time limit was imposed.
Trials in which either opponent turned around in the tube, both mice backed out without confrontation, or both mice squeezed past each other were not counted (all instances of trial "failure"). In the stranger encounter tube test, animals were completely naïve to the test and mistrials were not re-run (mistrials are listed in Supplementary information, Table S1). In the social encounter tube test, mistrials were rerun on a separate day to complete the within-cage hierarchy, but each opponent pair only underwent one successful trial. These paradigms were adopted to avoid any learning effects and to parallel testing procedures in reference 3 . Each animal completed only one tube test per day. Testing was arranged to ensure genotype groups and individual mice underwent trials balanced by side of entry. In the stranger encounter tube test, opponents were weight matched to minimize differences across the whole cohort. To maximize trial numbers, no trials were eliminated based on weight. In approximately 77% of encounters, the heavier mouse was less than 15% heavier than the lighter mouse. There were no significant differences in body weight between Grb10 +/p and WT mice in our colony across all three ages (2, 6 and 10 months) (See Supplementary Results). with Ninhydrin spray reagent (Sigma-Aldrich N1286) and scored using a 1 cm 2 grid overlay. All squares containing a scent mark were counted and used in a ratio against usable grid (total grid squares minus shredded sections and urine marks covering more than four consecutive squares). These scent marks/urine drops delineate territorial boundaries and contain chemical cues of social status. 15 The winner of each encounter possessed the higher ratio of squares containing sent marks to usable grid.

| Barbering
The Dhalia Effect, or the whisker barbering effect, describes the tendency for the dominant mouse in the cage to trim the whiskers from subordinates, resulting in cages with just one unbarbered mouse. 9 Barbering status was recorded at every behavioral testing session.
Barbering was identified as the specific removal of whiskers (partial or complete); facial overgrooming could occur independently of barbering, and was thus noted, but not sufficient to confer a "barbered" status.
F I G U R E 1 Experimental design. Four cohorts with both males and females (2, 6, 10 months and Isolation cohort at 10 months) underwent behavioral testing. Testing was limited to a 4-week period and ended at the age indicated in the cohort name. The order of experiments was stranger encounter tube test (S), social encounter tube test (T), urine marking test (U; males only), marble burying test (M; not described in this paper) and elevated plus maze (E). The isolation cohort underwent a 30-day isolation protocol (I) prior to the stranger encounter tube test (S) 3.2 | Grb10 +/p barbers were no more common than WTs

Garfield 2011 reported an increased incidence of barbering in cages with
Grb10 +/p mice. In our study, behavioral cages at 6 and 10 months with identifiable barbers (1 un-barbered to 3 barbered mice in the cage) were pooled to analyze the proportion of Grb10 +/p vs WT barbers (Supplementary information, Table S2). Binomial tests indicated the proportion of barbers who were Grb10 +/p was not statistically different from chance (0.5) in cages of either sex (Figure 2A

| Socially housed Grb10 +/p mice do not show a social dominance phenotype
In the stranger-encounter ( Figure 3A,B; Supplementary information, Table S3) and social encounter tube tests ( Figure 3C,E; Supplementary information, Table S4 & S5), binomial analysis indicated the proportion of wins for Grb10 +/p mice in all three age groups for both sexes were not significantly different to chance (0.5). Likewise, the proportion of Grb10 +/ p wins in the urine marking test was not statistically higher than chance in the 6-and 10-month cohorts. In the 2-month cohort, the proportion of Grb10 +/p wins in the urine marking test (0.70) at 2 months of age was statistically higher than chance (0.05), P = 0.01 (2-tailed), but this did not survive FDR corrections ( Figure 3D; Supplementary information, Table S6).
Rank hierarchies were also established in each cage using the social encounter tube and the urine marking tests (See Supplementary Figure S1). In the social tube test, there was no statistically significant difference between average within-cage rank for Grb10 +/p and WTs at 2, 6 or 10 months of age for males or females. In the urine marking test, there was no significant difference in average within-cage rank at 6 and 10 months. At 2 months, the difference in urine marking rank between Grb10 +/p mice (median average cage rank 0.667) and WTs (median average cage rank 0.333) was statistically significant, but this did not survive and indeed we have previously seen this in our lab. 22 However, there was no significant linear association between rank in the social tube test and rank in the urine marking test for the male behavioral cohorts 2 and 6 months of age (Figure 4). At 10 months of age there was a significant linear association between tube test rank and urine marking rank, χ 2 (1) = 7.176, P = 0.007, r = 0.409, n = 44. When this cohort was broken down by genotype group, a significant linear association was found for male Grb10 +/p (χ 2 (1) = 5.706, P = 0.017, r = 0.521) mice, but not for WTs.
Additionally, there was a significant linear association between tube test and barbering rank for male mice (pooled genotypes) 10 months of age (χ 2 (1) = 3.993, P = 0.046, r = 0.602, n = 12) ( Figure 5). When the cohort was broken down by genotype group, male WTs (χ 2 (1) = 4.091, F I G U R E 2 Whisker barbering in Grb10 +/p socially house mice. Proportions of whisker barbering subdivided by genotype in A, Male and B, Female behavioral cohorts. Barbering was not present at 2 months, but tended to increase with age P = 0.043, r = 0.905) but not Grb10 +/p mice had a linear association. All other associations between barbering and social tube (male and female) or barbering and urine ranking (male) mice were not significant ( Figure 4). Although the four associations of cage rank above were originally found to be significant, none survived FDR correction.

| DISCUSSION
Our primary goal was to assess social dominance behavior in grouphoused Grb10 +/p mice at multiple ages. Social housing provided a more ecologically relevant context for social dominance strategies optimal in close quarters. Group housed animals benefit from social hierarchies reducing costly conflicts, in contrast to isolation housing, where more territorial and aggressive confrontation strategies are more beneficial. 10, 23 We examined three cohorts, at 2, 6 and 10 months of age, to capture any variation in dominance or hierarchical behaviors that might depend on age. Barbering, for instance, was absent in our 2-month cohorts, and appeared in cohorts 6 and 10 months of age. Male and female mice underwent testing to determine whether sex-specific strategies were differentially impacted by paternal Grb10 deletion. 18 In both sexes and all three age groups, we found no difference between Grb10 +/p and WT socially housed mice in likelihood of winning matches in the stranger-encounter Lindzey tube test,  3 In contrast to males, our Grb10 +/p females were statistically significantly more likely to win in the stranger-encounter Lindzey tube test.
Our data suggest sex-specific effects of isolation on social dominance behaviors in our Grb10 +/p mice. Sex differences in the expression of (presumably) maternal Grb10 in muscle have been noted, 24 but as far as we are aware there are no known sex-differences in terms of paternal Grb10 expression in the brain, 25 22 Nevertheless, a different experimental set up is required to determine within-cage rank stability over time for social F I G U R E 4 No correlation between social dominance measures in mixed cages of Grb10 +/p and WT mice. Win frequency (0, 1, 2 or 3 wins) in the urine marking test was plotted against frequency in the social tube test for each male mouse. A, Males 2 months; B, Males 6 months; C, Males 10 months. There was initially a significant linear association at 10 months, but this did not survive FDR correction groups with Grb10 +/p animals. It is also possible Grb10 +/p mice alter the behavior of WT littermates, as is the case for Cdkn1c BACx1 and Nlgn3. 22,29 Our Grb10 +/p and WT balanced cage set up lacks an appropriate independent control group, like cages of Cdkn1c BAClacZ and WT mice, 22 to test this.
We have showed through robust and convergent testing at multiple ages, and in both sexes, that socially housed Grb10 +/p mice do not show a social dominance phenotype. Nevertheless, following social isolation there is an interaction with Grb10 expression that produces a change in social dominance related behaviors, with a sexually dimorphic direction of effects; critically the direction of effects was contrary to previous findings. 3 We also noted an absence of correlation of hierarchical rank between different dominance tests undertaken by Grb10 +/p containing cages, a pattern of behavior previously proposed to indicate instability of social rank. 22 Taken together, these findings suggest that paternal Grb10 may influence stability of social behavior. Nevertheless, although it is clear from the work here and others 30 that paternal Grb10 does impact on brain function generally, further work is required to determine the exact role played in social behavior.

ACKNOWLEDGMENTS
This work was supported by Wellcome grant 105218/Z/14/Z. ARI is part of the MRC Centre for Neuropsychiatric Genetics and Genomics (G0801418).
F I G U R E 5 No correlation between social dominance measures and barbering in mixed cages of Grb10 +/p and WT mice. Barbering status (0 -subordinate barbered mouse, 1dominant barber) was plotted against win frequencies in the social tube and urine tests (0, 1, 2 or 3 wins). Barbering plotted for A, male mice at 6 months against social tube and urine tests; B, male mice at 10 months against social tube and urine tests; C, female mice at 6 months against social tube test and D, female mice at 10 months against social tube test. There was no barbering at 2 months (See Figure 1)