Maternal care of heterozygous dopamine receptor D4 knockout mice: Differential susceptibility to early‐life rearing conditions

Abstract The differential susceptibility hypothesis proposes that individuals who are more susceptible to the negative effects of adverse rearing conditions may also benefit more from enriched environments. Evidence derived from human experiments suggests the lower efficacy dopamine receptor D4 (DRD4) 7‐repeat as a main factor in exhibiting these for better and for worse characteristics. However, human studies lack the genetic and environmental control offered by animal experiments, complicating assessment of causal relations. To study differential susceptibility in an animal model, we exposed Drd4 +/− mice and control litter mates to a limited nesting/bedding (LN), standard nesting (SN) or communal nesting (CN) rearing environment from postnatal day (P) 2‐14. Puberty onset was examined from P24 to P36 and adult females were assessed on maternal care towards their own offspring. In both males and females, LN reared mice showed a delay in puberty onset that was partly mediated by a reduction in body weight at weaning, irrespective of Drd4 genotype. During adulthood, LN reared females exhibited characteristics of poor maternal care, whereas dams reared in CN environments showed lower rates of unpredictability towards their own offspring. Differential susceptibility was observed only for licking/grooming levels of female offspring towards their litter; LN reared Drd4 +/− mice exhibited the lowest and CN reared Drd4 +/− mice the highest levels of licking/grooming. These results indicate that both genetic and early‐environmental factors play an important role in shaping maternal care of the offspring for better and for worse.

Extensive evidence indicates that poor parental care can contribute to increased vulnerability to develop later-life psychopathology in humans and impaired cognitive performance in rodents. 1,2 This vulnerability crucially depends on a complex cross-talk between an individual's genetic makeup and rearing environment. 3 While the genetic background of some individuals is related to a vulnerable phenotype in the face of early-life adversity, others appear to be more resilient.
Interestingly, individuals who are genetically more susceptible to the detrimental consequences of negative (rearing) conditions may also experience greater benefits from a positive and stimulating (rearing) environment. 4,5 This crossover effect for better and for worse, also called differential susceptibility, is supported by studies investigating the role of human allelic variation across a variety of susceptibility genes. 6 An example of such differentially susceptibility concerns the exon III 7-repeat polymorphism of the D2-like dopamine receptor D4 gene (DRD4-7R). In humans, this variant has been associated with reduced gene expression and efficiency 7,8 and acts as a susceptibility marker of dopamine-related genes. 6 Carriers of this variant have an increased risk of developing externalizing problems in relation to parental insensitivity 9 and chronic stress. 10 However, these individuals also benefitted most from enhanced positive parenting. 11 Meta-analytic evidence further supports an important role of dopamine-related genes in moderating susceptibility to both positive and negative rearing environments. 12 Of note, the DRD4 also plays a role in moderating parental care itself. 13,14 1.2 | Rodent models of impoverished or enriched rearing environments Studying differential susceptibility in humans is hampered by random genetic variability. Moreover, it is often difficult to randomly allocate individuals to specific environments while also taking genotype into account. Therefore, we set out to study the causal contribution of decreased Drd4 functioning to differential susceptibility with a truly randomized experiment in rodents, allowing strict control for both genetic variation and environmental factors. 15 By using heterozygous dopamine receptor D4 knock-out (Drd4 +/− ) mice, we aimed to mimic the reduced DRD4 efficiency observed in human DRD4-7R allele carriers.
We selected two rodent models developed to chronically induce alterations in the quality and quantity of parental care received by offspring. First, limited availability of nesting and bedding (LN) material to a mouse dam was used to induce an adverse early life environment; this model increases unpredictability of maternal care received by the pups, [16][17][18] leading to increased corticosterone levels in pups 19 and altered offspring development and behavior in adulthood. 20,21 Second, as beneficial and stimulating social rearing environment we selected a communal nesting (CN) condition, where two or more dams share care-giving behavior towards multiple litters. 22 In this condition, pups experience higher levels of nest occupancy by at least one dam 18,23 and can interact with peers as well as siblings. Mice reared in communal nesting conditions exhibit various neurobiological and behavioral characteristics that are indicative of improved social competences. 24

| Outcome parameters
In line with a previous study, 18 we focused on timing of puberty onset, a key moment in development that is malleable by environmental influences as part of an adaptive reproductive strategy. 25 Although adverse rearing conditions in females are linked to accelerated pubertal onset in humans 26 and rats, 27 such effects have not yet been observed in mice. 18,28 In human males, adverse rearing conditions had no effect on puberty onset, 29 while puberty onset in male rodents was either unaffected or delayed. 18,27,30 However, rodent models of early-life adversity (ELA) invariably decrease body weight gain, which is an important mediator of puberty onset. Therefore, it is unclear whether the delayed puberty onset observed in ELA reared animals is the result of decreased body weight gain or whether a relative acceleration irrespective of body weight exists in rodents as well.
A second outcome was maternal care provided by female offspring. In addition to sexual maturation, the theory submitted by Belsky et al 25 31 Variations in levels of licking/grooming (LG) behavior and arched-back nursing (ABN), core features of positive parenting in rodents, have been shown to affect corticosterone reactivity, hippocampal development and maternal care of the offspring. 31 In addition, the limited bedding/nesting model, which evokes changes in maternal care, results in aberrant patterns of maternal care of the offspring, 32 whereas mice reared in a communal nesting condition display improved maternal behavior towards their own pups. 33 Taken together, these studies highlight the importance of maternal care for offspring development, as well as the potential of maternal care to be shaped by the early-life environment, contributing to the intergenerational transmission of social behavior.
In this study, we tested heterozygous Drd4 knock-out (Drd4 +/− ) mice and control litter mates on susceptibility to both adverse (LN) and enriched (CN) rearing environments to model differential susceptibility in mice. Animals were examined on (a) puberty onset, to track early development, (b) their own maternal care towards the next generation as an indicator of transgenerational effects and (c) basal corticosterone levels, to investigate involvement of the hypothalamic-pituitary-adrenal-axis (HPA-axis) in differential susceptibility. Although puberty onset would be hypothesized to be accelerated in LN and delayed in CN reared animals according to life history theory, previous findings indicate that the opposite may be true in mice due to the strong effects of body weight. LN reared mice were hypothesized to display poor maternal care, whereas CN reared mice were hypothesized to show enhanced maternal care. To confirm differential susceptibility, these effects would have to be amplified in, or exclusive to, Drd4 +/− mice.

| Animals and housing
B6.129P2-Drd4 tm1Dkg /J (Drd4 +/− ) mice 34 were originally obtained from the Jackson Laboratory (Bar Harbor, Maine, USA) and bred inhouse with C57BL/6JOlaHsd (breeding colony, originally obtained from Harlan, France) mice for at least four generations before experiments started. All breeding was performed in our own animal facility.
Wild-type (wt) female C57BL/6 mice were allowed to breed with male  Table S1. Puberty onset and F1 maternal care were scored by a trained experimenter blind to rearing condition and genotype of the animals. Mice had ad libitum access to water and chow and were housed on a reversed LD cycle (lights off 08:00 AM, temperature 21-22 C, humidity 40%-60%). All experiments were performed in accordance with the EC council directive (86/609/EEC) and approved by the Central Authority for Scientific Procedures on Animals in the Netherlands (CCD approval AVD115002016644).

| Breeding conditions
Breeding was performed as described earlier. 18 In short, one male was paired with two females for 4 days, after which females were co-   building, self-grooming on nest, feeding and self-grooming off nest.
For observations during which the behavior did not qualify for one of these categories, only on or off nest location of the dam was scored. The entropy rate is obtained by calculating the probabilities of certain maternal behaviors predicting specific subsequent behaviors, in which higher entropy rates are indicative of higher unpredictability. In addition, unpredictability of maternal care specifically on the nest site was calculated by pooling all off-nest behaviors to enhance representation of the unpredictability rate as experienced by the offspring. Third, the average number of transitions from and to the nest site per observation was used as an index of fragmentation of maternal care. 19

| Puberty onset F1
As an external measure of puberty onset in males, mice were restrained and gently examined daily from P27 to P33 (10:00-12:00 AM) on the potential to fully retract the prepuce and expose the glans penis which was designated as puberty onset. 35 Female mice were scored daily from P24 to P36 for vaginal opening, here taken as sign of puberty onset. 36 All mice were weighed at puberty onset.

| Maternal care F1
During adulthood (>P70), female F1 mice were allowed to breed with a wild-type male as described for F0. All F2 litters were culled/cross-fostered to six pups and reared in standard nesting conditions. At P2, P9, P14 and P21, clean cages were provided and animals were weighed. Maternal care observations were performed as described for F0 maternal behavior. At P7 between 10:00-12:00 AM, pup retrieval behavior was measured using a 5 minute pup retrieval test as described earlier. 18 If a dam did not retrieve all three pups within 5 minutes, a latency of 300 seconds was assigned.

| Plasma corticosterone levels F1
To measure plasma corticosterone levels, all F1 dams were decapitated in random order between 1:00 and 5:00 PM at least 3 weeks after weaning of F2 litters. Trunk blood was collected in heparin containing tubes (Sarstedt, The Netherlands) on ice and centrifuged for 10 minutes (15 682 rcf) at 4 C. Plasma was collected and stored at −20 C until radioimmunoassay (MP Biomedicals, The Netherlands; sensitivity 3 ng/mL).

| Statistical analysis
Data are expressed as mean ± SEM. Values deviating >3.29 SD from the mean were defined as outlying and winsorized accordingly. 37 The entropy rate of one F0 LN dam was winsorized. Data were analyzed using SPSS 23 (IBM) and litter effects in all F1 measures were accounted for using the SPSS complex samples module. However, no Greenhouse-Geisser corrected repeated measures ANOVAs with breeding condition as the between-subject factor and postnatal day or observation as within-subject factors were used to analyze F0 maternal behaviors. Maternal behaviors from two observation sessions at P2 were analyzed separately to dissociate acute effects of novel environment exposure from more chronic alterations in maternal care. P2 maternal behavior, entropy rates and fragmentation were analyzed using a one-way ANOVA with breeding condition as the between-subjects factor. Pup retrieval latencies of F1 dams were analyzed using cox regression, as this method is preferred if a subset of animals fails to complete a certain task. 38 All other F1 measures were analyzed using a two-way ANOVA including rearing condition and genotype as independent variables. Pearson correlations were used for correlational data. Mediation analysis was conducted using the PROCESS v3 SPSS macro, 39 with rearing condition as a multicategorical independent variable and the SN group as the reference category. The day of puberty onset was used as dependent variable and body weight at weaning and received entropy rates as potential mediators. Significant mediation was assigned when 95% confidence intervals of mediation did not include zero. The maternal care of mouse dams was affected by environmental condition ( Figure 2, Table S1). Arched-back nursing (ABN) levels in LN dams were increased compared to CN dams (Figure 2A), while passive nursing was decreased in CN dams compared to SN dams ( Figure 2B).
Taking the sum of ABN and passive nursing together, total nursing levels displayed by individual CN dams were decreased compared to LN and SN dams ( Figure 2C), but feeding behavior in the CN condition increased ( Figure S1A). Although environmental conditions did not affect licking/grooming behavior from P3-8, LG levels were affected more acutely at P2 ( Figure 2D). Post hoc testing indicated that specifically pups in a LN setting were deprived from LG on this first day of novel environment exposure. Overall nest occupancy of LN dams was increased compared to SN and CN dams ( Figure 2E), but this was The overall unpredictability of behavior displayed by dams was not significantly affected by environmental condition ( Figure 2G).
However, unpredictability of behavior specifically on the nest site (on nest entropy rates) was altered ( Figure 2H). Post hoc comparisons revealed that the LN dams displayed increased unpredictability of maternal care compared to the SN and CN dams. Nesting condition also affected fragmentation of maternal care, measured by the number of transitions from and to the nest site ( Figure 2I); CN dams exhibited increased fragmentation compared to SN and LN dams.

| Effects of rearing conditions on early development
At P14, body weight of LN litters was decreased compared to SN and CN litters ( Figure 2J), an effect that remained at weaning in both males ( Figure 2K) and females (Figure 2l). Puberty onset was also affected by rearing condition in both males ( Figure 3A) and females ( Figure 3F); LN reared animals displayed a delay in puberty onset compared to SN and CN reared mice. In females ( Figure 3G), but not males  Mice that were exposed to LN rearing conditions during early development displayed decreased levels of arched-back nursing (ABN) towards their own offspring compared to SN-reared animals ( Figure 4A). While passive nursing levels were not affected by rearing condition (Figure S2A), total nursing behavior was decreased in LN reared mice compared to CN reared animals ( Figure S2B). In addition, the total time spent on the nest site was decreased in LN-reared animals compared to both SN and CN reared mice ( Figure 4B). A main effect of rearing condition was also observed for the percentage of time dams spent licking/grooming their own pups, a key maternal behavior; LN-reared dams spent less time licking/grooming than dams reared in a communal nesting environment (Figure 4c).
While F0 dams did not differ in total entropy rate, the total entropy rate of F1 maternal behavior was decreased in CN reared mice compared to dams reared in a SN environment ( Figure S2C). In addition, CN-reared dams displayed lower on-nest unpredictability rates compared to LN reared animals ( Figure 4C).  decreased in offspring from a LN reared mother compared to offspring from SN and CN reared dams ( Figure S2E), this was normalized at weaning at P21 ( Figure S2F). Finally, basal levels of blood plasma corticosterone were not affected by rearing condition ( Figure 4F).

| Moderation of rearing condition effects by Drd4 genotype
Different rearing conditions did not interact with Drd4 genotype to determine body weight at weaning (Figure 2) or sexual maturation ( Figure 3). In addition, basal corticosterone levels and most measures of maternal care were not affected by a gene-early environment interaction ( Figure 4). However, an interaction effect was observed for the percentage of time dams spent licking/grooming their own offspring ( Figure 4C). In line with the differential susceptibility theory, Drd4 +/− dams reared in the LN environment exhibited the lowest LG levels, whereas CN reared Drd4 +/− mice spent the most time licking/ grooming their own pups.

| DISCUSSION
In this study, we examined the causal role of Drd4 in differential susceptibility to the environment using a randomized experiment in rodents, allowing strict control for both genetic variation-using

| Modeling impoverished and enriched rearing environments
The pattern of F0 maternal care resulting from exposure to the LN condition was largely in line with earlier findings using this model. [16][17][18][19] While different pup-directed maternal behaviors remained relatively unaltered, the unpredictability of maternal behavior, particularly on the nest site, increased. In addition, pups in the LN condition were deprived from normal levels of licking/grooming upon first exposure to this condition on P2, whereas LG levels were similar to the SN and CN conditions from P3-P8. In contrast to other reports, but in line with previous findings from our lab, 18 fragmentation of maternal care was similar to control conditions, a difference that could be due to the difference in timing of observations. In this study, maternal behaviour was observed predominantly during the dark phase of the animals, whereas previous studies focused more on the light phase of the day/night cycle. 16,19 This difference in timing of observations is important as we observed, in line with earlier reports from our lab, 18

| Rearing conditions affect sexual maturation
The delayed puberty onset observed in both male and female LN reared mice was mediated by a decrease in body weight gain at weaning. The importance of body weight and leptin in regulating puberty onset is well-known for both humans 40,41 and rodents. 42 We therefore also measured body weight at puberty onset for the adolescent mice that were raised in different early life conditions. The minimal differences in body weight at puberty onset suggest that, irrespective of early life background and subsequent body weight at weaning, the majority of mice postpone the onset of puberty until a certain body weight is reached. This is in contrast to a recent study where body weight at vaginal opening was increased in female mice that were reared in a LN condition from P2-9. 28 However, because body weight at weaning of control groups is similar in both studies, this is unlikely to be a result of measurement differences. Future studies should therefore help to elucidate whether body weight at puberty onset is consistently affected by limited nesting rearing conditions.  54 Human studies also link the DRD4-7R genotype to alterations in components of the HPA-axis. Gene-early environment effects have been observed for basal cortisol in children, 53 as well as stress induced cortisol levels of young adults. 55 A prominent role for alterations in circulating basal corticosterone levels in adulthood is not supported by our data. However, stress reactivity was not assessed and could, at least in part, underlie the observed alterations in maternal care.
Other systems may also be critical in the mechanism underlying differential susceptibility. Recent studies using different molecular tools and mouse knock-in models have begun to unravel the exact function of the DRD4-7R in corticostriatal glutamatergic neurotransmission, enhancing our understanding of the Drd4 receptor and susceptibility to the environment. 56,57 Other studies used a wide array of techniques to show the involvement of other dopamine receptors in mediating the social deficits observed after severe early-life stress. 58 At a meta-analytic level, however, the effects of early-life adversity on the dopaminergic system appear limited, although significant for some parameters and areas. 59 It is important to note that none of the studies included in the meta-analysis examined Drd4 as a potential target, highlighting the lack of preclinical evidence on the role of Drd4 expression in mediating effects of adverse rearing conditions. The advances in our understanding of Drd4 functioning at a molecular level and the role of other dopamine receptors in regulating susceptibility will help to guide future studies into the role of DRD4.
Finally, there is increasing awareness that most consequences of early-life rodent models have small effect sizes, 21 which is also the case in our study. Although we have sizable group numbers compared to common practice in the field, we should take this into consideration and interpret the results with care. To increase statistical power in future experiments, animal numbers should be adapted to realistically expected effect sizes and animal ethical committees should be aware of this. 60 Moreover, more meta-analyses in this field should be stimulated and can help in designing future studies. 21

| CONCLUSION
The research presented here provides a translational approach to examine the contribution of the Drd4 gene in differential susceptibility. While other preclinical studies on differential susceptibility in socially monogamous prairie voles focused on the role of prenatal Research (Spinoza Prize). A preprint version of this article was uploaded at bioRxiv (http://www.biorxiv.org).

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available Rixt van der Veen https://orcid.org/0000-0002-0808-7396