Oxytocin receptor gene (OXTR) variant rs1042778 moderates the influence of family environment on changes in perceived social support over time

Running head: OXTR and social support across life span Oxytocin receptor gene (OXTR) variant rs1042778 moderates the influence of family environment on changes in perceived social support over time Henrik Dobewall1,2, Christian Hakulinen1, Liisa Keltikangas-Järvinen1, Laura PulkkiRåback1,3, Ilkka Seppälä4, Terho Lehtimäki4, Olli T. Raitakari5, and Mirka Hintsanen6,1,* 1 Department of Psychology and Logopedics, Faculty of Medicine, University of Helsinki, Helsinki, Finland. 2 Faculty of Social Sciences, Health Sciences, University of Tampere, Tampere, Finland 3 Helsinki Collegium for Advanced Studies, University of Helsinki, Helsinki, Finland. 4 Department of Clinical Chemistry, Fimlab Laboratories, Faculty of Medicine and Life Sciences, University of Tampere, Finland. 5 Research Collegium for Applied and Preventive Cardiovascular Medicine, University of Turku, Turku, Finland. 6 Research Unit of Psychology, University of Oulu, Oulu, Finland. *Correspondence to Prof. Mirka Hintsanen, Unit of Psychology, University of Oulu, Oulu, Finland. P.O.Box 2000 (Yliopistokatu 9), 90014 University of Oulu, Oulu, Finland; E-mail: [email protected]; gsm +358 50 569 5243. © 2018. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ This is the post print version of the article, which has been published in Journal of affective disorders 2018, 235, 480-488. The final publication is available via https://doi.org/10.1016/j.jad.2018.04.008 2 Highlights Perceived Social Support: •We examined environmental and genetic determinants of perceived social support •Reporting longitudinal changes from adolescence to middle adulthood •Family environment was strongly associated with the initial level of social support •Rate of change in perceived family support was dependent on OXTR genotype Abstract: Background: Lack of social support is an established risk factor across health outcomes, making it important to examine its family environmental and genetic determinants. Methods: In a 27-year follow-up of the Young Finns Study (N=2341), we examined with a latent growth curve model whether genes involved in the oxytocin signaling pathway—namely, oxytocin receptor gene (OXTR) variants rs1042778, rs2254298, and rs53576—moderate the effect of early-life social experiences on perceived social support across the life span. Mothers reported the emotional warmth and acceptance towards their children at baseline when the participants were from 3 to 18 years old (1980). Perceived family support and support from friends and peripheral sources were assessed in five follow-ups 18 years apart (1989-2007). Results: Maternal emotional warmth and acceptance predicted the initial level of perceived social support across subscales, while the rate of change in family support was affected by the family environment only if participants carried the T-allele of OXTR rs1042778. This gene-environment interaction was not found for the rate of 3 change in support from friends and peripheral sources and we also did not find associations between latent growth in perceived social support and OXTR variants rs53576 and rs2254298. Limitations: Selective attrition in perceived social support, maternal emotional warmth and acceptance, gender, and SES. Family environment was assessed by a non-standardized measure. Conclusions: OXTR rs1042778 polymorphism seems to contribute to changes in perceived family support in that way that some individuals (T-allele carriers) ‘recover’, to some extent, from the effects of early-life social experiences, whereas others (G/G genotype carriers) do not. Conflict of interest statement: All authors declare they have no conflicts of interest and have no financial interest. Contributors: All the authors have made a substantial contribution to designing or carrying out the research, writing or revising the manuscript, or providing guidance on the execution of the research. All analyses have been conducted by the first author. The paper has been read and approved by all the authors. The Young Finns Study has been financially supported by the Academy of Finland: grants 258578 (M.H.), 286284 (T.L.), 134309 (Eye), 126925, 121584, 124282, 129378 (Salve), 117787 (Gendi), and 41071 (Skidi); the Social Insurance Institution of Finland; Kuopio, Tampere and Turku University Hospital Medical Funds (grant X51001 for T.L.); Juho Vainio Foundation; Paavo Nurmi Foundation; Finnish Foundation of Cardiovascular Research; Finnish Cultural Foundation; Tampere 4 Tuberculosis Foundation (T.L.); Emil Aaltonen Foundation (T.L.); Yrjö Jahnsson Foundation (T.L.) and Signe ja Ane Gyllenberg’s Foundation (T.L.). The listed funding sources had no role in the design of this study and had any role during its execution, analyses, interpretation of the data, or decision to submit results. 5 Oxytocin receptor gene (OXTR) variant rs1042778 moderates the influence of family environment on changes in perceived social support over time Introduction Humans are highly social animals whose lives are dependent on other humans. A rich literature has demonstrated that social support is a key determinant in human health and well-being (Berkman, Glass, Brissette, & Seeman, 2000; Holt-Lunstad et al., 2015; Uchino, Cacioppo, & Kiecolt-Glaser, 1996; Yang et al., 2016). In addition, social support is especially relevant to mental health as it has been associated with better mental health both directly (Gariépy, Honkaniemi, & Quesnel-Vallée, 2016) and indirectly (Cohen & Wills, 1985). Broadly, social support can be divided into two categories, that is, one characterized by structural aspects (e.g., the size of social networks or frequency of contacts) and the other by emotional or informational aspects (perceptions of availability of support). Although some studies have examined social network changes (i.e., structural social support) across the lifespan (e.g., Wrzus, Hänel, Wagner, & Neyer, 2013), little is known about how perceptions of social support change over different developmental phases (Antonucci et al., 2014; Carstensen et al., 1999). Available evidence on social support perceptions across the lifespan is mostly based on cross-sectional studies (Pavlova, Körner, & Silbereisen, 2015; Prezza & Pacilli, 2002) or is limited to old age (Costa, Zonderman, & McCrae, 1985; Hakulinen et al., 2016; van Tilburg, 1998), yet, these studies have found that family support is rather stable across all ages while support from friends and peripheral sources decrease throughout adulthood. To our knowledge, no longitudinal study has focused on changes in perceived social support from adolescence to middle adulthood. Accordingly, little attention has been paid to family environmental and 6 genetic determinants of perceived social support across the lifespan (cf. Antonucci, Ajrouch, & Birditt, 2014). Family environment Adolescence and adulthood perceptions of social support might reflect internalized representations of the family environment (see, e.g., Herzberg et al., 1999) and, thus, child-rearing practices and the parent-child relationship quality are probably important for how individuals’ experience social support in later life. We argue that this is the case because especially the family environment is an origin of social behaviors and close relationships which are reflected in individuals’ perceived support across the lifespan. Most theories agree that emotional warmth between parent and child (i.e., their love and connectedness) is a main component of a beneficial family environment (Clark & Ladd, 2000; Maccoby, 1980; MacDonald, 1992; Schaefer, 1959). MacDonald (1992), for instance, conceptualizes warmth as a reward system that facilitates cohesive relationships within the family and parents’ investment in their children and their emotional bond. Another, yet related, basic characteristic of the family environment is parental acceptance of child’s feelings, opinions, and actions (Maccoby, 1980; Schaefer, 1959). Acceptance is a measure of parents’ tolerance and responsiveness toward the child. Low levels of acceptance further capture ambivalent child-rearing experiences (Donath, 2015; Shelton & Johnson, 2006), such as, parental perceptions of the child as demanding and requiring a lot of time and attention, and even perceiving the child in some situations as burdensome (Keltikangas-Järvinen, 2002). Thus, emotional warmth and acceptance are markers of the early-life social experience within a family, which reflect variation in normal parenting experienced 7 by most individuals during their childhood. To our best knowledge, however, the associations between maternal emotional warmth and acceptance and perceived social support have not been studied longitudinally to date. Genetic background of perceived social support With regard to specific genetic pathways, the role of oxytocin (OT) signaling is of recent interest in understanding human social behavior and close relationships (Kumsta & Heinrichs, 2013). The neuropeptide OT has been held responsible for regulating social cognition (Peltola et al., 2014; Skuse et al., 2014; Zink & MeyerLindenberg, 2012) as well as various social and affective traits and states (Ebstein, Knafo, Mankuta, Chew, & Lai, 2012; Olff et al., 2013). Perceived social support appears to have a biological makeup (i.e., is partly heritable) (Kessler, Kendler, Heath, Neale, & Eaves, 1992) and genes involved in OT signaling might contribute to inter-individual differences in perceived social support. Here, we focus on the most common type of genetic variation among individuals: single nucleotide polymorphisms (SNPs). SNPs represent naturally occurring variations in a single DNA building block. For example, an SNP may replace the nucleotide Guanine (G) with the nucleotide Adenine (A) or Thymine (T) in a certain stretch of DNA. These genetic differences might be associated with complex traits, such as perceived social support and might also modulate an individual’s susceptibility to environmental influences (U.S. National Library of Medicine, 2017). So far, three studies have investigated the relationship between oxytocin receptor gene (OXTR) variant rs53576 and various sources of social support, such as, experimental manipulated presence versus absence of a friend (Chen et al., 2011), the level of perceived unsupportive social interactions with parents and peers (McInnis, 8 McQuaid, Matheson, & Anisman, 2015), or the self-rated quality of seven aspects of support across relationships with family and nonfamily members (Hostinar, Cicchetti, & Rogosch, 2014). None of these studies, however, have found direct associations between OXTR rs53576 and perceived social support. Gene-environment (GxE) interactions The fact that previous research did not find direct effects of oxytocin pathway genes in social support perceptions might be explained by GxE interaction. It is possible that the associations between the family environment and perceived social support might be different in different OXTR genotype carriers, which makes it important to examine GxE interactions in perceived social support across adolescence and adulthood. Generally, research on GxE interactions has to differentiate between at least three complementary perspectives: The diathesis-stress perspective is guided by the theoretical assumption that some individuals are more vulnerable than others to adverse environments for genetic (or temperamental) reasons (Caspi et al., 2002; Rutter, Moffitt, & Caspi, 2006). A logical complement to this concept was recently coined vantage sensitivity (Pluess & Belsky, 2013), claiming that there are interindividual differences in the response to positive environmental influences as a function of opportunity characteristics. The differential susceptibility hypothesis, finally, states that some individuals are more susceptible than others to all environmental effects irrespective of their nature (Belsky & Pluess, 2009). Those individuals who carry certain “plasticity-alleles” are thus not just vulnerable to environmental risks (as the diathesis-stress perspective suggests) but also sensitive to effects of neutral or beneficial environments (as the concept of vantage sensitivity 9 implies). What all three perspectives have in common, is that for other individuals the same environmental influences might have only a weak effect, if any. Moreover, genes might not only modulate the effect an environmental determinant has on initial differences in a social phenotype but also influence whether these differences become more pronounced, decrease, or even disappear as individuals grow older. Therefore, common polymorphisms in OT pathway genes might moderate how salient the family environment is in perceptions of social support (Kumsta & Heinrichs, 2013). For instance, maltreated adolescents carrying the G/G genotype in OXTR variant rs53576 are shown to have lower perceived social support as compared to carriers of the A-allele, who are not affected by this environmental influence (Hostinar et al., 2014). Three candidate genes appear most promising for examining GxE interactions in perceived social support: The first two SNPs, rs53576 and rs2254298 (both G to A), have been targeted at in recent meta-analytical (Bakermans-Kranenburg & van Ijzendoorn, 2014; Li et al., 2015) and theoretical reviews (e.g., Brune, 2012). The third SNP, rs1042778 (G to T), is an early-discovered candidate gene (Israel et al., 2009), which has repeatedly been mentioned in literature on OXTR-social behavior linkages. For all three SNPs functional linkages to either alterations of oxytocinergic brain regions as responses to social cues (Inoue et al., 2010; Tost et al., 2010), OT plasma levels (Feldman et al., 2012), and/or DNA CpG methylation (Smearman et al., 2016) have been reported. The current study As early-life social experiences can lay a foundation for human development across the lifespan, the family environment might continue to influence individuals’ 16 from the current study. The frequencies of the three OXTR SNPs rs1042778 (p=.99), rs2254298 (p=.31), and rs53576 (p=.09) did not deviate from the Hardy-Weinberg equilibrium, yet the A-allele of rs2254298 was quite uncommon, and therefore carriers of A/A and G/A genotypes were combined in the subsequent analyses. For the two other OXTR SNPs, an additive genetic model was applied (SNPs are coded 0, 1, 2; counting the copies of the G-allele). By doing so, we assume that G/G genotype carriers have a linearly stronger genetic predisposition for social vulnerability/plasticity than carriers of one or two minor alleles of these OXTR SNPs, as the Aand T-alleles have repeatedly been shown to be associated with a social deficits and impairments (Bakermans-Kranenburg & van Ijzendoorn, 2014; Feldman et al., 2012; Hostinar et al., 2014; Inoue et al., 2010; Li et al., 2015; Smearman et al., 2016; Tost et al., 2010). Insert Table 1 about here Results Multiple indicator latent growth curve models of perceived social support For the two perceived social support scales, the global fit indices indicated that a linear trajectory describes the data well. RMSEA (<.05) and CFI suggested a good fit (≥.97) for family support (Table 2) across models. For support from friends and peripheral sources RMSEA (≤.06) suggested a good fit and CFI an acceptable fit (≥.92; Table 3). Insert Tables 2-3 about here 17 In the initial developmental models (Model 1), the point estimates of the latent slope indicated that participants in average increased in family support (.15) but decreased in support from friends and peripheral sources (-.13). The initial level (i.e., the latent intercept) of perceived social support (across scales) was higher in women than in men and in those participants with high socioeconomic status as compared to low socioeconomic status participants. The rate of change (i.e., the latent slope) of perceived support from friends and peripheral sources, however, decreased significantly faster for female participants. The older the participants were at baseline, the higher was their initial level of perceived support provided by friends and peripheral sources. The negative regression coefficient of age on latent slopes indicated that perceived family support increased somewhat slower and perceived support from friends and peripheral support decreased somewhat faster for those participants who entered the study at a younger age. The inclusion of a period effect for the 1998 wave, as expected, affected the intercept of perceived family support and the slope of perceived support from friends and peripheral sources significantly. Family environment, which combines mothers’ ratings of emotional warmth and acceptance, was strongly associated with the initial level of both perceived family support and perceived friends and more peripheral support (Model 2). There was, however, no direct effect of the family environment on the rate of change in perceived social support. We did not observe any significant direct effects of the three OXTR candidate genes on latent growth in social support (Model 3). In Model 4 interaction was found between the number of T-alleles of rs1042778 and variation in maternal emotional warmth and acceptance on the rate of 18 change of perceived family support, which remained significant when Bonferroni correction was conducted (p-value of <.017). For rs1042778 G/G genotype carriers, change in perceived family support over time was not dependent on family environment. For carriers of one or two copies of the T-allele of rs1042778, on the contrary, change in perceived family support over time was associated with the family environment. When maternal emotional warmth and acceptance were low, rs1042778 T-allele carriers increased at a faster rate in perceived family support, while the pattern was reversed if these participants were raised by mothers that described the family environment as beneficial. Insert Figure 2 about here The sample was further split into three groups according to high, average, and low levels of maternal emotional warmth and acceptance. We estimated for these groups their latent means to capture the change in family support across waves (T1T5). Figure 2 illustrates the consequences of this GxE interaction by comparing rs1042778 T/T genotype, G/T genotype, and G/G genotype carriers. The figure shows that rs1042778 G/G genotype carriers who were raised with high maternal emotional warmth and acceptance reported, across the lifespan, higher levels of perceived family support than those who grew up with low levels. G/T genotype carriers of rs1042778 also started at a level of perceived family support that was dependent on whether they were raised in a beneficial family environment or not. However, these differences decreased steadily as these T-allele carriers grew older, until, at T5 (when participants were from 30 to 45 years old), no visible differences remained. T/T genotype carriers 19 of rs1042778 were in the long run, compared to G/G genotype carriers, also less affected by the maternal emotional warmth and acceptance during childhood. Discussion Current study results indicate that the family environment may be directly associated with variation in perceived social support but not with its development over time. OXTR genotype was not directly associated with perceived social support, but we found a GxE interaction for the family support domain showing that the influence of family environment does not last for some individuals depending on their rs1042778 genotype. These GxE interactions were not observed for rs2254298 and rs53576 or support from friends and peripheral sources. Presented results are in line with the differential susceptibility hypothesis (Belsky & Pluess, 2009), stating that some individuals are genetically more susceptible to environmental influences irrespective of their nature, while for other individuals, the same environmental determinants might have no effect. Our findings indicate that, although there are initial differences in perceived social support, they disappear entirely for carriers of the G/T genotype of rs1042778 and weaken for T/T genotype carriers, while the initial differences in perceived family support seem to persist over time for individuals with the G/G genotype. Please note that, in our main analyses, we have followed an additive inheritance model, which assumes linear effects. Overall, the findings suggest that rs1042778 T-allele carriers are likely to ‘recover’ from the effects of low maternal emotional warmth and acceptance whereas G/G genotype carriers do not show such a pattern. In them, the effect of the family environment on social support seems to continue throughout the lifespan, which indicates that early-life social experience “programs” how G/G genotype carriers perceive social support later in life. This has importance for potential preventions. 20 Similar “programming” has been found for instance for environmental risk on children’s attachment (Bakermans-Kranenburg & van Ijzendoorn, 2007). Current findings should be replicated to verify whether the genetic moderation of OXTR rs1042778 is conditional to age, as suggested by our results. This observation suggests that, for some individuals, later life outcomes have their origin in childhood experiences, while others are less influenced by their family history. However, this lower susceptibility seems to come at a cost: it appears that for these individuals, a beneficial family environment does not continue to have a positive influence on later social support. Current results further add to the existing literature on GxE interaction in social support (Hostinar et al., 2014), which has found that OXTR variant rs53576 moderated the influence of adverse environmental influences. At the same time, they are at odds with these previous works as our environmental determinant did not interact with rs53576 polymorphism. Brune (2012) has further argued theoretically that OXTR polymorphism rs2254298 might confer differential susceptibility and not merely vulnerability for psychopathological characteristics of individuals. OXTR rs1042778, to our best knowledge, has previously not been proposed as a susceptibility gene. That we did not find a direct effect of OXTR rs1042778, rs2254298, and rs53576 is in line with earlier research on social support (Chen et al., 2011; Hostinar et al., 2014; McInnis et al., 2015). However, earlier studies have found polymorphisms in our three candidate genes to be related to differences in social phenotypes that are linked to social support, such as, loneliness (Lucht et al., 2009), trust behavior (Krueger et al., 2012) and, most importantly, emotional support seeking (Kim et al., 2010). Our GxE findings might explain why literature on the role of OXTR in human social behavior is inconclusive because rs1042778 polymorphisms 21 moderated the influence of an early-life social experience on perceived social support instead of having a direct genetic association. Thus, it is likely that the associations between social support and other social environments may also be different in different OXTR genotype carriers. The family environment had a direct effect on the initial level of both perceived social support domains. It may be that children form internal representations of maternal emotional warmth and acceptance, which in turn continue to influence perceptions of both family support and support from friends and more peripheral sources. Prior to this study family environmental determinants of perceived social support were essentially unknown (for an exception, see Herzberg et al., 1999), even though child-rearing practices and maternal warmth and aceptance have repeatedly been shown to affect the development of personality (e.g., Maccoby, 1980; Hakulinen et al., 2013) and the later adjustment of the offspring (e.g., Gluschkoff et al., 2017). We found in longitudinal data that family support increased and support from friends and peripheral sources decreased over the observed age span from 12 (age of the youngest age cohort at T1) to 45 (age of the oldest age cohort at T5) years. That perceived social support provided by friends and more peripheral sources decrease from adolescence to middle adulthood is in line with leading developmental models on social support (Antonucci, Ajrouch, & Birditt, 2014; Carstensen, Isaacowitz, & Charles, 1999) and matches available cross-sectional evidence (Pavlova, Körner, & Silbereisen, 2015; Prezza & Pacilli, 2002). Our findings that perceived family support actually increases from adolescence to middle adulthood rather than being stable is somewhat surprising (see Wrzus et al., 2013). Normative, age-related life events might explain this finding, as middle adulthood is the period in which participants 22 start to have children of their own, which in turn might increase the contact frequency to their parents (Bhattacharya et al., 2016). Limitations and Strengths The current study has some limitations. The study results were affected by attrition, which was somewhat selective for perceived social support and maternal environmental warmth and acceptance, but also gender and socio-economic status. Family environment was assessed by a non-standardized measure. The mother-reports of the emotional warmth and acceptance might have rather different meanings with children aged 18 years compared to aged 3 years, we, thus, standardized the environmental determinant within age group to overcome this limitation. However, we did not analyze the birth cohorts separately, as suggested by Mehta and West (2000), because that would have reduced the statistical power to detect genetic moderation. Further, we discovered a period effect for the first assessment of social support which altered participants’ answers across all ages and consequently had to be controlled for statistically. It was also not possible with our data to control for additional intervening variables, such as life events that might explain some of the observed changes in social support over time (e.g., marriage, divorce, or having children). Last, although around fifth of the participants has reported using antidepressants at some point of their life (Hintsa et al., 2016), most participants were healthy, and thus a generalization of current results to clinical settings is limited. The strengths of the current study include our longitudinal population-based data, analyzing change in the dependent variable over 18 years until the middle adulthood, and using four repeated measurements with 2341 participants based on a well-established scale that assesses functional social support in two meaningful 23 relationship domains. Most previous GxE interaction studies have not had enough statistical power to detect robust associations (Duncan & Keller, 2011). Further, the reported associations are not confounded by common method variance as the family environment was reported by mothers and adolescence, and adulthood social support was reported the participants. Finally, we examined the effect of three most promising polymorphisms in genes involved in the oxytocin signaling pathway instead of selecting a single SNP. Conclusions The current study showed that an OXTR candidate gene (rs1042778) moderates the influence of the family environment (i.e., maternal emotional warmth and acceptance) on perceived social support later in life. The observed GxE interaction has important theoretical implications. Participants who carry certain genes in the oxytocin signaling pathway—namely, the G/G genotype of OXTR variant rs1042778, profited from the more beneficial family environment during childhood, but the maternal emotional warmth and acceptance did not have an effect on changes in their perceptions of family support over time. Initial family support levels of carriers of the T-allele of rs1042778 were as well affected by their family environment, yet, the effect of this environmental determinant faded so that, in the long run, these individuals were less susceptible to the influence of maternal emotional warmth and acceptance than others. Taken together, our findings can be interpreted in a way that, with age, some individuals might become less affected by family environmental determinants of perceived social support for genetic reason, whereas others (G/G carriers) are “programmed” by early-life social experiences to perceive social support in certain way later in life. Our findings also have potential practical implications. If replicated (see, e.g., 24 Keers & Pluess, 2017), current findings suggest that intervention or prevention programs designed to improve childhood environment may lead to higher social support over the life course, particularly among those with genetic sensitivity. 25 Acknowledgments The Young Finns Study has been financially supported by the Academy of Finland: grants 258578 (M.H.), 286284 (T.L.), 134309 (Eye), 126925, 121584, 124282, 129378 (Salve), 117787 (Gendi), and 41071 (Skidi); the Social Insurance Institution of Finland; Kuopio, Tampere and Turku University Hospital Medical Funds (grant X51001 for T.L.); Juho Vainio Foundation; Paavo Nurmi Foundation; Finnish Foundation of Cardiovascular Research; Finnish Cultural Foundation; Tampere Tuberculosis Foundation (T.L.); Emil Aaltonen Foundation (T.L.); Yrjö Jahnsson Foundation (T.L.) and Signe ja Ane Gyllenberg’s Foundation (T.L.). 32 promotes human trust behavior. Frontiers in Human Neuroscience, 6:4. doi:10.3389/fnhum.2012.00004 Li, J., Zhao, Y., Li, R., Broster, L. S., Zhou, C., & Yang, S. (2015). Association of oxytocin receptor gene (OXTR) rs53576 polymorphism with sociality: A meta-analysis. PLoS One, 10(6), e0131820. doi:10.1371/journal.pone.0131820 Lucht, M. J., Barnow, S., Sonnenfeld, C., Rosenberger, A., Grabe, H. J., Schroeder, W., . . . Rosskopf, D. (2009). Associations between the oxytocin receptor gene (OXTR) and affect, loneliness and intelligence in normal subjects. Progress in Neuro-Psychopharmacology & Biological Psychiatry, 33(5), 860-866. doi:10.1016/j.pnpbp.2009.04.004 MacDonald, K. (1992). Warmth as a developmental construct: An evolutionary analysis. Child Development, 63(4), 753-773. doi:10.1111/j.14678624.1992.tb01659.x Makkonen, T., Ruoppila, I., Rönkä, T., Timonen, S., Valvanne, L., & Österlund, K. (1981). Operation Family (Child report No. A 34). Helsinki: Mannerheim League of Child Welfare. Mehta, P. D., & West, S. G. (2000). Putting the individual back into individual growth curves. Psychological Methods, 5(1), 23–43. doi:10.1037/1082989X.5.1.23 McInnis, O. A., McQuaid, R. J., Matheson, K., & Anisman, H. (2015). The moderating role of an oxytocin receptor gene polymorphism in the relation between unsupportive social interactions and coping profiles: Implications for depression. Frontiers in Psychology, 6, 1133. doi:10.3389/fpsyg.2015.01133 Olff, M., Frijling, J. L., Kubzansky, L. D., Bradley, B., Ellenbogen, M. A., Cardoso, C., . . . vanZuiden, M. (2013). The role of oxytocin in social bonding, stress 33 regulation and mental health: An update on the moderating effects of context and interindividual differences. Psychoneuroendocrinology, 38(9), 1883-1894. doi:10.1016/j.psyneuen.2013.06.019 Pavlova, M. K., Körner, A., & Silbereisen, R. K. (2015). Perceived social support, perceived community functioning, and civic participation across the life span: Evidence from the former East Germany. Research in Human Development, 12(1-2), 100-117. doi:10.1080/15427609.2015.1010351 Peltola, M. J., Yrttiaho, S., Puura, K., Proverbio, A. M., Mononen, N., Lehtimäki, T., Leppänen, J. M. (2014). Motherhood and oxytocin receptor genetic variation are associated with selective changes in electrocortical responses to infant facial expressions. Emotion, 14(3), 469-77. doi: 10.1037/a0035959. Pluess, M., & Belsky, J. (2013). Vantage sensitivity: Individual differences in response to positive experiences. Psychological Bulletin, 139(4), 901-916. doi:10.1037/a0030196 Prezza, M., & Pacilli, G. M. (2002). Perceived social support from significant others, family and friends and several socio‐demographic characteristics. Journal of Community & Applied Social Psychology, 12(6), 422-429. doi:10.1002/casp.696 Raitakari, O. T., Juonala, M., Ronnemaa, T., Keltikangas-Jarvinen, L., Rasanen, L., Pietikainen, M., . . . Viikari, J. S. (2008). Cohort profile: The cardiovascular risk in young Finns study. International Journal of Epidemiology, 37(6), 1220-1226. doi:10.1093/ije/dym225 Rutter, M., Moffitt, T. E., & Caspi, A. (2006). Gene-environment interplay and psychopathology: Multiple varieties but real effects. Journal of Child 34 Psychology and Psychiatry, 47(3-4), 226-261. doi:10.1111/j.14697610.2005.01557.x Savelieva, K., Keltikangas-Järvinen, L., Pulkki-Råback, L., Jokela, M., Lipsanen, J., Merjonen, P., ... & Hintsanen, M. (2016). Intergenerational transmission of qualities of the parent–child relationship in the population-based Young Finns Study. European Journal of Developmental Psychology. doi: 10.1080/17405629.2016.1230057 Schaefer, E. S. (1959). A circumplex model for maternal behavior. The Journal of Abnormal and Social Psychology, 59(2), 226-235. doi:10.1037/h0041114 Shelton, N., & Johnson, S. (2006). I think motherhood for me was a bit like a doubleedged sword: The narratives of older mothers. Journal of Community and Applied Social Psychology, 16(4), 316. doi: 10.1002/casp.867 Skuse, D. H., Lori, A., Cubells, J. F., Lee, I., Conneely, K. N., Puura, K., . . . Young, L. J. (2014). Common polymorphism in the oxytocin receptor gene (OXTR) is associated with human social recognition skills. PANAS, 111(5), 1987-1992. doi:10.1073/pnas.1302985111 Smearman, E. L., Almli, L. M., Conneely, K. N., Brody, G. H., Sales, J. M., Bradley, B., . . . Smith, A. K. (2016). Oxytocin receptor genetic and epigenetic variations: Association with child abuse and adult psychiatric symptoms. Child Development, 87(1), 122-134. doi:10.1111/cdev.12493 Smith, E. N., Chen. W., Kähönen, M., Kettunen, J., Lehtimäki, T., Peltonen, L., … Murray, S. S. (2010) Longitudinal genome-wide association of cardiovascular disease risk factors in the Bogalusa heart study. PLoS Genetics, 6(9):e1001094. doi:10.1371/journal.pgen.1001094 35 Tost, H., Kolachana, B., Hakimi, S., Lemaitrea, H., Verchinski, B. A., Mattay, V. S., . . . Meyer–Lindenberg, A. (2010). A common allele in the oxytocin receptor gene (OXTR) impacts prosocial temperament and human hypothalamic-limbic structure and function. PANAS, 107(31), 13936–13941. doi:10.1073/pnas.1003296107 Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser, J. K. (1996). The relationship between social support and physiological processes: a review with emphasis on underlying mechanisms and implications for health. Psychological Bulletin, 119(3), 488-531. doi:10.1037/0033-2909.119.3.488 U.S. National Library of Medicine (2017). What Are Single Nucleotide Polymorphisms (SNPs)?, Genetics Home Reference. Retrieved from http://ghr.nlm.nih.gov/handbook/genomicresearch/snp Wrzus, C., Hänel, M., Wagner, J., & Neyer, F. J. (2013). Social network changes and life events across the life span: A meta-analysis. Psychological Bulletin, 139(1), 53-80. doi:10.1037/a0028601. Van Tilburg, T. (1998). Losing and gaining in old age: Changes in personal network size and social support in a four-year longitudinal study. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 53(6), 313-323. doi:10.1093/geronb/53B.6.S313 Yang, Y. C., Boen, C., Gerken, K., Li, T., Schorpp, K., Harris, K. M. (2016). Social relationships and physiological determinants of longevity across the human life span. PANAS, 113(3), 578-583. doi: 10.1073/pnas.1511085112. Zimet, G. D., Powell, S. S., Farley, G. K., Werkman, S., & Berkoff, K. A. (1990). Psychometric characteristics of the Multidimensional Scale of Perceived 36 Social Support. Journal of Personality Assessement, 55(3-4), 610-617. doi:10.1080/00223891.1990.9674095 Zink, C. F., & Meyer-Lindenberg, A. (2012). Human neuroimaging of oxytocin and vasopressin in social cognition. Hormones and Behavior, 61(3), 400-409. doi:10.1016/j.yhbeh.2012.01.016 37 Table 1. Demographic Statistics of the Study Variables Variable No.of Cases Mean(%) SD Men 1058 (45) Women 1283 (55) Age in 1989 2341 19.45 5.02 Socioeconomic status 2322 0.02 0.66 Family environment (raw) 2341 4.22 0.47 rs1042778 T/T 352 (15) G/T 1112 (48) G/G 877 (37) rs2254298 A-allele 380 (16) G/G 1961 (84) rs53576 A/A 446 (19) A/G 1105 (47) G/G 790 (34) Perceived family support (T1) 1981 3.93 0.92 (T2) 1779 4.11 0.87 (T3) 1595 4.15 0.90 (T4) 1722 4.24 0.86 (T5) 1712 4.19 0.88 Perceived support from friends and peripheral sources (T1) 1986 3.91 0.93 (T2) 1779 4.21 0.85 (T3) 1595 4.31 0.81 (T4) 1728 4.22 0.88 (T5) 1712 4.19 0.90 Note. Perceived social support subscales were calculated as a mean of four / eight items. Family environment was calculated as a mean of seven items assessing maternal emotional warmth and acceptance. Socioeconomic status, measured with 38 parental education and income, was standardized within the initial sample (0 representing the average adult Finn at baseline). 39 39 Table 2. Perceived Family Support Model Intercept (SE) p Slope (SE) p Model 1 Developmental model Latent means 3.962 0.068 <0.001 0.152 0.050 0.002 Gender (male=1; female=0) -0.125 0.038 0.001 -0.017 0.028 0.559 Age 0.033 0.019 0.077 -0.071 0.014 <0.001 Socioeconomic status 0.051 0.022 0.018 0.027 0.016 0.089 Period effect for 1989 (T1=1; T2-T5=0) 0.180 0.068 0.008 -0.022 0.049 0.654 Model 2 Adding direct effect of family environment Emotional warmth and acceptance (EWA) 0.116 0.018 <0.001 -0.006 0.014 0.655 Model 3 Adding direct effects of OXTR rs1042778 (G to T) -0.010 0.027 0.712 -0.006 0.020 0.780 rs2254298 (G/G genotype=1; A-allele=0) -0.058 0.050 0.246 0.055 0.038 0.129 rs53576 (G to A) -0.047 0.026 0.072 0.013 0.019 0.503 Model 4 Adding GxE interactions EWA*rs1042778 -0.027 0.027 0.306 0.055 0.020 0.006 EWA*rs2254298 -0.042 0.048 0.381 0.043 0.036 0.223 EWA*rs53576 0.033 0.024 0.165 -0.021 0.018 0.240 40 40 Table 3. Perceived support from friends and peripheral sources Model Intercept (SE) p Slope (SE) p Model 1 Developmental model Latent means 4.616 0.064 <0.001 -0.134 0.049 0.006 Gender (male=1; female=0) -0.639 0.035 <0.001 0.090 0.028 0.001 Age 0.098 0.017 <0.001 -0.124 0.014 <0.001 Socioeconomic status 0.096 0.020 <0.001 -0.022 0.016 0.160 Period effect for 1989 (T1=1; T2-T5=0) -0.023 0.063 0.720 0.120 0.048 0.013 Model 2 Adding direct effect of family environment Emotional warmth and acceptance (EWA) 0.053 0.017 0.002 -0.009 0.013 0.487 Model 3 Adding direct effects of OXTR rs1042778 (G to T) 0.017 0.025 0.503 -0.015 0.020 0.438 rs2254298 (G/G genotype=1; A-allele=0) -0.011 0.047 0.807 0.002 0.037 0.961 rs53576 (G to A) -0.028 0.024 0.254 0.011 0.019 0.558 Model 4 Adding GxE interactions EWA*rs1042778 0.006 0.025 0.818 0.026 0.020 0.183 EWA*rs2254298 -0.030 0.043 0.484 0.015 0.034 0.655 EWA*rs53576 0.022 0.022 0.334 -0.010 0.018 0.561 41 41 Figure Captions Figure 1. Population diagram Figure 2. The effect of high and low maternal emotional warmth and acceptance during childhood on the latent means of perceived family support across five YFS waves; split by OXTR rs1042778 G/G genotype (a), G/T genotype (b), and T/T (c) genotype carriers. Each genotype and each interaction were analyzed separately.