Distinct gene regulatory signatures of dominance rank and social bond strength in wild baboons

Abstract

The social environment is a major determinant of morbidity, mortality and Darwinian fitness in social animals. Recent studies have begun to uncover the molecular processes associated with these relationships, but the degree to which they vary across different dimensions of the social environment remains unclear. Here, we draw on a long-term field study of wild baboons to compare the signatures of affiliative and competitive aspects of the social environment in white blood cell gene regulation, under both immune-stimulated and non-stimulated conditions. We find that the effects of dominance rank on gene expression are directionally opposite in males versus females, such that high-ranking males resemble low-ranking females, and vice versa. Among females, rank and social bond strength are both reflected in the activity of cellular metabolism and proliferation genes. However, while we observe pronounced rank-related differences in baseline immune gene activity, only bond strength predicts the fold-change response to immune (lipopolysaccharide) stimulation. Together, our results indicate that the directionality and magnitude of social effects on gene regulation depend on the aspect of the social environment under study. This heterogeneity may help explain why social environmental effects on health and longevity can also vary between measures. This article is part of the theme issue 'The centennial of the pecking order: current state and future prospects for the study of dominance hierarchies'.

Keywords: Papio cynocephalus; gene expression; social integration; social status.

Figure 1.

Strong, sex-specific signatures of dominance rank in white blood cell gene expression. (a) Study design: dominance rank (males and females) and social bond strength (females only) were evaluated for their relationship with white blood cell gene expression, generated from samples cultured for 10 h in the absence (baseline) or presence of lipopolysaccharide (LPS). (b,c) A gene expression-based elastic net model accurately predicts dominance rank for male (b; Pearson's R = 0.449, p = 8.46 × 10⁻⁴) and female (c; Pearson's R = 0.656, p = 1.31 × 10⁻⁶) baboons. The dashed circle denotes the outlier female AMB_2. (d) The effect estimates for the rank–gene expression association are negatively correlated in males versus females (Pearson's R = −0.536, p < 10⁻⁵⁰). Purple, green and yellow points correspond to genes passing a 10% false discovery rate for the male rank effect, female rank effect, or both, respectively. (e) Gene sets enriched for higher expression in high-ranking males are enriched for lower expression in high-ranking females, and vice versa. Enrichment in males is shown in purple; enrichment in females is shown in green. For all gene sets, enrichment score Bonferroni-corrected p-values are <0.005. Photo credits in (a): Elizabeth Archie (females) and Courtney Fitzpatrick (males). Stock images of LPS and blood draw tubes courtesy of BioRender.com. (Online version in colour.)

Figure 2.

Social status and social bond strength predict distinct patterns of gene expression. (a) Distribution of DSI effects (left) and rank effects (right) on gene expression at baseline for genes within the MSigDB Hallmark gene sets labelled at left. Genes within immune-related pathways (red/orange) are polarized towards higher expression in low-ranking females (positive effect sizes, because low rank is represented by high ordinal rank values). By contrast, genes in the same pathways show no pattern for association with DSI (small effect sizes centred on zero). Cellular metabolism and cell cycle-related gene sets (blue) tended to be more highly expressed with high rank (negative effect sizes) and low DSI (negative effect sizes). Translucent density plots indicate no significant bias in the direction of effects (binomial test p > 0.05). (b) Effect size bias for genes in the Hallmark inflammatory response and MYC (v1) target gene sets, for DSI and rank, respectively (binomial test for the inflammatory response: DSI p = 0.868; rank p = 2.35 × 10⁻¹²; MYC (v1): DSI p = 1.66 × 10⁻¹⁴; rank p = 1.53 × 10⁻²⁰). (c) Gene set enrichment analysis results for female DSI (green) and rank (pink) effects on the foldchange response to LPS stimulation. Inset: QQ-plot of the −log₁₀(p-value) for DSI and rank effects on the LPS response, relative to a uniform null distribution. We observe strong evidence for associations between DSI and the LPS response, but not for rank. (d) Example of FH, a key enzyme in the Krebs cycle that responds more strongly to LPS in high DSI females than low DSI females. (Online version in colour.)

Figure 3.

The gene expression signature of multidimensional social advantage. (a) Median gene expression for genes in the inflammatory response gene set illustrates that high-ranking animals exhibit lower inflammation-related gene expression regardless of social bond strength (main effect of rank = −0.65, p = 0.028). There is no main effect of DSI (p = 0.213), but the difference between high- and low-ranking females is greater when high-ranking females also have strong social bonds. (b) Median, rescaled gene expression levels per individual in the Hallmark IL6 signalling via JAK/STAT3 and the inflammatory response gene sets. Each row represents a different female, with rows stratified by median dominance rank and median DSI. All gene expression values in figure 3 are based on baseline samples. (Online version in colour.)