Transcriptome-wide analysis reveals different categories of response to a standardised immune challenge in a wild rodent

Abstract

Individuals vary in their immune response and, as a result, some are more susceptible to infectious disease than others. Little is known about the nature of this individual variation in natural populations, or which components of immune pathways are most responsible, but defining this underlying landscape of variation is an essential first step to understanding the drivers of this variation and, ultimately, predicting the outcome of infection. We describe transcriptome-wide variation in response to a standardised immune challenge in wild field voles. We find that genes (hereafter 'markers') can be categorised into a limited number of types. For the majority of markers, the response of an individual is dependent on its baseline expression level, with significant enrichment in this category for conventional immune pathways. Another, moderately sized, category contains markers for which the responses of different individuals are also variable but independent of their baseline expression levels. This category lacks any enrichment for conventional immune pathways. We further identify markers which display particularly high individual variability in response, and could be used as markers of immune response in larger studies. Our work shows how a standardised challenge performed on a natural population can reveal the patterns of natural variation in immune response.

Figure 1

Different categories of response to a standardised immune challenge. These are based on two overlapping sets of criteria: baseline-dependence of an individual’s response (blue background) and individual variability in response (yellow background). Arrows represent individual responses. No response (for reference): markers for which individuals (on average) show no response to stimulation (intercept not significantly different from zero; slope not significantly different from one). Uncorrelated response: markers for which responses of different individuals are variable and independent of their baseline (slope not significantly different from zero). Constant response: markers for which the responses of different individuals are (on average) constant and independent of their baseline (intercept significantly greater than zero; slope not significantly different from one). Baseline-dependent response: markers for which responses of different individuals vary as a function of their baseline, either as a linear function of their baseline (slope significantly different from one; slope greater than one is depicted but could equally be less than one), or as a quadratic function of their baseline (a saturating function is depicted but could equally be exponential). Convergent response: markers for which the coefficient of variation (CV) for baseline abundances is significantly greater than the CV for stimulated abundances across individuals (CV_base> CV_stim). Divergent response: markers for which the CV for stimulated abundances is significantly greater than the CV for baseline abundances across individuals (CV_stim> CV_base). Both convergent and divergent markers depicted as, but not limited to, markers for which response is uncorrelated.

Figure 2

Top 10 markers and enriched ontology terms of interest in each response category. Each box represents a category of response (as in Fig. 1). For each category, top 10 annotated markers for which we had the most confidence in their categorisation (markers were ranked on R² and p-values; see Supplementary Tables 2–6 for full set of parameters used to categorise and/or rank these markers) are listed, one or two of these are represented in plots showing stimulated versus baseline abundances (in counts per million) across individuals (solid line indicates significant relationship between baseline and stimulated abundance; dashed line indicates a slope equal to one and the same intercept for reference; note the differing axes of these plots). In the case of the convergent response category, which only included a total of six annotated markers, all markers are listed. Ontology terms of interest, from a functional enrichment analysis performed on all markers within a category (where possible), are also included (immune-related terms in black; non-immune-related terms in grey).

Figure 3

Map of the T-cell receptor signalling KEGG pathway for Mus musculus, with the colour of nodes representing the level of individual variability in response to stimulation with anti-CD3 and anti-CD28 antibodies in Microtus agrestis. Namely the p-value from an asymptotic test for the equality of variance in gene expression levels for baseline and stimulated samples (range = <0.001–0.97). Dark blue indicates high individual variability in response, whereas light blue or white indicates low individual variability in response. Grey nodes represent genes for which no information is available, either because they are unannotated in the M. agrestis genome, or because they are weakly expressed in the spleen.