Climate-driven immune gene expression profiles in Peromyscus leucopus: implications for bacterial infection dynamics
Description
Pinnae (ear tissue) samples from Peromyscus leucopus (N = 148) were collected by NEON staff in 2022 from eight sites (see Animals and Study Site section in the main text). Samples were used to obtain RNA (see RNA extraction in the main text) and measure immune gene expression (see Droplet Digital Polymerase Chain Reaction (ddPCR) section in the main text).
Files
Steps to reproduce
Statistical analyses All data was analyzed using IBM SPSS 29. Because our primary interest was to reveal whether and how climate conditions at NEON sites related to the competence of individual mice for Borrelia burgdorferi, not how climate affected single genes, we conducted a Varimax-normalized principal component analysis (PCA) on log10(N+1) transformed gene expression data to discern whether and how genes were expressed as a collective within individual mice. Following the Kaiser criterion, we considered only principal components with eigenvalues > 1 as appropriate for further investigation. This PCA revealed two PCs with eigenvalues > 1, PC1 (column W, PC1_Resistance) and PC2 (column X, PC2_Tolerance) based on their genetic composition (see details in Table S1). To discern how climate and other forces affected variation in gene expression (PC1 and PC2), we followed a model-fitting procedure. Linear mixed models were used to analyze resistance and tolerance PC scores separately, starting with the same omnibus model structure. We compared AICc scores of a series of progressively simpler models, starting from the omnibus model, including mouse sex (column D, Sex: male or female), mouse body mass (column O, Mean_BodyMass), and climate at a site (column Y, PC3_Climate PC score), and all possible interactions as fixed effects. Tick presence (column P, TicksPresence: presence or absence on an individual mouse) was also included as a simple fixed effect, but no interactions for this term were included as we lacked tick data for many mice (N = 54 of 148 total mice lacking tick data). We also included NEON sites as a random effect in all models to account for climate being a site-level variable (column B, Site_Alphabetic). Juvenile individuals (N = 3, rows 42, 43, and 75) were not used in the analysis, and empty cells represent the absence of data. Climate data The Excel file: climate_PCA.xlsx includes the site-level climate data that went into the climate PCA, the PCA loadings, and the metadata explaining what each climate variable refers to. The file named Climate PCA_code.R contains the R code used to download the 2022 climate data from NEON, run the PCA, and merge the climate PC scores column with the gene expression data. We used an oblimin rotation for this PCA, but the results were robust to other rotations (i.e., varimax). Figure Fig.1. The composite figure was generated in PowerPoint. The Map was generated using R, and the mouse image was modified from a photograph by wildlife photographer and writer Dr. William J. Weber (available through iStock). Fig. 2. The graph was generated using GraphPad Prism (version 10.2.3 – attached PC climate by NEON site). Fig. 3. The graph was generated using R (Resistance~Climate figures.R).
Institutions
- University of South FloridaFlorida, Tampa
- University of Wisconsin MadisonWI, Madison
- University of West AlabamaAL, Livingston