Data from: Spatiotemporal-social association predicts immunological similarity in rewilded mice
Data files
Nov 15, 2023 version files 6.51 MB
Abstract
Environmental influences on immune phenotypes are well-documented, but our understanding of which elements of the environment affect immune systems, and how, remains vague. Behaviors, including socializing with others, are central to an individual’s interaction with its environment. We therefore tracked behavior of rewilded laboratory mice of three inbred strains in outdoor enclosures and examined contributions of behavior, including associations measured from spatiotemporal cooccurrences, to immune phenotypes. We found extensive variation in individual and social behavior among and within mouse strains upon rewilding. And we found that the more associated two individuals were, the more similar their immune phenotypes were. Spatiotemporal association was particularly predictive of similar memory T and B cell profiles and was more influential than sibling relationships or shared infection status. These results highlight the importance of shared spatiotemporal activity patterns and/or social networks for immune phenotype and suggest potential immunological correlates of social life.
README: Data from: Spatiotemporal-social association predicts immunological similarity in rewilded mice
https://doi.org/10.5061/dryad.rjdfn2zjc
This dataset contains immune data, behavior data, and related R code for processing, analyzing, and plotting the data and model results. The immune data include flow cytometry data from lymphocytes, serum concentrations of cytokines, concentrations of cytokines produced by MLN cells following antigenic challenge, and complete blood count profiles. The data also include gut microbiome data, as measured via 16S sequencing.
This dataset goes with a preprint, Downie et al. (2023) (linked here), and the associated manuscript, provisionally accepted by Science Advances.
Description of the data and file structure
The data are presented in comma-separated-value (CSV) files, except the microbiome data; these are in a tab-delimited file. The full description of the methods for producing these data and flow cytometry gating schemes are available in Downie et al. (2023), as well as Oyesola et al. (2023; linked here). Below we provide some description for each file.
All NA entries in data files are places where the quantity either was not measured or was not known. NA values for parentage in metadata and CBC files are unknown in a few cases because mouse ID prior to release was unstable (lost ear tags during transfers between facilities preventing confident assignment of identity). If you have particular questions about these, please contact us.
mice.metadata.2021.block1.GxE.csv: This file contains the metadata for each mouse used in the rewilding experiment. This includes mouse ID ("Left.ear.tag") as well as secondary forms of ID used during the experiment ("Right.ear.tag"), genotype, the enclosure holding it during rewilding ("Wedge"), the cage it was raised in, parentage (when known – on occasion this is not known due to issues with maintaining stable mouse IDs prior to release), whether the mouse was challenged with Trichuris muris ("Infected"), and information about its RFID and capture status during the experiment. "Lost.RFID.1" and "Lost.RFID.2" indicate whether the mouse lost its RFID tag during the first and second portions of the experiment, respectively (most mice did not lose a tag during either). Values of "Y" for the "Infected" column indicate that the mouse received a challenge with T. muris at ~two weeks after release; values "N" indicate no challenge was given. "Missing.1" and "Missing.2" indicates a mouse that was not successfully captured during the first and second portions of the experiment, respectively. "Uncaptured" indicates that the mouse was not recaptured at the end of the experiment and therefore has no associated immune data from the endpoint. Values for "Uncaptured" and "Missing.2" are identical, since the second trapping session was the last trapping session, but they are used for different purposes in our code, so we have kept them separate.
mice.metadata.2021.block2.GxE.csv: Same as above, but for the second block (repetition) of the experiment.
2021.block1.clean.GxE.csv: This file contains the RFID check-in information for each mouse across the duration of the rewilding period (five weeks). The columns are wedge (reader number), timestamp (to the second), RFID (the RFID recorded by the reader), and mouse.
2021.block2.clean.GxE.csv: Same as above, but for the second block (repetition) of the experiment.
2021.plasma.cyt.csv: This file contains the concentrations of different cytokines in the plasma at the time of sacrifice, as well as some relevant pieces of mouse metadata taken from the metadata files and defined as in the descriptions there. The cytokines are IFNg, IL5, TNFa, IL6, IL17A, and IL22.
2021.MLN.cytokine.prod.csv: This file contains the concentrations of different cytokines after stimulation of mesenteric lymph node cells with various antigens in vitro to provoke cytokine responses, as well as some relevant pieces of mouse metadata taken from the metadata files and defined as in the descriptions there. The columns are organized by antigen: each cytokine's production in response to PBS (the control), followed by CD3/CD28, LPS, Candida albicans, Clostridium perfringens, Bacteroides vulgatus, and Trichuris muris. See the manuscript methods for the precise details of the stimulation assay and the full list of the cytokines assayed.
2021.CBC.data.compliant.csv: This file contains the readouts of complete blood count analyses for each mouse at different time points: prior to release into enclosures, two weeks after release, and five week after release (at trapout), as well as some relevant pieces of mouse metadata taken from the metadata files and defined as in the descriptions there. Not all timepoints are available for all mice due to mice not being captured at particular trapping sessions and trouble maintaining stable mouse IDs prior to release. Metadata from the metadata files is included, as well as age of mouse (in weeks) at time of sampling. Both total counts of different WBC types and percentages are included, along with data on RBC properties (unanalyzed in the manuscript). Details on method and machine for CBC analysis are in the manuscript methods.
2021.blood.flow.analysis.csv: This file contains flow cytometry results for analyses of lymphocyte (CD4 and CD8 T cell) populations in the peripheral blood, along with memory phenotypes (based on CD44 and CD62L expression), as well as some relevant pieces of mouse metadata taken from the metadata files and defined as in the descriptions there. The data include some metadata drawn from the above sheets as well as both counts of cells and percentages (which is which is clearly marked in column names). Percentages for CD44 and CD62L expression phenotypes are given as percentages of the total cell subtype (CD4 T, CD8 T, or B220+ B cells). For details on flow cytometry procedures, gating, and cell type classification, please see Downie et al. (2023) and Oyesola et al. (2023). Naïve cells are CD62L+, CD44-; effector memory cells are CD62L+, CD44+; central memory cells are CD62L-, CD44+; double negative cells are CD62L-, CD44-.
2021.MLN.flow.analysis.csv: This file is much the same as the above, but for cells in the mesenteric lymph nodes instead of the peripheral blood, as well as some relevant pieces of mouse metadata taken from the metadata files and defined as in the descriptions there. There are also a few additional cell types, including B220+ B cells (with CD44 and CD62L expression), NK cells, gd T cells, CD45+ cells, and CD45+ ILCs. Note that in this file all values are given strictly as percentages; percentages for CD44 and CD62L expression phenotypes are given as percentages of the total cell subtype (CD4 T, CD8 T, or B220+ B cells).
The R code provides instructions for integrating the data files together and for calculating pairwise immune similarity, which is the main form of analysis in the publication.
Sharing/Access information
The data can also be accessed at: https://github.com/aedownie/rewilding2021
These data, along with other data not yet shared, are also analyzed in a separate preprint, Oyesola et al. (2023) (linked here). The manuscript associated with that preprint is currently under review. We ask that individuals interested in analyzing these data for a publication please contact us directly if said manuscript has not been published.
Code/Software
The code is in six R scripts; the titles provide rough descriptions of the material in each script. We ran it on R v4.1.2 with RStudio v2023.03.0+386. We also employed several further packages at different stages:
- dplyr v1.0.10
- tidyr v1.2.1
- ggplot2 v3.4.0
- ggridges v0.5.3
- brms v2.16.3
- MetBrewer v0.2.0
- vegan v2.6-4
The scripts are organized in a sequence laid out below; in most cases the use of one script will depend on the output of the prior one, although not always.
CI.data.preprocessing.R: This script takes the RFID check-in data and mouse metadata and provides additional information for each check-in, such as the location of the RFID reader, the Julian night of the experiment, etc.
individual.behavior.calculation.R: This script uses the RFID check-in data (after updating via the previous script) and mouse metadata to identify various aspects of individual behavior from the check-ins, as shown mostly in Fig. 1 of the manuscript. This includes number of check-ins per night, roaming entropy, and minimum distance traveled per night.
social.association.calculation.R: This script uses the RFID check-in data (after updating from the first script) and mouse metadata to determine the pairwise strength of spatiotemporal association between mice. We use a variety of different ways of defining an association, and as a result the script is very long (we never developed an efficient way to do all the calculation for individual definitions from only a few commands, but in theory it's probably not too difficult to streamline). This script produces full.2021.assoc.info, a very important data frame for subsequent analyses.
immune.similarity.calculation.R: This script uses the various immune data spreadsheets, along with the mouse metadata to calculate for each pair of mice the similarity of their immune phenotypes, for various aspects of immune phenotype. These pairwise metrics are needed for calculating the relationship between pairwise spatiotemporal association and immune similarity. The calculations are slightly different for different aspects of immune defense. This script also contains code for calculating microbiome similarity.
statistical.analyses.R: This script uses the outputs from the three preceding scripts to explore statistical relationships between metrics of spatiotemporal association and immune similarity, as well as predictors of variation in spatiotemporal association and individual behavior. The various models are a mix of standard linear regressions for some quantities and beta regressions for other quantities, particularly the pairwise similarities.
results.plotting.R: This script contains the code necessary to reproduce the figures in the manuscript. It relies on the model outputs in the preceding script, as well as the output from the other scripts when plotting, for example, distributions of check-ins across the experiment or variation in behavior metrics between individuals.
Methods
This dataset includes the data and analysis code for Downie et al. (2023 preprint, 202? publication). It is a mixture of immune cell phenotypes, serum cytokines, MLN cytokine production, microbiome (sequenced via 16S), and behavioral data from RFID check-ins. Please see the preprint (or eventual manuscript) for details about the methodology and degree of processing. The behavioral data is largely unprocessed, while the flow cytometry data and ELISA data are post-processing.