A single-cell atlas of circulating immune cells over the first two months of age in extremely premature infants

Extremely premature infants (EPI) born prior to 30 weeks gestation are highly susceptible to infection. The trajectory of peripheral immunity in EPIs is poorly understood. Longitudinal analysis of immune cells from 250mL of whole blood at 1 week (n=7), 1 month (n=7), and 2 months (n=5) from 10 EPI was compared to healthy adults (n=6) and to neonatal cord blood (n=13). Single-cell suspensions from individual samples were split to perform single-cell(sc) RNA-, T- and B-cell receptor sequencing (seq), and phosphoprotein mass cytometry. Our scRNAseq data was integrated with existing data from full-term infants at 2 months of age. The trajectory of circulating T-, B-, myeloid, and natural killer cells in EPI infants over the first two months of life is distinct from full-term infants. Peripheral T cell development rapidly progressed over the first month of EPI life with an increase in the proportion of naïve CD4, regulatory, and cycling T cells, accompanied by increased STAT5 signaling compared to all other samples. Simultaneously, the transcription of IL2, which is essential for T cell growth and proliferation, increased in the lymphocytes, while IL7 and IL15 were highest in B cells and myeloid cells in EPI samples at 2 months of age. EPI memory CD4 T cells were dominated by ZBTB16 expression with a Th1 predominance compared to Th2 skewing of central memory-like T cells in full-term infants. Similarly, B cells from 2-month-old EPIs exhibited increased signatures of activation, BCR signaling, and differentiation compared to all other samples. Both B and T cells from 2-month-old EPIs had increased IFN signatures compared to full-term infants. Together, we demonstrate the feasibility of a robust multi-omic longitudinal analysis in EPIs from minute amounts of blood, developing a resource for studying early-life immune development.

Methods
In this study, we performed multi-omic analysis including mass cytometry by time of flight (CyTOF), single-cell RNA sequencing (scRNAseq), T cell and B cell receptor sequencing (TCR and BCR-seq) on circulating immune cells that were isolated from 100 to 250 microliters of blood obtained from extremely premature infants (EPI, n=10) and compared to cord blood from premature (n=5)  and full-term infants (n=5) as well as adult blood. The samples from EPI were obtained at 1 week, 1 month, and 2 months of life. 

Here we have included CyTOF data (.Rdata file format)  that has been gated on live, DNA+, CD45+ circulating leukocytes. Data was demultiplexed using Premessa (https://github.com/ParkerICI/premessa) and automated RPhenograph clustering with k=30 in Cytofkit2 (https://github.com/JinmiaoChenLab/cytofkit2) and included here is the Rdata file that includes analysis of ~15,000 cells that have been clustered based upon surface marker expression to identify 28 distinct populations of T cells, B cells, NK cells, innate lymphoid cells, Myeloid cells, and neutrophils from Adult blood n=3, full-term cord blood n=4,  preterm cord blood n=5, serial samples from 9 EPI infants collected at 1 week n=7, 1 month n=7, 2 months n=4. Using cytofkit2 this data can be visualized in a tsne format showing clusters as well as the heatmap expression of surface and phosphoprotein markers in each sample and among the groups.

We also included the code used to analyze scRNAseq, TCR, and BCR files that were analyzed in R and Python using various packages that are outlined in the description of each file. The raw data and processed data matrices are available for download at Gene Expression Omnibus (accession GSE271413). We performed quality control, batch correction, and clustering. Our scRNAseq data set was integrated with samples from full-term infants at 2 months (GSE204716). We performed a differentially expressed gene analysis based on infant exposure to in utero inflammation, and cell-to-cell communication based on age/timepoint of sample collection and determined TCR/BCR clonality and diversity among samples.