Fig 2: High GATA-1 protein abundance delineates epigenetic program for erythroid commitment in RBC developmental trajectory(A) BM aspirate is ficolled and enriched for CD34+ cells before gating for CD123– MEP population (CD34+/CD38+/CD10–/CD45RA–/CD123–) and selecting high- (25%–40%) and mid- (~lower 30%) GATA-1-expressing cells within each compartment.(B) GATA-1-high cells from CD34+/CD38+ and CD123– MEP compartments InTAC-seq data were simulated as scATAC counts and projected on scATAC UMAP space.(C) Putative erythropoiesis trajectory constructed from HSCs to late erythoid populations and overlaid on scATAC UMAP.(D) Heatmap of top variable TFs by ChromVAR deviation scores across constructed erythroid trajectory with the projected position of GATA-1-high InTAC-seq samples indicated in red as the point of GATA-1-high overlap, in blue as the before point, and in green as the after point. Top: line plot of InTAC-seq-denoted GATA-1-high-simulated scATAC cells as binned across pseudotime.(E) Top 20 genes significantly enriched (of fold change 2 and above) in integrated scRNA-seq data between the 3 bins, before, at, and after GATA-1-high overlap points in trajectory.(F) Summary schematic of continuous differentiation to erythrocytes in BM with downregulation of lymphoid/myeloid TF activity and gene expression programs and upregulation of erythroid TF activity and gene expression programs. High GATA-1 protein abundance overlaps epigenetic program shift to erythroid lineage commitment in human BM.

Fig 3: CD4+ and CD8+ T lymphocyte depletion blocks COPD-like pathology in pIgR-/- mice and is associated with reduced apoptosis of lung epithelial and endothelial cells.a, b Percentage of CD4+ or CD8+ lymphocytes among CD45+ cells in spleens of mice treated with weekly intraperitoneal injection of anti-CD4 or anti-CD8 antibodies from 4 to 8 months of age. c Measurement of mean linear intercept (MLI), a morphometric measurement of emphysema, or d measurement of VVairway, a morphometric measurement of small airway wall thickness, in pIgR-/- mice treated with anti-CD4 or anti-CD8 antibodies or rat IgG2b isotype control antibodies from 4 to 8 months of age. e TUNEL+ cells in the lung parenchyma of a pIgR-/- mouse (white arrows denote TUNEL+ cells). Scale bar = 100 µm. f Quantification of TUNEL+ cells in pIgR-/- and WT mice at the indicated ages. g Representative image of colocalization between TUNEL+ cells and pan-cytokeratin (marking epithelial cells) or CD34 (marking endothelial cells) in an 18-month-old pIgR-/- mouse. h Quantification of TUNEL/pan-cytokeratin and TUNEL/CD34 dual-expressing cells in pIgR-/- and WT mice at the indicated ages. i Quantification of TUNEL+ cells/field in 8-month-old pIgR-/- mice treated with anti-CD8 or anti-CD4 antibodies or isotype control antibodies between 4 and 8 months of age. a, b *p < 0.05 compared to untreated mice (t-test); n = 3–4 mice/group. c, d *p < 0.05 compared to isotype-control treated mice, **p < 0.05 compared to isotype-control treated mice (1-way ANOVA with Bonferroni post hoc test); n = 4–8 mice/group. f–h *p < 0.001 compared to age-matched WT mice (2-way ANOVA with Bonferroni post hoc test); n = 5–6 mice/group. i *p < 0.01 compared to isotype control, *p < 0.0001 compared to isotype control; (1-way ANOVA with Bonferroni post hoc test); n = 7–8 mice/group. Box-and-whisker plots represent median, interquartile range, and range.

Fig 4: High-dimensional, single-cell proteomic analysis of BM progenitors identifies TF and surface-marker trends in erythroid commitment(A) BM aspirate is ficolled and enriched for CD34+ cells before staining with antibodies to surface marker panel and key TF to capture single-cell protein abundance using mass cytometry. High-dimensional data with over 1 million cells were used to delineate heterogeneity in BM progenitors and find surface-marker surrogates for GATA-1 TF abundance for further functional validation.(B) Force-directed layout (ForceAtlas2) of density downsampled (to 250,000 cells) CD45+-gated BM progenitors colored by manually gated populations.(C) Normalized marker expression of key surface and TF proteins (in orange) across force-directed layout. Orange arrow denotes GATA-1-high region in single-cell map.(D) Leiden clusters of BM progenitors (resolution = 1) visualized on force-directed layout.(E) Barplot of frequency of manually gated BM progenitor populations across 11 Leiden clusters.(F) Normalized diffusion pseudotime calculation visualized on force-directed layout with trajectory from HSCs to erythroid-primed progenitors across Leiden clusters 3, 1, and 11 (in red).(G) Row-normalized heatmap of median marker abundance at 100 bins across diffusion pseudotime-aligned trajectory. Orange font and arrows indicate TF protein trends, and bold font with black arrows indicate key surface marker trends in trajectory.(H) Column-normalized heatmap of mutual information scores calculated on cells in Leiden clusters 2, 7, and 8 and normalized across key TF and surface-marker pairs.

Fig 5: GATA-1-high BM progenitors show strong priming for erythroid lineage(A) BM aspirate is ficolled and enriched for CD34+ cells before gating for CD34+/CD38+ cells and selecting GATA-1-high population (top ~8%) and remaining GATA-1-expressing cells (denoted as GATA-1 mid/low) (bottom ~87%).(B) InTAC-seq genome coverage plots at GATA1 and SPI1 loci for GATA-1-high and -mid/low BM progenitors.(C) Heatmap of chromVAR deviation scores across GATA-1 high and mid/low BM progenitors for top 50 most variable motifs.(D) MA plot of log2 fold change in accessibility between GATA-1 high and mid/low BM progenitors versus log2 mean number of reads in consensus peaks. Peaks with significant changes in accessibility are highlighted in red or blue.(E) Most significantly enriched TF motifs in differentially accessible peaks between GATA-1-high and -mid/low BM progenitors calculated using Fisher’s test.(F) (Top) Average accessibility of GATA-1 motif sites across all consensus ATAC-seq peaks binned by GATA-1 motif score. Accessibility is defined here as the area under the curve of a plot of bias-corrected, normalized Tn5 insertions centered at GATA-1 motif sites, integrated from -50 to +50 bp and excluding the TF footprint from -10 to +10 bp. (Bottom) Difference in GATA-1 motif accessibility between GATA-1 high and mid/low samples normalized to the accessibility in the GATA-1 mid/low population for each motif score bin.(G) UMAP of previously published and annotated BM scATAC dataset with Seurat clusters manually annotated as key BM populations.(H) Normalized GATA1 gene expression across BM progenitors in UMAP space (expression derived from scRNA-seq data integrated with scATAC-seq data).(I) Bulk BM progenitor InTAC-seq data simulated as scATAC counts and projected onto scATAC UMAP space.