Type II innate lymphoid cell plasticity contributes to impaired reconstitution after allogeneic hematopoietic stem cell transplantation

Abstract

Type II innate lymphoid cells (ILC2s) maintain homeostasis and barrier integrity in mucosal tissues. In both mice and humans, ILC2s poorly reconstitute after allogeneic hematopoietic stem cell transplantation (allo-HSCT). Determining the mechanisms involved in their impaired reconstitution could improve transplant outcomes. By integrating single-cell chromatin and transcriptomic analyses of transplanted ILC2s, we identify a previously unreported population of converted ILC1-like cells in the mouse small intestine post-transplant. Exposure of ILC2s to proinflammatory cytokines resulted in a mixed ILC1-ILC2 phenotype but was able to convert only a small population of ILC2s to ILC1s, which were found post-transplant. Whereas ILC2s protected against acute graft-versus-host disease (aGVHD) mediated mortality, infusion of proinflammatory cytokine-exposed ILC2s accelerated aGvHD. Interestingly, murine ILC2 reconstitution post-HSCT is decreased in the presence of alloreactive T cells. Finally, peripheral blood cells from human patients with aGvHD have an altered ILC2-associated chromatin landscape compared to transplanted controls. These data demonstrate that following transplantation ILC2s convert to a pro-pathogenic population with an ILC1-like chromatin state and provide insights into the contribution of ILC plasticity to the impaired reconstitution of ILC2 cells, which is one of several potential mechanisms for the poor reconstitution of these important cells after allo-HSCT.

Fig. 1. Ex vivo expanded murine ILC2s exhibit the phenotypic and functional properties of ILC2s.

a Schematic of ex vivo ILC2 expansion. Mice received 4 days of IP IL-25 before the isolation of lineage-negative cells from the peritoneum and mesenteric lymph nodes. Cells were cultured for 6 days in complete media with IL-7 and IL-33. b Scatter plot of H3K4me3 signal from − 300 bp to +500 bp around the TSSs of all transcribed genes and RNA abundance of the corresponding gene (Spearman correlation coefficient, r = 0.81) TPM values represent the average signal of three replicates. H3K4me3 signal represents the average of two replicates. c Distributions of H3K4me3 signal in cells isolated from murine small intestine at TSSs that demonstrated differential H3K4me3 specific to ILC1 or ILC2 cells (n = 2, both conditions, one-sided Mann Whitney U Test, p < = 5.37 × 10⁻⁸). Box plot lines represent min/max values. d Representative tracks of H3K4me3 signal at ILC2 lineage defining genes Gata3, Il13, Il4, Id2, Maf, and Areg. Black bars indicate regions of significant H3K4me3 enrichment (two-sided Wald Test (DESeq2), p adj < 0.05). e, f Following 6 days of expansion, ILC2s were stimulated and cells were characterized by multiparameter flow cytometry. Live, lineage-negative (CD3-, B220-, CD11b-, TER119-, Ly6G-) singlets were assessed for ST2, IFN-y, Tbet, GATA3, and IL-13 expression. e, f Show combined biological replicates where n = minimum of three animals and represent two-three independent experiments. Dotted lines indicate unstimulated controls. Box plots in (c) are centered at the median. The bounds of each box represent the interquartile range IQR, and the whiskers represent the min and max values. Outliers are not shown in boxplots.

Fig. 2. After allogeneic hematopoietic stem cell transplantation, a population of converted ILC1-like exILC2s emerges in the SI LP.

a Schematic of murine bone marrow transplantation model. Lethally irradiated B6D2 mice received allogeneic (C57BL/6) T cell depleted bone marrow plus total splenic T cells with activated GFP-expressing ILC2s. After 20 days, eGFP⁺ cells were isolated from the lamina propria (LP). b UMAP depicts embeddings of 5554 nuclei. Pre-transplant GFP⁺ ILC2s (blue) and post-transplant LP-isolated GFP⁺ cells (green) are shown. c Pre-transplant ILC2 and post-transplant cells are separately depicted. d Heatmap depicting differentially expressed genes (two-sided Wilcoxon rank sum test, p adj < 0.05, avglog2FC > 0) by comparing one pre-transplant cluster to the set of post-transplant clusters (Gene sets pre 1 – 3). Tick marks indicate genes shared across three pre-transplant gene sets (left) and post-transplant gene sets (right). e Gene ontology analysis was performed for each of the post-transplant associated gene sets (one-sided Fisher’s Exact test). The color indicates the proportion of differential genes found in each ontology. f The average expression of marker genes associated with each type of ILC (ILC1, ILC2, ILC3, and NK). g Distribution of per cell average marker expression is shown. Pre signifies pre-transplant and post- indicates post-transplant (one-sided Mann Whitney U Test, ***p < 2.2 × 10⁻¹⁶). h Regions of differential chromatin accessibility were identified between each pre- and post-transplant cluster from Fig. 2c. i–k Volcano plots showing the enrichment of DNA binding motifs at sites of increased chromatin accessibility in post-transplant clusters 1 (i), 2 (j), and 3 (k) relative to all the sites of open chromatin identified in the pre-transplant cell population. l, m Candidate regulators of pre- (l) and post- (m) transplant cells. n UMAP depicts a low-dimensional representation of the integrated RNA space, with each cell annotated with ILC cell type. o Scatter plot showing putative regulators of ILC1-like and ILC2s (Fig. 2n).

Fig. 3. In vitro skewing with IL-12 mediates ILC2 conversion resulting in the generation of a population of ILC1-like ILC2s.

a Schematic depicting ILC2s elicitation and isolation. b, c Representative flow cytometry plots of intracellular cytokine (b) and transcription factor expression (c) from cells expanded in vitro as described in 3 A. Error bars in (c) represent SEM. d ILC gene expression profiles based on mRNA abundance. e Differential sites of CA (two-sided Wald Test DESeq2), p adj < 0.05) between ILC2 and pcILC2s. The heatmap displays a z-score normalized ATAC signal. R1 and R2 indicate replicate numbers. f Representative tracks of normalized ATAC signal. Tick marks indicate differential sites of CA (two-sided Wald Test (DESeq2), p adj < 0.05). kb = kilobase. g UMAP of integrated RNA abundance signal from 7918 nuclei. ILC2s cultured with IL-7 and IL-33 (blue), ILC2s grown with proinflammatory cytokines, pcILC2 (orange). h Average chromatin accessibility at murine ILC2 marker genes.The Red dot shows mean value, ILC2-pcILC2: p < 2.2 × 10⁻¹⁶, ILC2-ILC1-like: p = 7.46 x 10⁻⁵. i Average per cell expression of genes associated with ILC1, ILC2, shared between ILC1/2, and ILC1/3. The analysis includes 4,758 ILC2 and 5641 pcILC2 nuclei. ***p < 2.2 × 10⁻¹⁶. j Candidate regulators of pcILC2 cells. k UMAP of integrated ILC2 and pcILC2s annotated with normalized gene expression. l UMAP of weighted nearest neighbor integration of single nucleus RNA and ATAC from in vitro ILC2s, pcILC2s, and post-transplant cells. The black arrow highlights ILC1-like cells. m, n Lethally irradiated B6D2 mice received T cell depleted bone marrow (BM, BM only), BM plus total splenic T cells (BM + T cells), BM plus T cells with activated ILC2s (BM, T cells + ILC2) or pcILC2s (BM, T cells + pcILC2). Mean ± SEM, n = 7–16. n Kaplan-Meier survival curve after allo-HSCT. Representative of 2 independent experiments with n = 6–10 mice. Log-rank (Mantel-Cox) test, ***P < 0.01. Boxplots (h, i), are centered at median. Box bounds represent the interquartile range (IQR), whiskers represent the min/max values, and outliers are not shown.

Fig. 4. Following allo-HSCT, blood cells from patients have decreased CA at sites of accessibility associated with ILC2 cells at the time of the diagnosis of aGVHD.

a Isolation of human ILC2s (hILC2s) from peripheral blood of healthy donors. Enrichment of ILC2s is followed by expansion of ILC2s and transdifferentiation for 14–21 days with IL-7 and IL-2 and IL-4, IL-33, IL-25 for hILC2s or IL-12, IL-18, IL-1β for proinflammatory cytokine human ILC2s (pc-hILC2s). b ATAC heatmap of z-score normalized ATAC signal of differential regions (two-sided Wald Test DESeq2), p adj < 0.05) of chromatin accessibility. R1 and R2 indicate replicate numbers. c Enriched motifs identified at differential sites of CA in hILC2 and pc-hILC2s (HOMER). d Average normalized ATAC signal was calculated at differential regions of chromatin accessibility that contain the GATA3 (left) or Tbet (right) binding motif. Dashed lines represent a second replicate. Rep 2 indicates replicate 2 (e) Blood was collected from patients 2–28 days prior to allogeneic HSCT. Post-transplant samples were collected at the time of aGVHD diagnosis (disease) or during a scheduled three-month follow-up (stable). Frozen buffy coats were thawed, and live cells were isolated by magnetic separation prior to snATAC-seq. f Differential analysis of aggregate ATAC signal over gene bodies +/− 2 kb for all genes. Volcano plot depicting log2FoldChange between pre- and post- samples for each gene score and − log10(p adj). Blue dots indicate p adj < 0.05 (two-sided Wald Test (DESeq2), p < 0.05). g Gene ontology (GO) analysis was performed for genes with differential ATAC signals for both the pre- and post-transplant cells. Bar plots indicate enrichment (− log10(p adj)) for the top 5 terms (One-sided Fisher’s Test). h Average normalized ATAC signal at hILC2-associated sites of CAs (Fig. 4b) was calculated per nucleus for each patient sample. Violin plots show the top 90% of the signal. Black dots indicate the mean of each distribution. i Average normalized ATAC signal at pc-hILC2-associated sites of CAs (Fig. 4b) per nucleus for each patient sample. Violin plots show the top 90% of the signal. Black dots indicate the mean of each distribution (***p < 0.001, two-sided t test).