Targeting the CD74 signaling axis suppresses inflammation and rescues defective hematopoiesis in RUNX1-familial platelet disorder

Abstract

Familial platelet disorder (FPD) is associated with germline RUNX1 mutations, establishing a preleukemic state and increasing the risk of developing leukemia. Currently, there are no intervention strategies to prevent leukemia progression. Single-cell RNA sequencing (n = 10) combined with functional analysis of samples from patients with RUNX1-FPD (n > 75) revealed that FPD hematopoietic stem and progenitor cells (HSPCs) displayed increased myeloid differentiation and suppressed megakaryopoiesis because of increased activation of prosurvival and inflammatory pathways. Bone marrow from patients with RUNX1-FPD contained an elevated cytokine milieu, exerting chronic inflammatory stress on HSPCs. RUNX1-FPD HSPCs were myeloid biased, had increased self-renewal, and were resistant to inflammation-mediated exhaustion. The bone marrow from patients with RUNX1-FPD showed high transcript and protein expression of CD74 at the preleukemic stage compared with that of healthy controls, which remained high upon patient transformation into leukemia. Further, CD74-mediated signaling was exaggerated in RUNX1-FPD HSPCs compared with healthy controls, leading to the activation of mTOR and JAK/STAT pathways with increased cytokine production. Genetic and pharmacological targeting of CD74 with ISO-1 and its downstream targets JAK1/2 and mTOR reversed RUNX1-FPD differentiation defects in vitro and in vivo and reduced inflammation. Our results highlight that inflammation is an early event in RUNX1-FPD pathogenesis, and CD74 signaling is one of the drivers of this inflammation. The repurposing of JAK1/2i (ruxolitinib) and mTORi (sirolimus) and promoting the advancement of CD74 inhibitors in clinical settings as an early intervention strategy would be beneficial to improve the phenotype of patients with RUNX1-FPD and prevent myeloid progression.

Fig. 1.. RUNX1-FPD HSPCs have skewed differentiation with higher myeloid bias and defective megakaryopoiesis.

(A) The experimental schematic. Freshly harvested BM cells from patients with FPD and HDs were received and processed to isolate HSPCs. In vitro assays for cell growth, differentiation, and colony formation ability were set up to characterize FPD versus HD phenotypically. (B) The in vitro culture of CD34⁺ cells for 14 days. The percentage of CD34⁺ and CD33⁺/13⁺ cells from FPD (n = 17 to 19) and HD (n = 10 to 12) were assessed using flow cytometry and represented as a percentage of live cells. CD33⁺/13⁺ was added later in the flow panel and thus represented fewer samples. Right, shows representative flow plots for the percentage of CD13 and CD33 on CD34⁺ cells. (C to F) In vitro differentiation assays for FPD (n = 21 or 22) and HD (n = 8 or 9) samples. CD34⁺ cells were seeded at 2000 cells per well in StemSpan II medium supplemented with specific cytokines for differentiation into myeloid [(C) and (D)], megakaryocyte (E), and erythroid (F) lineages. Flow cytometry analysis was done after 7 or 14 days, and data are represented as the percentage of live cells. Bottom shows representative flow plots for the percentage of indicated cell populations. (G) The colony-formation ability of FPD (n = 11) and HD (n = 14) samples. CD34⁺cells were seeded at 1000 cells per well density in duplicates using methocult H4434 (StemCell Technologies). Colonies were counted at day 14 (1°, left). Cells from the primary colony formation assays were collected and replated at the density of 100 thousand cells per well. Colonies were then counted on day 14 (2°, left). Bar graph showing CFU-GM and –M colony numbers of FPD and HD in primary plating (right). Granulocyte-macrophage progenitor (CFU-GM), macrophage (CFU-M), granulocyte (CFU-G), erythroid (BFU-E), and multipotent (CFU-GEMM) colonies were identified. (H) Representative flow charts and bar graphs for percentage of live cells that were CD14⁺ in primary colony formation assays from (G). Cells were collected from primary colony formation assay at day 14, and CD14⁺ percentage was assessed by flow cytometry. Statistical significance was calculated using Student’s t tests, and values indicating differences between FPD and healthy control samples are represented as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Fig. 2.. RUNX1-FPD stem and progenitor cells have a unique transcriptomic profile.

(A) UMAP of scRNA-seq analysis of ~122,021 fresh BM cells derived from individuals with FPD (n = 10). Data were processed using Seurat v3 integrateData function, and cells were assigned to 1 of the 18 cell clusters on the basis of the expression of cell-specific markers. (B) An overlay of FPD and HD UMAPs. UMAPs of an equal number of cells from FPD (n = 10, magenta) and HD (n = 4, teal) overlayed to visualize cellular differences in the identified cell populations. (C) The quantification of (B) UMAPs. Statistical significance was calculated using a two-way ANOVA test, and P values are shown within the bars. The percentage of cells in each cluster was calculated for each sample, and the mean values are shown in a bar graph. (D) Quantification of diffusion pseudotime. The percentage of cells in different clusters from each FPD and HD sample was compared with the total cell numbers and graphed on the y axis. Diffusion pseudotime, a measure of cell differentiation and lineage progression, was calculated and graphed on the x axis. The color bar on the top corresponds to specific cell populations. (E) RNA velocity data for FPD and HD. Violin plots display mean RNA velocity calculated for each cluster on the basis of scRNA-seq transcripts of canonical cluster defining genes. (F) Heatmap of DEGs in FPD cells compared with HD derived from scRNA-seq data. Genes up-regulated above 1.2-fold (P_adj < 0.05) in any indicated cell clusters were selected, and their expression across all indicated cell clusters is shown. (G) Gene network analysis of DEGs in HSCs from FPD compared with HD. The network was constructed using STRINGdb. The enrichment of genes in myeloid differentiation (yellow), cytokine signaling (green), and immune response (blue) pathways, as determined by Gene Ontology (GO) term search, is indicated by colored cells on the right of heatmap or in gene bubbles. Statistical significance for (E) was calculated using Student’s t test and Bonferroni correction with levels indicating differences between FPD and healthy control samples as *P ≤ 0.05 and ****P ≤ 0.0001. NK, natural killer; CTL, cytotoxic T lymphocyte; CLP, common lymphoid progenitor.

Fig. 3.. Up-regulation of inflammatory pathways and cytokines is observed in RUNX1-FPD HSPCs and their BM microenvironment.

(A) Enrichr pathway analysis using MsigDB Hallmark 2020 database for up-regulated DEGs (>1.2-fold change, P_adj < 0.05) in FPD HSCs, progenitors, and GMPs versus HD. The x axis shows a −log of P-adjusted values for the enrichment of genes in each pathway for indicated clusters. Pathways highlighted in red are related to inflammation. (B) Heatmap of leading genes for indicated pathways from (A). DEGs from indicated clusters are shown for the pathways. (C) Heatmap represents a log2-fold change in cytokine secretion in FPD (n = 40) compared with HD (n = 15). Extracellular fluid from FPD and HD fresh BM samples was collected and tested for the secretion of 65 cytokines using a Luminex assay. Fold differences in cytokine secretion were calculated for FPD over mean HD values for individual cytokines. Data were collected from three independent experiments. (D) Bar graph of significantly (P_adj < 0.05) up-regulated cytokines in FPD. Statistical analysis using a t test and Bonferroni correction was done to identify cytokines up-regulated above 1.5-fold change in FPD BM extracellular fluid compared with HD. (E to H) Quantification of colony formation ability (CFA) of FPD and HD CD34⁺ in the presence or absence (untreated, Unt) of inflammatory cytokines. Primary CD34⁺ cells were seeded at the density of 1000 cells per well and treated with LPS (E), CXCL-8, CCL-2, CCL-24, or IL-1β at 10 ng/ml (F). Colonies were counted at day 14 (1° CFA) and harvested for secondary plating at 100 thousand cells per well density for secondary colony-forming ability (2° CFA) at day 14 after seeding. CFU-GM, CFU-M, CFU-G, BFU-E, and CFU-GEMM colonies were identified. (G) The bar graph represents the number of CFU-GM colonies in FPD and HD samples with and without individual cytokine treatments as noted on the x axis. (H) The result of secondary plating for FPD cells treated with indicated cytokines. Statistical significance indicating differences between untreated and treated for FPD or healthy control samples for (E) to (H) as calculated using t test and indicated by *P ≤ 0.05 and **P ≤ 0.01.

Fig. 4.. Impact of MIF-CD74 signaling on skewed differentiation, activation of prosurvival, and inflammation in RUNX1-FPD.

(A) Heatmap representing the differential expression of cytokine receptors from scRNA-seq data. Data for the receptors of the up-regulated cytokines identified in FPD BM fluid (Fig. 3D) are shown for the indicated cell populations. Significantly (P_adj < 0.001) up-regulated (fold change >1.2) receptors are marked by ***. (B) The surface expression of CD74 in CD34⁺ cells of FPD (n = 10) and HD (n = 3) samples. Fresh BM cells were analyzed for the surface expression of CD34 and CD74 by flow cytometry. The populations are derived from live cells, and the bar graph shows the percentage of CD74⁺ in CD34⁺ cells. (C) CD74 levels were analyzed using immunohistochemistry (IHC) staining. RUNX1-FPD transformed to leukemia (MDS/CMML, n = 3), nontransformed RUNX1-FPD (n = 3), and HD (n = 4) BM biopsy slides were stained for CD74 (right), and CD74 expression was compared. For each slide, five different ROI s were analyzed, and data are shown as means with SE (left). (D) FPD (n = 1) and HD (n = 1) CD34⁺ cells were subjected to CRISPR editing to generate the knockout of CD74. The colony-formation ability of CD34⁺ cells upon CD74 knockout (sg-CD74) or nontargeting control (sg-NT) was assessed by plating 1000 cells per well in H4434 methocult (StemCell Technologies) using two or three technical replicates and quantification on day 14 after culture. (E) The results of colony-formation ability of HD (n = 8) and FPD (n = 12) CD34⁺ cells upon treatment with MIF (10 ng/ml). BM CD34⁺ cells seeded at 1000 cells per well density for primary plating (1° CFA) and 100 k cells per well for secondary plating (2° CFA). Statistical significance using Student’s t test indicates differences between FPD and healthy control samples. (F) The colony formation ability of HD (n = 5) and FPD (n = 7) CD34⁺ cells were assessed upon treatment with ISO-1 (25 nM) by seeding 1000 cells per well. (G) FPD MNCs (n = 15) were treated with ISO-1 at three different concentrations for 7 days. Megakaryocyte (CD41⁺/61⁺) population percentage was measured using flow cytometry (left). Representative flow charts for each condition are shown (right). (H) CD34⁺ cells from patients with FPD (n = 3) were cultured in vitro in the presence of MIF (10 ng/ml) or ISO-1 (100 nM) for 48 hours. Gene expression was measured using a nanostring inflammation panel, and DEGs were calculated for MIF or ISO-1–treated cells compared with vehicle-treated controls. Volcano plots (right) show genes up-regulated (magenta colored) or down-regulated (teal colored) by log2 fold change values. P values are graphed on the y axis. The heatmap (left) shows GSEA using DEGs in MIF or ISO-1–treated samples. (I) The phosphorylation of proteins in HD and FPD after ISO-1 treatment. CD34⁺ cells were cultured in the presence of ISO-1 (100 nM) for 48 hours. Phosphorylation of indicated proteins was measured through intracellular flow cytometry. Data show fold change differences in MFI of treated cells over untreated. n = 5 per group. Statistical significance using Student’s t test indicates the differences between treatments and untreated samples. Statistical significance was calculated using Student’s t tests and values indicating differences between FPD and healthy control samples or vehicle treatment and treated groups as *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

Fig. 5.. Digital spatial profiling shows up-regulation of mTOR/PI3K signaling in FPD.

(A and B) The network analysis of analyzed proteins using GeoMX DSP done on RUNX1-FPD (transformed MDS/CMML, n = 3 and nontransformed, n = 3) and HD (n = 4) control BM biopsy slides. MFIs were normalized to GAPDH. Using STRINGdb, the protein target markers were evaluated for network analysis. The up-or down-regulated proteins in patients transformed [(A) MDS/CMML] and nontransformed (B) FPD compared with HD controls are color-coded as red or blue, respectively. (C) The heatmap result of indicated proteins in FPD and HD samples, showing the fold change of FPD samples over average HD values. (D) The violin plots of MFI values normalized to GAPDH for indicated proteins across FPD and HD samples. (E) The correlation plots of CD74 expression and mTOR pathway proteins S6, p-S6, and p-4EBP1. CD74 expression for the indicated FPD (transformed and nontransformed) samples and HD controls was determined using IHC staining (shown in Fig. 4C). Normalized MFI values for S6, phospho-S6, and phospho-4EBP1 for FPD, MDS-C MML, and HD controls. Each color corresponds to a specific patient sample as shown in (C) and (D). Statistical significance was calculated using Student’s t tests and values indicating differences between FPD and healthy control samples or vehicle treatment and treated groups as *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

Fig. 6.. Restoring phenotypic defects in RUNX1-FPD through targeting MIF or downstream PI3K, mTOR, or JAK2 pathways.

(A and B) The results of pathway inhibition on differentiation of FPD cells. MNCs from FPD (n = 12) BM samples (20,000 to 30,000 cells per well) were cultured in the presence of inhibitors at three different concentrations selected on the basis of colorimetric cell viability assay (fig. S8A). In vitro cell differentiation was assessed after 7 days of culture using flow cytometry analysis. (A) Bar graphs show the mean % of live cells for indicated populations. (B) Flow plot representations for bar graphs shown in (A). (C) Quantification of colony formation ability of FPD CD34⁺ cells upon treatment with inhibitors. Colony formation was assessed by seeding CD34⁺ cells at 1000 cells per well density and counting colonies at day 14 after culture. CFU-GM, CFU-M, CFU-G, BFU-E, and CFU-GEMM colonies were identified. (D and E) The effect on phosphorylation of indicated proteins upon inhibitor treatment in FPD MNCs. Intracellular flow cytometry was used to measure phosphorylation of proteins in FPD MNCs (n = 4, duplicates or triplicates) upon treatment with AZD2014 (100 nM), rapamycin (0.1 nM), idelalisib (1000 nM), and ruxolitinib (250 nM), after 48 hours. (D) The representative flow plots for each phosphorylated moiety. (E) The heatmap of average fold changes in MFI of treated cells compared with vehicle. Statistical significance was calculated using a t test, and values indicating differences between treatment versus baseline are shown as *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

Fig. 7.. In vivo evaluation of MIF/CD74, mTOR, or JAK2 inhibition in Runx1^R188Q/+-mutated mice.

(A) The colony-formation ability of cells from WT and Het mice. LSK (lineage-Sca1⁺cKit⁺) cells were isolated from mice and seeded at 600 cells per well density in the presence of MIF (10 ng/ml) or ISO-1 (50 nM) using methocult M3434 (StemCell Technologies). The colonies were counted at day 7 after culture (1° CFA) and harvested for secondary plating at 100 k cells per well density (2° CFA). (B) The schematic of in vivo treatment. WT and Het mice (age and sex matched, n = 3 or 4) were treated with ISO-1 or ruxolitinib daily and rapamycin every other day for 8 weeks. (C) Platelet activation marker (CD62P⁺) positivity was measured in CD41⁺/CD61⁺ cells upon thrombin (0.1 U/ml treatment at the end of 8 weeks of mice treatment), and data are represented as a bar graph (left). The analysis was done by measuring CD62P percentage out of CD41⁺/61⁺ cell population. The representative flow charts of CD62P MFIs (right). The PB of WT and Het mice was treated with thrombin (0.1 U/ml) for 15 min before analysis. (D) BM cells were harvested, and the percentages of monocyte (CD11b⁺), neutrophils (Ly6-C⁺/−G⁺), and macrophages (CD11b⁺/F4–80⁺) from live cells were determined using flow cytometry. (E) Representative flow charts are shown for the indicated cell population for Het mice. Statistical significance was calculated using t test and are shown as *P ≤ 0.05 and **P ≤ 0.01.