Innate immune mechanisms hijacked by leukemia-initiating stem cells for selective advantage and immune evasion in Ptpn11-associated JMML

SUMMARY

Juvenile myelomonocytic leukemia (JMML) originates from mutated hematopoietic stem cells. The mechanism by which mutant stem cells are sustained, leading to leukemia development, remains elusive. By comprehensively examining transcriptomic profiles, cell compositions, developmental trajectories, and cell-cell interactions across various stages of tumor cell development in a mouse model of Ptpn11 mutation-associated JMML, we find that Ptpn11^E76K/+ mutant stem cells exhibit de novo activation of the myeloid transcriptional program and markedly increased expression of innate immunity-associated antimicrobial peptides and pro-inflammatory proteins, particularly S100a9 and S100a8. Biological experiments confirm that S100a9/S100a8 confer a selective advantage to mutant stem cells through autocrine effects and facilitate immune evasion by recruiting and promoting immune-suppressive myeloid-derived suppressor cells in the microenvironment. Importantly, pharmacological inhibition of S100a9/S100a8 signaling effectively impede leukemia development from Ptpn11^E76K/+ mutant stem cells. These findings collectively suggest that JMML-initiating cells exploit innate immune and inflammatory mechanisms to establish clonal dominance.

In brief

Zheng et al. demonstrate that hematopoietic stem cells harboring the Ptpn11^E76K mutation show de novo activation of the myeloid transcriptional program, resulting in high production of innate immunity-associated proteins, particularly S100a9 and S100a8. These pro-inflammatory proteins confer a selective advantage to the leukemia-initiating cells and contribute to immune evasion.

Graphical abstract

INTRODUCTION

Juvenile myelomonocytic leukemia (JMML), a pediatric myeloproliferative neoplasm, manifests as a clonal hematopoietic disorder characterized by the excessive production of myeloid cells. This disease originates from driver mutations acquired in hematopoietic stem cells (HSCs) and is propagated and sustained by these mutated stem cells, known as leukemia-initiating cells.^1–4 JMML has limited therapeutic options. Relapse remains the primary cause of treatment failure, most likely due to the persistence of therapy-resistant, self-renewing leukemia-initiating cells.^1–4 Addressing this issue is crucial for improving treatment outcomes in JMML patients.

Genetically, JMML is associated with mutations in genes encoding signaling proteins involved in the rat sarcoma (RAS)/extracellular signal-regulated kinase pathway, including protein tyrosine phosphatase, non-receptor type 11 (PTPN11), RAS, neurofibromin 1 (NF1), casitas B-lineage lymphoma (CBL), and others.^1–4 These mutations play a causal role in driving JMML development.^5–8 JMML arises from an HSC harboring a genetic mutation, but the molecular mechanisms by which the initially mutated stem cell acquires a competitive advantage and evades immune surveillance remain unexplored. Additionally, the specific reasons behind the propensity of a disease-associated mutation in the leukemia-initiating cell to induce a myeloid malignancy are not fully understood. Understanding these mechanisms could illuminate strategies for therapeutically targeting and eliminating JMML-initiating stem cells in established disease.

Of the genetic lesions identified in JMML, the protein tyrosine phosphatase PTPN11 (Src homology region 2 domain-containing phosphatase-2 [SHP-2]), a positive regulator of RAS signaling,^9,10 is the most frequently mutated (heterozygous).^11,12 Mutations in PTPN11 lead to a significant increase in the catalytic activity of SHP-2.^12,13 Patients carrying PTPN11 gain-of-function mutations have the worst prognosis among all subtypes of JMML.^14–17 To elucidate the mechanisms underlying the pathogenesis of PTPN11-mutated JMML, our laboratory created a conditional Ptpn11 allele in mice with the Ptpn11^E76K mutation, the most common PTPN11 mutation found in JMML,^11,12 and developed an inducible disease model.^6,18 Induction of the Ptpn11^E76K mutation in the hematopoietic system resulted in a JMML-like myeloproliferative neoplasm with complete penetrance, affirming the causative role of this mutation in JMML.⁶ In the present study, we take advantage of this unique disease model to investigate the cellular and molecular mechanisms involved in the initial pathological process following induction of the disease mutation. Our findings from single-cell transcriptomic profiling and experimental validations reveal an aberrant activation of innate immune responses in the mutated stem cells. These leukemia-initiating cells exploit innate immune and inflammatory mechanisms to gain a competitive advantage and evade antitumor immunity, ultimately leading to clonal dominance.

RESULTS

Aberrant activation of innate immune and inflammatory responses in Ptpn11^E76K/+ mutant stem cells

To explore the intricate mechanisms of JMML pathogenesis, we conducted a comprehensive single-cell RNA sequencing (scRNA-seq) analysis on bone marrow (BM) cells isolated from mice with an induced JMML-like myeloproliferative neoplasm (Ptpn11^E76K/+/Mx1-Cre)⁶ and wild-type (WT; Ptpn11^+/+/Mx1-Cre) control littermates, supplemented with 2% flow cytometry-sorted, HSC-enriched Lineage⁻Sca-1⁺c-Kit⁺ (LSK) cells. Utilizing gene expression pattern-based cell clustering, we identified 12 distinct cell clusters within the BM population on a t-distributed stochastic neighbor embedding (t-SNE) plot (Figure S1A). Clear distinctions among these clusters were evident in the heatmap representation of the expression patterns of the top 10 differentially expressed genes (DEGs) in each cluster (Figure S1B). By leveraging reference datasets,^19,20 we were able to further identify various hematopoietic cell types at different developmental stages in the clusters. Specifically, HSCs were localized to cluster 4, based on the expression of the known HSC marker genes Meis1, Hoxa5, Zfpm1, Mecom (Evi1), and Mpl. Additionally, we regrouped clusters with overlapping marker profiles: clusters 2, 5, and 6 were combined and reclassified as monocytes, while clusters 0, 1, and 3 were grouped as neutrophils. Granulocyte-macrophage progenitors (GMPs), megakaryocytic-erythroid progenitors, T cells, B cells, and other cell types were also identified (Figure S1C). Cell type-specific signature genes were indeed well represented in the identified cell clusters (Figure S2). Notably, Ptpn11^E76K/+ mutant HSCs exhibited reduced abundance, while monocytes and neutrophils displayed an increase compared to their WT (Ptpn11^+/+) counterparts (Figure S1C). The reduction of Ptpn11^E76K/+ HSCs and the enhanced myeloid cell production suggest hyperactivation of these mutant stem cells.

Further clustering of the HSC population based on gene expression profiles revealed two distinct subpopulations. One cluster showed high expression of Pbx1, Erg, Tal1, Meis1, Gfi1b, Mpl, and CD93, consistent with a long-term HSC (LT-HSC) identity^21,22 (Figure S3). Another cluster was classified as short-term HSCs (ST-HSCs). Based on this classification, the percentage of LT-HSCs decreased, whereas the percentage of ST-HSCs increased in Ptpn11^E76K/+ mice compared to their WT littermates (Figure 1A). Figure 1B presents the top 20 most significant DEGs between LT-HSCs and ST-HSCs. Notably, a total of 177 genes were significantly differentially expressed in Ptpn11^E76K/+ LT-HSCs compared to WT LT-HSCs (Figure 1C). Surprisingly, Gene ontology (GO) enrichment analyses of these DEGs highlighted the predominant elevation of defense reactions to bacterial infection, innate immune response, Toll-like receptor 4 (TLR4) signaling, and inflammation-associated pathways (Figure 1D). Gene set enrichment analysis (GSEA) revealed significant downregulation of gene sets typically upregulated in HSCs compared to GMPs²³ in Ptpn11^E76K/+ HSCs. In parallel, gene set variation analysis (GSVA) showed that genes associated with activated HSCs and multipotent progenitors were enriched in Ptpn11^E76K/+ HSCs, whereas genes in control HSCs were enriched for canonical HSC signature genes, retinoic acid response genes, and dormancy-associated genes (Figure 1E). These findings suggest that Ptpn11 mutant stem cells exhibit a loss of stemness, hyperactivation, and a priming toward differentiation. Similarly, 173 DEGs were identified in Ptpn11^E76K/+ ST-HSCs compared to WT ST-HSCs (Figure S4A), with Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis indicating dysregulation of antiviral immune response pathways, ribosome biogenesis, and spliceosome function (Figure S4B).

Figure 1. Aberrant activation of innate immune and inflammatory responses in Ptpn11^E76K/+ mutant stem cells.

(A) HSCs were segregated into 2 distinct cell clusters equivalent to LT-HSCs and ST-HSCs based on their gene expression profiles. t-SNE plots of WT (Ptpn11^+/+) and Ptpn11^E76K/+ mutant HSCs are shown.

(B) Heatmap displaying the top 20 DEGs in cell clusters. Each column represents one cell, and relatively high and low expression are represented in yellow and purple, respectively.

(C) Volcano plot depicting all DEGs in Ptpn11^E76K/+ LT-HSCs compared to WT counterparts. Significantly upregulated DEGs are labeled in red and downregulated DEGs in blue.

(D) Significant DEGs in Ptpn11^E76K/+ over WT LT-HSCs were subjected to GO enrichment analysis and sorted based on p values, with p < 0.05 considered significant. The size of the dots in the plot reflects the number of enriched genes in the respective terms, and the dot shape represents different GO classifications.

(E) GSEA plot illustrating downregulation of gene sets normally upregulated in HSCs compared to GMPs in Ptpn11^E76K/+ HSCs (left). The colored band represents enrichment score (ES) values, with red indicating positive enrichment in Ptpn11^E76K/+ HSCs and blue indicating negative enrichment in control HSCs. Right: GSVA results for enriched genes in Ptpn11^E76K/+ and control HSCs.

Further examination of stem cell self-renewal or differentiation-associated signature genes revealed widespread deregulation in Ptpn11^E76K/+ LT-HSCs and ST-HSCs as compared to WT counterparts (Figure 2A). Notable downregulated genes in Ptpn11^E76K/+ LT-HSCs included Hoxa5, Hoxa6, Hoxa7, and Hoxb8, while upregulated genes comprised Hoxb2, Hoxb3, and Hoxb4. Interestingly, upregulated expression of the myeloid differentiation-related genes Cebpb, Cebpe, Cebpg, and Cited2 was noticed in these mutant LT-HSCs. In Ptpn11^E76K/+ ST-HSCs, downregulated genes included Hoxa3, Gata1, Klf4, Cebpa, and Elane, while upregulated genes encompassed Mix, Irf5, Irf8, Ctss, Gata3, and Csf1r. The most significant DEGs in Ptpn11^E76K/+ LT-HSCs and ST-HSCs compared to WT counterparts are shown in Figure 2B. Surprisingly, myeloid cell-specific genes and genes associated with antipathogen and innate immune responses normally activated in myeloid cells, such as S100a9, S100a8, S100a6, S100a11, Retnlg, Ngp, Camp, Lcn2, Lyz2, Wfdc21, Chil3, and Pglyrp1, were highly expressed in Ptpn11^E76K/+ LT-HSCs. The expression levels of S100a9 and S100a8, also known as myeloid-related proteins 9 and 8, increased approximately 29- and 24-fold, respectively, standing out as the most strikingly upregulated among all significant DEGs in Ptpn11^E76K/+ LT-HSCs. Additionally, Cxcl2, also known as MIP-2α, a chemokine typically secreted by monocytes/macrophages and a powerful chemoattractant for polymorphonuclear leukocytes involved in many immune responses, including wound healing, cancer metastasis, and angiogenesis, was overexpressed in these mutant stem cells.

Figure 2. Proinflammatory proteins S100a9 and S100a8 and their receptors are highly expressed in Ptpn11^E76K/+ mutant stem cells.

(A) Heatmap illustrating the expression of selected signature genes in WT and Ptpn11^E76K/+ ST-HSCs and LT-HSCs.

(B) Heatmap displaying the expression of most significantly up- or downregulated genes in WT and Ptpn11^E76K/+ ST-HSCs and LT-HSCs.

(C) Violin plots illustrating the expression levels of the indicated cell surface markers in WT and Ptpn11^E76K/+ ST-HSCs and LT-HSCs.

Several cell surface molecules were differentially expressed in Ptpn11^E76K/+ LT-HSCs. Among the most significant DEGs, Cd52 and Cd9 were upregulated, while the transcriptional expression of the early stem/progenitor cell marker Cd34 was drastically reduced (Figure 2C). In addition, Cd33, P2ry14/Gpr105, and Gpr150 showed a marked upregulation in Ptpn11^E76K/+ LT-HSCs. These unique expression patterns of cell surface molecules in Ptpn11 mutant LT-HSCs hold promise for their utilization as therapeutic targets or biomarkers for JMML stem cells. Furthermore, Rage/Ager, the receptor for the S100a9/S100a8 heterodimer (calprotectin)²⁴ typically expressed on myeloid immune cells, exhibited substantial upregulation in Ptpn11^E76K/+ LT-HSCs, indicating potential autocrine feedback activities in these cells. Given that S100a9 expression was significantly upregulated in Ptpn11^E76K/+ LT-HSCs (Figure S5A), we sought to identify transcriptional factors potentially associated with this upregulation. To this end, we conducted Venn diagram data analysis involving the 177 DEGs in Ptpn11^E76K/+ LT-HSCs and 58 transcriptional factors related to S100a9. This analysis revealed Spi1 and Smarca4 (Figure S5B). Of the 47 dysregulated transcriptional factors in Ptpn11^E76K/+ LT-HSCs, Spi1 showed a significant upregulation, whereas Smarca4 was downregulated (Figure S5C), indicating that the elevated levels of Spi1 may have contributed to the observed overexpression of S100a9.

Profound impact on the myeloid lineage by the Ptpn11^E76K mutation

The influence of the Ptpn11^E76K mutation extended beyond the stem cell population, significantly affecting myeloid-committed GMPs. The percentage of GMPs in Ptpn11^E76K/+ mice was reduced (Figure S1C), a finding further validated by additional biological assays (Figure S6A, top). This reduction in mutant GMPs was likely associated with a decrease in mutant HSCs (Figure 1A) and their enhanced or spontaneous differentiation into myeloid lineage cells (Figure S6A, bottom). Gene expression profiling identified 4 distinct cell clusters in GMPs, revealing heterogeneity among these progenitors (Figure 3A). Interestingly, Ptpn11^E76K/+ GMPs exhibited a notable shift in cell composition, with cluster 3 emerging as an overrepresented subpopulation, constituting approximately 60% of the total. This cluster within the mutant GMP population appears to be an altered counterpart of cluster 1, a major cluster in the control group, exhibiting additional transcriptional changes. The heatmap representation of the top 10 DEGs in each cell cluster highlighted clear differences among these clusters, with cluster 1 enriched in Prom1, Clu, Mgam, Gpx3, and Slco4c1 and cluster 3 marked by high expression of Fbp1, Tmem53, Cracr2b, and Dlg2 (Figure 3B). Overall, 127 genes were significantly differentially expressed in Ptpn11^E76K/+ GMPs compared to their WT counterparts (Figure 3C). GO analysis of the DEGs underscored enrichment in innate immune and inflammatory pathways in Ptpn11 mutant GMPs (Figure 3D). These included pathways related to the positive regulation of immune response, neutrophil activation, neutrophil-mediated killing of bacteria, defense response to bacteria, and innate immune response. GSEA revealed an enrichment of genes typically associated with immune response, response to bacteria, and monocyte-dendritic cell progenitors²⁵ in Ptpn11^E76K/+ GMPs relative to WT GMPs (Figure 3E). Cluster 3, a major subpopulation within Ptpn11^E76K/+ GMPs, displayed high expression of genes such as Arl11, Fbp1, Slc31a2, Hnmt, Tmem53, and Cracr2b, among others (Figure S6B). Differential gene expression analysis between cluster 3 and cluster 1, a major subpopulation in WT GMPs, revealed 114 genes with distinct expression patterns (Figure S6C). KEGG pathway analyses illustrated the upregulation of genes involved in autoimmune responses, bacterial infection responses, natural killer cell-mediated cytotoxicity, neutrophil extracellular trap formation, and ribosome biogenesis, whereas downregulated pathways included phagosome, ribosome, RNA transport, spliceosome, RNA degradation, oxidative phosphorylation, and thermogenesis pathways in Ptpn11^E76K/+ GMPs (Figure S6D).

Figure 3. Enhanced innate immune signaling in Ptpn11^E76K/+ GMPs.

(A) GMPs were categorized into 4 different cell clusters based on their gene expression profiles. t-SNE plots of WT and Ptpn11^E76K/+ mutant GMPs are displayed.

(B) Heatmap illustrating the expression patterns of the top 10 representative genes in each cell cluster. Each column represents one cell, and relatively high and low expression are represented in yellow and purple, respectively.

(C) Volcano plot depicting DEGs in Ptpn11^E76K/+ GMPs. Significantly upregulated DEGs are labeled in red and downregulated DEGs in blue.

(D) DEGs in Ptpn11^E76K/+ GMPs were subjected to GO enrichment analysis and sorted by p values, with p < 0.05 considered significant. The size of the dots in the plot reflects the number of enriched genes in the respective terms, and the dot shape represents different GO classifications.

(E) GSEA plots depicting enrichments in gene sets associated with immune response, response to bacteria, and monocyte-dendritic cell (DC) progenitors in Ptpn11^E76K/+ GMPs. The colored band represents ES values, with red indicating positive enrichment in Ptpn11^E76K/+ GMPs and blue indicating negative enrichment in control GMPs.

We also examined the impact of the Ptpn11^E76K mutation on monocytes and neutrophils. Gene expression profile-based cell clustering demonstrated heterogeneity in monocytes. Seven distinct cell clusters were identified in monocytes (Figure 4A). The Ptpn11^E76K/+ monocyte compartment demonstrated notable changes in cell composition. The linker histone H1 family members (Hist1h2ab, Hist1h2af, Hist1h2bm, Hist1h2bn, Hist1h3b, and Hist1h3f), Sirpb1c, Ms4a8a, Apoe, Slfn5, Pla2g7, and P2ry6 were highly expressed in cluster 2 and cluster 3, which were unique in the Ptpn11 mutant monocyte population (Figure 4B). Similarly, neutrophils also exhibited heterogeneity, with altered cell composition in Ptpn11^E76K/+ mice (Figure 4C). Upregulated genes in Ptpn11^E76K/+ clusters included mitochondrial protein synthesis-associated Lars2; innate immunity-associated Chil5; the chemokine Ccl6; Arginase, type 2 (Arg2); and glycolysis-associated Ldhc, while downregulated genes comprised Lipg, Cmah, Qsox1, Calr, Pdia6, Sec61a1, Prok2, Wfdc17, Ifitm1, and others (Figure 4D).

Figure 4. Impact of the Ptpn11^E76K mutation on monocytes and neutrophils.

(A) Monocytes were divided into 7 different cell clusters based on their gene expression patterns. t-SNE plots of WT and Ptpn11^E76K/+ mutant monocytes are shown.

(B) Heatmap displaying the expression patterns of the top 10 representative genes in each cluster. Each column represents one cell, and relatively high and low expression are represented in yellow and purple, respectively.

(C) Neutrophils were segregated into 6 distinct cell clusters according to their gene expression profiles. t-SNE plots of WT and Ptpn11^E76K/+ neutrophils are displayed.

(D) Heatmap illustrating the expression patterns of the top 10 representative genes in each cluster.

We then explored whether the transcriptional landscape changes in Ptpn11^E76K/+ cells across different developmental stages shared commonality; the top 50 significant DEGs in Ptpn11^E76K/+ and WT stem cells, GMPs, monocytes, and neutrophils are shown in Figure 5A. Venn diagram data analysis for DEGs in the different cell populations identified 44 co-events (Figure 5B). Remarkably, these genes were consistently upregulated or downregulated in Ptpn11^E76K/+ cells throughout all developmental stages, without any exceptions (Figure 5C). This observation implies that they were cell-intrinsically dysregulated by the Ptpn11^E76K mutation. Many of these co-DEGs were associated with innate immune signaling and inflammatory pathways, including S100a11, Retnlg, and Lyz2. Interestingly, genes involved in ribosomal biogenesis, such as Rplp0, Rps3, and Rpl21, were upregulated, while Rpl41, Rpl37a, Rps28, Rpl38, Rpl23a, and Rps15 were repressed. Dysregulation of ribosomal biogenesis and function can collectively contribute to cellular abnormalities, genomic instability, and the development of malignancies.^26,27 These findings underscore that the impact on ribosomal function is a common pathological effect of the Ptpn11 mutation across different cell types.

Figure 5. Effects of the Ptpn11^E76K mutation on ribosomes across developmental stages.

(A) Heatmaps depicting the expression of the genes that are up- or downregulated in Ptpn11^E76K/+ HSCs, GMPs, monocytes, and neutrophils. Each column in the heatmaps represents one cell, and relatively high and low expression are represented in yellow and purple, respectively.

(B) Venn diagram data analysis was conducted on the DEGs in Ptpn11^E76K/+ HSCs, GMPs, monocytes, and neutrophils. Different colors were superimposed to represent shared genes, and the numbers of these shared genes are indicated.

(C) Bar graphs illustrating the results of the differential expression analysis for the shared DEGs (44) in Ptpn11^E76K/+ HSCs, GMPs, monocytes, and neutrophils. Upregulated genes are marked in red and down-regulated genes in black.

Altered developmental trajectories and cell-cell communication in Ptpn11^E76K/+ mutant stem cells

Monocle 2 was used to identify dynamically regulated genes through the differentialGeneTest function (the dpFeature method). These genes were then grouped into modules using Monocle’s built-in hierarchical clustering (hclust), which organizes genes based on shared expression dynamics along the pseudotime axis. Three distinct gene expression modules were identified during the differentiation process from stem cells to monocytes and neutrophils. Differences in the dynamic expression patterns of genes enriched in all modules were observed during the differentiation of Ptpn11^E76K/+ stem cells compared to WT HSCs (Figure S7). Notably, a larger number of genes showed dynamic expression changes within module 2, whereas fewer such changes were observed in module 3, reflecting distinct regulatory responses associated with the Ptpn11 mutation. GO enrichment analyses of three modules were conducted to assess the biological implications of these dynamic changes. Module 2 was enriched for translation-related genes, indicating enhanced translational activity during the differentiation of Ptpn11^E76K/+ stem cells. Module 3 showed decreased genes involved in type I/II interferon responses during mutant stem cell differentiation.

Dimensionality reduction was then conducted using the reduceDimension function, and cell trajectories were reconstructed with the orderCells function. The results revealed that the overall progression of Ptpn11^E76K/+ mutant stem cells to GMPs, monocytes, and neutrophils was largely comparable to that of control HSCs (Figure 6, first row). The diffusive trajectory within Ptpn11^E76K/+ GMPs showed minor overall alterations (Figure 6, second row). However, closer examination of individual cell clusters revealed that cluster 1 of Ptpn11^E76K/+ GMPs displayed enhanced differentiation within this subset (Figure 6, second row), which may explain its marked reduction compared to WT cluster 1. Clusters 2 and 4 of Ptpn11^E76K/+ GMPs also showed a minor shift toward later-stage subpopulations. Notably, Ptpn11^E76K/+ monocytes followed two distinct developmental trajectories, in contrast to the single pathway observed in WT monocytes. This suggests the emergence of diverse monocyte subpopulations along separate developmental routes in the mutant mice. The inferred pseudotime distribution for mutant monocytes showed accelerated differentiation/maturation (Figure 6, third row). In contrast, Ptpn11^E76K/+ neutrophils did not have significant changes in developmental trajectory or pseudotime distribution (Figure 6, fourth row), indicative of normal maturation in this population.

Figure 6. Altered developmental trajectories in Ptpn11^E76K/+ mutant stem cells.

Dimensionality reduction was conducted using the reduceDimension function, and cell trajectories were reconstructed with the orderCells function. The differentiation trajectories of WT and Ptpn11^E76K/+ HSCs to GMPs and further to monocytes/neutrophils are shown in the first row. Developmental trajectories for WT and Ptpn11^E76K/+ GMPs, monocytes, and neutrophils are shown in the second through fourth rows (left). The directions of differentiation for each cell type are indicated by arrows. Pseudotime distributions of GMPs and GMP clusters, monocytes, and neutrophils are shown on the right.

Cell-cell communication analyses based on ligand-receptor expression revealed increased signaling from neutrophils to HSCs in Ptpn11^E76K/+ mice compared to WT littermates, while signaling in the reverse direction—from HSCs to neutrophils—was reduced (Figures S8A and S8B). Furthermore, interactions among Ptpn11^E76K/+ stem cells increased relative to those in WT stem cells (Figures S8A and S8B). A closer examination of neutrophil-stem cell communications indicated that interactions mediated by interleukin 1β, transforming growth factor β, and Oncostatin M were enhanced in Ptpn11^E76K/+ mice compared to those in WT mice (Figure S8C), providing additional evidence that Ptpn11-mutated stem cells were situated in an inflammatory microenvironment.

De novo activation of the myeloid transcriptional program in Ptpn11^E76K/+ mutant stem cells

Cell identity and functional specificity are collectively governed by transcription factors and the expression levels of their target genes. Total unique molecular identifier counts were higher in Ptpn11^E76K/+ stem cells compared to their WT counterparts, indicating elevated overall transcriptional output (Figure S9), consistent with increased cellular activity in Ptpn11-mutated stem cells. To further elucidate the mechanisms through which the Ptpn11^E76K mutation influences cell behavior, we conducted single-cell regulatory network inference and clustering analysis (transcriptional factor regulon analysis). The activities of many transcription factors in Ptpn11^E76K/+ stem cells, GMPs, monocytes, and neutrophils were altered compared to those in their WT counterparts, as indicated by regulon activity scores. In Ptpn11^E76K/+ stem cells, the transcriptional activities of Atf3, Egr1, Jun, Jund, Klf6, Fos, and Gata2 were significantly decreased, while those of Irf7, Irf8, Maf, and Myc were increased (Figure S10A). Importantly, regulon specificity scores (RSSs), reflecting the association between regulon activities and cellular specificity, revealed that, among these differentially functioning transcription factors, the myeloid transcription factors Ets1, Cebpe, and Nfe2 were highly associated with the identity specificity of Ptpn11^E76K/+ stem cells as opposed to Tcf7l2, Relb, and Irf5 for WT HSCs. At the GMP level, the activities of the myeloid-specific transcription factors Cebpe, Cebpb, and Ets1 were markedly increased in Ptpn11^E76K/+ GMPs, and their cellular specificity was likely determined by Cebpe, Ets1, and Myc compared to Cebpe, E2f1, and Klf6 in WT GMPs (Figure S10B). The activities of transcription factors Irf7, Cebpb, Fos, Irf5, Irf8, Klf4, and Maf were significantly higher in Ptpn11^E76K/+ monocytes than in WT cells, and the identity specificity of Ptpn11 mutant monocytes was highly associated with transcription factors Irf7, Mafg, and Irf8, according to RSS (Figure S10C). Similarly, the distinction in transcriptional factor determinants influencing the specificity of Ptpn11^E76K/+ neutrophils (Mafg, Cebpb, and Junb) compared to those governing WT neutrophils (Maf, Irf8, and Junb) was also apparent (Figure S10D).

Consistent with the regulon results, Ptpn11^E76K/+ stem cells and GMPs demonstrated heightened cell cycling, as evidenced by the loss of quiescence (the G₀ phase in the cell cycle) and an increased number of cells in the G₂/M phase, based on single-cell transcriptomes and a reported predictor for allocating individual cells to the G₀, G₁/S, and G₂/M cell cycle phases²⁸ (Figure S11A). The cell division/replication-related histone H1 family members (Hist1h1c, Hist1h1d, Hist1h1e, and Hist1h2ae) and CDK1 were upregulated in both Ptpn11^E76K/+ stem cells and GMPs (Figure S11B). Additionally, GSEA revealed a significant enrichment of cell cycling-associated gene sets in Ptpn11^E76K/+ stem cells, with a similar trend observed in Ptpn11^E76K/+ GMPs (Figure S11C).

S100a9 and S100a8, aberrantly expressed in Ptpn11^E76K/+ stem cells, contribute significantly to leukemogenesis

Given the pronounced upregulation of S100a9 and S100a8 in Ptpn11^E76K/+ mutant stem cells (Figure 2B) and their diverse biological functions,^29,30 we investigated their potential role in JMML pathogenesis. We first confirmed a significant (>8-fold) increase in S100a9 and S100a8 expression in mutant stem cells from Ptpn11^E76K/+ mice compared to WT HSCs (Figure 7A). Next, we analyzed scRNA-seq data from a JMML patient with a PTPN11 mutation available in the Gene Expression Omnibus (GEO) database. Consistent with our findings in the mouse model (Figure 1A), the proportion of leukemic stem cells in the Lineage⁻CD34⁺ cell population was reduced in the patient compared to that of normal HSCs from a healthy donor (Figure S12A). Notably, lineage-specific genes, particularly myeloid genes, including S100a8 and S100a9, were markedly upregulated in leukemic stem cells (Figure S12B), in line with our observations in the mouse model. Similarly, analysis of bulk RNA-seq data from stem/progenitor cells of five JMML patients with PTPN11 mutations confirmed elevated expression of these genes (Figure S12C). Importantly, we experimentally verified that S100a9 and S100a8 expression in CD34⁺ leukemic stem/progenitor cells from seven PTPN11-mutated JMML patients was increased ~7-fold compared to normal CD34⁺ hematopoietic stem/progenitors (Figure 7B).

Figure 7. S100a9 and S100a8, aberrantly expressed in Ptpn11^E76K/+ stem cells, contribute significantly to leukemogenesis.

(A) mRNA levels of the indicated genes in the stem cells (Lineage⁻Sca-1⁺c-Kit⁺CD150⁺CD48⁻) sorted from WT and Ptpn11^E76K/+ mice were determined by RT-qPCR (n = 6 mice/genotype).

(B) CD34⁺ early hematopoietic progenitors isolated from healthy BM samples and PTPN11 mutated JMML patients were analyzed for mRNA levels of the indicated genes by RT-qPCR (n = 5–7 individuals/group).

(C) Stem cells freshly sorted from WT and Ptpn11^E76K/+ mice were cultured (5 × 10² cells) in StemSpan medium supplemented with stem cell factor (50 ng/mL), thrombopoietin (50 ng/mL), Fms-Like tyrosine kinase 3 ligand (50 ng/mL), and tasquinimod (5.0 μM) or vehicle. Seven days later, total cell numbers were determined (n = 3 mice/group).

(D) Colonies derived from WT and Ptpn11^E76K/+ mutant stem cells (5 × 10² cells) in the presence of tasquinimod (5.0 μM) or vehicle were determined by colony-forming unit assays as described in STAR Methods (n = 6 mice/group).

(E) Transmigration efficiency of PMN-MDSCs (CD11b⁺Ly6G⁺) toward WT and Ptpn11^E76K/+ stem cells in the presence of tasquinimod (5.0 μM) or vehicle were assessed as described in STAR Methods (n = 6 mice/group).

(F–Q) BM cells collected from Ptpn11^E76K/+/Mx1-Cre/mTmG mice were mixed with WT BM cells from congenic BoyJ mice at a 10:1 ratio and transplanted into lethally irradiated BoyJ mice (n = 13 mice/group). Four weeks post transplantation, tasquinimod or vehicle was administered to mice via drinking water (5 mg/kg body weight/day) for 4 weeks (F). Total mutant cells (GFP⁺) in the peripheral blood (PB) (G) and myeloid cells (Mac-1⁺) in the GFP⁺ mutant cell compartment (H) and the entire PB (I) were determined at the indicated time points. Mice were euthanized 4 weeks after the treatment. White blood cell (WBC) counts, neutrophils, monocytes, and red blood cell (RBC) counts in the PB were analyzed (J). Spleens were weighed (K). Total mutant cells (GFP⁺) and Mac-1⁺ myeloid cells in the GFP⁺ mutant compartment and the entire spleen and BM were determined (L and M). The pool size of GFP⁺ stem cells (GFP⁺Lineage⁻Sca-1⁺c-Kit⁺CD150⁺CD48⁻) in the BM, GFP⁺ early progenitors (GFP⁺Lineage⁻Sca-1⁺c-Kit⁺) in the spleen (N), cell cycling status (O), and apoptosis (P) in GFP⁺ stem cells were determined. (Q) BM sections prepared from tasquinimod- or vehicle-treated Ptpn11^E76K/+ mice were immunostained for stem cells (CD150⁺CD3⁻B220⁻Ter119⁻CD48⁻) (cyan) and PMN-MDSCs (CD11b⁺Ly6G⁺) (yellow) (n = 10 mice/group, with >76 stem cells/group examined in total). The distance between stem cells and the closest PMN-MDSCs was determined. Scale bars: 20 μM. Data are presented as mean ± SD of biological replicates. *p < 0.05, **p < 0.01, ***p < 0.001.

The overexpression of S100a9 and S100a8 by Ptpn11^E76K/+ mutant stem cells appeared to promote the growth of these mutant stem cells. Ptpn11^E76K/+ stem cells cultured in ex vivo expansion medium exhibited significantly accelerated proliferation compared to WT HSCs. However, this growth advantage was mitigated by tasquinimod, an inhibitor of S100a9/S100a8 that disrupts their interactions with the receptors Rage and TLR4^31,32 (Figure 7C), which were also highly expressed on these cells (Figure 2C). Additionally, the elevated capabilities of Ptpn11^E76K/+ mutant stem cells to form myeloid colonies compared to those of WT HSCs were substantially decreased by tasquinimod (Figure 7D). These findings suggest that S100a9 and S100a8 promote clonal expansion and increased myeloid cell production from Ptpn11-mutated stem cells through autocrine mechanisms, thereby counteracting the exhaustion of mutant stem cells caused by enhanced RAS signaling driven by the Ptpn11 mutation.

Previous studies have identified a significant role for S100a9 and S100a8 expressed in tumor cells in recruiting MDSCs, which are known for their association with immunosuppression and inflammation.^29,33,34 These heterogeneous cells co-express CD11b, Ly6G, and Ly6C myeloid lineage markers (polymorphonuclear MDSCs [PMN-MDSCs], CD11b⁺Ly6G⁺Ly6C^low; mononuclear MDSCs [M-MDSCs], CD11b⁺Ly6G⁻Ly6C^high). MDSCs are potent inhibitors of antitumor immunity, contributing to immune escape.^29,33,34 To investigate the potential interplay between Ptpn11^E76K/+ mutant stem cells and MDSCs, we conducted Transwell migration assays. As displayed in Figure 7E, Ptpn11^E76K/+ mutant stem cells demonstrated heightened chemoattracting activities for PMN-MDSCs (CD11b⁺Ly6G⁺) compared to WT HSCs. Notably, this effect was blocked by the S100a9/S100a8 inhibitor tasquinimod, indicating that the overproduction of S100a9 and S100a8 by Ptpn11 mutant stem cells may contribute to the recruitment of MDSCs to the stem cell microenvironment.

To test this possibility and further determine the role of S100a9 and S100a8 in the leukemogenic activities of Ptpn11^E76K/+ stem cells in an in vivo setting, we evaluated the therapeutic impact of the S100a9/S100a8 inhibitor tasquinimod using a widely used transplantation leukemia model. Ptpn11^E76K/+/Mx1-Cre/mTmG mice were generated by crossbreeding of Ptpn11^E76K/+/Mx1-Cre mice⁶ with lineage tracing mTmG transgenic mice,³⁵ which expressed red fluorescent protein but transitioned to green fluorescent protein (GFP) upon the induction of Cre expression (and the Ptpn11^E76K mutation) by polyinosinic:polycytidylic acid administration. To mimic clinical scenarios, we combined BM cells from Ptpn11^E76K/+/Mx1-Cre/mTmG mutant mice with WT BM cells from normal congenic BoyJ mice at a 10:1 ratio and transplanted mixed cells into lethally irradiated BoyJ mice. Four weeks post transplantation, when donor cells were engrafted, tasquinimod or vehicle was administered to mice via drinking water for 4 weeks (Figure 7F). Despite the high ratio of Ptpn11^E76K/+ mutant cells in the mixed donor cells, the hematopoietic reconstitution from Ptpn11^E76K/+ mutant stem cells in the recipient mice was approximately 50% due to the hyperactivation and significant depletion/exhaustion of the mutant stem cell population in the BM collected from mutant mice.⁶ Importantly, in response to tasquinimod treatment, a notable reduction in total mutant cells (GFP⁺) in the peripheral blood (PB) was observed (Figure 7G). Myeloid cells (Mac-1⁺) in the GFP⁺ cell compartment (Figure 7H) and the entire PB (Figure 7I) significantly decreased, indicating that the skewed myeloid differentiation of Ptpn11^E76K/+ mutant stem cells was largely rectified by blocking S100a9/S100a8 function.

Mice were euthanized after 4 weeks of treatment. White blood cell (WBC) counts in the tasquinimod-treated group significantly decreased, specifically in neutrophils and monocytes, with no apparent changes in red blood cell (RBC) counts (Figure 7J). Splenomegaly was also ameliorated in tasquinimod-treated mice (Figure 7K). The total number of mutant cells (GFP⁺) in the spleen as well as Mac-1⁺ myeloid cells in the GFP⁺ compartment and the entire spleen were all reduced (Figure 7L). Similar therapeutic effects were also observed in the BM (Figure 7M). Furthermore, we assessed the impact of the S100a9/S100a8 inhibitor on Ptpn11^E76K/+ mutant stem cells. As shown in Figure 7N, GFP⁺ Ptpn11^E76K/+ stem cells in the BM and early progenitor cells (LSK cells) in the spleen significantly decreased in the inhibitor-treated mice. Consistently, the cell cycling of hyperactive Ptpn11^E76K/+ stem cells was reduced by the treatment (Figure 7O). Moreover, apoptosis in these mutant stem cells increased in the inhibitor-treated mice (Figure 7P), suggesting that S100a9 and S100a8 played an important role in the survival of these mutant stem cells. Finally, we visualized Ptpn11^E76K/+ stem cells and surrounding cells in tasquinimod- or vehicle-treated mice and found that the distance between these mutant stem cells (CD150⁺CD11b⁻Ly6G⁻CD3⁻B220⁻Ter119⁻CD48⁻) (cyan) and the closest PMN-MDSCs (CD11b⁺Ly6G⁺) (yellow) was significantly increased following tasquinimod treatment (Figure 7Q), confirming that the recruitment of PMN-MDSCs to the microenvironment of Ptpn11 mutant stem cells was at least in part due to the overexpression of S100a9/S100a8 by these cells.

DISCUSSION

While considerable progress has been made in understanding the etiology of JMML, numerous questions remain, particularly concerning the cellular and molecular mechanisms that confer a selective advantage to the original leukemia-initiating cells. Understanding these mechanisms can illuminate how leukemia-initiating cells persist in established disease and how these tumor precursor cells may be effectively targeted and eliminated therapeutically. By undertaking a comprehensive characterization of the transcriptomic landscapes across all stages of tumor cell development in Ptpn11 mutation-associated JMML and substantiating our findings through experimental validation, we discovered that Ptpn11-mutated stem cells (leukemia-initiating cells) are primed by the myeloid transcriptional program and that innate immune and inflammatory responses are aberrantly activated in these cells. These mutant stem cells exhibit strikingly heightened expression of evolutionarily conserved genes that are typically activated in mature myeloid cells during pathogen defense, including antimicrobial peptides (Camp, Lcn2, Lyz2, Ltf, Chil3, and Pglyrp1) and essential trace metal-sequestering proteins (S100a9 and S100a8), which also function as pro-inflammatory proteins triggering and amplifying innate immune responses.^29,30

The innate immune system is conventionally activated through the recognition of pathogen-associated molecular pattern or damage-associated molecular pattern (DAMPs) proteins derived from host cells or damaged cells by pattern recognition receptors, including TLRs, on myeloid immune cells. These patterns play an important role in recruiting and activating myeloid immune cells, initiating inflammation to eliminate invading microorganisms.^29,30 S100a9 and S100a8, which show the most significant overexpression in Ptpn11^E76K/+ mutant stem cells, are also categorized as DAMPs. They preferentially heterodimerize to form calprotectin, which, like their monomeric/homodimeric forms, are endogenous ligands for TLR4, Rage/Ager, and CD33^24,36,37 on myeloid effector cells, activating intracellular signaling pathways and culminating in the production of inflammatory cytokines, chemokines, and antimicrobial peptides. Interestingly, the expression of Rage/Ager and CD33 is also markedly elevated on Ptpn11 mutant stem cells, producing autocrine effects. The autocrine effects of S100a9/S100a8 indeed contributed to the expansion of these mutant stem cells (Figure 7C). Moreover, given the well-characterized detrimental effects of inflammatory challenges on normal HSCs,^38,39 the pro-tumoral inflammatory milieu provides Ptpn11 mutant stem cells with a competitive advantage over normal counterparts, ultimately resulting in their clonal dominance.

S100a9 and S100a8 may also contribute to immune evasion of JMML-initiating stem cells by chemoattracting and expanding immunosuppressive MDSCs in the microenvironment. MDSCs are classically linked to immunosuppression, inflammation, and cancer, profoundly inhibiting T cell- and natural killer cell-mediated antitumor immunity through various mechanisms.^29,33,34 S100a9 is crucial for MDSC recruitment as MDSC accumulation in tumors is abolished in S100a9-null mice,⁴⁰ and expression of S100a9 in transgenic mice drives expansion and activation of MDSCs.³⁷ These immune-suppressive cells can also secrete abundant S100a9/S100a8 heterodimers, bind to their own surface receptors, and nurture an autocrine feedback loop that sustains MDSC recruitment, thereby maintaining immune suppression within the local microenvironment.²⁴ Moreover, S100a9 also contributes to antitumor immunity by inhibiting dendritic cell differentiation.⁴⁰ Ptpn11^E76K/+ mutant stem cells indeed demonstrate a strong ability to attract MDSCs, and this chemoattracting effect is diminished by the inhibitor of S100a9/S100a8 (Figures 7E and 7Q). Furthermore, administration of the S100a9/S100a8 inhibitor impedes leukemia development from Ptpn11^E76K/+ mutant stem cells (Figures 7F–7P). These results suggest that the overexpression of S100a9 and S100a8 by Ptpn11-mutated stem cells plays a pivotal role in the initial leukemogenic process in JMML.

Further investigations are necessary to elucidate how the Ptpn11^E76K mutation instigates a myeloid-specific transcriptional program and co-opts innate immune responses in the mutated stem cells. Shp-2 (encoded by Ptpn11) is predominantly localized to the cytosol and plays a prominent positive role in Ras signaling.^9,10 Since other genes that are mutated in JMML are also clustered in the Ras signaling pathway, it is conceivable that the Ptpn11 mutation causes pathogenic effects mainly through the Ras pathway. However, Shp-2 is also localized to the nucleus and the mitochondrion.^41–43 There is therefore a possibility that the Ptpn11^E76K mutation influences myeloid-specific transcriptomic activities through its nuclear and/or metabolic functions. The role of mutant Shp-2 in different cellular compartments may reveal novel avenues for understanding the diverse molecular mechanisms underpinning the aberrant activation of the myeloid transcriptional program in Ptpn11-mutated stem cells. Considering the distinctive subcellular localization of Shp-2 compared to other oncoproteins associated with JMML, it is important to ascertain whether dysregulated innate immune responses are also implicated in other JMML subtypes.

Another noteworthy finding of this study is the dysregulation of ribosomal biogenesis and function in Ptpn11^E76K/+ cells consistently throughout all stages, including mutant stem cells. Several ribosomal small and large subunit proteins displayed upregulation in Ptpn11^E76K/+ cells, consistent with the elevated protein translation essential for robust tumor cell growth. Intriguingly, there was a simultaneous decrease in the expression of certain ribosomal proteins. Recent research has revealed the heterogeneity of ribosomes, with different ribosome types displaying preferences for translating specific subsets of mRNAs.^26,27 Diminished expression of ribosomal proteins has the potential to disrupt ribosome formation and function. This can also contribute to malignancies through several mechanisms. The impairment in ribosomes can impact the synthesis of crucial regulatory proteins involved in cell growth, differentiation, and maturation, such as the tumor suppressor p53.^44–46 Moreover, reduced expression of specific ribosomal proteins and perturbed ribosome function can induce chronic ribosomal stress, triggering cellular dysfunction and genomic instability.²⁶ However, the precise mechanisms by which the Ptpn11 mutation selectively interferes with the expression of different ribosomal genes remain unclear.

In summary, our findings reveal previously unappreciated mechanisms in the initial phase of JMML leukemogenesis, where mutant stem cells exploit innate immune signaling to gain a selective advantage and evade antitumor immunity. The significant dysregulation of the proinflammatory proteins S100a9 and S100a8 in JMML-initiating cells underscores their pivotal role in orchestrating immune evasion and creating an inflammatory microenvironment conducive to leukemic progression. This insight offers valuable perspectives for developing therapeutic strategies to disrupt leukemic stem cells and improve treatment outcomes in JMML.

Limitations of the study

Despite efforts to enrich HSCs prior to RNA extraction and cDNA library preparation, the number of LT-HSCs captured in the Ptpn11-mutant sample was limited. Future studies should aim to improve the representation of LT-HSCs, especially if the focus is on this specific subset, as the majority of analyses in the current study were based on data from the broader HSC population. Additionally, the precise mechanisms by which the Ptpn11^E76K mutation activates a myeloid-specific transcriptional program and leads to elevated expression of innate immunity-associated proteins and peptides in mutant HSCs remain unclear. Finally, it will be important to investigate whether dysregulated innate immune responses in stem cells contribute to disease pathogenesis in other JMML subtypes as well.

RESOURCE AVAILABILITY

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Cheng-Kui Qu (cqu6@emory.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• The raw scRNA-seq data generated in this study have been deposited in the GEO database (https://www.ncbi.nlm.nih.gov/geo/) under accession number GSE266821.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

Ptpn11^{E76K Neo/+} conditional knock-in mice were generated in our previous study.⁶ Mx1-Cre⁺ (Strain #: 003556),⁴⁷ mTmG dual-fluorescent reporter transgenic mice (Strain #: 007676),³⁵ C57BL/6 mice (CD45.2⁺) (Strain #: 000664), and BoyJ mice (CD45.1⁺) (Strain #: 002014) were purchased from the Jackson Laboratory. All mice were kept under specific-pathogen-free conditions at Emory University Division of Animal Resources. Animal procedures complied with the NIH Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee.

Patient specimens

De-identified samples from PTPN11-mutated patients with JMML and pediatric healthy control’s normal BM biopsies were obtained from the University of California, San Francisco, and the Aflac Cancer and Blood Disorders Center Biorepository of Children’s Healthcare of Atlanta. Samples were obtained after written, informed consent under locally approved institutional review board research protocols and in accordance with the Declaration of Helsinki.

METHOD DETAILS

Single-cell transcriptome profiling

Fresh bone marrow (BM) cells were collected and pooled from Ptpn11^E76K/+/Mx1-Cre⁺ mice and Ptpn11^+/+/Mx1-Cre⁺ control littermates (3 mice/group) 8 weeks after polyinosinic:polycytidylic acid (pI-pC) administration and Ptpn11^E76K mutation induction, by which time a myeloproliferative disease had fully developed. To enrich for hematopoietic stem cells (HSCs), we sorted HSC-containing Lineage⁻Sca-1⁺c-Kit⁺ (LSK) cells from the BM using flow cytometry, and the sorted LSK cells were then added to BM cell samples at a proportion of 2%. The resulting cell mixture was subsequently processed according to the recommended protocol for the 10× Genomics Chromium Single Cell 3′ v3 scRNA-seq chemistry. In brief, scRNA-seq raw reads were obtained following the standard protocol for Chromium Single Cell 3ʹ Reagent Kits v3. Subsequently, the CellRanger 1 software from 10× Genomics was employed to identify cell-discriminating barcode sequence markers and unique molecular identifier (UMI) markers for different mRNA molecules within each cell. This process aimed to quantify the high-throughput single-cell transcriptome and conduct data quality statistics and comparisons against the original genome. Next, the Seurat 2 software package was utilized for further quality control (QC) and processing of the CellRanger results. In the QC step, delocalized cells were filtered by fitting a generalized linear model. Subsequently, the distribution of nUMI (unique molecular identifier counts), nGene (number of detected genes), and percent.mito (percentage of mitochondrial genes) was assessed to filter out low-quality cells, such as double cells, multiple cells, or dead cells, leaving only qualified 8,818 and 10,177 cells from WT and Ptpn11 mutant groups, respectively, for further bioinformatics analyses.

t-distributed stochastic neighbor embedding (t-SNE) visualization and cell identification

The single-cell transcriptome underwent principal component analysis (PCA) for linear dimensionality reduction. Subsequently, the PCA results were visualized in a two-dimensional space using t-SNE, a non-linear dimensionality reduction technique. The Seurat platform’s FindAllMarkers function was employed to identify marker genes for each cell classification relative to other cell populations. These identified genes serve as potential markers for each cell type. Visualization of the identified marker genes was carried out using the VlnPlot and FeaturePlot functions. Following the clustering process, the Single-R platform was utilized to assign cell types based on published datasets,^19,20 thereby enhancing the accuracy of cell type classification.

Gene set enrichment analysis (GSEA)

GSEA was conducted to identify gene sets enriched in specified cell types, such as HSCs (Hematopoietic Stem Cells) and GMPs (Granulocyte-Macrophage Progenitors). The analysis utilized the GSEA platform available at http://www.broadinstitute.org/gsea/index.jsp. To prepare input data for GSEA, the top 5000 variable genes in each group were selected using the Seurat “FindVariableGenes” function. Gene sets, including those from Gene Ontology (GO) or KEGG pathways, were obtained from the molecular signatures database (MSigDB).

Gene Set Variation Analysis (GSVA)

Gene Set Variation Analysis (GSVA) was performed to estimate the relative enrichment of predefined gene sets across individual samples. GSVA is a non-parametric, unsupervised method that transforms gene expression data from a gene-by-sample matrix into a gene set–by–sample matrix, allowing for the assessment of pathway or gene set activity within each sample independently. The analysis was conducted using the GSVA R package (version 1.44.0) with default parameters. Input expression data were normalized with log2 TPM. Enrichment scores were visualized and used for downstream comparisons between experimental groups.

Single-cell regulatory network inference and clustering (SCENIC) analysis

SCENIC analyses were performed using version 1.1.2.2, corresponding to RcisTarget 1.2.1 and AUCell 1.4.1. The motifs database for RcisTarget and GRNboost was utilized with default parameters. In detail, the analysis involved identifying over-represented transcription factor binding motifs on a given gene list using the RcisTarget package. Subsequently, the AUCell package was employed to score the activity of each group of regulons in each cell. This process enabled the inference and clustering of regulatory networks at the single-cell level, offering insights into the regulatory landscape of the analyzed cell populations.

To evaluate the cell type specificity of each predicted regulon, the regulon specificity score (RSS) was computed, employing the Jensen-Shannon divergence (JSD) as a measure of similarity between two probability distributions. Specifically, the JSD was calculated for each vector of binary regulon activity overlaps with the assignment of cells to specific cell types. The connection specificity index (CSI) for all regulons was determined using the scFunctions package, accessible at https://github.com/FloWuenne/scFunctions/.

Pseudotime analysis

Pseudotime analysis was performed using the Monocle 2 package (v2.9.0) to infer cell differentiation trajectories. First, the importCDS function was used to convert the Seurat object into a Monocle CellDataSet object. To select ordering genes, we applied the differentialGeneTest function, retaining genes with a q-value <0.01. Dynamically regulated genes were then grouped into modules using Monocle’s hierarchical clustering (hclust), which clusters genes based on shared expression dynamics along the pseudotime axis. Dimensionality reduction was conducted using the reduceDimension function, and cell trajectories were reconstructed with the orderCells function.

Cell-cell communication analysis

We utilized CellPhoneDB (v2.0) to identify biologically relevant ligand-receptor interactions from single-cell transcriptomic data. We defined a ligand or receptor as ‘expressed’ in a particular cell type if 10% of the cells of that type exhibited non-zero read counts for the ligand/receptor encoding gene. Statistical significance was assessed by randomly shuffling the cluster labels of all cells and repeating the aforementioned steps, thereby generating a null distribution for each ligand-receptor (LR) pair in each pairwise comparison between two cell types. Following 1,000 permutations, p-values were calculated using the normal distribution curve generated from the permuted LR pair interaction scores. To delineate networks of cell-cell communication, we connected any two cell types where the ligand was expressed in the former cell type and the receptor in the latter. The R package circlize was employed for visualizing the cell-cell communication networks.

Fluorescence-activated cell sorting (FACS) analysis and cell sorting

FACS analyses were performed on a Cytoflex flow cytometer (Beckman Coulter Life Sciences), following standard procedures. For HSC staining, BM cells were harvested, washed, and incubated for 30 min at 4°C in phosphate-buffered saline (PBS) with 2% fetal bovine serum (FBS) containing the following antibodies: anti-Mac-1 PerCP/Cyanine5.5 (Biolegend, 101228, clone M1/70),anti-Gr-1 Pacific Blue (Biolegend, 108430, clone RB6–8C5), anti-Ter119 PE (Biolegend, 116208, clone TER-119), anti-B220 PE (eBiosciences, 12–0452-83, clone RA3–6B2), anti-CD3 PE (BD Biosiences Pharmingen, 553064, clone 145–2C11), anti-Mac-1 PE (Biolegend, 101208, clone M1/70), anti-Gr-1 PE (eBiosciences, 12–5931-83, clone RB6–8C5), anti-Scal-1 PE/Cyanine7 (Biolegend, 108114, clone D7), anti-c-Kit APC/Cyanine7 (Biolegend, 105826, clone 2B8), anti-CD48 Percp (eBioscience, 46–0481-80, clone HM48–1), anti-CD150 AF647 (Biolegend, 115918, clone TC15–12F12.2). HSCs were defined as Lineage⁻Sca-1⁺c-Kit⁺CD150⁺CD48⁻. For apoptosis analyses, fresh BM cells were stained for HSCs, and then incubated with Annexin V-BV605 (BD Biosiences Pharmingen, 563974, clone Annexin V) (0.7 μg/mL) and 4′,6-diamidino-2-phenylindole (DAPI) (0.3 μg/mL). For the cell cycle analysis, fresh BM cells were stained for HSCs as above, fixed and permeabilized using a Cytofix/Cytoperm kit (BD Biosciences). The samples were then stained with Ki-67 BV605 (Biolegend, 652413) and further incubated with Hoechest 33342 (20 μg/mL). Data were collected on a Beckman Coulter CytoFLEX flow cytometer and analyzed with FlowJo (Tree Star). For cell sorting, BM cells were first lineage-depleted using a lineage depletion kit. Cells were then stained with fluorochrome-labeled antibodies. Sorting of specific cell populations was conducted using BD FACSAia II following standard gating strategies.

Colony-forming unit (CFU) assay

Freshly sorted HSCs (5 × 10² cells) were plated in triplicate in 35-mm dishes in 0.9% methylcellulose IMDM medium containing 15% FBS, Gln (10⁻⁴ M), β-mercaptoethanol (3.3 × 10⁻⁵ M), SCF (50 ng/mL), IL-3 (20 ng/mL), IL-6 (20 ng/mL), and EPO (3 Units/mL). After 12 days of incubation at 37°C in 5% CO2, myeloid colonies derived were counted under an inverted microscope.

Transmigration assay

Transmigration assays were conducted with 5.0 μm pore transwells (Corning). Briefly, HSCs freshly sorted from WT and Ptpn11^E76K/+ mice were suspended in StemSpan media (STEMCELL Technologies) containing 20% FBS, 50 ng/mL SCF, 50 ng/mL Flt3L, 20 ng/mL IL-3, and 20 ng/mL IL-6. Six hundred microliters of cell suspension (2 10³ cells) were loaded into lower chambers. The S100a9/S100a8 inhibitor tasquinimod was then added to the chamber (5.0 μM). CD11b⁺Ly6G⁺ myeloid cells freshly sorted from normal C57BL6 mice were labeled with carboxyfluorescein succinimidyl ester (CFSE) (1.0 μM), washed, and resuspended at 1 × 10⁶ cells/mL in the same medium as that in lower chambers but without the inhibitor. One hundred microliters of cell suspension were added to upper chambers. Cells were allowed to migrate across the membrane at 37°C in 5% CO₂ for 2 h. Both input cells, cells collected from the upper chamber, and cells collected from the lower chambers were analyzed by FACS. Migration efficiency was then calculated.

Immunofluorescence staining

Tissue sections were prepared from paraffin-embedded mouse femurs, deparaffinized, and rehydrated following standard protocols. The slides were stained with the following antibodies following standard procedures: anti-CD150 AF647 (Biolegend, 115918, clone TC15–12F12.2), anti-CD11b PE (Biolegend, 101208, clone M1/70), anti-Ly-6G AF488 (Biolegend, 127625, clone 1A8), anti-Ter119 FITC (Biolegend, 116206, clone TER-119), anti-CD3 FITC (Biolegend, 100306, clone 145–2C11), anti-B220 FITC (Biolegend, 103206, clone RA3–6B2), and anti-CD48 FITC (Biolegend, 103403, clone HM48–1) antibodies. Images were acquired using Leica Stellaris 8 and processed with ImageJ 1.54f software.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis

All statistical analyses were performed using GraphPad Prism. An unpaired two-tailed Student’s t-test was used to compare two groups. Data are presented as mean ± standard deviation (SD) of biological replicates (independent animals or independent experiments). Sample sizes (n) are indicated in the graphs and detailed in the figure legends. Statistical significance is denoted as follows: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), and p < 0.0001 (****).

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116087.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
PerCP/Cyanine5.5-Mac-1	Biolegend	Cat# 101228; PRID: AB_893232
Pacific Blue-Gr-1	Biolegend	Cat# 108430; PRID: AB_893556
PE-Ter119	Biolegend	Cat# 116208; PRID: AB_313708
PE-B220	eBioscience	Cat# 12-0452-83; PRID: AB_465672
PE-CD3	BD Pharmingen	Cat# 553064; PRID: AB_394597
PE-Mac-1	Biolegend	Cat# 101208; PRID: AB_312790
PE-Gr-1	eBioscience	Cat# 12-5931-83; PRID: AB_466046
PE/Cyanine7-Sca-1	Biolegend	Cat# 108114; PRID: AB_493596
APC/Cyanine7-c-Kit	Biolegend	Cat# 105826; PRID: AB_1626278
Percp-CD48	eBioscience	Cat# 46-0481-80; PRID: AB_10853483
AF647-CD150	Biolegend	Cat# 115918; PRID: AB_2239178
BV605-Annexin V	BD Horizon	Cat# 563974; PRID: AB_2869539
BV605-Ki-67	Biolegend	Cat# 652413; PRID: AB_2562664
AF647-CD150	Biolegend	Cat# 115918; PRID: AB_2239178
PE-CD11b	Biolegend	Cat# 101207; PRID: AB_312790
AF488 Ly-6G	Biolegend	Cat# 127625; PRID: AB_2561339
FITC-Ter119	Biolegend	Cat# 116205; PRID: AB_313706
FITC-CD3	Biolegend	Cat# 100306; PRID: AB_312670
FITC-B220	Biolegend	Cat# 103206; PRID: AB_312990
FITC-CD48	Biolegend	Cat# 103403; PRID: AB_313018
Chemicals, peptides, and recombinant proteins
Fetal bovine serum (FBS)	Thermo Fisher Scientific	Cat# SH3008803IH
DAPI	Santa Cruz Biotechnology	Cat# 28718-90-3
Hoechst 33342	Thermo Fisher Scientific	Cat# R37165
polyinosinic:polycytidylic acid (pI-pC)	Invivogen	Cat# tlrl-pic
IMDM	Thermo Fisher Scientific	Cat# 12440053
Glutamine	Coring	Cat# 25-005
β-mercaptoethanol	Sigma-Aldrich	Cat# M6250
Mouse stem cell factor (SCF) Recombinant Protein, PeproTech	Thermo Fisher Scientific	Cat# 250-03
Mouse IL-3 Recombinant Protein, PeproTech	Thermo Fisher Scientific	Cat# 213-13
Mouse IL-6 Recombinant Protein, PeproTech	Thermo Fisher Scientific	Cat# 216-16
Erythropoietin (EPO)	PeproTech	Cat# 100-64
StemSpan™ SFEM	STEMCELL Technologies	Cat#09600
Mouse FLT3L Recombinant Protein	Thermo Fisher Scientific	Cat# 250-31L
5.0 μm Pore Transwell	Coring	Cat# 3421
Tasquinimod	Sigma-Aldrich	Cat# SML2489
CFSE	Biolegend	Cat# 79898
Critical commercial assays
BD Cytofix/Cytoperm	BD Biosciences	Cat# 554714; PRID: AB_2869008
Lineage Cell Depletion Kit, mouse	Miltenyi Biotec	Cat# 130-090-858
Quick-RNA Miniprep Kit	Zymo Research	Cat# R1054
SuperScript First-Strand Synthesis System for RT-PCR	Invitrogen	Cat# 11904018
SYBR™ Green Master Mix for qPCR	Thermo Fisher Scientific	Cat# A25741
Experimental models: Organisms/strains
Mx1-Cre+	Jackson Laboratory	Strain #: 003556
Ptpn11E76K Neo/+	Generated in our laboratory	N/A
mTmG dual-fluorescent reporter transgenic mice	Jackson Laboratory	Strain #: 007676
C57BL/6	Jackson Laboratory	Strain #: 000664
BoyJ	Jackson Laboratory	Strain #: 002014
Ptpn11E76K/+/Mx1-Cre+	Generated in this study	N/A
Oligonucleotides
Forward primer for human S100A4 cDNA: 5′-CAGAACTAAAGGAGCTGCTGACC-3′	Synthesized in this study	N/A
Reverse primer for human S100A4 cDNA: 5′-CTTGGAAGTCCACCTCGTTGTC-3′	Synthesized in this study	N/A
Forward primer for human S100A6 cDNA: 5′-GTGACAAGCACACCCTGAGCAA-3′	Synthesized in this study	N/A
Reverse primer for human S100A6 cDNA: 5′-GGAAGTTCACCTCCTGGTCCTT-3′	Synthesized in this study	N/A
Forward primer for human S100A8 cDNA: 5′-ATGCCGTCTACAGGGATGACCT-3′	Synthesized in this study	N/A
Reverse primer for human S100A8 cDNA: 5′-AGAATGAGGAACTCCTGGAAGTTA-3′	Synthesized in this study	N/A
Forward primer for human S100A9 cDNA: 5′-GCACCCAGACACCCTGAACCA-3′	Synthesized in this study	N/A
Reverse primer for human S100A9 cDNA: 5′-TGTGTCCAGGTCCTCCATGATG-3′	Synthesized in this study	N/A
Forward primer for human S100A10 cDNA: 5′-AACAAAGGAGGACCTGAGAGTAC-3′	Synthesized in this study	N/A
Reverse primer for human S100A10 cDNA: 5′-CTTTGCCATCTCTACACTGGTCC-3′	Synthesized in this study	N/A
Forward primer for human S100A11 cDNA: 5′-CCAGAAGTATGCTGGAAAGGATG-3′	Synthesized in this study	N/A
Reverse primer for human S100A11 cDNA: 5′-CATCATGCGGTCAAGGACACCA-3′	Synthesized in this study	N/A
Forward primer for mouse S100A4 cDNA: 5′-AGCTCAAGGAGCTACTGACCAG-3′	Synthesized in this study	N/A
Reverse primer for mouse S100A4 cDNA: 5′-GCTGTCCAAGTTGCTCATCACC-3′	Synthesized in this study	N/A
Forward primer for mouse S100A6 cDNA: 5′-GCAAGGAAGGTGACAAGCACAC-3′	Synthesized in this study	N/A
Reverse primer for mouse S100A6 cDNA: 5′-CCTGATCCTTGTTACGGTCCAG-3′	Synthesized in this study	N/A
Forward primer for mouse S100A8 cDNA: 5′-CAAGGAAATCACCATGCCCTCTA-3′	Synthesized in this study	N/A
Reverse primer for mouse S100A8 cDNA: 5′-ACCATCGCAAGGAACTCCTCGA-3′	Synthesized in this study	N/A
Forward primer for mouse S100A9 cDNA: 5′-TGGTGGAAGCACAGTTGGCAAC-3′	Synthesized in this study	N/A
Reverse primer for mouse S100A9 cDNA: 5′-CAGCATCATACACTCCTCAAAGC-3′	Synthesized in this study	N/A
Forward primer for mouse S100A10 cDNA: 5′-GACAAAGGAGGACCTGAGAGTG-3′	Synthesized in this study	N/A
Reverse primer for mouse S100A10 cDNA: 5′-CTCTGGAAGCCCACTTTGCCAT-3′	Synthesized in this study	N/A
Forward primer for mouse S100A11 cDNA: 5′-GAAGGATGGAAACAACACTCAACT-3′	Synthesized in this study	N/A
Reverse primer for mouse S100A11 cDNA: 5′-CGTCACAGTTGAGGTCCAGCTT-3′	Synthesized in this study	N/A
Software and algorithms
scRNA-seq 10× Genomics platform	https://www.10xgenomics.com	Chromium Single Cell 3ʹ Reagent Kits v3
CellRanger 1 software	https://www.10xgenomics.com	N/A
Seurat platform’s FindAllMarkers	https://satijalab.org/seurat/	N/A
Single R platform	https://github.com/dviraran/SingleR	N/A
Gene set enrichment analysis (GSEA)	http://www.broadinstitute.org/gsea/index.jsp	N/A
SCENIC analysis	https://github.com/aertslab/SCENIC	version 1.1.2.2
scFunctions package	https://github.com/FloWuenne/scFunctions/	N/A
Monocle2 package (v2.9.0)	https://github.com/cole-trapnell-lab/monocle-release	N/A
CellPhoneDB (v2.0)	https://github.com/Teichlab/cellphonedb	N/A
Flowjo v10.8.1	https://www.flowjo.com/	N/A
ImageJ 1.54f	https://imagej.net/software/fiji	N/A
Raw scRNA-seq data	The Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/)	Accession number GSE266821
GraphPad Prism 10.2.2	Graphpad	N/A

Highlights.

• Ptpn11 mutated stem cells exhibit de novo activation of the myeloid transcriptional program

• Ptpn11 mutant stem cells display elevated expression of innate immunity-associated proteins

• S100a9/S100a8 provide selective advantages to mutated stem cells and promote immune evasion

• Inhibition of S100a9/S100a8 signaling suppresses leukemia development from mutant stem cells