Two distinct fetal-type signatures characterize juvenile myelomonocytic leukemia

Abstract

Juvenile myelomonocytic leukemia (JMML) is an aggressive clonal myeloproliferative neoplasm that affects infants and young children. The narrow window of onset suggests that age-related factors are involved in leukemogenesis. To investigate whether ontogeny-related features are involved in JMML oncogenesis, we compared the gene expression profile of hematopoietic progenitor cells isolated from JMML patients with that of healthy individuals at different stages of ontogeny. This analysis identified two main groups of JMML patients. In the first group, JMML progenitors exhibited a gene expression profile similar to that of embryo-fetal progenitors. Progenitors showed a strong monocytic identity as evidenced by the overexpression of monocytic/dendritic, inflammasome, and innate immune markers. This resembled the monocyte-predominant myelopoiesis characteristic of normal fetal hematopoiesis. However, in the second group, despite evidence of developmental dysregulation as indicated by the aberrant signature of the master oncofetal regulator LIN28B, JMML clustered separately from healthy prenatal and postnatal fractions. These findings highlight the intricate relationship between JMML and development, which will help inform future therapeutic approaches for this rare but severe form of leukemia.

Figure 1

Gene expression profile of JMML hematopoietic progenitors versus healthy prenatal and postnatal counterparts. (A) Unsupervised hierarchical clustering of sorted JMML hematopoietic progenitors (CMP, GMP, MEP) and their pre‐ or postnatal counterparts according to gene expression profile. Four clusters (C1 to C4) were defined. C1 contains FL samples (n = 9/9), FBM samples (5/6), and 16/47 JMML samples (CMP n = 5, GMP n = 3, MEP n = 8) from 8/16 patients. C2 contains healthy BM, 7/7 samples (CMP n = 3/3, MEP n = 4/4), and 2 JMML samples (#12). C3 contains GMP from 4/4 healthy BM, 1/4 FBM, JMML GMP (5/16 samples), and 1 JMML CMP. C4 contains 23/47 JMML samples (CMP n = 8, GMP n = 8, and MEP n = 7) from 10/16 patients and no healthy fetal or postnatal tissue (see also Supporting Information S1: Table 5). BM, postnatal bone marrow; FBM, fetal bone marrow; FL, fetal liver; WBC, white blood cell count. (B) Patients' age and hematological features by JMML group (see also Supporting Information S1: Tables 1 and 2).

Figure 2

(A) Vulcano plot showing differentially expressed genes upregulated in JMML_F (left) or JMML_L (right) according to log2 fold change (x‐axis) and q‐value (y‐axis). Differential gene expression analysis between JMML groups evidenced 1052 upregulated genes with a fold change higher than 1.5 and a q‐value lower than 0.1 in JMML_F versus 230 upregulated genes in JMML_L (listed in Table S3). Among genes most up‐regulated in JMML_F, MEFV, TBC1D9, NLRP12, and SCIMP code proteins involved in the pyrin inflammasome whereas CD14 and CD300E code monocytic markers. (B–C) Gene expression profiling in the JMML_F group. (B) GSEA plots show enrichment in monocyte, dendritic cells (TOP DC), and inflammasome signatures in the JMML_F group versus JMML_L (left) or versus healthy BM (right) (see also Table S4). (C) Histograms comparing gene expression (expressed as mean FPKM scores ±SD) of healthy samples across ontogeny (FL, FBM, BM), JMML_F and JMML_L (left panel). and correlation matrix of the 20 most upregulated genes in the JMML_F group (right panel). (D, E) Gene expression profiling in the JMML_L group. (D) GSEA plots show enrichment in LIN28B and WT1 signatures in JMML_L versus JMML_F. Histograms show expression levels (FPKM) of LIN28B (left) and WT1 (right) across ontogeny (FL, FBM, BM), JMML_F and JMML_L groups. (C) Histograms comparing gene expression of LIN28B transcriptional targets (expressed as mean FPKM scores ± SD) of healthy samples across ontogeny (FL, FBM, BM), JMML_F and JMML_L. BM, postnatal bone marrow; DC, dendritic cell; FBM, fetal bone marrow; FDR, false discovery rate; FL, fetal liver; FPKM, fragments per kilobase million; NES, normalized enrichment score; SD, standard deviation.

Figure 3

DNA methylation analyses. (A) Principal component analysis (PCA) of DNA methylation data obtained by RRBS for mononucleated cells from JMML samples (n = 16), fetal bone marrow (FBM; n = 2), and healthy postnatal bone marrow (BM; n = 2). The DNA methylation study distinguishes three groups of JMML according to low (Meth^low, n = 8), intermediate (n = 1), or high (Meth^high, n = 6) levels of methylation. JMML overexpressing LIN28B are all Meth^high (medium panel). Right panel shows a relationship between methylation status and transcriptional‐based groups. (B) LIN28B quantitation normalized by TPB expression using droplet digital PCR (ddPCR) shows that LIN28B expression is correlated with DNA methylation in JMML but not in healthy controls. (C) GSEA plot shows enrichment in the LIN28B signatures in Meth^high JMML versus Meth^low JMML. (D) Vulcano plot showing differentially expressed genes upregulated in Meth^low JMML (left) or Meth^high (right) according to log2 fold change (x axis) and q‐value (y axis). BM, postnatal bone marrow; DC, dendritic cell; FBM, fetal bone marrow; FDR, false discovery rate; FL, fetal liver; FNES, normalized enrichment score; FPKM, fragments per kilobase million; SD, standard deviation.