Abstract:The human genome contains 90 genes that encode either receptor or non-receptor tyrosine kinases, most of which are historically oncogenes activated by mutations, amplifications, in-frame indels, or rearrangements.1 Rearrangement of tyrosine kinases, which are often fused with a dimerization domain-containing fusion partner, is a potent mechanism for constitutive kinase activation that causes uncontrolled cell proliferation and is found in a variety of tumors. However, the rearrangement of the vascular endothelial growth factor receptor (VEGFR) family has never been reported in cancer, which may indicate that the obstruction of neoangiogenesis, which is critical for tumor development and requires VEGFRs' involvement, poses a significant obstacle for tumor cells to overcome. Here, we describe a de novo acute myeloid leukemia (AML) case in which a translocation between chromosome 3q12 and chromosome 13q12 leads to a gene fusion between FLT1 (also known as VEGFR1) and CMSS1 (Cms1 ribosomal small subunit homolog). The resulting chimeric protein contains an N-terminal fragment from CMSS1 and the intact tyrosine kinase domain of FLT1. CMSS1::FLT1 is cytoplasmic, constitutively activated, and promotes cell proliferation. In vitro cell models with forced expression of CMSS1::FLT1 were sensitive to an FLT1 kinase inhibitor Axitinib. A 79-year-old female patient presented with decreased WBC. Physical examination revealed no bruising or organomegaly. The complete blood counts showed WBC 1.8 × 109/L, Hb 105 g/L, MCV 109.2 fL, PLT 213 × 109/L, 3% blasts, 38.6% neutrophils, 47.5% lymphocytes, 13.3% monocytes, 0.6% eosinophils, and 0% basophils. Bone marrow aspirate smear revealed 21% of variably sized blasts with cytoplasmic pseudopod formation and 1–3 nucleoli. The myeloid elements were decreased and exhibited left-shift maturation, with some cells containing increased cytoplasmic granules. Erythroblasts were significantly increased, with some binucleated erythroblasts. Occasional erythrocytes with Howell–Jolly bodies were observed (Figure 1A). The myeloid to erythroid ratio was 0.35:1. Cytochemical analysis of blast cells was negative for myeloperoxidase (MPO) and periodic acid–Schiff (PAS). Bone marrow flow cytometry revealed 28% of CD34+ cells that were CD13+, CD33+, CD34+, CD38dim, CD117+, CD200+, HLADRdim, CD2−, cyCD3−, CD5−, CD7−, CD11b−, CD14−, CD16−, CD19−, CD56−, cyCD79−, and MPO−. The diagnosis of AML was made. Cytogenetic analysis showed a complex rearrangement between chromosome 3 and 13 as the sole change (46,XX,der(3)t(3;13)(q12;q12)inv(3)(p25q11.2),der(13)t(3;13)[10]/46,XX[10]) (Figure 1B). A targeted DNA-NGS assay of 128 leukemia-related genes identified BCORL1 p.A427Dfs17 (VAF 10.6%) and ETV6 p.R103Sfs9 (10.4%). CNV analysis revealed no apparent chromosome gain or loss. A targeted RNA-NGS assay with bait probes covering 81 leukemia-related genes did not find any fusion transcript. After obtaining the patient's informed consent and approval from our local Institutional Review Board (IRB), we performed a genome-wide mate-pair DNA sequencing, which revealed DNA fusion fragments consistent with the chromosome breakpoints seen by karyotype analysis, containing intron 1 of CMSS1 at 3q12 and intron 15 of FLT1 at 13q12 (Figure 1C). The expression of the CMSS1::FLT1 fusion transcript was confirmed by a RT-PCR assay with primers specific to CMSS1 (CMSS1_F1: TACCCGTGATGTTCTGC; CMSS1_F2 nest: TCGAGACCTGAGCTGAAA) and FLT1 (FLT1_R1: ACTTGCTGGCATCATAAG; FLT1_R2 nest: GAAGACCTTTTCATTTTTCGG), which showed two distinct bands in an agarose gel electrophoresis (Figure 1D). Sanger sequencing of the major band showed an in-frame fusion between CMSS1 exon 1 and FLT1 exon 16. The predicted chimeric protein contains an N-terminal fragment from CMSS1 and the entire tyrosine kinase domain from FLT1. The minor band consisted of CMSS1 exon 1, a 50 bp CMSS1 intronic sequence from intron 1, and FLT1 exon 16 (Figure 1E), which is likely from differential splicing because a classic acceptor splicing signal “AG” and a donor splicing signaling “GT” were observed at the 5′ and 3′ end of the inserted intronic fragment, respectively. The minor transcript is likely nonfunctional due to a shifted reading frame. A lentivirus carrying an MYC-tagged CMSS1::FLT1 was introduced into NIH3T3 cells. The subcellular localization of CMSS1::FLT1 was determined using immunofluorescence staining with an anti-MYC antibody, which revealed predominant cytoplasmic localization (Figure 1F). An immunoprecipitation of CMSS1::FLT1 was performed with MYC antibody, which showed a band corresponding to the expected size of CMSS1::FLT1 (72 Kd). Upon stripping and incubating with an anti-phospho-tyrosine antibody, the CMSS1::FLT1 was tyrosine phosphorylated (Figure 1G). In addition, MAPK, a known downstream target of FLT1 activation, is also phosphorylated (Figure 1H). These results are consistent with a ligand-independent kinase activation of the CMSS1::FLT1. CMSS1::FLT1-expressing cells showed significantly faster growth compared to cells expressing an empty vector, as determined by the Cell Counting Kit-8 (CCK-8) assay (Figure 1I). In contrast, cells expressing CMSS1::FLT1 were significantly more sensitive to treatment with Axitinib, an FLT1/VEGFR1 kinase inhibitor (Figure 1J). The VEGFR family comprises three members: FLT1 or VEGFR1, kinase insert domain receptor (KDR) or VEGFR2, and FLT4 or VEGFR3. These tyrosine kinase receptors possess seven immunoglobulin (Ig)-like extracellular domains, a transmembrane domain, and an intracellular split tyrosine kinase domain. Both FLT1 and KDR are predominantly expressed in vascular endothelial cells, while FLT1 is also expressed in monocytes and macrophages.2 FLT4 is mainly expressed in lymphatic endothelial cells. During embryonic development, FLT1 and KDR are required for angiogenesis and vasculogenesis, with KDR being the main signaling transducer and FLT1 a major regulator of VEGFA (ligand) levels. FLT4 plays a crucial role in lymphatic vessel development. Recent studies indicate that all three VEGFRs are important for neoangiogenesis in tumors. VEGFR amplification, activation mutation, and overexpression have been observed in various tumors, including lung adenocarcinoma, colon adenocarcinoma, melanoma, glioma, and endometrial endometrioid adenocarcinoma.3 FLT1 has been implicated in tumor metastasis by promoting the recruitment and activation of macrophages in the tumor microenvironment. FLT1 activation on macrophages can result in the secretion of cytokines and growth factors that promote tumor growth and invasion, as well as the recruitment of additional macrophages to the tumor site. FLT1 signaling on tumor cells can also contribute to metastasis by promoting the development of a pre-metastatic niche and enhancing tumor cell migration and invasion. In a mouse model expressing a deficient FLT1 with its tyrosine kinase domain deleted, highly metastatic 3LL-LLC lung cancer cells failed to produce lung metastasis. Macrophages with a tyrosine kinase-deficient FLT1 lost the capability of cell migration when induced by VEGF.4 While FLT1 is well known for its role in angiogenesis and tumor metastasis, the function of CMSS1 is unclear. RNA-Seq expression data from GTEx showed a ubiquitous expression pattern of CMSS1 in 53 different human tissues. Immunofluorescence staining with a CMSS1-specific antibody showed bright nucleoli staining in A-431, U2-OS, and U-251 MG cells (The Human Protein Atlas), consistent with an RNA-binding protein. RNA-binding proteins are important for hematopoiesis. Notably, the expression levels of 6 RNA-binding proteins, including CMSS1, reliably distinguished low-risk from high-risk patients with diffuse large B-cell lymphoma (DLBCL).5 Since dimerization is an important feature of RNA-binding proteins, it is likely that CMSS1 provides a dimerization motif for the CMSS1::FLT1 fusion protein, resulting in a ligand-independent dimerization and constitutive kinase activation of FLT1. The cellular signaling of FLT1 is not well-defined yet, however, PLCγ-PKC-MAPK signaling is required for VEGFA/KDR-induced endothelial proliferation. KDR 1175Y is phosphorylated when activated, which serves as a docking site for PLCγ. The PLC-γ then stimulates hydrolysis of phosphatidylinositol (PIP2), leading to protein kinase C (PKC) activation and subsequent raf-1/MAPK activation. Since FLT1 shares the same KDR 1175Y motif, a similar PLCγ-PKC-MAPK signaling likely contributes to the oncogenesis of CMSS1::FLT1.6 The activation of tyrosine kinase receptors is an important pathway implicated in the oncogenesis of myeloid neoplasms. Specifically, FLT3 is altered in roughly a quarter of AML cases. In myeloid/lymphoid neoplasms characterized by eosinophilia, rearrangements involve PDGFRA/B, FGFR1, JAK2, FLT3, and ABL1. Notably, tyrosine kinase inhibitors can successfully treat patients with these genetic anomalies. Our studies have shown that cells expressing CMSS1::FLT1 are sensitive to the tyrosine kinase inhibitor Axitinib, which targets VEGFRs and PDGFRs and is FDA-approved for advanced renal cell carcinoma. Additionally, inhibitors that target the PLCγPKC-MAPK signaling pathway, such as the PKC inhibitor Midostaurin and MAPK inhibitor Ulixertinib, may also be helpful. In summary, we reported the first documented case of cancer with VEGFR rearrangement, resulting in the activation of FLT1 signaling. Notably, these tumor cells are sensitive to the FLT1 inhibitor. Although this fusion is rare, it provides a unique opportunity to explore the downstream FLT1 signaling pathway, which plays a crucial role in angiogenesis and tumor metastasis. Xiaoshan Yang, Lingfeng Liu, and Xiaojun Chen performed experiments. Liying Zhang, Bingzong Li, and Yu Sun provided CNL patient samples and clinical data. Yu Sun and Sheng Xiao analyzed the data. Sheng Xiao, Hong Zhang, and Jun Li commented on the paper. Xiaoshan Yang and Sheng Xiao wrote the paper. Yu Sun and Sheng Xiao designed the research. All authors contributed to writing the paper by providing guidance and comments on its content. The authors declare no conflict of interest. The data that support the findings of this study are available from the correspondingauthor upon reasonable request.

Abstract:Acute myeloid leukaemia (AML) is a group of heterogeneous haematological malignancies caused by various genetic abnormalities, including chromosomal translocations and/or somatic mutations. Genomic sequencing efforts in AML, such as The Cancer Genome Atlas1 (TCGA) and the Beat AML programme,2 have identified an intensive catalogue of somatic mutations and gene fusions contributing to AML development. However, these sequencing-based investigations mainly pay attention to nuclear DNAs. Increasing evidence indicates that the altered energy and metabolic dependency of leukaemia cells is an important mechanism of AML pathogenesis.3, 4 Mitochondrial genomic alterations play critical roles in the reprogramming of mitochondrial metabolism.5, 6 However, the systematic understanding of somatic mitochondrial mutations and their interplay with the nuclear genome in AML remains poorly understood. In this study, we characterized somatic mitochondrial mutations and their interaction with nuclear driver mutations using whole-exome sequencing (WES) and/or RNA-seq data of 375 AML patients from the Beat AML (n = 175) and TCGA (n = 200) cohorts. We first applied an integrated approach (details in the Supporting Information) to call somatic mitochondrial DNA (mtDNA) mutations in 175 AML patients from the Beat AML cohort (mean coverage, ~346× in AML, ~414× in matched controls). A total of 134 somatic mtDNA mutations were identified, which were located in 130 mitochondrial nucleotide positions. A proportion of approximately 61% were located in protein-coding regions, followed by approximately 17% in two ribosomal RNA genes, approximately 14% in D-loop regions and approximately 8% in tRNA genes (Figure 1A). As previously reported, all 134 somatic mutations were heteroplasmic and most of them (~86%) had low variant allele frequency (VAF, <0.2). Notably, there were two heteroplasmic mutations (one in MT-CO3, missense mutation; the another in the D-loop) with a VAF of 0.9 or higher, suggesting that these two mutations might alter the cellular functions of AML cells due to the heteroplasmy threshold effect of mitochondria. In addition, the mutational signatures of AML mtDNAs were enriched in transition with A:T > G:C (~46%) and G:C > A:T (~46%) (Figure 1C). Among 82 somatic mtDNA mutations in protein-coding regions, about 68% were missense variants, suggesting that more than two-thirds of mutations could result in the alteration of coded amino acids. The top three mutated genes were MT-ND4, MT-ND5 and MT-CYTB, respectively (Figure 1D). Mutations of these three mitochondrial genes have also been found in solid cancers, such as prostate and lung cancer. Our results provided evidence that somatic mtDNA mutations might contribute to the pathogenesis of AML, consistent with the previous observation that mitochondrial Electron Transport Chain (ETC) gene mutations are negatively associated with clinical outcomes in AML.7 To understand the characterization of the transmission of mitochondrial mutations from DNA to the RNA level, we further used corresponding RNA-seq data to identify the mtDNA mutations at the mRNA level based on 51 AML patients from the Beat AML database with WES data. There were 84 mutations at mtDNAs found in 51 AML patients. Among these 84 mtDNA mutations, approximately 73% (61) of mutations were also found at the mitochondrial RNA level and the frequencies of the majority (~82%, 50 of 61) were consistent between mitochondrial DNA and RNA with less than 5% frequency difference (Figure 1E). The allele frequencies of most (~83% of 24) remaining heteroplasmic mutations that were lost at the mitochondrial RNA level were 0.1 or less, and there were only two mtDNA variants with an allele frequency of 0.25 or above, supporting that heteroplasmic DNA mutations with high allele frequency can be transcribed into mitochondrial RNA with high fidelity. Together, these results indicated the congruence of mutations between mitochondrial DNA and RNA in AML. Next, we performed an integrated analysis to investigate the interplay between mitochondrial and nuclear driver mutations/alterations in AML. As expected, somatic mitochondrial mutations were mutually exclusive with mutations in IDH1 (p < 0.05, Fisher's exact test; Beat AML cohort) (Figure 1F; see Table S1 for the contingency table), which has been reported to contribute to an enhanced mitochondrial oxidative metabolism.8 Notably, we also found that somatic mitochondrial mutations were mutually exclusive to the CBFB::MYH11 fusion gene in the Beat AML cohort (p < 0.05, Fisher's exact test) (Figure 1F; see Table S2 for the contingency table). To validate this finding, we further extracted the genomic data of nuclear and mitochondrial alterations from the TCGA cohort (n = 200). Consistent with the findings from the Beat AML cohort, we found complete mutual exclusivity between mitochondrial mutations and CBFB::MYH11 in the TCGA cohort (Figure 1F; see Table S3 for the contingency table). The mutually exclusive pattern indicated the existence of substitutable functional mutations, which often align with the same biological pathways.9 Therefore, it is deduced that the CBFB::MYH11 fusion gene is probably involved in the metabolism reprogramming associated with mitochondrial metabolism in AML. To validate this assumption, we retrieved differentially expressed genes driven by CBFB::MYH11 using public RNA-seq data on CBFB::MYH11 overexpressed cells versus the parental cells.10 A total of 985 upregulated genes and 1138 downregulated genes were found in CBFB::MYH11 overexpressed cells (Figure 2A). Gene ontology (GO) enrichment analysis showed that the CBFB::MYH11-downregulated genes were expectedly enriched by functional categories associated with neutrophil activation and regulation of cytokine-mediated signalization. Interestingly, we found that the CBFB::MYH11-upregulated genes were mainly enriched with metabolic pathways, that is, lipid, fatty acid and glycosphingolipid metabolic processes (Figure 2B). We then conducted enrichment analysis for metabolic pathways in the KEGG database and found that CBFB::MYH11-upregulated genes, including ABO, FUT1, GBGT1, GLA, ST3GAL3, ST3GAL5 and ST6GALNAC6, contributed to the activation of glycosphingolipid biosynthesis (Figure 2C). This pathway is known to play a compensatory role in mitochondrial dysfunction. This finding on the CBFB::MYH11-overexpressed cell line was further confirmed by two independent cohorts performed on primary AML patients (details in the Supporting Information). As shown in Figure 2D,E, most of the genes involved in glycosphingolipid biosynthesis identified above showed a tendency to be highly expressed in AML patients with CBFB::MYH11 compared to those without this fusion. Among them, mutations in ST3GAL5 have been reported to induce GM3 synthase deficiency and then alter the biosynthesis of glycolipid, and its aberrant expression contributes to mitochondrial dysfunction.11, 12 Given that CBFB::MYH11 is an oncogenic transcription factor, we asked whether the upregulation of these genes involved in the glycosphingolipid biosynthesis is caused by the direct binding of CBFB::MYH11. We then downloaded ChIP-seq data (GSE46044) for CBFB::MYH11 in AML.13 Interestingly, we found that the flanking regions (mainly promoter regions with ≤1 kb of transcription start site [TSS]) of all these seven genes show signal enrichment by CBFB::MYH11 in at least one sample. Particularly, the results showed that the promoter region (≤1 kb of TSS) of ST3GAL5 was enriched in all the ChIP-seq data (Figure 2F), indicating that the increased expression of ST3GAL5 could be directly regulated by the CBFB::MYH11 fusion protein. To validate the activation effect of CBFB::MYH11, we selected ST3GAL5 as the target gene because of its well-known function on mitochondrial dysfunction and the strong enrichment signals on chromatin-based on our analysis of ChIP-seq data. We constructed a CBFB::MYH11-expressing cell model in the haematopoietic cell line U937 (Figure 2G). Indeed, a significantly increased ST3GAL5 expression was detected by the quantitative reverse transcription polymerase chain reaction after the induction of CBFB::MYH11 expression (Figure 2H). Next, to determine whether CBFB::MYH11 transcriptionally activates ST3GAL5 by directly binding to the promoter region, a luciferase reporter assay was performed by transfecting the reporter plasmids containing a 500 bp-long promoter region of ST3GAL5 in U937 cells with and without CBFB::MYH11 overexpression. As shown in Figure 2I, the luciferase activity of the ST3GAL5 promoter was significantly increased in CBFB::MYH11-overexpressing U937 cells (p < 0.05) compared with parental cells. A recent study has shown that CBFB::MYH11 can activate target genes by interacting with RUNX1 on chromatin in abnormal myeloid progenitors.14 We also found a colocalization of CBFB::MYH11 and RUNX1 on the ST3GAL5 promoter region (Figure 2F). We then asked whether RUNX1 is indispensable in the CBFB::MYH11-mediated activation of ST3GAL5. As shown in Figure 2J, we performed RUNX1 knock-down experiments in CBFB::MYH11-expressing U937 cells and found that RUNX1 knock-down led to a significant decrease in ST3GAL5 expression (Figure 2J). Also, a reduced luciferase reporter activity was observed in CBFB::MYH11-expressing U937 cell after RUNX1 knock-down (p < 0.05; Figure 2K), and the consistent result was confirmed in the ME1 cell line carrying the CBFB::MYH11 fusion gene (Figure 2K). Collectively, these experiments suggested that CBFB::MYH11 increased the transactivation of ST3GAL5 to induce the glycosphingolipid pathway in a RUNX1-dependent manner. Taken together, this study established that the somatic mitochondrial mutations were mutually exclusive to the CBFB::MYH11 fusion gene in AML. Further mechanistic investigations showed that CBFB::MYH11 contributed to the activation of glycosphingolipid biosynthesis in a RUNX1-dependent manner by directly activating and binding to genes involved in this pathway. Given that AML patients with IDH1 mutations are mutually exclusive to mitochondrial mutations and maintain sensitivity to venetoclax,15 it should be possible to develop a novel therapeutic strategy by combining it with the mitochondrial metabolism as a target to treat AML patients with the CBFB::MYH11 fusion gene. Kankan Wang supervised the study and analysis; Chaoxian Zhao, Siyang Wang and Kankan Wang analysed the data and conducted experiments; Kankan Wang and Chaoxian Zhao wrote the manuscript; and all authors discussed the results and implications and reviewed the manuscript. This work was supported by the National Natural Science Foundation of China (81890994), the National Key R&D Programme of China (2019YFA0905902) and the Innovative Research Team of High-level Local Universities in Shanghai. The computations were run on the π 2.0 cluster supported by the Center for High Performance Computing at Shanghai Jiao Tong University. All authors declare no competing financial interests. Files containing somatic mitochondrial mutations in AML and the list of differentially expressed genes between CBFB::MYH11 overexpressed cells and the parental cells are available at our GitHub web site (https://github.com/Chaoxian/CBFB::MYH11_mutual_exclusivity). Appendix S1 Table S1 Table S2 Table S3 Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

CMSS1::FLT1 Rearrangement Leads to Ligand‐independent Activation of FLT1 Signaling in Acute Myeloid Leukemia

A Distinct Epigenetic Signature at Targets of a Leukemia Protein

Detection of FLT3/ITD, JAK2(V617F) and NPM1 gene mutations in chronic myelomonocytic leukemia

Mutual Exclusivity Between the Fusion Gene CBFB::MYH11 and Somatic Mitochondrial Mutations in Acute Myeloid Leukaemia

The IL-3, IL-5, and GM-CSF common receptor beta chain mediates oncogenic activity of FLT3-ITD-positive AML

The CBFβ-SMMHC/NRP1 Axis Regulates FLT3 and TGF-Beta Pathways in Inv(16) Acute Myeloid Leukemia

Identification of a T(x;17)(q28;q21) Generating a KANSL1-MTCP1 Gene Fusion Leading to Dysregulated Expression of MTCP1 in Acute Myeloid Leukemia.

Identification of Genes Abnormally Expressed in Human Mll-Af4 Leukemia.

Single cell proteogenomic sequencing identifies a relapse-fated AML subclone carrying FLT3-ITD with CN-LOH at chr13q

Identification of Cooperating Genetic Events in Acute Leukemia

A Leukemic Progression Model of Severe Congenital Neutropenia Uncovers a Novel Mechanism of AML Development Involving Elevated Inflammatory Responses, Mutation of CXXC4 and Decreased TET2 Levels

Molecular and Clinical Aspects of Acute Myeloid Leukemia with Inv(3)(q21q26)/t(3;3)(q21;q26) Carrying Spliceosomal Mutations

Differential Regulation Of C-Myc/Lin28 Discriminates Subclasses Of Rearranged Mll Leukemia

Signaling Pathway Mutations Cooperate with the PICALM/MLLT10 Fusion in a Knock-in AML Mouse Model

Profiling of somatic mutations and fusion genes in acute myeloid leukemia patients with FLT3-ITD or FLT3-TKD mutation at diagnosis reveals distinct evolutionary patterns

Global Increase in H3K79 Dimethylation in Murine and Human MLL-AF4 Lymphoblastic Leukemias.

High Expression of FLT3 is a Risk Factor in Leukemia

FLT3 Inhibition Downregulates EZH2 in AML and Promotes Myeloid Differentiation

ASXL1-Mutant Chronic Myelomonocytic Leukemia Is Associated with Increased Intratumoral Heterogeneity and Single-Cell Chromatin Co-Accessibility

The FLT3-ITD mutation and the expression of its downstream signaling intermediates STAT5 and Pim-1 are positively correlated with CXCR4 expression in patients with acute myeloid leukemia

Genes Similarly Abnormally Expressed in Both Human and Murine MLL-Associated Leukemia.