Abstract:Background The heterogeneity of acute myeloid leukemia plays an important role in the development of drug resistance. Minimal residual disease is the major cause of relapse for the patients who already received chemotherapy and stem cell transplantation. Some rare subgroups of leukemia cells harboring relapse-inducing genes were selected after chemotherapy. To unravel intra-tumoral heterogeneity and selective drug-resistance, single-cell RNA sequencing (scRNA-seq) was already performed on many solid tumors and blood cancer to achieve the high-resolution transcriptomic profiling on individual cells from a larger heterogeneous population. However, the mechanism underlying the reason of leukemia heterogeneity is still not validated. Methods Since single-cell suspension was obtained from bone marrow of acute myeloid leukemia samples, we used the 10x Genomics Chromium platform to capture transcriptomes of single cells on barcoded mRNA capture beads for massively parallel scRNA-seq. Data processing followed by Cell Ranger software pipeline to demultiplex cellular barcodes, and map reads to the genome and transcriptome hg38 using the STAR aligner. Unique molecular identifier (UMI) count matrix and quality control was performed using Seurat. The t-SNE map was calculated using Rtsne package. The related genes were overexpressed or knockdown in Thp1 cell line to investigate their roles in DNA damage response. Results We analyzed transcriptome data from above 50K single leukemia bone marrow cells across 3 patients during newly diagnosed, complete remission and relapse stages. These 3 patients belong to M4 or M5, according to French-American-British classification. The genotyping of them exhibited positive WT1-ABL, and negative for c-Kit, NPM1, FLT3-ITD and FL3-TKD. To define the landscape of cellular heterogeneity and its association with outcome in an unbiased manner, we performed unsupervised machine learning algorithm on near 50K single cells from leukemia bone marrow and identify one robust 14-cluster solution (from 0 to 13, Figure 1A) and the hallmark genes within each clusters (Figure 1B). The pattern exhibits distinct distribution on different stages (Figures 1C), indicating intra-tumoral heterogeneity during leukemia progression. The cluster 0 mainly contains newly diagnosed cells. Within cluster 0, several hallmark genes (Figure 2A), such as LILRB2(Leukocyte immunoglobulin like receptor 2) and TNFAIP2(TNF alpha induced protein 2) were already reported to be associated with AML. Interestingly, the subpopulation of PTAFR (Platelet-activating factor receptor) highly expressing cells were identified in newly diagnosed group, though whether PTAFR can induce acute myeloid leukemia is still elusive. Cells expressing these genes rarely occur at relapse region (Figure 1C), implying that these genes may be not related to relapse. However, cells expressing genes, such as RPS4Y1(Ribosomal protein S4, Y isoform 1), CDKN2A, KLF1 or GATA1 were mainly distributed in relapse group, and rare cells can be found in newly diagnosed group, implying that these genes may have potential role in chemotherapy resistance, and subclones expressing these genes expanded during leukemia progression. The high expression of KLF1 as erythropoietic transcription factor and GATA1 was associated with AML. Higher levels of GATA1 expression had poor outcomes. Next, we found these genes were associated with abnormal DNA damage response, which contributes to leukemia heterogeneity. These genes can regulate the pathway of DNA damage repair as the switch between NHEJ (Non-Homologous End Joining) and HR (Homologous recombination). Conclusion The application of scRNA-seq in cancer has been relatively limited in patients who achieve remission and appear relapse after therapy. We analyze heterogeneity of acute myeloid leukemia during progression and identify novel relapse subgroups in these heterogeneous populations. Our results show a greater degree of heterogeneity in AML tumor samples during leukemia progression, compared to traditional bulk RNA-seq, and highlights the potential of single-cell RNA-seq to identify novel subgroups in AML related to relapse, demonstrating the value of single-cell analysis for AML. Furthermore, we illustrate the roles of relapse-related genes in regulating DNA damage response, which facilitate better understanding of leukemia heterogeneity.

Abnormal DNA Damage Response Suggests Novel Fingerprint of Relapse in Acute Monocytic Leukemia Post Transplantation