Genetic characterization of stem cell-like cancer cell in the peri-tumoral regions and proliferative lymphocyte in the peripheral blood of patients with glioblastoma

Abstract

Recurrence is almost inevitable in glioblastoma (GBM) patients even after aggressive multimodal therapies. Most recurrences occur at the resection margin that is outside GBM's contrast-enhancing region (C region) of magnetic resonance imaging (MRI), and always left behind after surgery. Current knowledge about GBM cells outside C region is limited and thus impedes the post-operative therapy. In this study, single cell RNA-sequencing (scRNA-seq) was performed to dissect twelve GBM samples of C regions and outside C regions from four patients. We found a cluster of peculiar GBM cells outside C region different from the GBM cells with typical copy number variations (CNVs), which had only fragmental CNVs in chromosomes 1, 10, 11, 12 and 19, and were termed CNV 4 in this study. Furthermore, transcription factor-cofactor (TF-TCF) interaction network was constructed to uncover the transcriptional regulatory mechanism of GBM cells. The core signature of TF-TCF interaction network of CNV 4 GBM cells shows prognostic significance in GBM patients. Besides, sub-cluster of these peculiar GBM cells possess stem cell-like properties and have potential of tumorigenesis as revealed by in silico and in vitro analyses. To explore whether these peculiar GBM cells exist in the circulation of GBM patients, additional peripheral blood mononuclear cells (PBMC) samples from four GBM patients, three epilepsy patients and three healthy volunteers were analyzed by scRNA-seq. GBM cells were not found in PBMC, but peculiar proliferative immune cells were discovered in the PBMC of GBM patients but epilepsy patients and healthy volunteers. Collectively, these results indicated that peculiar GBM cells 3 outside C region are responsible for recurrence, and proliferative immune cells are specific to the PBMC of GBM patients. This study sheds light on the hidden and neglected parts of GBM, providing a new perspective to understand the nature of GBM recurrence, to develop monitoring strategy or promising immune therapy against GBM.Notably, most GBM cells of CNV patterns 1, 2 and 3 (77.44%, 95.73%, and 98.90%, respectively), whose CNVs were similar with classical GBM hallmarks, came from C regions, whereas GBM cells of CNV pattern 4 were found mainly in P and N regions (59.43% and 35.05%, respectively) (Chi-square test, P < 0.0001) ( Figure 1E). CNV patterns contribute to transcriptomic status of GBM cells in limited extent and show potential transitional relationshipsClassification of GBM cells according to CNV profiles was based on transcriptomic information, so we next analyzed the relationship between CNV patterns and transcriptomic sub-clusters. Totally, 6, 715 GBM cells were divided into 11 sub-clusters ( Figure 2A). Interestingly, sub-cluster 3 accounted for 93.93% of CNV pattern 3, and sub-cluster 5 accounted for 94.31% of CNV pattern 2, while CNV pattern 1 and 4 were made up of diverse sub-clusters in relatively balanced role in four CNV patterns, and MYC in CNV patterns 2, 3 and 4, while TP53 and E2F1 were specific to CNV pattern 1, SMAD2 s...