Heterogeneity in lung cancers by single‐cell DNA sequencing

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2023 The Authors. Clinical and Translational Medicine published by John Wiley & Sons Australia, Ltd on behalf of Shanghai Institute of Clinical Bioinformatics. genomes from seven lung adenocarcinomas (LUAD), seven lung squamous cell carcinomas (LUSC), and two small-cell lung carcinomas (SCLC). Our findings


F I G U R E 2
Copy number evolutionary interpretation on lung cancer tumors.(A) CNV subclone evolutionary tree for 16 lung cancer tumors.We marked key cancer-related genes for each branch.CNV amplified and deleted genes are colored in blue and red, respectively.(B-D) Progression-free survival curves of TCGA-LUAD patients stratified by CNV of APC (B), CEP89 (C) and FAT3 (D).
reflect lung tumors holding huge subclone diversity on copy number variations (CNVs) and complex structure variations (cSVs).
Lung cancer tops global cancer deaths, with intratumor heterogeneity (ITH) contributing to recurrence and resistance. 1 Single-cell DNA sequencing (scDNA-Seq) 2 provides a precise ITH perspective by profiling individual cells in multiple cancers, [3][4][5][6] but limited in lung cancer.Most lung cancer studies examine subclonal CNVs, 7 leaving cSV' ITH, and their role in lung cancer progression remains incompletely understood. 8,9his study analyzed 13,343 single-cell genomes from 16 lung tumors: 7 LUAD, 7 LUSC, and 2 SCLC (Figure 1, Table S1, Supporting Information Tables and Supporting Information Methods).We investigated the CNV landscape across all tumors.Tumor cell groups were obtained from hierarchical clustering (HC) using cell ranger-DNA. 6e assigned cells grouped by a leaf node in the HC cutdendrogram as cell clusters; that is, cells inside one cell cluster sharing similar CNVs.We identified 16 to 33 cell clusters per tumor (Figures S1-S3), yielding three to nine subclones per tumor and 72 subclones overall (Table S2).All tumors display polyclones, meaning they have at least two subclones.Subclones are denoted by dominant amplified (A), diploid (D), or lost (L) copy numbers (Table S2)."LX" indicates subclone has loss of heterozygosity in chromosome X.The largest subclone populates 1,273 diploid cells (LUAD03T-D), whilst we detected 23 small cell populations, that is, subclones under 10 cells (Table S2).We calculated the Gini index per subclone, with higher values reflecting greater CN dispersion across genomic regions in the subclone.Overall, cell numbers and Gini indices vary between subclones, illustrating CNV tumor heterogeneity in LUAD, LUSC, and SCLC (Figure 1 and Table S2).Moreover, hierarchical clustering of subclone CNVs revealed inter-patient similarities (Figures S4 and S5, Table S3 and Supporting Information Results).
Punctuated copy number evolution (PCNE) hypothesizes subclonal CNVs arise in short bursts of crisis In LUSC04T, the PCNE-produced subclones LX and A1 have 78.10% common genetic alterations, as does the minor subclone A2 (eight cells, 12.03% for A1, A2 and LX).In LUSC05T, 61.15% genetic alterations were recurrently observed in subclones A, LX, and L1 derived from PCNE.Subclone groups L1 and LX also share 23.68% common alterations.In SCLC01T, PCNE derived LX and A1, BCNE derived A2 and A3 also shares a plenty of genetic alterations (10.93% for A1 and LX; 15.60% for A1, A3, and LX; 14.18% for A1, A2 and LX; 31.92% for A1, A2, A3, and).The minor subclone L with six cells also shares several genetic alterations with the other four subclones.(C) Landscape of genetic alterations identified in multiple signal pathways associated with lung cancer throughout the cohort at the subclone level.The right panel demonstrates the subclone mutational frequency per gene for the cohort.The corresponding genes, gene families, and pathways are annotated on the right side.The bottom panel exhibits clinical metadata.
In brief, we used 10x scDNA-Seq to reveal extensive subclone diversity in CNVs and cSVs across LUAD, LUSC and SCLC.4][5] We suggest PCNE in three lung cancer subtypes, characterized by early genomics gains and losses, followed by BCNE.Two breakage-fusion-bridges duplicating oncogenes PLA2G4A and GBE1 were detected in LUSC, potentially linked to PCNE.cSVs are identified in lung cancers, especially with high frequency (75%) in LUSC, affecting two MHC-II genes (HLA-DRB5 and HLA-DRB1).Evolutionary analysis suggests these cSVs may occur before or during PCNE.Hence, our findings reflect extensive subclone diversity in lung tumors concerning CNVs and cSVs.
One study limitation is subclone detection dependent on 10x scDNA-Seq cell profiling, 6 partially repenting subclone diversity in tumors.Average single-cell coverage was low (Supporting Information Tables), potentially concealing SVs and cSVs in minor subclones.The issue is prevalent in scDNA-Seq, we aim to sequence more lung tumors to enrich findings.PCNE and BCNE hypotheses rely solely on observing cell phylogenies and subclone CNVs.Looking forward, we plan to mathematically model PCNE and BCNE processes for quantitative answers.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available at https://doi.org/10.57760/sciencedb.08329.

S U P P O R T I N G I N F O R M AT I O N
Additional supporting information can be found online in the Supporting Information section at the end of this article.

F I G U R E 3
Subclone-level genome aberration landscape across the cohort.(A) The number of InDel (small insertion and deletion), SV (DEL-ht, TRX-ht, DUP-th, TRX-th, INV-hh, TRX-hh, INV-tt and TRX-tt), and CSV (Chromothripsis and other CSV) in each tumor subclone.(B) The frequency of co-existence of InDel, SV, and CSV in different combinations of subclones for LUAD03T, LUAD04T, LUSC04T, LUSC05T and SCLC01T.The top, middle, and bottom layer presents InDel, SV, and CSV, respectively.In LUAD03T, the most genetic alterations coexist in subclones formed by PCNE (28.58% for D, L, LX and A; 40.81% for D, L and LX).In LUAD04T, subclones D, A1, and LX (formed by PCNE) share 57.34% common genetic alterations, subclone group D and LX have 40.14% common genetic alterations.
This work was supported by National Natural Science Foundation of China (Nos.81974363, 81772478 to Li Zhang; 81871890, 91859203, 92159302 to Weimin Li), CAMS Innovation Fund for Medical Science (No: 2019TX310002 to Weimin Li), Science and Technology Project of Sichuan (2022ZDZX0018 to Weimin Li), 1.3.5 Project for Disciplines of Excellence, West China Hospital, Sichuan University (ZYGD22009 to Weimin Li), and Key basic research projects of Shenzhen Science and Technology Innovation Commission (JCYJ20200109143216036 to Shuai Cheng Li).
Department of Pulmonary and Critical Care Medicine, Institute of Respiratory Health, State Key Laboratory of Respiratory Health and Multimorbidity, Frontiers Science Center for Disease-related Molecular Network, Precision Medicine Key Laboratory of Sichuan Province, West China Hospital, West China School of Medicine, Sichuan University, Chengdu, China 2 Department of Computer Science, City University of Hong Kong, Kowloon, China