Genomic and Functional Approaches to Understanding Cancer Aneuploidy

Abstract

Aneuploidy, whole chromosome or chromosome arm imbalance, is a near-universal characteristic of human cancers. In 10,522 cancer genomes from The Cancer Genome Atlas, aneuploidy was correlated with TP53 mutation, somatic mutation rate, and expression of proliferation genes. Aneuploidy was anti-correlated with expression of immune signaling genes, due to decreased leukocyte infiltrates in high-aneuploidy samples. Chromosome arm-level alterations show cancer-specific patterns, including loss of chromosome arm 3p in squamous cancers. We applied genome engineering to delete 3p in lung cells, causing decreased proliferation rescued in part by chromosome 3 duplication. This study defines genomic and phenotypic correlates of cancer aneuploidy and provides an experimental approach to study chromosome arm aneuploidy.

Keywords: aneuploidy; cancer genomics; genome engineering; lung squamous cell carcinoma.

Figure 1. Aneuploidy correlates with ploidy, genome doubling status, and tumor type

(A) Schematic of aneuploidy score. Step 1 is to generate copy number calls per sample, including somatic copy number alterations (SCNAs) of all sizes. Step 2 is to distinguish arm-level alterations within these SCNAs. Step 3 is to total the number of altered arms to generate the aneuploidy score. (B) Each tumor sample is organized by genome doubling status (blue = not doubled, green = 1 genome doubling, red = 2 or more genome doublings). X-axis is aneuploidy score, the sum of the number of altered chromosome arms. Y-axis is ploidy as determined by ABSOLUTE. Spearman’s rank correlation coefficient = 0.55. (C) X-axis is aneuploidy score, the sum of the number of altered chromosome arms. Y-axis is ploidy as determined by ABSOLUTE. Samples are separated by whole genome doubling status: samples without genome doubling (left, blue, Spearman’s rank correlation coefficient = -0.13), samples with one genome doubling (middle, green, Spearman’s rank correlation coefficient = -0.32), and samples with 2 or more genome doublings (right, red, Spearman’s rank correlation coefficient = 0.17). (D) Each tumor sample is organized by tumor type and genome doubling status (blue = not doubled, green = 1 genome doubling, red = 2 or more genome doublings). Samples are organized by tumor type, and ranked from least to most aneuploid samples within a tumor type. X-axis is aneuploidy score, y-axis is sample number. See also Figure S1 and Tables S1 and S2.

Figure 2. Aneuploidy score correlates with TP53 mutations and overall mutation rate

(A) Y-axis is -log₁₀ Bonferroni corrected p value for linear model coefficient of aneuploidy score. Dots represent every mutated gene. (B) X-axis is aneuploidy score. Y-axis is rate of non-silent mutations per megabase (square root). Blue samples have been called as microsatellite instability (MSI)-high or POLE mutated, whereas red samples do not have these features or have not been called. (C) Spearman correlation coefficients for aneuploidy score and mutation rate across TCGA tumor types, arranged from smallest to largest value. Tumor types in blue have MSI-high or POLE mutated samples. Average of correlation coefficients across cancer types is in purple. See also Figure S2 and Tables S3 and S4.

Figure 3. Aneuploidy score negatively correlates with immune infiltrate, which contributes to decreased expression of immune genes

(A) Spearman correlation coefficients for aneuploidy score and impurity across TCGA tumor types, arranged from smallest to largest value. Average of correlation coefficients across tumor types is in purple. (B) Spearman correlation coefficients for aneuploidy score and leukocyte fraction across TCGA tumor types, arranged from smallest to largest value. Average of correlation coefficients across tumor types is in purple. (C) Spearman correlation coefficients for aneuploidy score and non-leukocyte stroma across TCGA tumor types, arranged from smallest to largest value. Average of correlation coefficients across tumor types is in purple. (D) Heatmap of normalized enrichment scores for Hallmark gene sets in GSEA (Gene Set Enrichment Analysis), with FWER (family-wise error rate) p value < 0.01. Gray = not significant or not enriched. Predictor variables describe variables included in linear regression model for gene expression. Pathways are those identified from genes with significant coefficients in linear regression analysis. See also Figure S2 and Tables S4, S5, and S6.

Figure 4. Patterns of arm-level alterations cluster by tumor site, tissue of origin, and squamous morphology

(A) Matrix of mean arm-level alteration within each tumor type/subtype. Hierarchical clustering of tumor type by Pearson’s method. (B) Integrated genomics viewer (IGV) plots of chromosome 3 copy number alterations in lung squamous cell carcinoma or lung adenocarcinoma. Blue = loss, red = gain. Numbers for lung squamous and lung adenocarcinoma samples that had arm calls for both 3p and 3q. P values represent chi-square test for enrichment of co-occurrence of chromosome 3p loss and chromosome 3q gain. See also Figure S3 and Tables S7 and S8.

Figure 5. CRISPR-based approach can delete a chromosome arm in human immortalized cells

(A) Schematic of CRISPR and recombination based approach to delete chr_3p in vitro. A linearized plasmid containing homologous DNA, a puromycin selection marker, and an artificial telomere is co-transfected with a CRISPR-Cas9 construct to target DNA sequence adjacent to the centromere. Upon transfection, a double-strand break is produced and repaired by homologous directed recombination, removing a chromosome arm and replacing it with an artificial telomere. (B) qPCR measuring chr_3p gDNA normalized to chr_3q gDNA in single cell clones. Bars represent means with error bars +/- standard deviation. Light blue = single cell clones that were not transfected (NTF) and did not have chr_3p deletion, dark blue = single cell clones that were transfected (TF) but did not have chr_3p deletion, and red = single cell clones that were transfected and have chr_3p deletion. (* = p value < 0.05) (C) Whole genome sequencing output of HMMCopy, for one chr_3p deleted cell clone (top) and a non-deleted control clone (bottom). (D) Karyotype of one chr_3p deleted cell clone. Chromosome 3 is circled, and arrows point to chromosomal abnormalities. Mar = marker chromosome. See also Figure S4.

Figure 6. Chromosome 3p lung cells initially have slower proliferation, but normalize over time

(A) AALE cells were transfected, and one day later cells were selected with puromycin for cells that had incorporated transfected DNA. Cells were single cell cloned to isolate chr_3p deleted clones, and assayed before and after extensive passaging. (B) Proliferation curves were generated using CellTiter-Glo over 6 days (x-axis). Y-axis is relative luminescence units, normalized to Day 0. Data plotted are means with error bars +/- standard deviation. (* = p value < 0.01, ** = p value < 0.001) Cells were from Timepoint A. (C) Heatmap of normalized enrichment scores for Hallmark gene sets in GSEA. (D) Proliferation curves were generated using CellTiter-Glo over 6 days (x-axis). Y-axis is relative luminescence units, normalized to Day 0. Data plotted are means with error bars +/- standard deviation. Cells were from the higher passage population. (E) Karyotype from one of the chromosome 3p deleted clones. Chromosome 3 is circled, and arrows point to chromosomal abnormalities. (F) During passaging, cells have gained additional changes (*) to adapt to chr_3p deletion. See also Figure S4.