Total whole-arm chromosome losses predict malignancy in human cancer

Abstract

Aneuploidy is observed as gains or losses of whole chromosomes or chromosome arms and is a common hallmark of cancer. Whereas models for the generation of aneuploidy in cancer invoke mitotic chromosome segregation errors, whole-arm losses might occur simply as a result of centromere breakage. We recently showed that elevated RNA Polymerase II level over the S-phase-dependent histone genes predicts rapid recurrence of human meningioma and is correlated with total whole-arm losses relative to gains. To explain this imbalance in arm losses over gains, we have proposed that histone overexpression at S-phase competes with the histone H3 variant CENP-A, resulting in centromere breaks and whole-arm losses. To test whether centromere breaks alone can drive aneuploidy, we ask whether total whole-arm aneuploids can predict outcomes across different cancer types in large RNA and whole-genome sequencing databanks. We find that total whole-arm losses generally predict outcome, suggesting that centromere breakage is a major initiating factor leading to aneuploidy and the resulting changes in the selective landscape that drive most cancers. We also present evidence that centromere breakage alone is sufficient to account for whole-arm losses and gains, contrary to mitotic spindle error models for the generation of aneuploidy. Our results suggest that therapeutic intervention targeting histone overexpression has the potential to reduce aneuploidy and slow cancer progression.

Keywords: RNA sequencing; aneuploidy; centromeres; histones; whole genome sequencing.

Fig. 1.

Whole-arm losses correlate with recurrence in meningioma and across 33 TCGA cancer types. (A) Whole chromosome arm gains and losses were identified using CaSpER across 1,298 meningioma RNA-seq samples (8). Whole-arm aneuploidy is defined as the total number of chromosome arm gains or losses. Kaplan–Meier curves compare recurrence times among patients with low (fewer than three), medium (three to six), and high (more than six) levels of chromosome arm alterations. The log-rank test P-values, shown in the Lower-Left corner of each panel, assess the statistical significance of survival curve separation. (B) Chromosome arm aneuploidy was determined using the ABSOLUTE algorithm in 10,522 TCGA WGS samples, as reported in ref. . Disease-free time serves as the recurrence metric. (C) Relationship between median disease-free time and the average number of whole chromosome arm losses (Left) or gains (Right) across 33 TCGA cancer types. The solid lines represent linear regression fits. The kidney chromophobe (KICH) cancer type has a significantly longer disease-free time, exceeding the range displayed in the figure.

Fig. 2.

Percentage of whole-arm gains and losses in selected TCGA data. Each histogram bar represents the overall percentage of intact, gained, and lost chromosome arms in TCGA data for the indicated cancer type (SI Appendix, Table S1), where ACC is adrenocortical carcinoma, BRCA is breast cancer, LAML is acute myeloid leukemia, and GBM is glioblastoma. For clarity, only 4 of the 33 cancer types are shown. The full TCGA set of histograms is displayed in SI Appendix, Fig. S2. Vertical black lines separate whole chromosomes, where Chromosomes 13, 14, 15, 21, and 22 are acrocentrics with a short p arm (not displayed) and long q arm.

Fig. 3.

Acrocentrics and metacentrics gain or lose whole arms at similar frequencies in cancer. (A) A metacentric segregation; (B) An acrocentric segregation; (C) A metacentric segregation following a centromere break; (D) An acrocentric segregation following a centromere break. Acrocentric p chromosome arms are short and lack unique loci; thus, intact acrocentrics are not distinguishable from q arms resulting from centromere breaks in genomic studies. (E) TCGA WGS data by tumor type. Each dot represents a different autosomal chromosome arm (5 acrocentric long arms and 17 metacentrics).

Fig. 4.

Model for generation of centromere breaks leading to losses > gains. (A) Cancer progression: normal cells (gray) proliferate and differentiate to populate tissues, but aberrant induction of histone overexpression (black) promotes both hyperplasia and chromosome instability. Selection in hyperproliferating clones drives the frequencies of certain chromosomal abnormalities and the evolution of malignant cellular features (brown, dark blue, purple). (B) Model: defective centromeres compromised by centromeric histone displacement will break, leading to widespread aneuploidy (blue) through whole chromosome arm loss, arm gain, and through whole chromosome loss. The occurrence of micronuclei by encapsulation of fragmented chromosome arms further stimulates chromosomal instability.