Comparison of cell-free and small extracellular-vesicle-associated DNA by sequencing plasma of lung cancer patients

Abstract

Blood contains multiple analytes that can be used as liquid biopsy to analyze cancer. Mutations have been detected in DNA associated with small extracellular vesicles (sEVs). The genome-wide composition and structure of sEV DNA remains poorly characterized, and whether sEVs are enriched in tumor signal compared to cell-free DNA (cfDNA) is unclear. Here, using whole-genome sequencing from lung cancer patients we determined that the tumor fraction and heterogeneity are comparable between DNA associated with sEV (<200 nm) and matched plasma cfDNA. sEV DNA, obtained with size-exclusion chromatography, is composed of short ∼150-180 bp fragments and long >1000 bp fragments poor in tumor signal. The structural patterns of sEV DNA are related to plasma cfDNA. Mitochondrial DNA is relatively enriched in the sEV fractions. Our results suggest that DNA associated to sEV (including exosomes) is not preferentially enriched in tumor signal and is less abundant than cfDNA.

Keywords: Cancer; Cancer systems biology; Molecular genetics.

Graphical abstract

Figure 1

Experimental workflow and quality control of the EV separation (A) Schematic workflow of the AFC fractionation and DNA analysis. (B) Representative EM micrographs of the structures in fraction 1–5 (left panel) and fraction 12–15 (right panel). The white bar indicates the scale (200 nm). (C) Representative western blot with a range of EV-related targets for the different fractions. (D) Determination of the particle sizes (in nm) and concentration (in particles/mL) using a TRPS device depending on the AFC fractions.

Figure 2

DNA concentration is low in EV fractions (A) Concentration (color intensity) and size (bp) of the DNA in the different fractions determined using an electrophoresis device (patient Code53 is shown in this plot). (B) Concentration in ng/uL of the DNA in the different AFC fractions determined using an electrophoresis device for all patients. p values are indicated (paired Wilcoxon test). (C) Schematic explaining the different treatments applied to the EV fractions from four samples. (D) Concentration in ng/uL of the DNA in the different fractions depending on the treatment conditions.

Figure 3

EV fractions are not enriched in tumor-derived DNA (A) Heatmap of the log₂ratio copy-number aberrations calculated using sWGS data. Rows are samples, and columns are genomic positions (100 k bins). Columns are grouped depending on the chromosomes. Blue indicates a relative decrease in copy numbers and red an increase in copy numbers. The respective patients from which the samples are taken and the type of fractions are indicated as a left annotation. (B) Tumor fraction as estimated from the sWGS data using ichorCNA. p values are indicated (paired Wilcoxon test).

Figure 4

EV fractions are enriched in longer DNA fragments (A) Median DNA fragment sizes distribution for each of the AFC fractions and matched non-fractionated cfDNA samples. (B) Median log₂ratio of the DNA fragment size distribution in comparison to the median size distribution of non-fractionated plasma cfDNA samples. (C) Proportion of DNA fragment between 20 and 150 bp (P20_150). p values are indicated (paired Wilcoxon test). (D) Heatmap of the proportion of trinucleotides at the end of DNA fragments from the different AFC fractions and matched cfDNA plasma samples. Yellow indicates high proportions and blue low proportions. The respective patients from whom the samples are taken and the type of fractions are indicated as a left annotation. (E) Diversity in the proportion of fragment-end trinucleotides calculated using a Gini index for the DNA of the AFC fractions and matched cfDNA plasma samples. p values are indicated (paired Wilcoxon test).

Figure 5

EV fractions are relatively enriched in mitochondrial DNA fragments The boxplots indicate the proportion of mitochondrial DNA fragments per fractions. p values are indicated (paired Wilcoxon test).