miRNA profiling of circulating EpCAM(+) extracellular vesicles: promising biomarkers of colorectal cancer

Abstract

Cancer cells secrete small membranous extracellular vesicles (EVs) into their microenvironment and circulation. These contain biomolecules, including proteins and microRNAs (miRNAs). Both circulating EVs and miRNAs have received much attention as biomarker candidates for non-invasive diagnostics. Here we describe a sensitive analytical method for isolation and subsequent miRNA profiling of epithelial-derived EVs from blood samples of patients with colorectal cancer (CRC). The epithelial-derived EVs were isolated by immunoaffinity-capture using the epithelial cell adhesion molecule (EpCAM) as marker. This approach mitigates some of the specificity issues observed in earlier studies of circulating miRNAs, in particular the negative influence of miRNAs released by erythrocytes, platelets and non-epithelial cells. By applying this method to 2 small-scale patient cohorts, we showed that blood plasma isolated from CRC patients prior to surgery contained elevated levels of 13 EpCAM(+)-EV miRNAs compared with healthy individuals. Upon surgical tumour removal, the plasma levels of 8 of these were reduced (miR-16-5p, miR-23a-3p, miR-23b-3p, miR-27a-3p, miR-27b-3p, miR-30b-5p, miR-30c-5p and miR-222-3p). These findings indicate that the miRNAs are of tumour origin and may have potential as non-invasive biomarkers for detection of CRC. This work describes a non-invasive blood-based method for sensitive detection of cancer with potential for clinical use in relation to diagnosis and screening. We used the method to study CRC; however, it is not restricted to this disease. It may in principle be used to study any cancer that release epithelial-derived EVs into circulation.

Keywords: blood-based CRC detection; colorectal cancer; epithelial-derived extracellular vesicles; immunoaffinity; isolation; microRNA; non-invasive biomarkers.

Fig. 1

Characterization of isolated CRC cell-derived extracellular vesicles. (a) TEM analysis of EVs isolated from the bioreactor-cultivated SW620 CRC cell line. The mean size and standard deviation of 32 EVs is indicated. (b) SW620 cells and their secreted EVs were analysed for the expression of exosomal markers CD63, CD81, and Hsp90, as well as epithelial-specific EpCAM by immunoblotting using 3.6 µg protein. (c) Evaluation of the quantity of EpCAM and EGFR (negative control) on SW620 EVs by PLA. Serial dilutions of isolated SW620 EVs spiked into plasma from a healthy individual were performed and followed by PLA analysis. As positive controls serial dilutions of recombinant EpCAM and EGFR spiked into PLA buffer were used. Shown are mean values and standard deviations of 2 experiments. (d) Evaluation of the EpCAM immunoaffinity approach's ability to recover known levels of circulating EVs. As a measure of recovery, we used the number of miRNAs detected (Cq<40) in the RNA extracted from the vesicles isolated by applying the immunoaffinity-capture to series of plasma and serum samples spiked with incrementing amounts of SW620 EVs. IgG beads were used as a negative control. (e) The mean miRNA-assay Cq-values±S.E.M. measured after EpCAM or IgG immunoaffinity-capture of the plasma (left panel) and serum (right panel) SW620 EV spike-in series. From 3.2 µl of EV spike-in and upwards a linear relationship between the mean Cq value and the EV input level was observed. The detection thresholds for EpCAM were defined as 1 Cq-value below the IgG level of the 3.2 µl SW620 EV spike-in point. The dashed horizontal lines define the detection thresholds. Below the detection thresholds the signal (EpCAM) is significantly higher than the noise (IgG) as indicated by the grey-shaded area of the graph. P-values represent Mann–Whitney U tests (H₀: no difference between EpCAM and IgG Cq-values). (f) Recovery was also evaluated as the difference in average miRNA quantity (EpCAM vs. IgG beads) at each dilution point. Only miRNAs detected with Cq<40 in the EpCAM analysis were included. MicroRNAs with a signal (Cq<40) in all of the IgG spike-in series points were excluded as technical false positive assays. Depicted below the graph, is the number of miRNAs detected at each dilution point, excluding the false positive.

Fig. 2

Increased miRNA abundance associated with EpCAM⁺ extracellular vesicles in plasma and serum of CRC patients. Representative examples of miRNA amounts detected in EpCAM⁺-EVs in (a) plasma or in (b) serum from healthy individuals and CRC patients compared to the amount detected when SW620 EVs were spiked into (a) plasma or (b) serum from a healthy individual. The miRNA abundance in CRC plasma samples represented an acceptable signal as compared to the signal-to-noise window of EpCAM (signal) s. IgG (noise) bead spike-in series. (c) The number of miRNAs detected in EpCAM⁺-EVs isolated from plasma and serum of healthy individuals and CRC patients (Patient cohort I). Following criteria were used: i) All false positive miRNAs assays were excluded, ii) Only miRNAs detected in clinical samples at Cq<35.9 (plasma) and Cq<35.3 (serum) were included. Plasma from healthy individual 3 was excluded based on an extreme outlier profile in the initial quality check. (d) Box plot of the number of miRNAs detected from isolated EpCAM⁺-EVs (Student's t-test, p<0.05, **p<0.01).

Fig. 3

Analysis of altered miRNA profiles in EpCAM⁺-EVs from CRC patients. (a, b) Unsupervised Principal Component Analysis (PCA) demonstrating separation of samples based on their origin from CRC patients or healthy individuals from miRNA profiling of circulating EpCAM⁺-EVs. (a) Plasma, a total of 31 miRNAs were included in the analysis using the following selection criteria: i) miRNA expression in 6/6 cancer samples, ii) detected at Cq<35.9, and iii) exclusion of false positive miRNA assays. (b) Serum, a total of 7 miRNAs were included in the analysis using selection criteria: i) miRNA expression in 6/6 cancer samples, ii) detected at Cq<35.3, iii) exclusion of false positive assays. The patient ID number is depicted at each data point. (c, d) Unsupervised hierarchical clustering analysis based on the miRNAs from (a, b) (yellow-high expression, blue-low expression, n.d- not detected). (e) miRNAs significantly more abundant in EpCAM⁺-EVs from CRC patients than healthy controls (p≤0.05, Student's T-test). Shown are data for both plasma and serum.

Fig. 4

Circulating EpCAM⁺-EV-derived miRNAs are reduced following CRC surgery. (a) The number of miRNAs detected before and after surgery. EpCAM⁺-EVs were isolated from matched pre- and post-operation plasma from 7 stage III patients. From these, the abundance of 43 miRNAs associated with EpCAM⁺-EVs were profiled using a custom-designed Pick & Mix qRT-PCR panel. A miRNA was considered detected if it was below the previously defined threshold of Cq<35.9. Individual patient ID numbers are displayed. (b) 26 miRNAs were detected in all 7 pre-operation samples and shown are average abundance of these in pre- and post-operation samples. The increased Cq values observed in the post-operation samples relative to the pre-operation samples indicate reduced abundance. The data are depicted as the average abundance in all CRC samples. (c) A volcano plot of the 26 individual miRNAs showing the relationship between statistical significance (paired Student's t-test) and the fold-change between pre- and post-operation samples (▵Cq (pre – post)). The dashed line indicates p=0.05. MicroRNAs with distinct abundance in EpCAM⁺-EVs in plasma samples from CRC patients and healthy controls are marked in red. (d) Shown are paired pre- and post-OP Cq-values for 10 selected EpCAM⁺-EV miRNAs with significant difference in pre- and post-operation abundance.