Exploring circular MET RNA as a potential biomarker in tumors exhibiting high MET activity

Abstract

Background: MET-driven acquired resistance is emerging with unanticipated frequency in patients relapsing upon molecular therapy treatments. However, the determination of MET amplification remains challenging using both standard and next-generation sequencing-based methodologies. Liquid biopsy is an effective, non-invasive approach to define cancer genomic profiles, track tumor evolution over time, monitor treatment response and detect molecular resistance in advance. Circular RNAs (circRNAs), a family of RNA molecules that originate from a process of back-splicing, are attracting growing interest as potential novel biomarkers for their stability in body fluids.

Methods: We identified a circRNA encoded by the MET gene (circMET) and exploited blood-derived cell-free RNA (cfRNA) and matched tumor tissues to identify, stratify and monitor advanced cancer patients molecularly characterized by high MET activity, generally associated with genomic amplification.

Results: Using publicly available bioinformatic tools, we discovered that the MET locus transcribes several circRNA molecules, but only one candidate, circMET, was particularly abundant. Deeper molecular analysis revealed that circMET levels positively correlated with MET expression and activity, especially in MET-amplified cells. We developed a circMET-detection strategy and, in parallel, we performed standard FISH and IHC analyses in the same specimens to assess whether circMET quantification could identify patients displaying high MET activity. Longitudinal monitoring of circMET levels in the plasma of selected patients revealed the early emergence of MET amplification as a mechanism of acquired resistance to molecular therapies.

Conclusions: We found that measurement of circMET levels allows identification and tracking of patients characterized by high MET activity. Circulating circMET (ccMET) detection and analysis could be a simple, cost-effective, non-invasive approach to better implement patient stratification based on MET expression, as well as to dynamically monitor over time both therapy response and clonal evolution during treatment.

Keywords: Biomarker; Circular RNA; Molecular-targeted therapy resistance; Receptor tyrosine kinase.

Fig. 1

CircMET RNA identification. a, b Box plots of mean expression (a) and RNA-seq scores (b) of the indicated human MET circRNAs according to circAtlas [58] and circBase [54] dataset collections, respectively. CircMET is highlighted in red. c Mean junction ratio of MET circRNAs identified using circAtlas [58]. Mean junction ratio is defined as the ratio between back-splicing junction reads and the total number of reads aligned to the junction site. CircMET is highlighted in red. d RNA-seq scores of the indicated murine Met circRNAs according to circBase repository [54]. CircMet is highlighted in green. e Schematic representation of MET locus-derived circRNA conservation across vertebrates based on circAtlas [58] conservation analysis output. Filled boxes indicate the presence of the circRNA in each distinct species. Human circMET is reported in red, whereas murine circMET is reported in green. f, g Schematic representation of predicted MET locus-derived circRNA products containing human MET exon 2 (f) or mouse Met exon 3 (g) and related validation by PCR analysis in human gastric cancer (GTL16) and mouse sarcoma (#1640) cells, respectively. Intron–exon circRNA structure and predicted amplicon lengths are graphically represented on the left. Divergent non junction-spanning (non JS) primers were used to detect multiple exon 2- or exon 3-derived circRNAs. PCR bands corresponding to human and murine circMET are highlighted by a red and green box, respectively. h Sanger sequencing of human and murine PCR amplicons in the red and green boxes of panel f and g, respectively. Back-spliced junctions (JS) are highlighted

Fig. 2

CircMET RNA characterization. a PCR analysis of the indicated transcripts. MET Δex2 represents the 7-Kb MET transcript lacking exon 2. b Schematics of primer pairs designed to amplify linear MET exon 2 (convergent) or its deriving circMET RNA product (divergent). c PCR analysis on genomic DNA (gDNA) and complementary DNA (cDNA) derived from the indicated cell lines. Junction-Spanning (JS) divergent primers were used to specifically detect circMET, while convergent primers were used to detect linear MET mRNAs. ACTB was used as a control. d Schematic representation of the probe used for Northern blot analysis together with its location on the linear and circular MET RNAs and Northern blot on 7.5 µg total RNA from GTL16 cell line treated or not with RNase R. The linear and the circular RNA forms are indicated next to the gels with the ‘‘–’’ and ‘‘o’’ symbols, respectively. e Real-time PCR analysis of linear MET and circMET levels upon RNase R digestion in the indicated cell lines. f Real-time PCR analysis of circMET turnover in the indicated cells treated with Actinomycin-D (n = 5). g-i Real-time PCR analysis of circMET and linear MET mRNA levels during myogenic differentiation of RD18 human rhabdomyosarcoma cells conditionally expressing miR-206 (g), NIH 10T1/2 murine fibroblasts conditionally expressing MyoD (h), and murine muscle stem cells in proliferation and differentiation medium (i). j Quantification of circMET levels in nuclear and cytoplasmic fractions of the indicated cell lines. RNU48 was used as a nuclear positive control. Data are expressed as mean ± SEM. ^NSP > 0.05; **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test

Fig. 3

Assessment of circMET expression in cancer cells and primary tumors. a, b Box plots of abundance (number of reads) (a) and frequency (number of tumors) (b) of all circRNAs detected in tumor samples based on MiOncoCirc compendium-based analysis [17]. cicMET is indicated in red, along with its corresponding values. c Quantitative real-time PCR analysis of circMET, linear MET, and MET transcript lacking exon 2 (MET Δex2) in the indicated cell lines. d Correlation between circMET and MET linear mRNA levels measured by real-time PCR in cancer cell lines (r = Pearson correlation coefficient) (n = 16). e Dot plot of circMET levels measured by real-time PCR in cancer cell lines with and without MET amplification (n = 16). f Correlation between circMET and linear MET mRNA in tumor samples (ρ = Spearman correlation coefficient) (n = 237) based on MiOncoCirc data [17]. g Spearman rank correlation of the most abundant circRNAs (1^st quartile of the data included in panel b). CircMET coefficient is indicated with a dotted red line (ρ = 0.51). Data are expressed as mean ± SEM. **P < 0.01, Student’s t test

Fig. 4

CircMET evaluation reveals MET-driven acquired resistance and mirrors MET-targeted therapy response in vitro. a Gene Copy Number (GCN) analysis of the indicated parental (par.) and clonal (cl.), sensitive and resistant (res.) HCC827 cell populations. Red dashed line indicates the twofold threshold for MET amplification. b Dot plot of circMET levels measured by real-time PCR in 11 among HCC827 and WiDr parental and resistant cells with and without MET amplification. Triangles are referred to WiDr parental and resistant cells. c Proliferation assay of afatinib-resistant HCC827 clonal cells treated with the indicated inhibitors. Representative pictures are shown below (scale bar = 500 μm). d Soft-agar colony formation assay of afatinib-resistant HCC827 clonal cells treated with the indicated inhibitors. Representative pictures are shown below (scale bar = 500 μm). e, f MET GCN (e) and real-time PCR analyses (f) of afatinib-resistant HCC827 cell subclones upon 7 days of treatment with the indicated inhibitors. AFAT, afatinib; ERLOT, erlotinib; CTX, cetuximab. Data are expressed as mean ± SEM. ***P < 0.001, Student’s t test

Fig. 5

Circulating circMET in cancer patients correlates with the MET status of the corresponding tissue biopsies. a, b Immunohistochemical analysis of low- (a) and high- (b) MET-expressing human lung adenocarcinomas and mouse rhabdomyosarcoma (Mouse) specimens with the indicated antibodies (scale bar = 100 μm). c Dot plot of circMET levels measured by real-time PCR in FFPE tumor samples (n = 12). Samples included in the analysis are listed in Table S2. **P < 0.01, Student’s t test. Data are expressed as mean ± SEM. d In situ detection of circMET in a MET-amplified FFPE tumor sample (case #4) using padlock probes and rolling circle amplification (scale bar = 25 μm). e Absolute quantification of circMET levels measured by digital PCR in the indicated liquid biopsies of cancer patients (n = 16) and healthy donors (HD) (n = 2). Each sample was analyzed in duplicate: both circMET values are shown for each patient (colored dots), along with their average value (black lines) and SEM. Low and high circMET expressors are indicated in violet/green and red, respectively. Clinical information of the primary human samples analyzed is provided in Table S2

Fig. 6

CircMET allows non-invasive tracking of MET-driven acquired resistance and therapy response in cancer patients. a Upper panel: clinical history of lung adenocarcinoma case #5. Black and yellow arrows on CT scans indicate primary and secondary lung lesions, respectively, while green arrows indicate liver lesions. FISH for MET in the post-therapy biopsy (case #5R) is shown on the right. MET is labeled in green and centromeres in red. Lower panel: dPCR on circMET along with real-time PCR on EGFR T790M and EGFR Ex19del upon cell-free DNA (cfDNA) extraction from liquid biopsies longitudinally obtained at the indicated time points (N/A = not available). b Immunohistochemical and real-time PCR analyses on FFPE biopsies collected before and after therapy (case #5 and case #5R) (scale bar = 100 μm). c dPCR analysis of circMET upon cfRNA extraction along with ddPCR-based Copy Number Variation (CNV) assessment on cfDNA from liquid biopsies obtained at the indicated time points of a previously described colorectal cancer case [44, 47]. Data are expressed as mean ± SEM