Exosomal microRNAs as potential biomarkers for osimertinib resistance of non-small cell lung cancer patients

2.1 Ethics statement

This study was approved by the ethics committee of The Human Research Ethics Committee of Ramathibodi Hospital, Mahidol University (IRB no. EC600169). All patients and healthy controls have signed written informed consent.

2.2 Patient characteristics and blood collection

All patients were diagnosed with advanced NSCLC with EGFR mutations and were treated with 1 st or 2 nd generation of EGFR-TKI. Only patients whom developed EGFR T790M as the acquired resistance gene at disease progression were enrolled in this study. Blood samples were collected before starting osimertinib treatment (baseline sample). Osimertinib was given until the objective disease progression was documented, at that time, the second blood sample was collected (osimertinib-resistance sample). Blood samples were collected into 6 ml K3 EDTA tubes and centrifuged at 1,500 × g for 15 min at room temperature. Plasma was collected and stored at - 80 ∘ C until used. Plasma samples from age and sex-matched healthy control subjects were collected with the same protocol.

2.3 Plasma exosome isolation

Exosome isolation was conducted using ExoQuick (System Biosciences, Palo Alto, CA, USA). Briefly, 3 μ l of thrombin was mixed with 300 μ l of plasma, to eliminate fibrinogen, and incubated for 5 min. Samples were centrifuged at 17,000 × g for 10 min. Then, 250 μ l of supernatant was transferred to a new tube and mixed with 63 μ l of the ExoQuick reagent. After incubation at 4 ∘ C overnight, the mixture was then centrifuged at 1,500 × g for 30 min. The supernatant was removed and the exosome pellet was then centrifuged at 1,500 × g for 5 min to remove all traces of fluid. The pellet was subsequently resuspended in 25 μ l PBS, filtrated with a 0.22 μ m syringe filter, and stored at - 80 ∘ C until used. Protein concentrations were measured using a BCA protein assay kit (Thermo Scientific, Wilmington, DE, USA).

2.4 Electron microscopy

A formvar-carbon-coated 300 mesh copper grid (Electron Microscopy Sciences, Hatfield, PA, USA) was discharged with UV light for 15 min before use. The exosome suspension (5 μ l) was then placed on the grid and incubated at room temperature for 20 min. The sample was fixed with 1% glutaraldehyde for 5 min, followed by washing with distilled water 5 times. The samples were subsequently stained with 2% uranyl acetate for 1 min. Grids were allowed to dry at room temperature before visualization with a Hitachi HT7700 120 kV Compact Digital Biological TEM (Hitachi, Tokyo, Japan).

2.5 Western blot analysis

The exosome pellets were lysed with RIPA lysis buffer containing 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% TritonX-100, 1 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride (PMSF), and a protease inhibitor cocktail. The exosome samples were mixed with 6X Laemmli buffer (375 mM Tris-HCl pH 6.8, 6% SDS, 48% glycerol, 9% 2-mercaptoethanol, and 0.03% bromophenol blue), heated at 95 ∘ C for 5 min, and separated using 10% of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The proteins were then transferred to a polyvinylidene fluoride membrane and subsequently blocked with 10% non-fat dry milk for 1 h at room temperature to decrease non-specific protein binding. The membranes were incubated with antibodies for exosome markers, including CD81 (Santa Cruz Biotechnology, Dalla, TX, USA); Alix and Flottilin-1 (BD Biosciences, San Jose, CA, USA), overnight at 4 ∘ C. The membranes were washed 5 times with TBST before incubation with HRP-conjugated secondary antibodies. Signals were detected using a Luminata Crescendo Western HRP Substrate (Merck, Darmstadt, Germany) and exposed onto hyperfilm (Amersham, Little Chalfont, UK).

2.6 Small RNA extraction

Total small RNA was extracted from exosome samples using an miRNeasy mini kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. Briefly, 700 μ l of QIAzol Lysis was added to exosome sample and incubated for 5 min at room temperature. The resulting homogenate sample was mixed with 140 μ l of chloroform and centrifuged at 12,000 × g for 15 min at 4 ∘ C. Then, the sample was mixed with 525 μ l of 100% ethanol, loaded into an RNeasy Mini spin column, and centrifuged at 10,000 × g for 15 sec at room temperature. The column was subsequently washed with 500 μ l of RPE buffer and centrifuged twice at 10,000 × g. Total RNAs were eluted via the addition of 30–50 μ l of RNase-free water directly onto the RNeasy Mini spin column and centrifuged for 1 min at 10,000 × g. The quality and quantity of RNAs were examined using a Nanodrop-1000 Spectrophotometer (Thermo Scientific, Wilmington, DE, USA).

2.7 miRNA deep sequencing

The cDNA preparation and miRNA expression profiling were performed by QIAGEN (QIAGEN, Hilden, Germany). All small RNA extractions were packed in tightly sealed vial tubes and shipped on dry ice. Briefly, cDNA libraries were reverse transcribed from 1 μ g of total small RNA using the TruSeq Small RNA Sample Prep Kit. MiRNA deep sequencing was performed using Illumina MiSeq with 50-bp single-end reads. The short-read sequences that were obtained from next-generation sequencing (NGS) were extracted from adaptor sequences using fastx_clipper of the fastx toolkit (http://hannonlab.cshl.edu/fastx_toolkit/). The adaptor trimmed short-reads were mapped to the human reference genome sequence by the mapper.pl script. MiRNA mature sequences were obtained from miRBase (http://www.mirbase.org). Finally, the differential miRNA was analyzed by miRDeep2.pl script software.

2.8 Prediction miRNA target genes and pathway enrichment analysis

Target genes of dysregulated miRNAs were predicted by an online bioinformatics tool miRSystem (ver.20160513) containing 7 well-known miRNA target gene prediction algorithms: DIANA, miRanda, miRBridge, PicTar, PITA, rna22, and TargetScan [29]. Only the target genes predicted from at least three databases were chosen for further analysis. To identify significant signal transduction pathways of miRNA target genes, the putative genes were analyzed according to the KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway using WebGestalt (WEB-based Gene SeT AnaLysis Toolkit) [30, 31].

2.9 Reverse transcription and quantitative real-time PCR

First-strand cDNA synthesis was performed using the miScript II RT Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. Each 20 μ l reaction contains 1 μ g of RNA sample, 4 μ l of 5x miScript HiSpec buffer, 2 μ l of 10x miScript Nucleics Mix, 2 μ l of miScript Reverse Transcriptase Mix, and nuclease-free water. The mixture was incubated for 60 min at 37 ∘ C, followed by a 5 min incubation at 95 ∘ C to inactivate miScript Reverse Transcriptase. To probe for miRNA expression, samples were amplified using the miScript SYBR Green PCR Kit (QIAGEN, Hilden, Germany) and analyzed with an ABI PRISM7500 Sequence Detection System (Applied Biosystem, Bedford, MA, USA). Briefly, in a 25- μ l reaction, 2 μ l of template cDNA was mixed with 12.5 μ l of 2x QuantiTect SYBR Green PCR Master Mix, 2.5 μ l of 10x miScript Universal Primer, 2.5 μ l of 10x miScript Primer Assay, and RNase-free water. The expression of miRNA was analyzed using the – Δ CT method normalized to the expression of a small RNA (RNU6-2).

2.10 Statistical analysis

Exosomal miRNA expression levels were compared between those at baseline and those at the resistance stage of osimertinib using paired t-test. Statistical significance was considered as 0.05. The area under the ROC curve (AUC) was used as an accuracy index for evaluating the discriminating performance of exosomal miRNAs. Statistical significance was considered as 0.05. Data normalization and statistical analysis were performed using GraphPad Prism 8 (GraphPad Software Inc; San Diego, CA, USA). The Kaplan-Meier method was used to estimate the TTF and OS. Hazard ratios and 95% confidence intervals (95% CI) were assessed with Cox proportional-hazards model.

OPEN IN VIEWER

3.1 Characterization of exosomes derived from plasma of NSCLC patient

A total of 8 lung cancer patients (3 males and 5 females) were recruited to investigate exosomal miRNA profiles. All patients were treated with TKI and developed TKI-resistance harboring the T790M mutation. As shown in Table 1, the median patient age was 59.5 years. The median TTF was 11.5 months. Among them, 3 patients have EGFR L858R mutation while 5 patients have exon 19 deletion. Plasma samples were collected before osimertinib treatment (baseline) and at the initial point of the osimertinib-resistance (resistance). A total of 8 age- and sex-matched healthy controls were recruited for this study. Plasma exosomes were isolated using ExoQuick as outlined in Materials and Methods. Isolated exosome morphology was visualized using the transmission electron microscopy. In each group, exosomes displayed a spherical vesicular shape with diameters less than 200 nm (Fig. 1A). Exosome size was further confirmed by nanoparticle tracking analysis. The median size of exosomes derived from the plasma of lung cancer patients at baseline and the resistance stage were approximately 142.7 ± 1.4 nm and 140.8 ± 2.2 nm, respectively (Fig. 1B). The median size of exosomes derived from the plasma of healthy controls was 110.1 ± 1.4 nm. Exosome characterization was further confirmed via western blot to probe for the expression of exosomal protein markers. As shown in Fig. 1C, all exosome samples were positive for exosomal transmembrane protein markers (CD81, Flotillin-1) and a cytosolic protein marker (Alix). These results suggest that the isolated vesicles from all samples were indeed exosomes.

OPEN IN VIEWER

The plasma exosomal miRNA profiles of lung cancer patients before osimertinib treatment (baseline) and after the osimertinib resistance stage, as well as healthy controls, were analyzed by the NGS Illumina platform. On average, 13.1% reads across all samples (21–24 million mapped reads) mapped to the miRNAs in the human GRCh38 genome (Supplement Fig. S1). Among these, 2,409 miRNAs were characterized as known miRNAs, while 5,680 potential miRNAs were unknown. Differential miRNA expression analysis on the mapped miRNAs was analyzed with DESeq2. To remove the background from non-tumor cells, we compared the plasma exosomal miRNA profiles from lung cancer patients with healthy controls. As shown in Fig. 2A, 136 exosomal miRNAs derived from baseline lung cancer plasma were differentially expressed relative to healthy control miRNAs ( 0.05). Among these, 26 miRNAs were upregulated, and 110 were downregulated (Supplement Table S1). Likewise, the expressions of 164 plasma exosomal miRNAs from patients at the resistance stage were differentially expressed from the healthy control miRNAs ( 0.05) (Fig. 2B). Among these, 50 and 114 miRNAs were upregulated and downregulated at the resistance stage when compared to those of healthy control miRNAs, respectively (Supplement Table S2). Furthermore, we found that only 27 miRNAs were differentially expressed between baseline and the resistance stage (Fig. 2C). Moreover, 20 miRNAs were significantly upregulated at the resistance stage, whereas 7 miRNAs were significantly downregulated ( 0.05) (Fig. 2D). Exosomal miRNAs that were differentially expressed in each of the three pairs of comparison are illustrated in a Venn diagram (Fig. 2E). Each circle represents the number of differentially expressed exosomal miRNAs between 2 conditions (baseline vs healthy control = 136 miRNAs, resistance vs healthy control = 164 miRNAs and resistance vs baseline = 27 miRNAs). At the resistance stage, we found that 9 miRNAs were significantly upregulated and 1 miRNA was downregulated compared to both baseline and healthy controls, respectively (Supplement Table S3). Therefore, our results indicate that these 10 miRNAs might play a significant role in osimertinib resistance-related mechanisms and have potential as molecular biomarkers used for detecting osimertinib resistance in lung cancer patients.

3.3 Dysregulated miRNAs in plasma exosomes of osimertinib resistance patients are associated with diverse cancer-related signaling pathways

To understand the biological functions of differentially expressed miRNAs in lung cancer patients, we predicted target genes of the 10 lung cancer-dysregulated exosomal miRNAs using the miRSystem database. We found that 1,175 putative targets were regulated by these 10 miRNAs (Supplement Table S4). Pathway analysis demonstrated that many diverse pathways were associated with these dysregulated miRNAs (adjust p -value 0.05) (Fig. 2F). Many of them have been reported to be involved with cancer progression, such as the hedgehog signaling pathway, PI3K-Akt signaling pathway, Wnt signaling pathway, and non-small cell lung cancer pathway. In addition, Gene Ontology (GO) analysis was employed to evaluate the relative representations of different biological functions. We found that these genes belong to several cellular components and are involved with several biological processes (Fig. 2G). Taken together, these results show that exosomes derived from lung cancer patients are packed with highly regulatory miRNAs. These dysregulated miRNAs might account for the mechanisms underlying osimertinib resistance.

3.4 Validation of the expression of candidate exosomal miRNAs

To explore the potential of these dysregulated miRNAs as useful biomarkers for osimertinib resistance, we compared the expression of these 10 lung cancer-dysregulated exosomal miRNAs of 19 lung cancer patients at the baseline and the resistance stages (Table 1). This sample group consisted of 6 males and 13 females with median age of 67 years old. The median TTF was 12 months. Among them, 7 patients have EGFR L858R mutation while 12 patients have

OPEN IN VIEWER

exon 19 deletion. Of these 19 NSCLC patients, all except one had acquired T790M mutation after received 1 st or 2 nd generation EGFR-TKI. There was only one female whom had de novo L858R/T790M mutation. She got initial treatment with afatinib and had no tumor response. Osimertinib was prescribed for second line therapy in this patient. Using quantitative RT-PCR, we found that 6 exosomal miRNAs (hsa-miR-323-3p, hsa-miR-325, hsa-miR-1468-3p, hsa-miR-4433b-3p, hsa-miR-5189-5p, and hsa-miR-6513-5p) were significantly upregulated at the resistance stage relative to baseline (Fig. 3). These miRNAs may be regulatory factors that underlie resistance mechanisms and, thus, be potential resistance biomarkers. Next, a ROC curve was constructed to explore the predictive potential of the verified exosomal miRNAs for osimertinib resistance (Fig. 4). We found that measurement of hsa-miR-323-3p expression was successful in distinguishing between the baseline and resistance stages of osimertinib treatment with a sensitivity of 73.68%, a specificity of 89.47%, and an AUC of 0.878 ( 0.0001). Additionally, hsa-miR-1468-3p achieved 63.16% sensitivity, 94.74% specificity, and 0.886 AUC ( 0.0001); hsa-miR-5189-5p achieved 57.89% sensitivity, 94.74% specificity, and 0.928 AUC ( 0.0001); hsa-miR-6513-5p achieved 78.95% sensitivity, 94.74% specificity, and 0.934 AUC ( 0.0001). However, expression levels of hsa-miR-325 and hsa-miR-4433b-3p were not able to provide a marker for differentiating between baseline and osimertinib-resistance in NSCLC patients with AUCs of 0.715 ( p = 0.0832) and 0.637 ( p = 0.1485), respectively. These results indicate that 4 exosomal miRNAs (hsa-miR-323-3p, hsa-miR-1468-3p, hsa-miR-5189-5p, and hsa-miR-6513-5p) have the potential to be developed as molecular biomarkers for detecting osimertinib resistance in NSCLC patients.

To investigate the relationship between exosomal miRNA upregulation and clinical prognosis, survival analysis was performed on all 27 osimertinib-treated NSCLC patients (Table 1). TTF was defined as time since the initiation of osimertinib until stopping the drug due to disease progression or treatment toxicities. OS was defined as time since the initiation of osimertinib until death from any causes. The upregulation of our 4 previously identified miRNAs (hsa-miR-323-3p, hsa-miR-1468-3p, hsa-miR-5189-5p, and hsa-miR-6513-5p) was observed in 17 patients (Supplement Table S5). As shown in Fig. 5A, the median TTF of patients who had upregulation of all these 4 miRNAs was longer compared to the rest of the patients who did not have all 4 miRNAs upregulation (11.9 months vs. 8.8 months, HR, 0.72; 95% CI, 0.32–1.61, p = 0.43), as well as longer OS although it was not statistically significant (15.4 months vs. 11 months, HR, 0.86; 95% CI, 0.39–1.9, p = 0.71) (Fig. 5B). These data suggest that upregulation of hsa-miR-323-3p, hsa-miR-1468-3p, hsa-miR-5189-5p, and hsa-miR-6513-5p tended to associate with both TTF and OS. Moreover, the 10 lung cancer-dysregulated exosomal miRNAs were found to be upregulated in 6 out of 27 patients. The median TTF of these 6 patients was also longer compared to the other 21 patients in sample group (12 months vs 8.8 months, HR, 0.89; 95%CI, 0.35–2.27, p = 0.816) (Fig. 5C). Consistently, the median OS of the 6 patients with higher expression of the 10 candidate exosomal miRNAs was higher (16.2 months) compared to the other 21 patients (11.7 months) (HR, 0.71; 95%CI, 0.28–1.79, p = 0.47) (Fig. 5D). Although we found no statistical significance, lung cancer patients with upregulated candidate miRNAs tended to have longer TTF and OS than the patients without this upregulation.

OPEN IN VIEWER

OPEN IN VIEWER