Identification of exosomal miRNAs associated with the anthracycline-induced liver injury in postoperative breast cancer patients by small RNA sequencing

Study participants

We recruited 7 postoperative breast cancer patients at the National Cancer Center/Cancer Hospital, Chinese Academy of Medical Sciences (Beijing, China), from September 2017 to January 2018. Patients with any of the following conditions or previous treatments were ineligible: previous invasive breast cancer; non-breast cancer within 5 years before randomization, with the exception of carcinoma in situ of the cervix or colon, melanoma in situ, and skin basal-cell or squamous-cell carcinomas; any previous chemotherapy or radiotherapy for cancer; abnormal renal or hepatic function; and concurrent diseases interfering with planned laboratory tests, such as diabetes, hypertension, etc. Only those participants who received 4 cycles (every 3 weeks) of epirubicin or doxorubicin plus cyclophosphamide, followed 4 cycles (every 2 weeks) of paclitaxel were eligible for this study. The cumulative dose of epirubicin exceeded 320 mg/m². Written informed consent was obtained from all patients before enrollment, and the study was approved by the institutional ethical committee. National Cancer Center/Cancer Hospital, Chinese Academy of Medical Sciences granted Ethical approval to carry out the study within its facilities (Ethical Application Ref: 17-223/1774).

ALT detection and grouping of patients

In order to assess the liver function after chemotherapy, approximately 3ml of blood from each patient was collected in SST serum separation tubes (Becton Dickinson) and centrifuged at 3000 ×g for 10 min at 4 °C. The serum of all patients was analyzed for ALT using a Cobas c501 analyzer (Roche Diagnostics, Germany), and kits were procured by Roche. The criterion of liver injury was set as ALT>40U/L (Yu et al., 2017).

Sample collection and exosomes extraction

All of the whole blood samples from individuals were obtained and collected in K2EDTA tubes (Becton Dickinson) using standard venipuncture procedures. After centrifugation at 3,000 g for 15 min at 4 °C, the plasma was aspirated into micro tubes and stored at −80 °C fridge before use.

plasma-derived exosomes were isolated by differential centrifugation according to the protocol previously described (Théry et al., 2006). After thawing at 37 °C, plasma was centrifugated at 3,000 g at 4 °C for 15 min to remove cell debris. The supernatant was diluted by a seven-fold volume of phosphate-buffered saline (PBS), centrifuged at 13,000 g at 4 °C for 30 min, and filtered through a 0.22 µm filter to eliminate large particles. Then, the supernatant was ultracentrifuged using a P50A72-986 rotor (CP100NX; Hitachi, Brea, CA, USA) at 100,000 g at 4 °C for 2 h to pellet the exosomes. The pellet was resuspended in PBS and centrifuged again at 100,000 g at 4 °C for 2 h. After PBS washing, exosome pellets were resuspended in 100 µl PBS for TEM, NTA, and further research.

Exosomes characterize validation

The exosome pellets were observed and photographed by a transmission electron microscope (JEOL-JEM1400, Tokyo, Japan). Briefly, 10 µl exosomes solution was dropped onto a square copper grid and incubated for 10 min at room temperature. After washing with sterile distilled water, the exosomes were negatively stained with uranyl oxalate solution for 1 min. Then, the sample was dried under incandescent light for 2 min before viewing.

For the purpose of measuring size distribution and concentration, nanoparticle tracking analysis (NTA) was performed using the ZetaView PMX 110 (Particle Metrix, Meerbusch, Germany) and corresponding software NTA software (ZetaView 8.02.28) as recommended by the company.

Western blot analysis was performed to detect exosomal surface markers. Briefly, the exosome supernatant was denatured in sodium dodecyl sulfonate (SDS) buffer and then incubated with CD63, TSG101, Alix and Calnexin. The proteins were finally visualized with a Tanon4600 Automatic chemiluminescence image analysis system (Tanon, Shanghai, China).

MicroRNA sequencing and data analysis

Total RNA in plasma-derived exosomes was extracted and purified using a miRNeasy^® Mini kit (Qiagen, Cat. No. 217004) according to the manufacturer’s instruction. The degradation and contamination of RNA were monitored on 1.5% agarose gels. The concentration and purity of RNA were assessed using NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, DE) and RNA Nano 6000 Assay Kit of the Agilent Bioanalyzer 2100 System (Agilent Technologies, CA, USA). The concentration of RNA is more than 500 pg/ul in all samples. Library preparation and sequencing of miRNA were performed by Beijing ECHO BIOTECH Co. Ltd. In brief, after ligated with 3′ and 5′ adaptors, small RNAs were reverse-transcribed to cDNA and amplified by PCR. The products from PCR were sequenced on an Illumina HiSeq 2500 platform (Data S1). Considering that some miRNAs have little or no expression in some samples, only those with raw count value >3 in more than 75% of samples were retained for further analysis. The fold-change and P-value were calculated based on the miRNA counts using the package edgeR (Robinson, McCarthy & Smyth, 2010) of R and differentially expressed miRNAs were identified between groups of interest. P-value < 0.05 and —log2FC (fold change) —>1 were set as the cut-off criteria to screen out DE-miRNAs (Code S1).

Identification of AILI-related target genes

miRTarBase (Chou et al., 2018), an experimentally supported microRNA-target interactions database, was applied to predict the targets of the known miRNAs. The target genes of novel miRNAs were predicted using miRDB (Liu & Wang, 2019) and the screening criterion was set as score ≥80. Besides, we searched “drug-induced liver injury” in the Comparative Toxicogenomics Database (CTD, revision 15982) (Davis et al., 2019) and got the DILI-related genes (marker, mechanism or therapeutic genes of DILI). To narrow the research scope for further research, the intersection of predicted genes and DILI-related genes were identified as the AILI-related target genes of the DE-miRNAs.

Functional annotation and pathway enrichment analysis

Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis for AILI-related target genes of the DE-miRNAs were performed using the database for annotation, visualization and integrated discovery (DAVID 6.8) (Huang, Sherman & Lempicki, 2009). Only the GO terms and KEGG pathways with P < 0.05 were considered statistically significant.

PPI network and miRNA-hub gene network construction

To better understand the interaction among the AILI-related target genes, the protein-protein interaction (PPI) network was generated using the STRING database (Szklarczyk et al., 2015) and only the interactions with a combined score > 0.4 were considered significant. Then, the PPI network was visualized and analyzed using Cytoscape software (version 3.7.1) (Shannon et al., 2003). The Cytohubba plug-in of Cytoscape (version 0.1) (Chin et al., 2014) was used to identify important hub genes of the entire network. The maximal clique centrality (MCC) method was used to identifying hub objects in this study because this method could capture more essential proteins in the top-ranked list in both high degree and low degree proteins. After that, the miRNA-hub gene network was constructed by Cytoscape software.

Serum biochemical changes in patients

Seven postoperative breast cancer patients who underwent chemotherapy were recruited in our study. To assess liver function, serum ALT level was measured in all individuals before and after chemotherapy. According to the results of the serum ALT level (Fig. 1, Data S1), seven patients were divided into two groups: the non-liver injury group (n = 2) and the liver injury group (n = 5).

Verification of exosomes

After isolated from the plasma, exosomes were verified by TEM, NTA and western blot. The morphology of exosomes was visualized under a transmission electron microscope (TEM), and we found homogeneous cup-shaped vesicles with lipid bilayer membranes (Fig. 2A). We then performed a nanoparticle tracking analysis (NTA) to measure the size distribution of exosomes, and the results revealed the most widely distributed particle diameter was 122.9 nm (Fig. 2B). Additionally, exosomal surface markers were detected using western blot analysis. As shown in Fig. 2C and Data S2, exosome-positive markers (Alix, TSG101, and CD63) were identified, while the negative marker for exosomes (Calnexin) was absent in the isolated exosomes. These results demonstrated that exosomes were successfully isolated from the plasma.

Identification of DE-miRNAs and AILI-related target genes

We performed high throughput sequencing to detect the miRNA expression of exosomes and screened DE-miRNAs between non-liver injury group and liver injury group using edgeR with P < 0.05 and —log₂FC—>1. The results showed that a total of 2,031 miRNAs had been detected, including 1576 known miRNAs and 455 novel miRNAs. There are 30 DE-miRNAs were screened, of which 16 miRNAs (1 novel miRNA and 15 known miRNAs) were upregulated, and 14 miRNAs (1 novel miRNA and 13 known miRNAs) were downregulated. The volcano plots and heatmap of DE-miRNAs between the two groups are shown in Fig. 3. Then, we employed two miRNA-target interactions database, miRTarBase and miRDB, to predict the target genes of known and novel DE-miRNAs, respectively. As a result, 3823 predicted targets from miRTarBase and miRDB were identified as the target genes of 30 DE-miRNAs. Genes of DILI were downloaded from the CTD Database. A total of 433 DILI-related genes were recorded. To narrow the research scope, the intersection of predicted genes and DILI-related genes were identified as the AILI-related target genes of the DE-miRNAs. Finally, 79 AILI-related target genes were screened, including 55 and 36 targets for upregulated and downregulated DE-miRNAs, respectively (Fig. 3C).

Functional analysis of the AILI-related target genes

To evaluate the biological functions of these screened AILI-related target genes, GO annotation and KEGG pathway enrichment analysis were performed by using DAVID (Table S1).

Three functional categories were selected in GO annotation, including biological process (BP), cellular component (CC), and molecular function (MF). In the BP category, AILI-related target genes of upregulated miRNAs were significantly enriched in response to drug, positive regulation of MAPK cascade, and response to amino acid. In the CC category, these genes were significantly enriched in extracellular exosome, extracellular space, and cell–cell adherens junction. In addition, these genes were mainly enriched in growth factor activity, cadherin binding involved in cell–cell adhesion, and oxidoreductase activity in the MF category (Fig. 4A). Besides, AILI-related target genes of downregulated miRNAs were significantly enriched in cellular response to vascular endothelial growth factor stimulus, coronary vein morphogenesis, and response to lipopolysaccharide in the BP category. In the CC category, they were mainly enriched in extracellular exosome, myelin sheath, and platelet alpha granule lumen. Moreover, these genes were significantly enriched in enzyme binding, platelet-derived growth factor receptor binding, and chemoattractant activity in the MF category (Fig. 4B).

KEGG pathway analysis was further conducted for AILI-related target genes of DE-miRNAs. The most significantly enriched pathways of upregulated miRNAs included adherens junction, rheumatoid arthritis, and HIF-1 signaling pathway (Fig. 4A). The AILI-related target genes of downregulated miRNAs were enriched in some terms but none of them are significant (Fig. 4B).

PPI network and miRNA-hub gene network construction

The PPI network of AILI-related target genes was constructed using the STRING database (Fig. 5A). Then, we applied Cytohubba plug-in of Cytoscape to screen out the top 10 hub genes of the network using the MCC method (Table 1). In the PPI network, the hub genes were IL6, VEGFA, CCL2, HMOX1, IGF1, ACTB, CXCL1, SOD2, NOTCH1, and ARG1. Subsequently, the miRNA-hub gene network was constructed by Cytoscape software. As shown in Fig. 5B, the hub target genes are regulated by 6 DE-miRNAs and miR-1-3p regulates the most hub genes(n = 8). Our results suggest that miR-1-3p may play an important role in AILI.