Participants
This study was approved by the Clinical Research Ethics Committee of Shanghai Changhai Hospital (Shanghai, China) (no. CHEC2020-112). All of the clinical samples were obtained from Shanghai Changhai Hospital. Written informed consent was obtained from the participants before sampling. In this study, 488 participants, including localized ccRCCs (n = 226), patients with benign renal masses (n = 73), and healthy controls (n = 189), were recruited consecutively from August 2017 to July 2020. The renal masses were confirmed by surgical pathology and examined by two independent pathologists. Tumor stage and pathological grade were estimated according to the 8th TNM criteria proposed by the American Joint Committee on Cancer (AJCC) in 2017. Inclusion criteria for patients with renal masses were as follows: (1) age 20–80 years old; (2) CT scans prior to surgery; (3) a definite diagnosis by pathology; (4) ccRCC participants with clinical tumor stages I and II (localized renal carcinoma); (5) received surgical operation, did not undergo any anticancer treatment before sampling; (6) no history of other cancers; and (7) signed informed consent. Inclusion criteria for healthy controls were as follows: (1) age 20–80 years old, (2) underwent a health check-up and considered asymptomatic and healthy, (3) ultrasonic examination showed no mass, and (4) signed informed consent. The exclusion criteria were as follows: (1) previously diagnosed renal masses, (2) received ablation therapy, and (3) with other malignant tumors.
Study design
This study consisted of four phases: discovery, test, training, and validation. The participants in the discovery, training, and validation phases were consecutively enrolled, and they were randomly assigned into these groups. The participants in the test phase were randomly selected from the training phase. An overview of the workflow is summarized in Fig. 3. The clinical characteristics of the participants are summarized in Table 1. In the discovery phase, we investigated circulating emRNA profiling in ccRCCs (n = 12) and healthy controls (n = 22) by RNA-seq. Dysregulated (p < 0.05, fold change > 2 or < 0.5, FDR < 0.05) emRNAs in ccRCC were identified. The seven top upregulated emRNAs that were also related to ccRCC or/and multiple malignant tumors were selected as candidate biomarkers for further training and validation (see details of ‘The selection strategy of candidate biomarkers’ in Additional file 3: Supporting information). In the test phase, the expression levels of the candidate emRNAs were evaluated in localized ccRCCs (n = 16) and healthy controls (n = 20) by RT–qPCR. In the training phase, the expression levels of the significant emRNAs identified in the test phase were evaluated in another cohort of localized ccRCCs (n = 92) and healthy controls (n = 50) by RT–qPCR. A stepwise logistic regression analysis was used to identify the significant predictors and establish an emRNA-based ccRCC signature to differentiate ccRCCs from healthy controls. The signature was then validated in an additional cohort of localized ccRCCs (n = 106) and healthy controls (n = 97). In addition, we evaluated the diagnostic performance of the candidate emRNAs in distinguishing ccRCCs from patients with benign renal masses. In this phase, patients with benign renal masses (n = 73), including solid renal masses (n = 47) and cystic renal masses (n = 26), were included (the detailed clinical characteristics are summarized in Additional file 2: Table S1). Likewise, we applied stepwise logistic regression analysis to identify the significant predictors and establish an emRNA-based signature to differentiate ccRCCs from patients with benign renal masses, including solid and cystic masses.
Sample collection and processing
The patients with renal masses signed informed consent on the first day of admission, and their fasting peripheral blood was collected on the second morning. Healthy controls signed informed consent on the day of health check-up and their fasting peripheral blood was collected before the physical examination. Samples were stored in a 4 °C refrigerator and transported to the laboratory with ice. Blood samples were allowed to clot at room temperature for a minimum of 30 min and a maximum of 2 h. All samples were then centrifuged at 1600 × g for 15 min to separate the serum, and the serum (supernatant) was collected in 1.5-ml centrifuge tubes and numbered. The serum samples were immediately stored at − 80 °C until further processing.
Exosome isolation
An exoEasy Maxi Kit (No. 76064, Qiagen, Dusseldorf, North Rhine-Westphalia, Germany) was used to purify exosomes from serum according to the manufacturer’s instructions. First, the serum was filtered through a 0.22-μm filter membrane, an equal volume of XBP buffer wad added, and then it was gently inverted and mixed 5 times. After mixing, the above sample mixture was added to the adsorption column and centrifuged at 500 × g for 1 min at room temperature. The flow-through solution was discarded from the adsorption column. This procedure was repeated once. Then, approximately 5 ml of XWP buffer was added to the adsorption column for cleaning and centrifuged at 5000 × g for 5 min at room temperature. The adsorption column was discarded, and the serum exosomes were immobilized on the adsorption column membrane. After that, the adsorption column was placed into a new collection tube, and 400 μl XE buffer was added to the column for elution, then incubated at room temperature for 1 min, and centrifuged at 500 × g for 5 min. Finally, the eluate in the collection tube was added back into the adsorption column and incubated for 1 min at room temperature. The eluate was centrifuged at 5000 × g for 5 min and transferred to a 1.5-ml centrifuge tube. Exosomes could be used for further research or stored at − 80 °C.
Quality control of exosome isolation and verification
Transmission electron microscopy (TEM)
The purified exosome sample was diluted 100-fold with PBS (SH30256.01, HyClone, Logan, UT, USA) and used for electron microscopy (JEM-1400, JEOL, Akishima, Tokyo, Japan). A total of 20 μl of the diluted sample was pipetted, dropped to the center of a copper mesh, and allowed to stand for 20 min. Then, the liquid was blotted with filter paper to prepare the negative stain. Negative staining was performed with 1% phosphotungstic acid for approximately 10 s, and filter paper was used to blot the excess liquid. Then, 20 μl ddH2O was carefully added to the center of the copper mesh, and after 20 s, the remaining liquid was aspirated from the mesh with filter paper. The finished copper mesh was ready for exosome identification. TEM showed the presence of exosomes as rounded, biconcave-disk shaped, vesicle-like structures (Additional file 1: Fig. S1A).
Nanoparticle tracking analysis (NTA)
The extracted exosomes were diluted 1:200 (filtered PBS). The module detected by the Nano Sight 300 (Malvern Instruments Ltd., Malvern, UK) was washed with ddH2O 3 times, and then the module was purged with syringe pumping air 7–8 times to remove as much residual fluid as possible. One milliliter of the well-mixed and diluted exosome sample was aspirated into the module with a syringe, and the air bubbles in the module detection chamber were ejected. The number of images capturing repetitions was set to 3, and the duration of each capture was set to 60 s. NTA showed that the peaks of circulating exosomes from RCC, patients with benign masses, and healthy controls were 74 nm, 77 nm, and 79 nm, respectively, ranging from 50 to 200 nm (Additional file 1: Fig. S1B).
Western blot (WB)
A total of 50 μl RIPA lysis buffer (P0013C, Beyotime, Shanghai, China) and 0.5 μl of protease inhibitor were added to the extracted exosomes. The mixture was centrifuged at 15,000 × g for 20 min at 4 °C. The supernatant was collected, and the concentration was measured using a BCA kit (5,000,112, Bio–Rad, Hercules, CA, USA). Next, 5 × loading buffer (P0015, Beyotime, Shanghai, China) was added to the samples. Those samples were incubated in a 99 °C metal bath (Thermomixer comfort, Eppendorf, Hamburg, Germany) for 15 min. A PAGE Gel Fast Preparation Kit (PG112, EpiZyme, Shanghai, China) was used to make a 10% separating glue and 5% concentrated glue according to the manufacturer’s instructions. Then, 30-μg exosome protein samples were added to each well, and 5-μl protein marker (P0077, Beyotime, Shanghai, China) was added to the blank well. The machine (165–8029, Bio–Rad, Hercules, CA, USA) was set to 120 V, and electrophoresis was performed for 2 h. After electrophoresis was completed, and the gel was removed, stacked with a PVDF (IPFL00010, Millipore, Burlington, MA, USA) membrane, placed in a transfer tank, and filled with transfer buffer. The instrument was set to 300 mA for 2 h. After, the transfer was completed, the membranes were blocked with 5% BSA for 2 h at room temperature, and the membranes were cut at appropriate positions. Then, antibodies against CD9 (1:1000; AP1482D, ABGENT, Suzhou, Jiangsu, China), CD63 (1:500; AP5333B, ABGENT, Suzhou, Jiangsu, China), CD81 (1:4000; AM8557B, ABGENT, Suzhou, Jiangsu, China), TSG101 (1:2000; AM8662b, ABGENT, Suzhou, Jiangsu, China), and β-actin (1:5000; A5441, Sigma, St. Louis, MO, USA) were added and incubated at 4 °C overnight. The next day, the membranes were placed at room temperature for 15 min, and the primary antibody was removed. The membranes were washed 6 times with TBST for 5 min each time. Then, secondary antibodies were added, incubated for 2 h at room temperature, and washed 6 times with TBST for 5 min each time. Clarity Max Western ECL Substrate (P0018 FS, Beyotime, Shanghai, China) was added. Blots were imaged by an imaging device (e-Blot, Shanghai, China). WB showed that exosomal biomarkers, including CD9, CD63, CD81, and TSG101, were detectable in isolated exosomes (Additional file 1: Fig. S1C).
Circulating exosomal RNA purification and sequencing
Circulating exosomal RNA was purified using an exoRNeasy Serum/Plasma Maxi Kit (217,084, Qiagen, Dusseldorf, North Rhine-Westphalia, Germany). Total RNA sequencing utilized 0.25–50 ng of RNA as the input. gDNA removal was performed by adding HL-dsDNase (70,800–202, ArcticZymes, Tromsø, Norway) and reaction buffer (66,001, ArcticZymes, Tromsø, Norway). Libraries for total RNA sequencing were prepared using SMARTer Stranded Total RNA-Seq Kit v2-Pico Input Mammalian (634,413, Clontech, Mountain View, CA, USA). Compared to the manufacturer’s protocol, the fragmentation step was set to 2 min at 94 °C; hereafter, the option to start from highly degraded RNA was followed. Library preparation also included cDNA synthesis, 5 cycles of indexing PCR, ribosomal cDNA depletion, and 9–16 cycles of enrichment PCR. Each library was measured for size with an Agilent High Sensitivity DNA Kit (5067–4626, Agilent, Santa Clara, CA, USA) and concentration with a Library Quantification Kit (638,324, Clontech, Mountain View, CA, USA). Alternatively, libraries can be quantified by a Qubit dsDNA HS kit (Q32854, Thermo Fisher Scientific, Waltham, MA, USA). Libraries were combined into an equimolar pool, which was measured for size and concentration, and libraries were sequenced using the HiSeq 2500 platform (Illumina, San Diego, CA, USA). To ensure quality, 10-Gb raw data per library was necessary.
Sequencing data processing
The sequencing data were quality control processed by FastQC and trimmed by Trimmomatic with the following parameters: minimum length 30, sliding window 4, and required quality 15. Then, the clean data were mapped to the reference human genome (GRCh38.p13) by STAR with the Ensembl release-95 gene annotation database. Finally, we used htseq for expression quantification, and the R package Deseq2 to identify the differentially expressed genes between different groups. In order to reduce the false-positive rate of dysregulated genes, we performed multiple tests by Benjamini–Hochberg to obtain an adjusted p value.
DNA gel electrophoresis
Validation of PCR products was performed using DNA gel electrophoresis. First, 60 ml TAE and 1 g solid agarose were mixed and heated until completely transparent. When warm, 6 µl nucleic acid stain was added, shaken well, and poured into the glue making rack. The comb teeth were inserted until they were completely solidified, and then the comb teeth were removed. The agarose gel was placed into the electrophoresis tank, 1 × TAE solution was added until the liquid level was over the gel, and the air bubbles in the upper sample well were discharged. The samples were loaded. The electrophoresis conditions were typically set to 120 V for 30 min. After running, an image of the gel was obtained.
Reverse transcription and real-time quantitative polymerase chain reaction (PCR)
Circulating exosomal RNA was extracted and purified using an exoRNeasy Serum/Plasma Maxi Kit (217,084, Qiagen, Dusseldorf, North Rhine-Westphalia, Germany). RNA was reverse transcribed into cDNA according to the instructions of the PrimeScript™ RT reagent Kit (RR047A, Takara, Kyoto, Japan). The operation should be performed on ice, and the Master Mix should be prepared in an amount twice the number of reactions. Then, 10 µl should be dispensed into each reaction tube. The reverse transcription reaction conditions were as follows: 37 °C for 60 min, 85 °C for 5 s, and cooling on ice before use. The synthesized cDNA could be used immediately for subsequent qPCRs or temporarily stored at − 20 °C.
The primers and probes are presented in Additional file 2: Table S2. An ABI 7500 fluorescent PCR instrument (Applied Biosystems, Foster City, CA, USA) was used. Using 20 μl of the qPCR system, specifically 10 μl of 2 × qPCR Master Mix and 2 μl of cDNA solution, the working concentration of the upstream and downstream primers was 10 μM, the probes were used at a concentration of 5 μM, and the reaction was brought up to 20 μl with RNase-free water. The qPCR procedure was set as follows: 37 °C for 5 min, 95 °C for 10 min, and 40 cycles of 95 °C for 15 s and extension at 59 °C for 1 min.
EmRNA quantification
We applied the standard-curve quantitation method for emRNA quantification. This method is similar to previous approaches used for other RNA types [44]. Briefly, we synthesized the amplified fragments of each target gene. During the reverse transcription of these genes, we generated a standard curve using real-time quantitative PCR by testing synthesized transcripts at different copy number concentration gradients. By this means, we could calculate the copy numbers of target genes relative to a standard sample. Specifically, an initial amount of 400 µl serum was defined to extract RNA from exosomes. Then, the extracted RNA was dissolved in RNase-free water. Then, 20 µl of emRNA was processed for reverse transcription to yield 60 µl of cDNA. Then, we added 2 µl of cDNA to 20 µl of the fluorescence PCR system, with triplicate samples accessed to yield the mean CT value. At the same time, we synthesized the target RNA transcripts at copy numbers of 103, 104, 105, 106, and 107. These synthesized mRNAs were processed using the same procedure as the abovementioned transcripts (20 µl emRNA to 60 µl cDNA and 2 µl cDNA to 20 µl PCR system). These synthesized mRNAs yielded a standard curve, and we could calculate the copy number of the target emRNA by matching the CT value to the standard curve.
Statistical analysis
Baseline analysis of the clinical characteristics of the population enrolled in this study was processed by SPSS 21.0 (IBM Corporation, Armonk, New York, USA). The measurement data were tested for normality and variance homogeneity test, and the independent samples that met the normal distribution and were tested by t test in group design (the t' test was used for variance nonhomogeneity), and the Mann–Whitney U test was used if they did not meet the normal distribution; paired data were tested for normality of the mean difference, using the paired t test for normal distribution and the Wilcoxon signed rank test for nonnormality. The chi-square test or Fisher’s exact probability test was used for the count data.
Experimental results, such as comparative analysis of gene expression, one-way analysis for graphing, and calculation of p value, were performed by GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). Z-score (z = (x − μ)/σ; μ = average(); σ = stdevp()) was applied in data pre-processing to normalize the data. MedCalc 18.0 software (MedCalc Software, Ostend, Belgium) was used to establish a clinical diagnostic model by logistic regression and calculated the area under the curve (AUC), sensitivity, and specificity. The differences were considered statistically significant at p < 0.05. The area under the ROC curve values ranged from 0.5 to 1, with low diagnostic value between 0.5 and 0.7, moderate diagnostic value between 0.7 and 0.9, and high diagnostic value above 0.9.