The gene expression profiles of the GSE114564 dataset were obtained from the Gene Expression Omnibus database (www.ncbi.nlm.nih.gov.libproxy.ajou.ac.kr/geo/); gene expression profiles were analyzed with the GEO2R tool, using high-throughput sequencing to investigate the expression of 14 candidate reference genes in patients with different liver disease statuses. A heatmap of the reference genes was generated using the heatmap visualization tool Morpheus (https://software.broadinstitute.org/morpheus/). Suitable reference gene candidates for analyzing gene expression in HCC were identified using the list of housekeeping genes at genomics-online (https://www.genomics-online.com/resources/16/5049/housekeeping-genes); the gene accession numbers were obtained through the NCBI BLAST database (Table I). Kruskal-Wallis (non-parametric) followed by Dunn's post hoc test was used to determine statistical significance between non-tumor (normal, chronic hepatitis and liver cirrhosis) and HCC groups (early and advanced HCC). P<0.05 was considered to indicate a statistically significant difference.

The Exocarta database (http://www.exocarta.org) is a manually curated web-based overview of exosomal proteins, RNA and lipids. Exocarta, which is used to evaluate corresponding data, such as exosome characterization and molecular properties, was used to identify reference genes expressed in exosomes (28).

Sera and tissue samples were collected from the Biobank of Ajou University Hospital, a member of the Korea Biobank Network, between April 2015 and July 2019. Written informed consent was obtained from all study participants. Serum samples were collected from 20 healthy controls and 20 patients with HCC; 20 pairs of HCC tissues with 20 corresponding non-tumor tissue samples were also obtained from patients undergoing tumor resection surgery. These samples were immediately frozen in liquid nitrogen until use. Healthy controls were subjects 18 years of age or older without a history of viral hepatitis or alcoholic liver disease who visited the Ajou University Hospital for the purpose of regular health checkups. HCC was diagnosed based on the American Association for the Study of Liver Diseases practice guideline (29) or histopathologic findings. Subjects were excluded if they exhibited any evidence of other malignancy except HCC or viral coinfections with the human immunodeficiency virus. The patient clinical characteristics are presented in Table SI. All experiments were performed according to the Declaration of Helsinki and the study protocol was approved by the Institutional Review Board of Ajou University Hospital, Suwon, South Korea (approval no. AJRIB-BMR-KSP-16-365 and AJIRB-BMR-SMP-17-189).

To evaluate exosomes, Huh7 cells from the Korean Cell Line Bank were cultured in Dulbecco's modified Eagle's medium (GenDEPOT, LLC) supplemented with 10% fetal bovine serum (Invitrogen; Thermo Fisher Scientific, Inc.) and 1% penicillin-streptomycin (GenDEPOT, LLC). The cells were incubated at 37°C in a humidified atmosphere containing 5% CO₂.

Blood samples (5 ml each) were collected from 20 patients directly into serum collection tubes. The whole blood samples were centrifuged at 1,800 × g at room temperature for 10 min, and the resultant sera were aliquoted into 1.5 ml tubes. The samples were then centrifuged at 3,000 × g at 4°C for 15 min to remove cell debris prior to use.

Exosomes were isolated from human serum samples using ExoQuick (System Biosciences, LLC) according to the manufacturer's instructions (2).

Exosome presence and size were confirmed using TEM. Serum exosome samples were fixed with 2% glutaraldehyde and 4% paraformaldehyde for 2 h at room temperature, and then treated with 0.4% uranyl acetate at 4°C for 10 min. Thereafter, the exosomes were observed using a Sigma 500 electron microscope (Zeiss GmbH), and further examined using a NanoSight NS300 instrument (Malvern Panalytical Ltd.) equipped with a 405-nm laser, to determine the size and quantity of the isolated particles. A 60-sec video was generated at a frame rate of 30 frames/s, and particle movement was analyzed using NTA software (version 3.0, Malvern Panalytical, Ltd.). Each sample was analyzed three times and the average number of counts were used.

To validate the expression of exosomal protein markers, serum exosomes and Huh7 cell lysates were lysed in RIPA lysis buffer (Thermo Fisher Scientific, Inc.) containing the Halt protease inhibitor cocktail (Thermo Fisher Scientific, Inc.). Total protein concentration was quantified using the bicinchoninic acid assay method (Thermo Fisher Scientific, Inc.); equal amounts (10 µg) of protein sample were separated with 10% gel, and then transferred onto polyvinylidene difluoride membranes (MilliporeSigma). The membranes were blocked with 5% non-fat milk (in Tris-buffered saline and 0.1% Tween-20) for 1 h at room temperature, and then incubated with the following primary antibodies: Mouse anti-Alix (1:1,000; cat. no. sc53538; Santa Cruz Biotechnology, Inc.), mouse anti-CD81 (1:250; cat. no. 10630D; Invitrogen; Thermo Fisher Scientific, Inc.), rabbit anti-CD9 (1:2,000; cat. no. ab92726; Abcam) and mouse anti-BiP/GRP78 (1:1,000; cat. no. 610979; BD Biosciences). The resulting immune complexes were then probed using secondary horseradish peroxidase-conjugated anti-rabbit (cat. no. BR170-6515; Bio-Rad Laboratories, Inc.) or anti-mouse (cat. no. BR170-6516; Bio-Rad Laboratories, Inc.) antibodies. Luminescence was observed using the ChemiDoc™ Imaging System (Bio-Rad Laboratories, Inc.).

The NCBI BLAST database (https://blast.ncbi.nlm.nih.gov/Blast.cgi) was used for primer design. All primers were designed with target amplicons <200 bp in length. The primer sequences are listed in Table II. The specificity of these primer sets was confirmed using melting curve analysis (Fig. S1A).

Total RNA from the selected tissue samples was isolated using QIAzol reagent (Qiagen GmbH), and serum RNA was extracted from the selected blood samples using the TRIzol^® LS reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Exosomal RNA was isolated from serum using the SeraMir™ Exosome RNA Amplification kit (System Bioscience, LLC) according to the manufacturer's instructions. RNA concentration was quantified using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Inc.). Following the manufacturer's instructions, serum RNA (500 ng) was reverse transcribed into cDNA using the PrimeScript™ RT Master Mix (Takara Bio, Inc.), and exosomal RNA (50 ng) was reverse transcribed using the miScript II RT kit (Qiagen GmbH).

qPCR was performed using the amfiSure qGreen Q-PCR Master Mix (GenDEPOT, LLC) according to the manufacturer's instructions, on the CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratories, Inc.). Each sample was prepared in a total volume of 10 µl, containing 4 µl diluted cDNA template, 5 µl amfiSure qGreen Q-PCR Master Mix (GenDEPOT, LLC), and 500 nM of each primer. The PCR conditions were as follows: 95°C for 2 min, 40 cycles of 95°C for 15 sec, 58°C or 60°C for 34 sec, and 72°C for 30 sec, followed by a dissociation stage of 95°C for 10 sec, 65°C for 5 sec, and 95°C for 5 sec. Relative gene expression levels were calculated using the 2^−ΔΔCq method (30). All PCR reactions were performed in triplicate.

The stability of candidate reference gene expression was evaluated using the Excel-based software BestKeeper (https://www.gene-quantification.de/bestkeeper.html). All data processing was based on crossing point (CP). The stability rankings of the individual genes were determined according to the lowest standard deviations.

All experiments were performed independently in triplicate. Results are presented as the mean ± standard deviation or standard error of the mean. Statistical differences between groups were analyzed using paired Student's t-test for the tissue samples or Welch's t-test for the serum and serum exosome samples. All statistical analyses were performed using GraphPad Prism 5.0 (GraphPad Software Inc.) and P<0.05 was considered to indicate a statistically significant difference.

Expression levels of the 14 selected reference genes, analyzed using the next-generation sequencing multistage HCC RNA seq dataset GSE114564, are represented as a heat map based on liver disease status (Fig. 1A and Table SII). Differences in expression levels between the control group and the HCC group were identified in patients with different liver disease statuses. From the 14 genes, ACTB, GAPDH, HMBS, PPIA, RPLP0 and TBP were selected, as they did not show a statistically significant difference between the control and HCC groups (Fig. 1B and Table SII). For the exosome samples, the Exocarta database (http://www.exocarta.org/) was used to identify suitable reference genes from the five selected genes. TBP, which is not registered in the Exocarta database, was excluded from the final selected candidates.

The ACTB gene performs key functions of the cytoskeleton, such as cell motility and contraction (31). The GAPDH gene has glyceraldehyde-3-phosphate dehydrogenase and nitrosylase activities, and is involved in glycolysis and nuclear function. It also regulates the organization and assembly of the cytoskeleton (32,33). The HMBS gene supports the generation of hydroxymethylbilane synthase, and is indirectly involved in the production of heme (34). The PPIA gene catalyzes the cis-trans isomerization of proline imidic peptide bonds in oligopeptides, and is involved in apoptosis signaling through NF-κΒ, AKT1 and BCL2 upregulation (35,36). The RPLP0 gene encodes a ribosome, an organelle that catalyzes protein synthesis, is composed of a small 40S and a large 60S subunit, and is associated with pathologies including Chagas disease (37). Based on these results, the expression levels of 6 genes exhibited no statistical significance between the control and HCC groups. Among them, 5 genes were expressed in exosomes using the Exocarta database. The present study subsequently identified the molecular characteristics of those 5 candidate reference genes.

Following primer design using NCBI BLAST, and confirmation of specificity using melting curve analysis, all primers were observed as a single peak (Fig. S1A). The most suitable annealing temperature and mean Cq values were then selected (Fig. S1B).

Pure exosomes were identified by isolation from serum samples and characterization using TEM analysis (Fig. S2A). Furthermore, positive and negative protein markers of extracellular vesicles were confirmed through western blotting (Fig. S2B). Next, RT-qPCR analysis was used to evaluate the expression levels of the selected genes in the control and HCC groups. All samples were analyzed in triplicate, and Welch's t-test was performed with the average Cq values for each group. First, Cq values of the five selected reference genes were calculated in 20 healthy and 20 HCC tissues. The expression levels of PPIA (P=0.0076) showed the lowest significant difference between the control and HCC tissue groups, and the expression levels of ACTB (P=0.0011), GAPDH (P=8.92E-05), HMBS (P=0.0003), and RPLP0 (P=0.0003) indicated a more significant difference. (Fig. 2A). Next, Cq values of the selected reference genes in serum and serum exosome samples were estimated. Unlike the tissue samples, the expression levels of ACTB (P=0.0837), HMBS (P=0.0904), PPIA (P=0.2238) and RPLP0 (P=0.8058) showed no significant difference. However, similar to the tissue the samples, GAPDH (P=0.0233) indicated a significant difference in expression level between the control and HCC groups (Fig. 2B). Finally, the expression levels of the five reference genes were confirmed in exosomal RNA isolated from patient serum. Of these five genes, HMBS (P=0.0404) exhibited the least significantly different expression between the control and HCC serum exosome groups; however, the expression of ACTB (P=0.0001), GAPDH (P=0.0001), PPIA (P=0.0001) and RPLP0 (P=0.0001) indicated a substantially significant difference (Fig. 2C). Therefore, among the five reference genes identified, HMBS exhibited the least significant difference in expression between the control and HCC groups for blood samples (both serum and serum exosome).

BestKeeper analyses of the tissue, blood and serum samples were performed to investigate the stability of the five reference genes. Descriptive statistics of the derived CPs were calculated for each reference gene. CPs are direct results obtained from the threshold line crosses fluorescence plots for each of the samples. All CP data for all groups were compared throughout the study (38). Stability rankings for each sample were evaluated according to the coefficient of variance values of the BestKeeper analyses. As such, the most stable reference gene was identified to be HMBS. GAPDH, which is a commonly used reference gene, was found to be the least stable (Fig. 3A). In all 40 tissues and blood samples, HMBS had the most consistent CP values among five reference genes (Fig. 3B). The stability values obtained from the BestKeeper analyses are represented in Fig. 3C. Also, when performing NormFinder analysis (another tool for calculation of stability), HMBS exhibited the highest stability in tissue samples among the five candidate reference genes (data not shown). In conclusion, HMBS was selected as the most stable reference gene in tissue, serum and serum exosomes based on Bestkeeper, a software that identifies the suitable reference gene (39). Additionally, NormFinder analysis revealed that HMBS is the most stable reference gene for tissue samples.

In the present study, we found that HMBS is the most suitable reference gene for blood and tissue samples in HCC. This study will be helpful for future studies by finding suitable reference genes for RT-qPCR, used to detect gene expression widely. In this respect, we suggest that the expression stability of reference genes should be validated to obtain accurate and reliable results.