SK-HEP-1 and HuH-7 cell lines, derived from human hepatoma, were purchased from the American Type Culture Collection (ATCC) (Rockville, MD, USA) and the Health Science Research Resources Bank (Osaka, Japan), respectively. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Nissui Pharmaceutical, Tokyo, Japan) containing 10% heat-inactivated fetal bovine serum (Thermo Fisher Scientific, Kanagawa, Japan), penicillin (100 U/mL), streptomycin (100 μg/mL), and sodium bicarbonate (1.5 g/L) at 37 °C in a humidified atmosphere with 5% CO2 in air.
Samples were obtained with written informed consent from 14 patients who underwent curative hepatectomy for HCC between August 2002 and November 2007 in the Department of Digestive Surgery and Surgical Oncology, Yamaguchi University Graduate School of Medicine, Japan. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a prior approval by the Institutional Review Board for Human Use at Yamaguchi University Graduate School of Medicine. Ten samples each were used for RNA-seq and quantitative real-time polymerase chain reaction (qRT-PCR); six samples were common to both the analyses.
Induction of sphere cells
Cells were suspended in the sphere inducing medium, which was based on a neural stem cell medium . This medium, used to induce floating sphere cells, was DMEM/Nutrient Mixture F-12 Ham supplemented with 0.6% glucose, 10 mM HEPES, 2 μg/mL heparin, 0.1 mg/mL transferrin, 25 μg/mL insulin, 60 μM putrescine, 30 nM sodium selenite, 20 nM progesterone, 10 ng/mL human recombinant epidermal growth factor (all from Sigma-Aldrich Japan, Tokyo, Japan), 10 ng/mL basic fibroblast growth factor (Merck Millipore, Tokyo, Japan), 10 ng/mL leukemia inhibitory factor (Merck Millipore), 60 μg/mL N-acetyl-L-cysteine (Sigma-Aldrich), and 1/50 volume NSF-1 (Lonza, Tokyo, Japan).
Total RNA was isolated with the miRNeasy Mini Kit (Qiagen, Tokyo, Japan). Sequencing libraries were constructed using the TruSeq Stranded Total RNA with Ribo-Zero Gold LT Sample Prep kit (Illumina, Tokyo, Japan) according to the manufacturer’s instructions. Sequencing of paired-end fragments (75 bp × 2) was conducted on a NextSeq 500 sequencing platform (Illumina).
After a quality control step, the filtered short reads were mapped to the reference genome (hg38) with STAR (version 2.5.1b) . Strand-specific counts of fragments from each sample were obtained using RSEM (version 1.3.3)  and normalized with the trimmed mean of M-values method  using the TCC package [25, 26]. The edgeR (version 3.28.1) [27, 28] package was used to identify the differentially expressed genes (DEGs) based on a false discovery rate q-value threshold < 0.05.
Quantitative real-time polymerase chain reaction (qRT-PCR)
The mRNA expression was examined by qRT-PCR as described previously . qRT-PCR was performed using a LightCycler 480 Probe Master (Roche Diagnostics, Tokyo, Japan) and Universal ProbeLibrary (Roche Diagnostics) probes or a LightCycler 480 SYBR Green I Master (Roche Diagnostics) on a LightCycler 480 System II (Roche Diagnostics). The primers and probes used are listed in Supplementary Table S1. Amplification was performed in a two-step procedure and mRNA levels were measured quantitatively using the Δ/Δ threshold cycle method. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and phosphoglycerate kinase 1 (PGK1) were used as controls. Triplicate wells were analyzed for each assay.
Western blot analysis
Cells were lysed and the proteins (10 μg) were separated by SDS-PAGE on an 8% gel and transferred onto a Poly (vinylidene fluoride) (PVDF) membrane (Bio-Rad, Tokyo, Japan) as described previously . Membranes were blocked with 3% skim milk and treated with the primary antibodies, anti-RAB3B (ab55655; Abcam, Tokyo, Japan) and anti-VCP (anti-valosin-containing protein), (GTX113030, GeneTex, Alton Pkwy Irvine, CA, USA). The immunoreactive bands were visualized using an ECL Pro (PerkinElmer, Waltham, MA) and Amersham Imager (GE Healthcare, Tokyo, Japan), and quantified using the ImageJ software (National Institutes of Health, USA). VCP was used as the loading control because its levels are more stable compared to those of other loading controls, such as GAPDH and β-actin .
Genome editing for RAB3B
A guide RNA (gRNA) targeting a sequence in RAB3B (5′-GTTTCACCCGCTTCTCGTGA-3′) was constructed by in vitro transcription using a GeneArt Precision gRNA Synthesis Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions (Supplementary Fig. S1a). The gRNA and Cas9 mRNA (GeneArt CRISPR Nuclease mRNA, Thermo Fisher Scientific) were transfected into cells using Lipofectamine MessengerMAX (Thermo Fisher Scientific). Genomic DNA (gDNA) from isolated monoclonal clones was subjected to Sanger sequencing (Supplementary Fig. S1b). The RAB3B-edited clone derived from SK-HEP-1 was named RAB3B-KD. The RAB3B-KD cells were transfected with pcDNA3.1(−) (Thermo Fisher Scientific) harboring full-length RAB3B cDNA, pRAB3B, using Lipofectamine 3000 (Thermo Fisher Scientific) to generate RAB3B rescued KD/pRAB3B cells. A full-length RAB3B cDNA (NM_002867.3: position 214 to 873) with Kozak, 5′-flanking EcoRI, and 3′-flanking BamHI sequences was synthesized by FASMAC (Kanagawa, Japan) and it was inserted into the site downstream of the CMV promoter of pcDNA3.1(−).
Cell viability assay
The CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Tokyo, Japan), which includes a tetrazolium compound [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt; MTS] was used according to the manufacturer’s instructions. Cells in culture medium were incubated with one of the following anticancer drugs, 10 mM 5-fluorouracil (5-FU, Sigma-Aldrich), 250 nM docetaxel (Sigma-Aldrich), 2 μM doxorubicin (Sigma-Aldrich), 200 μM irinotecan hydrochloride (Sigma-Aldrich), 2.5 μM suberoylanilide hydroxamic acid (SAHA, Cosmo Bio), 75 μM sorafenib tosylate (ChemScene, Monmouth Junction, NJ), 25 μM lenvatinib mesylate (Carbosynth, Berkshire, UK), 50 μM regorafenib (ChemScene), or 20 μM cabozantinib S-malate (ChemScene), 24 h at 37 °C in an atmosphere of 5% CO2 in air. Cells incubated in culture medium alone were used as control. The optical density of the culture medium at 492 and 650 nm was measured by using an EnVision plate reader (PerkinElmer). The viability of cells treated with the anticancer drugs was calculated as a ratio with respect to their viability in the absence of anticancer drugs. Triplicate wells were analyzed for each assay.
Splenic injection of tumor cells
NOD-Rag1null IL2rγnull double mutant mice (NRG mice) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA) and maintained in a HEPA-filtered environment with autoclave-sterilized cages, food, and bedding. All animal studies were conducted in accordance with the Institutional Animal Care and Use Committee of Yamaguchi University and conformed to the Guide for the Care, Use of Laboratory Animals published by the United States National Institutes of Health (Bethesda, MD, USA), and ARRIVE guidelines.
The ability of cells to produce tumor nodules in the liver was studied subsequent to their implantation into the spleen of 8–12-week-old NRG female mice as described previously . After 8 weeks, injected mice were sacrificed and necropsied.
Quantitative analysis of exosomes
Cells were incubated for 24 h, after which the exhausted culture medium was replaced with fresh culture medium. At the same time, cell viability was measured from replicate plates using MTS assay and it was used for normalization of exosome quantification. For quantification of exosomes, the conditioned medium was collected and analyzed using a CD9/CD63 Exosome ELISA Kit, Human (Cosmo Bio) according to the manufacturer’s instructions. Signals were detected using an EnVision plate reader (PerkinElmer). If necessary, GW4869 hydrochloride hydrate (Cayman Chemical, Ann Arbor, MI) was added as an exosome inhibitor on the next day of cell seeding. Triplicate wells were analyzed for each assay.
Each experiment was repeated at least three times. Data are expressed as means ± standard deviation. Significant differences were evaluated by the Tukey–Kramer multiple comparison, paired t-test, or Fisher’s exact test, using the R version 3.6.3 software (the R project website, http://www.r-project.org/). A P value of < 0.05 was considered statistically significant.