Huh7 (TCHu182) and HepG2 (TCHu72) cell lines were purchased from Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. The culture medium of these 2 cell lines was Dulbecco’s Modified Eagle’s Medium supplemented with 10% fetal bovine serum (FBS) (Gibco, Carlsbad, CA, United States). The cells were cultured in an incubator at 37°C and 5% CO₂. Short hairpin (sh) RNAs were purchased from Shanghai GenePharma Co., Ltd. The sequences of shYRDC and shNC were used as follows: shYRDC: 5′- CAT​TCG​GAT​TCC​TGA​TCA​T -3′; and shNC: 5′- TTC​TCC​GAA​CGT​GTC​ACG​T -3′. Then, shYRDC and shNC were cloned, respectively, into the BamHI and EcoRI sites of the pGLV3 lentiviral vectors (GenePharm). The YRDC gene was amplified with oligonucleotides to add the NotI and BamHI sites with the primers 5′- AGG​GTT​CCA​AGC​TTA​AGC​GGC​CGC​GCC​ACC​ATG​TCT​CCG​GCG​CGT​CGG​TGC​AGG​GGG​ATG -3′ and 5′- ATC​AGT​AGA​GAG​TGT​CGG​ATC​CTC​ACA​GGT​AGG​ACG​CAT​GTG​AGG​GGA​GCA​GTC​C -3′. The amplified products were cloned into the pGLV5 lentiviral vector (GenePharm). The empty pGLV5 lentiviral vector was used as a control. Huh7 or HepG2 cells were seeded at 2×10⁴ cells/well in 24-well plates, and 24 h after seeding were treated with a linear range of puromycin concentration. Lentivirus was diluted 10-fold with DMEM (supplemented with 10% FBS) for transduction. Cells were incubated with diluted lentivirus solution (Liu YB. et al., 2018). After 24 h, the culture medium was changed to a complete medium and selected with puromycin for 4 days. Until control, the Huh7 cells or HepG2 cells completely died, the YRDC stably knockout and overexpression cells were established. The optimal screening concentration of puromycin was 1 μg/ml. The lentivirus infected cells were observed under a fluorescence microscope (EVOS M5000, Invitrogen) and then collected for analysis by quantitative real-time PCR (qPCR) and western blot.

The Cell Counting Kit-8 (Bimake) assay was used to estimate the viability of the cells. Huh7 and HepG2 cells were seeded in 96-well plates (5,000 cells/well). Next, different concentrations of lenvatinib (Selleck) or BGJ (Selleck) were added into each well separately and incubated for 48 h. After that, the culture medium containing lenvatinib or BGJ was sucked out. The cells were washed with phosphate buffer solution (PBS). Then, the mixture of 10 μl CCK-8 solvent and 90 μl of fresh medium was added to each well and incubated for 60 min. Finally, the absorbance at 450 nm was measured using a microplate reader (EON, BioTek Instruments, Inc.). The cell survival rate was calculated. The experiment was repeated four times.

The cells were seeded in 6-well plates, and 2 ml of cell suspension with a density of 2×10⁵ cells/ml was added to each well. When the cell growth density was around 90%, a 100 μl pipette tip was used to gently scrape off the cells to create a scratch. PBS was added to wash the floating cells, and subsequently, the medium with lenvatinib was added into each well. Photos were taken with the Leica 3000 microscope, using a magnification of 100 times, at 48 and 72 h. ImageJ 1.8.0 (Bethesda, United States) was used for image processing. The experiment was repeated three times.

The cells were seeded in 6-well plates at a density of 1,000 cells/per well. After the cells were stabilized, half of the wells were treated with 0.1 μM lenvatinib for 10 days. Lastly, the colonies were fixed with 4% paraformaldehyde (Beyotime, CHN) for 15 min and stained with crystal violet (Beyotime, CHN) overnight. Fifty cells are defined as a colony. The experiment was repeated three times.

Total RNA was isolated from the cells and tissues using RNAiso Plus (Takara) and reverse-transcribed using a PrimeScript™ RT reagent kit (Takara); the amount of RNA used in this experiment is 1,000 ng. The reverse-transcribed products were used as templates for qPCR with SYBR Green (Takara). Then, qPCR was performed using LightCycler^® 480 Instrument (Roche, Basel, Switzerland). Data were analyzed using the 2^-∆∆ct method, and the expression of GAPDH was used as normalization control. The primers (synthesized by Sangon Biotech, CHN) used in this study were listed as follows: YRDC, forward: 5′-GCA​AGC​GGA​CCC​TCA​AAC​AT-3′; reverse:5′-GCTCAACAAGGACCTAAACCCT-3′; GAPDH, forward: 5′-AGA​TCA​TCA​GCA​ATG​CCT​CCT​G-3′; reverse: 5′-TTG​GCA​TGG​ACT​GTG​GCA​TG-3′; KRAS, forward: 5′- GGG​GAG​GGC​TTT​CTT​TGT​GTA-3′; reverse: 5′- GTC​CTG​AGC​CTG​TTT​TGT​GTC-3′; FGFR2, forward: 5′-AGTCAAGTGGATGGCTCCAG-3′; and reverse: 5-ACAGTT-CGTTGGTGCAGTTG-3′. The experiment was repeated at least three times.

All animal experiments were performed in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Review Board of Central South University (Changsha, China). Forty female 5-week-old BALB/C nude mice (Beijing Vital River Laboratory Animal Technology Co., Ltd.) were fed under specific pathogen-free conditions at the Department of Laboratory Animals, Central South University. The mice were housed in a temperature-controlled environment of 23 ± 1°C with a 12 h light/dark cycle (lights on 08:00 AM) and given a standard mouse diet, with water provided ad libitum. Forty female 5-week-old BALB/C nude mice were randomly divided into four groups of 10 each. YRDC–KD, YRDC–OE, and their control Huh7 cells (NC–KD or NC–OE, 1 × 10⁷) were mixed with the Matrigel in a serum-free medium and then injected subcutaneously into the right lower quadrant of the nude mice. After 14 days, each group of mice was divided into two groups again. Lenvatinib (30 mg/kg) and solvent were intragastrically administered once a day. And, the tumor volume was measured every 2 days. All mice of YRDC–OE and its control were euthanized after 16 days; however, all mice of YRDC–KD and its control were euthanized after 26 days. Tumor nodules were removed and weighed. According to the formula: V = L × W²/2, where V is the volume (mm³), L is the maximum diameter (mm), and W is the shortest diameter (mm).

The isolated bulk tRNA was enzymatically hydrolyzed to nucleosides as described (Crain, 1990). In brief, we added 0.01 units of snake venom phosphodiesterase (Sigma), 10 units of nuclease P1 (Sigma), and 3 μl alkaline phosphatase (Sigma) in a 100 μl volume to digest approximately 100 μg of bulk tRNA. The digested extracts were directly injected in a HPLC–MS system for the separation and identification of the nucleosides as previously described (Pomerantz and McCloskey, 1990). HPLC–MS was performed with an ExionLC AD instrument (AB SCIEX, USA) coupled to a mass spectrometer (Triple Quad 6,500+, AB SCIEX, USA). A Discovery C₁₈ (250 mm, 4.6 mm, 5 μm) reverse-phase column (Waters) equipped with an Ultrasphere ODS guard column (Beckman) was used for lipid separation. The nucleosides were eluted at a flow rate of 1 ml/min and a column temperature of 30 ℃ with a gradient consisting of 10 mM ammonium acetate (A) and 40% acetonitrile (B) as follows: 0–7 min 99% A, 7–30 min 95% A, 30–40 min 85% A, 40–50 min 95% A, and 50–60 min 99% A. The mass spectrometer records in the cation mode. Using multiple/selected response monitoring modes (MRM/SRM), the collision energy is found to be 10 eV. For adenosine, the mass shifts from m/z 268–m/z 136, and for t⁶A, the mass shifts from m/z 413.2–m/z 281. The experiment was repeated three times.

The cDNA of human KRAS was synthesized and inserted between EcoRI and SalI sites of pET-28a (+) by GeneChem. The constructed pET-28a (+)-KRAS plasmid was linearized by dissecting with XhoI. The KRAS mRNA was transcribed from 1 μg linearized plasmid DNA in vitro using the MEGAscript T7 kit (Ambion) in the presence of m7G (5′) ppp (5′) G (CAP) as described by the manufacturer. The KRAS mRNA was purified by phenol/chloroform extraction, ethanol precipitation, and quantified by UV spectroscopy.

The specific depletion of endogenous tRNAs from commercial Flexi rabbit reticulocyte lysate (RRL) system (L4540, Promega) was achieved by ethanolamine–sepharose column chromatography according to the method of Jackson et al. (2001). All experimental steps were performed at 4 °C. 100 μl of 50% ethanolamine–sepharose slurry in buffer A (0.1 mM EDTA, 25 mM KCl, 1.1 mM MgCl₂, 10 mM NaCl, 10 mM HEPES-KOH, and pH 7.2) with 50 mM KOAc and 0.25 mM Mg(OAc)₂ was incubated with commercial RRL for 45 min at 4°C for the depletion of tRNA in commercial RRL. The desired supernatant (RRL depleted of endogenous tRNA) was collected by brief centrifugation at 1,500 g and was either used immediately for translation reactions or stored at −80 ℃.

Cellular tRNAs were prepared from four kinds of Huh7 cells by the procedure of Dittmar et al. (2006). In vitro translation was performed using a RRL system (L4540, Promega). 1 μg of luciferase mRNA (L4561, Promega) or 2 μg of KRAS mRNA was added to RRL containing 70 mM KCl, 2 mM DTT, 10 μM minus leucine, 10 μM minus methionine, and 40 U/μl RNase ribonuclease inhibitor. For RRL that is depleted of endogenous tRNA, 2.5 μg of extra tRNA isolated from the four kinds of Huh7 cells was respectively added to the reaction mixture. A total volume of 50 μL from this reaction mixture was incubated for 60 min at 30 ℃. To avoid aminoacyl tRNAs producing background bands (∼25 kDa), we added 0.2 mg/ml RNase A to the reaction (after completion) and incubate for 5 min at 30 ℃. Luciferase assay was performed according to the protocol by using a luminometer (Sirius, Berthold) to quantify the luciferase produced in the in vitro translation system. The KRAS protein produced in the in vitro translation system was measured by the western blot. The experiment was repeated three times.

The cells were collected from the cell flask into a 1.5 ml EP tube, and it was found that the number of cells is about 1 × 10⁷. Then, 1 ml RIPA lysis solution containing 10 μl protease inhibitor and 10 μL phenylmethanesulfonyl fluoride (P0013B, Beyotime) was added. The mixture was added to the cells and tissues under the conditions at 4 ℃ and fully dissolved for half an hour for protein extraction. The lysate was centrifuged at 8,000 g at 4°C for 15 min. Then, the supernatant was collected and the protein concentration was measured by BCA method. The protein samples in in vitro translation system were prepared by a special method. Once 50 μl translation reaction is completed, the 5 μl aliquot was removed, and it was added to 15 μl SDS sample buffer and heated at 75°C for 15 min to denature the proteins. The protein lysate (30 μg) was subjected to 10% SDS-PAGE and then electrotransferred onto the polyvinylidene difluoride membrane (PVDF, Millipore IPVH00010, Solarbio). And, the membrane was transferred to 5% skimmed milk for blocking for 1 h, followed by a membrane cutting process. Different bands were placed in an incubation box overnight in different primary antibodies. The main antibodies are as follows: β-actin (ab6276, Abcam), YRDC (sc-390477, SANTA CRUZ), and KRAS (12063-1-AP, Proteintech), and the ratios of them to the primary antibody dilution buffer (P0256, Beyotime) are 1:10,000, 1:800, and 1:5,000. The bands were washed the next day, and the secondary antibodies were incubated for 1 hour at room temperature. The ratio of β-actin and YRDC’s secondary antibody (SA00001-1, Proteintech) to the secondary antibody dilution buffer (P0258, Beyotime) is 1:10,000 and 1:5,000, and the ratio of KRAS’s secondary antibody (SA00001-2, Proteintech) to the secondary antibody dilution buffer (P0258, Beyotime) is 1:2,000. The bands were washed and soaked in the ECL kit (36208ES60, yeasen), and then ChemiDoc XRS + image analyzer (Bio-Rad, USA) was used to visualize the protein band, and Image Lab was used for immunoblot densitometric analysis. The experiment was repeated three times.

All data were described as mean ± standard deviation (mean ± SD). The comparison of two groups is carried out by Student’s t-test. All the other data were analyzed with one-way ANOVA followed by LSD when equal variances were assumed. The half maximal inhibitory concentration (IC₅₀) was calculated by GraphPad Prism six software. All statistical analyses were performed by SPSS 22.0 or GraphPad Prism 6 (GraphPad Software, Inc. La Jolla, CA, United States). p < 0.05 was considered statistically significant.

Stable YRDC knockdown (YRDC-KD), YRDC overexpression (YRDC–OE) and their respective control HCC cells (NC–KD or NC–OE), were successfully established in our previous study (Huang et al., 2019). The effects of lenvatinib on cell proliferation were measured by the CCK-8 cell viability assay in these four kinds of Huh7 or HepG2 cell models. Figure 1A showed that the Huh7 cells with YRDC knockdown were significantly less sensitive to lenvatinib-mediated growth inhibition than their control cells (IC₅₀: 1.379 vs 0.821 μM). However, the Huh7 cells with YRDC overexpression were a little higher, but not significantly sensitive to lenvatinib than their control cells (Figure 1B). The HepG2 cells had the same pattern as the Huh7 cells, but the IC₅₀ of lenvatinib to the former is higher than the latter (Figures 1C,D). Wound-healing assays revealed that the migration inhibition of lenvatinib to Huh7 cells was decreased by YRDC knockdown (Figures 2A–C). The clone formation assay further confirmed that Huh7 cells with YRDC knockdown had a less sensitivity to lenvatinib than their control cells (Figures 3A–C). These data suggested that the knockdown of YRDC in Huh7 or HepG2 cells allowed the cells to become more resistant to lenvatinib.

To test the effect of YRDC on lenvatinib resistance in Huh7 cells in vivo, four kinds of stable Huh7 cells with different expression of YRDC were injected into nude mice for the construction of xenograft models. In vivo growth of xenograft tumors was significantly inhibited by lenvatinib at a dose of 30 mg/kg in all models (Figure 4A). The results indicated that YRDC knockdown or lenvatinib treatment significantly inhibited the tumor growth, exhibited as declined tumor volume (Figure 4B), and decreased tumor weight (Figure 4C). However, the knockdown of YRDC could weaken the lenvatinib-mediated tumor growth inhibition, mainly reflecting in the inhibition of tumor weight (66 vs. 91%, Figure 4C). Meanwhile, YRDC overexpression cells showed more sensitivity to lenvatinib compared with its control cells when evaluated by tumor weight (78 vs. 67%) (Figure 4C), however, not significantly different when evaluated by tumor volume (Figure 4B). These results suggested YRDC expression could affect the lenvatinib sensitivity on tumor growth in vivo, which mainly reflect in tumor weight.

Since YRDC could be induced under stress, like the ischemia–reperfusion condition (Jiang et al., 2005), more evidence suggested that the resistance emerges soon after the initial treatment with lenvatinib (Fu et al., 2020). We wonder how YRDC changed under the treatment of lenvatinib. Interestingly, the mRNA and protein expression level of YRDC was remarkedly inhibited by lenvatinib in a time-dependent manner in Huh7 cells (Figures 5A,B). Subsequently, we further confirmed the inhibition effect of lenvatinib on YRDC expression in xenograft models (Figures 5C,D). Combined with the results above, we speculated that the inhibition effect of lenvatinib on YRDC could aggravate lenvatinib resistance in HCC cells with a low original expression of YRDC. Additionally, due to the inhibition of lenvatinib on YRDC, the expression of YRDC would be lower after lenvatinib treatment in original sensitive HCC cells (Figures 5A,B), and it may be an underlying cause due to which the resistance emerges soon after lenvatinib initial treatment. However, the detailed mechanisms of YRDC involved in lenvatinib resistance in HCC cells were still unknown.

The results mentioned above suggested that YRDC mediated the resistance of lenvatinib in HCC cells. Our previous study reported that the expression level of YRDC was positively correlated with the proliferation, migration, and invasion of Huh7 cells. These suggested that YRDC could be the target and the executor of lenvatinib antitumor. As we know, the antitumor effect of lenvatinib was realized via inhibiting tumor angiogenesis and directly blocking the RAS/RAF/MEK/ERK signaling pathway. Considering our previous finding that YRDC promotes HCC cells proliferation via regulating the activity of the RAS/RAF/MEK/ERK pathway, we speculated that this pathway may involve in YRDC-mediated lenvatinib resistance. Further experiments were conducted to investigate the correlation between them. Trametinib, a selective MEK inhibitor, was used in our following in vitro experiments. It showed that knockdown of YRDC attenuates cellular sensitivity to trametinib (Figure 6A). When combined with lenvatinib treatment, the cell mortality did not decrease any further than trametinib treatment alone (Figure 6A). In other words, there was no difference in the sensitivity of lenvatinib in YRDC knockdown cells and its control cells after trametinib treatment. All these results confirmed that the RAS/RAF/MEK/ERK pathway was involved in the YRDC-mediated lenvatinib resistance.

YRDC was an NNU-tRNA t⁶A modification enzyme. The low t⁶A modification of NNU-tRNA reduced the translation rate or increased the possiblity of the translation error. We analyzed the codon composition of all genes related in the lenvatinib antitumor pathway and found that the KRAS genes have the highest ANN codon ratio (35.98%, Table 1). In the four kinds of Huh7 cell models, the mRNA expression of the KRAS had no difference (Figure 6B); however, the protein level of the KRAS in YRDC knockdown was significantly lower compared with its control (Figure 6C). The protein level of KRAS in YRDC overexpression cells was slightly higher compared with its control cells. These results suggested that the low expression of YRDC may mainly reduce the level of NNU-tRNA t⁶A modification and further decrease the protein synthesis of KRAS gene with a high ANN codon ratio.

Then we analyzed the t⁶A modification level in the four kinds of Huh7 cell models with different YRDC expressions. As expected, the t⁶A level in YRDC knockdown cells was significantly lower by 18% when compared with its control cells, and the t⁶A level in YRDC overexpression cells was not significantly changed when compared with its control cells (Figure 6D).

To further confirm this point, we established the in vitro rabbit reticulocyte translation system. Firstly, we eliminated the endogenous tRNA from the rabbit reticulocytes translation system and performed the in vitro translation of the luciferase gene under the addition of tRNA isolated from the four kinds of Huh7 cells. The fluorescent activity in tRNA from YRDC knockdown was slightly lower than that of its control (p = 0.0985). The situation reversed between YRDC overexpression and its control (Figure 6E). The ANN ratio in the luciferase gene was 28%, which is close to the average level 26.1% in all genes expressed in humans. This result indicated that YRDC knockdown could reduce general protein synthesis.

We further performed the in vitro translation of KRAS. The cDNA of KRAS was cloned into the pET-28a (+) plasmid. The KRAS mRNA was obtained by the reverse transcription using pET-28a (+)-KRAS plasmid as a template. In the in vitro rabbit reticulocyte translation system, the level of the KRAS protein was significantly lower when added with the tRNA isolated from YRDC knockdown cells compared with the tRNA from its control cells (0.55 vs. 1.0). However, there is no significance in the KRAS synthesis in addition of tRNA from YRDC overexpression cells compared with tRNA from its corresponding control cells (Figure 6F).