The liver cancer cell lines Hep3B (RRID: CVCL_0326), SNU387 (CVCL_0250), SNU398 (CVCL_0077), and HepG2 (CVCL_0027) were obtained from ATCC. The mouse hepatocyte cell line AML12 (CVCL_0140) was also obtained from ATCC. Huh6 (CVCL_1296) and Huh7 (CVCL_0336) cells were purchased from the Japanese Collection of Research Bioresources Cell Bank (JCRB). All cells were obtained directly from cell banks that perform cell line characterizations and were passaged in our laboratory for fewer than 6 months after receipt. The cells were maintained according to instructions provided by the ATCC and JCRB and incubated with Len (CAS No.: 417716-92-8, ChemScene LLC, NJ, USA) and/or ASNase (ENZ-287, Prospec, NJ, USA) using the indicated concentration and time course. Relative cell viability was determined using a WST-8 method with Cell Count Reagent SF (#07553-15, Nacalai Tesque, Kyoto, Japan). Combination Index (CI) and dose-effect analyses were calculated using CompuSyn software (ComboSyn, Inc., Paramus, NJ, 07652 USA) according to the Chou-Talalay method (16). CI values were plotted against the fraction of affected cells (Fa) to represent the percentage of growth inhibition (CI-Fa plot). A CI value <1.0 indicates synergism of the combination. A lactate dehydrogenase (LDH) cytotoxicity assay was carried out using a Cytotoxicity LDH Assay Kit (#CK12, Dojindo Laboratories) according to the manufacturer’s instructions. Cells were incubated with indicated concentrations of drugs for 72 hours. Cytotoxicity was calculated as a percentage of the ratio of LDH release compared to controls having high levels of LDH.

RNA was extracted using TRIzol (#15596018, Invitrogen/Thermo Fisher Scientific, MA) and further purified using chloroform and isopropanol. RNA (1 μg) was used to generate cDNA with the PrimerScript RT cDNA Synthesis Kit (#RR036A, Takara Bio, Shiga, Japan). The expression of individual genes was quantified by real-time qPCR using SYBR FAST qPCR Master Mix (#KK4602, KAPA BIOSYSTEMS, MA) and a LightCycler 96 Real-Time PCR system (Roche Diagnostics, Mannheim, Germany). Expression levels were normalized to the housekeeping control gene GAPDH. Primers used for real-time qPCR analyses are listed in Supplementary Table 1.

Harvested liver cancer cells were homogenized in RIPA buffer and equal amounts of liver homogenates were fractionated by SDS-PAGE and transferred onto a PVDF membrane. The membranes were incubated with antibodies to GS (#G2781, Sigma-Aldrich, MO), ASNS (#14681-1-AP, Proteintech, IL), ASCT2 (#8057), p70S6K (#9202), phospho-p70S6K (Thr389, #9205), S6 (#2217), phospho-S6 (#4858), 4EBP1 (#9452), phospho-4EBP1 (#2855), PARP (#9532) (from Cell Signaling Technology, MA), and β-actin (Sigma-Aldrich, MO).

A LIVE/DEAD Cell Imaging Kit (#R37601, Invitrogen) and ROS (reactive oxygen species) Assay Kit-Highly Sensitive DCFH-DA kit (#R252; DOJINDO Laboratories. Inc., Kumamoto, Japan) were used according to the manufacturer’s instructions for the immunofluorescence analysis of liver cancer cell lines, including Huh6 cells, treated with Len, ASNase, or Len plus ASNase. Briefly, cells were incubated in a 96-well plate with the indicated treatment for 72 h, followed by incubation with the LIVE/DEAD reagent for 15 min at room temperature, or DCFH-DA solution for 30min at 37°C. A BZ-X800 Fluorescence Microscope (Keyence corporation, Osaka, Japan) was used to observe live or dead cells and to detect ROS.

The animal study was performed in accordance with Kyoto Prefectural University of Medicine (KPUM) guidelines for the care and use of live animals and was approved by the KPUM Institutional Animal Care and Use Committee (M2019-559). The mice used in this study for the xenograft model were female BALB/c nude mice aged 6-8 weeks and C57/BL6J male mice aged 20 weeks for determination of pharmacological impacts of ASNase on mature mice (Charles River Laboratories, Japan). The animals were maintained at KPUM in filter-topped cages with a 12 h dark/light cycle and given an autoclaved diet and water. To generate a xenograft tumor model, the flanks of 40 mice were inoculated with liver cancer cells (2.5 × 10⁶ cells) suspended in 100 μl PBS. Two weeks after the inoculation, mice having a tumor ≥3 mm were randomly divided into four groups: Control, Len, ASNase, and Len plus ASNase groups (n = 7–9 per each group). Len was administered orally daily (10 mg/kg in 3 mM HCl DMSO), while ASNase was injected intraperitoneally three times per week (5 U/g). The control group received saline instead of Len or ASNase. After 2 weeks of treatment, all the mice were sacrificed and analyzed. The relative tumor volumes were evaluated by comparing the final and initial volumes. Tumor volumes were calculated using the formula V = (L x W x W)/2, where V is tumor volume, W is tumor width, L is tumor length.

Data are presented as the mean ± standard deviation (SD) or the median with interquartile range, as indicated. For in vitro analysis, representative data from at least three independent experiments were shown. For the DEAD/LIVE and ROS stainings ( Figures 3 and 5 ), 8-10 field images were captured and analyzed. Differences in means and two categorical variables were analyzed by student’s t-test using JMP8.0 (SAS Institute Inc., NC). Multiple comparisons were analyzed by Turkey’s multiple comparisons test (GraphPad Prism 6, GraphPad Software, CA). Significance was set at P values < 0.05 (not significant (N.S.), P > 0.05; *, P < 0.05; **, P < 0.01).

We first examined the effect of Len, a TKI that is commonly used for HCC treatment, and ASNase on the survival of six liver cancer cell lines carrying specific mutations. Huh6, SNU398, and HepG2 cells harbor β-catenin mutations, while Hep3B and Huh7 cells have Fibroblast Growth Factor 19 (FGF19) mutations. SNU387 cells do not possess β-catenin or FGF19 mutations (Huh6: β-catenin+, FGF19−; Huh7: β-catenin−, FGF19+; SNU387: β-catenin−, FGF19−; SNU398: β-catenin+, FGF19−; Hep3B: β-catenin−, FGF19+; and HepG2: β-catenin+, FGF19−) (17). Notably, GS is a target of β-catenin (18), whereas FGF19 expression was reported to be associated with the clinical efficacy of Len (19).

Cell viability was measured using the WST-8 method after 72 hours of incubation with the indicated treatment. Len ( Figure 1A , blue lines) induced dose-dependent growth inhibition in the four cell lines. In particular, Hep3B and Huh7 cells, which harbor an FGF19 mutation, showed maximal inhibition (Fa 0.5: 0.28 μM for Hep3B and 2.84 μM for Huh7 cells, where Fa 0.5 is equivalent to the half-maximal inhibitory concentration (IC₅₀), which causes a 50% inhibition of the desired activity; Figure 1B ). Huh6 and HepG2 cells exhibited a moderate degree of growth inhibition (Fa 0.5: 10.49 and 14.25 μM, respectively; Figure 1B ). SNU387 and SNU398 cells were more resistant to Len than the other cell lines (Fa 0.5: 37.15 and 38.96 μM, respectively; Figure 1B ). Specifically, ASNase ( Figure 1A , red lines) caused a clear dose-dependent inhibition of the growth of SNU387 and SNU398 cells (Fa 0.5: 1.92 and 0.90 U/mL, respectively; Figure 1B ), whereas Huh7 and HepG2 cells were partially resistant to this treatment (Fa 0.5: 51.15 and 25.07 U/mL, respectively). Notably, Huh6 cells did not reach 50% growth inhibition (Fa 0.5: > 100 U/mL).

We also confirmed that ASNase indeed depleted asparagine and glutamine concentrations but increased aspartate and glutamate levels in the cell culture media ( Supplementary Figure 1A ).

To explore the possibility that treatment with ASNase combined with Len inhibits liver cancer cell growth to a greater extent than either drug alone, we treated liver cancer cells with ASNase and Len combined. The results from WST-8 analyses were used to determine the Combination Index (CI), which is the fraction of affected cells (Fa) ( Figure 2 ). A synergistic effect indicated by a CI value <1 for Len (0–50 μM) with ASNase (0–50 U/mL) was seen for all HCC cells tested (Fa 0.5), except for Huh7 cells (CI = 1.13). Surprisingly, the Len plus ASNase combination therapy had the highest synergistic effect on Huh6 cells, even though treatment of these cells with ASNase monotherapy did not reach Fa 0.5 ( Figures 1A, B ).

To confirm whether the combination of Len and ASNase induced cell death, LIVE/DEAD staining was performed on Huh6 cells ( Figure 3A ). The live cells showed intense and uniform green fluorescence, while the dead or dying cells exhibited a predominantly red nuclear fluorescence. In accordance with the WST-8 results, Len treatment had a moderate antitumor effect, whereas ASNase did not induce death of Huh6 cells. Importantly, the Len and ASNase combination treatment induced cell death to a much greater extent than the Len monotherapy ( Figure 3B ). These results are comparable with those of WST-8 analyses.

We also performed WST-8 assay for AML12, a normal hepatocyte cell line, and found that high dose of Len or ASNase, approximately 16 μM or 15 U/ml respectively, reduced cell proliferation, however, no synergistic effect indicated by a CI value <1 for Len (0–50 μM) with ASNase (0–50 U/mL) was achieved unlike in the case of HCC cell lines ( Supplementary Figures 2A–C ). Taken together, these results suggest that ASNase exerts synergistic antitumor effects in concert with Len on liver cancer cells and is particularly effective on Huh6 cells.

To investigate the mechanism by which ASNase reduces cell viability and induces cell death of liver cancer cells, we examined expression levels of GS and ASNS. Specifically, we analyzed messenger RNA (mRNA) expression levels of glutamate-ammonia ligase (GLUL), which encodes GS, and ASNS genes by real-time quantitative PCR (qPCR). SNU387 and SNU398 cells, which are sensitive to ASNase, showed lower GS expression than other cells, in addition, they also expressed higher ASNS levels than other cells at both the mRNA and protein level ( Figures 4A, B ). The amino acid contents in the culture media were analyzed and indicated in supplementary Figure S2D . The results suggest that SNU387 and SNU 398 consumed more extracellular glutamine (GLN) than other cells. Because of low GS expression, these two cell lines were more sensitive to asparagine (ASN) and GLN depletion by ASNase. The culture media from SNU387 and SNU 398 contained more ASN than other cells, which is consistent with higher ASNS expression. It is known that ASNS catalyzes the synthesis of asparagine (ASN) using glutamine (GLN). It may also lead to less GLN detected in the culture media from SNU387 and SNU 398 than others.

Interestingly, ASNase induced GS expression in cells with high GS expression ( Figure 4C ). Although ASNase alone was not effective on Hep3B, Huh6, or HepG2 cells, the combination of ASNase with Len synergistically suppressed viability of these cell lines.

We also examined the effects of ASNase on SNU387 cells with ASNS knockdown and Huh6 cells with GS knockdown. As expected, GS knockdown additively reduced the viability of Huh6 cells but not affect that of SNU387 cells in which GS expression was low. ASNS knockdown did not reduce the viability of both cell lines. The knockdown efficiency was confirmed with immunoblotting ( Supplementary Figures 3 , 4 ). Next, cross-silencing experiments of GS in SNU387 and ASNS in Huh6 were performed ( Supplementary Figure 4 ). These silencing did not affect the cell viability.

As Huh7 cells alone were resistant to the combination of ASNase and Len, we compared the expression levels of amino acid transporters in all cell lines. Among the selected transporters, expression of ASCT2 (SLC1A5), which is important for maintaining glutamine levels in tumor cells (20), was downregulated at both the mRNA and protein level in Huh7 cells ( Figures 4A, B ). In addition, Huh7 cells had the highest GS expression among all the liver cancer cell lines tested ( Figure 4B ). These results suggest that Huh7 cells do not depend upon extracellular glutamine for survival, but rather de novo glutamine synthesis or other metabolic alterations that provide energy sources.

Huh6, SNU398, and HepG2 cells harbor β-catenin mutations, and GS is a β-catenin target. Unexpectedly, we found that the presence of a β-catenin mutation did not correlate with GS expression level or sensitivity to ASNase ( Figure 4B ). Therefore, we further investigated the synergistic effects of ASNase plus Len as a combination therapy. A previous study reported that the inhibition of glutamine-dependent mechanistic/mammalian target of rapamycin complex 1 (mTORC1) activation ameliorates liver cancer development driven by β-catenin mutations (18). We and another group previously reported that mTORC1 activation is sufficient for the spontaneous development of HCC in a mouse model in which the tumor suppressors tuberous sclerosis complex (TSC) 1 or TSC2 are lost (21, 22). Notably, mTORC1 activation is highly prevalent in cancer, including in 45% to 50% of HCC cases (23). Therefore, we evaluated mTORC1 activation by measuring phosphor-p70 ribosomal S6 kinase (p-S6K), phospho-S6 ribosomal protein (p-S6), and phospho-eukaryotic translation initiation factor 4E-binding protein 1 (p-4EBP1) expression levels in cell lines treated with the various agents. We found that basal mTORC1 activity was relatively high in Huh7 and Hep3B cells and Len or ASNase monotherapy reduced its activity, however, the combination therapy showed no differences compared to the monotherapies. Although the combination therapy reduced mTORC1 activity in Huh7 cells, none of the treatments yielded significant mTORC1 activation or suppression in Huh6 cells. Therefore, mTORC1 may not be a critical determinant of the sensitivity of Len plus ASNase combination therapy.

We evaluated apoptotic cell death using poly (ADP-ribose) polymerase (PARP) as an apoptotic marker ( Figure 4C ). ASNase did not regulate PARP levels either with or without Len co-administration in Huh6 cells. Next, we focused on reactive oxygen species (ROS) production.

Interestingly, glutamine-depleted tumor cells are reported to be vulnerable to vitamin C treatment due to a significant increase in ROS production (24). To explore the effect of ASNase on liver cancer cells, we performed a reactive oxygen species (ROS) assay. Len treatment was associated with ROS production in Hep3B and Huh6 cells, but ASNase treatment induced more ROS production than did Len treatment ( Figures 5A, B ). Notably, the combination of Len and ASNase demonstrated substantially more ROS production than ASNase monotherapy. We also analyzed some anti-oxidant genes by real-time qPCR. Among them, HMOX1 and GSTM1 were upregulated by the combination treatment and indicated in Figure 5C . These results suggest that ASNase induces oxidative stress in HCC cells, and Len and ASNase combination therapy enhances this stress response that contributes to the antitumor effect. Oxidative stress may thus play a role in the anticancer effects of ASNase.

Next, we tested the effects of the ASNase combined with Len in an in vivo mouse model. As a preliminary experiment, we injected ASNase intraperitoneally into wild-type mice once a day for eight consecutive days, and then examined the blood amino acid profiles. As expected, these mice exhibited asparagine and glutamine depletion, whereas aspartate and glutamate concentrations were increased ( Supplementary Figure 1B ). Next, we inoculated Huh6 cells, in which the combination therapy was the most effective, into the flanks of BALB/c nude mice. When tumors were palpable (≥3 mm) at day 14, the mice were randomly divided into four treatment groups: Len, ASNase, Len combined with ASNase (Len + ASNase), or vehicle control (control) and treated for an additional 14 days ( Figure 6A ). Tumors arising from liver cancer cells in the control mice grew steadily, but mice treated with Len had slower tumor growth. Mice treated with the Len with ASNase combination therapy exhibited completely retarded tumor growth ( Figure 6B ). Although Len or ASNase monotherapy suppressed tumor growth, the Len plus ASNase combination therapy was more effective than either of the monotherapies alone.

To assess the proliferation status of xenograft tumors, we performed Ki67 immunostaining. Consistent with tumor development, the Len alone and Len plus ASNase combination treatments had significantly reduced Ki67-positive tumor cell counts compared to ASNase monotherapy or the control condition ( Figure 6C ).

Although ASNase is associated with several toxicities, including liver dysfunction (25), we did not detect liver injury in the mice treated with ASNase. In turn, Len yielded mild liver injury, but the Len plus ASNase combination therapy did not exacerbate this injury ( Supplementary Figure 1C ). Further, Len treatment led to a mild increase in body weight relative to the control treatment, but this difference was not significant ( Supplementary Figure 1C ).

These findings suggest that ASNase sensitizes liver cancer cells to Len treatment, leading to synergistic inhibitory effects without exacerbation of liver injury.

The precise mechanism by which Len and ASNase combination therapy is synergistically effective—even in cells that are resistant to each drug on its own—remains unclear. ASNase monotherapy was not effective in Huh6 cells in vitro; however, it suppressed tumor development in the xenograft mouse model. Len was also more effective in vivo vs. in in vitro settings. This discrepancy may be attributed to the effects of ASNase on the tumor microenvironment, including the blood supply, immune cells, and other mesenchymal cells.