Acquired lenvatinib-resistant HCC cells display increased cancer stemness

To establish HCC LR cell models, we exposed Hep-3B, HuH-7 and SNU-387 cells to increasing concentrations of lenvatinib (starting from 2 μM) for 3–6 months (Fig. 1a) and named them Hep-3B-LR, HuH-7-LR and SNU-387-LR. Compared with the parent strains, the LR strains acquired elevated IC50 values from 1.85 μM to 10.83 μM, 3.86 μM to 15.55 μM and 11.89 μM to 19.36 μM (Fig. 1b; Supplementary Fig. 1a). Nude mouse model bearing HuH-7 cell and HuH-7-LR cell xenografts were used to elucidate the effect of LR in vivo. HuH-7 cell xenografts showed a marked reduction in tumor volume following the administration of lenvatinib, while HuH-7-LR cell xenografts were decreased slightly (Fig. 1c, d). To avoid the side effects of lenvatinib, the weights of the mice were obtained, and no significant variation was found (Supplementary Fig. 1b). The colony formation assay further confirmed the LR phenotype (Fig. 1e; Supplementary Fig. 1c). Considering drug resistance as a crucial element of cancer stemness, we explored changes in stemness properties in HCC LR cells. Compared with the parent strains, the LR strains displayed significantly promoted migration ability (Fig. 1f; Supplementary Fig. 1d), formation of more and larger hepatospheres (Fig. 1g; Supplementary Fig. 1e) and upregulated mRNA expression of CSC markers (Fig. 1h; Supplementary Fig. 1f). We also examined the tumorigenicity of HuH-7 and HuH-7-LR cells in vivo (Fig. 1i). The estimated CI for the frequency of CSCs in HuH-7-LR cells was 1/251142, while that in HuH-7 cells was 1/2237898 (Fig. 1j). These results suggested that the LR strains displayed increased cancer stemness.

Fig. 1

Acquired lenvatinib-resistant HCC cells display increased cancer stemness. a HCC cells were subjected to escalating doses of lenvatinib over a period of 3–6 months in order to develop lenvatinib-resistant cell models. b The IC50 value of Hep-3B, Hep-3B-LR, HuH-7, and HuH-7-LR cells was determined via CCK-8 assay. Each point on the dose-response curves consisted of five replicates. c A vehicle or a Len treatment was administered to nude mice xenografted with HuH-7 cell (lenvatinib, 30 mg/kg) for 3 weeks (n = 5 per group). The isolated tumors were photographed after the mice were sacrificed. d Tumor volume of each group was measured twice a week. e Hep-3B, Hep-3B-LR, HuH-7, and HuH-7-LR cells were treated with the indicated concentrations of lenvatinib. Then, the remaining cells were stained after 2 weeks with crystal violet staining. The colony formations were photographed and counted. Histograms were used to statistically analyze changes. f The migration ability of Hep-3B, Hep-3B-LR, HuH-7, and HuH-7-LR cells was evaluated by transwell assay after incubation for 24 hours. g The sphere formation assay was used to evaluate the in vitro self-renewal ability. h The qRT-PCR assay was employed to measure the mRNA levels of stemness markers. i Limiting dilution xenograft formation of HuH-7 and HuH-7-LR cells in NOD/SCID mice (n = 5 per group). The isolated tumors were photographed, and the stem cell frequency (j) was calculated. Extreme limiting dilution analysis (ELDA) was used for the limiting dilution assay. The data from cell functional assays and RT-qPCR analysis were presented as mean ± SD of three individual experiments, and the data from animal experiments were presented as mean ± SEM. The Student’s t test was used for comparisons. *p < 0.05, **p < 0.01, ***p < 0.001. Scale bar: 100 μm. LR lenvatinib resistant, CCK-8 Cell Counting Kit-8, ns nonsignificant

HIF-1α pathway activation is responsible for acquired LR and increased cancer stemness in HCC

To investigate the molecular profile of the acquired HCC LR cells, we performed RNA sequencing and found that the HIF-1-centered transcriptional pathway was significantly activated (Fig. 2a, b). As the main subunit of HIFs, HIF-1α protein was found to be increased in HCC LR cells, while the HIF-2α protein had unstable changes (Fig. 2c; Supplementary Fig. 2a), which further supported the RNA-sequencing results. HIF-1α transcriptional activity was also increased in HCC LR cells (Fig. 2d; Supplementary Fig. 2b). The HIF-1α pathway is known to play a critical role in both glycolysis and cancer stemness. Thus, we tested the glycolysis capacity by a Seahorse XF Analyzer and found a higher extracellular acidification rate (ECAR) in HuH-7-LR and SNU-387-LR cells (Fig. 2e). In LR cells, the mRNA levels of glycolysis-related genes were elevated (Fig. 2f; Supplementary Fig. 2c). Moreover, knockdown of HIF-1α effectively decreased the migration ability (Fig. 2g), in vitro self-renewal ability (Fig. 2h) and the mRNA expression of CSC markers (Fig. 2i; Supplementary Fig. 2d). To identify the role of HIF-1α in LR, parent cells and LR cells were exposed to lenvatinib with or without HIF-1α knockdown. The results showed that HIF-1α knockdown could markedly increase the sensitivity of HCC LR cells to lenvatinib (Fig. 2j), but there was no significant difference in parent cells. In summary, these data revealed that activation of HIF-1α pathway promoted cancer stemness and LR in HCC.

Fig. 2

HIF-1α pathway activation is responsible for acquired LR and increased cancer stemness in HCC. a Top 156 upregulated DEGs of HuH-7-LR versus HuH-7 and SNU-387-LR versus SNU-387 were extracted from RNA-seq data. b Upon KEGG analysis show, it was revealed that the 156 upregulated DEGs were involved in associated with the HIF-1α pathway. c The expression of HIF-1α and HIF-2α was detected in WT and HCC LR cells via western blot analysis. d Changes in relative luciferase activity in WT and HCC LR cells were determined by luciferase reporter assay. e The ECAR of HCC LR cells and WT cells was identified using the Seahorse system. The means and SDs are represented on the column graphs. f qRT-PCR was used to measure the expression of glycolysis driver genes in WT and HCC LR cells. The effects of HIF-1α knockdown (shHIF-1α) or not (shNC) on HuH-7-LR and SNU-387-LR cell stemness were shown according to migration (g), in vitro self-renewal (h), and mRNA expression of stemness markers (i). j The WT and HCC LR cells with or without HIF-1α knockdown (shHIF-1α or shNC) were exposed to the indicated concentrations of lenvatinib for 48 hours. Trypan blue staining-based cell counting was used to detect lenvatinib sensitivity. The data were presented as mean ± SD of three individual experiments. The Student’s t test was used for comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar: 100 μm. DEGs differentially expressed genes, KEGG Kyoto Encyclopedia of Genes and Genomes, WT wild type, LR lenvatinib resistant, HIF-1α hypoxia-inducible factor-1α, HIF-2α hypoxia-inducible factor-2α, HCC hepatocellular carcinoma, ECAR extracellular acidification rate

MYH9 promotes LR and cancer stemness in HCC through stabilizing HIF-1α

To further determine the mechanism underlying the increased HIF-1α protein in HCC LR cells, we conducted immunoprecipitation with an anti-HIF-1α antibody on the whole cell lysate from SNU-387 and SNU-387-LR cells. After analyzing the results of silver staining and mass spectrometry sequencing, MYH9 was found to be a potential potent HIF-1α-interacting protein (Fig. 3a, b) in HCC LR cells. MYH9 was also increased in HCC LR cells (Fig. 3c; Supplementary Fig. 3a). The exogenous and endogenous interaction of HIF-1α and MYH9 was also confirmed in HEK-293T and HCC LR cells, respectively (Fig. 3d; Supplementary Fig. 3b). Immunofluorescence analysis revealed that the protein expression of HIF-1α and MYH9 was elevated in HuH-7-LR cells compared to that in HuH-7 cells and their colocalization was in cytoplasm (Fig. 3e). Given that HIF-1α could be stabilized by its interacting proteins, we speculated that MYH9 could bind to and stabilize HIF-1α. We demonstrated that HIF-1α was notably decreased in MYH9 knockdown HCC LR cells (Fig. 3f; Supplementary Fig. 3c). In parent HCC cells, overexpression of MYH9 increased HIF-1α protein expression (Fig. 3g; Supplementary Fig. 3d). But mRNA expression changes of HIF-1α were unstable with MYH9 knockdown in LR cells or MYH9 overexpression in HCC cells (Supplementary Fig. 3e, f). Moreover, knockdown of endogenous MYH9 facilitated HIF-1α protein degradation and inhibited its transcriptional activity in HCC LR cells (Fig. 3h; Supplementary Fig. 3i). Conversely, overexpression of MYH9 significantly delayed HIF-1α protein degradation and promoted its transcriptional activity in parent cells (Fig. 3i; Supplementary Fig. 3g, j). The above results indicated that MYH9 could interact and stabilize the HIF-1α protein.

Fig. 3

Identifying MYH9 as a HIF-1α-interacting protein and the effect of MYH9 on HIF-1α protein stabilization and ubiquitination. a Immunoprecipitated proteins of SNU-387 and SNU-387-LR cells were separated by SDS-PAGE and visualized by silver staining. b The top 10 protein scores from unique proteins that interact with HIF-1α were identified by LC-MS/MS analysis. c Western blot was conducted to assess the MYH9 expression in WT and HCC LR cells. d The interaction between endogenous MYH9 and HIF-1α was tested in Hep-3B-LR cells. IgG was used as a control. The MYH9 protein is indicated by the red arrow. e Colocalization of HIF-1α (green) with MYH9 (red) was observed by confocal microscopy in HuH-7 and HuH-7-LR cells. The nuclei were stained with DAPI. f Western blot analysis was used to assess the protein expression of MYH9 and HIF-1α following MYH9 knockdown (shMYH9) or shNC in HCC LR cells. g Western blot analysis was used to assess the protein expression of MYH9 and HIF-1α in WT HCC cells with MYH9 overexpression (MYH9) or vector. SNU-387-LR cells were treated with shMYH9 or shNC. h While Hep-3B cells were treated with MYH9 or vector (i). Both cell lines were then exposed to 10 μg/mL CHX for the designed time. The protein levels of HIF-1α and MYH9 were determined via western blot analysis. j SNU-387-LR and HuH-7-LR cells were treated with shMYH9 or shNC, and then exposed to MG132 (5 μM) for 2 hours. Western blot analysis was performed to assess the protein levels of HIF-1α and MYH9. k HuH-7-LR cells were cotransfected with shMYH9 or shNC and Ub (HA-tag) plasmids. l SNU-387 cells were cotransfected with MYH9 overexpression or vector and HA-Ub plasmids. Anti-HIF-1α antibody was used to isolate endogenous HIF-1α and anti-HA antibody was used to detect bound Ub. Exogenous MYH9 and HIF-1α expression in whole cell lysates was detected. m HuH-7-LR cells with shMYH9 or shNC were stably overexpressed with HIF-1α (wt) or mutated HIF-1α (P402A, P564A), HIF-1α (mut), which were with Flag-tag. Flag and MYH9 protein were determined by western blot, while the relative luciferase activity of HRE was analyzed (n). The data were presented as mean ± SD of three individual experiments. The Student’s t test was used for comparisons. **p < 0.01, ****p < 0.0001. Scale bar: 5 μm. IP immunoprecipitation, MYH9 non-muscle myosin heavy chain 9, WT wild type, LR lenvatinib resistant, HIF-1α hypoxia-inducible factor-1α, HCC hepatocellular carcinoma, CHX cycloheximide, HRE hypoxia-responsive element, ns nonsignificant

The proteasome-specific inhibitor MG132 rescued the HIF-1α protein from degradation in MYH9 knockdown cells (Fig. 3j; Supplementary Fig. 3h). Furthermore, MYH9 knockdown significantly enhanced polyubiquitination-induced HIF-1α degradation (Fig. 3k), while MYH9 overexpression significantly inhibited ubiquitin conjugation to HIF-1α (Fig. 3l). According to a previous study,9 we developed a stable HIF-1α mutant that is oxygen-independent by replacing two prolines with alanine in the oxygen-dependent degradation domain (ODDD) (P402 and P564). Mutated HIF-1α (P402 and P564) significantly rescued the protein level of HIF-1α in MYH9 knockdown cells (Fig. 3m). MYH9 knockdown had no influence on the transcriptional activity of the mutated HIF-1α (Fig. 3n). These results suggested that MYH9 stabilized the HIF-1α protein by preventing its ubiquitination and subsequent proteasomal degradation.

To ascertain the involvement of MYH9 in LR and cancer stemness of HCC cells, we further conducted a series of functional experiments. MYH9 knockdown effectively suppressed colony formation under lenvatinib treatment (Fig. 4a; Supplementary Fig. 4a) and migration (Fig. 4c; Supplementary Fig. 4c) and promoted lenvatinib-induced apoptosis (Fig. 4e) in HCC LR cells. In addition, MYH9 knockdown resulted in reduced hepatosphere formation (Fig. 4g; Supplementary Fig. 4e) and decreased mRNA expression of CSC markers (Fig. 4i; Supplementary Fig. 4g) and glycolysis-related genes (Fig. 4k; Supplementary Fig. 4i) in HCC LR cells. Conversely, MYH9 overexpression significantly promoted colony formation (Fig. 4b; Supplementary Fig. 4b) and migration (Fig. 4d; Supplementary Fig. 4d) and suppressed lenvatinib-induced apoptosis (Fig. 4f) in HCC cells. MYH9 overexpression also resulted in the opposite trend in hepatosphere formation (Fig. 4h; Supplementary Fig. 4f), expression of CSC mRNA expression of CSC markers (Fig. 4j; Supplementary Fig. 4h) and glycolysis-related genes (Fig. 4l; Supplementary Fig. 4j) in HCC cells compared to HCC LR cells. Importantly, MYH9 knockdown significantly sensitized HCC LR cells to lenvatinib in vivo (Fig. 4m, n). There was no notable variance in the body weights of mice (Fig. 4o). The expression of MYH9, HIF-1α and the proliferation marker Ki67 was suppressed by the combination of MYH9 shRNA and lenvatinib (Fig. 4p; Supplementary Fig. 4k). In summary, we demonstrated that MYH9 could stabilize HIF-1α and promote LR, glycolysis and cancer stemness in HCC.

Fig. 4

MYH9 promotes LR and cancer stemness in HCC. HCC LR cells with MYH9 knockdown (shMYH9) or shNC. a and WT HCC cells with MYH9 overexpression or vector (b) were cultured with the indicated lenvatinib concentrations. Following 14 days, crystal violet staining was used to determine the remaining cells. The influence of shNC and shMYH9 on HCC LR cell stemness was demonstrated through assessments of migration (c), flow cytometry (e), in vitro self-renewal (g) and the mRNA expression of stemness markers (i) and glycolysis driver genes (k). The influence of vector and MYH9 overexpression on WT HCC cell stemness was demonstrated through assessments of migration (d), flow cytometry (f), in vitro self-renewal (h) and the mRNA expression of stemness markers (j) and glycolysis driver genes (l). m HuH-7-LR cells with shMYH9 or shNC xenografted nude mice were treated with vehicle or Len (lenvatinib, 30 mg/kg) for 3 weeks after the tumor reached an average size of 50–100 mm3 (n = 5 per group). After 21 days, the mice were sacrificed. The isolated tumors were photographed. Tumor volume (n) and mouse weight of each group (o) were recorded every other week. p Immunohistochemistry was used to assess the expression of MYH9 and HIF-1α in the xenografts post-treatment. The data of cell functional assays and RT-qPCR analysis were presented as mean ± SD of three individual experiments, and the data of animal experiments were presented as mean ± SEM. The Student’s t-test was used for comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar: 100 μm., non-muscle myosin heavy chain 9; WT wild type, LR lenvatinib resistant, HIF-1α hypoxia-inducible factor-1α, HCC hepatocellular carcinoma, ns nonsignificant

Phosphorylated MYH9 (Ser1943) interacts with HIF-1α

Phosphorylation is a common modification of MYH9, and Ser1916 and Ser1943 are the two most reported phosphorylation sites.25 We generated S1916E and S1943E mutants (in which serine was substituted with glutamic acid to mimic the phosphorylated form) in constructs expressing myc-tagged MYH9. In the regulation of HIF-1α, phosphorylated MYH9 at Ser1943 had a more pronounced effect on the upregulation of HIF-1α protein than phosphorylated MYH9 at Ser1916 did (Fig. 5a). p-MYH9 (Ser1943) was also increased in HCC LR cells (Fig. 5b; Supplementary Fig. 5a). To investigate the interaction between p-MYH9 (Ser1943) and HIF-1α protein, we performed immunofluorescence staining and found that p-MYH9 (Ser1943) colocalized with HIF-1α in the cytoplasm (Fig. 5c). The protein expression data indicated that p-MYH9 (Ser1943) was primarily localized in the cytoplasm, while HIF-1α was present in both the nucleus and cytoplasm of HCC LR cells (Supplementary Fig. 5b). Additionally, we generated an S1943A mutant (in which serine was substituted with alanine to mimic the unphosphorylated form) in a construct expressing myc-tagged MYH9. The MYH9-S1943E mutant, had a more potent capability to bind HIF-1α than the MYH9-S1943A mutant (Fig. 5d). Moreover, MYH9-S1943E had a potent deubiquitination effect on the HIF-1α protein, while MYH9-S1943A did not (Fig. 5e). Furthermore, the wild type MYH9 and MYH9-S1943E did not affect the degradation of mutated HIF-1α (P402, P564) (Fig. 5f). These results indicated that p-MYH9 (Ser1943) dominates the deubiquitination regulation of the HIF-1α protein.

Fig. 5

Phosphorylated MYH9 at Ser1943 interacts with HIF-1α and promotes LR and stemness in HCC. a HIF-1α (Flag) and MYH9 (myc) plasmids or phosphorylation mimics at S1943 (S1943E) and S1916 (S1916E) were transfected into HEK-293T cells. Anti-myc antibody was used to detect MYH9, and anti-Flag antibody represented HIF-1α. b The expression of p-MYH9 (Ser1943) in WT and LR HCC cells was detected using western blotting. c Colocalization of HIF-1α (green) with p-MYH9 (Ser1943) (red) was observed by confocal microscopy in HuH-7 and HuH-7-LR cells. The nuclei were stained with DAPI. Scale bar: 5 μm. d HIF-1α interactions with MYH9, MYH9-S1943E and its dephosphorylation mimic (MYH9-S1943A) were tested. Vector (–), MYH9 (wt) and its mutants (myc-tag) were transfected into HEK-293T cells respectively along with HIF-1α plasmid (Flag-tag). The immunoprecipitation assay was conducted to isolate myc-tagged vector and MYH9 protein, and anti-Flag antibody was used to detect bound HIF-1α protein. The expression of Flag-HIF-1α, myc-MYH9 and β-actin in whole cell lysates was verified. e MYH9, MYH9-S1943E and MYH9-S1943A (myc-tag) were transfected into HEK-293T cells respectively along with Ub plasmid (HA-tag). The immunoprecipitation assay was conducted to isolate endogenous HIF-1α and anti-HA antibody was used to detected bound Ub. The expression levels of HIF-1α and my-tagged MYH9 in whole cell lysates was verified. f Vector (–), MYH9 (wt) and its mutants (myc-tag) were transfected into HEK-293T cells respectively along with mutated HIF-1α (P402A, P564A) (HIF-1α (mut)) plasmids (Flag-tag). 10 μg/mL CHX was added at the designed time. Western blot was used to assess the expression of HIF-1α and MYH9. g The expression of p-MYH9(Ser1943) and MYH9 in multiple HCC cell lines was analyzed by western blot. h The IC50 of multiple HCC cell lines was measured by a CCK-8 assay to determine cell viability. Each point on the dose-response curves represents five technical replicates. i The correlation between IC50 and the relative p-MYH9 (Ser1943) protein expression was examined. j Schematic of the construction of PDXs, primary cells and PDOs models from human HCC tissues. Created with BioRender.com. k Selection of patients with high and low expression of p-MYH9 (Ser1943) were detected by IHC. Scale bar: 100 μm. l NOD/SCID mice bearing PDX with high and low expression of p-MYH9 (Ser1943) were treated with vehicle or lenvatinib for 2 weeks (n = 5 per group). The isolated tumors were photographed. m Tumor volume of each group was measured twice a week. n Primary cells with high or low expression of p-MYH9 (Ser1943) were treated with lenvatinib under different concentrations for 48 hours in vitro. Alive cell numbers of each group were counted. Relative cell number was calculated. o The relative representative bright-field microscopy images of PDO with high or low expression of p-MYH9 (Ser1943) were photographed on Day 3, 5 and 7. p HE staining of PDO and IHC staining of GPC3 and Ki67 in PDO with high expression of p-MYH9 (Ser1943) were detected. Scale bar: 50 μm. The relevance analysis was conducted using Pearson’s correlation. The data of cell functional assays were presented as mean ± SD of three individual experiments, and the data of animal experiment were presented as mean ± SEM. The Student’s t test was used for comparisons. **p < 0.01. MYH9 non-muscle myosin heavy chain 9, WT wild type, LR lenvatinib resistant, HIF-1α hypoxia-inducible factor-1α, HCC hepatocellular carcinoma, CHX cycloheximide, CCK-8 Cell Counting Kit-8, IP immunoprecipitation, PDX patient-derived xenograft PDO patient-derived organoid

Targeting p-MYH9 (Ser1943) reverses LR and suppresses cancer stemness in HCC

Since MYH9 stabilized HIF-1α and promoted LR and p-MYH9 (Ser1943) specifically interacted with HIF-1α, we wondered whether p-MYH9 (Ser1943) could be an effective target for reversing LR and suppressing cancer stemness. We measured the expression level of p-MYH9 (Ser1943) in 11 liver cancer cell lines (Fig. 5g), and calculated the IC50 values of the HCC cell lines for lenvatinib by CCK-8 detection (Fig. 5h). A strong positive correlation (r = 0.837) was observed between p-MYH9 (Ser1943) expression and the IC50 values of the HCC cell lines for lenvatinib (Fig. 5i), revealing that p-MYH9 (Ser1943) may play an important role in driving HCC primary resistance to lenvatinib. Moreover, we constructed patient-derived HCC models, including patient-derived tumor xenografts (PDX), primary cells and PDO from human HCC tissues (Fig. 5j). According to the level of p-MYH9 (Ser1943) expression (Fig. 5k), PDX models with low p-MYH9 (Ser1943) expression demonstrated significantly greater sensitivity to lenvatinib compared to PDX models with high p-MYH9 (Ser1943) expression (Fig. 5l, m; Supplementary Fig. 5c). The primary HCC cell models displayed similar results (Fig. 5n). Since organoids are an excellent model for assessing stemness, we also conducted experiments with PDO models. Our findings revealed that PDO models exhibiting higher levels of p-MYH9 (Ser1943) expression demonstrated faster expansion compared to PDO models with lower p-MYH9 (Ser1943) expression (Fig. 5o, p).

Due to the lack of specific inhibitors of p-MYH9 (Ser1943), we screened several inhibitors reported by previous studies26 (Supplementary Fig. 6a). Compared to other inhibitors, the casein kinase-2 (CK2) inhibitor CX-4945 effectively inhibited the expression of p-MYH9 (Ser1943) and HIF-1α (Fig. 6a) and the transcriptional activity of HIF-1α (Fig. 6b). Functional assays showed that CX-4945 suppressed hepatosphere formation (Fig. 6c), migration (Fig. 6d) and mRNA expression of CSC markers (Fig. 6e) in HCC LR cells. CX-4945 combined with lenvatinib had a synergistic effect on suppressing colony formation and proliferation in HuH-7-LR, SNU-387-LR cells and primary HCC cells (Fig. 6f, g; Supplementary Fig. 6b, c). In vivo assays further demonstrated that CX-4945 could sensitize HCC LR cells to lenvatinib and inhibit the growth of xenograft tumors with acceptable adverse effect (Fig. 6h, i; Supplementary Fig. 6d). Immunohistochemical analysis of the xenograft tumors showed that the expression of p-MYH9 (Ser1943) and HIF-1α was significantly decreased in the CX-4945 and combination groups (Fig. 6j). Ki67 was suppressed by the combination of CX-4945 and lenvatinib (Supplementary Fig. 6e). Results from orthotopic HCC models also showed similar trends (Fig. 6k, l; Supplementary Fig. 6f). These results suggested that CX-4945 could inhibit the cancer stemness of LR cells in vitro and in vivo, which further proved that p-MYH9 (Ser1943)-mediated HIF-1α stability regulated the cancer stemness of HCC LR cells.

Fig. 6

p-MYH9 (Ser1943) is a therapeutic target for stemness and LR, and predicts poor prognosis and LR in patients with HCC. a HuH-7-LR and SNU-387-LR cells were exposed to the indicated concentrations of CX-4945. Western blot was used to assess expression of p-MYH9 (Ser1943), MYH9 and HIF-1α (b) The effect of CX-4945 treatment on the relative luciferase activity of HRE in HCC LR cells was also investigated. The influence of CX-4945 on HCC LR cell stemness was used to assess in vitro self-renewal (c), migration (d), the mRNA expression of stemness markers (e) and colony formation (f). g HuH-7-LR and SNU-387-LR cells were treated with DMSO or 2 μM CX-4945 combined with the indicated concentration of lenvatinib. Differences in sensitivity to lenvatinib and CX-4945 were observed through cell count using trypan blue staining. h HuH-7-LR cell xenografted nude mice were administrated with vehicle, Len (lenvatinib, 30 mg/kg), CX-4945 (20 mg/kg) and their combination daily for 3 weeks after the tumor volume reached an average size of 50–100 mm3 (n = 5 per group). Then, the mice were sacrificed and the isolated tumors were photographed. i Tumor volume of each group was measured twice a week. j IHC was used to detect the expression of HIF-1α and p-MYH9 (Ser1943) in the xenografts following treatment. k HuH-7-LR cell was injected into the sub-capsular space of right liver lobe nude mice to establish the orthotopic tumor model. After 1 week, mice were treated with vehicle, Len (lenvatinib, 30 mg/kg), CX-4945 (20 mg/kg) and their combination daily for 3 weeks (n = 5 per group). Then, the mice were sacrificed and the isolated livers were photographed. l Tumor volume was calculated. m p-MYH9 (Ser1943) expression was detected by IHC in normal adjacent to tumor tissue (NAT) and HCC samples from 171 patients. Based on IHC staining of p-MYH9 (Ser1943), patients were categorized into two groups, and subsequent overall survival curves (n) were plotted. o Patients with HCC treated with lenvatinib were separated into two groups depending on their p-MYH9 (Ser1943) protein expression, and subsequent progression-free survival curves were plotted. p Positive correlation between p-MYH9 (Ser1943) and HIF-1α IHC results in 215 HCC samples. The data of cell functional assays, Luciferase reporter assay and RT-qPCR analysis were presented as mean ± SD of three individual experiments, and the data of animal experiments were presented as mean ± SEM. The Student’s t-test was used for comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar: 100 μm. MYH9 non-muscle myosin heavy chain 9, LR lenvatinib resistance, HIF-1α hypoxia-inducible factor-1α, HRE hypoxia-responsive element, HCC hepatocellular carcinoma, IHC immunohistochemistry, DMSO dimethyl sulfoxide, ns nonsignificant

p-MYH9 (Ser1943) is upregulated in HCC and predicts poor prognosis and LR in patients with HCC

To verify the clinical significance of p-MYH9 (Ser1943), we measured its expression in 171 HCC and normal adjacent tumor tissues by IHC and found that the p-MYH9 (Ser1943) level was significantly upregulated in HCC, which indicates that p-MYH9 (Ser1943) is a potential therapeutic target in HCC (Fig. 6m; Supplementary Fig. 6g). Survival analysis showed that higher levels of p-MYH9 (Ser1943) expression were correlated with worse patient outcomes in terms of both overall survival and recurrence-free survival (Fig. 6n; Supplementary Fig. 6h). In 70 HCC patients taking lenvatinib, progression-free survival was longer in the low p-MYH9 (Ser1943) group (Fig. 6o). A univariate Cox proportional hazard analysis revealed that poor differentiation, AFP > 400 ng/mL, microvascular invasion, AJCC stages III-IV and high expression of p-MYH9 (Ser1943) were positively related to shorter overall survival time of HCC patients. A multivariate Cox proportional hazard analysis showed that p-MYH9 (Ser1943) expression and microvascular invasion served as independent prognostic factors for HCC patients (Supplementary Table 6). Similarly, high expression of p-MYH9 (Ser1943) was also an independent prognostic factor for HCC patients taking lenvatinib (Supplementary Table 7). Moreover, the protein expression of p-MYH9 (Ser1943) and HIF-1α was assessed in 215 HCC samples. The results showed a strong positive correlation between p-MYH9 (Ser1943) and HIF-1α (Fig. 6p; Supplementary Fig. 6i), further confirming the relationship between the two proteins. Thus, clinically, high p-MYH9 (Ser1943) expression in HCC could predict poor prognosis and LR.

p-MYH9 (Ser1943) recruits USP22 to deubiquitinate HIF-1α and promote LR

Because USP22 is an important deubiquitinating enzyme for HIF-1α9, we speculated that p-MYH9 (Ser1943) might recruit USP22 to deubiquitinate and stabilize HIF-1α. To this end, we first demonstrated that USP22 knockdown could abrogate high MYH9-induced HIF-1α upregulation and promote ubiquitin-mediated degradation of HIF-1α in the HCC parent type or LR cells (Fig. 7a, b). Additionally, the endogenous interaction between MYH9, HIF-1α and USP22 was confirmed in HuH-7-LR cells (Fig. 7c). Furthermore, we performed co-IP assays in HEK-293T cells and found that MYH9 could interact with USP22, especially MYH9-S1943E, but not MYH9-S1943A (Fig. 7d; Supplementary Fig. 7a). That suggested that p-MYH9 (Ser1943) could specifically recruit USP22 to deubiquitinate HIF-1α. Finally, we evaluated the efficacy of USP22 inhibition in reversing LR. USP22 knockdown or USP22 inhibition effectively promoted the sensitivity of HCC LR cells and primary HCC cells to lenvatinib (Fig. 7e and Supplementary Fig. 7c). In vivo, the USP22 inhibitor (S02) significantly reversed LR in HuH-7-LR cells (Fig. 7f, g). The weights were not significantly different (Supplementary Fig. 7b). The expression levels of HIF-1α and Ki67 were further decreased in the S02 groups (Fig. 7h). Interestingly, S02 had little effect on lenvatinib sensitivity in HuH-7 cells (Supplementary Fig. 7d–f), which further supported the significant role of USP22 in HCC LR cells. The above results indicated that p-MYH9 (Ser1943) could recruit USP22 to deubiquitinate HIF-1α and promote LR (Fig. 7i).

Fig. 7

p-MYH9 (Ser1943) recruits USP22 to deubiquitinate HIF-1α and promote LR. a HuH-7 cells with vector and MYH9 overexpression and HuH-7-LR cells, accompanied by USP22 knockdown (shUSP22) or not, the altered protein expression of MYH9, HIF-1α and USP22 was detected through western blot. b Both the shUSP22 and Ub (HA-tagged) plasmids were transfected into HuH-7 cells with MYH9 overexpression and HuH-7-LR cells. Anti-HIF-1α antibody was used to isolated the HIF-1α protein and anti-HA antibody was used to detected bound Ub. MYH9, HIF-1α and USP22 protein expression levels were determined in whole cell lysates. c The interaction of endogenous MYH9, HIF-1α and USP22 was detected in HuH-7-LR cells. IgG was used as a control. d USP22 (Flag-tag) plasmid was transfected with MYH9-S1943E (myc-tag) or MYH9-S1943A (myc-tag) plasmids in HEK-293T cells. The interaction of myc protein and Flag protein was detected. IgG was used as a control. e HuH-7-LR cells were transfected with shNC or shUSP22, and cells were treated with DMSO or 2 μg/mL USP22 inhibitor, S02 combined with the indicated concentration of lenvatinib. Variations in lenvatinib and USP22 inhibitor sensitivity were detected by trypan blue staining-based cell count. The data were presented as mean ± SD of three individual experiments. f HuH-7-LR cell xenografted nude mice were fed lenvatinib (30 mg/kg), S02 (10 mg/kg) and their combination daily for three weeks after the tumor volume reached an average size of 50–100 mm3 (n = 6 per group). The isolated tumors were photographed. g Tumor volume of each group was measured twice a week. After 21 days, the mice were sacrificed. The data were presented as mean ± SEM. h Immunohistochemistry was used to detect the expression of HIF-1α and Ki67 in the xenografts after various treatments. i Graphical summary of the molecular mechanisms involving MYH9, HIF-1α and USP22 in regulating of CSC properties in lenvatinib-resistant cells. The graph was created by Microsoft Powerpoint. The Student’s t-test was used for comparisons. *p < 0.05, ***p < 0.001. Scale bar: 100 μm. MYH9 non-muscle myosin heavy chain 9, LR lenvatinib resistant, HIF-1α hypoxia-inducible factor-1α, USP22 ubiquitin-specific protease 22, DMSO dimethyl sulfoxide, CSC cancer stem cell