USP29-mediated HIF1α stabilization is associated with Sorafenib resistance of hepatocellular carcinoma cells by upregulating glycolysis

Understanding the mechanisms underlying evasive resistance in cancer is an unmet medical need to improve the efficacy of current therapies. In hepatocellular carcinoma (HCC), aberrant expression of hypoxia-inducible factor 1 α (HIF1α) and increased aerobic glycolysis metabolism are drivers of resistance to therapy with the multi-kinase inhibitor Sorafenib. However, it has remained unknown how HIF1α is activated and how its activity and the subsequent induction of aerobic glycolysis promote Sorafenib resistance in HCC. Here, we report the ubiquitin-specific peptidase USP29 as a new regulator of HIF1α and of aerobic glycolysis during the development of Sorafenib resistance in HCC. In particular, we identified USP29 as a critical deubiquitylase (DUB) of HIF1α, which directly deubiquitylates and stabilizes HIF1α and, thus, promotes its transcriptional activity. Among the transcriptional targets of HIF1α is the gene encoding hexokinase 2 (HK2), a key enzyme of the glycolytic pathway. The absence of USP29, and thus of HIF1α transcriptional activity, reduces the levels of aerobic glycolysis and restores sensitivity to Sorafenib in Sorafenib-resistant HCC cells in vitro and in xenograft transplantation mouse models in vivo. Notably, the absence of USP29 and high HK2 expression levels correlate with the response of HCC patients to Sorafenib therapy. Together, the data demonstrate that, as a DUB of HIF1α, USP29 promotes Sorafenib resistance in HCC cells, in parts by upregulating glycolysis, thereby opening new avenues for therapeutically targeting Sorafenib-resistant HCC in patients.


Supple. Figure 1. Establishment of Sorafenib-resistant cell lines and RNA sequencing.
(a) IC50 for Sorafenib responsiveness of different HCC cell lines. Different patient-derived HCC cell lines were treated with increasing doses of Sorafenib, and the IC50 values for cell growth inhibition by Sorafenib were determined. Hep3B and Huh7 were selected as two Sorafenib-responsive and HLE and SNU 398 as moderate Sorafenib-resistant HCC cell lines. Suppl. Figure 2. HIF2a is not required for Sorafenib resistance. (a,b) Validation of knockdown of HIF1a with ON-TARGET siHIF1a in Sorafenib-resistant Huh7-IR (a) and Huh7-CR (b) cells. Huh7-IR and Huh7-CR cells were transfected with siCtrl or ON-TARGET siHIF1a, and knockdown efficiency was validated by immunoblotting. Culture of the cells in hypoxia was used as a positive control for high HIF1a expression. β-Tubulin was used as loading control. Results represent three independent experiments.
(c-f) HIF2a's loss of function has no effect on cell death with Sorafenib treatment. Huh7-IR (c) or Huh7-CR (e) cells were transfected with siCtrl or siHIF2a and treated with DMSO or with 6µM or 9µM Sorafenib for 2 weeks. Colony formation assays (c,e) and quantification of colony formation by crystal violet staining (d,f) did not revealed any effect on colony formation by the siRNA-mediated depletion of HIF2a. The ratio of cell viability between HIF2a-deficient and HIF2a-wildtype cells is given in blue numbers (d,f). n = 3 independent replicates. ns = not significant; Student's t-test. (g,h) Validation of knockdown of HIF2a with ON-TARGET siHIF2a in Huh7-IR (g) and Huh7-CR (h) cells. Huh7-IR and Huh7-CR cells were transfected with siCtrl or ON-TARGET siHIF2a, and knockdown efficiency was validated by immunoblotting. Culture in hypoxia was used as a positive control. β-Tubulin was used as loading control. Results represent three independent experiments. Suppl. Figure 3. USP29 has a positive correlation with HIF1a. (a) Identification of USP29 as a DUB of HIF1a. A mini-screen with siRNAs against a selected panel of DUBs in Sorafenib-resistant HLE cells revealed that USP29 is one of the most critical players in stabilizing HIF1a protein, as revealed by immunoblotting for HIF1a. Immunoblotting for β-Tubulin was used as loading control. Results represent three independent experiments. (b) Knockdown efficiencies of the siRNAs used in (a). Different ON-TARGET siRNAs targeting USP8, USP28, USP29, USP36, USP37, UCHL1 were transfected into HLE cells, and quantitative RT-PCR analysis were conducted to determine knock down efficiencies. n = 2 independent replicates. **, P < 0.01; ***, P < 0.001; Student's t-test.
(c) Loss of USP29 results in instability of HIF1a protein in Sorafenib-resistant SNU398 cells. Two distinct siRNAs against USP29 (siUSP29#1 and siUSP29#2) were transfected into SNU398 cells and the protein levels of HIF1a, and USP29 was determined by immunoblotting analysis. Immunoblotting for β-Tubulin was used as loading control. Results represent three independent experiments. (d) USP29 stabilizes HIF1a. HEK-293T cells were transfected with a plasmid encoding for Flag-HIF1a together with increasing amounts of plasmid encoding for Myc-USP29. HIF1a and USP29 protein levels were determined by immunoblotting against Flag (Flag-HIF1a) and Myc (Myc-USP29). Immunoblotting for β-Tubulin was used as loading control. Results represent three independent experiments. (e) Knockdown efficiency of siUSP29 expression. HLE cells were transfected with siCtrl and ON-TARGET siUSP29 and the expression and localization of USP29 was monitored by immunofluorescence microscopy analysis. DAPI was used to visualize nuclei. Results represent three independent experiments. Scale bar, 132.5µm. (f) Myc-tagged USP29 binds Flag-tagged HIF1a. A plasmid encoding Flag-HIF1a was transfected into HEK-293T cells together with a plasmid encoding Myc-USP29. Anti-Flag antibodies were used to precipitate (IP) Flag-tagged HIF1a, and the immunoprecipitates were immunoblotted (IB) for Flag (Flag-HIF1a) and for Myc (Myc-USP29). Input represents 1/10 of the lysate used for the immunoprecipitations. Results represent three independent experiments. The proteasome inhibitor MG132 (5µM) was added in these experiments to prevent protein degradation. (g) Flag-tagged HIF1a binds Myc-tagged USP29. A plasmid encoding for Myc-tagged USP29 was transfected into HEK-293T cells together with a plasmid coding for Flag-tagged HIF1a. Anti-Myc antibodies were used to precipitate (IP) Myc-tagged USP29, and the immunoprecipitates were immunoblotted (IB) for Myc (Myc-USP29) and for Flag (Flag-HIF1a). Input represents 1/10 of the lysate used for the immunoprecipitations. Results represent three independent experiments. The proteasome inhibitor MG132 (5µM) was added in these experiments to prevent protein degradation. (h) Myc-tagged USP29 binds endogenous HIF1a. Plasmids encoding Myc-tagged empty vector or Myc-tagged USP29 were transfected into HEK-293T cells. Anti-Myc antibodies were used to precipitate (IP) Myc-tagged USP29, and the immunoprecipitates were immunoblotted (IB) for Myc (Myc-USP29) and for endogenous HIF1a. Input represents 1/10 of the lysate used for the immunoprecipitations. Results represent three independent experiments. The proteasome inhibitor MG132 (5µM) was added in these experiments to prevent protein degradation.

Suppl. Figure 4. USP29 deficiency promotes Sorafenib-induced cell death in intrinsic Sorafenib-resistant HCC cell lines. (a,b) Loss of USP29 induces cell death upon Sorafenib treatment in Hep3B and Huh7 cells.
Hep3B (a) and Huh7 (b) cells were transfected with siCtrl or ON-TARGET siUSP29 and treated with 6µM Sorafenib for 18 hours. Immunoblotting for cleaved PARP visualizes the levels of apoptosis. β-Tubulin was used as loading control. Results represent three independent experiments. (c-n) USP29 deficiency diminishes cell survival in various Sorafenib-resistant HCC cell lines. Sorafenib-resistant cell lines HLE (c,d,e), SNU398 (f,g,h), SNU449 (i,j,k) and SNU475 (l,m,n) were transfected with siCtrl or ON-TARGET siUSP29 and plated for colony formation assays under treatment with different concentrations of Sorafenib (0µM, 3µM, 6µM or 0µM, 6µM, 9µM) for 2 weeks. Quantification of colony formation by crystal violet staining (d,g,j,m) revealed decreased cell survival upon loss of USP29 expression and Sorafenib treatment in comparison with controls. The ratio of cell viability between USP29-deficient and USP29wildtype cells is given in blue numbers (d,g,j,m). n = 3 independent replicates. ns = not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; Student's t-test. Knockdown efficiencies were validated by immunoblotting for USP29 (e,h,k,n). β-Tubulin was used as loading control. Results represent three independent experiments.

Suppl. Figure 5. Hypoxia rescues USP29 deficiency induced cell death in response to Sorafenib treatment in Sorafenib-resistant HCC cell lines. (a-d)
Hypoxia rescues USP29 deficiency-induced cell death in response to Sorafenib in Huh7-IR and Huh7-CR cells. Huh7-IR (a) and Huh-CR (b) cells were transfected with siCtrl or ON-TARGET siUSP29 every two days and plated for colony formation assays under treatment with different concentrations of Sorafenib (0µM, 6µM, 9µM) under hypoxia culture condition (1% O2, 94% N2, 5% CO2) for 2 weeks, cell medium was refreshed every other day. Quantification of colony formation by crystal violet staining (b,d) revealed no decreased cell survival upon loss of USP29 expression and Sorafenib treatment in comparison with controls. The ratio of cell viability between USP29-deficient and USP29-wildtype cells is given in blue numbers (b,d). n = 3 independent replicates. ns = not significant; Student's t-test.

Supplementary Tables
Suppl . Table I