N6-methyladenosine-modified USP13 induces pro-survival autophagy and imatinib resistance via regulating the stabilization of autophagy-related protein 5 in gastrointestinal stromal tumors

Abstract

Secondary resistance to imatinib (IM) represents a major challenge for therapy of gastrointestinal stromal tumors (GISTs). Aberrations in oncogenic pathways, including autophagy, correlate with IM resistance. Regulation of autophagy-related protein 5 (ATG5) by the ubiquitin-proteasome system is critical for autophagic activity, although the molecular mechanisms that underpin reversible deubiquitination of ATG5 have not been deciphered fully. Here, we identified USP13 as an essential deubiquitinase that stabilizes ATG5 in a process that depends on the PAK1 serine/threonine-protein kinase and which enhances autophagy and promotes IM resistance in GIST cells. USP13 preferentially is induced in GIST cells by IM and interacts with ATG5, which leads to stabilization of ATG5 through deubiquitination. Activation of PAK1 promoted phosphorylation of ATG5 thereby enhancing the interaction of ATG5 with USP13. Furthermore, N⁶-methyladenosine methyltransferase-like 3 (METTL3) mediated stabilization of USP13 mRNA that required the m⁶A reader IGF2BP2. Moreover, an inhibitor of USP13 caused ATG5 decay and co-administration of this inhibitor with 3-methyladenine boosted treatment efficacy of IM in murine xenograft models derived from GIST cells. Our findings highlight USP13 as an essential regulator of autophagy and IM resistance in GIST cells and reveal USP13 as a novel potential therapeutic target for GIST treatment.

Fig. 1. ATG5 promotes IM resistance and the malignant proliferation of GIST.

A Heatmap representation of differentially expressed proteins in patients with IM-resistant and IM-sensitive GIST. B Volcano plot for differentially expressed proteins in patients with IM-resistant and IM-sensitive GIST. C IHC for ATG5 and CD117. D Protein level of ATG5 in human GIST tissue from IM-resistant and IM-sensitive patients. E Quantification of ATG5 protein level in GIST cells. F RT-qPCR analyses of ATG5 mRNA levels in sensitive and resistant GIST tissues. G Quantitative RT-PCR analysis of ATG5 target genes from IM-sensitive and IM-resistant cells. H, I CCK8 proliferation assay verified the effect of ATG5 expression levels on the sensitivity of GIST cells to IM. J Kaplan–Meier plot of progression-free survival by ATG5 expression. K Receiver operating characteristic analysis of the risk of patients with IM-sensitive and IM-resistant GIST. L, M Cell cycle distribution and apoptosis rate of GIST cells transfected with vector, ATG5, shCtrl, or shATG5. Error bars represent the mean (n = 3) ±S.D. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 2. USP13 maintains ATG5 stability and interacts with ATG5.

A Flag-labeled DUBs (USP3, USP13, USP32, and OTUD4) were expressed in HEK293T cells, and cell lysates were analyzed by IP with Flag beads followed by western blotting with antibodies against ATG5 and Flag. B Increasing amounts of Flag-labeled wild-type USP13 or USP13-C345A were transfected into HEK293T cells, and cell lysates were analyzed by western blotting with antibody against ATG5. C Western blot analysis of ATG5 from IM-sensitive and IM-resistant cell lines. D GIST-T1R cells transfected with two independent USP13 shRNA were treated with or without the proteasome inhibitor MG-132 (20 μM for 8 h) and then USP13 and ATG5 were analyzed. E Western blotting analysis of ATG5 levels in GIST-T1R cells transfected with USP13 shRNA together with either shRNA-resistant (sh-res) Flag-labeled wild-type USP13 or USP13-C345A. F, G HEK293T cells were co-transfected with His-labeled ATG5 and Flag-labeled wild-type USP13 or USP13-C345A, treated with CHX (40 μg/ml), collected at the indicated times, and then subjected to western blotting with antibodies against His and Flag. Quantification of ATG5 levels relative to β-actin are shown. H, I GIST-T1R cells stably expressing control shRNA or shRNA-USP13 were treated with CHX (40 μg/ml), harvested at the indicated times, and then subjected to western blotting with antibodies against ATG5 and USP13. Quantification of ATG5 levels relative to β-actin are shown. J HEK293T cells were transfected with His-ATG5 alone or in combination with Flag-labeled wild-type USP13 or USP13-C345A. Cell lysates were analyzed by IP with Flag beads followed by western blotting with antibodies against His and Flag. a, vector; b, Flag-USP13; c, Flag-USP13-C345A. K Confocal images showing colocalization of USP13 (red) and ATG5 (green) in GIST cells. Nuclei were counterstained with DAPI (blue). L, M Cell lysates from GIST-T1R cellls were analyzed by IP using antibodies against USP13 and ATG5, then subjected to western blotting analysis. IgG was used as the isotype control. N Purified Flag-labeled wild-type USP13 or USP13-C345A were incubated with GST or GST-ATG5 coupled to glutathione-sepharose beads. Proteins retained on sepharose were then subjected to western blotting with the indicated antibodies. Recombinant GST-ATG5 was purified from bacteria and analyzed by SDS-PAGE and Coomassie blue staining. O Schematic representation of full-length (FL) Flag-labeled USP13, His-labeled ATG5, and various deletion mutants. P HEK293T cells were co-transfected with His-ATG5 and FL Flag-labeled USP13 or deletion mutants. Cell lysates were analyzed by IP with Flag beads followed by western blotting with antibodies against His and Flag. Q HEK293T cells were co-transfected with Flag-USP13 and His-labeled FL ATG5 or deletion mutants. Cell lysates were analyzed by IP with His beads followed by western blotting with antibodies against Flag and His. Error bars represent the mean (n = 3) ± S.D. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 3. USP13 deubiquitinates ATG5.

A GIST-T1S and GIST-T1R were co-transfected with HA-ubiquitin (HA-Ub), and cell lysates were subjected to denature-IP with ATG5 antibody, followed by western blotting with antibodies against HA and ATG5. Cells were treated with 20 μM MG-132 and with or without low-dose IM for 8 h before harvesting. B, C HEK293T cells or ICCs were co-transfected with His-ATG5, HA-ubiquitin (HA-Ub), and Flag-labeled wild-type USP13 or USP13-C345A, and cell lysates were subjected to denature-IP with His beads followed by western blotting with antibodies against HA and His. Cells were treated with 20 μM MG-132 for 8 h before harvesting. D GIST-T1R was co-transfected with the indicated shRNA and HA-Ub, and cell lysates were subjected to denature-IP with ATG5 antibody, followed by western blotting with antibodies against HA and ATG5. Cells were treated with 20 μM MG-132 for 8 h before harvesting. E Unubiquitylated or ubiquitylated His-ATG5 was incubated with wild-type GST-USP13 or GST-USP13-C345A coupled to glutathione-sepharose beads. His-ATG5 was subjected to denature-IP with His beads followed by western blotting with antibodies against HA and His. Recombinant GST-USP13 or GST-USP13-C345A were analyzed by SDS-PAGE. F GIST-T1S cells were co-transfected with His-ATG5, Flag-USP13, and HA-Ub Lys0, Lys48-only, or Lys63-only plasmids, and then the ATG5 ubiquitylation linkage was analyzed. G GIST-T1S cells transfected with wild-type Ub or Ub-Lys48R were cultured for 72 h in the presence of control shRNA or USP13 shRNA. Cell lysates were analyzed by western blotting using antibodies against ATG5 and USP13. H HEK293T cells were transfected with the vector plasmid or Flag-USP13, HA-Ub and wild-type His-ATG5 or K-to-R mutants and treated with a low-dose of IM. Samples were subjected to denature-IP with anti-Flag beads and then analyzed by immunoblot with an anti-HA or anti-Flag antibody. I HEK293T transfected with wild-type Ub, wild-type His-ATG5 or His-ATG5-K5R were cultured for 72 h in the presence of control shRNA or USP13 shRNA. Cell lysates were analyzed by western blotting using antibodies against ATG5 and USP13.

Fig. 4. USP13 enhances IM-resistance and malignant proliferation by the induction of protective autophagy in vitro and in vivo.

A, B GIST-T1S cells were transfected with a USP13 overexpression vector and empty control (NC), and GIST-T1R cells were transfected with USP13 shRNA and control (NC). Inhibition curves of GIST cells after treatment are indicated. C Western blot detected autophagy element expression levels in GIST-T1S and GIST-T1R cells after transfection as indicated. D Apoptosis-related protein levels were detected by western blot. E–G The autophagy flow and autophagosome in GIST cells were detected by confocal microscopy and transmission electron microscopy. H GIST-T1S was co-transfected with His-ATG5, HA-Ub, and Flag-USP13 in the absence or presence of Spautin-1 (1 μM) or a low-dose of IM and cell lysates were subjected to denature-IP with His beads followed by western blotting with antibodies against HA and His. Cells were treated with 20 μM MG-132 for 8 h before harvesting. I Western blot analysis of ATG5 in GIST-T1S treated with Spautin-1 (1 μM) or vehicle with or without MG-132. J GIST-T1S cells were treated with Spautin-1 (1 μM) or vehicle for 24 h, followed by CHX (40 μg/ml), harvested at the indicated times, and then subjected to western blotting with antibodies against ATG5. SE, short exposure; LE, long exposure. K–M Representative images from tumors in nude mice bearing GIST-T1S or GIST-T1R cells in different groups and the expression of autophagy markers monitored by western blot (n = 6 mice/ group). Scale bars: 1 cm. N–P Representative images of tumors in nude mice formed by the GIST-T1R cells in the different subgroups and the expression of autophagy and apoptosis markers monitored by western blot. Error bars represent the mean (n = 3) ± S.D. *P < 0.05, **P < 0.01, ***P < 0.001, respectively.

Fig. 5. PAK1 promotes ATG5’s binding with and deubiquitination by USP13.

A, B GIST-T1R cells were transfected with control shRNA or PAK1 shRNA, and cell lysates were analyzed by IP using antibodies against USP13 or ATG5, then subjected to western blot analysis. IgG was used as the isotype control. C, D GIST-T1R cells were treated with or without PAK1 inhibitor, and cell lysates were analyzed by IP using antibodies against USP13 and ATG5, then subjected to western blotting analysis. IgG was used as the isotype control. E, F HEK293T cells were transfected with Flag-USP13, His-ATG5, and wild-type PAK1 (WT) or PAK1 (K420R). Cell lysates were analyzed by IP and western blots were performed as indicated. G, H HEK293T cells were transfected with Flag-USP13, wild-type His-ATG5, His-ATG5-101A, His-ATG5-101D, or Myc-PAK1. Cell lysates were analyzed by IP and western blots were performed as indicated. I HEK293T cells were transfected with HA-Ub and Flag-USP13 plus control shRNA or PAK1 shRNA for 36 h. Then the cells were treated with MG-132 for 8 h. Cell lysates were denature-IP using an anti-His antibody and then analyzed by western blotting using an anti-HA antibody. J HEK293T cells were transfected with HA-Ub, and wild-type Myc-PAK1 or Myc-PAK1(K420R), and then treated with MG-132 for 8 h. Cell lysates were denature-IP using an anti-His antibody then analyzed by western blotting using an anti-HA antibody. K HEK293T cells were transfected with HA-Ub, Myc-PAK1, wild-type His-ATG5 His-ATG5-101A, or His-ATG5-101D, and then treated with MG-132 for 8 h. Cell lysates were denature-IP using an anti-His antibody then analyzed by western blotting using an anti-HA antibody. L HEK293T cells were transfected with HA-Ub, His-ATG5, Flag-USP13 with or without PAK1 inhibitor for 36 h. Then the cells were treated with MG-132 for 8 h. Cell lysates were denature-IP using an anti-HA antibody and then analyzed by immunoblotting.

Fig. 6. Knockout of PAK1 reverses the effect of USP13 in vitro and in vivo.

A, B The expression of autophagy and apoptosis marker proteins monitored by western blot. C Inhibition curves of GIST cells after treatment as indicated. D The effect of USP13 expression levels on the proliferation of GIST cells was examined by clone formation assay. E–G Cell cycle distribution, apoptosis rate, proliferation and the autophagy flow were assessed in GIST-T1S cells transduced with empty vector or USP13, and reconstituted with shCtrl or shPAK1. Error bars represent the mean (n = 3) ± S.D. *P < 0.05, **P < 0.01, ***P < 0.001, respectively. H–I Bioluminescence imaging of xenografted tumors and and quantitative analysis of tumor volume. Error bars represent the mean (n = 3) ± S.D. *P < 0.05, **P < 0.01, ***P < 0.001, respectively.

Fig. 7. METTL3-mediated m6A modification of USP13 mRNA maintains IGF2BP2-dependent stability.

A m⁶A peaks are enriched in the 3′-UTR region of USP13 mRNA as determined by m⁶A-seq. B–E qRT-PCR and western blot analysis of METTL3 and USP13 after METTL3 knockout or overexpression. F–G RNA immunoprecipitation-qPCR showing the enrichment of USP13 m⁶A after METTL3 knockout or overexpression. H The decay rate of USP13 mRNA after treatment with actinomycin D (2.5 mmol/L) for the indicated times with METTL3 knockout or overexpression. I Western blot analysis of ATG5 after METTL3 overexpression with or without shUSP13. J The mRNA levels in sh-Ctrl or sh-METTL3 GIST cells transfected with pGL3-USP13-3′-UTR wild-type or pGL3-USP13-3′-UTR-Mut1/2 reporters for 24 h. K–M USP13 mRNA levels in sh-Ctrl or sh-METTL3 GIST cells transfected with pcDNA-USP13-3′-UTR-WT/Mut1/2 for 24 h and then treated with actinomycin D for the indicated times determined by RT-qPCR. N The mRNA levels of USP13 in IGF2BP1/2/3 knockdown GIST cells were detected by qRT-PCR. O RNA immunoprecipitation-qPCR showing the enrichment of USP13 m⁶A after IGF2BP2 knockout. P The decay rate of USP13 mRNA after treatment with actinomycin D (2.5 mmol/L) for the indicated times with IGF2BP2 knockout. Q IGF2BP2 expression correlates positively with USP13 expression in GIST patients (linear regression). Error bars represent the mean (n = 3) ± S.D. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 8. Correlative expression of METTL3, USP13, PAK1 and ATG5 as prognostic indicators for clinical GIST.

A Representative images of IHC staining of METTL3, USP13, PAK1, and ATG5 in clinical GIST samples. Scale bars: 50 µm. B–E Correlation analyses of IHC data of GIST in (A). Statistical significance was determined by the chi-square test. A total of 42 GIST specimens were analyzed. F–H Kaplan–Meier analyses for GIST in (A). A total of 42 GIST specimens were analyzed. I The mechanistic scheme of this study.