Fig 2: Circ-PKD2 promotes Atg13-mediated autophagy by inhibiting miR-646 to increase the sensitivity of cisplatin.Schematic diagram depicting the proposed model in which circ-PKD2 regulates autophagy and cisplatin sensitivity.

Fig 3: CircHADHA released ATG13 inhibition by competitively recruiting miR-361. (A) The sequence of potential binding sites in circHADHA (7mer-m8 position), seed region of miR-361, and 3’UTR of ATG13. (B–D) The luciferase reporter between miR-361 and ATG13. (B) miR-361 did not regulate luciferase reporter activities of ATG13 in control HCoEpiC. (C) miR-361 mimic reduced, whereas miR-361 inhibitor increased, the luciferase activity of ATG13 compared with corresponding controls in HCoEpiC. (D) miR-361 did not regulate luciferase reporter activities of ATG13 with mutant in 3’UTR in HCoEpiC. (E) The measurement of ATG13 relative expression transfected by miR-361 with or without circHADHA overexpression. (Left panels) miR-361 negatively regulated ATG13 expression compared with corresponding controls in HCoEpiC. (Central panel) circHADHA overexpression increased ATG13 expression compared with corresponding control in HCoEpiC. (Right panels) circHADHA overexpression elevated ATG13 expression regardless of the transfection with miR-361 or control mimic. (F) Overexpression of circHADHA increased LC3B II and ATG13 expression at protein level by Western blot. Three independent experiments were performed. **P < 0.01; *P < 0.05; ns, no significant difference.

Fig 4: TFG‐LC3C interaction is important for ULK1 activity in autophagy induction AHeLa cells were co‐transfected with untagged TFG and HA‐tagged LC3A, LC3B or LC3C. After 1 h of starvation, protein extracts were immunoprecipitated by using an anti‐TFG antibody or IgG as negative control. LC3 subfamily and TFG were analysed by WB.BHeLa cells were nutrient‐starved for the indicated time periods in the presence or absence of NH4Cl for the indicated timepoints of starvation. TFG, ACTIN and LC3B‐II (as indicator of NH4Cl activity) were shown in the WB. Densitometry analysis of TFG over ACTIN protein levels was analysed in the graph (right). Data are expressed by mean ± SEM. Statistical analysis was obtained by two‐way ANOVA followed by Tukey’s multiple comparison test. ns, not significant (n = 3 independent experiments).CTFG LIR motives, obtained from the autophagy database iLIR with their sequence and position, are shown.DSchematic illustration of TFG wild type (TFGwtLIR) and TFG mutants (TFGmutLIR2 and TFGmutLIR3) is shown. The aromatic F and the hydrophobic L residues of the TFGwtLIR were changed into two alanine residues for both the TFG mutants.EPulldown assays were performed by using purified GST‐GABARAP, GST‐GABARAPL1 and GST‐GABARAPL2 isoforms, each incubated together with HA‐TFGwtLIR‐ or HA‐TFGmutLIR3‐ or Myc‐ULK1‐transfected HeLa cell lysates, respectively. Myc‐ULK1 was used as positive control. Ponceau staining reports GST‐GABARAP isoforms and GST‐GFP used as negative control. WB analyses show Myc‐ULK1, HA‐TFGwtLIR and HA‐TFGmutLIR3. Densitometry analysis of HA‐TFG normalized over GST‐tagged GABARAP isoforms (Ponceau) was performed and reported (right). Data are mean ± SEM. Statistical analysis was performed by two‐way ANOVA followed by Tukey’s multiple comparison test. ns, not significant (n = 3 independent experiments).FHeLa cells were transfected with the indicated siRNA, and LC3C mRNA relative levels were analysed by qPCR. Values are expressed as mean ± SEM. Statistical analysis was performed by unpaired t‐test. ****P < 0.0001 (n = 3 independent experiments).GPulldown assays were performed by using purified GST‐LC3A, GST‐LC3B and GST‐LC3C isoforms, each incubated together with HA‐TFGwtLIR‐ or HA‐TFGmutLIR2‐ or HA‐NDP52‐transfected HeLa cell lysates, respectively. HA‐NDP52 was used as positive control. Ponceau staining reports the indicated GST‐LC3 isoforms and GST‐GFP used as negative control. WB analyses show HA‐NDP52, HA‐TFGwtLIR and HA‐TFGmutLIR2 in association with LC3A and LC3C (left) and HA‐NDP52, HA‐TFGwtLIR and HA‐TFGmutLIR2 in association with LC3B (right) are shown. Densitometry analysis of HA‐TFG normalized over GST‐tagged LC3 isoforms (Ponceau) was performed and reported as indicated (right). All values are expressed as mean ± SEM. Statistical analysis was performed by two‐way ANOVA followed by Tukey’s multiple comparison test. ns, not significant (n = 3 independent experiments).HHeLa cells harbouring endogenously HA‐tagged LC3C were transfected with HA‐TFGwtLIR or HA‐TFGmutLIR2. After 30’ of starvation, cells were lysed, and protein extracts were immunoprecipitated by using an anti‐TFG antibody or IgG as negative control. LC3C and TFG were analysed by WB.IHeLa cells stably expressing GFP‐ATG13 together with BFP‐empty vector (EV) or BFP‐TFGwtLIR or BFP‐TFGmutLIR3 were co‐transfected with mCherry‐SEC16 together with siRNA targeting endogenous TFG (siTFG 3’UTR) or with a non‐targeting control (siCTR). Endogenously interfered cells were used to perform live‐cell imaging analysis (Fig 5I). WB analysis to detect endogenous or BFP‐TFG protein levels is reported.JHeLa cells stably expressing GFP‐ATG13 together with BFP‐TFGwtLIR or BFP‐TFGmutLIR3 were transfected with siRNA targeting endogenous TFG (siTFG 3’UTR) or with a non‐targeting control (siCTR) as indicated. Endogenously interfered cells were used to perform live‐cell imaging analysis (Fig 5J). WB analysis to detect endogenous or BFP‐TFG protein levels is reported.KHeLa cells harbouring endogenously HA‐tagged LC3C were starved for 30’. Protein extracts were immunoprecipitated by using an anti‐HA antibody or IgG as negative control. ULK1, TFG and HA‐LC3C protein levels were analysed by WB.L, MHeLa cells were transfected with unrelated siRNA (siCTR) or siRNA targeting LC3B (siLC3B) (L) or GABARAP (siGABARAP) (M). (L‐M) After 30’ of starvation, cells were lysed and protein extracts were analysed by WB for indicated markers. ULK1 over actin is reported in the graphs (right). Data are shown as mean value ± SEM. Statistical analysis was performed by unpaired Student’s t‐test. ns, not significant (n = 3 independent experiments).NHeLa cells harbouring endogenously HA‐tagged GABARAPL1 were transfected with siRNA targeting GABARAPL1 or unrelated siRNA (siCTR). After 30’ of starvation, cells were lysed. Part of protein extract was immunoprecipitated with anti‐HA antibody to evaluate the downregulation of GABARAPL1. Remaining protein extract was analysed by WB as shown. ULK1 over ACTIN is also reported (right). Data are shown as mean value ± SEM. Statistical analysis was performed by unpaired Student’s t‐test. ns, not significant (n = 3 independent experiments).OHeLa cells harbouring endogenously HA‐tagged GABARAPL2 were transfected with siRNA targeting GABARAPL2 or unrelated siRNA (siCTR). Cells were starved for 30’ and lysed. Protein extract was analysed by WB to detect the indicated markers. ULK1 over ACTIN is also reported (right). Data are as mean ± SEM. Statistical analysis was performed by unpaired Student’s t‐test. ns, not significant (n = 3 independent experiments).PHeLa cells harbouring endogenously HA‐tagged LC3C were transiently transfected with siLC3C or siCTR and grown in complete or starved (STV) media, in the presence or absence of CLQ for 1 h. LC3B‐II and ACTIN were analysed by WB. Densitometry analysis of LC3B‐II normalized over ACTIN is reported (bottom left). LC3C downregulation is also shown (bottom right). All data are reported as mean ± SEM. Statistical analysis was performed by two‐way ANOVA followed by Tukey’s multiple comparison test. *P < 0.05 (n = 3 independent experiments).QHeLa cells were co‐transfected with indicated siRNAs and cultured in starvation medium for 30’. Cells were lysed, and protein extracts were immunoprecipitated by using an anti‐GABARAP antibody or IgG as negative control. ULK1, GABARAP and TFG protein levels were analysed by WB. ULK1‐IP over GABARAP‐IP is reported in the graph (bottom). Data are mean ± SEM. Statistical analysis was performed by unpaired Student’s t‐test. ***P < 0.001 (n = 3 independent experiments).

Fig 5: NDs API improves the ATO-based therapy in vivo. HepG2 tumor-bearing nude mice were administered intravenously (i.v.) with NS (200 μL), NDs (5 mg kg−1), ATO (0.75 mg kg−1), ATO (1.5 mg kg−1), NDs–0.75 mg kg−1 ATO mixture, or NDs–1.5 mg kg−1 ATO mixture daily for 5 consecutive days every week, with 13-week cycles. a Images of livers/tumors or excised tumors from treated tumor-bearing mice. b Weight ratio of tumors from treated mice (n = 10 for NS, 0.75 mg kg−1 ATO and 1.5 mg kg−1 ATO-treated groups, n = 8 for NDs–0.75 mg kg−1 ATO or NDs–1.5 mg kg−1 ATO-treated groups, n = 7 for NDs-treated group). **P < 0.01, ***P < 0.001 by t-test, significantly different from NS; #P < 0.05, ##P < 0.01 by t-test, significantly different from ATO. c Immunoblots of autophagy-related proteins LC3-II, p62 (left); semi-quantified analysis (n = 3) in tumor tissues from mice after treatment (right). GAPDH was used as the loading control. Normalized band densities were shown below each band. d Representative images of TUNEL stains and its quantification by Image J. Scale bar: 50 μm. ***P < 0.001 by t-test, significantly different from NS; ##P < 0.01, ###P < 0.001 by t-test, significantly different from ATO. e, f Survival curves (e, n = 10) and whole body weight curves (f) of nude mice after various treatments (n = 20). *P < 0.05; **P < 0.01, ***P < 0.001 by t-test or one-way analysis of variance (ANOVA) using SPSS, significantly different from NS; #P < 0.05, ##P < 0.01 by t-test or one-way analysis of variance (ANOVA) using SPSS, significantly different from ATO. g Immunostaining of LC3 and ATG13 in tumor tissues of NDs–1.5 mg kg−1 ATO-treated mice. Schematic showing (left) and imaging (right) indicate that LC3/Atg13 puncta are in the periphery and around the blood vessels inside the tumors. Scale bars: 20 μm. Error bars are s.d.