* Supplementary Figure 1 A N.S B C D E basal Glycolytic reserve sh-cont sh-ATG5 10 20 30 40 50 ECAR (mPH/min) basal Glycolytic reserve sh-ATG7 60 B 1.3 10 18 27 35 44 52 61 70 79 87 96 minutes 50 100 200 250 150 sh-cont sh-ATG5 G O D N.S OCR (mpH/min) sh-ATG7 101 102 103 104 2289 4573 6867 Low flux High flux 9157 mCherry:EGFP Cell counts 101 102 103 104 2289 4573 6867 9157 High flux Low flux mCherry:EGFP Cell counts C D E Low flux High flux DAPI MAP1LC3B MERGE Relative glucose consumption High flux Low flux Relative Lac production 0.5 1.0 1.5 * Figure S1. High-autophagic flux cells exhibit low glycolysis, while low-autophagic flux cells exhibit high glycolysis. (A) ATG5 or ATG7 knockdown cells show no significant difference in basal and glycolytic reserve compared to control cells. (B) ATG5 or ATG7 knockdown cells show no significant difference in oxygen consumption rate (OCR) compared to control cells. OCR measured in SMMC7721 cells with control, ATG5- or ATG7-silencing using shRNAs. G, 10 mM glucose; O, 1 mM oligomycin; D, 50 mM 2-DG injection. The values are presented as the means ± SEM, n=3, N.S., not significant. (C) HepG2 cells were stably transfected with adenovirus expressing EGFP-mCherry-MAP1LC3B. Flow cytometry analysis of the ratio of mCherry:EGFP; the top and bottom 15% of cells were considered as high- and low-autophagic flux cells. (D) Immunofluorescence assessment of high- and low-autophagic flux cells. Expression of MAP1LC3B (green) and DAPI (blue) was visualized. Scale bars: 10 mm. (E) Relative glucose consumption, and lactate (Lac) production in the cell populations with high flux and low flux collected from (C). The values are presented as the means ± SEM, *p<0.05.

Supplementary figure 2 A B HK2 ATG5 TUBB sh-cont sh-ATG5 si-HK2 si-HK2-HK2 si-cont sh-cont sh-ATG5 si-cont si-HK2_1 si-HK2_2 HK2 ATG5 TUBB Figure S2. Immunoblots showing change of HK2 in ATG5 knockdown SMMC7721 cells with the indicated treatment. (A) SMMC7721 cells stably transfected with control or ATG5 shRNAs were transiently tranfected with siRNA against HK2, followed by immunoblot analysis of HK2 expression. (B) SMMC7721 cells stably transfected with control or ATG5 shRNAs were transiently tranfected with siRNA against HK2, then a plasmid carrying wild-type HK2 was introduced (synonymous mutations have been made in the target sequence of HK2), followed by immunoblot analysis of HK2 expression.

Supplementary figure 3 A B C D TUBB Baf A1 (100 nM) - + High flux Low flux HK2 MAP1LC3B SQSTM1 HK2 SQSTM1 Baf A1 (100 nM) 1 3 6 12 24 TUBB h C D si-cont si-LAMP2_1 si-LAMP2_2 HK2 Ub MG132 (10 mM) 6 12 TUBB h HK2 LAMP2A TUBB Figure S3. HK2 was degraded through an autophagy-dependent pathway. (A) Endogenous HK2 is upregulated upon Baf A1 treatment. SMMC7721 cells were assayed treated with Baf A1 (100 nM) for the indicated times. Cell lysates were analyzed by immunoblot. (B) The expression level of HK2 is high in low- flux cells compared to high- flux cells, and it accumulated more in high- flux cells that had been inhibited with Baf A1. Immunoblot analysis of high- and low- autophagic flux cells. (C) HK2 is not degraded by the 26S proteasome pathway. SMMC7721 cells were treated with the proteasome inhibitor MG132 (10 M) in for the indicated times, and the HK2 expression level was analyzed by immunoblot. (D) LAMP2 knockdown upregulates the expression level of HK2. SMMC7721 cells were silenced with LAMP2 siRNAs, and the HK2 expression level was analyzed by immunoblot.

Supplementary figure 4 A B IgG cont si-TRAF6 2 IP:HK2 HK2 HA-Ub - + 1 cell lysate TUBB FLAG-TRAF6 WT C70A FLAG B Supplementary figure 4 Figure S4. TRAF6 is the E3 ligase of HK2. (A) HK2 interacts with wild- type TRAF6, but not mutant TRAF6C70A. Wild- type TRAF6 or mutant TRAF6C70A was co-transfected with a plasmid expressing HA-ubiquitin (HA-Ub) into HEK293T cells. The cell lysates were then analyzed by IP immunoprecipitation by using anti-HK2 antibody. (B) 293T cells were subjected to immunoprecipitation IP analysis of HK2 ubiquitination after siRNA-mediated depletion of TRAF6.

Supplementary figure 5 T1 T2 T3 T 4 T5 T6 T7 SQSTM1 MAP1LC3B HK2 LDHA PFKP GAPDH PKM TUBB Figure S5. HK2, not other glycolytic enzymes, is correlated with SQSTM1 and MAP1LC3B in clinical liver cancer samples. Immunoblot analysis of HK2, LDHA, PFKP, GADPH, PKM, SQSTM1 and MAP1LC3B expression in tumors from 7 patients.