Fig 2: Mechanistic insights into ARIH2 loss-mediated EGFR-TKI resistance.(A) Mass spectrometry analysis of global protein changes between control and ARIH2-deficient HCC827 cells. (B) Immunoblots of indicated proteins in control and ARIH2-deficient HCC827 cells. ß-Actin was used as a loading control. (C) Quantitative RT-PCR analysis of relative mRNA levels of indicated genes in control or ARIH2-deficient HCC827 cells. Error bars represent mean ± SD; n = 4. (D) Immunoblots of indicated proteins showing ectopic expression of HA-METAP2 in HCC827 cells. GAPDH was used as a loading control. (E) Crystal violet staining colony formation assay of HCC827-Vector or HCC827-HA-METAP2 cell lines treated with DMSO, erlotinib (1 µM), or gefitinib (1 µM). (F) Quantification of colony formation in (E), shown as percentage of the HCC827-Vector sample. Mean (three biological replicates) ± SD is shown. (G) De novo protein synthesis of HCC827-Vector or HCC827-HA-METAP2 cells after treatment with DMSO or erlotinib (1 µM, 24 hr) as determined by L-azidohomoalanine (AHA) labeling. Cells were starved of methionine for 1 hr and incubated with AHA for 1 hr. Lysates were subjected to a Click-iT chemistry reaction to switch azido-modified nascent proteins to alkyne-biotin, and visualized by Streptavidin-HRP immunoblotting. ß-Actin was used as a loading control. (H) De novo protein synthesis of control or ARIH2-deficient HCC827 cells after treatment with DMSO or erlotinib (1 µM, 24 hr) as determined by AHA labeling. ß-Actin was used as a loading control. (I) Immunoblot of METAP2 in HCC827 cells upon proteasome inhibitor bortezomib treatment. ß-Actin was used as a loading control. (J) Immunoblots of ARIH2 and METAP2 in control or ARIH2 knockout HCC827 cells upon bortezomib treatment. ß-Actin was used as a loading control. (K) De novo METAP2 protein synthesis in control or ARIH2-deficient HCC827 cells as determined by AHA labeling and streptavidin pulldown. ß-Actin was used as a loading control. (L) Immunoblots of indicated proteins showing ectopic expression of HA-ALDOA and HA-PSAT1 in HCC827 cells. GAPDH was used as a loading control. (M) Crystal violet staining colony formation assay of HCC827-Vector, HCC827-HA-ALDOA or HCC827-HA-PSAT1 cells treated with DMSO, erlotinib (1 µM), or gefitinib (1 µM). (N) Quantification of colony formation in (M), shown as percentage of the HCC827-Vector sample. Mean (three biological replicates) ± SD is shown. Statistical significance was tested using unpaired two-tailed t test (C, F and N); *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001; ns, not significant.Figure 6—source data 1.Raw data from Figure 6.

Fig 3: TRPV1–4 Interact with FBP-Unoccupied Aldolase and Are Physically Required for Lysosomal AMPK Activation in Low Glucose(A) FBP at physiologically relevant concentrations abolishes the interaction between endogenous TRPV4 and aldolase. Light organelles were prepared from 2-h glucose-starved MEFs and aliquoted, followed by addition of FBP at concentrations indicated. Aldolases (represented by ALDOA) were immunoprecipitated, followed by immunoblotting using the indicated antibodies. Note that the polyclonal antibody raised in rabbit by His-tagged ALDOA was able to react with all three isozymes of aldolase, as validated in Figure S1G. The antibody against endogenous TRPV4 was validated in Figure S1B.(B) TRPV channels are functionally redundant in mediating AMPK activation under glucose starvation. TRPV1-/-, TRPV2-/-, TRPV3-/-, TRPV4-/-, TRPV-QKO MEFs, and wild-type (WT) MEFs as control, were incubated in DMEM with or without 25-mM glucose for 2 h, followed by analysis of p-AMPKa and p-ACC. AMPK activation was virtually abolished in TRPV-QKO MEFs in which no TRPV was detectable by the antibodies.(C) FBP dampens the association between endogenous TRPVs and aldolase. Lysates prepared from 2-h glucose-starved MEFs were mixed with 10 µM FBP. Aldolase was immunoprecipitated, followed by immunoblotting using the indicated antibodies. The antibody against endogenous TRPV1 was validated in Figure S2B (left panel).(D) TRPVs are required for inactivation of lysosomal v-ATPase in response to glucose starvation. TRPV-QKO MEFs were pre-loaded with LysoSensor Green DND-189 and Hoechst, and were then cultured in DMEM with 25-mM glucose (normal) or DMEM without glucose (GS) for 2 h. The relative fluorescent intensities of LysoSensor (normalized to the intensity of Hoechst) were then analyzed. Results are mean ± SEM; n value of each group is directly labeled on the bar, and the same below. p value was determined by Student’s t test. Representative images of this experiment are shown in Figure S2D.(E) TRPVs are required for the lysosomal translocation of AXIN. MEFs were glucose starved and the localization of AXIN was determined by immunofluorescent staining. Endogenous AXIN (green, validated previously, (Zhang et al., 2014)) and LAMP2 (red) were stained with goat anti-AXIN antibody and rat anti-LAMP2 antibody, respectively, and were imaged by confocal microscopy. “N” indicates nucleus.(F) TRPV proteins are required for the formation of Ragulator:AXIN:LKB1:AMPK complex. Endogenous LAMTOR1 in regularly cultured (in DMEM containing 25-mM glucose) or glucose-starved TRPV-QKO MEFs was immunoprecipitated, followed by immunoblotting.(G) TRPVs are required for the activation of AMPK in physiologically low glucose. TRPV-QKO MEFs were incubated in DMEM containing 1, 3, 5, or 10 mM glucose for 2 h, followed by analysis of p-AMPKa and p-ACC.(H) TRPV channels are specifically involved in the AMPK activation induced by glucose starvation. TRPV-QKO MEFs, along with its wildtype control, were starved for glucose (GS) or treated with 1 µM oligomycin, 50-µM dinitrophenol (DTP), 500 µM sorbitol, 1 mM H2O2 or 200 µM A769662, for 2 h, followed by analysis of p-AMPKa and p-ACC.(I and J) TRPVs are required for starvation-induced AMPK activation in mouse liver and primary hepatocytes. TRPV1-/- mice were injected with a combination of AAV-carried siRNAs against TRPV2, TRPV3, and TRPV4. After 4 weeks of injection, p-AMPKa and p-ACC levels in livers from 16-h starved mice (I) and in hepatocytes incubated in medium containing 3-mM glucose (resembling the concentration of plasma glucose detected in 16-h starved mice, as shown in Figure S2J) for 12 h (J) were analyzed.(K) Requirement of TRPVs for glucose restriction-induced AMPK activation is conserved in nematodes. Worms (3 days old) of the strain lacking TRPVs (osm-9; ocr-2), along with its wild-type control (N2), were cultured on normal NGM agar or NGM agar containing 5 mM 2-DG for 20 h (fed with live OP50 bacteria). The worms were then lyzed, and the levels of p-AMPKa and p-ACC were determined by immunoblotting.Experiments in this figure were performed at three times, except for (I), (J), and (K) twice.See also Figures S1 and S2.

Fig 4: Aldolase Binds to TRPV and Inhibits its Ca2+ Channel Activity in Low FBP(A) Aldolases are required for inhibition of TRPV channels. ALDO-TKD MEFs expressing TRPV4-GCaMP6s were incubated in glucose-free DMEM at 37°C. After 2 min in the live-cell incubation chamber, the images were captured by confocal microscopy at a regular interval. After 250 s of starvation, 5 µM AMG-9810 was added to the medium. The changes of TRPV4 activities were analyzed as described in Figure 2A. n = 3 cells from 2 dishes/experiment, and were graphed as mean ± SEM; p value by Student’s t test.(B and C) The FBP-constantly bound aldolase mutant ALDO-D34S fails to block the activity of TRPV4 in low glucose. MEFs (B) and HEK293T cells (C) expressing TRPV4-GCaMP6s were incubated in glucose-free DMEM at 37°C. After 2 min of incubation in the live-cell incubation chamber, the images were captured by confocal microscopy at a regular interval. For MEFs, after 350 s of starvation, 5 µM AMG-9810 was added to the medium, and the relative fluorescent intensities were analyzed and graphed (n = 3 cells from 3 dishes/experiment) in (B). For HEK293T cells expressing HA-ALDOA or its D34S mutant (red, labeled as #1 and #2 in C), representative images at t = 0 s and t = 1313 s (for ALDOA) or 1440 s (for ALDOA-D34S) were shown (upper panel) and the relative fluorescent intensities of GCaMP6s (green) were analyzed and graphed (lower panel, n = 5 cells from 2 dishes/experiment for ALDOA, and n = 7 cells from 3 dishes/experiment for ALDOA-D34S) in (C). Statistical analysis results were shown in mean ± SEM; p value by Student’s t test.(D) Schematic illustration of TRPV4 membrane topology. The K535 residue (in red) of TRPV4 forming the interface for aldolase, and the N-terminal region of aa 1–144 (blue) for binding to v-ATPase (as characterized below) are illustrated.(E) Glucose starvation fails to inhibit the channel activity of TRPV4-K535A that is defective in binding to aldolase. Cells were treated and the activity of TRPV4 was analyzed as in B, except that TRPV4-GCaMP6s-K535A was expressed in TRPV-QKO MEFs (n = 3 cells from 2 dishes/experiment). Statistical analysis data of E were shown as mean ± SEM; p value by Student’s t test.(F) Re-introduction of TRPV4-K535A to TRPV-QKO MEFs suppresses AMPK activation under glucose starvation. TRPV-QKO MEFs stably expressing TRPV4-K535A were regularly cultured or glucose starved for 2 h, followed by analysis of p-AMPKa and p-ACC.(G) The TRPV4-binding-defective, truncated ALDOA (aa 1–345) fails to trigger lysosomal AMPK activation. ALDO-TKD MEFs stably expressing ALDOA 1–345 or WT ALDOA as a control were regularly cultured or glucose starved for 2 h, followed by analysis of p-AMPKa and p-ACC.Experiments were performed for three times.See also Figure S6.

Fig 5: T-Cell-Specific Loss of AMPK Accelerates Development of T-ALL and Causes mTORC1 Hyperactivation(A) Tumor-free survival curves for mice with T cells of the four different genotypes. p values (log-rank, Mantel Cox) for survival curves that are significantly different from those of the tPTEN-/- tAMPK+/+ mice are shown.(B) Distribution of cell sizes, estimated by forward scatter in flow cytometry, of thymocytes from mice of two different genotypes at 37–42 days of age. The population of large cells in the tPTEN-/- tAMPK-/- sample indicates incipient lymphoma.(C) Signaling via AMPK, Akt, and mTORC1 in lymphomas from three mice of each genotype.(D) Quantification of blots from all mice analyzed; for pAMPK, AMPKa, pAMPK:AMPKa, pACC:ACC, and pAkt:Akt (Ser473), n = 8; for pRPS6:RPS6 and pEBP1:4EBP1, n = 6 to 10.(E) Quantification of blots (see Figure S1 for representative blots) analyzing expression of markers of cell proliferation and apoptosis in normal thymus or lymphomas of the indicated genotypes (n = 8–10).(F) Expression of HIF1A and the glycolytic enzymes aldolase A (ALDOA) and lactate dehydrogenase A (LDHA). (Top) Western blots from three mice. (Bottom) Quantification for all samples analyzed (HIF1A and LDHA, n = 3–5; ALDOA, n = 6–10.(G) Lactate levels in lymphoma-bearing thymus of the indicated genotypes.(A–G) Mean values ± SEM are shown; those significantly different from the Cre- control by one-way ANOVA are indicated by asterisks, and those significantly different from tPTEN-/- tAMPK+/+ mice are indicated by daggers (†).