Biguanide pharmacophore interacts with ATP synthase subunit e (ATP5I).

(A) Design of bio-inspired probe biotin functionalized biguanide (BFB) based on the structure of metformin (Met). (B) Immunoblots for the phosphorylation of AMPK (Thr172) and ACC (Ser79) in extracts from KP-4 pancreatic cancer cells treated with 2.5 mM Met or BFB for 16 hours. β-ACTIN was used as loading control. (C) Representative quantification of cell viability and growth with corresponding EC50 values of 3-day treatments with metformin (Met) or biotin functionalized biguanide (BFB) in KP-4 cells. Values represent the mean ± standard deviation of N=3. (D) Representative images of mitochondria and BFB localization in cells as in (B). Cells were treated with 1 mM of metformin (Met) or BFB for 16 hours and mitochondrial signal and BFB localization were analyzed by co-immunofluorescence using Streptavidin fluorophore conjugate and anti-TOMM20 antibody, scale bar= 10 μm. Cells untreated (-) and treated with 1 mM Met were used as negative controls. (E) Colocalization between TOMM20 (TOMM20-568) and Streptavidin (Strep-488) fluorophores was analyzed for the BFB condition from (D) through job plot intensity profile. (F) Pull-down validation experiments with streptavidin beads alone (-), D-biotin (B), BFA and BFB using antibody followed by immunoblot against ATP5I in cells as in in extracts from HEK-293T embryonic kidney cells. The whole cell lysate (WCL) was added as control. (G) Binding interactions studies of BFB with recombinant purified ATP5I (rATP5I) using Surface Plasmons of Resonance (SPR). Representative sensorgrams show affinity kinetics of BFB and rATP5I. BFB was exposed onto streptavidin immobilized sensor chip and several concentrations of rATP5I were added until saturation of the signal. RU: Resonance Units. (H) Binding affinity curve obtained from each steady state from (F). KD refers to the dissociation equilibrium constant and Rmax represent the theoretical maximum response.

(A) Synthesis of biotin-NHS (1); (a) NHS, EDC, DMF, room temperature (r.t) overnight (o/n). (B) Synthesis of biotin functionalized biguanide (BFB) chloride salt (3); (a) Dicyandiamide, TMSCl, MeCN, 160°C, 3h, 2: 6-aminohexylbiguanide hydrochloride salt; (b) Biotin-NHS, DIPEA, DMF, r.t, o/n. (C) Synthesis of biotin functionalized amine (BFA) chloride salt (5); (a) N-Boc-1,6-hexanediamine, DMF, r.t, o/n, 4: biotin functionalized N-Boc-amine; (b) HCl/MeOH, MeOH, r.t, o/n.

Proteins were isolated from streptavidin-coated beads after affinity purification using the biotinylated biguanide probe (BFB), followed by competitive elution with metformin (50 mM).

Mass spectrometry (MS/MS) analysis was performed to identify proteins potentially involved in BFB-specific interactions. The table reports the approximate molecular weight and total number of peptides identified for each protein.

(A) SDS PAGE analysis of recombinant ATP5I (rATP5I) after the purification process. Purity is estimated at ≥ 90 %. MM: molecular weight marker. (B) Gel filtration profile analysis of rATP5I using Superose 12 10/300 GL. DV represents column dead volume. mAU: milli-Absorbance Unit. (C) Calibration of the Superose 6 10/300 GL size-exclusion column for conformational analysis of the Nter-6×His-ATP5I construct. The calibration curve was generated by plotting the logarithm of the molecular weight (log(MW)) of standard proteins (thyroglobulin, γ-globulin, myoglobin, vitamin B12) as a function of their partition coefficient (Kav). Ve: elution volume; V : void volume (1.12 mL); Vt: total column volume (2.8 mL).

ATP5I knockout in pancreatic cancer cells alters the organization of the mitochondrial network.

(A) Immunoblot for the indicated proteins in extracts from clones of KP-4 cells expressing a control small guide RNA against GFP (sgGFP: control 1 and control 2) or two different sgRNAs against ATP5I (sgATP5I #1 and sgATP5I #2). GAPDH antibody was used as loading control. (B) Representative threshold images of mitochondrial morphologies visualized by TOMM20 immunofluorescence (scale bar = 10 μm). A magnified inset (yellow box) is shown for each image to highlight mitochondrial structural details. All images were analyzed using the Mitochondria Analyzer plugin in Fiji (ImageJ). (C–F) Quantitative analysis of key mitochondrial parameters: (C) number of branches, (D) total branch length, (E) number of branch endpoints and (F) mean branch diameter. Data represent mean ± standard deviation from N = 3 independent clones, with 50–100 cells analyzed per clone. (ns) not significant, P < 0.01 (**), and P < 0.0001 (****) using an unpaired Student’s t-test. (G) Representative Blue Native-PAGE followed by western blotting using an antibody against the β-subunit of the F domain of ATP synthase in KP-4 or U2OS cells treated with vehicle, metformin (10 mM, 16 h or 3 days), or in ATP5I knockout cells (ATP5I KO). Monomer* indicates the assembly intermediates of the F F -ATP synthase known to accumulate after disabling ATP5I.

(A) Immunoblot for the indicated proteins in extracts from clones of KP-4 cells expressing a control small guide RNA against GFP (sgGFP: control 1 and control 2) or two clones for each of the two different guides targeting ATP5I (sgATP5I #1: sgATP5I 1 and sgATP5I 2, and sgATP5I #2: sgATP5I 3 and sgATP5I 4). * is a non-specific band that migrates like 40 KD MTCO1 from Complex IV. GAPDH was used as loading control. (B) Relative qPCR quantification of the mRNAs encoding proteins representative of the five OXPHOS complexes (in Figure 2A) in clones of KP-4 cells expressing a control small guide RNA against GFP (sgGFP: control 1 and control 2) or two clones for each of the two different guides targeting ATP5I (sgATP5I #1: sgATP5I 1 and sgATP5I 2, and sgATP5I #2: sgATP5I 3 and sgATP5I 4). Values represent the mean ± standard deviation of three biological replicates. (C) qPCR quantification of mitochondrial genomic DNA (Mt) over cellular nuclear genomic DNA (Nu) in cells as in (A). Values represent the mean ± standard deviation of N=3.

Quantification of immunoblot in Figure 2G.

(A) The PVDF membrane was stained with Ponceau S to visualize total protein and subsequently imaged on Bio-Rad ChemiDoc XRS+ system. (B) Quantification. Using Bio-Rad Image Lab software, total band intensity was quantified for the band corresponding to the vestigial form of ATP synthase (monomer*) in figure 2G and normalized to the total band intensity of the corresponding lane in the Ponceau image. Values for each cell line were normalized to the “untreated” condition.

ATP5I knockout desensitizes pancreatic cancer cells to biguanides.

(A) Quantification of NAD+/NADH ratio in KP-4 cells expressing a control small guide RNA against GFP (sgGFP) or a representative clone of two different guides targeting ATP5I (sgATP5I #1 or sgATP5I #2). Values represent the mean ± standard deviation of N=3. (***) P < 0.001 using an unpaired Student’s t-test. (B) Relative quantification of oxygen consumption rate (OCR) over extracellular acidification rate (ECAR) by Seahorse analysis in cells as in (A). Values represent the mean ± standard deviation of at least N=3. (***) P < 0.001 using a paired Student’s t-test. (C) Immunoblot for total and phosphorylated levels of AMPK (Thr172) protein in extracts from cells as in (A). ATP5I confirms loss of expression in KO and GAPDH was used as loading control. (D) Growth curves of cells as in (A) measuring the relative number of cells over 6 days. Media was changed every two days. (E) Representative kinetic curves of OCR in cells as in (A) treated with 5 mM of metformin (Met) relative to control treated cells (dashed line) using Seahorse. (F) Representative kinetic curves of ECAR in cells as in (A) treated with 5 mM metformin (Met) relative to control treated cells (dashed line) using Seahorse. (G) Quantification of OCR/ECAR ratio fold change at 3 and at 6 hours from kinetic curves (E-F). Values represent the mean ± standard deviation of N=3. (ns) not significative, (*) P < 0.05, (**) P < 0.01, (****) P < 0.0001 using a repeated measures (RM) one-way ANOVA with Sidak’s multiple comparison test. (H) Representative growth of cells as in (A) exposed to different concentrations of metformin (Met) for three days with corresponding EC50 values of metformin. Values represent the mean ± standard deviation of N=3. (***) P < 0.001 and (****) P < 0.0001 using an unpaired Student’s t-test. (I) Representative kinetic curves of OCR in cells as in (A) treated with 100 μM phenformin (Phen) relative to control treated cells (dashed line) using Seahorse. (J) Representative kinetic curves of ECAR in cells as in (A) treated with 100 μM of phenformin (Phen) relative to control treated cells (dashed line) using Seahorse. (K) Quantification of OCR/ECAR ratio fold change at 3 and at 6 hours from kinetic curves (I-J). Values represent the mean ± standard deviation of at least three biological replicates. ns: not significative, (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001 using a RM one-way ANOVA with Sidak’s multiple comparison test. (L) Representative growth of cells as in (A) exposed to different concentrations of phenformin (Phen) for three days with corresponding EC50 values. Values represent the mean ± standard deviation of N=3. (****) P < 0.0001 using an unpaired Student’s t-test.

(A) Quantification of NAD+ concentration in KP-4 cells expressing a control small guide RNA against GFP (sgGFP) or a representative clone of two different guides targeting ATP5I (sgATP5I #1 or sgATP5I #2). Values represent the mean ± standard deviation of N=3. (***) P < 0.001, (****), P < 0.0001 using an unpaired Student’s t-test. (B) Quantification of NADH concentration in cells as in (A). Values represent the mean ± standard deviation of N=3. (ns) not significative, (**) P < 0.01 using an unpaired Student’s t-test. (C) Relative quantification of oxygen consumption rate (OCR) by Seahorse analysis in cells as in (A). Values represent the mean ± standard deviation of at least N=3. (ns) not significative, (*) P < 0.05 using a paired Student’s t-test. (D) Relative quantification of extracellular acidification rate (ECAR) by Seahorse analysis in cells as in (A). Values represent the mean ± standard deviation of at least N=3. (***) P < 0.001 using a paired Student’s t-test. (E) Representative cell growth of cells as in (A) treated with different concentrations of 2-D-deoxyglucose with corresponding EC50 values of 2-D-deoxyglucose treatment in cells as in (A). Values represent the mean ± standard deviation of N=3. (**) P < 0.01 using an unpaired Student’s t-test. (F) Representative cell viability curves with corresponding EC50 values of treatments of 2-D-deoxyglucose in combination without (-) and with different concentrations (1, 2.5 and 5 mM) of metformin (Met) in KP-4 cells expressing control sgGFP. Values represent the mean ± standard deviation of three biological replicates. (**) P < 0.01 using an unpaired Student’s t-test. (G) Total and phosphorylated levels of AMPK protein extracts from cells as in (A) treated with 2.5 mM or 5 mM of Met for 16 hours. β-ACTIN was used for loading control. (H) Growth curves of cells as in (A) supplemented with 100 μg/mL sodium pyruvate (Pyr) and 50 μg/mL uridine (Uri) by measuring the percentage of confluency over 6 days. Media was changed every two days.

Exogenous ATP5I enable the reorganization of mitochondrial network in ATP5I knockout pancreatic cancer cells.

(A) A representative immunoblots for the indicated proteins in KP-4 cells expressing exogenous ATP5I (exoATP5I: +) in control cells expressing a small guide RNA against GFP (sgGFP) or in ATP5I KO cells (clones of two different small guide RNAs: sgATP5I #1 sgATP5I #2) compared with the same cell lines without expression of exogenous ATP5I (-). GAPDH antibody was used as loading control. (B) Representative threshold images of mitochondrial morphologies visualized by TOMM20 immunofluorescence (scale bar = 10 μm) of ATP5I KO cells and their derivative re-expressing ATP5I. A magnified inset (yellow box) is shown for each image to highlight mitochondrial structural details. All images were analyzed using the Mitochondria Analyzer plugin in Fiji (ImageJ). (C–F) Quantitative analysis of key mitochondrial parameters: (C) number of branches, (D) total branch length, (E) number of branch endpoints and (F) mean branch diameter. Data represent mean ± standard deviation from N = 3 independent clones, with 50–100 cells analyzed per clone. (ns) not significant, P < 0.01 (**), and P < 0.0001 (****) using an unpaired Student’s t-test.

(A) Representative images of mitochondrial localization of ATP5I in KP-4 cells expressing exogenous ATP5I (exoATP5I: +) in control (clones with sgGFP) or a representative clone of each of two small guide RNA against ATP5I (sgATP5I #1 or sgATP5I #2) compared with the same cell lines without expression of exogenous ATP5I (-). Mitochondrial localization was analyzed by co-immunofluorescence using ATP5I and TOMM20 antibodies, scale bar= 10 μm. DAPI was used as a DNA counter stain. (B) Colocalization between TOMM20 (TOMM20-568) and ATP5I (ATP5I-488) signals was analyzed with job plot intensity profile.

Re-expression of ATP5I rescues metabolic profile and resensitizes ATP5I knockout pancreatic cancer cells to biguanides.

(A) Quantification of NAD+/NADH ratio in KP-4 cells expressing exogenous ATP5I (exoATP5I: +) in control sgGFP or a representative clone of two different small guide RNAs (sgATP5I #1 and sgATP5I #2) compared with the same cell lines without expression of exogenous ATP5I (-). Values represent the mean ± standard deviation of three biological replicates. (ns) not significative, (*) P < 0.05, (**) P < 0.01, (***) P < 0.001 using an ordinary one-way ANOVA with Sidak’s multiple comparison test. (B) Relative quantification of oxygen consumption rate (OCR) over extracellular acidification rate (ECAR) by Seahorse analysis in cells as in (A). Values represent the mean ± standard deviation of at least three biological replicates. (ns) not significative, (*) P < 0.05, (***) P < 0.001, (****) P < 0.0001 using a repeated measures (RM) one-way ANOVA with Sidak’s multiple comparison test. (C) Immunoblot of total and phosphorylated levels of AMPK (Thr172) protein in extracts from cells as in (A). GAPDH was used as loading control. (D) Intracellular ATP levels measured in cell lines as in (A). Data are presented as mean ± standard deviation. N = 2. ns: not significant. (E) Growth curves of cells as in (A) by measuring the relative number of cells over 6 days. Media was changed every two days. (F) EC50 values of metformin (Met) treatments in cells as in (A). Values represent the mean ± standard deviation of N=3. (ns) not significative, (**) P < 0.01, (****) P < 0.0001 using an ordinary one-way ANOVA with Sidak’s multiple comparison test. (G) EC50 values of phenformin (Phen) treatment in cells as in (A). Values represent the mean ± standard deviation of three biological replicates. (ns) not significative, (*) P < 0.05, (***) P < 0.001, (****) P < 0.0001 using an ordinary one-way ANOVA with Sidak’s multiple comparison test.

(A) Quantification of NAD+ concentration in KP-4 cells expressing exogenous ATP5I (exoATP5I: +) in control sgGFP clones or a representative clone of two different small guide RNAs against ATP5I (sgATP5I #1 and sgATP5I #2) compared with the same cell lines without expression of exogenous ATP5I (-). Values represent the mean ± standard deviation of three biological replicates. (ns) not significative, (*) P < 0.05, (****) P < 0.0001 using an ordinary one-way ANOVA with Sidak’s multiple comparison test. (B) Quantification of NADH concentration in cells as in (A). Values represent the mean ± standard deviation of three biological replicates. (ns) not significative using an ordinary one-way ANOVA with Sidak’s multiple comparison test. (C) Relative quantification of oxygen consumption rate (OCR) by Seahorse analysis in cells as in (A). Values represent the mean ± standard deviation of at least three biological replicates. (ns) not significative, (**) P < 0.01 using a repeated measures (RM) one-way ANOVA with Sidak’s multiple comparison test. (D) Relative extracellular acidification rate (ECAR) by Seahorse analysis in cells as in (A). Values represent the mean ± standard deviation of at least N=3. (ns) not significative, (*) P < 0.05, (****) P < 0.0001 using a RM one-way ANOVA with Sidak’s multiple comparison test. (E) EC50 values of 2-D-deoxyglucose treatment in cells as in (A). Values represent the mean ± standard deviation of N=3. (**) P < 0.01, (***) P < 0.001 using an ordinary one-way ANOVA with Sidak’s multiple comparison test.

(A) Representative kinetic curves of oxygen consumption rate (OCR) fold change in KP-4 cells expressing exogenous ATP5I (exoATP5I: +) in control sgGFP clones or a representative clone of each of two small guide RNA against ATP5I (sgATP5I #1 and sgATP5I #2) compared with the same cell lines without expression of exogenous ATP5I (-) treated with 5 mM metformin (Met) using Seahorse. (b) Representative kinetic curves of extracellular acidification rate (ECAR) fold change in cells as in (A) treated with 5 mM Met using Seahorse. (C) Quantification of OCR/ECAR ratio fold change at 3 and at 6 hours from kinetic curves in (A) and (B). Values represent the mean ± standard deviation of at least three biological replicates. (ns) not significative, (*) P < 0.05, (**) P < 0.01, (***) P < 0.001 using a repeated measures (RM) one-way ANOVA with Sidak’s multiple comparison test. (D) Representative kinetic curves of OCR fold change in cells as in (A) but treated with 100 μM phenformin (Phen) using Seahorse. (E) Representative kinetic curves of ECAR fold change in cells as in (A) but treated with 100 μM Phen using Seahorse. (F) Quantification of OCR/ECAR ratio fold change at 3 and at 6 hours from kinetic curves (D) and (E). Values represent the mean ± standard deviation of at least three biological replicates. (ns) not significative, (****) P < 0.0001 using a RM one-way ANOVA with Sidak’s multiple comparison test.

ATP5I Deletion Mimics Biguanide-Induced Bioenergetic Remodeling

(A) Representative oxygen consumption rate (OCR) profiles following sequential injection of oligomycin (Oligo), FCCP, and rotenone/antimycin A (Rot/AA) in control, ATP5I knockout (sgATP5I #1 and #2), and Rescue (sgATP5I + exoATP5I) cells treated with vehicle, 5 mM metformin (Met), or 100 µM phenformin (Phen). (B) Representative extracellular acidification rate (ECAR) profiles under the same conditions as in (A). (C) Quantification of basal mitochondrial respiration corresponding to the conditions in (A), N=3. (D) Representative confocal images of cells stained with the membrane potential sensitive dye TMRE (100 nM, 30 min at 37 °C in complete DMEM without phenol red) under control conditions, following ATP5I knockout (ATP5I KO), or after depolarization with FCCP (10 µM, 30 min prior to TMRE incubation), scale bar = 10 µm. (E) Quantification of TMRE fluorescence intensity per cell. Data are expressed as mean ± standard deviation and normalized to control levels. Results are from three independent experiments performed on separate days (n = 431 cells for control; n = 371 for FCCP; n = 536 for ATP5I KO). (ns) not significative and (***) P < 0.001, (****) P < 0.0001 using an unpaired Student’s t-test.

Chemogenomic screening of metformin reveals an imprint on F1Fo-ATP synthase.

A) Results of the pooled genome-wide CRISPR/Cas9 KO screen made in NALM-6 cells treated with 16 mM metformin or control. Data are represented as a Volcano plot of gene enrichment/depletion scores vs p-values from using the CRANKS algorithm. Some genes of interest are labeled. Enhancers of metformin growth inhibition with negative CRANKS scores below 2.5 (dashed line) are labeled blue, while suppressors with positive CRANKS scores above 2.5 (dashed line) are labeled red. B-C) Pairwise comparison of gene CRANKS scores obtained from screening metformin 16 mM in NALM-6 cells against that from screening either 70 nM rotenone or (C) 2 μM oligomycin A. D) Model for metformin action triggering the OMA1-DELE1-HRI pathway.