The speciﬁc AMPK ac/vator A-769662 ameliorates pathological phenotypes following mitochondrial DNA deple/on

AMP-acZvated protein kinase (AMPK) is a master regulator of cellular energy homeostasis that also plays a role in preserving mitochondrial funcZon and integrity. Upon a disturbance in the cellular energy state that increases AMP levels, AMPK acZvity promotes a switch from anabolic to catabolic metabolism to restore energy homeostasis. However, it is currently unclear how severe of a mitochondrial dysfuncZon is required to trigger AMPK acZvaZon, and whether sZmulaZon of AMPK using speciﬁc agonists can improve the cellular phenotype following mitochondrial dysfuncZon. Using a cell model of mitochondrial disease characterized by progressive mitochondrial DNA (mtDNA) depleZon and deterioraZng mitochondrial metabolism, we show that mitochondria-associated AMPK becomes acZvated early in the course of the advancing mitochondrial dysfuncZon,


Introduc/on
Mitochondrial disorders are a heterogeneous group of omen debilitaZng diseases, many of which manifest during childhood.They are caused by pathogenic variants in nuclear or mitochondrial genes that impair respiratory chain funcZon and/or mitochondrial ATP producZon either directly or indirectly.The first group includes geneZc variants in genes directly involved in oxidaZve phosphorylaZon (OXPHOS), while the second comprises genes whose gene products are essenZal for the maintenance or expression of mitochondrial DNA (mtDNA).With a few excepZons, there is currently no cure for mitochondrial diseases, and therapeuZc opZons are thus generally limited to management of the paZents' individual symptoms (Barcelos, Emmanuele, and Hirano 2019;Pitceathly et al. 2021).ConZnued research into potenZal treatments and their mechanisms of acZon is thus called for.
An experimental approach that has shown promise in animal models of mitochondrial disease is the general inducZon of mitochondrial biogenesis, achieved for example by acZvaZng the key energy sensor and metabolic regulator, the AMP-acZvated protein kinase (AMPK) (Viscomi et al. 2011;Peralta et al. 2016).AMPK becomes acZvated upon energeZc stresses manifesZng as a decreased raZo of ATP to AMP and/or ADP, and phosphorylates its targets to shim metabolism from anabolism to catabolism (Steinberg and Hardie 2023).This energy-metabolic reprogramming involves upregulaZon of lipid and glucose breakdown, increased autophagy and mitophagy, and a boost in lysosomal and mitochondrial biogenesis (Herzig and Shaw 2018).More chronic acZvaZon of AMPK induces the expression of a core group of genes including peroxisome proliferator-acZvated receptor ɣ coacZvator 1 ⍺ (PGC-1⍺), a posiZve regulator of mitochondrial gene expression from the nuclear genome (Malik et al. 2023;Reznick and Shulman 2006;Scarpulla 2002).In addiZon to transcripZonal sZmulaZon, AMPK is thought to sZmulate PGC-1⍺ acZvity through a combinaZon of direct or indirect pospranslaZonal modificaZons (Herzig and Shaw 2018), consequently promoZng the expression of mitochondrial transcripZon factor A (TFAM), the high-mobility group protein that is required for the maintenance and expression of mtDNA (Larsson et al. 1998;Parisi and Clayton 1991;Garstka et al. 2003).Increased TFAM levels lead to elevated mtDNA copy number, allowing for the overall expansion of funcZonal mitochondria (Ekstrand et al. 2004;Bonekamp et al. 2021).Thus, acZvaZon of the AMPK-PGC-1⍺ axis increases mitochondrial biogenesis to augment respiratory capacity in response to a low cellular energy state.
In line with this concept, the pharmacological acZvaZon of AMPK using the agonist 5aminoimidazole-4-carboxamide ribonucleoside (AICAR) has been shown to improve muscle funcZon in different models of mitochondrial myopathy caused by complex IV (CIV) defects (Peralta et al. 2016;Viscomi et al. 2011).However, neither of the above-cited studies found evidence for the expected increase in mitochondrial biogenesis in terms of elevated mtDNA copy number or mitochondrial mass, leaving the mechanisms underlying these posiZve effects of AICAR somewhat unclear (Viscomi et al. 2011;Peralta et al. 2016).In vivo, AICAR is converted to the inosine pathway intermediate ZMP that acts as an AMP analog.In addiZon to binding and sZmulaZng AMPK, ZMP modulates the acZvity of other AMP-regulated enzymes such as glycogen phosphorylase and fructose-1,6-bisphosphatase (Longnus et al. 2003;Young, Radda, and Leighton 1996;Vincent et al. 1991).Therefore, some of the posiZve effects of AICAR in models of mitochondrial disease might potenZally be apributable to AMPK-independent acZons of the drug.
More specific, non-AMP-mimeZc AMPK agonists such as A-769662 and compound 991 have been developed (Cool et al. 2006;Xiao et al. 2013;Göransson et al. 2007), but have not yet been as widely adopted as AICAR.Therefore, their impact on mitochondrial funcZon and especially mtDNA levels has not been conclusively determined.However, A-769662 and other specific AMPK agonists showed posiZve effects in a panel of paZent-derived cell lines with geneZcally heterogeneous forms of mitochondrial defects (Moore et al. 2020).The group of cell lines benefiZng from AMPK agonists included not only cell lines with direct defects in the OXPHOS machinery, but also ones manifesZng mitochondrial dysfuncZon due to defects in mtDNA maintenance.Moreover, a recent study reported an increase in both mitochondrial mass and mtDNA copy number in HEK293 cells treated with compound 991 (Malik et al. 2023).These findings raise the possibility that specific AMPK agonists may be beneficial even in cases with indirect OXPHOS defects resulZng from decreased mtDNA copy number, i.e. mtDNA depleZon.
In this study, we use a Flp-in T-Rex 293 cell line with inducible expression of PolɣA D1135A to address the role of AMPK acZvity during escalaZng mitochondrial dysfuncZon caused by severe mtDNA depleZon.We first delineate the Zming of events during the advancing decline in mtDNA levels caused by the expression of the dominant-negaZve D1135A variant.We demonstrate that AMPK acZvaZon is an early event that occurs before the cellular energy state or mitochondrial membrane potenZal is measurably affected, confirming that AMPK successfully restores energy homeostasis during the early stages of mitochondrial dysfuncZon.Notably, the observed acZvaZon is limited to AMPK molecules in the mitochondrial fracZon of the cell, while cytosolic AMPK is not affected.Next, we show that AMPK contributes to the maintenance of mitochondrial membrane potenZal even under basal condiZons, and that sZmulaZon of AMPK acZvity using the specific agonist A-769662 is sufficient to fully restore the mitochondrial membrane potenZal in mtDNA-depleted cells.This posiZve impact of A-769662 can be at least parZally apributed to a small increase in the levels of mtDNA and respiratory chain subunits; however, these benefits were insufficient to overcome the proliferaZon defect seen in conjuncZon with the mtDNA depleZon.Treatment with the AMPK agonist also elevated mitochondrial membrane potenZal in cell lines derived from paZents suffering from Polɣ-associated mitochondrial myopathy, but then through an mtDNA-independent mechanism.Our findings also uncover differenZal effects of the AMPK agonist in control cells and ones suffering from severe mtDNA depleZon.

Cell culture
Inducible Flp-in TM T-Rex TM 293 cell lines (female) carrying one integrated copy of myc-POLG (wt or D1135A) were generated previously (Wanrooij et al. 2007).Cells were grown in low-glucose DMEM medium (Gibco) containing 1 g/L glucose, 4 mM GlutaMAX TM , 1 mM sodium pyruvate, 50 µg/mL uridine, 10% heat-inacZvated fetal bovine serum (Gibco) and were culZvated in a 37°C humidity incubator at 8% CO2.Expression of myc-Polɣ was induced by the addiZon of 3 ng/mL doxycycline (Sigma) to the medium for the Zme periods indicated in the legends.Doxycyclinecontaining media was exchanged every 2-3 days.When indicated in the figures, AMPK acZvity was modulated by the administraZon of its pharmacological agonist, A-769662 (Sigma), while control, "untreated" cells received an equivalent volume of DMSO.The ρ 0 cells used as a control were generated by 44-day treatment of the myc-POLG D1135A Flp-in TM T-Rex TM 293 cell line with 150 ng/ml ethidium bromide, and the lack of mtDNA confirmed by qPCR (Fig. S1C).Growth of the ρ 0 cells was in high-glucose (4.5 g/L) DMEM media supplemented as above.
Human primary fibroblasts derived from two paZent biopsies bearing the PolɣA mutaZons Y955C or V1106A, both homozygous, were culZvated in the same condiZons as the Flp-in T-Rex 293 cells.The control cells used in parallel were sex-and age-matched.The research presented in this arZcle complies with all relevant ethical regulaZons.Wripen informed consent was obtained from all parZcipants or their guardians.This study was approved by the Queen Square Research Ethics Commipee, London, UK (09/H0716/76).

Cell prolifera/on
Growth curves were generated by seeding 20 000 cells/well in 12-well plates.Each day, cells from three wells per condiZon were trypsinized, cells were resuspended in medium, and counted using an automated cell counter (Countess II FL, Invitrogen).Trypan blue was used to assess cell viability and to exclude dead cells during cell counZng.

MtDNA copy number
Total DNA was isolated from cells at 80% confluency in a 6-well plate using the NucleoSpin® Tissue DNA isolaZon kit (Macherey-Nagel), according to the manufacturer's instrucZons.MtDNA copy number was analyzed essenZally as previously described (Repolês et al. 2021).Short targets in both the mtDNA and the nuclear DNA were quanZfied in duplicate by quanZtaZve real-Zme PCR using 6-12 ng total DNA in a 20 μl reacZon containing 0.2 μM forward and reverse primers and 10 μl of 2 × SyGreen Mix (qPCRBIO #PB20.14-05) in a LightCycler 96 instrument (Roche).Primer pairs targeted the 16S rDNA region in the mtDNA (forward GTCAACCCAACACAGGCATGCT, reverse CGTGGAGCCATTCATACAGGTCC); and a region of the single-copy XPC gene in the nuclear genome (forward GCTGGACCATCTGCTGAACCC, reverse TCCTTCCACCCCTCACCTTATGT).All primers were confirmed to target only the single target region of the genome using the BLAST tool.Cycling condiZons were: 95°C 180 sec, 35 cycles of (95°C 10 sec, 57°C 10 sec, 72°C 20 sec with signal acquisiZon), melZng curve (95°C 5 sec, 65°C 60 sec, heaZng to 97°C at 0.2°C/sec with conZnuous signal acquisiZon).Cq values determined by the LightCycler 96 somware (Roche) were used to calculate the copy number of mtDNA relaZve to nuclear DNA using the Pfaffl method (Pfaffl 2001) and ploped with GraphPad Prism somware (version 10; GraphPad Somware Inc, CA).

Cell frac/ona/on
To determine the sub-cellular localizaZon of (p)AMPK, two 10 cm plates per condiZon were grown to 80% confluence, harvested, washed twice with cold PBS, and then resuspended and incubated in 0.1 × hypotonic buffer (4 mM Tris-HCl pH 7.8, 2.5 mM NaCl, 0.5 mM MgCl2 containing protease and phosphatase inhibitors) for 10 min at 4°C.Cells were homogenised in a 1 mL glass/glass Wheaton Zght-fiƒng dounce (AcZve MoZf) unZl ca 90% of cells were disrupted, and samples made isotonic by addiZon of 1/10 vol of 10 × homogenizaZon buffer.Nuclear, cytosolic, and crude mitochondrial fracZons were obtained amer differenZal centrifugaZon.Briefly, nuclear fracZons were pelleted by centrifugaZon at 1 200 × g for 3 min, and the supernatant containing cytosolic and mitochondrial fracZons was transferred to new tubes.The centrifugaZon was repeated twice to remove any remaining nuclear material.From the resulZng supernatant, mitochondria were pelleted by centrifugaZon at 17 000 × g for 10 min and the supernatant was collected as the cytosolic fracZon.Proteins from the different fracZons were extracted and quanZfied as described above.

Blue na/ve PAGE (BN-PAGE) and CIV in-gel ac/vity
To assess the integrity and funcZon of the mitochondrial complexes and supercomplexes, mitochondria were isolated from harvested cells, solubilized, and resolved in blue naZve PAGE to further perform CIV in-gel acZvity determine and protein levels by immunobloƒng according to a protocol adapted from (Frezza, Cipolat, and Scorrano 2007).Briefly, cells from two 150 mm plates per condiZon were harvested at 90% confluence, suspended in IBc buffer (10 mM Tris-MOPS, 1 mM EGTA, 0.2 M sucrose) containing protease and phosphatase inhibitors, and homogenized in a 15 mL glass/glass Wheaton Zght-fiƒng dounce (AcZve MoZf) unZl they were 55-65% disrupted (posiZve for Trypan blue staining according to automated cell counZng (Countess II FL, Invitrogen).Mitochondrial pellets were obtained by differenZal centrifugaZon, resuspended in IBc buffer and the proteins quanZfied using the BCA method.Aliquots of mitochondrial fracZon were flash-frozen in liquid nitrogen and stored at -80°C for up to 2 weeks.On the day of the assay, mitochondrial pellets were thawed on ice and solubilized in NaZvePAGE TM buffer (Invitrogen) containing 4% digitonin and protease inhibitors.Amer 1.5 h incubaZon with solubilizaZon buffer, the samples were centrifuged at 16000 × g for 20 min and the supernatant containing solubilized mitochondria was mixed with NaZvePAGE TM G-250 sample addiZve (Invitrogen).Mitochondrial samples (100 µg for in-gel acZvity; 25 µg for WB) were loaded onto NaZvePAGE TM 3-12% Bis-Tris gel (Invitrogen) and resolved by BN-PAGE containing NaZvePAGE TM Running buffer (Invitrogen) and NaZvePAGE TM cathode buffer addiZve (Invitrogen).NaZveMark™ Unstained Protein Standard (Invitrogen) was used for molecular weight esZmaZon of protein in naZve gel electrophoresis.To assess the acZvity of CIV, the gel was incubated overnight with substrate soluZon (50 mM Na-phosphate buffer pH7.4,0.5 g/L 3,3'-diamidobenzidine tetrahydrochloride, 1.0 g/L bovine cytochrome c, 2 mg/L catalase, 75 g/L sucrose) unZl the brown signal appeared on the gel.Images of the gel were recorded using a high-quality imager (Epson PerfecZon V700 Photo).

Mitochondrial membrane poten/al (MMP)
Cells were grown to 70-80% confluence in 6-well plates.For each condiZon, one well was used for tetramethylrhodamine ethyl ester (TMRE) staining and another for 10-N-nonyl acridine orange (NAO) staining.To assess MMP, adherent cells were incubated with medium containing 200 nM TMRE for 30 min at 37°C.Amer incubaZon, cells were trypsinized and resuspended in PBS containing 1% BSA, and TMRE fluorescence was determined by flow cytometry (BD Accuri TM C6 Plus, BD Biosciences).TMRE fluorescence was also measured in cells further treated with the mitochondrial uncoupler BAM-15 (60 µM) for 30 min at 37°C to exclude non-mitochondrial TMRE fluorescence.Mitochondrial mass was esZmated by the fluorescence of cells stained with 10 nM NAO for 30 min at 37°C.The MMP was calculated by subtracZng the TMRE fluorescence of BAM-15 uncoupled cells from the respecZve coupled ones (ΔTMRE = TMREcoupled -TMREuncoupled) and the result was normalized by the NAO fluorescence (ΔTMRE /NAO).

Nucleo/de pool measurement
NucleoZde (dNTP and NTP) pools were measured in a single run using a recently developed isocraZc reverse phase HPLC-based technique (Ranjbarian et al. 2022), with minor modificaZons for the analysis of the relaZve levels of AMP, ADP and ATP (Purhonen, Hofer, and Kallijärvi 2024;Debar et al. 2023).Briefly, cells were culZvated in 10 cm plates to 50-75% confluence, washed with ice-cold PBS, and scraped off the plates in the presence of 0.5 mL icecold 80% (v/v) methanol in water.Cells were centrifuged for 1 min at 17 000 × g in a cooled centrifuge, and the supernatants containing the free nucleoZdes were collected in new tubes and stored at -80°C.The cell extracts were purified by solid phase extracZon (SPE) using Oasis WAX 3cc cartridges (Waters CorporaZon, Milford, MA, USA) in a similar manner as described before for the measurement of NTPs, dNTPs, and ADP (Ranjbarian et al. 2022), but with some minor modificaZons in the SPE step in experiments where also AMP was measured (Debar et al. 2023).The SPE-purified samples were subsequently analyzed by HPLC run at 1 ml/min using a 150 mm × 4.6 mm SunShell C18-WP HPLC column from ChromaNik Technologies Inc (Osaka, Japan).The aqueous mobile phase contained 5.8% (v/v) acetonitrile, 0.7 g/L tetrabutylammonium bromide as ion pairing agent and varying concentraZons of potassium phosphate at pH 5.6.The analysis of NTPs and dNTPs was performed with the isocraZc Fast Protocol (Ranjbarian et al., 2022), whereas the analysis of AMP, ADP and ATP was performed with a phosphate gradient (Purhonen, Hofer, and Kallijärvi 2024).

Cell cycle
Cells were grown to 70-80% confluence in 6-well plates, harvested and fixed with ice-cold 70% ethanol, and kept in freezer -20°C up to 1 week.For DNA staining, cells were washed with PBS and resuspended in staining soluZon containing 0.02 mg/mL propidium iodide, 0.1% Triton X-100, and 0.2 mg/mL RNAse in PBS.Cells were incubated with staining soluZon for 30 min at RT. Cell cycle was assessed by flow cytometry (BD Accuri TM C6 Plus, BD Biosciences) and data was analysed using FlowJo TM BD somware.

Cell death assay
Levels of cell death were measured in cells expressing Polγ WT and Polγ D1135A for 6 days using the FITC/Annexin V Dead Cell Apoptosis kit (Molecular Probes, Invitrogen) according to the manufacturer's instrucZons.Uninduced Polγ WT cells treated with 10 µM of camptothecin for 16 hours were used as a posiZve control for cell death.Cells were harvested, pelleted by centrifugaZon, washed once with 1 × PBS and resuspended in PBS to a concentraZon of 10 6 cells/ml.For each measurement, 10 5 cells were combined with 5 µl FITC Annexin V soluZon and 1 µl of 100 µg/mL propidium iodide soluZon.The samples were incubated at RT, protected from light, for 15 min.Amer incubaZon, 400 µL of annexin binding buffer was added to the mixture, gently mixed, and kept on ice.Samples were measured immediately amer the addiZon of the buffer on a BD Accuri C6 Plus (BD Biosciences) flow cytometer.Unstained and uninduced cells were used as a negaZve control.

Graphs and sta/s/cal analysis
All graphs and staZsZcal analyses were generated with GraphPad Prism somware (version 10; Graphpad Somware Inc, CA).For experiments comparing uninduced and induced Flp-in T-REx 293 cells across mulZple Zmepoints, data of all induced samples was normalized to the induced sample at day 0 to eliminate technical variaZon arising from different doxycycline preparaZons.To clearly indicate this, a verZcal doped line is used to divide the graphs into two halves, where data sets in each half are normalized to their respecZve 0-day sample.Results are expressed as mean ± standard deviaZon, and staZsZcal differences (unpaired t-test) indicated by asterisks (* P≤0.05, ** P≤0.01, *** P≤0.001, **** P≤0.0001).The number of biological replicates (n) is indicated in the figure legends.

Results
The progressive mtDNA deple/on induced by Polɣ D1135A expression impairs OXPHOS and cell prolifera/on within 4 to 6 days We used the previously established Flp-In T-REx 293 cell line to express the dominant-negaZve D1135A variant of the mitochondrial DNA polymerase PolɣA (henceforth referred to as Polɣ D1135A ) in a doxycycline-inducible manner (Wanrooij et al. 2007).A similar cell line expressing the wildtype PolɣA (Polɣ wt ) was used as a control.To beper mimic physiological energy metabolism and mitochondrial funcZon, the cells were maintained in low-glucose media (5 mM glucose).The expression of Polɣ wt and Polɣ D1135A was induced by the addiZon of 3 ng/ml of doxycycline, a low concentraZon that was selected because it decreased mtDNA levels to below 15% of starZng levels by day 3 in Polɣ D1135A -expressing cells (Fig. 1A).No leaky expression of Polɣ wt or Polɣ D1135A was observed in the absence of doxycycline (Fig. S1A-B).
To follow the dynamics of the mitochondrial and cellular effects of mtDNA depleZon, the expression of Polɣ D1135A was induced, and samples were collected 3, 6 and 9 days amer inducZon.Cells expressing Polɣ D1135A exhibited a considerably decreased copy number of 8-15% of starZng levels across all three post-inducZon Zmepoints (Fig. 1A), while the protein levels of respiratory chain (RC) subunits showed a decreasing trend over the Zme course (Fig. 1B-G).Subunits of complex IV and V were mildly affected already on day 3, and by day 6, all complexes except for the fully nuclear-encoded CII showed diminished protein levels.In line with the trend in the Flp-In T-REx 293 ρ 0 control cells, subunits of complexes I, III and IV were most severely affected, while the decrease in ATP5 (CV) levels was more moderate (ca 70% of iniZal protein levels remaining on day 9; see Fig. S1C-D for validaZon of the ρ 0 cells).In contrast to Polɣ D1135A , the expression of Polɣ wt had liple or no impact on copy number or RC complex levels, confirming that the observed effects were not related to an overwhelmed protein import system upon overexpression of a mitochondrially-targeted protein.Blue-naZve analysis confirmed a pronounced loss of assembled RC complexes and supercomplexes in Polɣ D1135A cells on day 6 of inducZon, including the appearance of a smaller subcomplex of CV that was also observed in ρ 0 cells by us and others (Fig. 1H-J) (Herrero-Mar•n et al. 2008).CIV acZvity on day 6 of inducZon was notably decreased compared to uninduced Polɣ D1135A cells (Fig. 1H-J).Accordingly, the mitochondrial membrane potenZal (MMP), measured using the caZonic dye tetramethylrhodamine methyl ester perchlorate (TMRE) and normalized to mitochondrial mass assessed by 10-N-nonyl acridine orange (NAO), diminished to 35 % of starZng levels by day 6 (Fig. 1K).Cell proliferaZon stagnated already on day 5, suggesZng that the decrease in OXPHOS and MMP observed in day 6 samples may be physiologically relevant already on day 5 when no samples were collected for protein or MMP analysis (Fig. 1L).

Mitochondria-associated AMPK is ac/vated early on during the course of the mitochondrial dysfunc/on
To further assess the effects of mtDNA depleZon on energy metabolism, we next used HPLC to quanZfy the levels of adenine nucleoZdes in Polɣ D1135A or Polɣ wt -expressing cells before inducZon, as well as 3 or 6 days amer the addiZon of doxycycline.In line with the observed MMP defect, the ADP level in Polɣ D1135A -expressing cells was elevated at day 6, and AMP levels showed a reproducible increase that however failed to reach staZsZcal significance (p=0.117;Fig. 2A).The cells' energy status, appraised by the raZo of ATP to the sum of AMP and ADP, thus decreased 2.5-fold compared to uninduced Polɣ D1135A cells (computed ATP/(AMP+ADP) raZo of 12.0 and 4.7 in uninduced and day-6 cells, respecZvely).Further analysis of nucleoZde levels revealed a decline in all pyrimidine nucleoside triphosphates (CTP, UTP, dCTP and dTTP) in Polɣ D1135A cells, which is an expected consequence of the respiratory chain deficiency impairing pyrimidine synthesis (Fig. S1E-F) (King and Apardi 1989;Grégoire et al. 1984).The Polɣ D1135A cells also exhibited acZvaZon of the ATM checkpoint kinase and an altered cell cycle profile (Fig. S1G-I).No signs of increased apoptosis were detected (Fig. S1J).These findings are in good agreement with the previouslyreported cell cycle alteraZons and ATM acZvaZon observed in response to mtDNA instability, and the fact that mitochondrial dysfuncZon can lead to nuclear DNA instability through a host of different mechanisms (Veatch et al. 2009;Hämäläinen et al. 2019;Marcon et al. 2022;Cao et al. 2022).
Given the altered ATP/(AMP+ADP) raZo in Polɣ D1135A -expressing cells, we next examined the status of the AMPK energy sensor, a fracZon of which localizes to the mitochondrial outer membrane (Drake et al. 2021).Specifically, we were interested in understanding when over the course of the progressive mitochondrial dysfuncZon in the Polɣ D1135A cells AMPK becomes acZvated and thus phosphorylated.Analysis of AMPK phosphorylaZon on threonine-172 in whole-cell lysates suggested that AMPK was not acZvated unZl day 9 of inducZon (Fig. 2B).However, evaluaZon of AMPK in different subcellular localizaZons revealed that mitochondriaassociated AMPK underwent acZvatory phosphorylaZon already on day 3, while the phosphorylaZon state of cytosolic AMPK remained at basal levels throughout the Zme course (Fig. 2C-D).Notably, the energy stress in Polɣ D1135A cells did not alter the parZZoning of AMPK between the cytosolic and crude mitochondrial fracZon, indicaZng no significant recruitment of addiZonal AMPK complexes to the mitochondria (Fig. 2D).As expected, the induced expression of Polɣ wt did not result in acZvaZon of AMPK in any cellular compartment (Fig. 2C).These results show that the energy defect caused by mtDNA depleZon in Polɣ D1135A cells primarily acZvates AMPK that is already associated with mitochondria.
The dynamics of the ensuing mitochondrial dysfuncZon in our experimental Zme course are summarized in Fig. 2E: the first effects on RC protein levels are observed on day 3 of inducZon, when the acZvaZon of mitochondria-associated AMPK also becomes apparent.Cell proliferaZon declines by day 5, and by day 6 the cells exhibit full-blown mitochondrial dysfuncZon involving diminished levels of most RC proteins, decreased MMP, and an overall decline in cellular energy state.

AMPK helps maintain basal MMP, and s/mula/on of AMPK ac/vity rescues the MMP loss in Polɣ D1135A cells
Based on the Zmeline in Fig. 2E, we focused our efforts on Polɣ D1135A cells on day 4 of inducZon, reasoning that this was a Zmepoint when the cells were suffering from mild mitochondrial dysfuncZon that could esZmate the situaZon in mild to moderate mitochondrial disease.We asked whether pharmacological modulaZon of AMPK acZvity could alleviate some elements of the mitochondrial dysfuncZon, and treated uninduced or induced Polɣ D1135A cells with the specific AMPK acZvator A-769662 for 48 or 72 h, as indicated at the top of Fig. 3A.Treatment with A-769662 resulted in the expected sZmulaZon of AMPK acZvity, as evidenced by the increased phosphorylaZon of a major AMPK substrate acetyl-coenzyme A carboxylase (ACC) on serine-79, a site specifically phosphorylated by AMPK (Davies, Sim, and Hardie 1990) (Fig. S2A-B).The phosphorylaZon status of AMPK itself did not change appreciably following A-769662 treatment (Fig. S2C).Notably, A-769662 treatment increased mitochondrial membrane potenZal in a Zme-dependent manner, with 72-h treatment increasing MMP to up to 140% of starZng levels in Polɣ D1135A -expressing cells (Fig. 3A).The impact of A-769662 on MMP was more pronounced in cells with a low starZng MMP (80% and 140% increase in uninduced and induced Polɣ D1135A cells amer 72 h, respecZvely), and was sufficient to restore the MMP of induced Polɣ D1135A cells to levels corresponding to the uninduced cells already amer 48 h of treatment.We conclude that A-769662-mediated AMPK acZvaZon is sufficient to correct a considerable drop in mitochondrial membrane potenZal, and that the sustained effect of the acZvator can be observed at least 24 h amer removal of the drug (Fig. 3A; 48 h treatment).
In contrast to AMP-mimeZc AMPK acZvators such as AICAR, A-769662 has been reported to be specific to AMPK (Xiao et al. 2013;Göransson et al. 2007).To confirm that the effect of A-769662 on MMP was AMPK-dependent, we next silenced the catalyZc ⍺-subunit of AMPK by siRNA treatment, and followed the impact of A-769662 on MMP in uninduced or induced Polɣ D1135A cells.The siRNA treatment resulted in efficient knockdown of AMPK⍺ as well as a clear drop in MMP in both uninduced and induced cells, indicaZng that AMPK is required for maintaining normal mitochondrial membrane potenZal (Fig. 3B-C).This finding is in good agreement with the reported impairment of mitochondrial respiraZon in mouse models lacking ⍺1 and/or ⍺2 subunits of AMPK (LanZer et al. 2014;Viollet et al. 2009).Further, as was seen in Fig. 3A, treatment with A-769662 increased the MMP in both the uninduced and induced cells, and the increase was more substanZal in the induced cells that started from a lower MMP (Fig. 3C).However, A-769662 had no impact on the MMP in cells transfected with an siRNA against AMPK, confirming that the sZmulatory effect of A-769662 on MMP is dependent on the presence of AMPK.
We have previously shown that the cell proliferaZon defect of Saccharomyces cerevisiae cells suffering from mitochondrial dysfuncZon can be rescued by increasing their MMP (Gorospe et al. 2022), and therefore now analyzed cell proliferaZon following A-769662 treatment in uninduced cells and ones induced for 6 days.As shown in Fig. 3D, the rescue in MMP observed amer A-769662 treatment of induced Polɣ D1135A cells was insufficient to improve cell proliferaZon, indicaZng that their growth defect was due to other consequences of the mitochondrial dysfuncZon than solely MMP loss.

The A-769662-mediated ac/va/on of AMPK improves mtDNA copy number and respiratory chain protein levels
Given the posiZve impact of A-769962 on the mitochondrial membrane potenZal of induced Polɣ D1135A cells, we sought to understand the molecular basis of this outcome by exploring the effect of A-769662 on the levels of mtDNA and RC subunits.A 72-h treatment of cells with A-769662 caused a small but staZsZcally significant increase in the mtDNA copy number of induced Polɣ D1135A cells, increasing mtDNA from 11% to 20% of the levels observed in uninduced cells (Fig. 4A).InteresZngly, A-769662 treatment did not alter mtDNA levels in uninduced cells.Analysis of RC protein levels in induced cells revealed a posiZve effect of A-769662 on the analyzed CI, CIII and CIV subunits, but no alteraZon in CII or CV subunits that were only mildly depleted relaZve to uninduced cells to begin with (Fig. 4B-G).As was observed with mtDNA copy number, acZvator treatment had a differenZal effect on induced and uninduced cells: the protein levels of the laper only increased in one case (CI), while they decreased in two cases (CIII and CIV) and were not affected for CII and CV.Taken together, these results suggest that specific AMPK sZmulaZon can alleviate the effects of mtDNA depleZon and that the downstream effects of A-769662-mediated AMPK acZvaZon can vary, likely depending on the energy metabolic context of the cell.
So far, our experiments were carried out in cells expressing a dominant-negaZve Polɣ mutant with highly deleterious effects on mtDNA maintenance and mitochondrial funcZon.Given that specific AMPK acZvaZon imparted some modest but promising effects even under this extreme scenario, we next applied A-769662 to cells with a milder and more clinically-relevant level of mtDNA depleZon.To this end, we made use of fibroblasts derived from paZents suffering from mitochondrial myopathy, as well as sex-and age-matched control cell lines.The mutant lines were homozygous either for PolɣA V1106A or for the PolɣA Y955C mutaZon that is the most common autosomal dominant mutaZon in POLG and a cause of progressive external ophthalmoplegia (PEO) (Rahman and Copeland 2019).Both mutant lines showed a depleZon of mtDNA to approximately 60% of the levels found in control cells (Fig. 4H; 61% and 54% in Polɣ Y955C and Polɣ V1106A cells, respecZvely).Treatment with A-769662 led to a slight increase that lacked staZsZcal significance in paZent cell lines, and no change or a decrease in the control lines (Fig. 4H).However, A-769662-treatment resulted in an elevaZon of MMP, indicaZng that it enhanced mitochondrial funcZon (Fig. 4I).Unlike the observaZons in uninduced vs. induced Polɣ D1135A cells in Fig. 3A, the impact of A-769662 was comparable in control and paZent cell lines, with treatment on average increasing MMP to 125% of starZng values.
Taken together, the experimental findings in Figures 3-4 indicate that specific AMPK acZvaZon using A-769662 has a beneficial impact on mitochondrial funcZon not only in the Flp-In T-REx 293 model system with severe mtDNA depleZon, but also in mitochondrial myopathyassociated cells (Fig. 3A; Fig. 4I).However, the mechanisms mediaZng the posiZve effects of pharmaceuZcal AMPK acZvaZon differed between our experimental models: while the increased MMP in the Polɣ D1135A -expressing cells was at least parZally mediated by an increase in mtDNA copy number (Fig. 4A), A-769662 did not significantly improve mtDNA levels in paZent fibroblasts with less severe mtDNA depleZon (Fig. 4H).The results of this study warrant expanding the repertoire of potenZal therapeuZc applicaZons of specific AMPK acZvators to include mitochondrial defects caused by mtDNA depleZon.

Discussion
Given its role as a master metabolic regulator tasked with ensuring energy homeostasis and adaptability to changing metabolic demands, AMPK naturally serves as a guardian of mitochondrial integrity and funcZon.This mulZfaceted role involves regulaZon of mitochondrial biogenesis, dynamics and morphology as well as the quality control of mitochondria through mitophagy (Herzig and Shaw 2018).AMPK signaling also helps determine mitochondrial mobility in neuronal axons, essenZally concentraZng mitochondria at sites of energeZc stress (Wapers et al. 2020).
In the current study, we concretely show that AMPK acZvity is essenZal for maintaining MMP in healthy cells, and that mitochondria-associated AMPK is acZvated early on during the progression of mitochondrial dysfuncZon, in line with its established role in restoring energy homeostasis (Fig. 2; Fig. 3C).Here it is relevant to note that the phosphorylaZon level of cytosolic AMPK was not affected, and that assessment of total cellular AMPK would yield a very different result in terms of the Zming of acZvaZon than that on mitochondria-associated AMPK (AMPK acZvaZon on day 9 vs. day 3 in total extracts and mitochondrial fracZons, respecZvely).Further promoZon of AMPK acZvity using the specific small-molecule acZvator A-769662 was sufficient to recover normal MMP in cells suffering from severe mitochondrial dysfuncZon following mtDNA depleZon (Fig. 3).Although A-769662 treatment improved many features of the induced Polɣ D1135A cells including mtDNA copy number and RC protein levels (Fig. 3-4), it did not correct the proliferaZon defect of these cells (Fig. 3D).In light of past work connecZng AMPK acZvaZon to a G1/S checkpoint through the p53-p21-cyclin E axis (Mitra et al. 2009;Jones et al. 2005;Mandal et al. 2005), the persisZng proliferaZon defect in AMPK-acZve cells is not enZrely surprising.On the other hand, the proliferaZon defect was not solely maintained by AMPK, since inhibiZng AMPK alone with the small-molecule inhibitor dorsomorphin was insufficient to rescue proliferaZon (Fig. S2D-H).It remains to be seen whether simultaneous inacZvaZon of AMPK and the correcZon of MMP through an AMPK-independent mechanism could be a feasible approach to restore normal proliferaZon in cells with severe mtDNA depleZon.
To our knowledge, this is the first report of a sZmulatory effect of A-769662 on mtDNA copy number.Previous findings regarding the effects of AMPK agonists on mtDNA levels have come to disparate results.While AMPK acZvaZon via administraZon of the AMP-mimeZc acZvator AICAR did not impact mtDNA levels in mice with RC defects (Peralta et al. 2016;Viscomi et al. 2011), treatment of HEK293T cells with compound 991 increased copy number by ~ 1.3-fold (Malik et al. 2023).It should however be noted that all these three studies were carried out in models with a normal starZng level of mtDNA, while in our study the largest effects of A-769662 on copy number were seen in the induced Polɣ D1135A cells with extreme mtDNA depleZon (ca 15% mtDNA remaining), while control cells or paZent fibroblasts with milder depleZon (ca 60% mtDNA remaining) showed no increase with A-769662 treatment (Fig. 4A and H).The actual impact of the various AMPK acZvators on mtDNA copy number can therefore not be directly compared across the different studies, and should be more thoroughly invesZgated in the future.Regardless, our findings call for exploraZon of the uZlity of specific AMPK acZvators as a potenZal experimental and/or therapeuZc avenue even in disorders caused by mtDNA depleZon.
The observed differenZal effects of A-769662 on control cells vs. Polɣ D1135A -expressing cells with mtDNA depleZon are intriguing.Why does AMPK acZvaZon lead to different outcomes depending on the cellular context?One possible explanaZon is that in mtDNA depleted cells, the increased AMP or ADP levels contribute to co-acZvaZon of AMPK such that a higher acZvity is reached than with A-769662 alone.This hypothesis is supported by the fact that AMP and A-769662 bind different sites of the heterotrimeric AMPK complex -while AMP binds the cystathionine-β-synthase (CBS) domains on the regulatory ɣ-subunit, A-769662 binds the allosteric drug and metabolite-binding site (ADaM) located between the kinase domain of the catalyZc ⍺-subunit and the carbohydrate-binding module of the β-subunit (Xiao et al. 2013).Bultot et al. observed considerable co-acZvaZon of AMPK complexes derived from mouse hepatocytes or C2C12 myotubes using acZvators that bound to these two disZnct sites (Bultot et al. 2016), so we speculate that the same may be occurring in the Polɣ D1135A -expressing cells treated with A-769662 in our study.The fact that no preferenZal MMP increases were seen in the paZent-derived fibroblasts over controls in Fig. 4I is well in line with this reasoning.Although we did not measure the adenine nucleoZde pools in the paZent fibroblasts, their MMP is indisZnguishable from that of the control cells, suggesZng that their mitochondrial ATP producZon and thus cellular energy status is in the normal range.
In humans, each of the three AMPK subunits is present in mulZple isoforms (two isoforms each of ⍺ and β, three of ɣ), theoreZcally giving rise to up to 12 disZnct heterotrimeric AMPK complexes.The different AMPK complexes may vary in terms of Zssue-specificity, subcellular localizaZon and/or response to stress sZmuli (Herzig and Shaw 2018).A shortcoming of A-769662 in the context of targeZng mitochondrial stress is that it only acZvates AMPK complexes containing a β1-, but not a β2-subunit (Scop et al. 2014), and is therefore not expected to exert an effect on AMPK complexes associated with skeletal muscle mitochondria that primarily contain β2-subunits (Drake et al. 2021).In contrast, compound 991 (also known as ex229) binds the same ADaM site on AMPK as A-769662 but is 5-10 Zmes more potent and shows no isoform specificity regarding the β-subunit (Lai et al. 2014;Xiao et al. 2013), highlighZng it as the preferenZal AMPK agonist for further studies.
The targeZng of a central regulator like AMPK as a treatment strategy comes with its own risks and challenges.In some studies, the chronic acZvaZon of AMPK has shown undesirable side effects such as cardiac and kidney hypertrophy and Alzheimer's disease (Wilson et al. 2021;Zimmermann et al. 2020).However, clinical trials and animal studies with other AMPK agonists did not reveal harmful side effects, indicaZng that acZvaZng this central regulator is feasible and safe with the right compounds (López-Pérez et al. 2021;Pinkosky et al. 2016;Ray et al. 2024;Cusi et al. 2021).In conclusion, the posiZve effects of AMPK sZmulaZon on mtDNA levels and/or mitochondrial membrane potenZal in cell lines with mild to severe mtDNA depleZon broaden the potenZal applicaZons of small-molecule AMPK agonists, and support further exploraZon of the effects of this group of compounds in the context of mitochondrial dysfuncZon.

Figure legends Figure 1 .
Figure legends

Figure 2 .
Figure 2. MtDNA loss leads to energy imbalance and ac/va/on of mitochondria-associated AMPK.(A) Levels of AMP, ADP and ATP in cells either uninduced (day 0) or expressing PolG wt (grey) or PolG D1135A (pink) for 3 and 6 days; n=4.(B) Immunobloƒng indicates levels of phosphorylated AMPK (p-AMPK) and total AMPK in cells overexpressing PolG wt and PolG D1135A .β-AcZn was used as loading control.A representaZve image of n=2 is shown.(C, D) Cell fracZonaZon was performed to show the subcellular localizaZon of AMPK.Immunobloƒng shows the distribuZon of p-AMPK and AMPK in the cytosolic and mitochondrial fracZon of cells overexpressing PolG wt (n=3) and PolG D1135A (n=4).QuanZficaZon of the p-AMPK/AMPK raZos is shown in the bopom panel.GAPDH and VDAC are shown as a purity control for the cytosolic and mitochondrial fracZons, respecZvely.TFAM levels provide an indirect esZmaZon of mtDNA copy number (see also Fig. 1A).(E) A schemaZc summarizing the advance of the cellular phenotypes associated with the progressive loss of mtDNA shown in figures 1 and 2. (n, number of biological replicates; ns, not significant; DOI, days of inducZon; * P≤0.05; ** P≤0.01).

Figure 4 .
Figure 4. Pharmacological ac/va/on of AMPK ameliorates pathological phenotypes in cells expressing mutant variants of Polγ.(A) RelaZve mtDNA copy number amer 72 h treatment with A-769662 (100 µM) reveals increased mtDNA levels in cells expressing Polγ D1135A (pink) for 4 days compared to its respecZve DMSO control.n=3.(B) Expression of subunits of the mitochondrial respiratory complexes (CI-CV) amer 48 h and 72 h treatment with A-769662 (100 µM) in uninduced cells and in cells expressing Polγ D1135A (myc-Polγ) for 4 days.Protein levels were assessed by WB using an OXPHOS cocktail anZbody.Protein molecular weight (kDa) are indicated on the lem and protein names (and their respecZve respiratory complexes) are indicated on the right.A representaZve image of n=3-4 is shown.(C, D, E, F and G) QuanZficaZon of the blots shown in (B); protein levels were normalized against GAPDH.(H) RelaZve mtDNA copy number from human primary fibroblast cells (2 controls; 2 paZent cell lines bearing the Polγ homozygous mutaZons Y955C and V1106A, respecZvely) was measured before and amer treatment with A-769662 (100 µM) for 24 h.n=6.Asterisks on untreated samples indicate the p-value of the comparison with the untreated Control 1 cell line; the asterisk on the A-769662-treated Ctrl2 sample marks the comparison with the untreated Ctrl2.(I) The mitochondrial membrane potenZal (MMP) of human primary fibroblast cells treated or not with A-769662 (100 µM) for 72 h was esZmated by flow cytometry as described in Fig. 1K.RelaZve MMP is expressed as ΔTMRE/NAO.n=4.(n, number of biological replicates; ns, not significant; * P≤0.05; ** P≤0.01; *** P≤0.001; **** P≤0.0001).