Given previous studies have shown that MBOAT7 is a risk locus for alcohol-associated cirrhosis (Buchs et al., 2015), we investigated whether active alcohol consumption was associated with alterations in MBOAT7 function. To address this, we measured both LPI substrates and PI products of the MBOAT7 enzymatic reaction in the circulation of healthy controls compared to confirmed heavy drinkers (Figure 1). Heavy drinkers were recruited and defined by an AUDIT score (Fleming et al., 1991) greater than 16, and compared to an age- and sex-matched healthy control population (Figure 1 - figure supplement 1). In agreement with genetic studies linking MBOAT7 variants to alcohol-associated cirrhosis (Buchs et al., 2015), we find that circulating levels of metabolic products of MBOAT7 including arachidonic acid- and eicosapentaenoic acid­containing phophosphatidylinositols (PI 38:4 & PI 38:5) are significantly reduced in heavy drinkers compared to age-matched healthy controls (Figure 1). Given MBOAT7 demonstrates specificity for polyunsaturated (PUFA) acyl-CoA substrates (Gijon et al., 2008), it is important to note that only select PUFA-containing MBOAT7 products (PI 38:4 and PI 38:5) were reduced in heavy drinkers, whereas all other molecular species of PI were unaltered. We also examined the circulating levels LPI substrates of MBOAT7 but found no significant differences between controls and heavy drinkers (Figure 1). These data show that excessive alcoholic intake is associated with reduced levels of MBOAT7 product lipids, which further bolsters the concept that MBOAT7 loss of function may be causally linked to ALD progression.

Although there is some emerging evidence that MBOAT7 genetic variants may predispose humans to alcohol-induced liver injury (Buchs et al., 2015), not all human studies have found a significant association (Zhang et al., 2018; Stickel et al., 2018; Beaudoin et al., 2021). Importantly, a causal relationship between MBOAT7 and alcohol-induced liver injury has never been established to date. To address this, we have studied ethanol-induced liver injury in mice selectively lacking Mboat7 in hepatocytes or myeloid cells, given the key roles that hepatocytes and myeloid cells play in the pathogenesis of ethanol-induced liver disease progression. To generate congenic hepatocyte-specific (MboatT^HSKO) and myeloid-specific (Mboat7^MSKO) Mboat7 knockout mice we crossed mice harboring a post-FLP recombinase conditionally targeted Mboat7 floxed allele (Anderson et al., 2013; Massey et al., 2023) to mice transgenically expressing Cre recombinase under the albumin promoter/enhancer (Postic et al., 1999) or Cre knocked into the M lysozyme locus (Clausen et al., 1999) respectively. These independent Mboat7^HSKO and Mboat7^MSKO lines were then backcrossed mice >10 generations into the C57BL/6J background and subsequently subjected to ethanol exposure. Compared to control mice (Mboat7^lox/flox), Mboat7^HSKO mice had significantly reduced Mboat7 mRNA and protein expression in the liver (Figure 2a,2b), but not in other tissues (data not shown; Massey et al., 2023). Hepatocyte-specific deletion of Mboat7 resulted in enhanced ethanol-induced increases in liver weight and high concentrations of plasma alanine aminotransferase (ALT) (Figure 2c,2d). Likewise, Mboat7^HSKO mice showed elevated hepatic steatosis scores and triglyceride levels under both pair-fed and ethanol-fed conditions (Figure 2e,2f). However, hepatocyte­specific deletion of Mboat7 did not significantly alter the mRNA expression for several proinflammatory cytokines/chemokines including tumor necrosis factor a (Tnfα), transforming growth factor p (Tgfα), monocyte chemoattractant protein 1 (Mcp1), or interleukins 1p (IL-1 fl) or 6 (IL-6) in either pair-fed or ethanol-fed conditions (Figure 2g). These data demonstrate that MBOAT7 function in hepatocytes is a critical determinant of ethanol-induced liver injury, but we also wanted to explore a potential role for MBOAT7 in non-parenchymal cells given the key roles that macrophages and neutrophils play in ALD.

It is important to note during the preparation of this manuscript, an independent study discovered a potential role for MBOAT7 in suppressing toll-like receptor (TLR) signaling and pro-inflammatory cytokine production in macrophages and Kupffer cells in the context of non­alcoholic fatty liver disease (NAFLD) (Alharthi et al., 2022). Furthermore, early studies examining the expression and substrate specificity for diverse lysophospholipid acyltransferases showed that MBOAT7-driven PI remodeling was highly active in human neutrophils where it can modulate the production of pro-inflammatory arachidonic acid-derived lipid mediators (Gijon et al., 2008). Therefore, we generated myeloid-specific (Mboat7^MSKO) Mboat7 knockout mice to further interrogate cell autonomous roles in ALD progression. First to confirm efficient deletion in myeloid cells, we isolated both bone marrow-derived and thioglycolate-elicited macrophages from control and myeloid-specific (Mboat7^MSKO) Mboat7 knockout mice, which confirmed essentially no detectable MBOAT7 protein in Mboat7^MSKO mice (Figure 2 - figure supplement 1). In contrast to the enhanced ethanol-induced liver injury seen in Mboat7^HSKO mice (Figure 2), myeloid-specific (Mboat7^MSKO) Mboat7 deletion resulted in unaltered ethanol-induced effects on 8 body weight, liver weight, circulating levels of aspartate and alanine aminotransferases (AST and ALT), liver triglyceride, and cytokine expression (Figure 2 - figure supplement 1). Collectively, these results demonstrate that MBOAT7 loss of function in hepatocytes, but not myeloid cells, facilitates ethanol-induced liver injury in mice.

Given the fact that ethanol exposure is well known to reorganize hepatic lipid metabolism, we performed comprehensive lipidomic profiling of the liver to understand how MBOAT7 could potentially shape ethanol-induced lipid metabolism in the liver. First, we used a targeted approach to measure the levels of MBOAT7 substrate LPIs and product PIs. Compared to control mice (Mboat7^lox/flox), Mboat7^HSKO mice had significant accumulation of palmitate- and oleate-containing LPI substrate lipids (LPI 16:0 and LPI 18:1), with large accumulation of LPI 16:0 under ethanol-fed conditions (Figure 3a). When we examined PI species, we confirmed previous findings that Mboat7^HSKO mice have reduced levels of the major arachidonic acid­containing PI (PI 38:4) and also striking reductions in PI 38:3 under both pair- and ethanol-fed conditions (Figure 3b). In addition, Mboat7^HSKO mice also have accumulation of several other PI species including PI 34:1, PI 36:1, PI 36:2, PI 38:6, and PI 40:6, some of which are exacerbated in the ethanol-fed group (Figure 3c). Although many of these changes in MBOAT7’s substrate LPIs and product PIs are expected based on MBOAT7’s substrate specificity and previous literature, there is a clear interaction between ethanol and MBOAT7 uncovered here that has not been observed in studies using NAFLD-related models (Tanaka et al., 2021; Thangapandi et al., 2021; Xia etal., 2021).

Given this unexpected interaction between ethanol exposure and MBOAT7 within the inositol-containing phospholipid pool, we examined the hepatic lipidome more broadly. Using untargeted lipidomics, we identified a striking remodeling of the global hepatic lipidome upon ethanol feeding in Mboat7^HSKO mice (Figure 3 & Figure 3 - figure supplements 1-11). Unexpectedly, we observed a large increase in the levels of endosomal/lysosomal lipids including bis(monoacylglycero)phosphates (BMPs) and their outer mitochondrial membrane precursor phosphatidylglycerols (PGs) in ethanol-exposed MboatT^HSKO mice (Figure 3e-3g; Figure 3 - figure supplement 3 & 4). In addition, MboatT^HSKO mice had elevated levels of cardiolipin species, which are known to localize to mitochondria (Figure 3e,3h; Figure 3 - figure supplement 5). Under the pair- and ethanol-fed conditions studies here, Mboat7^HSKO mice also exhibited some more minor alterations in certain species of phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), sphingomyelin (SM), ceramides (Cer), diacylglycerols (DAG), and ether-linked phospholipids (Figure 3e; Figure 3 - figure supplements 1,2,6-11). Although several recent studies examining hepatocyte-specific Mboat7^HSKO mice have found more limited effects on the global lipidome under experimental conditions designed to stimulate non-alcoholic steatohepatitis (NASH) (Tanaka et al., 2021; Thangapandi et al., 2021; Xia et al., 2021), here we show that upon ethanol exposure, hepatocyte MBOAT7 plays a major role in shaping endosomal/lysosomal lipid homeostasis.

Given the accumulation of endosomal/lysosomal lipids including BMPs seen in ethanol-fed Mboat7^HSKO mice, we hypothesized that ethanol may perturb lysosome function in a MBOAT7-driven manner to promote liver injury. It is well known that BMPs commonly accumulate in both drug-induced and genetic lysosomal storage disorders (Gruenberg 2020; Showalter et al., 2020; Hullin-Matsuda et al., 2014), and due to their cone-shaped structure BMPs can contribute to significant membrane asymmetry that impacts intracellular lipid sorting, apoptosis, and autophagic flux (Gruenberg 2020; Showalter et al., 2020; Hullin-Matsuda et al., 2014). At the same time, there is emerging evidence that chronic ethanol exposure can reduce the expression of transcription factor EB (TFEB), which is a master regulator of lysosomal biogenesis and autophagy-associated gene expression (Chao et al., 2018). Given the role that lysosomal dysfunction plays in ethanol-induced liver injury (Chao et al., 2018; Bala and Szabo 2018), and the unexpected accumulation of BMP lipids in Mboat7^HSKO mice, we next investigated TFEB-mediated lysosomal biogenesis and autophagy regulation in Mboat7^HSKO mice challenged with ethanol (Figure 4). First, Mboat7^HSKO mice showed elevated levels of key autophagy regulatory proteins LC3-I/II and p62 in the liver, particularly under ethanol-fed conditions (Figure 4a). Interestingly, Mboat7^HSKO mice also had increased total levels of the mammalian target of rapamycin (mTOR) (Figure 4a), which is a well-known master regulator of autophagic flux. The accumulation of p62 and LC3-I/II cannot distinguish between enhanced or defective autophagic flux, so we next examined potential alterations in lysosome abundance and function. Interestingly, both mRNA and protein levels of lysosomal marker proteins LAMP-1 and LAMP-2 were reduced in ethanol-fed Mboat7^HSKO mice (Figure 4a,4d). Furthermore, compared to control mice fed ethanol, we found that ethanol Mboat7^HSKO mice had generally reduced expressions levels of TFEB target genes and proteins associated with lysosome acidification and lipid turnover including ATPase H+ transporting V1 subunits A,H, and D (Atp6v1a, Atp6v1h, Atp6v1d), a galactosidase A (Gla), chloride channel 7 a (Clcn7), and mucolipin TRP cation channel 1 (Mcolnl) (Figure 4a,4d). Similarly, the mRNA expression, and total and nuclear protein abundance of TFEB was reduced in ethanol-fed Mboat7^HSKO mice compared to ethanol-fed Mboat7^lox/flox control mice (Figure 4a,4b,4d). In contrast, the expression of key autophagy related genes including Atg2b, Atg3, Atg7, Atg8/LC3, and Unc-51-like autophagy activating kinase 1 (Ulk1) were significantly elevated in ethanol-fed Mboat7^HSKO mice compared to ethanol-fed Mboat7^lox/flox control mice (Figure 4d).

We next investigated the cell autonomous effects of ethanol on wild type human Huh7 hepatoma cells or Huh7 cells genetically lacking MBOAT7 (MBOAT7A-Huh7). In agreement with what we found in mouse liver, MBOAT7&-Huh7 cells had reduced levels of total TFEB and lysosomal marker proteins including LAMP1-, LAMP-2, and ATP6V1A, (Figure 4 - figure supplement 1). Furthermore, MBOAT7A-Huh7 cells had increased levels of LC3 and total mTOR, which was particularly apparent upon ethanol exposure (Figure 4 - figure supplement 1). We next assessed lysosome protein degradation activity in wild type and MBOAT7A-Huh7 hepatoma cells by measuring the degradation of an exogenous lysosome indicator, quantified via median Bodipy/Alexa647 fluorescence ratio. In the absence of ethanol treatment, there was no significant difference in the lysosomal activity between wild type and MBOAT7A-Huh7 cells. Lysosome activity was increased by 42% in wild type cells with ethanol treatment vs. no treatment (p<0.001. Figure 4c), in agreement with increased mRNA levels of several lysosomal hydrolases (Figure 4d). However, the lysosome activity was decreased by 45% in MBOAT7A-Huh7 vs. wild type cells treated with ethanol (p<0.001, Figure 4c). Collectively, in the presence of EtOH, deletion of MBOAT7 in mouse or human hepatocytes results in defective TFEB-mediated lysosomal biogenesis and lysosome activity, which would be expected to lead to impaired autophagic flux.

Healthy Control and Heavy Drinking Patient Selection - Healthy controls or heavy drinkers with an AUDIT score greater than >16 were recruited from the Clinical Research Unit at the Cleveland Clinic or MetroHealth Hospital in Cleveland, Ohio based on medical history and physical examination. The study protocol was approved by the Institutional review board for the Protection of Human Subjects in Research at the Cleveland Clinic (IRB 17-718) and MetroHealth Hospital Cleveland (IRB 18-00911). All methods were performed in accordance with the internal review board’s guidelines and regulations, and written, informed consent was obtained from all subjects. Subject demographics are shown in Figure 1 - figure supplement 1.

To generate conditional Mboat7 knockout mice, we obtained “knockout first” (Mboat7^{tm1a(KOMP)Wtsi}) mice from Dr. Philip Hawkins (Anderson et al., 2013), and crossed these mice with mice transgenically expressing FLP recombinase to remove the NEO cassette resulting in a conditional Mboat7 floxed allele. The FLP transgene was then subsequently bred out of the line and resulting Mboat7^lox/WT mice, which were used to expand further downstream tissue-specific knockout lines. To generate congenic hepatocyte-specific (Mboat7^HSKO) and myeloid-specific (Mboat7^MSKO) Mboat7 knockout mice we crossed mice harboring a post-FLP recombinase conditionally-targeted Mboat7 floxed allele (Anderson et al., 2013; Massey et al., 2023) to mice transgenically expressing Cre recombinase under the albumin promoter/enhancer (Posticet al., 1999) or Cre knocked into the M lysozyme locus (Clausen et al., 1999) respectively. These independent Mboat7^HSKO and Mboat7^MSKO lines were then backcrossed mice >10 generations into the C57BL/6J background and subsequently subjected to ethanol exposure. Confirmation of sufficient backcrossing into the C57BL/6J background was confirmed by mouse genome single nucleotide polymorphism (SNP) scanning at the Jackson Laboratory (Bar Harbor, ME). Age- and weight-matched female (8-10 weeks old) control (Mboat7^flox/flox), hepatocyte­specific Mboat7 knockout mice (Mboat7^HSKO), or myeloid-specific Mboat7 knockout mice (Mboat7^MSKO) were maintained on a chow diet and randomized into pair- and ethanol-fed groups using the NIAAA model (Bertola et al., 2013). Briefly, mice were initially fed with control Lieber-DeCarli diet ad libitum for 5 days to acclimatize them to liquid diet. Afterward, ethanol (EtOH)-fed groups were allowed free access to the ethanol Lieber-DeCarli diet containing 5% (vol/vol) ethanol for 10 days, and control groups were pair-fed with the isocaloric substituted maltose dextrins as control diet. At day 11, ethanol-fed and pair-fed mice were gavaged in the early morning with a single dose of ethanol (5 g/kg body weight) or isocaloric maltose dextrin, respectively, and euthanized 6 hours later. (Bertola et al., 2013). All mice were maintained in an Association for the Assessment and Accreditation of Laboratory Animal Care, International-approved animal facility, and all experimental protocols were approved by the Institutional Animal Care and Use Committee of the Cleveland Clinic (IACUC protocols # 2018-2053 and # 00002499).

Hematoxylin and eosin (H&E) staining of paraffin-embedded liver sections was performed as previously described (Gromovsky et al., 2017; Warrier et al., 2015; Brown et al., 2010; Brown et al., 2008a; Brown et al. 2008b). Histopathologic evaluation was scored in a blinded fashion by a board-certified pathologist with expertise in gastrointestinal/liver pathology (Daniela S. Allende - Cleveland Clinic). H&E slides were scanned using a Leica Aperio AT2 Slide Scanner (Leica Microsystems, GmbH, Wetzlar, Germany) and images were processed using ImageScope (Aperio, Software Version 12.1)

Whole tissue homogenates were made from tissues in a modified RIPA buffer as previously described (Warrier et al., 2015; Helsley et al., 2019; Schugar et al., 2017; Lord et al., 2016), and protein was quantified using the bicinchoninic assay (Pierce). Proteins were separated by 4­12% SDS-PAGE, transferred to polyvinylidene difluoride membranes, and then proteins were detected after incubation with specific antibodies as previously described (Warrier et al., 2015; Helsley et al., 2019; Schugar et al., 2017; Lord et al., 2016) and listed in the Key resources table.

Tissue RNA extraction and qPCR analysis was performed as previously described (Helsley et al., 2019). The mRNA expression levels were calculated based on the AA-CT method using cyclophilin A as the housekeeping gene. qPCR was conducted using the Applied Biosystems 7500 Real-Time PCRsystem. All primer sequences are listed in the Key resources table.

To determine the level of hepatic injury in mice fed HFD, plasma was used to analyze alanine aminotransferase (ALT) levels using enzymatic assays as previously described (Helsley et al., 2019). Extraction of liver lipids and quantification of total plasma and hepatic triglycerides was conducted using enzymatic assays as described previously (Helsley et al., 2019; Gromovsky et al., 2017).

Lipids were extracted from liver samples as previously described (Jain et al., 2022). Samples were kept on ice throughout the extraction. Briefly, liver was homogenized in 500uL of 3:1:6 isopropanol:water:ethyl acetate containing internal standard in ceramic bead tubes (Qiagen #13113-50) using the TissueLyzer II (Qiagen #9244420). Samples were centrifuged at 16,000xg for 10min at 4°C and the lipid containing supernatant was transferred to a new 1.5mL tube. Lipid extracts were dried in a SpeedVac (Thermo Savant RVT5105) and resuspended in 150uL of methanol. Samples were kept at 4°C for no more than one week before analysis.

Quantitation of LPI and PI species was performed as previously described (Helsley et al., 2019) Briefly, LPI and PI standards (LPI 16:0, LPI 18:0, LPI 18:1, LPI 20:4, PI 38:4) and the two internal standards (LPI 17:1-d31, PI 34:1-d31) were purchased Avanti Polar Lipids. HPLC grade water, methanol and acetonitrile were purchased from Thermo Fisher Scientific. Standard LPI and PI species at concentrations of 0, 5, 20, 100, 500 and 2000 ng/ml were prepared in 90% methanol containing 2 internal standards at the concentration of 500 ng/ml. Samples were injected into the Shimadzu LCMS-8050 for generating the internal standard calibration curves. A triple quadrupole mass spectrometer (Quantiva, Thermos Fisher Scientific, Waltham, MA, USA) was used for analysis of LPI and PI species. The mass spectrometer was coupled to the outlet of an UHPLC system (Vanquish, Thermos Fisher Scientific, Waltham, MA, USA), including an auto sampler with refrigerated sample compartment and inline vacuum degasser. The HPLC eluent was directly injected into the triple quadrupole mass spectrometer and the analytes were ionized at ESI negative mode. Analytes were quantified using Selected Reaction Monitoring (SRM) and the SRM transitions (m/z) were 571 → 255 for LPI 16:0, 599 → 283 for LPI 18:0, 597 → 281 for LPI 18:1, 619 → 303 for LPI 20:4, 885 → 241 for PI 38:4, 583 → 267 for internal standard LPI 17:1, and 866 → 281 for internal standard PI 34:1-d31. Xcalibur software was used to get the peak area for both the internal standards and LPI and PI species. The internal standard calibration curves were used to calculate the concentration of LPI and PI species in the samples.

Untargeted lipidomics was performed using an Agilent 1290 Infinity II liquid chromatograph equipped with a Waters Acquity BEH C18 column (1.7 pm 2.1 x 100 mm) coupled to an Agilent 6546 Q-TOF mass spectrometer. Mobile phase Awas 60:40 acetonitrile:water and B was 90:9:1 isopropanol:acetonitrile:water with both phases buffered with 10mM ammonium formate and 0.1% formic acid. The gradient was as follows: starting at 15% B to 30% B at 2.40 min, 48% at 3 min, 82% at 13.2 min, 99% at 13.8 min then held at 99% until 15.4 min before equilibrating 15%, held until 20 min. Samples were analyzed in both positive and negative ionization in separate experiments with the following MS parameters: drying gas flowing 12 L/min at 250°C, nebulizer at 30 psi and sheath gas flowing 11 L/min at 300°C for positive mode. The sheath gas flow was 12 L/min at 375°C with a nebulizer pressure of 30 psi in negative mode. Both ionizations had the same voltage for capillary (4000 V), skimmer (75 V), fragmentor (190 V) and octupole (750 V). Reference masses (m/z = 121.05 and 922 for positive mode; 112.98 and 966.00 for negative mode) were continuously infused during sample runs for accurate mass calibration. Pooled samples were run in consecutive iterative MS/MS injections at a constant collision energy of 25V to collect spectra for lipid library creation using Agilent LipidAnnotator and MS1 data was collected for all individual samples. Peak integration was performed using Agilent Profinder (v8.0) software and further curation and internal standard normalization was performed using in­house R scripts as described previously (Jain et al., 2022). Untargeted data is reported in units of pmol lipid/g tissue. Principal component analysis and heatmaps were generated using MetaboAnalyst (Pang et al., 2021). The first and second principal components are plotted on the x- and y-axes, respectively, and sample treatment group is indicated by color. For heatmap generation, data was pareto-scaled and the normalized intensity is indicated by color with relative increase in red and decrease in blue. The top 70 lipid features by ANOVA p-value are plotted on the y-axis and samples are grouped by Ward clustering on the x-axis. Sample treatments are distinguished by color.

Targeted BMP and PG lipid analysis was conducted on an Agilent 1290 Infinity II LC coupled to an Agilent 6495C QQQ mass spectrometer. The same column, mobile phases, and gradient were used for targeted analysis as for the untargeted. The MS parameters for both ionizations were as follows: drying gas temp flow of 12 L/min at 250°C with nebulizer at 35 psi and sheath gas flow of 11 L/min at 300°C. Capillary voltage was kept at 4000 V with nozzle voltage of 500 V and collision energy = 20 V. The iFunnel high pressure RF was at 150/90 V and low-pressure RF at 60/60 V for positive/negative ionizations, respectively. A sample type specific lipid library was created by running pooled liver extract multiple times in positive ionization using multiple reaction monitoring (MRM) to scan for a list of BMP/PG lipid precursor masses ([NH4]⁺ adducts) with transitions for the respective fatty acid and diacylglycerol (minus 17 m/z) adducts (Grabner et al., 2020). BMPs were identified by the characteristic free fatty acid fragment being much higher intensity than the diacylglycerol fragment, whereas the opposite was true for PG. In addition, BMP lipids eluted between 0.3-0.5 min earlier than the isomeric PG counterpart. These distinguishing characteristics were validated using commercial BMP 18:1_18:1 and PG 18:1_18:1 standards. A dynamic MRM method was created using the retention time data gathered from positive ionization tests but in negative mode scanning for the [M-H]⁻ precursors and fatty acid fragments as BMP and PG are much more readily ionized in negative mode. Data was integrated in Agilent Quantitative analysis and peaks were manually adjusted and verified. Final units are in ng lipid/mg tissue.

Lysosome protein degradation activity was performed as previously described (Robinet et al., 2021). Briefly, DQ-ovalbumin (D12053, Thermofisher Scientific) was labeled with Alexa Fluor 647 succinimidyl ester (A20006, Thermofisher Scientific) at room temperature for 1 h (3:1 dye:protein mole ratio) to make lysosome protein degradation indicator. The reaction was stopped by incubating the conjugate with 1.5 M hydroxylamine (pH 8.5) for 1h at room temperature, and the conjugate was purified by extensive dialysis. Lysosome protein degradation indicator was validated in vitro by incubating with proteinase K and achieving significant increase of Bodipy/Alexa647 ratio (Robinet et al., 2021). The double labeled ovalbumin has increased Bodipy fluorescence upon its degradation by decreasing its self­quenching, while Alexa647 is not changed upon its degradation and is used to normalize for cellular uptake. Wild type or MBOAT7A-Huh7 hepatoma cells treated with or without 100 mM ethanol for 48 h were incubated with 10 pg/mL of lysosome indicator for 2 h, and lysosome activity was analyzed by flow cytometry in «10,000 cells with a LSRFortessa device (BD). Flowjo software was used to export data for each cell for ratiometric analyses, and the median Bodipy/Alexa647 ratio for each independent well was used for analysis.

Single comparisons between two groups were performed using two-tailed Student’s t tests with 95% confidence intervals. Comparisons involving multiple groups beyond binary comparison were assessed using One-way ANOVA with Tukey’s post-hoc test. All data presented as mean ± SEM. Values were considered significant at p< 0.05 (using superscripts), or ***p<0.001 and ****p<0.0001 in figure 1. JMP 17.0 statistical discovery software (SAS Institute, Cary, NC, USA) was used for all statistical analyses.