UBIAD1 and CoQ10 protect melanoma cells from lipid peroxidation-mediated cell death

Cutaneous melanoma is the deadliest type of skin cancer, although it accounts for a minority of all skin cancers. Oxidative stress is involved in all stages of melanomagenesis and cutaneous melanoma can sustain a much higher load of Reactive Oxygen Species (ROS) than normal tissues. Melanoma cells exploit specific antioxidant machinery to support redox homeostasis. The enzyme UBIA prenyltransferase domain-containing protein 1 (UBIAD1) is responsible for the biosynthesis of non-mitochondrial CoQ10 and plays an important role as antioxidant enzyme. Whether UBIAD1 is involved in melanoma progression has not been addressed, yet. Here, we provide evidence that UBIAD1 expression is associated with poor overall survival (OS) in human melanoma patients. Furthermore, UBIAD1 and CoQ10 levels are upregulated in melanoma cells with respect to melanocytes. We show that UBIAD1 and plasma membrane CoQ10 sustain melanoma cell survival and proliferation by preventing lipid peroxidation and cell death. Additionally, we show that the NAD(P)H Quinone Dehydrogenase 1 (NQO1), responsible for the 2-electron reduction of CoQ10 on plasma membranes, acts downstream of UBIAD1 to support melanoma survival. By showing that the CoQ10-producing enzyme UBIAD1 counteracts oxidative stress and lipid peroxidation events in cutaneous melanoma, this work may open to new therapeutic investigations based on UBIAD1/CoQ10 loss to cure melanoma.


Introduction
Although accounting for only 5% of all skin cancers, cutaneous melanoma represents the deadliest form of skin cancer with the highest levels of mutational load [1]. Melanoma cells have elevated ROS levels and elevated antioxidant defense compared to melanocytes, in which oxidative stress is already close to a survival threshold due to increased metabolism, melanin biosynthesis and ultraviolet radiation [2]. Melanoma tumors have evolved redox adaptive mechanisms to counteract oxidative stress, such as the activation of the nuclear factor erythroid 2-related factor 2 (NRF2), the upregulation of metabolic survival pathways, including serine biosynthesis and the pentose phosphate pathway, and the elevation of redox metabolites such as glutathione (GSH) and nicotinamide adenine dinucleotide phosphate (NADPH) [3].
The increase of ROS levels by giving oxidant treatments or by removing cellular antioxidant systems could serve as a therapeutic approach to trigger cell death in melanomagenesis and melanoma progression [4,5].
It has been recently suggested that the induction of lipid peroxidation (LP) could serve as a special target in melanomagenesis. Several protective mechanisms against lipid peroxidation in melanoma cells have been described by different studies [6]. For example, oleic acid from lymph, which is incorporated in plasma membrane phospholipids, was found to protect metastatic melanoma cells from lipid peroxidation [7]. Also, a striking upregulation of Sterol Regulatory Element-Binding Protein 2 (SREBP2), a lipogenesis regulator, and a decrease of lipid peroxidation levels have been found in cultured and in circulating single melanoma cells freshly isolated from blood specimens [8]. A similar role in protection against lipid peroxidation in BRAF-inhibitor resistant melanoma was shown for SREBP1, another master regulator of lipogenesis [9]. Another study [10] highlighted the role of aldo-keto reductases in melanoma survival, which convert aldehydes and ketones to their corresponding alcohols, which are in turn able to detoxify lipid peroxides and thus to inhibit cell death execution.
Therefore, increasing evidence indicates that tumor cells have evolved several defense mechanisms to suppress lipid peroxidation in parallel to the well-known defense mechanism mediated by GPx4 [11]. One of these newly discovered strategies is through the enzyme-mediated reduction of the plasma membrane CoQ10 [12]. It has been shown that ferroptosis suppressor protein 1 (FSP1, also called AIFM2) functions as an oxidoreductase to reduce ubiquinone (CoQ10) to ubiquinol (CoQ10H 2 ) mainly on the plasma membrane [11,13] Another very recent study published that dihydroorotate dehydrogenase (DHODH), located in the mitochondrial inner membrane, is important for the reduction of CoQ10 [14]. Along with these two enzymes, also NQO1 could play a significant role in CoQ10 reduction on the plasma membrane of melanoma cells, since it was published that NQO1 is important for melanoma survival [15]. Finally, it is important to mention that most current studies published that lipid peroxidation in cancer cells induces ferroptosis-dependent cell death, a newly described form of iron-dependent, regulated non-apoptotic cell death [16]. Overall, CoQ10 is an essential antioxidant involved in both mitochondrial bioenergetics and plasma membrane protection [17]. Indeed, CoQ10 protects membranes against oxidative damage directly by disrupting the lipid peroxidation chain and through maintaining the plasma membrane redox system [13,[18][19][20].
UBIAD1 is a novel transmembrane enzyme localized in the Golgi apparatus and the endoplasmic reticulum (ER) and it is responsible for the biosynthesis of plasma membrane CoQ10 [21,22]. Here we show for the first time that UBIAD1 plays an antioxidant tumor-promoting role in melanoma cells by supporting plasma membrane CoQ10 synthesis and, thus, contributing to protect membrane components against lipid peroxidation and cell death.
Cells were infected with lentiviruses containing pLKO empty plasmid (named as control) or pLKO plasmid containing shRNA-UBIAD1 or shRNA-NQO1. For most experiments, where not stated otherwise, the two genes were strongly silenced (named as UBIAD1 KD high or simply UBIAD1 KD and NQO1 KD high or simply NQO1 KD ). A lower level of silencing was instead used for co-silencing and dose-dependent experiments (referred to as UBIAD1 KD low and NQO1 KD low ). The relative level of silencing was validated by qRT-PCR (Fig. S4F).

Cell proliferation assay
1.5×10 3 SkMel28 and Mel Juso and 2×10 3 A375 cells/well were plated in 96-well plates in standard media and allowed to adhere 6 h before proceed with lentivirus transduction. Fresh media was replaced after 16 h and after 1 and 3 days of lentiviral infection. Every day, during the next 5 days from transduction, one plate was washed 3 times with 1× PBS, fixed in 3.7% paraformaldehyde for 10 min at room temperature (RT) and incubated in 0.1% of crystal violet solution in deionized water with gentle shaking. The plate was then washed 4 times with tap water and air-dried at RT for at least 24 h. Then 200 μL of 100% of methanol was added to each well and gently shaken on the rotator for 20 min to dissolve the crystal violet. The absorbance of each well was measured at a wavelength of 560 nm with a plate reader (Tecan).

Idebenone rescue experiments
Idebenone rescue was analyzed by growth curve assay of proliferation, by Western blotting assay of proliferation markers (RRM2 and cyclin A) and by lipid peroxidation FACS assay. Idebenone-dependent rescue analyses were performed on 96-well plates. 1.5×10 3 SkMel28 and Mel Juso and 2×10 3 A375 cells/well were plated in standard media and allowed to adhere 6 h. Concentration of idebenone (Tocris) was varied according to cell line: 1 μM idebenone for SkMel28, 10 μM idebenone for A375 and 50 nM idebenone for Mel Juso. For Western blot analysis of idebenone rescue adhered 2×10 5 cells on 6 cm dishes in standard media were treated with 100 nM for SkMel28, 200 nM for A375, 200 nM for Mel Juso of idebenone. Cells were harvested 3 days post-transduction. For FACS (Bodipy C11) assays 2×10 5 cells were treated in 6 cm dishes with 100 nM idebenone for SkMel28, 200 nM idebenone for A375 and 200 nM idebenone for Mel Juso. Cells were harvested 4 days post-transduction. For all treatments idebenone was added to cells at the same time with lentiviral transduction. Media supplemented with idebenone was replaced every 2 days.

Click-iT lipid peroxidation assay
Briefly, 1×10 5 cells were seeded in standard medium on glass coverslips in 24-well plates and let adhere overnight. Lentiviral-mediated UBIAD1 KD was performed on attached cells. Lipid peroxidation assay was performed according to the manufacturer's protocol of Click-iT staining (Thermo Fisher). DAPI staining for DNA visualization (300 nM) was done for 1 min in the darkness. Fluorescence was observed by Leica SP8 DLS microscopy using 40x magnification. Images were analyzed with Image J (National Institutes of Health).

Bodipy C11 lipid peroxidation assay
Lipid peroxidation was evaluated by flow cytometry using Bodipy C11 581/591 (Thermo Fisher). 1×10 5 cells for ctrl and 2×10 5 cells for UBIAD1 KD and NQO1 KD were seeded in standard medium in a 6 cm dish and let adhere for 6 h. UBIAD1 or NQO1 knockdown was induced by lentiviral transduction as previously described. After 4 days of infection cells were incubated for 15 min at 37 • C in the dark with 2.5 μM Bodipy C11 in 1× PBS supplemented with 5% FBS. As positive control, cells were treated with 100 μM of cumene hydroperoxide (CH) for 30 min.
Floating cells were collected and combined with adherent cells detached by trypsinization (Trypsin #ECB3052 EuroClone). Cellular pellets were washed 3 times with 1× PBS at 1000 g for 5 min at 4 • C. Eventually, 2×10 5 cells were resuspended in 200 μl of 1× PBS + 5% FBS and pipetted up and down to obtain a single-cell suspension. Flow cytometry analysis was performed using BD FACSCanto™ II Cell Analyzer (BD Biosciences). Non-treated cells without Bodipy C11 staining were used as blank. The ratio between oxidized (FITC-A) and reduced (PE-A) Bodipy C11 was measured. A minimum of 20,000 events were obtained per sample.

ROS analyses by DHE and DCFH-DA assay
Intracellular ROS levels were measured by flow cytometry after cells were stained with dihydroethidium (DHE) (#D23107, Thermo Fisher) (for superoxide) and dichlorodihydrofluorescein diacetate (DCFH-DA) (#C6827, Thermo Fisher) (for H 2 O 2 ). 1×10 5 cells for ctrl and 2×10 5 cells for UBIAD1 KD and NQO1 KD were seeded in standard media in 6 cm dishes and let adhere for 6 h. UBIAD1 or NQO1 knockdown was induced by lentiviral transduction as previously described. After 4 days of silencing cells were stained with DHE or DCFH-DA, according to manufacturer's recommendations: floating and attached cells were harvested by trypsin (Trypsin #ECB3052 Euro-Clone), rinsed with 1× cold-PBS, and then probed with DHE (10 μM) and DCFH-DA (300 nM) in 1× PBS+5% FBS for 10 min at 37 • C in the dark. Together with DHE or DCFH-DA staining, cells were incubated with

Apoptotic DNA fragmentation TUNEL assay
Apoptotic cells were detected by APO-BrdU™ TUNEL Assay Kit, with Alexa Fluor™ 488 Anti-BrdU (#A23210, ThermoFisher). 1×10 5 cells for ctrl and 2×10 5 cells for UBIAD1 KD were seeded in standard medium in 6 cm dishes and let adhere for 6 h. UBIAD1 knockdown was induced by lentiviral transduction as previously described. After 4 days of silencing cells were stained according to manufacturer's recommendations: floating cells were collected and combined with adherent cells detached by trypsinization. Cells were washed once in 1× PBS and fixed for 15 min on ice with 1% (w/v) PFA in 1× PBS. Then, cells were washed twice with 1× PBS, resuspended in ice-cold 70% (v/v) ethanol and incubated at − 20 • C for 2 h. After ethanol removal, samples were washed twice with wash buffer (blue cap) and resuspended in DNA-labeling solution.
To obtain the best staining, ¼ of the suggested reagent concentrations were used and incubation time was decreased to 15 min at 37 • C. At the end of incubation time, cells were washed twice with rinse buffer (red cap) and incubated 30 min at room temperature protected from light with Alexa Fluor 488 dye-labeled anti-BrdU antibody (to obtain the best staining, ¼ of the suggested antibody concentration was used). Then, cells were washed once in 1× PBS and resuspended in 500 μl of 1× PBS.
All centrifugations were performed for 5 min at 4 • C at 1000 g. Samples were analyzed within 3 h of completing the staining procedure using BD FACSCanto™ II Cell Analyzer (BD Biosciences) using FITC-A filter. A minimum of 50,000 events were collected per sample. As positive control, ctrl cells were treated with staurosporine (1 μM) for 16 h before proceeding with the staining procedure. Not stained control cells were used as blank.

Subcellular fractionation for plasma membrane CoQ10 measurements
2×10 5 cells for ctrl and 1×10 6 cells for UBIAD1 KD low were seeded in 10 cm dishes and let adhere for 6 h. UBIAD1 knockdown was induced by lentiviral transduction as previously described. After 5 days, cells were collected and washed 3 times with ice-cold 1× PBS by centrifugation at 4 • C at 600 g for 10 min. All the following steps were performed on ice.
Pellets were resuspended using 500 μl of ice-cold 1× PBS containing 0.1 M Tris-HCl pH 7.4, 0.1 M EGTA, 1 M sucrose and protease/phosphatase inhibitors (Complete Mini, Roche) and incubated 30 min on ice. Ten μl were saved, resuspended in RIPA buffer (#89900, Thermo Fisher) as Whole Cell fraction (W). Cells were homogenized using a pre-chilled potter (#RTCXH1.1, ROTH) till 80% of cells were lysed (approximately 40 strokes for A375 and 45 strokes for SkMel28). Percentage of cell lysis was checked using Trypan Blue staining. Cell lysates were centrifuged twice at 600 g for 5 min at 4 • C to remove unbroken cells and nuclei (pellets). Supernatants were centrifuged at 8,000 g for 20 min at 4 • C to obtain a pellet formed by crude mitochondria and a supernatant enriched in cytosolic organelles and plasma membrane. The pellet was then resuspended in 2 mM MgCl 2 -containing 1× PBS and centrifuged at 3,000 g for 15 min at 4 • C to pellet more pure mitochondria. Pellets were resuspended in RIPA buffer to form the Mitochondria fraction (M). The supernatants containing the cytosolic organelles and plasma membranes were centrifuged at 21,000 g for 90 min at 4 • C. Pellets containing plasma membrane and plasma membrane associates were resuspended in ice-cold PBS containing 0.1 M Na 2 CO 3 and 1 mM EDTA (pH 11.3) for 30 min on ice. Final centrifugation was performed at 21,000 g for 99 min at 4 • C. Plasma membrane pellets were resuspended in 30 μl of PBS forming the Plasma membrane fraction (PL). One third of the volume was saved and resuspended with RIPA buffer for Western blot analysis, while the remaining 2/3 were frozen in liquid nitrogen and stored at − 80 • C for mass spectrometry analysis.

UHPLC-MC/MS analyses of CoQ10 level
Cells were seeded in 10 cm plates in DMEM supplemented medium and let to reach 80% of confluence. Cells were washed with ice-cold 1× Results are presented as ng of CoQ10 per mg of proteins. The UHPLC/MS analysis was done on a Hybrid quadrupole-Orbitrap system (Q Exactive, Thermo Scientific) coupled to an UHPLC system (Ultimate 3000, Thermo Dionex) via a heated electrospray ionization source. Four μl of each sample was injected in the Accucore C18 100 × 2.1 (2.6 μm particle size) column. (Thermo Fischer Scientific) The UHPLC-/MS was operated at a flow rate of 250 μl/min with a gradient elution of 10 mM of ammonium formate in water (phase A) and 10 mM of ammonium formate in methanol/2-propanol 80:20 (phase B). Data were acquired in full MS/ddMS2mode with positive electrospray ionization mode. Similar procedure was used to analyze CoQ10 level in plasma membrane fraction. Ceramide 16:0 (Cer16:0) was used as internal control to normalize the CoQ10 amount among samples.

Lipid class quantification by Gas-Chromatography(GC) and GC-MS analyses
The following internal standards were used: trinonanoylglycerol and tripentadecanoylglycerol for triacylglycerols (TG), 5-α-cholestane for cholesteryl esters (CE), (3α, 5β)-cholestan-3-ol (epicoprostanol) for free cholesterol (CO). Lipids were extracted from 100 μl of sample using a chloroform-methanol solution in accordance with the Folch method (Folch J 1957). Extracted lipids were separated to classes by thin layer chromatography as previously reported [23]. The TG fraction was hydrolyzed with 2 ml of HCl-methanol (5%) (Merck). The resulting free fatty acids were trans-esterified and extracted with hexane. Hexane containing fatty acid methyl esters (FAME) was directly used for GC analysis on Agilent 5890 gas chromatograph (GC) equipped with a flame-ionization detector. FAMEs derived from TG were resolved with an Omegawax column (30 m × 0.25 mm internal diameter x 0.25 μm film thickness; Supelco) with injection of 1 μl running in on-column mode. The program for oven temperature was set up next: 60 • C for 3 min, increased 20 • C/min to 205 • C, then remained constant for 15 min. Temperature then increased 0.4 • C/min up to 213 • C, which was maintained for 10 min and finally increased to 240 • C at 5.0 • C/min and held for 8 min. Peaks were determined in relation to a reference standard mixture (GLC 461, Nuchek Prep).
CO and CE were eluted from the silica with 5 ml of mixture of chloroform-methanol 2:1. The solvent was collected and evaporated to dryness under a stream of N 2 . CE have been hydrolyzed by saponification and obtained free CO were extracted with hexane. In the end, both dried residues containing CO were derivatized in 50 μl of N,O-Bis

UBIAD1 expression in human cutaneous melanoma patient cohort
Publicly available RNA expression levels and associated clinicopathological and survival data of the skin cutaneous melanoma patients enrolled in the TCGA-SKCM cohort [24] were retrieved from cBioPortal [25,26]. Out of 470 patients, 10 were excluded from the analysis because of missing "Overall Survival" information or RNA expression data. Three patients had more than one biological sample available, the RNA expression profile from the sample with the lowest "SAMPLE_ID" was retained since it matched the "primary tumor" sample when both primary and metastatic samples were sequenced. Patients were then stratified according to UBIAD1 RNA expression levels in two groups: High, UBIAD1 RNA expression level above or equal to the median; Low, UBIAD1 RNA expression level below the median. The hazard ratio was calculated with a 95% confidence interval (95% CI) using Cox proportional Hazard regression both in univariable and in multivariable models to adjust for all the standard clinicopathological parameters by the "coxph" function of the "survival" package [27] in R [28]. The proportionality assumption of the hazards over time was tested with the Schoenfeld test as implemented in the "cox.zph" function of the "survival" package; none of the variables considered in the model violated the proportionality assumption. The forest plot was generated with the "forestmodel" package [29].

Statistical analyses
GraphPad Prism 8 was used for statistical analysis. For 2 experimental comparisons, 2-tailed unpaired Student's t-test was used. For multiple comparisons, 1-and 2-way ANOVA and 1-sample t-test were applied. Specified number of biological replicates are defined in the legends. A p-value of less than 0.05 was considered significant. Statistical significance is reported as exact p-value or ns, when not significant.

High UBIAD1 expression is associated with poor survival (OS) of melanoma patients and with melanoma cell lines
The role of the UBIAD1 metabolic enzyme in cancer progression has been poorly investigated. To define the role of UBIAD1 in melanoma progression and in the biology of real-life melanoma tumors, we evaluated the prognostic potential of its transcriptional levels in the TCGA Skin Cutaneous Melanoma (SKCM) cohort, which represents until now the largest collection of clinico-pathological information of SKCM patients with associated publicly available transcriptomics data (Fig. 1A). We observed a statistically different probability of overall survival (OS) in patients dichotomized at the median expression level of UBIAD1, with the group with the higher expression demonstrating a worse prognosis as compared to the group with lower expression (HR = 1.39; 95% CI = 1.06-1.82; p-value = 0.016). The poor prognostic behavior of the UBIAD1 high group persisted even in multivariable analysis (HR = 1.37; 95% CI = 1.04-1.8; p-value = 0.024) after correcting for standard prognostic factors such as tumor stage, age, and sex of the patient (Fig. 1B), thus defining the expression level of UBIAD1 as an independent prognostic factor.
Next, we investigated UBIAD1 expression across multiple human melanoma cell lines both at protein and mRNA levels. As control we analyzed normal human melanocytes (HEMa-LP and NHEM) and immortalized epidermal cells (HaCaT). We observed a significant expression of UBIAD1 protein in most of the melanoma cells tested, with BRAF-mutated cell lines (except for RPMI-7951) showing the highest amount of UBIAD1 ( Fig. 1C and Fig. S1A). UBIAD1 mRNA levels fluctuate among melanoma cells with the MM165, A375 and SkMel28 showing the highest levels (Fig. 1D). Intriguingly, the highest expression levels of UBIAD1 both at mRNA and protein levels were detected in SkMel28 and A375. UBIAD1 enzyme is responsible for the biosynthesis of the nonmitochondrial pool of CoQ10 and it plays an important role as an antioxidant enzyme [22]. Using high resolution mass-spectrometry we measured the levels of CoQ10 in melanocytes and melanoma cells (Fig. 1E). Compared to melanocytes, CoQ10 level is elevated in all melanoma cell lines except for SkMel24. Since CoQ10 could be also synthesized by COQ2, the mitochondrial homologue of UBIAD1, we analyzed mRNA expression of COQ2 in these cells (Fig. S1B). Overall, we observed that COQ2 mRNA level is similarly expressed by melanocytes and other melanoma cells, except for IPC-298, SkMel3 and A375 where COQ2 mRNA is significantly upregulated. Looking at the discrepancy between COQ2 and UBIAD1 mRNA levels in different cell lines, we conclude that there is no compensatory nor synergistic transcriptional regulation of CoQ10 synthesizing enzymes.
UBIAD1 can be localized both in the ER and the Golgi apparatus as previously demonstrated [22,30]. Such different localization has been associated with different functions of UBIAD1: in the ER it binds and stabilizes 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGCR), a key rate-limiting enzyme involved in cholesterol synthesis [30], while in Golgi it participates in CoQ10 synthesis [22]. We thus examined the subcellular localization of UBIAD1 in melanoma cell lines SkMel28 and A375 and in normal melanocytes HEMa-LP (Fig. 1F). Immunofluorescence data showed that UBIAD1 is mainly co-localized to the Golgi in melanoma cell lines compared to healthy melanocytes where it is mainly present in ER.
Overall, we conclude that UBIAD1 and CoQ10 are significantly expressed during melanoma progression as well as in melanoma cells carrying both BRAF and NRAS mutations. These results open the question whether UBIAD1 and CoQ10 might represent an important part of the redox systems which characterize melanoma progression and melanoma cells [5].
(caption on next page) L. Arslanbaeva et al.

UBIAD1 loss impairs melanoma cell proliferation and survival by reducing plasma membrane CoQ10
To understand the function of UBIAD1 in melanoma, we focused our studies on SkMel28 and A375 cell lines, carrying the BRAF V600E -mutation, and the Mel Juso cell line, carrying the NRAS Q61L -mutation. We then knocked-down UBIAD1 (UBIAD1 KD ) in these cell lines and performed a growth curve assay to determine whether UBIAD1 loss could affect melanoma proliferation. UBIAD1 KD resulted in a significant reduction of cell viability in these cell lines with strongest effects on SkMel28 and A375 cell lines ( Fig. 2A). To confirm these data, we analyzed the level of two markers of cell proliferation: cyclin A and ribonucleotide reductase subunit M2 (RRM2) a cell-cycle-regulated enzyme, that catalyzes the rate-limiting step in the de novo synthesis of DNA precursors [31] (Fig. 2B). Both cyclin A and RRM2 levels dropped in UBIAD1 KD conditions, confirming proliferation defects in UBIAD1 KD cells.
To further understand the effect of UBIAD1 KD in melanoma cells we examined a series of signaling pathways which play key roles in protein synthesis, survival, proliferation and metabolism (Figs. S2A-F) [32][33][34][35]. Compared to control cells, UBIAD1 KD affects phosphorylation of AKT (Ser473) protein, S6 signaling and AMPK in SkMel28, while UBIAD1 KD affect significantly only AKT and mTOR/s6 signaling in A375. Compared to control cells, UBIAD1 KD decreased the protein level of AKT and ERK1/2 in SkMel28 and Mel Juso, but not in A375. Hence, we observed different responses in all three cell lines. This experimental evidence raises the question of the involvement of different signaling pathways in each cell line and remains for now inconclusive.
UBIAD1 exerts an antioxidant function in cardiovascular tissues [22]. Thus, to better characterize the mechanism involved in UBIAD1-mediated survival of melanoma cells, we analyzed ROS levels in UBIAD1 KD cells (Fig. 2C). Our results showed that UBIAD1 loss increased ROS levels evaluated as DHE and DCFH-DA staining in both SkMel28 and A375 cells in a dose-dependent manner. Interestingly we noticed that non-viable UBIAD1 KD low and UBIAD1 KD were mainly positive to DHE staining that detects cytosolic superoxide, ONOO and •OH (Fig. S2G), while viable UBIAD1 KD low and UBIAD1 KD cells were mainly positive to DCFH-DA that detects hydrogen peroxide, hydroxyl radical, carbonate radical, and nitrogen dioxide (Fig. S2H) [36]. This suggests a temporal cascade in the generation of different reactive redox species that upon UBIAD1 knockdown leads to cell death.
Our group previously demonstrated that, being co-localized in the Golgi compartment, UBIAD1 is responsible for the biosynthesis of nonmitochondrial/plasma membrane CoQ10 [22]. Since our co-localization studies showed that UBIAD1 is mostly present in Golgi apparatus in BRAF-mutated melanoma cells, we examined CoQ10 levels after UBIAD1 KD in SkMel28 and A375 cells and we found them to be significantly reduced, as expected (Fig. 2D). To support the functional role of the non-mitochondrial fraction of CoQ10 after UBIAD1 KD , we isolated plasma membrane fractions (PL) from melanoma cells and measured CoQ10 levels (Fig. 2E). Here we detected a significant and important drop of CoQ10 in UBIAD1 KD cells compared to controls. Since UBIAD1 has been also associated with the regulation of cholesterol synthesis due to its putative ability to bind and stabilize the cholesterol biosynthetic enzyme HMGCR [37,38], we also measured free cholesterol (CO), cholesterol esters (CE) and triacylglycerols (TGs) (Fig. S2I). Surprisingly, UBIAD1 KD did not alter TGs, CE nor CO levels in these melanoma cells.
To confirm that the loss of UBIAD1-dependent CoQ10 synthesis is responsible for cell viability defects, we performed a rescue-viability assay by treating UBIAD1 KD cells with idebenone, a synthetic analog of CoQ10 with increased solubility, to emulate the same physiological mechanism of reduced CoQ10 [13]. Treatment with idebenone resulted in a significant rescue of the viability of SkMel28, A375 and Mel Juso after a mild KD of UBIAD1 (UBIAD1 KD Low ) (Fig. 2F). To further support that the CoQ10 analog, idebenone, is able to rescue or preserve cell viability and thus proliferation, we analyzed the level of two markers of cell proliferation, cyclin A and RRM2, as well as a marker of DNA damage (e.g. double-strand DNA breaks) as pH2AX (Fig. 2G). We found that idebenone treatment in UBIAD1 KD cells can significantly restore expression of these markers in all melanoma cell lines tested (except for RRM2 in Mel Juso), supporting the functional role of CoQ10 in UBIAD1-dependent melanoma survival.
Altogether, these data demonstrate that UBIAD1 is required to sustain survival and that it exerts this effect through the synthesis of CoQ10 in melanoma cells.

UBIAD1-mediated CoQ10 synthesis protects melanoma cells from lipid peroxidation
Reduced CoQ10 acts as a lipophilic radical-trapping antioxidant which detoxifies lipid peroxyl radicals [39]. To explore whether UBIAD1 deficiency induces lipid peroxidation in melanoma cells, we performed lipid peroxidation analyses using Bodipy C11 581/591 on SkMel28, A375 and Mel Juso cells after UBIAD1 KD (Fig. 3A). As positive control for lipid peroxidation we used cumene hydroperoxide (CH), since it can initiate and propagate lipid peroxidation [40]. UBIAD1 KD promoted significant lipid peroxidation in all melanoma cell lines. We also showed lipid peroxidation defects exerted by UBIAD1 loss in attached and living cells using Click-IT technology (Fig. 3B). Remarkably, UBIAD1 loss promotes lipid peroxidation in all tested melanoma cells lines. To demonstrate that the drop of CoQ10 that follows UBIAD1 KD is responsible for the induction of lipid peroxidation and thus of cell death, we performed rescue experiments by treating melanoma cells with idebenone (Fig. 3A-B). We observed that idebenone treatments were sufficient to rescue UBIAD1 KD -mediated lipid peroxidation as evaluated by Bodipy C11 and Click-iT lipid peroxidation assays.
Recent data showed that CoQ10 depletion leads to ferroptotic cell UBIAD1 level was normalized to β-actin and reported as relative to HEMa-LP (set at 1). One-sample t-test (hypothetical mean = 1) was used to quantify statistical significance. Error bars represent SEM, n ≥ 3. NRAS and BRAF mutation status was reported under each melanoma cell line: wt, wild type; het, heterozygous and hom, homozygous.
(D) qRT-PCR quantifications of UBIAD1 mRNA levels in a panel of cell lines. UBIAD1 mRNA level was normalized to β-actin and reported as relative to HEMa-LP (set at 1). One-sample t-test (hypothetical mean = 1) was used to quantify statistical significance. Error bars represent SEM, n ≥ 3. NRAS and BRAF mutation status was reported under each melanoma line: wt, wild type; het, heterozygous and hom, homozygous.
(E) HPLC-MS analyses of CoQ10 in different melanoma cell lines, normalized to total protein (TP) concentration. 1-w ay ANOVA with Dunnett's multiple comparisons test (HEMa-LP as a control) was used to quantify statistical significance. Error bars represent SEM, n = 6.
(F) Subcellular co-localization of UBIAD1. Confocal images showed prevalent UBIAD1 co-localization with ER marker calreticulin in HEMa-LP melanocytes and prevalent UBIAD1 co-localization with Golgi marker GM130 in melanoma cell lines SkMel28 and A375. Image scale bars = 15 μm.
death through the induction of lipid peroxidation [13]. We thus investigated whether UBIAD1 KD leads to ferroptosis. First, we performed AnnexinV/Propidium Iodide (PI) cell death staining. Analysis of AnnexinV/PI staining by flow cytometry allows to distinguish between "early apoptotic cells", which are AnnexinV-positive and PI-negative (AnnexinV + /PI − ) and "late apoptotic cells", which are instead AnnexinV + /PI + . Compared to control cells, UBIAD1 KD led to a significant increase of AnnexinV + /PI − signal in A375, but not in SkMel28 (Fig. 3C). On the contrary, both melanoma cell lines showed a highly significant increase of AnnexinV + /PI + staining after UBIAD1 KD . We also observed that in healthy melanocytes UBIAD1 KD did not affect Annex-inV + cells, but only slightly affected AnnexinV + /PI + cells. These data indicate that UBIAD1 KD can induce cell death in all cell types tested but with a stronger effect in melanoma cells compared to normal melanocytes.
To unequivocally confirm that UBIAD1 loss promotes apoptotic cell death, UBIAD1 KD melanoma cells were tested by TUNEL assay, which measures the level of nucleosomal DNA-fragmentation, a strong hallmark of apoptosis (Fig. 3D). SkMel28 and A375 melanoma cells were TUNEL-positive after UBIAD1 KD with UBIAD1 loss leading to apoptotic cell death in a dose-dependent manner.
These results suggest that by lowering the amount of plasma membrane CoQ10, UBIAD1 KD promotes the initiation and propagation of lipid peroxidation in melanoma cells and their subsequent apoptotic (but not ferroptotic) cell death.

NQO1 suppresses lipid peroxidation by regenerating the antioxidant form of CoQ10
Our data suggested a fundamental role of plasma membrane CoQ10 as an antioxidant preventing melanoma cell lines from lipid damage. To further investigate the mechanism of CoQ10-mediated lipid protection, we focused our attention on the role of NQO1 and FSP1, two plasma membrane NADH-dependent oxidoreductases. NQO1 possesses NAD(P) H:ubiquinone oxidoreductase activity and functions as a component of the plasma membrane redox system generating the reduced forms of CoQ10 (CoQ10H 2 or ubiquinol) [41]. It has been shown that NQO1 plays a key role in melanomagenesis and is highly expressed in melanoma [42,43]. FSP1 has been recently identified as an NADH-dependent oxidoreductase localized to the plasma membrane where it mediates the reduction of CoQ10 [13]. We thus sought to determine which plasma membrane ubiquinone reductase regenerates CoQ10H 2 and functions as a radical-trapping antioxidant suppressing the propagation of lipid peroxides in melanoma cells. For this purpose, we evaluated FSP1 and NQO1 protein and mRNA levels in a panel of melanoma cell lines ( Fig. 4A-C). FSP1 protein level was found to be significantly and similarly upregulated in all melanoma cell lines, highlighting its important role in melanoma survival. On the other hand, NQO1 protein level varied with significant upregulation only in RPMI-7951, SkMel24 and SkMel28 cell lines. We also compared FSP1 and NQO1 expression upon treatment with known inducers of different types of cell death: staurosporine (apoptosis), CH (lipid peroxidation) and RSL3, FIN56 and Erastin (ferroptosis) (Figs. S4A-B). UBIAD1 levels were also used as control (Fig. S4C). Interestingly, while FSP1 and UBIAD1 levels did not seem to be altered by these treatments, NQO1 expression was significantly stimulated by ferroptosis stimuli in SkMel28. These data suggested that NQO1 is differently regulated by different cell death conditions (apoptosis, ferroptosis or lipid peroxidation) making it an interesting target for further analyses.
To study the function of NQO1 in melanoma cells, we examined ROS and lipid peroxidation levels in SKMel28, A375 and Mel Juso cell after NQO1 KD . NQO1 loss promotes increase of ROS levels similarly to UBIAD1 KD (Fig. S4D). NQO1 loss promotes also lipid peroxidation as evaluated by Bodipy C11 staining (Fig. 4D). We then performed a rescue experiment with idebenone and we found that idebenone treatment was indeed able to rescue lipid peroxidation after NQO1 KD in melanoma cells (Fig. 4E). This evidence further indicates that NQO1 enzyme counteracts lipid peroxidation and ROS generation in melanoma cells.
To further explore the role of NQO1 in melanoma cells-survival, we knock-downed NQO1 in SkMel28 and A375 melanoma cells and performed cell proliferation assays. NQO1 KD resulted in a significant reduction of cell viability of both SkMel28 and A375 cell lines (Fig. 4F). Also in this case, both cyclin A and RRM2 levels dropped in NQO1 KD conditions, confirming a block in cell proliferation (Fig. S4E). To demonstrate the existence of a UBIAD1/CoQ10/NQO1 axis that regulates and maintains cell survival in ROS-dependent melanoma cell lines, we sought to understand whether the two enzymes could have a synergistic effect. For this purpose, we carefully titered UBIAD1 and NQO1 knockdown in SkMel28 and A375 melanoma cell lines at suboptimal level (UBIAD1 KD Low and NQO1 KD Low ) (Fig. S4F) to avoid complete cell death as for UBIAD1 KD and NQO1 KD ( Fig. 2A-B and Fig.4F). Next, we compared the effects of UBIAD1 and NQO1 silencing alone or in combination regarding cell viability and cell death (Fig. 4G). We found that a low level of UBIAD1 knockdown (UBIAD1 KD Low ) or NQO1 (NQO1 KD Low ) had less extreme effects on cell survival than those caused by a stronger KD (UBIAD1 KD or NQO1 KD ). However, the concomitant mild silencing of both UBIAD1 and NQO1 enzymes had a dramatic effect on cell survival (UBIAD1 KD Low + NQO1 KD Low ), suggesting a synergistic effect of the two enzymes in protecting melanoma cells from cell death.
Last, we evaluated the changes in mRNA expression of UBIAD1, (caption on next page) L. Arslanbaeva et al. NQO1 and FSP1 upon UBIAD1 KD or NQO1 KD in melanoma cells (Fig. 4H). Upon UBIAD1 KD , both SkMel28 and A375 lines showed a slight downregulation of NQO1. Differently, mRNA level of FSP1 did not change in SkMel28 upon UBIAD1 KD , but significantly increased in A375. This would probably mean that FSP1 could be engaged instead of NQO1 as CoQ10-reducing enzyme in this cell line, also considering the low level of NQO1 in A375 cells (Fig. 4B). NQO1 KD , on the other hand, did not induce any changes neither in UBIAD1, nor in FSP1 in both cell lines. We conclude that NQO1 is located downstream of UBIAD1 and cooperates with it to protect BRAF-mutated melanoma cells from lipid peroxidation and cell death.
In summary, we suggest that NQO1 protein is responsible for CoQ10 redox regeneration downstream of the UBIAD1/CoQ10 axis in melanoma cell lines. When melanoma cells experience UBIAD1/CoQ10/ NQO1 loss, ROS levels increase and promote lipid peroxidation as well as apoptotic cell death (Fig. 5A). The effect of UBIAD1/CoQ10 and NQO1 in survival of melanoma cell lines in vitro suggests to further consider these two enzymes as new therapeutic targets in melanoma research.

Discussion
Accumulating evidence indicates that melanoma cells overcome oxidative stress by developing different strategies [1,7]. The purpose of this study is to demonstrate that UBIAD1 is essential for melanoma survival by providing antioxidant protection through CoQ10 synthesis. UBIAD1 is a trans-membrane protein, localized in different cellular compartments and responsible for the biosynthesis of non-mitochondrial CoQ10 [22]. It was reported that UBIAD1 is involved in a variety of human diseases [44][45][46]. It also plays a tumor-suppressing role in bladder cancer [47,48]. Concerning melanoma, high UBIAD1 mRNA level is associated with poor prognosis (OS) in melanoma patients based on TCGA-Skin Cutaneous Melanoma datasets. To determine the functional role of UBIAD1 in melanoma progression we first tested mRNA and protein levels of UBIAD1 in a panel of melanoma cell lines and observed that UBIAD1 is significantly upregulated in melanoma cells with respect to melanocytes. By immunofluorescence we showed that UBIAD1 is co-localized mainly to Golgi and to a lesser extent to ER in melanoma cells with respect to melanocytes, consistently with previous data suggesting that the biosynthesis of non-mitochondrial CoQ10 by UBIAD1 is localized to Golgi [22,49]. We selected some BRAF-and NRAS-mutated cell lines such as SkMel28, A375 and Mel Juso to study the role of UBIAD1 in the antioxidant defense of melanoma.
In line with previous studies [22,50,51], we then showed that UBIAD1 KD -dependent depletion of plasma membrane CoQ10 blocks cell proliferation and triggers ROS increase, lipid peroxidation and apoptotic cell death in melanoma cells. Importantly, redox lipid balance, DNA damage and viability of UBIAD1 KD cells were rescued by restoring the antioxidant potential of plasma membranes through idebenone treatment.
Accumulating evidence indicated that UBIAD1 is tightly linked to cholesterol metabolism as some studies show that UBIAD1 suppresses intracellular cholesterol metabolism [45] and other studies report that UBIAD1 inhibits HMGCR degradation by direct interaction, thus elevating the cholesterol level [30]. Our mass-spectrometry analysis of different lipid metabolites did not display any significant difference between control and UBIAD1 KD samples regarding cholesterol and TAG, except a slight increase of cholesterol esters in A375 cell line upon UBIAD1 KD . While it was reported that UBIAD1 is responsible for the biosynthesis of vitamin K2 [52], we did not detect vitamin K2 in melanoma cells by HPLC measurement (data not shown), confirming that UBIAD1 is not required for generating Vitamin K2 in melanoma cells. Together, our results suggest that UBIAD1 serves as an antioxidant defense through the biosynthesis of non-mitochondrial CoQ10, but not by the regulation of cholesterol metabolism.
We then explored the dependency of melanoma cells on CoQ10 downstream of its synthesis. CoQ10 can be reduced by different enzymes in cancer cells, for example FSP1 in the plasma membrane [13] and DHODH in the mitochondrial inner membrane [14]. We hypothesized that in melanoma cells also NQO1, transcriptional target of NRF2, could be responsible for the regeneration of the reduced form of CoQ10, since NQO1 is part of a plasma membrane redox system and elevated levels of NQO1 are associated with poor melanoma patient outcome [15,41,53]. According to our hypothesis, we found that NQO1 KD induced ROS increase and lipid peroxidation and that NQO1 is required for survival in melanoma cells. Such defects resemble UBIAD1 KD conditions. Finally, we discovered that the simultaneous but mild knockdown of UBIAD1 and NQO1 have a synergistic effect with cell death levels comparable with the full KD of only one enzyme.
Recently, it was proposed a new model for antioxidant enzymes involved in defense against lipid peroxidation in cancer: GPx4 in the cytosol and mitochondria, FSP1 on the plasma membrane, and DHODH in mitochondria [14]. However, we suggest that the mechanism of antioxidant protection is more complicated and composed of more different antioxidant enzymes. The contribution of each enzyme is "context-dependent" on the chosen model of cancer, as the regulation of each enzyme is dependent on the ROS-modulators. Another important point is that cancer cells maintain high levels of ROS, very close to a death threshold. Thus, the impairment of even only one antioxidant pathway can trigger cell death. This view is supported by our findings that UBIAD1/CoQ10/NQO1 axis disruption is sufficient to trigger cell death in melanoma cell lines despite the presence of other antioxidant enzymes. However, it still remains to be further explored the functional cross-talk between UBIAD1 and NQO1 in melanoma progression and whether the blocking of these enzymes would represent a valid therapeutic approach to treat melanoma.

Conclusion
In this study, we showed that UBIAD1, a transmembrane enzyme localized in the Golgi apparatus, plays a key role in suppressing lipid peroxidation and promoting melanoma survival through CoQ10 synthesis and NQO1-dependent plasma membrane redox regulation. We envision UBIAD1 and NQO1 blockade as novel therapeutic strategies in melanoma. In particular, it could be interesting to study UBIAD1/ CoQ10/NQO1 axis in those situations where oxidative stress drives (A) Lipid peroxidation assay using Bodipy C11 581/591 FACS upon UBIAD1 KD in presence or absence of idebenone treatment (100 nM of idebenone for SkMel28, 200 nM of idebenone for A375 and 200 nM of idebenone for Mel Juso). Ratio between oxidized and reduced Bodipy C11 was measured and reported as fold change over control conditions (ctrl). For positive control, cells were treated for 30 min with 100 μM of cumene hydroxyperoxide (CH). One-way ANOVA with Tukey's multiple comparisons test was used to quantify statistical significance. Error bars represent SEM, n ≥ 3. (B) Quantification of lipid peroxidation in melanoma cells, induced upon UBIAD1 KD , assessed by Click-iT method. Statistical significance was quantified by 1-way ANOVA using Sidak's multiple comparisons test. Error bars represent SEM, n = 3. (C) Relative viability after UBIAD1 KD in melanocytes (HEMa-LP) and melanoma cells assessed by AnnexinV/PI flow cytometry after lentiviral transduction. Alive cells are defined as events negative for both PI and AnnexinV. Error bars represent SEM, n = 3. (D) Analysis of apoptosis by FACS TUNEL assay in melanoma lines SkMel28 and A375 upon UBIAD1 KD Low and UBIAD1 KD . Data are shown as % of apoptotic cells (TUNEL-positive) over singlets. As positive control, cells were treated with 1 μM of staurosporine for 16 h. Unpaired Student two-tailed t-test was used to quantify statistical significance between ctrl and UBIAD1 KD Low or UBIAD1 KD samples. Error bars represent SEM, n = 3. Fig. 4. NQO1 is required to suppress lipid peroxidation by regeneration of the antioxidant form of CoQ10 in melanoma cells. (A-B) Western blot analyses and relative quantifications of FSP1 (A) and NQO1 (B) protein levels in a panel of melanoma cell lines, normalized to protein level in melanocytes (HEMa-LP). 1-way ANOVA with Dunnett's multiple comparisons test (HEMa-LP as a control) was used to quantify statistical significance. Error bars represent SEM, n ≥ 4. (C) mRNA expression of FSP1 and NQO1 in a panel of melanoma cell lines. 1-way ANOVA with Dunnett's multiple comparisons test (HEMa-LP as a control) was used to quantify statistical significance. Error bars represent SEM, n = 3. (D) Lipid peroxidation assay using Bodipy C11 581/591 FACS upon NQO1 KD in melanoma cell lines SkMel28, A375 and Mel Juso. Ratio between oxidized and reduced Bodipy C11 was measured and reported as fold change over control conditions (ctrl). As positive control, cells were treated for 30 min with 100 μM of CH.
Unpaired Student two-tailed t-test was used to quantify statistical significance. Error bars represent SEM, n ≥ 3. (E) Idebenone rescue of lipid peroxidation upon NQO1 KD . Lipid peroxidation was assessed using Bodipy C11 FACS upon NQO1 KD in melanoma cell lines SkMel28 and A375 in presence or absence of idebenone (100 nM of idebenone for SkMel28, 200 nM of idebenone for A375). Ratio between oxidized and reduced Bodipy C11 was measured and reported as fold change over control conditions (ctrl). One-way ANOVA with Tukey's multiple comparisons test was used to quantify statistical significance. Error bars represent SEM, n ≥ 3. (F) NQO1 knockdown impairs viability of melanoma cells. Growth curves of melanoma cell lines SkMel28 and A375 upon NQO1 KD compared to control conditions (ctrl). 2-way ANOVA was used to quantify statistical significance. Error bars represent SEM, n = 3. (G) Co-silencing of NQO1 and UBIAD1 dramatically impairs viability of melanoma cells. Growth curves of melanoma cell lines SkMel28 and A375 upon mild NQO1 KD (NQO1 KD Low ), mild UBIAD1 KD (UBIAD1 KD Low ), or simultaneous NQO1 KD Low + UBIAD1 KD Low compared to control conditions (ctrl). 2-way ANOVA was used to quantify statistical significance. Error bars represent SEM, n ≥ 3.
(H) qRT-PCR quantifications of UBIAD1, NQO1 and FSP1 mRNA levels in UBIAD1 KD or NQO1 KD melanoma cells. One-sample t-test (hypothetical mean = 1) was used to quantify statistical significance. Error bars represent SEM, n ≥ 6. (A) Proposed mechanism of UBIAD1-and NQO1mediated protection against lipid peroxidation in the normal situation (on the left): non-mitochondrial CoQ10 is synthesized by UBIAD1 in Golgi and transported to plasma membrane, where it is reduced by NQO1 to protect against lipid peroxidation (the scheme is designed using Biorender.com). In UBIAD1 KD and NQO1 KD cells (on the right) loss of reduced plasma membrane CoQ10 leads to lipid peroxidation, increased ROS levels and apoptotic cell death.