Targeting Glycolysis with 2-Deoxy-D-Glucose and Lysosomal Integrity with L-Leucyl-L-Leucine Methyl Ester as Antimelanoma Strategy

1. Introduction

Melanoma is a highly aggressive skin cancer whose incidence has increased more rapidly than that of any other cancer type since the mid-1950s [1]. Global estimates indicate a current incidence of ~3.3 per 100,000 individuals [2]. While early-stage melanoma can often be successfully treated with surgical excision, metastatic melanoma remains a major therapeutic challenge, with a dismal 5-year survival rate of only 23% [3]. Approximately half of patients with advanced melanoma harbor activating BRAF mutations that drive MEK/ERK pathway activation and initially respond to BRAF inhibitors, but most relapse within months due to acquired resistance [4]. Therefore, there is an urgent need for novel and durable therapeutic strategies.

A hallmark of many cancers, including BRAF-mutant melanoma, is their strong reliance on aerobic glycolysis to meet the energy demands of rapid growth and proliferation [5,6]. In addition to enabling rapid ATP generation, glycolysis supplies key metabolic intermediates required for biomass synthesis, further fueling tumor expansion. This metabolic dependency renders melanoma cells particularly vulnerable to glycolysis inhibition. For instance, 2-deoxy-D-glucose (2DG), a glucose analog that competitively inhibits the glycolytic enzymes hexokinase (HK) and phosphoglucoisomerase (GPI) [7], has been shown to reduce ATP production and enhance the cytotoxic effects of several conventional [8,9] and experimental antimelanoma agents [10,11,12,13]. Importantly, 2DG is well tolerated in combination with chemotherapy and radiotherapy in patients with solid tumors [14,15,16].

In addition to metabolic rewiring, melanoma cells exhibit an increased number, size, and enzymatic content of lysosomes, accompanied by structural instability of the lysosomal membrane [17,18,19]. These altered lysosomes promote invasion and metastasis by secreting hydrolytic enzymes cathepsins that degrade the extracellular matrix [17,20]. They also contribute to drug resistance by sequestering and exporting weakly basic chemotherapeutics [17,20]. Moreover, as final effectors of the autophagy pathway, lysosomes support tumor growth by degrading damaged organelles and recycling their components into biosynthetic precursors and energy-yielding metabolites [20]. In contrast, elevated lysosomal content and cathepsin levels, together with membrane instability, may render melanoma cells particularly susceptible to lysosomal membrane permeabilization (LMP). LMP causes cathepsin release into the cytosol, where they non-specifically cleave intracellular proteins and initiate cell death [18,20]. Consequently, several LMP-inducing agents have been investigated for melanoma therapy [17,21,22,23,24,25,26,27]. L-leucyl-L-leucine methyl ester (LLOMe) is a lysosomotropic dipeptide that accumulates in lysosomes, where it is processed by cathepsin C into membrane-disrupting polymers, resulting in LMP and subsequent release of cathepsins into the cytosol [28,29,30,31,32]. To date, only one study has evaluated LLOMe’s antimelanoma potential, showing selective toxicity toward melanoma cells over primary melanocytes [33].

Combination therapy is routinely employed in oncology to simultaneously target multiple pathways, thereby increasing therapeutic impact and limiting resistance evolution [34]. Accordingly, our therapeutic strategy focuses on the simultaneous targeting of two specific vulnerabilities of melanoma cells: their reliance on glycolytic metabolism and the structural instability of their lysosomes. We demonstrated that 2DG and LLOMe synergistically induce severe energy depletion and mixed apoptotic–necrotic death of A375 melanoma cells by inhibiting both glycolysis and oxidative phosphorylation (OXPHOS), with 2DG primarily targeting glycolysis, while LLOMe predominantly impairs OXPHOS through LMP-induced cathepsin release and subsequent mitochondrial dysfunction. However, the therapeutic potential of the combined treatment was compromised by the high toxicity of LLOMe toward primary cells. Therefore, we further identified two LMP-inducing compounds that may represent more suitable candidates for synergistic combination with 2DG in future antimelanoma strategies.

2. Materials and Methods

2.1. Cell Culture

A375 melanoma cells with a BRAFV600E mutation (a kind gift from Dr. Jelena Grahovac, Institute for Oncology and Radiology of Serbia) and Normal Human Dermal Fibroblasts (NHDF; Merck, Darmstadt, Germany) were cultured at 37 °C in a humidified atmosphere with 5% CO₂ in high-glucose (4.5 g/L) DMEM with L-glutamine (Capricorn Scientific, Ebsdorfergrund, Germany), supplemented with 10% fetal bovine serum (FBS; Gibco, Thermo Fisher Scientific, Waltham, MA, USA), 1 mM sodium pyruvate, and antibiotic/antimycotic solution (both from Capricorn Scientific). After a 24 h resting period, cells were treated with 2-deoxy-D-glucose (2DG; Merck) and/or L-leucyl-L-leucine methyl ester (LLOMe; MedChemExpress, Monmouth Junction, NJ, USA) in the presence or absence of the following compounds: necrostatin-1, ferrostatin-1, wortmannin, Q-VD-OPh, bafilomycin A1, MG132, N-acetylcysteine (NAC), BAPTA-AM, sodium succinate dibasic hexahydrate, or L-carnitine (all from Merck); and pepstatin A or superoxide dismutase 1 (SOD1) (both from Santa Cruz Biotechnology, Dallas, TX, USA). Alternatively, cells were treated with 2DG in the presence or absence of mefloquine or siramesine (both from Merck), or with LLOMe in the presence of sodium dichloroacetate (DCA; Merck) or shikonin (MedChemExpress). A375 cells were seeded at 15 × 10³ cells/well in 96-well plates for crystal violet test, 25 × 10³ cells/well in 96-well plates for oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) measurement, 15 × 10⁵ cells/well in 24-well plates for flow cytometric analysis and fluorometry, 10 × 10⁵ cells/well in 24-well plates for fluorescent microscopy, 1 × 10⁶ cells/well in 24-well plates for cell transfection and immunoblotting. NHDF cells were seeded at 9 × 10³ cells/well in 96-well plates for crystal violet test.

2.2. Cell Viability

Cell viability was assessed by staining adherent cells with crystal violet (Merck), as previously described [35].

2.3. Synergism Assessment

The type of interaction between the two treatments (additive, synergistic, or antagonistic) was evaluated using the equation: α = (SF_2DG×SF_LLOMe)/SF_2DG+LLOMe, where SF_2DG and SF_LLOMe represent the surviving fractions after treatment with 2DG and LLOMe, respectively, and SF_{2DG +LLOMe} represents the surviving fraction after the combined treatment. Statistical significance of the α values was determined using a one-sample t-test against the theoretical value of 1. An α = 1 indicates an additive effect, α > 1 indicates synergism, and α < 1 indicates antagonism.

2.4. Apoptosis/Necrosis Analysis

Apoptotic and necrotic cell death were analyzed upon double staining with annexin V-FITC and 7-AAD (BD Biosciences, San Diego, CA, USA), according to the manufacturer’s protocol. Annexin V binds to phosphatidylserine on apoptotic cells, while 7-AAD labels necrotic cells with compromised membranes, enabling distinction of viable (annexin⁻/7-AAD⁻), apoptotic (annexin⁺/7-AAD⁻), and necrotic (annexin⁺/7-AAD⁺) populations. DNA fragmentation (sub-G0/G1 compartment), a hallmark of apoptotic cell death, was evaluated by cell cycle analysis of propidium iodide-stained cells exactly as previously described [36]. Both analyses were performed on the FACS Aria III flow cytometer, using FACSDiva 6.0 software for acquisition and FlowJo 10.7 software for analysis (BD Biosciences).

2.5. Lysosomal Staining

Lysosomes were stained with acridine orange (Thermo Fisher Scientific) or LysoTracker Red (MedChemExpress) according to the manufacturers’ instructions. Acridine orange-stained cells were analyzed using inverted fluorescent microscope (Leica Microsystems DMIL, Wetzlar, Germany), Leica Microsystems DFC320 camera and Leica Application Suite software (version 2.8.1), where lysosomes appeared as orange/red cytoplasmic vesicles, while nuclei and cytoplasm were stained green. Alternatively, acridine orange- and LysoTracker Red-stained cells were analyzed by flow cytometry using a FACS Aria III flow cytometer. The acidic lysosomal content was quantified as the mean red-to-green fluorescence ratio (FL3/FL1) for acridine orange, and as the mean red fluorescence intensity for LysoTracker Red (FL1). In both cases, results are presented relative to untreated control cells, which were arbitrarily set to 1.

2.6. Caspase Activation, Mitochondrial Membrane Potential and Superoxide Measurement

Caspase activation was assessed using the cell-permeable, FITC-conjugated pan-caspase inhibitor Z-VAD-FMK (ApoStat; R&D Systems, Minneapolis, MN, USA), by quantifying the increase in green fluorescence intensity. Mitochondrial membrane potential (MMP) was evaluated using JC-1 dye (MedChemExpress). In polarized mitochondria, JC-1 forms red fluorescent aggregates, whereas in depolarized mitochondria it remains in the green fluorescent monomeric form. MMP changes were expressed as the ratio of green to red fluorescence. Mitochondrial superoxide (O₂⁻•) was detected using MitoSOX Red (MedChemExpress), which selectively reacts with O₂⁻• to emit red fluorescence proportional to its concentration. All fluorescent probes were used according to the manufacturers’ instructions for flow cytometry staining, but prior to fluorescence measurement on a Hidex Sense microplate reader (Hidex, Turku, Finland), stained cells were transferred into black 96-well plates. Fluorescence was measured using appropriate filter sets with excitation/emission wavelengths of 488/530 nm for FITC-labeled Z-VAD-FMK and JC-1 monomers, 540/590 nm for JC-1 aggregates, and 510/580 nm for MitoSOX Red. Caspase activity and MitoSOX Red fluorescence were normalized to DAPI fluorescence (Thermo Fisher Scientific) as an estimate of cell number per well. All results are presented as fold change relative to untreated control cells arbitrarily set to 1.

2.7. Measurements of Oxygen Consumption and Extracellular Acidification Rates

OXPHOS activity and glycolytic flux in A375 melanoma cells were assessed by measuring the oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), using the MitoXpress Xtra Oxygen Consumption Assay and the pH-Xtra Glycolysis Assay, respectively (both from Agilent, Santa Clara, CA, USA), according to the manufacturer’s instructions. Fluorescence was recorded in real time for 1 hour using a Hidex Sense microplate reader in time-resolved mode, starting 2 hours after treatment. The rate of signal change (slope) across measurement cycles was calculated for each condition and used as a measure of OCR and ECAR. For comparative analysis, values were normalized to the slope obtained in the untreated control group, which was set to 1.

2.8. Intracellular ATP Quantification

The intracellular concentration of ATP was determined using a Hidex Sense microplate reader and the Luminescent ATP Detection Assay Kit (Abcam, Cambridge, UK), according to the manufacturer’s instructions. Luminescence values were normalized to crystal violet staining performed in a parallel plate to account for differences in cell number. Results are presented as fold change relative to untreated control cells, arbitrarily set to 1.

2.9. RNA Interference

Small interfering RNA (siRNA) targeting human HK2 and ATG5, and corresponding control siRNA (all from Santa Cruz Biotechnology) were transfected into A375 cells by electroporation in the 4D-Nucleofector X Unit, using the SF Cell Line 4D-Nucleofector X Kit and DC-135 program (Lonza, Basel, Switzerland), according to the manufacturer’s instructions. The cells were rested for 24 h before treatment.

2.10. Immunoblotting

Knockdown efficiency of HK2 and ATG5 was confirmed by immunoblotting exactly as previously described [36], using specific primary antibodies: anti-ATG5 (#12994, Cell Signaling Technology, Beverly, MA, USA), anti-HK2 (sc-130358, Santa Cruz Biotechnology), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (MA5-15738; Thermo Fisher Scientific) as a loading control. Peroxidase-conjugated anti-rabbit IgG (111-035-144) or anti-mouse IgG (115-035-146) (both from Jackson ImmunoResearch, West Grove, PA, USA) were used as secondary antibodies. Protein bands were visualized by enhanced chemiluminescence on a ChemiDoc MP Imaging System (Bio-Rad, Hercules, CA, USA).

2.11. In Silico Analysis of Gene Expression

Raw gene expression data were retrieved from the GEO dataset GSE3189 (platform: Affymetrix U133A; accession GDS1375). The study groups were balanced in terms of age and sex, and clinical/pathological characteristics are described in the original publication by Talantov et al. [37]. Normalized log₂ expression values for selected probe sets were downloaded directly from the GEO portal. Samples were categorized into three groups: normal skin (n = 7), benign nevi (n = 18), and primary melanoma (n = 45). Statistical comparisons between groups were performed using the Mann-Whitney U test (GraphPad Software, San Diego, CA, USA), while the data were presented as median values and interquartile range. To prioritize sensitivity over stringency in this exploratory analysis, no correction for multiple testing was applied, in line with recommendations for hypothesis-generating studies [38]. Statistical significance is indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

2.12. Statistical Analysis

Statistical significance of differences between treatments was assessed using Student’s t-test or one-way ANOVA followed by Tukey’s post hoc test for multiple comparisons, unless otherwise stated. A p value < 0.05 was considered statistically significant.

3. Results

3.1. 2DG and LLOMe Synergistically Reduce Viability of A375 Melanoma Cells

Given the upregulated glycolysis [5,6] and expanded lysosomal compartment [17,18,19] observed in melanoma cells, we investigated whether their simultaneous targeting could synergistically reduce cell viability. Cell viability of A375 cells treated with increasing concentrations of 2DG and LLOMe for 24 and 48 hours was evaluated using the crystal violet assay. The results showed that cell viability decreased in a dose- and time-dependent manner for both compounds individually, and more pronouncedly when used in combination (Figure 1A, D). To quantify potential synergistic effects, we calculated α-indices, where values above 1 indicate synergy (Figure 1B, E). Based on the α-index, the combination of 5 mM 2DG and 1 mM LLOMe was selected for subsequent mechanistic analyses. Phase-contrast microscopy showed that 2DG or LLOMe alone caused partial cell rounding, while their combination induced widespread rounding, detachment, and shrinkage, indicating extensive cell death (Figure 1C). Together, these findings demonstrate that co-targeting glycolysis and lysosomal integrity synergistically reduces melanoma cell viability in a dose- and time-dependent manner, highlighting a potentially exploitable vulnerability in melanoma cells.

3.2. 2DG+LLOMe Induces Mixed Apoptotic and Necrotic Death in Melanoma Cells

We next investigated the mode of cell death induced by 2DG, LLOMe and their combination. Ferrostatin-1 and necrostatin-1 failed to protect A375 cells from 2DG+LLOMe-induced cytotoxicity, suggesting that ferroptosis and necroptosis are not involved in the antimelanoma effect of the combined treatment (Figure 2A and B). Furthermore, autophagy inhibition by either wortmannin or ATG5 knockdown did not affect the viability of cells treated with 2DG, LLOMe, or their combination (Figure 2C, D), suggesting that the cytotoxic effects of these treatments are independent of autophagy modulation. Both 2DG and LLOMe induced caspase activation, with the strongest effect observed upon their combination, as measured by Apostat fluorescence using fluorometry (Figure 2F). Annexin V-FITC/PI flow cytometry showed that 2DG, LLOMe, and especially their combination increased the proportion of Annexin V⁺/PI⁻ apoptotic cells, while the combined treatment also elevated the proportion of Annexin V⁺/PI⁺ necrotic cells (Figure 2E). Moreover, the two agents synergistically induced apoptosis-associated DNA fragmentation, as evidenced by an increased sub-G0/G1 population in cell cycle analysis (Figure 2G). Finally, inhibition of caspases by Q-VD-OPh reduced the cytotoxicity induced by 2DG and/or LLOMe, supporting a key role for caspase-mediated apoptosis in these treatments. Collectively, our findings demonstrate that 2DG+LLOMe induce caspase-mediated apoptosis accompanied by necrosis, while ferroptosis, necroptosis, and autophagy-related pathways appear not to be involved.

3.3. Antimelanoma Effect of 2DG+LLOMe Is Mediated by Lysosomal Destabilization

As LLOMe is a well-known lysosome-destabilizing agent [39], we next investigated whether LMP contributes to cell death induced by the combination of 2DG and LLOMe. Phase-contrast microscopy revealed that cells treated with LLOMe or 2DG+LLOMe exhibited vacuole-like intracellular structures as early as 30 min after treatment, which may represent swollen lysosomes (Figure 3A). Fluorescence microscopy (Figure 3B) and flow cytometry (Figure 3C) showed that LLOMe alone or in combination with 2DG induced a red-to-green fluorescence shift of acridine orange (AO) 30 min after treatment, indicating decreased acidity of lysosomes or autolysosomes. Consistently, LysoTracker Red fluorescence intensity was markedly reduced within 30 min of LLOMe ± 2DG treatment (Figure 3D), further confirming rapid lysosomal damage. Bafilomycin A1, an inhibitor of lysosomal acidification and the entry of acidophilic agents into lysosomes [40], and MG132, a proteasome inhibitor that also targets cysteine cathepsins B, C, L, and S [41,42,43,44] suppressed the cytotoxicity of LLOMe-containing treatments (Figure 3E, F). The toxicity of LLOMe and 2DG+LLOMe was not reduced by pepstatin A, an inhibitor of aspartic proteases such as CTSD and CTSE [45,46] (Figure 3G). On the other hand, the Ca²⁺ chelator BAPTA-AM further enhanced LLOMe- and 2DG+LLOMe-induced cytotoxicity (Figure 3H), likely due to the requirement of Ca²⁺ for lysosomal membrane repair [39]. These findings indicate that the entry of LLOMe into lysosomes, lysosomal destabilization, and the activity of cysteine cathepsins are essential for the cytotoxic effect of LLOMe, both alone and in combination with 2DG.

3.4. 2DG+LLOMe-induced Cell Death is Mediated by LMP-dependent Mitochondrial Depolarization and Superoxide

Apoptosis and necrosis are frequently associated with oxidative stress [47,48] and mitochondrial membrane potential (MMP) loss [49,50]. We next examined if these processes are involved in 2DG and/or LLOMe induced cytotoxicity. An increased green-to-red fluorescence ratio of the JC-1 dye, measured using a fluorescence plate reader, demonstrated that LLOMe, both alone and in combination with 2DG, induced significant MMP loss, which was prevented by bafilomycin A1 and MG132, inhibitors of lysosomal acidification and cysteine cathepsins, respectively (Figure 4A). Furthermore, as evidenced by enhanced fluorescence in cells stained with MitoSOX Red, LLOMe-containing treatments also increased mitochondrial O₂•⁻ production, which was also attenuated in the presence of bafilomycin A1 and MG132 (Figure 4B). Pretreatment with the antioxidant N-acetyl cysteine (NAC) partially rescued cell viability, indicating that ROS contributes to 2DG+LLOMe-induced cytotoxicity (Figure 4C). On the other hand, pretreatment with superoxide dismutase (SOD) exacerbated 2DG- and/or LLOMe-induced cytotoxicity, indicating that O₂•⁻-derived oxidants, rather than O₂•⁻ itself, mediate 2DG+LLOMe-induced cell death (Figure 4D). Together, these findings indicate that cell death induced by LLOMe, both alone and in combination with 2DG, is mediated, at least in part, by LMP, which leads to mitochondrial damage manifested by membrane depolarization and O₂•⁻ production.

3.5. Combined Glycolytic and Mitochondrial Inhibition by 2DG and LLOMe Triggers Energetic Collapse and Loss of Viability

Since MMP loss is typically associated with mitochondrial dysfunction, we next measured OXPHOS following treatment with 2DG, LLOMe, or their combination. OCR measurements showed that LLOMe, both alone and more strongly in combination with 2DG, reduced OXPHOS (Figure 5A). In parallel, ECAR analysis revealed that 2DG strongly and LLOMe moderately suppressed glycolysis, while the most pronounced reduction in glycolytic activity was observed with the 2DG+LLOMe combination (Figure 5A). Accordingly, treatment with 2DG, more prominently with LLOMe, and most strongly with their combination led to a gradual, time-dependent reduction in ATP levels (Figure 5C). OXPHOS enhancers L-carnitine and succinate [51,52], partially rescued cell viability reduced by 2DG+LLOMe (Figure 5C), indicating that energy depletion contributed to the cytotoxic effect of the combined treatment. Furthermore, LLOMe synergized with other glycolytic inhibitors pyruvate dehydrogenase kinase (PDK) inhibitor dichloroacetate (DCA) [53] and pyruvate kinase M2 inhibitor (PKM2) shikonin (SHI) [54] in exerting an antimelanoma effect (Figure 5D, F). In addition, LLOMe potentiated the cytotoxicity induced by genetic knockdown of HK2, a key glycolytic enzyme and the primary target of 2DG [55] (Figure 5E), indicating that glycolysis inhibition by 2DG is critical for the synergistic antimelanoma effect with LLOMe. These results demonstrate that 2DG+LLOMe-induced cell death is driven by severe energy depletion resulting from combined inhibition of glycolysis and OXPHOS.

3.6. In Contrast to Mefloquine and Siramesine, LLOMe Exhibits Non-Selective Toxicity

To evaluate the therapeutic potential and selectivity of the 2DG+LLOMe combination, we compared its cytotoxicity in melanoma cells and primary human dermal fibroblasts (NHDF). As demonstrated by MTT test, fibroblasts were slightly less sensitive to 2DG than A375 cells, but significantly more sensitive to LLOMe (Figure 6A). Notably, both cell types exhibited similar sensitivity to the 2DG+LLOMe combination (Figure 6A), arguing against its therapeutic applicability. Therefore, we tested the melanoma selectivity of two other LMP-inducing agents, mefloquine and siramesine [20]. As shown in Figure 6B, fibroblasts were less sensitive to mefloquine, and especially to 2DG+mefloquine, than A375 cells (Figure 6B). Moreover, although siramesine alone killed A375 cells and fibroblasts to a similar extent, melanoma cells were more sensitive than fibroblasts to the 2DG+siramesine combination (Figure 6C). These findings suggest that, unlike LLOMe, the combination of 2DG with mefloquine or siramesine exhibits greater selectivity toward melanoma cells and may represent a more promising therapeutic approach.

3.7. Melanoma Progression in Patient Samples Is Associated with Enhanced Expression of Glycolytic Enzymes and Cathepsins

We analyzed publicly available gene expression data GSE3189 to compare glycolytic enzyme and lysosomal protease expression in normal skin, benign nevi, and primary melanoma. Among glycolytic enzymes, we observed significantly increased expression of HK isoform 3 (HK3), GPI, phosphofructokinase muscle type (PFKM), aldolase A (ALDOA), triosephosphate isomerase 1 (TPI1), GAPDH, and PKM, and lactate dehydrogenase A (LDHA) in melanoma samples compared to both benign nevi and normal skin. Moreover, GPI, PFKM, GAPDH, and PKM were also significantly upregulated in benign nevi relative to normal skin, suggesting that early metabolic changes may occur during the formation of nevi, while full activation of glycolysis characterizes melanoma progression. In contrast, the expression levels of HK1 and HK2 did not significantly differ between melanoma and healthy skin, whereas ALDOC expression was significantly reduced in melanoma compared to both healthy controls and benign nevi (Figure 7A).

Moreover, cathepsin B (CTSB), cathepsin D (CTSD), and cathepsin Z (CTSZ) were progressively upregulated from normal skin to nevi and melanoma, indicating stepwise lysosomal protease activation during melanocytic progression (Figure 7B). In contrast, the expression of cathepsin C (CTSC), the main mediator of LLOMe-induced cytotoxicity, was significantly reduced in melanoma compared to both nevi and normal skin. Interestingly, cathepsin L (CTSL) expression was decreased in benign nevi compared to both normal and melanoma samples (Figure 7B). Overall, the results indicate a global upregulation of glycolytic genes (9 out of 13 analyzed) and cathepsins (3 out of 6 analyzed) during the progression from normal skin to melanoma.

4. Discussion

The present study provides evidence that combined disruption of lysosomal integrity and glycolytic function leads to a profound metabolic collapse and cell death in melanoma cells. LLOMe disrupts mitochondrial function via cysteine cathepsin-dependent mechanisms, leading to OXPHOS suppression and partial glycolysis inhibition, while 2DG inhibits glycolysis. Their combined action induces a profound energy deficit, resulting in both apoptotic and necrotic cell death. Although this approach effectively reduced melanoma cell viability, the non-selective toxicity of LLOMe remains a critical limitation for its therapeutic application. This underscores the need to identify safer LMP inducers, and highlights mefloquine and siramesine as promising candidates for further investigation in combination with glycolysis inhibitors.

Consistent with our previous findings that 2DG synergizes with the lysosomal detergent N-dodecylimidazole (NDI) in killing B16 mouse melanoma cells [21], its combination with LLOMe similarly produced a synergistic reduction in the viability of A375 human melanoma cells (Figure 1A-E). Similarly to that study, the cytotoxicity of combined treatment was not mediated by necroptosis or ferroptosis (Figure 2A, B). 2DG has been shown to stimulate prosurvival autophagy [56,57] and to inhibit cytotoxic autophagy [58]. Although lysosomal damage is generally expected to impair autophagic flux by compromising lysosomal degradative capacity [59], we found no literature evidence supporting such an effect for LLOMe. Instead, LLOMe has been reported to induce lysophagy, a selective, cytoprotective autophagy that removes damaged lysosomes [39,60,61,62,63,64]. It is therefore plausible that 2DG may interfere with LLOMe-induced lysophagy. However, neither pharmacologic nor genetic inhibition of autophagy affected the cytotoxicity of 2DG, LLOMe, or their combination (Figure 2C, D), arguing against a role for autophagy modulation in mediating their cytotoxic effects. 2DG induces apoptosis [65,66], while LLOMe can trigger either apoptosis [32,39,67,68] or necrosis [44,69,70,71], depending on the dose and cell type. Consistent with this, we observed that both 2DG and LLOMe individually induced caspase-dependent apoptosis in melanoma cells (Figure 2E-H). However, their combination triggered both apoptotic and necrotic death, with apoptosis still predominating (Figure 2E). Given that apoptosis is a highly energy-dependent process, unlike necrosis [72], it is plausible that the strong energy depletion caused by combination of 2DG and LLOMe may have diverted part of the apoptotic response toward necrosis.

LLOMe is an acidophilic compound that is cleaved and polymerized by cathepsin C inside lysosomes, generating membranolytic products that trigger LMP [28]. Following LMP, some authors attribute LLOMe-induced cytotoxicity to the translocation of cathepsins B and L from lysosomes to mitochondria, where they degrade mitochondrial proteins and trigger MMP loss [29]. Others report that cysteine cathepsins B, L, S, K, and H activate pore-forming Bid and degrade anti-apoptotic proteins such as Bcl-2, Bcl-xL, and Mcl-1, thereby also promoting MMP loss [30,31]. Alternatively, cathepsins B and L were shown to activate caspases in cytoplasm [32]. We demonstrated that LLOMe, alone or in combination with 2DG, induces rapid lysosomal deacidification (Figure 3A-D), while cytotoxicity of these treatments was partially prevented by the V-ATPase inhibitor bafilomycin A1 (Figure 3E), which probably prevented accumulation of acidophilic LLOMe inside lysosomes. The Ca²⁺ chelator BAPTA-AM further enhanced LLOMe- and LLOMe+2DG-induced cytotoxicity (Figure 3H), which indirectly supports the involvement of LMP in cell death, given the established role of Ca²⁺ in promoting lysosomal membrane repair [39]. Partial protection was also observed with MG132, a proteasome inhibitor that also targets cathepsins B, C, L, and S [41,42,43,44], suggesting that one or more of cysteine cathepsins may contribute to the observed cytotoxicity. However, the only partial protection implies that cathepsin activity is not the sole mediator and that additional cathepsin-independent cytotoxic mechanisms are likely involved.

Mitochondrial depolarization is a key event in the intrinsic pathway of apoptosis, often preceding cytochrome c release and caspase activation, but it can also indicate irreversible mitochondrial dysfunction leading to necrosis when ATP levels are insufficient to support apoptotic execution [49]. In accordance with previous results [29,30], we demonstrated LLOMe triggers MMP loss (Figure 4A). Mitochondrial depolarization was even more pronounced in the presence of 2DG (Figure 4A), possibly due to the cell’s inability to maintain mitochondrial membrane potential under conditions of severe ATP depletion [73]. The partial prevention of MMP loss by bafilomycin A1 and MG132 suggests that lysosomal acidification, required for LLOMe accumulation in lysosomes, as well as the activity of cysteine cathepsins contribute to LLOMe-induced mitochondrial depolarization. However, whether LLOMe-induced mitochondrial depolarization results from cysteine cathepsin-mediated degradation of proteins that regulate mitochondrial permeability in the cytoplasm (as suggested by [30,31,32], or within mitochondria [29], remains to be clarified.

MMP loss can enhance O₂•⁻ production by disrupting electron flow through the electron transport chain [74], while O₂•⁻ itself exacerbates mitochondrial dysfunction by promoting permeability transition pore opening and further depolarization [75], thereby establishing a positive feedback loop. LLOMe was previously shown to stimulate both mitochondrial O₂•⁻ production and MMP loss in a cathepsin B-dependent manner [76]. Accordingly, we observed that LLOMe, alone or in combination with 2DG, significantly increased mitochondrial O₂•⁻ levels (Figure 4B), an effect that was partially prevented by bafilomycin A1 and MG132, further implicating lysosomal involvement in mitochondrial oxidative stress. Moreover, the non-specific antioxidant NAC protected against LLOMe±2DG-induced toxicity, while SOD exacerbated it (Figure 4C), suggesting that hydrogen peroxide (H₂O₂), a product of SOD-mediated O₂•⁻ dismutation [77], or its downstream products may contribute to oxidative damage. Lysosomes serve as major reservoirs of chelatable iron due to autophagic degradation of iron-containing proteins, and upon LMP, the released iron can catalyze Fenton reactions, generating highly reactive hydroxyl radicals (•OH) from H₂O₂ [78]. While •OH can initiate lipid peroxidation of membrane phospholipids, including those in lysosomes [78], the inability of the lipid radical-trapping antioxidant Ferrostatin-1 [79] to rescue cells from LLOMe or 2DG+LLOMe (Figure 2A) argues against lipid peroxidation as the primary cytotoxic mechanism, pointing instead to oxidative damage to proteins, mitochondria, or DNA.

Since ATP synthase relies on MMP as a driving force [73], MMP loss disrupts OXPHOS and impairs ATP production. Accordingly, LLOMe-induced MMP loss was associated with a decrease in OXPHOS activity (Figure 5A). Consistent with this, a previous study showed that LLOMe impaired mitochondrial respiration, which was reversed by cathepsin inhibition [29]. While 2DG has been shown to enhance OXPHOS in cells with metabolically flexible mitochondria as a compensatory response to glycolytic inhibition [80,81], it can suppress mitochondrial respiration in highly glycolytic cells by limiting pyruvate availability for the TCA cycle and subsequent OXPHOS [82,83]. In line with the latter, 2DG tended to reduce OXPHOS in our model, although this effect did not reach statistical significance (Figure 5A). As expected, the strongest inhibition was observed when both treatments were applied in combination (Figure 5A). Interestingly, although LLOMe was previously shown to transiently increase glycolysis after 1 hour in macrophages [29], it reduced glycolytic activity in our model (Figure 5A), suggesting that its metabolic effects may depend on cell type and duration of exposure. The observed glycolysis suppression by LLOMe may be attributed to cathepsin-mediated degradation of glycolytic enzymes [84] and/or a global metabolic decline caused by severe cellular damage [85]. As expected, the glycolytic inhibitor 2DG [7] suppressed ECAR and the strongest inhibition was observed following the combined treatment (Figure 5B). L-carnitine promotes ATP production by facilitating the transport of long-chain fatty acids into mitochondria for β-oxidation [86], while succinate supports mitochondrial respiration by acting as both a TCA cycle intermediate and an electron donor to complex II of the electron transport chain [87,88]. Both energy-boosting agents partially attenuated 2DG+LLOMe-induced cytotoxicity, indicating that energy depletion mediates the cytotoxicity of the combined treatment. However, the rescue was limited, most likely because mitochondria were already largely dysfunctional and unable to efficiently synthesize ATP. Inhibition of glycolysis by DCA, shikonin, or HK2 knockdown mimicked the effect of 2DG and enhanced LLOMe-induced cytotoxicity (Figure 5D-F), confirming that glycolysis suppression underlies the synergy and that 2DG could potentially be replaced by other glycolysis-targeting agents.

Our analysis of publicly available gene expression data is consistent with previous studies reporting upregulation of glycolytic enzymes GPI [89], ALDOA [90], GAPDH [91,92], PKM2 [92,93], and LDHA [94,95], as well as downregulation of ALDOC [96] in melanoma patients. However, while previous studies reported increased expression of the platelet form of PFK (PFKP) [97], our study is the first to show a progressive upregulation of its muscle isoform PFKM from healthy skin to benign nevi and malignant melanoma (Figure 7A). TPI1 was upregulated in metastatic uveal melanoma [98,99,100], but we demonstrated its increase in cutaneous melanoma (Figure 7A). Unlike earlier reports showing HK2 upregulation in melanoma cell lines relative to immune cells [101], we found reduced HK2 expression in benign nevi compared to healthy skin, which may reflect differences in the degree of malignant transformation and the choice of control groups. Importantly, while previous studies described HK3 overexpression in non-melanoma cancers [102,103,104,105,106], we demonstrated its increased expression in malignant melanoma relative to both benign and normal tissue (Figure 7A).

In line with previous findings [107,108,109,110,111,112], our analysis revealed a progressive upregulation of CTSB and CTSD from normal skin through benign nevi to malignant melanoma (Figure 7B), confirming their potential involvement in melanocytic transformation and tumor progression. While CTSZ has been reported to be elevated in metastatic uveal melanoma compared to non-metastatic tissue or primary uveal melanocytes [113,114], our analysis demonstrates its upregulation in cutaneous melanoma samples (Figure 7B). Importantly, although CTSC is upregulated in various non-melanoma cancers [115,116], our study revealed its decreased expression in malignant melanoma compared to healthy skin and nevi (Figure 7B). Fibroblasts represent a major population of skin-resident cells [117,118], and the higher CTSC expression in healthy skin than in melanoma tissue may explain the greater sensitivity of skin fibroblasts to LLOMe than melanoma cells (Figure 6A). In contrast, the cytotoxic effects of mefloquine and siramesine are primarily mediated by cathepsin B [119,120,121], whose expression increases during malignant transformation (Figure 7B), potentially accounting for the enhanced selectivity of these agents toward melanoma (Figure 6B, C). We did not observe increased expression of HK2, the primary target of 2DG [122] but rather of HK3, for which no evidence of 2DG sensitivity has been reported. However, we detected elevated levels of PGI, another enzyme affected by 2DG [123], which may contribute to the increased susceptibility of melanoma cells to 2DG compared to dermal fibroblasts.

5. Conclusions

In summary, our results indicate that LLOMe disrupts lysosomal integrity and, likely through cysteine cathepsin activity, induces mitochondrial dysfunction and suppression of OXPHOS, but also inhibits glycolysis to a lesser extent through an unknown mechanism. In addition 2DG inhibits glycolysis. The combination of 2DG and LLOMe causes a severe energy crisis, ultimately resulting in mixed apoptotic and necrotic cell death. Although LLOMe was effective, its considerable toxicity toward non-malignant cells limits its translational potential, whereas alternative lysosome-destabilizing agents such as siramesine and mefloquine show greater selectivity for melanoma cells, spare fibroblasts, and synergize with 2DG. These findings support the rationale for further exploration of selective LMP inducers in combination with glycolysis inhibitors as a promising therapeutic strategy against melanoma.