Abstract

   Pharmacological vitamin C (VC) has gained attention for its pro-oxidant
   characteristics and selective ability to induce cancer cell death.
   However, defining its role in cancer has been challenging due to its
   complex redox properties. In this study, using a human osteosarcoma
   (OS) model, we show that the redox-active property of VC is critical
   for inducing non-apoptotic cancer cell death via intracellular reactive
   oxygen species (ROS)-iron-calcium crosstalk and mitochondrial
   dysfunction. In both 2D and 3D OS cell culture models, only the
   oxidizable form of VC demonstrated potent dose-dependent cytotoxicity,
   while non-oxidizable and oxidized VC derivatives had minimal effects.
   Live-cell imaging showed that only oxidizable VC caused a surge in
   cytotoxic ROS, dependent on iron rather than copper. Inhibitors of
   ferroptosis, a form of iron-dependent cell death, along with classical
   apoptosis inhibitors, were unable to completely counteract the
   cytotoxic effects induced by VC. Further pharmacological and genetic
   inhibition analyses showed that VC triggers calcium release through
   inositol 1,4,5-trisphosphate receptors (IP3Rs), leading to
   mitochondrial ROS production and eventual cell death. RNA sequencing
   revealed down-regulation of genes involved in the mitochondrial
   electron transport chain and oxidative phosphorylation upon
   pharmacological VC treatment. Consistently, high-dose VC reduced
   mitochondrial membrane potential, oxidative phosphorylation, and ATP
   levels, with ATP reconstitution rescuing VC-induced cytotoxicity. In
   vivo OS xenograft studies demonstrated reduced tumor growth with
   high-dose VC administration, concomitant with the altered expression of
   mitochondrial ATP synthase (MT-ATP). These findings emphasize VC's
   potential clinical utility in osteosarcoma treatment by inducing
   mitochondrial metabolic dysfunction through a vicious intracellular
   ROS-iron-calcium cycle.

   Keywords: Osteosarcoma; Reactive oxygen species; Endoplasmic reticulum
   oxidoreductase 1 alpha; Inositol 1,4,5-trisphosphate receptors
   oxidative phosphorylation; Electron transport chain; Mitochondrial
   dysfunction; ATP

1. Introduction

   Osteosarcoma (OS) is the most common primary tumor of the bone and is
   an especially aggressive, predominantly pediatric cancer that is often
   fatal in both children and adults [[25]1,[26]2]. Currently, the
   standard treatment options for OS are surgery, chemotherapy, and
   radiation therapy, which may be used alone or in combination
   [[27]3,[28]4]. However, these conventional treatments can result in
   various adverse effects, such as cardiac toxicity, infertility, and
   kidney dysfunction [[29]5,[30]6]. Given these limitations, there is
   increasing demand for new and alternative approaches to treat OS
   without causing unwanted side effects.

   Vitamin C (VC) is a crucial nutrient essential for maintaining cellular
   physiology [[31]7]. Interestingly, it acts as an antioxidant at
   physiological concentrations (40–80 μM) in human plasma) and a
   pro-oxidant at high doses (10–20 mM), suppressing tumor growth
   [[32]8,[33]9]. The greater sensitivity of tumor cells to high-dose VC
   has been attributed to their lower capacity to neutralize reactive
   oxygen species (ROS) [[34]10]. As with other cancer types, increasing
   studies using osteosarcoma models have shown that high doses of VC can
   induce cell death independently or synergistically with conventional
   anti-cancer agents such as cisplatin and arsenic trioxide (ATO) to
   inhibit tumor cell proliferation and growth [[35][11], [36][12],
   [37][13], [38][14], [39][15], [40][16]]. Despite the growing evidence
   supporting the anti-cancer activity of VC, controversy surrounding its
   role in cancer may arise from its redox properties and dynamic
   interconversion between its reduced (ascorbic acid, AA) and oxidized
   (dehydroascorbate, DHA) forms. While current models suggest that DHA is
   the pharmacologically effective form [[41]17,[42]18], recent studies
   have shown varied effectiveness of this molecule in cancer cell models
   [[43][19], [44][20], [45][21]]. In addition, the lack of distinction
   between AA and DHA in certain previously reported outcomes adds
   confusion, leaving the contribution of the different redox forms to
   VC-induced cell death unclear [[46]22,[47]23].

   High-doses of VC have been shown to exhibit anti-cancer activity by
   generating extracellular hydrogen peroxide (H[2]O[2]), impacting
   redox-dependent signaling and metabolic pathways within cancer cells
   [[48]19,[49]24]. Catalytic metals, such as copper (Cu^+/Cu^2+) or iron
   (Fe^3+/Fe^2+), have been reported to mediate H[2]O[2] production via
   ascorbate oxidation [[50]19,[51]24]. Notably, compared to copper,
   intracellular labile iron has been shown in multiple cancer studies to
   facilitate increased susceptibility to VC, resulting in preferential
   cell death relative to normal cells [[52][25], [53][26], [54][27],
   [55][28], [56][29], [57][30]]. However, the exact involvement of the
   catalytic metals in VC-induced cytotoxicity in cancer models needs
   further clarification, as there are increasing numbers of reports with
   inconsistent findings regarding their roles, depending on the cancer
   cell type studied [[58][31], [59][32], [60][33], [61][34]].

   In addition to its pro-oxidant properties, VC has been shown to act as
   a cofactor for Fe^2+- and 2-oxoglutarate-dependent dioxygenases,
   including hypoxia-inducible factor (HIF) hydroxylases and DNA
   demethylases (e.g., ten-eleven translocation enzymes [TET1-3])
   [[62][35], [63][36], [64][37]]. While the involvement of these enzymes
   in tumor suppression is evident in various cancer models when using
   lower doses of VC (0.1–1 mM) [[65][38], [66][39], [67][40]], it remains
   unclear if these pathways contribute to acute cancer cell death with
   higher doses of VC [[68][39], [69][40], [70][41], [71][42], [72][43]].

   It has been reported that VC's anti-cancer effects involve various cell
   death mechanisms, including apoptosis, necroptosis, and autophagy,
   depending on concentration and cell type [[73]44]. Earlier studies
   suggested the involvement of caspase-dependent apoptosis or necrosis in
   VC-induced cancer cell death [[74][45], [75][46], [76][47]], but recent
   evidence indicates the potential involvement of non-canonical
   mechanisms such as ferroptosis, yielding mixed results [[77]48,[78]49].
   In this regard, an increasing number of studies suggest that
   intracellular Ca^2+ is a critical mediator of non-apoptotic forms of
   cell death (necroptosis, ferroptosis, parthanatos, pyroptosis) in
   response to ROS-inducing compounds [[79][50], [80][51], [81][52]].
   However, the precise interaction between intracellular Ca^2+ and ROS
   pathways in mediating VC-induced cancer cell death remains to be fully
   understood.

   Pharmacological VC has also been suggested to exert cytotoxic effects
   in cancer treatment through an additional mechanism involving the
   disruption of bioenergetics [[82]53,[83]54]. Indeed, recent studies
   have indicated that the cytotoxicity induced by pharmacological VC is
   associated with the overactivation of PARP1, triggered by ROS-mediated
   nuclear DNA damage [[84]55]. This overactivation leads to the
   consumption of NAD⁺ and subsequent depletion of ATP, ultimately
   resulting in mitotic cell death [[85]55]. However, subsequent
   investigations have indicated that the activation of PARP1 in response
   to ROS-induced DNA damage may not be essential, suggesting the
   involvement of alternative mechanisms of ATP depletion [[86]55,[87]56].
   In addition to cytosolic glycolysis, it has been well-documented that
   mitochondria play a crucial role in ATP production through oxidative
   phosphorylation (OXPHOS), which involves the mitochondrial electron
   transport chain (ETC) complexes (I to IV) whose components are encoded
   by mitochondrial DNA (mtDNA) [[88][57], [89][58], [90][59]]. However,
   the precise roles and mechanisms of mitochondrial pathways in
   VC-induced metabolic alterations and cell death are not yet clearly
   understood.

   In this study, we have shown that redox-reactive VC plays a crucial
   role in inducing non-apoptotic cell death via intracellular crosstalk
   between ROS-iron-calcium and mitochondrial metabolic pathways in human
   OS cells. Using 2D and 3D tumor models, we found that an oxidizable
   form of VC, relative to its reduced (AA) or oxidized (DHA) derivatives,
   exerts cancer cell specific cytotoxicity. Furthermore, live-cell
   analysis combined with pharmacological and genetic perturbation methods
   revealed that H[2]O[2] generated by VC upon its oxidation via iron acts
   in the endoplasmic reticulum (ER) by triggering Ca^2+ release through
   IP3Rs and subsequently provoking mitochondrial Ca^2+ overload and
   cytotoxic intracellular ROS production. Additional transcriptomic and
   metabolic analyses revealed that high-dose VC treatment comprehensively
   down-regulates mitochondrial OXPHOS-regulatory gene expression, which
   is accompanied by impaired mitochondrial metabolic functions and
   decreased ATP production. Correspondingly, orthotopic OS xenograft
   studies revealed that VC treatment reduces tumor growth and results in
   the marked alteration of mitochondrial ATP synthase expression. Our
   findings highlight the importance of the intricate interplay between
   the ROS, iron, and calcium pathways that target mitochondrial functions
   and metabolism in VC-induced cell death and tumor inhibition in human
   OS.

2. Results

2.1. Pharmacological doses of oxidizable vitamin C induce selective cell
death in human osteosarcoma cells

   Physiologic VC exists largely in its reduced (ascorbic acid [AA]) or
   oxidized (dehydroascorbic acid [DHA]) forms, with dynamic
   interconversion between them, to act on diverse cellular functions and
   processes ([91]Fig. 1A) [[92]60]. In contrast to its antioxidant
   function at physiological levels, VC at high doses (1–10 mM) has been
   reported to increase ROS and preferentially kill cancer cells
   [[93]9,[94]61,[95]62]. However, it is unclear which form of VC exerts
   this cytotoxic activity in tumors. Thus, to investigate the cancer
   cell-specific effects of different redox forms of VC, we subjected
   various OS cell lines (U–2OS, 143B, MNNG-HOS, Saos-2) derived from
   human patients as well as a human fetal osteoblastic cell line (hFOB
   1.19), as the non-malignant control, to comparative cell viability
   analysis. After treating these cell lines with plain VC, ascorbic acid
   2-phosphate (AA2P; a long-acting VC derivative that does not convert to
   DHA), or DHA at varying doses for 24 h, we found that plain VC at high
   doses (5–20 mM) markedly reduced cell viability in all the OS cell
   lines tested (U–2OS [IC[50] = 5.4286 mM], 143B [IC[50] = 4.8388 μM],
   MNNG-HOS [IC[50] = 7.7813 μM], Saos-2 [IC[50] = 2.9134 μM]), with much
   less of an effect on hFOB 1.19 control cells ([96]Fig. 1B–G). Notably,
   DHA treatment at 10–20 mM resulted in slightly reduced cell viability
   in some OS cells (U–2OS, 143B, Saos-2) but not in MNNG-HOS cells or the
   hFOB 1.19 control cells, while AA2P treatment induced no changes in
   cell viability in any of the cell lines at similar doses ([97]Fig.
   1B–G). Corresponding to the results, the clonogenic survival assay
   showed that high doses of plain VC (5–10 mM), but not 0.5 mM,
   completely abrogated single-cell derived colony formation, with the
   high dose effect being significantly blunted in hFOB 1.19 cells, while
   AA2P and DHA did not exhibit such inhibitory effects at similar doses
   ([98]Fig. 1H–L, [99]Supplementary Fig. 1). These findings align well
   with previous studies indicating that high doses of VC may display
   cancer cell-specific toxicity while preserving normal cells [[100]63].
   We also found that this VC-induced cell death was significantly blocked
   by pretreatment with catalase, a key antioxidant enzyme that inhibits
   ROS-induced cell damage by catalyzing the decomposition of hydrogen
   peroxide (H[2]O[2]) into water and oxygen [[101]64] ([102]Supplementary
   Fig. 2A). In accordance with these results, using live-cell brightfield
   microscopy analysis, we observed noticeable cellular shrinkage and
   blebbing only in response to high-dose (5–20 mM) VC treatment
   ([103]Supplementary Figs. 2B and C).

Fig. 1.

   [104]Fig. 1
   [105]Open in a new tab

   The oxidizable form of vitamin C induces cell death in human
   osteosarcoma cells at high doses. (A) Model of the interconversion
   between the reduced (ascorbic acid [AA]) and oxidized (dehydroascorbic
   acid [DHA]) forms of vitamin C (VC). (B–F) Alama blue cell viability
   analysis of (B) U–2OS, (C) 143B, (D) MNNG-HOS, (E) Saos-2, and (F) hFOB
   1.19 cells after 24 h of treatment with different doses (0.25–20 mM) of
   VC (red), ascorbic acid 2-phosphate (AA2P, light green), or DHA
   (brown), as indicated. **p < 0.001, ****p < 0.0001 by one-way ANOVA.
   (G) Dose-response curves of the VC-dependent cytotoxicity in OS cells,
   as indicated. (H-L) Clonogenic survival assay of (H) U–2OS, (I) 143B,
   (J) MNNG-HOS, (K) Saos-2, and (L) hFOB 1.19 cells. The assay was
   conducted after 10 days of incubation of single cells trypsinized and
   seeded at a density of 200 cells/well in a 24-well plate, following 3 h
   of treatment with varying doses (0.5–10 mM) of VC (red), AA2P (light
   green), or DHA (brown), as indicated. See representative images
   associated with these data in [106]Supplementary Fig. 1. **p < 0.001,
   ****p < 0.0001 by one-way ANOVA; ns, non significant. Data are
   representative of three independent experiments. (For interpretation of
   the references to color in this figure legend, the reader is referred