Background
In the Western world, B-cell chronic lymphocytic leukemia (CLL) is the most common adult leukemia. It is characterized by the accumulation of CD5+ B lymphocytes in the blood, bone marrow, and secondary lymphoid tissues [
1]. Until recently, the first-line therapy proposed to all patients was a combination of fludarabine-cyclophosphamide with CD20-specific immunochemotherapy (the antibody rituximab). The recent introduction of drugs targeting B-cell receptor signaling (such as the BTK inhibitor ibrutinib, the PI3K inhibitor idelalisib, and the BCL2 inhibitor venetoclax) has improved patient outcomes [
2]. However, these drugs are rendered less effective over time by the emergence of resistance through (i) acquired somatic mutations in the genes coding for BTK, PLCG2 and BCL2, (ii) increased expression of anti-apoptotic genes (
BCL2,
BCL2L1,
MCL1, etc.), and (iii) the CLL cells’ interaction with microenvironment [
2]. Accordingly, there is still a need for novel, less toxic, cost-effective and safe treatment strategies for CLL.
High-dose vitamin C (i.e. L-ascorbic acid (AA)) was suggested as a potential anticancer agent for the first time in the 1970s by Pauling and Cameron [
3]. More recently, preclinical data have confirmed the anticancer efficacy of AA and its selective cytotoxicity in different human cancers, both in vitro and in vivo [
4‐
6]. Although several in vitro preclinical studies have shown that pharmacologically achievable concentrations of ascorbate have cytotoxic effects on cancer cells [
4‐
6], AA has shown limited efficacy in some clinical trials [
7,
8] but good efficacy in others [
9]. In hematological malignancies, several studies have shown that AA is toxic for leukemic cells [
10,
11] but does less damage to healthy cells [
12,
13]. The major mechanism underlying AA’s anticancer activity is pro-oxidant damage through auto-oxidation. This leads to the generation of cytotoxic hydrogen peroxide (H
2O
2, i.e. a reactive oxygen species (ROS)) [
5,
12,
14]. High concentrations of ROS are cytotoxic, via damage to DNA and mitochondria and the activation of apoptotic pathways [
15].
The use of intravenous or oral administration route for AA has led to the impression that the compound’s anticancer effect is “controversial” because the route affects the maximum achievable concentration in plasma [
16]. Earlier studies suggested that AA is cytotoxic at millimolar concentrations, which are achievable by intravenous injection but not by oral administration [
4,
16]. However, revised interpretations and new knowledge about the pharmacokinetic properties of AA showed that a large oral dose can result in plasma concentrations of around 200 μM [
4,
16‐
18]. Furthermore, several pharmaceutical formulations of vitamin C (e.g. liposomal encapsulations) can achieve levels of up to 400 μM [
18,
19]. These new pharmacokinetic data and the results of previous clinical studies of orally administered vitamin C [
3,
20] prompted us to investigate the in vitro effect of orally achievable concentrations of AA on CLL B-cells.
In CLL, little is known about the molecular mechanisms by which AA induces cytotoxicity, its interaction with the CLL microenvironment, and AA’s influence on the effectiveness of chemotherapy and targeted therapies. The few preclinical studies to have investigated AA’s effect on CLL B-cells showed that high-dose of AA induces cytotoxicity in CLL B-cells [
21,
22]. Moreover, some CLL patients suffer from vitamin C deficiency (hypovitaminosis C), which is correlated with more aggressive disease [
23]. Given that CLL B-cells are known to be sensitive to oxidative stress mediated by H
2O
2 [
24‐
27], we hypothesized that a redox-inducing agent like AA might effectively kill leukemic cells and synergize with CLL treatments. Here, we performed a comprehensive study of the in vitro effect of 250 μM AA on primary highly purified CLL B-cells and two cell lines. We investigated the intrinsic and extrinsic mechanisms that lead to AA-cytotoxicity and resistance, and we evaluated the effect of combining AA with a panel of FDA-approved drugs. Overall, our study provides detailed mechanistic insights into AA’s action on CLL B-cells and identified drug combination strategies that might enhance the efficacy of treating CLL B-cells.
Methods
Patients
Chronic lymphocytic leukemia B-cells were isolated from the peripheral blood of 40 treatment-naïve patients diagnosed according to international guidelines (Binet stage A) (Table S
1). Only treatment-naïve Binet stage A patients were included in our study, given that treatment might have altered their B-cells’ response to ascorbic acid. Normal B-cells were isolated from donor lymphocyte infusions provided by age-matched healthy volunteers. Patients and healthy volunteers provided their written informed consent to participate to the study. The study was performed in accordance with the principles expressed in the Declaration of Helsinki. This study was conducted in compliance with French legislation on non-interventional studies.
Reagents
L-Ascorbic acid, dehydroascorbic acid (DHA), AA 2-phosphate (Asc-2P), catalase from human erythrocytes, deferoxamine (DFX), oligomycin A, and metformin were purchased from Sigma Aldrich. Sodium pyruvate (SP) was purchased from Thermo Fisher Scientific. Venetoclax (ABT-199), ibrutinib, idelalisib, fludarabine, and cyclophosphamide were purchased from Selleckchem. CPI-613 was purchased from Abcam.
Cell isolation and cell culture
Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density gradient centrifugation. CD19+ CD5+ B-cells were isolated from PBMCs using magnetic-bead-activated cell sorting, using a B-CLL cell isolation kit (Miltenyi Biotec). To evaluate the effect of AA on normal lymphocytes, naïve B-cells were isolated from donor lymphocyte infusions using a Naïve B-cell Isolation kit (Miltenyi Biotech). The purity assessed by CD19 expression on flow cytometry was around 98%. OSU-CLL cells were a gift from E. Hertlein and colleagues [
28]. JVM3 cells were purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ). Freshly isolated B-cells and CLL cell lines were cultured in RPMI 1640 medium (PAN Biotech, #P04–16500) with 10% fetal bovine serum (FBS) (PAN Biotech, #P30–3306) and L-glutamine, penicillin and streptomycin (1%) (Eurobio Scientific). When indicated, CLL B-cells were cultured in Iscove’s modified Dulbecco’s Medium (IMDM) (Merck, #FG0465) and alpha-MEM medium (Sigma Aldrich, #M4526) (
n = 7). Cells were cultured at a density of 4 × 10
5/ml in 48-well plates and were treated with either vehicle or AA or drugs for 24 h. To simulate hypoxia condition, cells were cultured in presence of 100 μM Cobalt(II) chloride hexahydrate (CoCl
2.6H
2O, Sigma Aldrich) for 24 h. Cells were then washed and incubated with AA for 24 h.
Cell viability assay
Cell viability and apoptosis were assessed using annexin V-APC/ 7-AAD staining (BD Biosciences). Fluorescence intensity was measured in a MACSQuant Analyzer (Miltenyi Biotech). Data were analyzed using Flow Jo software (version 10, Tree Star, Inc.). Cell viability of treated cells was normalized to vehicle condition (i.e. cell viability of vehicle treated cells for each patient is set to 100% and all data were indicated relative to this value). In some experiment (when indicated, ex. in drugs combination experiments), cell viability was assessed using CellTiter-Glo Luminescent Cell Viability Assay Kit (Promega).
Co-culture conditions with primary human bone marrow mesenchymal stem cells (MSCs)
MSCs were isolated from healthy donor bone marrow, as described by Naudot et al. [
29]. MSCs were seeded at 2 × 10
4 cells/ml in 24-well plates (Falcon) in alpha minimum essential medium (MEM) supplemented with 10% FBS, penicillin/streptomycin (1%), L-glutamine (1%) and 0.5 ng/ml basic fibroblast growth factor (bFGF) and incubated overnight to allow cells adhesion. Freshly isolated CLL B-cells were cultured alone or on MSCs at 4x10
5cell/ml (ratio: 20:1) in RPMI medium. CLL cells were co-cultured with MSCs for 6 h (
n = 12) or 24 h (
n = 6) prior to AA treatment (250 μM). After 24 h, CLL cells were carefully removed and cell viability was assessed as described above.
CLL B-cells were stimulated with CpG-ODN2006 (1.5 μg/ml) (Invivogen), CD40L (50 ng/ml) + IL-4 (50 ng/ml) (Miltenyi) or anti-IgM antibody (10 μg/ml) (Jackson ImmunoResearch) (
n = 7) or cultured in the presence of a combination of cytokines (as described in [
30]) (
n = 6) or in presence of 10% of the autologous patient’ serum (
n = 10) and treated with AA for 24 h before cell viability/apoptosis was assessed in an annexin V-APC/7-AAD flow cytometry assay.
Detection of ROS
OSU-CLL and JVM3 cell lines (2 × 105 cells) were treated with either vehicle or AA (250 μM) for 6 h and incubated with MitoSox™ (Thermo Fisher Scientific), a mitochondrial superoxide indicator, at 5 μM for 10 min at 37 °C (n = 7) according to the manufacturer’s user guide. Cells were then washed and fluorescence was analyzed with MACSQuant Analyzer (Miltenyi Biotech). Cells treated with H2O2 (50 μM) and AA (1 mM) were used as positive control. The fold change of mean fluorescence intensity (MFI) was then calculated in treated cells relative to controls.
Extracellular H2O2 (i.e. H2O2 in the medium) was measured after treatment with different concentrations of AA and at different time points in the presence or absence of catalase and SP, using the Pierce™ Quantitative Peroxide Assay Kit (Aqueous) (Thermo Fisher Scientific) according to the manufacturer’s user guide (n = 6). Absorbance was read in a GloMax® Discover Microplate Reader (Promega).
Glutathione levels
Levels of total cellular glutathione (GSH), oxidized form glutathione (GSSG) and the GSH/GSSG ratio were measured using the GSH/GSSG-Glo™ assay (Promega) in the OSU-CLL and JVM3 cell lines 2 h after treatment with 250 μM AA (n = 3). Luminescence was read using a GloMax® Discover Microplate Reader (Promega). Data were presented as the GSH/GSSG ratio.
mRNA extraction and gene expression analysis
Highly purified CLL B-cells, healthy donor B-cells (HD B-cells) and OSU-CLL and JVM3 cell lines were washed in PBS. mRNA was extracted using RNeasy Mini Kit (Qiagen), and 1 μg was reverse-transcribed using a High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). The relative mRNA expression of catalase was analyzed in a qPCR assay with TaqMan Universal PCR Master Mix (Thermo Fisher Scientific). The housekeeping genes beta-actin and GADPH were used as endogenous controls in the expression analyses. All PCR reactions were performed in triplicate. The TaqMan Gene Expression assays for catalase (Assay ID Hs00156308_m1), GAPDH (Hs02786624_g1), and β-actin (Hs01060665_g1) were purchased from Thermo Fisher Scientific.
Catalase knockdown with siRNA
MISSION® esiRNA human CAT and Control SiRNA (esiRNA targeting Renilla luciferase (RLUC)) were purchased from SIGMA Aldrich. 1 × 106 JVM3 cells were transfected with CAT siRNA or Ctrl SiRNA (188 nM) using Amaxa® Cell Line Nucleofector® Kit V (Lonza, Germany), program T-016. Transfected JVM3 cells were cultured in 12-well plate (1 × 106 cells/well) under standard culture conditions. Changes in catalase expression in JVM3 transfected cells were analyzed at 48 and 72 h post-transfection by western blot. At 72 h, the cells were then treated with ascorbic acid at different doses (250, 500 and 1000 μM) and were analyzed for cell viability by the CellTiter-Glo Luminescent Cell Viability Assay Kit (Promega).
Western blots
The cells were washed with PBS and lysed in RIPA buffer (Sigma). Cell lysates were centrifuged at 14000 rpm for 5 min, and supernatants were collected. After determination of the protein content in a BCA assay (Thermo Fisher Scientific), 50 μg of protein were separated using 10% SDS-PAGE and was transferred onto nitrocellulose membranes (Thermo Fisher Scientific). The membranes were incubated overnight at 4 °C with antibodies against cleaved and total poly-ADP-ribose polymerase (PARP, #9542; 1:1000, Cell Signaling Technology), catalase (#sc-271,803; 1:100, Santa Cruz Biotechnology), cleaved caspase-3 (#9664; 1:1000, Cell Signaling Technology), cleaved caspase-8 (#9496; 1/1000 Cell Signaling Technology), cleaved caspase-9 (#7237; 1/1000, Cell Signaling Technology), HIF-1α (#sc-13,515; 1:200, Santa Cruz Biotechnology) or β-actin (#sc-47,778; 1:500, Santa Cruz). Blots were then washed with TBS-buffer with 0.2% Tween and incubated with secondary antibodies against rabbit (Thermo Fisher Scientific), mouse (Sigma) or goat (Santa Cruz) antibodies (1:2500). Blots were developed using SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific).
Drugs combination with AA study and synergism determination
Synergism was evaluated with the Chou-Talalay Combination Index (CI) using the experimental design as recommended by Chou TC [
31]. Drugs were serially diluted in culture media and then cells were added to the media (in two replicates) and incubated for 24 h. Cells were seeded in 96-well plates at 1 × 10
5 cells per well. A constant molar ratio combination for drugs based on lethal concentration 50 (LC50) values was used. The cytotoxicity of these drugs or combinations was assessed by CellTiter-Glo assay (Promega). The combination indexes (CIs) and fraction affected (Fa) based on the Chou-Talalay method using CompuSyn software (CompuSyn Inc. Paramus, NJ, USA). CI values < 1 were considered as synergistic [
31]. For the data in Fig.
7b, synergism was determined using the coefficient of drug interaction (CDI) [
32,
33], which was calculated as CDI = AB/(A × B), where AB is the ratio of the two-drug combination group to the control group, and A or B is the ratio of the single-drug group to the control group. CDI values < 0.7 were considered as synergistic.
Statistical analysis
Data were expressed as the mean ± standard error of the mean (SEM). The data are expressed as a percentage with respect to that of vehicle-treated cells (control), which was set to 100%. However, statistical analyses were done with absolute viability data. All statistical analyses were performed with GraphPad Prism® software (version 5.0; GraphPad Software Inc., San Diego, CA, USA). Statistical significance was assessed in a one-way analysis of variance. A two-sided paired t-test was used to detect significant differences between groups. The threshold for statistical significance difference was set to p < 0.05 (* p values < 0.05; ** p values < 0.01 and ***p values < 0.001). No blinding and no randomization of samples were applied.
Discussion
Despite major progress in treatment, CLL is still an incurable disease. We wondered whether oral supplementation with AA (vitamin C) would have an impact on CLL B-cell survival. To study this effect in vitro, we used recent knowledge from pharmacokinetic studies showing that oral administration of AA can achieve a plasma concentration of around 250 μM [
16,
18]. Using this concentration, we (i) provided mechanistic insights into the cytotoxic effect of AA on CLL B-cells, (ii) investigated the role of the CLL microenvironment in resistance to AA’s effect, and (iii) suggested new therapeutic strategies. We showed that 250 μM AA selectively induced cell death in primary leukemic CLL B-cells by acting as a pro-oxidant and thus by leading to the release of H
2O
2 into the extracellular medium. These results are in compliance with previous observations in other cancers [
5,
12,
14]. The cytotoxic effect of AA via H
2O
2 generation was confirmed using an H
2O
2 formation inhibitor (the iron chelator DFX) and two H
2O
2 scavengers (catalase and SP). These molecules completely reversed AA’s cytotoxicity toward CLL B-cells. Knowing that H
2O
2 can enter the cell through passive diffusion [
4], we observed high intracellular and mitochondrial ROS levels and low GSH/GSSG ratios in AA-treated CLL cells vs. the control - confirming the presence of a redox alteration in the treated cells. On the molecular level, we showed that AA treatment induces cleavage of caspase-8 but not caspase-9; this suggests that apoptosis was extrinsically mediated and not mitochondrial [
50,
51]. Along with the superoxide anion (O
2•−) and the hydroxyl radical (
•OH), H
2O
2 is a ROS. The literature data show that ROS (including H
2O
2) mediate apoptosis through increasing the level of oxidative stress in cancer cells [
14,
15,
52]. Indeed, targeting intracellular redox homeostasis by increasing ROS levels in cancer cells is a promising treatment approach [
15]. CLL B-cells were shown to produce abnormally large amounts of ROS and to have impaired antioxidant defenses [
24,
27,
53]. Hence, these cells are vulnerable to molecules that perturb redox homeostasis [
46,
49].
Catalase is an anti-oxidant enzyme that protects normal cells against AA-mediated oxidative stress by degrading the generated H
2O
2 [
26,
34,
54]. Consequently, normal cells remove AA-generated H
2O
2 faster than tumor cells do [
55]. As shown here and by others [
56], CLL B-cells express lower levels of catalase than normal B-cells do. Nevertheless, we found that CLL B-cells from 7 of 40 patients (17.5%) were less sensitive to AA treatment; all these cases showed high levels of catalase expression. Importantly, it has been shown that CLL B-cells expressing high levels of catalase lead to a more aggressive disease [
56]. The role of catalase was further emphasized by the observed AA-resistance of the catalase-expressing JVM3 cell line; in contrast, the OSU-CLL cell line (which does not express catalase) was sensitive to AA. The knockdown of catalase expression in JMV3 cells sensitized these cells to the cytotoxic effect of AA indicating that catalase plays a role in the resistance to AA.
Recent studies have shown that the CLL microenvironment (including bone marrow MSCs) not only provides survival cues [
30,
57] but also protects against oxidative stress by modulating the expression of genes involved in redox homeostasis and by increasing glutathione synthesis [
24,
45,
46]. However, our present results show that AA was able to induce CLL B-cell death and thus overcome the anti-apoptotic protection provided by primary bone marrow MSCs. Furthermore, AA was also able to counter the survival support provided by another microenvironment cue from T cells (modeled here by incubation with CD40L and IL4). Moreover, hypoxia is shown to provoke resistance toward anticancer drugs. In CLL, malignant cells recirculate from normoxic peripheral blood to hypoxic tissues like lymph nodes and bone marrow, where hypoxia was shown to provide survival advantage [
46,
48]. We observed that chemically induced hypoxia by CoCl
2 provided survival advantages to CLL cells in presence of 250 μM AA. These could be explained by the intracellular pyruvate accumulation caused by the fact that under hypoxic environment CLL cells shift from oxidative phosphorylation to glycolysis as a source of energy [
46,
48]. Nevertheless, we were surprised to find that autologous serum protected against (but did not completely abolish) the cytotoxic effect of 250 μM AA. Furthermore, a still higher concentration of AA (500 μM) induced significant cell death in the presence of autologous serum. We have previously reported that culture of CLL B-cells with 10% autologous serum protected against apoptosis [
30]. However, in the context of AA treatment, this protective effect can be explained by the presence of H
2O
2 scavengers (such as catalase and pyruvate) in the human serum [
27,
36,
40,
47]. Indeed, plasma catalase activity was higher in CLL patients than in healthy subjects [
27] and was correlated with the disease’s aggressiveness. Here, we also showed that SP protects CLL B-cells from the cytotoxic effect of AA. Furthermore, SP has been demonstrated to directly protect CLL cells against oxidative stress and to increase cell viability after H
2O
2 treatment [
48]. We confirmed this effect by showing that the concentration of AA-generated H
2O
2 in the medium falls in the presence of SP. The pyruvate concentration in human plasma is around 100 μM, and this concentration increases after glucose uptake because pyruvate is a major product of glycolysis [
35,
58,
59]. Furthermore, glucose was shown to inhibit intestinal vitamin C transport ex vivo [
60], and blood glucose levels may interfere with the uptake of ascorbate by human neutrophils [
61]. These observations suggest that considering adequate dietary patterns is critical when delivering AA supplements orally. Hence, AA-induced cytotoxicity might be enhanced by a glucose-restricted diet that could increase intestinal AA uptake and reduce plasma pyruvate concentrations. On the same lines, recent data showed a synergistic effect between fasting-mimicking diet and vitamin C in KRAS mutated colorectal cancer [
6]. Moreover, our data showed that the culture medium influenced the CLL B-cells’ response to AA treatment. The cytotoxic effect of AA observed in basic RPMI 1640 medium was completely absent when cells were cultured in IMDM and alpha-MEM – both of which contain pyruvate as built-in component. This observation underlines the critical role of the cell culture medium in the cancer cells’ response to AA and may account for the conflicting results in the literature [
7,
8]. Importantly, our results suggest that publications about AA’s effect in vitro should always specify the exact reference of the cell culture medium and/or the exact compound added to the medium.
Given that previous studies generated conflicting findings on the influence of vitamin C on cancer treatments [
8], we investigated the compound’s influence on currently available drugs for CLL and hematological malignancies and drugs in the development pipeline. We observed that AA synergistically potentiates the cytotoxicity of ibrutinib, idelalisib, and venetoclax in primary CLL B-cells. Potentiation was also observed for drug candidates like CPI-613 (an alpha-ketoglutarate dehydrogenase inhibitor), the ATP synthase inhibitor oligomycin A, and metformin.
In CLL cells, perturbations in oxidative metabolism result in elevated levels of ROS; this is associated with a favorable prognosis and slower disease progression [
62]. The generation of oxidative stress might be useful for treating cancer directly or for enhancing sensitivity to other cancer drugs. It has been reported that BH3 mimetics can displace Bcl-2-bound glutathione, which thus inhibits the transport of glutathione into mitochondria and makes the cell more vulnerable to oxidative stress [
63‐
65]. In line with these previous observations, we observed synergistic CLL cell killing by a combination of venetoclax and AA. Mechanistically, the data suggest that this synergistic effect is linked to downregulation of
MCL1 expression by the two treatments. MCL-1 is an anti-apoptotic protein involved in resistance to venetoclax and ibrutinib. Trachootham et al. [
66] have shown that ROS decrease the expression of MCL-1 in CLL cells by inhibiting its glutathionylation. Therefore, the decrease in MCL-1 expression associated with AA–induced ROS favors the use of a combination therapy with venetoclax and AA. This finding might be of value in designing rational new treatment regimens by combining venetoclax with inducers of oxidative stress. Similarly, PI3K inhibition has been linked to increased oxidative stress in CLL cells through the inactivation of NRF2 [
67]. This effect might combine with ROS to target MCL-1 because the protein is more stable after phosphorylation by AKT [
68]. This might explain the synergistic cytotoxicity of idelalisib and AA for CLL B-cells. Furthermore, in addition to synergistic cytotoxic effect of ibrutinib/idelalisib with AA on CLL cells; ibrutinib and idelalisib induces the mobilization of leukemic cells from their protective tissue microenvironment to the blood circulation [
2], leading to the loss of this protective effect, and CLL cells eventually becoming more susceptible to cell death by AA.
Given that metabolic activity in CLL cancer cells results in an altered redox state [
69,
70], we decided to study the combination of ROS-inducing agent (i.e. AA) with drugs that targeting metabolic pathways such as the inhibitor of the tricarboxylic acid cycle CPI-613, the ATP synthase inhibitor oligomycin A and the electron transport chain complex I inhibitor metformin that target mitochondrial metabolism. CPI-613 and metformin are currently in clinical testing for hematologic malignancies including CLL [
69,
71]. Furthermore, CPI-613, oligomycin A and metformin showed synergistic effects with AA in killing CLL B-cells; hence, the combination of AA with drugs targeting mitochondrial metabolism might be a promising approach in CLL treatment.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.