High-dose vitamin C enhances cancer immunotherapy

We asked whether VitC could exert anticancer effects not only in a cancer cell–autonomous manner but also through modulation of antitumor immune responses. To address this, we studied several mouse cancer models including colorectal (CT26 and MC38), breast (TS/A and 4T1), melanoma (B16-F10), and pancreatic (PDAC). To explore the impact of the immune system on cancer growth, tumor volume was monitored in immunocompromised [nonobese diabetic–severe combined immunodeficient (NOD-SCID)] and immunocompetent syngeneic mice. Breast cancer cells were orthotopically injected in the mammary fat pad, whereas the other tumor cell lines were injected subcutaneously. Once the tumors reached around 100 mm³ in volume (typically 5 to 10 days), immunocompetent (Fig. 1A) and immunocompromised animals (Fig. 1B) were randomized to receive either control vehicle or high-dose VitC (4 g/kg per day intraperitoneally). Contrary to humans, mice are capable of synthesizing VitC, and measurements of endogenous VitC in mice resulted in basal plasma concentrations within the range of 0.005 to 0.011 mM, which were unaffected by tumor presence. One hour after VitC dosing, its plasma concentrations raised nearly 1000-fold to over 5 mM (fig. S1). We observed that, in most cases, tumor growth was delayed by the addition of VitC only in the presence of a fully competent immune system (Fig. 1A).

A dose threshold effect of VitC was seen at daily doses including and above 1.5 g/kg intraperitoneally, whereas lower doses were unable to affect growth of TS/A orthotopic breast tumors in immunocompetent mice (fig. S2). The tumor antiproliferative effect of VitC was maintained in the presence of antioxidant N-acetyl cysteine (NAC) at an oral dose of 1.2 g/kg, which abrogated VitC pro-oxidative effects, as indicated by the staining of 8-oxoguanine as a sensor of oxidative stress–induced DNA damage (fig. S3, A and B). The evidence that VitC exerts maximal anticancer therapeutic effects in immunocompetent, but not immunocompromised, mice suggests that VitC antitumor activity is also dependent on some immunomodulatory functions and not only on its pro-oxidant effects.

Immunocompromised NOD-SCID mice have impaired T and B lymphocyte development. T lymphocytes are the main effectors of tumor immune surveillance, and their modulation has therapeutic efficacy (27). We found that T lymphocytes isolated from the spleen of VitC-treated immunocompetent animals, when activated in vitro, produced higher interferon-γ (IFN-γ) concentrations in comparison with control mice (Fig. 2, A and B; see representative flow cytometry plots in fig. S4), suggesting that these lymphocytes may contribute to VitC immunomodulation.

Thus, to address the impact of T lymphocytes in the anticancer effects of VitC, we used two different approaches. On the one hand, we repeated the experiment shown in Fig. 1A in the presence of monoclonal antibodies (mAbs) targeting CD4- or CD8-positive T cells. An isotype antibody served as a control. We observed that anti-CD4 or anti-CD8 antibodies abolished the anticancer effect of VitC in breast TS/A and colorectal CT26 tumors (Fig. 2, C and D).

On the other hand, we adoptively transferred T lymphocytes from immunocompetent to immunocompromised animals. To this purpose, we first injected breast cancer cells (TS/A) in the mammary fat pad of syngeneic mice and, when tumors reached at least 100 mm³ in volume, treated the mice with VitC or control vehicle (Fig. 2E). After 30 days, spleens were explanted, and CD4 and CD8 T cells were isolated (28). In parallel, TS/A cells were implanted in immunocompromised mice. Five and 10 days after TS/A cell implantation, CD4 and CD8 T cells isolated from VitC- or control vehicle–treated immunocompetent mice were injected intravenously into immunocompromised animals (see representative flow cytometry plots of isolated cells in fig. S5). Adoptive transfer of CD4 T cells showed antitumor activity irrespective of previous VitC treatment (Fig. 2F), whereas adoptive transfer of CD8 T cells was able to impair tumor growth only when lymphocytes were isolated from VitC-treated animals (Fig. 2G). Because it is known that CD4 T cells can regulate CD8 T cells, we hypothesized that CD4 T cells could be required to costimulate CD8 T cells in the presence of VitC. To test this hypothesis, we repeated the experiment shown in Fig. 2G by first depleting CD4 T cells in immunocompetent mice, in which we then orthotopically inoculated breast TS/A cells. Once tumors engrafted, mice were treated with VitC and the isotype control antibody (fig. S6). Thirty days after cell injection, spleens were collected and CD8 T cells from vehicle- or VitC-treated mice that had been depleted of CD4 T cells, as well as from animals pretreated with the isotype control antibody, were transferred into immunocompromised NOD-SCID animals bearing TS/A orthotopic tumors. Consistent with results shown in Fig. 2G, CD8 T cells impaired tumor growth only when transferred from VitC-treated animals and not from isotype control mice. However, adoptive transfer of CD8 T cells from VitC-treated donor mice that had been concomitantly depleted of CD4 T cells did not induce antitumor activity in immunocompromised mice (Fig. 2H). Together, these results show that treatment with high-dose VitC delays tumor growth and suggest that this effect depends on T lymphocytes, primarily on CD8 T cells. They also suggest that the presence of CD4 T cells is required to engage the cytotoxic potential of CD8 T cells in VitC-treated model systems.

ICT can unleash the immune system and induce prolonged remissions in several tumors, but its efficacy is still very limited in some of the most prevalent malignancies such as breast and colon cancer (27, 29, 30). Intrigued by the finding that the antitumor activity of VitC was dependent on T cells, we assessed whether VitC could enhance the efficacy of ICT. Immune checkpoint modulators (anti–PD-1 and anti–CTLA-4 mAbs, ICT) alone and in combination were administered to mice bearing syngeneic pancreatic, breast, or colorectal tumors. In pancreatic PDAC and breast 4T1 models, the triple therapy combining VitC with anti–PD-1 and anti–CTLA-4 (VitC + ICT) induced tumor growth impairment compared to single treatments, but without eradicating the tumors (Fig. 3, A and B, and fig. S7). In the second breast cancer model (TS/A), combinatorial VitC + ICT induced prolonged tumor growth impairment (Fig. 3C). A subset of mice (8 of 13) that received the triple therapy rejected TS/A tumors and remained tumor-free for up to a year and eventually died without evidence of cancer, suggesting that the treatment had been curative (Fig. 3D and fig. S8A). In the subset of mice that displayed complete regression, no tumors developed even when they were later rechallenged with the same cancer cells (Fig. 3D), indicating that mice developed complete immune responses and that effective antitumor memory T cells had been expanded.

Treatment with VitC and immunomodulators in the previous experiments were administered when tumors were approximately 100 mm³ in volume. As previously reported by another group (31), CT26 tumors show relatively high mutational burden and are responsive to dual combinatorial CTLA-4 and PD-1 blockade when treatment is administered to animals bearing tumors between 400 and 600 mm³ in volume. To study the impact of combined VitC and ICT on the CT26 colon cancer model, we treated tumors of around 1000 mm³ in volume that are refractory to ICT. Despite the marked disease burden, the triple therapy induced strong tumor impairment and remission in several animals (7 of 13) (Fig. 3E). The subset of mice experiencing a complete response remained tumor-free up to a year, and no tumor developed even when they were later rechallenged with the same cancer cells, indicating effective antitumor immune memory (Fig. 3F and fig. S8B).

In support of the immunomodulatory functions of VitC, immunofluorescence analysis of TS/A breast tumors showed that VitC treatment induced tumor infiltration by both CD4 and CD8 T lymphocytes, which were further increased by combining VitC and ICT (Fig. 4, A and B).

To further characterize the immunological response observed after combining VitC and ICT, we explanted orthotopically grown TS/A breast tumors and isolated infiltrating immune cell fractions from control and treated mice. Flow cytometry analysis revealed that neither the CD45-positive nor the T regulatory cell fractions were modulated in tumors treated with VitC alone (fig. S9). Combined treatment with VitC and ICT induced activation of tumor-infiltrating lymphocytes, as shown by positive staining for the T cell activation marker CD69 and the effector/memory CD44 marker on CD4 and CD8 T lymphocytes (Fig. 4, C and D; see representative flow cytometry plots in figs. S10 and S11).

ICT is approved for the treatment of any tumor type displaying microsatellite instability, which is the result of MMR inactivation. Unfortunately, only a subset of MMRd tumors respond to immune checkpoint modulators, and among those that respond, only a fraction derive long-lasting benefits (3, 7, 8). We wondered whether VitC could improve the magnitude and durability of clinical benefit from immune therapies on MSI tumors. To this end, we used MMRd tumor models that we had previously developed by genetic inactivation of Mlh1 in colorectal and breast cancer murine cells (32). Mlh1 knockout (MLH1-KO) cells have increased mutational burden, augmented number of predicted neoantigens, and higher immunogenicity (32). Because we previously found that MLH1-KO cells grow slower than their parental counterpart in syngeneic mice (32), we initially injected them into immunocompromised animals until large tumors were established. Next, tumor samples were explanted, fragmented, and transplanted into multiple immunocompetent mice that were then randomized to receive either control vehicle or VitC, as shown in Fig. 5A. In parallel, matched Mlh1 wild-type (MLH1-WT) and MLH1-KO colorectal and breast tumors were also treated with VitC in immunocompromised animals. Once again, independently of the MLH1 status (and neoantigen burdens), VitC had no effect on tumor growth in immunocompromised mice (fig. S12). In immunocompetent hosts, the growth of MMRd was slower compared with MMR-proficient tumors, as expected (Fig. 5, B and C) (32). Notably, the effect of VitC alone was much more prominent in MMRd cancers than in their MMR-proficient counterparts (Fig. 5, B and C). These findings suggest that the antitumor effect of VitC is enhanced in mice with tumors harboring increased mutational/neoantigen burdens. The addition of individual immunomodulators (anti–PD-1 or anti–CTLA-4) to VitC improved anticancer responses in mice bearing sizeable MMRd tumors (Fig. 5D and figs. S13 and S14). The combination of VitC with anti–CTLA-4 induced complete tumor regression in most mice (fig. S13), and no relapses were seen for up to a year (Fig. 5E). Notably, after about 1 month of treatment, the effect of VitC and anti–CTLA-4 was comparable to the effect induced by the combination of anti–PD-1 and anti–CTLA-4 mAbs (Fig. 5D). Last, no tumors developed when mice bearing MMRd tumors that achieved complete response on VitC and ICT combination were later rechallenged with the same cancer cells (Fig. 5E). This indicates that these mice had developed protective immunity and immunological memory. Autoimmune reactions can be observed in animals treated with CTLA-4 inhibitors as a consequence of T regulatory cell depletion. Illness, chronic inflammation of ears and eyelids, and reduced motility have been observed during chronic T regulatory cell depletion in mice (33). We found that treatments were generally well tolerated, and no signs of autoimmune reactions were recorded in any of the treatment arms.

The objectives of this study were to assess whether high-dose VitC might exert anticancer activity through the immune system and to verify whether combinatorial treatment with ICT might limit tumor growth in mouse preclinical cancer models. The evidence that a fully competent immune system maximizes the anticancer effects of VitC was demonstrated by administering VitC to immunocompetent and immunocompromised tumor-bearing mice in parallel. We also showed that CD4 and CD8 T cells are the main mediators of the anticancer activity induced by VitC. We demonstrated this by directly depleting the CD4 and CD8 T lymphocytes in immunocompetent mice by administering specific depleting antibodies. The helper and cytotoxic roles of CD4 and CD8 T cells were verified by adoptive transfer experiments of lymphocytes from immunocompetent to immunocompromised mice. The cooperation of high-dose VitC and ICT was demonstrated by administering combinatorial treatments to mice bearing MMR-proficient and MMRd mouse tumors. Immunofluorescence analyses were performed to uncover that the addition of VitC to ICT increases the infiltration of CD4 and CD8 T lymphocytes. In all experiments, control and experimental treatments were randomly administered to age- and sex-matched mice. Tumor burden was monitored over time to assay responses to specific treatments and combinations. Animals were examined for toxicity by periodic observation, and samples were collected. Blinding was not used in this study. However, measurements of tumors were taken before the identification of the cages. The numbers of experimental replicates are indicated in the figure legends. Sample sizes were chosen empirically to ensure adequate statistical power and were in line with the standards for the techniques used in the study.

The TS/A breast cancer cell line was established from a moderately differentiated mammary adenocarcinoma that arose spontaneously in a Balb/c mouse (45). TS/A cells were provided by F. Cavallo (Molecular Biotechnology Center, University of Torino). CT26 is a mouse undifferentiated colon carcinoma, derived from Balb/c mice (46). CT26 cells were purchased from the American Type Culture Collection (ATCC). MC38 is a mouse colon adenocarcinoma line derived from a C57/BL6 mouse, and cells were provided by M. Rescigno (European Institute of Oncology). 4T1 is a spontaneous mammary adenocarcinoma derived from a Balb/c mouse and was purchased from ATCC (47). PDAC cells were isolated from FVB transgenic mice bearing pancreatic cancers with the following genotype: p48^cre, Kras^LSL-G12D, p53^R172H/+, and Ink4a/Arf^flox/+. PDAC cells were provided by D. Hanahan (48) (ISREC, EPFL, Lausanne). B16-F10 is a melanoma cell line derived from a C57/BL6 mouse, purchased from ATCC (49). CT26, MC38, 4T1, and PDAC cells were cultured in RPMI 1640–10% fetal bovine serum (FBS) plus 2 mM glutamine, penicillin (100 IU/ml), and streptomycin (100 μg/ml; Sigma-Aldrich). TS/A and B16-F10 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM)–10% FBS plus 2 mM glutamine, penicillin (100 IU/ml), and streptomycin (100 μg/ml; Sigma-Aldrich). All cell lines were tested for mycoplasma regularly. To ensure that the parental cell models were tumorigenic, before starting the experiments, all the lines were injected into matched syngeneic mice. On tumor formation, we reestablished in vitro cell cultures.

All animal procedures were approved by the Ethical Commission of the Candiolo Cancer Institute and by the Italian Ministry of Health, and they were performed in accordance with institutional guidelines and international law and policies. The number of mice included in the experiments and the inclusion/exclusion criteria were based on institutional guidelines. We observed tumor size limits (maximum allowable diameter of 20 mm) in accordance with institutional guidelines. Six- to 8-week-old female and male C57BL/6J, Balb/c, FVB, and NOD-SCID mice were used according to the approved protocol. Mice were obtained from Charles River. All experiments involved a minimum of five mice per group. Tumor size was measured every 4 days and calculated using the formula V = (d2 × D)/2 (d = minor tumor axis; D = major tumor axis) and reported as tumor volume (mm³; mean ± SEM of individual tumor volumes). Animals were kept under supervision by veterinary personnel throughout the entire duration of the experiments. Mice were checked at least three times a week for signs of illness, inflammation of ears and eyelids, and reduced motility, because these side effects had been previously reported for animals treated with CTLA-4–targeted mAbs (34). The investigators were not blinded; measurements were acquired before the identification of the cages. No statistical methods were used to predetermine sample size.

Ascorbate (Sigma-Aldrich) was prepared weekly by resuspending the powder in sterile water. Ascorbate was administered intraperitoneally 5 days per week at a dosage of 4 g/kg. The anti-mouse PD-1 (clone RMP1-14), anti-mouse CTLA-4 (clone 9H10), anti-mouse CD4 (YTS191), anti-mouse CD8a (YTS169.4), rat IgG2a, and polyclonal Syrian hamster immunoglobulin G (IgG) and rat IgG2b antibodies were purchased from BioXcell. Randomization was used for the experiments in which therapeutic effects had to be evaluated. Animals were treated intraperitoneally with 250 μg of anti–PD-1 antibody per mouse and 200 μg of anti–CTLA-4 antibody per mouse. Treatments were administered at the time points indicated in the graphs after checking for tumor establishment. In combinatorial treatments, VitC was administered starting with the first cycle of immunotherapy. Isotype controls were injected according to the same schedule. Anti-mouse CD4 and CD8a were used for depletion of T cells in immunocompetent mice. Anti-mouse CD4, CD8a, and matched isotype mAbs (400 μg per mouse) were injected intraperitoneally on the day of tumor inoculation. Depleting antibodies were administered (100 μg per mouse) on days 1 and 2 and then every 3 days since tumor cell injection. Depleting antibodies and matched isotypes were administered every 3 days throughout the course of the experiments. Flow cytometry analysis was performed every 3 days to assess the numbers of CD4⁺ and CD8⁺ cells in the bloodstream of mice. The fraction of CD4⁺ or CD8⁺ cells relative to CD45⁺ cells was around 20% before and 0.5% after the administration of depleting antibody. The low fraction of CD4⁺ and CD8⁺ cells (0.5%) was maintained throughout the entire experiment.

Mouse tumors were cut into small pieces, disaggregated with collagenase (1.5 mg/ml), and filtered through 70-μm strainers. Cells were stained with specific antibodies and Zombie Violet Fixable Viability Kit (BioLegend). Phenotype analysis was performed with the following antibodies purchased from BioLegend: anti-CD45–PerCp (peridinin chlorophyll protein) (30F11), anti-CD11b–APC (allophycocyanin) (M1/70), anti-CD3–PE (phycoerythrin)/Cy7 (17A2), anti-CD4–FITC (fluorescein isothiocyanate) (RM4-5), anti-CD8–PE or FITC (YTS156.7.7), anti-F4/80–APC (BM8), anti-CD49b–PE (DX5), anti-CD44–APC (IM7), anti-CD69–PE (H1.2F3), anti-CD62L–PE/Cy7 (MEL-14), anti-CD11c–FITC (N418), anti-CD28–PE (37.51), anti-CD25–APC (PC61), anti-CD127–PE/Cy7 (A7R34), and anti-FoxP3–PE (MF-14). For FoxP3 staining, cells were isolated and stained with surface antibodies for 30 min and then fixed and permeabilized using the FoxP3 Fix/Perm Buffer set (BioLegend). Cells were then stained with FoxP3-PE (BioLegend). For IFN-γ staining, cells were stimulated in vitro with the cell stimulation cocktail (eBioscience) and incubated with GolgiStop and GolgiPlug (BD Biosciences). After 6 hours of incubation, cells were washed and stained for extracellular markers. Then, cell permeabilization was performed by using the Cytofix/Cytoperm kit (BD Biosciences), and then the cells were stained for IFN-γ (XMG1.2, BioLegend). All flow cytometry was performed using the FACS Dako instrument and FlowJo software.

Detection of T cells was performed with a modification of the method for immunofluorescence of fresh frozen tissues described previously (32). In brief, tumor samples were included in Killik (Bio-Optica), serially cut (10 μm), and fixed using cold acetone:methanol (1:1). Samples were incubated for 1 hour in blocking buffer [1% bovine serum albumin and 2% goat serum in phosphate-buffered saline (PBS) with 0.05% Tween and 0.1% Triton X-100] and incubated overnight with anti-CD8 (clone YTS169 from Thermo Fisher) and anti-CD4 (clone RM4-5 from Thermo Fisher). For detection, anti-rat Alexa Fluor 647 was used (Thermo Fisher Scientific). Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Slides were then mounted using fluorescence mounting medium (Dako) and analyzed using a confocal laser scanning microscope (TCS SPE II; Leica).

The extraction procedure of ascorbic acid from plasma was carried out as previously reported (50), whereas the analytical part was developed on the basis of mass spectrometry technology, as recently reported (22, 23). Heparinized plasma samples, previously supplemented with 10% of metaphosphoric acid, were defrosted, vortexed, and centrifuged at 13,200 rpm. Ninety-five microliters of the supernatant was supplemented with 5 μl of solution (190 μg/ml) of ¹³C₆-l-ascorbic acid as internal standard (IS), dissolved in acetonitrile:0.1% formic acid (70:30), then combined with 900 μl of acetonitrile:0.1% formic acid (70:30), and vortexed for 1 min. After centrifugation at 13,200 rpm for 10 min at 4°C, 200 μl was transferred to microvials and 5 μl was injected into liquid chromatography–tandem mass spectrometry (LC-MS/MS) instrumentation consisting of an LC system Series 200 autosampler and micropump (Perkin Elmer) coupled to a triple quadrupole mass spectrometer API 4000 (SCIEX). Chromatographic separation was achieved on an Atlantis column T3 (2.1 mm × 150 mm, 3 μm) (Waters) fluxing mobile phase at a flow rate of 0.2 ml/min under gradient conditions. The mass spectrometer worked with electrospray ionization in negative ion mode and selected reaction monitoring, quantifying target ions mass/charge ratio (m/z) 175/115 for ascorbic acid and m/z 181/119 for IS. The limit of quantification was 0.0055 mM; on the day of analysis, a plasma standard calibration curve was prepared in the range of 0.0055 to 5.7 mM. Samples with concentration above 5.7 mM were reanalyzed (diluted 1:1).

Mice were euthanized, and their splenocytes were isolated as previously described (28). Briefly, spleens were minced and passed through a 70-μm cell strainer. Afterward, red blood cells were lysed with ACK lysis buffer (Gibco) and the remaining splenocytes were washed with magnetic-activated cell sorting (MACS) buffer. Magnetic bead sorting, using negative selection kit (Miltenyi), was used to acquire CD4⁺ and CD8⁺ T cells. The purity of the enriched cells was greater than 94%. Cells were dissolved in 100 μl of PBS and intravenously injected in an orthotopic model of breast cancer. Mice were injected twice with 5 million T cells by tail vein injection at days 5 and 10 since cancer cell injection.

In experiments where antioxidants were administered, NAC was administered by oral gavage (1.2 g/kg in PBS, pH 7.2) as previously described (51–53). To check antioxidant effects on tumors, 8-oxoguanine (Abcam, N45.1) staining was performed by immunohistochemistry on formalin-fixed paraffin-embedded (FFPE) sections (54).

Statistical analyses were performed using GraphPad Prism software. To determine statistical significance for tumor growth curves, normality and lognormality tests were performed for each experiment. In the case of a Gaussian-like distribution, Student’s t test for two-group comparison (P values were adjusted with Welch correction) and one-way analysis of variance (ANOVA) for more than two-group comparison (P values were adjusted with Tukey correction) were performed. In case of a non-Gaussian distribution, nonparametric tests were performed (P values were adjusted with Welch correction). For immunophenotypic analysis, normality and lognormality tests were performed. Statistical significance was calculated using one-way ANOVA (P values were adjusted with Tukey correction) in case of Gaussian-like distribution. Nonparametric analyses (P values adjusted with Welch correction) were conducted for datasets that failed to pass a normality test. The Kaplan-Meier method was used for survival analysis, and P values were calculated using the log-rank test (Mantel-Cox). All data are presented as mean ± SEM. Sample sizes were chosen to provide adequate power on the basis of our previous studies and literature surveys. The number of replicates and sample size for in vivo experiments were limited according to the requirements of the Italian Ministry of Health. Animal studies were performed in accordance with institutional guidelines and international law and policies. When therapy was applied, we performed randomization. In this case, only mice bearing tumors with a volume within 50% of the average size were included in the experiment. Original data are provided in data files S1 to S10.