Antioxidants are widely used to protect cells from damage induced by reactive oxygen species (ROS). The concept that antioxidants can help fight cancer is deeply rooted in the general population, promoted by the food supplement industry, and supported by some scientific studies. However, clinical trials have reported inconsistent results. We show that supplementing the diet with the antioxidants N-acetylcysteine (NAC) and vitamin E markedly increases tumor progression and reduces survival in mouse models of B-RAF– and K-RAS–induced lung cancer. RNA sequencing revealed that NAC and vitamin E, which are structurally unrelated, produce highly coordinated changes in tumor transcriptome profiles, dominated by reduced expression of endogenous antioxidant genes. NAC and vitamin E increase tumor cell proliferation by reducing ROS, DNA damage, and p53 expression in mouse and human lung tumor cells. Inactivation of p53 increases tumor growth to a similar degree as antioxidants and abolishes the antioxidant effect. Thus, antioxidants accelerate tumor growth by disrupting the ROS-p53 axis. Because somatic mutations in p53 occur late in tumor progression, antioxidants may accelerate the growth of early tumors or precancerous lesions in high-risk populations such as smokers and patients with chronic obstructive pulmonary disease who receive NAC to relieve mucus production.

Antioxidants including vitamins, carotenes, and minerals are found naturally in the diet and are added to food, cosmetic products, and pharmaceuticals. Antioxidants act as electron donors that neutralize reactive oxygen species (ROS) and other free radicals that may otherwise damage DNA and promote tumorigenesis ( 1 ). Consequently, popular wisdom—supported by numerous cellular and preclinical studies—holds that antioxidants protect against cancer ( 2 – 4 ). However, large randomized clinical trials have produced inconsistent results, and some studies indicate that antioxidants may even increase cancer risk ( 5 – 10 ). Moreover, recent genomic analyses of lung cancers have shown a high frequency of mutations in genes that activate an endogenous antioxidant program, suggesting that decreasing the amounts of ROS promotes tumor growth ( 11 , 12 ). Consistent with this notion, experimental studies show that oncogenes such as K-RAS and B-RAF promote tumor growth by stimulating NRF2-mediated transcription of endogenous antioxidant genes ( 13 , 14 ). Despite the striking discordance between the use of antioxidants and the lack of experimental support for their anticancer properties, no studies have yet examined their impact on tumor growth in state-of-the-art mouse models of cancer, including lung cancer—the most common form in humans ( 15 ).

( A and B ) Proliferation of human lung cancer cell lines expressing wild-type (A) or mutant (B) p53 in medium supplemented with 1 mM NAC or 100 μM Trolox. Values are the means of triplicate analyses per cell line. ( C ) Amounts of ROS in human lung cancer cell lines incubated for 4 days in medium supplemented with antioxidants, as judged by FACS analyses of DCF fluorescence (n = 4 analyses per cell line). ( D ) Western blots of lysates of cancer cell lines (wild-type p53) with antibodies against total and phosphorylated forms of p53 and phosphorylated H2AX (γH2AX). β-Tubulin was used as a loading control. ( E ) Western blots of lysates of cancer cell lines (mutant p53) with antibodies against p53. β-Tubulin was used as a loading control. Note that H358 and H1299 are p53-deficient and that the amount of p53 in H23 is not reduced by antioxidant treatment. ( F ) Proliferation of cell lines from the experiment in (A), incubated with lentiviruses expressing shRNAs targeting TP53 or containing a scrambled (SCR) sequence, in medium supplemented with NAC or Trolox. Values are the means of triplicate analyses per cell line and experiments with two different shTP53 lentiviral clones. *P < 0.05, **P < 0.01, ***P < 0.001. Exact P values are provided in table S2. Graphical data are presented as means ± SEM.

We next tested the impact of antioxidants on human lung cancer cell lines. NAC and Trolox increased the proliferation of cells expressing wild-type p53, but not that of cell lines with p53 mutations ( Fig. 4 , A and B). In wild-type p53 cell lines, the antioxidants reduced ROS, increased BrdU incorporation and the percentage of cells in S phase, had no impact on apoptosis, and reduced DNA damage and p53 expression ( Fig. 4 , C and D, fig. S7, A to C, and table S1). p53 in a cell line expressing a mutant form of the protein was not reduced by antioxidants ( Fig. 4E ). Knockdown of p53 expression with lentiviral short hairpin RNAs (shRNAs) in wild-type p53 cell lines abolished the antioxidant effect ( Fig. 4F , fig. S8, and table S1). Thus, human and mouse tumor cells respond similarly to antioxidants.

The tumor suppressor p53 regulates cell proliferation and is activated by ROS and DNA damage. We tested the possibility that reduced ROS and DNA damage in response to antioxidants would affect p53. Indeed, NAC and vitamin E markedly reduced the amounts of p53, as judged by Western blots of fibroblast and tumor lysates ( Fig. 3 , C and D). The amounts of phosphorylated H2AX Ser139 and ATM Ser1981 , markers of DNA damage, were reduced in lysates of antioxidant-treated cells and tumors ( Fig. 3 , C and D), consistent with the reduced amounts of 8-oxoguanine in tumors ( Fig. 2B , fig. S3B, and table S1). To determine whether p53 is required for the antioxidant effect, we bred Kras2 LSL/+ and Braf CA/+ mice on a background of a conditional p53 knockout allele (Trp53 fl/fl ) and used Cre to simultaneously activate oncogene expression and inactivate p53. Inactivation of p53 abolished the ability of antioxidants to increase the proliferation of oncogene-expressing fibroblasts in vitro and tumor cells in vivo ( Fig. 3 , E and F, table S1, and fig. S6D).

( A ) Proliferation of primary K-RAS G12D – (left) and B-RAF V600E –expressing (right) fibroblasts in medium supplemented with 250 μM (low) and 1 mM (high) NAC or 25 μM (low) and 100 μM (high) Trolox. Values are the mean proliferation of fibroblasts from three embryos per genotype assayed in triplicate. ( B ) The amounts of ROS in fibroblasts incubated for 4 and 12 days in medium supplemented with antioxidants, as judged by fluorescence-activated cell sorting (FACS) analyses of DCF fluorescence (n = 3 cell lines per genotype). ( C and D ) Western blots of fibroblast (C) and tumor (D) lysates with antibodies against total and phosphorylated forms of p53 and phosphorylated H2AX (γH2AX) and ATM Ser1981 . Actin and β-tubulin were used as loading controls. ( E ) Proliferation of primary p53-deficient K-RAS G12D – (left) and B-RAF V600E –expressing (right) fibroblasts in medium supplemented with 1 mM NAC or 100 μM Trolox. The cells were generated by incubating Kras2 LSL/+ Trp53 fl/fl and Braf CA/+ Trp53 fl/fl fibroblasts with Cre adenovirus. Control cells, designated as WT, were the parental cells incubated with a βgal adenovirus. Values are the mean proliferation of fibroblasts from three embryos per genotype assayed in triplicate. ( F ) Tumor burden in lungs of NAC-treated and littermate control conditional p53-deficient K-RAS G12D (left) and B-RAF V600E (right) mice 10 weeks after inhalation of Cre adenovirus. Numbers in bars = n. *P < 0.05, **P < 0.01, ***P < 0.001. Exact P values are provided in table S2. Graphical data are presented as means ± SEM.

To determine whether NAC and vitamin E increase the proliferation of cultured cells, we incubated Kras2 LSL/+ and Braf CA/+ primary mouse fibroblasts with a Cre adenovirus to activate the expression of K-RAS G12D and B-RAF V600E , and then added NAC or the soluble vitamin E analog Trolox to the culture medium. The antioxidants did not affect the proliferation of the parental Kras2 LSL/+ and Braf CA/+ fibroblasts (fig. S5A), but the proliferation of K-RAS G12D and B-RAF V600E cells increased in a dose-dependent fashion ( Fig. 3A ). The antioxidants increased BrdU incorporation and the percentage of cells in the S phase of the cell cycle, but did not affect apoptosis (fig. S5, B to D, and table S1). The proliferation of fibroblasts transformed by c-MYC also increased in response to NAC and Trolox (fig. S6 and table S1). ROS analyses revealed that whereas oncogene expression transiently suppressed the amounts of ROS, NAC and Trolox caused sustained suppression of ROS ( Fig. 3B and table S1). Thus, the response to antioxidants is similar in oncogene-expressing fibroblasts and lung tumor cells.

( A ) Left: Quantification of ROS in lung sections of antioxidant-treated and control K-RAS G12D mice as judged by dichlorofluorescein (DCF) fluorescence with the redox-sensitive probe CM-H 2 DCFDA [5-(and-6-)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester] [n = 5 fields of view per lung; five lungs per group for normal lung tissue (NLT); n = 25 tumors from five mice for Ctrl, NAC, and vitamin E (Vit E)]. a.u., arbitraty units. Right: Representative micrographs showing DCF fluorescence. ( B ) Left: Quantification of 8-oxoguanine–positive cells in lungs of antioxidant-treated and control K-RAS G12D mice (n = 2 to 3 tumors per lung; 10 lungs per group). Right: Representative immunohistochemistry micrographs showing 8-oxoguanine staining. ( C ) Left: Quantification of BrdU-positive cells in tumors from antioxidant-treated and control B-RAF V600E mice (n = 3 tumors per lung; five lungs per group). Right: Representative immunofluorescence micrographs showing BrdU-positive tumor cells. BrdU was injected into the peritoneal cavity 3 hours before the mice were sacrificed. ( D ) Quantification of BrdU-positive cells in lung tumors of K-RAS G12D mice treated with NAC for 9 weeks and sacrificed at 10 weeks (NAC Stop). Control mice received NAC for 10 weeks. Scale bars, 100 μm. **P < 0.01, ***P < 0.001. Exact P values are provided in table S2. Graphical data are presented as means ± SEM.

To determine whether NAC and vitamin E suppress the amounts of ROS in tumors, we quantified fluorescence in lung sections stained with a redox-sensitive probe. In untreated mice, ROS was lower in lung tumors than in the surrounding normal tissue ( Fig. 2A , fig. S3A, and table S1). This result is consistent with previous studies showing that oncogenes reduce ROS in tumor cells by activating endogenous antioxidants ( 14 ). ROS in tumors were further reduced by NAC and vitamin E ( Fig. 2A , fig. S3A, and table S1). The reduced ROS was accompanied by increased ratios of reduced (GSH) to oxidized (GSSG) forms of glutathione (fig. S3B and table S1). Consistent with the reduced ROS, NAC and vitamin E reduced the amounts of ROS-induced DNA damage in lung tumors, as judged by immunohistochemical analyses with antibodies against 8-oxoguanine ( Fig. 2B , fig. S3C, and table S1). Further immunohistochemical analyses revealed that tumors from antioxidant-treated mice contained more proliferating cells than did tumors from controls, as judged by the incorporation of 5′-bromo-2′-deoxyuridine (BrdU) and staining with antibodies against phosphorylated histone 3 (pH3) ( Fig. 2C , fig. S3D, and table S1). The increased proliferation was not associated with increased signaling of the RAS–RAF–extracellular signal–regulated kinase (ERK) pathway, because the percentages of tumors with high amounts of phosphorylated ERK1/2 did not differ between groups (fig. S3E). Apoptotic cells, evaluated with TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling) staining and antibodies against cleaved caspase-3, were essentially undetectable in antioxidant-treated and control tumors (fig. S4, A and B). The amounts of senescent or quiescent cells, evaluated with antibodies against p16, p19 ARF , and p21 CIP1 , were largely unaffected (fig. S4, C to E), although p19 ARF in B-RAF V600E tumors was reduced by antioxidants (fig. S4E). When antioxidant supplementation was stopped 1 week before the mice were sacrificed, the numbers of proliferating cells in tumors were reduced compared to tumors of mice receiving antioxidants throughout the experiment ( Fig. 2D and table S1). Thus, antioxidants increase tumor growth by reducing ROS and DNA damage and by promoting tumor cell proliferation.

Although both NAC and vitamin E are antioxidants, they have distinct molecular properties. Vitamin E is fat-soluble, regulates enzymatic activities, and is used as a dietary supplement, whereas NAC is water-soluble, participates in glutathione metabolism, and is used as a mucolytic agent ( 19 , 20 ). We hypothesized that the marked overlap in the effects of NAC and vitamin E on tumor growth and survival reflects their common antioxidant properties. If so, NAC and vitamin E supplementation should affect similar molecular pathways. Indeed, transcriptome sequencing (RNAseq) of K-RAS G12D tumors revealed that the transcriptional changes induced by NAC and vitamin E were highly overlapping ( Fig. 1G and fig. S2, A and B). Pathway analyses revealed that the antioxidants suppressed the expression of genes that participate in the endogenous ROS defense system ( Fig. 1 , G and H, fig. S2C, and table S1). Those genes were also suppressed by antioxidants in normal lung tissue (fig. S2D and table S1). Verified target genes for the transcription factor NRF2 ( 21 ) were enriched among the suppressed genes (fig. S2E), but the percentage of NRF2 targets was low, suggesting that the transcriptional response to antioxidants is also mediated by other factors. Verified p53 target genes were not enriched among the suppressed genes (fig. S2F).

( A and B ) Tumor burden (percent tumor area per lung area) in lungs from NAC-treated (A), vitamin E (Vit E)–treated (B), and littermate control K-RAS G12D mice 10 weeks after inhalation of Cre adenovirus. NAC (1 g/liter) was administered in the drinking water and vitamin E (0.1 and 0.5 g/kg) in the chow diet, starting 1 week after Cre adenovirus inhalation. ( C and D ) Tumor burden in lungs from NAC-treated (C), vitamin E–treated (D), and littermate control B-RAF V600E mice 6 weeks after inhalation of Cre adenovirus. NAC (1 g/liter) was administered 1 week after or 1 week before Cre adenovirus. ( E ) Kaplan-Meier plot showing survival of NAC-treated, vitamin E–treated, and littermate control B-RAF V600E mice (n = 20 to 23 per group). ( F ) Left: Tumor stage (stages 1 to 4) in lungs from mice with K-RAS G12D –induced lung cancer (n = 649 to 894 tumors in lungs from 17 to 33 mice per group). P values are for comparisons of the percentage of stage 3 and 4 tumors in antioxidant-treated and control mice. Right: Representative tumors in hematoxylin and eosin (H&E)–stained lung sections. ( G ) Ratio-ratio plot of K-RAS G12D tumor transcriptome sequencing data showing overlapping regulation of genes by NAC and vitamin E (n = 10 tumors from five mice per group). Statistical analyses identified 310 and 905 differentially expressed transcripts (false discovery rate, q, <0.05) in the NAC- and vitamin E–treated tumors, respectively. The direction of expression change (up or down) was consistent for most genes that were significantly regulated by either antioxidant (gray data points) and for all genes that were significantly regulated by both (blue data points). Unbiased pathway analysis of genes down-regulated by both antioxidants identified the “Metabolism of xenobiotics by cytochrome P450” [Kyoto Encyclopedia of Genes and Genomes (KEGG), P = 1.8 × 10 −6 ; red data points] category and the “Glutathione transferase activity” (GO; P = 2.4 × 10 −4 ) category, which is largely a subset of the former. ( H ) Expression of eight endogenous antioxidant genes identified in the RNAseq analyses. Numbers in bars = n. Scale bar, 500 μm. *P < 0.05, ***P < 0.001, ****P < 10 −30 . Exact P values are provided in table S2. Graphical data are presented as means ± SEM.

To define the impact of antioxidants in lung tumorigenesis, we administered N-acetylcysteine (NAC) in the drinking water to mice harboring a Cre-inducible endogenous oncogenic Kras2 LSL allele, 1 week after they inhaled a Cre adenovirus to activate K-RAS G12D expression in lung epithelial cells. K-RAS G12D mice develop multifocal tumors that vary in grade from epithelial hyperplasia and adenomatous hyperplasia to adenoma and adenocarcinoma ( 16 ). The mice were sacrificed 10 weeks after tumor initiation and were found to have 2.8-fold higher tumor burden than control mice ( Fig. 1A and table S1). We next tested the effect of vitamin E, a structurally unrelated antioxidant. Vitamin E supplementation of the diet increased tumor burden in a dose-dependent fashion ( Fig. 1B and table S1). To determine whether the effect of antioxidants on lung tumor growth occurs with a different oncogene and mouse line, we administered NAC and vitamin E to mice harboring a Cre-inducible endogenous oncogenic Braf CA allele ( 17 ). After Cre adenovirus inhalation, these mice express B-RAF V600E in lung epithelial cells and develop more tumors than K-RAS G12D mice, but with a lower histological grade ( 17 , 18 ). NAC increased the B-RAF V600E –induced tumor burden by 3.4-fold ( Fig. 1C and table S1); the effect was similar in mice given the antioxidant starting 1 week before Cre adenovirus ( Fig. 1C and table S1). In B-RAF V600E mice, vitamin E produced a dose-dependent effect similar to that in K-RAS G12D mice ( Fig. 1D and table S1). Moreover, NAC and vitamin E reduced the median and maximal survival of B-RAF V600E mice by 60 and 50%, respectively ( Fig. 1E ). Histological analyses revealed that the tumors of antioxidant-treated mice had a more advanced histological grade than tumors from control mice ( Fig. 1F and fig. S1, A and B).

DISCUSSION

This study demonstrates that antioxidant supplementation of the diet reduces ROS and DNA damage, prevents p53 activation, and markedly increases tumor cell proliferation and tumor growth in mice. The data demonstrate that tumor cells proliferate faster when oxidative stress is suppressed. This reasoning is consistent with previous studies showing that oncogenes stimulate NRF2-mediated expression of endogenous antioxidants, reduce ROS, and thereby increase tumor cell proliferation (13, 14, 22).

The antioxidants reduced the expression of genes involved in the endogenous ROS defense system. This result is consistent with the reduced amounts of ROS and oxidative DNA damage and the increased GSH/GSSG ratio. The simplest explanation for this result is that a feedback mechanism in lung cells down-regulates the endogenous ROS defense system when the amounts of ROS are suppressed by NAC or vitamin E.

Our data do not support a direct role for the reduced expression of endogenous antioxidant genes in the increased tumor growth. Instead, several lines of evidence suggest that reduced amounts of p53 mediate the antioxidant-induced increase in tumor growth. First, NAC and vitamin E reduced p53 in tumors and cultured mouse and human tumor cells. Second, antioxidants increased the proliferation of human lung cancer cells with wild-type, but not mutant, p53. Third, the ability of antioxidants to increase tumor cell proliferation was abolished when p53 was inactivated or suppressed by shRNAs. One potential explanation for the reduced amounts of p53 is that the antioxidants reduced oxidative DNA damage, γH2AX, and phospho-ATM and thereby removed potent stimuli for p53 activation and stabilization. However, we cannot rule out the possibility that additional factors are involved.

One limitation of the study is that the K-RAS and B-RAF models only allow us to study the impact of antioxidants on tumor progression, and not tumor initiation or prevention. In previous studies, antioxidants were protective against chemically induced lung cancer, and it is possible that high amounts of ROS are required for tumor development in that setting (23, 24). However, experimental studies and large clinical trials quite convincingly suggest that antioxidants, including isoflavones, carotenes, vitamins, and NAC, should not be recommended for the prevention of lung cancer and that their use may promote tumor growth (10, 25–27).

Another limitation is that although antioxidants accelerated the proliferation of human lung cancer cell lines via similar mechanisms as in mouse tumor cells, the precise clinical relevance of our findings is not yet clear. We speculate that because the antioxidant effect was dependent on p53, and TP53 mutations in humans are believed to occur late in tumor progression (28), antioxidants may accelerate the progression of early tumors and precancerous lesions. This would suggest that antioxidants are unsafe in patients with early stages of lung cancer and in people at risk of developing the disease. For example, this may be relevant to patients with chronic obstructive pulmonary disease, who are often smokers with an increased risk of developing lung cancer and ingest high amounts of NAC to relieve mucus production.