Nrf2 status affects tumor growth, HDAC3 gene promoter associations, and the response to sulforaphane in the colon
- 5.1k Downloads
The dietary agent sulforaphane (SFN) has been reported to induce nuclear factor erythroid 2 (NF-E2)-related factor 2 (Nrf2)-dependent pathways as well as inhibiting histone deacetylase (HDAC) activity. The current investigation sought to examine the relationships between Nrf2 status and HDAC expression in preclinical and translational studies.
Wild type (WT) and Nrf2-deficient (Nrf2−/+) mice were treated with the colon carcinogen 1,2-dimethylhydrazine (DMH) followed by 400 ppm SFN in the diet (n = 35 mice/group). WT mice were more susceptible than Nrf2−/+ mice to tumor induction in the colon. Tumors from WT mice had higher HDAC levels globally and locally on genes such as cyclin-dependant kinase inhibitor 2a (Cdkn2a/p16) that were dysregulated during tumor development. The average tumor burden was reduced by SFN from 62.7 to 26.0 mm3 in WT mice and from 14.6 to 11.7 mm3 in Nrf2−/+ mice. The decreased antitumor activity of SFN in Nrf2−/+ mice coincided with attenuated Cdkn2a promoter interactions involving HDAC3. HDAC3 knockdown in human colon cancer cells recapitulated the effects of SFN on p16 induction. Human subjects given a broccoli sprout extract supplement (200 μmol SFN equivalents), or reporting more than five cruciferous vegetable servings per week, had increased p16 expression that was inversely associated with HDAC3 in circulating peripheral blood mononuclear cells (PBMCs) and in biopsies obtained during screening colonoscopy.
Nrf2 expression varies widely in both normal human colon and human colon cancers and likely contributes to the overall rate of tumor growth in the large intestine. It remains to be determined whether this influences global HDAC protein expression levels, as well as local HDAC interactions on genes dysregulated during human colon tumor development. If corroborated in future studies, Nrf2 status might serve as a biomarker of HDAC inhibitor efficacy in clinical trials using single agent or combination modalities to slow, halt, or regress the progression to later stages of solid tumors and hematological malignancies.
KeywordsHDAC3 p16 Nrf2 Colon cancer Sulforaphane Broccoli
acetyl histone H4 lysine 12
human β-actin gene
murine β-actin gene
adenomatous polyposis coli/multiple intestinal neoplasia
broccoli sprout extract
cyclin-dependant kinase inhibitor 1A
cyclin-dependant kinase inhibitor 2A
cyclin-dependent kinase inhibitor 2B
cruciferous vegetable food frequency questionnaire
cytochrome P450 isoform 2E1
ethylene diamine tetraacetic acid
estrogen receptor 1
E26 avian leukemia oncogene 1,5′ domain
histone deacetylase 1
histone deacetylase 3
insulin-like growth factor 2 receptor
Kelch-like ECH-associated protein 1
O6-alkylguanine DNA methyltransferase
mutL homolog 1
nuclear factor erythroid 2 (NF-E2)-related factor 2
transformation-related protein 53
peripheral blood mononuclear cells
parts per million
protein kinase C, alpha
runt-related transcription factor 1
- Serpin b5
serine peptidase inhibitor
The Cancer Genome Atlas
Histone deacetylase (HDAC) enzymes have emerged as important regulators of cancer development [1, 2]. Downregulation of specific HDACs can increase global histone acetylation, turn-on epigenetically silenced genes, and trigger cell cycle arrest, apoptosis, or differentiation in cancer cells [3, 4, 5]. Pan-HDAC inhibitors are currently undergoing clinical evaluation as anticancer agents, but the quest continues for more specific HDAC inhibitors with improved efficacy towards hematological and solid tumors .
We reported that a natural compound, sulforaphane (SFN), targets HDAC3 for protein turnover in human colon cancer cells [6, 7, 8, 9, 10, 11, 12]. SFN is obtained from broccoli and other cruciferous vegetables that are rich sources of the precursor, glucoraphanin [13, 14]. SFN was first identified as an inducer of phase 2 detoxification enzymes, acting via the nuclear factor erythroid 2 (NF-E2)-related factor 2 (Nrf2) signaling pathway . However, Nrf2 in certain circumstances can preserve rather than attenuate cancer phenotypes [16, 17].
The current investigation sought to examine the relationships between Nrf2 status and HDAC expression in a widely used model of colon carcinogenesis [18, 19]. Preclinical experiments included post-initiation SFN treatment and highlighted a role for HDAC3 in regulating cyclin-dependant kinase inhibitor 2a (Cdkn2a/p16) expression. The findings were extended to human subjects according to their cruciferous vegetable consumption or supplement intake.
Dosing schedule and HDAC3 levels dictate anticancer outcomes in the colon
Nrf2 genetic background influences HDAC protein levels in colon tumors
Summary of mouse tumor studies
Colon tumor incidence
Tumor burden (mm3)
Average food consumption (g/day)
Mouse weight at end of study (g)
A 25-week study
23/24 (95.8 %)
5.3 ± 2.8
18.4 ± 15.9
3.5 ± 0.8
45.3 ± 4.9
15/21 (71.4 %)
2.8 ± 2.5*
7.1 ± 7.2*
3.9 ± 1.3
43.6 ± 3.8
10/11 (90.9 %)
7.2 ± 4.4
22.6 ± 18.3
3.8 ± 1.0
44.0 ± 2.9
B 35-week study
28/28 (100 %)
7.0 ± 3.7
62.7 ± 53.5
3.7 ± 0.7
51.3 ± 6.5
27/28 (96.4 %)
5.7 ± 4.0
26.0 ± 30.2*
3.6 ± 1.5
44.7 ± 7.0
31/32 (96.8 %)
4.0 ± 2.1
14.6 ± 24.8
3.1 ± 0.6
45.2 ± 8.0
28/31 (90.3 %)
3.0 ± 2.1
11.7 ± 13.5
3.2 ± 0.5
43.8 ± 6.5
Nrf2 status impacts HDAC3 levels on p16 in mouse colon tumors
Based on prior reports linking Cdkn2a and HDAC3 [23, 24, 25], we performed chromatin immunoprecipitation (ChIP) assays in vivo and observed HDAC3 interactions on the p16 proximal promoter region to be lower in colon tumors of Nrf2−/+ mice compared with WT (Fig. 6f, black vs. grey bars), especially after SFN treatment (green bars). A runt-related transcription factor 1 (Runx1) binding site, ~10 Kb upstream of the p16 promoter , had minimal HDAC3 interactions and served as a negative control for the ChIP assays (Fig. 6f, region “4”). Notably, the findings for HDAC3-dependent regulation of p16 did not represent a generic response of all highly dysregulated genes; HDAC3 interactions on Esr1, for example, were unaffected by SFN (Additional file 2: Figure S2).
p16 is regulated directly by HDAC3, but not Nrf2, in human colon cancer cells
HDAC3 and p16 are reciprocally regulated in humans after SFN intake
In mice, a single oral dose of SFN reduced HDAC3 protein expression and increased p16 protein level in splenocytes, supporting the use of these end points as potential biomarkers in systemic tissues (Additional file 2: Figure S3). Thus, HDAC3 and p16 protein expression changes also were examined in circulating peripheral blood mononuclear cells (PBMCs) from human volunteers (Fig. 8c). HDAC3 was reduced as early as 1 h after BSE consumption and continued to remain low up to 6 h later. Notably, the loss of HDAC3 coincided with increased p16 expression during this time period. On day 7, HDAC3 levels returned to baseline, whereas p16 remained elevated (Fig. 8c). This implies that cross-talk between HDAC3 and p16 can be uncoupled at later times, possibly via interactions of p16 with alternative HDACs, histone acetyltransferases, and/or their co-regulators [20, 21]. For example, HDAC3/SMRT inhibition and turnover has been observed to precede changes in other HDACs, such as HDAC6 . Lower SFN metabolite levels at day 7 vs. day 1 hinted at a possible compensatory mechanism following repeated daily SFN intake. Further studies are needed to examine the possible induction of pathways that favor enhanced SFN metabolism and/or excretion.
Based on findings from the short-term intervention trial with BSE, HDAC3 and p16 were examined in the context of more typical human dietary intake patterns (Fig. 8d). Patients scheduled for a screening colonoscopy were stratified as high vs. low cruciferous vegetable consumers based on a validated questionnaire . In PBMCs obtained immediately prior to colonoscopy, p16 mRNA levels were significantly higher for subjects reporting >5 vs. 0–1 servings of cruciferous vegetables per week (Fig. 8e). We next examined the corresponding colon biopsies for selected proteins of interest, i.e., p16, HDACs, and histone acetylation (Fig. 8f). With higher cruciferous vegetable intake, histone acetylation and p16 expression were increased, whereas HDAC1 levels were unchanged. Based on densitometry measurements of immunoblots, a significant inverse association was observed for p16 and HDAC3 in colon biopsies (Fig. 8g), whereas p16 and acetyl histone H4 lysine 12 (AcH4K12) (Fig. 8h) and HDAC3 and Nrf2 (Fig. 8i) were positively correlated.
SFN was first identified as a potent inducer of phase 2 enzymes, acting via the Nrf2 pathway to induce detoxification pathways that favor carcinogen excretion and elimination from the body [14, 15]. However, SFN also can be effective in post-initiation protocols , independent of carcinogen exposure. This is exemplified by the tumor suppression observed for SFN in genetic models, such as the adenomatous polyposis coli/multiple intestinal neoplasia (ApcMin/+) mouse, in which HDAC inhibition was identified as a contributing mechanism . Therefore, we examined the interplay between two key mechanisms implicated in the SFN antitumor activity, namely, Nrf2 induction and HDAC inhibition. Although pharmacological HDAC inhibitors have been tested in preclinical models of colon cancer , the role of Nrf2 was not examined. We specifically sought to test the hypothesis that Nrf2 status might affect HDAC3 protein expression levels in colon tumors, and thus the inhibitory response to SFN acting preferentially on HDAC3 [20, 21].
As in the ApcMin/+ mouse [10, 31], SFN suppressed tumorigenesis in the DMH model, and this was accompanied by reduced HDAC activity and HDAC3 protein expression in the colon tumors. These findings are in accordance with data from human colon cancer cells showing HDAC inhibition by SFN and its metabolites [11, 20], and by other dietary isothiocyanates , to involve HDAC3 protein turnover. Metabolites implicated in the HDAC3 turnover mechanism [11, 20, 21], such as SFN-Cys and SFN-NAC, were detected in tissues of SFN-treated mice, as reported . The corroboration of HDAC3 as a target of SFN in vivo is important considering the critical role of this HDAC in regulating colon cancer growth and tumorigenesis . Loss of tumor suppression in mice fed with SFN on alternating days (Fig. 1) might be related to the inability to sustain high enough SFN metabolite levels for effective HDAC3 inhibition. Indeed, SFN metabolites are cleared within 24 h in mice , as in human subjects consuming BSE (Fig. 8b).
Nrf2-deficient mice are generally more sensitive to carcinogens and agents that trigger chronic inflammation [32, 33, 34, 35]. However, Nrf2−/+ mice treated with DMH and observed for up to 35 weeks had a significantly reduced tumor burden and lower HDAC protein expression compared with WT animals. The differential response to the carcinogen likely was not attributable to genes that regulate DMH metabolism (Cyp2E1) or DNA repair (Mgmt), since their expression was similar in WT and Nrf2−/+ mouse colon (Additional file 2: Figure S4). In a recent study using urethane to initiate lung tumors , resistance to tumor growth was observed in Nrf2-deficient mice compared with WT. The enhanced tumor growth in WT mice adds to the discussion on pros vs. cons of Nrf2 signaling in different stages of cancer development [17, 36].
From the gene expression arrays, qRT-PCR and immunoblotting experiments, tumor suppressors p21 and p15 were induced only marginally, if at all, by SFN in the preclinical model reported here, in marked contrast to p16. For example, in gene expression arrays (Additional file 1: Table S1), p21 and p15 were attenuated slightly in the colon tumors from SFN-treated mice, whereas p16 was induced greater than 20-fold. Although p16 more typically is considered a tumor suppressor protein, high Cdkn2a/p16 levels have been detected in mouse colon tumors induced by azoxymethane, a metabolite of DMH , and in human benign tumors and high-grade malignancies . Overexpression of p16 has been linked to the so-called oncogene-induced senescence (OIS) in benign tumors or as a mechanism to arrest uncontrolled proliferation in advanced cancers . Early upregulation of p16 in some tumors might represent an attempt to correct for one or more dysregulated signaling pathways. We focused on p16 as a major target dysregulated in DMH-induced colon tumors and noted that SFN increased p16 expression in tumors of WT mice but had the opposite effect in Nrf2−/+ mice. Based on evidence that HDAC inhibition activates p16 [39, 40, 41], we confirmed HDAC3 interactions on p16 to be higher in colon tumors of WT vs. Nrf2−/+ mice (Fig. 6f), and inversely associated with p16 mRNA levels (Fig. 6a). In human colon cancer cells, HDAC3 knockdown increased p16 levels to a similar extent as SFN treatment, whereas Keap1 knockdown had no effect on p16 (Fig. 7). This suggested that p16 is regulated by HDAC3 but is not a direct target of Nrf2.
What, then, connects Nrf2 genetic background to altered HDAC3 levels on p16? A working model can be proposed (Additional file 2: Figure S5). We speculate that Nrf2 deficiency in mice, through mechanisms that remain to be clarified, attenuates the overall rate of colon tumor growth and global HDAC levels within the tumor. This in turn diminishes HDAC interactions on key genes dysregulated during tumor development, regardless of whether or not they are directly regulated by Nrf2 binding. Mechanisms affecting promoter methylation and transcription factor access [42, 43, 44, 45] might influence which genes are most altered, and the ultimate response to SFN treatment.
Nrf2 status in the mouse colon appears to serve as an arbiter of overall colon tumor growth, HDAC protein expression in colon tumors, and the response to HDAC inhibitor treatment mediated by downstream molecular targets such as p16. A key issue will be the extent to which Nrf2 status influences HDAC levels and p16 expression at different stages of human colon cancer development and the ultimate response to pan-HDAC or HDAC-selective inhibitors. This could have implications beyond the treatment of colorectal cancer, for example, in other solid tumors and hematological malignancies currently undergoing clinical evaluation with HDAC inhibitors . In the broad context of these various clinical trials and their overall aims, we believe that Nrf2 status is worthy of further investigation as a possible determinant of tumor growth and HDAC inhibitor responsiveness.
Animals and diets
Male WT or Nrf2−/+ mice at 8–10 weeks of age were randomized to 25 mice/group (pilot study) or 35 mice/group (main study). Tumors were induced by 1,2-dimethylhydrazine (DMH, Sigma-Aldrich), as reported . After DMH treatment, mice continued on regular AIN93 diet for an additional week before administering AIN93 diet or AIN93 diet supplemented with 400 ppm d,l-SFN (Toronto Research Chemicals, Inc.), either continuously or on alternating days. At end of the study, the colon was removed, and tumors were scored for number, size, and position by individuals blinded to the treatment. Tumor volume was calculated using the formula xy 2 * 0.5 (x = long diameter, y = short diameter). Total tumor burden per animal was calculated by adding the individual tumor volumes. Tumor and adjacent normal-looking tissue was removed and one portion was fixed in 10 % buffered formalin, while the other portion was flash-frozen in liquid nitrogen and stored at −80 °C. The work was approved by the Institutional Animal Care and Use Committee.
Frozen samples of colon tumors and adjacent tissue were thawed and subjected to immunoblotting using the methodology described previously [20, 21, 48]. Antibodies were for HDAC1 and HDAC3 (Santa Cruz), p16 (Proteintech), acetyl histone H4K12 and histone H4 (Cell Signaling), and β-actin (Sigma-Aldrich).
PCR arrays and qPCR
RT2 Profiler arrays were run as per manufacturer’s instructions (SA Biosciences, Qiagen, Valencia, CA, USA) on colon tumor samples (pooled, n = 6) and matched controls (pooled, n = 6). From the threshold cycle (C t) value, the relative gene expression of each target was normalized to human β-actin gene (ACTB) or murine β-actin gene (Actb) (β-actin gene in human and mouse, respectively). Two or more separate qRT-PCR experiments were performed to validate targets of interest, as reported .
The ChIP-IT Express Enzymatic kit (Active Motif, Carlsbad, CA) was used, as reported . Frozen tissue (100 mg) was cut into pieces, cross-linked with formaldehyde, and homogenized in order to isolate the nuclear fraction. DNA fragmentation was performed via enzymatic shearing, using a proprietary cocktail (Active Motif) that randomly cleaves between nucleosomes, generating DNA fragments of ~500 bp. Ten microliters of fragmented chromatin was kept as input while the remaining was immunoprecipitated (IP) with anti-HDAC3 antibody (Santa Cruz). After reversing the cross-linking and proteinase treatment, DNA was purified using QIAquick PCR Purification kits (Qiagen). PCR was run on a Roche LightCycler 480 II with preincubation for 5 min at 95 °C, then 45 cycles at 95 °C for 10 s, 60 °C for 10 s, and 72 °C for 10 s. Each experiment was repeated at least twice.
Formalin-fixed paraffin-embedded mouse colon tumor and adjacent normal tissue were processed for immunohistochemistry as reported . Slides containing 5 μm sections were rehydrated and placed in an Autostainer (Dako). After primary antibody to p16 (Proteintech) for 30 min and One-Step HRP Polymer anti-IgG (ImmunoBioscience) for 7 min, Nova Red (Vector Labs) was applied for 5 min followed by hematoxylin (Dako). Images were acquired on a Nikon E400 microscope equipped with a CCD camera.
HCT116 cells were obtained from American Type Culture Collection (Manassas, VA, USA) and validated as reported . Cells were transfected for 48 h with HDAC3 siRNA (Trilencer-27, OriGene), Keap1 siRNA (Sigma-Aldrich), control siRNA (OriGene), or Lipofectamine 2000 alone, using the manufacturer’s protocol (Invitrogen).
Phase “0” trial
A pilot intervention study was performed based on a prior protocol . Ten healthy subjects avoided cruciferous vegetables, starting 1 week before day 1 of the study and continuing through day 14. After an overnight fast, volunteers ate a standardized breakfast together with a broccoli sprout extract (BSE) supplement (IND #111736, 200 μmol SFN equivalents, n = 5) or a placebo (n = 5), once a day for 7 days. BSE and placebo were obtained from Johns Hopkins University (Baltimore, MD, USA), with SFN content validated by LC-MS/MS, as reported . The Institutional Review Board (IRB) approved the protocol, and all participants provided written consent.
Whole blood was collected into ethylene diamine tetraacetic acid (EDTA) Vacutainers (VWR, Radnor, PA, USA) at 0, 1, 3, and 6 h post-consumption on days 1 and 7 and on days 8, 9, and 14. After centrifuging at 2000 rpm for 30 min, plasma was removed, acidified with trifluoroacetic acid, and stored at −80 °C. Samples were analyzed for SFN metabolites as previously described . The remaining whole blood was processed to isolate peripheral blood mononuclear cells (PBMC), as described previously , and frozen at −80 °C.
Screening colonoscopy study
Men and women aged >50 years and scheduled for a screening colonoscopy were recruited based on cruciferous vegetable consumption (n = 28). Recruitment and data collection were implemented through the Oregon Clinical and Translational Research Institute and the Oregon Health & Science University Cancer Institute, with IRB approval and written consent from each participant. Subjects completed a validated  cruciferous vegetable food frequency questionnaire (CVFFQ) and had three 24-h dietary recalls over a 3-week period. The CVFFQ assessed intake over the previous 12 months, relating to number of servings, serving size, intake of raw and cooked vegetables, method of cooking, and use of condiments. Data from the CVFFQ were analyzed by the Arizona Diet, Behavioral, and Quality of Life Assessment Center, University of Arizona, Tucson, AZ, USA. Volunteers were stratified into low (0–1 serving/week, n = 5) and high (≥5 servings/week, n = 23) consumers. Dietary recalls were used to provide information on possible changes in diet between the CVFFQ and the clinic visit. Blood was obtained and processed as in the BSE trial. In addition, two biopsies of rectal colon and two of proximal colon were taken with standard forceps and placed in formalin or flash-frozen in liquid nitrogen.
Results were expressed as mean ± SD. Analysis of variance (ANOVA) was used for group comparisons, followed by Bonferroni’s multiple comparison test (GraphPad Prism v 5.04). Student’s t test was used for paired comparisons, with P < 0.05 considered as significant.
We thank Rong Wang, Hui Nian, Chris Larsen, Kate Cleveland, Hassaan Saeed, Tian-Wei Yu, and Soyoun Ahn for their technical assistance. Daniel Sudakin, M.D., Mary Garrard, R.N., Karen Hardin, Joshua Hay, and Matthew French assisted with the clinical trials. BSE supplement and placebos were from Drs. P. Talalay and J. Fahey (Johns Hopkins University, Baltimore, MD, USA). This work is supported by a grant CA090890 from the National Cancer Institute, P30 grants ES00210 and ES02351 from the National Institute of Environmental Health Sciences, and by a Chancellor’s Research Initiative from Texas A&M University. The open access publishing fees for this article have been covered by the Texas A&M University Online Access to Knowledge (OAK) Fund, supported by the University Libraries and the Office of the Vice President for Research.
- 1.Fredly H, Gjertsen B, Bruserud Ø. Histone deacetylase inhibition in the treatment of acute myeloid leukemia: the effects of valproic acid on leukemic cells, and the clinical and experimental evidence for combining valproic acid with other antileukemic agents. Clin Epigenetics. 2013;5:12.PubMedCentralCrossRefPubMedGoogle Scholar
- 12.Myzak MC, Tong P, Dashwood W-M, Dashwood RH, Ho E. Sulforaphane retards the growth of human PC-3 xenografts and inhibits HDAC activity in human subjects. Exp Biol Med. 2007;232:227–34.Google Scholar
- 20.Rajendran P, Delage B, Dashwood WM, Yu TW, Wuth B, Williams DE, et al. Histone deacetylase turnover and recovery in sulforaphane-treated colon cancer cells: competing actions of 14-3-3 and Pin1 in HDAC3/SMRT corepressor complex dissociation/reassembly. Mol Cancer. 2011;10:68.PubMedCentralCrossRefPubMedGoogle Scholar
- 24.Wang X, Feng Y, Xu L, Chen Y, Zhang Y, Su D, et al. YY1 restrained cell senescence through repressing the transcription of p16. Biochim Biophys Acta. 1783;2008:1876–83.Google Scholar
- 42.Chien WW, Ffrench M. Regulation of p16INK4a, senescence and oncogenesis. Med Sci. 2006;22:865–71.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.