Background
Epigenetic changes play a critical role in cancer development [
1‐
5]. These changes include the dysregulation of histone deacetylases (HDACs) and the altered acetylation status of histone and non-histone proteins [
6‐
8]. Efforts have been directed at reversing aberrant acetylation patterns in cancers through the use of HDAC inhibitors. HDAC inhibitors induce cell cycle arrest, differentiation, and apoptosis in cancer cells, some have anti-inflammatory activities, and a number have progressed to clinical trials [
8‐
12].
HDACs can be overexpressed in colorectal cancers and in several other cancer types [
13‐
18]. Silencing of HDACs, individually or in combination, has provided insights into the associated molecular pathways that regulate cell cycle transition, proliferation, and apoptosis [
14,
18‐
20]. In human colon cancer cells, silencing of HDAC3 resulted in growth inhibition, decreased cell survival, and increased apoptosis [
14]. Similar effects were noted for HDAC2 and, to a lesser extent, for HDAC1. Subsequent work [
18] identified a role for HDAC4 in regulating p21
WAF1 expression, via a corepressor complex involving HDAC4, HDAC3, and SMRT/N-CoR (silencing mediator for retinoid and thyroid hormone receptors/nuclear receptor co-repressor). Spurling
et al.[
16] reported that knockdown of HDAC3 increased constitutive, trichostatin A (TSA)-, and tumor necrosis factor (TNF)-α-induced expression of p21
WAF1, although HDAC3 silencing alone did not account for all the gene expression changes observed upon general HDAC inhibition. Cells with lowered HDAC3 expression had increased histone H4-K12 acetylation (H4K12ac) and were poised for gene expression changes [
16]. Ma
et al.[
20] observed that recruitment of p300 to the
survivin promoter led to the concomitant recruitment of other protein partners, including HDAC6, resulting in transcriptional repression. Thus, there is accumulating evidence for the involvement of multiple HDACs in colon cancer development.
HDAC activity and histone acetylation status can be influenced by dietary factors and their metabolites [
21‐
23]. For example, broccoli and broccoli sprouts are a rich source of glucoraphanin, the glucosinolate precursor of the cancer chemoprotective agent sulforaphane (SFN) [
24‐
28]. SFN has been reported to inhibit HDAC activity in human colon cancer cells [
29], and this was confirmed in prostate and breast cancer cells [
30,
31]. A structurally-related isothiocyanate also inhibited HDAC activity in human leukemia cells, resulting in chromatin remodeling and growth arrest [
32]. Combining these findings with the changes induced by SFN in NF-E2-related factor 2 (Nrf2) signaling [
24‐
28,
33], a "one-two" chemoprotective model can be proposed. In the first stage, SFN parent compound induces phase 2 detoxification pathways, and in the second stage SFN metabolites alter HDAC activity and histone status, leading to the unsilencing of tumor suppressors such as p21
WAF1[
34‐
36]. An unresolved question from our earlier studies [
29] was the fate of individual HDACs in SFN-treated colon cancer cells. If, indeed, SFN metabolites act as weak ligands for HDACs [
37], does this result in de-recruitment and/or turnover of specific HDAC proteins, and is this reversible? These questions were examined in the present investigation, along with the molecular mechanisms involved.
Discussion
This is the first investigation to examine the fate of individual HDACs in human colon cancer cells treated with SFN, with the dual aims of clarifying the mechanisms of the observed HDAC protein turnover and the timing of HDAC recovery following SFN removal. Pappa
et al.[
57] previously performed transient exposure experiments with SFN, observing that G
2/M arrest and cytostatic growth inhibition were reversible in the cell line 40-16. In the present study, HCT116 cells were plated at low density so as to allow HDAC changes to be followed for at least 72 h. Under these conditions, 6-24 h of SFN exposure followed by SFN removal resulted in the complete recovery of HDAC activity and HDAC protein expression, along with the normalization of histone acetylation and p21
WAF1 status. Although apoptosis induction was detected, most notably at higher SFN concentrations, caspase-3-mediated cleavage of HDAC3 [
39] was excluded as a contributing mechanism in the loss of HDAC3 protein. Other HDACs are known to be cleaved by caspases; for example, caspase-8-mediated cleavage of HDAC7 has been reported [
40]. HDAC7 and several other class II HDACs were unaffected at the protein level by SFN treatment; however, a formal examination of each caspase and its potential HDAC target(s) may be warranted.
Changes in HDAC6 were of interest because this HDAC has been described as a master regulator of cellular responses to cytotoxic insults [
42]. We performed several experiments on HDAC6 and observed the following: (i) HDAC6 protein loss was first detected at around 24 h post-SFN treatment (
versus 6 h for HDAC3); (ii) although delayed relative to other HDACs, HDAC6 was fully recovered by 72 h in the SFN reversibility studies; (iii) as with HDAC3, HDAC6 loss was not prevented by a cell-permeable pan caspase inhibitor; (iv) immunoprecipitation of HDAC3 followed by HDAC6 from whole cell lysates accounted for all of the HDAC inhibitory effects of SFN (Additional File
2); and (v) transient overexpression of HDAC6 in HCT116 cells completely blocked the increased tubulin acetylation associated with SFN treatment, as well as the induction of H4K12ac. Gibbs
et al.[
58] performed ectopic overexpression of HDAC6 in human prostate cancer cells, observing SFN-mediated inhibition of HDAC6 activity, HSP90 hyperacetylation, and destabilization of the androgen receptor. Decreased endogenous HDAC6 and HDAC3 protein expression was recently reported in SFN-treated prostate epithelial cells [
59], although the precise molecular mechanisms were not pursued. We conclude that HDAC6, along with its corepressor partners, is an important target for SFN action in human prostate and colon cancer cells. However, depletion of HDAC3 followed by HDAC6 (Additional File
2), or HDAC6 followed by HDAC3 (data not shown), suggested that HDAC3 accounted for approximately two-thirds and HDAC6 one-third of the SFN actions on HDAC activity in HCT116 cells. This observation coupled with the delayed loss and slower recovery of HDAC6 compared with HDAC3 suggested that HDAC3 plays a pivotal "sentinel" role, although HDAC6 mediating HDAC3 activity probably warrants further investigation.
In the present investigation, co-IP experiments indicated that dissociation of HDAC3/SMRT corepressor complexes occurred within 6 h of SFN treatment. SMRT and N-Cor are known to be regulated by distinct kinase signaling pathways [
48], resulting in corepressor complex disassembly and redistribution from the nucleus to the cytoplasmic compartment. Erk2, a mitogen-activated protein kinase, disrupts SMRT self-dimerization, releasing HDAC3 and other protein partners from the corepressor complex, thereby lowering transcriptional repression [
60]. SFN is known to activate kinase signaling pathways [
27,
61,
62], and we observed increased p-HDAC3 and p-SMRT in the nucleus within 6 h of SFN exposure, along with increased CK2 binding to HDAC3. In prior studies, phosphorylation of HDAC4 triggered its nuclear export and binding to 14-3-3 [
49]. In an analogous fashion, we now report, for the first time, that there was increased binding of 14-3-3 to HDAC3 following SFN treatment. This raises the possibility that 14-3-3 sequesters HDAC3 in the cytosolic compartment, pending the subsequent release and re-entry of HDAC3 into the nucleus (e.g., upon SFN removal).
Supporting this hypothesis were the results using phosphospecific antibodies to 14-3-3. The loss of cytoplasmic and nuclear p-14-3-3(T232) upon SFN treatment is consistent with this phosphorylation impeding interactions with client proteins, such as HDAC3, and indeed no p-14-3-3(T232) was pulled down with HDAC3 in the presence or absence of SFN treatment (Figure
8C). Loss of T232 phosphorylation upon SFN treatment would provide access to the adjacent nuclear export signal in 14-3-3 [
63], facilitating nuclear-cytoplasmic trafficking. On the other hand, phosphorylation of S58 in 14-3-3 shifts the pool of 14-3-3 towards more of the monomeric form, although some interaction of p-14-3-3(S58) with HDAC3 was detected. The current model (Figure
8D) proposes 14-3-3 interacting with HDAC3 phosphorylated at S424; however, other phosphorylation sites in HDAC3 may be involved, such as those associated with glycogen synthase kinase-3β [
64]. Based on the findings with class IIa HDACs [
56], 14-3-3 binding to HDAC3 might block the nuclear localization signal and facilitate cytoplasmic retention of HDAC3, leaving the nuclear export signal accessible to proteins such as CRM1 that further traffic HDAC3 from the nucleus to the cytoplasm. Additional work is needed to clarify this model, including the relative contributions of monomeric
versus dimeric 14-3-3, and the role of other known phosphorylation sites in 14-3-3 [
53‐
55].
Another interesting and novel observation was that SFN increased the binding of HDAC3 to Pin1. Pin1 knockdown completely blocked the SFN-induced loss of HDAC3, although this did not interfere with the induction of p21
WAF1. One explanation may be that HDAC1 and HDAC2 are the primary repressor HDACs of p21
WAF1[
65], and neither one interacted with Pin1 before or after SFN treatment (Additional File
1). Importantly, Pin1 binding to p-SMRT has been reported to trigger SMRT degradation [
52]. Proteins such as c-Myc and cyclin E use a common Pin1-interacting motif to allow turnover by the Fbw7 E3 ligase [
52], but this motif does not exist in SMRT [
52]. This suggests that a novel E3 ligase may be involved in the turnover of SMRT, and possibly HDAC3. There are estimated to be 500-1000 E3 ligases in human cells [
47], and further work is warranted to identify the E3 ligase involved in HDAC3 turnover. Although PYR-41 has been reported as an E1 inhibitor [
47], it also affects sumoylation pathways, which complicated the interpretation of PYR-41 effects on SFN-induced HDAC3 turnover in HCT116 cells. Interestingly, a selective inhibitor of CK2, 4,5,6,7-tetrabromo-2-azabenzimidazole, dose-dependently depleted Pin1 and concomitantly increased HDAC3 protein expression in HCT116 cells, further confirming the inverse association between these two proteins (P. Rajendran, data not presented).
Although the details are far from definitive and several questions remain, a model is proposed for SFN actions in human colon cancer cells (Additional File
3). Following SFN treatment, kinase signaling pathways facilitate CK2 recruitment to nuclear HDAC3/SMRT corepressor complexes resulting in the phosphorylation of HDAC3 and SMRT, complex dissociation, binding to 14-3-3 or Pin1, and trafficking from the nucleus to the cytoplasm. In the cytoplasmic compartment, sequestration of HDAC3 by 14-3-3 competes with a pathway involving Pin1-directed HDAC3 degradation. Upon SFN removal, it is postulated that HDAC3 and SMRT are released from 14-3-3 to re-enter the nucleus, reassembling the corepressor complexes to silence gene activation. Further work is needed to clarify the possible involvement of a unique E3 ligase that targets both HDAC3 and SMRT for simultaneous degradation. This model highlights the role of kinase signaling pathways triggered by SFN, but does not exclude direct actions of SFN or its metabolites on HDACs [
29]. For example, entry of SFN metabolites into the HDAC3 pocket might lead to conformational changes and/or altered protein interactions that facilitate CK2 binding. These mechanisms are under further investigation in SFN-treated colon cancer cells, including time-course analyses of histone marks and the phospho-acetyl switch [
66].
Methods
Cell culture and reagents
Human HCT116 colon cancer cells (ATCC, Manassas, VA) were cultured at 37°C, 5% CO2 in McCoy's 5A medium (Life Technologies, Carlsbad, CA) supplemented with 1% penicillin-streptomycin and 10% fetal bovine serum. SFN (Toronto Research Chemicals Inc. North York, ON, Canada) was prepared in DMSO and stored at a stock concentration of 10 mg/mL at -20°C. Chemical inhibitors leupeptin, ALLN, MG-132 (Sigma, St. Louis, MO) and PYR-41 (Calbiochem, San Diego, CA), were dissolved in DMSO (10 mM) and small aliquots (30 μl) were stored at -20°C. Z-VAD (OMe)-FMK was from SM Biochemicals LLC (Anaheim, CA). Cycloheximide and actinomycin D were purchased from Sigma (St. Louis, MO).
Cell Growth
Cells in the exponential growth phase were plated at a cell density of 5,000 cells per well in 96-well tissue culture plates. After attachment overnight, cells were treated with 15 μM SFN for selected times i.e., 2, 24, 48 and 72 h. At these time points cell viability was determined using the MTT assay, as described previously [
67], and cell number was counted using a Neubauer chamber.
Flow cytometry
Cells in the exponential growth phase were plated at a cell density of 0.1 × 106 cells in 60-mm culture dishes and treated with 0 (DMSO) or 15 μM SFN. Adherent and non-adherent cells were collected at different time points i.e., 3, 6, 9, 24, 48 and 72 h in cold PBS, fixed in 70% ethanol, and stored at 4°C for at least 48 h. Fixed cells were washed with PBS once and resuspended in propidium iodide (PI)/Triton X-100 staining solution containing RNaseA. Samples were incubated in the dark for 30 min before cell cycle analysis. DNA content was detected using EPICS XL Beckman Coulter and analyses of cell distribution in the different cell cycle phases were performed using Multicycle Software (Phoenix Flow Systems, San Diego, CA).
Cell lysates
Cells in the exponential growth phase were plated at a cell density of 0.1 × 106 cells in 60-mm culture dishes. After overnight incubation cells were treated with either 0 (DMSO) or 15 μM SFN. In some experiments a range of SFN concentrations was used (0, 10, 15, 25, 35 μM). Adherent and non-adherent cells were harvested by trypsinization at different time points, ranging from 2 to 72 h, and then washed with ice-cold PBS. Whole-cell extracts were prepared using lysis buffer containing 20 mM (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM sodium orthovanadate, and 1 μg/ml leupeptin. The harvested cell pellet obtained after centrifugation was resuspended in lysis buffer and frozen at -80°C for at least 15 min, thawed on ice, vortexed for 30s and centrifuged at 13,200 × g for 5 min. To study the reversibility of SFN effects, 0.1 × 106 cells in 60-mm culture dishes were treated with DMSO or 15 μM SFN for 6 or 24 h, and the media was replaced with fresh growth medium (containing no SFN) until harvest. Whole-cell extracts were prepared at 6, 24, 48 and 72 h, and samples were frozen at -80°C until further use. Cytoplasmic and nuclear lysates were prepared using NE-PER® Nuclear & cytoplasmic extraction reagent (#78833, Thermo scientific, Rockford, IL). The insoluble fraction was dissolved in SDS lysis buffer containing 65 mM Tris-HCl, pH 8.0, 2% SDS, 50 mM DTT, and 150 mM NaCl. Protease (Roche) and phosphatase (Sigma, St. Louis, MO) inhibitor cocktails were added immediately before use. Protein concentration of cell lysates was determined using the BCA assay (Pierce, Rockford, IL).
In vitro HDAC activity
HDAC activity was measured from whole cell lysates using the Fluor-de-Lys HDAC activity assay kit (Biomol, Plymouth Meeting, PA), as reported before [
68]. Incubations were performed at 37°C with 10 μg of whole-cell extracts along with the fluorescent substrate in HDAC assay buffer for 30 min. Assay developer was then added and the samples incubated at 37°C for another 30 min and read using a Spectra MaxGemini XS fluorescence plate reader (Molecular Devices), with excitation at 360 nm and emission at 460 nm. The results were expressed as AFU or AFU/μg protein.
Immunoblotting
Equal amounts of protein (20 μg/lane) were separated by SDS-PAGE on 4-12% Bis-Tris gel or 3-8% Tris acetate gel for larger proteins (NuPAGE, Invitrogen, Carlsbad, CA) and transferred to nitrocellulose membranes (Invitrogen, Carlsbad, CA). Membranes were saturated with 2% BSA for 1 h, followed by overnight incubation at 4°C with primary antibodies against β-actin (1:50,000 Sigma, #A5441), casein kinase-IIα (1:200, Santa Cruz, #9030), cleaved caspase-3 (1:1000, Cell Signaling, #9661), acetyl histone H4K12 (1:500, Upstate, #07-595), histone H4 (1:1000, Cell Signaling, #2592), HDAC1 (1:200, Santa Cruz, #7872), HDAC2 (1:200, Santa Cruz, #7899), HDAC3 (1:200, Santa Cruz, #11417), HDAC4 (1:200, Cell Signaling, #2072), HDAC6 (1:200, Santa Cruz, #11420), HDAC8 (1:200, Santa Cruz, #11405), HDAC10 (1:200, Biovision, #3610-100), phosphoHDAC3 (1:1000, Cell Signaling, #3815), HDAC3 N-19 (1:200, Santa Cruz, #8138), N-Cor (1:1000, Abcam, #ab24552), p21WAF1 (1:1000, Cell Signaling, #2947), PARP (1:1000, Cell Signaling, #9542), phosphoSMRT (pS2410, kindly provided by Dr. Marty Mayo, Univ. of Virginia, 1:1000), Pin1 (1:1000, Millipore, #07-091), SMRT (1:600, Millipore, #04-1551), acetyl α-tubulin (1:2000, Sigma, #T6793), α-tubulin (1:1000, Abcam, #ab7291), ubiquitin (1:3000, BD Pharmingen, #550944), pan14-3-3 (1:500, Santa Cruz, #629), p-14-3-3(T232) and p-14-3-3(S58), both used at 1:500 dilution (Epitomics Inc., Burlingame, CA). After washing, membranes were incubated with respective horseradish peroxidase conjugated secondary antibodies (Bio-Rad, Hercules, CA) for 1 h. Immunoreactive bands were visualized via Western Lightning Plus-ECL Enhanced Chemiluminescence Substrate (Perkin Elmer, Inc, Waltham, MA) and detected with FluorChem-8800 Chemiluminescence and Gel Imager (Alpha Innotech).
Immunoprecipitation
HCT116 cells were treated with either DMSO or 15 μM SFN with or without pre-treatment for 1 h with PYR-41 (50 nM). Cells were harvested after 6 or 24 h and either whole cell extracts or cytoplasmic and nuclear lysates from adherent and non-adherent cells were prepared as previously described. Protein concentration was determined by BCA assay (Pierce, Rockford, IL). Protein (500 μg) was precleared with Protein A Sepharose CL-4B (Amersham Biosciences) on a rotator at 4°C for 1.5 h. Pre-cleared supernatant was collected and immunoprecipitated overnight with anti-HDAC3 (2 μg, Santa Cruz, #11417) or anti-HDAC6 (2 μg, Santa Cruz, #11420) rabbit polyclonal antibody. Protein A Sepharose beads were collected and washed before immunoblotting with anti-HDAC3 (1:200), anti-SMRT (1:500), anti-phosphoSMRT (1:700), anti-Pin1 (1 μg/ml), anti-14-3-3 (1:500), and anti-casein kinase-IIα (1:100) antibodies. The supernatant depleted of HDAC3 and/or HDAC6 was collected and kept frozen at -80°C until used for HDAC activity assays. In some experiments, HDAC3 pulls-downs were followed by immunoblotting for p-14-3-3(T232) and p-14-3-3(S58), both at 1:250 dilution.
Overexpression and knock-down experiments
HDAC3 and HDAC6, as transfection-ready DNA in pCMV6-XL4 vector, and Pin1 siRNA (Trilencer-27) and control siRNA were from Origene (Rockville, MD). Cells were transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) at a ratio of 1:3-1:4 in reduced serum medium (OPTI-MEM, Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. SFN treatment started after 24 h of transfection. Immunoblotting was carried out with whole cell lysates prepared using lysis buffer.
Statistics
The results of each experiment shown are representative of at least three independent assays. Where indicated, results were expressed as mean ± standard error (mean ± SE), and differences between the groups were determined using Student's t-test. For multiple comparisons, ANOVA followed by the Dunnett's test was performed using GraphPad Prism. A p-value <0.05 was considered as statistically significant, and indicated as such with an asterisk (*) in the corresponding figure.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PR was responsible for the reversibility and HDAC overexpression data, the immunoblotting of corepressor proteins, and Pin1 knockdown studies, plus drafting of the manuscript. BD performed time-course, reversibility, and caspase-3 cleavage experiments. WMD performed qPCR studies of class I and class II HDAC s, plus co-IP assays and associated immunoblotting. TWY and BW assisted with HDAC activity assays. DEW, EH, and RHD were responsible for the overall study design and interpretation of data, plus final drafting and editing of the manuscript for publication.