Basal-like breast cancers are triple-negative (estrogen receptor negative, progesterone receptor negative, erythroblastic leukemia viral oncogene homolog 2 (ERBB2) negative) tumors with an aggressive clinical behavior that lacks effective molecular targets for therapy. We reported previously that the basal-like subtype cell lines display high constitutive nuclear factor (NF)-κB activation, whose inhibition in the basal-like subtypes suppressed their proliferation. Moreover, NF-κB-inducing kinase (NIK) is involved in the constitutive NF-κB activation. Here, we report that enhanced NIK expression, which is exclusively observed in the basal-like subtype rather than the luminal-like subtype or non-tumorigenic mammary epithelial cells, is caused by epigenetic alteration of the NIK gene. The stability of NIK mRNA and transcriptional activity driven by the NIK promoter are similar in the basal-like and luminal-like subtypes. However, histone H3 acetylation levels were up-regulated in the basal-like subtype. Furthermore, treatment of the luminal-like subtype with a histone deacetylase inhibitor, valproic acid, significantly increased NIK expression. Although DNA methylation of the NIK locus was not detected, NIK expression also increased when the luminal-like subtype was treated with 5-azacytidine, which inhibits histone H3-Lys-9 dimethylation in addition to DNA methylation. Taken together, these results suggest that the closed chromatin structure mediated by histone H3 methylation and deacetylation suppresses NIK expression in the luminal-like subtype, whereas disruption of these suppression mechanisms leads to enhanced NIK expression and the constitutive NF-κB activation in the basal-like subtype. Thus, NIK and genes induced by the NIK-mediated constitutive NF-κB activation could be therapeutic targets of basal-like breast cancer. (Cancer Sci 2010; 101: 2391–2397)
Breast cancers are categorized according to their profiles of gene expression as follows: luminal-like, ERBB2 overexpressing, basal-like, and normal breast like subtypes.(1) The luminal-like subtype is characterized as an estrogen receptor-positive (ER+) and a progesterone receptor-positive (PR+) phenotype. The basal-like subtype, which represents the ER−, PR−, ERBB2− phenotypes, shows a higher malignancy than other subtypes and exhibits a poor prognosis against various methods of therapy.(2,3) Therefore, it is necessary to find molecular signatures and signaling pathways that contribute to the malignancy of basal-like breast cancer.
The NF-κB family of transcriptional factors plays a critical role in inflammation, immunoregulation and cell differentiation.(4,5) This family consists of five members, including p50, p52, RelA, RelB and c-Rel, which form homomeric or heteromeric dimers to activate transcription of the target genes. NF-κB is made transcriptionally inactive by being sequestered in the cytoplasm when it forms complexes with the IκB family, including IκBα, IκBβ, IκBε and the p105 and p100 precursors of p50 and p52, respectively. Nuclear translocation of NF-κB can be driven by two distinct signaling pathways. In the canonical pathway, a large number of stimuli including various cytokines and bacterial and viral products induces IκB kinase (IKK) β-catalyzed phosphorylation and proteasomal degradation of IκBα, followed by nuclear translocation of mainly p50-RelA heterodimers. The non-canonical pathway is activated by receptors that are crucial in the formation of lymphoid organs and lymphocyte development such as the lymphotoxin (LT) β receptor, the receptor activator of NF-κB (RANK), and CD40. This pathway induces the IKKα-catalyzed phosphorylation of the C-terminal half of p100 that sequesters RelB in the cytoplasm, which leads to polyubiquitination-dependent processing of p100 to p52 and the nuclear translocation of the p52-RelB heterodimers.
Accumulating evidence indicates that aberrant NF-κB activation leads to tumorigenesis and cancer malignancy through the expression of genes involved in survival, metastasis and angiogenesis.(5,6) We have previously reported that the basal-like subtype, but not the luminal-like subtype, undergoes constitutive and strong activation of NF-κB, whose activation is transient in normal cells upon various physiological stimuli. Furthermore, reduction of the NF-κB activity of the basal-like subtype resulted in growth inhibition.(7) These results indicate that constitutive activation of NF-κB is involved in the malignancy of basal-like breast cancers and could be an effective therapeutic target.
Recently, several groups reported that NF-κB-inducing kinase (NIK), an activator of IKKα in the non-canonical pathway, is constitutively activated in a number of neoplasias. Multiple myeloma cells have several mutations and deletions in the TRAF2, TRAF3 and cIAP genes, whose products are normally involved in the degradation of NIK, resulting in the accumulation and activation of NIK proteins and thereby leading to constitutive activation of the non-canonical NF-κB pathway.(8) Aside from the post-translational regulation of NIK, NIK mRNA is significantly increased to constitutively activate the non-canonical pathway in adult T-cell leukemia and Hodgkin Reed–Sternberg cells.(9) We have also reported that the basal-like subtypes preferentially show high NIK mRNA expression, which is involved in the constitutive NF-κB activation. This result strongly suggests that inhibition of NIK expression is an effective therapeutic strategy to down-regulate the constitutive NF-κB activation in basal-like breast cancers. Therefore, the molecular mechanisms of regulation of the NIK mRNA expression in breast cancer need to be elucidated.
Materials and Methods
Cell culture, chemicals and adenovirus infection. The culture condition of all breast cancer cell lines has been described previously.(7,10) MG-132 (Calbiochem, San Diego, CA, USA), valproic acid (VPA) (Wako, Osaka, Japan) and 5-azacytidine (Sigma, St Louis, MO, USA) were purchased. Methods for adenovirus infection have been described previously.(7)
RT-PCR. RNA isolation and RT-PCR were performed as described previously.(11) The primer sets used were as follows: 5′-TTG GTT GGG GAG ATC GGC GCT TG and 5′-GGG GCT GAA CTC TTG GCT ATT CTC for NIK; 5′-GGT CAG AAG GAT TCC TAT GT and 5′-GGT GTT GAA GGT CTC AAA CA for β-actin; 5′-ATT GAG CGG AAG ATT CAA CT and 5′-TGT GGA TTT CTT GTC ATA GA for RelB.
Western blotting. Western blotting analysis was performed as described previously.(11) The antibodies used were anti-NIK (Cell Signaling Technologies, Beverly, MA, USA), anti-p52 (Millipore, Bedford, MA, USA) and anti-α-tubulin (Calbiochem).
Reporter construct and luciferase assay. The NIK promoter-driven luciferase reporter construct was generated by the insertion of several NIK promoter regions, which was determined based on the transcriptional start site of the NIK (MAP3K14) gene (NM_003954.2), into the pGL4.10 vector (Promega, Madison, WI, USA). Various NIK promoter regions were amplified from the human sperm genome DNA pool by PCR with appropriate primers. Luciferase assays were performed as described previously.(11)
Bisulfite sequencing analysis. Bisulfite sequencing analysis was performed as described previously.(12) The mixture of salmon sperm DNA (5 μg) and untreated or in vitro methylated NIK promoter (−8017/+76) in pGL4.10 (40 pg) using SssI methylase and S-adenosylmethionine (New England BioLabs, Beverly, MA, USA) was used as a negative or positive control, respectively. The NIK promoter region (−684/−313, −338/−79, −97/+451 and +432/783) of bisulfite modified MCF-7 cells genomic DNA (5 μg) and control samples were amplified using AccuPrime Taq DNA polymerase (Invitrogen, Carlsbad, CA, USA) with appropriate primers and subcloned in pT7blue vector (Novagen, Madison, WI, USA). Three independent clones of each region were sequenced.
Chromatin immunoprecipitation (ChIP) assay. A ChIP assay was performed as described previously.(13) The antibody against acetylated histone-H3 was purchased from Millipore. The immunoprecipitation efficiency of the NIK promoter region (−263 to −152) was analyzed by real-time PCR using a FastStart Universal SYBR Green Master (Roche, Indianapolis, IN, USA) according to the manufacturer’s instructions. The primer set used was 5′-ACA TCG TCC GGA AAT AGT GC and 5′-CCT ACG CCA ACC AAT GAG AC.
Statistics. Statistical significance was evaluated using a two-tailed Student t test. P-values <0.05 were considered significant.
NIK is highly expressed in basal-like subtype breast cancer cell lines. We have previously analyzed NIK mRNA expression in eight basal-like and two luminal-like subtypes of breast cancer cell lines and proposed that enhanced expression of NIK mRNA is observed more frequently in the basal-like subtypes.(7) Because the number of luminal-like subtypes analyzed in the previous paper was not sufficient to conclude that enhanced NIK expression is characteristic of the basal-like subtypes, we analyzed 12 luminal-like subtypes together with six basal-like subtypes. The basal-like subtypes showed higher NIK mRNA expression than all of the luminal-like subtypes tested (Fig. 1a, lower). Moreover, the expression levels of the NIK mRNA are closely correlated with the extent of the constitutive NF-κB activation reported in our previous data(7) (Fig. 1a). We then analyzed NIK protein levels in the presence of the proteasome inhibitor MG-132, because NIK protein is normally degraded so that it is kept at low levels by the TRAF2/TRAF3/cIAP complex in a proteasome-dependent manner.(14) The NIK protein is easily detected in the basal-like subtypes by the MG-132 treatment, whereas the NIK protein was scarcely detected in the luminal-like subtypes even in the presence of MG-132 (Fig. 1b). To further examine the correlation between NIK expression and the malignancy of the breast cells, we then compared NIK expression in a non-tumorigenic mammary epithelial cell line, MCF10A, with that of two breast cancer cell lines. Neither the NIK mRNA nor the NIK protein was detected in MCF10A, a non-tumorigenic mammary epithelial cell line, and MCF7 (luminal) cells, whereas NIK expression was dramatically enhanced at both mRNA and protein levels in MB231 (basal) cells (Fig. 1c,d). These results strongly suggest that the enhanced expression of NIK is a hallmark of basal-like subtype cell lines and is involved in the constitutive activation of NF-κB and in the aggressiveness of basal-like breast cancers.
Luminal-like and basal-like subtype cell lines exhibit similar NIK mRNA stability and transcriptional activity driven by the 5′ flanking region of the NIK gene. To elucidate the molecular mechanisms of enhanced NIK mRNA expression in the basal-like subtype, we first compared the stability of NIK mRNA between the luminal-like and the basal-like subtypes using actinomycin D to block new mRNA synthesis (Fig. 2a–c). The half-lives of NIK mRNA were approximately 3–4 h for both subtypes (Fig. 2c), indicating that mRNA stability is not significantly different between the luminal-like and basal-like subtypes. Additionally, we used published CGH-array data(15) to investigate NIK gene amplification. There are two probes (58.5 kbp upstream and 341.2 kbp downstream of the NIK gene) whose copy number may indicate that of the NIK gene. Intensities of these two probes indicate the normal copy number in several basal-like cell lines such as MB436, HCC1143, MB231, HCC1937, BT549 and HCC38, which are used in this study. Thus, these data suggest that the NIK gene is not amplified in the basal-like subtypes.
We then compared the transcriptional activity driven by the 5′ flanking region of the NIK gene. Progressive 5′ deletion mutations of the 8.1 kbp promoter were used to generate various reporter constructs (Fig. 2d). The two distinct regions, located between −71 and −25 and between −117 and −184, positively regulate the NIK promoter. However, transcriptional activities driven by the mutant promoters other than the −8017/+76 luc were similar between both subtypes (Fig. 2e). The −8017/+76 luc was activated more in the luminal-like than the basal-like subtype (Fig. 2e). However, we could not find any basal-like subtype-specific enhancer elements, which account for the enhanced NIK expression in the basal-like subtype, within the 8.1 kbp NIK promoter. Additionally, we examined the effect of NF-κB activity on NIK mRNA expression in the basal-like subtype because the activity of NF-κB is constitutively high in the basal-like subtype but not in the luminal-like subtype. While the expression of RelB, a known NF-κB target gene, was significantly reduced by the non-degradable IκBα super repressor (IκBαSR)-mediated inhibition of NF-κB in the MB436 (basal) cells,(7) NIK expression was hardly affected (Fig. 2f). Thus, the constitutive NF-κB activation is not a cause of enhanced NIK expression in the basal-like subtypes. Taken together, these data strongly suggest that transcriptional activity driven by the NIK promoter in the basal-like subtypes is similar to that in the luminal-like subtypes.
5-azacytidine and VPA induce NIK expression in the luminal-like subtype breast cancer cell lines. We did not detect any significant differences in the stability of NIK mRNA or in the transcriptional activity of the NIK promoter between the basal-like and the luminal-like subtypes, and thus speculated that NIK mRNA expression might be regulated by epigenetic alterations. To address this question, we first investigated CpG methylation on the NIK promoter. NIK mRNA expression was increased approximately fivefold by treatment with 5-azacytidine, a DNA methyltransferase inhibitor, in the luminal-like but not in the basal-like subtypes (Fig. 3a). However, bisulfite sequence analysis on the 1.4 kbp (−684/+783) fragment, which covers approximately 70% of the CpG island surrounding the NIK promoter, in MCF7 (luminal) cells revealed no CpG methylation (Fig. S1). It has recently been reported that RUNX3, a tumor suppressor gene, is suppressed by histone H3-Lys-9 dimethylation catalyzed by the G9a/G9a like protein 1 (GLP) complex without involving CpG methylation.(16) Furthermore, treatment of cells with a DNA methyltransferase inhibitor, 5-aza-2-deoxycytidine, reduced the level of histone H3-Lys-9 dimethylation and abolished this hypoxic RUNX3 silencing independently of cytosine methylation. Interestingly, our microarray data(17) indicate that expression of GLP, which is necessary for G9a/GLP histone methyltransferase activity,(18) was significantly reduced in the basal-like subtype when compared with the luminal-like subtype (Fig. 3b). These results strongly suggest that NIK expression is suppressed by histone H3-Lys-9 dimethylation in the luminal-like subtypes.
On the other hand, a ChIP assay revealed that histone acetylation in the 5′-regulatory region of the NIK promoter was significantly higher in MB231 (basal) cells than in MCF7 (luminal) cells (Fig. 4a). Moreover, treatment with the histone deacetylase (HDAC) inhibitor VPA significantly induced NIK mRNA expression in MCF7 (luminal) cells, but not in MB231 (basal) cells (Fig. 4b). Co-treatment with MG132 and VPA caused significant accumulation of the NIK protein in MCF7 (luminal) cells to a level similar to that in MG132-treated MB231 (basal) cells (Fig. 4c). Moreover, the VPA treatment scarcely affected activity of the NIK promoter in the transiently transfected reporter plasmid in MCF7 (luminal) cells (Fig. 4d). Processing of p100 to p52 was enhanced by the VPA treatment in MCF7 (luminal) cells (Fig. 4e), indicating that accumulated NIK protein by the VPA treatment activates the non-canonical NF-κB pathway. These data suggest that NIK expression is suppressed by HDAC-mediated histone deacetylation of the NIK promoter region in MCF7 (luminal) cells.
NIK expression is increased in sporadic basal-like subtype and BRCA-1-mutated primary breast cancers. To investigate NIK expression levels among the subtypes of primary breast cancer, we used a published dataset of human breast cancer specimens. Sørlie et al.(1) classified 97 breast cancer specimens by hierarchical clustering based on gene expression profiles that had been published previously.(19) These specimens were classified into 20 basal-like, 41 luminal-like, six ERBB2 over-expressing and four normal breast-like subtypes. Compared with total specimens, only the basal-like subtypes showed significantly elevated NIK expression (Fig. 5a). We then compared NIK expression in the total specimens with that in the specimens carrying BRCA1 germline mutations. Tumors with BRCA1 mutations showed a higher expression of NIK mRNA (Fig. 5b), which is consistent with the report that tumors with BRCA1 mutations are classified as basal-like subtypes based on their gene expression profiles.(20)
It is becoming increasingly apparent that gene expression is governed by epigenetic changes even without genetic abnormalities. Various aberrant epigenetic alterations result in gene expression patterns that are critical for cancer initiation and development.(21) Whether the chromatin structure takes an “open” or “closed” form is determined by histone acetyltransferases (HAT), HDAC and histone methyltransferases (HMT). Histone acetyltransferase-mediated acetylation of lysine residues in histones such as Lys-9 and Lys-14 in histone-H3 results in an open chromatin structure.(22) Histone deacetylase-mediated deacetylation leads to closed chromatin structure and transcriptional repression.(23) Histone methyltransferases transfer methyl groups to the lysine residues in histones such as Lys-4 and Lys-9 in histone-H3. In particular, Lys-9 methylation controlled by the G9a/GLP complex is implicated in gene repression.(24) Moreover, CpG dinucleotides in genome DNA can be methylated by DNA methyltransferases after the formation of the closed chromatin.(25)
In this paper, we demonstrate that NIK expression in the luminal-like subtype is significantly increased by treatment with VPA, a class I HDAC inhibitor (Fig. 4b,c), and that histone-H3 acetylation in the NIK promoter region is significantly higher in the basal-like rather than the luminal-like subtype (Fig. 4a). These data strongly suggest that NIK expression is constitutively suppressed by class I HDAC activity in the luminal-like subtype. We also demonstrate that treatment of 5-azacytidine, an inhibitor of CpG methylation and the G9a/GLP-catalyzed histone H3-Lys-9 methylation, increases NIK mRNA expression in the luminal-like subtypes (Fig. 3a), while no CpG methylation was observed in the NIK promoter region (Fig. S1). Moreover, expression of GLP was significantly reduced in the basal-like subtypes (Fig. 3b). Taken together, our results strongly suggest that NIK expression is suppressed by histone H3 deacetylation and histone H3-Lys-9 dimethylation in the luminal-like subtypes. NIK expression appears to be induced due to chromatin opening mediated by demethylation and acetylation of histone H3 in the basal-like subtype.
We analyzed NIK expression in primary breast cancer specimens in addition to the breast cancer cell lines. A significant augmentation of NIK expression was observed in the basal-like subtype specimens and tumors carrying BRCA1 mutations, which frequently display a basal-like phenotype.(20) There are several potential applications of this finding in the therapeutic strategy of human breast cancer. One strategy is to suppress NIK expression in basal-like breast cancers. We have previously reported that siRNA-mediated silencing of NIK resulted in reduced NF-κB activation, which is important for cell-cycle promotion in basal-like subtypes.(7) Therefore, proliferation of sporadic and BRCA1 mutation-induced basal-like breast cancers is expected to be inhibited by NIK suppression. The other strategy is to treat the basal-like subtypes with cIAP antagonists. cIAP is necessary for tumor necrosis factor-α–induced NF-κB activation in the canonical pathway, whereas cIAP induces degradation of NIK in the non-canonical pathway. cIAP antagonists induce degradation of the cIAP protein, which results in the accumulation of NIK proteins and non-canonical NF-κB activation, leading to the induction of TNF-α. Secreted TNF-α in turn induces apoptosis since NF-κB-mediated anti-apoptotic signaling in the canonical pathway is blocked by cIAP antagonists.(26,27) Our data showed that NIK is significantly accumulated by treatment with proteasome inhibitor MG132 only in the basal-like subtypes (Fig. 1d). Therefore, it is possible that cIAP antagonists induce sufficient accumulation of the NIK protein to activate the non-canonical pathway in the basal-like subtypes, which results in the abundant secretion of TNF-α to induce their apoptosis.
Finally, our results strongly suggest that expression of the NIK gene is regulated by its epigenetic alterations mediated by acetylation and methylation of histone H3. Further studies on the regulation of NIK expression in breast cancer cells are required to develop therapeutic strategies against basal-like breast cancers.
This study was supported by: Grants-in-aid for Scientific Research on Priority Areas (J.I.) and Young Scientists B (N.Y.), “Establishment of Consolidated Research Institute for Advanced Science and Medical Care” Project (N.Y., K.S.) and High-Tech Research Center; a Project for Private Universities: matching fund subsidy (K.S.) from the Ministry of Education, Culture, Sports, Science and Technology, Japan; a Waseda University Grant for Special Research Projects (N.Y., K.S.); the “Translational Research Program” from the New Energy and Industrial Technology Development Organization (S.W.); and research grants from the Princess Takamatsu Cancer Research Fund (J.I.) and Kanzawa Medical Research Foundation (N.Y.). M.Y. is a research fellow of the Japan Society for the Promotion of Science.