Background
Histone deacetylases (HDACs) are a class of total eighteen proteins (HDAC1–11 and SIRT1–7 in mammals) that deacetylate histones and non-histone proteins [
1]. HDACs play very important roles in diverse biological processes and related diseases such as cancers. High levels of HDACs is frequently associated with advanced cancers and poor prognosis [
2]. However, HDACs also display tumor suppressive effects in some cancers, e.g. an
HDAC2 truncating mutation was observed in human epithelial cancers [
3]. Low expression of HDAC10 is associated with poor prognosis of lung and gastric cancers [
4,
5].
Hdac2
−/− mice display a decreased intestinal tumors [
6] while liver-specific
Hdac3 KO mice develop hepatoma [
7]. Furthermore,
Hdac1 and
Hdac2 function as tumor suppressors on preleukemic stage, but oncogenes for leukemia maintenance in PML-RAR-mediated mouse acute promyelocytic leukemia [
8].
HDAC7 is a member of the HDAC family. Studies conducted on cell culture level by silencing or overexpressing
HDAC7 have shown that
HDAC7 is involved in the regulation of cell proliferation, apoptosis, differentiation and migration.
Hdac7
−/− mice are embryonic lethal due to a failure of angiogenesis (rupture of blood vessels) [
9]. There are only a few reports studying the role of
HDAC7 in cancers and even these scarce results were controversial. High HDAC7 protein level was observed in 9 out of 11 human pancreatic cancers [
10].
HDAC7 together with
HDAC1 is specifically over expressed in breast cancer stem cells (CSCs) and necessary to maintain CSCs [
11]. High
HDAC7 expression was associated with poor prognosis of 74 children with B-lineage CD10-positive acute lymphoblastic leukemia (ALL) (≥95% common-ALL/pre-B ALL) [
12]. All these results suggest an oncogenic function of
HDAC7 in these human cancers. However, low
HDAC7 expression was also observed in 75% of 28 pro-B-ALL samples [
13] and reported to be associated poor prognosis of lung cancer patients [
4]. This suggests a potential tumor suppressive function of
HDAC7 in some other cancers and/or at different stages of cancer development.
Signal transducer and activator of transcription 3 (STAT3) is a member of the STAT protein family. STAT3 activity is modulated by both acetylation and phosphorylation. In response to
cytokines or
growth factors, STAT3 is tyrosine-phosphorylated by receptor or nonreceptor kinases, forms dimmer and translocates into nucleus to activate the transcription of target genes implicated in a broad range of biological processes. In response to extracellular environmental factors including cytokines and nutrition, STAT3 acetylation is modulated by histone acetyl transferases/HDACs, e.g. p300/HDAC3, and therefore affecting its dimerization, phosphorylation, DNA binding and transactivation [
14]. The STAT3 mutated at key acetylation sites abolishes its ability of dimerization, and consequently impairs its tyrosine-phosphorylation and so forth [
14,
15]. STAT3 is thought to potently promote oncogenesis in a variety of tissues. The persistent tyrosine phosphorylation of STAT3 (pY-STAT3) and higher expression of STAT3 are observed in many human cancers and are often correlated with an unfavorable prognosis in these patients [
16]. However, a growing number of reports also suggest a tumor-suppressive function of STAT3 in some cancers. The pY-STAT3 level is negatively correlated with tumor size and distant metastases of papillary thyroid carcinomas [
17,
18]. High pY-STAT3 level is highly correlated with a better prognosis for soft tissue leiomyosarcoma, advanced rectal cancer and nasopharyngeal carcinoma [
19‐
21]. Studies on mouse cancer models also reveal that
Stat3 is a negative regulator for tumor progression in
Apc mutant mice and in
K-Ras mice [
22,
23]. Furthermore, STAT3 promotes oncoprotein EGFRvIII-induced glial transformation when it forms a complex with
EGFRvIII and conversely inhibits malignant transformation of astrocytes under
Pten deficiency condition [
24].
The role of
STAT3 in lung cancer development appears complex. The expression of constitutive STAT3 (STAT3C) induced inflammation and adenocarcinomas in mouse lung [
25]. STAT3 activation or high expression was initially reported to be associated with poor prognosis of lung cancer patients [
26,
27]. However, it has recently been demonstrated that low
STAT3 expression correlated with poor survival and advanced malignancy in human lung cancer patients with smoking history, and disruption of Stat3 signaling enhanced lung tumor initiation and malignant progression in mice [
23]. Furthermore, almost at the same time, Zhou, et al., have shown that
Stat3 can function as a tumor suppressor to prevent lung tumor initiation at an early stage of lung tumor development and an oncogene to facilitate lung cancer progression by promoting cancer cell growth at a late stage of lung cancer in the same
K-Ras mice [
28].
Here we report that Hdac7 functions as an oncogene in lung cancer. Higher HDAC7 protein level is observed in ~44% of human lung cancer samples and higher HDAC7 mRNA level is associated with poor prognosis of lung cancer patients. We also found that lung tumorigenesis was significantly inhibited in Hdac7
+/−/K-Ras mice. Hdac7 deficiency significantly inhibited proliferation and enhances apoptosis, respectively. The acetylation and phosphorylation of Stat3 were significantly enhanced in tumors from Hdac7
+/−/K-Ras mice and HDAC7-depleted human tumor cell lines. We demonstrated that the Hdac7 mutant-mediated tumor suppression was rescued by expressing dnStat3 in mouse lung tumors. Finally, our studies showed that HDAC7 directly interacted with and deacetylated STAT3. Our studies may shed a light on the design of new therapeutic strategies for human lung cancer.
Methods
Mice
Hdac7 PB (
PiggyBac) heterozygote mutant mice (thereafter called
Hdac7
+/− mice or
Hdac7 mutant mice), carrying an
Hdac7 mutant allele disrupted by insertion of a PB transposon in the intron between exon 1 and 2, was generated on the FVB/NJ background. Mapping information of PB insertion in
Hdac7 gene can be found in the PB mice database (
http://idm.fudan.edu.cn/PBmice).
LSL-K-Ras
G12D
mice (Stock No. 008179) were previously described [
29] and LSL-
K-Ras
G12D
allele was introduced into FVB genetic background through breeding with FVB mice for more than six generations (therefore called
K-Ras mice or control mice).
Hdac7
+/− mice were crossed with
K-Ras mice to generate
Hdac7
+/−/
K-Ras mice. All mice were maintained on 12/12-h light/dark cycles. Experiments were conducted with consent from the Animal Care and Use Committee of the Institute of Developmental Biology and Molecular Medicine at Fudan University, Shanghai, China.
Lung tumor induction, enumeration and tumor burden analysis
Lung tumors were induced by the method described previously [
30]. Briefly, 6-weeks-old
Hdac7
+/−/
K-Ras and control mice were infected with 5 × 10
7 plaque-forming units (PFU) of adenovirus (Ad-
Cre) expressing
Cre by intranasal inhalation, or 10
6 transforming unit (TU) of lentivirus (lenti-
Cre or lenti-
Cre-2A
-dnStat3) by tracheal instillation. Six weeks later, mouse lungs were retrieved and tumors on mouse lung surface were counted under dissection microscope. For tumor burden analysis, lungs were perfused through the trachea with 4% paraformaldehyde and fixed overnight followed by standard procedures for paraffin sections and H&E staining. Twelve randomly selected, lung sections/mice were scanned for 6 mice each genotype, total lung area occupied by tumor was measured and tumor burden was calculated as (area of lung section occupied by tumor)/(total area of section) in μm
2 using Image J (NIH, Bethesda, MD, USA).
TUNEL assay and EdU incorporation assay
TUNEL assay was performed following the manufacturer’s instructions for a kit (#G3250, Promega, Madison, WI, USA). For EdU (5-ethynyl-2′-deoxyuridine) assay, mice were injected intraperitoneally with 50 mg/kg EdU 6 weeks after Ad-Cre infection. Lung tissues were subjected to frozen section followed by EdU staining with a kit (#C10310–2, RiboBio, Guangzhou, China) 24 h after injection.
Plasmids
pFUW-Cre: A Cre fragment from MIGR1-Cre was cloned into of pFUGW to replace EGFP fragment. Plasmid pFUW-Cre-2A-dnStat3: A Cre fragment from pCS4-Cre and P2A–dnStat3 fragment from pCS4-dnStat3 were cloned into pFUGW to replace EGFP fragment. Plasmid pCMV6-Stat3-Flag: a PCR fragment of Stat3 was cloned into pCMV-Entry and in-frame fused to Flag tag. Plasmid pcDNA3.1-GST-Hdac7 (aa445–938): PCR fragments of GST and Hdac7 (aa 445–938) or Hdac7AWA (aa 445–938) were cloned into pcDNA3.1 and in-frame fused to Myc tag. Plasmid pLKO.1-shHDAC7: The selected ShHDAC7 targeting sequences were synthesized and cloned into pLKO.1. They are: shHDAC7 #1: ATCCGGGTGCACAGTAAATA, shHDAC7 #2: AAGTAGTTGGAACCAGAGAA, shHDAC7 #3: TCACTGACCTCGCCTTCAAAG.
Cell culture, virus preparation and infection
293T cells were cultured in as described [
31]. Human lung cancer cells A549, H1299, H2009 and H522 were cultured in RPMI 1640 with 10% FBS. Ad-
Cre was prepared and titrated as previously described [
32]. Recombinant lentiviruses were generated by co-transfecting lentivirus expression vectors and package plasmids (pCMV-VSV-G, pRSV-Rev and pMDLg/pRRE) into 293T cells and harvested 48 h later. Harvested lentiviruses for knockdown experiments were stored at −80 °C or directly used to infect human lung cancer cell lines.
Hdac7-silencing or scrambled cells were selected with puromycin. Lentiviruses for infecting mice were further concentrated by PEG 6000 as described previously [
33]. TU of lenti-
Cre and lenti-
Cre-2A
-dnStat3 was determine by infecting 293T cells carrying LSL-EGFP transgenes followed the methods published [
33].
MTT assay
Cells were plated at a density of 5000 cells/well in 96-well plate and cultured for desired time periods. The medium was then replaced by 200 μl of fresh RPMI 1640 and 20 μl MTT (5.0 mg/ml) and incubated for 4 h at 37 °C. 100 μl/well DMSO was added before reading OD 570 on a plate reader (Bio-Rad, Hercules, CA, USA).
Annexin V assay
H1299 cells infected shHAC7 or scrambled lentivirus were gently dissociated with trypsin/EDTA and stained for annexin V and 7AAD using the eBioscience™ Annexin V Apoptosis Detection Kit APC (#88–8007-74, Thermo Fisher Scientific) according to manufacturer’s instructions. The stained cells were immediately analyzed by flow cytometry (FACSCalibur, BD Biosciences). All annexin V positive cells were counted as apoptotic cells.
Soft agar assay
Soft agar assay was performed as previously described [
34]. Briefly, cells were plated at a density of 500 cells/well in 6-well plates and cultured for 14–21 days. Colonies were stained with MTT and counted using Colony Counter software (Tanon, Shanghai, China).
Quantitative PCR
Total RNA was extracted from wild type (wt) and Hdac7 mutant mouse lung tissue with TRIzol (Invitrogen, Carlsbad, CA, USA). Quantitative RT-PCR was performed with TaKaRa RNA PCR kit (TaKaRa, Dalian, China) and with Fast SYBR Green QPCR Master Mix on Mx3000P (Stratagene, San Diego, CA, USA) according to the manufacturer’s instructions and data were analyzed with the MxPro software. Expression of Gapdh was used as an internal control. The primer sequence is: Hdac7-F: CCCAGTGTGCTCTACATTTCCC, Hdac7-R: CACGTTGACATTGAAGCCCTC.
Antibodies, immunoblotting and immunoprecipitation
Antibodies against the following proteins were used for our studies: HDAC7 (#ab53101) and AKAP12 (#ab49849) from Abcam (Cambridge, UK); acetylated lysine (#9441), STAT3 (#8768), Phospho-STAT3 (Tyr705)(#9145), Acetyl-Stat3 (Lys685)(#2523), JAK1(#3344), JAK2(#3230), PKC(#2056), p53(#2524) from Cell Signaling (Danvers, MA, USA); anti-HA(#H3663), anti-FLAG(#F1804) and anti-β-Actin(A3854) from Sigma-Aldrich (St Louis, MI, USA); Myc (#SC-40) from Santa Cruz (Santa Cruz, CA, USA).
Standard Western blot protocol was adopted. Images were acquired with Tanon-5200 and the density of bands was determined with Image J. The methods for immunoprecipitation was previously described [
31].
GST pull-down assay and in vitro deacetylation assay
pCDNA3.1-GST-
Hdac7(aa445–938), pCDNA3.1-GST-
Hdac7(aa445–938)-AWA and pCMV6-
Stat3-Flag were transfected into 293 T cells separately. Cells were lysed with RIPA (Radio immunoprecipitation assay) buffer 48 h after transfection. GST-Hdac7(aa445–938) and GST-Hdac7(aa445–938)-AWA fusion proteins were purified using Glutathione Sepharose 4B (#17–0756-01 GE Healthcare), and Flag-tagged Stat3 was purified with ANTI-Flag M2 Affinity Gel (#A2220, Sigma) and eluted with 3X FLAG peptide (#F4799, Sigma). For in vitro pull-down assays, purified proteins were incubated together in Nonidet P-40 lysis buffer [
31] for 1.5 h, followed by washing with PBS for three times, and finally analyzed by Western blot.
In vitro deacetylation assay was performed as described previously [
35] with some modification. In brief, 0.5 μg purified Stat3 protein from 293 T cells was incubated with 0.5 μg purified GST-Hdac7(aa445–938) or GST-Hdac7(aa445–938)-AWA fusion proteins in deacetylation buffer (15 mM Tris–HCl, pH 8.0, 10 mM NaCl, 0.25 mM EDTA, 10 mM 2-mercaptoethanol, 10% (
v/v) glycerol) for one hour at 37 °C. The reaction products were subjected to Western blot with anti-acetylated lysine antibody.
Human lung cancer samples and prognosis analysis of lung cancer patients
Clinical lung cancer samples obtained from Huashan Hospital, Fudan University, Shanghai, China were used for immunoblot analysis.
For prognosis analysis of lung cancer patients, gene expression representative as FPKM (fragments Per Kilobase of transcript per Million mapped reads) derived from RNA-seq were downloaded from the TCGA project (
https://portal.gdc.cancer.gov/). Kaplan-Meier survival analysis and Cox’s proportional hazards regression were conducted by survival package in R (version: 3.2.3). Prognosis effect from
HDAC7 was estimated by Peto & Peto modification of the Gehan-Wilcoxon test and conducted by survdiff function from survival package (R). Hazard ratio (HR) and corresponding 95%CI were estimated with Cox’s proportional hazards regression.
Statistical analysis
Statistical analysis was conducted using an unpaired t test by GraphPad Prism (Graphpad, La Jolla, CA, USA). A p value < 0.05 was considered significant.
Discussion
Lung cancer is still the leading cause of cancer deaths worldwide [
46] due to lack of fully understanding of the molecular mechanisms of lung cancer development. Here we demonstrate that both lung tumor number and burden are dramatically reduced in
Hdac7
+/−/
K-Ras mice compared with those in control
K-Ras mice. We show that
HDAC7 silencing inhibits cell proliferation and anchorage-independent growth of human cancer cell lines. We have also observed higher HDAC7 protein level in ~ 44% human lung tumor samples and found that high
HDAC7 mRNA level in human lung cancer is correlated to poor prognosis. All these results from studies of mouse genetics, human lung cancer cell lines and clinic lung cancer patients strongly suggest that
HDAC7 play an oncogenic role in human lung cancer, but our conclusion is contradictory to a previous report by Osada, et al., claiming that high
HDAC7 mRNA level in human lung tumors was correlated to good progonosis [
4]. One possible reason for the discrepancy may come from the different samples size in these two studies. Only 72 human lung cancer samples were analyzed by Osada, et al., while data from 484 lung cancer samples from TCGA were evaluated by this study.
HDAC7 enhances the proliferation of HUVEC and cancer cells such as HeLa, HCT116 and MCF-7 probably by stimulating c-
Myc and inhibiting
p21 and
p27 expression [
37,
38].
HDAC7 can also protect mouse thymocytes and cerebellar granule neurons from apoptosis via repressing
Nur77 and
c-Jun expression, respectively [
39,
47]. In contrast,
HDAC7 has recently been reported to promote apoptosis of human pro-B-ALL and Burkitt lymphoma by down regulating
c-Myc expression [
13]. Our study shows that reduction of
Hdac7 expression in mice results in a decreased proliferation and increased apoptosis of mouse lung cancer cells. However, we only observed enhanced phosphorylation (activation) of Stat3 proteins but no obvious alteration at protein levels for Stat3, c-Myc, c-Jun, p21 and p53 (Additional file
1: Figure S1A). All these results suggest that
HDAC7 may regulate the proliferation and apoptosis of various cells through different molecular mechanisms.
During the period of preparing this manuscript, Peixoto, et al. reported that
HDAC7 high expression in glioblastoma (GBM) is associated with poor prognosis. They have also demonstrated that
HDAC7 silencing suppressed the tumor growth of GBM cell U87 in vivo mainly by inhibiting angiogenesis because
HDAC7 depletion had no effect on the proliferation of GBM cell U87 in vitro [
42]. Mechanistically, they also showed that
HDAC7-depletion inhibited angiogenesis by activating the expression of JAK1 and AKAP12, both of which can synergistically sustain the activity of STAT3 by inducing its phosphorylation (JAK1 tyrosine kinase) and protein expression (AKAP12) [
42]. Here we show that mouse
Hdac7 mutation suppresses lung tumor development in vivo and
HDAC7 silencing in human lung cancer cell lines inhibits their proliferation in vitro. Mechanistically, we have demonstrated
Hdac7 can directly interact with Stat3 and deacetylate Stat3 proteins, and decreasing
Hdac7 expression by mutation in mice or shRNA in human lung cancer cells results in enhanced acetylation and phosphorylation of Stat3 without significant effect on the expression of
JAK1 and
AKAP12 (Additional file
1: Figure S1B). Furthermore, STAT3 acetylation but not tyrosine phosphorylation has been shown to be required to silence expression of many tumor suppressor genes such as
SHP-1, CDKN2A by recruiting DNMT1 to methylate their promoter in cancer cells [
48‐
50]. Therefore, although
HDAC7-mediated angiogenesis may also play a role in lung tumor development, we think that reducing
Hdac7 expression is an important and sufficient factor to suppress lung tumorigenesis by inhibiting proliferation and enhancing apoptosis of tumor cells in mice, and probably in humans. This notion is supported by our observation that the expressions of cyclin D and cyclin E were significantly decreased in both lung tumors from
Hdac7
+/−
/K-Ras mice (Additional file
1: Figure S1A) and
HDAC7-depleted human lung cancer cell line H1299 (Additional file
1: Figure S1D). This notion is also further supported by recent findings that
HDAC7 expression is necessary to maintain breast and ovarian cancer stem cells in human and over-expression of
HDAC7 is sufficient to augment the CSC phenotype [
11].
Both positive and negative associations between STAT3 activation and survival of lug cancer patients or lung tumor progression have been reported [
23,
26,
27]. Furthermore, Zhou, et al., have recently shown that deletion of
Stat3 in
K-Ras mice enhanced lung tumor number but reduced lung tumor volume, suggesting that
Stat3 can function as a tumor suppressor and an oncogene at different stages of lung tumor development [
28]. Their study also showed that cell proliferation and
Cyclin D expression were decreased in the tumors from their
Stat3
−/−
/K-Ras mice in which
Stat3 was deleted. These results only provided cellular and molecular mechanisms underlying
Stat3 as an oncogene and failed to illustrate the cellular and molecular mechanisms underlying
Stat3 as a tumor suppressor. Here, our study results show that tumor number, tumor burden (Fig.
1c and d), and tumor size (unpublished data) were all decreased in
Hdac7
+/−
/K-Ras mice. Our study supports the notion that
Stat3 function as a tumor suppressor, but do not back the notion that Stat3 can also function as oncogene in lung tumor development because our results revealed that cell proliferation, Cyclin D and cyclin E expression were all decreased when Stat3 was more activated (Y705 phosphorylation) in the tumors from our
Hdac7
+/−
/K-Ras mice. Nevertheless, these two studies suggest that
Stat3 may function as a tumor suppressor in lung cancer under different genetic scenarios.
STAT3β, lacking a transcriptional activation domain at its C-terminal, is a product of differential splicing form of
STAT3 gene and displays a dominant negative function for the STAT3 [
51]. Zhang, et al., have shown that high STAT3β expression correlated with a favorable prognosis in patients with esophageal squamous cell carcinoma (ESCC). Expression of STAT3β substantially increased the Y705-phosphorylation, nuclear translocation, and DNA binding/promoter occupation of STAT3, but the transcriptional activity of STAT3 decreased by STAT3β [
52]. Here, we have not only shown that tumorigenesis was inhibited in
Hdac7
+/−
/K-Ras mice and Stat3 was more activated (more phosphorylation of Y705) in
Hdac7
+/−
/K-Ras tumors, but we also demonstrated that suppression of endogenous Stat3 activity in lung tumor cells, by expressing
dnStat3, reversed
Hdac7 mutant-mediated reduction of tumor number and burden in
Hdac7
+/−
/K-Ras mice. These results rule out the possibility that reduced tumorigenesis in
Hdac7
+/−
/K-Ras mice is due to increase expression of Stat3β because, if that is the case, expressing
dnStat3 should result in further reduction rather than enhanced tumorigenesis in
Hdac7
+/−
/K-Ras mice.
The catalytic activity of class IIa HDACs (including HDAC4, −5, −7 and −9) was initially attributed to the presence of class I HDACs co-purified along with the class IIa HDACs. Later, the catalytic domains of HDAC4 and HDAC7 purified from
E. coli that lacks histones and endogenous HDACs were demonstrated to have a low but measurable deacetylase activity on acetylated-lysine of histones [
53,
54] and a high deacetylase activity on trifluoroacetyl lysine, a class IIa-specific substrates in vitro [
53‐
55]. One of possible reasons proposed to explain the low catalytic activity of class IIa HDACs is that acetylated-lysine of histones is not a biological substrate for class IIa HDACs [
54]. Now we demonstrate that loss of
Hdac7 resulted in significantly enhanced Stat3 acetylation in both mouse primary tumors and human tumor cell lines. Our co-immunoprecitaion and pull down assay also show that HDAC7 protein directly interacts with Stat3 and the HDAC7 catalytic domain with AWA mutations failed to deacetylate Stat3 in the in vitro deacetylase assay. All these data suggest that the Stat3 protein may be a biological substrate for class IIa deacetylase HDAC7. However, the possibility that contribution of contamination of class I deacetylases such as HDAC3 cannot be ruled out completely since the HDAC7 catalytic domain used for our in vitro deacetylase assay was purified from mammalian 293T cells. Further experiments using HDAC7 purified from
E.coli and/or class I-selective HDAC inhibitors will help to solve whether Stat3 is a biological substrate for class IIa deacetylase HDAC7.