Background
Colorectal cancer (CRC) is one of the most common malignancies worldwide [
1]. Despite the progress in early diagnosis and therapy in the past decade, the overall survival of CRC patients remains low in the setting of metastasis [
2]. Metastasis is a complex process, and many cell-intrinsic identities and extrinsic microenvironment factors influence the metastatic potential of CRC cells [
3]. A better mechanistic understanding of metastasis may provide novel therapeutic targets to help improve the overall survival of CRC patients.
Epithelial to mesenchymal transition (EMT) is characterized by the loss of cell-cell adhesion and the gain of migratory and invasive traits [
4]. EMT has been reported to play important roles in many physiological and pathological processes [
5]. EMT frequently occurs at the initial stage of cancer metastasis. During EMT, epithelial cells evade extracellular constraints from adhesion molecules such as E-cadherin. Meanwhile, epithelial cells upregulate the expression of mesenchymal markers, such as fibronectin and ITGα5, to increase cell motility [
6]. EMT and related molecules have been suggested as potential prognostic or therapeutic targets in cancer [
7]. However, mechanistic understanding of EMT in metastasis is still limited.
Histone deacetylases (HDACs) are enzymes that catalyze the removal of acetyl groups from lysine residues in histones to suppress gene transcription. HDACs also have non-histone substrates. Eighteen classes of HDACs have been identified in mammals [
8], and are categorized as HDAC1–11 and SIRT1–7. Deregulated expression of HDACs in cancer has been extensively documented. Aberrant expression of classical (class I, II, IV) HDACs has been linked to the initiation and progression of a variety of cancers [
8]. In most cases, a higher level of HDACs is associated with advanced disease and poor outcome in tumor patients, presumably due to the loss of tumor suppressor functions [
9‐
11]. However, mutations and reduced expression of HDACs have also been reported in cancer.
HDAC1 somatic mutations were detected in 8.3% of dedifferentiated liposarcomas, and
HDAC4 homozygous deletions occurred in 4% of melanoma [
12,
13]. Low HDAC10 expression was associated with poor prognosis in lung and gastric cancer patients [
14,
15]. HDAC6 was downregulated in human hepatocellular carcinoma (HCC) tissues, and it down regulation was associated with poor prognosis in liver transplantation patients.
HDAC6 knockdown promoted angiogenesis in HCC [
16]. Truncating mutations and the loss of HDAC2 protein expression have been observed in human epithelial cancers with microsatellite instability [
17]. Furthermore, the ectopic expression of HDAC2 in
HDAC2 mutant cancer cells inhibited tumor cell growth in vitro and in vivo [
18]. Collectively, these findings support potential tumor suppressive functions of HDACs in cancer initiation and maintenance.
The roles of HDACs in EMT and cancer metastasis are context-dependent. HDAC1 promoted EMT in gallbladder cancer by binding with TCF12 [
19]. HDAC1 and HDAC2 were recruited by the transcriptional repressor ZEB1 to downregulate E-cadherin expression in pancreatic cancer [
20]. SIRT1 induced EMT and enhanced prostate cancer cell migration and metastasis by cooperating with ZEB1 [
21]. HDAC inhibitors (HDACis) have been developed as potential anticancer agents, and can restore the expression and function of tumor suppressors [
8]. However, HDACis also promote tumor progression, EMT and cancer metastasis in some models [
22]. Cell migration was dramatically enhanced by various classes of HDACis in 13 of 30 human breast, gastric, liver, and lung cancer cell lines examined in a dose-dependent manner [
23]. Metastasis was also enhanced in HDACi-treated mice through the activation of multiple PKCs and downstream substrates along with upregulated p21 expression [
23]. Nonselective HDACis such as Trichostatin A (TSA) and valproic acid (VPA) induced mesenchymal features in colon carcinoma cells with decreased expression in E-cadherin and increased expression in vimentin [
24].
In this study, we found that reduced expression of HDAC2 in CRC metastasis is associated with poor patient survival. HDAC2 deletion in CRC cells increased EMT-mediated metastasis in vivo and in vitro. Mechanistically, HDAC2 suppressed EMT and CRC metastasis through the inhibition of the H19/MMP14 axis. Taken together, these results establish HDAC2 as a novel metastasis suppressor in CRC.
Materials and methods
Dataset
The relative expression data of HDAC1, HDAC2, HDAC3 and HDAC8 were downloaded from the Oncomine public database (
www.oncomine.org). The dataset contains 330 primary sites and 43 metastatic CRC tissue samples.
Survival analysis of HDAC1, HDAC2, HDAC3 and HDAC8
We labeled TCGA samples as “high” or “low” according to whether the expression of HDAC1, HDAC2, HDAC3 and HDAC8 was higher or lower than the corresponding median value among all samples. The log-rank test was used to measure whether the survival time was significantly different between the “high” and “low” expression groups. Kaplan-Meier plots were made by Gene Expression Profiling Interactive Analysis (GEPIA:
http://gepia.cancer-pku.cn/).
CRC tissue sample and immunochemistry
Commercially available tissue microarray (TMA) slides (HLin-Ade075Met-01, Shanghai Biochip Co., Ltd., Shanghai, China) containing histologically confirmed tissues from CRC patients were purchased for immunohistochemistry (IHC) analysis. Specific primary antibodies against HDAC2 (Cell Signaling Technology) were used for IHC with a 2-step protocol.
Cell culture
DLD1, HCT116, SW480 and SW620 cells were obtained from ATCC. DLD1 and HCT116 cells were cultured in RPMI 1640 medium, while SW480 and SW620 were cultured in high glucose DMEM. All media were supplemented with 10% FBS, 100 μg/m L penicillin and 100 U/m L streptomycin. The DLD1
HDAC2 KO cell line was constructed in the lab of Professor Run-lei Du, who is our collaborator in this study [
25].
Microarray analysis
The total RNA from DLD1 and DLD1 HDAC2 KO cells was prepared for microarray analysis (n = 3 each). The Affymetrix microarray was used to detect mRNA and long noncoding RNA expression profiles. Microarray data were normalized using the RAM (robust multiple-array average) normalization method. The differentially expressed genes were determined with a threshold cutoff of 2-fold (p < 0.01).
Transwell migration assays
Tumor cell migration assays were performed according to the manufacturer’s instructions. Briefly, cells were harvested and resuspended in serum-free medium and then seeded onto Transwell inserts at a density of 100,000 cells/well. Then, the inserts were placed in a lower chamber filled with 600 μl culture media containing 10% FBS. Transwells were incubated for 24 h at 37 °C. Cells on the inside of the Transwell inserts were removed with a cotton swab; then, cells that migrated to the lower surface of the membrane were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. Photographs were taken from five random fields, and the cells were counted to calculate the average number of cells that had transmigrated.
Vector construction and luciferase reporter assay
The H19 promoter containing intact SP1 recognition sequences was PCR-amplified and subcloned into the KpnI and HindIII sites of the pGL-3-basic vector, and the vector was named pGL-3-H19. The pGL3-H19 vector with point mutations in the SP1 binding sites was synthesized by GenScript (Nanjing, China) and named pGL3-H19-mut (SP1). For the luciferase reporter assay, HEK293 cells cultured in 24-well plates were cotransfected with luciferase reporter plasmids and HDAC2 plasmids. Twenty-four hours posttransfection, HEK293 cells were lysed in lysis buffer. After centrifugation at 12,000 rpm for 5 min, the supernatant was transferred to a new tube. The luciferase activity was monitored by mixing 10 μl supernatant with 30 μl luciferase assay buffer and using a GloMax 20/20 Luminometer (Promega).
siRNA and gene transfection
The siRNAs were synthesized by RiboBio Company (Guangzhou, China). Oligonucleotide transfection was conducted by using Lipofectamine™ RNAiMAX transfection reagent (Life, USA) following the protocol recommended by the manufacturer. After 48 h posttransfection, the cells were collected and used for further investigations.
Chromatin immunoprecipitation (ChIP) and q PCR
ChIP was performed using a SimpleChIP® Enzymatic Chromatin IP Kit (CST) following the manufacturer’s instructions. Briefly, genomic DNA-protein complexes were immunoprecipitated using anti-HDAC2 antibody or normal rabbit IgG as a control. After enzyme digestion and sonication, the precipitated DNA was amplified by SYBR Green-based quantitative real-time PCR using primers encompassing the promoter regions of the
H19 gene. The ChIP PCR primers used were (the numbers in parentheses indicate the sequence regions corresponding to GenBank ID AF125183):
-
Primer 1: 5′-CCAGCCATGTGCAAAGTATG-3′ (9747–9766)
-
Primer 2: 5′-CCATCCTGGAATTCTCCAAA-3′ (9939–9920)
-
Primer 3: 5′-GCGGTCTTCAGACAGGAAAG-3′ (9468–9487)
-
Primer 4: 5′-CACGTTCCTGGAGAGTAGGG-3′ (9673–9654)
Co-immunoprecipitation
A co-immunoprecipitation assay was performed in the following steps. Briefly, the cells were washed with ice-cold PBS, lysed in NP-40 buffer containing cocktail and then centrifuged for 10 min at 12000 rpm and 4 °C. Anti-SP1/HDAC2 antibody or normal rabbit IgG was added to the cell lysate and incubated at 4 °C overnight. Then, 15 μl of protein A/G agarose beads was added to each tube, incubated at room temperature for 3 h and centrifuged for 3 min at 4000 rpm at 4 °C. A total of 30 μl of 2× SDS-loading buffer was added to the antigen-antibody-protein A/G agarose bead complex, which was boiled for 10 min. The sample was collected for subsequent SDS-PAGE and Western blotting.
RNA-binding protein immunoprecipitation
The anti-Ago2 RIP assay was performed using a Magan RIP™ RNA-Binding protein Immunoprecipitation Kit (Milipore) following the manufacturer’s instructions. Briefly, DLD1 HDAC2 KO cells were washed with cold PBS and lysis by RIP Lysis Buffer. After that, the cell lysates were incubated with antibody against Ago2 (Milipore, USA). The normal Mouse IgG was used as negative control. For RNA immunoprecipitation, the supernatant was incubated with the antibody-coated Sepharose beads overnight. The RNA bound to Ago2 antibody was extracted with TRIzol reagent (Invitrogen, USA) and detected by qRT-PCR.
Fluorescence in situ hybridization (FISH)
The sequence of the H19 probe was 5′-FAM/GCTGCTGTTCCGATGGTGTCTTTGATGTTGGGC/FAM-3′; this probe was used for FISH of H19 from B-NDG mouse pulmonary metastases. After the metastatic lung tissues were removed and cleaned, they were immediately fixed in an ISH fixation solution (DEPC water preparation) for 12 h. After tissue fixation, they were dehydrated by gradient alcohol and then embedded in paraffin. The tissue sections were used for H19 probe hybridization, and the nuclei were restained by DAPI. The sections were observed under a fluorescence microscope, and images were collected. The H19 probe is shown in green, while DAPI staining is shown in blue.
Animal study
B-NDG immunodeficient mice (with T cell, B cell and NK cell defects) were obtained from Beijing Biocytogen Biotechnology Co., Ltd. The B-NDG mice were randomly divided into two groups. A total of 5 × 106 luciferase-labeled CRC cells were injected intravenously into B-NDG mice via the tail vein. Four weeks later, the mice were anesthetized and injected intraperitoneally with fluorescein potassium salt (150 mg/kg), and 10 min later, the metastatic tumor was detected and photographed by a bioluminescent in vivo imager (VILBER Fusion FX7, France). The mice were sacrificed, and their lung tissues were removed for H&E staining, immunohistochemistry and FISH. The mean number of metastatic nodules in B-NDG mice with lung metastasis was calculated.
Statistical analysis
All experiments were repeated no fewer than 3 times. Experimental results are presented as the mean ± S.E.M. The statistical significance of comparisons between two groups was determined with a two-tailed Student’s t-test. P<0.05 indicated statistical significance.
Additional materials and methods
Discussion
Here, we reported a novel function of HDAC2 in suppressing CRC metastasis. The expression of HDAC2 was reduced in CRC metastasis tissues, and reduce expression is associated with poor survival of CRC patients. Loss of HDAC2 expression promoted EMT-mediated CRC metastasis via the H19/MMP14 axis (Fig.
6g). HDACs have been reported to be highly expressed in many tumors, and the development of cancer drugs targeting HDACs has been carried out for many years [
32]. Unfortunately, clinical trials have shown that HDAC inhibitors, as single agent, do not benefit patients with solid tumors including colorectal cancer [
33]. Several HDAC inhibitors were reported to induce EMT and metastasis [
24,
34]. These findings are consistent with our results, and support that some HDACs might have tumor suppressive functions. Therefore, more selective HDACi are likely needed in order for further clinical development in solid tumors.
HDACs are extended families of proteins regulating gene expression and cell physiology through many targets. We focused our studies on class I HDACs in CRC metastasis upon the finding of a negative correlation of HDAC2 expression and CRC patient survival. HDACs have been reported to regulate EMT and cancer metastasis either positively or negatively. Typical EMT markers include cell adhesion molecules such as E-cadherin, Vimentin, N-cadherin, ITGA5and Fibronectin, and transcription factors such as Snail1/2, ZEB1/2, and ZO-1 [
4,
35].Most HDACs were found to promote EMT and metastasis by binding to EMT-related transcription factors and directly inhibit the expression of epithelial markers such as E-cadherin. In this study, we also found a decrease of E-cadherin and increase of Fibronectin and ITGA5, but not an increase of Vimentin, ZEB1 and ZEB2 in the DLD1 HDAC2 KO microarray data. The reason is that not all types of cells or tissues share the same EMT markers, and this may be a feature of EMT induced by HDAC2 loss in CRC cells. Furthermore, we found a novel function of HDAC2 in suppressing CRC metastasis. HDAC2 regulates EMT and metastasis indirectly by inhibiting lncRNA H19 and MMP14. These findings reinforce context-dependent role of HDACs, and highlight the critical need to understand their cancer type specific roles.
LncRNA H19 is the first described human lncRNA and implicated in cancer initiation, progression and metastasis [
36]. Especially, H19 has been reported to up-regulated in CRC tissues compared with non-tumor tissues and H19 overexpression is closely associated with poor survival of CRC patients [
37‐
39]. Besides, several studies showed that H19 was upregulated in metastatic cancer tissues and could promote cancer metastasis, including colorectal cancer [
29,
40,
41]. However, the mechanistic insights of deregulated H19 expression in cancer are limited [
42,
43]. Our study firstly reported an inverse correlation between HDAC2 and H19 expression levels in CRC and verified H19 as a possible downstream target of HDAC2 in promoting EMT and MMP14 expression. H19 promotes EMT and cancer metastasis through a variety of mechanisms. H19 promoted glioma cell invasion by deriving miR-675 [
27]. H19 promoted EMT by sponging miRNAs in colorectal cancer [
29]. H19 increased bladder cancer metastasis by binding to EZH2 and inhibiting E-cadherin expression [
28]. In this study, we found that H19 promotes EMT and CRC metastasis by sponging miR-22-3P and upregulating MMP14. These findings support that HDACs and LncRNAs form gene expression regulatory networks to control CRC development. Actually, microarray data showed
HDAC2 deletion affects the expression of thousands of genes in CRC cells. Apart from the lncRNA H19 signaling axis, HDAC2 maybe still has other underlying molecular mechanisms in the regulation of CRC EMT and metastasis, and need more experiment to verify it in the future.
Several evidences showed that the expression of HDAC2 in CRC tissues is up-regulated compared with normal colon tissues [
44‐
46]. HDAC2 up-regulation has been reported to be a novel and important early event in CRC [
44]. These results indicate an oncogenic role of HDAC2 in CRC. However, in this study, we found HDAC2 expression decreases in metastastic CRC (compared with primary site CRC), and reduced HDAC2 expression is associated with poor survival of CRC patients. Actually, the occurrence and metastasis of cancer are two independent events. We think HDAC2 plays a dual role in the development of CRC, that is promoting the initiation of CRC but suppressing CRC metastasis. Explore the origins of dynamic HDAC2 expression in CRC is a valuable question in the future.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.