Background
The wnt-β-catenin pathway is a pro-growth and survival pathway that is active in multiple tumor types including chronic lymphocytic leukemia (CLL) [
1‐
3]. Activation of this pathway in CLL is the result of high wnt and frizzled expression [
4] along with epigenetic down regulation of wnt pathway antagonist genes including secreted frizzled-related protein (SFRP) family members, WIF1, DKK3 and APC [
5,
6]. The binding of wnts to their cognate receptors results in inhibition of GSK3β phosphorylation of β-catenin and its degradation. Stabilized β-catenin then translocates to the nucleus and interacts with lymphoid-enhancing (LEF) and T cell (TCF) transcription factors to activate transcription of wnt-target genes that include myc, LEF, cyclinD1, Cox-2, matrix metalloproteinase family members etc. [
7‐
11]. E-cadherin expression is also able to inhibit the β-catenin translocation to the nucleus as its intracytoplasmic domain binds β- catenin [
12,
13]. In a previous study we identified silencing of the E-cadherin gene in CLL specimens as an additional mechanism of wnt pathway activation [
14] as the ectopic expression of E-cadherin was sufficient to inhibit the active wnt-β- catenin pathway. Frequent loss of function of E-cadherin in CLL specimens and the activation of the wnt pathway highlights its role in CLL biology [
14]. Also as inhibition of the wnt pathway results in apoptosis of CLL cells this is an important pathway for developing treatment strategies for this disease [
15,
16].
Epigenetic modifications such as DNA methylation and histone modifications silence a number of genes and are involved in leukemia initiation and progression [
17,
18]. These modifications are heritable, reversible and alter expression patterns of a number of genes without altering any DNA sequences. In the case of chronic lymphocytic leukemia (CLL) a number of genes are reportedly silenced by epigenetic alterations including the wnt pathway inhibitor genes [
5,
6] and more recently micro RNA expression was found to be modulated by epigenetic changes [
19]. The inhibitors of these DNA epigenetic modifications are promising anticancer agents that allow re- expression of silenced genes, cell cycle arrest and apoptosis [
20,
21]. Histone deacetylase inhibitors (HDACi) are an example of drugs that are able to reverse epigenetic events and have useful clinical activity in various hematopoietic malignancies [
21‐
23] including CLL where exposure to these drugs results significant apoptosis [
24‐
26]. They increase the acetylation of the major histones H3 and H4 by inhibiting the histone deacetylases (HDACs) which leads to change in overall compactness of the chromatin and promotes accessibility of the DNA to the transcription factors and gene transcription [
22].
In this report we explored the possibility that E-cadherin down regulation could be due to epigenetic modifications and studied the effect of HDACi on CLL specimens. Our previous report indicates that E-cadherin down regulation in CLL could be due to excessive aberrant splicing resulting in an alternatively spliced, non-functional E- cadherin transcript that lacks exon 11 of the gene [
14]. This non-functional transcript has a premature termination codon and is degraded by the NMD pathway [
27,
28]. We explored this issue of aberrant splicing in the HDACi treated CLL specimens as well.
Methods
Cell culture and reagents
CLL specimens were obtained from CLL patients at the West Los Angeles VA hospital clinic after informed consent and an approval by the West Los Angles VA Medical Center Institutional Review Board. CLL cells were isolated by Ficoll-Paque gradient (GE healthcare, Piscataway, NJ) as manufacturer’s protocol and stored in liquid nitrogen. Specimens selected for analysis had more than 90% CLL cells in the PBMC isolate. For control samples, peripheral blood was obtained from four normal donors and PBMC (peripheral blood mononuclear cells) cells were isolated. HDACi MS-275 was obtained from ChemieTek, Indianapolis, IN and dissolved in DMSO. For HDACi treatment stored CLL specimens were taken in culture in RPMI1640 media with 10% fetal calf serum (complete media) and HDACi was added for 48 hours. To inhibit the NMD pathway and treat CLL cells with HDACi, cells were first treated with HDACi for 48 hours and emetine was added at a final concentration of 10μg/ml for the last eight hours. As a control, cells were treated with emetine alone for 8 hours. Annexin assay for analyzing apoptotic cells after HDACi treatment was performed with the Annexin V Apoptosis Detection kit II (BD Pharmingen, CA) as per manufacturer protocol.
Western blot and immunoprecipitation analysis
For western blot analysis cells were washed with cold PBS and disrupted in lysis buffer (Cell Signaling, MA) supplemented with protease inhibitor cocktail (Thermo Fisher Scientific, MA). They were lysed on ice and then sonicated for 10 seconds. Insoluble material was removed by centrifugation (10,000 g, 10 min) and protein concentrations determined by BioRad DC protein assay (Hercules, CA). Samples were mixed with SDS sample buffer and 20–30 μg aliquots resolved on SDS/PAGE gels. Following transfer to PVDF membranes immunodetection was performed with E-cadherin (BD Pharmingen, CA), Ac H3 (06–599, Millipore), Ac-H4 antibody (sc-34263, Santa Cruz Biotechnology).
Detection was performed with horseradish peroxidase-conjugated secondary antibodies and chemiluminescence (ECL plus, GE Healthcare and LAS Mini imager, Fuji). Immunoprecipitation with beta-catenin antibody (Cell Signaling, MA) was performed with the Classic IP kit by (Thermo Fisher Scientific, MA).
Real time PCR and transcript specific PCR analysis
E-cadherin expression in cells was analyzed by real time PCR analysis. The wild type E-cadherin transcript was quantified by real time PCR with the 5’ primer GGATGTGCTGGATGTGAATG that localizes to exon 10 of the E-cadherin gene, the 3’ primer CACATCAGACAGGATCAGCAGAA localizes to the exon 12 and the taqman probe TAACATATCGGATTTGGAGAGAC for the wild type E- cadherin transcript level, binds to the junction of exon 10-exon 11. The expression level of the skipped or aberrant transcript (transcript lacking exon 11) was determined by a 5’ primer that is at the junction of exon 10 and 12 (TATGGAACAGAAAATAACGTTC) and a 3’ primer in the exon 12 (TGTCATTCACATCAGACAGGAT) with a taqman probe (AACAGGGACACTTCTG). This primer set only amplifies the exon 11 skipped or the aberrant transcript. The ratio of these two transcripts in cells was calculated by the difference in their Ct values. Taqman probes for HDAC 1–9, LEF1 and cyclin D1 (Applied Biosystems, CA) were used to determine their relative expression in CLL and PBMC cells. Real time PCR for actin expression was used as a control and relative expression was determined by the method of Pfall [
29].
Chromatin immunoprecipitation assay
ChIP assays were performed according to protocols of the EZ-ChIPTM Chromatin Immunoprecipitation kit (Millipore, MA). Briefly, 1×106 CLL cells were fixed with 1% formaldehyde for 10 min at 37°C. The cells were washed extensively with PBS, and the chromatin was sheared by sonication (Branson sonicator) to 200–400 bp fragments. The cross-linked histone-DNA complex was immunoprecipitated with anti pan Acetylated-H4 (sc-34263, Santa Cruz Biotechnology), anti pan Acetylated-H3 antibodies (06–599, Millipore). Normal mouse IgG was used as negative controls. DNA was obtained from the cross-linked complex and was amplified by a SYBR green based real-time PCR (Applied Biosystems, CA reagents) with specific primer sets. Primer sequences will be made available on request. Data sets were normalized to ChIP input values. The percent of input DNA immunoprecipitated by each antibody was calculated using the following method: % input = 100*2Ctinput – CtIP. (Ct value of input DNA-Ct value of immunoprecipitated DNA).
Wnt reporter assay
Wnt activity in CLL specimens was determined by a luciferase reporter assay kit (SABiosciences, CA). This kit consists of a reporter construct expressing firefly luciferase with TCF/LEF binding sites and an identical negative control vector lacking TCF/LEF binding sites. Both constructs were transfected with a constitutively active renilla luciferase construct to control for transfection efficiency. CLL cells were thawed, cultured in complete medium for 24 hours, and then transfected by the Amaxa protocol (Nucleofector kit V, program U07). 8 hours after transfection cells transfected with the reporter construct were split into three wells. One well was left untreated and the other two were treated with either HDACi MS-275 (1μM concentration) or the GSK-3β inhibitor SB-216763 (1μM concentration) for another 24 hours. Cells were then lysed and analyzed by a Dual luciferase assay (Promega, Madison, WI). Data expressed as ratios of firefly to renilla to control for transfection efficiency.
Discussion
Our study with primary CLL specimens demonstrates that the frequent loss of E- cadherin gene expression in CLL specimens is due to histone epigenetic silencing of the E-cadherin promoter. This gene silencing can be reversed by HDACi that acetylate the histones in the promoter region of this gene and activate E-cadherin gene transcription. Furthermore this HDACi induced E-cadherin expression inhibits the wnt- β-catenin pathway in CLL by interacting with β-catenin and thereby inhibiting its transactivation function. An additional interesting finding of the study is the effect of HDACi on E-cadherin splicing as there is a preferential expression of correctly spliced E-cadherin transcripts.
HDACi increase the E-cadherin expression in a majority of CLL specimens as determined by real time PCR and this increase in the transcript is also translated into an increase in E-cadherin signal on the western blot analysis. The induction of E-cadherin expression by western blot analysis is variable and a potential reason could be the pro-apoptotic effect of the HDACi that inhibits protein translation in some CLL specimens. Lack of response to HDACi in two out of ten CLL specimens could be due to other mechanisms including E-cadherin promoter methylation that has been reported [
34]. With this significant increase in wild type E-cadherin RNA transcripts with HDACi we investigated whether there was a qualitative change in the RNA transcripts as a number of cellular alterations induced by HDACi such as gene transcription, change in chromatin structure are also known to alter gene splicing patterns [
30‐
32].
To study this we analyzed the HDACi treated CLL specimens for the two E- cadherin transcripts. Analysis of a number of CLL specimens shows a variable response of the exon 11 skipped transcripts to HDACi exposure. In some CLL specimens there is a complete loss of expression of this transcript and in some cases a decrease in expression of aberrant transcript while the total wild type transcript consistently increased in amount in the majority of CLL specimens (Figure
2C). In two CLL specimens the absolute amount of aberrant transcript also increases with HDACi exposure albeit the fold induction is less than the wild type transcript can be explained by a more robust HDACi effect on overall E-cadherin transcription than its ability to alter aberrant exon 11 splicing. Overall we observe a much smaller increase in the exon 11 skipped transcripts as compared to the correctly spliced transcript with HDACi exposure. This change in the ratio of the two transcripts is not due to a change in the activity of the NMD pathway as experiments with NMD blocker emetine and HDACi show similar change in ratios. With NMD blockade as there is no degradation of the aberrant transcript, both the transcripts should have a parallel increase in fold induction. The lack of similar fold induction of the two transcripts in the two CLL specimens with NMD blockade supports the argument that with HDACi there is a qualitative change in splicing pattern along with transcriptional induction.
Histones play an important role in splicing, they are positioned in a non-random manner on the genome with a higher concentration at the exon-intron junction and certain histone modifications are preferentially seen in the exons as compared to the surrounding introns [
35]. Histone epigenetic alterations are mainly found in the promoter regions but can also be detected in exonic regions [
36,
37]. With more than 95% of genes undergoing alternative splicing [
38,
39] in the human genome epigenetic histone deacetylation and other modifications such as methylation in cancer cells can alter splicing for the vast majority of genes and thereby alter the sequences of many RNA transcripts. This alteration in splicing can result in RNA isoforms and proteins with entirely different biological characteristics and is frequently observed in cancer cells [
40,
41]. For example the Bcl-x transcript is alternatively spliced to produce the anti-apoptotic Bcl-x (L) or the proapoptotic Bcl-x(s) and cancer cells often up regulate the anti-apoptotic Bcl-x (L) isoform which is associated with reduced sensitivity to chemotherapeutic drugs [
42]. The malignant cells therefore have the ability to modulate alternative splicing for their growth and survival advantage.
This change in splicing pattern could be due to many HDACi effects on the chromatin and gene transcription [
30,
31]. There is a change in chromatin compactness that in turn alters accessibility of DNA to transcription and splicing factors [
22]. It is also plausible that certain splicing factors are involved in this splicing defect and their expression is altered by HDACi exposure [
43]. Also as splicing and transcription occur simultaneously, changes in the transcription with HDACi would affect splicing as well. As HDACi clearly increase gene transcription of epigenetically silenced genes, this alone could change the splicing pattern of the E-cadherin gene. Studies have also shown that rate of transcriptional elongation can regulate splicing as well [reviewed in 30]. RNA Pol II promoter has a large C-terminal domain (CTD) that plays a role in splicing as its truncation results in splicing defects [
30]. The CTD is also subject to regulation by phosphorylation and functions as a docking protein for splicing factors as well [
30]. Differences in Pol II promoter structure thus results in changes in splicing [
44‐
46]. It is possible that the change in ratio of the two transcripts observed with HDACi in CLL specimens is predominantly due to change in E-cadherin transcription. As the change in transcription function of the HDACi cannot be separated from its effect on chromatin compactness and splicing factors, the precise mechanism by which HDACI induce splicing alteration in CLL is not clear.
Mutations of the spliceosome component SF3B1 in CLL cells have been recently identified in 7.5-15% of CLL specimens [
47]. It is proposed that these mutations could result in defective spliceosome function, dysregulated splicing, alterations in gene expression and isoform switch. The precise function of this mutation in CLL is still unclear but it is known that the mutations are associated with CLL specimens that are resistant to fludarabine [
48]. Further investigation of gene splicing in CLL cells is thus increasingly important to understand how the various RNA isoforms affect the biology of CLL and to develop treatment approaches to modify the splicing.
The role and the mechanism E-cadherin silencing in hematopoietic cells is not well known. Our findings clearly indicate that one of the mechanisms of the loss of its expression is histone hypoacetylation. The HDACi increase the acetylation of histones in the region of the E-cadherin promoter as determined by the ChIP assay and appears that this alteration is more pronounced as compared to the acetylation change at the exon 11 region. The overexpression of HDACs in CLL specimens also supports role of this epigenetic alteration in CLL. A recent report has analyzed HDAC expression in CLL specimens and correlated with prognosis [
49]. This report also describes upregulation of a number of different HDAC genes in CLL.
The re-expression of E-cadherin tumor suppressor gene can have a multitude of effects on tumor cells including adhesion, invasion, modulation of receptor kinase activity and wnt pathway activation [
50]. These downstream effects are better defined in the epithelial cells as compared to the hematopoietic cells. A signaling pathway in which there is a potential role of E-cadherin is the wnt pathway and multiple lines of evidence show that this is an important growth and survival pathway in CLL [
3‐
6]. Activation of wnt-pathway in CLL results in up regulation of target genes that are involved in cell proliferation, survival, adhesion and invasion [
51]. The studies in this report confirm the association between E-cadherin and β-catenin that is able to sequester beta-catenin in the cytoplasm which has been reported earlier in other tumor model systems [
12,
13]. The HDACi induced E-cadherin is a thus functional protein and is able to down regulate the wnt reporter activity in three out of four CLL specimens tested. There was also a down regulation of two important downstream effectors of the wnt pathway, namely LEF and cyclin D1 genes. The growth inhibition and annexin positivity observed with HDACi in CLL specimens is thus in part also due to wnt pathway inhibition.
HDACi reverse the epigenetic modifications, reactivate gene expression and cause apoptosis in vitro in different cell types including CLL specimens [
23‐
26]. Our own studies indicate significant apoptosis of CLL specimens with the HDACi MS-275. Studies in vivo have been reported in Tcl1 transgenic mice that develop a disease similar to CLL patients with elevated B cells, splenomegaly and infiltration of B cells in various organs [
52]. The activity of HDACi AR-42 was tested in a Tcl1 transplant model and treated mice showed reduction in peripheral leukocyte counts and survived longer [
53]. There are also encouraging results of clinical trial data in CLL patients with different HDACi [
54]. As these inhibitors have been approved for treatment of patients with T cell leukemia [
55] and are in clinical trials in other malignancies it is imperative that their mechanism of action is well understood.
Competing interest
The authors declare no competing financial interests.
Authors’ contributions
GJ performed experiments, analyzed data and wrote the paper; WL performed experiments, analyzed data and reviewed the paper. SS performed experiments, supervised the study and wrote the paper. All authors read and approved the final manuscript.