Roles of Achaete-scute Homologue 1 in Dkk1 and E-cadherin Repression and Neuroendocrine Differentiation in Lung Cancer

The proneural basic-helix-loop-helix protein achaete-scute homologue 1 (ASH1) is expressed in a very limited spectrum of normal and cancerous cells in a lineage-specific manner, including normal pulmonary neuroendocrine cells and lung cancer cells with neuroendocrine features. Our previous results indicated that ASH1 may play a crucial role in the growth and survival of lung cancers with neuroendocrine features, which prompted us to investigate the molecular function of ASH1 in relation to its involvement in carcinogenic processes. Herein, we report for the first time that ASH1 functions as a dual transcription factor by activating neuro-endocrine differentiation markers and also repressing putative tumor suppressors. This protein was found to inactivate DKK1 and DKK3, negative regulators of Wnt/B-catenin signaling, E-cadherin, and integrin B1 through ASH1-mediated deacety-lation and repressive trimethylation of lysine 27 (H3K27me3) of histone H3 in the promoter regions of DKK1 and E-cadherin. In addition, ASH1-transduced A549 adenocarcinoma cells exhibited markedly altered morphology characteristics compared with lung cancer cells with neuroendocrine features both in vitro and in vivo and also grew faster in vivo. Our results provide important clues for a better understanding of the molecular and cellular biological roles of ASH1 in the process of carcinogenesis of lung cancers with neuroendocrine features and warrant future investigations to shed light on the lineage-specific dependency of this transcription factor with dual functions.


Introduction
Lung cancer has long been the leading cause of cancer-related death in many economically developed countries, and a better understanding of the molecular pathogenesis of this fatal disease is greatly anticipated for preventive and/or therapeutic breakthroughs to drastically reduce the rate of mortality (1).Lung cancer is classified into two major classes, small cell lung cancer (SCLC) and non-SCLC (NSCLC), based on the clinicopathologic characteristics, whereas the latter is further divided into three major types of tumors with distinct morphologic features, adenocarcinomas, squamous cell carcinomas, and large cell carcinomas.It has also been well recognized that there are distinct types of lung tumors that exhibit neuroendocrine features, which comprise SCLCs and subsets of large cell carcinomas, such as large cell neuroendocrine carcinoma (LCNEC) and carcinoids.Although LCNEC is considered to be a disease entity distinct from SCLC, these two types of lung cancers share a great number of features, such as expression profiles and clinical characteristics (2)(3)(4), which suggest the possible roles of one or more genes related to neuroendocrine differentiation in the carcinogenic processes.
Achaete-scute homologue 1 (ASH1) is a basic-helix-loop-helix protein that is expressed in normal pulmonary neuroendocrine cells (PNEC) and lung cancers with neuroendocrine features (5,6).PNECs are proposed to play important roles in the development and growth regulation of lung epithelial cells by functioning as a stem-cell niche and chemoreceptors.It was previously shown that gene targeting of ASH1 resulted in loss of PNECs in the lung (7), whereas an ASH1 transgene promoted the development of lung tumors with neuroendocrine features in cooperation with SV40 large T antigen (8).We also reported that suppression of ASH1 expression by RNA interference induced cell cycle arrest and apoptosis in vitro and inhibited growth of lung cancer cells in vivo in an ASH1 expression-dependent manner (9), leading us to propose that ASH1 may be involved in ''lineage-specific'' survival and growth of lung cancers with neuroendocrine differentiation (10).However, the molecular mechanisms underlying the distinctive features of lung cancers expressing ASH1 remain largely unresolved.
In the present study, we aimed to elucidate the molecular functions of ASH1 in relation to its involvement in the carcinogenic processes of lung cancers with neuroendocrine differentiation.Herein, we report for the first time that ASH1 functions as a transcription factor with dual functions, leading to activation of neuroendocrine differentiation markers, as well as repression of putative tumor suppressors, including the Wnt signaling inhibitors DKK1 and DKK3, E-cadherin, and integrin h1.We also found that transcriptional repression of DKK1 and E-cadherin was associated with histone deacetylation and accelerated histone H3 lysine 27 trimethylation (H3K27me3).Together with our findings showing the biological effects of ASH1 expression in vitro and in vivo, these results provide important clues for a better understanding of how ASH1 may contribute to the development and acquisition of malignant properties of high-grade neuroendocrine lung tumors, such as SCLC and LCNEC.

Materials and Methods
Expression constructs.For the construction of expression plasmids, ASH1, DKK1, LEF1, HDAC1, and HDAC3 cDNAs were generated with reverse transcription-PCR (RT-PCR) and inserted into pCMV-puromycin or pcDNA3 (Invitrogen).For construction of lentivirus vectors, ASH1 and DKK1 cDNAs were inserted into CSII-CMV-MCS-IRES2-Blasticidin and CSII-EF-MCS-IRES2-Venus (improved YFP) vectors, respectively, which were kindly provided by Dr. H. Miyoshi (RIKEN BioResource Center).Venus was provided by Dr. A. Miyawaki (RIKEN Brain Science Institute).The myc-tag was attached at the NH 2 terminus of ASH1 cDNA.Lentiviruses were produced according to the protocol of Dr. Miyoshi.The constructed vectors were cotransfected with pCAG-HIVg and pCMV-VSV-G-RSV-Rev plasmids into 293T cells.Two days after transfection, the supernatant was harvested and concentrated using ultracentrifugation.The virus particle pellets were resuspended in RPMI medium and stored at À80jC.The virus titer was determined by counting Venus-positive cells or Blasticidin-resistant colonies.A549 cells were infected with ASH1 or empty lentiviruses.After infection, the cell lysates were consecutively harvested, and total RNA was extracted using an RNeasy kit (Qiagen), according to the manufacturer's instructions.
Expression profiling with microarrays.Using the RNA samples of both A549-ASH1 and A549-VC cells at 4 d after infection, cRNA was generated and labeled with Cy3 or Cy5 (GE Healthcare) using a low RNA fluorescent linear amplification kit (Agilent Technologies), according to the manufacturer's instructions.Labeled cRNAs were hybridized to an Agilent 44K whole human genome microarray, followed by confocal laser scanning (Agilent Technologies).Duplicate experiments of expression profiling were performed.
Luciferase assay.For analysis of h-catenin/TCF activity, TOP-flash and FOP-flash reporters (pGL3-OT with three wild-type TCF/LEF sites and pGL3-OF with mutated TCF/LEF sites, respectively, kindly provided by Dr. B. Vogelstein) were used.For generation of DKK1 and E-cadherin reporter constructs, genomic fragments corresponding to DKK1 (1,007 bp) and E-cadherin (364 bp) promoters were amplified with PCR and inserted into a pGL3 basic reporter (Promega).E-box motifs were disrupted using PCR amplification with Pfu TURO (Stratagene) and primers containing mismatches at E-box motifs.Luciferase reporter activity was measured as previously described in detail (11).In brief, A549 cells were grown at 3 Â 10 5 in a 3-cm dish and transfected in triplicate with 0.6 Ag of the expression vector, along with 0.02 Ag of a firefly luciferase reporter (TOP-flash, DKK1, or E-cadherin reporter) and 0.01 Ag of an internal control reporter, pRL-TK (Promega) using Lipofectamine 2000 (Invitrogen).To assess h-catenin/TCF activities, A549-ASH1 or A549-VC cells were transfected with a TOP-flash or FOP-flash reporter.The day after transfection, firefly luciferase and Renilla luciferase activities were measured using a dual-luciferase reporter assay system (Promega).Firefly luciferase activity was normalized with that of Renilla luciferase.
Chromatin immunoprecipitation assay.A chromatin immunoprecipitation (ChIP) assay was performed basically as described previously (11).Before harvesting, cells were treated with 1% formaldehyde for 10 min at 37jC, then harvested, and resuspended in SDS lysis buffer [1% SDS, 10 mmol/L EDTA, and 50 mmol/L Tris-HCl (pH 8.1)] for 10 min on ice.After a brief sonication, antibodies (Upstate Biotechnology) were added to the lysate.After an overnight incubation, immune complexes were collected using protein G-Sepharose 4 fast flow.The immunoprecipitated protein-DNA complexes were eluted with SDS/sodium bicarbonate and extracted using standard proteinase K/phenol extraction procedures.The obtained DNA was PCR-amplified using semiquantitative PCR with Platinum Taq DNA polymerase (Invitrogen) and primers for DKK1, E-cadherin, chromogranin B (CHGB), secretogranin 2 (SCG2), and b-actin promoters, applied using electrophoresis and measured with an ImageMaster-CL densitometer (GE Healthcare).The amount of DNA was also determined using real-time PCR with Power SYBR Green PCR Master Mix (Applied Biosystems) and a 7500 Real-Time PCR System (Applied Biosystems).The amounts of each promoter were normalized by that of the b-actin promoter and then compared between A549-ASH1 and A549-VC cells.In a ChIP assay of the DKK1 promoter of SCLC cell lines, the amount of DKK1 promoter was similarly normalized with that of the b-actin promoter, and its ratio to whole cell extract is presented.Both analyses showed consistent results.The antibodies used were anti-acetyl-histone H3 (K 9 /K 14 ), anti-acetyl-H3 (Lys 9 ), anti-dimethyl-H3 (Lys 4 ), anti-trimethyl-H3 (Lys 4 ), anti-dimethyl-H3 (Lys 9 ), anti-trimethyl-H3 (Lys 9 ), anti-monomethyl-H3 (Lys 27 ), and antitrimethyl-H3 (Lys 27 ) antibodies (Upstate).For immunoprecipitation of ASH1, the anti-myc-tag 9E10 antibody was used.

Results
Expression profiling of A549-ASH1 cell.To elucidate the molecular function of ASH1 in the carcinogenic processes of lung cancers with neuroendocrine differentiation, the ASH1 gene was introduced into a typical adenocarcinoma cell line, A549, using an IRES-blasticidin lentiviral vector.Comparison of expression profiles between A549 cells transduced with an ASH1-expressing virus (A549-ASH1) and control virus-infected A549 cells (A549-VC) resulted in the identification of 219 and 67 genes with upregulation and down-regulation, respectively (>4-fold; Supplementary Tables S1 and S2).Up-regulated genes included those directly related to neuroendocrine differentiation, such as CHGB and SCG2, whereas down-regulated genes comprised many that were related to cancer, including such putative tumor suppressors as DKK1, DUSP1, RASSF2, E-cadherin, IGFBP6, SOCS3, DKK3, and SEMA3C.Significant changes in the expression levels of CHGB, SCG2, DKK1, DKK3, E-cadherin, and integrin b1 at various time points after transduction of the ASH1-expressing virus were further substantiated by quantitative RT-PCR analysis (Figs. 1A, 3A, left, and 4A).The whole results of duplicate microarray analyses of A549-ASH1 and A549-VC cells are also presented in Supplementary Table S4.
ASH1-mediated repression of DKK1.Identification of significant down-regulation of the DKK1 and DKK3 Wnt signaling inhibitors prompted us to investigate the molecular and biological consequences of ASH1-mediated transcriptional repression.We observed that introduction of DKK1 significantly inhibited the growth of an ASH1-expressing and DKK1-repressed SCLC cell line, ACC-LC-172, whereas A549 cells, which express endogenous DKK1 but not ASH1, did not show any inhibited growth (Fig. 1B, top).Similarly, ACC-LC-80 with repressed DKK1 showed a significant reduction of Venus fluorescence intensity, which reflected the fraction of virus-infected cells indicated by the DKK1-IRES-Venus lentivirus, in contrast to virtually no effect in ACC-LC-76 expressing endogenous DKK1 (Fig. 1B, bottom).These results clearly showed the growth inhibitory effects of DKK1 in DKK1-negative and ASH1positive lung cancer cells, suggesting a functional consequence of DKK1 down-regulation in association with ASH1 expression.
It is possible that the down-regulation of DKK1 might be indirect, thus reflecting an inactivated state of Wnt/h-catenin signaling imposed by ASH1 expression, because it is known that DKK1 can be activated through a negative feedback mechanism in the Wnt/h-catenin signaling pathway.As shown in Fig. 1C (top left), ASH1 indirectly activated a TOP-flash reporter, which contains three TCF/LEF sites, but not a FOP-flash reporter in A549 cells.We also found that ASH1 expression did not affect activation by LEF1, a Wnt/h-catenin signaling mediator (Fig. 1C, top right).These findings indicate that ASH1 is able to activate the Wnt/h-catenin signaling pathway and DKK1 repression is unlikely to be a consequence of Wnt/h-catenin inactivation.
We also studied the activity of the DKK1 promoter, which contains Wnt/h-catenin signal-responsive TCF/LEF sites and a possible ASH1-binding element, E-box.DKK1 promoter activity was clearly down-regulated by ASH1 in A549 cells (Fig. 1C, bottom left), whereas LEF1 activated the DKK1 reporter, which is a negative feedback mechanism.Notably, LEF1-mediated activation of the DKK1 reporter, which contains both TCF/LEF-binding and ASH1binding sites, was significantly inhibited by ASH1 (Fig. 1C, bottom right), in contrast to the results with the TOP-flash reporter shown in Fig. 1C (top right).Together with the lack of any physical interaction between ASH1 and either h-catenin or LEF1 (data not shown), these findings suggest a direct inhibitory effect of ASH1 on DKK1 expression.
We did not observe changes in the DNA methylation status of the DKK1 promoter region in A549-ASH1 cells (data not shown), whereas significant changes in the chromatin structure of the DKK1 promoter were found (Fig. 1D).The results of a ChIP assay clearly showed direct binding of ASH1 to the DKK1 promoter, as well as transcription-repressing histone deacetylation and trimethylation of Lys 27 (H3K27me3) of histone H3 in A549-ASH1 cells.In addition, ASH1 was shown to form a complex with HDAC1 and HDAC3 (Supplementary Fig. S1A).Taken together, these findings strongly suggest that ASH1 has the ability to directly bind to the DKK1 promoter, which in turn recruits histone-modifying molecules, consequently inducing transcriptional repression of the DKK1 promoter.Because H3K27me3 is mainly catalyzed by EZH2 in the polycomb-repressing complex (12), the interaction between ASH1 and EZH2 was also studied.However, no clear evidence of such interaction was detected (data not shown).
DKK1 silencing in SCLC cell lines and tumor tissues.Next, we investigated whether the expression level of DKK1 had a relationship with that of ASH1 in lung cancer cell lines in vitro, as well as in lung cancer tumor specimens.DKK1 expression in SCLC cells with a generally high level of ASH1 expression was significantly lower than that in NSCLC (adenocarcinomas and squamous cell carcinoma) cell lines (Fig. 2A), whereas DKK1 downregulation was also observed in SCLC tumor specimens, but not in tumor tissues from adenocarcinomas and squamous cell carcinomas (Fig. 2B), demonstrating clear inverse correlations both in vitro and in vivo.DNA methylation of the DKK1 promoter region was also studied using a panel of lung cancer cell lines with methylation-specific PCR (MSP) analysis, which resulted in identification of DNA methylation in the DKK1 promoter in a limited fraction of cell lines, including cells from a single SCLC without DKK1 expression (Supplementary Fig. S1B and C).Analysis of the DNA methylation status at each CpG site in the representative cell lines yielded results consistent with those of MSP analysis (Fig. 2C; data not shown for A427).A ChIP assay was also performed, which showed the presence of histone deacetylation and induction of H3K27me3 in SCLC cell lines exhibiting DKK1 silencing without DNA methylation, whereas an SCLC cell line with DNA methylation, ACC-LC-80, exhibited induction of H3K9me2, but not H3K27me3 (Fig. 2D).These results suggest that

DKK1 silencing in SCLCs is mainly attributable to repressive histone modifications imposed by ASH1 expression rather than DNA methylation.
Repression of E-cadherin by ASH1.We also studied the expression of E-cadherin and integrin h1, which are crucial regulators of cell-cell interactions.Quantitative RT-PCR analysis confirmed the results of the microarray analysis, showing downregulation of E-cadherin and integrin h1 in response to introduction of ASH1 into A549 cells (Fig. 3A, left), whereas Western blot analysis also showed significant down-regulations of E-cadherin and integrin h1 proteins in ASH1-transduced A549 cells (Fig. 3A, right).The underlying mechanism for the transcriptional down-regulation of the E-cadherin gene was investigated using a luciferase reporter construct containing a PCR-amplified 369-bp-long genomic fragment of the E-cadherin promoter.Whereas ASH1 expression apparently reduced wild-type E-cadherin promoter activity, disruptions of E-boxes in the promoter region impaired the repressive effect of ASH1 (Fig. 3B).
Next, we turned our attention to the state of modification of the E-cadherin promoter in relation to transcriptional repression and found a lack of DNA methylation in ASH1-transduced A549 cells (data not shown).Then, the chromatin structure of the E-cadherin promoter was studied using a ChIP assay.ASH1 directly bound to the E-cadherin promoter and induced histone H3 deacetylation and H3K27me3 (Fig. 3C), as was similarly observed with the DKK1 promoter.Interaction of ASH1 with E-box1 was shown by results of a gel shift assay using an E-box1 probe, which was completely competed out by a cold E-box oligonucleotide (Fig. 3D).Addition of the antibody against myc-tag attached to the exogenous ASH1 protein clearly caused a super-shift of the band, implying binding of the ASH1 protein to the E-box1 element of the E-cadherin promoter.In contrast to the findings with the DKK1 and E-cadherin genes, we observed the opposite effects of ASH1 with the neuroendocrine marker genes CHGB and SCG2.ChIP assay findings revealed that transduction of ASH1 in A549 cells led to induction of H3 acetylation and a reduction of H3K27me3 (Fig. 4B), resulting in a strong induction of CHGB and SCG2 expression (Fig. 4A).
ASH1-induced phenotypic changes in vitro and in vivo.Next, we investigated the phenotypic consequences of ASH1 expression in human lung cancer cells.In monolayer cultures (Fig. 5A, top), A549-VC cells tightly attached to the dishes like typical epithelial cells, whereas A549-ASH1 cells exhibited a round-shaped morphology with reduced cell adhesion.In semisolid cultures (Fig. 5A, middle), A549-VC cells formed compact cell aggregates, whereas A549-ASH1 cells had loose cell-cell contacts.Interestingly, threedimensional cultures in collagen gels showed even greater morphologic differences between A549-VC and A549-ASH1, as A549-VC cells showed filopodia formation in contrast to the roundshaped cell morphology without any protrusions in A549-ASH1 cells (Fig. 5A, bottom).These ASH1-induced morphologic phenotypes closely resembled those of typically used SCLC cell lines, such as ACC-LC-48 (Fig. 5A).Immunofluorescent microscopic analysis Note that the shifted band was clearly inhibited by the cold competitor of the E-box1 oligonucleotides, but not by scrambled oligo, and a supershift (closed arrowhead) was seen using the anti-myc-tag antibody for detecting myc-tagged ASH1.also showed abundant expression of E-cadherin on the cell surface of A549-VC, whereas that was scarcely observed on A549-ASH1 cells, suggesting that the phenotypical alterations may be attributable to markedly reduced expression of cell adhesion molecules (Fig. 5B).
The effect of ASH1 on the subcellular localization of h-catenin was also studied using A549 cells transiently transfected with the ASH1 expression vector.Accumulation of h-catenin into the nuclei was observed in A549 cells expressing ASH1 (Fig. 5C, arrowheads), whereas nontransfected cells showed membrane-bound h-catenin proteins without nuclear accumulation, suggesting the activation of h-catenin signaling by ASH1.
We also examined the phenotypic consequences of ASH1 expression in vivo using A549-ASH1 and A549-VC xenografts.The former developed high-grade carcinomas with morphologic characteristics that showed neuroendocrine features, which are frequently seen in SCLCs, LCNECs, and carcinoid tumors.In contrast, tumors developed from the A549-VC xenografts formed cyst-like or tube-like structures, which were lined by columnar epithelia (Fig. 6A).Immunohistochemical analyses supported the morphologic characteristics, as a negative expression of E-cadherin and positive expression of neuroendocrine marker, known as synaptophysin, was observed in A549-ASH1 xenografts, whereas a contrasting pattern was observed in A549-VC xenografts (Fig. 6B and C).We also found accelerated tumor growth in A549-ASH1 xenografts compared with the A549-VC xenografts (Fig. 6D).
We also studied the universality and constancy of the effects of ASH1.In another adenocarcinoma cell line, SK-LC5, ASH1 overexpression induced alterations of gene expressions, i.e., down-regulation of DKK1 and up-regulation of SCG2 (Supplementary Fig. S2A), and the cells also underwent a phenotypic change into a round-shaped morphology (Supplementary Fig. S2B), similar to A549 cells, suggesting the universality of these ASH1 functions.In addition, dual-directional regulation of ASH1 on gene expression was consistently observed in stable A549-ASH1 cells when compared with the stable cell line A549-VC (Supplementary Fig. S2C and D), which was similar to the observation of early response just after virus infection.

Discussion
Our previous findings suggested an important role for ASH1 expression in lung cancers with neuroendocrine differentiation, including SCLCs and LCNECs (9), and led us to initiate the present study to elucidate the molecular function of ASH1 in relation to its involvement in carcinogenic processes.Our results showed a dual role of ASH1 as a transcriptional factor, leading to induction of a number of genes, such as neuroendocrine differentiation markers, as well as repression of multiple putative tumor suppressor genes, including the Wnt/h-catenin signaling inhibitors DKK1 and DKK3, as well as E-cadherin and integrin h1, which are cell adhesion molecules.In this regard, it should be noted that ASH1 is the first transcriptional repressor identified for members of the DKK gene family.We also showed that ASH1-mediated DKK1 and E-cadherin silencing were associated with the induction of histone H3 deacetylation and increased H3K27me3.Conversely, ASH1mediated induction of neuroendocrine differentiation markers was associated with histone H3 acetylation and a reduction in H3K27me3.
Polycomb repressive complex 2 (PRC2), which induces H3K27me3, is thought to play important roles in stem cell biology.Gene silencing in stem cell-related genes is frequent in cancer cells and has been suggested to be a possible reflection of cell fate, which is preprogrammed through H3K27me3 by PRC2 (13)(14)(15).In this context, DKK1 is a polycomb target gene and repressed by H3K27me3 in ES cells (16,17).ASH1 might play a role in carcinogenesis by expanding the population of cancer progenitor cells carrying preexisting H3K27me3, which may be in accordance with a hypothesis that PNEC is derived from multipotent epithelial progenitors that can differentiate into both Clara cells/pneumocytes and cells with a PNEC lineage (6).Alternatively, it is also possible that ASH1 may induce de novo H3K27me3, possibly through physical and/or functional interactions with PRC2.
Wnt signaling is thought to regulate the proliferation and renewal of stem cells, whereas dysregulated activation of Wnt/ h-catenin signaling has been implicated in carcinogenesis (18).DKK1 inhibits Wnt/h-catenin signaling through disruption of the complex formation of Wnt and its receptors, LRP5/6 and Frizzled (19), whereas DKK1 is transcriptionally induced by Wnt/h-catenin signaling itself as a negative feedback mechanism (20,21).Evidence for the potential involvement of DKK1 inactivation in human cancers is accumulating.The DKK1 gene was reported to be a frequent target for epigenetic silencing by DNA hypermethylation in colorectal cancers (21), whereas overexpression of DKK1 was shown to reduce in vitro and in vivo colorectal cancer cell growth (22).In contrast, the underlying mechanism of DKK1 silencing in lung cancers seems to be attributable mainly to repressive histone H3 modifications rather than DNA methylation, showing the existence of a lineage-specific preference imposed by the highly lineage-specific transcription factor ASH1.In addition to DKK1, we observed that ASH1 mediated the down-regulation of DKK3, which has also been reported to be inactivated in other cancers (19).Together with the increased growth of ASH1transduced A549 cells, the present findings support the notion A, bottom, induction of round-shaped morphology without protrusion in three-dimensional culture of A549-ASH1 on collagen gels in contrast to filopodia formation by A549-VC.Note that a typical SCLC cell line, ACC-LC-48, shows complete anchorage-independence and loose cell-cell interactions similar to A549-ASH1.B, immunofluorescent staining of E-cadherin, showing loss of E-cadherin in A549-ASH1 in contrast to the abundant expression in A549-VC.C, immunofluorescent staining of h-catenin, showing induction of nuclear accumulation of h-catenin in ASH1-transfected A549 cells (arrowheads ), suggesting the activation of h-catenin signaling.In contrast, plasma membrane-bound h-catenin is seen in cells lacking ASH1 expression.that ASH1 expression may favor cancer cell growth through repression of DKK1 and DKK3 and the consequential activation of Wnt/h-catenin signaling.The present findings of frequent silencing of DKK1 in SCLCs, but not in NSCLCs, seem to be consistent with a recent report on the expression of DKK1 in NSCLCs (23).
E-cadherin promotes cell-cell interactions and sequesters h-catenin in the cell membrane, whereas reduced E-cadherin expression has been postulated to play a role in cell migration and metastasis, as well as anchorage independence (24).E-cadherin is silenced in a number of types of malignancy, including lung cancers with neuroendocrine features (25,26).Several transcriptional repressors (the zinc finger proteins SNAI1/2 and ZEB1/2 and bHLH protein TWIST1) have been implicated in E-cadherin repression in cancers, which may in turn play a role in epithelial-mesenchymal transition (EMT) during tumor progression (27).Thus, our findings support the addition of ASH1 as a transcriptional repressor of E-cadherin.It is of interest that SCLC cells often show amoeboid migration through loose cellcell and cell-matrix interactions, which may contribute to the acquisition of highly metastatic features by SCLCs (28).It is also notable that the morphologic features of A549-ASH1 cells resembled the amoeboid migration phenotype, possibly because of the ASH1-mediated reduction of E-cadherin and integrin h1 expression.Although down-regulation of E-cadherin is frequently associated with EMT, ASH1-induced phenotypic changes do not seem to conform to the strict criteria of EMT, because the induction of mesenchymal markers such as N-cadherin and vimentin was not observed in A549-ASH1 cells (data not shown).Alternatively, ASH1-induced phenotypic changes might reflect the cell migration capacity of neuronal progenitors during embryogenesis.
In summary, the present findings support the notion that ASH1 has a role in the development of lung cancers with neuroendocrine features, not only due to its decisive role as a transcriptional activator in neuroendocrine differentiation, but also because of its transcriptional repressor activity that has effects on various putative tumor suppressor genes.The latter repressor activity leads to the inactivation of genes related to negative regulation of cell growth, such as DKK1, which consequently activates Wnt/h-catenin signaling, as well as those related to cell adhesion, such as E-cadherin and integrin b1, which may also confer cell migration/metastasis capabilities.A future study of the molecular functions of ASH1 is warranted, which would shed light on its involvement in carcinogenic processes, as well as the underlying mechanism of the lineagespecific dependency of this dual function transcription factor in lung cancers with neuroendocrine features.

Figure 1 .
Figure 1.ASH1-mediated repression of DKK1-inhibited cell growth of DKK1-negative SCLC cells.A, quantitative RT-PCR analysis of DKK1 and DKK3 with A549-ASH1 cells, showing prompt repression after infection with the ASH1-expressing virus.B, top, cell growth was significantly reduced by transfection of ASH1 into ACC-LC-172 SCLC cells lacking DKK1 expression, but not A549 adenocarcinoma cells expressing DKK1, as measured by a colorimetric assay (left) and colony formation assay (right ).B, bottom, Venus fluorescence intensity reflecting cell growth was significantly reduced in ACC-LC-80 SCLC cells with DKK1 silencing after infection with the DKK1-IRES-Venus lentivirus.Virtually, no effect was observed in ACC-LC-76 SCLC cells expressing DKK1.C, top, TOP-flash/FOP-flash assays, showing strong enhancement with the TOP-flash reporter containing TCF/LEF-responsive elements in A549-ASH1, but not A549-VC cells.C, top right, TOP-flash assay of A549 cells transfected with ASH1 and/or LEF1, showing no effect of ASH1 on LEF1-mediated activation of the TOP-flash reporter.C, bottom, reporter assay of the DKK1 promoter containing both TCF/LEF sites and a possible ASH1-binding E-box.Marked inhibition was seen with the introduction of ASH1.Note that LEF1-induced activation of the DKK1 reporter was also suppressed by ASH1.D, ChIP assays of the DKK1 promoter in A549-ASH1 and A549-VC cells, showing significant deacetylation and induction of H3K27me3 in A549-ASH1 cells.The ratios of the DKK1 PCR products between A549-ASH1 and A549-VC cells are shown after normalization with h-actin PCR products (top ) and each PCR product is also shown (bottom ).The ratio of PCR products with whole cell extracts (WCE ) was set at 1.

Figure 2 .
Figure 2. DKK1 silencing in SCLC cells and tumor tissues.A, inverse association between DKK1 and ASH1 expression in lung cancer cells, as determined by quantitative RT-PCR assays.AD, adenocarcinoma; SQ, squamous cell carcinoma.B, inverse association between DKK1 and ASH1 expression in primary lung cancers based on published microarray analysis data (29).C, bisulfite sequencing analysis of the DKK1 promoter in lung cancer cells and normal lung tissue, showing DNA methylation in a single SCLC cell line with DKK1 silencing.Open circle, nonmethylated CpG; closed circle, methylated CpG.D, ChIP assay of the DKK1 promoter in SCLC cells, showing histone H3 deacetylation in all SCLC cell lines with DKK1 silencing.H3K27me3 was seen in DKK-1-negative SCLC cell lines without DNA methylation.

Figure 3 .
Figure 3. Transcriptional repression of E-cadherin by ASH1.A, left, quantitative RT-PCR analyses of E-cadherin and integrin h1 in A549-ASH1 cells, showing repression after transduction of the ASH1-expressing virus.A, right, Western blot analyses of E-cadherin and integrin h1, showing marked reduction in expression levels of E-cadherin and integrin h1 in A549-ASH1 cells.Lamin B1 served as the loading control.B, reporter assay of the E-cadherin promoter after cotransfection with the ASH1 expression vector (gray column ) or control vector (white column ), showing reduction of repression by introduction of mutations to the E-boxes in the promoter.A schematic diagram of the E-cadherin promoter reporters (left) with disrupted E-boxes indicated by an X. Shaded boxes, E-boxes; ovals, CCAAT-box; circles, SP1-consensus motif.C, ChIP assay of the E-cadherin promoter performed in a similar manner for analysis of the DKK1 promoter in Fig. 1, showing induction of deacetylation and H3K27me3 in A549-ASH1 cells.D, GEMSA assay, showing a shifted band (open arrowheads ) specific to the A549-ASH1 nuclear extract.Note that the shifted band was clearly inhibited by the cold competitor of the E-box1 oligonucleotides, but not by scrambled oligo, and a supershift (closed arrowhead) was seen using the anti-myc-tag antibody for detecting myc-tagged ASH1.

Figure 4 .
Figure 4. ASH1-mediated transcriptional activation of neuroendocrine markers.A, quantitative RT-PCR analyses of CHGB and SCG2 in A549-ASH1 cells, showing increased expression after ASH1 expression induced by viral infection.B, ChIP assay of the CHGB and SCG2 promoters.The results show the induction of acetylation and reduction of H3K27me3 in A549-ASH1 cells, which were in marked contrast to the DKK1 and E-cadherin promoters.

Figure 5 .
Figure 5. Phenotypic changes in vitro and in vivo in response to the introduction of ASH1 in A549 adenocarcinoma cells.A, top, induction of round-shaped morphology with a reduction of attachment to culture dishes in monolayer culture of A549-ASH1 cells.A, middle, induction of loose aggregate in semisolid culture of A549-ASH1 in contrast to compact cell aggregates of A549-VC.A, bottom, induction of round-shaped morphology without protrusion in three-dimensional culture of A549-ASH1 on collagen gels in contrast to filopodia formation by A549-VC.Note that a typical SCLC cell line, ACC-LC-48, shows complete anchorage-independence and loose cell-cell interactions similar to A549-ASH1.B, immunofluorescent staining of E-cadherin, showing loss of E-cadherin in A549-ASH1 in contrast to the abundant expression in A549-VC.C, immunofluorescent staining of h-catenin, showing induction of nuclear accumulation of h-catenin in ASH1-transfected A549 cells (arrowheads ), suggesting the activation of h-catenin signaling.In contrast, plasma membrane-bound h-catenin is seen in cells lacking ASH1 expression.