Down regulation of the transcriptional mediator subunit Med1 contributes to the loss of expression of Metastasis associated dapk1 in human cancers and cancer cells

Abstract

DAPK1, a ca⁺²/calmodulin regulated serine/threonine kinase, is a major tumor suppressor, whose expression is lost in multiple tumor types. However, the mechanisms contributing to it are unclear. We have recently shown that CCAAT/Enhancer binding protein-β (C/EBP-β) is required for the basal and interferon γ (IFN-γ)-induced expression of dapk1 in many cell types. C/EBP-β interacts with the transcriptional Mediator, a multi-subunit complex that couples enhancer bound transcription factors to the basal transcriptional machinery in an IFN-γ dependent manner for regulating dapk1 expression. Specifically, the Med1 (TRAP220/PBP/DRIP220/CRSP220) subunit associates with the enhancer bound C/EBP-β at the CRE/ATF site of dapk1 in an IFN-γ dependent manner for stimulating gene expression. Therefore, we investigated if the mechanism responsible for the loss of dapk1 expression in human cancers involves a failure to recruit C/EBP-β and/or Med1 to the dapk1 promoter. We compared the relative occupancy of these factors at the dapk1 promoter at CRE/ATF sites in normal and cancer cell lines. A significantly lower binding of these factors to the CRE/ATF site of dapk1 promoter occurred in human cancer cell lines than in normal cells. We show that loss of Med1 expression correlates with a corresponding loss of dapk1 expression in a number of primary human lung carcinomas. Med1 levels were significantly lower in cancer cell lines than in normal controls. Importantly, we show that restoration of Med1 induces the expression of dapk1 in these cancer cells and also attenuates their metastatic potential in vivo. Our studies reveal a critical parameter limiting dapk1 expression in cancer cell lines.

Introduction

DAPK1 is a calcium calmodulin-dependent serine/threonine protein kinase, which suppresses tumor metastasis via induction of apoptosis and autophagy ^1-3. It was originally isolated as a regulator of IFN-γ-activated cell death using a genetic screen ², whose down-regulation inhibited IFN-γ-induced cell death in several cell types. However, the regulation of dapk1 is not fully understood. Recently, we identified that the expression of dapk1 is critically dependent on transcription factor CCAAT/Enhancer binding protein-β (C/EBP-β) ⁴. The C/EBP family of transcription factors regulates energy metabolism, immune response and cell growth and differentiation ^{5, 6}. Among its members, one protein, C/EBP-β exhibits remarkable functional diversity and plasticity due to in its unique ability to respond to a variety of extracellular and intracellular signals to mediate a number of responses ⁶. We found that both basal and IFN-γ-induced expression of dapk1, is critically dependent on C/EBP-β⁴. C/EBP-β regulates dapk1 promoter from at least two promoter sites, distal CBS and proximal CRE/ATF ⁴. IFN-γ induces the recruitment of C/EBP-β to CRE/ATF site of dapk1 promoter for inducing dapk1 expression ⁴.

C/EBP-β dynamically associates with various cellular factors for exerting its activity in a gene context and signal-specific manner ^7-10. We have recently reported that the Med1 (also known as TRAP220/PBP/DRIP220/CRSP220) subunit of the transcriptional Mediator complex ¹¹ interacts with C/EBP-β in an IFN-induced manner ^12-14. It is required for IFN-induced C/EBP-β-dependent expression of certain cellular genes including dapk1 ¹⁵. The Mediator protein complex regulates transcription from specific gene-enhancers in response to hormones and other extra-cellular signals. Deletion of the genes coding for its constituent subunits lead to a loss of a number of transcriptional events that participate in cell division, differentiation and metabolism ^16-19. Certain members of the MAPK family, specifically ERK1/2, play a central role in regulating IFN-induced interplay between Med1 and C/EBP-β by inducing phosphorylation of C/EBP-β at the critical T¹⁸⁹ residue and possibly of Med1 ¹⁵. It has been suggested that both C/EBP-β and Med1 undergo conformational changes following phosphorylation ^{15, 20, 21}. These events are critical for a recruitment of both factors to CRE/ATF sites and transcriptional up-regulation of dapk1 ⁴. Thus, C/EBP-β links Med1 to dapk1 in the IFNγ-induced signal transduction pathways.

As said above, dapk1 expression is down regulated in a number of human tumors ²². Although in several cases methylation appears to be one major mechanism, dapk1 expression is also silenced without apparent methylation ^23-27. Mutational inactivation of dapk1 is almost rare in most cancers ²², except in some familial chronic lymphocytic leukemias, where its expression is down-regulated by a unique polymorphism in nucleotide 6531 upstream of the A of translation initiation codon (c.1-6531A>G) ^{28, 29}.

Based on our recent observations that dapk1 gene expression is regulated by active recruitment of C/EBP-β and Med1 to its promoter at the CRE/ATF site in an IFN-γ dependent manner in normal cells, we hypothesize that a loss of these interactions in cancer cells results in a suppression of dapk1 expression. To investigate this aspect, we compared dapk1 promoter regulation in cancer vs. normal cells. In this study, we show that loss of Med1 expression directly correlates with a loss of dapk1 expression in a number of human lung cancers and cancer cell lines. Furthermore, in mouse tumor xenograft model, restoration of Med1 levels restrained the metastatic activity of a lung tumor cell line by upregulating DAPK1.

Materials and Methods

Reagents and Antibodies

Recombinant human IFN-γ was purchased from PBL Biomedical Laboratories. Total extracellular signal-regulated kinase (ERK) and diphosphorylated ERK (ppERK) antibodies were obtained from Cell Signaling Technology, Inc. Mouse monoclonal antibody against DAPK1 was purchased from Sigma-Aldrich, Inc, St.Louis, MO. Rabbit polyclonal antibody against Actin is from Sigma. Rabbit polyclonal antibodies against C/EBP-β; goat polyclonal antibodies against Med1,and bovine anti-goat IgG-horseradish peroxidase conjugate were purchased from Santa Cruz Biotechnology, Inc. Horseradish peroxidase conjugates of anti-rabbit and anti-mouse IgGs were obtained from GE Healthcare, Inc.

Cell culture

A549, BEAS-2B, MCF-7 and hTERT-HME1 were cultured as reported earlier ^{4, 30}. We compared the results obtained from A549, a human lung adenocarcinoma cell line to BEAS-2B, a non-oncogenic normal human bronchial epithelial cell line; and MCF-7, a human breast adenocarcinoma cell line, to hTERT-HME1, a normal human mammary epithelial cell line. Cells were treated with 20nM 5-Azacytidine (Sigma) for 48h, where indicated. ZR75-1 is a breast carcinoma cell line and MCF-10A, is a non-oncogenic breast epithelial cell line. The Lipofectamine plus reagent (Invitrogen, Inc, Carlsbad, CA) was used for transfecting specific genes into cells.

Gene expression analyses

The expression levels of specific transcripts were quantified using Real-time RT-PCR (qPCR) analysis with gene specific primers. PCR was performed with gene-specific primers in a 30 μl final reaction containing ~20ng of cDNA, 200 nM of primers, 0.6 units of Taq DNA polymerase, and SYBR Green dye. Expression of different genes was normalized to internal control transcripts rpl32 and rps24 coding for the ribosomal proteins as in our previous studies ^{4, 31}. The primers used for qPCR were: dapk1-forward: 5’- AATGGA GTTGGCGATTTCAGCGTG-3’ and dapk1-reverse: 5’-AAGGGACTTCAGGAAACTGAG CCA-3’; dapk2-forward: 5’ -TGCAGCCAAGTTCATCAAGAAGCG -3’ and dapk2-reverse: 5’-ACACTAGCTCAAGGATGAGCACCA-3’; irf1-forward: 5’-AGAGCAAGGCCAAG AGGAAGTCAT-3’ and irf1-reverse: 5’-AAGTCCTGCATGTAGCCTGGAACT-3’; rpl32 forward: 5’-TTAAGCGTAACTG GCGGAAACC-3’ and rpl32 reverse: 5’-CAGTAAGAT TTGTTGCACATCAGC-3’; and rps24-forward: 5’-GATGTCATCTTTGTATTTGGATTCAG-3’; and rps24-reverse: 5’-CTGACTTTCTTCAT TCTGTTCTTGCG-3’. Triplicate reactions were performed for each sample; and each experiment was repeated thrice with independent batches of RNA. Luciferase assays with dapk1-Luc, were performed as described elsewhere ⁴. All luciferase activities were normalized to an internal control reporter, CMV-β-galactosidase, for correcting variations in transfection efficiencies. Each experiment was repeated 3 times with multiple independent samples (n=4). Med1¹⁵ and dapk1-luc constructs ³² were described earlier. Widltype Med1and its mutant T¹⁴⁵⁷A/T¹⁰³²A were expressed¹⁵ as N-terminally flag-epitope-tagged proteins using the pCMV-3Xflag vector (Sigma_Aldrich, Inc).

Western Blot Analyses

Proteins were separated on SDSPAGE (8–10% gels), blotted onto a polyvinylidenedifluoride (PVDF) membrane (Millipore) and probed with appropriate antibodies as in our earlier studies ⁴. The primary and secondary antibodies were used at 1:1000 and 1:2000 dilutions, respectively, for Western blot analyses and bands were detected using ECL kits (Pierce). Under these conditions, these antibodies detected only specific bands with appropriate molecular weights. Hence, only the relevant portions of the blots were shown in the figures.

ChIP Assays

ChIP assays were performed as described in our recently published work ^{4, 15} using a commercially available kit (Upstate Biotechnology, Inc.). In brief, soluble chromatin complexes were incubated with 5μg of specific or nonspecific antibodies at 4°C overnight. After immunoprecipitation with specific or non-specific IgGs, initial PCRs were conducted with serial dilutions of input material from each immunoprecipitation to establish the appropriate cycling conditions to accurately compare template content across treatments. Appropriate numbers of PCR cycles are performed to ensure that the specific and mock-precipitated DNA is within the linear range of amplification (results not shown). Purified DNA from the input and IP samples were subjected to qPCR analysis with CRE/ATF site primers of the human dapk1 promoter: Fwd 5’-AGTCCTCAGAAATCTCATGCAAG -3’ and Rev 5’ ACATTAGAGTCCAAGACAGTA TGG 3’. DNA extracted from soluble chromatin was used as input control. Each experiment was repeated at least 3 independent times with multiple samples for time point (n=4) for ensuring the consistency of the results.

Tumor arrays

To study the loss of DAPK1 and Med1, we employed two different sets of lung tumor arrays on glass sides: 1) A commercially available lung tumor array, MH-358 (Imgenex, Inc) which contained 50 tumor samples (see results section for details) 9 non cancerous controls. 2) The second lung tumor array was generated and provided by the Tissue Array Research Program (TARP) of the National Cancer Institute, Bethesda, MD, which contained 100 lung tumors and 4 normal lung tissues as controls. Tumor collections were collected under HIPAA regulations at the respective array manufacturers and no personal information, other than age and sex, about the patients were revealed to the authors. All samples were stained using standard immunohistochemical methods with DAPK1 and Med1 specific antibodies and the corresponding secondary antibodies. Isotype control antibodies did not yield any specific signals(data not shown). Images (Magnification 100X) were captured using a Nikon Eclipse E800 microscope fitted with a digital camera. At least 4-5 random fields were screened for each tumor core arrayed on these slides and data were recorded using the SPOT Advanced Version 4.6 software (Diagnostic Instruments, Inc., Sterling Heights, MI). The intensity of staining in each field was measured as pixels. Data observed with controls was considered as ++ + (100% positive) and those observed tumors were designated as – (less than 5% positive) and −/+ (10-20% positive) to denote complete and partial loss of expression, respectively.

Tumor Studies

These experiments were conducted as in our earlier studies. Three to four-week-old athymic nude (nu/nu) NCr mice (Taconic) were used for these studies as described earlier ^{30, 33}. Procedures involving animals and their care were conducted in conformity with an institutionally approved protocol that is in compliance with United States national and international laws and policies. Pathogen free cell lines were used for these experiments. Cells restored with empty vector (pCMV-3Xflag) or the same vector expressing Med1¹⁵ were used for these studies. These plasmids were mixed with pcDNA 3.1 at molar ratio of 5:1 and then transfected into cells using the Lipofectamine plus reagent. Cells were selected with G418 for 3 weeks to generate stable clones. To avoid clone specific effects, we have used pools of stable clones (n=85) for these experiments. For metastasis experiments ~20,000 cells/mouse were injected via the tail vein. Animals (n=10) were monitored for metastatic tumor development in the lungs by necropsy, 6 weeks later. Tumors were found only in the lungs. For tumor growth measurements mice (n=10) were transplanted subcutaneously with tumor cells (1X10⁶) and tumor growth was measured as in our earlier studies ³⁴. Student's two-tailed t-test was used to assess the statistical significance of difference between different groups.

Results

Loss of dapk1 expression in human cancer cells

To determine if dapk1 expression is inhibited in human tumor cell lines, we compared its expression in a number of human carcinoma cell lines to two non-oncogenic BEAS-2B and hTERT-HME1, primary mouse embryonic fibroblasts (MEF) and mouse macrophage RAW264.7 cell line using Western blot analysis. Cell extracts were prepared and subjected to Western blot analysis with a monoclonal antibody specific for DAPK1. As shown in Fig. 1A, while the non-oncogenic cells hTERT-HME1, MEF and RAW264.7expressed DAPK1 normally, almost all tumor cells expressed no DAPK1 at all, except HeLa, which is known to express a low level of it. To further define if the loss of expression occurred at the mRNA level, we performed a qPCR analysis of its mRNA. Consistent with the western blot analysis, we noted a loss of DAPK1 mRNA expression in cancer cell lines not in non-oncogenic cell lines (Fig.1B). The expression of another member of the DAPK family, DAPK2, was unaffected in these cell lines. Similarly, the expression of DAPK3 was unaffected in these cells (data not shown). Thus, dapk1 expression is suppressed in several tumor lines.

Fig.1.

Loss of expression of DAPK1 in human tumor cells: (A) Expression of DAPK1 was investigated using a western blot analysis with equivalent amount of protein (150μg) from each sample. (B) A real time PCR analysis of the dapk1 and dapk2 mRNA expression using gene specific primers. All data were normalized to those of internal control mRNAs coding for the ribosomal proteins L32 and S24 as described in our recent studies⁴. (C) Effect of 5-Azacytidine (20nM) on dapk1 mRNA expression. Cells were treated with 5-Aza for 48h prior to performing a realtime PCR analysis with dapk1-transcript specific primers.

Based on these observations, we selected two cell lines, MCF-7 and A549 as representatives to further analyze a basis for the loss of DAPK1 expression and compared the data to hTERT-HME1 and BEAS-2B cells, respectively. Since methylation has been reported as one mechanism that suppresses dapk1 expression, we first analyzed its role. MCF-7 and A549 cells were treated with 5-Azacytidine, an inhibitor of DNA methylation and the expression of dapk1 mRNA was quantified using qPCR (Fig.1C). We also used hTERT-HME1 and BEAS-2B cells as controls for this study. In none of these cells, 5-Azacytidine did not enhance dapk1 expression above those observed in the controls. Thus, methylation does not appear to be reason for a loss of dapk1 expression. These cell lines, thus, represent a population of cancers in which dapk1 expression is lost without methylation. Therefore, we further investigated the basis for a loss of dapk1 expression in these cells.

IFN-γ induced ERK1/2 phosphorylation levels were comparable in these cell lines

Our previous studies showed a critical role for ERK1/2 in regulating the interactions between C/EBP-β and transcriptional Mediator in upregulating dapk1 expression ^{4, 15}. We, therefore, examined if a differential activation of ERK1/2 occurred in normal vs cancer cells (Fig.2). An equivalent amount of total protein (100 μg) from each cell line was used for a Western Blot analysis with ERK1/2 specific antibodies. In one group we compared the results obtained from A549 http://en.wikipedia.org/wiki/Cell_(biology) to BEAS-2B (Fig. 2A&2B). In the other group MCF-7 was compared to hTERT-HME1 (Fig. 2C&D). Although ERK1/2 are activated early (10 min after IFN treatment and data not shown), we compared these activities at 4 h and 8 h after IFN-γ-treatment because these are the times points at which dapk1 expression is elevated and its regulator C/EBP-β is activated and bound to dapk1 promoter. These analyses showed a comparable activation of ERK1/2 by IFN-γ. Thus, a differential activation of ERK1/2 could not account for a lack of dapk1 expression in A549 and MCF7 cells.

Fig 2.

IFN-γ induced ERK1/2 phosphorylation levels in tumor and normal cell lines: (A &B) BEAS-2B and A549cells were treated with human IFN-γ for indicated times. Western blot (WB) analysis was performed with specific antibodies after separating 100 μg of total protein by SDS-PAGE. (C & D) show a similar analysis in hTERT-HME1 and MCF-7 cells

IFN-γ induced recruitment of Med1 to the dapk1 promoter is ablated in tumor cells

We next examined if recruitment of C/EBP-β and Med1 to the promoter was negatively affected in cancer cells. To examine this aspect, cells were stimulated with IFN-γ for several hours and soluble chromatin was prepared after in situ cross-linking. ChIP was performed using C/EBP-β- and Med1- specific antibodies. Non-specific antibodies were used as controls. Fig.3A & Fig.3B show the typical ChIP profiles obtained in BEAS-2B cells. IFN-γ-treatment readily induced the binding of C/EBP-β and recruitment of Med1 to the dapk1 promoter. A very similar profile was obtained with hTERT-HME1 cells (data not shown). To quantify the differences between nononcogenic and tumor lines in terms of C/EBP-β and Med1 recruitment, a real-time PCR analysis of dapk1 promoter was performed, using the DNA fragments recovered from the ChIP products as templates, with the CRE/ATF site specific primers. An analysis of these data demonstrated a significantly lower recruitment of Med1 in both A549 and MCF7 cell lines than in BEAS-2B and hTERT-HME1 cell lines (Fig. 3D&F). In contrast, IFN-γ induced recruitment of C/EBP-β to this site was unaffected in these cell lines. In control cells, a time dependent recruitment of C/EBP-β and Med1 (Fig. 3C&E) was observed. Thus, a failure to recruit Med1 to the promoter appears to cause a loss of dapk1 expression.

Fig. 3.

IFN-γ induced recruitment of C/EBP-β and Med1 to the dapk1 promoter is ablated in tumor cell lines compared to non-oncogenic cells: ChIP assays were performed using a commercially available kit (Upstate Biotech, Inc.).(A&B) Typical PCR patterns obtained in ChIP assays with CRE/ATF-specific primers in hTERT-HME1 cells. For input control reactions, one-fifth of the soluble chromatin used for the ChIP analysis was employed. Twenty-six cycles of PCR were performed in each case. NR-IgG (normal rabbit IgG), NG-IgG (normal goat IgG), C/EBP and Med1 IgGs were used at 10μg each/reaction. (C-F) Real-time PCR analysis of “CRE/ATF” dapk1 promoter fragments recovered in ChIP assays performed with the indicated antibodies using chromatin prepared from IFN-γ treated cells at indicated time points. Each bar represents the mean abundance of dapk1 promoter fragments. SE of six separate reactions from two independent experiments were shown.

The differential recruitment is due to differences in Med1 expression levels

The loss of Med1 recruitment to the dapk1 promoter might be due to a defective expression of Med1 in cancer cells A549, MCF7, and ZR751 compared to non-malignant normal cell lines BEAS-2B and hTERT-HME1 and MCF-10A. To further investigate whether the loss of IFN-γ dependent recruitment of Med1 to the dapk1 promoter is due to a difference in protein expression levels, we analyzed protein expression levels in these cell lines. In 1^st group A549 was compared to BEAS-2B (Fig. 4A). In the 2^nd group MCF-7 was compared to hTERT-HME1 (Fig. 4B). In the 3^rd group, MCF-10A was compared to ZR75-1 (Fig. 4C). While C/EBP-β expression levels were comparable, Med1 expression was significantly lower (7-10 fold) in A549, MCF7 and ZR75-1 compared to BEAS-2B, hTERT-HME1 and MCF-10A. These results further support our previous observations that Med1 is required for IFN-induced C/EBP-β dependent expression of dapk1 ¹⁵. Consistent with the role for Med1 in regulating C/EBP-β driven responses, another study showed that C/EBP-β dependent adipocyte differentiation and gene expression were defective in med1^−/− cells ¹⁹.

Fig. 4.

The differential recruitment of Med1 to the dapk1 promoter is due to differences in Med1 expression levels. Western blot (WB) analysis was performed with specific antibodies after separating 100 μg of total protein by SDS-PAGE and Western transfer. (A) hTERT-HME1 and MCF-7 cells were treated with human IFN-γ or EGF (positive control) for indicated time periods and cell extracts were subjected to analyses with the indicated antibodies. (B) A similar experiment was performed in BEAS-2B and A549 cells (C) Western blot analysis of the indicated proteins in MCF-10A and ZR57-1 cells. Molecular masses are indicated on the right in each case.

Loss of DAPK1 and Med1 expression in primary human lung carcinomas

Since DAPK1 expression is lost in a number of tumors, we next examined if a correlation existed between the expression of DAPK1 and Med1. Since we were able to procure significant numbers of lung tumors from commercial and non-commercial sources, we screened two separate tumor arrays on glass slides with antibodies specific for DAPK1 and Med1 using immunohiostochemical methods. One of them, a commercially available lung tumor array (Imgenex, Inc), contained a total of 50 different tumor samples corresponding to 13 squamous cell carcinomas; 7 adenocarcinomas; 4 bronchioloalveolar carcinomas; 3 adenosquamous carcinomas; 2 small cell carcinomas; 1 combined small cell and adenocarcinoma; 3 large cell carcinomas; 4 mucoepidermoid carcinomas; 1 adenoid cystic carcinoma; 1 malignant mesothelioma; 1 mucinous adenocarcinoma; 10 metastatic carcinomas (7 of which were from lymph nodes; 2 from bones and 1 from soft tissue). Each of these tumors was isolated from a different patient diagnosed with lung cancer. The ages of these patients ranged from 34 to76 years with an average of 55.38 years. There were 39 males and 10 females in this group. The sex of one patient is unknown in this group. In addition to these tumors, this array contained 9 normal lung control tissues - 4 of which were derived from adjacent non-cancerous tissues and 5 from a location away from the primary tumors.

A second array with 100 lung tumors was obtained from the Tissue Array Research Program (TARP) of the National Cancer Institute, Bethesda, MD. This array contained- 39 squamous cell carcinomas; 40 adenocarcinomas; 4 bronchoalveolar carcinomas; 15 non-small cell lung carcinomas; and 2 other poorly differentiated carcinomas. There were 1 brain and 18 lymph node metstases in this group. There were 10 cases of adenocarcinoma, 7 cases of squamous cell carcinoma, 1 case of bronchoalveolar carcinoma and 1 case of non-small cell lung carcinoma in the metastatic group. The ages 63 of patients in this group ranged from 42 to 84 years with an average of 65.12 years. Of the 63 cases 32 were males and 31 were females. No age and sex data were available for the other 37 patients. This array had 4 normal lung tissues as controls.

Immunohistochemical studies on these arrays showed a direct correlation between DAPK1 and Med1 expression. In the Imgenex array 64% of tumors exhibited a complete loss of Med1 expression. Fig.5 shows some representative data. It is interesting to note that the same tumors also lost DAPK1 expression. In 16% of these tumors, we a observed a partial loss of expression. The loss of expression of Med1 occurred independently of the tumor type studied (table I). Thus, in 40/50 tumors in this array showed Med1 loss, which consistently matched with a loss of DAPK1 expression. In the TARP array similar percentages of tumors expressed low Med1. We observed a complete loss of expression in 63.6% of the 39 squamous cell carcinomas in these tumors (some examples are shown in supplementary figure Fig.S1). A partial loss of DAPK1 and Med1 expression was observed in 22.7% of the tumors. The remainder had normal expression of Med1. Studies with 40 adenocarcinomas 30% showed a complete loss and 40% had a partial loss. In 15 non-small cell lung carcinomas 46.16% showed complete loss and 45% showed partial loss of DAPK1 and Med1 expression. Even though the other tumor types had higher percentages of loss, their numbers are too low at this stage to make reliable conclusions. It is interesting to note that both metastatic and non-metastatic tumors lost DAPK1 and Med1 expression suggesting that an early reprogramming of transcriptional activators may have lead to a down regulation of growth suppressive gene expression.

Fig. 5.

Loss of DAPK1 and Med1 in primary lung tumors. Representative images of some select tumors from the array (MH-358, Imgenex, Inc), after staining with DAPK1 and Med1 specific antibodies, are shown (magnification:100X).

Table I.

Summary of the immunohistochemical analyses of the expression of DAPK1 and Med1 in human lung cancers

Array	Type of the tumor (n)	Complete loss n (%)	Partial loss n (%)
IMGENEX	Squamous cell carcinomas (13)	9 (69.20)	4 (30.70)
	Adenocarcinomas (7)	5 (71.40)	2 (28.50)
	Bronchoalveolar carcinomas (4)	3 (75.00)	1 (25.50)
	Adenosquamous carcinomas (3)	2 (66.60)	-
	Small cell carcinomas (2)	1 (50.00)	1 (50.00)
	Combined small cell adenocarcinoma (1)	1 (100.00)	-
	Large cell carcinomas (3)	2 (66.60)	1 (33.30)
	Mucoepidermoid carcinoma (4)	3 (75.00)	1 (25.00)
	Adenoid cystic carcinoma (1)	1 (100.00)	-
	Malignant mesothelioma (1)	-	1 (100.00)
	Mucinous adenocarcinoma (1)	-	-
	Metastatic carcinomas (10)	5 (50.00)	1 (30.00)
TARP	Squamous cell carcinomas (39)	25 (64.1)	9 (23.00)
	Adenocarcinomas (40)	12 (30)	16 (40.00)
	Bronchoalveolar carcinomas (4)	3 (75)	1 (25.00)
	Non-small cell lung carcinomas (15)	7 (46.6)	7 (46.60)
	Poorly differentiated carcinoma (2)	-	2 (100.00)
	Metastatic carcinomas (19)	8 (44.4)	7 (38.80)

Restoration of Med1 upregulates basal and IFN-induced expression of dapk1 in cancer cells

Since A549 cells had the lowest level of endogenous Med1, we next investigated if restoration of its expression would upregulate dapk1 expression. Two A549 cell lines- one transfected with empty vector and another with the same vector carrying Med1 were generated. Med1 was expressed as a flag-tagged protein in these experiments and verified prior to the employment of these cells for the following experiments (Fig.6A). Baseline and IFN-induced expression of dapk1 was measured in these cell lines using qPCR. As shown in fig.5B top panel, transfection of Med1 induced the expression increased the baseline and IFN-induced expression of dapk1 mRNA. In presence of Med1, the DAPK1 expression levels were enhanced compared to the vector transfected control (Fig. 6A, compare lane 2 and lane 1). A specific requirement of Med1 for the IFN-γ induced expression of dapk1 was further ascertained in a control experiment. The expression of another IFN-γ induced tumor suppressive protein IRF1 was studied in these cells (Fig. 6B bottom panel). IFN-induced the expression of IRF1 occurred equivalently either in the absence or in presence of Med1. We next examined the recruitment of Med1 to dapk1 promoter using ChIP assays (Fig. 6C). Indeed, IFN-γ-treatment further induced the already basal binding of Med1 to the dapk1 promoter. In the vector-control no such recruitment of Med1was observed.

Fig. 6.

Restoration of Med1 establishes dapk1 expression. (A) Western blot analyses of Med1 and DAPK1 expression. (B) Top and middle panels show the real time PCR analyses of dapk1 and IRF1 mRNAs. (C) A quantitative ChIP analysis of Med1 recruitment to the dapk1 promoter. Soluble chromatin was immunoprecipitated, after cross-linking the proteins in situ, using Med1-specific antibodies and the products were subjected to qPCR analysis with dapk1 promoter specific primers (D) Effect of Med1 on dapk1-Luc expression in A549 cells. Med1¹⁵ and dapk1-luc constructs³² were described earlier. Luciferase activities were quantified as in materials and methods.

To further support these studies, we performed a reporter expression assay in which dapk1-Luc, a luciferase reporter gene driven by the human dapk1 promoter, was employed. In these experiments, the dapk1-Luc was co-transfected with an empty vector, wildtype Med1 or a functionally-inactive Med1 mutant. As shown in Fig. 6D, co-expression of wildtype, but not the mutant, Med1 robustly upregulated basal and IFN-induced expression of dapk1-Luc. The empty vector, as expected, did not promote luciferase expression. Similar results were obtained in MCF-7 cells (data not shown). Thus, Med1 appears to be critical for inducing dapk1 expression in these cells. Consistent with these observations, Med1-transfectants were significantly more sensitive to IFN-γ-induced apoptosis than the corresponding vector control (data not shown).

Attenuation of the metastatic behavior of cells following re-expression of Med1

To test the biological relevance of our observations, we used A549 cells as an experimental model of metastasis. These cells metastasized in vivo as shown in our earlier study ³⁰. Equivalent numbers of A549 cells expressing either the empty expression vector or Med1 were administered intravenously into athymic nude mice. These experiments utilized pools of Med1 transfected clones to avoid clonal artifacts. Six weeks later, animals euthanized and a necropsy was performed for the formation of metastases in lungs (Table II). While a number of metastases were observed in the vector expressing cells, a significantly fewer metastases were observed in presence of Med1 (p<0.01 compared to vector expressing cells). These experiments also had a parallel positive control in which an A549 cell line transfected with a dapk1 expression vector. These cells also formed a significantly low number of metastases, compared to the vector control (p<0.01). Lastly, BEAS-2B cells, the negative control for this study, did not form any metastases and the transfection of Med1 into these cells did not change their ability to metastasize. Apart from the differences in their metastatic ability, the empty vector- and Med1- expressing cells formed tumors of similar size, when transplanted subcutaneously (Fig.7). Similar results were obtained with MCF-7, although this cell line is slightly weaker than A549 to metastasize (data not shown). Together, these data suggested that dapk1 expression achieved through Med1 expression reestablishes anti-metastatic action.

Table II.

Effect of Med1 on tumor Metastasis

Cells expressing	No. of Metastases	Average/mouse	Percent
BEAS-2B-Vector	0	0	0.00
BEAS-2B-Med1	0	0	0.00
A549-Vector	96	9.6	100.00
A549-Med1	71*	7.1*	73.50
A549-DAPK1	66**	6.6**	68.75

Athymic nude mice were injected 0.2×10⁵ in a volume of 0.1 ml via the tail vein. n=10 mice/group. Six weeks later mice were euthanized and necropsy was performed. Metastases in the lungs of each group were counted.

^*p<0.05

^**p<0.01 with respect to vector control.

Fig. 7.

Effect of DAPK1 expression on tumor growth in vivo. Athymic nude mice (n=10) were transplanted with the A549 tumor cell lines expressing either empty vector or Med1. Tumor growth was monitored over a period of several weeks and mean tumor volume ±SE data were plotted.

Discussion

DAPK1 was originally identified as a cell death mediator induced by IFN-γ. Since then several functions have been attributed to DAPK1 in cell differentiation and cell death ²². DAPK1 suppresses invasion and metastasis of tumor cells ³. Consistent with this, expression of DAPK1 is lost in a variety of tumor cell lines ³⁵ and was down regulated in 80% of B-cell lymphoma and leukemia cell lines; and also in 30-40% of cell lines derived from bladder carcinomas, breast carcinomas and renal carcinomas ¹. Although DNA methylation seems to be a major mechanism, dapk1 expression is also silenced without apparent methylation ^23-27. Mutational inactivation(e.g. deletions, insertions, or translocations) of the dapk1gene is almost rare in many human cancers ²², except, in some familial chronic lymphocytic leukemias, where dapk1 expression is down-regulated by a unique polymorphism ²⁹.

It has been suggested that transcriptional and/or translational mechanisms may be an important means of controlling DAPK1 expression. dapk1 is upregulated in response to various stimuli indicating its role in apoptosis ^{22, 32, 36}. We have recently identified that dapk1 expression is critically dependent on C/EBP-β in an IFN-γ regulated pathway ⁴. IFN-γ induces the recruitment of C/EBP-β to CRE/ATF sites for regulating dapk1 expression. We recently showed that the Med1 subunit of the transcriptional Mediator complex is required for IFN-induced C/EBP-β-dependent expression of certain cellular genes including dapk1 ¹⁵. Med1 associates with the enhancer bound C/EBP-β at the CRE/ATF site of dapk1 in an IFN-γ dependent manner for stimulating gene expression ¹⁵. Sequence analysis of human dapk1 promoter revealed a CRE/ATF site similar to that found in the mouse dapk1 promoter ⁴. The relevance of these observations to the loss of dapk1 expression in human tumors is unknown. It appears that only DAPK1, but not the another member of its family DAPK2, was inhibited in the tumor cells (Fig.1B) indicating a specific effect on this gene. Non-oncogenic and tumor cells expressed comparable levels of DAPK2. Since the assays used in this study require large number of fresh pure tumor cells, they could not be executed in primary tumor cells. Hence, we relied on tumor cell lines, which, like many metastatic tumors, lost dapk1 expression. Our analyses showed that neither C/EBP-β nor the activation of its upstream activator ERK1/2 were defective in these cells. Given that these factors mediate multiple biological responses, it is not a surprising result. In the case of DAPK1, loss of its expression is associated with DNA methylation. However, studies with methylation inhibitor 5-Azacytidine did not result in an upregulation of DAPK1 in normal and tumor cell lines (Fig.1). The lack of 5-Aza response was not due to mutations or loss of promoter elements in these cells (data not shown). Interestingly, we found that the recruitment Med1 to the dapk1 promoter was defective in these cells, which, in turn, was due to a low level expression of Med1 (Fig.4). This observation was not limited to laboratory cell lines. A majority of primary lung cancers (n=150) showed a consistent loss of Med1 and DAPK1 expression as revealed by immunohistochemical analyses (Fig.5, Fig.S1 and table I). These data suggest, Med1 loss contributes to the attenuation of antitumor responses and promotion of tumor growth. Indeed, transfection of Med1 upregulated the expression of DAPK1 (Fig.6). As said earlier, Med1 is critical for a number of transcriptional responses ^16-19.

We further investigated IFN-γ dependent recruitment of C/EBP-β and Med1 to the dapk1 promoter in non-oncogenic cell lines (BEAS-2B and hTERT- HME1) using ChIP assays (Fig 2A and 2B). In contrast to this, in tumor cell lines only Med1 was not recruited to the promoter in an IFN-γ induced manner. Such lack of recruitment of Med1 was due to a low level expression. Interference with the functioning of transcriptional co-activators has been documented in the case of viral oncoproteins in other studies ^{37, 38}. Deletion and mutations of transcriptional co-activator p300 has been shown in some colorectal and breast tumors ³⁹. Chromosomal alterations in some leukemias ^{40, 41}; and a loss of expression of certain transcriptional co-activators during hepatocarcinogenesis ^{42, 43} have been recently reported. In some forms of Her-2 positive breast cancers, loss of transcriptional co-activator ARA70 is observed ⁴⁴. Very little is known about the regulation of med1 gene. Constitutive expression of Med1 is found in many normal cells. It is unclear what causes down regulation in tumor cells. There are gross deletions in this gene in the tumor cell lines used in this study (data not shown). It is possible that some post-transcriptional mechanisms cause its down regulation, such as a small RNA. Indeed, a very recent study showed that the snoRNA ACA45 acts as a miRNA-like inhibitor of mediator subunit CDC2L6/CDK11 ⁴⁵. It is possible that similar or other miRNAs regulate Mediator subunit levels. Thus, limiting the levels of critical transcriptional regulators for the tumor suppressive genes could serve as a strategy for tumor proliferation. Indeed, in analogous situations, oncogenic alterations in basal transcription factor activities have been reported in other studies. For example, tumor suppressors like p53, PTEN and pRb inhibit TFIIIB activity for restraining RNA polymerase III activity ^46-50. In contrast, oncogenic proteins like myc- ras- and raf- stimulate its activity ^50-52, specifically when the tumor suppressors are mutated.

Restoration of Med1 levels permitted its recruitment to the dapk1 promoter and promoted gene expression (Fig.6). More importantly, the loss of dapk1 expression was not due to a global loss of IFN-induced responses, given a normal induction of IRF1 independently of Med1 levels (Fig.6). Lastly, the physiologic relevance of our data was highlighted by the observation that Med1-expressing A549 cells, like the DAPK1-restored A549 cells, formed significantly fewer metastases compared to the controls. A notable observation in this study is that Med1-restotation did not change the overall growth of tumors in the animals (Fig.7). Instead, it selectively suppressed metastatic behavior of cells (table II). The lack of complete suppression of metastasis by either Med1 or DAPK1 suggests that other cellular factors still may contribute to tumor metastasis. Conversely, MDA-MB231, a metastatic breast cancer cell line, had a normal level of Med1 like the controls but lacked dapk1 expression. This cell line may represent the class of metastatic tumors in which methylation acts a primary suppressor of dapk1 expression. Nonetheless, our studies for the first time couple Med1 to the DAPK1 expression and also identify a critical role for Med1 in tumor cell metastasis.