ABSTRACT
The increase in AP-1 activity is a hallmark of cell transformation by tyrosine kinases. Previously, we reported that blocking AP-1 using the c-Jun dominant negative mutant TAM67 induced senescence, adipogenesis, or apoptosis in v-Src-transformed chicken embryo fibroblasts (CEFs) whereas inhibition of JunD by short hairpin RNA (shRNA) specifically induced apoptosis. To investigate the role of AP-1 in Src-mediated transformation, we undertook a gene profiling study to characterize the transcriptomes of v-Src-transformed CEFs expressing either TAM67 or the JunD shRNA. Our study revealed a cluster of 18 probe sets upregulated exclusively in response to AP-1/JunD impairment and v-Src transformation. Four of these probe sets correspond to genes involved in the interferon pathway. One gene in particular, death-associated protein kinase 1 (DAPK1), is a C/EBPβ-regulated mediator of apoptosis in gamma interferon (IFN-γ)-induced cell death. Here, we show that inhibition of DAPK1 abrogates cell death in v-Src-transformed cells expressing the JunD shRNA. Chromatin immunoprecipitation data indicated that C/EBPβ was recruited to the DAPK1 promoter while the expression of a dominant negative mutant of C/EBPβ abrogated the induction of DAPK1 in response to the inhibition of AP-1. In contrast, as determined by chromatin immunoprecipitation (ChIP) assays, JunD was not detected on the DAPK1 promoter under any conditions, suggesting that JunD promotes survival by indirectly antagonizing the expression of DAPK1 in v-Src transformed cells.
IMPORTANCE Transformation by the v-Src oncoprotein causes extensive changes in gene expression in primary cells such as chicken embryo fibroblasts. These changes, determining the properties of transformed cells, are controlled in part at the transcriptional level. Much attention has been devoted to transcription factors such as AP-1 and NF-κB and the control of genes associated with a more aggressive phenotype. In this report, we describe a novel mechanism of action determined by the JunD component of AP-1, a factor enhancing cell survival in v-Src-transformed cells. We show that the loss of JunD results in the aberrant activation of a genetic program leading to cell death. This program requires the activation of the tumor suppressor death-associated protein kinase 1 (DAPK1). Since DAPK1 is phosphorylated and inhibited by v-Src, these results highlight the importance of this kinase and the multiple mechanisms controlled by v-Src to antagonize the tumor suppressor function of DAPK1.
INTRODUCTION
The Src nonreceptor tyrosine kinase has served as the prototypical kinase model for signaling in vertebrates. Understanding the fundamental basis for Src signaling has provided insight into both the normal function of kinase signaling in the cell as well as contributing to the understanding of human disease. In particular, high Src activity has been linked to poor prognosis and metastasis in breast and colon cancers (1, 2). Given Src's role in human disease, v-Src continues to serve as a model system for understanding intracellular signaling and has provided a system for studying transformation both in vitro and in animal models. It is well established that Src-dependent transformation induces profound changes in gene expression (3–5). A previous gene expression profiling study conducted by our group showed that v-Src induces expression changes in over 2,000 genes in two different primary cell types. Overexpression of a core subset of these genes was found to be correlated with poor prognosis in breast and lung cancer (4).
Signaling transduction cascades regulated by v-Src mediate the activation of transcription factors acting on promoter/enhancer regions. The roles of Ets, Stat3, and AP-1 in transformation have been documented through the inhibitory effect that their dominant negative alleles exert on v-Src or RasV12-dependent transformation (6–11). Inhibition of AP-1 activity in immortalized jun−/− mouse embryo fibroblasts (MEFs) renders cells resistant to transformation by activated Ras. Ectopic expression of c-Jun restores transformation of these MEFs in vitro and is capable of generating tumors in nude mice (12). Similarly, overexpression of JunD partially restores the transformation and tumor-generating potential in jun−/− MEFs (12). Knockout c-jun and junD MEFs also proliferate slowly and are prone to early senescence (12, 13). junD−/− MEFs, furthermore, are sensitized to apoptosis when treated with tumor necrosis factor alpha (TNF-α) (13). Taken together, these data illustrate the importance of the AP-1 family, particularly jun members, in transformation and survival.
Recently, our group demonstrated the pleiotropic action of AP-1 by inhibiting AP-1 activity through the expression of short hairpin RNAs (shRNAs) targeting c-jun, junD, and fra-2 or by repressing AP-1 via the c-Jun dominant negative allele TAM67. In normal chicken embryo fibroblasts (CEFs) TAM67 induces senescence; however, when the same cells are transformed by v-Src, they exhibit a pleiotropy where three distinct phenotypes are visible. In addition to senescence, a fraction of cells undergo adipogenesis while others undergo apoptosis (14). Since TAM67 can dimerize with Jun, Fos, and ATF family members (15–17), these heteromeric interactions may alter the activity of these proteins as well as c-Jun. Presumably, through a stochastic process, cells give rise to different phenotypes through differential inactivation of different AP-1 members. Resection of these distinct cell fates by shRNA showed that c-Jun, Fra-2, and JunD mediate distinct outcomes by antagonizing senescence, adipogenesis, and apoptosis, respectively (14). Significantly, apoptosis was not detected when c-Jun and Fra-2 were inhibited individually by shRNA expression. Therefore, by comparing the transcription profiles of TAM67 versus junD shRNA-expressing cells, in both transformed and untransformed backgrounds, we sought to identify genes regulating apoptosis in v-Src-transformed CEFs with inhibited AP-1 activity. Our microarray analysis identified a list of 11 candidate genes associated with the induction of apoptosis. Four genes in this list are members of the interferon (IFN) pathway. Among these, the death-associated protein kinase 1 (DAPK1) gene product was initially identified as a C/EBPβ-dependent proapoptotic protein in IFN-γ-induced cell death (18; reviewed in reference 19). DAPK1 is an activator of the Arf/p53 pathway and an established tumor suppressor (20–22). Furthermore, our lab has shown that inhibition of p53 by shRNA restores colony formation in v-Src-transformed CEFs expressing the JunD shRNA (14), suggesting that JunD-inhibited apoptosis is acting through DAPK1 in a p53-dependent manner. In this study, we show that inhibition of DAPK1 restores survival in v-Src-transformed CEFs with inhibited AP-1 activity. Furthermore, DAPK1 expression is inhibited by the expression of a dominant negative mutant of C/EBPβ and is antagonized by JunD/AP-1 expression.
RESULTS
Gene profiling of AP-1 inhibition in v-Src-transformed CEFs.To investigate the role of AP-1 in v-Src-mediated transformation, we undertook a gene profiling study to characterize the transcriptomes of v-Src-transformed CEFs expressing the c-Jun dominant negative mutant TAM67 or a JunD shRNA. It was reasoned that by comparing the transcriptomes of transformed CEFs repressing AP-1 activity using either the TAM67 dominant negative or the JunD shRNA, we could identify a gene or genes affecting the survival of v-Src-transformed cells. Since the induction of apoptosis is dependent on v-Src transformation, we performed these analyses at the permissive (P) and restrictive (nonpermissive, NP) temperatures, using cells infected with a temperature-sensitive mutant of Rous sarcoma virus (RSV). Therefore, CEFs infected with a temperature-sensitive RSV mutant, NY72-4, were coinfected with the TAM67-containing virus, the JunD shRNA-containing virus, or the RCASB green fluorescent protein (GFP) shRNA control virus. Following array normalization and probe set intensity estimation, two multiple pairwise analyses were conducted to identify genes of interest. The first set was a multiple pairwise analysis of genes differentially expressed by 2-fold or more between CEFs infected with NY72-4 and GFP shRNA at the permissive and restrictive temperatures, CEFs infected with NY72-4 and JunD shRNA at the permissive and restrictive temperatures, and between CEFs infected with NY72-4 and JunD shRNA or NY72-4 and GFP shRNA, both at the permissive temperature. Similarly, the second set was an analysis of differential gene expression between CEFs infected with NY72-4 and GFP shRNA at the permissive and restrictive temperatures, CEFs infected with NY72-4 and TAM67 at the permissive and restrictive temperatures, and CEFs infected with NY72-4 and TAM67 or NY72-4 and GFP shRNAs, both at the permissive temperature (Table 1 summarizes the number of probe sets corresponding to differentially expressed genes in each comparison; see Table S1 in the supplemental material for all significantly differentially expressed genes for both analysis sets). A total of 2,468 probe sets corresponding to 1,539 unique differentially expressed genes passed the significance criteria.
Summary of differentially expressed transcripts in comparisons indicated as determined by microarray analysis
In order to identify candidate genes involved in regulating apoptosis in v-Src-transformed cells, we considered the probe sets corresponding to genes that were differentially expressed in transformed CEFs expressing either TAM67 or the JunD shRNA in relation to control-transformed CEFs. Pathway perturbation analysis of orthologous genes in these two sets (misexpressed with TAM67 or JunD shRNA) and the set of genes differentially expressed between normal and transformed CEFs show 32 dysregulated pathways (Table 2). Thirteen pathways are shared between two or more sets, and eight are shared by all three (phosphatidylinositol signaling system, focal adhesion, extracellular matrix [ECM]-receptor interaction, TGF-β signaling pathway, pathways in cancer, natural killer cell-mediated toxicity, mitogen-activated protein kinase [MAPK] signaling pathway, and calcium pathway).
Common pathways found to be dysregulated in gene comparison setsa
Unsupervised hierarchal clustering was performed to identify gene clusters that were upregulated in transformed CEFs with repressed AP-1 activity but not in nontransformed cells or transformed CEFs with unperturbed AP-1 function (Fig. 1). Using this process, we identified a cluster of 18 probe sets corresponding to 11 unique annotated genes that were upregulated in transformed CEFs with repressed AP-1 activity (Fig. 1 and Table 3). Four of these genes are members of the interferon (IFN) pathway (DAPK1, IFIT5, OASL, and SERPINB2) and 3 of the 11 genes are regulated by C/EBPβ (DAPK1, SERPINB2, and IL-6) (18, 23, 24). In particular, DAPK1 is a C/EBPβ-regulated protein that mediates apoptosis in response to IFN-γ signaling (18). Since the inhibition of AP-1 enhances the activity of a promoter element controlled by C/EBPβ in v-Src-transformed CEFs (14, 25), DAPK1 is a candidate for the induction of cell death in these cells.
A cluster of 18 probe sets are upregulated in JunD shRNA- and TAM67-infected cells at the permissive temperature but not at the restrictive temperature and not in either of the GFP shRNA controls. (A) Unsupervised clustering of genes upregulated in NY72-4 JunD- and NY72-4 TAM67-infected CEFs grown at the permissive temperature in relation to control cells (NY72-4- and GFP shRNA-coinfected CEFs) grown at the permissive temperature. Examination of clusters reveals a group of 18 probe sets that are upregulated in JunD shRNA- and TAM67-infected cells at the permissive temperature but not at the restrictive temperature and not either of the GFP shRNA controls (indicated in red on the dendrogram). The scale on heat map indicates standardized expression values (−3 to 3 standard deviations). The plus sign indicates permissive temperature. (B) Gene-wise mean-centered expression values for genes found in the cluster. Each point indicates the mean standardized gene expression in a replicate array for the condition indicated at the bottom. Expression profiles were obtained for triplicate samples with the exception of NY72-4-infected CEFs with the JunD shRNA, which were examined in sextuplicate at the permissive temperature. Error bars show the standard errors of means of standardized expression values. NP and P indicate nonpermissive and permissive temperatures, respectively. (C) Line plot indicating gene-wise mean-centered expression values for genes found to be upregulated in NY72-4 JunD- and NY72-4 TAM67-infected CEFs grown at the permissive temperature in relation to control cells (NY72-4- and GFP shRNA-coinfected CEFs).
Mean log2 gene expression values for genes found to be upregulated in NY72-4 JunD shRNA- and NY72-4 TAM67-infected CEFsa
Validation of DAPK1 activation.To assess the induction of DAPK1 in response to AP-1 inhibition in transformed CEFs, we conducted reverse transcription-quantitative PCR (RT-qPCR) for the conditions used in the microarray analysis. Validation of expression by RT-qPCR showed that the DAPK1 expression increases 2- to 4-fold over control conditions and shows a similar profile of expression compared to that of the microarray results (Fig. 2A). Immunoblotting confirmed the 2-fold induction of DAPK1 in v-Src-transformed CEFs expressing the junD shRNA (Fig. 2B, lane 4, and data not shown). Similar 2- to 4-fold increases in DAPK1 expression were reported in response to IFN-γ stimulation and p53 activation elsewhere (18, 26).
Validation of DAPK1 expression. (A) Characterization and comparison of DAPK1 mRNA expression by RT-qPCR and microarray analysis. Red points indicate mean expression of DAPK1 on the microarray while the bar graph shows mean expression data as determined by RT-qPCR. Error bars indicate standard errors of means. Differences in DAPK1 expression levels as determined by qPCR were statistically significant between the GFP shRNA controls and TAM67- and JunD-shRNA-expressing experimental groups as well as between the NY72-4 TAM67 groups at permissive and nonpermissive temperatures (pairwise two-tailed t test, Bonferroni-corrected α = 0.05). Differences in DAPK1 expression levels were significant between the NY72-4 JunD-shRNA groups at an α of 0.1. (B) Immunoblotting validation of DAPK1 induction by v-Src under conditions of JunD inhibition by shRNA. The star indicates the position of a putative product of degradation of DAPK1 generated during sample preparation. NP and P indicate nonpermissive and permissive temperatures, respectively.
Repression of DAPK1 by shRNA restores transformation in v-Src-transformed CEFs with impaired AP-1 activity.Previous work in our lab showed that inhibition of JunD activity by shRNA increases apoptosis in v-Src-transformed CEFs approximately 12-fold and abrogates colony formation in soft agar (14). Coinhibition of JunD with p53 restores colony formation (14). To assess if DAPK1 functions in this pathway, a double shRNA construct was used to assay colony formation in transformed CEFs. Coinhibition of JunD and DAPK1 restored colony formation in transformed cells (Fig. 3A). Interestingly, this inhibition of JunD and DAPK1 produced colonies in greater numbers and larger sizes than those produced by the control alone (Fig. 3B and C), suggesting a synergistic mechanism of DAPK1 regulation, i.e., both via transcriptional control and by direct Src-mediated phosphorylation (see Discussion) (27).
v-Src-transformed CEFs with coinhibition of JunD and DAPK1 by shRNA form larger and more numerous colonies in soft agar. (A) Representative fields of a colony formation assay are shown. All CEFs are v-Src transformed and coinfected with the viruses indicated. (B) The number of colonies per field greater than 50 μm under each condition is shown. The mean number of JunD/DAPK1 double-knockdown CEF colonies is greater than that for the GFP shRNA control and the JunD single-knockdown CEFs (one-way analysis of variance followed by Tukey post hoc test, P < 0.05). The mean numbers of colonies per field for GFP shRNA, JunD, and JunD/DAPK1 are 4.22, 1.40, and 8.70, respectively. (C) The mean colony diameter of JunD/DAPK1 double-knockdown cells is greater than that of the GFP shRNA control and the JunD single-knockdown CEFs (Kruskal-Wallis test followed by Dunn's post hoc test, P < 0.05). Outlier data greater than 350 μm are not shown. Mean colony sizes for GFP shRNA, JunD, and JunD/DAPK1 are 138.6 μm, 113.7 μm, and 163.5 μm, respectively. Whiskers in both plots indicate 5th to 95th percentiles. Levels of significance in pairwise tests are indicated in brackets above box plots. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Representative fields from colony formation assays quantified in the experiments shown in panels B and C are shown at a magnification of ×20. Colony formation assays were quantified using ImageJ (59). All features in each field greater than 50 μm were counted. Feret's diameter was calculated and used for statistics. A minimum of nine fields per condition was used. Distribution normality was ascertained using the Shapiro-Wilk test followed by the appropriate analysis of variance (Kruskal-Wallis or analysis of variance). (D) Immunoblotting validation shows DAPK1 and JunD levels are reduced using single or double shRNA constructs as indicated.
The downregulation of AP-1 promotes transcriptional activation of the DAPK1 promoter in v-Src-transformed CEFs.In murine cells, the induction of DAPK1 by IFN-γ is modulated at the transcriptional level (18). To characterize the effect of AP-1 inhibition, we first examined the activity of a chicken DAPK1 promoter construct in CEFs expressing the junD shRNA and coinfected with NY72-4. JunD inhibition increased DAPK1 promoter activity at the permissive temperature but had no effect when temperature-sensitive v-Src activity was attenuated at 41.5°C (Fig. 4A). The DAPK1 promoter construct was also activated in a v-Src-dependent manner when AP-1 was inhibited by the expression of TAM67, the dominant negative mutant of c-Jun. However, this effect was abolished by coexpression of the Δ184 dominant negative mutant of C/EBPβ (Fig. 4B).
Control of the DAPK1 promoter by JunD and C/EBPβ. (A) Repression of JunD by shRNA activates the full-length DAPK1 promoter in v-Src-transformed CEFs. NY72-4-infected cells were coinfected with RCASB control virus or the JunD shRNA virus as shown. (B) Repression of C/EBPβ by the Δ184 dominant negative mutant of C/EBPβ abrogates activity of the full-length DAPK1 promoter in transformed and untransformed CEFs. NY72-4-infected cells were coinfected with RCASB control virus and TAM67 or Δ184 and TAM67 expression viruses as shown. Error bars indicate standard errors of means. Differences between experimental groups are statistically significant unless otherwise indicated (multiple pairwise t tests, Bonferroni-corrected, P < 0.05; n.s., not significant). A minimum of six replicates per condition was used. NP and P refer to nonpermissive and permissive temperatures, respectively. (C) Chromatin immunoprecipitation indicates that C/EBPβ is recruited to the DAPK1 promoter under conditions of AP-1 inhibition. NY72-4-infected CEFs were coinfected with either the RCASB control virus or TAM67 expression virus as indicated. The C/EBPβ-immunoprecipitated DAPK1 promoter region is PCR amplified under conditions of AP-1 repression in v-Src-transformed CEFs. Only background signal is observed in samples immunoprecipitated with rabbit IgG or anti-JunD antibody. (D) A distal intron sequence is not amplified when chromatin is precipitated using any of the antibodies indicated. Error bars indicate standard errors of means. RLU, relative light units.
Chromatin immunoprecipitation (ChIP) assays were performed with anti-C/EBPβ or anti-JunD antibodies or with rabbit IgG control serum. The JunD antibody was previously used to show interaction of JunD with the interleukin-8 (IL-8) promoter in response to v-Src activation (14). Analysis of the qPCR data shows that C/EBPβ is associated with the DAPK1 promoter in response to AP-1 inhibition in transformed CEFs (Fig. 4C). In contrast, the signal using anti-JunD does not exceed the background level seen in the IgG control, suggesting that JunD does not bind to the DAPK1 promoter under any condition. Amplification of the DAPK1 intron was an order of magnitude weaker than the promoter, indicating that the binding of C/EBPβ was not the result of contamination or nonspecific immunoprecipitation conditions (Fig. 4D). These data show that C/EBPβ is recruited to the DAPK1 promoter in v-Src-transformed cells under conditions of AP-1 repression. No evidence of JunD interaction with the promoter was seen, suggesting that the effect of AP-1/JunD on DAPK1 expression is indirect.
In mammals, IFN-γ-mediated apoptosis is dependent on C/EBPβ, whose activity is potentiated by extracellular signal-regulated kinase 1 and 2 (ERK1/2) (18). Analysis of the murine Dapk1 gene revealed that activation of the promoter is mediated through a proximal cyclic AMP (cAMP) response element (CRE) and a distal C/EBPβ-binding site (18). Work by Hu and colleagues (28) showed that ERK2 represses expression of IFN-γ-induced genes by binding to a modified C/EBPβ binding site known as a GATE (gamma-IFN-activated transcriptional element), originally characterized in the context of IFN-γ signaling (28). Analysis of the chicken DAPK1 gene revealed a potential TATA box and initiator element located upstream of an 800-bp GC-rich region adjoining the initiation codon and corresponding to the putative 5′ untranslated region of the DAPK1 mRNA. Two putative C/EBP sites, along with one CRE and a potential GATE, were also identified in the proximal promoter region (Fig. 5). Finally, a composite C/EBP binding site known as a cis-acting replication element (CARE) was located further upstream. CAREs are comprised of two-half sites that bind to C/EBPβ and ATF4 in response to stress and amino acid deprivation (reviewed in reference 29).
Structure of the upstream regulatory region of chicken DAPK1. (A) Schematic representation of DAPK1 promoter. The positions of a putative transcription start site (+1), TATAAA box, C/EBP binding sites, cAMP response element (CRE), and CARE are indicated. A GATE (in bold) is embedded within the −134 C/EBP binding site. Arrows indicate orientation of the element. (B) The sequence of the DAPK promoter is shown with relevant features indicated.
We performed a series of in vitro reporter assays to identify critical elements of the DAPK1 promoter required for activation in response to AP-1 inhibition. Deletion analysis showed that removal of the CARE attenuated reporter activity in CEFs but did not eliminate induction upon transformation in TAM67-expressing cells (Fig. 6A). Deletion of the candidate 5′ C/EBP site eliminated inducibility of the promoter as seen with constructs −146 and −134. Further deletion of a 38-bp region that includes the CRE rendered the promoter completely inactive (construct −45 in Fig. 6A).
Identification of regulatory elements of the DAPK1 promoter. (A) Quantification of Gaussia luciferase reporter activity in NY72-4-infected CEFs with 5′ deletion constructs of the DAPK1 upstream region. (B) Schematic representations of the deletion constructs used in the experiment shown in panel A. (C) Quantification of Gaussia luciferase reporter activity in NY72-4-infected CEFs with the DAPK1 promoter construct with mutations of candidate regulatory elements. (D) Schematic representations of the mutagenesis constructs used for the experiment shown in panel C. Cells were coinfected with the RCAS control virus or TAM67 expression virus as shown. Asterisks indicate statistical significance as determined by multiple pairwise t tests (Bonferroni corrected, P < 0.05). Error bars indicate standard errors of means. Six replicates per condition were used. NP and P refer to nonpermissive and permissive temperatures.
To ascertain the role of individual elements of the promoter, we generated point mutation constructs for the candidate distal C/EBP site, the GATE, and the CRE. Mutation of either the distal C/EBP site or the CRE (constructs μ146 and μ83) decreased reporter activity, whereas mutation of the GATE had no effect (Fig. 6C and D). Although the activity of the promoter region was greatly impaired, both the μ146 and μ83 mutants retained at least partial inducibility, suggesting the existence of multiple responsive elements regulated under conditions of AP-1 inhibition. The mutant CRE construct retained approximately four times more activity than the mutated C/EBP site, consistent with the deletion data and suggesting that the candidate distal C/EBP site is a more critical activator of DAPK1 transcription. Taken together, these data indicate that the CRE and the distal C/EBP element are the primary sites required for the activation of DAPK1. Although the function of the CARE was not assessed directly, deletion of the CARE and the promoter region upstream of the distal C/EBP site attenuated reporter activity but did not affect the inducibility by TAM67. This suggests that the CARE, or some other element upstream of the distal C/EBP site, acts to potentiate basal DAPK1 promoter activity but is not required for the induction under conditions of AP-1/JunD inhibition.
DISCUSSION
The activation of DAPK1 causes apoptosis in v-Src-transformed cells with JunD/AP-1 inhibition.In order to identify candidate genes involved in the inhibition of survival in AP-1-repressed v-Src-transformed cells, we carried out an expression profile analysis on v-Src-transformed primary CEFs with and without AP-1 inhibition. By using a c-Jun dominant negative allele (TAM67) and a junD shRNA to repress AP-1 activity, we were able to identify common targets expressed in v-Src-transformed CEFs. Analysis of the six conditions showed a total of 1,539 unique dysregulated genes. Pathway perturbation analysis of the comparison sets used for clustering showed a dysregulation of 13 pathways involved in cancer signaling, immunity, or cell adhesion. Five of these were uniquely shared between the AP-1-repressed comparison sets (Table 2), illustrating the similarity of pathway regulation between junD shRNA and TAM67. In several cell types, Src is a potent inducer of prosurvival genes of the Bcl2 family including Bcl2 and Bcl-xL (30–32). However, these genes were not misregulated in response to JunD/AP-1 inhibition in our experiments. Instead, clustering analysis revealed a set of 11 unique genes upregulated by the expression of TAM67 or the junD shRNA in v-Src-transformed cells but not in normal transformed or untransformed CEFs. Four of these genes (DAPK1, SerpinB2, OASL, and IFIT5) are induced in response to IFN-γ, and two of these IFN-γ-responsive genes (DAPK1 and SerpinB2) are targets of C/EBPβ, suggesting a common mechanism of regulation, i.e., through C/EBPβ (33). Consistent with this, downregulation of AP-1 enhances the activity of C/EBP-controlled promoter elements in CEFs (14, 34). DAPK1 has been shown to be indispensable in cell death mediated by IFN-γ, transforming growth factor β (TGF-β), TNF-α, DNA damage, and oxidative stress (reviewed in reference 35). JunD has a cytoprotective role in the response to oxidative stress in several cell types (36–41). Preliminary analyses indicated that v-Src-transformed CEFs expressing TAM67 or the junD shRNA have elevated levels of reactive oxygen species (our unpublished results). Therefore, the inhibition of JunD/AP-1 may generate conditions that stimulate both the expression and activity of DAPK1. By inhibiting the expression of DAPK1, we were able to rescue the colony formation phenotype of AP-1-inhibited v-Src-transformed CEFs.
Promoter analyses revealed that transcriptional activation of the DAPK1 promoter was dependent on two separate elements corresponding to a distal C/EBP site and a CRE (Fig. 6). This reflects the structure of the murine Dapk1 promoter, which is regulated by a proximal CRE and a distal C/EBPβ-binding site (18, 28). While the trans-acting proteins binding to the regulatory elements of the chicken DAPK1 promoter remain to be characterized, the results of chromatin immunoprecipitation assays and the effect of the dominant negative mutant Δ184-C/EBPβ suggest that C/EBPβ is involved in the induction of DAPK1 (Fig. 6). Significantly, the conservation of this regulatory mechanism across distant taxa emphasizes the important role that DAPK1 plays in regulating apoptosis. C/EBPβ is known to dimerize with the stress-responsive factor ATF4 and bind to CREs on other genes (42, 43). Cellular transformation induces oxidative stress, and DAPK1 has been shown to act in the oxidative stress response pathway mediated by protein kinase D (PKD) (44, 45; also see above). It is therefore possible that ATF4 cooperates with C/EBPβ in the regulation of DAPK1. Factors interacting with the C/EBP site and CRE are under investigation.
Binding of JunD was not observed in the ChIP assays, suggesting that the role of AP-1 in repression of DAPK1 is indirect. It was previously shown that overexpression of C/EBPβ antagonizes AP-1 activity and decreases AP-1 levels, while repression of C/EBPβ using a dominant negative mutant shows the inverse (34). More recently, we showed that the inhibition of AP-1 leads to an increase of C/EBP activity and vice versa (14).
C/EBPβ is controlled by an autoinhibitory mechanism regulated by several oncoproteins and signaling cascades (46). There is some conflicting evidence on the exact mechanism underlying this process, with some reports indicating that both the DNA binding activity and transcriptional activation function are controlled through this mechanism while others concluded that transcriptional activation is solely affected by the intramolecular inhibitory interaction (47, 48). Western blotting indicated that the expression of C/EBPβ is not stimulated by the inhibition of JunD/AP-1 (our unpublished data) while the results of ChIP assays demonstrated that C/EBPβ is recruited to the DAPK1 promoter in response to v-Src activation and AP-1 impairment in CEFs (Fig. 4C). The recruitment of C/EBPβ remains to be characterized but may depend in part on the interaction with unrelated transcription factors that are also regulated by v-Src and AP-1 inhibition. In particular, this may apply to factors interacting with the CRE.
While AP-1 does not interact directly with the DAPK1 promoter, AP-1 and a C/EBP factor may compete for common cofactors such as p300. Similar inhibition of gene expression by cofactor sequestration through AP-1 has been observed previously (49). This mechanism is presently under investigation.
Multiple pathways of DAPK1 regulation in v-Src-transformed cells.Src is known to inhibit DAPK1 by phosphorylating Y491 while the LAR phosphatase reverses this during periods of low Src activity (27). Transcriptional activation of DAPK1, as observed under our conditions of AP-1 inhibition, would therefore override the action of Src on DAPK1. The regulation of DAPK1 gene expression is also confounded by a significant amount of cross talk between the different pathways regulating DAPK1 activity, the pathways regulated by DAPK1 itself, and the transcription factors controlling the expression of DAPK1. ERK-mediated phosphorylation of DAPK1 on S735 (50) upregulates kinase activity while paradoxically, ribosomal S6 kinase (RSK), a target of ERK, decreases DAPK1 activity by phosphorylating S289. This is further complicated by the fact that DAPK1 can sequester ERK in the cytosol, thereby attenuating ERK activity and hence DAPK1 through a negative feedback loop. ERK also activates C/EBPβ by phosphorylating T189 in the transactivation domain (18). In v-Src-transformed CEFs, the activation of ERK, the phosphorylation of the JunD transactivation domain, and the activity of AP-1 are all markedly reduced by treatment with a MEK inhibitor (our unpublished results). Therefore, the inhibition of DAPK1 by tyrosine phosphorylation and the stimulation of both JunD/AP-1 and C/EBPβ by ERK may be required to ensure the survival of v-Src-transformed cells (14, 25). Finally, DAPK1 has been shown to upregulate p53 through p14Arf (20) even though DAPK1 itself is a target of p53 regulation (26). These observations suggest that cross-regulatory pathways control the activity of a small network of transcription factors that includes AP-1, C/EBPβ, and p53 and determines the fate of the cell. An imbalance in the activity of these factors alters the survival, proliferation, and differentiation state of v-Src-transformed cells (14).
The induction of apoptosis observed in response to the inhibition of JunD/AP-1 depends on DAPK1 and p53 since downregulation of any of these factors restores cell viability and the colony formation capacity of v-Src-transformed CEFs (Fig. 3 and 4) (14). Whether or not the induction of any of the genes coregulated with DAPK1 plays a role in the action of DAPK1 and the loss of viability of v-Src-transformed cells remains to be investigated (Table 2). IL-6, a target of C/EBPβ, is a mediator of the oncogene-induced senescence program triggered by the oncogenic BRAFE600 allele in human fibroblasts (51). The gene for PPP1R3C (protein phosphatase 1 regulatory subunit 3C) is hypermethylated in several cancers and is a candidate tumor suppressor gene (52–56). CBLB (Cas-Br-M retroviral transforming sequence b), a member of the Cbl family of ubiquitin ligases, promotes mono-ubiquitination of proteins targeted for lysosomal degradation (57). The increased expression of cathepsin K/L (Table 3) and the function of DAPK1 in the induction of autophagy also suggest a role for lysosomes in the impairment of transformation and cell viability (58). Thus, the downregulation of genes coregulated with DAPK1 by RNA interference may provide additional insight into the function of JunD/AP-1 in v-Src transformation and the mechanism of action of DAPK1.
MATERIALS AND METHODS
Cell culture.Early passages (n < 10) of CEFs were cultured at 41.5°C in Dulbecco's modified Eagle medium (DMEM) with 5% heat-inactivated calf serum (Cosmic calf serum; HyClone, Logan, UT), 5% tryptose phosphate broth, l-glutamine, and penicillin-streptomycin. CEFs were transfected with B-type RSV viral vectors expressing shRNAs (JunD, C/EBPβ, DAPK1, JunD-DAPK1 double shRNA, JunD-C/EBPβ double shRNA, and a GFP shRNA control) or the c-Jun TAM67 dominant negative mutant using the calcium phosphate method. CEFs were superinfected with the temperature-sensitive A-type NY72-4-RSV one passage following transfection and cultured for three passages or until the apoptosis phenotype was observed in response to AP-1 inhibition and transformation. Infected CEFs were cultured at the nonpermissive (NP) temperature of 41.5°C until the temperature shift was performed. Temperature shift was performed for 6 to 8 h at 37.5°C.
Colony formation assays.The assay was performed in 60-mm dishes with a lower layer of 0.5% low-melting-point agar in 1× DMEM containing 5% calf serum, 4% chicken serum, 5% tryptose phosphate broth, glutamine, and penicillin-streptomycin. A total of 105 cells were resuspended in the same medium with 0.35% low-melting-point agar and then were overlaid on the lower agar. The dishes were then incubated in a 37.5°C incubator containing 5% CO2 for 5 days or until colonies were visible. Images of colonies were documented with an inverted microscope (magnification of ×20). Colony formation assays were quantified using ImageJ software (59).
Generation of retroviral vectors for shRNA expression.Generation of the single and double shRNA RCASBP(B) constructs has been described elsewhere (14, 60). Several target sequences were selected for EGFP, junD, and DAPK1, subcloned into the modified microRNA operon in the transfer plasmid pRFPRNAi(-), and later transferred into the RCAS vectors according to the original supplier's instructions (ARK Genomics) (61). The control virus RCASBP(B)-shRNA-EGFP and RCASBP(B)-shRNA-junD (or simply RCASB-GFP shRNA and RCASB-JunD shRNA) were described previously (14); the following sequences were targeted for downregulation by shRNA: chicken JunD (nucleotides [nt]990 to 1010), GAAGAGCCTCAAGAGCCAGAA (NCBI RefSeq accession number XM_015300148.1 ); chicken DAPK1 (nt 298 to 319), CGGGAAGATATTGAGCGAGAA (NCBI RefSeq accession number NM_001278031.1 ); EGFP (nt 484 to 504), GCACAAGCTGGAGTACAACTA (GenBank accession number LC008492.1 ). Suppression of gene expression was ascertained by immunoblotting for the corresponding proteins.
Gene profiling analyses.RNA samples were isolated using TRIzol (Life Technologies) as described previously (4). All RNA samples were first analyzed by Northern blotting and probed for IL-8 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression to assess integrity. RNA quality was assessed by gel electrophoresis and examined by a Bioanalyzer (Agilent). RNA samples with an RNA integrity index of less than 9.7 were discarded. Microarray experiments were conducted at the Centre for Functional Genomics (CFG) at McMaster University, Hamilton, Canada. Biotinylated cRNAs were generated at CFG and hybridized to Affymetrix Chicken GeneChip arrays using standard Affymetrix protocols (EukGE-WS2v4). GeneChips were scanned using an Affymetrix GeneChip Scanner 3000. Feature intensity was quantified using Command Console software and exported to CEL format. All experimental groups consisted of three biological replicates except for NY72-4 RCASBP(B)-shRNA-ΔU6-JunD permissive (NY72-4 JunD shRNA P), which was conducted in sextuplicate. These samples undergo apoptosis rapidly upon transformation and show a greater intragenic variance in gene expression than the other samples. Therefore, the number of replicates was increased in order to increase sensitivity of downstream analysis.
CEL files were analyzed using Affymetrix Expression Console software, version 1.1. Array data were normalized, and probe intensities were estimated using the Affymetrix probe logarithmic intensity error (PLIER) algorithm. Twofold changes in gene expression levels between experimental conditions within each experiment were determined using log2-transformed expression values, and statistical significance of expression was determined using a Bonferroni-corrected two-tailed t test (α = 0.05). Gene clustering was performed using unsupervised hierarchal clustering by average Euclidean distance (62). Mean-centered gene expression values and clustering were carried out using PLIER-calculated values in dChip, version 2007 (63, 64). Pathway perturbation analysis was carried out using Pathway Express software (65) as described previously (4) except that orthologous human Entrez gene annotation was used instead of probe identification numbers. Annotations were retrieved from the Affymetrix NetAffx service (Affymetrix, Santa Clara, CA).
Antibodies and immunoblotting.Immunoblotting was performed as described before (14, 66) using the following antibodies: anti-Erk1 (sc-94; Santa Cruz Biotechnology), anti-JunD (sc-74; Santa Cruz Biotechnology), and anti-DAPK1 (sc-8163; Santa Cruz Biotechnology). Anti-chicken C/EBPβ was generated in-house and described previously (25). Chemiluminescence detection was used to visualize all antibody complexes (Lumina Forte; Millipore, Billerica, MA, USA), and signals were scanned, analyzed with ImageJ, and corrected for loading using Erk as a control (59). For ChIP, rabbit IgG and anti-JunD were obtained from Santa Cruz Biotechnology (sc-2027 and sc-74, respectively) and used as described previously (14). Anti-C/EBPβ antiserum used in ChIP was a gift from Achim Leutz (67). Results described in Western blotting analyses have been replicated in at least three independent sample sets prepared from different primary cultures of CEFs isolated from different embryos.
Real-time PCR.For expression validation, DNase I-treated RNA samples were reverse transcribed using a ProtoScript cDNA synthesis kit (New England BioLabs). PCR amplification and quantification were performed using PerfeCTa SYBR Green FastMix, Low ROX (Quanta Biosciences), on a Stratagene MX3000P real-time PCR instrument. Probe amplification was quantified using the ΔΔCT method (where CT is threshold cycle) (67) using GAPDH as the reference gene for array validation assays or input DNA for ChIP assays. The following oligonucleotides were used to amplify the targets indicated: 5′-GTCGGAGTCAACGGATTTGGCCG-3′ (GAPDH forward); 5′-ATGGCCACCACTTGGACTTTGCC-3′ (GAPDH reverse); 5′-TCAACCCAACAAGCATGGAACACCT-3′ (DAPK1 forward); 5′-CGGGATCCACGCTTTAGAAGCAACT-3′ (DAPK1 reverse); 5′-CCCTCAGGGCTGAGCAGTGCA-3′ (DAPK1 promoter forward); 5′-AGCATATGGCCCAGTGCTGG-3′ (DAPK1 promoter reverse); 5′-TCCCTAGGCCCCAGCTGCC-3′ (DAPK1 intron forward); 5′-CGGAGGGGCAGTGCTGAGCT-3′ (DAPK1 intron reverse).
ChIP assays.The JunD antibody protocol for chromatin immunoprecipitation (ChIP) was described previously (14) and was performed according to the method and reagents described in the EZ-ChIP protocol (Upstate Biotechnology-Millipore). Chromatin was prepared from CEFs infected with NY72-4 and RCASB-GFP shRNA or NY72-4 and RCASB-TAM67 at both permissive and restrictive temperatures. Chromatin corresponding to 500 mg of protein was immunoprecipitated using 1 mg of normal rabbit IgG, anti-C/EBP antibody, or anti-JunD antibody. Purified chromatin was amplified by real-time PCR as described above. One percent of the input from each sample was used to normalize PCR amplification. All PCR experiments were performed using a minimum of three replicates. The ChIP promoter primer set amplifies a 92-bp region encompassing the proximal C/EBP site, TATA box, and initiator region corresponding to nucleotides −69 to +24, as indicated in Fig. 5. The ChIP intron primer set amplifies a 97-bp region in the second intron of the DAPK1 gene spanning nucleotides 20244 to 20340 in the DAPK1 genomic sequence (NCBI RefSeq accession number NC_006127.4 ).
Transient reporter assays.pGLuc-derived reporter constructs and the RSV-β-galactosidase control plasmid (pRSV-βgal) (68) were cotransfected into NY72-4-infected CEFs along with RCASB-GFP shRNA, RCASB-TAM67, or RCASB-JunD shRNA as described in Results except for the experiment summarized in Fig. 4B. For this experiment pGLuc-359 and pRSV-βgal) were transfected into NY72-4-infected CEFs along with RCASB-GFP shRNA and RCASB-TAM67 or RCASB-Δ184-C/EBPβ and RCASB-TAM67. NY72-4-infected CEFs were seeded onto six-well plates and transfected by the DEAE-dextran method for 4 to 6 h and shocked for 2 min in a 10% dimethyl sulfoxide (DMSO)–phosphate-buffered saline (PBS) solution. The medium was replaced the following day, and half the cells were transferred to the permissive temperature. After 6 to 8 h at 37.5°C, CEFs were lysed in 70 μl of 250 mM Tris (pH 6.8) and 1% NP-40. Lysate was centrifuged to remove debris. Lysate (30 μl) was used for β-Gal normalization. Lysate was assayed for Gaussia luciferase using a Gaussia luciferase assay kit (New England BioLabs) as per the manufacturer's recommendations. Luminescence was quantified on a Berthold Lumat 9501 luminometer. For the JunD shRNA reporter assays (Fig. 4A), NY72-4-infected CEFs were transfected, fed the following day, and split 1:2 at 2 days following feeding. Half the cells were immediately shifted to the permissive temperature to induce v-Src transformation and harvested the next day. All reporter assays were conducted with a minimum of six replicates. Binding sites for candidate transcription factors were originally identified using Signal Scan, version 4.0 (69), and then filtered using our own transcription factor database.
Cloning.Reporter constructs were generated with the Gaussia luciferase vector system because these plasmids did not contain any cryptic v-Src-responsive regulatory elements (our unpublished results). pGluc-TATA was derived by inserting a DNA sequence consisting of a TATA box and the human β-globin initiator element into the BamHI site of the pGluc-Basic reporter plasmid (New England BioLabs) (68). DAPK1 promoter fragments were cloned into pGluc-TATA using directional cloning. The full-length DAPK1 promoter was PCR amplified from chicken genomic DNA and cloned into pCR2.1 (Invitrogen) using TA cloning and into pGluc-TATA as an EcoR1/BamH1 DNA fragment. The deletion mutants were PCR amplified from the pCR2.1 construct and subcloned into pGluc-TATA. Constructs −134, −83, and −45 had MfeI sites on the 5′ primers instead of EcoR1. MfeI generates cohesive, compatible ends that can be ligated into EcoR1-cut DNA. Binding site mutants were constructed by site-directed mutagenesis. Each oligonucleotide pair was designed to mutagenize at least two critical sites in the binding motif and to introduce a diagnostic restriction site for clone screening. PCR cloning of the full-length and truncated mutants was carried out using GoGreen Taq (Promega). Site-directed mutagenesis was done using Pfu Turbo (Stratagene). All clones were sequenced at the MOBIX sequencing facility at McMaster University. Table 4 describes the oligonucleotides used, and Table 5 gives the mutant transcription factor binding sites as used in transient reporter assays.
Oligonucleotides used in the construction of pGluc-reporter plasmidsa
Wild-type and mutated transcription factor binding sites as used in reporter assay experiments
Accession number(s).Array data are accessible through the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/ ) under accession number GSE69862 .
ACKNOWLEDGMENTS
We thank Colin Nurse for the use of his microscopy facility and Achim Leutz for providing the C/EBPβ antiserum used in chromatin immunoprecipitation assays.
FOOTNOTES
- Received 22 September 2016.
- Accepted 7 October 2016.
- Accepted manuscript posted online 19 October 2016.
Supplemental material for this article may be found at https://doi.org/10.1128/JVI.01925-16 .
- Copyright © 2016 American Society for Microbiology.
