In order to investigate the physiological outcomes of aberrant expression of Snail or of Slug in human breast cancer cells, we developed a recombinant adenovirus system permitting exogenous expression of these factors (15
). This system was used to assess the effects of aberrant Snail or Slug expression on epithelial architecture in MCF7 breast carcinoma cells that express endogenous Snail and Slug at very low levels. During initial characterization of this system, limiting amounts of Snail or Slug adenovirus were utilized, permitting simultaneous observation of infected and uninfected cells in single microscopic fields. Occludin, a transmembrane component of the tight junction, stained in the “chicken wire” pattern typical of epithelial cells (Fig. ) in the absence of infection or after infection with mock virus. In contrast, in cells infected with either Snail or Slug adenovirus, occludin staining was lost, and the cells were unable to form productive contacts with their neighbors (Fig. ). Immunoblot analysis of two well-characterized transcriptional targets of Snail (Fig. ) revealed a dramatic reduction of occludin (18
), a finding consistent with the immunocytochemistry, and a modest reduction of E-cadherin (3
). Slug expression resulted in qualitatively similar effects on E-cadherin and occludin albeit with slightly altered kinetics, probably relating to differences in the expression levels of Snail and Slug (Fig. ). The levels of repression of E-cadherin evident in this system differ somewhat from Snail/Slug-dependent repression of this molecule in MDCK cells (4
) but are consistent with reports that use breast carcinoma cells (13
FIG. 1. Expression of Snail or Slug leads to phenotypic alterations in epithelial cells. (a to i) MCF7 cells were infected with mock (a, b, and c), Snail (d, e, and f), or Slug (g, h, and i) adenovirus. The cellular distribution of green fluorescent protein (a, (more ...)
In addition to effects on junctional complex components, Snail and Slug also alter other features of epithelial cells, resulting in increases in migratory and invasive growth (16
). In Matrigel invasion assays, Snail- and Slug-expressing cells were approximately three times more efficient at invasion than mock-infected or uninfected cells (Fig. , top panel). Thus, many of the phenotypic alterations attributed to aberrant Snail or Slug expression are faithfully recapitulated in cells infected with the recombinant adenoviruses. In considering the significance of this threefold increase in invasion to cancer progression, we note that a recent report documented the capacity of fibroblast growth factor 8b overexpression to alter growth characteristics of MCF7. Overexpression of this molecule resulted in a 2.5-fold increase in an in vitro invasion assay and a five- to sixfold increase in tumor volume in a xenograft model (37
FIG. 2. Expression of Snail or Slug leads to increased invasion and protection from programmed cell death. (a) Snail, Slug or mock adenovirus-infected MCF7 cells were seeded on Matrigel invasion chambers. Noninfected cells were included as a control. Invading (more ...)
In epithelial cells, junctional complexes serve as platforms for cellular signaling (24
). Loss of the junctional complex after transcriptional repression of E-cadherin and occludin could result in the disruption of prosurvival signaling. To investigate the response of cells expressing Snail or Slug to genotoxic stress, the topoisomerase inhibitor ADR was used to induce DNA damage. After drug exposure, the relative apoptotic activity of parental MCF7 cells increased substantially (Fig. ). Expression of either Snail or Slug totally prevented ADR-induced cell death in MCF7, suggesting that aberrant expression of these transcription factors in epithelial cells induces alterations in both cell morphology and in the apoptotic response. In the context of cancer progression, alterations in the in vitro response to apoptotic stimuli can be correlated to tumor growth in animal models. In MDA435 breast carcinoma cells, overexpression of the antiapoptotic molecule Bcl-xL
resulted in a twofold decrease in apoptosis after transforming growth factor β treatment. In xenograft experiments, the capacity to form metastatic tumors was enhanced >5-fold (10
Given the function of Snail and of Slug as transcriptional repressors, their ability to interfere with the apoptotic program is predicted to result from alterations in gene expression. Since the tumor suppressor p53 plays an integral role in cellular responses to DNA damage, the role of Snail and Slug in p53 regulation was investigated. Expression of either Snail or Slug resulted in modest downregulation of steady-state levels of p53 mRNA (Fig. ), in keeping with the observation that the TP53 locus contains numerous potential binding sites for Snail and Slug. In addition to p53, mRNA levels for additional proapoptotic factors whose genomic loci contain potential binding sites for Snail and Slug were analyzed. Steady state mRNA levels of the apoptotic nuclease DFF40 (26
), as well as Mst4, a kinase implicated in the apoptotic response (8
), declined significantly in response to expression of Snail or of Slug (Fig. ). Occludin and E-cadherin, included as positive controls, demonstrated the expected Snail- or Slug-dependent decreases in mRNA levels (Fig. ). Expression of Snail also resulted in a modest decline in the steady-state level of p53 at the protein level (Fig. ). Similar results were obtained after expression of Slug or coexpression of Snail and Slug (Fig. ).
FIG. 3. Expression of Snail or Slug alters the apoptotic response through transcriptional repression of multiple targets. (a) Total RNA was extracted from MCF7 that were infected with Snail, Slug, or mock adenovirus and treated with ADR or vehicle. Steady-state (more ...)
Regulation of the cellular p53 levels is known to result from a complicated set of regulatory steps, including posttranscriptional regulation. After DNA damage, the p53 protein is stabilized (36
); therefore, the effect of Snail on p53 accumulation after ADR treatment was investigated. Mock-infected cells demonstrated the expected time-dependent accumulation of p53 (Fig. ). In contrast, Snail-expressing cells had reduced levels of p53 prior to treatment and had impaired accumulation of the protein following ADR treatment (Fig. ). Therefore, factors affecting p53 stability were examined (Fig. ). MDM2, an E3 ubiquitin ligase involved in p53 turnover (36
), was unaffected by Snail expression in the absence of drug. However, after ADR treatment, Snail-expressing cells exhibited a statistically significant increase in MDM2
transcript levels. Slug expression resulted in increased steady-state levels of MDM2 transcript regardless of drug exposure (Fig. ).
Factors implicated in the stabilization of p53 showed a mixed response. ARF, the alternative reading frame product of the p16INK4a
gene, stabilizes p53 independent of the DNA-damage response through inhibition of MDM2-dependent p53 turnover (36
mRNA levels were unaffected by ADR treatment or by expression of either Snail or Slug (Fig. ). The checkpoint kinase CHK2 has been implicated in p53 stabilization through phosphorylation of the protein after DNA damage in mouse cells (17
), although the relevance of this observation in human cells has recently been questioned (1
mRNA levels were not altered either by Snail expression, by Slug expression, or by ADR treatment (Fig. ). ATM
, the gene product mutated in ataxia telangiectasia, is induced after DNA damage and is implicated in p53 modification and stabilization (2
). In the absence of ADR treatment, expression of either Snail or Slug results in significant repression of mRNA coding for the ATM kinase (Fig. ). Thus, the observed reduction in p53 protein levels after expression of Snail or Slug likely resulted from the combinatorial effects of three distinct mechanisms: a decline in p53 mRNA levels, increased MDM2-mediated turnover, and the loss of stabilizing modifications from the ATM kinase. The failure of Snail or Slug to alter the abundance of CHK2 and ARF suggests that the effects of aberrant expression of these transcriptional repressors on the p53 response are mediated through specific regulatory molecules. Since both CHK2 and ARF contain multiple E-box sequences (data not shown) and yet fail to respond to Snail or Slug, not all potential target loci respond to aberrant expression of these factors.
As levels of p53 protein were clearly affected by expression of Slug and Snail, we also analyzed transcript abundance of genes activated by p53 in response to genotoxic stress. Transcripts from a subset of these genes, including p53AIP1
), and PUMA
), were not detectable in MCF7 under any circumstances (data not shown). Others, including p53DINP1
), and BAX
), were unaffected by Snail (Fig. and data not shown) or Slug (Fig. ). However, transcripts of the p53 targets BID
), and Caspase 6
) were significantly repressed by ectopic expression of Snail or Slug (Fig. ). Thus, in addition to affects on p53 itself, aberrant Snail or Slug expression led to transcriptional effects on other genes integral to the DNA damage-induced programmed cell death pathway.
The repression at the CDH1, OCLN, TP53, BID, PIG8, and DFF40 loci could result from either direct effects or indirect effects of Snail and Slug. ChIP assays were performed to determine whether aberrantly expressed Snail and Slug bound to these loci. E-cadherin and occludin served as positive controls. After Snail or Slug expression, ChIP was performed, and coprecipitated DNA was analyzed by PCR. At all responsive loci, E-box sequences (Snail or Slug binding sites) were precipitated, whereas control sequences were not (Fig. ). E-box DNA was amplified only after expression of the relevant repressor. Some loci containing E-boxes, such as APAF-1, are unresponsive, and the binding of Snail or Slug is not detected by ChIP. Thus, Snail and Slug can bind to the promoters of the TP53, BID, DFF40, OCLN, and CDH1 genes, the transcripts of which are all repressed after overexpression of Snail or Slug.
FIG. 4. Snail and Slug localization at responsive genes. ChIP analyses of MCF7 cells infected with Snail, Slug, or mock adenovirus were performed. P1 to P9 indicate PCR primer sets that amplify genomic DNA from the indicated loci. Primer sets P2, P4, P6, P7, (more ...)
In order to confirm these results in a more physiologic context, Snail protein was depleted in A375, a human melanoma cell line that expresses Snail at moderate to high levels and is wild type at the TP53 locus, by using adenovirus-based short hairpin RNA interference (shRNA). Infection with the shRNA-Snail adenovirus resulted in depletion of endogenous Snail mRNA with no apparent effect on Slug mRNA (Fig. , left panel). At the protein level, a decline in Snail protein was accompanied by moderate upregulation of p53 protein, as expected (Fig. , right panel). RT-PCR was performed for selected transcripts after Snail depletion (Fig. ). As expected, the endogenous Snail transcript declined. Statistically significant upregulation of p53, BID, PIG8, Caspase 6, DFF40, MST4, and OCLN was observed (Fig. ). No significant difference was observed in MDM2, ARF, and ATM (Fig. ). Thus, endogenous Snail represses steady-state levels of TP53, BID, PIG8, Caspase 6, DFF40, MST4, and OCLN transcripts.
FIG. 5. Depletion of endogenous Snail by using RNAi leads to increased p53 levels and increased sensitivity to apoptotic stimuli. (a) For the left panel, A375 cells were infected with either mock adenovirus or with shRNA-Snail adenovirus; control cells were uninfected. (more ...)
To determine whether endogenous Snail regulates DNA damage-induced programmed cell death, A375 cells were infected with shRNA-Snail or mock adenoviruses and treated with ADR. The presence of apoptotic cells (red fluorescence) was examined by an in vivo TUNEL assay in infected cells (green fluorescence) (Fig. ). Quantification of these data revealed a sevenfold increase in the number of apoptotic cells following adenoviral-mediated expression of the Snail-specific shRNA (Fig. ). These data suggest that Snail represses proapoptotic genes in the DNA damage response pathway, providing a prosurvival function.
FIG. 6. Snail knockdown results in an increased sensitivity to programmed cell death induced by ADR treatment. (a) A375 cells infected with shRNA-Snail or mock adenovirus were treated with ADR, and TUNEL-positive cells were detected by using immunofluorescence. (more ...)