|Home | About | Journals | Submit | Contact Us | Français|
The increased motility and invasiveness of tumor cells are reminiscent of epithelial-mesenchymal transition (EMT) that occurs during embryonic development, wound healing, and metastasis. In this study, we found that Snail is stabilized by the inflammatory cytokine TNFα through the activation of the NF-κB pathway. We demonstrated that NF-κB is required for the induction of COP9 signalosome 2 (CSN2) which, in turn, blocks the ubiquitination and degradation of Snail. Furthermore, we showed that the expression of Snail correlated with the activation of NF-κB in cancer cell lines and metastatic tumor samples. Knockdown of Snail expression inhibited cell migration and invasion induced by inflammatory cytokines and suppressed inflammation-mediated breast cancer metastasis. Our study provides a plausible mechanism for inflammation-induced metastasis.
It has become increasingly clear that chronic inflammation is tightly correlated with tumorigenesis as cancer has been viewed as “a wound that never heals” (Coussens and Werb, 2002; Jackson and Evers, 2006; Karin and Greten, 2005). Mounting evidence from various clinical and experimental studies has demonstrated that the inflammatory tumor microenvironment contributes a pivotal role not only to tumor development but also to metastasis. The inflammatory tumor microenvironment evolves as tumors grow; this includes the infiltration of immune cells and the activation of the inflammatory response. Inflammatory cells, particularly tumor-associated macrophages (TAMs), are usually found at the invasive front (the tumor–host interface) of more advanced tumors (Condeelis and Pollard, 2006). They facilitate angiogenesis, extracellular matrix breakdown, and tissue remodeling, and thus promote tumor cell motility. These TAMs also secrete pro-inflammatory cytokines, such as tumor necrosis factor-α (TNFα), to activate the major inflammatory response NF-κB pathway, which facilitates both tumor development and metastatic progression (Karin, 2006; Karin and Greten, 2005).
The increased motility and invasiveness of tumor cells are reminiscent of the events at EMT which is characteristic of embryonic development, tissue remodeling, and wound healing (Nieto, 2002; Peinado et al., 2007; Thiery and Sleeman, 2006). In the EMT process, epithelial cells acquire fibroblast-like properties and show reduced intercellular adhesion and increased motility (Thiery and Sleeman, 2006). A hallmark of EMT is the loss of E-cadherin expression. Loss of E-cadherin expression is often correlated with the tumor grade and stage (Cowin et al., 2005). Under physiological conditions, EMT takes place at the edge of injury during wound healing (Neilson, 2006). Similarly, this process occurs at the invasive front of many metastatic cancers (Christofori, 2006; Franci et al., 2006). These observations suggest that the migratory and invasive ability of tumor cells at the invasive front are initiated and propelled by an inflammatory microenvironment through the induction of EMT. Several transcription factors have been implicated in the control of EMT, including Snail/Slug, Twist, Goosecoid, δEF1/ZEB1, SIP1, and E12/E47 (Hartwell et al., 2006; Nieto, 2002; Yang et al., 2004). Snail was identified in Drosophila as a suppressor of the transcription of shotgun (an E-cadherin homologue) in controlling the formation of the mesoderm and neural crest (Nieto, 2002; Thiery and Sleeman, 2006). Absence of Snail is lethal because of severe defects at the gastrula stage during development (Carver et al., 2001). Expression of Snail represses expression of E-cadherin and induces EMT in MDCK and breast cancer cells, indicating that Snail plays a fundamental role in EMT and breast cancer metastasis (Batlle et al., 2000; Cano et al., 2000; Zhou et al., 2004). In addition, overexpression of Snail correlates with tumor grade, nodal metastasis, and tumor recurrence and predicts a poor outcome in patients with various cancers (Blanco et al., 2002; Cheng et al., 2001; Moody et al., 2005; Zhou et al., 2004). Furthermore, expression of stromal matrix metalloproteinase (MMP3) stimulates expression of Snail through the increased cellular reactive oxygen species (Radisky et al., 2005). This interesting discovery highlights the importance of the microenvironment in regulating Snail and in the initiation of EMT at metastasis. We recently showed that Snail is highly unstable with a short half-life and is regulated by protein stability and cellular location though GSK-3β mediated phosphorylation (Zhou et al., 2004). However, it is unclear what extrinsic signals regulated the activity of Snail and the induction of EMT at the tumor invasive front. Here, we examined the regulation of Snail and its role in cancer cell migration, invasion, and metastasis mediated by inflammation.
Since EMT occurs at the edges of the wound during healing and at the invasive front of metastatic cancers (Christofori, 2006; Franci et al., 2006; Neilson, 2006), both processes are influenced by stimuli that emanate from the inflammatory microenvironment, and because Snail is a major transcription factor for EMT induction (Nieto, 2002; Thiery and Sleeman, 2006; Zhou and Hung, 2005), we reasoned that tumor cell migration and invasion at the tumor-host boundary are induced by inflammation through Snail-mediated EMT induction. To test this hypothesis, we first examined the invasive ability of 18 cancer cell lines from breast, prostate, and colon cancers in response to macrophage conditioned medium. We found that the invasive ability for the majority of these cancer cell lines (15 out of 18 cell lines) was dramatically increased compared with cells cultured in regular medium (Figure 1A, Figure S1 and data not shown). Similar findings were also obtained when these cancer cells were directly co-cultured with macrophages (Figure S14 and data not shown). Interestingly, stable expression of Snail in two non-metastatic breast cancer cell lines, MCF7 and T47D (both contain little endogenous Snail) greatly increased the invasive ability of these cells (Figure 1A), indicating that Snail is a critical molecule for mediating inflammatory cytokine-induced invasion. Furthermore, when MCF7 and HEK293 GFP-Snail stable transfectants were cultured with macrophage conditioned medium, cell detachment and acquisition of an EMT-like morphology was noted (Figure 1B). Surprisingly, the intensity of the GFP-Snail in these cells was dramatically elevated (Figure 1B). We previously demonstrated that the half-life of Snail is about 25 min and is mainly regulated through β-Trcp-mediated proteasome degradation (Zhou et al., 2004). The enhanced intensity of GFP-Snail, after culture with macrophage conditioned medium, suggested that unknown cytokines from the conditioned medium induced protein stabilization of Snail. Of the many inflammatory cytokines that macrophages secrete, we tested the major cytokines, TNFα, IL-1β, IL-2, IL-6, and IFNγ, for their ability to stabilize Snail. When Snail/HEK293 cells were treated with different cytokines for 6 hr, we found that TNFα induced the stabilization of Snail to a similar degree as cells treated with proteasome inhibitor MG-132 (lane 2 vs lane 7, Figure 1C). Similar results were also found in Snail/MCF7, Snail/T47D, and Snail/ZR75 cells (Figure S2A). The cytokine-induced Snail stabilization appears to correlate with NF-κB activation (Figures S2B–S2D). IL-1β also induced Snail stabilization in Snail/MCF7 cells but this effect depended on cell types and IL-1β concentration (Figure S2A and S2B). We also combined the ineffective cytokines with TNFα and found no synergistic effect between TNFα and the other cytokines (Figure 1C), indicating that the effect on Snail stabilization by macrophage conditioned medium is mainly mediated by TNFα. To further confirm this observation, we treated four different Snail-expressing stable cell lines with TNFα and found that TNFα induced the stabilization of Snail to a similar level as the proteasome inhibitor MG-132 (Figure 1D). Moreover, we treated several cancer cell lines with TNFα and found that TNFα also induced the stabilization of endogenous Snail in these cells (Figure 1E). Interestingly, both endogenous and exogenous Snail induced by TNFα migrated faster on an SDS-PAGE gel compared with Snail induced by MG-132 (Figures 1D & 1E), suggesting that the mechanism of Snail stabilization appeared to be different with these two agents. To determine whether the different migrating bands represent different phosphorylated states of Snail, we immunoprecipitated Snail and treated it with protein phosphatase (Figure 1F). Similar to our previous report, Snail stabilized by MG-132 was in a phosphorylated form. However, Snail stabilized by TNFα remained in a non-phosphorylated form. We next measured the time course for Snail stabilization and found that the level of Snail was increased after 1 hr of TNFα stimulation and reached a maximum at about 4 hr (Figure 1G). However, the mRNA level of Snail did not demonstrate a significant elevation within 6 hr of TNFα treatment in MDA-MB231 and Snail/HEK293 cells (Figure 1H and data not shown). Thus, the initial and rapid induction of Snail suggested that the elevation of Snail by TNFα was mainly due to protein stabilization. Furthermore, the TNFα -stabilized Snail is functionally important for cell migration and invasion, because the stabilization of Snail in Snail/MCF7 cells by TNFα enhanced migration and invasion of these cells (Figure 1A and Figures S1 and S3). Collectively, these results clearly indicate that the inflammatory cytokine TNFα induces the stabilization of Snail in a non-phosphorylated, functional form and thus enhances cell migration and invasion.
Previously, we demonstrated that GSK-3β is a major kinase that phosphorylates Snail and induces the protein degradation of Snail (Yook et al., 2005; Zhou et al., 2004). To examine whether the stabilization of Snail by TNFα is mediated by regulation of GSK-3β activity, we treated Snail/HEK293 cells with the PI3K inhibitor LY294002 to activate GSK-3β prior to TNFα treatment (loss of GSK-3β phosphorylation at Serine 9 enhances GSK-3β activity). We found that TNFα induced stabilization of Snail regardless of the phosphorylated status of GSK-3β (Figure 2A), suggesting that TNFα -mediated Snail stabilization is independent of GSK-3β phosphorylation. To further investigate our findings, we knocked down the expression of GSK-3β in Snail/HEK293 cells using specific GSK-3β siRNA (Figure 2B). Downregulation of GSK-3β partially elevated the level of Snail (lane 2 vs lane 1, Figure 2B). However, this downregulation did not affect TNFα -mediated Snail stabilization (lane 3 vs lane 4, Figure 2B). Similarly, when Snail was expressed in GSK-3β−/− MEFs, it also became partially stabilized (lane 5 vs lane 2; Figure 2C), which is consistent with our previous finding that GSK-3β is the major kinase for mediating Snail degradation. Interestingly, TNFα-mediated Snail stabilization was further elevated in the absence of GSK-3β (lane 6 vs lanes 5 and 3, Figure 2C), indicating that TNFα and GSK-3β are two key factors for stabilization of Snail.
To determine which signaling pathway was involved in TNFα -mediated Snail stabilization, we next treated the cells with inhibitors of the MAPK/ERK, mTOR, p38, and JNK kinases since TNFα can induce the activation of these pathways. Although the stabilization of Snail by TNFα was not affected by these inhibitors, we found that an NF-κB inhibitor (Sanguinarine) completely blocked TNFα-stabilized Snail (Figure 3A). Similar results were also noted when these cells were treated with another commonly used NF-κB inhibitor BAY 11-7082 (Figure S4), suggesting that the activation of the NF-κB pathway is important for TNFα -mediated Snail stabilization. To further investigate this finding, we expressed wild-type (WT) or kinase-dead (KD) IKKα or IKKβ, p65, or IκB mutant in Snail/HEK293 cells followed by treatment with TNFα. Kinase-dead IKKβ partially suppressed the stabilization of Snail mediated by TNFα (lane 10, Figure 3B). Expression of the super-suppressor of IκB (Karin, 2006) completely inhibited the TNFα -stabilized Snail (lane 14, Figure 3B), while expression of p65 alone in these cells induced Snail stabilization (lane 11, Figure 3B). We also co-expressed p65 with GFP-Snail in HEK293 cells and found that the expression of p65 dramatically enhanced Snail protein levels as the expression of p65 was co-localized with GFP-Snail in the nucleus (Figure 3C). Collectively, these results indicate that the activation of the NF-κB pathway was critical for the stabilization of Snail. To confirm a causal relationship between p65 and Snail, expression of p65 was knocked down by siRNA in Snail/HEK293 cells. Knockdown of p65 expression suppressed the stabilization of Snail mediated by TNFα (Figure 3D). Similarly, knockdown of p65 expression in PC3 and HCT116 cells also blocked TNFα -induced endogenous Snail stabilization (Figure 3D). To rule out off-target effects caused by the siRNA, relevant data were also confirmed with a second siRNA that depletes p65 with a similar efficiency (Figure S5). In addition, when Snail was expressed in p65−/− MEF cells, we did not find the stabilization of Snail by TNFα (Figure 3E). However, when Snail was co-expressed with p65 in these cells, the stabilization of Snail by TNFα was restored (Figure 3E), confirming that p65 is required for the stabilization of Snail. Furthermore, when Snail was co-expressed with the E-cadherin promoter-driven luciferase construct, a mild suppression of E-cadherin luciferase was noted. Treatment with TNFα or co-expression with p65 greatly enhanced Snail-mediated E-cadherin promoter suppression, indicating that the enhanced suppression of the E-cadherin promoter was due to the stabilization of Snail mediated by TNFα or the co-expression of p65 (Figure 3F).
The p65 protein contains an N-terminal Rel-homolog domain (RHD) that is conserved among all five Rel family members and is required to associate with IκB and to form homo- or hetero-dimers among Rel family transcription factors (Chen and Greene, 2004; Gilmore, 2006). The C-terminus of p65 contains a DNA-binding region and is required for its transcription activity (Figure 4A) (Chen and Greene, 2004; Gilmore, 2006). To analyze the molecular mechanism for Snail stabilization by p65, we co-expressed wild-type (WT) or different deletion mutants of p65 with Snail in HEK293 cells. Interestingly, when the C-terminal region of p65 was deleted, the ability of p65 to induce the protein stability of Snail was completely abolished (Figure 4A), indicating that the transcription activity of p65 was critical for the stabilization of Snail. Consistent with this notion, we found that the Snail stabilization by TNFα was completely blocked when cells were pretreated with a transcription inhibitor actinomycin D (Act D; Figure 4B). As we showed that TNFα -stabilized Snail is mediated at the post-translational level (Figure 1G & 1H), the transcriptional dependence of p65 suggests that a mediator is involved in the stabilization of Snail and this potential mediator is induced by TNFα or p65 at the early time point. To further define the underlying mechanism, we washed away of TNFα from culture medium after 6 hr of cell incubation with TNFα and chased the degradation of Snail (Figures 4C & 4D). We found a slow downregulation of Snail (with a half-life more than 4 hr) after removal of TNFα, suggesting that TNFα treatment has initiated a program for the stabilization of Snail. However, this slow downregulation of Snail was suppressed by Act D even in the presence of TNFα, indicating that a mediator, which is transcriptionally induced by TNFα but is blocked by Act D, was required for the stabilization of Snail. Compatible with this idea, the accelerated downregulation of Snail by Act D was completely blocked by a proteosome inhibitor MG-132 (Figures 4C & 4D). In addition, we incubated cells with TNFα for 6 hr to induce a potential mediator and then treated the cells with protein translational inhibitor cycloheximide (Figure S6). We found that the rate of Snail degradation was slower than that from cells without TNFα pretreatment (Figure S6). Together, these results strongly suggest that a mediator was transcriptionally/translationally induced by TNFα or p65 to block the proteasome degradation of Snail.
Approximately 140 genes are regulated by p65, some of them are early response genes that are induced within 1 to 4 hr (Zhou et al., 2003). Because the protein stabilization of Snail by TNFα begins at 1 hr and reaches a maximum at 4 hr (Figure 1G), this suggests that the induction of genes that regulate Snail should occur within 1 hr of TNFα stimulation. There are about 50 genes that are induced within 1 hr of TNFα stimulation, such as IκBα (Zhou et al., 2003). We screened the majority of these genes for their ability to induce the stabilization of Snail (data not shown) and found that the expression of COP9 signalosome 2 (CSN2) was of particular interest. CSN2 is the second and most conserved subunit of the COP9 signalosome in all eukaryotes (Cope and Deshaies, 2003; Richardson and Zundel, 2005; Wolf et al., 2003). The eight CSN subunits share significant sequence homologies with the eight subunits of the 26S proteasome lid and are thought to cooperate with the ubiquitin/proteasome system in the regulation of protein stability by interacting with cullin-based ubiquitin ligases to control their activity. To test whether CSN2 is involved in the regulation of Snail mediated by TNFα or p65, we first examined the expression of CSN2 in cells by semi-quantitative PCR after TNFα stimulation. Consistent with a previous report (Zhou et al., 2003), the induction of CSN2 expression occurred within 1 hr of TNFα treatment and reached a maximum at 4 hr in several cell lines, such as HEK293, SKBR3, and MDA-MB231 cells (Figure 5A and Figure S7). The induction of IκB by TNFα, as a positive control, also began within 1 hr of stimulation. Similar results were also obtained when we performed Q-PCR analysis (data not shown). Consistent with the change in mRNA, TNFα also induced endogenous CSN2 expression in a time-dependent manner in SKBR3 and MDA-MB231 cells (Figure 5A and Figure S7). In addition, we noticed that the promoter region of CSN2 contains 4 conserved NF-κB binding sites. Expression of p65 induced the luciferase activity driven by the CSN2 promoter (Figure 5B). Deletion of the C-terminus of p65 abolished its ability to induce CSN2 (Figure 5B). Moreover, when CSN2 was co-expressed with Snail in HEK293 cells, it dramatically stabilized Snail protein to a similar degree as that by TNFα or MG-132 (Figure 5C). Surprisingly, the stabilized Snail mediated by CSN2 was in a non-phosphorylated form that is similar to TNFα but is different from MG-132 (Figure 5C). Similarly, when CSN2 was co-expressed with GFP-Snail, it enhanced the intensity of GFP-Snail in the nucleus (Figure 5D). To further determine whether CSN2 is required for TNFα -mediated Snail stabilization, we treated cells with two known CSN inhibitors, curcumin or emodin (Berse et al., 2004) and found that the stabilization of Snail by TNFα was suppressed (Figure 5E). In addition, when CSN2 expression was knocked down by siRNA, the TNFα-stabilized Snail was greatly suppressed in Snail/HEK293 cells (Figure 5F). Knockdown of CSN2 expression also significantly inhibited the stabilization of endogenous Snail induced by TNFα in PC3 and SKBR3 cells (Figure 5F). This finding was not caused by the off-target effects of siRNA as similar data were also obtained with a second siRNA that targets CSN2 with a similar efficiency (Figure S8). Furthermore, expression of exogenous CSN2 enhanced the steady-state of Snail whereas knockdown of endogenous CSN2 expression facilitated the degradation of Snail (Figure S9). Together, these results indicate that CSN2 is the critical mediator and is required for TNFα -mediated Snail stabilization.
Cullin is an essential scaffolding component of ubiquitin E3 ligases, and it is modified covalently by neddylation, which blocks the association of Cullin with a negative regulator CAND1 (Goldenberg et al., 2004; Richardson and Zundel, 2005). The activity of CSN is to remove the neddylation from cullins and to facilitate the association of cullin with CAND1 and, therefore, inhibit the ubiquitin E3 ligase activity of SCF complex. CSN2 is the most conservative subunits of CSN and it facilitates the functional assembly of CSN complex (Naumann et al., 1999; Schweitzer et al., 2007). The intrinsic de-neddylation function of CSN complex is executed by CSN5 that contains MPN metalloenzyme activity. Similar to CSN2, expression of CSN5 also induced the stabilization of Snail, indicating that the function of CSN complex is critical for Snail stabilization (Figure S10). Consistent with this finding, knockdown of CSN5 inhibited CSN2-mediated Snail stabilization (Figure S11). Interestingly, knockdown of CAND1, a negative regulator of cullins, also suppressed CSN2-mediated Snail stabilization (Figure S11). We previously demonstrated that the protein stability of Snail is regulated by the E3 ligase β-Trcp and the proteasome pathway; treatment with the proteasome inhibitor MG-132 stabilized Snail in a highly phosphorylated and ubiquitylated form (Zhou et al., 2004). Because stabilized Snail by TNFα or CSN2 occurs at a non-phosphorylated form (Figure 1F and Figure 5C) and because CSN2 affects ubiquitin E3 ligase, we analyzed whether the protein stabilization is mediated by suppression of Snail ubiquitylation. When ubiquitin was immunoprecipitated from the cell lysate, we found significantly less Snail ubiquitylation in cells treated with TNFα than in cells treated with MG-132 (Figure 6A). Similarly, when Snail was immunoprecipitated, we also found that the ubiquitylation of Snail was dramatically suppressed in cells treated with TNFα or expressing CSN2 compared with cells treated with MG-132 although total stabilized Snail proteins were similar (Figure 6B). Moreover, we found that the association of Snail with its E3 ligase, β-Trcp, was abolished in cells treated with TNFα or expression with CSN2 (Figure 6B). Because TNFα - and CSN2-stabilized Snail are unphosphorylated, we also examined the association of Snail with GSK-3β since GSK-3β is the main kinase for mediating Snail phosphorylation and degradation. Surprisingly, we found that the association of Snail with GSK-3β was diminished in cells treated with TNFα or CSN2 compared with MG-132 treatment (Figure 6C). Consistent with these findings, when endogenous GSK-3β was immunoprecipitated, the associated endogenous Snail was significantly reduced in cells treated with TNFα in comparison with that of MG-132 (Figure 6D). Similarly, when endogenous Snail was pulled down, the bound endogenous GSK-3β was lost in cells treated with TNFα (Figure 6E). Together, these results clearly demonstrated that the stabilization of Snail by TNFα or p65 was mediated by the transcriptional induction of CSN2 which, in turn, inhibited the association of Snail with β-Trcp and GSK-3β and thus suppressed its ubiquitylation and phosphorylation.
Having established that TNFα -stabilized Snail is mediated by NF-κB induced CSN2, we next examined whether the stabilization of Snail was required for cancer cell migration, invasion, and metastasis mediated by inflammation. We first established stable shRNA Snail knockdown cells from two highly metastatic cancer cell lines, MDA-MB231 and MDA-MB435. We achieved about 90% knockdown of endogenous Snail expression (Figure 7A). Although we did not detect the gain of expression of E-cadherin (data not shown), the expression of N-cadherin was dramatically decreased while the expression of two tight junction molecules, ZO-1 and claudin-1, was increased (Figure 7A). Consistent with these molecular changes, cell migration of both Snail knockdown cell lines was inhibited in a wound healing assay (Figure 7B and Figure S12). Importantly, cell migration induced by TNFα was also significantly suppressed (Figure 7B and Figure S12). To rule out off-target effects caused by the shRNA, relevant data were also confirmed with a second shRNA that depletes Snail with a similar efficiency in MDA-MB231 cells (Figure S13). On the invasion assays, although the invasiveness of MDA-MB231 and MDA-MB435 cells were dramatically increased when they were co-cultured with macrophage conditioned medium or macrophages (Figure 7C and Figure S14), the addition of TNFα antibody, but not the control IgG suppressed cell invasion induced by macrophages or macrophage conditioned medium (Figure 7C and Figure S14). These results further confirm our finding that TNFα is the major cytokine for inducing Snail stabilization and cell invasion (Figure 1). Most importantly, knockdown of Snail expression in these two cancer cell lines not only inhibited the invasiveness that is intrinsic to the metastatic cancers cells but also suppressed inflammation-enhanced invasion (Figure 7C and Figure S14).
We also tested our findings in a xenograft metastasis model in which a cancer cell line MDA-MB435 was used to generate pulmonary metastases, administration of lipopolysaccharide (LPS), an inducer of inflammation (Luo et al., 2004), greatly accelerated lung metastases in mice (Figure 8A, Figure S15, and S16). Knockdown of Snail expression suppressed both the intrinsic and inflammation-enhanced metastasis in these mice. Similar results were also obtained when we used another metastatic breast cancer cell line, MDA-MB231, in the xenograft metastasis assay (Figure 8A). Together, our results demonstrated that Snail is required for cell migration, invasion, and metastasis mediated by inflammation.
To further extend our findings in vivo, we determined whether there is a correlation between the expression of Snail and the activity of p65 in cancer cell lines and resected human breast cancer specimens. Because p65 is active in the nucleus and Snail is localized in the nucleus when it is stabilized (Zhou et al., 2004), we measured the expression of Snail and p65 in nuclear extracts from 14 different cancer cell lines. We found that the nuclear level of p65 was highly correlated with the protein level of Snail (Figure 8B). Likewise, in 110 cases of breast cancer specimens, the expression of p65 was highly correlated with the expression of Snail (Figure 8C and Supplementary Table 1). Thus, these results in human breast cancer tissues confirmed our observations in cell culture and in an animal model, lending further support to our hypothesis that the stabilization of Snail by p65 is required for the cell migration and invasion induced by inflammatory cytokines.
In this study, we showed that TNFα -induced Snail stabilization plays a critical role in inflammation induced EMT and cancer cell migration, invasion, and metastasis. Our study provides several insights into the regulation of EMT and metastasis. First, our study indicates that Snail-induced EMT is critical for inflammation-initiated invasion and metastasis. In the tumor-host boundary, the persistent recruitment of immune cells, such as macrophage and mast cells, is thought to establish an inflammatory tumor microenvironment. The remarkable interaction of macrophages with tumor cells has been visualized in vivo by multiphoton imaging and is found to enhance tumor cell dissemination and invasion at the invasive front (Condeelis and Pollard, 2006; Wyckoff et al., 2007). In addition, macrophage ablation in experimental tumor models greatly inhibited the occurrence of tumor metastasis (Lin et al., 2001). Although macrophages have been shown to enhance angiogenesis and MMP production and thus facilitate tumor metastasis (Lin et al., 2006; Luo et al., 2006), the underlying molecular mechanism for macrophage induced dissemination and invasion of tumor cells at the invasive front remains unclear. In our study, we found that co-culture of tumor cells with macrophages greatly enhanced the migration and invasion of tumor cells by inducing the EMT program through NF-κB mediated Snail stabilization (Figure 1). Knockdown of Snail expression not only inhibits TNFα-induced cancer cell migration and invasion in vitro but also suppresses LPS-mediated metastasis in vivo (Figure 7 and Figure 8). Recently, Karin and colleagues elegantly demonstrated that injection of LPS induced an inflammatory condition and enhanced NF-κB mediated tumor growth and metastasis of colon cancer in an experimental metastasis model (Luo et al., 2004). They further demonstrated that blocking the NF-κB pathway converts inflammation-induced tumor growth and metastasis to TRAIL-mediated tumor regression. This is an important finding because inflammation promotes tumor growth and metastasis as well as kills cancer cells. Consistent with this observation, we found that the introduction of LPS in mice greatly enhanced lung metastasis of breast cancer cells (Figure 8A). Knockdown of Snail not only blocked metastasis that is intrinsic to the metastatic breast cancer cells but also greatly suppressed inflammation-accelerated metastasis (Figure 8A). Given the fact that Snail possesses an anti-apoptotic function in addition to the induction of EMT (Kajita et al., 2004; Vega et al., 2004), future studies will explore whether knockdown of Snail induces apoptosis mediated by inflammation and thus contributes to the metastatic suppression. Similar to our finding, Olmeda et al recently showed that silencing of Snail expression by shRNA inhibited tumor growth and invasiveness (Olmeda et al., 2007). It will be interesting to determine whether the silencing of Snail expression in their system also suppresses inflammation mediated cell invasion and metastasis. Collectively, our study indicates that Snail stabilization and EMT induction, mediated by the inflammatory cytokine TNFα, is critical for metastasis and thus provides a plausible molecular mechanism for tumor cell dissemination and invasion at the invasive front.
Second, our study reveals a mechanism of Snail regulation by NF-κB in the process of inflammation-induced EMT and cancer metastasis. We and others previously showed that Snail is mainly regulated by GSK-3β -mediated phosphorylation and degradation (Yook et al., 2005; Zhou et al., 2004). In current study, we showed that TNFα-mediated Snail stabilization is through the induction of CSN2, which blocked the phosphorylation and ubiquitylation of Snail by disrupting its binding to GSK-3β and β-Trcp (Figure 5 and Figure 6), and thus results in Snail stabilization in a non-phosphorylated and non-ubiquitylated state (Figure 8D). These findings are in agreement with the notion that Snail is a labile protein and is subjected to delicate regulation by multiple extracellular signaling pathways to control its ubiquitylation and degradation (Figure 8D). Interestingly, dorsal, a drosophila homologue of p65, has been shown to be the key regulator of Snail and Twist during embryonic patterning and the innate immune response in fly (Furlong et al., 2001). In this study, we showed that activation of the NF-κB pathway is the major signal for the stabilization of Snail and the induction of EMT (Figure 1 and Figure 7). Activation of NF-κB correlated with the level of Snail in breast cancer cell lines and tumor samples (Figure 8). Thus, our results indicate that the signaling pathway between Snail and NF-κB is highly conserved from fly to mammal and highlight the notion that cancer cells utilize many developmental strategies for their proliferation and progression. Although Snail has recently been shown to be increased at the transcriptional level by the NF-κB pathway (Bachelder et al., 2005; Julien et al., 2007), our results indicated that the stabilization of Snail by TNFα also occurred at the post-translational level. Therefore, it is likely that NF-κB regulates the activity of Snail through both transcriptional and post-translational mechanisms at EMT (Huber et al., 2004). The post-translational regulation provides a rapid induction of Snail that is urgently needed, while the transcriptional regulation offers a persistent and lasting activity of Snail during chronic inflammation.
The involvement of CSN2 in regulating the stability of Snail is intriguing. CSN2 is the second and most conserved subunit of the COP9 signalosome in all eukaryotes (Cope and Deshaies, 2003; Richardson and Zundel, 2005; Wolf et al., 2003). Recent evidence indicated that CSNs cooperate with the ubiquitin/proteasome system in the regulation of protein stability by interacting with cullin-based ubiquitin ligases to control their activity. In line with this concept, Chang and colleagues found a 512-gene expression pattern as a wound response signature and showed that this signature was a powerful predictor of metastasis and death in diverse types of primary human tumors (Chang et al., 2005; Chang et al., 2004). They demonstrated that amplification of c-Myc and another subunit of COP9 signalsome, CSN5, were the critical genetic regulators for these large-scale transcriptional wound response signatures in cancer (Adler et al., 2006). Strikingly, overexpression of c-Myc and CSN5 in non-transformed mammary epithelial MCF10A cells induce an invasive morphology and confer invasive ability on these cells to invade through 3-dimensional basement membrane matrix. In addition, expression of these two genes activates the wound response program with a similar wound response signature (Adler et al., 2006). As c-myc is a transcriptional factor and CSN5 is a critical regulator of protein stability, we speculate that c-myc and CSN5 control wound response program at transcriptional and post-translational level, respectively. The critical role of CSNs in regulating protein stability is further supported by the observations that many important transcription factors and signal transduction regulators have short half lives (such as p53, c-Jun, HIF1, and IκBα) and are regulated through ubiquitylation and degradation. Expression of CSN2 or CSN5 induces the protein stabilization of c-Jun, IκBα, and p27 (Huang et al., 2005; Yang et al., 2002). In our study, we found that expression of CSN2 induced stabilization of Snail and knockdown of CSN2 expression blocks TNFα-mediated Snail stabilization (Figure 5 and Figure 6). Thus, the post-translation regulation of Snail, mediated by CSN during the inflammatory response, provides a delicate control of Snail during EMT and metastasis. Interestingly, Schweitzer et al recently found that CSN2 controlled NF-κB activity by de-ubiquitylating and stabilizing IκBα (Schweitzer et al., 2007). They found that expression of CSN2 induced the functional assembly of CSN complex. In addition, CSN2 disrupted the association of IκBα with β-Trcp and enhanced the de-ubiquitylating of IκBα by de-ubiquitinylase USP15. On contrary, knockdown of CSN2 expression enhanced TNFα -mediated highly phosphorylated and highly ubiquitylated IκBα. Their results are consistent with our findings. However, whether CSN2 also stabilizes other β-Trcp substrates in a similar fashion requires systematic and thorough investigation.
Third, our study suggests that EMT and MET (mesenchymal-epithelial transition) is a dynamic process that is controlled by an inflammatory microenvironment. By adopting a mesenchymal phenotype through EMT, individual carcinoma cells can infiltrate adjacent tissues, cross endothelial barriers, and enter the circulation through blood and lymphatic vessels. Once the cancer cells reach their secondary tissues or organs, they can revert back to an epithelial morphology through a MET process at the secondary tissues or organs that lack inflammatory stimuli. This model is consistent with the fact that, although some cancers are invasive and metastatic, their secondary metastasized lesions have well-differentiated epithelial characteristics. Our study also supports the view that EMT mainly occurred at the invasive front of metastatic carcinoma (Christofori, 2006; Franci et al., 2006). It is highly unlikely that EMT would occur in a whole tumor because there is lack of tumor-stroma interaction and an inflammatory microenvironment in the center of the tumor.
In summary, we showed that inflammation-triggered migration, invasion, and metastasis of tumor cells are mediated through NF-κB induced Snail stabilization. Knockdown of Snail expression inhibits inflammation-induced cancer cell migration and invasion in vitro and metastasis in vivo. Our study not only reveals a critical mechanism underlying inflammation-induced metastasis but also has important implications in the development of treatment strategies for metastatic cancers.
GSK-3β siRNA expression plasmid pkD-GSK-3β-V1 was purchased from Upstate Biotechnology. Smart pool siRNA against human CSN2 (siRNA-1), p65 (siRNA-1), CSN5 and CAND1 were from Dharmacon (Chicago, IL). A second validated siRNA against CSN2 (siRNA-2) or p65 (siRNA-2) were purchased from Qiagen (Valencia, CA). Primers corresponding to the human Snail (from 507 to 525) were cloned into the pSUPER vector to generate pSUPER-shRNA-Snail-1 expression plasmid (shRNA-1) for RNAi silencing. A second shRNA against Snail (shRNA-2; TRCN0000063819) were purchased from MISSION shRNA at Sigma-Aldrich (St Louis, MO). Information for other plasmids and antibodies used in this study are described in Supplemental Experimental Procedures.
Buffy coats that contain mononuclear cells were collected from the blood of healthy individual donors at Gulf Coast Regional Blood Center (Houston, Texas). Primary blood monocytes were isolated by density-gradient centrifugation through Ficol/Hypaque (Amersham Bioscience, Piscataway, NJ), suspended (8 × 106 cells/ml) in RPMI 1640 medium with 15% heat inactivated human serum (Sigma, St. Louis, MO), and seeded in flasks (Borovikova et al., 2000). After incubation for 2 hr at 37 °C, adherent cells were detached with 10 mM EDTA and resuspended (106 cells/ml) in medium supplemented with 40 ng/ml human MCSF (PeproTech Inc, Rocky Hill, NJ). Cells were allowed to differentiate for seven days in the presence of MCSF. On day seven, fresh medium without MCSF was added to the cells and continued to culture for 24 hr. Human macrophages were finally exposed to LPS (100 ng/ml, Sigma) for 24 hr. The culture medium was collected, centrifuged, stored in aliquots at −80°C, and defined as macrophage conditioned medium. For co-culture experiments, a Transwell inserts (0.2 µm pores, Nunc) that contains 2 × 105/ml macrophages were placed inside an upper well of the Boyden chamber that coated with Matrigel and seeded with various cancer cell lines.
Female ICR-SCID mice (6–8 wks old) were purchased from Taconic (Germantown, NY) and maintained and treated under specific pathogen-free conditions. All procedures were approved by the Institutional Animal Care and Use Committee at the University of Texas Medical Branch and conform to the legal mandates and federal guidelines for the care and maintenance of laboratory animals. Mice (6 for each group) were injected intravenously with breast cancer cell lines MDA-MB-231 (1 × 106 cells/mouse) or MDA-MB-435 (2 × 106 cells/mouse) cells via the tail vein. Three wks later, mice were injected intraperitoneally with 10 µg of LPS (serotype 055:B5; Sigma) in PBS or PBS alone. After an additional three wks (for MDA-MB435) or nine wks (for MDA-MB231), mice were sacrificed by cervical dislocation, and lungs were removed and weighed. Visible lung metastatic nodules were examined macroscopically or detected in paraffin-embedded sections stained with hematoxylin and eosin. Data was analyzed by Student's t-test; a p value <0.05 was considered significant.
The experiments were repeated at least two times. Results are expressed as mean ± SD or SEM as indicated. An independent Student’s t-test was performed to analyze the luciferase assay. Two-tailed student’s t-test was used to compare the intergroup. p < 0.05 was considered statistically significant.
A major challenge in cancer research is to identify the extrinsic signals and intrinsic factors that initiate metastasis. In this study, we found that the informatory cytokine TNFα dramatically enhanced the protein stabilization of Snail through NF-κB mediated CSN2 induction. We showed that the expression of Snail correlated with the activation of NF-κB and knockdown of Snail expression significantly inhibited cell migration and invasion induced by inflammation. Our finding demonstrates that TNFα -mediated Snail stabilization plays critical role in cell migration, invasion, and metastasis. Our study not only reveals a critical mechanism underlying inflammation-induced metastasis but also has important implications in the development of treatment strategies for metastatic cancers.
Other experimental procedures are described in the Supplemental Experimental Procedures.
We thank Dr. Warner C. Greene for providing wild-type and deletion mutants of p65. We also thank Dr. James R. Woodgett for providing immortalized wild-type and GSK-3β −/− MEFs. This work was supported by the John Sealy Memorial Endowment Fund, a pilot award from the ACS (IRG-110376), and grants from the Susan G Komen Foundation (KG081310) and NCI (RO1-CA125454) (to B.P. Zhou) and RO1 CA104748 and RO1 DK48498 from the NIH (to B.M. Evers). Y. Wu was supported by postdoctoral fellowship from NIH (T32CA117834).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.