These authors contributed equally to this work.
Promotion of tumor cell metastasis and vasculogenic mimicry by way of transcription coactivation by Bcl-2 and Twist1: A study of hepatocellular carcinoma†
Article first published online: 21 JUL 2011
Copyright © 2011 American Association for the Study of Liver Diseases
Volume 54, Issue 5, pages 1690–1706, November 2011
How to Cite
Sun, T., Sun, B.-c., Zhao, X.-l., Zhao, N., Dong, X.-y., Che, N., Yao, Z., Ma, Y.-m., Gu, Q., Zong, W.-k. and Liu, Z.-y. (2011), Promotion of tumor cell metastasis and vasculogenic mimicry by way of transcription coactivation by Bcl-2 and Twist1: A study of hepatocellular carcinoma. Hepatology, 54: 1690–1706. doi: 10.1002/hep.24543
Potential conflict of interest: Nothing to report.
- Issue published online: 28 OCT 2011
- Article first published online: 21 JUL 2011
- Accepted manuscript online: 11 JUL 2011 12:29PM EST
- Manuscript Accepted: 29 JUN 2011
- Manuscript Received: 12 MAY 2011
- National Natural Science Foundation of China. Grant Number: 30830049
- Cooperation project of China-Sweden. Grant Number: 09ZCZDSF04400
- Tianjin Natural Science Foundation. Grant Numbers: 08JCZDJC23500, 09JCYBJC12100
- 973 Program from the Ministry of Science and Technology of China. Grant Numbers: 2009CB918903, 2011CB933104
The antiapoptotic protein Bcl-2 plays multiple roles in apoptosis, immunity, and autophagy. Its expression in tumors correlates with tumor grade and malignancy. The recapitulation of the normal developmental process of epithelial-mesenchymal transition (EMT) contributes to tumor cell plasticity. This process is also a characteristic of metastatic cells and vasculogenic mimicry. In the present study we report functional and structural interactions between Bcl-2 and the EMT-regulating transcription factor Twist1 and the relationship with metastasis and vascular mimicry. Bcl-2 and Twist1 are coexpressed under hypoxia conditions. The Bcl-2 can bind to Twist1 in vivo and in vitro. This interaction involves basic helix-loop-helix DNA binding domain within Twist1 and through two separate domains within Bcl-2 protein. Formation of the Bcl-2/Twist1 complex facilitates the nuclear transport of Twist1 and leads to transcriptional activation of wide ranges of genes that can increase the tumor cell plasticity, metastasis, and vasculogenic mimicry. Finally, nuclear expression of Bcl-2 and Twist1 is correlated with poor survival of these patients in a cohort of 97 cases of human hepatocellular carcinoma. Conclusion: The results describe a novel function of Bcl-2 in EMT induction, provide insight into tumor progression, and implicate the Bcl-2/Twist1 complex as a potential target for developing chemotherapeutics. (HEPATOLOGY 2011;)
Bcl-2 is the founding member of a family of proteins that play important roles during development and disease. Bcl-2 family members exhibit either pro- or antiapoptotic behavior. They include six classical antiapoptotic proteins and three preapoptosis proteins that show canonical Bcl-2 homology domains (BH1-4). These domains interact to form functional hetero- or homodimers.1 Bcl-2 members interact to exert fine control over cell death. The classically defined interaction between Bcl-2 and Bax proteins at mitochondrial membrane voltage-dependent anion channels results in the release of Ca2+ and cytochrome C. Consequently, a cascade of cell death processes is initiated and leads to apoptosis and necrosis.2, 3 In addition, Bcl-2 family members also interact with other proteins to modify and regulate cellular metabolism, immune response, and autophagy.4-8 Taken together, these observations suggest that the Bcl-2 family operates by way of diverse mechanisms to regulate cell growth and death. In many malignant tumors, Bcl-2 importantly localizes at the endoplasmic reticulum, nucleus, and other nonmitochondrial sites. The antiapoptotic action of Bcl-2 is also associated with poor prognosis.9, 10 Nevertheless, the functional link of Bcl-2 to the mechanisms of tumor progression remain unclear.
Epithelial-mesenchymal transition (EMT) is a normal developmental process wherein phenotypic plasticity alters the properties of cell adhesion and migration. EMT has recently gained considerable attention as a mechanism leading to tumor metastasis.11-13 Our recent studies14 have implicated Twist1 and EMT in the formation of vasculogenic mimicry (VM) by human hepatocellular carcinoma cells (HCCs) in vivo and in vitro. The ectopic expression of Twist1 by HCCs resulted in the loss of E-cadherin, expression of the endothelial cell marker vascular endothelial (VE)-cadherin, and VM formation. These phenomena implicate the EMT plasticity of different mesoderm types in a specific tumor microenvironment, similar to vascular differentiation during embryonic development.
The Twist family of b-Helix-Loop-Helix (bHLH) transcription factors is known to regulate EMT in a variety of tumors.15, 16 The general mechanism of action of this regulation process involves the inhibition of E-cadherin transcription by interacting with its promoter. The expression of downstream mesenchymal marker molecules in cells is further triggered. As a result, phenotypic plasticity occurs and cell motility is activated.17, 18 In the specific tumor environment, Twist1 undergoes relocation into the nucleus to exhibit its transcriptional regulation effect. Twist1 has a nuclear localization signal (NLS). However, the mechanism that regulates nuclear translocation remains unclear. Accessory proteins required for translocation have not yet been identified.
In the present study we report that the Twist1-induced EMT in HCCs may operate in connect with antiapoptotic Bcl-2 under hypoxia conditions. These two proteins form a complex in vivo and synergistically activate the transcriptional activity of multiple downstream targets, which lead to vascular mimicry and tumor promotion.
Materials and Methods
Plasmid pcDNA3-Twist1-Flag has been described.19 Sequences of pcDNA3-Bcl2-Flag, pDSRed-Bcl2, pEGFP-Twist1, deletion mutants of Bcl-2, and Twist1 were subcloned, respectively, into pGEX-4T (see Supporting Materials).
The small interfering RNA (siRNA) coding oligos against human Twist1 and Bcl-2 were designed and verified to be specific to Twist1 and Bcl-2. The Twist1-siRNA-targeting sequence was AAGCTGAGCAAGATTCAGACC (siTwist1 nucleotides 505-525). The Bcl-2-siRNA-targeting sequence was CAGGACCTCGCCGCTGCAGAC (siBcl-2 nucleotides 200-221). The U6 promoter with the Twist1-siRNA or Bcl-2-siRNA insert was subcloned into pRNA-U6-Neo (Genscript, China). A nonsilencing siRNA sequence (target sequence AATTCTCCGAACGTGTCACGT) was used as the negative control as described.19
Cell Culture, Hypoxia, and Transfection.
The HepG2, PLC, SMMC7221, and 293 cell lines were purchased from the American Type Culture Collection. The cell culture for the proliferation and functions test were previously described.19 The details of the hypoxia culture and transfection are included in the Supporting Materials.
Cell Proliferation, Adhesion, Clone Formation Assay, Invasion, Wound Healing Assay, and Three-Dimensional Culture.
The details of these procedures are provided in the Supporting Materials.
Quantitative Reverse Transcription Polymerase Chain Reaction (PCR), Western Blot, Gelatin-Zymography, Flow Cytometry Analysis, and Immunofluorescence.
The details of these procedures are provided in the Supporting Materials.
Fluorescent Labeling by Bcl-2/Twist1.
pDSRed was used to express Bcl-2-DSRed fusion protein. pEGFP was used to express Twist1-EGFP fusion protein. HepG2 and 293 were immediately transfected. Laser scanning confocal microscopy was used to observe subcellular localization.
Coimmunoprecipitation and Glutathione S-transferase (GST) Pull-Down.
Cell lysates with 500 μg of protein prepared from HepG2 cells were cleaned with protein G/A beads before being subjected to coimmunoprecipitation (Co-IP) using 2 μg of Twist1 or Bcl-2 antibody. An equal amount of IgG was used as the negative control. Immunocomplexes were denatured by boiling in a sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer, and were separated in 6% SDS-PAGE gels for western blot using Twist1 and Bcl-2 antibodies.
The expression of Bcl-2 or Twist1 and of the serial deletion mutants of GST-Twist1 or GST-Bcl-2 were grown in bacteria. The GST-Twist1 and its deletion mutant protein were purified and immobilized on glutathione-sepharose 4B (GE Healthcare Bio-Science) and incubated overnight at 4°C with HepG2 extracts containing Bcl-2 (Flag-tag). The bound samples were washed thrice with buffer and subjected to western blot analysis with an anti-Flag antibody (see Supporting Materials for details).
Luciferase Reporter Gene Assay.
The plasmids pAP1-TA-luc, pSTAT3-TA-luc, and pNF-κB-TA-luc were used to determine the activation levels of AP1, STAT3, and nuclear factor kappaB (NF-κB) (see Supporting Materials).
Chromatin Coimmunoprecipitation (ChIP) Sequence and Complementary DNA (cDNA) Microarray.
The HepG2-control, HepG2-Twist1, HepG2-Twist1, and HepG2-Bcl2/Twist1 cells were used as samples. The ChIP-sequence method was employed to determine the effect of different treatment methods on Twist1 transcription combination sequences. The details of all the procedures are in the Supporting Materials.
Collection and Analysis of Patient Samples.
Tissue specimens were obtained from the Tumor Tissue Bank of the Tianjin Cancer Hospital. The specimens were from 97 patients who underwent hepatectomy for HCC between 2001 and 2005. The diagnoses of these HCC samples were verified by pathologists. Detailed pathologic and clinical data were collected for all samples, including the Edmondson tumor grade, metastasis, and survival duration. Paraffin-embedded tumor tissue samples were collected from patients who had not undergone therapy prior to the surgical operation on the tumor. The use of these tissue samples was approved by the Institutional Research Committee.
The details of the immunohistochemistry analysis are indicated in the Supporting Materials.
Murine Xenograft Model.
Six-week-old female NIH BALB/c-null mice were housed in the animal facilities of the Tianjin Medical University as approved by the Institutional Animal Care and Use Committee. HepG2 cells (107 cells/ml) were mixed with Matrigel (BD Bioscience) and subcutaneously injected into the backs of nude mice (0.1 mL/mouse). For 25 days the mice were monitored and tumor sizes were measured daily using a caliper. After 25 days the experiments were terminated because of the tendency of HepG2-Bcl2/Twist1 cells to become necrotic and form skin ulcers. After completing the observation the mice were sacrificed. Tumors were harvested and stored at −80°C for subsequent tests.
Yeast Two-Hybrid Analysis.
The details of the yeast two-hybrid analysis are in the Supporting Materials.
All data were evaluated using SPSS v. 13.5. Differences were considered significant at P < 0.05. The significant groups are marked with an asterisk in the figures.
Bcl-2 and Twist1 Show Similar Expression Kinetics Induced by Hypoxia.
Bcl-2 is an important mitochondrial membrane pore component that functions in a variety of proapoptotic stress responses, such as hypoxia. In the present study the growth response and hypoxia-induced up-regulation of Bcl-2 in the hepatoma cell lines HepG2, PLC, and SMMC7221, as well as in control cells, were examined. To prevent hypoxia-induced cell death and general protein degradation caused by energy depletion, each cell type was returned to normal oxygen conditions (hypoxia-normoxia group, H-N) after 24 hours of hypoxia. Each cell line showed a significant decrease in cell proliferation following hypoxia, which was reversed by normoxia conditions (H-N) to proliferation levels above control values (Fig. 1A). Notably, the proliferation rate at the terminal phase (72 hours) of the H-N group significantly increased compared with the normoxia alone control group. Migration and invasion assays showed similar responses (Fig. 1A). Cell migration and invasion decreased following hypoxia. In contrast, the H-N treatment caused an increase above the control group.
HepG2 cultures were also assessed for their abilities to undergo morphological conversion. This conversion leads to VM in a three-dimensional (3D) culture following hypoxia and after returning to normoxia. Cells grown under hypoxic conditions showed a modest level of conversion to “tube” formations, whereas those grown under control conditions did not show any indication of such 3D growth. In contrast, cells that were first treated under hypoxic conditions and were then returned to normoxia showed a robust conversion to 3D tube formations (Fig. 1B).
The expression levels of Bcl-2 and Twist following hypoxia and after returning to normoxia were assessed using quantitative PCR and western blot (Fig. 1C,D). Messenger RNA (mRNA) and protein levels showed expression peaks for Bcl-2 and Twist1 about 24 hours after cell hypoxia. The expression levels gradually decreased to undetectable levels at later timepoints. When hypoxia was relieved after 24 hours by returning to normoxia, Bcl-2 and Twist1 still had high expression levels. Taken together, these observations suggested that return to normoxia after hypoxia may trigger an increase in cell proliferation, movement, and molding. All these functional responses lead to VM, and may be mechanistically linked to the expression levels of Bcl-2 as well as Twist1.
Synergism Between Bcl-2 and Twist1 Leading to EMT.
The cellular response described above included EMT features. Therefore, EMT markers in HepG2 cells engineered to overexpress Bcl-2 and Twist1, separately or together, were assessed. The overexpression of either Bcl-2 or Twist1 alone showed no detectable effect on the abundance of endogenous Twist1 or Bcl-2, respectively (Fig. 2A). As expected, the overexpression of Twist1 resulted in a switch in the expression from E-cadherin to vimentin, indicative of EMT. Interestingly, this result was accentuated by the coexpression of both Bcl-2 and Twist1.
The HepG2-control, HepG2-Bcl2, HepG2-Twist1, and HepG2-BT (Bcl-2/Twist1 coexpression) cells were then analyzed for functional changes in proliferation, migration, and altered growth in a 3D culture. Compared with the Bcl-2 or Twist1 group and the control group, there was a significant increase in the activity of the Bcl-2/Twist1 group in terms of proliferation, migration, invasion, and clonigenicity (Fig. 2E). To rule out the effects of nonuniform transfection efficiency, stably transfected subclones for overexpressing Bcl-2 and Twist1, as well as for co-overexpressing both proteins were selected. The selected clones were then assayed for growth in a 3D culture to compare their characteristics when grown in and on 3D gel matrixes (Fig. 2F). HepG2-BT cultures showed growth structures similar with human umbilical vein endothelial cells when cultured in the “on gel” mode. When Bcl-2 and Twist1 were coexpressed they showed a pronounced and characteristic tube formation with smooth continuous cavities. In contrast, the single transfection with Bcl-2 or Twist1 and control groups showed no tubes when grown under either on gel and in gel modes, whereas HepG2-Twist1 exhibited modest tubal structures. Finally, HepG2-Bcl2 and the control groups had no tubal structure and only exhibited clone proliferation.
Additional evidence for VM was demonstrated by analyzing the vascular endothelial markers VE-cadherin, vascular endothelial growth factor receptor 1 (VEGFR1), VEGFR2, and matrix metalloproteinase (MMP) protease activities (Fig. 2B-D). Our previous work indicated the up-regulation of VE-cadherin expression in a 3D culture system based on the Twist1 transfection of HepG2 cells. However, no change in the 2D monolayer culture was observed. In the present study, when grown in a 3D culture, HepG2-BT had a dramatic up-regulation of VE-cadherin. This up-regulation was about two times as much as that of a single transfection with Twist1 (Fig. 2C). These results were confirmed by flow cytometry. In the HepG2-BT group, VE-cadherin-positive cells accounted for about 97.5%. In the single transfection Twist1 group these cells accounted for 88.7%. The VEGFR1-positive ratio of the HepG2-BT group was 99.1%, whereas that of the HepG2-Twist1 group was only 14.2%. The VEGFR2-positive ratio of the HepG2-BT group was 99.7%, whereas that of the HepG2-Twist1 group was only 39.2%. The HepG2-Bcl2 and control groups both had low-percent signals (Fig. 2D). MMP2 and MMP9 activities were highest in the HepG2-BT group (Fig. 2B). To examine the in vivo effects of the coexpression of Twist1 and Bcl-2 on tumor development, a murine xenograft model was used. HepG2 cells were utilized to establish xenografts in nude mice. Nodule formation and growth (volume) were monitored for 25 days. HepG2-BT showed the highest rate of tumor growth compared with HepG2-Twist1 and the control group (60% and 50%, respectively) (Fig. 3A). The western blot analysis of the excised tumors revealed that HepG2-BT had high expression levels of VE-cadherin and vimentin, suggesting that Twist1 and Bcl-2 can work synergistically to induce EMT in vivo (Fig. 3B). Taken together, these observations suggested a synergism between Bcl-2 and Twist1 can result in the increased proliferation, migration, invasion, and vasculogenic activities of tumor cells, and tumor growth in vivo.
Interaction Between Bcl-2 and Twist1 Leads to the Translocaton of Twist1 to the Nucleus.
The above observations led us to examine the underlying mechanisms of interaction between Bcl-2 and Twist1. A yeast two-hybrid system was used to evaluate the direct interactions between Bcl-2 and Twist1. The cDNA fragments encoding Twist1 and Bcl-2 were cloned into pGBKT7 (bait vector) and pGADT7 (prey vector), respectively; the protein binding between from bait and prey vectors are indicated by survival of Saccharomyces cerevisiae reporter strain AH109. As shown in Fig. 4A, only yeast strains containing expression vector for both Twist and Bcl-2 survived in the selective media, whereas these strains transfected with either Twist1 or Bcl-2 alone failed to produce a viable strain. These results demonstrated that Bcl-2 and Twist1 can interact and form a functional complex in yeast (Fig. 4A). To further demonstrate such protein-protein interaction in vivo, Co-IP was used to determine the protein complex in vivo. As shown in Fig. 4B, antibody against Twist1 will coprecipitate the Bcl-2. Similarly, antibody against Bcl-2 will also coprecipitate Twist1. Furthermore, their binding affinity was increased as evidenced by the level of increased pull-down protein after the hypoxia treatment in HepG2 for 24 hours (Fig. 4B). To differentiate exogenous protein expression from endogenously expressed protein, we used expression vector for Twist1 that is flag-tagged at its 5′ end. Our results showed increased expression as detected by anti-Flag antibody, demonstrating increased expression from exogenously transfected expression vector (Supporting Fig. s2).
To define the specific region of Twist1 involved in the interaction between Twist1 and Bcl-2, we used a series of expression vectors encoding different deletion mutants for Twist1 and Bcl-2. First, we expressed five different mutants of Twist1, N158, N121, N50, C112, and NLS; five different mutants of Bcl-2, N109, N138, N185, N203, and TM (expressed in Escherichia coli). The schematic of each mutant is shown in the top panel of Fig. 4C. The truncated protein at the amino acid 158 from N-terminus (N158) did not affect the binding of Twist1 to Bcl-1; however, further deletion into amino acid 121 abolished its binding, thus the binding can be maintained even after the first 112 amino acids were deleted, as shown by Twist1 C112 truncated protein. The above study defined the protein binding region to nine amino acids from 113 to 121 within Twist1 protein. These nine amino acids are located within the bHLH domain and play an important role in DNA binding and transcription activation.
We further mapped the regions of Bcl-2 to Twist1 using five expression vectors expressing deletion mutant proteins for Bcl-2. As shown in Fig. 4C, three deletion mutants from amino acid 109 to 185 resulted in loss of their binding to Twist1, whereas additional amino acids from 186 to 203 restored its binding to Twist1, suggesting that the region between amino acids 185 to 203 is required for Twist1 binding. Surprisingly, the C-terminal protein fragment from 201 to 233 as defined by TM is also sufficient to bind to Twist. The results define that the C-terminus from 185 to 233 has the binding sequence for Twist1. Taken together, our results defined nine amino acids within the bHLH domain and the C-terminus of Bcl-2 that are critical for their binding between these two proteins.
Next, we determined whether Twist1 and Bcl2 interaction can be directly visualized in vivo. To this end, we performed immunofluorescence staining against Twist1 and Bcl-2 and examined the colocalization of these two proteins within single living cells. The Twist1 expression is shown in green, whereas Bcl-2 expression is shown in red. The cells were cultured under hypoxia conditions. As shown in Fig. 4D, direct colocalization of Twist1 and Bcl-2 can be observed in multiple cells, as indicated by yellow fluorescence due to overlapping of red Bcl-2 and green Twist1. The strong yellow signal was observed in nucleus, although it can be observed in cytoplasm. To further demonstrate that the specific yellow signal is due to a specific interaction between Twist1 and Bcl-2 and not a false-positive colocalization due to high levels of endogenous nonspecific proteins, we constructed a fusion construct expression Twist1 and green fluorescence protein (GFP) and Bcl-2 with red fluorescence protein (RFP) designated Twist1-EGFP and Bcl-2-DSRed, respectively. These two constructs were cotransfected in the HepG2 and 293 cells. As shown in Fig. 4E, multiple yellow signals were observed, indicating colocalizations of Twist1 and Bcl-2 and Twist1 in these cells. Most of the localization regions appeared as points and were located around or in the middle of the nucleus. In the cytoplasm, rare colocalizations were observed. Bcl-2 was mostly located in the nearby cytoplasm and in the nuclear membrane around the nucleus, whereas Twist1 was mostly in the nucleus. However, when Bcl-2 and Twist1 are coexpressed the two combined into a protein complex and were present largely in the nucleus (Fig. 4E), suggesting that Bcl-2 may facilitate the nuclear transport of Twist1.
To further demonstrate the functional interaction between Bcl-2 and Twist1, we examined how Bcl-2 affects the nuclear transport of Twist1. The HepG2 cells transfected with vector expression short hairpin RNA (shRNA) against Bcl-2 and expression vector for Twist1 were fractionated into cytoplasmic and nuclear extracts, and western blot was used to examine the Bcl-2 and Twist1 distribution in these two compartments. As shown in Fig. 5A, as expected, expression of shRNA against Bcl-2 results in loss of protein Bcl-2 in both cytoplasm in nucleus, ectopic expression of Twist1 expression vector led to an increased expression of both cytoplasm and nucleus but predominantly in nucleus (Fig. 5A, right panel). However, when cells contain both Twist1 expression vector and shRNA against Bcl-2, the nuclear expression of Twist1 is completely attenuated. To further demonstrate whether Bcl-2 facilitates the nuclear transport of Twist, we examined these cells in hypoxia conditions and in the presence of overexpression of Bcl-2 rather than knockdown by shRNA. As shown in Fig. 5B, either hypoxia or ectopic expression of Bcl-2 can lead to up-regulation of expression of Twist1 with preferential expression in the nucleus. These results further support the interaction between Bcl-2 and Twist1; Bcl-2 could be an important cofactor to facilitate the transport of Twist1 to the nucleus (Fig. 5A,B).
Coexpressed Bcl-2 and Twist1 Cause Changes in Wide Signal Transduction.
To examine how interactions between Bcl-2 and Twist1 affect global gene expression, we examined the promoters bound to Twist1 using a ChIP-sequence analysis for HepG2-control, HepG2-Twist1, and HepG2-BT that are transfected with both Bcl2 and Twist1. The DNA fragments bound to Twist1 picked up by ChIP assay were sequenced. The results showed that the number of gene promoters that bound to Twist1 in the HepG2-BT expressing both Bcl-2 and Twist1 cells reached 100, whereas only 43 promoters were detected in HepG2 transfected with Twist1 expression vector alone (Fig. 6A). These genes are involved in many processes such as cell signal transduction, cell proliferation, angiogenesis, and cytoskeleton formation (detailed information is provided in the Supporting Materials, Table s7). To verify whether key signal transduction pathways were activated by the interaction of Bcl-2/Twist1, reporter gene plasmids with AP1, Stat3, and NF-κB activation sequences were used in the HepG2-BT and control cells. The results showed that the AP1 and Stat3 activities in the HepG2-BT group significantly increased. In contrast, the NF-κB transcriptional activity did not significantly change compared with the control and HepG2-Twist1 groups (Fig. 6B). The western blot analysis also showed similar results; a high level of c-Jun, p-c-Jun, as well as Stat3 was observed in the HepG2 cells expressing both Bcl-2 and Twist expression vector (HepG2-BT).
We also examined the global changes in mRNA for HepG2-control, HepG2-Bcl-2, HepG2-Twist1, and HepG2-BT using cDNA array. Cluster and comparative analyses showed a distinct pattern of mRNA expression when these cells exogenously expressed transfected Bcl-2, Twist1 or a combination of both (Supporting Fig. s3). Transfection of Bcl-2 led to an increase of multiple genes involved in embryonic development, transcription, cell adhesion, and may involve other biological functions in normal and cancer development (Supporting Table s6).
Coexpressed Bcl-2 and Twist1 Correlates with VM Formation and Tumor Metastasis.
We analyzed the expression of Bcl-2 and Twist1 in a cohort of 97 cases of hepatocellular carcinoma with detailed clinical and pathologic information (Supporting Tables s2, s3). Comparisons were made between the metastasis and nonmetastasis groups, as well as between the VM and non-VM groups. The results indicated that the Bcl-2 and Twist1 nuclei were positive, and that both had statistical significance (Tables s4, s5). The results showed that there was a correlation between the positive Bcl-2 and Twist1 nuclei, and that there was a statistical significance (Fig. 7A). Based on the analysis, the cytoplasmic expression of Bcl-2, Twist1, and Twist2 had no significant correlation with metastasis or VM formation. Correlation analyses were carried out for the relative proteins VE-cadherin, E-cadherin, MMP2, MMP9, HIF-1a, and VEGF. The results showed the correlation of HIF-1a, VEGF, VE-cadherin, and MMP9 with the nuclear expression of Bcl-2 and Twist1 (Fig. 7B).
The Kaplan-Meier survival analysis suggested that the positive Bcl-2 nucleus, Twist1 nucleus, VE-cadherin, and MMP9 were correlated with poor survival of these patients. Their survival time was also shorter than that of the negative group. The other indicators were not statistically significant (Fig. 7C).
Tumor cells can respond differently to various microenvironments, such as apoptosis, senescence, and plasticity.20 The key factor in a microenvironment is hypoxia. Hypoxia can induce the up-regulation of HIF-1 in tumor cells, activate numerous signal transduction pathways, promote tumor cell proliferation, induce EMT occurrence, or secrete VEGF. Consequently, a single tumor cell or tumor cell populations are forced to adapt to the changes caused by hypoxia.21-23 Tumor cells obtain blood for relieving hypoxia using a secretion factor or by simulating angiogenesis. Among them, tumor metastasis and VM formation are involved in tumor cells losing their epithelial adhesion molecules and obtaining a mesenchymal phenotype (as vimentin, N-cadherin, or VE-cadherin).19, 24-27 Our preliminary work19 has proven that VM exists in hepatocellular carcinoma, and that VM formation is correlated with the EMT-regulating factor Twist1. These findings indicate that VM formation may possibly be a part of EMT.25 Therefore, we named this process epithelial-endothelial transition (EET). The EET of tumor cells under a specific condition can be used for early vascular structuring to obtain the original blood source. Tumor cells are further remodeled or differentiated into endothelial cell-like tumor cells to participate in the construction of tumor microcirculation.28, 29
Hypoxia can directly induce the up-regulation of HIF-1. HIF can further combine with the promoter of Twist1 to promote its transcript expression, and further induce the occurrence of EMT.30, 31 Based on the immunohistochemical analysis of clinical pathological specimens, not all Twist1 cells were present in the nucleus. Instead, the vast majority is located in the cytoplasm,17, 32 although the underlying mechanism that regulates cytoplasmic and nuclear localization of Twist1 is not clear. It has been reported that the NLS in Twist1 can lead to the interaction between Twist1 and the nuclear membrane pore channel and the NLS can also induce Twist1 to enter the nucleus and act as a transcription factor.33 In the present study we provide data to show that the up-regulation of Twist1 reaches its peak level 24 hours after hypoxia, whereas the expression of Twist1 decreased after 24 hours, as many of these cells die due to continued hypoxia. When hypoxia was relieved after 24 hours, the high expression level of Twist1 can be sustained for more than 24 hours. Interestingly, the antiapoptotic protein Bcl-2 also exhibited an expression peak and trend similar to Twist1 within the same period (24 hours). This result indicated that Bcl-2 and Twist1 possibly acted during the stress phase in the same cell and followed similar kinetics. Bcl-2 and its family members have been found to mediate the apoptosis process. They have also been found to participate in protein modification and to form a complex with other proteins for participating in complicated processes of cell metabolism.2, 3, 34 In tumor tissues, Bcl-2 expression in the nucleus correlates with poor prognosis. Our data provide evidence that Bcl-2 may form a complex with Twist1 and synergistically to promote the transcription of downstream target genes which can lead a cascade changes in proliferation, adhesion, migration, infestation, clone formation, and tubal formation of tumor cells.
The formation of Bcl-2 with Twist1 as a protein complex to stimulate the transcription was unexpected. Bcl-2 has long been considered as a mitochondrial membrane protein. However, reports on the effects of Bcl-2 and its other family members in more complex biological processes are limited. The present study revealed that specific amino acids within Bcl-2 and Twist1 are involved in the binding of two proteins and form a novel functional complex and jointly enter the nucleus, which leads to changes of multiple downstream target genes. Such a heterodimer is more potent in stimulating the transcription of multiple downstream target genes than Twist1 alone. Although the detailed mechanisms for interaction between Twist1 and Bcl-2 are not clear at this time, we speculate that the following mechanisms may be involved. Bcl-2 may be initially associated with the nuclear membrane pore structure, and assists Twist1 in entering the nucleus. In the nucleus, Bcl-2 and Twist1 forms a protein complex and functions in synergy on the promoters of different target genes to regulate their transcription. As demonstrated in this study, the interaction between Bcl-2 and Twist leads to a marked synergistic biologic effect compared to Twist1 alone, as reflected in changes of multiple known target genes, global expression profile, and finally tumor growth. In summary, our study defines a novel mechanism for development of EMT and cancer development mediated by Twist1, and provides a foundation for the design of a novel inhibitor for this process in future investigations.
- 9Knockdown of eIF4E suppresses cell growth and migration, enhances chemosensitivity and correlates with increase in Bax/Bcl-2 ratio in triple-negative breast cancer cells. Med Oncol 2010., , , , , , et al.
- 21Hypoxia-inducible factor-1 as a therapeutic target in endometrial cancer management. Obstet Gynecol Int 2010; 2010: 580971., , , .
Additional Supporting Information may be found in the online version of this article.
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.