Transcription factor E2F1 promotes EMT by regulating ZEB2 in small cell lung cancer

Sixty SCLC biopsy tissue samples before treatment were obtained from the affiliated hospital of Binzhou Medical University from January 2014 to January 2015. All patients signed informed consent forms before providing tissue samples. This research was approved by the Medical Ethics Committee of Binzhou Medical University (No. 2013027). This study was performed according to the Declaration of Helsinki and to the relevant ethical guidelines for research on humans. The basic patient information is listed in Table 1. Human SCLC cell lines H446 (TCHu196, Chinese Academy of Sciences cell bank) and H1688 (TCHu154, Chinese Academy of Sciences cell bank) were stored in our lab. All cells were cultured in RPMI 1640 media (Gibco, Cat:89,984) with 10% FBS (Gibco, Cat: 26,140,079) and 100 U/ml penicillin and 100 μg/ml streptomycin.

All tissue sections were dewaxed in dimethylbenzene and rehydrated in an alcohol gradient ranging from 100% to 75%. Antigen retrieval was performed by placing samples in an EDTA antigen repair solution (ph = 9.0). Sections were then washed three times with PBS buffer (ph = 7.0). Next, the sections were incubated with a primary antibody at 4 °C overnight. The antibodies were used at the following dilutions: 1:100 for VIM (Cell signalling technology, Cat:5741), 1:150 for CDH2 (Cell signalling technology,Cat:13,116), 1:250 for CLDN1 (Cell signalling technology, Cat:13,255), 1:100 for CTNNB1 (Cell signalling technology, Cat:8480), 1:400 for CDH1 (Cell signalling technology, Cat:3195), 1:50 for E2F1 (Santa Cruz, Cat:sc-251), 1:200 for ZEB2 (Santa Cruz, Cat:271,984), ZEB1 (Cell signalling technology, Cat:3396), SNAI1 (Cell signalling technology, Cat:3879), and SNAI2 (Cell signalling technology, Cat:9585). The sections were washed three times with PBS and incubated with HRP-conjugated secondary antibodies for 40 min at 37 °C. Then, the DAB (diaminobenzidine) reaction, hematoxylin staining, differentiation with hydrochloric acid alcohol, dehydration and transparency steps were conducted in turn. All images were captured by Leica Microsystems CMS (DFC365 FX). All staining was scored according to our previous reports [2, 3]. In brief, the staining was quantified using a 4-value intensity score: 0 as negative; 1+ as weak; 2+ as moderate, 3+ as strong, and the percentage (0–100) of the extent of reactivity. A final score was obtained by multiplying the intensity and reactivity extension values (range, 0–300) [14].

Adenovirus (1 × 10⁹ titers) containing shRNA against E2F1 was provided by Gene Pharma Company. H446 cells were conventionally cultured in six-well plates. At a confluence of approximately 70%–80%, 1 × 10⁶ virus titers were added and mixed gently. After 12 h, the completed medium was added to cells and the previous medium was removed. After 72 h, puromycin (5 μg/ml) was added to the media. After 5 days, all cells were digested, diluted, and cultured in a 96-well plate to form monoclonal colonies. The medium containing puromycin (1 μg/ml) was changed every 3 days until the confluence was approximately 90%–100%. Cells were then digested and inoculated into 48-well, 24-well and 6-well plates. These cells were named H446-E2F1sh.

siRNAs targeting E2F1 was transfected into H1688 according to the methods stated in our previous report [2]. H1688 cells in which E2F1 was transiently silenced were named H1688-E2F1si.

Total RNA was extracted with RNAiso Plus (Takara, Cat: 9108), and cDNA was synthesized with the Prime Script RT reagent kit with gDNA Eraser (Takara, Cat: RR047A). Real-time PCR was performed according to the instructions provided for the SYBR Fast qPCR Mix (Takara, Cat: RR430A). The primers used are listed in additional file 1.

All cells were lysed with RIPA lysis buffer containing a protease inhibitor cocktail (Sigma, Cat: S8820). The protein concentrations were measured, and 50 μg of protein was run on an SDS-PAGE gel and electrophoretically transferred onto NC membranes. The membranes were blocked with 5% fat-free milk and incubated overnight at 4 °C with primary antibodies, including VIM (1:1000), CDH2 (1:1000), CTNNB1 (1:1000), CDH1 (1:1000), E2F1 (1:500), ZEB2 (1:500), α-tubulin (1:1000), β-actin (1:1000) and GAPDH (1:2000). The membranes were then washed three times and incubated for 40 min with HRP-conjugated secondary antibodies. Protein bands were detected by the ECL system [2, 3].

Cells were placed on slides, washed three times with PBS buffer, and fixed with ice-cold methanol and acetone (1:1). Next, the slides were blocked for 30 min with goat serum, washed three times with PBS-TX (PBS containing 1% Tritonx-100), and incubated overnight at 4 °C with primary antibodies. The slides were then incubated with Rhodamine or FITC-labelled fluorescent secondary antibodies and DAPI. Laser scanning confocal microscope (leica-LM7000) was used to observe cell morphology and capture pictures.

Genomic DNA was extracted from H446 cells, and the ZEB2 promoter was amplified by PCR and purified. The PCR fragment and pGL3-basic vector were digested with Nhe I and Bgl II enzymes, and T4 DNA ligase was used to ligate these two fragments together to construct the ZEB2 promoter reporter vector. Luciferase activity analysis was performed according to methods described in our previous report [2].

Distant metastasis in early stages is a typical feature of SCLC, and the occurrence of EMT is considered to be an early event in the process of tumour invasion and metastasis. Therefore, it is important to determine the expression pattern of EMT markers in SCLC tissue. We examined CDH1 expression and found that it was highly related with CTNNB1 expression (r = 0.9985, (p < 0.001) and that both proteins were located on the cellular membrane and absent from the cytoplasm and nucleus. CTNNB1 and CDH1 were expressed in 90% (54/60) of small cell lung cancer cells (Fig. 1a), in which weak expression (score between 10 and 20) was 11.11%, moderate (score between 20 and 100) was 70.37% and strong (score between 100 and 300) was 18.52% (Additional file 2: Figure S1). This result was consistent with other reports [15‐17]. CDH1 and CTNNB1 in bronchial epithelial cells as positive control were presented in Additional file 3: Figure S2. The Spearman analysis showed that CDH1, CTNNB1 were significantly relevant with tumour sizes (p = 0.03) and clinical stage (p < 0.001) not age (p = 0.461), gender (p = 0.335), smoking (p = 0.224). CTNNB1 has been previously observed in the nucleus when CDH1 was deregulated [18, 19]; however, we did not detect CTNNB1 in the nucleus (Fig. 1a, Additional file 2: Figure S1 and Additional file 3: Fig. S2), which was consistent with other reports [20]. This result supported that CTNNB1 did not transfer into the nucleus in SCLC samples. In additional, we analysed CDH1 and CTNNB1 expressions from the gene expression database of SCLC cell lines (https://​sclccelllines.​cancer.​gov/) [21] (Additional file 4: Figure S3), and found that their expressions were not related in SCLC cell lines (r = 0.01671, p = 0.443). This told us that the expression pattern of CDH1 and CTNNB1 was different between SCLC tissues and SCLC cell lines.

CLDN1 and CDH2 were not detected in any SCLC samples (0/60) (Fig. 1a). VIM was observed in all stromal cells (Fig. 1a) and 25% (15/60) tumour cells (Fig. 1b), and its expression was significantly related with tumour differentiation (p < 0.001) [22].

Our previous paper reported that E2F1 was highly expressed and was an independent and adverse prognostic factor for SCLC. We also found that E2F1 could directly regulate the expressions of RELA, MMPs and ADAM12 [2, 3, 9]. NF-κB, MMPs and ADAMs were found to be closely associated with EMT [23], indicating that E2F1 might promote EMT in SCLC.

In our previous research, we found that the cellular morphology was changed when E2F1 gene was silenced by specific siRNAs in SCLC. To further study cellular morphology, stable E2F1 gene knockdown cells were generated using an adenovirus containing shRNA against E2F1 in H446 cells (H446-E2F1sh cells). Eight clones were randomly tested for E2F1 expression. E2F1 was efficiently knocked down in clone-4 (Fig. 2a and Additional file 5: Figure S4), and clone-4 was then used to explore the relationship between E2F1 and EMT. The cellular morphology was significantly changed in clone-4 compared to negative control cells. When E2F1 was knocked down, cells changed from being grain-shaped to appearing slender and fibrous (Fig. 2b). The same change was observed when E2F1 was silenced using siRNA against E2F1 in H1688 cells (Fig. 2a, c and Additional file 5: Figure S4). Combined with our previous results [2, 3], we considered that E2F1 might regulate EMT to promote invasion and metastasis in SCLC.

Because a depletion of E2F1 changed the cell morphology, the expression of cytoskeletal proteins was tested. The expression levels of α-tubulin and β-actin were examined in H446 and H1688 cells by western blot and immunofluorescence. When E2F1 was knocked down in H446 cells, α-tubulin and β-actin were decreased (Fig. 3a and b). The same results were observed in H1688 cells when E2F1 was silenced by E2F1 specific siRNA (Fig. 3a and b). These results coincide with those shown in Fig. 2, indicating that E2F1 affects cytoskeletal protein expression in SCLC, further implying that the role of E2F1 is important in the process of EMT.

The E2F family has 8 members, from E2F1 to E2F8. Although they have similar DNA binding domains, the target genes controlled by E2F family members are different [24]. When E2F1 was stably knocked down in H446 cells, other E2F family members could have compensated for E2F1 function. We therefore tested the expression of other E2F family members with qPCR in H446 and H1688 cells. The results showed that the E2F1 expression was significantly knocked down (p = 0.0007), and the expressions of E2F2, E2F3, E2F5 and E2F8 mRNA was up-regulated, but not to statistically significant levels in H446-E2F1sh cells (Fig. 4). In H1688-E2F1si cells, E2F1 expression was significantly silenced (p = 0.00026) and the mRNA levels of other E2F family members were almost no changed (Fig. 4). These results showed that E2F family members did not compensate for E2F1 when E2F1 gene was stably knocked out in SCLC cell line.

A depletion of E2F1 changed the expression of cytoskeletal proteins, and other members of E2F1 family did not compensate E2F1 function. Next, epithelial and mesenchymal markers (CDH1, CTNNB1 and VIM, CDH2) were examined in SCLC cell lines. Real-time PCR showed that CTNNB1 was significantly increased, while VIM and CDH2 were significantly decreased in H446-E2F1sh cells. Additionally, CDH1 was decreased, but this change was not statistically significant (Fig. 5a). The protein levels of these EMT markers were consistent with the mRNA levels, with the exception of CDH1 (Fig. 5b, c and Additional file 6: Figure S5A). In H446-E2F1sh cells, we used two different CDH1 antibodies, but CDH1 was not detected by western blot and immunofluorescence. Although CDH1 was highly expressed in SCLC tissue samples (Fig. 1a), it was not present in H446 cells. Therefore, we selected the other SCLC cell line, H1688, to test for CDH1 expression. When E2F1 was transiently silenced by siRNAs in H1688 cells, CDH1 expression was significantly increased at both the mRNA and protein levels (Fig. 5d, Additional file 6: Figure S5B). To further certify that E2F1 could affect the expressions of EMT markers, we selected the lower E2F1 cell line A549 to verify this result. Compared with H1688 and H446, E2F1 expression was lower in A549 cells [2]. Next, E2F1 was transfected into A549 cells (named A549-E2F1). In A549-E2F1 cells, we found that CDH1 and CTNNB1 were decreased, and VIM and CDH2 were increased in mRNA and protein levels (Additional file 7: Figure S6). These results further indicated that E2F1 could affect the expressions of EMT related markers, and promote EMT occurrence.

The above results showed that E2F1 promotes EMT, but the mechanism by which it does so is unknown. Previous ChIP-seq studies that examined the target genes of E2F1 in SCLC found that CDH1, CTNNB1, VIM and CDH2 were not directly regulated by E2F1 [2, 25], indicating that E2F1 does not directly control EMT. In addition, ChIP-seq data showed that only ZEB2 not other known EMT transcription factors was a target of E2F1. We then investigated the expression of CDH1 inhibitory transcription factors ZEB1, ZEB2, SNAI1 and SNAI2 by immunohistochemistry. The results showed that ZEB2 was the highest among transcriptional repressors of CDH1 (Fig. 6). ZEB2 was found to be strong positive (score between 100 and 300) in 76.67% (46/60) SCLC tissue samples, moderate positive (score between 20 and 100) in 5% (3/60), weak expression (score between 10 and 20) in 6.67% (4/60), and negative expression (score 0) in 11.67% (7/60, Additional file 8: Figure S7). ZEB1 was mainly localized to mesenchymal cells (60/60), not tumour cells (0/60). SNAI1 was mainly localized to the vascular endothelium (55/60). SNAI2 was very weak and only found in a small number of tumour cells (2/60) (Fig. 6). These results were consistent with other reports [15, 18], and indicated that ZEB2 might play an important role in promoting EMT in SCLC.

Although ZEB1 and SNAI1 have been found to promote invasion and metastasis in SCLC [26, 27], ZEB1 and SNAI1 was very low in SCLC tissue. We analysed the expression of ZEB1, ZEB2, SNAI1 and SNAI2 in H446-E2F1sh and H1688-E2F1si cells by real-time PCR. ZEB2 mRNA was reduced by 60% and 79%, respectively in H446-E2F1sh and H1688-E2F1si cells, and ZEB2 protein was significantly decreased (Fig. 7a). These results suggest that E2F1 might promote EMT in SCLC by regulating ZEB2 expression. To test this hypothesis, we found that E2F1 was higher in these SCLC tissues where ZEB2 was strong positive (Fig. 7 b). Our ChIP-seq data (Additional file 9) [2] showed that ZEB2 was the only target EMT known transcription repressor factor regulated by E2F1. To further certify that E2F1 could regulate ZEB2 expression in SCLC, we constructed the dual luciferase reporter vectors containing ZEB2 promoter. After transfection, luciferase activity analysis showed that E2F1 could regulate ZEB2 expression in H446 and H1688 cells (Fig. 7c). These results further certified our hypothesis that E2F1 promotes EMT by regulating ZEB2 in SCLC.