A feedback loop between GATA2-AS1 and GATA2 promotes colorectal cancer cell proliferation, invasion, epithelial-mesenchymal transition and stemness via recruiting DDX3X

CRC cell lines used in this research included DiFi, SW620, DLD-1, HT-29, SW480 and HCT116. DiFi cell line and human normal colonic epithelial cells (HCoEpiC) were supplied by Shanghai Qincheng Biological Technology Co., Ltd. (Shanghai, China); the others were provided by ATCC (Manassas, VA). SW620 and SW480 cell lines were maintained in Leibovitz's L-15 Medium, HT-29 and HCT116 cell lines in McCoy's 5A Medium, and DiFi, DLD-1 and HCoEpiC cells in RPMI-1640 Medium. SW620, HT-29, HCT116 and DLD-1 cells were cultured in the medium containing 10% FBS with the supplementation of 1% penicillin/streptomycin in a humid incubator at 37 °C with 5% CO₂.

TRIzol Reagent (Invitrogen; Carlsbad, CA) was used for RNA isolation. Then, the extracted RNA was subjected to reverse transcription for cDNA synthesis using PrimeScript RT Reagent Kit (Takara Bio, Japan). SYBR^® Premix Ex Taq™ II (Takara Bio) was used for qPCR. Meanwhile, GAPDH was used as endogenous control. The calculation of gene expression level was based on the 2^−∆∆Ct method [23] and experimental data were displayed as mean ± standard deviation (SD). The experiment was repeated three times. Primer sequences were reported in Table 1.

For stably silencing GATA2-AS1, GATA2 or DDX3X expression, the specific short hairpin RNAs (shRNAs) to GATA2-AS1 or GATA2 or DDX3X and shRNA of negative control (sh-NC) were designed and supplied by GenePharma (Shanghai, China). The full-length cDNA sequence of GATA2 was inserted into the pcDNA3.1 vectors (Invitrogen), using empty vectors as control. These vectors were transfected into cells for 48 h using Lipofectamine 3000 (Invitrogen).

CRC cells (600 cells each well) were plated in 6-well plates and incubated for 12 days at 37 °C. Subsequently, cells were washed with phosphate buffered saline (PBS), and fixed by 4% paraformaldehyde for 15 min, followed by staining with 0.5% violet crystal for 10 min and colony counting by the manual method. The experiment was performed in triplicate.

CRC cells were plated in 24-well plates. We added 10 μM EdU into each well for incubation. Then, cells were fixed, washed, and added with Click-iT EdU Kit. DAPI was used for nuclear counterstain. Images were observed using a fluorescence microscope (Olympus). The experiment was carried out in triplicate.

Flow cytometry analysis was done for measuring the apoptosis of CRC cells. Transfected cells were collected, washed and then re-suspended in binding buffer (50 mL). Next, cell suspensions were added with staining solution containing Annexin V-FITC (1/500, BioVision, Milpitas, CA) and PI (1/500, Beyotime), followed by analysis utilizing a BD Biosciences FACSCalibur™ Flow Cytometer (San Diego, CA). The experiment was performed in triplicate.

Transfected cells were washed, fixed and then permeabilized, followed by TUNEL assays utilizing In Situ Cell Death Detection Kit (Roche) based on the supplier’s protocols. Briefly, cells were incubated in terminal dexynucleotidyl transferase (TdT) reaction cocktail (2 μL), and then treated with Click-iT reaction cocktail (50 μL). DAPI (1/2000) was used to counterstain the nuclei. The experiment was performed in triplicate.

Besides, apoptosis of mouse xenografts was evaluated by tissue TUNEL assay, which was performed as per the instructions for the TUNEL assay kit (KeyGen, Nanjing, China).

Cells (2 × 10⁵) were put into the upper chamber using an 8-mm pore size of transwell coated with Matrigel (50 μg/well, BD Biosciences). The medium with no serum was supplemented to the upper chamber. Complete medium (500 μL) with 20% FBS was utilized to treat the lower chamber. Subsequent to 24 h of incubation, we slightly wiped the cells on the surface of the upper membrane. The invaded cells into the lower chamber were subjected to 10-min fixation in 4% paraformaldehyde (PFA) and 30-min staining by crystal violet (500 μL). The invaded cells in 5 randomly selected fields were observed and imaged using an inverted microscope. The experiment was performed in triplicate.

Proteins were extracted utilizing RIPA lysis buffer (Thermo Fisher Scientific) and quantified by a BCA Protein Assay Kit (Abcam, Cambridge, MA). Subsequently, 1% SDS-PAGE was utilized for protein separation. Then, protein samples were transferred onto PVDF membranes (Millipore), which were sealed with 5% defatted milk in TBST, followed by overnight incubation with primary antibodies (Abcam) at 4 °C. In this assay, we used primary antibodies as follows: Anti-GATA2 (1/1000), Anti-E-cadherin (1/1000), Anti-N-cadherin (1/1000), Anti-Vimentin (1/1000), Anti-Nanog (1/1000), Anti-OCT4 (1/1000), Anti-DDX3X (1/1000), Anti-Ki-67 (1/1000), Anti-PCNA (1 µg/ml) and Anti-GAPDH (1/1000). Next, the membranes were subjected to 1 h of incubation with secondary antibodies (Abcam) at room temperature. Lastly, protein bands were visualized using ECL western blotting substrate (Invitrogen). The experiment was performed in triplicate.

Post transfection, CRC cells were plated on the Corning ultra-low attachment plates (Corning, NY) in medium with no serum, added with 20 ng/mL EGF, 20 ng/mL FGF, 4 mg/mL heparin and 2% B27 (Invitrogen) for 14 days of incubation, followed by analysis of the number and size of spheres. The experiment was performed in triplicate.

Nuclear and cytoplasmic fractions of cells were separated using PARIS kit (Life Technologies, Thermo Fisher Scientific) in the light of supplier’s requirements. The expression level of GATA2-AS1 in nuclear or cytoplasmic fraction was determined by RT-qPCR analysis. In this assay, U6 or GAPDH served as a positive control for nuclear/cytoplasmic fraction. The experiment was done in triplicate.

To determine the subcellular distribution of GATA2-AS1 in CRC cells, FISH assay was carried out utilizing FISH Tag™ RNA Red Kit (F32952, Invitrogen) as per the user guide. Cells were fixed and washed, followed by incubation with FISH probe specific for GATA2-AS1 (5 μL; RiboBio, Guangzhou, China) in hybridization buffer. Nuclear counterstain was done utilizing DAPI. The slides were observed with a fluorescence microscope. The experiment was done in triplicate.

ActD (Abcam) at a final concentration of 4 μM was used to treat cells. ActD was added to cells at 0 h, 4 h, or 8 h. RT-qPCR was used to quantify mRNA levels. The experiment was done in triplicate.

RNA pull down assay was conducted utilizing Pierce Magnetic RNA–Protein Pull-Down Kit (Thermo Fisher Scientific, Waltham, MA) as per the supplier’s protocols. The protein lysates were mixed with GATA2-AS1 or GATA2-AS1 AS or GATA2 or GATA2 AS, followed by addition of magnetic beads. RNA–protein mixture was analyzed using western blot. The experiment was carried out in triplicate.

Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, Bedford, MA) was utilized for RIP conforming to the supplier’s protocols. Cells were subjected to lysis with RIP lysis buffer. Subsequently, cell lysates were subjected to incubation with RIPA buffer containing magnetic beads conjugated with anti-DDX3X antibody or anti-IgG. The RNA extracted from the immunoprecipitates was purified for RT-qPCR analysis. The experiment was implemented in triplicate.

ChIP assay was conducted via a ChIP Assay Kit (Beyotime) following the guidelines of the supplier. Cells were cross-linked with 1% PFA for 10 min and then sonicated into 200–1000 bp fragments. Next, chromatin was immunoprecipitated with anti-GATA2 or IgG antibodies, followed by RT-qPCR. Each experiment was independently performed in triplicate.

The sequence of GATA2-AS1 promoter was amplified from CRC cells. Then, the fragments of Full length, Site 1-MUT, Site 2-MUT and Site 1 + 2 MUT were inserted into the pGL3 luciferase vector (Promega, Madison, WI) to construct reporter vectors. Then, these vectors were co-transfected with pcDNA3.1/GATA2 or sh-GATA2#1/2 into cells. After 48 h of transfection, luciferase activities were determined using Luciferase Assay Kit (Promega). The experiment was performed in triplicate.

A total of 20 BALB/c nude mice (five-week-old, male) were selected and randomly divided into 4 groups (5 mice for each group), two test groups and two control groups. Each mouse in test groups was injected with stably sh-GATA2-AS1#1-transfected HCT116 or SW480 cells (5 × 10⁶) while same amount of stably sh-NC HCT116 or SW480 cells were injected into each mouse in control groups. We used vernier caliper to measure the volume of xenografts in mice every 4 days. Finally, mice were sacrificed and then tumor tissues were extracted and used for subsequent analyses. Xenografted tumors were weighed by electronic scale. Animal study was approved by the Ethics Committee of the Third Xiangya Hospital of Central South University (2019-S336). Three independent assays were requested.

IHC staining was performed on 4-μm thick paraffin-embedded sections. Slides were dewaxed and antigen retrieval, and then incubated with primary antibodies (Abcam) including Anti-Ki-67 (0.1–5 µg/mL), Anti-PCNA (1/10,000–1/30,000), Anti-E-cadherin (1/500), Anti-N-cadherin (1 µg/mL), Anti-Nanog (1/100–1/250) and Anti-OCT4 (1/1000) overnight. Afterwards, secondary antibodies (1/200–1/2000, Abcam) were added for incubation at room temperature. Next, DAB Stain Kit (ZSBio, Beijing, China) was treated into slides and then counterstained by DAPI. The experiment was done in triplicate.

GEPIA (http://​gepia2.​cancer-pku.​cn) was used to analyze GATA2-AS1 expression in colon adenocarcinoma (COAD) samples (n = 275) and normal samples (n = 349). RNA-seq of GATA2-AS1 in 27 different normal tissues was analyzed by NCBI database (https://​www.​ncbi.​nlm.​nih.​gov). Potential miRNAs of either GATA2-AS1 or GATA2 were predicted by ENCORI (https://​rna.​sysu.​edu.​cn/​encori/​). In addition, UCSC database (http://​genome.​ucsc.​edu/​) was used to predict transcription factors of GATA2-AS1. JASPAR (http://​jaspar.​genereg.​net/​) was used to predict GATA2-binding sites on the sequence of GATA2-AS1 promoter.

To explore the expression profile of GATA2-AS1 in CRC, we used GEPIA2 (http://​gepia2.​cancer-pku.​cn) to analyze GATA2-AS1 expression in colon adenocarcinoma (COAD). Relative to 349 cases of normal tissues, GATA2-AS1 was significantly highly expressed in 275 cases of COAD tissues (Fig. 1A). GATA2-AS1 was lowexpressed in normal colon tissues according to the NCBI database (https://​www.​ncbi.​nlm.​nih.​gov) (Additional file 1: Fig. S1). The above data confirmed the aberrant elevation of GATA2-AS1 expression in CRC tissues. Moreover, RT-qPCR suggested that GATA2-AS1 expression was markedly increased in six CRC cell lines relative to that in human normal HCoEpiC (Fig. 1B). Intriguingly, we found that the expression of GATA2, nearby gene of GATA2-AS1, was also apparently elevated in CRC cell lines versus that in normal cell line (Fig. 1C). In particular, SW480 and HCT116 cell lines exhibited a relatively higher expression of GATA2-AS1 and GATA2 than other CRC cells did, so they were selected for follow-up experiments.

Besides, it has been suggested that some lncRNAs could modulate the expression of nearby genes [24]. Then, we investigated the regulatory relationship of GATA2-AS1 and GATA2 in CRC cells. Before that, two specific shRNAs for GATA2-AS1 were transfected into SW480 and HCT116 cells to silence GATA2-AS1 expression (Additional file 2: Fig. S2A). GATA2 mRNA and protein levels were overtly reduced when GATA2-AS1 was knocked down (Fig. 1D). Likewise, we also silenced GATA2 expression via transfection of sh-GATA2 plasmids (Additional file 2: Fig. S2B). It was found that GATA2 depletion obviously decreased the expression of GATA2-AS1 (Fig. 1E). These results suggested that GATA2-AS1 and GATA2 are expressed in CRC cells at high level and that their both expression is tightly regulated.

Next, we explored the impact of GATA2-AS1 or GATA2 on CRC cell proliferation and apoptosis. It was found that deficiency of GATA2-AS1 or GATA2 markedly hampered the proliferation capacity of CRC cells (Fig. 2A, B). The apoptosis of CRC cells was elevated in GATA2-AS1-silenced or GATA2-silenced SW480 and HCT116 cells (Fig. 2C, D). Altogether, both GATA2-AS1 and GATA2 promote CRC cell proliferation and repress cell apoptosis.

Moreover, we investigated the impacts of GATA2-AS1 and GATA2 on CRC cell invasion, EMT and stemness. Transwell assays showed that the number of invading cells was obviously reduced after the depletion of GATA2-AS1 or GATA2 (Fig. 3A). Besides, we analyzed the protein levels of EMT markers in GATA2-AS1-inhibited or GATA2-inhibited SW480 and HCT116 cells via western blot. Down-regulation of GATA2-AS1 or GATA2 led to an obvious elevation in the protein level of epithelial marker (E-cadherin) and a distinct reduction in those of mesenchymal markers (N-cadherin and Vimentin) (Fig. 3B). In addition, the data from western blot analysis and sphere formation assay disclosed that the protein levels of stem cell markers (Nanog and OCT4) and the number and size of spheres were overtly reduced after GATA2-AS1 or GATA2 knockdown (Fig. 3C, D). All above data indicated that GATA2-AS1 and GATA2 promote CRC cell invasion, EMT and stemness.

Furthermore, we investigated the regulatory mechanism between GATA2-AS1 and GATA2. The subcellular distribution of lncRNA is associated with its mechanism [25]. GATA2-AS1 was verified to be primarily localized in the cytoplasm of CRC cells (Fig. 4A, B), which implied the potential of GATA2-AS1 to post-transcriptionally regulate GATA2 expression. Notably, competing endogenous RNA (ceRNA) is a common post-transcriptional regulation mechanism functioning in the development of cancers [26]. Given that, we next explored miRNAs binding with GATA2-AS1 or GATA2; potential miRNAs were predicted by ENCORI (https://​rna.​sysu.​edu.​cn/​encori/​). However, there was no appropriate miRNA binding to both GATA2-AS1 and GATA2 (Additional file 2: Fig. S2C), which excluded ceRNA regulation mechanism. Subsequently, ActD was used to inhibit protein synthesis in the analysis of GATA2 mRNA stability. The results showed that GATA2-AS1 silencing obviously decreased GATA2 mRNA level in ActD-treated CRC cells (Fig. 4C). LncRNAs could recruit some proteins to stabilize mRNA [27]. Hence we carried out RNA pull down assay to explore potential RBPs of GATA2-AS1. A ~ 73 kDa protein was pulled down by GATA2-AS1 and identified as DDX3X via mass spectrometry (Fig. 4D). RNA pull down and RIP assays further confirmed the interaction between GATA2-AS1 and DDX3X (Fig. 4E, F). Also, it was found that DDX3X protein was pulled down by GATA2 (Fig. 4G); GATA2 was highly enriched in anti-DDX3X bound precipitates (Fig. 4H), indicating the binding between GATA2 and DDX3X. We next investigated the impact of DDX3X on the stability of GATA2 mRNA in CRC cells. We silenced the expression of DDX3X by transfection of sh-DDX3X#1/2 (Additional file 2: Fig. S2D). GATA2 mRNA and protein levels were significantly reduced after DDX3X depletion (Fig. 4I). Therefore, we hypothesized whether GATA2-AS1 regulates DDX3X to affect the expression of GATA2. GATA2-AS1 deletion had no marked effect on DDX3X expression at mRNA and protein levels (Additional file 2: Fig. S2E, F). Whereupon, we speculated that GATA2-AS1 might recruit DDX3X to regulate the stability of GATA2. GATA2 mRNA level was significantly decreased in ActD-treated cells after DDX3X silencing (Fig. 4J), indicating that DDX3X promotes the stability of GATA2 mRNA. All these data revealed that GATA2-AS1 recruits DDX3X to regulate the stability of GATA2 mRNA, affecting the expression of GATA2.

Based on the analysis of UCSC database (http://​genome.​ucsc.​edu/​), we obtained a list of putative transcription factors of GATA2-AS1 (Additional file 3: Fig. S3). Notably, GATA2 was one of the listed transcription factors. Thereby, we probed into the impact of GATA2 as a transcription factor on GATA2-AS1 expression at the transcriptional level. GATA2 was successfully overexpressed via transfection of pcDNA3.1/GATA2 (Fig. 5A). RT-qPCR showed that GATA2-AS1 expression was obviously elevated when GATA2 was overexpressed (Fig. 5B). With the help of JASPAR (http://​jaspar.​genereg.​net/​), we found two potential GATA2-binding sites on the sequence of GATA2-AS1 promoter (Fig. 5C, D). Site 1 or/and Site 2 mutations were performed for subsequent luciferase reporter assays. The luciferase activity of full length GATA2-AS1 promoter (Full length) or mutant GATA2-AS1 promoter (Site 1-MUT and Site 2-MUT) was significantly enhanced by GATA2 overexpression, while that of mutant GATA2-AS1 promoter (Site 1 + 2-MUT) was barely affected (Fig. 5E). Also, GATA2 silencing caused a decrease in the luciferase activity of GATA2-AS1 promoter (Full length) and mutant GATA2-AS1 promoter (Site 1-MUT and Site 2-MUT), but caused no marked change in that of mutant one (Site 1 + 2-MUT) (Fig. 5F). The above data indicated that both Site 1 and Site 2 are GATA2-binding sites in GATA2-AS1 promoter. Additionally, ChIP assay further confirmed the binding between GATA2-AS1 promoter and GATA2, as GATA2-AS1 promoter was overtly abundant in anti-GATA2 bound precipitates (Fig. 5G, H). Taken together, GATA2 transcriptionally activates GATA2-AS1 expression.

Besides in vitro assays, we established xenograft tumor models to verify the role of GATA2-AS1 in CRC tumorigenesis. Stably sh-GATA2-AS1#1 or sh-NC transfected CRC cells were subcutaneously injected into nude mice to construct mouse xenograft models. The tumor volume and tumor weight were significantly declined when GATA2-AS1 was knocked down (Fig. 6A, B; Additional file 4: Fig. S4A, B). Besides, IHC staining revealed that the expression of proliferative markers (Ki-67 and PCNA) was significantly inhibited in sh-GATA2-AS1#1 group compared to sh-NC group; TUNEL staining revealed that GATA2-AS1 inhibition promotes the apoptosis of xenograft tumors (Fig. 6C; Additional file 4: Fig. S4C). Moreover, the RNA levels of GATA2-AS1 and GATA2 were decreased in GATA2-AS1-inhibited group; the protein levels of GATA2, Ki-67 and PCNA were reduced when GATA2-AS1 was silenced (Fig. 6D; Additional file 4: Fig. S4D). Meanwhile, EMT and stemness were also repressed when GATA2-AS1 was knocked down (Fig. 6E; Additional file 4: Fig. S4E). All these data suggested that knockdown of GATA2-AS1 inhibits tumor growth, EMT and stemness in CRC.

Additionally, we performed rescue experiments to validate the impact of GATA2-AS1-GATA2 interaction on CRC cell proliferation and apoptosis. GATA2 overexpression reversed the inhibited proliferation ability of GATA2-AS1-silenced CRC cells (Fig. 6F, G). Besides, the increased apoptosis caused by GATA2-AS1 depletion was counteracted by GATA2 overexpression (Fig. 6H, I). In sum, GATA2-AS1 regulates CRC cell proliferation and apoptosis through the modulation of GATA2 expression.

Moreover, we conducted rescue experiments to verify the impacts of GATA2-AS1-GATA2 interaction on CRC cell invasion, EMT and stemness. As evidenced by transwell assays, GATA2 overexpression countervailed the repression of GATA2-AS1 depletion on CRC cell migration (Fig. 7A). Western blot revealed that the inhibition of GATA2-AS1 silencing on EMT was offset by GATA2 overexpression (Fig. 7B). In addition, the declined protein levels of Nanog and OCT4 and the reduced number and size of spheres in GATA2-AS1-silenced CRC cells were altered by up-regulation of GATA2 (Fig. 7C, D). Collectively, GATA2-AS1 facilitates CRC cell migration, EMT and stemness via up-regulating GATA2 expression.