Epidermal growth factor increases the interaction between nucleolin and heterogeneous nuclear ribonucleoprotein K/poly(C) binding protein 1 complex to regulate the gastrin mRNA turnover

Abstract

Gastrin, a gastrointestinal hormone responsible for gastric acid secretion, has been confirmed as a growth factor for gastrointestinal tract malignancies. High expression of gastrin mRNA was observed in pancreatic and colorectal cancer; however, the mechanism is unclear. Epidermal growth factor (EGF) was found to increase gastrin mRNA stability, indicating mRNA turnover regulation mechanism is involved in the control of gastrin mRNA expression. Using biotin-labeled RNA probe pull-down assay combined with mass spectrometry analysis, we identified the heterogeneous nuclear ribonucleoprotein K (hnRNP K) and poly(C) binding protein 1 (PCBP1) bound with the C-rich region in gastrin mRNA 3' untranslated region. Nucleolin bound with the AGCCCU motif and interacted with hnRNP K were also demonstrated. Under EGF treatment, we observed the amount of nucleolin interacting with hnRNP K and gastrin mRNA increased. Using small interfering RNA technology to define their functional roles, we found hnRNP K, PCBP1, and nucleolin were all responsible for stabilizing gastrin mRNA. Moreover, nucleolin plays a crucial role in mediating the increased gastrin mRNA stability induced by EGF signaling. Besides, we also observed hnRNP K/PCBP1 complex bound with the C-rich region in the gastrin mRNA increased nucleolin binding with gastrin mRNA. Finally, a novel binding model was proposed.

Figure 1.

EGF induces the expression and increased the gastrin mRNA stability in AGS cells. (A) EGF up-regulates the gastrin mRNA expression in a time-dependent manner. The AGS cells were plated at ∼40% confluence, and then the incubation medium was changed to serum starvation medium after 20 h. The cells were starved for 24 h and treated with EGF for 4, 8, and 16 h. At different time points, RNA was isolated and analyzed by RT-PCR. (B) Quantitative result of gastrin mRNA expression levels under EGF treatment. The -fold represents the mRNA expression level of EGF-treated cell divided by control cell. (C) EGF increases gastrin mRNA stability. AGS cells treated with or without 10 nM EGF for 4 h before the addition of 5 μg/ml actinomycin D. Total RNA was isolated, and gastrin mRNA level was determined using quantitative RT-PCR and normalized with the mRNA level of GAPDH. Quantitative results were plotted as a percentage of total gastrin mRNA at different time points compared with time point 0 h. Values are the means ± SE from three separate experiments.

Figure 2.

Characterization of the protein complexes interacted with gastrin mRNA 3′UTR. (A) The diagram illustrates the structure of human gastrin mRNA. (B) The structure of human gastrin 3′UTR was predicted by Vienna RNA Secondary Structure Prediction program (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi). (C) Sequences of three biotinylated RNA probes (probe A, B, and C) and mutant counterpart probes (probe D and F) used in RNA-EMSA. (D) Biotin-labeled probes A, B, and C were incubated with or without cytosolic extracts of AGS cells, and RNA-EMSA was performed. For competition assay, the reaction mixtures were added with RNA homopolymers [poly(rA), poly(rC), and poly(rU), respectively]. (E) Nonbiotinylated probe B (cold probe B) was added to reaction mixture as a competitor to examine the protein complex existing in probe B and C. (F) No protein complex was present in probe D or F (C-rich regions mutants of wild-type probes C and B, respectively).

Figure 3.

Identification of the RNA-binding proteins interacted with the gastrin mRNA. (A) Biotin-labeled probes A, B, and C were incubated with cytosolic extracts and pulled down by streptavidin beads. The mixtures were analyzed by SDS-PAGE combined with silver staining. (B) hnRNP K and PCBP1 were identified as major binding proteins interacted with gastrin mRNA 3′UTR by mass spectrometry analysis. (C) hnRNP K and PCBP1 interacted with probes B and C in vitro. The UV-cross-linking/biotin pull-down analysis and Western blot were performed with hnRNP K and PCBP1 antibodies. Beads represent the streptavidin beads control. (D) hnRNP K and PCBP1 bound to gastrin mRNA in vivo. AGS cells were treated with or without 10 nM EGF for 1 and 4 h, and then the cytoplasm proteins were harvested and incubated with anti-hnRNP K, anti-PCBP1, or anti-actin antibodies. RNA from immunoprecipitated complex was extracted, and the mRNA expression levels of gastrin and actin were measured by RT-PCR analysis (M, marker; P, positive control; N, negative control).

Figure 4.

Nucleolin binds with gastrin mRNA 3′UTR and hnRNP K. (A) Biotin-labeled RNA probes A, B, and C were incubated with cytosolic extract. The mixtures were analyzed by UV-cross link/biotin pull-down assay, followed by Western blot analysis by using nucleolin antibodies. (B) EGF increases the binding level of nucleolin with gastrin mRNA in vivo. AGS cells were treated with or without 10 nM EGF for 1 and 4 h, and the RNA immunoprecipitation assays were performed with anti-nucleolin or anti-actin antibodies. RNA was extracted and the mRNA expression levels of gastrin and actin were resolved by RT-PCR. (C) Quantitative results of RNA immunoprecipitation assay by using nucleolin antibodies under 10 nM EGF treatment for 0, 1, and 4 h. (D) hnRNP K forms a complex with nucleolin. AGS cell lysates were subjected to immunoprecipitation with anti-hnRNP K and anti-nucleolin antibodies. The immunoprecipitated complexes were treated with or without RNase and analyzed by SDS-PAGE and immunoblotting by using anti-nucleolin and anti-hnRNP K antibodies. The rabbit IgG (RIgG) and mouse IgG (MIgG) were used as negative controls. (E) EGF enhanced the interaction between nucleolin and hnRNP K. AGS cells were treated with or without 10 nM EGF for 4 h, and the protein immunoprecipitation assay was then conducted as described above.

Figure 5.

Identified the nucleolin binding sequence in the gastrin mRNA. (A) The diagram shows the putative nucleolin binding site (baseline). Biotinylated RNA probes were synthesized, including two wild-type probes, B and C, and four mutant counterparts: probe D (mutation of three C-rich sites of probe C), probe D-N (mutation of two C-rich sites of probe C and preservation of putative nucleolin binding site), probe E (the putative nucleolin binding site was mutated), and probe F (mutation of the C-rich region of probe B). (B) RNA probes B, C, D-N, E, and F were incubated with cytosolic extract, respectively. The mixtures were analyzed by UV-cross link/biotin pull-down assay combined with Western blot analysis by using nucleolin and hnRNP K antibodies.

Figure 6.

Silencing of hnRNP K, PCBP1, and nucleolin reduces gastrin mRNA expression level. (A) Three different kinds of siRNAs targeted each of hnRNP K, PCBP1, and nucleolin, respectively, were transiently transfected into AGS cells. The whole cell lysate was harvested, and Western blot analysis was performed using anti-hnRNP K, anti-PCBP1, and anti-nucleolin antibodies. (B) The gastrin mRNA expression levels were measured by quantitative real-time RT-PCR under different siRNA conditions. GAPDH mRNA expression level was used as an internal control for calibration. (C) The EGF inducibility of gastrin mRNA expression was calculated by dividing the quantity of gastrin mRNA expression level with EGF treatment to the quantity without EGF treatment under different siRNA conditions. (D) The AGS cells were transiently transfected with luciferase constructs bearing wild-type gastrin 3′UTR and mutated nucleolin binding site (3′UTRmNucleolin). After 24-h serum starvation, transfected cells were treated with or without 10 nM EGF for 6 h. Luciferase mRNA expression levels were measured by quantity real-time RT-PCR. (E) Tet-Off mRNA turnover assay to study the turnover rate with or without EGF treatment under reducing nucleolin expression. *p < 0.05, **p < 0.01, and ***p < 0.001 vs. si-control.

Figure 7.

Binding model of nucleolin interacted with gastrin mRNA. hnRNP K and PCBP1 bind with the C-rich regions of gastrin mRNA to form a hairpin-like structure. The putative nucleolin binding site is exposed in the loop region, and this structure has higher binding affinity than the linear structure with nucleolin.