Regulation of fibroblast growth factor-2 by an endogenous antisense RNA and by argonaute-2

Abstract

We have previously reported that elevated fibroblast growth factor-2 (FGF-2) expression is associated with tumor recurrence and reduced survival after surgical resection of esophageal cancer and that these risks are reduced in tumors coexpressing an endogenous antisense (FGF-AS) RNA. In the present study, we examined the role of the endogenous FGF-AS transcript in the regulation of FGF-2 expression in the human lung adenocarcinoma cell line Seg-1. FGF-2 and FGF-AS were temporally and spatially colocalized in the cytoplasm of individual cells, and knockdown of either FGF-2 or FGF-AS by target-specific siRNAs resulted in dose-dependent up-regulation of the complementary transcript and its encoded protein product. Using a luciferase reporter system, we show that these effects are mediated by interaction of the endogenous antisense RNA with the 3'-untranslated region of the FGF-2 mRNA. Deletion mapping identified a 392-nucleotide sequence in the 5823-nucleotide FGF-2 untranslated tail that is targeted by FGF-AS. Small interfering RNA-mediated knockdown of either FGF-AS or FGF-2 significantly increased the stability of the complementary partner mRNA, demonstrating that these mRNAs are mutually regulatory. Knockdown of FGF-AS also resulted in reduced expression of argonaute-2 (AGO-2) and a number of other elements of the endogenous micro-RNA/RNA interference pathways. Conversely, small interfering RNA-mediated knockdown of AGO-2 significantly increased the stability of the FGF-2 mRNA transcript and the steady-state levels of both FGF-2 mRNA and protein, suggesting a role for AGO-2 in the regulation of FGF-2 expression.

Fig. 1.

Schematic representation of the human FGF-2 and FGF-AS mRNA transcripts. The solid boxes in each transcript denote the open reading frames, and the arrows denote the direction of transcription. The FGF-AS mRNA is completely complementary over a span of 639 nucleotides (nts) to two widely separated regions (nucleotides 1183-1765 and 6160-6215) in the long 3′-UTR of the FGF-2 mRNA. The approximate target sites of the FGF-2 and FGF-AS siRNAs used in the present study are indicated.

Fig. 2.

FGF-2 and FGF-AS are temporally and spatially coexpressed. A, Coexpression of FGF-2 and FGF-AS in a single Seg-1 cell was confirmed by RT-PCR. Lane 1, FGF-2 (222 bp); lane 2, FGF-AS (125 bp); lane 3, E. coli carrier RNA; lane 4, no enzyme control (NEC); lane 5, no template control (NTC). The results shown here are representative of 10 individual cells analyzed. B, FGF-2 and FGF-AS are not segregated in nuclear and cytoplasmic compartments. Cytoplasmic and nuclear expression of FGF-2, FGF-AS, HIF-S, and HIF-AS were determined by semiQ RT-PCR of RNA extracted from isolated cytoplasmic and nuclear fractions. Relative mRNA expression was normalized to β-actin and expressed as the cytoplasmic/nuclear ratio. Results are expressed as the mean ± sem of three independent samples. ns, Not significant.

Fig. 3.

FGF-AS and FGF-2 siRNA knockdown reveals an inverse relationship between FGF-2 and FGF-AS mRNA and protein abundance. Cells were reverse transfected with the indicated concentrations of FGF-AS siRNA (A) or FGF-2 siRNA (B) for 72 h and mRNA expression determined by semiQ RT-PCR. Values were normalized to β-actin expression and are expressed as a percentage of the maximum mRNA expression observed for each gene of interest. Results are the mean ± sem of three independent transfections. Western blot analysis (C and D, upper panels) confirmed the knockdown of immunoreactive FGF-2 and up-regulation of NUDT6 by FGF-2 siRNA (C) and conversely knockdown of NUDT6 and up-regulation of FGF-2 by FGF-AS siRNA (D). Protein levels were normalized against tubulin expression in the same samples and expressed as a percentage of negative siRNA control (C and D, lower panels). Results are the mean ± sem of three independent transfections. *, P < 0.05; **, P < 0.01; ***, P < 0.001 vs. control.

Fig. 4.

FGF-AS regulates FGF-2 expression via the 3′-UTR of the FGF-2 mRNA. A, Schematic representation of the chimeric pCMV-LUC-FGF 3′-UTR luciferase reporter construct. The vector expresses a chimeric luciferase mRNA containing a 700-bp fragment of the 3′-UTR of the FGF-2 mRNA, including a 587-nucleotide sequence that is fully complementary to the FGF-AS mRNA. B, Increased luciferase activity in transfected Seg-1 cells after siRNA-mediated knockdown of endogenous FGF-AS. Cells were cotransfected with the pCMV-LUC-FGF 3′-UTR luciferase reporter construct and either negative control siRNA or FGF-AS siRNA for 72 h before processing for luciferase assay. Relative light units (RLU) were expressed as a fraction of controls. Results are the means ± sem of 10 samples. ***, P < 0.001 vs. control.

Fig. 5.

Deletion mapping of the FGF-2 3′-UTR. A, Diagram of various deletion constructs. The red boxes on FGF-2 mRNA denote overlap regions with FGF-AS. The dashed line denotes a long stretch of sequence that has no overlap with FGF-AS. Arrows in the constructs denote transcript orientation compared with FGF-2 sense transcript. Gray boxes denote fluorescent proteins (either GFP or dsRed). B, At 48 h after transfection, cell lysates were run on SDS-PAGE and blotted for GFP and tubulin. Unlabeled lanes contained lysates from nontransfected Cos-7 cells.

Fig. 6.

FGF-2 and FGF-AS mRNA are mutually regulatory. Cells were reverse transfected with 100 nm of either negative control siRNA (○, A and B), FGF-AS siRNA (•, A), or FGF-2 siRNA (•, B) for 72 h and then treated with the transcription inhibitor actinomycin D for 12, 18, and 24 h. RNA was harvested at the indicated times after actinomycin D treatments, and the amount of FGF-2 (A) or FGF-AS mRNA (B) remaining was determined by semiQ RT-PCR. The results were normalized against the level of β-actin mRNA in the same samples and plotted against time. Control (▵, A and B) indicates untransfected cells in the absence of actinomycin D. Results are the mean ± sem of four independent transfections. Asterisks denote values significantly different from untransfected controls without actinomycin D: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Fig. 7.

SemiQ RT-PCR validation of microarray hits after FGF-AS knockdown. siRNA knockdown of FGF-AS results in significantly reduced expression of FGF-AS, AGO-2, and PACT mRNAs and significantly increased expression of FGF-2 and SKP-2 mRNAs. Cells were reverse transfected with 100 nm negative control siRNA (white bars) or FGF-AS siRNA (gray bars) for 72 h. Results are the mean ± sem of four independent transfections. *, P < 0.05; **, P < 0.01; ***, P < 0.001 vs. control.

Fig. 8.

Effect of AGO-2 siRNA knockdown on FGF-2 expression. A, Cells were reverse transfected with 200 nm AGO-2 siRNA (+) or negative control siRNA (−) for 72 h before determination of steady-state mRNA levels of β-actin, AGO-2, or FGF-2. B, Western blot confirmation of AGO-2 knockdown at the protein level. Bar graph results in the lower panel are normalized against the level of tubulin in the same samples. **, P < 0.01; ***, P < 0.001. C, AGO-2 knockdown up-regulates FGF-2 protein levels. Cells were transfected with control siRNA or 100 nm siRNA against AGO-2 or FGF-2 for 72 h before processing for Western blot with anti-FGF-2 and antitubulin antibodies. FGF-2 levels in the lower panel are normalized against the level of tubulin in the same samples. D, Decay after transcriptional inhibition with actinomycin D, as described in the legend to Fig. 6 (○, control siRNA plus actinomycin D; •, AGO-2 siRNA plus actinomycin D; ▵, no siRNA and no actinomycin D). mRNA abundance was normalized to the level of β-actin mRNA in each sample. Results are the mean ± sem of four independent transfections. Asterisks denote values significantly different from untransfected controls without actinomycin D: *, P < 0.05; **, P < 0.01; ***, P < 0.001.