Distinct 3'UTRs differentially regulate activity-dependent translation of brain-derived neurotrophic factor (BDNF)

Abstract

Expression of the brain-derived neurotrophic factor (BDNF) is under tight regulation to accommodate its intricate roles in controlling brain function. Transcription of BDNF initiates from multiple promoters in response to distinct stimulation cues. However, regardless which promoter is used, all BDNF transcripts are processed at two alternative polyadenylation sites, generating two pools of mRNAs that carry either a long or a short 3'UTR, both encoding the same BDNF protein. Whether and how the two distinct 3'UTRs may differentially regulate BDNF translation in response to neuronal activity changes is an intriguing and challenging question. We report here that the long BDNF 3'UTR is a bona fide cis-acting translation suppressor at rest whereas the short 3'UTR mediates active translation to maintain basal levels of BDNF protein production. Upon neuronal activation, the long BDNF 3'UTR, but not the short 3'UTR, imparts rapid and robust activation of translation from a reporter. Importantly, the endogenous long 3'UTR BDNF mRNA specifically undergoes markedly enhanced polyribosome association in the hippocampus in response to pilocarpine induced-seizure before transcriptional up-regulation of BDNF. Furthermore, BDNF protein level is quickly increased in the hippocampus upon seizure-induced neuronal activation, accompanied by a robust activation of the tropomyosin-related receptor tyrosine kinase B. These observations reveal a mechanism for activity-dependent control of BDNF translation and tropomyosin-related receptor tyrosine kinase B signaling in brain neurons.

Fig. 1.

The BDNF long 3′UTR is a bona fide cis-acting translation suppressor. (A) Schematic of luciferase reporter constructs harboring full-length BDNF 3′UTR, short BDNF 3′UTR, or a deletion mutant that produces only BDNF long 3′UTR. Asterisks indicate polyadenylation sites. (B) BDNF long 3′UTR suppresses reporter translation in transfected CAD cells, an immortalized neuronal cell line derived from mouse brain. Top: RT-PCR demonstrates the expression of luciferase reporter mRNAs that carry the BGH 3′UTR, the BDNF full, long, or short 3′UTRs using primers indicated on the left. − and + indicate the absence and presence of reverse transcriptase. Bottom: Luciferase activity normalized to qRT-PCR reading of corresponding reporter mRNAs (one-way ANOVA, P < 0.05, n = 4; *P < 0.05, Tukey analysis). Error bars indicate SEM. (C) The long and short BDNF 3′UTRs differentially mediate luciferase translation when expressed in primary cultured hippocampal neurons (DIV3). Significantly lower luciferase activity was detected from cells transfected with the BDNF long 3′UTR construct compared with that from the BDNF short 3′UTR construct (*P < 0.05, Student t test, n = 6).

Fig. 2.

Endogenous long 3′UTR BDNF mRNA in mouse hippocampus was translationally repressed. (A) Cytoplasmic extracts from mouse hippocampi were prepared in the presence of MgCl₂ for linear sucrose gradient fractionation (15–45%) to separate translating polyribosomes (fraction 4–10) from nontranslating components including dormant mRNPs, ribosome subunits, and monoribosomes (fraction 1–3), monitored by absorption at OD254 (Top). qRT-PCR was performed to determine the distribution of the pan-BDNF mRNA, the long 3′UTR-BDNF mRNA, and the housekeeping GAPDH mRNA in each gradient fraction (Middle). Phosphorus 32–labeled semiquantitative RT-PCR products of the aforementioned mRNAs were visualized by a PhosphorImager (Bottom). (B) Hippocampal lysates were treated with EDTA to release mRNAs from polyribosomes to ribosome-free mRNPs. Note the disappearance of mono- and polyribosomes, and the accumulation of released 40S and 60S ribosome subunits on the linear sucrose gradient (Top). qRT-PCR for the long 3′UTR BDNF mRNA (Middle) and ³²P-labeled semiquantitative RT-PCR products (Bottom) for the long 3′UTR BDNF mRNA, the pan BDNF mRNA, and GAPDH mRNA were indicated respectively.

Fig. 3.

BDNF long 3′UTR sufficiently mediates activity-stimulated reporter translation. Laser confocal live cell imaging of cultured hippocampal neurons (DIV21) expressing d2EGFP fused with the SV40 3′UTR (A and B) or the short (C and D) or the long BDNF 3′UTR (E and F) before and after 10 min TEA exposure. (Scale bar: 40 μm.) Dotted boxes are magnified (2×) and shown as insets. The pseudocolor bar indicates fluorescent intensities. (G and H) Quantification of the changes in d2EGFP reporter fluorescence caused by TEA. A randomly selected neuron was imaged from each coverslip by z-sectioning through the entire cell (10–15 optical slices) and projected into a 2D image (maximum intensity) before and after TEA exposure. The total integrated fluorescent density in dendrites (G) and soma (H) of the same cell before and after 10 min TEA exposure was measured. The percent change of fluorescence was calculated by normalizing the integrated intensity after TEA treatment against that of the same cell before TEA treatment, and results were displayed graphically. The decrease of fluorescence for reporters that carry the BDNF short 3′UTR and the SV40 3′UTR may reflect photobleaching during confocal imaging. One-way ANOVA analysis: P < 0.05 for both soma and dendrites (n = 4). Error bars indicate SEM. **P < 0.01, Tukey analysis.

Fig. 4.

Pilocarpine-induced neuronal activation results in robust translational derepression of BDNF specifically mediated by the long 3′UTR. (A) Levels of pan-BDNF mRNA and long 3′UTR BDNF mRNA by qRT-PCR in rat hippocampus after pilocarpine-induced SE at the indicated time points. Basal levels of corresponding BDNF mRNAs normalized to that of the GAPDH mRNA were set at 100% (dotted line). qRT-PCR of long 3′UTR BDNF mRNA (B) and pan-BDNF mRNA (C) in sucrose gradient fractions of hippocampal extracts from control (■) or pilocarpine (Pilo)–treated rats (△) after 30 min of SE. P < 0.05, two-way ANOVA; *P < 0.05 Bonferroni posttest. (D) Semiquantitative RT-PCR of the 5′UTRs in BDNF mRNA indicates that exon IV and VI are the major 5′UTRs expressed in the adult rat hippocampus. Polyribosomal profiles of BDNF mRNAs carrying exon IV (E) and exon VI (F) determined by linear sucrose gradient fractionation qRT-PCR assay using hippocampal lysates from control or Pilo-treated rats after 30 min of SE. Error bars indicate SEM.

Fig. 5.

Increased BDNF protein and TrkB activation upon seizure-induced neuronal activation in the hippocampus associated with activity-dependent translation of the long 3′UTR BDNF mRNA. (A) Immunofluorescence of BDNF (green) is increased in hippocampal MF tract 30 min after pilocarpine-induced SE (arrows, Left Lower). The increase of BDNF signal is more clearly observed in the CA3 strata lucidum and radiatum under higher magnification (Center), which is quantified and graphically displayed (Right). Error bars indicate SEM; **P < 0.01 Student t test (n = 3). (B) pY816-TrkB immunofluorescence (green) is drastically increased in hippocampal MFs 30 min after pilocarpine-induced SE (arrows, Right). For all images, the nuclear staining by DAPI (blue) marks the principle neuron layers.