Dual regulation of miRNA biogenesis generates target specificity in neurotrophin-induced protein synthesis

Abstract

Control of translation is a fundamental source of regulation in gene expression. The induction of protein synthesis by brain-derived neurotrophic factor (BDNF) critically contributes to enduring modifications of synaptic function, but how BDNF selectively affects only a minority of expressed mRNAs is poorly understood. We report that BDNF rapidly elevates Dicer, increasing mature miRNA levels and inducing RNA processing bodies in neurons. BDNF also rapidly induces Lin28, causing selective loss of Lin28-regulated miRNAs and a corresponding upregulation in translation of their target mRNAs. Binding sites for Lin28-regulated miRNAs are necessary and sufficient to confer BDNF responsiveness to a transcript. Lin28 deficiency, or expression of a Lin28-resistant Let-7 precursor miRNA, inhibits BDNF translation specificity and BDNF-dependent dendrite arborization. Our data establish that specificity in BDNF-regulated translation depends upon a two-part posttranscriptional control of miRNA biogenesis that generally enhances mRNA repression in association with GW182 while selectively derepressing and increasing translation of specific mRNAs.

Figure 1. BDNF Increases P Body Formation in Soma and Dendrites of Hippocampal Neurons

(A) Endogenous GW182 (top, red) colocalizes with GFP-Dcp1a (middle, green) in neuronal dendrites; overlay (bottom). (B) P body formation in dendrites of hippocampal pyramidal neurons following mock (top) or BDNF stimulation (bottom, 100 ng/ml). (C) P body formation in cell somas following mock (top) or BDNF stimulation (bottom). t = min poststimulation. (D) Quantification and time course of percent change in GFP-Dcp1a P body numbers in neuronal dendrites following mock (open circles) or BDNF stimulation (closed circles) in the presence of Actinomycin D (0.5 µg/ml) to isolate changes due to translation. (E) Lysates from mock (−) or BDNF (+, 1 hr) stimulated neuronal cultures immunoprecipitated (IP) with GW182 antiserum (IP-GW) or isotype-control serum (IP-Ctrl). Input is 20% of IP’d protein. (F) Densitometric quantification from nine independent experiments, as in (E); mock condition (open bars) set as 1.0. (G) Total RNA, measured by A₂₆₀, recovered by GW182 IP from equal lysate inputs; mock (open bar) set as 1.0. BDNF increased GW182-associated RNA 2.62 ± 0.29-fold. All error bars represent SEM. *p < 0.05 by unpaired Student’s t test. Scale bars, 10 µm. See also Figure S1 and Movies S1–S3.

Figure 2. miRNA-Mediated Repression Is Enhanced by BDNF and Associated with BDNF Target Specificity

(A) Loss of GW182 function by shRNA targeting GW182 (GW182KD) or GFP-DNGW182 expression does not alter BDNF-enhancement of aggregate protein synthesis relative to control (uninfected) cells, or cells expressing scrambled GW182 shRNA or GFP alone. Total protein synthesis was monitored under mock (open bars) or BDNF (hatched bars, 100 ng/ml, 2 hr) stimulated conditions, and plotted relative to the control mock condition set as 1.0 (first open bar). A translation inhibitor (rapamycin, 20 µg/ml) demonstrates that observed changes are due to translation. (B) (Left) Immunoblotting for BDNF target proteins in neurons either uninfected or infected with lentivirus expressing GW182 shRNA (GW182KD) or a mismatched control shRNA (sh-Control-1). mCherry is coexpressed from the virus. (Right) Protein levels, normalized to β-tubulin, of representative BDNF-upregulated targets under mock (open bars) or BDNF (hatched bars, 100 ng/ml, 2 hr) stimulation in the presence or absence of GW182KD (control mock, white bars, set as 1.0); n = 6 independent experiments. (C) (Left) Immunoblotting for BDNF target proteins in neurons either uninfected or infected with lentivirus expressing GFP-DNGW182 or GFP. (Right) Protein levels, normalized to β-tubulin, of representative BDNF up- or downregulated targets under mock (open bars) and BDNF (hatched bars) stimulated conditions in cells expressing GFP-DNGW182, GFP, or control uninfected cells (control mock, white bars, set as 1.0); n = 6 independent experiments. (D) miRNA function is inhibited by GW182KD and GFP-DNGW182, but not by knockdown of LSm5 (LSm5KD). (Left) Luciferase activities of siRNA- or miRNA-reporter constructs in cells expressing reporter alone (−sh-CXCR4), or coexpressing reporter and CXCR4 shRNA (+sh-CXCR4), with or without GW182KD, GFP-DNGW182, or LSm5KD. Normalized luciferase values are shown relative to levels without sh-CXCR4 (set as 1.0). (Right) Diagram of reporter constructs. (E) LSm5 knockdown did not alter protein synthesis of representative BDNF targets. (Left) Immunoblotting for BDNF targets in neurons expressing control shRNA (sh-Control-2) or shRNA against LSm5 (LSm5KD) following mock (−) or BDNF (+) stimulation (100 ng/ml, 2 hr). (Right) Densitometric quantification of three independent immunoblots, normalized to GAPDH and plotted relative to mock-stimulated controls (sh-Control-2 -mock). (F) BDNF enhances repression of a miRNA-reporter by a small RNA hairpin (sh-CXCR4). Normalized luciferase values are shown for mock (open bars) or BDNF-stimulated (hatched bars) neurons coexpressing the miRNA reporter and either sh-Control-2 or a dose titration of sh-CXCR4. Low-dose sh-CXCR4 repressed the miRNA-reporter in BDNF-stimulated, but not mock-stimulated, conditions. All experiments done in the presence of Actinomycin-D. Error bars represent SEM. *p < 0.05 in comparison to reporter alone condition (− sh-CXCR4, D and F) or mock (open bars) by unpaired Student’s t test. See also Figures S2, S3, and S4.

Figure 3. BDNF Increases Dicer Levels and the Biogenesis of Mature miRNAs

(A) (Left) Percentage of miRNAs from TaqMan miRNA array with levels decreased over 50% (open bar) or increased over 2-fold (black bar) by BDNF (30 min, plus Actinomycin-D). (Right) Scatter plot of relative quantities (RQ) of individual miRNA species (solid circles) following BDNF relative to mock-stimulation. Red line = 1.0 or no change; each dot above the line represents a miRNA species increased by BDNF, each dot below the line represents a miRNA species decreased by BDNF. Normalization is to averaged reference RNAs U6snRNA, and snoRNA202, which are unchanged by BDNF; n = 3 separate miRNA array pairs for mock and BDNF conditions. (B) Immunoblot of cultured hippocampal neurons stimulated with BDNF for indicated min in the presence of Actinomycin-D. Dicer peaks near 20 and declines by 60 min. (C) (Top) BDNF enhances TRBP and ERK phosphorylation as shown by immunoblot for TRBP and phospho-Erk. Cultured hippocampal neurons were stimulated with BDNF for indicated min in the presence of Actinomycin-D. (Bottom) Lysates incubated with λ-phosphatase (λ-phos) as indicated demonstrate loss of phosphorylated TRBP (upper band). (D) P body appearance in dendrites of hippocampal pyramidal neurons expressing control (sh-control-2, top) or Dicer-targeting shRNA (DicerKD, bottom) following BDNF, t = min poststimulation. (E) Quantification and time course of P body numbers in Dicer-deficient (DicerKD, boxes) or control (sh-control-2, circles) expressing hippocampal neurons following mock (open shapes) or BDNF stimulation (closed shapes). (F) Quantification and time course of P body numbers in hippocampal pyramidal neurons treated with enoxacin (15 µM, closed circles) or oxolinic acid (15 µM, open circles). (G) The effect of Dicer loss on BDNF-regulated protein synthesis. (Top) Immunoblotting for BDNF target proteins in Dicer-wild-type (Dicer^flox/flox) or Dicer-deficient (Dicer^−/−) neurons. Cre^ERT2-expressing cells were treated with 4-hydroxy tamoxifen (800 nM) to induce recombination for 2.5 days before BDNF stimulation. Asterisk indicates nonspecific band. (Bottom) Densitometric quantification of immunoblots. All error bars represent SEM. *p < 0.05 by unpaired Student’s t test. See also Figure S4.

Figure 4. BDNF Induces Lin28, Selectively Diminishes Lin28-Regulated miRNAs, and Specifically Upregulates a Heterologous Reporter Containing Let-7-Binding Sites

(A) Lin28a (top) and Lin28b (bottom) immunoblots of lysates from cultured hippocampal neurons stimulated with BDNF for indicated min. (B) Time course of BDNF-induced reductions in Lin28-regulated miRNA levels by individual TaqMan qRT-PCR reactions in mock- (BDNF 0′) or BDNF-stimulated neurons. miRNA levels were normalized to 18 s rRNA and plotted relative to each mock-stimulated condition (set as 1.0). All samples underwent equal duration Actinomycin-D incubation prior to harvest. (C) Northern blot (left) and quantitation (right) of pre- and mature miRNA levels of a Lin28-target (Let-7a) or control miRNA (miR-17) in mock or BDNF-treated (30 min) neurons. (D) A binding site for Let-7 miRNAs in the 3′UTR of an mRNA confers upregulation of protein synthesis in response to BDNF. Neurons expressing Let-7 reporters containing two functional (Let-7 WT) or mutated (Let-7 Mut) Let-7 miRNA binding sites in the 3′UTR of firefly luciferase, or a reporter lacking miRNA binding sites were mock or BDNF stimulated (4 hr). Luciferase activities are normalized to coexpressed constitutive β-galactosidase activity and plotted relative to mock-stimulation for each reporter. All error bars represent SEM. *p < 0.05 by unpaired Student’s t test. (E) Predicted binding sites for Lin28-targeted miRNA. The presence of a Lin28-targeted miRNA binding site in the 3′UTR of transcripts for which translation is BDNF-upregulated (green), BDNF-downregulated (red), and BDNF-nonregulated (black) as predicted by TargetScan, PITA, Pictar, MiRanda, and miRwalk. Pink boxes denote the presence of a miRNA binding site in which the miRNA seed sequence (nucleotides 2–7) paired as a perfect or G-U wobble-containing match. Gray boxes denote the absence of a miRNA binding site. See also Figure S5.

Figure 5. Lin28 Is Required for Relief of miRNA-Mediated Repression and Selective Induction of BDNF-Upregulated mRNA Targets

(A) Loss of Lin28 prevents BDNF-induced decreases in mature Let-7a levels. Mature Let-7a levels were assessed by qRT-PCR from neurons infected with lentivirus expressing either control shRNA (sh-control-2) or shRNA targeting Lin28a (Lin28aKD) and mock or BDNF stimulated for 20 min (no Actinomycin-D); normalization was to control mock values (open bar, set as 1.0, n = 3). (B) Effect of Lin28a loss on BDNF-regulated protein synthesis. Immunoblotting of BDNF targets in control or Lin28a-deficient cells, mock or BDNF stimulated (left). Densitometric quantification of protein levels (right, top, n = 6 each condition). Total mRNA levels for both BDNF-upregulated or downregulated targets (right, bottom). (C) Effect of Lin28a KD on BDNF-induced association of protein and RNA P body components. Lysates were immunoprecipitated with anti-GW182 antibody in control (sh-Control-2) or Lin28a-deficient cells, mock or BDNF stimulated. Immunoblotting for co-IP’d Ago2 and Dcp1a (left) and densitometric quantification (right, top, n = 3). Total RNA from GW182 IP of equal lysate inputs from Lin28a knockdown (Lin28aKD) or control (sh-Control-2) neurons; mock (open bars) set as 1.0. BDNF-induced increase in GW182-associated total RNA remains intact after Lin28a knockdown (right, bottom). (D) Abundance of BDNF mRNA targets associated with GW182 in control (sh-Control-2) or Lin28a-deficient cells. In Lin28a-deficient neurons, mRNAs for BDNF-upregulated targets remain associated with GW182 in the presence of BDNF, whereas the response of mRNAs for BDNF-downregulated targets is unchanged. 18 s rRNA is nondetectable, ND. Error bars represent SEM. *p < 0.05 Student’s t test.

Figure 6. Lin28-Mediated Degradation of Let-7 Precursors Is Required for Induction of BDNF-Upregulated Targets and Neuronal Outgrowth

(A) Lentiviral-mediated expression of wild-type (Let-7^WT) or Lin28-resistant (Let-7^LR) Let-7 pre-miRNAs in neurons produced dose-dependent enhancement of mature Let-7a miRNA levels assessed by qRT-PCR and shown as fold change relative to infection with virus expressing GFP alone (gray bar, set as 1.0); 13 or 23 refers to viral dose. (B) Expression of Let-7^LR, but not Let-7^WT, blocks specificity of BDNF for upregulated targets. Reporter assays in mock (open bars) or BDNF stimulated (hatched bars) neurons with luciferase constructs fused to the 3′UTR from BDNF-upregulated targets (GluA1 or CaMKIIα) or a downregulated target (KCC2). (C) BDNF-induced dendrite outgrowth requires Lin28-mediated degradation of miRNA precursors. Dendrite complexity is quantitated for neurons expressing Let-7^WT (black circles) or Let-7^LR (red triangles) following mock (open shapes) or BDNF (25 ng/ml, closed shapes) treatment. *p < 0.05 by unpaired Student’s t test or unpaired one-way ANOVA between Let-7^WT and Let-7^LR in mock and BDNF conditions. (D) Soma size (left) and total dendritic length (right) did not significantly differ between Let-7^WT and Let-7^LR in mock or BDNF treatment. Error bars represent SEM. All experiments done in the presence of Actinomycin D. See also Figure S6.

Figure 7. Proposed Model for the Determination of mRNA Target Specificity in BDNF-Mediated Translation

(Left) In the absence of BDNF stimulation, both Lin28-targeted precursor miRNAs (GGAG, red) and non-Lin28-targeted precursor miRNAs (blue) are processed into mature miRNAs. mRNAs targeted for translational repression or degradation by these mature miRNAs accumulate in P bodies. (Right) BDNF induces both positive and negative regulation of miRNA biogenesis. In the presence of BDNF, TRBP phosphorylation and Dicer protein levels increase leading to a general enhancement of processing of precursor miRNAs (blue) into mature miRNAs. Increased abundance of mature miRNAs leads to an increase in targeting of mRNAs for repression and increases the number of P bodies in cells. However, Lin28a protein levels also increase in response to BDNF (far-right). Because Lin28a selectively prevents processing of its targeted precursor miRNAs (GGAG, red) into mature miRNAs, this population of miRNAs is diminished and mRNA targets of these miRNAs are no longer efficiently repressed and become more readily available for translation. The differential effects of BDNF on distinct miRNA populations can explain the selective increase in translation of only specific mRNAs in response to BDNF.