Fig 2: The microRNA-generating complex localizes at the neuronal endoplasmic reticulum Components of the RLC co-precipitate with ER markers in density gradient fractionation. The postnuclear supernatant of cortical neurons was loaded underneath a continuous Optiprep density gradient. ER markers, calnexin, ribophorin I, and Climp63, co-sedimented with ribosomes (RPS6, ribosomal protein subunit 6), mitochondria (Tim23), and RLC proteins at fractions 3–5. The bottom graph shows the density of each fraction.TRBP, PACT, and Ago2 partially co-localize with the ER marker mCherry-CLIMP63. Hippocampal neurons (10 DIV) expressing mCherry-CLIMP63 were immunostained with TRBP, PACT, or Ago2 antibodies. Boxed insets are magnifications of primary dendrites as depicted by the adjacent letters (scale bars; 10 µm).GFP-Dicer partially co-localizes with mCherry-CLIMP63 at the neuronal soma and primary dendrites of hippocampal neurons. Neurons were imaged at 10 DIV using total internal reflection fluorescence (TIRF) microscopy using a penetration depth of 150 nm. Boxed insets are magnifications of primary dendrites as depicted by the adjacent letters (scale bars; 2 µm or 10 µm).The majority of Dicer, TRBP, and PACT are associated with neuronal membranes. Sequential detergent extraction was used to separate cytoplasmic from membrane fractions in 7 DIV cortical neurons. The relative quantity of depicted proteins was analyzed in three independent experiments (**P < 0.005, t-test type 3, n = 3, error bars s.d.). Actin and ribophorin I were used as loading controls for cytoplasmic and membrane fractions, respectively. The dotted line is placed at 1 to indicate the fold change. Individual data points acquired in three independent experiments are shown as separate shapes.TRBP is cytoplasmic in primary astrocytes. Sequential detergent extraction was performed in primary astrocyte cultures. Actin and calnexin were used as loading controls for cytoplasmic and membrane fractions, respectively.The majority of Dicer, TRBP, and PACT proteins are present in cytoplasmic fractions in HEK293T cells, isolated using sequential detergent extraction.

Fig 3: Short BDNF stimulation leads to the redistribution of TRBP from the ER to the cytoplasm BDNF stimulation leads to a decrease in TRBP levels from ER microsomes. Cortical neurons were stimulated BDNF with or vehicle control, and ER microsomes were isolated using differential centrifugation. Protein intensity values were normalized to calnexin in each of three independent experiments (**P = 0.008, t-test type 3, n = 3, error bars; s.d.).Endogenous TRBP partially co-localizes with calnexin at the neuronal cell soma and primary dendrites. Hippocampal neurons were transfected with GFP at 7 DIV and fixed at 10 DIV following BDNF or control stimulation. Boxed insets are magnifications of the primary dendrite (15 µm in length). Arrows point to overlapping puncta (scale bars; 20 µm (GFP) or 10 µm).BDNF leads to a decrease in co-localization between TRBP and calnexin at the soma and primary dendrites of hippocampal neurons. Manders’ co-localization co-efficients between TRBP and calnexin were measured in the cell soma or primary dendrites (15 µm) using GFP as a morphological mask. Average co-localization values were normalized to controls in each experiment (**P < 0.008, t-test type 3, n = 4, error bars; s.d.).BDNF decreases mCherry-CLIMP63 co-localization with TRBP, but not with PACT or Ago2. Hippocampal neurons were transfected with GFP and mCherry-CLIMP63 at 7 DIV and immunostained 3 days later, following BDNF or control stimulation. Manders’ co-localization was measured at the cell soma using GFP as a mask as indicated (*P = 0.01, t-test type 3, n = 4, error bars; s.d.).BDNF leads to an increase in the levels of TRBP (*P = 0.03) and PACT (*P = 0.004) in the cytoplasmic fraction. Cortical neurons (7 DIV) were control- or BDNF-stimulated, and cytoplasmic fractions were isolated using digitonin. Protein intensity values were normalized to actin in each experiment. The position of the star corresponds to the TRBP band used for analysis (˜43 kDa) (t-test type 3, n = 4, error bars; s.d.).The BDNF-induced increase in cytoplasmic TRBP is dependent on intracellular calcium. Cortical neurons (7 DIV) were treated with vehicle control, BDNF and BAPTA-AM as indicated, and cytoplasmic fractions were isolated using digitonin. Protein intensity values were normalized to actin in each independent experiment (*P = 0.03, n = 3, error bars; s.d.).Data information: The dotted line is placed at 1 to indicate the fold change. Individual data points acquired in three independent experiments are shown as separate shapes.

Fig 4: BDNF affects Dicer binding to TRBP and PACT and bi-directionally regulates the total levels of a few miRNAs BDNF causes a decrease in the Dicer–TRBP interaction in co-IP experiments using a monoclonal mouse TRBP antibody or IgG control. Young cortical neurons were treated with BDNF or vehicle control for 20 min before lysis. The star depicts the right band for immunoprecipitated TRBP (43 kDa).BDNF does not affect the interaction of cytoplasmic Dicer and TRBP. Cytoplasmic fractions of stimulated cortical neurons were isolated, and co-IP was performed using anti-TRBP antibody or control IgG. The star depicts the right band for immunoprecipitated TRBP (43 kDa).BDNF does not affect global levels of microRNAs in small RNA sequencing. The average read counts (log2) of miRNAs from three independent experiments were plotted on the x-axis for the control and y-axis for the BDNF-treated conditions.BDNF stimulation bi-directionally regulates the levels of a few microRNAs in small RNA sequencing (P < 0.05, n = 3, t-test type 2, error bars are s.d.).BDNF does not lead to changes in 5p versus 3p strand selection. The read count ratio (log2) of 5p and 3p miRNAs, generated from the same precursor, was calculated from small RNA sequencing data. The averaged values for three independent experiments were plotted on the x-axis for the control and y-axis for the BDNF-treated conditions.

Fig 5: PACT suppresses mitobiogenesis.A and B, mitochondrial mass was determined from MitoTracker Green–stained cells (quantified by measuring mean fluorescence intensity (MFI) with ImageJ from at least 200 cells; a representative image shown. The scale bar represents 50 µm): (A) Prkra+/+, Prkra-/-, or Prkra-/- MEFs reconstituted with empty vector (Empty Vec.) or PACT (n = 8) and (B) HEK293T cells transfected with 100 nM scrambled or PRKRA siRNA (n = 5). C and D, HEK293T cells transfected with 100 nM scrambled or PRKRA siRNA: (C) Protein lysates were analyzed by Western blotting using specific antibodies for PGC1a, TOM70, antibody cocktail against ETC proteins, NRF1, PACT, and ß-actin (n = 3) and (D) Total genomic DNA was isolated and mtDNA (mitochondrial minor and major arc): nucDNA (B2M) ratio was analyzed by qRT-PCR (n = 3). E and F, HEK293T cells transfected with empty vector (Empty Vec.) or PACT plasmid: (E) Protein lysates were analyzed by Western blotting using specific antibodies for PGC1a, TOM70, antibody cocktail against ETC proteins, NRF1, PACT, and ß-actin (n = 3) and (F) Total genomic DNA was isolated and mtDNA (mitochondrial minor and major arc): nucDNA (B2M) ratio was analyzed by qRT-PCR (n = 3). Protein expression was calculated relative to ß-actin (for CL) or Ponceau S. (for MF) and depicted at the top of each blot. Data are mean ± SD. Unpaired t test with Welch’s correction or one-way ANOVA. *p = 0.05, **p = 0.01, ***p = 0.001. CL, cell lysate; Cyto, cytoplasmic fraction; ETC, electron transfer chain; MEF, mouse embryonic fibroblast; MF, mitochondrial fraction; mtDNA, mitochondrial DNA; nucDNA, nuclear DNA; siRNA, silencer RNA.