Brain-Derived Neurotrophic Factor protects neurons by stimulating mitochondrial function through Protein Kinase A

Abstract

Brain derived neurotrophic factor (BDNF) stimulates dendrite outgrowth and synaptic plasticity by activating downstream protein kinase A (PKA) signaling. Recently, BDNF has been shown to modulate mitochondrial respiration in isolated brain mitochondria, suggesting that BDNF can modulate mitochondrial physiology. However, the molecular mechanisms by which BDNF stimulates mitochondrial function in neurons remain to be elucidated. In this study, we surmised that BDNF binds to the TrkB receptor and translocates to mitochondria to govern mitochondrial physiology in a PKA-dependent manner. Confocal microscopy and biochemical sub-cellular fractionation assays confirm the localization of the TrkB receptor in mitochondria. The translocation of the TrkB receptor to mitochondria was significantly enhanced upon treating primary cortical neurons with exogenous BDNF, leading to rapid PKA activation. Showing a direct role of BDNF in regulating mitochondrial structure/function, time-lapse confocal microscopy in primary cortical neurons showed that exogenous BDNF enhances mitochondrial fusion, anterograde mitochondrial trafficking, and mitochondrial content within dendrites, which led to increased basal and ATP-linked mitochondrial respiration and glycolysis as assessed by a XF24e Metabolic Analyzer. BDNF-mediated regulation of mitochondrial structure/function requires PKA activity as treating primary cortical neurons with a pharmacological inhibitor of PKA or transiently expressing constructs that target an inhibitor peptide of PKA (PKI) to the mitochondrion abrogated BDNF-mediated mitochondrial fusion and trafficking. Mechanistically, Western/Phos-tag blots show that BDNF stimulates PKA-mediated phosphorylation of Drp1 and Miro-2 to promote mitochondrial fusion and elevate mitochondrial content in dendrites, respectively.

Effects of BDNF on mitochondrial function were associated with increased resistance of neurons to oxidative stress and dendrite retraction induced by rotenone. Overall, this study revealed new mechanisms of BDNF-mediated neuroprotection which entails enhancing mitochondrial health and function of neurons.

Introduction

Brain derived neurotrophic factor (BDNF) belongs to a family of neurotrophins that govern neuronal function, development, and survival (Reichardt, 2006). BDNF is synthesized in the endoplasmic reticulum as a precursor protein, then processed to pro-BDNF by cleavage of the signaling peptide. The terminal domain of the pro-BDNF is further cleaved by plasmin to generate the 13kDa active and mature form of BDNF (Lessmann and Brigadski, 2009). While binding of pro-BDNF to the p75 receptor is mainly implicated in apoptosis, the mature form of BDNF exerts its neurotrophic actions by binding to the Tropomyosin receptor kinase B (TrkB) (Costa et al., 2018). BDNF-induced TrkB receptor dimerization results in kinase activation followed by receptor autophosphorylation on the tyrosine residues Tyr-670/674/675/785 (Middlemas et al., 1994). Several intracellular adaptor proteins such as Shc2 bind to the phosphorylated and activated TrkB receptor via SH2 domains initiating multiple signaling cascades such as Ras/MAPK-ERK pathway, PI-3 kinase (PI3-K) pathway, and phospholipase C (PLC) pathway, which are involved in neuronal growth, maintenance, neuronal survival, dendrite outgrowth, and neuronal differentiation (Reichardt, 2006; You et al., 2010; Ahmed and Prigent, 2017). BDNF also acts as a self-amplifying cell signal to aid neuronal growth by inducing TrkB dependent activation of protein Kinase A (PKA) (Lai et al., 2019), triggering further synthesis and secretion of BDNF, and thereby augmenting BDNF function. In neurons, PKA is essential for maintaining neuronal functions, including mitochondrial homeostasis, bioenergetics, neuronal development, and neurotransmission (Cao et al., 2011; Piccini et al., 2015; Ould Amer and Hebert-Chatelain, 2018; Zhang et al., 2019). PKA signaling is well regulated and compartmentalized, given that PKA is targeted to sub-cellular domains by forming a complex with scaffolding proteins such as A-kinase anchoring protein (AKAP). Localized PKA signaling in the mitochondrion is enhanced when PKA binds to D-AKAP1 (murine homolog of human Dual specificity AKAP1), leadingto PKA-mediated phosphorylation and inhibition of the catalytic activity of the mitochondrial fission modulator dynamin-related protein (Drp1), a principal mechanism for mitochondrial stabilization and neuronal survival (Merrill and Strack, 2014). It has also been shown that increased PKA-mediated phosphorylation of Drp1 leads to increased resistance of neurons to toxic insults, including staurosporine, H₂O₂, and rotenone (Merrill et al., 2011). Additionally, mitochondrial trafficking throughout the neuron and the recruitment of mitochondria to regions with high metabolic demands are vital for the proper functioning of the neuronal network. It has been shown that PKA phosphorylates the mitochondrial adaptor protein Miro-2 leading to increased anterograde mitochondrial movement to the dendrites and retrograde movement to the soma (Das Banerjee et al., 2017). BDNF also plays a crucial role in synaptic plasticity by enhancing neuronal activity dependent upon synaptic strength through long-term potentiation (Aicardi et al., 2004; Lu et al., 2008).

While the binding of BDNF to TrkB activates a myriad of signaling pathways localized within the cell membrane and cytosolic compartments in neurons, there is evidence of a direct physiological role of BDNF in regulating mitochondrial structure/function. For instance, there have been reports of BDNF enhancing the respiratory coupling efficiency to increase ATP synthesis in neurons,thereby regulating mitochondrial efficiency. That study showed that BDNF enhances the respiratory control index of rat brain mitochondria in an in vitro synaptosome-mitochondrial preparation (Markham et al., 2004; Markham et al., 2014). BDNF stimulates mitochondrial biogenesis by enhancing peroxisome proliferator activated receptor gamma coactivator 1-alpha (PGC-1α) promoter activity via activation of mitogen-activated protein kinases (MAPKs) and CREB that is crucial for the formation and maintenance of hippocampal dendritic spines and synapses (Cheng et al., 2012). Alternatively, BDNF release is increased in the hippocampus via PGC-1α activation and secreted irisin (Wrann et al., 2013). In further support of the concept that BDNF modulates mitochondrial physiology, in endothelial cells, BDNF is reported to activate mitophagy through the HIF-1 α/BNIP3 signaling pathway (Jin et al., 2019). While there is increasing evidence that suggests that BDNF regulates mitochondrial function and dendrite health, the molecular mechanisms by which BDNF regulates mitochondrial motility, structure, and bioenergetics are unknown. In this study, we present new findings for the first time indicating that exogenous treatment with BDNF causes the translocation of TrkΒ to the mitochondria, thus promoting neuronal maintenance through stimulation of mitochondrial trafficking, dynamics, and bioenergetics. Furthermore, here we report the molecular mechanisms by which BDNF regulates mitochondrial trafficking, dynamics, content, and bioenergetics through the activation of PKA signaling and subsequent phosphorylation of OMM-localized substrates Drp1 and Miro2.

Materials and methods

Culture of primary cortical neurons:

primary cortical neurons were prepared as previously described (Soman et al., 2021). All experiments involving mice were performed in accordance with ARRIVE guidelines and were approved by the University of Nevada, Reno’s Institutional Animal Care and Use Committee (IACUC, Protocol # 20-09-1086-1). Briefly, E15–17 embryos were obtained from wild-type (WT) timed-pregnant female that underwent isoflurane-mediated euthanasia (RRID: IMSR_CRL:027) The E15 To minimize pain and distress, the embryos were placed in ice following extraction from the embryonic sac and their heads were immediately severed in ice using a scalpel.

Intact whole brains were then extracted from the severed heads of the E15 embryos. The cortices were micro-dissected from the brain and placed into ice-cold plating media [Neurobasal media (Thermo Scientific^™, USA, Catalog no. 21103049), 2% FBS, 2% B27 (Thermo Scientific^™, USA, Catalog no.- 17504044 ), 0.5mM Glutamax, 25μM Glutamic acid (Sigma-Aldrich, USA, Catalog no.- G1251), 100 U/ml penicillin/streptomycin (Thermo Scientific^™, USA, catalog no. 15140122)] and mechanically dissociated with repeated pipetting (12–15 times) with a 1mL pipettor. The cell density was determined using a hemocytometer, and subsequently, primary cortical neurons were plated on poly-L-Lysine (Thermo Scientific^™, USA, Catalog number- P4832)) coated sterile cell culture plates [Lab-Tek^™ IV Chamber Slides, catalog no. 177399 (0.5ml media/well)- 2 × 10⁶ cells/mL; Corning 12 well plate, catalog No. 07-200-82 (0.5ml media/well)- 2 × 10⁶ cells/mL; Corning 6 well plate, catalog No.07-200-83 (2ml media/well)- 3.5 × 10⁵cells/mL] in pre-warmed plating media. The plates containing primary cortical neurons were maintained at 37°C, 5% CO₂ / 95% humidity in a cell culture incubator. Approximately 1/3^rd of the culture medium was then changed every three days and replaced by fresh serum-free maintenance media [Neurobasal media (Thermo Scientific^™, USA), 2% B27 (Thermo Scientific^™, USA, 0.5mM Glutamax (Thermo Scientific^™, USA) 100 U/ml penicillin/streptomycin (Thermo Scientific^™,].

Animals:

Prior to undergoing euthanasia to prepare cultures of primary neurons, timed pregnant female mice were acclimated maintained under controlled temperature 25–26 °C, 12/12 h light/dark cycle, with complete access to food and water ad libitum and individually caged for at least four days prior to processing for euthanasia .Per transfection experiment or pharmacological assay, approximately 30–32 million total primary cortical neurons can be derived from 6–8 mouse embryos that were plated at a density of 150,000–175,000 per well for performing live confocal imaging assays or mitochondrial respiration assays, or plated at 1.5 million primary cortical neurons per well to perform Phostag/Western blot assays.

Number of animals used per experimiental design:

To minimize the number of animals used for research, several fluorescence imaging assays (mitochondrial length, content and dendrite length) were performed from primary cortical neurons isolated from the brain of the same timed pregnant female mouse. For all other assays, a different timed pregnant mouse was needed for performing biochemical assays (Phostag/Western blots) and mitochondrial respiration assays by culturing primary neurons in 6 well plates and 24 well plates respectively. Up to 4 timed pregnant mice were employed for Phostag/Western blots shown in figure 3E and figure 6A for a total of 8 animals. Overall, the complete research project necessitated the use of 40 timed pregnant mice which permitted the acquisition of data for at least 3 experiments per type of assay (figure 1; 4 timed pregnant mice), mitochondrial length assays (figure 2; 6 timed pregnant mice), assays of mitochondrial trafficking and content (figures 2–6; up to 16 timed pregnant mice), mitochondrial respiration assays (figure 7; 6 timed pregnant mice) and neuronal survival assays (figure 8; 3 timed pregnant mice).

Figure 3. BDNF promotes mitochondrial length through phosphorylation of Drp1.

(A) Representative epifluorescence images of primary cortical neurons transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml, 24hrs.), in the presence or absence of transient expression of GFP tagged Drp1-WT (Drp1-WT-GFP) & phosphorylation resistant (S656A)-mutant Drp1(Drp1-S656A-GFP) for 72hrs., scale bar = 10 microns. Transfected primary cortical neurons were identified based on GFP fluorescence with their elongated axons and complex dendritic arbors. The bar graph in (B) depicts the mean (±SEM) of mitochondrial length as a measure of mitochondrial fusion in dendrites. BDNF treatment was unable to increase mitochondrial length in the presence of the mutant Drp1-S656A-GFP. The bar graph in (C) depicts the mean (±SEM) of mitochondrial content in primary cortical neurons after exposure to exogenous recombinant human BDNF or transient transfection with the indicated plasmids. In brief, BDNF was partially able to rescue mitochondrial content when Drp1-WT was co-expressed but not by mutant Drp1-S656A-GFP, (D) Representative Phos-tag Western blot of phosphorylated-Drp1 and total Drp1 in cell lysates extracted from Primary cortical neurons treated with BDNF (50ng/ml, 24hrs.) & ANA12. The representative Western blot image suggests that exposing primary cortical neurons to exogenous BDNF increases the PKA-mediated phosphorylation of Drp1 leading to increased mitochondrial length. (E) Densitometric analysis of the Phos-tag Western blot data shown in (D) on the phosphorylation status of Drp1 normalized to Drp1. For B, (****: p<0.0001 vs. control, ****: p<0.0001 vs. BDNF, for C, (****: p<0.0001 vs. control, **: p<0.01 DRP1-WT vs. BDNF, ****: p<0.0001 DRP1-S656A vs. BDNF, each dot in the plot represents a mitochondrion from a sample n=100–200 mitochondria/condition from at least 13–15 transfected primary cortical neurons derived per timed pregnant mouse per assay per condition for each assay and pooled 3 assays, One-way ANOVA, for B, F= 201, Rsquared: 0.31, for C, F=31.2, Rsquared: 0.31, 15 comparisons, for E, F=31, Rsquared: 0.86, 3 comparisons, Tukey’s test with Bonferroni correction, two-tailed test).

Figure 6. Exposure of primary cortical neurons with BDNF enhances mitochondrial content in dendrites through PKA.

(A) Representative Phos-tag western blot for Miro-2 in lysates extracted from primary cortical neurons after treatment with 50ng/ml BDNF for 4 and 24hrs. The bar graph in (B) shows the means ±SEM of densitometric analysis of the immunoreactive bands for Miro-2 at the indicated conditions and normalized to control (Means ±SEM, *:p<0.05 vs Control, N=3 experiments, One-Way ANOVA, F=12, Rsquared: 0.81, 6 comparisons, Tukey’s test, two-tailed test). BDNF treatment significantly increases the phosphorylation of Miro-2. The graph in (C) depicts the mean (±SEM) of mitochondrial content in dendrites of primary cortical neurons transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml, 4hrs.), in the presence or absence of transient expression of Mito-PKI-GFP & PKI plasmids when compared to untreated transiently Mito-RFP transfected control. The graph in (D) depicts the mean (±SEM) of mitochondrial content in dendrites of primary cortical neurons transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml, 24hrs.), in the presence or absence of transient expression of Mito-PKI-GFP & PKI plasmids when compared to untreated transiently Mito-RFP transfected control. Mitochondrial content is defined by the percentage of a defined segment of a dendrite occupied by mitochondria. An appropriate level of mitochondrial content is an indication that dendrites have a sufficient number of mitochondria to maintain dendrite length as previously validated by our research group. (E) Representative kymograph of mitochondrial movement in primary cortical neurons treated with BDNF (50ng/ml, 24hrs.), BDNF (50ng/ml, 24hrs.) co-treated with H89 (0.5μM, 24hrs.) and untreated control, exhibiting significantly increased bi-directional movement of mitochondria (μm, x-axis) over a period of five minutes (min., y-axis) with BDNF treatment but does not with PKA inhibitor H89 co-treatment. The graph in (F) depicts the mean (±SEM) anterograde movement of mitochondria in dendrites of primary cortical neurons treated with BDNF (50ng/ml, 24hrs.), BDNF co treated with H89 (0.5μM, 24hrs.) when compared to Mito-RFP transfected control. In brief, the data shows that when compared to untreated Mito-RFP transfected control, treatment of primary cortical neurons with BDNF increased anterograde trafficking but not in neurons co-treated with the PKA inhibitor H89 (0.5μM). For B, C, D, F (*: p<0.05, ***: p<0.001, ****: p<0.0001 vs. control, **: p<0.01, ****: p<0.0001 vs. BDNF, n = 20–40 moving mitochondria in anterograde direction, from at least 13–17 primary cortical neurons pooled from 3 experiments, One-way ANOVA: for C, F=9.87, Rsquared: 0.40, 15 comparisons, for D, F=8.20, Rsquared: 0.67, 15 comparisons, for F, F=27.36, Rsquared: 0.29, 3 comparisons, two tailed- test,, post-hoc Tukey’s test with Bonferroni correctiont,).

Figure 1. TrkB receptor translocates to dendritic mitochondria of primary cortical neurons in response to BDNF treatment.

(A) Representative epifluorescence images of primary cortical neurons transfected with Mito-RFP (48hrs.) and TrkB-GFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml cell culture media, 4hrs & 24hrs.). scale bar = 10 microns. The bar graph in (B) depicts the mean (±SEM) of TrkB -GFP translocation events in the dendrites with mitochondria following treatment with BDNF (50ng/ml, 4hrs.). The bar graph in (C) depicts the mean (±SEM) of TrkB-GFP translocation events in the dendrites with mitochondria after BDNF treatment (50ng/ml, 24hrs.). The bar graph in (D) depicts the mean (±SEM) of Trkβ-GFP translocation events in the soma with mitochondria after BDNF treatment (50ng/mL, 4hrs.). The bar graph in (E) depicts the mean (±SEM) of TrkB-GFP translocation events in the soma with mitochondria after BDNF treatment (50ng/ml, 24hrs.). (For bar graphs shown in B-E, Means ±SEM, *: p<0.05 vs. Control, **: p<0.045 vs. Control, each dot on the plot indicates the mean number of events for each neuron and compiled from 13–17 transfected primary cortical neurons pooled from 4 experiments/ condition, One-way ANOVA, F=4.2, Rsquare: 0.16, post-hoc Tukey’s test with Bonferroni correction, two-tailed test). The data suggests that exposing primary cortical neurons to exogenous BDNF increases the translocation of TrkB-GFP with Mito-RFP-labeled mitochondria at 4hrs. post treatment and mitochondrial translocation of TrkB-GFP saturates at 24hrs. treatment of BDNF. (F) Representative Western blot of endogenous TrkB in mitochondrial fractions isolated from whole brains extracted from 10-month-old wt mouse. The membrane was stripped and re-probed for TOM20, an OMM-localized mitochondrial marker, to verify the purity of mitochondria relative to the cytosolic and lipid compartments. The representative WB from a total 4 assays shows that a fraction of endogenous TrkB is localized in the mitochondrial compartment.

Figure 2. BDNF promotes mitochondrial fusion in a PKA-dependent manner.

(A) Representative epifluorescence image of a primary cortical neuron transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml, 24hrs.), scale bar = 10 microns. The bar graph in (B) depicts the mean (±SEM) of mitochondrial length as a measure of mitochondrial fusion in dendrites of primary cortical neurons. The bar graph in (C) depicts the mean (±SEM) of mitochondrial length in neurons following transient transfection of GFP-tagged protein kinase A inhibitor peptide (PKI-GFP) or OMM targeted-PKI (Mito-PKI-GFP) to inhibit PKA with and without exogenous recombinant human BDNF (50ng/ml, 4hrs.) treatment. The bar graph in (D) depicts the mean (±SEM) of mitochondrial length in neurons after transfection of PKI-GFP or Mito-PKI-GFP, with and without treatment with exogenous recombinant human BDNF (50ng/ml, 24hrs.). In brief, the data shows that BDNF (50ng/ml, 4 & 24hrs.) increased mitochondrial length, but not in primary cortical neurons co-transfected with the PKA inhibitor PKI. For B-D, (****: p<0.0001 vs. control, ***: p<0.0001 vs. 4hrs. or **: p<0.001 vs. 24hrs. BDNF, each dot in the plot represents one mitochondrion analyzed, n=100–200 mitochondria/condition from at least 13–17 primary cortical neurons derived per timed pregnant mouse per assay per condition for each assay pooled from N=3 experiments/ condition One-way ANOVA: for B, F=29.71, Rsquare:, for C, F=59.42, Rsquared: 0.37, and for D, F=119, Rsquared:0.37, 0.05 post-hoc Tukey’s test with Bonferroni correction, two-tailed test)

Figure 7. BDNF increases oxidative phosphorylation in live primary neurons and in isolated mitochondria.

(A) A representative Seahorse oxygraph in primary cortical neurons treated with exogenous BDNF (75ng/ml, 2hrs.) followed by treatment with oligomycin (oligo), FCCP and rotenone (rot) to measure ATP-dependent, maximal, and mitochondrial-dependent OCRs respectively at the specified time points (Means ±SEM, n=5 wells per condition, *:p<0.02 vs. Control for baseline and maximal OCRs, Two-Way ANOVA, F=4.2, Rsquared:0.11., 6 comparisons, two-tailed) (B) compiled graph showing mean basal (acute) OCRs in primary cortical neurons treated with exogenous BDNF and the indicated treatments. (C) compiled graph of maximal mean OCRs in primary cortical neurons following exposure to FCCP. The graph shows BDNF increases mitochondrial production of ATP. (D). (E) graph of OCR shows BDNF increased proton leak compared to TrkB inhibitor ANA12. (F) graph of a representative experiment shows the mean ±SEM extracellular acidification rates (ECARs), a proxy of glycolysis, in neurons treated with BDNF (75ng/m; 2hrs.). The data shows that exposing neurons with BDNF treatment (75ng/ml, 24hrs.) enhanced glycolysis. (G) graph of ECAR shows BDNF increased Basal Glycolysis. (H) graph of ECAR maximal Glycolysis. For panels (A-H), BDNF treatment (75ng/ml, 2hrs.), BDNF treatment (75ng/ml, 2hrs.) & H89 (0.5μM, 24hrs.), ANA12 (200nM, 24hrs.), For B-H (means ±SEM, *: p<0.05, ****: p<0.0001, vs. Control, 4–5 replicate wells of neurons derived per single timed pregnant mouse per assay,One-Way ANOVA, For B, F= 4.78, Rsquared: 0.25, 6 comparisons, for C, F=4197, Rsquared: 0.99, 3 comparisons, for D, F=3.26, Rsquared: 0.18, 6 comparisons, for E, F=2.96, Rsquared: 0.17, 6 comparisons, for F, F= 6.31, Rsquared: 0.15, 6 comparisons, for G, Two Way ANOVA, F=61.56, Rsquared: 0.51, 6 comparisons, for H, Two Way ANOVA, F=3.23, Rsquared: 0.23, Tukey’s test with Bonferroni correction, two tailed tests). (I) Isolated mitochondria from mouse brain were treated with exogenous BDNF (100ng/50ul, 150ng/50ul, 1hr.) and the transmembrane potential (Δψm) was measured by employing a multimode fluorescent plate reader. The data shows that BDNF enhances Δψm. For some conditions, isolated mitochondria were also treated with FCCP to collapse Δψm or with malic acid to enhance the transmembrane potential by stimulating complex II dependent respiration. For I (Means ±SEM *: p<0.05, n=5 wells/ condition, One-way ANOVA, Tukey’s test; F= 289, Rsquared: 0.79, p: 0.0002, representative experiment of 3, two-tailed test)).

Figure 8. BDNF promotes mitochondria and protects dendrites from oxidative stress.

(A) Representative fluorescent images of primary cortical neurons treated with exogenous recombinant human BDNF (50ng/ml, 24hrs.), rotenone (60nM, 24hrs.), BDNF (50ng/ml, 24hrs.) & rotenone (60nM, 24hrs.) and immunostained with MAP2B (1:250) and TOM20 (1:250) to label dendrites and mitochondria respectively, scale bar = 10 microns. The graph in (B) shows mean (±SEM) dendrite length/neuron in neurons exposed to BDNF (50ng/ml, 24hrs.) in the presence or absence of rotenone (60nM, 24hrs.) The graph in (C) depicts mean (±SEM) integrated density of TOM20 in primary cortical neurons exposed to BDNF (50ng/ml, 24hrs.) in the presence or absence of rotenone (60nM, 24hrs.). It was observed that BDNF can rescue rotenone-induced loss of dendrites and of dendritic mitochondria (D) Representative fluorescent images of primary cortical neurons incubated with MitoSox for measuring mitochondrial reactive oxygen species in primary cortical neurons. The experimental conditions consisted of a control group, a rotenone (0.5 μM for 1hr.) treatment group, and a rotenone treatment combined with a BDNF (100ng/ml for 1 hour) pre-treatment group. The nucleus is stained with DRQ5. scale bar = 50 microns. The box plot graph in (E) is the graphical representation of (D) and shows enhanced mitochondrial ROS production after rotenone treatment reduced in BDNF treated neurons. . In B, C and E (*: p<0.05, **: p<0.01, ****: p<0.0001 vs. control, vs, **: p<0.01, ****: p<0.0001 vs. rotenone, n=30–40 primary cortical neurons/derived from a timed pregnant mouse per condition for each assay and pooled from 3 assays, One-way ANOVA: for B, F= 10.19, Rsquared: 0.12, 3 comparisons, for C, F=36.60, Rsquared: 0.60, 3 compairosn, for D, F=12.88, Rsquared: 0.39, 3 comparisons, Fisher’s LSD, two-tailed tests).

Transfection:

DNA plasmids that encode RFP targeted to mitochondria (Mito-RFP) and C-terminally tagged GFP to TrkΒ (TrkΒ -GFP) were purchased from (AddGene RRID: AddGene_83952 TrkΒ mEGFP was a gift from Ryohei Yasuda), C-terminally GFP tagged inhibitor of PKA targeted to the OMM (Mito-PKI-GFP; RRID: Addgene_118482) were provided by Dr. Stefan Strack (University of Iowa, Department of Pharmacology). For primary cortical neurons grown in Nunc^™ Lab-Tek^™ Chamber Slide System (Thermo Scientific^™, USA, Catalog no. 177399, 1μg of DNA plasmid was diluted in OPTIMEM media (Thermo Scientific^™, USA, Catalog no. 31985062) and mixed with Lipofectamine^™ 2000 Transfection Reagent ( Invitrogen^™, USA, Catalog no. 11668019) at a final concentration of 1%. Cells were subsequently treated with combined and diluted DNA: Lipofectamine mixture. 24hrs. following transfection, up to 2/3rd of the media was changed with pre-warmed maintenance media.

Extraction of mitochondria from brain tissue:

WT mice were perfused using PBS supplemented with glucose for the prevention of blood clots and to clear the brain tissue from serum albumins. The brain was weighed, and 200mg of tissue was dissected and used for mitochondrial isolation. The brain tissue was homogenized with a Dounce homogenizer (10–15 times) in ice-cold mitochondrial isolation buffer [3mM HEPES (pH 7.4, Sigma-Aldrich, Catalog no. 54457), 210mM Mannitol (Sigma-Aldrich, Catalog no. M4125), 70mM sucrose (Sigma-Aldrich, Catalog no. S0389), 0.2mM EGTA (Sigma-Aldrich, Catalog no. 324626); Add 1 ml mitochondrial isolation buffer /0.1 g of tissue] supplemented with protease inhibitor cocktail (Sigma-Aldrich, Catalog no. P8340) and 40mg bovine serum albumin (Sigma-Aldrich, Catalog no. 810531) without forming bubbles. Following homogenization, the homogenate was separated into multiple Eppendorf tubes on ice and centrifuged 4°C at 700 × g for 10 min, the supernatant was collected, whereas the resulting pellet was saved as the whole lysate fraction. The homogenized lysate was transferred to ice cold Eppendorf tubes and centrifuged at 600 × g for 10 min at 4°C. The supernatant was collected and transferred to new Eppendorf tubes and the pellet was stored as nuclear fraction which was labeled as lipid phase for Western blots. The supernatant was then centrifuged at 15,000 × g for 15 min at 4°C for subcellular fractionation, and the subsequent supernatant was transferred to new Eppendorf tubes and labeled as cytoplasmic fraction. The pellet was washed several times (3–5X) with mitochondrial isolation buffer and centrifuged at 15,000 × g for 15 min at 4°C. The pellet, being the mitochondrial fraction, was labeled and resuspended in storage buffer and stored at −80°C.

Western blotting and Mn²⁺ Phos-tag^™ SDS-PAGE:

Primary cortical neurons were harvested and lysed on ice using lysis buffer [50mM Tris-HCl (Sigma-Aldrich, Catalog no. 108319), 150mM NaCl (Sigma-Aldrich, Catalog no. S9888), 0.25% SDS (Sigma-Aldrich, Catalog no. L3771), 1mM EGTA (Sigma-Aldrich, Catalog no. 324626), 1mM EDTA (Sigma-Aldrich, Catalog no. E9884), 1% Triton^™ X-100 (Sigma-Aldrich, Catalog no. 93443), 10mM NaF (Sigma-Aldrich, Catalog no. 201154), 1mM dithiothreitol (Sigma-Aldrich, Catalog no. D9163), 0.5mM NaVO₃ (Sigma-Aldrich, Catalog no. 590088), 1X protease inhibitor cocktail (Sigma-Aldrich, Catalog no. 810531). Cell lysates were then centrifuged for 15min at 13,000 × g, and the resulting supernatants were transferred to sterile 1.5 mL microcentrifuge tubes. The protein amount in cell lysates was determined by employing the Pierce^™ BCA Protein Assay Kit per manufacturer’s instructions (Thermo Scientific^™, USA, Catalog no. 23225). Sample buffer [5% β-Mercaptoethanol (Sigma-Aldrich, Catalog no. 444203) in 4x Laemmli buffer (Bio-Rad Laboratories, USA, Catalog no. 1610747)] was added to each of the cell lysate samples and boiled for 5 min at 90°C. Proteins were then resolved in SDS/PAGE electrophoresis by using polyacrylamide gels composed of various acrylamide percentages, per the molecular weight of the proteins of interest to be analyzed (7.5%, 10%, 12.5%). The separated proteins were then transferred onto a PVDF transfer membrane (Thermo Scientific^™, USA, Catalog no. 22860) using a Trans-Blot^® SD Semi-Dry Transfer Cell system (Bio-Rad Laboratories, USA, Catalog no. 1703940) at 250 mA for 45min. The PVDF membrane was then blocked in TBST [20mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20 (Sigma-Aldrich, Catalog no.P1379) (v/v)] containing 5% skimmed milk or with 2–5% BSA for 2hrs. at RT with gentle shaking on an orbital rocker. The PVDF membranes were then incubated overnight at 4°C with primary antibodies for mouse anti-human TOM20 (1:1,2000, RRID:AB_628381), rabbit anti-human GAPDH (1:2,000, RRID:AB_2232048), rabbit anti-human TrkB (1:2,2000, :AB_1281171), mouse anti-human Drp1(1:2,000, RRID:AB_2924856) or for mouse anti-human Miro2 (1:2,000, RRID:AB_2179539), rabbit anti-human D-AKAP1 (1:500, RRID:AB_1267649) or with, or rabbit anti-human β-tubulin (1:5,000, RRID:AB_2210354). Following incubation with primary antibodies, the PVDF membranes were rinsed in TBST and subsequently incubated with the respective horseradish peroxidase (HRP)-conjugated secondary antibodies for 2hrs. at RT. Chemiluminescence substrate kit SuperSignal^™ Western Blot Enhancer (Thermo Scientific^™, USA, Catalog no. 46640) was used to detect immunoreactive proteins and were visualized by a detector [ChemiDoc^™MP Imaging System, Bio-Rad Laboratories, USA, Catalog no. 170–8280]. Densitometric analysis of immunoreactive bands in technical replicates of each protein marker of interest was performed by using ImageJ (RRID:SCR_003070) as previously published (Banerjee et al., 2021; Soman et al., 2021). The integrated density for each immunoreactive band of interest was normalized to β-tubulin.

Mn^2+/Phos-tag^™ (Wako Pure Chemical Industries Ltd, Japan, Cataog no. AAL-107S1) was employed for analyzing the phosphorylation of endogenous proteins-Miro-2 and Drp1 as previously published but with the following modifications (Das Banerjee et al., 2017). The major difference from the conventional SDS/PAGE electrophoresis pertains to the preparation of the acrylamide gel with the addition of Mn²⁺/Phos-tag. Also, following electrophoresis, the gels were treated with 10mM EDTA to remove Mn²⁺ for efficient transfer.

Image acquisition and analysis:

Epifluorescence microscopy: EVOS-FL Cell Imaging System (Thermo Fisher Scientific, Waltham, MA) with a 40X objective. The images were analyzed using Image J software and compatible plugin-Neuron J (Erik Meijering, Biomedical Imaging Group Rotterdam, Netherlands).

Mitochondrial trafficking:

To study mitochondrial trafficking, primary cortical neurons were transfected with 0.5 μg Mito-RFP for 24hrs. Up to 12–15 transfected primary cortical neurons were imaged for mitochondrial transport analysis. Transfected primary cortical neurons were identified by the red fluorescence emitted from Mito-RFP expression as well by their characteristic elongated axons and dendritic arbors that are noticeable by the “bleed through” extra-mitochondrial fluorescence emitted by Mito-RFP expression. The movement of mitochondria was recorded via live-cell imaging, using a Leica spinning disk confocal microscope containing a Plan-Apochromat 100X/oil objective. The mitochondrial movement was recorded every 10 seconds for four minutes. Kymographs were analyzed using Image J v 1.44 (Bethesda, MD, USA) and compatible plugin ‘Multiple Kymograph’ (J. Rietdorf, A. Seitz, EMBL, Heidelberg). In brief, the movement of mitochondria was characterized as ‘anterograde’ if mitochondria moved away from the cell body, and ‘retrograde’ for movement toward the cell body. Hence, to assign the direction of moving mitochondria for each neuron of interest in a consistent manner, the ‘movement tracks’ of mitochondria were always traced from left to right of the cell body. Mitochondrial velocity and distance were assessed for at least 50–70 mitochondria per neuron, from at least 15–17 transfected primary cortical neurons per condition. Importantly, to ensure rigor and reduce bias in the analysis due the subjective nature and intrinsic variability observed with fluorescence single cell imaging assays, all colocalization studies assays were performed blind to experimental conditions in a double blinded fashion. Specifically the experimental conditions were relabeled (using a numbering system “1–5” for relabeling slides) during data collection in a way so that the person collecting epifluorescence micrographs on the confocal microscope and the person analyzing the data from the fluorescent images was blinded to the experimental conditions after the relabeling of the images using a random number generator adopted by a macro. Following the completion of data analyses, the data that was tabulated in spreadsheets was “unblinded” and relabeled by using a legend that was safely stored during the collection of data..

Translocation studies:

Primary cortical neurons were co-transected with Mito-RFP and TrkΒ -GFP (1:1 plasmid ratio) for 24hrs. in Lab-Tek^™ IV chambered slides. Live-cell imaging was performed using a Leica spinning disk confocal microscope or an Olympus Fluoview1000 laser-scanning confocal microscope running Fluoview FV10-ASW software (Olympus Corporation, Waltham, MA, USA, RRID:SCR_014215), using a × 60 oil objective plus digital zoom. The translocationof TrkΒ-GFP puncta to mitochondria labeled with MitoTracker Red or with Mito-RFP was analyzed in at least 15–17 transfected primary cortical neurons by using ImageJ plugin Coloc2 program and manually by counting the number of yellow puncta (Dagda et al., 2008). Given that TrkΒ is localized to the outer mitochondrial membrane, mitochondria that were decorated or delineated with Trkβ-GFP but did not completely colocalized were counted as associated structured but included in the analysis. In addition to counting the mean number of TrkΒ-GFP puncta per neuron, the mean number of TrkΒ puncta that translocated to mitochondria was calculated within the dendrites or soma of neurons.

To ensure rigor and reduce bias in the analysis given the subjective nature of colocalization assays, all colocalization studies assays were performed blind to experimental conditions in a double blinded fashion. Specifically the experimental conditions were relabeled (using a numbering system “1–5” for relabeling slides) in a way so that the person collecting epifluorescence micrographs on the confocal microscope and the person analyzing the data from the fluorescent images was blinded to the conditions after relabeling of the images by using techniques that randomizes conditions (e.g. random number generator adopted by a macro). Following the data analyses, the data “unblinded” and relabeled by using a legend that was previously safely stored during the collection of data.

Analyzing bioenergetics:

The effect of BDNF on the bioenergetic status of primary cortical neurons was analyzed by studying oxidative phosphorylation (OXPHOS) and glycolysis using Seahorse XF24 Extracellular Flux Analyzer (Agilent, Santa Clara, CA). In brief, primary cortical neurons were plated (75,000 cells/well) in poly-L-Lysine pre-coated 24 well XF cell culture plates (Agilent Technologies catalog number 10077–004) and exposed to BDNF (35ng/well, 2hrs., Sigma-Aldrich, Catalog no. SRP3014), ANA12 (200nM, 2hrs., Tocris, Catalog no. 4781) and H89 (1μM, 2hrs., (Tocris, Catalog no. 2910) respectively which were injected from one of the injection ports into the wells (quadruplicate wells) via the Seahorse XF24 Extracellular Flux Analyzer. Cells were subsequently washed with pre-warmed XF assay media (Agilent Technologies catalog# 102340–100). After washing 3 x, cells were further incubated in the respective media (37 °C, 1 hr.) in a CO₂-free incubator to further purge CO₂ and allow temperature/pH equilibration before each set of measurements in the metabolic analyzer. Each plate contained four wells that were not seeded with neurons which served as blank controls. The real-time oxygen consumption rates (OCRs), which is an indicator of OXPHOS, and extracellular acidification rates (ECARs), which is an indirect measurement of glycolysis, were monitored by quantifying the following parameters: the non-mitochondrial oxygen consumption (the OCRs measured following injection with rotenone, SIGMA Aldrich, catalog # 83–79-4 and antimycin-A; SIGMA Aldrich, catalog # 1397–94-0 ), the basal respiration (the last OCR measured prior to exposing cells to the ATP synthase inhibitor oligomycin), the maximal respiration (the maximum OCR rate measurement obtained following exposure of cells with the mitochondrial uncoupler FCCP), the proton (H⁺) leak (the residual OCRs measured following oligomycin injection minus the OCRs obtained following rotenone/antimycin A injection), the ATP production (the baseline OCRs minus the OCRs obtained following exposing cells to oligomycin injection), and the mitochondrial reserve capacity (the maximal OCR minus the basal OCR). At the end of the Seahorse experiment, to allow a comparison among biological replicates, the data was normalized by cell number, as further described below. Immediately after the assay, the primary cortical neurons were fixed using 4% paraformaldehyde. Post fixation, the primary neurons were treated with DAPI and imaged using ImageXpress^® Nano Automated Imaging System (Molecular devices, USA, Catalog no. )automated microscope, to quantify cell counts for each well of the tissue culture plate as previously described (Soman et al., 2022). The metabolic parameters of the assay basal and maximal respiration, proton leak, and ATP production through oxidative phosphorylation—were calculated by using the Agilent/Seahorse XF Report Generator software and expressed as OCR in pmol/min and normalized to the cell counts obtained by using the ImageXPress Nano system. The results are illustrated as means ± standard error from at least three to six independent experiments performed in triplicate.

Assaying mitochondrial membrane potential:

To measure the transmembrane potential in vitro, isolated mitochondria were stained with TMRM (tetramethyl rhodamine methyl ester) to evaluate the effects of BDNF treatment on mitochondrial transmembrane potential. Briefly, mitochondria isolated from adult mouse brain were plated on a 96 black well plate at 15μg per well. The mitochondria were treated with FCCP (20μM) (Sigma-Aldrich, Catalog no. (Sigma-Aldrich, Catalog no. SRP3014) negative control, malic acid (5mM) (Sigma-Aldrich, Catalog no. (Sigma-Aldrich, Catalog no. 240176) positive control and BDNF (100ng/50ul,150ng/50ul) for time respectively, and loaded with TMRM (200nM) (Sigma-Aldrich, Catalog no. (Sigma-Aldrich, Catalog no. T5428) and incubated for 45 mins at RT. A plate reader was used to measure fluorescence at 573 to 590 nm wavelength.

Mitochondrial reactive oxygen species imaging in primary cortical neurons:

Primary cortical neurons were seeded on 4-well chamber slides (Lab-Tek) at a density of 75,000 neurons per well. The experimental conditions consisted of a control group, a rotenone (SIGMA Aldrich, catalog # 83–79-4) treatment group, and a rotenone treatment combined with a BDNF treatment group. At 5 DIV, the neurons were treated with BDNF (100ng/ml) for 1hr. Subsequently, the neurons were treated with 0.5 μM rotenone for 1 hr. to induce mitochondrial reactive oxygen species production. Finally, the neurons were incubated with 1 μM MitoSox ((Mitochondrial Superoxide Indicators, Invitrogen; catalog M36008) for 20 mins, followed by DRAQ5 (Thermo Scientific^™, USA, Catalog no. 62254) nuclear staining. The maintenance media was exchanged with fresh phenol red-free maintenance media prior to imaging. The images were acquired using an Olympus Fluoview1000 laser-scanning confocal microscope running Fluoview FV10-ASW software (Olympus Corporation, USA, RRID:SCR_014215), using a × 40 X oil objective.

Statistical analysis:

Unless indicated otherwise, results are expressed as mean ± SEM from three independent experiments. Data were analyzed by Student’s t-test (two-tailed) for pairwise comparisons, whereas multiple group comparisons were made by performing a One-way ANOVA followed by Bonferroni-corrected Tukey’s test and Dunnett’s multiple comparison test, using the GraphPad Prism software (version 6.0). P-values less than 0.05 were considered statistically significant. Please note that while no formal power analysis was done to calculate a sample size for all fluorescence single cell imaging experiments shown in figures 2–6 and 8 prior to starting the experiments, the number transfected neurons needed per assay in this study was based on our prior research papers that employed similar techniques which required at a minimum a sample size of 15 transfected neurons per condition to attain statistically significant data (Das Banerjee et al., 2017; Soman and Dagda, 2021).

Results

BDNF receptor TrkΒ localizes to mitochondria.

While it has been shown that BDNF influences mitochondrial structure and function (Markham et al., 2004), the molecular mechanisms by which this neurotrophic factor regulates mitochondrial physiology remains elusive. Given that BDNF stimulates mitochondrial movement (e.g., mitochondrial motility in axons) and TrkΒ complexes have been reported to localize to mitochondria, we wanted to investigate the molecular mechanisms by which BDNF regulates mitochondrial structure/function in the dendrites and the soma of neurons. Here, we hypothesized that BDNF regulates mitochondrial structure and function by promoting the early translocation of TrkΒ to the mitochondria in neurons. To test this hypothesis, we transiently transfected primary cortical neurons with a plasmid that expresses a C-terminally GFP-tagged TrkΒ via Lipofectamine-mediated transfection, in the absence and presence of exogenous treatment with recombinant human BDNF and tracked the localization of TrkΒ -GFP in live neurons via confocal microscopy. Indeed, we observed that a significant proportion of TrkΒ -GFP puncta translocated with mitochondria under basal conditions (Fig. 1 A–E). Furthermore, in the presence of BDNF, we noticed that some of the TrkBTrkB-GFP puncta either colocalized with mitochondria or surrounded mitochondria by forming distinct fluorescent rings that engulfed red fluorescent mitochondria within dendrites and soma (Fig. 1A); hence, given these distinct phenotypes, these GFP puncta that surrounded mitochondria, without showing perfect colocalization, were counted “translocation” events along with colocalized puncta within neurons. By quantifying the mean number of translocation events per neuron, we observed that exogenous treatment of BDNF for 4hrs. in the primary cortical neurons significantly elevated the translocation of TrkΒ to mitochondria in dendrites and the soma compared to untreated neurons (Fig. 1A–C). A less pronounced, non-significant increase in the mean number of translocation events in the dendrites and soma were observed at a later time point (24hrs.) of BDNF treatment (Fig. 1D–E). In the presence of ANA12, a pharmacological inhibitor of TrkΒ receptor, we noted that BDNF treatment for 4 hrs. was unable to significantly increase translocation of TrkΒ to mitochondria in the dendrites or the soma suggesting that binding and activation of TrkΒ is required for its translocation to the mitochondrion. It is worth noting that the significant increase in the mean number of translocation events in neurons treated with exogenous BDNF is not due to an increase in the total number of TrkB-GFP puncta as the mean number TrkB-GFP puncta in the dendrites or soma per neuron was not significantly increased in neurons exposed to BDNF (Suppl. Fig. 1). Overall, our data suggest that exposing neurons to BDNF increases the translocation of TrkΒ receptor to mitochondria, a physiological event that requires the binding of BDNF with Trkβ (Fig. 1B, 1D). In addition, this data is consistent with a previous report that TrkΒ colocalizes with mitochondria under basal conditions (Markham et al., 2004). To confirm the presence of endogenous TrkΒ within the cytosolic and mitochondrial compartments, we isolated mitochondria from the whole brains of 4-month-old mice by subcellular fractionation and immunoblotted for endogenous TrkΒ receptor and mitochondrial markers. In brief, we observed that Trkβ predominantly localizes in isolated mitochondria relative to the cytosolic compartment and highly enriched relative to the total tissue lysate, consistent with the localization of TrkΒ within cell membranes (Fig. 1F; Suppl. Fig. 1). Overall, these observations show that TrkΒ localizes to mitochondria, and its translocation can be enhanced via functional activation of the receptor by BDNF.

BDNF promotes mitochondrial fusion in dendrites in a PKA-dependent manner.

In primary mouse embryonic fibroblasts (MEFs), mitochondrial fusion is a protective mechanism against oxidative stress, given that the increased surface area of mitochondria is associated with decreased probability in the release of apoptotic factors, including cytochrome c (Gomes et al., 2011). Mechanistically, PKA increases mitochondrial fusion by phosphorylating S637 on Drp1 (Kim et al., 2019) and thereby restricting mitochondrial fission. We hypothesized that BDNF activates downstream PKA signaling to regulate mitochondrial structure/function in neurons. Indeed, we observed that exogenous treatment of BDNF in primary cortical neurons rapidly stimulates PKA activity in a rapid manner as it occurred within 5–15 minutes of treatment (Suppl. Fig. 2), a physiological effect that required the activation of TrkB as co-treating neurons with ANA12 negated BNDF-induced PKA activation. In addition, exogenous treatment of primary cortical neurons with BDNF increased the endogenous levels of the PKA scaffolding protein D-AKAP1 as measured by Western blot analyses of cell lysates from SH-SY5Y neuroblastoma cells treated with exogenous recombinant BDNF, indicating that BDNF mediates enhanced localized PKA activity at the mitochondria presumably by increasing the endogenous levels of D-AKAP1 (Suppl. Fig. 3). Next, we hypothesized that BDNF influences mitochondrial remodeling by stimulating PKA-mediated phosphorylation of Drp1 to inhibit its fission activity. In contrast, impeding PKA function is expected to terminate BDNF mediated mitochondrial fusion within dendrites. To test this hypothesis, primary cortical neurons were treated with exogenous BDNF at two time points (4 and 24 hrs.) and transfected with the following panel of plasmids to decrease PKA activity: PKI-GFP (C-terminally GFP tagged plasmid coding for Protein Kinase Inhibitor peptide), OMM-PKI-GFP (also known as Mito-PKI-GFP for the purposes of this paper, PKI targeted to the outer mitochondrial membrane via a TOM20 leader sequence), and mitochondrially targeted RFP (Mito-RFP) for assaying mitochondrial length and distribution using live-cell imaging with a confocal microscope. In brief, we observed that BDNF treatment significantly increased mitochondrial length in dendrites compared to control (Fig. 2B) in a bimodal manner. While exposing primary cortical neurons with 50ng/mL BDNF was able to significantly increase mitochondrial length (Fig. 2B), treatment with a higher concentration of BDNF (75ng/mL) did not significantly increase mitochondrial length showing a bimodal effect. In addition, we observed that treatment of BDNF 4hrs. (Fig. 2C) or 24hrs. (Fig. 2D) significantly increased mitochondrial length of dendritic mitochondria, but not when PKI, a specific endogenous inhibitor of PKA, or Mito-PKI-GFP, was transiently expressed in primary cortical neurons (Suppl. Fig. 4). Overall, these observations show that BDNF increases mitochondrial length in a PKA-dependent manner.

BDNF promotes mitochondrial length via phosphorylation of Drp1.

Thus far, our data suggest that the ability of BDNF to increase mitochondrial length is PKA dependent (Fig. 2, Suppl. Fig. 2, Suppl. Fig. 3). Given that PKA-mediated fusion of mitochondria is associated with PKA mediated phosphorylation of Drp1 at serine 637 (mouse) (Cribbs and Strack, 2007), we hypothesized that BDNF promotes mitochondrial fusion by enhancing PKA mediated phosphorylation of Drp1 and thereby restricting mitochondrial fission. To test this hypothesis, primary cortical neurons were treated with exogenous recombinant human BDNF and co-transfected with either wild-type (WT) Drp1 or a PKA phosphorylation site resistant mutant of Drp1 (with Drp1-S656A), and with Mito-RFP plasmids. 48hrs. post-transfection, the cells were analyzed for mean length and distribution of mitochondria within dendrites (mitochondrial content, as defined by the percentage of a consistent length of a dendrite occupied by mitochondria) of neurons through live cell imaging with a confocal microscope (Fig. 3A). Briefly, while exogenous treatment with BDNF alone increased mitochondrial length compared to control cells, BDNF treatment (4hrs.) partially increased the mean mitochondrial length and mitochondrial content in the presence of WT-Drp1-GFP, but not when Drp1-S656A-GFP was expressed (Fig. 3B, 3C). The data suggest that mitochondrial fission induced by transient expression of both GFP tagged Drp1 constructs can counteract the mitochondrial fusion stimulating effects of BDNF. It is worth noting that the abilities of the Drp1 (S656A) to completely negate the effects of BDNF in increasing mitochondrial length were not attributed to differences in expression of the individual plasmids in transiently transfected primary cortical neurons as both plasmids are expressed to similar levels as noted by their GFP fluorescence (Suppl. Fig. 5). Mechanistically, the data indicates that BDNF promotes mitochondrial fusion in dendrites via PKA mediated phosphorylation of Drp1 at S637. To biochemically verify that exogenous treatment of primary cortical neurons with BDNF promotes mitochondrial fusion via phosphorylation of Drp1, we employed Phos-Tag^™-mediated identification of phosphorylated proteins by performing an SDS-PAGE of cell lysates extracted from primary cortical neurons treated with exogenous human recombinant BDNF. Indeed, we observed that BDNF significantly increased the phosphorylation of Drp1 (Fig. 3D–E) within 4hrs. of treating primary cortical neurons as evident by the presence of a higher molecular weight species of Drp1. However, co-treatment of BDNF with ANA12, a pharmacological inhibitor of Trkβ, significantly reduced the ability of BDNF to phosphorylate Drp1, indicating that the BDNF mediated phosphorylation of Drp1 requires binding and activation of its cognate receptor (Trkβ). By using a phospho-specific antibody for the PKA site in Drp1 (S637), we were able to confirm that exogenous treatment of primary cortical neurons with BDNF increased the ratio of phosphorylated Drp1 relative to total Drp1 but not in the presence of ANA12 (data not shown). Mechanistically, these observations show that BDNF-mediated mitochondrial fusion and increased mitochondrial content (as defined by the percentage of a consistent length of a dendrite occupied by mitochondria) in dendrites involves PKA mediated phosphorylation of Drp1 at serine 637.

BDNF increases anterograde and retrograde mitochondrial trafficking in dendrites.

Mitochondria are dynamic organelles that undergo fission and fusion events but also traffic to areas of high energy demand, specifically in the dendrites and axons of neurons, to aid in functions such as calcium sequestering and to supply ATP (Sheng et al., 2012; Bartolák-Suki et al., 2017; Pallafacchina et al., 2018). In PINK1-deficient neurons, pharmacological and molecular approaches that enhance PKA activity within mitochondria can reverse mitochondrial pathology within dendrites induced by loss of endogenous PINK1, including reversing mitochondrial fission, loss of mitochondrial content in dendrites, decreased transmembrane potential while restoring mitochondrial trafficking in dendrites (Dagda et al., 2011; Das Banerjee et al., 2017). Given that enhanced PKA activity regulates mitochondrial trafficking, and content in dendrites, we hypothesized that BDNF modulates mitochondrial trafficking in dendrites via downstream activation of PKA. To test the hypothesis, we transiently transfected primary cortical neurons with Mito-RFP in the absence and presence of exogenous human BDNF, analyzed the trafficking and distribution of mitochondria by employing a spinning disc confocal microscope, and quantified mean velocity of moving mitochondria by analyzing kymographs assembled from stacks of epifluorescence images (Fig. 4A). In brief, we observed that exogenous treatment of primary cortical neurons with BDNF for 24hrs. significantly increased both anterograde and retrograde movement in dendrites when compared to untreated neurons (Fig. 4B, 4C). Interestingly, exposing primary cortical neurons to a higher concentration of BDNF (75ngs/ml) did not significantly enhance mean mitochondrial velocity compared to 50ngs/ml of BDNF (Fig. 4B, 4C). Overall, these observations show that BDNF enhances mitochondrial trafficking in a bi-directional manner in dendrites.

Figure 4. Exogenous recombinant BDNF treatment increases anterograde and retrograde mitochondrial trafficking to dendrites.

(A) Representative kymographs assembled from untreated primary cortical neurons treated with BDNF at 50ng/ml or 75ng/ml, for 24hrs. The data shows that exogenous administration of BDNF can significantly increase bi-directional movement of mitochondria (μm, x-axis) over a period of five minutes (min., y-axis). The bar graph in (B) depicts the mean (±SEM) mitochondrial anterograde velocity (μm/10s) in dendrites of primary cortical neurons in response to exogenous recombinant human BDNF (50ng/ml, 75ng/ml, 24hrs.) treatment. The graph in (C) depicts mean (±SEM) retrograde velocity of mitochondria in dendrites of primary cortical neurons in response to exogenous recombinant human BDNF (50ng/ml, 75ng/ml, 24hrs.) treatment. The data shows that treating neurons with 50ng/ml BDNF increases anterograde and retrograde trafficking when compared to control. For B (**: p<0.01 vs. Control), for C (*: p<0.05 vs. Control), each dot on the plot represents a mitochondrion, n = 20–40 moving mitochondria/condition from at least 13–18 transfected primary cortical neurons derived per timed pregnant mouse per assay per condition for each assay and pooled from three experiments One-way ANOVA: for B, F=4.6, Rsquared:0.06., 3 comparisons, for C: F=4.56, Rsquared:0.10, 3 comparisons, Tukey’s test with Bonferroni correction, two-tailed test).

BDNF increases anterograde and retrograde mitochondrial trafficking in a PKA dependent manner.

Next, to determine the extent that BDNF regulates mitochondrial trafficking by enhancing PKA activity (Suppl. Fig. 2), primary cortical neurons were treated with exogenous recombinant human BDNF and co-transfected with either PKI-GFP (C-terminally GFP tagged plasmid coding for protein kinase inhibitor peptide), or with OMM-PKI-GFP (mitochondrial targeted PKI), to inhibit PKA in the cytosolic or mitochondrial compartments respectively, and with Mito-RFP (RFP targeted to mitochondria) to label mitochondria. By performing time lapse spinning disc confocal microscopy, we observed that primary cortical neurons (Fig. 5A) treated with BDNF for 4 hrs., significantly increased mitochondrial anterograde movement, but this effect is transient as it is not significant at 24hrs. following treatment (Fig. 5B), consistent with our mitochondrial trafficking data shown in Figure 4. However, co-expression of PKI or of Mito-PKI significantly blocked the ability of BDNF in enhancing anterograde or retrograde mitochondrial movement (Fig. 5A, 5C). Overall, our observations show that bi-directional increase of mitochondrial trafficking in the dendrites is modulated by BDNF via PKA.

Figure 5. BDNF treatment increases anterograde and retrograde mitochondrial trafficking in dendrites in a PKA-dependent manner.

The graph in (A) depicts the mean (±SEM) of mitochondrial anterograde trafficking (μm/10s) in dendrites of primary cortical neurons transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml, 4hrs.), in the presence or absence of transient expression of Mito-PKI-GFP & PKI plasmids when compared to untreated transiently GFP transfected control. The graph in (B) depicts the mean (±SEM) mitochondrial anterograde trafficking (μm/10s) in dendrites of primary cortical neurons transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (100ng/ml, 24hrs.), in the presence or absence of transient expression of Mito-PKI-GFP & PKI plasmids when compared to untreated transiently GFP transfected control. The graph in (C) depicts the mean (±SEM) mitochondrial retrograde trafficking (μm/10s) in dendrites of primary cortical neurons transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml, 4hrs.), in the presence or absence of transient expression of Mito-PKI-GFP & PKI plasmids when compared to untreated transiently GFP transfected control. The graph in (D) depicts the mean (±SEM) mitochondrial retrograde trafficking (μm/10s) in dendrites of primary cortical neurons transfected with Mito-RFP (48hrs.) and treated with exogenous recombinant human BDNF (50ng/ml, 24hrs.), in the presence or absence of transient expression of Mito-PKI-GFP & PKI plasmids when compared to untreated transiently GFP transfected control. For A-D (*: p<0.05, ****: p<0.0001, vs. GFP control, **: p<0.01, ****: p<0.0001, vs. BDNF, each dot on the plot represents a mitochondrion, n = 20–40 moving mitochondria/ condition from at least 13–17 transfected primary cortical neurons derived per timed pregnant mouse per assay per condition for each assay and pooled from 3 experiments One-way ANOVA: for A, F=14.1, Rsquared: 0.29, 15 comparisons, for B, F=2.2, Rsquared: 0.38, 15 comparisons, for C, F=5.75, Rsquared: 0.15, 15 comparisons, for D, F=3.75, Rsquared: 0.07, 15 comparisons, Tukey’s test with Bonferroni correction, two-tailed test).

BDNF increases mitochondrial trafficking through PKA mediated phosphorylation of Miro-2.

Next, we investigated the molecular mechanisms by which BDNF regulates mitochondrial trafficking and content within dendrites. We have previously shown that mitochondrial PKA (D-AKAP1/PKA) phosphorylates the mitochondrial adaptor protein Miro-2 to stimulate bi-directional mitochondrial trafficking and mitochondrial content within dendrites of primary cortical neurons (Das Banerjee et al., 2017). To this end, we hypothesized that BDNF enhances bi-directional mitochondrial trafficking and content via PKA-mediated phosphorylation of Miro-2. To test our hypothesis, primary cortical neurons were treated with exogenous recombinant human BDNF in the presence or absence of ANA12, a pharmacological inhibitor of Trkβ, and analyzed for the phosphorylation of Miro-2 by performing Western blot/Phos-tag assays. Given that commercial antibodies that recognize the PKA phosphorylation sites in Miro-2 are currently not available, performing Western blot/Phos-tag assays is a well validated biochemical alternative to analyze the ability of BDNF, and other pharmacological compounds, to phosphorylate Miro-2. In brief, we observed that exogenous treatment of primary cortical neurons with BDNF for 4hrs. rapidly increased the phosphorylation of Miro-2 (Fig. 6A, 6B) as noted by the presence of higher molecular weight species detected with an anti-Miro-2 antibody. However, BDNF-mediated phosphorylation of Miro-2 was abolished in the presence of ANA12 indicating that BDNF promotes phosphorylation of Miro-2 by coupling with the TrkΒ receptor (Fig. 6A, 6B). Next, to determine whether the ability of BDNF to increase mitochondrial content within dendrites requires PKA activity, primary cortical neurons were co-transfected with PKI-GFP or GFP-OMM-PKI, and with Mito-RFP to visualize dendritic mitochondria by time lapse confocal microscopy. In brief, we observed that mitochondrial content (% of dendrites occupied by mitochondria) in the dendrites was increased significantly with 4hrs. BDNF treatment but not when PKI or Mito-PKI was co-expressed (Fig. 6C–D), suggesting that BDNF increases mitochondrial content by stimulating PKA activity within mitochondria. In another set of experiments, we transiently transfected primary cortical neurons with Mito-RFP and treated the neurons with exogenous BDNF, in the presence or absence of H89, a pharmacological inhibitor of PKA. Briefly, consistent with our transient transfection data (Fig. 6C–D), we observed that pharmacological inhibition of PKA by treatment of primary cortical neurons with H89 ceased the ability of BDNF to induce mitochondrial trafficking within dendrites (Fig. 6E, 6F). Overall, our collective imaging data shows that BDNF enhances bi-directional mitochondrial trafficking and content within dendrites in a PKA-dependent manner, presumably via mitochondrial PKA-mediated phosphorylation of Miro-2.

BDNF increases bioenergetics in live neurons and in isolated mitochondria.

Determining the bioenergetic status of neurons is a representative measurement of neuronal health (Theurey et al., 2019). BDNF mimetics can enhance mitochondrial respiration and mitochondrial biogenesis in skeletal muscles suggesting that BDNF directly regulates mitochondrial function (Wood et al., 2018). Given that BDNF regulates mitochondrial morphology, content and trafficking within dendrites in a PKA dependent manner (Fig. 2, Fig. 5) and the fact that exogenous BDNF can increase complex I activity in mitochondria isolated from brain-derived synaptosomes (Markham et al., 2004), we next hypothesized that exogenous treatment of primary cortical neurons with BDNF increases oxidative phosphorylation in a PKA dependent manner and by binding to the TrkΒ receptor. To test this hypothesis, primary cortical neurons plated on the Seahorse XF24 tissue culture plates were treated with exogenous BDNF at different time points in the presence or absence of H89 (pharmacological inhibitor of PKA), with/without ANA12 (pharmacological inhibitor of B) or co-treated with BDNF. Mitochondrial respiration and glycolysis were then measured by using an XF24eBioAnalyzer as previously published (Lujan et al., 2016; Grigoruţă et al., 2020; Soman et al., 2022). In brief, we observed exogenous treatment of primary cortical neurons with BDNF rapidly and significantly elevated oxygen consumption rates (OCRs) after 2hrs. of treatment. Specifically, exogenous treatment of primary cortical neurons with BDNF significantly increased both basal OCRs (Fig. 7A, 7C) and ATP-linked OCRs (Fig. 7B), while having no significant effects on maximal OCRs, or proton leak-associated OCRs (Fig. 7D, 7E). Unlike treatment with BDNF alone (Fig. 7B), co-treatment of primary cortical neurons with BDNF with ANA12 did not significantly increase either basal or ATP-linked OCRs (Suppl. Fig. 6). On the other hand, co-treating primary cortical neurons with H89 modestly reduced both basal and ATP linked OCRs suggesting that BDNF enhances OXPHOS in manner that does not require global PKA activity (Fig. 7B–C). Although ATP linked OCRs is a proxy of ATP production, the data suggest that BDNF may positively influence ATP production through the ATP synthase. In contrast, we observed that co-treating primary cortical neurons with ANA12 completely negated the effects of BDNF in stimulating basal or ATP-dependent OCRs, suggesting that BDNF regulates mitochondrial respiration by binding and activating TrkB receptor. Interestingly, we observed that exogenous treatment of primary cortical neurons with BDNF significantly elevated basal glycolysis (Fig. 7F, 7G) while only a non-significant effect was observed for maximal glycolysis as analyzed by measuring the extracellular acidification rates (ECARs) (Fig. 7E), a proxy or indirect measure of glycolysis that has been validated in primary neurons to be blunted by glycolysis inhibitors ( e.g. 2-deoxyglucose) (Lujan et al., 2016). . Overall, these data suggest that BDNF affects the overall bioenergetic states of neurons by enhancing both glycolysis and OXPHOS. To complement the mitochondrial respiration assays and to further corroborate the effects of BDNF on mitochondrial physiology, we measured the effects of treating isolated mitochondria derived from 4-month-old wild-type mice on transmembrane potential by using a multimode plate reader. In brief, whole brain mitochondria were co-treated with the red, potentiometric fluorescent dye TMRM to fluorescently quantify the mitochondrial transmembrane potential and with either exogenous recombinant BDNF (5 min.), treated with malic acid as a positive control for enhancing transmembrane potential, or with FCCP as a negative control to collapse transmembrane potential. The mean TMRM fluorescence was kinetically measured for up to 1hr. at 579/590 nm by using a multimode plate reader. In brief, we observed that exogenous treatment of isolated brain mitochondria with both 100ng/50uL of BDNF enhanced the TMRM fluorescence at mitochondrial membrane compared to untreated mitochondria. Indeed, treatment of isolated mitochondria with 100ng/50μL, but not with 150ng/50μL of BDNF, enhanced TMRM fluorescence in a time-dependent manner in a similar manner as treatment with malic acid (Fig. 7I). Therefore, in addition to remodeling mitochondria and trafficking in dendrites (Figs. 2–6), our collective bioenergetics data shows that exogenous BDNF directly modulates mitochondrial function via the TrkB receptor (Suppl. Fig. 6), which can localize to mitochondria (Fig. 1).

BDNF regulates mitochondria and protects dendrites from oxidative stress.

Exogenous BDNF is a strong neuroprotective neurotrophin as evident by its ability to reduce or delay neurodegeneration in genetic and chemical models of Parkinson’s disease (PD) and against a myriad of other toxic insults (Bifrare et al., 2005; Jiao et al., 2016; Colucci-D’Amato et al., 2020; Palasz et al., 2020). Like BDNF, elevating PKA signaling in either the mitochondrial (Banerjee et al., 2021) or cytosolic compartments can protect neurons from apoptosis induced by multiple toxic insults (Almeida et al., 2005). To this end, we hypothesized that exogenous treatment of primary cortical neurons with BDNF protects dendrites and mitochondria against oxidative stress induced by rotenone, a complex I inhibitor used to chemically model PD in vivo and in cultured primary neurons (Merrill et al., 2011). To test this hypothesis, primary cortical neurons were treated with exogenous BDNF and with rotenone at 20nM, a dose known to induce dendrite degeneration without causing overt cell death for 24hrs. (Das Banerjee et al., 2017). Following treatments, primary cortical neurons were fixed and stained with antibodies specific MAP2B to label dendrites and with TOM20 to label mitochondria respectively. The extent of dendrite loss was measured by analyzing for mean dendrite length of MAP2B positive structures, and for mitochondrial loss by measuring the mean intensity of somatic and dendritic mitochondria labeled with anti-TOM20 antibodies. In brief, while rotenone drastically reduced dendrite length and mitochondrial content in primary cortical neurons as expected (Fig. 8A–C), co-treatment of primary cortical neurons with exogenous recombinant BDNF significantly reversed both the loss of dendrites and of mitochondria induced by rotenone treatment (Fig. 8A, 8C).. Rotenone that can elicit an increase in the level of mitochondrial derived superoxide by blocking the electron flow from complex I to complex II (Sherer et.al., 2003). To identify the molecular mechanisms by which BDNF protects dendrites from rotenone mediated toxicity, we surmised that exposing neurons to pretreatment with BDNF may reduce the level of superoxide generated by dendritic mitochondria when exposed to complex I inhibitor rotenone. To measure the level of superoxide, we stained untreated live primary cortical neurons or treated (with BDNF or pretreated with BDNF and rotenone), with the red fluorescence permeable dye MitoSOX. The superoxide levels were analyzed by confocal microscopy and ROS levels were measured by image analyses of the integrated density of MitoSOX stained neurons and normalized to the number of DRAQ5-stained nuclei per epifluorescence field. In brief, we observed that pre-treating primary cortical neurons with BDNF for 4 hrs. efficiently reduced the level of superoxide induced by rotenone to the same extent as untreated cells (Fig. 8 D–E). Overall, these observations show that BDNF not only can significantly reverse the loss of dendrites, a measure of early stages of neurodegeneration, but can also protect mitochondria against overt oxidative stress induced by complex I inhibitor of mitochondria.

Discussion

Over the past two decades, a wealth of scientific data has identified canonical physiological roles of BDNF, when bound to the TrkB receptor, which include promoting synaptic plasticity, dendrite outgrowth, and neuronal survival in numerous brain regions including the cerebral cortex, hippocampus, olfactory bulb, basal forebrain, mesencephalon, brainstem and spinal cord (Lipsky and Marini, 2007; Bathina and Das, 2015; Shen et al., 2018). Although the canonical roles of BDNF are associated with touting signaling pathways in the cytosol and cell membrane, novel non-canonical roles of BDNF in other compartments have been identified including modulating different aspects of mitochondrial behavior and function (Markham et al., 2014; Marosi and Mattson, 2014). For instance, BDNF modulates mitochondrial trafficking of proteins and mRNA in dendrites (Righi et al., 2000), increases mitochondrial trafficking in the axons (Adachi et al., 2005), and regulates oxidative phosphorylation via modulating mitochondrial complex I and IV activity (Markham et al., 2004). PINK1, a dual localized mitochondrial/cytosolic kinase, interacts with PKA to stimulate intracellular levels of BDNF levels and its extracellular secretion to maintain neuronal functions (Soman et al., 2021). Pharmacological stimulation of PINK1 by treating neurons with kinetin has been shown to remodel dendrite morphology via regulation of PKA (Soman et al., 2021). In that particular study, we showed that the PINK1-PKA signaling axis modulates BDNF synthesis, maturation and its extracellular release to regulate dendrite outgrowth in neurons. Furthermore, PINK1 and PKA work in succession to drive mitochondrial morphology, content, trafficking to dendrites, and ultimately dendrite remodeling (Das Banerjee et al., 2017; Soman and Dagda, 2021). Previously, transient transfections studies in primary neurons have shown that PKA-mediated phosphorylation of Drp1 at S637 inhibits its ability to fragment mitochondria (Dagda et al., 2011), thereby allowing pro-fusion modulators like Mitofusin 1 and 2 (MFN ½) and optic atrophy 1 (OPA1) to fuse the outer mitochondrial membranes (OMM) (Zorzano et al., 2010), and inner mitochondrial membranes (IMM) respectively of mitochondria to promote mitochondrial fusion (Cribbs and Strack, 2007). Overall, these studies raised the possibility that physiological functions of BDNF, which can be modulated by a mitochondrial targeted kinase (PINK1) and PKA, include regulating signaling pathways in the mitochondrion to enhance neuronal health and survival. Here, for the first time, our data support a conceptual model that shows (Fig. 9) which depicts that BDNF regulates multiple aspects of mitochondrial behavior and function including mitochondrial trafficking in dendrites (Fig. 4), mitochondrial shape (Fig. 2), and mitochondrial bioenergetics (Fig. 7), leading to augmented mitochondrial health in neurons, a reduction in oxidative stress and dendrite loss (neurodegeneration) induced by the complex I inhibitor rotenone (Fig. 8). In addition, we have shown that a fraction of TrkB receptor, upon activation by exogenous BDNF, can rapidly and dynamically translocate from the cell membrane to mitochondria as assessed by confocal microscopy in live and in fixed neurons (Fig. 1; Supplementary Fig. S1). It is plausible that TrkB, which may compass various signaling endosomes, elicit the activation of downstream signaling cascades in the proximity of the OMM, and extended BDNF treatment (24hrs.) is required for significant translocation of TrkB to mitochondria to influence mitochondrial mobility, structure and function, presumably by increasing the endogenous levels of D-AKAP1, a mitochondrial-localized scaffold of PKA which redirects the endogenous pool of PKA from the cytosol to the mitochondrion (Suppl. Fig. 3). However, we acknowledge some gaps of our conceptual model. First, futures studies are warranted to determine the extent that translocation of TrkB to mitochondria requires activation of PKA in the mitochondrion. In addition, future studies are needed to identify molecular mechanisms by which BDNF elevates the level of endogenous D-AKAP1 in neurons and to identify the specific sub-compartment that TrkB is localized within the mitochondrion. Given that activation of PKA occurs in a rapid manner, specifically 5 minutes of treating neurons with exogenous BDNF (Figure S2), it is likely that PKA in the cytosol and mitochondrial compartments are activated due to increased cAMP produced via membrane bound adenylate cyclase bound of TrkB at the cell membrane; this concept is partly supported by the fact that co-treating neurons with ANA12 is sufficient to block PKA activity induced by exogenous BDNF (Figure S2). Overall, our data support a conceptual model that shows that BDNF increases the fused state of mitochondria (Fig. 2) presumably by inducing PKA-mediated phosphorylation Drp1 (Fig. 3) in a bimodal manner (Fig. 9). However, the extent that the translocation of TrkB to mitochondria is required for the effects of BDNF on mitochondrial physiology (e.g. mitochondrial fusion and trafficking) requires future experimental verification. Furthermore, our collective immunocytochemical data show that BDNF promotes mitochondrial fusion of dendritic mitochondria and increased mitochondrial content via eliciting PKA mediated phosphorylation of Drp1 and increasing the endogenous levels of D-AKAP1. However, the molecular mechanism by which BDNF promotes PKA mediated phosphorylation of Drp1 remains to be elucidated. It is plausible that BDNF binds to TrkB receptors to increase its mitochondrial translocation to activate PKA, presumably by increasing the localized concentration of cyclic AMP (Fig. 9). Indeed, it has been shown that adenylate cyclizes (AC) have been localized inside the mitochondria (Kumar et al., 2009), and BDNF TrkB may activate mitochondrial ACs. Similarly, it has been reported that stimulation of the mitochondrial sAC–cAMP–inner mitochondrial PKA (mt-sAC) signaling pathway increases mitochondrial respiration and ATP synthesis (Acin-Perez et al., 2009). The physiological implications of the hyper fused state of mitochondria are linked to increased resistance of neurons to oxidative stress, enhanced neuronal survival and reduced mitochondrial dysfunction as evidenced in various in vivo and cell culture models such as Parkinson’s and Alzheimer’s disease and excitotoxicity (Merrill et al., 2011; Meyer et al., 2017). Interestingly, we observed that BDNF significantly increased mitochondrial length at 25ng/ml but not at 37.5ng/ml. It is conceivable that the bi-modal effects of BDNF on mitochondrial fusion in dendrites is caused by PKA-mediated phosphorylation of all the available pool of Drp1 in mitochondrial scission sites within mitochondria, especially in light of the fact that PKA mediated phosphorylation of Drp1 does not inhibit its translocation to mitochondria or assembly of constriction rings (Merrill et al., 2011; Yu et al., 2019). However, future studies are warranted to determine the underlying mechanism by which BDNF promotes PKA mediated phosphorylation of Drp1.

Figure 9. Mechanistic conceptual model on the role of BDNF in regulating mitochondrial structure/function in neurons.

The representative figure shows 1. BDNF increases mitochondrial bi-directional trafficking by increasing PKA activity of phosphorylation of Miro-2. 2. BDNF increases mitochondrial fusion and length by increasing PKA activity of phosphorylation of Drp1. BDNF increases neuronal metabolism (oxidative phosphorylation and glycolsysis) and glycolysis, a process that requires activation of TrkB but independent of PKA activation.Based on our data, our conceptual model shows the mechanisms BDNF uses to regulate mitochondrial health as follows: 1) Binding of BDNF to TrkB, 2) activation of PKA (5–15 minutes) in the cytosol and mitochondrion which may occur the cell membrane by increasing the level of cyclic AMP, 3) increased mitochondrial ATP levels (15mins-2hrs.), 4) increased translocation to mitochondria (4–24 hrs.), 5) increased D-AKAP1 levels (24–48 hrs.) post treatment with BDNF, 6) then increased PKA-mediated phosphorylation of Drp1 and Miro2 leading to increased mitochondrial length (24hrs.), 7) trafficking and content in dendrites (24hrs.). While BDNF-mediated regulation of mitochondrial physiology requires the binding of BDNF to TrkB at the cell membrane and PKA activation, future studies will be required to understand the extent that mitochondrial translocation of TrkB or binding of BDNF to TrkB at the mitochondrion is required for the ability of BDNF to modulate mitochondrial trafficking, fusion and mitochondrial respiration. All responses lead to the physiological implication of increased resistance of mitochondria and dendrites to oxidative stress and increased neuronal survival.

In addition to promoting mitochondrial fusion, exposing primary cortical neurons to exogenous BDNF increased mitochondrial motility in a bi-directional manner, which leads to increased mitochondrial content in dendrites. Increased mitochondrial content, as a virtue of increased anterograde transport of mitochondria to proximal and distal sites of dendrites, is necessary to provide the energy in the form of ATP to maintain dendritic arbors as evident by an increased in oxidative phosphorylation and glycolysis as assessed by performing seahorse metabolic assays in cultured neurons treated with BDNF (Fig.7). To this end, it is conceivable that increased mitochondria availability in the dendrites allows for enhanced energy production necessary for powering critical biological functions such as localized translation of proteins, calcium buffering and signal transduction (Sheng, 2017; Leung et al., 2021). Indeed, given that BDNF enhances mitochondrial fusion and content by eliciting PKA mediated phosphorylation of Drp1, our data is consistent with other published observations that PKA mediated phosphorylation of Drp1 increases the fusion and content of mitochondria in dendrites to increase the number of dendrites while reducing the number of synapses (Dickey and Strack, 2011). Mitochondrial content in dendrites is significantly increased with exposure of neurons to BDNF. It is plausible that BDNF either slows down the turnover of mitochondria (mitophagy) in dendrites and/or drives an increase in biogenesis of mitochondria. Mitochondrial turnover and biogenesis are opposing physiological processes that work in an orchestrated manner to increase the quality and health of mitochondria. Mitochondria become damaged upon buffering high amounts of calcium or experiencing high levels of oxidative stress due to increased metabolic demand. The damaged/effete mitochondria would be sent to the soma to be degraded via mitophagy, which leads to increased mitochondrial biogenesis by employing raw, recycled biological materials/nutrients. To this end, it would be vital to study how mitophagy and mitochondrial turnover may be impacted by BDNF which leads to increased mitochondrial content in dendrites.

In addition, we observed that BDNF-mediated increased mitochondrial motility in dendrites requires PKA activity in the cytosolic and mitochondrial compartments as expressing constructs that express untargeted PKI or mitochondrially targeted PKI abrogated its effects. Furthermore, our collective immunocytochemical data show that BDNF promotes mitochondrial trafficking of dendritic mitochondria and increased mitochondrial content via eliciting PKA-mediated phosphorylation of Miro-2 and increasing the endogenous levels of D-AKAP1 (Fig. S2). However, the molecular mechanism by which BDNF promotes PKA mediated phosphorylation of Miro-2 remains to be elucidated and is recognized that this conceptual gap in our study warrants future studies. It is plausible that BDNF binding to TrkB leads to increased mitochondrial translocation of TrkB, increased PKA and D-AKAP1 levels and phosphorylation of Miro-2 and Drp1, which are PKA substrates. PKA-mediated phosphorylation of Miro-2 has been shown to be a molecular mechanism that drives increases mitochondrial anterograde transport in dendrites (Das Banerjee et al., 2017). In our study, we showed that exposure of primary cortical neurons to exogenous BDNF increased mitochondrial content in the dendrites but not in the presence of PKI or Mito-PKI (Fig. 6). Also, BDNF in the presence of PKA inhibitor H89 was unable to increase mitochondrial anterograde trafficking (Fig. 6). Evidently, BDNF activates mitochondrial PKA, which in turn increases Miro-2 phosphorylation leading to anterograde transport of mitochondria to the dendrites and locations of high energy demand.

BDNF enhances oxidative phosphorylation leading to increased ATP production and metabolic potential. Interestingly, in addition to increasing OXPHOS, BDNF was able to increase glycolysis. There is rationale that may explain how BDNF enhances glycolysis. Previously, other groups have shown that TrkB was shown to interact with the dopamine receptor D1 and adenosine A2AR receptors to enhance the cAMP-PKA signaling which activates downstream effectors leading to aerobic glycolysis (Ishii et al., 2018), also as previously observed in neuronal differentiation when neural progenitor cells (NPCs) were treated with BDNF (20ng/ml) and also significantly increased the levels of PGC-1a (Zheng et al., 2016), a protein known to upregulate hundreds of genes in glycolysis in cardiac cells (Rowe et al., 2010). In other skeletal muscle models, BDNF was shown to affect the muscle fiber program by increasing fast twitch glycolytic muscle which is a way to treat muscle degeneration pathologies (Delezie et al., 2019). Furthermore, pharmacological inhibition of TrkB by treating neurons with ANA12 shunts oxidative phosphorylation and glycolysis, leading to the concept that activated TrkB has a direct effect on mitochondrial bioenergetics. Here, we showed that BDNF modulates mitochondrial bioenergetics, presumably to increase ATP production, as shown by an increase in ATP synthase-derived OCR and increased mitochondrial energy capacity which can be compensatory mechanisms by which neurons elicit to meet instances of high “energy currency” (Fig. 7). However, the molecular mechanisms leading to enhanced bioenergetic flux following TrkBactivation through BDNF is not clear, and further studies delineating functional roles of BDNF in regulating electron transport chain are warranted. In addition, given some of the limitations of our conceptual model (Fig. 9), future in vitro studies are warranted to determine the extent that BDNF-mediated activation of TrkB in mitochondria is required or sufficient to increase mitochondrial respiration. Thus far, the treatment of primary neurons with exogenous BDNF depends on downstream activation of PKA to modulate mitochondrial structure/function, and this concentration of BDNF is associated with neuroprotective effects against oxidative stress. In the rotenone model of neurodegeneration, exogenous BDNF exerted both a reduction in the loss of dendrites and mitochondria in neurons exposed to high levels of oxidative stress induced by the complex I inhibitor rotenone (Fig. 8). In this in vitro chemical model of Parkinson’s disease, the neuroprotective effects of BDNF on dendrites is consistent its ability in reducing the levels of mitochondrial-derived superoxide induced by rotenone in BDNF-treated neurons (Fig. 8C–D). Furthermore, it is conceivable that BDNF increases the bioenergetic landscape of neurons by enhancing the activities of complex I and IV (Markham et al., 2004). Overall, our study shows that exogenous BDNF, by enhancing PKA mediated phosphorylation of OMM localized substrates (Drp1 and Miro-2), can modulate mitochondrial structure and mitochondrial health, effects associated with increased neuronal survival induced by oxidative stress. Given that the levels of BDNF, PKA activity and mitochondrial dysfunction play an etiological role in various brain-degenerative disorders, our work has implications in Parkinson’s disease, Alzheimer’s disease, and other brain-related diseases.

Overall, our study expands on the canonical model of BDNF mediated neuroprotection by remodeling mitochondria and enhancing mitochondrial health in dendrites by inducing the following sequence of molecular events in neurons: 1- Binding of BDNF to TrkB, 2- activation of PKA (5–15 minutes), 3 - increased ox/phos levels including basal and ATP synthase-derived OCR, (15 mins-2 hrs.) and glycolysis, 4- increased translocation to mitochondria (4–24 hrs.), 5-increased D-AKAP1 levels (24–48 hrs.) post treatment with BDNF, 6- then increased PKA-mediated phosphorylation of Drp1 and Miro2 leading to increased mitochondrial length (24hrs.), 7- trafficking and content in dendrites (24hrs.). 8- Finally, BDNF-mediated enhanced mitochondrial health is associated with increased resistance to oxidative stress and increased neuronal survival (Fig. 9).

Abbreviations used:

BDNF

Brain-derived neurotrophic factor

DRP1

Dynamin-related protein 1

D-AKAP1

Dual-specificity A-kinase anchoring protein 1

ECARs

Extracellular acidification rates

MIRO2

Mitochondrial rho GTPase 2

OCR

Oxygen consumption rates

OMM

Outer-mitochondrial membrane

PKA

Protein kinase A

PKI

Protein kinase inhibitor

SDS

Sodium dodecyl sulfate

TrkB

Tropomyosin receptor kinase B