Mechanism of neuroprotective mitochondrial remodeling by PKA/AKAP1

Abstract

Mitochondrial shape is determined by fission and fusion reactions catalyzed by large GTPases of the dynamin family, mutation of which can cause neurological dysfunction. While fission-inducing protein phosphatases have been identified, the identity of opposing kinase signaling complexes has remained elusive. We report here that in both neurons and non-neuronal cells, cAMP elevation and expression of an outer-mitochondrial membrane (OMM) targeted form of the protein kinase A (PKA) catalytic subunit reshapes mitochondria into an interconnected network. Conversely, OMM-targeting of the PKA inhibitor PKI promotes mitochondrial fragmentation upstream of neuronal death. RNAi and overexpression approaches identify mitochondria-localized A kinase anchoring protein 1 (AKAP1) as a neuroprotective and mitochondria-stabilizing factor in vitro and in vivo. According to epistasis studies with phosphorylation site-mutant dynamin-related protein 1 (Drp1), inhibition of the mitochondrial fission enzyme through a conserved PKA site is the principal mechanism by which cAMP and PKA/AKAP1 promote both mitochondrial elongation and neuronal survival. Phenocopied by a mutation that slows GTP hydrolysis, Drp1 phosphorylation inhibits the disassembly step of its catalytic cycle, accumulating large, slowly recycling Drp1 oligomers at the OMM. Unopposed fusion then promotes formation of a mitochondrial reticulum, which protects neurons from diverse insults.

Figure 1. PKA activators rapidly induce mitochondrial fusion.

TMRM or MitoTracker-stained PC12 cells (A–C) or hippocampal neurons (D) were treated for the indicated times with either vehicle or the listed compounds and mitochondrial morphology was quantified by blinded comparison to reference images (B–D, means ± S.E.M.). (A) Representative epifluorescence images show formation of interconnected mitochondria upon treatment of PC12 cells with forskolin/rolipram (forsk/roli, 20 µM/1 µM, 3 h). In the representative experiment shown in (B), rapid mitochondrial elongation by 50 µM forskolin is not affected by 100 µg/ml cycloheximide to inhibit protein synthesis. Long-term (20 h) forskolin or a cell-permeant cAMP analog (200 µM cpt-cAMP) promotes mitochondrial fusion in PC12 cells (C) and hippocampal neurons (D); summaries of three independent experiments with 20–30 cells per condition are shown.

Figure 2. Inducible inhibition of outer mitochondrial PKA antagonizes mitochondrial fusion and survival.

(A) Lysates of clonal PC12 cell lines (clone numbers listed) expressing outer-mitochondrial GFP-PKI (omPKI) or GFP-inhibitor-2 (omInh2) from a doxycycline (Dox)-inducible promoter were treated ± Dox (1 µg/ml, 48 h) or forskolin (forsk, 10 µM, 2 h) and immunoblotted for GFP, PKA substrates (RXX[pS/pT] antibody), and PP2A/C as a loading control. (B–D) PC12::omPKI and omInh2 cells were analyzed for mitochondrial morphology ± Dox induction for 2 d (representative confocal images of live cells stained with TMRM (B), reference image based length scores in (C), and digital morphometry in (D) of the same set of 31–38 cells). (E, F) PC12 cell lines were treated ± Dox for 2 d to induce omPKI or omInh2, followed by a 2 d challenge with staurosporine or H₂O₂. In (E), viability was scored by a colorimetric assay (tetrazolium reduction to formazan), while apoptotic nuclei were counted in (F). Bar graphs show means ± S.E.M. and are representative of at least three independent experiments. Student's t test comparisons are between ± Dox-treated cultures.

Figure 3. Mitochondrial PKA and AKAP1 promote neuronal survival and oppose mitochondrial fragmentation in vitro and in vivo.

(A) Hippocampal neurons were transfected with the indicated cDNA and shRNA plasmids (AKAP1ΔPKA  =  I310P,L316P [PKA binding defective]; Figure S2C). After 3 d, cells were treated ± 400 nM rotenone for 2 d, fixed, and analyzed by counting apoptotic nuclei in the transfected neuron population (means ± s.e.m. of n = 3–7 experiments). The inset shows two transfected neurons (green) with apoptotic nuclei and an untransfected neuron with normal nucleus (asterisk). (B, C) Representative confocal sections of TMRM-stained (mito) hippocampal neurons (B) and their mitochondrial length scores (C) 3 d after transfection with the indicated cDNA and shRNA constructs are shown (means ± s.e.m. of n = 3–5 experiments; Student's t test comparisons between GFP fusion proteins and omGFP and between AKAP1 and NS shRNAs). (D–F) Rats injected with lentivirus expressing mitochondrial (m)GFP and AKAP1-GFP into the hippocampus and striatum of left and right hemispheres, respectively, were analyzed 7–14 d later for mitochondrial shape. Perfusion-fixed cryostat sections immunolabeled for GFP (representative confocal image in (D), counterstained for nuclei with TOPRO-3 [blue]) were subjected to ImageJ software-based morphometry. Scatter plots (E, F) correlate form factor (inverse of circularity of individual mitochondria) with cumulative area:perimeter ratio (a measure of network connectivity). Each open symbol represents average shape metrics from 10–22 z-sections of one neuron; filled symbols are population averages (± s.d., 29–42 neurons per condition from 2 (E) and 3 (F) rats).

Figure 4. Mitochondrial restructuring underlies survival regulation by outer mitochondrial PKA.

(A, B) Hippocampal neurons were transfected with the indicated plasmid combinations and scored after 2–3 d for mitochondrial morphology (A, live TMRM stain) or after 5–6 d for apoptosis (B, % transfected neurons with condensed/fragmented nuclei). Means ± s.e.m. of n = 3–6 experiments are shown. (C) Model summarizing effects of PKA/AKAP1 on mitochondrial shape and neuronal survival.

Figure 5. cAMP and PKA/AKAP1 decrease mobility and promote mitochondrial translocation of Drp1.

(A) Confocal micrograph showing mixed cytosolic and mitochondrial localization of GFP-Drp1 in PC12 cells (mito, MitoTracker Deep Red). (B–D) FRAP analysis in PC12 cells shows opposite effects of PKA activation and AKAP1 knockdown on GFP-Drp1 dynamics. PC12 cells co-expressing GFP-Drp1 and either AKAP1-directed or control shRNA were treated ± forskolin/rolipram (25/1 µM, 1–3 h) and Drp1 turnover was measured by bleaching mitochondrial GFP-Drp1 in a 5×5 µm square and monitoring fluorescence recovery at 5 s intervals. (B) shows frames from representative cells (control, forsk/roli: control shRNA ± forskolin/rolipram; shAKAP1: AKAP1 shRNA #1), (C) shows averaged fluorescence recovery curves, and (D) plots Drp1 turnover as the ratio of mobile fraction (mFx) and 50% recovery time (t_1/2) derived from biexponential fits (R ²∼0.99) of individual recovery curves (means ± s.e.m. of 8–10 cells for each condition from a representative experiment). (E–F) Subcellular fractionation of Drp1. COS cells co-expressing GFP-Drp1 with either outer mitochondrial (om) PKA (+) or omGFP (−) were permeabilized with digitonin (500 µg/ml) and fractionated into a cytosolic (cyto) and a heavy membrane fraction containing mitochondria (mito). Fractions were immunoblotted for total Drp1, phospho-Ser_PKA Drp1 (pDrp1), and the mitochondrial marker TOM40 and analyzed by densitometry (E, representative blot; F, summary showing means ± s.e.m. of 6 independent experiments).

Figure 6. PKA shapes mitochondria through phosphorylation of Drp1 at Ser_PKA.

(A, B) PC12 cells expressing mitochondrial GFP (green) and either wild-type or S_PKAA-mutant Drp1 instead of endogenous Drp1 were treated ± forskolin/rolipram (25/2 µM, 3 h), fixed, and epifluorescence micrographs (representatives in A) were subjected to digital morphometry (means ± s.e.m. of ∼300 cells per condition from a representative experiment). (C–D) HeLa cells co-expressing the indicated constructs (om, outer mitochondrial) were fixed and processed for immunofluorescence for mitochondrial cytochrome oxidase II (mito, red) and GFP (green). Shown are representative images (C) and mitochondrial morphology analysis (D, means ± s.e.m. of ∼200–300 cells per condition from a representative experiment). (E–G) HeLa cells expressing WT or S_PKAD-mutant GFP-Drp1 were incubated for up to 4 h with the PKA inhibitor H89 (20 µM) and analyzed by quantitative immunoblotting for phosphorylated (pDrp1, Ser_PKA, and Ser_CDK) and total Drp1 (E) or by immunofluorescence for mitochondrial morphology (F, representative image; G, means ± s.e.m. of ∼200 cells/condition from a representative experiment). For immunoblot analysis only (E), trace amounts (5% plasmid) of PKA catalytic subunit were cotransfected to increase the signal strength with the phospho-Ser_PKA Drp1 antibody.

Figure 7. AKAP1 promotes mitochondrial elongation and neuronal survival by enhancing phosphorylation of Drp1 at Ser_PKA.

(A, B) COS cells co-expressing GFP-Drp1 and either empty vector, wild-type, or ΔPKA-mutant GFP-AKAP1_1–524 were stimulated for 45 min with increasing concentrations of forskolin/rolipram (rolipram at 1/30^th of the indicated forskolin concentrations), and total cell lysates were probed for phospho-Ser_PKA Drp1 and GFP (detecting Drp1 and AKAP1) on the same blot using a dual-channel infrared imager. Drp1 phosphorylation was quantified as the ratio of phospho- to total Drp1 signals normalized to the highest value and is shown as part of the representative blots (A) and the summary graph (B, means ± s.e.m. of 6 independent experiments). (C, D) HeLa cells co-expressing the indicated constructs (ΔPKA  =  I310P, L316P-mutant) were treated with forskolin/rolipram (25/2 µM, 3 h) and processed for immunofluorescence for mitochondrial cytochrome oxidase II (mito, red) and V5-tagged AKAP1 (green). Shown are representative images (C) and mitochondrial morphology analysis (D, mean ± s.e.m. of ∼200–300 cells per condition from a representative experiment). (E, F) PC12 cells were cotransfected with GFP-Drp1/shRNA expression plasmids and either AKAP1 cDNAs or shRNAs. Cultures were treated 48 h posttransfection with 0.5 µM staurosporine for 24 h, fixed, and stained with Hoechst 33342. Representative images in (E) show overlays of cell contours (bright field), GFP-Drp1 (green), and Hoechst-stained nuclei (blue), with transfected cells displaying normal nuclear morphology and many untransfected cells having condensed, apoptotic nuclei (arrows). Apoptosis was quantified as the percentage of GFP-positive cells with condensed or fragmented nuclei (E, means ± s.e.m. of 6–13 image fields with ∼300 transfected cells per condition from one experiment representative of three).

Figure 8. PKA/AKAP1 recruits Drp1 into large, slow turnover complexes by phosphorylating Ser_PKA.

(A–C) HeLa cells co-expressing wild-type or S_PKAA-mutant GFP-Drp1 and either wild-type or PKA-binding deficient (ΔPKA) AKAP1 were subjected to FRAP analysis. (A) Frames showing time lapse series of representative cells (green: GFP-Drp1, red: MitoTracker Deep Red) with enlargements of the 180 s frame demonstrating recovery of Drp1 into mitochondrial foci (arrows). (B) Average recovery curves of cells expressing wild-type GFP-Drp1 and either wild-type or mutant AKAP1, and (C) plots mitochondrial Drp1 turnover as the 50% recovery time calculated from biexponential fits of individual recovery curves (R ²∼0.992; means ± s.e.m. of 5–9 cells each from a representative experiment). (D) COS cells cotransfected with Drp1 and either omGFP or omPKA were treated for 5 min with the indicated concentration of the reversible, membrane permeant crosslinker dithiobismaleimidoethane (DTME), and cleared cell lysates were subjected to ultracentrifugation to sediment Drp1 complexes.

Figure 9. A GTPase-compromised Drp1 mutant accumulates in slowly recycling complexes at mitochondria.

(A) Drp1 domain diagram. (B–C) Endogenous Drp1 in HeLa cells was replaced by transfection of the indicated ratios of wild-type (WT) and T55D mutant GFP-Drp1 expression plasmids. Cells were fixed and analyzed for mitochondrial length and Drp1 colocalization with mitochondria using ImageJ software ,. (B) Representative images showing punctate localization of GFP-Drp1 T55D (green) on mitochondria (cytochrome oxidase subunit II antibody). (C) Correlation between mitochondrial length and mitochondrial localization of Drp1 as a function of Drp1 T55D expression (means ± s.e.m. of ∼200 cells per condition from a representative experiment). Wild-type and T55D-mutant Drp1 expression levels are similar and scale with plasmid amounts (inset). (D, E) Turnover of GFP-Drp1 WT and T55D was analyzed by FRAP as in Figure 8. (D) Frames showing representative time lapse series (green: GFP-Drp1, red: MitoTracker Deep Red, see Video S2) of HeLa cells expressing 100% WT or T55D Drp1, with the last frame expanded to show recovery of WT Drp1 into mitochondrial foci (arrows). (E) Average recovery curves (left) and curve fit-derived turnover (right, ratio of mobile fraction [mFx] and 50% recovery time [t_1/2]) from cells expressing varying ratios of WT and T55D-mutant GFP-Drp1 (mean ± s.e.m. of 12–20 cells each from a representative experiment). (F) COS cells expressing GFP-Drp1 WT or T55D were incubated with DTME (5 min, 500 µM), and cleared cell lysates were subjected to ultracentrifugation to sediment Drp1 complexes (∼2-fold increase with the T55D mutation).

Figure 10. PKA phosphorylation and T55D mutation prolong the lifetime of mitochondrial Drp1 foci.

(A, B) HeLa cells transfected with GFP-Drp1 (green), dsRed2/mito (COX8 matrix targeting sequence, red), ± PKA catalytic subunit were imaged for ≥1 h at 37°C, capturing images every 30 s. (A) Representative frames of time lapse series (see Video S3). Blue lines connect GFP-Drp1 punctae that could be tracked for at least 5 min; x symbols denote mitochondrial fission events. Note that GFP-Drp1 punctae often split with the fragmenting mitochondrion. (B) Cumulative frequency plot of Drp1 punctae lifetimes and average lifetimes (bar graph inset; means ± s.e.m. of 35–62 cells and 11,000 to 40,000 punctae per condition from two independent experiments). (C) Model of mitochondrial fusion by PKA/AKAP1. GTP-bound Drp1 translocates to mitochondria to assemble into oligomeric complexes. Drp1 assembly stimulates GTP hydrolysis, leading to mitochondrial fission and release of Drp1 into the cytosol to complete the cycle. OMM-anchored PKA/AKAP1 phosphorylates Drp1 at Ser_PKA, stabilizing the GTP-bound state to promote growth of Drp1 complexes to a size that is incompatible with membrane scission. Protein phosphatases (PP), including calcineurin, dephosphorylate Drp1 Ser_PKA to return Drp1 into its active, rapidly cycling state.