Calcineurin and huntingtin form a calcium-sensing machinery that directs neurotrophic signals to the nucleus

Abstract

When a neurotrophin binds at the presynapse, it sends survival signals all the way to the nucleus on signaling endosomes. These endosomes fuel their own journey with on-board glycolysis—but how is that journey initiated and maintained? Using microfluidic devices and mice, we find that the calcium released upon brain-derived neurotrophic factor (BDNF) binding to its receptor, tropomyosin receptor kinase B (TrkB), is sensed by calcineurin on the cytosolic face of the endosome. Calcineurin dephosphorylates huntingtin, the BDNF scaffold, which sets the endosome moving in a retrograde direction. In an in vitro reconstituted microtubule transport system, controlled calcium uncaging prompts purified vesicles to move to the microtubule minus end. We observed similar retrograde waves of TrkA- and epidermal growth factor receptor (EGFR)-bearing endosomes. Signaling endosomes in neurons thus carry not only their own fuel, but their own navigational system.

Fig. 1.. CaN is enriched on the surface of vesicles, colocalizes with HTT on Rab5⁺ endosomes, and interacts with HTT in vivo.

(A) Mass spectrometry analysis of vesicles purified from Thy1:p50-GFP mouse brains shows CaNA ranks highly among vesicle-associated proteins. LC-MS/MS, liquid chromatography–tandem mass spectrometry; KIF5C, kinesin family member 5C. (B) (i) Airyscan microscopy of isolated axons identified by the postfixation incubation with Alexa Fluor 488 conjugate of wheat germ agglutinin (WGA). Scale bar, 2 μm. (ii) Line scan analysis (left) and (iii) quantification of colocalization (right). The graph shows the endogenous colocalization of CaN with HTT compared to random colocalization in the same portion of axon (CT) (see Materials and Methods). a.u., arbitrary unit of normalized fluorescence intensity. Wilcoxon matched-pairs signed-rank test, **P = 0.021; n = 35 axons. Scale bar, 1 μm. (C) Confocal and two-dimensional stimulated emission depletion (2D-STED) images of free-cultured neurons at 5 days in vitro (DIV5) showing the colocalization of HTT and CaN. Scale bars, 1 μm. (D) Immunogold labeling of vesicles isolated from mouse brain shows the presence of both CaN (5-nm gold particles) and HTT (15-nm gold particles) on vesicles with a higher percentage of colocalization in large vesicles (diameter > 150 nm versus small vesicles < 60 nm). EM, electron microscopy. Mann-Whitney test, *P < 0.05; n_CTRL = 188 and n_CaN = 209. Scale bars, 50 nm. (E) Vesicular fractions (P3) were incubated with increasing concentrations of Proteinase K (PK) to digest proteins present on the surface of vesicles; only BDNF remains, meaning that it is within the vesicle. The predicted molecular weights (MWs) are indicated on the left, in gray. αSyn, α-synuclein.

Fig. 2.. CaN and HTT are present on TrkB signaling endosomes.

(A) Western blot of Thy1:p50-GFP brain fractions after successive centrifugations. The light membrane–enriched fraction (P3) contains endosomes, as shown by the presence of Rab5, HTT, CaN, and TrkB. (B) Immunoprecipitation (IP) of the endogenous Rab5 (αRab5) in endosomes purified from Thy1:p50-GFP mouse forebrains. Immunoglobulin G (IgG) served as a negative control. (C) 2D-STED microscopy of neurites shows the colocalization of Rab5 and CaN immunostaining on the same endosome. Scale bars, 1 μm. (D) (i) High- resolution microscopy of axons in a microfluidic chamber shows CaN and HTT present on Rab5-positive endosomes (n = 35 axons from two independent experiments). Scale bar, 2 μm. (ii) Line scan analysis and (iii) graph representing the percentage of triple colocalization. (E) Isolated axons were immunolabeled with antibody targeting overexpressed TrkB-mCherry (α mCherry), Rab5-YFP (α GFP), and endogenous CaN (α CaN). Graph at right shows that triple colocalization increased after 5 min of BDNF treatment. Mann-Whitney test, *P < 0.05; n = 50 to 54 from three independent experiments. Scale bars, 2 μm. (F) The plot profile of gray value intensity shows TrkB-mCherry, Rab5-YFP, and CaN peaks coincide, indicating colocalization in isolated axons. (G) 2D-STED microscopy confirmed CaN and TrkB-mCherry colocalization on vesicles. Scale bars, 1 μm. (H) Extracts of purified endosomes were incubated with Rab5 or TrkB antibodies (αRab5 or αTrkB) or immunoglobulin as control (IgG). CaN is enriched in Rab5- or TrkB-associated endosomes immunoisolated from the light membrane fractions of Thy1:p50-GFP mouse forebrains.

Fig. 3.. BDNF induces a wave of TrkB signaling endosomes.

(A) Corticostriatal connections form in microfluidic chambers in vitro. Kymographs, which show vesicle movement in both directions, were acquired at DIV10 to DIV12 when the neurons reach a mature stage, as shown by staining with synaptophysin I (SypI) and postsynaptic density 95 (PSD95). Scale bars, 20 μm (for kymograph) and 2 μm (for SypI/PSD95). LV, lentiviral vector. (B) Kymographs show movement of TrkB-mCherry vesicles before and 5 min after BDNF infusion (50 ng/ml). (C) BDNF treatment led vesicles to change their direction of movement from anterograde to retrograde. Kruskal-Wallis test, Dunn’s multiple comparisons test, *P < 0.05 and **P < 0.01; n = 30 to 40 axons from five independent experiments. (D) Analysis of axonal TrkB-mCherry vesicle transport from T0 and every 2.5 min thereafter. Each dot represents the average number of vesicles observed for each 100 μm of a given axon (n = 30 to 40 axons from five independent experiments). Left: Number of TrkB-containing anterograde vesicles remained fairly steady over time. Right: Number of retrograde vesicles peaked at T5. Kruskal-Wallis test, Dunn’s multiple comparisons test, *P < 0.05 and **P < 0.01. (E) After adding BDNF labeled with QDs (QDs-BDNF) to the synaptic compartment (top), we recorded the movement of TrkB-GFP and QDs-BDNF with dual acquisition at 5 and 75 min (bottom). Scale bars, 10 μm. (F) Distal TrkB-mCherry endosomes colocalized with Rab5-YFP, which is present on early endosomes; BDNF treatment in the synaptic compartment increased colocalization of TrkB endosomes with Rab5-positive vesicles. Unpaired two-tailed Student’s t test, **P < 0.01; n = 30 to 32 axons from three independent experiments. Scale bars, 2 μm.

Fig. 4.. The retrograde wave requires the activation of both TrkB and CaN.

(A) Axons were immunoisolated to analyze colocalization of TrkB-mCherry, Rab5-YFP, and p-TrkB before and after 5 min of BDNF (50 ng/ml) treatment. Mann-Whitney test, *P < 0.05; n = 70 to 80 axons from three independent experiments. Scale bars, 2 μm. (B) Inhibiting TrkB with K252a prevented the retrograde wave of TrkB-mCherry endosomes. Two-way analysis of variance (ANOVA), Sidak’s multiple comparisons test, *P < 0.05; n = 30 to 38 axons from three independent experiments. The graph represents the movement of TrkB-mCherry vesicles after BDNF infusion in control conditions (dark gray bar) and in the presence of the TrkB inhibitor K252a at 10 μM, added 30 min before BDNF infusion (light gray bar). (C) Inhibiting CaN with FK506 (1 μM added 30 min before BDNF) also prevented the retrograde wave. Two-way ANOVA, Sidak’s multiple comparisons test, **P < 0.01 and *P < 0.05; n = 25 to 38 axons from three independent experiments. (D) Depleting CaN with lentiviral infection of shCaNα+β (light gray) blocks TrkB retrograde wave compared to infection with shLuc (dark gray). Two-way ANOVA, Sidak’s multiple comparisons test, **P < 0.01, n = 30 to 42 axons from three independent experiments.

Fig. 5.. CaN-mediated dephosphorylation of HTT S421 is necessary for BDNF to induce a TrkB retrograde wave.

(A) Left: Immunoblot of cytosolic (S3) and vesicular (P3) fractions from three independent WT (HTT-WT), HTT-SA, or HTT-SD mouse brains. Unpaired two-tailed Student’s t test, *P < 0.05; n = 3 to 4 brains per genotype. HTT-SD mice have S421 replaced by aspartic acid (mimicking constitutive phosphorylation); HTT-SA mice have S421 replaced by alanine (mimicking the absence of phosphorylation). Right: HTT phosphorylation allows kinesin-1 (KIF5C) to associate with more vesicles. (B) Movement of vesicles in neurons isolated from HTT-WT (gray) and HTT-SD (lavender) mice. Two-way ANOVA, Sidak’s multiple comparisons test, ****P < 0.0001, **P < 0.01, and *P < 0.05; n = 30 to 40 axons from three independent experiments. The S421D mutation, which blocks HTT dephosphorylation, prevented BDNF from inducing a TrkB retrograde wave.

Fig. 6.. In vivo dendritic relocation of TrkB⁺ endosomes, synaptic maintenance, and enhancement of survival signals require HTT phosphorylation.

(A) Representative photo of the injection performed in the DLS. CTB and AAV-BDNF mCherry were unilaterally injected at p21, and, after 2 weeks, brains were sliced and immunostaining was performed using p-TrkB/PSD95 or p-ERK. (B) The number of p-ERK–positive cells in the cortex of WT and HTT-SD mutant mice was counted in the injected hemisphere and divided by the number of counted positive cells in the noninjected control hemisphere (CT). There were more positively marked cells in WT mice in the hemisphere that received the BDNF injection, but no changes were observed in either hemisphere in the HTT_SD mice. Two-way ANOVA, Sidak’s multiple comparisons test, **P < 0.01; n = 3 to 4 slices per brain and n = 3 brain per genotype. ns, not significant. Scale bars, 20 μm. (C) The number of adjacent p-TrkB/PSD95 spots in cortical dendrites is expressed as a ratio between the injected and the control hemisphere. The number of p-TrkB (Y816) puncta in close proximity to PSD95 protein reveals increased colocalization in WT (HTT-WT) but not HTT-SD mice. DAPI, 4′,6-diamidino-2-phenylindole. Unpaired two-tailed Student’s t test, *P < 0.05; n = 3 slices per brain and n = 3 brains per genotype. Scale bars, 5 and 1 μm (for inset).

Fig. 7.. Calcium-mediated TrkB activation is required to initiate the retrograde wave.

(A) Overexpression of the TrkB-mCherry WT or mutant Y816F in rat cortical neurons shows that lack of phosphorylation at Y816 blocks the retrograde transport of TrkB-mCherry endosomes upon BDNF induction (blue) compared to the WT situation (gray). Two-way ANOVA, Sidak’s multiple comparisons test, **P < 0.01; n = 26 to 38. (B) The GCaMP6f signal rose after addition of BDNF (50 ng/ml). Friedman nonparametric test followed by Dunn’s multiple comparisons test, ***P < 0.001 and ****P < 0.0001; n = 46 axons from four independent experiments. (C and D) Calcium events were not detectable at T0 in WT or Y816-overexpressing axons. BDNF treatment triggered a significant increase in calcium events from T2.5 until T10 in axons expressing TrkB-WT (dark gray) but not TrkB-Y816F (blue). Neurons expressing only endogenous TrkB and to which only GCaMP6f was added served as controls (light gray). Kruskal-Wallis test, Dunn’s multiple comparisons test, ***P < 0.001, **P < 0.01; n = 45 to 55 axons for WT and n = 68 for Y816F axons from at least three independent experiments. (E) There was no difference between WT and mutant (Y816F) axons in TrkB-mCherry and SypI colocalization. Mann-Whitney test, P = 0.1472; n = 56 to 63 neurites). Scale bars, 2 μm. (F) Analysis of vesicle movement in neurons treated with 100 μM 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester (BAPTA-AM) in the synaptic chamber. Two-way ANOVA, Sidak’s multiple comparisons test, *P < 0.05 and **P < 0.01; n = 20 to 37 axons from four independent experiments. Calcium chelation prevented the BDNF-mediated TrkB-mCherry retrograde wave.

Fig. 8.. CaN-HTT on the vesicular surface transduces changes in local Ca²⁺ concentration into retrograde transport.

(A) Microtubules were polymerized in vitro and their polarity marked by different concentrations of rhodamine-labeled tubulin at the minus (−) and plus (+) ends. Motile vesicles purified from offspring of p50-GFP or p50-GFP mice crossed with HTT-SD mice were added to microtubules seeded in flow chambers and their movement analyzed by total internal reflection fluorescence (TIRF) microscopy. (B) Purified vesicles were treated with increasing concentrations of CaCl₂ and Western blot membranes were incubated with the antibody recognizing HTT S421 phosphorylation. One-way ANOVA, Dunnett’s multiple comparisons test, *P < 0.05; n = 3. (C) Rhodamine-labeled microtubules were polymerized in vitro with lower concentrations of red tubulin during microtubule elongation to orient them. The labeled microtubules were then attached to a glass slide covered with anti-tubulin antibodies, incubated with the purified “motile” vesicle fraction in the presence of the calcium caging compound NP-EGTA, and lastly ultraviolet (UV)-flashed to release calcium. (D) Calcium uncaging increased the number of vesicles moving retrogradely in p50-GFP WT but not in p50-GFP × HTT-SD mouse brain. Two-way ANOVA, Sidak’s multiple comparisons test, *P < 0.05; n = 24 to 26 from five independent experiments. Scale bar, 5 μm. (E and F) Axonal TrkA-GFP and epidermal growth factor receptor (EGFR)–GFP trafficking in cortical neurons from 0 to 12.5 min after either (E) NGF (50 ng/ml) or (F) EGF (2.5 ng/ml) was added to the synaptic chamber. Transient retrograde waves were observed, similar to that seen for TrkB-GFP in response to its ligand BDNF. Kruskal-Wallis test, Dunn’s multiple comparisons test and one-way ANOVA, Holm-Sidak’s multiple comparisons test, *P < 0.05, **P < 0.01; n = 20 to 36 axons from three independent experiments.

Fig. 9.. Proposed model for retrograde endosomal TrkB signaling.

Upon BDNF binding, TrkB at the presynaptic membrane is activated via phosphorylation of Y816, triggering release of Ca²⁺ from intracellular storage. This local increase in Ca²⁺ concentration in turn activates CaN, which resides on the surface of BDNF-TrkB–activated signaling endosomes, and it dephosphorylates HTT at S421. This dephosphorylation causes kinesin to be released from the endosome, leaving dynein to mediate the retrograde transport of the endosomes. SVs, synaptic vesicles.