Retrograde transport of Akt by a neuronal Rab5-APPL1 endosome

Abstract

Long-distance axonal trafficking plays a critical role in neuronal function and transport defects have been linked to neurodegenerative disorders. Various lines of evidence suggest that the small GTPase Rab5 plays a role in neuronal signaling via early endosomal transport. Here, we characterized the motility of Rab5 endosomes in primary cultures of mouse hippocampal pyramidal cells by live-cell imaging and showed that they exhibit bi-directional long-range motility in axons, with a strong bias toward retrograde transport. Characterization of key Rab5 effectors revealed that endogenous Rabankyrin-5, Rabenosyn-5 and APPL1 are all present in axons. Further analysis of APPL1-positive endosomes showed that, similar to Rab5-endosomes, they display more frequent long-range retrograde than anterograde movement, with the endosomal levels of APPL1 correlated with faster retrograde movement. Interestingly, APPL1-endosomes transport the neurotrophin receptor TrkB and mediate retrograde axonal transport of the kinase Akt1. FRET analysis revealed that APPL1 and Akt1 interact in an endocytosis-dependent manner. We conclude that Rab5-APPL1 endosomes exhibit the hallmarks of axonal signaling endosomes to transport Akt1 in hippocampal pyramidal cells.

Figure 1

Rab5 endosomes display long-range movement. Primary hippocampal neurons were grown on a microfluidic chamber and transduced with mCherry-Rab5 lentivirus. Live cell imaging was performed at DIV 15. (A) Arrows point out the processive movement of a Rab5 labeled structure in a time series, bar 16.7 µm. (B) Kymograph shows the displacement of endosomes over time. Cell bodies are on the left side of the image. (C) Kymograph shows bi-directional movements of Rb5-positive endosomes in axon. (D) Distribution of displacements of mCherry-Rab5 labeled endosomes for the anterograde (red histogram) and retrograde (blue histogram) directions. (E) Speed distributions of mCherry-Rab5 labeled structures in the anterograde (red histogram) and retrograde (blue histogram) directions. Each distribution was decomposed into two log-normal components. Anterograde movement (red solid line) consists of two components, a slow component, μ = 0.33 ± 0.014 µm/sec, σ = 0.96 ± 0.04, <v> = 0.53 ± 0.03 µm/sec (dashed red lines) and a fast component, μ = 1.8 ± 0.1 µm/sec, σ = 0.13 ± 0.05, <v> = 1.84 ± 0.1 µm/sec (alternated dashed and dotted red lines). Similarly, retrograde movement (blue solid line) has a slow component, µ = 0.33 ± 0.04 µm/sec, σ = 0.98 ± 0.08, <v> = 0.53 ± 0.08 µm/sec (dashed blue lines) and a fast component μ = 1.6 ± 0.2 µm/sec, σ = 0.57 ± 0.06, <v> = 1.88 ± 0.19 µm/sec (alternated dashed and dotted blue lines). The fast anterograde and retrograde components constitute 2.5% and 30% of total movement events, respectively. The mean speeds for components were calculated by the formula $<v>=\mu .\exp \left(\frac{1}{2}\cdot {\sigma }^2\right)$. (F) Dependency of mean speed of the two log-normal components of endosome retrograde (blue) and anterograde (red) movement on (binned) integral intensity of Rab5, with slow components (square marked curves) and fast components (circle marked curves). Dashed black lines depict mean value of anterograde movement over all intensity bins. The fast retrograde speed is significantly different from the one of fast anterograde, Student’s-t p_value = 0.0014. (F) Dependency of fraction of movement events on (binned) integral intensity of GFP-Rab5 for fast anterograde (red) and retrograde (blue) motions. The t-test revealed significant difference between these dependencies (Student’s-t p_value = 0.004). Error bars indicate SEM. Statistical analysis was performed on 125 processive tracks from 4 movies from different experimental days.

Figure 2

Rab5 effectors localize to axons and dendrites. Primary hippocampal neurons were grown at low density supported by an astrocyte feeder layer. At DIV 14, neurons were fixed and immunostained for phosphorylated neurofilament-1 (pNF) and APPL1 (A), EEA1 (B), Rabankyrin-5 (C) or Rabenosyn-5 (D). The inset highlighted by the arrow shows a respective axon whereupon there is co-localization of APPL1, Rabankyrin-5 and Rabenosyn-5 with pNF but not with EEA1. Bars 10 µm.

Figure 3

APPL1 localization to endosomes is correlated to speed. Primary hippocampal neurons were grown and imaged as in Fig. 1. (A) Arrows point out the processive movement of one APPL1 labeled structure in a time series, bar 16.7 µm. Cell bodies are on the left side of the image. (B) Kymograph shows the displacement of endosomes over time. Statistical analysis was performed on 1603 tracks from 6 movies within the chamber from different experimental days. (C) Displacement distribution of mCherry-APPL1 labeled structures moving in anterograde (red histogram) and retrograde (blue histogram) direction. (D) Speed distributions of mCherry-APPL1 labeled structures in the anterograde (red histogram) and retrograde (blue histogram) directions. Anterograde movement (red solid line) consists of two components, a slow component, μ = 0.085 ± 0.002 µm/sec, σ = 1.4. ± 0.03, <v> = 0.24 ± 0.01 µm/sec (dashed red lines) and fast component, μ = 1.67 ± 0.1 µm/sec, σ = 0.43 ± 0.04, <v> = 1.84 ± 0.1 µm/sec (alternated dashed and dotted red lines). Retrograde movement (blue solid line) has two components, which parameters are: slow component, µ = 0.087 ± 0.002 µm/sec, σ = 1.38 ± 0.02 (dashed blue lines) and fast component μ = 0.94. ± 0.05 µm/sec, σ = 0.69 ± 0.03, <v> = 1.2 ± 0.1 µm/sec (alternated dashed and dotted blue lines). The fast anterograde and retrograde components constitute 6 ± 1.5% and 17 ± 1.5% of total movement events, respectively. (E) Dependency of mean speed of the two log-normal components of endosome retrograde (blue) and anterograde (red) movement on (binned) integral intensity of mCherry-APPL1, with slow components (square marked curves) and fast components (circle marked curves). Dashed black lines depict mean value of retrograde movement over all intensity bins. The fast retrograde speed is significantly different from the one of fast anterograde, Student’s-t p_value = 0.0004. (F) Dependency of fraction of movement events on (binned) integral intensity of mCherry-APPL1 for fast anterograde (red) and retrograde (blue) motions. The t-test revealed significant difference between these dependencies (Student’s-t p_value = 0.0004). Arrowheads: retrograde movement; bars 10 µm. Error bars indicate SEM.

Figure 4

APPL1 endosomes transport signaling cargo. Primary hippocampal neurons grown on a microfluidic chamber were transduced with mCherry-APPL1 and GFP-TrkB lentivirus. All live cell imaging experimens were performed at DIV 15. (A) Kymographs representing movie 4. Cell bodies are localized towards the left side of the image. Yellow and blue arrows on TrkB panel point to anterograde and retrograde directed tracks respectively. White arrow heads on merge image points on APPL1 fluctuation on APPL1- and TrkB- double-positive tracks. (B) Speed distributions of GFP-TrkB labeled structures in the anterograde (red histogram) and retrograde (blue histogram) directions. Anterograde movement (red solid line) consists of two components, a slow component, μ = 0.17 ± 0.01 µm/sec, σ = 1.1 ± 0.04, <v> = 0.31 ± 0.02 µm/sec (dashed red lines) and a fast component, μ = 2.16 ± 0.3 µm/sec, σ = 0.53 ± 0.12, <v> = 2.5 ± 0.45 µm/sec (alternated dashed and dotted red lines). Retrograde movement (blue solid line) has a slow component, µ = 0.2 ± 0.02 µm/sec, σ = 1.08 ± 0.06, <v> = 0.36 ± 0.04 µm/sec (dashed blue lines) and a fast component μ = 1.27 ± 0.24 µm/sec, σ = 0.53 ± 0.14, <v> = 1.5 ± 0.3 µm/sec (alternated dashed and dotted blue lines). (C) Speed distributions of GFP-TrkB and APPL1 double labeled structures in the anterograde (red histogram) and retrograde (blue histogram) directions. (D) Dependency of slow component velocity of double-positive mCherry-APPL1 and GFP-TrkB endosomes on integral intensity of mCherry-APPL1 per endosome for retrograde (blue) and anterograde (red) directions. (E) Primary hippocampal neurons grown on a microfluidic chamber were transduced with mCherry-APPL1 and GFP-Akt lentivirus. Time series and corresponding kymograph are shown. Bars 10 µm.

Figure 5

APPL1 and Akt1 interact at the neuronal growth cone. Hippocampal neurons were plated at low density and transduced with lentivirus to express the donor GFP-Akt1 alone or in the presence of the acceptor protein fused to mCherry, mCherry-APPL1 or mCherry-APPLΔPTB a truncated form of APPL1 lacking Akt binding domain. At DIV 1, neurons were transferred to an astrocyte feeder layer. Live cell FLIM measurements were performed at DIV 7. (A) Neurons expressing GFP-Akt1 alone display higher GFP lifetimes (yellow-red range) when compared to neurons expressing GFP-Akt1 and mCherry-APPL1 (green-blue range). The decrease in lifetime of GFP-Akt is evident in the axons. (B) FRET efficiency at the cell body and growth cones when GFP-Akt1 was co-expressed with mCherry-APPL or with mCherry-APPLΔPTB a truncated form of APPL1 lacking Akt binding domain. Error bars represent SEM from 3 independent experiments totalizing 60 cell measurements (*p < 0.001). (C) Dynasore treatment abolishes the decrease in lifetime of GFP-Akt1 in the presence of mCherry-APPL1. (D) Lifetime measurements of GFP-Akt1 alone or in the presence of mCherry-APPL1 before and after dynasore treatment. Error bars represent SEM from 3 independent experiments totalizing 90 cell measurements.