Sorting of Dendritic and Axonal Vesicles at the Pre-axonal Exclusion Zone

Abstract

Polarized sorting of newly synthesized proteins to the somatodendritic and axonal domains of neurons occurs by selective incorporation into distinct populations of vesicular transport carriers. An unresolved issue is how the vesicles themselves are sorted to their corresponding neuronal domains. Previous studies concluded that the axon initial segment (AIS) is an actin-based filter that selectively prevents passage of somatodendritic vesicles into the axon. We find, however, that most somatodendritic vesicles fail to enter the axon at a more proximal region in the axon hillock, herein referred to as the pre-axonal exclusion zone (PAEZ). Forced coupling of a somatodendritic cargo protein to an axonally directed kinesin is sufficient to drive transport of whole somatodendritic vesicles through the PAEZ toward the distal axon. Based on these findings, we propose that polarized sorting of transport vesicles occurs at the PAEZ and depends on the ability of the vesicles to acquire an appropriately directed microtubule motor.

Figure 1. Identification of the pre-axonal exclusion zone

(A) Z-stack reconstruction of confocal sections from DIV10 hippocampal neurons stained for endogenous TfR (red) and AnkG (green). (B, C) Magnified views of axons emanating from the soma (B) or a dendrite (C), and their neighboring dendrites. Graphs on the right show the corresponding TfR and AnkG intensity scans over a 10-μm line running from the soma to the axons or dendrites. In A–C, arrows point to the pre-axonal exclusion zone (PAEZ). (D) Images of axons emanating from the soma in neurons stained as in panel A at different times in culture. The PAEZ is the area delimited by dashed lines (images on the right). The top right picture marked with an asterisk is a higher intensity image of the top left image. (E) Graph representing the AIS area (green), PAEZ area (gray) and dendrite/axon (D/A) polarity index (red bars), calculated from 25 neurons at different days of culture in vitro. D/A values are the mean ± SD. (F) General view of the same DIV3 neuron shown in the top panels of D co-stained with phalloidin (grayscale, bottom) to highlight F-actin in the entire cytoplasm.

Figure 2. Exclusion of somatodendritic vesicles at the PAEZ

(A) Dual-color image of an axon and adjacent dendrite emanating from a live DIV10 neuron expressing TfR-GFP (green) and surface-stained with CF555-conjugated antibody to the AIS marker NF (red). Arrow points to the PAEZ. (B) Grayscale image of the TfR-GFP channel in A showing a photobleaching (PB) box (orange) and tracks (green) running from the soma (S) to the axon (A) or dendrite (D) along which vesicle movement was quantified. (C,D) Dual-color (left) and grayscale (right) kymographs generated from straightened lines along the tracks shown in B. Particles moving in anterograde and retrograde directions appear as lines with negative and positive slopes, respectively. See also Movie S1. (E, F) Corresponding TfR (green) and NF (red) line intensity scans at the 120-s point of the kymographs. (G) Scheme of the neuronal regions in which anterograde transport of TfR-GFP from the soma (S) to the dendrites (D) or axons (A) was quantified along 30 μm-long tracks. (H) Number of vesicles of TfR-GFP entering the neighboring dendrite, PAEZ and AIS over a 120-s period. A total of 216 vesicles in 15 neurons were analyzed. (I) Schematic representation of constructs used in the RUSH system (Boncompain et al., 2012): TfR fused to a streptavidin-binding peptide (SBP) and GFP (reporter), and streptavidin appended with the KDEL ER retention signal (ER hook). Biotin dissociates the reporter from the ER hook and allows for its transport through the secretory pathway. (J) DIV7 neuron co-expressing RUSH system constructs surface-stained for NF (red) and incubated for 20 min with biotin to release the TfR-SBP-GFP reporter from the ER and allow its concentration in the Golgi complex. (K) Grayscale image of the neuron in J showing trajectories of TfR-SBP-GFP vesicles budding from the TGN over a 120-s period from Movie S2. (L) Number of vesicles of TfR-SBP-GFP entering an average dendrite, PAEZ and AIS over a 120-s period. A total of 159 vesicles in 15 neurons were analyzed; values were expressed as percentage ± SEM. * P<0.001.

Figure 3. Axonal exclusion of somatodendritic vesicles in the absence of the AIS

(A) DIV3 neurons were co-transfected with plasmids encoding control shRNA or AnkG shRNA, together with TfR-GFP and mCherry-tubulin, and stained for AnkG on DIV8. (B) DIV3 neurons co-transfected with plasmids encoding control shRNA or AnkG shRNA, together with TfR-GFP, were surface-stained live with CF555-conjugated antibody to GFP prior to fixation and staining for AnkG on DIV8. Enlarged regions of axon are shown at the bottom of each image. Arrows indicate the position of the AIS and arrowheads trace the trajectory of the axon. (C) Graph representing AnkG intensity (%) and dendrite/axon (D/A) ratio for total and surface TfR-GFP quantified from 20 neurons and expressed as mean ± SD. * P<0.001. n.s., not significant. (D) Images of axons from live DIV8 neurons transfected as in A and stained for NF. Grayscale kymographs for TfR-GFP recorded every 500 ms for 120 s were generated from a straightened 30-pixel-wide line along a 50-μm path of an axon. Particles moving in retrograde direction appear as lines with positive slopes. See also Movie S3.

Figure 4. Selective exclusion of somatodendritic organelles and presence of KIF5 microtubule tracks in the PAEZ

(A) Left: Images of the soma-PAEZ-AIS transitional region from DIV10 neurons stained for AIS markers (AnkG or NF) and the indicated organellar markers (see Results section for description of these markers). Right: Corresponding fluorescence line intensity scans. (B) DIV10 neurons expressing GFP-Rigor-KIF5A (green) co-stained for AnkG (blue) and TGN38 (red), and corresponding line intensity scans. (C) DIV3 neurons expressing GFP-Rigor-KIF5A (green) co-stained for acetylated α-tubulin (red) and F-actin (grayscale).

Figure 5. Fusion of a KLC-binding sequence to somatodendritic proteins causes missorting of the chimeras to the axon

(A) Three copies of the KLC-binding sequence (KBS) from SKIP (Pernigo et al., 2013) or an inactive WD-to-AA mutant of it (KBS-mut) were fused to the cytosolic N-terminus of TfR or C-terminus of NiV-F, both tagged with fluorescent proteins (FP) (GFP or mCherry). (B, C) DIV7 neurons co-expressing mCherry-tubulin (Tub) (red) with TfR-GFP (B) or NiV-F-GFP (C), WT (left) or fused to KBS (middle) or KBS-mut (right) were immunostained for AnkG (blue). Enlarged regions of the somatodendritic domain and axon tips are shown in Figure S3. Dendrite/axon (D/A) and axon tip/dendrite (AT/D) ratios were quantified from 25 neurons and expressed as mean ± SD. * P<0.001. (D) DIV7 neurons co-expressing KBS-TfR-GFP (left) or NiV-F-KBS-GFP (right), mCherry-tubulin (Tub) (red) and the dominant-negative mutant KLC TPR-HA (cyan). Dendrite/axon (D/A) and axon tip/dendrite (AT/D) ratios were quantified from 25 neurons and expressed as mean ± SD. * P<0.001. In all images, arrowheads point to the AIS.

Figure 6. NiV-F-KBS and KBS-TfR chimeras redirect other somatodendritic proteins to the axon

(A) Expression of NiV-F-KBS-mCherry (red) redirects TfR-GFP (green) to the axon in DIV7 neurons. NiV-F-KBS-mCherry also redirects endogenous TfR to the axon (Figure S7). Dendrite/axon (D/A) and axon tip/dendrite (AT/D) ratios of wild type NiV-F and TfR proteins in neurons expressing NiV-F-KBS-mCherry, or the corresponding NiV-F-KBS-mut constructs. (B) Expression of KBS-TfR-mCherry (red) redirects NiV-F-GFP (green) to the axon in DIV7 neurons. Dendrite/axon (D/A) and axon tip/dendrite (AT/D) ratios of TfR and NiV-F proteins in neurons expressing KBS-TfR-mCherry, or the corresponding KBS-TfR-mut constructs. (C) Dendrite from a DIV10 neuron showing co-localization of TfR-mCherry (red) and NiV-F-GFP (green) to the same vesicles. (D) Grayscale kymographs for TfR-mCherry (left panel) and NiV-F-GFP (middle panel) corresponding to a DIV10 neuron recorded every 500 ms for 70 s were generated from a straightened 20-pixel-wide line along a 30-μm path of a dendrite. Co-movement of particles is indicated as yellow tracks on the right panel. (E) DIV7 neurons co-transfected with plasmids encoding mGluR1-GFP (grayscale in the middle and green in the right merged images) plus KBS-TfR-mCherry or KBS-TfR-mut-mCherry (grayscale in the left and red in the right merged images). Notice the redistribution of mGluR1-GFP to axons and axon tips upon KBS-TfR-mCherry but not KBS-TfR-mut expression. Dendrite/axon (D/A) and axon tip/dendrite (AT/D) ratios of mGluR1-GFP and NR2A-GFP proteins in neurons expressing KBS-TfR-mCherry, NiV-F-KBS-mCherry or the corresponding KBS-mut constructs. In all images, arrowheads point to the AIS. In all graphs the values are the mean ± SD from 25 neurons. * P<0.001.

Figure 7. Importance of tubulin acetylation and model for vesicle sorting at the PAEZ

(A) DIV7 neurons co-expressing GFP-Rigor-KIF5A (green) with mCherry-tubulin (Tub-WT) (left panels) or acetylation-mimic K40Q mutant α-tubulin (Tub-K40Q) (right panels). Notice that Tub-K40Q overexpression causes missorting of Rigor-KIF5A to dendrites. Arrowheads point to the AIS stained for AnkG (red). (B,C) Expression of mCh-Tub-K40Q (red) reduces axonal missorting of TfR-KBS-GFP (B, left panel, green on right panel) and NiV-F-KBS-GFP (C, left panel, green on right panel) in DIV7 neurons immunostained for AnkG (blue). Dendrite/axon (D/A) and axon tip/dendrite (AT/D) ratios were quantified from 25 neurons and expressed as mean ± SD. * P<0.001. (D) Schematic representation of vesicle sorting at the PAEZ. We propose that polarized sorting of transport vesicles begins at the PAEZ and depends on the ability of vesicles to acquire an axonally-directed microtubule motor. Sorting of newly-synthesized proteins to the axonal or somatodendritic domains of neurons involves selective incorporation into distinct populations of vesicular transport carriers budding from the trans-Golgi network (TGN) or recycling endosomes (RE) (Farías et al., 2012; Petersen et al., 2014). Axonal vesicles associate to axonally-directed microtubule (MT) motors such as kinesin-1 at the level of the PAEZ, resulting in movement of the vesicles towards the MT plus end in the distal axon. Somatodendritic vesicles, on the other hand, fail to enter the axon at the PAEZ because they lack the ability to bind axonal MT motors. Instead, these vesicles are driven to dendrites by association with other kinesins or dynein, which mediate transport along the mixed polarity MTs present in the dendrites.