Rab10 regulates the sorting of internalised TrkB for retrograde axonal transport

  1. Oscar Marcelo Lazo  Is a corresponding author
  2. Giampietro Schiavo  Is a corresponding author
  1. Department of Neuromuscular Diseases and UCL Queen Square Motor Neuron Disease Centre, UCL Queen Square Institute of Neurology, University College London, United Kingdom
  2. UK Dementia Research Institute at UCL, United Kingdom

Abstract

Neurons process real-time information from axon terminals to coordinate gene expression, growth, and plasticity. Inputs from distal axons are encoded as a stream of endocytic organelles, termed signalling endosomes, targeted to the soma. Formation of these organelles depends on target-derived molecules, such as brain-derived neurotrophic factor (BDNF), which is recognised by TrkB receptors on the plasma membrane, endocytosed, and transported to the cell body along the microtubules network. Notwithstanding its physiological and neuropathological importance, the mechanism controlling the sorting of TrkB to signalling endosomes is currently unknown. In this work, we use primary mouse neurons to uncover the small GTPase Rab10 as critical for TrkB sorting and propagation of BDNF signalling from axon terminals to the soma. Our data demonstrate that Rab10 defines a novel membrane compartment that is rapidly mobilised towards the axon terminal upon BDNF stimulation, enabling the axon to fine-tune retrograde signalling depending on BDNF availability at the synapse. These results help clarifying the neuroprotective phenotype recently associated to Rab10 polymorphisms in Alzheimer’s disease and provide a new therapeutic target to halt neurodegeneration.

Editor's evaluation

This important study, of interest to cellular neurobiologists, uses convincing microscopy methods to show that Rab10 GTPase is a new regulator of neurotrophin receptor trafficking and signaling. Defining how neurons respond to spatial extrinsic cues, such as neurotrophins, and relay this information long-distance to influence transcriptional events is an important topic in neurobiology.

https://doi.org/10.7554/eLife.81532.sa0

Introduction

Communication between cells depends on their ability to respond as integrated units to spatial and temporal signalling patterns. The complex morphology of neurons provides an unrivalled model to study how sorting and trafficking of signalling complexes coordinate local signalling at axon terminals and the propagation of messages to the cell body. A clear example is provided by neurotrophic factors secreted by target tissues and sensed by axon terminals, where, among other functions, they regulate local cytoskeletal dynamics to promote or cease axon elongation, induce branching, as well as synaptic maturation and plasticity (Andres-Alonso et al., 2019; Szobota et al., 2019; Woo et al., 2019). At the same time, a population of activated receptors are internalised and targeted to the retrograde axonal transport pathway within signalling endosomes (Barford et al., 2017; Villarroel-Campos et al., 2018), which propagate neurotrophic signalling towards the cell body, regulating gene expression, dendritic branching, and the balance between survival and apoptosis (Du and Poo, 2004; Pazyra-Murphy et al., 2009; Watson et al., 1999; Zhou et al., 2012). How local and central responses are coordinated across the massive distance from axon terminals to the soma is a crucial and still unanswered question for neuronal cell biology.

Brain-derived neurotrophic factor (BDNF), a member of the neurotrophin family, is widely expressed in the central nervous system, together with its receptor tropomyosin-related kinase B (TrkB). By activating diverse signalling cascades, comprising the phosphoinositide 3‐kinase (PI3K)‐Akt pathway, mitogen‐activated protein kinases (MAPKs), and phospholipase C‐gamma (PLCγ), BDNF and TrkB play a critical role in the formation, maintenance, and plasticity of neuronal circuits (Huang and Reichardt, 2003; Minichiello, 2009). Binding of BDNF at synaptic sites leads to endocytosis of TrkB and entry of the activated ligand–receptor complexes in early endosomes (Deinhardt et al., 2006). Whilst part of internalised TrkB recycle back to the plasma membrane, a pool of receptors is sorted to signalling endosomes and engage with the cytoplasmic dynein motor complex, which mediates the transport of these organelles to the soma (Andres-Alonso et al., 2019; Ha et al., 2008; Zhou et al., 2012). Independent lines of evidence indicate that this compartment is generated from early endosomes that then mature into a more specialised organelle escaping acidification and lysosomal degradation (Villarroel-Campos et al., 2018). Sorting of TrkB receptors from early to signalling endosomes constitutes the critical regulatory node controlling the intensity of the retrogradely propagated signal; however, to date, no clear mechanism controlling this sorting process has been elucidated.

The Rab family of monomeric GTPases plays a central role regulating post-endocytic trafficking of TrkB. Whilst early endosome formation is regulated by Rab5, maturation and processive transport of signalling endosomes in different neuronal models are controlled by Rab7 (Bucci et al., 2014; Burk et al., 2017; Deinhardt et al., 2006; Kucharava et al., 2020). However, it is currently unclear whether or not Rab7 is the only member of the Rab family necessary for the latter process. Because axonal retrograde signalling endosomes appear to be a diverse group of organelles (Villarroel-Campos et al., 2018), we hypothesised that other members of the Rab family also contribute to the segregation of TrkB and its sorting to retrograde axonal carriers.

In this work, we specifically focused on Rab10 since it has been shown that Rab10-positive organelles are transported both anterogradely and retrogradely along axons in hippocampal neurons (Deng et al., 2014). Previous work from our laboratory using an affinity purification approach to isolate neurotrophin signalling endosomes from mouse embryonic stem cell‐derived motor neurons found that Rab10 was significantly enriched in this axonal compartment, which is also characterised by the presence of Rab5 and Rab7 (Debaisieux et al., 2016). By manipulating Rab10 expression and activity in hippocampal neurons, as well as analysing the axonal dynamics of Rab10 organelles, we have explored its ability to regulate the sorting of TrkB to the retrograde axonal transport pathway and respond to increasing concentrations of BDNF, adjusting retrograde signalling on demand.

Results

Decreasing the expression of Rab10 in neurons

To manipulate Rab10 expression levels in hippocampal neurons, we used a lentiviral system encoding a doxycycline-inducible short hairpin RNA (shRNA). We monitored the effects of this virus on cell viability and expression levels of Rab10 and TrkB at different time points after addition of doxycycline. We observed that 48 hr of treatment consistently halved the number of cells per field compared to control group, whereas incubation for 24 hr did not affect neuronal density (Figure 1a and b). Rab10 immunoreactivity appeared significantly decreased at both 24 and 48 hr after addition of doxycycline compared to control lentivirus (Figure 1a and c). Finally, we monitored expression levels of TrkB receptors, to confirm that this system allows the study of the trafficking of the endogenous receptor upon Rab10 knockdown. We observed comparable levels of TrkB even at 48 hr of treatment with shRNA Rab10 (Figure 1d).

Doxycycline-inducible knockdown of Rab10 in hippocampal neurons.

(a) Representative fields of a primary mass culture of hippocampal neurons transduced with shRNA Rab10 versus control and treated with doxycycline for 48 hr. Cells have been immunolabelled for MAP2 (grey) and Rab10 (colour scale 0–255). Scale bar = 50 µm. (b) Cell density was quantified in 18 fields per treatment across three independent experiments, showing a significant decrease after 48 hr. Two-way ANOVA, F(1,68), p value for knockdown = 0.0270, p value for time = 0.0406, p value for interaction = 0.2114 (non-significant). The p values for Bonferroni multiple comparison tests, t(68), are indicated in the plot. (c) In the same experiments, immunoreactivity for Rab10 was quantified per cell at 12, 24, and 48 hr with doxycycline, and analysed using two-way ANOVA (p value for knockdown, time, and interaction <0.0001); p values for Bonferroni multiple comparison tests, t(68), are indicated in the plot and show a significant effect of the shRNA at 24 and 48 hr. (d) Representative low-magnification fields showing no difference on immunoreactivity for TrkB in hippocampal neurons treated with shRNA Rab10 versus control after 48 hr with doxycycline. The right panel shows zoomed boxes with TrkB (orange) and MAP2 (grey). Source data of the plots have been included in Figure 1—source data 1.

Figure 1—source data 1

Data tables for each plot presented in Figure 1 are given as individual CSV files.

https://cdn.elifesciences.org/articles/81532/elife-81532-fig1-data1-v1.zip

Rab10 is required for retrograde TrkB trafficking and signalling

To study axonal TrkB dynamics, we cultured hippocampal neurons in microfluidic chambers, which allow the cellular and fluidic compartmentalisation of axon terminals. After 7 days in vitro, neurons plated in one of the compartments (designated as ‘somatic’) displayed axons reaching the axonal compartment (Figure 2a). Maintaining a higher volume of media in the somatic compartment allows a micro-flow along the grooves, keeping any substance added to the axonal compartment confined. Using this experimental set-up, we incubated axon terminals with an antibody against the extracellular domain of TrkB and induced its endocytosis by adding 20 ng/mL BDNF. After 2.5 hr, we were able to detect axonal TrkB in the cell body of neurons stimulated with BDNF, but not in neurons depleted of BDNF by addition of an anti-BDNF-blocking antibody (Figure 2b), confirming the specificity of the antibody and the overall reliability of our retrograde accumulation assay.

Rab10 is required for retrograde TrkB trafficking and signalling.

(a) Schematic of two-compartment microfluidic chambers highlighting compartmentalisation of somata (left) and axon terminals (right). Micro-flow from somatic to axonal compartments provides fluidic isolation of the somatic compartment to probes added to the other chamber. (b) Representative images from the cell bodies of neurons incubated for 2.5 hr with anti-TrkB in the axonal compartment, with (+) or without (-) brain-derived neurotrophic factor (BDNF). Pink arrowheads indicate examples of retrogradely transported TrkB-positive organelles. Scale bar: 10 µm. (c) Neurons treated with the shRNA targeting Rab10 were compared to control transduced neurons. Immunofluorescence revealed similar neuronal density (see ßIII-tubulin in orange and nuclear staining in green, top panel), but a decrease in both, expression of Rab10 (grey, middle panel) and retrograde accumulation of TrkB after 2.5 hr (colour intensity scale, bottom panel). Scale bar: 50 µm. (d) Quantification of retrograde TrkB accumulation in three independent experiments show statistically significant differences (unpaired Student’s t-test, t(140), p<0.0001). (e) Correlation between expression level of Rab10 and retrograde TrkB accumulation in control and Rab10-knockdown neurons show a significant linear correlation (goodness-of-fit R2=0.61; Pearson r, XY pairs = 131, p<0.0001). (f) Axonal stimulation with BDNF for 2.5 hr leads to robust appearance of phosphorylated CREB in the nucleus of control neurons (left panel). This response was impaired in neurons depleted of Rab10 (middle panel), and rescued by the co-expression of a shRNA-resistant mutant Rab10 (right panel). Immunofluorescence for Rab10 is shown in grey, with the nuclei indicated with a pink mask, and nuclear phosphorylated CREB is shown in a colour intensity scale. Scale bar: 50 µm. (g) Quantification from three independent experiments showing the statistically significant effect of manipulating Rab10 expression on the levels of phosphorylated CREB in the nucleus (one-way ANOVA, F(2,280), p<0.0001; p values for the Bonferroni multiple comparison tests, t(280), are indicated in the plot). Source data of the plots have been included in Figure 2—source data 1.

Figure 2—source data 1

Data tables for each plot presented in Figure 2 are given as individual CSV files.

https://cdn.elifesciences.org/articles/81532/elife-81532-fig2-data1-v1.zip

Neurons transduced with the shRNA Rab10 and treated with doxycycline for 18–22 hr showed lower levels of endogenous Rab10 (Figure 2c, grey) and significantly reduced retrograde accumulation of TrkB compared with neurons expressing control lentivirus (Figure 2c and d). Since expression of Rab10 is variable among neurons, we tested the correlation between the levels of endogenous Rab10 with the retrograde accumulation of TrkB (Figure 2e). We found that, even though there was overlap, control and knocked down neurons clustered as expected. Moreover, when we take both populations together, we found a significant correlation between Rab10 expression and TrkB accumulation, strongly suggesting that Rab10 plays a role in retrograde axonal transport of internalised TrkB (Figure 2e).

To confirm the functional consequences of this decrease on the retrograde transport of TrkB, we treated neurons in the axonal compartment with BDNF and analysed the levels of phosphorylated cAMP response element binding protein (pCREB) in the nucleus. pCREB is a well-established proxy for neurotrophic signalling in neurons and has been shown to be critical for global neuronal responses to neurotrophins, such as BDNF-induced dendritic branching (González-Gutiérrez et al., 2020). Neurons treated with an shRNA directed against Rab10 showed a significant decrease in nuclear pCREB (Figure 2f and g), which was rescued by treatment of Rab10 knockdown cells with a lentivirus encoding shRNA-resistant Rab10, further confirming the specificity of this effect.

Rab10 associates transiently to TrkB-containing retrograde carriers

Given its critical role regulating retrograde accumulation of TrkB, we investigated whether Rab10 was present in retrograde signalling endosomes. As stated before, signalling endosomes are likely to be a heterogeneous pool of functionally related organelles with diverse molecular compositions (Villarroel-Campos et al., 2018). Most of these axonal carriers are positive for Rab7 (Deinhardt et al., 2006), which enables the recruitment of the dynein motor complex and their processive retrograde transport along microtubules (Ha et al., 2008; Zhou et al., 2012). Super-resolution radial fluctuations (SRRF) microscopy was used to examine the distribution of endogenous Rab10, Rab7, and Rab5 along the axon of neurons stimulated with 20 ng/mL BDNF for 30 min (Figure 3a). We found that Rab10 (orange) and Rab7 (green) consistently showed a very low degree of overlap (see yellow regions in Figure 3a and superposition of intensity peaks in Figure 3b). On the other hand, the super-resolution imaging revealed a population of organelles where Rab10 (orange) and Rab5 (purple) partially co-localise (see pink regions, Figure 3a), suggesting that Rab5-positive retrograde carriers (Goto-Silva et al., 2019), or stationary early endosomes, could also contain Rab10. To quantitatively assess whether co-localisation between endogenous Rab10 and these endosomal markers depended on BDNF signalling, we used confocal microscopy to evaluate co-localisation by using Manders index in neurons stimulated with BDNF (20 ng/mL for 30 min), as well as in neurons depleted of BDNF by treatment with a blocking antibody. Significance compared to randomised distributions was determined using confined-displacement algorithm (CDA) (Ramírez et al., 2010). Rab10 and Rab5 were confirmed to have around 20% co-localisation and a significant increase in the amount of Rab5 in Rab10 domains when stimulated with BDNF (Figure 3c). On the other hand, Rab10 and Rab7 were found to exhibit very low co-localisation levels, which was unaffected by BDNF stimulation, suggesting that Rab10 is more likely to be associated with early components of the endosomal system rather than mature and processive signalling endosomes. Similar co-localisation analysis has been done for the recycling endosomes markers Rab4 and Rab11 and included as supplementary material (Figure 3—figure supplement 1).

Figure 3 with 1 supplement see all
Internalised TrkB transiently co-localises with Rab10 in axons.

(a) Representative super-resolution radial fluctuations (SRRF) microscopy of an axon stained for endogenous Rab5 (purple), Rab7 (green), and Rab10 (orange). The inset at ×6 higher magnification shows examples of partial co-localisation of Rab10 with Rab5 and Rab7. Scale bar = 5 µm. (b) Intensity along the same axonal segment shown in (a) was plotted to show the correlation between the three markers. (c) Co-localisation of endogenous Rab10 and Rab5 was analysed comparing axons from starved neurons versus incubated 30 min with brain-derived neurotrophic factor (BDNF). Confined-displacement algorithm (CDA) has been used to compute Manders coefficients (Welch’s corrected unpaired Student’s t-test; M1: t(46.27), p-value = 0.0244; M2: t(48.94), p-value = 0.4794). Data points showing significant co-localisation compared to random distribution are marked with coloured dots. No significant differences were found between starved and BDNF-stimulated neurons. (d) Co-localisation of endogenous Rab10 and Rab7 was analysed in a similar experimental set-up (Welch’s corrected unpaired Student’s t-test; M1: t(72.32), p-value = 0.0621; M2: t(62.33), p-value = 0.1043). (e) Labelled HCT and anti-TrkB were co-internalised in the presence of BDNF for 30, 60, and 90 min, and their level of overlap with endogenous Rab10 in axons was evaluated using confocal microscopy. Relative areas positive for HCT and TrkB are shown in cyan (normalised to 30 min) and the fraction of the normalised area that was triple positive for HCT, TrkB and Rab10 is plotted in white. Whereas the double TrkB/HCT-positive area significantly increased by 90 min (one-way ANOVA, F(2,72), p-value <0.0001; Bonferroni multiple comparison test p value is shown in the plot), the triple TrkB/HCT/Rab10 surface remained low and fairly constant at all time points (one-way ANOVA, F(2,72), p-value = 0.2730; Bonferroni multiple comparisons test for 60 vs. 90 min, t(72), p-value >0.9999). (f) Representative SRRF microscopy from the same three time points. Double-positive puncta for TrkB (green) and HCT (purple) is indicated with cyan arrowheads. Triple-positive puncta of TrkB/HCT (cyan) and Rab10 (orange) are indicated with white empty triangles. Scale bar = 5 µm. Source data of the plots have been included in Figure 3—source data 1.

Figure 3—source data 1

Data tables for each plot presented in Figure 3 are given as individual CSV files.

https://cdn.elifesciences.org/articles/81532/elife-81532-fig3-data1-v1.zip

To specifically identify retrograde signalling endosomes containing TrkB, we took advantage of the property of the non-toxic carboxy-terminal domain of the heavy chain of tetanus neurotoxin (HCT) of being transported almost exclusively in retrograde axonal organelles in neurons (Deinhardt et al., 2006). We co-internalised HCT-Alexa Fluor647 and mouse anti-Flag antibodies in neurons transfected with a TrkB-Flag construct, and stimulated them with BDNF. Axonal double-positive TrkB/HCT puncta were considered signalling endosomes moving in the retrograde direction. We analysed the proportion of these organelles co-localising with endogenous Rab10. Figure 3e represents the relative area of TrkB/HCT and the triple-positive area of TrkB/HCT/Rab10 across three different time points (30, 60, and 90 min) post-endocytosis. As expected, the amount of TrkB in retrograde carriers increases with time (cyan). However, the proportion of TrkB present in Rab10 compartments is very low and remains constant across the duration of the experiment (white). These results provide another evidence suggesting that TrkB localises transiently on Rab10-positive membrane compartments en route to its delivery to axonal retrograde carriers. This observation is also supported by super-resolution images (Figure 3f), showing an increase in TrkB/HCT double-positive puncta (cyan arrowheads), but very few TrkB/HCT/Rab10 triple-positive puncta (white empty triangles) at 60 and 90 min.

Overexpressed Rab10 is co-transported with retrograde TrkB

The data presented so far support a model in which Rab10 is critical for retrograde transport of TrkB, but it does not specifically define a stable population of retrograde carriers. Rather, the evidence shown in Figure 3e and f suggests that TrkB is transiently associated to Rab10-positive membranes, opening the possibility that Rab10 participates in the sorting of internalised receptors to retrogradely transported signalling endosomes. To directly monitor this process, we overexpressed EGFP-Rab10 in primary hippocampal neurons. Since it is well documented that overexpression of Rab GTPases promotes their activity due to their increased association to membranes (Zhen and Stenmark, 2015), we expected that an increase in the abundance and/or activity of Rab10 would stabilise its association to TrkB retrograde carriers, allowing its visualisation by time-lapse microscopy. Neurons were co-transfected with EGFP-Rab10 and TrkB-Flag and, after 1 hr of starvation in non-supplemented Neurobasal medium, we internalised Alexa Fluor647-labelled anti-Flag antibodies in the presence of BDNF for 45 min (Figure 4a). Time-lapse confocal imaging of axon segments was performed at least 200 µm from the cell body. Sequential frames of a representative movie are shown in Figure 4b, where the retrograde co-transport of TrkB and EGFP-Rab10 is indicated with yellow arrowheads. Kymographs (Figure 4c) were generated for EGFP-Rab10 (orange) and TrkB-Flag (green). Tracks are shown in the bottom panel, with examples of co-transport indicated by yellow lines. Quantification of five independent experiments confirmed that approximately 60% of retrograde TrkB carriers are positive for Rab10 under these experimental conditions (Figure 4d). Interestingly, no anterograde TrkB/Rab10 double-positive compartments were observed, suggesting that TrkB is present in organelles with a strong retrograde bias.

Over-expressed Rab10 is co-transported with TrkB and p75 receptors in the axon.

(a) Hippocampal neurons in mass culture were co-transfected with EGFP-Rab10 and TrkB-Flag plasmids. Fluorescently labelled anti-Flag antibodies were internalised in the presence of brain-derived neurotrophic factor (BDNF) and their axonal dynamics monitored by time-lapse microscopy. (b) Representative images of the axon of a double transfected neuron, where a double-positive organelle for EGFP-Rab10 (orange) and anti-Flag (green) is indicated with yellow arrowheads. Scale bar = 5 µm. (c) Representative kymograph of the axon of a double transfected neuron during 5 min of imaging showing Rab10 and TrkB-Flag channels. Double-positive tracks (in yellow) show transport predominantly in the retrograde direction (right to left). Scale bar = 10 µm. (d) Quantification of five experiments showing the proportion of TrkB-containing mobile organelles that were positive for Rab10. No anterograde TrkB organelles were found; therefore, presence of Rab10 could not be determined (N.D.). (e) An equivalent experiment was performed by transfecting EGFP-Rab10 and visualising it together with endocytosed fluorescently-labelled anti-p75NTR antibodies. (f) Representative frames from a time-lapse movie displaying a double-positive organelle for Rab10 (orange) and p75NTR (green) moving anterogradely. Scale bar = 5 µm. (g) Representative kymograph showing Rab10 and p75NTR channels. Double-positive tracks (in yellow) show transport in both anterograde (left to right) and retrograde directions. Scale bar = 10 µm. (h) Quantification of five experiments displaying the proportion of p75NTR mobile organelles positive for Rab10 moving in the anterograde and retrograde direction. Source data of the plots have been included in Figure 4—source data 1.

Figure 4—source data 1

Data tables for each plot presented in Figure 4 are given as individual CSV files.

https://cdn.elifesciences.org/articles/81532/elife-81532-fig4-data1-v1.zip

To extend this analysis to the other physiological BDNF receptors, the internalisation of endogenous p75 neurotrophin receptor (p75NTR) was monitored in neurons expressing EGFP-Rab10. After depletion of trophic factors, p75NTR uptake was visualised by incubating neurons for 45 min with an Alexa Fluor647-labelled antibody against the extracellular domain of p75NTR, in the presence of BDNF (Figure 4e). Live-cell imaging of axon segments was done under conditions identical to those used for TrkB. Representative frames and kymographs of EGFP-Rab10 (orange) and internalised p75NTR (green) are shown in Figure 4f and g. Interestingly, p75NTR receptor can be found in both retrograde and anterograde Rab10 carriers. On average, 29.6% of anterograde and 15.5% of retrograde p75NTR-containing organelles were found positive for Rab10 across five independent experiments (Figure 4h).

BDNF promotes anterograde trafficking of Rab10-positive compartments

In agreement with the bidirectional transport observed for p75NTR carriers (Figure 4h), previous work suggested that Rab10-positive compartments travel along the axon in both directions (Deng et al., 2014). In light of these results, we hypothesised that the dynamics of axonal compartments containing Rab10 may be responsive to BDNF signalling to direct the sorting of TrkB to retrograde carriers. We therefore examined the axonal transport of Rab10-positive organelles in hippocampal neurons under two opposite conditions: depletion of BDNF using an anti-BDNF blocking antibody, followed by stimulation with 50 ng/mL BDNF (Figure 5—figure supplement 1).

Figure 5a shows representative frames of an EGFP-Rab10 organelle moving in the retrograde direction along the axon (white arrows) in the absence of BDNF. Five-minute segments of time-lapse microscopy have been used to generate kymographs (Figure 5b) at different time points of BDNF stimulation: before BDNF (top), immediately after BDNF addition (centre), and after 10 min of BDNF incubation (bottom). Traces have been colour-coded as retrograde (cyan), anterograde (pink), or stationary/bidirectional (yellow) to reveal changes in the direction bias of Rab10 organelles in the same axon, before, and after the addition of BDNF. Quantification of five independent experiments shows that BDNF-depleted axons exhibit a bias towards retrograde Rab10 transport, which significantly switches to anterograde after 10 min of stimulation with BDNF (Figure 5c).

Figure 5 with 4 supplements see all
Brain-derived neurotrophic factor (BDNF) regulates the directionality of Rab10 organelles.

Hippocampal neurons in mass culture were transfected with EGFP-Rab10 and depleted of BDNF for 60 min. (a) Representative axon of a live neuron showing retrograde (right to left) transport of a Rab10-positive organelle (white arrowheads) in the absence of BDNF. Scale bar = 5 µm. (b) In the panels on the left, representative kymographs (colour-coded as in a) are presented from the same axon upon BDNF depletion (top), immediately after the addition of 50 ng/mL of BDNF (middle), and 10 min thereafter (bottom). In the panels on the right, tracks have been traced and categorised as retrograde (cyan), anterograde (pink), or stationary/bidirectional (yellow). Scale bar = 10 µm. (c) The frequencies of tracks from each of the three categories have been quantified and plotted, comparing no BDNF and 10 min post-addition of 50 ng/mL BDNF. N = 14 axonal segments from 10 independent experiments. Unpaired Student’s t-test, t(14), showed a significant increase in anterograde carriers (p-value = 0.0150, *) at the expense of retrograde carriers (p-value = 0.003, **). Stationary and bidirectional carriers did not show any significant change (p-value = 0.4278). See Figure 5—video 1 for a video. Source data of the plots have been included in Figure 5—source data 1.

Figure 5—source data 1

Data tables for each plot presented in Figure 5 are given as individual CSV files.

https://cdn.elifesciences.org/articles/81532/elife-81532-fig5-data1-v1.zip

These surprising results reveal a novel mechanism ensuring a tight balance between retrograde and anterograde transport of Rab10 organelles, which is fine-tuned by BDNF. This result predicts that any local increase in BDNF release from post-synaptic compartments will increase the abundance of Rab10 organelles in the immediate vicinity of the source of BDNF.

BDNF increases the recruitment of KIF13B to Rab10 organelles in the axon

The uniform polarity of the microtubules in the axon of mature neurons implies that the directionality of axonal carriers largely depends on their association to anterograde or retrograde motor proteins (Guillaud et al., 2020; Kwan et al., 2008). To provide further mechanistic insights into the directional switch of Rab10 axonal carriers upon BDNF stimulation, we analysed the distribution of two anterograde microtubule-associated motor proteins predominantly found in central neuron axons, KIF5B and KIF13B (Yang et al., 2019). Figure 6a and d show representative high-resolution confocal images of axons where endogenous Rab10 (green) and either KIF13B or KIF5B (orange) have been detected. Neurons were either starved in the presence of a blocking anti-BDNF antibody (control) or treated for 30 min with 50 ng/mL BDNF (BDNF). Insets displaying two different views of three-dimensional reconstructions of axonal segments show Rab10 membranes positive for KIF13B (Figure 6a) or KIF5B (Figure 6d) in the absence and presence of BDNF. Interestingly, the axonal pool of KIF13B appears to decrease after stimulation with BDNF, whereas that of KIF5B increases under the same experimental conditions, suggesting that the total axonal content of these kinesins may be differentially regulated by BDNF signalling.

Figure 6 with 2 supplements see all
Brain-derived neurotrophic factor (BDNF) increases recruitment of KIF13B to Rab10 domains.

(a) The co-distribution of endogenous KIF13B (orange) and Rab10 (green) was monitored using high-resolution Airyscan confocal microscopy in axons of neurons with or without BDNF for 30 min. Top images correspond to maximum projection of z-stacks, scale bar = 5 µm. A 3D reconstruction of the inset area (grey frame) is shown for each top image on its original position and after turning the image 210° around the x axis. (b) KIF13B intensity was measured in the entire axon segment and in Rab10-positive areas, and the intensity ratio was plotted and analysed showing a significant enrichment of KIF13B in Rab10 areas upon 30 min of BDNF stimulation (Kolmogorov–Smirnoff nonparametric t test, t(79.5), p-value = 0.0177). (c) Ratio between KIF13B-positive area that overlaps with Rab10 from total KIF13B area was plotted, finding no difference between neurons starved or incubated with BDNF 30 min (Kolmogorov–Smirnoff nonparametric t test, t(79.5), p-value = 0.8024). (d) Co-distribution of KIF5B (orange) and Rab10 (green) was also analysed and displayed as in (a). Scale bar = 5 µm. Insets show 3D reconstructions on their original and rotated position. (e) Quantification of intensity ratio of KIF5B in Rab10 domains versus total KIF5B in the axon shows no significant difference between starved and BDNF-treated neurons (Kolmogorov–Smirnoff nonparametric t test, t(62.5), p-value = 0.0644). (f) Proportion of the KIF5B-positive area that overlaps with Rab10 is lower than KIF13B and is not altered by BDNF stimulation (Kolmogorov–Smirnoff nonparametric t test, t(62.5), p-value = 0.3738). (g) The interaction of HA-Rab10 and GFP-KIF13B from N2A cells is modulated by BDNF. Top panel: representative western blot showing the presence of both proteins in the lysate (input). Bottom panel: Western blot of co-immunoprecipitated samples from the same experiment using an antibody against the HA tag. (h) Quantification of the ratio between normalised KIF13B and Rab10 in three independent experiments. Western blots have been done in duplicate, and the corresponding paired experiments are indicated by data points of the same shade of grey. Groups were compared using paired Student’s t test, t(6), p-value = 0.0024. Source data of the plots have been included in Figure 6—source data 1.

Figure 6—source data 1

Data tables for each plot presented in Figure 6 are given as individual CSV files, as well as unedited representative blots used in Figure 6g.

Total cell extracts are labelled as input, and eluates from the immunoprecipitation are labelled as immunoprecipitation. Antibodies used for each membrane are indicated. Red frames indicate the bands that are shown in the main figure.

https://cdn.elifesciences.org/articles/81532/elife-81532-fig6-data1-v1.zip

Quantification of the intensity of KIF13B in Rab10-positive areas (Figure 6b) reveals a statistically significant 12.7% increase in the amount of the motor recruited to Rab10-positive organelles. On the other hand, Figure 6c shows that the ratio between the Rab10-positive and total KIF13B-positive area in the axon remained unchanged (7.3% of the area), suggesting that the proportion of double KIF13B/Rab10 compartments remains constant. In contrast, when we performed the same analysis for KIF5B, we found that, despite its overall increase in the axon upon BDNF stimulation, there is no significant change in the intensity ratio in Rab10-positive domains (Figure 6e). Consistently, Figure 6f shows that the proportion of KIF5B axonal organelles that overlap with Rab10 is lower than KIF13B (around 4% of the area) and remained unchanged after BDNF stimulation.

Prompted by the results shown in Figure 6b, we investigated whether KIF13B and Rab10 interact in a BDNF-dependent manner by expressing GFP-KIF13B and HA-Rab10 in Neuro-2A (N2A) cells. This mouse neuroblastoma cell line expresses TrkB and has been previously used in our laboratory to study trafficking and signalling of neurotrophic receptors (Terenzio et al., 2014). After 1 hr of starvation, transfected cells were treated or not with BDNF 50 ng/mL and then we used anti-HA-conjugated magnetic beads to pull down HA-Rab10. Lysates, as well as the co-immunoprecipitated samples eluted from the beads, were analysed by Western blot. A representative example is shown in Figure 6g, with the top panels showing the total amount of HA-Rab10 and GFP-KIF13B in the lysates (input). Co-immunoprecipitated HA-Rab10 and GFP-KIF13B eluted from the beads are displayed in the bottom panels (co-IP). Lysates from N2A cells only expressing GFP-KIF13B were used as controls. Quantification of three independent experiments (Figure 6h) show an average twofold increase in the KIF13B/Rab10 ratio when cells were stimulated with BDNF. A consistent change in pulled down KIF13B with no change of Rab10 is shown as separated plots in Figure 6—figure supplement 1.

Rab10 regulates the sorting of TrkB in early endosomes, with no effect on recycling

After endocytosis, TrkB accumulates in early endosomes, from which it is sorted either to the recycling route or to endosomal organelles with signalling capabilities (Deinhardt et al., 2006; Zhou et al., 2012). Rab10, on the other hand, has been shown to regulate multiple processes, including trafficking across early endosomes, the formation of specialised tubular endosomes, the recycling of cargoes back to the plasma membrane, as well as the targeting of plasmalemmal precursor vesicles (PPVs), among other functions (Babbey et al., 2006; Brewer et al., 2016; Deng et al., 2014; Etoh and Fukuda, 2019; Xu et al., 2014). To understand the mechanism linking Rab10-positive organelles with the retrograde axonal transport of TrkB in signalling endosomes, we designed experiments to discern between a potential role of Rab10 on recycling of internalised TrkB back to the plasma membrane, and sorting of TrkB out of early endosomes into retrograde transport carriers (Figure 7a).

Rab10 regulates sorting of TrkB out of early endosomes.

(a) Main hypotheses about the role of Rab10 regulating the sorting of TrkB to signalling endosomes include recycling back to the plasma membrane or sorting of TrkB receptors out of early endosomes to retrograde carriers. (b) Top: diagram of the experiment showing TrkB-Flag receptors on the axonal surface bound to anti-Flag antibodies (green). After internalisation, the remaining anti-Flag is removed from the surface and the labelled receptor that recycled to the plasma membrane is chased with a secondary antibody (orange). Bottom: representative examples of internalised TrkB (green) and recycled TrkB (orange) in axons from neurons transfected with EGFP or a Rab10 DN mutant. (c) Normalised recycling shows no difference between EGFP and Rab10 DN transfected neurons. Unpaired Student’s t-test, t(60.40), p-value = 0.3914. (d) Top: diagram of the internalisation of TrkB-flag labelled with anti-Flag antibodies (orange) to Rab5-positive early endosomes (green). Bottom: representative thresholded microscopy images from the axon of neurons transfected with EGFP or Rab10DN mutant. While the amount of orange puncta is similar in both conditions, yellow areas showing co-localisation of internalised TrkB and Rab5 are increased upon Rab10 DN expression. (e) Quantification of co-localisation between internalised TrkB-Flag and endogenous Rab5 is significantly higher in neurons expressing Rab10DN compared to EGFP. Unpaired Student’s t-test, t(38.22), p-value = 0.008. Significant co-localisation according to confined-displacement algorithm (CDA) (p-value < 0.05 compared to randomised signal) is shown with black circles, while inconclusive co-localisation (p-value > 0.05) is shown in grey. Scale bars = 5 µm. Source data of the plots have been included in Figure 7—source data 1.

Figure 7—source data 1

Data tables for each plot presented in Figure 7 are given as individual CSV files.

https://cdn.elifesciences.org/articles/81532/elife-81532-fig7-data1-v1.zip

To assess the contribution of Rab10 to the recycling of TrkB in the axon, hippocampal neurons were transfected with EGFP or a dominant negative mutant of Rab10 (Rab10T23N; referred to as Rab10DN) and TrkB-Flag. As illustrated in Figure 7b, endocytosis of anti-Flag M1 antibodies was allowed for 30 min in the presence of BDNF, and then the residual antibody still bound to the neuronal surface was removed using EDTA, which dissociates this antibody from the Flag peptide (Chen et al., 2005). Recycling receptors were then chased using Alexa Fluor647-conjugated secondary antibodies. After fixation, internalised TrkB-Flag was labelled with Alexa Fluor555-conjugated secondary antibodies. Comparison between recycling ratio (recycled/internalised) of EGFP- and Rab10DN-expressing neurons shows no significant differences (Figure 7c).

In contrast, if the sorting of TrkB to retrograde signalling carriers was regulated by Rab10, expression of Rab10DN would result in retention of a population of TrkB in early endosomes, driving an increase of co-localisation between internalised TrkB and Rab5. Therefore, we incubated neurons transfected with TrkB-Flag and either EGFP or Rab10DN, with anti-Flag antibodies for 30 min in the presence of BDNF, and then analysed TrkB/Rab5 co-localisation using Manders index and CDA (Figure 7d and e). Rab10DN caused a statistically significant increase of internalised TrkB in Rab5-positive domains from 13.2% ± 1.7 to 22.2% ± 2.8, and an increase in the reliability of the co-localisation measurements (CDA p<0.05) (Figure 7e). These results indicate that the probability of finding axonal TrkB in Rab5-positive early endosomes increases in neurons expressing Rab10DN, thus indicating that Rab10 activity modulates the sorting of activated TrkB receptors from axonal early endosomes to retrograde transport organelles.

Discussion

Our results unravel a novel role for Rab10 in regulating the sorting of internalised TrkB receptors to the retrograde axonal transport pathway. This function appears to be necessary not only for efficient trafficking of TrkB from axons to soma, but also for the propagation of neurotrophic signalling from distal sites to the nucleus, as shown by the decrease of BDNF-induced CREB activation upon Rab10 knockdown (Figure 2). CREB is a transcription factor supporting neuronal survival and differentiation, which is activated by phosphorylation downstream Akt and MAPK signalling (Wang et al., 2018). Some of the better characterised early response genes to neurotrophic factors (e.g., Egr1, Egr2, Arc, and cFos) are transcriptional targets of CREB, which is also required for BDNF-induced dendritic branching (Esvald et al., 2020; González-Gutiérrez et al., 2020; Kwon et al., 2011). Therefore, CREB phosphorylation is one of the best proxies for global responses to neurotrophic signalling. Long-distance activation of CREB has been reported from distal axons and dendrites, and the endosomal trafficking of phosphorylated TrkB and signalling interactors has been shown to play a crucial role in propagating neurotrophin signalling from the periphery to the nuclear pCREB (Cohen et al., 2011; Riccio et al., 1997; Watson et al., 1999). In our study, we found that Rab10 expression was critical for maintenance and survival of differentiated neurons since its downregulation for 48 hr led to a significant decrease in neuronal density (Figure 1). Similar effects have been reported upon overexpression of dominant-negative mutants of Rab5, the GTPase mediating the formation of early endosomes, but not for Rab11, which controls the recycling of receptors to the plasma membrane (Lazo et al., 2013; Moya-Alvarado et al., 2018), suggesting that these distinct arms of the endosomal network differentially impact on neuronal homeostasis.

Acting as a highly specialised network, Rab GTPases are master regulators of specific membrane trafficking events in eukaryotic cells (Stenmark, 2009). Rab10 is one of the few exceptions to this rule, and during the last 30 years, it has been associated to multiple trafficking pathways, including polarised exocytosis from early endosomes, exocytosis in adipocytes and neurons, endoplasmic reticulum dynamics, and the formation of tubular endosomes, to cite but a few (Chua and Tang, 2018). This multiplicity of functions is reflected by the diversity of its interactors and effectors, emphasising the importance of the specific context in which Rab10 operates. Here we have confirmed previous findings showing that Rab10-positives organelles are present in axons of hippocampal neurons and undergo bidirectional transport (Deng et al., 2014). Importantly, in this work we have demonstrated that the balance between anterograde and retrograde transport of Rab10-positive organelles is regulated by BDNF, indicating the ability of Rab10 to specifically respond to extracellular cues in a signalling context. A similar function of Rab10 balancing anterograde and retrograde transport has been shown to be required for correct dendritic patterning in Drosophila (Taylor et al., 2015).

We hypothesised that the differential recruitment of members of the kinesin superfamily of anterograde motors could explain a rapid change in the directionality of Rab10 organelles. Given that many kinesins distribute preferentially to distinct neuronal compartments (Hirokawa and Tanaka, 2015), we focused on two kinesins that have been found to be predominant in axons, KIF5B and KIF13B (Yang et al., 2019). We found that KIF13B, but not KIF5B, increased its relative abundance in axonal Rab10-positive domains (Figure 6a–f), which is in line with the recent evidence indicating a direct interaction between Rab10 and KIF13A and B, via a Rab-binding domain (RBD)-homology domain (Etoh and Fukuda, 2019). Moreover, when we used immunoprecipitation to investigate the interaction of KIF13B and Rab10, we found that their binding significantly increases in the presence of BDNF (Figure 6g and h). Interestingly, KIF13B and other members of the kinesin-3 family have been shown to yield super processive motion when acting as dimers (Soppina et al., 2014), helping to explain how even a relatively small increase in the recruitment of KIF13B onto axonal Rab10-positive organelle can trigger a rapid change in their transport directionality. Together with an increase of the recruitment of KIF13B and other kinesins, Rab10 effectors can also regulate the motor activity of cytoplasmic dynein (Taylor et al., 2015) or mediate the interaction with axonal myosins (Liu et al., 2013; Welz and Kerkhoff, 2019), where specific contribution to directionality and processivity of these organelles will require further exploration.

Although the main focus of this work has been demonstrating a novel role of Rab10 in the retrograde propagation of neurotrophic signalling, the recruitment of KIF13B onto Rab10-positive organelles is an important finding to start disentangling the mechanisms that control the BDNF-dependent switch between retrograde and anterograde transport. For example, the activity of Rab10 is known to be controlled by PI3K-Akt, a canonical BDNF/TrkB signalling pathway. In adipocytes and muscle cells, Rab10 is known to regulate the plasma membrane delivery of the glucose transporter GLUT4 in response to insulin, where activation of Akt leads to phosphorylation of the Rab GAP Akt substrate of 160 kDa (AS160). Rab10 associated to GLUT4-containing endosomes is kept inactive by AS160 until Akt signalling releases the brake and promotes fusion with the plasma membrane (Sano et al., 2007).

However, two key observations suggest that directionality of Rab10 organelles in the axon does not depend uniquely on Rab10 activity. First, both anterograde and retrograde organelles have membrane-bound Rab10 (Figure 5), which is generally accepted to be its GTP-bound active form. Second, we observed that the constitutively active Rab10Q68L mutant is transported predominantly in the retrograde direction in neurons depleted of BDNF (see Figure 5—figure supplement 1). Altogether, these findings indicate that whilst the GTP-bound conformation of Rab10 is necessary to drive its binding to the membrane of axonal organelles, this alone is not sufficient to determine the direction of transport. Signalling molecules associated to the membrane of the moving endosome are likely to be key defining its direction and processivity, as exemplified in Figure 5—video 2, which shows an anterograde Rab10-positive carrier running next to a stationary organelle, partially merging and continue moving together.

The LRRK2-dependent phosphorylation of Rab10 in the highly conserved switch II region has been proposed to affect its GTP/GDP cycle as well as its ability to bind effectors (Pfeffer et al., 1995; Xu et al., 2021). Moreover, this phosphorylation site has been shown to regulate the interaction of Rab10 with JIP3 and JIP4 (Waschbüsch et al., 2020), which are adaptors for the plus-end-directed microtubule-dependent motor kinesin-1 (Cockburn et al., 2018; Isabet et al., 2009). Interestingly, JIP3 has been shown to mediate the anterograde transport of TrkB in neurons and, by that mechanism, to enhance BDNF signalling (Huang et al., 2011; Sun et al., 2017). However, preliminary experiments showed that JIP3 and JIP4 association to Rab10 organelles is extremely low in hippocampal neurons and does not respond to BDNF stimulation (Figure 6—figure supplement 2). LRRK2-phosphorylated Rab10 has been also found to recruit myosin Va (Dhekne et al., 2021), providing other candidates for regulation of axonal trafficking. On the other hand, positioning of Rab10-positive membranes has been recently shown to determine their phosphorylation by LRRK2, suggesting that trafficking and composition of these organelles can be mutually regulated (Kluss et al., 2022). Whether LRRK2-mediated phosphorylation of Rab10 modulates the interaction with KIF13B is part of our ongoing research efforts and will not be discussed here in detail; however, it is worth mentioning that all three Rab10, KIF13B, and LRRK2 have been implicated in the formation of tubular endosomes (Bonet-Ponce et al., 2020; Etoh and Fukuda, 2019). Among the most interesting candidates linking BDNF-signalling and LRRK2 activity is Vps35, a component of the retromer complex that has been shown to modulate LRRK2 (Mir et al., 2018) and is recruited by TrkB binding to SorLA (Rohe et al., 2013), a sortilin family member. On the other hand, activation of Akt has been shown to compensate impairments of the insulin-stimulated GLUT4 trafficking associated with LRRK2 deficiency (Funk et al., 2019), opening the possibility that other downstream kinases could regulate Rab10 by LRRK2-independent mechanisms.

Once internalised, axonal TrkB reaches early endosomes, from where it can either recycle back to the plasma membrane, thus fine-tuning the response of nerve terminals to BDNF (reviewed in Andreska et al., 2020) or undergo sorting to the retrograde axonal transport route, propagating neurotrophic signals to the soma (Deinhardt et al., 2006; Ha et al., 2008; Zhou et al., 2012). Given that Rab10 downregulation decreased retrograde transport of TrkB and stimulation with BDNF promoted anterograde transport of Rab10, we reasoned that delivery of Rab10 to axon terminals facilitated the sorting of TrkB to retrogradely transported organelles. We demonstrated that Rab10DN expression increases the accumulation of internalised TrkB in early endosomes without significantly affecting recycling (Figure 7), suggesting that the pool of TrkB receptors available for local recycling and axonal transport are spatially segregated in the early endosome membrane, and thus indicating that functional Rab10 aids the exit of TrkB from early endosomes and its sorting to distinct axonal retrograde organelles.

When overexpressed, Rab10 is found at detectable levels in retrograde signalling endosomes positive for TrkB or p75NTR (Figure 4). An interesting observation is that only p75 receptors are present in Rab10-containing anterograde organelles, which bears the question on the role of the anterograde delivery of p75NTR to distal axons. In sympathetic neurons, it has been shown that p75NTR is rapidly mobilised to the plasma membrane upon stimulation with NGF via activation of Arf6 (Hickman et al., 2018), a small GTPase that has been shown to share common organelles with Rab10 in Caenorhabditis elegans (Shi and Grant, 2013).

At endogenous levels of Rab10, however, when we analysed triple co-localisation of internalised HCT and TrkB with endogenous Rab10, we found that the absolute amount of retrograde TrkB associated to Rab10 organelles remains low (Figure 3e), suggesting that this interaction is transient. This is in line with previous evidence showing that in MDCK cells, Rab10 mediates transport from early endosomes to a polarised trafficking route (Babbey et al., 2006). Moreover, in C. elegans, Rab10 is recruited to early endosomes where it downregulates Rab5, helping to select cargoes for delivery to recycling endosomes (Liu and Grant, 2015). According to this model, the transfer of TrkB from a stationary early endosome to a retrograde carrier would be preceded by a local increase of Rab10. In Figure 8b, we show an example of a live-cell imaging experiment in which a TrkB-positive endosome makes the transition from stationary to retrograde. We show the level of EGFP-Rab10 during the stationary phase (red), in the transition phase immediately preceding movement (yellow), and during processive retrograde transport (green). Crucially, Rab10 is recruited to this TrkB-positive axonal organelle few seconds before it starts moving retrogradely, strengthening the validity of our working model (Figure 8).

Model: role of Rab10 in the sorting of TrkB to retrograde axonal transport.

(a) At steady state, low concentrations of BDNF (grey side of the terminal) induce basal levels of TrkB internalisation. Rab10 supply (blue arrows) is in equilibrium and it mediates a baseline level of TrkB retrograde transport (pink arrows). Upon increase of BDNF (green side of the axon terminal), TrkB endocytosis as well as the proportion of Rab10 organelles moving towards the axon terminal are increased. Increased amounts of Rab10 result in further facilitation of the sorting of TrkB out of early endosome and an augmented flow of retrograde signalling carriers. (b) This model predicts that the transition from stationary early endosomes to processive retrograde carriers would be preceded by the focal recruitment of Rab10. The top panel shows an example of a kymograph of internalised TrkB-Flag in the axon. In the kymograph, three segments have been highlighted: (1) stationary phase, (2) transition phase, and (3) retrograde transport phase. The bottom panel shows the levels of EGFP-Rab10 in these segments, which clearly demonstrates the enhanced recruitment of Rab10 during the transition phase, before onset of transport.

Rab10 has been shown to participate in the biogenesis of tubular endosomes in mammalian cells (Etoh and Fukuda, 2019) and regulates endosomal phosphatidylinositol-4,5-bisphosphate in C. elegans, suggesting that this small GTPase modulates the membrane recruitment of factors altering membrane curvature and budding (Shi and Grant, 2013). During our live-cell imaging experiments, we observed Rab10-positive tubular organelles moving rapidly into proximal neurites, confirming that these structures are also generated in hippocampal neurons (Figure 5—video 3). In addition, it has been shown that snapin, an adaptor for cytoplasmic dynein recruitment to TrkB-signalling endosomes (Zhou et al., 2012), is also phosphorylated by LRRK2 (Yun et al., 2013), opening the possibility that Rab10 membranes constitute a specialised sorting domain. Interestingly, one of the well-characterised roles of KIF13B in the brain is the transport of phosphatidylinositol-3,4,5-biphosphate via its interactor centaurin-alpha (Hammonds-Odie et al., 1996), which is essential for axon specification (Horiguchi et al., 2006). Further research on the composition and cargo content of Rab10 organelles in neurons is therefore warranted.

Our data suggests that Rab10 regulates the amount of internalised TrkB that is sorted to retrograde signalling endosomes in response to the concentration of BDNF at axon terminals. Since BDNF is known to be released post-synaptically as a function of neuronal activity (Matsuda et al., 2009), retrograde neurotrophic signalling from the axon terminal is a substantial feedback mechanism regulating growth and survival, and therefore, ensuring that active circuits are preserved. To keep this feedback signal meaningful, any change in the availability of BDNF at the synapse must translate into proportional changes in the intensity of the signal arriving at the soma. We propose that Rab10-positive membranes deliver crucial components of the sorting machinery on demand. At steady state, the anterograde and retrograde flow of Rab10 compartments is in equilibrium (Figure 8; low BDNF). Further decrease in BDNF concentration makes retrograde transport of Rab10 carriers predominant, as observed in live-cell imaging experiments performed in neurons treated with anti-BDNF blocking antibodies (Figure 4; no BDNF). In contrast, adding BDNF reverts the directional bias to anterograde (Figure 8; high BDNF), which allows the delivery of appropriate levels of the sorting machinery to nerve terminals, thus increasing the efficiency of TrkB retrograde transport.

An independent example of anterograde delivery of components for the sorting of retrograde signalling molecules has been hypothesised for the bone-morphogenetic protein (BMP) pathway in Drosophila motor neurons. In this work, Khc-73, the fly orthologue of KIF13B, has been shown to regulate BMP sorting from early endosomes at the neuromuscular junction (Liao et al., 2018). This suggests that the proposed mechanism is evolutionary conserved in different neuronal types, increasing its potential as a therapeutic target for pathologies in which neurotrophic signalling from distal axons is impaired.

Little is known about the specific sorting machinery required for the biogenesis of signalling endosomes. Endophilins A1 and A3 have been shown to regulate the trafficking of TrkB across early endosomes and mediate survival signalling (Burk et al., 2017; Fu et al., 2011). Interestingly, endophilin A1 as well as Rab10 are known LRRK2 substrates, opening the possibility of both being found on the same organelle (Matta et al., 2012). Moreover, KIF13B has also been found to be enriched in early endosomes (Bentley et al., 2015). The characterisation of the sorting machinery delivered to axonal terminals by Rab10-positive carriers will be crucial to understand not only how this mechanism allows coordination between local signalling and global neuronal responses, but also how this process may fail in neurodegeneration. In this light, promoting the delivery of Rab10 organelles to nerve terminals may be explored as an novel therapeutic strategy for diseases in which the endolysosomal system is overloaded or dysfunctional, such as Alzheimer’s disease (Van Acker et al., 2019; Xu et al., 2018), or to increase the ability of axons to respond to trophic factors during regeneration.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
AntibodyGoat polyclonal anti-Rab10Santa Cruz BiotechnologiesCat#sc-6564; RRID:AB_22378441:50
AntibodyMouse monoclonal anti-Rab10AbcamCat#ab104859; RRID:AB_107112071:200
AntibodyRabbit monoclonal anti-Rab10Cell SignallingCat#8127; RRID:AB_108282191:200
AntibodyRabbit monoclonal anti-TrkBMerck (Millipore)Cat#AB9872; RRID:AB_112143171:200
AntibodyChicken polyclonal anti-tubulin ßIIISynaptic SystemsCat#302 306; RRID:AB_26200481:300
AntibodyRabbit monoclonal anti-phosphorylated CREBAbcamCat#ab32096; RRID:AB_7317341:250
AntibodyMouse monoclonal anti-Rab7AbcamCat#ab50533, RRID:AB_8822411:200
AntibodyRabbit polyclonal anti-Rab5AbcamCat#ab13253; RRID:AB_2997961:200
AntibodyRabbit polyclonal anti-KIF13BBiossCat#bs-12387R; RRID:AB_28952871:200
AntibodyRabbit polyclonal anti-KIF5BAbcamCat#ab5629; RRID:AB_21323791:200
AntibodyMouse monoclonal anti-GFP (B-2)Santa Cruz BiotechnologiesCat#sc-9996; RRID: AB_6276951:1000
AntibodyMouse monoclonal anti-HA (12CA5)Cancer Research UKCat#12CA5; RRID:AB_29207131:1000
AntibodyMouse monoclonal anti-Flag (M1)SigmaCat#F3040; RRID:AB_4397121:200
AntibodyRabbit polyclonal anti-p75NTRCancer Research UKCat#CRD5410; RRID:AB_28643251:200
Recombinant DNA reagentTET ON AdvanceTakara Bio (Clontech)Cat#630930
Recombinant DNA reagentpLVX shRNA Rab10This studyThe shRNA MSH031352 from GeneCopoeia targeting Rab10 has been cloned into a pLVX tight puro plasmid
Recombinant DNA reagentpLVX mCherry-Rab10This studyA mouse mCherry-Rab10 has been cloned into a pLVX tight puro inducible lentiviral vector using XbaI/NheI
Recombinant DNA reagentpLVX myc-Rab10 (shRNA resistant)This studyFrom the pLVX mCherry-Rab10, mCherry has been replaced by myc, and 3 silent mutations have been introduced
Recombinant DNA reagentpEGFP-C1ClontechDiscontinued
Recombinant DNA reagentEGFP-Rab10 WTDOI:10.1111/j.1462–5822.2010.01468.xRRID:Addgene_49472Marci Scidmore lab
Recombinant DNA reagentEGFP-Rab10 T23N; Rab10DNDOI:10.1111/j.1462–5822.2010.01468.xRRID:Addgene_49545Marci Scidmore lab
Recombinant DNA reagentEGFP-Rab10 Q68LDOI:10.1111/j.1462–5822.2010.01468.xRRID:Addgene_49544Marci Scidmore lab
Recombinant DNA reagentHA-Rab10 WTMRC Protein Phosphorylation and Ubiquitylation UnitDU44250Dario Alessi lab
Recombinant DNA reagentGFP-KIF13B10.1111/tra.12692RRID:Addgene_134626Marvin Bentley lab
Recombinant DNA reagentTrkB-FLAG10.1091/mbc.e05-07-0651Francis Lee lab

Neuronal cultures

Request a detailed protocol

Embryonic hippocampal neurons from C57BL/6 mice of either sex and embryonic age of 16–17 days were dissected adapting previously described protocols (Kaech and Banker, 2006). Dissection was performed in cold Hanks’ balanced salt solution (HBSS) and the tissue was collected in cold Hibernate E medium (Thermo Fisher, #A1247601). After incubating for 10 min in 300 µL of Accumax (Innovative Cell Technologies, #AM105) diluted in HBSS (1:1), tissue was washed in HBSS, resuspended in warm plating medium (Minimum Essential Medium supplemented with 10% horse serum, 0.6% glucose, and 2 mM glutamine), and mechanically dissociated by pipetting. 10,000–12,000 neurons per cm2 were then seeded on glass coverslips or microfluidic chambers, pre-coated with 1 mg/mL poly-L-lysine. Before coating, glass coverslips were treated overnight with NoChromix (Godax Laboratories), washed three times and sterilised in 70% ethanol. Microfluidic chambers were produced in-house as previously described (Restani et al., 2012; Sannerud et al., 2011). Polydimethylsiloxane inserts were fabricated from resin moulds, which are replicas of the master template produced by soft lithography, and then irreversibly bound to glass-bottom dishes (WillCo Wells, #HBSB-3512) by plasma treatment. Neurons were left in plating medium for 1.5 hr and then shifted to maintenance medium (Neurobasal supplemented with B27, 2 mM glutamine, 0.6% glucose, and antibiotics). Half of the culture medium was replaced by fresh medium every 3–4 days.

Immunofluorescence

Request a detailed protocol

Cells were washed in phosphate buffer saline (PBS) and fixed for 15 min in 3% paraformaldehyde (PFA) and 4% sucrose dissolved in PBS. Next, they were incubated in 0.15 M glycine dissolved in PBS for 10 min and then blocked and permeabilised simultaneously by incubation in 5% bovine serum albumin (BSA) and 0.1% saponin in PBS for 1 hr. Samples were incubated at 4°C overnight with primary antibodies diluted in 5% BSA, 0.05% saponin, 0.1 mM CaCl2, and 0.1 mM MgCl2 dissolved in PBS at the concentrations indicated in the Key resources table. Then, cells were washed three times with PBS and incubated for 90 min with Alexa Fluor-conjugated secondary antibodies 1:400 (Thermo Fisher) in 5% BSA, 0.05% saponin, 0.1 mM CaCl2, and 0.1 mM MgCl2 dissolved in PBS. 4′,6-diamidino-2-phenylindole (DAPI) was added with the secondary antibodies when appropriate. Finally, coverslips were washed in PBS and mounted with Mowiol.

Rab10 knockdown

Request a detailed protocol

Rab10 was knocked down by transducing 5 days in vitro (DIV) hippocampal neurons with an inducible shRNA Rab10 lentivirus (pTightPuro-shRNA Rab10) and its doxycycline-dependent regulator TET-ON Advance (Clontech). After 48 hr, they were treated with doxycycline 1 µg/mL and kept in the incubator for 12, 18, 24, 36, and 48 hr. Cell density, levels of Rab10 and TrkB, as well as the general health of the culture were analysed at 12, 24, and 48 hr to establish the optimal time frame for the following experiments.

Transfection and plasmids

Request a detailed protocol

Hippocampal neurons were transfected at 7 DIV using Lipofectamine 2000 (Thermo Fisher, Cat# 11668019). Experiments were carried out after 20–24 hr. The pEGFP-C1 plasmid is from Clontech (Addgene plasmid # 13031, RRID:Addgene_13031), the plasmids for EGFP-Rab10 WT (RRID:Addgene_49472), EGFP-Rab10 T23N (RRID:Addgene_49545), and EGFP-Rab10 Q68L (RRID:Addgene_49544) were a gift from Marci Scidmore (Huang et al., 2010), TrkB-FLAG plasmid was a gift from Francis Lee (Chen et al., 2005). Neuro-2a cells were transfected using Lipofectamine 3000 (Thermo Fisher, Cat# L3000001) and the experiments were done 48 hr later. The GFP-KIF13B plasmid (RRID:Addgene_134626) was a gift from Marvin Bentley (Yang et al., 2019), whilst the HA-Rab10 plasmid was provided by Dario Alessi and Miratul Muqit (Dundee University, DU44250).

Retrograde accumulation and signalling assays

Request a detailed protocol

Hippocampal neurons were seeded in custom-made microfluidic devices and after 5 DIV they were transduced with the inducible shRNA Rab10 and TET-ON Advance lentiviruses. At 7 DIV, dishes with overt axon crossing were selected and doxycycline 1 µg/mL was added to the cell bodies, 18–22 hr later the media was replaced with Neurobasal in somatic and axonal compartments to deplete cells from endogenous growth factors for 1 hr. For analysing the retrograde accumulation of TrkB, we added polyclonal antibodies against the extracellular domain of TrkB (1:50 rabbit anti-TrkB, Millipore, Cat# AB9872, RRID:AB_2236301) together with 20 ng/mL BDNF for 2.5 hr. After PFA fixation, the transport of internalised antibodies was revealed by incubating the somatic compartment with fluorescently labelled secondary antibodies. The same protocol was used to study retrograde propagation of neurotrophic signalling; after 1 hr of growth factor depletion, axons were stimulated with 20 ng/mL BDNF for 2.5 hr, and after fixation, phosphorylation of CREB in the nucleus was analysed by immunofluorescence. Transduction with a myc-Rab10 containing three silent mutations was used for rescue.

Co-localisation studies

Request a detailed protocol

To analyse the presence of two markers in the same organelle, we used confocal z-stack images (voxel size: 0.197 × 0.197 × 0.5 µm) and the confined displacement algorithm to measure Manders’ correlation index within axons and determine its statistical significance compared to random images of identical total intensity and shape (Ramírez et al., 2010). To compute random scenarios, seven random radial displacements were taken at a maximum radial distance of 12 pixels (a total of 353 samples), and histograms binning = 16. CDA was implemented by using the plugin from the GDSC University of Sussex (http://www.sussex.ac.uk/gdsc/intranet/microscopy/UserSupport/AnalysisProtocol/imagej/colocalisation). All the data points were plotted and the mean and standard error are indicated for each group and compared using Student’s t-test. Statistical significance of the individual data point is colour-coded (see figure legends).

Super-resolution radial fluctuations

Request a detailed protocol

High-fidelity super-resolution information was extracted from time series of 1000 confocal images per channel by using super-resolution radial fluctuations (SRRF) algorithm (Culley et al., 2018b). Super-resolution images were then quality-controlled by using Super-Resolution Quantitative Image Rating and Reporting of Error Locations (SQUIRREL) algorithm (Culley et al., 2018a). Implementation of the algorithms was done in FIJI by using the open-source plugin NanoJ-core (https://henriqueslab.github.io/resources/NanoJ).

Immunoendocytosis

Request a detailed protocol

Hippocampal neurons transfected with TrkB-Flag were kept in Neurobasal media for 1 hr and then incubated on ice with 1:50 mouse anti-Flag antibody (Sigma-Aldrich, Cat# F3040, RRID:AB_439712). In selected samples, 20 nM HCT was also added (Deinhardt et al., 2006). Internalisation of receptors was then induced by incubation with 50 ng/mL BDNF for 30 min at 37°C. Antibodies bound to receptors still at the cell surface were dissociated by washing twice for 1 min in PBS supplemented with 1 mM EDTA. In experiments to measure recycling of internalised receptors, neurons were further stimulated with BDNF for other 60 min in the presence of an Alexa Fluor647-conjugated anti-mouse secondary antibody, fixed, permeabilised, and incubated with an Alexa Fluor555-conjugated anti-mouse secondary antibody to detect total internalised receptor.

Axonal transport

Request a detailed protocol

Axonal transport of overexpressed fluorescent proteins and internalised fluorescent antibodies was analysed from confocal time series of 1 frame/s and a pixel size of ~0.1 × 0.1 µm2, captured during 5 min intervals at different time points by using a Zeiss LSM 780 NLO multiphoton confocal microscope with an oil immersion ×63 objective and equipped with an environmental chamber (Zeiss XL multi S1 DARK LS set at 37°C and environmental CO2). For these experiments, neurons were cultured on 25 mm coverslips kept in Neurobasal for 1 hr prior to cell imaging and mounted inside Attofluor chambers (Thermo Fisher Scientific, Cat# A7816) with BrightCell NEUMO photostable media (Sigma-Aldrich, Cat# SCM145) supplemented with 10 mM HEPES. Speed, pausing, and direction of labelled organelles were analysed from kymographs by using Kymoanalyzer set of macros from Encalada lab (https://www.encalada.scripps.edu/kymoanalyzer; Neumann et al., 2017).

Co-immunoprecipitation

Request a detailed protocol

Neuro-2A cells have been originally obtained from ATCC and modified to stably express TrkB-Flag (Terenzio et al., 2014). Their identity has been confirmed in the future batches by assesing their morphology and probing for TrkB-flag. They have been tested for mycoplasma contamination routinely. Cells transfected with plasmids encoding HA-Rab10 and GFP-KIF13B were starved for 1 hr, stimulated or not with BDNF for 30 min at 37°C, and then scrapped and incubated at 4°C for 30 min in lysis buffer containing 25 mM 4-morpholine-propanesulfonic acid sodium salt (MOPS) pH 7.2, 100 mM KCl, 10 mM MgCl2, 1% octyl-phenoxy-polyethoxyethanol (IGEPAL CA-630, Sigma-Aldrich, Cat# I3021) and HALT proteases and phosphatases inhibitor (Thermo Fisher, Cat# 78425). After clearing by centrifugation at 21,000 × g for 10 min, the detergent concentration was adjusted to 0.5%. Pre-washed anti-HA magnetic beads (Pierce, Cat# 88837) were incubated with lysates overnight at 4°C. Samples were washed in lysis buffer five times and elutes in loading buffer (NuPAGE LDS sample buffer, Thermo Fisher, Cat# NP0007, supplemented with 50 mM DTT) at 95°C for 10 min. The levels of immunoprecipitated Rab10 and KIF13B were analysed by Western blot using mouse anti-HA (Cancer Research UK, Cat# 12CA5, RRID:AB_2920713) and mouse anti-GFP B-2 (Santa Cruz Biotechnologies, Cat# sc-9996, RRID:AB_627695) antibodies.

Statistical analysis

Request a detailed protocol

Data generated in independent experiments were tested for normality and homoscedasticity to apply the appropriate corrections to the statistical tests. Each specific test, its degrees of freedom, and level of significance are indicated in the respective figure legends. Further details are summarised in the statistical annex (Table 1). Plots show mean ± standard error, and exact p-values are indicated when relevant.

Table 1
Statistical summary.
FigureVariableTestGroupsDegrees of freedomp value
Figure 1bCells per fieldTwo-way ANOVAKnock down; time; interactionF(1,68)0.0270; 0.0406; 0.2114
Multiple comparisonsControl: shRNA Rab10 at 24 hrt(68)>0.9999
Multiple comparisonsControl: shRNA Rab10 at 48 hrt(68)0.0183
cRab10 expressionTwo-way ANOVAKnock down; time; interactionF(1,653); F(2,653); F(2,653)<0.0001
Multiple comparisonsControl: shRNA Rab10 at 12 hrt(653)0.2012
Multiple comparisonsControl: shRNA Rab10 at 24 hrt(653)<0.0001
Multiple comparisonsControl: shRNA Rab10 at 48 hrt(653)<0.0001
Figure 2dTrkB accumulationUnpaired Student’s tControl: shRNA Rab10t(140)<0.0001
eTrkB accumulation and Rab10 expressionPearson rControl: shRNA Rab10XY pairs = 131<0.0001
gpCREB abundanceOne-way ANOVAControl, shRNA Rab10 and rescueF(2,280)<0.0001
Multiple comparisonsControl: shRNA Rab10t(280)<0.0001
Multiple comparisonsControl: rescuet(280)0.0336
Figure 3cCo-localisation Rab10 and Rab5Unpaired Student’s tM1 control: BDNFt(46.27)0.0244
Unpaired Student’s tM2 control: BDNFt(48.94)0.4794
dCo-localisation Rab10 and Rab7Unpaired Student’s tM1 control: BDNFt(72.32)0.0621
Unpaired Student’s tM2 control: BDNFt(62.33)0.1043
eArea of overlay HcT and TrkB (retrograde TrkB)One-way ANOVA30, 60, and 90 minF(2,72)<0.0001
Multiple comparisons30:60 mint(72)>0.9999
Multiple comparisons60:90 mint(72)0.0002
Area of overlay retrograde TrkB and Rab10One-way ANOVA30, 60, and 90 minF(2,72)0.2730
Multiple comparisons30:60 mint(72)0.5717
Multiple comparisons60:90 mint(72)>0.9999
Figure 5cDirection of Rab10 organellesUnpaired Student’s tAnterograde pre: post BDNFt(14)0.0150
Unpaired Student’s tRetrograde pre: post BDNFt(14)0.0030
Unpaired Student’s tNon-mobile pre: post BDNFt(14)0.4278
Figure 6bKIF13B intensity ratioKolmogorov–Smirnov Student’s tControl: BDNFt(79.5)0.0177
cKIF13B area occupancy ratioKolmogorov–Smirnov Student’s tControl: BDNFt(79.5)0.8024
eKIF5B intensity ratioKolmogorov–Smirnov Student’s tControl: BDNFt(62.5)0.0644
fKIF5B area occupancy ratioKolmogorov–Smirnov Student’s tControl: BDNFt(62.5)0.3738
hKIF13B co-immunoprecipitationPaired Student’s tControl: BDNFt(6)0.0024
Figure 7cRecycling of TrkBUnpaired Student’s tEGFP: Rab10 DNt(60.40)0.3914
eCo-localisation TrkB and Rab5Unpaired Student’s tEGFP: Rab10 DNt(38.22)0.0080

Software

Images were handled, edited, and analysed using ImageJ/FIJI (version 2.1.0, 1.53c). Figures were checked with Coblis (https://www.color-blindness.com/coblis-color-blindness-simulator/) using The Colour Blind Simulator algorithms from Matthew Wickline and the Human-Computer Interaction Resource Network. Palettes were adjusted to maximise visibility. The Orange/Green/Purple balanced look-up table was obtained from Christophe Leterrier’s GitHub repository (https://github.com/cleterrier/ChrisLUTs, copy archived at Lazo, 2023a). Data were imported, analysed, and sorted as R files using RStudio (version 1.0.44). GraphPad Prism for Mac (version 6.00, GraphPad Software) was used for running statistical analysis and generate the plots included in the figures. Illustrations were created with BioRender (http://www.biorender.com). Updated versions of ImageJ macros and R scripts used in this article, as well as the specific implementation of Kymoanalyzer used to analyse our datasets, can be found on our GitHub repository (https://github.com/omlazo; copies archived at Lazo, 2023b, Lazo, 2023c and Lazo, 2023d).

Data availability

Source tables for all the data in the manuscript have been submitted as a supplementary file (source data files 1-7).

References

Decision letter

  1. Suzanne R Pfeffer
    Senior and Reviewing Editor; Stanford University, United States
  2. Mitsunori Fukuda
    Reviewer; Tohoku University, Graduate School of Life Sciences, Japan

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Rab10 regulates the sorting of internalised TrkB for retrograde axonal transport" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife in its current form, but we would happily reconsider a new submission that addresses the reviewers' concerns in full.

As described below, the reviewers felt that the story, while potentially very interesting, was too preliminary in its current form; their specific concerns are described below.

Reviewer #1 (Recommendations for the authors):

Here, the authors use confocal microscopy, super resolution, and live imaging to report a role for Rab10 in the sorting of internalized TrkB receptors and in propagating the BDNF signal from axon terminals to the soma. The imaging analyses, specifically the super-resolution and live imaging, are well-done. However, the manuscript appears to be preliminary at this point; the authors draw strong conclusions about the mechanism of action of Rab10 in TrkB sorting that remain to be fully supported.

Concerns:

1. The authors see little co-localization between endogenous Rab10 and TrkB. The conclusion that Rab10 and TrkB are retrogradely co-transported is based on Rab10 over-expression, which raises concerns about the specificity of the interactions and artifacts related to over-expressing a Rab GTPase.

2. Other than P-CREB immunofluorescence experiments, the authors do not provide data to show the functional effects of Rab10 depletion on BDNF-mediated trophic effects. The authors state in the Discussion that Rab10 knockdown perturbs neuronal morphogenesis and survival, but do not show this data. These data are necessary to strengthen the conclusion that Rab10 is necessary to mediate BDNF-dependent trophic signaling. On a related note, the color intensity scales to show TrkB or P-CREB levels in cell bodies with Rab10 depletion are not adequate.

3. It is interesting that BDNF enhances anterograde Rab10 transport based on live imaging. Additional analyses are needed to strengthen the conclusion that Rab10 is enriched in distal axons and to define the mechanisms (phosphorylation?) underlying Rab10 recruitment/retention

4. The anterograde motor Kif13B is suggested to be involved in BDNF-induced anterograde transport of Rab10 based on confocal microscopy analyses in fixed neurons. However, this aspect is not fully developed. The authors should determine if Kif13B directly interacts with Rab10 and the necessity of Kif13B for anterograde Rab10 transport, retrograde transport of TrkB receptors or BDNF-mediated trophic effects.

5. The suggestion that Rab10 regulates "sorting" of internalized TrkB receptors is very interesting. The authors show nicely (in Figure 6) that dominant negative Rab10 (rab10DN) does not disrupt TrkB recycling in axons, but increases the co-localization of TrkB with Rab5. The images in Figure 6d also show that Rab5 domains appear to be larger in the Rab10 DN expressing neurons, which is consistent with early endosomes being perturbed. However, additional analyses are needed to define how exactly Rab10 coordinates with Rab5 and Rab7 in "handing off" TrkB from early endosomes to late endosomes in axons. Which axonal segments were imaged in the analyses in Figure 6? Was this in the distal axons? What happens to TrkB flux in mid-axons and retrograde accumulation in cell bodies in the Rab10 DN or shRNA expressing neurons? Does expression of Rab10 shRNA or Rab10 DN decrease co-localization of retrogradely transported TrkB with Rab7?

6. The authors should include important controls to show the specificity of Rab antibodies, and the knockdown of Rab10 using shRNA. In Figure 1f, the Rab10 levels (assessed by Rab10 immunofluorescence) appear to not be significantly affected by the kd or rescue, which raises questions about the antibody specificity, as well as efficacy of the kd.

Comments above address concerns about conclusions that are not fully supported by the data. Here are some specific suggestions for additional analyses plus other recommendations.

1. The authors should use additional measures to support that Rab10 associates/co-migrates with TrkB, for example by biochemical methods, or by performing live imaging of co-transport in neurons transfected with Rab10-GFP but with endogenous Rab10 knocked down to mitigate the OE concern.

2. The authors should strengthen the data to support that Rab10 accumulates in distal axons in a BDNF-dependent manner, using confocal microscopy or biochemical means. Live imaging analyses can be used to determine if Rab10 is being mobilized long-distance from cell bodies by the retrograde BDNF signal, or if this is a local effect of BDNF on Rab10 recruitment to the axon terminals.

3. In Figure 1c, the authors show reduced accumulation of axon-derived TrkB receptors in neurons expressing Rab10 shRNA. Images of the axons should be included to assess whether the TrkB receptors accumulate in distal axons, as proposed in the model (Figure 7)

4. The authors also propose a mechanism for BDNF-induced enrichment of Rab10 in axons that involves Rab10 phosphorylation. The manuscript would be strengthened by additional analyses to define mechanisms underlying Rab10 recruitment. The imaging analyses of Kif13B could be complemented with biochemical methods to show BDNF-induced association with Rab10.

5. Rab10 has been proposed to play a role in generating tubular endosomes-it would be of interest to use super-resolution imaging to define the morphology of TrkB/Rab10-positive endosomes in distal axons.

6. The authors should assess whether the Rab10 kd or expression of Rab10 DN has general effects on neuronal viability or affect total TrkB levels, and not specifically on retrograde transport of axonally-derived TrkB receptors.

Reviewer #2 (Recommendations for the authors):

Signaling from neurotrophins requires endocytosis of the ligand-bound receptors (such as NGF-TrkA or BDNF-TrkB), sorting into signaling endosomes and subsequent transport retrogradely along the axon back to the soma for signaling. Multiple Rab proteins have been implicated in this trafficking pathway, most prominently Rab5 and Rab7. This manuscript shows that Rab10 is also required for retrograde arrival of signaling endosomes in the soma and subsequent signaling. The most interesting finding is that Rab10-positive organelles show increased anterograde transport in response to BDNF and are not part of the TrkB-positive retrogradely moving signaling endosome itself. Interference with Rab10 traps TrkB in Rab5-positive early endosomes, and the authors propose that Rab10 is required for sorting of TrkB into signaling endosomes. Since Rab10 is largely in a distinct set of axonal organelles, the point of conversion of endocytosed TrkB and Rab10 is not clear. The paper stops short of decisively answering where Rab10 is active to promote sorting and what BDNF signaling does to change Rab10 motility patterns.

1) Rab10 was enriched in SEs by SILAC/mass spec (from a previous paper by this group) but now is not in the retrograde carriers. Can you discuss?

2) Are the anterograde Rab10 carriers endosomally derived? It would be important to characterize the cargos of these carriers. I think the observation that BDNF affects the directionality of the Rab10 carriers is very intriguing. Were these in microfluidic chambers? How is BDNF signaling conveyed to Rab10 carriers?

2) There is a lot of conjecture about Rab10 vesicles delivering machinery to the distal axon early endosome, but no experiment to address this. Where is TrkB in Rab10 interference conditions? It seems to be stuck with Rab5 and not sorting back to the axonal plasma membrane or into retrograde carriers. Is it accumulating in distal axon tips? It would be good to have live imaging of TrkB in Rab10-DN conditions along axons. The recycling experiments (Figure 6) are very interesting, but it looks like this is happening along the axon shaft. Is this correct?

3) It is not always clear what experiments are done in microfluidic chambers. Please specify in each figure.

4) Is p75 not in a complex with TrkB after BDNF binding? What is the interpretation/implication of p75 being in different carriers in the axon? I find this a very interesting observation, but the relevance is not explained or further explored.

5) The data with kinesins is not adding much to the understanding of the pathway. No interference with any kinesin is performed.

3) All the discussion of LRRK2 and phosphorylation is besides the point since the authors do not test involvement of this regulatory mechanism. This would be very interesting, but as is, the paper stops short of decisively answering where Rab10 is active to promote sorting and what BDNF signaling does to change Rab10 motility patterns.

There are several very interesting observations, but the scope of the study is somewhat limited.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Rab10 regulates the sorting of internalised TrkB for retrograde axonal transport" for further consideration by eLife. Your revised article has been evaluated by Suzanne Pfeffer (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission. We realize that this manuscript was already reviewed and resubmitted, and the reviewers hesitated to ask for more data, but they felt strongly that some additional work would be required to support the conclusions of your study. Note that it is not always possible to secure the same reviewers with each round of submission.

Essential revisions:

(The other issues that follow below are worthy of text clarification)

1) The fact that there is little colocalization between Rab10 and TrkB needs further characterization. That is not consistent with "a new Rab10 organelle" that mediates TrkB sorting and retrograde transport. Please test other recycling endosome Rabs, such as Rab11, Rab14, Rab10 to determine if these are really novel organelles or simply Rab11-recycling endosomes that also contain Rab10.

2) Please test whether Kif13B or Kif5B knock-down affects TrkB transport.

3) The IP experiment shown in Figure 6G-H is not very convincing. Can you blot for endogenous Kif13B co-precipitation with HA-Rab10? Second, it appears that there is more HA-Rab10 precipitated in BDNF-treated samples, thus, it is unclear whether slight increase in GFP-Kif13B co-precipitation is due to that rather than increased interaction with Rab10. Third, why there is some GFP-Kif13B in sample that does not express HA-Rab10? Please address.

Reviewer #1 (Recommendations for the authors):

In the revised manuscript, the authors faithfully addressed most of the reviewers' concerns. I have only few suggestions (see below).

1. Description of "Rab10-EGFP" and "Rab10-HA" should be considered, because EGFP and HA were tagged to the N-terminus of Rab10. Indeed, the authors have used "GFP-Rab5" in their previous paper (see ref. 12).

2. Several resources were missing in the Key resources table, e.g., shRNA-resistant Rab10, Rab10-Q68L, and JIP3/4 antibodies.

3. (Line 564) The reference may be mis-cited. Ref. 22 is correct?

Reviewer #2 (Recommendations for the authors):

This is a manuscript that focuses on understanding the machinery governing generation and transport of TrkB-containing endosomes, especially the involvement of Rab10 in the process. Since we are only beginning to understand the long-range signaling by endosomes in neurons, the topic is quite interesting and potentially significant. Additional data are needed to confirm association of endogenous Rab10 with TrkB-endosomes and the potential novelty of these structures.

1) Rab10 knock-down should be evaluated using western blotting or qPCR. It is much more quantitive method than microscopy. Additionally, how authors control for possible off-target effects? They used rescue in Figure 2f-g. Similar rescues should be used in other analyses in Figures 1-2.

2) Rescue images in Figure 2f are not particularly convincing. Not sure how authors can tell apart the neurons that are presumably expressing rescue plasmid.

3) As other reviewers pointed out there is very little colocalization between Rab10 and TrkB. That is not consistent with authors conclusions that Rab10 marks a new type of organelle that mediates TrkB sorting and retrograde transport. Authors speculate that these are transient associating but provide little evidence to support that. Overexpression of both, Rab10 and TrkB is hardly a strong evidence for that since even under these high overexpressed conditions the colocalization is still limited. Did authors try to overexpress (with TrkB) other recycling edosome Rabs, such as Rab11, Rab14, Rab10. Finally, Rab10 is related to Rab11, which is a well-established regulator of endocytic sorting and recycling. Does Rab10 colocalize with Rab11 (I suspect it will)? Are these really novel organelles or they are simply Rab11-recycling endosomes that also contains Rab10 (as it was shown in other experimental systems).

4) It is not clear why authors picked to analyze Kif5b and Kif13B. There are several other kinesins that were implicated in axonal transport. Why authors chose not to study them?

5) The change in Kif13B and Kif5B intensity in response to BDNF is very moderate at best. It does not help that authors do not show validation of anti-Kif5B or anti-Kif13B antibodies. It would also be good to test whether Kif13B or Kif5B knock-down affects TrkB transport.

6) IP experiment shown in Figure 6G-H is not very convincing. First, since authors used anti-Kif13B antibodies for immunofluorescent microscopy, why they did not blot for endogenous Kif13B co-precipitation with HA-Rab10. Second, it appears that there is more HA-Rab10 precipitated in BDNF-treated samples, thus, it is unclear whether slight increase in GFP-Kif13B co-precipitation is due to that rather than increased interaction with Rab10. Third, why there is some GFP-Kif13B in sample that does not express HA-Rab10?

7) Figure 7. Dominant-negative Rab mutants causes numerous non-specific effects. Since authors has Rab10 shRNA, all recycling experiments need to be done in Rab10-KD cells. Additionally, it is very puzzling that trapping TrkB in early endosomes did not affect recycling. Most plasma membrane receptors recycle by sequential transport from early endosomes to recycling endosomes. Consequently, I would expect that trappin TrkB in early endosomes would decrease its recycling.

Reviewer #3 (Recommendations for the authors):

Points #3, 5 and 6 of reviewer 2 seem very important.

The functional mechanism of Rab10 in BDNF signaling still remains a bit fuzzy since the mobilization of Rab10 in the axon terminal in response to BDNF is counterintuitive to explain the increased signaling to the soma, at least in the absence of further mechanistic insight. The model in Figure 8a thus remains speculative.

https://doi.org/10.7554/eLife.81532.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

We want to thank you and the Reviewers for their positive comments and very insightful suggestions with regard to our manuscript titled “Rab10 regulates the sorting of internalised TrkB for retrograde axonal transport”. Their advice has greatly helped us to clarify the key messages of our manuscript and prompted us to perform additional experiments to make our work more impactful and interesting for the readership of eLife.

We appreciate your kind words while summarising the importance of our work: “defining how neurons respond to spatial extrinsic cues, such as neurotrophins, and relay this information longdistance to influence transcriptional events is an important topic in neurobiology”. This showed us that the main message of our work has been well recognised and valued. Moreover, the Reviewers explicitly mentioned what for us is the main focus of this work: “to report a role for Rab10 (…) in propagating the BDNF signal from axon terminals to the soma” and “that Rab10 is also required for retrograde arrival of signalling endosomes in the soma and subsequent signalling”. Reviewer #2 adds that “the most interesting finding is that Rab10-positive organelles show increased anterograde transport in response to BDNF and are not part of the TrkB-positive retrogradely moving signalling endosome itself”, whilst Reviewer #1 recognises the quality of “the imaging analyses, specifically the super-resolution and live imaging”.

However, there is also consensus regarding the need of a better characterisation of the specific role of Rab10 in the system and the mechanism involving KIF13B in the anterograde transport of Rab10 organelles. In this amended version of the manuscript, we have added a new figure documenting the efficacy of the Rab10 knock down, the effects of its depletion in cell survival and TrkB expression (Figure 1). Moreover, we have expanded our analysis of the BDNFdependent interaction between Rab10 and KIF13B (Figure 6g-h), as well as provided supplementary videos and representative data to support our observations. This new evidence has been carefully calibrated to keep in focus what the Reviewers have also recognised as the key message of our paper: the novel role of BDNF-regulated axonal transport of Rab10 organelles to propagate neurotrophic signalling from the axon to the nucleus of central neurons.

We address below the specific points raised by the Reviewers:

1. Reviewer #1 points out that while we see ”little co-localization between endogenous Rab10 and TrkB (…), the conclusion that Rab10 and TrkB are retrogradely co-transported is based on Rab10 over-expression, which raises concerns about the specificity of the interactions and artifacts related to over-expressing a Rab GTPase”. We agree with the reviewer regarding the fact that endogenous TrkB and Rab10 are unlikely to be retrogradely co-transported in the same organelle, as they correctly conclude from Figure 3e-f. Our data suggest that there is a transient association between Rab10 and the endosome containing TrkB. Therefore, we overexpressed wild-type Rab10 to create an experimental situation in which this transient association is stabilised long enough to be observed as events of co-transport. The reviewer suggested that we should “use additional measures to support that Rab10 associates/co-migrates with TrkB, for example by biochemical methods, or by performing live imaging of co-transport in neurons transfected with Rab10-GFP but with endogenous Rab10 knocked down to mitigate the OE concern”. As mentioned in our manuscript, we obtained independent evidence of the association of Rab10 with neurotrophin signalling endosomes in our early work describing a quantitative proteomic analysis of these organelles using an affinity purification approach in mouse motor neurons (see reference 18). Consistently with the conclusions of this work, we found that Rab10 is not accumulated on the surface of these organelles during signalling endosome maturation, as found for Rab7, further supporting the view that Rab10 interacts transiently with these organelles (see also our response to Reviewer #2 point 7 below). Furthermore, transfection of Rab10-EGFP, even when endogenous Rab10 is knocked down, results in eight fold increase on its expression levels; additionally, biochemistry may provide inconclusive results, as it will fail to reveal dynamic interactions. For these reasons, we have provided as an additional proof of principle of the association of Rab10 to TrkB-positive endosomes, an example of a stationary TrkB-organelle which transiently accumulates Rab10-EGFP for few seconds, just before the onset of processive retrograde transport (Figure 8b). We are actively investigating the mechanism regulating the onset of motion in these organelles, and the molecular events leading to TrkB sorting into retrograde signalling endosomes.

2. Reviewer #1 also indicates that “other than P-CREB immunofluorescence experiments, the authors do not provide data to show the functional effects of Rab10 depletion on BDNF-mediated trophic effects”. While the complexity of retrograde trophic effects of BDNF are well documented [see references 1-3], we have decided to use levels of phosphorylated CREB (pCREB) in the nucleus since activation of this transcription factor constitutes the first node in which the three canonical signalling pathways of TrkB, PI3K-Akt, MAPKs and PLC-γ, converge [see reference 10]. Additionally, the majority of early genes regulated by BDNF depend on the phosphorylation of CREB at serine 133 to increase their expression [see references 19 and 31], making this node a relevant proxy for BDNF signalling and gene response. We have used the signalling readout as a complement to the accumulation of the internalised receptor itself, strengthening the physiological relevance of Rab10-depletion on TrkB transport. The reviewer also makes a very interesting point when focusing on discussion of Rab10 knockdown effect in morphology and survival. They advise that “these data are necessary to strengthen the conclusion that Rab10 is necessary to mediate BDNF-dependent trophic signaling”, adding that we should “assess whether the Rab10 kd or expression of Rab10 DN has general effects on neuronal viability or affect total TrkB levels, and not specifically on retrograde transport of axonally-derived TrkB receptors”. We thank the reviewer for raising these interesting issues. We have now included a new Figure 1, in which we analyse the changes in neuronal density after 24 and 48 hours of Rab10 knockdown. We showed that after 48 hours, there is a significant decrease in survival (Figure 1b), fully justifying our decision of performing this set of experiments between 18 and 24 hours. Importantly, we showed that the levels of endogenous TrkB remain stable even at 48 hours of Rab10 knockdown (Figure 1d), ruling out the depletion of TrkB as a potential confounding factor for the decreased pCREB levels and retrograde TrkB. Please also see point 4 for a discussion about the effects of Rab10 depletion on neuronal morphology. In addition, we followed the suggestion of the reviewer to include “important controls to show the specificity of Rab antibodies, and the knockdown of Rab10 using shRNA. In Figure 1f, the Rab10 levels (assessed by Rab10 immunofluorescence) appear to not be significantly affected by the kd or rescue, which raises questions about the antibody specificity, as well as efficacy of the kd”. In this regard, we have validated the significant decrease of endogenous Rab10 after 24 and 48 hours of treatment with doxycycline (Figure 1c), further optimising our immunofluorescence protocols at the same time.

3. Another interesting point raised by reviewer #1 focused on what happens to the relative concentration of Rab10 and TrkB in the axon. With regard to TrkB, the reviewer points out that: “In Figure 1c (now 2c), the authors show reduced accumulation of axon-derived TrkB receptors in neurons expressing Rab10 shRNA. Images of the axons should be included to assess whether the TrkB receptors accumulate in distal axons, as proposed in the model”. However, no accumulation of TrkB in the periphery is expected upon Rab10 knockdown. Indeed, the population of receptors undergoing retrograde transport is believed to be only a fraction of the total internalised TrkB, with the majority of the receptor engaging in local signalling and recycling back to the plasma membrane [see reference 5]. Moreover, anterograde flux of TrkB in hippocampal neurons has been shown to be regulated by retrograde BDNF signalling [see Zahavi, E. et al. 2021 Dev Cell, DOI: 10.1016/j.devcel.2021.01.010]; therefore, Rab10 knockdown is also expected to decrease the anterograde delivery of TrkB, thus halting any potential distal accumulation of this receptor. We have incorporated these considerations into the model shown in Figure 8a and the Discussion.

Regarding the distribution of Rab10, the reviewer suggests that “it is interesting that BDNF enhances anterograde Rab10 transport based on live imaging. Additional analyses are needed to strengthen the conclusion that Rab10 is enriched in distal axons and to define the mechanisms (phosphorylation?) underlying Rab10 recruitment/retention”. At this point it is important to emphasise that our model does not predict an absolute distal enrichment of Rab10 upon BDNF stimulation either, but a dynamic readjustment on demand of the availability of Rab10 at the site of endocytosis. At any given point, Rab10 carriers are moving anterogradely and retrogradely along the axon. This balance will be transiently shifted towards retrograde or anterograde upon changes in ligand concentration; therefore, the increase in the anterograde bias will proceed only until the demand has been met and it is very unlikely that it results in significant accumulation of Rab10 under physiological conditions. This is precisely why this phenomenon is best appreciated when we perform longitudinal live-cell imaging of axons before and after BDNF stimulation (Figure 5). Regarding the mechanism regulating Rab10 direction bias, we agree with the reviewer about phosphorylation potentially playing a crucial role. We have included this possibility in the discussion, and we have focused our experiments in providing evidence that KIF13B recruitment to Rab10 organelles is increased by BDNF. Following the suggestion of the reviewer that “additional analyses to define mechanisms underlying Rab10 recruitment” and that “the imaging analyses of Kif13B could be complemented with biochemical methods to show BDNF-induced association with Rab10”, we have performed co-immunoprecipitation experiments to show that BDNF significantly increases the interaction between Rab10 and KIF13B (new Figure 6g-h). In line with our interpretation of the imaging analysis in axons, these biochemical data suggest that the effect of BDNF in the recruitment of KIF13B to Rab10 organelles is robust and provides a plausible mechanism for the increase in the anterograde transport of these organelles.

4. Along the same line, both reviewers point out that “the anterograde motor Kif13B is suggested to be involved in BDNF-induced anterograde transport of Rab10 based on confocal microscopy analyses in fixed neurons. However, this aspect is not fully developed. The authors should determine if Kif13B directly interacts with Rab10 and the necessity of Kif13B for anterograde Rab10 transport, retrograde transport of TrkB receptors or BDNF-mediated trophic effects” and “the data with kinesins is not adding much to the understanding of the pathway. No interference with any kinesin is performed”. While the main message of our manuscript is unravelling a new role for axonal Rab10 organelles as dynamic regulators of retrograde TrkB trafficking and signalling, in this amended version of our manuscript, we have presented evidence showing that one specific kinesin, KIF13B, is preferentially recruited to Rab10 organelles upon BDNF stimulation. In stark contrast, the interaction of KIF5B, one of the most abundant axonal kinesins, remains unaffected. We have also responded to the request of confirming the BDNF-mediated interaction between Rab10 and KIF13B with a new set of biochemical experiments (Figure 6g-h).

Although we considered the idea of testing the effects of overexpressing the C-terminus of KIF13B lacking the first 441 residues (KIF13B DN), which includes the motor domain, we reasoned that these experiments may be particularly difficult to adequately control, given that KIF13B has a major role in establishing neuronal polarity [see reference 59]. Our preliminary experiments indeed confirm that the expression of KIF13B DN causes a decrease in morphological complexity that is qualitatively similar to the knockdown of Rab10 (see Author response image 1), further supporting our analysis.

Author response image 1

Comparison of the morphological changes of hippocampal neurons treated with shRNA directed against Rab10 or transfected with a dominant negative version of KIF13B. Neurons in (a) have been transduced with a doxycycline-inducible shRNA system to knock down Rab10 and treated with doxycycline for 48 hours, as indicated in the main Figure 1. In this example, the decrease of Rab10 expression is concomitant with a reduction of dendritic complexity similar to the observed in neurons that have been expressing a motorless mutant of KIF13B (KIF13B DN) for 48 hours (b). Scale bars, 50 µm.

5. Another aspect about which both reviewers had interesting insights and suggestions was the evidence about the role of Rab10 in sorting of TrkB from early endosomes. Reviewer #1 indicates that “the suggestion that Rab10 regulates "sorting" of internalized TrkB receptors is very interesting. The authors show nicely (in Figure 6 [now Figure 7]) that dominant negative Rab10 (rab10DN) does not disrupt TrkB recycling in axons, but increases the co-localization of TrkB with Rab5. The images in Figure 6d [now Figure 7d] also show that Rab5 domains appear to be larger in the Rab10 DN expressing neurons, which is consistent with early endosomes being perturbed. However, additional analyses are needed to define how exactly Rab10 coordinates with Rab5 and Rab7 in "handing off" TrkB from early endosomes to late endosomes in axons” and then adds the following questions: “Which axonal segments were imaged in the analyses in Figure 6? Was this in the distal axons? What happens to TrkB flux in mid-axons and retrograde accumulation in cell bodies in the Rab10 DN or shRNA expressing neurons? Does expression of Rab10 shRNA or Rab10 DN decrease co-localization of retrogradely transported TrkB with Rab7?”. In the same line, reviewer #2 adds that “there is a lot of conjecture about Rab10 vesicles delivering machinery to the distal axon early endosome, but no experiment to address this. Where is TrkB in Rab10 interference conditions? It seems to be stuck with Rab5 and not sorting back to the axonal plasma membrane or into retrograde carriers. Is it accumulating in distal axon tips? It would be good to have live imaging of TrkB in Rab10-DN conditions along axons. The recycling experiments (Figure 6) are very interesting, but it looks like this is happening along the axon shaft. Is this correct?”. We thank to the Reviewers for these comments. Indeed, the confocal images of both experiments shown in Figure 7 show that the expression of dominant-negative Rab10 results in enlarged Rab5-positive early endosomes accumulating internalised TrkB receptor, without affecting its internalisation rate nor its recycling to the plasma membrane. In light of these results, we propose that the increase of TrkB in early endosomes accounts for a specific deficits in its sorting to retrograde signalling endosomes. These experiments were done in mass cultures and mid-axons were imaged, since they represented the more isolated segments allowing analysis at the level of individual axons, something that is rarely possible in microfluidic chambers, where axons are bundled together within microgrooves. BDNF was added to the media in these cultures, therefore, both internalisation and recycling are observed along the entire axon and not uniquely at axon terminals. Interestingly, it is under these same exact conditions that the live-cell imaging of Rab10EGFP (Figure 5) has been done, showing that the BDNF-induced increase in net anterograde transport is not a consequence of a gradient of ligand, but an intrinsic feature of the system which we propose is the result from an increased interaction of Rab10 organelles with the anterograde motor KIF13B. In our model (Figure 8a), we have integrated the fact that the main source of BDNF in the nervous system are synaptic targets; therefore, this mechanism would operate mobilising Rab10 organelles towards nerve terminals or retaining them at TrkB internalisation hotspots until the system reaches equilibrium.

Regarding the hypothesis that Rab10 organelles deliver molecules essential for sorting to the distal axon, it is important to notice that Rab10 has been shown to recruit TBC-2 to the membrane of early endosomes, thus regulating the sorting of cargoes towards basolateral recycling [see reference 57]. Another additional example is the cooperation between Rab10 and Rab5 for the sorting of dense core vesicles in the Golgi apparatus [see Sasidharan, N. et al. 2012, PNAS, DOI: 10.1073/pnas.1203306109]. Interestingly, one of the best characterised roles of KIF13B is the anterograde transport of phosphoinositide-3,4,5-triphosphate (PIP3), which is crucial for the regulation of Akt signalling and endosomal maturation [see reference 58]. In Drosophila, the orthologue of KIF13B, Khc-73, has been also involved as a regulator of BMP retrograde signalling carriers [see reference 61]. As reviewer #2 pointed out (“Are the anterograde Rab10 carriers endosomally derived? It would be important to characterize the cargos of these carriers”), a more detailed and unbiased analysis of the composition of the pool of anterograde Rab10-positive organelles is warranted to identify specific components of the sorting machinery.

Although we agree with the Reviewers when predicting of a decreased interaction between TrkB and Rab7 in the absence of functional Rab10, we believe that this experiment would not add to the model, considering that we have already shown an increased accumulation of TrkB in Rab5 domains in the absence of functional Rab10 (Figure 7), and a significant decrease in retrograde TrkB delivered to the soma (Figure 2). The transition between a Rab5-Rab10 hybrid domain to Rab7-positive membranes (or “handing off”) fits with the almost complete exclusion between Rab10 and Rab7 that we show in Figure 3a-b. However, it is important to note that not all retrograde signalling carriers are Rab7-positive, and signalling endosomes have been shown to be heterogenous in nature, including Rab5-positive compartments moving in a less processive manner in the retrograde direction and carrying signalling molecules for long distances [see references 12 and 21]. Indeed, neurons expressing dominant-negative Rab10 display less processive retrograde transport of internalised TrkB compared to neurons transduced with wildtype Rab10 (see Author response image 2). Additionally, in Figure 8b, we have provided an interesting example of transient increase of Rab10 in a TrkB-containing organelle right after its transition between stationary and retrograde, which points in the same direction. These data and references have been included in the discussion of the revised manuscript.

Author response image 2
Retrograde transport of TrkB in neurons expressing Rab10 DN.

Using the same methods illustrated in Figure 4ad, TrkB-Flag was internalised and axons of neurons expressing Rab10 DN were imaged. (a) Example of a retrograde TrkB carrier. (b) Kymograph showing several cargoes with different degrees of processivity and pausing frequency.

6. Reviewer #2 suggested “to use super-resolution imaging to define the morphology of TrkB/Rab10-positive endosomes in distal axons”, considering that “Rab10 has been proposed to play a role in generating tubular endosomes”. We thank to the reviewer for this suggestion. We have included in the revised manuscript Supplementary Video 3, showing examples of tubular endosomes in the cell body of a living neuron. This video, which was recorded by using high resolution Airyscan confocal microscopy, clearly shows that the morphology of the Rab10 organelles is indeed very diverse

7. Reviewer #2 asked us to briefly explain the relationship between what we have found in this work and the mass spec analysis from signalling endosomes in Debaisieux et al. 2016 [see reference 18]: “Rab10 was enriched in SEs by SILAC/mass spec (from a previous paper by this group) but now is not in the retrograde carriers. Can you discuss?”. In this work, we have used the binding fragment of tetanus toxin (HcT) to isolate organelles containing this cargo from mouse stem cells-derived motor neurons after 10, 30 and 60 minutes of endocytosis. Whilst organelles isolated at early time points are enriched in endocytic and early endosomal markers, fractions isolated at later time points are positive for late endosomal markers. Hence, the abundance of Rab7 increase steadily at 30 and 60 min as shown in the figure below (see Author response image 3). In contrast, Rab10 first decreases and then remain constant at 30 to 60 mins, suggesting that it is not accumulated on these organelles during signalling endosome maturation. These data fit our confocal microscopy analysis using in hippocampal neurons shown in Figure 3e-f.

Author response image 3
Rab10 and Rab7 show different recruitment behaviour in signalling endosomes.

Cross correlation of the enrichment of >2,000 proteins detected from immunoisolated HcT-containing signalling endosomes purified from mouse stem cell-derived motor neurons. Rab7 has been highlighted as example of a protein that is enriched at later time points, whereas Rab10 is preferentially associated to early compartments. Modified from Debaisieux et al. 2016 [ref. 18].

8. Reviewer #2 noted that “It is not always clear what experiments are done in microfluidic chambers. Please specify in each figure”. We apologise for the confusion. We have now indicated in the revised manuscript when experiments have been performed in microfluidic chambers. When not indicated, they have been done in mass culture. The reviewer adds: “I think the observation that BDNF affects the directionality of the Rab10 carriers is very intriguing. Were these in microfluidic chambers? How is BDNF signaling conveyed to Rab10 carriers?”. As previously explained (see point 5), Rab10-EGFP live-cell imaging have been done in mass culture, so BDNF signalling can be acting along the entire axon. However, preliminary experiments in microfluidic chambers have shown that the same change in the direction bias occur when BDNF is added only to the terminals, suggesting that retrograde propagation of BDNF signalling is sufficient to inform Rab10-positive organelles in transit. Whilst the exact mechanism is not yet clear, coupling or merge of Rab10 organelles may facilitate the anterograde mobilisation of this compartment, as exemplified by the new Supplementary video 2, where a small stationary Rab10-positive organelle merge with a larger mobile compartment and the resulting Rab10positive combined compartment continues its anterograde motion.

9. Reviewer #2 also asked the following questions: “Is p75 not in a complex with TrkB after BDNF binding? What is the interpretation/implication of p75 being in different carriers in the axon? I find this a very interesting observation, but the relevance is not explained or further explored”. We thank the reviewer for highlighting this interesting aspect of the previous Figure 4. Whereas it is not entirely surprising that TrkB and p75 can be found in different organelles given their differential mechanisms of internalisation and post-endocytic dynamics; the presence of p75 alone in anterograde Rab10 carriers results intriguing. Bruce Carter’s group has previously shown that p75 is mobilised to the plasma membrane following TrkA activation with NGF in sympathetic neurons, which results in stronger TrkA signalling capabilities [see reference 55]. They show that this mobilisation of p75 depends on the activation of Arf6. Interestingly, in C. elegans it has been shown that Rab10 and Arf6 reside in the same subset of organelles. Whether concentration of p75 receptor in different domains of the axon is regulated by Rab10 organelles is an interesting possibility that we will be excited to explore in the future.

10. Reviewer 2# has pointed out that “all the discussion of LRRK2 and phosphorylation is besides the point since the authors do not test involvement of this regulatory mechanism”. We agree with the reviewer, and therefore, the discussion of this aspects has been summarised and focused on Rab10-phosphorylarion as a potential regulatory mechanism for the recruitment of motors and other effectors.

[Editors’ note: what follows is the authors’ response to the second round of review.]

We thank you and the reviewers for their positive comments and very insightful suggestions during the second round of revisions of our manuscript titled “Rab10 regulates the sorting of internalised TrkB for retrograde axonal transport”. We were delighted to learn that Reviewer #1 concerns were fully satisfied, and very grateful for the advice from Reviewers #2 and #3, who not only recognised the quality and interest of our work, but also helped us to further distil the key messages of our manuscript. Their comments prompted us to add extra controls and attempt new experiments, which made our work more robust and interesting for the readership of eLife.

Three main questions persisted from Reviewers #2 and #3; they are all addressed in this resubmission. The first concern was about the nature of the Rab10 axonal compartments, and whether they correspond to previously described recycling endosomes regulated by other Rab GTPases; the other two questions both focussed on the specificity of KIF13B as the kinesin mobilising anterograde axonal Rab10. In this revised version, we present novel evidence addressing these matters and other minor points raised by the Reviewers. At the same time, we were pleased to see that none of the questions concerned the key findings of our work: a novel mechanism by which BDNF regulates Rab10 anterograde delivery, and Rab10 mediates sorting of internalised TrkB to the retrograde axonal transport route.

Specifically,

1. Reviewers #2 and #3 agreed when saying that “the fact that there is little colocalization between Rab10 and TrkB needs further characterization”. Additionally, Reviewer #2 suggested an alternative interpretation of our data suggesting a model prescinding of a transient association of TrkB and Rab10 in axons. In figure 3, we show both by quantitative confocal microscopy (figure 3e) and by using super-resolution radial fluctuations (figure 3f) that after 30, 60 and 90 minutes upon BDNF treatment, endogenous retrograde TrkB and Rab10 partially co-localise, and this colocalisation does not increase with time. We also showed that this association can be stabilised if both proteins are overexpressed, so it can be also observed as events of retrograde co-transport (figure 4a-d). Moreover, in the last figure we documented an example where, even under overexpression conditions, the association between a TrkB-containing signalling endosome and Rab10 was still transient and linked to the processivity of the organelle (figure 8b). It is also noteworthy that Rab10 knock-down caused a significant decrease of retrograde transport of endogenous TrkB (figure 2d-e), further stressing the functional relevance of this transient association. Promted by the reviewers asking for additional proofs of this association, we attempted an unbiased proteomic approach based on the immunoisolation of Rab10-positive membranes from primary cortical neurons by using antibody-conjugated magnetic beads followed by mass spectrometry. Using this approach, we confirmed that TrkB is present in Rab10containing membranes. We are currently optimising the immunoisolation approach, and the full results will be made available in a follow-up study. Thus, we have used four different experimental setups to confirm that TrkB transiently associates with Rab10-containing organelles.

2. Along the same lines, Reviewers #2 and #3 pointed that the low level of co-localisation “is not consistent with "a new Rab10 organelle" that mediates TrkB sorting and retrograde transport”. We fully agree with this conclusion and we find very important to clarify that what we have called “Rab10 organelles” can be more precisely described as membrane domains enriched in Rab10, which eventually interact with the endo-lysosomal system. The use of a more precise description of the compartment is particularly important in light of the diverse distribution of Rab10 in multiple organelles and membrane subdomains that has been observed in different cell types [Chua, C. E. L., and Tang, B. L. (2018). Rab 10-a traffic controller in multiple cellular pathways and locations. J Cell Physiol, 233, 6483-94. https://doi.org/10.1002/jcp.26503].

We, therefore, have changed all the references to “Rab10 organelles” into “Rab10 compartments”, which better reflects the potentially heterogeneous nature of the Rab10-containing membranes in the axon. Regarding this aspect, Reviewers also pointed to Rab10 being related to Rab11 and recycling, so they suggest to “test other recycling endosome Rabs, such as Rab11, Rab14, Rab10 to determine if these are really novel organelles or simply Rab11-recycling endosomes that also contain Rab10”. Although Rab10 and Rab11 do not belong to structurally related groups of Rabs, they have been shown to converge in a sub-class of recycling endosomes [Gupta, K., Mukherjee, S., Sen, S., and Sonawane, M. (2022). Coordinated activities of Myosin Vb isoforms and mTOR signaling regulate epithelial cell morphology during development. Development, 149, https://doi.org/10.1242/dev.199363; Homma, Y., and Fukuda, M. (2016). Rabin8 regulates neurite outgrowth in both GEF activity-dependent and independent manners. Mol Biol Cell, 27, 2107-18. https://doi.org/10.1091/mbc.E16-02-0091]. However, there are several reasons why it is unlikely that Rab10-positive compartments we observe moving in the axon are Rab11-containing recycling endosomes. First, Rab11 organelles have been shown to be infrequent in the axon of hippocampal neurons, while we and others observe a large number of Rab10-positive axonal puncta. Second, blocking Rab10 function by overexpressing a dominant negative mutant (T23N) did not alter TrkB recycling in the axon. However, we fully agree with the reviewers regarding the need of a better characterisation of the Rab GTPases that are present in the Rab10-positive axonal membranes. Our unbiased proteomic characterisation of immunoisolated Rab10 compartments revealed the presence of more than 30 different Rabs, including those associated with the endolysosomal system (Rab4, Rab5, Rab7, Rab11). Importantly, because our starting material for immunoisolation was not pure axonal membranes, we decided to assess the co-localisation of endogenous Rab10 and other relevant Rabs by immunofluorescence. Confocal microscopy analyses confirmed a significant increase of Rab10 on Rab5 domains upon stimulation with BDNF, which has been added to the revised figure 3 (panel c).

To address the question of Reviewer #2, we have also analysed the co-localisation between Rab10 and the main Rabs associated with recycling endosomes, Rab4 and Rab11. These experiments were performed by using the same approaches previously described in the manuscript. Axon segments from three independent experiments were analysed. The data suggest that around 25% of Rab10-positive compartments also contained these slow recycling endosome markers. As shown on the left, the amount of Rab10 co-localising with Rab4 and Rab11 exhibited no changes upon BDNF treatment. However, the amount of Rab11 in Rab10-positive membranes increased after 30 minutes of BDNF treatment, suggesting that the distribution of the minute amount of Rab11 present in the axon is regulated by neurotrophic signalling and eventually converges into the same pool of Rab10-positive organelles. This is consistent with a scenario in which, at steady state, Rab10 is found in diverse axonal compartments, and upon treatment with BDNF, it relocalises to early endosomes positive for Rab5 and other endosomal Rabs.

3. Reviewer #2 raised questions about the reasons for studying KIF5B and KIF13B. Together with Reviewer #3, they suggested to analyse the effects of kinesin down-regulation on the transport of TrkB. In this regard, it is important to mention that we have not aimed to study direct effects of these or other kinesins on the axonal transport of TrkB, but to propose a mechanism for the BDNF-dependent regulation of Rab10 direction bias, the dynamics of which is, in turn, crucial to fine-tune TrkB sorting from early endosomes. However, it is important to notice that these kinesins have many other roles and, therefore, the prediction is that knocking them down will cause disruption of axonal dynamics of multiple cargoes. For example, anterograde transport of TrkB has been shown to be regulated by kinesin 1 [Huang, S. H., Duan, S., Sun, T., Wang, J., Zhao, L., Geng, Z., Yan, J., Sun, H. J., & Chen, Z. Y. (2011). JIP3 mediates TrkB axonal anterograde transport and enhances BDNF signaling by directly bridging TrkB with kinesin-1. J Neurosci, 31, 10602-14. https://doi.org/10.1523/JNEUROSCI.0436-11.2011; Sun, T., Li, Y., Li, T., Ma, H., Guo, Y., Jiang, X., Hou, M., Huang, S., & Chen, Z. (2017). JIP1 and JIP3 cooperate to mediate TrkB anterograde axonal transport by activating kinesin-1. Cell Mol Life Sci, 74, 4027-44. https://doi.org/10.1007/s00018-017-2568-z; Zahavi, E. E., Hummel, J. J. A., Han, Y., Bar, C., Stucchi, R., Altelaar, M., & Hoogenraad, C. C. (2021). Combined kinesin-1 and kinesin-3 activity drives axonal trafficking of TrkB receptors in Rab6 carriers. Dev Cell, 56, 494-508 e497. https://doi.org/10.1016/j.devcel.2021.01.010]; hence, the effects of KIF5B knock down will be necessarily complex and not only affect Rab10 trafficking. Similarly, KIF13B is crucial for establishing and maintaining neuronal polarity; therefore, knocking it down would also disrupt axon integrity [Nakata, T., & Hirokawa, N. (2007). Neuronal polarity and the kinesin superfamily proteins. Sci STKE, 2007, pe6. https://doi.org/10.1126/stke.3722007pe6]. Previously, we presented the results of a pilot experiment, in which we attempted to use a KIF13B mutant lacking the motor domain, which yielded a significant decrease in neurite complexity, emphasising the need to develop new tools to further explore this point. Instead, we attempted this experiment for a different reason: we wondered which interactors of Rab10 mediate its differential direction bias in response to BDNF. Since their interaction is regulated by phosphorylation, we initially focussed on the adaptors JIP3 and JIP4; however, we found little co-localisation between JIP3, JIP4 and Rab10, as documented in the supplementary material. We then looked at KIF13B because it was shown to preferentially interact with GTP-bound Rab10, and it is highly expressed in the axon of hippocampal neurons [Yang, R., Bostick, Z., Garbouchian, A., Luisi, J., Banker, G., & Bentley, M. (2019). A novel strategy to visualize vesicle-bound kinesins reveals the diversity of kinesin-mediated transport. Traffic, 20, 851-66. https://doi.org/10.1111/tra.12692]. KIF5B was chosen as control, because it is also robustly expressed in the axon of hippocampal neurons but lacks a binding site for Rab10. It is relevant to mention that, in contrast to Reviewer #2’s views, stating that “the change in Kif13B and Kif5B intensity in response to BDNF is very moderate at best”, we don’t focus on the change of intensity of the KIFs in the axon, but on their recruitment to Rab10positive domains. While KIF13B increased in Rab10 compartments compared to the rest of the cytoplasm, KIF5B displayed the opposite behaviour (figure 6b-e). As now explained in the Discussion, it is possible that this is not the only determinant of Rab10 direction bias, but the contribution of KIF13B would be enough to explain the BDNF-dependent changes of direction, providing a lead for further studies on the detailed mechanism controlling Rab10 dynamics.

4. Related to the previous point, Reviewers #2 and #3 also expressed concerns about the immunoprecipitation experiment shown in figure 6g-h. They mentioned that “it appears that there is more HA-Rab10 precipitated in BDNF-treated samples, thus, it is unclear whether slight increase in GFP-Kif13B co-precipitation is due to that rather than increased interaction with Rab10”. Upon quantification, we are pleased to report that there is a two-fold increase in Kif13B (p = 0.0024). The reviewer is absolutely right at noticing that a little more HA-Rab10 immunoprecipitated in the BDNF-treated sample in figure 6g; however, figure 6h displays the ratio between GFP-KIF13B and HA-Rab10 signals, which normalises data for variations in the efficiency of the immunoprecipitation. To better clarify this point, we have now included the quantification of the non-normalised GFP-KIF13B (showing a two-fold increase with a p value = 0.0067), and the quantification of the HA-Rab10 showing no differences (p value = 0.9240).

The reviewer also questions the presence of a slight band of GFP-KIF13B in the sample that does not expressed HA-Rab10. Indeed, a small amount of the GFP fusion protein is nonspecifically bound to the magnetic beads, which is a common occurrence in this type of experiments, yet the low intensity of this band does not affect the overall interpretation of our results.

5. The aim of the immunoprecipitation experiments was to show (i) that KIF13B and Rab10 can interact as part of a molecular complex, and (ii) that the interaction was positively regulated by BDNF. Both aspects were indisputably proven by the results shown in figure 6g-h, complementing the spatial correlation reported in figure 6a-c. However, the Reviewer #2 asked whether we could “blot for endogenous Kif13B co-precipitation with HA-Rab10”.

We did several pulldown experiments from Neuro 2A cells transfected with HA-Rab10. Although we were able to detect endogenous KIF13B in the input, we failed to reveal KIF13B in the immunoprecipitate (see Author response image 4). A possible explanation for this negative result might derive from a different localisation and/or expression of KIF13B in Neuro 2A cells compare to primary neurons. Using a lentiviral inducible system to efficiently express mCherry-Rab10 in hippocampal neurons, we used magnetic beads to immunoprecipitate mCherry-Rab10, but we again failed to observe endogenous KIF13B associated to the beads under our experimental conditions. Whilst disappointing, this result is hardly surprising, since our total lysates are not enriched in axonal membranes, and we did not deploy any effective mean to stabilise the transient interaction between GTP-bound Rab10 and KIF13B. We believe that determining the content of Rab10-positive membrane compartments by using immunoisolation and proteomics will provide a complete and unbiased characterisation of this Rab10 compartment; however, we hope that the Reviewers agreed with us that this aim goes beyond the focus of this manuscript.

Author response image 4

6. Reviewer #2 also raised the potential confounding effects of the Rab10 DN, which they suggested to mitigate by repeating all recycling experiments in neurons treated with shRNA for Rab10 (“all recycling experiments need to be done in Rab10-KD cells”). Far from showing confounding effects, our results revealed no effect of Rab10 DN in recycling of TrkB in axons. However, expression of Rab10 DN effectively induced an increase on TrkB/Rab5 colocalisation in axons, which is consistent with the reduction on retrograde accumulation observed in Rab10-KD neurons. On the same issue, Reviewer #2 argues that “it is very puzzling that trapping TrkB in early endosomes did not affect recycling. Most plasma membrane receptors recycle by sequential transport from early endosomes to recycling endosomes. Consequently, I would expect that trapping TrkB in early endosomes would decrease its recycling” We apologise for the lack of clarity explaining our model. The key assumption in this point of the reviewer is that TrkB receptors located in the early endosome are homogeneously distributed and equally available to be destined either to local recycling or retrograde trafficking. However, the sorting endosome is rich in membrane subdomains and this heterogeneity has been hypothesised to be crucial for cargo selection and trafficking control [Redpath, G.M.I, Betzler, V.M, Rosatti, P. and Rossy, J., (2020). Membrane Heterogeneity Controls Cellular Endocyctic Trafficking. Front. CellDev. Biol., 8, 757. https://doi.org/10.3389/fcell.2020.00757]. If retrograde and recycling pools of TrkB were differentially regulated in early endosomes, it is expected that blocking the sorting to retrograde transport would increase colocalisation of TrkB and Rab5, with no increase on recycling. Several lines of evidence indicate the potential of Rab10 and its effectors to define membrane sub-domains important for cargo selection and regulation of membrane curvature, budding and formation of tubules. We have included and discussed this literature in the manuscript, as well as rephrased any ambiguity on the interpretation of these experiment.

7. Finally, Reviewer #1 also provided very useful comments about the nomenclature of Rab10 constructs, a couple of missing resources in the table and asked for clarification about a reference. We have amended the manuscript accordingly.

https://doi.org/10.7554/eLife.81532.sa2

Article and author information

Author details

  1. Oscar Marcelo Lazo

    1. Department of Neuromuscular Diseases and UCL Queen Square Motor Neuron Disease Centre, UCL Queen Square Institute of Neurology, University College London, London, United Kingdom
    2. UK Dementia Research Institute at UCL, London, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    oscar.lazo@ucl.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4542-482X
  2. Giampietro Schiavo

    1. Department of Neuromuscular Diseases and UCL Queen Square Motor Neuron Disease Centre, UCL Queen Square Institute of Neurology, University College London, London, United Kingdom
    2. UK Dementia Research Institute at UCL, London, United Kingdom
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Project administration, Writing – review and editing
    For correspondence
    giampietro.schiavo@ucl.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4319-8745

Funding

Medical Research Council (MR/T001976/1)

  • Oscar Marcelo Lazo

Wellcome Trust (107116/Z/15/Z)

  • Giampietro Schiavo

Wellcome Trust (223022/Z/21/Z)

  • Giampietro Schiavo

UK Dementia Research Institute Foundation (UKDRI-1005)

  • Giampietro Schiavo

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. For the purpose of Open Access, the authors have applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.

Acknowledgements

We thank Dr James N Sleigh (UCL Queen Square Institute of Neurology) for critical reading of the first manuscript. This work was supported by the MRC Project grant MR/T001976/1 [OML], the Wellcome Trust Senior Investigator Awards (107116/Z/15/Z and 223022/Z/21/Z) [GS] and the UK Dementia Research Institute Foundation award UKDRI-1005 [GS].

Ethics

Animal experimentation: The use of mice during this study is regulated by the Home Office Project Licence PCF5CF564 granted to Prof. Schiavo on 19.12.2018. The number of animals used and the protocols for their handling minimising stress and suffering, are approved and monitored by University College London Biological Services. No human samples or personal data will be collected during this study. Procedures and research practices are in agreement with the UCL Code of Conduct for Research and Singapore Statement on Research Integrity (2010).

Senior and Reviewing Editor

  1. Suzanne R Pfeffer, Stanford University, United States

Reviewer

  1. Mitsunori Fukuda, Tohoku University, Graduate School of Life Sciences, Japan

Version history

  1. Preprint posted: April 7, 2021 (view preprint)
  2. Received: July 1, 2022
  3. Accepted: February 15, 2023
  4. Version of Record published: March 10, 2023 (version 1)

Copyright

© 2023, Lazo and Schiavo

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,285
    Page views
  • 143
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Oscar Marcelo Lazo
  2. Giampietro Schiavo
(2023)
Rab10 regulates the sorting of internalised TrkB for retrograde axonal transport
eLife 12:e81532.
https://doi.org/10.7554/eLife.81532

Share this article

https://doi.org/10.7554/eLife.81532

Further reading

    1. Computational and Systems Biology
    2. Neuroscience
    Tony Zhang, Matthew Rosenberg ... Markus Meister
    Research Article

    An animal entering a new environment typically faces three challenges: explore the space for resources, memorize their locations, and navigate towards those targets as needed. Here we propose a neural algorithm that can solve all these problems and operates reliably in diverse and complex environments. At its core, the mechanism makes use of a behavioral module common to all motile animals, namely the ability to follow an odor to its source. We show how the brain can learn to generate internal “virtual odors” that guide the animal to any location of interest. This endotaxis algorithm can be implemented with a simple 3-layer neural circuit using only biologically realistic structures and learning rules. Several neural components of this scheme are found in brains from insects to humans. Nature may have evolved a general mechanism for search and navigation on the ancient backbone of chemotaxis.

    1. Neuroscience
    Frances Skinner
    Insight

    Automatic leveraging of information in a hippocampal neuron database to generate mathematical models should help foster interactions between experimental and computational neuroscientists.