The GARP complex prevents sterol accumulation at the trans-Golgi network during dendrite remodeling

  1. Caitlin E. O’Brien
  2. Susan H. Younger
  3. Lily Yeh Jan
  4. Yuh Nung Jan

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Membrane trafficking is essential for sculpting neuronal morphology. The GARP and EARP complexes are conserved tethers that regulate vesicle trafficking in the secretory and endolysosomal pathways, respectively. Both complexes contain the Vps51, Vps52, and Vps53 proteins, and a complex-specific protein: Vps54 in GARP and Vps50 in EARP. In Drosophila, we find that both complexes are required for dendrite morphogenesis during developmental remodeling of multidendritic class IV da (c4da) neurons. Having found that sterol accumulates at the trans-Golgi network (TGN) in Vps54KO/KO neurons, we investigated genes that regulate sterols and related lipids at the TGN. Overexpression of oxysterol binding protein (Osbp) or knockdown of the PI4K four wheel drive (fwd) exacerbates the Vps54KO/KO phenotype, whereas eliminating one allele of Osbp rescues it, suggesting that excess sterol accumulation at the TGN is, in part, responsible for inhibiting dendrite regrowth. These findings distinguish the GARP and EARP complexes in neurodevelopment and implicate vesicle trafficking and lipid transfer pathways in dendrite morphogenesis.

Article activity feed

  1. Version published to 10.1083/jcb.202112108
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    1. General Statements

    We thank the reviewers for their helpful comments. We believe that we will be able to address all of their concerns and suggestions. We have highlighted our responses in the revision plan and the changes we have already made to the manuscript in blue text. For figures where we have added data or analyses at the request of reviewers, we have highlighted the corresponding text in the figure legends.

    2. Description of the planned revisions

    Reviewer #1

    2- In Figure 4, the two mutations appear to have statistically differential effects on Rab5 and Rab7 puncta even though the data mean and distribution seem very similar. Interestingly, in each case the non-significant effect is associated with a smaller sample size. Given that the overall sample sizes used are rather small for such highly variable data, this could easily cause a statistical anomaly due to sampling bias. The sample size should be made uniform across all genotypes and should ideally be at least doubled.

    We will repeat this staining to increase the n to at least double this number, and adjust our conclusions if need be, in the revised manuscript..

    3- Perhaps the most important issue related to Figure 6 where the authors find that there is no sterol accumulation at 96h APF in the Vps50 mutant. However, even that the dendritic phenotype is slower to appear in this mutant compared to the Vps54, are the authors sure that the accumulation is not just slower? This should be examined using the same temporal sequence used for Vps54 shown in Future 6 C. In addition, the fact that sterol accumulation returns to normal in the Vps54 mutant at 1 day, supports the notion of a delay phenotype (see point 1 above).

    These issues should be experimentally addressed to see if the data fully support the initial conclusions, or if the conclusions should be modified to suggest differential contribution of the two complexes to the process being studied and to a developmental delay phenotype.

    We had included the filipin staining for Vps50KO/KO at 1 day in Figure S4 A (which did not show a significant difference from control). We did not collect data for this genotype at 72hrs APF because the dendritic length phenotype didn’t appear until later, and so we did not include *Vps50KO/KO *in the full time-course in Fig 6 C. We will collect additional data so that we can include Vps50KO/KO at all timepoints in this figure in the revised manuscript.

    . Reviewer #2

    It is stated that loss of VPS50 and VPS54 only causes dendrite morphogenesis defects. However, the corresponding supplemental figure S2c (which is not referenced in the text), is not suited to address this question. Axonal arborization, in particular terminal arbors, are not visible in samples where multiple/all c4da axons are labeled simultaneously (Fig. S2c). Analogous to the dendrite analysis of c4da neurons single cell resolution is essential to examine this in a meaningful way. Likely, however, c4da neurons may not be a good choice to address this question.

    We should be able to get single cell resolution of the c4da axon terminals using MARCM. We already have two of the knockout lines recombined with FRTs (Vps53 and Vps54) for this analysis and we will make the third recombinant line so that we can use MARCM for all three lines to examine single-cell axon morphology, as suggested.

    Overall, I am concerned whether the data shown here can be generalized. The cd4a neurons are rather extreme cell types due to their very large dendritic compartment. It seems quite possible that many other neurons may not have a comparable sensitivity to the supply of lipids/sterols. This type of question can only be addressed if other types of neurons/dendrites are examined. Are class 2 or class 3 da neurons showing any defects in VPS mutants?

    Given that we see the phenotype emerge during the pupal stage, we want to analyze neurons that persist from the larval to adult stages. However, not all of the dendritic arborization neurons survive into adulthood- class I and II persist, while class III die during metamorphosis (Shimono et al., 2009). As we do not have adequate tools to for studying the class II neurons, we will examine dendrite morphology of the class I neurons in larvae and adults in our knockout lines. We would be happy to look at class III neurons at the reviewers request, but our analysis will necessarily be limited to the larval stage.

    Reviewer #3

    • Some of the experiments include multiple genotypes and so it would be important to show all in all figures. For example, figure 4B,D show four groups but figure 4F, presumably from the same set of animals, shows only three. Addition of the rescue genotype to 4F is particularly important here so should be shown. The same concern is valid for figure 5, where puncta number and area must be available.

    The data from Fig. 4 F (using a genetically encoded marker for lysosomes, UAS-spin-RFP) are not from the same samples as Fig. 4 B and D (staining). We did not include the rescue for Fig 4. F because the lysosome marker, the rescue transgenes and the neuronal membrane marker are all on the third chromosome. We will build additional fly stocks so that we can include the rescue in experiments looking at lysosome morphology.

    • This concern is amplified by the images in figure 6 of the filipin staining, that are more obviously perinuclear. However, the two sets of images in 6A and 6D, where co-staining with Golgin245 is shown, look very different. Improved images are required and it may be helpful to use supplementary information to show additional examples of the staining.

    The images in Fig. 6 A are maximum projections of z stacks while Fig. 6 D shows single confocal planes, making it easier to see the perinuclear Golgi ring. Because other reviewers wanted some additional experiments related to Fig. 6 that we plan to incorporate into this figure in the revised manuscript, we will address this comment in a future revision and include additional images in the supplement.

    • For the lipid regulation experiments in figure 7, please use an orthogonal approach to show that the Osbp and fwd RNAi had the expected effects on lipid accumulation.

    In addition to sterol, Osbp and fwd both affect levels of PI4P at the Golgi. We have obtained a transgenic PI4P sensor that we can use to show the effect of these manipulations on this lipid as well.

    3. Description of the revisions that have already been incorporated in the transferred manuscript

    Reviewer #1 While the data presented clearly support a role for GARP in regulating sterol levels to support dendritic growth, they do not inter current for suffice to exclude a role for EARP as important analyses to allow such a clear cut conclusion are either insufficient or missing. If the authors wish to maintain this claim - as suggested by the title of the manuscript - further analyses are essential.

    We don’t mean to argue the EARP complex doesn’t contribute to dendrite development at all – we do show it contributes to development in Fig 3, and as we discuss in the text.. We want to argue that the GARP and EARP complexes contribute to dendrite development by distinct mechanisms. Losing the GARP complex inhibits dendrite development by means of sterol accumulation at the TGN, which is what we are trying to highlight with our title. The reduced dendrite growth that we observe in EARP deficient neurons must occur by some other as yet unknown means. We apologize for the confusion and have reworded the title to read “Sterol accumulates at the trans-Golgi in GARP complex deficient neurons during dendrite remodeling.”

    1- Figure 3E shows that whereas both Vps50 and Vps54 mutations reduce dendritic complexity, the Vps54 phenotype appears earlier (96h APF). Furthermore, at 7 days dendrites appear to grow again but at a slower rate than controls. This begs the question of whether these mutations are causing a delay rather than a block in the regrowth after pruning and whether the growth will eventually be normal a few days later or whether it will stop at some point.

    We have included data for an additional adult timepoint (21 days) in the new Fig. 3 E. We also included graphs in which we show the statistics for each genotype over time (new Fig. S2 D-F), and discuss this analysis in the text (lines 186-195). We have also included a table of the p-values for each comparison in the Supplemental Materials (Table S2). From this analysis, we conclude that there is not a developmental delay in the knockouts, but rather a decrease in growth during the 72-96hrs APF and 1-7 day windows when the control neurons grow. We are unable to draw conclusions about the *rate *of growth as we analyzed neurons from different samples at each developmental timepoint, and not the same neurons over time.

    Reviewer #2

    It would be important to know, whether the dendrite morphogenesis defect is indeed a developmental patterning defect or rather a "scaling" defect due to the fact that da neurons increase their size (but not necessarily their projection pattern) during larval maturation.

    We have analyzed the larval data for coverage index – neuron area/hemisegment (receptive field) area as defined in (Parrish et al., 2009) to determine if there is a scaling defect at this stage in development. We do not observe a defect in scaling (Fig. S2 C) and discussed in lines 175-182.

    Reviewer #3

    • The statistical analyses generally look appropriate but it would be critical to clarify what N means in every case. For example in figure 2 the authors state n=8 without clarifying if this is n=8 animals or n=8 neurons. N should always be the number of animals, but then the n of independent cells counted should also be indicated. Typically, one would either pre-average per genotype or use a mixed model that includes N of animals and n of cells (or similar).

    For experiments analyzing dendrite morphology, n represents the number of neurons, as we have clarified in our figure legends. As per another reviewer’s request, we will increase the n for the organelle and filipin staining in our planned revision and specify fly and cell number at that time.

    • Please add details of how experiments were blinded to genotype

    The researcher was blinded to genotype during analysis. We have included that detail in our Methods section (line 566).

    • Some of the experiments include multiple genotypes and so it would be important to show all in all figures. For example, figure 4B,D show four groups but figure 4F, presumably from the same set of animals, shows only three. Addition of the rescue genotype to 4F is particularly important here so should be shown. The same concern is valid for figure 5, where puncta number and area must be available.

    We address the first portion of this comment in section 2, for additional experiments involving generating new fly lines. We have included data on puncta area, and mean fluorescence intensity for Rab5 and Rab7 in the supplement (Fig S3). We had already included the data on puncta number and area in Fig 5, but we have added the data on mean fluorescence intensity as well.

    • Related to figure 5, please provide validation of the staining of the TGN. Typically, one would expect trans Golgi to be close to the nucleus with at least some extended stacks. A Golgin245 knockout would be ideal.

    The Golgi in most Drosophila cells is typically found as discrete puncta dispersed throughout the cytosol like what we see in the Golgin245 staining, as opposed to the ribbon “stack of pancake” morphology typically seen near the nucleus in mammalian cells. For reference, please see Figure 6D in (Ye et al., 2007), Figures 2,4,5 in (Rosa-Ferreira et al., 2015), and observations reviewed in (Kondylis and Rabouille, 2009).

    The Golgin245 antibody was well characterized in the paper first describing it (Riedel et al., 2016) (colocalization with other Golgi markers, decreased staining with Golgin245 RNAi), but we would be happy to repeat this validation in the c4da neurons at the reviewer’s request. There do not appear to be Golgin245 mutant or KO lines available, so we would also use the Golgin245 RNAi.

    • For figures 6F, G please show examples of staining for late endosomes and lysosome with appropriate validation.

    Because several of our planned revisions relate to Fig. 6, we will include images for Fig. 6 F and G when we remake this figure to incorporate those planned revisions. To clarify, we used the same reagents to mark late endosomes and lysosomes in both Fig. 4 and Fig. 6. Like the Golgin245 antibody, the Rab7 antibody was developed by the Munro lab and characterized in (Riedel et al., 2016) (partial colocalization with the endosomal marker Hrs and with the lysosomal marker Arl8). Spinster (aka benchwarmer) is a known lysosomal transmembrane protein that colocalizes with Lamp1 (Dermaut et al., 2005; Rong et al., 2011). The fluorescently tagged spin transgenes were developed by the Bellen lab and have been frequently used to mark lysosomes. We would be happy to carry out additional validation experiments at the reviewer’s specification.

    • The title of figure 2 is inaccurate, at least if I understand the experiment, as it does not show neuron-specific knockout but instead whole body knockout with neuron rescue. Please rephrase.

    Because of the lethality of whole body Vps53KO/KO in adult flies, we analyze MARCM clonal neurons that are Vps53KO/KO in flies that are otherwise heterozygous (Vps53KO/+). To clarify this experiment, we have changed the title of Fig. 2 from “Neuron-specific knockout of Vps53 results in smaller dendritic arbors” to “Vps53KO/KO MARCM clonal neurons have smaller dendritic arbors”.

    • Figure 8 needs examples of the TGN and late endosome morphology.

    We have included these images in Figure

    The order appears different in Fig. 4 B & D because we only included the rescue for the KO that shows a phenotype for each staining. The genotypes included in Fig. 4 B are: +/+, Vps50KO/KO, Vps50KO/KO + rescue, and Vps54KO/KO. The genotypes included in Fig. 4 D are +/+, Vps50KO/KO, Vps54KO/KO, Vps54KO/KO + rescue. We have changed the shading of the bars corresponding to these rescue genotypes throughout the manuscript to make it easier to distinguish the two rescue conditions.

    4. Description of analyses that authors prefer not to carry out

    References Cited

    Dermaut, B., K.K. Norga, A. Kania, P. Verstreken, H. Pan, Y. Zhou, P. Callaerts, and H.J. Bellen. 2005. Aberrant lysosomal carbohydrate storage accompanies endocytic defects and neurodegeneration in Drosophila benchwarmer. Journal of Cell Biology. 170:127–139. doi:10.1083/jcb.200412001.

    Kondylis, V., and C. Rabouille. 2009. The Golgi apparatus: Lessons from Drosophila. FEBS Letters. 583:3827–3838. doi:10.1016/j.febslet.2009.09.048.

    Parrish, J.Z., P. Xu, C.C. Kim, L.Y. Jan, and Y.N. Jan. 2009. The microRNA bantam Functions in Epithelial Cells to Regulate Scaling Growth of Dendrite Arbors in Drosophila Sensory Neurons. Neuron. 63:788–802. doi:10.1016/j.neuron.2009.08.006.

    Riedel, F., A.K. Gillingham, C. Rosa-Ferreira, A. Galindo, and S. Munro. 2016. An antibody toolkit for the study of membrane traffic in Drosophila melanogaster. Biology Open. 5:987–992. doi:10.1242/bio.018937.

    Rong, Y., C.K. McPhee, S. Deng, L. Huang, L. Chen, M. Liu, K. Tracy, E.H. Baehrecke, L. Yu, and M.J. Lenardo. 2011. Spinster is required for autophagic lysosome reformation and mTOR reactivation following starvation. Proceedings of the National Academy of Sciences. 108:7826–7831. doi:10.1073/pnas.1013800108.

    Rosa-Ferreira, C., C. Christis, I.L. Torres, and S. Munro. 2015. The small G protein Arl5 contributes to endosome-to-Golgi traffic by aiding the recruitment of the GARP complex to the Golgi. Biology Open. 4:474–481. doi:10.1242/bio.201410975.

    Shimono, K., A. Fujimoto, T. Tsuyama, M. Yamamoto-Kochi, M. Sato, Y. Hattori, K. Sugimura, T. Usui, K. Kimura, and T. Uemura. 2009. Multidendritic sensory neurons in the adult Drosophila abdomen: origins, dendritic morphology, and segment- and age-dependent programmed cell death. Neural Dev. 4:37. doi:10.1186/1749-8104-4-37.

    Ye, B., Y. Zhang, W. Song, S.H. Younger, L.Y. Jan, and Y.N. Jan. 2007. Growing Dendrites and Axons Differ in Their Reliance on the Secretory Pathway. Cell. 130:717–729. doi:10.1016/j.cell.2007.06.032.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    O'Brien et al report how deficiency in GARP specific protein VPS54 or the EARP specific protein VPS50 affects the developmental dendritic remodeling of multidendritic class IV da (c4da) neurons in Drosophila. The main findings are that while both complexes play a role in dendritic remodeling, VPS54 deficiency leads to lipid accumulation in the trans-Golgi network (TGN). Manipulating sterols at the TGN affects dendritic remodeling suggesting that lipid accumulation is responsible for control of neuron morphology in this model. Overall, the data is interesting and the authors develop the experiments enough to be convincing on their major claims. However, a few aspects need clarification and perhaps revisiting conclusions.

    Major comments

    • The statistical analyses generally look appropriate but it would be critical to clarify what N means in every case. For example in figure 2 the authors state n=8 without clarifying if this is n=8 animals or n=8 neurons. N should always be the number of animals, but then the n of independent cells counted should also be indicated. Typically, one would either pre-average per genotype or use a mixed model that includes N of animals and n of cells (or similar).
    • Please add details of how experiments were blinded to genotype
    • Some of the experiments include multiple genotypes and so it would be important to show all in all figures. For example, figure 4B,D show four groups but figure 4F, presumably from the same set of animals, shows only three. Addition of the rescue genotype to 4F is particularly important here so should be shown. The same concern is valid for figure 5, where puncta number and area must be available.
    • Related to figure 5, please provide validation of the staining of the TGN. Typically, one would expect trans Golgi to be close to the nucleus with at least some extended stacks. A Golgin245 knockout would be ideal.
    • This concern is amplified by the images in figure 6 of the filipin staining, that are more obviously perinuclear. However, the two sets of images in 6A and 6D, where co-staining with Golgin245 is shown, look very different. Improved images are required and it may be helpful to use supplementary information to show additional examples of the staining.
    • For figures 6F, G please show examples of staining for late endosomes and lysosome with appropriate validation.
    • For the lipid regulation experiments in figure 7, please use an orthogonal approach to show that the Osbp and fwd RNAi had the expected effects on lipid accumulation.
    • Figure 8 needs examples of the TGN and late endosome morphology.

    Minor comments

    • The title of figure 2 is inaccurate, at least if I understand the experiment, as it does not show neuron-specific knockout but instead whole body knockout with neuron rescue. Please rephrase.
    • For ease of reading, it would be helpful to show genotypes in the same order in all figures (see 4B, 4D)

    Significance

    The advance here is to nominate lipid accumulation at the trans Golgi network (TGN) is sufficient to affect dendritic remodeling during development. Although the work is performed in a model system, it may have relevance to human neurodevelopmental disorders caused by mutations in the orthologous genes. The work will be of highest relevance to developmental neurobiologists, particularly those working on GARP or EARP mutations and those who use Drosophila as an appropriate model for neurodevelopment.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    This manuscript presents a solid genetic analysis of components of the GARP and EARP complexes. The analysis is focused on a specialized type of sensory neurons i.e. class IV da neurons in Drosophila larvae. The authors show that loss of multiple components (VPS50-54) disrupt dendrite morphogenesis in c4da neurons in distinct ways. Additional genetic interaction studies further support the notion of functional differences of GARP and EARP in vivo.

    Overall this is a solid study and with one exception (see below) I have little concern regarding the presented experiments. I do, however, find the exclusive focus on a highly specialized cell type c4da somewhat problematic.

    Concerns:

    Experimental concern: It is stated that loss of VPS50 and VPS54 only causes dendrite morphogenesis defects. However, the corresponding supplemental figure S2c ( which is not referenced in the text), is not suited to address this question. Axonal arborization, in particular terminal arbors, are not visible in samples where multiple/all c4da axons are labeled simultaneously (Fig. S2c). Analogous to the dendrite analysis of c4da neurons single cell resolution is essential to examine this in a meaningful way. Likely, however, c4da neurons may not be a good choice to address this question.

    It would be important to know, whether the dendrite morphogenesis defect is indeed a developmental patterning defect or rather a "scaling" defect due to the fact that da neurons increase their size (but not necessarily their projection pattern) during larval maturation.

    Overall, I am concerned whether the data shown here can be generalized. The cd4a neurons are rather extreme cell types due to their very large dendritic compartment. It seems quite possible that many other neurons may not have a comparable sensitivity to the supply of lipids/sterols. This type of question can only be addressed if other types of neurons/dendrites are examined. Are class 2 or class 3 da neurons showing any defects in VPS mutants?

    Significance

    At this point i am not convinced that the findings can be generalized. The c4da neuron is really an extreme cell type with a massive disproportionate increase in membrane extensions. This is rather unusual and other neuron types should be tested.

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary:

    O'Brien and colleagues use Drosophila dendrite development to dissect the roles of the GARP and EARP vesicular trafficking complexes in the development of neuronal morphology. By making complex-specific KOs they investigate the role of each complex in the growth, pruning and re-growth of sensory dendrites and conclude that the GARP, but not EARP, complex is required for proper dendrite development by limiting sterol accumulation in the neuronal TGN.

    Major comments:

    While the data presented clearly support a role for GARP in regulating sterol levels to support dendritic growth, they do not inter current for suffice to exclude a role for EARP as important analyses to allow such a clear cut conclusion are either insufficient or missing. If the authors wish to maintain this claim - as suggested by the title of the manuscript - further analyses are essential.

    1- Figure 3E shows that whereas both Vps50 and Vps54 mutations reduce dendritic complexity, the Vps54 phenotype appears earlier (96h APF). Furthermore, at 7 days dendrites appear to grow again but at a slower rate than controls. This begs the question of whether these mutations are causing a delay rather than a block in the regrowth after pruning and whether the growth will eventually be normal a few days later or whether it will stop at some point.

    2- In Figure 4, the two mutations appear to have statistically differential effects on Rab5 and Rab7 puncta even though the data mean and distribution seem very similar. Interestingly, in each case the non-significant effect is associated with a smaller sample size. Given that the overall sample sizes used are rather small for such highly variable data, this could easily cause a statistical anomaly due to sampling bias. The sample size should be made uniform across all genotypes and should ideally be at least doubled.

    3- Perhaps the most important issue related to Figure 6 where the authors find that there is no sterol accumulation at 96h APF in the Vps50 mutant. However, even that the dendritic phenotype is slower to appear in this mutant compared to the Vps54, are the authors sure that the accumulation is not just slower? This should be examined using the same temporal sequence used for Vps54 shown in Future 6C. In addition, the fact that sterol accumulation returns to normal in the Vps54 mutant at 1 day, supports the notion of a delay phenotype (see point 1 above).

    These issues should be experimentally addressed to see if the data fully support the initial conclusions, or if the conclusions should be modified to suggest differential contribution of the two complexes to the process being studied and to a developmental delay phenotype.

    Significance

    The study advances our understanding of the role of regulation of lipid storage in sculpting neuronal morphology during development.

  6. Version published to 10.1101/2021.09.29.462255 on bioRxiv