Development of dendrite polarity in Drosophila neurons
© Hill et al.; licensee BioMed Central Ltd. 2012
Received: 23 January 2012
Accepted: 14 September 2012
Published: 30 October 2012
Skip to main content
© Hill et al.; licensee BioMed Central Ltd. 2012
Received: 23 January 2012
Accepted: 14 September 2012
Published: 30 October 2012
Drosophila neurons have dendrites that contain minus-end-out microtubules. This microtubule arrangement is different from that of cultured mammalian neurons, which have mixed polarity microtubules in dendrites.
To determine whether Drosophila and mammalian dendrites have a common microtubule organization during development, we analyzed microtubule polarity in Drosophila dendritic arborization neuron dendrites at different stages of outgrowth from the cell body in vivo. As dendrites initially extended, they contained mixed polarity microtubules, like mammalian neurons developing in culture. Over a period of several days this mixed microtubule array gradually matured to a minus-end-out array. To determine whether features characteristic of dendrites were localized before uniform polarity was attained, we analyzed dendritic markers as dendrites developed. In all cases the markers took on their characteristic distribution while dendrites had mixed polarity. An axonal marker was also quite well excluded from dendrites throughout development, although this was perhaps more efficient in mature neurons. To confirm that dendrite character could be acquired in Drosophila while microtubules were mixed, we genetically disrupted uniform dendritic microtubule organization. Dendritic markers also localized correctly in this case.
We conclude that developing Drosophila dendrites initially have mixed microtubule polarity. Over time they mature to uniform microtubule polarity. Dendrite identity is established before the mature microtubule arrangement is attained, during the period of mixed microtubule polarity.
Many neurons are specialized to receive information through dendrites and send signals through axons. In order to establish and maintain functionally distinct compartments, different proteins must accumulate in axons and dendrites. Most neuronal protein synthesis takes place in the cell body, and so proteins must be moved to their place of function after synthesis. This long-distance transport is primarily microtubule-based[2–4], and so the arrangement of microtubules has the potential to control directional trafficking and neuronal polarity.
In primary cultures of mammalian neurons, microtubules in axons and dendrites have different arrangements. In axons, all microtubules have their plus ends directed away from the cell body (plus-end-out), and thus only plus end-directed kinesins can carry cargo into axons[1, 5]. In contrast, dendritic microtubules have mixed polarity[6, 7], and so any motor protein could carry cargo into dendrites.
In addition to the fundamental observation that microtubules have different arrangements in axons and dendrites, there are several studies that indicate a strong link between microtubule organization and dendrite identity. In cultured hippocampal neurons four to five unspecified neurites initially extend from the soma; all of these processes have plus-end-out microtubules. One of the processes then begins a rapid growth phase, and this process becomes the axon. Later, the remaining processes initiate growth as dendrites. The time at which many dendritic features are acquired and dendrites start their major outgrowth coincides with the time at which minus-end-out microtubules are added to the initial population of plus-end-out microtubules. Further evidence for the link between minus-end-out microtubules and dendrite identity was provided by phenotypic analysis of neurons in which levels of MKLP1 were reduced. In these neurons the number of minus-end-out microtubules in dendrites went down after antisense RNA treatment to reduce MKLP1, and at the same time dendritic shape and contents, including ribosomes, were lost. Thus there seems to be a strong link between mixed dendrite microtubule polarity and dendrite identity in cultured mammalian neurons.
However, the dendrites in which microtubules have been studied most extensively in vivo do not have mixed polarity. Drosophila neurons have axons and dendrites with polarized structure and function similar to mammalian neurons. Like mammalian neurons, Drosophila neurons have axons with uniform plus-end-out microtubules. However, dendrites of Drosophila larval sensory neurons[11–13], motor neurons, and interneurons have close to uniform minus-end-out polarity. Motor neuron dendrites in C. elegans neurons share this minus-end-out organization. This discrepancy in microtubule organization between cultured mammalian neurons and Drosophila and C. elegans neurons raises a number of questions, including whether polarized trafficking in invertebrate and mammalian neurons is fundamentally different, and whether a uniform minus-end-out neurite is capable of outgrowth. To begin to answer these questions we have analyzed the development of microtubule polarity in Drosophila neurons in vivo. We find that, as in mammalian neurons, Drosophila dendrites have mixed polarity early in their outgrowth. In fact it takes several days after dendrites are specified to attain uniform minus-end-out polarity. We also find that, as in mammalian neurons, major features of dendrite identity are established when dendritic microtubules are mixed.
In Drosophila, EB1-GFP can be expressed in neurons or subsets of neurons using the Gal4-UAS system. Most Gal4 drivers we tried did not, however, drive sufficient levels of EB1-GFP to detect comets in neurons at early stages of dendrite development. We did, however, find that 1407-Gal4 could be used to express EB1-GFP at early stages of dendrite outgrowth in Drosophila embryos (Figure1, stage 1). For analysis of later stages of dendrite development we used the pan-neuronal elav-Gal4 driver (Figure1).
For imaging microtubule dynamics in early stages of dendrite outgrowth before embryos were motile, we used a very simple mounting and imaging strategy. Dechorionated embryos under a thin layer of halocarbon oil were imaged on open coverslips suspended on the stage of an upright confocal microscope. This technique allowed us to image dendrites of neurons expressing EB1-GFP until late in embryonic development when embryos start to move inside the vitelline membrane. To image late-stage embryos, corresponding to stage 5 of dendrite development, we developed a strategy to immobilize living embryos.
To reduce mobility of late-stage embryos, we crossed a temperature-sensitive paralytic mutation (Shi ts1 ) into a genetic background that contained elav-Gal4 and UAS-EB1-GFP transgenes. Embryos with two copies of Shi ts1 and one copy of each of the transgenes were mounted for imaging, and were then warmed on the microscope stage with an objective heater. This allowed us to image EB1-GFP dynamics in paralyzed embryos that had developed up until imaging with normal neuronal function.
Imaging young larvae also presented a challenge as they are much more fragile than the second and third instar larvae we previously analyzed. We therefore used the Shi ts1 line for very young larvae as well and mounted these between an air-permeable membrane and cover slip. We were thus able to perform live imaging of microtubule dynamics throughout dendrite development (Figure1).
The late stage in dendrite development at which uniform polarity was acquired suggested that dendrite identity was established while microtubule polarity was mixed. Ribosomes localize predominantly to the cell body and dendrites in mammalian neurons, and very few are present in axons under normal circumstances. As ribosomes can therefore be used as markers of dendrites, we wished to determine whether ribosomes localized to Drosophila dendrites before or after they acquired uniform microtubule polarity.
When we examined the pattern of L10-YFP fluorescence during dendrite development, we could observe spots at branch points as soon as they were formed (Figure3A and B). Initially L10-YFP was primarily at branch points along main trunks of dendrites that emerged from cell bodies. Two days after larval hatching spots of fluorescence were also seen at more distal branch points (Figure3 A and B). This increase in distal L10-YFP could either reflect increased concentrations of L10-YFP that make all ribosome populations easier to visualize, or it could reflect an outward transport of ribosomes as uniform polarity in dendrites is achieved. In either case, L10-YFP was capable of localizing to major dendrite branch points at the time when microtubule polarity was mixed. We thus conclude that ribosome targeting to dendrites occurs during mixed microtubule polarity, and does not require uniform minus-end-out polarity.
Apc2-GFP is localized to cell bodies, dendrites, and some proximal axons, but not distal axons, in Drosophila central neurons and dendritic arborization neurons. It is one of the most polarized markers we have found that distinguishes Drosophila dendrites from axons. Within dendrites it is very specifically localized to dendrite branch points where Apc2 plays a role in controlling microtubule polarity. Apc2-GFP localized strongly to dendrites from very early stages (Figure4B and C). It also localized to nascent branches as they were forming, from the stage when new branches resemble filopodia. As the branches developed, Apc2-GFP became more distinctly localized to the branch point itself. Thus localization of Apc2-GFP in dendrites, and association with branches, was a very early developmental event that preceded uniform microtubule polarity.
As ribosomes, mitochondria and Apc2-GFP took their characteristic positions in dendrites before uniform microtubule polarity was established, we hypothesized that mixed microtubule polarity was sufficient to determine dendrite identity. To test whether uniform microtubule polarity was required for localization of organelles or proteins within dendrites, we compared marker distribution in control neurons and neurons with reduced levels of Kap3. Kap3 is an accessory subunit of the kinesin-2 motor[22, 23]. When levels of Kap3 or either of the two motor subunits of kinesin-2 are reduced, dendrite microtubule polarity is non-uniform. In the main trunk of the class I dendritic arborization neuron ddaE microtubule polarity changes from less than 5% plus-end-out to greater than 30% plus-end-out when Kap3 is targeted by RNAi. Dendrite shape is unchanged under these conditions. We therefore compared mitochondria and Apc2-GFP distribution in control and Kap3 RNAi ddaE neurons.
Similar experiments were performed to determine whether Apc2-GFP localized differently in mature neurons with uniform or mixed polarity (Figure5B). In control RNAi neurons more than 90% of branch points along the main trunk of the ddaE comb dendrite contained Apc2-GFP punctae. Similar numbers of branch points were occupied when Kap3 levels were reduced by RNAi (Figure5C). Thus two different markers were able to occupy their normal positions in dendrites that cannot acquire uniform microtubule polarity. This result is consistent with the experiments in embryos that showed Apc2-GFP and mitochondria localized to dendrites early in their development, before uniform microtubule polarity was established.
To test the link between uniform microtubule polarity and exclusion of DCVs from dendrites, we tested whether the number of ANF-GFP punctae would increase in neurons in which Kap3 levels were reduced and dendrite microtubule polarity was mixed. In both Kap3 and control RNAi neurons in 3-day-old larvae, ANF-GFP punctae were frequently found in dendrites up to the first branch point. However, only a few punctae were typically found beyond this point in either genotype (Figure6B), and no significant difference between genotypes was observed. Note that the number of punctae in larvae in Figure6B is higher than in 6A, probably because a stronger Gal4 driver was used in the RNAi experiment. We conclude that DCVs may be more slightly efficiently excluded from dendrites in mature neurons, but this does not seem to be due to the more uniform microtubule polarity.
This study is the first to examine how microtubule polarity develops in dendrites in vivo. In Drosophila dendritic arborization neurons, microtubule polarity was mixed during the major period of dendrite outgrowth during embryogenesis. Towards the end of embryogenesis and the beginning of larval life, the number of minus-end-out microtubules gradually exceeded that of plus-end-out microtubules, until 2 days into the larval period more than 90% of microtubules were minus-end-out. In mammalian cultured neurons, microtubules also have mixed polarity in dendrites as they grow out from the cell body. Thus microtubule polarity is similar in fly and mammalian dendrites during their development. This similarity suggests that conserved mechanisms could establish the basic layout of neuronal microtubules from flies to mammals.
To determine whether the change from mixed to uniform polarity in Drosophila dendrites affects localization of organelles or proteins, we analyzed a variety of markers during dendrite development. All were able to acquire their characteristic localization patterns in developing dendrites with mixed polarity. Again, this suggests similarity with mammalian neurons in culture, which can develop polarized dendrites with mixed microtubule orientation. Moreover specific cargoes can be delivered directly to mixed orientation dendrites in culture by polarized transport. Thus mixed microtubule polarity is sufficient to allow targeting of dendritic cargoes in both systems.
For analysis of axonal targeting we used ANF-GFP as it labels discrete neuropeptide vesicles that can be targeted relatively specifically to axons in da neurons (see Figure6B). During embryonic dendrite development some of these vesicles could be seen in dendrite arbors, and when the marker was expressed at relatively low levels with elav-Gal4 (Figure6A) very few were seen in larval dendrites. Thus DCVs may be slightly more efficiently excluded from dendrites in more mature neurons. To probe whether efficient exclusion of DCVs from dendrites might be related to acquisition of uniform microtubule polarity we compared the number of punctae in dendrites with mixed and uniform polarity microtubules. In both cases occasional punctae were seen in dendrites, but there was not a difference between the two sets. The lack of DCV escape into dendrites with mixed polarity is consistent with a recent study of DCV targeting to C. elegans dendrites, which also found that factors other than microtubule polarity were likely to control polarized targeting of these vesicles. It is possible, however, that microtubule polarity may contribute to efficient exclusion of other types of axonal cargoes from dendrites.
Two major models of polarized transport have been proposed to account for differential accumulation of proteins and organelles in axons and dendrites. One relies on plus end-directed kinesins to transport cargo into both axons and dendrites. In this model either different sets of motors, or different motor-cargo combinations are directed specifically to axons or dendrites[2, 3]. This type of kinesin-based model could account for trafficking in developing dendrites of flies and mammals, but becomes difficult to rationalize in mature Drosophila dendrites. The alternate model is that the major minus end-directed motor, dynein, plays a key role in transport from the cell body into dendrites. Indeed there is phenotypic evidence for a role for dynein in anterograde dendrite transport in Drosophila[12, 13], and dynein can also transport cargo into mammalian dendrites. It will be interesting to determine the relative contributions of kinesins and dynein to dendrite trafficking in future studies.
The finding that dendrite identity is established in flies and mammals when microtubules have mixed polarity raises the question of why Drosophila dendrites switch to minus-end-out polarity. The only current answers to this question are speculative. Having opposite polarity microtubules in axons and dendrites should increase efficiency of polarized transport and reduce targeting errors. So far we have not been able to detect differences in Drosophila dendrites with mixed and uniform polarity, but these may exist.
It is not yet clear whether mammalian dendrites ever have uniform minus-end-out polarity. One intriguing study has raised the possibility that very mature mammalian neurons in vivo may have regions of dendrites near the cell body that have uniform microtubule polarity. However, the second harmonic generation microscopy technique used could not distinguish between uniform polarity plus-end-out or minus-end-out microtubules, so it remains to be resolved whether mammalian dendrites can have minus-end-out polarity. It is clear, however, that developing Drosophila and mammalian neurons share a similar organization of microtubules, and that in both cases roughly equal numbers of plus and minus-end-out microtubules are present as dendrites grow.
We have analyzed microtubule polarity and marker distribution during development of dendrites in Drosophila neurons in vivo. We conclude that as they grow out from the cell body, dendrites initially have mixed microtubule polarity. This organization gradually changes to minus-end-out over the next two days. The dendritic and axonal markers examined were able to localize correctly before the final microtubule polarity was achieved.
Flies were kept at room temperature in standard media. Most transgenic lines have been previously described (see). Many of the lines were received from the Bloomington Drosophila Stock Center. RNAi lines were from the Vienna Drosophila RNAi Center (VDRC). A UAS-ANF-GFP insertion on the third chromosome was generously provided by David Deitcher.
The only line not previously described was the ribosomal marker UAS-EGFP-L10a. As a basis for this line, a plasmid containing an inframe fusion of EGFP and the mouse L10a ribosomal protein was obtained from Joshua Ainsley (Tufts University School of Medicine). The EGFP-L10a coding segment was subcloned into the pUAST transformation vector, and transgenic strains were generated using standard methods by Genetic Services, Inc.
In order to visualize microtubule polarity several different genetic backgrounds were used. For early stages of dendrite outgrowth 1407 Gal4 was used to drive expression of UAS-EB1-GFP. For older embryos and larvae, progeny from a cross between males of genotype shits1/Y; elav- Gal4, UAS-EB1-GFP and females of genotype shits1/shits1 were used. In order to visualize ribosomes in the da neurons, males with UAS-mcD8-RFP on chromosome II were crossed with females of the line UAS-L10-YFP, elavGal4/TM6 and non-tubby progeny were imaged. To visualize mitochondria, larvae from the line UAS-dicer, UAS-mcD8-RFP/CyO; elav-Gal4, UAS-mitoGFP/TM6 were used. Lastly, to visualize Apc2-GFP, males with an elav-Gal4 transgene were crossed to females from the line UAS-mCD8-RFP, UAS-dicer2; 221-Gal4, UAS-Apc2-GFP. This Apc2-GFP line was also used as a tester line for RNAi experiments, and progeny from this line crossed to a control RNAi targeting Rtnl2 (VDRC 33320) or an experimental RNAi targeting Kap3 (VDRC 45400) were analyzed. For analysis of mitochondria localization in conjunction with RNAi the UAS-dicer, UAS-mcD8-RFP/CyO; elav-Gal4, UAS-mitoGFP/TM6 line was crossed to the same pair of RNAi lines and progeny were analyzed. For analysis of ANF-GFP localization a tester line of genotype UAS-dicer, UAS-mcD8-RFP/CyO; 221-Gal4, UAS-ANF-GFP/TM6 was generated. It was crossed to RNAi lines targeting either Rtnl2 as a control or Kap3. For RNAi analysis embryos were collected overnight on either a yeasted apple cap or a cap containing standard Drosophila media, then aged for 3 days at 25°C.
Embryos were collected overnight on apple caps with yeast paste. The chorion was dissolved with 50% bleach solution for 2 min. After thorough rinsing with water, embryos were collected in heptane and transferred to a coverslip. Immediately after the heptane evaporated, embryos were covered in a thin layer of halocarbon oil 27 (Sigma). The coverslip was suspended with embryos open underneath it on the microscope stage. For EB1-GFP imaging in late stage embryos, an objective heater (BiOptechs) set to 36°C was used to warm the embryos as they were being imaged.
Young larvae were collected from apple caps within 2 h after hatching. They were then placed on air-permeable membrane supported on a metal slide and a coverslip was placed on top. An objective heater was used as for embryos.
For ‘2 day’ larvae imaging, embryos were collected overnight on a food cap. The food cap was then removed to a petri dish and larvae were aged for two days at 25°C before imaging. Larvae were then rinsed in Schneider’s media and transferred onto dried agarose pad on microscope slides. Larvae were then carefully rotated to dorsal side up and covered with glass coverslips taped onto the slide.
An Olympus FV1000 confocal microscope was used for all imaging. Timeseries were acquired using a 60× 1.4 NA objective.
For each embryo or larva, neurons in only one hemisegment were analyzed. All analysis was performed in ImageJ. For EB1-GFP dynamics, comets were only counted if seen in three consecutive frames. Each dendrite branch was watched independently for comets. Dendrites were categorized as those emerging directly from the cell body and those branching from another dendrite. For mito-GFP and Apc2-GFP, spots on branch points were counted as well as total number of branch points in the dendrites. These values were used to determine percentage of branch points occupied. For mito-GFP spot length was also recorded for every spot visible in the dendrites. The ImageJ measure function was used to determine length.
Photobleaching of L10-YFP was performed with the 488 nm laser on an Olympus FV1000 confocal microscope. Both L10-YFP and mCD8-RFP were bleached simultaneously with this laser at 100% power. Images of both channels were acquired every 3.3 s after bleaching. Analysis of recovery was performed in ImageJ. Small regions of interest within the bleach area, and also a background area without a fluorescent cell, were defined and the average intensity of the area was measured in the time series. The background value was subtracted from the value in the bleach area. The value of the area of interest was set to 100 before bleaching and 0 immediately after bleaching. Remaining values were normalized to this scale. Averages of these normalized values are plotted in the graph. Error bars represent the standard deviation at each timepoint.
Progeny of the analysis lines for Apc2-GFP and mitoGFP (see section on fly stocks and genetics) crossed to RNA hairpin lines were collected overnight on food caps. Caps were then transferred to a petri dish and aged for 3 days at 25°C. Larvae were mounted for live imaging on slides with dried agarose pads, and coverslips were taped on top of the larvae to restrain them. Images were acquired with an Olympus FV1000 confocal microscope. The localization of Apc2-GFP and mitoGFP was analyzed in the ddaE neuron. One ddaE neuron per animal was imaged and analyzed. Branch point occupancy was calculated along the primary branch of the dorsal comb-like dendrite. Mitochondria per unit length was also calculated in this part of the neuron. Measurements were performed using ImageJ.
For developmental analysis ANF-GFP was expressed together with mCD8-RFP using the pan-neuronal elav-Gal4 driver. Images were acquired as for other markers. For RNAi analysis the tester line UAS-mCD8-RFP, UAS-dicer2; UAS-ANF-GFP, 221-Gal4 was crossed to either Rtnl2 (VDRC 33320) or Kap3 (VDRC 45400) RNAi lines. Larvae were aged as for other RNAi experiments, and GFP punctae beyond the first branch point of the ddaE comb dendrite were counted in images acquired on a Zeiss LSM510 microscope.
Adenomatous polyposis coli
Fluorescence recovery after photobleaching.
We are very grateful to all members of the Rolls lab for technical help and ideas. The Bloomington Drosophila Stock Center and Vienna Drosophila RNAi Center are great resources, and we are indebted to them for many useful fly stocks. We are also grateful to David Deitcher for the ANF-GFP flies. This work was supported by an American Heart Association Scientist Development Grant, by March of Dimes Grant-in-Aid 1-FY10-359, and by R01 GM085115. MMR is a Pew Scholar in the Biomedical Sciences. FRJ and YH were supported by NIH R01 NS065900 and a NARSAD award from the Brain Research Foundation.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.