Axon regeneration helps maintain lifelong function of neurons in many animals. Depending on the site of injury, new axons can grow either from the axon stump (after distal injury) or from the tip of a dendrite (after proximal injury). However, some neuron types do not have dendrites to be converted to a regenerating axon after proximal injury. For example, many sensory neurons receive information from a specialized sensory cilium rather than a branched dendrite arbor. We hypothesized that the lack of traditional dendrites would limit the ability of ciliated sensory neurons to respond to proximal axon injury. We tested this hypothesis by performing laser microsurgery on ciliated lch1 neurons in Drosophila larvae and tracking cells over time. These cells survived proximal axon injury as well as distal axon injury, and, like many other neurons, initiated growth from the axon stump after distal injury. After proximal injury, neurites regrew in a surprisingly flexible manner. Most cells initiated outgrowth directly from the cell body, but neurite growth could also emerge from the short axon stump or base of the cilium. New neurites were often branched. Although outgrowth after proximal axotomy was variable, it depended on the core DLK axon injury signaling pathway. Moreover, each cell had at least one new neurite specified as an axon based on microtubule polarity and accumulation of the endoplasmic reticulum. We conclude that ciliated sensory neurons are not intrinsically limited in their ability to grow a new axon after proximal axon removal.

Most neurons in bilaterians cannot be replaced during the lifetime of the animal. One broadly conserved strategy to maintain long-term neuronal function is axon regeneration. In many tissues, regeneration refers to replacement of damaged cells by newly generated cells, often derived from activation of a stem cell reservoir. In most regions of bilaterian nervous systems, neuronal stem cells do not persist in mature animals. Instead, individual neurons can regenerate parts of themselves after irreversible damage.

Two major conserved patterns of axon regeneration have been described in invertebrate and vertebrate bilaterians. The most familiar of these has been studied extensively in the vertebrate peripheral nervous system (PNS). After transection of axons in a peripheral nerve, the part of the axon distal to the cut site degenerates through an active process termed Wallerian degeneration (Coleman and Hoke, 2020; Ding and Hammarlund, 2019). The remaining stump of the axon then reinitiates growth, and if regeneration is successful, will reconnect with a target cell (Chen et al., 2007; Smith et al., 2020). This pattern of axon regeneration from the stump has also been described in invertebrate model systems including the nematode Caenorhabditis elegans and the fruit fly Drosophila melanogaster (Brace and DiAntonio, 2017; Byrne and Hammarlund, 2017; Hao and Collins, 2017). The other cellular pattern of axon regeneration occurs when axons are completely removed from the cell body and a stump of 50 μm or less remains. In this case, the axon stump is not competent to initiate regrowth, and instead a dendrite is re-specified as an axon, which then initiates outgrowth. While much less studied, this pattern of regeneration of an axon from a converted dendrite also seems to be broadly conserved and has been described in vertebrates including rodents (Gomis-Ruth et al., 2008) and in the invertebrate D. melanogaster (Stone et al., 2010).

Most work on axon regeneration has been conducted in multipolar neurons with branched dendrites. However, neurons in which receptive function is conducted by modified cilia rather than branched dendrites are critically important for sensation in most animals, and axon regeneration of this type of neuron is not well investigated. Examples of ciliated sensory neurons in vertebrates include retinal rod and cone cells, olfactory sensory neurons and inner ear hair cells. Hair cells do not have an axon: ribbon synapses form at the base of the cell body, so axon regeneration is not relevant in this cell type. Rod and cone cells have a very short axon that seems unlikely to be susceptible to damage independently from the rest of the cell. Indeed, in these cells, most damage is due to phototoxicity, and the light-sensitive outer segment is replaced at a rate of about 10% a day to account for this type of damage (Wensel et al., 2021). In some vertebrates, including fish, damaged photoreceptors can actually be replaced in adults (Stenkamp, 2007). Cellular replacement is the major strategy for repair of olfactory sensory neurons, and the olfactory epithelium is one of the few areas of the mammalian nervous system that generates new neurons throughout life (McClintock et al., 2020). Thus, vertebrate ciliated sensory neurons seem to bypass the need for axon regeneration, either by dispensing with an axon altogether or by maintaining the ability to replace the entire damaged cell. One possible explanation for the evolutionary bypass of a need for axon regeneration in vertebrate ciliated neurons is that axon regeneration is somehow incompatible with a sensory cilium. Although it is unclear why this would be the case for regeneration from the axon stump after distal injury, the highly stereotyped nature of the sensory cilium could make it less amenable to conversion to a new axon after proximal axon injury.

Invertebrate animals rely perhaps more heavily on ciliated sensory neurons than vertebrate animals, and most of these neurons have long axons. For example, insects, including the model D. melanogaster, use ciliated neurons for olfaction (Jana et al., 2011) and hearing (Albert and Gopfert, 2015) like vertebrates, but also use ciliated sensory endings of neurons to detect mechanical stimuli including joint positioning and stretch (proprioception), taste and many types of touch (Kernan, 2007; Tuthill and Wilson, 2016). The model nematode C. elegans also relies heavily on ciliated sensory neurons to detect its surroundings. Out of the 302 neurons in the C. elegans hermaphrodite, 60 receive input from sensory cilia (Bae and Barr, 2008; Goodman and Sengupta, 2019; Maurya, 2022). These ciliated neurons mediate the vast majority of sensory input and respond to temperature, mechanical and chemical cues (Bae and Barr, 2008; Goodman and Sengupta, 2019). Motor neurons and non-ciliated sensory neurons regenerate axons in much the same way as vertebrate peripheral neurons in both C. elegans and Drosophila (Brace and DiAntonio, 2017; Byrne and Hammarlund, 2017; Hao and Collins, 2017). In Drosophila neurons with branched dendrites, as in mammalian neurons, injury of axons beyond about 50 μm from the cell body leads to regeneration from the axon stump, while injury closer to the cell body often leads to conversion of a dendrite into a regenerating axon (Rao and Rolls, 2017; Stone et al., 2010). It is less clear whether ciliated sensory neurons have the capacity to regrow axons after injury. One study in the locust tracked olfactory sensory neurons after nerve crush, which presumably severed axons at a distance of over 50 μm from the cell body, and found axons could regenerate back to the olfactory bulb (Bicker and Stern, 2020). This study suggests that ciliated sensory neurons regenerate axons after distal axotomy. Reports on the response to axon injury in ciliated C. elegans neurons are mixed. One study reported axon regeneration in motor neurons after injury, but not in ciliated neurons ASH and AWC (Gabel et al., 2008). In a different study, ASJ ciliated sensory neurons could make new axons after complete axon removal (Chung et al., 2016). Thus, although vertebrate sensory systems have evolved in a way that minimizes the requirement for axon regeneration in ciliated sensory neurons, this capacity may exist in other animals.

The Drosophila larva is a useful system for studying neuronal responses to different types of injury (Brace and DiAntonio, 2017; Hao and Collins, 2017). In particular, sensory neurons with branched dendrites have been shown to regenerate axons from the stump after distal axotomy (Stone et al., 2012) and to convert dendrites into axons after proximal axotomy (Rao and Rolls, 2017; Stone et al., 2010), and both types of regeneration rely on the conserved DLK/JNK signaling pathway (Stone et al., 2014; Stone et al., 2010) that is used for regeneration in C. elegans (Hammarlund et al., 2009), Drosophila motor neurons (Xiong et al., 2010) and vertebrate motor neurons (Adula et al., 2022). Ciliated sensory neurons are found near branched sensory neurons in the larval body wall, and their axons bundle together in nerves that transmit sensory information to the central nervous system. We therefore decided to use this system to determine whether ciliated neurons can regenerate like their non-ciliated neighbors after axon injury. We find that ciliated sensory neurons initiate regeneration from the axon stump after distal axotomy, and surprisingly can also regenerate after proximal axotomy. Indeed, after complete axon removal, the new axon can emerge from a variety of sites including the cell body and the cilium itself, demonstrating that ciliated sensory neurons can exhibit remarkable flexibility in response to injury.

Fly culture and stocks

Flies were maintained at 25°C in plastic vials or bottles with standard Drosophila cornmeal media. Many fly stocks were obtained from the Bloomington Drosophila Stock Center and the Vienna Drosophila Resource Center. UAS-Dcr2-nls-BFP was generated in our lab (Swope et al., 2022). Virgin female flies from the tester line (iav-Gal4, UAS-mCD8-mCherry, UAS-Dcr2-nls-BFP/TM6) were crossed to male flies from an RNAi line (see Table S1). Males from the lines UAS-EB1-GFP, puc-GFP/TM6 and UAS-Rtnl1-GFP/TM6 were also used. Crosses were kept in bottles capped with 35 mm Petri dishes filled with Drosophila media. Embryos were collected every 24 h on these caps and each cap was then incubated for 2 days at 25°C to generate larvae for injury or imaging.

Axon injury assay

Virgin females from the iav-Gal4 tester line were crossed to males from an RNAi line (see Table S1) or UAS-EB1-GFP. Embryos from these crosses were collected on 35 mm caps filled with standard Drosophila media. Larvae were isolated from caps that had been incubated for 2 days, and these larvae were washed with water to remove media. Next, one larva was mounted on a piece of dried agar on the middle of a microscope slide. It was positioned such that the right or left lateral side faced up. A 22×40 mm coverslip was placed on top of the larva with enough pressure to immobilize the larva and then secured with tape.

Imaging and laser injury were performed with a Zeiss inverted LSM800 confocal microscope using Zen software. A ×10 objective was used to locate larvae, and lch1 neurons in segments A2–A7 were located with a ×63 oil (1.4 NA) objective. A Micro Point pulsed UV laser was used to sever the axon of one lch1 neuron expressing the cell shape marker mCD8-mCherry under the control of iav-Gal4. For proximal axon injury experiments, the axon was severed less than 10 μm from the cell body. For distal axon injury experiments, the axon was severed at least 70 μm from the cell body. If the axon retracted immediately after injury, the larva was discarded or the 96 h data were excluded from analysis.

After injury, larvae were recovered, placed in media, and kept at 20°C for 96 h. For distal axon injury experiments, most larvae were also imaged at 24 h post-injury to check for complete severing of the axon. In both the proximal and distal axon injury experiments, regeneration was assessed 96 h post-injury. In some cases, larvae were anesthetized with isoflurane (99.9%, Patterson Veterinary) at the 96 h time point to ensure complete immobilization. A larva was placed in a 35 mm Petri dish nested inside a larger dish with a twisted Kimwipe between the two dishes. About 10 drops of isoflurane were added to the Kimwipe and the chamber was covered for about 5 min. When the larva was fully immobilized, it was mounted on a microscope slide in the same way as described above.

Z-stack images were collected immediately after injury (0 h) and 96 h after injury. All images were taken using a Zeiss ×63 oil objective.

Quantification of axon regeneration

The Z-stacks acquired during the axon injury experiments were collapsed into maximum intensity projections using ImageJ (https://imageJ.net/Fiji) (Schindelin et al., 2012). If the area of growth extended beyond one image, multiple images were stitched together using the stitching plugin (Preibisch et al., 2009) or using canvas X draw (https://www.canvasgfx.com). In general, the length of new growth was measured from the cell body and included all new growth. Cases where this approach was modified are as follows. (1) For proximal injury, if new growth originated from the axon stump and the boundary between new growth and the stump was clear, the stump was not included in the measurement of new growth. If the boundary was ambiguous, the stump and new growth were included in the measurement and the length of the stump at 0 h was subtracted from this measurement. (2) If neurites growing close together could not be distinguished in the projection, the Z-stacks were examined to separate the neurites. The length of regions that were ambiguous was measured as if the region were a single neurite. (3) If a regenerating neurite grew into the lch5 axon bundle, the length of the new growth included as much of the neurite that was possible to distinguish.

Puc-GFP assay

Adapting methods previously described (Stone et al., 2014), virgin females from the iav-Gal4 tester line were crossed to puc-GFP/TM6 males; 2 day old larvae were selected and mounted in the same manner as for the axon injury assay described above. A Zeiss inverted LSM800 confocal microscope was used for imaging and laser injury. A lch1 neuron in segment A2, A3 or A4 was selected, and an image of the nucleus was acquired using a Zeiss ×63 oil objective. After imaging, a Micro Point pulsed UV laser was used to sever the axon. For proximal injury, the axon was severed less than 10 μm from the cell body, and for dendrite injury, the dendrite was severed near the middle of its length. After injury, larvae were recovered, placed in standard media, kept at 25°C, and re-imaged at 24 h using a Zeiss ×63 oil objective.

For GFP quantification, Z-stacks were combined into a maximum intensity projection using ImageJ. Only images having at least three slices, including the brightest slice, that were able to be combined into a projection were included in the analysis. The average intensity within the nucleus was measured for each projection using ImageJ. The fold-change in puc-GFP was then calculated from the average nuclear fluorescence intensity at 24 h post-injury and the average nuclear fluorescence intensity before injury.

Microtubule polarity assay

To assess microtubule polarity in regenerated neurites, females from the iav-Gal4 tester line were crossed to UAS-EB1-GFP males and axon injury was performed as described for the axon injury assay. After injury, larvae were placed in media and kept at 20°C for 96 h.

At 96 h, larvae were anesthetized using isoflurane as described above. A Zeiss ×63 oil objective on a Zeiss inverted LSM800 confocal microscope was used to collect time series movies with frames obtained every 1.27 s. Movies were collected in the same way for uninjured neurons in 2 or 3 day old larvae.

Microtubule polarity was quantified using ImageJ. All regions of new growth were quantified in each neuron. A region was defined as the neurite between branch points or distal to the last branch point. The number of comets moving to or from the cell body was manually counted in each region. A region was classified as plus-end out if 75% of comets moved away from the cell body. To generate kymographs, the template matching plugin was used to stabilize the movies, a line was drawn along the region of interest, and the multikymograph plugin with line width set to 3 was used.

Rtnl1 localization assay

Virgin females from the iav-Gal4 tester line were crossed to UAS-Rtnl1-GFP/TM6 males to investigate the distribution of endoplasmic reticulum (ER) in regenerating neurites. Axon injury was performed in the same way as described for the axon injury assay. Larvae were incubated for 96 h at 20°C and 96 h images were collected using a Zeiss ×63 oil objective. The laser attenuation was adjusted per cell to avoid saturation.

Maximum intensity projections were assembled using ImageJ. The number of neurites with Rtnl1-GFP enriched at the tip was quantified for each neuron. For each neurite in a cell, a segmented line was drawn manually starting at the tip and extending 25 μm into the neurite toward the cell body. Any branch points were excluded from the line. A plot profile was created and assessed for any peaks in fluorescence within this 25 µm region. If peaks were at least 2× the fluorescence of the shaft region of the neurite, then the neurite was classified as having Rtnl1-GFP enriched at the tip.

Statistical analysis and graphing

Statistical analysis was performed using GraphPad Prism software (https://www.graphpad.com) or R version 4.1.2 (https://www.r-project.org/). Statistical tests used are described in the figure legends.

The lch1 ciliated sensory neuron survives proximal and distal axon injury

Injury of single axons can be performed in whole, living Drosophila larvae using laser microsurgery (Stone et al., 2010). Two major categories of ciliated sensory neurons are found in the abdominal segments of the larval body wall. External sensory (es) neurons have cilia that end in the cuticle, and chordotonal (cho) neurons act primarily as stretch receptors within the larval body to help coordinate movement (Caldwell et al., 2003; Hassan et al., 2019) and do not penetrate the cuticle (Bodmer and Jan, 1987; Campos-Ortega and Hartenstein, 1997; Ghysen et al., 1986). We chose to focus on the cho neurons as their dendrite, cell body and proximal axon can all be visualized in a relatively thin section of the body wall (Fig. 1A). Within each abdominal hemisegment of the larva there are eight of these cells. lch1 (lateral monoscolopidial chordotonal organ) is the most dorsal of these and relatively isolated (Fig. 1A,B), so we focused on this cell. To visualize lch1, we used iav-Gal4 to drive expression of an RFP-tagged membrane marker (mCD8-mCherry).

Fig. 1.

lch1 sensory neurons can survive both proximal and distal axon injury. (A) A whole live larva expressing mCD8-mCherry under control of iav-Gal4, mounted laterally to show the location of lch5 (purple arrow) and lch1 (orange arrow) in the context of the entire animal. The dorsal surface of the animal is at the top. (B) lch1 can be seen as an isolated neuron compared with the bundle of lch5 neurons. The red arrow indicates the site of distal axon injury after severing with a pulsed UV laser. The cell survives 24 h after injury and the blue asterisk indicates the site of the axon stump. (C) Examples of distal axon, proximal axon and cilium injury to lch1 neurons. Cells were injured at 0 h – 24 h later, survival after all three types of injury can be seen. Blue asterisks indicate the sites of proximal axon and cilium injury. (D) Quantification of survival 24 h post-injury: 50% of neurons that underwent cilium injury survived while all neurons that underwent distal or proximal axon injury survived. The number of neurons for each injury condition is indicated on the bars.

Fig. 1.

lch1 sensory neurons can survive both proximal and distal axon injury. (A) A whole live larva expressing mCD8-mCherry under control of iav-Gal4, mounted laterally to show the location of lch5 (purple arrow) and lch1 (orange arrow) in the context of the entire animal. The dorsal surface of the animal is at the top. (B) lch1 can be seen as an isolated neuron compared with the bundle of lch5 neurons. The red arrow indicates the site of distal axon injury after severing with a pulsed UV laser. The cell survives 24 h after injury and the blue asterisk indicates the site of the axon stump. (C) Examples of distal axon, proximal axon and cilium injury to lch1 neurons. Cells were injured at 0 h – 24 h later, survival after all three types of injury can be seen. Blue asterisks indicate the sites of proximal axon and cilium injury. (D) Quantification of survival 24 h post-injury: 50% of neurons that underwent cilium injury survived while all neurons that underwent distal or proximal axon injury survived. The number of neurons for each injury condition is indicated on the bars.

We first wished to test whether lch1 neurons would survive axon severing. We hypothesized that if these cells could not regenerate axons then they would die after losing axonal connections to targets in the central nervous system (CNS). We used a pulsed UV laser to precisely cut axons or the cilium (Fig. 1B,C) as this method has previously been used to sever axons and dendrites or neighboring dendritic arborization neurons. Dendritic arborization neurons do not die after laser microsurgery, but initiate regeneration after this type of injury to axons (Stone et al., 2010) and dendrites (Stone et al., 2014), and even removal of the axon and dendrites (Shorey et al., 2020). Axons of lch1 were severed either within 10 μm of the cell body (proximal injury) or greater than 70 μm from the cell body (distal injury). For comparison, cilia were severed about halfway along their length. Only one neuron was injured per animal. After injury, animals were returned to normal media, and were remounted for imaging 24 h later. Half of the cells with injured cilia died. However, all the cells in which the axon had been injured were still present (Fig. 1B–D). Thus, lch1 ciliated neurons survive even complete axon removal.

lch1 sensory neurons initiate outgrowth in response to axon severing

To determine whether ciliated lch1 neurons attempt to regenerate an axon, we monitored cells for longer time periods after laser microsurgery. After distal axotomy, most non-ciliated neurons regenerate from the severed stump. We used a threshold of 70 μm from the cell body as distal injury based on previous studies in mammals (Gomis-Ruth et al., 2008), nematodes (Gabel et al., 2008) and flies (Rao and Rolls, 2017). By 96 h after injury, clear growth from the stump had occurred in 19/33 animals and new axons followed two types of trajectories (Fig. 2). Axons from lch1 join a nerve that carries sensory axons centrally and motor axons to muscle targets in the periphery (Campos-Ortega and Hartenstein, 1997). In some instances, the new neurite that initiated from the stump followed the nerve towards the CNS (Fig. 2A). However, new neurites often left the main nerve to track a branch that led to a different peripheral target (Fig. 2B). Thus, after distal axotomy, lch1 neurons can grow from the axon stump, and the new neurite grows along the nerve, but not necessarily in the correct direction.

Fig. 2.

lch1 sensory neurons regenerate 96 h after distal axon injury. (A) Distal axon injury of lch1. The red arrow indicates the site of injury at 0 h. Neurons in B and C were injured in a similar manner at 0 h. At 96 h after injury, regeneration has initiated from the stump and regrowth occurs down the nerve bundle. The dashed green line indicates the region of new growth. (B) Distal axon injury of lch1. At 96 h after injury, regeneration has initiated from the axon stump and branches so that it follows a nerve branch to a different peripheral target as well as part of the nerve back to the CNS. The dashed green line indicates regions of new growth. (C) Distal axon injury of lch1. At 96 h after injury, the neuron has failed to regenerate. The blue asterisk indicates the axon stump at 96 h. (D) Quantification of the percentage of cells that failed to regenerate and those that regenerated either by following the nerve bundle or by following a peripheral nerve branch. Those classified as following the peripheral branch also often had some growth along the main nerve as shown in B. The number of neurons for each regeneration outcome is indicated.

Fig. 2.

lch1 sensory neurons regenerate 96 h after distal axon injury. (A) Distal axon injury of lch1. The red arrow indicates the site of injury at 0 h. Neurons in B and C were injured in a similar manner at 0 h. At 96 h after injury, regeneration has initiated from the stump and regrowth occurs down the nerve bundle. The dashed green line indicates the region of new growth. (B) Distal axon injury of lch1. At 96 h after injury, regeneration has initiated from the axon stump and branches so that it follows a nerve branch to a different peripheral target as well as part of the nerve back to the CNS. The dashed green line indicates regions of new growth. (C) Distal axon injury of lch1. At 96 h after injury, the neuron has failed to regenerate. The blue asterisk indicates the axon stump at 96 h. (D) Quantification of the percentage of cells that failed to regenerate and those that regenerated either by following the nerve bundle or by following a peripheral nerve branch. Those classified as following the peripheral branch also often had some growth along the main nerve as shown in B. The number of neurons for each regeneration outcome is indicated.

After confirming lch1 could regenerate from the stump like neurons with branched dendrites, we were curious whether it would be able to initiate growth after proximal axotomy. We used laser microsurgery to remove the axon, and then imaged the cell 96 h later. Most cells (36/38) grew new neurites (Fig. 3). Two things surprised us about the patterns of regrowth. First, these cells seem much more flexible in terms of the site from which growth initiated than expected. Neighboring dendritic arborization sensory neurons only initiate outgrowth from dendrites, not the stump or cell body, after proximal axotomy (Rao and Rolls, 2017; Stone et al., 2010). In contrast, lch1 neurons could initiate new growth from the cell body, the short axon stump or the base of the cilium (Fig. 3B–E). The other surprising aspect of the new growth was that it was often quite branched. Again, this distinguishes the growth from that of dendritic arborization neurons after proximal axotomy, which is largely linear (Rao and Rolls, 2017; Stone et al., 2010), and from growth after distal cuts, which tends to follow existing nerves with minimal branching (Fig. 2). In both cell types, the amount of outgrowth was quite variable and the new neurites wandered around the body wall (Fig. 3F; see also Rao and Rolls, 2017; Stone et al., 2010). The unexpected differences from regeneration of dendritic arborization neurons after proximal axotomy led us to question whether these cells were using a similar molecular pathway, and whether the new neurite was correctly specified as an axon.

Fig. 3.

Regeneration can initiate from the cell body, axon stump or cilium following proximal axon injury of lch1 sensory neurons. (A) Representative proximal axon injury of lch1. The red arrow indicates the site of injury at 0 h. Neurons in B–D were injured in a similar manner at 0 h. Green arrowheads show the site of emergence of the new axon. (B) Example lch1 showing that 96 h after proximal axon injury, regeneration can initiate from the cell body. (C) Example lch1 showing that 96 h after proximal axon injury, regeneration can initiate from the axon stump. (D) Example lch1 showing that 96 h after proximal axon injury, regeneration can initiate from the cilium. (E) Quantification of the percentage of cells that regenerated from the different sites following injury. The number of neurons in each case is indicated. (F) Quantification of lengths of new growth following proximal axon injury (box plot shows median, upper and lower quartiles and the minimum and maximum). The median length of new growth was 298 μm (n=38).

Fig. 3.

Regeneration can initiate from the cell body, axon stump or cilium following proximal axon injury of lch1 sensory neurons. (A) Representative proximal axon injury of lch1. The red arrow indicates the site of injury at 0 h. Neurons in B–D were injured in a similar manner at 0 h. Green arrowheads show the site of emergence of the new axon. (B) Example lch1 showing that 96 h after proximal axon injury, regeneration can initiate from the cell body. (C) Example lch1 showing that 96 h after proximal axon injury, regeneration can initiate from the axon stump. (D) Example lch1 showing that 96 h after proximal axon injury, regeneration can initiate from the cilium. (E) Quantification of the percentage of cells that regenerated from the different sites following injury. The number of neurons in each case is indicated. (F) Quantification of lengths of new growth following proximal axon injury (box plot shows median, upper and lower quartiles and the minimum and maximum). The median length of new growth was 298 μm (n=38).

lch1 neurons use the conserved DLK axon injury signaling pathway

Axon injury in Drosophila motor neurons (Xiong et al., 2010) and dendritic arborization neurons (Stone et al., 2014, 2010) activates a DLK/JNK/fos signaling pathway. One output of this pathway is transcription of the puc gene, which encodes a MAP kinase phosphatase (Xiong et al., 2010). A puc-GFP reporter accumulates in the nucleus of dendritic arborization neurons after axon, but not dendrite, injury (Stone et al., 2014) and acts as a specific readout of this pathway. We introduced the puc-GFP reporter into larvae with a red cell shape marker driven by iav-Gal4 and severed either the cilium or axon of lch1 cells (Fig. 4A). When monitored 24 h later, nuclear puc-GFP fluorescence was similar to that before injury in the cilium group and was elevated over twofold in the axon injury group (Fig. 4B). This specific increase in puc signal after axon injury is similar to that in dendritic arborization neurons (Stone et al., 2014) and consistent with activation of DLK/JNK/fos. To test whether this pathway was required for outgrowth in response to axon injury in lch1 neurons, we used cell type-specific RNAi to knock down DLK (Fig. 4C). Most control RNAi neurons extended neurites 96 h after injury. In contrast, most DLK RNAi neurons had no growth (Fig. 4D). A small subset of DLK RNAi cells did exhibit growth, likely because of incomplete knockdown. We conclude that even though neurites generated in response to proximal axotomy exhibit more branching than is typical of other regenerating axons and can emerge from a variety of sites, they depend on the core DLK/JNK/fos axon injury pathway.

Fig. 4.

The DLK pathway is initiated after proximal axon injury of lch1 sensory neurons. (A) Example images of an lch1 neuron expressing mCD8-mcherry and puc-GFP at 0 h and at 24 h after cilia injury or proximal axon injury. White arrowheads indicate the nucleus. (B) Quantification of the fluorescence change in nuclear puc-GFP 24 h after injury compared with that at the time of injury (mean±s.d.). The two groups were compared using a two-sample t-test. (C) Control RNAi and DLK RNAi lch1 neurons at 0 h and at 96 h after proximal axon injury. The site of injury is indicated by the red arrow at 0 h. The control RNAi neuron shows extensive regeneration at 96 h (indicated by the green dashed line) whereas the DLK RNAi neuron shows minimal to no regeneration at 96 h. The axon stump is indicated by the blue asterisk. (D) Quantification of lengths of new growth for both control RNAi and DLK RNAi. The median length of new growth for control RNAi was 143 μm (n=11) and that for DLK RNAi was 21 μm (n=12). The two groups were compared using a Mann–Whitney test.

Fig. 4.

The DLK pathway is initiated after proximal axon injury of lch1 sensory neurons. (A) Example images of an lch1 neuron expressing mCD8-mcherry and puc-GFP at 0 h and at 24 h after cilia injury or proximal axon injury. White arrowheads indicate the nucleus. (B) Quantification of the fluorescence change in nuclear puc-GFP 24 h after injury compared with that at the time of injury (mean±s.d.). The two groups were compared using a two-sample t-test. (C) Control RNAi and DLK RNAi lch1 neurons at 0 h and at 96 h after proximal axon injury. The site of injury is indicated by the red arrow at 0 h. The control RNAi neuron shows extensive regeneration at 96 h (indicated by the green dashed line) whereas the DLK RNAi neuron shows minimal to no regeneration at 96 h. The axon stump is indicated by the blue asterisk. (D) Quantification of lengths of new growth for both control RNAi and DLK RNAi. The median length of new growth for control RNAi was 143 μm (n=11) and that for DLK RNAi was 21 μm (n=12). The two groups were compared using a Mann–Whitney test.

Neurites grown in response to proximal axotomy have cellular features of regenerating axons

To more completely probe the identity of neurites that emerge after proximal axon severing, we examined two features that distinguish regenerating axons from dendrites in Drosophila dendritic arborization neurons. We previously found that the ER concentrates near regenerating axon, but not dendrite, tips (Rao et al., 2016). We therefore examined the distribution of the ER marker Reticulon-like1-GFP (Rtnl1-GFP) during lch1 regeneration. We found that after proximal axotomy all cells had ER near the tip in at least one neurite (Fig. 5A,B). Consistent with more branching growth than seen in dendritic arborization neurons, there were many cells that had more than one tip with strong Rtnl1-GFP signal (Fig. 5B). To further examine whether lch1 cells correctly specified new axons after proximal axotomy, we examined microtubule polarity. Axons are characterized by microtubules arranged with their more dynamic plus ends directed away from the cell body, while dendrites have mixed or minus-end out polarity (Baas and Lin, 2011; Rolls, 2022; Rolls and Jegla, 2015). To map the orientation of microtubules, we used GFP-tagged EB1. EB1 binds to microtubule ends when they are in the growth, but not shrinking, phase and so the direction of EB1 movement can be used to infer microtubule polarity (Stepanova et al., 2003). In uninjured lch1 axons, EB1-GFP comets moved away from the cell body (Fig. 5C,D) indicating axons had plus-end out microtubule polarity as expected. The pattern of microtubule growth was more complex after axon injury. First, many more EB1-GFP comets were seen in regenerating than in uninjured neurites (Fig. 5C,E). Axon injury increases microtubule dynamics in Drosophila dendritic arborization neurons (Stone et al., 2010) and rodent neurons (Kleele et al., 2014), so the increase we observed is consistent with previous results. Second, we noticed that many of the growing neurites had regions with different microtubule organization. Importantly, almost all (17/19) cells had a neurite with a plus-end out region (Fig. 5E,H), indicating that they had likely specified a new axon. However, many cells also had regions with mixed polarity, which is more characteristic of dendrites (Fig. 5F–H). These mixed regions tended to be either near tips or in regions that were branched. This combination of plus-end out and mixed regions with branching is reminiscent of dendritic arborization neurons after removal of all dendrites and the axon, which was interpreted as regeneration of both the axon and dendrites (Shorey et al., 2020). It is possible that removal of the axon in ciliated neurons triggers the cells to try to make branched dendrites, but the data are not strong enough to make a firm conclusion. The data do strongly support that ciliated neurons can initiate a program of axon regeneration, even after complete axon removal.

Fig. 5.

Regenerated neurites after proximal axon injury of lch1 sensory neurons have axon-like characteristics. (A) Example 96 h images of lch1 neurons expressing mCD8-mcherry and the endoplasmic reticulum marker Rtnl1-GFP. The axon of each neuron was injured proximal to the cell body at 0 h (not shown) and localization of Rtnl1-GFP in the regenerating neurites was assessed at 96 h. The accumulation of Rtnl1-GFP in neurite tips is indicated with white asterisks (one neurite on the left, two neurites on the right). (B) Quantification of the number of cells that had one, two or three neurites with Rtnl1-GFP enriched at the tip. (C) Microtubule plus ends were visualized using EB1-GFP in uninjured lch1 neurons. A representative kymograph generated from the axon of one uninjured neuron shows plus-end out comets (marked with cyan lines). (D) Quantification of the microtubule polarity of uninjured lch1 axons shows a 100% plus-end out microtubule array (n=17 comets in 20 neurons). (E) Microtubule polarity was tracked in the regenerated regions following proximal axon injury. An example kymograph of a plus-end out region (indicated by the purple box on the overview image) is shown. (F) An example kymograph of a mixed tip region (indicated by the blue box in the overview image). Here, the comets can be seen moving both toward and away from the cell body. (G) An example kymograph of a mixed branch region (indicated by orange box). Here, there is a region of neurite branching where comets are moving both toward and away from the cell body. (H) Quantification of the number of neurons that contain a plus-end out region. In addition, the number of neurons that contain a mixed tip, mixed branch or both mixed tip and mixed branch region is shown.

Fig. 5.

Regenerated neurites after proximal axon injury of lch1 sensory neurons have axon-like characteristics. (A) Example 96 h images of lch1 neurons expressing mCD8-mcherry and the endoplasmic reticulum marker Rtnl1-GFP. The axon of each neuron was injured proximal to the cell body at 0 h (not shown) and localization of Rtnl1-GFP in the regenerating neurites was assessed at 96 h. The accumulation of Rtnl1-GFP in neurite tips is indicated with white asterisks (one neurite on the left, two neurites on the right). (B) Quantification of the number of cells that had one, two or three neurites with Rtnl1-GFP enriched at the tip. (C) Microtubule plus ends were visualized using EB1-GFP in uninjured lch1 neurons. A representative kymograph generated from the axon of one uninjured neuron shows plus-end out comets (marked with cyan lines). (D) Quantification of the microtubule polarity of uninjured lch1 axons shows a 100% plus-end out microtubule array (n=17 comets in 20 neurons). (E) Microtubule polarity was tracked in the regenerated regions following proximal axon injury. An example kymograph of a plus-end out region (indicated by the purple box on the overview image) is shown. (F) An example kymograph of a mixed tip region (indicated by the blue box in the overview image). Here, the comets can be seen moving both toward and away from the cell body. (G) An example kymograph of a mixed branch region (indicated by orange box). Here, there is a region of neurite branching where comets are moving both toward and away from the cell body. (H) Quantification of the number of neurons that contain a plus-end out region. In addition, the number of neurons that contain a mixed tip, mixed branch or both mixed tip and mixed branch region is shown.

We hypothesized that neurons with sensory cilia might have more limited axon regenerative capacity than neurons with branched dendrites for two reasons. First, although mammals rely on sensory cilia to receive a variety of information types from the environment, the neurons that house them have strategies to avoid the need for axon regeneration. The cells are either routinely replaced or do not have long axons. Second, neurons with branched dendrites use one of these dendrites as the basis for a new, regenerating, axon after proximal axotomy. Conversion of a sensory cilium to a new axon seems unlikely as the microtubule organization is inflexibly built around a basal body, and most sensory neurons have a single cilium which would not be able to maintain its function if converted to an axon.

To probe the regenerative capacity of a ciliated sensory neuron, we performed laser microsurgery on lch1 in Drosophila larvae. After distal axotomy, these cells were able to initiate axon regeneration from the axon stump like many other neurons. Surprisingly, they also survived complete axon removal and showed remarkable flexibility in generating new neurites. The new processes could emerge from the base of the cilium, the cell body or the axon stump, although the cell body was the most common. Based on activation and requirement of the DLK/JNK/fos signaling pathway that coordinates axon injury responses in other cells, as well as subcellular analysis, we conclude that lch1 neurons can specify a new regenerating axon after axon removal. In many cases they also generated neurites or regions of neurites with more ambiguous identity. We found no evidence that the presence of a sensory cilium limits the capacity of neurons to respond to axon injury.

Initial reports on axon injury of ciliated neurons in C. elegans suggested that these cells may not be able to regenerate axons. Gabel et al. (2008) used laser microsurgery to compare responses of different neurons to axon injury. They found that the non-ciliated mechanosensory neuron AVM, as well as DA/DB and HSN motor neurons, can regenerate axons in the adult. In contrast, two ciliated sensory neurons, ASH and AWC, did not exhibit axon regeneration (Gabel et al., 2008). However, a later study on another ciliated cell of the same amphid type, ASJ, demonstrated that it exhibits DLK-dependent axon regeneration after proximal axotomy (Chung et al., 2016). Similar to regeneration after proximal axotomy of lch1, new growth could initiate from the cell body or base of the cilium (Chung et al., 2016). The likely explanation for the discrepancy between the initial report of no regeneration and the subsequent findings of regeneration is that in most C. elegans neurons, regeneration initiates quickly and is easily detectable 24 h after injury. This was the time point analyzed in the first study (Gabel et al., 2008). However, the later study tracked neurons longer and found that for the ASJ ciliated neuron, regeneration is often not detectable until 3 days after injury (Chung et al., 2016). So, in both C. elegans and Drosophila, ciliated sensory neurons can regrow axons after they are completely removed. Moreover, in both cases, regeneration uses the classic DLK axon injury signaling pathway. We conclude that the presence of a sensory cilium is compatible with axon regeneration, even after axon severing near the cell body.

We are grateful to members of the Rolls lab for thoughtful discussions and helpful technical advice throughout the study. The Bloomington Drosophila Stock Center (https://bdsc.indiana.edu/, supported by NIH P40OD018537) and Vienna Drosophila Resource Center (https://stockcenter.vdrc.at/) were extremely helpful in providing stocks used in this study.

Author contributions

Conceptualization: M.M.R.; Formal analysis: M.C.S., A.S.M.; Investigation: M.C.S., A.S.M.; Writing - original draft: M.C.S., A.S.M., M.M.R.; Writing - review & editing: M.C.S., A.S.M., M.M.R.; Visualization: M.C.S., A.S.M.; Supervision: M.M.R.; Funding acquisition: M.M.R.

Funding

Funding was provided in part by the National Institutes of Health, R01 GM085115. Deposited in PMC for release after 12 months.

Data availability

All relevant data can be found within the article.

Adula
,
K. P.
,
Shorey
,
M.
,
Chauhan
,
V.
,
Nassman
,
K.
,
Chen
,
S. F.
,
Rolls
,
M. M.
and
Sagasti
,
A.
(
2022
).
The MAP3Ks DLK and LZK direct diverse responses to axon damage in zebrafish peripheral neurons
.
J. Neurosci.
42
,
6195
-
6210
.
Albert
,
J. T.
and
Gopfert
,
M. C.
(
2015
).
Hearing in Drosophila
.
Curr. Opin. Neurobiol.
34
,
79
-
85
.
Baas
,
P. W.
and
Lin
,
S.
(
2011
).
Hooks and comets: the story of microtubule polarity orientation in the neuron
.
Dev. Neurobiol.
71
,
403
-
418
.
Bae
,
Y. K.
and
Barr
,
M. M.
(
2008
).
Sensory roles of neuronal cilia: cilia development, morphogenesis, and function in C. elegans
.
Front. Biosci.
13
,
5959
-
5974
.
Bicker
,
G.
and
Stern
,
M.
(
2020
).
Structural and functional plasticity in the regenerating olfactory system of the migratory locust
.
Front Physiol
11
,
608661
.
Bodmer
,
R.
and
Jan
,
Y. N.
(
1987
).
Morphological differentiation of the embryonic peripheral neurons in Drosophila. Roux's Arch
.
Dev. Biol
196
,
69
-
77
.
Brace
,
E. J.
and
Diantonio
,
A.
(
2017
).
Models of axon regeneration in Drosophila
.
Exp. Neurol.
287
,
310
-
317
.
Byrne
,
A. B.
and
Hammarlund
,
M.
(
2017
).
Axon regeneration in C. elegans: Worming our way to mechanisms of axon regeneration
.
Exp. Neurol.
287
,
300
-
309
.
Caldwell
,
J. C.
,
Miller
,
M. M.
,
Wing
,
S.
,
Soll
,
D. R.
and
Eberl
,
D. F.
(
2003
).
Dynamic analysis of larval locomotion in Drosophila chordotonal organ mutants
.
Proc. Natl. Acad. Sci. U.S.A.
100
,
16053
-
16058
.
Campos-Ortega
,
J. A.
and
Hartenstein
,
V.
(
1997
).
The Embryonic Development of Drosophila Melanogaster
.
Berlin
:
Springer
.
Chen
,
Z. L.
,
Yu
,
W. M.
and
Strickland
,
S.
(
2007
).
Peripheral regeneration
.
Annu. Rev. Neurosci.
30
,
209
-
233
.
Chung
,
S. H.
,
Awal
,
M. R.
,
Shay
,
J.
,
Mcloed
,
M. M.
,
Mazur
,
E.
and
Gabel
,
C. V.
(
2016
).
Novel DLK-independent neuronal regeneration in Caenorhabditis elegans shares links with activity-dependent ectopic outgrowth
.
Proc. Natl. Acad. Sci. U.S.A.
113
,
E2852
-
E2860
.
Coleman
,
M. P.
and
Hoke
,
A.
(
2020
).
Programmed axon degeneration: from mouse to mechanism to medicine
.
Nature Reviews Neuroscience
21
,
183
-
196
.
Ding
,
C.
and
Hammarlund
,
M.
(
2019
).
Mechanisms of injury-induced axon degeneration
.
Curr. Opin. Neurobiol.
57
,
171
-
178
.
Gabel
,
C. V.
,
Antoine
,
F.
,
Chuang
,
C. F.
,
Samuel
,
A. D.
and
Chang
,
C.
(
2008
).
Distinct cellular and molecular mechanisms mediate initial axon development and adult-stage axon regeneration in C. elegans
.
Development
135
,
1129
-
1136
.
Ghysen
,
A.
,
Dambly-Chaudiere
,
C.
,
Aceves
,
E.
,
Jan
,
L.
and
Jan
,
Y.
(
1986
).
Sensory neurons and peripheral pathways in Drosophila embryos
.
Rouxs Arch. Dev. Biol.
195
,
281
-
289
.
Gomis-Ruth
,
S.
,
Wierenga
,
C. J.
and
Bradke
,
F.
(
2008
).
Plasticity of polarization: changing dendrites into axons in neurons integrated in neuronal circuits
.
Curr. Biol.
18
,
992
-
1000
.
Goodman
,
M. B.
and
Sengupta
,
P.
(
2019
).
How Caenorhabditis elegans senses mechanical stress, temperature, and other physical stimuli
.
Genetics
212
,
25
-
51
.
Hammarlund
,
M.
,
Nix
,
P.
,
Hauth
,
L.
,
Jorgensen
,
E. M.
and
Bastiani
,
M.
(
2009
).
Axon regeneration requires a conserved MAP kinase pathway
.
Science
323
,
802
-
806
.
Hao
,
Y.
and
Collins
,
C.
(
2017
).
Intrinsic mechanisms for axon regeneration: insights from injured axons in Drosophila
.
Curr. Opin. Genet. Dev.
44
,
84
-
91
.
Hassan
,
A.
,
Sapir
,
L.
,
Nitsan
,
I.
,
Greenblatt Ben-El
,
R. T.
,
Halachmi
,
N.
,
Salzberg
,
A.
and
Tzlil
,
S.
(
2019
).
A change in ECM composition affects sensory organ mechanics and function
.
Cell Reports
27
,
2272
-
2280.e2274
.
Jana
,
S. C.
,
Girotra
,
M.
and
Ray
,
K.
(
2011
).
Heterotrimeric kinesin-II is necessary and sufficient to promote different stepwise assembly of morphologically distinct bipartite cilia in Drosophila antenna
.
Mol. Biol. Cell
22
,
769
-
781
.
Kernan
,
M. J.
(
2007
).
Mechanotransduction and auditory transduction in Drosophila
.
Pflugers Arch.
454
,
703
-
720
.
Kleele
,
T.
,
Marinkovic
,
P.
,
Williams
,
P. R.
,
Stern
,
S.
,
Weigand
,
E. E.
,
Engerer
,
P.
,
Naumann
,
R.
,
Hartmann
,
J.
,
Karl
,
R. M.
,
Bradke
,
F.
et al. 
(
2014
).
An assay to image neuronal microtubule dynamics in mice
.
Nat. Commun.
5
,
4827
.
Maurya
,
A. K.
(
2022
).
Structural diversity in a stereotypic organelle - Sensory cilia of Caenorhabditis elegans
.
J. Cell. Physiol.
237
,
2668
-
2672
.
Mcclintock
,
T. S.
,
Khan
,
N.
,
Xie
,
C.
,
Martens
,
J. R.
(
2020
).
Maturation of the olfactory sensory neuron and its cilia
.
Chem. Senses
45
,
805
-
822
.
Preibisch
,
S.
,
Saalfeld
,
S.
and
Tomancak
,
P.
(
2009
).
Globally optimal stitching of tiled 3D microscopic image acquisitions
.
Bioinformatics
25
,
1463
-
1465
.
Rao
,
K. S.
and
Rolls
,
M. M.
(
2017
).
Two Drosophila model neurons can regenerate axons from the stump or from a converted dendrite, with feedback between the two sites
.
Neural Dev.
12
,
15
.
Rao
,
K.
,
Stone
,
M. C.
,
Weiner
,
A. T.
,
Gheres
,
K. W.
,
Zhou
,
C.
,
Deitcher
,
D. L.
,
Levitan
,
E. S.
and
Rolls
,
M. M.
(
2016
).
Spastin, atlastin, and ER relocalization are involved in axon but not dendrite regeneration
.
Mol. Biol. Cell
27
,
3245
-
3256
.
Rolls
,
M. M.
(
2022
).
Principles of microtubule polarity in linear cells
.
Dev. Biol.
483
,
112
-
117
.
Rolls
,
M. M.
and
Jegla
,
T. J.
(
2015
).
Neuronal polarity: an evolutionary perspective
.
J. Exp. Biol.
218
,
572
-
580
.
Schindelin
,
J.
,
Arganda-Carreras
,
I.
,
Frise
,
E.
,
Kaynig
,
V.
,
Longair
,
M.
,
Pietzsch
,
T.
,
Preibisch
,
S.
,
Rueden
,
C.
,
Saalfeld
,
S.
,
Schmid
,
B.
et al. 
(
2012
).
Fiji: an open-source platform for biological-image analysis
.
Nat. Methods
9
,
676
-
682
.
Shorey
,
M.
,
Stone
,
M. C.
,
Mandel
,
J.
and
Rolls
,
M. M.
(
2020
).
Neurons survive simultaneous injury to axons and dendrites and regrow both types of processes in vivo
.
Dev. Biol.
465
,
108
-
118
.
Smith
,
T. P.
,
Sahoo
,
P. K.
,
Kar
,
A. N.
and
Twiss
,
J. L.
(
2020
).
Intra-axonal mechanisms driving axon regeneration
.
Brain Res.
1740
,
146864
.
Stenkamp
,
D. L.
(
2007
).
Neurogenesis in the fish retina
.
Int. Rev. Cytol.
259
,
173
-
224
.
Stepanova
,
T.
,
Slemmer
,
J.
,
Hoogenraad
,
C. C.
,
Lansbergen
,
G.
,
Dortland
,
B.
,
De Zeeuw
,
C. I.
,
Grosveld
,
F.
,
Van Cappellen
,
G.
,
Akhmanova
,
A.
and
Galjart
,
N.
(
2003
).
Visualization of microtubule growth in cultured neurons via the use of EB3-GFP (end-binding protein 3-green fluorescent protein)
.
J. Neurosci.
23
,
2655
-
2664
.
Stone
,
M. C.
,
Nguyen
,
M. M.
,
Tao
,
J.
,
Allender
,
D. L.
and
Rolls
,
M. M.
(
2010
).
Global up-regulation of microtubule dynamics and polarity reversal during regeneration of an axon from a dendrite
.
Mol. Biol. Cell
21
,
767
-
777
.
Stone
,
M. C.
,
Albertson
,
R. M.
,
Chen
,
L.
and
Rolls
,
M. M.
(
2014
).
Dendrite injury triggers DLK-independent regeneration
.
Cell reports
6
,
247
-
253
.
Stone
,
M. C.
,
Rao
,
K.
,
Gheres
,
K. W.
,
Kim
,
S.
,
Tao
,
J.
,
La Rochelle
,
C.
,
Folker
,
C. T.
,
Sherwood
,
N. T.
and
Rolls
,
M. M.
(
2012
).
Normal spastin gene dosage is specifically required for axon regeneration
.
Cell Reports
2
,
1340
-
1350
.
Swope
,
R. D.
,
Hertzler
,
J. I.
,
Stone
,
M. C.
,
Kothe
,
G. O.
and
Rolls
,
M. M.
(
2022
).
The exocyst complex is required for developmental and regenerative neurite growth in vivo
.
Dev. Biol.
492
,
1
-
13
.
Tuthill
,
J. C.
and
Wilson
,
R. I.
(
2016
).
Mechanosensation and adaptive motor control in insects
.
Curr. Biol.
26
,
R1022
-
R1038
.
Wensel
,
T. G.
,
Potter
,
V. L.
,
Moye
,
A.
,
Zhang
,
Z.
and
Robichaux
,
M. A.
(
2021
).
Structure and dynamics of photoreceptor sensory cilia
.
Pflugers Arch.
473
,
1517
-
1537
.
Xiong
,
X.
,
Wang
,
X.
,
Ewanek
,
R.
,
Bhat
,
P.
,
Diantonio
,
A.
and
Collins
,
C. A.
(
2010
).
Protein turnover of the Wallenda/DLK kinase regulates a retrograde response to axonal injury
.
J. Cell Biol.
191
,
211
-
223
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information