Epidermal and ECM Damage Following Pinch Injury Restricts Dendrite Regeneration in Drosophila

Visual Abstract

Abstract

Neuronal dendrites can be injured by a number of insults, but the cellular mechanism by which dendrites respond to tissue injury and undergo repair is poorly understood. Much of the field's progress has evaluated dendrite regeneration following laser injury. While precise, laser injury does not accurately model the real-world damage to surrounding tissue that would accompany neuronal injury. Here, we modify a pinch injury technique to injure both the dendrites and their surrounding tissues in both female and male Drosophila melanogaster larvae to better model what is observed following real-world neuronal trauma. We refined this technique such that only half of a sensory neuron's dendrites are injured, leaving the other half uninjured. Our data indicate that both dynamic and stable dendritic arbors regrow dendrites in response to pinch injury, and neurons primarily engage in compensatory regeneration whereby new branches are added on the uninjured half of the arbor. Comparing the regenerative response following pinch injury versus laser injury revealed that dendrites preferentially regrow into areas where the surrounding tissue was left intact and not into areas where the surrounding tissue was damaged by pinch. These results prompted us to evaluate the damage sustained to surrounding tissues. In examining non-neuronal tissues after pinch injury, we found damage to the epidermis and ECM, but not glia. We also observed a robust immune response in the pinched hemisegment. We conclude that the surrounding tissue damage combined with a sustained immune response creates a nonpermissive environment for dendrite regeneration following pinch injury.

Significance Statement

Neuronal dendrites are injured in clinical conditions, such as stroke, traumatic brain injury, and neonatal hypoxia. Dendrites also degenerate in the early stages of a number of neurodegenerative diseases. The role of surrounding tissues in dendrite regeneration is poorly characterized, especially considering that neuronal injury is typically accompanied by broad tissue damage. Our data evaluate dendrite regeneration following an injury that better mirrors real-world conditions and demonstrates that broad tissue damage diminishes a neuron's capacity to regenerate its dendrites. Our findings show that neurons preferentially regrow into intact, undamaged tissue environments, addressing a large gap in the field's knowledge: how damage to the surrounding tissue limits neuron regeneration after injury.

Introduction

Neurons are the main signal transducers of the nervous system. They are composed of two distinct structures: dendrites that receive signals and axons that send those signals. Unfortunately, dendrites can be injured by a number of insults like stroke, hypoxia, traumatic brain injury (TBI), and neurodegenerative disease. In response to this damage, dendrites are capable of regeneration. Following total dendrite laser resection, Drosophila melanogaster peripheral nervous system (PNS) dendrites can regenerate the same number of branches as age-matched uninjured control neurons (Song et al., 2012; Stone et al., 2014; Thompson-Peer et al., 2016). Similarly, Caenorhabditis elegans PNS PVD neurons can regenerate their dendrites via plasma membrane fusion (Oren-Suissa et al., 2017; Brar et al., 2022). Zebrafish spinal cord motor neurons regenerate their dendrites 5–7 d post laser microdissection, demonstrating that central nervous system (CNS) dendrites are also capable of regeneration (Stone et al., 2022). Albeit limited, dendrite regeneration in the mammalian CNS has been observed immediately following laser microsurgery in the mouse brain and spinal cord (Zhao et al., 2017) and following brain prick injury in the adult mouse cerebral cortex (Paveliev et al., 2016).

To investigate the mechanisms of dendrite regeneration, we use the dendritic arborization (da) sensory neurons of the Drosophila PNS (Grueber et al., 2002). These da neurons are grouped by class (I–IV) and innervate the epidermis of the larval body wall. Development of the da neural system depends upon permissive extracellular environments and signals from surrounding cell types and tissues. For example, an epidermally derived microRNA, bantam, regulates the scaling growth of class IV da neurons (Parrish et al., 2009; Jiang et al., 2014). Proper space-filling of class IV da neurons requires a permissive signal created by heparan sulfate proteoglycans (HSPGs) on the epidermal cell surface (Poe et al., 2017). The extracellular matrix (ECM) also plays a critical role in dendrite development in other systems (Dansie and Ethell, 2011; Levy et al., 2014; Long and Huttner, 2019; Heiman and Bülow, 2024).

In addition to normal development, recent studies hint that extracellular signals also regulate da dendrite regeneration. Epidermally derived bantam signaling restricts dendrite regeneration in class IV ddaC neurons (Song et al., 2012). Regenerated dendrites exhibit decreased ECM adherence, which is mediated by integrins and results in failure of self-avoidance mechanisms (Thompson-Peer et al., 2016). Further, dendrite regeneration in adult flies is constrained by ECM remodeling due to epidermally expressed matrix metalloproteinase 2 activity (DeVault et al., 2018). Alterations in ECM attachment may cause many of the morphological differences that newly regenerated dendrites display, such as decreased area coverage (Thompson-Peer et al., 2016).

Dendrite regeneration is a novel and growing field, but it has not been studied as extensively as axon regeneration, considering how traditional surgical injury approaches are not suitable for dendrite injury (Peterson and Benowitz, 2018; Varier et al., 2022; Hertzler and Rolls, 2024; Vaughn and Lee, 2024). Recent advancements in microscopy have overcome these challenges to studying dendrite regeneration (Song et al., 2012; Stone et al., 2014; Thompson-Peer et al., 2016; DeVault et al., 2018; Nye et al., 2020; Hertzler et al., 2023; Duarte et al., 2024; Hertzler and Rolls, 2024; Prange et al., 2024; Hwu et al., 2025), but these studies utilized a laser injury method, which is slow, has a low throughput, and does not accurately model real-world neuronal damage. Given that the role of extracellular cues in dendrite regeneration needs deeper discovery, how well an injury method models real-world trauma warrants consideration.

A more accurate means of modeling real-world injury has been demonstrated by developing an epithelial pinch wound assay to study wound healing and reepithelization in Drosophila larvae (Galko and Krasnow, 2004; Stevens and Page-McCaw, 2012; Burra et al., 2013; Tsai et al., 2017). An RNAi screen utilizing this pinch wounding assay identified genes involved in proper wound healing (Lesch et al., 2010), and it has been adapted to evaluate ECM wound healing (Ramos-Lewis et al., 2018).

Here, we have established a pinch injury technique to injure the da neurons and surrounding tissues. This injury method damaged class IV and class I da neurons, the epidermis, and ECM, while leaving the glia largely uninjured. Pinch injury also recruited hemocytes to the injured hemisegment. Response to this damage was stereotyped: dendrites preferentially regrew into empty territory where surrounding tissue was left intact, while surrounding tissue damage restricted dendrite regeneration. This pinch injury method was faster and more accurately modeled damage to the neuron, its dendrites, and surrounding tissue. Overall, we find that damage to the ECM and other surrounding tissues impedes dendrite regeneration.

Methods and Materials

Experimental model and subject details

Drosophila strains

Drosophila stocks were maintained at room temperature. The following fly strains were used in this study: Canton-S, ppk-CD4-tdGFP (second chromosome, BDSC #35842; Han et al., 2011), ppk-CD4-tdGFP (third chromosome, BDSC #35843; Han et al., 2011), Gal4^ppk (second chromosome, BDSC #32078; Grueber et al., 2003), UAS-CaMPARI2.L398T (second chromosome, BDSC #78319; Moeyaert et al., 2018), Gal4^2–21 (third chromosome, BDSC #26259; Grueber et al., 2003), UAS-CD4-tdGFP (second chromosome, BDSC #35839; Han et al., 2011), Gal4^2–21, UAS-CD4-tdTomato (third chromosome), ppk-CD4-tdTomato (second chromosome, BDSC #35844; Han et al., 2011), Gal4^A58 (third chromosome; Galko and Krasnow, 2004), viking-GFP Protein Trap (second chromosome, BDSC #98343; Morin et al., 2001), UAS-viking-GFP (third chromosome, courtesy of Noselli Lab; Van De Bor et al., 2015), Gal4^GMR38F11 (third chromosome, BDSC #50014; Jenett et al., 2012), UAS-Armadillo-GFP (third chromosome, BDSC #58725), armP::Armadillo-GFP (third chromosome, BDSC #8555; Orsulic and Peifer, 1996), Gal4^repo, UAS-mRFP (third chromosome; Sepp et al., 2001), Gal4^pxn (third chromosome, BDSC #600223; Stramer et al., 2005), UAS-CD4-tdTomato (second chromosome, BDSC #35841; Han et al., 2011), UAS-RedStinger⁶ (third chromosome, BDSC #8547; Barolo et al., 2004).

A complete list of fly stocks can be found in Table 1.

Table 1.

Key resources

Reagent or resource	Source	Identifier
Chemicals
DuPont Molykote High Vacuum Grease	Electron Microscopy Sciences	Catalog #60705
Glycerol ReagentPlus, ≥99.0% (GC)	Sigma Aldrich	CAS #56-81-5
Isoflurane	Piramal, Midwest Vet Supply	Item #193.33165.3
GeneMate LE Quick Dissolve Agarose	VWR	Catalog #E3119-500
Experimental models: organisms/strains
Drosophila: Canton-S	Lab Stock	FlyBase_FBsn0000274
Drosophila: WeeP304 (tau-GFP)	Melissa Rolls (Stone et al., 2008)	FlyBase_FBal0249389
Drosophila: w[1118]; P{w[+mC]=ppk-CD4-tdGFP}1b	BDSC	RRID:BDSC_35842, FlyBase_FBtp0068012
Drosophila:; ppk-Gal4^VK34 / CyOw; ppk-CD4-tdGFP / Tm6B	Lab Stock
Drosophila: w[1118]; P{y[+t7.7] w[+mC]=UAS-CaMPARI2.L398T}su(Hw)attP5	BDSC	RRID:BDSC_78319, FlyBase_FBti0199488
Drosophila: w[*]; betaTub60D[Pin-1]/CyO; P{?GawB}221w-	BDSC	RRID:BDSC_26259, FlyBase_FBti0114336
Drosophila: y[1] w[*]; P{w[+mC]=UAS-CD4-tdGFP}8M2	BDSC	RRID:BDSC_35839, FlyBase_FBti0143426
Drosophila:; + ; Gal4^2-21, UAS-CD4-tdTomato / Tm6B	Lab Stock
Drosophila: w[1118]; P{w[+mC]=ppk-CD4-tdTom}4a	BDSC	RRID:BDSC_35844, FlyBase_FBti0143430
Drosophila:;; A58-Gal4	Lab Stock	FlyBase_FBal0181674
Drosophila: w[1118]; P{y[+t7.7] w[+mC]=GMR38F11-GAL4}attP2	BDSC	RRID:BDSC_50014, FlyBase_FBti0135407
Drosophila: y[1] w[*]; P{w[+mC]=PTT-un}vkg[G00454]	BDSC	RRID:BDSC_98343, FlyBase_FBti0153267
Drosophila: +; sp / CyO; repo-Gal4, UAS-mRFP / TM6B	Lab Stock
Drosophila: w[*]; M{w[+mC]=UASp-arm.mGFP6}ZH-86Fb/TM6B, Tb[1]	BDSC	RRID:BDSC_58725, FlyBase_FBti0164941
Drosophila: w[*]; P{w[+mW.hs]=arm-GFP.P}83	BDSC	RRID:BDSC_8555, FlyBase_FBti0058765
Drosophila:;; UAS-Viking-GFP / Tm3, Sb	Stéphane Noselli	FlyBase_FBtp0125825
Drosophila: w[1118]; P{w[+mC]=Pxn-GAL4}3	BDSC	RRID:BDSC_600223, FlyBase_FBti0230342
Drosophila: y[1] w[*]; P{w[+mC]=UAS-CD4-tdTom}7M1	BDSC	RRID:BDSC_35841, FlyBase_FBti0143428
Drosophila: w[1118]; P{w[+mC]=UAS-RedStinger}6	BDSC	RRID:BDSC_8547, FlyBase_FBti0040830
Software and algorithms
ImageJ (FIJI)	Wayne Rasband	RRID:SCR_002285 https://fiji.sc
Adobe InDesign	Adobe	RRID:SCR_021799 https://www.adobe.com/products/indesign.html
GraphPad Prism	GraphPad	RRID:SCR_002798 https://www.graphpad.com
Zen Microscopy Software	Zeiss	RRID:SCR_013672 https://www.zeiss.com/microscopy/en/products/software/zeiss-zen.html
Microsoft Excel	Microsoft	RRID:SCR_016137 https://www.microsoft.com
Other
LSM 700 fluorescent confocal microscope	Zeiss	RRID:SCR_017377
LSM 780 fluorescent confocal microscope	Zeiss	RRID:SCR_020922
LSM 900 fluorescent confocal microscope with Airyscan 2	Zeiss	RRID:SCR_022263
LSM 980 fluorescent confocal microscope with Airyscan 2	Zeiss	RRID:SCR_025048
MAI TAI 2-photon laser	Spectra-Physics	Model EHP 1040S
Dissecting microscopes (Zeiss Stemi 2000, Nikon SMZ-10A, Olympus SZ60)	Zeiss, Nikon, Olympus	Models Stemi 2000; SMZ-10A; SZ60
Fiber-Lite Mi-150 Illuminator Series, 150 w Halogen light source (MI150 Mi 150)	Dolan-Jenner Industries	Catalog #660000391010
Eppendorf micropipettes	Thermo Fisher	SKU P-10MLR
Epredia Richard-Allan Scientific Cover Glass 22 mm × 22 mm	Thermo Fisher Scientific	Cat. # 22-050-235
Fisherbrand Premium Plain Glass Microscope Slides	Thermo Fisher Scientific	Catalog #12-544-1
LEICA MZ FLIII Fluorescent sorting microscope	LEICA	Model MZ FLIII
Super High Pressure Mercury Lamp Power Supply	Nikon	Model HB-10101AF
Fisherbrand Round Bottom Disposable Borosilicate Glass Tubes with Plain End	Thermo Fisher Scientific	Catalog #14-961-27
Parafilm M Sealing Film	Sigma-Aldrich	Catalog #HS234526B-1EA
Syringe, 10 ml, 5 ml	Thermo Fisher Scientific	Catalog #14-955-458; catalog #14-955-459
Drosophila food	UCI Fly Food Kitchen	N/A
Yeast, active, dry	Fisher Science Education	Catalog #S25632
Welch's 100% Grape Juice	Welch's	WEL35400
Drosophila Incubator	Powers Scientific	https://powersscientific.com/humidified-incubators/
Handheld UV lamp	Maxxeon	MXN02003

Table 1-1

Details of All Statistical Tests Information and details of each statistical test performed in this manuscript, including test type, group comparisons, p values, and post-hoc comparisons. Download Table 1-1, PDF file ^{(444.3KB, pdf)} .

Generation of fly lines and experimental crosses

Crosses were performed in an incubator at 22.5°C and 70% humidity, and eggs were collected on a plate made of grape juice and agarose (grape plate) with yeast paste to synchronize animal age. Cross progeny larvae for experiments were kept on grape plates in an incubator at 22.5°C and 70% humidity until used for injuries. After injury, larvae were individually housed in grape plates with yeast paste and maintained at 20°C and 70% humidity until the last imaging time point (72 h after injury). Both male and female larvae were used for all experiments. Assays to assess pinch or 2p injury and subsequent regeneration in class IV ddaC neurons were performed by crossing fly lines expressing ppk-CD4-tdGFP with Canton-S flies. Assays to assess pinch injury and subsequent regeneration in class I ddaE neurons were performed by crossing fly lines expressing Gal4^2–21 and UAS-tdTomato with Canton-S flies. Assays to assess damage to whole-body ECM were performed by crossing flies expressing ppk-CD4-tdTomato to flies expressing the viking-GFP protein trap. Assays to assess damage to A58-driven ECM were performed by crossing flies expressing ppk-CD4-tdTomato and Gal4^A58 to flies expressing UAS-viking-GFP. In Figure 4 and Extended Data Figure 4-3, assays to assess damage to epidermal cells were performed by crossing flies expressing ppk-CD4-tdTomato and Gal4^A58 to flies expressing UAS-Armadillo-GFP. In Figure 5 and Extended Data Figures 5-1 and 5-2B, assays to assess damage to epidermal cells were performed by crossing flies expressing ppk-CD4-tdTomato to flies expressing armP::Arm-GFP. In Extended Data Figure 5-2A, the assay to assess damage to epidermal cells was performed by crossing flies expressing ppk-CD4-tdTomato and Gal4^GMR38F11 with flies expressing UAS-Armadillo-GFP. Assays to assess damage to glia were performed by crossing flies expressing Gal4^repo and UAS-mRFP with flies expressing ppk-CD4-tdGFP. Assays to assess the immune response following injury (Fig. 4, Extended Data Fig. 4-5) were performed by crossing flies expressing Gal4^pxn and UAS-CD4-tdTomato with flies expressing ppk-CD4-tdGFP. Assays to quantify hemocyte number after pinch (Extended Data Fig. 4-6) were performed by crossing flies expressing ppk-CD4-tdGFP and Gal4^pxn with flies expressing UAS-RedStinger. Assays to assess CaMPARI conversion in class IV ddaC neurons were performed by crossing flies expressing UAS-CaMPARI2.L398T with flies expressing Gal4^ppk and ppk-CD4-tdGFP. Assays to assess CaMPARI conversion in class I ddaE neurons were performed by crossing flies expressing UAS-CaMPARI2.L398T with flies expressing UAS-CD4-tdGFP and Gal4^2–21.

Figure 4.

Pinch injury damages ECM, epidermal cells, and recruits hemocytes, but does not appear to damage glia. A, Schematic of a class IV ddaC neuron (green) with the other cell types and tissues it interacts with: epidermis (gray), glia (blue), ECM (purple), and hemocytes (orange). B, C, E, G, H, ddaC neurons (green) with other tissues in magenta following 2p-full bald (left) or pinch (right) injuries at 72 h AI. Dotted yellow outline highlights the region lacking dendrites due to pinch injury. B, Whole ECM protein trap (vkg-GFP); see Extended Data Figure 4-1 for individual channels. C, A58-driven ECM (Gal4^A58 > vkg-GFP); see Extended Data Figure 4-2 for individual channels and quantification. E, Epidermal tight junction marker (Gal4^A58 > Armadillo-GFP); see Extended Data Figure 4-3 for individual channels and counts of epidermal cells. G, Glial marker (Gal4^repo > CD4-tdTomato); see Extended Data Figure 4-4 for individual channels. H, hemocyte marker (Gal4^pxn > CD4-tdTomato); see Extended Data Figure 4-5 for individual channels. Yellow arrowhead identifies hemocyte accumulation. Scale bar, 100 µm. D, ΔF/F_background of vkg-GFP median fluorescent values following pinch (green) or 2p-full bald (blue) injuries at 24 and 72 h AI. Fluorescent values of pinched neurons are separated by pinch injured and spared sides. Fluorescent values of 2p-full bald neurons are arbitrarily separated into two sides, each representing “one-half injured arbor.” F, Number of epidermal cells near uninjured (purple), pinched (green), or 2p-full bald (blue) neurons at 24 and 72 h AI. Individual channels of whole ECM images are shown in Extended Data Figure 4-1. Individual channels of A58-driven ECM images and quantification are shown in Extended Data Figure 4-2. Individual channels of epidermal cell images and counts are shown in Extended Data Figure 4-3. Individual channels of glial images are shown in Extended Data Figure 4-4. Individual channels of hemocyte images are shown in Extended Data Figure 4-5. Quantification of hemocyte data is shown in Extended Data Figure 4-6.

Figure 5.

Surrounding tissue damage restricts dendrite regeneration. A, C, Half-2p (left side, blue) + half-pinch (right side, green) injured class IV ddaC neurons at 24 and 72 h AI with an ECM protein trap (vkg-GFP; A) or epidermal marker (armP::Armadillo-GFP; C). ddaC only images have been cropped for better visualization of neurons. Scar formation in the ECM at 72 h AI is outlined in white in A. Blue (half-2p) and green (half-pinch) outlines are drawn around each injured half of the class IV ddaC neuron to represent quantified dendrite area coverage (µm²) in B and D. Epidermal cells damaged by either injury paradigm (half-2p, blue; half-pinch, green) are outlined at both 24 and 72 h AI in C. Scale bar, 100 µm. B, D, Dendrite area coverage (µm²) of 2p injured side (blue) and pinch injured side (green) individual class IV ddaC neurons injured by both injury paradigms at 24 and 72 h AI with an ECM Protein Trap (vkg-GFP; B) or epidermal marker (armP::Armadillo-GFP; D). Data highlighting epidermal cell morphology before and after pinch injury is shown in Extended Data Figure 5-1. Data demonstrating improved epidermal cell visualization with armP::Arm-GFP is shown in Extended Data Figure 5-2.

Dendrite injury assays

All injury assays were performed in live, whole-mount larvae.

Pinch injury assay

Please refer to the schematic in Figure 1A for a visual explanation of the experimental flow-through of the pinch injury assay. Please refer to Movie 1 for a recording of a pinch injury and refer to Movie 2 for a close-up movie recording of a pinch injury. To perform the pinch injury assay, groups of 5–6 age-matched 96 h after egg lay (AEL) third instar larvae were collected onto a pad of 4% agarose, and the pad was transferred to a small, empty petri dish. The small petri dish was placed into a larger petri dish for anesthetization. Larvae were anesthetized with isoflurane (∼200 µl) for 45–50 s by placing a folded-up Kimwipe moistened with a small amount of isoflurane in the larger petri dish, exposing the larvae to the isoflurane vapor. After anesthetization, the larvae, sitting on the agarose pad, were transferred out of the petri dish and visualized under a light dissecting microscope. Larval segments (thoracic and abdominal) were counted until reaching abdominal segments 4 and 5. Larvae were rolled ∼45° to expose the right half of the dorsal side of the animal for pinch injury. A pair of blunted #5 forceps was used to pinch the animal. The forceps were held such that the lateral side of the tines of the forceps were what pinched the larvae, not the tips of the forceps. One tine of the forceps was aligned with the anterior side of abdominal hemisegment 4 (A4); one tine of the forceps was aligned with the posterior side of abdominal hemisegment 5 (A5). A slight scooping motion was used to properly pick up and pinch the two hemisegments together: the anterior side of A4 meets the posterior side of A5, with the cuticle region between the forceps popping out, resembling a “muffin top.” (Please see Fig. 1A for a cross-sectional reference of the larval “muffin top.”) By pinching abdominal hemisegments A4 and A5 together, a pinch injury will successfully remove half of the dendritic arbor of the class IV ddaC neuron in hemisegment A4 or A5, while the dendritic arbor of the other neuron will be damaged to a lesser degree. Typically, the lesser-injured neuron is only missing one-quarter of its dendritic arbor or some high-order branches. Occasionally, two successful pinch injuries can occur, wherein both neurons in hemisegments A4 and A5 have half of their dendritic arbors removed, but this was not the typical result observed following a pinch injury. The pinch was held with minimal pressure for 10 s and then released. For CaMPARI experiments, larvae were exposed to 20 s of UV light immediately after pinching. Following pinch injury, individual larvae were then housed in grape plates with yeast paste and maintained at 20°C and 70% humidity.

Figure 1.

Pinch injury damages peripheral sensory neurons in Drosophila larvae. A, Schematic of pinch injury technique. B, Larvae before and during pinch injury. Yellow arrowhead points to larval “muffin top” formed between tines of forceps. C, An uninjured (UI) class IV ddaC neuron at 24 h control and a pinched injured neuron at 24 h after pinch injury. Pinch injury removes half of the dendritic arbor (red-shaded area), leaving the other half spared (remaining green branches). Scale bar, 100 µm. D, Uninjured (UI) within-animal control neurons 24 h control and pinched neurons 24 h after injury (AI), expressing both CD4-tdGFP and CaMPARI, which photoconverts from green (unconverted CaMPARI) to red (converted CaMPARI). Yellow arrowhead points to soma of UI and pinched ddaC neurons. Scale bar, 100 µm for composite channels. Scale bar, 5 µm for inset of cell body in top left corner. E, Quantification of only red photoconverted CaMPARI fluorescence (normalized to background fluorescence) in uninjured (UI) within-animal control neurons (purple) and pinched neurons (green) at 24 h control or AI. F, Same as D, but class I ddaE neurons. G, Same as E, but class I ddaE neurons. Data for timing data for both injury paradigms is detailed in Extended Data Figure 1-1. Data demonstrating reproducibility of the pinch injury across users is detailed in Extended Data Figure 1-2.

Movie 1.

Download video file ^{(13.1MB, mp4)}

DOI: 10.1523/ENEURO.0058-26.2026.video1

Movie of pinch injury method. This movie shows the positioning of the third-instar Drosophila larvae in preparation for pinching. It also shows the alignment of the two tines of the forceps with abdominal hemisegments 4 and 5, along with the slight “scooping” motion required to pinch the two hemisegments together. [View online]

Movie 2.

Download video file ^{(6.1MB, mp4)}

DOI: 10.1523/ENEURO.0058-26.2026.video2

Close-up movie of pinch injury method. This movie shows a close-up version of the pinch injury technique as shown in Movie 1. [View online]

The following information details some common issues that result when the pinch injury is not performed properly along with some troubleshooting tips to help ensure survival of the larvae and reproducibility of the pinch injury. With respect to the amount of pressure to exert during pinch injury, a “minimal” amount of pressure should be exerted. “Minimal” pressure should be enough to close the forceps but not pop the larvae. “Minimal” pressure should feel as if any subtle relaxation of the finger muscles/release of pressure would allow the tines of the forceps to separate. If too much pressure is exerted during pinch injury, a scab may form by the 24 h AI timepoint. This scab will appear reddish brown in color and frequently forms a ring-like structure on the side of the larvae that was injured by pinch. Formation of a scab will make mounting more difficult because the epidermis and cuticle are weak where the scab has formed, and the larvae may pop. Formation of a scab will also make imaging more difficult, because it is not transparent. If larvae are younger than 96 h after egg lay (AEL), they may be too small, and the force of the scooping motion and pressure exerted by pinch may pop them. During pinch injury, if the larvae are left in a puddle of water, they will be too wet and slippery to be grabbed well for pinching. Conversely, if the larvae are too dry, the cuticle may stick to the forceps, and the larvae may be ripped apart when the tines of the forceps are separated. With respect to anesthetization with isoflurane, if the larvae are completely limp and fold in half when picked up following anesthetization, they have been exposed to too much isoflurane and will most likely not survive injury. At the 24 h AI timepoint, if the larvae are dead, white, and opaque, that is another likely indication that the larvae were exposed to isoflurane for too long. Additionally, if the larvae at the 24 h AI timepoint are dead with a splotchy brown coloration, that is a likely indication that too much force was exerted during pinch.

Two-photon injury assay

Larvae were immobilized for mounting by being sandwiched between an agarose pad and coverslip held together with vacuum grease (Thompson-Peer et al., 2016; Duarte et al., 2024; Prange et al., 2024; Hwu et al., 2025). Glycerol was used as the mounting media. Two-photon (2p) injury assays were performed with the Spectra-Physics Mai Tai two-photon tunable laser mounted on the Zeiss LSM980 confocal microscope. Neurons were imaged using 488 nm green or 561 nm red lasers and injured using the bleaching function with the 2p laser at 820 or 860 nm. Exposure to the laser lasted for ∼3 s. For the balding injury assay, the laser was focused on primary dendrite branches near the cell body to remove all branches with the fewest cuts possible without damaging the cell body. For the half-2p injury assay, the laser was focused on primary or secondary dendrite branches to remove half of the branches in the dendritic arbor with the fewest cuts possible. After laser injury, individual larvae were housed in grape plates with yeast paste and maintained at 20°C and 70% humidity. Successful balding injury was assessed by observing dendrite blebbing immediately after injury, comparing the 24 h AI images to the before injury images to assess if the arbors look different, and evaluating territory coverage of new arbors at 24 h AI.

Half-pinch + Half-2p injury assay

Larvae were pinched according to the details listed above in “Pinch Injury Assay.” One to two hours later, larvae were mounted as detailed above and the pinch injured neuron was subjected to 2p laser injury. By 2 h after pinch injury, the blebbing, degradation, and removal of damaged branches were observed before reliably performing a secondary 2p laser injury. All remaining branches were removed. After laser injury, individual larvae were housed in grape plates with yeast paste and maintained at 20°C and 70% humidity.

Imaging

Larvae were immobilized for mounting by being sandwiched between an agarose pad and coverslip held together with vacuum grease (Thompson-Peer et al., 2016; Duarte et al., 2024; Prange et al., 2024; Hwu et al., 2025). Glycerol was used as the mounting media. Images of injured and within-animal uninjured control neurons were taken on a Zeiss LSM700 confocal microscope, Zeiss LSM780 confocal microscope, Zeiss LSM900 confocal microscope, or a Zeiss LSM980 confocal microscope. For Extended Data Figures 5-1 and 5-2 only, animals were imaged 1–2 h before injury. Neurons were imaged at 24 h AI to confirm injury. Neurons were imaged at 72 h AI to observe the end time point of regeneration. Only neurons with obvious survival and absence of half the dendritic arbor were included in subsequent analyses and presented data.

Quantification and statistical analysis

Dendrite arbors were traced using the Simple Neurite Tracer (SNT) plugin in ImageJ to determine the number of dendrite branch tips and the total length of all the dendrite branches (Schindelin et al., 2012). The polygon selection tool in ImageJ was used to select and measure the total area of the dendritic arbor. The multipoint tool in ImageJ was used to count and label hemocytes. For UI control neurons, hemocytes were only included for quantification if they were located under the uninjured arbor. For pinched neurons, hemocytes were only included for quantification if they were located under the “spared” side of the arbor or were present in the area that the arbor would have covered, as if the neuron had been uninjured. Hemocyte quantification was split between “spared” and “injured” sides for pinched neurons and between two sections of “one-half total arbor” for UI controls. These values were determined by dividing the arbors in half, using the axon as a midpoint guideline.

For all graphs comparing means, values are plotted as mean ± standard error of the mean (SEM). For all graphs comparing means, individual data points represent individual neurons, except for Extended Data Figure 1-1A, in which individual data points represent the average time spent performing the pinch or half-2p injury assay. For paired data in Figure 1E,G, values are plotted for individual neurons with faded lines connecting the within-animal uninjured control neurons to each pinched neuron at 24 h AI. Larger, darker-colored points at 24 and 72 h AI represent the mean value. For paired data in Extended Data Figure 4-3B, D, and F, values are plotted for individual neurons with lines connecting repeated measurements of the same neuron at 24 and 72 h AI. For paired data in Figure 2B,C,F,G and Extended Data Figures 3-1B, 3-1C, 4-6B, and 4-6D, values are plotted for individual neurons with faded lines connecting repeated measurements of the same neuron at 24 and 72 h AI. Larger, darker-colored points at 24 and 72 h AI represent the mean value. Solid lines in Figure 2B,C,F,G and Extended Data Figure 3-1B,C represent the slope (m) between the mean 24 and 72 h AI values, which was calculated using the following equation:

$$m=(\mathrm{mean}\mathrm{value}\mathrm{at}72\mathrm{h}\mathrm{AI}-\mathrm{mean}\mathrm{value}\mathrm{at}24\mathrm{h}\mathrm{AI})/48\mathrm{h}.$$

Fold change in branch number was calculated using the following equation:

$$\mathrm{Fold}\mathrm{change}=\mathrm{\#}\mathrm{of}\mathrm{branches}\mathrm{at}72\mathrm{h}\mathrm{AI}/\mathrm{\#}\mathrm{of}\mathrm{branches}\mathrm{at}24\mathrm{h}\mathrm{AI}.$$

To calculate area coverage, we used the polygon selection tool in Fiji and used the axon as a midpoint guideline for which branches to include as part of the injured half of the arbor. Fold change in area coverage was calculated using the following equation:

$$\mathrm{Fold}\mathrm{change}=\mathrm{area}\mathrm{at}72\mathrm{h}\mathrm{AI}/\mathrm{area}\mathrm{at}24\mathrm{h}\mathrm{AI}.$$

To count the epidermal cells, we counted six apodemes on either side of the hemisegment, centered around the neuron's cell body, then drew “lines” between the two top apodemes and the two bottom apodemes. We counted all epidermal cells within the area created.

Figure 2.

ddaC and ddaE neurons grow following pinch injury. A, Class IV ddaC uninjured (UI) within-animal control (purple) and pinched (green) neurons at 24 and 72 h control or AI. Scale bar, 100 µm. B, C, Number of branch tips (top) and total branch length (bottom) of UI neurons (B) and pinch injured class IV ddaC neurons (C) at 24 and 72 h control or AI. D, Fold change in total branch number of UI and pinched class IV ddaC neurons. E, Same as A, but class I ddaE neurons. Scale bar, 100 µm. F, G, Same as B, C, but class I ddaE neurons. H, Same as D, but class I ddaE neurons.

Normalized F (fluorescence) of A58-driven vkg-GFP was calculated using the following equation:

$$\mathrm{Normalized}F=F\mathrm{gray}\mathrm{value}/\mathrm{mean}\mathrm{background}F.$$

If class IV ddaC or class I ddaE neurons expressing CaMPARI experienced photobleaching due to capturing multiple images, they were removed from the dataset. If multiple images had to be captured, the first image captured was used for analysis. Normalized F (fluorescence) of red converted CaMPARI was calculated using the following equation:

$$\mathrm{Normalized}F\mathrm{of}\mathrm{Converted}\mathrm{CaMPARI}=F\mathrm{mean}\mathrm{value}/\mathrm{mean}\mathrm{background}F.$$

ΔF/F_background was calculated using the following equation:

$$\mathrm{\Delta }F/F\mathrm{background}=(\mathrm{median}F\mathrm{at}72\mathrm{hrs}\mathrm{AI}-\mathrm{median}F\mathrm{at}24\mathrm{hrs}\mathrm{AI})/\mathrm{median}\mathrm{background}F.$$

Sample sizes for all figures are as follows. Figure 1E uninjured control n = 4, pinch n = 4; Figure 1G uninjured control n = 10, pinch n = 10; Extended Data Figure 1-1A 2p n = 3, pinch n = 3; Figure 2B n = 10; Figure 2C n = 11; Figure 2D uninjured control n = 10, pinch n = 11; Figure 2F n = 14; Figure 2G n = 16; Figure 2H uninjured control n = 14, pinch n = 16; Figure 3C pinch n = 11, 2p n = 13; Figure 3D pinch n = 11, 2p n = 13; Figure 3G n = 9; Figure 3H n = 9; Extended Data Figure 3-1B n = 12; Extended Data Figure 3-1C n = 13; Extended Data Figure 3-1D uninjured control n = 12, 2p n = 13; Figure 4B 2p n = 8, pinch n = 3; Figure 4C 2p n = 6, pinch n = 4; Figure 4D 2p at 24 h AI n = 6, 2p at 72 h AI n = 6, pinch at 24 h AI n = 7, pinch at 72 h AI n = 4; Figure 4E 2p n = 4, pinch n = 4; Figure 4F uninjured control n = 5, 2p n = 4, pinch n = 4; Figure 4G 2p n = 3, pinch n = 2; Figure 4H 2p n = 5, pinch n = 7; Extended Data Figure 4-1A n = 7; Extended Data Figure 4-1B n = 3; Extended Data Figure 4-1C n = 8; Extended Data Figure 4-2A and B uninjured control at 24 h AI n = 8, uninjured control at 72 h AI n = 6; Extended Data Figure 4-2C and D pinch at 24 h AI n = 7, pinch at 72 h AI n = 4; Extended Data Figure 4-2E uninjured control at 24 h AI n = 5, pinch at 24 h AI n = 7, uninjured control at 72 h AI n = 3, pinch at 72 h AI n = 4; Extended Data Figure 4-2F 2p at 24 h AI n = 6, 2p at 72 h AI n = 6; Extended Data Figure 4-2G uninjured control at 24 h AI n = 3, 2p at 24 h AI n = 6, uninjured control at 72 h AI n = 3, 2p at 72 h AI n = 6; Extended Data Figure 4-3A n = 5; Extended Data Figure 4-3B n = 5; Extended Data Figure 4-3C n = 4; Extended Data Figure 4-3D n = 4; Extended Data Figure 4-3E n = 4; Extended Data Figure 4-3F n = 4; Extended Data Figure 4-3G uninjured control n = 1, pinch n = 1; Extended Data Figure 4-4A n = 5; Extended Data Figure 4-4B n = 2; Extended Data Figure 4-4C n = 3; Extended Data Figure 4-5A n = 11; Extended Data Figure 4-5B n = 7; Extended Data Figure 4-5C n = 5; Extended Data Figure 4-6A and B n = 8; Extended Data Figure 4-6C and D n = 8; Extended Data Figure 4-6E uninjured control at 24 h AI n = 8, pinch at 24 h AI n = 8, uninjured control at 72 h AI n = 8, pinch at 72 h AI n = 8; Figure 5A and B n = 4; Figure 5C and D n = 3; Extended Data Figure 5-1A n = 4; Extended Data Figure 5-1B n = 4; Extended Data Figure 5-2A n = 1; Extended Data Figure 5-2B n = 4.

Figure 3.

ddaC neurons preferentially regenerate into empty territory following 2p injury. A, Schematics of pinch (top) and half-2p bald (bottom) injuries. B, Pinch (top) and half-2p bald (bottom) injured class IV ddaC neurons at 24 and 72 h AI. Black boxes identify empty territory. Scale bar, 100 µm. C, Number of branch tips (left) and total branch length (right) of class IV ddaC neurons injured by 2p (blue) or pinch (green). D, Fold change in area coverage of the injured half only of individual neurons following 2p or pinch injury. E, Schematic of a “half-pinch + half-2p” injury on the same class IV ddaC neuron. F, Half-2p (left side) + half-pinch (right side) injured class IV ddaC neuron at 24 and 72 h AI. Scale bar, 100 µm. G, Number of branch tips (left) and total branch length (right) of 2p injured side (blue) and pinch injured side (green) of individual class IV ddaC neurons at 24 and 72 h AI that were injured by both injury paradigms. H, Area coverage (µm²) of 2p injured side (blue) and pinch injured side (green) of individual class IV ddaC neurons at 24 and 72 h AI that were injured by both injury paradigms. Regeneration data following half-2p bald is shown in Extended Data Figure 3-1.

All sample sizes for paired graphs of 24 to 72 h have the same n and are analyzing the same neurons at both time points, except for the datasets collected for the A58-driven ECM (a few of which are missing the 72 h after injury time point). Nearly all injured neurons have a within-animal uninjured control neuron.

If necessary, image processing using the freehand selection tool in Fiji was performed to remove denticle belts in order to better visualize the dendrites beneath/near those structures. All images displayed are representative of the phenotype. Representative images were selected for their clarity.

Some schematics were created with BioRender.

Statistical analysis was performed using GraphPad Prism software (version 10.3.1). An unpaired Student's t test was used for analyses comparing two groups when both groups passed a D'Agostino–Pearson omnibus normality test. A two-tailed Welch's Student's t test was used for all analyses comparing two groups when one or both groups did not pass a D'Agostino–Pearson omnibus normality test. A paired Student's t test was used for all analysis comparing the same neuron at two time points. A one-way ANOVA with Holm–Šidák's multiple-comparisons correction was used for analyses comparing more than two groups, and a two-way ANOVA with Holm–Šidák's multiple-comparisons correction was used for analyses with more than two variables. For all statistical tests, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and ^nsp > 0.05.

Statistical tests for all figures are as follows. Details of each test are included in Extended Data Table 1-1. Figure 1E: paired t test; Figure 1G: paired t test; Extended Data Figure 1-1A: unpaired t test; Figure 2B: number of branch tips = paired t test, total branch length = paired t test; Figure 2C: number of branch tips = paired t test, total branch length = paired t test; Figure 2D: unpaired t test; Figure 2F: number of branch tips = paired t test, total branch length = paired t test; Figure 2G: number of branch tips = paired t test, total branch length = paired t test; Figure 2H: unpaired t test with Welch's correction; Figure 3C: number of branch tips = ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test, total branch length = ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test; Figure 3D: unpaired t test; Figure 3G: number of branch tips = ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test, total branch length = ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test; Figure 3H: ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test; Extended Data Figure 3-1B: number of branch tips = paired t test, total branch length = paired t test; Extended Data Figure 3-1C: number of branch tips = paired t test, total branch length = paired t test; Extended Data Figure 3-1D: unpaired t test with Welch's correction; Figure 4D: two-way ANOVA with Holm–Šidák's multiple-comparisons test; Figure 4F: ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test; Extended Data Figure 4-2E: two-way ANOVA with Holm–Šidák's multiple-comparisons test; Extended Data Figure 4-2G: two-way RM ANOVA with Holm–Šidák's multiple-comparisons test; Extended Data Figure 4-3B: paired t test; Extended Data Figure 4-3D: paired t test. Extended Data Figure 4-3F: paired t test; Extended Data Figure 4-6B: one-half total arbor = paired t test, one-half total arbor = paired t test, total arbor = paired t test; Extended Data Figure 4-6D: spared side = paired t test, injured side = paired t test, spared and injured sides = paired t test; Extended Data Figure 4-6E: 24 h AI = two-way ANOVA with Holm–Šidák's multiple-comparisons test, 72 h AI = two-way ANOVA with Holm–Šidák's multiple-comparisons test; Figure 5B: ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test; Figure 5D: ordinary one-way ANOVA with Holm–Šidák's multiple-comparisons test.

Data availability

The datasets generated and/or analyzed during the current study are available from the lead contact upon reasonable request. Traces of dendrite arbors will be uploaded to Neuromorpho.org and available there. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Results

Pinch injury damages da PNS neurons in D. melanogaster larvae

To develop an injury model that better mimics widespread tissue damage, we adapted a pinch injury method established by Burra et al. (2013). We modified the injury method such that cuticular puncture did not occur (Fig. 1A). To do this, we blunted a pair of #5 forceps and pinched the tissue laterally so that the larva's cuticle and epithelia bulged out between the tines of the forceps, which we termed a “larval muffin top” (Fig. 1A,B). The pinch injury was held for 10 s and typically lifted the larvae up into the air (Fig. 1B; Movies 1, 2). Compared with the traditional dendrite severing by a 2-photon (2p) laser, our new pinch injury method was significantly faster to perform (Extended Data Fig. 1-1A; 17.2 min per injury for 2p vs 4.2 min per injury for pinch). Throughout this manuscript, we will compare pinch-injured neurons to uninjured neurons and/or 2p-injured neurons. As such, representative images and data from pinch-injured neurons will be presented in green, uninjured neurons will be presented in purple, and 2p-injured neurons will be presented in blue. All images of neurons will be oriented such that the axon is pointing down and the region damaged by pinch will always be on the right side of the neuron/image.

Figure 1-1

Pinch Injury Technique Is Faster to Perform Than 2p Dendrite Severing. (A) Minutes required to perform injury to 1 neuron per animal for 2-photon (2p) injury assay (blue) and pinch injury assay (green). Download Figure 1-1, TIF file ^{(141.9KB, tif)} .

Figure 1-2

Pinch Injury Technique Is Reproducible Across Different Users. (A) Examples of pinch injury at 24 hrs after injury (AI), performed by three different individuals. Images shown are all class IV ddaC neurons expressing only CD4-tdGFP. Scale bar 100 µm. Download Figure 1-2, TIF file ^{(2.2MB, tif)} .

Following pinch injury, we first examined the extent of damage sustained by the PNS dendritic arborization (da) sensory neurons. We visualized the class IV da neuron, ddaC, using cell type-specific expression of a membrane-tagged GFP and imaged them at 24 h after pinch injury. We found that abdominal segment ddaC neuron A4 or A5 was successfully injured by our approach: one lateral half of the dendritic arbor was damaged, and the other half was left intact (Fig. 1C). Typically, we observed that two neurons were damaged by pinch injury. Most often, one neuron was successfully pinched, wherein half of the dendritic arbor had been removed, while the other neuron was damaged to a lesser extent, missing only part of the dendritic arbor. We did occasionally observe successful pinch injuries to neurons in both hemisegments A4 and A5, but this was not typical. Importantly, we found this pinch technique to be reproducible across different users (Extended Data Fig. 1-2A; examples of pinch injuries performed by three different persons). Figure 1C identifies the pinch-injured region (red-shaded area).

Next, we set out to determine whether pinch injury led to calcium influx, the first step of dendrite injury detection in da neurons, as we and others have shown previously (Duarte et al., 2024; Hertzler et al., 2024). We used CaMPARI (Calcium-Modulated Photoactivatable Ratiometric Indicator), a photoactivatable calcium indicator that photoconverts from green to red only in the simultaneous presence of calcium ions and UV light (Fosque et al., 2015). Immediately following pinch injury, larvae expressing CaMPARI in ddaC neurons were exposed to 20 s of UV light. Red converted CaMPARI fluorescence was detected in the soma and proximal primary dendrites in pinched ddaC neurons, but not in uninjured (UI) control neurons (Fig. 1D; Extended Data Table 1-1 details each statistical test performed and should be referred to for all figures presented in this manuscript). When we quantified the increase in red photoconverted CaMPARI fluorescence, pinched ddaC neurons had a significantly higher red fluorescence intensity of converted CaMPARI compared with uninjured controls (Fig. 1E). Class IV ddaC neurons function as nociceptors, involving a transient influx of Ca²⁺ into the dendrite as a response to pain and/or harsh touch (Hwang et al., 2007; Terada et al., 2016). To determine if the calcium influx detected by CaMPARI is due to injury or nociception, we repeated these pinch injuries with CaMPARI in the non-nociceptive class I ddaE proprioceptive neurons (Hughes and Thomas, 2007; Song et al., 2007; Cheng et al., 2010; He et al., 2019). Red converted CaMPARI fluorescence was detected in the soma of pinched class I ddaE neurons, but not uninjured control neurons (Fig. 1F). Quantification of the increase in red photoconverted CaMPARI fluorescence demonstrated that pinched ddaE neurons had a significantly higher red fluorescence intensity of converted CaMPARI compared with UI controls (Fig. 1G). These results demonstrate that Ca²⁺ influx due to pinch injury occurs in both proprioceptive and nociceptive neurons, strongly supporting the interpretation that Ca²⁺ influx is a response to dendrite injury rather than to nociceptive sensory activation. We conclude that the pinch technique causes calcium influx due to dendrite injury in these neurons, which is a critical step in dendrite injury detection and is similar to what has been observed immediately following 2p injury (Duarte et al., 2024; Hertzler et al., 2024).

Together, these results demonstrate that pinching injury can damage dendrites, in a highly reproducible and rapid manner.

ddaC and ddaE neurons (re)grow their branches following pinch injury

To assess how da neuronal dendrites respond to pinch injury, we first decided to evaluate the dendritic arbors of the class IV ddaC neurons. The dendritic arbors of class IV ddaC neurons are large, filling their receptive field by 48 h after egg lay (AEL) and are very dynamic (Grueber et al., 2002; Parrish et al., 2009). Further, they retain a robust capacity for dendrite regeneration (Song et al., 2012; Stone et al., 2014; Thompson-Peer et al., 2016; Hertzler et al., 2023; Duarte et al., 2024; Prange et al., 2024; Hwu et al., 2025). Pinch injury damaged half of the dendrite arbor and left the other half uninjured (Fig. 2A). Following pinch injury, the spared half of the arbor became visibly bushier as it gained branches over time (Fig. 2A). Specifically, ddaC neurons increased in total branch number and total dendrite length from 24 to 72 h after injury (AI; Fig. 2A,C). The growth on the spared half of the arbor was roughly similar to uninjured ddaC neurons (Fig. 2A,B), which continue to gain branches over time as the larvae grow and develop (Parrish et al., 2009).

In order to determine if the branches gained on the spared half of the pinched neuron were stimulated by injury or were simply the branch growth of normal development, we took two approaches. First, we looked at class I ddaE neurons, which are neurons that establish their arbors by 24 h AEL and are very static during later development (Sugimura et al., 2003). In the absence of injury, they do not change in branch number over time, though they do increase in branch length as the larvae grow (Fig. 2E,F). Thus, any significant increase in branch number following pinch injury would be indicative of regeneration. We found that pinched class I ddaE neurons did increase in total branch number and total dendrite length from 24 to 72 h after pinch, compared with the stable uninjured class I ddaE neurons that did not increase in total branch number during this time (Fig. 2E,F,G). Because asymmetric ddaE neurons lack a spared side, the injured arbor is the total dendrite arbor. As a second approach, we evaluated the growth rate of each neuron subtype, both class IV ddaC and class I ddaE neurons. We found that both had a greater fold change in total branch number (total branch number at 72 h AI divided by total branch number at 24 h AI) following pinch injury compared with their uninjured controls (Fig. 2D,H).

This result coupled with the regenerative response of class I ddaE neurons following pinch injury is indicative that both class IV ddaC and class I ddaE neurons exhibit regeneration of their dendrites following pinch injury and not simply continued developmental growth.

ddaC neurons preferentially regenerate into undamaged tissue

Next, we wanted to compare regeneration following pinch injury versus 2p injury, as dendrite severing by a high-powered laser is the “gold standard” for precise injury in the field (Hertzler and Rolls, 2024). To compare the two injury methods, we mimicked the amount of dendrite removal obtained with a pinch injury by using a 2p laser to cut off half of the class IV dendritic arbor (Fig. 3A,B; Extended Data Fig. 3-1A). We called this injury a “half-2p bald” as half of the neuron was left “bald” by removal of half of its dendrite arbor. Class IV ddaC neurons regenerated following this half-2p bald laser injury. They increased their total branch number and total branch length, and they had a significantly higher fold change in total branch number compared with their UI controls (Fig. 3B; Extended Data Figs. 3-1B–3-1D).

Figure 3-1

Half-2p Bald class IV ddaC Neurons Grow Into Undamaged Tissue. (A) Class IV ddaC uninjured (UI) within-animal control (purple) and half-2p (blue) injured neurons at 24 and 72 hrs control or AI. Scale bar 100 µm. (B-C) Number of branch tips (top) and total branch length (bottom) of uninjured (UI) within-animal control (B, purple) and half-2p bald (C, blue) neurons 24 to 72 hrs control or AI. (D) Fold change in total branch number of uninjured (UI) within-animal control (purple) and half-2p bald (blue) class IV ddaC neurons. Download Figure 3-1, TIF file ^{(4.1MB, tif)} .

When comparing the total number of branches and total dendrite length between pinched and half-2p bald class IV ddaC neurons, there were no significant differences at 24 or 72 h AI in the total number or length of dendrites (Fig. 3C). We noticed, however, that pinch injury resulted in fewer dendrites innervating the empty territory on the injured side relative to half-2p bald neurons (Fig. 3B, black boxes). Since this phenotype was not captured by calculating the total number of branches or total dendrite length (Fig. 3C), we therefore calculated the fold change in area coverage of only the injured side of the arbor. Neurons injured by half-2p bald experienced a greater fold change in area coverage on their injured side, reflecting our observation of an increased capacity for innervating empty territory following injury (Fig. 3D).

Given this difference in regenerative capacity, we next wanted to challenge individual neurons to regenerate their dendrites following both pinch and 2p injury. Class IV ddaC neurons were first pinched and 1–2 h later, the remaining arbor (the other half) was cut off with the 2p laser. This specific “half-pinch + half-2p” injury paradigm removed all branches from the class IV ddaC neuron and forced new dendrites to choose to regrow on one side or the other (Figs. 3E,F). For all quantifications of this injury paradigm, the axon was used as a midpoint guideline to separate the two halves of the arbor. Following this half-pinch + half-2p injury, there was no statistical difference between the two sides in branch number, dendrite length, or area coverage at 24 h AI (Fig. 3G,H). However, at 72 h AI, the half of the arbor responding to 2p injury regenerated significantly more than the half of the arbor responding to pinch injury, with respect to branch number, dendrite length, and area coverage (Fig. 3G,H).

Taking the results of these two experiments together, we conclude that ddaC neurons preferentially regenerate into the empty territory that had been injured by 2p injury. Since challenging individual neurons to regenerate following both injury paradigms led to differential regenerative outcomes, we hypothesized that extrinsic mechanisms may be causing poor reinnervation following pinch injury, either by active inhibition or absence of support.

Pinch injury damages other cell types and surrounding tissues

Following the differential regeneration of dendrites observed from the half-pinch + half-2p experiments, we next evaluated how the extracellular environment might be affecting regeneration following both injury paradigms. We examined the effects of 2p and half-pinch injury individually on the surrounding tissues (Fig. 4) and then examined tissue damage after combined half-pinch + half-2p injury together (Fig. 5). There are a number of cell types and tissues that interact with the da neurons (Fig. 4A). The da neuronal dendrites are suspended in an extracellular matrix (ECM) and innervate the larval epidermis. Glia wrap the axon, cell body, and proximal dendrites of the da neurons, and resident immune cells, hemocytes, are present. First, we examined the ECM.

Figure 4-1

Pinch Injury, but Not 2p-Full Bald Injury, Damages Whole ECM (viking/Collagen IV). (A-C) Uninjured (A), pinched (B), or 2p-full bald (C) class IV ddaC neurons with an ECM Protein Trap (vkg-GFP) at 24 and 72 hrs control or AI. Dotted yellow outlines highlight the region lacking dendrites due to pinch injury in (B). Scale bar 100 µm. Composite images of pinched or 2p-full bald neurons at 72 hrs AI are also shown in Fig 4B. Download Figure 4-1, TIF file ^{(14.8MB, tif)} .

Figure 4-2

Pinch Injury, but Not 2p-Full Bald Injury, Damages A58-Driven ECM (viking/Collagen IV). (A, C, F) Uninjured (A), pinched (C), or 2p-full bald (F) class IV ddaC neurons with A58-driven ECM marker (Gal4^A58 > vkg-GFP) at 24 and 72 hrs control or AI. Dotted yellow outlines highlight the region lacking dendrites due to pinch injury in (C). Composite images of pinched or 2p-full bald neurons at 72 hrs AI are also shown in Fig 4C. Scale bar 100 µm. (B, D) Normalized F of vkg-GFP over 250 µm starting from the cell body at 24 hrs control or AI (top) and 72 hrs control or AI (bottom). UI neurons (B) were arbitrarily separated into two sides, each representing “one-half total arbor”. Pinched neurons (D) were separated by spared (light green) and pinched (dark green) sides. (E, G) ΔF/F_background of vkg-GFP median fluorescence values at 24 and 72 hrs control or AI in uninjured (purple) or pinched (green) neurons (E), or uninjured (purple) and 2p-full bald (blue) neurons (G). Download Figure 4-2, TIF file ^{(16.9MB, tif)} .

Figure 4-3

Pinch Injury, but Not 2p-Full Bald Injury, Disrupts Epidermal Cell Morphology. (A, C, E) Uninjured (A), pinched (C), or 2p-full bald (E) class IV ddaC neurons with epidermal marker (Gal4^A58 > UAS-Armadillo-GFP) at 24 and 72 hrs control or AI. Epidermal cells included for quantification are outlined in yellow. Composite images of pinched or 2p-full bald neurons at 72 hrs AI are also shown in Fig 4E. Scale bar 100 µm. (B, D, F) Number of epidermal cells in hemisegment at 24 and 72 hrs control or AI near uninjured (B), pinched (D), or 2p-full bald (F) neurons. Comparison among uninjured and injured neurons and at each time point is also shown in Fig 4F. (G) Brightfield images of the cuticle above an uninjured neuron at 72 hrs control or pinched neuron at 72 hrs AI. Download Figure 4-3, TIF file ^{(20.2MB, tif)} .

Figure 4-4

Neither Pinch Nor 2p-Full Bald Injuries Seem to Dramatically Damage Glia. (A-C) Uninjured (A), pinched (B), or 2p-full bald (C) class IV ddaC neurons with glial marker (Gal4^repo > mRFP) at 24 and 72 hrs control or AI. Dotted yellow outlines highlight the region lacking dendrites due to pinch injury in (B). Scale bar 100 µm. Composite images of pinched or 2p-full bald neurons at 72 hrs AI are also shown in Fig 4G. Download Figure 4-4, TIF file ^{(8.9MB, tif)} .

Figure 4-5

Pinch Injury, but Not 2p-Full Bald Injury, Recruits Hemocytes. (A-C) Uninjured (A), pinched (B), or 2p-full bald (C) class IV ddaC neurons with hemocyte marker (Gal4^pxn > CD4-tdTomato) at 24 and 72 hrs control or AI. In the hemocyte channel for the uninjured neurons at 24 and 72 hrs control and for the 2p-injured neurons at 24 and 72 hrs AI, the top yellow box identifies the dorsal vessel cluster, the middle yellow box identifies the dorsal stripe, and the bottom yellow box identifies the lateral patch of hemocytes. For pinch injured neurons at 24 and 72 hrs AI, yellow arrowheads identify hemocyte accumulation. Dotted yellow outlines highlight the region lacking dendrites due to pinch injury in (B). Scale bar 100 µm. Composite images of pinched or 2p-full bald neurons at 72 hrs AI are also shown in Fig 4H. Download Figure 4-5, TIF file ^{(11.4MB, tif)} .

Figure 4-6

Pinch Injury Induces a Robust and Prolonged Immune Response. (A, C) Uninjured (A) or pinched (C) class IV ddaC neurons with hemocyte marker (Gal4^pxn > RedStinger) at 24 and 72 hrs control or AI. In the hemocyte channel for the uninjured neurons at 72 hrs control, the top yellow box identifies the dorsal vessel cluster, the middle yellow box identifies the dorsal stripe, and the bottom yellow box identifies the lateral patch of hemocytes. Dotted yellow outlines highlight the region lacking dendrites due to pinch injury in (C). Scale bar 100 µm. (B, D) Number of hemocytes at 24 and 72 hrs control or AI within the two halves (left and middle) or total (right) region of the hemisegment of uninjured (B) or the injured side (left), spared side (middle), or total (right) region of the hemisegment of pinched (D) neurons. (E) Number of hemocytes within the hemisegment of uninjured (purple) or pinched (green) neurons at 24 (top) and 72 (bottom) hrs control or AI. Download Figure 4-6, TIF file ^{(11.3MB, tif)} .

Figure 5-1

Epidermal Cells Appear to Undergo Fusion Following Pinch Injury. (A, B) Uninjured (A) or pinched (B) class IV ddaC neurons with epidermal marker (armP::Armadillo-GFP) before injury and 24 hrs control or AI. Epidermal cells are outlined in yellow. Dotted yellow outlines highlight the region lacking dendrites due to pinch injury in (B) at 24 hrs AI. Scale bar 100 µm. Download Figure 5-1, TIF file ^{(12.1MB, tif)} .

Figure 5-2

Arm-GFP Under Control of arm Promoter Labels Epidermal Cells Better Than UAS-controlled Arm-GFP Before and After Injury. (A, B) Class IV ddaC neurons with epidermal marker (Armadillo-GFP) under control of the Gal4-UAS system (A) or arm promoter (B) before injury and 24 hrs AI. Epidermal cells are outlined in yellow. Dotted yellow outlines highlight the region lacking dendrites due to pinch injury at 24 hrs AI. Scale bar 100 µm Download Figure 5-2, TIF file ^{(11.6MB, tif)} .

Throughout development, the larval da neurons adhere to components of the ECM through adhesion molecules like integrins (Kim et al., 2012). These interactions help constrain dendrites to a 2D space: within the ECM and between the musculature and epidermis (Grueber et al., 2002; Yasunaga et al., 2010; Han et al., 2012; Kim et al., 2012). Major components of the ECM include laminins and collagen IV, which are primarily produced by the fat body and hemocytes but also by the larval epidermis (Rodriguez et al., 1996; Han et al., 2012; Isabella and Horne-Badovinac, 2015; Ramos-Lewis et al., 2018).

Thus, to visualize the ECM, we chose to look at viking (vkg), which is homologous to the mammalian Collagen IV (Yasothornsrikul et al., 1997). We first used a protein trap fly line where the endogenous vkg gene is tagged with GFP. In uninjured hemisegments, vkg-GFP predominantly wrapped the muscle and muscle attachment sites; it also wrapped the axon, cell body, and proximal dendrites of the ddaC class IV neurons (Extended Data Fig. 4-1A). Following 2p injury, vkg-GFP was undamaged (Fig. 4B, Extended Data Fig. 4-1C). Following pinch injury, vkg-GFP was severely damaged (Fig. 4B, Extended Data Fig. 4-1B). The vkg-GFP wrapping the muscle and muscle attachment sites was significantly damaged, and there was no longer a consistent vkg-GFP signal across the hemisegment. Vkg-GFP continued to wrap the axon, cell body, and proximal dendrites of the spared half of the pinched arbor. However, there was a significant accumulation of vkg-GFP signal on the pinched half of the arbor that aligned with the pinched edge of the injured side of the arbor. This accumulation of vkg-GFP was reminiscent of a glial or fibrotic scar, which can create an inhibitory, nonpermissive environment for neurite regrowth (Liesi and Kauppila, 2002; Klapka and Müller, 2006; Tran et al., 2022).

Since the whole-body vkg-GFP had a significant amount of signal, mostly surrounding the muscle, it was difficult to see the ECM directly interacting with the dendrites. Thus, we sought to visualize vkg-GFP specifically expressed by A58-Gal4, which drives expression in the larval epidermis and, in later third instar stages, the fat body (Ramos-Lewis et al., 2018). Uninjured hemisegments had a smooth, even vkg-GFP fluorescence (F) signal that mimics the “cobblestone” pattern of the epidermis (Extended Data Fig. 4-2A, quantified in Extended Data Fig. 4-2B). The A58-driven vkg-GFP appeared unaltered by 2p injury (Fig. 4C, Extended Data Fig. 4-2F, quantified in Extended Data Fig. 4-2G). Thus, the A58-driven vkg-GFP was uninjured following 2p injury. When we compare pinched animals versus 2p injured animals (Fig. 4D) or versus uninjured animals (Extended Data Fig. 4-2E), we find that A58-driven vkg-GFP was enriched on the pinched side of the arbor compared with either half of an uninjured neuron or a neuron injured by 2p (Extended Data Fig. 4-2C). The enrichment of A58-driven vkg-GFP on the pinched half of the neuron increased from 24 to 72 h AI (Extended Data Fig. 4-2D,E). This was reminiscent of the enrichment of whole-body vkg-GFP observed in Figure 4B. As an additional note, we observed a slight increase in the A58-driven vkg-GFP fluorescence on the spared half at 72 h AI, most likely due to diffusion of vkg-GFP from the pinched side (Fig. 4D, Extended Data Fig. 4-2D,E). Thus, both endogenous and over-expressed viking accumulated and persisted at the site of injury, only after pinch injury.

Next, we wanted to look at the larval epidermis. The da neurons establish their arbors in a 2D space: within the ECM and between the musculature and epidermis (Grueber et al., 2002; Yasunaga et al., 2010; Han et al., 2012; Kim et al., 2012). The larval epidermis is a monolayer sheet of terminally differentiated epidermal cells and plays a crucial role in establishing the dendritic arbor of the da neurons (Parrish et al., 2009; Burra et al., 2013; Jiang et al., 2014; Poe et al., 2017). HSPGs expressed by the larval epidermis are required for stabilization of high-order dendrites in ddaC class IV neurons (Poe et al., 2017). Additionally, portions of ddaC class IV dendrite branches are enclosed by epidermal cells via Coracle to restrict dendrite outgrowth and mediate the occupancy of multiple classes of da neurons in a 2D space (Tenenbaum et al., 2017). To visualize the epidermis, we used a GFP-tagged version of Armadillo, the fly homolog of beta-catenin and an adherens-junction marker (Cox et al., 1996). Armadillo-GFP clearly outlined individual epidermal cells, and we additionally outlined epidermal cells included for quantification in yellow to highlight their morphology (Extended Data Figs. 4-3, 5-1). For uninjured hemisegments, the epidermis was a single sheet of polygonal epidermal cells (Extended Data Figs. 4-3A, 5-1A), and we found no change in the number of epidermal cells from 24 to 72 h control (Extended Data Fig. 4-3B). Following 2p injury, the epidermis appeared the same as uninjured hemisegments and did not change over time (Fig. 4E,F; Extended Data Fig. 4-3E,F). However, following pinch injury, the epidermis was damaged (Fig. 4E; Extended Data Figs. 4-3C, 5-1B, 5-2B). On the pinched side of the neuron, the epidermal cells were disfigured: close to the focal point of injury, they were small and crushed; next to the muscle attachment sites (apodemes), the epidermal cells were elongated and larger than normal. There were significantly fewer epidermal cells following pinch injury than there were for uninjured neurons (Fig. 4F). Even though we found damage to the epidermal cells following pinch injury, there was no damage to the overlying cuticle (Extended Data Fig. 4-3G). We examined the morphology of the epidermal cells before and after pinch injury (Extended Data Figs. 5-1, 5-2) and found that epidermal cells on the pinched side of the hemisegment could not be easily tracked or identified between images taken before injury (96 h AEL) and after injury (24 h AI; Extended Data Figs. 5-1B, 5-2B). On the spared side of a pinched hemisegment, the epidermal cells appeared uninjured, similar to uninjured and 2p-injured hemisegments, and did not change over time (Fig. 4E; Extended Data Figs. 4-3D, 5-1, 5-2B). Damage to epidermal cells may contribute to why pinch-injured neurons exhibit poor regrowth into empty territory.

Next, we looked at the glia. Dendritic arborization neurons have three layers of glia that wrap the axon, the soma, and the proximal region of the primary dendrites (Extended Data Fig. 4-4A; Yadav et al., 2019). Glia are important for establishing both the structure and function of ddaC class IV neurons. Following 2p injury, the glia were capable of wrapping the axon and soma (Thompson-Peer et al., 2016), but they seemed to have trouble wrapping the proximal segments of the newly regenerated primary dendrites, based on qualitative appearance (Fig. 4G, Extended Data Fig. 4-4C). Following pinch injury, we saw no obvious signs of direct injury to the glia (Fig. 4G, Extended Data Fig. 4-4B). On the spared side of the arbor, the glia appeared normal; on the pinched side of the arbor, the glia struggled to wrap the regenerated primary dendrites. The glia could, however, still wrap the axon and soma following pinch injury. Thus, glia appeared to not be directly damaged by either injury paradigm.

Lastly, we looked at the immune response following both injury paradigms. There are three types of immune cells (hemocytes) in Drosophila: plasmatocytes, crystal cells, and lamellocytes (Rizki, 1957; Rizki and Rizki, 1983, 1992; Tepass et al., 1994; Lebestky et al., 2000). Approximately 95% of all circulating hemocytes are plasmatocytes, and the remaining 5% are crystal cells. Lamellocytes are only stimulated by parasitism (Rizki and Rizki, 1992; Meister and Lagueux, 2003). While most hemocytes migrate through the hemolymph and have no specific localizations, in abdominal segments 1–7 (in which these injury experiments were performed), there are some groups of localized hemocytes in the epidermal-muscular pocket (Makhijani et al., 2011). Proper localization and survival of these resident hemocytes is supported by the da neurons (Makhijani et al., 2011).

These resident hemocytes include the dorsal vessel cluster (near the top of the ddaC class IV dendritic arbor), the dorsal stripe (near the soma), and the lateral patch (which extends from the bottom of the ddaC class IV dendritic arbor into the neighboring v'ada class IV arbor; Makhijani et al., 2011). These three patches of hemocytes were visible in uninjured hemisegments (yellow boxes in Extended Data Figs. 4-5A, 4-6A). Quantification of resident hemocytes near uninjured neurons indicated that there was no significant difference between the number of hemocytes on either side of the arbor or the total arbor at 24 or 72 h control (Extended Data Fig. 4-6B). Following 2p injury, there was no major accumulation of hemocytes near the regenerating arbor, and the three patches of hemocytes appeared normal (Fig. 4H, Extended Data Fig. 4-5C; yellow boxes in Extended Data Fig. 4-5C highlight the presence of normal resident hemocyte groups). Following pinch injury, however, there was major accumulation of hemocytes to the pinched hemisegment, on both spared and pinched sides of the ddaC class IV arbor (Fig. 4H; Extended Data Figs. 4-5B, 4-6C). There were no longer three discrete patches of hemocytes; instead, there was a large accumulation of hemocytes near the focal point of injury (Fig. 4H, Extended Data Fig. 4-5B). Quantification of the hemocyte influx into the pinched hemisegment showed that there was a significant increase in the number of hemocytes following pinch injury compared with uninjured neurons at both 24 and 72 h AI (Extended Data Fig. 4-6E), on both the pinched and spared half of the arbor or the total arbor area. When we looked at the count of hemocyte number over time, the number of hemocytes did not change from 24 to 72 h AI for the spared side of the pinched arbor or the total arbor (Extended Data Fig. 4-6D). There was a significant decrease in the number of hemocytes on the injured side of the pinched arbor from 24 to 72 h AI. The reduction of hemocytes over time on the injured side of a pinched arbor could indicate the start of a resolution of the immune response. Interestingly, resolution of an immune response following pinch injury has been reported to conclude by 24 h after wounding, which is faster than what we observed (Babcock et al., 2008; Burra et al., 2013). Overall, this influx of hemocytes to the injured hemisegment pointed toward a large immune reaction following the widespread tissue damage caused by pinch injury, representing another possible reason for poor dendrite regrowth into empty territory.

Damage to the ECM and epidermis restrict dendrite regeneration

To demonstrate that the discrete regenerative responses observed following the half-pinch + half-2p assay (Fig. 3) were, in fact, due to injury-specific differences in the surrounding tissue environment, we repeated the half-pinch + half-2p assay with the ECM viking-GFP protein trap or the Armadillo-GFP epidermal marker (Fig. 5). Regenerative outcomes following half-pinch + half-2p injury with either extracellular marker mirrored the results seen in Figure 3, where there was substantially more dendrite regrowth into the empty territory on the side of the neuron where the 2p-injury had been performed (Fig. 5A,C). There was a significant increase in the area coverage of the 2p-injured half of the regenerating arbor at 72 h AI compared with the pinch-injured side of the arbor (Fig. 5B,D). Importantly, the extracellular damage—either to the ECM or epidermal cells—could be visualized and aligned with the pinch-injured side of the regenerating arbors. The ECM “scar” following pinch injury is highlighted with a white outline at 72 h AI and aligned perfectly with the congested dendrites attempting to regenerate into the damaged territory (Fig. 5A). There was a stark contrast between the damaged and misshapen epidermal cell morphology of the pinch-injured versus the intact, cobblestone morphology of the 2p-injured sides of the hemisegment (Fig. 5C). By combining the half-pinch + half-2p injuries with these extracellular markers, the discrete differences in regenerative capacity are directly linked to the damage sustained by the extracellular tissues due to pinch injury. These data reinforce our conclusion that damage to the extracellular environment must be considered to effectively evaluate neuronal regeneration in a real-world context.

Discussion

Here, we have established a pinch injury method to injure dendrites that causes holistic tissue damage, effectively damaging the neurons and their dendrites, the ECM, and epidermis, while leaving the glia largely intact. This pinch injury method elicits a robust immune infiltration to the pinch-injured hemisegment. Importantly, the surrounding tissue damage restricts invasive dendrite regeneration into the empty territory, instead driving compensatory regeneration on the spared side of the arbor. This contrasts sharply with the invasive dendritic growth observed following 2p laser injury, where the epidermis and ECM remain undamaged. The discrete regenerative response to these two injury paradigms was further demonstrated in our half-pinch + half-2p injury paradigm: the pinched half of the class IV ddaC neuron exhibited reduced branch number, reduced branch length, and reduced area coverage compared with the 2p-injured half of the dendritic arbor. Performing the half-pinch + half-2p injury paradigm with the ECM and epidermal cell markers revealed that restricted dendrite regeneration spatially coincides with tissue damage, directly implicating the extracellular environment as a driver of poor regenerative outcomes. Given that the pinch injury damages not only neurons but also surrounding tissue, analogous to what occurs in traumatic brain injury (TBI), blunt force trauma, sports-related injuries, and blast injuries, this model provides a physiologically relevant platform to study neuronal regeneration (Hicks et al., 2010; Mustafa and Alshboul, 2013; George and Geller, 2018). In these real-world injuries, damage extends beyond neurons to include the ECM, supportive glial cells, and endothelial cells comprising the blood–brain barrier, making the pinch injury a more accurate model of acute neurotrauma than laser ablation alone (Gaudet et al., 2011; George and Geller, 2018; Yip et al., 2022).

The pinch injury method we describe here complements existing injury paradigms used to study neuronal regeneration. While laser ablation (either 2p or pulsed UV laser) has been instrumental in advancing our understanding of dendrite regeneration due to its precision and reproducibility (Duarte et al., 2024; Hertzler and Rolls, 2024; Prange et al., 2024; Hwu et al., 2025), it is fundamentally different from physiological injury in that surrounding tissue is left intact. Mechanical injury models such as axon crush injuries, commonly used to study peripheral nerve regeneration, cause widespread tissue damage similar to our pinch injury. Axon crush paradigms in Drosophila, vertebrate, and other invertebrate models have been widely used to study axon regeneration and have revealed that both the neuron and its local environment must be considered when evaluating regenerative capacity (von Bernhardi and Muller, 1995; Menorca et al., 2013; Rooney and Freeman, 2014; Akram et al., 2022; Bhattacharya, 2023; Pluta et al., 2025; Waller et al., 2025). Similarly, peripheral nerve injuries in mice, including sciatic nerve crush and pinch injuries, cause damage to neurons, Schwann cells, blood vessels, and ECM, triggering inflammatory responses that influence regenerative outcomes (Gaudet et al., 2011). Our dendrite pinch injury model parallels these axon injury paradigms but allows for spatial compartmentalization of damage to part of a single neuron. We liken this aspect of our pinch injury model to the novel twist injury, which uses a heat-pulled glass capillary tube to make a fine, hair-like structure which effectively severs one of the primary dendrites of PVD C. elegans neurons (Singh et al., 2026) or the removal of one first-order branch in a medicinal leech mechanosensory neuron, the dorsal pressure-sensitive neuron (P_D; Wang and Macagno, 1998). This unique feature of our pinch injury model—injuring only half of the dendritic arbor—enables direct within-neuron comparisons of regeneration in damaged versus undamaged tissue environments, providing insights that are harder to glean from whole-nerve injury models.

The accumulation of the Drosophila homolog of collagen IV, viking (vkg), and damage to the ECM that we observed following pinch injury may represent a scar-like structure that impedes dendrite regeneration. This vkg/collagen IV accumulation and scar-like structure is similar to the collagenous scar that forms in the ECM following dorsal pinch injury in Drosophila larvae (Ramos-Lewis et al., 2018). Further, in mammalian spinal cord injury (SCI) and TBI, glial scars form at the lesion site, characterized by reactive astrocytes, the deposition of chondroitin sulfate proteoglycans (CSPGs), and ECM remodeling (Siddiqui et al., 2022; Tran et al., 2022). Similarly, fibrotic scars composed of collagen and fibronectin create physical and chemical barriers to axonal regeneration following nerve injury (Ayazi et al., 2022). The vkg/collagen IV accumulation we observe in the pinched tissue shares features with these injury-induced scars, raising the possibility that it serves as a barrier to dendrite regrowth. However, the mechanism by which this damaged ECM might restrict dendrite regeneration remains unclear. Injured dendrites might be actively repelled by molecular cues within the damaged tissue, analogous to how CSPGs inhibit axon growth in the glial scar through receptor-mediated signaling (Ohtake and Li, 2015). Alternatively, the damaged epidermis and disrupted ECM architecture may simply fail to provide the adhesive substrate or trophic support necessary for dendrite attachment and growth (Sharma et al., 2023). Distinguishing between active repulsion and failure of permissive attachment is critical for understanding whether dendrites avoid the injured territory or are unable to invade it. Our half-pinch + half-2p injury paradigm, which reveals preferential growth on the 2p-injured side, suggests that the lack of tissue damage may be sufficient to promote regeneration, favoring a model in which the damaged tissue fails to support dendrite growth rather than actively repelling it.

Irregularities in epidermal cell morphology following wounding in Drosophila larvae have been described before (Kwon et al., 2010). We speculate that the disfigured epidermal cells we observed were damaged by pinch injury and subsequently experienced epidermal cell fusion (White et al., 2026). In the larval epidermis, wounding can lead to epidermal cell fusion and the formation of multinucleate syncytial cells (Galko and Krasnow, 2004; Wang et al., 2015; White et al., 2026). Wounding the adult Drosophila epidermis also leads to syncytial formation along with polyploidization further from the wound's edge (Losick et al., 2013; Losick, 2016). This hypothesis of polyploidization is attractive given that polyploidization as a result of endoreplication is required for epidermal cells in Drosophila larvae, acting as nonprofessional phagocytes, to successfully clear sensory dendrite debris following injury (Han et al., 2014; Huang et al., 2025).

The tissue damage following pinch injury is not static but rather evolves over time, potentially creating an increasingly restrictive environment for dendrite regeneration. We observe that vkg/collagen IV accumulation in the damaged ECM increases progressively following injury. This temporal pattern suggests that the barriers to regeneration may intensify as time progresses, rather than resolving to create a more permissive substrate for growth. The dynamic remodeling of the ECM following injury may be mediated in part by matrix metalloproteinases (MMPs), which degrade ECM components and play complex, time-dependent roles in regeneration (Yong, 2005; Phillips et al., 2014). In mammalian CNS injury, MMPs are rapidly upregulated in the acute phase and contribute to both beneficial effects, such as clearing debris and allowing cell migration, and detrimental effects, including disruption of the blood–brain barrier and exacerbation of inflammation (Andries et al., 2017). As such, the accumulation of vkg/collagen IV that we observe may reflect an active remodeling and reorganization of the ECM by MMPs, which, in turn, could progressively limit the ability of the ECM to form a permissive substrate for dendrite growth (DeVault et al., 2018; Furusawa and Emoto, 2021).

The immune response following pinch injury is also a highly evolving process. Prior to injury, there are two subsets of hemocytes aside from the lymph gland. Sessile hemocytes are present in hematopoietic pockets between the larval epidermis and musculature (Márkus et al., 2009; Makhijani et al., 2011; 2017). The second set of hemocytes include those that circulate within the hemolymph (Babcock et al., 2008), and these two subsets of hemocyte populations can exchange themselves, transitioning between circulatory and sessile states (Honti et al., 2010; Makhijani et al., 2011). Those that circulate within the hemolymph are described to patrol the larval body and only “happen upon” a wound when they bump into it as they circulate. At that point, those hemocytes adhere and begin proliferating in response to the damage (Babcock et al., 2008). Once the damage and wound are resolved (for an epidermal pinch wound, this is ∼24 h after wounding), the hemocytes are released back into circulation by the healed epidermis (Babcock et al., 2008; Burra et al., 2013). Our hemocyte accumulation data indicates a more prolonged immune reaction, where there exists a robust presence of hemocytes up to at least 72 h after pinch injury. The hemocyte accumulation is beginning to be resolved by 72 h after injury on the pinched side but is still high and stable on the spared side of an injured hemisegment.

What might account for these observations about hemocyte retention? Long-lived chemoattractants might continuously recruit hemocytes, structural features might impede hemocyte migration, and/or injury-induced alterations in hemisegment size might complicate the data. We consider these possibilities in turn.

Various chemoattractants could be responsible for prolonged hemocyte recruitment. Upon initial epidermal wounding in Drosophila larvae, damage-associated molecular patterns (DAMPs) are produced, such as reactive oxygen species (ROS) like H₂O₂ (Razzell et al., 2013; Shaukat et al., 2015; Niethammer, 2016; Evans et al., 2022). Production of ROS following epidermal cell damage is required for hemocyte activation and proliferation (Moreira et al., 2010; Razzell et al., 2013; Evans et al., 2022). Although ROS are immediately produced upon wound formation and could initially stimulate hemocyte recruitment and proliferation, they are short-lived, so they are unlikely to account for long-term hemocyte retention (Niethammer et al., 2009). Long-lasting signals could include actinin-β and/or the Pvr/Pvf pathway. Actinin-β is expressed by the da neurons and maintains the hematopoietic clusters near da neurons through TGF-β signaling in the hemocytes (Makhijani et al., 2017). Pvr is a receptor tyrosine kinase in Drosophila, related to platelet-derived growth factor receptor and vascular endothelial growth factor receptor, which can bind to any of three ligands in Drosophila: Pvf1, Pvf2, and Pvf3. Pvr is expressed by epidermal cells, and proper epidermal wound healing and epidermal cell migration requires hemolymph-derived Pvf1 to bind to Pvr on the wounded epidermis (Wu et al., 2009). Pvr is also expressed by hemocytes, and Pvf1 binding to Pvr on hemocytes is required for hemocyte spreading at the wound and epidermal wound closure (Tsai et al., 2022). Pvr also plays a role in neuron morphology, as class IV ddaC neurons have reduced branch number and branch length in Pvr, Pvf2, or Pvf3 mutants, and this signaling is important for class IV ddaC neuron function as larval nociceptors (Lopez-Bellido et al., 2019). Taking these data together, we can consider that pinch injury may induce Pvf ligand expression or release, activating Pvr on epidermal cells, hemocytes, and/or neurons. This could in turn promote prolonged hemocyte spreading at the epidermal wound, explaining our persistent hemocyte presence (Extended Data Fig. 4-6). It could also promote epidermal wound closure and healing following pinch injury and additionally be important for maintenance of the function of the remaining arbor of class IV ddaC neurons following pinch injury. The decrease in hemocyte number on the pinched half of the arbor could alternatively be a reflection of the respective sizes of each side of a pinched arbor. Due to the damage by pinch injury, the width of the injured side of a pinched arbor is smaller than that of the spared side. A decrease in physical space over time, as a result of repairing surrounding tissue following pinch injury, could reduce the number of hemocytes that could “fit” into the damaged territory. Finally, the damage to the ECM and the formation of a “scar” could impede hemocyte migration into the damaged half of the hemisegment, as lamininB mutants experience decreased hemocyte motility (Korabel et al., 2022).

While we have identified damage to the epidermis and ECM as key features of the pinch injury, the full extent of tissue damage and the cellular sources of ECM remodeling remain to be fully characterized. We used the Gal4^A58 line to visualize vkg/collagen IV accumulation following injury, and while Gal4^A58 is expressed in epidermal cells, it is also expressed in the fat body (Ramos-Lewis et al., 2018). Therefore, the vkg/collagen IV accumulation we observe may derive not only from epidermal cells but also from the fat body, which serves as a major source of ECM proteins and circulating factors in Drosophila (Rodriguez et al., 1996; Isabella and Horne-Badovinac, 2015; Ramos-Lewis et al., 2018). Distinguishing between these sources will be important for understanding the origin and regulation of the ECM response to injury. Moreover, vkg/collagen IV is just one of many ECM components, which we selected due to its high abundance and structural importance (Rozario and DeSimone, 2010). Other ECM components, including laminins, proteoglycans, and other collagens, are likely also disrupted or dysregulated following pinch injury. The complete ECM composition in the injured territory, and how it differs from uninjured tissue, may provide insight into what makes the environment restrictive for dendrite growth. The pinch injury damages all cells within the compressed tissue, not only the sensory neuron and individual other cell types we imaged here. It is possible we are damaging or killing other da neurons and cell types that normally provide trophic support, guidance cues, or other proregenerative signals that facilitate dendritic outgrowth. Uncharacterized aspects of the injury underscore the complexity of the tissue damage and highlight the need for comprehensive characterization.

Lastly, an important consideration is how the combined damage to neurons and surrounding tissue would affect sensory function. Both class IV da neurons and epidermal cells are mechanosensitive and work cooperatively to detect mechanical stimuli in Drosophila larvae (Luedke et al., 2024; Yoshino et al., 2025). Damage as we observe following pinch injury could compromise mechanosensation through multiple mechanisms, including direct loss of epidermal mechanosensory function and disruption of neuron-epidermis communication (Yang and Chien, 2019; Luedke et al., 2024; Yoshino et al., 2025). This has important implications for understanding recovery from injury in clinical cases, where damage to neurons, glia, and other supporting cell types occurs simultaneously. Even if dendrites successfully grow new branches after pinch injury, as we observed through compensatory growth on the uninjured side of the arbor, full restoration of sensory function might require neuronal regrowth as well as repair of epidermal cells and the re-establishment of neuron–epidermis interactions.

Synthesis

Reviewing Editor: Fabienne Poulain, University of South Carolina

Decisions are customarily a result of the Reviewing Editor and the peer reviewers coming together and discussing their recommendations until a consensus is reached. When revisions are invited, a fact-based synthesis statement explaining their decision and outlining what is needed to prepare a revision will be listed below. The following reviewer(s) agreed to reveal their identity: NONE.

In this manuscript, the authors provide a new experimental approach to study dendrite regrowth after injury in the Drosophila larval somatosensory system. Prior studies used a two-photon approach to sever dendritic arbors and assess dendrite regrowth. However, this approach does not affect the local environment of the neuron and as such, does not model realistic injuries that affect neighboring tissues. Here, using a novel pinch injury method, the authors damage the neuron asymmetrically, severing the dendrites and surrounding tissue (including epidermal cells, glial cells and extracellular matrix) on one side of an arbor and leaving the other unscathed. They found that the pinch injury alters substrate epidermal cells and immune cell distribution, leads to the accumulation of the ECM component collagen IV, and prevents dendritic regrowth, with dendrites preferentially regenerating into areas of intact tissues. Altogether, this is a well-presented description of a mechanical neuronal injury paradigm that is likely to be of broad interest and sets the stage for future studies on the contributions of environmental damages to neuronal regeneration. It is appropriate for publication in eNeuro, provided the following points are addressed.

Major points:

1. The authors state that the pinch induced damage is reproducible, but in contrast to the 2P approach, the damage induced by pinch could be very user dependent, even though the video recording of the pinch assay may help somewhat with this. The authors describe "minimal pressure" as sufficient to cause damage. To enhance reproducibility, the authors could perhaps report the range of effects seen with "maximal" and "medium" pressure, so that others could calibrate what minimal pressure is based on the effects they observe.

2. The authors perform CaMPARI imaging to show that pinching induces Ca influx into damaged class IV neurons, which is proposed to be an initial step in injury detection. However, since these are nociceptors, the Ca response could be a part of the normal sensory responses of these cells rather than related to injury. Does Ca accumulate as well in non-nociceptive neurons that are damaged by the pinch?

3. The authors claim that branches are added to the uninjured half of the arbor. However, the quantification appears to represent the total number of branches (and dendrite length, etc.) of the whole arbor (e.g. see Figure 2). Either the authors should clarify what is being quantified in the graphs, or they should break up their quantification to separate out the two halves of the arbor (injured and uninjured).

4. The epidermal cell data was very difficult to interpret. In the images in Figure 4, it was difficult to discern the individual epidermal cells and whether and how they were affected. It would be helpful to show pre- and post-injury images for an individual animal to illustrate the effect of the injury on epidermal cells. In their current images, the authors do not quantify epidermal cell morphology or highlight example(s) of disrupted epidermal cells to reflect their claims in the text (lines 489-492).

5. The authors claim that the regenerating dendrites grow into territory surrounded by intact tissue. This claim, however, is not quantitatively supported (and is also difficult to qualitatively discern from the images). In other experiments, the authors use fluorescent markers of the ECM and epithelial cells; it would be helpful to use these markers to support the claim that dendrites preferentially extend into intact epithelial territories.

6. The authors claim that there is an influx of hemocytes into the territory damaged by the pinch. However, the hemocyte data is not quantified. The authors need to quantify the hemocytes in control and injured animals, and, ideally, determine the correlation between the hemocyte distribution and the pinch-induced damage.

Minor points:

1. Some of the images and anatomy may be unfamiliar to a broader audience. The authors are encouraged to add arrows and arrowheads to draw the attention of readers to areas of interest in the figure panels. One example of this is 4-3C, where readers must sort out the smaller epidermal cells and longer ones without any guidance.

2. In figure 4-2D, it is not clear why the authors change to dark and light green when they are quantifying the magenta channel in the neighboring image.

3. In several figures the authors label the uninjured controls as "X hrs AI", which is confusing since there is no injury in these images and plots. For example in Figure 2, both images and plots follow this convention. Could the authors use "x hrs control" instead?

4. The authors should be careful not to make claims about data that is not quantified. For example, the authors state (lines 506-507) that "the glia have trouble wrapping the proximal segments of the newly regenerated primary dendrites (Figure 4G and Figure 4-4C)," but these data are not quantified. The authors should make it clear that this is speculation based on a qualitative evaluation.

5. It could be helpful for the authors to highlight the pinched side in their figures with a bracket or bar. In later figures, which include different channels, it would help the reader to identify where the pinch occurred.

6. Based on the methods, it seems that two neurons are likely affected in each pinch since the pinch occurs over two hemi-segments. The authors should specific how many neurons are affected during each pinch.

7. The authors should specify how they quantified the CaMPARI signal, especially as they are using GFP to mark the neuronal morphology.

8. Line 478: "coracle" likely refers to a protein, in which case the first letter should be capitalized.