Here, we present a protocol using the Drosophila sensory neuron – dendritic arborization (da) neuron injury model, which combines in vivo live imaging, two-photon laser axotomy/dendriotomy, and the powerful fly genetic toolbox, as a platform for screening potential promoters and inhibitors of neuroregeneration.
The regrowth capacity of damaged neurons governs neuroregeneration and functional recovery after nervous system trauma. Over the past few decades, various intrinsic and extrinsic inhibitory factors involved in the restriction of axon regeneration have been identified. However, simply removing these inhibitory cues is insufficient for successful regeneration, indicating the existence of additional regulatory machinery. Drosophila melanogaster, the fruit fly, shares evolutionarily conserved genes and signaling pathways with vertebrates, including humans. Combining the powerful genetic toolbox of flies with two-photon laser axotomy/dendriotomy, we describe here the Drosophila sensory neuron – dendritic arborization (da) neuron injury model as a platform for systematically screening for novel regeneration regulators. Briefly, this paradigm includes a) the preparation of larvae, b) lesion induction to dendrite(s) or axon(s) using a two-photon laser, c) live confocal imaging post-injury and d) data analysis. Our model enables highly reproducible injury of single labeled neurons, axons, and dendrites of well-defined neuronal subtypes, in both the peripheral and central nervous system.
The inability of axons to regenerate after an injury to the central nervous system (CNS), may lead to permanent disabilities in patients, and also plays a role in the irreversible neurological deficits in neurodegenerative diseases1,2,3,4,5. The CNS environment, as well as the intrinsic growth ability of neurons, determines whether axons are able to regenerate after trauma. Extracellular factors from oligodendrocyte, astroglial, and fibroblastic sources have been shown to impede neuronal growth4,6,7,8, but the elimination of these molecules only allows for limited sprouting5. Intrinsic regeneration signals can influence regenerative success5,9 and represent potential therapeutic targets, but these processes are still not well-defined at the molecular level. Increases in trophic factor signaling or elimination of endogenous brakes, such as the Pten phosphatase10, can result in axonal regeneration in certain circumstances. Combinations of different methods found to be individually effective also only provide limited overall recovery to date11,12,13,14. Therefore, there is a desperate need to identify additional pathways for targeted therapy. In addition to the initiation of axon regrowth, whether and how axons re-wire to the correct target, reform synapse specificity, and achieve functional recovery are important unanswered questions.
In summary, current understanding of the machinery dictating axon regeneration is still very fragmentary. Part of the problem is the technical difficulty of studying axon regeneration in mammals in real-time, an approach that is costly, time-consuming, and challenging for conducting large-scale genetic screens. Drosophila melanogaster, on the other hand, has proven to be an exceptionally powerful system for the study of complex biological questions. The fruit fly has been instrumental in defining genes and signaling pathways that are strikingly conserved in humans and has been a successful model for the study of human conditions, such as neurodegenerative diseases, through the vast molecular genetics tools available to manipulate gene function15. In particular, fruit flies are considered to be an ideal tool for the discovery of genes involved in neural injury and regrowth15,16. Several fly neural injury models have been developed, including adult-head or larval ventral nerve cord (VNC) stabbing with needles, larval VNC or nerve crush with forceps, larval neuron laser axotomy, olfactory receptor neuron removal, brain explants injury, and peripheral nerve lesion by wing severance15,17,18,19,20,21,22,23. Excitingly, recent work utilizing Drosophila injury models have advanced our understanding of the cellular and genetic pathways used by the nervous system to respond to neural injuries, some of which have been shown to be conserved in mammals24,25. Again, this emphasizes the utility of this model organism for identifying novel mechanisms of neural repair.
Described here is a two-photon laser-based Drosophila larval sensory neuron injury model. A two-photon laser was first used to cut axons in zebrafish in vivo in 200326. In the same year, the first laser dendriotomy was performed in Drosophila using a pulsed nitrogen laser27. Shortly afterward, several C. elegans labs used femtosecond lasers to establish models of axon regeneration28. In 2007, Wu and colleagues compared and reported the differences between laser injuries in C. elegans induced by various types of lasers29. In 2010, axon regeneration after laser axotomy was first shown to occur in Drosophila30. Building on this extensive laser injury literature, we have developed a fly neural injury model using the two-photon laser, which allows precise induction of injury to targeted sites with minimal perturbation of neighboring tissues, providing a relatively clean system to study both the intrinsic and extrinsic properties of neuroregeneration with single-cell resolution. Specifically, we have established a set of injury methods for dendritic arborization (da) sensory neurons in both the peripheral nervous system (PNS) and CNS. Da neurons can be grouped into four distinct classes distinguished primarily by their dendrite branching complexities: class I to IV31. Our published work shows that da neuron regeneration resembles mammalian injury models at the phenotypic and molecular level: da neurons display class specific regeneration properties, with class IV but not class I or III da neurons exhibiting regeneration in the PNS; class IV da neuron axons regenerate robustly in the periphery, but their regenerative potential is dramatically reduced in the CNS, thus resembling dorsal root ganglion (DRG) neurons in mammals; enhancing mTOR activity via Pten deletion or Akt overexpression enhances axon regeneration in the fly CNS19. Utilizing this injury model, we have been performing genetic screens and have identified the RNA processing enzyme Rtca as an evolutionarily conserved inhibitory factor for axon regeneration, linking axon injury to cellular stress and RNA modification20.
In the presented paradigm, the injury is induced via laser axotomy/dendriotomy of larval class IV or III da neurons, labeled by ppk-CD4-tdGFP or 19-12-Gal4, UAS-CD4-tdGFP, repo-Gal80, respectively. The injury is performed on 2nd to 3rd instar larvae at around 48 – 72 h after egg laying (h AEL). For PNS axotomy the lesion is targeted to the section of axon ~20 – 50 µm away from the cell body, for CNS axotomy to an area of ~20 µm in diameter at the commissure junction in the VNC, and for dendriotomy to the primary dendritic branch points. The same neuron is imaged at 8 – 24 h after injury (AI) to confirm complete transection, and at 48 – 72 h AI to assess regeneration. Through time-lapse confocal imaging, the degeneration and regeneration of individual axons/dendrites that have been injured in vivo can be monitored over time.
1. Preparation of Culture Plates and Bottles
2. Collection of Drosophila Larvae
3. Two-photon Injury and Confocal Imaging
4. Data Analysis
Da neurons show differential regeneration potential between the peripheral and central nervous system, as well as class specificity. This provides unique opportunities to screen for novel factors that are required for axon regeneration (using class IV PNS injury), as well as those that are inhibitory for regeneration (with class IV CNS injury and class III PNS injury).
Axon regeneration in the PNS
As an example, the characterization of regeneration of class III and class IV da neurons is described. These neurons are located bilaterally in each body segment. Multiple neurons can be injured in the same larva; typically, 3-4 neurons in the right side of abdominal segments A7-A2. Class III and class IV da neurons can be visualized by 19-12-Gal4, UAS-CD4tdGFP, repo-Gal80 and ppk-CD4tdGFP, respectively. Anesthetize and mount 48-72 h AEL larvae as described, adjusting the position of the larvae so that the da neurons of interest are facing up (Figure 1A and 1B). We usually injure the class III ddaF and class IV v'ada neurons (Figure 1C and 1E). Perform axotomy and recover larvae as described. Shortly after injury (AI), the larvae will have recovered from surgery and exhibit normal locomotion. The survival rate here is typically over 80%. Discard larvae that are dead or sick. Remount the remainder as described and assess degeneration. At 24 h AI, the distal axons should have completed degeneration at this time point19, and the axon stem will be readily visible (Figure 1D and 1F). Reimage the same larvae at 48 h AI for class IV da neurons or 72 h AI for class III da neurons to assess regeneration. We usually aim to assess at least 20 injured neurons per experimental condition. In (wild type) WT animals, whereas typically ~70% of severed class IV da neurons will have regenerated beyond the injury site (Figure 1F), class III da neurons fail to regrow, evidenced by growth cone stalling (Figure 1D).
Dendrite regeneration
We usually perform dendriotomy on class IV da neurons ddaC (Figure 2A). As shown in the schematic diagram, the injury is targeted to the primary dendritic branch point. Based on experience, when injured at 48 h AEL, ~50% of ddaC neurons regenerate their dendrites (Figure 2B). In the remaining 50%, neighboring dendrites invade and cover the vacant space. Additionally, the regeneration potential of these neurons is reduced if injured at a later developmental stage.
Axon regeneration in the CNS
For VNC axon injury, the survival rate of larvae varies substantially and depends upon the age at which injury is induced. Based on experience, larvae of 48-72 h AEL typically have the highest survival rate (>60%) among the different stages tested. Larvae younger than 48 h AEL survive poorly after injury, while in those older than 72 h AEL it is difficult to introduce injury in the VNC. Moreover, it is easier to induce injury at the posterior commissure segments than the anterior, as these posterior segments of the VNC are closer to the ventral surface and thus more accessible by laser (Figure 3A).
To injure class IV da neuron axons in the CNS, mount the larvae as previously described (Figure 3A). Under the microscope, locate the ladder-like structure of axon bundles that form part of the VNC (Figure 3A), and perform axotomy as outlined. Degeneration is confirmed at 8 h AI and regeneration are assessed at 24 and 72 h AI. As shown in Figure 3B, axons at 8 h AI have already started to degenerate, and at 24 h AI, axon regeneration is observed while axon debris can still be found around the injury sites. WT axons show limited regrowth in the VNC and fail to reconnect the gaps generated by the injury (Figure 3B). To quantify the regeneration ability of injured axons, the regrowth length after injury is measured and the length of the commissure segment (Y in Figure 3A) is used for normalization (Figure 3C).
Figure 1: Da neuron axon regeneration in the periphery displays class specificity. (A and B) Schematic drawing showing the position of the larvae. (C) Schematic drawing of class III da neurons. (D) Axons of the class III da neurons ddaF, labeled with 19-12-Gal4, UAS-CD4-tdGFP, repo-Gal80/+, fail to regrow. (E) Schematic drawing of class IV da neurons. (F) Axons of the class IV da neurons v'ada, labeled with ppk-CD4-tdGFP/+, regrow beyond the lesion site. (D and F) Red line indicates axon length while green dashed line marks the distance between the cell body and the axon converging point (DCAC). The blue dot marks the axon converging point. Scale bar = 20 µm. Please click here to view a larger version of this figure.
Figure 2: Da neuron dendrite regeneration. (A) Illustrative representation of class IV da neurons. (B) Illustrative representation of dendrite regeneration in class IV da neuron ddaC, labeled by ppk-CD4-tdGFP/+. Laser ablation is targeted to the primary branch point and is conducted at 48 h AEL. At 24 h AI injury transection of the neurite is confirmed, and at 72 h AI regeneration is quantified. Dendrites of ddaC neurons demonstrate substantial regrowth, with new dendritic branches sprouting from the severed stem to tile the vacant space. It is worth noting that new terminal branches are continuously added to the uninjured dendrites at this developmental stage. Scale bar = 20 µm. Please click here to view a larger version of this figure.
Figure 3: Da neuron axon regeneration in the VNC. (A) Schematic drawing of a Drosophila larva mounted on a slide and imaged under the microscope. Class IV da neuron axons in the VNC visualized in a ppk-CD4tdGFP/+ larva. Two candidate commissure segments are shown in the zoomed-in image and the schematic drawing. Each of them has two injury sites (red circles). (B) Confocal images of one injured segment imaged at 8, 24 and 72 h after injury (AI). Red lines depict the regrowing axons. (C) Measurement and normalization of regrowing axons. Scale bar = 20 µm. Please click here to view a larger version of this figure.
When setting up fly crosses, the number of females and males used can vary depending on the genotypes and the number of larvae needed for specific experiments. For WT flies, typical cross uses 10 females and 5 males. The collection window may be narrowed, depending on the accuracy of the larvae age required. For example, a 2-h collection period will yield larvae of a more homogenous population. In this case, using 20 or more virgin females to set up the crosses will help yield sufficient eggs. The yield from the first day is usually scarce, so collect the plate with eggs two or more days after setting up the chamber. When further culturing the plate with eggs, it is recommended to soak the tissue in the 60-mm Petri dish in 0.5% propionic acid solution instead of water, which will not only help maintain humidity in the dish but also avoid the growth of mold.
When setting up the devices for larvae anesthesia and mounting, make sure to use a glass dish because ether will melt through plastics. The glass dish is housed in a 15-cm plastic Petri dish in case of a leak. The tissue paper helps preserve ether in the dish so that the larva can be effectively knocked out by the ether vapor. Using amber dropping bottles to store aliquots of ether and adding ether in droplets with the glass dropper is recommended. Replenish the ether and always keep a layer of liquid ether at the bottom of the dish for optimal effect.
The timing of ether exposure is critical: under-anesthesia will lead to larva recovery in the middle of the imaging session and shaky images; an anesthesia overdose, or direct contact with the ether liquid, will cause lethality. To maximize survival and success rate, our rule of thumb is as follows: for PNS injury/imaging, take out the larva as soon as its tail stops twitching; for the CNS, wait until the entire larva becomes motionless, especially the head segments. Even a slight movement will interfere with the injury and imaging of VNC axons. Older larvae tend to take longer to knock out. It usually takes less than 2 min for larvae younger than 72 h AEL and 2-5 min for larvae older than 72 h AEL.
When setting up the intensity of the two-photon laser for injury, the value is determined by the tissue fluorescence signal obtained from the "Live" scan. The injury is introduced by "Continuous" scan as in step 3.3. The injury-induced increase in fluorescence is due to autofluorescence at the injury site and serves as a good indicator of the severance of the neurite. Typically, it takes 1-10 s to see the fluorescence spike. It is critical to control the scan time once fluorescence is elevated. For PNS injury, it is essential to stop scanning immediately. Prolonged exposure will enlarge the injury site and damage neighboring tissues. However, this does provide the opportunity to adjust scan time and manipulate the severity of the injury. A successful injury should have a lesion size diameter smaller than 3-4 μm. For VNC axon injury, the VNC axons are embedded much deeper compared to the da neuron PNS axons/dendrites, which are right underneath the skin, and thus require higher laser intensity. We usually leave the scan on for a few more seconds. This is to ensure that the entire axon bundle is severed. If the radius of the lesion sites is more than half the width of the commissure bundle, such injury sites are counted as unsuccessful. Such larvae have a low survival rate and will not be included in the analysis.
For axon regeneration, the regeneration ability is similar across larval stages. But for dendrite regeneration after a single dendrite cut, the regeneration potential is reduced after 72 h AEL19. Thus, axon injury is typically performed at 48 h-72 h AEL and dendrite injury at 48h AEL, with the assessment of regeneration at 120h AEL. Larvae of 24h AEL can also be used, but they require more careful handling, given their smaller size. Larvae pupate after 120 h AEL, making imaging more challenging. Therefore, our endpoint is usually 120 h AEL.
What is the interrelationship between degeneration and regeneration? For PNS injury, at 24 h AI, normally the distal axon/dendrite in WT has completed Wallerian degeneration while regeneration has not started at this time point. Therefore, we believe that in WT larvae, degeneration of severed neurites only has a very limited impact on regeneration, if at all. For VNC injury, axons at 8 h AI have already started to degenerate, and at 24 h AI, axon regeneration is observed while axon debris could still be found around the injury sites. The debris does not seem to block regeneration. However, it is possible that under certain circumstances, there may be an overlap or even a crosstalk between degeneration and regeneration. In fact, it has been reported that in aged mice, debris clearance after peripheral nerve damage is slower than that in young animals. Concomitantly, slower reinnervation of the neuromuscular junction was observed, which may be attributed to the greater number of obstructions regenerating axons encounter in the old animals. Surprisingly, however, axons from aged animals regenerate quickly and reinnervate neuromuscular junction sites efficiently when not confronted with debris32. This suggests that facilitating debris clearance might be a potential strategy to promote regeneration.
Compared to other neuroregeneration models, the fly sensory neuron injury model has unique advantages. Mouse models usually take weeks to months to perform and are not suitable for conducting large-scale genetic screens; C. elegans only has a primitive central nervous system which may not closely recapitulate the regeneration barriers in the mammalian CNS; different from mammals, zebrafish CNS axons regenerate robustly. The rapid life cycle of flies, the versatility of fly genetics, accessibility and stereotypical patterning of axons/dendrites of fly sensory neurons, and the characteristic regeneration properties of fly sensory neurons – subtype specific regeneration in the PNS and limited regeneration in the CNS – make Drosophila da neurons an attractive model for studying neuroregeneration. Moreover, recent studies from crushed mouse retinal ganglion cells (RGCs) also suggest that neuronal subtypes bear distinct regeneration competence; some RGC subtypes can regenerate, whereas others in the seemingly homogenous nerve bundle fail to regrow33. This important finding suggests that neuronal type-specific strategies should be exploited to promote regeneration and functional recovery and argues strongly that the induction of axon injury and subsequent regeneration analysis should be performed in a neuronal subtype-specific manner. Furthermore, the cellular and molecular determinants for this regeneration type-specificity remain largely unknown19,33. Therefore, the da neuron injury model offers the ideal paradigm to tackle these issues.
Compared to axon regeneration, studies focusing on dendrite regeneration are much scarcer. Dendrite injury does occur, such as in traumatic brain injury, stroke, and many forms of neurodegeneration, yet almost nothing is known about dendrites' ability to repair and reform neural connections. The da neuron injury model again provides a highly accessible system that displays stereotypical patterning, class specificity, and temporal regulation19, to explore this direction.
It is also worth mentioning that while it is possible to injure a labeled single axon in the PNS, the way injury is performed in the CNS results in the lesion of a bundle of axons. If desired, MARCM34 or the FLP-out clone35 approach may be used to label single axons in the CNS. Additionally, when glial cells are simultaneously labeled with mRFP (Repo-Gal4, UAS-mRFP), the accumulation of glia processes is observed specifically at the lesion site. Furthermore, the expression of Ptp99A, the fly homolog of chondroitin sulfate proteoglycan (CSPG) phosphacan/ PTPRZ1, is up-regulated at the lesion site. Ptp99A co-localizes with glial cells and surrounds the injury site, forming a ring-like structure similar to what has been reported for astroglial scars in mammals19,36,37. In conclusion, the da neuron injury model, when combined with markers of other cell types such as glial cells or immune cells, will allow in vivo real-time surveillance of the multicellular interactions between an injured neuron and its surrounding environment.
While this fly larvae sensory neuron injury model provides us opportunities to find potential neuroregeneration regulators both in the PNS and CNS, it still has several limitations. First, it is not of high throughput at the present stage. Typically, 5-6 genotypes could be screened by one person in one week. It needs to be optimized in order to perform an unbiased screen. Second, as the ether anesthesia on larvae can last from several min to no more than twenty min, it is not optimal for long-term imaging. Thus, specific time points that are representative of axon degeneration and regeneration respectively are chosen for imaging. Third, even though this protocol is introducing a way to injure axons precisely, the possibility of damage to surrounding tissues cannot be completely ruled out. In the VNC, these tissues can be glial cells, axons, and dendrites of other neurons. To minimize this potential caveat, we minimize the injury sites, apply the lowest laser power possible, and perform the same procedures in parallel to both control and experiment groups. Fourth, in the process of setting up the CNS injury paradigm, different injury sites were tested, including the sites close to the axon termini that we chose to use and the entry point of axons into the VNC. The entry points were found to be more difficult to injure because they are deeper in the tissue and more likely to move. Thus, for these practical reasons, we opt for the current method, which is more consistent, better controlled, and good for larger-scale experiments. One potential concern, as stated above, is the possibility of injuring neighboring tissues, such as the postsynaptic components and the glial cells. On the other hand, it is an outstanding question how the damaged surrounding tissues influence the degeneration and regeneration of axons. For example, how the glial scar affects axon regeneration. This presents another reason why our injury model may closely resemble injury models in mammals, in which axons and tissues around the lesion sites are usually injured simultaneously.
Laser injury followed by time-lapse microscopy is a sensitive assay for studying axon/dendrite regeneration. However, one main concern of this assay is the perceived cost, which ranges from <$10K for low-end solid-state pulse lasers, $25-100K for femtosecond lasers to >$100K for two-photon lasers. There are several good discussions about different laser systems used for axotomy29,38,39,40,41,42,43,44,45. To summarize, conventional lasers are optimal for cutting axons within about 30-50 µm from the surface. There will be more collateral damage with the nano and pico second lasers compared with the femtosecond laser, especially as the depth of the target area increases45. For injuring axons in the VNC, the depth of which is typically about 50-100 µm, it is essential to minimize tissue damage. In this case, the two-photon laser is ideal, which focuses the laser power to the focal plane, reducing collateral tissue damage without compromising tissue penetration. In conclusion, the two-photon system is costly but offers the best precision and tissue preservation. However, if only the PNS axons are the target for axotomy, conventional pulse lasers may be a more affordable alternative.
The authors have nothing to disclose.
We thank Jessica Goldshteyn for technical support. Work in the Song lab is funded by the NIH grant R00NS088211, and the Intellectual and Developmental Disabilities Research Center (IDDRC) New Program Development Award.
Diethyl ether, ACS reagent, anhydrous | Acros Organics | AC615080010 | |
Halocarbon 27 Oil | Genesee Scientific | 59-133 | |
Phosphate buffered saline (PBS), 20x Concentrate, pH 7.5, supplier # E703-1L | VWR | 97062-948 | |
Agar powder, Alfa Aesar, 500GM | VWR | AAA10752-36 | |
Grape juice | Welch’s | ||
Ethanol 95% (Reagent Alcohol 95%) | VWR | 64-17-5 | |
Acetic acid | Sigma-Aldrich | A6283 | |
Propionic Acid | J.T.Baker | U33007 | |
Cover Glasses: Rectangles | Fisher Scientific | 12-544-D | 50 mm X 22 mm |
Zeiss LSM 880 laser scanning microscope | Zeiss | ||
Zen software | Zeiss | ||
Chameleon Ultra II | Coherent |