Exercise-Induced Differential Transcriptional Output of AMPK Signaling Improves Axon Regeneration and Functional Recovery
-
Author contributions: S.B. and A.G.R. designed research; S.B. performed research; S.B. contributed unpublished reagents/analytic tools; S.B. analyzed data; S.B. and A.G.R. wrote the paper.
Abstract
In adulthood, the regenerative capacity of the injured brain circuit is poor, thereby preventing functional restoration. Rehabilitative physical exercise is a promising approach for enhancing the behavioral recovery after neuronal injury to both the central and peripheral nervous systems. The metabolic energy sensor AMPK acts as a mediator for the exercise benefit in Caenorhabditis elegans axon regeneration. However, the mechanistic understanding of upstream and downstream components of AMPK signaling in the physical exercise-mediated enhancement of axon regeneration is still unclear. Here, we addressed this question by combining swimming exercise with laser axotomy of C. elegans posterior lateral microtubule (PLM) neurons. Using a genetically encoded ATP sensor iATPsnFR1.0, we observed that immediately after swimming exercise, ATP level is decreased both in neuron and muscle. Further, we found that AICAR-mediated AMPK activation is sufficient to promote axon regeneration and functional recovery. The PAR-4/Liver kinase B1 acts upstream of AMPK to improve functional recovery through swimming exercise. We also found that the transcriptional regulators DAF-16 and MDT-15 mediate the beneficial effects of swimming by acting downstream of AMPK. MDT-15 functions within neuron to mediate the benefit of AMPK activation, whereas DAF-16 acts both in neuron and muscle to promote functional restoration. Additionally, we demonstrated that swimming exercise induces nuclear localization of DAF-16 in an AMPK-dependent manner. Our results showed that neuronal and non-neuronal arms of AMPK signaling play an integrative role in response to physical exercise to promote functional recovery after axon injury.
Significance Statement
Finding ways to promote functional recovery after nervous system injury has been challenging, as the axons of adult neurons cannot regenerate well. Rehabilitation therapy is a promising approach to improve the health condition of patients with nervous system injuries. Swimming exercise can promote axon regeneration in the roundworm C. elegans, which is mediated by the metabolic energy sensor AMP Kinase. In this study, using sensory neurons in worm, we found that PAR-4/Liver kinase B1 acts upstream of AMPK, whereas the transcription factor DAF-16/FOXO and the transcriptional coregulator MDT-15 act as downstream signaling arms in muscle and neuron tissues. Excitingly, both of these arms could be harnessed through agonist-mediated activation of AMPK to promote functional recovery in adulthood.
Introduction
Injury to the central nervous system (CNS) axons results in functional impairment due to axon regeneration failure caused by multiple extrinsic and neuron-intrinsic factors (He and Jin, 2016; Zheng and Tuszynski, 2023). The peripheral nervous system (PNS) shows better regeneration than CNS, although behavioral recovery is limited (Gordon and English, 2016). Clinically validated rehabilitative therapy is promising to attain functional recovery after spinal cord injury (Jung et al., 2016; Wagner et al., 2018; Griffin et al., 2023). It also enhances axon regeneration after injury in the PNS (Molteni et al., 2004; Gordon and English, 2016; Maugeri et al., 2021). Exercise is linked to axon regeneration through the reduction of inhibitors, activation of proregenerative pathways, reduced inflammation, and increased blood flow (Maugeri et al., 2021). However, the cellular and molecular mechanisms of how physical exercise improves axon regeneration and functional recovery are still unclear.
Caenorhabditis elegans is an excellent model system to address this gap, as many axon regeneration regulators are being discovered using this system (Yanik et al., 2004; Ghosh-Roy et al., 2010; Byrne and Hammarlund, 2017). Moreover, it is shown using this model that axon regeneration in its early stages of life leads to functional recovery, whereas it is inhibited in old age (Abay et al., 2017; Basu et al., 2017, 2021; Ding and Hammarlund, 2018). Additionally, swimming paradigm mimics major features of mammalian exercise in C. elegans (Laranjeiro et al., 2017, 2019). Interestingly, swimming exercise after PLM neuron axotomy promotes axon regrowth and functional restoration in both the early and old stage of life, which is dependent on the metabolic sensor AMPK (Kumar et al., 2021). The metabolic sensor AMPK (AMP-activated protein kinase) monitors the ATP level through the competitive binding of AMP/ATP in its γ subunit (Steinberg and Hardie, 2023). AMPK is essential for swimming exercise-mediated improvement in behavioral recovery, fitness preservation, and muscle health in C. elegans (Kumar et al., 2021; Campos et al., 2023; Chen et al., 2023). Phosphorylation of Thr172 in the α subunit of AMPK is essential for its enzymatic activity (Hardie et al., 2016), and it is catalyzed by the Liver Kinase B1 (LKB1) and CAMKK2/CAMKKβ (Hawley et al., 2003, 2005; Woods et al., 2003, 2005). Activated AMPK maintains energy homeostasis through the regulation of metabolic enzymes and transcription factors (Herzig and Shaw, 2018). It is not yet clear how AMPK is activated in neurons and other tissues following physical exercise. Also, the upstream and downstream signaling components of the AMPK pathway in exercise-mediated enhancement of the functional rewiring process remain unclear.
In this study, we utilized the combined paradigm of laser-induced PLM neuron axon injury, swimming exercise, genetics, and pharmacology to identify the upstream and downstream components of AMPK signaling that are required to improve axon regeneration. We found that immediately after a 90 min swimming session, the ATP level is decreased both in neuron and muscle with a concomitant activation of AMPK. We have also shown that acute activation of AMPK using AICAR treatment is sufficient to promote regrowth from the injured axon tip, fusion between proximal and distal axons, and guidance to the ventral nerve cord. Furthermore, we found that LKB1/PAR-4 acts upstream of AMPK following swimming exercise to mediate exercise benefits in regeneration. We also show that FOXO/DAF-16 and MDT-15/PGC1α constitute critical downstream transcriptional effectors through which AMPK promotes functional recovery. Nuclear localization of DAF-16 is enhanced both in neuron and muscle following a swimming session in an AMPK-dependent manner. Additionally, using tissue-specific RNAi, we demonstrated that DAF-16 is required in both muscle and neuron, while MDT-15 is required only in neurons to regulate rewiring events. Our study revealed an integrative role of neuron-specific and non-neuronal AMPK signaling arms in physical exercise-mediated enhancement of axon regeneration.
Materials and Methods
Genetics and Caenorhabditis elegans strains
Standard conditions were followed to grow and maintain the strains using Nematode Growth Media (NGM) at 20°C (Brenner, 1974). For some sets of experiments, worms were maintained at alternative temperatures as mentioned in the respective places. Standard E. coli OP50 strain was used for growing the worms. Detailed genotypes of the C. elegans strains used are denoted in Table S1. Standard nomenclature notion was followed to describe genotypes of the C. elegans strains used in this study. For example, aak-2(lf) denotes that a loss of function allele, i.e., ok524, was used in which AAK-2 is not functional. These genetic strains were acquired from the Caenorhabditis Genetics Centre (CGC) and genotyped using an appropriate strategy involving PCR, RFLP and sequencing.
Synchronization of worms
Fifty to seventy gravid worms were transferred to freshly seeded NGM plates containing OP50 bacteria, and the worms were allowed to lay eggs for 2 h. After 2 h, the adults were removed from the plate and embryos were grown until the L4 stage. Then 60–70 L4 worms were transferred to fresh OP50 seeded NGM plates containing 50 μM 5-fluoro deoxyuridine (FUDR; Sigma; catalog #F0503; Basu et al., 2017). The worms were transferred to a new FuDR plate after 3 d. The adult day was counted from the time of L4 stage worm transfer for example after 24 h from the time of transfer; the worms are labeled as day 1 adults (A1 stage). For the experiments in this study, multiple life stage worms were used as specified in respective places.
Swimming exercise paradigm
We used a single session of swimming exercise in our study which is considered as an acute physical exercise paradigm (Laranjeiro et al., 2017, 2019; Kumar et al., 2021). For a swimming session, the worms were subjected to 1× M9 buffer in a 96-flat bottom plate for 90 min. A single worm was kept in a single well containing 200 µl buffer for the swimming session (swimming well). The plate was gently shaken at a 10 min interval to avoid the settlement of worms. After the swimming session was over, the worms were recovered on NGM plates using a Pasteur pipette. The control nonswimming group worms were kept on an unseeded NGM plate for the duration of the swimming session and were then transferred to OP50 containing NGM plates after 90 min. We primarily used a 90 min swimming session for our experiments as the ATP levels significantly dropped after this and worms switched to an episodic mode of swimming as per previous reports (Ghosh and Emmons, 2008; Chaudhari and Kipreos, 2017; Kumar et al., 2021).
Laser-assisted axotomy
Appropriate aged worms were immobilized on a 5% agarose bed prepared on top of a glass slide. Worms were physically immobilized in a 0.1 μm polystyrene bead solution (Polysciences 00876-15) while sandwiched between the agarose and a coverslip. We axotomized only a single PLM neuron axon for all the experiments at a 50–60 µm distance from the cell body as described previously (Basu et al., 2017). The gap created by two shots of laser pulse was ∼7–10 µm. The “cut side” and the “control side” are the sides that correspond to the side of axotomized PLM neuron and uninjured PLM neuron, respectively. We used a Bruker system with a Spectra-Physics femtosecond multiphoton laser setup for performing simultaneous imaging and axotomy as described previously (Basu et al., 2017). The system uses two Spectra-Physics (Mai Tai with DeepSee) automated depression-compensated femtosecond lasers which are tunable in the wavelength range of 690–1,020 nm. The axotomy was performed using a 720 nm laser micropulse, GFP-labeled PLM neurons were imaged with a 920 nm laser while for mCherry-labeled PLM neuron, imaging was performed with a 1,020 nm laser. A 3 mm galvanometer system was used for axotomy, while a 6 mm galvanometer scanning system was used for the imaging.
Alternatively, some sets of axotomy experiments were performed using a pulsed UV laser (Andor, Oxford Instruments) coupled to a Nikon Ti2 Eclipse microscope under a 100× oil objective (NA 1.4). The axons were severed similarly to the abovementioned specification.
Gentle touch assay
Gentle touch assay was performed for functional assessment after axon injury and in the context of aging according to the previously published method (Chalfie and Sulston, 1981; Chalfie et al., 1985; Basu et al., 2017). We gave 10 alternative anterior and posterior gentle touch with the fine tip of an eye-lash and recorded the reversal response. We scored the response by giving the anterior touch to the forward-moving worm and the posterior touch to a backward-moving worm. The anterior touch was given in the area between the pharynx and vulva whereas the posterior touch was given between vulva and anus. The positive reversal response was denoted as 1 and 0 for the corresponding no response. The posterior touch response index (PTRI) was calculated by taking the ratio of total positive response out of the total touch given which is 10 in our case as reported previously (Basu et al., 2017).
Correlation of functional recovery with axon regeneration at a single neuronal level
After 3 h of axotomy, each worm was kept on a single NGM plate and PTRI was assessed from each side of the worm. Then, worms were kept at 20°C for 24 h followed by PTRI measurement. After the 24 h PTRI was assessed, the worms were transferred to an agarose bed glass slide and the individual worm was immobilized in a drop of 10 mM levamisole solution. The anatomical regeneration pattern was assessed in an epifluorescence microscope (Nikon Ti2 eclipse) or confocal microscope (Nikon-A1 confocal system). The functional recovery was correlated with axon regeneration using the “Recovery Index.” The recovery index was calculated by the ratio of PTRI at 24 h to PTRI at 3 h for a single side representing the touch response of a single PLM neuron (Data S1).
Imaging of regenerated PLM axons
The regenerated PLM axons were imaged after 24 h of axotomy. The worms were immobilized on a 5% agarose pad using a 10 mM levamisole solution. The imaging was done using a Nikon Ti2 epifluorescence microscope or a Nikon A1 confocal system. In the epifluorescence system, PLM axons were imaged under a 40× oil (NA 1.1) objective using a Zyla EMCCD camera at the exposure value of 200 ms. GFP-labeled PLM axons were imaged using the appropriate filter set and z-scanning with a step size of 0.5 µm.
In the confocal system, PLM axons were imaged using a 40× air (NA 0.9) objective. The GFP-labeled axons were imaged using a 488 nm solid-state laser powered on 20%. The PMT power for the 488 nm channel was 100 with an offset value set at 20. PLM axons were imaged at a z-scanning step size of 0.5 µm. For the mCherry-labeled axons, imaging was done using a 561 nm laser power to 20%. The PMT power for the 561 nm channel was 110 with an offset value at 15. A similar 0.5 µm z-scanning step was used. The 24 h timepoint was used for the classification of PLM axon regeneration events. The fusion events were determined based on the suggested parameters like distal intactness, pattern of regrowth, and reconnection with distal parts (Basu et al., 2017; Neumann and Hilliard, 2019). Ventral targeting and other classes of nonfusion regrowth was classified according to previous guideline (Basu et al., 2021).
For regrowth analysis, the worms were imaged similarly at 10 h timepoint postaxotomy. The regrowth length was calculated from the z-stacked image using ImageJ with the plugin Simple Neurite Tracer (SNT).
Molecular biology and transgene
The cytoplasmic ATP sensor (iATPsnFR1.0) with Ruby was amplified from the plasmid cyto-Ruby3-iATPSnFR1.0 (Addgene plasmid #102551) which was a gift from Baljit Khakh (Lobas et al., 2019). The specific sequence was amplified using PCR by the primer set (Forward: 5′-TCCGAATTCGCCCTTCTGGAATTCATGGTGTCTAAGGGC-3′, Reverse: 5′-GTCGAATTCGCCCTTGGCCGCCTACTTCATTTCGG-3′) and cloned into the pCR8 entry vector through in-Fusion cloning (catalog #638945; Takara Bio). The resulted entry vector plasmid pNBR83 was used for LR coning reaction according to the manufacturer's protocol (catalog #11791020; Thermo Fisher Scientific) with the gentle touch sensory (mec-4 promoter) destination vector (pCZGWY553) and body wall muscle (myo-3 promoter) destination vector (pCZGWY925). The resulted plasmid pNBRGWY194 (Pmec-4:: cyto-Ruby3-iATPSnFR1.0) and pNBRGWY195 (Pmyo-3:: cyto-Ruby3-iATPSnFR1.0) was microinjected (20 ng/μl) for generating respective transgene.
The AMPK sensor AMPKAR was amplified using the plasmid (Addgene plasmid #35097) received as a gift from Lewis Cantley using primer set (Forward: 5′-TCCGAATTCGCCCTTGCCACCATGGTGCGGGGT-3′ and Reverse: 5′-GTCGAATTCG CCCTTGAATTCTTACTCGATGTTGTGGCGG) cloned into a pCR8 entry vector resulting in pNBR84 (pCR8::AMPKAR). Using LR reaction pNBRGWY196 (Pmyo-3::AMPKAR) vector was created as described above. This vector was microinjected at 25 ng/μl concentration similar to the reported method (Carman et al., 2022) for creating expression transgene.
Imaging of cytoplasmic ATP sensor
The sensor (iATPsnFR1.0) is an ATP sensor based on circularly permuted superfolder GFP inserted into the ε-subunit of F0F1-ATPase, which reports ATP binding via fluorescence changes (Lobas et al., 2019) and has been used in C. elegans (Kenny-Ganzert et al., 2025). The worms expressing the ATP sensor were recovered from swimming well and immediately mounted on an 8% agarose pad and immobilized using the polystyrene bead solution. One worm was mounted on the slide and imaging was carried out quickly. The swimming experiments and the sample preparation was carried near to the imaging setup to minimize the time gap. The nonswimming control animals were also imaged in a similar manner. For the azide control group, animals were incubated for 5–7 min in 5% sodium azide (#S202, Sigma) prior to imaging, previously shown to reduce muscle ATP in C. elegans (Galimov et al., 2018).
The imaging was carried out using a Nikon confocal microscope equipped with AX Galvano scanner and multichannel spectral detectors. The worms were imaged using a 60× oil (NA 1.42) objective using 488 and 561 nm laser in the minimal cross-talk configuration. The raw laser power was 10 and 5, respectively. The pinhole size was 52.1 μm, and dwell time 2 μs was used for imaging. GFP and mRuby channel settings were used at 35 detector gain. For the imaging of muscle reporter, raw laser power 1 was used and detector gain 35 for GFP and 30 for mRuby.
The intensity was calculated by drawing an ROI over the PLM cell body in the middle section slice (Fig. 1C,D). The intensity from the GFP channel (iATPsnFR1.0 signal) and mRuby channel was used for calculating the ratio. In case of the muscle, ROI was drawn over the top posterior muscle for intensity calculation (Fig. 1E,F).
Imaging of AMPK sensor
The AMPK activity sensor (AMPKAR) detect change in fluorescence resonance energy transfer (FRET) between ECFP and Venus based on the AMPK-dependent phosphorylation on the substrate peptide (Tsou et al., 2011) and it is used in C. elegans (Carman et al., 2022). The worms expressing the AMPK sensor (AMPKAR) in body wall muscle were recovered from swimming well and immediately mounted on a 5% agarose pad and immobilized using levamisole solution. One worm was mounted on the slide and imaging was carried out quickly. The swimming experiments and the sample preparation was carried near to the imaging setup to minimize the time gap. The nonswimming control animals were also imaged in a similar manner. For the AICAR group, animals were treated with 1 mM AICAR as described below section just prior to imaging. The imaging was carried out using a Nikon confocal microscope equipped with AX Galvano scanner and multichannel spectral detectors. The worms were imaged using a 60× oil (NA 1.42) objective using 445 nm laser for ECFP stimulation and Venus channel detection settings with a detector gain of 35 in the minimal cross-talk configuration. The raw laser power was 25. The pinhole size was 52.6 μm, dwell time 2 μs, and z-step 0.5 μm.
The anterior muscles (region anterior to the pharynx) were analyzed for the Venus signal and the signal puncta were analyzed using ImageJ tool. Briefly, the z-stacked images is background subtracted with a rolling ball radius of 50 pixels and thresholded at default setting. Then particle analysis (>0.2 μm2) was performed through particle analysis tool of ImageJ (Fig. 1G,H).
AICAR treatment on paralyzed worms
The animals were given a short exposure (∼15 s) to 5 mM of levamisole hydrochloride to temporarily paralyze them after 3 h postaxotomy PTRI was assessed. This brief levamisole treatment was effective in perturbing swimming during the 90 min swimming session in the swimming well. This short levamisole treatment does not affect gentle touch response (Kumar et al., 2021). Therefore, we used this paradigm to test the effect of AICAR treatment on functional axon regeneration. These paralyzed worms were subjected to a swimming well containing a specified concentration of AICAR (Sigma-Aldrich; catalog #A9978) in 1× M9 buffer. For the control groups, paralyzed worms were kept only in 1× M9 buffer for the duration of the swimming session. After the session, worms were recovered to NGM plates containing OP50 bacteria. For the aging paradigm, AICAR treatment was done similarly for the specified aged animals. Dorsomorphin dihydrochloride (Tocris, catalog #3093) was used in the swimming well at 10 μM working concentration during the swimming exercise session.
Thrashing frequency analysis of swimming worms
We recorded videos of swimming worms using a Leica MC 120 HD camera for the thrashing frequency analysis. We recorded for 30 s at 15 frames/s using LAS V4.4 software at 1.25× magnification in the camera attached to a Leica stereo microscope M165 FC. We calculated body bends per second from the video analysis by using the ImageJ software plugin wrMTrck (http://www.phage.dk/plugins/wrmtrck.html) and converted the values to thrashing frequency (thrashes/minute). One body bend was defined as a bending direction change at the midbody section (Kumar et al., 2021).
Tissue-specific RNAi condition
We used a similar induction method for efficient RNAi as reported earlier (Singh et al., 2024). We restreaked the required gene from the library stored at −80°C on an LB agar plate containing 50 µg/ml carbenicillin and 12.5 µg/ml tetracycline and followed by overnight incubation at 37°C. The bacteria was restreaked again before the inoculation into liquid culture. The bacteria were grown in 4 ml LB broth containing 50 µg/ml carbenicillin and 12.5 µg/ml tetracycline until the OD600 reached 0.7–0.8 value. Then the culture was centrifuged at 4,300 × g for 2 min to harvest the bacteria, and the supernatant was thoroughly dried. The pellet was resuspended in 1 ml of sterilized 1× M9 containing 1.5 mM IPTG, 50 µg/ml carbenicillin, and 12.5 µg/ml tetracycline. This resuspended culture was seeded to the previously prepared NGM plates containing the same concentration of IPTG, carbenicillin, and tetracycline. Approximately 330 µl of resuspended culture was seeded onto a 60 mm NGM plate whereas 50 µl was seeded onto a 35 mm NGM plate. For preparing these NGM plates, after pouring the NGM with the abovementioned chemicals, the plates were allowed 24 h room temperature solidification followed by 24 h incubation at 4°C. The seeded bacteria culture was dried in the laminar hood for 2 h and then incubated at 25°C for 36 h. After the incubation was over, 7–10 L4 worms of the required genotype were transferred to a 60 mm plate and kept at 20°C for growth. In each batch, control genes (dhc-1 and ama-1) were used to ensure the induction and optimum RNAi condition. We used the L4 progenies from the RNAi plates for the experiments. For muscle-specific RNAi, we used NR350 strain which is developed based on muscle-specific rescue of RDE-1 using hlh-1 promoter in rde-1(lf) background and is an established strain for this purpose (Qadota et al., 2007). For neuron-specific RNAi, we used the TU3595 strain which is reported to be optimized for efficient RNAi in gentle touch neurons (Calixto et al., 2010).
Aging of RNAi worms
We picked L4 worms from the RNAi plates described in the above section and transferred them to RNAi bacteria-seeded NGM plates containing the previously mentioned concentration of carbenicillin, tetracycline, IPTG, and also contained FUDR (50 mM). The worms were aged to the required adulthood stage in such plates and worms were then transferred to new plates at 3 d intervals. At the specific adulthood stage, worms were picked and subjected to swimming and nonswimming condition. Following swimming worms were recovered in RNAi plates and after 24 h functional assessment was performed.
Axotomy and functional recovery assessment of RNAi worms
The PLM neuron axotomy was performed with worms picked from the RNAi plates according to the above mentioned procedure. After the axotomy, worms were recovered on RNAi bacterial plates and kept for 3 h for recovery from the stress of the axotomy procedure. Then we separated each axotomized worm into a 35 mm RNAi bacterial plate and assessed the gentle touch behavior as per the method described previously. After the behavioral assessment, a swimming session was introduced to a group of worms while the control group did not undergo a swimming session. Following the swimming session, worms were recovered back onto the RNAi plates and kept in the incubator for 24 h for axon regeneration. Following 24 h, again gentle touch response assay was performed to assess the functional recovery and then worms were subjected to the microscopic analysis for anatomical characterization of axon regeneration.
Imaging of DAF-16 localization pattern after swimming
We used either the GFP or mKate2 endogenous CRISPR tagged in the daf-16 locus reporter strain to analyze the effect of swimming exercise on DAF-16 localization (Aghayeva et al., 2020; Zhang et al., 2022). A single worm was mounted on a 5% agarose pad using 10 mM levamisole and immediately imaged within 7–10 min using a Nikon confocal microscope equipped with AX Galvano scanner and multichannel spectral detectors. The sample preparation steps were followed to minimize the external stress. The microscope room was maintained at 20°C 30 min before the experiment and throughout the experiment. The control worms were kept in the unseeded plate for 90 min while the swimming group underwent a swimming session in the well. The worms were imaged using a 60× oil (NA 1.42) objective in the minimal cross-talk configuration mode.
DAF-16-GFP animals expressing mRuby muscle reporter were imaged using 488 and 561 nm laser powered to 5 value. GFP and mRuby channel settings were used with detector gain at 35. The pinhole size was 65.2 μm, dwell time was 2 μs, and z-step 0.5 was used for the imaging. DAF-16 localized to the nucleus of the body wall muscle posterior to pharynx were counted in a blindfolded manner.
DAF-16-mKate2 animals were imaged using 488 and 561 nm laser powered to 0.2 and 15, respectively. GFP and mkate2 channel settings were used with detector gain at 25 and 50, respectively. The pinhole size was 65.2 μm, dwell time was 2 μs, and z-step 0.5 was used for the imaging. An circular ROI (1 micron dia) was drawn on the central region of PLM neuron cell body, and the mean intensity ratio between the channels was calculated.
Statistics
We used GraphPad Prism software version 9.0.2 for all the statistical analysis in this study. We used an unpaired t test with Welch's correction for two-way comparison. We used Fisher's exact test for comparing proportions and ANOVA (nonparametric) with a post hoc Tukey's multiple-comparisons test for comparing three or more groups. Details of the tests are described in the figure legends. The values represent mean ± standard deviation. Each bar of the graph represents sample numbers (n) accumulated over the total number of biological replicates (N) in a given experiment.
Results
A 90 min swimming session leads to a cellular drop in ATP level in body wall muscle and the neuron
Swimming exercise in C. elegans is an established physical exercise paradigm that mimics key features of exercise in mammal (Laranjeiro et al., 2017, 2019). The increased activity during the swimming session requires high energy consumption (Chaudhari and Kipreos, 2017; Laranjeiro et al., 2017; Kumar et al., 2021). A single 90 min swimming session improves regeneration and functional recovery after laser axotomy of C. elegans posterior gentle touch sensory (PLM) neurons (Kumar et al., 2021). To understand how swimming exercise affects the energetics of neuron, we analyzed the intracellular ATP level using a genetically encoded fluorescent ATP reporter (iATPsnFr1.0) that is recently used in C. elegans (Fig. 1A,B; Lobas et al., 2019; Kenny-Ganzert et al., 2025). Our analysis immediately after 90 min swimming session showed a reduced intensity of iATPsnFr1.0 in PLM neuron cell body and also in body wall muscle without any change in the intensity of diffusible Ruby3 reporter (Fig. 1C–F). As a result, the ratio of intensity between iATPsnFr1.0 and Ruby3 is significantly decreased in both PLM cell body and body wall muscle just after the swimming session (Fig. 1C–F). Sodium azide treatment is reported to reduce the intracellular ATP concentration (Galimov et al., 2018; Morton et al., 2024), and our analysis also shows a decrease in the intensity ratio between iATPsnFr1.0 and Ruby3 after azide treatment, providing the validation of our assay (Fig. 1C–F). This ATP depletion after exercise is consistent with the previous finding that high activity during swimming exercise leads to high ATP consumption at the organism level (Chaudhari and Kipreos, 2017; Kumar et al., 2021). Our analysis shows that cellular ATP level also drops in neurons following a swimming session, along with the body wall muscle.
Swimming exercise induces ATP depletion in neuron and body wall muscle. A, The Schematic highlights the analysis of swimming exercise effects on ATP consumption in body wall muscle and neuron. B, Swimming exercise paradigm to analyze cellular ATP through imaging of the cytoplasmic ATP sensor (iATPsnFR1.0) immediately following a 90 min swimming session. C, Images of the PLM neuron cell body show a reduced fluorescence intensity of the ATP sensor (iATPsnFR1.0) immediately after a 90 min swimming session, whereas the soluble marker (Ruby3) does not change. The reporters are represented in the Fire LUT for better representation. Worms treated for 5 min in 5% sodium azide are used as a positive control. Scale bar, 5 μm. D, Quantification of the ratio of the fluorescence intensity (iATPsnFR1.0/Ruby3) in PLM neuron cell body in control nonswimming, swimming, and azide treatment condition. N = 3, n = 10–25 number of PLM neuron. E, Images of the posterior body wall muscle show a reduced fluorescence intensity of the ATP sensor (iATPsnFR1.0) immediately after a 90 min swimming session whereas the soluble marker (Ruby3) does not change. The ATP reporter is represented in the Fire LUT for better representation. Worms treated for 5 min in 5% sodium azide is used as a positive control. Scale bar ,20 μm. F, Quantification of the ratio of the fluorescence intensity (iATPsnFR1.0/Ruby3) in posterior body wall muscles in control nonswimming, swimming, and azide treatment condition, N = 3, n = 24–34 number of posterior muscle. G, Images showing the AMPK activation through AMPKAR reporter in body wall muscle immediately after swimming. The image is captured in venus fluorescence channel setting while the stimulation is through 433 nm laser. Temporarily paralyzed worms are treated with 1 mM AICAR as a positive control condition. Scale bar, 20 μm. H, Quantification of venus puncta through image particle analysis tool shows a significant increase following swimming session suggests AMPK activation in body wall muscle. Statistics, for D, F, H, *p < 0.05, **p < 0.01, ***p < 0.001 ANOVA with Tukey's multiple-comparison test. Error bars represent SD; ns, not significant.
As a consequence of a decrease in cellular ATP level following a swimming session, the AMPK might get activated. To test whether AMPK activation is triggered by the swimming session, we used the AMPKAR fluorescence reporter previously employed in C. elegans (Tsou et al., 2011; Carman et al., 2022). We saw an increased level of AMPKAR puncta in the body wall muscle following the swimming session (Fig. 1G,H). We used AICAR as an agonist, reported to directly activate AMPK (Corton et al., 1995; Kim et al., 2016), and previously used in C. elegans (Ju et al., 2022) to validate the assay. Similar observation of increased number of puncta was made when worms were treated with AMPK agonist AICAR (Fig. 1G,H). These observations strongly support that swimming exercise activates cellular energy sensor AMPK/AAK-2 as a consequence of cellular ATP drop.
AMPK activation through AICAR treatment improves functional recovery after PLM axon injury
Injured C. elegans neuron regenerate within 24 h, resulting in partial functional recovery (Abay et al., 2017; Basu et al., 2017, 2021; Ding and Hammarlund, 2018). In our experimental paradigm, the functional recovery is represented as “recovery index” which is the ratio of the PTRI at 24 h postaxotomy to the PTRI at 3 h postaxotomy (Fig. 2A; Basu et al., 2017; Kumar et al., 2021). The metabolic sensor AMPK/AAK-2 (AMPK α subunit) is essential for the swimming-mediated improved recovery, as this improvement is absent in aak-2 (lf) worms as opposed to wild-type worms (Fig. S1A), consistent with previous finding (Kumar et al., 2021). In C. elegans, both aak-1 and aak-2 encode for the alpha subunit of the AMPK complex (Apfeld et al., 2004). However, the swimming-mediated improved functional recovery remained unaltered in aak-1(lf) mutant (Fig. S1A). Also, according to CeNGEN (C. elegans Neuronal Gene Expression Network) dataset, the PLM neuron at threshold 2, aak-1 shows 0 transcripts whereas aak-2 shows 16.692 transcripts per million (Taylor et al., 2021). This strongly suggests the involvement of aak-2, ruling out the involvement of aak-1 in swimming-mediated improved recovery.
Direct activation of AMPK through acute AICAR treatment improves axon regeneration. A, Schematics of axon injury paradigm of the posterior gentle touch sensory neuron (PLM) in combination with swimming exercise. After a laser-induced break is introduced to the PLM axon, a posterior touch response assay was performed to record the PTRI followed by a 90 min swimming session. At 24 h of axotomy, PTRI is assessed followed by imaging of the regeneration. An alternative paradigm replaces the swimming session with the specified modifications. Short-term levamisole treatment administered to worms prior to the swimming session suppressed their physical activity in the swimming well, thereby blocking the beneficial effects of exercise (Paradigm 1). AICAR treatment was administered in the swimming well under this suppressed activity condition (Paradigm 2). B, Schematics highlights uncharacterized AMPK signaling in the context of exercise-mediated enhanced axon regeneration and behavioral recovery. C, Bar graph showing the effect of AICAR treatment on recovery index (24 h postaxotomy PTRI/3 h postaxotomy PTRI). Early stage adulthood (day 1) stage regeneration is characterized. N = 3–6 independent replicates, n = 18–55 number of worms. D, Representative image of PLM axon regeneration at 10 h postaxotomy following swimming exercise and 1 mM AICAR treatment. The orange arrow shows the position of the axotomy. Scale bar, 20 µm. E, Axon regrowth characterization at 10 h postaxotomy shows an enhancement following 1 mM AICAR treatment. N = 3, n = 18–28. F, Representative images depicting major classes of the PLM regrowth observed at 24 h postaxotomy. The regrowth type is classified as “Fusion” when the regenerated proximal axon fuses with the distal end and rescues it as distinguished by distal intactness. The regrowth type is classified as “Ventral targeting” when the regenerated axon is targeted and projected toward the ventral nerve cord as distinguished by the depth and spread on the cord. The other types of regrowth include distal targeting, straight anterior posterior axis regrowth, and complex pattern multiple branch regeneration events. The orange arrow shows the position of the axotomy. Scale bar, 20 µm. G, Quantification of the percentage of Fusion events after 1 mM AICAR treatment. N = 3–4 independent replicates, n > 20 number of worms. H, Quantification of the percentage of Ventral targeting events after 1 mM AICAR treatment. N = 3–4, n > 20 number of worms. Statistics, for C *p < 0.05, **p < 0.01, ***p < 0.001 ANOVA with Tukey's multiple-comparison test. For E **p < 0.01 unpaired t test. For G, H *p < 0.05 Fisher's exact test. Error bars represent SD; ns, not significant.
Next, we tested whether the direct activation of AMPK after axotomy can replace the swimming session itself to promote functional recovery (Fig. 2A,B). Although metformin has been reported to promote recovery in this context (Kumar et al., 2021), its complex mechanisms of action and broad pleiotropic effects complicate the interpretation of the finding (Rena et al., 2017; LaMoia and Shulman, 2021; Foretz et al., 2023). We administered a specific agonist of AMPK, AICAR (Corton et al., 1995; Kim et al., 2016), in a suppressed physical activity paradigm, in which a brief levamisole exposure immediately before the swimming session attenuates high locomotor activity in the swimming-well and thereby abolishes the beneficial effects of exercise (Fig. 2A,C; Kumar et al., 2021). We found that 1 mM AICAR treatment in this perturbed swimming session paradigm significantly improved the recovery index in the wild-type background (Fig. 2C). However, AICAR treatment in aak-2 (lf) animals could not significantly enhance functional recovery, suggesting that the enhancement occurs through AMPK activation (Fig. 2C). AICAR treatment improved regrowth length at 10 h postaxotomy signifying enhanced axon regrowth rate due to AMPK activation (Fig. 2D,E). The pattern of the axon regeneration events at 24 h postinjury can be broadly classified into a few categories, like “fusion,” “ventral targeting,” and “other types” of regrowth, which include straight, multiple branching, and dorsal regrowth (Fig. 2F–H; Basu et al., 2021). AICAR treatment mimicked the enhanced functional rewiring parameters seen due to the swimming session (Fig. 2F–H). For example, both the AICAR treatment and swimming-session enhance the “fusion” and “ventral targeting” events (Fig. 2G,H) that cumulatively contribute to the improved gentle touch response recovery (Fig. 2C). Therefore, we used the population behavioral recovery as a parameter to evaluate the roles of molecular components upstream and downstream of AMPK in mediating the exercise benefit.
Previously, it has been reported that the regeneration significantly declines with age, as seen with reduced recovery index in day 3 adults (Basu et al., 2017). Interestingly, acute AICAR treatment could overcome this age-related decline in recovery index at the day 3 life stage in an AAK-2-dependent manner (Fig. S1B). Similarly, in the background of a gain-of-function mutant of aak-2 (CA-aak-2), which constitutively activate AAK-2 (Burkewitz et al., 2015), the functional recovery after axon injury is significantly enhanced as compared with wild type both in early adulthood (L4 injury) and late adulthood (A3 injury) stage (Fig. S1C). Conversely, treatment with Dorsomorphin (Compound C), a well-known inhibitor of AMPK activity (Zhou et al., 2001; Berry et al., 2020) during the swimming session (Fig. S1D), blocked the enhanced functional recovery due to the swimming session, validating that AMPK activation is a key mediator of swimming exercise benefits (Fig. S1E).
We wanted to verify the effect of AICAR treatment in an alternate functional recovery paradigm that also utilizes the AMPK function. So, as previously reported, the PTRI decreases significantly as the worm ages from day 4 adult stage (A4) to day 5 adult stage (A5; Basu et al., 2017; Kumar et al., 2021). But with a 90 min swimming session at the A4 stage, such aging-related decline was overcame (Fig. S1F; Kumar et al., 2021). AICAR treatment of the temporarily paralyzed worms resulted in the preservation of posterior gentle touch response in the A5 stage (Fig. S2G,H). Conversely, brief treatment of dorsomorphin during the swimming session hinders the preservation of PTRI conferred by swimming exercise (Fig. S1I). These results established AMPK as a key regulator of exercise's effect on axon regeneration and behavioral improvement in adulthood.
Liver kinase B1/PAR-4 acts upstream of AMPK in swimming-mediated enhancement of axon regeneration
Next, we aimed to identify the enzyme that acts upstream of AMPK during swimming exercise (Fig. 3A). The AMPK enzyme is phosphorylated at the Threonine-172 position for its activation either by the liver kinase B1 (LKB-1) or by CAMKK2 (Ca2+/CaM-dependent kinase 2 (Hawley et al., 2003, 2005; Woods et al., 2003, 2005; Shaw et al., 2004). In C. elegans, the orthologues for liver kinase B1 and CAMKK2 are known as PAR-4 and CKK-1, respectively. The scaffolding protein PRY-1 or AXL-1, the worm orthologue of Axin, is a key mediator phosphorylation of AMPK by LKB-1 (Zhang et al., 2013; Chen et al., 2017; Mallick et al., 2020). We found that the recovery index is enhanced due to swimming exercise even in the absence of either ckk-1 or axl-1 (Fig. 3B,C). The functional improvement of touch sensation in the aging paradigm also supported this finding as improved PTRI in day 5 adults due to swimming session was not hampered in ckk-1 or axl-1 mutant and was comparable with the wild-type background (Fig. S2A,B).
Swimming-mediated improvement in functional recovery after axon injury of PLM neuron requires liver kinase B1/PAR-4. A, Schematics showing upstream regulators of AMPK that can be involved in swimming-mediated improvement in axon regeneration. B, Paradigm to evaluate the role of upstream candidates using genetic mutants. The PLM axon was severed using a controlled laser pulse at the L4 stage and functional assessment was done by analyzing the PTRI values after 24 h postaxotomy. C, Bar graph showing the recovery index values for wild-type, ckk-1(lf), and axl-1(lf) worms at 24 h postaxotomy with the introduction of a 90 min swimming exercise. N = 3–4 independent replicates, n = 26–38 number of worms. D, Bar graph showing the recovery index values for wild-type and par-4 (ts) worms at 24 h postaxotomy with the introduction of a 90-minute swimming exercise. The worms were grown until the L4 stage at permissive temperature (15°C) and after axotomy, were subjected to nonpermissive temperature (25°C). N = 3–4, n = 30–43. E, Image showing the axonal regrowth at 10 h postaxotomy with or without swimming exercise for wild-type and par-4 (ts) worms. After axotomy, worms were maintained at a nonpermissive temperature (25°C). The orange arrow shows the position of the axotomy. Scale bar, 20 µm. F, Quantification of regrowth length at 10 h as represented in panel E. N = 3–4, n = 13–24. G, Quantification of the percentage of Fusion events after swimming exercise of par-4(ts) worms at 25°C. N = 3–4 independent replicates, n > 20 number of worms. H, Quantification of the percentage of Ventral targeting events after swimming exercise of par-4(ts) worms at 25°C. N = 3–4, n > 20 number of worms. I, Improved behavioral recovery mediated by constitutive activation of AMPK (CA-aak2) is not affected by par-4(ts). N = 3, n = 28–41. J, Effect of swimming exercise on the recovery index of neuronal knockdown of pry-1. N = 3, n = 15–25. Statistics, for C, D, F, I *p < 0.05, **p < 0.01 ANOVA with Tukey's multiple-comparison test. For G, H *p < 0.05 Fisher's exact test. For J, *p < 0.05 unpaired t test; error bars represent SD; ns, not significant.
Since the absence of par-4 leads to embryonic lethality (Kemphues et al., 1988), to test the role of PAR-4, we used a temperature-sensitive allele of par-4, which displays a mutant phenotype at 25°C (Fig. 3B; Watts et al., 2000). We saw a significant enhancement in the recovery index in the wild-type animals due to a swimming session at 25°C which was significantly perturbed in the par-4 (ts) animals (Fig. 3D). Although the basal level axon regeneration length in par-4 (ts) at 10 h is comparable with wild type (Fig. 3E,F), swimming-mediated increment is not observed in par-4 (ts) (Fig. 3E,F). Accordingly, swimming exercise-induced enhancement of fusion and ventral targeting events is not seen in the absence of functional PAR-4 (Fig. 3G,H).
Similarly, swimming-mediated improvement of touch sensation in older worms at nonpermissive (25°C) temperature was affected in par-4 (ts) (Fig. S2C,D). To test whether par-4 acts upstream of the activation of AMPK, we evaluated the effect of par-4(ts) in the background with constitutively active AAK-2 (CA-aak-2). We found that CA-aak-2 in the par-4 (ts) still shows an improved recovery at nonpermissive 25°C at a comparable level with the CA-aak-2 background by itself (Fig. 3I). However, activation of AMPK through AICAR treatment in the par-4(ts) mutant fails to mimic the exercise effect (Fig. S2E–H). This is consistent with the fact that AICAR-mediated activation of AMPK would also at least need PAR-4-kinase activity. Also, only at 15°C, an improvement in posterior touch response was observed in aged animals following AICAR treatment, but not at 25°C (Fig. S2G,H). This supports the role of PAR-4 in AICAR-mediated AMPK activation, similar to previous reports (Shaw et al., 2004). Additionally, the knockdown of pry-1 using RNAi perturbs swimming session-mediated enhancement in functional recovery following axotomy of PLM (Fig. 3J). Overall, these results showed that PAR-4/LKB-1 and its scaffold PRY-1/axin are required for swimming exercise-mediated enhancement in sensory function in older age and also in behavioral recovery through axon regeneration.
Transcriptional regulators, DAF-16 and MDT-15, mediate AMPK's beneficial effects after swimming exercise
The activation of AMPK following physical exercise leads to an adaptive response by regulating transcription factors along with acute metabolic changes (Canto and Auwerx, 2010; Herzig and Shaw, 2018; Trefts and Shaw, 2021). AMPK directly phosphorylates the FOXO3/DAF-16 transcription factor in mammals and C. elegans (Greer et al., 2007a,b). In mice, AMPK also directly phosphorylate the CREB family transcription factor (CRH-1 in the worm) and histone deacetylase HDAC5 (HDA-4 in the worm; McGee et al., 2008; Thomson et al., 2008; Cho et al., 2013). AMPK also regulates mitochondrial biogenesis and transcription by directly phosphorylating peroxisome-proliferator-activated receptor γ coactivator 1α (PGC-1α), an important transcriptional coactivator (Jager et al., 2007). In worms, a direct orthologue of PGC-1α is not reported, but MDT-15 acts as a functional orthologue (Taubert et al., 2006). We assessed these molecules’ role using loss-of-function mutants in swimming exercise-mediated functional improvement (Fig. 4A). The absence of either crh-1 or hda-4 did not affect the enhancement in recovery index through swimming exercise (Fig. 4B), ruling out the roles of CREB and HDAC5 in this process. However, the swimming-mediated improvement in behavioral recovery is not observed in daf-16 and mdt-15 mutants (Fig. 4B). The enhanced regrowth at 10 h postaxotomy after swimming exercise is not observed in both the daf-16 (lf) and mdt-15 (lf) mutants unlike the wild type while the basal level regrowth is comparable (Fig. 4C,D). The swimming quality represented by thrashing frequency is comparable with the wild type for both groups (Fig. S3A). Additionally, the loss in recovery of PTRI in old daf-16 (lf) after a comparable swimming session (Fig. S3B–D) strengthened the role of DAF-16 being a mediator of exercise's beneficial effects. Impaired swimming in the aged mdt-15 (lf) animals resulted in exclusion from this paradigm. Additionally, the AICAR treatment failed to enhance the recovery index in the daf-16 (lf) and mdt-15 (lf) animals, indicating the roles of these transcriptional regulators downstream to AMPK (Fig. 4E). Also, AMPK agonist treatment does not enhance the PTRI in aged animals (Fig. S3E,F). These results indicate that DAF-16 and MDT-15 act as downstream effectors of AMPK in the context of swimming exercise.
DAF-16 and MDT-15 are required for swimming-mediated enhanced functional recovery after PLM axon injury. A, Schematics showing important downstream effectors of AMPK which may mediate the positive effects of swimming for functional improvement. B, Bar graph showing functional recovery represented as the recovery index for the mentioned genetic mutant worms after a 90 min swimming session. Axotomy was performed on L4 stage and regeneration analysis on day 1 adulthood stage. N = 3–4 independent replicates, n = 23–37 number of worms tested. C, Image showing the axonal regrowth at 10 h postaxotomy with or without swimming exercise for daf-16 (lf) and mdt-15 (lf) worms. The orange arrow shows the position of the axotomy. Scale bar, 20 µm. D, Quantification of regrowth length at 10 h postaxotomy as represented in C. N = 3–4, n = 17–29. E, AICAR (1 mM) treatment does not improve the functional recovery of daf-16 (lf) and mdt-15 (lf) worms. N = 3–4, n = 16–45. Statistics, for B, D, E, *p < 0.05, **p < 0.01 ANOVA with Tukey's multiple-comparison test. Error bars represent SD; ns, not significant.
DAF-16 is required in both muscles and neurons, while MDT-15 only in neurons for improved functional recovery after swimming exercise
The tissue-specific rescue experiments suggested that AMPK is required in both muscles and neurons for exercise-mediated improved functional recovery (Kumar et al., 2021). To understand the tissue-specific roles of the transcription factors that act downstream of AMPK signaling in the swimming exercise, we performed muscle and neuron-specific RNAi of daf-16 and mdt-15 (Fig. 5A; Qadota et al., 2007; Calixto et al., 2010). The RNAi protocol is efficient for knocking down targeted gene as evident from the drastic intensity reduction of endogenously tagged DAF-16 fluorescence reporter in all tissue in case of global RNAi and only in muscle for muscle-specific RNAi (Fig. S4A,B). The functional recovery after axotomy was not enhanced through swimming paradigm when muscle-specific RNAi was done for aak-2 and daf-16 unlike the control vector (L4440) group (Fig. 5B). On the other hand, RNAi of mdt-15 in body wall muscle did not affect the enhanced functional recovery due to swimming (Fig. 5B). The aging paradigm also complemented this finding (Fig. S4C,D). The knockdown of upstream kinase LKB1/PAR-4, the scaffold protein for PAR-4, axin/PRY-1, and AAK-2 in muscle did not improve the PTRI of old animals upon swimming session (Fig. S4D). These results also showed that knocking down DAF-16 but not MDT-15 in muscle inhibits the swimming-mediated PTRI improvement (Fig. S4D).
DAF-16 acts in body wall muscle and neuron, whereas MDT-15 acts only in neuron for swimming-mediated enhancement in functional recovery after axon injury. A, Paradigm schematics showing assessment of functional recovery after axotomy of L4 worms in RNAi condition. B, Bar graph showing the recovery index for vector control (L4440), aak-2, daf-16, and mdt-15 muscle-specific RNAi at 24 h postaxotomy. Axotomy was performed at the L4 stage. Muscle-specific RNAi strain rde-1(ne219); kzIs20 [(Phlh-1::rde-1 + sur-5p::NLS::GFP); tbIS222 (Pmec4:: mCherry)]. N = 3 independent replicates, n = 17–37 number of worms tested. C, Bar graph showing the recovery index for vector control (L4440), aak-2, par-4, daf-16, and mdt-15 neuron-specific RNAi at 24 h postaxotomy. Axotomy was performed at the L4 stage. Neuron-specific RNAi strain: [sid-1(pk3321) him-5(e1490); lin-15B(n744); uIs72 (pCFJ90 (Pmyo-2::mCherry) + Punc-119::sid-1 + Pmec-18::mec-18::GFP)]. N = 3, n = 16–32. Statistics, for B, C **p < 0.01 unpaired t test. Error bars represent SD; ns, not significant.
Similarly, neuronal knockdown of both aak-2, par-4, and daf-16 hindered the improvement in recovery index through swimming session after axonal injury as compared with the control condition (Fig. 5C). Similar observation was made upon neuronal RNAi of mdt-15 (Fig. 5C). These RNAi experiments conclude that DAF-16 has a broad role across both neuronal and muscle tissues whereas MDT-15 has a neuron-specific role for swimming-mediated improved recovery.
Neuronal functions of both DAF-16 and MDT-15 are required for accelerated regrowth and ventral targeting due to swimming exercise
Since functional recovery after axotomy is contributed by multiple parameters such as accelerated regrowth, ventral targeting and fusion event, we wanted to understand the tissue-specific functions of DAF-16 and MDT-15 which might be influencing the regeneration parameters. We found that the enhanced axon regrowth at 10 h postaxotomy due to swimming exercise remained unaffected by the muscle-specific knockdown of either daf-16 or mdt-15 (Fig. 6A,B). However, neuronal knockdown of each of daf-16 and mdt-15 perturbs the enhanced regrowth mediated by swimming (Fig. 6C).
DAF-16 regulates swimming-induced axon regeneration from body wall muscle and neuron, whereas MDT-15 from the neuron. A, Schematics showing paradigm for the analysis of regeneration parameters. B, Regrowth length at 10 h postaxotomy in muscle-specific RNAi condition with or without a swimming session. N = 3 independent replicates, n = 10–14 number of worms tested. C, Regrowth length at 10 h postaxotomy in neuron-specific RNAi condition with or without a swimming session. N = 3, n = 10–15. D, Percentage of ventral targeting events in muscle-specific RNAi condition with or without a swimming session. Ventral targeting is the type of regeneration in which the regenerated PLM axon is targeted toward the ventral nerve cord where developmentally synaptic connection occurs. N = 3, n > 20 worms tested. E, Percentage of ventral targeting events in neuron-specific RNAi condition with or without a swimming session. N = 3, n > 20 worms tested. F, Percentage of fusion observed in muscle-specific RNAi condition with or without swimming exercise. G, Percentage of fusion observed in neuron-specific RNAi condition with or without swimming exercise. N = 3, n > 20 worms tested. H, The paradigm of AICAR treatment for the axotomized RNAi worms. I, Recovery index of the AICAR-treated neuronal aak-2 knockdown worms. N = 3, n = 13–28. Muscle-specific RNAi strain rde-1(ne219); kzIs20 [(Phlh-1::rde-1 + sur-5p::NLS::GFP); tbIS222 (Pmec4:: mCherry)]. Neuron-specific RNAi strain: [sid-1(pk3321) him-5(e1490); lin-15B(n744); uIs72 (pCFJ90 (Pmyo-2::mCherry) + Punc-119::sid-1 + Pmec-18::mec-18::GFP)]. Statistics, for B, C, I *p < 0.05 unpaired t test. For D, E, F, and G *p < 0.05, ***p < 0.001 Fisher's exact test. Error bars represent SD; ns, not significant.
Muscle-specific knockdown of aak-2 and mdt-15 did not attenuate the exercise-induced enhancement of ventral targeting, which remained comparable with the vector control (Fig. 6D). In contrast, neuronal knockdown of these genes abolished this exercise-mediated enhancement (Fig. 6E). However, the muscle-specific knockdown of daf-16 significantly reduces improvement in ventral targeting events due to swimming as opposed to its neuronal knockdown (Fig. 6D,E). The enhanced fusion events due to exercise session were perturbed in both neuronal and muscle-specific knockdown of daf-16 and aak-2. However, knockdown of mdt-15 only in neuron perturbed the fusion events (Fig. 6F,G).
These results suggest that the transcriptional axis of AMPK signaling involving DAF-16 acts both in neuron and muscle to contribute to the enhanced axon regrowth, fusion, and ventral axon targeting, whereas MDT-15 acts exclusively in neuron to impart various regeneration parameters.
Since our findings implicate differential axis of AMPK in two tissues, we speculated that these two axes could be epistatic or parallel for this process. To address this, we treated the animals that underwent neuronal RNAi for aak-2 with AICAR after axotomy (Fig. 6H,I). We saw that AICAR treatment could still improve functional recovery in this background (Fig. 6I). This result strongly suggested that the non-neuronal function of AMPK can act as a parallel axis to boost the functional recovery after axon regeneration.
Swimming session promotes nuclear entry of DAF-16 through AMPK
To understand whether the swimming session directly affects DAF-16, we looked at the localization of the CRISPR-tagged endogenous DAF-16::GFP (Zhang et al., 2022) before and after an exercise session. Usually DAF-16: GFP is localized in many tissues, including neurons, muscle, hypodermis, and intestine (Fig. 7A) as reported previously. We found that following a 90 min swimming session, the nuclear localization pattern becomes prominent in various tissues, including muscle cells, as compared with the control condition (Yellow arrows, Fig. 7A; Fig. S5). The number of round-shaped nuclear spots gets significantly increased in the post-swimming group in the wild-type background. The same experiment in aak-2 mutant background revealed that the basal level of round-shaped nuclear spots in muscle is not changed (Fig. 7A,B). However, the swimming-induced increase in such spots was abrogated in aak-2(lf) background (Fig. 7A,B).
Swimming exercise induces DAF-16 nuclear localization in muscle and PLM neuron through AMPK. A, Representative image showing enhanced DAF-16 nuclear localization after swimming exercise. DAF-16 localization is visualized by CRISPR-tagged GFP. The yellow arrows show the DAF-16 nuclear localization in muscle which is significantly higher with swimming. Increased nuclear localization of DAF-16 in muscle is not observed in the absence of AMPK. Body wall muscles are represented in magenta through the expression of mRuby. The yellow arrowheads mark the DAF-16 localized to the nuclei of somatic (intestine, hypodermis, neuron) cells. Orange arrowheads depict the puncta caused by nonspecific autofluorescence. Scale bar, 50 µm. B, Quantification showing the number of muscle nuclei observed after swimming exercise in wild-type and aak-2 (lf) genetic background. N = 3 independent replicates, n =10–18 number of worms tested. C, Representative images showing DAF-16 localization in the PLM neuron cell body. The yellow arrow points to the nuclear region of the cell body, which is mostly devoid of fluorescence signal in the control condition, as opposed to the swimming group. Scale bar, 10 µm. D, Ratiometric quantification of mkate2 to GFP in the center region of PLM cell body, N = 3, n = 12–17. Statistics, for B and D **p < 0.01, ***p < 0.001 ANOVA with Tukey's multiple-comparison test. Error bars represent SD; ns, not significant.
Similarly, we also assayed the effect of swimming exercise on the localization of DAF-16:mKate2 reporter (Aghayeva et al., 2020) in the PLM neuron before and after the swimming session (Fig. 7C,D). We noticed that the DAF-16: mKate2 is localized in the PLM cell body (Fig. 7C, dotted trace), with an exclusion from nucleus (Fig. 7C, yellow arrow, round hollow spot devoid of fluorescence). However, following a swimming session, we saw an increase in DAF-16: mKate2 localization in the nucleus (Fig. 7C, yellow arrow). Diffusible GFP expressed in touch neuron was used as a control to quantify the relative change in DAF-16::mKate2 localization in the central region of the cell body of PLM where usually nucleus is located. We found that in the wild-type background, the ratio of DAF-16::mKate2 versus GFP in this region is increased in the post-swimming group as compared with the control (Fig. 7D). Such an increment was not observed in the aak-2 mutant (Fig. 7C,D). This finding shows that DAF-16 activity following swimming exercise is regulated by AMPK signaling. Perhaps nuclear entry of DAF-16 is triggered by the activation of AMPK following a swim session.
Discussion
In this study, we have shown that cellular ATP level is dropped in response to swimming exercise, which in turn activates the metabolic energy sensor AMP Kinase in a Liver Kinase B-1/PAR-4-dependent manner. This activation of AMPK/AAK-2 engages two signaling axes, one in neuron and the other in muscle tissues. The FOXO family transcription factor DAF-16 acts both in neurons and muscle, whereas PGCα/MDT-15 acts only in neurons to accelerate regrowth from the injured axon tip, promote fusion between the proximal and distal axon, and enhance proper targeting of regenerated axon toward its target area. Therefore, physical exercise exerts multiple regeneration-promoting effects leading to enhanced functional recovery following PLM axon injury (Fig. 8). Excitingly, both of these signaling arms could be activated by AMPK agonist AICAR as an exercise mimetic. Therefore, AMPK signaling in body wall muscle and neurons plays an integrative role for swimming-mediated enhancement in axon regeneration.
Schematics showing the summary of the AMPK signaling components required in body wall muscle and neuron for enhanced functional recovery after swimming exercise.
AMPK is a master regulator for maintaining cellular energetic balance by regulating ATP production and expenditure (Herzig and Shaw, 2018). AMPK activation after exercise has been linked to the regulation of metabolism and health benefits (Spaulding and Yan, 2022). Evidence suggests this role is evolutionarily conserved. In C. elegans, swimming exercise robustly increases levels of phosphorylated AMPK (p-AMPK; Chen et al., 2023). Furthermore, functional analyses indicate that AMPK is required for two key exercise-mediated outcomes: the enhancement of axon regeneration after injury and the preservation of physical fitness during aging (Kumar et al., 2021; Campos et al., 2023). In this study, we observed a significant ATP reduction following a 90 min swimming session in the body wall muscle and the neuron. Further, our data using the AMPK sensor also pointed to the AMPK activation in the muscle just after the swimming session. This strongly suggested a role of AMPK downstream of physical exercise.
We found that acute activation of AMPK through AICAR treatment improved functional recovery after axon injury mimicking swimming exercise. In contrast, inhibition of AMPK activation through dorsomorphin blocked such enhancement. This strengthens the role of AMPK as a major regulatory hub for exercise-mediated beneficial effects. More interestingly, it suggests that AMPK activation through an AICAR-like agonist is a promising strategy to improve axon regeneration following an injury. AICAR treatment in mice leads to improvement in cognition and motor coordination, both in the young and aged mice (Kobilo et al., 2014). Further investigation of AICAR treatment in other models can help develop strategies for enhancing axon regeneration.
AMPK activation is a highly regulated process that requires upstream factors including the kinases that phosphorylate it at the T172 position. Using a temperature-sensitive mutant of PAR-4/LKB-1, we found that it is required for swimming-associated improvement in both aging and axon regeneration. Upregulation of LKB-1 in the mouse CNS resulted in improved axon regeneration and enhanced locomotor functional recovery partly through AMPK (Ohtake et al., 2019). Our finding that PAR-4 is required for swimming-mediated recovery in the posterior gentle touch response supports its axon regeneration-promoting role.
Activated AMPK restores energy balance by acutely regulating metabolism and initiating an adaptive response. AMPK regulates the activity of diverse downstream molecules through direct phosphorylation (Steinberg and Hardie, 2023). We found that DAF-16 (orthologue of FOXO) and MDT-15 (functional orthologue of PGC1α) are important downstream effectors for swimming-mediated enhancement of axon regeneration. Consistent with this observation, we found a significant enhancement of nuclear localization of DAF-16 in muscle and neuron in a AMPK-dependent manner following a swimming exercise. Previous reports also show increased DAF-16 localization in the nucleus after exercise (Chaudhari and Kipreos, 2017; Chen et al., 2023). AMPK regulates DAF-16 activity through direct phosphorylation at multiple sites (Greer et al., 2007a). Our finding supports a mechanistic model in which exercise-induced AMPK activation leads to DAF-16 activation resulting in improved functional recovery. Swimming exercise improves the fitness of aged animals by active regulation of mitochondrial dynamics (Campos et al., 2023). In aged animals lacking the insulin receptor/DAF-2, DAF-16 activation promotes preservation of muscle mitochondrial morphology, higher mitochondrial abundance, and elevated intracellular ATP (Wang et al., 2019). Swimming exercise also induces autophagy in worms and DAF-16 is linked to this process (Campos et al., 2023; Chen et al., 2023). Mitochondria and autophagy both regulates axon regeneration potential in rodents and worms (Cartoni et al., 2016; Han et al., 2016; Ko et al., 2020). Transcriptional regulation of dlk-1 by DAF-16, which codes for the Dual Leucine Zipper Kinase DLK-1 enhances axon regrowth in older worms (Byrne et al., 2014). DAF-16 promotes transcription of many other neuronal genes involved in cellular transport, axon guidance, and signaling (Kaletsky et al., 2016). Additionally, DAF-16 regulates netrin-mediated guidance of regenerating axons to the ventral nerve cord for successful behavioral recovery (Basu et al., 2021).
Our finding also shows that neuronal activity of MDT-15 in promoting axon regeneration in response to swimming exercise. MDT-15 acts as a functional orthologue of PGC1α in C. elegans (Taubert et al., 2006). MDT-15 encodes a subunit of the transcriptional mediator complex and regulates fatty acid metabolism genes, fasting-induced gene expression, xenobiotic and heavy metal detoxification genes, and oxidative stress response (Taubert et al., 2006, 2008; Yang et al., 2006; Goh et al., 2014; Lee et al., 2019; Mao et al., 2019; Shomer et al., 2019). Understanding how MDT-15 regulates axon regeneration is an interesting direction to explore further.
We also observed that when AMPK is knocked down in neurons, AICAR treatment could boost the non-neuronal function of AMPK to promote axon regeneration. This opens up an exciting opportunity to integrate the multiple signaling arms of AMPK across various tissues.
Swimming exercise or activation of AMPK through AICAR leads to accelerated axonal regrowth and enhanced axon fusion and ventral targeting events. The improved behavioral recovery is a cumulative effect of all these positive effects. Axonal fusion might be an efficient way of neuronal circuit repair. Although fusion-type repair has not been clearly demonstrated in mammalian models, recent findings provide compelling evidence that fusion could be encouraged in the peripheral nerve of both mouse and human (Nemani et al., 2024; Ko et al., 2025). Also, retargeting and synaptic re-establishment are critical components of behavioral restoration after axon regeneration. Physical exercise through AMPK signaling promotes these events, leading to improved behavioral recovery is therefore relevant toward finding a potential therapeutic strategy.
Footnotes
- Received January 14, 2025.
- Revision received November 25, 2025.
- Accepted January 5, 2026.
-
We thank the NBRP, Japan and Caenorhabditis Genetics Center (CGC) for strains. CGC was supported by the NIH Office of Research Infrastructure Programs (P40 OD010440). We thank Shirshendu Dey and Swagata Dey for maintaining the 2-photon and the micro point UV laser setup, respectively. We thank Arnab Mukhopadhyay for his guidance in RNAi experiments and constructive feedback. We thank Kavinila Selvarasu and Sumit Kumar Mahapatra for the help in RNAi experiments and microinjections, respectively. We thank Sruthy Ravivarma for verifying some of the behavioral experiments in a blinded manner. We thank Dikshalee Bassi and Chakshu Mangal for reading the manuscript and providing feedback. We would like to thank everyone in the lab for their support and discussions. This work was supported by the NBRC core fund from the Department of Biotechnology, The India Alliance DBT Wellcome Senior Fellowship (Grant #IA/S/22/1/506243) to A.G.R. and a grant from Science and Engineering Research Board (SERB: CRG/2019/002194) to A.G.R.
-
The authors declare no competing financial interests.
-
This paper contains supplemental material available at: https://doi.org/10.1523/JNEUROSCI.0124-25.2026
- Correspondence should be addressed to Sibaram Behera at sibaram.b92{at}gmail.com or Anindya Ghosh Roy at anindya{at}nbrc.ac.in.
- Copyright © 2026 the authors
