Regulation of UNC-40/DCC and UNC-6/Netrin by DAF-16 promotes functional rewiring of the injured axon

The adult nervous system is vulnerable to accidental damage and injury. Neuronal injury leads to a wide range of behavioral deficits, such as loss of sensation or fatal paralysis. Regeneration from the injured proximal stump and subsequent growth towards the target tissues can lead to a functional recovery (He and Jin, 2016; Laha et al., 2017). For example, after axotomy, Caenorhabditis elegans motor neurons can reach the target muscles, which leads to a recovery in locomotion (Yanik et al., 2004; El Bejjani and Hammarlund, 2012). Similarly, in fish, swimming behavior is restored as the axon regrows after spinal cord injury (Becker et al., 1997; Rasmussen and Sagasti, 2017). Axon regeneration potential in an organism declines with age (Geoffroy et al., 2016; Verdú et al., 1995; Basu et al., 2017) for multiple reasons (Fawcett, 2020). First, the intrinsic capacity of the neurons to regenerate diminishes with age (Verdú et al., 1995), and then the external microenvironment poses challenges to the regeneration of the axon (Brosius Lutz and Barres, 2014). Therefore, functional recovery is limited in adulthood. Manipulation of intrinsic pathways and increasing the neural activity together showed some promising effects in the functional restoration of retinal ganglion cells (Lim et al., 2016; Bei et al., 2016). However, the big question about the limiting factors in functional restoration after the accidental damage of axonal processes is still unanswered.

In adulthood, injured axons face a non-permissive environment for correct navigation of the regenerating growth cone (Fawcett, 2020; Fitch and Silver, 2008; Benowitz and Popovich, 2011). Inhibitory axon guidance cues are upregulated following injury in the central nervous system (Giger et al., 2010; Liu et al., 2008). Although nerve injury can trigger expression of axon guidance ligands Slit, Netrin and their receptors, such as Robo and DCC (Chen et al., 2011, 2020; Wang et al., 2019), it is unclear how they are regulated during regeneration. For functional recovery to occur, a regrowing axon must integrate into a neuronal circuit through the formation of proper synapses. Several lines of evidence suggest that the injured neurons can regenerate a synapse (Vidal-Sanz et al., 1987; Oliphint et al., 2010), although it is unclear whether the synaptic transmission can take place through these synapses. Furthermore, it is challenging to correlate behavioral recovery with the anatomical features of axon regeneration at a single neuron level.

Experiments using model organisms have been useful in deciphering the molecular mechanisms underlying the neuronal response to injury (Richardson and Shen, 2019; He and Jin, 2016). The molecular players essential for the regrowth of injured axons have been identified in the past decade (Mahar and Cavalli, 2018; Fawcett and Verhaagen, 2018; Blanquie and Bradke, 2018; Richardson and Shen, 2019). C. elegans has been established as an efficient model for studying the mechanism of axonal regeneration (He and Jin, 2016). Conserved Ca²⁺/cAMP cascade and the Dual leucine zipper kinase (DLK-1) pathway are essential for regrowth (Ghosh-Roy et al., 2010; Hammarlund et al., 2009; Yan et al., 2009). Many other permissive and inhibitory factors have been identified using motor and sensory neurons (Nix et al., 2014; Chen et al., 2011; Kim et al., 2018). For example, the axon regeneration potential of motor neurons declines with age, and is controlled by insulin/IGF signaling pathway (IIS) (Byrne et al., 2014). A clear understanding of the link between axon regeneration and functional rewiring is still lacking. Axotomy of DA9 motor neurons leads to the formation of synapse-like structures in dendritic locations, and behavioral recovery was abnormal (Ding and Hammarlund, 2018). On the other hand, after axotomy of the posterior lateral microtubule (PLM) neuron, self-fusion between the proximal and distal fragments leads to a rapid functional restoration (Basu et al., 2017; Abay et al., 2017). It is unclear whether the injured axons that cannot fuse to their distal counterparts could reach the correct destination and form a synapse. Also, it is unclear whether these events would lead to functional recovery, and mechanistic details underlying these processes remain elusive.

Here, using PLM neuron as a model, we correlated the anatomical features of regrowth with functional recovery at the single neuron level. We found that the injured proximal ends, which could reach the ventral nerve cord (VNC) and form a synapse-like structure, contribute to recovery in the touch sensation. Efficient ‘ventral targeting’ and functional restoration are dependent on the transcription factor DAF-16, which upregulated the Netrin-receptor UNC-40 in the neurons in response to axotomy. In contrast, IIS contributed to the age-related decline in the expression of Netrin/unc-6 in ventral muscle in a DAF-16-dependent manner. Our work elucidates how axon guidance cues are regulated by insulin signaling towards the functional rewiring of an injured neuron.

Among the six touch receptor neurons, two PLMs are responsible for sensing the gentle touch in the posterior part of C. elegans (Bounoutas and Chalfie, 2007; Basu et al., 2017; Chalfie et al., 1985). A backward moving worm responds to a posterior touch by moving forward. During this response, the PLM neuron inhibits backward movement through an inhibitory chemical synapse to AVA and generates forward movement through an electrical synapse to PVC (Chalfie et al., 1985; Wicks and Rankin, 1995) (Fig. 1A). Previous studies have demonstrated that an axotomized PLM neuron displays two different patterns of regeneration (Fig. 1A) (Basu et al., 2017; Ghosh-Roy et al., 2010; Neumann et al., 2011). In the first category, the regrowing proximal axon can recognize and fuse with its distal part, known as ‘fusion’. This self-fusion phenomenon correlates with the rapid restoration of lost touch sensation following axotomy (Basu et al., 2017; Abay et al., 2017) (Fig. 1B). In the second category, the proximal end regrows without fusing with the distal part, named ‘non-fusion’ events. In this case, the distal end often undergoes degeneration (Ghosh-Roy et al., 2010; Neumann et al., 2011), and functional recovery was not seen at 24 h post-axotomy (Basu et al., 2017) (Fig. 1B). We investigated whether this ‘non-fusion’ class of regrowth events would eventually lead to a functional recovery at later time points. We found that the posterior touch response index (PTRI) at 48 h post-axotomy was significantly raised to a value of 0.63±0.23 (mean±s.d.) from 0.48±0.16, obtained at 24 h post-axotomy (Fig. 1B; ***P<0.001, Tukey's multiple comparison test). We noticed that some of the events in the whole cohort of the ‘non-fusion’ category showed PTRI values comparable with that of the ‘fusion’ category (Fig. 1B). We speculated that these events might represent the regrowth events with correct targeting. We noticed that the proximal part often regrew towards the ventral cord by making a branch (arrowhead, Fig. 1Ca) or turning towards the ventral side (arrowhead, Fig. 1Cb). We classified these events as ‘ventral targeting’ (Fig. 1C). In other ‘non-fusion’ events, the proximal axon regrew either towards the dorsal side (Fig. 1Cc) or towards the anterior direction (Fig. 1A), or sprouted multiple small branches (Fig. 1Cd). To compare the functional recovery across various regrowth patterns, we used the ‘recovery index’ (RI) as the ratio of the ‘PTRI at 24 h or 48 h post-axotomy’ to the ‘PTRI just after axotomy’. Any RI value above 1 (green dotted line, Fig. 1D) indicates a sign of recovery. We noticed that the RIs of ‘ventral targeting’ events at 24 h and 48 h were 1.69±0.78 and 2.22±1.09 (mean±s.d.), respectively (Fig. 1D), whereas the RIs corresponding to all the other categories were below 1 (Fig. 1D), indicating no functional restoration. Overall, the RI of the ‘non-ventral’ events was significantly lower than the ‘ventral targeting’ events (Fig. S1A). The recovery values corresponding to the ‘ventral’ category were comparable with that of the ‘fusion’ class at 48 h post-axotomy (Fig. 1D). Moreover, we saw that the percentage of ‘ventral targeting’ events increased from 27% at 24 h to 45% at 48 h post-axotomy (**P<0.01, Fisher's exact test). This possibly explains the overall increase in the PTRI values at 48 h in the ‘non-fusion’ events (Fig. 1B).

The PLM neuron has a ventral branch (arrowhead, Fig. 1A), which makes en passant synapse to AVA interneuron at the VNC (Schaefer et al., 2000; White et al., 1986). We speculated that the ‘ventral targeting’ might lead to an enrichment of the presynaptic proteins at the tip of the regrowing axon. Therefore, we analyzed the localization of presynaptic reporter GFP::RAB-3 (Luo et al., 2014) in different regrowth events. In the case of ‘ventral targeting’ events, we saw an accumulation of GFP::RAB-3 at the VNC (white arrowheads, Fig. 1E). This accumulation pattern resembled the RAB-3 punctae seen at the end of the synaptic branch of PLM neuron (arrowheads, Fig. 1F), which we termed as ‘developmental synapse’. In the case of a developmental synapse, before axotomy, the ratio of GFP::RAB-3 versus a cytosolic marker mScarlet was 1.70±0.61 (mean±s.d.) at the region of interest (ROI) placed near the synapse (rectangular ROI, Fig. 1F,G). This value was comparable with the ratio obtained for the ‘ventral targeting’ events during regeneration (Fig. 1E,G). The normalized value of GFP::RAB-3 at the tip of the regrowing axon in the ‘non-ventral’ class (Fig. S1B) was significantly lower than the ‘ventral targeting’ events (Fig. 1G). The RI corresponding to the ‘ventral targeting’ events with the RAB-3 enrichment in the transgenic background expressing GFP::RAB-3 was 1.83±0.84, which was significantly higher than the value obtained in the case of ‘non-ventral’ events (Fig. S1C). This phenomenon was similar to what has been seen in muIs32 [Pmec-7-GFP] transgenic background (Fig. 1D). This suggested that the functional recovery related to ‘ventral targeting’ of the injured PLM axon is not the artefact of the transgenic reporter. Another presynaptic reporter, ELKS-1::TagRFP (Zheng et al., 2014), also showed an enrichment at the VNC in the case of ‘ventral targeting’ events (arrowheads, Fig. S1D). Similarly, the postsynaptic receptor GLR-1::GFP (Hoerndli et al., 2015) aligned along the ventral cord in the ‘ventrally targeted’ axon (arrowheads, Fig. S1E). This helped us to conclude that the regrowth towards the ventral side and enrichment of presynaptic machinery at the ventral cord leads to the functional recovery during axon regeneration.

Axon regeneration potential in adulthood is limited (Verdú et al., 1995; Kang and Lichtman, 2013; Graciarena et al., 2014). A previous study in touch neurons showed that functional restoration through axonal self-fusion decreases with age (Basu et al., 2017). Therefore, we investigated whether the accuracy in axon guidance during regrowth would decrease in adulthood. We found that the RI corresponding to the ‘non-fusion’ events gradually decreased with age (Fig. 2A). The RI at A3 stage became 0.91±0.57 (mean±s.d.), which is significantly lower than the value obtained at L4 stage (***P<0.001, Tukey's multiple comparison test) (Fig. 2A). Investigation of the regrowth pattern revealed that the percentage of ‘ventral targeting’ events was significantly decreased when axotomy was carried out at the A3 stage (Fig. 2B), although the total regrowth was unperturbed (Fig. 2C). Moreover, the total regrowth length in the ‘non-ventral’ and ‘ventral’ class was comparable (Fig. 2C). Another phenomenon we noticed was that the RI corresponding to the ‘ventral targeting’ events at the A3 stage was significantly decreased compared with the L4 stage (**P<0.01, Tukey's multiple comparison test; Fig. 2D). This suggested that ‘ventral targeting’ is not enough for functional recovery in an older animal.

We saw that at the A3 stage, in case the regrowing axons reached the VNC, the longitudinal growth along the cord (double-headed yellow arrows, Fig. 2E) was less compared with that at the L4 stage (Fig. 2F). Moreover, the ratio of mCherry::RAB-3 (Sood et al., 2018) versus GFP in the rectangular ROI placed at the VNC (Fig. 2E) was significantly decreased at the A3 stage (***P<0.001, Tukey's multiple comparison test; Fig. 2G). In the background expressing mCherry::RAB-3 reporter, the percentage of ‘ventral targeting’ and the RI corresponding to the ‘ventral targeting’ event significantly decreased at the A3 stage (Fig. S2A,B). This observation strengthens the conclusion that the age-related decrease in ventral guidance during axon regeneration is not an artefact of transgenic background.

Efforts in the past decade helped to identify the pathways regulating the axon regeneration potential in the worm (Nix et al., 2014; Chen et al., 2011; Kim et al., 2018). However, how axon regeneration pathways control functional rewiring has not yet been addressed. Removal of the cell-intrinsic inhibitory pathways enhances the regenerative outgrowth following axotomy (He and Jin, 2016). We investigated whether the mutants with enhanced axon regrowth ability (Chen et al., 2011; Nix et al., 2014; Knowlton et al., 2017; Zou et al., 2013; Kosmaczewski et al., 2015; Hubert et al., 2014; Alam et al., 2016; Table S1) would have better functional recovery. We also tested the mutants with enhanced lifespan (Murphy and Hu, 2013; Byrne et al., 2014; Taub et al., 2018; Senchuk et al., 2018) and healthspan (Bansal et al., 2015; Yuan et al., 2012; Table S1), as their role in touch neuron regeneration is not clear yet. We compared the recovery indices of ‘non-fusion’ events at L4 and A3 stages to find the possible regulators for axon regeneration in adulthood. Although some of the mutants showed an enhanced functional restoration in the L4 stage (Fig. S3A), the recovery indices in those mutants at A3 stage were not significantly different from the value obtained in the wild-type control (Fig. S3B). There were a few exceptions, such as akt-1(lf), sgk-1(lf), rsks-1(lf) and let-7(lf), which showed an RI above 2. This was significantly higher than the RI in wild type at L4 and A3 (Fig. S3A,B).

The AKT-1 kinase acts downstream to the insulin receptor DAF-2 (Fig. 3A) to regulate the lifespan of C. elegans (Murphy and Hu, 2013). It was observed that the insulin signaling downregulates the axon regrowth capacity of D-type motor neuron in older animals (Byrne et al., 2014). Therefore, we tested daf-2 (e1368ts) and the mutants that inactivate downstream kinases in IIS, such as age-1, pdk-1, akt-1 and akt-2. All of these mutants showed significantly enhanced functional recovery (***P<0.001, Tukey's multiple comparison test) compared with the wild-type control at A3 stage (Fig. 3B). Reduced IIS promotes longevity through the upregulation of the transcriptional activity of DAF-16 (Libina et al., 2003; Murphy and Hu, 2013) (Fig. 3A). Consistently, we found that the enhanced RI seen in the daf-2 mutant was completely dependent on daf-16 (Fig. 3B). We examined whether the enhanced functional recovery due to loss of IIS correlates with the enhanced ‘ventral targeting’. The mutants for IIS displayed a higher percentage of ‘ventral targeting’ events than wild type (Fig. 3C). Additionally, the recovery indices corresponding to the ‘ventral targeting’ events were significantly higher in these mutants (Fig. S3C). In daf-2 mutant, the longitudinal growth along the VNC was 53.50±14.27 µm (mean±s.d.) compared with 14.0±8.34 µm in wild type (***P<0.001, Mann–Whitney comparison t-test). More RAB-3 punctae were accumulated along the VNC in daf-2(lf) (arrowheads, Fig. 3D). A similar observation was also made in the mutants of IIS at L4 stage. For example, both the RI and the percentage of ‘ventral targeting’ events were significantly enhanced in these mutants, even at L4 stage (Fig. 3E,F). Therefore, IIS controls the guidance of the regrowing PLM axon irrespective of age. The percentage of ‘ventral targeting’ was significantly reduced in the daf-16 single mutant background at L4 stage (***P<0.001, Fisher's exact test; Fig. 3G), although total regrowth was not reduced (Fig. S3D). Often, in daf-16(lf), we noticed a multibranch regrowth pattern (Fig. 3H) that did not sufficiently grow towards the VNC. Consistent with these data, the RI in daf-16(lf) at the L4 stage was significantly less than the wild type (Fig. 3I). Therefore, our observation indicated that FOXO family transcription factor DAF-16 is required for ventral guidance and functional restoration independent of insulin signaling.

As DAF-16 is expressed in multiple tissues, including neurons, it was important to test the tissue-specific requirement of this protein in the axon regrowth of PLM neuron. As daf-16(mu86) affects both the ‘a’ and ‘f’ isoform of daf-16 (Chen et al., 2015), we expressed both isoforms for the rescue experiments. We expressed them in daf-16-expressing cells, intestine, neurons, muscle and epidermis in the daf-16 mutant background. The recovery indices due to pan-neuronal and touch neuron-specific expression of daf-16f in daf-16(lf) were 1.50±1.14 and 1.74±1.34 (mean±s.d.), respectively. These values were significantly higher (***P<0.001, Tukey's multiple comparison test) than the daf-16 mutant (Fig. 4A). The drop in the percentage of ‘ventral targeting’ was also rescued by the neuron-specific expression of daf-16f (Fig. 4B). Similarly, the neuron-specific expression of daf-2 significantly rescued the enhanced functional restoration and ‘ventral targeting’ events seen in daf-2(m596) (Fig. S4A,B). This was consistent with the previous report on motor neuron regeneration (Byrne et al., 2014). However, surprisingly, the muscle-specific expression of daf-16f could also rescue the phenotype in daf-16(lf) to a similar extent (Fig. 4A,B). Simultaneous expression of the f isoform in touch neuron and muscle resulted in a synergistic increase in the RI and ‘ventral targeting’, which was akin to the rescue of daf-16(lf) using its own promoter (Fig. 4A,B). The lack of longitudinal regrowth along the ventral cord in daf-16(lf) was rescued by both neuron and muscle-specific expression of daf-16f (Fig. 4C,D). Expression of daf-16f in the intestine or in the epidermis did not result in any improvement in the phenotype in the daf-16 mutant (Fig. 4A,B). Neuron-specific expression of the daf-16a isoform also rescued the ‘ventral targeting’ and RI value in daf-16(lf) (Fig. S4C,D). However, when daf-16a isoform was driven in muscle it showed a milder rescue, indicating that the N-terminal region of the f isoform was needed for the optimal function in muscle cells. Tissue-specific rescue experiments suggested that the functional rewiring process is independent of IGF-1/IIS activity in the intestine. However, the function of IIS in axon regeneration is not solely neuron intrinsic as found in motor neurons (Byrne et al., 2014). Neuron-intrinsic activity of DAF-16 is facilitated by its muscle-specific activity in the PLM neuron.

Age-dependent decline in the regeneration of D-type motor neuron was correlated to the drop in the expression of dlk-1, which is also a transcriptional target of DAF-16 (Byrne et al., 2014). We saw that DLK-1 is essential for axon regeneration and functional recovery both in wild-type and daf-2(lf) backgrounds (Fig. S4E,F). DLK-1 is sufficient to promote axon regrowth after neuronal injury (Yan et al., 2009; Hammarlund et al., 2009). However, the role of DLK-1 MAP kinase in the guidance of regrowing axons is unclear. We tested whether DLK-1 is also sufficient to promote ‘ventral targeting’ and functional restoration. Upon overexpression of DLK-1 in neuron, axon regrowth was significantly enhanced in both wild-type and daf-16(lf) backgrounds (Fig. 4E,F). However, neither ‘ventral targeting’ nor functional restoration were enhanced in the same backgrounds (Fig. 4G,H). Upon the overexpression of DLK-1, the injured axons regrew straight in the anterior direction (arrowheads, Fig. 4E). Therefore, the percentage of ‘ventral targeting’ events was significantly reduced upon DLK-1 overexpression in the wild-type background (Fig. 4G). This correlated with the reduced value of the RI due to overexpression of DLK-1 (Fig. 4H). This indicated that the ‘ventral targeting’ phenomenon related to the function of DAF-16 probably does not involve its target DLK-1.

DAF-16 regulates many biological processes in adulthood, including longevity by promoting the expression of its transcriptional targets (Murphy, 2006; Kaletsky et al., 2016). The neuronal targets of DAF-16 have been identified by sequencing the mRNAs isolated from neurons in the daf-16 mutant (Kaletsky et al., 2016). The authors found ∼9000 neuronal genes differentially regulated by DAF-16. We performed a gene ontology (GO) analysis on these 9000 genes using the specific GO term (GO:0097485) involving ‘neuronal projection and guidance’, which revealed 110 genes (Table S4). Among these, unc-40 and unc-6, are related to axon guidance. The receptor for the chemoattractant ‘netrin’/UNC-6, also known as ‘Deleted in colorectal cancer’ (DCC)/UNC-40, regulates ventral guidance of growth cone during axon development (Chisholm et al., 2016; Kennedy et al., 1994; Ishii et al., 1992; Chan et al., 1996). The unc-6/unc-40/slt-1 pathway antagonizes the Wnt planar polarity signaling to establish the ventral branch in PLM neuron (Chen et al., 2017). The promoter of the unc-40 gene has the binding site ‘TGTTTA’ for DAF-16 (Fig. 5A), which was also experimentally verified (Gerstein et al., 2010; Celniker et al., 2009). After axotomy, we found that the ‘ventral targeting’ events were drastically reduced in both unc-6 and unc-40 mutants (Fig. 5B); however, axon regrowth was not reduced (Fig. S5A). As these mutants show a partially penetrant phenotype of having no branch (Fig. S5B,C), we used the PLMs with a branch for our axotomy experiments. Similarly, the RI was also reduced in these mutants (Fig. 5C). Enhanced ‘ventral targeting’ and RI seen due to the overexpression of daf-16f in the neuron were significantly reduced in the absence of unc-40 or unc-6 in the transgenic background (Fig. 5B,C). Similarly, the loss of unc-6 reduced the enhanced ‘ventral targeting’ and RI obtained due to the overexpression of daf-16f in the muscle (Fig. 5B,C). A similar result was also obtained in akt-1(lf) background (Fig. 5B,C), which upregulates DAF-16 activity (Paradis and Ruvkun, 1998).

We speculated that ventral guidance might be compromised in the daf-16(lf) due to a reduced expression of unc-40. Additionally, in response to axotomy, unc-40 could be upregulated in PLM neuron in a daf-16-dependent manner. We looked at the transcriptional cum translational reporter of UNC-40, Punc-40::UNC-40::GFP (Chan et al., 1996), in wild type and daf-16(lf). We compared the intensity of UNC-40::GFP in the cell body of PLM neuron before and after axotomy, with respect to a diffusible marker mScarlet (Yellow dashed ROIs, Fig. 5D). The normalized intensity of UNC-40::GFP before axotomy was significantly increased in the daf-2(e1368ts) background compared to the wild-type control in a daf-16-dependent manner (Fig. 5D,F). However, the mean intensity of mScarlet was unaffected in the mutant (Fig. S5E). This ratio was significantly lower in daf-16(lf) than wild type before axotomy (Fig. 5D,F). After axotomy, the intensity of UNC-40::GFP in the cell body showed a steady increase with time (Fig. 5D,F; Fig. S5D). Similarly, the UNC-40::GFP intensity gradually increased at the tip of the cut axon following axotomy (rectangular ROIs, Fig. 5E,G; Fig. S5F). However, this axotomy-driven steady increase in UNC-40::GFP, both in the cell body and the axon tip, was significantly perturbed in the daf-16(lf) background (Fig. 5D-G; Fig. S5D-G). Therefore, as the unc-40 level is low, the regrowing axons in the daf-16 mutant may not be sufficiently attracted towards the ventral side in response to ‘Netrin’ cues.

To understand the muscle-specific role of DAF-16 in the ventral guidance of regenerating axons, we speculated that DAF-16 could regulate the guidance cues that are secreted by the muscle. This was also reflected in the transcriptomics data, which suggested that the level of unc-6 was increased due to loss of daf-2 in a daf-16-dependent manner (Kaletsky et al., 2016; Table S4). UNC-6 is expressed and secreted from the ventral muscles (Ishii et al., 1992; Weinberg et al., 2018). We investigated how unc-6 expression is regulated in adulthood using a fosmid reporter of unc-6 (unc-6::SL2::NLS::YFP::H2B), which has been shown to express in ventral muscles (Asakura et al., 2010; Weinberg et al., 2018). The nuclei of the 16 ventral muscles along the VNC showed the localization of this reporter (white arrowheads, Fig. 6A) (Asakura et al., 2010). We named the four pairs of nuclei from tail to vulva as ‘posterior nuclei’, and the rest of the four pairs from the vulva to pharynx as ‘anterior nuclei’ (arrowheads, Fig. 6A). The expression of this reporter was considerably reduced at the A3 stage (arrowheads, Fig. 6B) to negligible levels. Consequently, the number of visible nuclei significantly dropped at the A3 stage (Fig. 6F). Also, in the observable nuclei, unc-6 reporter intensity was reduced at A3 compared with L4 (Fig. 6G). For example, the intensity in M1 muscle dropped significantly at the A3 stage (**P<0.01, Mann–Whitney comparison t-test) (Fig. 6G). A similar observation was also made in the nuclei of the anterior M8 muscle (Fig. 6G).

To test whether the unc-6 expression is regulated by IIS, we looked at the expression of the reporter in daf-2(e1368ts) and daf-2(e1368ts); daf-16(lf) backgrounds. We observed that in a non-permissive temperature, 25°C, the intensity of the reporter in daf-2(e1368ts) was significantly higher than that in wild type at the A3 stage (***P<0.001, Tukey's multiple comparison test) (Fig. 6B,C,H). Next, we found that the intensity of the reporter corresponding to M1, M2 and M8 in daf-2(e1368ts);daf-16(lf) was significantly reduced compared with the daf-2(e1368ts) single mutant (Fig. 6D,H). Further, we noticed that at the L4 stage, there was a significant decrease in the reporter intensity in daf-16(lf) (Fig. 6I). Conversely, the intensity of the reporter was significantly enhanced when daf-16f was overexpressed in muscles (Fig. 6E,I). Overall, we concluded that the expression of unc-6 in muscle is regulated by DAF-16 activity, and this regulation is crucial for the guidance of injured proximal stump towards the VNC.

Comprehensively, our work shows that proper guidance and synapse formation during the regrowth of injured PLM axon leads to functional restoration. The role of the chemoattractant Netrin/UNC-6 and its receptor UNC-40/DCC is crucial for ventral guidance of the injured proximal stump. This process is controlled by the DAF-2-DAF-16 axis acting in both neuron and muscle, but not in the intestine. DAF-16 plays an instructive role in upregulating the expression of UNC-40/DCC in injured PLM. DAF-16 also controls the expression of Netrin/UNC-6 in ventral muscles. In daf-16 mutant, the misguided proximal stump cannot form a synapse-like structure. Therefore, the ventral guidance is an important step in the functional rewiring process in injured PLM neurons (Fig. 7).

This study establishes that the axon regrowth of PLM neuron following axotomy can lead to functional restoration irrespective of self-fusion events between injured proximal and distal ends as described previously (Basu et al., 2017; Abay et al., 2017). Functional restoration correlates with the proper guidance of the proximal ends to the VNC and the accumulation of presynaptic proteins, such as RAB-3 and ELKS-1 at the nerve endings. The enrichment pattern of the presynaptic components is comparable to the original synaptic branch. The localization of postsynaptic protein GLR-1 was also observed at the VNC, juxtaposed by the enrichment of the new presynaptic machinery. This is consistent with the previous observation in the lamprey spinal cord that regeneration of new synapse leads to a functional recovery (Oliphint et al., 2010). Another recent study using DA9 motor neuron showed an abnormal accumulation of synaptic proteins in dendrites after axotomy, which did not support functional recovery (Ding and Hammarlund, 2018). Therefore, PLM would be a good system to find the limiting factors in the rewiring process. However, only 40% of the events corresponds to correct targeting at the L4 stage, which further drops to 15% at the A3 stage. This indicates that there are inhibitory components that need further investigation.

It is intriguing that synapse formation at the VNC promotes functional recovery after axonal injury in PLM neuron. This might apparently raise the question of how a chemical synapse can help restore the functional loss, as the electrical synapse to the PVC neuron is sufficient for the behavioral response to posterior touch (Chalfie et al., 1985). Although the gap junction in PLM plays a driving role in the posterior touch response, the chemical connection to AVA was predicted to pause the backward movement upon applying posterior touch stimuli (Chalfie et al., 1985; Bounoutas and Chalfie, 2007). This was further supported by the evidence that the chemical synapse plays an inhibitory role in a tap-response assay (Wicks and Rankin, 1995). Moreover, laser microsurgery of the synaptic branch alone leads to a 30% reduction in touch response (Basu et al., 2017). Therefore, ‘ventral targeting’ and synapse formation during the regeneration of PLM may help regain the posterior touch sensation.

We found that insulin signaling controls the functional restoration after the axotomy of PLM. Loss of insulin receptor daf-2 or its downstream kinases promotes the functional restoration through enhancement of the ‘ventral targeting’ events, which is dependent on the FOXO family transcription factor DAF-16. This is consistent with the finding that IIS contributes to the age-dependent decline in growth cone formation during the regeneration of D-type motor neuron (Byrne et al., 2014). However, we found that IIS controls the guidance of regenerating axon irrespective of age, and that DAF-16 is essential for proper guidance and functional recovery. A previous study suggested that enhanced regrowth in the daf-2 mutant is dependent on the MAPKKK dual leucine zipper kinase-1 and that the expression of dlk-1 is driven by transcriptional activity of DAF-16 (Byrne et al., 2014). Our data indicate that although DLK-1 is sufficient to promote the regrowth of injured axon, it does not support accuracy in the regrowth. Therefore, we identified a DLK-1-independent step in axon regeneration that is highly coordinated by the axon guidance molecules. We provided evidence that DAF-16 regulates the expression of UNC-40 and UNC-6 in neuron and muscle, respectively. This is also consistent with the previous finding that UNC-6 is required for the accuracy of the post-injury regrowth of AVM neuron (Gabel et al., 2008). DAF-16 plays several roles in neurodevelopment, including axon growth (Christensen et al., 2011; Kennedy et al., 2013; Grossman et al., 2013) and neuronal migration (Kennedy et al., 2013) in both a cell-autonomous and non-autonomous manner. It is also required for neuronal remodeling due to hypoxia (Pocock and Hobert, 2008) and food deprivation (Calixto et al., 2012; Caneo et al., 2019).

We were intrigued to find that DAF-16 is required for the ‘ventral targeting’ of the PLM axon during regeneration, irrespective of DAF-2. The promoter of the unc-40 gene has a binding site for DAF-16 (Gerstein et al., 2010; Celniker et al., 2009). We found that the DAF-16 promotes the expression of UNC-40 after axotomy. How might DAF-16 be regulated upon axonal injury? A previous study indicated that in response to heat stress, C-Jun-N terminal kinase-1 (JNK-1) helps in the translocation of DAF-16 to the nucleus and extends lifespan in a pathway parallel to IIS (Oh et al., 2005). Moreover, the JNK-1 MAP kinase pathway is required for axon regeneration in C. elegans (Li et al., 2012; Nix et al., 2014). Upon axon injury, JNK-1/DLK-1 gets activated by local injury signal, such as cAMP (Hao et al., 2016); therefore, it can activate DAF-16.

Axon guidance cues are often modulated to improve axon regeneration after traumatic injury in the central and peripheral nervous system (Giger et al., 2010; Becker and Becker, 2007). The expression and localization pattern of these cues are different in adulthood as opposed to the pattern seen during early development (Hilton and Bradke, 2017). These cues are expressed aberrantly after the lesion of the nervous system (Giger et al., 2010; Chen et al., 2020; Dun and Parkinson, 2017). Both Netrin and Slit play important roles in axon regeneration after sciatic nerve transection (Dun et al., 2019). Recently, it has been seen that following axon injury, let-7 miRNA negatively regulates the expression of slit and dcc in peripheral neurons (Wang et al., 2019). Our finding gives novel insight into the regulation of axon guidance ligands and receptors by DAF-16.

C. elegans strains were grown on nematode growth medium (NGM) agar plates at 20°C using standard methods (Brenner, 1974). The loss-of-function alleles were denoted as ‘lf’ throughout the paper except for the daf-2 mutant. In this study, we used two temperature-sensitive alleles of daf-2. One of them is daf-2(e1368ts), and the other is daf-2(m596ts). Both daf-2(e1368ts) and daf-2(m596ts) mutants were grown in permissive temperature at 20°C. When L2 staged larva of these mutants are transferred to the non-permissive temperature of 25°C they produce Dauer larvae (Riddle et al., 1981). This characteristic was used to make single and double mutants involving daf-2(lf). All the strains used in the study are described in Table S2. The strains containing the newly generated transgenic extrachromosomal arrays are mentioned in Table S3.

The axotomy experiments were performed either at the late larval (L4) stage or with 3-day-old adult worms (A3 stage). During axotomy experiments, the worms were immobilized using 0.2 µl of 0.1 µm diameter polystyrene beads (Polysciences, 00876-15) in 5% agarose pads (Basu et al., 2017). Axotomy and imaging were performed using a Bruker Ultima two-photon microscope (Basu et al., 2017). The 60× water immersion objective (NA= 1.1) was used for imaging and axotomy of PLMs. The laser wavelengths of 920 nm and 1020 nm were used for imaging the GFP- and mScarlet-labeled PLMs, respectively. The laser wavelength of 720 nm was used for cutting the axon (Fig. 1A). The cutting laser was set at a pulse width of ∼ 80 fs, an irradiation pulse width of 20 ms, a lateral point spread function (PSF) of ∼400 nm, and a z-axis PSF of ∼1.5 μm. For most of the experiments, both the PLM axons located on the left and right sides of the worm were severed, unless otherwise mentioned. The gentle touch response assay was then performed at 3 h and 24 h or 48 h post-axotomy (Fig. 1A). Usually, two small cuts of 6 µm apart were introduced at a 50 μm distance from the cell body of PLM (Fig. 1A; Basu et al., 2017).

The gentle touch response assay (Chalfie et al., 2014; Chalfie and Sulston, 1981; Basu et al., 2017) was performed for checking the function related to the PLM neuron before and after axotomy. The PTRI (Chalfie and Sulston, 1981; Chalfie et al., 1985) corresponding to both PLML and PLMR were obtained by performing the assay for the left and right side of the worm, respectively. According to the method described before, the worms were subjected to the anterior and posterior touches alternatively with an eyelash tip (Chalfie et al., 1985, 2014; Chalfie and Sulston, 1981). Anterior touch was given to a forward-moving worm, and a backward-moving animal was subjected to the posterior touch. In response to both the anterior and posterior touches, the worm reverses its direction (Chalfie and Sulston, 1981). A total of ten alternate anterior and posterior touches were given on either side of the worm (Hobert et al., 1999; Chalfie and Sulston, 1981). The response was denoted as ‘1’ and no response as ‘0’ (Basu et al., 2017). The PTRI was quantified as the ratio of the number of responses scored divided by the total number of touches applied. We calculated the PTRI values at 3 h, 24 h and 48 h post-axotomy for each side of the worm. To express the extent of functional recovery quantitatively, we divided the PTRI value measured at 24 h or 48 h post-axotomy by the PTRI obtained just after axotomy (3 h post-axotomy). The functional recovery was defined as the RI (Fig. 1D; Fig. S1A). Therefore, an RI value above 1 (green dashed line, Fig. 1D) denotes successful functional recovery.

For correlating the RI corresponding to a given side of the worm with the regrowth pattern, we imaged the regeneration event from the same side using a point-scanning confocal microscope LSM510 Meta (Zeiss) after the touch response assay was conducted at 24 h or 48 h post-axotomy (Fig. 1A). The worms were mounted on a 5% agarose pad with 10 mM levamisole hydrochloride (Sigma-Aldrich, L0380000) for imaging (Fig. 1A). Typically z-sections were scanned at 0.5 µm intervals under a 63× oil objective (NA=1.46). Approximately 56% of 480 nm laser and 68% of 543 nm laser were used for imaging the muIs32 [Pmec-7-GFP] and shrEx209 [Pmec-4-mScarlet] reporters, respectively. The captured z-slices were pseudocolored using ImageJ (Fiji) to represent the depth across the z-sections (Fig. 1C). Broadly, the regrowth patterns were divided into ‘fusion’ and ‘non-fusion’ events (Fig. 1A). The axons categorized in the ‘non-fusion’ class followed different trajectories of regrowth (Fig. 1C). The ‘non-fusion’ category was further classified into two subcategories, ‘ventral targeting’ events (Fig. 1Ca,Cb) and ‘non-ventral’ regrowth (Fig. 1Cc,Cd). The PLM neuron has a ventral branch (arrowhead, Fig. 1A), which makes synapse onto the postsynaptic interneuron (Fig. 1A,F; Chen et al., 2017). This allowed us to set the dorsal-ventral axis of the worm (Fig. 1A) while analyzing the direction of regrowth at 24 h or 48 h post-axotomy. The axons that regrew up to 35-45 µm depth in a ventral direction and fasciculated along the VNC were characterized as ‘ventral targeting’ events, whereas the axons regrowing up to the same depth in the dorsal direction were identified as ‘dorsal targeting’ events. When multiple branches emanated from the cut tip, the event was named as a ‘multibranch’ event. When the proximal stump regrew straight towards the anterior side without any deviation (‘non-fusion’ event in Fig. 1A), it was termed as ‘straight regrowth’.

First, the optical sections corresponding to a regeneration event were projected together. Then, the longest neurite length was measured from the cell body to the tip of the regenerated axon. To calculate the actual regrowth length at 24 h or 48 h post-axotomy, the distance from the cell body to the injury site was measured and subtracted from the distance between the cell body and the regrowth tip. We used Image J for calculating the regrowth length of the regenerated PLM axon. The total fasciculation along the ventral cord (Double-headed yellow arrows, Fig. 2E) corresponding to the ‘ventral targeting’ events was measured as ‘longitudinal regrowth along the VNC’ (Fig. 2F).

For the experiments related to the temperature-sensitive daf-2 mutants, the wild-type and mutant worms were grown at the permissive temperature of 20°C until L4 stage, and were then transferred at a non-permissive temperature of 25°C for 3 days to obtain A3 stage (day-3 old) worms for axotomy experiments. After axotomy was performed, the worms were transferred back to 25°C for 24 h or 48 h before the assessment of PTRI and imaging. For performing the experiments with daf-2(e1368ts) mutants at the L4 stage, L3-staged worms were transferred from 20°C to 25°C and were allowed to grow until late L4 stage before the quantitative imaging of UNC-40-GFP intensity in PLM (Fig. 5D-G; Fig. S4D-G).

For the tissue-specific expressions of daf-16f and daf-16a isoforms, we first prepared gateway (Thermo Fisher Scientific) entry clones of daf-16f and daf-16a. We obtained the daf-16 cDNAs from the Yuji Kohara cDNA consortium (https://nematode.nig.ac.jp/dbest/keysrch.html). pCR8::daf-16f (NBRGWY106) and pCR8::daf-16a (NBRGWY96) were prepared by PCR8 cloning. For making pCR8::daf-16f (NBRGWY106), PCR was performed with the primers 5′-GACATGCAAGCGTGGAACTGTCGTG-3′ and 5′-TGATGATTTAATAAAAATCAAATTTGG-3′, using daf-16f cDNA (yk3444n05) as the template, followed by ligation into the pCR8/GW/TOPO vector (Thermo Fisher Scientific, K2500-20). Similarly, for making pCR8::daf-16a (NBRGWY96), PCR was performed with the primers 5′-GGATGAACGACTCAATAGACGACG-3′ and 5′ TGATGATTTAATAAAAATCAAATTTGG-3′ using daf-16a cDNA (yk3193h05) as a template. The pCR8 plasmids were recombined with PCZGY553 [Pmec-4::Gateway destination vector], PCZGY925 [Pmyo-3::Gateway destination vector], PCZGY44 [Pdpy-7::Gateway destination vector] and PCZGY66 [Prgef-1::Gateway destination vector] using LR clonase (Thermo Fisher Scientific, K2500-20) for expressing daf-6a/f in touch neuron, muscle, epidermal cells and all the neurons, respectively. These plasmids were injected either into wild-type or daf-16(lf) background with a concentration of 10 ng/µl. The pttx3-RFP was used as a co-injection plasmid at a concentration of 40 ng/µl mixed with the pBSK (105-110 ng/µl). To make Pmec-4-mScarlet, first, the coding sequence of mScarlet was amplified using primers 5′-GGCTGCTCTTCGATGGTCTC-3′ and 5′-CTGGGTGCTCTTCGCTACTTGTAG-3′, using Addgene plasmid pMS050 as a template, followed by ligation into pCR8/GW/TOPO. pCR8::mScarlet (PNBRGWY52) was recombined with pCZGWY553 [Pmec-4::Gateway destination vector] using LR Clonase to obtain Pmec-4::mScarlet (PNBRGWY54). PNBRGWY54 was injected at a concentration of 5 ng/µl along with 40 ng/µl of pttx3::RFP plasmid as the co-injection marker and 120 ng/µl of pBSK plasmid in the N2 strain to make Pmec-4::mScarlet containing transgenic worms. pCR8::dlk-1 (PNBRGWY14) was recombined with pCZGWY553 [Pmec-4::Gateway destination vector] using LR Clonase to construct Pmec-4::dlk-1 (PNBRGWY13). PNBRGWY13 was injected at a concentration of 0.5 ng/µl along with 40 ng/µl of pttx3::RFP plasmid as the co-injection marker and 120 ng/µl of pBSK plasmid in wild-type and daf-16(lf) strains.

Comparative transcriptomics data of daf-2(lf) and daf-2(lf); daf-16(lf) from Kaletsky et al. (2016) were analyzed using an enrichment analysis tool (www.wormbase.org/tools/enrichment/tea/tea.cgi) (Angeles-Albores et al., 2016). The q value threshold was kept as 0.1 (Kaletsky et al., 2016). Fold change values of relevant GO term genes were analyzed and interpreted. Transcriptomics data from Kaletsky et al. (2016) showed differential expression of ∼9000 neuronal genes. We further applied the GO term (GO: 0097485) relevant to neuronal projection and guidance and found ∼110 genes (Table S4).

The worms were immobilized on a 5% agarose pad with 10 mM levamisole. The jsIs821 [Pmec-7::GFP::RAB-3] (Luo et al., 2014; Mondal et al., 2011) and shrEx209 [Pmec-4::mScarlet] reporters were simultaneously imaged using a LSM510 Meta confocal microscope (Zeiss) before and 24 h after axotomy. The confocal planes were scanned at 0.5 μm intervals under a 63× oil objective (NA=1.46). Approximately 65% of 488 nm laser and 70.2% of 543 nm laser were used for imaging the GFP::RAB-3 and mScarlet reporters, respectively. The mean intensity of GFP::RAB-3 and mScarlet was measured from the ROI of 3 µm drawn at the tip of the regrowing axon (Fig. 1E; Fig. S1B). To quantify the GFP::RAB-3 intensity at the synaptic region of the ventral branch, a similar ROI was placed at the ventral cord region at the end of the branch (Fig. 1F). The mean intensity was quantified from the z-stacked image after background subtraction using the Image J. For comparing the GFP::RAB-3 intensities across various regrowth events, GFP::RAB-3 intensity was normalized by the mean intensity of mScarlet from the same ROI (Fig. 1E,F). Similarly, Pmec-4::mcherry::RAB-3 (tbIs227) reporter (Sood et al., 2018) in muIs32 (Pmec-7::GFP) background was used to image the different regrowth events at L4 and A3 stages at 24 h post-axotomy (Fig. 2E,G). The simultaneous imaging of the PLM neuron expressing muIs32 and tbIs227 (Pmec-4::mCherry::RAB-3) was performed using a 60× oil objective (NA=1.4) of a Nikon A1 plus confocal microscope at 0.5 µm slice intervals. Approximately 0.8% of 488 nm laser and 1% of 543 nm laser were used for imaging the muIs32 and Pmec-4::mCherry::RAB-3 (tbIs227) reporters, respectively. For correlating the recovery indices (Fig. S1C;Fig. S2B) to the GFP/mCherry::RAB-3 localization, the touch response assay was performed before proceeding with the confocal imaging.

The PLM neuron expressing icIs132 [Punc-40::UNC-40::GFP] (Chan et al., 1996) and shrEx209 [Pmec-4-mScarlet] was imaged using a Zeiss LSM510 Meta confocal microscope in the wild-type, daf-2(e1368ts) and daf-16(lf) backgrounds. Images were collected before and at 0 h, 3 h, 5 h and 8 h post-axotomy. Three to five z slices were taken at a 0.5 μm interval using a 63× oil objective (NA=1.46). Three planes were z projected for the quantification of UNC-40::GFP and mScarlet from PLM cell body. With regards to the axon tip, the number of z slices that were imaged became higher depending on the extent of regrowth. Approximately 18% of 488 nm laser and 65.5% of 543 nm laser were used for imaging the UNC-40::GFP and mScarlet reporters, respectively. Detector gains used for imaging with 488 nm and 543 nm lasers were 685 and 732, respectively. We captured the images in 8 bits. Therefore, the saturation limit for the image intensities was 256 arbitrary units for both 488 nm and 561 nm lasers. The images were collected below the saturation limit (Fig. S4D-G). The mean intensity of UNC-40::GFP and mScarlet in the PLM cell body were determined from an ROI surrounding the cell body (Fig. 5D). The mean intensity in the axon was measured from a line ROI of 5 µm length placed at 50 µm distance (Fig. 5E) from the cell body (before axotomy) or the tip of the regenerating axon (Fig. 5E,G). The background intensities were determined from the same ROIs placed inside the worm away from PLM. The background correction was performed before determining the normalized intensities (Fig. 5F,G).

PLM neuron expressing muIs32 [Pmec-7::GFP] and jsIs1075 [Pmec-7-TagRFP::ELKS-1] (Zheng et al., 2014) reporter was imaged using a Zeiss LSM 510 Meta confocal microscope at 24 h post-axotomy (Fig. S1D). Confocal planes were imaged at 0.5 µm z intervals under a 40× air objective (NA=0.75). The excitation laser powers were 66% and 78% for 488 nm and 543 nm, respectively. Detector gains were 650 and 738 for 488 nm and 543 nm lasers, respectively.

PLM neuron expressing shrEx209 [Pmec-4::mScarlet] and AVA neuron expressing akIs141 [Prig3::GLR1::GFP] reporter (Hoerndli et al., 2015) were imaged using a Nikon A1 plus confocal microscope at 24 h post-axotomy (Fig. S1E). Confocal planes were scanned at 1 µm intervals under a 60× oil objective (NA=1.4). Excitation power was at 5% and 0.5% for 488 nm and 561 nm lasers, respectively.

The unc-6 reporter (OtIs638) [unc-6 (fosmid):: NLS::YFP::H2B] (Weinberg et al., 2018) was imaged in wild-type, daf-2(e1368ts), daf-2(e1368ts);daf-16(lf) and Pmyo-3::daf-16f (shrEx314) backgrounds using a Nikon A1 plus confocal microscope (Fig. 6A-E). Images were captured using a 60× oil objective (NA=1.4) at 2-µm z-section intervals. Approximately 5% 488 nm laser power was used to image unc-6(fosmid)::YFP and 2% 561 nm laser power was used for imaging red fluorescence. We captured the images in 12 bits. Therefore, the saturation limit is 4096 arbitrary units for both 488 nm and 561 nm lasers (Fig. 6G-I). Approximately 10 to 12 z planes were merged for collecting the intensity from all eight pairs of muscle nuclei. The four pairs of muscle nuclei from anal region to vulva are termed as M1, M2, M3 and M4. The other four pairs of nuclei expanding from vulva to pharynx are named as M5, M6, M7 and M8 (Fig. 6A). To remove the autofluorescence, we subtracted the z-projected image of the red channel (561 nm) from the z-projected image of the green channel (488 nm). The mean intensity of the unc-6 reporter was determined from a circular ROI of 2 µm radius surrounding the nuclei M1, M2 and M8 (Fig. 6A-E) after background correction. The images were collected below the saturation limit.

The worms at the L4 stage were transferred to OP50-seeded 60 mm NGM plates containing 50 mM of 5′ fluro-2′deoxyuridine (FuDR; Sigma-Aldrich, F0503) (Sutphin and Kaeberlein, 2009). The presence of food was monitored every 2 days. If the food became a little less, the worms were transferred to a newly seeded plate. The number of living worms and dead worms was calculated every day as the L4 s were transferred for lifespan determination. This counting of worms was carried out until and unless there were no living animals. Before performing axotomy of age-related mutants, we verified whether the lifespan of these mutants in our growth condition matched with the previous studies (Berman and Kenyon, 2006).

Statistical analysis for all experiments in this paper was conducted using GraphPad Prism (GraphPad Prism 8.2.1). Three or more arrays were compared using ANOVA (non-parametric) with Tukey's multiple comparisons test. Two conditions were compared using a Mann–Whitney t-test (non-parametric). Fisher's exact test was used to compare the percentage values of multiple sets in the contingency plots. For all the plots, ‘n’ (the number of samples) and ‘N’ (the number of independent replicates) are mentioned in the figure panels and the legends, respectively. Data are mean±s.d.

We thank Yuji Kohara for cDNAs; the National BioResource Project, Japan; and the Caenorhabditis Genetics Center (CGC) for strains. The CGC is supported by the National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440). We also thank Keerthana Ponniah for cloning and strain constructions; Bhavani Shankar Sahu, Swagata Dey and Arnab Mukhopadhyay for their comments on the manuscript; and A Mukhopadhyay, S. P. Koushika, K. Babu and M. Nonet for help with strains.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.198044