Abstract
The mechanisms that govern the ability of specific neurons to regenerate their axons after injury are not well understood. In Caenorhabditis elegans, the initiation of axon regeneration is positively regulated by the JNK–MAPK pathway. In this study, we identify two components functioning upstream of the JNK pathway: the Ste20-related protein kinase MAX-2 and the Rac-type GTPase CED-10. CED-10, when bound by GTP, interacts with MAX-2 and functions as its upstream regulator in axon regeneration. CED-10, in turn, is activated by axon injury via signals initiated from the integrin α-subunit INA-1 and the nonreceptor tyrosine kinase SRC-1 and transmitted via the signaling module CED-2/CrkII–CED-5/Dock180–CED-12/ELMO. This module is also known to regulate the engulfment of apoptotic cells during development. Our findings thus reveal that the molecular machinery used for engulfment of apoptotic cells also promotes axon regeneration through activation of the JNK pathway.
SIGNIFICANCE STATEMENT The molecular mechanisms of axon regeneration after injury remain poorly understood. In Caenorhabditis elegans, the initiation of axon regeneration is positively regulated by the JNK–MAPK pathway. In this study, we show that integrin, Rac-GTPase, and several other molecules, all of which are known to regulate engulfment of apoptotic cells during development, also regulate axon regeneration. This signaling module activates the JNK–MAPK cascade via MAX-2, a PAK-like protein kinase that binds Rac. Our findings thus reveal that the molecular machinery used for engulfment of apoptotic cells also promotes axon regeneration through activation of the JNK pathway.
Introduction
A fundamental and conserved property of neurons is their ability to regenerate their axons after injury, an ability that is modulated by interactions between the intrinsic axon growth machinery and the local extracellular environment. In adult mammals, the PNS regenerates relatively efficiently, whereas CNS axons regenerate poorly (Chen et al., 2007). This distinction between central and peripheral axon regeneration has been attributed to the combined effects of extrinsic signals provided by the inhibitory glial environment and intrinsic growth capabilities (Case and Tessier-Lavigne, 2005). Because intrinsic regeneration signals can influence regenerative success, they represent potential therapeutic targets to enhance regeneration. However, our understanding of the intrinsic signaling pathways that promote regeneration in the adult nervous system remains limited.
The nematode Caenorhabditis elegans has recently emerged as an attractive model with which to dissect the mechanisms of axon regeneration (Yanik et al., 2004; Hammarlund et al., 2009). Its amenability to genetic analysis makes possible the discovery and elucidation of novel signaling pathways involved in regeneration. Indeed, genetic screens have identified many such genes and pathways, including those specifically required for adult axon regrowth (Chen et al., 2011; Nix et al., 2014). Recent genetic studies have shown that, in C. elegans, the JNK–MAPK pathway functions as a key intrinsic regulator of the initiation of regeneration and may be involved in the sensing of axonal damage (Nix et al., 2011; Li et al., 2012, 2015). The MAPK signaling pathways respond to a variety of extracellular stimuli, generating a diverse array of physiological responses within the cell (Ichijo et al., 1997; Ninomiya-Tsuji et al., 1999), and thus are involved in the regulation of cell proliferation, differentiation, regeneration, stress response, and apoptosis. The C. elegans JNK pathway is composed of the components MLK-1 MAPKKK, MEK-1 MAPKK, and KGB-1–JNK. In a typical MAPK cascade, activation of the upstream MAPKKK is a critical control point for signal specificity and amplification (English et al., 1999). In the case of axon regeneration, we showed recently that TPA-1, a C. elegans protein kinase C (PKC), functions upstream of MLK-1 in the KGB-1 pathway regulating regeneration (Pastuhov et al., 2012). TPA-1 phosphorylates MLK-1 on Ser-355 of the kinase domain activation loop, leading to its activation.
The C. elegans KGB-1–JNK pathway also regulates the stress response to heavy metals (Mizuno et al., 2004; Fujiki et al., 2010). In this pathway, the protein kinase MAX-2, related to yeast Ste20, also activates MLK-1 through direct phosphorylation of Ser-355. Furthermore, the Rac-type GTPase MIG-2 functions upstream of MAX-2 in the KGB-1-mediated stress response pathway. In this study, we show that MAX-2 and another Rac-type GTPase, CED-10, regulate axon regeneration upstream of MLK-1. We also demonstrate that the integrin α-subunit INA-1 activates the JNK pathway through the signaling complex CED-2–CED-5–CED-12, thereby activating axon regeneration. The INA-1–CED-10 signaling pathway was shown previously to regulate the engulfment of apoptotic cells during development (Hsu and Wu, 2010). Therefore, the conserved INA-1–CED-10 signaling unit used for engulfment of apoptotic cells is also used to promote axon regeneration.
Materials and Methods
C. elegans strains.
The C. elegans strains used in this study are listed in Table 1. All strains were maintained on nematode growth medium plates and fed with bacteria of the OP50 strain, as described previously (Brenner, 1974).
C. elegans strains used in this study
Plasmids.
Punc-25::ced-10(G12V) and Punc-25::ced-10(T17N) were generated by cloning the ced-10 cDNA isolated from a cDNA library (Kawasaki et al., 1999), mutagenizing it to generate the GTP and GDP-bound form, respectively, and inserting the resulting constructs into the pSC325 vector. Punc-25::ina-1 and Punc-25::ina-1::gfp were made by cloning the ina-1 cDNA isolated from a cDNA library (Kawasaki et al., 1999) and inserting it into the pSC325 vector. For Punc-25::ina-1::gfp, the last coding codon of ina-1 was ligated in frame with the gfp coding sequence obtained from the pPD95.75 vector. Punc-25::max-2 was made by inserting max-2 cDNA into the pSC325 vector. Punc-25::mlk-1(S355E) and Pmyo-2::dsredmonomer were described previously (Arimoto et al., 2011; Pastuhov et al., 2012). The GAL4 DBD-CED-10(G12V) and GAL4 DBD-CED-10(T17N) plasmids were generous gifts from Dr. Kozo Kaibuchi. The GAL4 AD-MAX-2 plasmid has been described previously (Fujiki et al., 2010). To make the GAL4 AD-MAX-2(CRIB) plasmid, the DNA region corresponding to the 41–91 aa of MAX-2 protein was amplified by PCR and subsequently subcloned into the pACTII vector.
Transgenic animals.
Transgenic C. elegans animals were obtained by a standard microinjection method (Mello et al., 1991). Pmyo-2::dsredmonomer (25 ng/μl), Punc-25::ced-10(G12V) (25 ng/μl), Punc-25::ced-10(T17N) (25 ng/μl), Punc-25::max-2 (25 ng/μl), Punc-25::mlk-1(S355E) (50 ng/μl), and Punc-25::ina-1::gfp plasmids were used in kmEx466 [Punc-25::ced-10(G12V) + Pmyo-2::dsredmonomer], kmEx467 [Punc-25::ced-10(T17N) + Pmyo-2::dsredmonomer], kmEx468 [Punc-25::max-2 + Pmyo-2::dsredmonomer], kmEx469 [Punc-25::mlk-1(S355E) + Pmyo-2::dsredmonomer], and kmEx472 [Punc-25::ina-1::gfp + Pmyo-2::dsredmonomer]. A fosmid fragment containing the dlk-1 gene (25 ng/μl) was used in kmEx524 [dlk-1 fosmid + Pmyo-2::dsredm] as described previously (Li et al., 2012). An extrachromosomal array carrying Pmax-2::flag::max-2 has been described previously (Fujiki et al., 2010).
Immunoblotting.
Worm extracts were prepared by boiling harvested worms in SDS sample buffer. Standard techniques were used for immunoblotting (Kawasaki et al., 1999). Anti-phospho-KGB-1 (anti-pKGB-1) and anti-KGB-1 antibodies have been described previously (Mizuno et al., 2004).
Stress sensitivity.
The assay for heavy metal toxicity (100 μm Cu2+) was performed as described previously (Mizuno et al., 2004).
Axotomy.
Young adult or L4 stage hermaphrodite animals were immobilized with 20 mm levamisole solution in M9 buffer on a 2% agarose pad under a coverslip. D-type motor neurons expressing green fluorescent protein (GFP) were imaged with a fluorescence microscope. Selected commissural axons, mainly posterior D-type neurons, were severed using a 440 nm MicroPoint ablation Laser System from Photonic Instruments. The animals were transferred to an agar plate and remounted for fluorescent imaging ∼24 h after surgery. Both Zeiss Axioplan fluorescent and Olympus FV-500 confocal laser microscopes were used for observation. Axons that grew a distance of 5 μm or more were scored as regenerated. The proximal axon segments that showed no change after 24 h were counted as not regenerated. At least 20 animals with one to two axotomized commissures were observed for most experiments. Two-tailed p-values were calculated using Fisher's exact test.
Microscopy.
Standard fluorescent images of transgenic animals were observed under a Zeiss Plan-APOCHROMAT X100 objective of a Zeiss Axioplan II fluorescent microscope and photographed with a Hamamatsu 3CCD camera. Confocal fluorescent images were taken on an Olympus FV-500 confocal laser-scanning microscope with a 100× objective.
Localization of INA-1::GFP in growth cones.
Two or three axons in L4 stage hermaphrodite N2;In[Punc-47::mcherry]; Ex[Punc-25::ina-1::gfp + Pmyo-2::dsred] animals were axotomized as described above and imaged 6 h later with an FV-500 confocal laser microscope. Nonaxotomized L4 stage animals served as a control. For each axotomized axon, the fold enrichment (FE) of INA-1::GFP in proximal ends of injured axons, FE(INA-1)end, was calculated as follows:
GFPend, GFPaxon, and GFPbg were the mean fluorescent intensities of GFP in the proximal end of the injured axon and the proximal fragment of the axotomized axon excluding the proximal end and the background, respectively. mCherryend, mCherryaxon, and mCherrybg were the corresponding mean intensities of mCherry in the proximal end of the injured axon and the axon region below the proximal end and the background, respectively. All fluorescent intensities were measured with ImageJ. To calculate FE(INA-1)end for nonaxotomized axons, a region of the axon near the midline, which corresponded to the site of axotomy, was marked GFPend and mCherryend and the region of the axon below it was used to measure GFPaxon and mCherryaxon. The difference between FE(INA-1)end for axotomized and nonaxotomized axons was tested for statistical significance with Mann–Whitney test. The box plot for FE(INA-1)end was made with R-studio (http://www.rstudio.com/).
Yeast two-hybrid assay.
The Saccharomyces cerevisiae reporter strain PJ69-4A (James et al., 1996) was cotransformed with plasmids expressing the GAL4 DBD-CED-10(G12V or T17N) and GAL4 AD-MAX-2(full-length or CRIB) fusion proteins. The transformants were plated onto synthetic medium lacking histidine and incubated at 30°C for 4 d. Interaction of the pairs of fusion proteins transactivates the HIS3 reporter gene and allows their growth on the plate.
Counting of apoptotic cells.
Differential interference contrast images of at least 20 animals in each of the 1.5-fold and 2-fold stages were obtained and apoptotic cells were counted.
Results
TPA-1 functions redundantly with MAX-2 in the KGB-1–JNK pathway to regulate the stress response to heavy metals
We have shown previously that MAX-2, a C. elegans Ste20-related protein kinase, acts upstream of MLK-1 in the stress response to heavy metals by phosphorylating it on Ser-355 (Fujiki et al., 2010). This phosphorylation is important for the activation of MLK-1 kinase activity in the KGB-1 pathway (Fig. 1A). However, animals with a max-2(nv162)-null mutation remain partially resistant to stress induced by copper ions (Cu2+) (Fig. 1B) (Fujiki et al., 2010), raising the possibility that another protein kinase may play a redundant role with MAX-2 in mediating the response to heavy metals. In addition to the heavy metal stress response, the C. elegans MLK-1–MEK-1–KGB-1 pathway also regulates axon regeneration (Nix et al., 2011). In axon regeneration, TPA-1, a C. elegans PKC, interacts with and phosphorylates MLK-1 at the Ser-355 residue (Pastuhov et al., 2012). We therefore investigated whether TPA-1 is also involved in the KGB-1-mediated stress response pathway. The k501 allele of the tpa-1 gene contains a missense mutation causing a change of the proline at 544 in the kinase domain to a serine. Animals carrying the tpa-1(k501) mutation were placed on agar plates containing a high Cu2+ concentration and their development monitored for any signs of an altered response to Cu2+. We found that tpa-1(k501) mutants were weakly sensitive to Cu2+ compared with wild-type animals (Fig. 1B). To test whether TPA-1 regulates the heavy metal stress response redundantly with MAX-2, we constructed max-2(nv162); tpa-1(k501) double mutants and found that these showed Cu2+ sensitivity stronger than the single max-2(nv162) mutant (Fig. 1B). Sensitivity to Cu2+ was also observed in tpa-1(fr1) mutants (Fig. 1B). The fr1 allele contains a nonsense mutation, which results in a premature stop codon at Trp-187, suggesting that fr1 is probably a null mutation. We found that max-2(nv162); tpa-1(fr1) double mutants showed enhanced Cu2+ sensitivity, which was similar to that observed in kgb-1 mutants (Fig. 1B). Therefore, TPA-1 and MAX-2 regulate the response to Cu2+ stress redundantly.
Heavy metal stress sensitivity in C. elegans mutants. A, Schematic diagram of the KGB-1–JNK pathway. B, Copper sensitivity. Each animal was cultured from embryogenesis on NGM plates containing 100 μm Cu2+ (+) and seeded with bacteria of the OP50 strain. The percentages of worms reaching adulthood 4 d after egg laying are shown with SEs (error bars). WT, Wild-type. C, KGB-1 activity. N2 WT, kgb-1(km21), max-2(nv162), tpa-1(k501), and max-2(nv162);tpa-1(k501) animals were treated with 100 μm Cu2+. Extracts prepared from each animal were immunoblotted with anti-pKGB-1 and anti-KGB-1 antibodies.
To verify the redundancy of TPA-1 with MAX-2 in the KGB-1 pathway, we examined KGB-1 activation. KGB-1 is activated by MEK-1-mediated phosphorylation at the Ser-198 and Tyr-200 residues located in the KGB-1 kinase domain (Mizuno et al., 2004, 2008). The anti-pKGB-1 antibody, which recognizes the phosphorylated form of KGB-1, was used to monitor KGB-1 activation in C. elegans. Western blot analysis with anti-pKGB-1 antibody clearly detected the phosphorylated form of KGB-1 in wild-type animals treated with Cu2+ ions (Fig. 1C, lane 2). In contrast, animals harboring the kgb-1(km21) deletion mutation exhibited undetectable levels of KGB-1 protein expression or activation (Fig. 1C, lane 1). In the single max-2(nv162) or tpa-1(k501) mutants, phosphorylation of KGB-1 in response to Cu2+ treatment was somewhat lower than wild-type (Fig. 1C, lanes 3 and 4). However, in the max-2(nv162); tpa-1(k501) double mutants, KGB-1 activity was markedly reduced compared with either single mutant (Fig. 1C, lane 5). Together, these results suggest that MAX-2 and TPA-1 activate the KGB-1 response to heavy metal stress redundantly by directly phosphorylating MLK-1 Ser-355 and upregulating MLK-1 kinase activity.
MAX-2 and TPA-1 regulate axon regeneration mainly in the L4 and young adult developmental stages, respectively
As observed previously (Pastuhov et al., 2012), axon regeneration in tpa-1(k501) mutants was reduced only in young adult animals, not in L4 larvae (Fig. 2A, Table 2). Therefore, TPA-1 is required mainly in the adult stage for axon regeneration of D-type motor neurons after laser surgery. In contrast, MLK-1 is required for axon regeneration at both stages (Fig. 2A, Table 2), suggesting that another factor may act as an activator of MLK-1 to promote axon regeneration mainly in the L4 stage. Our analysis of MAX-2 raised the possibility that it might be a candidate for such an MLK-1 activator. To test this possibility, we investigated whether MAX-2 is involved in the KGB-1-mediated axon regeneration pathway. Although the axon guidance defect was observed in max-2(nv162) mutants (Lucanic et al., 2006), we found that the D-type motor neurons of max-2(nv162) mutants in the L4 stage were impaired in their regeneration response (Fig. 2A,B, Table 2). This result is consistent with a recent report of extensive screening by Nix et al. (2014). Furthermore, Chen et al. (2011) performed a systematic screen of mutants looking for defects in axon regeneration and identified max-2 among several genes that regulate axon regeneration positively in a glutamatergic touch sensory neuron. These results support the idea that MAX-2 is generally required during regeneration. The effect of the max-2 mutation on axon regeneration in the L4 stage was stronger than that in the young adult stage (Fig. 2A, Table 2). In addition, the max-2;tpa-1 double mutants were defective in axon regeneration in both the L4 and adult stages (Fig. 2A, Table 2). However, the axon guidance defect observed in max-2(nv162) mutants was not enhanced by the tpa-1(k501) mutation. Therefore, MAX-2 and TPA-1 regulate axon regeneration mainly in the L4 and adult stages, respectively. Although the expression levels of the max-2 gene in whole animals are constant during various developmental stages (Spencer et al., 2011), it has been shown that max-2 is expressed in ventral cord neurons during early development, but not at the young adult stage (Lucanic et al., 2006). These results suggest that TPA-1 takes the place of MAX-2 to activate MLK-1 in axon regeneration at the adult stage.
Effects of MAX-2 and TPA-1 on axon regeneration. A, Percentages of axons that initiated regeneration 24 h after laser surgery in L4 or young adult stage. Error bars indicate 95% confidence intervals (CI). *p < 0.05, **p < 0.01, ***p < 0.001 as determined by Fisher's exact test. NS, Not significant. B, Representative D-type motor neurons in wild-type and max-2(nv162) mutant animals 24 h after laser surgery in L4 stage. In wild-type animals, the severed axon had regenerated growth cone (arrowhead). In max-2 mutants, the severed axon failed to regenerate (arrow). Scale bar, 10 μm.
Raw data for axon regeneration
We next investigated whether MLK-1 functions downstream of MAX-2 in axon regeneration. We generated a phosphomimetic form of MLK-1 [MLK-1(S355E)] in which Ser-355 has been mutated to glutamate. We found that the expression of MLK-1(S355E) by the unc-25 promoter in D-type neurons was able to suppress the regeneration defect of max-2;tpa-1 double mutants at either the L4 or adult stage (Fig. 2A, Table 2). These results suggest that MLK-1 functions cell autonomously downstream of MAX-2.
C. elegans axon regeneration is regulated by the p38 and JNK–MAPK pathways, which consist of DLK-1 (MAPKKK)–MKK-4 (MAPKK)–PMK-3 (p38 MAPK), and MLK-1(MAPKKK)–MEK-1(MAPKK)–KGB-1 (JNK–MAPK), respectively (Fig. 3A) (Hammarlund et al., 2009; Yan et al., 2009; Nix et al., 2011). We have shown previously that there is cross talk between the PMK-3 and KGB-1 MAPK pathways (Nix et al., 2011). Overexpression of dlk-1 can activate the KGB-1 pathway by cross-activating MEK-1 (Fig. 3A). Consistent with this, we found that dlk-1 overexpression suppressed the regeneration defect in max-2 mutants (Fig. 3B, Table 2). In addition, we have demonstrated that the DLK-1–p38 MAPK pathway activates axon injury-induced expression of svh-2 via the bZip transcription factor CEBP-1 (Fig. 3A) (Li et al., 2015). The svh-2 gene encodes a Met-like receptor tyrosine kinase, which activates the KGB-1–JNK pathway through tyrosine phosphorylation of MLK-1 (Li et al., 2012). Activation of the MLK-1–MEK-1 pathway requires the dual phosphorylation of MLK-1; that is, SVH-2-mediated tyrosine phosphorylation and MAX-2-mediated serine phosphorylation. In dlk-1 mutants, svh-2 is not induced by axon injury (Li et al., 2015). Therefore, it can be expected that tyrosine phosphorylation of MLK-1 may not occur in dlk-1 mutants overexpressing MAX-2. As expected, we found that overexpression of max-2 by the unc-25 promoter was unable to suppress the regeneration defect in dlk-1 mutants (Fig. 3B, Table 2). Axons in wild-type animals overexpressing MAX-2 regenerated normally (Fig. 3B, Table 2).
Effects of max-2 and dlk-1 overexpression on axon regeneration. A, Relationship between the JNK and p38 MAPK pathways that regulate axon regeneration in C. elegans. Left, Diagram of the two MAPK pathways essential for axon regeneration, with crosstalk between kinases indicated with arrows. Right, DLK-1–p38 MAPK pathway activates axon injury-induced expression of svh-2 via the bZip transcription factor CEBP-1. B, Percentages of axons that initiated regeneration 24 h after laser surgery in L4 stage are shown. Error bars indicate 95% confidence intervals (CI). **p < 0.01 as determined by Fisher's exact test. NS, Not significant.
CED-10 functions upstream of MAX-2 in axon regeneration
MAX-2 is a member of the PAK (p21-activated kinase) family, which is regulated by small GTPases (Fujiki et al., 2010). These GTPases act as molecular switches that interconvert between two states: the active, GTP-bound state and the inactive, GDP-bound state. PAKs have been shown to bind to the GTP-bound form of Cdc42/Rac family GTPases through an N-terminal Cdc42/Rac-interactive binding (CRIB) domain (Dan et al., 2001) and MAX-2 also contains this CRIB motif (Fujiki et al., 2010). In the stress response pathway, the GTP-bound form of the Rac-type GTPase MIG-2 interacts with and activates MAX-2 (Fujiki et al., 2010). We therefore investigated whether MIG-2 would also be involved in the axon regeneration pathway, but found that the mig-2(mu28) mutation had only a weak effect on axon regeneration (Fig. 4A, Table 2). However, we found that another Rac-type GTPase, CED-10, was able to bind to MAX-2. Using a yeast two-hybrid assay, we observed that MAX-2 interacted through its CRIB domain with the activated, GTP-bound form of CED-10, but not with the GDP-bound form (Fig. 4B).
Effect of CED-10 on axon regeneration. A, Percentages of axons that initiated regeneration 24 h after laser surgery in L4 stage. Error bars indicate 95% confidence intervals (CI). *p < 0.05, **p < 0.01, ***p < 0.001 as determined by Fisher's exact test. NS, Not significant. B, Two-hybrid assay for the interaction of MAX-2 with CED-10. The reporter strain PJ69-4A was cotransformed with expression vectors encoding GAL4 DBD-CED-10(T19N) (GDP), GAL4 DBD-CED-10(G14V) (GTP), GAL4 AD–MAX-2, and GAL4 AD-MAX-2(CRIB) as indicated.
We next investigated whether CED-10 functions in axon regeneration. We observed that animals harboring the ced-10(n3246) mutation exhibited a decrease in the axon regeneration of D-type motor neurons (Fig. 4A, Table 2). The n3246 allele replaces a glycine at 60 with arginine. Because Nix et al. (2014) reported that the ced-10 mutation had no effect on axon regeneration, we examined a different allele of the ced-10 gene, n1993. The n1993 allele replaces a valine at 128 with glycine. The ced-10(n1993) mutant was also defective in axon regeneration (Fig. 4A, Table 2). It can be expected that CED-10 cycles between the GTP-bound active and the GDP-bound inactive forms. Consistent with this, we found that expression of the GTP-bound active CED-10(G12V) by the unc-25 promoter in D-type neuron, but not of the GDP-bound inactive CED-10(T17N), was able to rescue the ced-10 defect in axon regeneration (Fig. 4A, Table 2). Furthermore, the ced-10(n1993 or n3246) mutation had no effect on the morphology of D-type motor neurons. These results suggest that CED-10 acts to promote regeneration in the damaged neuron in a cell-autonomous manner.
To address whether ced-10 and max-2 function in the same pathway, we constructed max-2(nv162);ced-10(n3246) double mutants. The double mutants did not show any enhanced defect in axon regeneration compared with the single max-2(nv162) or ced-10(n3246) mutants (Fig. 4A, Table 2), suggesting that CED-10 and MAX-2 function in the same pathway. Furthermore, we found that overexpression of the max-2 gene from the unc-25 promoter in D-type motor neurons was able to suppress the ced-10 defect, whereas expression of GTP-bound active CED-10(G12V) from the unc-25 promoter failed to suppress the max-2 defect (Fig. 4A, Table 2). These results suggest that MAX-2 functions cell autonomously downstream of CED-10 in the axon regeneration pathway.
Previous genetic analyses have shown that CED-10 is important for the engulfment of apoptotic cells during development (Hsu and Wu, 2010). We therefore investigated whether MAX-2 and MLK-1 would be required for the removal of apoptotic cells. A time course analysis of the number of cell corpses present during embryogenesis showed no obvious difference between mlk-1(km19) mutants and wild-type animals at both the 1.5-fold and 2-fold stages (Table 3). In contrast, both the ced-10(n3246) and max-2(nv162) mutations caused an increase in the number of cell corpses (Table 3). However, we found that the max-2 mutation enhanced the defect in engulfment of apoptotic cells in ced-10 mutants (Table 3). Furthermore, overexpression of max-2 by its own promoter failed to suppress the ced-10 defect in engulfment of apoptotic cells (Table 3). These results suggest that MAX-2 and CED-10 function in independent pathways for cell corpse engulfment. Therefore, in the engulfment pathway, the CED-10 GTPase does not transduce signals to the JNK pathway.
Numbers of apoptotic cells at embryonic stages
CED-2, CED-5, and CED-12 act upstream of CED-10 in axon regeneration
In the engulfment pathway, CED-10 GTPase is activated by the intracellular signaling proteins CED-2, CED-5, and CED-12 (Fig. 5A) (Hsu and Wu, 2010). CED-2, CED-5, and CED-12 are similar to the SH2–SH3 domain-containing adaptor protein CrkII, Dock180, and ELMO, respectively (Wu and Horvitz, 1998; Reddien and Horvitz, 2000, 2004; Gumienny et al., 2001). In mammals, Dock180 and ELMO form a complex that functions as a guanine nucleotide exchange factor (GEF) for Rac-GTPases (Brugnera et al., 2002). It has been proposed that CED-5/Dock180 and CED-12/ELMO may interact with CED-2/CrkII to form a trimeric complex and activate the CED-10/Rac (Gumienny et al., 2001; Wu et al., 2001). We therefore next investigated whether CED-2, CED-5, and CED-12 function in axon regeneration. We used two different mutant alleles of each gene, which are all nonsense mutations. The e1752 and n1994 alleles encode CED-2 proteins that terminate at Trp-153 and Arg-102, respectively. The n1812 and n2002 alleles terminate at Glu-28 and Arg-962, respectively, of the CED-5 protein. The k149 and k156 alleles terminate at Arg-38 and Trp-195, respectively, of the CED-12 protein. We observed that all of these mutants showed a lower frequency of axon regeneration after neuronal injury compared with wild-type animals (Fig. 5B, Table 2). Furthermore, none of the ced-2, ced-5, or ced-12 mutations had any apparent effect on the morphology of D-type motor neurons.
Effect of the signaling module CED-2–CED-5–CED-12 on axon regeneration. A, The CED-2–CED-5–CED-12 complex functioning upstream of CED-10. In the molecular pathway regulating the engulfment of dead cells in C. elegans, the CED-2–CED-5–CED-12 complex acts as a GEF for CED-10. B, C, Percentages of axons that initiated regeneration 24 h after laser surgery in L4 stage. Error bars indicate 95% confidence intervals (CI). *p < 0.05, **p < 0.01, ***p < 0.001 as determined by Fisher's exact test. NS, Not significant.
If the CED-2–CED-5–CED-12 complex functions in axon regeneration as a GEF to activate CED-10, then we would expect that the expression of a constitutively active mutant of CED-10 would suppress the ced-5 phenotype. We expressed CED-10(G12V), a mutant CED-10 locked in the GTP-bound state, in D-type motor neurons using the unc-25 promoter and found that it indeed suppressed the ced-5 defect in axon regeneration (Fig. 5C, Table 2). In contrast, similar expression of CED-10(T17N), a GDP-bound inactive form, did not suppress ced-5 (Fig. 5C, Table 2). Furthermore, we found that the max-2(nv162);ced-5(n1812) double mutants were no more defective in axon regeneration than the single ced-5(n1812) mutant. In addition, this ced-5 defect was suppressed by overexpression of the max-2 gene from the unc-25 promoter in D-type neurons (Fig. 5C, Table 2). These results suggest that the CED-2–CED-5–CED-12 complex acts as a GEF for the CED-10 GTPase and functions cell autonomously upstream of MAX-2 in the axon regeneration pathway.
INA-1 and SRC-1 regulate axon regeneration upstream of the JNK pathway
In the apoptotic cell engulfment pathway in C. elegans, the CED-1 receptor or the integrin α-subunit INA-1 functions upstream of CED-10 (Fig. 6A) (Zhou et al., 2001; Hsu and Wu, 2010). To investigate which receptor acts in the CED-10 pathway leading to axon regeneration, we examined ced-1(e1754) and the ina-1(gm39) mutants. The e1754 allele of the ced-1 gene contains a nonsense mutation, which causes termination of the protein at Gln-32. The gm39 allele is the ina-1 weak mutant allele, which encodes a substitution of the amino acid from glycine to glutamic acid at 119 (Baum and Garriga, 1997). Although the ced-1 mutation had no effect on axon regeneration, we found that the frequency of axon regeneration after axon injury was lower in ina-1(gm39) mutants compared with wild-type animals (Fig. 6B, Table 2). The ina-1(gm39) mutation did not affect the morphology of D-type motor neurons. To determine whether INA-1 can act in a cell-autonomous manner, we expressed the ina-1 cDNA from the unc-25 promoter in ina-1(gm39) mutants. The ina-1 defect was rescued by expression of ina-1 in D-type motor neurons (Fig. 6B, Table 2). These results demonstrate that INA-1 functions cell autonomously.
Effects of components used for dying-cell-recognition signaling on axon regeneration. A, INA-1–SRC-1 pathway and the CED-1 receptor functioning upstream of CED-10. In the molecular pathway regulating the engulfment of dead cells in C. elegans, the integrin α-subunit INA-1 and the CED-1 receptor transmit signals to CED-10. B, Percentages of axons that initiated regeneration 24 h after laser surgery in L4 stage. Error bars indicate 95% confidence intervals (CI). *p < 0.05, **p < 0.01, ***p < 0.001 as determined by Fisher's exact test. NS, Not significant.
To determine whether ina-1 acts upstream of the JNK pathway in axon regeneration, we overexpressed MAX-2 in D-type motor neurons from the unc-25 promoter. Overexpression of max-2 significantly rescued the axon regeneration defect seen in ina-1(gm39) mutants (Fig. 6B, Table 2). In addition, we found that expression of a phosphomimetic form of MLK-1 [MLK-1(S355E)] in D-type motor neurons by the unc-25 promoter also suppressed the ina-1 defect (Fig. 6B, Table 2). These observations suggest that INA-1 likely acts upstream of MAX-2 and MLK-1 in axon regeneration.
We next examined INA-1 localization in D-type motor neurons during axon regeneration. For this purpose, we expressed the Punc-25::ina-1::gfp gene in wild-type animals and performed fluorescent imaging analysis in D-type neurons. In the absence of axon injury, we observed that INA-1::GFP distributed uniformly throughout D-type motor axons (Fig. 7A). At 6 h after laser ablation of the axons, significant amounts of INA-1::GFP accumulated at the proximal ends of the injured axons (Fig. 7A,B).
Localization of INA-1::GFP in D-type motor neurons after axon injury. A, Fluorescent images of axons before (0 h) and 6 h after laser surgery in wild-type animals carrying Punc-25::ina-1::gfp and Punc-47::mcherry are shown. The proximal end of the injured axon is boxed (middle) and magnified (right). Scale bars, 2 μm. B, Quantification of the relative fluorescence levels of INA-1::GFP between the proximal ends and internal region. The data are presented as box-blot graph with median (thick line within the box) and interquartile range (edge of box). A white circle indicates an outlier. The numbers (n) of axons examined are shown below. The difference between FE(INA-1)end for nonaxotomized (0 h) and axotomized (6 h) axons was tested for statistical significance with Mann–Whitney test. *p < 0.05.
Previous genetic and biochemical analyses have shown that the nonreceptor tyrosine kinase SRC-1 mediates apoptotic cell engulfment signaling from the INA-1 cytoplasmic domain to CED-2, a component of the CED-2–CED-5–CED-12 module that activates CED-10 (Fig. 6A) (Hsu and Wu, 2010). We therefore investigated whether src-1 participates in axon regeneration. The cj293 allele of the src-1 gene encodes a protein that is deleted between aa 136 and 337, essentially a truncation at 137. Therefore, this mutation lacks the SH2 and kinase domains, suggesting that the cj293 allele is null. The src-1(cj293) mutant had a slight defect in axonal guidance during development. We found that the src-1(cj293) mutation was indeed defective in axon regeneration (Fig. 6B, Table 2). Similar to ina-1 mutants, the src-1 defect in axon regeneration was also suppressed by overexpression of MAX-2 or expression of MLK-1(S355E) in D-type motor neurons from the unc-25 promoter (Fig. 6B, Table 2). Therefore, SRC-1 regulates axon regeneration upstream of MAX-2 and MLK-1. Together, these results suggest that INA-1 and SRC-1 act upstream of the JNK pathway in axon regeneration. The src-1(cj293) mutation caused a greater reduction in regeneration than the ina-1(gm39) mutation. Although this could be due to a difference in allele strength, it also raises the possibility that SRC-1 may mediate signals from sources other than INA-1.
Discussion
The C. elegans JNK pathway is composed of MLK-1 MAPKKK, MEK-1 MAPKK, and KGB-1–JNK and regulates both the stress response to heavy metals and axon regeneration (Mizuno et al., 2004; Nix et al., 2011). MLK-1 is activated by phosphorylation of Ser-355 in the activation loop of its kinase domain. In the stress response to heavy metals, phosphorylation of MLK-1 on Ser-355 is catalyzed by the Ste20-related kinase MAX-2 (Fujiki et al., 2010), whereas in the axon regeneration pathway, MLK-1 Ser-355 phosphorylation is catalyzed by the PKC homolog TPA-1 (Pastuhov et al., 2012). Therefore, two different kinases, MAX-2 and TPA-1, can activate MLK-1 by phosphorylating the same Ser-355 site. Do these kinases function redundantly or independently? We have demonstrated previously that max-2-null mutants, unlike animals harboring mutations in kgb-1, mek-1, or mlk-1, remained partially resistant to heavy metal stress (Fujiki et al., 2010). Furthermore, the max-2 mutation caused a smaller decrease in KGB-1 activity than the mlk-1 mutation. These observations raised the possibility that another kinase may function redundantly with MAX-2 in transducing the heavy metal stress signal to MLK-1. In this study, we show that combining the tpa-1 mutation with the max-2 mutation resulted in greater sensitivity to heavy metal stress and significantly lower KGB-1 activity. Therefore, TPA-1 functions redundantly with MAX-2 in the heavy metal stress response. TPA-1 is necessary for axon regeneration of D-type motor neurons in the young adult stage, but not in the L4 stage (Pastuhov et al., 2012). Given that MLK-1 is required for axon regeneration in both the L4 and adult stages, we hypothesized that MAX-2 may activate MLK-1 mainly in the L4 stage to promote axon regeneration, and indeed we observed this to be the case.
Ste20-like protein kinases are regulated by small GTPases (Dan et al., 2001). These GTPases are active when bound to GTP and inactive after hydrolysis of the bound GTP to GDP. One family of Ste20-like kinases, PAK, has been shown to bind the GTP-bound form of the Cdc42/Rac family of GTPases through a CRIB domain found in PAK kinases (Bagrodia and Cerione, 1999). The PAK family member MAX-2 contains a CRIB motif in its N-terminal domain and is regulated in the stress response pathway by MIG-2, a member of the Rac family of small GTPases (Fujiki et al., 2010). Conversely, in the axon regeneration pathway, MAX-2 is regulated by another Rac-type small GTPase, CED-10. The finding that MAX-2 interacted with the GTP-bound form of CED-10, but not with the GDP-bound form, suggests that CED-10 may be involved in transmitting axon injury signals to MAX-2. This model is further supported by genetic analysis showing that the defect in axon regeneration observed in ced-10 mutants was not enhanced by the max-2 mutation, but was suppressed by overexpression of the max-2 gene. Therefore, CED-10 and MAX-2 appear to act in the same pathway and function to promote axon regeneration by activating the KGB-1–JNK pathway (Fig. 8). Previous studies have revealed that CED-10 is required for the UNC-40/DCC-dependent developmental axon attraction and outgrowth, as well as the ventral guidance of regenerating adult axons, of AVM neurons (Gitai et al., 2003; Gabel et al., 2008). However, they are distinct processes because the unc-40 mutation has no effect on axon regeneration in D-type motor neurons (Nix et al., 2014).
Schematic model for the regulation of axon regeneration by INA-1–CED-10–JNK signaling pathway.
In C. elegans, CED-10 is involved in the engulfment of apoptotic cells (Hsu and Wu, 2010). The intracellular signaling proteins CED-2, CED-5, and CED-12 function as a GEF to activate CED-10 (Gumienny et al., 2001; Wu et al., 2001). The integrin α-subunit, INA-1, recognizes apoptotic cells and mediates signaling through the CED-2–CED-5–CED-12 complex in the engulfing cells to promote cell corpse removal. Furthermore, the nonreceptor tyrosine kinase SRC-1 provides a link between INA-1 and the downstream signaling module CED-2–CED-5–CED-12, leading to CED-10 activation (Hsu and Wu, 2010). In this study, we demonstrate that these intracellular signaling components also function upstream of MAX-2 in the axon regeneration pathway. Although the max-2 mutation also affected the engulfment of apoptotic cells, MAX-2 and CED-10 function in parallel pathways for cell corpse engulfment. Therefore, CED-10/Rac-GTPase activates two separate signaling pathways, one that leads to axon regeneration and involves the JNK pathway and another that leads to the engulfment of apoptotic cells, but does not involve the JNK pathway (Fig. 8).
More recently, Nichols et al. (2016) demonstrated that the apoptotic machinery, including CED-2–CED-5–CED-12 and CED-10, regulates PLM axon degeneration. Therefore, the core molecular machinery used for the engulfment of apoptotic cells regulates both axon regeneration and degeneration. In the process of axon degeneration, the transmembrane receptor CED-1 functions non-cell autonomously in the surrounding hypodermis, which acts as the engulfing tissue for the severed axon. Conversely, CED-1 does not participate in axon regeneration, although INA-1 is involved in this process. INA-1 recognizes apoptotic cells through its extracellular domain and mediates signaling via its intracellular domain to promote apoptotic cell internalization (Hsu and Wu, 2010). Apoptotic cells display “eat me” signals, consisting primarily of phosphatidylserine (PS) on the outer leaflet of the plasma membrane that are recognized by engulfment receptors (Segawa and Nagata, 2015). Mammalian integrins can bind to externalized PS on apoptotic cells indirectly via bridging molecules such as the secreted MFG-E8 (Hanayama et al., 2002; Akakura et al., 2004). However, C. elegans does not appear to have an MFG-E8 homolog and it remains to be determined whether INA-1 recognizes externalized PS on the surface of apoptotic cells. Previous studies have identified the C. elegans secreted, transthyretin-like protein TTR-52 as a bridging molecule between PS on the surface of the apoptotic cell and the CED-1 receptor on the membrane of the engulfing cell (Wang et al., 2010). A recent study indicated that PS is exposed by axon severing and promotes axonal fusion in PLM neuron through a TTR-52–CED-1-dependent signaling pathway (Neumann et al., 2015). Our results suggest that the INA-1-mediated signaling pathway regulating the engulfment of apoptotic cells has been evolutionarily coopted for the regulation of axon regeneration. These findings raise the possibility that the externalized PS generated by axon severing is also recognized by INA-1 indirectly via a bridging molecule to activate the CED-2–CED-5–CED-12 pathway. In this model, axonal injury induces the presentation of PS on the surface of the vesicles generated from the distal fragment of the severed axon. INA-1 may then recognize PS via a bridging molecule and activate the CED-2–CED-5–CED-12 pathway. To verify this possibility, it would be necessary to identify the nature of the signal that promotes axon regeneration and identify which cells generate the signal.
Footnotes
This work was supported by the Ministry of Education, Culture and Science of Japan (K.M.); MEXT (Grant-in-Aid for Scientific Research on Innovative Areas “Homeostatic regulation by various types of cell death,” 15H01375 to N.H.); the Mitsubishi Foundation (K.M.); the Naito Foundation (N.H.); the Daiko Foundation (N.H.); and the Astellas Foundation (N.H.). S.P. is an overseas researcher under a postdoctoral fellowship from the Japan Society for the Promotion of Science. We thank M. Bastiani and P. Nix for providing information and for helpful discussion and K. Kaibuchi of the Caenorhabditis Genetic Center and C. elegans Knockout Consortium for materials.
The authors declare no competing financial interests.
- Correspondence should be addressed to either Kunihiro Matsumoto or Naoki Hisamoto, Division of Biological Science, Graduate School of Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan. g44177a{at}nucc.cc.nagoya-u.ac.jp or i45556a{at}cc.nagoya-u.ac.jp
References
In this issue
