|Home | About | Journals | Submit | Contact Us | Français|
Growth cone guidance and synaptic plasticity involve dynamic local changes in proteins at axons and dendrites. The Dual Leucine zipper MAPKKK (DLK) has been previously implicated in synaptogenesis and axon outgrowth in C. elegans and other animals. Here we show that in C. elegans DLK-1 regulates not only proper synapse formation and axon morphology, but also axon regeneration, by influencing mRNA stability. DLK-1 kinase signals via a MAPKAP kinase, MAK-2, to stabilize the mRNA encoding CEBP-1, a bZip protein related to CCAAT/enhancer binding proteins, via its 3′ UTR. Inappropriate upregulation of cebp-1 in adult neurons disrupts synapses and axon morphology. CEBP-1 and the DLK-1 pathway are essential for axon regeneration after laser axotomy in adult neurons, and that axotomy induces translation of CEBP-1 in axons. Our findings identify the DLK-1 pathway as a regulator of mRNA stability in synapse formation and maintenance and also in adult axon regeneration.
Neurons respond to environmental stimuli and insults in a compartmentalized manner. Local protein synthesis in dendrites and axonal growth cones has emerged as a major mechanism allowing compartmentalized responses in growth cone guidance (Piper and Holt, 2004) and neuronal plasticity (Sutton and Schuman, 2005). Several mRNAs are known to be transported and localized to growth cones or axons; transport of such mRNAs often is mediated by their 3′ untranslated regions (3′UTRs) (Lin and Holt, 2008).
However, it remains controversial whether mature neurons employ axonal regulation of mRNA and local protein synthesis. mRNA binding proteins, such as Zip-code binding proteins (ZBP) and cytoplasmic polyadenylation element binding protein (CPEB) are abundant in growing neurites, but are mostly undetectable in axons and synapses of mature neurons (Brittis et al., 2002; Leung et al., 2006). Polyribosomes are rarely seen in axons (Giuditta et al., 2002). Nonetheless, several reports suggest axonal mRNA regulation must occur in mature neurons. In Aplysia neurons, mRNA for the peptide neurotransmitter Sensorin is concentrated at synapses upon contact with target motor neurons (Lyles et al., 2006), and local secretion of Sensorin stimulates synapse maturation (Hu et al., 2004). A number of mRNAs are up-regulated in axons of injured adult dorsal root ganglion (DRG) neurons (Wang et al., 2007). Regulation of Ran GTPase via local translation of RanBP1 is implicated in retrograde signaling of axon injury (Yudin et al., 2008). Local translation thus can transmit injury signals and initiate local repair processes (Wang et al., 2007).
CCAAT/enhancer binding proteins (C/EBP) are widely expressed bZip domain transcription factors with long-studied roles in cell proliferation, differentiation and stress (Ramji and Foka, 2002). In neurons the transcriptional roles of C/EBP proteins have been linked to learning and memory (Alberini et al., 1994; Chen et al., 2003). Learning and memory tasks trigger activation of Erk or p38 kinases, leading to phosphorylation of specific C/EBP isoforms. mRNAs of murine and leech C/EBP are also up-regulated following axonal injury, and murine C/EBPβ can activate the transcription of an α-tubulin gene associated with injury responses (Korneev et al., 1997; Nadeau et al., 2005). The pathways that induce C/EBP after injury are largely unknown.
We have been studying the function of the conserved ubiquitin E3 ligase RPM-1 in synaptogenesis and axon formation in C. elegans. C. elegans neurons have simple unipolar or bipolar axon trajectories and form synapses en passant. For example, the ALM and PLM mechanosensory neurons have a long axon that bifurcates into a branch exclusively forming synapses and another branch transducing mechanoreception (Chalfie, 1995). RPM-1 regulates the organization and stabilization of presynaptic terminals, and axon termination both in mechanosensory and motor neurons (Schaefer et al., 2000; Zhen et al., 2000). A major target of RPM-1 ubiquitination is the dual-leucine zipper kinase DLK-1 MAPKKK, which acts in a MAPK cascade consisting of the MAPKK MKK-4 and the p38 kinase PMK-3 (Nakata et al., 2005). By controlling the level of DLK-1, RPM-1 keeps the activity of the DLK-1 cascade at optimal levels. This negative regulation of the DLK pathway by ubiquitin-mediated protein degradation is conserved in Drosophila and mammalian neurons (Collins et al., 2006; Jin and Garner, 2008; Lewcock et al., 2007).
The targets of the DLK-1/p38 cascade have not previously been identified. Here we report the identification of MAK-2, a member of the MAPKAP kinase family, and CEBP-1, a member of the C/EBP class of bZip factors as effectors of the C. elegans DLK-1 cascade. MAPKAPKs are conserved Ser/Thr kinases that are direct targets of p38 and Erk kinases (Gaestel, 2006). We find that cebp-1 mRNA is destabilized by RPM-1 via the DLK-1/MAK-2 cascade, acting on the cebp-1 3′ UTR. The DLK-1/MAK-2/CEBP-1 pathway is essential for regenerative regrowth of mature axons following laser axotomy, in part by regulating axonal cebp-1 mRNA stability and translation.
Previous screens for suppressors of rpm-1(lf) neuronal defects uncovered many alleles of the MAPK genes dlk-1, mkk-4, and pmk-3 (Nakata et al., 2005). By analyzing a large number of additional rpm-1 suppressors, we identified two new loci (see Supplemental Data). The suppressor mutation ju637 was mapped to a region containing mak-2, which encodes a MAP kinase-activated protein kinase (MAPKAPK) related to murine MAPKAPK2 (MK2) (Figure 1A, S1A). ju637 results in a replacement of Phe at an invariant Leu (L219) in the kinase domain (Figure 1A). Two deletion alleles of mak-2 suppressed rpm-1(lf) to a similar degree as did ju637 (Figure 1C, S1A), suggesting all three alleles eliminate mak-2 function. A mak-2 reporter was expressed in the nervous system (Figure S1B). By expressing mak-2 under the control of neuron-type specific promoters we found that mak-2 functions autonomously in neurons (Figure 1B, C). These results show that, like dlk-1, mkk-4, and pmk-3, loss of mak-2 function suppresses the synaptic and axon defects of rpm-1 mutants. mak-2 mutants did not exhibit overt neuronal or behavioral phenotypes (Figure 1B, C).
MAPKAPKs are activated by p38-dependent phosphorylation of Thr and Ser residues (Gaestel, 2006), two of which are conserved among all MAPKAPKs and correspond to Thr250 and Thr362 of MAK-2 (Figure 1A). In a yeast two-hybrid assay, we found that MAK-2 interacted with PMK-3 (Figure S2A). This interaction was also observed in co-immunoprecipitation studies in HEK293T cells (Figure S2B), consistent with the idea that MAK-2 is a target of PMK-3. We next tested whether MAK-2 function depends on phosphorylation. Transgenes expressing a predicted non-phosphorylatable form of MAK-2, in which both Thr residues were mutated to Ala, failed to rescue suppression of rpm-1(lf) by mak-2(lf); mutation of either Thr residue partially abrogated rescuing activity (Figure 1E). In contrast, a phosphomimetic form of MAK-2, in which both Thr were mutated to Glu, caused gain-of-function defects resembling those of rpm-1(lf) mutants (Figure 1D). The AA and EE mutations did not appear to alter MAK-2 expression levels (Figure S1E). Further, a kinase-dead MAK-2(K103R) lacked rescuing activity (Figure 1E). These data suggest MAK-2 activation via phosphorylation of the two Thr residues is critical for its function in synapse development and axon termination.
We mapped the rpm-1 suppressor mutations ju659 and ju640 to the left arm of chromosome X (Figure S3A). Transgenes containing cosmid W05H7 rescued suppression of rpm-1(lf) by ju659 (not shown). We localized rescuing activity to a 7.5 kb DNA region containing the gene D1005.3, which encodes a protein of 358 aa containing a basic-leucine-zipper (bZip) domain most closely related to those of the C/EBP family (Figure 2A, S3B, C) (Landschulz et al., 1988). We refer to this gene as cebp-1. ju659 changes Arg290 to Cys and ju640 changes Ser289 to Leu in the basic region of the bZip domain (Figure 2A, S3B). Both residues are highly conserved among bZip proteins; previous studies have shown that the Arg is critical for DNA binding (Kim et al., 1993). The cebp-1(tm2807) deletion allele suppressed rpm-1(lf) to a similar degree as did ju659 (Figure 2C, S3A). Neuronal expression of cebp-1 rescued the suppression of rpm-1(lf) in a cell-autonomous manner (Figure 2C). CEBP-1 lacking the bZip domain lacked rescuing activity (Figure 2D). bZip proteins usually function as dimers, and mutations impairing DNA binding often have dominant-negative effects (Ramji and Foka, 2002). Transgenic overexpression of cebp-1 containing either the ju659 (R290C) or ju640 (S289L) mutation partly suppressed rpm-1(lf) defects (Figure 2E). Although ju659 and ju640 are recessive to wild-type for suppression of rpm-1, these findings suggest both mutations may have dominant negative activities by either inhibiting endogenous CEBP-1 or its binding partners. These data indicate CEBP-1 acts similarly to MAK-2 and the MAP kinases in neurons.
To establish the order of mak-2 and cebp-1 function in the dlk-1 pathway, we performed genetic epistasis analysis. Double mutants between null mutations in mak-2, or cebp-1, or pmk-3 developed superficially normal synapses and axons (not shown). pmk-3;mak-2 or mak-2;cebp-1 double mutants suppressed the neuronal defects of rpm-1(lf) to a level comparable to that of each single mutant (Figure S4A). Overexpression of dlk-1 causes abnormal synapse development and axon termination defects resembling those of rpm-1(lf) (Nakata et al., 2005). This gain of function effect was largely eliminated in mak-2(lf) or cebp-1(lf) mutants (Figure S4B). Additionally, the gain-of-function effect of phospho-mimetic mak-2(EE) in wild type animals was suppressed by cebp-1(lf) but not by pmk-3(lf) (Figure S4B). Thus, both mak-2 and cebp-1 act downstream of the dlk-1/mkk-4/pmk-3 cascade, and cebp-1 acts downstream of mak-2 (Figure S4C).
In the course of transformation rescue of cebp-1, we observed that cebp-1 transgenes lacking the 3′ untranslated region (UTR) often caused gain-of-function defects such as touch axon overextension and uncoordinated movement, reminiscent of those resulting from constitutive activation of the DLK-1 pathway (Nakata et al., 2005). Analysis of cebp-1 cDNAs revealed that cebp-1 transcripts contain a long 3′ UTR (Figure S5A). To test whether the DLK-1 pathway might regulate cebp-1 mRNA levels, we first performed quantitative RT-PCR and found that cebp-1 mRNA was increased in rpm-1(lf) compared to wild type animals or MAP kinase mutants (Figure 3A). The increase of cebp-1 mRNA levels in rpm-1(lf) mutants was eliminated by loss of function in dlk-1, pmk-3 or mak-2 (Figure 3A), indicating that activation of the MAP kinases caused by rpm-1(lf) results in up-regulation of cebp-1 mRNA. cebp-1(ju659) did not affect the increase of cebp-1 mRNA level caused by rpm-1(lf) (Figure 3A), suggesting that the effect on cebp-1 mRNA does not involve transcriptional autoregulation of cebp-1. Expression of a cebp-1 promoter reporter was unaltered in rpm-1(lf) compared to the wild type (Figure S6A, B). We further compared the stability of cebp-1 mRNA by qRT-PCR on animals cultured in the presence of α-amanitin to block transcription (Sanford et al., 1983). The half-life of cebp-1 mRNA was increased three-fold in rpm-1 mutants compared to wild type (Figure S6C). This increase in stability was abolished in dlk-1;rpm-1 animals, but not in rpm-1;cebp-1 animals. We infer that RPM-1 acts via the DLK-1 pathway to regulate cebp-1 mRNA stability.
To test whether the 3′ UTR of cebp-1 is required for cebp-1 mRNA regulation, we expressed mCherry-tagged full-length cebp-1 with its own 3′ UTR or with a heterologous 3′ UTR, driven by the unc-25 promoter (Supplemental Data). In rpm-1(lf) mutants the fluorescence intensity of mCherry::CEBP-1 with the cebp-1 3′ UTR was over 6-fold higher than in wild type (Figure 3B). This increase in fluorescence intensity reflected a corresponding increase in transgene mRNA levels in rpm-1 (Figure S6D, E). mCherry::CEBP-1 fluorescence intensity was also significantly increased by transgenic expression of MAK-2(EE) but not MAK-2(AA) (Figure 3B). The 3′ UTR of the myosin gene unc-54 is known to stabilize mRNAs (Fire et al., 1990). However mCherry::CEBP-1 transgenes containing the unc-54 3′ UTR did not show increased fluorescence intensity in rpm-1(lf) (Figure 3C). Indeed, the cebp-1 3′ UTR was sufficient to confer rpm-1-dependent regulation on mCherry (Figure 3D). Expression of cebp-1 with the unc-54 3′ UTR consistently caused more severe defects in motor neuron synapses (Figure 3E) and touch neuron morphology (Figure 3F) than did transgenes with the cebp-1 3′ UTR. Overexpression of cebp-1 in muscles did not generate neuronal abnormalities (Figure 3F), suggesting regulation of cebp-1 mRNA by rpm-1 and the MAP kinases occurs in neurons. Thus, activation of the DLK-1 cascade, either by loss of function in rpm-1 or by constitutive activation of MAK-2, can stabilize the cebp-1 mRNA via its 3′ UTR.
We next examined localization of MAK-2 and CEBP-1 proteins in touch and motor neurons. Expression of mCherry tagged MAK-2 or CEBP-1 in motor neurons fully rescued rpm-1 suppression phenotypes of mak-2(lf) or cebp-1(lf), respectively (Figure 1B, ,2B),2B), and did not cause overt defects in wild type animals (Figure 4A, B), suggesting the expression levels were close to physiological. mCherry::MAK-2 was consistently present in cell bodies and synapses (Figure 4A), but rarely seen in axon commissures (not shown). MAK-2 was broadly localized at motor neuron synapses, and significantly colocalized with synaptobrevin/SNB-1 but not with RPM-1 (Figure 4A; Figure S7A–C). Unexpectedly, mCherry::CEBP-1 was also detected at synapses, in addition to the cytoplasm and nucleus of the soma (Figure 4B). At synapses, mCherry::CEBP-1 colocalized with SNB-1, but not with RPM-1 (Figure S7D, E). MAK-2 and CEBP-1 partly colocalized at synapses (Figure S7F). In touch neurons, CEBP-1 was present in the soma, in the synaptic area, and at discrete regions along axons (Figure 5A).
As RPM-1 and DLK-1 are present at synapses (Abrams et al., 2008; Nakata et al., 2005), the synaptic localization of MAK-2 and CEBP-1 suggested that the DLK-1/MAK-2 cascade might act locally to promote cebp-1 mRNA stability. We first tested this by manipulating MAK-2 localization. As predicted, removing the nuclear localization signal (NLS) caused MAK-2 to accumulate in the somatic cytoplasm, whereas removing the nuclear export signal (NES) confined MAK-2 to the nucleus (Figure S1D). Neither manipulation affected the synaptic localization of MAK-2 (Figure S1D). Both constructs could fully rescue the rpm-1 suppression phenotype of mak-2(lf) (Figure 1E). Phospho-mimetic MAK-2(EE), which causes abnormalities in synapses, was also localized at synapses but excluded from the nucleus (Figure S1E). These data suggest nucleo-cytoplasmic shuttling may not affect MAK-2 synaptic function and that activated MAK-2 might function at synapses. We therefore attempted to target activated MAK-2(EE) to motor neuron synapses by tagging MAK-2 with the PHR domain from RPM-1, which can target GFP to synapses (Abrams et al., 2008). mCherry::PHR::MAK-2 was concentrated at synapses, adjacent to SNB-1::GFP, and weakly detected in the somatic cytoplasm and excluded from the nucleus (Figure 4C). Synaptic targeting of activated MAK-2(EE), but not of wild type MAK-2 or inactive MAK-2(AA), disrupted presynaptic morphology to a similar degree as did untargeted MAK-2(EE) (Figure 4D, E). These results suggest that, in addition to possible roles in the soma and nucleus, MAK-2 function may be regulated at synapses.
The above observations raised the possibility that cebp-1 mRNA might be localized at synapses. To test this, we used an in vivo mRNA imaging method based on the high-affinity binding of the coat protein of the MS2 bacteriophage to a 19 nucleotide RNA stem-loop structure (Bernardi and Spahr, 1972; Bertrand et al., 1998). GFP::NLS::MS2 fluorescence was observed in cell bodies and only faintly along the axons of touch neurons and motor neurons (Figure 5B, S8A, B). As background fluorescence of GFP::NLS::MS2 in touch neuron axons was much lower than in motor neurons, we focused on cebp-1 mRNA in the touch neurons. Upon coexpression of cebp-1 mRNA containing six copies of the MS2 binding site in the 3′ UTR, we detected punctate GFP::NLS::MS2 fluorescence along axons and at synapses (Figure 5B, S8A, B). A control reporter containing MS2 binding sites in the unc-54 3′ UTR did not cause an increase of GFP::NLS::MS2 in axons (Figure 5B). The intensity of GFP::NLS::MS2 in axons was further increased in rpm-1(lf) mutants, and reduced in dlk-1(lf) mutants (Figure 5B, C). The half-life of the transgenic MS2 binding site-tagged cebp-1 transcripts was similar to that of endogenous cebp-1 mRNA, and was likewise increased in rpm-1 mutants (Figure S8C). We also saw GFP::NLS::MS2 fluorescence increase in the synaptic regions of the motor neurons upon coexpression with MS2 binding site-tagged cebp-1 mRNA (Figure S8A). These findings suggest that cebp-1 mRNAs may be present in axons and presynaptic regions, and that activation of the DLK-1 cascade could locally stabilize cebp-1 mRNA.
Our observations that MAK-2 and CEBP-1 are present at synapses of adult neurons raised the possibility that this pathway may also act in adults. To test this, we examined the effects of acute adult overexpression of cebp-1 on synaptic morphology, exploiting the temperature sensitive splicing of a mec-2 intron (Poon et al., 2008; M. Chalfie, personal communication). Transgenic animals expressing cebp-1 lacking its 3′ UTR were cultured at 15°C until late L4 stage, and then shifted to 25°C for the next 24 hours to overexpress cebp-1 in adults. In control experiments the mec-2 intron became permissive for expression within 1 hour of temperature shift (Figure S9A). These transgenic animals displayed fewer synapses, of abnormal morphology, compared to controls (Figure 4F). Likewise, PLM neurons showed axon overextension and retraction of the synaptic branch upon induction of cebp-1 overexpression in adults (Figure S9B). We infer that regulation of the DLK-1/CEBP-1 pathway is required continuously in mature neurons to maintain axon and synapse morphology.
Animals lacking components of the DLK-1 pathway develop morphologically normal synapses and axon trajectories, and show normal locomotion, egg-laying, and touch responses. To test whether this pathway has a later role in mature nervous system, we assayed regeneration after laser axotomy in adult mechanosensory axons (Wu et al., 2007). In wild type adults, axotomized PLM axons exhibit robust regrowth of ~100 μm in 24 h (Figure 6A). In contrast, PLM regrowth was essentially blocked in animals lacking each MAP kinase or CEBP-1 (Figure 6A). In these mutants, axons that failed to regenerate did not display growth cones, unlike wild type axotomized axons (Figure S10A). ALM neurons also failed to regrow in these mutants (data not shown). The MAPK/CEBP-1 pathway is required in adult axons for regrowth, because animals in which we induced expression of a dominant negative cebp-1(R290C) by heat-shock in late L4 stage showed reduced regeneration (Figure 6B). Elevating the activity of the DLK-1 pathway by overexpression of dlk-1(+) enhanced regeneration in a cebp-1 dependent manner (Figure 6A). Moreover, touch neuron specific expression of mak-2(+), but not of the kinase dead mak-2(K103R), rescued the mak-2(lf) regrowth defect (Figure 6A), indicating that this pathway acts cell autonomously. We conclude that the DLK-1 pathway is required for regrowth of mature touch neuron axons, and that activity of DLK-1 and CEBP-1 may be rate limiting for regrowth after injury.
We next addressed whether the DLK-1 pathway acts via similar mechanisms in regenerative responses in mature neurons as in developing neurons. We tagged CEBP-1 with the photoconvertible fluorescent protein Dendra (Gurskaya et al., 2006) to distinguish newly synthesized CEBP-1 from pre-existing protein. In uninjured neurons, we observed a low level of synthesis of new Dendra::CEBP-1 within 4 hours of photoconversion (Figure 7A). Following axotomy, within 2 hours, the appearance of new Dendra::CEBP-1 was accelerated both in the severed distal fragments and in the soma and proximal axon (Figure 7B, D). Animals laser-wounded in adjacent epidermis did not show elevated synthesis of Dendra::CEBP-1 (Figure 7B). The appearance of newly synthesized CEBP-1 in distal axon fragments suggests that the translation machinery is present in axons. Consistent with this interpretation, cycloheximide blocked new synthesis of CEBP-1 in axon fragments (Figure 7B, D). dlk-1 mutants did not display new Dendra within the same time course (Figure 7B, D). mak-2 mutants showed partly reduced synthesis of new Dendra (Figure 7B, D), consistent with the partial block of regeneration in mak-2 mutants (Figure 6A). These data suggest axotomy induces activation of DLK-1/MAK-2, which in turn promotes local translation of cebp-1 mRNA. Dendra::CEBP-1 transgenes in which the 3′ UTR of cebp-1 was deleted were not upregulated after axotomy (Figure 7C, D). To test if the 3′ UTR of cebp-1 was responsible for the induced local synthesis, we expressed Dendra with the cebp-1 3′UTR or the unc-54 3′ UTR. Axotomy resulted in elevated synthesis of new Dendra from transgenes containing the cebp-1 3′ UTR, but not from transgenes with the unc-54 3′ UTR (Figure 7E, F). Lastly, to address whether new synthesis of CEBP-1 protein is required for axon regeneration, we induced expression of cebp-1(+) with its own 3′ UTR in cebp-1(lf) mutants following axotomy, and observed partial rescue of axon regeneration (Figure S10C). Induced expression of cebp-1 with the unc-54 3′UTR did not rescue the cebp-1 phenotype (Figure S10C). These observations indicate that axotomy induces local activation of the DLK-1 kinase cascade, which results in local synthesis of CEBP-1 protein.
The DLK kinases have been known to be present in axons and synaptic terminals for over a decade (Holzman et al., 1994; Mata et al., 1996), but only recently have their functions in neurons been revealed. Through genetic studies in C. elegans and Drosophila, these kinases have been identified as substrates of the conserved PHR family E3 ligases (Collins et al., 2006; Nakata et al., 2005). Regulation of the DLK kinases by ubiquitin-mediated degradation also plays roles in axon pathfinding and synapses in vertebrates (Bloom et al., 2007; D’Souza et al., 2005; Lewcock et al., 2007). However the output of this pathway is not well understood. We have shown that in C. elegans neurons the DLK-1 pathway acts via the MAPKAP kinase, MAK-2, to promote stabilization of the mRNA of the bZip protein CEBP-1, resulting in enhanced CEBP-1 translation. This mechanism is also required for the regenerative responses of adult neurons.
A key outcome of the DLK-1 pathway is to stabilize cebp-1 mRNA, leading to its increased translation at synapses and in axons. Local protein synthesis is now well established as a mechanism in synaptic plasticity in mature neurons, following the identification of polyribosomes in dendrites (Steward and Levy, 1982; Sutton and Schuman, 2005). Local translation in growth cones of developing neurons has also been linked to growth cone turning, although possibly not in all situations (Lin and Holt, 2008; Roche et al., 2009). Whether protein synthesis occurs in mature axons has been unclear, largely because of a lack of evidence for polyribosomes in axons (Piper and Holt, 2004). Nonetheless, local synthesis of some proteins is reported in axons of select neurons, and expression of many mRNAs increases in injured adult axons (Di Giovanni et al., 2005; Willis et al., 2007; Xiao et al., 2002; Wang et al., 2007). Our studies indicate that local regulation of mRNA and protein translation in axons is functionally important in mature neurons both under normal and injured conditions.
How might the DLK-1 pathway regulate cebp-1 mRNA stability? Studies in non-neuronal cells have shown that murine MK2 can regulate stability of mRNAs such as TNF-α and interleukins; known targets of MAPKAPKs include mRNA binding proteins such as tristetraprolin and HuR (Gaestel, 2006). Future studies may reveal which components of the RNA turnover machinery are relevant in cebp-1 mRNA stabilization.
Although the DLK/CEBP-1 pathway has a restricted role in developmental axon outgrowth, we show here that this entire pathway is essential for regenerative regrowth in adult touch neurons, consistent with recent findings in motor neurons (Hammarlund et al., 2009). Our results further suggest a model in which unknown injury signals activate DLK-1, which then triggers regrowth by promoting cebp-1 mRNA regulation and translation. As C. elegans axonal transport occurs at rates comparable to those of much larger neurons in other animals (Zhou et al., 2001), it is unexpected that C. elegans neurons use local synthesis for responses to acute localized stimuli such as axotomy. By analogy to our findings at synapses, DLK-1-dependent upregulation of CEBP-1 by axotomy may involve local stabilization of cebp-1 mRNA, although for technical reasons we have not been able to visualize this directly. Activation of the DLK-1 pathway by axotomy could have additional effects, such as direct stimulation of translation.
In injured Aplysia neurons ApC/EBP is activated by phosphorylation via a MEK-ERK cascade, which involves retrograde transport of the ‘injury signal’ to the nucleus (Sung et al., 2001). bZip proteins of the Jun and ATF subfamilies are also implicated in mammalian axon regeneration (Herdegen et al., 1997; Raivich et al., 2004). C/EBPβ mRNA and protein phosphorylation are elevated after injury, and C/EBPβ is required for injuryinduced upregulation of α-tubulin and GAP-43 (Nadeau et al., 2005). However, as yet there has been no demonstration that any of these bZip proteins are required for regenerative growth, apart from Jun which has a partial requirement (Raivich et al., 2004). The finding that DLK-1, its effector kinases, and CEBP-1 are essential for regeneration suggests their mammalian homologs may be required for axonal injury responses, and that local mRNA stability and translation could be a conserved axonal response to injury.
It is unexpected that CEBP-1 is present at synapses and in axons. A related bZip protein does not appear to be synaptically localized, suggesting this localization of CEBP-1 is specific (D.Y. & Y.J., unpublished data). CEBP-1 joins an increasing list of nuclear proteins associated with synapses and neurites. For example, Ribeye, an isoform of the transcriptional repressor C-terminal Binding Protein (CtBP2), is found at the ribbon synapse active zone (Schmitz et al., 2000). The homeodomain proteins Emx2 and Engrailed2 localize to presynaptic termini of olfactory neurons and Xenopus neuron growth cones, respectively, and interact with the translation factor eIF4E (Nedelec et al., 2004; Brunet et al., 2005). The bZip protein CREB can be locally translated in axons and transported retrogradely (Cox et al., 2008). At this stage, our data do not distinguish whether CEBP-1 has two functions, one locally at the axon and synapses, and the other at the nucleus. The observations that abnormally high levels of CEBP-1 disrupt synapse development and maintenance imply that CEBP-1 is tightly regulated. That this regulation is in part achieved via local regulation of mRNA stability suggests CEBP-1 could form a link between synapse dynamics and gene expression. Such dual modes of regulation have been long studied in CREB-mediated neural plasticity (Harris, 2008; Lonze and Ginty, 2002).
Improved sensitivity in mRNA detection has led to the identification of an increasing number of axonally localized mRNAs (Lin and Holt, 2008; Wang et al., 2007). However, a mechanistic understanding of how such mRNAs are targeted to axons and activated for translation is only beginning (Kiebler and Bassell, 2006; Krichevsky and Kosik, 2001). Transport and subcellular localization of mRNAs depends on both sequence tags in the mRNA itself and its interactions with carrier proteins. We find that the cebp-1 3′ UTR is essential both for its axonal delivery and its post-transcriptional regulation in axons, and contains multiple adenine/uridine-rich motifs associated with unstable mRNAs (Wilusz et al., 2001). Several mammalian C/EBP mRNAs also contain similar nucleotide motifs in their 3′ UTR (Figure S5B). Future work will address the multiple levels at which CEBP-1 is regulated in synaptogenesis and regeneration.
We maintained C. elegans strains on NGM plates at 20–22.5°C as described in Brenner (1974). The characterization of the suppressors is in Supplemental Data. The mutations and transgenes are: dlk-1(ju476) I, mkk-4(ju91) X, pmk-3(ok169, ju485) IV, mak-2(tm2927, ju637, ok2394) IV, cebp-1(tm2807, ju659, ju640) X; juIs1[Punc-25-SNB-1::GFP] for GABA motor neuron synapses (Hallam and Jin, 1998); muIs32[Pmec-7-GFP] (Ch’ng et al., 2003) and zdIs5[Pmec-4-GFP] (Clark and Chiu, 2003) for touch neuron morphology. Other transgenes are described in Supplemental Data and listed in Supplemental Table 2.
We scored fluorescent reporters in living animals using a Zeiss Axioplan 2 microscope equipped with Chroma HQ filters. For quantification of touch neuron morphology, 30–90 animals were analyzed. Images of juIs1 synapse morphology, cebp-1 mRNA localization, and MS2::GFP reporters were collected on one-day old adults immobilized in 1% 1-phenoxy-2-propanol (TCI America, Portland, OR) in M9 buffer using a Zeiss LSM510 confocal microscope (Supplemental Data). We cut PLM axons using a near-infrared Ti-Sapphire laser (KMLabs, Boulder, CO) as described (Wu et al., 2007).
In heat shock experiments, wild type or Phsp::cebp-1(R290C) transgenic worms were grown at 20°C until late L4, heat shocked at 35°C for 45 min, recovered at 20°C for one hour before axotomy. Regrowth of axons was imaged 24h post-axotomy at 20°C. In temperature shift induction experiments, wild type or Prgef-1-intron-cebp-1 transgenic animals were cultivated at 15°C until late L4, then half were shifted to 25°C and half kept at 15°C. For effects on axotomy and motor neuron synapse maintenance, worms were cultivated for 24 hours after temperature shift. For effects on touch neuron axon maintenance, animals were grown 60 hours after temperature shift.
We photo-converted Dendra using a 405 nm laser (Blue Sky Research, Milpitas, CA) at 1 mW power and 40× objective for 30 seconds. Images were captured under the same conditions using 488 nm and 543 nm laser, and spinning-disk imaging as above. We analyzed two independent transgenes for the same construct, n > 6 animals per line in each experiment. We measured fluorescence intensity (ImageJ) at the cell body, in the axon 50–60 μm from the cell body (no axotomy groups), or in the first 10 μm axon distal to cutting point (for axotomy groups). We quantified the normalized fluorescence intensity increase, NIC = (Fn−F0)/(F−1−F0), where Fn = fluorescence intensity n hours after photo-conversion; F0 = fluorescence intensity immediately after photo-conversion; F−1 = fluorescence intensity before photo-conversion. We set the average NICs of wild type worms to 1. Changes are shown as a multiple of the average NIC of wild type.
In comparisons of measurements such as puncta density, fluorescence intensity, or axonal regrowth we first tested for normality using a D’Agostino-Pearson test. For comparisons of two groups we used a two-tailed Student’s t test. Comparisons involving multiple groups used one-way Anova and Bonferroni or Dunnett post tests in Graphpad Prism (GraphPad Software, La Jolla, CA). To compare variables such as axon termination proportions we used the Fisher exact test. To correct for multiple comparisons in 2 × k tables we used a Monte Carlo simulation approach followed by the Marascuilo procedure, as implemented in xlstat (Addinsoft, New York, NY). Correlation coefficients in Figure S7 were calculated using GraphPad Prism.
We thank J. Kniss and X-M. Wang for their assistance with the rpm-1 suppressor complementation tests, S. Mitani and the C. elegans gene knockout consortium for deletion alleles, Y. Kohara for cDNA clones, M. Hansen for advice on cycloheximide, K. Shen for the mec-2 ts intron construct, M. Chalfie for sharing unpublished results, L. Gavis and R. Singer for the MS2 clones, Z-L. Qiu, A. Ghosh and G. Patrick for tissue culture facilities and advice. Some strains were provided by the Caenorhabditis Genetics Center, funded by the NIH National Center for Research Resources. We are grateful to Jin and Chisholm lab members for discussions. Supported by NIH R01 awards NS35546 to Y.J. and NS57317 to A.D.C. D.Y. and Z.W are associates and Y.J. is an Investigator of the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.