1. Neuroscience
Download icon

CELF RNA binding proteins promote axon regeneration in C. elegans and mammals through alternative splicing of Syntaxins

  1. Lizhen Chen
  2. Zhijie Liu
  3. Bing Zhou
  4. Chaoliang Wei
  5. Yu Zhou
  6. Michael G Rosenfeld
  7. Xiang-Dong Fu
  8. Andrew D Chisholm  Is a corresponding author
  9. Yishi Jin  Is a corresponding author
  1. University of California, San Diego, Division of Biological Sciences, United States
  2. Howard Hughes Medical Institute, University of California, United States
  3. University of California, San Diego, School of Medicine, United States
Research Article
Cite this article as: eLife 2016;5:e16072 doi: 10.7554/eLife.16072
Voice your concerns about research culture and research communication: Have your say in our 7th annual survey.
7 figures, 3 videos and 7 additional files


Figure 1 with 3 supplements
UNC-75 is required cell autonomously for axon regeneration.

(A) Images of regenerating PLM axons at 24 hr post axotomy; anterior is to the left and dorsal up. Red asterisk: PLM cell body; red arrow: injury site. (B) Quantitation of PLM axon regrowth 24 hr post laser axotomy, normalized to wild type. The unc-75 alleles md1309 and md1344 display defects in axon regrowth similar to e950. The unc-75(e950) PLM regrowth defect is rescued by multicopy and single copy unc-75(+) transgenes. Statistics: One-way ANOVA with Bonferroni post test. (C) unc-75(e950) animals showed reduced axon regrowth at all time points examined, and reduced growth rate 0–24 hr post axotomy. (D) Representative image series from time-lapse movies of the tip of a regenerating PLM axon from wild-type and unc-75(e950) animals starting at 14 hr post axotomy; see Video 1 and 2. Red and orange arrows point to the ends of regenerating axons at 14 and 15h post axotomy respectively. (E) unc-75(e950) is defective in regeneration of GABAergic motor neurons [marked by Punc-25-GFP(juIs76)]; this was rescued by Prgef-1-UNC-75 (juSi76). Images of GABAergic motor neuron commissures at 24 hr post axotomy. DD2 and VD4 were axotomized, and VD3 was uncut. Red arrowheads indicate the ends of regenerating or non-regenerating axons. Scale: 10 μm. Bar charts showing reduced regrowth of DD2 and VD4 neurons; N = 30–52.

Figure 1—figure supplement 1
unc-75 alleles and role of nuclear localization.

(A) Schematic of unc-75 deletion mutations. The deletion e950 removes exons 1–5 as well as 5’ upstream sequences; the breakpoint of the deletion is not known. md1309 is a deletion affecting exons 4–7 but the exact boundary is not known. More information can be found in Loria et al., 2013. md1344 is a 780 bp deletion that deletes most of intron 8 and the first 15 bp of exon 9. By determining the sequences of transcripts from unc-75(md1344) we found that splicing of intron 8 is altered, resulting in the inclusion of 26 bp of intron 8 before exon 9 and deletion of 15 bp of exon 9. md1344 causes a frame shift in the resulting transcript, leading to expression of a truncated UNC-75 protein containing aa 1–472. (B) Images of PLM cell body showing GFP::UNC-75 protein localization. Wild type UNC-75 is predominantly localized to the nucleus. A truncated protein containing 472 amino acids corresponding to that encoded by md1344 allele localizes to cytoplasm. Scale bar: 5 µm. (C) Normalized PLM regrowth 24 hr post axotomy. Wild type UNC-75 cDNA expressed using the unc-75 promoter was able to rescue the regrowth defect in unc-75 mutant, but a mutant cDNA lacking the NLS (nuclear localization signal) failed to rescue. (D) Representative images of unc-75(e950) and unc-75(e950); Prgef-1-FLAG::UNC-75(juIs369) (CZ14662, as used for neuronal UNC-75 CLIP-seq) animals. The Unc phenotype in unc-75(e950) was rescued by juIs369. Scale: 500 µm.

Figure 1—figure supplement 2
Regenerative growth cones in unc-75 and other mutants.

(A) Representative PLM axon regrowth images of indicated genotypes. Asterisk: PLM cell body; red arrow: injury site. Scale: 10 μm. (B) Percentage of regenerating axons with a growth cone-like structure at 24 hr post axotomy. Regenerative growth cones were observed at significantly higher frequency in unc-75 and synaptic transmission mutants compared to WT. Statistics: One-way ANOVA with Bonferroni post test. N ≥ 3 experiments, each experiment involved 10 or more animals.

Figure 1—figure supplement 3
unc-75 acts in parallel to other axon regeneration pathways.

(A) exc-7 is not required for axon regrowth and does not enhance unc-75(0). N ≥ 10. Statistics: One-way ANOVA with Bonferroni post test. (B) Normalized PLM regeneration 24 hr post axotomy. The dlk-1(0); unc-75(0) double mutant resembles dlk-1(0) single mutant (no growth cone and no axon extension). (C) Overexpression of DLK-1 or loss of EFA-6 can partially rescue the defects of unc-75(e950) mutant, suggesting that UNC-75 functions in a parallel pathway. (D) RT-qPCR data showing no change in dlk-1 or efa-6 transcripts in unc-75(e950) mutant. N = 4. Two sets of efa-6 primers (one set at the 3’ end, the other at the middle of mRNA) were used.

Figure 2 with 1 supplement
CLIP-seq of UNC-75 in C. elegans neurons.

(A) Flow chart of CLIP-seq analysis that identified 533 potential mRNA targets bound by UNC-75. (B) The top two motifs enriched in UNC-75 CLIP-seq peaks, based on MEME analysis of the sequence of bound mRNAs. (C) UNC-75 CLIP-seq peaks in unc-75, nrx-1, and unc-41, displayed using the UCSC Genome Browser. Splicing variants are shown under each genomic locus. Sequence of the annotated CIMS peak is shown next to the peak. 's' in peak number stands for 'substitution'. Red and blue tags indicate the two different gene orientation on chromosomes. Scale, 5 kb.

Figure 2—figure supplement 1
Purification of UNC-75 and CELF2 bound RNA using CLIP.

(A) Autoradiogram showing size-separated crosslinked UNC-75-RNA complexes following complete digestion with high, or partial digestion with low amounts of micrococcal nuclease, immunopurification with an anti-FLAG antibody (or IgG for control) and 5’ end radiolabeling. The red box depicts the areas on the nitrocellulose membrane from which crosslinked RNAs were purified for reverse transcription and deep sequencing. (B) Autoradiogram showing size-separated crosslinked CELF2-RNA complexes following digestion with micrococcal nuclease at different dilution. The red box depicts the areas on the nitrocellulose membrane from which crosslinked RNAs were purified for reverse transcription and deep sequencing.

Figure 3 with 3 supplements
UNC-75 regulates alternative splicing of unc-64/Syntaxin in neurons.

(A) UNC-75 CLIP-seq peaks in the unc-64 locus; genomic track display from UCSC Genome Browser. (B) In wild-type animals, UNC-64A::GFP is strongly expressed in most neurons. In unc-75(0) mutants UNC-64A::GFP is expressed at lower levels in most neurons. Images of nerve ring and head neurons. (C) UNC-64B::GFP is not expressed in the nervous system in wild-type background but is ectopically expressed in head neurons in unc-75(0). (D) Deletion of the 38 bp UNC-75 binding site in intron 7 results in neuronal expression of UNC-64B::GFP in wild-type background. (E) Images of anterior portion of animals expressing the UNC-64A/B splicing reporter in wild-type and unc-75(0) mutant backgrounds. unc-75(0) is e950. The splicing reporter contains the 3’ part of unc-64 genomic sequence (boxes are exon7, 8a and 8b, lines are intron 7a and 7b); RFP is inserted at the end of exon 8a and GFP inserted at the end of exon 8b. For B-E, scale bar 20 μm. Bar charts show quantitation of fluorescence intensity in the nerve ring region (ROIs shown in dashed boxes). Statistics: Student’s t-test. N=5–10.

Figure 3—figure supplement 1
UNC-64 expression is regulated by UNC-75.

(A) Positions of unc-64 exons and PCR primers used in RT-qPCR. Primers for common exons (black arrow) and isoform specific exons (red arrow for a isoform and green arrow for b isoform) used in RT-qPCR were indicated. Relative mRNA expression levels of unc-64 detected by RT-qPCR are shown in the bar graph. Statistics: Student’s t-test. N = 4–8. (B) Immunostaining using an antibody recognizing the UNC-64 N terminus. Quantification of expression in the nerve ring (ROI enclosed by dashed line) is shown in the bar graph. Statistics: Student’s t-test. N=5–10.

Figure 3—figure supplement 2
unc-64 RNA splicing in wild type and unc-75 mutants.

3’ RNAseq data are displayed using the UCSC Genome Browser. The two alternatively spliced isoforms of unc-64 (a and b) are shown at the bottom. Solid blue boxes represent exons and lines with arrowheads represent introns. For reads from RNA-seq, each block represents one tag/read. Two blocks linked by a line represent one tag mapped to two adjacent exons separated by an intron (showed as the line linking two blocks). There are reads mapped to intron 7a and 7b in both wild type and unc-75(e950) animals. In unc-75 mutants more reads mapped to intron 7a and fewer mapped to intron 7b, consistent with reduced expression of unc-64a isoform and increased unc-64b isoform. The inserted table shows the number of RNA-seq reads mapped to intron 7a and intron 7b from four biological replicates.

Figure 3—figure supplement 3
UNC-64A and UNC-64B have distinct roles in neuronal function.

Representative images and quantification of locomotion velocity of animals with indicated genotypes. Locomotion defects in unc-64(md130) and unc-64(e246) mutants were rescued by expression of UNC-64A cDNA driven by pan neuronal promoter (Prgef-1), but not by expression of UNC-64B using the same pan-neuronal promoter. Statistics: One way ANOVA with Bonferroni post test. Scale: 500 µm.

Figure 4 with 1 supplement
unc-64 is required cell autonomously for PLM axon regeneration.

(A) Normalized PLM regrowth 24 hr post axotomy. PLM axon regeneration is reduced in mutants with unc-64 partial loss of function alleles md130 and e246, as well as null allele js115. These alleles affect both isoforms (Saifee et al., 1998). unc-64(md130) regeneration phenotypes are rescued by pan-neural expression of UNC-64A or UNC-64B, but not by UNC-64∆TM. Expression of UNC-64A or B in a wild type background does not affect PLM regeneration. Statistics, One-way ANOVA followed by Bonferroni's Multiple Comparison Post Test. N ≥ 10. (B) Schematic illustration of two strategies to generate unc-64 mutation in touch neurons. The lethality of unc-64(js115) is rescued by oxEx705 or juSi316. juSi316 was crossed to Pmec-7-nCre to delete transgenic UNC-64 in touch neurons. (C) Representative images of animals with indicated genotypes. js115; oxEx705 animals were viable but severely Unc, while js115; juSi316 animals were viable and slightly Unc. (D) Representative PLM regrowth images 24 hr post axotomy. Asterisks: PLM cell body; red arrow, injury site. Scale bar: 100 μm.

Figure 4—figure supplement 1
UNC-64 is not required for PLM development.

PLM neurons labeled by muIs32(Pmec-7-GFP) in young adult animals of indicated genotypes. Cre-induced deletion of the floxed unc-64 allele in juSi316 did not affect PLM development. Red asterisks: PLM cell body. White arrow points to the terminus of PLM axon.

Overexpression of UNC-64A suppresses unc-75 neuronal phenotypes.

(A-B) Quantitation (n ≥ 10 per genotype) and images of PLM axon regeneration 24 hr post axotomy; scale, 10 μm. Asterisks: PLM cell body; red arrow, injury site. Transgenic UNC-64 expression using 9 kb genomic DNA encoding both UNC-64A and UNC-64B was able to partially rescue the regeneration defect of unc-75(0). Pan-neuronal expression of UNC-64A cDNA, but not UNC-64B, significantly increased axon regrowth in unc-75(0) mutants. (C-D) Quantitation of locomotion velocity and images of animals with indicated genotypes; scale, 0.5 mm. N ≥ 10 per genotype; statistics: One way ANOVA with Bonferroni post test. Genomic DNA or UNC-64A cDNA driven by Pan-neuronal promoter partially rescues the unc-75 locomotor phenotype; expression of UNC-64B or UNC-64ΔTM cDNA does not rescue. e950 is used in unc-75(0).

Figure 6 with 1 supplement
CELF proteins play conserved roles in axon regeneration.

(A) Expression of full length cDNAs of mouse CELF2 or CELF4 with mCherry tag in C. elegans touch neurons significantly rescues the unc-75 axon regrowth defect. Normalized PLM axon regrowth 24 hr post axotomy, N = 14–28. Based on mCherry fluorescence the CELF2 and CELF4 transgenes are expressed at similar levels (not shown). (B) Celf2 transcript levels in DRG neurons decrease during postnatal development, whereas Celf4 levels increase. Expression was normalized to P1, and mouse β-Actin was used as internal reference. Statistics, One-way ANOVA followed by Bonferroni's Multiple Comparison Post Test. N = 4–6. (C) Expression of Celf2 transcripts in DRG of 2 month old mice is significantly enhanced 3 days after sciatic nerve injury. Ratio of the crushed side to the uncrushed side within the same animal is plotted. Statistics, Student’s t-Test. (D) Mutation of Celf2 impairs axon regeneration in DRG PV+ neurons. Confocal images of longitudinal sciatic nerve sections distal to the lesion, stained with anti-SCG10 (green) at 3 days post crush. tdTomato expression is from Rosa26-lox-STOP-lox-tdTomato and marks neurons in which Cre was active. Enlarged images of the boxed areas are shown on the right; white dashed line marks the lesion site. Scale bar: 0.5 mm. (E) Quantitation of SCG10 intensity in tdTomato positive axons at different distances from the lesion site, normalized to SCG10 intensity proximal to the lesion. 6 control and 5 mutant animals were analyzed. Statistics, One-way ANOVA followed by Bonferroni's multiple comparison post test.

Figure 6—figure supplement 1
Celf2 is required for neurite growth in mouse DRG neurons.

(A) Dendrogram created using Clustal Omega showing mouse CELF1-6, C. elegans UNC-75 and ETR-1, as well as Drosophila Bruno. (B) Schematic illustration of mouse gene targeting vector and the Celf2flox allele. Black boxes represent exons and solid lines represent introns. Exon 3 is flanked by two loxP sites. (C) Loss of CELF2 impairs neurite growth of E13.5 dorsal root ganglia (DRG) explants. '-' indicates null allele derived from crossing to ZP3-Cre. (D) Quantification of the average neurite length of each DRG explant. N= 5 animals for control and 5 for mutant. DRGs from each animal were cultured in two wells. Statistics, Student’s t-test. Scale bar, 300 µm. (E) Nestin-cre driven tissue-specific knockout of Celf2 causes reduced animal size. (F) Cultured adult DRG neurons from Celf2 mutant show significantly reduced neurite growth, compared to DRG neurons from littermate controls. In this in vitro axon regeneration paradigm, primary cultured neurons were resuspended and re-plated then fixed after 24 hr. (G) Quantitation of average DRG neurite length in each well. N= 5 animals for control and 5 for mutant. DRGs from each animal were cultured in two wells. Statistics, Student’s t-test. Scale bar, 200 µm.

CELF2 regulates expression of specific neuronal Syntaxin isoforms.

(A) Genome browser tracks displaying CELF2 CLIP-seq peaks on the stx2 and stx16 gene loci. Red and blue tags indicate the two different gene orientation on chromosomes. CELF2 binding peaks are mapped to introns near alternatively spliced exons. Isoform labeling is consistent to gene annotation on Ensembl. Red arrows under the exons implicate primers used for RT-qPCR in panel b. (B) Transcript levels of Syntaxin genes were measured by RT-qPCR in E15.5 control and Celf2-/- brains. Statistics, Student’s t-test, N = 5–6. Expression of the alternatively spliced isoform 002 of stx2 and isoform 001 of stx16 was significantly decreased in Celf2-/- constitutive mutants.



Video 1
Time-lapse movie of the tip of a regenerating PLM axon from a wild type animal, starting at 14 hr post axotomy, ending at 15 hr post axotomy.
Video 2
Time-lapse movie of the tip of a regenerating PLM axon from an unc-75(e950) mutant, starting at 14 hr post axotomy, ending at 15 hr post axotomy.
Video 3
Time-lapse movie of the tip of a regenerating PLM axon from an unc-64(md130) mutant, starting at 14 hr post axotomy, ending at 15 hr post axotomy.

Additional files

Supplementary file 1

mRNA targets identified by UNC-75 CLIP-seq in C. elegans neurons

Supplementary file 2

Comparison of UNC-75 targets identified in this study with those defined by RNAseq.

Supplementary file 3

Summary of expression patterns of mouse CELF transcripts based on Allen Brain Atlas.

Supplementary file 4

mRNA targets identified by CELF2 CLIP-seq in N2a cells.

The top 2919 CLIP-seq peaks are ranked by significance score.

Supplementary file 5

C. elegans strains, transgenes, and clones.

Supplementary file 6

Primers used for RT-qPCR analyses.

Supplementary file 7

Raw measurement data of regrowth.


Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)