IRES-mediated translation of cofilin regulates axonal growth cone extension and turning

Abstract

In neuronal development, dynamic rearrangement of actin promotes axonal growth cone extension, and spatiotemporal translation of local mRNAs in response to guidance cues directs axonal growth cone steering, where cofilin plays a critical role. While regulation of cofilin activity is well studied, regulatory mechanism for cofilin mRNA translation in neurons is unknown. In eukaryotic cells, proteins can be synthesized by cap-dependent or cap-independent mechanism via internal ribosome entry site (IRES)-mediated translation. IRES-mediated translation has been reported in various pathophysiological conditions, but its role in normal physiological environment is poorly understood. Here, we report that 5'UTR of cofilin mRNA contains an IRES element, and cofilin is predominantly translated by IRES-mediated mechanism in neurons. Furthermore, we show that IRES-mediated translation of cofilin is required for both axon extension and axonal growth cone steering. Our results provide new insights into the function of IRES-mediated translation in neuronal development.

Keywords: axonal growth cone; cofilin; internal ribosome entry sites.

Figure 1. 5′UTR of cofilin harbors IRES activity

1. A

   5′UTR of cofilin shows strong IRES activity, while reversely oriented 5′UTR of cofilin lacks IRES activity. This was measured by using a bicistronic reporter system. Cap‐dependent (renilla luciferase, RLUC) and cap‐independent (firefly luciferase, FLUC) translation can be distinguished by determining the ratio of FLUC to RLUC. FLUC cannot be synthesized unless DNA fragment inserted into the intercistronic region contains IRES activity. FLUC and RLUC activities were normalized to β‐galactosidase activity, which is used as an independent transfection control. For IRES activity measurement, Neuro2A cells were transfected with pRF (backbone of bicistronic reporter system), pRF containing 5′UTR of cofilin (5′UTR), or reversely oriented 5′UTR of cofilin (R‐5′UTR). *P < 0.0001, N = 5 independent experiments. Values shown are mean ± SEM and are tested for statistical significance by one‐way ANOVA followed by Bonferroni post hoc test.

2. B

   Both pCMV‐β‐gal vector and pRF (backbone of bicistronic reporter system) containing 5′UTR of cofilin (5′UTR) or 5′UTR lacking a CMV promoter (∆CMV) were transfected into Neuro2A cells, and dual‐luciferase assays were performed. FLUC and RLUC activities were normalized to β‐galactosidase activity. N = 3 independent experiments. Values shown are mean ± SEM.

3. C, D

   pCMV‐β‐gal vector and the indicated plasmids were co‐transfected with siRNA targeting RLUC. After 48‐h transfection, Neuro2A cell lysates were used for measuring luciferase activity. FLUC and RLUC activities were normalized to β‐galactosidase activity. RLUC siRNA reduced RLUC but not FLUC in cells transfected with psi‐Check2 (control). While RLUC siRNA significantly decreased RLUC, FLUC still exhibited strong activity in cells transfected with cofilin 5′UTR. Apaf‐1 5′UTR also showed FLUC activity compared to that of cofilin R‐5′UTR. Note that Apaf‐1 IRES activity is significantly lower than that of cofilin. N = 3. Values shown are mean ± SEM and are tested for statistical significance by one‐way ANOVA followed by Bonferroni post hoc test.

4. E

   A secondary structure of 5′UTR of cofilin having 145 nucleotides was prepared with mfold software, which produced three different regions containing loop domains. Deletion of nucleotides (∆1–34, ∆62–84, or ∆91–111) removed the D1, D2, or D3 loop, respectively. Also, deletion of D1 together with D2 (∆1–84) or all loops including D1, D2, and D3 (∆1–92) was generated.

5. F

   RLUC and FLUC activities were measured with different mutations of cofilin 5′UTR. N = 5. Values shown are mean ± SEM.

6. G

   Bicistronic mRNA transcripts prepared by in vitro transcription were transfected into Neuro2A cells and incubated for 24 h, and then, IRES activity was measured. Values are normalized to the cells containing mRNA transcripts prepared from pRF vector used in (A and E). m⁷G: 7‐methyl‐guanosine. *P < 0.0001, N = 4. Values shown are mean ± SEM and are tested by one‐way ANOVA followed by Bonferroni post hoc test.

Figure 2. IRES‐mediated cofilin translation extends axon growth

1. A bicistronic reporter system using mCherry and eGFP fluorescent reporters was adopted to determine IRES activity in hippocampal primary neurons. 5′UTR of cofilin, reversely oriented 5′UTR (R‐5′UTR), or 5′UTR with ∆1–34 (∆1–34) was inserted to the intercistronic region of the bicistronic reporter vector, respectively. These vectors were transfected to neurons at DIV1 (days in vitro) and IRES activity measured at DIV3. Growth cones were outline by white line. Scale bar = 2 μm.

2. IRES activity was determined by the ratio of mCherry/eGFP signals in axonal growth cones at DIV3. *P < 0.0001, N = 15 neurons. Values shown are mean ± SEM and are tested for statistical significance by one‐way ANOVA followed by Bonferroni post hoc test.

3. Cofilin was expressed in hippocampal primary neurons by IRES‐mediated translation. 5′UTR‐cofilin‐eGFP or 5′UTR(∆D1)‐cofilin‐eGFP construct was inserted into the intercistronic region. The vectors were transfected at DIV1, and the axon length of neurons was measured at DIV3. Scale bar = 50 μm.

4. IRES‐mediated cofilin expression significantly increased axonal length compared to that of neurons lacking IRES‐mediated cofilin translation. *P < 0.0002, N = 42 for neurons containing 5′UTR‐cofilin‐eGFP, and n = 35 for neurons having 5′UTR(∆D1)‐cofilin‐eGFP. Values shown are mean ± SEM and are tested for statistical significance by one‐way ANOVA followed by Bonferroni post hoc test.

5. Real‐time quantitative PCR for mCherry and eGFP mRNA transcripts was performed on vectors used in (B). N = 3 independent experiments. Values shown are mean ± SEM and are tested by one‐way ANOVA followed by Bonferroni post hoc test.

Figure 3. Cofilin synthesis in neurons utilizes IRES‐mediated translation mechanism

1. Primary hippocampal neurons were transfected with a vector that expressed FLAG either under regulation of cofilin 5′UTR or cofilin 5′UTR with depleted D1 (∆1–34) region at DIV 1. eGFP was co‐transfected. FUNCAT‐PLA was performed at DIV2 and azidohomoalanine (AHA) was treated for 2 h. AHA replaced methionine in neurons taking up AHA, which was labeled by biotin. The FUNCAT‐PLA method using combination of biotin‐labeled AHA and FLAG antibody was able to detect newly synthesized FLAG. Anisomycin blocked to synthesize new FLAG peptide in neurons containing cofilin 5′UTR‐FLAG, and neurons containing cofilin‐5′UTR (∆1–34)‐FLAG failed to express FLAG. Scale bar = 20 μm.

2. Neuro2A cells treated with RAD001 showed little effect on cofilin expression, while cycloheximide (CHX) reduced cofilin translation. Conversely, a downstream target of mTOR, phospho‐rpS6, showed no phosphorylation after 3 h of RAD001 treatment, while cycloheximide did not affect the level of phospho‐rpS6.

3. Expression of eIF4E and c‐Myc was reduced by eIF4E siRNA transfection to Neuro2A cells, while cofilin level was increased. Cell lysates were prepared 3 days after transfection of siRNA. *P < 0.05, N = 4 independent experiments. Values shown are mean ± SEM and are tested by t‐test.

Source data are available online for this figure.

Figure 4. Cofilin expression prefers IRES‐mediated translation to cap‐dependent translation

1. A, B

   The D1 region of cofilin 5′UTR was deleted using the CRISPR–Cas9 method; then, the genome‐edited cell line was selected. Blue bars indicate the location of paired sgRNAs. Red arrowheads show expected deletion of the genomic DNA. Genotyping analysis identified a Neuro2A cell line lacking the D1 region.

2. C

   Cofilin protein expression was effectively reduced in the genome‐edited cell line compared to the control.

3. D

   Left: Quantification of cofilin expression from (C). *P < 0.0001, N = 3 independent experiments. Right: Real‐time quantitative PCR for cofilin mRNA transcripts in the genome‐edited cell line showed a slight decrease. *P < 0.04, N = 3 independent experiments. Values shown are mean ± SEM and are tested by t‐test.

Source data are available online for this figure.

Figure 5. Overexpression of cofilin by IRES‐mediated translation increases axonal growth cone extension

1. A

   To verify the role of IRES activity in regulation of cofilin translation, cofilin 5′UTR, 5′UTR lacking D1 region, or 5′UTR having mutations in D1 was inserted to the upstream of renilla luciferase gene in psi‐Check2 dual‐reporter vector, which contains two promoters for RLUC and LUC+ (modified firefly luciferase gene), respectively. The plasmids were transfected into Neuro2A cells, and dual‐luciferase assays were performed.

2. B

   Insertion of 5′UTR increased RLUC, while no insert, 5′UTR lacking D1 region, or mutated 5′UTR did not increase RLUC activity. *P < 0.0001, N = 4. Values shown are mean ± SEM and are tested for statistical significance by one‐way ANOVA followed by Bonferroni post hoc test.

3. C

   Real‐time quantitative PCR for RLUC and LUC+ mRNA transcripts showed no difference between cells transfected with the indicated constructs, N = 3. Values shown are mean ± SEM and tested for statistical significance by one‐way ANOVA followed by Bonferroni post hoc test.

4. D, E

   Vectors containing 5′UTR‐cofilin, 5′UTR‐cofilin lacking IRES activity (∆1–34), or no 5′UTR‐cofilin were transfected to hippocampal primary neurons at DIV2, and axon length at DIV3 was measured. Axon length was significantly increased in neurons containing 5′UTR‐cofilin, while neurons with 5′UTR‐cofilin lacking IRES (∆1–34) showed no axon extension compared to that of the control. *P < 0.001, N = 30 for each condition. Values shown are mean ± SEM and are tested for statistical significance by one‐way ANOVA. Scale bar = 50 μm.

5. F, G

   GFP expression level in primary neurons containing the vector used above (D). *P < 0.01, N = 3. Values shown are mean ± SEM and are tested by one‐way ANOVA followed by Bonferroni post hoc test.

Source data are available online for this figure.

Figure 6. nPTB is required for IRES‐mediated cofilin translation

1. A

   Neuro2A cell extracts were used to test whether nPTB is binding to the biotinylated cofilin 5′UTR. Streptavidin–biotin RNA affinity purification was performed, and purified samples were separated by SDS–PAGE and immunoblotted with antibody against nPTB. For competition assay, nonlabeled cofilin 5′UTR (10×) was also co‐incubated.

2. B

   IRES activity was determined with a bicistronic reporter system used in Fig 1A together with control siRNA or nPTB siRNA in Neuro2A cells. *P < 0.001, N = 3. Values shown are mean ± SEM and are tested for statistical significance by two‐way ANOVA followed by Bonferroni post hoc test.

3. C

   Cofilin expression level was decreased in Neuro2A cells containing nPTB siRNA compared to that of control siRNA. GAPDH was a loading control. Amounts of cofilin mRNA transcripts were not altered. N = 3. Values shown are mean ± SEM, and tested for statistical significance by one‐way ANOVA followed by t‐test.

4. D, E

   Cofilin expression was reduced in primary hippocampal neurons containing nPTB siRNA compared to that of neurons having control siRNA. nPTB siRNAs and GFP vector were co‐transfected to neurons at DIV1 and analyzed at DIV3. Immunocytochemistry was performed using antibodies against cofilin and nPTB, respectively. Scale bar = 20 μm. *P < 0.05, **P < 0.01, N = 18 for each condition. Values shown are mean ± SEM and are tested for statistical significance by t‐test.

5. F, G

   Axon length was significantly decreased in neurons containing nPTB siRNA. Experimental condition was the same as in (D). Scale bar = 50 μm. *P < 0.001, N = 28 for each condition. Values shown are mean ± SEM and are tested for statistical significance by t‐test.

6. H

   The Dunn chamber experiments produced trajectory plots on axonal growth cone turning response to Sema3A gradient. Axons containing nPTB siRNA did not turn away from Sema3A gradient, while axons having control siRNA repelled from Sema3A. Sema3A gradient was established to increase along the y‐axis, and all axons are re‐positioned to it accordingly for clarity. Black lines show the initial 10 μm of axons, and different colors represent the initial turning angles of axonal growth cones over a 2‐h recording period. To reduce complexity of the figure, only 10 representative traces are shown. N > 12 transfected neurons were analyzed for each group.

7. I

   Axonal growth cones of neurons containing nPTB siRNA did not show any response in the Dunn chamber assay regardless of the presence of Sema3A gradient, while axonal growth cones with control siRNA were notably turned away from Sema3A gradient. *P < 0.001. Values shown are mean ± SEM and are tested for statistical significance by two‐way ANOVA followed by Bonferroni post hoc test.

Source data are available online for this figure.