Noncanonical Modulation of the eIF2 Pathway Controls an Increase in Local Translation during Neural Wiring

Abstract

Local translation is rapidly regulated by extrinsic signals during neural wiring, but its control mechanisms remain elusive. Here we show that the extracellular cue Sema3A induces an initial burst in local translation that precisely controls phosphorylation of the translation initiation factor eIF2α via the unfolded protein response (UPR) kinase PERK. Strikingly, in contrast to canonical UPR signaling, Sema3A-induced eIF2α phosphorylation bypasses global translational repression and underlies an increase in local translation through differential activity of eIF2B mediated by protein phosphatase 1. Ultrasensitive proteomics analysis of axons reveals 75 proteins translationally controlled via the Sema3A-p-eIF2α pathway. These include proteostasis- and actin cytoskeleton-related proteins but not canonical stress markers. Finally, we show that PERK signaling is needed for directional axon migration and visual pathway development in vivo. Thus, our findings reveal a noncanonical eIF2 signaling pathway that controls selective changes in axon translation and is required for neural wiring.

Keywords: PERK; Sema3A; axon; axon branching; axon guidance; eIF2B; eIF2α; local translation; retinal ganglion cell; unfolded protein response.

Graphical abstract

Figure 1

eIF2α Phosphorylation Underlies Sema3A-Induced Upregulation of Axonal Protein Synthesis (A and B) IF representative images (A) and quantification (B) for total-eIF2α and p-eIF2α in growth cones treated with Tg (15 min) or Sema3A (10 min) (unpaired t test). (C and D) IF representative images (C) and quantification (D) for puromycin in growth cones incubated with puromycin and co-treated with Tg (15 min) or Sema3A (10 min) and ISRIB (one-way ANOVA with Bonferroni’s multiple comparisons test). Error bars indicate SEM. Scale bars, 5 μm. See also Figure S1.

Figure 2

pSILAC-SP3 Reveals 75 Nascent Proteins Regulated by the Sema3A-p-eIF2α Pathway (A) Schematic of the pSILAC-SP3 methodology applied to somaless retinal axons. (B) Subset of NSPs regulated in response to Sema3A by p-eIF2α. Only significant NSP changes are shown (blue, downregulation; red, upregulation; p < 0.10). (C) KEGG pathway analysis (red, upregulated pathway; blue, downregulated pathway; cutoff ≥ 2 proteins per pathway). (D) Network-based cluster analysis of the enriched NSP changes induced by Sema3A-p-eIF2α signaling and their associated functional classes (blue nodes, downregulated NSPs; red nodes, upregulated NSPs; light blue edges, interactions known from databases; purple edges, interactions experimentally determined; green stars, NSPs belonging to the “response to stress” category; false discovery rate [FDR] < 0.05). (E and F) IF representative images (E) and quantification (F) for Gsn, Actb, and RpL7a in growth cones pre-incubated with ISRIB for 30 min and co-stimulated with Sema3A for 15 min (one-way ANOVA with Bonferroni’s multiple comparisons test). (G) Correlation analysis of pSILAC- and qIF-derived detection of protein changes (r = 0.92). Error bars indicate SEM. Scale bar, 5 μm. See also Figure S2 and Table S1.

Figure 3

Sema3A-Induced Initial Wave of Local Protein Synthesis Elicits eIF2α Phosphorylation via PERK (A) Retinal axons were immunostained for PERK, HRI, and PKR. (B and C) IF representative images (B) and quantification (C) for total-eIF2α and p-eIF2α in growth cones co-treated with Sema3A and CHX, PP242, or U0126 for 10 min (one-way ANOVA with Bonferroni’s multiple comparisons test). (D and E) IF representative images (D) and quantification (E) for p-eIF2α in growth cones co-treated with Sema3A and GSK2606414 (GSK) for 10 min (one-way ANOVA with Bonferroni’s multiple comparisons test). (F and G) IF representative images (F) and quantification (G) for puromycin in growth cones co-incubated with puromycin, Sema3A and GSK for 10 min (one-way ANOVA with Bonferroni’s multiple comparisons test). (H and I) IF representative images (H) and quantification (I) for puromycin in growth cones of embryos injected with CoMO or PERK MO co-incubated with puromycin and Sema3A for 10 min (one-way ANOVA with Bonferroni’s multiple comparisons test). Error bars indicate SEM. Scale bars, 5 μm. See also Figure S3.

Figure 4

Sema3A and Canonical UPR Signaling Differentially Control Translation by Distinct Modulation of eIF2B (A and B) IF representative images (A) and quantification (B) for eIF2Bε in growth cones co-treated with Sema3A and CHX (10 min) or Tg (15 min) (one-way ANOVA with Bonferroni’s multiple comparisons test). (C) RT-PCR for Actb (positive control; Turner-Bridger et al., 2018), Brn3a (negative control; Yoon et al., 2012), and eIF2Bε mRNAs. (D and E) IF representative images (D) and quantification (E) for p-eIF2Bε (Ser539) in growth cones co-treated with Sema3A and tautomycin (TM) or U0126 (10 min) or treated with Tg (15 min) (one-way ANOVA with Bonferroni’s multiple comparisons test). (F and G) IF representative images (F) and quantification (G) for puromycin in growth cones co-incubated with puromycin, Sema3A and TM for 10 min (one-way ANOVA with Bonferroni’s multiple comparisons test). (H and I) Immunoblot (H) and quantification (I) of puromycin signal intensity in lysates of intact brains of stage 35-36 embryos incubated with TM for 30 min and puromycilated over the last 15 min of the treatment (unpaired t test). Error bars indicate SEM. Scale bars, 5 μm. See also Figure S4.

Figure 5

Spatially Polarized Phosphorylation of eIF2α Mediates Sema3A-Induced Chemorepulsion (A) Turning assay. Arrows indicate the position of the pipette. (B) Cumulative distribution of turning assay outcome. A polarized gradient of Sema3A was generated, and ISRIB, GSK, or TM were bath-applied. Positive values indicate attraction, and negative values indicate repulsion (unpaired t test). (C) Turning assay with somaless axons. Arrows indicate the position of the pipette. Eye explants were removed immediately prior to the experiment. (D) Cumulative distribution of turning assay outcome. A polarized gradient of Sema3A was generated, and ISRIB was bath-applied. Positive values indicate attraction, and negative values indicate repulsion (unpaired t test). (E) Growth cone immunostained for p-eIF2α with a line dividing the near and far sides. Arrowheads indicate the 90° polarized gradient of Sema3A. (F) Cumulative distribution assessing the asymmetric increase of p-eIF2α with the near:far ratio method (unpaired t test). (G) Asymmetric increase of p-eIF2α, assessed by the center of mass method (unpaired t test). (H) Sema3A-induced repulsive growth cone model. p-eIF2α increases on the near-stimulus side, controlling the β-actin polarized decrease (Cagnetta et al., 2018), thus helping with asymmetric cytoskeleton deconstruction and filopodium collapse. Error bars indicate SEM. Scale bar, 5 μm. See also Figure S5.

Figure 6

PERK Signaling Is Required for Visual Pathway Development In Vivo (A) Experimental outline to investigate the contribution of axonal PERK in RGCs only and Slit1 in the optic tract pathway substrate. Unilateral MO injection leads to targeted KD in half of the nervous system. (B) Schematic of axons navigating the optic tract and reaching the tectum. TPB, tectal posterior boundary; TAB, tectal anterior boundary; TPA tectal projection angle; MDT, mid-diencephalic turn; A, anterior; P, posterior; OC, optic chiasm; Tec, tectum; Di, diencephalon; Hy, hypothalamus; Tel, telencephalon. (C–G) Representative images of DiI-filled stage 41 retinotectal projections in Control MO (C), unilateral KD of PERK in the axons (D), unilateral KD of Slit1 in the optic tract substrate (E), or both (F and G) (Ax, axon; Br, brain). (H) Cumulative distribution of MDT angle measurements in unilateral KD of PERK in the axons or Slit1 in the optic tract substrate or both (one-way ANOVA with Bonferroni’s multiple comparisons test). (I) Penetrance for MDT angles of less than 45° (Fisher’s exact test). (J) Cumulative distribution of TPA measurements in unilateral KD of PERK in the axons or Slit1 in the optic tract substrate or both. Positive values indicate angles pointing toward the TPB, and negative values indicate angles pointing toward the TAB (one-way ANOVA with Bonferroni’s multiple comparisons test). (K) Penetrance of posterior tectum avoidance, measured as TPA < mean TPA in CoMO (i.e., −8.6°) (Fisher’s exact test). (L) Single RGC axons in the tectum and line drawings of the corresponding trajectories shown by color-coded branch order: white, axon shaft; branches: red, primary; blue, secondary; yellow, tertiary. (M) Number of axon branches in the various orders and total number of branches in the PERK morphants (two-way ANOVA). (N) Length of axon branches in the PERK morphants (unpaired t test). (O) Formulation of axon complexity index (ACI). Color indicates the branch order as in (L). (P) ACI in the PERK morphants (Fisher’s exact test). Error bars indicate SEM. Scale bars, 100 μm (C–G) and 20 μm (L). See also Figure S6.

Figure 7

eIF2Bε Constitutes a Pivotal Node between the Responses to Canonical Stress and Sema3A Sema3A induces an initial (5 min) wave of local translation independent of the eIF2 pathway mediated by ERK-1/2 and mTOR. Simultaneously, eIF2Bε is locally translated in an ERK-1/2-mTOR-independent manner. The rapid increase in local protein synthesis triggers eIF2α phosphorylation via PERK at 10 min stimulation. Within this time course, ERK-1/2 represses GSK-3β and activates PP1, dephosphorylating eIF2Bε and increasing eIF2B activity. The engagement of p-eIF2α and increased eIF2B GEF activity generates a specific level of ternary complex higher than in the canonical stress response, resulting in uORF-independent selective translation of 75 NSPs, upregulating global translation. +p, phosphorylation, −p, dephosphorylation, ↑ and ↓, axonal translation upregulation and downregulation; dashed lines, indirect activation following a rise in local protein synthesis; dotted lines, interaction; dashed circles, timing.