Regulation of axon guidance by compartmentalized nonsense-mediated mRNA decay

Abstract

Growth cones enable axons to navigate toward their targets by responding to extracellular signaling molecules. Growth-cone responses are mediated in part by the local translation of axonal messenger RNAs (mRNAs). However, the mechanisms that regulate local translation are poorly understood. Here we show that Robo3.2, a receptor for the Slit family of guidance cues, is synthesized locally within axons of commissural neurons. Robo3.2 translation is induced by floor-plate-derived signals as axons cross the spinal cord midline. Robo3.2 is also a predicted target of the nonsense-mediated mRNA decay (NMD) pathway. We find that NMD regulates Robo3.2 synthesis by inducing the degradation of Robo3.2 transcripts in axons that encounter the floor plate. Commissural neurons deficient in NMD proteins exhibit aberrant axonal trajectories after crossing the midline, consistent with misregulation of Robo3.2 expression. These data show that local translation is regulated by mRNA stability and that NMD acts locally to influence axonal pathfinding.

Figure 1. Robo3.2 Protein Levels Are Induced by Floor Plate Signals

A) Schematic representation of the Robo3.2 transcript and its expression pattern in spinal cord commissural neurons. Robo3.2 protein, green, is detected exclusively in the postcrossing segments of commissural axons. B) Schematic of the half open book explant system. Half open book explants from E10.5 mouse spinal cords were cultured with (+FP) or without (−FP) the floor plate (indicated in pink). (C and D) Immunostainings of Robo3 isoforms in −FP and +FP explants. Robo3.2 protein is detected only +FP axons (C). Quantification of results in C (D) (Robo3.2 staining, −FP n=41 explants, +FP n=42 explants; Robo3.1 staining, −FP n=37, +FP n=40). (E-H) Schematic of floor plate-conditioned media (FCM) experiment (E). FCM is sufficient to induce Robo3.2 expression in precrossing axons (F, right panels). High power images depicting prominent Robo3.2 labeling at axonal tips (G, indicated by arrows). Quantification of Robo3.2 levels following treatment with FCM (n=26 explants) indicated a 10-fold increase in Robo3.2 protein levels compared to untreated −FP axons (n=23 explants) (H). Data represented as mean +/− SEM. Scale bar, (C, F) 150 μm; (G) 60 μm. See also Figure S1.

Figure 2. Robo3.2 mRNA Is Translationally Repressed Prior to Midline Crossing

A) Robo3.2 is not a target of NMD in precrossing commissural neurons. qRT-PCR did not display a significant change in Robo3.2 mRNA levels in cycloheximide-treated cells compared to untreated samples. NMD target Arc was increased 7-fold upon treatment with cycloheximide for 4 hr. Data are represented as mean +/− SEM, n=3 biological replicates/condition, ***p < 0.001. (B) Robo3.2 is not translated in precrossing neurons. Polysome sedimentation was performed from E10.5 – FP explants. In addition to RNA absorbance profiles, FMRP immunoblotting was used as a marker for polysomes. Robo3.2 mRNA was detected primarily in lighter fractions that are not associated with translating ribosomes. EDTA, which results in disruption of polysomes, relocalized FMRP and Robo3.1 transcripts to the non-translating fractions. The position of these markers is identical to the position of Robo3.2, confirming that Robo3.2 mRNA is in non-translating fractions. See also Figure S2.

Figure 3. Robo3.2 mRNA Is Transported into Pre- and Postcrossing Commissural Axons

(A-D) Detection of endogenous Robo3.2 mRNA in commissural axons by FISH. Antisense riboprobes against Robo3.2 mRNA resulted in punctuate labeling along axons in both −FP (A, B) and +FP (C, D) explants. Tau (red) immunolabeling was used to visualize axons. High power images show prominent labeling at the distal tips of the axons (B and D, indicated by arrows). Scale bar, (A and C, axons) 150 μm; (A and C, cell bodies) 60 μm; (B and D) 60 μm. (E) Schematic of a microfluidic chamber that is used to isolate commissural axons from half open book explants. Half open book explants were cultured in the cell body compartment. The microgrooves in microfluidic devices ensure that no cell bodies enter into axonal compartment. (F and G) Detection of endogenous Robo3.2 mRNA in purified commissural axons by RT-PCR. Robo3.2 transcripts were detected in axons of both −FP and +FP explants. RhoA mRNA and gamma-actin mRNA were used as positive and negative controls, respectively. Quantitative analysis of endogenous Robo3.2 mRNA in purified −FP and +FP axons (G) (n=3 biological replicates (65 explants/replicate). Consistent with the data in F, Robo3.2 mRNA is present in pre- and postcrossing axons. **p < 0.01. See also Figure S3.

Figure 4. Robo3.2 Is Locally Translated in Commissural Axons

(A and B) Robo3.2 is locally translated in postcrossing axons. +FP explants were cultured in microfluidic chambers. Axonal treatment of cycloheximide (12 hr, 10 μM) resulted in a more than 90% reduction in Robo3.2 protein levels in postcrossing axons (B) (n=120 axons/condition). Data are represented as mean +/− SEM, ***p < 0.001. (C and D) FCM induces local translation of Robo3.2. Schematic of the experimental design (C). Application of FCM to severed −FP axons resulted in prominent axonal labeling of Robo3.2 protein (D). This effect was blocked by application of 10 μM cycloheximide, indicating that Robo3.2 induction in axons is translation dependent. (E-G) Robo3.2 is locally translated in postcrossing axons in the absence of commissural cell bodies. Schematic representation of the experimental design (E). +FP Axons from Wld^S mice were assayed to monitor Robo3.2 in spinal cord explants in which the cell bodies were transected from the axons before axons reach the midline. Severed axons from Wld^S grew through the floor plate with no degeneration. Severed axons induced Robo3.2 protein after crossing the midline in the absence of cell bodies (F, right panels). Scale bar, (A) 75 μm, (D) 200 μm, (F, G) 60 μm. See also Figure S4.

Figure 5. NMD Regulates Robo3.2 Protein Levels in Postcrossing Commissural Axons

(A) Robo3.2 mRNA becomes a target of NMD following exposure to floor plate-conditioned media. FCM resulted in 70% lower levels of Robo3.2 mRNA in commissural cell bodies compared to control-treated explants. This reduction was blocked by treatment with 10 μM cycloheximide suggesting that Robo3.2 degradation upon FCM is NMD-dependent. (B-D) Upf2 (B) Upf1 (C) and Smg1 (D) are localized to axons, with increased levels at axonal tips. (E-G) FCM treatment resulted in higher Robo3.2 levels in −FP axons from Upf2 cKO compared to −FP axons from control (E). Quantifications of results in E (F) (110 axons per control (n=3) and mutant (n=4) embryos). Quantification of Robo3.2 protein in severed axons following treatment with FCM (G) (460 control axons (n=11 explants, 4 embryos) and 410 Upf2 cKO axons (n=10 explants, 3 embryos)). (See also Figures S6A and S6B). (H and I) Robo3.2 immunostaining in +FP axons from control and Upf2 cKO (H). Robo3.2 is 3.5-fold higher in postcrossing axons of Upf2 cKO compared to control axons (I) (211 axons, control (n=11 axonal areas, 4 embryos) and 217 axons, mutant (n=13 axonal areas, 5 embryos) embryos). (J and K) Quantification of Robo3.2 mRNA by qRT-PCR in isolated unsevered (J) and severed (K) axons following induction of Robo3.2 translation by FCM. Axons were harvested from control and Upf2 cKO explants that were cultured in microfluidic chambers. Both Upf2 cKO −FP axons that were treated with FCM or Upf2 cKO +FP axons that encountered the floor plate have higher levels of Robo3.2 mRNA compared to control axons (J) (Upf2 cKO −FP axons (n=38 explants, 5 embryos), control −FP axons (n=30 explants, 4 embryos); Upf2 cKO +FP axons (n=41 explants, 5 embryos), control +FP axons (n=35 explants, 4 embryos). qRT-PCR for Robo3.2 mRNA in isolated −FP severed axons following FCM treatment (K) (74 control (n=3 embryos) and 77 Upf2 cKO (n=3 embryos) explants. Data are represented as mean +/− SEM, *p < 0.05, **p < 0.01 and ***p < 0.001 Scale bar, (B-D) 75 μm, (E, H) 100 μm. See also Figures S5 and S6.

Figure 6. NMD Regulates Postcrossing Axon Behavior

(A) Schematic of precrossing and postcrossing axon behavior. Axons were visualized by DiI at E13.5. Postcrossing axons were binned into three categories based on their distance from midline: 0-75 μm, 75-275 μm, and >275 μm. (B-D) Upf2 cKO axons exhibited normal precrossing behavior, but more lateral postcrossing trajectories than control axons (B). Many more Upf2 cKO axons are seen >275 μm from the midline compared to control axons (C) (459 axons, control (n=5) and 559 axons, Upf2 cKO (n=6) embryos). Boxed areas indicate examples of organized (control panel) or altered (mutant panel) trajectories (D). (E-G) Schematic of electroporation in spinal cord open book cultures (E). Electroporation of dominant-negative Upf1 in commissural neurons resulted in aberrant guidance similar to Upf2 cKO axons (F). Lateral distributions of postcrossing axons following dominant-negative Upf1 electroporation (G) (153 axons, control (n=4) and 131 axons, Upf2 cKO (n=3) embryos). Data are represented as mean +/− SEM. Scale bar, 200 μm (B), 75 μm (D, F).

Figure 7. NMD Machinery Is Localized to Growth Cones of Several Types of Neurons

(A -D) Upf2, Upf1 and Smg1 localize to axons and growth cones of peripheral (A) and central (B) nervous system neurons. Quantifications of the fluorescent intensities of NMD proteins in individual axons of dorsal root ganglia (C) and hippocampal (D) neurons. Data are represented as mean +/− SEM. Scale bar, 100 μm (A), 30 μm (B).