Subcellular second messenger networks drive distinct repellent-induced axon behaviors

Abstract

Second messengers, including cAMP, cGMP and Ca²⁺ are often placed in an integrating position to combine the extracellular cues that orient growing axons in the developing brain. This view suggests that axon repellents share the same set of cellular messenger signals and that axon attractants evoke opposite cAMP, cGMP and Ca²⁺ changes. Investigating the confinement of these second messengers in cellular nanodomains, we instead demonstrate that two repellent cues, ephrin-A5 and Slit1, induce spatially segregated signals. These guidance molecules activate subcellular-specific second messenger crosstalk, each signaling network controlling distinct axonal morphology changes in vitro and pathfinding decisions in vivo.

Fig. 1. Ephrin-A5 induces an elevation of cGMP and an increase in the frequency of Ca²⁺ transients in lipid rafts.

a Strategy to identify the cGMP source. The cGMP biosensor ^ThPDE5^VV is expressed in RGC axons alone (top left) or together with either the non-targeted cGMP scavenger SponGee (bottom left) or its lipid raft-targeted (Lyn-SponGee, top right) or -excluded (SponGee-Kras, bottom right) variants. When expressed alone, ^ThPDE5^VV monitors cGMP from the entire cytoplasm. SponGee prevents the sensor to report changes in cGMP concentration. Lyn-SponGee or SponGee-Kras lead ^ThPDE5^VV to monitor cGMP changes from all cellular compartments excluding the vicinity of either lipid rafts or the non-raft plasma membrane, respectively. b After ephrin-A5 exposure, ^ThPDE5^VV alone or co-expressed with SponGee-Kras monitors a cGMP elevation. By contrast, when co-expressed with SponGee or Lyn-SponGee, the FRET signal is not affected by ephrin-A5, similarly to vehicle (PBS)-exposed axons. A nitric oxide (NO)-induced cGMP elevation at the end of each recording ensures biosensor functionality and axon viability. Right traces: magnification of the dashed line-enclosed portion of the left traces. Image color-code: from low cGMP (blue) to high cGMP (red/white). Traces: mean ± s.e.m. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. *P < 0.05; **P < 0.01; Kruskal–Wallis test followed by Dunn’s post-hoc test. Scale bar, 10 µm. c Strategy to identify local changes in Ca²⁺ concentration. The Ca²⁺ sensor Twitch2b (top left), its lipid raft-targeted (top right) or -excluded (bottom) variants are expressed in RGCs. They report Ca²⁺ changes from all cytosolic sources, lipid rafts and the non-raft fraction of the plasma membrane, respectively. d An elevation in the Ca²⁺ transient frequency is detected by Twitch2b after ephrin-A5 but not after vehicle (PBS) exposure. This observation is reproduced with the lipid raft-targeted Twitch2b, but not when using its lipid raft-excluded equivalent. An ionomycin (iono)-induced Ca²⁺ elevation at the end of each recording ensures biosensor functionality and axon viability. Images illustrate a detected Ca²⁺ transient. Color-code: from low Ca²⁺ (blue) to high Ca²⁺ (red/white). Representative traces and individual data points (n indicated on the graphs) are shown. *** P < 0.001; two-tailed Wilcoxon test. Scale bar, 10 µm. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 2. Slit1 induces a cAMP reduction in the vicinity of the plasma membrane, outside lipid rafts.

a Strategy to identify local changes in cAMP concentration. The cAMP FRET sensor H147 (left), its lipid raft-targeted (middle) or -excluded (right) variants are expressed in RGCs. They report cAMP changes from anywhere in the cytosol, from lipid rafts and from the non-raft fraction of the plasma membrane, respectively. b A reduction in the cAMP concentration is detected by the biosensor H147 after Slit1 exposure but not after vehicle (PBS) addition to the culture medium. This observation is reproduced with the lipid raft-excluded H147 (H147-Kras), but not when using its lipid raft-targeted equivalent (Lyn-H147). A forskolin (Fsk) stimulation leading to a cAMP elevation was achieved at the end of each recording to verify the functionality of the biosensor and the viability of the axon. The portion of the left traces enclosed in the dashed rectangles is shown magnified in the right part of the panel. Images illustrating the change in the FRET ratio between before (−20 s) and after (+2 min) the PBS or Slit1 stimulation are color-coded from low blue to high red/white. Traces: mean ± s.e.m. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. *P < 0.05; ***P < 0.001; two-tailed Mann–Whitney test. Scale bar, 10 µm. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 3. Slit1 induces a cGMP elevation and an increase in the frequency of Ca²⁺ transients in the vicinity of the plasma membrane, outside lipid rafts.

a When retinal axons are exposed to Slit1, ^ThPDE5^VV alone or co-expressed with Lyn-SponGee monitors an elevation of cGMP. By contrast, when co-expressed with SponGee or SponGee-Kras, the FRET ratio is not affected by Slit1, similarly to axons that are not exposed to Slit1 and express ^ThPDE5^VV. A nitric oxide (NO) stimulation leading to a cGMP elevation was achieved at the end of each recording to verify the functionality of the biosensor and the viability of the axon. The portion of the left traces enclosed in the dashed rectangles is shown magnified in the right part of the panel. Traces: mean ± s.e.m. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. b An elevation in the frequency of Ca²⁺ transients is detected by Twitch2b after Slit1 exposure but not after vehicle (PBS) addition to the culture medium. This elevation is reproduced with the lipid raft-excluded Twitch2b, but not when using its lipid raft-excluded equivalent. An ionomycin (iono) stimulation leading to a Ca²⁺ elevation was achieved at the end of each recording to verify the functionality of the biosensor and the viability of the axon. Representative traces and individual data points are shown. The number of quantified axons is indicated on the graphs. a,b * P < 0.05; *** P < 0.001; a Kruskal–Wallis test followed by Dunn’s post-hoc test, b two-tailed paired Wilcoxon test. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 4. Lipid raft-restricted second messenger network downstream of ephrin-A5.

a The ephrin-A5-induced elevation of the cGMP concentration monitored using ^ThDPE5^VV is reduced by preventing local cAMP signals in lipid rafts (Lyn-cAMP Sponge). By contrast, reducing Ca²⁺ downstream signaling in this cellular domain does not affect cGMP changes. A nitric oxide (NO) stimulation was achieved at the end of each recording to verify the functionality of the biosensor and the viability of the axon. b The ephrinA5-induced elevation in the Ca²⁺ transient frequency monitored by Lyn-Twitch2b was prevented by scavenging cAMP in lipid rafts (Lyn-cAMP Sponge), whereas altering the downstream signaling of cGMP in this subcellular domain did not impact the ephrin-A5-induced Ca²⁺ transients. An ionomycin (iono) stimulation was achieved at the end of each recording to verify the functionality of the biosensor and the viability of the axon. c The lipid raft-restricted cAMP signals induced by ephrin-A5 were monitored using Lyn-H147. The reduction in the cAMP concentration was not affected by preventing the cGMP or Ca²⁺ downstream signaling in the same cellular compartment using Lyn-SponGee and Lyn-SpiCee, respectively. A forskolin (Fsk) stimulation was achieved at the end of each recording to verify the functionality of the biosensor and the viability of the axon. d Overall model of the lipid raft-restricted second messenger network downstream of ephrin-A5. Exposing growth cones to this axon guidance molecule leads to a combined modulation of cyclic nucleotides and Ca²⁺ that is restricted to the vicinity of lipid rafts. This network is characterized by a non-reciprocal influence of cAMP on cGMP and Ca²⁺ signaling. a,c The portion of the left traces enclosed in the dashed rectangle in is shown magnified in the right part of the panel. Traces: mean ± s.e.m. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. b Representative traces and individual data points are shown. The number of quantified axons is indicated on the graphs. a–c * P < 0.05; ** P < 0.01; *** P < 0.001; Kruskal–Wallis test followed by Dunn’s post-hoc test. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 5. Lipid raft-excluded second messenger network downstream of Slit1.

a The Slit1-induced elevation in cGMP imaged using ^ThDPE5^VV is prevented by the blockade of Ca²⁺ signaling next to the non-raft plasma membrane (SpiCee-Kras). By contrast, blocking cAMP signaling outside lipid rafts (cAMP Sponge-Kras) does not affect the Slit1-induced cGMP changes. A nitric oxide (NO) stimulation was achieved at the end of each recording to verify the biosensor functionality and the axon viability. b The elevation in the Ca²⁺ transient frequency induced by Slit1 outside lipid rafts and recorded using Twitch2b-Kras is prevented by scavenging either cAMP or cGMP outside lipid rafts (cAMP Sponge-Kras or SponGee-Kras, respectively). An ionomycin (iono) stimulation was achieved at the end of each recording to verify the biosensor functionality and the axon viability. c The juxta-membrane lipid raft-excluded cAMP signals induced by Slit1 were monitored using the biosensor H147-Kras. The reduction in the cAMP concentration is dampened by preventing cGMP downstream signaling in the same cellular compartment using SponGee-Kras, but not by reducing Ca²⁺ signaling with SpiCee-Kras. A forskolin (Fsk) stimulation was achieved at the end of each recording to verify the biosensor functionality and the axon viability. d Overall model of the juxta-membrane lipid raft-excluded second messenger network downstream of Slit1. Exposing growth cones to this axon guidance molecule leads to a combined modulation of cyclic nucleotides and Ca²⁺ that is restricted to the vicinity of the non-raft domain of the plasma membrane. This network is characterized by complex interactions including a cAMP influence on Ca²⁺, a control of cGMP elevation by Ca²⁺ transients and cGMP influencing both cAMP and Ca²⁺ signals. a,c The portion of the left traces enclosed in the dashed rectangle in is shown magnified in the right part of the panel. Traces: mean ± s.e.m. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. b Representative traces and individual data points are shown. The number of quantified axons is indicated on the graphs. a–c * P < 0.05; ** P < 0.01; *** P < 0.001; Kruskal–Wallis test followed by Dunn’s post-hoc test. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 6. Lipid raft-specific and -excluded scavenging of second messengers prevent the collapse of growth cones induced by ephrin-A5 and Slit−1, respectively.

a Ephrin-A5 induces the collapse of mRFP- and SponGee-Kras-expressing axons, whereas the non-targeted and lipid-raft targeted variants of SponGee (Lyn-SponGee) prevent growth cone collapse. b When lacking a targeting sequence or when restricted to lipid rafts, SpiCee prevents the ephrin-A5-induced growth cone collapse, in contrast to the lipid raft-excluded variant of SpiCee (SpiCee-Kras). c Slit−1 induces the collapse of mRFP-expressing retinal growth cones. The proportion of collapsing axons is not affected by the expression of Lyn-SponGee but is reduced by SponGee-Kras or the cytosolic SponGee. d The collapse of retinal growth cones exposed to Slit−1 is reduced by the expression of SpiCee (not targeted) or its lipid raft-excluded variant (SpiCee-Kras), but not by Lyn-SpiCee. e Slit−1-induced growth cone collapse is prevented by cAMP Sponge or cAMP Sponge-Kras, but not by Lyn-cAMP-Sponge. Axons were immunolabeled with a βIII-tubulin and a Ds-Red (mRFP) antibody. The latter reports the expression of SponGee (a,c), SpiCee (b,d) or cAMP Sponge (e). Scale bars, 10 µm. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. *** P < 0.001; One way ANOVA followed by Dunnett post-hoc test. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 7. Ephrin-A5 and Slit1 induce distinct morphological changes of axonal growth cones in vitro.

The growth of axons exposed to PBS is not affected (top row). Ephrin-A5 induces a growth cone collapse followed by a prompt retraction (middle row). Axons exposed to Slit1 exhibit a collapse of the growth cones but in contrast to axons encountering ephrin-A5, do not retract within the 20 min recorded (bottom row). Traces: mean ± s.e.m. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. *** P < 0.001; Kruskal–Wallis test followed by Dunn’s post-hoc test. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 8. Imposing lipid raft-restricted and -excluded cyclic nucleotides modulation mimics the axon behavior induced by ephrin-A5 and slit1, respectively.

a Lyn-bPAC-expressing growth cones retract when exposed to successive pulse of blue light whereas control axons are insensitive to this light stimulation. By contrast, bPAC-Kras-expressing growth cones collapse but do not retract. The blue line denotes the time of light exposure both in the image sequences and in the traces. b When exposed to blue light, Lyn-BeCyclOp induces the collapse of a limited fraction of growth cone. Light activation of BeCyclop-Kras also leads to growth cone collapse, but in a large majority of axons. Lyn-BeCyclOp-expressing collapsing growth cones retract more than the axons of BeCyclOp-Kras-electroporated neurons. a,b Traces: mean ± s.e.m. Box-and-whisker plot elements: median, upper and lower quartiles, 10th and 90th percentiles. * P < 0.05, ** P < 0.01, *** P < 0.001. Top graphs χ² test followed by χ² post-hoc tests; bottom graphs, One way ANOVA followed by Dunnett post-hoc test. Source data, number of replicates and P values are provided as a Source Data file.

Fig. 9. Lipid raft-targeted and -excluded scavenging lead to misguided retinal axons in the SC and at the optic chiasm, respectively.

a Lyn-SpiCee- and Lyn-SponGee-expression lead to overshooting axons in the inferior colliculus at P15, by contrast to mRFP-, SpiCee-Kras- and SponGee-Kras-expression. The orange dashed line highlights the position of the posterior end of the superior colliculus (SC). The inferior colliculus (IC) is above this line. The top row images are magnifications of the regions of the bottom row images indicated by the black dashed squares. Scale bars: top row, 250 µm; bottom row, 500 µm. b SpiCee-Kras- and SponGee-Kras-expressing axons (cyan) exit the optic chiasm labeled with TAG1 (magenta), by contrast to the axons of mRFP-, Lyn-SpiCee and Lyn-SponGee-electroporated RGCs. An excess of retino-retinal axons is also detected in SpiCee-Kras- and SponGee-Kras-electroporated animals, as compared to mRFP-, Lyn-SpiCee- and Lyn-SponGee-electroporated RGCs. The top row highlights the mRFP channels in which electroporated axons are seen. Closed arrowheads, axons exiting the optic chiasm; open arrowheads, retino-retinal axons. Scale bar, 200 µm. a,b ** P < 0.01; *** P < 0.001; χ² test followed by χ² post-hoc tests. Source data are provided as a Source data, number of replicates and P values are provided as a Source Data file.