Spatial and temporal second messenger codes for turning

Xavier Nicol1,2, Kwan Pyo Hong, and Nicholas C. Spitzer

Neurobiology Section, Division of Biological Sciences, Kavli Institute for Brain and Mind, University of California at San Diego, La Jolla, CA 92093

Edited* by Lynn T. Landmesser, Case Western Reserve University, Cleveland, OH, and approved July 5, 2011 (received for review January 6, 2011)

Cyclic AMP (cAMP) and calcium are ubiquitous, interdependent se- generated in growth cones in response to Netrin-1, the principal cond messengers that regulate a wide range of cellular processes. guidance molecule required for attraction of commissural During development of neuronal networks they are critical for the by the spinal floor plate. We report a major difference in first step of circuit formation, transducing signals required for axon cAMP/calcium interplay between filopodia and the growth cone fi pathfinding. Surprisingly, the spatial and temporal cAMP and cal- center. Optogenetic elevation of cAMP in lopodia but not in fi cium codes used by molecules are unknown. Here, growth cone centers drives growth cone attraction. Both lopo- we identify characteristics of cAMP and calcium transients gener- dial and growth cone center signaling pathways are mediated by ated in growth cones during Netrin-1–dependent axon guidance. In activation of the same Netrin-1 , Deleted in Colorectal (DCC). We show that cAMP dynamics are also essential filopodia, Netrin-1–dependent Deleted in Colorectal Cancer (DCC) for midline crossing of commissural axons in vivo. receptor activation induces a transient increase in cAMP that causes a brief increase in calcium transient frequency. In contrast, activa- Results tion of DCC in growth cone centers leads to a transient calcium- We identified spatial and temporal second-messenger codes that dependent cAMP increase and a sustained increase in frequency of fi regulate axon guidance using dissociated cell cultures of Xenopus calcium transients. We show that lopodial cAMP transients regu- spinal neurons. Application of a local Netrin-1 gradient to growth fi late spinal axon guidance in vitro and commissural axon path nd- cones of these cultured neurons attracts axons in a cAMP- and ing in vivo. These growth cone codes provide a basis for selective calcium-dependent manner that requires the DCC Netrin-1 re- activation of specific downstream effectors. ceptor (8, 16) mediating both attraction and repulsion in Xenopus spinal neurons (22). In contrast to previous studies (16, 23), we cyclic AMP dynamics | calcium dynamics | compartmentalization | did not include serum in the culture medium because it blocked photoactivated adenylyl cyclase alpha | cyclic AMP oscillations some of the growth cone calcium signals observed in vivo (Fig. S1 A and B). The response to Netrin-1 and its dependence on cAMP yclic AMP (cAMP) is a major cellular second messenger that and calcium were robust: Netrin-1 is an attractant in the control Cactivates and integrates multiple intracellular signaling path- condition, converted into a repellent when adenylyl cyclase is ways. Microdomains of cAMP are generated in cardiac myocytes inhibited, and does not affect the direction of axon growth in the (1), regional domains of cAMP are present in neurons (2, 3), and absence of extracellular calcium (Fig. S1C). cAMP diffusion is restricted by rapid degradation by phospho- We used a plasma membrane-targeted Epac2-camps (pmEpac2- fl diesterases (4) that limit the duration of cAMP signaling (5, 6). camps) FRET probe (24) to monitor uctuations in cAMP con- However, little is known about spatial compartmentalization and centration. Growth cones expressing pmEpac2-camps responded temporal dynamics of cAMP in neuronal growth cones, despite to bath application of forskolin, an activator of transmembrane the importance of cAMP for axon pathfinding in response to adenylyl cyclases, with an increase in the CFP:YFP ratio (Fig. S2 a wide range of molecular guidance cues, including Netrin-1 (7– and Movie S1). Although we did not calibrate the sensor, we 11), (12), Slits, and ephrins (13). demonstrated that saturation was not reached with the resting Calcium is another ubiquitous second messenger involved in concentration of cAMP in growth cones (Fig S2). Targeting axon pathfinding, mediating responses to axon guidance mole- pmEpac2-camps to the plasma membrane reduces differences in fluorescence intensity between the center of the growth cone and cules. In addition to axon steering, calcium modulates axon out- fi growth and retraction (14, 15). A sustained gradient of calcium lopodia, and allows monitoring of cAMP in both compartments across the growth cone is thought to be generated by asymmetric (Fig. S3). However, measurement of cAMP levels with this Epac- activation of axon guidance receptors, and to be the relevant based sensor could miss detection of cAMP levels that would be calcium signal for axon pathfinding (16). Spontaneous fast and revealed with a PKA-based sensor. spatially restricted filopodial calcium transients are also sufficient to steer axons (17, 18). The frequency of slower transients in the Dynamics of cAMP and Calcium Signals in Filopodia. When growth cones were stimulated by Netrin-1, cAMP increased transiently in entire growth cone controls the rate of axon extension in vivo and fi ∼ ∼ in vitro (19). The regulation of these transients by axon guidance lopodia 1 min after stimulation, reaching a maximum by 4 min (Fig. 1A). Although stimulation was maintained, the cAMP cues has not been investigated. ∼ Filopodia are critical for axon pathfinding (20), and clues to level returned to baseline 10 min after the onset of Netrin-1 their operation are provided by subcellular localization of signal- application. This effect was not due to desensitization of the ing components. Filopodial enrichment of regulatory subunit II of probe, because pmEpac2-camps reported a sustained increase in protein kinase A, a major effector of cAMP, is required for growth cAMP concentration (Fig. S2). The cAMP response to Netrin-1 cone attraction mediated by intracellular gradients of cAMP or PACAP (21). Modulation of spontaneous calcium transients in filopodia regulates axon turning (17). Temporal regulation of Author contributions: X.N. and N.C.S. designed research; X.N. and K.P.H. performed re- cAMP synthesis is also important for axon responses to ephrin-As search; X.N. analyzed data; and X.N. and N.C.S. wrote the paper. (14), and the frequency of calcium transients in the growth cone The authors declare no conflict of interest. center that includes the lamellipodium regulates the rate of axon *This Direct Submission article had a prearranged editor. extension (19). However, little is known about regulation of filo- 1Present address: Institut National de la Santé et de la Recherche Médicale, U839, Uni- podial second messenger signals and their temporal dynamics by versité Paris 6, Institut du Fer à Moulin, 75005 Paris, France. diffusible axon guidance cues. 2To whom correspondence should be addressed. E-mail: [email protected]. We have investigated the subcellular localization, temporal This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. dynamics, and local interactions of cAMP and calcium signals 1073/pnas.1100247108/-/DCSupplemental.

13776–13781 | PNAS | August 16, 2011 | vol. 108 | no. 33 www.pnas.org/cgi/doi/10.1073/pnas.1100247108 Downloaded by guest on September 28, 2021 ) Netrin-1

was blocked when it was replaced with a vehicle or when cultures -1 A 12.5 C 50% were preincubated in a solution containing a function-blocking ΔF/F 2 against the DCC Netrin-1 receptor (Fig. 1 A and B). 10.5 10 s The frequency of fast filopodial calcium transients (17) followed 1 the same dynamics as the cAMP transient (Figs. 1 C and D and 8.5 control 2A, Fig. S4,andMovie S2). Starting from a baseline (6.8 ± 0.7 Netrin-1 −1 0 6.5 min ) before Netrin-1 stimulation, the frequency began rising ∼2 Δ R/R (%) cAMP Netrin-1 Filopodial calcium min after the initiation of stimulation and achieved its maximum SQ 22536 Netrin-1 −1 4.5 transient frequency (min (10.9 ± 1.8 min ) ∼4 min after the onset of Netrin-1 application; -5 0 5 10 15 20 it returned to baseline after ∼14 min (Fig. 2A and Movie S2). time (min)

Stimulation with vehicle or exposure to anti-DCC prevented this forskolin fi C D time (min) increase in lopodial calcium transient frequency (Fig. 1 and ). 1% Netrin-1 -0.5 0 +4 +4.5+14 +14.5 Ratiometric calcium imaging that avoids artifacts due to volume B ΔR/R

fi – 2+ 5 min

changes con rmed the Netrin-1 induced increase in frequency of ) fi -1 before lopodial calcium transients. The increase in frequency was D E Netrin-1 fi ** after greater in proximal than in distal lopodia (Fig. S4). 0 mM Ca Netrin-1 16 ** 2 * 2 mM Ca2+ 1.5 cAMP Drives Generation of Filopodial Calcium Transients. Because 0 mM Ca2+ cAMP and calcium regulate each other and modulate axon turning 1.5 * 12 1 (25), we investigated the interplay between cAMP and calcium 1 during Netrin-1–induced signaling in filopodia. To determine 8 fi 0.5 0.5 whether cAMP regulates production of lopodial calcium tran- 4 transient frequency Spontaneous filopodial ***

sients, we suppressed its synthesis with SQ22536, a blocker of Δ R/R (%) cAMP 0

transmembrane adenylyl cyclases, and stimulated its synthesis with 0 Normalized filopodial calcium 0 calcium transient frequency (min -0.5 before SQ 22536 - + - SQ 22536 - + - forskolin. SQ22536 reduced and forskolin increased the frequency 2-6 min 10-14 min of spontaneous filopodial calcium transients (Fig. 2 C and D). In stimulation forskolin - - + forskolin --+ – contrast, both treatments precluded the Netrin-1 induced in- Fig. 2. The Netrin-1–induced elevation of cAMP drives generation of filo- crease in frequency (Fig. 2 C and E). Blocking transmembrane podial calcium transients. (A) The increase in calcium transient frequency adenylyl cyclases prevents the Netrin-1–dependent increase in (dashed trace) follows the increase in cAMP (solid trace). (B) Calcium-free calcium transient frequency, whereas constantly stimulating them medium does not affect the transient Netrin-1–induced elevation of cAMP. likely elevates cAMP to a level that induces the maximal calcium (A and B) Average of 10 or more growth cones. Dashed line as in Fig. 1A.(C transient frequency so that further stimulation by Netrin-1 is not and D) Spontaneous filopodial calcium transients occur less frequently in SQ22536-treated growth cones (red trace) and more often in forskolin- treated neurons (green trace). (C) Dashed lines as in Fig. 1C.(E) Application of the transmembrane adenylyl cyclase inhibitor SQ22536 or forskolin pre- A cAMP C calcium vents the increase in frequency of calcium transients induced by Netrin-1 1% Netrin-1 50% Netrin-1 (blue trace). (D and E) n ≥ 60 filopodia. Error bars, SEM; *P < 0.05, **P < 0.01, ΔR/R ΔF/F NEUROSCIENCE 2+ ***P < 0.001. (B and D) Paired Student t test. (E) ANOVA. 2+ 5 min 10 s 2 mM Ca 2 mM Ca

2+ – vehicle vehicle possible. These experiments indicate that Netrin-1 dependent 2+ stimulation of filopodial calcium transients is driven by transient elevation of cAMP. We monitored cAMP in Netrin-1–stimulated 2 mM Ca

2 mM Ca Netrin-1 2+ growth cones in calcium-free medium to determine whether cal- 2+ Netrin-1 cium influx regulates cAMP synthesis. The absence of calcium did –

+ α DCC time (min) not affect the cAMP response, suggesting that the Netrin-1 2 mM Ca + α DCC 2 mM Ca -0.5 0 +4 +4.5+14 +14.5 dependent increase in cAMP concentration is independent of extracellular calcium (Fig. 2B). A role for calcium release from ** intracellular stores is not excluded. BD2 * 2 mM Ca2+ before stimulation 1.5 vehicle after stimulation 1.5 αDCC cAMP and Calcium Dynamics in Growth Cone Centers. We next fo- 1 1 cused on the dynamics of signaling events occurring in the center fi 0.5 0.5 of the growth cone, excluding lopodia. Application of Netrin-1 to 0 growth cones elicited an increase in cAMP concentration in cAMP Δ R/R (%) cAMP 0 growth cone centers with a time course delayed and prolonged

-0.5 before transient frequency 2-6 min 10-14 min fi ∼ stimulation Netrin-1 - + + compared with that in lopodia, starting to rise 2 min after α ∼ Normalized filopodial calcium DCC --+ stimulation and reaching its maximum by 5 min before decaying (Fig. 3 A and B). This increase was blocked by replacement of Fig. 1. In filopodia, Netrin-1 induces a brief cAMP transient and a brief Netrin-1 with vehicle or with the DCC function-blocking antibody. increase in frequency of calcium transients. (A and B) The cAMP concentra- fi The frequency of calcium transients in the growth cone center tion increases transiently in lopodia after growth cone stimulation with increased in a sustained manner upon stimulation with Netrin-1 Netrin-1 (blue trace), whereas no signal is detected when Netrin-1 stimula- (Movie S3), but not in response to vehicle, and was fully sup- tion is replaced by application of vehicle (gray trace) or when neurons are pressed by a higher concentration of anti-DCC (Fig. 3 C and D). preincubated with a DCC function-blocking antibody (1 μg/mL, orange trace). Average of 10 or more growth cones. Signal is considered positive when its amplitude exceeds twice the SD of the baseline for 5 min before Calcium Drives Elevation of cAMP in Growth Cone Centers. Growing neurons in calcium-free culture medium blocked both the stimulation (dashed lines). (C and D) Netrin-1 induces a transient increase in – frequency of filopodial calcium transients (blue trace) that is absent when Netrin-1 induced increase in cAMP concentration (Fig. 4A) and Netrin-1 is replaced by a vehicle (gray trace) or in the presence of the anti- the calcium transients (26). In contrast, neither blockade of DCC function-blocking antibody (1 μg/mL, orange trace). Signals are con- cAMP synthesis by SQ22536 nor forskolin-induced increase of

sidered positive when they exceed 20% ΔF/F0 (dashed lines; greater than cAMP concentration affected the frequency of calcium tran- twice the SD of the baseline). n ≥ 60 filopodia. Error bars, SEM; *P < 0.05, sients in the growth cone center in response to Netrin-1 (Fig. 4 B **P < 0.01. (B and D) Paired Student t test. and C). Thus, the elevation of cAMP is downstream of calcium,

Nicol et al. PNAS | August 16, 2011 | vol. 108 | no. 33 | 13777 Downloaded by guest on September 28, 2021 1 A C A 2+ cAMP calcium * 2 mM Ca 100% 0.5% Netrin-1 0.5 0 mM Ca2+ 0.5% 2+ ΔF/F Netrin-1 ΔR/R

Netrin-1 2+ ΔR/R 5 min 5 min 0 5 min 2+ 2 mM Ca before cAMP Δ R/R (%) -0.5 2-6 min 10-14 min

vehicle 0 mM Ca 2+ stimulation 2 mM Ca )

-1 * vehicle 100% Netrin-1 * *

2 mM Ca B C ΔF/F 40 2+ 2+ Netrin-1 10 min 30 control Netrin-1 before 1 μ g/ml + α DCC 2 mM Ca 2 mM Ca 20 Netrin-1

Netrin-1 2+ Netrin-1 after 2+

SQ 22536 10 Netrin-1 Netrin-1 4 μ g/ml Growth cone calcium + α DCC transient frequency (h 2 mM Ca 0 + α DCC

2 mM Ca SQ 22536 - + - forkolin - - + forskolin ) -1 before Netrin-1 B D 40 * after Netrin-1

1 2 mM Ca2+ 30 ** D brief increase cAMP * vehicle Netrin-1 DCC in Ca2+ transient αDCC transient 0.5 20 frequency

10 Filopodia 0 2+

Growth cone calcium 0 Ca cAMP transient transient frequency (h cAMP Δ R/R (%) -0.5 before 2-6 min 10-14 min Netrin-1 + - + + Netrin-1 DCC stimulation αDCC - - 1 μg/ml 4 μg/ml sustained increase in center Ca2+ transient frequency

Fig. 3. In the growth cone center, Netrin-1 induces a longer cAMP transient cone Growth and a sustained increase in frequency of calcium transients that are larger Fig. 4. The Netrin-1–induced calcium drives elevation of cAMP in the center and longer in duration. (A and B) The cAMP concentration increases tran- of the growth cone. (A) Calcium-free culture medium blocks the transient siently in the growth cone center after Netrin-1 stimulation (blue trace), elevation of cAMP in growth cone centers. Average of 10 or more growth whereas no signal is detected when Netrin-1 stimulation is replaced by ap- cones. Dashed line as in Fig. 1A.(B and C) Application of the transmembrane plication of culture medium (gray trace) or after incubation with a function- adenylyl cyclase inhibitor SQ22536 (red trace) or forskolin (green trace) does μ blocking antibody to DCC (1 g/mL, orange trace). Average of 10 or more not affect the Netrin-1–dependent elevation of calcium transient frequency. growth cones. Dashed line as in Fig. 1A.(C and D) Netrin-1 stimulation (B) Dashed lines as in Fig. 1C. Error bars, SEM; *P < 0.05. (A) Paired Student t induces a sustained increase in frequency of calcium transients in growth test. (C) Wilcoxon test. (D) Summary of signaling pathways. In filopodia, cone centers (blue trace). The spontaneous calcium transient frequency is Netrin-1 drives a transient increase of cAMP concentration through its DCC not affected by application of vehicle to growth cones (gray trace). The receptor; cAMP elicits a brief rise in frequency of filopodial calcium tran- – Netrin-1 induced increase in frequency is blocked after preincubation with 4 sients. In growth cone centers, Netrin-1, through the DCC receptor, drives μ ≥ g/mL of anti-DCC antibody (brown trace). n 5 growth cones. (A and C) a transient calcium-dependent increase in cAMP and a sustained cAMP- < < Dashed lines as in Fig. 1 A and C. Error bars, SEM; *P 0.05, **P 0.01. (B) independent increase in the frequency of calcium transients. Paired Student t test. (D) Wilcoxon test.

contrast, asymmetric elevations of cAMP in the growth cone center and the order of cAMP and calcium signaling is inverted be- are not sufficient to induce growth cone turning. Overlap of exci- tween the filopodia and growth cone center (Fig. 4D). tation wavelengths of pmEpac2-camps and PACα prevented comparison of the amplitude of PACα-induced cAMP transient cAMP Transients in Filopodia Steer Axons. To investigate further the with that elicited by Netrin-1. spatial code by which cAMP regulates axon guidance, we used mCherry-tagged photoactivated adenylyl cyclase (27) (mCherry- cAMP Transient Frequency Codes Growth Cone Responsiveness. We PACα; Fig. S5A and Movie S4) to generate localized elevations of examined temporal coding of axon turning by cAMP by exposing cAMP mimicking the duration of those stimulated by Netrin-1 mCherry-PACα–expressing growth cones to pulses of blue light at − (27). Exposure of quadrants of growth cones expressing mCherry- frequencies of 0–12 h 1. These growth cones did not turn spon- PACα to blue light for 3 min every 20 min induced turning of axons taneously in the absence of stimulation. Increasing the frequency toward the stimulated side (Fig. 5 A, E,andF), mimicking the of cAMP transients led to an increase in the percentage of turning effect of Netrin-1 in similar culture conditions (8, 16, 22, 23) (Fig. axons, and the maximum response was achieved with a frequency S1) and demonstrating that asymmetrical and pulsatile elevations of three transients per hour, similar to that of spontaneous tran- of cAMP are sufficient for axon turning. The same stimulation sients observed in cell bodies (5) (Fig. 6). Neither the turning protocol applied to mCherry-expressing growth cones had no ef- success rate (percentage of axons turning more than 7.5°) nor the fect on the direction of axon outgrowth (Fig. 5 B, E,andF). We turning angle was affected by a further increase in frequency. next determined the contributions of cAMP elevations in filopodia Moreover, the frequency of cAMP transients did not affect the and growth cone centers. Mimicking the Netrin-1–driven cAMP turning angle of individual, responsive axons (Fig. S6), suggesting increase in filopodia on one side of the growth cone induced axon that the frequency of cAMP transients codes for an all-or-none turning toward the stimulated side (Fig. 5 C, E,andF). In contrast, switch, inducing growth cones to turn. The distinct thresholds for increasing cAMP in filopodia on one side of growth cones in cul- individual growth cones may reflect cell type-specific sensitivity for ture medium lacking calcium did not induce axon turning (Fig. 5 E axon guidance cues. and F), in agreement with the dependence of filopodial calcium transients on cAMP signals (Fig. 4D). Stimulation of one side of Sustained Changes in cAMP Levels Disrupt Commissural Axon the growth cone center in the presence of extracellular calcium also Pathfinding in Vivo. Netrin-1 is the major attractive axon guid- failed to induce turning (Fig. 5 D–F). Local stimulation of PACα ance molecule required for commissure formation in a wide range produces an asymmetric and transient elevation of cAMP (Fig. S5 of animal species, including Caenorhabditis elegans, Drosophila B and C), likely because of hydrolysis by phosphodiesterases (4). melanogaster, and mammals (28–30). Calcium signaling is required Thus, filopodial cAMP signals that mimic the duration of those for proper ventral midline crossing by commissural axons in the elicited by Netrin-1 are sufficient to steer growth cone turning. In Xenopus spinal cord (31), but the role of cAMP in their guidance

13778 | www.pnas.org/cgi/doi/10.1073/pnas.1100247108 Nicol et al. Downloaded by guest on September 28, 2021 α 60 α 60 40 A -15 s PAC +80 min C -15 s PAC +80 min E * * 30 ( μ m) ( μ m) distance distance 40 40 20 10

turning angle (°) 0 20 20 α α α α Fliopodia only -10 PACmcherryPAC PAC PAC filopodia ++-+ + growth cone +-++ - μ -20 0 20 center 10 m -20 0 20 μ Filopodia + growth cone center distance (μm) distance ( m) calcium + + ++ - -15 s mCherry +80 min 60 α 60 B D -15 s PAC +80 min F 100 ( μ m) ( μ m) distance

distance 80 40 40 filopodia + growth 60 cone center

cumulative filopodia only distribution (%) 40 growth cone 20 20 center only filopodia only 20 0 mM Ca2+ mCherry Growth cone center only -20 0 20 -20 0 20 -25 0 25 50 75

Filopodia + growth cone center distance (μm) distance (μm) turning angle (°)

Fig. 5. Optogenetically generated cAMP transients in filopodia but not growth cone centers encode growth cone turning in vitro. (A) A brief cAMP elevation induced by blue light illumination of one side of a PACα-expressing growth cone is sufficient to induce a change in direction of axon outgrowth (Left). Tra- jectories of PACα-expressing axons stimulated for 3 min three times per hour in filopodia and growth cone center (Right). The initial orientation of the axons (before stimulation) is aligned with the y axis and superimposed on the same origin. (B) Illumination of filopodia and the growth cone center expressing mCherry does not induce growth cone turning (Left). Trajectories from mCherry-expressing growth cones are evenly distributed around the initial orientation of the axon (Right). (C) Stimulation of PACα in filopodia alone is sufficient to induce axon turning toward the stimulated side of the growth cone. (D) Elevation of cAMP in the growth cone center does not stimulate axon turning. (E) Average turning angle and (F) cumulative distribution of angles of axons expressing PACα or mCherry and illuminated on filopodia, growth cone center, or in both. (E and F) n ≥ 8 growth cones. Error bars, SEM; *P < 0.05, Kruskal–Wallis test.

has been contentious. In vitro, cAMP induces a switch in the re- Discussion sponse to Netrin-1 (8), but genetic deletion of soluble adenylyl Our experiments identify spatial and temporal second-messenger cyclase fails to produce a defect in guidance of spinal commissural codes regulating signaling in growth cones in response to a guidance neurons in mice (32). We imaged growth of Xenopus laevis spinal molecule. Use of an Epac-based FRET probe (24) enables de- commissural axons to determine whether cAMP signaling regu- tection of small signal-averaged cAMP transients in growth cones. lates axon guidance in vivo (Fig. S7A). In control conditions, Netrin-1 elicits brief, extracellular calcium-independent elevation NEUROSCIENCE commissural axons grow to the midline and cross it before turning of cAMP in filopodia that stimulates increases in filopodial calcium anteriorly or posteriorly (31) (Fig. 7 A and D and Fig. S7B). De- transients. Netrin-1 also stimulates brief calcium-dependent ele- creasing cAMP levels with SQ22536 prevents axons from crossing vation of cAMP and prominent calcium transients in growth cone the midline. The majority of axons join the floor plate before centers. Thus, the relation between cAMP and calcium signaling turning and growing longitudinally (Fig. 7 B and D and Fig. S7B), depends on the domain of the growth cone. cAMP transients im- indicating that cAMP signaling is required for normal commissural posed in filopodia control the turning response by regulating the axon guidance. A sustained increase in cAMP concentration generated by application of forskolin leads to a similar defect (Fig. 7 C and D and Fig. S7B). Analysis of growth cone velocity indicates that cAMP signaling is required for commissural growth cone A -15 s +80 min

pathfinding, with only a modest effect on axon outgrowth (Fig. 7 E -1

and F). Elimination of cAMP modulation both with SQ22536 and 12 h forskolin suggests that appropriate guidance of spinal commissural α stimulation 10 μm axons requires critical cAMP dynamics. PAC

Local Pulses of cAMP Regulate Axon Pathfinding in Vivo. To de- * 100 B * C 100 25 termine whether the spatial localization and temporal dynamics of * 80 cAMP identified in vitro regulate axon pathfinding in vivo, we 80 20 α 60 expressed PAC in half of an X. laevis embryo and imaged spinal 60 15

commissural axons approaching the ventral midline (Fig. S7A). cumulative 40

40 10 distribution (%) control Control axons grow perpendicular to the midline (Fig. 8 A and D). 0.75 h-1 -1 turning angle (°) turning axons (%) 20 5 20 1.5 h Blockade of Netrin-1 signaling using the DCC function-blocking 3 h-1 -1 antibody causes axons to turn away from the perpendicular to the 0 0 12 h 012 3 12 -25 0 25 50 midline, likely because of the response to other axon guidance frequency (h-1) -5 turning angle (°) cues not present in culture (Fig. 8 B and D). Local and pulsatile blue light illumination of the distal part of PACα-expressing Fig. 6. Temporal coding of turning by optogenetically generated cAMP −1 growth cones, in an area equivalent to filopodia location in vitro, is transients. (A) High-frequency cAMP elevations (12 h ) induced by blue α fi sufficient to maintain the orientation of axon outgrowth toward light illumination of one side of a PAC -expressing growth cone are suf - cient to induce a change in direction of axon outgrowth. (B) Turning success the midline despite blockade of DCC (Fig. 8 C and D). Thus, re- rate (percentage of axons turning more than 7.5°) and turning angle after − stricted and transient elevations of cAMP can replace Netrin-1 cAMP pulses at frequencies of 0–12 h 1. n ≥ 7 growth cones. Error bars, SEM; signaling to maintain commissural axon growth directionality *P < 0.05, Kruskal–Wallis test. (C) Cumulative distribution of angles of PACα- in vivo, using the cAMP codes described in vitro. expressing axons illuminated at frequencies of 0–12 h−1.

Nicol et al. PNAS | August 16, 2011 | vol. 108 | no. 33 | 13779 Downloaded by guest on September 28, 2021 A 0h 1h30 3h 110 A 0h 1h20 B 0h 1h20

55 ( μ m) distance

-150 -50 50 150 control α DCC μ control -55 distance ( m)

50 µm -110 50 µm B 110 0h 1h20

55 ( μ m) C D 90 ** * distance distance (μm) 60 -150 -50 50 150 30 -55 angle (°) SQ 22536 0 -110 αDCC - + + α α DCC + PAC 110 PACα C stimulation - - +

55 ( μ m) distance distance (μm) Fig. 8. Local and pulsatile optogenetic elevation of cAMP concentration rescues defects of commissural axon pathfinding in vivo. (A) Before crossing -150 -50 50 150 the ventral midline, Alexa 488 dextran-filled commissural axons grow per- forskolin -55 pendicular to the midline (dashed line at the Alexa 568 border). Arrowheads

-110 identify growth cones. (B) Application of a blocking antibody against DCC prevents axons from growing perpendicular to the midline. Blue light illu- mination of growth cones that do not express PACα (blue areas; uninjected before midline at midline side of the animal, not labeled with Alexa 568) does not rescue this defect. D E control F control 60 *** SQ 22536 SQ 22536 (Inset) Outlined arrowhead indicates enlarged growth cone. (C) Local and ) control ) 60 *** 60 -1 -1 forskolin forskolin α SQ 22536 * *** pulsatile illumination (blue areas; 3 min three times per hour) of PAC - 30 forskolin 40 40 *** expressing growth cones (PACα and Alexa 568-coinjected side of the animal) μ m) distance (μm) maintain the orientation of axon growth toward the midline. (D) Mean 0 60 120 20 20 angle of axon trajectory deviation from the perpendicular to the midline. velocity ( μ m.h velocity ( μ m.h Twelve or more trajectories were scored for each condition. Error bars, SEM;

-30distance ( 0 0 **P < 0.01, *P < 0.05, ANOVA.

-60 transverse transverse velocitylongitudinalvelocity velocitylongitudinalvelocity to distinct microdomains and can account for compartmentaliza- Fig. 7. Disruption of cAMP signaling prevents midline crossing of commis- tion of cAMP signals (34). Different adenylyl cyclases can account sural axons in vivo. (A) Alexa 488 dextran-filled commissural axons imaged in for distinct signals: the genetic deletion of soluble adenylyl cyclase the intact spinal cord of control embryos grow to the ventral midline (dashed does not affect commissural axon trajectories (32) although it is line at the Alexa 568 border), cross it, and turn rostrally or caudally along the required for Netrin-1–dependent axon outgrowth (11), but alter- contralateral ventral fascicle (Left). Trajectories of control growth cones ation of transmembrane adenylyl cyclase activity induces the fl (Right). Orange horizontal bar represents the oor plate. (B) Commissural defects in spinal commissural axon guidance reported here. axons in a SQ 22536-treated spinal cord grow to the ventral midline but fail to Uneven distribution of cAMP- or calcium-responsive elements cross it. (C) Treating the spinal cord with forskolin also prevents axons from is also likely to contribute to compartmentalization of downstream crossing the midline. (A–C) n > 25 growth cones. All trajectories are shown in Fig. S7B.(D) Trajectories of individual axons were aligned to obtain an av- second-messenger signaling. PKA and Epac, two of the main erage trajectory for each experimental condition, and rostrally and caudally cAMP effectors, regulate distinct signaling pathways during axon guidance (7) and are confined to specific subcellular compart- turning trajectories were superimposed. Averaged trajectories demonstrate fi a failure of midline crossing when spinal cords are treated with SQ22536 or ments (35). The difference in cAMP af nity of PKA and members forskolin. More than 25 trajectories were scored for each condition. (E)To of the Epac family further distinguishes these cAMP-dependent distinguish effects on axon directionality and growth, we quantified the signaling pathways (7, 36). It is still unclear whether the FRET transverse velocity (toward the midline) and the longitudinal velocity (parallel probe we used reflects cAMP signaling independently of down- to the midline) of axons from spinal cords exposed to control, SQ22536-, or stream effectors or is specific to one of the cAMP pathways, likely forskolin-containing medium. The transverse velocity was reduced by high or an Epac-dependent pathway, because the sensor is based on low cAMP concentration before the midline and (F) further suppressed within Epac2. A PKA-based sensor might reveal a different distribution a 40-μm–wide area around the ventral midline. In contrast, the longitudinal of signals in growth cones in response to Netrin-1. velocity was only slightly affected by modifying the cAMP concentration. We show that the cAMP and calcium codes used to induce (E and F) Error bars, SEM; ***P < 0.001, *P < 0.05, ANOVA. growth cone steering are based on transient signals of these second messengers. Netrin-1–induced calcium transients have not been reported. This discrepancy may have resulted from the use of se- frequency of filopodial calcium transients. Previous studies of axon rum-containing culture medium that blocks calcium transients in turning initiated by calcium signals in lamellipodia may have stim- growth cone centers observed in vivo (19) (Fig. S1A) and from the ulated calcium signals in filopodia as well (16, 25). In contrast, high image acquisition rate required to detect fast filopodial cAMP transients generated in the growth cone center do not par- transients. Use of serum-free medium leads to observation of ticipate in axon steering. Calcium transients stimulated by Netrin-1 growth cone center calcium transients in Xenopus, Helisoma,and in growth cone centers bear close resemblance in amplitude, du- mammalian CNS neurons in vitro (26, 37–39). Signaling by tran- ration, and frequency to those that regulate axon extension (19, 33). sients allows different downstream effectors to be sensitive to Our findings extend compartmentalization of cAMP from car- different frequencies. Conversion of transients into persistent diac myocytes, neurons, and a variety of cell lines (1–3, 34) to effects has been described for synaptic plasticity, in which the growth cones and demonstrate how partitioning of signaling cAMP pathway is required for long-term changes (40, 41). Tran- enables regulation of dual functions by a single ligand. Differences sient signals have been proposed to lead to a long-lasting effect via in calcium and cAMP signal generators (calcium channels and a network of interacting signaling pathways in which cAMP and adenylyl cyclases) are common and might be the basis of our calcium are nodes (42). Integrators needed to persistently activate observations. Different subtypes of adenylyl cyclases are targeted downstream effectors could include calcineurin for regulation of

13780 | www.pnas.org/cgi/doi/10.1073/pnas.1100247108 Nicol et al. Downloaded by guest on September 28, 2021 axon outgrowth by growth cone center calcium transients (43) and The signaling events we describe may mediate responses to calpain for signals in filopodia (18). Our results suggest that the other guidance cues. We propose that cAMP and calcium tran- integrator for cAMP is activated above a threshold frequency of sients encode distinct growth-cone behaviors based on their transients, but not by an optimal frequency. localization in subcellular domains and on their frequency. Extinction of filopodial cAMP and calcium signals during sus- Guidance cues may activate one or the other of these pathways tained stimulation can account for desensitization of growth cones based on distinct receptors or clusters of receptors (22, 37) in to guidance cues and zig-zag navigation (23). Desensitization would restricted regions that confine the activation of downstream allow long-range pathfinding and adaptation of axons to different pathways (46). concentrations of guidance cues, and the same molecules would enable sustained control of axon extension. This corroborates the Materials and Methods dependence on IP3 receptor activation of axon outgrowth (similar Photostimulation of PACα was achieved by high-frequency, repetitive illu- to growth cone center calcium transients) but not axon steering mination of selected compartments of the growth cone using a 488-nm laser fi (dependent on lopodial calcium transients) (37). line. Imaging of spinal commissural axons in vivo was performed using Imposing a constant high or low concentration of cAMP in vivo embryos injected with Alexa 568 in a single blastomere at the two-cell stage leads to similar defects in trajectories of spinal commissural axons, (to label the midline) and with Alexa 488 in both V1 blastomeres at the indicating the importance of accurate regulation of cAMP levels. eight-cell stage (to label the dorsal spinal cord and commissural axons). The cAMP regulates several signaling pathways important for com- ventral surface of the spinal cord was then imaged after abdomen removal missural axon guidance, such as initiation of repulsion by sem- (Fig. S7A). A detailed description of experimental procedures is provided in aphorins after midline crossing by decreasing PKA activity (44), SI Materials and Methods. and responses of axons to a midline attractant such as Netrin-1 (this study). Sensitivity to repulsive cues is driven by a sustained ACKNOWLEDGMENTS. We thank Dr. M. Roe for the gift of pmEpac2 cAMPs change in overall concentration of cAMP (44) that may regulate generated from the Epac2-camps sensor of Dr. M. J. Lohse, and Dr. G. Nagel targeting of receptors to the cell surface (45). We find that local for the gift of PACα. We are grateful to members of our laboratory for axon pathfinding driven by Netrin-1 in vivo can be regulated by thoughtful discussion and to Darwin Berg, Michaël Demarque, Davide fi Dulcis, Patricia Gaspar, David Gomez-Varela, Kurt Marek, and Yimin Zou spatially and temporally restricted cAMP signals identi ed for helpful critical reading of the manuscript. This work was supported by in vitro. This finding suggests that it is possible to steer growth Fondation pour la Recherche Médicale and Marie Curie fellowships (to X.N.) cones in vivo through focal manipulation of intracellular signals. and by National Institutes of Health Grant NS15918 (to N.C.S.).

1. Zaccolo M, Pozzan T (2002) Discrete microdomains with high concentration of cAMP 24. Nikolaev VO, Bünemann M, Hein L, Hannawacker A, Lohse MJ (2004) Novel single chain in stimulated rat neonatal cardiac myocytes. Science 295:1711–1715. cAMP sensors for receptor-induced signal propagation. J Biol Chem 279:37215–37218. 2. Hockberger P, Yamane T (1987) Compartmentalization of cyclic AMP elevation in 25. Ooashi N, Futatsugi A, Yoshihara F, Mikoshiba K, Kamiguchi H (2005) Cell adhesion neurons of Aplysia californica. Cell Mol Neurobiol 7:19–33. molecules regulate Ca2+-mediated steering of growth cones via cyclic AMP and rya- 3. Shelly M, et al. (2010) Local and long-range reciprocal regulation of cAMP and cGMP nodine receptor type 3. J Cell Biol 170:1159–1167. in axon/dendrite formation. Science 327:547–552. 26. Gu X, Olson EC, Spitzer NC (1994) Spontaneous neuronal calcium spikes and waves 4. Zaccolo M, et al. (2006) Restricted diffusion of a freely diffusible second messenger: Mech- during early differentiation. J Neurosci 14:6325–6335. anisms underlying compartmentalized cAMP signalling. Biochem Soc Trans 34:495–497. 27. Schröder-Lang S, et al. (2007) Fast manipulation of cellular cAMP level by light in vivo. 5. Gorbunova YV, Spitzer NC (2002) Dynamic interactions of cyclic AMP transients and Nat Methods 4:39–42. spontaneous Ca2+ spikes. Nature 418:93–96. 28. Ishii N, Wadsworth WG, Stern BD, Culotti JG, Hedgecock EM (1992) UNC-6, a laminin- 6. Dunn TA, et al. (2006) Imaging of cAMP levels and protein kinase A activity reveals that related protein, guides cell and pioneer axon migrations in C. elegans. Neuron 9:873–881. NEUROSCIENCE retinal waves drive oscillations in second-messenger cascades. JNeurosci26:12807–12815. 29. Mitchell KJ, et al. (1996) Genetic analysis of Netrin in Drosophila: Netrins guide – 7. Murray AJ, Tucker SJ, Shewan DA (2009) cAMP-dependent axon guidance is distinctly CNS commissural axons and peripheral motor axons. Neuron 17:203 215. fi regulated by Epac and protein kinase A. J Neurosci 29:15434–15444. 30. Sera ni T, et al. (1996) Netrin-1 is required for commissural axon guidance in the – 8. Ming GL, et al. (1997) cAMP-dependent growth cone guidance by netrin-1. Neuron developing vertebrate . Cell 87:1001 1014. 19:1225–1235. 31. Shim S, et al. (2005) XTRPC1-dependent chemotropic guidance of neuronal growth – 9. Höpker VH, Shewan D, Tessier-Lavigne M, Poo M, Holt C (1999) Growth-cone at- cones. Nat Neurosci 8:730 735. traction to netrin-1 is converted to repulsion by laminin-1. Nature 401:69–73. 32. Moore SW, et al. (2008) Soluble adenylyl cyclase is not required for axon guidance to – 10. Corset V, et al. (2000) Netrin-1-mediated axon outgrowth and cAMP production re- netrin-1. J Neurosci 28:3920 3924. 33. Gu X, Spitzer NC (1995) Distinct aspects of neuronal differentiation encoded by fre- quires interaction with adenosine A2b receptor. Nature 407:747–750. quency of spontaneous Ca2+ transients. Nature 375:784–787. 11. Wu KY, et al. (2006) Soluble adenylyl cyclase is required for netrin-1 signaling in nerve 34. Willoughby D, Cooper DMF (2007) Organization and Ca2+ regulation of adenylyl cy- growth cones. Nat Neurosci 9:1257–1264. clases in cAMP microdomains. Physiol Rev 87:965–1010. 12. Terman JR, Kolodkin AL (2004) Nervy links protein kinase A to plexin-mediated 35. Gloerich M, Bos JL (2010) Epac: Defining a new mechanism for cAMP action. Annu Rev repulsion. Science 303:1204–1207. Pharmacol Toxicol 50:355–375. 13. Nicol X, Muzerelle A, Rio JP, Métin C, Gaspar P (2006) Requirement of adenylate 36. Peace AG, Shewan DA (2011) New perspectives in cyclic AMP-mediated axon growth cyclase 1 for the ephrin-A5-dependent retraction of exuberant retinal axons. J Neu- and guidance: The emerging epoch of Epac. Brain Res Bull 84:280–288. rosci 26:862–872. 37. Li L, Hutchins BI, Kalil K (2009) Wnt5a induces simultaneous cortical axon outgrowth and 14. Nicol X, et al. (2007) cAMP oscillations and retinal activity are permissive for ephrin repulsive axon guidance through distinct signaling mechanisms. J Neurosci 29:5873–5883. signaling during the establishment of the retinotopic map. Nat Neurosci 10:340–347. 38. Gomez TM, Snow DM, Letourneau PC (1995) Characterization of spontaneous calcium 15. Yamada RX, et al. (2008) Long-range axonal calcium sweep induces axon retraction. transients in nerve growth cones and their effect on growth cone migration. Neuron J Neurosci 28:4613–4618. 14:1233–1246. 16. Hong K, Nishiyama M, Henley J, Tessier-Lavigne M, Poo M (2000) Calcium signalling in 39. Williams DK, Cohan CS (1995) Calcium transients in growth cones and axons of cul- – the guidance of nerve growth by netrin-1. Nature 403:93 98. tured Helisoma neurons in response to conditioning factors. J Neurobiol 27:60–75. 17. Gomez TM, Robles E, Poo M, Spitzer NC (2001) Filopodial calcium transients promote 40. Nguyen PV, Abel T, Kandel ER (1994) Requirement of a critical period of transcription – substrate-dependent growth cone turning. Science 291:1983 1987. for induction of a late phase of LTP. Science 265:1104–1107. 18. Robles E, Huttenlocher A, Gomez TM (2003) Filopodial calcium transients regulate growth 41. Tzounopoulos T, Janz R, Südhof TC, Nicoll RA, Malenka RC (1998) A role for cAMP – cone motility and guidance through local activation of calpain. Neuron 38:597 609. in long-term depression at hippocampal mossy fiber synapses. Neuron 21: fi 19. Gomez TM, Spitzer NC (1999) In vivo regulation of axon extension and path nding by 837–845. – growth-cone calcium transients. Nature 397:350 355. 42. Bhalla US, Iyengar R (1999) Emergent properties of networks of biological signaling fi 20. Zheng JQ, Wan JJ, Poo MM (1996) Essential role of lopodia in chemotropic turning pathways. Science 283:381–387. of nerve growth cone induced by a glutamate gradient. J Neurosci 16:1140–1149. 43. Graef IA, et al. (2003) Neurotrophins and netrins require calcineurin/NFAT signaling to 21. Han J, Han L, Tiwari P, Wen Z, Zheng JQ (2007) Spatial targeting of type II protein stimulate outgrowth of embryonic axons. Cell 113:657–670. kinase A to filopodia mediates the regulation of growth cone guidance by cAMP. J 44. Parra LM, Zou Y (2010) Sonic hedgehog induces response of commissural axons to Cell Biol 176:101–111. Semaphorin repulsion during midline crossing. Nat Neurosci 13:29–35. 22. Hong K, et al. (1999) A ligand-gated association between cytoplasmic domains of 45. Bouchard JF, et al. (2004) Protein kinase A activation promotes plasma membrane UNC5 and DCC family receptors converts netrin-induced growth cone attraction to insertion of DCC from an intracellular pool: A novel mechanism regulating commis- repulsion. Cell 97:927–941. sural axon extension. J Neurosci 24:3040–3050. 23. Ming GL, et al. (2002) Adaptation in the chemotactic guidance of nerve growth cones. 46. Katsuki T, Ailani D, Hiramoto M, Hiromi Y (2009) Intra-axonal patterning: Intrinsic com- Nature 417:411–418. partmentalization of the axonal membrane in Drosophila neurons. Neuron 64:188–199.

Nicol et al. PNAS | August 16, 2011 | vol. 108 | no. 33 | 13781 Downloaded by guest on September 28, 2021