Axonal growth cones migrate along the correct paths during development, not only directed by guidance cues but also contacted by local environment via cell adhesion molecules (CAMs). Asymmetric Ca2+ elevations in the growth cone cytosol induce both attractive and repulsive turning in response to the guidance cues (Zheng, J.Q. 2000. Nature. 403:89–93; Henley, J.R., K.H. Huang, D. Wang, and M.M. Poo. 2004. Neuron. 44:909–916). Here, we show that CAMs regulate the activity of ryanodine receptor type 3 (RyR3) via cAMP and protein kinase A in dorsal root ganglion neurons. The activated RyR3 mediates Ca2+-induced Ca2+ release (CICR) into the cytosol, leading to attractive turning of the growth cone. In contrast, the growth cone exhibits repulsion when Ca2+ signals are not accompanied by RyR3-mediated CICR. We also propose that the source of Ca2+ influx, rather than its amplitude or the baseline Ca2+ level, is the primary determinant of the turning direction. In this way, axon-guiding and CAM-derived signals are integrated by RyR3, which serves as a key regulator of growth cone navigation.
During neuronal development and regeneration, the growth cone navigates the axon toward its target by monitoring environmental cues along its path. Such environmental cues can be diffusible, substrate-bound or cell surface molecules, which serve as either attractants or repellents upon binding to relevant receptors expressed on the growth cone surface (Tessier-Lavigne and Goodman, 1996; Huber et al., 2003). Many guidance molecules, including netrin-1, NGF, brain-derived neurotrophic factor, and myelin-associated glycoprotein (MAG), influence the direction of axon extension through cytosolic Ca2+ signals in the growth cone (Song and Poo, 1999; Hong et al., 2000). It has also been shown that a focal and unilateral elevation of cytosolic free Ca2+ concentration ([Ca2+]c) in the growth cone is sufficient to induce its turning to the side with elevated [Ca2+]c (Zheng, 2000). Interestingly, extracellular gradients of netrin-1 or MAG, which normally induce attractive or repulsive growth cone turning, respectively, increase [Ca2+]c that is highest on the side of the growth cone facing the source of the netrin-1/MAG gradients (Hong et al., 2000; Henley et al., 2004). This observation indicates that a [Ca2+]c gradient across the growth cone can trigger turning to the side with higher [Ca2+]c (attraction) as well as to the side with lower [Ca2+]c (repulsion). What, then, determines the growth cone responses to the [Ca2+]c gradient? An important distinction might be the amplitude of Ca2+ signals, with high-amplitude signals mediating attraction and low-amplitude signals mediating repulsion (Nishiyama et al., 2003; Henley et al., 2004; Wen et al., 2004). An earlier study showed that baseline [Ca2+]c is a determinant of the growth cone responses to [Ca2+]c increases: normal baseline [Ca2+]c (∼130 nM) favors attraction, and low baseline [Ca2+]c (∼60 nM) favors repulsion (Zheng, 2000). However, because cytosolic Ca2+ signals above some threshold level activate the ryanodine receptors (RyRs) on the internal Ca2+ stores (for review see Bouchard et al., 2003), high-level Ca2+ signals are likely to trigger secondary Ca2+ influx to the cytosol via RyRs, a process called Ca2+-induced Ca2+ release (CICR). Therefore, it is possible that the differential turning responses of the growth cone to Ca2+ signals depend primarily on the source of Ca2+ influx rather than the amplitude or the baseline level of Ca2+ signals. Answering this question is an important step for identifying downstream Ca2+ effectors and understanding the precise mechanisms of growth cone navigation.
Whilst guided by molecular guidance cues, a migrating growth cone also makes adhesive contact with its environment via cell adhesion molecules (CAMs). CAMs expressed by neurons can be divided into three large families: integrins, cadherins, and the immunoglobulin superfamily (IgSF) members. Cadherins and the majority of IgSF CAMs interact homophilically, i.e., with the same type of molecules present on adjacent cells (Takeichi, 1991; Brummendorf and Rathjen, 1994). In contrast, integrins interact heterophilically, i.e., with extracellular matrix molecules, such as laminin (Letourneau et al., 1994). Upon ligand binding, different sets of CAMs expressed at the growth cone are likely to influence its turning response to guidance cues. For example, growth cone attraction to netrin-1 is converted to repulsion by laminin (Hopker et al., 1999). Furthermore, neuropilin-1 forms a cis-dimer with L1, an IgSF CAM, on the growth cone membrane and mediates repulsive turning against semaphorin 3A (Castellani et al., 2000). This repulsive response is converted to attraction by trans binding L1 with neuropilin-1 (Castellani et al., 2002). These results raise the possibility that CAMs influence the Ca2+ signals that mediate growth cone steering in response to guidance cues.
In this paper, we report two major findings: (1) the primary determinant of the turning direction is the source of Ca2+ influx to the growth cone cytosol. RyR-mediated CICR triggers attractive turning, whereas Ca2+ signals without CICR induce repulsive turning. Consistent with this in vitro result, the RyR is also implicated in axon guidance in vivo. (2) CAMs influence growth cone steering by regulating the RyR activity via cAMP and protein kinase A (PKA). Growth cones migrating on an L1 or N-cadherin substrate have active RyRs and make an attractive turn in response to Ca2+ signals, whereas growth cones on a laminin substrate show the opposite behavior. Our results indicate that axon-guiding and CAM-derived signals merge at the level of RyRs, which serve as key regulators of growth cone navigation.
Ca2+-induced growth cone turning on different substrates
Spatially restricted Ca2+ signals in a growth cone of a dorsal root ganglion (DRG) neuron were produced by focal laser–induced photolysis (FLIP) of a caged Ca2+ compound, o-nitrophenyl EGTA (NP-EGTA) compounded with Ca2+ in the cytosol. Time-lapse images of [Ca2+]c using Calcium Green-1 (CG-1) demonstrated a focal and transient elevation of [Ca2+]c in response to a single laser pulse that photolyzed NP-EGTA (Fig. 1 A). This type of experimental manipulation has been used extensively to study the physiological roles of Ca2+ signals in axon growth and guidance (Gomez and Spitzer, 1999; Gomez et al., 2001; Wen et al., 2004). Consistent with a previous study (Zheng, 2000), repetitive FLIP of NP-EGTA on one side of the growth cone resulted in attractive turning on both L1 and N-cadherin substrates (Fig. 1, B and D; and Fig. 2 A). The attractive turning was sometimes preceded by lamellipodial protrusion (Fig. 1 B, 3 min) and filopodial elongation (not depicted), as reported previously (Lau et al., 1999; Cheng et al., 2002). In contrast, the same FLIP treatment caused repulsive turning on a laminin substrate (Fig. 1 F and Fig. 2 A). As a control, simultaneous loading of 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA), a fast Ca2+ chelator, completely negated the turning responses to NP-EGTA photolysis, indicating that Ca2+ released from NP-EGTA, but not byproducts of photolyzed NP-EGTA, was responsible for growth cone turning (Fig. 2 and see Fig. 7). These results indicate that, depending on the substrates, focal Ca2+ signals in growth cones induce either attractive or repulsive turning. The three substrates (L1, N-cadherin, and laminin) did not influence the subpopulations of surviving neurons as shown in a previous study (Nishimura et al., 2003), excluding the possibility that the differential turning behavior of growth cones was dependent on neuronal cell types.
Substrate dependency of Ca2+-induced growth cone turning is accounted for by differential cAMP activities
Hopker et al. (1999) previously reported that netrin-1–mediated attraction of retinal growth cones was converted to repulsion by laminin. These authors also explained that the amount of cAMP in growth cones, as assessed by immunocytochemistry, was decreased in the presence of laminin, suggesting that cAMP might mediate cross talk between growth cone guidance and adhesion. Therefore, we first compared cAMP levels in DRG neurons cultured on three different substrates—L1, N-cadherin, and laminin. A competitive immunoassay showed that the neurons contained higher cAMP levels on L1 and N-cadherin than on laminin (Fig. 3 A). To quantify the activity of cAMP in the growth cone, we used FlCRhR, which changes the efficiency of fluorescence resonance energy transfer from fluorescein to rhodamine, in response to cAMP binding to the PKA regulatory subunits (Adams et al., 1991). Consistent with our immunoassay data, the ratio images (F520/F580) showed that the cAMP activities in the growth cones were higher on L1 and N-cadherin than on laminin (Fig. 3, B and C). Next, the effect of cAMP modulation on Ca2+-induced growth cone turning was examined. Bath application of Rp-cAMPS, a membrane-permeable cAMP antagonist, converted Ca2+-induced attraction to repulsion on both L1 and N-cadherin substrates (Fig. 1, C and E; and Fig. 2 B). An inhibition of PKA, a cAMP-dependent protein kinase, by KT5720 had a similar result (Fig. 2 B). Sp-cAMPS, a cAMP agonist that activates PKA, converted Ca2+-induced repulsion to attraction on laminin (Fig. 1 G and Fig. 2 B). These results indicate that differential cAMP–PKA activities account for the substrate dependency of Ca2+-induced growth cone turning, with high cAMP–PKA activities favoring attraction and low cAMP–PKA activities causing repulsion.
Because PKA has been shown to phosphorylate RyRs and facilitate CICR in other biological systems (Yoshida et al., 1992; Valdivia et al., 1995; Wehrens and Marks, 2003), we assessed whether Ca2+-induced growth cone attraction depended also on RyR-mediated CICR in addition to cAMP–PKA. As shown in Fig. 2 B, Ca2+ signals caused growth cone repulsion even on L1 and N-cadherin substrates if neurons were pretreated with a high dose (100 μM) of ryanodine to trap RyRs in the closed state (Zucchi and Ronca-Testoni, 1997), suggesting that RyRs cooperated with cAMP–PKA in Ca2+-mediated growth cone attraction.
cAMP facilitates RyR-mediated CICR in growth cones
To test whether cAMP–PKA facilitated CICR in growth cones, we analyzed the amplitude of Ca2+ signals by CG-1 imaging synchronized with FLIP. Focal [Ca2+]c was quantified by averaging CG-1 ΔF/F0 within a 2-μm-diameter zone centered by the FLIP site (Fig. 4). A transient elevation of [Ca2+]c was observed only in the first charge-coupled device (CCD) frame after FLIP (0.8–16.5 ms after a laser pulse; see Materials and methods). This elevation should be the sum of Ca2+ liberation from NP-EGTA and secondary Ca2+ influx to the cytosol, if applicable. To dissect the two different Ca2+ sources, we compared the amplitude of [Ca2+]c elevations in the same growth cones before and after pharmacological treatments. In a growth cone on laminin, for example, a train of five laser pulses produced Ca2+ signals of similar amplitude that increased after treating the growth cone with Sp-cAMPS (Fig. 4, A and B). In contrast, Ca2+-signal amplitude decreased after a growth cone on N-cadherin was treated with 100 μM ryanodine (Fig. 4, C and D). The amplitude of five Ca2+ elevations (ΔF/F0 spikes) after five laser pulses was averaged and compared in the same growth cone before and after a 5-min treatment with various pharmacological reagents (Fig. 5). The Sp-cAMPS treatment augmented Ca2+ signals in growth cones only on a laminin substrate, presumably because the downstream cascade of cAMP was fully active on L1 and N-cadherin, even in the absence of Sp-cAMPS. This augmentation by Sp-cAMPS on laminin was abolished by a simultaneous treatment with 100 μM ryanodine, suggesting that Sp-cAMPS facilitated an additional increase of [Ca2+]c via ryanodine-sensitive channels, such as RyR-mediated CICR. Consistent with this idea, the treatment with either Rp-cAMPS or ryanodine attenuated Ca2+ signals in growth cones on L1 and N-cadherin but not on laminin.
Among the three RyR isoforms (types 1, 2, and 3), only the type 3 isoform (RyR3) has been detected in DRGs by Western blotting (Lokuta et al., 2002). Therefore, we analyzed DRG neurons derived from RyR3 knockout mice (Futatsugi et al., 1999) to further confirm that the RyR is involved in the augmentation of Ca2+ signals induced by cAMP. As reviewed by Bouchard et al. (2003), there has been accumulating evidence for the presence of functional RyRs in the axons and presynaptic terminals. Our immunocytochemical analysis showed the punctate distribution of RyRs in growth cones with strong expression in the central domain and weak expression in the lamellipodia (Fig. 6, A, B, D, and E). RyR expression and distribution were not influenced by culture substrates (unpublished data). Although the primary antibody recognized all three RyR isoforms (Kuwajima et al., 1992; Futatsugi et al., 1999), the immunoreactivity to this antibody was reduced or sometimes absent in RyR3-deficient growth cones (Fig. 6, C and F), further confirming that RyR3 is a predominant isoform in DRG neurons. On a laminin substrate, the Sp-cAMPS treatment augmented the amplitude of FLIP-induced Ca2+ signals in wild-type growth cones, whereas the same Sp-cAMPS treatment did not have that effect in RyR3-deficient growth cones (Fig. 6 G). On L1, the Rp-cAMPS treatment attenuated the amplitude of Ca2+ signals in wild-type growth cones but not in RyR3-deficient growth cones (Fig. 6 H). Collectively, our results indicate that the growth cone regulates the efficiency of RyR3-mediated CICR through cAMP–PKA upon contact with different substrates.
RyR3 controls Ca2+-mediated turning of growth cones
Our results up to this point suggested that cAMP–PKA controlled Ca2+-induced turning of growth cones by modulating the efficiency of RyR3-mediated CICR. Therefore, we next performed the axon turning assay using RyR3-deficient neurons. On L1 and N-cadherin substrates that activated the cAMP pathway, RyR3-deficient growth cones showed Ca2+-mediated repulsion, whereas wild-type growth cones exhibited attraction (Fig. 6 I). The most important evidence for the involvement of RyR3 downstream of the cAMP pathway came from the assay on a laminin substrate, in which Ca2+-mediated repulsion of wild-type growth cones, but not of RyR3-deficient growth cones, was converted to attraction by Sp-cAMPS (Fig. 6 I). Collectively, growth cone–substrate interactions regulate cAMP-activated RyR3-mediated CICR that determines the turning response to Ca2+ signals.
Because RyR-mediated CICR augmented the amplitude of FLIP-induced Ca2+ signals (Fig. 5), it remained to be determined whether the turning behavior of growth cones depended on the amplitude or the source of Ca2+ signals. Therefore, we tested whether attractive turning of growth cones could be switched to repulsion when the Ca2+-signal amplitude was lowered by coloading neurons with NP-EGTA and increasing concentrations of BAPTA followed by laser irradiation (Fig. 7). Elevating BAPTA acetoxymethyl ester (AM) concentrations caused attenuated or undetectable Ca2+ signals (CG-1 ΔF/F0) in growth cones on N-cadherin and decreased the attractive turning angle to zero without causing a significant repulsive turn. This result indicated that lowering the amplitude of Ca2+ signals was not sufficient to induce growth cone repulsion if the cAMP–PKA pathway was active. We also tested whether growth cone repulsion on laminin could be converted to attraction by increasing the Ca2+-signal amplitude. Previous work demonstrated that FLIP-induced Ca2+ signals in the growth cones of Xenopus laevis spinal neurons on laminin caused attractive or repulsive turning when the growth cones had been loaded with 6 or 2 μM NP-EGTA, respectively (Wen et al., 2004). However, under our experimental conditions, wild-type mouse DRG growth cones on laminin still showed repulsive turning (−13.3 ± 7.4°, n = 11) even if loaded with the high concentration (6 μM) of NP-EGTA followed by FLIP. This result was further confirmed with RyR3-deficient growth cones on N-cadherin. In this set of experiments, wild-type and RyR3-deficient growth cones were loaded with 2 and 6 μM NP-EGTA, respectively. The baseline [Ca2+]c in the growth cones was measured by Fura-2 ratio imaging, and the amplitude of [Ca2+]c elevations immediately after FLIP was determined by CG-1 single-wavelength measurements as described previously (Lev-Ram et al., 1992). The baseline [Ca2+]c in wild-type growth cones was 202 ± 17 nM (n = 50), which rose to 643 ± 48 nM (n = 8) after FLIP. Similarly, [Ca2+]c in RyR3-deficient growth cones rose from 207 ± 12 nM (n = 51) to 861 ± 106 nM (n = 11) in response to FLIP. Statistical analysis showed no significant difference between the two groups of growth cones, indicating that the Ca2+-signal amplitude in RyR3-deficient growth cones with 6 μM NP-EGTA loading was equivalent to or slightly higher than that in wild-type growth cones with 2 μM NP-EGTA loading. Then, the two groups were tested for their turning response to FLIP-induced Ca2+-signals. RyR3-deficient growth cones with 6 μM NP-EGTA exhibited repulsive turning (−14.6 ± 8.1°, n = 10), whereas wild-type growth cones with 2 μM NP-EGTA showed attraction (26.5 ± 8.6°, n = 7) in parallel experiments. The difference was statistically significant (P < 0.05). This result indicated that, depending on the occurrence of RyR3-mediated CICR, growth cones were either attracted or repelled by [Ca2+]c elevations of the equivalent amplitude on top of the same baseline level. We conclude that the source of Ca2+ influx, rather than its amplitude or the baseline Ca2+ level, is a decisive factor that regulates the direction of growth cone steering: RyR3-mediated CICR triggers attraction, whereas Ca2+ signals without CICR induce repulsion.
RyR3 is implicated in DRG axon guidance in vivo
To test for the involvement of RyRs in growth cone steering in vivo, we analyzed DRG axon trajectories in the spinal cord of RyR3 knockout mice. Because our in vitro experiments were performed using DRG neuronal cultures supplemented with NGF, the corresponding subpopulation of DRG axons in vivo was labeled by an antibody against calcitonin gene-related peptide (CGRP), a marker for the NGF-dependent TrkA-positive nociceptive neurons (Snider, 1994; Averill et al., 1995). As reported previously (Gibson et al., 1984), there are two major types of CGRP-positive axon projections in the spinal cord: those that terminate in the superficial dorsal horn (laminae I and II) and those that project deeply in the dorsal gray commissure (dorsal to the central canal; Fig. 8). Although the CGRP-positive superficial projections were not affected by RyR3 deficiency, we found a significant difference in the deep projections between wild-type and RyR3 knockout mice. In the wild-type spinal cord, CGRP-positive axons projecting toward the dorsal gray commissure extended horizontally or in a slightly dorsal trajectory away from the central canal (Fig. 8, A and B). In the RyR3-deficient spinal cord, however, the axons projected ventrally toward the central canal (Fig. 8, C and D). This phenotype was consistently observed in six knockout animals in comparison with six wild-type littermates. The similar pathfinding error of nociceptive axons was observed when the spinal cord was treated with a function-blocking antibody against NgCAM (a chick homologue of L1) or its binding partner, axonin-1 (Perrin et al., 2001), suggesting that homophilic or heterophilic L1 interactions are required for these axons to be guided correctly by extracellular cues.
During development, the precise pattern of neuronal networks depends on the guided advance of growth cones along correct paths toward their targets. The growth cone serves as the sensory and motile apparatus for translating extracellular cues into directional elongation of the axon. A cytosolic Ca2+ signal is one of the most important messengers that relay extracellular information to the motile behavior of growth cones (Davenport et al., 1993; Kuhn et al., 1998; Takei et al., 1998; Gomez and Spitzer, 1999; Doherty et al., 2000; Gomez et al., 2001; Tang et al., 2003). Elongating axons could respond to the cytosolic Ca2+ signal in multiple ways, depending on the source of Ca2+ influx to the cytosol and its amplitude and frequency. The work we present here demonstrates the importance of the source of Ca2+ signals in determining the turning direction of growth cones, in which RyR-mediated CICR triggers attraction, whereas Ca2+ signals without CICR induce repulsion. We also show that CAMs regulate RyR activity through cAMP and PKA and that the growth cone is able to change its turning behavior on different substrates. The most significant point of our work is the direct demonstration, using gene-manipulated mice, that RyR3 is a downstream component of the cAMP–PKA pathway and that RyR3 controls the bidirectional turning responses induced by Ca2+ signals of the same polarity in growth cones. In this way, axon-guiding and CAM-derived signals are integrated at the level of RyR3, which may provide growth cones with the flexibility of generating distinct responses to the same guidance cue in different environmental situations in vivo. This concept has been supported by our experiment that showed the altered projection of DRG axons in the RyR3-deficient spinal cord, although axon-guiding molecules involved in this phenotype remain to be identified. Antibody perturbation experiments by Perrin et al. (2001) suggested that impaired L1 interactions with its binding partners caused a similar guidance error of DRG axons, which is consistent with our concept that L1 controls growth cone turning responses to guidance cues by modifying cytosolic Ca2+ signals.
Because the occurrence of CICR and the amplitude of Ca2+ signals are tightly associated, it has been challenging to dissect these two factors experimentally. Although our data support the idea that the source of Ca2+ signals is the primary determinant of the turning direction of growth cones, several previous studies have suggested that the Ca2+-signal amplitude is the important factor. For example, Henley et al. (2004) conducted clever experiments in which different levels of [Ca2+]c gradients were created across a growth cone by unilateral application of a Ca2+-selective ionophore through a micropipette in the presence of distinct preset Ca2+ concentrations in bathing media. Their study showed that high- or low-amplitude Ca2+ signals were sufficient to trigger attraction or repulsion of the growth cone, respectively. In their experiments, however, low-amplitude Ca2+ signals were produced after treating the cells with thapsigargin that depleted Ca2+ from the internal stores, raising the possibility that impaired CICR might have been responsible for the repulsion. Another example comes from studies by Zheng and colleagues (Zheng, 2000; Wen et al., 2004), in which they lowered the absolute level of cytosolic Ca2+ signals by removing Ca2+ from external bathing media. The growth cone repulsion observed under this experimental condition might have been attributable to the indirect depletion of internal Ca2+ stores. Therefore, the results in these two papers do not contradict our proposal that RyR-mediated CICR triggers attraction, whereas Ca2+ signals without CICR induce repulsion. A study by Henley et al. (2004) showed that MAG-induced growth cone repulsion was blocked by inhibiting Ca2+ influx from ryanodine-sensitive channels but not from plasmalemmal channels. Furthermore, Nishiyama et al. (2003) reported that netrin-1–induced attraction was associated with high-amplitude Ca2+ influx through plasmalemmal L-type channels. These results initially appear to be in conflict with our conclusions; however, we used a direct [Ca2+]c elevation approach that bypassed activation of the MAG or netrin-1 receptors on the plasma membrane and could avoid cross talk among different intracellular signals. Therefore, growth cone turning assayed by extracellular application of guidance molecules cannot be compared simply with our study that focuses on [Ca2+]c elevations and their downstream events. We propose that Ca2+-signal localization at the molecular level is a decisive factor for switching the steering direction of growth cones. Perhaps Ca2+ signals confined to nanodomains/microdomains activate distinct sets of molecular targets (Augustine et al., 2003), depending on the location of open Ca2+ channels as well as on the distance between these channels and Ca2+-binding proteins. Although recent papers have identified some of the downstream effectors of Ca2+ (Robles et al., 2003; Wen et al., 2004), much more work will need to be done to understand the whole picture of switch-like mechanisms that navigate the growth cone bidirectionally in response to Ca2+ signals of different localization and amplitude.
Another important question concerns the causal relationship between cAMP–PKA and Ca2+ signals. Wen et al. (2004) demonstrated that the cAMP pathway acts downstream of [Ca2+]c elevations, negatively regulating the repulsive signaling cascade that involves calcineurin and phosphatase-1. This inhibition by cAMP allows [Ca2+]c elevations to activate only the attractive signaling components, thereby switching growth cone repulsion to attraction. In contrast, the cAMP pathway acts upstream of Ca2+ signals in growth cones guided by netrin-1 or MAG. Activation of the cAMP pathway increases the basal [Ca2+]c in growth cones, which converts MAG-induced repulsion to attraction because MAG-induced [Ca2+]c elevations can be superimposed on top of the higher basal [Ca2+]c (Henley et al., 2004). Also, growth cone attraction by netrin-1 depends on L-type Ca2+ channel activation by cAMP–PKA (Nishiyama et al., 2003). The present study has demonstrated that the cAMP pathway activates RyRs and allows primary [Ca2+]c elevations to induce the secondary Ca2+ influx from the internal stores, thereby acting at the level of Ca2+ signaling. Although the cAMP pathway may exert multiple roles at multiple steps in signal transduction cascades, we believe that the major mode of cAMP action, at least in DRG growth cones, is on the RyR-mediated CICR because RyR3-deficient axons did not respond to cAMP activation in either the Ca2+ imaging (Fig. 6 G) or the turning assay (Fig. 6 I).
In summary, this study shows that CAMs control Ca2+-mediated steering of growth cones via cAMP and RyRs. Upon contact with differing extracellular environments, growth cones could change their turning responses to axon guidance cues by modulating the efficiency of RyR-mediated CICR. In this way, CAMs and axon-guiding molecules cooperate to navigate axons along their correct paths. Because activating cAMP pathways has been shown to overcome myelin-associated inhibitors of axon regeneration (e.g., MAG) in the damaged central nervous system (Cai et al., 1999; Neumann et al., 2002; Qiu et al., 2002), our findings further suggest that CAMs may play a role in regenerating axons, presumably through a mechanism similar to that used by cAMP-dependent switching of growth cone repulsion to attraction.
Materials And Methods
DRGs dissected from embryonic day 9 chicks or postnatal day 0 mice were dissociated as previously described (Nakai and Kamiguchi, 2002) and plated on a glass-based dish coated with 9 μg/cm2 laminin (Life Technologies) or CAM-Fc chimeric proteins that consisted of the whole extracellular domain of a CAM (L1 or N-cadherin) and the Fc region of human IgG. The L1-Fc cDNA construct was provided by Dr. Vance Lemmon (University of Miami, Miami, FL), and the N-cadherin-Fc construct by Dr. Patrick Doherty (King's College London, London, UK). Production of CAM-Fc and preparation of CAM-Fc–coated dishes were performed as described previously (Kamiguchi and Yoshihara, 2001). The neuronal cultures were maintained in Leibovitz's L-15 medium (Life Technologies) supplemented with N-2 (Life Technologies), 50 ng/ml NGF (Promega), and 750 μg/ml BSA (Life Technologies) in a humidified atmosphere of 100% air at 37°C.
The 355-nm Nd:YAG laser beam (a pulse width of 5 ns; New Wave Research) was introduced into an inverted microscope (TE300; Nikon) through the random-scanning photolysis system (Hamamatsu Photonics) and was focused on a growth cone using a 100× Plan Fluor (NA 1.3, oil objective). Spatial restriction of FLIP (∼1 μm in diam in the xy plane and ∼2 μm in thickness in the z direction) was demonstrated by irradiating caged fluorescein maleimide (Dojindo) immobilized in acrylic polymer.
Growth cone turning assay
DRG neurons were incubated, unless otherwise noted, with 2 μM NP-EGTA AM (Invitrogen) and 0.025% Cremophor EL (Sigma-Aldrich) for 30 min and then washed. After an additional 1.5-h incubation, an area within a lamellipodium in one side of the growth cone was irradiated with a laser (∼35 nJ/pulse) every 3 s while time-lapse differential interference contrast (DIC) images of the growth cone were acquired with a CCD camera. Growth cone turning was assessed by tracing the middle point of the distal edge of the growth cone central domain. On L1 and N-cadherin substrates, the original (preFLIP) direction of axon extension was defined as a straight line connecting the positions of the growth cone at the onset and 15 min before the onset of FLIP. The direction during FLIP was defined as a straight line connecting the positions of the growth cone at the onset and at the end of repetitive FLIP for 60 min. The growth cone turning angle was defined by the angle between the original direction and the direction during FLIP. Because axons extended faster on a laminin substrate, the turning angle on laminin was determined using the positions of the growth cone at 10 min before, at the onset of, and at 30 min after repetitive FLIP. Included in this assay were only those growth cones with net extension >5 μm during the preFLIP periods and without collapse during the time-lapse imaging.
In some experiments, various concentrations of BAPTA AM (Invitrogen) were loaded into neurons before the turning assay (a 20-min incubation with BAPTA AM and Cremophor EL followed by a 30-min incubation). The following reagents were applied to some cultures 30 min before the turning assay: 20 μM Sp-cAMPS (Calbiochem), 20 μM Rp-cAMPS (Calbiochem), 100 μM ryanodine (Calbiochem), and 100 nM KT5720 (Calbiochem).
Imaging FLIP-induced Ca2+ signals
Dextran (10,000 D) conjugated with CG-1 (3 mg/ml; Invitrogen) was introduced into DRG neurons by trituration loading as described previously (Sydor et al., 1996). After NP-EGTA loading, FLIP-induced Ca2+ signals were imaged with a high-speed CCD camera (HiSCA; Hamamatsu Photonics) coupled with a Gen IV image intensifier (Hamamatsu Photonics). The images were background subtracted and expressed as the change in CG-1 fluorescence relative to the baseline before FLIP (ΔF/F0) using the AquaCosmos software version 2.0 (Hamamatsu Photonics). Each image was acquired by a 15.7-ms exposure with CCD binning set at 2 × 2. The sampling rate was 63.7 Hz. To synchronize FLIP with CG-1 imaging, a laser shot was triggered by a transistor–transistor logic signal that was generated by AquaCosmos and controlled in timing by a pulse generator (Nihon Kohden). Typically, a CCD exposure was initiated 0.8 ms after the laser shot (∼90 nJ/pulse). This interval was sufficiently long to prevent laser-induced autofluorescence from affecting the CG-1 image. Focal Ca2+ signals were evaluated by averaging ΔF/F0 values within a 2-μm-diam zone centered by the FLIP site. After a 5-min incubation with 20 μM Sp-cAMPS, 20 μM Rp-cAMPS, and/or 100 μM ryanodine, FLIP-induced Ca2+ signals were again analyzed in the same growth cone under the same experimental conditions.
In some experiments, FLIP-induced Ca2+ signals were quantified by converting the CG-1 ΔF/F0 value to [Ca2+] using the following formula (Lev-Ram et al., 1992): [Ca2+] = ([Ca2+]rest + Kd × f/fmax)/(1 − f/fmax), where f = ΔF/F0 and fmax = ΔFmax/F0. The dissociation constant (Kd) of dextran-conjugated CG-1 was determined as 413 nM (37°C, pH 7.2) by using the calcium calibration buffer kit with magnesium #2 (Invitrogen). The resting Ca2+ concentration ([Ca2+]rest) was measured by Fura-2 ratio imaging as described in the next section. The ΔFmax was the maximal change of CG-1 fluorescence, which was determined in every growth cone by applying 10 μM of calcium ionophore A23187 (Sigma-Aldrich) in media with a saturating amount of Ca2+ (1.26 mM).
Fura-2 ratio imaging
DRG neurons were incubated with 2 μM Fura-2 AM (Invitrogen) and 0.025% Cremophor EL for 30 min. After an additional 1-h incubation, Fura-2 was excited at two wavelengths (340 and 380 nm), and emission images were acquired at 510 nm. After background subtraction, the ratio of fluorescence intensity (R = F340/F380) was measured in a growth cone and converted to [Ca2+] using the following standard formula (Grynkiewicz et al., 1985): [Ca2+] = Kd × (Ffree/Fbound) × (R − Rmin)/(Rmax − R). The dissociation constant (Kd) of Fura-2 was determined as 207 nM (37°C, pH 7.2) by using the calcium calibration buffer kit with magnesium #2 (Invitrogen). The constant Ffree/Fbound was the fluorescence ratio of the Ca2+-free and -saturated forms of Fura-2 upon 380-nm excitation. Rmin designated the minimum R measured in the absence of Ca2+, and Rmax the maximum R at a saturating Ca2+ concentration. These parameters were determined in growth cones by applying 10 μM A23187 in media with a saturating amount of Ca2+ (1.26 mM) followed by perfusion with Ca2+-free buffer containing 2 mM EGTA. Values for Ffree/Fbound, Rmin, and Rmax were 2.15, 0.32, and 4.12, respectively.
Intracellular cAMP levels in neurons were measured using a competitive nonacetylation immunoassay kit (GE Healthcare), according to the manufacturer's instructions.
We used fluorescence ratio imaging of FlCRhR to measure the activity of cAMP in the growth cone cytosol. FlCRhR (20 ± 5 μM; Invitrogen) was pressure-injected into a DRG neuron with a micropipette. After 3 h of incubation, FlCRhR was excited using a band-pass filter (484/15 nm), and two single-wavelength emission images (F520 and F580) were simultaneously acquired using a band-pass (520/30 nm) and a log-pass filter (570 nm), respectively. After background subtraction, the ratio of F520/F580 was calculated and averaged within the growth cone using AquaCosmos.
DRG neurons were fixed with 4% formaldehyde in PBS for 40 min, permeabilized and blocked with 0.1% Triton X-100 and 10% horse serum in PBS for 1 h, and then incubated with rabbit polyclonal anti-RyR antibody (anti-C2; 2 μg/ml; Kuwajima et al., 1992) overnight at 4°C. Primary antibody binding was detected using Alexa 488–conjugated goat anti–rabbit IgG secondary antibody (10 μg/ml; Invitrogen).
Postnatal day 0 mice were deeply anesthetized and perfused with 4% formaldehyde in 0.1% Na-phosphate buffer. The lumbar spinal cord was dissected, postfixed overnight in the same fixative at 4°C, and infiltrated overnight with 25% sucrose in PBS at 4°C. 16-μm-thick cryostat sections were collected and stored at −20°C. After blocking and permeabilization with 10% horse serum and 0.1% Triton X-100 in PBS, the sections were incubated sequentially with rabbit polyclonal anti-CGRP antiserum (1:5,000 dilution; overnight at 4°C; Sigma-Aldrich) and Alexa 488–conjugated goat anti–rabbit IgG (10 μg/ml).
Data were expressed as the mean ± SEM. Statistical analyses were performed using Prism version 3.0a (GraphPad Software). Unless otherwise noted, a comparison between two groups was performed by an unpaired t test and a comparison among three or more groups by a one-way analysis of variance followed by Tukey's posttest. P values <0.05 were judged statistically different.
We would like to thank Dr. Vance Lemmon for providing us with the L1-Fc construct and Dr. Patrick Doherty for the N-cadherin-Fc construct.
This study was partially supported by a Grant-in-Aid for Scientific Research (C) of the Japan Society for the Promotion of Science (16500210) to H. Kamiguchi.
Abbreviations used in this paper: AM, acetoxymethyl ester; BAPTA, 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid; [Ca2+]c, cytosolic free Ca2+ concentration; CAM, cell adhesion molecule; CCD, charge-coupled device; CG-1, Calcium Green-1; CGRP, calcitonin gene-related peptide; CICR, Ca2+-induced Ca2+ release; DIC, differential interference contrast; DRG, dorsal root ganglion; FLIP, focal laser–induced photolysis; IgSF, immunoglobulin superfamily; MAG, myelin-associated glycoprotein; NP-EGTA, o-nitrophenyl EGTA; PKA, protein kinase A; RyR, ryanodine receptor.