Filopodial Calcium Transients Promote Substrate-Dependent Growth Cone Turning

See allHide authors and affiliations

Science  09 Mar 2001:
Vol. 291, Issue 5510, pp. 1983-1987
DOI: 10.1126/science.1056490


Filopodia that extend from neuronal growth cones sample the environment for extracellular guidance cues, but the signals they transmit to growth cones are unknown. Filopodia were observed generating localized transient elevations of intracellular calcium ([Ca2+]i) that propagate back to the growth cone and stimulate global Ca2+ elevations. The frequency of filopodial Ca2+ transients was substrate-dependent and may be due in part to influx of Ca2+ through channels activated by integrin receptors. These transients slowed neurite outgrowth by reducing filopodial motility and promoted turning when stimulated differentially within filopodia on one side of the growth cone. These rapid signals appear to serve both as autonomous regulators of filopodial movement and as frequency-coded signals integrated within the growth cone and could be a common signaling process for many motile cells.

Growth cone filopodia are sensory/motor protrusions that first detect changes in the molecular milieu during axon pathfinding (1–3). Activation of receptors at their tips by interaction with ligands may generate signals that are transmitted back to the growth cone where they are transduced into changes in motility, but both the identity of these signals and whether they regulate filopodial movement directly are unknown. Although slow spontaneous changes in [Ca2+]i within growth cones have been imaged at 0.1 Hz (4, 5), here we analyzed fast Ca2+ dynamics within individual filopodia of culturedXenopus spinal neurons (6) by high-speed (8 Hz) confocal imaging of magnified regions of Fluo-4 loaded growth cones (7). Most filopodia (55 ± 14%) of neurons grown on poly-d-lysine (PDL) generated brief Ca2+ transients (317 ± 85 ms) during 30-s imaging periods (Fig. 1, A and B) (n = 22 filopodia from seven growth cones); active filopodia generated transients at 9.2 ± 1.1 min−1. Ca2+ transients detected by ratiometric imaging of Fluo-4 and Fura-Red fluorescent signals (n = 9) (8,9) (Fig. 1, C and D) (movie 1, 10) ruled out artifactual fluorescence elevations due to movement or volume changes. Ca2+ elevations occurred locally at the tip or along the length of a filopodium and spread to the growth cone with no apparent decrement at 12.5 ± 0.6 μm/s (n = 12) (Fig. 1, E and F) (movie 2, 10), consistent with either propagated surface entry or intracellular diffusion in a weakly buffered environment (11). Filopodial transients appeared similar to growth cone transients (4, 5) in that they depended on Ca2+ influx through non–voltage-gated channels, but they were distinct because they were not amplified by Ca2+ release from internal stores (Fig. 1G).

Figure 1

Spontaneous filopodial Ca2+ transients visualized by high-speed imaging of Fluo-4 fluorescence. (A) Time-lapse pseudocolor confocal images of Fluo-4 loaded filopodia. Regions selected for quantification at the tip, middle, and base of one filopodium are shown at 0 ms. A local Ca2+ transient at the tip occurs at 125 ms (arrow). Fluorescence intensity is coded on a linear pseudocolor scale. (B) Normalized fluorescence changes within regions of this filopodium captured at 125-ms intervals. Ca2+transients occur at high frequency at the tip and move proximally toward the growth cone. Arrowhead indicates transient visualized in (A). F/F0 indicates Fluo-4 fluorescence (F) relative to baseline fluorescence (F0). (C) Pseudocolor ratio images of filopodia loaded with Fluo-4 and Fura-Red. (D) Fluorescence ratio increases within regions of the filopodium in (C) demonstrate that these transients are not due to artifacts of movement or volume changes. Note that elevated Ca2+ initiated at the tip causes a small elevation of Ca2+ at the distal region of this growth cone (base). Arrowhead indicates transient shown in (C). (E) Pseudocolor profile images show that elevated Ca2+ (colored arrows) moves from tip to base of this filopodium within 375 ms. (F) Fluorescence along the filopodium in (E) normalized to resting levels [linescan between white arrowheads in 0-ms panel in (E)] and smoothed with a five-point moving average. The velocity of propagation is 19.2 μm/s. Synchronous fluctuations are artifacts due to scanning at high zoom. (G) Ca2+ transients in filopodia on tissue culture plastic are generated by Ca2+ influx through a non–voltage-gated channel and are not affected by store release. Ca2+ influx is required because the average transient frequency decreased in reduced [Ca2+]o, and transients were abolished in 0 mM Ca2+, with 10 mM NiCl in the presence of 2 mM Ca2+ or by intracellular Ca2+ chelation with BAPTA. Filopodial transient frequency was unaffected by voltage-gated Ca2+ channel blockers (VGC) (50 μM NiCl, 10−7 M nifedipine, 10−6 M Ω-conotoxin, and 60 nM Ω-agatoxin) or Ca2+ store release with thapsigargin (TG). Neurons were grown on poly-d-lysine (PDL) in 2 mM Ca2+ except where indicated and imaged at 12 to 24 hours in culture. n > 20 filopodia for each condition. Bar in (A), (C) and (E), 2 μm.

To determine the dependence of fast Ca2+ transients on molecules that regulate neurite outgrowth, [Ca2+]iwas imaged in filopodia of neurons grown on eight different substrates (12). Transient frequency varied from one substrate to another whereas duration and amplitude were similar (Fig. 2A) (Web table 1, 10), indicating that these Ca2+ elevations could provide a specific signaling mechanism in growth cone filopodia. Growth cones did not generate global Ca2+ transients on tenascin-C (TN), fibronectin (FN), or vitronectin (VN), indicating that on certain substrates local filopodial transients are not sufficient to produce global elevations (Web table 1,10). In physiological extracellular Ca2+, the frequency of fast local transients was highest on TN and declined as the TN concentration ([TN]) was lowered (Fig. 2A inset), suggesting that TN stimulates transient production.

Figure 2

Frequencies of local filopodial Ca2+ transients are regulated by the substrate, in part through integrin receptors. (A) The average frequency of transients for all filopodia (including inactive ones) of neurons grown in 2 mM Ca2+ on artificial substrates (TC, PDL), cell adhesion molecules [N-cadherin (Cad) and L1], and on ECM components LN, FN, TN, and VN (50 μg/ml). For some substrates (L1, Cad, LN, PDL, and TC) the frequency of global growth cone transients correlated with the local filopodial transient frequency, but no correlation was seen for others (FN, VN, TN). TN stimulated the highest frequency of transients in filopodia, in a dose-dependent manner (inset). n ≥ 10 filopodia for each condition. (B) Soluble RGDS produced a dose-dependent, biphasic elevation of [Ca2+]i in 75% of growth cones tested (arrowhead; traces are averages from 10 growth cones) and stimulated increased frequency of filopodial and veil transients. Antibodies to β1 integrin did not block spontaneous filopodial Ca2+ transients, perhaps due to restricted antibody access. (C) Pseudocolor profile images show that elevated Ca2+ (arrow) in response to 1 mM soluble RGDS spreads through a growth cone veil in 1 s. Bar, 2 μm. Fluorescence intensity is coded on a linear pseudocolor scale. (D) Normalized fluorescence changes within the active veil in (C) captured at 125-ms intervals. Ca2+ transients occurred at high frequency after the addition of RGDS (arrow). Gap in trace represents 14 s during application. Neurons in (B) through (D) were grown on TN (5 μg/ml) in 2 mM Ca2+.

Because extracellular matrix (ECM) components, cell adhesion molecules, and artificial substrates stimulate Ca2+ transients in filopodia, Ca2+ entry could be activated by multiple receptor types. To determine whether a particular class of receptor is sufficient to generate these fast Ca2+ transients, we examined the role of β1 integrin receptors that bind many ECM proteins and trigger Ca2+ fluctuations (7,13). The tripeptide sequence arginine-glycine-aspartate (RGD), contained in many ECM molecules including TN (14) and recognized by a subset of integrins (15), is sufficient to stimulate adhesion and Ca2+ influx in non-neuronal cells (16). Soluble RGDS (S, serine) elevated [Ca2+]i in growth cones (Fig. 2B) and increased mean filopodial Ca2+transient frequency from 2.8 ± 0.4 to 4.3 ± 0.8 min−1 (P < 0.05; n = 26 active and inactive filopodia from seven growth cones). The effects of RGDS were specific because the non–integrin binding peptide RGES (E, glutamate) neither elevated [Ca2+]i nor altered filopodial transient frequency and antibodies that block the function of β1 integrins (17) prevented RGDS-mediated Ca2+ elevations. In addition to its effects on the frequency of filopodial Ca2+ transients, RGDS evoked transients involving larger regions in the growth cone veil (36%,n = 11) (Fig. 2, C and D) (movie 3, 10). These findings show that engagement of integrins on neuronal growth cones can activate Ca2+ influx and oscillations and suggest that integrin clusters (18, 19) may be the sites of fast local Ca2+ transients.

Integrins promote Ca2+ influx and adhesion, which lead to increased cell attachment, spreading, and migration (15). To determine whether the frequency of fast local Ca2+transients regulates filopodial motility, we measured the lifetime from initiation to retraction of filopodia (20, 21) of growth cones growing on three concentrations of TN that elicit varying frequencies of Ca2+ transients (Fig. 2A inset). As the [TN] increased, the average filopodial lifetime also increased, whereas the average axon length decreased. The effect of TN is Ca2+-dependent because blocking local transients with the Ca2+ chelator BAPTA reduced filopodial lifetimes (Fig. 3A). Regulation of filopodial lifetime by Ca2+ transients could be due to effects on the actin cytoskeleton (22) or adhesion through increased integrin affinity or clustering (avidity) (13). To test whether Ca2+ transients regulate integrin avidity, β1 integrin clustering was examined before and after chelation of Ca2+with BAPTA during the time when filopodial lifetimes are reduced (23, 24). Blocking endogenous filopodial Ca2+ transients with BAPTA had no detectable effect on integrin clustering at filopodial tips (Fig. 3, B and C). However, localized Ca2+ transients were consistently initiated at sites of punctate integrin clusters (15/15 clusters from 4 growth cones) (Fig. 3, D and E) (Web fig. 1, 10), suggesting that they stabilize filopodia through integrin activation.

Figure 3

Local Ca2+ transients reduce filopodial motility and slow neurite outgrowth. (A) Increased [TN] prolonged filopodial lifetimes and shortened neurite lengths. The effect of [TN] depended on Ca2+ transients because 100-nM BAPTA reduced their effects on lifetime (n ≥ 95 filopodia from three growth cones for each condition) with no effect on resting [Ca2+]i(5). (B and C) BAPTA-loaded (C) and unloaded control (B) growth cones were double labeled for β1 integrin (red) and actin (green). Neither the proportion of filopodia with integrin clusters at their tip (arrows) (52 versus 44% of 239 total filopodia in 30 growth cones; P = 0.3, Fisher's exact test), nor the size of these clusters (0.15 ± 0.01 μm2 versus 0.16 ± 0.01 μm2,n = 114 total clusters; P = 0.4) was significantly different, indicating that Ca2+ transients do not affect motility by increasing receptor clustering. (D and E) Sites of Ca2+ transient activity colocalized with β1 integrin clusters. (D) Pseudocolored maximum projection of a 30-s time-lapse sequence encompassing 240 confocal images of a Fluo-4 loaded growth cone. Bright regions indicate areas of Ca2+ transient activity (arrows) (Web fig. 1,10). (E) The same growth cone now labeled for β1 integrin shows that regions of transient activity occur at sites of β1 integrin clusters (arrows). Bar in (C) indicates unit of measure in (B) and (C), 2 μm; bar in (D) indicates unit of measure in (D) and (E), 2 μm. Neurons were grown on TN (10 μg/ml) in 2 mM Ca2+.

To determine whether Ca2+ transients generated in filopodia promote global growth cone transients, we exploited differences in transient frequencies evoked by laminin (LN) and tissue culture plastic (TC). Because growth cones exhibit a higher frequency of both local and global Ca2+ transients on TC than LN, we investigated the distance at which Ca2+ dynamics change as growth cones encounter a LN-TC border (25,26). The frequency of Ca2+ transients in growth cones on LN increased as they approached TC, suggesting that filopodial contact with TC is sufficient to induce global Ca2+ changes (Web fig. 2A, 10). The signals promoting global Ca2+ transients may be local Ca2+ transients because these elevations occurred at a higher frequency within individual filopodia that project onto TC compared with LN (27). To investigate whether filopodial input stimulates the production of global Ca2+transients, global transient frequency was measured in growth cones at this boundary in the presence of 100 nM cytochalasin D to disrupt actin filaments. In the absence of filopodia, the frequency of these transients did not increase until growth cones were within 5 μm of the border (Web fig. 2A, 10). Because Ca2+ transients precede growth cone turning in vivo (28), we analyzed whether they induce axon turning at a LN-TC border in vitro. Most axons turned to remain on LN when transients were allowed, whereas most crossed into TC when global elevations were blocked in either 0.2 mM Ca2+ or when all transients were blocked with BAPTA (Web fig. 2B, 10). These results indicate that when filopodia contact a second substrate, local Ca2+ signals spread to the growth cone and stimulate global Ca2+ elevations that promote growth cone turning.

To test whether filopodial Ca2+ transients on one side of the growth cone are sufficient to promote turning, we examined growth cones at a boundary between high (50 μg/ml) and low (5 μg/ml) TN (25). Because growth cones do not generate global Ca2+ elevations on TN, Ca2+-dependent turning at a high [TN]/low [TN] border would likely result from differences in the frequency of filopodial Ca2+ transients on varying [TN] (Fig. 2A). Seventy-nine percent of axons turned to remain on a low [TN] under control conditions (n = 28), whereas only 33% turned when filopodial transients were blocked with 100 nM BAPTA (n = 27; P < 0.002, Fisher's exact test). To assess differences in local transient frequencies at this border, Fluo-4 loaded growth cones turning away from a high [TN] were imaged (Fig. 4, A and B). Filopodia in contact with a high [TN] exhibited both a higher incidence and frequency (69%, 6.3 ± 0.8 min−1,n = 26) of transients compared to those contacting a low [TN] (54%, 4.7 ± 0.4 min−1, n= 37, P < 0.05). Thus a step-gradient of TN directs axon growth in a Ca2+-dependent manner by stimulating filopodial Ca2+ transients on one side of the growth cone.

Figure 4

Local filopodial Ca2+ transients regulate turning. (A) Fluo-4 loaded growth cone turning at a border between high and low [TN]. Border is identified by TRITC fluorescence (50 μg/ml TN). Filopodia on either side of the boundary were monitored for Ca2+ transient activity (boxes). Bar, 5 μm. (B) Normalized fluorescence changes within regions of the filopodia in (A) captured separately at 125-ms intervals. The frequency of Ca2+ transients is higher in the filopodium on 50 μg/ml TN than those on 5 μg/ml TN. (C) Fluo-4/NP-EGTA double-loaded growth cone placed adjacent to a spot within which filopodia are exposed to UV pulses (yellow circle) whereas those outside are not. Bar, 10 μm. (D) Normalized fluorescence changes within regions of the filopodia in (C) captured sequentially at 125-ms intervals. Ca2+ elevations at a frequency of 6 min−1 are generated only in filopodia within the spot. (E through K) Fast Ca2+ transients imposed in filopodia on one side of a growth cone promote turning to the other side. Unloaded control [(E) through (G)] and NP-EGTA loaded [(H) through (J)] growth cones at the onset of UV pulses [(E) and (H)] and after 30 min growth toward the spot [(F) and (I)]. (G) and (J) Superimposed profiles of unloaded control growth cones (G) and those loaded with NP-EGTA (J) after 5 to 15 μm of growth toward a region exposed to pulsed UV light at 6 min−1 (yellow circles). In only one instance did a growth cone loaded with caged-Ca2+ penetrate the pulsed region. Bar in (J) indicates unit of measure in (E), (F), (H), and (I), 20 μm; bar in (J) indicates unit of measure in (G) and (J), 10 μm. (K) Cumulative distribution of turning angles for growth cones loaded and unloaded with caged-Ca2+. Vertical dashed line represents the average turning angle of growth cones loaded with caged-Ca2+ but not exposed to UV light. Neurons in (C) and (E) through (J) were grown on 50 μg/ml LN in 2 mM Ca2+. (G) and (J), n = 10.

Because different concentrations of TN may affect axon growth in ways unrelated to filopodial Ca2+ transients, we focally released caged-Ca2+ within filopodia on a homogeneous substrate to test whether local differences in filopodial transient frequencies are sufficient to produce turning (29). Growth cones loaded with caged-Ca2+ (NP-EGTA) or unloaded controls were positioned so their forward projecting filopodia extended into a 20-μm spot that was pulsed with UV light (310 to 410 nm) for 200 ms at 10-s intervals. Because nonphotolyzed NP-EGTA chelates spontaneous transients due to its low dissociation constant (K d) (80 nM) (30), Ca2+transients were generated only in those filopodia projecting into the spot (Fig. 4, C and D). After 5 to 10 μm of growth, the direction of axon outgrowth was deflected by 37° ± 8° (n = 10) in all growth cones loaded with caged-Ca2+, but only by 14° ± 3° (n = 10) in unloaded controls (Fig. 4, E and K) (P < 0.02). These results show that filopodial Ca2+ transients imposed on one side of the growth cone are sufficient to induce turning away from that side.

Our findings indicate that growth cones exhibit spontaneous cytosolic Ca2+ signals over a wide range of spatial and temporal domains that have diverse effects on pathfinding behaviors. The amplitudes of Ca2+ gradients imposed within the palm of growth cones also regulate attractive or repulsive turning (31, 32). The frequency of Ca2+ transients in filopodia depends on substrate type and concentration, underscoring their biological significance. Ca2+ elevations detected in isolated filopodia in contact with guidance cues may be initial signaling events necessary for adhesion (33). Integrin-mediated Ca2+ influx through non–voltage-gated channels seems to contribute to Ca2+ signals that result in decreased filopodial motility, possibly by feedback activation of integrins (13). Although integrins appear to be sufficient to stimulate Ca2+transients, they are not necessary given activation by non–integrin binding substrates. Filopodial Ca2+ transients are similar to other types of elemental signaling processes such as vesicular transmitter release (34) and Ca2+ puffs and sparks (35), because they can be summed to produce global actions on certain substrates. The autonomous effects of local Ca2+ signals on filopodial motility and substrate-specific ability to induce global elevations of [Ca2+]i together provide versatile control of growth cone behavior.

  • * To whom correspondence should be addressed. E-mail: tmgomez{at}

  • Present address: Department of Anatomy, University of Wisconsin Medical School, Madison, WI 53706, USA.

  • These authors contributed equally to this work.

  • § Present address: Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720, USA.


View Abstract

Navigate This Article