Activation of ADF/cofilin mediates attractive growth cone turning toward nerve growth factor and netrin-1

Abstract

Proper neural circuitry requires that growth cones, motile tips of extending axons, respond to molecular guidance cues expressed in the developing organism. However, it is unclear how guidance cues modify the cytoskeleton to guide growth cone pathfinding. Here we show acute treatment with two attractive guidance cues, nerve growth factor (NGF) and netrin-1, for embryonic dorsal root ganglion and temporal retinal neurons, respectively, results in increased growth cone membrane protrusion, actin polymerization, and filamentous actin (F-actin). ADF/cofilin (AC) family proteins facilitate F-actin dynamics, and we found the inactive phosphorylated form of AC is decreased in NGF- or netrin-1-treated growth cones. Directly increasing AC activity mimics addition of NGF or netrin-1 to increase growth cone protrusion and F-actin levels. Extracellular gradients of NGF, netrin-1, and a cell-permeable AC elicit attractive growth cone turning and increased F-actin barbed ends, F-actin accumulation, and active AC in growth cone regions proximal to the gradient source. Reducing AC activity blunts turning responses to NGF and netrin. Our results suggest that gradients of NGF and netrin-1 locally activate AC to promote actin polymerization and subsequent growth cone turning toward the side containing higher AC activity.

INTRODUCTION

Accurate wiring of neural networks requires that axonal growth cones follow paths determined by extrinsic guidance cues (Letourneau and Cypher, 1991; Mitchison and Kirschner, 1988; Song and Poo, 1999; 2001; Tessier-Lavigne and Goodman, 1996). These cues direct growth cones by binding receptors to trigger signaling cascades that produce cytoskeletal modifications (Bentley and O’Connor, 1994; Huber et al., 2003; Kozma et al., 1997; Lin et al., 1994; Lowery and Van Vactor, 2009; Tanaka and Sabry, 1995; Theriot and Mitchison, 1991). Despite the identification of numerous cues and signaling cascades, the roles of downstream actin regulatory proteins are poorly understood (Dent and Gertler, 2003).

Neurotrophins, which affect multiple aspects of neuronal development, are also guidance cues (Casaccia-Bonnefil et al., 1999; Chen et al., 1999; Davies, 2000; Henderson, 1996; Schuman, 1999). Studies in vitro (Gunderson, 1985; Gunderson and Barrett, 1979; Letourneau, 1978) and in vivo (Glebova and Ginty, 2004; Hassankhani et al., 1995; Menesini Chen et al., 1978; Patel et al., 2000) show that the neurotrophin nerve growth factor (NGF) is an attractive cue for sensory and sympathetic neurons and is necessary for target innervation. Netrins are another well-characterized group of guidance cues that induce attraction or repulsion depending on receptor expression (Kennedy et al., 1994; Moore et al., 2007; Round and Stein, 2007).

It is proposed that attractive cues, such as NGF and netrin-1, favor actin stabilization and polymerization in the growth cone proximal side (Dent et al., 2004; Gallo and Letourneau, 1997, 2000; Lebrand et al., 2004; Lin and Forscher, 1993; O’Connor and Bentley, 1993; Quinn et al., 2008; Sabry et al., 1991; Seeley and Greene, 1983). However, the actin-binding proteins that mediate this actin reorganization are poorly understood. Actin depolymerizing factor and cofilin (AC) family proteins remodel actin filaments by enhancing assembly/disassembly dynamics (reviewed by Andrianantoandro and Pollard, 2006; Chan et al., 2009; Sarmiere & Bamburg, 2003). AC is inactivated by phosphorylation at a conserved serine3 residue by multiple kinases, including LIM kinases 1 and 2 (Arber et al., 1998; Toshima et al., 2001; Yang et al., 1998), and is activated by dephosphorylation via multiple phosphatases (Endo et al., 2003; Huang et al., 2008; Nagata-Ohashi et al., 2004; Niwa et al., 2002; Ohta et al., 2003). Furthermore, AC proteins are inhibited by phosphatidylinositol phosphates in a pH-dependent manner (Frantz et al., 2008).

Several studies suggest that AC may mediate some axonal guidance. AC proteins have roles in axon extension (Bamburg and Bray, 1987; Birkenfeld et al., 2001; Kuhn et al., 2000; Meberg and Bamburg, 2000; Meberg et al., 1998), and are implicated in regulating filopodial responses to neurotrophins (Chen et al., 2006; Gehler et al., 2004). Moreover, a direct role for AC in growth cone guidance was recently shown with Xenopus laevis spinal neurons’ response to bone morphogenic proteins (Wen et al., 2007). In carcinoma cells, AC activation stimulates local actin polymerization during chemotaxis (Ghosh et al., 2004; Mouneimne et al., 2004; 2006).

Here we report that NGF (sensory neurons), and netrin-1 (retinal ganglion cells) stimulate plasma membrane protrusion and actin polymerization in growth cone leading margins, while increasing active AC. Direct increases in active AC also increase protrusion and growth cone F-actin levels. Local increases in active AC within the growth cone induce attractive turning, and reducing AC activity blunts turning to NGF or netrin-1. These data suggest that two important attractive cues promote growth cone turning by activating AC and inducing actin polymerization in the growth cone region proximal to the attractive cues.

METHODS

Materials

F-12 medium, B27 additives, poly-D-lysine (MW >300,000), jasplakinolide, Alexa Fluor 488 DNase1, Alexa Fluor 488- and 568–phalloidin, Alexa Fluor 488 and 568 secondary antibodies were purchased from Invitrogen. NGF, recombinant chick netrin, and L1–Fc were purchased from R & D Systems. Fc Fragment was purchased from Jackson Immuno Research. Chariot was purchased from Active Motif. White leghorn fertilized chicken eggs were purchased from Hy-Line North America, LLC. Cytochalasin D and all other reagents were purchased from Sigma-Aldrich, unless otherwise indicated.

Neuronal culture

Glass coverslips were coated with 100ug/ml poly-D-lysine, rinsed with water, coated with 1% nitrocellulose dissolved in 100% amyl acetate (Fischer Scientific), dried, and coated overnight with 4 μg/ml recombinant L1–Fc mixed with 8 μg/ml Fc in phosphate buffered solution (PBS; Roche). Video dishes were made by gluing a coverslip (18 mm × 18 mm; Gold Seal) over a hole (5mm) in the bottom of a culture dish (Falcon 35 mm × 10 mm), allowed to dry, rinsed with water, and coated as described. Embryonic day 7 (E7) dorsal root ganglia (DRG) and temporal retina were removed from chick embryos according to procedures approved by the University of Minnesota Institutional Animal Care and Use Committee. Neural tissues were cultured on experimental substrates in F-12 with B27 additives and buffered to pH 7.4 with 10 mM HEPES. Neural tissues were cultured overnight in a humidified incubator at 37°C.

Xenopus laevis embryos were obtained as described previously (Gomez et al., 2003) and staged according to Nieuwkoop and Faber (1994). For spinal cord cultures, neural tubes were dissected from stage 22 embryos and explants were cultured in a 1x modified Ringer’s solution. Explants were plated on acid-washed glass coverslips coated with 25 μg/ml laminin (Sigma) as described previously (Gomez et al., 2003). Cultures were fixed 12–16 h after plating with 4% Paraformaldehyde in Krebs+Sucrose Fixative (4% PKS) for 15 minutes and rinsed with PBS.

Preparation of NGF-coated beads

NGF and BSA were covalently coupled to 6μm carboxylated beads (Polysciences, Inc) as described previously (Gallo et al., 1997).

Neuronal transfection

DRG and retinal tissue was dissociated as described in Roche et al., 2009. Approximately 2 × 10⁶ cells were transfected with one of the following: 1 ug of plasmid co-expressing GFP and shRNA against human cofilin (control), 1 μg of a plasmid encoding the fluorescent F-actin probe, green fluorescent protein (GFP)–utrophincalponin homology domain (UtrCH) kindly provided by Dr. W. M. Bement (University of Wisconsin, Madison, WI) (Burkel et al.,2007); 2 ug constitutively-active LIMK with 1 ug GFP; or 1ug ADF shRNA using the G-13 program of the Amaxa Biosystems Nucleofector, and the chicken neuron nucleofector reagents. Cells were cultured 24–48 h, as indicated.

RNA interference (RNAi) was performed using a DNA vector to express small hairpin RNAs (shRNA) from a polymerase III promoter. Chick ADF siRNA oligonucleotides were designed using established criteria based on a corresponding sequence used successfully in mouse (Garvalov et al., 2007; Hotulainen et al., 2005) and inserted into the pSuper expression vector (Brummelkamp et al., 2002) containing the H1 polymerase III promoter. The sequence used in this study is 5′-GTGGAAGAAGGCAAAGAGATT -3′. The H1 pol III-shRNA cassette was then subcloned into pAdtrack plasmids which co-express GFP from a separate promoter, aiding the identification of transfected cells.

Time-lapse microscopy and quantitative fluorescence determinations

A Spot digital camera mounted on an Olympus XC-70 inverted microscope, and MetaMorph software (Molecular Devices) were used for all image acquisitions. In any one experiment, all images were acquired in one session. For collection of fluorescent images, exposure time and gain settings on the digital camera were kept constant, and image acquisition and analysis was performed as described in Roche et al., (2009). In MetaMorph, a line tool was used to outline the terminal 25 μm of each distal axon and growth cone, and background intensity value was subtracted from the fluorescence intensity value of the accompanying neuronal measurement.

Video and kymograph analysis

E7 DRG or retinal explants were cultured 24 h, or transfected cells were cultured 48 h on L1-coated video dishes. A growth cone was selected, and imaged at 10 s intervals. When a cue reagent was added (40ng/ml NGF or Chariot+XACA3 for DRGs, or 500ng/ml netrin for retina), a blank frame was taken to mark the time of addition (as seen as a black line in the kymograph). The video was then opened in ImageJ and three one pixel-wide lines were drawn perpendicular to the growth cone leading edge: one in line with the neurite axis (labeled 0°), and one on either side of the 0° line, at a 30° angle (labeled −30° and 30°). The angle of the 30° lines was taken at the approximate growth cone center at the video start. Using the “stacks → reslice” function, a kymograph was generated and enlarged for better visualization.

Retrograde flow measurements

To determine retrograde actin flow rates, DRGs were transfected with GFP-actin, cultured overnight on L1, and imaged every 3 s for 3 min before and after the addition of 40 ng/ml NGF. Kymographs were generated for the distal 10 μm of growth cones (n=8), and flow rates were measured by tracking bright GFP-actin features, which are formed by unequal incorporation of GFP- and non-GFP actin monomers into polymerized filaments at the leading edge. Measuring retrograde flow using the movement of these features has been described previously, including a demonstrated sensitivity of this retrograde flow to the myosin-II inhibitior blebbistatin (Chan and Odde, 2008).

Immunocytochemistry

Neuronal cultures were fixed and blocked as described in Roche et al., 2009. Coverslips were incubated with primary antibodies diluted in PBS containing 10% goat serum overnight at 4° C. Affinity purified antibody 12977 (ADF) was used at 1:200 dilution; 4321 (phospho-ADF/cofilin) and 3981 (XAC1) were used at 1:1000 dilution. Because both 12977 and 4321 are rabbit antibodies, we compared staining between groups of neuronal growth cones for which immunostaining and image acquisition parameters were kept identical. Coverslips were then rinsed 3 times in PBS and incubated in PBS rinse for 1 h. For labeling F-actin, Alexa Fluor 568–phalloidin was applied at a 1:20 dilution, and mixed with secondary antibodies: Alexa Fluor 568 goat anti-rabbit or anti-mouse antibodies at 1:1000 dilutionin PBS with 10% goat serum for 1 h. For labeling G-actin, 25 ng/ml Alexa Fluor 488 DNase1 was applied to coverslips for 1 h. For staining total ADC in growth cones subjected to gradients, a pipette coated with NGF, netrin, or BSA was presented to one side of growth cones for 5 min, pipette was removed and incubated 1 min in permeabilization buffer (described below) with 100nM phalloidin, fixed and stained.

Antibodies against γ-actin (7577) and β-actin pAb 2963 were generated in the lab of James Ervasti (University of Minnesota) and were kind gifts; β-actin antibody A1978 was purchased from Sigma and FITC-conjugated AC15 was purchased from Abcam. Generation of γ-actin Ab 7577 was described previously (Hanft et al, 2006) and was used at 1:100 dilution, and pAb 2963 was raised against the unique N-terminal peptide sequence of beta-actin and affinity purified against platelet actin as described in Prins et al., (2008), and was used at 1:25, For Xenopus and chick cultures, coverslips were incubated overnight at 4° C, rinsed in PBS, incubated 1 h in secondary. All staining, rinses, and image acquisitions were done in parallel. After rinsing 3 times in PBS, coverslips were mounted in SlowFade reagent (Invitrogen).

Pharmacological Inhibitors

Cultures were treated with 20 nM cytochalasin D (Invitrogen), 40 nM jasplakinolide, or 500 nM K252a (Calbiochem) for 3 min prior to NGF addition. Controls were treated with the same volume DMSO vehicle, which never exceeded 5 μl/ml.

Barbed end labeling

This protocol was adapted from Chan et al., (1998). In summary, neurons were cultured overnight and treated as indicated. At room temp, the culture media was gently aspirated and a permeabilization buffer (1% BSA, .025% saponin, 138 mM KCL, 10 mM Pipes, 0.1 mM ATP, 3 mM EGTA, 4 mM MgCl2, pH = 6.9) with 100nM Alexa-fluor 488 phalloidin (Invitrogen) was added for 1 min, after which the buffer was aspirated and buffer containing 0.45 uM Rhodamine non-muscle actin (Cytoskeleton) was added for 4 min, after which the buffer was removed and cells were fixed with 4% paraformaldehyde and .05% glutaraldehyde, rinsed, and imaged.

Recombinant proteins and protein loading

Recombinant XAC proteins were generated as described previously (Gehler et al., 2004). Proteins were delivered into cells using Chariot reagent (Active Motif; Morris et al., 2001), according to the manufacture’s instructions. Briefly, 6 μl Chariot was complexed with 1 μg BSA, XACWT, or XACA3 for 1 h, then added to the culture medium or immobilized onto a nitrocellulose-coated micropipette.

Growth cone turning assay

Turning was assessed as described previously (Roche et al., 2009). Briefly, micropipette tips were dipped in a 1% nitrocellulose solution, dried, then dipped several times in a PBSsolution of 1 μg/ml NGF, BSA, XACA3, Chariot+XACA3, Chariot+BSA, or 10ug/ml netrin (retina). A growth cone was imaged at 30 or 60 s intervals for 15 min. The micropipette tip was thenpositioned 100 μm from a growth cone at a 45° angle to the direction of axon elongation. Images were acquired 45 min after introducing the micropipette tip. NGF or Chariot+XACA3 was added to the culture medium at the beginning of some of the videos. Growth cone turning angles were determined as the change in direction of growth cone migration between the beginning and end of the image acquisition period (Ming et al., 1997).

Effects of Chariot-A3 gradient on GFP–UtrCH distribution in DRG growth cones

This was performed as described previously (Roche et al., 2009).

Statistical analysis

Parameters of population values are reported as mean +/− SEM. All statistical analysis was by unpaired Student’s t test, Mann-Whitney U test, or ANOVA for multiple comparisons.

RESULTS

Nerve growth factor and netrin-1 induce growth cone leading edge protrusion

Guidance cues regulate growth cone motility. To assess the effects of attractive cues on protrusive activity, we cultured embryonic day 7 (E7) chick dorsal root ganglion (DRG) or temporal retina explants overnight on L1CAM, then imaged growth cones at 10 sec intervals for 5 min before and 5 min after adding 40ng/ml NGF (for DRGs) or 500ng/ml netrin (for retina) to the medium (Lebrand et al, 2004). We then generated kymographs of motility at 3 locations at the growth cone leading edge, in line with the neurite axis (0°) and at 30 degrees to either side of the axis (−30° and 30°). As seen in the kymographs, membrane protrusion from DRG and retinal growth cones increased within seconds of guidance cue addition (Figure 1A, B).

Figure 1.

Nerve growth factor and netrin-1 induce leading edge protrusion. Embryonic day 7 (E7) DRG (A, C) or temporal retinal explants (B, D) were cultured overnight on L1CAM. A growth cone was imaged every 10 s for 5 min before and after adding 40ng/ml NGF (A) or 500ng/ml netrin-1 (B). Kymographs at three locations were generated using a line perpendicular to growth cone leading edge: one in line with the neurite axis (0°) and 30° to either side (−30° and 30°). In panels C and D kymographs were quantified to track position of the leading edge before and after addition of NGF (C) or netrin (D), and averaged across 5 growth cones per condition. Scale bars, 10μm.

To better quantify this membrane protrusion we plotted the leading edge position in pixel numbers relative to the time of cue addition. These spatial positions were then averaged across five growth cones for each condition. As shown in Figure 1C and D, membrane protrusion is stimulated at all 3 kymograph locations at the growth cone leading edge following NGF or netrin addition.

Nerve growth factor and netrin-1 increase growth cone F-actin and free barbed ends

Since growth cone membrane dynamics depend on the underlying actin cytoskeleton, we next examined the changes in actin organization that accompany this increased membrane protrusion. We visualized monomeric actin (G-actin), using fluorescein-conjugated DNase I (Barak et al., 1981), and fluorescent phalloidin-stained filamentous actin (F-actin) in growth cones treated with control media, NGF (DRG), or netrin (retina). This method was validated using cytochalasin D and jasplakinolide, agents that prevent and promote actin polymerization, respectively (Figure S1). Fluorescence intensities were quantified by tracing growth cones, subtracting background, and determining the mean growth cone integrated pixel intensity. Compared to controls, DRG growth cones treated with NGF for 5 min had a relative decrease in G-actin and increase in phalloidin-stained F-actin (Figure 2A, C). Similar changes were seen in temporal retina growth cones treated 5 min with netrin (Figure 2B, D). These data show an increased ratio of phalloidin to DNase I labeling (for DRGs, −NGF=0.85 ±0.07, +NGF=2.38 ±0.37; for retina, −netrin=3.96±0.42, +netrin=8.42±1.70), suggesting a shift from G-actin to F-actin in growth cones responding to NGF or netrin.

Figure 2.

Nerve growth factor and netrin-1 increase growth cone F-actin, and NGF increases free F-actin barbed ends. DRG (A, C) or retinal explants (B, D) were treated 10 min with 40ng/ml NGF (A) or 500ng/ml netrin (B), then fixed. Quantitative fluorescence measurements of growth cone F-actin (Alexa Fluor 568-phalloidin; red) and G-actin (fluorescein-conjugated DNase1; green) were performed. Panels C and D show integrated growth cone fluorescence intensity of F-actin and G-actin in experimental populations. Panel E shows DRG growth cones stained to label F-actin (Alexa Fluor 488-phalloidin, green) or free actin barbed ends (Rh-actin, red). Panel F shows integrated rhodamine-actin growth cone intensity. Statistical significance using Student’s t test. Data are means ± SEM; *P<0.05; ***P<0.0001. Scale bars, 10μm.

An increase in F-actin could result from decreased actin filament depolymerization and/or an increase in actin polymerization. To assess the number of F-actin barbed ends available for polymerization, we adapted a method for the first time to primary neurons in which cells are briefly permeabilized with saponin and then rhodamine-actin is added to polymerize the free F-actin barbed ends (Chan et al., 1998). DRG growth cones treated 15 min with 40ng/ml NGF before permeabilization showed significantly more rhodamine-actin addition to free F-actin barbed ends at the leading edge compared to growth cones of neurons treated with medium alone or grown overnight in 40ng/ml NGF (Figure 2E, F). Both cytochalasin D (20nM), which caps barbed ends, and K252a (500nM), an inhibitor of Trk receptors, blocked the NGF-induced increase in F-actin barbed ends available for rhodamin-actin.

The amount of F-actin at the leading growth cone margin is increased by actin polymerization and decreased by retrograde actin transport. Thus, a decrease in retrograde actin flow might contribute to the net F-actin increase that is induced by NGF. We therefore measured retrograde flow at growth cone leading margins by tracking the movement of bright F-actin features in kymographs of DRG neurons transfected to express GFP-actin (Chan and Odde, 2008). We found that retrograde actin flow was not significantly different after adding 40 ng/ml NGF (10. 5 ± 0.6 μm/min; n=9) than before NGF addition (11.4 ± 0.5 μm/min; n=9) (Figure S2). Together, these data suggest that NGF and netrin increase F-actin levels in DRG and retinal growth cones, respectively by stimulating F-actin polymerization.

Gradients of nerve growth factor and netrin-1 locally increase F-actin and free barbed ends

We next tested whether an increase in F-actin occurs locally in growth cones encountering extracellular gradients of NGF and netrin. Gradients were created by coating a micropipette tip with a protein solution and then placing the pipette tip near a growth cone (Roche et al., 2009). Figure S3A shows that a gradient of rhodamine-BSA was rapidly released from a nitrocellulose-coated micropipette tip, was established at 1 min, was sustained at 5, 15, and 15 min, and extended beyond the 100 μm distance from the pipette tip to a theoretical growth cone. Producing gradients this way, DRG growth cones turn toward an NGF source (Figure 3A, 7B, 8B) and retinal growth cones turn toward a netrin source (Figure 8C; Movie S1).

Figure 3.

Growth cones exposed to a gradient of NGF or netrin-1 exhibit asymmetrical F-actin free barbed end distribution. DRG growth cones turn toward NGF (A), data is taken from panel 7B of this paper. Panel B depicts method used to quantify F-actin (in 3C, D, 6B) distribution across growth cones by measuring integrated intensities in 20×20 pixel-wide regions. Micropipette releasing protein is located to the right side of growth cones. For panel C, a micropipette releasing BSA, NGF, or netrin is introduced to one side of a DRG or retinal growth cone for 5 min, fixed, stained for F-actin (Alexa-488 phalloidin), and pseudo-colored. Distribution was quantified by taking the integrated intensities proximally (region 3) and distally (region 1), relative to the growth cone center (region 2). Panel E shows phalloidin (green) and barbed end (red) labeling on growth cones subjected to a gradient of NGF (DRG), netrin (retina) or BSA for 5 min. Barbed end labeling distribution was performed taking the Rh-actin intensity linescan along growth cone membrane border, dividing in half, and computing the integrated ratio of proximal/distal intensity. Statistical significance using Student’s t test. Data are means ± SEM; *P<0.05; **P<0.001. Scale bars, 10μm.

Figure 7.

Growth cones exposed to a gradient of cell-permeable AC exhibit attractive turning. Panel A shows time-lapse images of a DRG and a retinal growth cone turning toward a gradient of chariot-A3 diffusing from the micropipette. Time shows minutes relative to addition of the micropipette. Panel B shows average turning angles of DRG growth cones to various gradients. Statistical significance using Student’s t test. Data are means ± SEM; *P<0.05; **P<0.001. Scale bars, 20 μm

Figure 8.

Reducing AC activation or protein levels blocks NGF- and netrin-induced membrane protrusion and guidance. Panel A shows DRGs that were transfected with GFP-RNAi control, GFP + CA-LIMK, or ADF-RNAi and cultured 48 h on L1. Transfected growth cones were imaged at 10 s intervals for 5 min before and after the addition of 40ng/ml NGF, and kymographs of the leading edge were generated. Quantification of membrane dynamics were performed as in Fig. 1 (n=5). DRG (B) or retinal neurons (C) were transfected with GFP or GFP + CA-LIMK, cultured 48 h, and subjected to a gradient of NGF (B) or netrin-1 (C), and mean turning angles were measured. In B, non-transfected DRG growth cones were treated with 500nM Trk inhibitor K252a 3 min prior to NGF-pipette introduction. Statistical significance using Student’s t test. Data are means ± SEM; *P<0.05; **P<0.001. Scale bars, 10 μm.

To assess the effects of guidance cue gradients on growth cone F-actin distribution we fixed growth cones 5 min after the introduction of BSA-, NGF-(DRG), or netrin-(retina) coated micropipettes, then labeled F-actin with fluorescent phalloidin. To quantify asymmetries in F-actin distribution, we measured the integrated phalloidin intensity in 20×20 pixel-wide boxes in three regions across the growth cone width, and calculated the distal (region1/region2) and proximal (region3/region2) intensity ratios relative to the center (Figure 3B). Pseudo-colored images of fluorescent phalloidin staining intensity in growth cones indicated that F-actin content is higher in growth cone sides proximal to an NGF (DRG) or netrin (retina) source, but this asymmetry was not present in growth cones exposed to a BSA gradient (Figure 3C). Quantification of this phalloidin staining further indicated higher F-actin levels in regions proximal to the NGF or netrin source, but not a BSA source (Figure 3D). These data support the hypothesis that netrin and NGF gradients induce increased F-actin in growth cone regions closer to the attractive cue source.

To determine if local increases in free F-actin barbed ends also occur, we performed our barbed end labeling assay on growth cones subjected to a NGF (DRG) or netrin (retina) gradient for 5 minutes. Growth cones from DRGs had increased barbed ends proximal to the NGF source, where no such asymmetry occurred with a BSA gradient (Figure 3E). A similar increase was also found in retinal growth cones subjected to a netrin gradient. Because recent studies in Xenopus laevis growth cones have shown local protein synthesis plays a role in growth cone guidance (Leung et al., 2006; Yao et al, 2006), we bath-applied the protein synthesis inhibitor cycloheximide at 20μM (see image in panel E) five minutes prior to introduction of the NGF gradient, and found this local increase in F-actin barbed ends still occurred. The majority of barbed end labeling was at the leading edge, so to quantify the distribution we traced a 5 pixel-wide line from the distal base of the growth cone along the leading margin to the proximal base, and obtained an intensity profile (Figure 3E). This intensity line was then divided, and the pixel intensities along the line segments proximal and distal to the cue were summed. We found the proximal/distal ratio of barbed end staining intensity was significantly higher in DRG growth cones exposed to an NGF gradient versus a BSA gradient (Figure 3E). Together, these data suggest NGF (DRG) and netrin (retina) locally stimulate actin polymerization proximal to the cue source.

ADF/cofilin family proteins are activated by NGF and netrin

We next investigated the role of AC family proteins in this guidance cue-induced F-actin increase. AC proteins are inactivated by phosphorylation at serine 3. Upon dephosphorylation, AC proteins bind to and modify actin filaments. To determine if acute NGF addition affects the level of active AC, we measured the levels of total chick ADF and inactive phosphorylated ADF/cofilin (pAC) in growth cones via immunocytochemistry, using FITC as a control for total protein. Previous work has shown that expression of ADF in chick brain greatly exceeds (about 10 fold) expression of cofilin (Devineni et al., 1999). Compared to untreated growth cones, 5 min treatment with NGF had no significant effect on the total amount of ADF in growth cones, but did significantly lower the total levels of phospho-AC (Figure 4A). Similarly, 5 min treatment of retinal neurons with netrin significantly reduced total phospho-AC levels but not total ADF (Figure 4B).

Figure 4.

ADF/cofilin family proteins are activated by NGF and netrin-1. DRG (A) or retinal explants (B) were treated with medium or 40ng/ml NGF (A) or 500ng/ml netrin-1 (B) for 5 min, fixed and stained for chick ADF (ab 12977) or phospho-ADF/cofilin (pAC; ab 4321), and FITC (total protein). Mean growth cone integrated intensity values were normalized to FITC. Panel C shows DRG growth cones incubated with BSA- or NGF-conjugated beads (pseudo-colored red from phase image) for 1 h, fixed, and stained for total protein (FITC), and pAC (green). Average intensities shown for growth cone regions under BSA or NGF beads, as well as regions of the same growth cone not under a bead, and normalized to FITC. Panel D shows pAC and total ADF (ab 12977) staining after growth cones were subjected to a NGF (DRG), netrin (retina) or BSA gradient for 5 min, permeabilized for 1 min with phalloidin, then fixed and stained. Quantifications were performed as in Figure 3B, and experiments were done on three separate days. Statistical significance using ANOVA. Data are means ± SEM; *P<0.05. Scale bars, 10 μm.

Having found that global addition of NGF reduces the content of inactive phospho-AC in DRG growth cones, we wanted to know if a local NGF source stimulates local AC activation. To do so we examined the levels of phospho-AC in regions of growth cones that touched a bead with surface-bound NGF (Gallo and Letourneau, 1998; 2004). DRG cultures were incubated with 6 μm beads conjugated with BSA or NGF for 60 minutes, then fixed and stained for phospho-AC, FITC (total protein), and fluorescent phalloidin binding (F-actin). Compared to growth cone regions that contacted a BSA bead or growth cone regions that did not contact NGF- or BSA- beads, there was a significantly lower phospho-AC level in growth cone regions that contacted an NGF bead (Fig 4C).

Despite the above demonstrations that phospho-AC levels are reduced by global NGF or netrin and reduced near contact with NGF-coupled beads, we were not able to detect local phospho-AC staining differences in growth cones exposed to gradients of NGF or netrin. Several reasons may account for this. Compared to conditions where small growth cone regions were contacting a local and stable NGF source presented on a bead (Fig 4C), the gradient of NGF across the growth cone width using the micropipette is comparatively shallow. Furthermore, the fixation of a relatively small (20kDa) and soluble protein such as phospho-AC, which does not bind F-actin, may not be rapid enough to preserve its distribution in this shallow gradient, and these small differences may be undetectable using our immunocytochemical methods. Although Wen et al (2007) demonstrated asymmetric phospho-AC distribution in Xenopus growth cones responding to BMPs, these growth cones turned away from regions of higher AC activity, suggesting the gradient of pAC may be steeper in this repulsion behavior compared to a growth cone turning toward higher AC activity. Furthermore, Xenopus growth cones have less actin compared to our chick growth cones (Fig 10), which suggests differences in actin regulation exist between the two species.

Figure 10.

Chick DRG growth cones contain higher levels of actin compared to Xenopus spinal neuronal growth cones. Cultured chick DRG and Xenopus spinal neurons were labeled with antibodies against γ-actin (ab 7577), β-actin (AC15, Abcam; A1978 Sigma), actin (2963, James Ervasti), Alexa-Fluor 488 DNase (green), or Alexa-fluor 568 phalloidin (red). Where primary antibodies were not conjugated to FITC (γ-actin, A1978, 2963), FITC was used as a control for total protein. Integrated growth cone intensities were measured and normalized to chick. Statistical significance using Student’s t test. Data are means ± SEM; *P<0.05; **P<0.001; ***P<0.0001. Scale bars, 10 μm.

However, the asymmetric labeling of barbed ends in this gradient (Fig 3E) suggests asymmetric AC activation does occur. Thus, in order to probe localized AC activity in our conditions used for growth cone turning, we exposed growth cones to a NGF or netrin gradient for 5 min, and applied our phalloidin-containing permeabilization buffer for 1 minute prior to fixation. Under these conditions, all phospho-AC staining was lost, presumably because inactive phospho-AC is not bound to F-actin and diffuses from the growth cone upon permeablilization. We then stained for total ADF, assuming that active ADF would be bound to F-actin, and not lost upon permeabilization. Under these conditions, we found significantly higher ADF staining proximal to the NGF (DRG) or netrin (retina) source, with no such asymmetry in growth cones exposed to a gradient of BSA (4D). Together, these data suggest that NGF and netrin activate AC in DRG and retinal growth cones, respectively, in both a global and localized manner.

Increased AC activity results in increased F-actin and barbed ends

Results thus far suggest that increased AC activity contributes to the NGF- and netrin-1 – induced increase in growth cone F-actin. To test this, we used the Chariot reagent, which forms cell-permeable complexes with proteins, to load purified recombinant Xenopus ADF/cofilin into our cultured neurons. Chariot-mediated incorporation of these proteins into growth cones was confirmed immunocytochemically, using an antibody against Xenopus AC (XAC1) that has little cross-reactivity with chick AC (Shaw et al, 2004). Compared to chick DRGs incubated 1 hour with Chariot alone or wild-type Xenopus AC (XACWT) not complexed with Chariot, growth cones treated with Chariot-XACWT had significantly more XAC1 staining (Fig 5A). Thus, using this method we loaded DRG and retinal neurons with Chariot complexes of XACWT or a “constitutively-active” XAC with a serine3 to alanine mutation that renders it non-phosphorylatable (XACA3; Gehler et al., 2004). We assessed G- and F-actin levels as before, and found in both neuronal types that the Chariot-mediated incorporation of XACWT or XACA3 increased phalloidin-stained F-actin levels and reduced DNase I staining of G-actin (Figure 5B, C). The addition of NGF to XAC-loaded DRG growth cones further increased F-actin levels, presumably through activation of endogenous AC. Netrin addition to XAC-loaded retinal growth cones did not significantly alter F-actin levels, which could be due to actin properties in retinal growth cones that limit the extent of F-actin increase upon XAC-loading and netrin addition. In support of this, the magnitude of XAC-induced F-actin increase in retinal neurons was smaller compared to DRGs. Taken together, these data demonstrate that directly elevating AC activity increases F-actin in DRG and retinal growth cones.

Figure 5.

Increased AC activity results in increased F-actin, decreased G-actin, and increased F-actin barbed ends. One μg of XACWT or XACA3 in PBS was complexed with Chariot reagent in dH20 for 30 min. Panel A shows chick DRG explants treated 1h with chariot, XACWT alone, or chariot+XACWT, and fixed and stained with the anti-Xenopus AC antibody XAC1. The leftmost pair of images show a growth cone of a Xenopus spinal neuron stained robustly with XAC1. The histogram shows integrated XAC1 growth cone intensities for chick DRGs shown. For panels B–D, Chariot alone or with recombinant proteins added to DRG (B and D) or temporal retina (C) explants 1 h before addition of control media, 40ng/ml NGF (B, D) or 500ng/ml netrin (C) for 15 min. Panels B and C show growth cones fixed and stained for F-actin (Alexa Fluor 568-phalloidin, red) and G-actin (fluorescein- DNase1, green), and mean growth cone fluorescence intensities are expressed normalized to Chariot, media controls. Panel D shows DRG growth cones that were permeabililzed with Alexa Fluor 488-phalloidin for 1 min (green), then incubated 4 min with 0.45 μM rhodamine-G actin (red) before being fixed and imaged. Mean growth cone fluorescence intensities of phalloidin staining and Rh-actin were determined. Statistical significance using ANOVA are indicated with “*”, significance using Student’s t test are indicated with “†”. Data are means +/− SEM; *P<0.05; **P<0.001; ***P<0.0001. Scale bars, 10 μm.

Because the severing activity of AC proteins can increase the number of F-actin barbed ends available for polymerization, we assessed if incorporation of AC into growth cones increases F-actin barbed ends. We treated DRG growth cones 1 h with Chariot alone or Chariot complexed with XACWT or XACA3 and then assayed for F-actin barbed ends. Compared to controls, XACWT- or XACA3-treated growth cones had significantly higher levels of free barbed ends (Figure 5D). In addition, NGF treatment of XACWT- or XACA3-treated neurons further increased barbed end labeling, likely due to activation of endogenous chick AC. In addition, the morphology of NGF-treated growth cones was larger and more flattened than growth cones incubated with Chariot-XAC alone (Fig 5D), suggesting that NGF induces additional changes that contribute to the change in growth cone shape.

Taken together, these data show that direct elevation of active AC increases F-actin levels in DRG and retinal growth cones, mimicking the effects seen with NGF and netrin. Furthermore, the direct elevation of AC also increases F-actin barbed ends in DRG growth cones.

Gradients of cell-permeable active AC induce local increases in F-actin and attractive growth cone turning

Our data show global and local addition of the guidance cues NGF and netrin increase growth cone F-actin and AC activity, and directly increasing AC activity also increases growth cone F-actin. We next examined growth cone turning behavior, F-actin distribution and AC asymmetry in growth cones responding to gradients of cell-permeable Chariot-XACA3. Figure S3B shows that Chariot-rhodamine-BSA complexes diffused similarly from micropipette tips as rhodamine-BSA (Figure S2A). Gradients of Chariot-XACA3 induced attractive turning of both DRG and retinal growth cones (Figure 7A, movies S2 and S3), while growth cones did not turn in a Chariot-BSA gradient (Figure 7B). Importantly, growth cones also did not turn in a gradient of extracellular XACA3 that was not complexed with Chariot, indicating that growth cone turning is induced by an intracellular, but not extracellular, asymmetry of AC activity. To show that this method does produce an intracellular asymmetry in XAC, we fixed growth cones 15 min after introduction of a micropipette coated with Chariot-XACWT or XACWT alone, then stained the growth cones with an antibody selective for Xenopus AC (XAC1). Quantification of growth cone XAC1 staining distribution showed that growth cones exposed to a gradient of XACWT without Chariot had no asymmetry in XAC1 staining, whereas growth cones subjected to a Chariot-XACWT gradient had higher XAC1 staining proximal to the micropipette (Figure 6A). These results indicate that growth cones turn toward the side with higher intracellular AC activity. To determine if this local increase was associated with locally increased F-actin, DRG growth cones were subjected to pipette tips coated with Chariot-XACA3 or XACA3 alone for 15 min, fixed and then stained with fluorescent phalloidin. Similar to our results with an NGF or a netrin gradient, fluorescent phalloidin staining was more intense in the growth cone regions proximal to a Chariot-XACA3 source but not a source of XACA3 alone (Figure 6B).

Figure 6.

Growth cones exposed to a gradient of XACA3 exhibit asymmetrical F-actin distribution. For panel A, DRG growth cones were subjected to XACWT or chariot complexed with XACWT diffusing from a micropipette for 15min, fixed and stained with XAC1, and quantified as in Fig 3B. Data was taken from three separate days. Panel B shows F-actin (Alexa Fluor 488-phalloidin) distribution (pseudo-colored) in DRG growth cones subjected for 15 min to XACA3 or chariot+XACA3 diffusing from a micropipette, and quantified as in Figure 3B. Panel C shows distribution of GFP-UtrCH or GFP in transfected DRG growth cones that were exposed at one side to a pipette tip releasing NGF (data as reported in Roche et al, 2009), chariot-A3 or chariot-BSA complexes. An image was taken before and 5 min after pipette introduction. The distribution of fluorescence is expressed as the intensity along a 15 pixel-wide line across the growth cone width. Statistical significance using ANOVA. Data are means +/−SEM; *P<0.05. Scale bars, 10 μm

F-actin distribution in live cells can be visualized by transfection to express an F-actin binding domain of utrophin conjugated to GFP (GFP-UtrCH; Burkel et al., 2007). To examine F-actin distribution in real time we transfected DRG neurons with GFP-UtrCH, and as reported in Roche et al. (2009), the GFP-UtrCH fluorescence intensity was higher in the proximal side of growth cones at two min after introducing an NGF gradient (6C; Roche et al, 2009). A BSA gradient produced no asymmetry in the distribution of GFP-UtrCH intensity. We used this approach to confirm results based on phalloidin staining that F-actin is more concentrated in the growth cone region proximal to a chariot-XACA3-releasing pipette tip (Figure 6B). DRG neurons were transfected to express GFP-UtrCH, and we assessed F-actin distribution in live growth cones that were subjected to a gradient of Chariot-XACA3. A fluorescence image was taken before and 5 min after introducing a Chariot-XACA3-coated pipette tip to one side of a growth cone. To analyze the response, a 15 pixel wide line was drawn across a growth cone and the “linescan” function was used to generate fluorescence intensity measurements across the growth cone width (in Metamorph). To quantify GFP-UtrCH distribution, the proximal/distal intensity ratios of growth cones before and after introducing the pipette were computed. For GFP-UtrCH expressing growth cones exposed to a Chariot-BSA gradient, this ratio was 1.12±0.05 SE (n=9), and for growth cones exposed to Chariot-XACA3 the ratio was 1.40±0.09 (n=10), which was significantly different from growth cones exposed to Chariot-BSA (p<0.05) (Figure 6C). As a control for volumetric changes after introducing the Chariot-XACA3 gradient, we measured fluorescence intensity across GFP only-transfected growth cones. The proximal/distal GFP ratio of 0.90±0.09 (n=8) was not significantly different from GFP-UtrCH transfected growth cones exposed to a Chariot-BSA gradient, suggesting that the increase in proximal UtrCH intensity induced by the Chariot-XACA3 gradient was not due to a local increase in volume.

Taken together these studies indicate that gradients of NGF, netrin and Chariot-XACA3 induce attractive growth cone turning and increased F-actin distribution in growth cone regions proximal to the pipette tips. In addition, F-actin is more concentrated in growth cone regions containing higher AC activity. This is consistent with our earlier findings (Figure 4C) that staining for phospho-AC (inactive AC) is lower in growth cone regions contacting NGF-beads than in regions free of NGF-beads or in regions contacting BSA-beads, as well as the increased permeabilization-resistant ADF in growth cone regions proximal to a NGF or netrin gradient (Figure 4D).

Our evidence suggests that NGF induces attractive DRG growth cone turning by locally activating AC, which stimulates local actin polymerization in the growth cone region proximal to NGF. In support of this, we showed that a gradient of cell-permeable AC induces growth cone turning with both increased AC content and increased F-actin in the growth cone side toward the AC source. To further examine the role of asymmetric AC activity in mediating growth cone turning, we introduced an NGF gradient to growth cones in the presence of globally applied Chariot-A3 to raise AC activity throughout the growth cone and diminish NGF-induced asymmetry in AC activity (Figure 7B). We found the average turning angle to NGF was reduced compared to NGF turning in the absence of Chariot-A3, although the difference was not statistically significant (p=0.099). In like manner, the mean turning angle to a Chariot-A3 gradient was also reduced by global elevation of NGF, compared to A3 turning in the absence of NGF (but not statistically significant, p=0.23). In both cases, NGF gradient with global Chariot-A3 and Chariot-A3 gradient with global NGF, the average turning angle was significantly greater than turning toward a Chariot-BSA gradient (p<0.05). These data suggest that a small asymmetry in ADF/cofilin activity across a growth cone can elicit attractive turning, but the reduced turning angles observed with diminished AC asymmetry provides further support that turning toward positive cues is mediated by intracellular asymmetry of AC activity. Furthermore, activation of additional proteins downstream of NGF signaling, including paxillin (Khan et al, 1995), integrins (Grabham and Goldbert, 1997), and microtubule plus end binding protein APC (Zhou et al, 2004) also likely affect growth cone trajectories in addition to AC activation.

Reducing AC activation or protein levels blocks NGF- and netrin-induced growth cone protrusion and turning

If local AC activation is necessary for attractive turning toward NGF and netrin, we would expect that blocking AC activation will block attractive responses to these cues. To test this we transfected neurons with either a constitutively-active LIMK construct (CA-LIMK) to maintain high levels of phosphorylated inactive AC, with a plasmid expressing GFP and a hairpin RNA that generates a small interfering siRNA against chick ADF, or with a GFP- control RNAi plasmid. As expected, neurons transfected with CA-LIMK and cultured 24hrs had higher levels of phospho-AC compared to GFP-transfected controls (Figure S4A). Neurons transfected with ADF-RNAi for 48hrs showed reduced ADF immunostaining, compared to GFP- control RNAi transfected cells or non-transfected cells (Figure S4C), and Western blot analysis showed ADF protein level was reduced in ADF-RNAi transfected chick embryo fibroblasts (Figure S4B).

We then examined growth cone responses to NGF in transfected DRG growth cones. As expected, in GFP-control RNAi DRG growth cones, leading edge protrusion increased immediately after adding NGF. Remarkably, growth cones expressing CA-LIMK or ADF RNAi had no protrusion increase following NGF addition (Figure 8A). Furthermore, compared to GFP controls, CA-LIMK expressing DRG neurons showed reduced attractive turning to NGF (Figure 8B), and the turning of retinal growth cones toward a netrin-coated micropipette was also reduced in CA-LIMK expressing retinal neurons (Figure 8C). Bath application of K252a, an inhibitor of the NGF receptor TrkA, also abolished turning toward NGF. Although the growth cones of CA-LIMK transfected DRG neurons advanced slower than GFP controls (1.4 +/− 0.04 μm/min for GFP, 1.18 +/− 0.16 μm/min for CA-LIMK), growth cones were observed 60 min, long enough to advance more than 60μm and complete a turning response. These data suggest that AC activation by NGF or netrin is necessary for membrane protrusion and attractive turning to NGF or netrin.

Restoring active AC to CA-LIMK and ADF RNAi growth cones rescues membrane protrusion

To determine if the block of NGF-induced membrane protrusion was specifically caused by the reduced AC activity, we added Chariot+XACA3 protein to GFP-control RNAi, CA-LIMK, or ADF RNAi-expressing growth cones, and examined leading edge dynamics. As seen in Figure 9, the addition of Chariot-XACA3 proteins increased growth cone protrusion in DRG neurons transfected with either of the three plasmids. These data support the notion that the reduction in NGF-induced protrusion seen in CA-LIMK and ADF RNAi-expressing growth cones is due to reduced AC levels and not to other factors associated with the constructs.

Figure 9.

Restoring active AC to CA-LIMK- and ADF-RNAi-expressing growth cones rescues membrane protrusion. DRG neurons were transfected with GFP-RNAi control, GFP + CA-LIMK, or ADF-RNAi, and cultured 48hr on L1. Transfected growth cones were imaged every 10 sec for 10 min before and after the addition of chariot complexed with XACA3. Quantification of membrane dynamics were performed as in Fig. 1. Scale bars, 10 μm.

Actin protein levels are higher in chick DRG growth cones compared to Xenopus laevis spinal neuronal growth cones

Our data suggest a local increase in AC activity can induce growth cone turning toward the side with higher activity. This finding is in direct contrast with that of Wen et al. (2007), where Xenopus spinal neuronal growth cones responding to bone morphogenic protein 7 (BMP7) turned toward the side with lower AC activity and away from the side with higher AC activity. Recent studies have found differing effects on actin organization at different ratios of AC to actin (Andrianantoandro and Pollard, 2006; Blanchoin and Pollard, 1999; Chan et al., 2009; Chen et al., 2004; Pavlov et al., 2007). We therefore sought to compare actin protein levels in chick and Xenopus spinal neuronal growth cones. Using immunocytochemistry, we stained chick DRG and Xenopus spinal neuron cultures simultaneously, using multiple antibodies against β- and γ-actin. The very high sequence conservation for β- and γ-actin among vertebrates suggests that these antibodies should recognize these actin isoforms similarly in frog and chick neurons. The integrated (sum) intensity was measured and normalized to total protein (FITC), except with FITC-conjugated AC15. Measurements of integrated growth cone intensities indicated lower levels of both γ-actin and β-actin in Xenopus growth cones compared to chick (Figure 10). These differences remained consistent with multiple antibodies, and when measured as a ratio to total protein using FITC (Figure 10). Additionally, we found Xenopus growth cones had higher levels of DNase and lower levels of phalloidin staining, suggesting Xenopus growth cones have a higher G-actin:F-actin ratio compared to chick DRG growth cones.

We do not know the relative concentrations or basal activity of AC in chick versus Xenopus, nor possible differences in additional actin-binding proteins. Although further research is needed to determine if these actin differences are sufficient to account for these disparate results, it is not unreasonable to suggest that basal differences in actin expression could affect the net response of growth cones responding to guidance cues. For example, recent studies have shown local protein synthesis is important in growth cone guidance (Campbell and Holt, 2001; Wu et al., 2005; Piper and Holt, 2004), and specifically, local translation of β-actin mRNA in Xenopus growth cones (Leung et al., 2006; Yao et al, 2006). However, we recently reported normal growth cone guidance with chick and mouse neurons in the presence of protein synthesis inhibitors (Roche et al, 2009). These data suggest that in growth cones containing less actin, guidance cue-induced local protein synthesis may be more essential for growth cone turning compared to growth cones with higher actin levels, where reorganization of existing actin is sufficient. In the present study, these relative differences in actin levels and organization could affect how activated AC modifies actin filament dynamics.

DISCUSSION

Here we have shown that two attractive cues, NGF and netrin, increase leading edge protrusion in DRG and retinal growth cones, respectively, through increases in F-actin and actin polymerization. These guidance cues decrease inactive (phosphorylated) AC levels in growth cones. Direct elevation of AC activity in DRGs also increases free actin barbed ends and F-actin levels. A diffusible gradient of NGF, netrin or cell-permeable AC stimulates proximal F-actin assembly in growth cones, and elicits attractive turning. Reducing AC activity or protein level blocks guidance cue-induced membrane protrusion and blunts guidance cue-induced attractive turning for DRG and retinal growth cones. These data support a model in which NGF and netrin stimulate attractive turning through local activation of AC, which promotes actin polymerization in the growth cone side toward the guidance cue.

The regulation of actin organization and dynamics by guidance cue signaling has been a focus of growth cone research (for review, see Lowery and Van Vactor, 2009; Pak et al., 2008). Applied globally, repulsive cues reduce growth cone F-actin and protrusion, while attractive cues promote protrusion and increased F-actin (Brown and Bridgman, 2009; Fan et al., 1993; Gallo and Letourneau, 2000; Kalil and Dent, 2005; Roche et al, 2009; Zhou and Cohen, 2003). Specifically, global application of NGF and netrin-1 have been linked to increases in F-actin (Seeley and Greene, 1983; Shekarabi and Kennedy, 2002; Goldberg et al., 2000; Lebrand et al., 2004). More akin to guidance situations, local application of NGF and netrin, were shown to locally promote protrusive activity and increased F-actin (Dent et al., 2004; Gallo et al., 1997; Gallo and Letourneau, 1998; 2000; Chang et al., 2006), and recent studies have shown gradients of netrin-1 and brain-drived neurotrophic factor induce proximal increases in β-actin synthesis in Xenopus growth cones (Leung et al., 2006; Yao et al., 2006). However, in the nearly two decades of work linking F-actin increases to attractive cues, we understand relatively little about the mechanisms underlying these changes. And despite the identification of numerous actin-binding proteins with diverse effects on actin organization, links between specific proteins downstream of specific guidance cues are not well characterized (Dent and Gertler, 2003; Huber et al., 2003; Lowery and Van Vactor, 2009; Pak et al., 2008; Suter and Forscher, 1998).

In this paper we looked more closely at the role of actin polymerization in growth cone responses to attractive guidance cues. We quantified the increased F-actin in growth cones exposed to globally applied NGF and netrin and the distribution of increased F-actin in growth cones exposed to gradients of these cues. To probe the specific changes in actin organization responsible for these increases, we measured, for the first time, the effects of guidance cues on F-actin barbed ends, which are the sites of actin polymerization. We found that barbed ends are increased throughout growth cones by global NGF exposure and locally in growth cone margins that are proximal to gradients of NGF (DRG) or netrin (retina). We also measured retrograde actin flow, which if reduced, could contribute to increased F-actin in the growth cone leading margin. We found that NGF exposure did not change the retrograde actin flow rate, suggesting increased actin polymerization is the primary mechanism for the increased F-actin induced in the growth cone leading margin by attractive cues. In migratory cells, similar local stimulation of leading edge actin polymerization is critical for cell polarity and directed migration to chemoattractants (Charest and Firtel, 2007; Pollard and Borisy, 2003; Weiner et al., 1999).

Our results indicate that local activation of the actin regulatory protein ADF/cofilin (AC) is a necessary component of attractive growth cone guidance. Previous work showed that NGF and the related neurotrophin BDNF activate AC in conjunction with neurotrophin-induced increases in motility and filopodial dynamics (Chen et al., 2006; Gehler et al., 2004; Meberg et al., 1998). Similarly, we found that global addition of NGF or netrin reduced total growth cone levels of inactive phospho-AC, and regions of DRG growth cones contacting NGF-coupled beads also had reduced phospho-AC. When growth cones were exposed to NGF (DRG) or netrin (retina) gradients and permeabilized before fixation to release inactive phospho-AC, subsequent immunostaining for remaining ADF bound to the actin cytoskeleton was concentrated in the growth cone region toward the attractive cue, further indicating that AC activity was higher proximal to an attractant. When we reduced AC activity by RNAi or by expressing constitutively active LIMK to inactivate AC by phosphorylation, growth cones failed to respond to attractive guidance cues, indicating the necessity of AC activity for attractive turning to NGF or netrin.

Similar to the effects of NGF and netrin exposure, we found the content of F-actin barbed ends in DRG and retinal growth cones was increased by incorporation of cell-permeable AC, and when DRG or retinal growth cones were subjected to a gradient of cell-permeable AC, the asymmetric incorporation of AC resulted in both increased F-actin and attractive growth cone turning toward the source of cell-permeable AC. This demonstrates the sufficiency of elevated AC activity for barbed end creation, actin polymerization, and attractive growth cone turning, and is similar to the role of AC in promoting actin polymerization at the front of chemotaxing carcinoma cells (Ghosh et al., 2004; Mouneimne et al., 2004; 2006; Zebda et al., 2000).

Biochemical studies have revealed the complexity of AC in regulating actin filament turnover (for review, see Carlier et al., 1997, Van Troys et al., 2008). Some studies suggest AC activity enhances monomer loss from F-actin pointed ends, thereby replenishing the actin monomer pool (Carlier et al. 1997; Cramer, 1999; Maciver et al., 1998), while other studies attribute increased actin polymerization to AC-mediated severing of F-actin that increases barbed ends for monomer addition (Du and Frieden, 1998; McGough et al. 1997). The ratio of AC:actin is another factor in actin dynamics. At low AC levels relative to actin, cofilin binds actin independently, promoting filament turnover and polymerization at new barbed ends. (Andrianantoandro and Pollard, 2006). At intermediate concentrations, cofilin binds actin cooperatively and rapidly severs actin filaments (Andrianantoandro and Pollard, 2006; Blanchoin and Pollard, 1999; Chan et al., 2009; Pavlov et al., 2007). At high relative concentration cofilin can both nucleate and rapidly sever actin but then stabilizes filaments as they become saturated with cofilin (Andrianantoandro and Pollard, 2006; Chen et al., 2004; Pavlov et al, 2007). Beyond these studies with purified components, the net roles of AC become more complex and varied within cells, where multiple factors, including the G-actin pool size, components that regulate AC activity, and the actions of additional actin regulatory proteins are integrated in determining actin filament organization and dynamics in migrating cells (Bernstein and Bamburg, 2010; Ressad et al., 1999; Sarmiere and Bamburg, 2003).

Such complexity in regulating actin filaments may indicate why our result concerning AC activity and growth cone turning differs from that of Wen et al. (2007), where Xenopus growth cones responded to BMPs by turning toward the side with lower AC activity. We probed other differences between Xenopus and chick growth cones and found Xenopus growth cones have lower levels of actin and lower F-actin/G-actin ratios compared to chick growth cones, which may impact how actin is re-organized when AC activity increases. An additional consideration is that BMP-receptor signaling may involve different pathways and activities than what is activated downstream of netrin-1 and NGF. For example, netrin-mediated actin polymerization requires the anti-capping activity of Ena/VASP (Gitai et al., 2003; Gomez and Robles, 2004; Lebrand et al., 2004). The GTPase Rac, which is activated by both NGF and netrin (Gitai et al., 2003; Ridley et al., 1992), activates SCAR to enhance the actin nucleating activity of the Arp2/3 complex (Machesky et al., 1999). The actin nucleating and F-actin branching actions of the Arp2/3 complex may interact with the filament severing and debranching actions of AC in a variety of ways to regulate actin organization during lamellipodial and filopodial protrusion (Chan et al., 2009; Blanchoin et al., 2000; Ichetovkin et al., 2002; Ressad et al., 1999; Svitkina and Borisy, 1999). Further, the relative activities of F-actin capping and actin-sequestering proteins directly control the fundamental elements of actin polymerization, free barbed ends and G-actin monomer. Thus, the integration of all actin regulatory activities may dictate that different levels of AC activity should exist in different growth cones to mediate attractive growth cone turning. In Xenopus growth cones increased filopodial protrusion and F-actin accumulation may require low AC activity, while in chick growth cones, increased AC activity promotes actin polymerization to drive protrusion and attractive turning to NGF and netrin.

This paper does not address the mechanisms by which NGF and netrin regulate AC activity, which is becoming a more complex topic (Bernstein and Bamburg, 2010). NGF- and netrin-induced dephosphorylation of AC could occur through activation of a phosphatase or a decrease in kinase activity. Several phosphatases activate AC, including chronophin (Gohla et al., 2005), and isoforms of slingshot, which was recently implicated in carcinoma cell chemotaxis (Eiseler et al, 2009; Niwa et al., 2002). Inactivation of AC is mediated by LIM and TES kinases (Toshima et al., 2001; Yang et al., 1998). LIMK is active when phosphorylated by PAK1, PAK4, or Rho-kinase (ROCK), and is inhibited by dephosphorylation by SSH phosphatase or ubiquitination by Rnf6 (for review, see Bernard, 2007). In addition, several feedback loops may play important roles, such as SSH activation by F-actin, or LIMK inactivation by SHH. Additional regulators of AC include 14-3-3 proteins, which promote inactive phospho-AC by directly binding to phospho-AC and LIMK (Gohla and Bokoch, 2002), and incorporation of 14-3-3 into retinal growth cones blocks BDNF-mediated activation of AC (Gehler et al., 2004). AC activity is also regulated by pH, phosphoinositides, and other proteins such as tropomyosin or Aip1 (for review, see Sarmiere and Bamburg, 2003). In summary, multiple regulators affect AC activity.

In summary, our data support a model in which two important cues, NGF and netrin, induce attractive growth cone turning by locally activating AC and stimulating actin polymerization in growth cone regions proximal to the guidance cue source. Further studies are needed to clarify how the activities of multiple actin-binding proteins and associated regulatory factors are integrated to mediate the effects of guidance cues on growth cone trajectories.

Supplementary Material

Supp Fig s1. Figure S1.

FITC-conjugated DNase reliably measures G-actin levels. DRG explants were cultured overnight on L1 and treated 10 min with control media, 20nM cytochalasin D (CytD), or 40nM Jasplakinolide (jasp) and fixed. Immunochemistry against F-actin (Alexa Fluor 568-phalloidin; red) and G-actin (fluorescein-conjugated DNase1; green) was performed. DNase 1 intensity increases and decreases with CytD and jasp, respectively. Phalloidin intensity does not increase with jasp, possibly because phalloidin and jasp compete for the same F-actin binding site. Statistical significance using Student’s t test. Data are means ± SEM; *P<0.05.

Supp Fig s2. Figure S2.

NGF does not significantly alter retrograde actin flow. DRGs were transfected with GFP-actin and growth cones were imaged every 3 s for 3 min before and after 40n/ml NGF addition. Flow rates before and after NGF were calculated by tracking bright actin features, as shown in sample kymographs.

Supp Fig s3. Figure S3.

A micropipette coated with nitrocellulose and protein establishes a diffusible gradient. A micropipette was coated with nitrocellulose and 1 μg/ml rhodamine-BSA (A) or rhodamine-BSA complexed with Chariot reagent (B) and lowered into a dish with medium. Fluorescent images were taken after 1, 5, 10, and 15 min. The concentric gradients were measured using the “linescan” function in Metamorph. Insets show the gradients between 68–102μm from the micropipette tip.

Supp Fig s4. Figure S4.

Expressing constitutively-active LIMK (CA-LIMK) or ADF hRNA results in decreased AC activity. A) Dissociated DRGs were transfected with GFP or GFP + CA-LIMK, cultured 24 hr, fixed and stained with pAC. Bar graph shows mean growth cone pAC intensities in GFP- and CA-LIMK-transfected growth cones. B) Western blot of extracts from chick embryo fibroblasts infected with hRNA-expressing adenoviral vectors for the time indicated before preparing cell extracts. Control vector expresses a rat hRNA that does not silence the chick protein. Two different loading volumes of the 120 hr extracts are shown for the chick ADF silencing vector. Infection efficiency for these chick fibroblasts is about 80% (based on GFP co-expression from the virus) suggesting silencing at 72 hr is about 65% and at 120 hr is about 80%. C) DRGs transfected with ADF hRNA, cultured 48hr, then fixed and stained with pAC (red). Note the level of anti-pAC staining of the adjacent non-transfected cell. Scale bars, 10 μm. Statistical significance using Student’s t test. Data are means ± SEM; *P<0.05.

Supp Movie s1. Movie S1.

A nitrocellulose-coated micropipette incubated with netrin-1 induces attractive turning when lowered to one side of an extending retinal growth cone. Video shows one frame every 2 min, for a total duration of 78 min.

Supp Movie s2. Movie S2.

A nitrocellulose-coated micropipette incubated with Chariot-A3 complexes induces attractive turning when lowered to one side of an extending DRG growth cone. Video shows one frame every 2 min, for a total duration of 38 min.

Supp Movie s3. Movie S3.

A nitrocellulose-coated micropipette incubated with Chariot-A3 complexes induces attractive turning when lowered to one side of an extending retinal growth cone. Video shows one frame every 2 min, for a total duration of 64 min.