Abstract
Axons navigate through the embryo to construct a functional nervous system. A missing part of the axon navigation puzzle is how a single axon traverses distinct anatomic choice points through its navigation. The dorsal root ganglia (DRG) neurons experience such choice points. First, they navigate to the dorsal root entry zone (DREZ), then halt navigation in the peripheral nervous system to invade the spinal cord, and then reinitiate navigation inside the CNS. Here, we used time-lapse super-resolution imaging in zebrafish DRG pioneer neurons to investigate how embryonic axons control their cytoskeleton to navigate to and invade at the correct anatomic position. We found that invadopodia components form in the growth cone even during filopodia-based navigation, but only stabilize when the axon is at the spinal cord entry location. Further, we show that intermediate levels of DCC and cAMP, as well as Rac1 activation, subsequently engage an axon invasion brake. Our results indicate that actin-based invadopodia components form in the growth cone and disruption of the invasion brake causes axon entry defects and results in failed behavioral responses, thereby demonstrating the importance of regulating distinct actin populations during navigational challenges.
SIGNIFICANCE STATEMENT Correct spatiotemporal navigation of neuronal growth cones is dependent on extracellular navigational cues and growth cone dynamics. Here, we link dcc-mediated signaling to actin-based invadopodia and filopodia dynamics during pathfinding and entry into the spinal cord using an in vivo model of dorsal root ganglia (DRG) sensory axons. We reveal a molecularly-controlled brake on invadopodia stabilization until the sensory neuron growth cone is present at the dorsal root entry zone (DREZ), which is ultimately essential for growth cone entry into the spinal cord and behavioral response.
Keywords: DREZ, invasion, neuron, pathfinding, zebrafish
Introduction
Throughout the nervous system, neurons extend axons that migrate long distances and surpass distinct tissue environments to reach their final destination. Nearly 100 years ago, Ramón y Cajal observed growth cones that must steer these axons. Further work has determined that growth cones of the first axons in a tract, pioneer axons, form more complex morphologies as compared with later-developing axons that aid in navigational decision-making (Cajal, 1911; Sotelo, 2003). Since Cajal's observations, countless molecules have been identified as important for axon navigation. However, how axons employ distinct populations and arrangements of cytoskeletal components in their growth cones to bypass specific and various embryonic environments during their expedition still needs more understanding. We approach this gap by investigating the navigation of dorsal root ganglia (DRG) pioneer axons.
Generally, growth cones are composed of actin filaments that can fluidly be recycled to change actin state, ultimately allowing for growth cone motility. Actin-binding proteins (ABPs) regulate organization and polymerization of actin filaments for development of filopodia, lamellipodia, and invadopodia. These distinct actin populations then dynamically work together in the growth cone during navigation. Filopodia at the tip of the growth cone extend or retract with lamellipodia to achieve proper pathfinding (Shekarabi and Kennedy, 2002; Mattila and Lappalainen, 2008; Norris and Lundquist, 2011; Demarco et al., 2012), microtubules at the base of the growth cone interact with F-actin to employ growth cone steering (Dent and Kalil, 2001), and invadopodia are employed to invade across boundaries (Santiago-Medina et al., 2015; Nichols and Smith, 2019a,b). While a litany of axon guidance molecules have been identified through investigation of an axon's final path (Chédotal, 2019), how distinct actin populations are molecularly controlled during growth cone movement remains a missing but critical link to understanding how axons navigate through the vertebrate embryo. To date, this link has largely been gathered through two-dimensional in vitro models, which especially poses an issue for actin structures that form in three-dimensional space, like invadopodia (Santiago-Medina et al., 2015; Nichols and Smith, 2019a,b). The gap in our understanding of how the axon moves through the complicated and diverse embryonic tissue is underscored by recent discovery investigating 3D navigation (Ming et al., 1997; Norris and Lundquist, 2011; Santiago-Medina et al., 2015; Akin and Zipursky, 2016; Santos et al., 2020).
Using the zebrafish DRG, we studied the molecular control of filopodia and invadopodia in the growth cone as axons navigate. The DRG pioneer axons navigate dorsally in the periphery until they reach the dorsal root entry zone (DREZ) where they encounter and invade the glial limitans that form the spinal cord boundary, thereby experiencing three different tissue environments during development (Golding et al., 1997; Nichols and Smith, 2019a). Using time-lapse imaging, we reveal an essential invasion brake that inhibits actin stabilization of invadopodia until the pioneer axon approaches the DREZ. During this process, transient invadopodia form during filopodia-based navigation. Invadopodia then stabilize at the DREZ, forming a coordinated invadopodia state to ultimately allow for growth cone entry into the spinal cord. We define a role of dcc, Rac1 and cAMP in reducing transient invadopodia in the growth cone until the axon reaches the DREZ, ultimately leading to coordinated invadopodia stabilization and growth cone entry into the spinal cord. Further, experimental perturbation of these molecules leads to a deficient behavioral response to sensory stimuli. These findings fill the gap in knowledge of how axons molecularly balance modes of navigation from filopodia-based movement to invadopodia-mediated growth cone entry along their pathfinding journey.
Materials and Methods
Experimental model and subject details
All animal studies were approved by the University of Notre Dame Institutional Animal Care and Use Committee. Zebrafish strains were as follows: Tg(sox10:meGFP) (Smith et al., 2014), Tg(gfap:nsfb-mcherry) (Johnson et al., 2016), Tg(gfap:gfp) (Bernardos and Raymond, 2006), Tg(sox10:gal4) (Hines et al., 2015), Tg(uas:lifeact-gfp) (Helker et al., 2013), Tg(tnfa:gfp) (Marjoram et al., 2015), Tg(ngn1:gfp) (McGraw et al., 2008), dcczm130198 (Jain et al., 2014), Tg(sox10:lifeact-gfap);dcczm130198+/−, Tg(gfap:nsfb-mcherry);dcczm130198+/−, and Tg(tnfa:gfp);dcczm130198+/−.
Embryos were produced from pairwise mating and kept at constant darkness in 28°C in egg water (Kimmel et al., 1995). Embryos were staged by hours postfertilization (hpf) or days postfertilization (dpf), and either sex were used for all experiments.
In vivo imaging
Embryos were manually dechoriantated at 48 hpf and exposed to 3-amino-benzoic acid ester (Tricaine) for anesthetization. Embryos were immersed in 0.8% low-melting agarose and mounted laterally on their right side on the glass cover of a 35-mm Petri dish. Images were acquired on a custom built spinning-disk confocal microscope as previously described (Nichols et al., 2019b). Time-lapse images were taken every 5 min for 24 h starting at 48 hpf. Super-resolution, DeSOS, images and movies were acquired as described previously (Zhang et al., 2019). The only enhancement to presented images were changes to brightness and contrast.
Whole-mount larval zebrafish RNAScope
The following probes were used: ntn1b (1:50, 80 µl, C2, ACD) and dcc (1:50, 80 µl, C1, ACD). Protocol was adapted from Nichols et al. (2019b). Larvae were fixed in 4% PFA at 25°C for 30 min, then transferred to a new Eppendorf tube and dehydrated with 25%, 50%, and 100% methanol at 10 min each. Larvae were left in 100% methanol at −20°C overnight and then rehydrated with 50% and 25% methanol in PBSTw (PBS, 0.1% Tween 20) at 10 min each. Next, larvae were air dried for 30 min, then washed twice in PBSTw for 5 min. Larvae were then permeabilized in proteinase K (10 mg/ml, 1:2000 in PBSTw) at 25°C for 6 min and immediately washed three times with PBSTw for 10 min at 25°C. Larvae were then incubated with the probe (two drops of C1 or 80 µl of C2/C3 in diluent buffer) at 40°C overnight. After incubation, larvae were washed twice with SSCTw (saline-sodium citrate buffer, 0.1% Tween 20) and then washed with 4% PFA for 10 min at 25°C. Larvae were transferred to a new Eppendorf tube and washed three times with SSCTw for 10 min each at 25°C. Next, larvae went through a series of incubations in two drops of Amp1 at 40°C for 30 min, two drops of Amp2 at 40°C for 30 min, two drops of Amp3 at 40°C for 15 min, two drops of HRP-C1/2/3 at 40°C for 30 min, opal fluorophore 640 (1:500) at 40°C for 30 min, and two drops of Multiplex FLv2 HRP blocker at 40°C for 30 min. Larvae were washed twice with SSCTw at 25°C for 10 min between each incubation. Fish were then incubated in two drops of DAPI at 4°C, washed twice with SSCTw at 25°C for 10 min, and finally stored in 50% glycerol and stored in 4°C until imaging.
Photoactivatable Rac1
A working solution of a tol2-4xnr UAS-PA Rac1-mcherry plasmid (Addgene plasmid #41 878) was diluted to 12 ng/µl. An injection mix of pa-Rac1 with 25 ng/µl of transposase mRNA was injected into single-cell Tg(sox10:gal4);Tg(UAS:lifeact-gfp) animals. At 48 hpf, animals were screened for mcherry expression in DRG neurons on the confocal. Embryos positive for mcherry were either not exposed or exposed to 445-nm light to activate Rac1 (Nichols and Smith, 2019a). Only DRG neurons labeled with pa-Rac1-mcherry were scored.
Construction of DCC-tdTomato and testing cell-autonomy
The expression plasmid for DCC-tdTomato was synthesized by Genewiz. The synthesized DNA includes tol2 sites that flank the cDNA for zebrafish dcc that is upstream of tdTomato (Kwan et al., 2007). To generate cell-specific expression, the plasmid was linearized with EcoRV and cell-specific promoters were inserted with NEB HiFi Cloning Enzyme mix. The cell-specific promoter was amplified from p5e-uas construct and inserted into the linearized plasmid (Kwan et al., 2007). The resulting plasmid, psCJS16, was injected into one-cell Tg(sox10:gal4) animals to express DCC-tdTomato in DRG pioneer neurons. Individual DRG neurons that express tdTomato were subsequently used for analysis.
Drug treatments
The chemical reagents used for this experiment were sp-cAMP (Fisher Scientific, catalog #11681510UMO), rp-cAMP (Fisher Scientific, catalog #1168145UMOL), NSC23766 (EMD Biosciences, catalog #553502). Stock solutions of these reagents were kept at −20°C with concentrations of 1 mm (sp-cAMP), 60 µm (rp-cAMP), and 25 µm (NSC23766). Embryos were dechorianted at 36 hpf and primarily treated with 100 µm (sp-cAMP; Muramatsu et al., 2010), 25 µm (rp-cAMP; Ming et al., 1997), or 1 µm (NSC23766; Araiza-Olivera et al., 2018). Rescue experiments in cAMP quantification were treated with both 100 µm (sp-cAMP) and 25 µm (rp-cAMP) at the same time. Co-treatment to test Rac1 and cAMP were treated with 1 µm (NSC23766) and 100 µm (sp-cAMP) at the same time. All control animals were exposed to 1% DMSO in egg water.
Experimental design and statistical analysis
Slidebook software was used to create maximum intensity projections for all images and movies. Quantification of all data were completed using ImageJ. All graphs represent the mean and SEM for all representations of error. Statistical values and n values can be found in each figure and figure legend. Sample numbers were determined based on standards in the field. All statistics comparing two groups were gathered from unpaired t test analysis. All statistics comparing more than two groups were gathered from one-way ANOVA analysis. Data throughout text is reported as the mean ± SEM, followed by type of statistical analysis. Scientists were blind to the experimental versus control group during quantifications. Experiments with dcc mutants were quantified before genotyping, and matching of genotype and phenotype was done before figure generation.
Entry quantifications
Radial glial boundary
Glial limitans were visualized using animals expressing mCherry under the gfap promoter, and DRG neurons were identified using animals expressing meGFP under regulatory regions of the Sox10 promoter (Nichols and Smith, 2019a). Representative images of entry were gathered using IMARIS, and double transgenic animals were rotated 90°. Quantification of entry was determined by using images of double transgenic animals rotated 90° using Slidebook software for a y-orthogonal view to visualize the glial limitans and navigating axons. These images were converted to an RGB-Tiff file and analyzed in ImageJ. The edge of the glial limitans was indicated in the y-plane, and the z-plane was the horizontal plane of the y-orthogonal image through the growth cone. A line was drawn through the growth cone, and intensity of both gfap+ glial limitans and sox10+ pioneer axon was measured over distance and normalized to the background. Normal entry was indicated by a sox10+ pioneer axon peak being present medial to the gfap+ glial limitans peak, indicating the axon has passed through the glial limitans.
Tg(tnfa:gfp) entry quantifications
Entry was also visualized using Tg(tnfa:gfp) as an output (Nichols and Smith, 2019a). Tg(tnfɑ:gfp) animals were used for initial screening and were injected at the one-cell stage with four gRNAs for ntn1b, ntn1a, dcc, or lamb1a (Wu et al., 2018). Tg(tnfɑ:gfp) animals were also bred into a dcc+/− background, and incubated in NSC23766, NSC23766 + sp-cAMP, sp-cAMP, rp-cAMP, or DMSO at 36 hpf. All animals were visualized at 3 dpf using a Zeiss Axiozoom microscope equipped with filter cubes: DAPI, GFP, and RFP, and a monochrome camera run by Zen software. Number of tnfɑ+ DRG neurons was then quantified on ImageJ to indicate entry, and the number was compared for all groups.
Actin quantifications
For all actin quantifications, Tg(sox10:lifeact-gfp) animals were either injected with pa-Rac1, incubated in DMSO, NSC23766, sp-cAMP, rp-cAMP, sp-cAMP + rp-cAMP, pa-Rac1 + rp-cAMP, or were in either a dcc−/− background or dcc+/− background treated with DMSO, sp-cAMP, rp-cAMP, or injected with sox10:dcc. They were then time lapsed in 30-µm stacks at 48 hpf for 24 h. Maximum projections were made to collapse all z-planes unless otherwise noted, and quantifications were performed in ImageJ. In quantifications, only DRG neurons that had all 80 time points present within the time lapse were scored, which equates to one to two DRG per animal.
Filopodia length
Several filopodia projections during the invasion time point were scored in quantifications to ensure a range of lengths were represented. Length was determined in ImageJ by tracing a line from the tip to the bottom of the projection, and measuring µm in the x-y-plane.
Filopodia and invadopodia correlations
A total of 80 time points (400 min) per DRG were used to correlate filopodia number and invadopodia number over time. The point of entry was determined by bifurcation of the axon, and changes were tracked 40 time points before and after entry. Total change in filopodia and invadopodia number per time point was determined by calculating the difference in number (per actin group) between the most recent time point and the preceding time point.
Overall filopodia length was calculated by finding the sum length of each filopodia per time point. Correlation between filopodia length and invadopodia number was calculated by finding the difference in length between the most recent time point and the preceding time point. Invadopodia difference was calculated as described above.
Invadopodia calculations
Y-orthogonal images of Tg(sox10:lifeact-gfp) axons navigating to the glial limitans were generated by digitally rotating images 90° using Slidebook software (Nichols and Smith, 2019a). For invadopodia number quantifications, projections in the z-plane of Lifeact-GFP at the growth cone extending away from the axon and toward the spinal cord were termed invadopodia. Definitions of “transient” versus “stable” were then confirmed by measuring the intensity values of Lifeact-GFP using the ImageJ plugin, MTrackJ, and normalized to the background intensity. The point of entry was determined by bifurcation of the axon, and intensity of the growth cone was tracked 40 time points before and after entry, for a total of 80 time points. These intensity values were then plotted to form the Coordinated Invadopodia State Graphs. Intensity values collected from the coordinated invadopodia state graphs were plotted in an Excel sheet, and the average of the highest and lowest intensity points was calculated for each DRG to normalize each calculation. This average of the highest and lowest Lifeact-GFP intensities was then plotted on the graph to determine a threshold line to determine invadopodia state. Lifeact-GFP intensity peaks below the threshold line were labeled as transient, and Lifeact-GFP intensity peaks above the threshold line were labeled as stable.
Difference graphs
Difference graphs were calculated by creating the coordinated invadopodia state graphs with threshold lines as described above. Each point on the x-axis equates to 5 min, and the first 30 time points (150 min) indicated navigation, and time points 30–50 indicated invasion (100 min). The difference of either transient or stable invadopodia was then scored between navigation and invasion. A positive number indicated a primary role during the navigation phase, and a negative number indicated a primary role during the invasion phase.
Duration of coordinated invadopodia state
Lifeact-GFP intensity values and invadopodia threshold lines were collected and determined from the coordinated invadopodia state graphs as described above, and any peak above the threshold line was considered to be a stable invadopodia involved in coordinated invadopodia state. Each point on the x-axis equates to 5 min, and the width of each of these peaks indicates time in minutes. The combined width of all peaks above the threshold line was then added together to calculate duration of time of coordinated invadopodia state per DRG.
Behavior assay
AB animals were used for initial testing of behavior assay and were raised at 28°C until scored at either 2, 3, 4, or 5 dpf. To assay thermosensitivity and behavior, single animals were removed from 28°C egg water and placed into the test arena with a pipette tip. All egg water was removed from the test arena to reduce temperature fluctuations of the new temperature condition before the immediate gentle addition of temperature conditioned egg water. The initial acclimation of 2 s is not considered for the analysis. Videos to record behavior were assayed from 2 to 60 s after exposure to new temperature conditions. Water temperatures were measured before each animal was placed into the new condition. Recordings were processed through FLOTE software to divide animals into four segments (Burgess and Granato, 2007). Tail curvature magnitude quantified as the sum of angle A, which is measured between the second and third segments, and angle B, which is measured between the third and fourth segments.
For the behavioral assay, AB animals were either incubated in sp-cAMP, rp-cAMP, both, or DMSO at 36 hpf, or were in a dcc+/− background treated with DMSO, sp-cAMP, or rp-cAMP. Animals were raised at 28°C until being assayed as described above. Movies were then analyzed on ImageJ, and the duration of each shiver was calculated by determining the number of frames the shiver lasted. The combined number of frames from each shiver over the total number of frames was then used to calculate the percentage of time shivering.
Results
Filopodia and invadopodia dominate at distinct time points during DREZ assembly
To understand how growth cone actin populations dynamically transition over time, we visualized filopodia and invadopodia in DRG pioneer axon growth cones during pathfinding to the DREZ. We first asked whether both filopodia and invadopodia can form at the same time, or whether each is a distinct binary event. To do this we performed 24-h time-lapse imaging in a super-resolution format (Zhang et al., 2019). We imaged Tg(sox10:lifeact-gfp) animals at 48 hpf to visualize actin populations in DRG growth cones. We then quantified the number of filopodia and invadopodia events for 40 time points before and after entry (Fig. 1A). In these graphs, navigation to the DREZ is represented by the first 150 min, while the entry of the growth cone at the DREZ is represented by the following 100 min. Actin-containing structures were scored as invadopodia if they met all three of the following criteria: (1) bright Lifeact-GFP puncta could be spatiotemporally differentiated from each other; (2) these puncta formed medially-projecting protrusions when examined in 90° rotated z-stacks; and (3) normalized intensity of Lifeact-GFP corresponded with 1 and 2 (Nichols and Smith, 2019a). We observed that filopodia form an average of 14.50 ± 0.68 projections during navigation to the DREZ, while invadopodia form transiently with an average of 1.25 ± 0.37 puncta (p < 0.0001, n = 8 DRG, t test). Conversely, the number of filopodia projections decreases to an average of 3.50 ± 0.93 while at the DREZ, while invadopodia increase in number to an average of 4.25 ± 0.37 puncta (p = 0.1489, n = 8 DRG, t test; Fig. 1B). To test whether these transient actin-based events during navigation are bona fide invadopodia, we used immunohistochemistry on Tg(sox10:lifeact-gfp) animals during navigation to test whether transient actin localized to Cortactin (Fig. 1C). We found significant localization of Cortactin and basally-projecting transient actin-structures, with an average of 66.67% of transient events containing positive Cortactin staining (n = 12 DRG; Fig. 1D). Unlike a complete binary phenomenon, invadopodia and filopodia formed transiently during both navigation to and approachment at the DREZ.
Figure 1.
Filopodia and invadopodia stabilize at distinct time points. A, left, Representative confocal maximum projection of Tg(sox10:lifeact-gfp) from a 24-h time lapse starting at 48 hpf. Right, Graph of filopodia (blue) and invadopodia (green) number over time (n = 4 DRG). White dotted box denotes DREZ insets. White arrow denotes actin concentrates. Solid color lines denote representative graphs. Dashed colored lines denote mean ± SEM. Blue box denotes time at the DREZ. B, Quantification of filopodia and invadopodia before, during, and after entry (n = 8 DRG). C, Representative confocal maximum projection of Tg(sox10:lifeact-gfp) animal at 48 hpf localizing to Cortactin staining. D, Quantification of percent of transient invadopodia during navigation that have positive Cortactin staining (n = 12). E, Tracing of growth cone Lifeact-GFP intensity before, during, and after entry (n = 4 DRG). Blue box denotes time at the DREZ. Light purple arrow denotes transient invadopodia. Dark purple arrow denotes stable invadopodia. Blue line denotes representative graph. Dashed black line denotes mean ± SEM. F, G, Confocal maximum projections of super-resolution Tg(sox10:lifeact-gfp) from a 24-h time lapse starting at 48 hpf. Representative images of filopodia (blue arrows) and invadopodia (green arrows) in the x-y-plane and z-plane during (F) navigation and (G) at the DREZ. H, Quantification of difference in invadopodia number between navigation and time at the DREZ (n = 4 DRG). Positive numbers indicate the dominant role in navigation. Negative numbers indicate the dominant role during time at the DREZ. I, J, Correlation between invadopodia number and (G) filopodia number or (H) filopodia length over time (n = 4 DRG). Scale bars: 10 μm (A, C, F, G). A, B, E, H–J, Data are represented as mean ± SEM.
The discovery that transient invadopodia form even during filopodia-dominant navigation led us to hypothesize that molecular mechanisms could limit the stabilization of invadopodia via a brake-like mechanism until the axon reaches the DREZ. To first explore this possibility, we quantified Lifeact-GFP in the pioneer axon. A few small intensity peaks formed randomly during navigation, which corresponded to transient invadopodia formation. This next transitioned into multiple small peaks within a much larger actin-based growth cone event while at the DREZ. These quantitated individual peaks corresponded with many invadopodia events in the growth cone, suggesting stabilization of invadopodia (Fig. 1E–G). Hereafter, we refer to the spatiotemporal precision of invadopodia stabilization as the coordinated invadopodia state.
To further test the hypothesis that transient invadopodia form during navigation to the DREZ but stabilize while at the DREZ, the difference of transient and stable invadopodia in the growth cone between navigation to the DREZ and while at the DREZ were calculated separately. If more invadopodia form during navigation as compared with growth cone entry into the spinal cord at the DREZ, the difference in invadopodia state will be a positive number. Conversely, if less invadopodia form during navigation as compared with at the DREZ, the difference in invadopodia state will be a more negative number. Positive numbers thus indicate employment of invadopodia during navigation, and negative numbers indicate invadopodia stabilization while at the DREZ. Our findings show invadopodia formation during navigation to the DREZ was predominantly transient, showing a difference of an average of 9.75 ± 0.63 as compared with an average of −5.00 ± 0.82 of stable invadopodia (p < 0.0001, n = 4 DRG, t test; Fig. 1H). These results are consistent with the hypothesis that pioneer axon entry is dependent on the coordinated invadopodia state, specifically at the DREZ.
Since both invadopodia and filopodia are actin-based structures, we hypothesized that as the brake on invadopodia is released at the DREZ, actin is used to produce invadopodia instead of filopodia. There were two most likely possibilities: either abundance of invadopodia comes at the expense of filopodia abundance, or invadopodia stabilization comes at the cost of filopodia length. To dissect these possibilities, we scored and plotted the following changes at each time point over four DRG: invadopodia abundance versus filopodia abundance and invadopodia abundance versus filopodia length. We reasoned that a positive correlation of either graph would identify the more likely mechanism. The correlation between invadopodia abundance versus filopodia abundance was weak (r < 0.001, simple linear regression, n = 4 DRG; Fig. 1I), inconsistent with the possibility that invadopodia stabilization is linked to filopodia genesis. However, invadopodia abundance versus filopodia length showed a strong correlation (r = 0.83, simple linear regression, n = 4 DRG; Fig. 1J), indicating that invadopodia stabilization is likely linked to filopodia length. Knowing that entry also requires a strict coordinated invadopodia state, we next sought molecules that could instruct the invadopodia during navigation to the DREZ versus growth cone entry at the DREZ.
Rac1 inhibits coordinated invasion events
Rac1 drives disassembly of invadopodia (Moshfegh et al., 2014; Nichols and Smith, 2019a). We therefore hypothesized that Rac1 regulates coordinated invadopodia state by driving invadopodia disassembly until the axon reaches the DREZ. To test this, we injected a photoactivatable-Rac1 construct, uas:PA-Rac1-mcherry, into Tg(sox10:gal4)(uas:lifeact-gfp) animals at the one-cell stage. Rac1 was then either photoactivated with 445-nm light before growth cone entry as previously described, or without exposure as a control (Nichols and Smith, 2019a). To test whether Rac1 promotes invadopodia disassembly while the growth cone navigates to the DREZ, we first quantified the number of invadopodia puncta in photoactivated embryos as compared with non-photoactivated embryos. These results showed an average decrease in invadopodia puncta in the photoactivated growth cone, from 3.25 ± 0.25 puncta to 1.50 ± 0.53 puncta (p < 0.0001, n = 8 DRG, t test; Fig. 2A). Given our observation that invadopodia and filopodia are linked, we next hypothesized that activating Rac1 would lead to the actin distribution favoring filopodia length. To address this, we next quantified the length of filopodia while at the DREZ. Activating Rac1 led to an average increase in length from 3.40 ± 0.20 to 4.59 ± 0.27 µm (p = 0.0011, n = 20 filopodia over 5 DRG, t test; Fig. 2B). Taken together, these data are consistent with the hypothesis that Rac1 promotes filopodia length and invadopodia disassembly during navigation, and reverses while at the DREZ.
Figure 2.
Rac1 inhibits coordinated invadopodia state. A, Quantification of invadopodia number during time at the DREZ (n = 8 non-pa-Rac1 DRG; n = 8 pa-Rac1 DRG; n = 5 DMSO DRG; n = 5 NSC23766 DRG). B, Quantification of filopodia length during time at the DREZ (n = 20 non-pa-Rac1 filopodia; n = 20 pa-Rac1 filopodia; n = 16 DMSO filopodia; n = 16 NSC23766 filopodia). C, Quantification of difference of transient invadopodia between navigation and time at the DREZ (n = 5 non-pa-Rac1 DRG; n = 5 pa-Rac1 DRG; n = 4 DMSO DRG; n = 5 NSC23766 DRG). D, Quantification of difference of stable invadopodia between navigation and time at the DREZ (n = 5 non-pa-Rac1 DRG; n = 5 pa-Rac1 DRG; n = 4 DMSO DRG; n = 5 NSC23766 DRG). E, Quantification of duration of coordinated invadopodia state of stable invadopodia (n = 4 non-pa-Rac1 DRG; n = 5 pa-Rac1 DRG; n = 4 DMSO DRG; n = 5 NSC23766 DRG). A–E, Black dots denote non-photoactivated Rac1. Purple dots denote photoactivated Rac1. Gray dots denote DMSO. Magenta dots denote NSC23766. F–I, left, Confocal maximum projection of Tg(sox10:lifeact-gfp) from a 24-h time lapse starting at 48 hpf in a (F) non-photoactivated Rac1, (G) photoactivated-Rac1, (H) DMSO, or (I) NSC23766 condition. White dotted box denotes DREZ insets. White arrow denotes actin concentrates. Right, Tracing of growth cone Lifeact-GFP intensity before, during, and after entry in a (F) non-photoactivated Rac1 (n = 4), (G) photoactivated-Rac1 (n = 5), (H) DMSO (n = 4), or (I) NSC23766 (n = 5) condition. Blue box denotes time at the DREZ. Blue line denotes representative graph. Dashed black line denotes mean ± SEM. Scale bars: 10 μm (F–I). A–I, Data are represented as mean ± SEM.
If Rac1 functions to inhibit invadopodia until the axon reaches the DREZ, activating it should also reduce the temporal domination of transient and stable invadopodia. The difference of transient invadopodia during navigation and while at the DREZ showed an average decrease from 10.60 ± 0.51 to 1.60 ± 1.21, indicating a lack of transient invadopodia during navigation (p = 0.0001, n = 5 non-photoactivated DRG, n = 5 photoactivated DRG, t test; Fig. 2C). Stable invadopodia were also reduced while at the DREZ, with an average increase from −4.20 ± 0.73 to 1.80 ± 0.86 (p = 0.0007, n = 5 non-photoactivated DRG, n = 5 photoactivated DRG, t test; Fig. 2D). These results are consistent with the hypothesis that Rac1 controls the stabilization state of invadopodia during the navigation process to the DREZ.
We next explored that activating Rac1 would lead to an uncoordinated invadopodia state by decreasing invadopodia stabilization. To test this, Rac1 was activated with 445-nm light in 48 hpf Tg(sox10:lifeact-gfp) animals and time lapsed for 24 h. We then measured the intensity of Lifeact-GFP at the growth cone tip 40 time points before and after it was located at the DREZ. In control embryos, measurements of Lifeact-GFP intensity showed a peak of intensity while at the DREZ with multiple small spikes within the large peak, which lasted for an average of 73.75 ± 3.15 min (Fig. 2E,F; Movie 1). This differed from activated animals, in which invadopodia never stabilized to form one large peak, and the time the growth cone spends while at the DREZ decreased to an average duration of 36.00 ± 5.79 min (p = 0.0011, n = 4 non-photoactivated DRG, n = 5 photoactivated DRG, t test; Fig. 2E,G; Movie 2).
Time lapse of control growth cone. Confocal maximum projection of 400-min time lapse of control 48 hpf Tg(sox10:lifeact-gfp) animal. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of photoactivated-Rac1 growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal with photo-activated Rac1. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
We next performed complementary Rac manipulations to spatiotemporally manipulate endogenous Rac1 levels. In addition to modulating actin populations in embryonic animals later in development, Rac1 also regulates directed migration of neural crest cells before 36 hpf (Matthews et al., 2008). We therefore treated Tg(sox10:lifeact-gfp) animals with 1 µm a Rac1 inhibitor, NSC23766, at 36 hpf to circumvent its earlier developmental role in neural crest migration, and then performed 24-h time-lapse imaging at 48 hpf. Actin dynamics were then scored as previously discussed. Rac1 inhibition led to an average increase in invadopodia puncta from 2.60 ± 0.24 (n = 5) to 4.00 ± 0.32 (p = 0.0081, n = 5 DRG, t test; Fig. 2A) and an average decrease in filopodia length from 3.55 ± 0.17 µm (n = 4 DRG; Fig. 2B) to 2.57 ± 0.26 µm (p = 0.0033, n = 16 filopodia over 5 DRG, t test). We next asked whether inhibition of Rac1 resulted in an imbalance of transient and stable invadopodia states during navigation and while at the DREZ. Transient invadopodia had an average change in difference from 9.25 ± 0.25 (n = 4) to 0.20 ± 0.37 (p < 0.0001, n = 5 DRG, t test; Fig. 2C), while stable invadopodia had an average change in difference from −5.00 ± 0.41 (n = 4) to 2.60 ± 0.24 (p < 0.0001, n = 5 DRG, t test; Fig. 2D). This resulted in an uncoordinated invadopodia state with the formation of many stable peaks during both navigation and while the growth cone was at the DREZ (Fig. 2I), culminating in an increase of average duration from 65.00 ± 3.54 min (n = 4) to 173.00 ± 14.46 min (p = 0.0003, n = 5 DRG, t test; Fig. 2E). Together, these data support Rac1 as a driving force in coordinating the brake on invadopodia until the axon reaches the DREZ.
Precise levels of cAMP regulate the invadopodia brake
To better understand the molecular mechanism that controls invadopodia in the growth cone, we next performed a pilot drug screen to identify molecules that could impact growth cone entry into the spinal cord. Previous data has established upregulation of tumor necrosis factor-ɑ (tnfɑ) in pioneer axons at the DREZ, and is not expressed when growth cone entry into the spinal cord does not occur (Smith et al., 2017). We therefore used tnfa expression in DRG neurons as an output for growth cone entry into the spinal cord. In this assay, we count DRG neurons located in all segments on the right side of the animal. Through this we identified that Tg(tnfa:gfp) animals treated with a cAMP agonist of PKA, sp-cAMP, demonstrate a 2+ DRG neurons (Fig. 3A). Given the tnfa assay implicating cAMP/PKA in entry, we next tested the hypothesis that increasing or decreasing cAMP would impede the precise control on invadopodia stabilization. We first hypothesized that increasing cAMP would cause a decrease in invadopodia while at the DREZ, while decreasing cAMP with the rp-cAMP antagonist would lead to the opposite phenotype. To test this, Tg(sox10:lifeact-gfp) animals were treated with either 100 µm sp-cAMP, 25 µm rp-cAMP, or DMSO at 36 hpf. Larvae were then time lapsed for 24 h at 48 hpf, and the number of invadopodia puncta were scored. The number of invadopodia puncta while the growth cone was at the DREZ decreased from an average of 2.67 ± 0.33 puncta in the control group (n = 6 DRG) to 0.57 ± 0.20 and 1.67 ± 0.21 in the sp-cAMP and rp-cAMP groups, respectively (p = 0.0002, n = 7 DRG, t test; p = 0.0296, n = 6 DRG, t test; Fig. 3B). These changes in invadopodia came at the expense of filopodia, with an average increase in length from 3.47 ± 0.21 to 4.28 ± 0.13 µm (sp-cAMP) and 4.44 ± 0.34 µm (rp-cAMP; p = 0.0037, n = 10 filopodia over 5 DRG, t test; p = 0.0237, n = 10 filopodia over 5 DRG, t test; Fig. 3C). These data are inconsistent with a hypothesis that a binary increase or decrease of cAMP controls stabilization of invadopodia at the DREZ, but rather revealed that invadopodia stabilization may be sensitive to intermediate levels of cAMP.
Figure 3.
Precise levels of cAMP regulate the invadopodia brake. A, Images of Tg(tnfa:gfp) animals at 72 hpf treated with DMSO, 100 μm sp-cAMP, or 25 μm rp-cAMP. B, Quantification of invadopodia number during time at the DREZ (n = 6 DMSO DRG; n = 7 sp-cAMP DRG; n = 6 rp-cAMP DRG; n = 5 sp-cAMP + rp-cAMP DRG; n = 4 pa-Rac1 DRG). C, Quantification of filopodia length during time at the DREZ (n = 10 DMSO filopodia; n = 10 sp-cAMP filopodia; n = 10 rp-cAMP filopodia; n = 10 sp-cAMP + rp-cAMP filopodia; n = 20 pa-Rac1 + rp-cAMP filopodia). D, Quantification of difference of transient invadopodia between navigation and time at the DREZ (n = 5 DMSO DRG; n = 5 sp-cAMP DRG; n = 5 rp-cAMP DRG; n = 5 sp-cAMP + rp-cAMP DRG; n = 4 pa-Rac1 + rp-cAMP DRG). E, Quantification of difference of stable invadopodia between navigation and time at the DREZ (n = 4 DMSO DRG; n = 5 sp-cAMP DRG; n = 5 rp-cAMP DRG; n = 5 sp-cAMP + rp-cAMP DRG; n = 4 pa-Rac1 + rp-cAMP DRG). F, Quantification of duration of coordinated invadopodia state of stable invadopodia (n = 5 DMSO DRG; n = 5 sp-cAMP DRG; n = 4 rp-cAMP DRG; n = 5 sp-cAMP + rp-cAMP DRG; n = 4 pa-Rac1 + rp-cAMP DRG). B–F, Black dots denote DMSO. Dark blue dots denote 100 μm sp-cAMP. Green dots denote 25 μm rp-cAMP. Light blue dots denote 100 μm sp-cAMP + 25 μm rp-cAMP. Pink dots denote pa-Rac1 + 25 μm rp-cAMP. G–K, left, Confocal maximum projection of Tg(sox10:lifeact-gfp) from a 24-h time lapse starting at 48 hpf in a (G) DMSO, (H) 100 μm sp-cAMP, (I) 25 μm rp-cAMP, (J) 100 μm sp-cAMP + 25 μm rp-cAMP, or (K) pa-Rac1 + 25 μm rp-cAMP-treated animal. White box denotes DREZ insets. White arrow denotes actin concentrates. Right, Tracing of growth cone Lifeact-GFP intensity before, during, and after entry in a (G) DMSO (n = 4), (H) 100 μm sp-cAMP (n = 5), (I) 25 μm rp-cAMP (n = 4), (J) 100 μm sp-cAMP + 25 μm rp-cAMP (n = 5), or (K) pa-Rac1 + 25 μm rp-cAMP (n = 4) treated animal. Blue box denotes time at the DREZ. Blue line denotes representative graph. Dashed black line denotes mean ± SEM. Scale bars: 10 μm (A, G–K). A–K, Data are represented as mean ± SEM.
To next ask whether cAMP levels controlled the stabilization during the navigation to the DREZ, larvae were treated with either sp-cAMP or rp-cAMP, and the difference of transient and stable invadopodia were scored via intensity of Lifeact-GFP at the growth cone between navigation and at the DREZ. Treatment with sp-cAMP led to a temporal imbalance of both transient and stable invadopodia, with an average decrease in transient difference from 10.60 ± 0.75 (n = 5 DRG) to −0.20 ± 0.73 (Fig. 3D), and an average increased positive stable difference from −3.20 ± 0.37 (n = 5 DRG) to 3.60 ± 1.63 (p < 0.0001, n = 5 DRG, t test; p = 0.0036, n = 5 DRG, t test; Fig. 3E). This imbalance of stabilization also occurred with the rp-cAMP treated larvae, with an average difference of 1.20 ± 1.07 transient invadopodia (Fig. 3D) and 2.20 ± 1.39 stable invadopodia (p < 0.0001, 5 DRG, t test; p = 0.0057, n = 5 DRG, t test; Fig. 3E). These data are consistent with the hypothesis that invadopodia stabilization, via cAMP, is precisely controlled until the axon reaches the DREZ.
To determine how cAMP then impacts the coordinated invadopodia state, we next hypothesized that modulating cAMP levels would alter the timing of invadopodia stabilization. To test this, we treated Tg(sox10:lifeact-gfp) animals at 36 hpf with 100 µm sp-cAMP, 25 µm rp-cAMP, or DMSO, and measured Lifeact-GFP intensity. Both treatments led to an uncoordinated invadopodia state with many invadopodia peaks forming during navigation and while the growth cone was at the DREZ, rather than one primary peak at the DREZ. This led to an increase in the duration of time spent at the DREZ from an average of 75.00 ± 10.61 min in DMSO animals (n = 4 DRG; Fig. 3F,G), to 285.00 ± 37.38 min in sp-cAMP-treated animals (Fig. 3F,H; Movie 3), and 310.00 ± 22.27 min in rp-cAMP-treated animals (p = 0.0019, n = 5 DRG, t test; p < 0.0001, n = 4 DRG, t test; Fig. 3F,I; Movie 4). These data are consistent with the hypothesis that cAMP regulates the coordinated invadopodia state at the DREZ by destabilizing invadopodia during navigation.
Time lapse of sp-cAMP-treated growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal treated with 100 μm sp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of rp-cAMP-treated growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal treated with 25 μm rp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Since both increasing and decreasing cAMP leads to the same phenotype, we tested the possibility that an intermediate level of cAMP regulates the brake on the coordinated invadopodia state. We hypothesized that if an intermediate level of cAMP is required, then co-treating with sp-cAMP and rp-cAMP would restore control of invadopodia. To test this, we treated Tg(sox10:lifeact-gfp) animals with both 100 µm sp-cAMP and 25 µm rp-cAMP at 36 hpf, then time lapsed at 48 hpf. Co-treatment led to a rescue in the average number of invadopodia puncta to 2.80 ± 0.37 (p = 0.0.7597, n = 5 DRG, t test; Fig. 3B), as well as filopodia length to 3.22 ± 0.25 µm (p = 0.4498, n = 10 filopodia over 5 DRG, t test; Fig. 3C). We next tested whether a co-treatment rescued both the transient invadopodia-dominated navigation to the DREZ, as well as the stable invadopodia state while at the DREZ. Both states were rescued, with transient invadopodia having an average positive difference of 10.40 ± 0.60 (p = 0.840, n = 5 DRG, t test; Fig. 3D) and stable invadopodia having an average negative difference of −3.20 ± 0.58 (p > 0.9999, n = 5 DRG, t test; Fig. 3E). This led to a rescue of the coordinated invadopodia state, with one primary peak forming while the growth cone was at the DREZ for an average duration of 76.00 ± 12.29 min (p = 0.9541, n = 5 DRG, t test; Fig. 3F,J). Together, these data are consistent with the hypothesis that intermediate levels of cAMP are necessary for the precise temporal control of the invadopodia brake.
cAMP functions within the same pathway as Rac1
In order to next test the hypothesis that cAMP acts within the same pathway as Rac1 to control the invadopodia brake, we treated Tg(sox10:gal4)(uas:lifeact-gfp) animals injected with pa-Rac1 with rp-cAMP, and exposed the animal to 445-nm light before growth cone entry. We first scored invadopodia number and filopodia length, and found the combination of rp-cAMP treatment with activation of pa-Rac1 led to a rescue of invadopodia number to an average of 3.00 ± 0.41 (Fig. 3B), as well as filopodia length to an average of 3.15 ± 0.16 µm (p = 0.5447, n = 4 DRG, t test; p = 0.2474, n = 20 filopodia over four DRG, t test; Fig. 3C). We next sought to test a temporal rescue of the stable invadopodia states and found that the difference between transient invadopodia once again became more positive with an average of 9.00 ± 1.78 (p = 0.3982, n = 4 DRG, t test; Fig. 3D). Additionally, stable invadopodia were rescued to an average negative difference of −4.25 ± 1.18 (p = 0.3801, n = 4 DRG, t test; Fig. 3E). This ultimately led to a rescue of the coordinated invadopodia state, with the formation of one primary peak while the growth cone was at the DREZ for an average duration of 63.75 ± 3.15 min (p = 0.3484, n = 4 DRG, t test; Fig. 3F,K; Movie 5). This data indicates that perturbing cAMP can alter the cell specific manipulation of Rac1. Taken together, this supports the hypothesis that cAMP and Rac1 function within the same pathway, and suggests that cAMP may activate Rac1 to control actin-state dynamics at the growth cone. However, we cannot definitively rule out the possibility that cAMP does so in a non-cell-autonomous mechanism.
Time lapse of photoactivated-Rac1 growth cone treated with rp-cAMP. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal with photo-activated Rac1 and treated with 25 μm rp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
DCC identified as candidate for actin-regulated entry
We next explored receptor/ligand molecular pairs that could control cAMP and Rac1 to regulate invadopodia. We performed a pilot genetic screen of molecules with a role in axons and used tnfɑ expression as an output for entry. Using CRISPR/Cas9 screening, Tg(tnfɑ:gfp) animals were injected at the one-cell stage with a pool of candidate gRNA, and the number of Tg(tnfa:gfp)+ DRG were quantified (Fig. 4A). Animals were injected with only Cas9 as controls. In control animals, 100.00% of animals had over 25 Tg(tnfa:gfp)+ DRG along the right side of the spinal cord. In contrast, injections of gRNAs to Netrin signaling components led to 53.00% (ntn1a), 30.00% (ntn1b), and 58.00% (dcc) of Tg(tnfa:gfp) animals having less than seven Tg(tnfa:gfp)+ DRG (Fig. 4B). To ensure indels were created, a pool of these animals were validated by T7E1 assay. To further validate the candidates from the screen we first tested whether Tg(tnfa:gfp) was reduced in dcczm130198, which has an early stop codon (Jain et al., 2014). Since our data from cAMP manipulations demonstrated cAMP control was not binary, we also hypothesized that the signaling mechanism that controls cAMP may be sensitive to different doses of other pathway components. To test this, we visualized Tg(tnfa:gfp) animals in dcc+/−. Scoring of GFP in these animals demonstrated a decrease in tnfa+ DRG neurons as compared with wild-type animals (Fig. 4C). These data demonstrate a potential haploinsufficiency of dcc signaling in modulating growth cone entry at the DREZ.
Figure 4.
DCC identified as candidate for actin-regulated entry. A, Schematic representation of screen using Tg(tnfa:gfp) animals. B, Quantification of genetic screen showing percentage animals with the given number of Tg(tnfa:gfp)+ DRG (n = 53 ntn1a animals; n = 53 ntn1b animals; n = 26 dcc animals; n = 43 lamb1a animals). C, Images of Tg(tnfa:gfp) animals in wild-type and dcc+/− backgrounds at 72 hpf. D, Confocal maximum projects of the anterior, middle, and posterior Tg(gfap:gfp) animals with RNAScope of ntn1b. E, Quantification of ntn1b puncta per anatomical location (n = 3 animals). Green dots denote anterior. Blue dots denote middle. Purple dots denote posterior. F, Confocal maximum projections of Tg(ngn1:gfp) animals with RNAScope of dcc. Dotted orange box denotes inlet of 4 μm DRG plane. White arrowheads denote localization with dcc puncta. G, Quantification of percent of DRG that contain dcc puncta (n = 10 wild-type DRG; n = 6 dcc+/− DRG; n = 3 dcc−/− DRG). H, Quantification of number of dcc puncta per DRG over anatomic location (n = 7 animals). Green dots denote anterior. Blue dots denote middle. Purple dots denote posterior. I, Quantification of percentage of pioneer axons in dcc−/− and dcc+/− animals. Scale bars: 10 μm (C, D, F). E, H, Data are represented as mean ± SEM.
To understand whether dcc and its ligand, netrin, could control invadopodia during navigation and entry, we explored the possibility that both the ligand and receptor could be spatiotemporally located to alter invadopodia of DRG growth cones. To test this, we first performed RNAscope in situ hybridization with a ntn1 probe on 48 hpf Tg(gfap:gfp) animals. Our data showed expression of ntn1 within the gfap+ radial glial limitans, primarily in the floor plate (Fig. 4D,E). We performed a similar RNAscope analysis using a dcc probe on Tg(ngn1:gfp) to visualize neurons (Fig. 4F), and scored the percentage of DRG containing dcc puncta. We found that around 80% of DRG neurons contained dcc puncta (Fig. 4G), demonstrating that dcc is expressed in DRG neurons. dcc was also localized to the spinal cord as previously described (Jain et al., 2014). We also used the dcc probe at 48 hpf on dcczm130198 (Jain et al., 2014), in a Tg(sox10:lifeact-gfp) background to visualize the DRG neurons, and found a decrease to 38% and 0% in DRG neurons that contain dcc puncta in both dcc+/− and dcc−/−, respectively (Fig. 4G). These data confirm that dcc+/− animals demonstrate intermediate expression levels of dcc and together support the hypothesis that ntn1 and dcc are spatiotemporally present to regulate invadopodia-driven growth cone entry.
To provide insight into the potential mechanism of how invadopodia-mediated entry could be modulated, we explored whether ntn1 and dcc were expressed at different levels in axons before, during, and after entry. We reasoned that a likely mechanism of how DCC signaling could regulate entry is by upregulating the dcc receptor in DRG neurons precisely when navigation is occurring, but not during invasion of the growth cone at the DREZ. To test this possibility, we took advantage of the developmental concept that maturity of the spinal cord occurs in an anterior to posterior direction. At 48 hpf, zebrafish animals display DRG at varying degrees of maturation, with anterior-most DRG that have growth cones that have entered into the spinal cord, and posterior-located DRG that have not initiated pioneer axons. We then used the dcc probes on 48 hpf Tg(ngn1:gfp) animals, and scored expression anterior to posterior within the same animals. Anterior was defined as the area encompassed by the first 10 DRG located closest to the head, middle was defined as the next 10 DRG, and posterior was defined as the section covering the last 10 DRG closest to the tail. These results indicated significant differences (p = 0.0489, n = 18, 16, and 13 DRG, one-way ANOVA), with an average of 5.50, 4.90, and 4.40 dcc puncta per DRG that could be detected between anterior, middle, or posterior spinal locations (Fig. 4H). Similar analysis with ntn1b showed an increase in ntn1 puncta along the floor plate of the spinal cord (p < 0.0001, n = 3 DRG, one-way ANOVA; Fig. 4E), with an average of 9.33 puncta in the anterior, 68.00 puncta in the middle, and 196.70 puncta in the posterior, demonstrating that more immature DRG neurons that express less dcc interact with spinal cord regions with higher ntn1 expression. Our detection of dcc in neurons that have not entered the spinal cord is inconsistent with the hypothesis that this dcc-mediated process is controlled by the precise temporal upregulation of dcc transcripts when the pioneer axon is either navigating to the DREZ or is located at the DREZ.
DCC signaling controls invadopodia in pioneer axons
DCC has been shown to act as a key receptor in axonal navigation, but its role in regulating invadopodia at the growth cone is unknown (Duman-Scheel, 2009; Jain et al., 2014; Junge et al., 2016). To address this gap, we examined how dcc affects invadopodia during pioneer axon navigation and while at the DREZ. To first test this we time-lapse imaged wild-type siblings, dcc+/−, or dcc−/− in a Tg(sox10:lifeact-gfp) background at 48 hpf, and scored the growth cone. In these movies, we scored that 98.00% of wild-type DRG neurons formed pioneer axons, which decreased to 66.66% in dcc+/− and 20.00% in dcc−/− (Fig. 4I). However, all DRG neurons were present. Given that formation of a pioneer axon and subsequent growth cone is required to study invadopodia, heterozygous mutants serve as a more useful model to study actin dynamics than homozygous mutants.
To better understand the role of dcc in regulating the invadopodia brake, we next hypothesized that dcc is necessary to the formation of invadopodia while at the DREZ. Using wild-type siblings, dcc+/−, or dcc−/− in a Tg(sox10:lifeact-gfp) background, we time lapsed animals at 48 hpf and quantified the number of invadopodia puncta. We found an average decrease in puncta while the axon was at the DREZ from 3.00 ± 0.26 (n = 6 DRG) in wild-type siblings to 1.29 ± 0.29 in dcc+/− (p = 0.0011, n = 7 DRG, t test) and 1.25 ± 0.48 in dcc−/− (p = 0.0079, n = 4 DRG, t test) siblings (Fig. 5A), indicating that dcc impacts the control of invadopodia in pioneer axons. dcc paralleled Rac1 and cAMP analysis, with a decrease in invadopodia puncta accompanied by an increase in filopodia length (Fig. 5B). Together, these data implicate dcc in the pathway regulating the invadopodia brake.
Figure 5.
DCC signaling controls invadopodia in pioneer axons. A, Quantification of invadopodia number during time at the DREZ (n = 6 wild-type DRG; n = 4 dcc−/− DRG; n = 7 dcc+/− DRs; n = 3 dcc+/− + sox10:dcc DRG; n = 5 dcc+/− + sp-cAMP DRG; n = 6 dcc+/− + rp-cAMP DRG). B, Quantification of filopodia length during time at the DREZ (n = 10 wild-type filopodia; n = 10 dcc−/− filopodia; n = 10 dcc+/− filopodia; n = 10 dcc+/− + sox10:dcc filopodia; n = 10 dcc+/− + sp-cAMP filopodia; n = 10 dcc+/− + rp-cAMP filopodia). C, Quantification of difference of transient invadopodia between navigation and time at the DREZ (n = 5 wild-type DRG; n = 4 dcc−/− DRG; n = 5 dcc+/− DRG; n = 3 dcc+/− + sox10:dcc DRG; n = 5 dcc+/− + sp-cAMP DRG; n = 5 dcc+/− + rp-cAMP DRG). D, Quantification of difference of stable invadopodia between navigation and time at the DREZ (n = 5 wild-type DRG; n = 4 dcc−/− DRG; n = 5 dcc+/− DRG; n = 3 dcc+/− + sox10:dcc DRG; n = 5 dcc+/− + sp-cAMP DRG; n = 5 dcc+/− + rp-cAMP DRG). E, Quantification of duration of coordinated invadopodia state of stable invadopodia (n = 4 wild-type DRG; n = 4 dcc−/− DRG; n = 4 dcc+/− DRG; n = 3 dcc+/− + sox10:dcc DRG; n = 5 dcc+/− + sp-cAMP DRG; n = 5 dcc+/− + rp-cAMP DRG). A–E, Black dots denote wild type. Purple dots denote dcc−/−. Orange dots denote dcc+/−. Red dots denote dcc+/− + sox10:dcc. Blue dots denote dcc+/− + sp-cAMP. Green dots denote dcc+/− + 25 μm rp-cAMP. F–J, left, Confocal maximum projection of Tg(sox10:lifeact-gfp) from a 24-h time lapse starting at 48 hpf in a (F) wild-type, (G) dcc−/−, (H) dcc+/−, (I) dcc+/− + sox10:dcc, or (J) dcc+/− + sp-cAMP-treated animal. White box denotes DREZ insets. White arrow denotes actin concentrates. Right, Tracing of growth cone Lifeact-GFP intensity before, during, and after entry in a (F) wild-type (n = 4), (G) dcc−/− (n = 4), (H) dcc+/− (n = 4), (I) dcc+/− + sox10:dcc (n = 3), or (J) dcc+/− + sp-cAMP (n = 5) treated animal. Blue box denotes time at the DREZ. Blue line denotes representative graph. Dashed black line denotes mean ± SEM. K, left, Confocal maximum projection of Tg(sox10:lifeact-gfp) at 48 hpf localizing with DCC. Right, Quantification of DCC localization to the DRG soma, transient invadopodia, and stable invadopodia (n = 6 DRG). Scale bars: 10 μm (F–K). A–K, Data are represented as mean ± SEM.
If dcc controls invadopodia stabilization until the axon reaches the DREZ, we would expect both transient and stable invadopodia during the navigation to the DREZ to be altered. dcc+/− and dcc−/− animals showed a decrease in the difference of transient invadopodia from an average of 10.20 ± 0.37 (n = 5 DRG) to 3.20 ± 0.86 and 2.00 ± 0.71 (p < 0.0001, n = 5 DRG, t test; p < 0.0001, n = 4 DRG, t test; Fig. 5C), indicating a lack of navigational presence. The difference of stable invadopodia also became more positive from an average of −2.40 ± 0.40 (n = 5 DRG) to 2.60 ± 2.41 and 2.00 ± 0.71 (p = 0.0024, n = 5 DRG, t test; p = 0.0007, n = 4 DRG, t test; Fig. 5D), indicating premature stabilization of the invadopodia. These data are consistent with the idea that invadopodia control through the entire navigation process is dictated by dcc. If the above hypothesis is correct, then dcc should impact the coordinated invadopodia state. To test this, we performed 24-h time-lapse imaging of 48 hpf wild-type siblings, dcc+/−, and dcc−/− animals in a Tg(sox10:lifeact-gfp) background. In wild-type sibling animals, one large peak was observed while the growth cone was at the DREZ, for an average duration of 78.75 ± 10.87 min (n = 4 DRG; Fig. 5E,F). However, in both dcc+/− and dcc−/− animals invadopodia over-stabilized, resulting in an average duration of 218.18 ± 24.78 min (p = 0.0021, n = 4 DRG, t test; Fig. 5E,H; Movie 6) and 260.00 ± 35.12 min (p = 0.0024, n = 4 DRG, t test; Fig. 5E,G), respectively. These results thus indicate that dcc is necessary for the coordinated invadopodia state.
Time lapse of dcc+/− growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
To test that dcc functions cell-autonomously within the DRG, we next hypothesized that rescuing dcc only in pioneer DRG cells would also rescue actin dynamics. To test this, Tg(sox10:gal4)(uas:lifeact-gfp); dcc+/− animals were injected with uas:dcc-tdTomato at the one-cell stage and only tdTomato+ neurons were scored. Animals were imaged, quantified, and then genotyped, ensuring that quantification was completed blind to the genotype. Actin dynamics in DRG cells that express dcc-tdTomato were rescued as compared with dcc+/− animals without dcc-tdTomato, with the average number of invadopodia being rescued to 2.67 ± 0.33 (Fig. 5A), and the average filopodial length being rescued to 3.43 ± 0.26 µm (p = 0.0233, n = 3 DRG, t test; p = 0.0002, n = 10 filopodia over three DRG, t test; Fig. 5B). To further test rescue of transient invadopodia during navigation, and stable invadopodia while at the DREZ, the difference was scored as previously described. Transient invadopodia were restored to an average positive difference of 9.00 ± 1.00 (p = 0.0053, n = 3 DRG, t test; Fig. 5C), and stable invadopodia were restored to an average negative difference of −3.33 ± 0.33 (p = 0.0065, n = 3 DRG, t test; Fig. 5D). Rescue of actin population dynamics led to one primary Lifeact-GFP intensity peak while the growth cone was at the DREZ, for an average duration of 80.00 ± 14.43 min (p = 0.0072, n = 3 DRG, t test; Fig. 5E,I; Movie 7). These data are consistent with the hypothesis that dcc functions cell autonomously within the DRG to control the invadopodia brake.
Time lapse of dcc+/− growth cone injected with sox10:dcc. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal injected with sox10:dcc. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
To further explore the potential mechanism of how dcc functions to regulate transient versus stable invadopodia, we next scored localization of DCC-tdTomato while the growth cone was navigating to and located at the DREZ. To acquire this spatial information, DCC-tdTomato was imagined in pioneer neurons for 24 h. In this analysis, 77.83% of DCC-tdTomato puncta were detected in the soma and 22.17% localized to transient invadopodia (Fig. 5K). We could not detect DCC-tdTomato localized to stable invadopodia (p = 0.0032, n = 6 DRG, t test; p < 0.0001, n = 6 DRG, t test). The simplest explanation for these data are that DCC may regulate the balance of transient versus stable invadopodia by localizing at transient invadopodia to drive their disassembly.
Increasing cAMP rescues dcc phenotypes
To determine the placement of dcc in the signaling pathway, we next explored whether dcc functions in the same pathway as cAMP. To test this, Tg(sox10:lifeact-gfp); dcc+/− animals were treated with either 100 µm sp-cAMP or 25 µm rp-cAMP at 36 hpf, and time lapsed at 48 hpf. The number of invadopodia and filopodia length were then scored. Quantification of these movies showed that treatment with rp-cAMP did not rescue either actin population, with an average of 1.67 ± 0.33 invadopodia puncta (p = 0.0101, n = 6 DRG, t test; Fig. 5A) and an average filopodia length of 7.41 ± 0.48 µm (p < 0.0001, n = 10 filopodia over 6 DRG, t test; Fig. 5B). However, treatment with sp-cAMP rescued invadopodia number to an average of 3.00 ± 0.32 puncta (p > 0.9999, n = 5 DRG, t test; Fig. 5A) and an average filopodia length to 3.36 ± 0.22 µm (p = 0.9377, n = 10 filopodia over 5 DRG, t test; Fig. 5B) while the axon was at the DREZ. Stabilization states of invadopodia similarly were rescued with sp-cAMP. Transient invadopodia with sp-cAMP treatment had an average positive difference of 8.40 ± 1.63 (Fig. 5C), and stable invadopodia had an average negative difference of −1.40 ± 0.68 (p = 0.3134, n = 5 DRG, t test; p = 0.2398, n = 5 DRG, t test; Fig. 5D). This was not the case in rp-cAMP-treated animals, with an average decreased positive difference to 0.20 ± 1.88 transient invadopodia (Fig. 5C) and an average positive difference of 3.80 ± 1.43 stable invadopodia (p = 0.0008, n = 5 DRG, t test; p = 0.0031, n = 5 DRG, t test; Fig. 5D). Finally, the coordinated invadopodia state was scored and showed sp-cAMP treatment in dcc+/− led to one primary Lifeact-GFP intensity peak forming while the growth cone was at the DREZ, and rescued the duration to an average of 89.00 ± 10.42 min (p = 0.5214, n = 5 DRG, t test; Fig. 5E,J; Movie 8), demonstrating that increasing cAMP can rescue the reduction of dcc. In the rp-cAMP group, invadopodia became over-active, and lasted for an average duration of 287.00 ± 43.06 min (p = 0.0041, n = 5 DRG, t test; Fig. 5E; Movie 9). These results are consistent with the hypothesis that dcc functions in the same pathway as cAMP.
Time lapse of dcc+/− growth cone treated with sp-cAMP. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal treated with 100 μm sp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of dcc+/− growth cone treated with rp-cAMP. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal treated with 25 μm rp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Disruption of coordinated invadopodia state disrupts growth cone entry
We next sought to test the hypothesis that modulation of the coordinated invadopodia state leads to defects in growth cone entry. Using Tg(tnfa:gfp) animals, we selected several manipulations that disrupt invadopodia stabilization and tested growth cone entry (Fig. 6A). Tg(tnfa:gfp) animals were first treated at 36 hpf with either the Rac 1 inhibitor (1 µm NSC23766), the Rac1 inhibitor and cAMP agonist (1 µm NSC23766 + 100 µm sp-cAMP), the cAMP agonist (100 µm sp-cAMP), or the cAMP antagonist (25 µm rp-cAMP). The number of Tg(tnfa:gfp)+ DRG neurons was quantified at 72 hpf (Fig. 6B), and DMSO-treated animals were used as a control. Consistent with the idea that a proper coordinated invadopodia state is a necessity for growth cone entry, animals with Rac1 inhibition (n = 11) had an average of 18.73 ± 1.78 Tg(tnfa:gfp)+ DRG neurons as compared with DMSO control animals (n = 6), which had an average of 29.33 ± 0.99 Tg(tnfa:gfp)+ DRG neurons (p = 0.0008, t test). A reduction of Tg(tnfa:gfp)+ DRG neurons was also detected in animals to an average of 15.00 ± 1.14 or 14.20 ± 1.16 Tg(tnfa:gfp)+ DRG neurons in the sp-cAMP and rp-cAMP groups (p < 0.0001, n = 5 DRG, t test; p < 0.0001, n = 5 DRG, t test). To test whether cAMP and Rac1 functioned in the same pathway to drive axon entry, animals were treated with both sp-cAMP and NSC23766 (n = 13) and demonstrated a rescue to an average of 28.62 ± 0.54 Tg(tnfa:gfp)+ DRG neurons (p < 0.0001, t test). Together, these data indicate that cAMP likely functions in the same pathway as Rac1. Next, Tg(tnfa:gfp);dcc+/− animals were scored, and an average decrease to 6.75 ± 0.48 Tg(tnfa:gfp)+ DRG neurons from an average of 29.31 ± 0.49 Tg(tnfa:gfp)+ DRG neurons in wild-type animals (n = 6) was also observed (p < 0.0001, n = 4 DRG, t test; Fig. 6B). To test that dcc functions in the same pathway as Rac1, Tg(tnfa:gfp);dcc+/− animals were treated with NSC23766 (n = 5). These animals had an average of 8.60 ± 3.08 Tg(tnfa:gfp)+ DRG neurons (p = 0.6146, t test), consistent with the hypothesis that Rac1 functions within the same signaling cascade as dcc (Fig. 6B). However, Tg(tnfa:gfp);dcc+/− animals alone had a more severe phenotype as compared with sp-cAMP (p = 0.0005), rp-cAMP (p = 0.0010), and Rac1 inhibited (p = 0017) animals, suggesting that dcc also functions in a separate pathway to regulate entry. While all manipulations impacted growth cone entry, further analysis revealed that all axons still bifurcated in the absence of entry (Fig. 6C). Taken together, these data suggest that the coordinated invadopodia state is necessary for growth cone entry, and implicate Rac1, cAMP, and dcc as essential molecular regulators of this event.
Figure 6.
Disruption of coordinated invadopodia state disrupts axon entry. A, Images of Tg(tnfa:gfp) animals at 72 hpf treated with DMSO, 1 μm NSC23766, 1 μm NSC23766 + 100 μm sp-cAMP, or in a dcc+/− background treated with 1 μm NSC23766. B, Quantification of number of Tg(tnfa:gfp)+ DRG in animals treated with DMSO (n = 6), 1 μm NSC23766 (n = 11), 1 μm NSC23766 + 100 μm sp-cAMP (n = 13), 100 μm sp-cAMP (n = 5), or 25 μm rp-cAMP (n = 5), or wild-type animals (n = 6), dcc+/− animals (n = 4), or dcc+/− animals treated with 1 μm NSC23766 (n = 5). Black dots denote DMSO. Magenta dots denote NSC23766. Cyan dots denote NSC23766 + sp-cAMP. Blue dots denote sp-cAMP. Green dots denote rp-cAMP. Gray dots denote wild type. Purple dots denote dcc+/−. Orange dots denote dcc+/− + NSC23766. C, Quantification of percentage of DRG neurons that bifurcate in pa-Rac1 (n = 5), 1 μm NSC23766 (n = 8), 100 μm sp-cAMP (n = 11), 25 μm rp-cAMP (n = 6), or dcc+/− animals (n = 14). D, IMARIS rendering of the pioneer axon (green arrow) and radial glial boundary (purple arrow) in the x-y-plane and the y-orthogonal view. Dotted orange line represents the line of intensity used to determine intensity numbers of transgenes in entry quantifications. E, Quantification of percent of axons that entered the spinal cord in Tg(sox10:meGFP);Tg(gfap:nsfb-mcherry) animals treated with DMSO (n = 6 DRG), 100 μm sp-cAMP (n = 9 DRG), or 25 μm rp-cAMP (n = 9 DRG), or in a dcc−/− (n = 6 DRG) or dcc+/− background treated with DMSO (n = 7 DRG), 100 μm sp-cAMP (n = 6 DRG), 25 μm rp-cAMP (n = 4 DRG), injected with sox10:dcc (n = 4), or injected with sox10:dcc and treated with rp-cAMP (n = 4). F, G, Intensity tracings of y-orthogonals show relationship between the pioneer axon and glial limitans. Green arrows denote pioneer axon. Purple arrows denote glial limitans. Scale bars: 10 μm (A, D). B, Data are represented as mean ± SEM.
To further test entry phenotypes, we used a second complementary approach to score an inability of the pioneer axon to cross the radial glial boundary (Fig. 6D; Nichols and Smith, 2019a,b). To test this we treated Tg(sox10:meGFP);Tg(gfap:nsfb-mcherry) animals with 100 µm sp-cAMP, 25 µm rp-cAMP, or DMSO at 36 hpf and imaged at 72 hpf. In these animals, the glial limitans boundary is marked by mcherry, and the pioneer axon by meGFP. We used a published entry quantification, which marks the spatial location of the meGFP+ axon as either medial or lateral to the mCherry+ glial limitans by drawing a line of intensity at the tip of the growth cone (Fig. 6D). We found that treatment with sp-cAMP or rp-cAMP led to a more lateral location in 77.78% (n = 9 DRG) and 66.67% (n = 9 DRG) of pioneer axons as compared with the radial glial boundary, indicating the axon has not entered the spinal cord. This differed from DMSO-treated animals, in which 0.00% (n = 6 DRG) of pioneer axons had a more lateral location (Fig. 6E,F). In Tg(sox10:meGFP);Tg(gfap:nsfb-mcherry) animals in either a dcc+/− or dcc−/− background, 71.43% (n = 7 DRG) and 83.33% (n = 6 DRG), respectively, of DRG neurons had a more lateral location, indicating that dcc is required for entry and again confirming that dcc displays haploinsufficiency for growth cone entry (Fig. 6E,G). To confirm that dcc functions cell-autonomously to regulate growth cone entry, Tg(sox10:gal4)(uas:lifeact-gfp); dcc+/− animals were injected with uas:dcc-tdTomato at the one-cell stage and entry was scored in tdTomato+ neurons. This led to a rescue, with 100% of DRG growth cones completing entry (n = 4 DRG; Fig. 6E,G). Together with published results of growth cone entry defects when Rac1 is cell-autonomously manipulated (Nichols and Smith, 2019a), these data are consistent with the hypothesis that modulation of Rac1, cAMP, and dcc leads to improper growth cone entry.
To next explore that a rescue in the coordinated invadopodia state equates to a rescue in DRG growth cone entry, we then treated Tg(sox10:meGFP);Tg(gfap:nsfb-mcherry);dcc+/− animals with either 100 µm sp-cAMP or 25 µm rp-cAMP at 36 hpf and scored growth cone and radial glial intensity at 72 hpf. The results recapitulated the previous phenotypes of dcc+/− animals treated with cAMP, with rp-cAMP causing 0.00% of growth cone entry (n = 4 DRG) and sp-cAMP rescuing entry in 100.00% of DRG growth cone entry (n = 6 DRG; Fig. 6E). To further approach the cell-specificity of cAMP manipulation, Tg(sox10:gal4)(uas:lifeact-gfp); dcc+/− animals were injected with uas:dcc-tdTomato and treated with 25 µm rp-cAMP at 36 hpf. Given cell-autonomous rescue of growth cone entry in tdTomato+ neurons, a defect in growth cone entry would most likely indicate that cAMP also functions within the DRG. Results indicated 0% of DRG growth cones entered (n = 4 DRG) past the glial limitans (Fig. 6E,G). While we cannot rule out the possibility that cAMP signaling in neighboring cells impacts DRG dynamics, the simplest explanation of these results is that cAMP acts cell-autonomously to regulate growth cone entry. With two complementary approaches, these results are consistent with the hypothesis that the coordinated invadopodia state, controlled by Rac1, cAMP, and DCC, drives growth cone entry.
Molecules that disrupt invadopodia state and growth cone entry cause behavioral defects
To test how DCC-dependent pioneer axon growth cone entry impacts the function of DRG neurons, we next asked whether sensory-specific behavior is altered. We therefore developed a behavioral assay to measure the animals response to exposure to sensory stimuli (Fig. 7A). We focused on behavior as a result of exposure to 4°C water, which has previously been shown to depend on intact DRG neurons in zebrafish (Nichols and Smith, 2020). To explore this response, larvae zebrafish were transferred in 23°C water to a testing arena. All the water was removed from the testing arena followed by the immediate addition of either 4°C or 23°C water. During exposure, we recorded their behavioral movements at 250 frames per second for 60 s. For the analysis, we omitted the first 2 s of video after the water was added to eliminate any movements that occurred as a result of the flow of water onto the fish. Upon exposure to 4°C, animals displayed a shivering movement, which is best represented by collapsing all individual movie frames into a single image (Fig. 7B). In contrast, animals exposed to 23°C in this assay mostly remain stationary. To distinguish this shivering behavior from normal turning and swimming movements, we used FLOTE software (Burgess and Granato, 2007) to measure the tail curvature by dividing each larva into four segments and calculating the sum of the angles between segments 2 and 3, angle A, and 3 and 4, angle B (Fig. 7C,D). Turn and swim movements have regular rhythm and much greater magnitude than shivers, which are erratic, very low-magnitude movements (Fig. 7E,F). As FLOTE was unable to reliably detect these subtle shiver movements automatically, for subsequent analyses we counted both the number and duration of shiver bouts manually. To explore the developmental timing of this behavioral response, we measured the number of cold shivering events in animals at 2–5 dpf after exposure to either 4°C or 23°C water (Fig. 7G). At 2 dpf, when DRG pioneer axons have yet to enter the spinal cord but mechanosensitive Rohon Beard neurons are present, animals did not display a cold shivering response. The shiver response is specific to the cold stimulus, as turn/swim movements were largely unaffected at 4°C, with an average of 0.75 ± 0.31 and 0.54 ± 0.13 turn/swim events occurring between exposure to 4°C versus 23°C (p = 0.4659, n = 12 fish at 23°C and 24 at 4°C, t test; Fig. 7I). At 3 dpf, after DRG pioneer axons have entered the spinal cord, our analysis shows a higher number of shiver events with an average of 9.52 ± 1.19 shivers seen after 4°C exposure, compared with an average of 0.76 ± 0.44 shivers at 23°C exposure (p < 0.0001, n = 21 fish at 23°C and 42 fish at 4°C, t test). The increased number of shiver events in 4°C compared with 23°C exposure was also observed at 4 and 5 dpf, suggesting this behavior is maintained across development. Similar observations were also obtained when measuring the percentage of time shivering at 3 dpf (Fig. 7H), with 24.43% shivers seen after 4°C exposure, compared with 0.95% shivers at 23°C exposure (p < 0.0001, n = 21 fish at 23°C and 42 fish at 4°C, t test). This increase in percentage of time shivering was also maintained at 4 and 5 dpf. This establishes that zebrafish develop a distinct shivering behavior specific to exposure to 4°C after DRG pioneer axons have entered the spinal cord.
Figure 7.
Molecules that disrupt coordinated invadopodia state and growth cone entry cause behavioral defects. A, Schematic of the behavioral assay. B, Representative collapsed image from 60-s movies of animals exposed to either 4°C or 23°C showing animals display a shivering movement after exposure to 4°C. C, Representative image of FLOTE segmentation used for the behavioral analysis. D, Method of calculating tail curvature. Curvature magnitude analyzed as the sum of angle A, angle between segments 2 and 3, and angle B, angle between segments 3 and 4. E, Representative graph of typical “turn/swim” bout. F, Two representative plots of spine curvature during shiver events. Note the scale on the y-axis demonstrates a shorter displacement then seen in turn bouts. G, Number of shivers quantified at 4°C (n = 24 at 2 dpf, 42 at 3 dpf, 42 at 4 dpf, and 48 at 5 dpf) and 23°C (n = 12 at 2 dpf, 21 at 3 dpf, 21 at 4 dpf, and 24 at 5 dpf). H, Total percentage of time shivering quantified at 4°C (n = 24 at 2 dpf, 42 at 3 dpf, 42 at 4 dpf, and 48 at 5 dpf) and 23°C (n = 12 at 2 dpf, 21 at 3 dpf, 21 at 4 dpf, and 24 at 5 dpf). I, Number of “turn/swim” bouts at 4°C (n = 24 at 2 dpf, 42 at 3 dpf, 42 at 4 dpf, and 48 at 5 dpf) and 23°C (n = 12 at 2 dpf, 21 at 3 dpf, 21 at 4 dpf, and 24 at 5 dpf). G–I, Cyan dots denote 4°C. Pink dots denote 23°C. J, Representative collapsed image from 60-s movies of animals exposed to 4°C and either treated with DMSO, 100 μm sp-cAMP, 25 μm rp-cAMP, or in a dcc+/− background treated with DMSO or 100 μm sp-cAMP. K, Total percentage of time shivering quantified at 4°C and 23°C (n = 4 sp-cAMP animals; 4 rp-cAMP animals; 5 dcc+/− animals; 3 dcc+/− +100 μm sp-cAMP animals). Cyan dots denote 4°C. Pink dots denote 23°C. L, Schematic representation of DCC-mediated invadopodia brake. Scale bars: 10 μm (B, J). G–I, K, Data are represented as mean ± SEM.
Using this behavioral assay, we first hypothesized that treatment with rp-cAMP or sp-cAMP would cause a lack of shivering in response to 4°C. In this assay, animals were treated with either 100 µm sp-cAMP, 25 µm rp-cAMP, or DMSO from 36–72 hpf. At 3 dpf, animals were then immersed in 4°C water and recorded for 60 s (Fig. 7J). In these movies, the overall duration of time shivering was scored as a percentage. Both sp-cAMP-treated and rp-cAMP-treated animals exhibited lower shivering durations in 4°C water as compared with control (Fig. 7K), with control fish spending an average duration of 39.00% of overall time shivering and sp-cAMP and rp-cAMP exhibiting an average of 0.00% (p < 0.0001, n = 4 animals, t test) and 18.00% (p = 0.0029, n = 4 animals, t test) of time shivering. We next performed the same assay but with dcc+/− animals (Fig. 7J). Again, overall duration of the time spent shivering decreased to an average of 17.00% (p = 0.0179, n = 5 animals, t test; Fig. 7K). Given that sp-cAMP treatment in dcc+/− rescued growth cone entry, we next hypothesized that treatment with sp-cAMP in dcc+/− animals would rescue behavior. As expected, the overall time spent shivering was rescued to an average of 41.00% (p = 0.3961, n = 3 animals, t test; Fig. 7K). Taken together, the simplest explanation for the data are that Rac1, cAMP, and dcc are crucial to regulate invadopodia precisely at the DREZ, thereby modulating entry and, consequently, behavior.
Discussion
The precise control of distinct growth cone structures is essential to DRG growth cone entry into the spinal cord. By using time-lapse imaging of actin populations with genetic pathway manipulations, we have identified DCC, cAMP, and Rac-1 as critical upstream regulators of the coordinated invadopodia state. Although it is well known that cytoskeletal dynamics in the growth cone are necessary for axon extension, motility, and guidance, the majority of research has focused on filopodia and lamellipodia. Only recently have invadopodia been identified at the growth cone (Santiago-Medina et al., 2015; Nichols and Smith, 2019a). Our results introduce the modulation of invadopodia at the growth cone by elaborating on the key molecular regulators that control the stabilization of invadopodia during the axon entry process. As a result of this molecular brake on invadopodia during navigation, DRG neurons can travel to the DREZ, then shorten their filopodia projections and stabilize the invadopodia to break through the glial limitans.
Given our data and based on previous literature, we hypothesize that levels of activation of the receptor, DCC, increases cAMP. The activation of cAMP, likely via PKA, then determines whether Rac1 is activated to drive invadopodia disassembly. In this way, various levels of DCC activation correspond with different states of actin (Fig. 7L). Examining the location of netrin1 mRNA, we detected netrin primarily localized to the floor plate during pioneer axon navigation. If Netrin were secreted from the floor plate and acted as a long-range diffusible factor as previously proposed (Kennedy et al., 1994), Netrin would be expected at lower concentrations in dorsal locations, potentially resulting in differential levels of DCC activation as the pioneer axon navigates dorsally to the DREZ. Studies in Caenorhabditis elegans have found that the presence of UNC-6 (Netrin-1) leads to stabilization of UNC-40 (DCC), which undergoes oscillatory clustering and increasing F-actin production in the absence of UNC-6 (Ren et al., 1999; Wadsworth, 2002; Wang et al., 2014a). It is thus possible that a Netrin-1 gradient regulates DCC activation via stabilization of clustering.
Alternatively, recent data have shown that neuronal progenitors from the ventricular zone (VZ) are the source of Netrin-1 for commissural axon navigation guiding axons via haptotaxis (Dominici et al., 2017; Varadarajan et al., 2017). DREZ defects can be visualized in similar manipulations, however the pioneer events of navigation were not fully investigated. Although we did not detect netrin mRNA in the SV regions during pioneer axon navigation, it is possible that Netrin localizes there later in development as additional axons extend to the DREZ, causing the well-described DREZ defects (Varadarajan and Butler, 2017). Since our experiments cannot distinguish the source of Netrin, future investigation regarding the source of Netrin that guides specific axonal populations will be important.
Our findings defining the necessity for the coordinated invadopodia states are corroborated by previously published literature in C. elegans, which demonstrates that anchor cell (AC) entry into the vulval epithelium requires actin invasion to occur in a coordinated process (Ziel et al., 2009; Hagedorn et al., 2013; Wang et al., 2014a,b; Lohmer et al., 2016). In that context, UNC-40 localized to the area of invasion in order for invadopodia to create a larger hole to displace and invade the vulval tissue. Once this has been accomplished, invadopodia then disassemble. However, in unc-40 mutants, invadopodia cannot form a distinct invasion protrusion and instead create many small holes at delayed time points, never displacing the basement membrane or fully disassembling. This mechanism could be similar at the DREZ, causing dcc mutants to not generate a large enough hole in the glial limitans for the axon to travel through. Further work showed that unc-40 recruits the Rac GTPase effector, ced-10, to the AC to promote F-actin formation for invasive structures (Wang et al., 2014b). While Rac1 is shown to promote invadopodia formation in this context, other studies in cancer (Moshfegh et al., 2014) and DRG growth cone (Nichols and Smith, 2019a) display an inhibitory role of Rac1 on invadopodia stabilization. Our results corroborate the latter, showing a key role of Rac1 in driving invadopodia disassembly during navigation.
Our data show that the inhibition of invadopodia until the proper time is mediated by dcc. While many studies have used homozygous dcc mutants to examine actin populations, one possible reason this phenomenon has not previously been observed may be attributed to the nuanced levels of cAMP and DCC necessary for the balance of filopodia and invadopodia. If the levels of DCC activation are regulating the binary switch of navigation from filopodia to invadopodia, then the DRG neurite of homozygous mutants may fully retract to the cell soma since a lack of filopodia would lead to the incapability of the neurite specifying as a single DRG pioneer axon (Zhang et al., 2019). This possibility is supported by our data, showing that dcc−/− DRG neurons have a reduction of axons. As such, a heterozygous dcc mutant could best represent the intermediate phenotype to study filopodia and invadopodia switches. The ability to visualize the pioneer axon during assembly of the DREZ helped to identify a key role of DCC in control of invadopodia components, and may explain the mechanism by which disorganization of the DREZ occurs in Dcc mutants in mice (Watanabe et al., 2006; Varadarajan and Butler, 2017). The concept that Netrin receptors function with spatiotemporal precision to guide sensory axon dynamics is best represented by the work performed by Watanabe et al. (2006), where the authors propose that Netrin is transiently expressed during the “waiting period” to inhibit axon outgrowth into the dorsal mantle layer via Unc5c. We suggest a similar mechanism before entry into the dorsal spinal cord, and propose that dcc temporally functions within transient invadopodia of the pioneer axon to inhibit invadopodia components until the axon approaches the DREZ. Whether dcc functions differently in pioneer axons versus secondary axons during assembly of the DREZ is an interesting topic for future investigation.
Here, we show that actin populations at the growth cone must form at distinct time points for the coordinated invadopodia state to occur. We hypothesize that the process by which invadopodia become inhibited in order for filopodia to increase in length is regulated as follows: DCC increases cAMP/PKA, which activates Rac1 to inhibit invadopodia. Given that both increasing and decreasing cAMP leads to uncoordinated invadopodia states, our data suggests the cAMP must be present at intermediate levels to maintain the invadopodia brake and promote filopodia elongation. We hypothesize that the brake is released when cAMP is present at lower or higher levels, leading cAMP to switch roles and instead promote invadopodia stabilization and inhibit filopodia length. While the mechanism of cAMP regulation of the invadopodia brake has not been described previously, cAMP has been shown to have distinct phenotypes dependent on its spatiotemporal level of activation (Nicol et al., 2011; Kobayashi et al., 2013). Lower and higher intracellular levels of cAMP lead to repulsive or attractive growth cone turning, respectively, in the presence of Netrin-1 (Ming et al., 1997; Höpker et al., 1999; Henley et al., 2004; Forbes et al., 2012). Furthermore, DCC activation in filopodia has been shown to cause a transient increase in cAMP leading to a brief increase in calcium, while DCC activation in the center of the growth cone leads to a calcium-dependent increase in cAMP, and consequently sustained increase in calcium (Nicol et al., 2011). It is therefore possible that the location of cAMP activation could also contribute to the invadopodia brake, a future topic that will be interesting to pursue.
While the previous literature supports a singular, cell-autonomous, pathway with DCC, Rac1 and cAMP, our data do not exclude the possibility that a non-cell-autonomous function of cAMP could be controlling invadopodia states and growth cone entry. Nevertheless, our data showing that cAMP manipulations can induce a phenotype in dcc cell-autonomously rescued DRG neurons suggests a degree of specificity. The cell-specific manipulations of Rac1 and DCC, however, are consistent with the model that Rac1 and DCC in DRG cells control invadopodia states and axon entry of DRG pioneer axons. We additionally report behavioral phenotypes associated with experimental manipulation of the invadopodia brake pathway. Although pharmacological treatment with cAMP has been used previously to determine behavioral deficiencies in zebrafish (Wolman et al., 2014), a caveat to this standard remains to be the inability to determine cell-autonomous function. Furthermore, the mosaicism associated with both pa-Rac1 and the sox10:dcc rescue construct leads to DRG neurons entering at varying amounts compared with wild-type conditions. Therefore, cell-autonomous conclusions cannot be associated with behavioral phenotypes, and further investigation is required to elucidate cell-specificity.
Taken together, our data incorporate a new mechanism by which axonal navigation is regulated through distinct temporal formation and disassembly of filopodia and invadopodia. In addition to the role of a chemoguidance receptor, DCC activation modulates the coordinated invadopodia state. Understanding the nuances behind the control on these specific actin populations at the growth cone provides further insight as to how axons can balance between navigation to targets and invasion into tissue environments during their developmental odyssey.
Footnotes
This work was supported by the University of Notre Dame, the Elizabeth and Michael Gallagher Family (C.J.S.), Centers for Zebrafish Research and Stem Cells Regenerative Medicine at the University of Notre Dame (C.J.S.), the Indiana Spinal Cord and Brain Injury Research with the Indiana State Board of Health (C.J.S.), the Alfred P. Sloan Foundation (C.J.S.), the National Institutes of Health Grants DP2NS117177 (C.J.S.) and R01NS116354-01A (K.C.M.) and the North Carolina State University (K.C.M.). We thank Brent Redford, Andrew Barazia, Sam Connell, and 3i for fielding imaging questions; the Wingert and Smith labs for their comments and reagent guidance; and Deborah Bang and Karen Heed for zebrafish care; Michelle Wang for her help constructing figures; Addgene for UAS-PA Rac1-mcherry plasmid (plasmid #41878); and Michael Granato for dcc mutants.
The authors declare no competing financial interests.
References
- Akin O, Zipursky SL (2016) Frazzled promotes growth cone attachment at the source of a Netrin gradient in the Drosophila visual system. Elife 5:e20762. 10.7554/eLife.20762 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Araiza-Olivera D, Feng Y, Semenova G, Prudnikova TY, Rhodes J, Chernoff J (2018) Suppression of RAC-1 driven malignant melanoma by group A PAK inhibitors. Oncogene 37:944–952. 10.1038/onc.2017.400 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bernardos RL, Raymond PA (2006) GFAP transgenic zebrafish. Gene Expr Patterns 6:1007–1013. 10.1016/j.modgep.2006.04.006 [DOI] [PubMed] [Google Scholar]
- Burgess HA, Granato M (2007) Modulation of locomotor activity in larval zebrafish during light adaptation. J Exp Biol 210:2526–2539. 10.1242/jeb.003939 [DOI] [PubMed] [Google Scholar]
- Cajal R (1911) Histology of the nervous system of man and vertebrates. New York: Oxford University Press. [Google Scholar]
- Chédotal A (2019) Roles of axon guidance molecules in neuronal wiring in the developing spinal cord. Nat Rev Neurosci 20:380–396. 10.1038/s41583-019-0168-7 [DOI] [PubMed] [Google Scholar]
- Demarco RS, Struckhoff EC, Lundquist EA (2012) The Rac GTP exchange factor TIAM-1 acts with CDC-42 and the guidance receptor UNC-40/DCC in neuronal protrusion and axon guidance. PLoS Genet 8:e1002665. 10.1371/journal.pgen.1002665 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dent EW, Kalil K (2001) Axon branching requires interactions between dynamic microtubules and actin filaments. J Neurosci 21:9757–9769. 10.1523/JNEUROSCI.21-24-09757.2001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dominici C, Moreno-Bravo JA, Puiggros SR, Rappeneau Q, Rama N, Vieugue P, Bernet A, Mehlen P, Chédotal A (2017) Floor plate-derived netrin-1 is dispensable for commissural axon guidance. Nature 545:350–354. 10.1038/nature22331 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Duman-Scheel M (2009) Netrin and DCC: axon guidance regulators at the intersection of nervous system development and cancer. Curr Drug Targets 10:602–610. 10.2174/138945009788680428 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Forbes EM, Thompson AW, Yuan J, Goodhill GJ (2012) Calcium and cAMP levels interact to determine attraction versus repulsion in axon guidance. Neuron 74:490–503. 10.1016/j.neuron.2012.02.035 [DOI] [PubMed] [Google Scholar]
- Golding J, Shewan D, Cohen J (1997) Maturation of the mammalian dorsal root entry zone - from entry to no entry. Trends Neurosci 20:303–308. 10.1016/S0166-2236(96)01044-2 [DOI] [PubMed] [Google Scholar]
- Hagedorn EJ, Ziel JW, Morrissey MA, Linden LM, Wang Z, Chi Q, Johnson SA, Sherwood DR (2013) The netrin receptor DCC focuses invadopodia-driven basement membrane transmigration in vivo. J Cell Biol 201:903–913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Helker CSM, Schuermann A, Karpanen T, Zeuschner D, Belting H-G, Affolter M, Schulte-Merker S, Herzog W (2013) The zebrafish common cardinal veins develop by a novel mechanism: lumen ensheathment. Development 140:2776–2786. 10.1242/dev.091876 [DOI] [PubMed] [Google Scholar]
- Henley JR, Huang KH, Wang D, Poo MM (2004) Calcium mediates bidirectional growth cone turning induced by myelin-associated glycoprotein. Neuron 44:909–916. 10.1016/j.neuron.2004.11.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hines JH, Ravanelli AM, Schwindt R, Scott EK, Appel B (2015) Neuronal activity biases axon selection for myelination in vivo. Nat Neurosci 18:683–689. 10.1038/nn.3992 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Höpker VH, Shewan D, Tessier-Lavigne M, Poo M, Holt C (1999) Growth-cone attraction to netrin-1 is converted to repulsion by laminin-1. Nature 401:69–73. 10.1038/43441 [DOI] [PubMed] [Google Scholar]
- Jain RA, Bell H, Lim A, Chien CB, Granato M (2014) Mirror movement-like defects in startle behavior of zebrafish dcc mutants are caused by aberrant midline guidance of identified descending hindbrain neurons. J Neurosci 34:2898–2909. 10.1523/JNEUROSCI.2420-13.2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Johnson K, Barragan J, Bashiruddin S, Smith CJ, Tyrrell C, Parsons MJ, Doris R, Kucenas S, Downes GB, Velez CM, Schneider C, Sakai C, Pathak N, Anderson K, Stein R, Devoto SH, Mumm JS, Barresi MJF (2016) Gfap-positive radial glial cells are an essential progenitor population for later-born neurons and glia in the zebrafish spinal cord. Glia 64:1170–1189. 10.1002/glia.22990 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Junge HJ, Yung AR, Goodrich LV, Chen Z (2016) Netrin1/DCC signaling promotes neuronal migration in the dorsal spinal cord. Neural Dev 11:19. 10.1186/s13064-016-0074-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kennedy TE, Serafini T, de la Torre JR, Tessier-Lavigne M (1994) Netrins are diffusible chemotropic factors for commissural axons in the embryonic spinal cord. Cell 78:425–435. 10.1016/0092-8674(94)90421-9 [DOI] [PubMed] [Google Scholar]
- Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schilling TF (1995) Stages of embryonic development of the zebrafish. Dev Dyn 203:253–310. 10.1002/aja.1002030302 [DOI] [PubMed] [Google Scholar]
- Kobayashi T, Nagase F, Hotta K, Oka K (2013) Crosstalk between second messengers predicts the motility of the growth cone. Sci. Rep 3:1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kwan KM, Fujimoto E, Grabher C, Mangum BD, Hardy ME, Campbell DS, Parant JM, Yost HJ, Kanki JP, Chien CB (2007) The Tol2kit: a multisite gateway-based construction Kit for Tol2 transposon transgenesis constructs. Dev Dyn 236:3088–3099. 10.1002/dvdy.21343 [DOI] [PubMed] [Google Scholar]
- Lohmer LL, Clay MR, Naegeli KM, Chi Q, Ziel JW, Hagedorn EJ, Park JE, Jayadev R, Sherwood DR (2016) A sensitized screen for genes promoting invadopodia function in vivo: CDC-42 and Rab GDI-1 direct distinct aspects of invadopodia formation. PLoS Genet 12:e1005786. 10.1371/journal.pgen.1005786 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Marjoram L, Alvers A, Deerhake ME, Bagwell J, Mankiewicz J, Cocchiaro JL, Beerman RW, Willer J, Sumigray KD, Katsanis N, Tobin DM, Rawls JF, Goll MG, Bagnat M (2015) Epigenetic control of intestinal barrier function and inflammation in zebrafish. Proc Natl Acad Sci USA 112:2770–2775. 10.1073/pnas.1424089112 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Matthews HK, Marchant L, Carmona-Fontaine C, Kuriyama S, Larraín J, Holt MR, Parsons M, Mayor R (2008) Directional migration of neural crest cells in vivo is regulated by Syndecan-4/Rac1 and non-canonical Wnt signaling/RhoA. Development 135:1771–1780. 10.1242/dev.017350 [DOI] [PubMed] [Google Scholar]
- Mattila PK, Lappalainen P (2008) Filopodia: molecular architecture and cellular functions. Nat Rev Mol Cell Biol 9:446–454. 10.1038/nrm2406 [DOI] [PubMed] [Google Scholar]
- McGraw HF, Nechiporuk A, Raible DW (2008) Zebrafish dorsal root ganglia neural precursor cells adopt a glial fate in the absence of neurogenin1. J Neurosci 28:12558–12569. 10.1523/JNEUROSCI.2079-08.2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ming GL, Song HJ, Berninger B, Holt CE, Tessier-Lavigne M, Poo MM (1997) cAMP-dependent growth cone guidance by netrin-1. Neuron 19:1225–1235. 10.1016/S0896-6273(00)80414-6 [DOI] [PubMed] [Google Scholar]
- Moshfegh Y, Bravo-Cordero JJ, Miskolci V, Condeelis J, Hodgson L (2014) A Trio Rac1 Pak1 signalling axis drives invadopodia-disassembly. Nat Cell Biol 16:574–586. 10.1038/ncb2972 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Muramatsu R, Nakahara S, Ichikawa J, Watanabe K, Matsuki N, Koyama R (2010) The ratio of “deleted in colorectal cancer” to “uncoordinated-5A” netrin-1 receptors on the growth cone regulates mossy fibre directionality. Brain 133:60–75. 10.1093/brain/awp266 [DOI] [PubMed] [Google Scholar]
- Nichols EL, Smith CJ (2019a) Pioneer axons employ Cajal's battering ram to enter the spinal cord. Nat Commun 10:562. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nichols EL, Smith CJ (2019b) Synaptic-like vesicles facilitate pioneer axon invasion. Curr Biol 29:2652–2664. 10.1016/j.cub.2019.06.078 [DOI] [PubMed] [Google Scholar]
- Nichols EL, Smith CJ (2020) Functional Regeneration of the Sensory Root via Axonal Invasion. Cell Reports 30:9–17. e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nicol X, Hong KP, Spitzer NC (2011) Spatial and temporal second messenger codes for growth cone turning. Proc Natl Acad Sci USA 108:13776–13781. 10.1073/pnas.1100247108 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Norris AD, Lundquist EA (2011) UNC-6/netrin and its receptors UNC-5 and UNC-40/DCC modulate growth cone protrusion in vivo in C. elegans. Development 138:4433–4442. 10.1242/dev.068841 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ren XC, Kim S, Fox E, Hedgecock EM, Wadsworth WG (1999) Role of netrin UNC-6 in patterning the longitudinal nerves of Caenorhabditis elegans. J Neurobiol 39:107–118. [DOI] [PubMed] [Google Scholar]
- Santiago-Medina M, Gregus KA, Nichol RH, O'Toole SM, Gomez TM (2015) Regulation of ECM degradation and axon guidance by growth cone invadosomes. Development 142:486–496. 10.1242/dev.108266 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Santos TE, Schaffran B, Broguière N, Meyn L, Zenobi-Wong M, Bradke F (2020) Axon growth of CNS neurons in three dimensions is amoeboid and independent of adhesions. Cell Rep 32:107907. [DOI] [PubMed] [Google Scholar]
- Shekarabi M, Kennedy TE (2002) The netrin-1 receptor DCC promotes filopodia formation and cell spreading by activating Cdc42 and Rac1. Mol Cell Neurosci 19:1–17. 10.1006/mcne.2001.1075 [DOI] [PubMed] [Google Scholar]
- Smith CJ, Morris AD, Welsh TG, Kucenas S (2014) Contact-mediated inhibition between oligodendrocyte progenitor cells and motor exit point glia establishes the spinal cord transition zone. PLoS Biol 12:e1001961. 10.1371/journal.pbio.1001961 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Smith CJ, Wheeler MA, Marjoram L, Bagnat M, Deppmann CD, Kucenas S (2017) TNFa/TNFR2 signaling is required for glial ensheathment at the dorsal root entry zone. PLoS Genet 13:e1006712. 10.1371/journal.pgen.1006712 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sotelo C (2003) Viewing the brain through the master hand of Ramón y Cajal. Nat Rev Neurosci 4:71–77. 10.1038/nrn1010 [DOI] [PubMed] [Google Scholar]
- Varadarajan SG, Butler SJ (2017) Netrin1 establishes multiple boundaries for axon growth in the developing spinal cord. Dev Biol 430:177–187. 10.1016/j.ydbio.2017.08.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Varadarajan SG, Kong JH, Phan KD, Kao TJ, Panaitof SC, Cardin J, Eltzschig H, Kania A, Novitch BG, Butler SJ (2017) Netrin1 produced by neural progenitors, not floor plate cells, is required for axon guidance in the spinal cord. Neuron 94:790–799.e3. 10.1016/j.neuron.2017.03.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wadsworth WG (2002) Moving around in a worm: netrin UNC-6 and circumferential axon guidance in C. elegans. Trends Neurosci 25:423–429. 10.1016/S0166-2236(02)02206-3 [DOI] [PubMed] [Google Scholar]
- Wang Z, Linden LM, Naegeli KM, Ziel JW, Chi Q, Hagedorn EJ, Savage NS, Sherwood DR (2014a) UNC-6 (netrin) stabilizes oscillatory clustering of the UNC-40 (DCC) receptor to orient polarity. J Cell Biol 206:619–633. 10.1083/jcb.201405026 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang Z, Chi Q, Sherwood DR (2014b) MIG-10 (lamellipodin) has netrin-independent functions and is a FOS-1A transcriptional target during anchor cell invasion in C. elegans. Development 141:1342–1353. 10.1242/dev.102434 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Watanabe K, Tamamaki N, Furuta T, Ackerman SL, Ikenaka K, Ono K (2006) Dorsally derived netrin 1 provides an inhibitory cue and elaborates the “waiting period” for primary sensory axons in the developing spinal cord. Development 133:1379–1387. 10.1242/dev.02312 [DOI] [PubMed] [Google Scholar]
- Wolman MA, de Groh ED, McBride SM, Jongens TA, Granato M, Epstein JA (2014) Modulation of cAMP and Ras signaling pathways improve distinct behavioral deficits in a zebrafish model of neurofibromatosis type 1. Cell Rep 8:1265–1270. 10.1016/j.celrep.2014.07.054 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wu RS, Lam II, Clay H, Duong DN, Deo RC, Coughlin SR (2018) A rapid method for directed gene knockout for screening in G0 zebrafish. Dev Cell 46:112–125.e4. 10.1016/j.devcel.2018.06.003 [DOI] [PubMed] [Google Scholar]
- Zhang Y, Nichols EL, Zellmer AM, Guldner IH, Kankel C, Zhang S, Howard SS, Smith CJ (2019) Generating intravital super-resolution movies with conventional microscopy reveals actin dynamics that construct pioneer axons. Development 146:dev171512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ziel JW, Hagedorn EJ, Audhya A, Sherwood DR (2009) UNC-6 (netrin) orients the invasive membrane of the anchor cell in C. elegans. Nat. Cell Biol 11:183–189. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Time lapse of control growth cone. Confocal maximum projection of 400-min time lapse of control 48 hpf Tg(sox10:lifeact-gfp) animal. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of photoactivated-Rac1 growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal with photo-activated Rac1. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of sp-cAMP-treated growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal treated with 100 μm sp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of rp-cAMP-treated growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal treated with 25 μm rp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of photoactivated-Rac1 growth cone treated with rp-cAMP. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp) animal with photo-activated Rac1 and treated with 25 μm rp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of dcc+/− growth cone. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of dcc+/− growth cone injected with sox10:dcc. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal injected with sox10:dcc. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of dcc+/− growth cone treated with sp-cAMP. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal treated with 100 μm sp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.
Time lapse of dcc+/− growth cone treated with rp-cAMP. Confocal maximum projection of 400-min time lapse of 48 hpf Tg(sox10:lifeact-gfp); dcc+/− animal treated with 25 μm rp-cAMP. Time lapse covers 200 min before and after growth cone entry. Green open circle denotes growth cone.