In vivo imaging of cell behaviors and F-actin reveals LIM-HD transcription factor regulation of peripheral versus central sensory axon development
Development of specific neuronal morphology requires precise control over cell motility processes, including axon formation, outgrowth and branching. Dynamic remodeling of the filamentous actin (F-actin) cytoskeleton is critical for these processes; however, little is known about the mechanisms controlling motile axon behaviors and F-actin dynamics in vivo. Neuronal structure is specified in part by intrinsic transcription factor activity, yet the molecular and cellular steps between transcription and axon behavior are not well understood. Zebrafish Rohon-Beard (RB) sensory neurons have a unique morphology, with central axons that extend in the spinal cord and a peripheral axon that innervates the skin. LIM homeodomain (LIM-HD) transcription factor activity is required for formation of peripheral RB axons. To understand how neuronal morphogenesis is controlled in vivo and how LIM-HD transcription factor activity differentially regulates peripheral versus central axons, we used live imaging of axon behavior and F-actin distribution in vivo.
We used an F-actin biosensor containing the actin-binding domain of utrophin to characterize actin rearrangements during specific developmental processes in vivo, including axon initiation, consolidation and branching. We found that peripheral axons initiate from a specific cellular compartment and that F-actin accumulation and protrusive activity precede peripheral axon initiation. Moreover, disruption of LIM-HD transcriptional activity has different effects on the motility of peripheral versus central axons; it inhibits peripheral axon initiation, growth and branching, while increasing the growth rate of central axons. Our imaging revealed that LIM-HD transcription factor activity is not required for F-actin based protrusive activity or F-actin accumulation during peripheral axon initiation, but can affect positioning of F-actin accumulation and axon formation.
Our ability to image the dynamics of F-actin distribution during neuronal morphogenesis in vivo is unprecedented, and our experiments provide insight into the regulation of cell motility as neurons develop in the intact embryo. We identify specific motile cell behaviors affected by LIM-HD transcription factor activity and reveal how transcription factors differentially control the formation and growth of two axons from the same neuron.
KeywordsGrowth Cone Peripheral Axon Calponin Homology Domain Central Axon Neuronal Morphogenesis
dominant negative cofactor of LIM
green fluorescent protein
Tag red fluorescent protein
utrophin calponin homology.
Neurons must develop characteristic and often complex morphologies to form functional circuits of the nervous system. Development of proper neuronal structure requires precise regulation of multiple cellular processes, including axon formation, outgrowth and branching [1, 2, 3, 4]. Axon formation creates a distinct molecular and functional compartment of the neuron and the position of axon initiation is often critical for establishing neuronal polarity . Axons initiate and extend led by highly protrusive growth cones that sense environmental guidance cues. As growth cones advance, the axon shaft undergoes consolidation, a process that suppresses protrusive activity behind the growth cone and maintains axon polarity . Axon branching and/or the extension of multiple axons allow neurons to innervate multiple targets [2, 6]. The formation of axon branches in specific locations and prevention of branching at inappropriate sites requires local regulation of signals that promote or suppress protrusions. These processes are critical for neuronal development, yet the mechanisms that regulate axon formation, consolidation and branching in the complex in vivo environment remain poorly understood.
Precise regulation of filamentous actin (F-actin) dynamics is essential for the cell motility processes underlying neuronal morphogenesis [1, 4, 7, 8]. A necessary first step in axon formation is accumulation of F-actin at the initiation site, which was first shown in situ during axonogenesis in grasshopper sensory neurons . In addition, the organization of F-actin into filopodia is a key prerequisite for neurite formation in cultured mouse cortical neurons . In Caenorhabditis elegans, the specific site of F-actin accumulation and axon formation is determined by extracellular UNC-6/netrin, which locally activates a signaling pathway that controls the actin regulatory protein MIG10/lamellipodin [11, 12]. In mammalian in vitro models of neuronal polarization, actin regulation is also a critical convergence point for polarity signals (reviewed in [13, 14]). Nonetheless, our understanding of the mechanisms controlling the process of initial axon emergence from the cell body and the underlying actin rearrangements in vivo is far from complete.
During outgrowth, the polarized structure of axons must be established and maintained by consolidation along the axon shaft and restriction of F-actin-based protrusions to the growth cone [1, 4]. In vitro, consolidation is mediated in part by Rho-kinase inhibition of F-actin at the growth cone neck . Along the axon shaft, calpain proteolytic activity is required to inhibit protrusive activity by reducing levels of cortactin, an actin regulator . Thus, active suppression of F-actin protrusive activity along the axon shaft is required for proper axon extension and to prevent ectopic branching. To date, the process of axon consolidation in vivo has not been investigated.
The formation of axon branches also relies on dynamic rearrangements of the actin cytoskeleton [4, 6, 17, 18, 19]. Axon branching can occur by at least two distinct modes: growth cone bifurcation or interstitial branching off the axon shaft [2, 6, 17]. Although bifurcation is necessary for sensory axon target innervation in vivo[20, 21], the cytoskeletal rearrangements involved in this process are largely unknown. Interstitial branching appears to be the predominant branching mode in most neuron types studied thus far [2, 6, 22, 23, 24]. During interstitial branching in cultured cortical neurons, F-actin accumulates at branch initiation points and actin polymerization is required for branch formation [25, 26]. Branch promoting factors have been shown to act through derepression of calpain-mediated consolidation, thereby releasing the inhibition of F-actin polymerization . Still, little is known about the dynamics of axon branching or the underlying F-actin rearrangements in vivo. Studies using live imaging of F-actin in vitro have been instrumental to our understanding of axon formation, outgrowth and branching [1, 4, 15, 25, 26]. Here, we image F-actin during neuronal morphogenesis in intact embryos, where axons must integrate multiple signals that influence their motility.
Recent studies have brought to light the importance of intrinsic transcription factor regulation in neuronal morphogenesis . Transcription factor expression not only defines neuronal identity, but also can dictate axon trajectory choices downstream of identity decisions. For example, specific combinations of LIM homeodomain (LIM-HD) and homeobox domain (Hox) transcription factors are expressed in subsets of spinal motor neurons, and these determine the motor axon trajectories and innervation of specific muscle targets [28, 29, 30, 31]. In retinal ganglion cells, cell-specific expression of Zic2, a zinc-finger transcription factor, and Isl2, a LIM-HD transcription factor, regulate the midline crossing choice of retinal ganglion cell axons at the optic chiasm [32, 33]. These transcription factors mediate axon pathway choices at least in part by regulating expression of Eph family axon guidance receptors [30, 34, 35]. In general, however, the mechanisms by which transcription controls axon motility remain largely unknown.
The zebrafish Rohon-Beard (RB) spinal sensory neurons provide an ideal model to investigate mechanisms controlling neuronal development in vivo because of their unique and stereotyped morphology [36, 37]. RB neurons extend two types of axons, centrals and peripherals, with distinct trajectories and behaviors . Each neuron extends ascending and descending central axons that grow longitudinally within the spinal cord to form the dorsal longitudinal fasciculus, and a single peripheral axon that exits the spinal cord and branches in the skin. The differential pathways of central and peripheral RB axons are determined in part by differing responses to extracellular signals . Furthermore, LIM-HD transcription factors of the Islet family and their cofactors, CLIMs, are essential intrinsic regulators of RB neuron morphology [39, 40]. Ubiquitous expression of a dominant negative CLIM (DN-CLIM) protein disrupts LIM-HD transcriptional activity and results in the reduction or elimination of RB peripheral axons without an apparent effect on the central axons or RB cell fate [39, 40]. Thus, these transcription factors are selectively required for peripheral axon formation and/or guidance. In this study, we analyze how LIM-HD transcription factors control cell motility processes and the F-actin cytoskeleton during neuronal morphogenesis in vivo.
To investigate the mechanisms controlling RB morphology and to understand how two axons from one cell are guided differently, we used live imaging of axon behaviors and F-actin distribution in vivo. We were able to visualize changes in F-actin distribution during dynamic processes such as axon consolidation and branching. We show that extensive F-actin protrusive activity precedes formation of the peripheral axon, and that F-actin concentrates at the site of peripheral axon initiation. In addition, we found that DN-CLIM differentially affects the motility of peripheral and central axons. In most cases, peripheral axons failed to initiate in DN-CLIM-expressing embryos; however, those that formed had significantly reduced growth rates and branching. In contrast, central axon growth rates were faster in DN-CLIM-expressing embryos. Finally, we found that disruption of LIM-HD activity did not affect F-actin-based protrusive activity or the ability of F-actin to accumulate at peripheral axon initiation sites, indicating that F-actin dynamics were unperturbed by DN-CLIM. However, DN-CLIM did influence the location of F-actin accumulation and axon initiation, suggesting LIM-HD activity may function in axon positioning.
Live imaging of individual RB neuron development
The RB neurons compose the primary sensory system in the trunk of anamniotic vertebrate embryos [36, 41]. RB cell bodies lie in bilateral rows in the dorsal spinal cord and begin extending central axons between approximately 16 and 17 hours post-fertilization (hpf) . Peripheral axons form approximately 2 hours after central axon initiation and continue to extend and branch over a period of several hours. To visualize the axon morphology of individual RB neurons, we used a transient mosaic cell labeling approach . We injected plasmid DNA encoding a membrane-targeted fluorophore (GFP-CAAX or TagRFP-CAAX) driven by a cis-regulatory element from the neurogenin1 gene (-3.1 ngn1)  (see Materials and methods). This method allows visualization of the entire neuron structure during all stages of development.
During their initial outgrowth (phase I), within approximately 25 μm from the cell body, the central axons are thick and display protrusive activity along the entire length of the axon shaft (Figure 1B,D). The central growth cones extend relatively slowly during this phase (mean phase I extension rate = 12.8 ± 2.1 μm/h). In the second phase of central axon outgrowth, when axons are between approximately 25 and 50 μm long, membrane protrusions along the proximal axons are reduced and the axons display a more polarized morphology, with consolidated axon shafts and distinct protrusive growth cones (Figure 1C,D; Additional file 1). This morphological transition is accompanied by an increase in extension rates (mean phase II rate = 23.1 ± 1.1 μm/h). During this phase, central growth cones come into contact with axons from neighboring RB neurons and grow along them to form the dorsal longitudinal fasciculus (Figure 1E,F). In addition, transient filopodial protrusions continue to form along the axon shaft during this phase, and these can make lateral contact with other central axons (Figure 1F), suggesting they could mediate recognition and fasciculation. In the third phase, after central axons have reached a length of 50 μm or greater, their growth cones extend more rapidly (mean phase III rate = 50.3 ± 2.1 μm/h). Although filopodial protrusions continue to form along the central axons in phase III, their frequency in general decreases in proportion to increasing central axon length (Figure 1D).
Formation of the peripheral axon is a particularly important step of RB development because it is the point at which discrete central and peripheral compartments with different axon trajectories and behaviors are established. The peripheral axon forms orthogonally to the central axons, exits the spinal cord, and grows into a very different environment than the central pathway. Peripheral axons initiate during phase II or III of central axon outgrowth (Figure 1D), and become established during phase III. Single cell labeling allowed us to determine the site of peripheral axon formation. Like the central axons, the peripheral axon forms at the basal edge of the RB neuron; however, the precise anterior-posterior site at which it forms is variable from cell to cell. In 40% of cells (n = 47), the peripheral axon emerged directly from the cell body, and in the other 60% it arose as a branch from one of the central axons. The branches did not form preferentially from either the ascending or descending central axon, but initiated from both axons in equal proportions. However, in the majority of neurons (83%), the peripheral axon initiation site was restricted to a region near the cell body. On average, the distance from the cell body center to the peripheral axon initiation site was 10.7 ± 0.67 μm (approximately one cell body diameter). Thus, it appears there is a specific compartment of the RB cell that is capable of generating a peripheral axon.
Imaging F-actin distribution during RB development
Motility comparison between control neurons expressing a fluorophore only versus neurons expressing the UtrCH F-actin reporter
Average central growth rate (μm/h)
Average peripheral growth rate (μm/h)
Average number of filopodia per hour
23.8 ± 1.4 (n = 8)
40.9 ± 2.7 (n = 19)
40.9 ± 2.4 (n = 8)
22.4 ± 1.8 (n = 8)
40.6 ± 2.5 (n = 7)
37.1 ± 2.3 (n = 9)
Additional file 3:Movie of F-actin distribution in a developing wild-type RB neuron. Movie of a -3.1ngn1:mch-utrCH-expressing RB neuron in Figure 3A. Images were acquired every minute over 240 minutes between 17 and 21 hpf. A peripheral axon grows out of the focal plane during part of the movie.(MOV 664 KB)
Imaging RB development in embryos with disrupted LIM-HD transcription factor activity
Previous studies examining LIM-HD regulation of RB development concluded that DN-CLIM did not affect RB central axons [39, 40]. However, our live imaging showed that central axon growth rates were in fact affected in DN-CLIM-expressing embryos. We discovered that during phase II of central axon extension, when peripheral axons normally initiate outgrowth, central axon growth rates are significantly faster in DN-CLIM-expressing embryos (Figure 5G). These results indicate that DN-CLIM has specific, opposite effects on the motility of peripheral versus central axons. Moreover, these data argue that the effects of DN-CLIM on peripheral axon outgrowth are not simply due to a developmental delay or to poor health of the neurons.
Imaging F-actin distribution during RB development in embryos with disrupted LIM-HD transcription factor activity
Axon behaviors and F-actin dynamics in wild-type RB neurons
In this study, we were able to directly visualize F-actin during axon development in vivo using the UtrCH biosensor. This reporter has been shown to bind selectively to F-actin without stabilizing it , and has revealed F-actin dynamics during multiple cell motility processes, including wound healing and cytokinesis in Xenopus embryos [49, 50], neural crest cell or neutrophil migration in zebrafish [51, 52], and growth cone turning in cultured neurons . Here we show that neural-specific expression of UtrCH provides excellent spatial-temporal resolution of F-actin during axon development in vivo. Moreover, the UtrCH probe did not disrupt actin-based neuronal motility when expressed mosaically under control of the -3.1ngn1 cis-regulatory element. Our experiments provide the first characterization of F-actin dynamics during axon consolidation, initiation, and branching in vivo.
Consolidation is essential for establishing and maintaining axon shape and polarity, yet how this process occurs in vivo is not understood. Live imaging of newly forming RB central axons revealed that these axons undergo a dynamic transition in their consolidation state during the first few hours of outgrowth. Our results suggest that consolidation and acquisition of polar axon morphology are not immediate processes but occur gradually as axons extend. Furthermore, our finding that transient F-actin-based filopodia continue to form along the axon shaft after many hours suggests RB axons do not become completely consolidated in the in vivo environment. This state likely reflects a fine balance between signals that maintain active consolidation and those that stimulate F-actin protrusions and branching. What are the mechanisms that control consolidation in vivo? In cultured neurons, consolidation is an active process, requiring Rho-kinase-mediated inhibition of actin polymerization at the growth cone neck , and calpain-mediated degradation of actin regulatory factors in the axon shaft . Moreover, calpain inhibition in adult mice induces excess dendritic branching in hippocampal neurons , supporting a role for this pathway in vivo. Although calpain activity can be induced by low protein kinase A levels in the axon shaft , the upstream signals that activate consolidation or that control protein kinase A levels are not known. Our results suggest that the stimuli triggering consolidation are not immediately active in vivo. RB central axons initially extend along neuroepithelial cells before contacting and fasciculating with each other . Perhaps axons must reach a critical length and/or engage sufficient adhesive contacts with neuroepithelial cells to activate consolidation signals. Alternatively, contact with other central axons may stimulate consolidation. Our RB model will be useful for future studies to define the molecular signals that regulate consolidation of pioneering axons in vivo.
Axon initiation is an important first step in neuronal morphogenesis, yet few studies have examined this process in vivo. Our observations that F-actin protrusive activity and accumulation precede RB peripheral axon emergence are consistent with previous findings that remodeling of the actin cytoskeleton has a key role in neurite initiation in situ and in vitro[9, 10]. The specific position of axon initiation also is critical for neuronal morphology and polarity in vivo. We found that although F-actin-based filopodial protrusions can occur all along the central axons, only those in a cellular compartment containing the basal cell body and proximal central axons typically lead to peripheral axon initiation, suggesting this compartment is uniquely competent to generate a peripheral axon. There are several potential mechanisms controlling this axon initiation site. For example, filopodia in the compartment near the cell body may be favored for microtubule invasion because of better access to the microtubule organizing center. Indeed, both the formation of F-actin-based filopodia and their invasion by microtubules are required for neurite initiation in cultured neurons [10, 54]. In addition, the inherent apical-basal organization of the spinal neuroepithelium likely influences RB polarity and may thus define the basal position of peripheral axon initiation. Extracellular signals also likely play a role in determining the axon initiation site. Our finding that somite boundary regions are favorable for peripheral axon exit from the spinal cord suggests they may be a localized source of growth-promoting cues. Extracellular UNC-6/netrin defines axon initiation position in C. elegans by directing asymmetric activation of an actin-regulatory protein, MIG-10/lamellipodin . However, little else is known about extracellular signals that determine axon positioning in any in vivo system. On the whole, the precise location of axon initiation likely involves coordinated activity of external signals that promote and position outgrowth as well as intrinsic regulatory mechanisms that lead to localized remodeling of the cytoskeleton.
In addition to initiating axons at the correct position, neurons must regulate the number of axons they extend. In a subset of our imaging experiments, we observed transient initiation of multiple peripheral neurites from the basal cell body or central axons, all but one of which retracted after the successful axon exited the spinal cord. One potential explanation for this behavior is that axons encounter a signal upon spinal cord exit that suppresses supernumerary peripheral axons. Spinal cord exit is an important transition in the peripheral axon pathway, marked by a dramatic change in growth cone size and behavior. As they exit, peripheral axons must grow through the basal lamina, a layer of extracellular matrix enriched in laminin, which is a well-known regulator of axon outgrowth and guidance (for example, [55, 56, 57, 58, 59, 60, 61]). In zebrafish bashful (bal)/laminin-α1 mutants, supernumerary axons initiate from ectopic positions in midbrain neurons, suggesting laminin could act to regulate axon number or polarity [60, 61]. Although peripheral RB axon trajectories were normal in bal embryos, their arbors appeared denser [60, 61], suggesting they could have supernumerary peripheral axons. However, individual neuron morphologies have not been examined in these mutants. The extracellular matrix has key roles in regulating cell motility in multiple cell types, and thus is likely to influence both axon number and axon positioning.
Our live imaging has also provided insight into mechanisms underlying axon branching in vivo. Axon branches can form by two modes: growth cone bifurcation or interstitial collateral formation from the axon shaft [2, 6, 17]. We found that one of the first steps in peripheral axon branching is sustained F-actin accumulation, which is consistent with previous F-actin imaging during interstitial axon branching in vitro. Several studies of other neuron types have shown that interstitial branching begins by formation of filopodia along the axon shaft (reviewed in ). Moreover, previous work showed that in cultured neurons filopodia are preceded by F-actin patches [15, 62, 63] that form in association with localized microdomains of phosphatidylinositol-3,4,5-triphosphate (generated by phosphoinositide 3-kinase) . After the formation of F-actin accumulations and filopodia, microtubules must invade the filopodia for branch maturation [6, 19]. Microtubule invasion requires local microtubule severing [64, 65], dynamic instability , and transport of short microtubules into the branch . Moreover, interactions between actin and microtubules are required for branch development [19, 26], suggesting localized F-actin accumulations have an important function in directing microtubule capture and branch formation.
Final branch morphology does not necessarily reflect how a branch formed. Thus, distinguishing between bifurcation versus interstitial branching mechanisms requires live imaging with high temporal resolution. Previous live imaging of axon branching in vivo in mammalian cortex , in zebrafish or Xenopus optic tectum [46, 47], or of zebrafish trigeminal sensory axons  showed that interstitial branching appears to be the predominant mode, although occasional growth cone splitting was observed in some studies [46, 67]. We found that both growth cone bifurcations and interstitial branches commonly occur in RB peripheral axons as they arborize in vivo. RB neurons may employ bifurcation to lay out their overall arbor territory, and interstitial branching to fully arborize and fill the territory. In addition, we found that the two branching modes occur simultaneously within one neuron. In contrast, DRG sensory neurons, which serve a function analogous to RBs, display bifurcation and interstitial branching at distinct stages of their outgrowth [20, 21], and recent evidence suggests the two branching modes may be controlled by different molecular mechanisms [6, 20]. It will be important to determine the extent to which these branching modes employ common versus distinct molecular mechanisms. Multiple extracellular cues have been identified that can promote axon branches in other systems (reviewed in [2, 6]). To date, the only branching factor identified that affects RB axons is Slit2, which can promote excess secondary branching of peripheral RB arbors when overexpressed . The ability to combine live imaging with molecular manipulation makes the zebrafish RB model ideal for future studies to determine mechanisms of axon branching and the molecular differences in bifurcation versus interstitial branching.
LIM-HD transcription factor activity differentially regulates RB axons
The importance of transcriptional regulation in axon development has only recently come to light, and little is known about the molecular and cellular steps between transcription and axon behavior. Previous studies implicated LIM-HD transcription factors in RB peripheral axon formation [39, 40]. Here we show that these transcription factors function in several cell motility processes during RB morphogenesis, and we show how they differentially affect peripheral versus central axon behavior. LIM-HD activity is required for the initiation and outgrowth of the peripheral axon but not the central axons. Although most RB neurons did not initiate a peripheral axon in DN-CLIM embryos, F-actin accumulations and filopodia still formed in these neurons, suggesting LIM-HD activity does not affect the actin remodeling underlying axon initiation. Instead, our results suggest that LIM-HD activity regulates a subsequent step in axon development, perhaps controlling the ability of microtubules to invade filopodia and/or mediating interactions between the microtubule and actin cytoskeleton. Furthermore, our finding that F-actin accumulations and initiating neurites can form in ectopic apical locations on the RB cell body in DN-CLIM embryos suggests that LIM-HD activity contributes to the basal positioning of peripheral axons. The fact that DN-CLIM specifically affects peripheral and not central axon initiation suggests that distinct mechanisms may direct initiation of these two axon types.
DN-CLIM affects not only peripheral axon initiation, but also secondary branching in the periphery. This finding suggests that LIM-HD transcription factor activity could control multiple processes that occur at different stages of RB development. Alternatively, axon initiation and branching may be variations of the same process and may share similar underlying mechanisms mediated by LIM-HD activity. Indeed, in many cases the peripheral axon forms as a branch from the central axon. Both axon initiation and branching require precise control over actin rearrangement and microtubule invasion into the developing branch [4, 5, 6, 7, 8, 26, 54]. Moreover, several extracellular signals, such as netrin [11, 25], are known to stimulate both processes in other systems. Axon branching and the underlying cytoskeletal changes are influenced both by branch promoting cues [2, 6, 18] and by suppressive signals that prevent ectopic branching [16, 69]. LIM-HD activity could potentially modulate axon responses to either branch-stimulating or branch-suppressing signals.
LIM-HD transcription factor activity also differentially affects the motility of RB axons during outgrowth. In contrast to peripheral axons, whose growth was slowed by DN-CLIM, central axon growth rates were increased when LIM-HD activity was disrupted. This result could indicate that LIM-HD activity has a specific function in suppressing central outgrowth. Alternatively, the change in central axon growth rate may result from a general increase in available cellular resources when a peripheral axon does not extend. Consistent with the former hypothesis, differential neurite outgrowth ability of cultured Xenopus spinal neurons was shown to be mediated by an active intracellular signaling mechanism, and not simply by limited cellular resources . Although we cannot distinguish between these possibilities, our results support the idea that an intrinsic transcriptional mechanism can selectively regulate the behavior of two axons from a single neuron.
How might LIM-HD transcription factors control the differential behavior of central versus peripheral axons? One possibility is that LIM-HD transcription factors regulate differential localization or trafficking of receptor or signaling molecules to one RB axon versus the other. Some evidence exists for molecular differences between central and peripheral RB axons. For instance, the cell surface glycoprotein TAG-1 (transiently expressed axonal glycoprotein 1)/Contactin-2 is required specifically for central axon extension, while extracellular Sema3D is repulsive to peripheral but not central axons . In Xenopus RB neurons, focal adhesion kinase, which regulates adhesions and F-actin dynamics, is required for peripheral but not central RB axon growth . To date, few downstream targets of LIM-HD transcription factors have been identified in zebrafish: plexinA4 (a semaphorin receptor) , dpysl3/crmp4 (a semaphorin signaling component) , and contactin-1. Dpysl3 cooperates with Sema3D to regulate peripheral axon outgrowth , PlexinA4 plays a role in mediating Slit2-induced secondary branching of peripheral RB axons , and contactin-1 functions in regulating Xenopus RB axon fasciculation . These latter processes - branching and contact-mediated axon behaviors - represent key differences between central and peripheral axons, further supporting the idea that LIM-HD transcription factors control differential behavior of the axons. There are likely multiple downstream genes regulated by LIM-HD transcription factors that function together to control axon behavior. Indeed, inhibition of any one of the known target genes does not phenocopy the DN-CLIM effect, consistent with this idea. Future work to identify the essential downstream targets of LIM-HD transcription factors, and how they cooperate in RB neurons will be critical for understanding mechanisms of transcriptional regulation of neuronal morphogenesis.
Our live imaging experiments provide insight into the cell motility processes underlying neuronal morphogenesis in vivo. We characterize cell behaviors and the dynamics of F-actin distribution during axon initiation, consolidation and branching in the natural in vivo environment. Our results demonstrate that LIM-HD transcription factor activity differentially regulates the motile behavior of two axons from one neuron. In addition, we show that LIM-HD activity does not affect F-actin protrusive activity or the ability of F-actin to accumulate during axon initiation and branching, but may have a role in peripheral axon positioning. The zebrafish RB model will be ideal for future studies to investigate molecular mechanisms that control these processes in vivo.
Materials and methods
Adult zebrafish (Danio rerio) were maintained in a laboratory breeding colony on a 14:10 h light:dark cycle. Wild-type AB strain and transgenic Tg(-3.1ngn1:gfp) (provided by U Strähle)  or Tg(-3.1ngn1:gfp-caax) lines with labeled primary sensory neurons were used for experiments. Embryos were raised at 23 to 28.5°C and staged as described previously . Animals were handled in accordance with guidelines set forth by NIH and IACUC. Our animal use protocol was approved by the University of Wisconsin Animal Care and Use Committee.
DNA expression constructs were generated with the Multisite Gateway® Cloning System (Invitrogen, Carlsbad, CA, USA) using zebrafish-compatible Tol2 transposon vectors [76, 77, 78] (Tol2 kits provided by K Kwan, C-B Chien and N Lawson), as described .
To drive expression in Rohon-Beard neurons, a 3.1-kb cis-regulatory element from the zebrafish neurogenin1 gene (-3.1ngn1)  was subcloned from pCS2:ngn3.1-GFP (provided by U Strähle) into the SpeI/MslI sites of p5E-MCS (Tol2 kit). The resulting construct, p5E-ngn(-koz), lacks the kozak translational start site sequence, and was subsequently used to generate all expression vectors in this study.
For transgenesis and cell labeling, we used fluorophores GFP or TagRFP (Axxora LLC, San Diego, CA, USA) fused to a CAAX box prenylation sequence, which targets proteins to the plasma membrane. Tol2 kit middle entry vectors  were used to generate expression vectors. Two-way gateway recombination cloning using pTolDestR4-R2pA generated RB-specific expression constructs pEXP-3.1ngn1:gfp-caax or pEXP-3.1ngn1:tagrfp-caax, referred to as GFP-CAAX and TagRFP-CAAX, respectively.
For F-actin labeling, the F-actin-binding calponin homology domain of utrophin with amino-terminal mCherry was amplified from pCS2+mCh-utrCH (gift from B Burkel and W Bement) and cloned by Gateway recombination into pDONR221 (Invitrogen). The resulting middle entry vector (pME-mCherry-UtrCH) was subsequently used in a two-way gateway recombination reaction to generate pEXP-3.1ngn1:mCherry-UtrCH (referred to as mCh-UtrCH).
5'-Capped RNA was synthesized from template DNA in vitro using the mMessage mMachine kit (Ambion, Austin, TX, USA). Tol2 transposase RNA was generated using plasmid pCS2FA-transposase DNA (Tol2 kit), as described . DN-CLIM RNA was generated using plasmid pCS2+DN-CLIM DNA (gift from I Bach), as described .
A transgenic line, Tg(-3.1ngn1:gfp-caax), expressing membrane-targeted GFP in all RB neurons was generated by co-injecting 50 pg (pEXP-3.1ngn1:gfp-caax) plasmid DNA along with 25 pg of Tol2 transposase mRNA into wild-type AB embryos at the one-cell stage. Injected G0 embryos were screened by epifluorescence and raised to adulthood. F1 progeny were screened for stable transmission of the gfp-caax transgene and GFP-positive embryos were raised to establish the line. F2 or F3 embryos were used for experiments.
Cell and F-actin labeling by transient transgenesis
Membrane-targeted fluorophores and mCh-UtrCH were used to visualize cell motility and F-actin, respectively, in developing RB neurons. Injection of plasmid DNA at the one-cell stage results in transient mosaic expression. Embryos were microinjected with 10 to 25 pg of DNA encoding GFP-CAAX, TagRFP-CAAX or mCh-UtrCH in RB-specific expression vectors. Embryos containing individually labeled RB neurons were dechorionated and sorted under epifluorescence illumination using a Nikon AZ100 dissecting microscope equipped with a 4× objective (at 40× magnification). RB neurons in the central region of the trunk (somites 3 to 14) were selected for imaging.
LIM-HD activity was disrupted by ubiquitous expression of a dominant negative cofactor of LIM (DN-CLIM) . Approximately 150 pg of DN-CLIM mRNA was injected into wild-type or transgenic embryos at the one-cell stage. Embryos were sorted at approximately 16 hpf for DN-CLIM-associated eye and midbrain-hindbrain boundary phenotypes , and those with morphological defects were selected for further analysis. Efficacy of DN-CLIM was assessed after imaging by examining all RB neurons and confirming the peripheral RB axon outgrowth phenotype.
Fixed embryo imaging
To analyze mature RB morphology and the extent of peripheral axon branching, selected embryos with individually labeled RB neurons were raised to 24 hpf. Embryos were fixed in 4% paraformaldehyde overnight at 4°C and washed stepwise into 70% glycerol in phosphate-buffered saline. Embryos were whole mounted and imaged on an Olympus Fluoview1000, IX81 confocal microscope equipped with a 60× oil immersion objective (NA 1.35).
Live embryo imaging
Embryos were anesthetized with 0.02% 3-amino benzoic acid ethylester (tricaine) then mounted in 1% low melting point agarose in 10 mM HEPES-buffered E3 embryo medium, as described . Images were captured using an Olympus Fluoview1000, IX81 confocal microscope equipped with a 60× oil immersion objective (NA 1.35).
Z-stacks of 0.5- to 2-μm step sizes encompassing a total of approximately 25 to 30 μm into the trunk were taken of lateral or dorsal-lateral mounts to visualize the RB peripheral and central axon pathways. For time-lapse imaging, embryos ranged in age from 16 to 19 hpf at the beginning of the experiment and were imaged for durations of 2 to 8 hours at 28°C, with Z-stacks captured at 1 to 2 minute intervals.
Image analysis and quantifications
Images were processed and analyzed using Volocity software (Perkin Elmer, Waltham, MA, USA), and figures assembled with Adobe Photoshop (Adobe Systems, Inc., San Jose, CA, USA). Movies were processed and built with Volocity and ImageJ software, and are played at a rate of 4, 6 or 9 frames per second depending on the time intervals used during acquisition. In figures with time-lapse images, time is displayed relative to the first image in the series shown.
Axonal growth rates were calculated as distance over time, and are reported as μm/h. Distance was determined by measuring axon length in XYZ dimensions at two time points of axon extension and calculating their difference. Only axons that extended through the field of view for at least 20 minutes were measured. Central axon lengths were measured from their initiation site on the cell body to the growth cone tip. To compensate for drift during imaging, fixed embryonic features, such as RB cell body or somite boundary position, were used as positional landmarks. Peripheral axon lengths were measured from their initiation site (on the cell body or central axon) to the leading growth cone of the peripheral arbor. For peripheral axons, growth rates were measured starting at the time that the axon reaches the skin and grows ventrally while branching.
Peripheral branching was measured by counting the number of axon termini per axon arbor/neuron in mosaically labeled or transgenic embryos, and expressed as number of axon tips per neuron. The extent of peripheral RB axon branching was assessed in both fixed and live wild-type embryos, and exclusively in live DN-CLIM embryos. To remain consistent, only peripheral axon arbors that extended past the horizontal myoseptum (approximately 80 μm in the ventral direction) were quantified for extent of branching, generally between 21 and 24 hpf depending on the anterior-posterior axial position of the RB neuron within the trunk.
F-actin protrusive activity was measured by counting the number of F-actin-containing filopodia that formed along RB central axons during the time when a peripheral axon formed, or would be expected to form in the case of DN-CLIM-expressing embryos (phase II). Filopodia were defined as visible protrusions, labeled with mCh-UtrCH, with a length of at least 1.5 μm, and were counted over a 20-minute time window. Protrusive activity is reported as number of filopodia per hour.
Statistical and graphical analyses were performed using Microsoft Excel software. Axon growth rates, extent of axon branching, and protrusive activity are reported as mean ± standard error of the mean. Statistical significance was assessed using a two-tailed t-test.
This work was supported by NIH grant NS042228 (MCH). The confocal microscope was acquired with an NIH shared instrumentation grant (S10RR023717, PI Bill Bement, Department of Zoology, University of Wisconsin). We would like to thank Dustin Tryggestad and John Irwin for fish care, and Brennan Boettcher and Xuejiao Tian for technical assistance. We thank Bill Bement and Brian Burkel for the mCh-UtrCH actin probe, Ingolf Bach, Catherina Becker and Thomas Becker for the DN-CLIM construct, Kristen Kwan, Chi-Bin Chien and Nathan Lawson for Tol2 kit vectors, Uwe Strähle for the ngn1 promoter and transgenic fish.
- 8.Dent EW, Gupton SL, Gertler FB: The growth cone cytoskeleton in axon outgrowth and guidance. Cold Spring Harb Perspect Biol. 2010, 3: pii a001800-Google Scholar
- 40.Segawa H, Miyashita T, Hirate Y, Higashijima S, Chino N, Uyemura K, Kikuchi Y, Okamoto H: Functional repression of Islet-2 by disruption of complex with Ldb impairs peripheral axonal outgrowth in embryonic zebrafish. Neuron. 2001, 30: 423-436. 10.1016/S0896-6273(01)00283-5.CrossRefPubMedGoogle Scholar
- 42.Andersen E, Asuri N, Clay M, Halloran M: Live imaging of cell motility and actin cytoskeleton of individual neurons and neural crest cells in zebrafish embryos. J Vis Exp. 2010, 36: pii 1726Google Scholar
- 62.Ketschek A, Gallo G: Nerve growth factor induces axonal filopodia through localized microdomains of phosphoinositide 3-kinase activity that drive the formation of cytoskeletal precursors to filopodia. J Neurosci. 2010, 30: 12185-12197. 10.1523/JNEUROSCI.1740-10.2010.PubMedCentralCrossRefPubMedGoogle Scholar
- 73.Gimnopoulos D, Becker CG, Ostendorff HP, Bach I, Schachner M, Becker T: Expression of the zebrafish recognition molecule F3/F11/contactin in a subset of differentiating neurons is regulated by cofactors associated with LIM domains. Gene Expr Patterns. 2002, 2: 137-143. 10.1016/S0925-4773(02)00304-0.CrossRefPubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.