Microtubule-associated protein 1b is required for shaping the neural tube

Background Shaping of the neural tube, the precursor of the brain and spinal cord, involves narrowing and elongation of the neural tissue, concomitantly with other morphogenetic changes that contribue to this process. In zebrafish, medial displacement of neural cells (neural convergence or NC), which drives the infolding and narrowing of the neural ectoderm, is mediated by polarized migration and cell elongation towards the dorsal midline. Failure to undergo proper NC results in severe neural tube defects, yet the molecular underpinnings of this process remain poorly understood. Results We investigated here the role of the microtubule (MT) cytoskeleton in mediating NC in zebrafish embryos using the MT destabilizing and hyperstabilizing drugs nocodazole and paclitaxel respectively. We found that MTs undergo major changes in organization and stability during neurulation and are required for the timely completion of NC by promoting cell elongation and polarity. We next examined the role of Microtubule-associated protein 1B (Map1b), previously shown to promote MT dynamicity in axons. map1b is expressed earlier than previously reported, in the developing neural tube and underlying mesoderm. Loss of Map1b function using morpholinos (MOs) or δMap1b (encoding a truncated Map1b protein product) resulted in delayed NC and duplication of the neural tube, a defect associated with impaired NC. We observed a loss of stable MTs in these embryos that is likely to contribute to the NC defect. Lastly, we found that Map1b mediates cell elongation in a cell autonomous manner and polarized protrusive activity, two cell behaviors that underlie NC and are MT-dependent. Conclusions Together, these data highlight the importance of MTs in the early morphogenetic movements that shape the neural tube and reveal a novel role for the MT regulator Map1b in mediating cell elongation and polarized cell movement in neural progenitor cells. Electronic supplementary material The online version of this article (doi:10.1186/s13064-015-0056-4) contains supplementary material, which is available to authorized users.


Background
The neural tube, the precursor of the central nervous system, derives from the neurectoderm through a process known as neurulation. In anterior regions of mouse, chick and Xenopus embryos, conserved aspects of this process entail thickening of the neural ectoderm to shape the neural plate, elevation of the edges of the neural plate to form neural folds and convergent extension of the neural plate that narrows and elongates the neural ectoderm [1][2][3][4] and contributes to neural groove formation. The neural folds on either side of the neural plate eventually fuse at the dorsal midline and separate from the overlying non-neural ectoderm to shape the neural tube [5]. Mechanisms of teleost neurulation are often thought to diverge from primary neurulation due to the initial formation of a solid rod (and hence absence of a neural groove), which only later cavitates to give rise to a neural tube [6]. A common misconception is that the neural rod is assembled from the coalescence of neurectodermal cells that exhibit mesenchymal properties (reviewed in [3]), akin to secondary neurulation in mammals. However, closer examination of this process in zebrafish revealed that the neural tube derives in fact from a bilayered neural plate, albeit incompletely epithelialized, that infolds as a continuous sheet. The two sides of the neural plate are closely juxtaposed during infolding, explaining the absence of a neural groove. Thus, medio-lateral positions of cells in the deep layer of the neural plate correlate with dorso-ventral positions in the neural tube [7,8]. In this regard, neural tube formation in zebrafish is similar to primary neurulation in mammals, which also entails the folding of an epithelialized neural plate.
As in other vertebrates [9][10][11], the zebrafish neural plate undergoes neural convergence and extension. However, in zebrafish, narrowing and elongation of the neural anlage is not limited to the neural plate stage, since convergence also drives infolding of the neural plate to shape the neural rod and extension occurs concomitantly with this event. This later convergence event (referred to henceforth as NC, for neural convergence) is driven by polarized migration towards the dorsal midline and cell elongation along the medio-lateral (prospective apico-basal) axis. Failure to undergo proper NC, as a consequence of disruption of the planar cell polarity (PCP) pathway, results in severe neural tube defects in zebrafish [12], highlighting the importance of this early stage of neural tube formation.
The cellular mechanisms underlying NC were first revealed in Xenopus and zebrafish, owing to early access and transparency (zebrafish) of the embryo. In Xenopus, explant assays have revealed that migration of deep neural cells in the medial neural plate is mediated by monopolar protrusions (filopodia and lamellipodia) directed towards the midline [11,13,14]. We have previously demonstrated that cells in the zebrafish neural plate also extend medially-oriented protrusions and elongate as they converge towards the midline [8]. Narrowing of the neural plate in mice involves cell elongation [15] and cellular rearrangements [10,16] that are driven by polarized apical boundary rearrangement and bipolar protrusive activity at the basal pole of cells [9]. Thus, the ability of neuroepithelial cells to form polarized protrusions appears to be an essential and conserved aspect of neural tube morphogenesis, the molecular underpinnings of which remain poorly understood.
Many inroads have been made in understanding how the microtubule (MT) network contributes to cell polarity during migration [17]. MTs are dynamic heteropolymers of αand β-tubulin, existing in alternating states of active polymerization and depolymerization known as dynamic instability [18,19]. These cytoskeletal elements establish the position of cortical polarity (manifested as actin-rich lamellipodia in migrating cells) via multiple pathways [20,21]. Key to MT-mediated establishment of cellular asymmetry is the polarized (radial) organization of these structures, with slow-growing minus-ends anchored at the centrosome and the faster growing plus-ends clustered at the leading edge, adjacent to the cell cortex [22]. In addition to their role in cell migration, dynamic MTs play an active role in cell elongation and maintenance of homeostatic length [23]. The role of stable MTs in cellular dynamics is less well established.
MT stability and dynamics are regulated in part by microtubule-associated proteins (MAPs). Members of the MAP1 family bind along the entire MT lattice. MAP1B, a founding member of this family, is posttranslationally cleaved into a heavy chain (HC) and a light chain (LC1) [24]. The heavy chain contains domains for actin, MTs and LC1 binding [25][26][27] and can therefore crosslink MTs and microfilaments [28,29]. The light chain also binds MTs and actin and regulates the cytoskeleton [30,31]. MAP1B proteins were first identified based on their MT-stabilizing properties [31][32][33]. However, unlike tau, MAP1B preferably associates with dynamic (tyrosinated) MTs, helping to maintain a pool of dynamic MTs required for axonal elongation [34,35]. This activity of MAP1B is controlled by several kinases, including Glycogen synthase kinase-3β (GSK-3β), which increases MAP1B MT binding and dynamicity [36]. The poor MT stabilizing properties of MAP1B combined with its ability to promote MT dynamics, suggest MAP1B function differs from the other MAPs [35]. MAP1B is also expressed prior to other members of this family in the nervous system [37][38][39][40], as it is observed in neuronal progenitors prior to their last mitotic division [41]. Despite this early expression and function in promoting MT dynamics, MAP1B has not been implicated in early stages of neural tube development.
We investigate here whether zebrafish Map1b plays a role in the polarized cell movements that shape the neural rod during NC. Our studies reveal that MTs undergo major changes during neural tube formation, as they become progressively more stable and elongated. The perturbation of cell elongation and polarized migration following nocodazole and paclitaxel treatments suggests that the regulation of MT stability during NC is essential for proper completion of this process. To gain insight into underlying mechanism, we characterized the function of Map1b, previously shown to promote MT dynamicity in axons. map1b is expressed earlier than previously reported, in the developing neural tube and underlying mesoderm. Loss of Map1b function using morpholinos (MOs) or δMap1b, encoding a truncated Map1b protein product, resulted in delayed NC and duplication of the neural tube, a defect previously observed in PCP mutants in which NC is also defective [12]. We observed a loss of stable MTs in these embryos that is likely to contribute to the NC defect. Lastly, we reveal that Map1b mediates cell elongation in a cell autonomous manner and polarized protrusive activity, two cell behaviors that underlie NC and are MTdependent. Together, these data highlight the importance of MTs in the early morphogenetic movements that shape the neural tube and reveal a novel role for the MT regulator Map1b in mediating cell elongation and polarized cell movement in neural progenitor cells.

Microtubules undergo dramatic changes during neurulation
During early stages of neurulation, MTs appear to undergo global morphological changes. Immunolabeling with anti-β−tubulin (anti-β-tub, a marker for the total MT population) revealed that at the neural plate stage (tb-1 som), when neural cells extend polarized protrusions towards the midline, MTs are distributed throughout the cytosol (Fig. 1a, a'), consistent with the radial organization we previously reported [42]. In contrast, at the neural keel stage (so named because of the keel shape adopted by the neural tissue as it transition from a neural plate to a neural rod, 4-5 som) and neural rod (12-13 som) stage, MTs organize into long linear arrays (bundles), which align along the future apico-basal axis of neural cells, coincident with epithelialization that occurs following NC (Fig. 1b-c') [43].
The organizational changes observed in MTs suggest that they become increasingly stable as neurulation Fig. 1 Microtubules become increasingly stabilized during neurulation. Hindbrain sections of embryos at the neural plate (tb-1 som) (a, a', d, d'), neural keel (4-5 som) (b, b', e, e') and neural rod (12-13 som) (c, c', f, f') stages immunolabeled with anti-β-tub (total MTs) in green (a-c') and anti-glu-tub (detyrosinated MTs) in red (d-f'). (a'-c') and (d'-f') Higher magnification of boxed areas in (a-c) and (d-f), respectively. Scale bars: 10 μm progresses. In order to investigate the distribution of stable MTs at different stages of neurulation, embryos were immunolabeled with anti-detyrosinated-tubulin (glu-tub). Glu-tub antibodies recognize stable, detyrosinated MTs by binding to the exposed carboxy-terminal glutamic acid of α-tubulin (α-tub) in MT polymers [44,45]. At the neural plate stage, glu-tub labeling is diffuse and more punctate than β-tub (Fig. 1d, d'). However, by the neural keel stage (Fig. 1e, e') detyrosinated MTs organize into linear structures, which become more accentuated at the neural rod stage (Fig. 1f, f '). The distribution of glu-tub labeling thus implies that detyrosination of MTs occurs initially in discrete foci along the MT polymers that subsequently expand to include the entire polymer. These observations suggest that stable MTs increase over time, reaching elevated levels in the epithelialized neural tube. To quantify the relative abundance of stable MTs, we analyzed the ratio of glu-tub (stable MTs) to αtub (total MTs) at the neural plate (tb), neural keel and neural rod stages and found this ratio to be highest at the neural rod stage (Additional File 1: Figure S1).

Microtubules are required for NC
The striking increase in the levels of detyrosinated MTs suggests that the stability of these cytoskeletal elements is regulated during neural tube development and likely to be important for neural tube morphogenesis. To test this, we treated early neural stage (2-3 som) embryos with nocodazole (17 and 32 μM) or paclitaxel (50 μM), which destabilize and hyperstabilize MTs respectively, and analyzed the effect of these drugs on cell behaviors at the neural keel stage (4-5 som). The efficacy of these drugs was first confirmed by immunolabeling with anti-βtub, which revealed that the linear organization of MTs was disrupted following both treatments (Additional file 2: Figure S2).
In order to investigate whether MTs play a role in polarized cell migration, time-lapse confocal imaging was carried out at the neural plate (tb-1 som) stage using control and nocodazole-treated embryos mosaically expressing mGFP. Cells from untreated embryos exhibit an elongated appearance with protrusions oriented medially, as previously described [8]. In contrast, cells from nocodazole-treated embryos were rounded in shape and failed to migrate in a directional manner, as a result of randomized membrane protrusions (Additional files 3 and 4). This observation suggests that MTs are not required for the formation of protrusions but are rather implicated in the proper polarization of these extensions.
Together these findings identify MTs as key mediators of cell elongation and polarized cell movement during NC. They further suggest that regulation of MT stability is tightly controlled during early development, pointing to a potential role for microtubule-associated proteins (MAPs) in this process.
Zebrafish map1b is expressed in the developing neural tube Mammalian MAP1B is one of the earliest MAPs to be expressed in the developing nervous system and hence a good candidate for mediating early morphogenetic movements during neural tube formation. Gene ontology analysis revealed a high level of sequence similarity between zebrafish Map1b and its orthologues in chick, mouse, rat and human. The regions of highest conservation (98 % identity) comprise a stretch of 550 amino acids in the N-terminus and 120 amino acids in the C-terminus [31]. Mammalian MAP1B contains two MT-binding domains, each composed of multiple repeats of KKEE or KKEI/V motifs [26,30]. Domain analysis of zebrafish Map1b revealed the presence of the conserved KKE signature repeats at the N-terminus, in the region encoding the heavy chain. Furthermore, synteny analysis showed that zebrafish map1b is located in a conserved region of the genome.
In order to determine whether zebrafish map1b is expressed during neurulation, we performed wholemount in situ hybridization. We observed that map1b is broadly distributed at the neural plate (tb-1som), neural keel (4-5 som) and neural rod (8-10 som) stages ( Fig. 4a-c). Its expression appears to be in a gradient that is highest in the mesoderm at the neural plate stage (Fig. 4a). By the neural keel and rod stages the level of map1b increases and expression expands dorsally, as map1b is present throughout the developing neural tube (Fig. 4b, c). To confirm that the signal observed at these developmental stages is specific, in situ hybridization using a sense riboprobe was also performed. No labeling was observed with the latter (Fig. 4a'-c').
These observations indicate that map1b is expressed earlier than previously reported, in undifferentiated neural progenitor cells undergoing NC.
To evaluate whether map1b disruption results in later developmental defects, map1b MO1 and MO2 (10 ng)injected embryos were imaged at 24 hpf. We observed a disorganization of the hindbrain region in these embryos, characterized by absence of morphological landmarks (Additional file 5: Figure S3) that were more pronounced with MO1 than MO2. Sectioning through the hindbrain of MO1-injected embryos revealed that the disorganization was caused by a partial (n = 4 out of 10 embryos) or full (n = 1 out of 10 embryos) duplication of the neural tube (Fig. 5b). This striking phenotype was first observed in PCP mutants [12] and is thought to be a consequence of delayed NC [47,51]. In addition to the disorganized hindbrain, we observed a shortened body axis (Additional file 5: Figure S3), a phenotype often associated with impaired convergent extension in the axial mesoderm [52].

Truncated Map1b lacking the MT-binding-domain causes delayed NC
In order to confirm the role of Map1b in mediating NC using a MO-independent method, we designed a construct based on a mouse mutation thought to function as a dominant-negative allele [31,53,54]. The zebrafish mutant construct, δmap1b, encodes the first 571 aa of Map1b, which includes the Map1b light chain (LC1) binding domain (in the heavy chain region) but not the MT-binding domain (6A). Mice that are heterozygous for this mutation have a spectrum of phenotypes including slow growth rates and small eyes, while their homozygous siblings die during embryogenesis [53]. Interestingly, zebrafish embryos injected with δmap1b RNA (25, 50, 75 and 100 ng/μl) exhibit an increasingly severe reduction in body and eye size with higher concentrations of RNA (Additional file 6: Figure S4B), suggesting that the truncated Map1b protein functions in a similar manner in both species.
The more severe phenotype in δmap1b-injected embryos compared to map1b MO-injected embryos suggests that δMap1b functions in a dominant-negative manner. Given that δ Map1b retains the ability to bind to LC1 [31], which is implicated in the regulation of MT stability and other MT-independent processes [31,55], it is likely that depletion of this peptide accounts for some of the pronounced defects observed in injected embryos. In addition, the unbound endogenous heavy chain in δmap1b-injected embryos may also play a contributing role.
Together these findings reveal a previously unknown role for Map1b in mediating NC in the neural ectoderm. The fully or partially duplicated neural tube observed in map1b MO-injected embryos is consistent impaired NC. Furthermore, the fact that a neural tube, albeit abnormal, forms in Map1b-depleted embryos indicates that Map1b is required for the timely progression rather than completion of neurulation.

Loss of Map1b results in the loss of stable microtubules
Previous studies indicate that Map1b maintains a dynamic population of MTs that promotes axonal growth [48,56]. If Map1b plays a similar role in the neuroectoderm, its loss-of-function should result in increased MT stability. To test this prediction, neural keel (4-5 som) stage uninjected, standard MO-injected, map1b MO2injected and δmap1b-injected embryos were sectioned and labeled with anti-β-tub (total MTs), anti-glu-tub (stable, detyrosinated MTs) and DAPI (nuclei) and imaged at the hindbrain level. Unlike the dramatic perturbation of the MT network observed following treatments with nocodazole and paclitaxel (Additional file 2: Figure S2), the overall organization of MTs visualized with β-tub labeling in map1b MO2-and δmap1binjected embryos appeared similar to controls (Fig. 6a), although in some samples (insets in Fig. 6a b1, c1) the β-tub-labeled bundles were less well defined than those of uninjected (inset in Fig. 6a a1) and standard MO (data not shown).
Enhanced MT stability is often manifested in either an increase in MT bundle length or the average number of MT bundles/cell. Quantification of the length of glu-tub-and β-tub-positive bundles in the different treatment groups did not reveal any differences (data not shown). However, the average number of stable, glu-tub-positive bundles per cell (nucleus) revealed an unexpected decrease in map1b MO2-and δmap1binjected embryos relative to uninjected embryos (Fig. 6b) (uninjected embryos: 3.2 ± 0.23, n = 13 embryos; standard MO-injected embryos: 2.8 ± 0.69, n = 3 embryos; map1b MO2-injected embryos: 2.18 ± 0.23, n = 6 embryos; δmap1b-injected embryos: 2.19 ± 0.27, n = 9 embryos). Although a decrease in bundle number in experimental groups relative to standard MO-injected embryos was also observed, this number was not statistically significant, most likely due to the smaller sample size of the latter (Fig. 6b). These observations suggest that in the context of NC, Map1b functions to stabilize MTs rather than promote MT dynamics.
A decrease in total bundle number per cell was also observed with β-tub labeling, but only in δmap1b-injected embryos (uninjected embryos: 8.16 ± 0.85, n = 14 embryos; standard MO-injected embryos: 11.5 ± 1.6, n = 5 embryos; δmap1b-injected embryos: 6.18 ± 0.54, n = 9 embryos) (Fig. 6c). This reduction is unlikely to reflect the loss of stable MTs from the total MT population given that a decrease in total MTs was not observed in map1b MO2injected embryos. Thus, δMap1b may disrupt total MTs (dynamic and stable) whereas map1b-MO2 alters stable MTs specifically. However, a more likely explanation, is that the apparent loss of total MT bundles in δmap1b-injected embryos reflects subtle changes in the organization of the MT cytoskeleton (bundles that are less well defined) that make the automated quantification method less accurate.
To more directly tease apart the role of Map1b in regulating stable versus dynamic MTs, we also analyzed the levels and distribution of tyrosinated α-tub (dynamic MTs) in map1b-MO2-and δmap1b-injected embryos. We did not observed an obvious difference in the number and organization of tyr-tub MTs in these embryos (map1b MO2-injected embryos: n= 4 embryos; δmap1binjected embryos: n= 3 embryos) relative to controls (uninjected embryos: n = 5 embryos). We did however notice an increase in the number of puncta labeled with anti tyr-tub (arrowheads in Additional File 7: Figure S5). Since re-tyrosination has been reported to occur on the non-assembled tubulin dimer pool [57], these puncta may correspond to depolimerized tubulin. Overall these observations suggest that loss of Map1b does not impact dynamic MTs.
To further test whether Map1b is implicated in MT stabilization, we performed a nocodazole sensitization test in map1b MO2-injected (4 ng) embryos. However, treatment of these embryos with a low dose of nocodazole (1 μg/ml) did not worsen NC defects (Fig. 5d). This may be due to intrinsic differences in the mechanisms by which these molecules alter MT properties. Alternatively, Map1b may function via additional MTindependent mechanisms to promote NC.

Map1b functions cell-autonomously to regulate cell elongation
In order to identify the cellular mechanisms underlying delayed NC in Map1b-deficient embryos, we analyzed the morphology of hindbrain cells mosaically expressing mGFP at the neural keel (4-5 som) stage. We observed that in contrast to control (uninjected) cells (Fig 6a a-a'), cells in map1b MO1 (data not shown), map1b MO2-and δmap1b-injected embryos failed to elongate (Fig. 7a bb' , c-c'), as was observed with drug-treated embryos (Fig. 3c, d). LWR measurements of control and Map1bdeficient cells revealed a significant difference in cell shape (LWR of uninjected cells = 3.70 ± 0.1, n = 111 cells from 7 embryos; LWR of cells from embryos injected with map1b MO2 = 2.78 ± 0.2, n = 43 cells from 5 embryos; LWR of cells from embryos injected with δmap1b = 2.41 ± 0.1, n = 135 cells from 9 embryos) (Fig. 7b). Thus, Map1b is required for cell elongation during NC and its ability to regulate MTs may underlie this process.
Since map1b is expressed in the mesoderm and neuroectoderm (Fig. 4), it is possible that the widened neural plate of Map1b-deficient embryos is an indirect consequence of defective convergent extension movements in the mesoderm [52]. If Map1b functions in a cell-autonomous manner in the neuroectoderm to regulate cell elongation, then Map1b-deficient cells isochronically transplanted into WT hosts are expected to be rounded. Conversely, isochronic transplantation of control (WT) cells into Map1b-deficient embryos should not impact the ability of these donor cells to adopt their correct elongated morphology.
To perform the first isochronic transplantation experiment, donor embryos were injected with RNA encoding mRFP with or without map1b MO1 (10 ng) and host embryos were injected with mGFP DNA (which is mosaically expressed). Cell shapes of both donor (red) and host (green) cells were analyzed in hindbrain sections of host embryos fixed at the neural keel stage (4-5 som). We observed that transplanted control cells were similar in shape to WT host cells (LWR of control cells = 3.37 ± 0.5, n = 100 cells from 18 embryos; LWR of host cells = 3.52 ± 0.69, n = 88 cells from 14 embryos; Fig. 7c, d a-a'), whereas transplanted Map1b-deficient cells consistently appeared rounder than WT host cells (LWR of map1b MO1-injected cells = 1.51 ± 0.27, n = 135 cells from 14 embryos; LWR of WT host cells = 3.63 ± 0.52, n = 92 cells from 14 embryos; Fig. 7c, d b-b.2'). There was no apparent bias to the location of transplanted cells in the neural tube of their hosts, eliminating position as a contributing factor to differences in cell shape. These data reveal that Map1b functions cell autonomously to regulate cell elongation in the neural ectoderm.
The reciprocal isochronic transplantation could not be completed as the Map1b-deficient hosts did not survive the transplantation.

Map1b is required for polarized migration during NC
To test whether Map1b plays a role in polarized cell migration during NC, mGFP-labeled cells in control (uninjected) and map1b MO1(10 ng)-injected embryos were imaged using time-lapse microscopy ( Fig. 8a and Additional files 8 and 9). Cell tracing revealed that control cells (from uninjected embryos) were initially rounded and gradually elongated as they approached the midline. In contrast, cells in map1b MO1-injected embryos took longer to elongate, consistent with the LWR measurements of cells in fixed preparations. In addition, their migration towards the midline was delayed ( Fig. 8b; n = 3 embryos; 6-9 cells/embryo).
In order to determine whether delayed migration in map1b MO1-injected embryos was caused by defective protrusive activity, as observed in drug-treated embryos, the angular distribution of plasma membrane extensions was quantified and ploted. At the neural plate (tb-1 som) and neural keel (4-5 som) stages, membrane protrusions of control cells were biased towards the medio-lateral axis, whereas the membrane protrusions in Map1b-deficient cells were less polarized and failed to align with the medio-lateral axis (Fig. 8c). Together these findings suggest that Map1b mediates both cell elongation and the polarized orientation of protrusive activity, two cell behaviors that are MT-dependent.

Discussion
MTs are required for NC Cellular dynamics during convergent extension movements in vertebrates are powered by actin polymerization, cell-cell adhesion and cell-extracellular matrix (ECM) interactions [52]. Since MTs play a prominent role in cell migration [17,22], it seems intuitive that they would also be implicated in the mechanics of cellular rearrangements during convergent extension in the mesoderm and neurectoderm. There is some experimental evidence supporting MT-mediated cellular rearrangement in the mesoderm, however, there is a dearth of data on the involvement of MTs during NC. With respect to the mesoderm, disruption with nocodazole in early gastrula Xenopus embryos prevents mediolateral intercalation, involution and convergent extension of the marginal zone (the precursor of the mesoderm) [58]. Kwan and Kirschner further demonstrated that treatment of Xenopus dorsal marginal zone explants with nocodazole but not taxol prevents lamellipodia formation, indicating that the bulk of polymerized tubulin rather than MT dynamics (which would be altered by both drug treatments) is important for convergent extension [59]. In zebrafish embryos, MTs are known to mediate cell-cell contacts and initiation of planar polarity, by localizing PCP pathway component Prickle in a polarized manner during mesodermal convergent extension [60].
In amniotes, narrowing of the neural plate is brought about by a combination of cell elongation and intercalation. In chick embryos, treatment of the neural plate with nocodazole prevents cell lengthening along the apico-basal axis, resulting in a wider neural plate [61]. While it is currently unknown whether MTs also power cell intercalation in amniotes, the medio-lateral oriented basal protrusions that drive cellular rearrangement in the mouse epithelialized neural plate [9] are reminiscent of MT-dependent polarized basal protrusions in C. elegans epithelial cells undergoing dorsal closure [62], raising the possibility that MT-based mechanisms may also be employed to narrow the neural in chick and mouse embryos.
We have previously shown that MTs in neural plate cells have a radial organization, which is characteristic of migratory cells. Following NC, MTs become linear, an architecture often observed in epithelial cells [42]. We report here that the levels of stable MTs steadily increase as neurulation proceeds. Our functional analysis using MT-disrupting drugs further suggests that proper regulation of MT stability is essential for both cell elongation and polarized migration during NC. In addition, the fact that protrusive activity is still observed (albeit random) in nocodazole-treated embryos, indicates that MTs are required for polarization but not the formation of these membrane extensions.
Thus, despite the more prominent mesenchymal properties of zebrafish neural plate cells relative to their amniote counterparts, evidence suggests that MTs and their regulators play a central role in driving NC in vertebrates.

Map1b promotes stable microtubules
The prominent changes in MT stability during neurulation are likely to be regulated by microtubule associated proteins. We show here that stable/detyrosinated MTs are lost in Map1b-deficient embryos. The apparent selective reduction in detyrosinated MTs (observed in map1b MO2-injected embryos) argues against a role for Map1b as a general MT stabilizing factor. Rather, Map1b may protect stable MTs or promote the α-tub detyrosination event that is revealed by glu-tub labeling. Despite the focus on MTs in this study, it is likely that Map1b also influences the actin cytoskeleton. In this regard, Map1b is known to bind actin in addition to MTs [25,63], thereby crosslinking the two cytoskeletons. Furthermore, Rac1 and Cdc42 are downstream effectors of Map1b [64] that are both implicated in the crosstalk between actin and MTs [65].

Map1b regulates distinct cell behaviors during NC
We have previously shown that, during NC, cells elongate as they migrate towards the dorsal midline. These cellular dynamics are accompanied by extensive protrusive activity polarized along the medio-lateral axis [8].
We found that despite the broad distribution of mRNA in mesodermal and ectodermal cells, Map1b is required cell autonomously for cell elongation in the neural tissue. A recent study has shown that a polarized population of dynamic MTs is required for cell length maintenance in the zebrafish neural tube [23]. While our studies reveal a role for Map1b in promoting stable MTs rather than MT dynamicity, it is likely that both MT populations contribute to cell elongation.
In addition to cell elongation, directional migration is also defective in Map1b-depleted embryos. Impaired migration may also be attributable to abnormal MTs, as stable MTs, anchored at the cell cortex, are thought to function as tracks to deliver regulators of actin polymerization to the leading edge [17]. Whether the same population of Map1bregulated MTs mediates cell elongation and migration in the zebrafish neural tube is unclear.
Analysis of protrusive activity in Map1b-depleted embryos revealed a lack of biased orientation along the medio-lateral axis and ectopic persistent protrusions on the anterior and posterior pole of neural cells. This abnormal protrusive activity may underlie the delay in cell elongation and migration. It is unclear how Map1b biases protrusive activity medially. However, the recent finding that Map1b binds and sequesters EB3 in the cytosol of developing neuronal cells [72] raises an interesting possibility. EB3 is a MT-plus end binding protein that is enriched in growth cones and has been shown to coordinate the interaction between F-actin (required for protrusive activity) and dynamic MTs during neuritogenesis [73]. Furthermore, EB3-capped MT plus ends orient towards the leading edge in migrating cells, possibly in response to an extracellular signal [17]. In this context, Map1b may regulate polarized protrusive activity by controlling the levels of EB3 available to associate with MT plus ends. In the absence of Map1b, increased binding of EB3 to MTs plus ends could cause the formation of ectopic F-actin nucleation.
Despite the significant increase in neural plate width in map1b MO-injected embryos, the neural tube eventually forms (albeit abnormally), indicative of a delay rather than blockage of NC. A similar outcome was also observed following depletion of PCP pathway components, suggesting compensatory mechanisms that ensure proper completion of neural development.
Whether map1b function during NC is conserved remains to be determined, as neural tube defects have not been reported in mouse map1b knockouts [74], possibly due to functional redundancy among MAP family members [75,76] or distinct cellular mechanisms underlying the narrowing of the neural plate. Despite these differences, loss of Map1b function in mice also causes a delay rather than a blockage in neural development [34].

Conclusions
We show that MTs become progressively more stable as neurulation progresses. Drug treatments that either destabilize or hyperstabilize MTs impair NC by disrupting cell elongation and polarization, indicating that the regulation of MT stability is a key event during neural tube development. We demonstrate that the microtubule-associated protein Map1b is broadly expressed during neurulation and promotes stable MTs. Furthermore, loss of Map1b function causes a delay in NC, cell autonomous disruption of cell elongation, impaired directional migration and polarized protrusive activity. Based on these findings, we propose that Map1b enables NC at least in part by maintaining a population of stable MTs.
Collectively, these studies identify map1b as a key regulator of early morphogenetic movements in the neural tube. It will be interesting in the future to identify the signaling pathways that function upstream of Map1b to control the MT cytoskeleton during NC.

Zebrafish strains
Studies were performed using wildtype (AB) strains. All experiments were approved by the University of Maryland, Baltimore County's Institutional Animal Care and Use Committee (IACUC) and were performed according to national regulatory standards.

Embryo staging
Staging was done according to [77]. Stages of neurulation were defined as previously described [8].
The δmap1b construct, a codon optimized sequence encoding the first 571 aa of zebrafish Map1b, was synthesized by Genewiz based on the published zebrafish map1b sequence (accession # XM_003198629). δmap1b was subsequently subcloned into the pCS2+ vector.
For RNA injections, mGFP or mRFP expressing plasmids were linearized with NotI and transcribed using SP6 mMESSAGE mMACHINE kit (Ambion, cat. no. AM1340). 50 pg of RNA was injected into one-to fourcell stage embryos.

Cell transplantation
Transplantation was performed as described in [82]. 50-100 cells from donors were transplanted isochronically into the animal pole of host embryos at the sphere to dome stage.

Time-lapse confocal microscopy
Time-lapse microscopy was performed as previously described [83]. Embryos were imaged using a Leica confocal microscope (Leica SP5 TCS 4D) at 30 s-1 min intervals. Images were analyzed using the Leica LAS software, Image J (NIH) and Adobe PhotoShop.
For cell shape analysis, mGFP or mRFP RNA/DNA injected embryos were sectioned and either imaged directly (mRFP) or immunolabeled with anti-GFP prior to imaging. All fluorescently labeled sections were imaged using an SP5 confocal microscope (Leica SP5 TCS 4D).
Wholemount in situ hybridization ISH was conducted as described [85]. Plasmid containing dlx3 (obtained from Igor Dawid, NIH) was linearized with NotI and T7 polymerase was used to generate anti-sense probe. map1b (in pCRII-TOPO vector) was linearized using NotI and KpnI for sense and anti-sense probes, respectively. T7 and SP6 polymerases were used to generate anti-sense and sense probes respectively. ISH labeled embryos were sectioned and imaged using a Zeiss Axioscope2 microscope.

Measurements and statistical analysis
Length-to-Width ratios LWRs were calculated as previously described [8].

Tracing and quantification of cell behaviors
Cell tracing of single cells over multiple time frames was performed using Metamorph (MolecularDevices).

Protrusion analysis
Stacks of images from time-lapse microscopy were flattened to a single frame per time point using Volocity v5.5 (Perkin-Elmer) and exported as tifs. Interphase cells that stayed in frame for the extent of the movie were selected. Using a plugin created for imageJ (NIH), individual cells were threshholded and the outline traced. Each cell was then divided into 8 segments based on the centroid and the orientation to the midline. Finally individual frames were overlaid and any new membrane extensions were counted for each frame. The percent protrusions for each section was then calculated and plotted in Mathematica v9 (Wolfram).

Statistical analysis
InStat (GraphPad) was used to run statistical analysis on data sets. Student's T-test, ANOVA followed by Bonferroni or Kruskal-Wallis test followed by Dunn's post-hoc test were used to analyze data groups as appropriate.

Quantification of Western band intensity
Band intensity was determined using digital scans followed by analysis with ImageJ (NIH). The area used to measure was constant for all experiments.

Quantification of MTs
Measurements of MT bundle length and number of MT bundles/nuclei were done using Volocity (Perkin-Elmer). A maximum intensity projection of the total confocal sections (40 μm) imaged using the Leica SP5 was generated. The neural keel was cropped to quantify only MT signal within the neural tissue. Filters that identified nuclei (DAPI), stable MTs (glu-tub) and total MTs (β-tub) based on the standard deviation of intensity was used. To compensate for objects close together, an automated algorithm that separates object was run as part of the filter. The measurements were automatically collected and analyzed in excel Additional files Additional file 1: Figure S1. Microtubule stability increases during neurulation. (A) Western blot of whole cell lysates blotted for glu-tub (stable MTs) and α-tub (total MTs) at neural plate (tb-1 som), neural keel (4-5 som) and neural rod (12 som) stages. Two bands (1)

Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions PJ, VO and SS: performed the majority of the experiments and contributed to data analysis and manuscript preparation. EH: carried out the analysis of dynamic changes in MT organization in WT embryos and the initial studies on the effect of MT-disrupting drugs on cell elongation and migration. EV and JW worked with SS to perform Map1b loss of function studies using primarily δmap1b but also map1b MO2 and performed data quantification for multiple experiments. RJM performed the analysis and quantification of the protrusive activity, assisted in the quantification of MT stability, carried out the statitiscal analyses for most experiments and helped to assemble figures. NSA designed δMap1b and assisted with the molecular work on this construct. AM characterized the function of Gsk3β. RMB supervised the research and data analysis and wrote the manuscript. All authors read and approved the final manuscript.