In situ observations of the development of hippocampal and cortical neurons indicate that final axon-dendrite identity is defined at the time of generation of the first two, oppositely positioned, neurites. Quite differently, in vitro studies demonstrated that axonal fate is defined by the stochastic selection of one of the multiple minor neurites for fast outgrowth. By analyzing the fate of all neurites, starting at the time of emergence from the cell body, we demonstrate that polarity is defined at the bipolar stage, with one of the two first-appearing neurites acquiring axonal fate, irrespective of how many other neurites later form. The first two neurites have, as in vivo, the highest growth potential, as cutting the axon results in the growth of a new axon from the neurite at the opposite pole, and cutting this induces regrowth from the first. This temporal and spatial hierarchical definition of polarized growth, together with the bipolar organization of microtubule dynamics and membrane transport preceding it, is consistent with polarity being initiated by an intrinsic program. In this scenario, molecules required for axon specification would act at one of the first two neurites and extrinsic cues will be required for final commitment of polarity.

Introduction

The formation of neurites that occurs immediately after the last mitosis is a crucial event in the development of the brain. In fact, the first neurites support the migration of the newly generated neuron away from the ventricular or sub-ventricular zone to reach their final destination (Hatten, 1999; Kriegstein and Noctor, 2004; Nadarajah et al., 2003; Nadarajah et al., 2001; Nadarajah and Parnavelas, 2002; Noctor et al., 2004; Parnavelas et al., 2002). Although the process of neurite formation and differentiation has been the subject of analysis for several decades (Da Silva and Dotti, 2002), only recently has the whole process been observed in individual neurons, from the time of the neurogenic division until the time of reaching their final position (Hatanaka and Murakami, 2002; Noctor et al., 2004). From these and previous studies (Buchman and Tsai, 2007; Kriegstein and Noctor, 2004; O'Rourke et al., 1992) it becomes clear that neurons display a strong polar phenotype during all stages of development. Newborn neurons first attain a bipolar phenotype shortly after the final division in the ventricular zone, with a leading and a trailing neurite (LoTurco and Bai, 2006; Nadarajah et al., 2001). Neurons then migrate radially to the sub-ventricular zone, where additional neurites extend (Noctor et al., 2004; Tabata and Nakajima, 2003). A subsequent second phase of bipolar migration towards the cortical plate is guided by the leading neurite forming opposite to the trailing neurite, the latter attaining the morphological characteristics of the axon (Nadarajah et al., 2002; Noctor et al., 2004). As the two architecturally most conspicuous domains of a mature neuron, i.e. axon and apical dendrite, present the same orientation of the initial trailing and leading neurites, generated in the ventricular zone, it appears reasonable to hypothesize that the final axis of neuronal polarity, i.e. axonal and dendritic identity, is a consequence of the intrinsic mechanisms that determine the initial bipolar shape, required to guarantee migration.

Quite different from the above view, data based on observations of the development of neuronal polarity in dissociated hippocampal neurons in vitro support the notion that polarity is established by the stochastic selection of one of the multiple neurites of the so-called stage-2 neurons for rapid axonal growth at the stage-2 to stage-3 transition (Bradke and Dotti, 2000; Dotti et al., 1988). This stochastic selection of the axon can be guided by environmental cues since axonal growth can be forced to occur from any of the presumptively equal multiple neurites that decorate the cell body of stage-2 neurons (Bradke and Dotti, 1999; Esch et al., 1999; Esch et al., 2000; Shelly et al., 2007). Mechanistically, it is assumed that this neurite is selected because it most efficiently responds to the growth-positive environmental signals (Arimura and Kaibuchi, 2007; Bradke and Dotti, 2000; Craig et al., 1992; Wiggin et al., 2005). This view would imply that those events underlying polarized growth, such as destabilization of actin (Bradke and Dotti, 1999), the stabilization of microtubules (Ferreira and Caceres, 1989), axonal-directed membrane traffic (Bradke and Dotti, 1997) and the accumulation of a number of specific proteins (reviewed by Wiggin et al., 2005) would be exclusively regulated in stage-2 neurons, by differentially regulating the activity of extracellular cues. This view is weakened by the fact that neurons polarize in vitro, where attachment factors and soluble cues are uniformly distributed, thus giving rise to the assumption that the polarity axis of growth must be defined intrinsically. This was demonstrated most elegantly in neuroblastoma cells by Frank Solomon in the early 1980s (Solomon, 1981) and in neurons, by the showing that the axon arises from the sprout that grows closest to the centrosome in hippocampal neurons (de Anda et al., 2005) and cerebellar granule cells (Zmuda and Rivas, 1998). Still, acceptance of this view is not total, and present polarity studies concentrate on the identification and role of molecules capable of inducing growth from any neurite (or inhibiting growth of the axon), without taking into account the possible intrinsic predisposition of the neurites to the effect of such `polarizers'.

Fig. 1.

Axon fate predisposition is present at the bipolar stage. (A) Neocortical slice of an E18 mouse brain that had been transfected with a cytoplasmic, fluorescent protein (Venus-EGFP) at E15 by in utero electroporation, showing migrating cells (lower panel). The upper panel shows the distribution of the axonal marker tau-1. The lower panel shows Venus-EGFP fluorescence merged with the tau-1 image. (B) Maximal intensity projection of a confocal z-stack of a cell in the upper intermediate zone (boxed in A). The GFP-positive cell has a typical bipolar morphology of migrating neurons (arrowheads indicate trailing neurite). The numbers at the left (1-7) mark the places along the neuron that were selected for y-z scans shown in C. (C) Views along the y-z plane. Arrowheads show a colocalization of GFP and tau-1. (D) GFP and tau-1 intensity-profiles from the selected places 1-7 confirm a colocalization of GFP and tau-1 immunoreactivity in the distal portion of the trailing process of the migrating neuron. The profile was obtained by drawing a line that crossed the middle of the GFP signal. Bar, 100 μm (for A) and 10 μm (for B,C).

Fig. 1.

Axon fate predisposition is present at the bipolar stage. (A) Neocortical slice of an E18 mouse brain that had been transfected with a cytoplasmic, fluorescent protein (Venus-EGFP) at E15 by in utero electroporation, showing migrating cells (lower panel). The upper panel shows the distribution of the axonal marker tau-1. The lower panel shows Venus-EGFP fluorescence merged with the tau-1 image. (B) Maximal intensity projection of a confocal z-stack of a cell in the upper intermediate zone (boxed in A). The GFP-positive cell has a typical bipolar morphology of migrating neurons (arrowheads indicate trailing neurite). The numbers at the left (1-7) mark the places along the neuron that were selected for y-z scans shown in C. (C) Views along the y-z plane. Arrowheads show a colocalization of GFP and tau-1. (D) GFP and tau-1 intensity-profiles from the selected places 1-7 confirm a colocalization of GFP and tau-1 immunoreactivity in the distal portion of the trailing process of the migrating neuron. The profile was obtained by drawing a line that crossed the middle of the GFP signal. Bar, 100 μm (for A) and 10 μm (for B,C).

In this work we show, by a number of different in vitro and in vivo assays, that the mechanisms that govern the bipolar phenotype in situ determine the establishment of a bipolar neuronal growth axis and that the establishment of this bipolar growth axis is intrinsically defined.

Results

Axon fate predisposition is present at the bipolar stage, in vivo and in vitro

To show if axon and dendrite specification for multipolar neurons, like those of the hippocampus or cortex, is defined at the bipolar stage in situ, we transfected ventricular precursor cells of the E15 neocortex by in utero electroporation of a Venus-EGFP plasmid. The anatomy of the progeny of the transfected cells was analyzed 3 days later (Fig. 1). Newly generated neurons in the ventricular zone, which are negative for the precursor marker nestin (supplementary material Fig. S1A,B), have a rhomboid appearance, with the longest axis perpendicular to the ventricle. Two protrusions are evident in such cells, one extending towards the subventricular zone and one at the opposite pole, facing the ventricle. Once in the upper intermediate zone (see boxed cell in Fig. 1A shown in detail in Fig. 1B), neurons display a pronounced bipolar morphology, with a long and thick neurite facing towards the cortical plate and a thin, oppositely located neurite that expresses the axonal marker tau-1 (microtubule-associated protein tau, MAPT) (Fig. 1A-D). Since the immunoreactivity of tau-1 in that thin neurite becomes obscured by those of other non-transfected neurons, we analysed, by confocal microscopy, the level of colocalization of the tau-1 signal and the Venus-EGFP signal, which completely overlap in the thin ventricular-oriented but not in the opposite thicker neurite (Fig. 1C,D). Since those two neurites form well before the other neurites, the restriction of tau-1 to the neurite that originally was directed towards the ventricle strongly indicates that axonal fate was determined at the bipolar stage.

To directly test the above prediction, we used dissociated embryonic hippocampal neurons in vitro (Dotti et al., 1988), in which the fate of neurites generated at different times can be easily visualized. Tracing the early stages of development in individual neurons by time-lapse video microscopy revealed that the first two neurites arise at opposite poles (Fig. 2A,B). This was confirmed by experiments performed in parallel in fixed cells in which freshly sprouted neurite buds were visible by their high F-actin content: in 88.4% (n=86, supplementary material Fig. S2) of neurons with two neurites these were opposite to each other. More importantly, our time-lapse studies revealed that, in spite of the outgrowth of several neurites after the first two, in 71% of the neurons the axon grew from the first neurite (one example is shown in Fig. 2A) and in 23% from the second, opposite neurite (one example is shown in Fig. 2B n=52 cells) and only in the remaining 6% of the observed neurons from a different location. From these results it becomes clear that the first two neurites have an increased competence to become the axon. In support of this, we observed that the extension of the first neurite is mostly coupled to the retraction of the opposite, second-formed neurite or vice versa (supplementary material Fig. S3, n=6 cells), without any clear association with the development of the other neurites.

Fig. 2.

Neurites and the axon are generated in a stereotypical sequence. (A,B) Time-lapse video microscopy of individual neurons showing that they extend the first two sprouts opposite to each other. (A) An example of a neuron in which the first sprout becomes the axon (arrowheads). (B) Example of a neuron that extended the axon (2; arrowheads) opposite to the first sprout (1). (C) Neurons migrating radially from an aggregate culture in Matrigel. The neurons exit the explant by extending one leading neurite (open arrowhead, 0'). The oppositely located trailing neurite is visible (solid arrowhead). (D) Migrating neurons in Matrigel cultures were fixed and stained with the axonal marker tau-1. Open arrow, leading neurite; arrowhead, trailing neurite. (E,F) Individual neurons were followed from the formation of the first bud (inset) until stage2. Stage-2 cells were fixed and the mean intensity of APC immunoreactivity measured in all neurite growth cones. In 70% of the observed neurons (33 cells) APC mean fluorescence was maximal either in the neurite appearing first (one example shown in E) or in the neurite appearing second (one example shown in F) neurite. Bars, 10 μm.

Fig. 2.

Neurites and the axon are generated in a stereotypical sequence. (A,B) Time-lapse video microscopy of individual neurons showing that they extend the first two sprouts opposite to each other. (A) An example of a neuron in which the first sprout becomes the axon (arrowheads). (B) Example of a neuron that extended the axon (2; arrowheads) opposite to the first sprout (1). (C) Neurons migrating radially from an aggregate culture in Matrigel. The neurons exit the explant by extending one leading neurite (open arrowhead, 0'). The oppositely located trailing neurite is visible (solid arrowhead). (D) Migrating neurons in Matrigel cultures were fixed and stained with the axonal marker tau-1. Open arrow, leading neurite; arrowhead, trailing neurite. (E,F) Individual neurons were followed from the formation of the first bud (inset) until stage2. Stage-2 cells were fixed and the mean intensity of APC immunoreactivity measured in all neurite growth cones. In 70% of the observed neurons (33 cells) APC mean fluorescence was maximal either in the neurite appearing first (one example shown in E) or in the neurite appearing second (one example shown in F) neurite. Bars, 10 μm.

The temporal difference in axon specification observed between neurons in vivo, where axonal characteristics become evident at the bipolar stage, and in vitro, in which the axon extends only after several neurites have formed (see above) (Dotti et al., 1988), might result from the incapacity of neurons grown in vitro to undergo migration. In support of this, hippocampal neurons seeded not on the usual poly-lysine substrate but in Matrigel, which supports migration, developed the typical bipolar architecture and saltatory movement that migrating neurons show in situ (Nadarajah and Parnavelas, 2002), with a clear leading (Fig. 2C, open arrowheads) and trailing neurite (Fig. 2C, filled arrowheads). At later stages one of these principal neurites differentiated into an axon before any other neurite had formed (Fig. 2D). From all the above we find it reasonable to conclude that the neuronal polarity axis is intrinsically instructed at the bipolar stage. That would imply that molecules found to be required for axonal growth (Wiggin et al., 2005), would concentrate to one of the two first neurites, not stochastically. To test this, we analyzed the subcellular distribution of the microtubule (MT) stabilizing protein adenomatous polyposis coli (APC), which is known to be transported in the future axon before axonal outgrowth becomes visible (Gärtner et al., 2006b). To this end, the sequence of neurite generation was monitored in live neurons which were fixed and immunolabelled for APC after 1 day in vitro, the time point when most of the neurons were in the multipolar stage 2. Consistent with our concept of bipolar predisposition, 70% (n=33 neurons) of the analyzed multipolar stage-2 cells, showed APC concentrated maximally in the growth cone of the first (Fig. 2E) or the second (Fig. 2F) neurite to sprout.

MT dynamics and membrane transport have a bipolar organization

To define the mechanisms by which the first or second neurites are conferred with maximal growth capacity, we analyzed the dynamics of MT polymerization and membrane transport, before and during the process of polarization. In freshly plated neurons, after nocodazole washout MT repolymerization occurs asymmetrically in a bipolar fashion: maximally towards the area closest to the MT-nucleation centre and to a lesser degree in the opposite direction, as shown in the images taken of fixed neurons after different times of MT repolymerization (Fig. 3A). The quantification of those results (n=19 cells) is shown in Fig. 3B. The MT organisation remained bipolar even in already polarized neurons (n=20 cells; quantification in Fig. 3C; one example for a stage-3 neuron is shown in the supplementary material Fig. S4B). Next we analysed, by different methods, whether the transport of membranes is polarized during the development of the neuron. Initially we specifically labelled post-Golgi elements, by loading neurons with BODIPY-ceramide. An example of one neuron with three sprouts in which the dynamics of the post-Golgi carriers was observed over time is shown in Fig. 3D. Here the membrane transport is mainly bipolar with a high preference toward the largest sprout. Further, static measurements of the polarized accumulation of post-Golgi carriers in stage-2 neurons also provided evidence of bipolar transport, with the highest accumulation of membrane in the neurite facing the Golgi (quantification: Fig. 3E; n=15 cells). The same was shown by live-imaging of overall membrane transport using a membrane-bound GFP (supplementary material Movie 1). This bipolarity is, as shown for the MT repolymerization, also preserved in fully polarized stage-3 neurons (quantification: Fig. 3F; n=5 cells): For that purpose we used phase-contrast microscopy in living neurons (Bradke and Dotti, 1997) to analyse the number of membrane carriers passing through a defined segment of proximal neurites. The data suggest that the first two neurites have a greater chance to form the polar axis of growth, because of the bipolar arrangement of MT polymerization and membrane traffic, which represent the major neurite growth-promoting forces.

Actin filament instability in the growth cone of a single neurite of stage-2 neurons is sufficient and necessary for this neurite to become the axon, probably by permitting MTs and membranes to invade the growth cone area and to induce growth (Bradke and Dotti, 1999). Therefore, if the organisation of MT polymerization and membrane trafficking is indeed organized in a bipolar manner, actin depolymerisation in all growth cones of stage-2 neurons should result in preferential outgrowth from the first two neurites, at least during the early stages after actin cytoskeleton disruption. To test this, we added the actin depolymerising drug cytochalasin D (cytoD) to neurons in which the time of appearance of all neurites was known. Time-lapse video microscopy analysis of such cells revealed that shortly after the drug was added, outgrowth occurred mainly from the first two neurites (Fig. 3G).

Fig. 3.

Neuronal cytoarchitecture is arranged in a graded bipolar manner. (A-C) MT repolymerization in early plated neurons (3-4 hours in culture). (A) Neurons were treated for 2-3 hours with nocodazole and MT repolymerization was visualized at different times after nocodazole washout in fixed cells. Between 2 and 5 minutes, MTs polymerize mainly towards the pole opposite to the nucleus where the centrosome is located (pole 1 in B). After 10 and 15 minutes, MTs polymerize still towards the opposite pole (pole 2 in B) but less to the sides (pole 3 and 4 in B). (B) Quantification of the amount of MTs in the different poles of the neurons after repolymerization as shown in A (n=19). Neurons were divided in populations in which MTs sprouted towards one pole (as in A 2' and 5') and those with bipolar-orientated MTs (as in A 10'and 15': two poles). (C) The same experiment as in A and B was performed in stage-3 neurons (n=11: 1 pole, n=9: 2 poles). (D) Live cell analysis of post-Golgi membrane traffic after labelling with BODIPY-ceramide. In a cell with three sprouts (1,2,3) membrane traffic is preferentially directed towards the largest lamellipodium (0',1), at later times also towards the opposite pole (2), followed by sporadic traffic to a third pole (3). (E) Quantification of mean intensities of BODIPY-ceramide fluorescence in neurites from cells with three neurites (n=15 cells). (F) The traffic of vesicles and membrane compartments in live cells into all neurites of polarized stage-3 neurons was imaged by phase-contrast microscopy. The number of membrane carriers travelling within 1 minute through a defined proximal and distal neurite segment (n=5 cells) was counted. (B,C,E,F) Statistical analysis: ANOVA test followed by Tukey's multiple comparison test ***P<0.001, **P<0.01, *P<0.05 comparing the values to the first column or second as indicated. (G) F-actin disruption using cytoD (addition of 2 μM cytoD at 0 minutes) in cells with three sprouts, in which the order of the appearance of the sprouts was monitored before (–180 minutes, 0 minutes). CytoD addition leads to the highest growth from the first (1) and second (2) sprout. (H) Neurons were transfected with centrin-1–GFP before plating and centrosome position was monitored after 24 hours in neurons in which initially the first sprout faced the centrosome. These data are presented in the frequency distribution (n=30) graph showing the movement of the centrosome in degrees, with respect its initial position. Bars, 10 μm.

Fig. 3.

Neuronal cytoarchitecture is arranged in a graded bipolar manner. (A-C) MT repolymerization in early plated neurons (3-4 hours in culture). (A) Neurons were treated for 2-3 hours with nocodazole and MT repolymerization was visualized at different times after nocodazole washout in fixed cells. Between 2 and 5 minutes, MTs polymerize mainly towards the pole opposite to the nucleus where the centrosome is located (pole 1 in B). After 10 and 15 minutes, MTs polymerize still towards the opposite pole (pole 2 in B) but less to the sides (pole 3 and 4 in B). (B) Quantification of the amount of MTs in the different poles of the neurons after repolymerization as shown in A (n=19). Neurons were divided in populations in which MTs sprouted towards one pole (as in A 2' and 5') and those with bipolar-orientated MTs (as in A 10'and 15': two poles). (C) The same experiment as in A and B was performed in stage-3 neurons (n=11: 1 pole, n=9: 2 poles). (D) Live cell analysis of post-Golgi membrane traffic after labelling with BODIPY-ceramide. In a cell with three sprouts (1,2,3) membrane traffic is preferentially directed towards the largest lamellipodium (0',1), at later times also towards the opposite pole (2), followed by sporadic traffic to a third pole (3). (E) Quantification of mean intensities of BODIPY-ceramide fluorescence in neurites from cells with three neurites (n=15 cells). (F) The traffic of vesicles and membrane compartments in live cells into all neurites of polarized stage-3 neurons was imaged by phase-contrast microscopy. The number of membrane carriers travelling within 1 minute through a defined proximal and distal neurite segment (n=5 cells) was counted. (B,C,E,F) Statistical analysis: ANOVA test followed by Tukey's multiple comparison test ***P<0.001, **P<0.01, *P<0.05 comparing the values to the first column or second as indicated. (G) F-actin disruption using cytoD (addition of 2 μM cytoD at 0 minutes) in cells with three sprouts, in which the order of the appearance of the sprouts was monitored before (–180 minutes, 0 minutes). CytoD addition leads to the highest growth from the first (1) and second (2) sprout. (H) Neurons were transfected with centrin-1–GFP before plating and centrosome position was monitored after 24 hours in neurons in which initially the first sprout faced the centrosome. These data are presented in the frequency distribution (n=30) graph showing the movement of the centrosome in degrees, with respect its initial position. Bars, 10 μm.

Although bipolar asymmetric organization of MT polymerization and membrane traffic can certainly underlie the sequential and opposite generation of the first two neurites, it is still possible that this is controlled by the position of the centrosome, moving from one pole to the other as shown in cerebellar granule neurons (Zmuda and Rivas, 1998). To analyze this possibility, we tracked the centrosome position in developing neurons expressing the centrosomal protein centrin 1 fused to GFP (centrin-1–GFP) (Piel et al., 2000). This study revealed that the centrosome remains rather stationary during polarization: in 23 out of 30 neurons the centrosome remained static or underwent only small movements of 15-30° (quantification of the results: Fig. 3H, some examples are shown in supplementary material Fig. S5), suggesting that maximal growth rate is the consequence of the bipolar organization of MT polymerization and membrane transport.

Intracellular bipolar organization determines axon growth potential

The above series of results demonstrate that young neurons have a robust intracellular bipolar organization that causes axon outgrowth preferentially from one of these neurites. However, our data also show that axonal growth occurs in most cases from the first neurite. This is also supported by the observation that CytoD addition to cells with two buds resulted in maximal growth from the first neurite (supplementary material Fig. S6). To directly test if axonal fate is the consequence of the intracellular organization of membrane traffic and MT polymerization in the direction of the first neurite, we cut the axon. Our prediction was that, if the bipolar growth forces were sufficient and necessary, a new axon would form from the second pole, not randomly. Indeed, 24 hours after the axotomy, 69% of the cells (n=187 neurons) extended a new axon from the neurite at the opposite pole. One example of such a sequence of axotomy and regrowth is shown in Fig. 4A. In 18% of the neurons the new axon grew from a neurite at the same pole as the original axon, and in 13% of the cells from a different neurite without fixed position in relation to the original axon (the quantification of 187 neurons is shown in Fig. 4B). Thus in 87% of the axotomized neurons the new axon grew according the bipolar predisposition. To further test the importance of the intrinsic bipolar growth organization, we extended this experiment and cut the second, oppositely grown axon, which had appeared after the cut of the first axon. In a high proportion of neurons (eight out of 16 cells), the axon regrew from the opposite pole, i.e. the stump of the original sectioned axon (see Fig. 4C). In four out of 16 cells the axon regrew from the second axon and only in four out of 16 cells from a random position.

Fig. 4.

Axon lesion leads to axon formation at the opposite pole. (A) A polarized neuron (0'; axon: arrowheads), the axon of which was cut (5', arrow), develops a new axon opposite to the transected one (24 hours, arrowheads). Axonal identity was confirmed by tau-1 immunoreactivity (right panel). (B) Frequency distribution representing the site of axon regeneration after axotomy (n=187). 0° is the position of the original axon. Bin size 20°. (C) Double axotomy: after the first transection (5', arrow) the new axon formed opposite (24 hours, arrowheads) to the first one (0'). 24 hours after cutting this new axon (24 hours, arrow), axonal regrowth occurred again from the site of the initial axon (48 hours, arrowheads). Axonal identity was confirmed by tau-1 immunoreactivity (right lower panel). In eight out of the 16 cells that survived the double axotomy, the new axon formed from the original pole (from the stump of the original axon) in four cells or from a neurite that was close to the original axon in the other four cells. In four of the remaining cells the new axon grew from the `second' pole and in the last four cells a new axon grew from a random position. Bars, 10 μm.

Fig. 4.

Axon lesion leads to axon formation at the opposite pole. (A) A polarized neuron (0'; axon: arrowheads), the axon of which was cut (5', arrow), develops a new axon opposite to the transected one (24 hours, arrowheads). Axonal identity was confirmed by tau-1 immunoreactivity (right panel). (B) Frequency distribution representing the site of axon regeneration after axotomy (n=187). 0° is the position of the original axon. Bin size 20°. (C) Double axotomy: after the first transection (5', arrow) the new axon formed opposite (24 hours, arrowheads) to the first one (0'). 24 hours after cutting this new axon (24 hours, arrow), axonal regrowth occurred again from the site of the initial axon (48 hours, arrowheads). Axonal identity was confirmed by tau-1 immunoreactivity (right lower panel). In eight out of the 16 cells that survived the double axotomy, the new axon formed from the original pole (from the stump of the original axon) in four cells or from a neurite that was close to the original axon in the other four cells. In four of the remaining cells the new axon grew from the `second' pole and in the last four cells a new axon grew from a random position. Bars, 10 μm.

To confirm that the growth of the new axon from the opposite neurite was the consequence of a change in the direction of membrane traffic, the intensity of Golgi-derived membrane carriers labelled with the dye BODIPY-ceramide was analyzed in sectioned neurons. This study revealed that membrane trafficking was most intense to the pole opposite to the cut axon, preceding growth of the neurite at this pole (n=15 cells; one example shown in Fig. 5A). To rule out the involvement of centrosome translocation to the opposite site, which could trigger regrowth from this site after axon ablation, the axon of centrin-1–GFP-expressing neurons was cut and centrosome movement was monitored by fluorescence microscopy in live cells. As the example in Fig. 5B shows, the centrosome remained stationary after axon sectioning, at the base of the original axon. The quantification of the angle of centrosome movements before axon sectioning and after axonal regrowth is shown in Fig. 5C (n=26 neurons).

Fig. 5.

Axonal respecification is preceded by the reversion of membrane flow and does not involve rotation of the centrosome. (A) Live cell analysis of post-Golgi membrane traffic visualized by labelling with BODIPY-ceramide after recovery from axotomy (2-3 hours after cutting: arrow) shows that more traffic is directed towards the opposite side (open arrowheads) from the sectioned axon (original axon at 0': arrowheads, n=15). (B) Neurons were transfected with centrin-1–GFP to mark the centrosome position (open arrowhead), and the centrosome was observed before axotomy and after axon regrowth. Arrowheads, axons; arrow, axotomy; open arrowheads, centrin-1–GFP. Bars, 10 μm. (C) Quantification of the results in B: frequency distribution showing centrosome position after axotomy with respect to its original position when the axotomy was performed (bin size 10°; n=26 neurons).

Fig. 5.

Axonal respecification is preceded by the reversion of membrane flow and does not involve rotation of the centrosome. (A) Live cell analysis of post-Golgi membrane traffic visualized by labelling with BODIPY-ceramide after recovery from axotomy (2-3 hours after cutting: arrow) shows that more traffic is directed towards the opposite side (open arrowheads) from the sectioned axon (original axon at 0': arrowheads, n=15). (B) Neurons were transfected with centrin-1–GFP to mark the centrosome position (open arrowhead), and the centrosome was observed before axotomy and after axon regrowth. Arrowheads, axons; arrow, axotomy; open arrowheads, centrin-1–GFP. Bars, 10 μm. (C) Quantification of the results in B: frequency distribution showing centrosome position after axotomy with respect to its original position when the axotomy was performed (bin size 10°; n=26 neurons).

Discussion

We have demonstrated that the first two neurites, which form at opposite poles of the cell, have the highest axonal potential with the first being the most likely to develop into the axon. This temporal and spatial hierarchy of neurite formation and the accompanying bipolar organization of MT repolymerization and membrane traffic constitute, per se, a direct indication that the instructiveness for polarized growth is intrinsically defined. Moreover, one would predict that if the instruction for polarized growth was environmentally controlled, first and second neurites would form from the same pole of the cell, which faces the environmental signal, not at opposite poles, as it occurs. Our results of intrinsic polarity predisposition are supported by early experiments demonstrating that neuronal asymmetry in neuroblastoma cells is defined intrinsically (Solomon, 1981) as well as by studies in grasshopper neurons (Lefcort and Bentley, 1989), in cerebellar granule neurons (Rivas and Hatten, 1995) and by a recent study in primary neurons (de Anda et al., 2005). That polarized growth is driven by intrinsic mechanisms, is supported by observations in situ, showing that migrating neurons become bipolar immediately after their generation (supplementary material Fig. S1A,B) and that such bipolar organization is used for migration and later to confer axonal and apical dendrite identity (Fig. 1) (Hatanaka and Murakami, 2002; Noctor et al., 2004).

The robustness of this bipolar organisation observed in situ becomes evident in our observation that the bipolar phenotype is well preserved in neurons developing in vitro (Fig. 2) in the absence of migration, without developing a leading and trailing neurite. The importance of bipolar organization is even more evident in neurons grown on laminin or tenascin, in which a large population are bipolar with the axon growing from one of the two neurites (Lochter and Schachner, 1993). This is also demonstrated here in neurons allowed to migrate in vitro (Fig. 2C,D). Such neurons display a strong bipolar morphology and the axon grows from one of these poles. It has also been thought for a number of years to be the case in vivo, but until now it has only been supported by morphological evidence, that axon formation takes place when bipolar neurons migrate, in the intermediate zone (Noctor et al., 2004; Shoukimas and Hinds, 1978). Here we confirm the prediction in vivo, showing that neurons display a neurite positive for the axonal marker tau-1 already at the bipolar stage in the upper intermediate zone (Fig. 1).

The `gross' determinant of the preferential growth at the first and second neurites seems to be the bipolar organization of MTs: maximal MT repolymerization occurs towards the place where the first neurite forms, followed by the opposite direction (Fig. 3A-C). Reflecting this organization, post-Golgi compartments labelled by BODIPY-ceramide and membrane compartments in general show the same graded distribution (Fig. 3D-F). This initial intracellular polarization may lead, with time, to a graded accumulation of growth-supporting molecules at both poles. A consequence of this would be that the first and second neurites can eventually respond more efficiently to an environmental growth signal acting on all neurites. This view is supported by the stronger growth response of these two neurites to the addition of the actin depolymerising drug cytoD (Fig. 3G). Considering that axonal growth in vitro occurs only after several neurites have formed, the intrinsic mechanism would only account for polarity predisposition. In fact, numerous studies have demonstrated that efficient growth is regulated by extracellular contacts (Esch et al., 1999; Polleux et al., 1998; Zhang et al., 2007). Since, however, the first and second neurite have the highest chances to become the axon in a homogeneous environment (Figs 2, 4), final polarized growth would be the consequence of subtle intrinsic-extrinsic crosstalk with the intrinsic programme determining the principal cell axis.

Fig. 6.

Stages of neuronal polarity development. The scheme represents the different stages of neuronal polarity development of hippocampal neurons in vitro. See Discussion for details.

Fig. 6.

Stages of neuronal polarity development. The scheme represents the different stages of neuronal polarity development of hippocampal neurons in vitro. See Discussion for details.

An open question from this work concerns the `fine' molecular mechanisms by which the first neurite later becomes the axon. This asymmetry may be caused by a polarized accumulation of the same molecules that play a role in apical-basal segregation of neuronal precursor cells in the ventricular zone, such as Par3 or aPKC (Huttner and Kosodo, 2005; Wodarz, 2005). However, we did not see such molecules concentrated at the pole where the first or second neurite form (see example in supplementary material Fig. S7), suggesting that, if polarity is triggered by a qualitative event, rather than a quantitative one (see above), the responsible molecules might be different from those already identified in other systems. This notion needs further testing.

In the end, our new data combined with recent observations (de Anda et al., 2005) allow us to modify the current view (Dotti et al., 1988) of how hippocampal neurons differentiate in vitro. This model is shown schematically in Fig. 6. In this model, the neuronal polarity axis is `marked' in the immediate post-mitotic, round cell through a polarized organization of the cytoplasm (stage 0: polarity instruction). This stage is followed by the appearance of a sprout from the vicinity of the centrosome and subsequently the bipolar axis becomes visible by the sprouting of a second neurite from the opposite poles (stage 1 monopolar/bipolar: polarity preference). Then, and despite the outgrowth of additional neurites, during the multipolar stage 2, sudden growth occurs from one of the two predisposed neurites (stage 3: polarity commitment). Thus this model presents a new view of the timing and mechanism of polarity instruction: we show that the site of axon outgrowth is predefined already at the initial stages and until the formation of the first two neurites the process seems to be exclusively intracellularly driven. Since the attention of most of the recent work on that subject was mainly focussed at the stage 2 to 3 transition, i.e. the axon outgrowth, this early process was never studied before. At stage 2, several polarity molecules accumulate selectively in one neurite (Arimura and Kaibuchi, 2007), thus becoming differentially responsible to growth and towards extracellular cues. Most likely, the preferential accumulation of these molecules follows the bipolar intracellular organization that we describe here and as we show for APC. Therefore, in the homogeneous culture condition, the first and second neurite have the highest probability to become the axon unless selected neurites are differentially challenged with axon growth-promoting cues (Esch et al., 1999; Esch et al., 2000; Polleux et al., 1998).

Materials and Methods

In utero electroporation

All experiments were approved by the Institutional Animal Care and Use Committee of Massachusetts Institute of Technology. Pregnant Swiss Webster mice were anesthetized by intraperitoneal injections of ketamine 1% and xylazine 2 mg/ml (0.01 μl/g body weight), uterine horns were exposed and plasmid (1 μg/μl; pCAGIG-Venus-EGFP) mixed with Fast Green (Sigma) was microinjected in the lateral ventricles of embryos. Five current pulses (50 msecond pulses/950 msecond intervals) were delivered across the head of the embryos (35 V).

Hippocampal cultures

Hippocampal neurons from E17 or E18 mouse embryos were prepared (Banker and Goslin, 1988) and plated at a density of 2,500 cells per cm2 on poly-L-lysine (PLL) or on laminin (20 ng/ml; Invitrogen)-coated coverslips. Neurons were transfected in suspension using nucleofection (Amaxa) (Gärtner et al., 2006a). Transfected neurons were kept in suspension for 1-2 hours to allow transgene expression and plated after careful resuspension on poly-L-lysine-coated gridded dishes (MatTek). Centrin-1–GFP cDNA was kindly provided by M. Piel (Piel et al., 2000). The development of individual neurons was observed on Cellocate coverslips (Eppendorf). For migration studies neuronal aggregates were made by shaking dissociated hippocampal neurons prepared as described above, at 350 rpm overnight. Cell aggregates were collected by sedimentation and plated in a 1:3 dilution in Matrigel (BD Biosciences).

Immunofluorescence

Dissociated neurons

Neurons were fixed with 4% paraformaldehyde (PFA; with 1.44 M sucrose, 1 M MgCl2, 100 mM EGTA) at 37°C for 10 minutes or for 3 minutes followed by fixation for 3 minutes in methanol at –20°C. Cells were permeabilized for 3 minutes in 0.1% Triton X-100-PBS. After blocking in 2% FBS, 2% BSA and 0.2% fish gelatine in PBS, neurons were incubated with the primary antibodies.

Cortical sections

Embryos were perfused (4% PFA), their brains removed and kept overnight in 4% PFA and afterwards in 30% sucrose-PBS (4°C). Brains were embedded in OCT compound. Cryosections, 30-40 μm thick, were labelled overnight at 4°C with primary antibodies.

The following antibodies were used: anti-α-tubulin antibody (Calbiochem); anti-tau-1 (MAB3420, Chemicon); anti-nestin (Chemicon); anti-pericentrin (Covance) and rabbit anti-APC (kindly provided by Inke Näthke, Cancer Research UK, Dundee, UK). F-actin was detected with TRITC-conjugated phalloidin (Sigma), nuclei by Hoechst 33342 staining (Invitrogen). Alexa-Fluor-350, -488 or -568-conjugated secondary antibodies (Molecular Probes) were used. Colocalization was determined by comparison of single confocal z-sections using a Zeiss LSM 510 microscope.

Live cell imaging

Neurons were plated into slide flasks (Nunc) filled with medium and their differentiation followed for up to 48 hours by time-lapse video microscopy (1 frame per 5 minutes) using an inverted microscope in a temperature controlled heated chamber. For centrosome imaging in live cells, centrin1-GFP-transfected neurons were plated on 3.5-cm glass bottom dishes with gridded coverslips (MatTek) and observed using a confocal microscope (Zeiss) or by fluorescence microscopy.

Live cell analysis of membrane traffic

Imaging of post-Golgi compartments using BODIPY-ceramide

Neurons were loaded for 10 minutes at 37°C with 5 μM BODIPY-ceramide (Molecular Probes) (Bradke and Dotti, 1997), washed and BODIPY-ceramide fluorescence was monitored 10-30 minutes later on a heated stage in a closed aluminium chamber. The mean intensity in each sprout or neurite was measured and normalized by using ImageJ.

Live imaging of membrane traffic

For imaging of membrane traffic using phase-contrast microscopy, images were captured in 1-second intervals for 100 seconds. With this method, very dynamic round and tubular membrane compartments are easily visible (Bradke and Dotti, 1997). The amount of membrane traffic over time was analyzed by counting the number of membrane carriers passing through a proximally (immediately at the neurite shaft) and a distally (approximately half way along the neurite length) located neurite segment of 10 μm.

Measuring repolymerization of microtubules after nocodazole-induced depolymerization

At 2-3 hours (stage 0) or at 48 hours (stage 3) after plating, neurons were treated for 2 hours with 7 μM nocodazole, which leads to a complete depolymerization of microtubules (MTs). Repolymerization of MTs was visualized by indirect immunofluorescence at different times (2-15 min) after nocodazole washout. Before fixation cells were treated for 1-2 minutes with 0.01-0.5% Triton X-100 MT-stabilizing buffer (MBS: 60 mM PIPES pH 7, 2 mM MgCl2, 10 mM EGTA) to extract free cytoplasmic MTs. For quantification, neuronal spheres were divided in four quarters using the centrosome position as the centre and considering the first quarter as ±45° with respect to a line drawn through the middle of the cell axis. Pole 1 is the one opposite to the nucleus in the centrosome area. Pole 2 is the pole where the nucleus is occupying more space, and poles 3 and 4 are perpendicular to 1 and 2 (see supplementary material Fig. S4A). The mean intensity in each quarter/pole was measured using ImageJ.

Axon lesioning

The axon was cut by quickly moving a microinjection needle, fitted in a micromanipulator (Eppendorf), over the glass surface orthogonally to the axon, leaving an axonal stump not longer than the minor neurites. Individual neurons were identified by their location on a gridded coverslip and monitored before and after the cut and after the outgrowth of a new axon after approximately 24 hours. Regrowths from exactly the same neurite were not taken into account for this quantification since they would also occur if the axon was not sectioned short enough with respect to the other neurites (Goslin and Banker, 1989).

Acknowledgements

We thank P. Malatesta (IST Genova) for programming the ImageJ plugin. F.C.d.A. was supported by an EMBO long-term fellowship and A.G. by an Otto-Hahn fellowship of the MPG. L.-H.T. is an investigator of the Howard Hughes Medical Institute.

References

References
Arimura, N. and Kaibuchi, K. (
2007
). Neuronal polarity: from extracellular signals to intracellular mechanisms.
Nat. Rev. Neurosci.
8
,
194
-205.
Banker, G. and Goslin, K. (
1988
). Developments in neuronal cell culture.
Nature
336
,
185
-186.
Bradke, F. and Dotti, C. G. (
1997
). Neuronal polarity: vectorial cytoplasmic flow precedes axon formation.
Neuron
19
,
1175
-1186.
Bradke, F. and Dotti, C. G. (
1999
). The role of local actin instability in axon formation.
Science
283
,
1931
-1934.
Bradke, F. and Dotti, C. G. (
2000
). Establishment of neuronal polarity: lessons from cultured hippocampal neurons.
Curr. Opin. Neurobiol.
10
,
574
-581.
Buchman, J. J. and Tsai, L. H. (
2007
). Spindle regulation in neural precursors of flies and mammals.
Nat. Rev. Neurosci.
8
,
89
-100.
Craig, A. M., Jareb, M. and Banker, G. (
1992
). Neuronal polarity.
Curr. Opin. Neurobiol.
2
,
602
-606.
Da Silva, J. S. and Dotti, C. G. (
2002
). Breaking the neuronal sphere: regulation of the actin cytoskeleton in neuritogenesis.
Nat. Rev. Neurosci.
3
,
694
-704.
De Anda, F. C., Pollarolo, G., Da Silva, J. S., Camoletto, P. G., Feiguin, F. and Dotti, C. G. (
2005
). Centrosome localization determines neuronal polarity.
Nature
436
,
704
-708.
Dotti, C. G., Sullivan, C. A. and Banker, G. A. (
1988
). The establishment of polarity by hippocampal neurons in culture.
J. Neurosci.
8
,
1454
-1468.
Esch, T., Lemmon, V. and Banker, G. (
1999
). Local presentation of substrate molecules directs axon specification by cultured hippocampal neurons.
J. Neurosci.
19
,
6417
-6426.
Esch, T., Lemmon, V. and Banker, G. (
2000
). Differential effects of NgCAM and N-cadherin on the development of axons and dendrites by cultured hippocampal neurons.
J. Neurocytol.
29
,
215
-223.
Ferreira, A. and Caceres, A. (
1989
). The expression of acetylated microtubules during axonal and dendritic growth in cerebellar macroneurons which develop in vitro.
Brain Res. Dev. Brain Res.
49
,
205
-213.
Gärtner, A., Collin, L. and Lalli, G. (
2006a
). Nucleofection of primary neurons.
Meth. Enzymol.
406
,
374
-388.
Gärtner, A., Huang, X. and Hall, A. (
2006b
). Neuronal polarity is regulated by glycogen synthase kinase-3 (GSK-3{beta}) independently of Akt/PKB serine phosphorylation.
J. Cell Sci.
119
,
3927
-3934.
Goslin, K. and Banker, G. (
1989
). Experimental observations on the development of polarity by hippocampal neurons in culture.
J. Cell Biol.
108
,
1507
-1516.
Hatanaka, Y. and Murakami, F. (
2002
). In vitro analysis of the origin, migratory behavior, and maturation of cortical pyramidal cells.
J. Comp. Neurol.
454
,
1
-14.
Hatten, M. E. (
1999
). Central nervous system neuronal migration.
Annu. Rev. Neurosci.
22
,
511
-539.
Huttner, W. B. and Kosodo, Y. (
2005
). Symmetric versus asymmetric cell division during neurogenesis in the developing vertebrate central nervous system.
Curr. Opin. Cell Biol.
17
,
648
-657.
Kriegstein, A. R. and Noctor, S. C. (
2004
). Patterns of neuronal migration in the embryonic cortex.
Trends Neurosci.
27
,
392
-399.
Lefcort, F. and Bentley, D. (
1989
). Organization of cytoskeletal elements and organelles preceding growth cone emergence from an identified neuron in situ.
J. Cell Biol.
108
,
1737
-1749.
Lochter, A. and Schachner, M. (
1993
). Tenascin and extracellular matrix glycoproteins: from promotion to polarization of neurite growth in vitro.
J. Neurosci.
13
,
3986
-4000.
LoTurco, J. J. and Bai, J. (
2006
). The multipolar stage and disruptions in neuronal migration.
Trends Neurosci.
29
,
407
-413.
Nadarajah, B. and Parnavelas, J. G. (
2002
). Modes of neuronal migration in the developing cerebral cortex.
Nat. Rev. Neurosci.
3
,
423
-432.
Nadarajah, B., Brunstrom, J. E., Grutzendler, J., Wong, R. O. and Pearlman, A. L. (
2001
). Two modes of radial migration in early development of the cerebral cortex.
Nat. Neurosci.
4
,
143
-150.
Nadarajah, B., Alifragis, P., Wong, R. O. and Parnavelas, J. G. (
2002
). Ventricle-directed migration in the developing cerebral cortex.
Nat. Neurosci.
5
,
218
-224.
Nadarajah, B., Alifragis, P., Wong, R. O. and Parnavelas, J. G. (
2003
). Neuronal migration in the developing cerebral cortex: observations based on real-time imaging.
Cereb. Cortex
13
,
607
-611.
Noctor, S. C., Martinez-Cerdeno, V., Ivic, L. and Kriegstein, A. R. (
2004
). Cortical neurons arise in symmetric and asymmetric division zones and migrate through specific phases.
Nat. Neurosci.
7
,
136
-144.
O'Rourke, N. A., Dailey, M. E., Smith, S. J. and McConnell, S. K. (
1992
). Diverse migratory pathways in the developing cerebral cortex.
Science
258
,
299
-302.
Parnavelas, J. G., Alifragis, P. and Nadarajah, B. (
2002
). The origin and migration of cortical neurons.
Prog. Brain Res.
136
,
73
-80.
Piel, M., Meyer, P., Khodjakov, A., Rieder, C. L. and Bornens, M. (
2000
). The respective contributions of the mother and daughter centrioles to centrosome activity and behavior in vertebrate cells.
J. Cell Biol.
149
,
317
-330.
Polleux, F., Giger, R. J., Ginty, D. D., Kolodkin, A. L. and Ghosh, A. (
1998
). Patterning of cortical efferent projections by semaphorin-neuropilin interactions.
Science
282
,
1904
-1906.
Rivas, R. J. and Hatten, M. E. (
1995
). Motility and cytoskeletal organization of migrating cerebellar granule neurons.
J. Neurosci.
15
,
981
-989.
Shelly, M., Cancedda, L., Heilshorn, S., Sumbre, G. and Poo, M. M. (
2007
). LKB1/STRAD promotes axon initiation during neuronal polarization.
Cell
129
,
565
-577.
Shoukimas, G. M. and Hinds, J. W. (
1978
). The development of the cerebral cortex in the embryonic mouse: an electron microscopic serial section analysis.
J. Comp. Neurol.
179
,
795
-830.
Solomon, F. (
1981
). Specification of cell morphology by endogenous determinants.
J. Cell Biol.
90
,
547
-553.
Tabata, H. and Nakajima, K. (
2003
). Multipolar migration: the third mode of radial neuronal migration in the developing cerebral cortex.
J. Neurosci.
23
,
9996
-10001.
Wiggin, G. R., Fawcett, J. P. and Pawson, T. (
2005
). Polarity proteins in axon specification and synaptogenesis.
Dev. Cell
8
,
803
-816.
Wodarz, A. (
2005
). Molecular control of cell polarity and asymmetric cell division in Drosophila neuroblasts.
Curr. Opin. Cell Biol.
17
,
475
-481.
Zhang, X., Zhu, J., Yang, G. Y., Wang, Q. J., Qian, L., Chen, Y. M., Chen, F., Tao, Y., Hu, H. S., Wang, T. et al. (
2007
). Dishevelled promotes axon differentiation by regulating atypical protein kinase C.
Nat. Cell Biol.
9
,
743
-754.
Zmuda, J. F. and Rivas, R. J. (
1998
). The Golgi apparatus and the centrosome are localized to the sites of newly emerging axons in cerebellar granule neurons in vitro.
Cell Motil. Cytoskeleton
41
,
18
-38.

Supplementary information