Asymmetric membrane ganglioside sialidase activity specifies axonal fate

Share Embed


Descrição do Produto

ARTICLES

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

Asymmetric membrane ganglioside sialidase activity specifies axonal fate Jorge Santos Da Silva1, Takafumi Hasegawa2, Taeko Miyagi2, Carlos G Dotti1,3 & Jose Abad-Rodriguez1 Axon specification triggers the polarization of neurons and requires the localized destabilization of filamentous actin. Here we show that plasma membrane ganglioside sialidase (PMGS) asymmetrically accumulates at the tip of one neurite of the unpolarized rat neuron, inducing actin instability. Suppressing PMGS activity blocks axonal generation, whereas stimulating it accelerates the formation of a single (not several) axon. PMGS induces axon specification by enhancing TrkA activity locally, which triggers phosphatidylinositol-3-kinase (PI3K)- and Rac1-dependent inhibition of RhoA signaling and the consequent actin depolymerization in one neurite only. Thus, spatial restriction of an actin-regulating molecular machinery, in this case a membrane enzymatic activity, before polarization is enough to determine axonal fate.

The specification of a single axon from the population of early neurites is fundamental to the establishment of axonal and somatodendritic territories and is thus critical for neuronal function1. An early event of neuronal polarity is the destabilization of the actin cytoskeleton in only one of the multiple growth cones of unpolarized neurons2. It has been proposed that such localized change prompts polarized membrane trafficking and protrusion of microtubules into the selected neurite to sustain outgrowth and thus establish polarity3. Numerous molecules have been suggested to be involved in neuronal polarization4. Many accumulate in the already-specified axon, and their overexpression enhances elongation, such as Par3/Par6/aPKC complex5, whereas others, such as CRMP-2, induce multiple axons when overexpressed6. These results are consistent with the view that in steady-state situations, rather than polarizing growth, these molecules confine elongation to a previously chosen neurite. This is essential to establish polarity, in the sense that continuous elongation permits the acquisition of specific axonal characteristics, such as the particular microtubular and membrane polarity1. How the elongation machinery is spatially restricted to work in a single growth cone of the pre-polarized neuron is, however, unknown. Establishment of a spatial membrane landmark is the first step in cell polarization7,8. This onset event is upstream of other processes such as cytoskeletal changes or membrane trafficking that then support and consolidate a polarized cellular domain. This type of priming event occurs in epithelial cells9, yeast10 and migrating fibroblasts11. Here we describe a landmark mechanism for establishing neuronal polarity that depends on the locally enriched membrane activity of the gangliosideconverting enzyme PMGS in one growth cone. This induces the spatially restricted regulation of molecules involved in axon outgrowth (such as PI3K5, Rac1 and RhoA7) that modify local actin stability to define axonal fate.

RESULTS Early segregation of PMGS to a single neurite Polarized membrane flow is essential for the induction of neuronal polarity12. Thus, the generation of a spatially different plasma membrane may be important for determining the local changes in actin dynamics that are necessary to trigger axon formation2. PMGS specifically hydrolyzes gangliosides to generate plasma membrane areas that are enriched in GM1, a modulator of intracellular signaling13–15. Thus, we analyzed PMGS distribution before and during neuronal polarization. We took advantage of primary cultures of embryonic (E) 17 hippocampal neurons, as these cells are unpolarized (all neurites have similar lengths) at 24 h in culture (stage 2) and become distinctively polarized 24 h later, as one neurite extends faster than the others to become the future axon (stage 3)16. In a significant number of unpolarized (stage 2) neurons (57.1 7 4.6%, n ¼ 40), PMGS was enriched in the growth cone of only one process (Fig. 1a and see Supplementary Fig. 1 online). To determine whether such polarized distribution could correspond to that of the neurite that will become the axon, we analyzed the content of polymerized actin in the same cells. According to previous findings, the growth cone with the lowest actin filament content in an unpolarized neuron will become the axon2. In fact, PMGS and F-actin signal intensities showed an inverse correlation: the neurite with the lowest actin filament content had the highest PMGS concentration, and those neurites with more F-actin had less PMGS (Fig. 1a and Supplementary Fig. 1). To determine whether the polarization of PMGS to a single neurite of unpolarized neurons marks the axon, we evaluated its distribution in slightly (stage 2+) and evidently (stage 3) polarized neurons. In stage 2+ cells, PMGS accumulated along and at the tip of the longest neurite, the one destined to become the axon (Fig. 1b and Supplementary Fig. 1). This was evident in most cells (74.3 7 4.9%, n ¼ 37;

1Cavalieri Ottolenghi Scientific Institute, University of Turin, 10043 Orbassano, Turin, Italy. 2Miyagi Prefectural Cancer Center, Division of Biochemistry, 47-1 Nodayama, Medeshima-Shiode, Natori, Miyagi, 981-1293 Japan. 3Center for Human Genetics, Catholic University of Leuven and Flanders Interuniversitary Institute for Biotechnology, 3000 Leuven, Belgium. Correspondence should be addressed to J.A.-R. ([email protected]) or C.G.D. ([email protected]).

Published online 17 April 2005; doi:10.1038/nn1442

NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION

1

ARTICLES

a

b

1 PMGS fluorescence intensity at tip (a.u.)

1 2

3

5

PMGS

F-actin

* PMGS

0

4

F-actin

Less F-actin

Others

F-actin

PMGS

3

1

2

3

4

F-actin fluorescence intensity at tip (a.u.) 0

5

PMGS fluorescence intensity at tip (a.u.)

F-actin

PMGS fluorescence intensity at tip (a.u.)

PMGS

1

2

3

c

4

5

* 0 Longer

Shorter

0

d

1

F-actin

1

2

2

F-actin

3 3

PMGS HA

1

F-actin

3.5

*

0 Axons

Dendrites

F-actin fluorescence intensity at tip (a.u.)

PMGS

PMGS fluorescence intensity at tip (a.u.)

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

F-actin

1 PMGS

4

* 0 OEPMGS neurites

OEPMGS axon

Control neurites

Figure 1 Early segregation of PMGS to a single neurite correlates with axonal fate and reduced F-actin content. (a) Stage 2 neurons (24 h in vitro). PMGS was enriched in one neurite tip only (arrow), which correlated with localized reduction of F-actin (see line graph and insets corresponding to numbered growth cones). In this neuronal population, neurite tips with less F-actin (future axons) accumulated more PMGS (bar diagram; n ¼ 47). (b) Stage 2+ neurons (36 h in vitro). PMGS enrichment occurred in the future axon (arrows and graph; n ¼ 37 cells/three experiments), especially in areas of high actin turnover such as lamellipodia and the growth cone (open triangles at right). (c) Stage 3 neurons (48 h in vitro). PMGS segregated to the newly formed axon (arrows and graph; n ¼ 31 cells/three experiments) at the tip and within actin-related structures (open triangles). PMGS accumulation was detected where F-actin was less abundant. Scale bar, 10 mm. In a–c, *P o 0.001. (d) Hemagglutinin (HA)-PMGS transfection (OE PMGS) accelerated specification of a single axon (24 h in vitro), as compared with neighboring untransfected (control) cells. F-actin was significantly reduced in the OE PMGS axons (n ¼ 66), as compared with the OE PMGS neurite tips and with the tips of all neurites in control cells (n ¼ 63; *P o 0.001, compared with OE PMGS neurites and with control neurites). Data are mean + s.d.; a.u., arbitrary units.

Supplementary Figs. 1 and 2). This enrichment was particularly clear in the distal regions of the growth cone that had the lower F-actin content (Fig. 1b). In polarized neurons (stage 3), PMGS was also restricted to discrete areas along and at the tip of the newly formed axon (84.9 7 4.1% of cells, n ¼ 31), where actin filaments were less abundant (Fig. 1c and Supplementary Fig. 1). In our analysis, axons were defined as tau-positive processes that were longer than 40 mm and more than three times the length of any other neurite (thus cells with at least one neurite longer than 40 mm were considered to be polarized; Supplementary Fig. 2). These results indicate a strong correlation between PMGS enrichment and axon specification. To directly ascertain this, we carried out a time-lapse analysis of living stage 2 and 2+ neurons. In these cells, the future axon is readily identifiable by the larger growth cone area, the higher motility of this cone and the reduced content of actin filaments. In addition, in stage 2+ cells, the future axon is further distinguishable by its greater length in comparison with the others neurites2,12,16–20. Cells were imaged for 30 min at 2-min intervals, and the area and motility of all growth cones

2

were determined along the imaging period (Supplementary Videos 1 and 2 and Supplementary Fig. 3). This, together with the detection of F-actin density per cone, allowed the unequivocal determination of the future axon. Notably, PMGS enrichment coincided with this neurite in living cells (Supplementary Videos 1 and 2 and Supplementary Fig. 3). This first series of results indicates that the polarization of the membrane enzyme PMGS can act as a landmark for neuronal polarization. PMGS activity specifies the single axon If the early (stage 2) spatial segregation of PMGS is required to induce polarization, then increasing its activity should influence axon specification. Considering that transient transfection in cultured neurons allows for ectopic expression to start a few hours into the differentiation program and considering that endogenous PMGS is already polarized at very early time points, we envisaged two possible outcomes. First, the overexpressed protein could be delivered equally to all neurites (unlike the endogenous protein) and this would result in the formation of

ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

ARTICLES iRNA

a

Non-targeting RNA

* 0

PMGS

0

iRNA

ntRNA

iRNA

ntRNA

Neurite length (mm)

25

GFP

0

2

3

F-actin

3 2

F-actin

4.5 PMGS F-actin

PMGS fluorescence intensity at tip (a.u.)

2

1

1

0

PMGS

1

2

3

F-actin fluorescence intensity at tip (a.u.)

b

PMGS

0

Growth cone number

c

0:00

0:10

*

d

0:00

0:10

* *

0:00

* *

0:10

e

0:15

* F-actin

0:15

* PMGS

*

0:00

0:15

0:15

*0:10 *

F-actin

PMGS

Growth cone area (change in percentage from time 0:00)

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

130

PMGS fluorescence intensity (a.u.)

Tau

Cells with axons (percentage of total)

40

60

Control perfused cones Control non-perfused cones NeuAc2en perfused cones NeuAc2en non-perfused cones

40 20 0 –20 –40 –60 –80

Perfusion time (0:05 intervals)

Figure 2 PMGS activity specifies the single axon by reducing local F-actin concentration. (a) Hippocampal neurons were cotransfected with pEGFP-N1 cDNA plus specific iRNA targeted against PMGS (iRNA) or nontargeting iRNA (nt-iRNA) and were immunostained with Tau-1 and anti-PMGS. Data are mean + s.d. from three experiments evaluated by two independent observers. At least 100 iRNA- and 100 nt-iRNA-transfected cells were analyzed per experiment. *P o 0.001. (b) PMGS inhibition by NeuAc2en treatment (48 h in vitro) blocked axon formation (compare with Fig. 1c). Arrow indicates single PMGS-positive growth cone. F-actin concentration is not reduced (graph). (c) Local buffer perfusion of a single growth cone (asterisk indicates likely future axon) of a stage 2 neuron. Small panels on the left show general growth cone structure. (d) Local NeuAc2en perfusion induced growth cone area reduction (small panels at left) and favored local actin polymerization. Polarized distribution of PMGS was not affected. (e) Quantification of the change in growth cone area over perfusion time for the whole neuronal population. Local PMGS inhibition markedly reduces growth cone area. (n ¼ 19). Scale bar, 10 mm.

multiple axon-like neurites (fast growing, tau positive). Second, overexpression could result in preferential trafficking of the overexpressed protein to one neurite (in a manner similar to the endogenous protein), producing the acceleration of axon specification. We found that increased PMGS activity, as stimulated by hemagglutinin-PMGS overexpression21, induced growth of a single axon at 24 h (Fig. 1d) and did not affect the growth of the other neurites (future dendrites; Supplementary Fig. 4). In this situation, 83.6 7 5.1% neurons (n ¼ 47) were polarized, a percentage that was reached only after 48 h in nonoverexpressing cells (compare lengths in Fig. 1a,c). Hemagglutinin-PMGS was still asymmetrically enriched in the newly formed axon and was still found to enrich at the neurite tip of this process (Fig. 1d and Supplementary Fig. 4). Consistent with the finding that excess

NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION

PMGS induced higher enzymatic activity locally, the product GM1 was also more abundant in the specified axon (Supplementary Fig. 4). Additionally, the axon showed a five- to sixfold reduction in F-actin content, as compared with the average in all other neurite tips of the transfected cells and in all neurite tips of controls (Fig. 1d). To test if PMGS needs to reach an activity threshold to trigger polarity, its synthesis was precluded by interference RNA (iRNA; Fig. 2a and Supplementary Fig. 5). Neurons were cotransfected with EGFP-cDNA and PMGS-targeting or non-targeting iRNAs and were plated, and the morphological phenotype and polarization of the tau protein were analyzed after 48 h. Only the cells that were transfected with PMGS-targeting iRNA showed a strong reduction in PMGS expression (Fig. 2a and Supplementary Fig. 5). Phenotypically, most

3

ARTICLES HA-PMGS

HA-PMGS

OE PMGS + NeuAc2en

HA

OE PMGS + NGF + NeuAc2en

Cells with neurites longer than 40 µm (percentage of total)

OE PMGS + NeuAc2en

HA

Tubulin

OE PMGS + NGF + NeuAc2en

of these cells seemed to be unpolarized. Careful, double-blinded quantitative analysis showed that only 4.8% of the transfected cells showed a neurite longer than 40 mm (Fig. 2a), as compared with 24.2% of neurons transfected with nontargeting iRNA. This percentage of polarized cells in the nontargeting transfected cells was identical to that observed in control untransfected cells, reflecting the normal scenario for cells maintained in vitro during 48 h16. That the PMGS-targeting iRNA interfered with axon generation is indicated by the fact that 95.2% of these cells did not present tau polarization (Fig. 2a). The remaining 4.8% had some degree of tau polarization, which indicates that the penetrance of the inhibition was not total. Unlike its effect on axonal generation, PMGS iRNA transfection did not affect the length of the minor neurites (Fig. 2a). To confirm that the observed effects are due to the inhibition of PMGS enzymatic activity and not to the absence of the PMGS protein, neurons were exposed to the specific PMGS inhibitor 2,3,dehydro-2, deoxy-N-acetylneuraminic acid (NeuAc2en; see ref. 22 and references therein), which blocks enzymatic activity without changing transcription levels. After 48 h of treatment, neurons remained morphologically unpolarized (85.7 7 1.6% unpolarized cells, n ¼ 41; Figs. 1c and 2b and Supplementary Fig. 4). Notably, PMGS was still distributed in a polarized fashion, but no localized reduction of F-actin content was observed (Fig. 2b and Supplementary Fig. 1). These results, together with the overexpression data, indicate that PMGS activity may be needed to reach a certain local threshold to trigger the necessary actin changes that underlie axon specification. To test this further, we locally perfused NeuAc2en at the tip of the future axon of stage 2 and 2+ neurons, to thus inhibit PMGS activity in a spatially restricted fashion.

c

90

** *

60

30

0

Control OE PMGS +NGF

+BDNF OE PMGS

+NT3

Control + NGF + NeuAc2en

Control + NGF

Tubulin

Figure 3 NGF signaling requires PMGS activity. (a) Neurons that overexpressed hemagglutinin-PMGS were treated for 24 h with NGF, BDNF or NT3. Data are mean + s.d.; n ¼ 4 experiments; *P o 0.001 as compared with buffer-treated controls; **P o 0.001 as compared with OE PMGS. (b) PMGS inhibition prevented polarization. NGF did not potentiate PMGS effect (compare with a) when PMGS activity was inhibited (+ NGF + NeuAc2en). (c) By itself, NGF induced axon specification at 24 h (left panel), but PMGS inhibition (right panel) precluded the NGF effect. (d) Summary analysis of findings illustrated in b and c. Data are mean + s.d.; n ¼ 3; *P o 0.001, +NGF is compared with respective –NGF control; **P o 0.001, +NeuAc2en effects are compared with respective starting backgrounds (adjacent to the black bar). Scale bar, 10 mm.

4

Tubulin

OEPMGS + NT3

OEPMGS + BDNF

Control OE PMGS

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

b

Tubulin

OE PMGS + NGF

Tubulin

Tubulin

d

Cells with neurites longer than 40 µm (percentage of total)

HA-PMGS

a

Tubulin

90

*

+ NeuAc2en 60

30

**

**

* **

0 –NGF Control

+NGF

–NGF

+NGF

OE PMGS

The future axons in these cells were identified by their larger growth cone size and greater motility. Local buffer perfusion did not affect the structure and average area of the growth cone of the future axon (Fig. 2c,e and Supplementary Fig. 6) nor did it modify the distribution of actin filaments (which were less abundant in the future axon; Fig. 2c and Supplementary Fig. 6). The enrichment of PMGS in the future axon was similarly not altered (Fig. 2c and Supplementary Fig. 6). Conversely, local inhibition of PMGS activity triggered actin polymerization at the tip of the future axon, although the distribution of the enzyme was not affected (Fig. 2d and Supplementary Fig. 6). This local stabilization of the actin cytoskeleton correlated with a marked retraction of the perfused growth cone as illustrated by a sharp reduction in growth cone area (Fig. 2d,e and Supplementary Fig. 6; refs. 23–26). Notably, neither buffer nor NeuAc2en altered the growth cone area of nonperfused growth cones of the same cells (Fig. 2e and Supplementary Fig. 6). Together with the previous data, these results indicate that suprathreshold PMGS activity that is spatially restricted to the tip of one neurite specifies a single neurite to become the axon. This paves the way to define which actin-regulating molecular mechanism is particularly recruited and activated by PMGS to trigger axon specification. PMGS modulates the effects of NGF during polarization PMGS activity regulates plasma membrane synthesis of GM1, a ganglioside known to modulate neurotrophin Trk receptor–mediated signaling27. Hence, we analyzed whether PMGS activity could influence or be influenced by the neurotrophins nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF) or neurotrophin-3 (NT3).

ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

ARTICLES

a

b Control Control OE PMGS + NGF

OE PMGS + NGF

+ NGF Anti-HA

+

+

HA-PMGS



+

+ AP

– Lys

+

+

+

+

+



c

Chtx-B + anti-HRP HA-PMGS

+ –

– +

+ +

pTrkA

pTrkA

HA

HA

Mann2

Mann2

Lys

TrkA

Lys

d

IP

HA

IP Lys

Mann2

1 2

Control (st2+, 36 h)

3

4

5

OE PMGS (36 h)

2

3

6 + NeuAc2en (36 h)

pTrkA

PMGS

Phase

Control (st2, 36 h)

IP

IP

pTrkA

1

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

– +

Figure 4 PMGS leads to localized activity of TrkA receptors. (a) Western blot of neuronal membrane (24 h). Buffer-treated neurons had low amounts of pTrkA (Control). Hemagglutinin-PMGS overexpression increased pTrkA amounts (OE PMGS). NGF increased pTrkA slightly but clearly in controls (Control + NGF) and potentiated TrkA activation induced by PMGS overexpression (OE PMGS + NGF). (b) Western blot of neuronal extract immunoprecipitations (IP) with anti-hemagglutinin (HA). Hemagglutinin-PMGS interacted with pTrkA (lane 4), an interaction that was potentiated by NGF (lane 2). Hemagglutinin-PMGS expression (lane 1), beads alone without antibody (lane 3) or absence of cell lysate (–Lys; lane 6) did not yield pTrkA signals. Alkaline phosphatase (+AP) reduced pTrkA phosphorylation (lane 5). TrkA amounts were similar in all conditions. (c). Western blots of neuronal extract immunoprecipitations. Endogenous GM1 precipitated with pTrkA, albeit at very low amounts (lane 1). More GM1 (produced by hemagglutinin-PMGS overexpression) precipitated pTrkA more efficiently (lane 3). No signal was detected when beads were incubated without antibodies (lane 2). Chtx-B, HRP–cholera toxin B subunit. The unrelated protein mannosidase 2 (Mann2) was unaffected on blots of lysates (Lys) for every background in a–c. (d) Endogenous PMGS enriched pTrkA at the tip of the future axon (Control stage 2 and stage 2+, arrowheads). PMGS protein accumulated along and at the tip of the early axon that was induced by PMGS overexpression. This favored the concentration of activated TrkA receptors in this neurite (OE PMGS, arrowheads). Under PMGS inhibition, activated TrkA receptors did not show a polarized distribution. Scale bar, 10 mm.

NGF alone caused a mild acceleration of polarization (Fig. 3b,d), yet this was not evident for BDNF or NT3 (Fig. 3a and Supplementary Fig. 5). Furthermore, PMGS inhibition precluded the NGF polarizing effect, which hinted at a particular association between NGF and PMGS. To test this further, we determined whether the acceleration of polarization that was induced by PMGS overexpression (Fig. 2) would be affected by the addition of NGF. In fact, this PMGS-induced polarization was significantly enhanced in the presence of NGF but not in the presence of BDNF or NT3 (Fig. 3a). The above results led us to hypothesize that PMGS may induce polarity by regulating the activity of TrkA, the receptor that mediates NGF signaling28. In cells cultured for 24 h in vitro, the amount of activated (phosphorylated) TrkA (pTrkA) was low. This amount increased slightly when neurons were stimulated with NGF (Fig. 4a).

NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION

In addition, overexpression of PMGS produced a major increase in pTrkA, which was further enhanced by NGF (Fig. 4a). This was consistent with the observed potentiation of PMGS phenotypic effects by NGF (Fig. 3a). Because PMGS polarizes to one neurite, we hypothesized that modulation of TrkA activity depends on its interaction with PMGS. In fact, we showed that PMGS interacts with TrkA, especially after NGF stimulation, but it did not interact with TrkB and TrkC (Supplementary Fig. 5), which is consistent with the phenotypic data shown above regarding BDNF and NT3 (Fig. 3a and Supplementary Fig. 7). This indicates that PMGS may preferentially interact with pTrkA. Indeed, pTrkA coprecipitated with hemagglutinin-PMGS, particularly in the presence of NGF (Fig. 4b). Conversely, pTrkA was not coprecipitated when extracts were hydrolyzed with alkaline phosphatase, indicating

5

Control

OE PMGS

OE PMGS + NeuAc2en HA Membrane RhoA Mann2 total RhoA

**

100

100

* * HA

0

*

0

OE PMGS + NeuAc2en

Phase

HA

*

OE PMGS OE PMGS + RhoA-L63 alone

myc-L63

Stage 2

+ Saponin

PMGS

PMGS

RhoA

10

RhoA

PMGS

RhoA

Ratio fl. axon*/ fl. dendrite* (log x)

RhoA

Stage 2+

PMGS

OE PMGS

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

– Saponin

OE PMGS+RhoA-L63

Phase

OE PMGS

c

OE PMGS alone

b

Cells with neurites >40 µm (percentage of the total)

a

Membrane RhoA (percentage of control)

ARTICLES

– –

+ +

– –

+ +

PMGS

+

1

0.1

Stage 2

Stage 2+ PMGS

RhoA

– +

1

0.01

PMGS



OE PMGS RhoA

RhoA

Figure 5 PMGS inactivates the small GTPase RhoA. (a) Western blots of neuronal membrane fractions (24 h). Buffer-treated cells (Control) showed detectable amounts of membrane-associated (active) RhoA. RhoA dissociated from the membrane under hemagglutinin-PMGS overexpression (OE PMGS) but remained membrane bound if PMGS activity was inhibited in this background (+NeuAc2en, 18 h). The total content of RhoA (total RhoA) and of the unrelated protein mannosidase 2 (Mann2) remained unaffected in every condition (n ¼ 3; *P o 0.001, compared with Control; **P o 0.001, compared with OE PMGS). (b) PMGS-triggered polarity at 24 h (OE PMGS alone) was precluded if constitutively active RhoA-L63 was cotransfected (OE PMGS+RhoA-L63, unpolarized cell in panels and graph). Changes in RhoA activity did not alter the polarized distribution of PMGS to one neurite (HA panels), although the constitutively active RhoA was evenly distributed to all neurites (Supplementary Fig. 4). *P o 0.001, compared with OE PMGS alone; n ¼ 75 cells/condition. (c) Neurons had less membrane-associated RhoA in the future axon after cytosolic extraction (+ Saponin) as compared with control neurons (– Saponin). Quantification of this effect in the population is summarized by determination of fluorescence ratios of both PMGS and RhoA, between PMGS-rich (axons*) and PMGS-poor (dendrites*) neurites. Data are mean + s.d. Scale bar, 10 mm.

that PMGS interacted with the phosphorylated form of TrkA (Fig. 4b). Together with the morphological results (Fig. 3), these biochemical data indicate that PMGS interacts and modulates TrkA activity. The fact that GM1 interacts and facilitates the activation of TrkA29 prompted us to test in our experimental model if PMGS activity might modulate TrkA by means of this ganglioside. In fact, phosphorylated TrkA was precipitated with GM1 when PMGS was overexpressed (Figs. 3 and 4c). These results confirmed that PMGS could modulate TrkA activity. Considering that PMGS creates a polarity landmark in differentiating neurons, the result of the above situation would be the localized accumulation of TrkA activity in the same growth cone in which PMGS accumulates. In fact, endogenous pTrkA, although detected throughout the cell, was enriched at the tip of the neurite that accumulated endogenous PMGS (Fig. 4d and Supplementary Fig. 8). TrkB and TrkC were not enriched at the tip of the future axon (Supplementary

6

Fig. 9), in concurrence with the previous findings (Fig. 3a and Supplementary Fig. 5). Furthermore, endogenous pTrkA preferentially accumulated in the prematurely specified axon of PMGS-overexpressing cells, especially at the tip, where most of the PMGS protein was detected (Fig. 4d). In agreement with the above data, a reduction of PMGS activity, although not affecting the local enrichment of PMGS in one neurite, did induce a loss of spatially restricted TrkA activity (Fig. 4d). These findings indicate that local PMGS activity induces a robust concentration of TrkA signaling in the future axon. PMGS targets the actin regulatory protein RhoA How can the polarized activity of PMGS induce spatially restricted changes to modify the underlying actin cytoskeleton? NGF-stimulated TrkA is known to inactivate RhoA30, a local regulator of actin dynamics in young hippocampal neurons31. Because active RhoA is membrane bound whereas inactive RhoA is cytosolic32,33, we tested whether the

ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

ARTICLES

a

OE PMGS alone

OE PMGS + wortman.

c Control

OE PMGS

OE PMGS + Rac N17 HA

RhoA Mann2

HA

HA

b

d Control

OE PMGS + Rac N17

Control OE PMGS OE PMGS + wortman. + wortman.

RhoA Mann2 HA

*

Flag-N17

100

** 100

Cells with neurites >40 µm (percentage of total)

200

Membrane RhoA (percentage of control)

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

HA

50

* 0

0

Control OE + wortman. PMGS

OE PMGS + wortman.

OE PMGS + Rac N17

activity of PMGS influences the association of RhoA with the membrane. Significantly less RhoA was bound to the membrane in neurons transfected with hemagglutinin-PMGS, as compared with control cells (Fig. 5a). Notably, RhoA dissociation from the membrane was prevented when overexpressing neurons were grown in the presence of the PMGS activity inhibitor (Fig. 5a). In agreement with this, the total amount of RhoA-GTP in whole cells and in purified growth cones was decreased when PMGS was activated and was increased when PMGS was specifically inhibited (data not shown). Furthermore, the accelerated axon specification that was produced by overexpression of PMGS alone was not observed in neurons that were cotransfected with the constitutively active Myc-RhoA L63 (Fig. 5b), highlighting the functional relevance of our findings. In addition, the severe arrest of axon growth that was caused by constitutively active RhoA31,34 (Supplementary Figs. 8 and 9) was not reversed by the increased activity of PMGS, confirming that RhoA was a downstream effector of PMGS. In fact, the polarized distribution of PMGS to one neurite was not impaired by constitutively active RhoA expression (Supplementary Fig. 7). To test if local PMGS activity correlated with a spatially restricted modification of RhoA activity, we removed most of the cytosolic material (including the cytosolic RhoA) to unveil the membranebound pool of RhoA. RhoA was reduced at the membrane of the neurite tip of the future axon of control stage 2 and 2+ neurons and at the axon of cells overexpressing hemagglutinin-PMGS (Fig. 5c). From these results, we concluded that PMGS activity induces inactivation of RhoA in one neurite to trigger axon formation. PMGS inactivates RhoA through PI3K and Rac1 In immortalized cell lines TrkA-dependent inactivation of RhoA requires PI3K and its downstream effector Rac1 (ref. 30). We therefore tested if this pathway links PMGS activity and the local modifications of RhoA. PI3K activity reduction with wortmannin35 inhibited the polarization of cells that overexpressed PMGS (Fig. 6a), although it did

NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION

OE PMGS alone

Merge

Figure 6 PI3K and Rac1 couple TrkA and RhoA in PMGS-induced signaling. (a) PI3K inhibition (OE PMGS + wortman.) hindered PMGS-induced early polarity (OE PMGS alone). PI3K inactivity did not change the enrichment of PMGS in one of the early neurite tips (see Supplementary Fig. 7). (b) Western blots of neuronal membrane fractions (24 h) show a pool of active RhoA in buffertreated controls (see Fig. 5a). PI3K inhibition precludes RhoA inactivation in such control cells (Control + wortman.) and in cells that overexpressed hemagglutinin-PMGS (OE PMGS + wortman.) and that otherwise had reduced levels of active RhoA (OE PMGS; see Fig. 5a). Data are mean + s.d.; n ¼ 3; dashed line (100%) is membrane-bound RhoA in control; *P o 0.001, compared with control; **P o 0.001, compared with OE PMGS. (c) Rac1 inactivity (dominantnegative Flag–Rac N17 cotransfection; OE PMGS + Rac N17) precluded RhoA inactivation that was induced by PMGS activity (OE PMGS). (d) In fact, Rac N17 expression (Flag-positive cell; gray bar) precluded PMGS-induced early polarity at 24 h (black bar) Data are mean + s.d.; n ¼ 4; *P o 0.001. Alterations of Rac1 activity did not change polarized localization of PMGS (see Supplementary Figs. 7 and 8 for details). The unrelated protein mannosidase 2 (Mann2) was unaffected in every background on blots. Scale bar, 10 mm.

not affect the accumulation of PMGS in one neurite (Supplementary Fig. 7). Although a considerable amount of RhoA was membrane bound in control untransfected cells, the presence of wortmannin increased membrane-bound RhoA (Fig. 6b). Conversely, when PMGS was overexpressed, membrane-bound RhoA was significantly lower than that of untransfected cells (Figs. 5a and 6b), although if wortmannin was present, membrane-bound RhoA was no longer reduced (Fig. 6b). The involvement of PI3K in PMGS-dependent RhoA inactivation was further tested by considering the activity of the PI3K downstream effector Rac1 (ref. 30). When a dominant-negative form of Rac1 (Rac N17) was expressed in neurons with enhanced PMGS activity, most RhoA remained associated with the membrane, in contrast to what was observed in cells that overexpressed PMGS alone (Fig. 6c). The induction of axon formation by PMGS overexpression was also inhibited in cells that coexpressed the Rac1 mutant (Fig. 6d and Supplementary Figs. 8 and 9). PI3K and Rac1 are enriched in the forming axon of differentiating hippocampal neurons5,36. Notably, the localized enrichment of PMGS was not affected by general loss of Rac1 activity (Supplementary Fig. 7), and membrane-bound Rac1 was enriched in the neurite tip that accumulated more PMGS (data not shown). These findings strongly indicate that localized activity of PMGS within one neurite can modulate the activities of PI3K and Rac1 (and thus of RhoA) to specify its axonal fate. The ROCK/profilin IIa complex is a target of PMGS Because the localized activity of PMGS affects actin filament stability, we determined how the PI3K/Rac1/RhoA system impinged on actin. A complex that links RhoA and actin, formed by the RhoA-specific kinase ROCK and the brain-specific actin-binding protein profilin IIa (PIIa) is involved in regulating actin dynamics during early neuronal differentiation31. Thus, the ROCK/PIIa complex might be an important link between PMGS and actin. We determined whether PMGS activity

7

– –

+ –

– +

+ +

ROCK

IP

PIIa

IP

1

+

3



Control

PIIa

NeuAc2en

PIIa

D P

*

2.4

* 1.0 Control

4

Phase

HA

AP

NeuAc2en

50

0

* OE PMGS OE PMGS + OE ROCK alone

HA

Phase

HA

GFP

50

* 0 OE PMGS OE PMGS + OE PIIa alone

Figure 7 PMGS activity modulates the ROCK/profilin IIa actin-regulating complex. (a) Western blot of neuronal extract immunoprecipitations with anti-RhoA. Overexpression of hemagglutinin-PMGS reduced the interaction between RhoA and its target ROCK/PIIa complex (compare lanes 1 and 3). RhoA activation (GTPgS treatment) precluded this effect (compare lane 3 and 4 and see lane 2 for control). (b) Myc-ROCK overexpression (Myc- and hemagglutinin-positive cells, bottom panels) precluded PMGS-induced axon formation at 24 h (hemagglutinin-positive cell, upper panels). (c) GFPPIIa overexpression (GFP- and HA-positive cell, bottom panels) prevented early axon specification that was induced by PMGS at 24 h (hemagglutinin-positive cell, upper panels). (d) Quantification of the effects described in b and c across neuronal populations. Data are mean + s.d.; n ¼ 4; *P o 0.001. (e) PMGS inhibition (NeuAc2en, lower panel, right bar) increased the active (phosphorylated, P) form of PIIa, as seen by two-dimensional western blots of neuronal extracts probed with anti-PIIa. The more negative spot (P) is identified as the more phosphorylated form (in comparison with the more dephosphorylated form, D) by hydrolysis with alkaline phosphatase (AP, middle bar). Data are mean ratios between P and D forms of PIIa + s.d.; n ¼ 3; *P o 0.001; scale bar, 10 mm.

affects the capacity of RhoA to recruit this complex. In steady-state conditions, RhoA interacted with ROCK and PIIa, and this interaction was favored by RhoA activity (Fig. 7a). We then observed that increased PMGS activity reduced the capacity of RhoA to recruit ROCK/PIIa, an effect that could be blocked by enhancing RhoA activity (Fig. 7a). To test this further, we increased the activity of ROCK and PIIa in PMGS overexpression backgrounds. Cotransfection of Myc-ROCK and hemagglutinin-PMGS blocked the PMGS-alone phenotype, precluding axon formation (Fig. 7b,d). Similarly, PIIa gain of function in a PMGS overexpression background averted axon formation, as only a few cotransfectants showed neurites longer than 40 mm (Fig. 7c,d). It is of note that the polarization arrest caused by PMGS inhibition (Fig. 2b,d,e and Supplementary Fig. 5) could be reverted by inactivation of the ROCK/PIIa complex (Supplementary Fig. 7). These results indicate that the RhoA-directed ROCK/PIIa complex is a downstream target of PMGS activity. This complex favors actin polymerization when ROCK is activated to phosphorylate PIIa31. Notably, PMGS inhibition favored PIIa phosphorylation (Fig. 7e), indicating that PMGS activity was required to inactivate the ROCK/PIIa complex (by inactivation of RhoA) to induce actin depolymerization. Although other RhoA-dependent processes might well be involved, these results establish a link between the activity of PMGS and the local regulation of actin dynamics. DISCUSSION Many molecules have been implicated in the establishment of neuronal polarity. The main evidence for such a role is their accumulation in the already-specified axon and their capacity to induce multiple axons on overexpression4. Although this indicates that such molecules support the crucial process of axon elongation, it remains unknown how they induce polarized growth at endogenous amounts. A likely scenario in

8

myc

100

100

Phase

PIIa

AP

Ratio P (PIIa)/D (PIIa)

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

e

2

HA

OE PMGS + OE PIIa

Lys

OE PMGS + OE ROCK

Phase Mann2

d OE PMGS alone

GTPγS HA-PMGS

c

Cells with neurites >40 µm (percentage of total)

b OE PMGS alone

a

Cells with neurites >40 µm (percentage of total)

ARTICLES

neurons might be similar to that which operates in other cell types, where the localization of the polarity signal arises from a polarity landmark in the membrane8–10. We show here that the transmembrane enzyme PMGS is segregated to a single neurite before morphological polarization and specifies axon formation. We demonstrate that increased PMGS accelerates the specification of the axon, whereas the contrary occurs when the protein production is precluded by iRNA or when the enzymatic activity of PMGS is specifically blocked. Furthermore, the effect of localized application of NeuAc2en is similar to that of bath application and, in both cases, neurites without PMGS enrichment are not affected. This strongly suggests that the role of PMGS activity is specific to the one neurite where it accumulates, which ultimately becomes the axon. We thus interpret these findings to indicate that PMGS acts as a polarity landmark. Neuronal polarization requires asymmetric destabilization of actin filaments at the unpolarized stage2. Our results indicate that PMGS can bring about such an event by spatially restricting TrkA-mediated signaling. This is important for the local inactivation of a RhoA signaling cascade by PI3K and Rac1, thereby establishing a link with the stability of the actin cytoskeleton (Supplementary Fig. 10). These results also explain how Par3/Par6/aPKC, by means of PI3K, can facilitate axon elongation5,37–39 by localizing to a single neurite. We thus propose that axon determination depends on the early polarization of membrane molecules that can, because of their intrinsic activities, locally recruit polarity-sustaining proteins that are otherwise spread throughout the cell (such as Par3/Par6/aPKC40 or PI3K, Rac1 and RhoA). The polarization of a given membrane protein (such as PMGS) would suffice to create a localized variation of signal strength in comparison with other neurites, which can then allow for the rapid, differential and polarized changes in actin dynamics that are required for the extension of one neurite and thus prompt neuronal polarity.

ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

ARTICLES Because another process can grow as the axon if the original one is sectioned41, it is probably the capacity of enriching the polarity signal over the threshold in the neurite that is important for determining its growth. In fact, PMGS overexpression leads to the regrowth of the sectioned axon21. This raises the issue of how the primordial enrichment of PMGS in one neurite occurs. It is still rather controversial whether early protein polarization depends on early molecular sorting or on a bulk membrane flow mechanism42. We favor the view that the transmembrane PMGS might be enriched by bulk flow toward the site with a higher rate of membrane addition43 rather than being delivered by active sorting. (PMGS overexpression would imaginably disrupt a molecular sorting machinery by saturation, but in fact it still results in polarized trafficking; Fig. 2 and Supplementary Fig. 1.) Vectorial movement of membranes might hypothetically favor the accumulation of PMGS in one neurite, allowing a threshold concentration to be reached faster and the process of axon formation to initiate in one tip preferentially. Indeed, in hippocampal neurons, a vectorial bulk membrane flow to the future axon has been described12. Whether the signal for polarized membrane flow is dependent on intrinsic or extrinsic cues remains an open question (Supplementary Fig. 9). In summary, our results highlight the importance of early membrane asymmetry as an upstream mechanism to trigger morphological asymmetry. It will be important in the future to determine whether other types of membrane enzymatic activities have similar roles to that shown here for PMGS and whether there are other membrane activities that favor actin polymerization and quiescence in the neurites that acquire dendritic fates. METHODS Hippocampal neuron culture. Primary hippocampal neurons from rat embryos were cultured as described17. For morphological studies, cells were plated onto poly-L-lysine-covered glass cover slips in minimal essential medium and N2 supplement (MEM-N2). For biochemical purposes, cells were cultured with MEM-N2 in 3-cm tissue culture dishes coated with poly-L-lysine. Where noted, NGF (50 ng/ml; Sigma), BDNF (50 ng/ml; Peprotech) or NT3 (50 ng/ml; Sigma) were added to cells after plating. As described31, ROCK was inhibited with Y-27632, and PIIa was reduced with specific antisense morpholino oligonucleotides. Membrane PMGS activity. To estimate PMGS activity, assay mixtures containing GM1 and GT1b (Matreya; 50 nmol of ganglioside-bound sialic acid), 0.2 mg BSA, 15 mmol sodium acetate (pH 4.6), 0.2 mg Triton X-100 and membrane samples in 0.2 ml were incubated for 1 h at 37 1C. Gangliosides were purified on C18 cartridges (Merck) and were separated on TLC plates (55:45:10 (v/v/v) CHCl3/methanol/0.2% CaCl2). Ganglioside bands were visualized with resorcinol-HCl reagent as described44 and were quantified by densitometry. Transient transfections. Expression plasmids coding mouse hemagglutininPMGS21, constitutively active RhoA (Myc-RhoA-L63; a gift from J. Settleman, Massachusetts General Hospital, Charlestown, MA), dominant-negative Rac1 (Flag-Rac N17; a gift from J. Hartwig, Harvard Medical School, Boston, MA), pPIIa-EGFP (PIIa-GFP31), pEF-Bos-Myc/ROCK (a gift from K. Kaibuchi, Nagoya University, Japan) were transfected into hippocampal neurons by electroporation (20 mg of DNA/300 ml of cell suspension) using a Genepulser (Bio-Rad) operated at 250 V and 250 mF. Cells were washed with PBS-CLAP (CLAP: chymotrypsin, leupeptin, antipain and pepstatin), scraped, pelleted and lysed by ultrasonication in cold PBS-CLAP. A postnuclear suspension was then centrifuged for 1 h at 100,000g to pellet the membrane fraction. Wortmannin (1 mM; Sigma) was used before membrane purification where noted. Western blotting. Extracts were loaded onto SDS-PAGE or two-dimensional acrylamide gels and transferred to nitrocellulose filters. These filters were then incubated with rat monoclonal antibody to hemagglutinin (3F10; Roche), mouse monoclonal antibody to RhoA (anti-RhoA; 26C4; Santa Cruz Biotechnology), mouse monoclonal antibody to mannosidase 2 (a gift from

NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION

G. Griffiths, European Molecular Biology Laboratory, Heidelberg, Germany), mouse monoclonal antibody to phospho-TrkA (E6; Santa Cruz Biotechnology), rabbit polyclonal antibody to TrkA (Chemicon), rabbit polyclonal antibody to TrkB (Chemicon), rat antibody to TrkC (US Biological), rabbit polyclonal antibody to PIIa45 (anti-PIIa) or mouse monoclonal antibody to ROCK (BD Transduction Laboratories). The filters were incubated with horseradish peroxidase (HRP)-labeled anti–rat or anti–mouse antibodies (Amersham) and were then developed using ECL (Amersham). Quantification was performed by densitometry of autoradiograms using NIH Image 1.62. Blots were normalized against the unrelated protein mannosidase 2 or were analyzed using total density as 100% for two-dimensional gels. Data were analyzed by the Student’s t-test. Immunoprecipitations. For immunoprecipitations, extracts that were recovered in cold lysis buffer (1% Triton X-100, 100 mM NaCl, 2 mM EDTA, 10 mM Tris-HCl, 1 mM Na3VO4, pH 7.5, and protease inhibitors) were incubated overnight at 4 1C with monoclonal antibody to hemagglutinin (12CA5, Roche), with anti-RhoA (26C4) or with a polyclonal antibody to HRP (ICN Biomedicals) that was preincubated with GM1 bound to HRP–cholera toxin B subunit (Sigma). Complexes were precipitated by centrifugation after incubation with protein A– and protein G–Sepharose beads, respectively. Beads were washed with cold lysis buffer, with a high-salt buffer (same as lysis buffer but with 500 mM NaCl and no Triton X-100) and then with low-salt buffer (same as immunoprecipitation buffer (10 mM Tris-HCl; 2 mM EDTA; 1 mM Na3VO4, pH 7.5) but without NaCl and Triton X-100). Before immunoprecipitation, extracts were treated with alkaline phosphatase (Roche), NGF (Sigma) or GTPgS (Pierce Chemical) where indicated. The immunoprecipitated complexes were separated by SDS-PAGE and were subjected to western blotting. Morphological analysis. For immunofluorescence detection, the following antibodies, together with most of those used for western blotting, were used: mouse monoclonal antibody to a-tubulin (Amersham), mouse monoclonal antibody to tau (Tau-1; Roche), mouse monoclonal antibody to Myc (9E10; Developmental Studies Hybridoma Bank), mouse monoclonal antibody to Flag (M2; Sigma), rabbit polyclonal antibody to PMGS21 (anti-PMGS) and rabbit polyclonal antibody to Rac1 (Santa Cruz Biotechnology). F-actin was detected with Alexa 568–conjugated phalloidin (Molecular Probes), and Alexa-conjugated secondary antibodies (Molecular Probes) were also used. For detection of membrane-associated proteins, cytosolic extraction with saponin (Sigma) was carried out. In brief, 5- to 10-sec pulses of 0.001% saponin in MSB (60 mM PIPES, pH 7.0; 2 mM MgCl2; 10 mM EGTA) were applied to live cells that had been rinsed in warm HBSS. After further rinsing, cells were fixed for fluorescence analysis as described above. Images were captured under a DMIRE2 microscope (Leica), and the lengths of neurites were measured (two observers; at least three experiments) using the Qfluoro software (Leica). Fluorescence intensity of F-actin and PMGS was measured using NIH Image software by considering the individual background fluorescence and the total fluorescence of each fluorescence channel from each individual image. Tubulin staining was used to control for volume. Fluorescence plots (including scatter plots) were obtained by drawing a tangential line to the neurite tips of each cell in each fluorescence channel and then measuring the intensity of every pixel crossed by the pre-established tangential line. For ratio analysis, total fluorescence of the area of the neurite tip (marked based on fluorescence channels and phase contrast images) was measured in each fluorescence channel from each individual image by considering background and total fluorescence for each channel. The data were analyzed by Student’s t-tests (two-tailed distribution and two-sample unequal variance). In the case of fluorescence ratios, these values were used to derive ratios between the fluorescence of individual images. These values were plotted using logarithmic scales for clarity. PMGS synthesis inhibition by iRNA. RNA oligonucleotides and nontargeting small RNA (nt-iRNA; siCONTROL) were obtained from Dharmacon RNA Technologies, Inc. All iRNA candidates were evaluated and scored by the SIRNA algorithm according to rational design criteria described previously (http://www.mpibpc.gwdg.de/abteilungen). The top scored candidate was selected against the target sequence 5¢-GGATTAACCTAGGCATCTA-3¢ beginning at nucleotide 1280 of the mouse PMGS (Neu3) open reading frame

9

ARTICLES

© 2005 Nature Publishing Group http://www.nature.com/natureneuroscience

sequence. Neurons in suspension were transfected by electroporation with iRNA or nt-iRNA plus pEGFP-N1 cDNA (Clontech; 20 mg RNA or DNA/300 ml cell suspension) using a Genepulser operated at 300 V and 450 mF. At 48 h after plating, cells were fixed and immunolabeled for tau and PMGS. Two independent observers took pictures of random fields containing EGFP-expressing cells (at least 100 cells per condition and experiment; three experiments). Local perfusion. Individual growth cones were locally perfused based on a previous description2. Two micropipette holders where placed in close proximity to allow a flux between the ejection needle and the sucking needle. Needles were prepared with a needle puller (Sutter Instruments). A peristaltic pump (Pharmacia Amersham) simultaneously drove application and sucking of solutions at very low velocities to reduce cell disturbance. Solutions were prepared in HBSS with bromphenol blue (Sigma), which fluoresces in red and thus allows precise detection of flux shape. The controlled flux was positioned in close vicinity to one particular growth cone of neurons grown in gridded coverslips (Eppendorf). Growth cones were chosen based on their higher area and density (for stage 2) and also on their being at the tip of the larger neurite (for stage 2+). Local buffer or NeuAc2en fields were applied for 15 min. Images were taken at time 0 and then every 5 min. At 15 min, cells were processed for fluorescence analysis as described above. Growth cone areas were measured for every cell (NIH Image) and were analyzed using Microsoft Excel v. X (growth cone area percentage variations are in comparison with initial growth cone at time 0, considered as 100%). The validity of our choice of growth cone was confirmed in control cells by low F-actin content and higher PMGS accumulation. Images were prepared by overlaying the fluorescent image in red (flux detection) on the phase contrast images. The 15-min time point is illustrated by F-actin and PMGS stainings. Note: Supplementary information is available on the Nature Neuroscience website.

ACKNOWLEDGMENTS We thank B. Hellias and E. Cassin for technical assistance; M. Giustetto and H. Vara for help with the local perfusions. J.S.S. is supported by an FCT/PRAXIS XXI scholarship (Portuguese Ministry of Science and Technology). Part of this work is supported by EU Grant Apopis (FP6-2002–LIFESCIHEALTH). COMPETING INTERESTS STATEMENT The authors declare that they have no competing financial interests. Received 7 December 2004; accepted 29 March 2005 Published online at http://www.nature.com/natureneuroscience/

1. Craig, A.M. & Banker, G. Neuronal polarity. Annu. Rev. Neurosci. 17, 267–310 (1994). 2. Bradke, F. & Dotti, C.G. The role of local actin instability in axon formation. Science 283, 1931–1934 (1999). 3. Bradke, F. & Dotti, C.G. Changes in membrane trafficking and actin dynamics during axon formation in cultured hippocampal neurons. Microsc. Res. Tech. 48, 3–11 (2000). 4. Horton, A.C. & Ehlers, M.D. Neuronal polarity and trafficking. Neuron 40, 277–295 (2003). 5. Shi, S.H., Jan, L.Y. & Jan, Y.N. Hippocampal neuronal polarity specified by spatially localized mPar3/mPar6 and PI 3-kinase activity. Cell 112, 63–75 (2003). 6. Inagaki, N. et al. CRMP-2 induces axons in cultured hippocampal neurons. Nat. Neurosci. 4, 781–782 (2001). 7. Da Silva, J.S. & Dotti, C.G. Breaking the neuronal sphere: regulation of the actin cytoskeleton in neuritogenesis. Nat. Rev. Neurosci. 3, 694–704 (2002). 8. Nelson, W.J. Adaptation of core mechanisms to generate cell polarity. Nature 422, 766–774 (2003). 9. Yeaman, C., Grindstaff, K.K. & Nelson, W.J. New perspectives on mechanisms involved in generating epithelial cell polarity. Physiol. Rev. 79, 73–98 (1999). 10. Chant, J. & Herskowitz, I. Genetic control of bud site selection in yeast by a set of gene products that constitute a morphogenetic pathway. Cell 65, 1203–1212 (1991). 11. Nobes, C.D. & Hall, A. Rho GTPases control polarity, protrusion, and adhesion during cell movement. J. Cell Biol. 144, 1235–1244 (1999). 12. Bradke, F. & Dotti, C.G. Neuronal polarity: vectorial cytoplasmic flow precedes axon formation. Neuron 19, 1175–1186 (1997). 13. Hasegawa, T. et al. Molecular cloning of mouse ganglioside sialidase and its increased expression in Neuro2a cell differentiation. J. Biol. Chem. 275, 14778 (2000). 14. Kalka, D., von Reitzenstein, C., Kopitz, J. & Cantz, M. The plasma membrane ganglioside sialidase cofractionates with markers of lipid rafts. Biochem. Biophys. Res. Commun. 283, 989–993 (2001).

10

15. Miyagi, T. et al. Molecular cloning and characterization of a plasma membrane– associated sialidase specific for gangliosides. J. Biol. Chem. 274, 5004–5011 (1999). 16. Dotti, C.G., Sullivan, C.A. & Banker, G.A. The establishment of polarity by hippocampal neurons in culture. J. Neurosci. 8, 1454–1468 (1988). 17. Goslin, K. & Banker, G. Rat hippocampal neurons in low-density cultures. in Culturing Nerve Cells (eds. Banker, G. & Goslin, K.) (MIT Press, Cambridge, Massachusetts, USA, 1991). 18. Bartlett, W.P. & Banker, G.A. An electron microscopic study of the development of axons and dendrites by hippocampal neurons in culture. I. Cells which develop without intercellular contacts. J. Neurosci. 4, 1944–1953 (1984). 19. Bartlett, W.P. & Banker, G.A. An electron microscopic study of the development of axons and dendrites by hippocampal neurons in culture. II. Synaptic relationships. J. Neurosci. 4, 1954–1965 (1984). 20. Banker, G.A. & Cowan, W.M. Further observations on hippocampal neurons in dispersed cell culture. J. Comp. Neurol. 187, 469–493 (1979). 21. Rodriguez, J.A., Piddini, E., Hasegawa, T., Miyagi, T. & Dotti, C.G. Plasma membrane ganglioside sialidase regulates axonal growth and regeneration in hippocampal neurons in culture. J. Neurosci. 21, 8387–8395 (2001). 22. Kopitz, J., Muhl, C., Ehemann, V., Lehmann, C. & Cantz, M. Effects of cell surface ganglioside sialidase inhibition on growth control and differentiation of human neuroblastoma cells. Eur. J. Cell Biol. 73, 1–9 (1997). 23. Lin, C.H., Espreafico, E.M., Mooseker, M.S. & Forscher, P. Myosin drives retrograde F-actin flow in neuronal growth cones. Neuron 16, 769–782 (1996). 24. Kozma, R., Sarner, S., Ahmed, S. & Lim, L. Rho family GTPases and neuronal growth cone remodelling: relationship between increased complexity induced by Cdc42Hs, Rac1, and acetylcholine and collapse induced by RhoA and lysophosphatidic acid. Mol. Cell. Biol. 17, 1201–1211 (1997). 25. Baas, P.W. & Ahmad, F.J. Force generation by cytoskeletal motor proteins as a regulator of axonal elongation and retraction. Trends Cell Biol. 11, 244–249 (2001). 26. Zhang, X.F., Schaefer, A.W., Burnette, D.T., Schoonderwoert, V.T. & Forscher, P. Rhodependent contractile responses in the neuronal growth cone are independent of classical peripheral retrograde actin flow. Neuron 40, 931–944 (2003). 27. Duchemin, A.M., Neff, N.H. & Hadjiconstantinou, M. Induction of Trk phosphorylation in rat brain by GM1 ganglioside. Ann. NY Acad. Sci. 845, 406 (1998). 28. Kaplan, D.R., Hempstead, B.L., Martin-Zanca, D., Chao, M.V. & Parada, L.F. The trk proto-oncogene product: a signal transducing receptor for nerve growth factor. Science 252, 554–558 (1991). 29. Rabin, S.J. & Mocchetti, I. GM1 ganglioside activates the high-affinity nerve growth factor receptor trkA. J. Neurochem. 65, 347–354 (1995). 30. Nusser, N., Gosmanova, E., Zheng, Y. & Tigyi, G. Nerve growth factor signals through TrkA, phosphatidylinositol 3-kinase, and Rac1 to inactivate RhoA during the initiation of neuronal differentiation of PC12 cells. J. Biol. Chem. 277, 35840–35846 (2002). 31. Da Silva, J.S. et al. RhoA/ROCK regulation of neuritogenesis via profilin IIa–mediated control of actin stability. J. Cell Biol. 162, 1267–1279 (2003). 32. Kranenburg, O., Poland, M., Gebbink, M., Oomen, L. & Moolenaar, W.H. Dissociation of LPA-induced cytoskeletal contraction from stress fiber formation by differential localization of RhoA. J. Cell Sci. 110, 2417–2427 (1997). 33. Leung, T., Manser, E., Tan, L. & Lim, L. A novel serine/threonine kinase binding the Ras-related RhoA GTPase which translocates the kinase to peripheral membranes. J. Biol. Chem. 270, 29051–29054 (1995). 34. Bito, H. et al. A critical role for a Rho-associated kinase, p160ROCK, in determining axon outgrowth in mammalian CNS neurons. Neuron 26, 431–441 (2000). 35. Ui, M., Okada, T., Hazeki, K. & Hazeki, O. Wortmannin as a unique probe for an intracellular signalling protein, phosphoinositide 3-kinase. Trends Biochem. Sci. 20, 303–307 (1995). 36. Santos Da Silva, J., Schubert, V. & Dotti, C.G. RhoA, Rac1, and cdc42 intracellular distribution shift during hippocampal neuron development. Mol. Cell. Neurosci. 27, 1–7 (2004). 37. Billuart, P., Winter, C.G., Maresh, A., Zhao, X. & Luo, L. Regulating axon branch stability: the role of p190 RhoGAP in repressing a retraction signaling pathway. Cell 107, 195–207 (2001). 38. Luo, L. et al. Differential effects of the Rac GTPase on Purkinje cell axons and dendritic trunks and spines. Nature 379, 837–840 (1996). 39. Nakayama, A.Y., Harms, M.B. & Luo, L. Small GTPases Rac and Rho in the maintenance of dendritic spines and branches in hippocampal pyramidal neurons. J. Neurosci. 20, 5329–5338 (2000). 40. Nishimura, T. et al. Role of the PAR-3–KIF3 complex in the establishment of neuronal polarity. Nat. Cell Biol. 6, 328–334 (2004). 41. Dotti, C.G. & Banker, G.A. Experimentally induced alteration in the polarity of developing neurons. Nature 330, 254–256 (1987). 42. Ledesma, M.D. & Dotti, C.G. Membrane and cytoskeleton dynamics during axonal elongation and stabilization. Int. Rev. Cytol. 227, 183–219 (2003). 43. Pfenninger, K.H. et al. Regulation of membrane expansion at the nerve growth cone. J. Cell Sci. 116, 1209–1217 (2003). 44. Schnaar, R.L. & Needham, L.K. Thin-layer chromatography of glycosphingolipids. Methods Enzymol. 230, 371–389 (1994). 45. Witke, W., Sutherland, J.D., Sharpe, A., Arai, M. & Kwiatkowski, D.J. Profilin I is essential for cell survival and cell division in early mouse development. Proc. Natl Acad. Sci. USA 98, 3832–3836 (2001).

ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

Lihat lebih banyak...

Comentários

Copyright © 2017 DADOSPDF Inc.