Disregulated RhoGTPases and Actin Cytoskeleton Contribute to the Migration Defect in Lis1-Deficient Neurons

Deficient migration of Lis1+/-cerebellar granule neurons on glial or neuritic fibers

Because the major mode of neuron movement in cortical regions is along radially oriented glial fibers, we first examined Lis1-deficient neurons moving on glia (Fig. 1A-E). Glia-guided migration was confirmed by dual immunolabeling of cultures using anti-βIII-tubulin for neurons and anti-GFAP for glia (Fig. 1B). By time-lapse video microscopy, Lis1+/- cells exhibited a 50% reduction in average velocity compared with wild type (Fig. 1C). This was associated with a reduced maximal velocity and increased time spent at rest (Fig. 1D,E)

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Lis1+/- cerebellar granule neurons display motility defects in both glia- and neurite-guided migration assays. A, Wild-type neurons cover a longer distance when migrating on glia than Lis1+/- neurons, as shown by displacement of the cell body (arrow) over the period of 1 hr (video-enhanced phase-contrast microscopy, 30 min between planes). B, Glia-guided assay cultures stained with βIII-tubulin (neurons; red) and GFAP (glia; green) antibodies. C, Average speed of Lis1+/- neurons (n = 17) compared with wt neurons (n = 43). D, The maximal speed of neurons was decreased in Lis1+/- neurons, whereas the percentage of time at rest (E) was increased. F, In reaggregate cultures, neurons cover a longer distance when migrating on neurites than Lis1+/-, neurons as shown by displacement of the cell body (arrow) over the period of 1 hr (video-enhanced phase-contrast microscopy, 30 min between planes). G, Dual labeling with βIII-tubulin (neurons; red) and GFAP (glia; green) antibodies show reaggregate cultures are >95% neurons. Average speed (H), maximal velocity (I), and time spent at rest (J) are all negatively affected in Lis1+/- neurons on neurites (n = 12) compared with wt neurons (n = 21). ^*p < 0.01.

In certain locations, neuronal motility in brain also occurs along neuronal fibers (Wichterle et al., 1997; Komuro and Rakic, 1998). A reaggregate assay of granule cells on neurites enables observation of more neurons undergoing vectorial movement (Fig. 1F-J). Double immunostaining with anti-βIII-tubulin and anti-GFAP revealed that >95% of cells in the reaggregate cultures were neurons (Fig. 1G). Migration along neurite fascicles extended from the aggregates was recorded by high-resolution time-lapse video microscopy (Fig. 1F,H-J). As in classical studies of glia-guided neuronal movement, this migration was saltatory, with periods of movement interspersed with rest (Edmondson and Hatten, 1987). There was a nearly fourfold reduction in the average speed of migration of Lis1+/- neurons along neurite fascicles, compared with wt littermates (Fig. 1H). The deficient migration of Lis1+/- cells was characterized by both a reduced maximal velocity and prolonged periods at rest (Fig. 1I,J).

Having validated the migration defect by time-lapse video microscopy of granule cells on glia as well as neurites, additional experiments examined whether a migration defect could be detected in reaggregate cultures by a simpler assay involving assessment of single frames taken at 24 hr in culture. Figure 2 compares the distribution of granule cells that had migrated along the neurite fascicles that were extended from clusters of Lis1+/+ granule neurons (Fig. 2A), compared with similar sized clusters of Lis1+/- cells (Fig. 2B). By 24 hr, significantly fewer cells had migrated from the Lis1+/- reaggregates at all distances measured along the neurite fascicles, and an 88% reduction (p < 0.002) in the number of neurons was found at a distance of 150 μm from an aggregate cluster (Fig. 2C). These time-lapse and single-frame experiments demonstrated that the migration deficit could be detected by determining the distribution of cells in the reaggregate assay at 24 hr and that this measure was relevant to neuronal movement. Therefore, this bin assay, which was conducted more rapidly, provided a reasonable representation of neuronal motility.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Fewer Lis1+/- cerebellar granule neurons migrate out of reaggregate clusters, and axonal fascicles are shortened. After 24 hr on laminin, wt neurons (A) actively move out from the reaggregate clusters and start migrating along neurite fascicles. Far fewer cells migrated from clusters of Lis1+/- neurons (B). C, Cell numbers in every 50 μm radial bin are significantly reduced in Lis1+/- (n = 60 clusters) compared with wild type (n = 60). D, Axons extending from clusters of Lis1+/- neurons (n = 826 neurite fascicles) are significantly shorter compared with wild type (n = 847 neurite fascicles). Scale bar, 50 μm. ^*p < 0.01.

This reaggregate assay also demonstrated a defect in axon extension by Lis1+/- granule cells after 24 hr in culture (Fig. 2D). The mean fascicle length was 175 μm in Lis1+/- clusters, compared with 210 μm in wt cells, a difference of 15% (p < 0.0001). However, the number of neurite fascicles per cluster did not differ significantly.

Lis1+/-neurons display abnormalities of the actin cytoskeleton

The time-lapse video microscopy studies revealed altered leading processes and filopodia in Lis1-deficient neurons, suggesting abnormalities in the actin-based cytoskeleton. Therefore, we sought to determine whether alterations in F-actin could be associated with Lis1 haploinsufficiency. Cultured cerebellar granule neurons were immunostained with βIII-tubulin antibody, whereas F-actin was visualized with rhodamine-phalloidin. The cytoskeletal organization of isolated, migrating neurons is similar to those moving on glia (Rivas and Hatten, 1995). Therefore, measurements were made on neurons isolated on the coverslip, to permit reliable quantification of fluorescence. Granule neurons were selected for analysis on the basis of migratory morphology, as described by Rivas and Hatten (1995). The diameter of the soma had to be <12 μm with scant cytoplasm, and the length of the leading neurite had to be <40 μm, consistent with a leading process rather than a growth cone.

Lis1 has been implicated as a regulator of microtubules. Therefore, we first sought to evaluate the microtubule component of the cytoskeleton. Staining of neurons with βIII-tubulin antibody did not reveal obvious abnormalities in the organization of the microtubule cytoskeleton in Lis1+/- neurons (Fig. 3A,B). The distribution of Tuj1 staining in Lis1-deficient granule neurons appeared indistinguishable from wt cells, and the fluorescence intensity was preserved (wt, 41 cells: 6.6 × 10⁵ ± 0.6 × 10⁵; Lis1+/-, 33 cells: 6.2 × 10⁵ ± 0.6 × 10⁵).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

F-actin is reduced at the leading edge and increased in the soma of Lis1+/- neurons, whereas the number and length of filopodia are reduced. Cultured cerebellar granule neurons stained with rhodamine-phalloidin (red) and βIII-tubulin (green). A, Wild-type neurons have a characteristic leading edge (l.e.) containing several filopodia and a cell body (soma) with little F-actin. B, Lis1+/- neurons have a leading edge dramatically reduced in size, fewer filopodia, and an increased amount of F-actin in soma (yellow overlap of labels). Filopodia length (C) and number (E) are reduced in Lis1+/- cells (n = 33) compared with wt cells (n = 41). Whereas the total amount of F-actin labeling was unchanged (D), the amount of F-actin at the leading edge (F) was reduced in Lis1+/- cells. Scale bar, 10 μm. a.u., Arbitrary units. ^*p < 0.01.

Phalloidin-labeled F-actin was visibly reduced at the leading edge in Lis1+/- neurons (Fig. 3A,B). The amount of F-actin present in Lis1+/+ and Lis1+/- neurons was estimated by summation of the phalloidin labeling in optical sections recorded along the z-axis of the entire volume of either the reconstructed neuron or its leading process. Compared with wild type, F-actin in Lis1+/- neurons was significantly reduced in the leading process (34% of control; p < 0.001), whereas the total amount of phalloidin staining across the entire cell was unchanged (Fig. 3D,F). In Lis1+/- neurons, the decrease in phalloidin staining in the leading process was accompanied by an increase in fluorescence intensity in the cell body (note the increased yellow overlap of tubulin and actin in the Lis1-deficient cell body in Fig. 3 B), reduced length, and a reduced number of filopodia (Fig. 3C,E).

Lis1+/-cerebellum contains reduced levels of activated Rac1 and Cdc42 and increased level of activated RhoA

Additional experiments investigated the activities of Rho GTPases, which are known to regulate the actin cytoskeleton (Schmidt and Hall, 1998). Rac1 and Cdc42 have been shown to promote actin polymerization at the leading edge of motile cells, whereas RhoA antagonized this activity, promoting retraction of the leading edge and assembly of stress fibers (Schmitz et al., 2000). Therefore, the activities of these Rho GTPases were compared in Lis1+/- and Lis1+/+ neurons. As summarized in Figure 4, Lis1-deficient cerebellar tissue contained levels of activated Rac1 that were 43% (43.06 ± 3.67%) of wt controls and levels of Cdc42 that were 18% (17.52 ± 4.74%) of wt cells (Fig. 4A,B,D). Thus, Lis1 haploinsufficiency was associated with a large downregulation of the activated forms of Rac1 and Cdc42, as measured by pull-down assays with PAK1-PBD that specifically binds activated forms of Rac1 and Cdc42. Consistent with the role of RhoA as an antagonist of Rac1/Cdc42 activity, GTP-bound RhoA was elevated in Lis1+/- cerebella more than three times that found in wt neurons (345.4 ± 28.26%; Fig. 4C,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Reduced levels of active Rac1, Cdc42, and increased active RhoA in Lis1+/- cerebellum. A-C, Pull-down assays reveal decreased GTP-bound Rac1 (A) and Cdc42 (B), whereas GTP-bound RhoA is increased in brain lysates of Lis1+/- animals. Total levels of Rac1, Cdc42, and RhoA proteins are unchanged. D, Activity of Rho GTPases expressed as percentage of control level is reduced in Lis1+/- cells (n = 3 experiments; 12 animals per genotype). ^*p < 0.01.

Inhibitors of the RhoA effector kinase p160ROCK upregulate activated Rac1/Cdc42 and rescue defects in the actin cytoskeleton, migration, and axon elongation in vitro

The observation that active Rac1 and Cdc42 were severely reduced in Lis1+/- neurons prompted investigation of whether RhoA influences Rac1 and Cdc42 activity in granule neurons in a manner similar to other systems. The RhoA-associated kinase p160ROCK has been the best characterized downstream effector of RhoA (Bishop and Hall, 2000). Thus, pharmacological p160ROCK inhibitors HA1077 and Y27632 were used to examine whether RhoA activity regulated levels of activated Rac1/Cdc42 in these Lis1+/- neurons (Fig. 5). Treatment of granule cell cultures with either or both HA1077 andY27632 restored activity of Rac1 and Cdc42 in Lis1+/- neurons to wt levels and enhanced these activities in wt neurons (Fig. 5C,D). The response to the two drugs together was slightly more robust with less variability among experiments, and so the combination of compounds was used in the data presented. Whereas the levels of activated Rac1 and Cdc42 in wt cerebellar neurons were doubled by ROCK inhibitors, these activated GTPases in Lis1-deficient neurons quadrupled, with Rac1 rising above wt untreated levels, although not quite as high as wt neurons treated with ROCK inhibitors (Fig. 5C,D). To further validate the involvement of Rho A in this activity, separate cultures were treated with C3 transferase, a potent direct inhibitor of Rho (Williamson et al., 1990). Administration of C3 had similar biochemical effects to the p160ROCK inhibitors (Fig. 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

p160ROCK inhibitors or C3 restore normal levels of active Rac1 and Cdc42 in Lis1+/- neurons. A, B, Treatment of either wt or Lis1+/- neurons with either p160ROCK inhibitors (R.I.) or RhoA inhibitor (C3) upregulates the amount of GTP-bound Rac1 (A) and Cdc42 (B). Treatment of wt neurons with p160ROCK inhibitors increases levels of active Rac1 (C) and Cdc42 (D) twofold and restores wt levels of these GTPases in Lis1+/- neurons (n = 3 experiments; 4 cultures of each genotype per condition or 16 cultures per experiment). Treatment of cultures with C3 transferase, a direct inhibitor of RhoA, produces a similar upregulation of Rac1 and Cdc42 (C, D). ^*p < 0.01 compared with untreated wt control; #p < 0.01 compared with untreated Lis1+/- control.

We also examined whether pharmacological interference with RhoA signaling could restore the motility deficit observed in Lis1-deficient neurons (Fig. 6). Exposure of cultured Lis1+/- neurons to inhibitors of the RhoA effector kinase p160ROCK had no effect on the total F-actin content (Fig. 6F). However, the inhibitors restored F-actin at the leading edge to wt control levels, (Fig. 6H) and restored filopodia number and length to wt values for both control and drug-treated conditions (Fig. 6E,G). These results confirmed that the modulation of Rho GTPases can eliminate cytoskeletal abnormalities associated with Lis1 deficiency.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

p160ROCK inhibitors rescue defects in the actin cytoskeleton of Lis1+/- neurons. A-D, Cultured cerebellar granule neurons stained with rhodamine-phalloidin (red) and βIII-tubulin (green). A, Wild-type neurons have a characteristic leading edge (l.e.) containing several filopodia and a cell body (soma) with little F-actin. B, Application of p160ROCK inhibitors to wt cells increases F-actin content in the leading edges. C, Control Lis1+/- neuron has a leading edge dramatically reduced in size, fewer filopodia, and an increased amount of F-actin in soma. D, Lis1+/- neurons treated with p160ROCK inhibitors look more like wt cells. Filopodia length (E) and number (G) are reduced in Lis1+/- cells (n = 33) compared with wt cells (n = 41), and p160ROCK inhibitors restore these parameters to wt values (wt, n = 33; Lis1+/-, n = 33). Whereas the total amount of F-actin labeling was unchanged (F), the amount of F-actin at the leading edge (H) was reduced in Lis1+/- cells. Scale bar, 10 μm. a.u., Arbitrary units. ^*p < 0.01.

This biochemical effect of pharmacological inhibition of RhoA was functionally associated with restoration to normal of Lis1 +/- granule cell migration and of neurite elongation (Fig. 7). As shown in Figure 7E, compared with untreated wt neurons, the migration of Lis1+/- neurons along neurite fascicles was again significantly decreased at all distances from a cell cluster. In addition, the mean length of neurite fascicles extended from Lis1+/- cell aggregates was significantly shortened by 15% compared with Lis1+/+ neurons (Fig. 7F; p < 0.0001). Importantly, HA1077/Y27632 dramatically enhanced motility of Lis1+/- neurons along neurites, increasing the numbers of neurons at the farthest distances from the cluster to near wt levels (Fig. 7E, compare 7A,B and 7C,D). Interestingly, the number of either wt or Lis1+/- neurons found at distances of 0-50 μm from the cluster was little affected by the ROCK inhibitors. However, the number of Lis1-deficient neurons counted at 101-200 μm was greatly enhanced by the drug, nearing the treated wt control levels (Fig. 7E,G). Overall, suppression of Rho activity restored the distribution of treated Lis1+/- cells to most closely approximate the treated wt cells (Fig. 7G). These ROCK inhibitors also restored the axon length of the Lis1-deficient clusters to wt levels (Fig. 7F, compare 7A,B and 7C,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

p160ROCK inhibitors enhance motility of wt cerebellar granule neurons and restore motility of Lis1+/- neurons. A-D, Reaggregate clusters of cerebellar granule neurons cultured on laminin for 24 hr. Treatment of granule neurons with p160ROCK inhibitors resulted in the enhanced migration out of the wt (A vs C) and Lis1+/- (B vs D) clusters. Treated Lis1+/- clusters (D) appeared more like wt clusters (A) with respect to the number of migrating neurons and length of axonal fascicles. E, Number of cells moved out from the clusters expressed as a percentage of wt control was greatly enhanced by 160ROCK inhibitors and brought untreated Lis1+/- to wt values (n = 60 clusters were analyzed for each genotype and treatment). F, Length of axonal fascicles was increased on p160ROCK inhibitor treatment in both wt neurons (n = 847 and 1032 axonal fascicles in control and treated groups, respectively) and Lis1+/- neurons (n = 826 and 1056 axonal fascicles in control and treated groups, respectively). G, Distribution of cells across 50 μm bins radiating from the Lis1+/- clusters was largely restored with more cells populating distant bins. Scale bar, 50 μm. ^*p < 0.01.

Lis1 deficiency interferes with motility and small GTPase activities in fibroblasts as well

Lis1 is a ubiquitously expressed protein, although the effects of its loss are most prominent in brain development. We, therefore, asked whether haploinsufficiency of Lis1 in non-neuronal cells would produce a similar reduction in the activity of Rho GTPases and whether that would be reflected in a motility defect. As shown in Figure 8, Lis1+/- fibroblasts displayed a reduction of activated Rac1 to 46% (45.73 + 6.45%) and Cdc42 to 22% (22.13 + 0.33%) of wt levels (p < 0.005). Activated RhoA levels were increased in Lis1-deficient fibroblasts to 300% of control (291.36 + 26.05%; Fig. 8 D). Unlike the fourfold reduction in motility of Lis1+/- neurons, however, fibroblasts monitored by time-lapse video microscopy displayed only a 30% decrease in average speed (p < 0.0005), compared with wt cells (Fig. 8C). Thus, the biochemical disturbance of Rho-GTPase activity was similar in Lis1-deficient neurons and fibroblasts. Interestingly, however, the degree to which cell movement was impaired was far more dramatic for the neurite-guided neurons.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Like neurons, Lis1+/- fibroblasts exhibit motility defects and alterations in Rho GTPase activities. A, B, Wild-type fibroblasts (n = 68 cells; left) migrate faster on laminin than Lis1+/- fibroblasts (n = 37 cells; right), as shown by video-enhanced phase-contrast microscopy. C, Average speed of migration is reduced in Lis1+/- fibroblasts. D, Pull-down assay reveals decreased activity of Rac1 and Cdc42, whereas RhoA activity is increased (n = 3 experiments; three cultures of each genotype were analyzed per experiment). Scale bar, 10 μm. ^*p < 0.01.

Coordinate abnormalities in Lis1+/-cells: impaired motility, defects in actin cytoskeleton, and disregulation of Rho GTPases

Time-lapse video microscopy of cerebellar granule cell migration demonstrated that glia-guided migration of Lis1-deficient cells was impaired. It also confirmed that motility of Lis1+/- granule neurons along neurites is impaired (Bix and Clark, 1998; Hirotsune et al., 1998). In the present study, this was caused by both decreased maximum velocity of movement and increased time spent at rest, suggesting that Lis1-deficient neurons have problems with both initiating movement and in their cycling through the steps of motility.

Another efficient measure of motility was the distribution of granule cells on neurites extended from reaggregate clusters after 24 hr in culture (Hirotsune et al., 1998). The number of granule neurons at a distance of ≥100 μm from the aggregate was dramatically lower in Lis1+/- granule cells. This assay in reaggregate clusters was validated by time-lapse video microscopic observations, and this simpler assay facilitated the evaluation of pharmacological manipulation on granule neuron motility.

Previous studies revealed Lis1 interactions with microtubules and with the dynein motor protein complex (Morris et al., 1998b; Horesh et al., 1999; Sapir et al., 1999; Caspi et al., 2000; Feng et al., 2000b; Sasaki et al., 2000; Smith et al., 2000; Han et al., 2001). However, the regulation of actin polymerization is another key component of cell motility (for review, see Lauffenburger and Horwitz, 1996). The altered appearance of filopodia and elaboration of a leading edge of migration during video microscopy prompted our examination of the actin cytoskeleton in Lis1-deficient neurons, using phalloidin staining. Indeed, the distribution of F-actin within the leading edge of neurons lacking Lis1 was abnormally low, suggesting altered regulation of actin polymerization in particular cell compartments. Because the F-actin cytoskeleton is a primary downstream target of Rho GTPases (Aspenstrom, 1999), we examined the small GTPases in Lis1+/- cells for alterations.

The present study provides the first indication that Lis1 is required to maintain physiological activities of the Rho GTPases that are central regulators of actin polymerization. Pull-down assays used GST fusion proteins to specifically bind active GTP-bound Rho, Rac1, and Cdc42. Relative to wild type, activated Rac1 and Cdc42 levels were reduced by 57% and 82%, respectively, at the same time that GTP-bound RhoA was increased threefold in Lis1+/- cerebellar tissue, as well as in cultured neurons and fibroblasts. Thus, the disregulation of GTPases observed in Lis1-deficient cells was not an artifact of culture or migration substratum.

In Lis1+/- granule neurons, fluorescence labeling of F-actin in the leading margin was reduced whereas it was increased in the somitic compartment, so that total F-actin content was unchanged. This was consistent with the observed enhanced RhoA activity, because RhoA has been shown to antagonize Rac1/Cdc42-mediated actin polymerization (Grosheva et al., 2001). It is Rac1/Cdc42 that regulate actin polymerization at the leading edge of migrating cells (Nobes and Hall, 1995). Moreover, RhoA has been thought to promote the formation of contractile actin stress fibers, which may account for the enhanced F-actin labeling in granule cell somata.

Rac1 and Cdc42 promote the formation of the lamellipodia and filopodia, respectively (Nobes and Hall, 1995). The present analysis focused on filopodia, because they are easily recognized and reliably quantified. Consistent with the reductions in Cdc42 activity, Lis1+/- neurons contained 58% fewer filopodia, and average filopodial length was 30% shorter. Filopodia sample the environment of migrating neurons, and a 30% reduction in length translates into a 50% reduction in the potential sampling area for detection of guidance cues. This altered distribution of F-actin, the reduced filopodia number and length, support the significance of the disregulation of Rho GTPases caused by haploinsufficiency of Lis1.

Cross-talk between Rho GTPases

The particular regulatory relationships between Rho GTPases are cell type specific. In Swiss 3T3 fibroblasts, Rho GTPases have been placed in a hierarchical cascade in which Cdc42 activates Rac1, which activates Rho (Nobes and Hall, 1995). Activation of Rac1 by Cdc42 has also been shown in N1E-neuroblastoma cells and primary astrocytes (Kozma et al., 1997; Etienne-Manneville and Hall, 2001), and RhoA is activated by Rac1 in primary neurons (Li et al., 2002). In N1E-115 neuroblastoma, MCDK, NIH 3T3, and SV80 cells, Rac1 and Cdc42 inhibit RhoA activity, whereas in some cells this inhibition is mutual (Kozma et al., 1997; Leeuwen et al., 1997; Sander and Collard, 1999; Sander et al., 1999; Grosheva et al., 2001). RhoA also suppresses Rac1 activity in NGF-stimulated PC12 cells (Yamaguchi et al., 2001). Finally, in Xenopus optic tectal neurons, Cdc42 inhibits RhoA, and RhoA inhibits Rac1 (Li et al., 2002).

Here, the interdependence of Rho GTPases in cerebellar granule neurons was tested through inhibition of RhoA signaling. The direct inhibition of RhoA by C3 induced Rac1 and Cdc42 activities in wt neurons and approached wt activations in Lis1-deficient cells. Pharmacological inhibitors of p160ROCK induced twofold increases in GTP-bound Rac1 and Cdc42 in wild type and restored activity of Rac1 and Cdc42 to wt levels in Lis1+/- neurons (increases of 3.8-fold and 6-fold, respectively). Strikingly, restoration of Rac1 and Cdc42 activities by inhibition of RhoA signaling largely eliminated the defects in the actin cytoskeleton and neurite-guided neuronal migration. The relationship between overactivated RhoA and the downregulation of Rac1/Cdc42 was present in both Lis1+/- neurons and fibroblasts. Furthermore, the effects of inhibition of RhoA or p160ROCK established the antagonism of Rac1/Cdc42 by RhoA activity in granule neurons.

Lis1 and the modulation of small GTPases in axon elongation

Rho GTPases play an essential role in axon growth, growth cone navigation, and the organization of F-actin in growing neurites (Kuhn et al., 2000). Axon growth in Lis1+/- neurons was, therefore, examined by measuring the length of axonal fascicles of cerebellar granule neurons after 24 hr in culture. Although the foreshortening of fascicles was not dramatic at 15%, it was nevertheless significant (p < 0.0001) and completely reversible with the agents used here. This again implicated Lis1 in modulating the actin-based cytoskeleton, because manipulation of RhoA and Rac1/Cdc42 fully restored normal measures.

Previous work has indicated a crucial role for Rho kinase (p160ROCK) in axon outgrowth of cultured granule neurons (Bito et al., 2000). In those studies, inhibition of p160ROCK accelerated axonal initiation and increased the size and motility of growth cones without changing the rate of axon elongation. The present data are consistent and suggest that increased RhoA activity in Lis1-deficient neurons is responsible for delays in axonal initiation. However, whether other factors, such as increased frequency of retraction, affects neurite growth in Lis1+/- neurons remain to be determined (Billuart et al., 2001).

Possible role of Lis1 in the coordinated regulation of cytoskeletal components.

Based on the data presented here and other published studies, two models for Lis1 function may be proposed (Fig. 9). In the first scenario, Lis1 interacts directly with RhoA to downregulate its activity and signaling to Rac1 and Cdc42 (Bishop and Hall, 2000). This would be consistent with the observation that cells with decreased Lis1 display higher levels of activated RhoA. In addition to its impact on actin polymerization, RhoA could influence microtubule dynamics. This mechanism would agree with previous reports that RhoA regulates microtubules by influencing the initial polarization of microtubules (Wittmann and Waterman-Storer, 2001). Thus, Lis1 could simultaneously influence microtubule organization and actin. We did not find evidence of direct Lis1-RhoA binding in Western pull-down assays. However, this does not rule out the possibility of low-affinity binding interactions that are not detectable in this method. More likely, Lis1 could interact with one or more of the numerous modulators of Rho GTPases (Kjoller and Hall, 1999), including the guanine nucleotide exchange factors, the GDP dissociation inhibitors, and GTPase activating proteins. Identification of such interactions will require extensive biochemical investigation beyond the scope of the present study.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Possibilities for the interaction between Lis1 and Rho GTPases in migrating neurons.

An indirect influence of Lis1 on RhoA activity via Lis1 modulation of microtubules could also account for the present observations (Fig. 9). In this scenario, Lis1, which binds microtubules in vitro (Sapir et al., 1999), could act to modulate microtubule dynamics. In turn, microtubules have been shown to regulate RhoA activity, thereby influencing the actin cytoskeleton (Liu et al., 1998). In addition, microtubule growth has been suggested as a stimulus of Rac1 activity, thereby promoting protrusion of the leading edge of migrating cells via actin polymerization (Waterman-Storer et al., 1999).

Data presented here further enrich the emerging picture of the role of Lis1 in neuronal migration. The present studies indicate influences of Lis1 on the actin-based cytoskeleton, not just microtubule function, and suggest a role for Lis1 in the coordinate regulation of the actin and microtubule cytoskeletal components to promote cell motility. It remains to be determined whether these Lis1 influences on actin occur by direct signaling to Rho GTPases or are via microtubule-based signaling mechanisms.