A role for cytoplasmic dynein and LIS1 in directed cell movement

Cytoplasmic dynein is a molecular motor associated with diverse subcellular structures, such as membranous organelles and kinetochores, which it transports or pulls toward the minus end of microtubules. Two lines of recent evidence have implicated dynein in a new and potentially general role in directed cell movement. First, dynein interacts with the LIS1 protein (Faulkner et al., 2000; Sasaki et al., 2000; Smith et al., 2000) mutations in which cause lissencephaly, a disease resulting from incomplete migration of neural progenitor cells from the ventricular zone during early brain development (Reiner et al., 1993). A second line of evidence implicating dynein itself in cell movement has come from analysis of cytoskeleton reorganization in wounded monolayer cultures (Etienne-Manneville and Hall, 2001; Palazzo et al., 2001), a system which provides an excellent means for controlling the timing and direction of cell movement. An early step in wound healing involves the reorientation of the centrosome, the major microtubule organizing structure, to a position ahead of the nucleus and toward the leading cell edge (Gundersen and Bulinski, 1988). Microtubules located at the front of the cells also become preferentially stabilized, and the cells migrate to close the wound. Injection of antidynein antibody or overexpression of the dynactin subunit dynamitin each interfered with centrosome reorientation (Etienne-Manneville and Hall, 2001; Palazzo et al., 2001) without affecting microtubule stabilization or organization (Palazzo et al., 2001).

These data implicated dynein in the reorientation process. However, the mechanism by which dynein contributed to this behavior, and a possible direct role in cell translocation were not assessed. We now report that dynein, its associated regulatory complex dynactin, and LIS1 are enriched at the leading cell edge in wounded NIH3T3 fibroblast monolayers during MTOC reorientation and subsequent cell migration. Inhibition of dynein, dynactin, and LIS1 interfere not only with reoriention of the microtubule network, but also with persistent directed cell migration as well.

To monitor the behavior of cytoplasmic dynein during cell migration, we performed immunofluorescence microscopy using antibodies to dynein or its accessory proteins. We detected a striking enrichment of dynein at the leading edge of the cell monolayer as the cells migrated to close the wound (Fig. 1, a–c, arrows; Fig. 2 j; Fig. 3 b). Dynactin, which has been implicated in dynein targeting (Echeverri et al., 1996) and processivity (King and Schroer, 2000) was also enriched at these sites (Fig. 1, d–f, arrows), where it colocalized with dynein (Fig. 1, g–i, arrows). In many cells, dynein and dynactin were enriched at regions toward which microtubules were directed (Fig. 1, a–f). Both punctate and diffuse staining were observed. The latter pattern could be seen even in regions containing few or no microtubules (Fig. 1, j–l, arrowheads). It was often observed in regions of lamellipodial protrusion as judged by the presence of membrane ruffles visualized by phase-contrast microscopy (Fig. 2, a–c arrows; and Fig. S1 A). The staining did not overlap precisely with the ruffles. Furthermore, in many cells, dynein and dynactin were enriched relative to the membrane marker CD44 (Fig. 2 m–o; Fig. S1, B and C; Perschl et al., 1995), and leading edge staining was clearly observed by confocal (Fig. S1 D) and total internal reflection fluorescence microscopy (TIRF; see Fig. 3).

Punctate dynein and dynactin staining was also observed throughout the cell, but was enriched at the leading edge of cells in the recovering wound. Some of these immunoreactive spots were associated with the ends of microtubules (Fig. 1, j–l, arrows). This pattern, however, was morphologically distinct from the elongated regions of dynein and dynactin seen at the plus ends of growing microtubules in vertebrate cells (Vaughan et al., 1999). Furthermore, antibodies such as the polyclonal anti-IC used in the current paper fail to produce the elongated patterns, and serve as selective markers for the cortical dynein structures observed here.

Actin and the cortical protein IQGAP1 (not depicted) were also enriched at sites of dynein and dynactin concentration, though their detailed distributions were distinct from that of the motor protein complexes (Fig. 2, d–f). In the well-spread lamellipodia of chick embryo fibroblasts, the region of dynein and dynactin enrichment was within the zone where the actin-rich lamellipodium encounters microtubule ends (Fig. 2, p–r and not depicted). No apparent colocalization between dynein and the focal adhesion protein vinculin could be detected (Fig. 2, g–i). Of considerable interest, LIS1 exhibited virtually the same pattern as dynein and dynactin throughout the leading edge of wounded NIH3T3 cell monolayers (Fig. 2, j–l), as it does in the cell cortex of mitotic epithelial cells (Faulkner et al., 2000).

In NIH3T3 cells, reorientation of the microtubule network occurs within 1–2 h of recovery from wounding (Gundersen and Bulinski, 1988; Palazzo et al., 2001). Both dynein and dynactin were enriched at the leading edge after 20 min of recovery, though staining appeared to increase steadily for several hours afterward. Thus, dynein and dynactin were present early enough at the leading cell edge to mediate reorientation of the microtubule network though why they continued to accumulate subsequently was uncertain.

Leading edge dynein and dynactin staining were absent in serum-starved cells (Fig. 3 a), which exhibit neither reorientation of the microtubule network nor cell migration (Gundersen et al., 1994; Palazzo et al., 2001). Serum addition triggers orientation of the microtubule network (Palazzo et al., 2001) and restored leading edge dynein staining (Fig. 3 b, arrows).

Reorientation of the microtubule network can be induced without lamellipodial protrusion by use of lysophosphatidic acid (LPA; Palazzo et al., 2001). Surprisingly, leading edge staining was not clearly detected in LPA (Fig. 3 c). Similar results were obtained in the presence of serum plus cytochalasin D, which also allows for reorientation of the microtubule network without forward cell movement (Nagasaki et al., 1992; Palazzo et al., 2001). To determine whether lower levels of dynein and dynactin could be involved in the reorientation process, we used TIRF microscopy, which increases the detectability at the base of the cells due to the high signal to noise ratio achieved by this system. Staining was considerably more punctate than observed by epifluorescence. In the presence of serum, spots could be clearly observed enriched at the leading edge relative to other cell regions in close contact with the substratum (Fig. 3, d–o; Fig. S2, A and B), and many of them were associated with microtubules (Fig. 3, d–f, m, and n, arrows). Similar staining was observed after treatment with cytochalasin D (Fig. 3 o) or induction by LPA (Fig. 3, j–n), indicating that dynein is indeed present at the leading edge in conditions allowing for MTOC reorientation. In these cases, striking spots of dynein and dynactin could be observed at microtubule ends (Fig. 3, m and n, arrowheads). Leading edge enrichment was not clearly observed by TIRF microscopy in serum-starved cells (Fig. S2). We note that the number of spots and, therefore, the overall intensity of staining were higher at the leading edge of serum-stimulated cells, making the enrichment of dynein and dynactin at the leading edge more readily apparent in the presence of serum (Figs. 1 and 2 and Fig. 3 e).

The appearance of the bright, diffuse leading edge dynein and dynactin staining pattern in the presence of serum, which alone allows for forward cell migration during wound healing, suggested a potential novel role for the motor protein in this process. To test this possibility, we overexpressed a GFP tagged version of the dynamitin subunit of dynactin, which was shown to inhibit dynein function and block microtubule network orientation (Echeverri et al., 1996; Burkhardt et al., 1997; Palazzo et al., 2001; Dohner et al., 2002; though see Deacon et al., 2003). Overexpressing cells showed a marked decrease in motility (Fig. 4, A and D), and fell behind the wound edge during recovery (Fig. 4 C; Video 1). A similar effect was also produced by overexpressing cDNA encoding the NH₂-terminal portion of LIS1 (Fig. 4, C and D), which produced a pronounced mitotic phenotype without affecting the distribution of Golgi elements (Tai et al., 2002). We also injected cells with the well-characterized 70.1 mAb to the cytoplasmic dynein intermediate chains, which inhibits dynein function (Burkhardt et al., 1997; Faulkner et al., 2000; Yvon et al., 2001). Cells were injected with the antibody 2 h after wounding, sufficient time for centrosome orientation to have occurred (Palazzo et al., 2001). The cells were followed by time lapse microscopy and subsequently fixed for immunofluorescence. Again, migration was less efficient compared with controls (Fig. 4 C). In all three cases the rate of migration was clearly decreased (Fig. 4 D). Although ruffling activity and lamellipodial protrusion persisted in most cells, these activities were clearly abolished or reduced in 36% of dynein inhibited cells (Videos 1 and 2).

To test for other possible effects on cytoplasmic dynein function, the organization of the Golgi apparatus was examined 4 h after the 70.1 antibody was injected. The Golgi apparatus remained condensed in the perinuclear region in 89% of the injected cells (n = 55) versus 92% of noninjected controls (n = 48; Fig. S2 C), a result comparable to the insensitivity of the Golgi apparatus to this antibody during microtubule reorientation at short times of wound healing (Palazzo et al., 2001). However, the Golgi apparatus remained oriented toward the leading edge in most (76%; n = 55) of the injected cells, similar to control cells (84%; n = 51).

Our results reveal an enrichment of dynein, dynactin, and LIS1 at the leading cell edge during wound recovery in what appear to be two distinct subcellular pools, potentially involved in two distinct functions. The punctate staining observed along microtubules and at microtubule ends is observed during the early phase of the wound healing process and under conditions that support the reorientation of the microtubule network but do not stimulate migration. The punctate structures appear to represent sites of attachment between microtubules and cortical dynein, and may represent the loci at which dynein pulls on microtubules to reorient the microtubule network and associated organelles (Fig. 5). Cortical dynein has been detected in different systems (for review see Dujardin and Vallee, 2002). Dynein and dynactin have been identified at the cell cortex in dividing cells where they orient the mitotic spindle through its astral microtubules (Busson et al., 1998; Faulkner et al., 2000). How this dynein pool may be related to that reported in the current paper is uncertain. Dynein and dynactin have also have been implicated in the reorientation of the centrosome–nucleus complex in two-cell stage Caenorhabditis elegans embryos, where they may act either from the site of the midbody from the prior cell division or from the region of the cortex lying between the two cells (for review see Dujardin and Vallee, 2002). Dynein has been phenotypically implicated in maintaining centrosome position during interphase in amoeboid and nonmotile epithelial cells (Koonce et al., 1999; Burakov et al., 2003), a potentially related phenomenon, though the sites from which dynein may act in these cases were not determined. Spots of dynein or its regulatory proteins have also been seen at the tips of microtubules reaching the cortex during meiotic nuclear oscillations in Schizosaccharomyces pombe (Yamamoto et al., 1999), in Saccharomyces cerevisiae buds (Lee et al., 2003; Sheeman et al., 2003), and at the hyphal tip of filamentous fungi (Xiang et al., 1995; Minke et al., 1999). Dynein and dynactin have been localized to other cortical sites, such as adherens junctions (Ligon et al., 2001). These structures are absent from the leading edge of the migrating fibroblasts used in our paper, and, therefore, unlikely to be responsible for the currently observed leading edge localization. Furthermore, although microtubules interact with focal adhesions (Kaverina et al., 1998), this behavior was found to involve kinesin rather than dynein (Krylyshkina et al., 2002).

The bright, more diffuse staining we observe at the leading edge of migrating fibroblasts is detectable in the early stages of wound healing, but continues to accumulate during the subsequent phase of cell migration, and is abolished by conditions that interfere with this latter process. These observations support an additional novel role for dynein in cell translocation. This possibility was confirmed by use of dynamitin and LIS1 overexpression, as well as dynein antibody injection. Dynein and dynactin were enriched within regions of lamellipodial protrusion. Because dynein is only known to produce force in conjunction with microtubules, our observations supports the existence of a novel pool of dynein associated with the actin-rich cortical cytoskeleton which could be available for capture of microtubules entering this region.

Interference with cytoplasmic dynein, dynactin, or LIS1 resulted in decreased cell migration, revealing a persistent requirement for dynein and its regulatory proteins in forward migration. Limited microtubule depolymerization also inhibits forward cell movement in this assay (Liao et al., 1995), and there may well be a common physiological basis for these responses. To determine whether the migration defects produced by dynein inhibitors resulted from effects on the orientation of the microtubule cytoskeleton, we allowed wound healing to proceed for 2 h before injecting antidynein antibody. Forward cell migration was inhibited despite the completion of the microtubule reorientation phase of the wound healing process. No change in the orientation of the Golgi apparatus was observed. This observation indicates that an important component of the cell's biosynthetic machinery remained oriented toward the leading cell edge, and suggested that the same was true for the microtubule organizing center. Thus, inhibition of cell migration does not appear to involve disruption of the microtubule cytoskeleton and may involve a distinct and novel dynein function.

What precisely this function may be remains to be fully elucidated. Recent evidence has revealed a retrograde flux of actin filaments toward the cell center from the leading lamellipodium in migrating cells, which produces a backward force on microtubules and, presumably, their associated organelles (Salmon et al., 2002). Leading edge dynein might potentially serve as a holdfast for those microtubules which invade the actin network at the front of the cell. Consistent with this hypothesis, bidirectional, actomyosin-mediated movements of microtubules in PtK2 cells were enhanced by dynein inhibition (Yvon et al., 2001).

It is also possible that the link of leading edge microtubules to cortical cytoplasmic dynein serves as a means of communication between the microtubule and actin cytoskeleton. As envisaged for cortical dynein, kinetochore dynein interacts with the plus ends of microtubules. From this site, it both pulls on chromosomes and removes checkpoint proteins (Howell et al., 2001), regulating cell cycle progression. Conceivably, tension produced through the interaction of microtubules with cortical dynein could serve to regulate actin based motility, through a tension sensing or other signaling mechanism. The activation of lamellipodial protrusion by microtubule growth (Waterman-Storer et al., 1999), and the recent finding that LIS1 haploinsufficiency affects filamentous actin organization at the leading edge (Kholmanskikh et al., 2003) are potentially consistent with this hypothesis.

Type I lissencephaly, which is caused by mutations in the human LIS1 gene, is thought to involve a defect in the migration of differentiating neurons in the developing brain (for review see Morris, 2000). This possibility is supported by an effect of reduced LIS1 expression on nuclear movements within cerebellar granule reaggregate cultures (Hirotsune et al., 1998; Kholmanskikh et al., 2003), and free translocation of brain derived fibroblasts (Kholmanskikh et al., 2003). Based on evidence for the codistribution of LIS1 with dynein at kinetochores and the mitotic cell cortex, as well as a pronounced LIS1 mitotic phenotype, we have suggested a role for LIS1 in the timing and orientation of progenitor cell divisions (Vallee et al., 2000). The finding of LIS1 at the leading edge of the migrating cells in the current paper identifies a new site for dynein and LIS1 colocalization. Inhibition of cell migration by a dominant negative LIS1 fragment supports a dynein regulatory role this site. Further work will be directed at testing this possibility.

NIH3T3 fibroblasts were grown to confluence in the presence or absence of bovine calf serum (Palazzo et al., 2001) and treated with 0.25 μM cytochalasin D or 1 μM LPA as required (Palazzo et al., 2001). GFP-dynamitin and GFP-LIS1-N cDNAs (Palazzo et al., 2001; Dohner et al., 2002; Tai et al., 2002), or monoclonal antidynein intermediate chain antibody concentrated at 18–20 mg/ml (clone 70.1; Sigma-Aldrich) were injected, respectively, in the nucleus and the cytoplasm using a semi-automatic microinjector (Eppendorf) as described previously (Faulkner et al., 2000). Primary chick embryo fibroblasts were prepared from E10 embryos and grown in sparse culture in medium 199 (Sigma-Aldrich) supplemented with 5% FCS and 1% chicken serum on fibronectin-coated coverslips. Live cells were recorded at 37°C in a 5% CO₂ atmosphere using inverted microscopes (model DMIRBE; Leica) equipped with an incubation chamber. Images were collected with a CCD camera (model ORCA 100; Hamamatsu) piloted by Metamorph (Universal Imaging Corp.), or a CCD camera (model DC F350; Leica) piloted by the Leica FW4000 software. Confocal pictures were acquired on a confocal microscope (model MRC1024; Bio-Rad Laboratories). TIRF images were collected using a microscope (model Eclipse TE2000-u; Nikon) with a Turnkey TIRF module. Images were processed using Metamorph and Photoshop softwares.

Cells were fixed after wounding by immersion in −20°C methanol for 6 min, or 20 min in 3% PFA (Fig. 2, j–l and Fig. 3, g–i) followed by 1 h incubations in primary and secondary antibodies diluted in PBS containing 0.05% BSA. Antibodies included polyclonal antidynein IC (Vaughan et al., 1999), anti-CD44 (Perschl et al., 1995), anti-p150^GLUED, and anti-GM130 (Transduction Labs), antitubulin (DM1A), -actin (AC-40), and -vinculin (all from Sigma-Aldrich), monoclonal anti-LIS1 (provided by O. Reiner, Weizmann Institute, Rehovot, Israel), polyclonal anti-IQGAP1 (provided by G. Bloom, University of Virginia, Charlottesville, VA), polyclonal anti–Arp-1 (provided by D. Meyer, University of California, Los Angeles, Los Angeles, CA), Alexa green–conjugated secondary antibodies (Molecular Probes Inc.), and Cy3- and Cy5-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories).

Two videos illustrating NIH3T3 cell migration defects after dynein inhibition. Video 1 shows cell migration after overexpression of GFP-dynamitin, and Video 2 shows cell migration after injection with antidynein antibody. Two additional figures are also provided. Fig. S1 shows the distribution of dynein relative to membrane ruffles, as well as the quantification of dynein enrichment at the leading edge. Fig. S2 shows that dynein is enriched at the leading edge using TIRF microscopy, as well as the normal distribution of the Golgi apparatus after injection of antidynein antibodies.

We thank Drs. Orly Reiner, David Meyer, and George Bloom for antibody reagents; Yu Chen, Anna Kalinovsky, and Célia Clouet for technical contributions; and Dr. Jan De Mey for the use of microscopy equipment in the later stages of this project.

This work was supported by grants to R.B. Vallee (NIH GM47434 and HD61982, and a March of Dimes Birth Defects Foundation grant); to G.G. Gundersen (NIH GM62939); and to D.L. Dujardin (Human Frontiers Science Program Fellowship).