Immunolocalization of Fascin, an Actin‐Bundling Protein and Glial Fibrillary Acidic Protein in Human Astrocytoma Cells

Abstract

Fascin is a 55‐kDa globular protein that functions to organize filamentous‐actin into parallel bundles. A role for fascin in cell migration has led to its study in many tumor types. In this report, we investigate fascin in astrocytomas. We show that fascin is expressed in astrocytes and in a panel of human astrocytoma cell lines. Immunofluorescence analysis demonstrates that fascin and the intermediate filament protein, glial fibrillary acidic protein (GFAP), are both expressed in the perinuclear region and within cytoplasmic processes of astrocytes and astrocytoma cells. Amino acid residues within the NH2 terminus of GFAP can undergo phosphorylation; these modifications regulate intermediate filament disassembly and occur during cytokinesis. We show that fascin and specific phosphorylated species of GFAP colocalize within dividing cells. Finally, we demonstrate that fascin co‐immunoprecipitates with GFAP and that immunocomplex formation is preferential for GFAP phosphorylated at serine residues 8 and 13. These data show that fascin and GFAP are immunolocalized regionally within cells and tumors of astrocytic origin and suggest that their binding may occur during dynamic reorganization of intermediate filaments.

INTRODUCTION

Astrocytomas are the most common tumors arising in the central nervous system and account for 65% of all primary brain tumors. The majority of astrocytomas are malignant neoplasms that are highly proliferative, invasive neoplasms that infiltrate diffusely into regions of normal brain. The invasion of astrocytoma cells into brain involves a complex series of molecular events including the elaboration of proteolytic enzymes that can degrade extracellular matrix (ECM) 14, 38, 40, 52, 59, the adhesion of astrocytoma cell surface receptors to the ECM or to other cell types 10, 17, 24, 50 and the activation of small cytoskeletal GTPases which serve as molecular motors to propel an astrocytoma cells along or through ECM components 18, 35, 56, 60. Certain cytoskeletal GTPases, such as the Rho‐GTPases, work in concert to reorganize the actin microfilament system to form lamellipodia and filopodia (2). These cellular protrusions are dynamic structures that can extend, remodel and retract in response to cell stimuli. The stability of these cellular protrusions is maintained by actin and a network of actin‐binding proteins. One such actin‐binding protein is fascin, a 55‐kDa globular protein which enables filamentous‐actin to aggregate side‐by‐side into bundles 6, 64.

Within cells, fascin colocalizes with actin in filopodia, stress fibers, microspikes and microvilli 1, 63, 65. Filopodia containing fascin and actin bundles actively extend and retract during cell movement 9, 63. Using antibodies to block the actin‐binding site of fascin, cell spreading and migration are prevented 2, 3, 33. We have shown that fascin downregulation by siRNA technology results in decreased cellular attachment to matrix and migration in glioma cell lines (26). Taken together, these data demonstrate a role for fascin in cell migration.

Interestingly, fascin expression is upregulated in a number of human carcinomas including pancreatic 36, 58, non‐small cell lung carcinoma 43, 44, ovarian 5, 8, colon (29), esophageal (22), skin (19), gastric (22) and breast 20, 21. Several of these studies have shown that fascin expression is associated with the most clinically aggressive tumors. In non‐small cell lung carcinoma, gastric, esophageal and breast carcinomas, fascin expression correlates with poor patient prognosis and/or decreased disease‐free survival 22, 23, 45, 68. Other tumor types where upregulation of fascin has been documented include lymphoid 12, 30, 48 and brain 46, 51. We have previously shown that high‐grade astrocytomas demonstrate increased fascin expression compared with low‐grade astrocytomas (46).

To further our understanding of a role for fascin in human astrocytomas, we examined how fascin interacts dynamically with another cytoskeletal protein in astrocytoma cell culture, glial fibrillary acidic protein (GFAP). We show that fascin and GFAP colocalize within the cytoskeleton of astrocytoma cells. The possible implications of this colocalization in terms of astrocytoma migration and invasion will be discussed.

MATERIALS AND METHODS

Cell lines

Permanent human astrocytoma cell lines, U251, U343 and U373, have been previously characterized. HeLa cells were purchased from Clontech (Mississauga, ON, Canada). All cell lines were grown in high‐glucose Dulbecco's minimal essential medium with 10% fetal bovine serum. Human neural progenitor cells were obtained from Cambrex (East Rutherford, NJ, USA). Cells were grown in neural progenitor media (Cambrex) and supplemented with 10% fetal bovine serum for 7 days to cause astroglial differentiation. For immunostaining studies, neurospheres were plated on 10‐mm glass coverslips that were precoated with PEI reagent (Cambrex).

Western blotting

Astrocytoma cells were lysed in TNTE buffer (50 mM Tris‐HCl pH 7.5, 150 mM NaCl, 0.5% Triton‐X 100, 1 mM EDTA). Lysates were incubated on ice for 15 minutes followed by centrifugation at 13000 rpm for 15 minutes. Protein concentration of cellular lysates was determined using the Bio‐Rad Protein Assay reagent (Bio‐Rad, Mississauga, ON, Canada) as per manufacturer's instructions and quantitated by spectrophotometry. For western blot analysis, 10 µg of total protein were separated on a 10% sodium dodecyl sulfate polyacrylamide gel. Proteins were electrophoretically transferred onto Immobilon‐P membrane (Millipore, Mississauga, ON, Canada) and blocked in phosphate‐buffered saline (PBS) containing 5% skim milk powder and 0.1% Tween‐20 for 1 h followed by western blotting. All antibodies used for western blot analysis were diluted in PBS containing 5% skim milk powder and 0.1% Tween‐20. A dilution of 1:1000 was used for all antibodies described: mouse anti‐GFAP (Pharmingen, Mississauga, ON, Canada), mouse anti‐fascin (DAKO, Mississauga, ON, Canada) and mouse anti‐actin (Sigma, Mississauga, ON, Canada). The secondary antibody used was goat anti‐mouse horseradish peroxidase at a 1:5000 dilution (Bio‐Rad). Antibody binding was assayed using an enhanced chemiluminescence detection system (Amersham, Baie d'Urfe, QC, Canada) as per manufacturer's specifications.

Fluorescence immunocytochemistry

U343 astrocytoma cells and differentiated human neurospheres were grown on 10‐mm glass coverslips to 50% confluency. Coverslips were washed in PBS, and cells were prepared for immunostaining by fixing in 4% paraformaldehyde for 30 minutes followed by permeabilization with 0.25% Triton‐X 100 for 5 minutes. Cells were blocked using 2% normal donkey serum (Sigma) for 30 minutes prior to incubation with primary antibodies. Rabbit anti‐GFAP (DAKO), mouse anti‐GFAP phospho serine‐8 (Ser8) (MBL, Cambridge, MA, USA), mouse anti‐GFAP phospho Ser13 (MBL), mouse anti‐GFAP phospho threonine‐7 (Thr7) (MBL) and mouse anti‐fascin (DAKO) antibodies were used for primary immunostaining. Counterstains were performed using goat anti‐rabbit Alexa fluor‐594, goat anti‐rabbit Alexa fluor‐488, goat anti‐mouse Alexa fluor‐488, goat anti‐mouse Cy3 and goat anti‐mouse Alexa fluor‐647 antibodies (Invitrogen, Mississauga, ON, Canada). Double‐immunostaining using primary antibodies from the same species was performed using the Zenon labeling method (Invitrogen) as per manufacturer's instructions. Non‐specific immunoglobulin G (IgG) used in immunocytochemistry control experiments included mouse IgG (Sigma) and rabbit IgG (Jackson ImmunoResearch, West Grove, PA, USA).

Spinning disc confocal immunofluorescence microscopy

Immunofluorescence images were captured using a Zeiss Axiovert 200 inverted microscope (Carl Zeiss Inc., Hamburg, Germany), equipped with a Hamamatsu Orca AG CCD camera (Hamamatsu City, Japan) and spinning disc confocal scan head (Quorum Technologies, Ontario, Canada). Image acquisition was performed using Velocity 3.0 software (Improvision Inc. Waltham, MA, USA). Final image compilation was performed using Photoshop CS2 (Adobe Systems Inc., San Jose, CA, USA).

Image analysis of cells in mitosis

Dividing U343 cells were identified and imaged from a random population of cells plated on coverslips and immunostained as described above. Images were captured on the system described above and subject to analysis as mentioned.

Co‐immunoprecipitations

Parental U343 cells were lysed in TNTE buffer (50 mM Tris‐HCl pH 7.5, 150 mM NaCl, 0.5% Triton‐X 100, 1 mM EDTA). Co‐immunoprecipitations were performed using the following antibodies: monoclonal anti‐fascin (DAKO), monoclonal anti‐GFAP (Pharmingen), monoclonal anti‐GFAP phospho Thr7 (MBL), monoclonal anti‐GFAP phospho Ser8 (MBL), monoclonal anti‐GFAP phospho Ser13, monoclonal anti‐actin (Sigma) and monoclonal anti‐B23 (Sigma). Non‐specific mouse IgG (Sigma) was also used in immunoprecipitation control experiments. For co‐immunoprecipitation experiments, 750 µg of total cell lysate was incubated with 0.5 µg of specific antibody or mouse IgG at 4°C for 2 h. Protein G‐sepharose resin (Sigma) was then added to the mixture and incubated for an additional hour to bind antibody–protein complexes. The resin was washed three times using TNTE wash buffer (containing 0.1% Triton‐X 100), and proteins were eluted from the resin by boiling in Laemmli sample buffer containing 4 mM dithiothreitol.

Protein eluates were separated on a 10% sodium dodecyl sulfate polyacrylamide gel, electrophoretically transferred onto Immobilon‐P membrane (Millipore) and western blotted using an anti‐fascin antibody (DAKO) or anti‐GFAP antibodies as described above. Antibody binding was assayed using an enhanced chemiluminescence detection system as described.

RESULTS

Fascin is expressed in human astrocytoma cell lines

Astrocytoma cell lines were assayed by western blot analyses to demonstrate fascin expression (Figure 1A). The expression of fascin is abundant in all cell lines examined. GFAP and actin expression were also detected in these cell lines (B and C, respectively). HeLa cell lysate was used as a positive control for fascin expression; as expected, GFAP is not expressed in HeLa cells (B, lane 1).

Figure 1.

Protein expression analysis of fascin, glial fibrillary acidic protein (GFAP) and actin. Whole cell lysates from astrocytoma cell lines including U251, U343 and U373 demonstrate fascin (A), GFAP (B) and actin (C) protein expression. While there is a single band detected using this anti‐GFAP antibody in U251 and U373 cell lysates, there are multiple bands in the U343 cell lysates (B, lane 3). As expected, HeLa cell lysates express both fascin (A, lane 1) and actin (C, lane 1) but not GFAP (B, lane 1). Molecular weight markers are indicated in kDa.

GFAP and fascin regionally co‐immunolocalize in U343 astrocytoma and neural progenitor cells

The U343 astrocytoma cell line was used in immunocytochemistry studies. Anti‐fascin immunofluorescence analysis revealed a punctate cytoplasmic staining pattern with increased positivity within cell processes and membrane protrusions (Figure 2A). GFAP appears as a meshwork of cytoplasmic filaments oriented around the nucleus and extending into large cytoplasmic processes (Figure 2B). By double‐immunofluorescence labeling, both GFAP and fascin are most apparent in the perinuclear region and within large cytoplasmic processes (excluding small membrane protrusions) (Figure 2C). A magnified view of a single cell process (A′–C′) reveals that fascin is expressed throughout the microspike (A′: arrow and C′) while GFAP is not (B′ and C′).

Figure 2.

Immunofluorescent colocalization of fascin and glial fibrillary acidic protein (GFAP) in U343 astrocytoma cells. Panels A–C show a single cell double‐stained for fascin and filamentous GFAP and the merged image. Diffuse fascin (red) staining in the cytoplasm and within cell processes is evident (A). B. Filamentous GFAP expression (green) is shown throughout the cytoplasm and within the perinuclear region of astrocytoma cell. C. A merged composite image of the cell shown in A and B demonstrates generalized areas of GFAP and fascin colocalization (yellow) which are prominent in the perinuclear region. Panels A′–C′ show magnification of the cell area denoted by the asterisk in panel A. Fascin and overlapping filamentous GFAP staining is shown; only fascin can be seen at the terminal end of the cell protrusions (A′: arrow), while GFAP staining is absent in the same region (B′). Fascin and GFAP expression overlap (yellow) is shown in the merged image (C′). Panels D–F show double‐staining with anti‐fascin and anti‐GFAP phospho serine‐8 (Ser8). Fascin and GFAP phospho Ser8 staining is cytoplasmic and within cell protrusions (D and E). Panels D′–F′ are the magnified cell protrusions corresponding to the asterisked area in D. Again, fascin staining appears to reach further in the microspikes (D′: arrow and F′) compared with GFAP phospho Ser8 (E′ and F′). Panels G–I show a cell double‐stained with antibodies to fascin and GFAP phospho Ser13. Panels G′–I′ are magnified areas corresponding to the asterisked region in G. As shown above, fascin is expressed throughout the cytoplasm (G) and cell processes (G′: arrow). GFAP phospho Ser13 is also expressed in the cytoplasm (H); however, staining is less evident in the terminal portions of the cell process (H′). A merged image displays areas of signal overlap (I and I′). All corresponding double‐staining images were captured at the same focal plane using spinning disc confocal microscopy.

The NH2 region of GFAP contains a number of residues that can undergo phosphorylation, including Thr7, Ser8 and Ser13. Phosphorylation at these sites mediates assembly and disassembly of GFAP filaments (reviewed in 55). To characterize the localization of phospho‐GFAP species and fascin, we performed fluorescence immunocytochemistry (Figure 2D–I). Using antibodies specific for GFAP phosphorylated at serine residues 8 (Figure 2E and E′) and 13 (Figure 2H and H′), we show that the staining pattern is punctuate and cytoplasmic. Of interest, we observe that phospho‐GFAP, as with filamentous GFAP, is absent in the microspikes (E′) and smaller cell protrusions (H′); as expected, the expression of fascin is positive throughout these structures (D′ and G′: arrows). Immunostaining control experiments were performed using non‐specific IgG from appropriate species and the corresponding fluorescent‐tagged secondary antibodies to rule out any background staining (Figure 3).

Figure 3.

Immunofluorescence staining experiment controls. A. Purified rabbit and mouse immunoglobulin G counterstained with anti‐rabbit Alexa 488 and anti‐mouse Cy3. B. Purified mouse immunoglobulin G counterstained with anti‐mouse Cy3 and anti‐mouse Alexa 647. Both panels demonstrate lack of non‐specific fluorescence labeling. Images were captured using the same spinning disc microscope settings as those used to capture other images shown (2, 4, 5).

To determine the expression pattern of fascin and GFAP in normal astrocytes, neural progenitor cells were serum‐treated, and GFAP immunopositivity was used as a marker for astrocytic differentiation (Figure 4). Fascin immunostaining in these cells appears in a punctate pattern throughout the cell but is concentrated in the perinuclear region and within membrane protrusions and microspikes (Figure 4A and B). Double‐immunofluorescence analysis shows that fascin and GFAP are both found predominantly in the perinuclear region of the differentiated neural progenitor cells (Figure 4A and C). Positive fascin and GFAP staining is evident along extended cytoplasmic processes (Figure 4B and C); however, the microspikes are stained exclusively for fascin (as indicated by arrows in Figure 4A) and exclude any GFAP staining.

Figure 4.

Localization of fascin and glial fibrillary acidic protein (GFAP) in human neural progenitor cells serum‐treated for 7 days; GFAP positivity was used as a marker for astrocytic differentiation. A. An overlay, double‐immunofluorescence image shows an astrocyte co‐stained for GFAP (red) and fascin (green). A yellow color in the cytoplasm and within regions of the cell processes (arrows) demonstrates areas where both GFAP and fascin expression are shared. A single process (boxed area) is magnified and discussed in B, C, and D. B. Fascin staining is observed in the cell process in addition to the smaller microspikes originating from the main process (as shown by the arrows). C. GFAP staining is shown along the length of the process. D. The overlay image shows colocalization of GFAP and fascin within the astrocytic process. All images were captured at the same focal plane.

Localization of fascin and GFAP in dividing U343 cells

Earlier studies have demonstrated that phosphorylated species of GFAP (including phospho Thr7, Ser13 and Ser34) localize at the contractile ring within the cleavage furrow of dividing cells 37, 41. We also find Thr7 and Ser13 at the cleavage furrow (Ser34 was not tested). As reported previously, Ser8 is not detected at the cleavage furrow; rather it is expressed throughout the cytoplasm at the G2M transition 31, 41. Also found at the actin‐rich cleavage furrow are a number of actin‐binding proteins including coronin, myosin II, profilin, anillin, actin‐depolymerizing factor/cofilin, cyclase‐associated protein, talin, radixin, cortexillin and myosin V 11, 47, 57. We examined the localization of fascin in dividing U343 cells and show a cytoplasmic staining pattern that is not limited to the cleavage furrow (Figure 5). Double‐immunostaining of fascin and phospho‐GFAP species demonstrates that fascin expression overlaps with GFAP phospho Ser8 (Figure 5B) and Ser13 (Figure 5C). The expression of GFAP phospho Thr7 and fascin does not appear to overlap in dividing cells (Figure 5A).

Figure 5.

Localization of fascin and glial fibrillary acidic protein (GFAP) in dividing U343 cells. A. Fascin and GFAP phospho threonine‐7 (Thr7). GFAP phospho Thr7 is expressed at the cleavage furrow of dividing cells, while fascin is expressed in the cytoplasm. Fascin expression is absent in the areas of GFAP phospho Thr7 (arrows). B. Expression of GFAP phospho serine‐8 (Ser8) is cytoplasmic and overlaps with fascin. C. GFAP phospho Ser13 expression is limited to the cleavage furrow in dividing cells; the expression overlaps with fascin. D. Filamentous GFAP is expressed throughout the cytoplasm and colocalizes with fascin. Corresponding images for panels A–D are all taken at the same focal plane. 4′,6‐diamidino‐2‐phenylindole nuclear staining is shown for all four cells.

Molecular interaction of fascin and GFAP

Using co‐immunoprecipitation followed by western blotting analyses on whole cell lysates extracted from parental U343 cells expressing endogenous GFAP and fascin, we show that fascin co‐immunoprecipitates with GFAP (Figure 6). Anti‐GFAP co‐immunoprecipitation followed by anti‐fascin western blotting confirms that these proteins exist in a complex (A, lane 3); conversely, anti‐fascin co‐immunoprecipitation followed by anti‐GFAP western blotting also demonstrates that fascin is in an immunocomplex with GFAP (B, lane 2). As a control experiment, anti‐actin co‐immunoprecipitation was performed and western blotted with anti‐fascin antibody (A, lane 5) and anti‐GFAP antibody (B, lane 5). As expected, actin co‐immunoprecipitates with fascin (A, lane 5) but not with GFAP (B, lane 5).

Figure 6.

Fascin co‐immunoprecipitates with glial fibrillary acidic protein (GFAP). Total cell lysates were prepared from U343 cells and subjected to co‐immunoprecipitation (IP) followed by western analyses as indicated. Panel A (lane 1) demonstrates the presence of fascin in the control lysate. Anti‐fascin IP followed by anti‐fascin western blotting confirms the presence of fascin in the immunocomplex (lane 2). Anti‐GFAP IP followed by anti‐fascin western blotting indicates that these proteins can be co‐immunoprecipitated (lane 3). Lanes 4–6 represent control IPs. Lane 4 is an IP using anti‐B23 (nucleolar protein); as expected, no presence of fascin is detected upon western blotting. Lane 5 shows anti‐actin IP producing a complex that contains fascin following western blotting. Lane 6 shows a control IP using mouse non‐specific immunoglobulin G (IgG) demonstrating the absence of non‐specific binding. Panel B shows the membrane from A, stripped and reprobed with anti‐GFAP antibody. We confirm the presence of GFAP in the control lysates (lane 1). Lane 2 shows that an anti‐fascin IP pulls down GFAP. Lane 3 confirms that GFAP can be detected by western blot following anti‐GFAP IP. As described above, lanes 4–6 are control IPs. We do not detect an immunocomplex containing GFAP and actin (lane 5).

Interaction of fascin with phospho‐GFAP species

As we have shown that fascin appears to colocalize with GFAP phosphorylated at serine residues 8 and 13 and co‐immunoprecipitate with GFAP, we sought to determine whether fascin can co‐immunoprecipitate with phosphorylated species of GFAP. Using phospho‐specific GFAP antibodies, we performed co‐immunoprecipitations followed by anti‐fascin western blotting on U343 cell lysates to determine binding ability (Figure 7). These studies demonstrate that fascin is found in a complex with GFAP when it is phosphorylated at either serine residue 8 or serine residue 13; however, fascin does not co‐immunoprecipitate with GFAP phospho Thr7. The latter is also supported by our data showing that fascin and GFAP phospho Thr7 do not colocalize. The membrane shown in Figure 7 was stripped and reprobed with indicated anti‐phospho‐GFAP antibodies and confirms presence of respective phospho‐GFAP species in the immunocomplexes (data not shown).

Figure 7.

Fascin co‐immunoprecipitates with certain phospho‐GFAP species. Co‐immunoprecipitation analyses demonstrating the interaction of fascin with various phosphorylated species of GFAP. Fascin co‐immunoprecipitates with GFAP phosphorylated at serine residues 8 and 13 (lane 5 and 8), however, not with GFAP phosphorylated at Thr7 (lane 2). Specificity of IPs is confirmed by the lack of signal in IgG control lanes (lanes 3, 6 and 9). GFAP, glial fibrillary acidic protein; Ig, immunoglobulin; IP, immunoprecipitation; Ser, serine; Thr, threonine.

DISCUSSION

In this study of fascin and GFAP, we have shown that these proteins share regional colocalization within the cytoplasm of astrocytoma cells by fluorescent immunocytochemistry analyses and that the proteins can be co‐immunoprecipitated. When different phosphorylated GFAP species were tested for their ability to co‐immunoprecipitate with fascin, we show that fascin is in an immunocomplex with GFAP phospho Ser8 and Ser13, but not Thr7; these data are supported by the corresponding fluorescent immunocytochemical data. Taken together these findings suggest that fascin and GFAP co‐immunoprecipitate and colocalize in neoplastic and non‐neoplastic cells of astrocytic origin.

We were interested in determining if fascin could bind GFAP because of the prominent structural role GFAP has in the cytoskeleton of astrocytes and astrocytoma cells 53, 54. GFAP is a 50‐kDa intermediate filament (IF) protein that provides structural support to the cytoskeleton of astrocytes 15, 16. In normal astrocytes, GFAP stabilizes the cytoskeleton and helps to maintain cell shape through its interactions with the nuclear and plasma membranes. These interactions are likely mediated by a host of putative intermediate filament‐associated proteins (IFAPs), although few of these have been well characterized. One such example of an IFAP which binds to GFAP is plectin 49, 61, a 300‐kDa protein first identified in the rat C6 astrocytoma cell line. Plectin interacts with actin via a NH2‐terminal actin‐binding domain and with GFAP via its COOH‐terminal (25). Plectin deficient mice show severe defects in muscle integrity (4). Other IFAPs have also been implicated in disease states; for example, Yang et al have described an essential IFAP that connects neurofilaments to actin (66). Mice deficient in this linker protein, termed BPAG1n, undergo sensory neurodegeneration and demonstrate a dystonia musculorum phenotype (34). The recent elucidation of the role of IFAPs in various diseases creates new opportunities to explore the functional importance of these elements in cell growth, differentiation and survival.

We have shown that fascin co‐immunoprecipitates specifically with phosphorylated forms of GFAP. In general, the phosphorylation of IFs can lead to filament reorganization, altered solubility, localization within specific cellular domains, binding with other proteins and mediation of tissue specific functions (reviewed in 32). Just as with other IFs, GFAP does not exist in one static form; rather there is a continuous and dynamic shift from an assembled (filamentous) state to a disassembled (soluble) state 13, 27, 28. Phosphorylation of key amino acids within the NH2 region of GFAP regulates the disassembly from a filamentous, insoluble form into a soluble form. Serine and threonine residues in the NH2‐terminal region of GFAP act as substrates for phosphorylation by a number of serine/threonine kinases including cdc2, protein kinase A, protein kinase C, CaM kinase II and Rho‐kinase 7, 28, 31, 39, 67.

The mechanisms regulating the activity of fascin binding to actin are not well understood. Fascin contains two actin‐binding domains permitting tight packing of filamentous‐actin. The binding of actin to fascin is regulated by phosphorylation at the NH2 terminus of fascin, a process which inhibits actin bundling. The major phosphorylation site, Ser30, has a protein kinase C consensus motif, (S/T)X(K/R), which is present in all fascins 42, 62.

The identification of the fascin–GFAP association, is of interest in that it forms a potential link between IFs and actin microfilaments within the cytoskeleton of astrocytoma cells; our observation that fascin complexes with phosphorylated forms of GFAP suggests that this interaction might take place during dynamic reorganization of IFs. Interestingly, phosphorylation of fascin inhibits its binding to actin and the process of actin bundling. It is tempting to speculate that phosphorylation of both fascin and GFAP is required to facilitate the interaction between these two molecules—the phosphorylation of fascin to release it from actin bundles and the phosphorylation of the NH2 terminus of GFAP to enable GFAP to disassemble into its soluble form. If the interaction between GFAP and fascin is assisted by phosphorylation, then presumably dephosphorylation of each by phosphatase activity could enable fascin to return to actin bundles and GFAP to return to its assembled form.

To begin to understand the nature of the fascin–GFAP interaction, we performed imunofluorescence analyses to study their localization in U343 cells. Neural progenitor cells differentiated into astroglia show that filamentous GFAP and fascin share the same general localization in the cytoplasm and in cell processes. Interestingly, in astrocytoma cells, the expression of filamentous GFAP and fascin regionally overlaps within the cytoplasm in a predominantly perinuclear distribution. However, filamentous GFAP expression is not apparent in the terminal ends of cell protrusions and microspikes. Similarly, the expression of phospho‐GFAP, just as with fascin, is diffusely distributed throughout the astrocytoma cytoplasm, and in cell protrusions, however, it is absent in the terminal ends of cell protrusions and microspikes. It is known that phosphorylation of serine and threonine residues at the amino terminus of GFAP occurs during cytokinesis 28, 67. Accordingly, we find strong expression of phospho‐GFAP species in astrocytoma cells undergoing cytokinesis. In these cells, there is overlap of GFAP and fascin expression. These data suggest that GFAP, in its phosphorylated form, and fascin are associated during the process of cytokinesis.

As previously mentioned, we and others have shown by immunohistochemistry that high‐grade astrocytomas express increased levels of fascin compared with low‐grade astrocytomas or normal brain. The overexpression of fascin in tumor cells has been proposed as a mechanism by which they can invade and infiltrate surrounding normal tissue. Our future studies are now directed toward increasing our understanding of how the binding of GFAP to fascin may modulate not only cytoskeletal reorganization but also the process of tumor invasion by astrocytoma cells.