Actinin-4 is preferentially involved in circular ruffling and macropinocytosis in mouse macrophages: Analysis by fluorescence ratio imaging

Ruffling, macropinocytosis and phagocytosis are forms of cell motility mediated by filamentous actin (F-actin) polymerization and its reorganization (Swanson et al., 1999; Swanson and Watts, 1995). Such motility requires a variety of actin-binding proteins to sever, bundle and/or cross-link F-actin (Matsudaira, 1991; Otto, 1994; Schmidt and Hall, 1998; Stossel, 1993). α-Actinin is one actin-binding protein which can cross-link F-actin into F-actin bundles or networks and also connect F-actin to the plasma membrane (Djinovic-Carugo et al., 1999; Matsudaira, 1991; Otto, 1994). So far, four isoforms of human α-actinin have been identified and characterized at the gene level. Actinin-1, encoded by ACTN1, is a classical isoform of nonmuscle α-actinin which primarily localizes to focal adhesion plaques. It binds F-actin bundles (stress fibers) to the plasma membrane via other proteins such as talin, vinculin and integrin (Chen and Singer, 1982; Honda et al., 1998; Langanger et al., 1984; Lazaride and Burridge, 1975; Meigs and Wang, 1986). It was also reported that nonmuscle α-actinin localized in various F-actin-based structures including stress fibers, ruffling membranes, phagocytic cups, cleavage furrows, receptor capping sites and contractile vacuoles of Dictyostelium discoideum (Allen and Aderem, 1996; Bretscher and Lynch, 1985; Fujiwara et al., 1978; Furukawa and Fechheimer, 1994; Geiger and Singer, 1979; Meigs and Wang, 1986). However, it is not known if different isoforms of α-actinin have different subcellular localizations. Actinin-2 and -3, encoded by ACTN2 and ACTN3 respectively, are muscle α-actinin which cross-link F-actin together in the region of the Z-discs of striated muscle cells (Beggs et al., 1992).

Actinin-4 is a recently characterized isoform of nonmuscle α-actinin (Honda et al., 1998). Using the monoclonal antibody (mAb) HCC-Lu-632, which specifically recognized human actinin-4, it was shown that the localization of actinin-4 was different from that of actinin-1 in cancer cell lines. Unlike actinin-1, actinin-4 did not localize to focal adhesion plaques or adherens junctions; its cytoplasmic localization was closely associated with enhanced motility of cancer cells and might predict the metastatic potential of human cancer. Interestingly, this isoform was found to be translocated into the nucleus in several cancer cell lines upon inhibition of phosphoinositide 3-kinase (PI3-kinase) by wortmannin (Honda et al., 1998). These findings suggested that actinin-1 and actinin-4 might play distinct roles in cellular motility and stability (adhesion) and might be regulated by different signal transduction cascades. Expression of actinin-4 mRNA was detected in almost all human tissues, but its precise intracellular localization in cells other than cancer cells has not yet been examined. It is of particular interest to determine actinin-4 localization in highly motile cells such as macrophages, since actinin-4 seems to be implicated in cell motility rather than cell adhesion.

Macrophages show active membrane ruffling and macropinocytosis as well as phagocytosis. The ruffling can be divided into at least two types: cell edge ruffling and dorsal surface ruffling (Ridley, 1994; Swanson and Watts, 1995). Most ruffles are continuously initiated at the cell edge. The cell edge lamellipodial ruffling seems to be associated with cell spreading, locomotion and chemotaxis. Some edge ruffles undergo a smooth centipetal movement, known as retrograde flow (Heath and Holifield, 1991; Mitchison and Cramer, 1996), then shift to dorsal surface ruffles. The dorsal surface ruffling is closely correlated with macropinocytic activity. In particular, circular ruffles formed on the dorsal surface are known to be precursor forms of macropinosomes (Swanson and Watts, 1995). Thus, each type of ruffling seems to be associated with a different cell function and controlled by a different signaling pathway (Allen et al., 1997; Cox et al., 1997; Hall, 1994; Ridley, 1994; Ridley et al., 1992), although both types of ruffles may be closely related to each other. Therefore, it would be important to clarify any differences in the molecular composition and machinery among the morphological phenotypes of ruffles.

Fc-mediated phagocytosis, in which pseudopodial extensions of the cell surface cover opsonized particles and enclose them to form phagosomes, occurs by regulated actin polymerization and reorganization. We recently distinguished two component activities of phagocytosis: pseudopod extension to form the phagocytic cup and a purse-string-like contraction that closes the phagosome, and also showed that several different classes of myosins are present in phagosomes (Araki et al., 1996; Swanson et al., 1999). Accordingly, the characteristic membrane transformations such as extension, curvature and contraction must have distinct mechanisms regulated by different kinds or isoforms of mechanochemical proteins, including myosins and actin-binding proteins. Therefore, we have focused on the possibility that each α-actinin isoform may contribute differently to cellular functions including ruffling, macropinocytosis and phagocytosis.

Since most actin-binding proteins are concentrated in the F-actin-enriched regions, the amount of each is greatly affected by the amount of F-actin. This fact is disadvantageous when comparing the functional relationship between actin and actin-binding proteins in different structures containing different amounts of F-actin. In fact, even though conventional fluorescence microscopy could demonstrate the localization of F-actin-associated proteins, it is hard to know the protein concentrations relative to F-actin. Moreover, conventional fluorescence microscopy and biochemical analyses may overlook the possibility that some actin-binding proteins are associated with F-actin at a high rate in some regions where a small amount of F-actin is present. Therefore, we have newly applied the ratio imaging technique to the analysis of the functional relationship between an actin-binding protein and F-actin, since ratio imaging can avoid the effects of other parameters such as cytoplasmic volume and F-actin amount (Araki and Hatae, 2000; Dunn and Maxfield, 1998). The ratio of the fluorescence intensity of FITC-immunolabeled actinin-4 to that of rhodamine phalloidin-labeled F-actin could indicate how much functional contribution actinin-4 made to F-actin-bundling, without the effect of F-actin content. We demonstrate here that actinin-4 preferentially participates in circular ruffling and has a role in macropinocytosis and phagocytosis in mouse macrophages.

Monoclonal antibody (mAb) against actinin-4 (Clone HCC-Lu-632 mouse IgM) was produced as previously described (Honda et al., 1998). Rhodamine phalloidin, Alexa 488-anti-rat IgG and fixable fluorescein-dextran M_r 10×10³ (FDx10) were purchased from Molecular Probes, Inc. (Eugene, OR). Bovine serum albumin (BSA, Fraction V), anti-α-actinin (actinin-1) mAb (Clone BM-75.2 mouse IgM), Texas Red anti-mouse IgM, FITC-anti-mouse IgM, horseradish peroxidase (HRP)-anti-mouse IgM, Agarose anti-mouse IgM were from Sigma Chemical Co. (St Louis, MO). Rabbit anti-cathepsin D serum was a gift from Dr Sadaki Yokota (Yamanashi Medical School, Japan). Anti-LAMP-1 (110 kDa lysosomal-associated membrane glycoprotein) mAb (Clone 1D4B) developed by Dr J. T. August was obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Biological Sciences, University of Iowa. Dulbecco’s-modified essential medium (DME) and fetal bovine serum (FBS) were from Gibco BRL (Grand Island, NY). All other reagents were purchased from Wako Pure Chemical (Osaka, Japan), unless otherwise indicated.

Bone marrow-derived macrophages were obtained from femurs of C3H HeJ mice as previously described (Swanson, 1989). After 6 or 7 days of culture, macrophages were harvested from dishes and plated onto 12 mm circular coverslips in 24-well culture dishes, or 10 cm dishes, then incubated overnight in medium lacking M-CSF (DME-10F: DME with 10% heat-inactivated FBS).

Other cultures, a human vulval epidermoid cancer cell line A431, a human monocytic leukemia cell line HL-60 and a human histiocytic lymphoma cell line U937, were grown in 10-cm dishes in DME-10F.

Thirty minutes before experiments, DME-10F was replaced with Ringer’s buffer (RB) consisting of 155 mM NaCl, 5 mM KCl, 1 Mm MgCl₂, 2 mM Na₂HPO₄, 10 mM glucose, 10 mM Hepes, pH 7.2 and 0.5 mg/ml BSA.

Macropinocytosis was stimulated by the addition of human recombinant M-CSF (2,000 unit/ml) to the macrophage culture (Araki et al., 1996; Racoosin and Swanson, 1992; Swanson, 1989). To label macropinosomes, some macrophages were incubated in RB containing 1 mg/ml fixable FDx10 and M-CSF (2,000 unit/ml) for 2-5 minutes at 37°C, followed by a brief rinse in PBS and fixation as described previously (Araki et al., 1996; Araki and Swanson, 1998). For FDx10 pulse/chase experiments, macrophages were pulse-labeled with 1.0 mg/ml fixable FDx10 for 2 minutes as described above, then quickly rinsed in RB and chased in FDx10-free RB + M-CSF for 0-10 minutes at 37°C. Some macrophage cultures were fed with 2 μm diameter latex (polystyrene) beads (Polyscience, Inc., Warrington, PA) and incubated for 5-30 minutes at 37°C to allow phagocytosis.

Cells were washed in cold PBS and scraped with a rubber policeman in lysis buffer (10 mM Hepes, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1 mg/ml NaN₃, 0.5 mM Na₃VO₄, 1 mM PMSF, 1 μg/ml leupeptin, 1 μg/ml pepstatin A) and homogenized on ice in a Dounce homogenizer with a tight-fitting pestle. After 30 minutes extraction on ice, the sample was centrifuged at 12,000 g for 15 minutes and the supernatant was recovered. For western blot analysis, the cell lysates (50 μg protein) were separated by 10% SDS-PAGE (Laemmli, 1970) and transferred to nitrocellulose membrane (Advantech Toyo, Japan). Actinin-4 and -1 proteins were revealed using mAb HCC-Lu-632 and mAb BM-75.2, respectively. After incubation with primary antibodies for 3 hours at room temperature, the blots were detected using HRP-conjugated anti-mouse IgM and TMB substrate (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD).

For immunoprecipitation, the macrophage cell lysate was incubated with anti-actinin-1 (BM-75.2) mAb overnight at 4°C, then agarose-conjugated anti-mouse IgM was added. After 3 hours incubation at 4°C, the agarose was washed 4 times with Tris-buffered saline (TBS) containing 1% Triton X-100 and 0.2% Tween-20, then lysed in Laemmli buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 5% glycerol, 715 mM 2-mercaptoethanol, 0.0025% bromophenol blue). The sample was separated by SDS-PAGE and analyzed by western blotting as described above.

Macrophages cultured on glass coverslips were usually fixed with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, 5% sucrose for 30 minutes at room temperature. For the combination of endocytic labeling and immunofluorescence, 0.1% glutaraldehyde was added to the usual fixative, so that FDx10-labeling would be efficiently retained in macropinosomes. After fixation, cells were rinsed four times 5 minutes in PBS and two times 5 minutes in 0.25% NH₄Cl in PBS, to quench free-aldehyde groups. The fixed cells were treated with a blocking solution consisting of 1% normal goat serum and 0.5% BSA in permeabilizing buffer (0.25% Triton X-100 or 1% saponin in PBS). The first and second antibodies were diluted in the blocking solution and incubated at room temperature for 90 minutes and 60 minutes, respectively. As primary antibodies, we used mouse anti-actinin-4 mAb HCC-Lu-632 at 1:20 dilution, mouse anti-actinin-1 mAb BM-75.2 (1:100 dilution), rat anti-LAMP-1 mAb 1D4B (1:20 dilution) and rabbit anti-cathepsin D polyclonal antibody (1:500 dilution). As negative controls, normal mouse IgM or normal rabbit serum at the same concentration was substituted for the specific antibody. As secondary antibodies we used FITC or Texas red-conjugated anti-mouse IgM, Alexa 488-conjugated anti-rat IgG and FITC-conjugated anti-rabbit IgG (all 1:250 dilution). The indirect immunofluorescence protocol was previously described in detail (Araki and Swanson, 1998). To visualize F-actin, rhodamine-phalloidin (5 units/ml) was added to the secondary antibody dilution. We carefully determined the optimal dilution of antibodies and phalloidin by titration and their saturation kinetics. The specimen coverslips were mounted on glass slides using Antifade (Molecular Probes, Inc., Eugene, OR) and observed by a confocal laser microscope (Olympus GB200) or an epifluorescence microscope (Nikon TE300).

For digital image analysis, 8 bit confocal images of both fluorescein and rhodamine were obtained by simultaneous dual wavelength excitation (488 nm and 568 nm, respectively). Laser intensity, aperture size, gain and black level settings were carefully determined so that each signal was enough to be visualized, but any of the pixels in the interest area were not saturated. We confirmed that FITC and rhodamine images were exactly aligned in the same focal plane; FITC and rhodamine signals were detected only in the 488 nm channel and the 568 nm channel, respectively; and there was little signal in the other channel (Araki and Hatae, 2000). Also, essentially the same results were obtained when FITC- and rhodamine-labeling were replaced with Texas red- and NBD-labeling, respectively. These images saved as TIFF files were processed by MetaMorph Imaging System (Universal Imaging Co., West Chester, PA) to get the ratio images of actinin-4/F-actin and sometimes the reversal (F-actin/actinin-4). The ratio images were shown as spectrum pseudocolor corresponding to the ratio value. Quantitative and statistical analysis of ratio values was carried out at more than 150 different regions of ruffles in 30 cells using the line scan program of MetaMorph software.

Macrophages were scrape-loaded with either an antibody against the central rod of actinin-4 or -1, control mouse myeloma IgM, or PBS only, as previously described (McNeil et al., 1984; Swanson et al., 1999). Briefly, macrophages cultured on a 35-mm dish were scrape-loaded by first washing with PBS, then replacing it with 50 μl of IgM (0.5 mg/ml in PBS) and scraping the cells off the dish with a rubber policeman. Cells were washed in PBS then replated on a 24-well dish at a cell density of 2.5×10⁵ cells/well. After 60 minutes incubation in DME-10F, macrophages were incubated with FDx10 (1.0 mg/ml in DME-10F), a fluid-phase pinocytic probe, in the presence of M-CSF (2,000 U). Although 2-5 minutes labeling of FDx10 was used to label only macropinosomes in other morphological experiments, we chose a 30 minute labeling time for spectrofluorometry, since the signal/noise ratio of FDx10 at 5 minutes was not high enough to assess the inhibitory effect. Then, macrophages were thoroughly washed and lysed in a lysis buffer. The amount of FDx10 in the cell lysate was measured by a fluorescence spectrophotometer (Hitachi 650-40) as previously described (Araki et al., 1996; Araki and Swanson, 1998). Macropinocytic activity was revealed as the amount of intracellularly accumulated FDx10 normalized by the total cell protein.

Macrophages were cultivated on plastic sheets and stimulated with M-CSF for 5 minutes before fixation. Cells were fixed with 2% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, containing 6% sucrose for 1 hour at room temperature. Fixed cells were rinsed, postfixed with 1% osmium tetroxide and conventionally processed for scanning electron microscopy as previously described (Araki et al., 1996). Specimens were observed by a Hitachi S-900 SEM.

Specific reactivity of mAb HCC-Lu-632 with human actinin-4 was previously determined using human epithelial cell lines including A431, GST-fusion proteins expressed in E. coli and in vitro translation products of cDNA of actinin-4 and actinin-1 (Honda et al., 1998). To confirm its reactivity with mouse actinin-4, western blot and immunoprecipitation analyses were performed using mouse macrophage lysates as well as human cell lines. Western blot analysis revealed that mAb HCC-Lu-632 reacted with a single protein of approx. 100 kDa in the whole cell lysate of mouse macrophages as well as human epithelial cell line A431, monocytic cell line HL-60 and histiocytic cell line U937 (Fig. 1A). Although monoclonal antibody against chick actinin-1 (mAb BM-75.2) also detected a single band at the same molecular mass (Fig. 1B), mAb BM-75.2 immunoprecipitable products were not reacted with mAb HCC-Lu-632 (Fig. 1C). Normal mouse IgM, used as a control, did not detect 100 kDa protein in any cell lysates nor mAb BM-75.2 immunoprecipitable products (data not shown). These indicate that mAb HCC-Lu-632 cross-reacts with mouse actinin-4, but not with mouse actinin-1.

Scanning EM displayed the many ruffles that formed at the peripheral edge and the dorsal surface of M-CSF-stimulated macrophages. The morphology of the dorsal ruffles was varied from straight to curved or circular (Fig. 2A). As previously shown by phase-contrast time-lapse video microscopy of living macrophages (Araki et al., 1996; Swanson and Watts, 1995), circular ruffles are most frequently formed by inward curving of lateral edge ruffles; these then close on top to become phase-bright macropinosomes. Newly formed macropinosomes move toward the perinuclear region while gradually shrinking in size. Circular ruffles were defined by a phase-dark rim, which distinguished them from the phase-bright macropinosomes (Fig. 2B). Two to five minutes pulse-labeling with the fluid-phase probe FDx10 labeled macropinosomes but not circular ruffles (Fig. 2C). Rhodamine-phalloidin labeling showed that circular ruffles were rich in F-actin, but FDx10-labeled macropinosomes had very little associated F-actin (Fig. 2D). This indicates that F-actin dissociated from macropinosomes shortly after circular ruffles close into intracellular organelles. In this paper, we defined circular ruffles as phase-dark-lined, FDx10-unlabelable and F-actin-rich profiles; we defined macropinosomes as phase-bright, FDx10-labelable and F-actin-poor profiles.

Control and M-CSF-stimulated macrophages showed essentially the same features of α-actinin localization. We described only M-CSF-stimulated macrophages here, since these cells displayed more ruffles and the features of α-actinin localization were more pronounced.

Conventional confocal images of actinin-4 immunofluorescence and rhodamine-phalloidin-stained F-actin showed that actinin-4 was abundant in F-actin-rich cell cortical regions such as ruffles of the cell margin and the dorsal surface of macrophages (Fig. 3A-C). However, the color merged image and its line scan quantitative analysis indicated that the intensities of both fluorescences were not always correlative (Fig. 3C,C′). A confocal image of actinin-4 immunofluorescence shows the total amount of actinin-4 in cytoplasm optically sliced at 0.5 μm-thickness, but does not reveal the relative amounts of actinin-4 to F-actin. The relative amounts of actinin-4 to F-actin were determined by fluorescence ratio imaging analysis. The ratio image topographically revealed that the actinin-4/F-actin ratio was especially high at circular ruffles (Fig. 3D). Conversely, the actinin-4/F-actin ratio was lower at the peripheral edge ruffles. This can be confirmed by the reversal ratio (F-actin/actinin-4) image in which peripheral edge lamellipod-like ruffles showed a high ratio value (Fig. 3E). In peripheral edge ruffles, some regions curving inward showed a higher actinin-4/F-actin ratio than did straighter regions (Fig. 4A,B). These ratio images indicate that actinin-4 contributes to F-actin-bundling more in circular or curved ruffles than in straight ruffles such as flat lamellipodia.

Some vacuolar structures, presumably macropinosomes, were also labeled with anti-actinin-4 to variable degrees; such structures did not contain abundant F-actin (Fig. 4A-C). Ratio imaging showed that the actinin-4/F-actin ratio was considerably higher in the vicinity of macropinosome-like structures, owing to the diminished F-actin levels (Fig. 4B,C). Other cytoplasmic regions were weakly positive for actinin-4. These findings were further confirmed by a statistical analysis of ratio values in different regions of 30 cells (Table 1). By a statistical t-test analysis, it was shown that ratio values in curved ruffles and circular ruffles were significantly higher than that in straight ruffles (P<0.01). The difference in ratio values between curved ruffles and circular ruffles was not significant (P>0.1).

Immunofluorescence using mAb BM-75.2, which recognizes actinin-1, a classical isoform of α-actinin, indicated that actinin-1 was in all F-actin-containing structures. These included peripheral edge ruffles, dorsal surface ruffles and circular ruffles (Fig. 4D). Since the fluorescence intensities of F-actin and actinin-1 correlated (Fig. 4F), the ratio image was less contrasted (Fig. 4E). Thus, actinin-1 appeared to localize more uniformly, relative to F-actin.

Confocal images of optical sections containing the basal plasma membrane also showed different localizations of two α-actinin isoforms. Unlike fibroblasts, smooth muscle cells and epithelial cell cultures, macrophages do not have stress fibers, but instead focal adhesion plaque-like structures corresponding to cell-substratum contact sites (Fig. 5). It was reported that these F-actin-rich adhesion structures, termed podosomes or F-actin dots, included several actin-binding proteins such as vinculin, fimbrin, talin and α-actinin, similar to focal adhesion plaques (Correia et al., 1999; Marchisio et al., 1987). Our confocal observation clearly revealed that actinin-4 did not localized in podosomes (Fig. 5A,B), while actinin-1 did (Fig. 5C,D).

To confirm the association of actinin-4 with macropinosomes, macrophages were pulse-labeled with FDx10 for 5 minutes, fixed and immunolabeled for actinin-4 using Texas red-labeled secondary antibody. Most FDx10-labeled macropinosomes that were relatively large and located at cell periphery were associated with actinin-4, although perinuclear pinosomes were scarcely labeled for actinin-4 (Fig. 6A-C). Such actinin-4-negative pinosomes might be macropinosomes at a later stage. The quantitative data of FDx10 pulse-labeling and chase experiments supported this finding. As shown in the graph in Fig. 7A, more than 70% of 0-2 minutes old (2-minutes pulse/no chase) newly formed macropinosomes showed the actinin-4 association. The actinin-4 positive fraction decreased with increasing macropinosome age (chase time). Less than 20% of 10-12 minutes old (2-minutes pulse/10-minutes chase) macropinosomes were immunolabeled for actinin-4. No association of actinin-1 with any FDx10-labeled macropinosomes was observed by confocal microscopy (Fig. 6D-F).

Racoosin and Swanson (1993) reported that early macropinosomes matured to late macropinosomes and finally merged with tubular lysosomes, all within approximately 15 LAMP-1 and cathepsin D. Fluorescence microscopy and quantitative analysis revealed that more than half of the macropinosomes labeled for actinin-4 were negative for LAMP-1 and were generally located at the cell periphery. The majority of LAMP-1-positive macropinosomes located at the perinuclear region were relatively small (< c.a. 2 μm) and not associated with actinin-4 (Fig. 7B-G). However, macropinosomes showing both actinin-4 and LAMP-1 labeling were also occasionally seen at the perinuclear region. Counting of 200 actinin-4 associated macropinosomes indicated that 34.9% of total actinin-4-associated macropinosomes was positive for LAMP-1. Curiously, macropinosomes having both minutes. During the sequential processes of macropinosome maturation and fusion with tubular lysosomes, delivery of LAMP proteins to macropinosomes precedes delivery of cathepsins (Racoosin and Swanson, 1993). In order to clarify the stages of macropinosome maturation at which actinin-4 associates, we further characterized actinin-4-associated macropinosomes by dual immunolabeling with the endosomal/lysosomal markers such as LAMP-1 and actinin-4 were considerably large (>3 μm the most frequent size) in spite of the perinuclear location (Fig. 7E-G). Only a small percentage (3.75% of total actinin-4-associated macropinosomes) was positive for cathepsin D. Association of actinin-4 with cathepsin D-positive tubular lysosomes was hardly seen. Taken together, these observations indicate that the dissociation of actinin-4 from late macropinosomes begins before macropinosomes gain LAMP-1 and is nearly complete before macropinosomes become cathepsin D-positive by fusing with tubular lysosomes.

Our previous study showed that both anti-actinin-4 (HCC-Lu-632) and anti-actinin-1 (BM-75.2) monoclonal antibodies recognized the central rod domain of α-actinin. Since the central rod accounts for antiparallel dimerization of α-actinin to form a rod-shaped molecule with an actin-binding domain at either end (Djinovic′-Carugo et al., 1999; Matsudaira, 1991), these monoclonal antibodies may perturb F-actin bundling in living cells. To address the functional contribution of α-actinin isoforms to macropinocytosis more directly, we compared the effect of scrape-loaded antibodies (IgM) against actinin-4 and -1. After scrape-loading with IgM (0.5 mg/ml), cells were incubated with 1.0 mg/ml FDx10 for 30 minutes to allow fluid-phase pinocytic uptake. Anti-actinin-4 IgM loading significantly reduced intracellular accumulation of FDx10 by fluid-phase pinocytosis for 30 minutes, compared with anti-actinin-1 IgM or control IgM (Fig. 8). Higher concentrations (1.0 mg/ml) of anti-actinin-1 and control IgM also did not show significant inhibition on macropinocytosis (data not shown).

Macropinocytosis and phagocytosis are mechanistically similar (Araki et al., 1996; Swanson and Watts, 1995), although they differ in quality of internalized substances; fluid and particles, respectively. In phagocytosis, pseudopod extension along a particle surface results in phagocytic cup formation. The phagocytic cup then closes into the phagosome. Finally, the phagosome becomes a phagolysosome by fusing with lysosomes. We observed F-actin and actinin-4 redistributions in macrophages during phagocytosis of latex beads. Latex beads were fed to macrophages to allow phagocytosis for 10-30 minutes. Since latex beads could be confirmed by the bright-field diffraction mode of confocal microscopy, phagocytic cups or phagosomes were distinguishable from circular ruffles or macropinosomes. Also, phagocytic cups could be distinguished from phagosomes, since F-actin was enriched in phagocytic cups, whereas F-actin was scarcely seen around phagosomes and phagolysosomes. Immunofluorescence by anti-actinin-4 showed that actinin-4 localized in phagocytic cups and phagosomes 10 minutes after latex beads feeding (Fig. 9A-C). Ratio imaging indicated that actinin-4/F-actin was higher at phagocytic cups and some phagosomes than straighter ruffles (Fig. 9D). However, after 30 minutes, most latex bead-containing structures, presumed to be at a late stage of phagosomes or phagolysosomes, were unlabeled for actinin-4 (data not shown).

Consistent with a previous report (Allen and Aderem, 1996), actinin-1 was also enriched around phagocytic cups. However, ratio imaging showed that actinin-1/F-actin was not so high as actinin-4/F-actin. Moreover, actinin-1 was hardly seen around intracellular phagosomes (not shown).

It is well known that α-actinin cross-links F-actin to form F-actin bundles (stress fibers) and mediates membrane-F-actin interactions via other proteins such as talin, vinculin and integrin. Nonmuscle α-actinin, presumed to be actinin-1, is primarily localized in focal adhesion structures of cultured fibroblasts, epithelial cells (Honda et al., 1998; Mangeat and Burridge, 1984; Meigs and Wang, 1986) and macrophages (Marchisio et al., 1987). It was also reported that α-actinin participated in cadherin-mediated cell-cell adhesion via α-catenin at adherens junctions in epithelial cells (Knudsen et al., 1995). These adhesion complexes stabilize the cell on adherent substrates or among epithelial tissues. Glück et al. (1993, 1994) reported that transfection with actinin-1 cDNA suppressed cell motility and tumorigenicity. These studies indicate that α-actinin is a structural component, maintaining cell location, rather than contributing to force-generation. In some immunofluorescence studies, α-actinin was also observed in dynamically moving structures including leading edges, phagocytic cups, phagosomes and contractile vacuoles of Dictyostelium discoideum (Allen and Aderem, 1996; Furukawa and Fechheimer, 1994). However, it is possible that the antibodies against α-actinin that showed multiple localizations recognized multiple isoforms of the protein with the same molecular mass. In our observation on macrophages, actinin-1 was observed in all F-actin-containing structures, including dorsal and peripheral edge ruffles, phagocytic cups and basal podosomes. However, the ratio images indicated that actinin-1 was distributed homogeneously in such structures, relative to F-actin. On the contrary, actinin-4 was shown to be most concentrated in special regions of ruffles such as curved and circular ruffles on the dorsal surface of macrophages. Consistent with the previous findings in cancer cells (Honda et al., 1998), actinin-4 was not concentrated in focal adhesion structures, called podosomes in macrophages, as was actinin-1. Judging from the localization of actinin-4, it is unlikely that actinin-4 contributes to stabilization of the cell on the adherent substrate. Conversely, actinin-4 preferentially localizes to actively moving structures such as dorsal ruffles. This is consistent with the proposal that actinin-4 predicts the metastatic potential of cancer, since active membrane ruffling of cancer cells is considered to be a parameter of tumor cell invasion and metastasis (Jiang, 1995). In an epithelial cancer cell line, actinin-4 was concentrated at the leading edge of a motile epithelial sheet during wound healing (Honda et al., 1998). However, our ratio imaging demonstrated that actinin-4 was not concentrated in peripheral edge ruffles of macrophages. This discrepancy may be due to other differences between cancer cells and phagocytic cells. In many cell types, including cancer cells, ruffling is observed in response to certain extracellular factors and leading edge lamellipod-like ruffles are required for directed cell migration. In macrophages showing amoeboid patterns of movement, however, edge ruffling is constitutively observable even when they are not migrating. Moreover, PI3-kinase inhibitors suppress growth factor-induced ruffling in some cell types (Kotani et al., 1994; Wennstrom et al., 1994) but not in macrophages (Araki et al., 1996). Thus, the leading edge lamellipodial extension during cell migration and the peripheral edge ruffling constitutively shown in macrophages may be functionally different. Though peripheral edge ruffles of macrophages have a low actinin-4/F-actin ratio on the average, some portions have a high actinin-4/F-actin ratio. This may imply that peripheral edge ruffling of macrophages consists of more than a single phenotype for different cell functions and some such as inward curving ruffles, are transit forms from peripheral ruffles to dorsal circular ruffles.

Our study indicated that the actinin-4/F-actin ratio was high around newly formed macropinosomes. Macropinosomes are formed from circular ruffles by closure of the ruffle’s tip. As shown in Fig. 2, newly formed macropinosomes immediately lost most of the associated F-actin, while circular ruffles were enriched with F-actin. Nevertheless, actinin-4 retained a small amount of F-actin around macropinosomes for a short while. This suggests that actinin-4 preferentially contributed to actin-bundling around early macropinosomes. Newly formed macropinosomes with variable sizes (often as large as 2-5 μm in diameter) move toward the perinuclear region, shrink in size and ultimately merge with tubular lysosomes. During this process, macropinosomes acquire lysosomal membrane glycoproteins such as LAMP-1 (<5 minutes) and then several minutes later the cation-independent mannose-6-receptor and markers of lysosomal content such as cathepsins derived from tubular lysosomes (Racoosin and Swanson, 1993). Our FDx10 pulse/chase experiment showed that actinin-4 gradually dissociated from macropinosomes. Comparing the time course of actinin-4 dissociation in Fig. 7A with previous published data (Racoosin and Swanson, 1993), it appears that actinin-4 dissociation occurs during acquisition of LAMP proteins. However, dissociation of actinin-4 from macropinosomes seemed to be unnecessary for LAMP-1 delivery to macropinosomes, because some actinin-4 positive macropinosomes (<35%) showed LAMP-1 in their membranes. It is noteworthy that macropinosomes having both actinin-4 and LAMP-1 are generally larger than macropinosomes having only LAMP-1. This implies that the actinin-4 association with the macropinosomal membrane may be dependent on not only the degree of macropinosome maturation but also the size of macropinosomes.

What is the role of actinin-4 around macropinosomes? It has been known that F-actin immediately dissociates from the plasma membrane when circular ruffles close into macropinosomes. However, our quantitative analysis by line scan on a macropinosome (Fig. 4C) indicated that F-actin was apparently associated with the macropinosome, although the amount of F-actin was much less than in circular ruffles. This finding is not surprising because F-actin facilitates endosome trafficking and/or endosome/lysosome fusion (Durrbach et al., 1996). From our ratio image analysis, it is conceivable that actinin-4 predominantly cross-links F-actin and forms an actinin-4-F-actin network surrounding macropinosomes. A previous report revealed that a cortical α-actinin-F-actin network prevented deformation of cell shape against osmotic stress (Rivero et al., 1996). A rheological study with a torsion pendulum showed that the viscoelastic properties of F-actin gels cross-linked by α-actinin differed from gels cross-linked by the 120 kDa gelation factors: α-actinin-F-actin networks responded to deformation with a strongly damped oscillation, whereas the gelation factor-F-actin networks reacted in a more elastic, weakly damped way (Janssen et al., 1996). Moreover, a mutation study using Dictyostelium indicated that lack of α-actinin caused a significant reduction of the viscoelastic response after deformation at high frequency (Eichinger et al., 1996). These results suggest that α-actinin is responsible for organizing the actin cytoskeleton against fast and strong impacts. Consistent with these properties of α-actinin-F-actin gels, it is considered that the actinin-4-F-actin network surrounding macropinosomes may provide a constructive brace to maintain a large vacuolar structure in the cell. Larger vacuoles may require more structural resistance to the cytoplasmic pressure (viscosity) during centripetal movement. Although we cannot rule out the possibility that the actinin-4–F-actin interaction may be directly involved in the contraction (shrinkage) force of macropinosomes and/or the intracellular transport of macropinosomes, there has been no report indicating that α-actinin is directly involved in such force generating machinery.

In the process of phagocytosis, actinin-4 was enriched in phagocytic cups and some phagosomes but not in late phagosomes or phagolysosomes. The relationship between actinin-4 and F-actin in phagocytic cups and phagosomes seems to be comparable to that in circular ruffles and macropinosomes, respectively. Since the processes of macropinocytosis and phagocytosis are mechanistically similar (Araki et al., 1996; Swanson et al., 1999; Swanson and Watts, 1995), it seems likely that actinin-4 plays the same role in both macropinocytosis and phagocytosis. Taken together, the localization of actinin-4 in macrophages correlated with actively moving structures involved in endocytosis but not in migratory movements, whereas actinin-4 in cancer cells is associated with cell migration.

Actinin-4 shows a high degree of similarity to actinin-1 (80% nucleotide and 86.7% amino acid similarity. The amino acid sequences indicate several domains conserved among α-actinin family members, including actin-binding domains, pleckstrin-homology (PH) domains, central rod domains and two EF-hand calcium regulation domains. This indicates that both actinin-1 and actinin-4 actin-binding activities could be commonly regulated by Ca²⁺ and phosphoinositides (Djinovic′-Carugo et al., 1999; Honda et al., 1998; Matsudaira, 1991). Molecular properties such as molecular flexibility, the ability to bind other molecules and F-actin cross-linking angles may affect the cell structural configurations generated by F-actin gel network (Djinovic′-Carugo et al., 1999). However, such properties have not been obtained for actinin-1 or -4.

Different localizations of α-actinin isoforms suggest that their recruitment to specialized regions could be regulated by different signals. Recruitment of actinin-4 to actively moving structures should be more highly regulated by rapid signal transduction than actinin-1 which is present homogeneously, or constitutively at stable structures such as basal adhesion podosomes. Honda et al. (1998) reported that PI3-kinase inhibition caused actinin-4 translocation into the nucleus from cytoplasm in some but not all cancer cell lines. In our preliminary observation of macrophages treated with wortmannin, a potent PI3-kinase inhibitor, the nuclear translocation of actinin-4 was not observed. Instead, actinin-4 labeling somewhat decreased in macrophages by wortmannin, suggesting that actinin-4 dissociated from the actin cytoskeleton (N. Araki et al., unpublished data). Therefore, we expect that PI3-kinase may be involved in signal transduction for actinin-4 recruitment and/or F-actin-binding.

The small GTP-binding proteins Cdc42, Rac and Rho regulate the formation of distinct F-actin-based structures (Guillemot et al., 1997; Ridley and Hall, 1992). For example, Rac regulates lamellipodium formation and membrane ruffling, while Cdc42 induces the formation of filopodia (Allen et al., 1997; Cox et al., 1997; Hall, 1994; Ridley et al., 1992). The various forms of ruffling may be accounted for by different signals and machinery components.

Our scrape-loading experiments revealed that the antibody against the central rod domain of actinin-4 significantly reduced the rate of macropinocytosis in macrophages, suggesting that actinin-4 facilitates macropinocytosis, though it is still unknown whether actinin-4 is definitely required for macropinocytosis. Anti-actinin-1 loading showed less effect on macropinocytosis. These results strongly support the notion that F-actin bundling by actinin-4 functionally contributed to macropinocytosis more than that by actinin-1. However, we cannot argued that actinin-1 was not involved in macropinocytosis. A recent study using F-actin cross-linking proteins mutant strains of Dictyostelium showed double mutants lacking α-actinin and gelation factor ABP120 or 34 kDa actin-bundling protein exhibited a reduced rate of fluid phase endocytosis, while single mutants lacking α-actinin did not, suggesting that α-actinin function may be guaranteed by more than a single molecule (Rivero et al., 1999). There is likely some redundancy of function among these actin cross-linking proteins. Further studies using dominant negative mutants or gene knockout of individual isoforms of actin-binding proteins and signal transduction-related molecules would be required for a full understanding their interplay and physiological roles in cell functions.

To our knowledge, this is a first report that reveals a difference in molecular composition between dorsal circular or curved ruffles and edge straight ruffles. The present study can provide insight into the distinct roles of α-actinin isoforms in specific cellular functions and a clue to resolving how ruffling in special forms occurs. Studies of signaling mechanisms of actinin-4 recruitment to specialized F-actin-containing structures such as circular ruffles are now in progress using macrophages and other cell types.

The authors thank Dr Joel A. Swanson, University of Michigan Medical School for critical reading of this manuscripts and helpful advice, Drs M. Tokuda and K. Kawakami for help with immunoprecipitation and cell culture, Dr S. Yokota for a gift of anti-cathepsin D antibody and Drs M. Hamasaki, T. Ishida and T. Toyosima for helpful discussion. Secretarial and technical services provided from Ms H. Yamamoto and Mr K. Yokoi are greatly appreciated. This study was supported by a Grant-in-Aid for Scientific Research from Japan Society for the Promotion of Science and by a grant from the Ichiro Kanehara Foundation (to N.A.).