Indirect Role for COPI in the Completion of Fcγ Receptor-mediated Phagocytosis*

Recent evidence suggests that extension of pseudopods during phagocytosis requires localized insertion of endomembrane vesicles. The nature of these vesicles and the processes mediating their release and insertion are unknown. COPI plays an essential role in the budding and traffic of membrane vesicles in intracellular compartments. We therefore assessed whether COPI is also involved in phagosome formation. We used ldlF cells, a mutant line derived from Chinese hamster ovary cells that express a temperature-sensitive form of εCOP. To confer phagocytic ability to ldlF cells, they were stably transfected with Fc receptors type IIA (FcγRIIA). In the presence of functional COPI, FcγRIIA-transfected ldlF cells effectively internalized opsonized particles. In contrast, phagocytosis was virtually eliminated after incubation at the restrictive temperature. Similar results were obtained impairing COPI function in macrophages using brefeldin A. Notably, loss of COPI function preceded complete inhibition of phagocytosis, suggesting that COPI is indirectly required for phagocytosis. Despite their inability to internalize particles, COPI-deficient cells nevertheless expressed normal levels of FcγRIIA, and signal transduction appeared unimpeded. The opsonized particles adhered normally to COPI-deficient cells and were often found on actin-rich pedestals, but they were not internalized due to the inability of the cells to extend pseudopods. The failure to extend pseudopods was attributed to the inability of COPI-deficient cells to mobilize endomembrane vesicles, including a VAMP3-containing compartment, in response to the phagocytic stimulus.

Phagocytosis of microbial pathogens by leukocytes is an essential component of the host defense against infection. Microorganisms become internalized into a membrane-bound vacuole called a phagosome, which subsequently matures upon fusion with endosomes and lysosomes into a powerful microbicidal organelle (1). The phagocytic capacity of neutrophils and macrophages is remarkable; individual cells can take up multiple and/or very large particles. As a result, the area of membrane internalized is significant and has in some cases been estimated to approach or exceed the total initial surface area of the phagocyte (2). Despite the internalization of a considerable membrane expanse, no net loss of surface has been detected, and, in fact, surface gains have been documented electrophysiologically (3) and by spectroscopic means (4). These observations therefore suggest that active exocytosis of endomembranes accompanies phagocytosis.
Pharmacological studies have in fact suggested that secretion of endomembranes is essential for optimal phagocytosis. First, inhibitors of phospholipase A 2 precluded phagocytosis, with a concomitant accumulation of clear vesicles under the site of particle attachment (5). Second, Cox et al. (6) found that inhibition of phosphatidylinositol 3Ј-kinase by wortmannin, which can inhibit exocytosis, prevented phagocytosis of large particles, with comparatively minor effects on the ingestion of smaller ones. Third, botulinum and tetanus toxins, which interfere with SNARE-mediated membrane fusion, were found to induce partial inhibition of phagocytosis (7). Finally, it was shown that exocytic vesicles enriched in VAMP3, a v-SNARE that mediates secretion of recycling vesicles (8), were locally secreted at sites of phagocytosis (9). Jointly, these studies suggest that extension of pseudopods during particle engulfment involves the focal exocytosis of endomembrane vesicles.
The mechanisms mediating the formation and traffic of these putative vesicles are not understood. However, existing information regarding endomembrane traffic may be applicable to the genesis of phagosomes. In particular, COPI has been convincingly shown to participate in the budding and traffic of vesicles between the Golgi complex and the endoplasmic reticulum (ER) 1 (10 -12) as well as vesicles involved in recycling and endosome maturation (13,14). COPI exists as a protein complex in the cytosol and can assemble on docking sites of membranes, where it promotes the fission of vesicles (15,16). We therefore considered the possibility that COPI may participate in the process of phagosome formation.
To this end, we used brefeldin A (BFA), which inactivates the GTP exchange factors of adenosine ribosylation factor, to inhibit COPI or the Chinese hamster ovary (CHO) mutant cell line called ldlF, originally isolated by Hobbie et al. (10). These cells harbor a temperature-sensitive mutation in the gene encoding ⑀COP, a component of COPI (10,11). When incubated at the permissive temperature (34°C), such cells express functional ⑀COP, while incubation at the restrictive temperature (Ն39°C) results in destabilization and rapid degradation of the mutant ⑀COP, thereby inactivating COPI (10,11). While extremely useful for the study of COPI function, ldlF cells are not phagocytic. In order to analyze the role of COPI in phagocytosis, ldlF cells were stably transfected with Fc receptors (Fc␥RIIA). Such heterologous transfection of opsonin receptors was shown earlier to confer phagocytic capacity to nonmyeloid cells, including CHO cells (17)(18)(19). We found that, like CHO cells, ldlF cells transfected with Fc␥RIIA (called FcR-ldl hereafter) effectively internalized IgG-opsonized particles when grown at the permissive temperature. However, following down-regulation of COPI, phagocytosis was progressively and drastically inhibited. Evidence is provided that COPI is indirectly required for phagocytosis by maintaining a VAMP3containing endomembrane pool that appears to be required for pseudopod extension during Fc␥R-mediated phagocytosis.
Cell Culture and Handling-The murine macrophage cell lines, J774 and RAW 264.7, were obtained from the ATCC (Manassas, VA). RAW macrophages were transfected with Fugene 6, as described by the manufacturer (Roche Molecular Biochemicals). ldlF cells, bearing a thermolabile mutation in the ⑀COP, were the generous gift of Dr. M. Krieger. Wild-type CHO and ldlF cells were stably transfected with Fc␥RIIA cDNA using calcium phosphate, yielding FcR-CHO and FcRldl cells, respectively. All cell lines were grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 5% penicillin-streptomycin (Life Technologies, Inc.) and maintained under 5% CO 2 . Unless otherwise indicated, FcR-CHO cells were grown at 37°C, and FcR-ldl cells were grown at 34°C. For down-regulation of COPI, FcR-ldl cells were placed at the restrictive temperature of 39°C for the specified periods. Alternatively, cells were incubated with 100 M BFA for the indicated periods, followed by phagocytosis in the continuous presence of BFA.
Assessment of Phagocytosis and Particle Adherence-To assess phagocytosis, SRBC were opsonized with goat or rabbit anti-SRBC IgG (1:40 for 1 h at 37°C) and then washed in cold phosphate-buffered saline to remove unbound IgG. The opsonized SRBC were added to plated phagocytic cells (ϳ10 SRBC/cell) and incubated for 1 h at 37°C. Extracellular SRBC were then removed by a brief (30-s) hypotonic lysis in water, and internalized SRBC, which resist lysis, were quantified under Nomarski optics. To assess particle adherence, FcR-CHO cells or FcRldl cells were incubated with opsonized SRBC for 30 min at 4°C. Nonadherent SRBC were removed by washing in phosphate-buffered saline, and the number of bound SRBC/cell was quantified under light microscopy. Zymosan particles and 0.8-m latex beads were opsonized with 1 mg/ml human IgG and washed as described above.
Immunofluorescence and F-actin Staining-To label F-actin, cells were fixed for up to 3 h with 4% paraformaldehyde and then permeabilized with 0.1% Triton X-100 and incubated with rhodamine-phalloidin (0.01 unit/ml phosphate-buffered saline). To stain for ␣-mannosidase II, calnexin, catalase, and phosphotyrosines, fixed and permeabilized cells were incubated with a 1:100, 1:1000, 1:500, or 1:100 dilution of the respective antibodies for 1 h and subsequently incubated for 1 h with fluorochrome-conjugated secondary antibodies used at 1:1000. To stain for GFP, fixed, nonpermeabilized VAMP3-GFP-transfected cells were incubated with a 1:600 dilution of rabbit anti-GFP antibody followed by Cy3-conjugated anti-rabbit at 1:1000. FcR-ldl (not permeabilized) cells were stained with a 1:50 dilution of anti-Fc␥RIIA and undiluted anti-␤ 1 -integrin antibodies to quantitate surface expression of Fc␥RIIA and ␤ 1 -integrin, respectively. Fluorescence was analyzed using either a Leica model TCS4D or a Zeiss LSM 510 laser confocal microscope. Composites of confocal images were assembled and labeled using Photoshop (Adobe, Mountain View, CA) and Microsoft Powerpoint software. Quantification of immunofluorescence was performed using Scion Image (Scion Image, MA).
Quantification of Surface Fc␥RIIA and ␤ 1 -Integrin-Surface ␤ 1 -integrin was quantified using fixed, nonpermeabilized cells attached to glass coverslips by confocal microscopy. This was accomplished by acquiring confocal optical slices at a constant interval and integrating the total cellular fluorescence by stacking the collected slices. Fluorescence intensity of the reconstructed images was quantified using Scion Image. For flow cytometry, the cells were immunostained as above and then scraped off the coverslips in ice-cold divalent cation-free phosphatebuffered saline. After washing, the cells were analyzed as in Ref. 7.
Measurement of Free Cytosolic Calcium ([Ca 2ϩ ] i )-Cells grown on glass coverslips were incubated overnight at either 34 or 39°C and then loaded with fura-2 by incubation with 10 M of the parental acetoxymethyl ester for 30 min. Coverslips were then mounted in a thermostatted Leiden holder on the stage of a Zeiss IM-35 microscope, equipped with a ϫ 63 oil immersion objective. The microscope set-up has been previously described in detail (18). Calibration of fluorescence ratio versus [Ca 2ϩ ] i was performed as described (20). All measurements were at 37°C.
Scanning and Transmission Electron Microscopy-For scanning electron microscopy, cells were fixed in 2% glutaraldehyde and postfixed with 1% OsO 4 . Following washing, the cells were dehydrated in a graded series of ethanol. The samples were then critical point-dried in a Pelco CPD-2 critical point drying device and mounted on EM stubs with colloidal silver glue. They were then coated with evaporated gold/ palladium with a Pelco sputter coater model 3 for 50 s at 18 mA. The samples were then examined with a Philips XL 30 scanning electron microscope.
For transmission EM, cells were fixed as for scanning EM. The cells were washed extensively and then en bloc stained with 1% aqueous uranyl acetate for 30 min. Following washing, the samples were dehydrated through a graded series of ethanol and then embedded in Epon as we have described previously (21). Thin sections were cut and stained with lead citrate and uranyl acetate and observed with a Philips CM-12 electron microscope.
Receptor-mediated Endocytosis and Recycling of 125 I-Transferrin-FcR-ldl cells were either maintained at 34°C or incubated at 39°C for Ն12 h and then serum-starved for 1 h. The rate of receptor-mediated endocytosis of 125 I-Tfn was then measured by incubating cells with 0.4 Ci/ml of 125 I-Tfn for either 10 or 30 min. Extracellular 125 I-Tfn was then washed and stripped with phosphate-buffered saline solution at pH 3.0. To measure the rate of 125 I-Tfn recycling, cells were incubated for 15 or 30 min in the absence of 125 I-Tfn after a pulse of 30 min. All steps were performed at 37°C. The cells were then lysed, and the amount of internalized 125 I-Tfn was quantified with a ␥ counter.

Modulation of ⑀COP Expression in FcR-ldl Cells-
To study the role of COPI in phagocytosis, ldlF cells were stably transfected with Fc␥RIIA, yielding FcR-ldl cells. To verify that the mutation characteristic of ldlF cells persisted in FcR-ldl cells, the ⑀COP content of both cell types was assessed by immunoblotting following incubation at the permissive (34°C) or restrictive (39°C) temperature. For comparison, wild-type CHO stably transfected with Fc␥RIIA (FcR-CHO cells) were also analyzed. Tubulin, which is constitutively expressed in these cells in a temperature-independent manner, was used as a reference. When grown at 34°C, both ldlF and FcR-ldl express ⑀COP, although at levels that are 2-4 times lower than that found in FcR-CHO cells (Fig. 1A). A similar differential expression of ⑀COP at 34°C was reported earlier between ldlF and wild-type CHO cells (12). Incubation at the restrictive temperature for Յ18 h resulted in the complete disappearance of ⑀COP in ldlF and FcR-ldl cells but had no detectable effect on the expression of this protein in FcR-CHO cells (Fig. 1A), as found earlier for wild-type CHO cells (12).
The disappearance of ⑀COP was also apparent by analyzing the functional consequences of incubation at 39°C. As illustrated in Fig. 1B, i, the Golgi complex normally displays a tight juxtanuclear structure composed of cisternae and vesicles. Maintenance of this structure is known to depend on the continued function of COPI, which in turn requires the presence of ⑀COP (13,14). Degradation of ⑀COP in FcR-ldl cells was associated with dispersal of the juxtanuclear complex, resulting in a diffuse punctate staining of ␣Ϫmannosidase II-reactive vesicles ( Fig. 1B, iv). Dispersal of the Golgi complex was due to disappearance of ⑀COP and not to the temperature shift itself, since FcR-CHO cells were found to preserve their juxtanuclear Golgi cisternae after overnight incubation at 39°C (Fig. 1B, iii).
In contrast to the Golgi complex, incubating cells at 39°C did not appear to cause detectable changes to peroxisomes, mitochondria, and the ER. The punctate distribution of catalase, a marker of peroxisomes, was indistinguishable in cells incubated at 34 and 39°C (not shown), implying that peroxisomes are able to import and retain this soluble protein in their lumen. Furthermore, rhodamine-123 accumulated normally in the mitochondria of cells incubated at 39°C (Fig. 1C), indicating that the mitochondrial membrane potential was unaffected, which implies that the activity of the respiratory chain is normal. In addition, the reticulate morphology of the ER, revealed by immunostaining of calnexin, was similarly unaltered by prolonged incubation at the restrictive temperature (not illustrated). These results were consistent with ultrastructural analysis of thin sections by transmission EM, which showed no alterations in the structure or distribution of mitochondria or ER in cells treated at 39°C (100 sections from three different experiments; not illustrated). Together, these observations imply that treatment of the cells at the restrictive temperature for Յ15 h does not induce wholesale, nonspecific disorganization of the cellular ultrastructure and that only organelles dependent on COPI for their homeostasis, such as the Golgi apparatus, undergo visible alterations, as shown previously (11,13,14).
Effect of ⑀COP Depletion on Phagocytosis-Previously, it had been shown that ldlF cells stably transfected with Fc␥RIIA were able to internalize IgG-opsonized particles (17). Therefore, we next investigated whether COPI is required for phagocytosis. FcR-ldl cells were maintained at either 34 or 39°C and then exposed to IgG-opsonized SRBC to initiate phagocy-tosis. As illustrated in Fig. 2A and quantified in Fig. 2C, FcRldl cells maintained at the permissive temperature internalize SRBC effectively; phagocytosis occurred in upwards of 50% of the cells. This compares favorably with the phagocytosis efficiency of FcR-CHO cells (ϳ30%; Fig. 2C). The greater phagocytic ability of FcR-ldl cells is most likely attributable to clonal differences. Depletion of ⑀COP by overnight incubation at 39°C profoundly reduced the ability of FcR-ldl cells to perform phagocytosis. SRBC were observed in only 4 Ϯ 3% of the ⑀COPdepleted cells. By contrast, FcR-CHO cells internalized SRBC slightly more effectively following incubation at 39°C than when maintained at 34°C. These observations imply that the effect noted in FcR-ldl cells is not due to the temperature per se but instead that the presence of COPI is essential for optimal phagocytosis.
Consistent with the findings of Daro et al. (13) in ldl cells, the and in ldlF cells that were either untransfected or transfected stably with the human Fc␥RIIA receptor (FcR-ldl). The cells were either maintained at the permissive temperature (34°C) or preincubated for Յ18 h at the restrictive temperature (39°C), as indicated. B, effect of ⑀COP depletion on the morphology of the Golgi complex. FcR-CHO cells (i and iii) and FcR-ldl cells (ii and iv) were pretreated at 34°C (i and ii) or 39°C (iii and iv) as above and immunostained with antibodies to ␣-mannosidase II. C, mitochondrial membrane potential is intact in ⑀COP-containing (i) or -depleted (ii) FcR-ldl cells. The presence of an internally negative mitochondrial membrane potential was detected by the accumulation of rhodamine-123. Bar, 10 m. cells stably transfected with Fc receptors (FcR-ldl) were found to take up and recycle Tfn, albeit at reduced rates. As shown in Fig. 2D, the initial rate of uptake was Ͼ2-fold lower in ⑀COPdepleted cells, although the extent of uptake after 30 min was only ϳ30% lower. Also in agreement with Daro et al. (13), we found that the slower rate of Tfn uptake was compensated by a reduced rate of recycling (Fig. 2D, lanes 5-8), accounting for the nearly normal Tfn content at steady state. The reduced recycling of transferrin is consistent with the proposed role of COPI in traffic along the endocytic pathway.
Effect of ⑀COP Depletion on Cell Morphology-During the course of the phagocytosis experiments described above, a striking morphological difference between cells incubated in different conditions became apparent. When incubated at the restrictive temperature, FcR-ldl cells were found to become more rounded (Fig. 2, compare A and B). As before, the effect was attributable to the depletion of COPI and not to the incubation at 39°C, since FcR-CHO cells maintained their spread morphology at this temperature (not shown).
Further evidence that this effect was specifically related to loss of COPI was obtained by treating FcR-CHO and FcR-ldl cells (maintained at 34°C) with the fungal metabolite BFA. BFA associates with and blocks the activity of nucleotide exchange factors that regulate ARF, thereby preventing the association of coatomer subunits with membranes (23,24). Treatment with 100 M BFA for 30 -60 min induced rounding of both cell lines (not shown).
The cell rounding observed upon inactivation of COPI was associated with an apparent decrease in net surface area. This was determined by confocal microscopy, quantifying the amount of cell-associated FM1-43 (not shown), a dye that becomes fluorescent when it intercalates into the plasmalemma, thereby providing a measure of cell surface area (6,25). The decrease in ⑀COP-depleted cells was reproducible and statistically significant, but since fluorescence is not a simple function of the area, the change in plasmalemmal surface was not quantified precisely.
Several separate lines of evidence argue that rounding of COPI-deficient cells is not an indication of detrimental effects on cell function or viability. First, the cells remained impermeant to vital dyes and to the fluorescent dye FM1-43. Second, the cytosolic free calcium concentration, a sensitive measure of cell integrity and well being, was unaltered by depletion of ⑀COP (see below). Third, and as reported earlier for ldlF cells (13,14), FcR-ldl cells treated overnight at 39°C displayed endocytosis and recycling of transferrin (Fig. 2D). Fourth, the rate of incorporation of [ 35 S]methionine/cysteine into newly synthesized proteins was not significantly changed (not shown). Fifth, peroxisomes, the ER, and mitochondria in ⑀COP depleted cells resembled those of control FcR-ldl cells, and the mitochondrial potential appeared unaffected. Sixth, cells responded to the addition of stimuli with increased tyrosine phosphorylation and actin assembly (see below). Finally, the effects of incubation at 39°C were reversible. These findings imply that within the time period examined, suppression of COPI function does not cause generalized detrimental effects on the cell.
Effect of COPI Depletion on Receptor Expression and Function-In view of the apparent loss of surface area, the effects of COPI depletion on phagocytosis could result from alterations in Fc receptor expression, distribution, or function. This possibility was considered in the experiments illustrated in Fig. 3. At the permissive temperature, Fc␥RIIA receptors were clearly observable by immunostaining on the plasmalemma of FcR-ldl cells (Fig. 3A). Neither the distribution (Fig. 3B) nor the amount of receptors appeared to be altered following incubation at the restrictive temperature. Quantitation of receptors by flow cytometry indicated that similar levels of Fc␥RIIA were exposed after incubation at 34 or 39°C for 18 h in both FcR-CHO and FcR-ldl cells (Fig. 3E).
The ability of the receptors to engage their ligand was also assessed, quantifying the number of opsonized SRBC bound to the surface of the Fc␥RIIA-transfected cells. As illustrated by the scanning electron micrographs in Fig. 3, C and D, and summarized in Fig. 3E, SRBC bound readily to FcR-ldl cells both before and after depletion of ⑀COP by overnight incubation at 39°C. Comparable results were obtained with FcR-CHO cells (Fig. 3E).
The Effect of Brefeldin A on Phagocytosis by FcR-ldl and Macrophage Cells-Our results indicated that functional COPI was required for phagocytosis in FcR-ldl cells. These findings appear to be in conflict with earlier findings of Zhang et al. (26), who reported that in RAW 264.7 macrophages phagocytosis was insensitive to BFA. BFA is a specific inhibitor of GTP exchange factors of the ARF family of GTPases, which are essential for COPI function (23,24,27). We therefore tested the effect of BFA on phagocytosis in FcR-ldl cells and macrophages. The inhibitor induced a rapid dispersal of ␣-mannosidase II from a perinuclear distribution (Fig. 4A) to a diffuse intracellular appearance (Fig. 4B), resembling that found in ⑀COPdepleted cells. Despite the demonstrated effect of BFA on COPI function, phagocytosis was virtually unaffected when measured 2 h after the addition of the drug (Fig. 4C), consistent with earlier observations (26).
The primary difference between the two protocols used to inactivate COPI is the time involved in the development of the inhibitory effect. To resolve the apparent inconsistency, we compared the rate of disappearance of ⑀COP with the onset of inhibition of phagocytosis. The results are summarized in Fig.  4D. Significantly, the course of disappearance of ⑀COP preceded the complete ablation of phagocytosis. After 6 h at 39°C, when ⑀COP was no longer detectable, phagocytosis was dimin-ished but still clearly observable (approximately half-maximal; see Fig. 4D). Similarly, while no significant effects were noted at the early stages of BFA incubation, phagocytosis was inhibited by nearly 35% after 3 h and by 100% after 20 h (Fig. 4D). BFA also inhibited phagocytosis in professional phagocytes; particle ingestion was inhibited by 50% and by Ͼ90% in J774.1 and RAW cells incubated with the ARF antagonist for 6 and 12 h, respectively (not shown). Importantly, these data argue that a significant attenuation of phagocytosis can be observed at early time periods (4 -8 h) after COPI deficiency and further suggest that our observations are not due to generalized, nonspecific detrimental effects on cell function due to prolonged absence of COPI.
However, these findings imply that COPI is not directly involved as an essential component of the phagocytic process. Instead, loss of COPI function abolishes phagocytosis indirectly, perhaps by gradual depletion of an essential component required for the late stages of signal transduction, actin polymerization, and/or pseudopod extension. There is precedent for depletion of membrane components in ldlF cells, which were initially isolated on the basis of reduced ldl receptor expression at the restrictive temperature (10). Reduction of these essential components may conceivably occur by inhibition of the biosynthetic pathway when COPI is inactivated (11). Nonetheless, treatment of cells with cycloheximide for 24 h did not preclude phagocytosis, although protein synthesis was demonstrably arrested (not shown).
We also investigated if the prolonged absence of functional COPI was equally effective at blocking phagocytosis of smaller particles. RAW cells were treated with BFA for 8 -10 h and then allowed to internalize either SRBC, which have a diameter of ϳ4 m, or 0.8-m latex beads coated with IgG, the latter approaching the size of bacterial pathogens. At this time, phagocytosis of SRBC was blocked by ϳ65% Ϯ 10% relative to control cells. The uptake of 0.8-m beads was also clearly inhibited by depletion of ⑀COP, although the inhibition was significantly smaller (40 Ϯ 9%; Fig. 4E).
Effect of ⑀COP Depletion on Receptor Signaling-The preceding results indicate that the inhibition of phagocytosis induced by depletion of COPI cannot be attributed to a decrease in the number of Fc␥RIIA receptors or in their ability to ligate SRBC. Instead, the effect of COPI is probably exerted at a later stage, affecting signal transduction or downstream effectors. These possibilities were analyzed next. Clustering of Fc␥ receptors upon interaction with IgG-opsonized particles leads to the activation of tyrosine kinases and to accumulation of F-actin around nascent phagosomes (18,28,29). The occurrence of these steps was tested in COPI-deficient cells. As shown in Fig.  5, A and B, accumulation of tyrosine-phosphorylated proteins was detected at sites of particle attachment to FcR-ldl cells pretreated at 34°C (not shown) and 39°C. Moreover, activation of downstream effectors was also readily observable. Accumulation of F-actin in the phagosomal cup, which is evident in the phagocytosis-competent FcR-ldl cells kept at 34°C (Fig.  5D), was in fact more pronounced in cells shifted to 39°C (Fig.  5E). In fact, scanning EM revealed that SRBC were often found tightly attached to "pedestals" that protruded from the ⑀COPdepleted FcR-ldl cells (Fig. 5C). F-actin was concentrated at the junction between the opsonized particles and these pedestals (Fig. 5E).
We also investigated whether increase in cytoplasmic free Ca 2ϩ concentration ([Ca 2ϩ ] i ) was normal in COPI-deficient cells. An increase in [Ca 2ϩ ] i mediated by activation of phospholipase C and release of Ca 2ϩ from intracellular stores is one of the earliest consequences of Fc receptor ligation (30,31). Although not essential for phagocytosis, this [Ca 2ϩ ] i transient is invariably associated with receptor clustering and activation in professional phagocytes and nonprofessional engineered phagocytes (32,33). In FcR-ldl cells maintained at 34°C, [Ca 2ϩ ] i spikes were similarly triggered by binding of IgG-opsonized particles (Fig. 5F, squares). Such spikes were absent in ldl cells lacking Fc receptors (Fig. 5F, circles). Importantly, [Ca 2ϩ ] i transients were elicited by particles in FcR-ldl cells pretreated at 39°C (Fig. 5G). Similar observations were obtained with FcR-CHO cells preincubated at both temperatures (not illustrated). In contrast, [Ca 2ϩ ] i transients were not observed upon the addition of opsonized particles to the parental wild-type CHO (not shown) and ldlF cells (Fig. 5F, circles), which do not express Fc␥RIIA. These findings imply that COPI is not essential for this early response and confirm the notion that incubation at 39°C does not have generalized detrimental effects on cells.
Effect of ⑀COP Depletion on Integrin Expression-Integrins are important in cell spreading (34) and have been reported to act synergistically with Fc receptors to promote phagocytosis (35)(36)(37). Since COPI-deficient cells were morphologically altered (i.e. rounded), we hypothesized that integrin function was compromised in these cells. We therefore assessed the presence and distribution of ␤ 1 -integrins. We found that focal contacts were formed in FcR-ldl cells incubated at both 34 and 39°C (Fig. 6, A and C). Actin stress fibers were often seen to abut these focal adhesions under both conditions, although the length of the stress fibers was reduced in ⑀COP-deficient cells (Fig. 6, B and D). The total content of ␤-integrins in unstimulated FcR-ldl cells was compared at 34 or 39°C using flow cytometry of suspended cells and also by quantitative fluorescence microscopy of adherent cells. No significant difference was found using the former method, and only a small decline was observed microscopically (Fig. 6E). Similarly, the F-actin content was unaffected by incubation at 39°C.
Ultrastructural Analysis of Phagocytosis in COPI-containing or -depleted Cells-We next used transmission EM to compare the ultrastructure of normal and COPI-deficient cells during the course of phagocytosis. FcR-ldl cells preincubated at either 34 or 39°C were exposed briefly to opsonized SRBC, fixed, and processed for EM. As described earlier for both professional (38) and engineered phagocytes (18), the early stages of phagocytosis in cells with functional COPI are characterized by extension of elaborate pseudopods that surround and ultimately engulf the particles (Fig. 7, A and B). In these cells, SRBC located well within the cytosol (often near the nucleus; Fig. 7B) were seen frequently, probably representing complete (sealed) phagosomes. Such structures were virtually never observed in COPI-deficient cells (e.g. Fig. 7, C and D). Nevertheless, adherent SRBC were found routinely on the surface of the COPI-depleted cells (Fig. 7, C and D). Of note, the pseudopods extending from these were much shorter, if present at all. Often, the SRBC were apposed to flat regions of the plasma membrane. These sites of apposition were sometimes raised above the surface of the cells, corresponding to the "pedestals" noted in Fig. 5. In a significant proportion of cells that were pretreated at 39°C, clear vesicular structures were found to accumulate beneath the bound SBRC (small arrowheads in Fig. 7D and inset). These structures were not found at 34°C.
Focal Secretion during Phagocytosis in COPI-deficient Cells-Pseudopod extension during Fc␥R-mediated phagocytosis is thought to depend on focal exocytosis of endomembranes (3,6). In fact, focal exocytosis of VAMP3-containing secretory vesicles was shown recently to occur in phagocytic cups (9). Because prolonged depletion of COPI appears to limit the extension of pseudopods around opsonized particles, we postulated that inhibition of phagocytosis may result from the impaired delivery of endomembrane vesicles to the site of phagosome formation. To test this notion, the distribution and mobilization of VAMP3 was analyzed in RAW macrophages by transient transfection of a VAMP3-GFP chimeric construct. The predicted transmembrane topology of the chimera is shown in Fig. 8A. The construct retains the cytosolic domain of VAMP3, which dictates its intracellular targeting, while the GFP module is in the lumen of endosomes and appears on the exofacial side of the plasma membrane upon exocytosis. As a result, the occurrence of exocytosis of VAMP3-containing vesicles can be detected by staining intact (nonpermeabilized) cells with anti-GFP antibodies as described (9).
VAMP3-GFP was found both in endomembrane vesicles and on the surface of RAW cells performing phagocytosis (Fig. 8,  B-D). Importantly, as reported in FcR-CHO cells (8), VAMP3-GFP accumulated in the membrane of nascent phagosomes (Fig. 8C). The accumulated VAMP3-GFP was inserted on the plasma membrane and not simply concentrated in vesicular form below the surface. This was confirmed by the observation that the phagosomal VAMP3-GFP was accessible to anti-GFP antibodies in intact cells (Fig. 8D). In contrast, VAMP3-GFP was often intracellular in cells treated with BFA, distributed in tubules or cisternae (Fig. 8E). Importantly, little accumulation was generally observed at the sites of aborted phagocytosis where particles attached (Fig. 8, F and G). These results suggest that COPI deficiency precludes the mobilization of a VAMP3-expressing endomembrane pool necessary for pseudopod extension during phagocytosis. DISCUSSION The central observation described in this paper is that phagocytosis was impaired when COPI function was inhibited. The requirement for COPI was documented by two independent approaches: the use of a mutant cell line expressing a thermolabile form of ⑀COP and the inhibition of ARF with BFA. It is noteworthy that within the time frame studied, COPI inactivation did not produce nonspecific deleterious effects on cell viability or function. Thus, COPI-deficient cells accumulated Tfn, had seemingly normal mitochondrial potential, and synthesized proteins at nearly normal rates, and the morphology of the mitochondria, ER, and peroxisomes was unaltered. More importantly, the cells retained the ability to bind SRBC, triggered cytosolic calcium changes, and induced the focal assembly of F-actin at sites of frustrated phagocytosis. We believe, instead, that COPI plays a specific and discrete role in the events leading to phagocytosis. A recent study demonstrated that interference with ARF6 function depressed phagocytosis (26). However, it is unlikely that our results are related to ARF6, because this isoform of ARF has not been linked to COPI function and, in addition, the exchange factors that control ARF6 are thought to be BFA-insensitive (39).
The inhibition of phagocytosis lagged behind the disappearance of ⑀COP or the inactivation of ARF nucleotide exchange factors. Near normal phagocytosis was seen for up to 2 h after the addition of BFA, which enters the cells and interferes with ARF function rapidly, and almost 50% of the phagocytic rate remained immediately after ⑀COP was virtually eliminated. For these reasons, we felt that COPI was necessary for the maintenance and/or localization of a critical pool of elements of the phagocytic machinery, instead of playing a direct role in phagocytosis. We initially considered whether the plasmalemmal Fc receptors were present and responsive. Direct assessment by flow cytometry confirmed the presence of nearly normal quantities of surface receptors, and their ability to signal upon cross-linking was verified by three different observations: (a) the accumulation of phosphotyrosine in the phagocytic cup (Fig. 5, A and B); (b) the initiation of calcium spikes (Fig. 5, F and G), and (c) the induction of F-actin polymerization (Fig. 5, D and E). Hence, neither absence nor inactivation of Fc receptors can account for the inhibition of phagocytosis.
Integrins have been reported to act synergistically with Fc receptors to promote phagocytosis (35)(36)(37). A possible role of integrins was also suggested by morphological changes undergone by cells following COPI inactivation. The rounding of the cells was consistent with diminished integrin activity, inasmuch as integrins are important in cell spreading (34). However, the expression of ␤ 1 -integrins quantified by flow cytometry or confocal microscopy was affected only modestly, if at all, by ⑀COP depletion, and focal adhesions were clearly visible in adherent cells grown at 39°C (Fig. 6, A and C). Actin stress fibers were seen to abut these focal adhesions, and the total content of F-actin of unstimulated FcR-ldl cells was essentially identical whether grown at 34 or 39°C (Fig. 6, B and D). Therefore, gross alterations in integrin function are unlikely to account for the inhibition of phagocytosis associated with COPI depletion.
The occurrence of seemingly normal signaling after engagement of the Fc receptors suggests that the inhibitory effect of COPI depletion is exerted downstream. Some insight as to the possible site of action was gained from ultrastructural analysis. Two features are noteworthy. First, unlike the cells with normal COPI, which extended long and convoluted pseudopodia, those lacking COPI extended only short pseudopods or none at all (Fig. 7). Second, clear vesicles were frequently seen to accumulate under the sites of frustrated phagocytosis in COPIdeficient cells (Fig. 7D). These observations resemble findings of Cox et al. (6) and Lennartz et al. (5). In the former study, actin polymerization but not pseudopod extension was seen when phosphatidylinositol-3Ј-kinase was inhibited. In the latter, impairment of phopholipase A 2 also limited the extension of pseudopods and in addition promoted the accumulation of small electron lucent vesicles underneath the sites of particle attachment. In both cases, it was suggested that failure to perform exocytosis was responsible for the inability of the cells to extend pseudopods. In accordance with this interpretation, there is mounting evidence that secretion of endomembranes is involved in pseudopod extension (see Introduction). It is therefore conceivable that delivery of vesicles required for pseudopod elongation is impaired by neutralization of COPI and ARFs.
Recycling endosomes are likely to be one of the endomembrane pools affected by COPI. These membranes have been linked to phagocytosis by two observations: (a) impairment of Rab11, a GTPase that regulates fusion of recycling endosomes (38,39), attenuated phagocytosis (40) and (b) exocytic vesicles containing VAMP3 were locally inserted into phagocytic cups (9). Interestingly, treatment with BFA and genetic ablation of COPI lead to defects in transferrin receptor recycling, implicating COPI in the recycling pathway ( Fig. 2D and Refs. 13, 41, and 42).
These findings prompted us to analyze the effects of COPI inactivation on the distribution and mobilization of VAMP3containing vesicles. We found that treatment with BFA drastically reduced the amount of VAMP3-GFP integrated into the plasma membrane of RAW cells. Of particular significance, the localized secretion of VAMP3-GFP-containing vesicles at the sites of phagocytosis observed in untreated cells was largely eliminated by BFA (Fig. 8, F and G). We concluded that COPI is required to maintain a pool of endomembrane vesicles that need to insert into the plasma membrane to promote the extension of pseudopods around the phagocytic particle.
In summary, our findings and those of others suggest that COPI and ARF are important for the homeostasis of the endocytic recycling compartment. Recycling endosomes, in turn, seem to provide at least part of the membrane required for pseudopod extension. Last, as in the case of wortmannintreated cells, our observations dissociate the polymerization of actin from the extension of pseudopods during phagocytosis, suggesting that multiple events are required for successful pseudopodial elongation.