FcγR-Mediated Phagocytosis Stimulates Localized Pinocytosis in Human Neutrophils¹

Professional phagocytes, comprised of monocytes, macrophages, and neutrophils, are key to the innate immune defense system and, by removing apoptotic bodies, also contribute to tissue remodeling. Neutrophils often mount the initial response to infection because of their rapid chemotactic response toward bacterial peptides and inflammatory cytokines. Upon reaching the infected area, neutrophils curb pathogen activity by ingestion of microorganisms, free radical synthesis, cytokine release, and degranulation (1, 2).

The antimicrobial responses of phagocytes are triggered by surface receptors that recognize either conserved patterns on the surface of microorganisms or opsonins that coat them. The latter receptors include FcγR, which are responsible for the phagocytosis of IgG-opsonized particles (1, 3, 4). Particle engulfment is triggered by FcγR clustering, which induces localized activation of Src family and Syk tyrosine kinases at the phagocytic cup. These initial events are followed by stimulation of phosphatidylinositol 3-kinase (PI3K)⁵ and phospholipase Cγ (4, 5), which hydrolyses phosphatidylinositol-4,5-bisphosphate into diacylglycerol and inositol-1,4,5-trisphosphate. The latter mediator is responsible for the rise in the free cytosolic Ca⁺² concentration ([Ca⁺²]_i) observed during FcγR-mediated phagocytosis (6, 7, 8). Rac and Cdc42, members of the Rho family of small GTPases, are then activated and coordinate actin remodeling at the sites of phagocytosis, culminating in the engulfment of the microbe into an intracellular vacuole or phagosome (9, 10, 11, 12).

In neutrophils, FcγR signaling also causes degranulation. Neutrophils possess at least four types of secretory organelles: primary (azurophilic), secondary (specific), and tertiary (gelatinase) granules and secretory vesicles (1, 13). Primary granules are enriched in lysosomal hydrolases and myeloperoxidase, and they can be identified by the presence of CD63 on their membrane. Secondary granules contain lactoferrin and lysozyme and express CD66b on their membrane. Tertiary granules contain gelatinase, while secretory vesicles are rich in albumin and alkaline phosphatase (1). These organelles do not necessarily undergo secretion simultaneously, since the signals leading to their exocytosis differ in type and/or activation threshold (14, 15).

Exocytosis of multiple types of secretory organelles contributes additional surface area to the target membrane. In other systems that undergo similar acute and vigorous secretion, such as chromaffin cells, the net area of the membrane is maintained approximately constant by the concomitant activation of endocytosis (16, 17, 18, 19). Endocytosis also serves to retrieve components of the secretory machinery to be reused in subsequent rounds of stimulation. For these reasons, endocytosis (pinocytosis) is also likely to be activated during FcγR-mediated phagocytosis. It is noteworthy, however, that, unlike chromaffin cells, phagocytes are capable of focal secretion during particle engulfment, targeting the secreted material to the area of the plasma membrane where phagosomes are being generated or to the lumen of formed phagosomes (20, 21). It is therefore conceivable that localized signals may, in fact, trigger focal pinocytosis during phagocytosis. Indeed, clathrin, dynamin, and amphiphysin were detected around the phagosomal cup (22, 23, 24). To test these hypotheses we studied whether pinocytosis is, in fact, activated during FcγR-mediated phagocytosis and, if so, whether it occurs locally at or near nascent phagosomes. In addition, we analyzed the signals leading to membrane retrieval.

Cytochalasin B, colchicine, PMA, thapsigargin, EGTA, fibronectin, fMLP, and human IgG were obtained from Sigma-Aldrich (St. Louis, MO). Pefabloc SC was purchased from Roche (Indianapolis, IN). Ionomycin and wortmannin were obtained from Calbiochem (La Jolla, CA). Zymosan, Lucifer Yellow (LY), Indo-1/AM, and BAPTA-AM were obtiained from Molecular Probes (Eugene, OR). Latex beads were purchased from Bangs Laboratories (Carmel, IN). Mouse anti-CD63 and anti-CD66b Abs were obtained from Caltag (San Francisco, CA), Serotec (Oxford, U.K.), and the Hybridoma Developmental Studies Bank (Iowa City, IA). Fluorochrome-conjugated anti-human and anti-mouse Abs were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA) and Molecular Probes.

Human neutrophils were isolated from heparinized blood from healthy donors by Ficoll-Hypaque gradient centrifugation as previously described (25, 26) or using the 1-Step Polymorph Isolation kit (Accurate Chemical and Scientific, Westbury, NY). Contaminating RBC were removed by NH₄Cl lysis when required, and neutrophils were then counted using a Coulter counter (model ZM; Hialeah, FL). Neutrophils were maintained in either HEPES-buffered RPMI or complete HBSS at room temperature until use, within 5 h of isolation. When required, cells were washed with Ca²⁺-free HBSS supplemented with 1 mM MgCl₂. Cytoplasts and karyoplasts were prepared as described previously (27).

Zymosan and latex beads were opsonized with 1–2 mg/ml human IgG for at least 1 h and were washed three times with PBS. Particles were then added to adherent or suspended neutrophils to initiate phagocytosis. When in suspension, cells and particles were rapidly cosedimented by centrifugation to synchronize phagocytosis. To observe fluid phase endocytosis (pinocytosis) during particle ingestion, phagocytosis proceeded in the presence of 1 mg/ml LY for the indicated times and was arrested by paraformaldehyde fixation. The phagocytic index was quantified by counting the number of internalized particles per 100 cells. Pinocytosis was quantified by flow cytometry or by measuring the endocytic index, defined as the number of neutrophils with at least three distinct LY-labeled vesicles. During quantification of pinocytosis, early time points were employed to minimize the contribution of phagosome-derived vesicles.

Following the desired treatment, neutrophils were fixed with 4% paraformaldehyde for 15 min, and extracellular particles were identified by staining with Cy3- or Cy5-conjugated anti-human Abs for 30 min at 1/1000. To stain for total CD63 and CD66b, cells, cytoplasts, and karyoplasts were permeabilized with 0.1% Triton X-100 for 10 min, followed by blocking for 1 h with 5% donkey serum and incubation for 1 h with anti-CD63 or anti-CD66b mAbs diluted to 1/100 and 1/200, respectively. After washing, the cells were stained using the respective secondary Abs for 1 h, washed, and mounted using mounting medium (DAKO, Carpenteria, CA). Where specified, permeabilization was omitted from the above protocol to detect exofacial CD63 or CD66b. Samples were analyzed using an epifluorescence microscope (model DM-IRB; Leica, Rockleigh, NJ) or a LSM 510 laser scanning confocal microscope (Zeiss, New York, NY) equipped with a ×100 oil immersion objective. Images were prepared using Adobe PhotoShop 6.0 and Illustrator 10.0 (Adobe Systems, San Jose, CA).

Endocytic uptake of LY and secretion of CD63/CD66b were quantified using a FACScan flow cytometer (BD Biosciences, Mountain View, CA). Preparation of samples was performed as described above, but cells were diluted in PBS and maintained in suspension. For every sample, at least 10,000 ungated cells were counted. Selection of the population of interest was performed after the acquisition of raw data using LYSIS II analysis software as described previously (28).

[Ca²⁺]_i was quantified by spectrofluorometry using Indo-1 as previously described (26, 28). Briefly, neutrophils were loaded with 1 μM Indo-1/AM for 30 min at 37°C, washed, and maintained in HCO₃⁻-free, Ca²⁺-free medium (140 mM NaCl, 5 mM KCl, 10 mM glucose, 1 mM MgCl₂, and 10 mM HEPES, pH 7.4). Where noted, 1 mM EGTA or 1 mM CaCl₂ was added to the medium. Calibration of [Ca²⁺]_i was accomplished by adding 10 μM ionomycin, followed by 2 mM CaCl₂ to attain maximal fluorescence, and subsequently 2 mM MnCl₂ to quench Indo-1 fluorescence for determination of autofluorescence and scattering.

Intracellular calcium depletion was accomplished by pretreatment of cells with 100 nM thapsigargin or 1 μM ionomycin in nominally Ca²⁺-free medium containing 1 mM EGTA for 25 min at 37°C before stimulation. Alternatively, cells were pretreated with 10 μM BAPTA/AM in Ca²⁺-free medium containing 1 mM EGTA for 30 min at 37°C before stimulation.

FcγR-mediated phagocytosis was shown to induce secretion that is preferentially targeted to the phagocytic cup in macrophages and neutrophils (20, 21). We therefore analyzed whether localized retrieval of membranes also occurs during phagocytosis, using human neutrophils as a model system. When unstimulated, these cells have a remarkably low rate of spontaneous pinocytosis, facilitating the detection of stimulation-induced events. Indeed, when resting neutrophils were incubated with the fluid-phase marker LY for 15 min, very few neutrophils (<5%) were visibly labeled (not illustrated). Upon exposure to IgG-opsonized beads, distinct LY-containing vesicles were noticeable in many of the cells (Fig. 1). Note that pinocytic vesicles were present in cells associated with beads (arrows in Fig. 1), but not in adjacent cells that failed to bind beads. Pinocytic events were observed as early as 30 s during synchronized phagocytosis and seemed to precede sealing of the phagosome. As shown in Fig. 1, at the time when pinocytosis was clearly discernible (A) the opsonized particles were still accessible to Abs added extracellularly (B and inset), implying that phagocytosis was still in progress. It is also noteworthy that at the early stages the endocytic vesicles accumulated in the vicinity of nascent phagosomes, where they were probably formed. Pinocytosis was also stimulated by IgG-opsonized and unopsonized zymosan (not illustrated). The latter suggests that mannose- and/or β-glucan receptor-mediated phagocytosis can likewise induce pinocytosis.

As phagosomes sealed, indicated by the inaccessibility of the particles to Abs (Fig. 1, D and inset), the number of LY-containing vesicles increased, and they distributed more widely throughout the cell (Fig. 1 C). The continued formation of labeled vesicles may represent ongoing pinocytic activity at the cell surface, but may alternatively result from fission of membranes from sealed, LY-containing phagosomes (29, 30). Accordingly, LY trapped along with the particles disappeared gradually as the phagosomes matured.

Phagocytosis of IgG-opsonized particles is stringently dependent on remodeling of the actin cytoskeleton and is obliterated by treatment with cytochalasins (5). On the other hand, the secretion that accompanies receptor activation in phagocytic cells is, in fact, potentiated by cytoskeletal disruption (21). These divergent effects of cytoskeletal inhibitors provided a means of distinguishing whether endocytosis was tightly linked to either the particle engulfment or secretory processes.

Neutrophils were pretreated with cytochalasin B or D and then exposed to IgG-opsonized beads or zymosan in the presence of LY. As expected, capping the barbed end of filamentous actin filaments with cytochalasin virtually eliminated phagocytosis (Fig. 2). Remarkably, the pinocytosis induced by interaction with opsonized particles not only persisted, but was, in fact, more noticeable than in control cells (Fig. 2, A–D). Similar results were obtained whether opsonized latex (Fig. 2, A and B) or opsonized zymosan (Fig. 2, C and D) was used as the phagocytic target. It is noteworthy that despite the failure of the cells to ingest particles, the pinocytic events occurred preferentially in the immediate vicinity of the adherent particles. These observations imply that closure of the phagosomes is not essential for receptor-induced pinocytosis and that actin assembly is not involved in targeting the pinocytic events.

The resistance, indeed the potentiation, of pinocytosis observed in cytochalasin-treated cells is consistent with the idea that granule secretion may be associated with the increased fluid phase uptake. Because the polarized secretion of primary granules in neutrophils requires an intact microtubular network, we tested the effects of colchicine on particle-induced pinocytosis (21). We found that the efficiency of phagocytosis and pinocytosis decreased somewhat in colchicine-treated cells (Fig. 2, G and H). More importantly, although no systematic quantitation was attempted, it was clear that the pinocytic vesicles no longer accumulated to the same extent in the vicinity of the adherent opsonized particles (Fig. 2, E and F). Jointly, these observations indicate that while actin-dependent particle engulfment is not required for focal pinocytosis, secretion of granules may play a role in the induction of pinocytosis.

In neutrophils, granule secretion can be elicited by artificially increasing [Ca²⁺]_i, bypassing the activation of surface receptors. This strategy was used to further explore the relationship between secretion and pinocytosis. As shown in Fig. 3,A, [Ca²⁺]_i could be readily increased beyond the resting physiological level by addition of ionomycin, a Ca²⁺ ionophore, or thapsigargin, an inhibitor of sarco(endo)plasmic reticulum calcium ATPase-type Ca²⁺-ATPases (Fig. 3,A). The levels attained suffice to induce exocytosis of all secretory granules and vesicles of human neutrophils (14, 15) (our unpublished observations). Addition of ionomycin (Fig. 3,B) and thapsigargin (Fig. 3 C) also induced a remarkable burst of fluid phase endocytosis in >90% and ≥85% of the cells, respectively. Of note, in both instances the LY-containing vesicles were homogeneously dispersed throughout the cell.

Secretion in neutrophils is markedly inhibited by antagonists of PI3K (31, 32, 33). On the other hand, most endocytic processes are either insensitive or only modestly affected by inhibition of PI3K (34, 35). This enabled us to test the causal relationship between these events. As illustrated in Fig. 3 E, pretreatment of the cells with 100 nM wortmannin greatly depressed the formation of LY-containing vesicles in cells stimulated with IgG-coated particles. Wortmannin was also a powerful antagonist of ionomycin-dependent pinocytosis (not shown). These effects are unlikely to result from a direct impairment of pinocytosis and could instead be an indirect result of the inhibition of secretion, which may be a necessary precursor to the stimulation of fluid phase uptake. Supporting this idea, wortmannin was shown to block ionomycin-induced exocytosis in pituitary gonadotrophs (36). Nevertheless, a direct inhibitory effect of wortmannin on pinocytosis cannot be discounted, since in some systems fluid phase uptake was reportedly inhibited by the PI3K antagonist (37, 38, 39).

While suggestive of a relationship between secretion and pinocytosis, the evidence provided by wortmannin and calcium ionophores is circumstantial. We therefore sought an approach to more directly test the nature of the relationship between these processes. To this end, we used a preparation of enucleated and degranulated neutrophils, originally developed by Roos et al. (27). Degranulated cell fragments, called cytoplasts, can be obtained by sedimentation of cells through a discontinuous Ficoll density gradient, which induces fission of the cells in two components: a dense karyoplast that contains the nucleus and secretory granules, and a lighter cytoplast fraction that is enriched in cytosol and light membranes, including the plasmalemma. The process is conservative, so that no cellular components are lost, and remarkably the cytoplasts retain the ability to perform phagocytosis and to mount a respiratory burst (27). The distribution of primary and secondary granules in cytoplasts and karyoplasts is compared with that of intact neutrophils in Fig. 4. As expected, both primary (CD63) and secondary granule markers (CD66b) are abundant in intact cells and inside karyoplasts. By contrast, no CD63 was detectable in cytoplasts, and CD66b was only detectable in the limiting membrane of a fraction of the cytoplasts. The appearance of CD66b on the membrane is indicative of some degranulation during the centrifugation procedure.

We proceeded to test the uptake of LY in cytoplasts. Like intact cells, unstimulated cytoplasts have very low rates of fluid phase uptake; we were unable to detect pinocytosis even after 15 min of incubation with LY (Fig. 5, A and B). In accordance with the results reported by Roos et al. (27), we found that cytoplasts were capable of phagocytosis (Fig. 5 D). Opsonized yeast (zymosan) particles were used for these experiments because they are porous and capable of trapping fluid phase markers. Indeed, when LY was present at the time of phagocytosis but removed thereafter, the probe was found trapped in the phagosome, confirming that sealing of the phagocytic vacuole had occurred. Importantly, pinocytic vesicles were not found in cytoplasts up to 10 min after phagocytosis.

These results suggest that the pinocytosis elicited by FcR cross-linking requires the presence and probably the secretion of granules. However, it is possible that the cytoplast isolation procedure may have directly impaired their pinocytic ability. This was tested using PMA, an activator of protein kinase C that is a potent activator of endocytosis in a variety of cells, including phagocytes (40, 41, 42). As illustrated in Fig. 5, E and F, PMA effectively induced LY uptake in cytoplasts. Together, these observations point to an essential role of degranulation in the stimulation of pinocytosis by IgG-opsonized particles. Because LY persisted for extended periods inside the cytoplast phagosome, we conclude that clearance of soluble phagosomal contents also requires prior fusion with secretory organelles and/or occurs by “kiss-and-run” with such organelles (30).

Several studies have demonstrated that elevated [Ca²⁺]_i is required for secretion during neutrophil stimulation (21, 43, 44). Moreover, detailed analysis of the [Ca²⁺]_i dependence of secretion has revealed that the threshold of activation of individual granule types varies in the order: primary granules secondary granules tertiary granules secretory vesicles (14, 15). We took advantage of the known [Ca²⁺]_i dependence of exocytosis to verify the relationship between secretion and the induction of pinocytosis and to try to identify the granule types involved.

As shown in Fig. 6,A, when neutrophils suspended in Ca²⁺-free medium were treated with ionomycin or thapsigargin, they underwent a transient increase in [Ca²⁺]_i, attributable to Ca²⁺ release from internal stores, followed by extrusion across the plasmalemma. In both instances, [Ca²⁺]_i had returned to baseline within 5 min, implying depletion of the mobilizable Ca²⁺ stores. Accordingly, subsequent stimulation of FcR (Fig. 6,A, arrowhead) failed to produce any detectable changes in [Ca²⁺]_i, which contrasts with the sharp [Ca²⁺]_i peak induced by aggregation of FcR in Ca²⁺ replete cells (Fig. 6 A, upper right trace).

Prior depletion of Ca²⁺ stores decreased the ability of opsonized particles to induce primary granule secretion (Fig. 6 B). Surface exposure of CD63 in response to FcR clustering assessed by flow cytometry was reduced by ≥65%. An even more pronounced inhibition was obtained in cells loaded with the Ca²⁺-buffering agent BAPTA.

Secondary granule secretion assessed by surface exposure of CD66b was affected by the Ca²⁺ depletion manipulations in a different manner (Fig. 6,C). First, pretreatment with thapsigargin alone sufficed to stimulate exocytosis, and an even larger response was elicited by ionomycin. This secretion occurred in response to the transient [Ca²⁺]_i increase triggered by the Ca²⁺-mobilizing agents (see Fig. 6,A). In accordance with this interpretation, no such effect was induced by BAPTA. These findings are in good agreement with the lower [Ca²⁺]_i threshold for activation of secondary granules (15, 45, 46). Subsequent stimulation of the depleted cells with opsonized particles produced an additional stimulation that, although reduced, brought the total secretion of CD66b to levels similar to or higher than those recorded in control cells (Fig. 6 C).

Despite extensive secretion of CD66b during ionomycin- or thapsigargin-mediated Ca²⁺ depletion (Fig. 6,C), the rate of pinocytosis in such cells was virtually unaffected when particulate stimuli were omitted (Fig. 6,D). However, subsequent addition of opsonized particles to Ca²⁺-depleted cells induced a sizable increase in pinocytosis, although the maximal rates attained were lower than in Ca²⁺-replete cells (Fig. 6 D). Together, these results suggest that pinocytosis correlates well with the secretion of primary, but not secondary, granules.

Our results demonstrate that FcγR-mediated phagocytosis signals pinocytic uptake at phagocytic sites. The stimulation of pinocytosis occurs before and independently of phagosome formation, since 1) vesicles trapping LY were clearly discernible in cells with unsealed phagocytic cups; and 2) pretreatment of cells with cytochalasin abolished phagocytosis, yet greatly stimulated pinosome formation. Therefore, while fission of these vesicles may be akin to that mediating phagosome maturation, the phenomenon reported here clearly precedes phagosome sealing and remodeling.

While not requiring completion of phagocytosis, LY-stained endosomes were nevertheless formed predominantly on or very near the patch of membrane juxtaposed to the opsonized particle. These results are consistent with observations that clathrin, amphiphysin II, and dynamin-2 localize to phagocytic cups (22, 23, 24). Moreover, when exposed to soluble immune complexes, FcγR undergo receptor-mediated endocytosis, a key process in Ag processing and in the genesis of inflammation (3, 47). By analogy, it is conceivable that cross-linking of FcγR by the opsonized particles initiates receptor-mediated endocytosis. However, this would require detachment of the IgG from the opsonized particle or disengagement of the receptor-ligand complex, which entails cessation of signaling. We regard this mechanism as improbable because, unlike receptor-mediated endocytosis, pinosome formation required elevation of cytosolic calcium and was sensitive to inhibitors of PI3K and because it was absent in cytoplasts.

Instead, our results suggest that pinocytosis was coupled to the occurrence of exocytosis. In accordance with this idea, stimulation of secretion with calcium ionophore promoted extensive pinosome formation. Moreover, the enhancement of pinocytosis noted in cells treated with cytochalasin is reminiscent of the stimulation of secretion that this drug induces in neutrophils (48). Our results, in addition, point to a central role of primary (lysosomal) granules in the induction of pinocytosis. Briefly, the calcium sensitivity profile (Fig. 6) and the preferential occurrence of pinocytosis in the immediate vicinity of the phagosomal cup (Figs. 1 and 2) closely parallel the established behavior of primary granules (14, 21). Of note, Fittschen and Henson (49) previously reported that endocytosis can also be triggered in neutrophils by chemotactic peptides and that it correlates with primary granule secretion.

The coincident occurrence of fluid phase endocytosis and secretion may reflect parallel, yet independent, events, which may share common signaling elements and are therefore similarly sensitive to pharmacological interventions. On the other hand, the events may be sequential and causally related. We believe that pinocytosis is at least partly dependent on prior secretion, to the extent that degranulated cytoplasts failed to form pinocytic vesicles. A similar consecutive and causal relationship between secretion and endocytosis has been postulated for neurons and endocrine cells (16, 17, 18, 50). It is currently unclear whether soluble contents or membrane-associated components of the secretory granules are the factors that prompt endocytosis. Transmembrane proteins of the granules may serve as nucleation sites for the assembly of endocytic coats, such as clathrin. On the other hand, proteases or other enzymes released from the granules may induce pinocytosis by cleaving exofacial membrane components. In this regard, protease inhibitors have been reported by several authors to inhibit phagocytosis (Refs. 48, 49, 50 and our own unpublished observations using Pefabloc).

What is the functional purpose of pinocytosis during FcγR-mediated phagocytosis? Pinocytosis may have a role in recycling membrane components such as soluble N-ethylmaleimide sensitive factor attachment receptors for use in subsequent rounds of secretion. While this may not be a critical response in neutrophils, which have a short biological half-life, it may play an important role in the case of macrophages. Alternatively, pinocytosis during FcγR-mediated phagocytosis may participate in the initiation of the inflammatory response or in Ag processing and presentation (3, 51). The latter is a critical aspect of macrophage function (52) and is also observed in neutrophils treated with GM-CSF, IL-3, or IFN-γ, which express MHC class II and can activate T cells both in vitro and in vivo (53, 54, 55, 56, 57). Lastly, it is possible that the membrane fission observed during phagocytosis represents an early stage of phagosome maturation. This premature remodeling would be exacerbated when phagocytosis is frustrated in cytochalasin-treated cells. Indeed, a proportion of the LY-containing vesicles detected in our experiments may have originated during the maturation of sealed phagosomes. Both clathrin and coatomer protein I, which have been shown to contribute to phagosomal recycling (58, 59), could contribute to budding of the observed vesicles. Possible additional mechanisms include caveolae-like structures or other membrane coats perhaps related to sorting nexins (60, 61). Alternatively, fluid may have exchanged during the process of “kiss-and-run” (30), whereby secretory organelles, probably including primary granules, would transiently fuse with the phagosomal membrane.

In summary, we have described the induction of focal pinocytosis at sites of phagosome formation. Such endocytosis often precedes and is independent of phagosome sealing and correlates with the localized secretion of primary granules. The signals that trigger pinocytosis may be generated by cross-linking of FcγR, but components delivered to the membrane by exocytosis also appear to be essential, to the extent that cytoplasts are capable of phagocytosis, yet fail to activate pinocytosis. The functional significance of the accelerated pinocytic uptake remains to be defined, but a role in the early stages of phagosome maturation appears likely. In this regard, it would be of interest to monitor phagosomal maturation in cytoplasts, where the initial fission events appear to be lacking.