|Home | About | Journals | Submit | Contact Us | Français|
Actin waves that travel on the planar membrane of a substrate-attached cell underscore the capability of the actin system to assemble into dynamic structures by the recruitment of proteins from the cytoplasm. The waves have no fixed shape, can reverse their direction of propagation and can fuse or divide. Actin waves separate two phases of the plasma membrane that are distinguished by their lipid composition. The area circumscribed by a wave resembles in its phosphoinositide content the interior of a phagocytic cup, leading us to explore the possibility that actin waves are in-plane phagocytic structures generated without the localized stimulus of an attached particle. Consistent with this view, wave-forming cells were found to exhibit a high propensity for taking up particles. Cells fed rod-shaped particles produced elongated phagocytic cups that displayed a zonal pattern that reflected in detail the actin and lipid pattern of free-running actin waves. Neutrophils and macrophages are known to spread on surfaces decorated with immune complexes, a process that has been interpreted as “frustrated” phagocytosis. We suggest that actin waves enable a phagocyte to scan a surface for particles that might be engulfed.
The actin cortex of a cell is a dynamic organelle capable of forming a variety of membrane-anchored structures specialized in cell motility, substrate-interaction, division or intracellular transport. The differentiation of the cell cortex into distinct areas can be induced by external signals, but it can also occur autonomously. For instance, the differentiation of a motile cell into a front and a tail does not require any signal from the environment,1 although the orientation of this differentiation may be determined by external gradients. Filopodia are also spontaneously formed in the absence of a trigger at the site of their protrusion, and actin waves arise spontaneously along the plasma membrane of a cell on a planar surface.2 On the other hand, the formation of a phagocytic cup is locally induced by the attachment of a particle to the cell surface and mediated by trans-membrane signals.3,4
The goal of the present report is to link two activities of the actin system to each other, the self-organization of actin waves and the induction of phagocytic cups. In phagocytosis a particle may be taken up as a nutrient or as a stimulant of immune responses. Engulfment of the particle is initiated by the accumulation of actin in a protrusion of the cell surface, the phagocytic cup, which extends along and finally closes around the particle. Actin filaments are enriched at the extending phagocytic cup, especially at its rim, and disassemble when the particle is completely engulfed.5–9 The differentiation of the cell cortex into a central area delineated by the rim of the cup is directed by the site of particle attachment, and the size of this area is controlled by the volume of the particle. Quite in contrast, actin waves are formed on the planar membrane area of a substrate-attached cell. They are propagated in changing directions over the inner surface of the membrane in contact with a uniform substrate, which means that the waves have no fixed position or shape.2,10 In addressing the role of these dynamic actin structures, we compare cells in a state of extensive wave generation with cells that are in the process of phagocytosing yeast particles.
Our studies were conducted using cells of Dictyostelium as professional phagocytes. In these microbial cells actin assembly and phosphoinositide signaling play similar roles in phagocytosis as in mammalian macrophages. Phosphoinositides are concentrated at the cytosolic face of membranes, where they act as key players in signaling pathways controlling phagocytosis and other processes that involve remodeling of the cytoskeleton and membranes.11–14 By direct interaction with specific binding modules, e.g., pleckstrin homology (PH) domains, each phosphoinositide recruits a different group of signaling proteins to its resident membrane. Two phosphoinositides, phosphatidylinositol 4,5-bisphosphate and phosphatidylinositol 3,4,5-trisphosphate, hereafter called PI(4,5)P2 and PIP3, play important and distinct roles in regulating actin dynamics.12,15 PI(4,5)P2, the predominant phosphoinositide of the plasma membrane, is the substrate of PI3-kinases, which convert PI(4,5)P2 to PIP3. In phagocytic cups, PIP3 is transiently enriched together with PI(3,4)P2, the product of its degradation by inositol 5-phosphatases.16
The domain of CRAC, a cytosolic regulator of adenylyl cyclase in Dictyostelium,17 binds with high specificity to PIP3 and PI(3,4) P2.16,18 Thus, PHcrac-GFP, a fusion of GFP to the PH domain of CRAC, labels the membrane of forming phagosomes and disappears shortly after the phagosome seals.18–20
Using PHcrac-GFP together with mRFP-LimEΔ, a probe for filamentous actin,21 we examined cells in the process of generating actin waves on a planar surface. We found that these waves separate two different areas of the cell surface: an inner area that bound PHcrac-GFP, corresponding to the interior of a phagocytic cup, and an external area that was not labeled by PHcrac-GFP, corresponding to the plasma membrane external to a phagocytic cup. In order to define the patterns of actin and PHcrac along the length of a cup, we fed cells of Dictyostelium discoideum with a mutant of Saccharomyces cerevisiae that forms rod-shaped particles.22 When these long yeast particles were taken up, the probes bound to the surrounding phagosome membrane in an axial direction, along a membrane tube extending from the rim to the base of the cup. Furthermore, very long particles were often not completely engulfed but eventually released, making pattern dynamics recognizable during both the extension and retraction of a cup. We show that the complex 3-dimensional patterns generated in phagocytosis are paralleled by the planar wave patterns, and propose that the waves' function is to search for particles to be taken up.
Actin waves are one of several actin-based structures present during normal movement of Dictyostelium cells on a planar glass surface.2 An efficient method to enhance wave formation is the depolymerization of actin by incubating cells with latrunculin A; during the reorganization of actin structure and function after removal of the drug, the cells pass through a period of excessive wave formation before normal cell motility recovers.10,23 Figure 1A shows that under the conditions of wave formation, the cell cortex differentiates into two phases sharply separated from each other by the actin waves. The membrane area circumscribed by a wave is strongly decorated with PHcrac (green). The actin wave (red) forms a boundary between this inner area and the external one, which does not bind PHcrac (see also movie 1).
Since PHcrac recognizes both PI(3,4,5)P3 and PI(3,4)P2, we also examined cells labeled with GFP-TAPP, which binds specifically to PI(3,4)P2.18 This probe displayed no difference in labeling of the inner and external areas, or the waves separating them, indicating that PI(3,4,5)P3 is the phosphoinositide that distinguishes these two areas in wave-forming cells. When the waves propagate in one or the other direction, the inner and external areas undergo reciprocal changes in size with negligible changes in the total area of the substrate-attached plasma membrane. This interconversion of membrane areas implies up and downregulation of PIP3 at the sites of actin waves.
The zonal arrangement of PIP3 and actin labels in wave patterns can be viewed as a 2-dimensional projection of their arrangement in phagocytic cups. The inner leaflet of the cup membrane is enriched in PIP3, and a ring of filamentous actin at the rim of the cup is equivalent to a wave in separating the nascent phagosome membrane from the outer area of the cell surface (Fig. 1B and C, and movie 2). The two cells shown in Figure 1B and C expelled a long yeast particle after partial uptake. Concomitantly the cup area decorated with PHcrac expanded and shrank before the actin band disappeared.
The actin waves always co-localize with the edge of the PHcrac-decorated area, where PIP3 forms a steep gradient between the inner and external areas. This rule holds for waves propagating in any direction and either expanding or diminishing the size of the inner area (Fig. 2A and B). It also holds for the de-novo formation of a circular wave, which is initiated by the focal appearance of PIP3 in association with actin. While the PIP3-enriched area expands, an actin wave is built up at its border (Fig. 2C). The concerted evolution of PIP3 and polymerized actin suggests an autocatalytic circuit of wave initiation, as previously demonstrated for the formation of a leading edge.24 Indeed, PIP3 production and formation of actin waves are interdependent. The plasma membrane became depleted of PIP3 upon the depolymerization of actin with latrunculin A, and regained PIP3 in the form of bursts when actin polymerization was allowed to recover from this treatment, as shown in Figure 2C. Similarly, actin waves and the high PIP3 enrichment of the inner territory disappeared within 2 minutes after addition of the PIP3-kinase inhibitor LY-294002, and both recovered within 3 minutes after removal of the drug (data not shown).
The rule that actin waves are linked to a gradient of PIP3 is best demonstrated by the toroid-like configuration shown in Figure 3A and movie 3. At the beginning of this time series, most of the large area of the cell surface was decorated with PHcrac, indicating it was enriched in PIP3. Gradually the label declined in the middle of the area until a critical steepness was reached and a new actin wave was inserted. This co-evolution of PIP3 and actin patterns shows two distinctive features: (1) the formation of an actin wave is coupled to a steep PIP3 gradient, and (2) there is a size limit for a uniform inner territory beyond which it becomes unstable.
Assuming that the inner territory of the wave pattern corresponds to the membrane area of a phagocytic cup, one would predict that the cup area might also have a size limit, and its expansion beyond this limit would result in the insertion of an actin band similar to the wave pattern in Figure 3A and movie 3. To test this, we have fed Dictyostelium cells with highly elongated yeast particles and recorded the patterns of actin and PIP3 along the large cups shaped by these particles. In Figure 3B and movie 4, the two ends of an elongated particle have been partially ingested by two cells that are expressing only the marker for filamentous actin. Initially, the phagosome surrounding the left-hand end of the particle was completely labeled with actin, but as the particle was drawn deeper into the cell, the actin label disappeared from the bottom of the phagosome (59- and 193-second frames), generating a tube of actin with two strongly labeled ends. This particle was eventually released by both cells.
Figure 3C and movie 5 show the successful uptake of a long particle by a cell expressing markers for both actin and PIP3. This particle, 16 µm in length, was drawn into the cell to a depth of 10 µm until it contacted the cortex on the far side, whereupon the cell completed ingestion by extending its own length to encompass the particle. Throughout the process of phagocytosis, actin filaments were highly enriched at the growing edge of the cup. During the initial stage of uptake, actin filaments diminished along the portion of the phagosome membrane that had already been internalized. However, in the later stage of uptake, a strongly labeled new band of actin filaments formed at the phagosome membrane deep within the cell (Fig. 3C, 15- and 31-second frames); this band persisted until uptake had been completed (Fig. 3C, 55-second frame). A PHcrac-labeled tube of phagosome membrane connected the two bands of actin, while the basal part of the cup below the second actin band was depleted of PHcrac. Thus, PHcrac was enriched between two bands of actin about 8 µm separated from each other, resembling the toroid configuration in a planar wave.
Actin waves can produce force to push the plasma membrane forward,10,23 and phagocytes need to produce force to extend the membrane area around a particle. Figure 4 relates actin waves to the generation of force in a phagocytic cup. On a planar substrate we often observed patterns in which wave fronts escaped from the major wave and propagated independently into the inner area (Fig. 4A). A similar behavior is displayed by the phagocytic cup shown in Figure 4B and movie 6. When the time series began, the Dictyostelium cell had partially ingested an elongated yeast particle. Actin filaments were enriched in two bands, one at the rim of the phagocytic cup and the other deeper in the cell, with a lesser amount of actin along the phagosome membrane between the two bands but none detectable below the second band (Fig. 4B, 0 s frame). Over the next several minutes, waves formed along the actin-rich zone of the phagosome membrane and moved toward the rim of the cup (Fig. 4B, 169 to 185 s frames). Simultaneously, the yeast particle was pushed in the opposite direction against the back of the cell, indicating that during the period of wave generation force was produced that facilitated uptake of the particle (Fig. 4C). Eventually, the cell failed to engulf this particle, which turned out to be too long; wave formation ceased, and the particle was released.
Phagocytosis does not require myosin-II,25,26 and the formation of actin waves is independent of this conventional myosin.10 To determine whether the same is true for the actin pattern along phagocytic cups, we fed myosin-II-null cells with long yeast particles. Movie 7 shows that actin initially decorated the entire cup and subsequently condensed into a tube. Finally, the inner ring moved like a wave toward the rim of the cup. In summary, the actin dynamics resembled in all respects that in wild-type cells.
The analogous patterns of 2-dimensional waves and 3-dimensional phagocytic cups suggest that phagocytic structures are spontaneously generated on a planar substrate surface. If this view is correct, the period of extensive wave formation should be a period of high phagocytic activity. This implies that, after the depolymerization of actin, phagocytosis would recover earlier than normal cell motility. To test this hypothesis, we supplied cells at early stages of recovery with yeast particles and recorded uptake of the yeast (Fig. 5 and movies 8 to 11).
During the stage of wave-formation, cells proved to be highly active phagocytes. This is indicated by the increase in the number of particles ingested during the period of intense wave formation at 20 to 50 minutes of recovery from actin depolymerization (Fig. 5A). Often, but not always, a wave regressed when a particle was taken up. The wave-forming cell of Figure 5B had already taken up four yeast particles, and took up a fifth one as the wave regressed (Movie 8). A direct link between wave formation and particle uptake is evident in Figure 5C, which shows the conversion of a wave into a cup upon contact with a particle (see also movie 10).
Another actin-dependent activity involving membrane internalization is fluid uptake. For cells in the wave-forming state we observed both extensive formation of macropinocytic cups (Fig. 5B) and clathrin-associated micropinocytosis (unpublished data). Together these data indicate that the capacity for endocytosis is one of the earliest activities to recover during the repolymerization of actin. Since in the time window of extensive wave formation the cells have a high potential of forming phagocytic structures even on a planar surface, they can be used to study this activity by viewing their substrate-attached surface essentially independent of cell motility.
The results presented here link a phenomenon of self-organization in the actin system to the established biological function of this system in phagocytosis. We show that the structure of a phagocytic cup is mimicked by the relationship of planar actin waves to PIP3-enriched areas of the plasma membrane, as outlined in Figure 6A. An interrelation of actin polymerization and PIP3 localization in the plasma membrane has been noted for several actin-based cellular functions: chemotaxis,27,28 phagocytosis,18–20 cytokinesis29,30 and actin wave propagation. In the keratocyte-like cells of a Dictyostelium mutant, but also in wild-type cells, actin waves were found to be correlated with PIP3 waves.31 In these cases, the actin waves were usually chased by PIP3 waves, which suggested that effectors recruited by PIP3 may induce the depolymerization of actin filaments. Under our conditions the actin waves separated two areas of the plasma membrane, an inner area enriched in PIP3 from an external area depleted of this lipid. When a wave expanded, PIP3 followed its back; when the wave retracted, PIP3 disappeared at its front (Figs. 2A and B, and movie 1).
The peak of actin polymerization along a spatial gradient of PIP3, illustrated in Figures 2 and and3A3A and movie 3, may be relevant not only to phagocytosis. It may also be instrumental in translating a PIP3 gradient formed in response to chemoattractant into a directional movement of the cell. In the membrane of a cell exposed to a gradient of chemoattractant, PIP3 synthesis is induced in the form of a crescent with the highest concentration pointing to the source of attractant and gradual decline toward both sides.27 The actin-driven response of cells in a shallow gradient of attractant is characterized by split pseudopod formation.32 Similarly, in the absence of external cues, cells move persistently by split pseudopodia that are extended alternately right/left/right from the previous one.33 Thus, it appears that the actin-based formation of pseudopodia is focused on the lateral slopes of a PIP3-enriched membrane area, similar to the actin waves on a planar cell surface and the actin rings in a phagocytic cup (Fig. 6A).
A relationship of cell spreading on a planar surface to phagocytosis of a curved particle has repeatedly been proposed (reviewed in ref. 34). Neutrophils adhere to immune complexes along a surface that they cannot phagocytose. In response to interaction with the functionalized surface, lysosomal constituents, which are normally delivered into a closed phagosome, are released into the environment. Under these conditions, the attempted phagocytosis has been considered to be “frustrated” by the large size of the surface.35 Frustrated phagocytosis has also been reported for macrophages that interact through their Fc receptors with ligands on a surface.36 Here we consider the propagation of actin waves as a mechanism that enables a phagocyte to scan a surface for particles to be taken up. As scanning machines the actin waves propagate in different directions along the entire substrate-attached membrane of a cell by the incorporation of actin monomers and the depolymerization of actin filaments in a defined 3-dimensional pattern.10
Actin waves have been shown to apply force on the cell border to push the membrane forward.23 Likewise, the actin structures in a phagocytic cup appear to apply force for internalization of the particle against membrane tension. The long particles employed in this study required a substantial area of the membrane for their uptake; in fact the particles were often too long to be completely engulfed. A rough estimate assuming a spherical cell with a smooth surface suggests for a phagocytosing cell, like the one in Figure 4B, that at least 20 percent of the total membrane area is reversibly pulled into the phagocytic cup.
Upon cessation of actin mediated force production at the cup surface, the pressure for internalization is relieved and the particle is expelled. We observed two modes of redistribution of the excess membrane after the often forceful release of a particle: global formation of surface ruffles (Fig. 4, 308 and 327 s frames) and less often local bleb formation within the boundary of an actin ring (Fig. 1B and movie 2) as diagrammed in Figure 6B. The sudden conversion of an invaginated membrane area confined by an actin ring into a protrusion illustrates most clearly that the uptake machinery has acted against tension. In all cases that we recorded the particle was either completely engulfed or eventually released. Thus, phagocytes must have mechanisms to recognize that a particle is not suitable for uptake. The decision for release might be based on a clock counting the time spent in the state of an open cup, as suggested by Figure 4B or on a safety-valve opened when internal pressure exceeds a certain limit, as suggested by Figure 1B.
In conclusion, the patterns generated by wave forming cells of Dictyostelium correspond to two-dimensional projections of phagocytic cups, indicating that the cup structure can be spontaneously generated on a planar surface by differentiation of the cell cortex into two areas separated by actin waves. The inner area is distinguished by its high PIP3 content from the external area. These areas expand or shrink reciprocally when the waves propagate into one direction or the other, paralleling the increase and decrease of cup area during the uptake and expulsion of a particle. The temporal evolution of actin wave and phagocytic cup patterns indicates that spatial gradients of PIP3 are translated into sharp peaks of polymerized actin.
PHcrac-GFP27 and mRFP-LimEΔ,21 LimEΔ-GFP37only, or GFP-TAPP18 and mRFP-LimEΔ were expressed in Dictyostelium discoideum strain AX2-214. The strain used for Figures 3B and and4B4B carried a VacuolinB-null mutation that is not relevant to the work presented. The AX2-derived mutant HS2205 deficient in myosin-II heavy chains38 was transfected with LimEΔ-GFP. Cells were cultivated in Petri dishes with nutrient medium and washed for imaging in 17 mM K/Na-phosphate buffer, pH 6.0 (PB).10 Experiments were performed at 23 ± 2°C. For recovery of actin polymerization, cells were incubated for 15–20 minutes with 5 µM latrunculin A (Invitrogen), and imaged after replacement of the drug with PB. For treatment of wave forming cells with Ly 294002 (Sigma L9908), a 30 mM stock solution in DMSO was diluted with PB to 50 µM. After incubation for 5 to 10 minutes, the drug was replaced with PB.
For phagocytosis assays, cells were fed with long particles of the conditional cdc42 S. cerevisiae mutant DDY 1306,22 grown in YPD at 27°C with 3 percent formamide or, for Figure 5, with spherical S. cerevisiae particles (Sigma YSC2) boiled for 30 minutes in a waterbath and labeled with TRITC. For Figure 5A, 1 × 109 yeast particles per ml of PB were vortexed and filtered through gauze of 8 µm pore size to remove aggregates. 10 µl of this suspension and 1 ml of PB were added to a coverslip within a ring of 20 mm diameter on which the Dictyostelium cells were attached. To visualize waves on the bottom of the cells together with all particles taken up, confocal images were acquired in 4 planes separated from each other by 2 µm.
Wave patterns in Figures 1A, ,22 and and3A3A and and4A4A were imaged using through-objective TIRF microscopy essentially as described.10 GFP and mRFP were simultaneously excited at 491 nm such that emissions of both fluorophores were recorded from the same evanescent field. All other images were obtained by confocal scanning fluorescence microscopy. For Figure 3C, cells were imaged using an Ultraview ERS spinning disc confocal set (Perkin Elmer LAS) on a Nikon TE-2000 inverted microscope equipped with a Plan Apo VC 100x/1.4 NA objective. In this system, channels were recorded sequentially onto an EM-CCD camera using 488 nm excitation, 527/55 nm emission and 568 nm excitation, 615/70 nm emission for the GFP and mRFP labels, respectively. For Figure 5A and B, an Eclipse TE300 microscope with a 100x/1.4 oil Plan Apo objective (Nikon), an Ultraview spinning disc scanner (Perkin Elmer), an ORCA ER model C4742-95-ERG camera (Hamamatsu Photonics) was used. Figures 1B and C, C,3B3B and and4B4B were recorded on a Zeiss LSM 510 as described,20 and Figure 5C was acquired on a Zeiss LSM 410 confocal microscope.
Our work was supported by the Max Planck Society and grants in the SPP 1128 program of the Deutsche Forschungsgemeinschaft to G.G. and Stefan Diez, and by National Science Foundation grant MCB-0344541 to M.C. We acknowledge the equipment and technical support of the Imaging Core Facility at the Oklahoma Medical Research Foundation. We appreciate the advice of Stefan Diez and thank Hans Meinhardt, Tübingen, for stimulating discussions, Jana Prassler for expert assistance, Carol Parent, NIH, for the PHcrac-GFP vector and Cornelis Weijer, Dundee, for the TAPP-GFP vector.
Previously published online as a Cell Adhesion & Migration E-publication: http://www.landesbioscience.com/journals/celladhesion/article/9708