Receptor Activation Alters Inner Surface Potential During Phagocytosis

See allHide authors and affiliations

Science  21 Jul 2006:
Vol. 313, Issue 5785, pp. 347-351
DOI: 10.1126/science.1129551


The surface potential of biological membranes varies according to their lipid composition. We devised genetically encoded probes to assess surface potential in intact cells. These probes revealed marked, localized alterations in the charge of the inner surface of the plasma membrane of macrophages during the course of phagocytosis. Hydrolysis of phosphoinositides and displacement of phosphatidylserine accounted for the change in surface potential at the phagosomal cup. Signaling molecules such as K-Ras, Rac1, and c-Src that are targeted to the membrane by electrostatic interactions were rapidly released from membrane subdomains where the surface charge was altered by lipid remodeling during phagocytosis.

The plasma membrane of mammalian cells contains about 20 mol % of anionic lipids on the inner leaflet. The preferential accumulation of negative charges creates an electric field, estimated at 105 V/cm, that strongly attracts cationic molecules, including peripheral membrane proteins (1). This electrostatic interaction has been best documented for the myristoylated alanine-rich C kinase substrate (MARCKS), which interacts with the plasmalemma through a polycationic domain, in conjunction with a myristoyl anchor (2). The realization of this charge-dependent anchorage led to the postulation of an “electro-static switch” model (2), which predicts that the formation and stability of electrostatic associations can be regulated by changes in the charge of either the cationic protein complex or the anionic lipid layer.

Little is known about the regulation of the electrostatic potential of the plasmalemma. It is not clear whether the surface potential of the cytoplasmic leaflet undergoes regulated changes and, if so, whether such changes play a role in modulating the association of cationic proteins. This paucity of information is due to the absence of methods to monitor the surface potential of the inner membranes of intact cells.

Phagocytosis is associated with extensive remodeling of the plasma membrane lipids (3). Such lipid changes could potentially alter the overall charge of the plasmalemma and may therefore serve as an “electrostatic switch” to modulate the interaction with cationic ligands, which could in turn affect the phagocytic response. To investigate this possibility, we developed means to assess the electrostatic potential of the inner aspect of the plasma membrane in intact cells. We designed several polycationic fluorescent probes that are selectively targeted to the plasmalemma by virtue of its unique negative surface charge. One probe was modeled after the C terminus of K-Ras, which was shown to associate with the membrane in a charge-dependent manner (4). To ensure that the probe was not phosphorylated, we mutated all serine and threonine residues to alanine to create a second probe (K-pre, Fig. 1A). In a third probe, all lysines were substituted by arginines to avoid ubiquitination (R-pre, Fig. 1A).

Fig. 1.

Design and characterization of surface potential probes. (A) Structure of surface potential–sensitive probes. Amino acid abbreviations: A, Ala; C, Cys; D, Asp; F, Phe; G, Gly; I, Ile; K, Lys; L, Leu; M, Met; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp. (B) Transfer of R-pre from donor liposomes (100 mol % PC) to acceptor liposomes containing PC (98 mol %), PC/PE (78/20 mol %), PC/PS (78/20 mol %), PC/PA (78/20 mol %), or PC/PIP2 (96/2 mol %). F/Fo, fluorescence intensity ratio. (C) Transfer of R-pre to PC (98 mol %; navy) or PC/PS (78/20 mol %; other colors) acceptor liposomes in medium supplemented with the indicated NaCl concentrations. (D) Partition coefficient (Kp; relative to PC) of R-pre binding to liposomes containing PC (100 mol %), PE, PS, or PA (20 mol % each), or PIP2 (2 mol %) in the presence (red) or absence (blue) of 1 M NaCl. (E) Macrophages transfected with R-pre, KRϕ, or K-myr. (F) Macrophages transfected with K-Ras tail containing +8, +4, and +2 charges. Scale bars, 2 μm. (G) Effect of charge on the membrane/cytosol ratio of K-Ras tail variants.

An in vitro assay assessed the effect of surface potential on the affinity of R-pre for pure lipid bilayers (5). R-pre labeled with bimane, a solvochromic dye, partitioned preferentially to liposomes containing anionic lipids (Fig. 1B). Inclusion of phosphatidylinositol 4,5-bisphosphate (PIP2) or phosphatidylserine (PS), at concentrations resembling those in the plasmalemma, or phosphatidic acid (PA) increased the partition coefficient by factors of 10, 51, and 87, respectively, relative to liposomes containing only phosphatidylcholine (PC) with or without phosphatidylethanolamine (PE) (Fig. 1D). Moreover, progressive elevation of the ionic strength reduced the interaction of R-pre with anionic liposomes and minimized the affinity difference relative to uncharged liposomes (Fig. 1, C and D, and fig. S1B).

We next expressed a genetically encoded form of R-pre conjugated to red fluorescent protein (RFP) in RAW264.7 macrophages. R-pre–RFP associated almost exclusively with the inner aspect of the plasma membrane (Fig. 1E), as found for the tail of K-Ras (Fig. 1F) (6). We also used polybasic constructs containing an N-terminal myristoylated sequence (K-myr and Nt-Src) or containing an amphiphilic helix (KRϕ) instead of a farnesylation site (Fig. 1A) (6, 7). Like R-pre, KRϕ partitions preferentially to anionic liposomes in an ionic strength–dependent manner (fig. S1, A and C). When expressed in macrophages, KRϕ (Fig. 1E), K-myr (Fig. 1E), and Nt-Src (fig. S5A) localized to the plasma membrane, which implies that the common feature of these constructs—namely their positive charge—was a primary determinant of their targeting. Accordingly, progressive elimination of cationic residues resulted in graded detachment of K-Ras tail–derived mutants from the plasmalemma (Fig. 1, F and G) (6).

Three approaches were used to show that the probes responded to changes in the electric field at the inner surface of the plasma membrane. First, cells were treated with an ionophore to elevate cytosolic calcium, which shields the surface charge of the membrane and induces PIP2 hydrolysis through activation of phospholipase C (PLC). The extensive degradation of PIP2 was verified with the use of PH-PLCδ-GFP, a green fluorescent protein (GFP)–tagged probe for this phosphoinositide (Fig. 2A). Calcium also activates the lipid scramblase, resulting in net translocation of PS to the outer leaflet (Fig. 2A). In parallel with the changes in anionic lipid composition and distribution, ionomycin induced a pronounced dissociation of R-pre, KRϕ, and K-myr from the inner surface of the membrane (Fig. 2A and movie S1). Redistribution of the probes was not due to wholesale remodeling or disruption of the membrane. This was established with the use of three different genetically encoded fluorescent markers retained at the plasma membrane by hydrophobic interactions: glycosylphosphatidylinositol (GPI)–linked GFP, a transmembrane chimeric protein (GT46), and a farnesylated and diacylated GFP termed Palm (fig. S1D). When coexpressed with the cationic probes, the distribution of these markers remained essentially unaltered after treatment with ionomycin (Fig. 2A), which confirmed the integrity of the plasmalemma. Quantitation of the effects of ionomycin on the distribution of the probes is shown in Fig. 2B.

Fig. 2.

Effect of manipulating surface potential on probe distribution. (A) Distribution of probes in cells treated with ionomycin (top), antimycin/deoxyglucose (middle), or dibucaine (bottom). Insets: cells before ionomycin. Scale bars, 2 μm. (B) Quantification of effect of treatments on probe distribution (ratio of membrane fluorescence of probes specified). Data are means ± SE (n > 20).

An alternative means of reducing surface charge is to inhibit the ongoing formation of polyphosphoinositides by depleting cellular adenosine triphosphate (ATP). Simultaneous impairment of glycolysis and mitochondrial respiration was accompanied by dissociation of PH-PLCδ-GFP from the membrane, indicating loss of PIP2 (Fig. 2A). All three cationic probes detached from the plasma membrane of ATP-depleted cells, whereas GPI-GFP, GT46, and Palm remained unaltered (Fig. 2, A and B).

Flipping of PS can be induced by the anesthetic dibucaine, as indicated by the marked increase in annexin-V binding (Fig. 2A). Loss of PS from the inner leaflet, together with the positive charges contributed by dibucaine itself, sufficed to displace R-pre, KRϕ, and K-myr from the membrane (Fig. 2A). Because the three probes responded similarly to all treatments, regardless of their mode of hydrophobic interaction with the membrane, we conclude that their dissociation occurred in response to changes in surface charge. These probes are therefore suitable for monitoring the surface potential of the inner aspect of the plasma membrane.

We next applied the cationic probes to study membrane remodeling during phagocytosis. Confocal imaging was used to monitor ingestion of immunoglobulin G–opsonized latex particles by macrophages expressing the surface potential probes. The hydrophobically anchored markers were used along with the cationic probes to control forremodelingcausedbyfusion and fission events. The cationic and hydrophobically anchored probes localized to the nascent phagocytic cup (Fig. 3A and fig. S2, B and C). Both probes were also present in pseudopods as they progressed along the sides of the particle, although the cationic probes were often partially depleted from the base of the cup at this intermediate stage (Fig. 3A and fig. S2, A to C). At more advanced stages of internalization, a further reduction of the cationic probes was observed, which was invariably followed by near-total depletion shortly after completion of particle engulfment (Fig. 3A; fig. S2, B and C; and movies S2 and S3). By contrast, a substantial fraction of the hydrophobically anchored probes remained associated with the phagosome even after internalization was completed (Fig. 3A and fig. S2, B and C). When calculated 3 min after initiation of phagocytosis, the phagosome-to-bulk (unengaged) plasma membrane ratios for R-pre (0.15 ± 0.02), KRf (0.20 ± 0.03), and K-myr (0.20 ± 0.02) were significantly (P < 0.001) lower than those for GPI (0.58 ± 0.04), GT46 (0.59 ± 0.04), and Palm (0.52 ± 0.03) (Fig. 3, B to D). The depletion of the cationic probes was more profound than expected on the basis of remodeling, suggesting alterations in the surface potential of forming phagosomes.

Fig. 3.

Charge and lipid changes during phagocytosis. (A) Time course of R-pre/GPI-GFP redistribution. Asterisks indicate latex beads. (B to D) Fluorescence ratio of probes in phagosomal membrane/unengaged plasma membrane, calculated at onset of phagocytosis (0 s) and after 180 s. Data are means ± SE (n > 20); *P < 0.001. (E) Redistribution of PH-PLCδ and GPI-GFP during phagocytosis. (F) Distribution of PS (white) during phagocytosis. Green, total beads; red, extracellular beads. (G) Fluorescence ratio of PH-PLCδ, GPI-GFP, or annexin-V in phagosomal membrane/unengaged plasma membrane. Data are means ± SE (n ≥ 16); *P < 0.001. (H) Distribution of R-pre (red) and PH-PLCδ (green) in absence (left) or presence (right) of LY294002. In (A), (E), and (H), insets show separately the fluorescence of RFP (red) and GFP (green) of the area boxed in the main panels. In (F), insets show (from left to right) PS, external beads, and total beads. Scale bars, 2 μm.

We investigated whether changes in anionic lipid composition or distribution account for the alterations in surface potential of the phagosomal membrane. PIP2 was markedly depleted from forming phagosomes (Fig. 3, E and G, and movie S4). Moreover, dissociation of the cationic probe (R-pre) could be prevented when the loss of PIP2 was impaired (Fig. 3H).

Because PS contributes ∼30% of the charge on the plasma membrane, its fate during phagocytosis was also studied. Annexin-V was used to monitor the distribution of PS at the onset of phagocytosis and 3 min thereafter. In nonpermeabilized cells, there was no discernible binding of annexin-V; this result implies that little PS is present on the outer monolayer before, during, or after phagocytosis. To gain access to PS in the inner leaflet, we fixed cells and gently permeabilized them with saponin. PS was clearly detectable at the base of nascent phagosomes but appeared greatly depleted from formed phagosomal vacuoles (Fig. 3, F and G). Similar results were obtained with a PS-specific antibody (fig. S4). Jointly, the metabolism of PS and phosphoinositides could account for the changes in surface potential during phagocytosis.

Could the change in surface potential have physiological consequences? Molecules attracted to the membrane by its negative potential are anticipated to dissociate, possibly altering signal transduction and cytoskeletal structure. The fact that certain members of the Ras superfamily (e.g., K-Ras, Rac1) contain a polybasic domain gives credence to this concept (8). K-Ras constitutively associates with the plasma membrane by both prenylation and a polycationic domain in its hypervariable region (Fig. 4A) (8). The importance of the positive charges in this region was validated by introduction of three negatively charged residues, which resulted in partial dissociation of the protein from the plasmalemma (Fig. 4A). Moreover, K-Ras responded to changes in surface potential induced by ionomycin, whereas H-Ras, which is dually palmitoylated, was unaffected (Fig. 4A). As anticipated, H-Ras was retained in sealed phagosomes to an extent comparable to that of the GPI-anchored marker (Fig. 4, A and B). In sharp contrast, K-Ras was virtually absent from formed phagosomes (Fig. 4, A and B, and movie S6).

Fig. 4.

Surface potential modulates guanosine triphosphatase localization. (A) Top: Distribution of full-length K-Ras, K-Ras-3E, and H-Ras before and after ionomycin treatment. K-Ras 3E is a mutant form of K-Ras with three additional negative charges (see fig. S6 for structure). Bottom: Distribution of H-Ras and GPI-GFP (left) or K-Ras and H-Ras (right) during phagocytosis. (B) Phagosome/bulk membrane ratio of GPI-GFP, H-Ras, or K-Ras at onset of phagocytosis (0 s) or after 180 s. Data are means ± SE (n > 20); *P < 0.01. (C) Full-length prenylated Rac1 partitions preferentially (by a factor of 2.3 ± 0.7; n = 8) to beads coated with PC/PS (80/20 mol %) relative to beads coated with PC (100 mol %). (D) Distribution of Rac1(Q61L), Rac1(Q61L)-6Q, or Rac1(Q61L)–H-Ras tail before and after ionomycin treatment. From left to right: plain Rac1(Q61L); Rac1(Q61L)-6Q; plain Rac1(Q61L), treated with ionomycin; Rac1(Q61L)–H-Ras tail, treated with ionomycin [see fig. S6 for structures of Rac1(Q61L)-6Q and Rac1(Q61L)–H-Ras tail]. (E) Redistribution of Rac1 and Palm during phagocytosis. (F) Phagosome/cytosol ratio of wild-type Rac1 at 0 and 180 s. Data are means ± SE (n > 20); *P < 0.001. (G) Redistribution of Rac1(Q61L) and Palm during phagocytosis. (H) Phagosome/membrane ratio of Rac1(Q61L) at 0 and 180 s. Data are means ± SE (n > 20); *P < 0.01. Scale bars, 2 μm.

Like K-Ras, Rac1 also contains a polybasic domain (9). Recombinant prenylated Rac1 bound preferentially to anionic (PS/PC-coated) beads relative to beads coated with PC only (Fig. 4C). Moreover, mutation of the cationic residues in the polybasic region to glutamine resulted in dissociation of constitutively active Rac1(Q61L) from the plasmalemma (Fig. 4D), and Rac1(Q61L) localization was sensitive to changes in surface potential, whereas a mutant with the polybasic domain substituted by the hydrophobic tail of H-Ras was not (Fig. 4D). Rac1 is of particular importance to Fc receptor-mediated phagocytosis and accumulates at the base of forming phagosomes, detaching rapidly upon sealing (Fig. 4, E and F, and fig. S7A) (10). Rac1(Q61L) also detached from sealing phagosomes with kinetics indistinguishable from those of wild-type Rac1 (Fig. 4, G and H, and movie S7). Because Rac1(Q61L) is constitutively bound to guanosine triphosphate (GTP), its dissociation from phagosomes was not due to nucleotide hydrolysis or cessation of nucleotide exchange. Instead, release waslikelymediatedbytermination of its electrostatic association with the plasmalemma. Accordingly, the C-terminal tail of Rac1 containing the polybasic domain behaved similarly (fig. S7B).

Our data indicate that the surface potential of the inner leaflet of the membrane decreases locally during phagosome formation. The change is attributable primarily to depletion of PIP2 and PS, but depletion of phosphatidylinositol 4-phosphate was also observed (fig. S3 and movie S5). Activation of inositide lipases, kinases, and phosphatases occurs during phagocytosis and bacterial invasion (3), readily accounting for the changes in PIP2. PS could be converted to PE by decarboxylation or could be externalized during phagocytosis by scramblases and/or efflux pumps.

Our results also indicate that the anchorage of important signaling molecules, including K-Ras and Rac1, can be modulated focally by localized changes in surface potential. Other proteins anchored electrostatically to the membrane, such as MARCKS, are equally susceptible to the charge alterations that accompany lipid remodeling. Indeed, we also obtained evidence for localized detachment of the tyrosine kinase c-Src (fig. S5, B and C).

The consequences of altered surface charge in other important biological phenomena must be considered. Activation of phosphoinositide metabolism, elevation in cytosolic calcium, and PS flipping occur after stimulation of multiple receptors and channels as well as during apoptosis. The effect of such responses on inner surface potential may be measurable with the use of approaches like the one described here. Cycles of membrane dissociation/reassociation may add a layer of functional control to complement the traditional biochemical mode of regulation of signaling proteins.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Movies S1 to S7

References and Notes

Stay Connected to Science

Navigate This Article