Report

Regulation of Phagosome Maturation by Signals from Toll-Like Receptors

See allHide authors and affiliations

Science  14 May 2004:
Vol. 304, Issue 5673, pp. 1014-1018
DOI: 10.1126/science.1096158

Abstract

In higher metazoans, phagocytosis is essential in host defense against microbial pathogens and in clearance of apoptotic cells. Both microbial and apoptotic cells are delivered on a common route from phagosomes to lysosomes for degradation. Here, we found that activation of the Toll-like receptor (TLR) signaling pathway by bacteria, but not apoptotic cells, regulated phagocytosis at multiple steps including internalization and phagosome maturation. Phagocytosis of bacteria was impaired in the absence of TLR signaling. Two modes of phagosome maturation were observed, constitutive and inducible; their differential engagement depended on the ability of the cargo to trigger TLR signaling.

Phagocytosis is an ancient form of host defense (1) performed by specialized phagocytes, such as macrophages (26). Although the phagocytosis of bacteria and apoptotic cells relies on the same cellular machinery (7), the immunological response differs (46), as determined primarily by the engagement of TLRs (6, 8).

To examine whether TLR signaling regulates phagocytosis, we compared macrophages from wild-type (WT), MyD88–/–, and TLR2x4–/– mice (generated from TLR2–/– × TLR4–/– mice crossed to homozygosity). Phagocytosis was performed in serum-free media to eliminate contributions of Fc and/or complement receptors (6, 9). MyD88–/– and TLR2x4–/– macrophages were unresponsive to inactivated Escherichia coli (fig. S1). Fluorescent bacteria were used to monitor phagocytosis in a time- and dose-dependent manner by a fluorescence activated cell sorter (FACS) assay (10) (fig. S2). MyD88–/– and TLR2x4–/– macrophages internalized recombinant green fluorescent protein (GFP)–expressing E. coli at reduced levels compared with WT, as measured by their lower GFP mean fluorescence intensity (MFI) (Fig. 1A). No differences were observed at 4°C, indicating that internalization, rather than adherence, was affected (11). MyD88–/– and TLR2x4–/– macrophages showed similar defects in internalization of heat-inactivated Salmonella typhimurium. The use of Salmonella allowed definitive discrimination between adherent and internalized bacteria when we stained with an antibody against Salmonella (fig. S3). Defects were observed only when permeabilized cells were stained, which indicated impaired internalization and not bacterial binding. A similar defect was observed in the presence of serum, which suggested that the contribution of TLRs is dominant during phagocytosis of serum-opsonized bacteria (Fig. 1B). These differences are likely to have been underestimated, because intact TLR-dependent antimicrobial mechanisms in WT macrophages resulted in bacterial degradation (figs. S4 and S10, and text below). A defect was also observed in phagocytosis of Staphylococcus aureus (fig. S5). Phagocytosis in MyD88–/– and TLR2x4–/– macrophages was not impaired in general, because the phagocytosis of inert microspheres that did not activate TLRs (11) was similar to that of WT (Fig. 1C) (11). In addition, no differences in actin polymerization or activation of Cdc42 and Rac1 were observed (11). TLR signaling stimulated phagocytosis independent of gene transcription, as it occurred within 30 min and was unaffected by the presence of actinomycin D (fig. S6).

Fig. 1.

Impaired phagocytosis in the absence of a TLR-MyD88 signal. Bone marrow (BM)–derived macrophages given E. coli–GFP at a ratio of 1:100. (A) (a) FACS histograms representative of at least 10 experiments and showing phagocytosis in serum-free conditions by WT (red), MyD88–/– (tinted, top panels), or TLR2x4–/– BM-derived macrophages (tinted, bottom panels) at 37°C. GFP fluorescence is plotted on the x axis, on a logarithmic scale ranging from 100 to 104, and cell number is plotted on the y axis. (b) Mean fluorescence intensities (MFIs) of GFP from (a). (B) MFIs showing phagocytosis of serum-opsonized E. coli–GFP by WT or MyD88–/– macrophages at various times in the presence of serum [similar results in TLR4–/– macrophages (11)]. Plot legend is as in (C). (C) MFIs at 2 hours after phagocytosis of FluoSpheres under serum-free conditions by WT or MyD88–/– macrophages, as indicated in the legend.

Phagocytosed bacteria are initially contained within phagosomes that mature into phagolysosomes (1215). LysoTracker (Molecular Probes) selectively labels late endosomes and lysosomes (16), and it colocalizes with the lysosomal-associated membrane protein LAMP (17, 18) (fig. S7). We monitored the maturation of phagosomes containing E. coli–GFP by their ability to colocalize with LysoTracker red over time. Whereas E. coli colocalized with LysoTracker in 100% of WT macrophages at 1 hour, colocalization was not observed in more than 50% of MyD88–/– or TLR2x4–/– macrophages at this time point (Fig. 2A). Similarly, intracellular staining for heat-killed S. typhimurium in WT macrophages showed numerous clusters of internalized bacilli (arrows) that colocalized with LAMP-2 (Fig. 2B, top inset). In TLR2x4–/– macrophages, less Salmonella were internalized, and individual bacilli (arrows) were discernible that did not colocalize with LAMP-2 (Fig. 2B, bottom inset).

Fig. 2.

Block in phagosome maturation in the absence of a TLR-MyD88 signal. (A) Peritoneal macrophages loaded with LysoTracker red were given E. coli–GFP at a ratio of 1:10. Fluorescence micrographs show WT, MyD88–/–, or TLR2x4–/– macrophages at 1 hour. Scale bar, 10 μm. (B) BM-derived macrophages activated by interferon-γ (IFN-γ) given heat-killed S. typhimurium at a ratio of 1: 50. Fluorescence micrographs show WT or TLR2x4–/– macrophages at 30 min stained with antibodies against both Salmonella and LAMP-2. “Intact,” staining of intact cells for surface bacteria; “permeabilized,” additional staining for internalized bacteria in permeabilized cells. Insets in right panels show boxed areas enlarged. Nuclei in (A) and (B) stained with 4',6-diamidino-2-phenylindole (DAPI). Scale bars, 5 μm. All micrographs are compiled serial Z-stack images 0.4 μm apart. (C) Electron micrographs of WT and MyD88–/– BM-derived macrophages fixed 1 hour after phagocytosis of S. aureus. (D) Electron micrographs of BM-derived WT, MyD88–/–, TLR4–/–, or TLR2–/– macrophages 1 hour after phagocytosis of S. aureus. Scale bar in WT also for TLR4–/– and TLR2–/– micrographs. All scale bars in C and D, 500 nm. (E) Quantification of LAMP-1–gold immunolabel along either S. aureus–phagosomal (black) or endocytic vesicle membranes (patterned) in WT versus MyD88–/– macrophages 1 hour after phagocytosis.

We next examined the phagocytosis of S. aureus by electron microscopy (EM). Whereas recognition of E. coli can involve signaling adaptors other than MyD88, recognition of S. aureus depends solely on TLR2 and MyD88 (19). Therefore, TLR2–/– and MyD88–/– macrophages were compared with WT and TLR4–/– macrophages. All macrophages were capable of internalizing S. aureus, but more MyD88–/– than WT macrophages contained no bacteria (table S1). In addition to S. aureus–containing phagosomes, numerous vacant vacuoles were observed in WT cells (arrowheads, Fig. 2, C and D) that were absent in MyD88–/– macrophages. These vacuoles were absent in uninfected macrophages and most likely represent a postdegradative compartment, because many contained remnants of bacterial degradation. A clear space was present between S. aureus and the phagosomal membrane in WT macrophages (spacious phagosomes) (arrows, Fig. 2, C and D), in contrast to the close apposition of S. aureus with the phagosomal membrane (tight phagosomes) in MyD88–/– macrophages (arrows, Fig. 2, C and D). Some phagosomes were in contact with electrondense lysosomes, which suggested intact lysosomal docking, but stalled fusion with phagosomes, in the absence of TLR signaling (open arrowheads, Fig. 2, C and D). In TLR2–/– macrophages, S. aureus were also within tight phagosomes (arrows, Fig. 2D) and showed normal docking but deficient fusion with lysosomes (open arrowheads, Fig. 2D), which was consistent with the recognition of S. aureus by TLR2. Notably, TLR4–/– macrophages were similar to WT (arrows, Fig. 2D), which was consistent with the lack of TLR4 involvement in S. aureus recognition (19).

By immuno-EM, numerous LAMP-1–gold particles were observed along the membranes of phagosomes in WT macrophages, confirming that phagosomes here had acquired the lysosomal marker (13, 17, 18) (Fig. 2E and fig. S8). In contrast, phagosomes in MyD88–/– (Fig. 2E) and TLR2x4–/– macrophages (11) did not label appreciably with LAMP-1–gold. Rather, gold particles were detected along the membranes of endocytic vesicles adjacent to S. aureus phagosomes. Collectively, the data suggest a block in phagosome maturation into late endosomal and lysosomal stages in the absence of TLR signaling.

We next asked whether TLR signaling affected phagocytosis of apoptotic cells. WT, MyD88–/–, and TLR2x4–/– macrophages could form phagocytic cups around fluorescent apoptotic cells similar to those formed around E. coli–GFP (fig. S9). No differences were observed in phagocytosis of apoptotic cells among WT, MyD88–/–, and TLR2x4–/– macrophages. On average, 40% of macrophages had formed conjugates with more than one apoptotic cell within 1 hour, where the bulk of the apoptotic cell body remained outside the macrophage (Fig. 3A). At 2 hours, apoptotic cells colocalized with LysoTracker, and the entire apoptotic cell was internalized (Fig. 3B). In contrast, E. coli had already colocalized with LysoTracker in the majority of WT macrophages at 1 hour, but only in 40 to 50% of MyD88–/– and TLR2x4–/– macrophages (as discussed for Fig. 2A). At 2 hours, >80% of MyD88–/– and TLR2x4–/– macrophages now showed colocalization of E. coli with LysoTracker, whereas most E. coli–GFP in WT macrophages were degraded (Fig. 3C).

Fig. 3.

The kinetics of phagosome maturation is determined by the engagement of TLRs during phagocytosis. Representative fluorescence and differential interference contrast (DIC) micrographs showing phagocytosis of carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled apoptotic cells at 30 min (A) and at 2 hours (B) by WT, MyD88–/–, or TLR2x4–/– peritoneal macrophages. (C) Phagocytosis of E. coli–GFP at 2 hours. (D) Percent of WT, MyD88–/–, or TLR2x4–/– peritoneal macrophages containing phagolysosomes quantified by fluorescence microscopy. Percentages are means of four separate experiments. (E) Phagolysosomal fusion kinetics in WT macrophages during phagocytosis of E. coli–GFP (1:10), apoptotic cells (1:1), apoptotic cells with 10 ng/ml LPS, or apoptotic cells with unlabeled E. coli (1:10), as stated in the legend. (F) Simultaneous phagocytosis of either Texas Red–conjugated E. coli or S. cerevisiae with CFSE-apoptotic cells by WT BM-derived macrophages at 30 min or 2 hours as indicated. (G) Simultaneous phagocytosis of E. coli–GFP with CFSE-apoptotic cells, distinguished by size and shape, by WT macrophages at 30 min. In (A, B, C, and G), lysosomes were loaded with LysoTracker red. In (C, F, and G), nuclei were stained with DAPI. All micrographs are compiled serial Z-stacks 0.4 μm apart and represent at least two experiments. All scale bars, 10 μm.

The pattern and kinetics of phagocytosis of apoptotic cells was markedly different from that of E. coli (11). Phagosomes containing E. coli colocalized with acidic compartments at an increased rate in WT compared with MyD88–/– and TLR2x4–/– macrophages (Fig. 3D). In contrast, the maturation rate of phagosomes containing apoptotic cells was the same in all macrophages (Fig. 3D, also see Fig. 3, A and B); these findings emphasized that TLR- and MyD88-deficient macrophages did not have a generalized defect in phagocytosis. E. coli–LysoTracker co-localization was evident within 30 min in WT macrophages (Fig. 3D), whereas apoptotic cell–LysoTracker colocalization occurred after 1.5 to 2 hours (Fig. 3B). These data suggest two modes of phagosome maturation, constitutive and inducible, with the inducible mode controlled by TLRs.

Lysosomes acquire enhanced antimicrobial activity and antigen proteolysis after phagocyte activation (6, 15, 20). In the absence of TLR signaling, MyD88–/– macrophages were impaired in bacterial killing compared with WT (fig. S10A). Furthermore, MyD88–/– dendritic cells were less efficient in presenting bacterial antigens for CD4 T lymphocyte activation (fig. S10B). Thus, phagosome fate controlled by TLRs has functional consequences, some of which may reflect the kinetics of maturation.

We next asked if the inducible mode of phagosome maturation was phagosome autonomous or whether it could also be triggered by TLRs activated at the plasma membrane or a different phagosome. We followed maturation of phagosomes containing apoptotic cells with or without stimulation with lipopolysaccharide (LPS). Stimulation of TLR4 with LPS was not sufficient to trigger inducible phagosome maturation (Fig. 3E). Simultaneous phagocytosis of apoptotic cells and bacteria showed that they did not colocalize, which suggested that they were internalized into separate phagosomes (Fig. 3F). Particle size was not a factor, because simultaneous phagocytosis of apoptotic cells and Saccharomyces cerevisiae, which also engages TLRs (8), did not result in colocalization either (Fig. 3F). At 1 hour and within the same cell, only phagosomes containing bacteria colocalized with LysoTracker at the inducible rate, and the rate of maturation of apoptotic cell–containing phagosomes was unaffected (Fig. 3G). Thus, the TLR-induced signal appeared to emanate from phagosomes containing cargo that engaged TLRs. Those phagosomes matured at an inducible rate, whereas phagosomes containing apoptotic cells matured at a constitutive rate. This rate was not influenced by the activation of TLR signaling within the same cell from a different phagosome that carried bacteria, or from the plasma membrane where the TLR was engaged by a nonparticulate ligand like LPS. This suggested that phagosome maturation was stimulated by a TLR signal that was spatially confined such that only phagosomes containing cargo that engaged TLRs were subject to inducible maturation.

Which TLR signaling pathway is responsible for inducing phagosome maturation? p38 mitogen-activated protein kinase (MAPK), which is activated by TLRs (8) (fig. S11), can modulate the rate of endocytic traffic by regulating activity of guanyl-nucleotide dissociation inhibitor (GDI) on Rab proteins (21, 22). Specific p38 inhibitors impaired the ability of macrophages to efficiently phagocytose E. coli, as compared with controls (Fig. 4A). Furthermore, in the presence of p38 inhibitors, but not inactive control, no significant colocalization of E. coli and LysoTracker was detected, which indicated a block in phagosome maturation (Fig. 4, B and C). Thus, MyD88-dependent activation of p38 is involved in TLR-induced phagosome maturation.

Fig. 4.

p38 MAP kinase inhibitors impair phagocytosis and inhibit phagosome maturation. (A) WT BM-derived macrophages treated with the indicated p38 inhibitors or inactive control SB202474 and then given E. coli–GFP at a ratio of 1:100. FACS histogram overlays show red, no E. coli; dashed, inhibitors but no E. coli; open, E. coli; tinted, E. coli and inhibitors. Plots represent three experiments. Unaffected class II levels show intact cell integrity in the presence of inhibitors. (B) Fluorescence micrographs of WT peritoneal macrophages treated with the indicated p38 inhibitors or inactive control, shown 30 min after the addition of E. coli–GFP at a ratio of 1:10. Lysosomes were loaded with LysoTracker red before inhibitors were added. Nuclei stained with DAPI. All micrographs are compiled Z-stacks 0.4 μm apart. Scale bar, 10 μm. (C) Quantification by fluorescence microscopy of the percent of WT peritoneal macrophages containing phagolysosomes with or without the indicated p38 inhibitors or inactive control at 30 min. Percentages are means for three independent experiments.

Finally, many intracellular pathogens evade phagolysosomal fusion in macrophages by using a variety of strategies (23). Our findings identify TLR-MyD88-p38 signaling as a potential pathway that may be targeted to avoid phagolysosomal fusion.

Supporting Online Material

www.sciencemag.org/cgi/content/full/304/5673/1014/DC1

Materials and Methods

Figs. S1 to S11

Table S1

References and Notes

References and Notes

View Abstract

Navigate This Article