Enhanced Dendritic Cell Antigen Capture via Toll-Like Receptor-Induced Actin Remodeling

See allHide authors and affiliations

Science  20 Aug 2004:
Vol. 305, Issue 5687, pp. 1153-1157
DOI: 10.1126/science.1099153


Microbial products are sensed through Toll-like receptors (TLRs) and trigger a program of dendritic cell (DC) maturation that enables DCs to activate T cells. Although an accepted hallmark of this response is eventual down-regulation of DC endocytic capacity, we show that TLR ligands first acutely stimulate antigen macropinocytosis, leading to enhanced presentation on class I and class II major histocompatibility complex molecules. Simultaneously, actin-rich podosomes disappear, which suggests a coordinated redeployment of actin to fuel endocytosis. These reciprocal changes are transient and require p38 and extracellular signal–regulated kinase activation. Thus, the DC actin cytoskeleton can be rapidly mobilized in response to innate immune stimuli to enhance antigen capture and presentation.

Immature DCs respond to pathogen-derived products (1, 2) by initiating a program of maturation that induces their migration to lymphoid organs and culminates in the enhanced expression of major histocompatibility complex (MHC)–peptide complexes, costimulatory molecules, and cytokines necessary for T cell activation (35). Much of the DC response to microbial products is dependent on changes at the transcriptional level (6, 7). However, it is becoming clear that activation of TLRs can also trigger faster responses, for example, to the cellular machinery that supports antigen processing and presentation. Thus, endosomal/lysosomal proteolysis as well as membrane transport and fusion reactions are boosted (812), ubiquitinated proteins accumulate within specialized structures (13), and antigen presentation on both class I and class II MHC molecules is enhanced (14, 15). Nevertheless, many of these adaptations take several hours to become evident. A further well-documented characteristic of DC maturation is the progressive down-regulation of endocytosis (1618); yet, this seems paradoxical because, a priori, increased antigen capture upon encounter with a pathogen would seem desirable. To investigate this further in primary DCs, we undertook a detailed analysis of early events following TLR activation (19).

Murine bone marrow–derived DCs (BMDCs) and spleen-derived DCs (SDCs) were activated with TLR ligands, and their ability to take up the endocytosis marker fluorescein isothiocyanate (FITC)–dextran was assessed at early time points. Surprisingly, both types of DC accumulated several times as much FITC-dextran during a short pulse after lipopolysaccharide (LPS) treatment (Fig. 1A). However, this enhancement was transient, peaking after 30 or 45 min of LPS stimulation in BMDCs and SDCs, respectively (Fig. 1B, left). Longer-term down-regulation of endocytic capacity followed this acute stimulation (Fig. 1B, right). Ligands that stimulate DCs through other TLRs all transiently increased pinocytosis in BMDCs (Fig. 1C), but only if the appropriate TLR was expressed (fig. S1). We tested the lipopeptide Pam3CS(K)4 (TLR2), poly I:C, which is a mimic of double-stranded RNA (TLR3) and an unmethylated CpG-containing oligonucleotide (ODN1668) mimicking bacterial DNA (TLR9). TLR4- and TLR3stimulated FITC-dextran uptake persisted in DCs lacking the signaling adaptor MyD88, whereas the TLR9- and TLR2-dependent responses were abolished (Fig. 1D) (20). These results are consistent with other data indicating that TLR4 is coupled to both MyD88-dependent and independent signaling pathways, whereas TLR2 and TLR9 signaling is MyD88-dependent. TLR3 is known not to use the MyD88 adaptor (21).

Fig. 1.

Different TLR ligands transiently stimulate endocytosis in SDCs and BMDCs. (A) SDCs and BMDCs, with or without prior exposure (30 to 60 min) to 50 ng/ml LPS, were incubated with FITC-dextran for 10 min, washed, and analyzed by fluorescenceactivated cell sorting. (B) As in (A), but FITC-dextran uptake (10 min pulse) was measured after different times of acute (left) or long-term (right) LPS treatment. (C) BMDCs were exposed to oligonucleotides ODN 1668 or control ODN 1720, poly I:C, or Pam3CSK, and FITC-dextran uptake was measured as in (A). Fold stimulation (mean fluorescence intensity values) is relative to FITC-dextran uptake in unstimulated DCs. (D) BMDCs from wild-type or MyD88-deficient mice were exposed to TLR ligands, and FITC-dextran uptake was assayed as in (A) and (B). Means ± SD of triplicates shown. (E) Combined phase contrast and fluorescence microscopy of DCs with or without LPS stimulation for 30 min, then exposure to FITC-dextran for 10 min. Scale bars, 10 μm.

Observation of DCs by video and fluorescence microscopy revealed the likely basis for enhanced FITC-dextran accumulation in LPS-exposed DCs: Membrane-ruffling activity was stimulated, leading to an enhanced accumulation of FITC-dextran–filled macropinosomes compared with control cells (Fig. 1E and movie S1). LPS-stimulated FITC-dextran uptake was abolished by cytochalasin D, which depolymerizes actin filaments (20). These results show that down-regulation of DC endocytosis by microbial products is actually preceded by a short-term enhancement of actin-dependent endocytic capacity.

To establish whether acute stimulation of antigen capture also enhanced antigen presentation, we pulsed DCs briefly (30 to 60 min) with antigen, either concurrently with LPS or for the same length of time but before LPS exposure (Fig. 2A). DCs were then chased for various times and cocultured with T cells. We assessed antigen presentation on class I MHC molecules using immune complexes containing ovalbumin (Ova-IC) (22) and on class II MHC molecules using tetanus toxin C fragment (TTCF). Coadministration of Ova-IC with LPS resulted in significantly enhanced cross-presentation of the ovalbumin peptide SIINFEKL to the T cell hybridoma B3Z compared with Ova-IC administration followed by LPS (Fig. 2B). This difference was not seen in DCs lacking TLR4 (Fig. 2C). Similarly, enhanced presentation on class II MHC was observed when TTCF was coadministered with LPS versus sequential administration of TTCF, then LPS (Fig. 2D). Similar results were obtained using the TLR2 ligand Pam3CS(K)4 in both SDCs and BMDCs (Fig. 2, E and F).

Fig. 2.

Acute stimulation of antigen capture leads to enhanced antigen presentation. (A) Protocols for coadministration versus sequential administration of antigen and TLR ligands. Presentation of the ovalbumin epitope SIINFEKL or the tetanus-toxin epitope SGFNSSVITYPDAQLVP was detected by using T cell hybridomas B3Z (B and C) or 5B12 (D to F) on class I or class II MHC, respectively (34). B3Z cells were added to cells fixed after a 6-hour chase; 5B12 cells were added to live DCs after antigen-pulse manipulations. T cell stimulation was measured either by β-galactosidase luminescence assay (B3Z) or by enzyme-linked immunosorbent assay measurement of interleukin-2 release (5B12). DCs were either from spleen [(C) and (E)] or from bone marrow [(B), (D), and (F)]. Antigen was either ovalbumin immune complex (Ova IC, expressed as concentration of antibody to ovalbumin) or TTCF. In (D), dendritic cell numbers were titrated after pulsing with 0.5 μg/ml TTCF. Means ± SD of triplicate determinations are shown.

TLR activation not only stimulated membrane-ruffling activity but also had a dramatic effect on F-actin–rich structures called podosomes. Podosomes are found in arrays on the ventral surface of macrophages (23, 24), osteoclasts (25), and DCs (2628). They are distinct from focal adhesions and are thought to be involved in cell migration and tissue invasiveness (24). A high proportion of DCs displayed clusters of podosomes (Fig. 3A), each encircled by a characteristic ring of vinculin (Fig. 3B). We expressed actin–green fluorescent protein (GFP) in living DCs and monitored the behavior of podosome clusters. After photobleaching, podosomes rapidly reincorporated GFP-actin, which demonstrates that in DCs, as in other cells (23, 25, 29), podosomes are sites of rapid actin turnover (Fig. 3C and movie S2).

Fig. 3.

DC podosome clusters are highly dynamic and acutely sensitive to TLR signaling. (A) Podosomes in SDCs. (B) Note characteristic actin core (red phalloidin staining) and peripheral (green) vinculin ring. Scale bar, 5 μm. (C) SDCs were transiently transfected with GFP-actin, and discrete areas (white dotted circles) were photobleached. Times are in seconds. (See also movie S2.) (D) Podosomes disassemble in LPS-treated cells. Scale bar, 10 μm. (E) Quantitation of TLR signaling–dependent podosome disassembly in MyD88 +/+ and –/– DCs scored at 40 min. (F) TLR4 signaling triggers a cycle of podosome disassembly and reassembly in living cells. Time points are in minutes, with LPS added at 13 min. (See also movie S3.)

Although podosomes were a stable feature of the actin cytoskeleton in immature DCs when monitored over many hours, TLR activation induced podosome disassembly by 30 min in most DCs (Fig. 3, D and E) (30). Again, this effect was MyD88-independent for TLR4 and TLR3 but MyD88-dependent for TLR9 and TLR2 (Fig. 3E) (20). Interestingly, a more extended analysis revealed that the destabilization of podosomes induced by TLR signaling was transient. Podosomes disappeared precipitously 10 to 20 min after LPS challenge but then began to reappear at about 40 min and had almost fully recovered after 1.5 to 2 hours, even though LPS was still present (Figs. 3F and 4A and movie S3). Broadly, the same results were obtained in BMDCs, except that the disappearance and recovery of podosomes was slower (Fig. 4B). This cycle of podosome loss and recovery correlated inversely with the acute phase of enhanced endocytosis (Fig. 4, A and B). Thus, the few cells that retained podosomes after 30 min of LPS stimulation generally had fewer macropinosomes. Conversely, cells with large numbers of macropinosomes seldom had podosomes (Fig. 4C).

Fig. 4.

Enhanced FITC-dextran uptake and podosome disassembly show reciprocal and reversible kinetics in SDCs (A) and BMDCs (B). (C) After 25 min of LPS, DCs retaining podosomes exhibit fewer pinosomes, and vice versa. FITC-dextran was included for the last 10 min. LPS-triggered podosome disassembly (D and E) and enhanced FITC-dextran uptake (F) is blocked in the presence of the combined ERK (PD184352) and p38 (SB203580) MAP kinase inhibitors. Typical images shown in (D) and quantitation in (E) and (F). Means ± SD of five experiments are shown in (E) and (F). Analysis was performed after 30 to 40 min of TLR activation. Scale bars, 10 μm.

TLR ligands are known to activate protein and lipid-kinase signaling pathways leading ultimately to de novo gene transcription. However, as expected given the rapidity of the effects observed, actinomycin D had no effect on either enhanced endocytosis or podosome disassembly (fig. S2). The PI3-kinase inhibitor wortmannin blocked LPS-stimulated pinocytosis, but also basal pinocytosis (20), presumably because PI3 kinase is required for the terminal stages of macropinocytosis (17, 31). It is difficult, therefore, to assess its role in LPS-stimulated pinocytosis. Wortmannin had no effect, however, on podosome disassembly (fig. S3). Both extracellular signal–regulated kinase 1/2 (ERK1/2) and stress-activated protein kinase 2 (SAPK2)/p38α mitogen-activated protein (MAP) kinases were activated by LPS in SDCs, and kinetics similar to the actin rearrangements were observed (fig. S3). Podosome disassembly, triggered by LPS, was completely blocked in the combined presence of PD184352, an inhibitor of MEK1 (the upstream activator of ERK1/2) and SB203580, an inhibitor of SAPK2/p38 (Fig. 4, D and E). These inhibitors did not block the appearance of DC maturation markers (fig. S4) and, notably, they did not inhibit rapid actin cycling through podosomes, demonstrating that they were not toxic (fig. S5). Because podosome disassembly and enhanced ruffling/macropinocytosis could be driven by distinct or similar signaling pathways, we tested the effect of MAP kinase inhibitors on LPS-stimulated FITC-dextran uptake. As shown in Fig. 4F, this was also substantially blocked by combined inhibition of p38 and ERK kinases. When tested individually, SB203580 and PD184352 partially inhibited LPS-stimulated endocytosis. SB203580 partially blocked podosome disassembly, whereas PD184352 had little effect (Fig. 4E).

Our results reveal previously unknown features of the TLR-induced DC maturation program at its earliest stages, including a transient phase of enhanced endocytosis that can be used to boost antigen capture and presentation. Although sampling of “self” by nonactivated DCs is emerging as an important mediator of tolerance under steady-state conditions (32), the capacity to up-regulate antigen capture under infectious conditions may favor presentation of pathogen-derived peptides. Our data further suggest that to fuel increased actin-dependent endocytosis, DCs disassemble other actin-rich structures, particularly podosomes. However, further studies are needed to demonstrate an obligatory link between these two events. Podosomes are still somewhat enigmatic elements of the actin cytoskeleton that grow, divide, and fuse with each other in a highly dynamic fashion at the leading lamella of macrophages (23) and in the differentiating osteoclast (25). Podosomes are proposed to play an important role in cell migration and tissue invasiveness (24), so it is particularly intriguing that TLR activation in DCs can induce dramatic perturbations in podosome stability and dynamics. Our data imply that optimal antigen sampling and DC migration are mutually exclusive events. Although podosomes returned after transiently disappearing, mature murine DCs (∼30 hours of LPS) lacked podosomes (20), consistent with studies in human DCs (28).

LPS and other TLR ligands are well known to activate MAP kinase cascades (33), but here we report that acute modulation of the DC actin cytoskeleton is a downstream consequence. How MAP kinase activation simultaneously regulates enhanced ruffling/ pinocytosis and podosome disassembly remains to be determined. We did not observe major effects on the activation state of Rac or Cdc42 (fig. S6), so it seems likely that additional regulators of actin dynamics are targeted by the MAP kinases. Inhibition of both ERK and p38 kinases was required to fully block TLR-triggered effects on actin, which suggests that key downstream substrates can be activated by either pathway or that distinct ERK and p38 substrates deliver the effects observed. Because podosomes are continually turning over their actin content, consumption of the available actin pool and/or associated cofactors after a primary stimulation of actin-dependent endocytosis might indirectly result in podosome disassembly. Alternatively, separate signals may drive enhanced ruffling/pinocytosis and podosome disassembly in DCs, which may nonetheless be interdependent through a finite actin pool.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

Movies S1 to S3


References and Notes

Stay Connected to Science

Navigate This Article