The Small GTPase Arf6 Is Essential for the Tram/Trif Pathway in TLR4 Signaling

Background: The small GTPase Arf6 affects LPS-induced cytokine secretion.

Results: Arf6 regulates transport of the TLR4 adaptor protein Tram and internalization of LPS.

Conclusion: Arf6 plays an essential role in regulating the Tram/Trif-dependent TLR4 pathway.

Significance: Knowing how TLR4 is modulated is crucial for understanding innate immunity.

Abstract

Recognition of lipopolysaccharides (LPS) by Toll-like receptor 4 (TLR4) at the plasma membrane triggers NF-κB activation through recruitment of the adaptor proteins Mal and MyD88. Endocytosis of the activated TLR4 allows recruitment of the adaptors Tram and Trif, leading to activation of the transcription factor IRF3 and interferon production. The small GTPase ADP-ribosylation factor 6 (Arf6) was shown to regulate the plasma membrane association of Mal. Here we demonstrate that inhibition of Arf6 also markedly reduced LPS-induced cytokine production in Mal^−/− mouse macrophages. In this article, we focus on a novel role for Arf6 in the MyD88-independent TLR4 pathway. MyD88-independent IRF3 activation and IRF3-dependent gene transcription were strictly dependent on Arf6. Arf6 was involved in transport of Tram to the endocytic recycling compartment and internalization of LPS, possibly explaining its requirement for LPS-induced IRF3 activation. Together, these results show a critical role for Arf6 in regulating Tram/Trif-dependent TLR4 signaling.

Introduction

Toll-like receptors (TLRs)³ are transmembrane receptors essential to our innate immunity (1). Ten human TLRs are known, all recognizing different pathogen-associated molecular patterns (PAMPs) of extracellular and intracellular pathogens. TLR4 recognizes lipopolysaccharides (LPS) of Gram-negative bacteria (2) while TLR2 recognizes bacterial lipoproteins (3). Upon binding of LPS, TLR4 signaling is initiated by the recruitment of adaptor proteins. Depending on the adaptor usage, TLR4 signal transduction goes via the Myeloid differentiation factor 88 (MyD88)-dependent or -independent pathway. The former pathway starts with the recruitment of MyD88 adaptor-like (Mal) to the plasma membrane, where it links MyD88 to the activated receptor via its Toll-IL-1 receptor (TIR) interaction domain (4–9). Further propagation of the MyD88-dependent signaling cascade results in activation of different transcription factors such as NF-κB and AP-1, inducing the production of several pro-inflammatory cytokines (10). The MyD88-independent pathway requires internalization of TLR4, promoting the binding of Tram (Trif-related adaptor molecule) to TLR4, which in turn recruits Trif (TIR-domain-containing adaptor-inducing interferon-β), resulting in the activation of interferon-regulatory factor 3 (IRF3) and the production of type I interferons (11–14). TLR4 is unique among the TLR family for its ability to use both the MyD88-dependent and -independent pathways (15).

Kagan et al. demonstrated that Mal binds to phosphatidylinositol 4,5-bisphosphate (PIP2) in the plasma membrane via an N-terminal PIP2-binding domain (7). The small GTPase Arf6 can regulate the activity of phosphatidylinositol 4-phosphate 5-kinase (PI5K), which results in PIP2 production at the plasma membrane (16). An Arf6 inhibitory peptide suppressed LPS-induced production of the chemokine KC, suggesting a role for Arf6 in regulating TLR4 signaling (7). Overexpression of a dominant negative mutant Arf6 (Arf6 T27N) led to a redistribution of Mal to Arf6 containing vesicles and a model was proposed where Arf6 controls TLR4 signaling via the MyD88-dependent pathway by regulating Mal localization (7). However, a later study failed to observe a role for Arf6 in LPS-induced cytokine production or maturation of primary mouse dendritic cells (17).

The functions of Arf6 are not limited to activation of PI5K. Arf6 regulates internalization, targeted delivery and recycling of endosomes (18). It is involved in cell migration through membrane ruffle formation and podosome formation (16, 17, 19). Arf6 also affects uptake of pathogens via endocytosis and recycling of β1 integrins (20), membrane delivery during phagocytosis, actin remodeling (21, 22), intracellular transport of MHCI molecules (21), and the control of Fcγ receptor-mediated phagocytosis (23, 24). Recently, Arf6 was shown to be essential for two receptors that are related to TLR4. Arf6 is required for TLR9 signaling via the regulation of cellular uptake of CpG oligodeoxynucleotides (ODN) (25). Arf6 also mediates interleukin-1β (IL-1β) induced vascular endothelium stability (26). Both TLR9 and the IL-1β receptor use MyD88, but not Mal for their signal transduction (25–27).

The function of Arf6 in intracellular trafficking processes and the role of Arf6 in TLR4-related receptor systems that do not require Mal, suggest that the function of Arf6 in TLR4 signaling may not be limited to Mal recruitment. In this study, we therefore further analyzed the role of Arf6 in TLR4 signal transduction. We indeed found that the function of Arf6 in TLR4 signaling is not limited to Mal recruitment. Arf6 is essential for NF-κB- dependent cytokine production, but also for the MyD88-independent pathway by controlling IRF3 phosphorylation and IRF3-dependent gene transcription. We demonstrate that this effect is related to the role of Arf6 in the internalization of LPS and LPS-induced transport of Tram.

MATERIALS AND METHODS

Plasmids

pcDNA3.1-Arf6 wild-type (WT) was provided by Dr. T. Roberts (Dana-Farber Cancer Institute, Boston) via Addgene (Addgene plasmid 10834) (28). pcDNA3.1-Arf6 (T27N) and pcDNA3.1-Arf6 (Q67L) were generated via the QuickChange^TM site-directed mutagenesis method (Stratagene) with the oligonucleotides 1–2 (T27N) and 3–4 (Q67L) listed in supplemental Table S1. The pNFconluc reporter was a gift from Dr. Alain Israel (Institut Pasteur, Paris, France). The Gal4-luciferase reporter (p55-UASG Luc), pEF-BOS-Gal4-IRF3, and pEF-BOS-Gal4-DBD constructs were a kind gift of Dr. T. Fujita (The Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan) (29). pCAGGS-E-Tram and pMET7-HA-Rab11a were generated by amplification of full-length Tram and Rab11a from HEK293 cDNA, via primers 5–6 and 7–8. After NotI/XhoI (Tram) or NotI/XbaI (Rab11a) digestion, the fragment was cloned in the pCAGGS (Tram) or pMET7 vector (Rab11a). pLVTHM-Arf6 shRNA was created by MluI/ClaI digest on pLVTHM and ligation of the Arf6 shRNA (oligonucleotides 9–10), pLVTHM-siREN2 (negative control shRNA targeting Renilla luciferase, referred to as control shRNA) was created by MluI/ClaI digest on pLVTHM and ligation of the annealed oligonucleotides 11 and 12. Lentiviral packaging (pCMVR8.74), pseudotyping (pMD2.G), pLV-tTR-KRAB-IRES-RED and pLVTHM constructs were provided by dr. D. Trono (Tronolab, CH-1015, Lausanne, Switzerland) through Addgene (plasmid 22036, plasmid 12259, plasmid 12250 and 12247, respectively) (30).

Cell Culture

HEK-Blue^TM hTLR4 cells (InvivoGen), HEK293T cells, immortalized Mal-knock-out macrophages (a kind gift of dr. Jonathan Kagan, Harvard Medical School, Boston, MA), immortalized Mal-knock-out macrophages transduced with wild-type Mal (31) and A549 cells were grown using DMEM with 10% fetal bovine serum and 50 μg/ml gentamicin (Invitrogen) or HEK-Blue^TM Selection (InvivoGen) for HEK-Blue^TM hTLR4 cells. HEK-Blue^TM hTLR4 cells are HEK293 cells stably expressing TLR4 and its co-receptors MD2 and CD14. Primary mouse cells and THP-1 cells were grown in the same circumstances in RPMI supplemented with 1 mm sodium pyruvate. All cells were maintained in a 5% CO₂ humidified atmosphere at 37 °C. The immortalized Mal-knock-out macrophages are C57/BL6 background cells characterized by Nagpal et al. (32). These cells do not show detectable Mal expression on Western blot (33) and display decreased TNFα production upon LPS or Pam2 stimulus (32). We previously transduced these immortalized macrophages with wild-type Mal, which increases their responsiveness toward TLR4 and TLR2 ligands (31, 34).

GAL4-IRF3 and NF-κB Luciferase Reporter Assay

IRF3-mediated transcription was tested with a GAL4-IRF3/GAL4-luciferase reporter system (29). For a typical luciferase reporter assay, 10⁴ HEK-Blue^TM hTLR4 cells were seeded in a black 96-well plate (Nunc), in triplicates per condition, 24 h before transfection with 200 ng of the indicated Arf6 mutant, 5 ng pCAGGS-E-Tram, 10 ng of pEF-BOS-GAL4-IRF3 or pEF-BOS-GAL4-DBD and 100 ng of Gal4-luciferase reporter using a standard calcium phosphate precipitation procedure. 48 h later, transfected cells were left untreated or stimulated with 100 ng/ml lipopolysaccharide (LPS) from E. coli K12 (InvivoGen) for 16 h. Luciferase activity from triplicate samples was measured by chemiluminescence with a TopCount luminometer (Canberra-Packard). NF-κB luciferase reporter assays were performed according to a similar protocol. 3 × 10⁵ HEK-Blue^TM hTLR4 or HEK293-TLR2 (HEK293T cells overexpressing TLR2) cells were seeded in a 6-well plate (Nunc) and transfected with 1.2 μg of the indicated constructs and 300 ng of pNFconluc reporter. 24 h later, transfected cells were seeded in black 96-well plates (8 × 10⁴ cells per well). The next day, cells were either left untreated or stimulated with 100 ng/ml LPS (HEK-Blue^TM hTLR4) or 1 μg/ml LTA (HEK293-TLR2) for 3 h.

Cell Viability Assay

Cell viability was determined using the CellTiter-Glo® Luminescent Cell Viability Assay (Promega) according to the manufacturer's instructions.

Primary Macrophage and Dendritic Cell Culture

Bone marrow-derived macrophages (BMDM) were differentiated from bone marrow progenitor cells of C57BL/6J mice that were cultured in IMDM containing 10% heat-inactivated FBS, 30% L cell-conditioned medium, 100 units/ml penicillin, and 100 μg/ml streptomycin at 37 °C in a humidified atmosphere containing 5% CO₂. Alternatively, C57BL/6J bone marrow progenitors were differentiated in vitro into bone marrow-derived dendritic cells (BMDC) by culturing them in RPMI containing 10% heat-inactivated FBS, 100 units/ml penicillin, and 100 μg/ml streptomycin supplemented with 1000 units/ml recombinant GM-CSF (VIB Protein Service Facility, Belgium) at 37 °C in a humidified atmosphere containing 5% CO_2. After 6 days of incubation, cells were collected and plated in 96-well plates in IMDM containing 10% heat-inactivated FBS and 1% nonessential amino acids, in the presence of antibiotics.

Arf6 Inhibitory Peptide and SecinH3 Assay

Arf6 inhibitory peptide was synthesized in-house according to the sequence described previously by Kagan et al. (GKVLSKIFGNKEMRIRQIKIWFQNRRMKWKK with an N-terminal myristoyl group, the cell-permeating domain of the Drosophila antennapedia protein is underlined) (7). A control peptide, in which the Arf6 sequence was scrambled, was also designed in-house with sequence GFGKNIEMSLKVRKIRQIKIWFQNRRMKWKK with an N-terminal myristoyl group. Peptides were solubilized in DMSO and diluted in serum-free media at a final concentration of 5 μm. The medium was sonicated for better solubilization of the peptides (Vibra Cell^TM, 3 mm tip, 100 Watt output, 2 s pulses for 30 s). Cells were pretreated with sonicated media:peptide mixture for 3 h and stimulated, in the presence of peptides, with 100 ng/ml LPS, 1 μg/ml LTA, 100 ng/ml Pam3CSK4, or 10 ng/ml Pam2CSK4 for 3 to 6 h (time frames respectively for tumor necrosis factor-alpha (TNFα) or IL-1β ELISA) or left untreated. Supernatants were collected, and the production of cytokines was determined by ELISA. A similar approach was used for SecinH3 assays where cells were pretreated with 12.5 or 25 μm SecinH3 (or the same volume of DMSO as control; final percentage DMSO 0, 5, or 1%, respectively) for 3 h and stimulated with 100 ng/ml LPS or left untreated for another 3 h (TNFα ELISA).

Quantification of TNFα, IL-1β, and RANTES Secretion via ELISA

For detection of secretion of mouse TNFα or human TNFα, IL-1β, and RANTES, cells were seeded at a density of 2 × 10⁴ cells per 96-well (differentiated with 1 μm phorbol 12-myristate 13-acetate (PMA) (Sigma) final concentration in case of THP-1 cells) and stimulated for respectively 3, 3, 6, or 24 h with TLR2 or TLR4 ligands or left untreated. The same procedure was used for THP-1 cells, immortalized Mal-knock-out macrophages, primary dendritic cells and primary macrophages. Supernatants were collected and assayed for human TNFα, IL-1β, or RANTES and mouse TNFα, using the DuoSET ELISA kit from R&D Systems, according to the manufacturer's instructions.

Transduction of THP-1 Cells for Inducible Arf6 Silencing

Lentiviral particles were produced using a classic calcium phosphate transfection (35). Briefly, 24 μg of pLV-tTR-KRAB-IRES-RED was cotransfected with 18 μg of pCMVR8.74 and 7.2 μg of pMD2.G in HEK293T cells seeded the day before in 75 cm² flasks (Nunc) in DMEM medium (Invitrogen) with 10% fetal bovine serum (Invitrogen). The supernatant containing the viral particles was harvested 48 and 72 h after transfection. 24 h before transduction, 10⁴ THP-1 cells were seeded into 24-well plates (Nunc) in 0.5 ml RPMI medium (Invitrogen) + 10% FBS. The virus-containing HEK293T supernatants were centrifuged for 3 min at 720 × g to remove cell debris, and the supernatant was subsequently filtered over a 0.45 μm nitrocellulose filter (Millex). The virus was concentrated by ultracentrifugation at 90,000 × g, 2 h 10 min, 4 °C in a SW28 rotor (Beckman), and resuspended in 100 μl of RPMI medium. THP-1 cells were transduced by adding the virus together with 8 μg/ml polybrene (Sigma) using a spin-infection protocol (1 h at 200 × g). Cells were cultured for 10 days after the addition of 1 ml of fresh medium. dsRED-positive cells were sorted using a DakoCytomation Mo-Flo fluorescence-activated cell sorter (Agilent Technologies). The resulting THP-1 pLV-tTR-KRAB-IRES-RED cells were supertransduced with pLVTHM-Arf6 shRNA or pLVTHM-control shRNA using the same transduction protocol. Sorting and selection of GFP-positive cells was performed after 96 h treatment with 10 μg/ml doxycyclin (DOX) (Duchefa Biochemie) to obtain the final THP-1-Arf6shRNA and THP-1-shRNA cells. After sorting, the cells were cultured for at least 2 weeks to restore Arf6 expression. For Arf6 silencing experiments, the THP-1-Arf6shRNA and THP-1-controlshRNA cells were differentiated with 1 μm PMA and treated with 10 μg/ml DOX for 96 h.

Cell Lysis and Immunoblotting

For Western blot analysis, total lysates from 6-well plates were prepared as previously described (36). The following primary antibodies were used: anti-Arf6 (3A-1) from Santa Cruz Biotechnology, anti-β-actin (AC-74 and A2066) from Sigma and anti-phospho-IRF-3 (Ser-396) (4D4G) from Cell Signaling Technology. Dylight 800- or Dylight 680-conjugated secondary antibodies (Pierce) were used. Targeted proteins on the blots were visualized using the Odyssey infrared imaging system (LI-COR Biosciences).

Immunofluorescence Staining and Confocal Microscopy

For immunofluorescence staining, 10⁵ HEK-Blue^TM hTLR4 cells were seeded on cover slips in 6-well plates and transfected after 24 h via standard calcium phosphate precipitation. For detection of Arf6 plus Tram, the cells were transfected with 100 ng of pCAGGS-E-Tram DNA and 500 ng of wild-type pMET7-HA-Arf6 DNA. For detection of Tram plus Rab11a, the cells were transfected with 250 ng of pMET7-HA-Rab11a plasmid plus 100 ng of pCAGGS-E-Tram and 250 ng pMET7 mock vector. To test the effect of Arf6 mutants, the cells were transfected with 500 ng of pMET7-HA-Arf6-T27N or pMET7-HA-Arf6-Q67L plasmids. The same procedure was used for immunofluorescence staining of A549 cells. For this assay, 10⁵ A549 cells were seeded on cover slips in 6-well plates and transfected after 24 h via X-tremeGENE HP DNA transfection reagent (Roche).

48 h after transfection, cells were rinsed with phosphate-buffered saline (PBS) and fixed for 15 min at room temperature in 4% paraformaldehyde. After three washes with PBS, cells were permeabilized with 0.1% Triton X-100 in PBS for 10 min and blocked in 1% bovine serum albumin (BSA) in PBS for another 10 min at room temperature. Samples were then incubated for 1 h at 37 °C with primary antibody. After four washes in PBS, cells were incubated for 20 min at room temperature with secondary antibody. Three additional washes with PBS were performed. Images were acquired using a 60 × 1.35 NA objective on an Olympus IX-81 laser scanning confocal microscope and analyzed using Fluoview 1000 software. To determine colocalization, the Pearson's correlation coefficient was calculated in Fluoview 1000 and colocalizing pixels were visualized (37, 38).

Primary antibodies used for staining were rat anti-HA antibody (3F10, Roche Diagnostics, Arf6 and Rab11a detection), goat anti-E (A190-132A, Bethyl Laboratories, Tram detection), and rabbit anti-golgin 245 (C-13, Santa Cruz Biotechnology). Nuclear DNA was stained with DAPI. Secondary antibodies used were donkey anti-rat Alexa Fluor 594, donkey anti-rabbit Alexa Fluor 594, and donkey anti-goat Alexa Fluor 488.

To test the effect of LPS stimulation, the HEK-Blue^TM hTLR4 and A549 cells were incubated for 30 min with 100 ng/ml LPS before fixation. To determine internalization of LPS, cells were incubated with 5 μg/ml Escherichia coli LPS serotype 055:B5 Alexa Fluor 488 conjugate (Invitrogen) for 1 h before fixation.

RNA Isolation, cDNA Synthesis, and Quantitative Real-time PCR Analysis

Total RNA isolation, cDNA synthesis, and quantitative real-time PCR were performed as described earlier (39). Relative mRNA transcript levels were assessed by real-time quantitative PCR using the LightCycler 480 Probes Master (Roche) in combination with the Universal ProbeLibrary probes (for a complete list of primers and probes, see supplemental Table S1). Each sample was run as a technical duplicate. EIF2B2 and PPARA were validated as most stable reference genes in our experimental setup using the geNorm software (Biogazelle). Both genes served as internal controls and were used to normalize for differences in each sample. Relative mRNA levels were quantified using LightCycler 480 Software 1.5 (Roche) and qBasePlus software (Biogazelle).

Statistical Analysis

Statistical analysis was performed using GraphPad Prism Version 5.00. Differences between groups were analyzed using a one-way analysis of variance followed by a Bonferroni posttest. Data were considered significantly different when p < 0,05.

RESULTS

Interfering with Arf6 Suppresses LPS-induced Cytokine Production

Peptides comprising residues 2–17 of Arf6 interfere with processes regulated by this GTPase (40). Kagan et al. showed that a myristoylated fusion peptide composed of residues 2–17 of Arf6 fused to the cell-permeating domain of the Drosophila antennapedia protein inhibits LPS-induced production of the chemokine KC in MEFs (7). We tested the effect of this peptide on LPS-induced secretion of IL-1β and TNFα in different cell types.

Treatment of PMA-differentiated THP-1 macrophages, bone marrow-derived dendritic cells (BMDC) and bone marrow-derived macrophages (BMDM) with this Arf6 inhibitory peptide suppressed LPS-induced production of IL-1β and TNFα (Figs. 1, A and B, and 3C) without affecting cell viability (data not shown). A scrambled peptide, serving as a control, had no effect on LPS-induced cytokine production, demonstrating the specificity of the Arf6 inhibitory peptide. Svensson and co-workers (17) reported that retroviral infection of BMDCs with Arf6 mutants does not affect intracellular production of TNFα, IL-6, and KC. In contrast, we find that treatment of BMDCs with Arf6 inhibitory peptide does suppress LPS-induced production of IL-1β and TNFα (Fig. 1B).

FIGURE 1.

An Arf6 inhibitory peptide and Arf6 silencing suppress LPS-induced IL-1β and TNFα secretion and mRNA expression in different cell types. PMA-differentiated THP-1 cells (A) and BMDCs (B) were treated with 5 μm Arf6 inhibitory peptide or 5 μm Arf6 scrambled peptide for 3 h. The treated cells were stimulated with 100 ng/ml LPS for 3 h (TNFα detection) (B) or 6 h (IL-1β detection) (A) or left untreated. TNFα and IL-1β in the medium were determined via ELISA in triplicate. C, PMA-differentiated THP-1-Arf6shRNA cells were treated with 10 μg/ml doxycyclin for 96 h or left untreated. Western blotting confirms the decreased Arf6 protein levels. The cells were stimulated with 100 ng/ml LPS for 6 h or left untreated. IL-1β in the medium was determined via ELISA in triplicate. Data represent the average ± S.D. of three (A), four (B), or five (C) separate experiments expressed as percentage of LPS stimulated in absence of peptide. For qPCR data (D), PMA-differentiated THP-1-Arf6shRNA or THP-1-controlshRNA cells were stimulated for 3 h with 0 or 100 ng/ml LPS after treatment with 10 μg/ml doxycyclin for 96 h. Data represent the average fold induction of mRNA expression ± S.D. in three experiments expressed as percentage of doxycyclin unstimulated. HEK-Blue^TM hTLR4 cells (E) were transiently cotransfected with an NF-κB luciferase reporter (pNF-conluc) and plasmids encoding wild-type Arf6, Arf6 T27N, Arf6 Q67L, or with a pcDNA3.1 vector (mock control). After transfection, cells were left untreated or were stimulated with 100 ng/ml LPS for 3 h. Western blot shows the expression level of Arf6 constructs. Data represent 100% × (fold induction of luciferase activity of sample/fold induction of luciferase activity of the LPS-stimulated pcDNA3.1 mock) ± S.D. of three experiments. **, p < 0,01: significantly different versus both Arf6 wild-type and mock (E). ***, p < 0,0001: significantly different versus both untreated and scrambled peptide (A, B) or versus without doxycyclin (C, D).

FIGURE 3.

An Arf6 inhibitory peptide suppresses LPS-induced cytokine production in Mal^−/− mouse macrophages before and after lentiviral reconstitution with wild-type Mal. Mal^−/− mouse macrophages (A), Mal^−/− mouse macrophages transduced with wild-type Mal (B) and BMDMs (C) were treated with 5 μm Arf6 inhibitory peptide or 5 μm Arf6 scrambled peptide for 3 h. Cells were stimulated with 100 ng/ml LPS or left untreated. TNFα in the medium was determined via ELISA in triplicate. Data represent the average ± S.D. of three experiments expressed as percentage of LPS stimulated untreated. ***, p < 0,0001: significantly different versus both untreated and scrambled peptide.

To confirm the results obtained with the Arf6 inhibitory peptide, we next tested the effect of Arf6 silencing on LPS-induced cytokine production. An inducible knockdown of Arf6 was obtained by transducing THP-1 cells with a pLV-tTR-KRAB-IRES-RED construct and supertransducing them with pLVTHM-Arf6 shRNA. Upon treatment of these THP-1-Arf6shRNA cells with 10 μg/ml doxycyclin for 96 h, Arf6 protein levels were reduced (Fig. 1C). Silencing of Arf6 resulted in a decrease of LPS-induced IL-1β protein production (Fig. 1C) and inhibited IL-1β and TNFα mRNA expression (Fig. 1D), suggesting that Arf6 interferes with TLR4-dependent processes that lead to transcription of these cytokines. Our data thus confirmed the effect of Arf6 inhibition on TLR4-dependent cytokine production (7). The effects of Arf6 on IL-1β and TNFα mRNA levels are probably linked to decreased NF-κB activation, as dominant negative and constitutively active Arf6 mutants strongly inhibit NF-κB activation in HEK-Blue^TM hTLR4 cells (Fig. 1E).

Interfering with Arf6 Does Not Affect TLR2-induced Cytokine Production

Neither the Arf6 inhibitory peptide, nor Arf6 silencing affected TLR2-mediated cytokine production, indicating that Arf6 inhibition specifically interfered with a TLR4-dependent process and not with a general cytokine production process (Fig. 2, A and B). NF-κB activation was also unaffected upon overexpression of Arf6 mutants in HEK293-TLR2 cells, further emphasizing the TLR4-specific effects of Arf6 (Fig. 2E).

FIGURE 2.

An Arf6 inhibitory peptide and Arf6 silencing do not influence TLR2 ligand-induced IL-1β and TNFα secretion and mRNA expression in THP-1 cells. A, PMA-differentiated THP-1 cells were treated with 5 μm Arf6 inhibitory peptide or 5 μm Arf6 scrambled peptide for 3 h. The treated cells were stimulated with 1 μg/ml LTA, 100 ng/ml Pam3CSK4, or 10 ng/ml Pam2CSK4 for 6 h or left untreated. IL-1β in the medium was determined via ELISA in triplicate. Data represent the average ± S.D. of three or four (PAM2CSK4) experiments expressed as percentage of untreated plus ligand. For qPCR data (B), PMA-differentiated THP-1-Arf6shRNA and THP-1-controlshRNA cells were treated with 10 μg/ml doxycyclin for 96 h or left untreated. The cells were stimulated with 1 μg/ml LTA for 3 h or left untreated. TNFα in the medium was determined via ELISA in triplicate. Data represent the relative fold induction ± S.D. of three experiments expressed as percentage of unstimulated. HEK293-TLR2 cells (E) were transiently cotransfected with an NF-κB luciferase reporter (pNF-conluc) and plasmids encoding wild-type Arf6, Arf6 T27N, Arf6 Q67L, or with a pcDNA3.1 vector (mock control). After transfection, cells were left untreated or were stimulated with 1 μg/ml LTA for 3 h. Western blot shows the expression level of Arf6 constructs. Data represent (fold induction of luciferase activity of sample/fold induction of luciferase activity of the LPS-stimulated pcDNA3.1 mock) × 100% ± S.D. of three experiments.

TLR2 uses Mal to initiate the MyD88-dependent TLR pathway. As Arf6 was proposed to affect TLR signaling by regulating Mal localization, the lack of an effect of Arf6 inhibition on TLR2-induced cytokine secretion was somewhat unexpected. However, Mal may be dispensable at higher TLR2 ligand concentrations (41).

An Arf6 Inhibitory Peptide Suppresses LPS-induced Cytokine Production in Mal^−/− Cells

To test the hypothesis that Arf6 inhibition influences TLR4 signaling via its effect on Mal localization (7), we examined the effect of the Arf6 inhibition in Mal^−/− cells (32), Mal^−/− cells transduced with wild-type Mal (31) and BMDMs. All three cell types were incubated with Arf6 inhibitory peptide and the effect on LPS-induced cytokine production was tested. Notably, we still observed a decrease in LPS-induced cytokine production in Mal^−/− cells (Fig. 3), showing that Arf6 inhibition can affect TLR4-induced cytokine secretion independently of Mal. As already suggested by Kagan et al. (7), Arf6 indeed seems to have multiple role in TLR4 signaling. In this article, we investigate one of these roles: a new role of Arf6 in the MyD88-independent TLR4 signaling pathway.

Blocking Arf6 Strongly Inhibits LPS-induced IRF3 Phosphorylation and IRF3-dependent Signaling

The MyD88-independent TLR4 pathway results in phosphorylation and activation of IRF3, followed by IRF3-dependent gene transcription of e.g. type I interferons (11, 13–15).

Incubation of PMA-differentiated THP-1 cells with Arf6 inhibitory peptide completely abrogated LPS-induced IRF3 phosphorylation (Fig. 4A), demonstrating a new role for Arf6 in regulating the TLR4/MyD88-independent pathway. We next tested the effect of overexpressing a dominant negative Arf6 mutant (Arf6 T27N, unable to bind GTP) and a constitutively active Arf6 mutant (Arf6 Q67L, GTPase-deficient) on IRF3 activation by LPS stimulation of HEK-Blue^TM hTLR4 cells (InvivoGen) (42, 43). The Arf6 mutants were cotransfected with a Gal4-luciferase reporter and Gal4-DBD (negative control) or Gal4-IRF3 construct. Overexpression of Arf6 T27N and Arf6 Q67L inhibited activation of this IRF3-reporter upon LPS stimulation (Fig. 4B). IFN-β, RANTES and ISG15 are IRF3-dependent genes (44, 45). The Arf6 inhibitory peptide strongly affected the mRNA expression of these IRF3-dependent genes, as determined by qPCR (Fig. 4C). Silencing of Arf6 by treatment of PMA-differentiated THP-1-Arf6shRNA cells with doxycyclin again leads to an inhibition of LPS-induced IRF3-dependent gene expression (Fig. 4D). This was also reflected in the decrease of LPS-induced RANTES production in doxycyclin-treated THP-1-Arf6shRNA (Fig. 4E). Taken together, these data demonstrate the importance of Arf6 in regulating IRF3-dependent TLR4 signaling.

FIGURE 4.

Blocking Arf6 inhibits IRF3 phosphorylation and IRF3-dependent gene expression in TLR4 signaling. A, PMA-differentiated THP-1 cells were treated with 5 μm Arf6 inhibitory peptide or 5 μm Arf6 scrambled peptide for 3 h. The treated cells were stimulated with 100 ng/ml LPS or left untreated for different time intervals. Phosphorylation status of IRF3 was checked via Western blot with an anti-phospho-IRF3 antibody. A representative experiment is shown (n = 3). B, HEK-Blue^TM hTLR4 cells were transiently cotransfected with a Gal4-luciferase reporter and a Gal4-IRF3 or Gal4-DBD construct (negative control) together with plasmids encoding Arf6 T27N, Arf6 Q67L, or a pcDNA3.1 vector (mock). After transfection, cells were left untreated or were stimulated with 100 ng/ml LPS for 16 h. Data represent (fold induction of luciferase activity of sample/fold induction of luciferase activity of the LPS stimulated pcDNA3.1 mock) * 100% ± S.D. of three experiments. For qPCR data (C), PMA-differentiated THP-1 cells were treated with 5 μm Arf6 inhibitory peptide or 5 μm Arf6 scrambled peptide for 3 h. The treated cells were stimulated with 100 ng/ml LPS or left untreated for 3 h. Data represent the relative fold induction ± S.D. of three experiments expressed as percentage of unstimulated. D, PMA-differentiated THP-1-Arf6shRNA or THP-1-controlshRNA cells were treated with 10 μg/ml doxycyclin or left untreated for 96 h followed by 3 h stimulation with 0 or 100 ng/ml LPS. mRNA expression was determined via qPCR. Data represent the average fold induction of mRNA expression ± S.D. of three experiments expressed as percentage of doxycyclin unstimulated. E, doxycyclin-treated and PMA-differentiated THP-1-Arf6shRNA or THP-1-controlshRNA cells were stimulated with 100 ng/ml LPS for 24 h or left untreated. RANTES in the medium was determined via ELISA in triplicate. Data represent the average ± S.D. of three experiments expressed as percentage of untreated. ***, p < 0,0001: significantly different versus mock (B), versus scrambled peptide (C) or versus without doxycyclin (D, E).

Arf6 Mutants Affect Internalization of LPS

The endocytosis of LPS and TLR4 is essential for the MyD88-independent TLR4 pathway. Without endocytosis, the adaptor proteins Trif and Tram are unable to interact and IRF3-dependent gene transcription is impaired (46). In a similar way, TLR9 requires internalization of its ligand CpG DNA. Arf6 regulates this internalization of CpG DNA and is essential for TLR9 MyD88-dependent signal transduction (25). We therefore tested whether Arf6 modulation also affects the internalization of LPS.

HEK-Blue^TM hTLR4 cells were transfected with dominant negative (T27N) or constitutively active (Q67L) Arf6 mutants and incubated with an Alexa Fluor 488 fluorescent conjugate of LPS. The localization of fluorescently labeled LPS in transfected cells was determined via confocal microscopy. Cells overexpressing Arf6 T27N display a decreased internalization of fluorescently labeled LPS (Fig. 5). Arf6 Q67L overexpression results in a significant increase of internalized LPS. This demonstrates that Arf6 is indeed important for LPS internalization via TLR4. Internalization of CpG and LPS are both enhanced by the Arf6 Q67L mutant. In contrast, the Arf6 Q67L mutant has opposite effects on TLR9 versus TLR4 signaling: it stimulates TLR9-mediated signaling in line with the increased CpG uptake, but blocks TLR4 Trif-dependent signaling (Fig. 4B), despite increased internalization of LPS. We therefore examined additional effects of Arf6 in the Trif-dependent pathway of TLR4.

FIGURE 5.

Arf6 mutants affect internalization of LPS. HEK-Blue^TM hTLR4 cells were transiently transfected with dominant negative (T27N), constitutively active (Q67L) and wild-type Arf6. A pcDNA3.1 construct was used as mock control. Cells were stimulated for 1 h with an Alexa Fluor 488 fluorescent conjugate of LPS (green). Cells were then fixed, permeabilized, and labeled with an anti-HA antibody for detection of Arf6 (red). For each mutant, 50 transfected cells were counted (A) and the position of fluorescently labeled LPS was determined. Average of two independent experiments are plotted ± S.D. ***, p < 0,0001: significantly different versus mock, wild-type Arf6 and Arf6 T27N. B, confocal image of membrane-associated (left) and internalized LPS (right). Color code: red, Arf6-HA, green, fluorescently labeled LPS. Scale bar: 5 μm.

Arf6 Is Involved in the Transport of Tram toward the Endocytic Recycling Compartment upon LPS Stimulation

In monocytes, TLR4 is present in the endocytic recycling compartment (ERC) where it colocalizes with Rab11a (47). 30 min after stimulation with E. coli particles, this TLR4 pool is recruited to E. coli phagosomes, together with Tram, Trif, and IRF3 in a Rab11a-dependent way (47). Arf6 plays roles in endocytic recycling, is present in the endocytic recycling compartment and can cooperate with Rab11 (18, 48–50). We therefore studied the role of Arf6 in the intracellular localization of Tram and the relocation of Tram upon LPS stimulation of TLR4 via confocal microscopy.

Tram is N-terminally myristoylated and is reported to localize at the plasma membrane in unstimulated cells (51). We find that wild-type Arf6 and Tram partially colocalize in the vicinity of the plasma membrane of LPS unstimulated HEK-Blue^TM hTLR4 cells (Fig. 6C1). Our data show a relocation of Tram from the plasma membrane to a juxtanuclear cell compartment upon 30 min of LPS stimulation in HEK-Blue^TM hTLR4 cells (Fig. 6A). This Tram-positive compartment is close to the trans-Golgi network, as shown by immunostaining of golgin 245 (Fig. 6B1). However, Tram and golgin 245 do not colocalize, as indicated by low Pearson's correlation coefficient (r) and visualization of colocalizing pixels (Fig. 5B1) (37, 38). Co-expression of Rab11a or Arf6 WT did not affect the LPS-induced juxtanuclear accumulation of Tram and revealed strong colocalization of the Tram-containing juxtanuclear compartment with Rab11a and Arf6 WT (Fig. 6, B2, B3). Together, these data show that LPS stimulation induces Tram relocation to a juxtanuclear Rab11a- and Arf6-positive cell compartment, close to the trans-Golgi network, which fits the description of the ERC (42, 52, 53). Overexpression of the dominant negative Arf6 T27N mutant impaired the colocalization of Arf6 and Tram, as was shown by the low Pearson's correlation coefficients in both unstimulated and stimulated HEK-Blue^TM hTLR4 cells (Fig. 6, C3, C4). Furthermore, Arf6 T27N altered the localization of Tram. Upon Arf6 T27N overexpression, Tram in unstimulated cells was not expressed at the plasma membrane anymore. In cells overexpressing Arf6 T27N, LPS stimulation still causes accumulation of Tram in an intracellular compartment, but the distance of this compartment to the nucleus is considerably larger, suggesting a defect in the transport of Tram to the ERC (Fig. 6, C3, C4). Overexpression of the constitutively active Arf6 mutant Q67L caused a redistribution of Tram in Arf6-positive vesicles in the cytoplasm in both unstimulated and stimulated conditions (Fig. 6, C5, C6). This way, Arf6 Q67L disturbs the transport of Tram to the ERC upon LPS stimulation. Similar observations of Arf6 mutants on the distribution of Tram were made in A549 cells, human alveolar epithelial cells endogenously expressing TLR4 (Fig. 7) (54, 55).

FIGURE 6.

Arf6 is involved in the transport of Tram toward the endocytic recycling compartment upon LPS stimulation in HEK-Blue^TM hTLR4 cells. For immunostaining, HEK-Blue^TM hTLR4 cells were transiently transfected and afterward fixed, permeabilized, and labeled with an anti-HA-tag antibody (for detection of Arf6 or Rab11a, red), an anti-E-tag antibody (for detection of Tram, green), or a golgin 245 marker (for detection of trans-Golgi network, red) and DAPI. Color code: blue, DAPI staining, green, E-tag staining, red, HA-staining or golgin 245 marker staining, and white, colocalizing pixels. Scale bar: 5 μm. Representative images are shown. HEK-Blue^TM hTLR4 cells were transfected with E-tagged Tram and stimulated for 30 min with 100 ng/ml LPS (A2) or left untreated (A1) to study relocation of Tram (green) upon LPS stimulation. After LPS treatment, E-tagged Tram (green) accumulates in a compartment in the vicinity of the trans-Golgi network, as visualized by staining for golgin 245 (red) (B1). However, hardly any colocalization between E-tagged Tram and golgin 245 is detected (B1, right panel) Transfection with HA-tagged Rab11a (red) (B2) or wild-type HA-tagged Arf6 (red) (B3) and LPS stimulation was used to investigate the nature of this compartment. The LPS-induced Tram-containing compartment colocalizes with Arf6 and Rab11 (B2, B3, right panel). The effect of wild-type Arf6, Arf6 T27N, and Arf6 Q67L on Tram localization was studied by co-transfection of Tram and Arf6 constructs (C) upon which the distance of Tram to the nucleus and Pearson's correlation coefficient for colocalization were calculated.

FIGURE 7.

Arf6 is involved in the transport of Tram toward the endocytic recycling compartment upon LPS stimulation in A549 cells. For immunostaining, A549 cells were transiently transfected and afterward fixed, permeabilized and labeled with an anti-HA-tag antibody (for detection of Arf6, red), an anti-E-tag antibody (for detection of Tram, green) and DAPI. Color code: blue, DAPI staining, green, E-tag staining, red, HA-staining. Scale bar: 5 μm. Representative images are shown. A549 cells were cotransfected with wild-type Arf6, Arf6 T27N, or Arf6 Q67L and Tram upon which the distance of Tram to the nucleus and Pearson's correlation coefficient for colocalization were calculated.

DISCUSSION

In this study, we further investigated the role for Arf6 in TLR4 signaling. We show that Arf6 modulation has Mal-independent effects on LPS-induced cytokine secretion and that Arf6 is important for the Trif-dependent pathway of TLR4.

Different TIR-domain-containing receptors seem to use Arf6 in different ways. The IL-1β receptor uses a MyD88-ARNO-Arf6 pathway to affect vascular endothelium cell-cell interactions, but Arf6 is not required for the IL-1β-induced NF-κB activation (26). We find that TLR2-dependent NF-κB activation is also independent of Arf6. In contrast, TLR9-dependent NF-κB activation is strongly dependent on Arf6, as Arf6 regulates the internalization of CpG DNA (25, 27). In this study, we show a similar effect of Arf6 on internalization of LPS, in line with a role of Arf6 in Trif dependent signaling of TLR4. However, Arf6 also plays an additional role in the cellular localization of Tram and in the LPS-induced migration of Tram to the endocytic recycling compartment, where Tram colocalizes with Arf6 and Rab11.

Husebye et al. reported the colocalization of TLR4 and Rab11a in the endocytic recycling system of monocytes and showed that Rab11a is required for transport of TLR4, Trif, Tram, and IRF3 to E. coli phagosomes (47). Arf6 and Rab11a are both involved in endocytic recycling and Arf6 and Rab11a cooperate in multiple processes, such as Fcγ receptor-mediated phagocytosis in macrophages (56), the completion of abscission in cytokinesis (48) and stimulation-dependent recycling of integrin β1 (49). We suggest that a similar cooperation of Arf6 and Rab11a is possibly involved in trafficking of LPS and TLR4 and activation of IRF3. In this respect, it will be interesting to test whether FIP3 and FIP4 also have a role in the trafficking processes related to TLR4 signal transduction. FIP3 and FIP4 are dual effectors for Arf6 and Rab11a, and FIP3, Arf6 and Rab11a can form a ternary complex in vitro (48, 50). FIP3 directly interacts with the dynein light intermediate chain 1 of the dynein 1 motor protein complex, which directs minus-end-directed microtubule-based transport of cargo associated with FIP3 (57). Interestingly, the TBK1 and IKKϵ kinases that phosphorylate IRF3 in the TLR4 Trif-dependent pathway are able to phosphorylate FIP3 and antagonize or redirect transport via FIP3 (58, 59). This led to the hypothesis that these kinases may regulate transport of TLR4 to phagosomes via FIP3 (58).

The different functions of Arf6 require a specific guanine exchange factor (GEF), which induces the release of GDP, and the appropriate GTPase activating protein (GAP), which stimulates the hydrolysis of GTP (60). It is unclear which Arf6 GAP and GEF proteins are involved in TLR4 signal transduction. AIP1 was proposed as a novel Arf6 GAP involved in TLR4 signal transduction, but the data on the Arf6 GAP activity of AIP1 were revoked (61). IL-1β can activate a MyD88-ARNO-Arf6 signaling pathway and the Arf6 GEF ARNO/cytohesin-2 was shown to interact directly with MyD88 (26). Zhu et al. predicted that the MyD88-ARNO-Arf6 pathway may apply for Toll-like receptors (26). SecinH3 is a small molecule antagonist of the cytohesins family of Arf6 GEFs, including ARNO (26, 62). SecinH3 had no effect on LPS-induced cytokine production (data not shown). Other specific Arf6 GEFs, not inhibited by SecinH3, are probably involved in TLR4 signaling.

In summary, our data firmly establish a role of Arf6 in TLR4 signal transduction in different cell types. Arf6 affects aspects of the MyD88-independent pathway. It regulates LPS internalization and LPS-induced relocation of Tram, which is required for the MyD88-independent TLR4 signaling cascade. Identification of the Arf6 effectors and the Arf6 GEFs and GAPs involved in the TLR4-mediated processes in this article will provide new insights in the regulation of intracellular transport of TLRs, their ligands and their signal transduction molecules.