LPS or ethanol triggers clathrin- and rafts/caveolae-dependent endocytosis of TLR4 in cortical astrocytes



Toll-like receptor 4 (TLR4) activation and signalling in glial cells play critical roles in neurological disorders and in alcohol-induced brain damage. TLR4 endocytosis upon lipopolysaccharide (LPS) stimulation regulates which signalling pathway is activated, the MyD88-dependent or the TIR-domain-containing adapter-inducing interferon-β (TRIF)-dependent pathway. However, it remains elusive whether ethanol-induced TLR4 signalling is associated with receptor internalization and trafficking, and which endocytic pathway(s) are used in cortical astrocytes. Using the adenoviral over-expression of TLR4GFP, confocal microscopy and the imagestream technique, we show that upon ethanol or LPS stimulation, TLR4 co-localizes with markers of the clathrin and caveolin endocytic pathways, and that this endocytosis is dependent on dynamin. Using chlorpromazin and filipin as inhibitors of the clathrin and rafts/caveolae endocytic pathways, respectively, we demostrate that TRIF-dependent signalling relies on an intact clathrin pathway, whereas disruption of rafts/caveolae inhibits the MyD88- and TRIF-dependent signalling pathways. Immunofluorescence studies also suggest that lipid rafts and clathrin cooperate for appropriate TLR4 internalization. We also show that ethanol can trigger similar endocytic pathways as LPS does, although ethanol delays clathrin internalization and alters TLR4 vesicular trafficking. Our results provide new insights into the effects of ethanol or LPS on TLR4 signalling in cortical astrocytes, events that may underlie neuroinflammation and brain damage.


The results demonstrate that ethanol or lipopolysaccharide (LPS) triggers toll-like receptor 4 (TLR4) endocytosis by caveolae and clathrin-dependent pathways in astrocytes. We proposed that while clathrin is the protein responsible for TLR4 internalization, caveolin-1/lipid rafts membrane microdomains are required for TLR4 signaling. The results provide new insights into the effects of ethanol on TLR4 signalling in astrocytes, events that may underlie neuroinflammation.

Abbreviations used









toll-like receptor

Toll-like receptors (TLRs) are pattern-recognizing receptors which enable the recognition of conserved structural motifs in a wide array of pathogens. Activation of TLRs triggers specific signalling pathways, leading to the induction of genes that encode inflammation-associated molecules and cytokines to allow the initiation of the anti-microbial response (Akira and Takeda 2004). Among these receptors, TLR4, which recognizes lipopolysaccharide (LPS), is the only TLR which, upon activation, recruits different adaptor proteins, triggering two different signalling pathways referred to as MyD88-dependent and TIR-domain-containing adapter-inducing interferon-β (TRIF)-dependent. The MyD88-dependent pathway is mediated by adaptors MyD88 and TIRAP (Mal), and it induces mitogen-activated protein kinase and nuclear transcription factor kappa B. The TRIF-dependent or MyD88-independent pathway uses adaptors trif-related adaptor molecule and TRIF, and through the activation of transcription factor interferon regulatory factor 3 (IRF3), it generates interferon-β (IFN-β) and IFN-inducible genes such as IP10. Both signalling events lead, in turn, to the induction of various immune and inflammatory genes (Akira and Takeda 2004), which play central roles in the innate immune response.

Endocytosis of receptors has emerged as an important regulatory mechanism for innate immune responses and in signal transduction processes (reviewed in Scita and Di Fiore 2010). In particular, TLR4 LPS-activation at the plasma membrane has been described to initiate the binding of Mal, leading to MyD88-dependent signalling, while the translocation of TLR4 into endosomes brings about the displacement of Mal and the binding of trif-related adaptor molecule, triggering TRIF-dependent signalling in mouse macrophages and Ba/F3 cells (Kagan et al. 2008; Tanimura et al. 2008; reviewed in McGettrick and O'Neill 2010).

Several internalization routes have been described in mammalian cells, including the clathrin-dependent and the rafts/caveolae-dependent endocytic pathways (reviewed in Le Roy and Wrana 2005). In clathrin-mediated endocytosis, the clathrin coat is dissociated and uncoated vesicles fuse with early endosomes to sort endocytosed molecules. Then, some receptors are recycled back to the plasma membrane, whereas others are transported to the Golgi apparatus for signal processing or are destined to lysosomes for degradation (Maxfield and McGraw 2004). However, the raft/caveolae endocytic pathway is mediated by caveolin-dependent membrane invaginations known as caveolae (Cheng et al. 2006; Parton and Simons 2007). Lipid rafts and rafts/caveolae are membrane microdomains rich in cholesterol and sphingolipids, and act as receptor signalling platforms that are important integrators of signal events and trafficking (Parton and Simons 2007). TLR4 endocytosis has been shown to be mediated by the clathrin-dependent pathway in HEK293 cells (Husebye et al. 2006) and by the caveolin-dependent pathway in the CHO cell line (Shuto et al. 2005). Some studies have demonstrated that lipid rafts are essential for LPS-induced TLR4 recruitment and receptor signalling (Triantafilou et al. 2002) since raft-disrupting agents inhibit LPS-induced TLR4 signalling.

We previously demonstrated that LPS and ethanol (EtOH) are capable of activating TLR4 signalling in astrocytes (Blanco et al. 2005; Blanco et al. 2008; Alfonso-Loeches et al. 2010) and microglial cells (Fernandez-Lizarbe et al. 2009). TLR4 activation has been associated with the translocation of TLR4 and signalling proteins into lipid rafts/caveolae upon LPS or EtOH stimulation, and treatment with lipid rafts disrupting agents abolishes the LPS- and ethanol-induced activation of the TLR4 signalling pathway (Blanco et al. 2008; Fernandez-Lizarbe et al. 2008). These results indicate the potential role of lipid rafts in TLR4 endocytosis and signalling, although the precise endocytic pathway involved remains unknown.

TLR4 activation and signalling in glial cells play critical roles in many neurodegenerative diseases (reviewed in Glass et al. 2010; Buchanan et al. 2010) and in alcohol-induced brain damage. Indeed, we demonstrated that chronic EtOH treatment causes gliosis, induction of inflammatory mediators, myelin dysfuctions and neural death (Alfonso-Loeches et al. 2010, 2012). TLR4-deficient mice are protected against the ethanol-induced activation of glial cells and the production of brain cytokines and neuronal death (Alfonso-Loeches et al. 2010, 2012). However, it remains to be clarified whether ethanol-induced TLR4 signalling is associated with receptor internalization and trafficking, and which endocytic pathway is used to promote TLR4 signalling in astrocytes.

Using an adenoviral vector (TLR4GFP), which induces a functional TLR4 over-expression in astrocytes, here we show that upon LPS or EtOH treatment, TLR4 can be endocyted by both clathrin- and rafts/caveolae-dependent pathways, and that EtOH treatment impairs TLR4 endocytosis and trafficking. We also provide evidence of how the disruption of rafts/caveolae blocks both Myd88-dependent and TRIF-dependent signalling, while the inhibition of clathrin-dependent pathway abolishes TRIF-dependent signalling. These results provide new insights into TLR4-mediated signalling and inflammatory response mechanisms in astrocytes, events which may underlie neurodegeneration, alcohol-induced neuroinflammation, and brain damage.



Female Wistar rats (Harlan Ibérica, Barcelona, Spain) were used for cell isolation. After mating, dams were placed into separate cages during the gestation period. Food and water were provided ad libitum. All the animal experiments were approved by the local Research and Ethics Committee of the Príncipe Felipe Research Center (CIPF). All the experimental procedures were carried out in accordance with the ARRIVE guidelines and the guidelines approved by the EC Council Directive (86/609/ECC) and by Spanish Royal Decree 1201/2005.

Primary culture of cortical astrocytes, microglia and treatments

Primary cultures of rat cortical astrocytes from 21-day foetuses or newborn pups (less that 12-h-old) were prepared as previously described (Guerri et al. 1990; Renau-Piqueras et al. 1989) (see Figure S1 for a detailed protocol). Cells were used after 10–11 days in culture. The purity of cultured astrocytes was assessed by immunostaining using anti-glial fibrillary acidic protein (GFAP) and anti-CD11b to label astrocytes and microglia, respectively. Astrocyte cultures were found to be 98% ± 0.5 GFAP-positive and 2% CD11b-positive (see Figure S2).

Rat microglial cultures were prepared as described previously (Fernandez-Lizarbe et al. 2009). Confluent mixed glial cultures were subjected to mild trypsinization (0.05%) in the presence of 0.25 mM EDTA and 0.5 mM Ca2+. This results in the detachment of astroglial cells and leaves a population of firmly attached cells identified as > 98–99% microglia, as determined by immunofluorescence using anti-CD11b, anti-GFAP and anti-MAP-2.

To assess EtOH or LPS stimulation, foetal bovine serum was replaced with 1 mg/mL bovine serum albumin in astrocyte cultures or N2-supplement (Gibco, Rockville, MD, USA) in microglial cultures 18 h before treatments. Then cells were stimulated with 50 mM of EtOH (Merck, Madrid, Spain) or 50 ng/mL of recombinant LPS (Escherichia coli 026:B6; Sigma-Aldrich, Madrid, Spain). For endocytosis inhibition, astrocytes were pre-incubated at 37°C for 30 min with different inhibitors: dynasore (DYN) (40 μM) or chlorpromazin (15 μg/μL) or filipin (5 ng/mL) (all obtained from Sigma-Aldrich). Next, they were stimulated with EtOH or LPS. At the time indicated for each experiment, cells were harvested and used for specific determinations.

Adenoviral construction and infection

A recombinant adenovirus was produced as described elsewhere (Pascual et al. 2008). Briefly, TLR4 mRNA was amplified from rat cortex cDNA with PfuUltra High-Fidelity DNA Polymerase (Stratagene, La Jolla, CA, USA). Primers for TLR4, forward: 5′- TTG AGC TCA TGA TGC CTC TCT TGC -3′ and reverse: 5′- TAA GGG CCC TCA GGT CAA AGT TGT TGC TTC -3′, amplified a predicted 2525-bp DNA fragment. The forward primer includes a SacI site, while the reverse primer includes an ApaI site (underlined). The agarose-purified PCR product was double-digested with ApaI/SacI and was directionally ligated into the pEGFP-C3 vector (Clontech). TLR4GFP construction was extracted from pEGFP-C3 using the NheI and XmaI sites, treated with Klenow enzyme to generate blunt ends, and it was subcloned into adenoviral shuttle vector pAC-CMVpLpA, which was previously digested with BamHI and was Klenow-treated. The correct orientation of the insert was confirmed by a sequence analysis. Plasmid construct pAC-CMVpLpA-TLR4GFP was co-transfected with vector pJM17 to obtain recombinant adenoviruses that were plaque-purified, expanded and titrated as previously described (Jover et al. 2001).

Cultured astrocytes were re-plated 24 h prior to infection at a confluence of 1.5 × 105 cells per mL. Infection was performed at 150 multiplicity of infection for 2 h at 37°C. Cells were then washed and fresh medium was added. At 48–72 h post-infection, the TLR4GFP protein expression was confirmed by fluorescence microscopy and cells were harvested according to the subsequent procedure.

Retrotranscription and standard PCR

Cells in 10-mm culture dishes were lysed in 1 mL of Tri-Reagent solution (Sigma-Aldrich) and RNA was isolated following the manufacturer's instructions. The amount of purified RNA was estimated by measuring absorbance at 260 nm and its purity was assessed by the 260/280 nm ratio in a Nanodrop device (ND-1000, NanoDrop Technologies, Thermo Fisher Scientific, Barcelona, Spain). RNA integrity was examined by agarose gel electrophoresis. The RNA of each sample (1 or 4 μg) was subjected to DNAase treatment (Invitrogen, Carlsbad, CA, USA) and was reverse-transcribed using the Transcriptor First Strand cDNA Synthesis kit (Roche Molecular Biochemicals, Indianapolis, IN, USA). The RNA (1 μg) from the infected astrocytes was used for standard PCRs to check the mRNA expression of the full-length TLR4GFP construction.

Quantitative real-time RT-PCR

cDNA was diluted by a factor of two and 1 μL of each sample was amplified in triplicate in a rapid thermal cycler (LightCycler Instrument; Roche Diagnostics, Indianapolis, IN, USA) in 10 μL of LightCycler 480 SYBR Green I Master mix (Roche Molecular Biochemicals) using 0.5 μM of each oligonucleotide. Primers were designed using the Primer Blast program (NCBI) by covering at least one intron to avoid false results because of the amplification of residual genomic DNA. Samples underwent 40–50× RT-PCR cycles (2 min at 95°C; 10 s at the melting temperature and 18 s at 72°C). For each primer pair, the melting temperature was optimized. The sequences of primer pairs used in this study were: CD11b, 5′- CTG GGA GAT GTG AAT GGA G-3′ (forward) and 5′- ACT GAT GCT GGC TAC TGA TG-3′ (reverse); IFNβ, 5′- TGG ACC CTC CAC ATT GCG TTC C-3′ (forward) and 5′- TCT TCT CCA TCT GTG ACG GGT GC-3′ (reverse); IL1β, 5′- TGA TGT TCC CAT TAG ACA GC-3′ (forward) and 5′- GAG GTG CTG ATG TAC CAG TT-3′ (reverse); IP10, 5′- CCG GAA TCT GAG GCC ATC AAG AG C-3′ (forward) and 5′- ATT GGG AAG CCT TGC TGC TGG-3′(reverse); MHCII, 5′- AGA GAC CAT CTG GAG ACT TG-3′ (forward) and 5′- CAT CTG GAG TGT TGT TGG A-3′ (reverse); PPIA, 5′- GCG TCT GCT TCG AGC TGT TTG C-3′ (forward) and 5′- ACC ACA TGC TTG CCA TCC AGC-3′ (reverse). The mRNA levels of housekeeping gene PPIA were used as an internal control for normalization. The relative quantification of the PCR products was done using the LightCycler 480 quantification software (Roche Molecular Biochemicals).

Immunocytochemistry and confocal microscopy

Infected astrocytes were plated on 16-mm diameter glass coverslips treated with poly-D-Lysine (Sigma-Aldrich). Cells treated with or without LPS or EtOH for 5, 30, 60, or 180 min were fixed with 4% formaldehyde in phosphate-buffered saline for 10 min and were blocked for 30 min with tris buffered saline and tween-20 containing 3% bovine serum albumin. Cells were then co-incubated with mouse or rabbit anti-GFP (1 : 50; Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-TLR4 (1 : 50; Santa Cruz Biotechnology), rabbit anti-cav-1 (1 : 50; Abcam, Cambridge, UK), mouse anti-clathrin (1 : 50; BD Bioscience, Madrid, Spain), mouse anti-β-COP (1 : 50; Abcam), mouse anti-EEA1 (1 : 100; Santa Cruz Biotechnology), or mouse anti-58 K (1 : 300; Santa Cruz Biotechnology) overnight at 4°C. Following extensive washes, appropriate AlexaFluor®-labelled secondary antibody (Invitrogen) mixes were used by avoiding cross-reactions between them. Negative controls were obtained by omitting primary antibodies. Live astrocytes were marked with Lysotracker® RedDND-99 (1/12 000; Molecular Probes, Eugene, OR, USA) for 30 min at 37°C and were visualized at 37°C with a warm stage apparatus (Leica, Mannheim, Germany). Cell nuclei were detected by incubation with Hoechst 33342 (1 : 20 000; Molecular Probes). Immunofluorescence images were collected by a confocal microscope (LCS Lite TCS- AOBA- SP2; Leica), analysed with LCS Lite v.2.61 (Leica, Mannheim, Germany) and quantified by a Metamorph software analysis (Molecular Probes). To analyse the random overlap, images were processed using the Adobe Photoshop CS2 software (Adobe Systems Incorporated, San José, CA, USA). At least one coverslip per treatment from three different cultures was assessed.

Western blot analysis

Cultured astrocytes were resuspended in lysis buffer (1% NP-40, 20 mM Tris-HCl pH 8, 130 mM NaCl, 10 mM NaF, 10 μg/mL aprotinin, 10 μg/mL leupeptin, 10 mM dithiothreitol, 1 mM Na3VO4 and 1 mM phenylmethylsulfonyl fluoride). An equal amount of cell lysate of each sample was loaded onto sodium dodecyl sulphate–polyacrylamide gel electrophoresis, blotted onto polyvinylidene fluoride membranes and was then blocked with 50 g/L dried milk in tris buffered saline and tween-20. Membranes were incubated overnight with the following antibodies: rabbit anti-calnexin (1 : 2000; Abcam), mouse anti-complex II (1 : 5000, Invitrogen), rabbit anti-laminA/C (1 : 1000; Cell Signaling Technology, Danvers, MA, USA), mouse anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH, 1 : 3000; Millipore Bioscience Research Reagents, Temecula, CA, USA), goat anti-TLR4 (1 : 50, Santa Cruz Biotechnology), or mouse anti-GFP (1 : 50, Santa Cruz Biotechnology). After washing, blots were incubated with the appropriate secondary antibody. Proteins were visualized with either the enhanced chemiluminescence system (ECL Plus, GE Healthcare, Madrid, Spain) or alkaline phosphatase conjugate (Sigma-Aldrich). Band densities were quantified using the AlphaImager 2200 software (Alpha Innotech, San Leandro, CA, USA).


Infected primary cultured astrocytes were treated with lysis buffer for 30 min on ice. Cellular extracts were pre-cleaned with 10 μL of protein A/G-agarose (Santa Cruz Biotechnology) for 1 h and were then incubated overnight with 2 μg of rabbit anti-GFP (Santa Cruz Biotechnology) or pre-immune serum. Then, protein A/G-agarose beads were added to the samples and were incubated for 2 h. Precipitated beads were washed, solubilized by boiling in sodium dodecyl sulfate buffer, and samples were analysed by western blot.

Subcellular fractionation

Astrocytes were lysed by being sequentially passed through a 25G and a 21G syringe in a 0.25 M sucrose solution. The cell lysate was firstly centrifuged to eliminate intact cells and nuclei (800 g, 10 min). Then, the supernatant was centrifuged (15 000 g, 10 min) to obtain the fraction containing mitochondria, endosomes, and lysosomes (pellet-1) and the cytosolic fraction. The cytosolic fraction was re-centrifuged at 50 000 g for 1 h to obtain the cytosol- and the microsomal fraction- (pellet-2) containing membranes, including endoplasmic reticulum membranes. Both pellet-1 and pellet-2 were resuspended in 0.25 M of sucrose solution, and after centrifugation at 15 000 g (10 min) or 30 000 g (60 min), respectively, pellets were dissolved in radioimmunoprecipitation assay buffer. To obtain the astrocytes homogenate, cells were resuspended with lysis buffer (1% NP-40, 20 mM Tris-HCl pH 8, 130 mM NaCl, 10 mM NaF, 10 μg/mL aprotinin, 10 μg/mL leupeptin, 10 mM dithiothreitol, 1 mM Na3VO4, and 1 mM phenylmethylsulfonyl fluoride).


The primary cultured TLR4GFP-expressing astrocytes (48 h post-infection) grown in 10 cm-diameter plates were stained with a specific organelle marker according to the manufacturer's protocols (Lysotracker®RedDND-99 (1/12.000), ER-TrackerRed (1,1/1.000) or Cholera Toxin Subunit B (CT-B)-Alexa Fluor® 594 (1/1.000), all obtained from Molecular Probes) and were resuspended in 150 μL of cold PBS. Cell imagery was done in an ImageStream multispectral imaging flow cytometer (Amnis Corporation, Seattle, WA, USA). A minimum of 10 000 total events was collected per sample. Classifiers were set to eliminate cell debris and clusters prior to the data analysis. After acquisition, a compensation matrix was applied to all the data to correct for spectral overlaps. All the analyses were completed in a population of spectrally compensated, single focused cells. The degree of co-localization between green fluorescent protein (GFP) and each probe was quantitatively assessed by the Bright Detail Similarity Algorithm (Beckman Coulter, CA, USA) available in the ImageStream analysis software (IDEAS® v.4.0.735; Amnis Corporation).

Flow cytometry

Cortical astrocytes in six-well plates were incubated with or without LPS or EtOH for 30 min, trypsinized, blocked with 1% mouse serum for 15 min and stained with the mouse monoclonal anti-TLR4-PE antibody (1 : 350, 76B357.1; Santa Cruz Biotechnology) for 30 min on ice. TLR4GFP-expressing astrocytes were stimulated, trypsinized and then directly incubated with 3 μg/mL of propidium iodide for 10 min in the darkness at (25°C). Cells were washed and immediately processed in a Cytomics FC500 flow cytometry (Beckman Coulter, Madrid, Spain). Data were analysed and plotted using the CytF500 software (Beckman Coulter, CA, USA).

Cytokine analysis

TNFα levels were determined in the cell culture media of infected astrocytes (72 h post-infection) treated with either LPS or EtOH for 24 h. TNFα levels were evaluated by ELISA using standard kits from Diaclone Research (Bionova Científica, Madrid, Spain) following the manufacturer's procedure.

Statistical analysis

Statistical analyses were performed using the GraphPad Software (La Jolla, CA, USA). Data represent mean ± SEM. p-values were calculated with a two-way anova followed by a Dunnet's post hoc test or a Newman–Keuls' post hoc test when comparing treatments with the control or treatments among them, respectively. A two-tailed Mann–Whitney test was used to statistically analyse the qPCR data. The number of independent experiments is indicated in parentheses (n).


Infection with adenoviral vector TLR4GFP induces a functional over-expression of TLR4 in astrocytes

To evaluate the possibility that EtOH can promote TLR4 endocytosis in cortical astrocytes, we prepared a recombinant adenoviral vector carrying a GFP-tagged TLR4 (TLR4GFP). To assess functional over-expression, we infected astrocytes and checked the size and integrity of mRNA and protein. We confirmed that infected primary astrocytes fully expressed both TLR4GFP mRNA (Fig. 1a) and the complete fusion protein (Fig. 1b). The immunofluorescence analysis also demonstrated that TLR4 and GFP signals co-located within the plasma membrane of infected astrocytes (Fig. 1c), suggesting that the adenoviral expression of TLR4GFP does not cause degradation of this protein within the cell.

Figure 1.

Adenoviral vector Ad-TLR4GFP induces a functional over-expression of toll-like receptor 4 (TLR4) in primary cultured rat astrocytes. (a) Full-length mRNA expression of the TLR4GFP construction by RT-PCR. Full-length mRNA is amplified only in infected astrocytes. (b) Full-length fusion protein TLR4GFP by immunoprecipitation, followed by western blotting. A representative immunoblot showing the expression and integrity of the TLR4GFP fusion protein, n = 2. (c) Co-localization of TLR4 (red) and GFP (green) signals in TLR4GFP-expressing astrocytes (n = 3). Scale bar = 5 μm. (d) Astrocytes were infected with Ad-GFP or Ad-TLR4GFP and were stimulated with lipopolysaccharide (LPS) for 3 h to detect IL1β mRNA by quantitative RT-PCR and for 24 h to detect the release of TNFα by ELISA (n = 3–9). ***p < 0.001 vs. each control.

Finally, to assess whether the adenoviral expression or the GFP tag alters the TLR4 function, uninfected or infected astrocytes were stimulated with LPS, and the TLR4 response was evaluated by measuring the IL-1β and TNF-α levels. Notably, LPS significantly increased the IL-1β and TNF-α levels in both the uninfected and infected astrocytes, indicating that infection or GFP tag does not alter the TLR4 response (Fig. 1d).

TLR4 GFP is internalized and sorted into intracellular astrocytic compartments in response to LPS or EtOH

To assess whether TLR4 can be internalized in response to LPS or ethanol, the astrocytes expressing TLR4GFP were stimulated with LPS or EtOH for 30 min and then fluorescence intensity was measured by flow cytometry. Considering that GFP fluorescence intensity diminishes with vesicular acidification (Bizzarri et al. 2009; Haupts et al. 1998; Kneen et al. 1998), we expected to obtain diminished GFP fluorescence upon receptor internalization and vesicular acidification during receptor trafficking (Johnson et al. 1993). The results in Fig. 2a demonstrate that stimulation with either LPS or EtOH significantly diminished the mean GFP fluorescence intensity in stimulated astrocytes when compared with non-stimulated cells, suggesting that TLR4GFP is internalized after stimulation with either LPS or ethanol. GFP fluorescence also diminished after LPS or EtOH stimulation using the ImageStream technology (data not shown). To further corroborate TLR4GFP internalization, we performed the same experiment, but TLR4GFP-expressing astrocytes were preincubated with an anti-TLR4-Phycoerythrin (PE) antibody to specifically measure TLR4 in the cell plasma membrane. The data in Fig. 2b confirm that either LPS or EtOH stimulation diminished mean TLR4-PE fluorescence intensity, which supports that TLR4GFP is internalized upon LPS or EtOH stimulation.

Figure 2.

Toll-like receptor 4 (TLR4) receptor is endocytosed after ethanol (EtOH) and lipopolysaccharide (LPS) stimulation. TLR4GFP-expressing astrocytes were treated with LPS (50 ng/mL) or EtOH (50 mM) for 30 min (a) GFP intensity was analysed by flow cytometry using propidium iodide (PI) to discard dead cells (n = 8) (b) Flow cytometry analysis of TLR4 surface staining on treated and untreated cells assessed with the anti-TLR4-phycoerythrin (PE) antibody shows a reduction of TLR4 from the plasma membrane after LPS or EtOH treatment (n = 6). (c) Subcellular fractionation of TLR4GFP-expressing astrocytes after LPS or EtOH stimulation. Fractions were analysed by immunoblotting for TLR4GFP levels and for specific subcellular markers of each fraction: Calnexin (Microsomes), Complex II (Endosomes), Lamin A (Nucleus) and GAPDH (cell homogenate) that were used as loading controls. A representative immunoblot of four different experiments and its quantification is shown. *or #p < 0.05, **or ##p < 0.01 vs. control.

To further determine TLR4GFP internalization, we assessed the subcellular distribution of TLR4GFP in the astrocytes stimulated with LPS or EtOH treatment for 5 and 30 min. As shown in Fig. 2c, both LPS and EtOH promoted a rapid increase (within 5 min) in the TLR4GFP levels in the endosomes fraction, containing endosomes, mitochondria and lysosomes, and in the microsomal fraction, containing Endoplasmic Reticulum (ER) membranes. Nevertheless, the intracellular trafficking of TLR4 depended on the treatment. Thus, whereas LPS induced the up-regulation of TLR4GFP in both the endosomal and microsomal fractions upon 5 min of LPS-stimulation and then decreased at 30 min, EtOH treatment induced a similar increase in the TLR4GFP levels in the endosomal fraction upon 5 and 30 min of EtOH stimulation, while it caused a higher TLR4GFP expression at 30 min than at 5 min in the microsomal fraction.

We also observed that either LPS or ethanol stimulation up-regulates the TLR4 levels in the homogenate. The mechanisms involved in the LPS-induced TLR4 expression are not clear because while some studies have reported an increase of TLR4 in both microglial and human mononuclear phagocytes (Bosisio et al. 2002; Fernandez-Lizarbe et al. 2013; Muzio et al. 2000), others have found a reduced TLR4 expression upon LPS treatment in mice macrophages (Medvedev et al. 2000; Nomura et al. 2000). Differences in the cell type used or LPS treatment might explain these contradictory results.

In summary, the above results indicate that TLR4GFP activation by either LPS or EtOH triggers TLR4 endocytosis, and that EtOH can alter or delay the intracellular trafficking of the TLR4 receptor.

EtOH or LPS promotes TLR4GFP endocytosis via the caveolae- and clathrin-dependent pathways

The TLR4 receptor has been shown to enter cells via clathrin-mediated endocytosis in HEK293 cells (Husebye et al. 2006). Nevertheless, previous work in our lab indicated that the TLR4 receptor rapidly translocates to lipid rafts/caveolae domains upon LPS or EtOH treatments in cultured astrocytes, suggesting that TLR4 can also be internalized via caveolae (Blanco et al. 2008). Therefore, to provide further insights into the TLR4 endocytic pathways in primary cultures of astrocytes, we performed immunofluorescence experiments in TLR4GFP-expressing astrocytes, and we then assessed the co-localization of the anti-GFP signal with different organelle markers of the two main endocytic pathways: the caveolin- and clathrin-dependent pathways. Specifically for caveolin-dependent endocytosis, we used: caveolin-1 (cav-1), 58K (Golgi marker) and β-COP (ER) marker. To evaluate clathrin-dependent endocytosis, we utilized the following markers: clathrin, EEA1 (early endosomes marker) and Lysotracker (in vivo lysosome marker). TLR4GFP-expressing astrocytes were then analysed by confocal microscopy.

The analysis of the co-localization of TLR4GFP with caveolin-dependent markers (Fig. 3 and Figure S3a) indicated that LPS prompted a rapid increase (5 min) in TLR4GFP/cav-1 co-localization, which persisted until 30 min of treatment. However for longer periods (60-180 min, see Figure S3a), the percentage of TLR4GFP/cav-1 co-localization lowered and coincided with an increase in TLR4GFP/58K co-expression, suggesting that TLR4 moves from the plasma membrane to the Golgi apparatus (GA) during later periods. Likewise β-COP, a protein associated with the non-clathrin-coated vesicles involved in the transport from the ER to GA (Pelkmans and Helenius 2002; Le Roy and Wrana 2005), was found to co-localize with TLR4GFP soon after LPS stimulation (5 min) and persisted up to 180 min later. These data suggest that TLR4 can be internalized via caveolae and can traffic from the plasma membrane to the endoplasmic reticulum and to the Golgi complex. When astrocytes were stimulated with EtOH (50 mM), the percentage of TLR4GFP/cav-1 co-localization at 5 and 30 min was lower than the results obtained with LPS stimulation. In addition, EtOH seems to alter or delay the TLR4 trafficking from the plasma membrane to the ER.

Figure 3.

Toll-like receptor 4 (TLR4) receptor is endocyted in a caveolin- dependent manner upon lipopolysaccharide (LPS) or ethanol (EtOH) treatment. TLR4GFP-expressing rat astrocytes were treated with LPS (50 ng/mL) or EtOH (50 mM) for 5, 30, 60, or 180 min (see Figure S2 to see 60 and 180 min) and then were fixed and stained for GFP (green), nuclei (blue) and specific markers for endocytosis (red): cav-1, 58K (GA marker) and β-COP (ER marker). A representative photomicrograph of at least three different cultures is shown. 30–110 cells with double staining were analysed by confocal microscopy. The random overlap (R.O.) analyses of each marker are presented as grey lines. **p < 0.01, ***p < 0.001 vs. control (unstimulated) for LPS treatment; #p < 0.05, ##p < 0.01 vs. control (unstimulated) for EtOH treatment. Scale bar = 25 μm and 10× for amplifications.

The confocal microscopy studies data also illustrated that both LPS and EtOH promote TLR4 internalization via clathrin (Fig. 4 and Figure S3b), as demonstrated by the fast cellular co-localization of TLR4/clathrin upon stimulation. Thus upon 5 min of LPS or EtOH treatment, TLR4GFP appeared at the plasma membrane and colocalized with clathrin (TLR4GFP/clathrin). TLR4GFP/EEA1 immunostaining appeared maximal up to 30 min. Then during subsequent time periods (60 and 180 min, see Figure S3b), TLR4GFP was sorted to the late endosomes/lysosomes, as demonstrated by the co-expression of TLR4GFP with Lysotracker. An analysis of the data obtained with ethanol-stimulated astrocytes revealed that EtOH delays the internalization of TLR4 via clathrin and slowed up its intracellular transport between the plasma membrane to EEA1, as shown by the delay in the co-localization of TLR4GFP with clathrin and EEA1.

Figure 4.

The toll-like receptor 4 (TLR4) receptor is endocyted in a clathrin-dependent manner upon lipopolysaccharide (LPS) or ethanol (EtOH) treatment. TLR4GFP-expressing rat astrocytes were treated with LPS (50 ng/mL) or EtOH (50 mM) for 5, 30, 60, or 180 min (see Figure S2 to see 60 and 180 min) and were then fixed and stained for GFP (green), nuclei (blue) and specific markers for endocytosis (red): clathrin and EEA1. For the Lysotracker® experiments, alive TLR4GFP-expressing cells were stimulated and stained with the lysosomal marker in red and were analysed under in vivo conditions. A representative photomicrograph of at least three different cultures is shown. 30–110 cells with double staining were analysed by confocal microscopy. The random overlap (R.O.) analyses of each marker are presented as grey lines. *p < 0.05, **p < 0.01, ***p < 0.001 vs. control (unstimulated) for LPS treatment; #p < 0.05, ##p < 0.01 vs. control (unstimulated) for EtOH treatment. Scale bar = 25 μm and 10× for amplifications.

To evaluate the significance of the above co-localization we analysed the co-localization of the TLR4GFP signal with each cell marker used after displacing one of the channel signals 1 μm on the x- and y-axes. The results of these analyses indicate that co-localization did not significantly change with either treatment (LPS or ethanol) or time (Figs 3 and 4, grey lines). Therefore, the results indicate that the co-localization of TLR4GFP with each marker does not simply result from a random overlap of both signals, but reflects real co-localization in response to LPS or EtOH treatment.

To further confirm the confocal microscopy results, we performed studies with the imagestream system, which combines the capabilities of microscopy and flow cytometry, and allows quantitative image-based cellular assays in large and heterogeneous cell populations. To evaluate the endocytosis pathway of TLR4GFP in primary culture astrocytes, we used fluorescently labelled cholera toxin subunit B (CT-B, a marker of caveolae-mediated endocytosis) and ER-Tracker (a marker of the Endoplasmic Reticulum) to assess the caveolin-dependent pathway. Lysotracker was used to evaluate the clathrin-dependent pathway. The co-localization results of the imagestream experiments indicate a significant increase in the TLR4GFP/CT-B and TLR4GFP/ER-Tracker co-expressions at 30 min upon LPS stimulation, and that co-immunostaining decreased at 60 min. The maximal co-localization of TLR4GFP/Lysotracker was observed at 60 min (Fig. 5). Taken together, the confocal microscopy and imagestream flow cytometry results suggest that TLR4GFP is endocytosed via the caveolin and clathrin-dependent pathways in the primary culture of astrocytes in response to LPS or EtOH stimulation.

Figure 5.

TLR4GFP co-localization with endocytic markers after lipopolysaccharide (LPS) or ethanol (EtOH) treatment by ImageStream Technology. TLR4GFP-expressing astrocytes were stained in vivo with CT-B (lipid raft marker), ER-Tracker (ER marker) or Lysotracker® (lysosome marker) in red. Then, cells were stimulated with LPS (50 ng/mL) or EtOH (50 mM) for 5, 30 or 60 min and were analysed by ImageStream flow cytometry (n = 4–7). The total number of cells in which the colocalization of TLR4GFP occurred with the different markers is: CT-B = 315–661, ER = 300–368 and Lysotracker = 115–189 (cells). *p < 0.05, **p < 0.01 vs. control.

EtOH or LPS activates the TLR4 TRIF-dependent pathway in cortical astrocytes

Our previous results demonstrated that both LPS and EtOH are capable of activating the MyD88-dependent pathway in cortical astrocytes (Blanco et al. 2005, 2008). However, it is uncertain whether the TRIF-dependent pathway is functional in cultured astrocytes. Previous studies have shown that TLR4 initiates nuclear factor kappa B activation from the plasma membrane, but that subsequent TLR4 translocation to the endosomes is required for IRF3 activation (Kagan et al. 2008; Tanimura et al. 2008). Therefore in view of the relationship between TRIF signalling and TLR4 endocytosis, we assessed whether this pathway is activated in cortical astrocytes stimulated with LPS or ethanol. To answer this question, the expression of two IRF3-activated genes, IFNβ and IP10, were evaluated in the astrocytes treated with LPS or EtOH for 180 min. The results in Fig. 6a show that both LPS and EtOH treatments significantly increased the IFNβ and IP10 mRNA levels in astroglial cells, although the expression levels of these genes induced by LPS were much higher than those induced by ethanol.

Figure 6.

TRIF-dependent pathway is activated in astrocytes upon lipopolysaccharide (LPS) or ethanol (EtOH) stimulation. (a) Non-infected astrocytes were stimulated with LPS (50 ng/mL) or EtOH (50 mM) for 3 h and gene expression was measured by quantitative real-time PCR. An increment in the IP10 and IFNβ mRNA levels was detected in response to both stimuli (n = 4–6). (b) CD11b and MHCII mRNA expression were evaluated in non-infected astroglial and microglial cultures in resting and LPS-stimulated cells (n = 4). The low expression and non-LPS-response of the CD11b and MHCII microglial markers in cultured astrocytes indicate lack of microglia contamination. *p < 0.05, **p < 0.01, ***p < 0.001 vs. astroglial control, ##p < 0.01, ###p < 0.001 vs. microglial control.

Although our highly enriched astroglial cells were 98% GFAP-positive (see 'Methods' and Figure S2), we attempted to eliminate the possibility that the expressions of IFNβ and IP10 were because of microglia contamination. Using quantitative real-time PCR, we measured the expression of microglia markers CD11b (constitutive and activated microglia) and MHCII (a marker of microglia activation) in both cultured astrocytes and primary cultures of microglia. As shown in Fig. 5b, whereas the expressions of CD11b and MHCII mRNA in astrocytes with or without LPS were very low, the CD11b mRNA levels in microglia cultures were 6 times higher than those in astrocytes cultures, and both CD11b and MHCII markedly increased upon LPS treatment. These results support the purity of our highly enriched astroglial cultures and demonstrate that both LPS and EtOH are capable of activating the TRIF-dependent pathway in primary cultured astrocytes.

Dynamin is necessary for TLR4 signalling in cortical astrocytes in response to LPS or EtOH

LPS-induced endocytosis of the TLR4 receptor appears essential for its signalling functions (see McGettrick and O'Neill 2010 for a review), we therefore sought to evaluate whether the endocytosis mediated by LPS or EtOH in astrocytes was associated with TLR4 signalling. To assess TLR4GFP endocytosis in astrocytes, and as receptor-mediated endocytosis is dependent on GTPase dynamin, a protein which plays a critical role in catalysing membrane fission in endocytosis (Hinshaw 2000), we treated cells with DYN, a specific inhibitor of dynamin. We first confirmed that DYN was able to inhibit TLR4GFP internalization by a flow cytometry analysis. After taking into account that GFP fluorescence intensity at the plasma membrane is higher than during intracellular vesicular transportation because of pH changes (Haupts et al. 1998; Kneen et al. 1998; Bizzarri et al. 2009), the results in Fig. 7A indicate that whereas stimulation with either LPS or EtOH significantly diminished superficial TLR4GFP, DYN treatment abolished the LPS- or ethanol-induced reduction of membrane TLR4GFP, suggesting an inhibition of TLR4GFP internalization. Furthermore, DYN treatment also abolished the increased mRNA expression of genes IL1β and IFNβ in response to LPS and EtOH treatment (Fig. 7b, c). These results indicate that dynamin is necessary for TLR4 endocytosis and MyD88- and TRIF-dependent signalling in response to LPS or EtOH in cortical astrocytes.

Figure 7.

Effects of inhibition of dynamin, rafts/caveolae and clathrin-mediated endocytosis on toll-like receptor 4 (TLR4) signalling. Cells were pre-treated with dynasore (DYN) (dynamin inhibitor) or chlorpromazine (CPZ) (clathrin-mediated endocytosis inhibitor) or filipin (FIL) (raft/caveolae-mediated endocytosis inhibitor) before stimulation with lipopolysaccharide (LPS) (50 ng/mL) or ethanol (EtOH) (50 mM). (a) TLR4 endocytosis was evaluated 30 min after stimulation by measuring GFP intensity in TLR4GFP-expressing cells by flow cytometry, excluding the PI-positive cells. The mRNA expressions of IL1β (b) and IFNβ (c) were evaluated 3 h after stimulation in non-infected astrocytes by quantitative real-time PCR (n = 3–7). *p < 0.05, ***p < 0.001 vs. Control; #p < 0.05 vs. LPS or EtOH treatment.

Inhibition of rafts/caveolae or clathrin-dependent pathways selectively affects TLR4 signalling

To provide further insights into TLR4 endocytic pathways in astrocytes upon LPS or EtOH stimulation, we pre-treated cells with inhibitors of the clathrin or rafts/caveolae endocytic pathways before stimulation with LPS or ethanol. Chlorpromazin (CPZ) was used as a specific inhibitor of the clathrin-dependent endocytosis (Wang et al. 1993; Yao et al. 2002; Ivanov 2008), while filipin (FIL) was utilized as a specific inhibitor of the caveolin-dependent pathway (Schnitzer et al. 1994; Orlandi and Fishman 1998; Singh et al. 2003). TLR4 endocytosis was evaluated by measuring the TLR4GFP expression at the cell plasma membrane (GFP intensity) in the presence or absence of CPZ and FIL by flow cytometry. The results in Fig. 7a depict how both inhibitors, CPZ and FIL, abolished LPS or ethanol-induced TLR4GFP internalization, as observed by the lack of GFP signal reduction when compared with stimulated cells in the absence of inhibitors. We next investigated whether the inhibition of TLR4 endocytosis abolished MyD88-dependent and TRIF-dependent TLR4 signalling. To this end, the IL1β and IFNβ mRNA levels were measured in the LPS- or ethanol-stimulated astrocytes with or without inhibitors. The results in Fig. 7b illustrate that the IL1β gene expression was affected only by caveolin pathway inhibitor FIL. However, the gene expression of IFNβ lowered by inhibiting the endocytosis pathway with either FIL or CPZ. These results indicate that the caveolae- and clathrin-dependent endocytic pathways were involved in TLR4 endocytosis in the primary cultures of cortical astrocytes. However, while CPZ abolished the TRIF-dependent pathway, as demonstrated by the reduction in the IFNβ expression, FIL abolished both MyD88-dependent and TRIF-dependent pathways, as illustrated by the lower IL1β and IFNβ expression, respectively. These results suggest that the disruption of lipid rafts might affect TLR4 internalization and signalling.

Rafts/Caveolae and clathrin cooperate in the internalization of the TLR4 receptor

We finally evaluated whether caveolae- and clathrin-dependent pathways could interact in the endocytosis of TLR4. For this purpose, we evaluated if the inhibition of rafts/caveolae pathway with FIL could affect TLR4/clathrin co-localization and whether inhibition of clathrin pathway with CPZ could impair TLR4/cav-1 co-localization. Our results demonstrate that when lipid rafts were disrupted with FIL treatment, the LPS-induced increase in TLR4GFP/clathrin co-localization was abolished and that the co-localization of these proteins went below the control levels. In addition, when the assembly of the clathrin-coated pits was inhibited by CPZ treatment, we observed that the increment in TLR4GFP/cav-1 co-localization diminished, although it was still significantly greater that the control values (Fig. 8a). These results indicate that while lipid rafts seemed necessary for clathrin-mediated TLR4 internalization, inhibition of the clathrin pathway by CPZ was not necessary, but affected TLR4GFP/cav-1 co-localization, which only occurred at the plasma membrane (Fig. 8b), suggesting that both rafts/caveolae and clathrin are linked to and cooperate in appropriate TLR4 internalization and endocytosis. Consequently, we observed an increment in cav-1/clathrin co-localization after LPS or EtOH stimulation, specifically at the plasma membrane (Fig. 8d). However, we were unable to find co-localization of cav-1 and EEA1 after LPS or EtOH stimulation (Fig. 8c).

Figure 8.

Clathrin and rafts/caveolae cooperate in toll-like receptor 4 (TLR4) internalization and endocytosis. (a) TLR4GFP-expressing astrocytes were treated with chlorpromazine (CPZ) or filipin (FIL) and were then stimulated with lipopolysaccharide (LPS) (50 ng/mL) or ethanol (EtOH) (50 Mm). Cells were then fixed and stained with GFP (green) and cav-1 (red) or clathrin (red) for 5, 30 or 60 min. Nuclei were stained with Hoechst (blue). The co-localization of either clathrin or cav-1 with TLR4GPF was analysed by confocal microscopy. A representative photomicrograph of at least three cultures is shown. **p < 0.01 vs. control for LPS treatment; #p < 0.05, ##p < 0.01 vs. control for EtOH treatment. The random overlap (R.O.) analyses of each marker are presented as grey lines. (b) The TLR4GFP/cav-1 co-localization within the plasma membrane or cytosol was evaluated by confocal microscopy in the cells stimulated with LPS or EtOH. (c) EEA1/cav-1 co-localization in cells with or without stimulation with LPS or EtOH. (d) Clathrin/cav-1 co-localization within the plasma membrane (arrows) or cytosol was evaluated by confocal microscopy in the cells stimulated with LPS or EtOH. Cav-1 (green),clathrin (red) and nuclei were stained with Hoechst (blue). A representative photomicrograph of at least three different cultures is shown. *p < 0.05, **p < 0.01 vs. control. Scale bar = 25 μm and 10× for amplifications.


Our previous results demonstrated that, by interacting with membrane lipid rafts, ethanol is capable of acting as a TLR4 agonist by triggering the TLR4 signalling cascade and cytokine production in glial cells (Blanco et al. 2005; Fernandez-Lizarbe et al. 2009; Alfonso-Loeches et al. 2010), effects which contribute to the neuroinflammation and brain damage induced by EtOH intake (Alfonso-Loeches et al. 2010, 2012). However, whether ethanol-induced TLR4 signalling is associated with receptor internalization and trafficking, and which endocytic pathway(s) is(are) used in glial cells, in particular in astrocytes, remain unknown. Using highly enriched cultured astrocytes, the present study provides evidence for the first time that, upon LPS or EtOH stimulation, TLR4 is internalized and endocytosed by the caveolae and clathrin pathways, and that each endocytic pathway differentially affects TLR4 signalling pathways. The results suggest that both endocytic pathways cooperate in TLR4 internalization.

Previous studies have shown that TLR4 can be endocytosed by either the clathrin-dependent pathway in HEK293 (Husebye et al. 2006) or the caveolin-dependent pathway in CHO cells (Shuto et al. 2005). However, whether both pathways coexist in the same cell remains unknown. Using an adenoviral vector, which induces a functional TLR4 over-expression in astrocytes, flow cytometry and subcellular fractionation, we demonstrate that both LPS and EtOH are able to trigger TLR4 internalization from the plasma membrane to distinct cellular compartments, including endosomes/lysosomes and the ER. Co-localization immunofluorescence studies further demonstrated that either LPS or EtOH triggers a rapid TLR4GFP/clathrin co-localization, followed by a TLR4GFP movement to early endosomes (TLR4GFP/EEA1 co-localization). Finally, the receptor is delivered to late endosomes/lysosomes, as demonstrated by the maximal co-localization of TLR4GFP with Lysotracker upon 60 min of stimulation in confocal and imagestream studies.

The present work supports the role of lipid rafts/caveolae in TLR4 internalization/endocytosis induced by LPS or EtOH in astrocytes. Indeed using confocal microscopy and imagestream flow cytometry, we observe that either LPS or EtOH triggers a rapid increase in TLR4GFP co-localization with caveolae markers, such as cav-1 and CT-B, which persists until 30 min. It is noteworthy that the image analyses of the confocal data reveal that an increase in TLR4GFP/cav-1 co-localization (Fig. 8d) occurs at the plasma membrane, but not in the cytosol, suggesting that the TLR4 receptor translocates to caveolin-1-rich membrane microdomains. These results agree with our previous findings, demonstrating that EtOH triggers TLR4 activation by interacting with lipid rafts/caveolae in astrocytes (Blanco et al. 2008), and they support the involvement of lipid rafts in LPS-induced TLR4 recruitment and signalling (Triantafilou et al. 2002). We also observe that TLR4 moves from the plasma membrane to the Golgi apparatus for longer periods, as demonstrated by an increase in TLR4/58K (GA marker) at 60–180 min.

Our results support the notion that lipid rafts/caveolae and clathrin cooperate for appropriate TLR4 internalization, based on: (i) cav-1/clathrin co-localize after LPS or EtOH stimulation only at the plasma membrane; (ii) inhibition of lipid rafts integrity, by using FIL, impairs clathrin-mediated TLR4 internalization; (iii) inhibition of the clathrin pathway by CPZ affects TLR4GFP/cav-1 co-localization time-course patterns, suggesting that clathrin is important for the TLR4 endocytosis of the lipid-rafts/caveolae complex. Taken together, the results suggest that cav-1 and clathrin-positive vesicles are generated soon after TLR4 internalization at/near the plasma membrane, and that then TLR4 is sorted to the EEA1-endosomes leaving the lipid rafts/caveolae. Indeed, we were unable to detect cav-1/EEA1 colocalization after stimulation.

We further show that the TLR4 endocytosis pathway is associated with signalling. Firstly, we demonstrate that TLR4 internalization and TRIF-dependent signalling are dynamin-dependent, which coincides with other studies in different cell types (Kagan et al. 2008; Aksoy et al. 2012). However, we also observe that DYN reduces Myd88-dependent signalling. In line with this, recent studies have related dynamin to the caveolae function (Yao et al. 2002; Ferguson et al. 2009). Astrocytes are cells remarkably rich in caveolae (Megias et al. 2000), which might explain the differential effects reported by other studies using other cell types. We also show that the disruption of lipid rafts, by using FIL (Orlandi and Fishman 1998; Singh et al. 2003), significantly decrease the IL1β and IFNβ mRNA expressions, suggesting that lipid rafts integrity is essential for both Myd88-dependent and TRIF-dependent signalling. Nevertheless, when astrocytes are treated with an inhibitor of clathrin-mediated endocytosis, CPZ (Ivanov 2008), the expression of IFNβ, but not of IL1β, is abolished, indicating that clathrin-dependent endocytosis is necessary for TLR4 TRIF-dependent signalling. These results indicate that lipid rafts integrity is crucial for TLR4 endocytosis and signalling, and they also link TRIF-dependent signalling with clathrin-dependent endocytosis. Indeed, several receptors have been described to employ distinct endocytic pathways to mediate different functions and signalling (Hartung et al. 2006).

On the basis of our TLR4 endocytosis and signalling findings, we hypothesize that upon LPS or EtOH stimulation, TLR4 translocates to membrane lipid rafts/caveolae and increases plasma membrane TLR4GFP/cav-1 co-localization. The TLR4 complex, embedded in a lipid raft/caveolae environment, is then endocytosed via clathrin-coated vesicles (CCV). These CCV mature into EEA1-positive cav-1-negative vesicles and sort TLR4 to different endocytic compartments, such as the GA, the ER and lysosomes. We propose that clathrin is the protein responsible for TLR4 complex internalization, while cav-1 is a physically/functionally related protein associated with the TLR4 membrane microdomain that is required for receptor signalling. This hypothesis might explain why lipid rafts integrity is necessary for clathrin internalization and for TLR4 TRIF-dependent signalling since the disruption of lipid rafts by FIL abolishes the initiation of clathrin endocytosis internalization.

The present findings also evidence that EtOH is capable of triggering similar endocytic pathways, as LPS does in astrocytes. Nevertheless, EtOH delays CCV formation, as demonstrated by slower TLR4GFP/clathrin co-localization upon EtOH stimulation, and also seems to delay TLR4 intracellular vesicular trafficking.. Our previous studies demonstrated that EtOH alters the actin cytoskeleton in astrocytes (Guasch et al. 2003; Minambres et al. 2006), affecting intracellular protein trafficking (Megias et al. 2000; Guerri et al. 2001). Therefore, alterations in the actin cytoskeleton in astrocytes might mediate an ethanol-induced delay in vesicular trafficking during endocytosis.

The present findings also demonstrate that the astrocytes expressed the TRIF-dependent pathway and that both LPS and ethanol are able to activate the IFN pathway. Nevertheless, the natural TLR4-ligand LPS stimulates IFN to a much greater extent than ethanol. The potential physiological role of the slight response induced by ethanol in this pathway is presently unknown, although ethanol may be capable of activating IFN signalling under non-infectious conditions, which seems to be a key step for inducible NO synthase gene expression and NO biosynthesis (Schilling et al. 2002). Indeed, ethanol induces inducible NO synthase by activating TLR4 signalling in astrocytes (Blanco et al. 2004).

Astrocytes, the most abundant glial cell population of the CNS, are important contributors to the inflammatory immune response within the brain in response to microbial challenges or injury. They display an array of receptors involved in innate immunity (see rev. Farina et al. 2007), including toll-like receptors (Bowman et al. 2003; Esen et al. 2004; Carpentier et al. 2005; El-Hage et al. 2011) and their response to LPS (Park and Murphy 1994; Bowman et al. 2003; Esen et al. 2004; Carpentier et al. 2005; El-Hage et al. 2011). Nonetheless, some studies have suggested that cultured astrocytes express neither TLR4 nor TLR4 signalling in the absence of microglia contamination (Holm et al. 2012; Barbierato et al. 2013). In this study, we used highly enriched cultured astrocytes (98% GFAP+) and we believe that the results obtained are mainly because of astroglial TLR4 internalization and signalling. In previous studies using the same astrocyte protocol, we observed, by western blotting, that EtOH as LPS is capable of recruiting TLR4 and IL-1RI into isolated lipid rafts, which suggests the expression and function of TLR4 in astrocytes (Blanco et al. 2008). Indeed, the expression of TLR4 in GFAP+ cells is also observed in the cultured astrocytes used herein (see Figure S2b). Activation of TLR4 by EtOH also occurs in vivo since EtOH intake promotes astrogliosis, the release of inflammatory mediators and TLR4-deficient mice are protected against ethanol-induced neuroinflammation and brain damage (Alfonso-Loeches et al. 2010). Therefore, the present results extend previous findings by demonstrating that EtOH is also capable of triggering internalization and TLR4 endocytosis

In summary, our findings shed light on the role of lipid raft/caveolae and clathrin in LPS or ethanol-induced TLR4 endocytosis in cortical astrocytes. By considering the role of glial TLR4 activation in neurodegenerative disorders and brain damage induced by ethanol, understanding TLR4 receptor internalization and cellular trafficking mechanisms will open up new ways to modulate cell signalling pathways and the control of receptor functions in astrocytes and in neurodegenerative disorders.

Acknowledgements and conflict of interest diclosure

We thank M. March and M.J. Morillo for their excellent technical assistance. In addition, we thank the Confocal Service and the Cytometry Services at the CIPF for their help. This work has been supported by grants from the Spanish Ministries of Science and Innovation (SAF-2009-07503), Spanish Ministries of Economy and Competitivity (SAF-2012-33747), Spanish Ministries of Health, Carlos III Institute, FEDER funds (RTA-Network) and PNSD (2010I037), and from the Generalitat Valenciana-Conselleria de Educación: PROMETEO/2009/072 and the M. Lautenschaläger Award.

The authors have no conflict of interest to declare.