Role of Human CD36 in Bacterial Recognition, Phagocytosis and Pathogen-Induced C-Jun N-Terminal Kinase (JNK) - Mediated Signaling ¹

Reagents

All media, sera, and antibiotics were obtained from Invitrogen (Carlsbad, CA). The MAPK inhibitors PD98059, SB203580, SP600125, peptide inhibitors, JNKI and JNKIII, and corresponding negative control peptides were purchased from EMD Biosciences (San Diego, CA). Inhibitors for NF kappa B and D-JNKi1 inhibitor (D-retro-inverso form of the cell-permeable JNK peptide inhibitor) were obtained from BioMol International (Plymouth Meeting, PA). All LPS preparations and S. aureus LTA were from Sigma (St. Loius, MO). Anti-CD36 monoclonal antibody FA16 was purchased from Abcam Inc. (Cambridge, MA). Synthetic amphipathic peptides were synthesized by a solid-phase procedure (23, 24). Peptide sequences were described in a previous report (25). Human high-density lipoproteins (HDL) and LDL were purchased from EMD Biosciences. Extensively oxidized LDL (oxLDL) was prepared by incubation with 5 µM CuSO₄ at 37 °C for 24 h as described in (26). All cell trackers and reactive fluorescent dyes were obtained from Invitrogen.

Cell Cultures

HeLa (Tet off) cells were transfected with Fu-GENE-6 (Roche Diagnostics, Indianapolis, IN), using the expression plasmid pTRE2 (Clontech, Palo Alto, CA), encoding a human CD36 protein (pTRE2-CD36). Cells were co-transfected with pTRE2-hCD36 and pTK-Pur (Clontech), using a 1:20 ratio, and selected with 400 µg/ml puromycin. Puromycin-resistant cells were screened for the expression of the human CD36 protein utilizing mouse anti-hCD36 antibody (Abcam) in Western blotting. The HEK293 cell line was stably transfected with CD36 pIRES-hrGFP-2a plasmid (Stratagene, La Jolla, CA) followed by selecting cells with the highest green fluorescent protein expression. Normal human skin fibroblasts (ATCC, Manassas, VA), hCD36-overexpressing HeLa cells, HeLa (Tet-off) cells (Clontech) as well as human embryonic kidney cell line HEK293 (ATCC, Manassas, VA), both wild type and CD36-overexpressing, were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), 2 mM glutamine, 100 IU/ml penicillin, 100 µg/ml streptomycin, and 100 µg/ml G418 at 37 °C in a 5% CO₂ humidified atmosphere. Rat CD36-null (CD36 (−/−)) and wild type (CD36 (+/+)) macrophages were isolated from bone marrow cells of spontaneously hypertensive Wistar-Kioto male rat (SHR/NCrl) and its control strain Wistar-Kyoto/NCrl (Charles River Laboratories), respectively. The macrophages were differentiated by growing them in RPMI 1640, 20% FCS, in the presence of 10 ng/ml of mouse macrophage colony-stimulating factor (M-CSF) and 10 ng/ml of mouse IL-4 for 7 days.

Fluor-labeled ligand uptake

Native and oxidized lipoproteins, peptides, and bacteria were conjugated with Alexa Fluor® 488, using a protein labeling kit (Invitrogen) following the vendor's instructions. LTA from S. aureus was labeled with BODIPY TMR-X, whereas S. minnesota Re 595 LPS was labeled with a BODIPY FL, SE labeling kit, both from Molecular Probes (Eugene, OR) following the manufacturer's suggested procedure with previously reported modifications (25, 27). All bacterial uptake and phagocytosis studies were performed by using DMEM containing 2 mg/ml bovine serum albumin (BSA) without antibiotics. HeLa cells were incubated with bacteria at ≈50 labeled bacteria per cultured cell. After a brief centrifugation of the plate to accelerate bacterial sedimentation, the cells were incubated at 37°C for 1 h and then washed extensively with phosphate-buffered saline (PBS), detached with Cellstripper dissociation solution (Mediatech, Herndon, VA), fixed with 4% paraformaldehyde, and analyzed by a fluorescence-activated cell sorter (FACS, model A; Hitachi). A similar protocol, excluding the centrifugation step, was used for uptake experiments with other fluoro-labeled ligands.

Confocal microscopy imaging

Confocal microscopy experiments designed to demonstrate bacterial co-localization with various intracellular compartments were conducted as reported before (5, 25, 27). Briefly, 8-well microscopy slides with cultured cells were added with Alexa 488-labeled bacteria, centrifuged (3000 × g for 5 min) to accelerate bacterial sedimentation, and further incubated for 30–60 min. After three washings with PBS to remove unattached bacteria, slides were incubated with fluorescent Lysotracker Red. After a 5- to 30-min incubation with trackers, slides were washed with PBS, fixed with 4% paraformaldehyde and sealed or viewed immediately without fixation. Visualization was achieved by using Alexa Fluor 488/594-labeled anti-mouse/rabbit IgG secondary antibodies (Invitrogen). To assess subcellular localization, images were obtained with a Zeiss 510 laser scanning confocal microscope (Zeiss, Jena, Germany), using a krypton-argon-Omnichrome laser with excitation wavelengths of 488 and 568 nm for Alexa-488 and Alexa-568 labels, respectively.

CD36 expression manipulations

For CD36 knockdown or overexpression, human skin fibroblasts were electroporated in the presence of On-Targetplus Smartpool SiRNA CD36 (Dharmacon) (5 µg/10⁶ cells) or hCD36 coding plasmid (2 µg/10⁶ cells), utilizing a Nucleofector II instrument and Nucleofector kit T following recommended procedures (www.amaxa.com). Cells treated with a non-targeting siRNA served as a negative control. Following the nucleofection, cells were cultured for 24 h in DMEM containing 10% FCS followed by overnight incubation in serum-free media before stimulant additions.

Detection of activated JNK1/2 by Western blot analysis

Wild type and CD36-overexpressing HEK293 cells were grown in 6-well culture plates to confluence. Before the JNK1/2 activation assay, the cells were incubated overnight in serum-free DMEM. The cells were stimulated for varying periods of time with LPS (10 ng/ml) or E. coli (2 ng/ml) at 37 °C. After stimulation, the culture medium was immediately aspirated; the cells were placed on ice and washed three times with ice-cold PBS. Afterwards, the cells were scrapped into 100 µl of lysis buffer (20 mM Tris-HCl, pH 7.5, 2 mM EDTA, 0.5% (v/v) Triton X-100, 1 mM NaF, 1 mM Na₃VO₄, 50 mM 2β-mercaptoethanol, 5 mM MgCl₂, 1 mM phenylmethylsulfonyl fluoride, and 2% (v/v) protease inhibitor mixture set III (EMD Biosciences). After a 10-min incubation on ice, the samples were sonicated for 7–8 s and centrifuged at 12,000 rpm for 10 min at 4 °C. The cell extracts were collected and mixed with the 2× SDS sample buffer. The samples were separated on SDS-PAGE in 8% Tris-glycine pre-cast gels (Invitrogen) and then transferred to nitrocellulose membranes. The membranes were blocked with Tris-buffered saline containing 0.1% Tween 20 and 5% (w/v) nonfat dry milk and further incubated either with the primary anti-phospho JNK1/2 antibodies or with antibodies that recognize both active and inactive JNK forms (BioMol International) overnight at 4°C. Immunoreactive bands were detected with a horseradish peroxidase-conjugated secondary antibody and enhanced chemiluminescence (GE Healthcare, Piscataway, NJ). Alternatively, the immunoreactive bands were detected using an alkaline phosphatase-conjugated secondary antibody and Western Blue Stabilized Substrate for Alkaline Phosphatase (Promega, Madison, MI).

Other procedures

Cytokine secretion was analyzed in culture media after a 20-h incubation utilizing a commercial enzyme-linked immunosorbent assay kits for human IL-8 (Invitrogen), rat IL-6 and rat TNF-alpha (Pierce Biotechnology). Bacterial preparations from different species were standardized according to the bacterial lysates’ protein content measured by the Bradford dye (Bio-Rad Laboratories, Hercules, CA) protein assay (28). CelLytic B Plus Kit (Sigma-Aldrich) was used for efficient bacterial lysis and protein extraction.

Uptake of FITC-labeled ligands is enhanced in hCD36 transfected vs. mock transfected HeLa cells

To test the functional activity of CD36 as a lipoprotein receptor, we measured cellular uptake of fluorescently labeled HDL, LDL, and oxLDL in HeLa cells transfected with hCD36 using FACS analysis. Compared to mock transfected cells, expression of hCD36 markedly increased the uptake of fluorescently labeled lipoproteins (Fig. 1). Synthetic amphipathic helical peptides have been previously shown to be the ligands for another CD36 family member – the Cla-1 (SR-BI) receptor (29). Fig. 1 shows that the uptake of Alexa 488-labeled L-37pA and D-37pA α–helical peptides was also increased in CD36-overexpressing vs. control cells. As previously reported by Stuart et al. (6), murine CD36 can act as a phagocytic receptor for Gram-positive and, to a lesser extent, Gram-negative bacteria in HEK 293 cells. According to Philips et al. (5), HEK293 cells transfected with murine CD36 are capable of taking up E. coli and to a lesser extent S. aureus. In order to see whether hCD36 has any preference for internalization of Gram-positive vs. Gram-negative bacteria, we evaluated the cellular uptake of the various Alexa 488-labeled bacteria in hCD36 transfected vs. control HeLa cells. We found the most pronounced increases in the uptake of the Gram-negative E. coli K-12 and K-25 strains as well as the Gram-positive Enterococcus faecalis (Fig. 2). Although the increases were less pronounced, internalization of two other tested Gram-negative bacteria - Klebsiella pneumoniae and Salmonella typhimurium - was also enhanced in CD36-overexpressing vs. control HeLa cells (Fig. 2). In contrast, the difference in S. aureus internalization between CD36 and mock-transfected cells was masked by a relatively high bacterial uptake in control cells, possibly due to the presence of CD36-independent mechanisms of bacterial uptake in HeLa cells. Altogether these results suggest that hCD36-mediated uptake is not selective for Gram-positive or Gram-negative bacteria, and hCD36 functions as a receptor capable of recognizing and mediating intracellular uptake of either type of bacteria.

Regarding the intracellular uptake of fluorescently-labeled LPS and LTA, we observed a 5–10-fold increase in internalization of Bodipy 558/568-LPS in CD36-transfected vs. mock HeLa cells, while the levels of Bodipy-LTA uptake were comparable in the two cell types (Fig. 1).

CD36 mediates bacterial phagocytosis in CD36-overexpressing HeLa cells

CD36-mediated bacterial uptake was further analyzed in immunofluorescence studies with confocal microscopy. hCD36-transfected HeLa cells were incubated with Alexa 488-E. coli (K-12 strain) followed by co-localization with the lysosomal marker, LysoTracker Red (Fig. 3A–C). The merged image (Fig. 3C) demonstrated co-localization of E. coli K-12 with the lysosomal compartment in hCD36-overexpressing HeLa cells. These results confirm the involvement of CD36 in bacterial phagocytosis and provide visual evidence that CD36-dependent bacterial uptake in HeLa cells is at least partially associated with lysosomal degradation of bacteria. No appreciable bacterium uptake was found in control hCD36-negative HeLa cells.

Bacterium and LPS-induced secretion of IL-8 is increased in hCD36-overexpressing HEK293 cells

We next tested the ability of various live bacteria to induce IL-8 secretion via a CD36-dependent pathway. All Gram-negative bacteria tested induced marked (5- to 20-fold) dose-dependent increases of IL-8 secretion in CD36-transfected HEK293 cells. Control cells demonstrated low or no IL-8 secretion upon exposure to the bacteria (Fig. 4). The response of CD36-overexpressing cells to Gram-positive bacteria was either weak or non-existent, except for a robust response induced by Listeria monocytogenes. Incubation of CD36-overexpressing HEK293 cells with increasing doses of LPS, a structural component of the cell wall of Gram-negative bacteria, resulted in an even more pronounced stimulation of IL-8 secretion than the treatment with whole bacterium preparations (Fig. 5A). Control cells were unresponsive to LPS in concentrations from 1 to 100 ng/ml. While higher doses of LPS were able to induce detectable levels of IL-8 in control cells as well, the responses were 10- to 20-times lower than those in CD36-transfected cells. Various preparations of LPS, derived from wild strains of E. coli (serotype 0111:B4 and 055:B5) or S. enterica (serotype Minnesota) as well as LPS Re 595 mutant (from S. enterica serotype Minnesota) were efficient stimulators of IL-8 secretion in CD36-HEK293 cells (data not shown). Unlike LPS, LTA was a less potent stimulator of cytokine secretion in CD36-transfected cells. Only a moderate IL-8 (1.5- to 1.7-fold) increase was observed in CD36-overexpressing cells upon LTA stimulation when compared to control cells (Fig. 5B). Another well-known CD36 receptor ligand, oxLDL, was recently reported to mediate signaling via a CD36-dependent pathway (17). In our experiments, oxLDL induced IL-8 secretion in CD36-HEK293 cells in a dose-dependent manner (Fig. 5C).

Bacterium and LPS-induced secretion of IL-8 is decreased by CD36 siRNA knockdown in human fibroblasts

It has been previously reported that monocytes from CD36 knockout mice differentiated into macrophage demonstrate a 30–50% decreased response to LPS and Gram-negative bacteria when compared to normal cells (6). To study the role of CD36 in the cells moderately expressing TLR2/4, we manipulated CD36 expression in human fibroblasts. As seen at Fig. 6, cells treated with CD36 siRNA demonstrated 40–50% decrease in IL-8 release when stimulated with LPS and Gram-negative E. coli. A similar decrease in IL-8 secretion was observed with Gram-positive S. aureus. Nucleofection with scrambled siRNA had no effect. Importantly, CD36 overexpression was associated with a 2.5–3-fold increase in IL-8 secretion levels following cell treatment with LPS and both bacteria - E. coli and S. aureus, when compared to control cells.

Bacterium-, LTA- and LPS-induced cytokine production is significantly impaired in CD36 (−/−) rat macrophages

To evaluate the role of CD36 in the inflammatory response to bacteria and their derived ligands in cells with high levels of TLR2/4 expression, we measured IL-6 and TNF-α secretion in CD36 (+/+) and CD36 (−/−) rat macrophages following a 20-h cell treatment with different inflammatory stimuli. CD36 (−/−) cells demonstrated significantly decreased levels of IL-6 (60–70% reduction) and TNF-α (40–60% reduction) upon stimulation with E. coli K12, S. aureus as well as their cell wall components, LPS and LTA, when compared to CD36 (+/+) macrophages (Fig 7.). Similar levels of cytokine production impairment occurred in the absence of a functional CD36 with both Gram-negative and Gram-positive bacteria, as well as with LPS and LTA. These data indicate that CD36 plays an important role in macrophage cytokine response to both Gram-negative and Gram-positive pathogens.

Amphipathic α-helical peptides block LPS and E. coli (K-12)-induced IL-8 secretion in CD36-overexpressing HEK293 cells

Amphipathic α-helical peptides, that were found to be high affinity CD36 agonists in our uptake experiments, were also tested as potential inhibitors of CD36-dependent IL-8 secretion induced by LPS and bacteria. Two amphipathic α-helical motif containing peptides, L-37pA and D-37pA, demonstrated a 60–80% inhibition of cytokine secretion in CD36-overexpressing HEK293 cells challenged with LPS Re595 from mutant S. enterica or E. coli K-12 bacteria. In contrast, the L3D-37pA peptide, which contains three D-amino acid substitutions that disturb the amphipathic α-helical motif, did not affect IL-8 secretion induced by either of the two CD36 ligands (Fig. 8).

Effects of NF-κB and MAP kinase inhibitors on LPS - and E. coli (K-12) - induced cytokine secretion in CD36-overexpressing HEK293 cells and normal rat macrophages

In order to elucidate the potential mechanisms of the observed increase in IL-8 secretion induced by LPS and bacterial stimuli in CD36-overexpressing cells, we tested various pharmacological inhibitors of different signaling cascades. In particular, inhibitors of the NF-κB and MAP kinase pathways, known to be linked to cytokine production were evaluated. As seen from Table I, except for sulfasalasine with its highest concentration, none of the NF-κB inhibitors tested was capable of blocking LPS- or E. coli- induced IL-8 secretion in CD36-overexpressing HEK293 cells. These results present indirect evidence that the pro-inflammatory CD36-mediated response in this type of cells may occur independently from the TLR4 receptor, known to mediate ligand-induced signaling primarily via NF-κB activation (30). Moreover, neither SB203580, an inhibitor of p38 MAP kinase, nor PD98059, a selective inhibitor of MAP kinase kinase (MEK), was able to inhibit LPS or E. coli-induced IL-8 production in hCD36-overexpressing cells and only a weak 10–15% reduction of IL-8 levels was observed in the presence of PD98059 (Table II). IL-8 secretion in hCD36 overexpressing HEK293 cells treated with the JNK1, −2 and −3 inhibitor SP600125, or the JNKi1 inhibitor (cell-permeable peptide, consisting of a C-terminal sequence derived from the JNK-binding domain, and an N-terminal peptide containing the HIV-TAT_48–57 sequence), was reduced by 50–70% compared to mock-transfected cells. These data provide indirect evidence that the CD36-dependent pro-inflammatory response induced by LPS or E. coli in HEK293 cells predominantly involves activation of the JNK signaling pathway. Current data indicates that both NF-kB and JNK pathways are required for inducible IL-8 regulation in most cell types studied (31). However, the relative contribution of these pathways to the activation of cytokine expression may significantly vary depending on the specific ligand and cell type. To estimate the potential contribution of the different signaling pathways to the LPS- and E. coli-induced IL-6, another cytokine, whose expression in many respects is regulated similar to IL-8 (32), we tested the effects of the MAP kinase and NF-κB inhibitors in differentiated rat macrophages, as a cell model with high expression of both TLR4 and CD36 receptors. Our results in Table III demonstrate that both the MEK inhibitor PD98059 and p38 MAPK inhibitor SB203589 moderately (up to 40–50%) blocked LPS and E. coli-induced IL-6 secretion, whereas JNK1/2/3 inhibitor SP600125 caused the most pronounced reduction (up to 80–90%) of cytokine production in rat macrophages challenged with either LPS or E. coli K-12. PDTC, one of two NF-kB inhibitors tested also potently (up to 60–75%) decreased LPS and E. coli- induced IL-6 secretion in rat macrophages.

JNK1/2 inhibitors and amphipathic α-helical peptides block LPS and E. coli K-12-induced JNK1/2 activation assessed by immunoblotting

In order to confirm the role of CD36 in LPS and E. coli-induced JNK activation, we compared the levels of JNK1/2 phosphorylation in control and hCD36-overexpressing HEK293 cells by immunoblotting. Fig. 9 shows that, while JNK1/2 was transiently phosphorylated in both cell types, CD36-overexpressing cells demonstrated 3- to 8-times higher levels of JNK phosphorylation vs. control cells following treatment with increasing doses of LPS or E. coli (Fig. 8C) as well as within 1-h of treatment with either 100 ng/ml LPS (Fig. 9A) or 2 ng/ml of E. coli (Fig. 9B). A 1-h treatment of CD36-overexpressing cells with JNK inhibitors, D-JNK peptide inhibitor (Fig. 10A) or SP600125 (Fig. 10B), prior to a 45-min incubation of cells with LPS or E. coli, reduced JNK1/2 phosphorylation by 65–80 %.

To provide further evidence of CD36-dependence of LPS- and bacterium-induced JNK phosphorylation, we tested the ability of amphipathic α-helical peptides, previously shown to be CD36 receptor agonists, to alter JNK1/2 activation induced by LPS and E. coli. Both α-helical peptides, L-37pA and D-37pA, efficiently decreased the levels of JNK1/2 phosphorylation induced by LPS and E. coli in CD36-overexpressing cells, while the L3D-37pA peptide with a disturbed amphipathic α-helical motif was ineffective (Figs. 10, C and D).