Tumor necrosis factor alpha and adalimumab differentially regulate CD36 expression in human monocytes

Peripheral blood mononuclear cells (PBMCs) were isolated from the cytapheresis residues obtained from healthy donors by density gradient on Lymphoprep (AbCys, Paris, France) according to the manufacturer's instructions. The monocytes were isolated from the PBMC by adhesion [25]. The PBMCs were cultured in macrophage-serum-free medium (M-SFM; Gibco Invitrogen, Cergy Pontoise, France) supplemented with L-glutamine at a concentration of 10⁷ cells/ml for 1.5 hours at 37°C with 5% CO₂ in a humid atmosphere. The non-adherent cells were eliminated via three PBS (Eurobio, Les Ulis, France) washes. The adherent cells (>85% of monocytes [26]) were then cultivated in M-SFM in 96-well trays (Becton Dickinson, Le Pont-De-Claix, France) with 0.125 × 10⁶ monocytes/0.125 ml for reactive oxygen species (ROS) assay, in 24-well trays with 0.5 × 10⁶ monocytes/0.5 ml for flow cytometry tests and gel shift assays, and in 12-well trays with 10⁶ monocytes/ml for reverse transcription (RT)-PCR tests.

The isolation of the F(ab')2 fragment from adalimumab (Abbott France, Rungis, France), a human anti-TNFα IgG1 monoclonal antibody, was carried out by pepsin digestion using the ImmunoPure F(ab')2 Preparation Kit (Pierce by Interchim, Montluçon, France) according to the manufacturer's instructions. The purity of the F(ab')2 fragment obtained after adalimumab digestion was verified by migration of the final specimen on a 12% denaturant acrylamide gel, according to the manufacturer's instructions.

In time-course experiments, monocytes were incubated for 6 to 12 and 24 h with M-SFM alone or with M-SFM containing TNFα (10 ng/ml) or adalimumab (1 μg/ml). In additional assays, monocytes were incubated for 24 h with M-SFM alone or with M-SFM containing: human recombinant TNFα at increasing concentrations (0.1, 1 or 10 ng/ml; BD Biosciences Pharmingen, Le Pont-De-Claix, France); or a combination of TNFα (10 ng/ml) or adalimumab (1 μg/ml; a biologically relevant concentration used in human therapeutics [27]; Abbott France) with either TNFα (10 ng/ml), GW9662 (2 μM; Cayman Chemicals by Spi-Bio, Montigny le Bretonneux, France), Trolox^® (1 μM; Sigma-Aldrich, Saint Quentin Fallavier, France), or diphenylene iodonium chloride (DPI; 1 μM; Calbiochem by VWR International, Fontenay sous Bois, France). Rituximab, a human anti-CD20 IgG1 monoclonal antibody (Roche, Meylan, France) was used at the same concentration as adalimumab (1 μg/ml) as control antibody. To investigate the relative contributions of Fab and Fc fragments in the induction of CD36 membrane expression by adalimumab, monocytes were incubated for 24 h with the F(ab')2 fragment of adalimumab (0.8 μg/ml). Quantification of membrane expression of CD36 on monocytes was carried out using flow cytometry according to the following protocol: the monocytes were washed once with PBS and then incubated for 15 minutes at 4°C in 5 mM PBS EDTA and collected by aspirating and refilling the wells. The monocytes were then incubated with the anti-CD36 monoclonal antibody labeled with phycoerythrin (BD Bioscience Pharmingen), used at a ratio of 10 μl per 0.5 × 10⁶ cells. The background staining was evaluated using a control isotype labeled with phycoerythrin (BD Bioscience Pharmingen) (data not shown). The region of interest of the monocyte population, comprising over 3,000 cells, was isolated on morphological criteria of cell size and granularity. The presence of strong CD14 expression, a characteristic of monocytes, was verified within the region of interest (data not shown). The quantification of membrane expression of CD36 was obtained from the geometric mean of the fluorescence measured [23].

Monocytes were incubated with M-SFM alone or with M-SFM containing TNFα (10 ng/ml) for 5, 30 or 60 minutes and then stimulated, or not, by a synthetic ligand of PPARγ, rosiglitazone (5 μM). Nuclear proteins were then isolated according to the following procedure: the monocytes were lysed at 4°C in a hypotonic buffer (10 mM Hepes Free Acid^® (Sigma-Aldrich), 10 mM KCl, 0.5 mM EDTA, 1 mM MgCl₂, 0.1 mM EGTA) supplemented by anti-proteases (Complete^®, Roche Diagnostics). Igepal^® (10%; Sigma-Aldrich) was added. The cytoplasmic extracts (supernatants) were isolated after centrifugation and the pellet was replaced in a hypertonic buffer, (20 mM Hepes Free Acid^®, 400 mM NaCl, 0.5 mM EDTA, 1 mM MgCl₂, 0.1 mM EGTA) supplemented with anti-proteases (Complete^®) in order to extract the nuclear proteins. The proteins were dosed according to the Bradford method.

The oligonucleotide (Santa Cruz Biotechnology by Tebu-Bio, Le Perray en Yvelines, France) used for the shift had the following sequence: 5'-CAAAACTAGGTCAAAGGTCA-3', with the underlined sequence corresponding to the PPAR DNA consensus binding sequence. It was labeled with [γ-³²P]ATP at 37°C using T4 polynucleotide kinase (Promega France, Charbonnières-les-Bains, France) and purified on appropriate columns (Quiagen, Courtaboeuf, France). The probe was labeled at 30,000 to 40,000 cpm/μl.

For the DNA protein reaction, 3 μg of proteins mixed at ambient temperature with the binding buffer (2 mM Hepes Free Acid^®, 50 mM NaCl, 0.5 mM EDTA, 1 mM MgCl₂, 4% glycerol, 2 μg/ml bovine serum albumin, 0.5 mM dithiothreitol), 3 μl of labeled oligonucleotides and 0.3 μg of poly (dI-dC) (Sigma-Aldrich) were added in a final volume of 25 μl and incubated for 20 minutes at room temperature. The specimens were placed on 5% non-denaturing acrylamide gel and set to migrate for 2.5 h at 180 V. The gel was dehydrated under vacuum and exposed by autoradiography.

Monocytes were incubated for 4 h with M-SFM alone or with M-SFM containing either TNFα (10 ng/ml), adalimumab (1 μg/ml), or rosiglitazone (5 μM) with or without pretreatment with TNFα (10 ng/ml). The monocytes were lysed in TRIzol^® Reagent (Invitrogen) and the mRNA was extracted using the chloroform/isopropanol/ethanol standard procedure. To ascertain that RNA preparations were genomic DNA-free, a negative control reaction was systematically included in which the sample was substituted with water.

PCR for CD36 and β-actin cDNA was performed with the LC FastStart DNA master SYBR Green I (Roche Diagnostics). Amplification and detection were performed in a LightCycler^® system (Roche Diagnostics) as follows, according to the manufacturer's instructions. Reaction mixture (20 μl) was incubated initially for 8 minutes at 95°C to activate the Fast Start Taq DNA; amplifications were performed for 40 cycles (15 s at 95°C and 30 s at 69°C) for CD36 and β-actin. The primers were designed with the software Primer Express (Applied Biosystems, Foster City, USA). The primers were: 5'-TGT-AAC-CCA-GGA-CGC-TGA-GG-3' (sense) and 5'-GAA-GGT-TCG-AAG-ATG-GCA-CC-3' (antisense) for CD36; 5'-CCT-CAC-CCT-GAA-GTA-CCC-CA-3' (sense) and 5'-TGC-CAG-ATT-TTC-TCC-ATG-TCG-3' (antisense) for β-actin.

Real-time PCR data are represented as Ct values, where Ct is defined as the crossing threshold of PCR using Light-Cycler^® Data Analysis software. For calculating relative quantification of CD36 mRNA expression, we used the following procedure: ΔCtCD36 = CtSample - CtControl and ΔCtβ-actin = CtSample - CtControl; then, ΔΔCt represents the difference between ΔCtβ-actin and ΔCtCD36 calculated by the formula ΔΔCt = ΔCtβ-actin - ΔCtCD36; finally, the N-fold differential expression of CD36 mRNA samples compared to the control is expressed as 2^ΔΔCt.

Monocytes were incubated for 1 h with Hanks balanced salt solution (HBSS; Eurobio, Les Ulis, France) alone or with HBSS containing adalimumab (1 μg/ml), TNFα (10 ng/ml) or F(ab')2 (0.8 ng/ml). ROS production was quantified by chemiluminescence in the presence of 5-amino-2,3-dihydro-1,4-phthalazinedione (66 mM; Luminol^®, Sigma-Aldrich) using a thermostatically (37°C) controlled luminometer (Wallac 1420 Victor2, Finland) [26]. The generation of chemiluminescence was monitored continuously for 1 h. Results are expressed as total chemiluminescence emission (area under the curve).

All flow cytometry and real time PCR experiments were performed at least three times. The values are expressed as the mean ± standard deviation (SD). A Wilcoxon test was used to assess the significance of differences between two conditions. All p values are two-sided, and p values less than 0.05 are considered significant.

To study the effect of TNFα on the regulation of membrane expression of CD36, monocytes were treated with TNFα (10 ng/ml) for increasing periods of time and membrane expression of CD36 was quantified using flow cytometry. Figure 1a shows that TNFα did not influence membrane expression of CD36 after 6 h of cell culture in comparison to basal conditions (mean ± SD: 91.8 ± 9.2 versus 94.2 ± 11.2, -4%, p = 0.5). After 12 h, TNFα decreased membrane expression of CD36 (46.5 ± 6.9 versus 82.9 ± 18.5, -44%, p = 0.04). The strongest effect was observed after 24 h of cell culture, with a 67% TNFα-induced decrease of CD36 membrane expression (35.3 ± 15.5 versus 106.9 ± 11.3, p = 0.0004). Monocytes were then treated with TNFα at increasing concentrations for 24 h. Figure 1b shows that TNFα reduced membrane expression of CD36 in a dose-dependent manner: -9% (91.6 ± 10.3 versus 100.3 ± 10.9, p = 0.3), -29% (71.2 ± 13.1 versus 100.3 ± 10.9, p = 0.002) and -59% (41.4 ± 4.5 versus 100.3 ± 10.9, p = 0.003) for 0.1, 1 and 10 ng/ml TNFα, respectively, in comparison to basal conditions.

Next, we investigated whether the reduction of membrane expression of CD36 induced by TNFα could be inhibited by adalimumab. Figure 1c shows that adalimumab (1 μg/ml) inhibited the effect of TNFα (10 ng/ml) on membrane expression of CD36. Furthermore, adalimumab independently increased membrane expression of CD36 by 59% (194.9 ± 42.3 versus 122.8 ± 23.2, p = 0.03) in the presence of TNFα and by 90% (233.7 ± 49.7 versus 122.8 ± 23.2, p = 0.04) in the absence of TNFα, in comparison to basal conditions.

To assess the specificity of adalimumab's effect on membrane expression of CD36, we used rituximab, a human anti-CD20 IgG1 monoclonal antibody, as a control antibody. Figure 1d shows that adalimumab increased CD36 membrane expression (155.3 ± 24.1 versus 97.3 ± 11.7, +60%, p = 7 × 10^-5) while rituximab did not influence it (101.3 ± 18.5 versus 97.3 ± 11.7, +4%, p = 0.48), in comparison to basal conditions. Figure 1e shows that adalimumab did not affect CD36 expression after 6 and 12 h, while it increased CD36 expression by 92% after 24 h of cell culture (161.5 ± 10 versus 84 ± 12.4, p = 0.0003) in comparison to basal conditions.

To study the effect of TNFα and adalimumab on CD36 mRNA expression, the monocytes were treated with TNFα or adalimumab and the relative quantification of CD36 mRNA was carried out by RT-PCR. Figure 2 shows that TNFα (10 ng/ml) reduced CD36 mRNA expression by 72% (mRNA relative level ± SD: 0.28 ± 0.05, p = 0.002), while adalimumab (1 μg/ml) increased CD36 mRNA expression by 96% (1.96 ± 0.2, p = 0.02) in comparison to basal conditions.

Since PPARγ is a transcription factor that plays a key role in inducing membrane expression of CD36 on human monocytes [19], we investigated its implication in the regulation of CD36 expression by TNFα. We tested the hypothesis that PPARγ activation is inhibited by TNFα using gel-shift assays. The monocytes were incubated with M-SFM alone or with M-SFM containing TNFα (10 ng/ml) and stimulated, or not, with a synthetic ligand of PPARγ, rosiglitazone (5 μM), and PPARγ activation was analyzed by gel-shift assay. Figure 3a shows a basal activation of PPARγ that was inhibited by TNFα. Rosiglitazone increased the activation of PPARγ and this effect of rosiglitazone was inhibited by TNFα.

To evaluate the consequences of the inhibition of PPARγ activation by TNFα, we assessed the effect of TNFα on the induction of CD36 mRNA by rosiglitazone. Monocytes were incubated with M-SFM alone or with M-SFM containing TNFα (10 ng/ml) and stimulated, or not, with rosiglitazone (5 μM), and the relative quantification of CD36 mRNA expression was carried out by RT-PCR. Figure 3b shows that TNFα reduced CD36 mRNA expression (-72%, 0.28 ± 0.05, p = 0.002), while rosiglitazone increased CD36 mRNA expression (+46%, 1.46 ± 0.2, p = 0.02). The combination of TNFα and rosiglitazone inhibited the effect of the rosiglitazone by reducing CD36 mRNA expression (-59%, 0.41 ± 0.2, p = 0.002) in comparison to basal conditions.

Since redox signaling is involved in the regulation of CD36 expression in human monocytes [21, 22], we analyzed its role in the repression of CD36 membrane expression induced by TNFα (Figure 3c,d). Monocytes were incubated with TNFα (10 ng/ml) and ROS production was quantified by chemiluminescence for 1 h. Figure 3c shows that TNFα induced a two-fold increase in ROS production in comparison to basal conditions (45,173 ± 3,966 versus 19,207 ± 4,115, p = 0.03). The role of ROS production in the regulation of CD36 expression induced by TNFα was analyzed using an antioxidant. Monocytes were treated with TNFα (10 ng/ml), or with an antioxidant derived from vitamin E, Trolox^® (1 μM), or with a combination of TNFα and Trolox^®. Membrane expression of CD36 was then quantified using flow cytometry. Figure 3d shows that membrane expression of CD36 was not significantly modified by Trolox^® (76.6 ± 12 versus 86.2 ± 6.8, -12%, p = 0.11) in comparison to basal conditions. TNFα decreased CD36 expression (35.9 ± 7.8 versus 86.2 ± 6.8, -52%, p = 1 × 10^-5) and this effect was not affected by Trolox^® (42. 3 ± 4 versus 86.2 ± 6.8, -60%, p = 6 × 10^-6).

Since PPARγ is a transcription factor that plays a key role in inducing membrane expression of CD36 on human monocytes [19], we investigated its implication in the induction of CD36 membrane expression by adalimumab (Figure 4a). The monocytes were treated with adalimumab (1 μg/ml) or with a PPARγ antagonist, GW9662 (2 μM), or with a combination of adalimumab and GW9662, and CD36 expression was quantified using flow cytometry. Figure 4a shows that adalimumab increased CD36 membrane expression (233.7 ± 43.7 versus 122.8 ± 23.2, +70%, p = 0.02), while GW9662 did not significantly decrease CD36 membrane expression (93.8 ± 31.8 versus 122.8 ± 23.2, -24%, p = 0.2), in comparison to basal conditions. The combination of adalimumab and GW9662 did not inhibit the effect of adalimumab on CD36 membrane expression, which remained increased (205.5 ± 24.3 versus 122.8 ± 23.2, +67%, p = 0.003). We assessed the effect of adalimumab on PPARγ activation and did not observe any effect in gel shift experiments (data not shown).

To evaluate the role of redox signaling in the induction of CD36 expression observed with anti-TNFα, monocytes were incubated with adalimumab (1 μg/ml) and ROS production was quantified by chemiluminescence for 1 h. Figure 4b shows that adalimumab induced a two-fold increase in ROS production in comparison to basal conditions (37,095 ± 1,693 versus 19,207 ± 4,115 p = 0.008). The role of redox signaling in CD36 expression was investigated by using an antioxidant. The monocytes were treated with adalimumab (1 μg/ml) or with an antioxidant derived from vitamin E, Trolox^® (1 μM), or with a combination of adalimumab and Trolox^®, and membrane expression of CD36 was quantified using flow cytometry. Figure 4c shows that adalimumab increased CD36 membrane expression (168.6 ± 18.2 versus 86.23 ± 6.8, +96%, p = 0.001), while Trolox^® did not significantly modify it (76.7 ± 12 versus 86.23 ± 6.8, -11%, p = 0.06), in comparison to basal conditions. By contrast, the combination of adalimumab and Trolox^® inhibited the effect of adalimumab on CD36 membrane expression, which returned to levels observed in basal conditions (81.3 ± 17 versus 86.23 ± 6.8, -6%, p = 0.5).

Since NADPH oxidase is a key enzyme of oxidative metabolism, inducing production of free radicals in monocytes [28], we investigated its implication in the induction of CD36 membrane expression by adalimumab (Figure 4d). Monocytes were treated with adalimumab (1 μg/ml) or with an NADPH oxidase inhibitor, DPI, or with a combination of adalimumab and DPI, and the membrane expression of CD36 was quantified using flow cytometry. Figure 4d shows that adalimumab increased CD36 membrane expression (188.6 ± 46 versus 88 ± 10.9, +114%, p = 0.002), while DPI decreased it (56.9 ± 4 versus 88 ± 10.9, -35%, p = 0.0003), in comparison to basal conditions. In contrast, the combination of adalimumab and DPI inhibited the effect of adalimumab on CD36 membrane expression, which returned to levels observed in basal conditions (108.3 ± 25.4 versus 88 ± 10.9, +23%, p = 0.07).

The biological effects of monoclonal antibodies, such as adalimumab, involve both Fab and Fc fragments. The interaction between the Fc fragments of monoclonal antibodies and the Fc gamma receptor (FcγR) can activate redox signaling via NADPH oxidase [29, 30]. To investigate the relative contributions of Fab and Fc fragments in the induction of CD36 membrane expression by adalimumab, we removed the Fc region from adalimumab through pepsin digestion and isolated the F(ab')2 region (Figure 5a). The monocytes were treated with the purified F(ab')2 fragment from adalimumab at an equimolar concentration to that of 1 μg/ml adalimumab (0.8 μg/ml of F(ab')2 fragment being equivalent to 1 μg/ml of adalimumab), and the membrane expression of CD36 was quantified using flow cytometry. Figure 5b shows that the F(ab')2 fragment of adalimumab increased membrane expression of CD36 (140 ± 20.7 versus 87.5 ± 7.4, +60%, p = 0.005) in comparison to basal conditions. Finally, we investigated whether the F(ab')2 fragment of adalimumab promoted ROS production. Monocytes were incubated with the F(ab')2 fragment of adalimumab (0.8 μg/ml) and ROS production was quantified by chemiluminescence for 1 h. Figure 5c shows that the F(ab'2) fragment induced a two-fold increase in ROS production in comparison to basal conditions (39,826 ± 6,927 versus 21,873 ± 3,834, p = 0.01).