Anti-C1q Autoantibodies from Systemic Lupus Erythematosus Patients Induce a Proinflammatory Phenotype in Macrophages

Systemic lupus erythematosus (SLE) is characterized by B cell hyperactivity, a variety of Abs directed against autoantigens (e.g., intracellular components or plasma proteins), the formation of immune complexes (ICs), and aberrant complement activation (1, 2).

Inherited and acquired complement deficiencies are associated with the development and pathogenesis of SLE. Particularly, primary deficiencies in the early components of the classical pathway (CP; C1q, C4, C2) are strongly linked to SLE. Homozygous C1q deficiency provides the strongest genetic susceptibility for the development of SLE, underscoring that complement plays a major role in the pathogenesis of SLE (3). However, most SLE patients do not suffer from primary C1q deficiency, but in general, ongoing complement activation is considered to be responsible for low or undetectable complement levels. A possible reason for low C1q levels is anti-C1q autoantibodies (anti-C1q), which are present in 20–50% of unselected SLE patients; their occurrence correlates with low complement levels and lupus nephritis (4–6). Although these autoantibodies are linked to renal involvement, there is no direct evidence about how these autoantibodies contribute to the pathogenesis of lupus nephritis. Animal models suggested that renal inflammation is only induced by anti-C1q in combination with preformed glomerular C1q-containing ICs, requiring both complement activation and FcγR triggering (7). Recently, Pang et al. (8) demonstrated that affinity-purified anti-C1q from lupus patients bind to C1q on early apoptotic cells and, thereby, decrease the uptake of apoptotic cells by THP-1 cells. However, the direct downstream effect of anti-C1q on professional phagocytes is not well understood.

C1q, the recognition molecule of the C1 complex, acts as the initiator of the CP of the complement system (9). Beyond complement activation, C1q serves as a regulatory protein during inflammatory processes, including autoimmunity. The regulatory functions also affect cell differentiation, chemotaxis, migration, and survival (10–13). Moreover, C1q participates in the clearance of apoptotic cell material. C1q can directly bind to the surface of apoptotic cells via its globular heads (14, 15). The collagen-like tails then interact with phagocytes via C1q receptors (16), facilitating the engulfment of apoptotic cells (17, 18). Additionally, studies from different research groups demonstrated that C1q enhances the uptake of apoptotic cell material, as well as skews the cytokine profile released by phagocytic cells toward a less inflammatory response during phagocytosis (17–20). During the uptake of early and late apoptotic cells, C1q exerted a potent inhibitory effect on macrophage-mediated inflammation (21). These data suggest that C1q is crucial in limiting inflammation during the uptake of apoptotic cells. Furthermore, C1q polarization of macrophages might induce an anti-inflammatory phenotype.

Macrophages play an important role in host defense, inflammatory processes, tissue remodeling, and homeostasis. The functional profile of macrophages is determined by their activation and exposure to environmental factors, such as cytokines and growth factors, during their differentiation from monocytes into macrophages (22). In vitro, monocytes can be polarized into different macrophage subtypes by specific cytokines. Based on the stimulus providing an activation signal, macrophages can be divided into a continuum between two functionally polarized states: proinflammatory macrophages (M1) and anti-inflammatory macrophages (M2). The polarized macrophages exhibit functional differences as evident by their phenotypic profiles such as cytokine release and surface markers (22–24). M1 macrophages mainly produce proinflammatory cytokines and phagocyte microorganisms and are often linked to tissue injury and inflammation, whereas M2 macrophages display, in general, a low Ag-presenting capacity, inhibit and prevent T cell activation, and are associated with tissue repair and fibrosis (25, 26). However, polarization of macrophages is not permanent; rather, it is dependent on the milieu and is reversible (27).

Functional defects in the cells of the monocyte–macrophage lineage from SLE patients are well known, although the underlying molecular mechanism is not fully understood. SLE patient–derived macrophages exhibit a defect in the phagocytosis of apoptotic cell material correlating with low complement levels (28, 29). Additionally, monocytes from SLE patients have an abnormal cytokine-secretion profile in response to apoptotic cells, independent of the monocyte’s phagocytic efficiency or the patient’s disease state (30). These reports underscore that phagocytes and complement play key roles in the pathogenesis of lupus.

The aim of the study was to determine the simultaneous interaction of human monocyte-derived macrophages (HMDMs) with bound autoantibodies, on the one hand, and the immune-regulatory C1q molecule, on the other hand, using SLE patient–derived high-affinity anti-C1q, which were found to correlate with disease activity.

A cohort of 17 SLE patients (Table I) and 15 healthy control donors was included in the study. All SLE patients fulfilled at least 4 of 11 criteria of the American College of Rheumatology (31, 32). Collection and use of serum samples were approved by the local Ethics Committee (EKZ No. 110/04; 130/05).

HMDMs were derived from CD14⁺ monocytes isolated from fresh buffy coats (Blood Transfusion Centre, Basel, Switzerland). PBMCs were isolated by Ficoll gradient centrifugation (Histopaque 1077; Sigma-Aldrich, St. Louis, MO). CD14⁺ monocytes were isolated from PBMCs using CD14 MicroBeads (MACS; Miltenyi Biotec, Bergisch Gladbach, Germany), according to the manufacturer’s instructions (the average purity of the CD14⁺ monocyte fraction was always >95–98%, as assessed by flow cytometry). HMDMs were generated from CD14⁺ monocytes by culture in DMEM supplemented with 1% penicillin/streptomycin (DMEM+; both obtained from Life Technologies, Invitrogen, Carlsbad, CA) and 10% normal human serum (NHS; pooled from 40 healthy donors). The culture was maintained in 5% CO₂ at 37°C for 7 d, and the media were exchanged every 2–3 d.

A total of 15 ml heparinized venous blood was obtained from SLE patients (Table II) after written consent, according to the local Ethics Committee (EKZ No. 2014/125). All SLE patients fulfilled ≥4 of 11 criteria of the American College of Rheumatology (31, 32). CD14⁺ monocytes were isolated and differentiated as described above with one exception: adherent monocytes were differentiated into macrophages in DMEM+ supplemented with 10% autologous serum for 7 d. For the analysis of LPS-induced cytokine release, the patient’s differentiated macrophages were stimulated as described below. For the stimulation, the patient’s own anti-C1q also were included in the experimental setting.

After 7 d, HMDMs were harvested using ice-cold PBS (Life Technologies, Invitrogen), resuspended in DMEM+ supplemented with 0.1% human serum albumin (HSA-DMEM; Sigma-Aldrich) at 0.5 × 10⁶ cells/ml, and used for stimulation experiments. Ninety-six–well plates (MaxiSorp; Nalge Nunc International, Roskilde, Denmark) were coated with purified C1q (immobilized C1q [imC1q]; Complement Technology, Tyler, TX) or human serum albumin (HSA) at 5 μg/ml in coating buffer (0.1 M sodium carbonate buffer; pH 9.6) overnight at 4°C. The plates were washed twice with PBS, and C1q-coated wells were incubated with 100 μg/ml purified total IgG (purified by Protein G affinity columns) from SLE patients or healthy donors diluted in high-salt buffer (PBS/1 M NaCl) for 1 h at 37°C. Before use, each IgG preparation was centrifuged at 14,000 × g for 30 min at 4°C. After washing the plates four times with PBS, HMDMs (100 μl; 0.5 × 10⁶ cells/ml; 50,000 cells/well) were added to the wells, cells were allowed to adhere for 60 min at 37°C (where indicated, 10 ng/ml LPS [Escherichia coli: O127:B8; Sigma-Aldrich] or DMEM-HSA was added), and cells were incubated for 18 h.

For FcγR-blocking experiments, FcγRIIs were blocked by incubating HMDMs with 8 μg/ml anti-FcγRII [anti-human CD32 F(ab′)₂, clone 7.3; Ancell, Stillwater, MN] for 30 min at 4°C in DMEM-HSA before proceeding with stimulation experiments, as described above.

The morphology of HMDMs was assessed using the above-described in vitro model and an Olympus IX50 inverted phase-contrast microscope (Olympus, Hamburg, Germany) with original magnification ×10 and ×20. The size of cell aggregations was quantified using ImageJ software 1.47. Data are expressed relative to the cell size formed by HMDMs incubated on imC1q alone, which was set to 1.

After the stimulation of HMDMs on different coatings for 18 h, supernatants (SNs) were collected, centrifuged to remove cellular debris, and stored at −80°C until analysis. The concentrations of IL-1β, IL-6, TNF-α, and IL-10 were measured using Opt ELISA kits (BD Biosciences, San Diego, CA), according to the manufacturer’s instructions. All samples were analyzed in duplicates.

After stimulating HMDMs on different coatings for 18 h using the in vitro model described above, cells were collected and washed twice with ice-cold FACS buffer (PBS/1% BSA [Sigma-Aldrich]/1 mM sodium azide). Cells were resuspended in FACS buffer at 1 × 10⁶ cells/100 μl. Nonspecific binding to FcγRs was blocked by incubating cells with 2 mg/ml human IgG/1 × 10⁶ cells for 45 min at 4°C. For FACS analysis of surface markers, the following mouse mAbs conjugated with FITC were used: anti-CD14 (Immunotools, Friesoythe, Germany), anti-CD80 and -CD86 (both from BD Biosciences), anti-CD206 (BioLegend, Fell, Germany), and MHC class II (Immunotools), mouse mAbs conjugated with PE: anti-CD273 and -CD274 (both from BioLegend), and anti-Mer tyrosine kinase (MerTK; R&D Systems, Minneapolis, MN), or mouse mAb conjugated with allophycocyanin: anti-CD163 (BioLegend). In each experiment, parallel staining with appropriate isotype-matched controls IgG1-FITC, IgG2a-FITC, IgG1-allophycocyanin (all from Immunotools), or IgG2b-PE (BioLegend) was performed. For staining, cells were resuspended at 5 × 10⁵ cells/100 μl in FACS buffer and stained with FITC-, PE-, or allophycocyanin-conjugated Abs or their matched isotype control for 45 min at 4°C. Cells were washed twice with FACS buffer and resuspended in FACS buffer. For each dataset, 10,000 events in the viable cell gate (propidium iodide [PI]; Sigma-Aldrich; exclusion of dead cells) were acquired using a BD Accuri (BD Biosciences) and analyzed using FlowJo software (TreeStar, Ashland, OR). The final geometric mean fluorescence intensities (gMFIs) were calculated by subtracting the gMFI of the corresponding isotype control from the gMFI of the sample.

The endocytic activity of stimulated HMDMs was measured by analyzing the uptake of FITC-conjugated dextran ([FITC-dextran] molecular mass: 40,000 kDa; Sigma-Aldrich). A total of 5 × 10⁵ cells/100 μl stimulated HMDMs was incubated with 0.5 mg/ml FITC-dextran in media for 0, 30, or 60 min at 37 or 4°C, respectively, to measure specific uptake versus nonspecific binding. HMDMs were washed three times with FACS buffer, and the uptake of FITC-dextran was analyzed by flow cytometry using a BD Accuri.

Jurkat T cells were cultured in RPMI 1640 supplemented with 1% penicillin/streptomycin, 1% l-glutamine, 26 mM HEPES, and 10% FCS (complete RPMI; all reagents from Life Technologies, Invitrogen). Apoptosis was induced by UV light treatment (Stratalinker 1800; Stratagene) at 254 nm for 1 min at a cell concentration of 2 × 10⁶ cells/ml. Irradiated Jurkat T cells were cultured for an additional 16 h in complete RPMI to obtain early apoptotic cells or in FCS-free complete RPMI to obtain late apoptotic cells. Prior to apoptosis induction, Jurkat T cells were fluorescently labeled with 5 μM CFSE (Invitrogen, Molecular Probes), according to the manufacturer’s instructions. Apoptotic cells were characterized by double-staining for annexin V (AnV; Immunotools) and PI. For phagocytosis assays, HMDMs were primed for 18 h as described above, harvested, and washed twice with DMEM+. CFSE-labeled apoptotic Jurkat T cells were coincubated with differently primed HMDMs (10⁵ cells) at a 1:1 ratio for 0 or 30 min at 37 or 4°C in a total volume of 200 μl phagocytosis buffer (DMEM+/26 mM HEPES/5 mM MgCl₂). Phagocytosis was stopped by adding cold FACS buffer, and unphagocytosed Jurkat T cells were washed away. HMDMs were further stained with an allophycocyanin-conjugated mAb against CD14 (Immunotools), and the uptake of apoptotic Jurkat T cells was analyzed by flow cytometry. The percentage of CD14⁺CFSE⁺ double-positive cells was used to assess the percentage of HMDMs that phagocytosed (incubated at 37°C) and/or bound (incubated at 4°C) apoptotic Jurkat T cells. The uptake of apoptotic Jurkat T cells by HMDMs was expressed as the percentage of phagocytosis, which was defined as CD14⁺CFSE⁺ cells/total CD14⁺ cells × 100.

Data are expressed as mean ± SEM, unless stated otherwise. Statistical analyses were performed using the Wilcoxon matched-pair test or the Mann–Whitney U test to compare two groups. To compare more than two groups, we used one-way ANOVA, followed by the Bonferroni posttest, as indicated. Data were analyzed using GraphPad Prism software (GraphPad, La Jolla, CA). A p value <0.05 was considered statistically significant.

CD14⁺ monocytes were cultured for 7 d to obtain HMDMs. Cells were then harvested and reseeded on wells with different coatings. The cell morphology on these coatings was assessed using phase-contrast microscopy.

HMDMs stimulated for 18 h on HSA or imC1q showed different morphologies (Fig. 1A). Cells on imC1q exhibited a circular shape compared with those plated on HSA, which had a more elongated and spindle-like shape. This observation is consistent with previous findings on changes in the morphology of monocytes plated on imC1q versus HSA (19).

Intriguingly, when HMDMs were plated on anti-C1q bound to imC1q (imC1q+SLE IgG), the cells still had a circular cell shape, but, additionally, they formed cell aggregates (Fig. 1A). This effect was anti-C1q specific, because HMDMs plated on bound anti-C1q derived from anti-C1q⁺ healthy donors also formed aggregates (data not shown). However, no cell aggregation was observed when using ICs (BSA–human IgG anti-BSA) (Fig. 1A). Furthermore, HMDMs plated on coated purified SLE IgG or normal human (NH) IgG also did not form aggregates (Fig. 1A). The formation of cell aggregates was not due to increased cell death because double staining for AnV and PI did not reveal any significant difference between the priming conditions (data not shown).

Comparing total IgG derived from four SLE patients with IgG from four healthy donors confirmed that the formation of cell aggregates by HMDMs on SLE patient’s anti-C1q can be observed in several patient’s anti-C1q but not in anti-C1q⁻ healthy donors (imC1q+NH IgG) (Fig. 1B). Next, we studied the effects of anti-C1q from a entire cohort of SLE patients with variable anti-C1q titers and healthy donors (Fig. 1C). The average size of cell aggregates formed by HMDMs on bound anti-C1q was 1.8-fold greater compared with HMDMs plated on imC1q alone or on imC1q+NH IgG (imC1q+SLE IgG: 1.79 ± 0.015, imC1q+NH IgG: 1.17 ± 0.082, p = 0.0037).

Beyond complement activation, C1q was described to act as an anti-inflammatory regulator in immune cell processes. For example, regarding cytokine secretion and cytokine mRNA levels, C1q induces a decreased inflammatory response in phagocytes in combination with a proinflammatory stimulus, such as LPS (18, 21).

Therefore, we next assessed whether anti-C1q bound to imC1q changed the LPS-induced cytokine profile secreted by HMDMs compared with imC1q or HSA alone. For this purpose, HMDMs were incubated on HSA, imC1q, or imC1q with different anti-C1q⁺ SLE IgG or anti-C1q⁻ NH IgG, in the presence or absence of the TLR4 ligand LPS, for 18 h (Fig. 2).

HMDMs generally did not produce detectable cytokine levels in the absence of LPS, independent of the coatings (data not shown).

However, in accordance with previous reports (18, 21), imC1q significantly downregulated LPS-induced secretion of proinflammatory cytokines, such as IL-1β, IL-6, and TNF-α, compared with HSA alone (IL-1β/IL-6: p = 0.031, TNF-α: p = 0.0435) (Fig. 2A–C, upper panels). The suppressive effect of imC1q alone was reversed by SLE patient–derived anti-C1q bound to imC1q, as evidenced by a significant upregulation of proinflammatory cytokine levels compared with imC1q+NH IgG (IL-1β/TNF-α: p = 0.0001, IL-6: p = 0.0003) (Fig. 2A–C, lower panels). Induction of a proinflammatory cytokine response could also be observed when NH IgG from anti-C1q⁺ healthy donors was used (Fig. 2A–C, lower panels, arrows).

In contrast, imC1q upregulated the LPS-induced level of the anti-inflammatory cytokine IL-10 compared with HSA coating (IL-10: p = 0.02) (Fig. 2D, upper panel). Interestingly, SLE patient–derived anti-C1q significantly attenuated LPS-induced IL-10 production compared with imC1q+NH IgG (IL-10: p = 0.0354) (Fig. 2D, lower panel). Again, healthy donor–derived anti-C1q induced a similar effect as anti-C1q from SLE patients (Fig. 2D, lower panel, arrows).

When SLE IgG of anti-C1q⁻ patients was incubated on imC1q, no differences from NH IgG were detectable for any of the tested cytokines (data not shown). In addition, we found that IgG anti-C1q levels of SLE patients correlated with the LPS-induced secretion of all cytokines tested (Supplemental Fig. 1).

Taken together, imC1q-bound autoantibodies shifted the LPS-induced cytokine levels toward an inflammatory response, as reflected by an upregulation of IL-1β, IL-6, and TNF-α secretion, which was accompanied by a slight downregulation of IL-10 secretion compared with imC1q with control NH IgG.

Several reports suggested that FcγR engagement by deposited ICs in kidneys is crucial for the development of lupus nephritis (7, 33). Given the close correlation between lupus nephritis and anti-C1q, we hypothesized that FcγRs are of crucial importance for the proinflammatory effects of anti-C1q, as described previously. Because IgG2 is the predominant anti-C1q class and can trigger CD32 (FcγRII) (34, 35), we preincubated HMDMs or not with a CD32-blocking Ab. Next, HMDMs were incubated as outlined above, and LPS-induced cytokine levels were analyzed (Fig. 3). Blocking of FcγRII led to a decreased secretion of the proinflammatory cytokines down to levels as observed in the absence of anti-C1q (IL-1β: 84.3 ± 13.5%, p = 0.021; IL-6: 63.1 ± 9.8%, p = 0.0003; TNF-α: 86.0 ± 5.3%, p = 0.0011) (Fig. 3A–C). In contrast, IL-10 secretion was not significantly altered by blocking CD32 (IL-10: −22.2 ± 7.4%, p = 0.075, NS) (Fig. 3D).

Because the effect of anti-C1q is expressed as the induction of a proinflammatory and activating response, as reflected by morphological changes and LPS-induced cytokine secretion levels, we next assessed whether anti-C1q also have an effect on surface markers expressed by HMDMs. We evaluated the expression of several surface markers, including CD14, costimulatory receptors (CD80, CD86, CD273, CD274), mannose receptor (CD206), scavenger receptor (CD163), and MHC class II (Fig. 4).

Fully differentiated HMDMs were primed as described previously and stained for surface markers. Untreated HMDMs and HMDMs stimulated with LPS were used as controls for M2 macrophages and M1 macrophages, respectively.

HMDMs incubated on imC1q significantly upregulated CD14 and CD163 compared with untreated control cells (CD14: p = 0.011; CD163: p = 0.048), whereas CD86 and CD274 were downregulated (CD86: p = 0.0036, CD274: p = 0.048). HMDMs primed on anti-C1q bound to imC1q showed a tendency to upregulate MHC class II compared with imC1q alone (MHC class II: p = 0.0625, NS). On the contrary, expression of CD163 was reversed by HMDMs stimulated on bound anti-C1q compared with imC1q alone (CD163: p = 0.030) (Fig. 4, left panels). No expression of CD80 and no differences in the expression levels of CD206 and CD273 could be detected independent of the priming conditions that were used (data not shown).

To mimic the inflammatory environment to which macrophages might be exposed in the tissue of SLE patients, an inflammatory signal was provided to the cells by adding LPS.

In response to LPS priming, HMDMs altered their phenotype by upregulating CD14, CD80, and CD274 (CD14: p = 0.031, CD80: p = 0.0049, CD274: p = 0.013). Furthermore, TLR4 triggering by LPS led to a downregulation of CD86 and CD163 (CD86: p = 0.031, CD163: p = 0.040). The expression of CD206 was unaffected by LPS (Fig. 4, right panels).

HMDMs that were incubated on imC1q in the presence of LPS tended to express higher levels of CD80 compared with cells primed with LPS alone (CD80: p = 0.063, NS). In contrast, the combination of imC1q and LPS led to a decreased expression of CD163, CD274, and MHC class II (CD163: p = 0.06, NS; CD274: p = 0.05, MHC class II: p = 0.031). Anti-C1q bound to imC1q in combination with TLR4 triggering by LPS increased the expression of CD80, CD274, and MHC class II compared with imC1q and LPS (CD80: p = 0.024, CD274: p = 0.016, MHC class II: p = 0.015). The expression of CD206 and CD273 was not affected by bound anti-C1q compared with imC1q in combination with LPS (Fig. 4, right panels).

Using anti-C1q from two additional SLE patients [SLE IgG (2) and SLE IgG (3)] led to similar results (data not shown). In general, we did not observe a difference in the expression of surface markers by HMDMs primed on C1q-coated wells incubated with NH IgG compared with cells incubated on imC1q alone (data not shown).

Independent of the stimulation condition used, HMDMs exhibited macrophage characteristics, such as the expression of CD14 and CD163, but no expression of dendritic cell (DC)-related markers (e.g., no CD83 neoexpression upon LPS stimulation was detected).

In summary, HMDMs incubated on imC1q exhibited a CD14^high, CD86^low, CD163^high, CD274^low phenotype, consistent with an M2-like phenotype. In the presence of LPS, imC1q priming led to a phenotype characterized by CD14^int, CD80^int, CD274^low, MHC class II^low expression, similar to traits of M1 and M2. Furthermore, imC1q-bound anti-C1q priming resulted in a CD14^high, CD86^low, CD163^low, MHC class II^high phenotype, consistent with a more M1-like phenotype. The combination of anti-C1q and TLR4 stimulation by LPS induced a CD14^high, CD80^high, CD274^high, MHC class II^high phenotype in HMDMs, also consistent with a more M1-like phenotype.

It is well known that C1q bound to apoptotic cells facilitates their uptake by phagocytes (17, 18). C1q might be concentrated in tissues as a result of local production by DCs and macrophages. Consequently, C1q might be present in the fluid phase or deposited on cell surfaces in tissues (so-called “tissue-deposited C1q”). Deposited C1q might have unique functions as a priming agent for HMDMs or as a target for anti-C1q, thus influencing the ability of HMDMs to phagocytose dying cells (36).

To investigate whether C1q and anti-C1q bound to imC1q affect the endocytic capacity of macrophages, primed HMDMs were incubated with FITC-dextran for 30 min (Fig. 5A) and 60 min (Fig. 5B). As expected, HMDMs primed with LPS alone exhibited a slightly downregulated ability to endocytose FITC-dextran compared with untreated cells (30 min: p = 0.061, NS; 60 min: p < 0.05). In contrast, imC1q priming alone slightly increased the endocytic activity of HMDMs compared with untreated control cells (30 min: p = 0.2305, NS; 60 min: p = 0.0793, NS). Interestingly, HMDMs plated on bound anti-C1q derived from two SLE patients reduced the endocytosis of FITC-dextran to the level observed in LPS-primed HMDMs [untreated versus imC1q+SLE IgG(1/2): 30 min/60 min: p < 0.05, imC1q versus imC1q+SLE IgG(1/2): 30 min/60 min, p < 0.001].

Next, we investigated whether priming of HMDMs on imC1q compared with cells primed on bound anti-C1q had an effect on their phagocytosis rate. For this purpose, Jurkat T cells were labeled with CFSE prior to UV treatment to induce apoptosis. Early or late apoptotic Jurkat T cells were incubated with primed HMDMs (Fig. 6). Early apoptotic cells were characterized as AnV⁺PI⁻ (Fig. 6A, left panel). Routinely, ∼50–70% of early apoptotic cells were obtained. In contrast, late apoptotic cells were defined as AnV⁺PI⁺ (Fig. 6A, right panel).

In general, untreated HMDMs phagocytosed only low numbers of apoptotic Jurkat T cells (early: 16.5 ± 4.3%; late: 13.4 ± 2.4%) (Fig. 6B, 6C). In addition, we could not detect a significant difference in the uptake of early and late apoptotic cells by any priming condition used. However, HMDMs primed with LPS had downregulated phagocytic ability compared with untreated HMDMs (data not shown). In contrast, additional exposure to imC1q led to an increased phagocytosis of early and late apoptotic cells compared with untreated control cells (early: 34.5 ± 7.3%, p = 0.029; late: 28 ± 1.3%, p = 0.0057). Furthermore, HMDMs incubated on anti-C1q bound to imC1q displayed a significantly lower phagocytosis rate of apoptotic cells compared with imC1q-primed cells (early: 21.1 ± 4.3%, p = 0.048; late: 18.7 ± 2.2%, p = 0.017).

The downregulated phagocytosis rate of HMDMs incubated on imC1q-bound anti-C1q further underscores that anti-C1q might have a direct effect on HMDMs, altering the clearance of apoptotic cells.

Because HMDMs primed on bound anti-C1q exhibited a lower phagocytosis rate compared with cells stimulated on imC1q, we further analyzed MerTK expression by differentially primed HMDMs (Fig. 7). C1q triggered an upregulation of MerTK in murine macrophages that was accompanied by an increased phagocytosis rate of apoptotic cells by macrophages (37). Additionally, it was demonstrated that an efficient uptake of apoptotic cells by HMDMs is MerTK dependent (38).

HMDMs incubated on imC1q upregulated MerTK compared with untreated control cells (p = 0.017), whereas anti-C1q bound to imC1q suppressed MerTK expression compared with imC1q (p = 0.0048) (Fig. 7A). HMDMs stimulated with LPS also exhibited downregulated MerTK expression compared with control cells (p = 0.012) (Fig. 7B). Again, in the presence of a proinflammatory stimulus, imC1q slightly upregulated MerTK levels of HMDMs, whereas the combination of anti-C1q and TLR4 triggering by LPS slightly suppressed MerTK expression.

Thus far, all experiments were performed using healthy donor–derived cells. We next studied cells derived from SLE patients at different stages of disease (Table II), because cells of the monocyte–macrophage lineage of SLE patients are known to exhibit functional defects (28, 29).

The morphology of SLE patient–derived HMDMs incubated on different coatings followed a similar pattern as that observed for healthy donor–derived HMDMs (Supplemental Fig. 2).

Similar to healthy donor–derived HMDMs, HMDMs obtained from SLE patients secreted lower LPS-induced levels of proinflammatory cytokines (IL-1β, IL-6, TNF-α) when they were incubated on imC1q compared with HSA (IL-1β: p = 0.016, IL-6/TNF-α: p = 0.031) (Fig. 8A–C, upper panels). The anti-inflammatory effect of imC1q alone was abolished when SLE-derived HMDMs were incubated on anti-C1q bound to imC1q, as reflected by a significant upregulation of all proinflammatory cytokines tested compared with imC1q+NH IgG (IL-1β/IL-6/TNF-α: p = 0.029) (Fig. 8A–C, lower panels). Thus, these results were in accordance with the LPS-induced cytokine profiles of healthy donor–derived HMDMs (Fig. 2A–C).

In addition, secretion of the anti-inflammatory cytokine IL-10 was increased by cells stimulated on imC1q compared with those incubated on HSA (p = 0.031) (Fig. 8D, upper panel). However, when SLE patient–derived HMDMs were stimulated on anti-C1q bound to imC1q, they exhibited enhanced LPS-induced IL-10 secretion compared with imC1q+NH IgG (p = 0.029) (Fig. 8D, lower panel).

Interestingly, SLE macrophages also reacted strongly to the combination of self–anti-C1q bound to imC1q (data not shown). This phenomenon was observed in the majority of patients and was dependent on anti-C1q levels measured in the serum of patients (Table II).

Taken together, these results demonstrate that the anti-C1q–triggered secretion of proinflammatory cytokines was similar between SLE- and healthy donor–derived HMDMs in the same experimental settings. However, LPS-induced IL-10 secretion was slightly increased in SLE-derived HMDMs when incubated on bound anti-C1q compared with healthy donor control cells.

Anti-C1q are believed to be pathogenic in SLE, in particular with regard to lupus nephritis. Anti-C1q were found in the glomeruli of patients with lupus nephritis and are believed to contribute to renal inflammation (4–7, 39, 40). Nevertheless, their biological and pathogenic properties in this inflammatory disease are not well defined. In this context, limited information was available about the effect of imC1q-bound anti-C1q on HMDMs. We demonstrated that SLE patient–derived anti-C1q bound to imC1q induce a proinflammatory cytokine response, as reflected by increased LPS-induced production of IL-1β, IL-6, and TNF-α, as well as suppressed IL-10 secretion, thereby reversing the effect of imC1q alone. Additionally, bound autoantibodies induced downregulation of CD163 and upregulation of the LPS-induced expression of CD80, CD274, and MHC class II. In addition, HMDMs primed on anti-C1q bound to imC1q displayed a significantly lower phagocytosis rate of apoptotic cells, accompanied by reduced MerTK expression, compared with imC1q-primed HMDMs. Thus, bound anti-C1q altered the C1q-dependent suppression of macrophage-mediated inflammation by inducing a proinflammatory phenotype.

Analogous with Th cell nomenclature, macrophage subsets are classified as M1 and M2 subsets, which are associated with different functions (25, 26, 41). However, translation of these macrophage phenotypes into disease models might be oversimplified. In vivo, macrophages are constantly encountering various signals. Therefore, it might be possible that macrophages exhibit a phenotype showing both M1 and M2 characteristics, and that multiple phenotypes coexist. Additionally, polarization of macrophages is thought to be partially reversible, and, in response to their microenvironment, macrophages can express constantly changing phenotypes, also termed plasticity (25, 41). In fact, no characteristic macrophage phenotype could be defined in lupus nephritis. Different reports showed that infiltrating macrophages/DCs in murine lupus nephritis are very heterogeneous (42, 43). Nevertheless, mononuclear cells play a role in the pathogenesis of organ-related diseases, such as lupus nephritis, and are associated with chronic tissue damage and injury (44).

Infiltration of mononuclear cells plays a role in the progression of lupus nephritis and is associated with poor prognosis in SLE patients (45). Data from murine models of lupus nephritis suggest that macrophages and DCs infiltrating the kidneys represent a heterogeneous group and are derived from circulating peripheral monocytes (42, 43). Sahu et al. (42) found that the dominant macrophage subtype was not M1 or M2 and concluded that this phenotype might reflect an overall failure to resolve inflammation present in kidneys during flares. Additionally, the onset of proliferative glomerulonephritis in mice was associated with an upregulation of chemokine and cytokine expression, mediating further infiltration of activated DCs and monocytes into the kidneys. Macrophage subsets primarily secreted inflammatory cytokines (IL-1, IL-6, TNF-α) and expressed CD11b, CD80, and CD86, confirming an activated macrophage phenotype. The investigators suggested that the renal phenotype of macrophages resembles a M2b phenotype, which is induced by FcγR ligation in addition to TLR-4 triggering by LPS (43). Other investigators concluded that mononuclear phagocytes infiltrating the kidneys have an aberrant activation profile that contributes to renal tissue damage by mediating local inflammation, as well as excessive tissue remodeling (46). We found that imC1q-bound anti-C1q induced the production of proinflammatory cytokines (IL-1β, IL-6, TNF-α), which was accompanied by significant upregulation of LPS-induced CD14, CD80, CD274, and MHC class II, resembling more closely a M1-like phenotype.

Moreover, certain cytokines are implicated in the pathogenesis of SLE and lupus nephritis, such as IL-6, IL-10, IL-17, type I IFNs, and TNF-α (47). In the context of a disease, cytokines are thought to play a role as mediators of inflammation and tissue damage. For example, in experimental IC glomerulonephritis, monocytes infiltrating the kidneys secrete IL-1, which can trigger TNF-α secretion and lead to tissue injury (48). Additionally, urinary levels of IL-6 and IL-8 were higher in patients suffering from lupus nephritis compared with patients without renal involvement or healthy controls, suggesting local production of these particular cytokines (49). Furthermore, analysis of isolated cell populations from nephritic mouse kidneys during flares demonstrated increased expression levels of IL-1, IL-6, IL-10, and TNF-α in gene-expression arrays (43). In addition to the anti-C1q–triggered increased production of IL-1β, IL-6, and TNF-α, we found that LPS-triggered IL-10 secretion was suppressed in healthy donor–derived HMDMs when incubated on bound anti-C1q. In contrast, SLE patient–derived HMDMs showed enhanced anti-C1q–triggered IL-10 production. Interestingly, serum levels of IL-10 are also elevated and even correlate with disease activity in SLE patients (50, 51). In vitro, monocytes and B cells from SLE patients spontaneously secrete high levels of IL-10 (52, 53). Based on these observations, IL-10 is considered to be involved in the pathogenesis of lupus (54). In the context of a chronically inflamed environment, it may be possible that IL-10’s anti-inflammatory properties are lost, and high IL-10 levels themselves become pathogenic. Thus, SLE patient–derived HMDMs seem to have a defect in their phagocytosis efficiency, as well as in their cytokine response and, thereby, in the regulation of inflammation.

Beyond host defense, the complement system has an important function in the recognition and removal of apoptotic cell material. The efficient and fast clearance of dead cell material is crucial to avoid inflammatory and autoimmune processes (55–57). Moreover, C1q is an essential molecule in the clearance of apoptotic cells by bridging apoptotic cells and phagocytes (16–18). In this context, it was hypothesized that the binding of anti-C1q to C1q might interfere in the phagocytosis process. We demonstrate that bound anti-C1q suppress the C1q-dependent increased phagocytosis of apoptotic cell material by HMDMs. The downregulated phagocytosis rate of HMDMs incubated on imC1q-bound anti-C1q supports the hypothesis that anti-C1q have a direct effect on HMDMs by altering the clearance of apoptotic cells by direct binding of C1q bound to apoptotic cells or by interfering indirectly with the uptake of apoptotic cells by binding to imC1q (e.g., deposited in tissues) and suppressing the clearance of apoptotic cells by inducing a less efficient phagocytic macrophage phenotype. It is well known that not all macrophage subsets display the same phagocytic efficiency. IL-10–producing macrophages, reflecting a M2-like phenotype, preferentially clear early apoptotic cells and are more efficient in phagocytosis compared with other subsets (58). We found that healthy donor–derived HMDMs primed on imC1q were superior in clearing apoptotic cells and produced higher levels of IL-10 compared with anti-C1q⁻ and unprimed control cells.

C1q has a nonhepatic origin and is produced mainly by DCs and macrophages (59). C1q might accumulate in tissues during inflammatory processes as a result of local production by infiltrating DCs and macrophages (60, 61). The local synthesis and availability of this freshly synthesized C1q might exert different effects on local cells and their effector functions. It might even be possible that the release of C1q during the phagocytosis of apoptotic cells is upregulated, which, in turn, could increase, for example, MerTK expression itself and thus, facilitate the clearance of apoptotic cells. Moreover, efficient uptake of apoptotic cells by HMDMs was demonstrated to be MerTK dependent (38). Additionally, two recent studies by Galvan et al. (37, 62) showed that imC1q can trigger an upregulation of MerTK expression and its ligand, grow-arrest specific 6, in murine macrophages. In this context, upregulation of MerTK expression was accompanied by an increased rate of phagocytosis of apoptotic cells by macrophages (37). Analogously, we found that imC1q increased MerTK expression in HMDMs, which correlated with an increased ability to phagocytose apoptotic cells. In contrast, bound anti-C1q reduced MerTK expression, which was accompanied by a lower phagocytosis rate. The reduced MerTK expression induced by anti-C1q might reflect an indirect mechanism by which these autoantibodies interfere in the uptake of apoptotic cell material by macrophages, resulting in an increased apoptotic cell load. This indirect effect might even potentiate the direct effects on C1q. In fact, impaired and inefficient clearance of apoptotic material was proposed as a mechanism underlying SLE pathogenesis, causing an accumulation of dead cell material (63, 64).

An important limitation of our study is its in vitro character that does not necessarily reflect the human in vivo situation. However, our experiments were performed using patient-derived high-affinity Abs to correlate with disease activity (65), as well as patient-derived macrophages. In addition, our data on anti-C1q are in line with other studies demonstrating that autoantibodies from SLE patients can modify the response of phagocytes (8, 66, 67).

In conclusion, we show that imC1q, as well as bound anti-C1q, skew the polarization of HMDMs into different phenotypes and that anti-C1q play a critical role as polarizing agents of HMDMs by inducing a proinflammatory phenotype and reversing the anti-inflammatory properties of imC1q alone. In addition, anti-C1q seem to directly and indirectly affect the phagocytic capacity of macrophages. Our results provide new insights into the pathogenic mechanisms of anti-C1q and their possible role in SLE.

We thank Dr. M. Fischer (Immunobiology Group, Department of Biomedicine, University Hospital Basel) for help and the Blutspendezentrum (University Hospital Basel) for providing buffy coats.

The authors have no financial conflicts of interest.