Extraordinarily potent proinflammatory properties of lactoferrin-containing immunocomplexes against human monocytes and macrophages

Lactoferrin (LTF), an important first line defense molecule against infection, is a common target for humoral autoimmune reactions in humans. Since LTF is a multifunctional protein capable of activating innate immune cells via various surface receptors, we hypothesized that LTF-containing immune complexes (ICs) (LTF-ICs), likely formed in patients with high titer anti-LTF autoantibodies, could possess unique monocyte/macrophage-activating properties compared with other ICs. ELISA analysis on serum samples from rheumatoid arthritis (RA) patients (n = 80) and healthy controls (n = 35) for anti-LTF autoantibodies confirmed a positive correlation between circulating LTF-specific IgG and RA. ICs between human LTF and LTF-specific IgG purified from patient sera or immunized rabbits and mice, but not control ICs, LTF or Abs alone, elicited strong production of TNF-α and IL-1β by freshly fractionated human peripheral blood monocytes and monocytes-derived macrophages. Furthermore, LTF-ICs utilized both membrane-anchored CD14 and CD32a (FcγRIIa) to trigger monocyte activation in an internalization-, Toll-like receptor (TLR)4- and TLR9-dependent manner, and also that LTF-IC-induced cytokine production was blocked by specific inhibitors of caspase-1, NF-κB and MAPK. These results uncover a possible pathway for LTF-ICs perpetuating local inflammation and contributing to the pathogenesis of autoimmune diseases by triggering activation of infiltrating monocytes or tissue macrophages in vivo.


Results
Anti-LTF Abs in RA sera. Serum samples from 80 RA patients and 35 healthy subjects (NHS) were analyzed for levels of IgM and IgG against huLTF using ELISAs ( Fig. 1A and B). The data acquired has a normal distribution, and for anti-LTF IgG, but not IgM, the difference between RA and NHS groups was statistically significant (p < 0.01). When cutoff value was set at 0.43 (OD492 nm), 75% and 20% of the sera were positive for LTF-specific IgG in the RA and NHS groups, respectively. RA sera with highest levels of anti-LTF IgG were selected for affinity purification of human anti-LTF autoantibodies, and the resultant products (RA-IgG, mostly IgG1), characterized using huLTF-based ELISA (Fig. 1C) and SDS-PAGE electrophoresis (Fig. 1D), were subjected to subsequent functional analysis (see below). Values are the mean OD492 nm ± SEM from triplicate wells. Horizontal lines represent group averages. NS: not statistically significant. RA sera with highest levels of anti-LTF IgG were selected and pooled for affinity purification of human anti-LTF IgG autoantibodies using Protein A-and huLTF-Sepharose 4B columns. The resultant huLTF-specific IgG (RA-IgG) was compared with the sample of flow through and also IVIG (commercial) in huLTF-based ELISAs, **p < 0.01 compared with IVIG control (C). Samples of RA sera, RA-IgG and IVIG were also analyzed by Coomassie blue-stained SDS-PAGE 10% gel (D).
SCIentIFIC REpoRts | 7: 4230 | DOI: 10.1038/s41598-017-04275-7 LTF-ICs, but not huLTF or LTF-Abs alone or control ICs, effectively induce TNF-α and IL-1β production by human monocytes in vitro. Monocytes in the peripheral blood differentiate into macrophages following their homing to tissues; and upon activation produce cytokines such as TNF-α and IL-1β that mediate much of the inflammatory pathology in vivo. The reported effectiveness of soluble huLTF to activate human monocytes/macrophages is controversial [7][8][9][10][11] . In this study, human peripheral blood CD14 + CD11b + monocytes ( Fig. 2A, upper panels), freshly fractionated from PBMCs of healthy donors, were stimulated with huLTF in the presence or absence of RA-IgG for 24 h, followed by ELISA quantitation of TNF-α in the culture supernatant. Treatment of human monocytes with 1:1 mixture of huLTF and RA-IgG (RA-IgG-IC) resulted in significant TNF-α secretion compared to that treated with huLTF, RA-IgG or a mixture of huLTF with normal IgG (IVIG), which is similar to that cultured in medium alone (Fig. 2B). These results indicate that IC formation between huLTF and specific IgG is essential for the activation of monocytes to produce proinflammatory cytokines. Moreover, huLTF-specific IgG from rabbits (L3262, polyclonal IgG) and mice (M860, mAb of IgG1 subclass) 32 were as effective as, if not better than, RA-IgG in synergizing huLTF in the activation of human monocytes (Fig. 2D). Given that both monocytes and macrophages accumulate in inflamed joints of RA patients, we next addressed the question whether macrophages are similarly responsive as monocytes to LTF-IC stimulation. Human macrophages differentiated from monocytes in vitro by cytokine stimulation for 7 days ( Fig. 2A, lower panels) were subjected to LTF-IC stimulation. Like monocytes, these macrophages responded to RA-IgG-IC, LTF + L3262 and LTF + M860, but not to LTF alone or in combination with control Abs (Fig. 2C and E).
Mouse mAbs against OVA (Clone M562) or bovine serum albumin (BSA) (Clone J1), produced in house and highly specific for the respective immunizing Ags and of IgG1 subclass, were employed for preparation of control ICs. While huLTF plus M860 (M860-IC) or huLTF plus L3262 (L3262-IC) elicited significant levels of TNF-α and IL-1β secretion by human monocytes in vitro, equal proportion mixtures (15 μg/ml for each protein) of OVA plus M562 (OVA-IC), or BSA plus J1 (BSA-IC) did not (Fig. 3A). Taking this further, preformed ICs of LTF-M860 and BSA-J1, fractionated using Sephadex columns and assayed using SDS-PAGE electrophoresis (Fig. 3B), were serially titrated against human monocytes for cytokine induction. The dose-response study again confirmed and differentiated macrophages (M Ф , lower panels) were stained with FITC-conjugated mAbs against human CD14 or CD11b followed by FACS analysis. (B) Freshly fractionated human monocytes and (C) human monocytes-differentiated macrophages were incubated with 15 μg/ml huLTF, or RA-IgG, or RA-IgG-IC (mixture of huLTF plus RA-IgG), or huLTF plus IVIG (15 μg/ml for each protein), for 24 h. (D) Fractionated human monocytes and (E) human monocytes-differentiated macrophages were incubated with 15 μg/ml huLTF, or mouse anti-huLTF mAb M860, or rabbit anti-huLTF polyclonal IgG L3262, or a mixture of huLTF plus M860 (LTF + M860) or L3262 (LTF + L3262) for 24 h. Cells cultured in medium only (Med) or stimulated with 3 μg/ml LPS were included as controls. TNF-α concentration in the culture supernatant was then determined by ELISA. Data are mean concentration ± SEM from triplicate wells and these are representatives of at least 3 independent experiments. *p < 0.05, **p < 0.01 compared with medium controls. the potency and specificity of the LTF-IC in the induction of TNF-α and IL-1β secretion by human monocytes (Fig. 3C). Kinetic analysis showed that TNF-α and IL-1β secretion by human monocytes under M860-IC stimulation peaked at 16-18 h, whilst that in the BSA-IC group remained barely detectable throughout the 48 h period (Fig. 3D). Moreover, a 1 h pulse treatment with M860-IC, but not with BSA-IC, was sufficient to trigger TNF-α and IL-1β responses lasting over 24 h (Fig. 3E). Based on statistical analysis of results from several independent experiments, there is no significant difference between TNFα levels in monocyte cultures given 1 h pulse stimulation or in the presence of M860-IC for the entire duration.
Synergistic effect of LTF-IC with LPS. LTF is well known for its ability to sequester LPS activity though competing with soluble CD14 (sCD14) and/or LPS-binding protein (LBP) for LPS binding [3][4][5][6] or counteracting the NF-κB signaling pathway associated with LPS activation of phagocytic cells [33][34][35][36] . Surprisingly, we found that LTF-ICs exhibited a clear synergistic effect with LPS in activating human monocytes. Human monocytes cultured with mixtures of suboptimal concentration LPS (1 μg/ml) and L3262-IC, or RA-IgG-IC, produced significantly more TNF-α than the additive effects of the two stimulants individually ( Fig. 4A and B). As expected, neither huLTF nor LTF-Abs alone enhanced LPS activity in similar experiments (data not shown). These results indicate a synchronizing effect between signals triggered by LPS and LTF-ICs in monocytes.

Requirement for CD32a (FcγRIIa), mCD14 and TLR4 in LTF-IC-induced human monocyte activation.
ICs are known to interact with the FcγR-expressing innate immune cells (e.g. DCs, monocytes/macrophages, neutrophils) via Fc portion of the IgH chain 22,23 . Indeed, when mAb M860 was pretreated with papain or pepsin (both are capable of cleaving the Fc portion of IgG, Supplemental Figure S1A) its ability to synergize with huLTF in activating human monocytes was almost completely lost (Supplemental Figure S1B). To further identify the FcγR(s) required for LTF-IC function, blocking mAbs against FcγRI (CD64), FcγRIIa (CD32a), or FcγRIII (CD16) were used in functional inhibition assays. TNF-α production by monocytes stimulated with LTF-M860 ICs, or RA-IgG-IC, was significantly blocked by αCD32a, but not by αCD64 or αCD16, mAb ( Fig. 5A and B). Note that, in previous studies, within the concentration range employed herein, the anti-CD16 and anti-CD64 mAbs exhibited strong suppression effect on monocyte activation triggered by CD16 and CD64 signaling, respectively 37,38 . Additionally, CD32a blockade did not have any detectable effect on LPS-induced responses in similar experiment (Fig. 5C). These results suggest that CD32a, but not CD16 and CD64, is an indispensable surface receptor for LTF-IC-mediated monocyte/macrophage activation.
It has been shown that IC-mediated FcγR cross-linking alone is by far insufficient to trigger full activation of myeloid cells and result in their proinflammatory cytokine production 39,40 . The potent monocyte/ macrophage-stimulating activity of LTF-IC suggests that huLTF in the complex can also deliver activation signaling, presumably through direct interaction with cell surface LTF-Rs and/or intracellular sensors. Supported by an earlier report that huLTF can bind with a high affinity to sCD14 in vitro 36 , we hypothesized that glycophosphotidyl inositol (GPI)-anchored CD14 (mCD14), a surface marker for phagocytic cells, is a likely receptor for LTF-ICs. Consistent with this notion, recombinant sCD14 effectively blocked the binding of FITC-labeled huLTF-ICs to human monocytes in vitro (Fig. 5D). Furthermore, mAb against human CD14, but not that against other potential LTF-Rs such as low-density lipoprotein receptor-related protein-1 (LRP-1) and nucleolin-1 41,42 , almost completely abolished the monocyte-activating effect of LTF-ICs in functional assays in a dose-dependent manner ( Fig. 5E and F). It should be emphasized that the inability of anti-LRP-1 and anti-nucleolin-1 Abs in preventing LTF-IC-mediated monocyte activation excludes the possibility that anti-CD14 mAb blocked LTF-IC function through its Fc fragment.
It has been shown that the GPI-anchored mCD14 relies on Toll-like receptor 4 (TLR4) for signaling following LPS ligation 43 . Interestingly CLI-095, a specific TLR4 inhibitor, was capable of suppressing LTF-IC-induced TNF-α production by human monocytes in a dose-dependent fashion, while it had no effect on the TNF-α production induced by zymosan which activates monocytes/macrophages through dectin-1 rather than TLR4 44 (Fig. 5G). These results collectively suggest that LTF-ICs may employ, in addition to FcγRIIa, the CD14/TLR4 complex (major components of the LPS signaling machinery) as a key signaling receptor during monocytes/ macrophages activation.
To exclude the possibility that mCD14 cross-linking alone by aggregated huLTF could activate monocytes/ macrophages without FcγR participation, biotinylated huLTF (LTF-Bio), which could be cross-linked by avidin to form dimmers, tetramers and higher molecular weight oligomers, was used to treat human monocytes. As shown in Supplemental Figure S1C-E, when avidin was added to LTF-Bio-treated cells, no significant increase in intracellular TNF-α expression was observed, thereby supporting our "co-ligation" hypothesis that simultaneous engagement of FcγRIIa and mCD14 on the cell surface is key to the potent pro-inflammatory monocyte/ macrophage-stimulating properties of LTF-ICs. Intracellular signaling mechanisms in monocytes following LTF-IC ligation. The kinetic study results (Fig. 3E) indicate that, following colligation with LTF-Rs and FcγRs on the surface of the responding cells, LTF-ICs could be internalized and continue to function through intracellular sensors. Indeed, FITC-huLTF-containing ICs were efficiently internalized by human monocytes following 1 h incubation at 37 °C, whilst FITC-huLTF alone, or FITC-OVA-IC, was poorly captured and internalized in the same experiment ( Fig. 6A-C). Moreover, MDC, a potent chemical inhibitor of endocytosis, suppressed TNF-α production by human monocytes in vitro (Fig. 6D). It is also of interest to note that the internalized LTF-IC (or LTF thereof) co-localized with lysosomes, but not ER, of the cell (Fig. 6E-K). Means et al. showed that DNA-containing SLE-ICs transiently co-localized to intracellular lysosomes containing FcγRIIa and TLR9, thereby defining a pathway in which FcγRIIa delivers SLE-ICs to intracellular compartment containing TLR9 in DCs 30 . We reasoned that a similar pathway could be employed in LTF-IC-mediated monocyte/macrophage activation. Although TLR9 expression in resting monocytes is low, but it is significantly increased following LTF-IC activation as revealed by semi-quantitative RT-PCR analysis on TLR9-encoding mRNA (Fig. 7A) and also intracellular staining for TLR9 protein and FACS analysis (Fig. 7B). Importantly, ODN-A151, a suppressive ODN capable of competitively inhibiting TLR9, effectively down-regulated TNF-α production by human monocytes cultured with M860-IC (Fig. 7C). Furthermore, TNF-α production by LTF-IC-activated monocytes was also blocked by inhibitors against p38 MAPK (SB203580) and NF-κB (BAY11-7082) dose-dependently (Fig. 7D), providing additional evidence for the involvement of LPS signaling pathways in LTF-IC-induced monocyte activation. It has recently been reported that IL-1β secretion by monocytes/macrophages depends on inflammasome activation 45 , of which caspase is a key enzyme responsible for pro-IL-1β cleavage. Interestingly, Z-WEHD-FMK, a specific caspase-1 inhibitor, strongly suppressed IL-1β, but not TNF-α, production by human monocytes under stimulation with LTF-ICs ( Fig. 7E and F).

Discussion
In this report we have shown a positive correlation between the prevalence of circulating anti-LTF IgG autoantibodies and RA in humans (Fig. 1), albeit there is much variation in the levels of anti-LTF autoantibodies in the RA sera and a small percentage of healthy subjects also have relatively high titer serum IgG against LTF. Some  analyze if there is clear correlation between clinical indicators (e.g. disease severity, drug usage) and LTF-specific autoantibody levels, much larger sample sizes of RA sera would be required to address these questions in full. We are also developing assay systems for simple and rapid quantitation of LTF-ICs in human sera, which may become useful for further studies in this direction.
The main finding of the present study is that ICs between huLTF and LTF-specific IgG from RA patients are potent activators of human monocytes/macrophages via co-ligation of CD14 and FcγRIIa in an inernanlization-, TLR4-and TLR9-dependent manner. Given that LTF concentration can be elevated significantly in damaged tissues 20,21 , the concentration range of LTF-ICs (10-30 μg/ml) used in this study is pathophysiologically relevant. LTF-ICs containing IgG from mice and rabbits are as effective as that containing human Abs in activating human monocytes/macrophages in vitro. This can be explained by the fact that rabbit IgG and mouse IgG1 in ICs can bind human FcγRs with relatively high affinity 22,23 . In order to further elucidate the roles of LTF-ICs in autoimmune arthritis, in vivo experiments employing relevant animal models (e.g. collagen-induced arthritis in mice) would be needed. Given that FcγRIIa exists in humans but not in mice, such experiments would require hFcγRIIA-trangenic animals. Interestingly, hFcγRIIA-trangenic mice, prepared by Tan Sardjono et al., develop spontaneous multisystem autoimmune disease and hypersensitivity to antibody-induced inflammation 46 .
The proinflammatory properties of LTF-ICs are unique amongst endogenous ICs as activators of innate immune cells in the following aspects: (i) LTF-ICs are over 10 folds more potent than ordinary ICs in eliciting proinflammatory cytokine responses in human monocytes/macrophages; (ii) A brief pulse (30 min) with LTF-ICs causes long lasting cytokine responses in human monocytes; (iii) In contrast to free huLTF which is capable of sequestering LPS activity in vitro, LTF-ICs synergize with LPS in activating human monocytes/macrophages. Our results, together with recent advances endorsing pivotal roles for ICs in the development of RA [28][29][30][31] , may help us to better understand the pathogenesis of autoimmune arthritis. Since LTF-Abs are found in patients with various systemic autoimmune diseases [13][14][15][16][17][18][19] , a similarly important role of LTF-ICs in a much wider spectrum of inflammatory autoimmune disorders can be anticipated.
Membrane-bound CD14 is apparently a major surface receptor for LTF-ICs. We found that the monocyte-binding and -activating effects of LTF-ICs are susceptible to heparin inhibition (Supplemental Figure S2), which is in agreement with previous reports that LTF bound to monocytic cells as well as various cationic ligands such as DNA and LPS in a heparin-sensitive way 6,36,47 . The GPI-anchored CD14 molecule constitutively resides in lipid rafts (microdomains) on the membrane surface of monocytes/macrophages 43 . Following LPS ligation, adaptor and signaling molecules such as TLR4 and MD2 are recruited to the site of LPS ligation within the lipid rafts, resulting in the formation of LPS sensing and signaling complex (LSSC) responsible for an enhanced (focused) signaling event though the NF-kB and MAPK pathways 43 . Since CD14 and TLR4 are the major components of LSSC in lipid rafts, we propose that the LPS signaling mechanisms are triggered during LTF-IC activation of monocytes/macrophages and also that LTF-IC-mediated cross-talk between CD32a and LSSC (or related receptor complex) acts as a strong signal for monocytic cell activation and subsequent internalization and intracellular triggering processes. A schematic representation of the co-ligation model is presented in Fig. 8. Following internalization, huLTF (or fragments thereof) gains access to intracellular sensors such as TLR9 and inflammasome, resulting in the enhanced functional state of the cell. Thus, LTF-IC-activated monocytes and macrophages in the joints could play significant roles in the pathogenesis of RA by secreting large amounts of pro-inflammatory cytokines such as TNF-α and IL-1β. High concentration of LTF found in synovial fluid of RA patients 20 is a likely result of neutrophil accumulation and activation because LTF is a major component of secondary granules of neutrophils 1 . Based upon these observations, we further hypothesize that factors in RA synovium including LTF-IC, monocyte/macrophage activation, proinflammatory cytokine secretion, neutrophil accumulation/triggering, LTF release and additional LTF-IC formation will form a positive-feedback loop leading to arthritogenic damage in vivo.
An interesting analogy can be drawn between LTF-ICs and DNA-containing ICs (DNA-ICs). In the case of DNA-IC-mediated DC activation, the internalized CpG-containing dsDNA triggers activation cascade through cooperation of CD32 and TLR9 28,30 . Another case in point is RA-specific autoantibodies to citrullinated proteins complexed with fibrinogen, which could induce macrophage secretion of TNF-α through FcγRIIa and TLR4 engagement 28,29,31 . It will be of considerable interest to examine if such immunologically active ICs could collaborate (synergize) with each other to endorse further enhanced inflammatory reactions, as seen in the case of LTF-IC-pretreated monocytes responding more vigorously to stimulation with otherwise non-stimulatory OVA-IC (data not shown).
The inhibitory hFcγRIIb is expressed only poorly on 20% blood monocytes 23,48 , therefore its potential influence on in vitro studies employing blood monocytes and monocyte-derived macrophages is negligible. However, it is also evident that tissue macrophages may express hFcγRIIb at higher level 49 , therefore more complex intracellular signaling through FcγRIIa as well as FcγRIIb could be triggered by LTF-IC in synovial macrophages. To further elucidate the role of LTF-IC in autoimmune arthritis, in vivo experiments employing hFcγRIIA-trangenic and/or FcγRIIb-knockout mice would be needed.
Taken together, LTF-ICs can be considered novel proinflammatory mediators that can elicit strong proinflammatory cytokine production by monocytes/macrophages, thereby contributing to the pathogenesis of the autoimmune diseases such as RA. This idea should facilitate the development of therapies of autoimmune diseases with specific targets including the binding sites, the dual ligation requirement, the internalization steps, and the intracellular sensors in LTF-IC-mediated activation of innate immune cells.

Materials and Methods
Antibodies and reagents. Monoclonal Ab M860 was prepared in our laboratory as described elsewhere 32 .
Monoclonal Abs M562 and J1 against OVA and BSA (Sigma), respectively, were also generated in this laboratory using the same procedure. For preparation of anti-huLTF IgG from RA patients (RA-IgG), IgG from 6 pooled plasma samples shown by ELISA to contain high levels of anti-LTF antibodies were sequentially purified by affinity chromatography on LTF-S4B (prepared in our laboratory) and Protein A-S4B columns (Pierce). The eluted IgG fractions were concentrated by centrifugation with buffer exchange to PBS (Amicon Ultra, Millipore) and were depleted of endotoxin by filtration through a polymyxin B column (Detoxigel). IgG concentrations were determined by optical density at 280 nm; IgG was aliquoted, and stored at −80 °C. For preparation of preformed LTF-ICs, huLTF (2 mg/ml) and M860 (2 mg/ml) were mixed in a sterile tube with gentle rotation at 37 °C for an hour. The mixture was then loaded onto a Sephadex Superfine G-75 column for separation of the IC from the uncoupled Ab and antigen. IC was eluted by 0.9% NaCl solution and the elution was collected every 0.5 ml. Samples of the elutions were separated by SDS-PAGE and stained with Coomasie blue. The elutions of IC were pooled, desalted and concentrated. Endotoxin was removed by polymyxin B coupled beads repeatly and the level of endotoxin in IC is below 1 EU/mg which was detected by Chromogenic LAL Endotoxin Assay Kit (Genscript). Control ICs between BSA and J1 were prepared similarly.
Rabbit anti-huLTF polyclonal IgG (L3262), rabbit IgG, huLTF, LPS, OVA, heparin, FITC, MDC and zymosan were from Sigma-Aldrich. PE-labeled anti-CD11b and Alexa Fluor 647-labeled anti-human CD14 mAbs were from Biolegend. APC-labeled mouse anti-human TLR9 mAb (eB72-1665) was from BD Biosciences. Mouse mAbs against human CD16 (B73.1) or CD64 (10.1) were from eBioscience. Mouse anti-human mAb CD32a (IV.3) was from Stem Cell Technologies. Mouse anti-human CD14 mAb (134620) and Z-WEHD-FMK were from R&D. Mouse mAbs against human nucleolin-1 and LRP1 were from Santa Cruz. F(ab)' 2 and Fab were prepared using immobilized pepsin and papain (Thermo scientific). Recombinant soluble human CD14 was from Peprotech. Biotin and avidin were from Thermo Scientific. ODN-A151 and CLI-095 were from Invivogen, BAY11-7082, SB203580, lysotracker and ER tracker were from Beyotime. Sample collection. Peripheral blood was collected from patients with RA (n = 80; 20-93 y of age; mean = 57.4 y of age; 48 females) attending the Department of Rheumatology, Peking University people Hospital (Beijing, China) between 2008 and 2009. Patients were diagnosed according to the 1987 criteria of the American College of Rheumatology 50 . All blood samples were collected while their disease was in an active phase. The blood samples were processed within 18 h of collection and the cell-free sera were stored at −80 °C until use. Serum samples from healthy volunteers between 20 and 50 y of age (n = 35, mean age, 30.6 y) were included as controls.
Cell isolation and culture. For preparation of monocytes from human peripheral blood mononuclear cells (PBMCs), venous blood from healthy donors were 2-fold diluted using RPMI 1640 medium and overlaid on Ficoll lymphocyte separating solution (Sigma-Aldrich) followed by centrifugation at 400 × g for 30 min at room temperature. Cells in the interphase were collected and washed. Monocytes were isolated using Monocyte Isolation Kit (Miltenyi Biotec) according to manufacturer's instructions of positive and negative selection. The resultant cells were 90% CD14-positive and viable as evidenced by flow cytometric analysis and microscopic observation. Purified monocytes were cultured in "R10" medium: RPMI 1640 supplemented with 10% (v/v) Fetal calf serum (FCS, HyClone Laboratories), penicillin/streptomycin (100 U/ml), L-glutamine (2 mM), and 2-ME (5 × 10 −5 M). Cells were cultured at a concentration of 5 × 10 5 cells/ml of medium at 37 °C and 5% CO 2 .