Address correspondence and reprint requests to Ramune Morkuniene, Institute of Neurosciences, Lithuanian University of Health Sciences, Eiveniu 4, LT-50009 Kaunas, Lithuania. E-mail: email@example.com
Beta amyloid (Aβ) oligomers are thought to contribute to the pathogenesis of Alzheimer's disease. However, clinical trials using Aβ immunization were unsuccessful due to strong brain inflammation, the mechanisms of which are poorly understood. In this study we tested whether monoclonal antibodies to oligomeric Aβ would prevent the neurotoxicity of Aβ oligomers in primary neuronal-glial cultures. However, surprisingly, the antibodies dramatically increased the neurotoxicity of Aβ. Antibodies bound to monomeric Aβ fragments were non-toxic to cultured neurons, while antibodies to other oligomeric proteins: hamster polyomavirus major capsid protein, human metapneumovirus nucleocapsid protein, and measles virus nucleocapsid protein, strongly potentiated the neurotoxicity of their antigens. The neurotoxicity of antibody-oligomeric antigen complexes was abolished by removal of the Fc region from the antibodies or by removal of microglia from cultures, and was accompanied by inflammatory activation and proliferation of the microglia in culture. In conclusion, we find that immune complexes formed by Aβ oligomers or other oligomeric/multimeric antigens and their specific antibodies can cause death and loss of neurons in primary neuronal-glial cultures via Fc-dependent microglial activation. The results suggest that therapies resulting in antibodies to oligomeric Aβ or oligomeric brain virus proteins should be used with caution or with suppression of microglial activation.
Immunization against Aβ in Alzheimer's disease carries a significant risk of neuroinflammation, the mechanisms of which are poorly understood. In this study we show that oligomeric antigens such as Aβ and viral proteins complexed with their specific antibodies can exert neurotoxic effects by Fc-dependent microglia activation. This suggests that therapies using antibodies against oligomeric proteins should be used with caution.
Alzheimer disease (AD) is the most common dementia disorder in the elderly. There is substantial evidence that the disease arises from the accumulation of amyloid-beta (Aβ) peptide in the brain, which may form soluble oligomeric species and insoluble Aβ fibrils. The Aβ oligomers have been found to be the most toxic forms of Aβ peptide in neuronal cultures and animal AD models (Lambert et al. 1998; Deshpande et al. 2006; Cizas et al. 2010). The levels of soluble Aβ were found to be increased in brains from AD patients and correlated highly with disease severity (Mclean et al. 1999). Accordingly, Aβ has been suggested as a potential therapeutic target in AD. A number of in vitro and in vivo studies have indicated the beneficial effects of immunotherapy for AD including immunization with Aβ or administration of Aβ antibodies. For example, Aβ antibodies were shown to be effective in reducing the toxicity of Aβ oligomers in cell cultures (Kayed et al. 2003), to prevent Aβ plaque formation and synaptic impairment in transgenic AD animal models (Schenk et al. 1999; Morgan et al. 2001; Hillen et al. 2010) and to slow cognitive decline in human AD patients (Hock et al. 2003).
Several mechanisms have been proposed to account for the protective effects of Aβ antibodies. It has been suggested that anti-Aβ may act in the periphery to bind plasma Aβ leading to an efflux of Aβ from the brain (DeMattos et al. 2001) or if the antibodies cross the blood–brain barrier (BBB), they may trigger Fc-receptor mediated phagocytosis of Aβ (Bard et al. 2000), or they may induce disaggregation of Aβ fibrils (Solomon et al. 1996).
Despite possible beneficial effects, many of the clinical trials of Aβ vaccines were terminated because of the development of meningoencephalitis, an inflammatory reactions in brain, in some AD patients (Nicoll et al. 2003; Orgogozo et al. 2003). In addition, Aβ antibody-induced encephalitis and vasogenic edema have been found to develop in transgenic animal models of AD (Furlan et al. 2003; Minami et al. 2010). The precise mechanism(s) by which anti-Aβ antibodies induce neurotoxicity and/or neuroinflammation remain unclear.
The aim of this study was to elucidate whether complexes of specific antibodies with Aβ aggregates may prevent or cause neurotoxicity in mixed neuronal-glial cultures and whether microglia are involved in this process.
We found that antibodies when in complexes with Aβ or other oligomeric/multimeric antigens exert strong toxic effects on neuronal cells in vitro which are dependent on microglial activation. Our study not only provides new insight into the mechanism of possible Aβ neurotoxicity in the presence of Aβ-specific antibodies in AD brain but may also be important clinically in the development of better and safer vaccines for AD treatment.
Materials and methods
Monoclonal antibodies, peptides, and reagents
Murine monoclonal antibody against N-terminal epitope of Aβ1-42, reactive both with Aβ1-42 and Aβ1-40 (clone #11E12, subtype IgG2b), was generated by hybridoma technology after immunization of BALB/c mice with Aβ1-42 oligomers sized 1–2 nm (Dalgediene et al. 2013). Monoclonal antibodies against recombinant oligomeric proteins: hamster polyomavirus major capsid protein VP1 (clone #6D11, IgG2a), human metapneumovirus (MTP) nucleocapsid (N) protein, (clone #4A2, IgG2b), measles (Me) virus N protein (clone #10F7, IgG2b) were described previously (Zvirbliene et al. 2006, 2007; Petraityte-Burneikiene et al. 2011). Monoclonal antibody against Aβ42, subtype IgG1, clone #G2- 13 was purchased from Millipore (Temecula, CA, USA), rabbit polyclonal Aβ1-42 antibody was from Abbiotec LLC (San Diego, CA, USA). The F(ab')2 fragments of monoclonal antibodies were prepared using Pierce™F(ab')2 Preparation Kit (Thermo Fisher Scientific, Rockford, IL, USA) according to the manufacturer's recommendations.
Peptides and reagents are described in Figure S1.
Preparation of synthetic Aβ aggregates
Various Aβ1-42 forms (oligomers, monomers, fibrils) were generated as described previously (Cizas et al. 2010), (see Figure S1). The size and morphology of the preparations of Aβ1–42 oligomers and fibrils were characterized by AFM as described (Cizas et al. 2010).
Neuronal-glial cell culture
Primary mixed neuronal-glial cultures from rat cerebella (referred to as CGC cultures below) were prepared from 7–8 days old male and female Wistar rats, or from mouse cerebella from 4–6 days old BALB/c both genders mice as described previously (Bal-Price and Brown 2001). Rats and mice were bred and maintained at our University animal house under controlled conditions. The procedures used in this study were approved by The State Food and Veterinary Service of the Republic of Lithuania in accordance with European Convention for the protection of vertebrate animals used for experimental and other purposes, and ARRIVE guidelines.
After 7 days in vitro cultures were used for experiments. In average, the rat cultures contained 8.5 ± 2.1% microglia, as assessed by staining with isolectin GS-IB4, 7.3 ± 4.9% astrocytes and 84.2% cerebellar granule neurons, as assessed by cellular morphology.
Treatment of CGC cultures and cell viability assay
For toxicity assays the cultures were treated for 24 h with full-length monoclonal antibodies or the respective F(ab')2 fragments (2 μg/mL) with or without the corresponding antigens: Aβ1-42 or Aβ1-40 (1 μM, hereinafter concentration of oligomeric antigens is indicated as an amount of peptide monomer per volume, or 4.5 μg/mL), VP1, MTP-N, MeN proteins or synthetic peptides representing their fragments (all at 2 μg/mL concentration). To allow formation of antibody–antigen complexes antibodies were pre-incubated with the respective antigens in culture medium for 10 min prior to the treatment of CGC.
Lipopolysaccharide (LPS) contamination test was done with LPS-binding antibiotic polymyxin B (1000 U/mL). Solutions of antibody complexes (antibody #11E12 + Aβ1-42 or anti-VP1 + VP1) were pre-incubated with polymyxin B for 30 min at 37°C and then added to CGCs at concentrations indicated above.
For depletion of microglia, CGC cultures were incubated for 4 h with 15 mM lysosomotropic reagent l-leucine-methyl-ester (LME) which selectively removes microglia (Neniskyte et al. 2011). Then medium was replaced with medium from sister cultures and further incubations were performed as described above.
The viability of neurons in cultures after 24 h incubation was assessed by propidium iodide (PI) and Hoechst 33342 staining using a fluorescence microscope OLYMPUS IX71S1F-3 (OLYMPUS AMERICA INC., Melville, NY, USA) as described in Cizas et al. 2010. Cells were counted in at least five microscopic fields per well (two wells per treatment). Data are expressed as percentage of specific neuronal cells of the total number of neuronal cells per field.
Neuronal densities in CGC cultures were assessed by counting neuronal cells in at least five microscopic fields/well and expressed as an average number of neuronal cells per field.
ELISA for TNF-α
After indicated time of treatment with antibodies–antigens complexes, the aliquots of culture media were collected from CGCs. TNF-α content in cell culture medium was determined using the TNF-α (Rat) ELISA kit (IBL International GmbH, Hamburg, Germany), assaying the samples according to manufacturer's protocol.
Data are expressed as mean ± SE of 3–11 experiments on separate CGC cultures. Statistical comparison between experimental groups was performed using Student's t-test. A value of p < 0.05 was considered statistically significant.
Monoclonal antibody #11E12 increases the neurotoxicity of Aβ
We investigated the effect of antigen–antibody complexes formed by Aβ1-42 and its specific monoclonal antibody on the viability of neurons in mixed neuronal-glial cultures. For this, monoclonal antibody #11E12 (2 μg/mL) generated against oligomeric forms of synthetic Aβ1-42 (Dalgediene et al. 2013), was pre-incubated with different Aβ1-42 aggregates – either large, non-toxic oligomers (with average size of particles of 4–10 nm z-height) or Aβ1-42 monomers, or Aβ1-42 fibrils (all at 1 μM concentration) – and then added to cultures for 24 h. As shown in Fig. 1a, complexes of #11E12 antibody with Aβ1-42 oligomers caused strong increases in both neuronal apoptosis (40%) and necrosis (49%) so that only 11% of neurons remained viable. Similar decreases of neuronal viability were observed when CGC cultures were incubated with #11E12 antibody plus Aβ1-42 monomers or Aβ1-42 fibrils: the neuronal viability decreased to 42% (antibody + monomers) and to 35% (antibody + fibrils) because of extensive apoptosis and necrosis (Fig. 1a). When either antibody #11E12 or Aβ1-42 oligomers (Fig. 1a) were added separately at the same concentrations, there was no neuronal death. Aβ1-42 monomers and fibrils added by themselves were also non-toxic for CGC as we demonstrated in our previous study (Cizas et al. 2010).
Complexes of #11E12 + Aβ1-42 oligomers or #11E12 + Aβ1-42 fibrils were strongly neurotoxic even at significantly lower concentrations of the antibody in CGC cultures: the neuronal viability was 52 ± 15% and 47 ± 19%, respectively, with 0.5 μg/mL of antibody, and 30 ± 14% when 0.2 μg/mL #11E12 with 2 μM Aβ1-42 oligomers was used. Vehicle without Aβ1-42 added together with #11E12 had no effect on cell viability (data not shown).
Aβ1-42 tends to aggregate quickly when added to aqueous solutions (Snyder et al. 1994). We observed that 10 min pre-incubation of Aβ1-42 monomers in Dulbecco's modified Eagle's medium caused formation of various oligomeric forms of Aβ1-42 (see Figure S2). This was confirmed by AFM data showing that oligomers of various sizes were formed immediately after addition of Aβ1-42 monomers to phosphate buffer, and remained nearly at the same aggregation state during 24 h incubation (see Figure S3a–c). In contrast, Aβ1-40 has been reported to have a substantially lower rate of oligomerization/fibrilization (Snyder et al. 1994) and we also detected that during 10 min – 2 h incubation in Dulbecco's modified Eagle's medium Aβ1-40 stayed mostly in monomeric form (see Figure S2). As the epitope of monoclonal antibody #11E12 is at the N-terminal end of the Aβ molecule and this antibody forms complexes with Aβ1-40 (Dalgediene et al. 2013), we investigated whether toxicity of immune complexes depends on the oligomerization status of the Aβ1-40. As can be seen in Fig. 1b, complexes of #11E12 antibody + Aβ1-40 oligomers were highly toxic and caused death of nearly all neurons in the culture during 24 h incubation because of necrosis (68%) and apoptosis (38%), though added alone Aβ1-40 oligomers did not cause any significant neuronal death. In contrast, the #11E12 antibody+Aβ1-40 monomers had no effect on neuronal viability in CGC (Fig. 1b) indicating that oligomeric/multimeric structure of Aβ may be essential determinant for neurotoxicity of complexes of the antibody with Aβ.
To test whether the cytotoxic effect of the antibody–antigen complex is caused by the binding of the Fc domain of #11E12 antibody to the Fcγ receptors (FcγR) on microglia present in cell cultures causing their activation, we generated the F(ab')2 fragment of #11E12 antibody and tested the effect of F(ab')2 (2 μg/mL) complex with Aβ1-42 oligomers on the viability of neurons in CGC cultures. As shown in Fig. 1a, 24 h exposure of cells to F(ab')2 #11E12 + Aβ1-42 oligomers did not cause any significant cell death: the viability of neurons was very similar to that of the control cells. This suggests that the Fc region of the #11E12 antibody in complex with Aβ1-42 is crucial for inducing neuronal death in neuronal-glial cultures.
Interestingly, as shown in Fig. 1c, the exposure of cells to #11E12 antibody + Aβ1-42 oligomers not only induced cell death but also decreased the total number of neurons in CGC cultures. After 24 h exposure of CGC cultures to antibody plus Aβ1-42 oligomer complexes neuronal densities decreased by 23% as compared with control cultures. Similarly, the number of neurons decreased by 24% in cultures treated with #11E12 + Aβ1-42 monomers and by 18% in those treated with #11E12 + Aβ1-42 fibrils as compared with the control (Fig. 1c). In contrast, immune complex of F(ab')2 #11E12 + Aβ1-42 oligomers did not reduce neuronal density over 24 h. The intact #11E12 antibody itself had no effect on cell number in CGC cultures (Fig. 1c).
To rule out the possible effect of contamination with bacterial LPS of the antibody or Aβ1-42 preparations causing immune reaction and neurotoxicity (Rietschel et al. 1994), we performed experiments in which antibody #11E12 and Aβ1-42 preparations were pre-incubated with 1000 U/mL polymyxin B, a potent antibiotic that binds to and neutralizes LPS (Cooperstock 1974). Polymyxin B did not prevent the toxic effect of the complexes of #11E12 antibody with Aβ1-42 oligomers: the percentage of viable cells was similar (14% and 19%) in #11E12 + Aβ1-42 and #11E12 + Aβ1-42 + polymyxin B groups and was statistically significantly lower than in control cultures incubated without the immune complexes (see Figure S4). This suggests that the neurotoxic effect of #11E12 antibody + Aβ1-42 oligomers is not caused by LPS contamination.
Complexes of other oligomeric proteins with the respective antibodies reduce the viability of neurons in CGC cultures
To investigate whether complexes of other oligomeric proteins bound to their specific antibodies are able to induce similar neurotoxicity, we used monoclonal antibodies raised against recombinant virus-like particles (VLPs) and nucleocapsid-like particles (NLPs) that represent highly ordered oligomeric structures (Gedvilaite et al. 2000; Zvirbliene et al. 2007). When CGC cultures were incubated with such monoclonal antibody #6D11 raised against hamster polyomavirus major capsid protein VP1 pre-incubated with Aβ1-42 oligomers or fibrils, no significant loss of neuronal viability was detected (Fig. 2a), indicating that specific antibody–antigen interaction was necessary to cause neuronal death in cell cultures. However, complexes of anti-VP1 #6D11 with its target antigen VP1 protein forming VLPs decreased neuronal viability to 37% compared to 80% viability of control cultures (Fig. 2a). Polymyxin B did not change the toxic effect of the immune complexes anti-VP1 and VP1 (Fig. 2a), indicating that there was no contribution of LPS to neurotoxicity. Removal of the Fc region of the anti-VP1 monoclonal antibody #6D11 reduced its toxicity: complexes formed by the F(ab')2 fragments plus VP1 did not cause any significant cell death (Fig. 2a). Thus, the neurotoxicity of anti-VP1 + VP1 required the Fc part of the IgG molecule.
In the next series of experiments, we used two other monoclonal antibodies: #10F7 raised against the MeN protein and antibody #4A2 raised against the MTP-N (both added at concentrations of 2 μg/mL). Both proteins are self-assembled to NLPs (Zvirbliene et al. 2007; Petraityte-Burneikiene et al. 2011). The complexes of anti-MTP-N + MTP-N and anti-MeN + MeN reduced neuronal viability to 50% and 67%, respectively, whereas complexes of the F(ab')2 fragments of the respective antibodies with their specific antigens [F(ab')2 anti-MTP-N + MTP-N and F(ab')2 anti-MeN + MeN], did not cause any significant neuronal death during a 24 h incubation (Fig. 2a). The antibodies by themselves (without antigens) were non-toxic (data not shown).
Neuronal loss was also observed in cultures treated with the immune complexes formed by these antibodies and their oligomeric antigens. After incubation of cultures with anti-VP1 + VP1, anti-MTP-N + MTP-N, and anti-MeN + MeN, neuronal densities were reduced by 14, 14, and 22%, respectively, compared with control cultures after 24 h (Fig. 2b). And again, complexes of antigens with respective F(ab')2 fragments had no significant effect on neuronal densities during 24 h incubations (Fig. 2b).
Altogether, these results suggest that complexes of the monoclonal antibody #11E12 with Aβ1-42 of any state of aggregation, as well as complexes of other monoclonal antibodies with their specific oligomeric antigens (anti-VP1 #6D11 + VP1, anti-MTP# 4A2 + MTP-N and anti MeN #10F7 + MeN) exert strong neurotoxic effect on neuronal-glial cultures and that the Fc region of these antigen-bound antibodies is essential to cause neuronal death.
Complexes of antibodies plus antigens reduce the viability of mouse neurons
To test whether cross-species reactivity is not involved in neurotoxicity when rat CGC cultures are treated with immune complexes formed by mouse monoclonal antibodies, we performed analogous experiments using primary neuronal-glial cultures isolated from mouse cerebellum. Similar as in rat cultures, we found the substantial decline of viability of mouse neurons when CGC cultures were treated with both complexes #11E12 antibody+Aβ1-42 oligomers and anti-VP1 + VP1 (to 1.25% and 32.3%, respectively, compared to 66.8% neuronal viability in control) (Fig. 3). The antibodies or antigens added separately had no effect on neuronal viability (Fig. 3). This indicates that mouse primary neurons are similarly sensitive to mouse antibody + antigen complexes as rat primary neurons.
Complexes of #11E12 antibody plus synthetic peptides representing monomeric Aβ fragments have no effect on neuronal viability
To investigate more deeply whether the neurotoxicity of antibody–antigen complexes was dependent on the oligomeric/multimeric structure of the antigen, we used short peptides of Aβ1-42 protein representing the epitope of monoclonal antibody #11E12. Based on the epitope mapping studies, the binding site of #11E12 antibody was localized at the amino (N)-terminus of Aβ1-42, between amino acid (aa) residues 1 and 6, and the strong interaction between the #11E12 antibody and Aβ1-13 peptide was estimated with an indirect ELISA (Dalgediene et al. 2013). Therefore, as a model of the monomeric Aβ antigen we employed synthetic peptides spanning the N-terminal sequence of Aβ1-42: Aβ1-6 peptide and Aβ1-13 peptide (aa 1–6 and aa 1–13 of Aβ, respectively). Their monomeric state was confirmed by AFM experiments – Aβ1-6 and Aβ1-13 in phosphate buffer did not form any detectable aggregates (see Figure S3d and e). To evaluate the cytotoxicity of immune complexes formed by other monomeric antigens, we have used synthetic peptide MeN440-452 representing the fragment of MeN protein recognized by the monoclonal antibody #10F7 (Zvirbliene et al. 2007). As shown in Fig. 4, complexes #11E12 antibody + Aβ1-6, #11E12 antibody + Aβ1-13 and anti-MeN + MeN440-452 had no effect on neuronal viability, and neuronal densities were not affected (data not shown).
These results confirm that oligomeric/multimeric structure of antigen is an essential factor in neurotoxicity of antigen–antibody complexes.
Antibody plus antigen-induced neurotoxicity is microglia-dependent
To test whether #11E12 antibody plus Aβ1-42–induced neuronal death was dependent on the presence of microglia, we performed experiments on glial-neuronal cultures from which microglial cells were selectively eliminated by incubating them with LME. In these experiments, cell cultures were pre-incubated with 15 mM LME for 4 h. This resulted in destruction of microglial cells whereas neurons and astrocytes remained intact. Then the cultures were exposed to #11E12 + Aβ1-42 oligomers for further 24 h. As shown in Fig. 5, there was no significant loss of neuronal viability over 24 h period in microglia-depleted cultures treated with #11E12 + Aβ1-42 compared with the control cultures. This suggests that the #11E12 antibody + Aβ1-42-induced neuronal death in neuronal-glial cultures was microglia-dependent.
Complexes of #11E12 antibody plus Aβ1-42 stimulate microglial proliferation in cultures
As shown in Fig. 6a, the complex of #11E12 + Aβ1-42 oligomers increased the number of microglia in neuronal-glial cultures from 9.8 cells/field in control to 13.6 cells/field after 24 h. A similar effect was observed when cell cultures were incubated with complexes of #11E12 + Aβ1-42 monomers or #11E12 + Aβ1-42 fibrils: the number of microglia was increased to 13.3 cells/field and 14.2 cells/field, respectively (Fig. 6a). The complex of F(ab')2 #11E12 + Aβ1-42 oligomers had no effect on microglial proliferation (Fig. 6a), indicating a requirement for the Fc region of the antibody to induce microglial proliferation. Aβ1-42 oligomers, monomers or fibrils in the absence of #11E12 antibody did not change the number of microglia during 24 h (Fig. 6a).
Microglial proliferation (up to 13.6 cells/field) was also observed when neuronal-glial cultures were incubated with anti-MeN + MeN complexes but not with F(ab')2 anti-MeN + MeN for 24 h (Fig. 6b). The other two antibodies, anti-VP1 + VP1 and anti-MTP-N + MTP, intact and F(ab')2, had no statistically significant effect on microglial proliferation over 24 h (Fig. 6b). All antibodies tested alone or in complexes with short monomeric synthetic peptides (Aβ1-6, Aβ1-13 and MeN440-452) did not change the number of microglia in cultures (data not shown).
Complexes of antibodies with oligomeric/multimeric antigens cause release of TNF-α
As the complexes of #11E12 antibody + Aβ1-42 can cause microglia-dependent neuronal death, loss of neurons and proliferation of microglia, and taking into account that the Fc region of #11E12 antibody is important to induce these effects, we tested whether the antibody–antigen complexes activated the microglia to release pro-inflammatory cytokines. First, we measured TNF-α levels in the cell culture media collected after 1, 2, 4, 8, and 24 h incubation with #11E12 + Aβ1-42 oligomers. As shown in Fig. 7a, TNF-α was undetectable in control as well as after 1 to 8 h incubation of cells with immune complexes, however, a significant increase in TNF-α in culture medium (up to 98 ± 26 pg/mL) was observed after 24 h incubation with #11E12 antibody + Aβ1-42. In contrast, Aβ1-42 oligomers alone (Fig. 7a) or #11E12 alone (data not shown) did not induce production of TNF-α during 24 h period.
A much higher TNF-α concentration, 676 ± 284 pg/mL, was detected in the medium of cells treated with anti-VP1 + VP1 complex for 24 h (Fig. 7b). The other two investigated immune complexes, anti-MTP-N + MTP-N and anti-MeN + MeN, moderately increased TNF-α level in culture medium, to 74 ± 18 and 67 ± 28 pg/mL, respectively, after 24 h (Fig. 7b). VP1, MTP-N or MeN or their antibodies alone did not induce production of TNF- α during 24 h (data not shown).
Importantly, anti-inflammatory agent minocycline at 500 μM concentration completely blocked the #11E12 antibody + Aβ1-42-induced TNF-α release and partially protected against immune complexes-induced neuronal death (see Figure S5).
Effects of monoclonal anti-Aβ1-42 of IgG1 subtype and rabbit polyclonal anti-Aβ1-42 complexes on CGC cultures
The next series of experiments were performed to investigate further the mechanism of the toxicity of antibody–antigen complexes on cultured neurons. We used a commercial mouse monoclonal antibody (anti-Aβ42, clone #G2-13) that recognizes linear Aβ1-42 peptide at the carboxy (C) terminus (Ida et al. 1996). According to the manufacturer's information, this antibody is of IgG1 subtype and therefore might have different affinity to Fc receptors of rat macrophages as compared to the antibody #11E12 of IgG2b subtype (Nimmerjahn and Ravetch 2006). The second control antibody used in our experiments was rabbit polyclonal antibody specific to Aβ1-42 sequence but lacking specificity to rat Fc receptors. Similar to experiments described above, we treated CGC cultures with complexes anti-Aβ42 # G2-13 + Aβ1-42 oligomers or with rabbit polyclonal anti-Aβ1-42 + Aβ1-42 oligomers for 24 h and measured neuronal viability, number of neurons and microglia, and TNF-α level. As shown in Fig. 8, the anti-Aβ42 #G2-13 + Aβ1-42 oligomers decreased neuronal viability from 89% (control culture) to 61%. As expected, the immune complexes of rabbit polyclonal anti-Aβ1-42 + Aβ1-42 oligomers did not influence neuronal viability. None of the antibodies added separately had an effect on cell viability after 24 h (Fig. 8). The number of neurons and microglia cells, and medium TNF-α level did not change after incubation of cultures with complexes anti-Aβ42 #G2-13 + Aβ1-42 oligomers and rabbit polyclonal anti-Aβ1-42 + Aβ1-42 oligomers for 24 h (data not shown). These data suggest that neurotoxicity of immune complexes of antibodies with Aβ1-42 oligomers does not strongly depend on the isotype of immunoglobulin involved, but does depend on the specific interaction of the complexes with Fc receptors on the competent cells.
In this study we show that antigen–antibody complexes formed by Aβ1-42 and specific monoclonal antibodies of IgG isotype can cause death and loss of neurons in primary neuronal-glial cultures by a mechanism that involves microglial activation by the Fc region of antibody–antigen complexes. This conclusion is supported by our findings that the neurotoxic effects of immune complexes were abolished by the removal of microglia from cultures or by removal of the Fc region of #11E12 monoclonal antibody, suggesting that activation of microglial FcγR was required for microglial inflammation and neuronal death. These effects were seen not only with the antibody raised against Aβ1-42 but also with other antibodies raised against recombinant oligomeric antigens – VLPs and NLPs – when they were bound to their target antigens. The neurotoxicity of the antibodies was lost when the same antibodies were applied in complexes with short peptides that do not form oligomeric structures. Therefore, our study demonstrates for the first time that the multimeric structure of antigen–antibody complexes is essential to causing neurotoxicity in mixed neuronal-glial cultures.
Understanding this mechanism may have important implications for developing new strategies for prevention and treatment of AD. The antibody-based immunotherapy to treat AD either via the immunization with Aβ or by the use of Aβ antibodies have shown controversial results in many pre-clinical and clinical studies, however, why the treatment complications such as meningoencephalitis (Nicoll et al. 2003; Orgogozo et al. 2003; Ferrer et al. 2004) does occur in some immunization cases is unclear. The immune response mechanism based on T-cell activation has been suggested which may initiate toxic events (Orgogozo et al. 2003). Passive immunization does not involve T-cell activation, however, significant microglia activation and Fc receptor dependent-phagocytosis was observed in AD patients (Nicoll et al. 2003; Ferrer et al. 2004) and in AD transgenic mice (Bard et al. 2000; Wilcock et al. 2004). Furthermore, auto-antibodies to Aβ were significantly elevated in AD patients and in serum of APP transgenic mice, which have been shown to enhance the neurotoxicity of Aβ (Nath et al. 2003).
Our data are in agreement with these findings and suggest that the neurotoxic effect may result from the interaction of oligomeric complexes of Aβ with antibodies causing Fc receptor mediated activation of microglia. The involvement of microglia in antibody-mediated removal of Aβ aggregates in primary cultures was recently documented for anti-Aβ antibodies (Mohajeri et al. 2006) and human intravenous immunoglobulin IVIG, which contained naturally occurring auto-antibodies, including antibodies against Aβ peptides (Magga et al. 2010). It was shown that peripheral administration of IVIG (Magga et al. 2010) or anti-Aβ antibodies (Koenigsknecht-Talboo and Landreth 2005) caused their penetration into mouse brain tissue and binding to Aβ deposits together with co-localization of microglia. This suggests that peripheral antibodies indeed may cross BBB and interact with the Aβ aggregates and may activate microglia cells. Recently, it has been shown that immune complexes formed in mice brain parenchyma induced a neuroinflammatory response and behavioral changes through recruitment and activation of microglia via FcγR (Teeling et al. 2012). Activated microglia, via production of reactive oxygen species and other inflammatory mediators, may be sufficient to induce neuronal death (Brown and Neher 2010) and neuronal loss (Neniskyte et al. 2011).
We have previously determined that the epitope of #11E12 antibody was localized at the N-terminus of Aβ1-42, between amino acid residues 1 and 6, and that #11E12 antibody reacted with the peptide representing the N terminal sequence (aa 1–13) (Dalgediene et al. 2013). However, the complexes of #11E12 antibody plus Aβ1-6 or #11E12 antibody plus Aβ1-13 were not toxic to cultured neurons. This indicates that neurotoxicity of the antibody–antigen complex is depended on the structure of antigen. Multimeric antigens may bind multiple antibodies, which may bring together multiple FcγR and thereby activate them to induce inflammation. Whereas monomeric antigens may bind single antibodies recruiting single FcγR resulting in minimal inflammation. Therefore, development of short Aβ immunogens that target monomeric Aβ may be a promising approach for AD treatment.
We also showed that the cytotoxic effects of antibody–antigen complexes disappeared when microglia cells were removed from the culture or when the Fc part of the antibody was cleaved off, indicating the involvement of microglia FcγR. These data are consistent with results obtained with deglycosylated anti-Aβ, which were shown to reduce microglia phagocytosis and cytokine production because of reduced affinity to FcγR in primary cultured microglia (Takata et al. 2007). In addition, deglycosylated anti-Aβ (Carty et al. 2006) or F(ab')2 fragments (Tamura et al. 2005) were able to suppress the microhemorrhages in AD mice. In accordance, our data showed that minocycline, which is known to inhibit microglial activation in animal models of neurodegenerative diseases (Fan et al. 2007), was neuroprotective against anti-Aβ + Αβ induced release of TNF-α and neuronal death.
Rodent microglia cells have four types of FcγR (Ravetch and Kinet 1991; Nimmerjahn et al. 2005). The IgG subclass specificities of mouse and rat phagocytic cells FcγR are likely to be more similar to each other (Song et al. 1990). FcγRII lack phagocytic activity (Takai et al. 1994), however, FcγRI, FcγRIII, and FcγRIV exhibit affinity to mouse IgG (Fossati-Jimack et al. 2000; Nimmerjahn et al. 2005). All three subtypes of IgG were able to reduce Aβ deposits in mouse microglia cultures (Bard et al. 2003). Moreover, the deleterious effects of anti-Aβ in AD mice also have been obtained with different IgG subtypes: microhemorrhage with IgG2a (Schroeter et al. 2008), encephalomyelitis with IgG1, IgG2a (Furlan et al. 2003), and IgG2b (Carty et al. 2006), microglia activation with IgG1 (Wilcock et al. 2004). We used primary rat and mouse cultures and found cytotoxic effects of different mouse IgG subtype complexes, IgG2b (anti-Aβ, clone #11E12, anti-MTP-N, clone #4A2 and anti-MeN, #10F7), IgG2a (anti-VP1,clone #6D11), and IgG1 (anti-Aβ42, clone #G2-13) indicating that neurotoxic effect was not animal species dependent as mouse antibody complexes were similarly toxic to rat as well as to mouse neurons. However, the capacity of FcγR to bind various IgG subtypes may be different, the high-affinity FcγRI is capable to bind preferentially IgG2a (Sears et al. 1990), the low-affinity FcγRIII binds IgG1, IgG2a, and IgG2b (Weinshank et al. 1988), and a recently identified high-affinity FcγRIV binds IgG2a and IgG2b (Nimmerjahn et al. 2005; Bruhns 2012). These divergent activities might have contribution to antibody-mediated cytotoxicity (Nimmerjahn and Ravetch 2006). Indeed, our study suggests that rat microlia FcγR can bind oligomeric antigen–antibody of different mouse IgG isotypes and initiate the sequence of toxic reactions at different level. In contrast, the complexes of polyclonal rabbit anti-Aβ42 were non-toxic to cultured neurons suggesting that rat microglia do not recognize rabbit antibodies possibly because of a low affinity for rat FcγR.
We used two different monoclonal Aβ1-42 antibodies, the clone #11E12, that targeted N-terminal epitope and clone #G2-13, that targeted the C-terminus, and both in complexes with Aβ1-42 oligomers induced death of neurons in cultures indicating that the epitope position does not play any role in oligomeric antigen–antibody complex-induced cytotoxicity. Our results are in agreement with those obtained in passive immunization studies where microhemorrhage appearance and microglia activation followed treatment with N-terminal Aβ antibodies (Racke et al. 2005) as well as after C-terminal Aβ antibodies (Carty et al. 2006).
We did not find any significant differences in induction of neuronal death when either Aβ1-42 oligomers, fibrils, and monomers bound to the specific monoclonal antibody #11E12 were used most likely because of rapid oligomerization/fibrilization of Aβ1-42 monomers in aqueous solutions. However, our data revealed the difference in toxicity caused by complexes of #11E12 antibody with monomeric and oligomeric forms of Aβ1-40. The aggregation of Aβ1-40 monomers has been shown to be slow in aqueous media (hours to days) in contrast to rapid aggregation (minutes) of Aβ1-42 (Snyder et al. 1994), therefore, the Aβ1-40 monomeric preparation were likely to be in the monomeric state during experiments and therefore did not cause neurotoxicity when complexed with antibodies.
Our data also demonstrate that various oligomeric proteins when complexed with specific antibodies such as the antibodies raised against recombinant VLPs and NLPs can induce microglial Fc-receptor dependent death of cultured neurons. In contrast, complexes of short linear epitope MeN440-452 bound to anti-MeN were non-toxic to neurons. This confirms the finding that antigen oligomeric conformation of immune complexes is crucial in causing cell death in vitro. Such mechanism may be involved in neuroinflammatory responses caused by various viral infections particularly in the cases of disruption of BBB leading to penetration of viral oligomeric proteins and virus-specific antibodies into brain. Human polyomaviruses include the JC virus that infects the brain causing progressive multifocal leukoencephalopathy (PML) accompanied by brain inflammation and neuronal loss. Measles virus can also infect the brain causing acute encephalitis or more rarely subacute sclerosing panencephalitis, driven by brain inflammation and neuronal loss (Hayashi et al. 2002). If endogenous antibodies to oligomeric viral proteins enter the brain, they might cause neuroinflammation and neurotoxicity by the mechanism described here, and if so pathology might be limited by inhibiting microglial activation e.g. with minocycline.
The data presented here demonstrate that oligomeric/multimeric antigens complexed with their specific antibodies can exert cytotoxic effects on primary neurons by Fc-dependent microglia activation. Thus, if these antibodies cross the BBB and bind their oligomeric/multimeric antigens in the brain, they may induce microglia activation and neurotoxicity by this mechanism. Elucidation of the detailed mechanism of antibody–antigen complex-mediated neuroinflammation may help limit the neurotoxicity and potentially increase the effectiveness of immunotherapy for AD.
Authors thank Prof. Guy C. Brown for discussions and critical reading of the manuscript. This study was supported by the Research Council of Lithuania (grant LIG-04/2012 MALPAMA).
The authors declare no conflict of interest.
Authors' contributions: RM, AZ and VB designed the research and wrote the manuscript, ID, GV, PC, SJ, RJ, MJ and GB performed the research. All authors contributed to the final version of the manuscript.