Alzheimer’s disease is a severely debilitating disease of high and growing proportions. Hypercholesterolaemia is a key risk factor in sporadic Alzheimer’s disease that links metabolic disorders (diabetes, obesity and atherosclerosis) with this pathology. Hypercholesterolaemia is associated with increased levels of immunoglobulin G against oxidized lipoproteins. Patients with Alzheimer’s disease produce autoantibodies against non-brain antigens and specific receptors for the constant Fc region of immunoglobulin G have been found in vulnerable neuronal subpopulations. Here, we focused on the potential role of Fc receptors as pathological players driving hypercholesterolaemia to Alzheimer’s disease. In a well-established model of hypercholesterolaemia, the apolipoprotein E knockout mouse, we report increased brain levels of immunoglobulin G and upregulation of activating Fc receptors, predominantly of type IV, in neurons susceptible to amyloid β accumulation. In these mice, gene deletion of γ-chain, the common subunit of activating Fc receptors, prevents learning and memory impairments without influencing cholesterolaemia and brain and serum immunoglobulin G levels. These cognition-protective effects were associated with a reduction in synapse loss, tau hyperphosphorylation and intracellular amyloid β accumulation both in cortical and hippocampal pyramidal neurons. In vitro, activating Fc receptor engagement caused synapse loss, tau hyperphosphorylation and amyloid β deposition in primary neurons by a mechanism involving mitogen-activated protein kinases and β-site amyloid precursor protein cleaving enzyme 1. Our results represent the first demonstration that immunoglobulin G Fc receptors contribute to the development of hypercholesterolaemia-associated features of Alzheimer’s disease and suggest a new potential target for slowing or preventing Alzheimer’s disease in hypercholesterolaemic patients.
The pre-dementia phase of Alzheimer’s disease can be modelled in transgenic mice carrying human mutations (Ashe and Zahs, 2010). Such models are considered useful in the development of therapeutics aimed at slowing the transition from asymptomatic Alzheimer’s disease—in which the neuropathology is already initiated—to full-blown Alzheimer’s disease. On the other hand, the amplifying effects of secondary changes are now thought to confound therapeutic interventions and given that Alzheimer’s disease develops over decades, it is clearly important to improve understanding of the mechanisms that initiate this pathology. Animal models burdened with risks associated with the development of Alzheimer’s disease serve this goal better than those involving overexpression of molecules that are causally associated with Alzheimer’s disease.
The neuropathological characteristics of Alzheimer’s disease include amyloid β depositions and accumulations of abnormally hyperphosphorylated tau protein in selected brain regions; in addition, Alzheimer’s disease brains show abundant signs of microvascular damage and pronounced inflammation (Bertram et al., 2010). Current evidence suggests that amyloid β initiates a disease process that progresses to cognitive impairment and that tau mediates cognitive dysfunction (Ashe and Zahs, 2010). Although the pathways contributing to increased amyloid β levels appear to differ according to predisposing risk factors, it is reasonable to assume that these pathways converge at some point before triggering unbalanced amyloid β metabolism; identification of such an upstream hub would facilitate research and therapeutic developments.
Patients suffering from sporadic Alzheimer’s disease show elevated levels of immunoglobulin G (IgG) autoantibodies against non-brain antigens (Ounanian et al., 1990), and express receptors for the constant Fc region of IgG (FcγR) (Bouras et al., 2005) in neuronal populations that are vulnerable to the disease (Morrison and Hof, 2002). Importantly, increased circulating and brain IgG levels are found in individuals at risk of developing Alzheimer’s disease (Ballard et al., 2011), including advanced age (Listì et al., 2006), stress (Nakata et al., 2000) and metabolic disorders such as diabetes, obesity and atherosclerosis (Lu et al., 2001; Turk et al., 2001; Haroun and El-Sayed, 2007; Okamatsu et al., 2009; Winer et al., 2011). Abnormal modifications of proteins, such as oxidation and non-enzymatic glycosylation, are also observed during early stages of metabolic disorders (Haroun and El-Sayed, 2007; Winer et al., 2011); these result in an adaptive immune response with subsequently sustained high levels of circulating autoantibodies (mainly IgG isotype) (Doyle and Mamula, 2001) and the corresponding immune complexes formed by antibody–antigen interaction.
Antibody penetrance into the brain is severely limited in physiological conditions; however, weakening of the blood–brain barrier allows entry of immunoglobulins and immune complexes, as has been observed in individuals with metabolic syndrome, and risk for Alzheimer’s disease, and also in experimental models of Alzheimer's disease-related pathologies (Methia et al., 2001; Kuang et al., 2004; Hafezi-Moghadam et al., 2007; Bake et al., 2009; Diamond et al., 2009). Thus, we hypothesize that brain FcγR activation may be a major convergence point for sporadic Alzheimer’s disease triggered by the occurrence of two pathological phenomena: IgG overproduction and blood–brain barrier leakage.
In the mouse, four FcγR classes that differ in affinity, specificity and function have been described. Activating FcγRs (I, III and IV) associate with the common γ-chain which harbours the immunoreceptor tyrosine-based activation motif and elicit an immune response upon ligand (IgG or immune complexes) binding; in contrast, the inhibitory FcγRIIb contains the immunoreceptor tyrosine-based inhibition motif and nullifies cell activation (Nimmerjahn et al., 2005; Nimmerjahn and Ravetch, 2008). FcγRs have been localized in neurons, microglia, astroglia and oligodendrocytes (Bouras et al., 2005) and have been shown to play an important role in myelination (Nakahara et al., 2003).
Hypercholesterolaemia, a common feature in metabolic disorders, is characterized by autoimmune responses triggered by most forms of modified low density lipoproteins (Burut et al., 2010). Interestingly, hypercholesterolaemic mice display several neuropathological hallmarks of Alzheimer’s disease, including increased levels of amyloid precursor protein and amyloid β, abnormally hyperphosphorylated tau and inflammation, together with cognitive impairments (Oitzl et al., 1997; Veinbergs et al., 1999; Crisby et al., 2004; Rahman et al., 2005; Bjelik et al., 2006).
In this study, we investigated the potential role of FcγR in the aetiopathogenesis of Alzheimer’s disease. Given that a multiplicity of risk factors may be represented in hypercholesterolaemia and the fact that Alzheimer’s disease is associated with IgG overproduction and blood–brain barrier leakage, studies were performed in apolipoprotein E (apoE) knockout mice, which display overt hypercholesterolaemia (Zhang et al., 1992). Therefore, the features characteristic of Alzheimer’s disease were comparatively studied between single apoE knockout (apoE−/−) mice and double knockout (DKO) mice (Hernandez-Vargas et al., 2006), which are gene deficient in both apoE and γ-chain, the common subunit necessary for assembly, cell-surface localization and functionality of activating FcγRs (Nimmerjahn and Ravetch, 2008).
Materials and methods
Soluble IgG immune complexes (Oreskes and Mandel, 1983) were obtained by heat aggregation (30 min at 63°C) of monomeric mouse IgG (Cappel, MP Biomedicals) and subsequent centrifugation to eliminate insoluble immune complexes as described (Hernandez-Vargas et al., 2006). Culture media, supplements and transfection reagent were purchased from Lonza and Life Technologies; SP600125 and U0126 from Stressgen Bioreagents Corp.; small interfering RNA from Santa Cruz Biotechnologies. The 9E9 FcγR blocking antibody was generously provided by Dr J.V. Ravetch (The Rockefeller University, NY) (Nimmerjahn et al., 2005). Primary antibodies used were: synaptophysin, microtubule associated protein 2, and tau phosphorylated at Ser199/202 (Millipore); tau doubly phosphorylated at Ser202/Thr205, Ser199/202 and Ser205/208 (AT8 antibody) (Pierce Biotechnology Inc.); glial fibrillary acidic protein (GFAP; Sigma-Aldrich); a neoepitope in amyloid β, to exclude cross-reactions with its precursor amyloid precursor protein (IBL); FcγRIV (Santa Cruz Biotechnologies); mouse IgG (Amersham); β-site amyloid precursor protein cleaving enzyme 1 (BACE1; ProSci Inc.); insulin degrading enzyme (Abcam); CD11b (Abcam); α-tubulin (Sigma-Aldrich); and glyceraldehyde-3-phosphate dehydrogenase (GAPDH, Millipore). Assays for real-time PCR were from Life Technologies.
All animal studies conformed to Directive 2010/63/EU of the European Parliament and were approved by the Institutional Animal Care and Use Committee. Wild-type and apoE knockout mice were purchased from Jackson Laboratory; the DKO mice (gene deficient in both apoE and γ-chain) were generated in our laboratory as previously described (Hernandez-Vargas et al., 2006). Wild-type, apoE−/− and DKO mice were fed a standard mouse lab chow diet (Panlab) and allowed to age. Mice from each group were sacrificed at 12 months (wild-type, n = 5; apoE−/−, n = 8; DKO, n = 7) and 21 months (wild-type, n = 5; apoE−/−, n = 12; DKO, n = 11). Mice brains were obtained after deep anaesthesia with a mixture (0.01 ml/g body weight, intraperitoneally) of ketamine (10 mg/ml) and xylazine (1 mg/ml) and perfusion with saline.
Mice were housed three to five per cage and allowed to habituate to the cage environment for 2 weeks before behavioural testing. Housing conditions were kept constant until the end of behavioural testing. All behavioural tests were carried out under dim light in a sound-proof room by the same researcher and were conducted in the following order: elevated plus maze, object location test.
Elevated plus maze
Mice were allowed to freely explore the maze for 8 min. Anxiety index was calculated as the time spent in open arms relative to the total time spent in closed and open arms and in the centre. Locomotor activity index was calculated as frequency of entry into closed arms.
Object location test
One day after the elevated plus maze test, mice were habituated to the empty arena placed in a room with extra-maze cues for 15 min during 5 days before training. In the training phase, mice were exposed to two identical objects for 5 min. After a delay of 14 min, the time spent exploring the objects in a new (novel) and in the old (familiar) locations was recorded during 3 min (test phase). To analyse cognitive performance, a location index (or relative exploration time) was calculated as previously described (Murai et al., 2007) with a slight modification: to/Te, where to = time spent exploring either the displaced or the non-displaced object and Te = total time spent exploring both objects.
Total serum cholesterol concentrations were measured using standard enzymatic methods (Invitrogen). Total serum immunoglobulins were measured by sandwich ELISA using specific antibodies recognizing mouse IgG1, IgG2a/c and IgG3 (BD Biosciences).
Primary neuronal cultures
Mouse cortical neurons from prenatal embryonic Day 17 were mechanically dissociated and plated on poly-l-lysine-coated coverslips or plates. Cell suspension (106 cells/ml) was seeded in Dulbecco’s modified Eagle medium containing 5% foetal bovine serum, 0.05% GlutaMAX™ and 1% penicillin/streptomycin, and then cultured in Neurobasal® medium supplemented with 2% B27, 1% GlutaMAX™ and 0.1% penicillin/streptomycin. On Day 5 in vitro, 10 µM cytosine arabinoside was added. Contaminating glial cells in neuronal cultures accounted for <2% (astroglia and microglia combined) on the day of use. Cultures at Day 12 were stimulated with IgG immune complexes (150 μg/ml). In some experiments, cells were pretreated for 1 h with mitogen-activated protein kinases inhibitors [c-Jun N-terminal kinase (JNK) inhibitor, SP600125, 5 × 10−5 M; extracellular regulated mitogen-activated protein kinase (ERK) inhibitor, U0126, 10−5 M] or FcγRIV blocking antibody (9E9 or irrelevant hamster IgG, 1 µg/106 cells; Nimmerjahn et al., 2005) before stimulation. For knock-down experiments, cells were transfected with 60 pM of small interfering RNA (FcγRIV or irrelevant) and 8 µl of Lipofectamine™ 2000 in culture medium. Small interfering RNA knocking-down efficiency checked by real-time PCR was 72.1 ± 0.3% at 24 h after transfection (n = 3).
Brains were fixed in 4% paraformaldehyde for 4 h. Cryostat sections (10 µm) on slice were used for immunohistochemistry. For antigen-retrieval in amyloid β immunostaining, sections were pretreated with 88% formic acid for 5 min. Non-specific binding sites were blocked by incubating sections in 1% bovine serum albumin and 5% pre-immune serum diluted in 0.5% Tween-20 in PBS for 1 h at room temperature in slight orbital agitation. Primary antibody was incubated in 0.5% Tween-20 (FcγRIV 1:50, amyloid β 1:100, GFAP 1:2000; CD11b 1:20) for three overnights at 4°C in slight orbital agitation.
For immunocytochemistry, fixation in 2% paraformaldehyde (10 min, room temperature) was performed before incubation of primary antibodies in PBS overnight at 4°C in slight agitation (FcγRIV 1:50, microtubule associated protein 2 1:100, amyloid β 1:100, GFAP 1:2000, CD11b 1:20). For both in vivo and in vitro studies, fluorescent secondary antibodies (1:100 in PBS) were incubated at room temperature for 1 h with orbital agitation. Incubation with 4′-6-diamidino-2-phenylindole (DAPI; 1:10 000 in PBS) was then performed (10 min, room temperature). Confocal images were obtained by sequential scan.
Cytosolic proteins were extracted from cells (Ortiz-Munoz et al., 2010) and the cerebral cortex from mice (Rodrigo et al., 2004) and resolved on SDS–PAGE, transferred onto polyvinylidene fluoride membranes and immunoblotted with specific antibodies (Ortiz-Munoz et al., 2010). Optical densities of individual bands were normalized to loading controls (GAPDH or α-tubulin) and n-fold changes were obtained by normalization against results obtained under basal conditions (untreated cells) or control (wild-type) mice, as appropriate.
Messenger RNA expression
Total RNA from cells and tissues was extracted and retro-transcribed as previously described (Ortiz-Munoz et al., 2010). Gene expression was analysed in duplicate by real-time PCR on a TaqMan® ABI 7500 sequence detection system (Applied Biosystems) and normalized to housekeeping 18S transcripts. Results are given as n-fold changes, relative to control groups. The relative expression levels of activating and inhibitory FcγR were calculated according to the formula (FcγRI + FcγRIII + FcγRIV)/FcγRIIb and expressed as a percentage of control values.
Statistical significance was tested by unpaired Student’s t-test and one-way ANOVA followed by an appropriate post hoc least significant difference comparison test (GraphPad Prism software). A value of P < 0.05 was considered to be statistically significant.
FcγR deficiency restores cognitive and synaptic status and attenuates Alzheimer-like pathology in hypercholesterolaemic mice
Western blot studies in brain lysates demonstrated very low levels of IgG in middle-aged (12-month-old) wild-type mice and high levels in apoE−/− and DKO (deficient in apoE and γ-chain genes) mice, with both knockout strains showing a similar content (Fig. 1A). ApoE−/− and DKO groups showed comparable serum cholesterol levels (460 ± 27 mg/dl versus 475 ± 20 mg/dl, P > 0.05). Furthermore, serum antibody levels confirmed increased brain IgG levels in apoE−/− and DKO mice, as compared to wild-type mice. The two strains of knockout mice (apoE−/− and DKO) did not differ in their IgG subclass (IgG1, IgG2 and IgG3) expression profiles (Fig. 1B), i.e. they shared a similar immune imprint.
Given that high levels of IgG reportedly upregulate FcγR expression in different inflammatory disease models (Nimmerjahn and Ravetch, 2008), we next examined whether experimental hypercholesterolaemia altered FcγR gene expression in the brain. Consistent with the increase in IgG levels, we found an increased expression of activating and inhibitory FcγRs, mainly of the FcγRIV isoform, in brains from apoE−/− animals, when compared with wild-type mice (Fig. 1C); this resulted in a net activating profile (activating:inhibitory ratio, percentage versus wild-type mice: cortex, 6.06; hippocampus, 13.99). In contrast, apart from changes in the expression of FcγRIII, hypercholesterolaemic DKO mice did not show any difference in FcγR expression in the cortex and hippocampus, compared to wild-type animals (Fig. 1C).
Middle-aged hypercholesterolaemic apoE−/− mice showed intracellular amyloid β immunostaining in pyramidal neurons of the hippocampus and temporal cortex (Fig. 2A and C), consistent with the early neuroanatomical and cellular distribution of intracellular amyloid β deposition found in patients with Alzheimer’s disease (Gouras et al., 2000; Gyure et al., 2001; Fernandez-Vizarra et al., 2004). Immunohistochemical localization of amyloid β was similar to that of FcγRIV (Fig. 2D and E), the major receptor expressed in hypercholesterolaemic mice brains; FcγRIV staining was observed in pyramidal neurons of the hippocampus, temporal cortex, but also of the cingulate cortex (data not shown) of apoE−/− mice, but not in other types of neuron or glial cells. ApoE−/− mice also showed a significant increase in tau hyperphosphorylation in the cerebral cortex, as compared to age-matched wild-type controls (Fig. 2H). Importantly, deletion of activating FcγR in DKO mice attenuated amyloid β immunostaining (Fig. 2A and C) and tau hyperphosphorylation (Fig. 2H) in all neuroanatomical areas studied. Interestingly, FcγR deficiency was also protective against amyloid β deposition in aged (21-month-old) mice (Fig. 2B). Furthermore, lack of functional FcγRs significantly reduced the gene expression of the inflammatory cytokine tumour necrosis factor α (TNFα) and the monocyte chemoattractant protein 1 (MCP1, also known as Ccl2; Fig. 2J) as well as the astroglial marker GFAP (Fig. 2I) in cerebral cortex of hypercholesterolaemic mice. Immunohistochemical studies revealed that astrogliosis was restricted to the hippocampus (Fig. 2F), the region most affected by amyloid β deposition. By contrast, microgliosis, detected in the hippocampus (Fig. 2G) and across several cortical areas (data not shown), did not correlate with the pattern of amyloid β deposition or FcγRIV overexpression and was not affected by FcγR deficiency.
A robust loss of synapses was found in apoE−/− mice (Fig. 3A), as judged by the expression of synaptophysin; currently, synaptic loss appears to be the best correlate of cognitive dysfunction in Alzheimer’s disease (Masliah et al., 2001; Reddy et al., 2005; Gomez et al., 2010). Importantly, FcγR deletion completely rescued synaptophysin protein levels in hypercholesterolaemic mice (Fig. 3A). This recovery was shown to be functionally relevant as shown by performance in the object location test, a measure of hippocampus-medial temporal lobe related learning and memory (Eichenbaum et al., 2007). In contrast to apoE−/− mice, DKO mice and wild-type mice discriminated between novel and familiar locations to similar extents, indicating intact cognitive function in the DKO mice (Fig. 3B). Specifically, differences observed in spatial learning and memory were not a function of total object exploration times in the training (wild-type, 15.7 ± 6.1 s; apoE−/−, 17.3 ± 5.9 s; DKO, 5.3 ± 1.2 s, n = 4–10) and test phases (wild-type, 10.3 ± 3.2 s; apoE−/−, 7.5 ± 2.1 s; DKO, 4.0 ± 1.6 s, n = 4–10) or of different relative exploration times of the two objects during the acquisition phase of the task (P > 0.05 in all cases). Furthermore, the observed differences in cognitive performance could not be explained by increased anxiety or decreased locomotor activity, as revealed in the elevated plus maze test (Fig. 3C).
Engagement of neuronal FcγR in primary cultures induces Alzheimer-like pathology
The role of neuronal FcγR in hypercholesterolaemia-associated Alzheimer-like pathology was further studied in primary neuronal cultures from wild-type, apoE−/− and DKO mice. In wild-type and apoE−/− neurons, IgG immune complexes induced a large increase in the expression of genes for all activating FcγR isoforms, mainly FcγRIV (IV > I ≈ III); smaller effects were observed with regard to the inhibitory FcγRIIb isoform, resulting in a net activating profile (activating/inhibitory ratio, n-fold versus basal: wild-type, 2.58; apoE−/−, 1.57; Fig. 4A). Immmunofluorescence confirmed the induction of the activating FcγRIV isoform by IgG immune complexes (Fig. 4B). In contrast, immune complexes failed to induce activating FcγR expression in neurons from DKO mice; these neurons showed an upregulation of inhibitory FcγRIIb only (Fig. 4A and B). In parallel, amyloid β accumulation was observed after FcγR engagement in wild-type pyramidal neurons (Fig. 5A). Immune complexes also induced tau hyperphosphorylation at epitopes that are considered important in Alzheimer’s disease (Fig. 5B and C). Furthermore, significant synaptic loss was found in immune complex-stimulated primary neurons from wild-type and apoE−/− mice (Fig. 5D). Importantly, FcγR deficiency attenuated amyloid β deposition and tau hyperphosphorylation (Fig. 5A–C), protected primary neurons against synapse loss (Fig. 5D) and significantly blocked the expression of inflammation-related genes after stimulation with immune complexes (Fig. 5F).
To directly assess the involvement of FcγRIV, the most inducible activating FcγR found in vivo and in vitro, inhibition experiments were performed with small interfering RNA and neutralizing antibody. FcγRIV small interfering RNA efficiently abrogated intracellular amyloid β accumulation (Fig. 5E), tau phosphorylation (Fig. 5C) and synaptic loss (Fig. 5D) in response to immune complexes. However, FcγRIV small interfering RNA partially reduced MCP1 gene expression without significant effects on TNFα (Fig. 5F and G). Similarly, blocking FcγRIV with a highly specific antibody prevented tau phosphorylation and synaptic loss and attenuated MCP1 protein expression (Fig. 5C, D and G). Furthermore, and consistently with our in vivo results, confocal microscopic studies using specific markers of astro- and microglial cells (GFAP and CD11b, respectively) failed to reveal FcγRIV immunostaining in either astroglia or microglia (data not shown, note that cultures contained <2% of glial cells), suggesting that neurons are responsible for the observed responses following treatment with immune complexes.
Role of BACE1, JNK and ERK pathways in the neuroprotective effects of FcγR deficiency
Both increased production and decreased clearance of amyloid β have been implicated in sporadic Alzheimer’s disease in humans (Yang et al., 2003; Li et al., 2004; Mawuenyega et al., 2010). Accordingly, we investigated whether the enzymes BACE1 and insulin degrading enzyme were modulated by hypercholesterolaemia and FcγR functional deletion. Western blot analysis for BACE1 in the cerebral cortex from apoE−/− mice revealed a significant increase in the 140-kDa band, with a parallel decrease in the 70-kDa band (Fig. 6A), suggestive of an increase in homodimeric BACE1 activity in hypercholesterolaemia (Schmechel et al., 2004; Jin et al., 2010). Interestingly, the effects of hypercholesterolaemia on BACE1 were abolished when the FcγR gene was deleted (Fig. 6A). On the other hand, insulin degrading enzyme levels remained unchanged under hypercholesterolaemic conditions and were unaffected by FcγR deletion, as shown by western blot analysis (Fig. 6B).
Activation of JNK and ERK pathways have been implicated in synaptic plasticity (Curtis and Finkbeiner, 1999), amyloid β production and tau phosphorylation (Sato et al., 2002; Colucci-D'Amato et al., 2003) and are known to be involved in FcγR signalling (Song et al., 2002,; Luo et al., 2010). In our in vivo model, hypercholesterolaemia was associated with activation of JNK and ERK pathways; in contrast, JNK and ERK activation levels were similar in wild-type and FcγR-deficient mice (Fig. 7A). In vitro, immune complexes induced a sustained activation of both JNK and ERK; these changes were seen before and during expression of Alzheimer’s disease markers and synaptic loss in wild-type cultures, and were precluded by FcγR deficiency (Fig. 7B). Moreover, pretreatment of neurons with specific inhibitors of JNK and ERK significantly attenuated the tau hyperphosphorylation and synaptic loss induced by immune complex treatment (Fig. 7C and D).
This study used an animal model of hypercholesterolaemia to investigate the upstream pathways that lead to neuropathological features characteristic of Alzheimer’s disease, i.e. those preceding inappropriate increases in amyloid β. Targeting early pathological mechanisms before the onset of accumulation of amyloid β and associated amplifying events, such as neuroinflammation and vascular changes, can be expected to facilitate therapeutic interventions. While genetic and molecular data indicate the important role of amyloid β in the initiation of Alzheimer’s disease (Selkoe, 2001; Hardy, 2006), a number of recent clinical trials, aimed at reducing amyloid β burden in mild-to-moderate stages of dementia, have failed (Shepardson et al., 2011b). Accordingly, the importance of focusing on factors that induce amyloid β accumulation and in advance of secondary changes has become even more evident.
It is now recognized that hypercholesterolaemia, often found in diabetes, obesity and atherosclerosis, is a key risk factor in sporadic Alzheimer’s disease (Ballard et al., 2011; Shepardson et al. 2011a). In support of this, several studies have shown that the molecular, neuroanatomical and cognitive changes found in Alzheimer’s disease are recapitulated in hypercholesterolaemic mice (Oitzl et al., 1997; Veinbergs et al., 1999; Crisby et al., 2004; Rahman et al., 2005; Bjelik et al., 2006). The work presented here represents the first to suggest a pivotal role for brain FcγR in the pathogenesis of sporadic Alzheimer’s disease. We show that hypercholesterolaemia results in a significant increase in brain IgG levels with a parallel increase in activating FcγR gene expression. Furthermore, hypercholesterolaemic mice exhibited increased protein expression of FcγRIV isoform in pyramidal neurons vulnerable to intraneuronal amyloid β deposition in the hippocampus and temporal cortex, regions similarly affected in the brains of patients with Alzheimer’s disease (Gouras et al., 2000; Gyure et al., 2001; Fernandez-Vizarra et al., 2004). Notably, we show that FcγRIV is more widely expressed than amyloid β; besides the hippocampus and temporal cortex, we observed FcγRIV staining in the cingulate cortex, an area that is affected in early-stage Alzheimer’s disease (Barnes et al., 2007; Pengas et al., 2010). These findings are complemented by the observation that cross-linking of FcγR with IgG immune complexes leads to amyloid β accumulation, hyperphosphorylation of tau at sites associated with Alzheimer’s disease and synaptic loss in primary neuronal cultures. An important finding to emerge from this study is that deletion of FcγR function in hypercholesterolaemic mice prevents learning and memory impairments. These cognition-protective effects were associated with a reduction in amyloid β deposition, abnormal tau hyperphosphorylation and synapse loss. Interestingly, the above mentioned behavioural and neuronal consequences of FcγR deletion appear to occur independently of brain IgG and blood cholesterol levels since neither of these parameters are altered in FcγR null mutant mice.
Support for the view that the FcγRIV isoform is responsible for mediating the detrimental effects of immune complexes on neurons stems from experiments in which selective FcγRIV inhibition (small interfering RNA silencing and blocking antibody) rescued immune complex-induced synaptic loss and the accumulation of amyloid β and hyperphosphorylated tau. It is important to note that our analysis revealed that FcγRIV expression is restricted to neurons both in vivo and in vitro, as no signal was detected in astro- and microglia. The results, obtained in primary neuronal cultures, indicate that MCP1 and TNFα are unlikely to exert substantial influence over the development of amyloid β accumulation, tau pathology and synaptic loss in primary neurons: although FcγRIV knock-down in neurons did not preclude cytokine (MCP1 and TNFα) induction by immune complexes, this inflammatory response was not followed by the aforementioned neuropathological features. Nevertheless, a potentially protective role of these cytokines cannot be ruled out at present. In this context, it is worth mentioning that MCP1, a cytokine involved in the recruitment, proliferation and activation of glial cells, may trigger gliosis in the hippocampus; since astro- and microglia do not express the only hypercholesterolaemia-responsive FcγR isoform (FcγRIV), gliosis in the hippocampus cannot be attributed to activation of FcγR. If early inflammatory responses serve to retard hypercholesterolaemia-associated Alzheimer’s disease pathology, interesting therapeutic opportunities could be explored by exploiting the differential regulation of inflammation and Alzheimer-related pathology by FcγR isoforms; importantly, these would not necessarily interfere with the potentially protective functions of glial cells.
Our hypothesis that IgG is involved in the onset of certain sporadic forms of Alzheimer’s disease is consistent with recent data suggesting that factors that induce weakening of the blood–brain barrier may be causally related to neurodegenerative disease; since this barrier gets leaky during normal ageing (Hafezi-Moghadam et al., 2007), hypercholesterolaemia would add to or multiply the risk of developing Alzheimer’s disease. Indeed, a recent study in mice showed that disruption of the blood–brain barrier, and subsequent deposition of IgG in the brain, causes synaptic loss and memory impairment (Bell et al., 2010). Loss of integrity of the blood–brain barrier, as well as IgG deposits and/or FcγR expression, are also associated with Alzheimer’s disease risk factors other than ageing, namely atherosclerosis, obesity and diabetes (Methia et al., 2001; Kuang et al., 2004; Hafezi-Moghadam et al., 2007; Bake et al., 2009; Diamond et al., 2009). Moreover, the brain is among the most commonly affected organs in patients with immune-mediated diseases, such as systemic lupus erythematosus, subacute endocarditis and hepatitis C infection (Harris and Cobbs, 1996; Lister and Hickey, 2006; Carvalho-Filho et al., 2012), in which neuronal dysfunction may result from direct immune effects (autoantibody binding to cell surface, circulating immune complex deposition and inflammation) on brain resident cells via FcγR and complement activation. Although these immune conditions might influence the likelihood of developing Alzheimer’s disease and other dementias, further research in this field is clearly worthwhile.
In light of strong evidence that amyloid β production is increased (Yang et al., 2003; Li et al., 2004) and cleared with reduced efficiency (Mawuenyega et al., 2010) in sporadic Alzheimer’s disease, we here explored the mechanisms through which FcγR activation may lead to increased levels of cerebral amyloid β. We focused on BACE1, whose homodimerization and association with amyloid precursor protein leads to the initial cleavage of amyloid precursor protein (Schmechel et al., 2004; Jin et al., 2010) and whose expression and activity is upregulated in patients with sporadic Alzheimer’s disease (Yang et al., 2003; Li et al., 2004). We observed increased amounts of homodimeric BACE1 in hypercholesterolaemic mice, an effect that was abolished when the FcγR gene was deleted. Further, in line with previous work that reported that BACE1 homodimerization depends on its phosphorylation at specific serine/threonine residues (Walter et al., 2001), we observed sustained activation of JNK and ERK in the brains of hypercholesterolaemic apoE−/− mice, and in cultured neurons that were exposed to immune complexes; activated JNK and ERK have been previously implicated in the generation of amyloid β and the abnormal hyperphosphorylation of tau (Sato et al., 2002; Colucci-D'Amato et al., 2003).
In summary, our study describes a novel mechanism that may trigger certain sporadic forms of Alzheimer’s disease. The results point to FcγR as a potential target for preventative intervention, at least in cases where hypercholesterolaemia is a risk factor.
Although preventive approaches in those patients may primarily include reduction of hypercholesterolaemia, recent evidence indicates that lipid-lowering drugs are inefficacious treatments for Alzheimer’s disease (Shepardson et al., 2011b). Alternatively, strategies targeting antibody production and FcγR balance and activities could determine the net functional effect and hence retard disease onset and progression. In future, it will be also important to determine whether levels of IgG immune complexes can serve as early biomarkers of sporadic Alzheimer’s disease and to assess the potential contribution of FcγR signalling to forms of Alzheimer’s disease that are not directly associated with hypercholesterolaemia.
Spanish Ministry of Science (SAF2009/11794), Fondo de Investigaciones Sanitarias (FIS PI10/00072, RECAVA RD06/0014/0035), Fundacion Renal Iñigo Alvarez de Toledo, Spanish Society of Nephrology and Lilly Foundation. P.F.-V. was supported by postdoctoral fellowships from FIS (Sara Borrell program) and Caja Madrid Foundation.
The authors thank Dr J. V. Ravetch (The Rockefeller University, New York) for generously providing FcγRIV blocking antibody, Dr M.P. Sanchez and A.M. Garcia-Cabrero (Laboratory of Neurology, IIS-FJD, Madrid) and Drs C. Lopez-Menendez and L. Sanchez-Ruiloba (IIB Alberto Sols, CSIC-UAM) for antibody samples and skilled advice with mice and primary cultures.
β-site amyloid precursor protein cleaving enzyme 1
c-Jun N-terminal kinase
extracellular regulated mitogen-activated protein kinase
glial fibrillary acidic protein
monocyte chemoattractant protein 1
receptor for the constant Fc region of immunoglobulin G
tumour necrosis factor α