Background
The main neuropathological hallmarks of Alzheimer’s disease (AD) are synaptic dysfunction, neuron loss, amyloid plaques, composed of aggregated Aβ peptide, and neurofibrillary tangles, containing hyperphosphorylated forms of the protein tau [
1]. Preclinical findings in animal models indicate that active and passive immunization against classical molecular markers of AD might represent suitable therapeutic strategies [
2,
3]. However, in clinical trials, Aβ-targeted immunotherapies have shown limited efficacy against AD cognitive symptoms [
4,
5], and have been associated with various adverse effects such as microhemorrhages, vasogenic oedema and aseptic meningitis [
6].
Intravenous immunoglobulin (IVIg), which is composed of over 98% human immunoglobulin G (hIgG), is used in the treatment of an increasing number of diseases and is generally safe and well tolerated [
7]. Natural autoantibodies against Aβ peptide and oligomers have been reported in the blood of healthy individuals and in IVIg preparations [
8,
9]. Initial evidence of IVIg efficacy comes from pilot studies in which IVIg improved cognition and reduced Aβ in the cerebrospinal fluid (CSF) in AD patients [
10,
11]. Results from a large phase III clinical trial (ClinicalTrials.gov identifier: NCT00818662) presented at the 2013 Alzheimer’s Association International Conference indicate that an 18-month treatment with IVIg acted on plasma and positron emission tomography (PET) biomarkers, but did not improve cognitive scores in mild to moderate AD patients at the doses studied [
12]. However, subgroup analyses unveiled improved cognitive endpoints in APOE4 carriers [
12], suggesting clinical benefit in a subpopulation representing almost 40% of AD patients [
13]. These large clinical trials also confirmed the good safety profile of IVIg in AD patients [
12,
14]. Finally, a retrospective case-controlled study using anonymous medical data indicates that IVIg-treated patients have a 42% lower incidence rate of dementia than an untreated population [
15]. Overall, these preliminary data strongly suggest that further clinical studies better powered for subgroup analyses and using larger doses of IVIg are warranted in AD.
Regardless of the results of the clinical assays, the amount that would be needed to treat the 24 million patients afflicted worldwide [
16] precludes a widespread use of this product which has to be purified from the plasma of healthy donors. Preclinical research in animal models is thus essential to identify potential pharmacological targets, and to develop a replacement therapy to avoid a massive shortage of plasma that would ensue the use of IVIg as a first-line treatment for AD. The present investigation aimed at replicating the beneficial effects of IVIg in the 3xTg-AD mouse model to decipher its potential mechanisms of action.
Discussion
Our results are consistent with IVIg-induced improvement of behavioral function, reduction of Aβ*56 oligomer levels and immunomodulation in the 3xTg-AD mouse model, without altering the non-amyloid aspects of AD neuropathology. To our knowledge, this is the first demonstration that chronic administration of IVIg can strikingly decrease levels of the pathogenic oligomer Aβ*56 in association with reduced expression of peripheral CX3CR1 and attenuation of behavioral deficits in a mouse model of AD. IVIg also displayed strong immunomodulatory properties, leading to a correction of immune abnormalities frequently observed in AD and animal models.
The use of IVIg in AD was initially motivated by the hypothesis that it contains natural, polyclonal, conformation-specific antibodies against Aβ. This view is supported by the lower titer of anti-Aβ antibodies found in the blood of AD patients compared to controls [
8,
9]. We thus analyzed the impact of IVIg on various parameters of brain amyloid pathology and found no significant reduction of either Aβ40 or Aβ42 in both soluble and insoluble protein fractions from treated mice, consistent with a recent report in which IVIg treatment in the AβPPswe/PS1ΔE9 mouse model of AD failed to decrease Aβ concentrations in the hippocampus [
38]. However, we observed a 22% decrease in the soluble Aβ42/Aβ40 ratio following IVIg treatment in 16-month-old 3xTg-AD mice. This finding is interesting in view of the fact that, in familial AD, most known APP mutations increase the Aβ42/Aβ40 ratio without necessarily changing the total concentration of Aβ peptides formed, shifting the proteolysis of APP in favor of Aβ42, which is more prone to oligomerization [
39]. Furthermore, an
in vitro study of APP and Aβ processing in familial AD indicates that the Aβ42/Aβ40 ratios correlate inversely with the age of onset of AD [
40]. In the Tg2576 mouse, a reduction of spine density, a decline in long-term potentiation, fear conditioning impairments and an increase in Aβ42/Aβ40 ratio precede amyloid plaque deposition [
41]. Moreover, an approximate 30% increase in the insoluble Aβ42/Aβ40 ratio is associated with spatial memory deficits following a partial loss of glutamate transporter 1 in the AβPPswe/PS1ΔE9 mouse model [
42]. Consistent with these findings, a substantial decrease in the soluble Aβ*56 oligomer species was also observed in IVIg-treated 3xTg-AD mice. There is no consensus on the actual relevance and toxicity of the various Aβ oligomers associated with AD pathogenesis. The Aβ*56 species are found at the AD synapses [
43] and are elevated in the CSF of cognitively normal adults at greater risk for AD [
44]. In animal models, intracerebral administration of Aβ*56 produces cognitive impairments in a concentration-dependent manner [
45,
46]. In addition, Aβ*56 levels show a better association with learning/memory deficits than plaque load [
25] in most transgenic AD models. Finally, in cognitively intact elderly subjects, Aβ*56 correlates positively with soluble pathological tau species and negatively with the postsynaptic proteins, drebrin and fyn kinase, suggesting that Aβ*56 may play a pathogenic role very early in the pathogenesis of AD [
47]. The present data, in line with lower incidence rate of dementia in IVIg-treated patients [
15], suggests that IVIg impedes accumulation of Aβ oligomers possibly by an effect on their production, aggregation, degradation or clearance, and might prevent AD in the pre-clinical stage. Furthermore, although not significant in our study, Puli and colleagues [
38] reported a significant rise in the soluble levels of Aβ40 and Aβ42 peptides in the AβPPswe/PS1ΔE9 mouse model following an 8-month treatment with IVIg that would be consistent with decreased Aβ oligomer/monomer ratio following IVIg injections.
In addition to its anti-Aβ action, it can be hypothesized that the immunomodulatory effect of IVIg contributes to its effect in the CNS [
8]. Indeed, IVIg administration increases C5a brain levels [
48] and reduces the expression of the CD45 marker in a sub-population of microglial cells in mice, in association with increased neurogenesis [
38]. We found that chronic IVIg treatment steadily decreases the CD4/CD8 cell ratio in 3xTg-AD mice, as previously reported in a mouse model of Parkinson’s disease [
24]. Such a decrease in the CD4/CD8 cell ratio was also reported in IVIg-treated patients [
49], suggesting that it may actually provide a clinically relevant index of IVIg efficacy. Interestingly, the natalizumab-induced decrease in CD4/CD8 ratio in multiple sclerosis patients is significantly related to clinical response [
50,
51]. In AD patients, however, the actual relevance of the CD4/CD8 ratio is still much debated. Indeed, increased [
52,
53], unchanged [
54] and even decreased [
55] CD4/CD8 ratios have been reported in AD patients compared to controls. Thus, the clinical implications for AD progression of the IVIg-induced decrease of the CD4/CD8 ratio observed here remain to be further established. We also report evidence that IVIg decreases the protein levels of YKL-40 in the cortex of older mice. Although the function of YKL-40 is still under study, it has been associated with local neuroinflammation in acute and chronic diseases [
33], and is increased in the CSF of AD patients [
56]. We also report altered pro/anti-inflammatory ratios of cytokines and chemokines in 3xTg-AD mice. In agreement with the extensive body of evidence showing an imbalance in cytokine and/or chemokine production in the blood, CSF or brain of individuals diagnosed with AD [
57], we observed a rise in the ratios of IL-5, IL-12, MCP-1 and GM-CSF over IL-10 in the brain of 3xTg-AD mice. Furthermore, although IVIg treatment had no effect on GM-CSF/IL-10 and MCP-1/IL-10, it reduced both IL-5/IL-10 and IL-12/IL-10 ratios. Atopy is a genetic predisposition to hypersensitivity reactions against common environmental allergens, which manifests itself as asthma, dermatitis or rhinitis [
58]. Whereas an increased IL-5/IL-10 ratio has been reported in patients with atopic asthma [
59], atopy itself is associated with a modest rise in the risk of dementia [
58], in support of an inflammatory component in the etiology of neurodegenerative dementia.
Finally, recent studies have linked the anti-inflammatory effects of IVIg to terminal α2,6-linked sialic acid residues on the N-linked glycans of a sub-population of the IgG fragment crystallizable (Fc) domain [
60‐
62]. These sialylated IgG bind to the human receptor dendritic cell-specific intercellular adhesion molecule-3-grabbing non-integrin (DC-SIGN) or its murine orthologue, specific intracellular adhesion molecule-grabbing non-integrin receptor 1 (SIGN-R1) and, in animal models of autoimmune diseases, this interaction leads to the expression of anti-inflammatory cytokines and receptors [
61]. Therefore, sialylated IgG might recapitulate the immunosuppressive action of IVIg. If confirmed, the sialylated fraction could provide a suitable alternative to this blood-derived product. Interestingly, the SIGN-R1 receptor is also expressed on mouse microglia [
63] and therefore could interact with sialylated IVIg in the brain. However, it is noteworthy that the results of IVIg treatment in the 3xTg-AD model were not limited to immunosuppression suggesting that sialylated IVIg fractions could hardly account for all the effects observed on AD markers. Taken together, these observations support the view that IVIg acts, at least in part, on maintaining brain immune homeostasis, providing a favorable environment against neurodegenerative diseases.
The interaction between CX3CR1 and its exclusive ligand fractalkine is required for the physiological trafficking of circulating monocytes to organs, and is an important regulator of autoimmune inflammation and antigen-specific leukocyte recruitment from the blood and the bone marrow to the CNS [
34]. Interestingly, IVIg is used in the treatment of idiopathic thrombocytopenic purpura and increased expression of CX3CR1 has been observed in these patients [
7,
64]. Thus, our data support the reduction of CX3CR1 expression on peripheral leucocytes as a new mechanism of action for IVIg. Studies on the role of the fractalkine pathway in animal models of AD have generated somewhat contradictory results. CX3CR1 genetic depletion was found to protect from neuronal loss in the 3xTg-AD model [
37] and reduce β-amyloid deposition by activating its phagocytosis in CRND8, APPPS1 and R1.40 mice [
35,
36], whereas it exacerbates tau phosphorylation and aggregation, and enhances cognitive deficits in hTau and hAPP mice [
65,
66]. These results strongly suggest that CX3CR1 signaling regulates microglial activity and neuropathological processes in AD. However, genetic knockouts of CX3CR1 hardly mimic the actual trends in CX3CR1 expression in physiological conditions and during the progression of the disease. In the brain, the fractalkine pathway mediates the communication between neurons, which produce fractalkine, and microglia, which express CX3CR1, as binding of the two inhibits microglial activation [
67]. In the present study, IVIg administration had no effect on brain expression levels of CX3CR1 or fractalkine, as detected by Western blot analysis. However, our data show an 11 to 13% decrease in CX3CR1
+ cells in the bone marrow from IVIg-treated 3xTg-AD mice, which was correlated with a reduction of both Aβ*56 concentration and Aβ42/Aβ40 ratios in the cortex. A recent study by Smolders and colleagues [
68] reported an enrichment of CD8+ cells in the corpus callosum in humans and associated CD4/CD8 ratio reduction. This increase in CD8+ T cells was associated with higher expression of the CX3CR1 chemokine receptor and the authors suggested that it could serve in the homing of T cells to the brain parenchyma. The fact that the concentration of T cells is low in the brain could explain our failure to detect an increase in CX3CR1 protein levels in the IVIg-treated mice. Therefore, the fractalkine pathway clearly warrants further investigation as a therapeutic target of IVIg or other compounds in neurodegenerative diseases.
Competing interests
FC and RB have received funding from Grifols (Mississauga, ON, Canada). The funding sources had no involvement in the study design, and in the collection, analysis or interpretation of the data. The remaining authors declare that they have no competing interests.
Authors’ contributions
ISA designed the experiments, performed the animal studies and most of the postmortem analyses, analyzed the data, and wrote the manuscript. IP participated to the animal studies, performed flow cytometry analyses and ELISPOT experiments. CT performed immunoblot analyses. KC performed ELISA quantification. RB provided funding and scientific input on IVIg, and revised the manuscript. FC provided funding, conceived and designed the study, analyzed the data, and wrote the manuscript. All authors read and approved the final version of the manuscript.