Aβ accumulation is suggested to be a key event leading to development and progression of AD. Production of these toxic proteins is associated with neuronal apoptosis, reactive astrocytes, activated microglia and production of inflammatory molecules. Innate immune response plays a crucial role in the course of AD. However, the exact role of activated microglia is still under debate. These cells can release neurotoxic mediators or can be neuroprotective in promoting the clearance of toxic Aβ from the brain. Several lines of evidence have shown that microglia activated via TLRs can induce beneficial effects on AD progression. We have previously demonstrated that TLR2 is important to delay the cognitive decline in a mouse model of AD . The role of TLR2 in microglial activation was also highlighted in the study from Jana et al. . A small proportion of microglia in APPswe/PS1 mice brain expresses TLR2 mRNA, which is a reliable index of pro-inflammatory signaling in myeloid cells [20, 21]. Other induced innate immune receptors in AD brains include CD14, TLR4, TLR5, TLR7 and TLR9 [9, 20, 25–28]. Phagocytosis of Aβ by activated microglia is significantly increased in presence of ligands that activate TLRs [11, 29–31]. A functional interaction was also demonstrated between fibrillar Aβ1-42 peptides and the innate receptor CD14 and this interaction resulted in microglial activation . Furthermore TLR2, TLR4 and CD14 are involved in fAβ phagocytosis . Activation of TLR/MyD88 pathway may therefore act as a natural defense mechanism in presence of toxic Aβ and restrict disease progression.
Quantification of MyD88 mRNA revealed an upregulation of this gene in response to Aβ as depicted by a significant increase in APPswe/PS1 mice compared to WT. This suggests a potential function of this gene in AD. To study the role of MyD88 in AD, we used MyD88 knockout mice. Unfortunately, APPswe/PS1 mice deficient in MyD88 gene were not viable. This suggests that MyD88 signaling cascade acts as a natural defense mechanism to prevent Aβ toxicity in this transgenic model, even during mouse development. However it will be important in future studies to test the effects of MyD88 deficiency in other mouse models of AD, because the possibility remains that such a lethally depends on the mouse lines used in this study.
Thus, we created APPswe/PS1 mice heterozygous for the MyD88 gene. This model showed a significant decreased expression of the MyD88 transcript. In our previous report, we have shown that APPswe/PS1 mice deficient in TLR2 expression accelerated spatial and contextual memory impairments. Here, we demonstrate that a reduction in MyD88 expression also induced such behavioral impairments. Analysis of hippocampus-based spatial working learning was assessed in T-water maze. In this behavioral test, the submerged platform is the positive reinforcement. Mice were first trained to learn the position of the platform in the maze. Forty-eight hours later their reversal learning abilities were then challenged, as the submerged platform was placed at the opposite side of the maze. APPswe/PS1-MyD88+/- mice were impaired in reversal training, as seen with a higher delay and number of errors made to reach the criterion compared to APPswe/PS1 mice. These data demonstrate the importance of the MyD88-dependent pathway in a context of memory impairment induced by Aβ.
Area occupied by amyloid-β plaques was calculated in the brain of APPswe/PS1-MyD88+/- mice. We found that disease severity demonstrated by behavioral impairment in the T-water maze did not correlate well with plaque load in the brain. Indeed, APPswe/PS1-MyD88+/- mice had fewer plaques at 6 and 9 months compared to APPswe/PS1 mice. Historically, accumulation of Aβ plaques was indicative of AD progression in human and mouse models. However, as mentioned above, accumulating evidence suggests that soluble and small oligomeric forms of Aβ within human brain is more closely associated with disease severity . Soluble Aβ in different protein pools has deleterious effects in the brain and promotes disease progression by inducing changes in synaptic functions, behavioral deficits and promoting neuronal degeneration [32–34]. We were able to show that partial MyD88 gene deletion caused significant increases in soluble Aβ species in extracellular, intracellular and membrane-associated enriched protein pools. Nevertheless, we cannot rule out the possibility that the membrane-associated enriched proteins can also contain insoluble Aβ, because of the 3% SDS concentration used in the buffer. Although sAPPα fragment is known to be a neurotrophic factor when it is secreted in the extracellular space [35, 36], Aβ has been shown to promote microtubule transport impairments in neurons and subsequently sequester sAPPα inside the cell and prevent its secretion . Accordingly, sAPPα fragment levels were higher in intracellular proteins fractions of APPswe/PS1-MyD88+/- mice while there was no significant difference in extracellular-enriched proteins level.
It is tempting to propose that senile plaques may serve as an inert reservoir of Aβ, thus protecting neurons from soluble oligomeric forms of Aβ . These data regarding the plaque load are in agreement with our previous study in the APPswe/PS1-TLR2-/- mice that had less Aβ plaques than their control APPswe/PS1 littermates . As in APPswe/PS1-MyD88+/- mice, memory capacities did not correlate with plaque load in the brain of APPswe/PS1-TLR2-/- mice. TLRs and MyD88 signaling may have a more direct role on the clearance of soluble oligomeric Aβ.
Analyze of blood monocyte subsets revealed changes in APPswe/PS1 mice in a context of partial MyD88 deficiency. These mice exhibited significant relative decrease in Gr1+ and increase in Gr1- monocytes. Globally, there was a 30% reduction in the ratio of Gr1+/Gr1- monocytes in blood of APPswe/PS1-MyD88+/- compared to APPswe/PS1 mice. As explained earlier, inflammatory monocytes (CX3CR1
LoCCR2+Gr1+) and resident monocytes (CX3CR1
HiCCR2-Gr1-) seem to play opposite roles. Inflammatory monocytes are recruited quickly to inflammatory sites and differentiate into tissue specific macrophages and dendritic cells, whereas resident monocytes infiltrate tissues in an inflammation-independent fashion . This process has been described in different tissues including brain in mouse models of multiple sclerosis  and axonal injury . Moreover, inflammatory monocytes have also been shown to be the main precursors for microglial cells under inflammatory conditions of the CNS [41, 42]. MCP-1 (CCL2) is a major chemiokine involved in the recruitment of monocytes in AD patients [43, 44] and in mouse models of this pathology [45, 46]. Since interaction between MCP-1 and its binding CCR2 receptor stimulates the mobilization of bone marrow-derived inflammatory monocytes , a decrease production of this chemokine may be the mechanism underlying the 30% reduction in the ratio of Gr1+/Gr1- monocytes in blood of APPswe/PS1-MyD88+/-. In this regard, MyD88 signaling is essential for the transcriptional activation of MCP-1 in response to bacterial ligands [15, 48], brain injury  and endogenously produced toxic proteins . Moreover, MyD88-deficient mice have impaired monopoiesis during bacterial infection resulting in significant reduction in blood, splenic and bone marrow progenitors of inflammatory monocytes . This suggests an important role for MyD88-mediated monocyte homeostasis under inflammatory conditions.
On the other hand, it seems that inflammatory CX3CR1
LoCCR2+Gr1+ monocytes differentiate preferentially in M1-like macrophages while resident CX3CR1
HiCCR2-Gr1- monocytes are likely to become M2-like macrophages [50, 51]. M1 macrophages are known to have high phagocytic, proteolytic and inflammatory functions while M2 is suggested to be an anti-inflammatory subset, which is known to promote functions such as tissue remodeling, angiogenesis and matrix deposition . M1-like macrophages would therefore be more efficient to clear Aβ from the brain in a context of chronic inflammation induced by accumulation of cerebral Aβ. Taking that into consideration, a decreased Gr1+/Gr1- ratio in the pool of available monocytes of APPswe/PS1-MyD88+/- mice may be associated with less M1-like macrophages and more M2-like cells and explain the higher levels of soluble Aβ in brain of APPswe/PS1-MyD88+/- mice. However, such a possible mechanism has to be fully investigated.
Low IL-1β mRNA levels in the brain of APPswe/PS1-MyD88+/- mice compared to APPswe/PS1 provide evidence that MyD88 is a critical molecule for the regulation of such inflammatory pathway in response to Aβ accumulation. This cytokine is crucial to orchestrate the inflammatory response by microglia  and a sustained production of IL-1β was reported to be beneficial by reducing Aβ pathology in a mouse model of AD . We consequently believe that the innate immune response by microglia is compromised in a context of MyD88 deficiency, which may prevent their adequate activation to fight against and clear Aβ peptides. These data converge into a primary role of MyD88-signaling pathway to act as a natural protective mechanism in presence of toxic Aβ.
In summary, we show here that MyD88 plays a significant role in the evolution of AD. Partial MyD88 deficiency worsens cognitive deficit in APPswe/PS1 mice, as well as increases brain soluble oligomeric Aβ. Moreover, APPswe/PS1 mice deficient in MyD88 have reduced IL-1β gene expression and altered blood monocyte homeostasis leading to a decrease in the inflammatory population. Altogether, these results clearly demonstrate the crucial role of functional MyD88 signaling and support the importance of TLRs to prevent or delay AD pathology.