Inflammation’s Dark Side
Taking the broad view, Piet Eikelenboom at VU University Medical Center, Amsterdam, The Netherlands, discussed the close links among aging, inflammation, and sporadic AD. Eikelenboom’s group was the first to describe inflammation and the complement system in Alzheimer’s disease in the 1980s (see ARF related news story), and he sees a resurgence of interest around the topic now. Eikelenboom argued that inflammation precedes AD. Eikelenboom pointed out that the diffuse amyloid plaques commonly found in non-demented elderly people contain numerous inflammation-related molecules, such as activated complement proteins, ApoE, clusterin, and α1-antichymotrypsin (see Eikelenboom et al., 2006; Eikelenboom et al., 2011). Most of these proteins have since been genetically tied to AD (see AlzGene top results), and scientists increasingly voice a suspicion that a person’s particular set of risk alleles might exert their effect in young and middle adult life, before the actual disease process begins. People whose parents had AD are more susceptible to inflammation in midlife, producing greater quantities of pro-inflammatory cytokines when their systems are challenged (see van Exel et al., 2009). This suggests that genetic risk factors for AD go hand-in-hand with inflammation-prone systems.

As AD develops, fibrillar Aβ deposits accumulate, appearing around Braak Stage 3 or 4 and before any clinical symptoms. Activated microglia and astrocytes surround these plaques and pump out cytokines that promote inflammation, Eikelenboom said. Thus, fibrillar plaques may play a key role in triggering an inflammatory reaction, an idea discussed by several presenters. Also, several studies have found that systemic infections worsen AD symptoms and accelerate degeneration and cognitive decline (see Perry et al., 2007; Holmes and Cotterell, 2009; Holmes et al., 2009). As Eikelenboom described it, infections can put “inflammatory pressure on the brain.”

Eikelenboom suggested that the process of aging itself may put the body into a low-grade inflammatory state. This would prime microglia and allow a late-life infection or injury to activate these cells and precipitate rapid presentation of serious disease. For example, something as simple as a hip fracture can cause delirium in elderly people due to a neuroinflammatory response involving acute-phase proteins, Eikelenboom said (see Eikelenboom et al., 2002). The general scheme is not limited to AD, Eikelenboom believes. An elderly person at risk for type 2 diabetes can rapidly develop diabetes after a late-life infection or injury, while somebody at risk for AD would come down with this disease after a late-life challenge.

Activated microglia also damage the cholinergic system, Eikelenboom noted. This system normally inhibits the inflammatory activation of microglia (see Hwang et al., 2010). Microglia thus escape cholinergic control and ramp up further, causing more neurodegeneration and more damage to cholinergic neurons (see van Gool et al., 2010). In this model, then, inflammatory processes are not just a response to neurodegeneration, but play a key role in driving late-onset AD.

Pathology indicates, too, that inflammation is not limited to AD. Neuroinflammation is an early feature of almost all neurodegenerative diseases, said Annemieke Rozemuller, also at VU University. She presented postmortem data from studies comparing patients with AD to others who had various prion diseases such as Creutzfeldt Jakob and Gerstmann Straussler Scheinker's disease. In both AD and prion diseases, she said, amyloid plaques in brain and blood vessels contain activated complement proteins and α1-antichymotrypsin. Activated microglia cluster around prion plaques, and Rozemuller saw deposits of tau and ubiquitin in prion-riddled brains. Rozemuller concluded that inflammatory changes are related to the process of amyloid deposition, regardless of the type of amyloid.

Good Inflammation, Bad Inflammation—Ways to Tell Them Apart
Despite all the data showing the dark side of inflammation, it can be protective, too. Stimulated microglia can devour Aβ, helping keep plaques in check (see ARF related news story and ARF news story). Researchers have just begun to tease apart what factors make microglia harmful or helpful, and there is not a clear consensus yet. Researchers recently discovered that knocking out the microglial-expressed immune phosphatase CD45 caused microglia to secrete more pro-inflammatory cytokines and fail to chew up Aβ, and mice lost more neurons (see ARF related news story).

At AD/PD, Todd Golde at the University of Florida, Gainesville, described data that go against conventional wisdom, showing that pro-inflammatory cytokines can promote phagocytosis and reduce plaques. His team uses viral vectors to express various immune factors in the brains of TgCRND8 AD model mice, creating “somatic brain transgenic” animals. This method is quicker and cheaper than breeding traditional transgenic mice, Golde noted. Injection at different ages produced different expression patterns. Injection into newborns yielded the broadest expression, throughout midbrain and forebrain. The scientists first tested the pro-inflammatory cytokine IL-6, which is elevated in AD brains. They injected newborns and examined brains at five months. Instead of seeing the expected worsening of pathology, they saw less Aβ deposition (see Chakrabarty et al., 2010). They also saw widespread activation of astrocytes and microglia, more microglia associated with plaques, and more phagocytosis of Aβ. Overall levels of APP or Aβ were unchanged. The researchers saw similar effects with the pro-inflammatory cytokines IFNγ and TNFα (see Chakrabarty et al., 2010; Chakrabarty et al., 2011). In contrast, preliminary studies show the anti-inflammatory cytokines IL-10 and IL-4 both enhanced Aβ deposition, resulting in more and/or larger plaques. Although the results suggest that some inflammation can help limit Aβ deposits, Golde noted that it is unclear if this actually protects the brain. Inflammation could be causing problems, too, and more research is needed to determine if the animals fare better overall. Nor is it clear how microglial subtypes contribute to this picture, Golde said.

Golde described another mouse model to illustrate the point that the effects of a particular cytokine depend on where and when it acts. When Golde’s team expressed IFNγ primarily in mouse choroid, the mice developed calcium deposits in the basal ganglia reminiscent of human Fahr’s syndrome. This syndrome includes Parkinson’s-like symptoms, and, indeed, the mice showed parkinsonism as well as profound progressive degeneration of the nigrostriatal tract. Being presented with a parkinsonian model after merely expressing IFNγ in the choroid was totally unexpected, Golde said. Since IFNγ levels rise in response to viral infections, this may explain why Fahr’s syndrome can develop after viral infection of the brain, Golde noted. IFNγ has been reported to be elevated in Parkinson’s disease (see, e.g., Mount et al., 2007), suggesting that this mechanism could be important in general PD as well, Golde said. He added that this mouse model places specific types of inflammation and inflammatory mediators on the pathway toward neurodegeneration (Chakrabarty et al., in press). The effect of IFNγ contrasts with that of IL-6, which does not drive neurodegeneration despite causing massive gliosis.

To determine whether inflammation is good or bad in particular models, scientists may have to dissect the specific signaling pathways involved. Saskia van der Vies at VU University described one. Her team discovered that interleukin-1 receptor-associated kinase 4 (IRAK-4) and its substrate, the kinase IRAK-1, were elevated in late-stage AD brains. These kinases are expressed in microglia and astrocytes, van der Vies said, where they mediate signaling through the interleukin-1 (IL-1) receptor and through toll-like receptors. Downstream, the kinases activate the transcription factor NF-κB, which itself increases the production of other cytokines including monocyte chemotactic protein-1 (MCP-1). MCP-1 goes up with age in people (see ARF related news story) and is elevated in AD (see Galimberti et al., 2006). It has been fingered as a culprit in chronic inflammation (see Sokolova et al., 2008). Van der Vies’s team stimulated microglia and astrocytes while inhibiting IRAK-4 activity, and found that the glia were unable to make MCP-1 but continued to gobble up Aβ. This indicates that IRAK-4 activity is essential for MCP-1 production in both astrocytes and microglia, van der Vies said. Since IRAK-4 inhibition stops production of a harmful cytokine while leaving presumably helpful phagocytosis intact, inhibiting this kinase could be promising therapeutically, van der Vies suggested.

Makoto Higuchi at the National Institute of Radiological Sciences, Chiba, Japan, presented another way to separate good and bad microglial functions. He also points an accusing finger at MCP-1. The harmful subset of microglia express the 18-kDa translocator protein (TSPO; also known as peripheral benzodiazepine receptor), Higuchi said. Conveniently for scientists, TSPO radioligands exist to visualize this protein in the living brain using positron emission tomography (PET). This method reveals TSPO-positive microglia almost exclusively around plaques in the AD brain in both mice and humans (see Yasuno et al., 2008).

To characterize the effects of this microglial subtype, Higuchi’s team prepared a microglial cell line clone that expressed high levels of TSPO, as well as one that expressed very little TSPO. When the researchers implanted the low TSPO clone into the brains of APP transgenic mice, they saw less Aβ and higher levels of brain-derived neurotrophic factor, whereas injection of high TSPO-expressors seemed to prevent neighboring glia from mopping up Aβ and to worsen plaques.

To investigate what makes the TSPO-expressors harmful, Higuchi’s team analyzed their cytokine profile and found that they make high levels of MCP-1. Higuchi noted that the enzyme glutaminyl cyclase (QC) converts MCP-1 to a more stable form (see ARF related news story and ARF AD/PD story). The same enzyme creates a stable, pyroglutamate form of Aβ that has been proposed to initiate the formation of Aβ plaques (see, e.g., ARF related SfN story). This makes QC a good therapeutic target, Higuchi noted, since QC seems to promote both amyloid deposition and harmful inflammation.

In mouse experiments, anti-Aβ therapy reduces amyloid deposits, but also increases TSPO, Higuchi said. He concluded that antibodies stimulate both protective microglia that phagocytose Aβ as well as harmful microglia that secrete MCP-1. Since MCP-1 seems to mediate the harmful effects of TSPO-expressing microglia, Higuchi’s team combined anti-Aβ therapy with anti-MCP-1 antibodies in APP transgenic mice. The combination provided long-lasting removal of Aβ without any increase in TSPO-positive microglia, Higuchi said.

The Flip Side: Microglia and Tau
If you think you are beginning to understand microglial subtypes and their respective effects on pathology, consider experiments looking at the effect of microglia on tau. In the PS19 mutant human tau-overexpressing mice, TSPO-positive microgliosis occurs as well, Higuchi said. MicroPET TSPO signals were actually higher in tauopathy mice than in amyloid mice, and they preceded neuronal loss, Higuchi said, suggesting that a TSPO signal could eventually serve as a “toxicity alarm” in AD brains (see Maeda et al., 2011).

David Morgan at the University of South Florida, Tampa, also presented evidence that inflammation interacts quite differently with tau tangles than it does with amyloid plaques. It’s known that in APP transgenic mice, stimulation of microglia with the generic inflammatory agent lipopolysaccharide (LPS) activates phagocytosis and reduces amyloid. More recently, Morgan’s team looked at what happens in the Tg4510 tauopathy model mouse. This strain seems to model a slightly later stage of AD, with more neurodegeneration, brain shrinkage, and poor memory, than do amyloid models, which reflect early, preclinical AD (see Dickey et al., 2009). Tauopathy mice normally have less microglial activation than do amyloid mice, and it shows up later, Morgan said in Barcelona, but in the Tg4510 mice, microglial stimulation with LPS was harmful, increasing the amount of phosphorylated tau (see Lee et al., 2010). LPS injections dramatically increased the numbers of CD45-positive, activated microglia, which is the subtype that has beneficial effects in amyloid mice (see ARF related news story).

Since many scientists believe that tau mediates the harmful effects of Aβ (see, e.g., ARF related news story and ARF news story), a treatment that decreases one while increasing the other might not help the patient. “LPS injection enhances degradation of amyloid deposits in APP mouse brain but exacerbates pathology in tau-depositing mice. That makes it hard to decide what you want to do therapeutically with microglia,” Morgan said.—Madolyn Bowman Rogers.