When told what to kill, the complement system deftly deals the final blows—helping the immune system purge away pathogens and infected cells. But in the nervous system, debate rages as to whether this army of small blood proteins lives up to its auspicious name. In the decades since Japanese researchers first spotted complement components in Aβ deposits peppering the brains of Alzheimer patients (Ishii and Haga, 1984), a bewildering mix of evidence has accumulated to suggest that in AD, the complement system can both harm and help. In the June 18 issue of the Journal of Neuroscience, work by Cindy Lemere of Brigham and Women's Hospital in Boston and colleagues at Harvard Medical School adds to the evidence supporting the complement system’s beneficial role in AD. In older AD mice overexpressing mutant human amyloid precursor protein (APP) and lacking the central complement component C3, the authors report increased Aβ deposition, significant neurodegeneration, and a shift in microglia activation toward a more alternative (M2) phenotype, compared to APP mice with an intact complement system. Other scientists note that the findings should be interpreted with caution due to key pathological differences in AD patients and mouse models of the disease.
Given the link between aberrant Aβ accumulation and brain inflammation in AD, many in the field had long believed inflammation contributes to the disease process. Brain inflammatory events are mediated in large part by microglia—phagocytic cells that course through the central nervous system in search of damaged neurons and rare pathogens that get past the blood-brain barrier. Activated complement proteins spur microglia into action, causing them to unleash proinflammatory molecules that may harm surrounding nervous tissue in AD and other diseases. In this scenario, the complement system can be seen as an unwanted instigator of neurodegeneration. Support for this view comes from a study by researchers at the University of California, Irvine, showing reduced AD pathology in APP transgenic mice that lack C1q, a component key to triggering the classical complement cascade (Fonseca et al., 2004). More recent work by Ben Barres and colleagues at Stanford University, Palo Alto, California, suggests that the complement cascade helps prune excess synapses during neural development and that abnormal re-expression of key complement proteins C1q and C3 later in life may contribute to the synaptic loss evident in AD (Stevens et al., 2007 and ARF related news story).
On the flip side, a PNAS paper (Wyss-Coray et al., 2002) shows that brain-targeted inhibition of C3 in APP mice resulted in greater Aβ deposition and neurodegeneration. These and other more recent studies (Shaftel et al., 2007 and ARF related news story) have proposed a helpful role for inflammation in AD by demonstrating that microglia can be coaxed to clear amyloid plaques.
In the 2002 PNAS work, C3 activity was blocked by expressing soluble complement receptor-related protein y (sCrry) in the brains of AD transgenic mice expressing human mutant APP. In the current study, the Harvard team recapitulates and extends those findings in AD transgenic mice (expressing human APP with Swedish and Indiana mutations) that completely lack C3 (i.e., APP;C3-/- animals). Led by joint first authors Marcel Maier and Ying Peng, the researchers analyzed mice at 8, 12, and 17 months of age. In younger (8- to 12-month-old) animals, C3 deficiency had no effect on the neuropathological changes induced by the APP transgene. However, quantitative immunohistochemical and biochemical analyses revealed that 17-month-old APP;C3-/- mice had nearly twice the Aβ plaque load in hippocampus and mid-frontal cortex, compared with APP animals. Using Western blotting to detect APP and its cleavage products in mouse brain samples, the researchers showed that the absence of complement C3 had no effect on APP processing in older (12- to 17-month-old) APP;C3-/- animals relative to mice with normal C3.
When they probed the brains of the older animals for neuronal loss, the scientists found a roughly 10 percent decrease in NeuN-positive hippocampal neurons in APP;C3-/- compared with APP mice. This change was statistically significant, the authors found, but minor reductions in MAP2 immunoreactivity (a marker for neuronal dendrites and cell bodies) and synaptophysin levels (a measure of synaptic integrity) in the APP;C3-/- mice were not.
The researchers examined a battery of cell-surface markers and secreted factors to characterize the microglia in the brains of the mice. Compared to APP mice with functional C3, C3-deficient APP mice had increased microglial activation (measured by CD45 immunoreactivity) in both hippocampus and mid-frontal cortex. Upon closer examination, the activated microglia in older APP;C3-/- mice showed a number of characteristics that reflect the more alternative, or M2, microglial/macrophage phenotype—namely, increased brain levels of IL-4 and IL-10, and reduced levels of CD68, F4/80, inducible nitric oxide synthase, and tumor necrosis factor.
“This is a very timely and important paper that supports a more sophisticated view of the role of inflammation in amyloid deposition,” wrote David Morgan of the University of South Florida, Tampa, in an e-mail to ARF (see full comment below). In a review (Morgan et al., 2005), Morgan raised awareness in the AD field that microglia can adopt at least two distinct activation states with different consequences for the surrounding tissue (for examples of protective and detrimental effects of microglia, see ARF related news story). “The real question,” Morgan says, “is the extent to which this will translate to the human circumstance.”
This concern was shared by Barres, who led the 2007 study showing that two complement proteins (C1q and C3) essential for neural development can come back to haunt later in life when their synapse-pruning activity drives neurodegeneration in AD and other diseases. Compared to what happens in the human disease, mouse models of AD display “very little synapse loss” and “relatively little complement activation,” Barres noted in an e-mail to this reporter. “We clearly need much better animal models.”
Pat McGeer of the University of British Columbia, Vancouver, offers a reason for the weaker complement activation in mice compared with AD patients: in AD transgenic mice, Aβ deposits get marked for destruction, but the final stage of the alternative complement pathway (i.e., assembly of the membrane attack complex) occurs to a much lower extent than it does in human AD brains, and can be difficult to detect (Schwab et al., 2004). In human AD, opsonization—the antibody-coating process that labels Aβ plaques for ruin—happens alongside assembly of the membrane attack complex, and studies have implicated these events in nerve damage in and around senile plaques, McGeer explained in an e-mail to ARF (see full comment below).
Though it is hard at this point to determine the clinical relevance of the APP;C3-/- mouse findings, the study’s suggestion that complement plays a beneficial role in AD could affect the interpretation of recent clinical trials showing the failure of anti-inflammatory drugs to protect people from AD (for example, see ADAPT results in ARF related news story). If an anti-inflammatory drug interferes with the activity of C3, it may end up promoting amyloidosis rather than protecting against it, suggested Giulio Pasinetti of Mount Sinai School of Medicine, New York.
Animal models with additional AD pathologies, such as tauopathy mice with pronounced neuronal loss, will be key to sorting out the role of inflammation in AD.—Esther Landhuis