Scores of studies have made the case for microglial involvement in AD, but it remains controversial under which conditions these immune cells intensify or curb pathogenesis (see ARF related news story on El Khoury et al., 2007 and Fan et al., 2007; ARF related news story on Ryu et al., 2009 and Koenigsknecht-Talboo et al., 2008). Much of the evidence suggesting microglia are beneficial comes from observations that the activated phagocytes migrate toward and chew up Aβ plaques in the brains of AD transgenic mice. Once spurred into action, though, microglia can also become troublemakers, unleashing proinflammatory cytokines and chemokines that harm surrounding cells.

In the current study, first authors Martin Fuhrmann, Tobias Bittner, and colleagues homed in on the mischievous side of microglia. They used two-photon imaging to look in vivo at these cells and their interactions with neighboring neurons in the brains of coauthor Frank LaFerla’s 3xTg mice, which overexpress pathogenic forms of presenilin-1 (M146V), amyloid precursor protein (Swedish), and tau (P301L). To make the neurons and microglia light up under the microscope, the researchers crossed the 3xTg mice with two additional transgenic lines—one that expresses yellow fluorescent protein (YFP) in subsets of layer III and V cortical neurons (Feng et al., 2000), and another carrying a green fluorescent protein (GFP) knock-in at the CX3CR1 locus (Jung et al., 2000). Inserted at this position, GFP not only labels microglia and other myeloid cells, but also interferes with neuron-microglia crosstalk in the new quintuple-transgenic model (5xTg). CX3CR1 is found on microglia and intercepts signals from CX3CL1 (aka fractalkine), a cytokine-like molecule expressed by neurons.

Two-photon imaging of four- to six-month-old YFP-expressing 3xTg mice revealed neuronal loss over two to four weeks of observation. The neurodegeneration was moderate—1.8 percent of YFP-positive layer III neurons lost over 28 days—but the fact it was seen at all is news in 3xTg mice, which develop plaques and tangles, but until this study never had detectable neuron loss. “The data are quite striking,” said Terrence Town, University of California, Los Angeles, who was not involved in the work. “The time-lapse images very clearly show that neurons are dying in the 3xTg mice, but in the fractalkine receptor knockout 3xTg mice, the neurons are persisting.” Moreover, in 3xTg mice, microglia swarmed in and moved more quickly around neurons that were about to be eliminated, whereas in CX3CR1-deficient 3xTg animals, microglia in similarly close proximity to neurons (i.e., within 100 micrometers) were comparatively fewer and slower. To Town, the data suggest that fractalkine/CX3CL1 signaling somehow calls microglia to neurons that are destined for death. “When you stop that pathway, you keep microglia from homing to those marked neurons,” he said.

What remains unclear, though, is whether microglia initiate the neuronal death cascade or simply serve as downstream mediators. Herms finds this a “most intriguing question,” but acknowledges it is hard to dissect. “In fact, we think that microglia do both—facilitate nerve cell death from a certain point of no return and subsequently remove the cell,” he wrote.

On a conceptual level (i.e., “Are microglia good or bad for AD?”), this idea departs from a recent report suggesting that microglia are irrelevant to formation and clearance of amyloid plaques in AD mice. In that study, co-led by Mathias Jucker, University of Tübingen, and Frank Heppner, Charite-Universitaetsmedizin, Berlin, researchers used a suicide gene approach to kill microglia for two to four weeks in two AD models (APP/PS1 and APP23). They found no change in plaque size or number (Grathwohl et al., 2009 and ARF related news story).

Considered alongside his own findings, Herms sees the two studies as complementary, even though they address different aspects of microglial involvement in AD. “If microglia have no role in plaque clearance, as suggested by Grathwohl et al., our study gives a hint at what their role might be instead—assisting neuron loss in AD,” he proposed. Furthermore, Herms and colleagues did measure insoluble Aβ40 levels in four- to six-month-old CX3CR1-deficient 3xTg mice and found no appreciable difference compared with 3xTg. This could be interpreted as saying the CX3CR1 receptor does not play a role in Aβ clearance or maintenance—or, more speculatively, that microglia have no part in these processes, Herms suggested.

A possible sticking point with the Aβ analyses is the fact that they were done in mice lacking hyperphosphorylated tau or extracellular amyloid plaques. At that age (four to six months), it is too early to determine whether CX3CR1 deficiency affects microglial control of amyloid deposition, noted Pritam Das of Mayo Clinic, Jacksonville, Florida, in an e-mail to ARF. Herms said his team is analyzing plaque load with respect to CX3CR1 loss in aged 5xTg mice but cannot comment on this issue at present.

Meanwhile, CX3CR1’s role in AD is turning out to be a complex story. At a Keystone symposium called Alzheimer’s Disease Beyond Aβ, held 10-15 January 2010 at Copper Mountain, Colorado, parallel studies led by Joseph El Khoury of Harvard Medical School and by Richard Ransohoff and Bruce Lamb of the Cleveland Clinic, Ohio, showed that loss of CX3CR1 led to reduced amyloid deposition in APP/PS1 mice. In another study from Lamb’s lab, CX3CR1 knockouts crossed with humanized tau (hTau) transgenic mice showed worse tau pathology compared with CX3CR1-sufficient hTau mice (see ARF related conference story). “It appears that the altered microglial reaction in CX3CR1 knockout mice is a double-edged sword, producing better amyloid phagocytosis and worse tau pathology,” Ransohoff wrote in an e-mail to ARF (see full comment below and Ransohoff, 2009).

One aspect of the microglial effects in the APP/PS1 studies—namely, migration—did seem to jibe with Herms’s data, Town noted. El Khoury reported at Keystone that CX3CR1 knockout APP/PS1 mice had fewer mononuclear phagocytes associating with Aβ plaques. In Herms’s study, lack of CX3CR1 also slowed migration of phagocytes—but toward neurons.—Esther Landhuis.

Reference:
Fuhrmann M, Bittner T, Jung CK, Burgold S, Page RM, Mitteregger G, Haass C, LaFerla FM, Kretzschmar H, Herms J. Microglial Cx3cr1 knockout prevents neuron loss in a mouse model of Alzheimer’s disease. Nature Neuroscience. 21 March 2010. Abstract