The findings may come as a surprise, given indications that microglia can be recruited to tackle Aβ deposits (see Boissonneault et al., 2009; ARF related news story on El Khoury et al., 2007), but it gets at the fundamental question of whether resident microglia in the brain have any effect on plaque dynamics—a heavily debated issue, according to Terrence Town, University of California, Los Angeles. “This is an elegant study. I think that it definitively tells us that microglia are not important in clearance or formation of amyloid plaques,” Town told ARF, though he emphasized also that the work does not rule out the possibility that peripheral macrophages might be involved in plaque turnover or that brain microglia could be spurred into action. “While CNS resident microglia may not be able to influence plaque progression in the absence of further manipulation, that does not mean that if we devise a therapeutic or genetic strategy to manipulate these cells that that would not have an effect on plaques,” he said.

To ablate microglia from the brain, Jucker, Heppner, and colleagues made use of mice engineered to produce the “suicide” thymidine kinase from herpes simplex virus (HSVTK). The kinase converts some antiviral drugs, such as ganciclovir (GCV), into toxic nucleotide analogs that insinuate into growing DNA and kill dividing cells. Joint first authors Stefan Grathwohl and Roland Kälin crossed HSVTK mice with two AD mouse models—the APP/PS1 mouse, which has aggressive plaque pathology, and the APP23 mouse, which develops plaques more slowly. The HSVTK was driven by the CD11b promoter. This promoter restricts kinase expression to cells of the myeloid lineage including brain microglia, but the researchers generated chimeric mice carrying congenic wild-type bone marrow to keep peripheral myeloid cells alive. This ensured that only resident brain microglia were ablated in the crosses.

Grathwohl and colleagues used two different GCV treatments to ablate brain microglia in the APP/PS1-TK mice. Given orally at five months to the chimeric mice, GCV led to a 30 percent reduction in microglia in the neocortex. The morphology or number of Aβ plaques did not change as a consequence. Using a micropump to deliver GCV directly into brain ventricles (for the ventricle infusions, APP/PS1-TK mice were not bone marrow chimeric), the researchers achieved a 90 percent reduction in brain microglia over two to four weeks. Again, they found no change in plaque dynamics. The researchers also adjusted the timing of the microglial decimation to before or after plaques emerged. “It didn’t matter whether we depleted the microglia first or whether we had amyloid depositing mice with plaques and then depleted the microglia; the plaques just didn’t care whether they had microglia or not,” said Jucker.

It also didn’t matter which mouse model the researchers used. With the less aggressive APP23 crosses, the researchers also achieved about 95 percent ablation of brain microglia after pumping GCV into brain ventricles of 17- or 24-month-old mice. They found no change in congophilic-positive plaque load or the number and morphology of amyloid-associated dystrophic neurites. The latter observation argues against activated microglia being involved in neuronal damage. “This was a very carefully designed study. They used multiple approaches, two different mouse models of Alzheimer’s, and I think the results are compelling,” said Town.

One major limitation to the study is the length of observation. Because GCV eventually becomes toxic, the researchers were only able to follow its effects for up to four weeks, leaving open the question of whether a longer microglial depletion would have some effect on plaques. “That’s a fair criticism,” said Jucker but he added that studies of microglial infiltration into the brain have seen effects within two weeks (see ARF related news story on Simard et al., 2006) and that plaques can pop up literally overnight (see ARF related news story on Meyer-Luehmann et al., 2008), “so I have trouble to believe that after, say, 16 weeks we would see anything different,” he said, though he is planning to investigate longer times. Town thinks that the researchers tried their best to see a positive effect. “The time points they chose to administer GCV tended to be during the initiation phase of Aβ deposition, so you would expect that if any phase of amyloidosis would be sensitive to this kind of manipulation, it would be then,” he said.

“This is a very exciting study that gives us a lot of thought for what is really going on in the brain with amyloid, and I think we will be busy discussing this for a long time,” suggested Tony Wyss-Coray, Stanford University, California, who also wondered if resident microglia can be induced to attack amyloid. “The study shows that if you don’t do anything, then microglia are not involved in plaque turnover. But that doesn’t exclude that they could be clearing amyloid in human brains,” he suggested. Jucker said he doubts that happens in the absence of some other initiating event. “This goes back almost 20 years to the studies of Henryk Wisniewski,” said Jucker. Wisniewski found Aβ fibrils in brain microglia of AD patients only if the patients had also suffered a stroke (see Wisniewski et al., 1991), suggesting that some additional event was necessary to goad microglia into gobbling up Aβ.

What that event might be in humans is unclear, though inflammatory responses aid microglia clearance of Aβ in transgenic mice (see, e.g., Wyss-Coray et al., 2001). And just recently, Pritam Das and colleagues at the Mayo Clinic, Jacksonville, Florida, reported that they can induce massive gliosis and suppression of Aβ deposition by administering the pro-inflammatory cytokine interleukin-6 (IL-6) to transgenic animals (see Chakrabarty et al., 2009). “We are excited about that because it shows that inflammation does not promote more amyloid or more Aβ, which was the hypothesis for a long time,” said Das. Immunotherapy is currently an active therapeutic strategy and that might stimulate microglia, as well.

In vitro, too, microglia are well known for phagocytosing Aβ. Gary Landreth’s group at Case Western Reserve University in Cleveland, Ohio, have identified specific receptors that mediate microglial responses to Aβ (see ARF related news story on Reed-Geaghan et al., 2009). “That would, of course, create hope that one could try to find the right receptor or target to activate these cells. If it is true that they are not doing anything in the brain, but they do in cell culture, they are targets I would want to pursue,” suggested Wyss-Coray.

If it is true that microglia are normally oblivious to amyloid plaques, then why do they surround them? “My own theory is that they are making a glial scar,” suggested Das. “The microglia surround the plaques, and astrocytes come on top and cordon off the area, essentially protecting the rest of the brain. If the microglia cannot remove plaques, it makes sense that they would try to protect the rest of the brain from damage.” Though ablation of microglia did not result in neurodegeneration, noted Jucker, he did agree with Das’s speculation.

Perhaps the next step is to question the role of the astrocytes, suggested Wyss-Coray. “A number of people, including us, have shown that astrocytes can degrade amyloid.” (See Wyss-Coray et al., 2003.) However, given their specialized role in modulating neurotransmission, ablating astrocytes without having catastrophic effects on the mice might be a tall order.—Tom Fagan.

Reference:
Grathwohl SA, Kalin RE, Bolmont T, Prokop S, Winkelmann G, Kaeser SA, Odenthal J, Radde R, Eldh T, Gandy S, Aguzzi A, Staufenbiel M, Mathews PM, Wolburg H, Heppner FL, Jucker M. Formation and maintenance of Alzheimer's disease beta-amyloid plaques in the absence of microglia. Nature Neuroscience 2008 October 18. Abstract