12 November 2012. Neurons are a minority of cells in the brain, with glia, including microglia and astrocytes, dominating. It may seem due time, then, that that majority is drawing the attention of Alzheimer’s and other neurodegenerative disease researchers. The topic stood out as a theme at the Society for Neuroscience 42nd annual meeting, held 13-17 October in New Orleans. Scientists are slowly beginning to understand the interplay between glial-neuronal crosstalk and AD pathology, though much more work needs to be done.
Glia and Inflammation
The interplay between glia and Aβ is one aspect of AD pathology that continues to mystify. Although researchers know that glia mop up Aβ deposits, and that the peptide can render the cells proinflammatory and potentially toxic (see ARF related news story), the signals involved are poorly understood. Toll-like immune receptors (TLRs) on microglia seem to have a hand in this (see ARF related news story on Reed-Geaghan et al., 2009), but knocking out individual receptors has not painted a clear picture. Gary Landreth at Case Western Reserve University, Cleveland, Ohio, explained that TLRs are part of a very large molecular complex that fails to form properly when the receptors are absent, confounding interpretation of the knockout phenotypes.
Landreth’s group took a different approach. Instead of targeting TLRs, they went for interleukin receptor-associated kinase 4 (IRAK4), a member of a family of kinases that mediate signaling through the TLR complexes. As presented on a poster, first author Brent Cameron and colleagues replaced the normal mouse gene for IRAK4 with one that codes for an inactive form. “While knocking out TLRs and co-receptors ends up disrupting the complex, using this knock-in strategy we allow the complexes to form naturally, but they cannot signal,” Landreth told Alzforum.
Cameron knocked the inactive IRAK4 into APP/PS1-21 mice obtained from Mathias Jucker’s lab at the University of Tubingen, Germany. These mice aggressively deposit Aβ beginning at about six weeks of age. The researchers found that Aβ pathology in IRAK4-negative APPPS1 mice and their kinase-competent littermates seemed the same at four months of age. By eight months, however, the IRAK4 mutants accumulated 40 and 30 percent less soluble and insoluble Aβ42, respectively, in the brain. They also had fewer and smaller amyloid plaques (see Cameron et al., 2012).
How does loss of the kinase signal translate into reduced Aβ? The researchers looked to inflammatory responses that might exacerbate Aβ production. But instead of finding a reduction in proinflammatory markers, they found that microglia from IRAK4-negative eight-month-old transgenics expressed higher amounts of both pro- and anti-inflammatory markers compared to APP/PS1 controls. “We interpret this to mean that pro-inflammatory microglia may be associated with plaques, but a few millimeters away there are glia that are not involved,” said Landreth. That suggests that loss of IRAK4 renders microglia generally more quiescent, while still allowing them to respond to perturbances in their environment, such as plaques.
Support for this idea came from analysis of interferon response factors (IRFs). These transcription factors are master regulators of toll-like receptor responses. Recently, scientists discovered that IRF4 and IRF5 have opposing actions on microglia. IRF4 promotes an anti-inflammatory (M2) state (see Ahyi et al., 2009), while IRF5 induces pro-inflammatory (M1) responses and suppresses M2 responses (Krausgruber et al., 2011). Cameron and colleagues found that, beginning at four months, the IRAK4-deficient animals began to switch to an M2 state. They made 75 percent less IRF5 at four months, and by eight months, 40 percent more IRF4 and 55 percent less IRF5. While the data could be interpreted to mean that loss of toll-like receptor signaling favors the M2 state, Landreth admitted that the story is complex. The challenge in determining the role of glia in pathology lies in understanding their heterogeneity, he suggested. “Microglia only care about what is happening in their vicinity,” he said. “Until we get the right tools, such as better histochemistry or single-cell analysis, we will not be able to understand that complexity.”
Kiran Bhaskar, now at the University of New Mexico, Albuquerque, addressed a different type of glial signaling. When he worked with Bruce Lamb at the Cleveland Clinic in Ohio, Bhaskar reported that fractalkine, a peptide released by neurons, seems to quench glial inflammatory feedback on those same neurons. Fractalkine activates the glial CX3CR1 receptor. When the researchers crossed CX3CR1 knockout mice with animals expressing human mutant tau, hyperphosphorylation and aggregation of tau ran rampant in response to an inflammatory stimulus. Neurodegeneration rose as well (see ARF related news story). The findings suggested that fractalkine signaling keeps tau pathology in check. On the flipside, Bhaskar predicted that tau pathology would exacerbate inflammatory glial signaling.
In New Orleans, he showed what happens when he took tau out of the picture.
Knocking out tau protected primary neurons against activated microglia. Lipopolysaccharide (LPS), a potent inducer of glial inflammation, triggered cell death in mutant human tau-positive neurons co-cultured with CX3CR1-negative microglia, but tau-negative neurons were mostly unaffected. Neurons lacking tau showed little increase in the apoptosis markers caspase 3 and annexin V in comparison to control neurons. Similarly, in two-month-old tau-negative mice, only half as many cells expressed caspase 3 in the dentate gyrus as in tau-positive animals. Furthermore, LPS barely activated microglia in tau knockout mice, as judged by staining with the glial marker Iba1. The cellular protection afforded by knocking out tau seemed to translate into behavioral gains. LPS-treated CX3CR1 knockout mice spent more time in the open field than wild-type animals, indicating a lack of inhibition, but on removal of tau the animals mostly avoided the open, just like wild-type mice. All told, the work suggests that tau supports inflammatory crosstalk between neurons and glia.
Bhaskar is unsure why knocking out neuronal tau prevented microglial activation, especially since there should be no fractalkine signaling between the neurons and the CX3CR1-negative glia. “At this point, I think communication between neurons and microglia is independent of CX3CL1-CX3CR1 signaling, but may be dependent upon other possible signaling pathways,” he told Alzforum (e.g., see Ransohoff and Cardona, 2010). Bhaskar also said that other cell markers of glial activation depended on neuronal tau. He plans to explore other neuron-glial signaling pathways that might depend on tau.
Astrocytes and Neurotransmission
Astrocytes are the other major glia in the brain. They, too, respond to Aβ. At SfN, researchers led by Rheinallt Parri at Aston University, Birmingham, U.K., and Kelly Dineley at the University of Texas, Galveston, presented data showing that Aβ acts on astrocytic acetylcholine receptors to boost release of glutamate, which then signals to neurons. First, Parri showed that acute hippocampal slices from young, nine- to 14-day postnatal Tg2576 mice generated more spontaneous astrocyte calcium spikes than slices from wild-type animals. Adding Aβ1-42 (300 pM) to wild-type slices increased astrocytic activity, and that boost was blocked by an α7-nicotinic acetylcholine receptor (α7-nAChR) antagonist. On the other hand, adding Aβ1-42 did not further increase activity in transgenic slices, indicating that the receptors were already saturated. Parri believes that astrocyte activity releases glutamate, which can then activate adjacent neurons. Patch-clamp recordings from hippocampal neurons revealed slow inward currents (SICs) consistent with glial glutamate transmission to neuronal NMDA receptors. Inward currents in Tg2576 slices persisted for longer than in wild-type tissue and could be induced by an α7-nAChR agonist. Parri claims that increased astrocytic activity in the Tg2576 animals leads to enhanced gliotransmission. Since others have proposed that Aβ production occurs with synaptic activity, these results suggest a potential role for endogenous Aβ to modulate glutamate neurotransmission via interaction with astrocytic α7-nAChRs, suggested Dineley. “The take-home message from this work is that astrocyte gliotransmission becomes dysfunctional early, before any effect on cognitive function,” she said.
Researchers from South Korea looked to a different transmitter in astrocytes—γ-aminobutyric acid. On her poster, Seonmi Jo, from C. Justin Lee’s lab at the Korea Institute for Science and Technology, Seoul, reported finding high levels of GABA in reactive astrocytes in APPswe/PS1 transgenic animals. Jo used a GABA-specific antibody to detect a fivefold increase above astrocytic levels in wild-type mice. Interestingly, astrocyte glutamic acid decarboxylase and GABA transporters appeared normal in the transgenic mice, apparently ruling out uptake or synthesis from glutamic acid as explanations for GABA increases. Instead, the researchers found elevated putrescine and monoamine oxidase, which can metabolize the polyamine to GABA. Closing the loop, they found that the astrocytes from the transgenic animals lacked the transaminase that normally metabolizes GABA.
The data suggest a buildup of GABA in astrocytes in the APP/PS1 mice. How might this unconventional finding fit with AD pathology? Using in-situ microdialysis, Jo and colleagues detected 60 percent more GABA in the dentate gyrus of the transgenics, as well as tonic inhibition by GABA. A GABA receptor antagonist boosted neural transmission in the transgenic animals but not in wild-type mice. The researchers were unable to show that the elevated GABA impairs learning and memory in these transgenic mice. Even so, they suggested that astrocytic GABA might be worth exploring as a therapeutic target in AD.—Tom Fagan.