“I think it is interesting that out of a large variety of molecules that could have been involved [in astrogliosis], they were able to identify neutral sphingomyelinase, and they were able to show that the interaction between glia and neurons, instead of being protective, can be just the opposite,” said Tobias Hartmann in an interview with ARF. Hartmann, from the University of Saarland, Homburg, Germany, previously reported that soluble Aβ42 can activate neutral sphingomyelinase in neurons (see ARF related news story on Grimm et al. 2005). That was more of a physiological response, Hartmann told ARF, since it occurs at very low, 1-nanomolar concentrations of Aβ. The response described in the current work is more pathological, he suggested. Jana and Pahan challenged astroglia with high concentrations of fibrillar Aβ, trying to mimic the conditions the cells might be subjected to in proximity to amyloid plaques in the brain.

The authors report that a combination of 1-μM fibrillar Aβ, prepared by allowing Aβ1-42 peptides to aggregate in solution, together with 10 ng/ml of the pro-inflammatory cytokine interleukin 1 β (IL-1β), acutely elevated levels of neutral sphingomyelinase in primary human astrocytes. (The acidic version of this enzyme did not rise.) A trans-well culture system, in which glia and neurons share medium but don’t touch, revealed that IL-1β/fibrillar Aβ-activated astrocytes (FAA) induce neutral sphingomyelinase activity in neurons as well. Ceramide, a product of sphingomyelin breakdown, also ran amok in the neurons, which began to undergo apoptosis as judged by TUNEL staining. Two other measures of cell death—release of lactate dehydrogenase and mitochondrial damage—further indicated that neurons in the co-culture were under duress. The authors found less robust responses when they treated the cells with fibrillar Aβ alone. The authors chose to use the IL-1β combination to better mimic in vivo events, as IL-1β is elevated in the AD brain (see Griffin et al., 1989).

To test if sphingomyelinase was a cause or an effect of neuronal cell death in this trans-well system, the researchers knocked down its expression in the primary neurons with antisense oligonucleotides. Silencing neutral sphingomyelinase, but not its acidic form, protected the neurons against activated glia, placing this enzyme upstream in the cell death cascade. GW4869, an inhibitor of neutral but not acidic sphingomyelinase, similarly protected neurons, whereas imipramine, an inhibitor of the acidic form, did not.

What is it about glia activated by fibrillar Aβ that sets off neuronal neutral sphingomyelinase? The authors looked to soluble factors the glia might release. Ceramide is a ready suspect, since sphingomyelinase is elevated in the glia in response to Aβ; however, even though Jana and Pahan found more ceramide inside activated glia than normal glia, it was not released into the culture medium. In contrast, nitric oxide (NO) is released from activated astroglia, and the researchers found reasons to believe that it may mediate the neuronal response. An NO scavenger protected neurons from the activated astroglia, while an NO donor activated neutral sphingomyelinase in primary neurons. Moreover, silencing neutral sphingomyelinase in astroglia completely prevented activation of the inducible nitric oxide synthase gene by Aβ/IL-1β and the ensuing production of NO.

“This is a pretty comprehensive and definitive study,” said Terrence Town, University of California, Los Angeles, in an interview with ARF. “The key question in AD neuroinflammation has been whether we should promote or oppose gliosis to treat the disease. At least in their system, it looks like we should be opposing it.” Town, who was not involved in the study, pointed out some outstanding questions, most notably whether the same sequence of events takes place in vivo.

The authors started addressing this question by injecting fibrillar Aβ into the cortex of male wild-type mice. Within six hours, neutral sphingomyelinase activity went up, inducible nitric oxide synthase turned on specifically in glia, and neuronal apoptosis became detectable starting at 24 hours after injection. “These results clearly suggest that in the Aβ-microinjected model, glial activation is followed by neuronal apoptosis,” write the authors.

Town cautioned that the glial activation and neuronal apoptosis may not be causally related. “They do have temporal data to suggest that astrogliosis kicks off neuronal apoptosis, but no direct evidence,” he said, noting that Aβ could be damaging the neurons and glia independently but at different time scales. “But on the translational side, I don’t think this detracts at all from using neutral sphingomyelinase inhibitors as an AD therapeutic approach, in principle,” he said.

While the interplay in vivo remains to be worked out, the authors did show that astroglia can affect neurons dramatically, “which is something that is too often overlooked,” noted Hartmann. “Since there are 10 times more glia in the brain than neurons, if glia can damage neurons, then that becomes an important factor.”—Tom Fagan.

Reference:
Jana A, Pahan K. Fibrillar amyloid-beta-activated human astroglia kill primary human neurons via neutral sphingomyelinase: Implications for Alzheimer’s Disease. J. Neuroscience. 22 September 2010; 301:12676-12689. Abstract