13 February 2009. Infamous among Alzheimer disease researchers for delivering the final cut that unleashes Aβ from its parent protein in neurons, γ-secretase has less-proclaimed physiological functions, too. A few of these came out in a recent paper from researchers led by Ben Barres at Stanford University in Palo Alto, California. In the 26 November issue of Neuron, the scientists unveiled a surprising role for γ-secretase in glia, the non-neuronal cells that provide nourishment and protection in the nervous system. “The idea that [γ-secretase] might have a completely different role in glia is novel,” Barres said in an interview with ARF. “Its role is to inhibit myelination. It's acting as an on-off switch.” In the same study, the researchers showed that young oligodendrocytes initiate myelination within a remarkably brief time window and that astrocytes enhance the speed and degree of myelin wrapping.
These findings came as byproducts of fulfilling a greater goal Barres has had since setting up his lab at Stanford 16 years ago—the development of a rapidly myelinating culture system. Most existing in-vitro systems use mixtures of embryonic CNS cells to myelinate endogenous axons over a period of weeks. Slow and hard to interpret, these assays have yielded little mechanistic insight as to how myelination happens, Barres said. Developing a more defined system using co-cultures of purified neurons and glia was the main goal of this work.
First author Trent Watkins, at Genentech in South San Francisco, and colleagues started by isolating perinatal rat retinal ganglion cells, one of the few CNS neurons with established purification and culture methods. Visualized by electron microscopy and various lipophilic dyes, early signs of myelination appeared on the ganglion cell axons after six days of co-culture with oligodendrocytes derived from precursors in a rat optic nerve prep. The researchers had to tweak conditions to boost efficiency, though, because most of the oligodendrocyte precursors in the starting population either strayed into the astrocyte lineage or otherwise failed to differentiate into myelinating cells.
Barres’s group had shown previously that Notch1 signaling blocks differentiation of oligodendrocyte precursors and that retinal ganglion cell axons express the Notch ligand Jagged1 (Wang et al., 1998). Fingering Notch1 as a possible culprit behind the failed oligodendrocyte differentiation in their co-cultures, the scientists spiked the cultures with DAPT, which inhibits γ-secretase, a protease required for Notch1 activation. This did the trick, sending more than 70 percent of oligodendrocyte precursors into the business of myelination. When co-culture conditions were adjusted to allow more astrocytes to form, myelination picked up speed and intensity. And, importantly from a logistical standpoint, the researchers were able to adjust the co-culture protocol to use mouse cells in place of rat neurons or glia—a key feature for their subsequent experiments using existing genetic models.
To get a sense for when during the oligodendrocyte maturation process myelination occurs, the scientists isolated these glial cells at various developmental stages from mice or rats, and cultured them with mouse retinal ganglion cells. For an even closer look, they transfected oligodendrocyte precursors with a membrane-targeted green fluorescent protein and caught them in the act of myelination using time-lapse microscopy. Together, these experiments suggest that myelination capacity peaks during early oligodendrocyte differentiation and that once the process begins, it tends to happen in a single burst within 12 to 24 hours.
“The surprise is that when an oligodendrocyte decides to myelinate, it does so in a very short period of time. This implies there's a nuclear program,” Barres told ARF. “Up to now, it's been assumed that the decision to myelinate is just governed by a local interaction between an oligodendrocyte process and its axon. This is clearly not so. This has huge implications for how myelination is controlled.”
To address how myelination is regulated at the molecular level, Barres and colleagues turned back to γ-secretase inhibition. Having shown that pharmacological blockade of Notch1 signaling using the γ-secretase inhibitor DAPT lifted the brakes on oligodendrocyte differentiation and promoted myelination, the scientists performed the genetic version of this experiment—by culturing Notch1-deficient glia with rat neurons. In this scenario, the absence of Notch1 activity stimulated oligodendrocyte differentiation but did not seem to drive myelination. Adding DAPT into this system restored the myelination boost seen in the earlier co-cultures with wild-type glia. “Together, these results implicate glial γ-secretase in the regulation of myelination in at least two ways: in the control of differentiation by Notch1 signaling and in the Notch1-independent modulation of myelin segment initiation,” the authors write.
“This is a great story with an important new technique to study myelination,” wrote Bart de Strooper of K.U. Leuven in Belgium, who was not involved with the new work. “A fallout of this study, i.e., that γ-secretase inhibition promotes myelination in two ways, will stimulate without any doubt investigations to explore the potential of γ-secretase inhibitors in demyelination disorders.” In collaboration with Stephen Miller at Northwestern University in Evanston, Illinois, Barres’s team has preliminary data showing that inhibition of γ-secretase markedly reduces disease severity in animal models for multiple sclerosis.
Whether the new study has relevance to AD is less clear, though several lines of evidence offer food for thought. Interestingly, the β-secretase BACE1—which partners with γ-secretase to cleave Aβ out of its parent protein APP—is also implicated in myelination. Several years ago, Christian Haass at University of Munich, Germany, and colleagues reported that BACE1 triggers a period of robust myelination in developing peripheral motoneurons (Willem et al., 2006 and ARF related news story). The scientists found that BACE1 knockout mice accumulate in their brains an excess of full-length neuregulin-1 protein, a substrate shared by BACE1 and γ-secretase. Subsequent studies have found that BACE1 delivers the extracellular cut that modulates signals to neighboring Schwann cells (Savonenko et al., 2008 and ARF related news story), while γ-secretase releases neuregulin’s intracellular domain (Dejaegere et al., 2008 and ARF related news story) to provide a cell-autonomous signal. De Strooper noted that this bidirectional signaling leads to similar schizophrenic-like behavior in BACE1- and Aph1B-γ-secretase-deficient mice, and wonders whether BACE1 and γ-secretase could act dually in a similar fashion within the context of myelination.
The enzymes do appear to have opposing roles in this regard—BACE1 drives myelination, γ-secretase inhibits it—but they act in different cell types (BACE1 in neurons, γ-secretase in glia). So far, Barres’s group has not seen evidence of antagonism in their co-cultures. “We don't think our effects are BACE-dependent,” he said. “When we tried BACE inhibitors in our system, we didn't see any effects.” He and colleagues are working hard to identify the relevant γ-secretase substrate in their myelination system. Highly expressed in oligodendrocytes, APP is “one of the most interesting candidates,” Barres said. But the jury is still out.
The notion that myelination may be tied with AD has been around for some time, and draws support from several recent papers. In one study (Ringman et al., 2007), presymptomatic individuals with familial AD mutations in presenilin 1 (PS1) were found to have lower white matter volume and integrity in areas that myelinate late during brain development, compared with their non-carrier relatives. In another study (Bartzokis et al., 2007), decreased white matter integrity appeared in asymptomatic subjects whose ApoE4 genotype put them at increased risk for late-onset AD. That the white matter defects in each study showed up prior to the appearance of known AD structural biomarkers (e.g., brain or hippocampal atrophy) supports the idea that “myelin breakdown and loss precedes brain atrophy and is the driving force behind AD,” wrote George Bartzokis of the University of California, Los Angeles, in an e-mail to ARF. (For a review on this hypothetical model of AD, see Bartzokis, 2004).—Esther Landhuis.
Watkins TA, Emery B, Mulinyawe S, Barres BA. Distinct stages of myelination regulated by γ-secretase and astrocytes in a rapidly myelinating CNS co-culture system. Neuron. 2008 Nov 26;60(4):555-69. Abstract