Haass’s group at Ludwig Maximilian University got onto the trail of myelin when they established a developmental profile of BACE-1 expression in the mouse. They saw no BACE expression in adult mice, but sky-high expression in the two weeks after birth. “It was a difference like night and day and night,” recalled Haass. “That gave us the idea that something must happen functionally after birth with BACE.”

In parallel, the scientists knocked down BACE in zebra fish and noticed a movement phenotype, where the fish no longer swam away as normal ones do in response to a shock. This pointed to a problem with the peripheral nervous system. Because myelination begins in earnest right after birth, it came up as a suspect at this point. Turning back to mice, the researchers made sections of the sciatic nerve of the BACE-1 knockout mice generated by Bart de Strooper’s lab in Belgium and saw that it was much less extensively myelinated than in wild-type mice. Some axons in the BACE-less sciatic nerve had no myelin around them, while others had an abnormally thin sheath. The axons in their sciatic nerve were also bundled aberrantly.

This hypomyelination persisted into adulthood. It is not a quirk of this particular mouse strain, either. Martin Citron at Amgen in Thousand Oaks, California, checked in adult mice of his groups’ BACE-1 single and BACE-1/2 double knockouts, and reported at the ICAD conference that their myelin, too, was thinner by about a third than that of wild-type. BACE-2 knockout mice do not show this phenotype, however, so it is specific to BACE-1. The mice show no overt motor deficit, but Citron noted that they have yet to be stressed or subjected to detailed strength tests.

The hypomyelination phenotype is well known in the literature. It resembles the phenotype of a heterozygous neuregulin-1 knockout (homozygous neuregulin deletion is lethal), as well as that of knockouts of ErbB, a neuregulin-1 ligand. This hinted that BACE function could lie in neuregulin-1 signaling.

Next, the German scientist, which included Linna Rabe and Michael Willem, looked for neuregulin-1 in mouse brain lysates. They reasoned that if it were a substrate for BACE, then in knockout mice the full length neuregulin-1 protein would accumulate. They indeed found unprocessed neuregulin-1 in the lysate and with further mechanistic work found evidence suggesting that BACE cuts neuregulin-1 twice, once in front of and once right behind neuregulin-1’s EGF domain. This releases the EGF domain from the axon, possibly to signal to the Schwann cell, Haass speculated.

As always with first data from basic science, their implications for drug development are hard to predict. Citron, Haass, and other scientists noted that they do not see it as a show stopper for BACE as a target, because the effect occurs mostly in development. Importantly, wholesale genetic knockouts represent a more extreme phenotype than a calibrated pharmacological reduction of BACE levels. Even so, after brain injury, BACE is known to become upregulated and conceivably aids in remyelination of damaged fiber tracks. AD patients who get a stroke or, more immediately relevant, a peripheral nerve injury, while on a BACE inhibitor, might in theory recover less well than they would without the drug. This is far from clear at this point, as the knockout mice have a peripheral phenotype, not a central one. Oligodendrocytes, which myelinate central axons in the brain, may be very different with regard to BACE and neuregulin. But the specter of a myelination side effect has put a caveat that bears watching on the radar screen of the many drug companies engaged in developing BACE inhibitors.

Haass believes that signaling the myelination of peripheral neurons is a major physiological function of BACE-1. An evolutionary tidbit supporting this view lies in the fact that invertebrates have no BACE-1. Myelination evolved with vertebrates, and BACE evolved in parallel.

The relevance of this finding for AD pathogenesis needs more research. White matter degeneration shows up as a frequent, if underappreciated, feature of early AD in various imaging modalities. Indeed, 46 presentations dealt with white matter changes in cognitive impairment and AD at the ICAD conference alone. Readers may want to consider George Bartzokis’s provocative hypothesis that AD is a disease of myelin breakdown that proceeds in a reverse ontogenetic order of myelination; that is, areas that are myelinated last in teenage development lose myelin first during AD pathogenesis, and early-myelinating areas in early childhood degenerate last in the course of AD (Bartzokis et al., 2003; Bartzokis, 2004). This is food for thought in a broader context, but the difference between peripheral and CNS BACE action precludes a direct analogy at this point.

The Haass group’s finding is equally relevant for trauma and schizophrenia. In particular, neuregulin-1 is one of the genetic risk factor genes for schizophrenia. By logic, BACE may then well be involved in schizophrenia, as well. As a next step, Haass is beginning to look for BACE upregulation and increases in neuregulin signaling in fresh postmortem schizophrenia brain tissue. This discovery, then, has brought a new lab to the growing field of schizophrenia research. The closely intertwined goings-on between molecular players in AD and schizophrenia represent an intriguing aspect in this research: BACE-1 cleaves neuregulin, and the protease that takes the baton from BACE-1 during AD pathogenesis, γ-secretase, is known to cleave a neuregulin receptor on Schwann cells, that is, ErbB4.

In summary, this work brings to light a broader parallel between APP and BACE function: Both have an important physiological role in development (neuronal migration in the case of APP, see related news story, myelination in the case of BACE), whereas in adult brains they appear to play useful repair roles in response to stressors such as ischemia or injury. Then where, exactly, do things first go awry?—Gabrielle Strobel.