As the two enzymes responsible for cleaving Aβ peptide out of its parent amyloid precursor protein (APP), β-secretase (BACE1) and γ-secretase have long captured the attention of AD drug developers. The problem is that these proteases cut things other than APP, and inhibiting them not only curtails Aβ production but also shuts down other critical biological processes. Wong and others have reported that BACE1 knockout mice have hypomyelinated axons (Willem et al., 2006), impaired synaptic function and cognition (Laird et al., 2005), as well as schizophrenia-like behaviors (Savonenko et al., 2008). Blocking γ-secretase activity in mice throws off Notch and epidermal growth factor signaling, causing skin tumors and gastrointestinal side effects (Li et al., 2007). Wong’s lab showed that reducing BACE1 or γ-secretase activity by just 30 to 50 percent was enough to prevent most of the unwanted side effects—however, with dampened CNS benefits. “We thought, Could we increase the efficacy in terms of reducing amyloidosis while limiting the side effects?” Wong said. “Since we knew both enzymes are required for generation of Aβ, perhaps they work in an additive manner.”

First author Vivian Chow and colleagues put this question to the test by creating three new lines of APP/PS1 mice: one lacking an allele of BACE1, one lacking an allele of Aph-1a (an essential component of the γ-secretase complex), and one heterozygous for both BACE1 and Aph-1a. Measuring Aβ load using biochemical and immunohistochemical analyses, the researchers found far less amyloid in the brains of BACE1+/- Aph-1a+/- APP/PS1 mice than in either of the single heterozygote APP/PS1 mice. This relative trend held up even as the animals aged to 19-23 months. APP/PS1 mice normally show advanced stages of Aβ pathogenesis by that age. Furthermore, amyloid reduction seemed to correlate with functional benefits. Compared with the APP/PS1 single heterozygotes, BACE1+/- Aph-1a+/- AD mice were better protected from memory deficits, to an extent that their performance in a Morris water maze was comparable to that of wild-type mice. Importantly, the double heterozygote APP/PS1 mice had a normal lifespan and seemed free of the other untoward side effects seen in animals completely lacking either secretase.

For the most part, the above outcomes were expected, Wong told ARF. However, “we were a little bit surprised that we got such a dramatic effect functionally,” he said. Consistent with them being key mediators of synaptic dysfunction, the researchers found reduced amounts of oligomeric Aβ in brain tissue from BACE1+/- Aph-1a+/- AD mice, whereas brain levels of Aβ oligomers hardly changed in single heterozygote AD mice. “That might be one reason why we got such a robust recovery of learning and memory in the double reduced mouse models,” Wong said.

Whatever the reason, the fact that modest reduction of both secretases could curb amyloid accumulation and preserve cognition, while avoiding harmful side effects in AD mice, suggests that combination strategies may eventually hold promise in the clinic. It remains to be seen whether the benefits of secretase-targeting compounds offered to AD patients in mid- to late life mirror the effects of genetic deletions present from birth in mice. In the meantime, the current study provides an important framework for evaluating future experimental drugs. “Genetic models define the limits of the mechanism-based toxicities,” Wong said. If compounds carry other toxicities not seen in the animal models, that would suggest the effects are due to the compound itself and may not relate to inhibition of APP processing.

The combination therapy could have benefits above and beyond blocking Aβ production. In a PNAS paper published in December, Ralph Nixon and colleagues at the Nathan Kline Institute in Orangeburg, New York, showed that another APP fragment—C-terminal product of β-secretase cleavage (βCTF)—is toxic early in AD (see ARF related news story). If true, this would mean that manhandling Aβ isn’t enough, offering further support for a multipronged attack on APP processing. As Wong noted, Nixon’s study suggests that “lowering the activity of γ-secretase alone may not be sufficient to deal with both problems of Aβ and βCTF accumulation.” (See full comment.)

Nixon agrees. “I think the stories are very compatible,” he said of his and Wong’s studies in an e-mail to ARF. “The partial inhibition of both secretases is quite interesting and could easily be viewed as implicating βCTF as well as Aβ. In addition to the mechanism Wong proposes of additive/synergistic effects on Aβ lowering, our data would suggest that an additional explanation for the combined inhibitor effects on memory deficits is a reversal of separate adverse effects of βCTF and Aβ on cognitive performance.” (See full comment below)

In the meantime, understanding why BACE1 inhibitors can be so toxic would speed their clinical development, which currently lags behind that of γ-secretase inhibitors. Teasing out BACE’s physiological roles would surely help. Toward this end, a team led by Dennis Selkoe, Brigham and Women’s Hospital, Boston, has used quantitative proteomics to identify 63 potential new substrates for BACE1. First author Matthew Hemming and colleagues probed two BACE1-overexpressing epithelial cell lines (HEK and HeLa) for proteins whose levels rose in parallel with increased BACE1 activity. More than half of the proteins are involved in intercellular communication or are receptors with recognized biological functions. If these are real substrates, then they may highlight additional concerns about strongly inhibiting BACE1, Wong noted. However, since the studies were done using cell lines that express artificially high levels of BACE1, the physiological relevance of the identified proteins remains in question, he said.

Other scientists did not see the BACE1 overexpression in these studies as a caveat as much as a possible means for discovering roles for β-secretase in conditions where its levels go up, such as traumatic brain injury and energy deprivation. The study “is particularly interesting because it demonstrates that increased β-secretase activity results in alteration of normal cellular functions, e.g., missorting of lysosomal hydrolases and increased cleavage of cell adhesion molecules, which may be detrimental for the cell,” wrote Giuseppina Tesco, Tufts University, Boston, in an e-mail to ARF. (See full comment below.) “Given that BACE1 is elevated in AD brains, increased cleavage of additional substrates may contribute to AD pathology. Thus, β-secretase inhibition is expected to be beneficial in conditions associated with increased levels of BACE1.”—Esther Landhuis.

References:
Chow VW, Savonenko AV, Melnikova T, Kim H, Price DL, Li T, Wong PC. Modeling an Anti-Amyloid Combination Therapy for Alzheimer’s Disease. Sci Transl Med. 2010 Jan 6;2(13):13ra1. Abstract

Hemming ML, Elias JE, Gygi SP, Selkoe DJ. Identification of Beta-Secretase (BACE1) Substrates Using Quantitative Proteomics. PLoS ONE. December 2009;4(12):e8477. Abstract