A recent breakthrough in the treatment of the initial stages of Alzheimer’s disease (AD) is marked by the first approved disease-modifying drugs, which are all monoclonal antibodies recognizing aggregates of the amyloid β peptide (Aβ) (1). The approval vindicates the importance of Aβ as a therapeutic target for AD, with the production, aggregation, and clearance of Aβ as strategic focal points (1). However, whereas an increased production of Aβ clearly causes early onset AD (2), chemical inhibition of the proteases (i.e., β-secretase and γ-secretase) responsible for Aβ production have so far failed to yield a successful drug, despite extensive efforts by multiple pharmaceutical industries (3). Aβ peptide is carved out of the amyloid precursor protein (APP) through sequential cleavage by β-secretase and γ-secretase (Figure 1A). β-Site APP cleaving enzyme (BACE1) has been identified as the main enzyme performing the critical amyloidogenic β-secretase cut in human and rodent brain (4), while the role of its homologue BACE2 remains poorly understood. BACE1 and BACE2 are both type I transmembrane aspartyl proteases, have a similar length, and share 59% identify in the amino acid sequence (5), which is the reason why chemical inhibitors designed to inhibit BACE1 in most cases also inhibit BACE2, to a varying extent (3). Early studies in mouse brain established BACE1 as having strong expression in mouse neurons, with the expression of BACE2 in the brain remaining very low (6). BACE2-knockout mice (unlike those with BACE1 deficiency) did not show decreased amyloid plaque load in AD models (7). These findings resulted in a relative neglect of BACE2 as a role player in AD and rendered the cross inhibition of BACE2 by the BACE1 inhibitors less important for AD treatment. However, multiple studies have since shown that human neurons express much more BACE2 than mouse neurons (8, 9). Studies in a variety of human cell-line models overexpressing or silencing BACE2 revealed a reproducible effect that dose of BACE2 has on the level of secreted Aβ: BACE2 overexpression decreases secreted Aβ levels (10–14), while siRNA silencing of BACE2 increases Aβ levels (15). However, the in vivo effects of an unwanted cross inhibition for BACE2 were difficult to prove and impossible to monitor because of the lack of an easily accessible pharmacodynamic target of BACE2 cleavage. Unlike BACE1, whose cleavage of APP and SEZ6L could be pharmacodynamically measured in cerebrospinal fluid (CSF) and partly observed in blood (16, 17), the only verified in vivo targets of BACE2 until now remained TMEM27 and PMEL, whose cleavages by BACE2 altered glucose homeostasis in β-islet cells and pigmentation in melanocytes, respectively. The manuscript by Schmidt et al., a multidisciplinary work led by Lichtenthaler and published in this issue of the JCI (18), identifies an easily accessible pharmacodynamic marker for BACE2 activity (not shared with BACE1) in human, nonhuman primate, and rodent plasma samples.

sSEZ6L and sVEGFR3 provide pharmacodynamic markers specific to BACE1 and BACE2 cleavage activities. (A) BACE1 and BACE2 predominantly cleave APP or Aβ at specific sites. BACE inhibitor drugs block activity of both proteases. (B) BACE1 cleavage of SEZ6L releases sSEZ6L into the plasma, while BACE2 cleavage of VEGFR3 releases sVEGFR3 into the plasma and lymphatic vessels. Serum levels of sSEZ6L and sVEGFR3 specifically reflect BACE1 and BACE2 activity, respectively.