The β-secretase is critically involved in the generation of the amyloid β-peptide (Aβ), a major component of amyloid plaques in the brain of Alzheimer's disease (AD) patients. β-Secretase has been identified in 1999 by several groups as a novel type of aspartic protease and dubbed beta-site amyloid precursor protein (APP) cleaving enzyme 1 (Bace1), memapsin 2, and Asp2 [for review (Kandalepas and Vassar 2012)]. Bace1 is a ~ 70 kDa type 1 membrane protein with a globular luminal domain containing two characteristic D(T/S)G(T/S) catalytic site motifs of aspartic proteases, a single transmembrane domain, and a short cytoplasmic tail (Walter 2006; Vassar et al. 2009). This protease shares all characteristics previously identified for the beta-secretase, including cleavage specificity for the APP, pronounced expression in neuronal cells, an acidic pH optimum, and localization in vesicular compartments. A second protease of similar structure was also identified and termed Bace2 (or memapsin 1 or Asp1) (Kandalepas and Vassar 2012). Bace1 and Bace2 are related to aspartic proteases of the pepsin family, but are unique in having a transmembrane domain which anchors these enzymes in cellular membranes (Fig. 1). Despite their overall structural similarity, Bace1 and Bace2 differ in their cleavage specificity within APP. While Bace1 cleaves at the Asp1 residue of the Aβ domain and thereby initiates Aβ production, Bace2 cleaves APP much more efficiently in the middle of the Aβ domain and thereby rather reduces the generation of Aβ in cell culture models (Yan et al. 2001; Fluhrer et al. 2002). The crucial role of Bace1 in Aβ generation in vivo has been fully confirmed in studies with Bace1 knock-out mice. Crossing Bace1 knock-out to APP transgenic mice completely inhibited formation of Aβ plaques and also prevented cognitive deficits usually observed in mice over-expressing APP (Luo and Yan 2010; Kandalepas and Vassar 2012). However, the deletion of Bace2 did not alter generation of Aβ, indicating that this protease only plays a minor if any role in the metabolism of APP in vivo (Dominguez et al. 2005).
Bace1 and Bace2 also show very different expression pattern. Consistent with a prominent role in Aβ generation in the brain, Bace1 is predominantly expressed in neuronal cells, but also in many other cell types at moderate or low levels. In contrast, Bace2 expression is low in most brain regions, but particularly high in several peripheral secretory tissues, including pancreas, prostate, stomach, and thyroid (Kandalepas and Vassar 2012). The differences in expression and substrate specificity suggest that both proteases also exert distinct physiological functions.
In the Journal of Neurochemistry (DOI: 10.1111/jnc.12198), van Bebber et al. (2013) used the zebrafish model to identify and characterize physiological functions of both Bace variants. Zebrafish Bace1 shows 74% identity to the human ortholog with specifically high conservation in the active site motifs and the cytoplasmic domain. While the amino acids in the active site are critical for the proteolytic reaction and substrate selection, the cytoplasmic tail of Bace1 contains amino acid motifs important for the subcellular trafficking, including a binding motif for monomeric adaptor proteins of the GGA family and a phosphorylation site for protein kinase CK1 (Walter 2006).
In zebrafish, Bace1 mRNA and protein expression increase during embryonic development, and is also detected in larval and adult stages. To study the physiological function of Bace1, loss of function mutations were introduced by genome editing using zink finger nuclease technology. To specifically assess the role of Bace1 in myelination, Bace1−/− fishes were crossed to a transgenic line expressing a membrane-bound variant of GFP (mGFP) under control of the myelin-specific claudin k promoter thereby allowing the investigation of myelin forming oligodendrocytes and Schwann cells in live animals. Bace1−/− larvae showed reduced myelination of anterior and posterior lateral line neurons by Schwann cells in the peripheral nervous system (PNS), while oligodendrocytes that enwrap Mauthner axons in the central nervous system (CNS) appeared normal, suggesting Bace1 might play a specific role in the myelination in the PNS. Initial data from mouse models also suggested a specific role of Bace1 in myelination of the PNS (Willem et al. 2006), while other studies also demonstrated altered myelination in the CNS in Bace1−/− mice (Hu et al. 2006). It is also interesting to note that Bace1 deficient mice reveal an increased frequency of spontaneous and kainite-induced epileptiform seizures (Kobayashi et al. 2008; Hitt et al. 2010; Hu et al. 2010), suggesting impaired regulation of neuronal network activity. However, it remains to be determined, whether these phenotypes are indeed related to altered myelination in the CNS. It has also been shown that Bace1 cleaves the β-subunits of voltage gated sodium channels (Vassar et al. 2009), which might also contribute to neuronal signal propagation.
As the proteolytic processing of Neuregulin (Nrg) 1 type III by Bace1 plays a critical role in myelination in mice (Hu et al. 2006; Willem et al. 2006), van Bebber et al. also compared the phenotypes of Nrg1 type III knock-down and Bace1−/− knock-out in the zebrafish model. The knock-down of Nrg1 type III expression induced a very similar, yet more severe phenotype of hypomyelination of anterior lateral line neurons as compared to the Bace1 knock-out fishes, suggesting that Nrg1 type III and Bace1 act in the same signaling pathway that regulates myelination.
A phenotypic similarity between the Bace1 knock-out and Nrg1 type III knock-down variants was also observed regarding mechanosensory neuromasts, both having increased number of neuromasts as compared to the corresponding wild-type genotypes. Neuromasts are specialized organs in the lateral line of fishes and some other aquatic animals consisting of hair cells that sense water movement. As this cell type in the lateral line organ appears to be related to sensory hair cells in the auditory and vestibular system, it will be interesting to assess whether Bace1 might also serve a function in hearing and spatial orientation in mammals.
Although the phenotypic similarities of Bace1−/− and neuregulin−/− zebrafish are very suggestive for a functional relation of both proteins, further investigations, for example, additional genetic reconstitution experiments with the Nrg1 type III processing products in a Bace−/− background, would be needed to proof whether Bace1 mediated proteolytic processing of Nrg1 type III is indeed required for proper myelination and neuromast development. It will also be interesting to assess whether other proteases could partially compensate for the loss of Bace1 in cleavage of Nrg1 type III, which might explain the more severe phenotype of Nrg1 deficient fishes as compared to that of Bace1 knock-outs.
Van Bebber et al. also extended their work in the zebrafish model to the characterization of physiological functions of Bace2. The deletion of Bace2 induced dilatation of melanophores and altered migration of these cells resulting in abnormal pigmentation. No significant effect on myelination or the number of neuromasts was observed in Bace2−/− fishes, indicating that the two Bace homologs in zebrafish exert distinct functions. This could be further supported by direct comparison of Bace1 and Bace2 single knock-out with Bace1/Bace2 double knock-out fishes. Interestingly, the neuromast phenotype observed in Bace1−/− fishes was not further enhanced in Bace1/2 double knock-outs. Likewise, the pigmentation phenotype of Bace2−/− fishes caused by aberrant melanophore morphology and migration was also unchanged in Bace1/2 double knock-out fishes. Together, the current data demonstrate that Bace1 and Bace2 have specific and non-redundant functions in zebrafish.
While the function of Bace1 in myelination appears to be conserved from fish to mice, it will be interesting to test whether the role of Bace2 in melanophore migration identified in zebrafish could also be established in mammals. So far, only one physiological function for Bace2 has been demonstrated in mice. Bace2 cleaves the plasma membrane protein Tmem27 that stimulates the proliferation of pancreatic β cells and secretion of insulin (Esterhazy et al. 2011). Accordingly, Bace2 dependent shedding of Tmem27 negatively regulates β cell proliferation and insulin secretion. Bace1 did not cleave Tmem27, also supporting separate functions of both proteases. Interestingly, the specific inhibition of Bace2 led to improvement of blood glucose homeostasis and increased β cell mass in an obesity mouse model, suggesting that Bace2 could be targeted in diabetes therapy (Esterhazy et al. 2011).
The identification of physiological functions of Bace1 and Bace2 also has important implications for the targeting of these enzymes in AD therapy. Although myelination mainly occurs during post-natal development, re-myelination is also important for axonal regeneration during adulthood. In that regard it is interesting to note that the deletion of Bace1 affects re-myelination upon sciatic nerve injury in mice (Hu et al. 2008).
The development of pharmacological inhibitors for Bace1 is challenging, because of the unusual large catalytic cleft in the active site of this protease. In addition, to efficiently target Aβ production in the brain, compounds suitable for clinical application need to penetrate the blood–brain barrier (Ghosh et al. 2012). Because of intensive research in academia and industry, promising Bace1 inhibitors have been developed and some are currently tested in clinical trials for AD therapy. However, it was reported recently that a orally available non-peptidic Bace1 inhibitor caused lipofuscin-like deposition in the retina of rats (May et al. 2011). Retinal aberrations were also observed in Bace1−/− mice or upon treatment of wt mice with a Bace1 inhibitor (Cai et al. 2012). Interestingly, deletion of Bace2 also induced defects in the mouse retina. However, while Bace1 deletion mainly affected the neural retina, Bace2 deficiency specifically affected the choroidal vasculature (Cai et al. 2012). The molecular mechanisms underlying these effects are unclear, but might involve differential cell and tissue specific expression of both proteases and cleavage of different protein substrates.
The number of identified protein substrates for Bace proteases is steadily increasing, suggesting their involvement in multiple biological processes (Luo and Yan 2010; Kandalepas and Vassar 2012). Thus, it is very important to analyze potential effects of pharmacological inhibition of Bace proteases in order assess the safety and specificity of compounds designed for AD therapy. The identified distinct physiological functions of Bace1 and Bace2 in zebrafish, thus, could not only help to further dissect molecular mechanisms but also facilitate the development and characterization of pharmacological inhibitors in vivo in a vertebrate organism.