Biochemical Society Transactions


Expression and activity of β-site amyloid precursor protein cleaving enzyme in Alzheimer's disease

J.A. Johnston, W.W. Liu, S.A. Todd, D.T.R. Coulson, S. Murphy, G.B. Irvine, A.P. Passmore


Several lines of evidence indicate that the Aβ peptide is involved at some level in the pathological process that results in the clinical symptoms of AD (Alzheimer's disease). The N-terminus of Aβ is generated by cleavage of the Met-Asp bond at position 671–672 of APP (amyloid precursor protein), catalysed by a proteolytic activity called β-secretase. Two ‘β-secretase’ proteases have been identified: BACE (β-site APP-cleaving enzyme) and BACE2. The cause of sporadic AD is currently unknown, but some studies have reported elevated BACE/β-secretase activity in brain regions affected by the disease. We have demonstrated that robust β-secretase activity is also detectable in platelets that contain APP and release Aβ. This review considers the current evidence for alterations in β-secretase activity, and/or alterations in BACE expression, in post-mortem brain tissue and platelets from individuals with AD.

  • Alzheimer's disease (AD)
  • amyloid precursor protein (APP)
  • β-site amyloid precursor protein-cleaving enzyme (BACE)
  • cholesterol
  • platelet
  • post-mortem brain protease


Aβ peptides are found aggregated in the cortical amyloid plaques associated with AD (Alzheimer's disease) neuropathology [1]. These peptides are generated when APP (amyloid precursor protein) undergoes proteolysis between positions 671–672 (APP 770 numbering), and anywhere between positions 710–715, to give rise to 39–43 amino acid Aβ [2]. This is represented schematically in Figure 1. Cleavage of the Met-Asp bond at 671–672 (Asp1 of Aβ) generates the N-terminus of Aβ and is catalysed by a protease activity known as β-secretase. Two ‘β-secretase’ proteases have been identified and are known as BACE (β-site APP cleaving enzyme) and BACE2 [36]. Several lines of evidence indicate that Aβ is involved in the sequence of pathological events that culminates in the clinical symptoms of AD. Fibrillar Aβ is a major component of amyloid plaques, and lower-order assemblies of Aβ peptides have been shown to compromise hippocampal long-term potentiation in vivo [7]. Studies of families where AD is inherited in an autosomal dominant fashion have revealed point mutations in the APP, presenilin-1 and presenilin-2 genes that co-segregate with the disease. These mutations cause the generation of more Aβ, or more of the longer, 42 amino acid, Aβ42 [8,9]. However, the majority of people affected by AD (probably >97% [10]) lack a strong family history of the disease and are therefore classified as ‘sporadic’ cases. The cause of sporadic AD remains unknown. The APP and presenilin mutations demonstrate that elevated or altered Aβ production is sufficient to cause the disease, and it is possible that other factors serve to elevate Aβ production in sporadic AD. Increased β-secretase activity would be predicted to elevate Aβ production and elevated BACE/β-secretase activity has been reported in brain regions affected by AD [1114]. We have recently identified an increase in β-secretase activity in platelets prepared from individuals with AD, as compared with age-matched controls. This review considers the current evidence for alterations in BACE expression, and/or alterations in β-secretase activity, in tissues from individuals with AD.

Figure 1 A schematic representation of APP processing

Alternatively spliced regions are shown in pale grey. The presence or absence of these exons generates isoforms ranging from 695 to 770 amino acids in length. APP is processed along at least two parallel pathways, the α- and β-secretase pathways. Full-length APP is either cleaved within the Aβ peptide, catalysed by α-secretase, or at the N-terminus of the Aβ peptide, catalysed by β-secretase. These cleavage sites are indicated by arrows. This results in the production of soluble secreted APP (APPs α or β) and a C-terminal cell-associated ‘stub’ that is a γ-secretase substrate. γ-Secretase cleavage results in generation of p3, or Aβ, peptides in the α- or β-secretase pathways respectively.


Catalysis of APP cleavage to generate the N-teminus of Aβ was initially called ‘β-secretase’ activity, and represents the rate-limiting step in Aβ production. In 1999, a protease was described that fulfilled all the criteria for β-secretase. This enzyme was identified independently by four groups and was named BACE, Asp2 or memapsin 2 [35,15]. A schematic representation of BACE is shown in Figure 2. A BACE-homologue, BACE2, was identified shortly afterwards [6]. Both BACE and BACE2 are aspartyl proteases that can cleave APP at Asp1 and Glu11 of Aβ. BACE seems to be more widely expressed in the brain [16], and BACE knockout animals do not produce detectable levels of Aβ, suggesting that BACE2 does not catalyse much APP cleavage in vivo [17]. BACE is 501 amino acids long, with a signal peptide (1–21), a prodomain (22–45) and a single transmembrane domain (461–477). The 23 amino acid prodomain is removed by furin or a furin-like activity [18], and BACE is also phosphorylated, N-glycosylated and palmitoylated [19,20] during maturation (Figure 2). The active site involves Asp93 and Asp289 and is on the luminal/extracellular side of the membrane, as is the β-secretase cleavage site on APP. BACE could cleave APP while membrane-inserted in the lumen of an acidic intracellular compartment, most likely to be endosomes, since its optimal pH is acidic (∼4.5). The active site Asp residues are located in the centre of the substrate-binding cleft, which accommodates eight substrate residues [21]. A loop known as the ‘flap’ partially covers the cleft and may regulate substrate access to the active site [21].

Figure 2 A schematic representation of BACE

The amino acid numbering is shown, and the prodomain (positions 22–45) is shaded pale grey. The active site Asp93 and Asp289 residues are indicated (*). N-glycosylation sites (+) are present at Asn153, Asn172, Asn223 and Asn354, and a phosphorylation site at Ser498 (¤). BACE can be palmitoylated at Cys478, Cys482 and Cys485 at the intracellular/cytosolic face of the membrane (region indicated by a solid bar). Cys residues form three intramolecular disulphide bonds between positions 216 and 420, 278 and 443, and 330 and 380 (indicated by brackets).

β-Secretase activity and BACE levels in the brain

Several post-mortem brain studies have investigated β-secretase activity, BACE protein levels and BACE mRNA levels in regions affected by AD. In a study of approx. 60 AD and 20 control brains, increased levels and enzymatic activity of BACE were identified in AD [11]. Using an antibody, BACE was captured and quantified, and β-secretase activity was assayed in temporal cortex, frontal cortex and cerebellum. BACE levels increased by approx. 15% in frontal and temporal cortices, accompanied by significant increases in enzymatic activity of 13% in the frontal cortex and 63% in the temporal cortex. Temporal cortex β-secretase activity correlated with the duration of illness and no changes were observed in the cerebellum. Analysis of β-secretase activity in a separate study of temporal cortex from 15 AD and 16 controls also reported a significant 85% increase in β-secretase activity in the AD group [22]. A separate BACE capture assay detected significant 2–3-fold elevations of BACE protein in temporal cortex and hippocampus and increased β-secretase activity in AD (∼2-fold), with no change in cerebellar BACE [13,14].

Other studies have addressed the issue of BACE mRNA expression in post-mortem brain. BACE mRNA expression was reported to be unchanged in frontal cortex from 16 AD patients as compared with 10 controls using quantitative PCR, even though a 2.7-fold increase in BACE protein levels was observed using Western immunoblotting [12]. Two other studies have reported unchanged BACE mRNA expression in the disease [23,24], although one report found some evidence for increased BACE mRNA in AD brains using Northern blotting and RNase protection assay [14].

Taken together, these reports indicate that β-secretase activity and BACE protein levels are elevated in regions of the brain affected by AD pathology, but provide no ready consensus on the degree of elevation. There is less evidence for increased BACE mRNA expression in AD, although changes in BACE translation efficiency have been reported [25]. Some reports also identified differences between the degree of increase in β-secretase activity and the degree of increase in BACE protein levels [11], suggesting that β-secretase activity may be regulated by other factors besides BACE expression. A study of BACE expression in a transgenic mouse model of AD (Tg2576) indicated that overall BACE levels were not significantly different between the Tg2576 mice and controls [26]. This indicates that increased expression of BACE is not required in order to produce the large increases in Aβ production observed in these mice, and provides further evidence that something other than BACE expression can regulate β-secretase activity. In addition, BACE immunoreactivity was observed in reactive astrocytes in the vicinity of plaques in the older Tg2576 mice [26], while expression was exclusively neuronal in young Tg2576 mice and controls. Neuronal and astrocytic BACE expression also increased following traumatic brain injury in rats [27]. More recently, a study of BACE expression in hippocampus and frontal cortex of AD and controls indicated that although no overall changes in BACE immunoreactivity were observed, there was an increase in BACE-immunoreactive astrocytes close to plaques in AD [28]. This indicates that some of the differences between the previous studies of BACE levels may be due to the methods employed, which did not distinguish between neuronal and astrocytic BACE expression. Further studies will be required in order to elucidate whether the altered expression of BACE reported in regions of the brain affected by AD is due to altered neuronal expression patterns, or loss of neuronal cells and gain of glial cells expressing BACE.

What other factors, besides increases in BACE expression, could regulate β-secretase activity? β-Secretase activity in vivo must be partly regulated by access of BACE to APP. Increased β-secretase activity observed in vitro in the presence of a peptide substrate could also be caused by changes in enzyme kinetics in the disease, or changes in the levels of an as yet unidentified endogenous regulator. Heparan sulphate has been shown to interact with BACE and regulate β-secretase activity, possibly by affecting the flap region and inhibiting access of APP to the active site [29]. In addition, membrane-associated BACE has been reported to form dimers in vivo, which were more catalytically active than monomeric BACE [30,31]. The post-mortem brain studies reported to date did not investigate the kinetics of the protease activity in the disease, as compared with controls. In the studies where individual data points are shown for β-secretase activity, a large range in BACE activity is evident amongst the individuals studied; with up to 15-fold differences in activity [11,22]. The reason for this large inter-individual variation in β-secretase activity, and the mechanism underlying the increases in activity observed in AD post-mortem brain, are as yet unknown.

Platelet APP processing in AD

Full-length and processed APP is present in human platelets, which contain BACE and release Aβ [3236]. Previous studies revealed a reduced ratio of higher molecular mass (∼130 kDa) to lower molecular mass (∼110 kDa) APP-immunoreactive bands, following gel electrophoresis of whole platelet homogenate [3640]. This alteration in platelet APP isoform ratios correlated with declining cognition [41] and was improved following treatment with statins [42] and acetylcholinesterase inhibitor [43]. BACE immunoblotting of platelets has also revealed alterations in AD. Two BACE bands (36 and 57 kDa) were identified, and the 36 kDa band decreased in AD [36]. The precise molecular identities of the BACE, or APP, isoforms detected in this way have not yet been identified.

We have investigated β-secretase activity directly in platelets (J.A. Johnston, W.W. Liu, S. Todd, D.R. Coulson, S. Murphy, C. Foy, G.B. Irvine and A.P. Passmore, unpublished work), using a peptide substrate based on the wild-type APP sequence at the β-secretase site [MCA-EVKM-DAEFK-(DNP)-NH2]. We identified a significant elevation of platelet membrane β-secretase activity in the AD group, as compared with controls. Considerable inter-individual variability in β-secretase activity was also identified, with initial rates in some samples severalfold higher than in others. Interestingly, platelet membrane β-secretase activity correlated with cognitive decline, as assessed by minimental status examination, whereby individuals with lower MMSE scores had higher β-secretase activity. Platelet membrane β-secretase activity also correlated with platelet membrane cholesterol levels, with increased activity found in individuals with higher cholesterol levels. Recently, a separate study of platelet APP in AD demonstrated a significant increase in the cell-associated APP fragment produced following β-secretase cleavage (CTF99) AD [43]. This indicates that increased platelet membrane β-secretase activity may be relevant in vivo, producing more β-cleaved APP and presumably more Aβ.

Aβ levels in AD

Increased β-secretase activity would be predicted to increase Aβ production, but this would not necessarily increase levels of Aβ, since mechanisms are also in place to degrade Aβ and these may be able to maintain a constant level of the peptide [44]. Given that the capacity of these systems may vary in different tissues, increased Aβ production could cause the slight increases in peptide concentration that have been identified in some tissues where capacity may be weaker. The post-mortem brain β-secretase activity studies that have also measured Aβ levels present evidence both for [14] and against [11] a correlation between cortical Aβ levels and cortical β-secretase activity. Platelets are one of the major contributors of Aβ to the plasma, although there is also considerable evidence for movement of Aβ across the blood–brain barrier [45]. A number of previous studies investigated whether levels of Aβ were increased in plasma in AD. These studies demonstrated either no change in plasma Aβ, or small significant increases in plasma Aβ40 or Aβ42 in AD (reviewed in [45]). All studies reported a large inter-individual variability in Aβ levels and a broad overlap between the study groups. This agrees with our data showing a large range in platelet β-secretase activity between individuals, although as stated above, many other factors will influence Aβ levels.

Cholesterol and AD

Several lines of evidence suggest that cholesterol can modify Aβ production and influence risk for AD. There is a well-established increase in risk for AD in individuals carrying the ε4 allele of APOE, an important cholesterol transport protein [46]. Epidemiological studies identified a reduced incidence of AD in individuals treated with cholesterol-lowering statins, while reduction of cholesterol levels in vitro and in vivo shunted more APP into the α-secretase pathway (see Figure 1), and reduced Aβ production (see [47] for a recent review). Intracellular cholesterol distribution is also important, and reduction of cholesterol ester levels also reduced Aβ production [48]. In addition, some of the effects of statins on APP processing have recently been attributed to effects on isoprenoid levels, as well as cholesterol levels [49]. APP and BACE have been reported to co-localize on lipid rafts, and therefore an increase in membrane cholesterol would be predicted to increase the chance of BACE encountering APP and generating Aβ. Our study also found that platelet membrane β-secretase activity correlated positively with membrane cholesterol levels. We used an exogenous substrate, suggesting that this effect occurred independently of increased co-localization of BACE and APP in lipid rafts, although a cholesterol-dependent alteration in BACE subcellular localization cannot be ruled out (since we assayed membrane β-secretase activity). Alternatively, there may be a cholesterol-dependent change in β-secretase kinetics; either via a direct effect of cholesterol, or via altered BACE dimerization or interactions with endogenous regulator(s) of activity. In contrast with previous reports, a recent investigation of APP and BACE localization in vitro and in vivo reported that while BACE partly localized to rafts, APP did not. The authors also demonstrated increased β-secretase cleavage in reduced membrane cholesterol conditions [50]. The reason for these differences is unclear but future studies will aim to elucidate the exact links between APP processing and cholesterol metabolism.


An increase in β-secretase activity would be predicted to increase Aβ production, and this is known to be capable of causing AD in individuals with inherited forms of the disease. There is now some evidence for elevated β-secretase activity in post-mortem brain tissue and platelets from individuals affected by sporadic AD. This elevated activity does not directly correlate with increased levels of BACE protein. In addition, β-secretase activity measured in brain tissue and platelets shows large inter-individual variability. What is the basis of this variability? Cholesterol, cholesterol esters and/or isoprenoids may be involved at some level but the exact mechanism is currently unclear. By studying these issues in more detail, valuable information could be generated about endogenous regulators of β-secretase activity and Aβ generation. This would enable us to address whether this regulation is altered in AD, and perhaps uncover new therapeutic targets.


We thank the Alzheimer's Society (U.K.), and the Research and Development Office, Health and Personal Social Services, Northern Ireland for their generous financial support.


  • Proteins in Disease: A Focus Topic at BioScience2005, held at SECC Glasgow, U.K., 17–21 July 2005. Edited by B. Austen (St George's Hospital Medical School, London, U.K.), C. Connolly (Dundee, U.K.), B. Irvine (Belfast, U.K.), M. Sugden (Queen Mary, London, U.K.) and V. Zammit (Hannah Research Institute, Ayr, U.K.).

Abbreviations: AD, Alzheimer's disease; APP, amyloid precursor protein; BACE, β-site APP-cleaving enzyme


View Abstract