The search to find therapeutic targets in Alzheimer's disease (AD) has been dominated for over 25 years by research into the roles in the initiation and progression of dementia of the amyloid beta protein (Aβ) [1, 2], derived from the β-pathway of amyloid beta protein (AβPP) cleavage. This has been driven by several lines of evidence:

1. the genetic evidence from familial forms of AD (FAD) where fully penetrant mutations in the presenilin (PSENs) and the amyloid β precursor protein (AβPP) genes are qualitative markers of disease and are associated with younger ages of disease on-set [3];
2. the neuropathological deposition of Aβ in senile plaques and as cerebral amyloid angiopathy associated with dementia in both FAD and sporadic AD [4, 5];
3. evidence from animal and cell culture models of AD based on the genetic mutations.

Taken together, the evidence has been interpreted to give Aβ a causal role in the development of dementia in humans and that modulation of Aβ is a primary therapeutic target. This approach has never been fully accepted by the AD research community [6-12] and epidemiological population based studies of ageing consistently find complex relationships between age, amyloid pathology, in-life factors such as education, and dementia status [13-17]. The recent failures of clinical trials demand that we re-examine the amyloid approach in detail. Of particular relevance to this re-examination is the question - What is Aβ?

Superficially, this question seems irrelevant - the body of literature reporting on Aβ in AD is vast and Aβ is assumed to be a well-defined molecular concept. However, when mapping the AβPP proteolytic system from a systems biology approach it becomes difficult to assign a single node to "Aβ" [18] suggesting a more complex model is required.

The first difficulty is the heterogeneity of what we understand as the Aβ amino acid sequence. Most researchers accept that Aβ40 and Aβ42 have different associations with AD; however, a detailed investigation of Aβ-related AβPP proteolytic fragments in experimental settings reveals a multitude of associated soluble peptides [19] few of which have been systematically investigated with respect to AD. Some fragments are known to cross-react with commonly used antibodies introducing confounding in interpretations of immunoassays and immunohistochemistry for Aβ, of which the perhaps most concerning is the confounding by P3-40 and P3-42 (derived from the alternate α-pathway of AβPP cleavage) in cerebrospinal fluid based biomarkers relating to C-terminal Aβ and in neuropathological diagnostic protocols using the anti-Aβ antibody 4G8 [20]. Despite known reaction with various antibodies raised against the Aβ C-terminal, no study has investigated the extent of confounding due to P3-42 and/or P3-40 with these antibodies. The enhanced reactivity profile of 4G8 when compared to both 6E10 and 6F3D illustrated in Alafuzoff et al [21] may be due to its reactivity with P3 type fragments in addition to Aβ- type fragments. The current practice of interpreting immunoreactivities seen with commonly used antibodies as "Aβ" without controlling for the other fragments misleads the entire amyloid based research approach. What do these different reactivities mean and how do we translate findings relating to Aβ between studies using different antibodies? Are we all measuring the same Aβ?

A second difficulty is the heterogeneity of Aβ aggregation state, including monomers, dimers, oligomers and fibrils. No experimental approach currently measures Aβ in all possible aggregation states so that any measure of Aβ may be missing specific aggregations with particular relevance to oligomeric forms. Aβ-type fragments of any sequence length in any aggregation state in relation to AD have not been systematically investigated in humans.

Considerations of P3-42 highlight a third difficulty - that of solubility. Evidence suggests that while P3-40 is seen in the soluble compartment, P3-42 is not [19], though it has been detected neuropathologically in fleecy amyloid deposits [22]. Differences in solubility are also seen between Aβ40 and Aβ42. The consequences of these differences in solubility and the effects of compartmentation remain to be clarified.

A fourth difficulty arises due to post translation modifications of Aβ. Taking the immunoreactivity profiles of the anti-Aβ antibodies 6E10 and 6F3D seen in human brain tissues [21], beyond considerations of aggregation state, is the lower reactivity of 6E10 also associated with N-terminal truncations or other modifications [23]? What significance do these modifications have for the physiological roles of Aβ?

A fifth difficulty arises when assigning functions to specific fragments from the AβPP proteolytic system. Most investigations focus on Aβ alone without taking the complexity of the AβPP proteolytic system into account however, this neglects the contributions from full length AβPP and other proteolytic fragments derived from AβPP including the N-terminal sAPPα released following α-cleavage and sAPPβ released following β-cleavage. Given that AβPP is rate limiting [24], any change towards the β-pathway that results in increased production of Aβ-type fragments necessarily involves loss of function in full length AβPP and/or α-pathway. It then becomes difficult to assign causal roles to gain of function of Aβ without controlling for loss of function in full length AβPP and/or products of the α-pathway. Our understanding of the roles of Aβ in AD is currently confounded by our lack of understanding of how Aβ sits within the wider context of the whole AβPP proteolytic system [18, 20, 25].

The research community as yet has no systematic approach to the definition of Aβ either in theory, e.g., how many nodes are required in a systems biology based model of the AβPP proteolytic system—or in practice—e.g., which Aβ are we measuring in immunoassays? Aβ is currently a poorly defined concept associated with multiple confounding factors which undermine our understanding of "Aβ". Without an understanding of what Aβ is, we cannot say what roles Aβ plays in human AD with any certainty with important consequences for amyloid based research. Despite strong pressures to include amyloid based immunoassay biomarkers in clinical settings, none are specific enough at a molecular level to take account of sequence, aggregation state, solubility and post translation modifications, none have been validated in the human population, and their diagnostic and prognostic usefulness is uncertain [26]. It is essential to identify and clarify ambiguities in our understanding of the amyloid based approach if we are to understand the recent failures and build a better foundation for future research. It is now well past the time for the AD research community as a whole to have an open and honest discussion, however difficult that might be, to re-visit the decades of accumulated evidence. What do we actually know about the roles of "Aβ" in all its isoforms in AD and how do we know it?

References:
[1] Hardy JA, Higgins GA (1992) Alzheimer's disease: the amyloid cascade hypothesis. Science 256, 184-185.
[2] Hardy J, Selkoe DJ (2002) The amyloid hypothesis of Alzheimer's disease: progress and problems on the road to therapeutics. Science 297, 353-356.
[3] Brouwers N, Sleegers K, Van Broeckhoven C (2008) Molecular genetics of Alzheimer's disease: an update. Ann Med 40, 562-583.
[4] Hyman BT, Phelps CH, Beach TG, Bigio EH, Cairns NJ, Carrillo MC, Dickson DW, Duyckaerts C, Frosch MP, Masliah E, Mirra SS, Nelson PT, Schneider JA, Thal DR, Thies B, Trojanowski JQ, Vinters HV, Montine TJ (2012) National Institute on Aging-Alzheimer's Association guidelines for the neuropathologic assessment of Alzheimer's disease. Alzheimers Dement 8, 1-13.
[5] Montine TJ, Phelps CH, Beach TG, Bigio EH, Cairns NJ, Dickson DW, Duyckaerts C, Frosch MP, Masliah E, Mirra SS, Nelson PT, Schneider JA, Thal DR, Trojanowski JQ, Vinters HV, Hyman BT, National Institute on A, Alzheimer's A (2012) National Institute on Aging-Alzheimer's Association guidelines for the neuropathologic assessment of Alzheimer's disease: a practical approach. Acta Neuropathol 123, 1-11.
[6] Regland B, Gottfries CG (1992) The role of amyloid beta-protein in Alzheimer's disease. Lancet 340, 467-469.
[7] Joseph J, Shukitt-Hale B, Denisova NA, Martin A, Perry G, Smith MA (2001) Copernicus revisited: amyloid beta in Alzheimer's disease. Neurobiol Aging 22, 131-146.
[8] Rottkamp CA, Atwood CS, Joseph JA, Nunomura A, Perry G, Smith MA (2002) The state versus amyloid-beta: the trial of the most wanted criminal in Alzheimer disease. Peptides 23, 1333-1341.
[9] Lee HG, Casadesus G, Zhu X, Takeda A, Perry G, Smith MA (2004) Challenging the amyloid cascade hypothesis: senile plaques and amyloid-beta as protective adaptations to Alzheimer disease. Ann N Y Acad Sci 1019, 1-4.
[10] Robinson SR, Bishop GM (2002) Abeta as a bioflocculant: implications for the amyloid hypothesis of Alzheimer's disease. Neurobiol Aging 23, 1051-1072.
[11] Robinson SR, Bishop GM, Munch G (2003) Alzheimer vaccine: amyloid-beta on trial. Bioessays 25, 283-288.
[12] Turner PR, O'Connor K, Tate WP, Abraham WC (2003) Roles of amyloid precursor protein and its fragments in regulating neural activity, plasticity and memory. Prog Neurobiol 70, 1-32.
[13] Brayne C, Richardson K, Matthews FE, Fleming J, Hunter S, Xuereb JH, Paykel E, Mukaetova-Ladinska EB, Huppert FA, O'Sullivan A, Dening T (2009) Neuropathological correlates of dementia in over-80-year-old brain donors from the population-based Cambridge city over-75s cohort (CC75C) study. J Alzheimers Dis 18, 645-658.
[14] MRC-CFAS (2001) Pathological correlates of late-onset dementia in a multicentre, community-based population in England and Wales. Neuropathology Group of the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). Lancet 357, 169-175.
[15] Tanskanen M, Makela M, Notkola IL, Myllykangas L, Rastas S, Oinas M, Lindsberg PJ, Polvikoski T, Tienari PJ, Paetau A (2017) Population-based analysis of pathological correlates of dementia in the oldest old. Ann Clin Transl Neurol 4, 154-165.
[16] Hamasaki H, Honda H, Okamoto T, Koyama S, Suzuki SO, Ohara T, Ninomiya T, Kiyohara Y, Iwaki T (2017) Recent Increases in Hippocampal Tau Pathology in the Aging Japanese Population: The Hisayama Study. J Alzheimers Dis 55, 613-624.
[17] Savva GM, Wharton SB, Ince PG, Forster G, Matthews FE, Brayne C (2009) Age, neuropathology, and dementia. N Engl J Med 360, 2302-2309.
[18] Hunter S, Brayne C (2012) Relationships between the amyloid precursor protein and its various proteolytic fragments and neuronal systems. Alzheimers Res Ther 4, 10.
[19] Wang R, Sweeney D, Gandy SE, Sisodia SS (1996) The profile of soluble amyloid beta protein in cultured cell media. Detection and quantification of amyloid beta protein and variants by immunoprecipitation-mass spectrometry. J Biol Chem 271, 31894-31902.
[20] Hunter S, Brayne C (2017) Do anti-amyloid beta protein antibody cross reactivities confound Alzheimer disease research? J Negat Results Biomed 16, 1.
[21] Alafuzoff I, Pikkarainen M, Arzberger T, Thal DR, Al-Sarraj S, Bell J, Bodi I, Budka H, Capetillo-Zarate E, Ferrer I, Gelpi E, Gentleman S, Giaccone G, Kavantzas N, King A, Korkolopoulou P, Kovacs GG, Meyronet D, Monoranu C, Parchi P, Patsouris E, Roggendorf W, Stadelmann C, Streichenberger N, Tagliavini F, Kretzschmar H (2008) Inter-laboratory comparison of neuropathological assessments of beta-amyloid protein: a study of the BrainNet Europe consortium. Acta Neuropathol 115, 533-546.
[22] Thal DR, Sassin I, Schultz C, Haass C, Braak E, Braak H (1999) Fleecy amyloid deposits in the internal layers of the human entorhinal cortex are comprised of N-terminal truncated fragments of Abeta. J Neuropathol Exp Neurol 58, 210-216.
[23] Kummer MP, Heneka MT (2014) Truncated and modified amyloid-beta species. Alzheimers Res Ther 6, 28.
[24] Moore S, Evans LD, Andersson T, Portelius E, Smith J, Dias TB, Saurat N, McGlade A, Kirwan P, Blennow K, Hardy J, Zetterberg H, Livesey FJ (2015) APP metabolism regulates tau proteostasis in human cerebral cortex neurons. Cell Rep 11, 689-696.
[25] Hunter S, Martin S, Brayne C (2016) The APP proteolytic system and its interactions with dynamic networks in Alzheimer's disease. Methods Mol Biol 1303, 71-99.
[26] McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Jr., Kawas CH, Klunk WE, Koroshetz WJ, Manly JJ, Mayeux R, Mohs RC, Morris JC, Rossor MN, Scheltens P, Carrillo MC, Thies B, Weintraub S, Phelps CH (2011) The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement 7, 263-269.