Protein Phosphorylation is a Key Mechanism in Alzheimer’s Disease

INTRODUCTION

Post-translational modifications like protein phosphorylation and ubiquitination act as the gatekeepers of cellular processes. It is now widely accepted that aberrant protein phosphorylation is involved in the pathogenesis of a wide range ofdiseases among them neuropathological disorders [1, 2]. Neuropathologies are complex and anomalies can occur at the molecular and cellular level, affecting strategically distinct brain regions. There are hundreds of diseases of the nervous system, among them Alzheimer’s disease (AD), amyotrophic lateral sclerosis, Creutzfeldt-Jakob disease, familial adenomatous polyposis, frontotemporal dementia, Huntington’s disease, Machado–Joseph disease, and Parkinson’s disease.

AD has been associated with abnormal phosphorylation of specific key proteins, as has been well documented for tau (microtubule-associated protein tau) [3, 4]. Of note, protein phosphorylation is recognized as the major post-translational modification through which numerous physiological processes are regulated. However, in AD the exact role played by abnormal protein phosphorylation in the disease etiology, is not completely understood. It is particularly relevant that protein phosphorylation, via complex signaling cascades, can regulate neuronal plasticity, neurotransmission, and consequently compromise memory and learning. Ultimately protein phosphorylation can contribute to disease associated processes [1, 6]. Signaling cascades are precisely controlled by dynamic reversible protein phosphorylation and depend on the precise balance between protein kinase and protein phosphatase (PP) activities. Human genome sequencing predicts more that 500 protein kinases and around 150 protein phosphatase genes [7, 8]. The kinases are subdivided into two families, the Ser/Thr-kinases with 428 members and the Tyr-kinases with 90 members [9, 10]. For the phosphatases, four distinct families have been described, the protein Tyr-phosphatases (PTPs) [11], the specificity phosphatases dual specificity protein phosphatases [12] and two families of protein phosphatases, serine and threonine (PSPs) [13], of these, around 107 are PTPs [11] and around 40 are PSPs [14, 15]. The apparent lower number of PSPs is compensated for by an array of regulatory and targeting subunits, which confer target and substrate specificity, with more than 200 of these phosphatase interacting proteins (PIPs), thus far identified [16–20]. Originally kinases were considered the regulators of signal transduction-mediated events; however, it is now recognized that protein phosphatases have an equally important role. Together, kinases and phosphatases represent a central mechanism regulating major cellular events. Brain is the human tissue expressing the highest levels of protein kinases and phosphatases [21, 22]. Consistently abnormal protein phosphorylation states are a hallmark in specific disease conditions, and consistently altered phosphatase and kinase activities have been reported in the brains of AD patients [23].

ALZHEIMER’S DISEASE SIGNS, HALLMARKS, AND PROTEIN PHOSPHORYLATION

AD is a neurodegenerative multifactorial disorder first described in 1906 by the German pathologist Alois Alzheimer. Symptoms include progressive memory loss and decline in other cognitive functions. Affected individuals suffer personality changes, such as behavioral and emotional disturbances, and the capacity to execute social and personal activities declines. Initial memory impairment evolves to disorientation, speech abnormalities, hallucinations, apraxias, among other signs. At later stages, patients become completely dependent. Typically other diseases can become lethal, in particular pneumonia, which is the principal cause of death in ADpatients [24].

The majority of AD cases are sporadic (SAD), but rare, familial, early-onset autosomal dominant forms of AD (FAD) have been described. Mutations or polymorphisms in genes encoding AβPP (amyloid precursor protein), PS1/PSEN1 (presenilin-1), and PS2/PSEN2 (presenilin-2) [25] are related to FAD. Although more than 25 mutations in the AβPP gene have been described, these, as already mentioned, are related to FAD, but are rare when all AD patients are considered. Nonetheless, these mutations suggest that abnormal AβPP processing can contribute to the disease condition. Of note, these mutations occur in the flanking region of the Aβ domain. Duplication of the AβPP gene also appears to contribute to the early-onset of the disease, explaining why individuals with Down’s syndrome/trisomy 21 (chromosome encoding the AβPP gene) have an increased risk of developing AD [26]. Missense mutations in AβPP represent less than 0.1% of all the AD cases. Missense mutations, in the secretases, PS1 (chromosome 14) and PS2 (chromosome 1) genes are related to early onset (between 40 and 60 years) and aggressive AD forms. Studies revealed that mutations in these two genes shift the substrate specificity of PS1 and PS2 from Notch to AβPP; resulting in increased Aβ production [27]. Additionally, the presence of two alleles APOE4 (allele 4) is one of the most important genetic risk factors for SAD [28]. The presence of APOE4 precipitates the onset of the disease. Contrastingly, APOE2 (allele 2) seems to have a protective effect against the disease [29]. Non-genetic factors, like regular use of nonsteroidal anti-inflammatory drugs, wine and coffee consumption, and regular physical activity have been associated with a lower risk of developing AD; whereas aging and low educational levels, were associated with increased risk of AD incidence [30]. Elevated serum levels of cholesterol and LDL also correlated with increased amounts of Aβ in the brain and subsequently increase the risk of developing AD [31].

Senile plaques (SPs), intraneuronal neurofibrillary tangles (NFTs), and neuropil threads (abnormal neurites) have been extensively described as AD hallmarks. These lesions are found in specific brain regions associated with memory and learning processes namely the neocortex, entorhinal cortex, and hippocampus. The presence and distribution of NFT, SP, and synaptic degeneration correlate with the degree of cognitive decline [32, 33].

Of particular relevance, both histopathological hallmarks of proteinacious deposits in AD; the NFTs and SPs, can be correlated to phosphorylation events. NFTs are a consequence of hyperphosphorylated tau, whereas SPs have at their core the Aβ peptide, whose production can be modulated by the phosphorylation state of AβPP [34–37]. A third hallmark, which should be considered in AD, is synaptic dysfunction that appears to precede the deposition of NFTs and SPs [38]. Synaptic signaling cascades inevitably involve protein phosphorylation mediated events, thus the third hallmark, like NFTs and SPs involves anomalous phosphorylation processes. Consistently, postmortem analysis of AD brains reveals a tendency toward decreased phosphatase levels, and increased kinases levels, together favoring conditions for hyperphosphorylated proteins [6, 39].

tau PROTEIN AND ALZHEIMER’S DISEASE

NFTs and neuropil threads are composed of aggregated abnormal paired helical filaments (PHFs) of hyperphosphorylated tau protein. tau is a microtubule-associated protein, essential in microtubule dynamics, neurite outgrowth, and axonal transport. It is regulated in a phosphorylation dependent manner. In its hyperphosphorylated state, tau sequesters normal tau and other microtubule-associated proteins, leading to microtubule destabilization and polymerization. Consequently, axonal transport and neurotransmission are compromised, affecting particularly synapses and contributing to a decline in cognitive functions. Moreover, hyperphosphorylated tau self assembly [40] leads to small deposits (pretangles) that adopt a β-sheet conformation in PHFs. In turn, these assemblies form into large NFTs, whereby tau undergoes additional modifications, namely, truncations, glycations, and cross-linking by transglutaminases [41, 42]. Therefore, signaling cascade alterations, leading to abnormal protein phosphorylation or aggregation can potentiate NFT formation and neuronal degeneration.

Although tau is involved in many cellular functions; among the most important is perhaps tubulin polymerization. Protein phosphorylation regulates the binding of tau to tubulin, whereby phosphorylation of the former alters its conformation and causing it to detach from microtubules [43, 44]. Cross talk between tau and AβPP is an intense area of research, this may involve protein phosphorylation, particularly if one considers that Aβ has been shown to influence tau phosphorylation [45].

tau possesses a large number of potential phosphorylation sites (Table 1) at serine, threonine, and tyrosine residues [46, 47]. For the longest brain tau isoform (441 amino-acids) more than 80phosphorylation sites have been described ([39] and Table 1). tau phosphorylation at normal physiological conditions controls a variety of processes such as microtubule binding and microtubule assembly [48], neurite outgrowth [49], axonal transport [50], and cell sorting [51]. The proline-directed protein kinases (PDPK) are the major proteins involved in tau protein phosphorylation. These include glycogen synthase kinase 3 (GSK-3), mitogen activated protein kinase (MAPK), tau-tubulin kinase, cyclin-dependent kinases such as CDK2 and CDK5, and stress-activated kinases (SAP kinases). Kinases from the Non-PDPK group with activity toward tau protein, include the microtubule-affinity regulating kinase (MARK), Ca^2 +/calmodulin-dependent protein kinase II (CaMK-II), cyclic-AMP-dependent kinase (PKA), casein kinase II (CK2), and protein kinase C (PKC) [23, 53]. Of note, tau extracted from the brain of AD patients exhibited 45 phosphorylation sites, 29 serines (S), 13 threonines (T), and 3 tyrosines (Y) (Fig. 1), the majority of which can be modified by GSK-3 [54]. However, if one globally considers the findings published, identifying the tau phosphorylation sites, many more have been documented and these are summarized in Table 1.

OPEN IN VIEWER

Fig.1

tau phosphorylation sites identified in brains of AD patients. Diagrammatic representation of the most common tau residues (Y in orange; S in purple and T in blue) shown to be phosphorylated in the AD brain. The kinases responsible for the respective phosphorylations are also indicated (when the information was available). The KXGS motifs are indicated *259-QIGS-262, **290-QCGS-293, ***321-QCGE-324, and ****353-QIGS-356.

Phosphorylation within the microtubule-binding domain, at the KXGS motifs (red asterisks Fig. 1), reduces the binding of tau to microtubules. More specifically the phosphorylation at residues S262 and S356 ‘breaks’ the binding between tau and microtubules [55]. Likewise, phosphorylation at T231 by GSK-3β also plays a role in diminishing the ability of tau to bind to microtubules [56]. Given that so many kinases are involved in tau phosphorylation, tau might be primed by a given kinase before subsequent phosphorylation by another kinase [54]. For instance, an example from Down’s syndrome, identified that DYRK1A phosphorylates tau at several sites including T181, S199, S202, T205, T212, T217, T231, S396, S400, S404, and S422; these phosphorylations can prime tau for further phosphorylations by GSK-3 at T181, S199, S202, T205, and S208 but not by CDK5 and PKA [57–59]. Sequential phosphorylations have long been shown to be relevant and a key event in modulating protein functions [60].

AβPP AND ALZHEIMER’S DISEASE

AβPP is one of three members of a gene family, which includes APLP1 (Amyloid-like protein 1) and APLP2 (Amyloid-like protein 2) where only AβPP contains the Aβ domain. This protein is a ubiquitously expressed type I transmembrane glycoprotein, encoded by a single gene on chromosome 21q21. Multiple isoforms exist resulting from alternative splicing of exons 7, 8, and 15 of the AβPP mRNA. There are three major AβPP isoforms (AβPP695, AβPP751, and AβPP770), consisting of 695, 751, and 770 amino acids, respectively [61]. The 695 isoform is the only one that lacks a kunitz protease inhibitor (KPI) domain in its extracellular domain and is the predominant form in neuronal tissues.

AβPP can be processed by a non-amyloidogenic and an amyloidogenic pathway. In the non-amyloidogenic pathway, AβPP is cleaved by α-secretase within the Aβ domain. This results in the shedding of nearly the entire ectodomain (releasing a secreted AβPP fragment, sAβPPα) and generation of a membrane anchored α-C-terminal fragment (C83). The latter is cleaved by the γ-secretase complex [62–64], releasing a non-toxic p3 peptide and the AβPP intracellular domain (AICD) polypeptide fragment. AβPP cleavage by activation of α-secretase is the major and ubiquitous pathway of AβPP metabolism in most cells. Estrogen, testosterone, various neurotransmitters, growth factor and protein kinase C (PKC) are able to regulate theα-cleavage pathway [65]. Zinc metalloproteases like TACE/ADAM17, ADAM9, ADAM10 and MDC-9 can all cleave AβPP at the α-secretase site [66].

In the amyloidogenic pathway, particularly enriched in neurons, AβPP is first cleaved by β-secretase, releasing sAβPPβ and β-C-terminal fragment (C99). Subsequently this fragment is cleaved by the γ-secretase complex; giving rise to the Aβ peptide and the AICD fragment. β-secretase can also cleave AβPP within the Aβ domain to produce a C89 truncated Aβ species [67]. BACE1 (Beta-secretase 1) and BACE2 (Beta-secretase 2) are two enzymes capable of cleavage at the β-site. BACE 1is the major β-secretase in the brain and is the key rate-limiting enzyme that initiates Aβ formation. Overexpression of BACE1 in cell culture has been shown to increase the amount of β-secretase cleavage products [68]. BACE2 shows similar substrate specificity but is not as highly expressed in the brain [69].

AβPP can be phosphorylated at multiple sites in both extracellular and intracellular domains. In neuronal cells, AβPP695 is phosphorylated at serine, threonine, and tyrosine residues. In the intracellular domain 8 putative phosphorylation residues have been described (Fig. 2): Y653, T654, S655, T668, S675, Y682, T686, and Y687 [70–72]. In the extracellular domain, two phosphorylatable residues have been identified, S198 and S206 [73]. The consequence and physiological relevance of phosphorylation events at each of these residues is not clearly understood. However, correlations between phosphorylation at specific residues and AβPP fate are starting to be described. For example, phosphorylation at Y687 is relevant for AβPP endocytosis and subsequent Aβ production [36]. Likewise, S655 phosphorylation determines the fate of AβPP with respect to lysosomal targeting or retrograde transport to the Golgi via a retromer mediated process [74], and T668 phosphorylation is important to regulate AβPP binding to other proteins [75].

OPEN IN VIEWER

Fig.2

AβPP phosphorylation sites identified in brains of AD patients. Diagrammatic representation of AβPP (isoform 695 numbering) and the residues (Y in orange, S in purple and T in blue) shown to be phosphorylated in the AD brain. The kinases responsible for the respective phosphorylations are also indicated (when the information was available). Several motifs are indicated; GFLD the growth-factor like domain, the E1 and E2 domain, CuBD the copper binding domain, the pentapeptide RERMS domain, CAPPD the central AβPP domain, the Aβ domain shown in red, the AβPP C-terminal fragment and the 3 C-terminal motifs *682-YENPTY-687, **667-VTPEER-672, and ***653-YTSI-656.

Several kinases have been shown to be relevant to AβPP biology, either by directly phosphorylating AβPP or by phosphorylating other relevant substrates. Among the first reported was protein kinase C (PKC) [76–79]. Direct activation of PKC by phorbol esters, which mimics the effects of diacylglycerol, resulted in increased sAβPP release and inhibited Aβ formation [76, 80–83]. AβPP can be phosphorylated on the C-terminal domain and a number of kinases fulfilling this function have been identified. PKC phosphorylates AβPP S655 [63], whereas CDK5 and CDC2 phosphorylate AβPP at T668 in neurons [84, 85]. T668 can also be phosphorylated in vivo by a number of protein kinases, namely GSK-3,JNK3/SAPK1b, [84, 87]. In vitro, the tyrosine kinases, TRKA and c-Abl, have also been shown to phosphorylate AβPP on Y682 [72, 88].

OTHER MAJOR PHOSPHORYLATED PROTEINS IN ALZHEIMER’S DISEASE

Several other phosphorylatable proteins of relevance to AD have been described, among them the secretases. The γ-secretase complex consists of at least four different proteins: Presenilins (PS), Nicastrin, APH-1, and PEN-2 proteins. In order to form an active γ-secretase complex, Nicastrin, a type I transmembrane glycoprotein, and APH-1, form a dimeric subcomplex to which PS binds. Subsequently, PEN-2 is incorporated into the complex and PS is cleaved into two stable fragments; the N-terminal fragment and a C-terminal fragment. The active γ-secretase is thus formed [89–91]. ERK1/2 appears to be an endogenous negative regulator of γ-secretase, probably via direct phosphorylation of Nicastrin [92]. The γ-secretase can also be activated by tumor necrosis factor α, which promotes JNK phosphorylation of Nicastrin and PS1 [93].

PS is responsible for the catalytic activity of the γ-secretase complex as described above. The former has two homologues, PS1 and PS2, mutations of which are the most common cause of early-onset FAD, as previously discussed. PS1 is an integral membrane protein, with nine predicted transmembrane domains, localized in the ER, Golgi, and plasma membrane [94]. PS1 can be phosphorylated by several protein kinases [95]. Phosphorylation of PS1 at S353 and S357 by GSK-3β may influence binding to β-catenin [96]. PS1 phosphorylation at S397 by GSK-3β regulates C-terminal fragment levels [97]. Physiologically distinct functions can occur via PKA-mediated phosphorylation, which strongly inhibits proteolytic processing of PS1 by caspase activity during apoptosis, reducing the progression of apoptosis [98]. To add further complexity to the system, studies have shown that PS1 can stimulate PI3K/AKT signaling, thus promoting phosphorylation and inhibiting GSK-3β with the effect of suppressing GSK-3β mediated tau overphosphorylation found in AD. It appears that in FAD mutations PS1-dependent PI3K/Akt activation is inhibited favoring GSK-3β mediated tau phosphorylation [99].

In the near future, many other phosphorylated proteins associated with AD are likely to be forthcoming. Recent work by Henriques et al. [100] showed that when cells were exposed to Aβ, the recovery of 73 phosphorylated proteins increased and that of 68 phosphorylated proteins decreased, in comparison to cell models not exposed to the toxic peptide. These candidates, include protein kinases and phosphatases, and provide a set of potential candidates for therapeutic and diagnostic strategies.

THE RELEVANCE OF PROTEIN PHOSPHATASES IN ALZHEIMER’S DISEASE

Given the dynamic nature of protein phosphorylation systems, it is perhaps not surprising that the association of PPs with AD has been well established and is increasing. Of note, altered PP expression levels have been reported in AD. Most of the phosphatase activity can be attributed to the Ser/ThrPPs [101].

In the human brain PP2A, PP5, PP1, and PP2B have all been shown to regulate tau phosphorylation, accounting for approximately 71%, 11%, 10%, and 7%, of tau dephosphorylation, respectively [102]. Thus, PP2A is the major tau phosphatase. In many cases, all the above-mentioned phosphatases can dephosphorylate the same tau residues, albeit with different efficiencies toward different sites. tau dephosphorylation by PP1, PP2A, and PP5 exhibited K(m) values around 8–12 microm, whereas that for PP2B was around five times higher [102]. In the AD brain PP1, PP2A, and PP5 phosphatase activity were decreased whereas PP2B increased. Among these the most significant is PP2A, as it can regulate phosphorylation at multiple tau sites. The relevance of PP2A was shown by using metabolically competent rat brain slices as a model, and inhibiting PP2A with okadaic acid (OA), induced AD-like hyperphosphorylation and the accumulation of NFTs [103].

PP2A is also involved in Aβ production. OA induced inhibition of PP2A in N2a cells, increased AβPP phosphorylation at T668 and the secretion of both sAβPPα and sAβPPβ were enhanced. However in some of these experiments it is not clear whether these effects are preferentially mediated by PP1 [104] and/or PP2A [105].

It is particularly noteworthy that PP1 is highly enriched in dendritic spines [22]. This observation led to the discovery of spinophilin, which binds PP1, targeting it to the post-synaptic density. Of note, synaptic loss is presently the best correlation for AD, and the best post-synaptic marker is PP1. Subsequently, many PP1 interacting proteins (PIP) have been identified [18, 19], some of which are brain specific or enriched. Three mammalian isoforms, expressed in virtually all tissues, have been described, PP1α, PP1β, and PP1γ1, [106]. However, their expression levels vary in different brain regions where it appears they are localized to specific subcellular compartments [21, 107]. A recently described PIP that is of specific neuronal relevance, is FE65. The latter which binds AβPP was shown to also bind PP1 [108], forming the complex AβPP:FE65:PP1, thus bringing the phosphatase to the proximity of AβPP and modulating dephosphorylation of the latter. In essence, FE65 is an important bridging protein between AβPP and PP1 and even more so, when one considers that a brain specific FE65 has also been described [109]. Of all the above mentioned phosphatases PP1 is an important phosphatase in AD related pathology, further strengthened by the finding that Aβ can inhibit PP1 [110] and PP1 can bind both AβPP and tau, as discussed below.

THE AMYLOID CASCADE HYPOTHESIS

As already discussed SPs are extracellular deposits found in AD brains, mainly composed of Aβ peptide aggregates, although Aβ is associated with AD it is also found in normal aging. SPs are likewise observed in individuals with mild cognitiveimpairment at a higher level than in non-AD older adults [111] and it is a strong predictive factor of conversion to AD. Aβ deposition spreads from regions with early deposits to regions which receive their neuronal input. A strong correlation exists between the degrees of Aβ deposition and clinical symptoms. Indeed, several studies support the amyloid cascade hypothesis which defends that Aβ accumulation in the central nervous system is the first event that initiates the pathogenic cascade culminating in neurodegeneration and neuronal death typical of AD [112]. Clearly, in this hypothesis (Fig. 3), AβPP and PS mutations contribute to higher Aβ production favoring disease onset [113]. Longer forms of this peptide increase its aggregatory properties and seeding is an important aspect for peptide aggregation [114]. Aβ accumulation involves oligomerization, aggregation processes, and amyloid deposition; these events are related with the above mentioned histopathological (SPs) and clinical manifestations of the disease [42]. In the amyloid cascade hypothesis, the following three aspects are significant: Aβ deposition is central to the disease process; Aβ deposition is an early event; mutations in AβPP and other players involved in the cascade contribute to FAD and SAD.

OPEN IN VIEWER

Fig.3

Diagrammatic representation of AD pathology related hypotheses. Overview of the most documented cascade hypotheses for the onset of AD pathology. PS1, Presenilin 1; PS2, Presenilin 2; NOS, nitric oxide synthase expression; NO, nitric oxide; ROS, reactive oxygen species; IDE, insulin degrading enzyme; SPs, senile plaques; NFTs, neurofibrillary tangles. Dashed black line denotes that Aβ can influence tau hyperphosphorylation.

According to this hypothesis, genetics, age, and environment factors can all lead to an imbalance in Aβ production and clearance. Among these, the greatest genetic risk factor is APOE4, which appears to contribute to late onset cases [28]. Also central to the hypothesis are the secretases, particularlyγ-secretase, which regulates levels of Aβ production. It is no less important that the physiological responses of these molecular players can be regulated by their phosphorylation state. For instance, phosphorylation of γ-secretase can modulate its activity [92, 93] and AβPP phosphorylation can regulate the levels of Aβ production [36]. However, the amyloid cascade hypothesis cannot explain all of the histopathological hallmarks associated with AD, like for instance NFTs, although some studies suggest that Aβ itself can render in NFT formation [115, 116]. Additionally if one considers that, as explained above, Aβ can inhibit PP1 and or PP2A [110], this would subsequently result in tau hyperphosphorylation (dashed black arrow, Fig. 3) and thus the AD related hallmarks can all be explained by the amyloid cascade hypothesis (Fig. 3). Furthermore, Aβ can activate MAPK and GSK-3β that can, in turn, phosphorylate tau [117]. This is consistent with the dual pathway hypothesis as described below. In fact Aβ can influence many other molecular events, such as those related to cytoskeletal organization [118, 119], and this can also potentially contribute to AD.

Associated with Aβ deposition a variant of the amyloid cascade hypothesis has evolved [120]. The hypothesis defends that the formation Aβ oligomers (oAβ) is neurotoxic and can cause synaptic damage. The latter is a third AD hallmark which precedes SPs and NFTs deposits. This is attractive given that amyloid plaque deposition does not necessarily correlate with AD onset or neuronal loss [121, 122]. This hypothesis places oAβ as a neurotoxic agent with the capacity to mediate toxic effects, and whereas SPs are neurotoxically inert oAβ is not. It is noteworthy that oAβ can affect several important neuronal receptors such as the insulin receptor and nicotinic receptors. For example, it can induce the reduction or abolition of long term potentiation, and when injected into rodent brains it can induce impaired cognitive function [123–129]. Furthermore the oAβ species may have a role in the seeding of amyloid plaques (reviewed in [130]).

ALTERNATIVE AD HYPOTHESIS

Several other hypotheses have been proposed to explain AD (reviewed in [131]). Among them the dual pathway hypothesis [132] that attempts to refine the amyloid cascade hypothesis, particularly with respect to SAD. The dual pathway hypothesis proposes that upstream factors may drive both Aβ and tau-mediated pathologies. With this in mind, downstream treatments targeting both Aβ and tau pathology will ultimately be of therapeutic benefit. In this hypothesis, Aβ and tau deposits increase, although the authors of the model defend that a decrease in Aβ clearance could be operating. It is worth reinforcing, as stated above, that protein phosphorylation appears to be a key event in SPs and NFTs formation and thus of relevance to this model.

Another hypothesis is the mitochondrial cascade hypothesis. The latter proposes that age-related mitochondrial dysfunction will lead to AD [133–135]. In fact several studies have confirmed mitochondrial damage in AD patients brains’ [136, 137]. Consistently AD cybrid cells have increased Aβ production [138] and reactive oxygen species [139]. These AD cybrids include platelet mitochondria from an AD subject in a cell line (p0) blocked for mitochondrial DNA replication [140]. It appears that mitochondria trigger the abnormal onset of neuronal degeneration and cell death associated with AD [141]. A range of anomalies have been detected in AD patients, including mitochondrial dysfunction, increased oxidative stress, and apoptotic neurons. Besides providing the cell with ATP, mitochondria play a significant role in regulating cell death and thus can contribute to the above mentioned pathological AD hallmarks. Mitochondria perform electron transport via the electron transport chain (ETC) complex (I, II, III, and IV). This enables the harnessing of energy from mobilized free electrons, and drives proton translocation. An additional complex (V) permits protons to re-access the matrix, coupling the energy from this proton flux to ADP phosphorylation. Thus protein phosphorylation is also relevant in mitochondrial function [142]. Furthermore mitochondrial function was shown to determine tau phosphorylation. In fact, tau phosphorylation is promoted in AD fibroblasts exposed to an ETC uncoupler [143]. To summarize, the mitochondrial cascade hypothesis proposes that a mitochondrial deficiency in AD brains underpins an increase in Aβ production and can even promote tau phosphorylation. However, despite genome wide association studies, genes encoding mitochondrial proteins have not yet been found [144], and in fact this hypothesis cannot explain all the panoply of AD pathology. Furthermore, placing mitochondria at the apex of AD pathology remains controversial.

The metabolism cascade hypothesis [145, 146] proposes that the underlying cause of AD is cerebral glucose hypometabolism. The authors developed a rat model in which injecting streptozotocin intracerebroventricularly [147] associated with decreased glucose/energy brain metabolism and learning and memory deficits. Subsequent work, showed that insulin signaling in the brain is significantly impaired in AD [148], leading to the term that AD is a ‘Type 3 Diabetes’. Both insulin and IGF-1 (insulin/insulin-like growth factor I) stimulate Aβ release from neurons, and IGF-I promotes brain amyloid clearance. In addition, insulin and IGF-I levels are altered in AD and cell sensitivity toward insulin and possibly IGF appears to be reduced in these individuals. Insulin is likely to exert two distinct effects on brain Aβ. It stimulates neuronal Aβ release and at the same time contributes to extraneuronal Aβ accumulation by competing for insulin degrading enzyme (IDE). The latter enzyme can also regulate Aβ extracellular levels. Therefore, the net action of insulin is to increase brain Aβ. It appears that preserving synaptic connectivity requires insulin signaling and the latter may also play a role in neuronal stem cell activation and neuronal ‘resilience’. The oAβ can bind to and antagonize components of the insulin signaling pathway resulting in increased activity of GSK-3β, a known tau kinase [149]. This is an attractive hypothesis; involving complex phosphorylation mediated responses, cognitive dysfunction and insulin resistance, and is presently the focus of much research [150].

Microglia, astrocytes, and even neurons appear to be involved in the inflammatory processes associated with AD, giving rise to the inflammatory cascade hypothesis. Aβ was shown to activate microglia leading to increased cell surface expression of the major histocompatibility complex II and increased secretion of the pro-inflammatory cytokines; interleukin-1β, interleukin-6, tumor necrosis factor α, and the chemokins: interleukin-8, macrophage inflammatory protein-1 α, and monocyte chemo-attractant protein-1 [151]. Many other cytokines can be released in response to Aβ.

Additionally, microglia can participate in Aβ degradation by releasing IDE. At sites of Aβ deposition, astrocytes will cluster and secrete a range of factors like interleukins, prostaglandins, coagulation factors, and protease inhibitors. Furthermore, Aβ induced a phagocytic response in microglia and nitric oxide synthase expression, leading to increased nitric oxide and neuronal damage. In turn, nitric oxide can inhibit IDE activity [152]. The nonfibrillar Aβ_1 - 42 form is the predominant form found in diffuse plaques. Astrocytes in contact with diffuse plaques were shown to accumulate Aβ_1 - 42, but also cell debris, from degenerated synapses and dendrites. Microglia activation in AD brains appears to be involved not only in Aβ degradation (“good phagocytic phenotype”), but also in the production and release of reactive oxygen species and pro-inflammatory cytokines and chronic microgliosis (“bad phagocytic phenotype”) [153]. In turn inflammatory mediators can also affect Aβ deposition and even tau phosphorylation levels (Domingues et al., submitted). These alterations appear to contribute to neuronal dysfunction and death and therefore to disease progression in a vicious cycle.

Neurons themselves can express high levels of classical pathway complement and pro-inflammatory products that trigger inflammatory processes. In essence, the complement system, cytokines, chemokines, and acute phase proteins appear to contribute to the inflammatory response in AD. Similar to other hypothesis mentioned above, protein phosphorylation can be an important factor. For instance, in FAD and SAD, GSK-3 over-activity accounts for memory impairment, tau hyperphosphorylation, increased Aβ production, and inflammatory responses. Whether neuroinflammation is a primary cause or secondary effect in AD remains to be elucidated [154].

The observation that AD brains have a disorganized and reduced capillary and vascular network [155–157] lead to the vascular hypothesis. The vascular hypothesis proposes that AD develops when advancing age and the presence of vascular risk factors converge. Cerebral microvascular pathology and cerebral hypoperfusion can potentially trigger cognitive and degenerative changes in AD. It is noteworthy that AD risk factors include hypertension and diabetes, both exhibiting significant vascular morbidities [158]. Vascular damage and Aβ clearance do colocalize but more work is needed to establish if this is a chance or a neuropathological link, and for one to determine if amyloid plaques lead to vascular damage, or if vascular damage results in plaque formation. Additionally, according to this hypothesis tau hyperphosphorylation and NFTs development may be triggered either by secondary consequences of cerebrovascular damage (e.g., hypoperfusion/hypoxia insult), by Aβ-mediated neurotoxicity or a concomitant action of both events [159].

The cell cycle re-entry hypothesis is a general hypothesis whereby an age-related increase in neuronal DNA damage underlies neurodegenerative diseases. Given that neurons are post-mitotic cells they need to sustain their genomic integrity for life. Of note, mitogen kinases, which play a key role in cell cycle control, have increased expression in the AD brain [160]. In fact, AβPP appears to have a role in activating neuronal cell cycle proteins and a failure in the regulation of this pathway occurs in neurons in AD brains. Also supportive of this hypothesis is the interaction of AβPP with the adaptor protein BP1 (amyloid precursor protein AβPP binding protein 1/AβPP-BP1), a cell cycle protein that regulates mitotic transition from S- to M-phase. Phosphorylation of AβPP at T668 is required for protein interaction. The complex AβPP: AβPP-BP1 activates a pathway leading to the conjugation of NEDD8 (Neural Precursor Cell Expressed, Developmentally Down-Regulated 8), which is a ubiquitin-like protein playing an important role in cell cycle control and embryogenesis. NEDD8 mediated neddylation of cullins enhances ubiquitin ligase activity, promoting polyubiquitination and proteasomal degradation of cyclins and other regulatory proteins. The neddylation pathway promotes AβPP-mediated cell cycle entry and apoptosis which appears to be relevant in FAD [161]. Of note, overexpression of AβPP-BP1 pushes neurons into the S-phase causing DNA replication and expression of the cell cycle markers CDC2 and cyclin B1 [162–164]. It has also been reported that phosphorylation of AβPP at T668, known to occur during the G2/M phase, is required for the interaction between AβPP and AβPP-BP1 [165]. It would appear that cell cycle re-entry is pivotal in AD, given that these markers preclude the appearance of NFTs and SPs [166]. In other words cell cycle re-entry may be a causal factor in AD [167].

Another interesting finding is that specific antibodies raised to PHF of tau purified from AD brain, cross-reacted with epitopes in dividing cells [168]. Other lines of research showed that infecting differentiated neurons with oncogenes c-myc and ras, promoted them to initiate division and this resulted in DNA duplication and increases in anti-phospho-tau immunoreactivity [169]. Of note, the concept that tau phosphorylation promotes tangle formation in neurons attempting to re-enter the cell replication cycle has been well documented. Also consistent with the cell cycle re-entry hypothesis, is the finding that in postmitotic neurons, DNA repair defects can result in neurological abnormalities including neurodegeneration [170].

As briefly discussed above, there is significant evidence supporting the different AD hypotheses. However, a physiologically relevant process in all cases appears to be protein phosphorylation. Thus, by addressing protein phosphorylation dependent processes one may identify potential cues of diagnostic and therapeutic potential.

tau INTERACTIONS AND PHOSPHORYLATION MEDIATED EVENTS

Signaling pathways involve a dense network of protein:protein interactions, and the reaction rates depend on protein concentrations and associations/dissociations, among other factors. Protein phosphorylation can modulate the nature and the strength of protein:protein interactions, consequently regulating protein binding and the signaling pathways. Protein phosphorylation at or near a binding site may directly affect the binding energy of the complex. On the other hand, phosphorylation at a site outside a binding domain may cause a long-range conformational change. This allosteric mechanism can affect the binding of interacting proteins, as observed for glycogen phosphorylase [171, 172]. Recent advances in proteomics and the implementation of a systems biology approach have unraveled interactomes of many proteins. It is clear that protein:protein interactions are central to cellular functions and anomalies thereof can contribute to pathological conditions, among them AD. Thus analyzing the interactome of proteins relevant to this pathology will provide fundamental information regarding the phosphorylation dependent complexes formed and subsequently how these might contribute to the diseaseprocess.

The tau protein is a dipole, possessing two domains of opposite charge [173] and these are important in determining internal folding and aggregation [174], as well as the protein’s interaction with microtubules and other binding partners (Figs. 1 and 4). The C-terminus binds to microtubules [175] and the region of 150–240 residues is the ‘proline-rich domain’ that can serve as a target of proline-directed kinases as well as binding sites for proteins with SH3 domains [174]. The N terminus does not bind microtubules; it is the ‘projection domain’ (Fig. 1) [176] and it interacts with other cytoskeletal elements, mitochondria and the neuronal plasma membrane[177–179].

OPEN IN VIEWER

Fig.4

Merged protein:protein interaction network for tau and AβPP. Human interactors for the AβPP (gene APP) and the tau (gene MAPT) were retrieved from the public databases (MINT and IntAct), further interactions were included as follows: Domingues et al. [203] for RANBP9, Sumioka et al. [207] for YWHAG, Zheng et al. [208] for APPBP2, and for the tau interactors: PIN1 [214], HSP90 [183], HSP70 and HSC70 [184], FKBP52 [185] and SNCA [238]. Two nodes from the mouse interactome were also included; in one case the mouse tau bound to the human PPP2R2 and in the other to the mouse Ppp1ca. The network was developed using Cytoscape 3.4 and the following gene ontologies for molecular function are indicated: GO 16301 kinase activity (17 red nodes); GO 16791 phosphatase activity (2 green node); GO 51219 phosphoprotein binding (3 nodes with a blue circumference); GO 4860 protein kinase inhibitor activity (3 nodes with a pink circumference); GO 19903 protein phosphatase binding (10 nodes with a green circumference). Gene ontologies were retrieved using the BINGO ‘plug in’, and the PP1 binding proteins from [18, 19].

The tau interactome revealed 35 interacting proteins (Fig. 4), which were retrieved from the public databases IntAct and MINT as well as from specific reports, as indicated below. The output from the databases refers the gene, but for consistency in the text, the interactions are discussed at the protein level. The tau protein:protein interaction network has (Fig. 4); 12 nodes (red) with kinase activity, two that bind phosphoproteins (blue circumference), one with phosphatase activity (green node), and four nodes (green circumference) that interact with protein phosphatases. Analysis of the tau interacting proteins (Fig. 4) clearly reveals that tau binds to AβPP directly. These highlighted nodes represent proteins that bear some of the gene ontologies most overexpressed in the tau interactome, and are also relevant to protein phosphorylation events. The protein:protein interaction network reflects interactions for the human proteins; however, two examples from the mouse tau interactome were included. The PP2R2A (serine/threonine-protein phosphatase 2A 55 kDa regulatory subunit B alpha isoform), because the mouse tau bound to the human PP2R2A, and PP2 is the most important tau phosphatase, and the mouse interaction tau:Ppp1ca was also included for the same reason, many tau residues are dephosphorylated by PP1.

Among the 12 tau interactors having kinase activity (Fig. 4 red nodes), the expected kinases like GSK-3, CDK5, and MARKs are present but among them is also FYN. tau interacts with FYN [180], which is regulated by tau phosphorylation [181]. It is plausible that tau may localize FYN to the postsynaptic compartment, leading it to dephosphorylate the NMDA receptor causing increases in Ca^2 +, and culminating in pathological neurotoxicity [182]. Thus one can draw direct parallels between the tau interactome and AD associated anomalies.

Given the relevance of molecular chaperons in regulating tau post-translational processing, it is not surprising that HSP90 [183], HSP70, and HSC70 [184] are found in the tau interactome (Fig. 4). HSP90 regulates GSK-3β promoting tau phosphorylation, potentially favoring hyperphosphorylation of the latter. Furthermore, HSP90 can, in combination with FKBP52, produce oligomeric tau preventing clearance and increasing its aggregation potential. FKBP52 is another tau binding protein [185]. HSP70 has a dual role with tau, stabilizing it to microtubules and promoting its degradation in association with CHIP (chaperone-associated ubiquitin ligase). HSC70 can likewise regulate tau’s microtubule association [184]. It appears that HSC70 is important in maintaining tau conformation as a protective mechanism preventing the tau self-assembly process associated with NFTs and a major histopathological hallmark in AD.

AβPP INTERACTIONS AND PHOSPHORYLATION MEDIATED EVENTS

AβPP can integrate into the plasma membrane (Fig. 1) and interact both intracellularly and extracellularly to regulate various signal transduction mechanisms. It appears that the N terminal domain binds to phosphoinositide-rich domains on the surface of hippocampal neurons [186]. The RERMS domain is also important from an interaction perspective, and is briefly discussed. Many AICD binding proteins have been identified, and some of these may present novel potential targets for therapeutic intervention [187, 188].

The N-terminus of AβPP encompasses a cysteine-rich globular domain (E1), and an acidic α-helix-rich domain (E2) [189]. Two distinct regions have been identified within the E1 domain, the growth factor-like domain (GFLD) and the copper/metal binding domain (CuBD) [190, 191]. The GFLD corresponds to amino acid residues 23–128, this was crystallized several years ago [191] with the copper/metal binding domain, the structure contains nine β-strands and one α-helix and is well conserved across the AβPP family. The disulfide bridge between the residues C98 and C105 stabilizes a β-hairpin loop, and is critical for neurite outgrowth [190] and MAP kinase activation [192]. Several basic residues within this domain give rise to the heparin binding domain (HBD) which is a hydrophobic pocket that could potentiate aprotein-binding site or a dimerization site [191]. Although the precise function of AβPP is still unclear it may act as a receptor, as a growth factor [191], or may bind to an extracellular matrix component [190]. The CuBD is adjacent to the HBD and can bind several metal ions [193]. On its C-terminal the E1 domain has an acidic region of unknown function, which is rich in glutamic acid and aspartic acid residues and contains a stretch of seven threonine residues [194].

This acidic region connects the E1 to the E2 domain. The E2 domain, also called the central AβPP domain (CAβPPD), can readily dimerize [195] and may be involved in AβPP self-association. This domain consists of six α-helices [196] and provides potential binding sites for other binding partners. The E2 domain contains the RERMS sequence, which may account for the growth-promoting properties of AβPP [197]. The E2 domain also has a highly conserved heparan sulfate proteoglycan-binding site [189, 198] and a number of putative metal-binding sites [199, 200]. Metal binding is likely to hold the E2 domain in a rigid conformation. The longer isoforms of AβPP (AβPP751 and AβPP770) contain a Kunitz-type protease inhibitor (KPI) domain and an Ox-2 antigen domain, only in the latter. In general terms, it appears that the AβPP N-terminus may have an important role in cell adhesion and extracellular interactions, but whether any of these events can be mediated by protein phosphorylation remains to be seen. From a phosphorylation mediated functional perspective, the AICD is the most interesting region [201, 202]. It contains at least three functionally important motifs (Fig. 1) enabling AβPP interaction with several binding-partners, 86 of which are represented in Fig. 4. The interactome was retrieved from the public databases IntAct and MINT, additional interactors were included as individually identified throughout the text. Given the pivotal role of protein phosphorylation in the different biological processes played by AβPP, it is perhaps not surprising that its interactome with 86 nodes includes seven nodes (red) with kinase activity, three with protein kinase inhibitor activity (nodes with pink circumference), two which bind phosphoproteins (blue circumference), one with phosphatase activity (green node), and seven nodes (green circumference) that interact with protein phosphatases.

Such protein interactions regulate AβPP/AICD function, localization, processing, and Aβ production [203, 204]. The highly conserved 682-YENPTY-687 AβPP motif (AβPP 695 numbering) is a sorting motif involved in clathrin-mediated endocytosis and found in many tyrosine receptor kinases, non-receptor tyrosine kinases, low-density lipoprotein-receptor related family proteins, and integrins [205, 206]. The YENPTY domain is recognized by proteins containing phosphotyrosine interaction domains, namely, the FE65 (Fig. 4 APBB1) family, c-Jun N-terminal kinase interacting protein, X11 family (Fig. 4 APBA1), SHC family (Fig. 4 SHC1 and SHC3). To the 667-VTPEER-672 AβPP motif, binds the protein 14-3-3γ (YWHAG) that is highly expressed in the brain, skeletal muscle, and heart [207]. Another important interaction is that with PAT1 (protein interacting with AβPP tail 1, also denoted APPBP2), a microtubule interacting protein, which binds to the 653-YTSI-656 motif [208]. The three above mentioned domains all contain T, S, and/or Y residues which can be phosphorylated, thereby modulating protein:protein interactions as discussed below.

AICD has an unstable conformation in solution and the interaction with a binding-partner will stabilize its structure. This property has been termed binding promiscuity. The stability of the AICD-ligand complex can be influenced by AICD phosphorylation and this can influence the binding of interacting proteins at the above-mentioned domains, or vice versa. For instance, T668 phosphorylation is essential for FE65 (APBB1) binding to the YENPTY AβPP domain, and probably involves GSK-3 mediated phosphorylation [209]. Subsequent work unraveled the existence of a trimeric complex (AβPP:FE65:PP1). The authors demonstrated that FE65 is the bridging protein in the complex formed, and therefore a PP1 interacting protein (Fig. 4 APBB1). This interaction correlated with AβPP T668 phosphorylation state, consistent with the role whereby PP1 was recruited and could dephosphorylate T668 [108]. The same group also showed that the phosphorylation state of Y687 could influence the binding of FE65. The AICD also interacts with RANBP9 via the NPXY internalization motif [203]. Subsequent work, in cell culture, showed a RANBP9-TIP60 interaction that is important for nuclear targeting.

The JNK pathway is another signaling cascade highly relevant to AD, which typically targets c-Jun, ATF2 and Elk-1, JNK3 (SAPK1b/MAPK10). JIP-1b (Fig. 4 MAPK8IP1) can recruit JNK to AβPP, and JNK3 causes phosphorylation of the latter on T668 [87, 210]. Furthermore JIP-1b binds the TPR domain of kinesin light chain, thus AβPP can associate with kinesin-I via JIP-1b. Hence it would appear that JIP-1b is a linker protein between kinesin-I motor protein and the cargo receptor AβPP [207].

COMMON NODES AND CROSSTALK BETWEEN tau AND AβPP

Considerable scientific effort has focused on identifying molecular factors common to the development of SPs and NFTs (Fig. 3). It is particularly noteworthy that upon merging the protein:protein interaction networks of tau and AβPP a sub-set of nodes common to both interactomes is evident. As for the individual interactomes, in the merged interactome, phosphorylation related gene ontologies are well represented. Of these 17 nodes (red) have kinase activity, three with protein kinase inhibitor activity (nodes with pink circumference), three that bind phosphoproteins (blue circumference), two with phosphatase activity (green node) and ten nodes (green circumference) that interact with protein phosphatases.

Six nodes were identified (Fig. 4), exhibiting a direct interaction between tau and AβPP. In fact, both of these proteins can bind to proteins involved in other neuropathological conditions, for instance SNCA (α-Synuclein) and PRNP (prion protein) (Fig. 4). The protein SNCA, found in Lewy bodies, is typically associated with Parkinson’s disease; however, it may also be involved in AD [211]. SNCA interacts with synaptic vesicles at presynaptic terminals, functionally resembling tau protein, to which it binds, priming it for kinases. In particular, SNCA forms a complex with tau and GSK-3β. Additionally SNCA interacts directly with tau and stimulates its phosphorylation by PKA [212]. GSK-3α is another shared node, and as discussed throughout this review, this is an important kinase able to phosphorylate tau and AβPP. Likewise YWHAZ (14-3-3 protein zeta/delta) binds to both tau and AβPP. This is an adaptor protein involved in the regulation of a wide range of signaling pathways, with many interactors and typically recognizes a phosphoserine or phosphothreonine motif. YWHAZ can simultaneously bind AICD and FE65 facilitating FE65-dependent gene transactivation by enhancing the association of AICD with FE65 [207]. In tau, the binding with YWHAZ promotes the interaction of the former with a number of kinases (reviewed in [213]). Thus YWHAZ acts as a facilitator.

PIN1 (peptidyl-prolyl cis-trans isomerase) is another node in the ‘common sub-set’. Lu et al. [214] reported tau:PIN1 binding but in the database searches a PIN1P1 (putative PIN1-like protein) has been identified, thus both nodes are included in Fig. 4. Further research is needed to clarify if these two nodes correspond to the same protein. PIN1 binds to hyperphosphorylated tau [214, 215], promoting its dephosphorylation at CDK5 phosphorylation sites, in particular S202, T205, S235, and S404 [216]. It is interesting to note that PIN1 can modulate PP2A activity, which is a major tau phosphatase. PIN1 appears to regulate not only the phosphorylation, but also the conformation of its substrates. PIN1 can catalyze prolyl isomerization of specific pS/T-P motifs both in CDC25C and tau facilitating PP2A mediated dephosphorylation [217]. The relevance with respect to AβPP stems from the finding that neurons exposed to Aβ, exhibit PIN1 activation, with a consequential dephosphorylation of tau on T231 [218]. PIN1 is likewise responsible for the transient regulation of S199, S396, S400, and S404 tau phosphorylation in response to Aβ. Furthermore PIN1 can regulate Aβ production by binding to AβPP when the latter is phosphorylated on T668 [219, 220]. PIN1 promotes AβPP turnover by inhibiting GSK-3 activity [221]. Likewise relevant to the cell cycle re-entry hypothesis was the finding that PIN1 can upregulate cyclin D1 expression. This can, in turn, facilitate the transition of neuronal quiescent cells to the G1 phase [215]. Given PIN1’s essential role in the G0/G1 transition it may provide a good target for novel therapeutic strategies.

Another common node identified was EGFR (Epidermal growth factor receptor; also know as HER-1 or ErbB-1); although intrinsically associated with cancer, it has similarly been associated with neurometabolic conditions, like AD and diabetes. In neurodegeneration, altered EGFR expression levels have been observed. EGFR appears to mediate the effects of EGF/TGF-α, associated with neuronal differentiation [222], survival [223–225], and glial proliferation [226] and appears to mediate Aβ-related toxicity [227]. PS1 regulates the EGFR pathway [228]. Mutations in the PS1 and PS2 genes contribute significantly to early onset FAD [229]. In fact, EGFR levels increased in fibroblasts that were deficient in both PS1 and PS2, but transfecting with PS1 reversed this effect [228]. The loss of PS1 can stimulate the activation of EGFR and β-catenin pathways, contributing to neurodegeneration and aberrant cell cycle re-entry [228].

As a consequence of EGFR signaling activation, the transcription of several genes and regulators can be altered. In particular, a panel of miRNAs essential for metastatic phenotypes and neurodegeneration appears to be involved. The miRNAs have been termed “guardians” of the genome, as they help maintain cellular genomic stability. Reports have identified two miRNAs, miR-221 and miR-222, as downstream targets of the EGFR-RAS-RAF-MEK pathway [230, 231]. MiR-221/222 can modulate cell cycle progression by repressing cell cycle inhibitor proteins p27/Kip1 and p57, which facilitates cell proliferation and self-renewal [232]. The ability to modulate the cell cycle may explain the role of miR-221/222 in neurodegenerative disorders and in the apoptotic death of damaged neurons. Cell cycle control is among the hypotheses for AD (Fig. 3) and preventing neurons from entering high-risk states may provide attractive therapeutic strategies. Given the central role of these miRNAs in neurodegenerative disorders, they provide potential novel molecular therapies to be explored in thefuture.

CLOSING REMARKS

The criteria for diagnosing dementia require the presence of multiple cognitive deficits in addition to memory impairment. The diagnostic approach for AD begins with collecting the clinical history, carrying out a physical examination and cognitive testing. At this phase, there is a clinical recognition of a progressive memory decline, and a decrease in the patient’s ability to perform daily living activities, personality changes, and behavioral problems may also become evident. During the diagnostic process, interviewing friends and family may be helpful and an important tool to assist in the diagnosis. Neuroimaging techniques are evermore used as tools for AD diagnosis, namely magnetic resonance imaging (MRI), computerized tomography, and more recently positron emission tomography (PET). In MRI, structural brain changes can be visualized due to substantial neuronal loss. MRI exams are routinely requested and are helpful to rule out other possible causes of dementia such as brain tumor and vascular lesions [233]. PET assesses the Aβ deposition in brain using a radioactive compound, where the Pittsburgh compound B is the present gold standard [234]. This compound binds specifically to Aβ and does not bind to neurofibrillary tangles or Lewy bodies at the concentrations achieved during PET scan [235]. Nonetheless, diagnostic confirmation is based on post-mortem observation of the specific pathological lesions like NFT, SP, and synapse dysfunction and loss. The present focus is on identifying molecular abnormalities that contribute to the deposition of SPs and NFTs as early as possible in the etiologyof AD.

Neurochemical biomarkers are increasingly used in assisting with the complementary AD diagnostics. The ‘gold-standard’ monitors levels of Aβ peptides (in particular Aβ_1 - 42) and phosphorylated tau in the cerebrospinal fluid [236, 237]. It is evident that protein phosphorylation contributes significantly to the occurrence of both these biomarkers; tau by direct phosphorylation and as described above AβPP phosphorylation can influence levels of Aβ production. Novel biomarkers are presently the subject of intense study as well as identifying novel biomarker candidates in peripheral tissues. Blood based biomarkers are particularly sought, given that lumbar puncture is an additional burden for the patient. Here too, investigating protein phosphorylation events may provide very promising candidates. As described above, given the number of kinases, phosphatases, and the various substrates involved, predicting an outcome as a result of any single phosphorylation/dephosphorylation event can prove to be an inefficient exercise and it follows that we move toward a system biology approach. It is clear, however, that the phosphorylation state of a key set of proteins may provide fundamental cues to anomalous cellular processes and thus serve as highly specific diagnostic biomarkers. In closing, it is worth emphasizing that by merging the interactomes for tau and AβPP, a sub-set of nodes that are presently being pursued for their therapeutic value became clearly evident. In fact, these candidates are particularly important as they bridge/interact with the proteins fundamental to the two central histopathological hallmarks associated with AD.