Introduction

Alzheimer’s disease (AD) is the leading cause of dementia, a condition that gradually destroys brain cells and leads to progressive decline in mental function. As the society is graying, the number of elderly people at risk of developing dementia is growing rapidly as a consequence of increasing life span. Recently, the prevalence of AD and its demographic development in the near future has been analyzed by a panel of 12 international experts (Ferri et al. 2005). They estimated that 24.3 million people have dementia today, with 4.6 million new cases of dementia every year (one new case every 7 s).

Histopathologically, AD is characterized by accumulation of misordered proteins in the form of two types of insoluble fibrous material: extracellular senile plaques and intracellular neurofibrillary deposits, originally reported by Alois Alzheimer (Alzheimer 1907). Advances in the proteomics in 1980s allowed the identification of the constituents of these structures at the molecular level. Both appeared to result from aberrant folding of proteins, senile plaques being composed of amyloid β (Aβ) (Glenner and Wong 1984) and neurofibrillary deposits (or tangles, NFT) being composed of the tau protein (Grundke-Iqbal et al. 1986; Wischik et al. 1988a, 1988b).

Tau and Aβ became the objects of intense research. Currently, it is assumed that both of these proteins are responsible for the development of AD pathology. The role of Aβ in the AD pathology was proposed in the Aβ cascade hypothesis (for detailed review see Seabrook et al. 2007; and Walsh and Selkoe 2004), which also became the basis for the development of therapeutic approaches and drug discovery for AD.

In our work we focused on tau protein, which is essential for the propagation of neurofibrillary degeneration in AD and is also able to induce neurofibrillary pathology in other tauopathies (Baker et al. 1997; Goedert et al. 1998; Hutton et al. 1998; Rapoport et al. 2002; Roberson et al. 2007).

Tau Proteome

Tau belongs to the family of microtubule-associated proteins (MAPs) and is involved in establishing neuronal polarity and axonal outgrowth during development and for maintaining axonal morphology and axonal transport in mature neurons (Caceres and Kosik 1990; Dawson et al. 2001; Esmaeli-Azad et al. 1994; Hanemaaijer and Ginzburg 1991; Harada et al. 1994).

At the proteomics level, tau belongs to the family of the intrinsically disordered proteins (IDPs) that are characterized by the absence of a rigid three-dimensional structure (Skrabana et al. 2006; Uversky 2002). However, IDPs attain defined structure upon binding to their partners. Thus, not surprisingly, the IDPs are mostly found between signaling and regulatory proteins that have multiple interaction partners.

The intrinsically disordered character endows tau with high structural flexibility allowing interaction with many proteins. This interaction flexibility is further increased by the presence of six isoforms in the brain (Goedert et al. 1989). Further multiplication of tau forms is achieved through many posttranslational modifications. Eighty-five phosphorylatable Ser/Thr/Tyr residues are present on tau, out of which 45 were already found to be phosphorylated in AD (Hanger et al. 2007). Tau isolated from AD is ubiquitinated and oxidized (Horiguchi et al. 2003; Morishima-Kawashima et al. 1993). Finally, AD tau is also characterized by truncation (Gamblin et al. 2003; Novak et al. 1991, 1989).

Thus, in AD, the combination of all of these modifications leads to a population of hundreds of multiply modified tau proteins, better characterized by the term “AD tau proteome” (Fig. 1). Consequently, giving this multitude of tau species, the following questions arise:

Fig. 1
figure 1

The AD tau proteome. Tau is expressed as six isoforms in adult human brain. This complexity is multiplied by phosphorylation at least 45 known sites in AD, truncation, ubiquitination, and other known modifications taking place in AD. This multitude of tau proteins makes it difficult to identify the key molecular steps of the transition of the disordered normal tau into misordered aggregated protein

Full size image
  1. (a)

    Which of these individual tau proteins are responsible for the initiation of neurodegeneration in AD and related tauopathies?

  2. (b)

    Which represent protective attempts of a neuron against the environmental insults?

  3. (c)

    What are the pathological and protective pathways leading to AD?

In order to address this complex task, decoding the AD tau proteome, we have to undertake a multipart proteomic approach. This approach should take place at four main platforms: functional, structural, signaling and, interaction neuroproteomics.

Functional Neuroproteomics

The appearance of truncated tau is a characteristic feature of AD and other tauopathies (Binder et al. 2005; Novak 1994). We have identified in sporadic AD a truncation event in tau that converts the intrinsically disordered normal tau into a mis-disordered Alzheimer’s tau protein with toxic gain of function (Novak et al. 1989; Zilka et al. 2006).

In order to determine the toxic potential of Alzheimer’s tau, we expressed this protein as a transgene in the rat brain (Zilka et al. 2006). Expression of the mis-disordered tau protein led to the induction of a complete tau cascade of neurofibrillary degeneration (Table 1), consisting of tau hyperphosphorylation, formation of argyrophilic tangles and sarcosyl-insoluble tau complexes with a typical A68 triplet (Fig. 2). These pathological changes resulted in the neuronal dysfunction consequently leading to the progressive neurobehavioral impairment and premature death (Hrnkova et al. 2007; Koson et al. 2008). Furthermore, the rat model of AD also revealed that expression of the Alzheimer’s tau protein led to the accumulation of reactive oxygen species and sensitized rat cortical neurons to cell death induced by oxidative stress (Cente et al. 2006). This indicates that truncation of tau may precede oxidative stress in the pathogenesis of neurodegenerative diseases such as AD and other tauopathies.

Table 1 Comparison of tau pathological cascade in human Alzheimer’s disease and the transgenic rat model of tauopathy
Full size table
Fig. 2
figure 2

Expression of Alzheimer’s tau in the rat brain induces complete tau cascade of neuropathology including the sarcosyl-insoluble tau with the A68 triplet characteristic of human AD neurofibrillary degeneration. The western blot represents sarcosyl insoluble brain extracts from wild type (WT), transgenic rat expressing Alzheimer’s tau (TG) and human AD brain. The blot was probed with a pan tau antibody (DC25) or with PHF1 antibody specific for phosphorylated Ser396/Ser404. Numbers on the left indicate positions of molecular mass markers (kDa)

Full size image

Thus, we have discovered the inducer and driving force of pathogenesis leading to full development of neurofibrillary degeneration of Alzheimer’s type, including histopathological hallmarks—NFTs, oxidative stress, neuronal dysfunction, and death.

The expression of Alzheimer’s tau in the rat brain also induced its hyperphosphorylation, which seems to be an integral part of the tau cascade. The hyperphosphorylation occurred immediately after its expression started, while sarcosyl-insoluble tau appeared only late in the life of the transgenic animals (Zilka et al. 2006). Thus, there are additional steps except truncation and hyperphosphorylation that ultimately lead to the formation of the insoluble misordered tau.

In order to identify the events leading neurofibrillary pathology, we employ molecular mapping of the tau modifications and structures using state-of-the-art proteomics approaches.

Structural Neuroproteomics

Inherent physical properties of the intrinsically disordered protein tau have precluded so far its detailed structural analysis by X-ray crystallography or nuclear magnetic resonance. The tau protein undergoes disorder-to-misorder transition in AD, whereby a highly soluble protein forms insoluble aggregates, neurofibrillary tangles, which contain polymeric structures. Determination of the structure of the polymerized tau—paired helical filaments (PHFs) will provide vital information for therapeutic intervention to halt the pathological process. Many laboratories struggle to determine the structure of the core PHF using different techniques, such as circular dichroism, nuclear magnetic resonance, X-ray diffraction, and some others. Several models of the arrangement of tau molecules within PHF or in solution have been suggested (Berriman et al. 2003; Inouye et al. 2006; Mukrasch et al. 2007; Ruben et al. 1991). In spite of that effort, the structure of misordered tau long escaped its determination.

In order to solve the structure of misordered tau assembled into PHF, we have used a specific monoclonal antibody MN423 raised against the PHF core, the protease resistant product of proteolytic cleavage of Alzheimer’s PHF, which retains its characteristic morphological features and consists mainly of tau protein (Novak et al. 1989). The MN423 has unequivocal specificity for tau ordered into the PHF structure (Khuebachova et al. 2002; Skrabana et al. 2004; Wischik et al. 1988a, 1988b) and represents an in vivo imprint of the core PHF. The antibody, serving as a mold, induced disorder-to-misorder transition of tau upon interaction, i.e., forcing the tau fragment to adopt the PHF core conformation. Crystallization of the complex of monoclonal antibody MN423 with the C-terminal hexapeptide of the core PHF tau led to the identification of its structure at a resolution of 1.65 Å (Sevcik et al. 2007).

This structural analysis suggests important role of the core PHF C-terminus for PHF assembly. This approach shows new possibilities in crystallographic analysis of intrinsically disordered proteins involved in pathogenesis of neurodegenerative diseases. Using several monoclonal antibodies recognizing different parts of the PHF core could lead to the solution of the 3D structure of the whole PHF core.

It is reasonable to expect that this approach will help to reveal the structural principles underlying the tau protein assembly into PHF and possibly will facilitate rationale drug design for inhibition of Alzheimer neurofibrillary changes.

Signaling Neuroproteomics

Hyperphosphorylation of tau is an important “cascade-like” molecular event implicated in tau pathology and a prerequisite for generation of insoluble forms—misordered tau assembled in paired helical filaments in AD and other tauopathies (Iqbal et al. 2005; Pevalova et al. 2006). Alonso et al. (1994, 1996) have shown that hyperphosphorylated tau isolated from human brains exhibits severely reduced ability to bind microtubules (loss of function), and on the other hand gains ability to sequester normal tau and prevents it from executing its functions (gain of toxic function).

However, the role of (hyper)phosphorylation in the transition of tau into its misordered pathological form is still not fully elucidated, since phosphorylation of the individual Ser/Thr residues has different effects on tau function (microtubule binding) and self-assembly into filaments. Some reports showed reduced self-assembly of tau protein phosphorylated in MT domain or Pro-rich domain (Eidenmuller et al. 2001; Schneider et al. 1999), while others reported a weakly promoting effect (Liu et al. 2007). It has recently been found that although hyperphosphorylation of tau is essential for tau induced toxicity, no single phosphorylation event plays a dominant role in controlling tau toxicity (Steinhilb et al. 2007a, 2007b). Finally, many protein kinases are able to phosphorylate tau in vitro and in vivo (for recent reviews see Iqbal and Grundke-Iqbal 2008; Pevalova et al. 2006).

These findings showed that tau is an acceptor of many modification/regulatory events originating in multiple signaling pathways. Thus, as it is typical for IDPs, the high number of regulatory sites (more than 45 phosphosites already identified on tau) allows tau to efficiently participate in both one-to-many and many-to-one signaling pathways (Uversky et al. 2008). They also suggest that the final outcome of tau phosphorylation depends on the combination of various phosphorylated sites on one tau molecule.

In order to understand these regulatory signals, it is necessary to decode the combination of signaling phosphosites on tau—the “tau phosphosignalome”. Towards this goal, we employed state-of-the-art proteomic approaches in combination with the rat model of AD (Zilka et al. 2006). In this model, the transgene Alzheimer’s tau undergoes further modifications (phosphorylation) and becomes misordered, aggregated in AD-like PHF. First, rapid, efficient, and simple procedure for purification of tau proteins from AD brain tissue had to be developed, which would faithfully preserve all phosphorylated residues on tau and thus render tau amenable to further analysis (Ivanovova et al. 2008). Next, efficient separation techniques have to be designed in order to separate and purify the individual populations of tau proteins, based on their phosphorylation pattern and truncation. For their classification, both bottom-up and top-down proteomics approaches have to be employed. The bottom-up approach usually starts with a gel-based protein separation followed by excision of the protein band of interest, in-gel digested with trypsin and the resulting pool of peptides analyzed by mass spectrometry for protein identification and the presence of modifications (Domon and Aebersold 2006). This method can provide information on all existing modifications, e.g., phosphorylation, in the pool of tau proteins in the sample. However, it cannot provide information on the combination of the individual phosphosites (or other modifications) that exists simultaneously on one tau molecule, the key information that distinguishes normal tau from misordered tau.

Therefore, it is necessary to analyze intact tau proteins by top-down mass spectrometry. This technology will provide the essential information on the abundance of the individual tau protein forms, information on the combination of phosphosites existing on tau in diseased brains as well as identity of additional modifications like truncation (Siuti and Kelleher 2007). This approach needs to be supported with high-end instrumentation like Fourier-transform tandem MS (FTMS/MS), Orbitrap, and latest MALDI-TOF MS/MS, which provide both high resolution and high mass accuracy (for a review see Aebersold and Mann 2003).

These approaches would allow us to understand the pathological metamorphosis of normal tau into misordered tau during the disease progression by decoding AD tau signalome. The pathways will be exploited as potential drug targets. The tau molecular species will also be utilized as biomarkers for development of preclinical AD diagnostic assays.

Interaction Neuroproteomics

The intrinsically disordered proteins are characterized by a large number of interacting partners. Tau is no exception to this rule, since many proteins have already been found to interact with tau in vitro and some also in vivo. These include structural proteins like tubulin and actin, peptidyl-prolyl isomerase, motor protein complexes, protein chaperones, cytoplasmic and membrane-bound kinases, phosphatase PP2A, proteases, S100 calcium binding protein B, α-synuclein, and other potential targets (Baudier and Cole 1988; Binder et al. 2005; Dickey et al. 2007; Iqbal et al. 2005; Jensen et al. 1999; Lee 2005; Lim and Lu 2005; Magnani et al. 2007; Mesco and Timiras 1991). Thus, given the high number of phosphorylation sites on tau and diversity of interaction partners, tau most likely sits in a crossroad of many signaling pathways, a typical many-to-one and one-to-many signaling protein (Uversky et al. 2008).

For that reason, uncovering the molecular underpinnings of tau neurodegeneration requires not only mapping the “AD tau signalome” (see above), but it is necessary to map the crosstalk between tau, normal and pathologically altered, and the proteome of a neuronal and glial cell. We set up a proteomics platform for purification and identification of tau interaction partners, the “tau interactome”. The approach is based on immunopurification of tau complexes, their resolution on 2-dimensional electrophoresis and mass spectrometry. We took advantage of the rat model of AD expressing Alzheimer’s tau. This approach already yielded more than 40 interacting partners of tau, many of which are differentially expressed between wild type and transgenic animals (Hanes et al., manuscript in preparation). The differentially interacting proteins will provide essential information on both protective and pathological pathways that are active at different stages of the disease process. Furthermore, these pathways might become targets for therapeutical intervention to halt the progression of the disease or even reverse it.

The second part of the interaction proteomics involves kinetic measurements of tau-partner interactions using optical biosensors (Visser and Heck 2008). The variability in the interaction should provide information on the structure of the tau proteins and the level of “misorder” caused by various modifications, a kind of staging of the transition process from disordered to misordered tau.

Conclusion

The prerequisite for a successful accomplishment of the aforementioned proteomics strategy for decoding tau proteome is that all four platforms work in concert, i.e., there is sufficient flow of information between them. In addition, analysis of tau proteome has to be done in both AD model systems and AD patients. The model systems, from cellular to animal, have advantage of precise staging of the pathology, known pathological starter (transgene, environmental insult), unlimited supply, sample homogeneity and existence of a wild type counterpart of the same genetic background. The human samples have to be used, however, for validation of all observations made in the model systems.

Modern era of proteomics brings about rapid development and implementation of novel technologies in protein analysis. Decoding AD tau proteome will uncover novel targets and detailed knowledge on molecular nature of neurodegeneration induced by pathologically misordered tau protein.