In-vivo and in-vitro techniques used to investigate Alzheimer's disease

Abstract

Alzheimer's disease (AD) is one of the most common neurological diseases. It is characterized by the presence of β-amyloid peptides and highly phosphorylated tau proteins in the brain. The study of AD became more reliable with the development of new advances in both in-vivo (e.g. positron emission tomography, single-photon emission computed tomography, magnetic resonance imaging, functional magnetic resonance imaging, diffusion-weighted magnetic resonance imaging, and diffusion and perfusion magnetic resonance imaging) and in-vitro (e.g. circular dichroism, Fourier transform infrared spectroscopy, dye binding, transmission electron microscopy, scanning transmission electron microscopy, scanning tunnelling microscopy, atomic force microscopy and fluorescence resonance energy transfer) methods. These methods are used in the study of the pathogenesis and diagnosis of AD. Each method has its own advantages and limitations. This review explains the significance of different methods in understanding the pathogenesis of AD.

Keywords:

• Alzheimer's disease
• β-amyloid (Aβ) peptide
• in-vivo methods
• in-vitro methods

Introduction

Polypeptides are folded into a three-dimensional (3D) native structure to enable their biological functions. The secondary structure (e.g. α-helix and β-sheet) is a local spatial arrangement of the amino acid without involving any interaction with the side-chain, while the tertiary structure is a folded, global spatial arrangement of amino acids which involves the interaction of the side-chain. Often, the proteins fail to fold into their native tertiary conformation, giving rise to a misfolded state which is toxic. In most proteopathies, a change in the 3D folding (conformation) increases the tendency of a specific protein to bind to itself and form an aggregate, which is resistant to clearance and interferes with the normal function of the affected organs. Usually, disease-causing molecular configurations involve an increase in the β-sheet secondary structure of the protein. The abnormal deposition of protein in brain tissue is a feature of several age-related neurodegenerative diseases, including Alzheimer's disease (AD), Parkinson's disease and Huntington's disease (Schulz & Dichgans 1999). Although the composition and location of protein aggregates differ between diseases, the existence of protein aggregates in most age-related neurodegenerative diseases suggests that protein deposition is toxic to neurons. Protein deposition may cause damage directly, interfere with intracellular trafficking or sequester proteins that are important for cell survival.

AD is one of the most common neurological diseases. Various models have been presented to explain the genesis of AD. The following sequence of events takes place: Amyloid Precursor Protein (APP), Aβ42, fibrillar Aβ and oligomers, amyloid plaques, inflammation and neurofibrillary tangles (NFTs), neuron death and synapse loss. In AD, tau proteins become hyperphosphorylated and lose the capacity to bind to microtubules, resulting in NFTs. Presenilin-1 (PS-1) is present in the endoplasmic reticulum/Golgi complex and cleaves the gamma-secretase site. Abnormal proteins from the PS1 and PS2 genes apparently influence the gamma-secretase enzyme, causing more Aβ₄₂ peptide formation.

The various methods used to understand AD can be classified into in-vivo and in-vitro techniques. In-vivo imaging methods include positron emission tomography (PET), single-photon emission computed tomography (SPECT), voxel-based morphometry (VBM), magnetic resonance imaging (MRI) (Zamrini et al. 2004), magnetic resonance perfusion imaging (MRPI), functional magnetic resonance imaging (fMRI), diffusion and perfusion imaging, diffusion-weighted imaging (DWI) and magnetic resonance spectroscopy (Zamrini et al. 2004). PET and SPECT were especially developed to diagnose AD by detecting amyloid plaques, while other brain imaging techniques were developed to evaluate the structure or function of the brain. In these methods, AD is diagnosed indirectly by the interpretation of data. In AD, β-amyloid (Aβ) protein undergoes partial folding, creating a similar unstable intermediate that undergoes conformational transformation into a β-sheet-rich structure and self-assembly.

In vitro techniques used to study this process include nuclear magnetic resonance (NMR)-based methods such as solution-phase and solid-phase NMR; diffraction-based methods such as X-ray crystallography and X-ray fibre diffraction; spectroscopy-based techniques such as X-ray absorption spectroscopy (XAS), circular dichroism (CD) spectroscopy and Fourier transform infrared spectroscopy (FTIR); fluorescence-based approaches such as Congo red binding and birefringence, thioflavin-T and thioflavin-S fluorescence, and fluorescence resonance energy transfer (FRET); microscopy-based methods such as transmission electron microscopy (TEM) (Tomaselli et al. 2006), scanning transmission electron microscopy (STEM), scanning tunnelling microscopy (STM) and atomic force microscopy (AFM); and electron spin resonance (ESR)-based methods.

The present review explains the principles of and advancements in these methods, which are associated with the study of AD.

Brain imaging techniques for in-vivo investigation of Alzheimer's disease

In-vivo diagnosis of AD can be established by postmortem histopathological examination. Postmortem brains of AD patients show neuropathological characteristics such as senile plaques and NFTs, which contain Aβ peptides and highly phosphorylated tau proteins, respectively (Klunk 1998; Selkoe 2001). This section deals with diverse in-vivo methods and covers all the brain imaging techniques currently available in the field.

Positron emission tomography

PET is based on the selective uptake of very small (picomolar) quantities of biological substances, which are labelled with a positron emitter such as carbon-11, oxygen-15, nitrogen-13 or fluorine-18 (Ziegler 2005). Fluorine-18 is used in the form of ¹⁸F-labelled fluorodeoxyglucose (FDG). On the basis of the uptake pattern of FDG, baseline PET differentiates AD from other forms of dementia. Deficits in temporoparietal metabolism are seen in patients with AD but are absent in patients with other forms of dementia (Salmon et al. 1994; Okamura et al. 2001). The radioligand N-methyl-¹¹C-2-(4-methylaminophenyl)-6-hydroxybenzothiazole (also known as ¹¹C-PIB or ¹¹C-6-OH-BTA-1) has been developed for use in PET (Klunk et al. 2001; Mathis et al. 2004). PIB-PET has been shown to be superior to FDG-PET in distinguishing AD from healthy controls (Ng et al. 2007). The best regions for FDG-PET (hippocampus) and PIB-PET (middle frontal gyrus) have high diagnostic agreement and equal value in the diagnosis of AD. Using ¹¹C-PIB, it is possible to visualize the uptake pattern and to quantify the amount of Aβ present in the brain (Klunk et al. 2004). Several groups have shown that healthy controls and patients with AD exhibit differential ¹¹C-PIB uptake (Klunk et al. 2004; Kemppainen et al. 2006; Mintun et al. 2006; Rowe et al. 2007). Possible reasons for this are incorrect clinical diagnosis of AD (Rousset et al. 1998; Zamrini et al. 2004), controls who have incipient AD, unknown ligand specificity (Zhou et al. 2007), partial volume effects (Kanetaka et al. 2004), inappropriate quantification (Ng et al. 2007; Zhou et al. 2007), and variability and specificity of anatomical voxels of interest to describe and quantify Aβ binding.

The cerebral metabolic rate of glucose metabolism (MRglc) using [¹⁸F]fluoro-2-deoxyglucose ([¹⁸F]FDG-PET) and Aβ pathology using Pittsburgh compound B ([¹¹C]PIB-PET) are also assessed in the diagnosis of AD (Li et al. 2008). Studies with PET have suggested that many other radiotracers showing affinity for Aβ in-vivo can be used to detect amyloid in-vivo. They include 2-(3-[¹⁸F]fluoro-4-methyaminophenyl)benzothiazol-6-ol (GE-067, flutemetamol) (Koole et al. 2009), (E)-4-(N-methylamino)-49-(2-(2-(2-[¹⁸F]fluoroethoxy)ethoxy)ethoxy)-stilbene (BAY94-9172, florbetaben) (Villemagne et al. 2011), (E)-4-(2-(6-(2-(2-(2 [¹⁸F]fluoroethoxy)ethoxy)ethoxy)pyridin-3-yl)vinyl)-N-methylaniline (AV-45, florbetapir) (Choi et al. 2009; Lin et al. 2010; Poisnel et al. 2012), 2-(2-[¹⁸F]fluoro-6-(methylamino)pyridin-3-yl)benzofuran-5-ol (AZD4694) (Cselenyi et al. 2012), [¹¹C]SBI-13 (Ono et al. 2011), [¹¹C]BF-227 (Kudo et al. 2007), [¹¹C]MeS-IMPY (Seneca et al. 2007), [¹⁸F]FDDNP (Shoghi-Jadid et al. 2005), [¹⁸F]AV-1 (florbetaben, BAY94-9172) (Rowe et al. 2008), AZD2184, (2-[6-(methylamino) pyridin-3-yl]-1,3-benzothiazol-6-ol) (Johnson et al. 2009), [¹¹C]AZD2184 (Johnson et al. 2009; Nyberg et al. 2009; Andersson et al. 2010), [18F]AZD4694 (Jureus et al. 2010; Cselenyi et al. 2012) and AZD2995 (Forsberg et al. 2008; Swahn et al. 2010). Both AZD2995 and AZD2184 detect Aβ with high affinity and specificity and also display a lower degree of non-specific binding than that reported for PIB. Overall, [¹¹C]AZD2184 seems to be an Aβ radioligand with higher uptake and better group separation than [¹¹C]AZD2995 (Forsberg et al. 2013). Similarly, among florbetapir F-18 ([¹⁸F]AV-45), [¹⁸F]flutemetamol ([¹⁸F]GE067), florbetaben ([¹⁸F]BAY94-9172) and [¹⁸F]FDDNP, florbetapir F-18 ([¹⁸F]AV-45) is widely used as a research biomarker in AD (Koole et al. 2009; Jagust et al. 2010; Kung et al. 2010). Florbetapir F-18 demonstrates high affinity and specificity to Aβ. It is important to note that only some of the probes described above are applicable for testing in humans. The probes applicable to humans includes [¹⁸F]FDG, Pittsburgh compound-B and AV-45 florbetapir, while others are either under clinical trial or under in- vitro studies. The structures of chemical probes applicable to humans are given in Table 1.

Table 1. Structure of chemical positron emission tomography probes that are applicable for investigating Alzheimer's disease in humans.

Serial
Number
(SN)	Chemical probe	Structure
1	[^18F]Fluoro-2-deoxyglucose
2	Pittsburgh compound B
3	AV-45 florbetapir

Single-photon emission computed tomography imaging

Cerebral SPECT is based on brain uptake of technetium 99m-based lipid-soluble radionuclides such as ethyl cysteinate dimer or hexamethylpropylene amine oxime. It is a widely available method for the evaluation of brain perfusion with a rotating gamma camera. This method yields semiquantitative or relative values for comparison within and between patients (Pickut et al. 1999). SPECT imaging also aids in the differential diagnosis of patients with dementia. Iodinated (123-I) pyridyl benzofuran derivatives show excellent affinity for Aβ(1–42) aggregates and intensely labelled Aβ plaques in autoradiographs of postmortem AD brain sections. Several radioiodinated and ^99mTc-labelled ligands for Aβ plaques have been reported, but unfavourable pharmacokinetics in-vivo (e.g. low uptake into the brain and slow washout) have prevented clinical studies (Ono et al. 2011). [¹²³I]6-Iodo-2-(49-dimethylamino)-phenyl-imidazo[1,2-a]pyridine (IMPY) is the first SPECT imaging compound tested in humans (Kung et al. 2002, 2004; Newberg et al. 2006). [¹⁸F]FPYBF-2 shows selective affinity for Aβ plaques and good pharmacokinetics in the brain (Ono et al. 2011). More recently, the development of ^99mTc-labelled pyridyl benzofuran derivatives as SPECT probes targeting Aβ plaques has been reported (Cheng et al. 2012). Ono et al. (2013) also suggested the feasibility of in-vivo Aβ imaging with [¹²³I]8 in the brains of AD patients.

Both PET and SPECT are non-invasive techniques which enable the presymptomatic diagnosis of cerebral β-amyloidosis, risk prediction of AD and monitoring of the effectiveness of anti-amyloid therapies (Klunk 2011). PET is similar to SPECT in its use of radiolabelled probes and detection of γ-rays. However, the probes used in SPECT emit γ-radiation that is measured directly, whereas PET probes emit positrons that annihilate electrons, causing two γ-photons to be emitted in opposite directions. The PET scanner detects these emissions coincidentally to provide higher resolution images than SPECT. SPECT scans are less expensive than PET because they utilize longer lived radioisotopes such as ^99mTc (6 h) or ¹²³I (13 h) than PET. Therefore, SPECT would be more appropriate than PET for routine diagnostic use.

Voxel-based morphometry

Voxel-based techniques are used to determine differences in the structural and functional imaging of the brain by analysing the pathophysiology and disease-associated functional changes (Herholz 2003; Mosconi 2005). VBM permits the comparison of local grey matter density at every voxel in an image and has been developed to quantify brain atrophy in AD (Baron et al. 2001; Burton et al. 2002; Ishii et al. 2005). In a MRI-VBM study, the grey matter density of the amygdala–hippocampal complex and bilateral temporal and frontal gyri was significantly lower in the mild AD group than in the healthy control group (p < 0.01) (Kawachi, Ishii et al. 2006).

Magnetic resonance imaging

The basic principle of MRI is similar to NMR. The human body contains water and its protons are excited by normal frequency radiowaves (typically 40–130 MHz). Excited protons relax back into the ground state and produce a relaxation pattern, i.e. free-induction decay (FID), that is dependent on the position of water in the body. Different locations of water have different chemical environments around the protons, which lead to different relaxation patterns. The relaxation pattern is converted into a magnetic resonance image by Fourier transformation. The voxel intensity of a given tissue depends on the proton density of that tissue. The higher the proton density, the stronger would be the FID response signals. The MRI contrast also depends on two other tissue-specific parameters, namely longitudinal relaxation time (T₁) and transverse relaxation time (T₂). T₁ measures the time required for the magnetic moment of the displaced nuclei to return to equilibrium, and T₂ indicates the time required for the FID response signal from a given tissue type to decay.

MRI-derived markers of AD show relatively high diagnostic accuracy for AD but clinically insufficient predictive value for the prediction of progression from mild cognitive impairment (MCI) to AD (Killiany et al. 2002; Kantarci & Jack 2003; Pennanen et al. 2004; Csernansky et al. 2005; Stoub et al. 2005). Structural MRI has been used to detect spatial patterns of atrophy in AD (DeCarli et al. 1995; Duchesne et al. 2008; Kloppel et al. 2008; Misra et al. 2009; Vemuri et al. 2008). T1-weighted structural MRI was used to determine the cortical thickness and to provide anatomical information in a navigated transcranial magnetic stimulation study (Niskanen et al. 2011). The relationship between cortical thickness and motor cortex excitability has been studied in patients with AD or MCI and in healthy controls (Plant et al. 2010; Niskanen et al. 2011). High-resolution T1-weighted magnetization produced a rapidly acquired gradient echo 3D sequence showing a pattern of atrophy in agreement with the findings of a range of previous independent MRI-based (Meguro et al. 2001; Scahill et al. 2002) and other neuropathological studies (Price et al. 1991).

Magnetic resonance perfusion imaging

MRPI uses rapid T2*-weighted imaging of the brain during intravenous injection of a bolus of paramagnetic contrast material. This technique enables the measurement of several haemodynamic parameters, including relative cerebral blood volume, flow and transit time (Rosen et al. 1989; Rempp et al. 1994). MRPI measurement of cerebral blood volume was found to be similar to changes in the cerebral metabolic rate of glucose consumption as determined by PET in patients with dementia. MRPI may be a lower cost alternative to PET (Gonzalez et al. 1995). The sensitivity and specificity of MRPI are superior to those of cerebral SPECT, with 87–95% sensitivity for patients with mild or moderate AD and 88–95% specificity in control subjects (Harris et al. 1996, 1998; Maas et al. 1997; Bozzao et al. 2001).

Functional magnetic resonance imaging

In fMRI, regional brain activity is measured on the basis of local changes in deoxyhaemoglobin concentration in response to various stimuli and tasks (Courtney et al. 1997). In brief, rapid T2*-sensitive imaging, mainly gradient-echo, echo-planar imaging, is performed after a stimulus or performance of a specific task and rest periods. After this step, a voxel-by-voxel statistical comparison is performed between the images obtained during the stimulus/task periods and those obtained during the rest periods. This creates a statistical activation map that can be ‘thresholded' and presented as a colour overlay on anatomical T1-weighted images.

The results showed an increased (Bookheimer et al. 2000) or decreased (Smith et al. 1999; Small et al. 2000a) activation in the ‘at-risk group' compared with activation in a ‘'control group'. The authors of studies in which increased activation is seen in the AD or at-risk groups explain their findings in terms of a compensatory-recruitment hypothesis. This hypothesis presumes that there is enough healthy neural tissue present to accommodate the task. In the presence of substantial neuronal loss, decreases in activation may result (Bookheimer et al. 2000). In summary, the results of fMRI studies of cognitively impaired or at-risk patients have shown increases or decreases in the intensity and extent of activation, compared with findings in control groups. In comparison with controls, patients in the AD group and one-third of patients in the MCI group showed diminished activation throughout, while the other two-thirds of the MCI group showed a decrease (Thulborn et al. 2000), which supports the compensatory-recruitment hypothesis (Saykin et al. 1999). Additional activation in the AD group compared with the control group also supports the compensatory-recruitment hypothesis (Small et al. 2000b). In the AD group, greater atrophy associated with greater activation supports the compensatory-recruitment hypothesis as well. However, decreased activation in the high-risk group relative to that in the low-risk group in the middle and posterior inferior temporal regions does not support the compensatory recruitment hypothesis (Smith et al. 1992; Johnson et al. 2000).

Diffusion and perfusion magnetic resonance imaging

Diffusion imaging is based on thermal molecular movement. This provides qualitative and quantitative parameters that are highly sensitive to the microstructural and pathophysiological alterations occurring in association with specific pathological conditions (Bozzao et al. 2001). Perfusion MRI is used to differentiate patients with possible or probable AD (Harris et al. 1998). Reduced perfusion (regional cerebral blood volume) has been observed in the temporoparietal and sensorimotor cortices of AD patients compared with age-matched controls.

Sandson et al. (1999) reported decreased anisotropy in the posterior white matter and increased apparent diffusion coefficient (ADC) values in hippocampi of patients with AD compared with a control group. Harris et al. (1991) documented that dynamic susceptibility contrast MRI is capable of detecting haemodynamic deficits with high sensitivity and specificity in AD patients. Many MRI studies have shown that there is an increase in cerebrospinal fluid and a reduction in cerebral volume in patients with AD in comparison with healthy volunteers (Tanna et al. 1991; Jack et al. 1992; Wahlund et al. 1993). Rusinek et al. (1991) reported a significant reduction of grey matter in affected patients compared to age-matched control volunteers. Tanabe et al. (1997) used a semiautomatic segmentation of magnetic resonance images of the brains of AD patients and showed significant brain atrophy to be attributable to loss of cortical grey matter.

Diffusion-weighted magnetic resonance imaging

DWI is sensitive to the microscopic motion of water molecules in tissue. Its primary application is in the evaluation of acute cerebral ischaemia, but it has also been applied to the study of patients with AD (Hanyu et al. 1997, 1998; Sandson et al. 1999). Kantarci et al. (2001) demonstrated statistically significant differences in mean diffusibility between the AD group and the control group. Pathological changes in the hippocampus are manifested by an increased ADC in these regions. Kantarci et al. (2001) found that at a fixed specificity of 80%, the sensitivity is only 57% for distinguishing patients with AD from controls using the hippocampal ADC.

A more advanced application of DWI is diffusion tensor MRI. Diffusion tensor imaging enables measurement of the directionality or asymmetry of microscopic water movement in tissue (Pierpaoli et al. 1996). Such asymmetric diffusion (i.e. anisotropy) is seen in normal white matter, owing to the integrity of white matter tracts that preferentially allows diffusion of water parallel rather than perpendicular to the tracts. The primary application of this technique is in the evaluation of cerebral white matter tracts (Melhem et al. 2002), but it has also been used in studies of AD (Rose et al. 2000). Results demonstrate a statistically significant reduction in white matter integrity throughout the brain, with relative sparing of the motor tracts, reflecting the known pathological and clinical findings in AD (Rose et al. 2000).

Magnetic resonance spectroscopy

NMR spectroscopy yields well-resolved natural abundance proton and decoupled phosphorus spectra from small volumes of the brain in-vivo (Ross et al. 1997). [¹H]NMR spectroscopy is used in many neurological disorders, where it has been successful in localizing the foci of epileptic seizures (Hetherington et al. 1995), categorizing tumours (Preul et al. 1996), and studying the relationship between energy metabolism and neurotransmission (de Graaf et al. 2004). Furthermore, it has been used to characterize cerebral metabolic alterations in individuals with MCI and AD patients (Schuff et al. 1997; Griffith et al. 2008; Kantarci et al. 2009; Zhang et al. 2009; Rupsingh et al. 2011; Watanabe et al. 2012) and in several transgenic AβPP animal models for AD (Choi et al. 2007; Woo et al. 2010; Mlynarik et al. 2012).

Proton NMR spectroscopy allows non-invasive assessment of a number of local metabolite levels in brain tissue (Shonk et al. 1995; Ross et al. 2005; Kantarci 2007, 2008). It also allows in-vivo assessment of N-acetylaspartate (NAA), aminobutyric acid, glutamine, glutamate, myo-inositol, glycine, mobile choline moieties, creatine and phosphocreatine, lipids and lactate (Ross et al. 1997; Schuff et al. 1997; Griffith et al. 2008). NAA is mainly present in neurons in the central nervous system. NAA is generally considered as a marker of neuronal function (Chen et al. 2000) and neuronal viability, and therefore reduction of NAA has been used as an indicator of the progression of neurodegenerative pathology in AD patients, and to differentiate stable MCI from progressive MCI (Kantarci et al. 2007; Moffett et al. 2007; Jessen et al. 2009). Elevated myo-inositol has been associated with inflammatory processes (Govindaraju et al. 2000). Various [¹H]MRS studies have found increased levels of myo-inositol in the temporal, parietal and occipital lobes of AD patients (Rupsingh et al. 2011; Watanabe et al. 2012). myo-Inositol is a sugar alcohol that is thought to be a marker for osmotic stress, astrogliosis and microglial activation (Govindaraju et al. 2000), and an increase in cerebral myo-inositol levels is therefore associated with inflammatory processes. Significant differences in hippocampal myo-inositol/glycine concentrations between 12-month-old wild-type and AβPP-PS1 mice were not observed, but correlation analysis revealed a positive correlation between the levels of Aβ-40 and Aβ-42 aggregates and myo-inositol/glycine levels in the AβPP-PS1 animals, suggesting the possible involvement of Aβ in the inflammatory process (Jansen et al. 2013). Furthermore, disturbances of several other metabolites have been found in AD patients, although the reports are inconsistent. Some studies identified elevated choline-containing compounds (Chitcholtan et al. 2008) and creatine (Tomaselli et al. 2006) in AD patients (Pfefferbaum et al. 1999; Huang et al. 2001; Kantarci et al. 2004), whereas others did not (Schuff et al. 1997).

The proton spectrum in AD is different from normal, with elevated myo-inositol/creatine and reduced NAA/creatine (Miller et al. 1993; Shonk et al. 1995). Stokes and Hawthorne (1987) found a marked decrease in the membrane lipid phosphatidyl inositol in the postmortem AD brain. Jolles et al. (1992) showed reduced activity of the inositol polyphosphate (Perdomo et al. 2012) enzyme, phosphatidylinositol kinase and postulated a specific defect of the inositol polyphosphate cascade in the AD brain. Several groups (Miller et al. 1993; Moats & Shonk 1995) have identified a 20–30% increase in the concentration of myo-inositol in mild to moderate AD, as opposed to other dementias (Moats & Shonk 1995). Shonk and Ross (1995), using [¹H]MRS in adults with Down's syndrome, observed that significant elevation of myo-inositol occurs before the onset of dementia and without any parallel reduction in the neuronal marker NAA. Furthermore, in a single patient with both Down's syndrome and dementia, the magnetic resonance spectrum was indistinguishable from that of AD patients, with elevated myo-inositol and reduced NAA concentrations (Shonk & Ross 1995).

In-vitro techniques used to investigate Alzheimer's disease

Amyloid deposits have common structural characteristics but the contributing amyloid proteins have different primary structures (Dobson 2004; Makin & Serpell 2005; Pastor et al. 2008). The proteins that cause amyloid-associated diseases can be divided into two classes based on their native structures: globular proteins and natively unfolded proteins. In the first group, amyloid formation appears to require a step of partial unfolding, leading to an unstable intermediate (or intermediates) that self-associates into β-sheet-rich toxic aggregates. In the second group, proteins (e.g. Aβ protein, α-synuclein and tau) undergo partial folding, creating a similar unstable intermediate that undergoes conformational transformation into a β-sheet-rich structure and self-assembly (Fezoui & Teplow 2002). The amyloid deposits show characteristic Congo red binding, cross-β X-ray diffraction patterns and unbranched fibrillary morphology by electron microscopy (Makin & Serpell 2005). The different in-vivo methods that are used to study AD are discussed below.

Nuclear magnetic resonance-based methods

Solution-state nuclear magnetic resonance

Solution-state NMR is used for studying the 3D structures of soluble samples. Hydrogen-1 (¹H), carbon-13 (¹³C) and nitrogen-15 (¹⁵N) are common markers for protein which can be used to study amyloidogenic peptide and protein monomers. Oligomerization of Aβ results in broad signals that eventually disappear because of the increase in tumbling time (Zeeb & Balbach 2004). Changes in chemical shift, linewidth, cross-peaks and the nuclear Overhauser effect (NOE) can be used to monitor the transition of monomers into oligomers (Dyson & Wright 2004; Zeeb & Balbach 2004). NMR is used to identify the folding of a protein, such as the secondary structures and long-range interactions among different regions (Esposito et al. 2005; Sung & Eliezer 2006; Tomaselli et al. 2006). NOE signal intensity, J-coupling constants and chemical shift are very useful for calculating distances and torsion angles of correlated atoms (Shao et al. 1999; Tomaselli et al. 2006). Because of the heterogeneous nature of amyloidogenic protein oligomers and the non-availability of the higher concentrations of Aβ needed for NMR experiments, it is difficult to study its oligomerization by solution-state NMR (Mandal, Pettegrew et al. 2006); however, full-length Aβ has been studied by solution-state NMR in aqueous solution at neutral pH (Hou et al. 2004; Yan & Wang 2006).

Solid-state nuclear magnetic resonance

Solid-state NMR is used to study insoluble and non-crystalline solid-state molecular systems. The absence of isotropic tumbling in solid-state samples implies that dipole–dipole couplings and chemical shift anisotropies are not averaged, resulting in broadening of the linewidth in solid-state NMR spectra relative to solution-state NMR, and resulting in lower resolution. The absence of tumbling enables the effects of anisotropic or orientation-dependent interactions to be studied. Cross-polarization (CP), high-power proton decoupling and magic-angle spinning (MAS) are standard techniques used to obtain high-resolution solid-state NMR spectra. The isotropic chemical shift values obtained from CP-MAS experiments can be used to determine site-specific secondary structures. In fibrillar samples, CP-MAS is useful for observing rigid fibrillar parts, whereas dipolar dephasing MAS is used to detect soluble components of mobile parts (Kamihira et al. 2000; Naito et al. 2004). Rational resonance has been used for determining homonuclear–internuclear distances, and rational-echo double resonance has been used to determine heteronuclear–internuclear distances (Lansbury et al. 1995; Tycko & Ishii 2003). Other techniques, such as radiofrequency-driven recoupling and dipolar-assisted rational resonance, are also useful for obtaining folding information of amyloid fibrils (Balbach et al. 2002). Solid-state NMR studies have also been conducted on full-length Aβ, and Aβ fragments including Aβ(34–42), Aβ(10–35) and transthyretin (TTR105–115) (Tycko 2004, 2006a, 2006b, 2006c; Naito & Kawamura 2007; Heise 2008).

Diffraction-based methods

X-ray diffraction crystallography

X-ray crystallography studies the atomic structures of crystals using diffraction patterns of X-ray radiation directed at homogeneous single crystals in which molecules exist in highly ordered repetitive units. The unit cells within a crystal give the diffraction signals, which are converted into 3D electron-density maps. The positions of the atomic nuclei in the crystal are deduced from the electron-density maps. On the basis of these data and complementary chemical information, 3D models of the molecules or macromolecular assemblies in the crystal have been obtained. X-ray crystallography has been used to determine the structures of amyloidogenic proteins that form a stable fold, i.e. non-amyloid form (Rosano et al. 2005). The structures of 30 short segments of amyloidogenic proteins, including Aβ, tau, islet amyloid polypeptide (IAPP) (Colinge et al. 2005) and α-synuclein, have been determined and found to be organized in a steric zipper (Nelson et al. 2005; Sawaya et al. 2007).

X-ray fibre diffraction

Unlike X-ray crystallography, which requires high-quality single crystals, X-ray fibre diffraction can be executed on pulverized or lower quality crystalline samples. The X-ray diffraction patterns of almost all amyloid fibres exhibit broad equatorial reflections at approximately 10 Å and meridional reflections at around 5 Å (Sunde et al. 1997). This is a characteristic of a β-sheet structure, in which the β-strands are arranged perpendicular to the fibril axis (cross-β arrangement).

Spectroscopy-based methods

X-ray absorption spectroscopy

This technique provides information about the local environment and electronic state around heavy atoms in biomolecules. When a high-energy X-ray beam hits the sample, certain energy is absorbed and electrons are excited and ejected from their orbitals (photoelectrons). The energy absorption occurs at defined energy levels which correspond to the binding energy of the electrons in the heavy atom. When a photoelectron leaves the absorbing atom, its wave is backscattered by the neighbouring atoms. The XAS is the plot of absorption coefficients of the heavy atom against the incident X-ray energy. XAS spectra typically are divided into two regions, designated as X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS).

The pathologies of AD involve interaction of the Aβ protein with metal ions such as Cu²⁺, Zn²⁺ and Fe³⁺ (Gaeta & Hider 2005). Information on metal-binding amyloid-peptide and metal binding sites is very useful in understanding the function of transition metals in peptide aggregation and redox chemistry in this disease (Gaeta & Hider 2005; Streltsov 2008). A major cause of neurotoxicity in AD is reactive oxygen species, the production of which in the brain may involve Aβ–Cu ion complexes (Smith et al. 2007). Stellato et al. (2006) reported EXAFS studies of Aβ40–Cu²⁺ complexes and concluded that Cu²⁺ is penta-coordinated to three nitrogen atoms of three histidines (His6, His13 and His14) and two oxygen atoms, one belonging to tyrosine (Tyr10) and the other from either a water molecule or an acidic residue. A number of studies have linked the function of cellular prions to their ability to bind Cu²⁺, and this plays a role in the copper uptake from the extracellular environment to endosomes and the inverse release process (Brown et al. 1997). EXAFS experiments also suggest that a square planar four-nitrogen–Cu²⁺ coordination is formed in prion–Cu²⁺ complexes (Morante et al. 2004; del Pino et al. 2007), in which two nitrogen atoms are contributed by imidazole rings of histidine residues and the other two nitrogen atoms are deprotonated backbone atoms.

Circular dichroism spectroscopy

CD spectroscopy measures the differential absorption of circularly polarized light by asymmetric centres as a function of wavelength. Far-ultraviolet CD spectra (180–250 nm) of peptides and proteins are highly sensitive to secondary structures, while near-ultraviolet CD spectra (250–350 nm) reveal contributions of aromatic side-chains and disulfide bonds and provide information about tertiary structure (Greenfield 2006). The aggregation of amyloidogenic proteins into protofibrils (Harper et al. 1997; Walsh et al. 1997) and fibrils is supplemented by the abundant formation of β-sheet conformation, which is detected by the appearance of an absorption minimum at 215–218 nm in the far-ultraviolet CD spectrum. CD is used for qualitative determination of amyloid assemblies and kinetics of conformational transitions associated with aggregation. CD can be used for determination of the secondary structure because of its tolerance of a wide range of pH and temperature, e.g. low pH compromises the binding of dyes such as Congo red and thioflavin T to amyloid samples.

Fourier transform infrared spectroscopy

FTIR spectroscopy is used for the study of the secondary structure of proteins. A strong absorption band at around 1600–1700 cm⁻¹ in the infrared spectra of proteins has been assigned to the vibration of C=O bonds in the polypeptide main chain, which provide information about the secondary structural elements. This region is also called the amide-I region. Similarly, C–N stretching and N–H bending vibrations give rise to absorption bands at 1550 cm⁻¹ (amide-II) and 1250 cm⁻¹ (amide-III), which are useful for the assignment of secondary structure. After deconvolution of the amide-I band contour, secondary structure content can be estimated by taking into account the relationship between peak position and the type of secondary structures.

Amyloidogenic protein regions have been detected in fibrils at 1626 and 1632 cm⁻¹ for parallel β-sheet structures and at 1690 cm⁻¹ for antiparallel β-sheets. A random coil (an extended, irregular structure with distinct spectroscopic features) has a band at 1644 cm⁻¹ while the α-helix has a band at 1654 cm⁻¹. Helical conformations also contribute to band intensities at 1644 and 1662 cm⁻¹, hence it is difficult to distinguish α-helix from random coil. Non-α and non-β structures appear at < 1660 cm⁻¹. Although strong signals corresponding to particular secondary structures exist in this region, information on infrared spectra obtained in aqueous buffers is masked by a strong signal at 1630 cm⁻¹ due to water bending vibrations. FTIR is used for the study of amyloid assemblies and kinetics of the conformational transitions.

Fluorescence-based methods

Congo red binding and birefringence

Congo red staining is commonly used for the identification of amyloid in tissue samples (Elghetany & Saleem 1988; Elghetany et al. 1989; Jin et al. 2003). An alkaline solution of Congo red stains amyloid with an intense red colour, which leads to a shift in the wavelength of maximal absorption from 490 nm to 540 nm upon binding to amyloid samples. This characteristic helps in the spectroscopic detection of amyloid and measurement of fibril formation. Binding of Congo red to amyloid yields a unique blue–green birefringence under cross-polarized light, enabling the visual determination of amyloid formation. This phenomenon was first observed by the Belgian physician Paul Divry in 1927, while studying degeneration of the ageing brain. It is hypothesized that the Congo red dye intercalates between β-strands parallel to the peptide chains and perpendicular to the fibril direction (Klunk et al. 1989). Binding also occurs through interaction between the dye's negatively charged sulfonate groups with the positive N-terminals of polypeptides, in such a way that the dye's axis is perpendicular to the length of the peptide and parallel to the direction of the fibril axis (Wu et al. 2007). High salt concentrations and alkaline pH are required for binding and birefringence (Brigger & Muckle 1975).

Thioflavin T and thioflavin S fluorescence

Thioflavin T has both polar and non-polar functional groups, which form micelles in aqueous solutions with the non-polar dimethylaminophenyl groups in the core and the charged benzothiazole groups exposed to the solvent. Thioflavin T micelles bind to β-sheet structures, resulting in relatively intense fluorescence signals compared to unbound thioflavin T (Khurana et al. 2005). Vassar and Culling (1959) introduced the use of this dye in 1959, for detecting amyloid tissue in the kidney. In the unbound state, excitation and emission occur at 342 nm and 430 nm, respectively, while when bound to amyloid samples, the excitation and emission of thioflavin T undergo a bathochromic shift to 444 nm and 482 nm, respectively (LeVine 1999). Thioflavin T has been used to characterize amyloid aggregates in vitro and ex vivo, whereas uncharged thioflavin derivatives have been used to detect amyloid in-vivo (Klunk et al. 2001). This dye can also be used as a staining agent for fluorescence microscopy and confocal microscopy of amyloid proteins (Abe & Nakanishi 2003; Ban & Goto 2006).

Fluorescence resonance energy transfer

FRET is based on the energy transfer between a donor and an acceptor when they are in the resonance state. The efficiency of FRET depends on the distance between fluorophores (donor and acceptor); therefore, it is useful for measuring intramolecular and intermolecular distances, protein folding and protein–membrane interactions.

Kim and Lee (2004) have reported FRET experiments in which both fluorescence donor and acceptor have conjugated to Aβ (Talafous et al. 1994; Kirshenbaum & Daggett 1995; Brunger 1997; Coles et al. 1998; Shao et al. 1999; Riek et al. 2001; Fezoui & Teplow 2002; Zeeb & Balbach 2004; Tomaselli et al. 2006; Chen et al. 2007), for the investigation of conformational transitions during Aβ assembly. Amine-reactive 5-(and 6-)carboxytetramethyl rhodamine was introduced as the donor to the N-terminus of Aβ(11–25) and 4-dimethylaminophenylazophenyl-4-maleimide was attached to the C-terminus as the acceptor through a cysteine linker. This system is used to monitor conformational changes during Aβ fragment assembly in real time. FRET measurement has also been used to study the structure and dynamics of α-synuclein under diverse conditions (Lee et al. 2004; Munishkina & Fink 2007).

Microscopy-based methods

Transmission electron microscopy

In TEM, a cathode ray source emits an electron beam, which is accelerated and focused by electrostatic and electromagnetic lenses. When the electrons pass through an electron-transparent specimen, an image is formed from the electrons transmitted through the specimen, magnified and focused by an objective lens, and appears on an imaging screen. The image gives information about the structure of the specimen. Negative staining, rotary shadowing and cryo-electron microscopy have been used to exchange the image and contrast. Steven and Belnap (2005) used TEM to characterize various assembly states of Aβ, including low molecular weight Aβ (Bitan et al. 2003a, 2003b), small Aβ oligomers (Roher et al. 1996; Modler et al. 2003), paranuclei (Bitan et al. 2003a, 2003b), protofibrils (Roher et al. 1996; Walsh et al. 1999; Ward et al. 2000; Nichols et al. 2002; Kheterpal et al. 2003) and fibrils (Wood et al. 1996; Ward et al. 2000). Cryo-electron microscopy has been used to resolve the structure of Aβ42 fibrils, which comprise regularly spaced (∼45 Å) protofilaments that twist regularly along the fibril axis (Luhrs et al. 2005).

Scanning transmission electron microscopy

STEM uses a field emission gun, which delivers a subnanometre beam of 100 kV electrons on to a specimen. The electron beam scans the specimen and scattered electrons are collected by detectors located behind the specimen. The image of the specimen is generated as the focused beam moves step by step over the specimen. STEM has emerged as an excellent tool to characterize the homogeneity and structural properties of transient quaternary structure intermediates in the fibril-formation pathway of amyloid proteins. STEM is used for studies of Aβ and α-synuclein (Lashuel & Wall 2005) and to compare the mass per length (MPL) ratio of Aβ fibrils and protofibrils (Goldsbury et al. 2000, 2005; Petkova et al. 2005).

Scanning tunnelling microscopy

In STM, an extremely sharp conductive tip (ideally terminating in a single atom) traces the contours of the surface of a sample with atomic resolution. An x-y-z piezoelectric translator controls the movement of the tip in three dimensions. When the probe is brought sufficiently close to the sample's surface and a weak voltage (a few millivolts to a few volts) is applied, a small electric current is generated by electrons tunnelling across the distance between the sample and tip. This distance is kept constant (a few nanometres) by mechanisms feeding back to the piezoelectric elements. The magnitude of the current depends exponentially on the distance between the probe and the surface. The tunnelling current rapidly increases as the distance between the tip and the surface decreases. The feedback signal applied to a piezoelectric element provides a measure of the molecular surface contour (Hansma et al. 1988). Initial low-resolution STM studies of Aβ showed the ribbon-like filamentous nature of Aβ fibrils (Shivji et al. 1995). Later, it was shown that Aβ fibrils contain laterally associated filaments that exhibit a right-handed twist (Wang et al. 2003). Another study examined the structure of Aβ40 monomers, dimers and oligomers on a surface of atomically flat gold (Losic et al. 2006). Structures of approximately 3–4 nm and 6–7 nm length have also been found, which correspond to monomeric and dimeric Aβ, respectively (Losic et al. 2006).

Atomic force microscopy

In AFM, a cantilever tip scans the surface contour of the specimen, and upon contact, a repulsive force in the piconewton to nanonewton range bends the cantilever upwards. A laser beam focused on the end of the cantilever detects the extent of bending, and deflection of the laser is translated to force units by a photodetector. By keeping the force constant while scanning across the surface, the vertical movement of the tip generates the surface contour, which is recorded as the topographic map of the sample. AFM has been used to study the assembly of amyloidogenic Aβ proteins (Goldsbury & Green 2005). IAPP fibrils comprise of two protofilaments (Goldsbury & Green 2005). An advantage of AFM is that it allows continuous monitoring of the growth of oligomers (Blackley et al. 2000) and fibrils in solution (Lee et al. 2008).

Electron spin resonance

ESR characterizes the energy levels of a system with unpaired electrons in an externally applied magnetic field (Boas et al. 2008). ESR measures the spin signals of electrons, whereas NMR measures the spin signals of atomic nuclei. More sensitive measurement can be expected in ESR than in NMR because unpaired electrons do not normally exist in biological samples, making the background noise very low. Thus, ESR spectroscopy has higher sensitivity and a lower signal-to-noise ratio than NMR. Using phenyl-tert-butylnitrone and ESR, Butterfield (2003) evaluated the ability of Aβ to produce radicals, a mechanism putatively involved in Aβ-induced toxicity. It has also been suggested that radical production by Aβ alloforms containing familial AD-related substitutions at position 22 is associated with pathogenesis (Murakami et al. 2005). A similar ESR spin-trapping method has been used for other proteins, such as α-synuclein (Tabner et al. 2001). Intermolecular distance analysis in Aβ fibrils (Torok et al. 2002) and intramolecular distance measurements in Aβ monomers (Murakami et al. 2007) have been reported. Structural investigations of IAPP (Jayasinghe & Langen 2004) and α-synuclein (Chen et al. 2007) have also been performed using this method.

Conclusions

There is a need for efficient methods that can be used in the diagnosis and study of the pathogenesis of AD. Various methods are currently used to understand AD but they have their limitations. Both in-vivo and in-vitro methods help in the understanding of AD and the mechanism of pathogenesis by Aβ and tau proteins. Advancement in these methods will improve the diagnosis and treatment of AD.

Funding

Dr Vishvanath Tiwari would like to thank UGC, India, for start-up grant [UGC/BSR/F.No.30-18/2014] and SERB, DST, India, for start-up grant [SB/YS/LS-07/2014].

Disclosure statement

No potential conflict of interest was reported by the authors.