An Ally in Alzheimer's Disease Research
- Fig. 1: Aggregation of the Aβ peptide from unstructured monomers (A), to oligomers rich in antiparallel β-sheets (B), and to amyloid fibers with cross-β architecture. Hypothetical scheme showing the kinetics of appearance/disappearance of the three aggregation states depicted in panels A–C.
- Fig. 2: Infrared absorption (bottom) and second derivative spectra (top) of 200 µM Aβ42 at alkaline p2H (panel A, 10 mM NaO2H, p2H > 11) and neutral p2H (panel B, 100 mM sodium phosphate, p2H 7). Samples were prepared in 2H2O and their spectra were recorded using 50-µm infrared cells. Inset: CD spectra of samples A (continuous line) and B (dashed line).
- Fig. 3: Time-resolved infrared spectroscopy of the pH-induced aggregation of Aβ40. (A) Photolysis of “caged” sulfate with UV light, leading to production of a sulfate anion and a proton. (B) Difference spectra (spectra after the flash minus reference spectrum before the flash), showing time-dependent loss of random coil (1638 cm-1) and formation of β-sheets (1623 cm-1 and part of the band at 1687 cm-1). The photolysis byproduct, 2-nitrosoacetophenone, also contributes to the band at 1687 cm-1. The disappearance of the NO2 vibration in the “caged“ compound gives rise to a sharp, negative band at 1527 cm-1.
- Fig. 4: Theoretical scheme and infrared absorption spectra showing formation of a β-hairpin from an Aβ monomer containing one (A, B) and two (C, D) 13C=18O labels within its sequence. In the singly labelled peptide, the labelled residue can remain either in a random coil conformation in the β-hairpin (A), so that its absorption does not change (~1585 cm-1), or be incorporated into one of the two strands (B), shifting its absorption to lower wavenumbers because of inter-strand hydrogen-bonding. In the doubly labelled peptide, the two labelled residues can be incorporated into any two positions of the β-hairpin, yielding the same band as in (B), or be close in space and aligned roughly in the same direction, thereby coupling and giving rise to a band that is at even lower wavenumbers that for the uncoupled signals in (B) and (C).
- © freshidea - Fotolia.com
Alzheimer's disease (AD) ranks among the most common and widespread neurodegenerative disorders in the western societies, and is expected to affect more than 1 % of the global population in the forthcoming decades. Although more than a century has passed since its discovery by neuropathologist Alois Alzheimer in 1906, no treatment is currently available that can reverse or simply stop the progression of the disease, with AD-affected individuals living, on average, as little as seven years after diagnosis.
Although the hallmark of AD is represented by the accumulation, in the brain, of microscopic extracellular deposits composed of amyloid fibrils of the 40-42 residue-long amyloid-β (Aβ) peptide, it is now widely recognised that the severity of the observed symptoms rather correlates with the concentration of transient extracellular assemblies of the Aβ peptide known as oligomers (olígos = few, merós = parts) . Unlike the bulky fibrils, oligomers are small enough to be soluble and diffuse locally across the brain, where they are thought to interfere with synaptic structure and function . Despite two decades of extensive investigation, however, the structure and exact mode of toxicity of Aβ oligomers have not yet been unequivocally elucidated. Several model structures have been proposed, but a general consensus is far from being reached.
Detailed knowledge of oligomer structure and kinetics of formation is essential to develop and test novel drugs for their ability to disrupt toxic oligomers (fig. 1). Infrared spectroscopy is to date among a restricted array of spectroscopic techniques that simultaneously provide structural and kinetic details of forming oligomers, as well as the only type of spectroscopy that can study both soluble and insoluble protein samples. This makes infrared spectroscopy an invaluable tool in the fight against Alzheimer's disease.
From Molecular Vibrations to Structural Details
Absorption of infrared radiation by oscillating atoms is what gives rise to an infrared spectrum. The information content of an infrared spectrum is large enough so that the structures of simple molecules can be deduced from their infrared absorption spectra.
Because proteins, however small they may be, contain several thousands of atoms, extensive spectral crowding and band overlapping prevent their three-dimensional structure from being obtained by mere interpretation of their infrared spectra. Nevertheless, remarkable structural details can be deduced by interpretation of infrared bands arising from vibrations of the peptide groups of a protein or peptide, such as the so called "amide" bands. The amide I band (1700-1600 cm-1, or ~6 µm) arises predominantly from the C=O stretching vibration within a peptide group. The exact position of the amide I band of a given peptide group is largely influenced by coupling to nearby peptide groups. Because coupling between two or more peptide groups depends on how these are arranged in space, different types of secondary structures yield different, well-defined amide I absorptions. As a result, the secondary structure of a protein can be assessed and quantified by analysis of its infrared absorption spectrum. However, an infrared spectrum is sensitive to more details than just secondary structure. Solvent exposure, local deformations and supramolecular assembly contribute to, and can often be inferred from the infrared spectrum of a protein.
Additional structural information, such as the stability of the overall tertiary fold, can be deduced by analysis of the amide II band, arising mainly from the peptide N-H bending, in hydrogen/deuterium exchange (HDX) experiments.
Infrared Spectroscopy: Simple Yet Powerful
The most simple approach in infrared spectroscopy of proteins is to record absorption spectra of a protein sample prepared in normal (H2O) or heavy (2H2O) water. The use of heavy water as a solvent removes the strong H-O-H bending vibration (~1640 cm-1) which largely overlaps with the amide I band, thereby allowing for a more accurate secondary structure analysis.
In research on the Aβ peptide, infrared spectroscopy provides a quick diagnostic tool to observe formation of oligomeric assemblies from monomers in random coil conformations [3,4]. Figure 2 shows the absorption (bottom) and second derivative (top) spectra of synthetic Aβ42 at alkaline (A, p2H > 11) and neutral (B, p2H = 7) p2H values. Absorption bands appear as negative, narrower signals in second derivative spectra. Under alkaline conditions, Aβ42 predominantly adopts unstructured, random coil conformations. The amide I band is, indeed, dominated by a broad absorption at 1643 cm-1, together with a sharp signal at 1673 cm-1 originating from trifluoroacetic acid (TFA), a chemical often used in peptide synthesis. The second derivative spectrum clearly shows the absence of other types of secondary structure. Lowering the p2H to neutral values dramatically alters the amide I lineshape (panel B). The broad absorption at 1643 cm-1 shifts to 1624 cm-1 and sharpens, suggesting formation of β-sheets from random coils. The second derivative spectrum shows appearance of a weaker signal at 1686 cm-1, revealing the antiparallel arrangement of the strands. Although similar results could have been obtained by CD spectroscopy (fig. 2, inset), the information content of an infrared spectrum far exceeds that of a CD spectrum. For instance, the distinction between parallel and antiparallel β-sheets is not clear-cut in CD as it is in infrared. Additionally, the position of the main β-sheet band provides insight on the approximate size and twist of the sheet, and allows to discriminate between narrow/twisted sheets (~1630 cm-1) and wide/flat ones (<1620 cm-1).
Time-resolved Infrared Spectroscopy and "Caged" Compounds
One of the most remarkable advantages brought about by modern Fourier-transform infrared spectrometers is the high temporal resolution achieved by such instruments: high-quality spectra can be recorded within a few milliseconds, resulting in sampling rates of tens or even hundreds of spectra per second. Although this is often not sufficient to monitor the fastest structural changes in proteins or peptides, processes occurring on a slower timescale, such as oligomerisation and aggregation, can conveniently be studied. A pivotal aspect becomes, then, to precisely synchronise data acquisition with induction of oligomerisation/aggregation. A cunning way to induce in situ aggregation of the Aβ peptide is to quickly lower the pH from alkaline to near physiological or slightly acidic values. This can be performed directly in the infrared cell by UV-irradiation of the photolysable compound 1-(2-nitrophenyl)ethyl sulfate, often referred to as "caged" sulfate or "caged" proton (fig. 3A). In unbuffered solutions, this process leads to the sub-millisecond release of a proton per photolysed molecule, inducing acidification of protein samples to pH values that can, in theory, be as low as the pK of the released sulfate anion (~2). This approach, originally developed by Dr. Corrie and Prof. Barth , has been used to monitor structural changes during the early aggregation of short versions of the Aβ peptide [6,7] as well as Aβ40 (fig. 3B).
Pinpoint Accuracy With Isotopic Labels
The intrinsic width of the C=O band originating from individuals residues within a given type of secondary structure, together with the delocalization of each normal mode along several amide groups, makes it so that these cannot be discriminated from one another, but rather give rise to a cumulative band within the amide I region. Through solid-phase synthesis, however, it is possible to "label" specific carbonyls in the Aβ peptide with heavier isotopes, such as 13C, 18O or both. This does not impair the labelled residue from adopting its native secondary structure configuration, but shifts the vibrational frequency of the 13C=O (as well as C=18O and 13C=18O) groups to lower (~ 40-50 cm-1) wavenumbers with respect to their "unlabelled" positions, thereby isolating its signal from the unresolved amide I lineshape. The use of site-specific isotopic labels in Aβ research has successfully been used to determine the organisation of β-strands in aggregates of short versions on the Aβ peptide  and the prion protein  containing 13C=O labels (for a comprehensive review, see ref. ), but combined structural and kinetic information of oligomer formation remains a task for the future.
The use of isotopic labels allows (1) to tentatively assign the type of secondary structure adopted by the labelled residue by observing the redshift induced by hydrogen bonding; (2) to extrapolate kinetic data by using time-resolved approaches, such as the one based on caged sulfate; and (3) to identify close contacts within an unknown structure by measuring the extent of coupling between any two labelled residues. An approach for the study of the Aβ oligomers, based on these rationales, is shown in figure 4.
Although long considered unfashionable, infrared spectroscopy is now becoming more and more popular in amyloid research and is bound to become one of the approaches of choice in research on Aβ. Significant advantages over other more widespread techniques include:
- Very low sample requirements: Experiments can be performed with as little as 1-2 µg of sample.
- High temporal resolution: The kinetics of stable or transient species can be detected and characterised with millisecond resolution.
- High information content: An infrared spectrum contains both structural information and information on amino acid side chains, which allows to study several processes, e.g. binding of ligands.
- Low running costs: Research-grade spectrometers are available for under € 20,000 and chemicals required for infrared measurements are relatively inexpensive (100 mL 2H2O < € 100).
 Benilova I. et al.: Nature Neuroscience 15(3), 349-57 (2012)
 Haass C. and Selkoe D.J.: Nature Reviews Molecular Cell Biology 8, 101-112 (2007)
 Cerf E. et al.: Biochemical Journal 421(3), 415-23 (2009)
 Sarroukh R. et al.: Biochimica et Biophysica Acta 1828(10), 2328-38 (2013)
 Barth A. and Corrie J.E.: Biophysical Journal 83(5), 2864-71 (2002)
 Perálvarez-Marín A. et al.: Journal of Molecular Biology 379(3), 589-96 (2008)
 Mandal P. et al.: Journal of Physical Chemistry B 116(41), 12389-97 (2012)
 Petty S.A. and Decatur S.M.: Journal of the American Chemical Society 127, 13488-89 (2005)
 Petty S.A. and Decatur S.M.: Proceedings of the National Academy of Sciences 102(40), 14272-77 (2005)
 Moran S.D. and Zanni M.T.: The Journal of Physical Chemistry Letters 5(11), 1984-93 (2014)