Several spectroscopic methods for the determination of protein secondary structure are available. CD spectroscopy is an established technique that works in the far UV region and gives distinct spectra for the structure elements random coil,α-helix and β-sheet.
The resulting spectrum of a protein will be a combination of all the structural elements
2. CURRENT STATE OF RESEARCH 2.5. PROTEIN STRUCTURE DETERMINATION
that are present. In addition to structure, protein stability towards heat, pH or solvents can be analyzed with this technique. There are complementary techniques to CD spectroscopy, such as vibrational spectroscopy. These techniques can give additional insight into the details of the protein's secondary structure and changes within such as dierentiation between inter- and intramolecularβ-structures.
Raman- and Infrared spectroscopy are two main vibrational techniques. Raman spec-troscopy is based on a change in polarizability during vibration. This is for example the case for the symmetric stretching vibrations of CO2. The interaction of a molecule with a photon can lead to elastic or inelastic scattering. In the case of elastic scat-tering the photon does not lose energy and no vibrational transition is excited in the molecule. Elastic scattering contributes more than 99%to all scattering events. Vibra-tional transition occurs in the case of inelastic scattering. The two types of inelastic (Raman) scattering dier between a gain (Stokes) or a loss (Anti-Stokes) of energy of the incident photon, depending on the change in the rotational and vibrational energy of the molecule. The most characteristic bands in a Raman spectrum result from the CONH group of the protein backbone, thus revealing details about the secondary struc-ture. Bands of the C=O stretching vibration are found in the amide I region between 1600 cm−1 and 1690 cm−1. The amide II region spans from 1480 cm−1to 1580 cm−1and is the ngerprint region for coupled CN stretching and NH bending vibrations of the peptide group, as is the amide III region at 1230 cm−1 to 1300 cm−1.[153] -helical struc-tures typically have signals at 1662 cm−1 - 1655 cm−1 and 1272 cm−1 - 1264 cm−1 while β-structures produce bands at 1674 cm−1 - 1672 cm−1 and 1242 cm−1 - 1227 cm−1.[154]
There are several examples for the use of Raman spectroscopy in protein structural analysis. In 1986, Vogel and Jähnig determined the content of dierent secondary structure elements in the outer membrane proteins porin, maltoporin and OmpA from E. coli with the help of Raman spectroscopy.[155] Carey et al. describe the technique of Raman microscopy, which uses Raman dierence spectroscopy to monitor structural changes in crystalline proteins while soaking the crystal in a ligand solution.[156] Res-onance Raman dierence spectra of the heme cofactor in Cytochrome c Oxidase were
2. CURRENT STATE OF RESEARCH 2.5. PROTEIN STRUCTURE DETERMINATION
used to identify intermediates in the oxidative cycle that converts molecuar oxygen to water.[157]
2.5.1 IR spectroscopy
In contrast to Raman spectroscopy, Infrared spectroscopy is an absorption spectroscopy.
It gives information on molecule structure, protonation states and solvent-interaction.
Electromagnetic radiation in the infrared-region (the mid-IR region ranges from 400 to 4000 cm−1) is absorbed by a molecule if it induces a vibrational transition. The resonant frequencies depend on bond strengths, on the masses of the atoms and on vibronic coupling.1. Whether a molecule is IR active or not depends on whether the transition to an excited state is linked to a change in dipole moment. This is the case for asymmetric molecules. Molecules with a center of symmetry may, but do not necessarily have to be IR inactive. The symmetrical stretching motion of CO2, for example, can not be detected by IR spectroscopy, whereas the asymmetrical stretching and the bending vibrations can. The magnitude of the dipole moment depends on the bond length and the extent to which the charges in the molecule are distributed.
The important IR region regarding protein vibrations lies between 1000 cm−1 and 2000 cm−1. The wavenumber ν˜ is used rather than the wavelength λ for historical reasons and is connected to the frequencyν via the speed of light νc = ˜ν.
2.5.2 Characteristic IR vibrations of proteins
The CO stretching vibration dominates the amide I vibrations which are located be-tween 1600 cm−1and 1700 cm−1. Due to the fact that coupling occurs between dierent residues the protein structure determines the position of the absorption bands in the amide I region. Structural elements that can be distinguished in the amide I region includeα-helices, 310-helices, parallel and anti-parallel β-sheets,β-turns, γ-turns, ran-dom coils and loops. For a list of secondary structure elements and their characteristic
1Vibronic coupling refers to the interaction between electronic and nuclear vibrational motion in a molecule
2. CURRENT STATE OF RESEARCH 2.5. PROTEIN STRUCTURE DETERMINATION
absorption bands see (Table 2.3). When D2O is used as a solvent in case of H2O this is indicated by referring to the amide I' region.
Table 2.3: Absorption band positions of protein secondary structures in the amide I region.
Secondary Structure Amide I frequency Amide I' frequency
α-helix[158] 1657-1648 cm−1 1660-1642 cm−1 β-sheet[158] 1641-1623 cm−1 1638-1615 cm−1 β-sheet, antiparallel[158] 1695-1674 cm−1 1694-1672 cm−1 β-turn, loops[158] 1686-1662 cm−1 1691-1653 cm−1 aggregated strands,
intermolecular β-sheets[159] 1615 cm−1 aggregated strands,
turns and loops[159] 1685 cm−1
random coil[158] 1657-1642 cm−1 1654-1639 cm−1
open loops[160] 1645 cm−1 1638 cm−1
CO stretching vibrations are not the only vibrations in the amide I region. The widely used solvent H2O shows strong absorption in the mid-amide I region due to HO bending (Figure 2.9). The interpretation of a protein spectrum is dicult if the intensity of the H2O band overlays other bands. To minimize the H2O signal the use of measuring cells with small path lengths can be helpful. However, very high protein concentrations are necessary in return in order to obtain a signicant IR signal from the secondary structures. If the intensity of the H2O band in the amide I region to below A = 1 OD it may be possible to carefully subtract the water signal from the recorded spectrum.
The use of D2O is a common approach to eliminate the H2O signal. Deuterated water absorbs at lower frequencies than H2O, as does HDO Figure 2.9. Additionally, the signal of the protein backbone's CO groups that interact with the solvent shifts to lower wavenumbers when H2O is exchanged for D2O.
2. CURRENT STATE OF RESEARCH 2.5. PROTEIN STRUCTURE DETERMINATION
Figure 2.9: FTIR transmission spectra of H2O, D2O and HDO. Spectra were recorded in a CaF2 cell with a pathlength of 15µm. The absorption of H2O in the amide I region is well above 1 OD. Subtraction of the H2O band therefore is not reliable.
2.5.3 Instrumentation and experimental setup
In dispersive IR spectroscopy, the protein sample is irradiated sequentially with various single wavelengths. In FTIR spectroscopy all frequencies pass the sample simultane-ously and an interferogram is detected. An algorithm called a Fourier transformation converts it into a single beam spectrum. The acquisition of the interferogram requires a Michelson-Interferometer.
A Michelson-interferometer is an optical set-up that consists of a source, a beamsplit-ter, one static mirror, one moveable mirror, a compensating plate and a detector Fig-ure 2.10. The source emits polychromatic radiation which strikes a beamsplitter that transmits half of the beam and reects the other half. The transmitted beam (A) is rst reected by the static mirror and then by the beamsplitter. Beam B is reected by a moveable mirror and then passes through the beamsplitter. The coherence length of white light is very small. Therefore, the optical pathlengths for beam A and B must be equal. A compensating plate ensures that beam A and B pass through the same length of material which the beamsplitter consists of. When beam A and B are combined
2. CURRENT STATE OF RESEARCH 2.5. PROTEIN STRUCTURE DETERMINATION
Figure 2.10: Schematic representation of a Michelson-Interferometer. The beam from the light source is split by a beamsplitter into ray A and ray B. The beams are reected by mirrors and re-combined with a phase dierence that depends on the position of the moveable mirror. The compensating plate ensures that beam A and beam B travel the same pathlengths. Adapted from[161]
again an interference pattern is created. The interferogram passes the sample which absorbs part of the energy. The transmitted part reaches a detector. The detected interferogram is converted to an infrared spectrum by Fourier transformation.
For FTIR measurements in transmission mode a reference spectrum is usually recorded in addition to each sample spectrum. The reference spectrum can be subtracted au-tomatically or manually. The reference and the sample cell consist of an IR-inactive material such as KBr, NaCl or CaFl.
In order to reduce background noise in an FTIR measurement and to increase the protein signal ATR-FTIR spectroscopy can be applied Figure 2.11. The protein is attached to a crystal that is of an inert material, i.e. silicon or germanium. In general, an ATR crystal is referred to as an IRE (internal reection element). The incident beam reaches the IRE at an angle that ensues total reection. At the reection point an evanescent wave penetrates into the sample that covers the crystal surface. Penetration depth depends on the IRE material and is a few micrometers and depicted in. Only the part of sample that is penetrated by the evanescent wave shows up in the resulting
2. CURRENT STATE OF RESEARCH 2.5. PROTEIN STRUCTURE DETERMINATION
Figure 2.11: The sample is placed on an IRE. The IR-beam enters the IRE from be-low and is reected on its surface. An evanescent wave at the reec-tion point probes the sample on the crystal. Multiple reecreec-tions increase the overall pathlength at which the sample is penetrated. Adapted from http://las.perkinelmer.com/content/TechnicalInfo/TCHFT IRAT R.pdf
Material Depth of penetration at 45 ◦, 1000 cm−1 (µm)
Table 2.4: Penetration depths of IRE materials. Table modied from.[162]
FTIR spectra. The number of reections determines the overall pathlength. If the protein is concentrated on the IRE surface by immobilization, precipitation or else, the local concentration is signicantly increased compared to the rest of the sample volume.
2.5.4 Immobilization of proteins on an ATR element
Several methods exist to increase the local protein concentration near the surface of an IRE. Although proteins can be immobilized to surfaces by dierent means[163] this section focuses on protein immobilization via a Polyhistidine-tag (tag). The His-tag is a modication at the protein's C- or N-terminus. It consists of at least six Histidines which have the ability to bind to a Ni2+ or Co2+ ion that is linked to a chelator. The classic applications for this technique are, i.e. protein purication, pulldown-assays and immunolabeling. In ATR-FTIR spectroscopy the His-tag is a convenient tool to immobilize proteins to an IRE surface. Many successful approaches
2. CURRENT STATE OF RESEARCH 2.6. FOLDING STUDIES ON OMPS
were made to immobilize His-tagged proteins to various surfaces. For example, Ataka et al. attached His-tagged Cytochrome c Oxidase to a gold electrode via a linker with a nitrilotriacetic acid head group that is able to chelate Ni2+ ions.[164] Schartner et al. used a similar method to attach the GTPase Ras and photosystem I (PS I) to a Germanium crystal..[165]It is also possible to coat a surface with a lipid layer containing NTA-modied lipids that are able to bind His-tagged protein via Ni2+-chelation.[166]
Due to TtoA and TtOmp85 being solubilized in detergent for the folding studies, the NTA-modied lipid approach was not suitable for this work. Thus, the attachment of a linker with an NTA-head group was performed to immobilize TtoA. The method will be thoroughly discussed in section 5.2.