Presentation on theme: "Applications of NMR spectroscopy in structural biology - protein structure determination - ligand screening for drug discovery - analysis of mobility in."— Presentation transcript:
Applications of NMR spectroscopy in structural biology - protein structure determination - ligand screening for drug discovery - analysis of mobility in proteins and protein-ligand interactions - macromolecular complexes - protein folding - imaging
Protein mobility Proteins are not always rigid. In this example, the protein structure changes in the left part upon phosphorylation (=attachment of a phosphate group to the side chain of a specific amino acid). The picture shows the superposition of the backbone structure of the proteins with and without phosphate bound. The structures were determined by NMR. Science 291, 2429 (2001)
Protein structure determination This example shows that two proteins can have a ‘random coil’ conformation, yet bind specifically to each other and, when they bind, assume a defined three-dimensional structure. Spectra a and b are 15 N-HSQC spectra of the individual proteins (black) and the same proteins in the complex. In a, protein A had been labelled with 15 N, but not protein B. (so only peaks of Protein A are visible). In b, protein B had Been labelled with 15 N, but not protein A (so only peaks of protein B are visible). In both spectra, the narrow distribution of 1 H chemical shifts in the free proteins (black peaks) indicates ‘random coil’ conformation. The much wider distribution of 1 H chemical shifts in the complex (red peaks) is characteristic of a folded protein.
How to label a protein with 15 N and 13 C: Express the protein in E. coli, using a medium containing 15 NH 4 Cl as the only nitrogen source and 13 C-glucose as the only carbon source. It’s straightforward and not too expensive. Why label? The most abundant isotopes of nitrogen and carbon are 14 N (>99%) and 12 C (>98%), but 14 N has a spin 3/2 which means that its nuclear magnetization relaxes too quickly to be useful for NMR, and 12 C has spin 0, i.e. no magnetic moment at all. 15 N and 13 C are non-radioactive, naturally occurring isotopes (0.3% of nitrogen and 1% of carbon). They have a spin ½ (like 1 H) and are great for NMR. Compounds enriched to >99% with these rare isotopes are commercially available. Labelling with 15 N and 13 C allows to record 3D and 4D NMR spectra efficiently which yield better resolution than 2D NMR spectra. Hence, larger proteins can be studied by NMR, if the proteins are labelled.
Ligand screening for drug development Recipe: -label protein with 15 N -record 15 N-HSQC spectra with and without ligand -the shifted peaks are from those amino acids for which the chemical environment is changed by the presence of the ligand = great method for the identification of ligand binding sites
Amino-acid selective labelling of proteins Why: The assignment of cross-peaks in 15 N-HSQC spectrum to individual amino-acid residues in the protein takes time, If the protein is uniformly labelled with 15 N. With selective labelling, individual cross-peaks can be assigned very quickly. How: Express protein with a mixture of amino acids, only one or two of them isotope labelled. Example: PpiB is an enzyme in E. coli which isomerizes peptide bonds involving proline. The amino acid sequence is shown below, the 15 N-HSQC spectrum of the uniformly labelled protein at the left. Labelling with 15 N-arginine (commercial compound) results in a 15 N-HSQC spectrum with only 5 cross-peaks.
15 N-HSQC spectrum of 15 N-arginine labelled PpiB It’s a much simpler spectrum, but we still need to know which peak belongs to which arginine in the amino acid sequence. Assignment of the 15 N-HSQC cross-peak of Arg87 by double-selective labelling Recipe: -protein made with 15 N-arginine (R) and 13 C-alanine (A) -only Arg87 is preceded by an alanine, i.e. only the 15 N of Arg87 couples to 13 C -an HNCO spectrum transfers magnetization selectively from H N to 15 N and 13 C, and contains only one cross-peak
H N - 15 N cross peaks in 15 N-arginine labelled PpiB H N - 15 N cross peak of Arg87 in 15 N-arginine/ 13 C-alanine double-labelled PpiB Addition of a ligand (signals from the ligand are circled). Only the cross-peak of Arg87 shifts, showing that the ligand binds near Arg87. HSQC HNCO HSQC + ligand Arg87
Screening for protein-binding ligands, if the protein is very large Transverse magnetization relaxes much faster for systems of high molecular weight than for small molecules. Fast relaxation means broad NMR signals that disappear during longer pulse sequences. In these NMR spectra, the protein signal has been suppressed intentionally (by relaxation), so that only the signals from the ligand cocktail are left. Quart. Rev. Biophys. 32, 211 (1999) Mixture of 9 small molecules Mixture of 9 small molecules + protein. The signals of the molecule binding to the protein disappeared, because the magnetization from small molecules relaxes like the protein, once the molecules are bound. Difference between (a) and (b) NMR spectrum of pure compound
Structures of protein-ligand complexes A is a protein B is a ligand If we know where B binds to A (e.g. from chemical shift changes observed between A with and without ligand), we still don’t know the orientation of B with respect to A. Attachment of a paramagnetic ion to A causes enhanced relaxation (and therefore signal broadening) of the protons nearby.
Example: protein-DNA complex J. Am. Chem. Soc. 125, 6634 (2003) in which orientation does the protein bind to DNA? recipe: synthesize DNA with a chemical group which binds metal ions (here: EDTA). Two different DNA molecules were synthesized, with the EDTA group at opposite ends (green and red). Ca 2+ is not paramagnetic. Mn 2+ is paramagnetic. 15 N-HSQC spectra of the 15 N-labelled protein show that some cross-peaks broaden in the presence of Mn 2+.
Hydration dynamics J. Am. Chem. Soc. 111, 1871 (1989) and Science 254, 974 (1991) NOESY experiments have shown that the residence times of hydration water molecules on protein surfaces is much shorter than 500 ps. Therefore, protein solvation presents no kinetic hindrance to protein function. Here is the proof: Part of a NOESY spectrum recorded with a special type of water suppression, so that NOEs between the water protons and the protein protons can be observed: position of the water resonance Surprisingly few cross-peaks are observed between the water and the protein, because only those NOEs show up which are to water molecules buried deeply inside the protein structure. NOEs with surface hydration water can be observed in a peptide which is too small to have internal water molecules. The NOESY cross-peaks with water are weak and negative even under conditions, where all intra-peptide cross-peaks are positive. NOE theory shows that negative cross-peaks occur only within small molecules or for very short-lived intermolecular interactions. Cross-section through NOESY cross-peaks with water Conventional 1 H NMR spectrum
Protein folding Protein folding can be studied by NMR spectroscopy in different ways (Acc. Chem. Res. 31, 773 (1998)). Here is one: - freeze-dry protein - redissolve in 100% D 2 O - measure NMR spectra of amide protons as a function of time The signal intensity of the amide protons will go down as they exchange with deuterium from the solvent. Even the amide protons most deeply buried in the interior of the protein structure exchange with time, because protein structures unfold every now and again, exposing the amides to the solvent. This amide proton exchange experiment thus yields information about the most stable and least stable parts of the structure. Under suitable conditions, the hydrogen exchanges each time the protein unfolds. In this way the frequency of unfolding events can be measured. H N exchange experiment, monitored by a series of 2D COSY spectra J. Mol. Biol. 160, 343 (1982)
Imaging (MRI) Anatomical images Usually, only the 1 H NMR spectrum of water is recorded; the contrast in the images is based on different water properties: concentration, T 1 relaxation time, T 2 relaxation time, flow rate.
recipe: take the difference between two images which are recorded while the subject does and does not perform a certain task (e.g. finger tapping) the way it works: Blood vessels widen in brain areas of increased activity. With the in-flow of blood, concentrations of oxygenated hemoglobin rise. The iron in hemoglobin is paramagnetic. The 1 H NMR signal of water relaxes more quickly in the Presence of a paramagnetic ion. Functional imaging by MRI Angiography by MRI recipe: inject a paramagnetic compound into the blood vessel to enhance the relaxation rate of water in the blood
Principle of MRI Record 1 H NMR spectrum of water in the presence of a magnetic field gradient. This means that the magnetic field B 0 (and, hence, the Larmor frequency of the protons) varies as a function of coordinates -> the frequency spectrum shows the water distribution as a function of coordinates.
Simple pulse sequence to image a 2D plane: - a frequency-selective pulse applied during a B 0 gradient in the z-direction excites only spins with a certain z-coordinate - recording the FID in the presence of a B 0 gradient in the y-direction yields the water signal intensity as a function of the y-coordinate - the experiment is performed as a 2D experiment, where the intensity of a B 0 field gradient in the x-direction is systematically increased from FID to FID. (Increasing the gradient strength is equivalent to increasing its duration, but not its strength.) - 2D FT yields a picture of the water distribution in the xy-plane. A mathematical description can be found at