Experimental Methods for Structure Determination PDF

Experimental methods for structure determination Prof. Heinrich Sticht Bioinformatik Institut für Biochemie Friedrich-Alexander-Universität Erlangen-Nürnberg  Gallery of protein structures from: https://hms.harva...

Experimental methods for structure determination Prof. Heinrich Sticht Bioinformatik Institut für Biochemie Friedrich-Alexander-Universität Erlangen-Nürnberg  Gallery of protein structures from: https://hms.harvard.edu/news/folding-revolution  How are protein structures determined by experiment?  How can we predict protein structures?  How can we analyze and interpret structural information?  How is protein structure related to function?  The Protein Data Bank (PDB)  databank for experimentally determined 3D-dimensional structures  recently also supplemented by computed structure models (accessed 20.09.2023)  Experimental methods used for structure determination There are three major experimental methods for structure determination:  X-ray crystallography  cryo electron microscopy (EM)  nuclear magnetic resonance (NMR)  Structure determination by X-ray crystallography Prof. Heinrich Sticht Bioinformatik Institut für Biochemie Friedrich-Alexander-Universität Erlangen-Nürnberg  Steps of an X-ray structure determination  protein crystallization  irradiation of the crystal with X-rays  collection of a diffraction pattern resulting from scattered X-rays  computation of electron density maps  construction of an atomic protein model  iterative model refinement  Protein crystallization Parameters that critically affect crystal formation:  pH  temperature  protein concentration  protein purity  type of the solvent and precipitant  presence of ions or ligands Principle: Crystals form when proteins precipitate slowly in supersaturated solutions A precipitant (e.g. polyethylene glycol) is used to reduce protein solubility and create the supersaturated state Method: Hanging-drop method  Protein solution in buffer with precipitant (low conc.)  Drop of protein solution on glass plate  Reservoir with precipitant (high conc.)  Vapor diffusion (towards equilibrium between drop and reservoir)  Loss of water in drop  Increasing concentration of precipitant in drop  Supersaturation of protein  Properties of protein crystals A crystal represents an ordered array of proteins The asymmetric unit is the smallest portion of a crystal structure, from which the complete unit cell can be generated by symmetry operations A crystal is built from a large number of identical unit cells: smallest repeating unit having the full symmetry of the crystal structure unit cells are loosely packed against each other large holes or channels between the individual molecules: filled with disordered solvent molecules (usually water) High solvent content in the crystal + only small crystal packing  relatively close to the physiologic structure  Measuring Xray diffraction 5 diffraction pattern crystal is exposed to an x-ray source primary beam of usually monochromatic x-rays (defined wavelength) recording of X-ray diffraction pattern on a detector (image film or electronic detector) crystal is rotated during to get information from different angles x-ray primary beam is scattered by the electrons of the atomic shell during the scattering process, the electron starts oscillating and sends a spherical wave out the wavelength and energy of the scattered wave do not change (elastic scattering)  most of the scattered x-rays cancel each other out  those in certain directions will add up and produce spots on the detector What is the requirement for the emergence of a diffracted beam?  Bragg’s law BC + CD = 2d·sinθ = n·λ primary beam hits the crystal with an incident angle θ the crystal will reflect (scatter) back with the same angle θ a constructive interference (i.e. a detectable diffracted beam) will occur, if 2d·sinθ = n·λ (Bragg's law) where: λ is the wavelength of the x-ray d is the spacing of the crystal layers θ is the incident angle n is an integer  The ‘phase problem’ 6 each diffracted beam is defined by three properties: amplitude: intensity of the spot wavelength: defined by the x-ray source phase: related to its interference (positive or negative) with other beams  all three properties need to be known to calculate the position of the atoms by Fourier transformation  Solutions for the ‘phase problem’ 1) Multiple isomorphous replacement (MIR) : 6 soaking of the crystal with heavy atoms heavy-metals diffuse into the holes/channels and bind to protein side chains no change of structure of the molecule (= isomorphous replacement) use intensity differences to deduce the positions of the heavy atoms in the crystal unit cell calculate phase of heavy metals (works because heavy metals contain more electrons  stronger scattering) positive/negative interference with protein spots  phase of all diffracted beams detector a) Observation: the signal of the protein gets amplified by the presence of the heavy metal  the protein and heavy metal must have the same phase  known phase of the heavy metal (negative phase)  the phase of the protein must be negative (phase known) b) Observation: the signal of the protein is weakened by the presence of the heavy metal  the protein and heavy metal must have the opposite phase  known phase of the heavy metal (positive phase)  the phase of the protein must be negative  Solutions for the ‘phase problem’ 2) Multiwavelength anomalous dispersion (MAD) requires crystals containing heavy atoms (as in MIR) or Selenomethionine (SeMet) instead of Met uses synchrotron radiation (tunable wavelength) Physical background of MAD: o the scattering from an atom is usually largely independent of wavelength o however, each atomic type has a few ‘absorption edges’ around which the scattering varies rapidly (in amplitude and phase) with wavelength o by varying the wavelength around the absorption edge for an atomic type, the contribution from those atoms to the total scattering can be varied from: https://doi.org/10.1038/npg.els.0002722 comparison of the intensity of diffracted beams collected at different wavelengths phase information can be recovered by the same method as for MIR data  The electron density map 7 Fourier- Trans- formation electron density map is obtained from the diffraction pattern by a mathematical transformation (Fourier-Transformation: requires wavelength, amplitudes and phases) contour plot indicating those regions in the crystal where the electrons on the molecule are located quality of the map depends on the resolution of the diffraction data (mainly determined by the quality of the crystals)  Resolution of diffraction data Resolution: Smallest distance of atoms to appear as separate maxima in the electron density plot  At higher resolutions, more sophisticated structural features can be seen: o ~ 6 Å – shape of the macromolecule o < 3 Å – the polypeptide chain can be traced; well- defined water molecules are visible o < 1.5 Å – positions of individual non-hydrogen from: doi:10.1111/j.1742-4658.2007.06178.x atoms start to become resolved (length of a C-C bond ~ 1.5 Å) o < 1 Å – direct location of hydrogen atoms in the electron-density map PDB statistics (Oct 2023) o 0.77 Å – physical limit, when using copper Kα X-ray radiation (λ = 1.542 Å) o < 0.77 Å – only very few protein structures available As a general rule of thumb, we have more confidence in the location of atoms in structures with resolution values that are small, called "high-resolution structures". from: https://pdb101.rcsb.org/learn/guide-to-understanding-pdb-data/resolution Types of structural representations 8 a) Wire model shows the path of the polypeptide backbone suitable for structure comparison (overlay) b) Ribbon model highlights the secondary structure elements c) Ball-and-stick model shows all individual atoms of each amino acid suitable to highlight amino-acid interactions d) Space-filling representation each non-hydrogen atom is shown as a sphere of its van der Waals radius e.g. suitable to identify ligand binding pockets e) Surface representation “smoothed” representation of the protein surface electrostatic properties of the surface are highlighted in color (red = negatively charged; blue = positively charged)  The R-factor R-factor (‚residual factor’ or ‘reliability factor’) 10 measure of the agreement between the crystallographic model and the experimental X-ray diffraction data (i.e., it is a measure of how well the refined structure would predict the observed diffraction data) 0.0 for exact agreement 0.59 for total disagreement usually between 0.15 and 0.20 for a well-determined protein structure Rfree most useful global measure of model-to-data agreement calculated analogous to the R-factor calculated only for ~1000 randomly selected reflections that were never used for model refinement warning if Rfree exceeds R by more than 7% (may indicate over-fitting of the experimental data, or may result from a serious model defect)  Additional measures of quality Root-mean-square deviations (RMSDs) from stereochemical standards 11 indicates how much the model deviates from geometrical parameters that are considered typical, or represent chemical common sense based on previous experience medium-to-high-resolution structures are expected to have a RMSD of 0.02 Å for bond length (high values >0.03 Å indicate errors in the model) Other quality parameters sidechain outliers (percentage of residues with unusual sidechain conformations) steric clashes (number of pairs of atoms which are unusually close together) ramachandran plot outliers (residues with unusual φ/ψ backbone torsion angles)  Comparison of X-ray and Cryo-EM from: 10.1002/pro.3022 because of an electron’s strong interaction with each atom’s Coulomb potential, individual molecules within a specimen can be imaged directly the EM density of a biological molecule can be reconstructed from a set of 2D projections of the molecule with common structure, imaged at various orientations by the electron microscope the magnified image already includes the full structural information of the molecule cryo-EM can be used to examine non-crystalline structures and conformational heterogeneous states structural quality (resolution, R-Factor,...) is calculated as for crystal structures  Structure determination by nuclear magnetic resonance (NMR)- spectroscopy Prof. Heinrich Sticht Bioinformatik Institut für Biochemie Friedrich-Alexander Universität Erlangen-Nürnberg  The nuclear spin Nuclei are composed of neutrons and positively charged protons  the nucleus exhibits a magnetic moment, resulting from the rotation around its axis (= spin)  the type of motion is determined by the spin quantum number I  three different situations exist: 1. The number of protons and neutrons is even:  Nuclear spin I = 0  no resulting magnetic dipole/multipole.  these nuclei are magnetically inactive and not suitable for NMR-spectroscopy  Examples: 12C, 16O, 32S,... 2. The number of protons and neutrons is odd:  integer nuclear spin I > 0  existence of a magnetic quadrupole moment (observed for all nuclei with I > ½ )  hampers signal detection  Examples: 2H, 6Li, 14N,... 3. The number of protons or the number of neutrons is odd:  half-integer nuclear spin I = 1/2, 3/2, 5/2,...  Examples: I = 1/2: 1H, 13C, 15N, 19F, 31P, 77Se, 113Cd, 119Sn,... I = 3/2: 7Li, 9Be, 11B, 23Na, 33S, 35Cl, 37Cl,... I = 5/2: 17O, 25Mg, 27Al, 55Mn,...  I = ½ nuclei are particular suitable for NMR-spectroscopy, because they exhibit a dipole moment, but no quadrupole moment Natural abundancy of the I = ½ isotope  Nuclei in an external magnetic field What happens, if a nucleus is exposed to an external homogeneous magnetic field?  For a nucleus with spin I=½ only two quantized states ±½ exist (generally, the number of allowed states is 2 I + 1).  In the absence of an external field, these two states are degenerated  In the presence of an external field, this degeneration no longer exists  The energetic difference between the two states is proportional to the strength of the external magnetic field.  The energetic difference also depends on the type of nucleus investigated (determined by its gyromagnetic constant γ ) Nuclei in an external field Which phenomenon is detected by NMR-spectroscopy?  The transition between two allowed states (in case of I=½ between the states ±½) How is the experiment done? Irradiation of external energy, which results in a distinct signal  two setups are possible: 1. Irradiation of a constant energy hν; systematic change of the strength of the external magnetic field 2. Constant external magnetic field; systematic change of the irradiated energy hν. (see right picture)  In both setups, the energy absorbed by the sample can be detected as a distinct signal One-dimensional 1H-NMR spectrum Why do protons differ in their resonance frequencies? 1. Different chemical environment (bound to C, N,...) 2. Different spatial environment  buried in the 3D-structure vs. Solvent exposed  ring current effects of aromatic residues (Phe, Tyr, Trp, His)  hydrogen bonds  effect of the secondary structure  One-dimensional and two-dimensional NMR-spectra - The resonances present in the 1D-spectrum are found on the diagonal of the 2D-spectrum - The 2D-spectrum contains additional resonances outside the diagonal (termed „crosspeaks“) - these crosspeaks contain information about the coupling of nuclei Two-dimensional 1H-1H NMR-spectrum of a protein Crosspeaks located on the same horizontal or vertical lines frequently belong to the same spin system (i.e. the respective protons share a „common property“ like belonging to the same amino acid or being close in space) Identification of protons belonging to the same amino acid Approach: Exploit the scalar (‚spin-spin‘) coupling = Coupling of two nuclei through the covalent bonds of the molecule  depending on the type of bond (+ bond geometry)  larger coupling constants allow a more efficient transfer of magnetization  scalar coupling is mainly restricted to protons within the same amino acid Suitable two-dimensional 1H-1H-NMR experiments: - COSY (‚correlation spectroscopy‘): Detects neighbors maximally separated by three bonds - TOCSY (‚total correlation spectroscopy‘): Detects the spin systems of entire amino acids TOCSY Couplings in amino acids Hα2 Hα1 COSY  Identification of protons that are close together in space Approach: Exploit the „nuclear Overhauser effect“ (NOE) Signals of the protons from amino acid i can be detected on HNi+1 - The „nuclear Overhauser Effect“ (NOE) results from dipole-dipole relaxation - The effect can only be detected for protons in close spatial proximity (< 5 Å) - The NOE-Intensity depends on the distance of the protons (I ∝ r -6)  allows an accurate distance measurement - NOE-Information can be used for the structure calculation of proteins and protein-ligand complexes - Typical experiment: NOESY Principle of protein structure calculation  Structure Calculation List of unambiguous structural restraints input into distance geometry or simulated annealing protocol  a set of 30-100 structures are calculated that are consistent with restraints  structures are refined by restrained molecular dynamics or energy minimization Initial structures usually of poor quality due to inadequate numbers of NOEs or incorrectly assigned NOEs.  both problems can be reduced using heteronuclear NMR with isotope-labelled proteins structures help to assign NOEs that were ambiguous, and fix incorrect ones. Repeat this process iteratively. 15-25 “best” structures are selected for NMR model.  Assessing Structural Quality 1998 IUPAC Task Force recommended the following structural statistics be reported: 1. Number and type of NOEs used {intraresidue, sequential, medium range (≤5 residues apart), long range (>5 residues apart), intermolecular} 2. Number of torsion angle restraints 3. Number of hydrogen bond restraints 4. Maximum restraint violation and the average violation per constraint 5. Deviations from idealized geometry (I.e., unusual bond lengths or bond angles 6. Precision of structures: RMSD with respect to the mean structure (backbone versus all heavy atoms) 7. Percentage of residues falling into allowed regions of φ/ϕ space 1 and 6 are the best indicators of structural quality. Goal: a) >20 restraints per residue b) 0.3-0.6 Å rmsd for backbone atoms, 0.5-0.8 Å rmsd for heavy atoms  A guide for judging the ‘resolution’ of NMR-derived protein structures Assessment Very high Medium High resolution Low resolution criterion resolution resolution Restraints per > 18 14–18 10–15 < 10 residuea Backbone rmsd < 0.3 0.3–0.5 0.5–0.8 > 0.8 (Å)b Heavy-atom rmsd < 0.75 0.75–1.0 1.0–1.5 > 1.5 (Å)b Ramachandran Plot quality (%)c > 95 85–95 75–85 < 75 Example PDB file 1TVJ 2IL8 2FE0 1LMM Goal: a) >20 restraints per residue b) 0.3-0.6Å rmsd for backbone atoms, 0.5-0.8Å rmsd for heavy atoms Very rough rule of thumb: an NMR structure calculated with ≥20 restraints per residue is equivalent to a 2-2.5 Å crystal structure But… long range restraints are much more important than medium range, sequential or intraresidue ones for making a high quality NMR structure  A typical representation of an NMR structure 20 structures superimposed, all consistent with the available data Backbone (green) and core side chains (blue) usually better defined than the solvent-exposed side chains (orange) and the chain termini.  Ill-defined regions may indicate conformational dynamics in solution or a lack of data in that region  Dynamics can be confirmed by relaxation measurements  Proteins are not static! Dynamics can be substantial and functionally important  Results can strongly depend on the fitting procedure Ensemble of 20 NMR-derived structures of ω-ACTX-Hv2a, a specific blocker of insect voltage-gated calcium channels. (A) In this view, the structures have been overlaid to minimize the backbone rmsd over all 45 residues; (B) An overlay over the backbone atoms of residues 3–32 only reveals that the poor fit in part (A) results from inclusion of the highly disordered C- terminal region in the rmsd calculation. When this region is excluded, the structured N-terminal core, which includes three disulfide bonds (not shown), is clearly visible.  The size-limitation in NMR-spectroscopy The signal dispersion increases only moderately with protein size  there is a larger overlap of the signals for larger proteins  unambiguous assignment of resonances is hampered for large proteins  in most cases NMR-spectroscopy can only be used to determine the 3D-structure of smaller proteins (< ~250 amino acids) References: (1) http://www.physik.uni-augsburg.de/lehrstuehle/exp4/FP_A/material/FP08.pdf (2) http://schwalbe.org.chemie.uni-frankfurt.de/sites/default/files/oc1p/NMR_Vortrag.pdf (3) http://opus4.kobv.de/opus4-ubbayreuth/files/146/DISS_F_BAUER.PDF (4) „Bioanalytik“ von Joachim W. Engels, Friedrich Lottspeich, Spektrum Akademischer Verlag, ISBN:3- 8274-2942-0 (5) Schweimer K, Hoffmann S, Bauer F, Friedrich U, Kardinal C, Feller SM, Biesinger B, Sticht H. (2002) Structural investigation of the binding of a herpesviral protein to the SH3 domain of tyrosine kinase Lck. Biochemistry 41:5120-30. (6) Bauer F, Schweimer K, Meiselbach H, Hoffmann S, Rösch P, Sticht H. (2005) Structural characterization of Lyn-SH3 domain in complex with a herpesviral protein reveals an extended recognition motif that enhances binding affinity. Protein Sci. 14:2487-98. (7) Bauer F, Hofinger E, Hoffmann S, Rösch P, Schweimer K, Sticht H. (2004) Characterization of Lck- binding elements in the herpesviral regulatory Tip protein. Biochemistry 43:14932-9. (8) “Introduction to Protein Structure” Carl Branden, John Tooze, Garland Science, ISBN: 9780815323044 (9) http://onlinelibrary.wiley.com/doi/10.1002/anie.200300595/abstract (10) http://www.ncbi.nlm.nih.gov/pubmed/21214860 (11) http://www.chem.umass.edu/~thompson/Courses/BioStruct/Fall2011/LectureSlides/Oct13nmr.pdf (12) http://www.mol.biol.ethz.ch/groups/wider_group/publications/WIDERbiotechn_M29.1278.pdf (13) http://www.nmr2.buffalo.edu/resources/edu/matr/nmr2_2004.pdf (14) http://www.cs.duke.edu/brd/Teaching/Bio/asmb/current/2papers/Intro-reviews/flemming.pdf Take-home-messages I: General aspects of X-ray crystallography The steps of an X-ray structure determination include:  protein crystallization  irradiation of the crystal with X-rays  collection of a diffraction pattern resulting from scattered X-rays  computation of electron density maps  construction of an atomic protein model  iterative model refinement Parameters that critically affect crystal formation:  pH  temperature  protein concentration  protein purity  type of the solvent and precipitant  presence of ions or ligands  Take-home-messages II: Interpretation of X-ray structures Resolution: Smallest distance of atoms to appear as separate maxima in the electron density plot o ~ 6 Å – shape of the macromolecule o < 3 Å – the polypeptide chain can be traced o < 1.5 Å – positions of individual non-hydrogen atoms start to become resolved o < 1 Å – direct location of hydrogen atoms in the electron-density map o 0.77 Å – physical limit, when using copper Kα X-ray radiation o < 0.77 Å – only very few protein structures The appearance of electron density as a function available of the resolution of the experimental data Measures of quality include:  R-factor (measure of the agreement between the crystallographic model and the experimental X-ray diffraction data; 0.0 for exact agreement; 0.59 for total disagreement)  Rfree (global measure of model-to-data agreement calculated only for ~1000 randomly selected reflections that were never used for model refinement)  Root-mean-square deviations (RMSDs) from stereochemical standards (indicates how much the model deviates from geometrical parameters that are considered typical, based on previous experience)  sidechain outliers (percentage of residues with unusual sidechain conformations)  steric clashes (number of pairs of atoms which are unusually close together)  ramachandran plot outliers (residues with unusual φ/ψ backbone torsion angles)  Take-home-messages III: General aspects of NMR - most NMR experiments rely on the magnetic properties of 1H, 15N, and 13C - the proton resonance frequency depends on the chemical and spatial environment of the proton - COSY and TOCSY are NMR-experiments aim to identify protons belonging to the same amino acid - NOESY experiments can detect the spatial proximity of protons up to a distance of ~ 5 Å - NOE distance information is the key source of information for the calculation of protein 3D-structures Antiparallel β-sheet stabilized by hydrogen bonds (red dashed lines). Proton pairs, for which NOE signals can be detected, are indicated by arrows.  Take- home-messages IV: Comparison of NMR and X-Ray NMR: X-Ray: - mg-amounts of protein - mg-amounts of protein (pure, homogeneous) (pure, homogeneous) - May require isotope labelling - May require labelling with heavy for larger proteins (15N, 13C) metals or Seleno-Methionine - Size limitation - No size limitation (< ~250 amino acids) - Requires good solubility - Requires crystals (~ 1mg/ml) - Well-suited to study dynamical - Investigation of dynamical processes processes is very difficult