NMR assignments and structure determination
15 likes11,005 views
This document is a comprehensive tutorial on protein NMR (Nuclear Magnetic Resonance) structure determination, emphasizing the importance of NMR for analyzing proteins in solution when crystal structures are unavailable. It outlines the multi-step process from sample preparation to sequential and side-chain assignments, collections of NMR restraints, and detailed analysis of spectra for amino acid identification. Additionally, the tutorial highlights the calculation and validation of NMR structures using various restraints and experimental data, stressing the need for careful refinement and validation of the resulting models.
NMR assignments and structure determination
- 1. NMR TUTORIAL 1I PROTEIN NMR ... from data to structure. for 2nd year Biochemists by Christiane Riedinger
- 2. why do we solve NMR structures ? • ... can’t get a crystal structure :-) • study the protein in solution • obtain information about dynamics in general: • infer function from structure • map interaction sites of ligands • molecular mechanisms C.Riedinger 2009
- 3. The process of NMR structure determination 1. the sample 2. sequential assignment 3. side-chain assignment 4. collection of NMR restraints 5. NMR structure calculation 6. structure validation C.Riedinger 2009
- 4. 1. The Sample • protein of interest (now up to ~60kDa), ideally well folded • high concentration (mM), no aggregation/precipitation, stable over time • isotopically labelled, 15N, 13C, 2H ... • keep salt concentration low, more signal • recent advances for larger / less soluble / less well folded proteins: decrease overlap - multi-dimensional spectra high MW - TROSY effect improve s/n - cryogenic probe-heads C.Riedinger 2009
- 5. 2. NMR assignments (general) example protein • we want to know which atom each peak in our spectra correspond to (that’s MANY!!) • 1st carry out sequential assignment know the sequence! (obtain chemical shift of backbone amides) • then side-chain assignment (obtain chemical shift of all other atoms) C.Riedinger 2009
- 6. 2. sequential assignments (side-chains with amides) •1H/15N HSQC spectrum: one peak for each backbone amide (and side-chains containing amides) • distribution of peaks depends on chemical environment of amide in the protein • aim: assign each peak in HSQC to an amide in the primary sequence of your protein. • in order you achieve this, a set of 3D experiments are acquired... protein sequence C.Riedinger 2009
- 7. concept of a 3D spectrum: • label HN frequencies with the frequency of a third nucleus, for example side-chain carbons • “spread out” HSQC along a third dimension • (of course there are other 3D spectra too, which are not HN based...) C.Riedinger 2009
- 8. concept of a 3D spectrum: 15N 1H • label HN frequencies with the frequency of a third nucleus, for example side-chain carbons • “spread out” HSQC along a third dimension • (of course there are other 3D spectra too, which are not HN based...) C.Riedinger 2009
- 9. concept of a 3D spectrum: 15N C 13 1H • label HN frequencies with the frequency of a third nucleus, for example side-chain carbons • “spread out” HSQC along a third dimension • (of course there are other 3D spectra too, which are not HN based...) C.Riedinger 2009
- 10. concept of a 3D spectrum: 15N C 13 1H • label HN frequencies with the frequency of a third nucleus, for example side-chain carbons • “spread out” HSQC along a third dimension • (of course there are other 3D spectra too, which are not HN based...) C.Riedinger 2009
- 11. generally, two types of spectra are collected: e.g. CBCACONH e.g. CBCANH links HN frequencies with Ca and Cb links HN frequencies with Ca and Cb chemical shifts of previous residue (i-1) chemical shifts of previous residue (i-1) in sequence in sequence and those of itself (i) C.Riedinger 2009
- 12. generally, two types of spectra are collected: Spectrum A: IDENTIFIES amino acid e.g. CBCACONH e.g. CBCANH i-1 i links HN frequencies with Ca and Cb links HN frequencies with Ca and Cb chemical shifts of previous residue (i-1) chemical shifts of previous residue (i-1) in sequence in sequence and those of itself (i) C.Riedinger 2009
- 13. generally, two types of spectra are collected: Spectrum A: Spectrum B: IDENTIFIES amino acid CONNECTS amino acid to its neighbours in the sequence e.g. CBCACONH e.g. CBCANH β α i-1 i i-1 i links HN frequencies with Ca and Cb links HN frequencies with Ca and Cb chemical shifts of previous residue (i-1) chemical shifts of previous residue (i-1) in sequence in sequence and those of itself (i) C.Riedinger 2009
- 14. Analysing the CBCACONH and CBCANH • look along the carbon dimensions for each peak in the HSQC in both 3D spectra (this is called “strip plot”) “strip plot” • for the CBCACONH (A) you will see two peaks, the Ca and Cb of the previous residue in the sequence Cb (i) • for the CBCANH (B) you will ideally see four peaks, the Ca and Cb of the previous residue as well as those of itself Cb (i-1) • by comparing the two strips, you can identify which peaks belong to the previous residue (i-1) or to itself (i). • from the chemical shifts (=the position of the peaks), you Ca (i-1) can take a guess as to which amino acids you are dealing with. Ca (i) A B 13C 15N 13C C.Riedinger 2009 1H
- 15. Analysing the CBCACONH and CBCANH • look along the carbon dimensions for each peak in the HSQC in both 3D spectra (this is called “strip plot”) “strip plot” • for the CBCACONH (A) you will see two peaks, the Ca and Cb of the previous residue in the sequence Cb (i) • for the CBCANH (B) you will ideally see four peaks, the Ca and Cb of the previous residue as well as those of itself Cb (i-1) • by comparing the two strips, you can identify which peaks belong to the previous residue (i-1) or to itself (i). • from the chemical shifts (=the position of the peaks), you Ca (i-1) can take a guess as to which amino acids you are dealing with. Ca (i) A B 13C 15N 13C C.Riedinger 2009 1H
- 16. Identifying Amino Acids from Chemical Shifts • due to their chemical structure, side-chain atoms of amino acids have dispersed chemical shifts • for a complete list of average chemical shifts, see http://www.bmrb.wisc.edu/ref_info/ statful.htm • based on Ca and Cb chemical shifts alone, some amino acids such as alanines, serines/ threonines and glycines can be unambigously identified, others can be at least narrowed down to a few possibilities: C.Riedinger 2009
- 17. Placing strips in the sequence... 13C A B A B A B A B A B 13C • now you compare the strips of different amide peaks with each other • for the i-peaks in the CBCANH, you should find matching peaks in a different strip of the CBCACONH • for the i-1 peaks in the CBCACONH, you should find matching i-peaks in a different strip of the CBCANH • remember: the chemical shifts of side-chain atoms of a residue X will be identical, whether they are “seen” from the position of the previous amide or itself! • now compare a chain of strips and their possible amino acid types to the primary sequence. That’s it! C.Riedinger 2009
- 18. Placing strips in the sequence... 13C A B A B A B A B A B 13C • now you compare the strips of different amide peaks with each other • for the i-peaks in the CBCANH, you should find matching peaks in a different strip of the CBCACONH • for the i-1 peaks in the CBCACONH, you should find matching i-peaks in a different strip of the CBCANH • remember: the chemical shifts of side-chain atoms of a residue X will be identical, whether they are “seen” from the position of the previous amide or itself! • now compare a chain of strips and their possible amino acid types to the primary sequence. That’s it! C.Riedinger 2009
- 19. Real data... In reality it’s not as pretty! C.Riedinger 2009
- 20. 3. side-chain assignments • after completing the backbone assignments, you proceed to the side-chain assignments, i.e. the remaining carbon atoms and hydrogen atoms of your protein’s side-chains • for the HN-based experiments, you might again analyse the data in a strip plot • you can identify the different side-chain hydrogen atoms based on their chemical shift: C.Riedinger 2009
- 21. • here is a list of experiments you might acquire (of course there are many more!): • note: not all of them are HN based, these spectra are analysed by starting with a carbon- hydrogen pair with known chemical shifts • for example, if you know the chemical shifts for the Ca and Ha of a particular residue X, you move to this position in the HCCH-COSY and collect the chemical shifts for the adjacent hydrogens in the side-chain • use the carbon HSQC to collect any final missing assignments, but also to obtain the most precise chemical shift, as this spectrum will have the highest resolution (it’s just a 2D) C.Riedinger 2009
- 22. side-chain assignments The carbon HSQC shows all side-chain carbon- hydrogen pairs in one spectrum. Going through all your data, you will eventually obtain a full list of chemical shifts for all assignable atoms in your protein. 1H/13C HSQC C.Riedinger 2009
- 23. 4. NMR restraints: Define the structure We usually obtain the following structural restrains from NMR data: 1. the NOE (measures distance between two atoms) 2. torsion angles (define the rotation around bonds) 3. Residual dipolar couplings (RDCs) (provide global, orientational restraints) C.Riedinger 2009
- 24. 1. the NOE (nuclear Overhauser effect) • is strictly local phenomenon!!! • measure a change in intensity of one resonance when neighbouring nucleus is perturbed identify peaks • distance dependent! NOE~1/r6 assign to proton-pair • spins that are less than 5A apart calibrate to distance • intensity of cross-peaks ~ separation of the nuclei • translate peak intensity into distance! (calibration required) • separation ranges: 1.8-2.7A, 1.8-3.3A, 1.8-5A (strong, medium, weak) • lower bound = VDW radius C.Riedinger 2009
- 25. 2. Torsion angles • torsion angles, dihedral angle (the angle defined between two planes) • phi, psi and omega(=180*) • define the orientation of the backbone! • torsion angles are related to J-couplings through the Karplus equation • J-couplings = scalar through-bond couplings ... formation of multiplets • measure in Hz • or predict from carbon chemical shifts! (TALOS, part of NMRPipe) C.Riedinger 2009
- 26. 2. Torsion angles C.Riedinger 2009
- 27. 3. Residual Dipolar Couplings decoupled, isotropic J coupling, isotropic • global restraints!! • interaction between two magnetic dipoles • the DC depends on the distance between splitting = J-coupling (Hz) two nuclei (r) and the angle of the bond J-coupling plus RDC! (psi) relative to the magnetic field (B0) using alignment medium • measure the orientation of a bond with respect to magnetic field! splitting = J-coupling + RDC (Hz) C.Riedinger 2009
- 28. C.Riedinger 2009
- 29. 5. NMR Structure Calculation - experimental restraints: NOEs, torsion angles, RDCs - other known parameters: VdW radii, bond angles = “force field” / “target function” - describes potential energy - 3D energy landscape - impossible to enumerate all solns! - aim of structure calculation determine the energetic minimum that combines the empirical force field with experimental restraints. - minimise!!!!
- 30. Minimisation of the Target Function • Cartesian space or torsion angle space • t.a.s: only rotation around angles, bond lengths not affected • molecular dynamics: overcome local energy barriers with Ekin • = simulate elevated temperature! • “Simulated annealing” • start with high Ekin (maximise sampling of conformational space) • then cool down • generate a number of models that are in agreement with the experimental data Note: this picture was obtained from “the internet” a few years ago, so unfortunately, there is no C.Riedinger 2009 reference. All other figures have been made by me.
- 31. The result PDB code: 1Z1M C.Riedinger 2009
- 32. 5. The work is not done... Refinement... Validation... • some structure calculations use simplified force-fields that introduce errors that need to be fixed afterwards • a very useful structure validation tool is part of the “WhatIf” molecular modelling package: http://swift.cmbi.ru.nl/servers/html/index.html Simply submit a PDB file and the software will check for most common errors in nomenclature, packing, planarity, torsion angles... • try your favourite structure of the PDB database and see how good it is! C.Riedinger 2009