Contents

• Data processing

• 2D and nD data

• Making things look nicer

• Experiment examples

• Triple resonance experiments

• Metabonomics

• Screening

• Other software

• AMIX

• AUREMOL

2D processing

Workflow:

• Transform – xfb

• Phase correct

• Baseline – abs1, abs2, abs2.water

• Set contours – levcalc

• Peak pick – pp

• Integrate – int2d

Making it look nice

• Improved spectra = improved peak picking

• QFIL baseline correction in F1 to remove water stripe

• WATERGATE to improve water suppression – wg in pulse sequence / parameter set name

• Strip transforms to transform only part of the spectrum – particularly useful for 3D, and 15N HSQC

• T1away! Nonlinear processing, so bad for integration, but may be useful for peakpicking

Water “suppression” - QFIL

• Parameter bc_mod controls baseline correction

• NO= nothing

• QUAD=dc offset correction

• QPOL/QFIL = filter out zero frequency component

• Need to also set bcfw

• Use COROFFS if water not in the centre of the spectrum

Cropping the spectrum – strip transform

•Usually, centre of detected dimension is water frequency

•Usually, detect amide protons – all have shift > 4.7

•Less points = faster data manipulation

•Useful when viewing cubes/planes of cubes

•Parameters:

•STSR = first output point number, zero = left of spectrum

•STSI = number of output points, SI/2 = left half

•Note that points relate to SI , so to show only right half of dimension, set STSR to SI/2 and STSI to SI/2

More dimensions

• As 2D, but:

• Need to get phase parameters from 2D planes

• Only real part of processed data kept (so phase in advance)

• Strip transform more commonly used (reduced processed data size)

• Note that processing takes time!

• ftnd command handles up to 6D data, at least

• Slices of nD data can be processed with xfb

• Peak picking/integration still possible

FTND

• General FT of nD data

• Options – ftnd a b c

• a = dimensions to process and order, e.g. 31 processes f1 and f3, 0 = process all directions in default order

• b = output procno

• c = dlp – use delayed linear prediction

• Example: ftnd 321 998 dlp

• Transform f3, f2, and f1, store result in procno 998, use dlp.

• DLP ensures no distortions arise from linear prediction

• In Topspin 1.3, use ft3d for 3D data

Extracting subsets

• For nD data, one may want to view a subset, e.g. 2D plane from 3D cube

• projcbp, projcbn, sumcb = positive, negative and sum cube projections

• projplp, projpln, sumpl = positive, negative and sum planes

• To extract single planes/cubes, use xfb/ft3d

• Promts for plane/cube orientation and number

Peak picking/integration

• pp brings up appropriate dialogue box – any number of dimensions

• Set thresholds using contour levels – remember to save!

• Check and adjust manually!

• Correlated windows useful

• Can import peak list from another dataset for comparison

• Can annotate peaks

• int for integration

Experiments – naming/information

• Pulprog.info – describes Bruker naming system

• NMR guide – pulse sequences, description, theory

• 3D triple resonance manual (help->manuals)

• Examples:

• HNCAGP3D – both intra- and inter-residue HN->N->Ca correlation

• HNCOCAGP3D – only inter-residue HN->N->Ca correlation

• C_CANCO_3D – carbon-detected Ca->N->CO

Pulprog.info

•Lives in $topspinhome/exp/stan/nmr/lists/pp

•Can see from edpul in 2.1

•Describes two-letter codes used for naming Bruker sequences

3D/triple resonance experiment families

• Backbone assignment, e.g.:

• HNCO – inter-residue connection

• HNCA – both intra- (strong) and inter-residue (weaker)

• HN(CO)CA – only inter-residue

• HN(CA)CO – both

• Backbone-sidechain, e.g.:

• HN(CO)CACB – Ca / Cb have opposite phase

• TOCSY experiments for whole sidechain

• Coupling constant measurements – IPAP

• NOESY type experiments

Worked examples – HNCA/HN(CO)CA

• HNCA shows intra- and inter-residue correlations

• HN(CO)CA only shows inter-residue correlations

•1JNCa = 11 Hz

•2JNCa = 7 Hz

•1JNCO = 15 Hz

•2JNCO = <2 Hz

Example: Hymenistatin

• 8 residue cyclic peptide

• Sequence: -Pro-Pro-Tyr-Val-Pro-Leu-Ile-Ile-

• No terminal residues, but prolines (no NH group) provide key to assignment

• Few peaks, so can work with 2D planes of 3D experiments

• H-Ca plane of HNCA and HNCOCA for backbone assignments

• H-N plane for amide N

Interpreting HNCA/HN(CO)CA pair

Green = intra

HN(n)->Ca(n)

Black = inter

HN(n)->Ca(n-1)

Interpreting HNCA/HN(CO)CA pair

Vertical correlation:

HN(n)Ca(n) <-> HN(n)Ca(n-1)

Horizontal correlation:

HN(n)Ca(n) <-> HN(n+1)Ca(n)

Interpreting HNCA/HN(CO)CA pair

• Follow connections – vertical, then horizontal, then vertical – move to increasing residue number

• Ca peak only in HNCA:

• next residue is proline

• next residue is N-methylated

• current residue is terminus

• (inter) Ca peak missing in HNCA = preceding residue is proline

• No peaks seen for a proline residue preceding another proline

-Pro-Pro-Tyr-Val-Pro-Leu-Ile-Ile-Interpreting HNCA/HN(CO)CA pair

Interpreting HNCA/HN(CO)CA pair

Other aspects of biological NMR

• Bio-NMR is not just 3D triple resonance!

• Screening experiments to observe ligand binding

• SAR by NMR (15N HSQC)

• Saturation transfer difference

• WATER-LOGSY

• Metabonomics

• Study of relationships between metabolites and diseases etc.

• Can be combined with LC/MS

Screening - STD

• Saturation transfer difference (STD) – identify ligands which bind to your protein

• Does not require much protein (has been done on 1 nmolar!)

• Observe ligand signals – large excess of ligand useful

• No labelling required

• Can observe competition between ligands

• Some information about binding site on ligand

• Suitable for large proteins (ideally >20kDa)

Patent: Bernt Meyer, University of Hamburg, Germany

Ref: Mayer & Meyer JACS 123 (2001) 6108-6117

Review: Angew. Chem. Int. Ed., 42 (2003, ) 864-890

1. Selective irradiation of a well

separated protein signal

2. Spin diffusion quickly spreads the

saturation to all protein-protons

3. Intermolecular NOE transfers

saturation to ligand-protons at the

binding site

4. Exchange between bound

(saturated) and free ligands

allows further ligand saturation if

ligand in excess

3. intermolecular NOE

1. selective irradiation

2. spin diffusion

protein

ligand

H

HH

HHH

H

H

H

H

H

H

HH

HHHH

H

HH

H

H

H

H

H

H

H

HH

H

STD – how it works

STD - results

• Screening: identification of the binding component(s) from a mixture through positive NOE signals

• Epitope mapping: parts of the ligand in contact with the protein give strongest response

• Competition: add a known binder, response from candidate reduced if it binds at correct site

STD - implementation

• Parameter sets e.g. STDIFFESGP

• Frequency list: protein on-res (e.g. <0 ppm – away from ligand) and off-res (e.g. –20ppm)

• Can include CPMG (t2) filter to remove broad protein signals (e.g. stdiffesgp.3)

• Standard bruker sequences make pseudo 2D

• Need to take the difference afterwards

• Sequence available with automatic subtraction (i.e. output spectrum is difference) on request

Metabonomics

• Statistical analysis of metabolite solutions, e.g. urine, plasma

• Need good quality data!

• Parameter sets available in TS2 which take full advantage of hardware/software improvements

• Take care of temperature regulation/shimming

• Water suppression – noesygppr1d

• Good suppression, integratable spectra

• Not a NOESY! Mixing time short, d8=10msec

Metabonomics parameter sets

• MET_NOEGPPR1D – noesy presat

• MET_DIFFUFILT – diffusion filtered (remove smaller molecules)

• MET_CPMGPR1D – CPMG t2 fiter (remove protein signals)

• MET_COSYGPPR – COSY, with presat

• Parameter sets use improved digital filters

• Noesygppr1d spectra – use akp0.noe for phasing

Example – water suppression

AUREMOL

AMIX – mixture analysis/statistics