Protein NMR terminology COSY-Correlation spectroscopy Gives experimental details of interaction between hydrogens connected via a covalent bond NOESY-Nuclear.

Protein NMR terminology

COSY- Correlation spectroscopyGives experimental details of interaction between hydrogens connected via a covalent bond

NOESY-Nuclear Overhauser effect spectroscopyGives peaks between pairs of hydrogen atoms near in space (1.5-5 Å)(and not necessarily sequence)

CCHHCNHH

CHHNH

C N

O

C

H

C

CH3

H

N C

O

H

C N

H

CH2

CHH3C CH3

O

H

CH2

H

COO-NH 7.09.0 7.0

1

2

3

4

5

Fingerprint region

Ala

COSY

TOCSY

H

H

H

NOE NOE

Walkalongthesequence

N

H

C

H

C

O

CH2

N

H

C

dαN

dαN - Connects CαH of residue i to NH of i+1dN - Connects CH of resdiue i to NH of i+1dNN - Connects NH of residue i to NH of i+1

dNNdαNdNN

dN

dN H

CH

H3C CH3

C

O

N

H

C

H

CH3

dN

dNN

dαN

Connectivites by NOE

LEU VAL ALA

αHi-NHi+1αHi-NHi+3

H

i

i+3

Hi+2

NOE

HN

An α-helix can be recognisedby repeating patterns of shortrange nOes. A short range nOeis defined as a contact betweenresidues less than five apart inthe sequence (sequential nOesconnect neighbouring residues)

For an α-helix we see αHi-NHi+3

and αHi-NHi+4 nOes.

i+4

Assignment of secondary structural segments

• sequential stretches of residues with consistent secondary structure characteristics provide a reliable indication of the location of these structural segments

A -strand is distinguished by strong CαHi-NHi+1contacts andlong range nOes connecting the strands.

A long range nOe connects residues more than 5 residues apartin the chain.

A real example.

The rat fatty acid acyl carrierprotein. Involved in fatty acidbiosynthesis and part of alarger subunit, the synthase,

Is it structured by itself??

GDGEAQRDLVKAVAHILGIRDLAGINLDSSLADLGLDSLMGVEVR

QILEREHDLVLPIREVRQLTLRKLQEMSSKAGSDTELAAPKSKN

NHi-NHi+1

αΗi−NHi+1Ηi−NHi+1

NHi-NHi+2

αHi-NHi+2

αHi-NHi+3

αHi-NHi+4

αHi-Hi+3

NHi-NHi+1αΗi−NHi+1

Ηi−NHi+1

NHi-NHi+2

αHi-NHi+2αHi-NHi+3

αHi-NHi+4

αHi-Hi+3

0- 00000- - - - - - - - - +- 0- 0- - 0+- - +0+- - - +00+- 0- - - - - - + ++ - - - - - - - +- - +++++ +++- +++ - - - - - - - - - - -

- - - - - 0+- +0++- - 0- - 00+- - - - - - - - 00000000+0+00- 00- - ++- +++ - - - - +- + - - +- +++++++ +++

D D D D D

D D D DDD

DD D

D D DDD

CSIJ

CSIJ

Summary of the Sequential and Secondary NOEs observed forrat FAS ACP - most definitely structured

So I have assigned the NMR spectrum and connected the amino acids. I have a good idea of the secondary structure.What next??At this point we notice there are still many nOes we have not assigned on the 2D spectrum. These are neither sequential or short range. They are long-range and connect residues more more than 5 amino acids apart (But still close in space!).

NH2CHCH2CHOOCH2COHO

Asn Gly

Glu

HNCHCHO

HNCHCCH2OCNOHH

Identified as an asparagine amino-hydrogen from COSY spectra

NOE indicated the asparagine amino-hydrogen is near a glutamate acidic hydrogen

Schematic showing long range nOes in the lac headpiece protein

What next? STRUCTURE CALCULATIONS

•From NOE I know close atom-atom distances, but that doesn’t give a structure

•The information you have up to this stage is a list of distance constraints

•The structure can be determined by inputting this information to computer minimization software.

•The computer program also contains information about amino acids, bond lengths/angles and standard information about atom-atom interactions such as minimum distance (i.e. Van der Waals radii)

•With all this information you can generate a model of the structure.

Important: NMR gives you a number of possible solutions (all almost identical, rmsd <1Å), This can range from 5-20 models

X-ray crystallography give one average structureNMR structures can be averaged to give one average structure as well

! Thr7 NH

assign (resid 7 and name HN )(resid 75 and name HD1* ) 4.0 2.2 0.5

assign (resid 7 and name HN )(resid 75 and name HD2* ) 4.0 2.2 0.5

! Leu10 NH

assign (resid 10 and name HN )(resid 75 and name HD2* ) 5.0 3.2 0.5

assign (resid 10 and name HN )(resid 75 and name HD1* ) 3.3 1.5 1.0

!Arg72 NH

assign (resid 72 and name HN )(resid 31 and name HD1* ) 5.0 5.0 0.5

assign (resid 72 and name HN )(resid 31 and name HD2* ) 3.3 1.5 0.5

assign (resid 72 and name HN )(resid 31 and name HB* ) 4.0 4.0 1.5

assign (resid 72 and name HN )(resid 31 and name HA ) 4.0 4.0 1.0

! Leu 75 NH

assign (resid 75 and name HN )(resid 10 and name HD1* ) 4.5 4.5 1.0

Excerpt from an NOE table for Actinorhodin Polyketide ACP - 1997This file contained ~ 700 lines of nOe restraints

The simulated annealing protocol - begin by simulating a 1000Kheat bath and generate an extended model strand

Apply the distance restraints from the NOE data (perhaps 1000restraints for a protein of 90 amino acids). Weight the nOes tofavour the formation of local secondary structure and later longrange structure. Allow chain to move through itself

Start

30 ps

Start to cool the system and increase the penalty for bad contacts.

Minimize the final structure to see if it satisfies all the nOes

20 ps

A simulated annealing trajectory over the first few picoseconds

4 helices begin to‘condense’

Unfolded

Correctly folded

Challenges for Interpreting3D Structures

• To correctly represent a structure (not a model), the uncertainty in each atomic coordinate must be shown

• Polypeptides are dynamic and therefore occupy more than one conformation– Which is the biologically relevant one?

Representation of Structure Conformational Ensemble

UncertaintyRMSD of the ensemble

Neither crystal nor solution structures can be properly represented by a single conformation

Intrinsic motions

Imperfect data

Representations of 3D Structures

C

N

These 2D methods work for proteins up to about 100 amino acids,and even here, anything from 50-100 amino acids is difficult.

We need to reduce the complexity of these 2D spectra.

1

HN

1 6

O

1 2

C

1 4

N

1 2

Cα

1 2

Cα

1 4

N

1

HN

1

Hα

R1

1 6

O

1 2

C

1

Hα

R2

We can start by replacing 14N with 15N, a spin 1/2 nucleus.

Run a ‘COSY’ type experiment that correlates an amide protonwith the 15N nuclei.

This is a heteronuclear experiment, I.e. we are looking at twodifferent nuclei, a 1H and a 15N nucleus. The ‘COSY’ typeexperiment is beyond the scope of these lectures but is knownas HSQC, or heteronuclear single quantum coherence spectroscopy.

This refers to how the magnetisation is transferred from the 1H to the15N.

So how well dispersed are the 15N shifts? Is it worth trying to separateour spectra out based on their differences?

1H-15N HSQC of rat FAS ACP

Why?•The more we understand about a protein and its function, the more we can do with it. It can be used for a new specific purpose or even be redesigned too carry out new useful functions (biotechnology & industry).

•We can use this knowledge to help understand the basis of diseases and to design new drugs (medicine & drug design).

•The more knowledge we have how proteins behave in general, the more we can apply it to others (protein families etc)

A case study - Leukocyte function associated protein-1 (LFA-1)

This protein is involved in tethering a leukocyte to a endothelium,allowing migration through the tissue to a site of inflammation.

One domain of LFA-1, the I-domain is 181 amino acids and undergoes a conformational change where helix 7 slides down theprotein, switching it into an active open form. This open formis competent for cell surface binding.

If we can stop this switch, we may have an anti-inflammatorymechanism

Inflammation (chronic) is responsible for asthma and arthritis.

S

N

N

N

O

N

O-O

A B

C

D

O

Developed small molecule inhibitors and test binding

Weak bindingM to mMsee a migration of the peaks

A more successful inhibitor- nM ‘tight’ binding.

See unbound and boundpopulations

Solve NMR structure of complex…

Helix 7 isprevented fromshifting

NMR is a diverse tool with which we can study protein structure.

It gives us information in solution under ‘physiological’ conditions

2D and 3D techniques combined with modern assignment methodshave allowed proteins up to 40 kDa to be solved.

The power of NMR lies not just with its ability to solve structuresbut also its ability to probe binding of ligands and partner proteinsin ‘real’ time.

Many aspects we have not had time to deal with. NMR reveals howproteins move in solution - can see domains flexing with differenttimescale motions. These often correlate with binding patcheson the protein.

Textbook I recommend reading.

J Evans - Biomolecular NMR Spectroscopy.

Chapter 4. Protein Structure, pages 147-174. After p174numerous examples of NMR structures, labelling etc.

Chapter 2. More high level NMR approach - descriptionof how pulse sequences (I.e. COSY, TOCSY, HNCA etc) work.Beyond the scope of the course but may be of interest.

Chapter 3. Details of calculations - for you details not importantbut will give you more of an idea of how we use the NMR data tocalculate the structure.