Lecture 10: NMR Applications

baconossifiedMechanics

Oct 29, 2013 (3 years and 5 months ago)

118 views

Lecture 10: NMR Applications

We may know a
tiny bit

about NMR theory, but what
bioanalytical problems can we adress with NMR?

Protein Structures
: NMR is one of two main ways of
acquiring molecular images of proteins

Organic Analysis: Specific organic molecules give
characteristic signals

in terms of chemical shifts and peak
splitting due to J coupling

Solid state structures: NMR can be used on solids! But we
need to use a special trick, which we’ll talk about

Bioanalytical Applications

Protein/Protein, Protein/Ligand Interactions
: NMR can be
used to ‘track down’ binding interfaces

Protein dynamics:
NMR is the only way to get atomic level
information about how proteins move.

So You Want To Be An NMR Spectroscopist…

If you’re a bioanalytical NMR spectroscopist, here’s the
typical runup to an experiment:

1) Grow up your protein with the appropriate label. You’ll
either be expressing your protein in
bacteria

(probably
E.
coli
) or
yeast

(probably
S. Cerevisiae
)

For 2D NMR: Probably
15
NH
4
Cl
or

13
C
-

-

-
G汵cose

For 3D NMR: Probably
15
NH
4
Cl
and

13
C
-

-

-
G汵cose

2⤠Purify your 污be汥d protei渠⡰robab汹 His
6

or GST tag)

3) Dilute to desired conc. (probably in water, around 1


10
mM), add 5
-
10%
D
2
O
, 100 uM
DSS

4) Make sure the protein is stable under these conditions,
place in
NMR tube
, throw into instrument

So You Want To Be An NMR Spectroscopist…

6)
Shim

the instrument: We have to make the magnetic field
perfectly homogenious across the sample or equivalent
nuclei will have
different spins
!

The computer can do this for you using the ‘gradient’ approach discussed
last time.
Shims

are extra magnetic coils with their fields pointed in
essentially every direction relative to the sample. They can therefore ‘add’
or ‘subtract’ to the big huge magnetic field as needed to even it out.



shim x

shim z

5) Set the temperature, check tuning/matching

So You Want To Be An NMR Spectroscopist…

7) Calibrate the hard pulse length for 90
°
:

time to
360
°
/4!!

-
y

z



z

-
y



8) You’re ready to go! Load and calibrate the experiment you
want to do!! You may have to work out the appropriate
power and/or duration of certain soft pulses, depending on
the experiment and water suppression scheme.

Good tutorial on biological nmr:
http://www.nmr.sinica.edu.tw/Cours/Course20040906/NMRExperiments_
LargerMolecule.pdf

More on Gradients: DOSY

We’ve said that you can destroy magnetization using a
gradient pulse. But you can also reconstitute it by using the
same pulse

with the
opposite phase

at a later time:

Of course, this will only work if the nuclei remain
essentially
stationary

over the course of the wait period. This is the
basis for
diffusion measurements

(DOSY) by NMR!

Water Suppression

Biological experiments are carried out in water. If we want
to see protons from our sample we’re going to need to
strongly suppress

the water signal.

Here are a few ways of doing that:

Watergate
:

Selective ‘soft’ pulses on
H
2
O protons

1
H

G
z

Defocus
everything

Defocus water,
refocus not water

3
-
9
-
19
Watergate
:

Water Supression

Flipback Watergate:

Puts water on z before first gradient
pulse

Pre
-
saturation:

Lengthy, continuous ‘soft’ excitation
of water offset

NMR of Peptides

Now that we’ve suppressed our water signal, we can take
some spectra of peptides in water.

If we do a simple water suppression pulse
-
acquire experiment,
we may see something like this:

NMR of Peptides: TOCSY

That can be useful


we have some idea of what we’re
looking at… but which peak corresponds to which specific
proton?

TOCSY

TOCSY
tells us
which
amino acid

belongs to
which peak

NMR of Peptides: NOESY

But we still don’t know the amino acid sequence. For that
we need to look at ‘through space’ interactions:

NOE

NOEs are a
relaxation effect
.
As such they are dependent
on the
correlation time
:

1D Saturation Transfer

Saturation transfer is a simple technique that can be used to
determine if and
how

something
small

is binding to
something
very big
.

Saturation here is the same as pre
-
saturation in water suppression. It
involves continually hitting a select
frequency with a
train

of soft pulses:

Sat. pulse

Protein NMR: We Need More Nuclei

1D and H
-
H coupling experiments are all well and good
when you’ve got < 200 protons, but proton signals are not
really all that well
dispersed
.

We’re going to need to use couplings between two different
nuclei (heteronuclear NMR). Since we’re dealing with
proteins our options are most likely
13
C (very expensive!)
and
15
N (expensive).

The most dispersed signals would be the carbons, but
nitrogen is much cheaper!

Thus, the most common type of protein NMR spectrum is
an HSQC which usually correlates the
amide nitrogen

with
the
amide proton
. Thus there is
one peak per residue

(except
proline!!)

A
13
C Protein Spectrum

Here’s a
13
C Spectrum of an SH3 domain (aprox. 70 a.a.):

HSQC Spectra

Here’s the pulse sequence for the HSQC experiment:

And here’s the result:

This is ‘the’ HSQC
for properly folded
Sso

Acylphosphatase
(104 a.a.)

I

S

x/
-
x

y

Detecting and Localizing Ligand Binding

Most analytical techniques work hard to tell us ‘if’
something is binding to our protein of interest. NMR not
only tells us that, but
where
!

The most common way of measuring this is by ‘ligand
titration’ experiments which amount to monitoring the
HSQC as a function of ligand concentration.

Low Ligand

Med Ligand

High Ligand

Protein Dynamics by NMR: H/D Exchange

Once you’ve got an HSQC, you can study slow (minutes to
days) conformational dynamics by NMR.

To do this, you calibrate your HSQC for speed, ‘buffer
exchange’ your protein into 90%+ D
2
O,
RUN

to the NMR
instrument, drop your sample in, quick re
-
shim and
GO
!!

Here’s the result:

No D
2
O

1
st

HSQC after D
2
O

t = 60 min

HDX results

Since we know which backbone protons correspond to
which signals, we can identify which are more protected:

H/H Exchange: CLEANEX

CLEANEX is a cool HD exchange technique that uses water
protons instead of
2
H!

H

H

H

O

H

H

H

H

Backbone and water protons are exchanging all the time

Instead of exciting all protons except water, we
only

excite
water

These magnetized protons now exchange onto the protein…

And we use
that

magnetization to transfer to
15
N

Results of CLEANEX

In order for CLEANEX to work, exchange has to occur faster
than the relaxation of protons on the protein. This means
mid
-
to
-
low milliseconds range:

Sequencing Proteins by NMR

The HSQC gives us a spectrum in which each amino acid is
distinguishable, but doesn’t tell us much about which amino
acid they are, and in what order. To do that, we need to
extend our analysis into the
13
C plane. 3D NMR!!

Sequential Assignment by NMR

To do ‘sequential assignments’, we use pairs of J
-
coupling
-
based 3D experiments, the most common pair is:

C

1

HNCA

C

2

C

C

O

O

N

N

H

H

HNCOCA

C

1

C

2

C

C

O

O

N

N

H

H

Getting Structural Info: The CSI

The Chemical Shift Index (CSI) is a quick way of assessing
secondary structure:


RESIDUE TYPE HA CA CB CO


Ala 4.35 52.5 19.0 177.1


Cys 4.65 58.8 28.6 174.8


Asp 4.76 54.1 40.8 177.2


Glu 4.29 56.7 29.7 176.1


Phe 4.66 57.9 39.3 175.8


Gly 3.97 45.0
-

173.6


His 4.63 55.8 32.0 175.1


Ile 3.95 62.6 37.5 176.8


Lys 4.36 56.7 32.3 176.5


Leu 4.17 55.7 41.9 177.1


Met 4.52 56.6 32.8 175.5


Asn 4.75 53.6 39.0 175.5


Pro 4.44 62.9 31.7 176.0


Gln 4.37 56.2 30.1 176.3


Arg 4.38 56.3 30.3 176.5


Ser 4.50 58.3 62.7 173.7


Thr 4.35 63.1 68.1 175.2


Val 3.95 63.0 31.7 177.1


Trp 4.70 57.8 28.3 175.8


Tyr 4.60 58.6 38.7 175.7

To your observed shifts, give score:

+1

if >.7 ppm
higher

than CSI value

-
1

if >.7 ppm
lower
than CSI value

0

if within
-
.7 to +.7 of CSI value

Four

shifts in a row at
-
1 HA and/or
+1 CA/CO = minimum for
Helix

Three

shifts in a row at +1 HA and/or
-
1 CA/CO = minimum for

-
stra湤

A汬lot桥r re杩g湳nare de獩杮ated
random coil

Getting Structural Info: NOEs

In NMR, the Nuclear Overhauser Effect is the effect that one
nucleus has on the relaxation of another. The intensity of
this effect is directly related to the proximity of the
interacting nuclei:

Just like in J coupling, NOE coupled nuclei will experience
an
oscillating

phase

at each other’s offsets

.

‘is proportional to’

absolute distance between
the interacting nuclei

correlation function


describes
attenuation (or buildup) of the NOE due
to the relative motions of the nuclei

So the internuclear distance effects the size of the NOE

This tells us which nucleus is interacting with which, but a 3D
experiment (e.g. HSQC
-
NOE) is required to distinguish

THE HSQC
-
NOE experiment

Here’s the most common NOE
-
based experiment for
structure elucidation:

Has the advantage of not requiring
double labeling

Gives us a set of
inter
-
proton
distance contraints

We know which amide proton is
which and which amide protons
are nearby (
1.6


6 Å
).

More Structural Info: Angle Restraints

A network of NOEs from an HSQC
-
NOE is a start, but
there’s plenty it doesn’t tell us, particularly which way the
side chains are pointed.

One ‘cheating’ way is to use

/


va汵es co湳iste湴 wit栠t桥
seco湤ary structures derived from C, but t桥re are weak
co湳trai湴s

A better way is to do a ‘residual dipolar coupling’ experiment
in which the sample is placed in a medium, such as
polyacrylamide or phage coat particles, that causes a
net
alignment with the magnetic field
.

B
0

Residual Dipolar Couplings

Through bond (J) dipolar couplings have well defined
frequencies
called

coupling constants
. It is, in fact, by using
coupling constants that we pass magnetization around
through bonds (such as in a TOCSY).

I



J
1,2

J
1,2

The magnitude of the coupling constant
depends on the orientation of the interaction
with respect to the big huge magnetic field.

By measuring the coupling constant (which
can only be done in alignment media), we can
figure out the bond angle with respect to B
0.

difference between
aligned and
unaligned J

angle with respect
to B
0

Results of Residual Dipolar Couplings

Here’s what RDC’s look like. We have to
run our HSQC without decoupling.

Now we have distance constraints and some
bond angles. Combined, these are sufficient
to allow us to parameterize a model protein
structure
.

The next step is to plug our distance and
angular constraints into a computer program
that uses a
molecular mechanics

force field to
find the lowest energy structures that meet
the constraints.

NMR Structure Results

Here’s what an NMR ‘structural ensemble’ looks like:

You can then take the average structure
or the ‘best’ structure (the one that best
fits the constraints) to give you a final
structure:

Note that, unlike x
-
ray crystallography,
this is a structure from the protein
in
solution
.

There are currently 4,448 protein
structures in the BioMagResBank database.

Dynamics by NMR

Since we’re looking at our protein in solution, it should also
be moving around roughly like it does
in vivo
. NMR will
allow us to get site specific information about these
movements.

NMR is also the only method by which motions on virtually
all relevant time
-
scales are observable:

Virtually every type of
protein motion/activity is
covered.

Very Fast Motions: T
1

and T
2

We talked about longitudinal (T
1
) and transverse (T
2
)
relaxation (biophysicists call them R
1

and R
2
)

To make a very long story short, you can get a general
description of the conformational freedom for nuclei in a
protein by mapping out the
spectral density function

J
(


which is directly related to T
1

and T
2
.

The spectral density function at any particular frequency


i猠re污ted to t桥
order parameter


via t桥 fo汬lwi湧
re污tio湳桩n

An order parameter of 1 indicates complete restriction of fast
timescale motions while
S

= 0 indicates completely
unrestricted motion.


m

= correlation time



= correlation time + timescale of
bond vibration (~ns)

Slower Motions: Relaxation Dispersion NMR

Until recently, there was a big hole in the timescale
accessible to NMR measurements. And it was centered right
on the all important millisecond timescale.

The reason is that the equation for the spectral density function

becomes
underdetermined

when an
additional term to account for
conformational exchange is added.

The answer was developed at nearby U of T in the Kay group.
They advanced a technique called Carr
-
Purcell
-
Meiboom
-
Gill
(CPMG) Relaxation Dispersion NMR.

CPMG relaxation dispersion

The key to CPMG relaxation dispersion is that the
contribution to J(

⤠from R
ex

is suppressed by the
application of a train of


pu汳es.

As a co湳eque湣e, t桥
contribution to
J
(

⤠from R
ex

alone can be measured and the
frequency of the motion causing
R
ex

can be acquired.

Chem. Rev.
2006,
106, 3055
-
3079

A major advantage of CPMG RD
is that is sensitive even to low
population protein folding
intermediates.

Solid State NMR

Remember we said we can’t see big stuff in solution by
NMR because to correlation time is too long and thus T
2

is
too fast. Well, what about solids?

They aren’t tumbling at all, so they have infinitely long

c

and
thus an (almost) infinitely fast T
2
. (Also recall that, after an
initial rise, T
1

goes
down

with increasing

c
).


BUT


this T
2

relaxation is almost all
secular
, meaning that it
is due primarily to dipolar couplings

In solution, random tumbling causes these dipolar couplings,
which are vector quantities, to
cancel each other out

Fortunately, some very clever people thought of another way
of making this happen…

Solid State NMR

In solid state NMR, we tilt the sample to the ‘
magic
’ angle,
which is 54.74
°

relative to B
0
.


=
54.74
°

B
0

And then we spin it around that angle at very high
frequency. Thus the name of this type of NMR


‘Magic
Angle Spinning’.

To be effective, this spinning has to be close to or above the
offset frequency


of t桥 湵c汥i bei湧 ob獥rved.

We don’t use this too much in bioanalytical chemistry… YET!

We’re Done!!!

At times, you might have
felt like this…

But now you’re almost
like THIS!