In-cell NMR
in solution and solid state
Sina Reckel
7th CCPN conference
2nd of August, 2007
Protein detection inside living cells
•
buffers are usually chosen according to best NMR spectra rather than
biological conditions
Æ detection inside living cells provides most natural environment
1
conformational changes
- FlgM, α-synuclein (Pielak group)
3
Posttranslational modifications
in eukaroytic cells
(Serber and Selenko)
2
Protein/ligand-interactions
- STINT-NMR (Burz et al. 2006, JMB 355, 893-897)
Solution NMR approach
•
uniform 15N labelling gives well resolved spectra
•
viscosity of cytosol increases apparent molecular weight only 2fold
BUT:
a)
cell sedimentation prevents longer experiments (>2h)
9 gel entrapment
b)
many proteins are involved in high molecular weight complexes
Æ tumbling rate is too slow and no 15N detection is possible
9 selective methyl group labelling
9 solid state NMR
c)
uniform 13C labelling gives rise to high amount of background
In-cell sample preparation
(1) overexpression and labelling in
suitable cells
e.g. E. coli, Yeast
(2) injection of isotopically labelled protein into either cells
or cell extract
e.g. Xenopus laevis oocytes
Serber et al. (2006), Nature Protocols 1, 6: 2701
E. coli sample preparation
• 50 ml LB culture
• 37°C until OD600= 0.8
cell disruption
• change to 50 ml M9 medium
• induction with 0.8 mM IPTG
• 37°C for ~4h
• harvest and wash cell pellet
• use pellet for solid state or
resuspend in 600ul buffer/D2O
purification
Influence of the expression level
Gel entrapment prevents sedimentation
•
embedding bacterial cells into agarose does not impair spectral
quality
15N,1H-TROSY
cell suspension in buffer
of in-cell U-15N labelled NmerA
cells in agarose
Background signals in U13C labelled samples
¾ Rifampicin to inhibit bacterial protein
biosynthesis
¾ Pulsed field gradient HSQC for filtration
of macromolecules
U13C-labelled Calmodulin, in-cell
Æ similar pattern in uninduced samples
- gradient minimizes resonances from small, fast diffusing molecules while
retaining those of the protein
1% GZ
80% GZ
in-cell CaM
13C-methyl
methionine
In-cell vs. purified
Calmodulin
Methyl groups
- higher sensitivity as compared to 15N-1H spin system
15N-lysine
labelled CaM
13C-methyl
Met labelled CaM
Serber et al (2004)
FK506 binding protein
- methyls as probes for high molecular weight complexes
in-cell 15N TROSY of U-15N labelled FKBP
13C-HSQC
13C-methyl
of
Met labelled FKBP
in lysate
In vitro simulation of unspecific interactions
Titration Experiments
7
6
6
800µM FKBP
5
+180 g/l BSA
"Met. 1,
(1,93/17,0)ppm
intensity
5
4
4
3
2
1
3
2
3
"Met. 2,
(1,94/16,9)ppm
3
"Met. a.s.,
(1,915/17,6)ppm
2
1
1
800µM
FKBP
+96 g/l
BSA
1
1
0
+180 g/l
BSA
+300 g/l
BSA
titrated BSA conc.
+300 g/l BSA
+96 g/l BSA
R. Hänsel
Thioredoxin is part of large complexes
• α-ketobutyric acid serves as precursor substance to label the δ-methyl group
of isoleucines (Kay 2001)
Solid state NMR provides the missing link
•
1st experiments using inclusion bodies were successful
•
at low temperatures even soluble proteins are detectable
In vitro inclusion bodies
13C-CP
In-cell inclusion bodies
spectra
aliphatic
groups
CP MAS, NmerA 225 K
C=O
Cα
*
CP MAS, NmerA 280 K
aromats
200
300
250
200
150
150
100
50
100
0
ppm
50
300
250
200
ppm
0
150
100
50
0
ppm
Selective labelling is the way to go
•
uniform labelling lacks the required selectivity to look at site specific
residues
metabolic side
product
K37
145
15N-CP
•
140
135
130
125
120
115
110
105
100
ppm
spectrum of 15N-lysine labelled NmerA at 235K
attempts to reduce scrambling using DL39 cells were not successful
13C
•
detection provides more sensitivity
natural abundance interferes with protein signals Æ 12C glucose (13C
depleted) as carbon source necessary
1-13C-Ile76
1-13C-Ile7
1-13C-Ile90
1-13C-Ile56
184
13C-CP
182
180
178
176
174
172
170
168
166
164
162
ppm
experiment of 1-13C Isoleucine labelled FKBP at 250K
13C-CP
•
vs. DP experiments
different experiment types allow filtering of more mobile residues
13Cε-Met 49
13Cε-Met 29
free aa
13Cε-Met 29
13Cε-Met 49
13Cε-Met 66
intracellular
scrambling
intracellular
scrambling
13Cε-Met 66
22
20
18
16
14
12
13C-CP
10
8
6 ppm
22
20
18
16
14
12
10
8
spectrum (left) and 13C-DP spectrum (right)
of 13C-methyl methionine labelled FKBP
6 ppm
Necessity of a negative control
•
control experiment is required
¾ to evaluate amount of background labelling
¾ to have a final prove the signals really come from the protein of
interest
Problem: non-induced cells undergo normal cell metabolism
Æ high background thus not comparable!
• rifampicin addition does not show effect on spectra
• induction of an “invisible” protein does not work for solid state, low
temp experiments
Cell survival in solution and solid state
Î Cell lysis in solution can lead to false positive results!
- If protein is released from the cell, signals can arise that are invisible in
the in-cell spectrum
Æ after an experiment supernatant should always be checked again!
- solid state experiments run under MAS conditions and often take several
days Æ shock-freezing inevitable
- plating - tests after 1d at 10,000 Hz spinning revealed 75% survival rate
Future perspectives
I.
expansion to eukaryotic cells Æ Xenopus oocytes
•
cooperation with MPI/Frankfurt
•
experiments in solution have already been shown (S. Serber,
P. Selenko)
II.
optimisation of cell environment inside the NMR tube
•
III.
oxygen deficiency leads to anaerobic metabolism and pH changes
control experiment
•
to evaluate amount of background signals
Acknowledgements
People in Frankfurt
Robby Hänsel
Daniel Basting
Frank Löhr
Christoph Kaiser
Volker Dötsch
Clemens Glaubitz
Zach Serber,
Stanford University