Structure of Outer Membrane Protein
OmpG
(and how we got there)
Lukas Tamm
University of Virginia
MEMBRANE PROTEINS ARE ABUNDANT
AND IMPORTANT
BUT KNOWLEDGE DATABASE IS LAGGING FAR
BEHIND THAT OF SOLUBLE PROTEINS
• ~ 30 % of all open reading frames are expected to be
membrane proteins
• ~ 0.6 % of all structures in the PDB are membrane
proteins (280/46,818) – 133 unique MPs
• ~ 60 % of all current drugs target membrane proteins
(channels, GPCRs, transporters etc.)
Outer Membrane Protein A (OmpA) of E. coli
Functions ascribed to OmpA:
• Structural protein: connects the outer membrane to
periplasmic peptidoglycan
• Ion channel/pore: single channel conductance in vitro
response to osmotic stress in vivo
• Receptor for various bacteriophages (K3, Ox2)
• Mediates bacterial conjugation
15N-1H
TROSY NMR spectrum of OmpA TM
domain in DPC micelles at 750 MHz
Fold of the OmpA TM domain by NMR spectroscopy
19 kDa
177 residues
in DPC micelle
of ~46 kDa
Arora et al. (2001)
Nat Struct Biol 8, 334
Interacting Gate Side-Chains in OmpA
See: Hong, Szabo, & Tamm
(2006) Nature Chem. Biol.
11:627-635
Arg 138
Glu 52
Glu 128
Lys 82
How about NMR of larger
membrane proteins?
larger porins of bacterial outer
membranes
Outer membrane protein G
(OmpG)
• facultative porin in outer membrane of E. coli
• facilitates uptake of large oligosaccharides
• only up-regulated when LamB is down-regulated
• monomeric
• 280 residues (33 kDa)
• engineerable ion channel
• potential use in biosensor development
• no crystal structure available when project started
boiled
+ Pr-K
refolded
Initial attempts resulted in poor
TROSY spectra despite apparent
good refolding
33 kDa
28 kDa
following extraction and refolding protocols by
Conlan and Bayley
Jason Schmittschmitt
Spring 2004
Further Optimization of Refolding
• Optimizing the folding of OmpG, using different
methods to determine the refolding efficiency:
Methods for determining refolding:
SDS-PAGE band shift essay
CD
Trp Fluorescence
• Testing different conditions for NMR sample
preparation:
micelle type and concentration, pH etc
15N-labeled proteins, TROSY spectra
DPC vs. !-OG at various pH
8
6.5
6.5
10
9
8
6.5
6.5
Proteinase
K digest
9
control
10
refolding
pH =
15mM DPC, 37ºC overnight
70mM !-OG, 37ºC overnight
OmpG: 25 µM
OmpG: 25 µM
Effect of pH on TROSY in !-OG
10 mM Tris, pH 9
10 mM phosphate, pH 6.3
• Lower pH leads to 20% increase in peak numbers
Summary of OmpG Refolding
• The effect of folding aids tested so far is insignificant
• OmpG concentration has some effect, in general,
lower concentration is better for achieving a high
level of refolding
• The best conditions:
– 10 mM Tris, pH 9.0
– 70 mM !-OG (~2%)
– 37ºC, overnight or longer
Planar bilayer (single channel)
recordings of OmpG
pA
Time (s)
Buffer: 1M KCl, 10mM Tris, pH 7.2,
OmpG stock: 25 µM in 70 mM !-OG, 10 mM Tris, pH 9
Applied voltage: 40 mV, sampling rate: 1kHz
OmpG is a porin
pA
Time (s)
# of events
5000
4000
3000
2000
0.8 nS in 1 M KCl
pH 7.4 at 40 mV
1000
20
40
60
80
100
pA
120
140
160
180
200
15N,13C,2H-labeled
15N-1H
TROSY with 13C decoupling
OmpG sample
15N-1H
TROSY without 13C decoupling
• Triple labeling and sample refolding are successful
Long-term stability of OmpG in !-OG
• The average linewidth increased by 3% after 15 hours at 40 ºC
• Samples show a faint whitish precipitate after several days at 40 ºC
Detergent exchange to DPC
• Fresh concentrated samples refolded in !OG were
diluted into DPC micelle solutions
• Allowed to sit overnight at room temperature
• Mixed OmpG/DPC/!-OG solution were concentrated
in Amicon using a 30K NMWL membrane
• Diluted and concentrated one more round with DPC
micelle containing buffer
15N-1H
TROSY of 15N,D-OmpG
exchanged into DPC measured at 500 MHz
•
•
OmpG is still properly folded in DPC and general features are the
same
Peaks are broader probably due to the larger size of the
OmpG/DPC compared to OmpG/!-OG complex
Triple-resonance TROSY experiments at
800 MHz
• HNCA, HN(CO)CA
HN(CA)CB, HN(COCA)CB
HNCO, HN(CA)CO
• Different versions of these experiments were tested
for the best spectral quality
HNCA Strip Plot (71-92)
HN(CA)CB Strip Plot (71-92)
HNCO and HN(CA)CO Strip Plot (71-81)
OmpG: 33 kDa, 280 residues
800 MHz TROSY
234 residues assigned, 9 partially assigned, 12 tentatively assigned = 255/280
C" and C! secondary chemical shifts
•
•
•
Plot of three-bond averaged secondary chemical shifts versus residue numbers
The strong negative !C"-!C# values identify 14 beta strands.
The markedly positive !C"-!C# region indicates the existence of a short helix between
strands 7 and 8. However, this helix it was not confirmed in the structure calculation.
133 long range NOEs define topology
and strand connectivity of OmpG
from 15N-1H-1H NOESY-TROSY and 15N-15N-1H TROSY-NOESY-TROSY experiments
NOE measurements and distance restraints
• 316 unique NOE distances
– 137 sequential
– 46 medium-range
– 133 long-range (more than 4 residues apart)
• 262 dihedral angle restraints
• 55 H-bonding pairs (220 distances)
Ensemble of 10 Lowest-Energy Structures
w/o H-bonds
with H-bonds
Including H-bonds improves backbone r.m.s.d. from
2.54 ± 0.47 to 1.67 ± 0.29 Å (!-sheet and turn residues)
Comparison with X-ray crystal structures
closest-to-mean NMR
(2JQY)
pH 6.3
closed state (2IWW)
pH 5.6
open state (2IWV)
pH 7.5
open state (2F1C)
pH 5.5
Comparison of NMR and Crystal
Structures
r.m.s.d. accuracy of NMR vs. different crystal structures
1.65 to 1.71 Å
Backbone Dynamics by Heteronuclear NOEs
Ribbon Representation of Lowest Energy Conformer of
OmpG (largest membrane protein - 33 kDa - solved by NMR to date)
Liang and Tamm (2007) PNAS 104:16140
What’s next?
Improve quality of NMR structure by measuring
RDCs and PREs
Can we learn more about gating from
measurements of membrane protein dynamics?
Specifically, can we explain pH-dependent
gating?
What is the role of the lipid matrix in all of this?
Gels to introduce weak alignments to measure
Residual Dipolar Couplings of OmpG
7%
5%
4.3 %
3.6 %
Original charged gel # Swelling in H2O # Cut # Dried # Re-Swelling
in protein/micelle solution # Sample made # Reposition piston
Two types of acrylamide-based copolymer
gels are used to obtain multiple alignments
1. Negatively charged gels (7%):
2-Acrylamido-2-methyl-1-propanesulfonic acid (50$S and 75$S)
2. Positively charged gels (6% and 7%):
(3-Acrylamidopropyl)trimethylammonium chloride (25+M, 50+M, 75+M)
Change of Alignments
50-S, 13mm
isotropic
92.8 Hz
95.4 Hz
50-S, 15mm
101.3 Hz
(8.5 Hz)
109.4 Hz
(16.6Hz)
98.9 Hz
(3.5Hz)
50-S, 17mm
100.8 Hz
(5.4 Hz)
95.8 Hz
(3 Hz)
95.2 Hz
(-0.2 Hz)
HNCO based RDC experiments
HN-N
N-C’
C’-C"
A46
RDC = 12.6 Hz
-3.8 Hz
-2.6 Hz
1.2 Hz
0 Hz
N80
RDC = -21.9 Hz
Refinement of OmpA Structure with Residual
Dipolar Couplings
measured in two different alignments
in two different (+,–) charged compressed gels
421 RDC couplings:
74 + 71
72 + 74
73 + 70
1D
1D
1D
HN
NC’
C’C"
plus:
90 NOE restraints
142 dih. angle restraints
68 H-bond restraints
rmsd precision: 0.62 ± 0.16
rmsd accuracy: 1.11 ± 0.06
Cierpicki et al. (2006) JACS 128, 4389
Conformations of periplasmic turns
quite well defined by RDCs
type I or II
type I or II
type I
Parallel Site-Directed Spin-Labeling and Paramagnetic Relaxation
Enhancement to Refine Solution Structure of OmpA
spin-labels introduced at 11 sites:
Solomon-Bloembergen equation:
& K
r = $ sp
$% R2
,
3- c )#
** 4- c +
'
2 2 '!
1 + . h - c (!"
+
I para
R2 exp(! R2sp t )
=
R2 + R2sp
I dia
1/ 6
Liang et al. (2006) JACS 128:4389
Correlation Between Experimental NMR Distances Measured
by PRE and Best Fit Distances to Paramagnetic Centers in
Crystal Structure with Grafted MTSSL
Fitted Distance (Å)
N46R1
T88R1
+ T132R1
A11R1
M53R1
T95R1
L139R1
Diagonal
± 2Å limits
Experimental Distance (Å)
Structures of OmpA Calculated With and Without PRE Restraints
no PRE restraints
90 NOE restraints
142 dih. angle restraints
68 H bond restraints
rmsd precision: 1.25 ± 0.29
rmsd accuracy: 1.62 ± 0.19
320 PRE restraints
90 NOE restraints
142 dih. angle restraints
68 H bond restraints
rmsd precision: 0.85 ± 0.17
rmsd accuracy: 1.09 ± 0.12
In Summary (NMR):
• Solution NMR has gained a significant role in structural biology
of membrane proteins (now extended to 33 kDa protein)
• Functions of membrane proteins may be studied by NMR-based
dynamical studies
• Interactions with specific lipids and substrates may be studied in
ways that may not be accomplished by crystallography
• NMR of membrane proteins still not easy: sample preparation
continues to be difficult, new NMR methods continue to be
developed and tailored for use with membrane proteins,
especially "-helical membrane proteins (harder)
• Limitations of detergent micelles as solvents: excellent to get a
good start, but also try bicelles, ss-NMR, and SDSL-EPR in
bilayers for verfication
Thanks to:
NMR Structure and Dynamics
Binyong Liang
Ashish Arora
Ming-tao Pai
Membrane Protein Folding
Heedeok Hong
Ricardo Flores, Sangho Park
Dennis Rinehart
Jörg Kleinschmidt
Single Channel Recordings
Ashish Arora, Heedeok Hong
Binyong Liang, Luiz Salay
Collaborators:
John Bushweller, Tomek Cierpicki, Jeff Ellena, Gabor
Szabo
(U. of Virginia)
Funding:
NIGMS
Chapter on solution
NMR of membrane
proteins
Recent review
by Ashish Arora in
Tamm and Liang
“NMR of Membrane
Proteins in Solution”
Progress in NMR
Spectroscopy
48, 201-210 (2006)
Wiley-VCH 2005